The Smallest Biomolecules: Diatomics and their Interactions with Heme Proteins

THE SMALLEST BIOMOLECULES Diatomics and their Interactions with Heme Proteins

This page intentionally left blank

THE SMALLEST BIOMOLECULES Diatomics and their Interactions with Heme Proteins

Edited by ABHIK GHOSH Department of Chemistry, University of Tromsø Tromsø, Norway

Amsterdam • Boston • Heidelberg • London • New York • Oxford Paris • San Diego • San Francisco • Singapore • Sydney • Tokyo

Elsevier Radarweg 29, PO Box 211, 1000 AE Amsterdam, The Netherlands The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK First edition 2008 Copyright © 2008 Elsevier B.V. All rights reserved No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the publisher Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone (+44) (0) 1865 843830; fax (+44) (0) 1865 853333; email: [email protected]. Alternatively you can submit your request online by visiting the Elsevier web site at http://elsevier.com/locate/permissions, and selecting Obtaining permission to use Elsevier material Notice No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress ISBN: 978-0-444-52839-1 For information on all Elsevier publications visit our website at books.elsevier.com Printed and bound in Italy 08 09 10 11 12 10 9 8 7 6 5 4 3 2 1 The cover depicts a DFT model of a unique dinitrosylheme intermediate, which is discussed in the chapter by Peter Ford.

Working together to grow libraries in developing countries www.elsevier.com | www.bookaid.org | www.sabre.org

Contents

Preface

PART I

ix

INTRODUCTORY OVERVIEWS

1. Mammalian Myoglobin as a Model for Understanding Ligand Affinities and Discrimination in Heme Proteins John S. Olson and Abhik Ghosh

1 3

2. A Surfeit of Biological Heme-based Sensors Marie-Alda Gilles-Gonzalez and Gonzalo Gonzalez

18

3. NO and NOx Interactions with Hemes Peter C. Ford, Susmita Bandyopadhyay, Mark D. Lim, and Ivan M. Lorkovic

66

PART II

ELECTRONIC STRUCTURE AND SPECTROSCOPY

4. CO, NO, and O2 as Vibrational Probes of Heme Protein Active Sites Thomas G. Spiro, Mohammed Ibrahim, and Ingar H. Wasbotten 5. Nuclear Resonance Vibrational Spectroscopy — NRVS W. Robert Scheidt, Stephen M. Durbin, and J. Timothy Sage 6. EPR and Low-temperature MCD Spectroscopy of Ferrous Heme Nitrosyls Nicolai Lehnert

PART III

ASPECTS OF HEMOGLOBINS (EXCEPT HEME NOx INTERACTIONS)

93 95

124

147

173

7. Protoglobin and Globin-coupled Sensors Tracey Allen K. Freitas, Jennifer A. Saito, Xuehua Wan, Shaobin Hou, and Maqsudul Alam

175

8. Neuroglobin and Cytoglobin Thomas Hankeln and Thorsten Burmester

203

vi

Contents

9. Extreme pH Sensitivity in the Binding of Oxygen to Some Fish Hemoglobins: The Root Effect T. Brittain

219

10. Microbial Hemoglobins: Structure, Function, and Folding Changyuan Lu, Tsuyoshi Egawa, Dipanwita Batabyal, Masahiro Mukai, and Syun-Ru Yeh

235

PART IV

267

HEME

NOx INTERACTIONS

11. The Reaction between Nitrite and Hemoglobin: The Role of Nitrite in Hemoglobin-mediated Hypoxic Vasodilation Daniel B. Kim-Shapiro, Mark T. Gladwin, Rakesh P. Patel and Neil Hogg

269

12. Nitric Oxide Dioxygenase: An Ancient Enzymic Function of Hemoglobin Paul R. Gardner and Anne M. Gardner

290

13. Respiratory Nitric Oxide Reductases, NorB and NorZ, of the Heme–Copper Oxidase Type Walter G. Zumft

327

14. Nitric Oxide Reductase (P450nor ) from Fusarium oxysporum Andreas Daiber, Hirofumi Shoun and Volker Ullrich 15. Nitric Oxide Interaction with Insect Nitrophorins and Possibilities for the Electron Configuration of the {FeNO}6 Complex F. Ann Walker

354

378

16. Bioinorganic Chemistry of the HNO Ligand Filip Sulc and Patrick J. Farmer

429

PART V

463

SELECTED ENZYMES AND SENSORS

17. Ligand-Protein Interactions in Mammalian Nitric Oxide Synthase Denis L. Rousseau, David Li, Eric Y. Hayden, Haiteng Deng and Syun-Ru Yeh

465

18. CooA: A Paradigm for Gas-sensing Regulatory Proteins Gary P. Roberts, Robert L. Kerby, Hwan Youn and Mary Conrad

498

19. Soluble Guanylyl Cyclase and Its Evolutionary Relatives Eduardo Henrique Silva Sousa, Gonzalo Gonzalez, and Marie-Alda Gilles-Gonzalez

524

Contents

vii

20. Resonance Raman Studies of the Activation Mechanism of Soluble Guanylate Cyclase Biswajit Pal and Teizo Kitagawa

540

21. Insights into Heme-based O2 Sensing from Structure–Function Relationships in the FixL Proteins Kenton R. Rodgers, Graeme R.A. Wyllie, and Gudrun S. Lukat-Rodgers

564

Index

597

This page intentionally left blank

Preface

This is not a book about NO, or about hemoglobin, or indeed about heme-based sensors per se. Of course, it covers all of these subjects and more, but above all, the book’s raison d’être is to present a multifaceted overview of the interaction of diatomic ligands with heme proteins. An overarching goal is to build bridges among disciplines, to bring about a meeting of minds. The contributors to this book hail from diverse university departments and disciplines – chemistry, biochemistry, molecular biology, microbiology, zoology, physics, medicine and surgery, bringing with them diverse viewpoints on heme–diatomic interactions. My hope is that the juxtaposition of this diversity will lead to increased exchanges of ideas, approaches, and techniques across traditional disciplinary boundaries. In other words, the goal is to inspire bioinorganic chemists to think about physiological problems and biomedical researchers to think along bioinorganic lines and so on. The book starts with two introductory overviews. The first, by Olson and Ghosh, recounts the upheaval in recent years of our conception of how heme proteins discriminate among the three very similar-sized diatomics CO, NO, and O2 . The old picture that an oxygen-carrier such as myoglobin discriminates against CO by forcing the bound ligand into a high-energy bent conformation has given way to a view where protein electrostatics (especially hydrogen-bonding interactions) is the key agent of discrimination. The second chapter, by Gilles-Gonzalez provides an overview of the rapidly burgeoning field of heme-based sensors, a major theme of the book. The next set of chapters on electronic structure and spectroscopy focus on wellestablished techniques – resonance Raman (Spiro and coworkers) and EPR (Lehnert) as well as on a new spectroscopy, nuclear resonance vibrational spectroscopy (Scheidt and Sage), that promises to become a part of the bioinorganic chemist’s toolbox. Chapters 6–10 focus on a variety of somewhat specialized aspects of globins, including protoglobins and globin-coupled sensors (Alam), neuroglobin and cytoglobin (Burmester and Hankeln), root effect hemoglobins (Brittain), and hemoglobins from unicellular organisms (Yeh). The next six chapters (nos. 11–16) provide a fairly comprehensive overview of heme– NOx interactions. Chapter 11 (Kim-Shapiro et al.) focuses on a major biomedical story that is currently unfolding, namely the role of nitrite in hemoglobin-mediated hypoxic vasodilation. Similarly, Chapter 12 addresses another recent development, the role of hemoglobin as a nitric oxide dioxygenase. The next few chapters cover other important topics: respiratory nitric oxide reductases (Zumft), P450nor (Daiber, Shoun, Ullrich), insect nitrophorins (Walker), and the bioinorganic chemistry of the HNO ligand (Sulc and Farmer). The book concludes with a set of chapters (nos. 17–21) on specific enzymes and sensors: nitric oxide synthase (Rousseau et al.), CooA (Roberts et al.), soluble guanylate cyclase (Sousa et al. and Pal and Kitagawa), and FixL (Rodgers and Lukat-Rodgers).

x

Preface

Early versions of several of the chapters appeared as review articles in the Journal of Inorganic Biochemistry (JIB), where many of them quickly ranked among the most downloaded JIB articles. Indeed, this success, along with the considerations outlined above, ultimately led me to put together this volume. My sincerest thanks go to all authors for taking time out of their busy schedules to contribute to this unique collaborative project. In addition, I am most grateful to Prof. John Dawson of the University of South Carolina for much helpful advice. In conclusion, to the broader heme community, I would say: I hope you find this book useful and enjoyable; should any comments or criticisms occur to you, please do not hesitate to contact me. Abhik Ghosh Tromsø, June 2007

Part I Introductory Overviews

This page intentionally left blank

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 1

Mammalian Myoglobin as a Model for Understanding Ligand Affinities and Discrimination in Heme Proteins John S. Olsona and Abhik Ghoshb a

Department of Biochemistry & Cell Biology, Rice University, Houston, TX, USA Department of Chemistry and Center for Theoretical and Computational Chemistry, University of Tromsø, N-9037 Tromsø, Norway b

1. INTRODUCTION The binding of the diatomic gases O2 , CO, and NO to the iron atom in heme proteins is involved in a wide variety of crucial physiological functions, including respiration, O2 transport and storage, vasoregulation, neuronal function, transcriptional control, and heme degradation, all of which are discussed in the chapters of this book. Each system has its own unique structural and physiological features; however, there are certain fundamental principles associated with ligand binding to heme iron that can be generalized into a useful mechanistic framework. Recombinant mammalian myoglobin has served as a convenient model system for exploring these general features [1–5], and the purpose of this initial chapter is to describe the currently accepted mechanism for O2 , CO, and NO binding to myoglobin as a basis for interpreting ligand binding to the wide variety of heme proteins discussed in the rest of this book. Many of the chapters in this book are devoted to the reactions of NO, including both reversible binding to ferric and ferrous forms and the dioxygenation of NO by bound O2 . Thus, for the most part, our focus is on O2 and CO binding, but the resultant ideas and mechanisms are applicable for understanding NO binding and NO dioxygenation. Both of these latter processes are key secondary in vivo functions of mammalian Mbs and Hbs and have been reviewed extensively by us and others over the past 10 years [6–8]. Most substrates have complex shapes and charge distributions that are complemented by binding sites on protein surfaces, which lead to high affinity and specificity. Proofreading to obtain the right binding partner occurs during the association process, and poor substrates are rejected because they do not fit or have the right electrostatic complementarities. In the case of the simple diatomic gases, ligand discrimination only occurs during and after coordination with the iron atom. As explained below, steric hindrance plays only a small role in this discrimination, in part because the differences between the bent Fe O O and Fe N O geometries and the mostly linear Fe C O complexes are small and because angular deformation of these complexes is not as unfavorable as once thought. The key factor causing ligand discrimination is differential electrostatic stabilization of the partial charges on the bound ligand atoms, which are small in the case of

4

J.S. Olson and Abhik Ghosh

bound CO, moderate in the case of NO, and large in the case of O2 [2,9–15]. At the same time, protein regulation of distal iron accessibility or proximal coordination geometry can affect the absolute affinities of all three ligands over 10,000-fold ranges compared to simple, unhindered hexacoordinate complexes with the same proximal base [2,16–21].

2. LIGAND CONFORMATION AND DISCRIMINATION Our understanding of ligand stereochemistry and deformability in heme proteins and their importance in ligand discrimination has increased enormously over the last 10–15 years [9,22]. The textbook explanation, based on early MbCO crystal structures, is that the protein forces the heme-bound CO into a high-energy bent conformation, whereas such a conformation is natural for bound O2 . On the basis of high-resolution recombinant MbCO crystal structures [23,24], structures for native MbCO that take into account heme disorder [25], and IR absorption and photoselection studies [26,27], this picture changed dramatically in the 1990s. These studies indicated a stiff, upright FeCO unit with an Fe C O angle ≥160 . Thus, the high FeCO bending frequency, ∼550 cm−1 , of carbonylhemes (which is higher than the Fe C stretching frequency of ∼480–510 cm−1 ) suggested an essentially nondeformable FeCO unit. However, this picture also turned out to be flawed. In the mid-1990s, using DFT calculations, Ghosh and Bocian showed that FeCO units are in fact extremely flexible with respect to cooperative tilting and bending of the MXO unit but remarkably stiff when tilting and bending occur in opposite directions, the tilting and bending angles being defined in Fig. 1 [28,29]. This idea was soon confirmed by Spiro and coworkers [30], while Nakamoto and coworkers [31] provided a molecular orbital explanation for this effect. Although cooperative tilting and bending hardly disrupts M X O -bonding, out-of-phase tilting and bending does. In light of the cooperative tilting and bending potential energy surface (for which a mathematical expression was given by Ghosh and Bocian [28]), the 550 cm−1 FeCO vibration could be assigned to the high-energy out-of-phase tilting and bending mode. More recently, the concept of cooperative tilting and bending has been extended to NO complexes [22]. Small-molecule crystallography [32] and DFT calculations [22] for Fe(II) NO porphyrins have revealed a small but distinct tilting of the Fe N(O) vector τ

β

O C N

Fe

N N

N His

Fig. 1. Definition of the tilting () and bending () angles, relative to the heme normal.

Mammalian Myoglobin

5

(in addition to FeNO angles ∼140 ). Fe(III) NO porphyrins are effectively low-spin d6 complexes, and the FeNO unit is generally linear. However, the presence of strong-field sixth ligands such as thiolate (as in some ferriheme proteins such as nitrophorin 4 [33] as well as in a synthetic model complex [34]) or aryl ligands (as in a synthetic model complex [35]) also leads to substantially tilted (∼9 ) and bent FeNO units with FeNO angles ∼156 (i.e., ∼24 ). Regardless of the details of the minimum-energy geometry, the key point is that cooperative tilting and bending provides a very low-energy pathway for deformation of the diatomic ligand. Thus, DFT calculations showed that the upright conformation of an S=0 Fe(Por)(NO)(Ar) (Por = porphyrin, Ar = aryl) model complex is just 1 kcal mol−1 higher in energy, relative to the observed bent conformation [22]. The overall conclusion emerging from all these studies is that any reasonable amount of deformation (i.e.  + ≈35 ) cannot account for the ∼4 kcal mol−1 by which Mb discriminates in favor of O2 and against CO, relative to protein-free heme. In contrast, although the entropic effects involved are difficult to estimate theoretically, DFT calculations of hydrogen bond strengths for heme-bound CO, NO, and O2 ligands do a fair job of accounting for the observed discrimination energies [36–38].

3. WATER OR ENDOGENOUS LIGAND DISPLACEMENT Ligand binding to myoglobin involves a minimum of four steps, which are shown in Fig. 2, and this mechanism serves as a model for all single domain globins

Mb

deoxyMb(H2O) Kentry

KH2O H2O

Mb

MbXs

X Kbond Kstabilization MbX

Fig. 2. Mechanism of ligand binding to mammalian myoglobin. The four-step mechanism originally described by Olson and Phillips [2] was based on the crystal structures of sperm whale deoxyMb (2mgl, [24]), the photoproduct of low temperature MbCO crystals (1ABS, [48,49]), and room temperature sperm whale MbCO (2mgk, [24]). Similar photoproduct structures for Mb•••X have now been observed at room temperature in time-resolved X-ray crystallography studies ([50,51,61–63]). The lower states, deoxymyoglobin with an “empty” distal pocket, Mb, and liganded myoglobin without any stabilization of the bound ligand, MbX cannot been “seen” by crystallography for the wild-type protein, but can be simulated by mutation of the distal histidine to an apolar amino acid that cannot stabilize either internal water or bound ligands [24]. (see Plate 1.)

6

J.S. Olson and Abhik Ghosh

and gas-binding heme proteins. The first step involves the displacement of water noncovalently bound in the distal pocket of deoxyMb [1,2,24], creating an open or “empty” distal pocket containing pentacoordinate heme iron. In other globins and heme proteins or in the oxidized form of most myoglobins and hemoglobins, water or internal His, Tyr, Lys, Pro, or Met side chains may be coordinated directly to the iron and must also be displaced before the diatomic gases can bind. Well-characterized examples of O2 and CO binding proteins containing endogenous ligands to regulate rates and affinities of ligand binding are nonsymbiotic plant hemoglobins [39,40], neuroglobins [41–44], and CooA [45–47], which are described in the succeeding chapters of this book. In mammalian deoxyMbs, the inhibitory effect of displacing noncovalently bound water is about 10-fold and applies to the binding of all three diatomic gases, whereas in metMbs, the coordinated water inhibits ferric ligand binding roughly 1000-fold [2].

4. LIGAND ENTRY IN THE DISTAL PORTION OF THE HEME CAVITY The second step in binding involves ligand movement into the protein to form an intermediate in which the gas is trapped in the active site near the heme group. This “docking” or “B” state is readily observed as a discrete spectral species in low temperature or timeresolved room temperature FTIR and X-ray crystallographic experiments [48–54]. In terms of an equilibrium mechanism, this Mb•••X species represents noncovalent binding of the ligand inside the protein matrix. The ligand is not always at the initial “docking” or B state shown in the middle panel of Fig. 2, but is rapidly equilibrating between all the available positions in the distal pocket, including the “docking site” and the Xe4 cavity located deeper in the pocket behind Leu(B10)29, Ile(G8)107, and Val(E11)68 [4,55–59]. The equilibrium constant for noncovalent binding into the “empty” active site, Kentry , depends primarily on the volume of the distal cavity. Larger cavities result in greater equilibrium constants [55–57,60].

5. IRON-LIGAND BOND FORMATION The third step in ligand binding involves bond formation with the iron atom. The equilibrium constant for this process, Kbond , is determined primarily by the intrinsic reactivity of the ligand molecule and then by the ease of in-plane movement of the iron atom and its steric accessibility. All three of these factors contribute significant enthalpic barriers to the internal binding of CO and O2 . The first factor is governed by the quantum mechanics of iron-ligand coordination [64–68]. In the case of NO, there is little or no enthalpic barrier regardless of the exact position of the iron atom [64,66,69]. In general, the intrinsic strengths of the final bonds are Fe NO>>Fe CO>>Fe O2 , and we previously estimated Kbond values for wild-type Mb to be ∼103 , ∼106 , and ∼1010 for O2 , CO, and NO binding, respectively [2]. Kbond is also regulated by protein structure, including the proximal base-heme coordination geometry and steric interactions between distal amino acids and the bound ligand.

Mammalian Myoglobin

7

Finally, there is the additional unfavorable requirement of fixing the position of the ligand right above the iron atom during internal rebinding instead of allowing it to be spatially distributed in the distal portion of the heme pocket. This fourth factor governing Kbond involves the entropic barrier to bond formation from within the protein [57]. The Mb•••X intermediate is, in effect, a relatively long-lived (10 to several 100 ns) transition state. Increasing the equilibrium constant for its formation, Kentry , by making the distal cavity larger and more entropically favorable will decrease Kbond by roughly the same factor, due to the unfavorable entropic effect of confining the ligand next to the iron atom. Thus, to a first approximation, increasing the size of the distal cavity will increase Kentry , decrease Kbond , and cause little change of the product Kentry Kbond . Decreasing the volume will have the opposite effects on the individual constants, but again cause little change in Kentry Kbond and overall ligand affinity (see Equation 1 below). However, the size of the distal cavity can have large effects on the rates of ligand association and dissociation, depending on the rate-limiting steps for these processes [55,57].

6. ELECTROSTATIC STABILIZATION OF BOUND LIGANDS The fourth step in ligand binding to Mb involves stabilization of the bound ligand by hydrogen bonding or electrostatic interactions with polar residues in the active site. Describing favorable electrostatic interactions as a separate equilibrium step is a somewhat arbitrary but useful formalism. In the case of hydrogen bond donation by the distal histidine, it may physically be a separate step. First, in deoxyMb, the major imidazole tautomer is N-H with the N lone electron pair accepting a hydrogen bond from noncovalently bound water [10]. Thus, hydrogen bond donation to bound ligands in the MbXs state requires tautomerization of the His(E7)64 side chain. Second, ligand entry into the distal pocket requires transient upward and outward rotation of the distal histidine to create a channel from solvent over the heme propionates and into the active site. Closure requires the imidazole side chain to rotate back down toward the heme plane and, after the iron coordination, further movement to within less than 3.0 Å of the ligand atoms. Third, it is clear that the distal histidine can occupy multiple conformational states when CO is bound as manifested by the multiple A states reported for MbCO [10,70]. In the case of MbO2 , this heterogeneity is reduced to two very similar conformations in which the N atom is 0.3 Å closer to the O(2) atom than in any of the MbCO orientations, indicating an inward movement of the His64 side chain [10]. Similar minor conformational heterogeneity is seen in sperm whale and horse heart MbNO complexes, and in these structures, the conformation of the Fe(II) N O complex also varies, due presumably to subtle changes in the sequences of the two different species [71,72]. The fourth reason for making stabilization a separate step is to provide a simple way of distinguishing it from unfavorable steric and electrostatic interactions. For example, if water is removed from the deoxyMb structure shown in Fig. 2 and then O2 is bound to the iron atom, the O(2) atom will clash sterically with the N atom of the distal histidine. More importantly, the imidazole will still be present as the NH tautomer, with the N lone pairs pointing a negative field vector at the partially negative bound O2 , destabilizing it. Both of these factors would cause Kbond to be

8

J.S. Olson and Abhik Ghosh

small. However, His64 moves away to a more optimal position and isomerizes to the N-H tautomer for optimal formation of a strong and favorable hydrogen bond that appears to stabilize bound oxygen by ∼1000-fold in Mb. Thus, in the mechanism in Fig. 2, unfavorable distal electrostatic and steric effects are lumped with proximal effects in the Kbond term, and any active site relaxations that lead to favorable polar effects are evaluated separately as the Kstabilization term. The alternative is to have a branched scheme in which the Mb•••X intermediate converts to stabilized and destabilized bound states, each of which would have a different Kbond term, one significantly larger than the other. The ratio of these constants would be equal to the value of Kstabilzation in the linear scheme and describe isomerization between the destabilized and stabilized states.

7. FeC–O STRETCHING FREQUENCY, ELECTROSTATIC FIELDS, AND O2 DISSOCIATION RATE CONSTANTS A fifth reason for describing electrostatic stabilization as a separate step is that it allows a simple way of interpreting and correlating FTIR spectra of CO complexes and O2 dissociation rate constants. As shown in Fig. 3 and described in detail by Phillips et al. [10], the area-averaged C−O value is an empirical measure of the electrostatic field in the vicinity of bound ligands and can be correlated with O2 dissociation rate and

35

4

(B) Correlation of log(kO2) with νC

O

A3

25 20 Positive

10

3

R2 = 0.93

A1

log(kO2) s–1

Electrostatic field (kcal/mol)

30

15

5

(A) Correlation of Electrostatic Field with νC O for SW MbCO

A0

5

2

SW Mb R2 = 0.81

1

Ascaris D1

CerHb, AscHb R2 = 0.85 Lba(.......) R2 = 0.50

(+)

(0)

0

0 –1

Neutral

–5

–2

–10

(–)

Negative –15 1900

1920

1940

νc

o

1960

(cm–1)

1980

2000

–3 1900

1920

1940

νc

o

1960

1980

2000

(cm–1)

Fig. 3. Correlations between the electrostatic field at the second ligand atom, C–O stretching frequency, C−O , and the O2 dissociation rate constant (KO2 ) for sperm whale Mb, soybean Lba, and Cerebratulus Hb mutants. (A) The strong linear dependence on the calculated electrostatic fields and the observed stretching frequencies of bound CO for 20 different mutants of sperm whale Mb [10]. (B) Correlation between log(KO2 ) and C−O for Mb [10], Lba [73], CerHb mutants (Olson, Blouin, Moens, Salter, Hale, Nienhaus, and DeWilde unpublished), and Ascaris domain 1 Hb [74].

Mammalian Myoglobin

9

equilibrium constants. There is a strong inverse linear relation between the CO stretching frequency of bound CO and the electrostatic field at the second ligand atom, calculated from the MbCO crystal structure (Fig. 3A, [10]). More importantly, there is also a linear correlation between C−O and the logarithm of the oxygen dissociation rate constant for over 30 different Mb mutants (Fig. 3B). Lower C−O peaks imply more positive fields, which in turn preferentially stabilize bound O2 . This stabilization is observed most directly by a decrease in the thermal rate of O2 dissociation, which requires disruption of the hydrogen bond or electrostatic interaction. The linear relationship between C−O and log (KO2 ) indicates that C−O is also a good empirical predictor of oxygen affinity, and in the latter case, the relationship is reciprocal; low C−O frequencies predict high KO2 values [10].

8. FORMALISM FOR INTERPRETING LIGAND BINDING CONSTANTS The factors governing O2 binding to heme proteins can be interpreted using the following equation as a framework:  Koverall =

1 1 + KH2 O H2 O

 Kentry Kbond 1 + Kstabilization 

(1)

This equation was derived from the scheme in Fig. 2, as described in Olson and Phillips [2]. In the case of mammalian myoglobins, ligand entry is inhibited by the presence of a noncovalently bound water molecule hydrogen bonded to His(E7)64 (Fig. 2).  The fraction of “empty” protein, Mb in Fig. 2, is given by  1/ 1 + KH2 O H2 O , which is ∼0.10 at pH 7.0, 20 C. Kentry is small, ∼20 M−1 , and proportional to the size of the distal pocket, whereas Kbond is large and depends on the nature of the ligand molecule, varying in the order NO>>CO>O2 [2]. Kbond is also strongly dependent on the proximal geometry of the His(F8) Fe bond, which in the case of Mb is unfavorable (Fig. 4B). The final step involves preferential stabilization of the highly polar Fe + − O2 − complex by hydrogen bonding to HisE7 (Figs. 2 and 4A). We have estimated that Kstabilization is ∼1000 in wildtype MbO2 based on 500–1000-fold increases in the rate of Fe O2 bond dissociation when the distal histidine is replaced with an apolar amino acid (Table 1, HisE7 to Phe mutation for SW Mb [2]). In contrast, the FeCO complex is neutral. Although the stretching frequency of bound CO is shifted to lower frequencies in the presence of positive field vectors, the change in bond order is small, and Kstabilization for CO is only 2–5 in wild-type MbCO. Thus, hydrogen bond donation by His(E7) results in a net 100-fold stabilization of bound O2 (i.e., Kstabilization • 1/ 1 + KH2 O H2 O  ≈ 1000 • 0 1 = 100), whereas the polarity of His(E7) inhibits CO binding due to the unfavorable effect of stabilizing water in the distal pocket of deoxymyoglobin (i.e., Kstabilization • 1/ 1 + KH2 O H2 O  ≈ 2 − 5 • 0 1 = 0 2 − 0 5 [2].

10

LeuB10

TyrB10 HisE7

SW MbO2

SW Mb

LbaO2

ValE11

LeuE11 ValG8

IIeG8

HisE7

eclipsed

HisF8

HisF8

(A)

Cerebratulus HbO2 GlnE7

Cerebratulus Hb

staggered

staggered

ThrE11

GlnE7

AlaG8

HisF8

Fig. 4. Proximal geometries and electrostatic interactions in four globins. Panel A, Distal pockets of SW MbO2 (2mgm); postulated structure of soybean LbO2 , (1bin, [73]). Domain 1 of Ascaris HbO2 (1ash, [83]); and Cerebratulus HbO2 (1kr7,[84]). Panel B, Proximal His(F8) plane orientations of the globins shown in Panel A. Only SW Mb shows an eclipsed orientation with the edge of the imidazole ring directly beneath two of the pyrrole nitrogens, a conformation that inhibits in-plane movement of the iron atom. (see Plate 2.)

J.S. Olson and Abhik Ghosh

HisF8

Ascaris Hb TyrB10

IIeE11

PheG8

staggered

(B)

TyrB10

Ascaris HbO2

Soybean Lba

Globin

kO2 M−1 s−1

1. SW Mb Wild-type HisE7 to Phe HisF8 to Gly(+Im)

17 74 20

2. Soybean Lba Wild-type HisE7 to Phe TyrB10 to Phe HisE7 to Arg HisF8 to Gly(+Im)

130 130 75 250 120

3. Ascaris Hb Native Wild-type TyrB10 to Phe

1 5 2 8 40

kO2 s−1 15 10 000 3 8 5 6 280 0 75 0 20 15 0 004 0 013 2 0

4. Cerebratulus Hb Wild-type ThrE11 to Val

240 30

180 0 18

5. Human HbA R state subunits T state subunits

40–150 6–10

16–40 2000–4000

KO2 M−1 1 1 0 0074 5 2 23 0 45 100 1200 8 370 220 20 1 3 170 2 4 0 003

kCO M−1 s−1

kCO s−1

0 51 4 5 1 2

0 020 0 054 0 011

15 24 11 80 11 0 21 0 35 2 7 20 3 0 4–7 0.05–0.10

0 0084 0 013 0 0068 0 0060 0 019 0 018 N.D. N.D. 0 048 0 0070 0.005–0.010 0.1–0.2

KCO M−1

M(KCO /KO2 )

27 83 110

25 11000 21

1800 1800 1600 13000 580

78 4000 16 11 73

12 N.D. N.D. 580 430 700 0.5

Mammalian Myoglobin

Table 1. Effects of mutagenesis on O2 and CO binding to three animal globins, soybean Lba, and human HbA. The values for Mb, Lba, Ascaris Hb, and Cerebratulus Hb were measured at pH 7.0, 20 C. The values for human HbA were measured at pH 7.4, 25 C. The rate constants for sperm whale myoglobin, soybean Lba, Ascaris suum Hb, and Cerebratulus Hb were taken from references [21,76–79] respectively. For the HisF8 to Gly mutations, Kundu et al. [21] added external imidazole to provide a proximal based with a flexible geometry, following the strategy pioneered by Doug Barrick [20,82] and developed further by Boxer and coworkers [16,88]. The data for R and T state human hemoglobin A subunits were taken from Unzai et al. [89] as summarized in [2]. N.D., not determined

0 032 N.D. N.D. 450 2 5 290 170 11

12

J.S. Olson and Abhik Ghosh

9. REGULATION OF O2 AFFINITY AND LIGAND DISCRIMINATION IN SOYBEAN Lba Comparisons of the distal pocket structures and proximal geometries of sperm whale MbO2 and soybean LbaO2 are shown in Fig. 4, and serve as an example of the use of Equation 1 for interpreting the observed differences between the O2 and CO affinities and the effects of amino acid replacements for these two globins. Rate and equilibrium constants for O2 and CO binding to Lba are also given in Table 1. Myoglobin is the classic example of an O2 storage protein with an affinity for O2 in between that of hemoglobin and cytochrome oxidase, allowing it to store oxygen during blood flow and release it during muscle contraction [5,75–78]. Leghemoglobin evolved both to reduce the O2 tension around nitrogen fixing bacteria in the root nodules and to facilitate O2 diffusion by rapid release of the ligand to allow bacterial respiration [79,80]. As described above, bound O2 is stabilized in myoglobin by donation of a strong hydrogen bond from NH of His(E7). Replacement of His(E7) with Phe results in an 1000-fold increase in the rate constant for O2 dissociation and a shift of the C O peak from ∼1945 to 1964 cm−1 [10,81]. This strong favorable electrostatic interaction compensates for the unfavorable eclipsed geometry of the proximal Fe His(F8) bond, giving sperm whale Mb a moderate O2 affinity KO2 ≈ 1 M−1 . Replacement of His(F8) with Gly and the addition of free imidazole cause a fivefold increase in KO2 due to greater rotational freedom of the proximal base [21,82], which supports the view that the eclipsed Fe His(F8) geometry in the wild-type protein reduces iron reactivity. In soybean Lba, the Tyr(B10) side chain pulls the His(E7) side chain away from bound ligands, weakening electrostatic interactions, and, as a result, the His(E7)Phe mutation only produces a sevenfold increase in kO2 [21,86]. Replacing Tyr(B10) with Phe allows a stronger interaction with His(E7) and reduces kO2 from ∼6 to 0.8 s−1 . The most remarkable Lba mutation is His(E7) to Arg, which increases O2 affinity from 23 to 1100 M−1 (P50 ≈1 nM) because the guanidinium group can hydrogen bond with both TyrB10 and bound O2 [73]. In addition, Lba has a staggered Fe His(F8) geometry, which accounts for its high O2 affinity [90]. When His(F8) is replaced with Gly and free imidazole, KO2 for Lba decreases almost fourfold, an effect opposite to that seen for sperm whale myoglobin, which has an eclipsed proximal geometry [73].

10. COMPARISONS BETWEEN ASCARIS SUUM HEMOGLOBIN DOMAIN 1 (AscHb) AND CEREBRATULUS LACTEUS MINI-HEMOGLOBIN (CerHb), TWO GLOBINS CONTAINING TyrB10 As shown in Fig. 4 and Table 1, two strong hydrogen bonds to bound oxygen are observed in domain 1 of Ascaris suum hemoglobin (AscHb), which has a remarkably low O2 dissociation rate constant (0.004 s−1 ), shows a very high affinity for O2 (KO2 = 370 M−1 , P50 ≈ 3 nM), and binds O2 30 times more avidly than CO (Table 1). All of these characteristics indicate that this hemoglobin functions to scavenge O2 to protect the parasitic intestinal worm, which is an obligate anaerobic in its adult stage. Mutation of Tyr(B10) to Phe in AscHb results in a 200-fold increase in kO2 , and a similar increase

Mammalian Myoglobin

13

is observed when Gln(E7) is mutated to Leu [87,91,92]. Perhaps, the most remarkable characteristic of AscHb is its low M value, KCO / KO2 ≈ 0 03, which suggests significant evolutionary pressure to discriminate in favor of O2 binding and against CO binding. This strong discrimination may be a protective mechanism for preventing inactivation of the O2 scavenging function by CO, which is being generated by heme degradation in the gut. Neuronal Cerebratulus lacteus Hb (CerHb) has an active site that looks very similar to that for AscHb (Fig. 3) and suggests high O2 affinity. However, this globin shows remarkably high rates of O2 association and dissociation and a moderate affinity, which is identical to that of mammalian myoglobins (Table 1, [84]). This moderate affinity correlates with the role of CerHb in storing and rapidly releasing O2 in nerve tissue during periods of anoxia [93]. The distal pocket of CerHb contains Tyr(B10) and Gln(E7), suggesting that multiple hydrogen bonds to the bound O2 are possible (Fig. 3). However, CerHb also contains an unusual polar Thr at the E11 position, which is normally occupied by an apolar amino acid. Pesce et al. [84,94] have shown that the Thr(E11) hydroxyl O atom pulls the Tyr(B10) hydroxyl proton away from the bound ligand, causing the nonbonded electrons of the Tyr(B10) hydroxyl O atom to point toward the bound ligand. This unfavorable negative field adjacent to the bound ligand accounts for the high frequency of the C−O peak of CerHbCO (C−O = 1979 cm−1 [84]). Mutating Thr(E11) to Val disrupts the interaction with Tyr(B10), which allows the Tyr OH to form a strong hydrogen bond with bound ligands, decreases kO2 1000-fold, from 180 to 0.2 s−1 , and decreases C−O from 1979 to 1930 cm−1 (Table 1, [84]). In contrast, the proximal geometry in CerHb is very favorable for high iron reactivity. Thus, the negative field of the TyrB10 hydroxyl O atom and its sterically unfavorable proximity to the bound ligand negate the favorable proximal geometry, resulting in a moderate Mb-like O2 affinity. These examples from plant and animal globins demonstrate that Equation 1 and the scheme in Fig. 2 allow interpretation of remarkably different kinetic and equilibrium ligand binding properties. The differences between the R and T state forms of human hemoglobin have been thoroughly studied (Table 1). Again, the changes in affinity can be considered in terms of the four-step mechanism [2]. Most workers in the field agree that the major cause of the change in ligand affinity in Hb in going from the R to the T state is due to increases in strain and unfavorable geometry of the Fe-proximal HisF8 bond [19,95,96]. These proximal changes decrease the affinity of all ligands by ∼300–1000-fold, with little change in ligand discrimination (i.e., M values in Table 1). Even more impressive mutagenesis tests of the utility of Equation 1 have been carried out by Hargrove and coworkers on hexacoordinate Hbs from cyanobacteria, plants,  and  animals (for a review, see [80]). In these cases, the inhibitory 1/ 1 + KH2 O H2 O term is replaced by a 1/(1+KH ) term where KH is the isomerization constant that defines the equilibrium ratio of hexacoordinate to pentacoordinate conformations, usually involving HisE7 coordination to the iron atom. A similar term can be used to interpret CO binding to CooA, and in this case, the endogenous ligand is a Pro N atom [45]. The remaining chapters in this book develop these ideas for interpreting ligand binding to a wide variety of heme proteins involved in gas sensing, transport, storage, and consumption.

14

J.S. Olson and Abhik Ghosh

11. NO BINDING TO FERRIC AND FERROUS HEME PROTEINS Nitric oxide is unique among the diatomic gases because it can bind to both the ferric and ferrous forms of heme proteins [6,97–100]. In general, binding to Fe(II) is much stronger, but its reaction with Fe(III) can have physiological relevance as described in some of the succeeding Chapters. There are large differences between the kinetic constants for NO binding to the ferric and ferrous forms. In the case of sperm whale Mb, kNOFeIII = 12 s−1 at 20 C, pH 7 for metMbNO, whereas kNOFeII = 0 00010 s−1 for MbNO [2,101], reflecting an 100,000-fold difference in stability of the Fe(II) NO versus Fe(III) NO bond. The bimolecular rates of association are also quite different:   kNOFeIII = 0 080 M−1 s−1 , whereas kNOFeII = 22 M−1 s−1 [2,6,101]. The latter difference is primarily a reflection of direct water coordination to the iron atom in the ferric form, which makes the 1/ 1 + KH20 [H2 O]) term ∼1/1000 for NO binding to metMb. This  interpretation is supported by the ∼300–1000-fold increase in kNOFeIII when HisE7 is  changed to an apolar amino acid or glutamine [2,6,97,101]. In contrast, kNOFeII only increases ∼10-fold as a result of the same mutations. In both the ferric and ferrous state, replacing the distal histidine with an apolar amino acid increases the NO dissociation rate constant by roughly a factor of 10: from 0.0001 to 0.001 s−1 in the case of the ferrous state; and from 12 to ∼150 s−1 in the ferric state. The net result is little change in NO affinity for ferrous Mb, when the distal histidine is replaced with apolar amino acids, whereas in the ferric state, NO affinity increases over 100-fold due to the loss of inhibitory coordinated water [2,101]. Another interesting feature of the metMbNO state is that the bound ligand is stabilized by the N-H tautomer of HisE7 where the nonbonded electron pair on N stabilizes the positive charge on the Fe(III) N O+ complex [102].

12. SUMMARY In general, O2 , CO, or NO binding to a heme protein involves four steps: (i) displacement of endogenously bound ligands to create an open or “empty” distal pocket; (ii) ligand movement into the protein to form an intermediate in which the gas is trapped in the active site near the heme group; (iii) bond formation with the iron atom; and (iv) stabilization of the bound ligand by electrostatic interactions. The first step provides a simple negative mechanism for regulating the affinity of all three ligands and is observed in many globins, enzymes, and heme protein gas sensors. In some cases, the deoxygenated protein is hexacoordinate (i.e., nonsymbiotic plant Hbs, neuroglobin, CooA, etc.), with the sixth ligand being provided by the protein to reduce ligand affinity into a physiological relevant regime. The second step involves the formation of a transition state in which the ligand is very weakly and noncovalently bound in protein cavities near or in the active site. Stabilization of this noncovalently bound ligand by increasing the size of the internal cavities can increase rates of association and dissociation but, as expected for a transitionlike state, has little effect on overall ligand affinity. Unlike the binding of substrates with more complex shapes and charge distributions, the first two steps in ligand binding to heme proteins do not allow discrimination between O2 , CO, and NO because the physical sizes and polarities of these gas molecules are virtually indistinguishable at physiological temperatures. Discrimination only occurs in the third and fourth steps, bond formation

Mammalian Myoglobin

15

and electrostatic stabilization. The intrinsic chemical differences between the ligands dominate the differences in affinities of the gaseous ligands for heme iron, with the bond strength varying in the order Fe NO>>Fe CO>>Fe O2 . Protein structure can markedly alter the absolute equilibrium constants for internal bond formation by altering steric accessibility on the distal side of the heme group and the ease of in-plane iron movement and extent of backbonding through changes in proximal coordination geometry. Discrimination in favor of O2 binding is achieved primarily in the last step of ligand binding by preferential stabilization of the more polar Fe + − O2 − complex through placement of hydrogen bond donors or a positive electrostatic field vector pointing toward the bound O atoms. A negative electrostatic field vector pointing toward the ligand binding site has the opposite effect, preferential discrimination against O2 binding and only small effects on NO and CO binding.

ACKNOWLEDGMENTS The original research in the chapter was supported by United State Public Health Service grants GM 35649 (JSO) and HL 47020 (JSO), grant C-612 from the Robert A. Welch Foundation (JSO), and grants from the Research Council of Norway (AG). We could like to thank Dr. Jayashree Soman for preparing the molecular graphics in Figures 2 and 4.

REFERENCES [1] Springer, B.A., Sligar, S.G., Olson, J.S. and Phillips, G.N., Jr. (1994) Chem. Rev. 94, 699–714. [2] Olson, J.S. and Phillips, G.N., Jr. (1997) J. Biol. Inorg. Chem. 2, 544–552. [3] Suzuki, T. and Imai, K. (1998) Cell Mol. Life Sci. 54, 979–1004. [4] Brunori M. and Gibson, Q.H. (2001) EMBO Rep. 2, 674–679. [5] Dou, Y. Maillett, D.H., Eich, R.F. and Olson, J.S. (2002) Biophys. Chem. 98, 127–148. [6] Eich, R.F., Li, T., Lemon, D.D. et al. (1996) Biochemistry 35, 6976–6983. [7] Gardner, P.R. (2005) J. Inorg. Biochem. 99, 247–266. [8] Olson, J.S., Foley, E.W., Rogge, C. et al. (2004), Free Radic. Biol. Med. 36, 685–697. [9] Spiro, T.G. and Kozlowski, P.M. (2001) Acc. Chem. Res. 34, 137–144. [10] Phillips, G.N., Jr., Teodoro, M., Li, T. et al. (1999) J. Phys. Chem. B. 103, 8817–8829. [11] Franzen, S. (2002) J. Am. Chem. Soc. 124, 13271–13281. [12] Park, E.S., Andrews, S.S., Hu, R.B. and Boxer, S.G. (1999) J. Phys. Chem. B 103, 9813–9817. [13] Park, E.S. and Boxer, S.G. (2002) J. Phys. Chem. B 106, 5800–5806. [14] Ibrahim, M., Xu, C. and Spiro, T.G. (2006) J. Am. Chem. Soc. 128, 16834–16845. [15] De Angelis, F., Jarzecki, A.A., Car, R. and Spiro, T.G. (2005) J. Phys. Chem. B 109, 3065–3070. [16] Decatur, S.M., DePillis, G.D. and Boxer, S.G. (1996) Biochemistry 35, 3925–3932. [17] Barrick, D. (2000) Proteins 39, 291–308. [18] Barrick, D. and Dahlquist, F.W. (2000) Proteins 39, 278–290. [19] Barrick, D., Ho, N.T., Simplaceanu, V. and Ho, C. (2001) Biochemistry 40, 3780–3795. [20] Barrick, D., Ho, N.T., Simplaceanu, V. et al. (1997) Nat. Struct. Biol. 4, 78–83. [21] Kundu, S., Snyder, B., Das, K. et al. (2002) Proteins 46, 268–277.

16

J.S. Olson and Abhik Ghosh

[22] Ghosh, A. (2005) Acc. Chem. Res. 38, 943–954. [23] Quillin, M.L., Li, T., Olson, J.S. et al. (1995) J. Mol. Biol. 245, 416–436. [24] Quillin, M.L., Arduini, R.M., Olson, J.S. and Phillips, G.N., Jr., (1993) J. Mol. Biol. 234, 140–155. [25] Stec, B. and Phillips, G.N., Jr., (2001) Acta Cryst. D 57, 751–754. [26] Lim, M., Jackson, T.A. and Anfinrud, P.A. Science (1995) 269, 962–966. [27] Ivanov, D., Sage, J.T., Keim, M. et al. (1994) J. Am. Chem. Soc. 116, 4139–4140. [28] Ghosh, A. and Bocian, D.F. (1996) J. Phys. Chem. 100, 6363–6367. [29] Vangberg, T., Bocian, D.F. and Ghosh, A. (1997) J. Biol. Inorg. Chem. 2, 526–530. [30] Spiro, T.G. and Kozlowski, P.M. (1998) J. Am. Chem. Soc. 120, 4524–4525. [31] Papai, I., Stirling, A., Mink, J. and Nakamoto, K. (1998) Chem. Phys. Lett. 287, 531–534. [32] Wyllie, G.R.A. and Scheidt, W.R. (2002) Chem. Rev. 102, 1067–1089. [33] Roberts, S.A., Weichsel, A., Qiu, Y. et al. (2001) Biochemistry 40, 11327–11337. [34] Xu, N., Powell, D.R., Cheng, L. and Richter-Addo, G.B. (2006) Chem. Comm. 19, 2030–2032. [35] Richter-Addo, G.B., Wheeler, R.A., Hixson, C.A. et al. (2001) J. Am. Chem. Soc. 123, 6314–6326. [36] Sigfridsson, E. and Ryde, U. (2002) J. Inorg. Biochem. 91, 101–115. [37] Sigfridsson, E. and Ryde, U. (1999) J. Biol. Inorg. Chem. 4, 99–110. [38] Tangen, E., Svadberg, A. and Ghosh, A. (2005) Inorg. Chem. 44, 7802–7805. [39] Hargrove, M.S., Brucker, E.A., Stec, B. et al. (2000) Structure 8, 1005–1014. [40] Trent, J.T., III, Hvitved, A.N. and Hargrove, M.S. (2001) Biochemistry 40, 6155–6163. [41] Trent, J.T., III, Watts, R.A. and Hargrove, M.S. (2001) J. Biol. Chem. 276, 30106–30110. [42] Hankeln, T., Ebner, B., Fuchs, C. et al. (2005) J. Inorg. Biochem. 99, 110–119. [43] Pesce, A., Dewilde, S., Nardini, M. et al. (2003) Structure 11, 1087–1095. [44] Vallone, B., Nienhaus, K., Matthes, A. et al. (2004) Proc. Natl. Acad. Sci. U.S.A. 101, 17351–17356. [45] Puranik, M., Nielsen, S.B., Youn, H. et al. (2004) J. Biol. Chem. 279, 21096–21108. [46] Ibrahim, M., Kerby, R.L., Puranik, M. et al. (2006) J. Biol. Chem. 281, 29165–29173. [47] Pinkert, J.C., Clark, R.W. and Burstyn, J.N. (2006) J. Biol. Inorg. Chem. 11, 642–650. [48] Schlichting, I., Berendzen, J., Phillips, G.N., Jr. and Sweet, R.M. (1994) Nature 371, 808–812. [49] Schlichting, I. and Chu, K. (2000) Curr. Opin. Struct. Biol. 10, 744–752. [50] Srajer, V., Ren, Z., Teng, T.Y. et al. (2001) Biochemistry 40, 13802–13815. [51] Schotte, F., Soman, J., Olson, J.S. et al. (2004) J. Struct. Biol. 147, 235–246. [52] Lim, M., Jackson, T.A. and Anfinrud, P.A. (1997) Nat. Struct. Biol. 4, 209–214. [53] Nienhaus, G.U., Mourant, J.R., Chu, K. and Frauenfelder, H. (1994) Biochemistry 33, 13413–13430. [54] Nienhaus, K., Olson, J.S., Franzen, S. and Nienhaus, G.U. (2005) J. Am. Chem. Soc. 127, 40–41. [55] Olson, J.S. and Phillips, G.N., Jr. (1996) J. Biol. Chem. 271, 17593–17596. [56] Olson, J.S., Soman, J. and Phillips, G.N.J. (2007) IUBMB Life 59, 552–562. [57] Scott, E.E., Gibson, Q.H. and Olson, J.S. J. Biol. Chem. 276, 5177–5188. [58] Srajer, V., Ren, Z., Teng, T.Y. et al. (2001) Biochemistry 40, 13802–13815. [59] Teng, T.Y., Srajer, V. and Moffat, K. (1997) Biochemistry 36, 12087–12100. [60] Carver, T.E., Rohlfs, R.J., Olson, J.S. et al. (1990) J. Biol. Chem. 265, 20007–20020. [61] Schmidt, M., Nienhaus, K., Pahl, R. et al. (2005) Proc. Natl. Acad. Sci. U.S.A. 102, 11704–11709. [62] Aranda, R.T., Levin, E.J., Schotte, F. et al. (2006) Acta Cryst. D 62, 776–783. [63] Bourgeois, D., Vallone, B., Arcovito, A. et al. (2006) Proc. Natl. Acad. Sci. U.S.A. 103, 4924–4929.

Mammalian Myoglobin [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75] [76] [77] [78] [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95] [96] [97] [98] [99] [100] [101] [102]

17

Franzen, S. (2002) Proc. Natl. Acad. Sci. U.S.A. 99, 16754–16759. Ye, X., Yu, A., Georgiev, G.Y. et al. (2005) J. Am. Chem. Soc. 127, 5854–5861. Champion, P.M., Rosca, F., Ionascu, D. et al. (2004) Faraday Disc. 127, 123–135. Jensen, K.P. and Ryde, U. (2004) J. Biol. Chem. 279, 14561–14569. Strickland, N. and Harvey, J.N. (2007) J. Phys. Chem. B 111, 841–852. Ionascu, D., Gruia, F., Ye, X. et al. (2005) J. Am. Chem. Soc. 127, 16921–16934. Vojtechovsky, J., Chu, K., Berendzen, J. et al. (1999) Biophys J 77, 2153–2174. Copeland, D.M., West, A.H. and Richter-Addo, G.B. (2003) Proteins 53, 182–192. Brucker, E.A., Olson, J.S., Ikeda-Saito, M. and Phillips, G.N., Jr. (1998) Proteins 30, 352–356. Kundu, S., Blouin, G.C., Premer, S.A. et al. (2004) Biochemistry 43, 6241–6252. Das, T.K., Friedman, J.M., Kloek, A.P. et al. (2000) Biochemistry 39, 837–842. Wittenberg, J.B. and Wittenberg, B.A. (2003) J. Exp. Biol. 206, 2011–2020. Wittenberg, J.B. and Wittenberg, B.A. (1990) Annu. Rev. Biophys. Biophys. Chem. 19, 217–241. Brunori, M. (2001) Trends Biochem. Sci. 26, 21–23. Brunori, M. (2001) Trends Biochem. Sci. 26, 209–210. Wittenberg, J.B., Appleby, C.A., Bergersen, F.J. and Turner, G.L. (1975) Ann. New Tork Acad. Sci. 244, 28–34. Kundu, S., Trent, J.T., III and Hargrove, M.S. (2003) Trends Plant Sci. 8, 387–393. Li, T. Quillin, M.L., Phillips, G.N., Jr. and Olson, J.S. (1994) Biochemistry 33, 1433–1446. Barrick, D. (1994) Biochemistry 33, 6546–6554. Yang, J., Kloek, A.P., Goldberg, D.E. and Mathews, F.S. (1995) Proc. Natl. Acad. Sci. U.S.A. 92, 4224–4228. Pesce, A., Nardini, M., Ascenzi, P. et al. (2004) J. Biol. Chem. 279, 33662–33672. Rohlfs, R.J., Mathews, A.J., Carver, T.E. et al. (1990) J. Biol. Chem. 265, 3168–3176. Kundu, S. and Hargrove, M.S. (2003) Proteins 50, 239–248. De Baere, I., Perutz, M.F., Kiger, L. et al. (1994) Proc. Natl. Acad. Sci. U.S.A. 91, 1594–1597. Decatur, S.M., Franzen, S., DePillis, G.D. et al. (1996) Biochemistry 35, 4939–4944. Unzai, S., Eich, R., Shibayama, N. et al. (1998) J. Biol. Chem. 273, 23150–23159. Hargrove, M.S., Barry, J.K., Brucker, E.A. et al. (1997) J. Mol. Biol. 266, 1032–1042. Kloek, A.P., Yang, J., Mathews, F.S. et al. (1994) J. Biol. Chem. 269, 2377–2379. Goldberg, D.E. (1999) Chem. Rev. 99, 3371–3378. Vandergon, T.L., Riggs, C.K.,Gorr, T.A. et al. (1998) J. Biol. Chem. 273, 16998–17011. Geuens, E., Dewilde, S., Hoogewijs, D. et al. (2004) IUBMB Life 56, 653–656. Perutz, M.F. (1990) Annu. Rev. Physiol. 52, 1–25. Barrick, D., Lukin, J.A., Simplaceanu, V. and Ho, C. (2004) Methods Enzymol. 379, 28–54. Sharma, V.S., Isaacson, R.A., John, M.E. et al. (1983) Biochemistry 22, 3897–3902. Laverman, L.E. and Ford, P.C. (2001) J. Am. Chem. Soc. 123, 11614–11622. Laverman, L.E., Wanat, A., Oszajca, J. et al. (2001) J. Am. Chem. Soc. 123, 285–293. Ford, P.C. and Lorkovic, I.M. (2002) Chem. Rev. 102, 993–1018. Foley, E.W., Ph.D. (2005) Dissertation, Houston, TX: Rice University. Miller, L.M., Pedraza, A.J. and Chance, M.R. (1997) Biochemistry 36, 12199–12207.

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 2

A Surfeit of Biological Heme-based Sensors Marie-Alda Gilles-Gonzalez and Gonzalo Gonzalez Department of Biochemistry, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75390-9038, USA

Abstract In a great variety of organisms throughout all kingdoms of life, the hemebased sensor proteins are the key regulators of adaptive responses to fluctuating oxygen, carbon monoxide, and nitric oxide levels. These signal transducers achieve their responses by coupling a regulatory heme-binding domain to a neighboring transmitter. The past decade has witnessed an explosion in the numbers of these modular sensory proteins known, from just two recognized members, FixL and soluble guanylyl cyclase (sGC), to six broad families comprising more than 100 sensors. Heme-based sensors so far feature six different types of heme-binding modules: the heme-binding PAS domain, globin-coupled sensor (GCS), CooA, heme-NO-binding (HNOB) domain, heme-binding GAF domain, and heme-associated ligand-binding domains (LBD) of the nuclear-receptor class. The transmitters for coupling to such heme-binding domains include protein-histidine kinases, serine-threonine kinases, nucleotide cyclases, cyclic-nucleotide phosphodiesterases, chemotaxis methyl-carrier protein receptors, and transcription factors of the basic helix-loophelix, helix-turn-helix, and zinc-finger classes. Some well-studied sensors are the FixL, EcDos, AxPDEA1, NPAS2, HemAT, CooA, sGC, Tar4, DevS, DosT and E75 proteins. This chapter elaborates the defining characteristics of heme-based sensors, examines the advances on those proteins, and discusses the regulatory hypotheses postulated for those sensors. A general, “helix-swap,” model is also proposed for signal transduction by PAS domains. Keywords: Cyclic di-GMP; EAL domain; FixJ; Guanylyl cyclase; GGDEF domain; Hemoglobin; H-NOX domain; Myoglobin; Oxygen sensor; Response regulator; Sensor kinase; SONO domain; Two-component regulatory system; Mycobacterium tuberculosis DevR

1. INTRODUCTION Since the advent of genomic analyses and recombinant-DNA techniques, a greater variety of ligand-binding heme proteins have been discovered than all previous years put together. This has expanded our knowledge of the strategies with which these proteins achieve their architectures and control their properties. This has also made us realize that much of what we thought we could conclude about ligand-binding heme proteins may have been the characteristics of individual classes of proteins. For example, numerous

A Surfeit of Biological Heme-based Sensors

PAS Family

19

GCS Family

Histidine Protein Kinase 5 Rhizobial FixLs (bacterial) Nitrogen fixation and alternative oxidases M. thermoautotrophicum FixL (archaeal) C. crescentus FixL (bacterial) Plus to over 10 bacterial homologs Second Messenger A. xylinum PDEA1 (bacterial) Cellulose production E. coli, A. caulinodans (2 bacterial) Plus to over 12 bacterial homologs

DNA Binding Mammalian NPAS2 (eukaryotic) Circadian rhythm

Methyl Carrier Protein H. salinarum (archaeal) B. halodurans (bacterial) C. crescentus (bacterial) and B. subtilis HemATs (bacterial) Aerotaxis Plus to over 13 bacterial homologs

Second Messenger A. ferrooxidans (bacterial) Plus to over 15 bacterial homologs

Serine-Threonine Kinase Mammalian HRI (eukaryotic) Globin-subunit translation

HNOB Family Second Messenger, Mononucleotide Mammalian sGC (eukaryotic) Vasodilation, etc. C. elegans GCY-35 (eukaryotic) Aerotaxis D. Melanogaster Gyc88E/89Da,b (eukaryotic)

Methyl Carrier Protein T. tengcongensis (H-NOX/SONO) (bacterial) C. botulinum Cb-SONO (bacterial) Ddes2822Dde and CAC3243 (bacterial)

CooA Family DNA Binding R. rubrum (bacterial) C. hydrogenoformans (bacterial) C. vinelandii (bacterial) D. hafniense (bacterial) D. vulgaris (bacterial) CO metabolism Plus 4 bacterial homologs

Histidine Protein Kinase Over 5 bacterial homologs

LBD Family GAF Family

DNA Binding

Histidine Protein Kinase M. Tuberculosis DevS and DosT (bacterial) Latency signaling Plus 5 mycobacterial homologs

D. melanogaster E75 (eukaryotic) Moulting of larvae Plus 11 eukaryotic homologs

Fig. 1. Families of heme-based sensors. A distinctive heme-binding domain defines each family of sensors. Subgroups within the families couple their heme-binding domain to different transmitters for signal transduction. Those proteins specifically named are ones that have been purified and established as heme proteins. The physiological functions, if known, are highlighted in green. The last line in each category notes the kingdom membership and approximate numbers of additional members expected from sequence homology.

20

M.A. Gilles-Gonzalez and G. Gonzalez

heme-protein enzymes are actually signal transducers where the heme center is directly concerned with regulation rather than catalysis [1–14]. The traditional relations of the affinities of hemoglobins for ligands, i.e., O 2  CO ≪ NO, do not extend to other heme proteins designed for ligand binding, or even O2 binding [7,15,16]. Scores of proteins with hexacoordinate heme iron function in reversible binding of ligands rather than transfer of electrons [7,17–27]. For at least one class of O2 -binding heme proteins, the iron-histidine bond stretches determined by resonance Raman spectroscopy do not correlate with O2 affinity [28]. The protein scaffold for a heme may consist entirely of -strands, a mixture of -helices and -strands, or -helices differing in number and arrangement from those in myoglobins [14,29–33]. Study of a greater assortment of these ligand-binding heme proteins should reveal some unifiying principles about their behaviors. Although currently it is again difficult to rationalize those behaviors, it is also quite exciting to consider them from first principles. Many of the changes in thinking about heme-based sensors began with their recognition as a distinct functional class and the expectation that their diverse requirements for sensing would lead to a broad range of characteristics [1]. Discovery of heme-based sensors has rapidly accelerated (Fig. 1) [1–3,6,7,11,17–19,34–36].

2. WHAT CONSTITUTES A BIOLOGICAL HEME-BASED SENSOR? In a biological heme-based sensor, a regulatory heme-binding domain or subunit controls a neighboring transmitter region of the same protein [1,37–39]. Such signaltransducing heme proteins govern adaptative responses to fluctuations in O2 , CO, or NO: all three being diatomic gases that are now appreciated as physiological messengers. As a class of heme proteins, the sensors are distinct from the carriers of gases and the catalysts of oxygen-atom and electron transfer reactions [1]. The transmitter regions of heme-based sensors typically feature modules that also transduce signals in many nonheme proteins; such modules include protein-histidine kinase, serine-threonine kinase, cyclic-dinucleotide phosphodiesterase, nucleotide cyclase, chemotaxis receptor, and DNA-binding transcription-factor activities [1–12,17–19,34–36,40,41]. The regulatory domains, on the other hand, have several architectural and sequence motifs that are entirely novel for heme binding, such as PAS (Per-Arnt-Sim), cAMP-receptor-like modified globin, GAF (cGMP-regulated cyclic nucleotide phosphodiesterases, adenylate cyclases, and bacterial transcriptional regulator F hlA), and LBD (nuclear-receptor class ligand-binding domains) [14,29,31,33,42]. Heme-based sensors provide excellent models for study of signal transduction, with the iron center supplying a built-in probe of the sensor’s status and the transmitter reporting the switching potential of any state of the heme. An effective sensor must: • Bind its signal ligand over the concentration range appropriate for “switching” an activity; • Switch an activity reversibly on binding of its signal ligand, i.e., trigger a change from an active to an inactive state, or vice versa (Fig. 2); • Discriminate against false signals if this is physiologically necessary (Figs. 2B and C); • Switch its conformation on binding of its signal ligand.

A Surfeit of Biological Heme-based Sensors

21

Switching by true signal:

Activity (e.g. kinase)

Activity (e.g. kinase)

Response to alternative signal:

(A) No discrimination. False signal can bind and switch off.

(B) Discrimination at switching step. False signal binds but does not switch off; it also prevents true ligand from binding.

(C) Discrimination at binding step. False signal does not bind. Sensor stays in unliganded, on state.

Fig. 2. Switching and discrimination by heme-based sensors. This cartoon illustrates the possible responses to true signal (closed circles) and to alternative ligands (open circles) by a hypothetical sensor that is normally active in the unliganded state. The top of the figure shows switching by the true signal. The three lower panels show the likely responses to alternative ligands: absence of discrimination (A), binding without switching (B), exclusion of the ligand (C). To apply the same figure to a hypothetical sensor that is normally inactive in the unliganded state, reverse all the switches. (see Plate 3.)

2.1. Signal Binding in an Appropriate Range and Discrimination at the Binding Step Heme-based sensors are biochemical tools for physiological adaptation. As such, they recognize their signal based principally on the organism and environment in which they have evolved to function. For example, the RmFixL O2 sensor (Kd ∼ 50 M) triggers the expression of Sinorhizobium meliloti nif and fix genes as the Rhizobia encounter hypoxic zones (50–100 M O2 ) of symbiotic root nodules [1,43].

22

M.A. Gilles-Gonzalez and G. Gonzalez

In addition to detecting their true signal in the correct range, some sensors have adapted to exclude other signals. That is, they discriminate at the ligand-binding stage, if this is required (Fig. 2C). For example, a sensor of one ligand in a high background of several others may need to exclude “false” ligands as well as detect its true signal. The mammalian soluble guanylyl cyclase (sGC) is a case in point. This sensor has the difficult task of detecting miniscule levels of NO in the presence of much higher levels of O2 . Soluble guanylyl cyclase has adapted, in ways that are still not entirely understood, to bind NO and CO but not O2 [15,44–46]. This adaptation is essential because even if O2 cannot switch sGC, this protein must reject O2 binding simply to avoid becoming saturated with O2 . In contrast to the mammalian sGC, an NO sensor designed to work anaerobically would not need to exclude O2 , since discriminating against an absent signal offers no adaptive advantage. It is therefore possible that some relatives of sGC in anaerobic organisms are quite acceptable NO sensors in vivo even if they do not reject O2 binding in vitro [41]. Similarly, an O2 sensor in an organism that normally encounters only a miniscule background of CO or NO, compared to O2 , should not need to feature any adaptation for preventing CO or NO saturation. Hence, the ligand-binding properties of sensors in vitro cannot, by themselves, be a basis for establishing their true signals.

2.2. Switching an Activity on Binding of Signal Since the job of a sensor is to switch an activity when it encounters its signal (Fig. 2), the most convincing evidence for a ligand as a true signal should be the response that it causes a sensor to initiate. For example, the transcription factor CooA is clearly a CO sensor because it responds to CO in a physiologically relevant range and induces expression of the genes required for Rhodospirullum rubrum to consume CO as a carbon source (Fig. 3) [17,48,49]. It is important to appreciate that an observation of switching does not by itself guarantee that one has found the true signal for a sensor. For example, even though oxidation of the heme iron can inactivate many sensors, this finding cannot be taken to imply that these proteins are sensors of redox potential, unless there exists

CO CooA active

+ cooM K L X U H

CooA inactive

+ cooF S C T J

cooA

Fig. 3. CooA-regulated gene-expression cascade [47]. When deprived of light and oxygen, the photosynthetic bacterium R. rubrum can grow on CO as its sole energy source by inducing the expression of the cooMKLXUH and cooFSCTJ operons. These genes encode a multiprotein complex for oxidizing CO to CO2 . The CooS, CooF, and CooH proteins are known to be a CO dehydrogenase, a CooS-associated iron–sulfur protein, and a CO-tolerant hydrogenase, respectively.

A Surfeit of Biological Heme-based Sensors

23

evidence that they can be oxidized in-vivo. On the whole, although determinations of the ligands that can or cannot cause switching are essential for understanding a sensor’s chemistry, switching of a sensor can only mean that a specific heme ligand is the true signal if the natural environment of the sensor contains this ligand, and if the sensor triggers an appropriate response at a ligand concentration that the organism would reasonably encounter.

2.3. Discrimination Against False Alarms at the Switching Step To communicate binding of a true signal, a sensor may need to discriminate against false alarms. Section 2.1 has already treated how a heme-based sensor could reject a false signal by binding it poorly or not at all (Fig. 2C). An alternative way to discriminate is by designing a sensing mechanism that cannot cause effective switching if a false signal binds (Fig. 2B). Suppose we define the “inhibition factor” for an inhibitory ligand as the ratio of the activities of the unliganded and the fully liganded forms of a sensor [4]. The inhibition factors for different ligands then provide a measure of any discrimination at the switching step [4]. For example, FixL proteins clearly discriminate against NO regulation at the switching step (Fig. 2B). FixLs bind NO much better than O2 [1], but binding of NO causes only a twofold inhibition of the FixL kinase activity toward the FixJ protein, whereas binding of O2 shuts down this activity with an inhibition factor of over 100-fold [4]. When measuring inhibition or activation factors, unless an experimenter carefully verifies that sufficient ligand is added to saturate the heme, a mutation that lowers affinity for the ligand may easily be mistaken for one that decouples the heme from its transmitter. That is, a change in the concentration of the liganded form should not be confused with a change in its activity.

2.4. Switching Conformation on Binding of Signal Whenever a ligand switches an activity, a conformational change has surely occurred, even if such a change is not observable with the conditions and methods used. The converse of this statement, however, is untrue. An observable change in protein state, by spectroscopy or any other means, does not prove that switching of an activity has occurred. The only real measure of the regulatory potency of a ligand is the activity of a sensor with and without this ligand. So far, for spectroscopic studies requiring high concentrations of proteins, it has been easiest to prepare the isolated heme-binding domains of sensors, and in particular, their ferric unliganded forms (i.e., the FeIII or met species). Consequently, multiple high-resolution structures are available for hemebinding domains, but only one structure has been solved for a full-length sensor, and no structures of fully “on” and “off” conformations are available to compare for any complete sensor [4,5,14,29,31,33,50–55]. To prepare on- and off-states of sensors, which typically correspond to the unliganded and liganded ferrous forms or vice versa, crystals are usually grown for the more easily obtainable met forms and subsequently exposed to reducing agents and to ligands. Some caveats about these procedures are that conformational switching: (i) may not manifest fully in the truncated proteins; (ii) may not be possible for some crystal forms; or (iii) may, in some crystals, require conditions

24

M.A. Gilles-Gonzalez and G. Gonzalez

different from the ones determined in solution. For example, compared to solutions of the Bradyrhizobium japonicum FixL heme-binding domain (BjFixLH), some crystals of BjFixLH have much lower affinities for ligands [51]. The EcDos heme-binding domain (EcDosH) dimerizes in solution as well as in crystals, shows no sign of negative cooperativity between the two hemes in solution, and has a reasonably high affinity for O2 (Kd ∼ 10 M) [7]. Nevertheless, in the crystalline form, only one monomer out of each EcDosH dimer was discovered to bind O2 in air [54].

3. FAMILIES OF HEME-BASED SENSORS Given the great biological importance of O2 and the fact that most organisms rapidly adapt to its fluctuations, the most common application of heme-based sensors is expected to be O2 sensing. The currently known sensors group into six families, based on their heme-binding domains: heme-PAS, CooA, GCS (globin-coupled sensor), HNOB (hemeNO binding), GAF, and LBD [14,37–42,56] (Fig. 1).

3.1. The Heme-PAS Family Heme-PAS domain–containing sensors occur in all kingdoms of life, are extraordinarily versatile, and are relatively well studied. Members are already known that respond specifically to O2 or CO and whose heme-binding domains couple to three dissimilar but widespread activities [1,3,6,7,18,28,37–39]. These sensors demonstrably transduce signals and mediate adaptive responses by diverse strategies, including chemical modification of proteins, control of second-messenger levels, and regulation of macromolecular interactions (Fig. 4) [3,4,6,9,18].

3.1.1. The FixL Proteins 3.1.1.1. FixL-like Proteins Regulate Microaerobic Adaptation by Rhizobia and Many Other Bacteria For nearly 10 years, the FixLs were the only proteins known to feature a heme-PAS domain [37,43,59–69,]. In Rhizobia, where the fixL genes were first discovered, they have three distinct, but not mutually exclusive functions: • Enabling the symbiotic Rhizobia to survive O2 starvation; • Restricting the expression of the rhizobial nitrogen fixation (nif, fix) genes to hypoxic conditions; • Limiting the expression of the bacterial denitrification genes to hypoxic conditions. Thus, a drop in O2 tension leads a specific rhizobial species to induce one or more prominent classes of genes: those encoding the alternative high O2 -affinity terminal oxidases required for microaerobic respiration, those directly encoding subunits of the nitrogenase enzymes, or those encoding the denitrification enzymes [43,59–68]. RmfixL regulates the first two sets of genes in S. meliloti [59–61]. BjfixL most directly controls

A Surfeit of Biological Heme-based Sensors

RmFixL

Histidine Kinases

BjFixL

25

PAS

PAS

PAC

MtDos

PAC

PAS

PAS

EcDos

PAS

PDEA1

PAS

PAC

HATPase

HisKA

PAC

PAS

PAC

PAS

HisKA

HATPase

PAC

PAS

PAC

HATPase

DUF2

DUF1

PAC

Phosphodiesterases

bHLH DNA Binding

NPAS2

HLH

PAC

GAF

DUF1

A

B

PAS

PAS

DUF2

PAC

Fig. 4. Domain organization of known heme-binding PAS proteins [37,38]. PAS domains are depicted as purple rectangles, and the heme-PAS domains are highlighted with red circles. Note that each protein-histidine kinase or phosphodiesterase contains only one heme-PAS domain. The enzymatic subdomains of the protein-histidine kinases (HisKA/HATPase) are shown in green and those of the cyclic dinucleotide phosphodiesterases (DUF1 and DUF2, corresponding to GGDEF and EAL, respectively) in brown. In NPAS2, an N-terminal basic-helix-loop-helix DNAbinding region (HLH) is followed by two heme-binding PAS domains, PAS-A and PAS-B, and a C-terminal region of unknown function. Domain nomenclatures, symbols, and protein organizations are according to the simple modular architecture research tool (SMART) from the European Molecular Biology Laboratory [57,58]. (see Plate 4.)

the expression of the alternative oxidases and denitrification enzymes in B. japonicum [62–65]. The regulation of microaerobic respiration by FixLs is probably more universal than their control of nitrogen fixation. Comparisons of protein sequences suggest that many other FixL-like proteins exist, not only in Rhizobia, but also in non-nitrogenfixing organisms such as the bacteria Caulobacter crescentus and Rhodopseudomonas palustris, and the archaea Halobacterium salinarum and Achaeoglobulus fulgidus. These sensors are largely uncharacterized. Homologs of FixJ likewise exist in these organisms and may be regarded as consisting of two types [69]. Some, called FixJs, are true substrates of FixL and often coexpressed with FixLs [59,60,62,66]. Others, called FixTs, are inhibitors of FixL and usually independently expressed [69,70]. The inhibition by FixT is entirely independent of the FixL heme status or the O2 level [70].

26

M.A. Gilles-Gonzalez and G. Gonzalez

3.1.1.2. The Enzymatic Reactions of FixL: All Phosphoryl-transfer Steps Occur in a FixLJ Complex All of the FixLs so far examined are heme-regulated protein-histidine kinases [3–5, 70–79]. In RmFixL or BjFixL, saturation of the heme-PAS domain with O2 (Kd = 50–140 M) inactivates the kinase [4,5,71,73–75]. By contrast, the deoxy state of either protein is catalytically competent to transfer a -phosphoryl group from ATP to a conserved aspartate residue of the transcription factor FixJ [4,5,70,71,73–78]. Sensor kinases can react with ATP in the absence of their second substrates, i.e., the response regulators, to form a phosphorylated intermediate [80,81]. It is common to measure the rate of this “half-reaction” and to assume that its kinetics and regulation reflect those of the formation of a similar reaction intermediate during the physiologically relevant turnover reaction. However, it is also common to observe that the phosphorylated enzyme intermediate prepared in this way is quite unstable when presented with its protein substrate and more likely to hydrolyze than to transfer to this protein. For some sensor kinases, a “phosphatase” activity is proposed to be the principal means of regulation. Whatever may be the case for other sensor kinases, it is clear that for FixL, the formation of a FixL2 :FixJ2 (or “FixLJ”) complex precedes all the phosphorylation steps and even the detection of O2 [73,74]. The rate of formation of the phospho-FixL intermediate in the presence of FixJ is 8 times faster than the rate of formation of the phosphorylated intermediate obtained by reaction with ATP alone [73,75]. Furthermore, inhibition of this so-called “autophosphorylation” by ligands is substantially altered by the presence of FixJ [74]. The generation of phospho-FixJ during turnover occurs very efficiently and without any detectable “phosphatase” activity and proceeds to completion, i.e., until 50% of the FixJ (or all of the dimeric FixJ) is phosphorylated [5,73]. By contrast, attempts to make phosphorylated response regulators by presenting them to sensor kinases “prephosphorylated” in a separate reaction with ATP now appear to be, in many cases (including that of FixL and FixJ), a quite efficient means of generating free phosphate [73,76]. (1) Complexation (2) Autophosphorylation (3) Phosphoryl transfer

FixL2 + 2FixJ + 2 ATP FixL2 :FixJ2 :2 ATP P-FixL2 :FixJ2 :2 ADP

FixL2 :FixJ2 :2 ATP P-FixL2 :FixJ2 :2 ADP FixL2 + P-FixJ 2 + 2 ADP

While “autophosphorylation” reactions can be useful in a qualitative way, any attempt to describe a quantitative response of the FixL/FixJ system to ligands must include FixJ, even though the reaction steps that directly involve FixJ appear to be independent of the heme state. To ensure that FixJ concentrations stay relatively constant for a long enough period to measure an accurate turnover number, one must ensure that there is enough substrate for FixL to “turn over” at least 10 times. Needless to say, the kinetics are sufficiently complex, especially for inhibited and partially inhibited FixL, to require a complete time course for the accurate modeling of the reaction rates [4,5,73–75]. FixL2 Turnover

2 FixJ + 2 ATP

P-FixJ2 + 2 ADP

To examine whether O2 inhibits this enzymatic reaction by lowering the affinity of FixL for one or both of its substrates, Gilles-Gonzalez and colleagues directly measured the affinities of oxy- and deoxy-FixL for ATP and FixJ by fluorescence spectroscopic methods [75]. Thus, they discovered that O2 exerts its inhibitory effect by altering the

A Surfeit of Biological Heme-based Sensors

27

reactivity of the FixL:FixJ:ATP enzyme-substrate complex, rather than by interfering with its formation. 3.1.1.2.1. Nucleotides and other reaction products do not affect the O2 saturation of FixL. The O2 affinity of a truncated RmFixL, i.e., RmFixLT , in the RmFixLT J complex has been reported to drop to less than a third of its normal value on exposure to 200 M ATP or ADP [79]. This observation was interpreted as “an enhanced reciprocating kinase reaction;” however, it has the more drastic implication that FixL would not sense O2 in physiological levels of ATP and ADP, since more than half of the FixL would be in the enzymatically active deoxy state at all physiological O2 tensions, including air saturation [52]. This is clearly not the case. Multiple observations of stringent regulation of FixJ turnover in air have been reported, and one of those reports has come from the same research group that proposed the “enhanced reciprocating kinase reaction” [4,5,70, 73–75,78]. Furthermore, if the fraction of active FixL increased rapidly during a reaction as the reaction products accumulated, the time courses for in vitro assays would be extremely strange and definitely not fit any standard curve. Assays of heme-based sensors lose all validity unless some care is taken to verify that the state of the heme (which, by definition, controls the activity of the sensor) remains unchanged over the course of any given assay. This is not for exotic considerations, such as possible reaction product– induced changes in affinity, but for protection against mundane mishaps such as oxidation of the heme iron or a broken seal in a gas-tight apparatus. Such routine observations clearly show that the heme state in FixL is quite indifferent to any reactants or products of the phosphorylation reactions [75]. Our best guess is that the observation attributed to an effect of nucleotides was due to artifactual generation of deoxy-FixL on addition of high concentrations of dithiothreitol to keep the protein reduced under sealed conditions. 3.1.1.2.2. The heme-iron oxidation state influences enzymatic turnover by some FixLs. There are many intriguing differences in the chemistries of FixL proteins, and one of the more striking of these is in the effect of heme-iron oxidation state on the kinase activity [4,5,74]. For BjFixL, oxidation of the deoxy form (FeII ) to the met form (FeIII ) did not affect the rate at which the protein turned over FixJ to phospho-FixJ (Table 1a) [4,5]. For RmFixLT , on the other hand, oxidation to the met form inhibited the FixJ turnover more than 100-fold (Table 1a) [74]. There was no significant effect of oxidation on the reaction of either protein with ATP alone (Table 1a), providing further support for the conclusion that the phospho-FixL intermediate that is produced within a FixL:FixJ complex is not chemically identical to that produced by reaction with ATP alone [4,74]. Neither the RmFixLH nor the BjFixLH heme-PAS-domain structures show any obvious conformational differences between the deoxy and met form [51,53]. So the basis for the peculiar sensitivity of RmFixLT to oxidation state is not yet understood, although this is probably not relevant in the bacteria; in whole Escherichia coli, for example, intracellular FixL exists exclusively in the reduced form (M.-A. Gilles-Gonzalez, unpublished observation).

3.1.1.3. Control of Enzymatic Turnover by Heme Ligands 3.1.1.3.1. Discrimination imposed at the regulation step. The FixL proteins readily bind ligands of ferrous and ferric heme (Table 2) [1,72,85,86]. Ferrous FixL binds O2 , CO, and NO, and the ferric protein binds CN− and imidazole [1,72,85,86]. Molecular oxygen stabilizes the inactive conformation of FixL so effectively that there remains

28

M.A. Gilles-Gonzalez and G. Gonzalez

Table 1. Influence of oxidation state and ligands on the activities of S. meliloti and B. japonicum FixLs at 23 C and pH 8.0 [4,5,74] Autophosphorylation (% h−1 )

(a)

RmFixLT BjFixL

FeII

FeIII

Ratio

33 33

33 33

10 10

Turnover (h−1 ) RmFixLT BjFixL

12 26

(b)

Oxygen Imidazole Cyanide Carbon monoxide Nitric oxide

0.12 24

100 11

Inhibition factors for turnover RmFixLT

BjFixL

R220A BjFixL

R206A BjFixL

>100 >260 2 2.7 2.3

>100 4.7 >2700 1.4 1.9

— 10 1.7 1.1 0.90

— — 140 1.1 0.90

(a) Effect of oxidation state on the autophosphorylation and turnover activities. (b) Inhibition of turnover caused by binding of heme ligands. Turnover represents the RmFixJ molecules phosphorylated per hour by one FixL molecule. For revised assay methods, see Sousa et al. [5]. Inhibition factors were computed by dividing the activity of the unliganded form (met or deoxy) by the activity of the same oxidation state bound to ligand. The activities of liganded forms were measured for protein over 99% saturated with ligand, except for O2 . Since it was not feasible to saturate RmFixLT and BjFixL completely with O2 , the O2 inhibition factors are conservative estimates based on subtracting the contribution of the deoxy fraction; actual inhibition factors may be much higher. For R206A BjFixL and R220A BjFixL, the O2 inhibition factors at standard temperature and pressure were not measurable because of the extraordinarily low affinities of these proteins.

essentially none of the active conformation (Table 1b) [5]. By contrast, CO and NO allow one-third to one-half of the FixL molecules to occupy the active conformation even when saturated (Table 1b) [4,74]. The most likely basis for the greater efficacy of O2 is the highly polarized bond that it forms with the heme iron, as suggested by the strong cyanide inhibition of BjFixL [4,5]. This polarization of the heme pocket on binding of O2 or CN− causes entry of the highly conserved G -2 arginine (Arg 220 in BjFixL, Arg 214 in RmFixL) into this pocket to form a hydrogen bond to the bound ligand [51]. An apparently strong inhibition of RmFixLT by imidazole needs special consideration, because, unlike BjFixL, oxidation of RmFixLT to the ferric form is strongly inhibitory even without a ligand [74]. That is, imidazole acts on a quite different initial state in met-RmFixLT than in met-BjFixL. Unfortunately, high-resolution structures have not been obtainable for the liganded (switched off) derivatives of the RmFixL heme-PAS domain [53]. More importantly, there is no structure for a FixL:FixJ complex for either FixL, or even a structure that includes the kinase. It is not clear what features of the regulatory conformational change can occur despite the absence of the kinase region.

A Surfeit of Biological Heme-based Sensors

29

Table 2. Parameters for binding of ligands to BjFixLs and SWMbs with and without polar stabilization Protein

R206A BjFixLa R220A BjFixLb R220A BjFixLHc BjFixLd Chelated protohemee SWMbe H64F SWMbe

CN−

O2 Kd (M)

kon (M−1 s−1 )

koff (s−1 )

Kd (M)

kon (M−1 s−1 )

koff (s−1 )

350 1500 1250 142 67 083 130

016 0002 14 014 0063 16 75

29 64 1750 20 4200 15 10000

1.7 16 — 0.94 — 1.3 920

75 × 10−5 38 × 10−6 — 11 × 10−4 — 32 × 10−4 12 × 10−7

13 × 10−4 63 × 10−5 — 12 × 10−4 — 40 × 10−4 11 × 10−4

Full-length R206A BjFixL ligand-binding data are from Sousa et al., 25 C, pH 8.0 [5]. Full-length R220A BjFixL ligand-binding data are from Dunham et al., 25 C, pH 8.0 [4]. c R220A BjFixLH O2 -binding data are from Balland et al., 25 C, pH 7.4 [82]. d For full-length BjFixL, O2 -binding data are from Gilles-Gonzalez et al., 25 C, pH 7.5 [1]; cyanide-binding data are from Dunham et al., 25 C, pH 8.0 [4]. e Data on O2 binding to chelated protoheme in benzene and sperm-whale myoglobins are from Olson and Phillips, 20 C, pH 7.0 [83]; data on cyanide binding to sperm-whale myoglobins are from Dou et al., 20 C, pH 7.0 [84]. a

b

3.1.1.3.2. Nature of the low inhibition by CO. As shown in Table 1b, the ligands (O2 and CN− ) that induce entry of the distal arginine into the BjFixL heme pocket have much higher inhibition factors than those that form a less polar bond or one with a different polarity [4,5,51,74]. Before the appearance of these enzymatic studies, FixL was postulated not to sense CO, and a reported failure of CO to regulate RmFixLT appeared to support this hypothesis [50,77]. The enzymatic study of RmFixLT that failed to observe an effect of CO had examined the reactions of FixL with ATP, rather than turnover reactions, and the levels of FixL phosphorylation reported, including that of the “on-state” deoxy form, were extraordinarily low [77]. The original proposal that FixL does not sense CO was based, not on kinase measurements, but on crystal structures that showed no substantive differences between the deoxy and carbonmonoxy forms of the BjFixL heme-binding domain [50]. Interestingly, these crystals were reported to have a CO affinity so astonishingly low as to suggest that the protein structure might have been somehow “locked” in the deoxy state [50]. Later, Key and Moffat examined BjFixLH crystals at room temperature, where the CO affinities were comparable to the affinities measured in solution, and they thus managed to observe a CO-induced structure relaxation [52]. Room-temperature crystal structures for deoxy- and carbonmonoxyBjFixLH, each state from two crystallographic space groups at 1.8 and 2.0 Å resolution, showed that binding of CO to the heme iron leads to a requisite and far-reaching movement of the leucine 236 residue [52]. This displacement of a distal leucine increased the flexibility of the FG loop, disordering the arginine 206 guanido group and permitting movements of the heme propionate 6 and aliphatic residues of the FG-loop, H -strand, and I -strand. These motions were propagated more than 15 Å to a helix outside of the PAS domain. These results are consistent with leucine 236 of BjFixL serving as

30

M.A. Gilles-Gonzalez and G. Gonzalez

a general-ligand detector, somewhat analogously to the way the displaceable methionine 95 residue functions in EcDos [52,54]. Photoacoustic-calorimetry studies additionally found that volume changes too substantive to be limited to the heme pocket occur in BjFixLH on CO photodissociation and rebinding [87].

3.1.1.4. The Heme-PAS Serves as an Electrostatically Driven Switch It is increasingly clear that the entire BjFixL heme pocket behaves as an electrostatically driven switch that forbids kinase activity in the O2 and CN− -bound states [5]. Even the wild-type protein has an O2 affinity so low as to suggest that the protein matrix actively depresses O2 binding [1,4]. The effects of side chains such as the arginine 206 and 220 residues, that form an integral part of this regulatory switch, are farther reaching than initially presumed from the initial structures of wild-type BjFixLH. Both the arginine 206 and arginine 220 residues are critical, not only for kinase regulation, but also for binding of O2 in air [4,5]. Although the propionate 6 carboxylate is the heme’s nearest contact to the arginine 206 guanido group, this residue nevertheless contributes an approximately twofold enhancement to the binding affinity [5]. The importance of the arginine 220 residue is even greater; without this arginine to assist ligand binding as part of the signal-transduction mechanism, the O2 affinity would become vanishingly low (Kd ∼ 15 mM) [4,232]. It is quite surprising that the affinity of wild-type BjFixL for O2 (Kd ∼ 100 M) is about the same as that of a heme for O2 when no stabilizing hydrogen bond is supplied [83], and it is even more startling that R220A BjFixL has an affinity 10 times lower than that [4]. Gilles-Gonzalez and colleagues have postulated that O2 binding and sensing by BjFixL is influenced more strongly by the interaction of the arginine 220 residue with the heme edge than by its hydrogen bond to bound O2 . In R220A BjFixLH, the heme structure radically differs from that in wild-type BjFixL or indeed any other known O2 -binding heme protein. The Fe His bond is extremely long (2.7 Å), the plane of the heme in the deoxy state is as flat as that of normal heme proteins in the oxy state, and this heme plane is rotated with respect to the proximal histidine plane on two axes [4]. Interestingly, there does exist a correlation between the O2 on-rate, a resonance Raman propionate mode, and the polarity of the residue at position 220 [88]. No other polar side chain appears capable of substituting entirely for a guanido group at that position. In particular, the arginine-propionate interaction confines unbound O2 to the heme pocket so effectively that approximately 90% of the O2 photolyzed by a laser flash-off typically rebinds in about 5 ps [82,89,90].

3.1.2. AxPDEA1: A Heme-PAS-controlled Phosphodiesterase Regulating Biofilm Production 3.1.2.1. Physiological Role It is reasonable to expect that aerobic processes that are costly and irreversible will be rapidly shut off during hypoxia. The coupled synthesis and excretion of cellulose by Gluconoacetobacter xylinum (formerly Acetobacter xylinum) appears to be just such a process [91]. Although AxPDEA1 had been known since 1990 to regulate G. xylinum cellulose production, 11 years would pass before its recognition as a heme-based sensor of O2 by Gilles-Gonzalez, Benziman, and their colleagues [6,92].

A Surfeit of Biological Heme-based Sensors

31

3.1.2.2. A Phosphodiesterase Activity Specific for a Novel Second Messenger The C-terminal half of AxPDEA1 is a “GGDEF/EAL” region (Pfam designations DUF1/DUF-2; discussed under Section 3.1.3.2.1), where the EAL domain is an enzyme specific for cleaving cyclic bis(3 -5 )diguanylic acid, also called c-di-GMP (Fig. 4) [91–94]. This dinucleotide subtrate consists of two GMP moieties cyclically joined “head to tail” by 5 -3 phosphodiester bonds [93]. The enzyme rapidly hydrolyzes c-di-GMP to linear pGpG, but it is essentially inert toward the well-known cyclic mononucleotide second messengers cAMP and cGMP [92]. Binding of the c-di-GMP to a bacterial cellulose synthase confers a 200-fold allosteric activation, without which this enzyme is essentially inactive [91,92]. Thus, the function of AxPDEA1 amounts to a biological removal of c-di-GMP, and thus an indirect “inhibition” of the cellulose synthase. Domains over 20% identical in sequence to the AxPDEA1 enzymatic region are ubiquitous in Bacteria, including quite a few pathogens and many species not known to make cellulose [6–8,94–96]. Over 500 proteins are already known to contain this region. In cases where physiological and genetic data are available, they correlate these proteins with global changes in metabolism, such as the switch of C. crescentus from stalked to swarmer cells and the formation of biofilms by E. coli, Salmonella typhimurium, Pseudomonas aeruginosa, and Vibrio cholerae [97–102]. Although biochemical studies are still limited to relatively few proteins, the studies done so far confirm their roles either as cyclases or phosphodiesterases specific for c-di-GMP and support the view that c-di-GMP is an important second messenger [6–9,92–95,102–105].

3.1.2.3. Importance of the Sensory Heme-PAS Domain Although many physiological factors control the balance of c-di-GMP degradation and synthesis in G. xylinum, the AxPDEA1 protein governs the dependence of cellulose synthesis on O2 [6]. The initial suggestion that AxPDEA1 might be a heme-based sensor came from a report of the sensitivity of G. xylinum cellulose synthesis to the culture aeration and a strong resemblance (30% sequence identity, 50% homology) of the AxPDEA1 N-terminal region to the FixL heme-binding domain [6,7,91,94]. No sign of a heme absorption had been reported, however, for the purified AxPDEA1 protein in many years of study, and this was due to the heme being lost during purifications of this protein. Gilles-Gonzalez, Benziman, and their colleagues designed a purification strategy that yielded the holoprotein, and this allowed them to measure the ligand-binding properties of the heme and examine ligand regulation [6]. The equilibrium dissociation constant for binding of O2 was found to be 10 M, and saturation of the heme with O2 inhibited the enzymatic activity [6]. These findings agreed with observations that static cultures of G. xylinum are clear, whereas their surface is blanketed with a cellulose pellicle. Cleavage of c-di-GMP prevails throughout the hypoxic culture, essentially blocking cellulose synthesis, whereas accumulation of c-di-GMP at the air–water interface allows the construction of an elaborate biofilm of microcrystalline cellulose. The closest known relative of AxPDEA1 is the E. coli Dos protein (EcDos) [7]. A sequence identity of more than 40%, extending over the heme-PAS as well as the enzymatic regions, is observed between the two proteins. Indeed, EcDos contains heme [7]. Interestingly, EcDos has the same affinity as AxPDEA1 for O2 (Kd ∼ 10 M) despite featuring a different heme-iron coordination in the unliganded state [7,106]. In AxPDEA1, the iron atom is pentacoordinate, and the ligand enters an unoccupied site [7,28]. In EcDos, the iron atom is predominantly hexacoordinate, and the ligand gains access to its

32

M.A. Gilles-Gonzalez and G. Gonzalez

binding site by “displacing” a resident methionine side chain [54,106]. The implications of heme-iron hexacoordination in some sensors are discussed under Section 3.1.7.2.

3.1.3. EcDos: an E. coli Sensor for Rapid response by Preexisting Proteins? 3.1.3.1. Discovery and Sensing Range After discovering that the former yddU gene of E. coli encodes a heme-PAS protein, Gilles-Gonzalez and colleagues hypothesized that this gene plays a role in biofilm production and renamed it EcDos, or “E. coli direct oxygen sensor,” for the following reasons [7,37]: • The closest known relative of EcDos is the proven O2 sensor AxPDEA1 (33% aa identity); • The affinity of EcDos for O2 (Kd ∼ 10 M) falls within a range of concentrations that E. coli normally encounters; • The affinity of EcDos for CO (Kd ∼ 10 M) is too low for sensing any normal physiological exposure of E. coli to this gas; binding of NO is exceedingly slow (∼0002 M−1 s−1 ), and the affinity for this ligand probably also falls outside any physiological range of NO concentrations; • In vivo, EcDos remains ferrous regardless of the stage of growth or media O2 tension, ruling out any possibility of its functioning as a redox sensor (Unpublished results of Delgado-Nixon and Gilles-Gonzalez).

3.1.3.2. Enzymatic Activity 3.1.3.2.1. Predictions from the protein sequence. The C-terminal half of EcDos is an enzyme for c-di-GMP degradation composed of a silent “GGDEF” domain together with an intact “EAL” domain (Fig. 4) [7,8]. Benziman and colleagues were first to note that the cyclases for c-di-GMP synthesis are widespread in bacteria and can be identified by a conserved 210-residue domain with an invariant Gly-Gly-Asp-Glu-Phe (GGDEF, designated “DUF-1” in Pfam, Fig. 4) [94,95]. The two acidic residues within the GGDEF consensus were since found to be involved in catalysis [103,104]. In wildtype EcDos, both are neutralized, with EGTQF replacing the GGDEF motif. Benziman and colleagues likewise noted that the phosphodiesterases for c-di-GMP degradation are ubiquitous in bacteria and may be recognized by a conserved 250-domain with a signature Glu-Ala-Leu (EAL, Pfam designation DUF-2) (Fig. 4) [94]. 3.1.3.2.2. False starts. An early report proposed that EcDos differs from all other biochemically characterized EAL proteins in being a cAMP, rather than a c-di-GMP, phosphodiesterase; the same report also hypothesized that EcDos senses redox potential rather than O2 [107]. The EcDos heme is easily kept reduced in vitro by mild reducing agents such as dithiothreitol, as may be expected from its myoglobin-like redox potential of +64 mV [108]. Consequently, it is hard to imagine the usefulness of a cellular redox sensor that would switch at this potential. As for the reported cAMP phosphodiesterase, this activity was about 50 times lower for purified EcDos than for the well-established E. coli cAMP phosphodiesterase CpdA in crude extracts of E. coli [109,110]. Some arguments were later offered, attempting to connect the dos gene to regulation of cAMP

A Surfeit of Biological Heme-based Sensors

33

levels on the basis of observations that the overexpression of dos to levels vastly surpassing that of cpdA correlated with a drop in the cAMP levels, and the deletion of dos correlated with a rise in the cAMP levels [111]. However, the levels of E. coli cAMP could not be restored to normal by complementing a dos deletion strain with the wild-type dos gene [111]. Given that cAMP levels respond to many types of cellular stress, a simple explanation for these observations is that the deletion and massive overexpression of dos lead to stresses in E. coli that correlate with changes in the cAMP concentration. 3.1.3.2.3. Activity of EcDos toward c-di-GMP. Presentation of c-di-GMP to EcDos did indeed show a robust activity of the enzyme toward this unusual nucleotide. Gomelsky and colleagues found that the EcDos EAL domain (Dos540−807 ) linearized c-di-GMP to pGpG with a turnover rate of 100 min−1 but was entirely inert toward cyclic mononucleotides such as cAMP [8]. The c-di-GMP turnover rate that they measured compared well to the rate of 190 min−1 that Benziman and colleagues reported for purified apoAxPDEA1 [94]. Subsequently, the same group that had earlier claimed a cAMP phosphodiesterase activity reported 3-minute single-time point assays of c-di-GMP hydrolysis from which turnover rates of 27, 67, 126, and 143 min−1 could be estimated for the FeIII , FeII , FeII O2 , and FeII CO forms of full-length EcDos, respectively [9]. These results qualitatively demonstrate coupling between the heme and c-di-GMP phosphodiesterase, although the activation by ferrous heme ligands would appear to be rather weak and nonspecific. Interestingly, ligand binding appeared to stabilize the active conformation of EcDos, in contrast to AxPDEA1 for which O2 was found to stabilize the inactive state [6]. The quantitative effect of ligands on EcDos may be much larger than these experiments indicate. Complications such as a reaction lag phase, product degradation, or product inhibition will introduce large errors in rates based on single time-point assays. For all but the simplest reactions, an accurate quantitation of enzyme regulation by ligands requires careful and complete reaction time courses. This is definitely the case for the FixL heme-regulated kinase [5]. It is also possible that some as yet unidentified factors serve to amplify the differences in the activities of different EcDos states. The cellular activities of GGDEF and EAL enzymes are usually stringently regulated, and factors such as signal binding, covalent modification, and allosteric effects of substrates and products can play a considerable role in this regulation [6,102–105].

3.1.3.3. Physiological Counterpart

The dos gene belongs to a two-gene operon 5 -yddV-dos-3 , where only the stop codon of yddV separates the two genes. The yddV-encoded protein is alternately called EcGreg, YddV, or YhcK. This protein is unusual in containing a C-terminal GGDEF domain without even a silent accompanying EAL. This feature was noted quite early during studies of GGDEF domains, and the yddV gene (then called yhcK) was extensively exploited as a reagent for complementing the absence of other c-di-GMP synthases in vivo or for examining the physiological effects of c-di-GMP overproduction [95,98]. As expected from the membership of yddV and dos in the same operon, the two genes are jointly transcribed [112]. This would seem paradoxical, since the Dos phosphodiesterase and EcGreg cyclase activities are expected to counter each other. This need not be so if the two proteins’ activities are ultimately determined, not by their in vivo levels, but by their responses to environmental signals.

34

M.A. Gilles-Gonzalez and G. Gonzalez

3.1.3.4. Examinations of Knockout and Overexpressing Strains Despite considerable effort to determine the functions of the yddV and dos genes, the physiological roles of this regulatory couple remain an open question. Two groups have reported physiological effects in E. coli of deleting the dos gene, deleting the entire yddV-dos operon, overexpressing the dos gene, or overexpressing yddV [111,112]. Their results are summarized in Table 3. Both studies suffered from significant shortcomings, including: a failure to complement the effects of the deletion(s) by supplying the

Table 3. Physiological effects of imbalances in yddV and dos expression in E. coli Strain

Genetic effects examined

Observations, compared to corresponding wild-type strain

Study

W3110

Effects of deleting dos on levels of c-AMP levels and growth morphology.

The dos deletion strain had 26-fold higher aerobic cAMP levels and a tendency toward filamentous growth in early stationary phase. No attempt was made to complement the dos deletion with a moderately expressed dos on a low-copy vector. No effects were reported for anaerobic conditions, presumably because EcDos could not be detected.

[111]

W3110

Effects of deleting the yddV-dos operon or the dos gene, or of overexpressing the yddV gene on growth morphology.

The yddV-dos deletion strain had no observable phenotype. The dos deletion and yddV-overexpressing strains formed filaments and swimmed poorly, based on optical and transmission electron microscopy, and plate swimming assays. Complementation of the dos deletion was not shown.

[112]

BL21(DE3) Effects of overexpressing the dos gene (on a pET plasmid) on c-di-GMP levels and growth morphology.

This extraordinarily high expression of wild-type dos correlated with only a modest drop (fivefold) in the cAMP levels. This also appeared to abolish filamentous growth. This was described as a “partial complementation.”

[111]

MG1655

Effects of overexpressing yddV on c-di-GMP levels and the gene expression pattern.

The yddV-overexpressing strain showed a 6–10-fold rise in c-di-GMP, the induction of over 40 genes, and the repression of a similar number of genes. These included genes associated with membrane components (e.g., ompC), cell division, sugar metabolism, iron uptake (e.g., fur), capsule synthesis, phospholipid synthesis, resistance to superoxide (e.g., soxS), and resistance to acid conditions (e.g., gadX, gadE).

[109]

TOB-1

Effects of overexpressing yddV on cellulose production.

The yddV-overexpressing strain showed increased production of cellulose on Calcofluor-agar plates.

[112]

A Surfeit of Biological Heme-based Sensors

35

wild-type gene(s) [111,112]; a lack of attention to associated changes of the growth media during hypoxia, especially the pH [111,112]; and an overwhelming use of standard E. coli laboratory strains that cannot form biofilms [113,114]. In particular, the effects of dos and yddV-dos knockouts were not examined in TOB-1, although this was the only E. coli strain used that could make substantive and cellulosic biofilms [97]. Consequently, both studies fell short of their original goal to link yddV to biofilm production. Membrillo-Hernandez and colleagues did successfully show, in agreement with several other studies, that the overexpression of yddV in a cell is tantamount to raising its c-di-GMP concentration [95,98,112]. They further associated this elevation in c-di-GMP to broad changes in E. coli gene expression, based on hybridization of purified mRNAs against antisense genomic microarrays [112]. Though the gene-expression effects are interesting and valuable, methods to identify direct targets of c-di-GMP binding in E. coli are likely to be yet more informative. Analogous microarray studies of G. xylinum gene expression would have entirely failed to identify an effect of c-di-GMP on the target bacterial cellulose synthase, since c-di-GMP enhances the activity of this enzyme rather than its production.

3.1.3.5. Postulated Physiological Roles The ability of EcDos to influence preexisting proteins allosterically via c-di-GMP, rather than control their de novo synthesis, probably represents the most important difference between this protein and other known E. coli O2 sensors, such as the transcription factor Fnr [37,115,116]. It is much easier, with modern microarray methods, to detect changes in gene expression than to identify enzymes whose activities are modified. One must not forget that any effect on enzyme activity that influences the cellular state will exert secondary effects on gene expression. E. coli is a facultative anaerobe that can experience extended periods of hypoxia or large transient swings in O2 concentration coinciding with its host’s ingestion of food. Therefore, it is reasonable to expect E. coli to feature sensors for governing both long-term adaptation to prolonged hypoxia and short-term adaptation to intermittent oxic conditions. If O2 scarcity is transient, it is wasteful to accomplish a temporary shutdown of inessential O2 -consuming processes by degrading and resynthesizing proteins rather than by temporarily inhibiting their activities. Like other known GGDEF- and EAL-domain proteins, EcGreg and EcDos probably govern a bacterial transition from a sessile to a motile state [91,98–102]. Study of these fascinating bacterial cyclases and phosphodiesterases will likely improve the understanding of processes such as biofilm formation and may yield valuable applications, such as new antibiotics.

3.1.4. NPAS2: A classical bHLH-PAS Transcription Factor Implicated in Higher Brain Functions 3.1.4.1. A Heme-PAS Coupled Mammalian Transcription Factor The domain organization of full-length NPAS2 places this protein in the bHLH-PAS class of eukaryotic transcription factors [117–119]. Most of these proteins do not require any cofactor to assist their responses to environmental signals [120]. Examples include the aryl hydrocarbon receptor (AhR) that binds a variety of xenobiotics and the hypoxiainducible factor 1 alpha (HIF-1) that indirectly responds to O2 levels [121,122]. The bHLH-PAS proteins are typically about 800-residues long, with an N-terminal basic

36

M.A. Gilles-Gonzalez and G. Gonzalez

helix-loop-helix DNA-binding region (bHLH, about 100 residues) immediately preceding a pair of PAS domains (PAS-A and PAS-B, ∼300 total residues) (Fig. 4) [120]. In NPAS2, each of the PAS domains binds heme [18]. As a rule, bHLH-PAS proteins bind to transcriptional enhancer sequences in DNA and activate gene expression only as heterodimers with a small set of general partners that are themselves also bHLH-PAS proteins [120]. Such partners can form heterodimers with multiple alternative proteins, and the competitiveness of a bHLH-PAS protein for “mates” changes in response to environmental signals. This variation is achieved either by changing the concentration of a bHLH-PAS protein or its affinity for partners. For NPAS2, the heterodimeric partner is the BMAL1 protein [118,119]. Gilles-Gonzalez, McKnight, and their colleagues used CO to demonstrate that binding of a heme ligand to NPAS2 is directly coupled to the formation of a NPAS2:BMAL1:DNA complex [18]. The PAS-A and PAS-B domains were found to bind CO with 10-fold different affinities (Kd values of ∼2 and 20 M, respectively), but they appeared to have no influence on each other’s ligand-binding behavior (Fig. 5).

(A) 0.12

0.8

PAS-A/B

0.06 0.00

0.4

Deoxy

Absorbance

0.0 0.12

0.8

PAS-A 0.06

0.4

Rate observed (s–1)

CO

(B) 80

PAS-A

70 60 kon = 0.4 μM–1s–1 40 20

kon = 0.04 μM–1s–1

7 0 0

100

200

300

PAS-B 400

500

Carbon monoxide (μM)

0.00

[CO], µM

0.0 0.12

0.8

1

3 10

(C)

30

BMAL1:BMAL1

PAS-B 0.06

0.4

0

HOLO NPAS2:BMAL1

0.00

BMAL1:BMAL1

0.0

300

400

500

600

700

NPAS2:BMAL1

APO

Wavelength (nm)

Fig. 5. Influence of CO on the mamalian bHLH-PAS transcription factor NPAS2 [18]. The doublet in the 500–600 nm region of the absorption spectra of the deoxy forms (A, solid lines) indicates heme-iron hexacoordination and the provision of two axial ligands to the heme iron by residues from the protein. The absorption spectra of the carbonmonoxy forms (A, dashed lines) indicate stable binding of CO. Limiting of the rates of CO association is illustrated in (B) (open symbols, independently measured heme-PAS domains; closed symbols, deconvoluted rates from protein containing both heme-PAS domains). In NPAS2 with heme at CO level consistent with saturation of the PAS-A domain inhibits the formation of a productive NPAS2:BMAL1:DNA complex (C, top), but in NPAS2 without heme, CO does not influence the formation of this complex (C, bottom).

A Surfeit of Biological Heme-based Sensors

37

From studies of the effect of CO on DNA binding, productive NPAS2:BMAL:DNA complexes were found to disappear between 1 and 3 M CO (Fig. 5). This transition corresponded roughly to the expected range for CO detection by this domain but was quite sharp when compared to the saturation of the PAS-A domain with CO. For PAS-B, the affinity for CO was such that this domain would not be expected to sense physiological levels of CO, which are expected to be at best in the micromolar range [18]. This suggests at least two possibilities for PAS-B signaling: additional factors might be needed to enhance the affinity for CO; alternatively, the signal ligand for this domain might be another physiological gas, such as O2 or NO.

3.1.4.2. Does NPAS2 Play a Role in Neurotransmission? Carbon monoxide is considered one candidate for the native signal of NPAS2, but sensing of other gaseous ligands by this protein has not been conclusively ruled out [18]. For NPAS2 to sense CO for neurotransmission, this gas would have to diffuse to a NPAS2containing neuron from a CO-generating nearby neuron over a very short synaptic gap [123]. If not, hemoglobin would scavenge the CO. Alternatively, the CO could come from within the same neuron where it would likely be produced by heme oxygenase 2 (HO-2) on the endoplasmic reticulum membrane. Although CO has traditionally been regarded as a breakdown product of heme degradation by heme oxygenase, this view is increasingly being reconsidered [124,125]. Several isoforms of HO exist in mammals [126,127]. The HO-2 isoform is peculiar to neurons and quite unlike the HO-1 enzyme of the liver and spleen [126,127]. Studies of HO-2 knockout mice suggest that the purpose of HO-2 is not to recycle heme but to turn over heme so as to generate CO for neurotransmission, analogously to the way nitric oxide synthase generates NO for vasodilation and other processes [128]. A physiological link between NPAS2 and HO-2 is suggested by the observation that in NPAS2−/− mice, the regions of the forebrain that can no longer express NPAS2 show enhanced expression of HO-2 [129].

3.1.4.3. Does NPAS2 Link Higher Brain Functions to Circadian Rhythm? In NPAS2 knockout mice, the forebrain regions unable to express NPAS2 also fail to manifest a normally rhythmic expression of the period (PER2) gene [129]. This finding is taken as one line of support for the involvement of NPAS2 in circadian rhythm. Additional support comes from a finding of in vivo heme regulation of PER1 and PER2 gene expression by a mechanism involving the NPAS2 and PER2 proteins [130]. A close relationship of NPAS2 to proteins that control circadian rhythm also implicates NPAS2 in this process. For example, the circadian-control protein CLOCK is the nearest known evolutionary relative of NPAS2; the physiological partner of NPAS2, i.e., BMAL1, can also partner with CLOCK [118,119]. In fact, NPAS2:BMAL1 heterodimers and CLOCK:BMAL1 heterodimers are equally competent at activating the expression of the period (PER) and cryptochrome (CRY) genes [118,119]. The PER and CRY proteins in turn inactivate the NPAS2:BMAL1 and CLOCK:BMAL1 heterodimers by an unknown mechanism, thus providing feedback for circadian cycling. The main difference between NPAS2 and CLOCK appears to be their tissue distributions. NPAS2 is abundant in the somatosensory cerebral cortex but absent from the suprachiasmatic nucleus, whereas CLOCK resides in many brain regions but is most abundant in the suprachiasmatic nucleus. Mice deficient in CLOCK exhibit profound circadian rhythm defects, whereas mice deficient in NPAS2 show subtle behavioral changes that implicate this protein

38

M.A. Gilles-Gonzalez and G. Gonzalez

in higher brain functions [129]. For example, normally nocturnal wild-type mice adapt their rhythms, possibly by entraining on food intake, to become active and to have an appetite during the day, if they are fed only during the daytime; by contrast, NPAS2−/− mice entirely lose their drive and appetite if they are fed exclusively during the day [129]. The knockout mice never adapt to restricted feeding, to the point where more than 80% of them become severely ill from starvation. A lack of NPAS2 also correlated with changes in electroencephalogram activity during nonrapid eye movement sleep and an attenuation of PER2 expression after sleep deprivation [131]. These observations suggest an involvement of NPAS2 in sleep homeostasis, possibly via an effect on cryptochrome levels [131,132].

3.1.5. Structural Information on the Heme-PAS Family 3.1.5.1. The Per-ARNT-Sim Sequence Motif The PAS sequence motif is not limited to heme binding or heme-ligand detection but is the hallmark of a versatile sensory domain found in more than 1300 different signaling proteins [133,134]. This motif encompasses 100–130 residues, with a middle “variablelength” region (10–30 residues) separating an N-terminal “core” from a C-terminal region, each with about 50 residues. In the entire PAS sequence, there are only about nine highly conserved residues. These are principally small residues such as glycines, aspartates, and asparagines, and seven of them occur in the core. The nonconserved regions of the PAS are those most indicative of specific sensing functions. For example, the middle variable-length and C-terminal regions of bacterial O2 -sensing PAS domains include all of the direct contacts of these proteins with the heme and heme ligands, such as the proximal histidine of heme attachment and the distal residues that line the ligand-binding pocket. In BjFixL and AxPDEA1, conservation of these regions results in a sequence identity of about 36% between these two eubacterial heme-PAS domains [6,37]. By contrast, the BjFixL and NPAS2 heme-PAS domains, which originate from evolutionarily more distant organisms, share the more typical PAS relationship of less than 12% sequence identity [18,38].

3.1.5.2. The PAS Structural Fold The possibility of a conserved PAS fold first arose from the nearly simultaneous determinations of the PAS sequence motif and a high-resolution structure of a PAS domain, that of the blue-light sensor photoactive yellow protein (PYP) [133–135]. The crystals structures of the O2 -sensing BjFixLH and voltage-sensing human ether-a-go-go-related gene (HERG) proteins experimentally showed the existence of a common fold in these sensors with widely different functions [29,136]. Most of the conserved small residues in PAS domains were found to delineate the boundaries of secondary-structure regions. By contrast, the nonconserved regions were found to participate more intimately in signal detection. Some of the proteins contained cofactors in their nonconserved regions. The FixL proteins have their heme moiety linked to a histidine imidazole [29,53,137,138]; PYP contains a parahydroxycinnamate cofactor linked to a cysteine thiol [135], and HERG has no cofactor [136]. Most PAS-domain proteins are anticipated not to require cofactor augmentation for sensing, although the ones with cofactors have been most appealing for study. The availability of crystal and NMR structures of PAS domains has rapidly accelerated [54,55,139–143].

A Surfeit of Biological Heme-based Sensors

39

3.1.5.3. Heme-PAS-domain Structures For the heme-PAS domains, the following high-resolution crystal structures are currently available: • BjFixLH: deoxy (FeII ), oxy (FeII O2 ), carbonmonoxy (FeII CO), nitrosyl (FeII NO), met (FeIII ), and imidazolemet (FeIII Imid) forms [29,51,52]. • R206A BjFixL met form [5]. • R220A BjFixLH met form [4]. • RmFixLH: deoxy and met forms [53]. • EcDosH asymmetric FeII /FeII O2 dimer, deoxy form, and met form [54,55]. For all of these proteins and other PAS domain–containing proteins, the overall threedimensional structure of the PAS domain can be compared to a baseball glove with its thumb resting against the rest of a six-fingered hand: the palm being a set of short -helices (C ,D ,E ), the thumb being a longer helix (F ), and the other fingers being a network of five antiparallel -strands (A , B , G , H , and I ) [29]. In the hemePAS proteins, the heme is pinched, like a ball, principally between the thumb (a 16–17 residue F -helix) and the index (G -strand) and “middle” (H -stand) fingers (Fig. 6). The strongest link between the heme and protein is the coordination bond of the heme iron to the proximal histidine [137]. This histidine is the most conserved residue of heme-PAS proteins (F 3 histidine, or His 200 in BjFixL, His 194 in RmFixL, His 77 in EcDos) [29,37,53–55,138]. The distal heme pocket, where the ligands bind, is highly hydrophobic due to the large aliphatic side chains protruding into this pocket from the FG-loop and the G - and H -strands [29,37,53–55]. Oxygen-sensing heme-PAS-domain distal pockets typically also feature a well-conserved arginine about 20 residues after the proximal histidine (the G -2 arginine, or R220 in BjFixL, R214 in RmFixL, R97 in EcDos) (Fig. 6) [4,29,37,53–55]. This distal arginine directly interacts with bound O2 (Fig. 6) [4,29,54]. The heme iron of EcDos is coordinated, not only to a histidine on the proximal side, but also to a displaceable methionine residue on the distal side [106]. This displaceable residue, or DR, occurs near the start of the G -strand. For the methionine 95 sulfur to coordinate to the heme iron, the orientations of the F -helix and G -strand in EcDos were predicted to be quite distorted from their positions in BjFixLH [106]. A crystal structure of ferrous EcDosH, half-saturated with O2 , shows that this is accomplished by distorting the popypeptide backbone [54].

3.1.6. Models of PAS Regulation Inspired by FixL 3.1.6.1. Hypotheses Based on Crystal Structures

On binding of O2 or CN− to BjFixLH, the biggest movement seen is a 1.6-Å displacement of a loop linking the proximal F -helix of heme attachment to the distal G -strand lining the ligand-binding region [29,51]. This FG-loop movement is accompanied by flattening of the heme, a slight displacement of some apolar side chains out of the heme distal pocket (Ile 215, Ile 238), and a reorganization of the polar residues directly or indirectly hydrogen bonded to the heme propionates (Arg 206, His 215, and Arg 220). Several alternative triggering mechanisms have been proposed for those changes. An early hypothesis that the more planar heme triggers kinase inhibition has been ruled out by the discovery of a very planar heme in deoxy-R220A BjFixL, even though

40

M.A. Gilles-Gonzalez and G. Gonzalez (A)

(B) G

G

H

H D212

R220

H214

I

D212

C

A

I

H214 R220

B

B H200 R206

H200 R206

D

E

(C)

BjFixL RmFixL Mt Dos AxPDEA1 EcDos

F alpha helix

proximal

197 191 181 64 74

220 214 201 87 97

D

E F

F

BjFixL RmFixL Mt Dos AxPDEA1 EcDos

C

A

FG loop

0

1

2

3

4

5

6

7

8

9

10

11

12

13

R R L R R

S H E H P

R E K E A

H H H H H

D D D D P

S G R E

Y Y F Y Y

I L I I

S Q N D R

R R R R H

Y Y I N N

R M I R R

T A D E E

T T R T G

S G G G

D E E H K

–2

–1

G beta 0 1

2

3

–1

0

1

2

3

R R R R

I V E E

V V V V L

T S E E Q

G G G F L

K Q V T E

T T R Y K

F F V I I

P P P C W

M M L G T

H K E E R

R R R R K

R K K A K

D D D D D

G G G G G

T S T E S

Distal P K I N A

H R S R R

H beta 4 5 Distal L L I L F

S A S S A

I I I I V

I I M V E

G G V G G

I I T T M

G D D S S

6

7

8

9

10

I V V L L

G G F S S

E E S K K

M M W V V

Q R N Q S

Fig. 6. Structural elements implicated in conformational switching by the BjFixL heme-PAS domain, and their occurrence in proven heme-PAS proteins. In unliganded “on-state” BjFixLH (A), the G -2 arginine (Arg 220) on the distal side of the heme forms a hydrogen bond to the heme propionate 7, and the F 9 arginine (Arg 206) on the proximal side interacts with the FG loop [29,51,52]. In liganded “off-state” BjFixLH (B), the G -2 and F 9 arginines switch their hydrogen-bonding interactions to the bound O2 and the heme propionate 6, respectively [29,51]. The structures of the known on-state deoxy and met forms are very similar; likewise, the structures the known off-state oxy and cyanomet forms closely resemble each other. Structures are compared for PDB files 1XJ3 and 1LT0; on versus off, respectively. An alignment is shown (C) for the F-helix and FG-loop sequences of B. japonicum FixL, S. meliloti FixL, Methanobacterium thermoautotrophicum Dos, G. xylinum PDEA1, and E. coli Dos. The absolutely conserved F 3 residue (H200 in BjFixL), or proximal histidine, coordinates the protein to the heme iron. The conserved G -2 arginine (R220 in BjFixL) and usually basic F 9 residue (R206 in BjFixL) strongly influence affinity and regulation [4,5]. The G -2 arginine alternately interacts with the heme propionate 7 in the unliganded on-state or with bound O2 or CN− in the off-state; the F 9 residue (R206 in BjFixL) alternately interacts with the FG-loop in the BjFixL on-state or the heme propionate 6 in the off-state. (see Plate 5.)

this protein is 3 times more enzymatically active than deoxy-BjFixL [4,29,72]. Another model proposed the displacement of bulky distal residues as the regulatory trigger [144]; however, the activities of FixL proteins with distal-pocket residue substitutions revealed no correlation between regulation and the sizes of those side chains [145]. A more recent model suggests that the main regulatory trigger is the entry of the G -2 arginine (Arg 220

A Surfeit of Biological Heme-based Sensors

41

in BjFixL) into the heme distal pocket to form hydrogen bonds to regulatory ligands [51]. This model was inspired by high-resolution crystals structures showing that the G -2 arginine forms a salt bridge to the heme propionate 7 in unliganded BjFixL, but that bound O2 and CN− cause this arginine to enter the distal pocket (Fig. 6) [29,51].

3.1.6.2. Views of Regulation from Enzymatic Measurements Direct measurements of regulation indicate that it may be a mistake to imagine a single “trigger” switching FixL from an “on” to an “off” conformation. Some FixL variants, like R220A BjFixL, are more active than wild-type FixL in the deoxy state [4]. Since it is unreasonable to have an inhibition factor less than 1, this implies that in normal FixL, even the “noninhibiting” states have some heme-binding domains in the inhibiting conformation. It is quite likely that each FixL species, liganded or unliganded, represents an equilibrium between conformations that inhibit the partner kinase and conformations that do not. Moreover, there probably exist multiple factors that can shift this equilibrium toward the inhibiting conformations. When these factors act in concert, almost all of the FixL will be in the “off” state; but even in the best of circumstances, FixL will never be entirely in the “on” state. Strong regulators of BjFixL, such as O2 and CN− , show inhibition factors better than 100-fold; poor regulators such as NO and CO also inhibit this protein, but with factors of less than fourfold (Table 1) [4,5]. The arginine 206 and 220 residues (the F -9 and G -2 arginines, respectively) do indeed play important roles in adjusting the relative stabilities of the “off” and “on” conformations in FixL, but these residues are not absolutely required for a response to ligand [4,5]. It is not practical to measure the O2 inhibition of the R206A and R220A BjFixL mutants because their extremely low affinities for O2 (Kd ∼ 035 mM for R206A BjFixL and 1.5 mM for R220A BjFixL) leave them predominantly in the deoxy (active) state even in pure O2 (Table 2) [4,5]. Saturation with cyanide inhibited the wild-type BjFixL much more strongly than the R206ABjFixL or R220A BjFixL variants (Table 1) [4,5]. This finding of defective inhibition without the F -9 and G -2 arginines argues strongly for these residues being important for control of the kinase activity [4,5].

3.1.7. Insights from Comparisons of Heme-PAS Sensors 3.1.7.1. Salt-bridge Alteration as a Mechanism for a Regulatory Conformational Change The alteration of the salt bridge between the G -2 arginine and the heme propionate 7 during binding of O2 is a conserved feature of the BjFixLH and EcDosH proteins [51,54]. This salt-bridge swapping mechanism may be maintained in other heme-PAS O2 sensors, most of which feature this distal arginine. Interestingly, the sense of the saltbridge alteration differs for BjFixLH and EcDosH, even though in both cases hydrogen bonding of the G -2 arginine to O2 follows binding of O2 [51,54]. In BjFixLH the G -2 arginine (Arg 220) must rupture a salt bridge to the heme propionate 7 in the deoxy state to form a hydrogen bond to O2 in the oxy state (Fig. 6) [4,51]. In EcDosH, the displaceable residue (the DR, or Met 95) must rupture its coordination from the heme iron in the deoxy state to allow the G -2 arginine (Arg 97) to form hydrogen bonds to O2 and the heme propionate 7 in the oxy state [106]. It is likely that the displacement of the DR out of the heme pocket provides the basis for a general detection of ligand binding, whereas the entry of the G -2 arginine into the pocket to shield the negative dipole

42

M.A. Gilles-Gonzalez and G. Gonzalez

of bound O2 supplies the basis for a selective response to this ligand. Such a mechanism should best respond to ligands that carry a net or partial negative charge while bound. Although the crystal structures of the BjFixL heme-PAS provide some insights into plausible regulatory mechanisms, the subtlety of the changes and absence of a transmitter region to support these models do require caution. It is quite likely that the regulatory conformational changes will be more dramatic in a full-length protein that allows for longer-range interactions. The correspondence of the structural effects in the heme-PAS domain to the ligand-binding and signal transduction effects in the full-length protein confirms the importance of some residues, but this does not rule out the possibility that some unidentified residues might be equally important.

3.1.7.2. Displacement of a Coordinated Residue as a Basis for Conformational States In the PAS-A and PAS-B domains of NPAS2, each heme iron is coordinated to protein residues at both of its axial positions (Fig. 5A) [18]. The DR is not yet known for either of the PAS domains in NPAS2, although in EcDos it is well established as being the methionine 95 residue [106]. This is because the well-conserved residues of bacterial heme-PAS proteins have no clear equivalents in NPAS2, and structures are not yet available for NPAS2. Resonance Raman and mutagenesis studies of NPSA2 PAS domains have suggested a variety of possibilities for ligation of the iron atoms [146,147]. So far, three independent amino acid substitutions in the PAS-A domain (H119A, C170A, and H171A), and one in the PAS-B domain (H335A) have resulted in heme-iron atoms with a more pentacoordinate nature. Such experiments presume that alterations of distant side chains cannot perturb the coordination of the metal center: an assumption that may be incorrect, given the known plasticity of heme-PAS domain structures [4,5,51,52,54]. Indeed, a NPAS2 metal center was found to be affected even by its neighboring bHLH domain, with absorption shifts and heme destabilization reported for the isolated PAS-A domain (FeIII state) compared to a bHLH-PAS-A construct [148]. Despite the intrinsic hexacoordination of the heme-iron atoms in NPAS2 and EcDos, standard heme ligands readily displace their DRs, triggering a requisite and substantial movement of these residues. The two well-defined and stable states of the DR, i.e., coordinated versus not coordinated, were proposed for EcDos to provide the basis for corresponding protein conformational states, and the occurrence of a DR in a sensor is generally assumed to implicate this side chain in triggering regulatory conformational changes [7,106]. The structure of a mixed FeII /FeII O2 EcDosH dimer indeed shows that a displacement of the methionine 95 DR by O2 substantially alters the protein’s conformation [54]. The FG-loop becomes more distorted and less ordered than during BjFixLH binding to O2 , and this distortion extends to the EcDosH G -strand containing both the DR and the distal arginine [54]. In EcDos the DR appears to have a more complex function than that of a simple competing ligand [106]. Substitution of this residue (Met 95) with isoleucine failed to raise the affinity for all ligands by the same factor, contrary to what would be expected for lifting a simple competition [106]. In addition, the off-rate of the DR from the EcDos heme iron was sufficiently fast (∼90 s−1 ) that it never limited the association rates of gaseous ligands at experimentally practical concentrations [7]. For NPAS2, the PAS-A and PAS-B domains had limiting rates of CO association of 70 s−1 and 7 s−1 , respectively: a behavior quite unlike that of EcDosH (Fig. 5B) [18]. These rates correspond to the dissociation of each DR and hence the rates of formation

A Surfeit of Biological Heme-based Sensors

43

of each pentacoordinate form, which no ligand association can exceed. The hexacoordination of the metal center in some sensors implies that ligand binding always entails a competition between the signal ligand and a weakly bound endogenous ligand. Given the high effective concentration of the endogenous ligand, i.e., the DR, if it were not weakly bound, signal binding could not occur. It is reasonable to assume that physiological factors, such as the presence of regulatory partners or small allosteric effectors, might in some cases modulate the affinity and mechanistic behavior of a DR. This would provide a means for the physiological environment to reset the switching point of a sensor.

3.1.7.3. Monomer–dimer Equilibria as a Means of Conformational Switching For the mammalian bHLH-PAS proteins (Section 3.1.4), including NPAS2, switching is known to be achieved by allowing or forbidding formation of heterodimers [18,118– 120,142]. One contact surface for these dimers is believed to reside in the PAS domain itself [142]. Could a regulatory mechanism involving dimerization and monomerization via PAS domains be a general feature of PAS domain–containing proteins, including FixL? If we take this model very literally, the short answer is no. Full-length FixL proteins, not deleted of their kinase, remain dimeric whether or not ligands are bound. The physiological sensing complex, FixL2 :FixJ2 , shows no sign of dissociating upon O2 binding, as determined by fluorescence spectroscopy, and consequently, altered oligomeric states do not represent a mechanistic route for FixL signaling [75]. Furthermore, the association of FixL with FixJ, which is certainly an important point of regulation, cannot be a PAS–PAS interaction, since FixJ lacks a PAS domain (Section 3.1.1.2) [73,75]. Nevertheless, there is compelling evidence that ligand binding to FixL can disrupt some PAS–PAS interactions. This issue has been somewhat muddled by conflicting reports on two different FixLs and a variety of FixL heme-PAS truncations from different research groups. These are RmFixL119−266 [1], RmFixL123−251 [53], RmFixL127−260 [149], and BjFixL141−270 (also called BjFixLH and corresponding in sequence to RmFixL135−264 ) [1]. BjFixLH crystallizes as a monomer, whether it is liganded or unliganded, ferrous or ferric [29,51,52]. By contrast, RmFixLH forms dimers when unliganded and monomers if liganded. Analytical gel filtration experiments have shown that CO saturation can cause RmFixL127−260 to monomerize from a dimeric deoxy state [149]. This observation is consistent with crystallization of the unliganded ferric RmFixL123−251 as a dimer [53], even though NMR spectroscopy studies of the heme-binding pocket structure in cyanometRmFixL119−266 , which required monomeric species in millimolar concentrations, were quite feasible [137]. The reported failure to obtain liganded crystals of RmFixLH by diffusing ligands into the deoxy or met forms also suggests that ligands may induce quite large conformational changes in this protein [53].

3.2. The Helix-swap Model: A General Mechanism for Regulation by PAS Domains If the PAS–PAS interactions of heme-PAS fragments are recapitulated in the corresponding full-length proteins, then why don’t regulatory ligands also cause these full-length versions, e.g., BjFixL, to monomerize? Perhaps they cannot do so because a ligandindependent dimerization interface within the transmitter domain is sufficiently extensive to hold the dimer together. Indeed, in the protein-histidine kinases and in NPAS2,

44

M.A. Gilles-Gonzalez and G. Gonzalez

the main dimerization determinants are thought to reside in the “HisKA” and bHLH regions, respectively (Fig. 4) [150,151]. Nevertheless, even without monomerization, a breach of PAS–PAS contact could readily trigger switching. An interesting recent finding along those lines comes from NMR spectroscopy and limited proteolysis studies of a blue-light sensor, the oat phototrophin 1 (NPH1) [141]. This sensor contains a flavinmononucleotide (FMN) binding PAS region coupled to a kinase. Blue-light exposure of a NPH1 PAS domain having a C-terminal extension (NPH1404−560 ) showed this signal to trigger the PAS domain to destabilize an -helix within the C-terminal extension but outside of the canonical PAS structure [141]. These results suggested that PAS domains alternate between a closed conformation where they constrain a target protein region and an open conformation where they release this region. For NPH1404−560 , the regions of interaction were mapped to the solvent-exposed surface of the PAS -sheet and a sequence linking the PAS domain to a kinase. Reports on RmFixLH and EcDosH, on the other hand, have shown the same PAS surface to interact with a helix preceding the core canonical PAS structure and mediating PAS–PAS contacts [53–55]. One might therefore ask whether PAS domains exclusively govern PAS–PAS contacts in some proteins and PAS-transmitter contacts in others. We propose instead that, during signaling, PAS domains alternate between both types of interactions by a “helix-swap mechanism” (Fig. 7). Such a mechanism would permit a PAS domain to interact with a helix of a target transmitter in the closed conformation and to swap this interaction for one with a helix of another PAS domain in the open conformation. Specifically, the helix-swap mechanism would have the following features (Fig. 7).

(B) (A)

I

r

itte

sm

ACTIVE FORM

n

Sensor PAS

Tr

ter

II

II

a Tr

mit

Sensor PAS

an s

I

II INACTIVE FORM

Fig. 7. The helix-swap mechanism of regulation by PAS domains. In the hypothesized “helixswap mechanism,” the external surface of the -sheet in each PAS domain can engage in two mutually exclusive interactions with the helical regions I and II. The association of Region I to the PAS domain leads to a PAS/PAS interaction that displaces the transmitter and frees it to adopt an open, active, conformation (A). The association of Region II to the PAS domain results in a PAS/transmitter interaction that confines the transmitter to a closed, inactive, state (B). Electrostatic rearrangements, originating within the PAS domain, and triggered by the presence or absence of signal, stabilize these alternative interactions by altering the PAS-surface structure and potential.

A Surfeit of Biological Heme-based Sensors

45

• The PAS domain is bounded by two regions, I and II, either of which can adopt a helical conformation, and for either of these regions, the helical conformation is stabilized by its interaction with the PAS. • In the open, active, conformation of a sensor, the PAS domain interacts with region I resulting in a PAS–PAS dimerization. • In the closed, inactive, conformation of a sensor, the PAS domain interacts with region II, resulting in a PAS-transmitter interaction that inactivates the target transmitter. • These interactions of the PAS domain with regions I and II are mutually exclusive. This model is applicable to many different sensors, it is consistent with a variety of seemingly contradictory observations, and it is readily testable. It makes at least two important predictions. First, depending on the way a PAS domain is truncated, it will be discovered to interact with either of two alternative protein regions, and it will be dimeric while interacting with region I and monomeric while interacting with region II. Second, the conformational changes of PAS domains will be manifested most fully in protein fragments sufficiently long to alternate between the two main conformations available to those domains. For crystallization experiments, this implies that once a PAS domain has crystallized in one conformation, it will remain frozen in that conformation, and the introduction of signal in the crystal will elicit changes far short of those that are possible in solution. On the other hand, the full extents of the regulatory conformational changes brought on by the signals should be quite observable for proteins examined in solution or crystallized in the presence of their signal.

3.3. The CooA Protein 3.3.1. Evolutionary Origin and Adaptive Function The CooA protein from the purple nonsulfur bacterium Rhodospirullum rubrum is an unusual CO-sensing transcription factor that governs the oxidation of CO (Fig. 3) [17,152]. Early experiments conclusively showed the following. CooA is closely related to the cAMP-receptor proteins (CRP), with the CooA sequence being about 30% identical to that of E. coli CRP [152]. The metabolism of CO by R. rubrum is an anaerobic process [152]. The specific function of the CooA transcription factor, determined from genetic and physiological studies, is to induce the genes for producing a complex system of enzymes that oxidize CO to CO2 (Fig. 3) [152]. The deoxy form is an inactive species that does not bind DNA [17]. The carbonmonoxy form, on the other hand, can activate transcription from specific promoters in R. rubrum DNA [17,152–154].

3.3.2. Heme-iron Coordination and Protein Structure CooA is typically isolated in the unliganded deoxy form, or “off-state” when purified anaerobically to avoid its oxidation by O2 [17]. Under such conditions, the absorption spectra indicate a low-spin and hexacoordinate ferrous heme iron, as do resonance Raman and magnetic circular dichroism spectra [155,156]. Mutagenesis experiments, together with spectroscopic studies, identified the proximal axial ligand to the heme iron in the

46

M.A. Gilles-Gonzalez and G. Gonzalez

ferrous form as histidine 77 [155–158]. A crystal structure of full-length R. rubrum CooA in the deoxy state showed a homodimer resembling the E. coli catabolite repressor protein (CRP), as anticipated from the protein sequence [31]. In this structure, the heme of each monomer was held in a domain analogous to the cAMP-binding domain of CRP. As expected, the histidine 77 residue supplied the proximal axial ligand to the heme iron in each monomer. Interestingly, the proline 2 residues from each partner monomer supplied the distal heme-iron axial ligands: a quite unprecedented ligation for a heme protein [31]. The heme-binding domains, although symmetrically organized, were linked to a pair of asymmetrically arranged DNA-binding domains by their respective “C-helices,” one continuous and one discontinuous. A comparison of this asymmetric deoxy-CooA structure to a more compact and symmetric CRP structure has been put forth to suggest that in the carbonmonoxy form of CooA, the CO-bound heme alters the conformations of the C-helix to promote a more symmetric arrangement of the DNA-binding region. Ultraviolet-resonance Raman studies have found that an indole side chain belonging to the C-helix (Trp 110) goes from being exposed to solvent in the deoxy state to being buried in the CO-bound state [159]. This has been interpreted as a rotation of the helix during signal transmission, although alternative explanations, such as partial unwinding of a small region containing tryptophan 110, are also possible. Another -helix, helix E, contains a glutamate residue (Glu 167) that probably binds a divalent cation (Mg2+ ) and can engage in an interesting synergism with the CO-bound heme in establishing high-affinity DNA binding [160]. A forthcoming crystal structure of Carboxydothermus hydrogenoformans CooA might shed some light on such details of the activation of CooA and help to answer the following questions [161]. Is the observed asymmetry of deoxy-CooA in crystals a necessary aspect of the off-state or simply the result of crystal-packing forces? Does the structure of a CO-bound “on-state” differ from that of DNA-bound CRP?

3.3.3. Carbon Monoxide Binding Measurements of the dynamics of CO binding to the heme iron in CooA show this binding to be strongly cooperative, with a Hill coefficient n = 14 [49]. The association of CO to the first subunit in CooA is thought to lead to conformational changes sufficiently large to decrease dramatically the affinity of the DR in the second subunit for the heme iron. The P50 value for binding of CO was found to be 2.2 M, and the fitted Adair constants were K1 = 016 M−1 and K2 = 13 M−1 [49].

3.3.4. Structure-function Studies Employing Mutagenesis The relatively early solution of a crystal structure for deoxy-CooA, together with the development of powerful in vivo screens for transcriptional activation at target promoters, led to an initial focus on mutagenesis of the R. rubrum protein. Targets for analysis were residues of the C-helix that line the heme-distal pocket and those whose side chains directly coordinate to the heme iron. Thus, it was found that the only indispensable residue was the histidine 77 that supplies the imidazole side chain for coordination to the iron atom in the ferrous state [157]. In the ferric state, the proximal axial ligand to the heme iron is not the imidazole at position 77 but a thiolate from a cysteine at position 75 [156–158]. This residue does not affect CO regulation or activity, but it is essential

A Surfeit of Biological Heme-based Sensors

47

for the extremely low redox potential of CooA (−300 mV) that allows CO sensing to be disabled by oxygen [157,162]. The seemingly unique ligation of an N-terminal proline (Pro 2) to the iron atom on the distal side could be substituted by ligation of any neutral peptide N-terminal amino group, without much ill effect on DNA binding or transcription activation [163]. Deletion of residues from the N-terminal end of the protein, together with changes of the heme-distal pocket, led to sterically less constrained variants that could bind cyanide, proline analogs, and histidine analogs [164–167]. Interestingly, the only alternative ligand found to activate such a reengineered CooA was imidazole, and though this ligand could bind to the ferrous as well as ferric CooA variants, it only activated the ferrous state [165,167].

3.3.5. Examination of Natural Variants The discovery of seven additional homologs over 50% identical in sequence to R. rubrum CooA (RrCooA) is likely to yield a watershed of information about this type of sensor [56]. Analyses of these CooA sequences rapidly revealed that none had an N-terminal proline, all had retained a proximal histidine, and six lacked the proximal cysteine implicated in a redox-driven proximal-ligand switch in RrCooA. Though the CooA from C. hydrogenoformans (ChCooA) was the only one with the potential for proximalcysteine (Cys 80) ligation, the iron atom in this protein, whether FeII or FeIII , was exclusively coordinated to a proximal histidine residue (His 82) [168]. Ferrous ChCooA did not discriminate against NO activation: the NO-saturated protein activated the transcription of a target promoter and bound DNA in vitro with an affinity identical to that of the CO bound form [168].

3.3.6. Oxygen-disabled CO Sensing in R. rubrum CooA In a trivial sense, any heme-based sensor of a ferrous ligand will become disabled by oxidation. However, disabling of CO sensing by oxidation in R. rubrum has a special physiological significance. Since R. rubrum is a facultative aerobe that metabolizes CO solely under anaerobic conditions, economy would call for this organism to disable CO activation in air [152]. Oxidation of ferrous RrCooA to the inactive ferric state is accompanied by a switch in the proximal axial ligand from neutral histidine 77 to the cysteine 75 residue [156–158]. This switch in proximal ligation is not required for inactivation of DNA binding; ferric CooA is inactive even in variants without the alternate proximal ligand. However, the stabilization of increased charge on the ferric iron upon switching ligation from the neutral histidine to the anionic thiolate results in an extremely low redox potential (−300 mV) [162]. RrCooA is so easily oxidized that no report exists of a stable oxy form. Binding of O2 to RrCooA always leads to oxidation, as does binding of this ligand to free heme. In contrast, an obligate anaerobe such as C. hydrogeniformans would have no need of a mechanism for disabling aerobic CO activation. Not surprisingly, the proximal ligation of ChCooA is to a neutral histidine in both oxidation states [168]. The redox potential of ChCooA (+200 mV) is such that it should remain reduced in vivo [162]. Thus, the CooA proteins provide an excellent illustration of how the redox potential, ligand affinities, physiological ligands, and built-in capacity for discrimination can conspire to determine the ultimate output of a heme-based sensor (Table 4).

48

M.A. Gilles-Gonzalez and G. Gonzalez

Table 4. The outputs from two different CooA sensors Treatment

Reduced Oxidized CO added to reduced state NO added to reduced state O2 added to reduced state CN− added to oxidized state a

R. rubrum CooA

C. hydrogeniformans CooA

Coordination

Status

Coordinationa

His77 FeII Pro2 Cys75 FeIII Pro2 His77 FeII CO FeII NO Cys75 FeIII Pro2 Cys75 FeIII Pro2

Off Off On Off Off Off

His His His His Not Not

FeII N FeIII N FeII CO FeII NO reported reported

Status Off Off On On — —

Known from electron paramagnetic resonance and magnetic circular dichroism spectra [168]. In every case, a coordination to histidine 82 is expected for ChCooA; in the unliganded forms, the additional coordination is to a neutral N-terminal amino group from the polypeptide.

3.3.7. Regulatory Mechanisms of CooA Proteins It is quite likely that alternative regulatory mechanisms will be found to operate in CooA proteins, depending on the functions they serve. It is even possible that only some CooAs have evolved to sense CO. So far, studies of CooAs, natural and mutagenized, suggest that their only activatable state is the one with proximal-histidine coordination of the heme iron. Nitric oxide readily binds to RrCooA and to ChCooA, and this ligand surely displaces the DR of each protein; however, NO fails to activate RrCooA but fully activates ChCooA (Table 4) [168,169]. If this difference is due to the formation of a hexacoordinate nitrosyl-ChCooA, as opposed to a pentacoordinate nitrosyl-RrCooA, as does occur, then activation of the protein would indeed require displacement of a DR from the distal side as well as histidine ligation of the iron atom on the proximal side. Though ChCooA has not been reported to bind O2 , its occurrence in an obligate anaerobe suggests that this protein would not have evolved a mechanism to exclude O2 , and its high reduction potential implies that it should remain reduced on exposure to O2 . If so, then does O2 also activate transcription of ChCooA? Given that no redox-driven proximal-ligand exchange occurs in this protein, can binding of cyanide activate the ferric form? More generally, might CooAs identify their cognate ligand by marshalling a distal residue into an electrostatic interaction with the bound ligand after displacing a DR, analogously to the recognition of O2 by FixL and Dos? The aggregate answers to questions such as these will likely provide a clearer picture of the CooA heme-pocket requirements for binding and discrimination.

3.4. The Globin-coupled Sensors 3.4.1. The HemAT Proteins 3.4.1.1. Origins and Physiological Roles The discovery of the HemAT proteins by Alam and colleagues established that a globinlike fold can function in regulation as well as ligand transport [34]. Initially, they found two interesting heme-containing aerotaxis transducers [34]. The first, HemAT-Hs, was from the salt-loving archaeon Halobacterium salinarum and 489-residues long. The

A Surfeit of Biological Heme-based Sensors

49

second, HemAT-Bs, was from the soil bacterium Bacillus subtilis and 432-residues long. The N-terminal sequences of both proteins were weakly homologous to sperm-whale myoglobin. These regions are now known to characterize a broad class of globin-coupled sensors (Fig. 8) [40]. Another sequence of about 250 residues, at the C-terminal ends of the proteins, was about 30% identical to the cytoplasmic-binding domain of the E. coli methyl-accepting chemotaxis protein Tsr and could even be recognized in HemAT-Hs by antibodies against Tsr [34]. There is as yet no in vitro direct demonstration of coupling of HemAT heme state to methyl acceptance or CheA kinase activity. By disrupting, overexpressing, or reintroducing the hemAT genes in strains of H. salinarum or B. subtilis and examining the movement of these organisms in an O2 gradient, it was possible to show conclusively that HemAT-Hs mediates an aerophobic response in H. salinarum, whereas HemAT-Bs controls an aerophilic response in B. subtilis [34]. The evolutionary range of the HemATs is quite large. Additional HemATs from the fresh-water bacterium Caulobacter crescentus and the facultatively alkaliphilic bacterium Bacillus halodurans have already been demonstrated to contain heme (Fig. 8) [35,36].

3.4.1.2. Absorption Spectra and Heme-iron Coordination The full-length HemAT-Hs and HemAT-Bs proteins, as well as their isolated hemebinding domains, have been overproduced and purified [34,35,171]. Their deoxy, oxy, and carbonmonoxy forms showed characteristic myoglobin absorption spectra, with the deoxy spectra indicating a pentacoordinate heme iron. Studies of N-terminal regions of increasing lengths determined the first 195 residues of the proteins to be essential for stable heme binding [35]. Mutagenesis and spectroscopic studies identified the histidine 123 residue as the proximal axial ligand in each protein [35].

3.4.1.3. Structural Data

Crystal structures of deoxy (FeII ; 2.15-Å resolution) and cyanomet (FeIII CN− ; 2.7-Å resolution) forms of the HemAT-Bs heme-binding domain showed these two forms to consist of nearly identical homodimers of a globin lacking a D-helix [33]. Analytical ultracentrifugation studies of the full-length HemAT-Bs showed that in solution, this protein folds as a highly assymetric rod-shaped dimer with an axial ratio greater than 10 [172]. Biochemical data are as yet unavailable to verify whether the deoxy and cyanomet species examined in the crystals indeed correspond to “on” and “off” states of the heme-binding domain. Even so, the unusual shape of HemAT-Bs does suggest that a slight breakdown in symmetry as that observed between the two forms, e.g., a shift of a heme-pocket phenolic side chain (Tyr 70) in one subunit and minor changes in a helix of that subunit at the dimer interface, might cause a large conformational change [172].

3.4.1.4. Ligand Binding At least two conformers of HemAT-Bs coexist in any liganded state, as indicated by methods such as resonance Raman, Fourrier-transform infra red, and electron paramagnetic resonance spectroscopies [173–175]. This was first noted from studies of O2 binding, which revealed one binding component with a Kd of 1–2 M and koff of 50–80 s−1 , and another with a Kd of 50–100 M and koff of about 2000 s−1 [172,176]. Not surprisingly, normal O2 -binding behavior by this heme pocket is quite intolerant of mutagenesis. Substitutions of a distal tyrosine (Tyr 70, i.e., the B10 residue), a distal threonine (Thr 95),

50

M.A. Gilles-Gonzalez and G. Gonzalez ApPgb 1000 663

MaPgb CaPgb TfPgb

622 1000

T.elongatus GmGCS GsGCS MgGReg SfGReg

1000

978

EcGReg 737

AvGReg 617

1000 647

BbGReg BpeGReg BpaGReg

EAC21812 CvGRegA 1000

EAK60420 936 1000

545 1000

EAJ19547 EAK35357 VvGReg CvGRegB HemAT-Rr

HemAT-MmB

548

521 836

HemAT-MmA McpM 820

582

740

HemAT-At McpB HemAT-Rs

511

HemAT-Na VHb BfGReg AfGReg 698 1000

HemAT-Bs HemAT-Bh 1000

586

HemAT-Ba HemAT-Bc HemAT-Ch

HemAT-Mg HemAT-Hs SWMb

0.1

HRI Rabbit

Fig. 8. Amino acid neighbor-joining phylogenetic tree of globins [40,170]. The horizontal scale bar represents 0.1 substitution per site. The protoglobins are designated by the first letters of their genus and species of origin followed by “Pgb.” For example, ApPgb indicates Aeropyrum pernix Pgb; the other Pgbs originate from Ma = Methanosarcina acetivorans, Ca = Chloroflexus aurantiacus, and Tf = Thermobifida fusca. Proteins that couple a globin to a methyl carrier chemotaxis receptor domain (MCP) are designated by “HemAT-” followed by the first letters of their genus and species of origin. For example, HemAT-Bs indicates the HemAT from Bacillus subtilis; the other HemATs originate from Hs = Halobacterium sp. NRC-1, Ba = Bacillus anthracis,

A Surfeit of Biological Heme-based Sensors

51

and even a histidine on the proximal side that forms a hydrogen bond to a heme propionate, all influence hydrogen bonding to bound O2 [172–175]. Phillips and colleagues made the interesting suggestion that HemAT-Bs might govern graded aerotatic responses under hypoxic as well as aerobic regimes [172]. A dose-response study encompassing a range of O2 concentrations sufficiently broad to test this hypothesis has not been reported.

3.4.2. The Broader Family of Globin-coupled Sensors From phylogenetic studies, it is clear that the HemATs belong to a family of sensors that includes at least 30 members and is not exclusively composed of chemotaxis receptors (Figs. 1 and 8) [40]. In this family, a conserved globin-like heme-binding domain is coupled to a variety of transmitters [35, 40]. These transmitters include domains the GGDEF (Pfam designation DUF1) and EAL (Pfam designation DUF2) domains, which are predicted to be enzymes for regulating second-messenger levels, and several other domains (e.g., Pfam designations: HAMP, STAS) of as yet unknown function. A novel Acidithiobacillus ferrooxidans protein with an AxPDEA1-like phosphodiesterase region (Section 3.1.2) and a GCS heme-binding domain has already been demonstrated to bind heme [35].

3.4.3. Is the Eukaryotic Initiation Factor 2 Alpha Kinase a Globin-coupled Sensor? 3.4.3.1. Physiological Role Eukaryotic initiation factor 2 alpha kinases (eIF-2 kinases) globally control protein synthesis in diverse tissues, including liver, kidney, and testis (Fig. 9) [177,178]. These kinases prevent formation of eIF2:eIF2B dimers by phosphorylating their eIF2 substrate at a serine (Ser 51) (Fig. 9) [179]. Blockage of the eIF2:eIF2B dimerization effectively stops translation initiation, since only this dimer can exchange bound GTP for GDP: a requisite step in recycling eIF2. The eIF2 protein itself also forms

Fig. 8. (Continued) Bh = Bacillus halodurans, Bc = Bacillus cereus, Ch = Carboxydothermus hydrogenoformans, Mm = Magnetospirillum magnetotacticum (two proteins: A and B), Rs = Rhodobacter sphaeroides, Rr = Rhodospirillum rubrum, At = Agrobacterium tumefaciens, Na = Novosphingobium aromaticivorans, and Mg = Magnetococcus sp. MC-1. Globin domains occur together with HAMP and MCP domains in HemAT-At, HemAT-Na, HemAT-Mg, and the Caulobacter crescentus McpB and McpM proteins. Proteins that couple a globin to GGDEF and/or EAL domains, for second-messenger regulation, are designated by the first letters of the genus and species of origin followed by “Greg.” For example, EcGReg indicates Escherichia coli GReg; the other GRegs originate from Bb = Bortedella bronchiseptica, Bpa = Bordetella parapertussis, Bpe = Bordetella pertussis, Av = Azotobacter vinelandii, Sf = Shigella flexneri 2a str.301, Af = Acidithiobacillus ferrooxidans, Bf = Burkholderia fungorum, and Vv = Vibrio vulnificus. Proteins that couple a globin to an unidentified domain carry the more general designation of GCS. Additional abbreviations are: Mb = myoglobin; HRI = heme-regulated inhibitor of translation; Fhb = flavohemoglobin; Nhb = neuroglobin; TrHb = truncated Hb; Lb = leghemoglobin.

52

M.A. Gilles-Gonzalez and G. Gonzalez

GTP

60 S ribosomal subunit

eIF2 Translation initiated on 40 S subunit

GDP

+

Translation continued on 80 S ribosome

Pi

HRI

P

(B)

eIF2

P

eIF2B

(A)

GDP

eIF2

GTP exchange GDP

eIF2B

GTP

+

GDP

eIF2 eIF2B

GDP

eIF2

eIF2

EIF2B

eIF2B does not dissociate from phosphorylated eIF2 substrate (cannot turnover)

Fig. 9. Biochemical roles of HRI as an eIF-2 kinase for controlling globin translation. During translation in eukaryotic cells, the initiation factor 2, i.e., eIF-2, expends a GTP to mediate binding of the charged initiator t-RNA to the 40 S ribosomal subunit. The fate of the GDP-bound eIF-2 is controlled by kinases specific for its -subunit. In erythroid cells, the eIF-2 kinase is HRI. (A) When HRI senses that the cell cannot support globin production, it becomes an active kinase that phosphorylates the -subunit of eIF-2 bound to GDP; this inhibits a crucial recycling of eIF-2 and brings translation to a halt. (B) When HRI senses that the cell can support globin production, the HRI kinase is switched off, and it cannot phosphorylate eIF-2. The eIF2B protein interacts with the unphosphorylated GDP-bound eIF-2 and catalyzes an exchange of GDP for GTP in this protein; the active GTP-bound eIF-2 supports another round of globin mRNA translation initiation.

a homodimer as part of its own activation mechanism, and its dimerization requires multiple autophosphorylations [180]. The heme-regulated inhibitor (HRI) is an eIF-2 kinase found exclusively in red blood cells. By the time developing erythrocytes fully mature, their hemoglobin levels must reach nearly 5 mM. Yet, even a small surplus of free heme is toxic, and likewise, a buildup of apo-globin leads to protein aggregates that damage erythrocytes and inhibit their development. Thus, in these cells, the production of globin - and -chains must closely match the synthesis of protoporphyrin IX and the availability of iron. The erythroid specific eIF-2 kinase HRI is thought to link the translation of globin mRNAs to the availability of heme [181,182]. Free heme, i.e., heme not bound protein, is barely soluble and exists in cells only in miniscule (submicromolar) concentrations. At higher concentrations, heme tends to associate with any hydrophobic surface. Moreover, heme, by catalyzing formation of reactive oxygen species, is so highly toxic that it is exploited as a cytotoxic agent in cancer treatment [183]. Any effect of hemin on a protein that requires concentrations beyond 1 M should be considered strictly nonphysiological. Unfortunately, much of the “regulation” reported for HRI is based on hemin concentrations around 10 M [184,185]. Significantly, all of the inhibitory effects attributed to heme are irreversible, i.e., activity is not restored by removing the heme.

A Surfeit of Biological Heme-based Sensors

53

3.4.3.2. Heme Location and Coordination Chen, Matts, and their colleagues demonstrated, by applying gentle affinity-purification methods, that HRI contains a stably bound heme [186,187]. In particular, they showed HRI to have the characteristic absorption of a protein with low-spin ferrous heme iron: a Soret band at 424 nm, a -peak at 530 nm, and an -peak more intense than the -peak at around 560 nm [186,187]. Addition of CO produced a carbonmonoxy spectrum typical of myoglobins, whereas exposure to NO yielded a pentacoordinate complex with a spectrum resembling that of sGC [187]. Formation of the pentacoordinate NO-bound complex was successfully verified by electron-spin resonance spectroscopy [188]. Attempts to prepare a stable O2 complex of HRI have so far been futile because of a rapid oxidation to the ferric form on exposure to O2 [189]. Each 626-residue monomer of HRI is claimed to possess two heme-binding sites: a stable site in the N-terminal domain (residues 1–154) and an unstable site in the middle of the C-terminal kinase region (residues 232–420) [186,190]. The heme affinity of this unstable “site” (Kd ∼ 5 M), however, is too low for it to have any physiological significance. The first quarter of HRI, containing the true heme-binding site, is homologous to the heme-binding domains of the GCSs and groups phylogenetically with sperm-whale myoglobin (Personal communication from T. Freitas, Fig. 8). The residues of this domain that coordinate axially to the heme iron are not yet known. In alignments with globins, the proximal axial ligand of rabbit HRI best corresponds to histidine 81, but this residue is very near to two other histidines (His 76 and 79). Attempts to disrupt heme binding by mutagenizing these positions have been unsucessful. So far, no satisfactory role has been found for HRI’s resident heme. Since holo-HRI is active in the absence of added hemin, the Kd for binding of heme to its genuine site must be vastly lower than the concentrations at which the hemin inhibition is observed [186]. Thus, the stable binding site for heme cannot be involved in hemin detection. Might the resident heme instead mediate regulation by gaseous ligands?

3.4.3.3. Regulation by Heme Ligands? Matts and colleagues have shown that concentrations of the NO donor 6-(2-hydroxyl1-methyl-2-nitrosohydrazino)-N -methyl-1-hexanamine (NOC-9) over 100 M enhance the enzymatic activity of HRI, whereas levels of CO nearing one atmosphere (∼1 mM) inhibit this activity [187]. These results led them to suggest that HRI may be a hemebased sensor for control of protein synthesis by diffusible gases. Since these effects of ligands on HRI activity are manifested at levels of the ligands that are extraordinarily high for any physiological processes and well above the saturation limits for the heme, it remains unclear whether HRI can specifically respond to binding of NO or CO at a heme center. A response to O2 can also not be entirely ruled out at this stage.

3.4.4. The Ancestor of Hemoglobin A particularly exciting development in the study of globin-coupled sensors was the discovery of two globin-sized, stand-alone, archaeal heme proteins that are clearly related to the GCSs (Fig. 8) [170]. These globins from Aeropyrum pernix (ApPgb) and Methanosarcina acetivorans (MaPgb) fit the characteristics of the predicted ancestor of mammalian hemoglobins, i.e., the protoglobins. They demonstrably bind O2 , CO, and NO, but they are rapidly oxidized by O2 [170]. They feature a proximal histidine (His F8), a distal cysteine (Cys E19), and a distal tyrosine (Tyr B10). The cysteine E19 and

54

M.A. Gilles-Gonzalez and G. Gonzalez

tyrosine B10 residues are postulated to be ancient adaptations for H2 S binding and NO detoxification, respectively, that became lost from modern O2 -binding hemoglobins.

3.5. Guanylyl Cyclase and its Microbial Relatives 3.5.1. Subunit and Domain Organizations of sGC The soluble guanylyl cyclase (sGC) from mammals was celebrated for nearly 30 years before recognizing its bacterial ancestry [2,15,191–193]. The mammalian sGC is a heterodimeric enzyme composed of two homologous subunits,  and , with the 1 and 1 isoforms being the ones most commonly encountered [194]. Nevertheless, sGC contains only one heme [15]. The histidine 105 residue of the 1 -subunit supplies the only axial coordination to the heme iron [195,196]. Each sGC subunit is now thought to have the following organization: (i) an N-terminal domain that binds the heme in the 1 -subunit but apparently binds no prosthetic group in the 1 -subunit; (ii) a middle domain that contains the main determinants of dimerization; (iii) a C-terminal domain that has the catalytic region [193]. The latter domain is the one most conserved between the two subunits.

3.5.2. Regulation by NO and CO The sGC protein responds to NO and CO by synthesizing the second messenger 3 5 cyclic guanosine monophosphate (cGMP) from GTP [2,15,191–193]. This enzymatic activity is greatly enhanced if specific allosteric potentiators, such as the synthetic organic molecule 3-(5 hydroxymethyl-2 -furyl)-1-benzylindazole (YC-1) are supplied along with NO or CO [44,197,198]. YC-1 and other molecules, such as pyrazolopyridine (BAY41-2272), are thought to interact with the 1 -subunit of sGC [197]. Physiological potentiators of NO and CO regulation reminiscent of these synthetic molecules have not been found. The cGMP produced by sGC acts on many downstream effectors, including protein kinases, ion channels, and cyclic-nucleotide phosphodiesterases that directly control vascular smooth muscle tone, neurotransmission, and many other processes [193,199]. For example, the impotence remedy sildenafil (Viagra) is a specific inhibitor of a cGMP-regulated cGMP phosphodiesterase [200]. Because of the initially identified functions of sGC in mammals, this protein was long supposed to occur exclusively in higher animals and was viewed as an important but singular case of a signal transducer that responds to gaseous ligands. In recent years, however, the discovery of the other heme-based sensor families, localization of the sGC heme-binding region to the first 385 residues of the 1 -subunit, and the refusal of this fragment to yield to crystallization attempts, inspired a hunt for sGC microbial relatives [37,196,201].

3.5.3. A Bacterial Origin for sGC? 3.5.3.1. A Heme-NO Binding Sequence in Microbial Proteins By searching the National Center for Biotechnology Information (NCBI) database of nonredundant protein sequences for homology to the sGC heme-binding region, Aravind and colleagues found a broad family of proteins with the sequence characteristics predicted for relatives of sGC [41]. Their findings suggest a bacterial origin for sGC.

A Surfeit of Biological Heme-based Sensors

55

In particular, many bacterial genomes, but no archaeal genomes, were found to encode regions homologous to the N-terminal domains of the sGC subunits (approximately residues 1–185). Whereas the eukaryotic “heme-NO binding” (HNOB) domains are coupled exclusively to guanylyl-cyclase-like regions, the bacterial HNOB regions are found together with: methyl-carrying chemotaxis receptor regions like the ones in HemATs, protein-histidine kinase regions like the ones in FixLs, and at least two other unidentified activities (Fig. 1) [41].

3.5.3.2. Not all Proteins with an HNOB Contain Heme Although the name HNOB implies heme binding, the -subunit of sGC does not bind heme [196,201], and at least two bacterial “HNOB proteins,” one from Rhodobacter sphaeroides (Rhsp22958463) and the other from Magnetococcus sp. (Mcsp 22999020), lack the conserved histidine residue predicted to be the proximal heme-iron axial ligand [41]. It is possible that these microbial proteins bind heme at a different site or without coordinating to the heme iron. If they do not bind heme, then another possibility still is that they have functions quite different from their heme-binding counterparts: possibly in responding to other physiological signals.

3.5.3.3. Not all Proteins With an HNOB have an Attached Transmitter Some HNOB genes encode just a short heme-binding region [41]. These shorter polypeptides may function as stand-alone heme proteins or as the regulatory subunits of multisubunit sensors. If they represent subunits of larger sensors, their roles within those proteins will remain a fundamental question, regardless of how much can be learned about them in isolation.

3.5.3.4. Current Views of sGC Regulation For an inclusive hypothesis about sGC regulatory mechanisms that considers this sensor’s responses to small-molecule stimulants as well as to ligands, see the chapter in this volume on guanylyl cyclases by Sousa and colleagues. NO binding to sGC causes rupture of the Fe His coordination, resulting in a pentacoordinate nitrosyl species. The formation of pentacoordinate heme, which was long assumed to be the basis for NO sensing, is not required for sGC activation [44].

3.5.3.5. Proven Heme-binding HNOB Proteins from Bacteria Two different research groups identified heme-containing bacterial HNOB proteins and renamed the class “heme nitric-oxide/oxygen sensors” (H-NOX) and “sensors of nitric oxide” (SONO) because of the unusually high affinities of individual proteins for either O2 or NO [202–205]. It is important to note the following: • Although CO is not entertained as a regulatory ligand for these proteins, they all bind CO; • A specific in vitro response of an activity to O2 or NO is not yet demonstrated for any protein of this class. As discussed earlier under Section 2, discrimination requires switching of an activity following binding of ligand. It is important to measure switching directly and not to presume it solely from ligand-binding data.

56

M.A. Gilles-Gonzalez and G. Gonzalez

3.6. Gas-responsive Nuclear Receptors 3.6.1. Physiological Roles of the Drosophila E75 Protein In 2005, Krause and colleagues reported a discovery of heme in the Drosophila nuclear receptor E75 [19]. This was quite exciting, not only for the field of heme-based sensors, but also for the area of nuclear receptors, which is currently the subject of intense pharmacological interest. Although many of these receptors are known from genetic and physiological studies to be critical to development, metabolism, and reproduction, most of these proteins are currently classified as “orphan receptors” because their ligands stubbornly defy identification. Of the nearly 20 known Drosophila nuclear receptors, ligands have been recognized for only two: the ecdysone receptor EcR and the E75 proteins [19,206]. Both participate in a common pathway that additionally involves a retinoid-like nuclear receptor called ultraspiracle (USP) and another receptor called HR3 [207–210]. Specifically, one of the responses of a EcR:USP heterodimer to a pulse of the insect hormone ecdysone is to activate the expression of the gene that encodes the E75 receptor [207]. An E75:HR3 couple then triggers a cascade of events leading to sweeping lifestyle changes for the insect such as larval molts and metamorphoses [211,212].

3.6.2. Domain Organization All nuclear receptors contain at least two conserved domains: one that binds DNA and another that senses a ligand [206]. This organization is quite reminiscent of NPAS2 and CooA, except that the nuclear-receptor domains belong to evolutionary families quite distinct from bHLH-PAS and CRP [17,18]. In the nuclear receptors, the DNA-binding domain (called DBD; Pfam designation ZnF) belongs to the zinc-finger class [213,214]. The ligand-binding region (called LBD; Pfam designation HOLI) also indicates a conserved fold [42]. This region participates in interactions of the receptor, not only with its ligand, but also with its regulatory partners. A signal-binding pocket is identifiable from the many available crystal structures of LBDs with their ligands [42].

3.6.3. Heme Characteristics Krause and colleagues overproduced the E75 LBD region (E75341−602 ) in E. coli and discovered it to contain heme from its red color, characteristic heme protein absorption, and binding of a cofactor with the theoretical mass of heme [19]. For both the FeII and FeIII unliganded states, the absorption spectra suggested a low-spin hexacoordinate heme iron with both of its axial ligands contributed by residues from the protein. Therefore, like EcDos, NPAS2, and CooA, the E75 protein features a displaceable residue (DR) that probably participates in its switching mechanism [7,17,18,31,54,55,106]. Attempts to identify the axial ligands to the heme iron by mutagenesis have so far mostly served to indicate that the heme pocket of E75 is extraordinarily sensitive to changes of the protein sequence, including amino acid substitutions in the DNA-binding domain [19,215]. Electron paramagnetic resonance studies confirmed the suggestion, from the absorption spectra, that the FeIII state contains low-spin and hexacoordinate heme iron; these studies suggest that in the FeIII state, the ligation of the iron atom is heterogenous,

A Surfeit of Biological Heme-based Sensors

57

and one of the side chains involved in this coordination is a cysteine thiolate [215]. On the basis of the absorption spectra, E75 stably binds CO [19]. By contrast, the spectra resulting from exposure to NO show that binding of this ligand causes the rupture of both iron-coordination bonds from the protein and the generation of a pentacoordinate nitrosyl heme.

3.6.4. Signal Transduction In vitro, the FeIII state of E75 could not stabilize a 19-amino-acid peptide related to DHR3, whereas the FeII state raised this peptide’s melting temperature by approximately 3 C, and the FeII CO state by about 6 C [19]. In vivo, NO appeared to relieve a supression, imposed by unliganded E75, on the expression of HR3 target genes [19]. Is the signal for E75 the oxidation state of the heme iron, a heme ligand, or both? Although this question is not yet fully anwered, the data strongly argue that E75 is a heme-based sensor, and that such sensors can govern lifestyle changes, from the symbioses of bacteria to the metamorphoses of insects.

3.7. Expected Heme-GAF Involvement in Mycobacterium tuberculosis Latency 3.7.1. Physiological Roles of the DevS and DevR Proteins Latent infections with Mycobacterium tuberculosis can persist for decades, and currently afflict one-third of the human population [216]. Such infections represent an enormous reservoir of this pathogen that is key to its success. An important breakthrough in M. tuberculosis research was the relative development in 1996 of an in vitro model of dormancy that induces bacterial stasis with a “hypoxic shiftdown” [217]. Though this model suggests that the latency of M. tuberculosis may be linked to its adaptations to hypoxia, the biochemical trigger of the dormancy remains unknown; involvements of NO and O2 in this process have been hypothesized [218–220]. In the in vitro model, over 40 genes are induced, many of which are demonstrably required for the persistence of M. tuberculosis infections in vivo [221–223]. The transcription factor DevR, a response regulator of the two-component class, controls the expression of this large suite of genes [222–226]. The status of DevR is in turn governed by DevS: a protein-histidine kinase that is produced jointly with DevR and recognizes DevR as its substrate [222,224,226,227]. Thus, the DevS/DevR system is quite reminiscent of FixL/FixJ.

3.7.2. DevS is a Heme-containing Protein-histidine Kinase Djordjevic and colleagues found the DevS protein to contain heme [14]. This discovery of heme in a protein that possesses a C-terminal protein-histidine-kinase region like the one in FixL strongly suggested that the heme is coupled to the kinase (Fig. 10). Though the domain organization of DevS is reminiscent of FixL, an interesting difference between the two proteins is that FixL binds heme in a PAS domain, whereas DevS binds heme in the first of two N-terminal GAF domains (Fig. 10) [14,29]. Until the

58

M.A. Gilles-Gonzalez and G. Gonzalez 1

DevS

100

200

GAF

GAF

BjFixL

PAS

PAC

PAS

PAC

HisKA_3

HisKA

HATPase_c

HATPase_c

Fig. 10. Schematic representation of the M. Tuberculosis DevS domain organization and its comparison to B. japonicum FixL. Both DevS and BjFixL feature a C-terminal protein-histidine kinase region (HisKA plus HATPase_c regions). In DevS, this region is preceded by two tandem GAF domains, whereas in BjFixL, it is preceded by two tandem PAS domains. The heme in DevS is held in the first GAF domain, whereas the heme of BjFixL is held in the second PAS domain. Domain nomenclatures, symbols, and protein organizations are according to the simple modular architecture research tool (SMART) from the European Molecular Biology Laboratory [57,58].

nearly simultaneous reports of a heme cofactor in DevS and of a nonheme iron center in the NO sensor NorR, GAF domains were thought to be dedicated to binding of cyclic nucleotides and photopigments [14,228–231]. The observation that a similar domain can alternatively bind heme, iron–sulfur clusters, or cyclic nucleotides is not unprecedented and was first made for CooA, which is evolutionarily related to both the iron–sulfur binding protein Fnr and the cAMP binding protein CRP [17,31]. Until recently, kinase reactions had been examined only for DevS variants lacking a sensory function, due to difficulties in obtaining the full-length holo-protein in soluble form [12,13]. Both DevS and a close homolog of this protein called DosT are now known to sense O2 exclusively, with this ligand binding more avidly to DevS (Kd = 3 M) than to DosT (Kd = 26 M) [233].

3.8. An Array of Heme-based Sensors Many organisms employ a variety of heme-based sensors. The genome of C. crescentus, for example, encodes proteins corresponding to a FixL, a HemAT, and a HNOB. R. rubrum has the CooA protein, a globin-coupled chemotaxis receptor (HemAT-Rr), and several heme-PAS proteins with their heme-binding domains linked to prokaryotic CheY-like regions. Azorhizobium caulinodans possesses a FixL and a heme-PAS-coupled PDEA. Magnetospirillum magnetotacticum has two globin-coupled chemotaxis receptors (HemAT-MmA and HemAT-MmB) and features seven different heme-PAS sensors; the heme-binding domains of the latter signal transducers are coupled to protein-histidine kinases, PDEAs, and a domain (HAMP) of unknown function. Though it will probably take years to decipher the possible physiological roles and interactions of such arrays of heme-based sensors, it is already certain that they form a crucial component of signal transduction in most living organisms.

A Surfeit of Biological Heme-based Sensors

59

4. FUTURE DIRECTIONS 4.1. Full-length Sensors An important task will be to study regulatory mechanisms for physiologically relevant forms of complete heme-based sensors. Given the modularity of these sensors and their frequent partnership with other proteins, their study will require the application of multiple tools, including genetics, physiology, enzymology, and various kinds of spectroscopy. Previously, sensors and other proteins were discovered by their physiological functions and purified from their activities. Their biochemistries were examined with the aim of understanding their mechanistic underpinnings and the reactions relevant to their in vivo functions. Their high-resolution structures were determined after a great deal had been learned about their functions, with care being taken to obtain relevant and rationalizable structures. Now “orphaned” sensors, without an established signal or function, can be discovered primarily from their sequences and purified on the basis of the properties of recombinant tags. Microarray studies may suggest a panoply of potential targets for the sensors. Structures, particularly of domains, are often solved quite early in the study of sensors and may also suggest biological functions. Indirect approaches can serve as guides to possible signal molecules and biochemical activities, keeping in mind that there exist many examples where a domain with a well-characterized function in one organism proved to have a completely different function in another organism. Consequently, today it is more important than ever to avoid the appeal of indirect approaches and focus on directly measuring the physiology and biochemistry of a sensor.

4.2. Conformational Changes Relevant to Regulation Areas that will require some caution include the assignment of regulatory ligands to novel sensors, the identification of the “on” and “off” states, and the determination of regulatory conformational changes. Ligands cannot be assigned to sensors entirely from their in vitro behavior. A ligand that binds to the heme may not cause switching. A ligand that regulates in vitro might never be encountered in vivo, although its effects could inform on the sensing mechanism. States of a sensor that appear to be “regulated” in vitro might simply be inactive. Ligands that trigger a spectroscopic change may not necessarily cause switching. Alternatively, switching (Section 2.2) might be incompatible with the design of some recombinant fragments or the conditions required for some types of spectroscopy, such as low temperatures, high salt, crystalline states, or oxidized species. As one tackles the study of these exciting proteins with increasingly sophisticated tools, it will be important to keep on hand the simple ideas that the essence of a heme-based sensor is to couple a heme-binding region to an activity, and its purpose is to regulate a physiological adaptation.

ACKNOWLEDGMENTS The authors thank Eduardo Sousa, Jason Tuckerman, Elhadji Dioum, and Olivier Belzile for their comments, Tracey Freitas for a phylogenetic analysis of the HRI protein,

60

M.A. Gilles-Gonzalez and G. Gonzalez

and the U.S. Public Health Service Grant HL-64038, the NRI Competitive Grants Program/USDA Award 2002-35318-12515, and Welch Foundation Grant No. I-1575 for support.

REFERENCES [1] Gilles-Gonzalez, M.A., Gonzalez, G.M., Perutz, F. et al. (1994) Biochemistry. 33, 8067–8073. [2] Craven, P.A. and DeRubertis, F.R. (1978) J. Biol. Chem. 253, 8433–8443. [3] Gilles-Gonzalez, M.A., Ditta, G.S. and Helinski, D.R. (1991) Nature. 350, 170–172. [4] Dunham, C.M., Dioum, E.M., Tuckerman, J.R. et al. (2003) Biochemistry. 42, 7701–7708. [5] Gilles-Gonzalez, M.A., Caceres, A.I., Sousa, E.H.S. et al. (2006) J. Mol. Biol. 360, 80–89. [6] Chang, A.L., Tuckerman, J.R., Gonzalez, G. et al. (2001) Biochemistry. 40, 3420–3426. [7] Delgado-Nixon, V.M., Gonzalez, G. and Gilles-Gonzalez, M.A. (2000) Biochemistry 39, 2685–2691. [8] Schmidt, A.J., Ryjenkov, D.A. and Gomelsky, M. (2005) J. Bacteriol. 187, 4774–4781. [9] Takahashi, H. and Shimizu, T. (2006) Chem. Lett. 35, 970–971. [10] Chefalo, P.J., Oh, J., Rafie-Kolpin, M. et al. (1998) Eur. J. Biochem. 258, 820-830. [11] Uma, S., Yun, B.G. and Matts, R.L. (2001) J. Biol. Chem. 276, 14875–14883. [12] Saini, D.K., Malhotra, V. and Tyagi, J.S. (2004) FEBS Lett. 565, 75–80. [13] Saini, D.K., Malhotra, V., Dey, D. et al. (2004) Microbiology. 150, 865–875. [14] Sardiwal, S., Kendall, S.L., Movahedzadeh, F. et al. (2005) J. Mol. Biol. 353, 929–936. [15] Stone J.R. and Marletta, M.A. (1994) Biochemistry. 33, 5636–5640. [16] Duff, S.M., Wittenberg, J.B. and Hill, R.D. (1997) J. Biol. Chem. 272, 16746–16752. [17] Shelver, D.R., Kerby, L., He, Y. and Roberts, G.P. (1997) Proc. Natl Acad. Sci. U.S.A. 94, 11216–11220. [18] Dioum, E.M., Rutter, J., Tuckerman, J.R. et al. (2002) Science. 298, 2385–2387. [19] Lam, J., Reinking, M.M., Pardee, K. et al. (2005) Cell. 122, 195–207. [20] Arredondo-Peter, R., Hargrove, M.S., Sarath, G. et al. (1997) Plant Physiol. 115, 1259–1266. [21] Couture, M.T., Das, K.P., Savard, Y. et al. (2000) Eur. J. Biochem. 267, 4770–4780. [22] Scott, N.L. and Lecomte, J.T. (2000) Protein Sci. 9, 587–597. [23] Burmester, T., Weich, B., Reinhardt, S. and Hankeln, T. (2000) Nature. 407, 520–523. [24] Burmester, T., Ebner, B., Weich, B. and Hankeln, T. (2002) Mol. Biol. Evol. 19, 416–421. [25] Trent, J.T., III and Hargrove, M.S. (2002) J. Biol. Chem. 277, 19538–19545. [26] Kundu, S.J., Trent, T. III and Hargrove, M.S. (2003) Trends Plant Sci. 8, 387–393. [27] Wittenberg, J.B., Bolognesi, M., Wittenberg, B.A. and Guertin, M. (2002) J. Biol. Chem. 277, 871–874. [28] Tomita, T., Gonzalez, G., Chang, A.L. et al. (2002) Biochemistry. 41, 4819–4826. [29] Gong, W., Hao, B., Mansy, S.S. et al. (1998) Proc. Natl Acad. Sci. U.S.A. 95, 15177–15182. [30] Weichsel, A.J., Andersen, F.D., Champagne, E. et al. Nat. Struct. Biol. 5, (1998) 304–309. [31] Lanzilotta, W.N., Schuller, D.J., Thorsteinsson, M.V. et al. (2000) Nat. Struct. Biol. 7, 876–880. [32] Pesce, A., Couture, M., Dewilde, S. et al. (2000) EMBO J. 19, 2424–2434. [33] Zhang W. and Phillips G.N., Jr. (2003) Structure. 11, 1097–1110. [34] Hou, S., Larsen, R.W., Boudko, D. et al. (2000) Nature. 403, 540–544. [35] Hou, S., Freitas, T., Larsen, R.W. et al. (2001) Proc. Natl Acad. Sci. U.S.A. 98, 9353–9358. [36] Hou, S., Belisle, C., Lam, S. et al. (2001) Extremophiles. 5, 351–354. [37] Gilles-Gonzalez, M.A. (2001) IUBMB Life. 51, 165–173. [38] Gilles-Gonzalez, M.A. and Gonzalez, G. (2004) J. Appl. Physiol. 96, 774–783.

A Surfeit of Biological Heme-based Sensors

61

[39] Gilles-Gonzalez, M.A. and Gonzalez, G. (2005) J. Inorg. Biochem. 99, 1–22. [40] Freitas, T.A., Hou, S. and Alam, M. (2003) FEBS Lett. 552, 99–104. [41] Iyer, L.M., Anantharaman, V. and Aravind, L. (2003) BMC Genomics. 4, 5–12 (Pub Med ID: 12590654). [42] Renaud, J.P. and Moras, D. (2000) Cell. Mol. Life Sci. 57, 1748–1769. [43] Soupene, E., Foussard, M., Boistard, P. et al. (1995) Proc. Natl Acad. Sci. U.S.A. 92, 3759–3763. [44] Friebe, A., Schultz, G. and Koesling, D. (1996) EMBO J. 15, 6863–6868. [45] Lawson, D.M., Stevenson, C.E., Andrew, C.R. and Eady, R.R. (2000) EMBO J. 19, 5661–5671. [46] Ballou, D.P., Zhao, Y., Brandish, P.E. and Marletta, M.A. (2002) Proc. Natl Acad. Sci. U.S.A. 99, 12097–12101. [47] Roberts, G.P., Thorsteinsson, M.V., Kerby, R.L. et al. (2001) Prog. Nucleic Acid Res. Mol. Biol. 67, 35–63. [48] He, Y., Gaal, T., Karls, R. et al. (1999) J. Biol. Chem. 274, 10840–10845. [49] Puranik, M., Nielsen, S.B., Youn, H. et al. (2004) J. Biol. Chem. 279, 21096–21108. [50] Gong, W., Hao, B. and Chan, M.K. (2000) Biochemistry. 39, 3955–3962. [51] Hao, B., Isaza, C., Arndt, J. et al. (2002) Biochemistry. 41, 12952–12958. [52] Key, J. and Moffat, K. (2005) Biochemistry. 44, 4627–4635. [53] Miyatake, H., Masahiro, M., Park, S.Y. et al. (2000) J. Mol. Biol. 301, 415–431. [54] Park, H.J., Suquet, C., Satterlee, J.D. and Kang, C. (2004) Biochemistry. 43, 2738–2746. [55] Kurokawa, H., Lee, D.S., Watanabe, M. et al. (2004) J. Biol. Chem. 279, 20186–20193. [56] Youn, H., Kerby, R.L., Conrad, M. and Roberts, G.P. (2004) J. Bacteriol. 186, 1320–1329. [57] Letunic, I., Goodstadt, L., Dickens, N.J. et al. (2002) Nucleic Acids Res. 30, 242–244. [58] Letunic, I., Copley, R.R., Schmidt, S. et al. (2004) Nucleic Acids Res. 32, Database issue D142–144. [59] Virts, E.L., Stanfield, S.W., Helinski, D.R. and Ditta, G.S. (1988) Proc. Natl Acad. Sci. U.S.A. 85, 3062–3065. [60] David, M., Daveran, M.L., Batut, J. et al. (1988) Cell 54, 671–683. [61] De Bruijn, F.J., Rossbach, S., Bruand, C. and Parrish, J.R. (2006) Environ Microbiol. 8, 1371–1381. [62] Anthamatten D. and Hennecke, H. (1991) Mol. Gen. Genet. 225, 38–48. [63] Mesa S., Bedmar, E.J., Chanfon, A. et al. (2003) J. Bacteriol. 185, 3978–3982. [64] Sciotti, M.A., Chanfon, A., Hennecke, H. and Fischer, H.M. (2003) J. Bacteriol. 185, 5639–5642. [65] Robles, E.F., Sanchez, C., Bonnard, N. et al. (2006) Biochem. Soc. Trans. 34, 108–110. [66] Kaminski, P.A. and Elmerich, C. (1991) Mol. Microbiol. 5, 665–673. [67] Lopez, O., Morera, C., Miranda-Rios, J. et al. (2001) J. Bacteriol. 183, 6999–7006. [68] Boesten, B. and Priefer, U.B. (2004) Microbiology 150, 3703–3713. [69] Crosson, S., McGrath, P.T., Stephens, C. et al. (2005) Proc. Natl Acad. Sci. U.S.A. 102, 8018–8023. [70] Garnerone, A.M., Cabanes, D., Foussard, M. et al. (1999) J. Biol. Chem. 274, 32500–32506. [71] Gilles-Gonzalez, M.A. and Gonzalez, G. (1993) J. Biol. Chem. 268, 16293–16297. [72] Gilles-Gonzalez, M.A., Gonzalez, G. and Perutz, M.F. (1995) Biochemistry. 34, 232–236. [73] Tuckerman, J.R., Gonzalez, G. and Gilles-Gonzalez, M.A. (2001) J. Mol. Biol. 308, 449–455. [74] Tuckerman, J.R., Gonzalez, G., Dioum, E.M. and Gilles-Gonzalez, M.A. (2002) Biochemistry. 41, 6170–6177. [75] Sousa, E.H., Gonzalez, G. and Gilles-Gonzalez, M.A. (2005) Biochemistry. 44, 15359–15365. [76] Lois, A.F., Weinstein, M., Ditta, G.S. and Helinski, D.R. (1993) J. Biol. Chem. 268, 4370–4375.

62

M.A. Gilles-Gonzalez and G. Gonzalez

[77] Akimoto, S., Tanaka, A., Nakamura, K. et al. (2003) Biochem. Biophys. Res. Commun. 304, 136–142. [78] Saito, K., Ito, E., Hosono, K. et al. (2003) Mol. Microbiol. 48, 373–383. [79] Nakamura, H., Kumita, H., Imai, K. et al. (2004) Proc. Natl Acad. Sci. U.S.A. 101, 2742–2746. [80] Stock, J.B., Ninfa, A.J. and Stock, A.M. (1989) Microbiol. Rev. 53, 450–490. [81] Stock, A.M., Robinson, V.L. and Goudreau P.N. (2000) Annu. Rev. Biochem. 69, 183–215. [82] Balland, V., Bouzhir-Sima, L., Kiger, L. et al. (2005) J. Biol. Chem. 280, 15279–15288. [83] Olson, J.S. and Phillips, G.N. (1997) J. Biol. Inorg. Chem. 2, 544–552. [84] Dou, Y., Olson, J.S., Wilkinson, A.J. and Ikeda-Saito, M. (1996) Biochemistry. 35, 7107–7113. [85] Winkler, W.C., Gonzalez, G., Wittenberg, J.B. et al. (1996) Chem. Biol. 841–850. [86] Mansy, S.S., Olson, J.S. Gonzalez, G. and Gilles-Gonzalez, M.A. (1998) Biochemistry. 37, 12452–12457. [87] Miksovska, J., Suquet, C., Satterlee, J.D. and Larsen, R.W. (2005) Biochemistry. 44, 10028–10036. [88] Balland, V., Bouzhir-Sima, L., Anxolabéhère-Mallart, E. et al. (2006) Biochemistry. 45, 2072–2084. [89] Liebl, U., Bouzhir-Sima, L., Negrerie, M. et al. (2002) Proc. Natl Acad. Sci. U.S.A. 99, 12771–12776. [90] Jasaitis, A., Hola, K., Bouzhir-Sima, L. et al. (2006) Biochemistry. 45, 6018–6026. [91] Ross, P., Mayer, R. and Benziman, M. (1991) Microbiol. Rev. 55, 35–58. [92] Ross, P., Mayer, R., Weinhouse, H. et al. (1990) J. Biol. Chem. 265, 18933–18943. [93] Egli, M., Gessner, R.V., Williams, L.D. et al. (1990) Proc. Natl Acad. Sci. U.S.A. 87, 3235–3239. [94] Tal, R., Wong, H.C., Clahoon, R. et al. (1998) J. Bacteriol. 180, 4416–4425. [95] Ausmees, N., Mayer, R., Weinhouse, H. et al. (2001) FEMS Microbiol. Lett. 204, 163–167. [96] Ryjenkov, D.A., Tarutina, M., Moskvin, O.V. and Gomelsky, M. (2005) J. Bacteriol. 187, 1792–1798. [97] Zogaj, X., Nimtz, M., Rohde, M. et al. (2001) Mol. Microbiol. 39, 1452–1463. [98] Simm, R., Morr, M., Kader, A. et al. (2004) Mol. Microbiol. 53, 1123–1134. [99] Romling, U. (2005) Cell. Mol. Life Sci. 62, 1234–1246. [100] Aldridge, P., Paul, R., Goymer, P. (2003) Mol. Microbiol. 47, 1695–1708. [101] Hickman, J.W., Tifrea, D.F. and Harwood, C.S. (2005) Proc. Natl Acad. Sci. U.S.A. 102, 14422–14427. [102] Tamayo, R.A., Tischler, D. and Camilli, A. (2005) J. Biol. Chem. 280, 33324–33330. [103] Chan, C., Paul, R., Samoray, D. et al. (2004) Proc. Natl Acad. Sci. U.S.A. 101, 17084–17089. [104] Christen, M., Christen, B., Folcher, M. et al. (2005) J. Biol. Chem. 280, 30829–30837. [105] Christen, B., Christen, M., Paul, R. et al. (2006) J. Biol. Chem. 281, 32015–32024. [106] Gonzalez, G., Dioum, E.M., Bertolucci, C.M. et al. (2002) Biochemistry. 41, 8414–8421. [107] Sasakura, Y., Hirata, S., Sugiyama, S. et al. (2002) J. Biol. Chem. 277, 23821–23827. [108] Watanabe, M., Kurokawa, H., Yoshimura-Suzuki, T. and Sagami, I. (2004) Eur. J. Biochem. 271, 3937–3942. [109] Amin, N. and Peterkofsky, A. (1995) J. Biol. Chem. 270, 11803–11805. [110] Imamura, R., Yamanaka, K., Ogura, T. et al. (1996) J. Biol. Chem. 271, 25423–25429. [111] Yoshimura-Suzuki, T., Sagami, I., Yokota, N. et al. (2005) J. Bacteriol. 187, 6678–6682. [112] Mendez-Ortiz, M.M., Hyodo, M., Hayakawa, Y. and Membrillo-Hernandez, J. (2006) J. Biol. Chem. 281, 8090–8099. [113] Prigent-Combaret, C., Brombacher, E., Vidal, O. et al. (2001) J. Bacteriol. 183, 7213–7223. [114] Brombacher, E., Dorel, C., Zehnder, A.J. and Landini, P. (2003) Microbiology 149, 2847–2857.

A Surfeit of Biological Heme-based Sensors [115] [116] [117] [118] [119] [120] [121] [122] [123] [124] [125] [126] [127] [128] [129] [130] [131] [132] [133] [134] [135] [136] [137] [138] [139] [140] [141] [142] [143] [144] [145] [146] [147] [148] [149] [150] [151] [152] [153] [154] [155] [156] [157]

63

Green, J., Scott, C. and. Guest, J.R (2001) Adv. Microb. Physiol. 44, 1–34. Beinert, H. and Kiley, P.J. (1999) Curr. Opin. Chem. Biol. 3, 152–157. Zhou, Y.D., Barnard, M. and Tian, H. et al. (1997) Proc. Natl Acad. Sci. U.S.A. 94, 713–718. Hogenesch, J.B., Gu, Y.Z., Jain, S. and Bradfield, C.A. (1998) Proc. Natl Acad. Sci. U.S.A. 95, 5474–5479. Reick, M. Garcia, J.A., Dudley, C. and McKnight, S.L. (2001) Science 293, 506–509. Gu, Y.Z., Hogenesch, J.B. and Bradfield, C.A. (2000) Ann. Rev. Pharmacol. Toxicol. 40, 519–561. Denison, M.S. and Nagy, S.R. (2003) Ann. Rev. Pharmacol. Toxicol. 43, 309–334. Semenza, G.L. (2004) Physiology 19, 176–182. Boehning, D. and Snyder, S.H. (2002) Science 298, 2339–2340. Maines, M.D. and Gibbs, P.E. (2005) Biochem. Biophys. Res. Commun. 338, 568–577. Boehning, D., Moon, C., Sharma, S. et al. (2003) Neuron 40, 129–137. Maines, M.D. (1997) Ann. Rev. Pharmacol. Toxicol. 37, 517–554. Maines, M.D. (2000) Cell Mol. Biol. 46, 573–585. Baranano, D.E. and Snyder, S.H. (2001) Proc. Natl Acad. Sci. U.S.A. 98, 10996–11002. Dudley, C.A., Erbel-Sieler, C., Estill, S.J. et al. (2003) Science 301, 379–383. Kaasik, K. and Lee, C.C. (2004) Nature 430, 467–471. Franken, P., Dudley, C.A., Estill, S.J. et al. (2006) Proc. Natl Acad. Sci. U.S.A. 103, 7118–7123. Wisor, J.P., O’Hara, B.F., Terao, A. et al. (2002) BMC Neurosci. 3, 20. Zhulin, I.B., Taylor, B.L. and Dixon, R. (1997) Trends Biochem. Sci. 22, 331–333. Taylor, B.L. and Zhulin, I.B. (1999) Microbiol. Mol. Biol. Rev. 63, 479–506. Pellequer, J.L., Wager-Smith, K.A., Kay, S.A. and Getzoff, E.D. (1998) Proc. Natl Acad. Sci. U.S.A. 95, 5884–5890. Morais-Cabral, J.H., Lee, A., Cohen, S.L. et al. (1998) Cell 95, 649–655. Bertolucci, C., Ming, L.J., Gonzalez, G. and Gilles-Gonzalez, M.A. (1996) Chem. Biol. 3, 561–566. Monson, E.K., Ditta, G.S. and Helinski, D.R. (1995) J. Biol. Chem. 270, 5243–5250. Amezcua, C.A., Harper, S.M., Rutter, J. and Gardner, K.H. (2002) Structure 10, 1349–1361. Crosson, S., Rajagopal, S. and Moffat, K. (2003) Biochemistry 42, 2–10. Harper, S.M., Neil, L.C. and Gardner, K.H. (2003) Science 301, 1541–1544. Erbel, P.J., Card, P.B., Karakuzu, O. et al. (2003) Proc. Natl Acad. Sci. U.S.A. 100, 15504–15509. Reinelt, S., Hofmann, E., Gerharz, T. et al. (2003) J. Biol. Chem. 278, 39189–39196. Perutz, M.F., Paoli, M. and Lesk, A.M. (1999) Chem. Biol. 6, 291–297. Mukai, M., Nakamura, K., Nakamura, H. et al. (2000) Biochemistry 39, 13810–13816. Uchida, T., Sato, E., Sato, A. et al. (2005) J. Biol. Chem. 280, 21358–21368. Koudo, R., Kurokawa, H., Sato, E. et al. (2005) FEBS J. 272, 4153–4162. Mukaiyama, Y., Uchida, T., Sato, E. et al. (2006) FEBS J. 273, 2528–2539. Rodgers, K.R., Tang, L., Lukat-Rodgers, G.S. and Wengenack, N.L. (2001) Biochemistry 40, 12932–12942. Dutta, R., Qin, L. and Inouye, M. (1999) Mol. Microbiol. 34, 633–640. Rutter, J., Reick, M., Wu, L.C. and McKnight, S.L. (2001) Science 293, 510–514. Shelver, D., Kerby, R.L., He, Y. and Roberts, G.P. (1995) J. Bacteriol. 177, 2157–2163. He, Y., Gaal, T., Karls, R. et al. (1999) J. Biol. Chem. 274, 10840–10845. Aono, S., Nakajima, H., Saito, K. and Okada, M. (1996) Biochem. Biophys. Res. Commun. 228, 752–756. Vogel, K.M., Spiro, T.G., Shelver, D. et al. (1999) Biochemistry 38, 2679–2687. Dhawan, I.K., Shelver, D., Thorsteinsson, M.V. et al. (1999) Biochemistry 38, 12805–12813. Shelver, D., Thorsteinsson, M.V., Kerby, R.L. et al. (1999) Biochemistry 38, 2669–2678.

64

M.A. Gilles-Gonzalez and G. Gonzalez

[158] Aono, S., Ohkubo, K., Matsuo, T. and Nakajima, H. (1998) J. Biol. Chem. 273, 25757–25764. [159] Kubo, M., Inagaki, S., Yoshioka, S. et al. (2006) J. Biol. Chem. 281, 11271–11278. [160] Youn, H., Thorsteinsson, M.V., Conrad, M. et al. (2005) J. Bacteriol. 187, 2573–2581. [161] Komori, H., Satomoto, K., Ueda, Y. et al. (2006) Acta Crystallograph. Sect. F Struct. Biol. Cryst. Commun. 62, 471–473. [162] Inagaki, S., Masuda, C., Akaishi, T. et al. (2005) J. Biol. Chem. 280, 3269–3274. [163] Clark, R.W., Youn, H., Parks, R.B. et al. (2004) Biochemistry 43, 14149–14160. [164] Youn, H., Kerby, R.L. and Roberts, G.P. (2003) J. Biol. Chem. 278, 2333–2340. [165] Youn, H., Kerby, R.L. and Roberts, G.P. (2004) J. Biol. Chem. 279, 45744–45752. [166] Pinkert, J.C., Clark, R.W. and Burstyn, J.N. (2006) J. Biol. Inorg. Chem. 11, 642–665. [167] Clark, R.W., Youn, H., Lee, A.J. et al. (2007) J. Biol. Inorg. Chem. 12, 139–146. [168] Clark, R.W., Lanz, N.D., Lee, A.J. et al. (2006) Proc. Natl Acad. Sci. U.S.A. 103, 891–896. [169] Reynolds, M.F., Parks, R.B., Burstyn, J.N. et al. (2000) Biochemistry 39, 388–396. [170] Freitas, T.A.K., Hou, S., Dioum, E.M. et al. (2004) Proc. Natl Acad. Sci. U.S.A. 101, \1–\2. [171] Piatibratov, M., Hou, S., Brooun, A. et al. (2000) Biochim. Biophys. Acta 1524, 149–154. [172] Zhang, W., Olson, J.S. and Phillips G.N. Jr. (2005) Biophys. J. 88, 2801–2814. [173] Pinakoulaki, E., Yoshimura, H., Yoshioka, S. et al. (2006) Biochemistry 45, 7763–7766. [174] Pinakoulaki, E., Yoshimura, H., Daskalakis, V. (2006) et al. Proc. Natl Acad. Sci. U.S.A. 103, 14796–14801. [175] Yoshimura, H., Yoshioka, S., Kobayashi, K. et al. (2006) Biochemistry 45, 8301–8307. [176] Aono, S., Kato, T., Matsuki, M. et al. (2002) J. Biol. Chem. 277, 13528–13538. [177] Crosby, J.S., Chefalo, P.J., Yeh, I. et al. (2000) Blood 96, 3241–3248. [178] Berlanga, J.J., Herrero, S. and de Haro, C. (1998) J. Biol. Chem. 273, 32340–32346. [179] Sudhakar, A., Ramachandran, A., Ghosh, S. et al. (2000) Biochemistry 39, 12929–12938. [180] Bauer, B.N., Rafie-Kolpin, M., Lu, L. et al. (2001) Biochemistry 40, 11543–11551. [181] Wallis, M.H., Kramer, G. and Hardesty, B. (1980) Biochemistry 19, 798–804. [182] Chen, J.J. and London, I.M. (1995) Trends Biochem. Sci. 20, 105–108. [183] Chakraborti, A.S. (2003) Mol. Cell. Biochem. 253, 49–54. [184] Hronis, T.S. and Traugh, J.A. (1986) J. Biol. Chem. 261, 6234–6238 [185] Hronis, T.S. and Traugh, J.A. (1981) J. Biol. Chem. 256, 11409–11412. [186] Chefalo, P.J., Oh, J., Rafie-Kolpin, M. et al. (1998) Eur. J. Biochem. 258 820–830. [187] Uma, S., Yun, B.G. and R.L. (2001) Matts, J. Biol. Chem. 276, 14875–14883. [188] Igarashi, J., Sato, A., Kitagawa, T. et al. (2004) J. Biol. Chem. 279, 15752–15762. [189] Igarashi, J., Sato, A., Kitagawa, T. et al. (2003) Biochim. Biophys. Acta. 1650, 99–104. [190] Rafie-Kolpin, M. Chefalo, P.J. Hussain, Z. et al. (2000) J. Biol. Chem. 275, 5171–5178. [191] Ignarro, L.J., Barry, B.K., Gruetter, D.Y. et al. (1981) Biochim. Biophys. Acta. 673, 394–407. [192] Lewicki, J.A., Brandwein, H.J., Mittal, C.K. et al. (1982) J. Cyclic Nucleotide Res. 8, 17–25. [193] Krumenacker, J.S., Hanafy, K.A. and Murad, F. (2004) Brain Res. Bull. 62, 505–515. [194] Harteneck, C., Koesling, D., Soling, A. et al. (1990) FEBS Lett. 272, 221–223. [195] Wedel, B., Humbert, P., Harteneck, C. et al. (1994) Proc. Natl Acad. Sci. U.S.A. 91, 2592–2596. [196] Zhao, Y., Schelvis, J.P., Babcock, G.T. and Marletta, M.A. (1998) Biochemistry 37, 4502– 4509. [197] Stasch, J.P., Becker, E.M., Alonso-Alija, C. (2001) et al. Nature 410, 212–215. [198] Lee, Y.C., Martin, E. and Murad, F. (2000) Proc. Natl Acad. Sci. U.S.A. 97, 10763–10768. [199] Moncada, S. (1999) J.R. Soc. Med. 92, 164–169. [200] Boolell, M., Allen, M.J., Ballard, S.A. et al. (1996) Int. J. Impot. Res. 8, 47–52. [201] Wedel, B., Humbert, P., Harteneck, C. et al. (1994) Proc. Natl Acad. Sci. U.S.A. 91, 2592–2596. [202] Pellicena, P., Karow, D.S., Boon, E.M. et al. (2004) Proc. Natl Acad. Sci. U.S.A. 101, 12854–12859.

A Surfeit of Biological Heme-based Sensors [203] [204] [205] [206] [207] [208] [209] [210] [211] [212] [213] [214] [215] [216] [217] [218] [219] [220] [221] [222] [223] [224] [225] [226] [227] [228] [229] [230] [231] [232] [233]

65

Karow, D.S., Pan, D., Tran, R. et al. (2004) Biochemistry 43, 10203–10211. Gray, J.M., Karow, D.S., Lu, H. et al. (2004) Nature 430, 317–322. Nioche, P., Berka, V., Vipond, J. et al. (2004) Science 306, 1550–1553. Thummel, C.S. (1995) Cell 83, 871–877. Segraves, W.A. and Hogness, D.S. (1990) Genes Dev. 4, 204–219. White, K.P., Hurban, P., Watanabe, T. and Hogness, D.S. (1997) Science 276, 114–117. Swevers, L., Ito, K. and Iatrou, K. (2002) J. Biol. Chem. 277, 41637–41644. Hiruma, K. and Riddiford, L.M. (2004) Dev. Biol. 272, 510–521. D’Avino, P.P. and Thummel, C.S. (1998) Development 125, 1733–1745. Dubrovskaya, V.A., Berger, E.M. and Dubrovsky, E.B. (2004) Gene 340, 171–177. Freedman, L.P. and Luisi, B.F. (1993) J. Cell Biochem. 51, 140–150. Meier, C.A. (1997) J. Recept. Signal Transduct. Res. 17, 319–335. de Rosny, E., de Groot, A., Jullian-Binard, C. et al. (2006) Biochemistry 5, 9727–9734. Dye, C., Scheele, S., Dolin, P. et al. (1999) JAMA 282, 677–686. Wayne, L.G. and Hayes, L.G. (1996) Infect. Immun. 64, 2062–2069. Boshoff, H.I., Myers, T.G., Copp, B.R. et al. (2004) J. Biol. Chem. 279, 40174–40184. Schnappinger, D., Ehrt, S., Voskuil, M.I. et al. (2003) J. Exp. Med. 198, 693–704. Shi, L., Jung, Y.J., Tyagi, S. et al. (2003) Proc. Natl. Acad. Sci. U.S.A. 100, 241–246. Malhotra, V., Sharma, D., Ramanathan, V.D. et al. (2004) FEMS Microbiol. Lett. 231, 237–245. Sherman, D.R., Voskuil, M., Schnappinger, D. et al. (2001) Proc. Natl. Acad. Sci. U.S.A. 98, 7534–7539. Voskuil, M.I., Schnappinger, D., Visconti, K.C. et al. (2003) J. Exp. Med. 198, 705–713. Boon, C. and Dick, T. (2002) J. Bacteriol. 184, 6760–6767. Florczyk, M.A., McCue, L.A., Purkayastha, A. et al. (2003) Infect. Immun. 71, 5332–5343. Park, H.D., Guinn, K.M., Harrell, M.I. et al. (2003) Mol. Microbiol. 48, 833–843. Dasgupta, N., Kapur, V., Singh, K.K. et al. (2000) Tuber. Lung Dis. 80, 141–159. D’Autreaux, B., Tucker, N.P., Dixon, R. and Spiro, S. (2005) Nature 437, 769–772. Ho, Y.S., Burden, L.M. and Hurley, J.H. (2000) EMBO J. 19, 5288–5299. Galperin, M.Y., Nikolskaya, A.N. and Koonin, E.V. (2001) FEMS Microbiol. Lett. 203, 11–21. Martinez, S.E., Bruder, S., Schultz, A. et al. (2005) Proc. Natl Acad. Sci. U.S.A. 102, 3082–3087. Tanaka, A., Nakamura, H., Shiro, Y. and Fujii, H. (2006) Biochemistry 45, 2515–2523. Sousa, E.H., Tuckerman, J.R., Gonzalez, G. et al. (2007) Protein Sci. 16, 1708–1719.

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 3

NO and NOx Interactions with Hemes Peter C. Ford, Susmita Bandyopadhyay, Mark D. Lim, and Ivan M. Lorkovic Department of Chemistry and Biochemistry, University of California, Santa Barbara, CA 93106 U.S.A.

Abstract In this chapter, we will discuss the mechanistic studies of NO and NOx (NO− 2, NO2 , N2 O3 , etc.) reactions with heme proteins and heme model compounds, with the goal of providing further insight into the related reactions occurring in mammalian biology. •

Keywords: nitric oxide; nitrogen oxides; reaction mechanisms; hemes; heme models; heme proteins; metalloporphyrins.

ABBREVIATIONS Por2− TPPS TMPS TPP2− TmTP2− OEP2− TpivPP2− TMeOPP2− TClPP2− TCl2 PP TF5 PP sGC NOS eNOS iNOS Hb metHb Mb metMb

porphyrinato anion meso-tetrakis(4-sulfonatophenyl)-porphyrinato anion (tetra(4-sulfonato-mesityl)porphyrinato) meso-tetraphenylporphyrinato meso-tetra(meta-tolyl)porphyrinato octaethylporphyrinato “picket fence porphyrin” = meso-tetrakis(o-pivalamidophenyl)porphyrinato meso-tetra(4-methoxyphenyl)porphyrinato meso-tetra(4-chlorophenyl)porphyrinato meso-tetra(2,6-dichlorophenyl)-porphyrinato meso-tetra(perfluoro-phenyl)porphyrinato soluble-guanylyl cyclase nitric oxide synthase endothelial nitric oxide synthase inducible nitric oxide synthase hemoglobin methemoglobin myoglobin metmyoglobin

NO and NOx Interactions with Hemes

RBC P2− PPIX2− cGMP GSH GSNO DFT TTP NMR irr

67

red blood cell(s) porphyrinato protoporphyrin IX dianion guanylyl monophosphate glutathione S-nitrosoglutathione density functional theory tetra(p-tolyl)porphyrinato nuclear magnetic resonance irradiation wavelength

1. INTRODUCTION: REACTIONS OF NO AND NOX IN SOLUTIONS Nitric oxide (NO) is known to play an important role in mammalian biology including neurotransmission, blood pressure control, and immune response [1]. As a result, there has been a tremendous outpouring of research publications related to the medical consequences of nitric oxide biochemistry and a growing interest in the roles (both known and postulated) of other NOx derivatives including nitroxyl (HNO) [2], nitrogen dioxide − (NO2 ) [3–5], nitrite (NO− 2 ) [6], peroxynitrite (ONOO ) [7], and various nitrosoamine (R2 NNO) and nitrosothiol (RSNO) derivatives [8]. Notably, much of the biochemistry of NO and of the related NOx species is concerned with the reactions and interactions of these with metal centers [9], especially heme proteins. For examples, the primary source of endogenous NO is the oxidation of a guanidine nitrogen of arginine at the heme center of nitric oxide synthase, and the cardiovascular regulatory role of NO is defined by its reactions with the ferroheme of soluble-guanylyl cyclase (sGC) [10]. NO is also known to inhibit certain heme-based enzymes [11,12], and is carried (and released) by the ferriheme centers of nitrophorins, which are saliva proteins of certain blood-sucking insects [13]. In this context, the present chapter will outline and discuss the quantitative chemistry of NO with heme proteins and iron porphyrin heme models and to survey similar studies with various NOx with the goal of providing a chemical basis for elucidating the role(s) of these species in mammalian biology. It has been established quantitatively by Feelisch and coworkers that steady state − species found in mammalian tissue and fluids include nitrate (NO− 3 ), nitrite (NO2 ), S-nitrosothiols (RSNO) such as S-nitrosoglutathione and S-nitroso proteins, nitrosoamine derivatives, and nitrosylhemes [14,15]. Except for the first two, the concentrations of these NO-derived products are quite low, but there is remarkable variation in their relative concentrations from one organ or type of tissue to another. Their respective roles continue to be the source of extensive investigation and discussion. For example, nitrite has been shown to induce vasodilation [16]. Although the mechanism of this dilation has yet to be conclusively established, the role of NO− 2 may be to serve as another source of NO under conditions of low oxygen tension [17] (see below and Chapter 11 on “The reaction between nitrite and hemoglobin: The role of nitrite in hemoglobin-mediated hypoxic vasodilation” by Kim-Shapiro, Gladwin, Patel, and Hogg). Endogenous NO is formed by the oxidation of arginine by various isoforms of nitric oxide synthase (NOS), and possibly in part from nitrite, as noted above. The continuous

68

P.C. Ford et al.

production by the constitutive enzyme eNOS in endothelial cells gives low steady state concentrations. Although certain analytical techniques have suggested higher steady state NO concentrations in blood plasma, values as low as ∼4 nM were reported by a more recent study by Garthwaite [19]. Much higher concentrations (∼M) are produced transiently by macrophages and neutrophils by the inducible nitric oxide synthase iNOS during immune response to pathogen invasion [20]. What happens to the NO once produced? In the cardiovascular system, the primary sink has been long thought to be the reaction with the oxyhemoglobin (Equation 1) in the red blood cells (RBC), although the analogous reaction with muscular oxymyoglobin could also be a logical sink [21]. In buffered aqueous solutions, the large secondorder rate constants for the reactions of NO with HbO2 and MbO2 are comparable (9 × 107 M−1 s−1 and 4 × 107 M−1 s−1 , respectively) [22]. Another suggested pathway for NO removal from plasma is oxidation by a copper protein with nitric oxide oxidase functionality that would serve to balance the NO and NO− 2 concentrations [6,23]. In whole blood, the reactions of NO with the HbO2 in RBC are influenced by the heterogenous nature of this system and issues such as membrane diffusion [24]. NO + HbO2  → metHb + NO− 3

(1)

The principal target of NO in cardiovascular regulation is the FeII (PPIX) chromophore of sGC (Equation 2, PPIX = protoporphyrin IX dianion). NO coordination leads to conformational changes in this enzyme that activate the protein toward the synthesis of cyclic guanylyl monophosphate (cGMP) [25]. The in vitro “on” reaction with sGC (Equation 2) is very fast with a second-order rate constant kon = 14 × 108 M−1 s−1 , while the “off” reaction is slow (koff = 8 × 10−4 s−1 ) [26,27]. Thus, the equilibrium constant for formation of nitrosyl sGC is quite large (>1011 M−1 ) as generally seen with ferrous heme models such as FeII (TPPS) (TPPS = meso-tetrakis(4-sulfonatophenyl)porphyrinato anion) [28] and ferrous heme proteins such as myoglobin (Mb) [29,30]. The more remarkable feature about sGC is that it does not bind dioxygen (O2 ) [31–34]; thus, it acts as a very specific NO sensor, and recent studies by Boon and Marletta have concluded that the absence of a tyrosine or histadine in the distal pocket of sGC allows it to differentiate between NO and O2 [31–34]. The fast “on” reaction is necessary to effect timely activation given the low steady state concentrations of NO. On the other hand, the “off” reaction has importance as a likely mechanism for downregulating the enzyme, and there are indications that this may be significantly accelerated by cGMP, the product of the enzymatic reaction [35,36]. sGC + NO

kon koff

sGC(NO)

(2)

It is an often-stated myth that nitric oxide reacts extremely rapidly with dioxygen. However, while such autoxidation is a possible sink for NO in oxygenated media, the third-order kinetics dictate that the reaction rate is very dependent on the conditions, most importantly, the NO concentration. In the gas phase and in nonprotic media such as organic solvents, the product of autoxidation is nitrogen dioxide (Equation 3), while in aqueous solution the product is nitrite (Equation 4) [37]. For the latter, it is generally thought that the nitrite is formed by the hydrolysis of an N2 O3 intermediate. One pathway

NO and NOx Interactions with Hemes

69

for forming N2 O3 would be the reaction of NO2 with excess NO [38], although doubts have been raised regarding the intermediacy of NO2 during aqueous autoxidation [39]. 2NO + O2 → 2NO2

(3)

4NO + O2 + 2H2 O → 4NO− 2 + 4H+

(4)

The kinetics of both aqueous and nonaqueous NO autoxidation are reflective of a third-order reaction that is first order in O2 concentration and second order in nitric oxide concentration (Equation 5, kaut = 9 × 106 M−2 s−1 ) [37]. dNO = −kaut NO2 O2  dt

(5)

The second-order dependence on [NO] would make aqueous autoxidation a minor player as an NO sink under the nM conditions of cardiovascular regulation, but greater importance might be expected under conditions of immune response, given the significantly higher [NO] values in that case. However, another issue of considerable importance derives from the heterogenous nature of cells. Since both NO and O2 are much more soluble in aprotic hydrophobic media than in water, partitioning between the cytoplasm and the membranes of the cell will concentrate NO and O2 in the lipid membranes. Thus, although the membranes constitute a small percentage of the cell’s volume, a substantial fraction of NO autoxidation is likely to occur in such locations, and the autoxidation product in that hydrophobic environment would likely be NO2 [24]. Thus, there is a serious need to evaluate the chemistry of NO2 as well as of NO with different possible substrates in biological media. Another potential NO sink, especially under immune response conditions, is the − reaction with superoxide ion (O− 2 ) to give peroxynitrite (ONOO , Equation 6). This − reaction is certain to play a role, if O2 is produced simultaneously with NO in the same locale, since the rate constant in solution approaches that of a diffusion limited process [40,41]. Peroxynitrite undergoes pH dependent decomposition, reportedly to NO2 and hydroxyl radical ( OH) as well as isomerization to nitrate ion [42]. The decomposition of ONOO− is promoted by the presence of carbon dioxide, and NO2 plus the carbonate anion radical CO− 3 are the reported products of such decomposition [43]. Peroxynitrite coordination to metals promotes isomerization to nitrate [44,45]. •

•

− NO + O− 2 → ONOO

(6)

Physiological nitrite ion is present both in tissue and in fluids, and its potential roles are drawing increasing attention [7]. As noted above, NO− 2 has recently been shown to induce vasodilation in human subjects [16,46,47]. In conjunction with this discovery, it was proposed that the mechanism of such effects might be the reaction of NO− 2 with deoxyhemoglobin to form NO [46–48]. Extensive studies by Gladwin, Patel, Hogg, and Kim-Shapiro ([17,18], also, see Chapter 11 by Kim-Shapiro et al. in this book) have confirmed an earlier report by Doyle [48] that Hb is oxidized by nitrite with the release of NO. In addition, they have shown kinetics effects due to the allosteric properties of hemoglobin in different stages of oxidation ([17,18] and Chapter 11). However, in

70

P.C. Ford et al.

whole blood, the Hb is localized in the red blood cells, and it is difficult to see how NO generated by Hb reduction of nitrite escapes the RBC to diffuse to the vascular tissue and effect vasodilation. Such escape from the RBC would seem challenged by the high reactivity of NO with both Hb and HbO2 , unless there is a specific mechanism for such transport. Nonetheless, free NO has been observed as the product of the reaction of nitrite with RBC at moderate oxygen tension [47]. In this context, it is notable that several groups have suggested that the nitrite reaction with deoxyhemoglobin leads to formation of S-nitrosothiols in the RBC in addition to free NO [18,49]. If NO− 2 conversion to NO is responsible for nitrite-induced vasodilation, one should consider the reduction of NO− 2 by muscular myoglobin as another possible pathway for NO production. Interestingly, nitrite may play some other roles in the reactions with heme proteins. For example, it has also been found to catalyze the NO reduction of the ferriheme proteins met-hemoglobin (metHb) and met-myoglobin (metMb) to the ferrous analogs [50–52]. Thus, the different pathways by which these higher nitrogen oxides interact with heme systems and the potential importance of these to mammalian functions emphasize the need to understand their interactions with bioinorganic models. This chapter is intended to be an overview, not a comprehensive review, of NO and NOx interactions with heme models in various media and with heme proteins. More detailed reviews focusing on the interactions of NO with metal centers are listed in references [9,53–60].

2. THE FORMATION AND DISSOCIATION OF FERRIC AND FERROUS PORPHYRIN NITROSYL COMPLEXES Fe(Por)(NO). The ferrous porphyrin complexes are {FeNO}7 systems [61] characterized as a low-spin iron(II) (S = 1/2) with a neutral NO [62]. Selected properties of ferroheme model compounds are listed in Table 1 [63–68]. Scheidt and coworkers have shown that Fe(TPP)(NO) (I, TPP2− = meso-tetraphenylporphyrinato2− , see Fig. 1) has a bent nitrosyl coordination with an Fe N O bond angle of 149.2 and an Fe N(O) bond length of 1.72 Å [63]; however, “representative values” are 143.4 and 1.728(5) Å, respectively. Optimized DFT calculations of an Fe(P)(NO) model (where P2− is the porphine dianion) by Ghosh et al. give a calculated Fe N O bond angle of 143.8 [69] in good agreement with representative value, but a shorter calculated Fe N(O) bond at 1.692 Å. Six coordinate Fe(Por)(L)(NO) ferroheme nitrosyl model complexes tend to have somewhat more acute Fe N O angles (137 –140 ) and longer Fe N(O) bond lengths (1.74–1.76 Å) [57]. A much greater range of Fe N O angles and Fe N(O) bond lengths have been noted for nitrosyl complexes of ferroheme proteins (see summary in Table 2 of ref. [60]), although these may be somewhat dependent on the resolutions at which the protein crystal structures were determined. Nonetheless, a recent determination [70] of the 1.30 Å–resolution crystal structures of horse heart myoglobin nitrosyl complex, hh-Mb(NO) offered the surprising conclusion that the heme nitrosyl structure for this protein is remarkably dependent on the mode of preparation. When this was prepared from the reaction of hh-metMb with nitrite/dithionite, the Fe N O angle of the resulting complex is 144 with an Fe N(O) bond length of 1.87 Å. However, when prepared from reaction of reduced hh-Mb with NO, the resulting Fe N O angle is 120 with an Fe NO bond length of 2.13 Å, closer to values perhaps expected

Complex

Fe N

O bond angle

NO cm−1 (solvent)

UV-vis data ( in M−1 cm−1 ) in CHCl3 soret (nm)

FeII (TPP)(NO) I FeII (TPP)(NO)2 II FeIII (TPP)(NO2 )(NO)III FeIII (TPP)(NO3 )V FeIII (TPP)(NO3 )(NO)VI

149.2 (ss) 134.3 (DFT) 169.3∗ (ss) N/A Unknown

1681 1695 1884 N/A 1909

(CHCl3 ) (CHCl3 ) (CHCl3 ) (CHCl3 )

405 418 431 412 431

(100 × 103 ) (140 × 103 ) (180 × 103 ) (113 × 103 ) (180 × 103 )

Q0 (nm) 606 578 578 574 578

(87 × 103 ) (20 × 103 ) (10 × 103 ) (29 × 103 ) (10 × 103 )

Reference

NO and NOx Interactions with Hemes

Table 1. Selected X-ray, IR, and UV-vis data for certain Fe(TPP) complexes

Q1 (nm) 537(28 × 103 ) 540 (14 × 103 ) 544 (18 × 103 ) 513 (115 × 103 ) 544 (18 × 103 )

[63] [64,65] [66] [67] [68]

ss = derived from solid state structure; DFT = density functional theory calculations of porphine; ∗ = from picket fence porphyrin.

71

72

P.C. Ford et al. R1

R2

CH3 H3C C CH3 CO

R2

TpivPP

R2

R1 R2

NH –H

R1 N

R2

M2+

N

R2

N

–H

R2

TPP

R1

R2

SO3–

TPPS

N

R1

R2

OEP

–H

–H

–CH2CH3

Fig. 1. Some Fe(Por) complexes discussed in this chapter (Por2− is a general abbreviation for a porphyrinato ligand).

for an HNO complex. The authors explained the differences in terms of effects of the hh-Mb(NO) distal pocket in stabilizing local Fe NO conformational minima [70]. Fe(TPP)(NO) has been synthesized by the reductive nitrosylation of FeIII (TPP)(Cl) with NO in methanol (see Scheme 1) [62] and by reaction of sodium nitrite with Fe(TPP)(Cl) in the presence of a reducing agent [71]. Low temperature infrared spectral studies of sublimed porphyrin layers show that I can also be formed from the direct reaction of Fe(TPP) with NO [72]. Reaction of Angeli’s salt (sodium trioxodinitrate, Na2 N2 O3 ) with FeIII (TPP)(Cl) in methanol was also reported to result in I [73]. Since Angeli’s salt decomposes to give NO− /HNO, Fe(Por)(NO) would be the product expected for the reaction of HNO with ferri-porphryins [74,75]. FeIII(TPP)(Cl) + NO

FeIII(TPP)(Cl)(NO)

FeIII(TPP)(Cl)(NO) + MeOH

FeII(TPP) + MeONO + HCl

FeII(TPP) + NO

Fe(TPP)(NO)

Scheme 1. Reductive nitrosylation of Fe(TPP)(Cl).

Laser flash photolysis (irr = 355 and 532 nm) of FeII (TPP)(NO) in toluene solution results in the reversible photolabilization of NO (Equation 7) with a quantum yield ( ) of 0.5 at both wavelengths [76]. The Fe(TPP) thus formed reacts with excess NO via bimolecular kinetics to regenerate the starting complex with a very large second-order rate constant kNO = 52 × 109 M−1 s−1 (300 K). FeII(TPP)(NO)

k NO[NO]

FeII(TPP) + NO

(7) hν

NO and NOx Interactions with Hemes

73

Continuous photolysis (350<irr <550 nm) of solid FeII (TTP)(NO) (TTP2− = tetra (p-tolyl)porphyrinato) and FeII (OEP)(NO) in KBr pellets at 25 K has been shown to result in the linkage isomerization of the nitrosyl to form the isonitrosyl ( 1 –oxygen coordinated) derivative [77]. One can envision this happening via NO photolabilization to generate the {FeII (TPP),NO} geminate pair in the solid matrix. Recombination must be fast, and we would suggest that this is not very selective, since some of the isonitrosyl isomer is a product (Scheme 2). However, once the isonitrosyl complex is formed, the activation barrier for isomerization back to the more stable Fe NO species must be sufficiently large to make the reaction very slow at 25 K.

caged pair

O

NO

N hν

Fe

Fe fast fast slow

N O Fe

Scheme 2. Likely scenario for the photochemical formation of the isonitrosyl complex Fe(TPP)(ON) at 25 K (the circle represents a porphyrinato ligand such as TPP2− ).

FeII (TPP)(NO) can coordinate, albeit weakly, to ligands such as pyridine (py) in the position trans to the nitrosyl group. Reaction with large excess of L displaces NO to give Fe(Por)L2 (e.g., Equation 8) [78,79]. FeTPPNO + 2py → FeTPPpy2 + NO

(8)

In aerated pyridine, the “picket fence” analog Fe(TpivPP)(NO) (Fig. 1) was found to form the nitro derivative Fe(TpivPP)(py)(NO2 ). This oxidation was reversed upon addition of triphenylphosphine [80], presumably by oxygen atom abstraction from the nitro group to regenerate the nitrosyl complex plus Ph3 PO. The reaction of excess NO with Fe(TPP)(NO) has been reported to give the nitro nitrosyl complex Fe(TPP)(NO2 )(NO), possibly via the disproportionation of NO [81,82]. Similar disproportionation reactions have been characterized for ruthenium and osmium [83,b–86], and manganese analogs [87]. However, studies in this laboratory [88] found that the ambient temperature reaction of the iron(II) complex I with carefully purified NO follows a different pattern. Solutions of Fe(TPP)(NO) in chloroform, methylcyclohexane, and toluene display no observable change in the infrared spectrum of I ( NO = 1681 cm−1 in CHCl3 ) in the presence of NO in large excess at room temperature, if great care has been taken to exclude higher NOx and traces of air [88]. However, when the temperature of these solutions was lowered to 179 K, this NO band disappears and a second, more intense

74

P.C. Ford et al.

one at 1695 cm−1 appears. This band splits into three peaks with a 1:2:1 intensity ratio when a 1:1 mixture of 15 NO:14 NO was used. The process was reversed by raising T or by evacuating the cell to remove excess NO. We have interpreted these observations in terms of the reversible formation of the dinitrosyl complex Fe(TPP)(NO)2 (II) [62,65,66] under excess NO (Equation 9). FeTPPNO + NO  FeTPPNO2 I

(9)

II

Only when traces of air were admitted to such solutions (thereby effecting the autoxidation of NO to NO2 and N2 O3 under these conditions) did the IR bands characteristic to the nitro nitrosyl complex III ( NO = 1884 cm−1 ) appear. The latter process apparently represents the reaction of I with N2 O3 to give Fe(TPP)(NO2 )(NO) (Equation 10). FeTPPNO + N2 O3  FeTPPNONO2  + NO I

(10)

III

The conclusions of the above studies were substantiated by investigating the reaction of NO with thin porous layers of Fe(TPP) that had been sublimed onto a solid substrate. FTIR studies with carefully purified NO demonstrated the formation of Fe(TPP)(NO) exclusively [72,89]. The spectrum of Fe(TPP)(NO2 )(NO) was observed when higher Nx Oy were deliberately introduced. Fe(TPP)(NO)2 . Formation of the bis(nitrosyl) complex II was first suggested by Wayland and Olson who found that the EPR signal characteristic of Fe(TPP)(NO) diminished when a solution under an NO atmosphere was cooled to 77 K [62]. As described above, Lorkovic et al. [64] demonstrated the reversible appearance of a new nitrosyl stretch in the IR spectrum at 1695 cm−1 ( ∼ 1600 M−1 cm−1 ) at the expense of the NO of I (1681 cm−1 , 800 M−1 cm−1 ) upon cooling a CHCl3 solution containing NO and I. Reversible changes in the optical spectra paralleled the IR spectral changes. The 1 H NMR spectrum of analogous solutions in toluene-d8 demonstrated the shifting and sharpening of the porphyrin protons as the temperature was lowered in a manner consistent with the formation of a diamagnetic {Fe(NO)2 }8 species according to Equation 9. Similar changes did not occur in the absence of excess NO. From the temperature dependence of the line broadening, the equilibrium constant (K = 23 and 3100 M−1 at 253 and 77 K, respectively) and the enthalpy of dinitrosyl formation from the mononitrosyl (−6.7 kcal mol−1 ) could be determined [64]. Although the changes in the IR, optical, and NMR spectra were fully consistent with the formation of the dinitrosyl II in low-temperature solutions of I under NO, there remained an interesting ambiguity with regard to the IR spectra. The observation of a single NO band with an intensity about twice that of the mononitrosyl complex appeared to suggest a centro-symmetric structure, since point group symmetries without an inversion center would be expected to display two such bands [64]. The simplest geometry that would fit this suggestion would be a trans-Fe(TPP)(NO)2 complex in which the two M NO bonds angles are colinear. However, linear Fe NO bonds appeared to be counterintuitive given the {Fe(NO)2 }8 electronic configuration and relatively low NO frequency, so an alternative trans-anti configuration with C2h symmetry was considered. In this context, independent density functional theory (DFT) computations for a

NO and NOx Interactions with Hemes

75

Fig. 2. Fe(P)(NO)2 (P = porphine) structure optimized using DFT (B3LYP/3–21g) [65].

Fe(porphine)(NO)2 model were performed in two laboratories and both concluded that the lowest energy structure with this composition is not centro-symmetric but has a trans-syn geometry (Fig. 2) [65,90]. For such a structure, both the symmetric and asymmetric NO stretching modes are IR allowed, but only one strong band was seen. The ambiguity with regard to this observation was resolved by DFT frequency calculations predicting the observed asymmetric band to be about 2 orders of magnitude more intense than NO (symmetric) for the trans-syn geometry [65]. The analogous ruthenium complex Ru(TPP)(NO)2 has also been observed and has a higher formation constant than does the iron homolog [65].

3. TRANSFORMATIONS OF COORDINATED NOX Fe(TPP)(NO2 )(NO). The nitro nitrosyl complex III was first reported by Yoshimura who observed the formation of a new species with a NO = 1884 cm−1 upon bubbling of a chloroform solution of Fe(TPP)(NO) with NO [81]. Although other groups have reported similar pathways to III, it was noted above that the reaction of I with NO does not give III, if the gas streams are carefully purified to remove higher oxides and contamination by oxygen is scrupulously avoided. Instead, there is an equilibrium with the dinitrosyl complex Fe(TPP)(NO)2 that is more readily seen at low temperature. When N2 O3 was introduced to such solutions, the formation of III was immediate (Equation 10) [88,89]. Reaction of N2 O3 with the -oxo dimer [Fe(TPP)]2 O is also reported to lead to the eventual formation of III [91]. The equilibrium constant for the reaction of Fe(TPP)(NO) with N2 O3 to form Fe(TPP)(NO2 )(NO) (Equation 10) has been measured to be 160 (298 K) by spectroscopic methods in toluene solutions in the presence of excess NO. Similar experiments were carried out for a series of analogous Fe(Por)(NO) complexes, and K values for reactions analogous to Equation 10 are listed in Table 2. Simple inspection suggests that the electron-withdrawing porphyrin substituents, as evidenced by the increasing NO values, give the smaller the equilibrium constant K. Mössbauer measurements on solid III were interpreted in terms of a low-spin Fe(II) species consistent with the {FeNO}6 (S = 0) electronic configuration [91]. This and

76

P.C. Ford et al.

the nitrosyl stretch ( NO = 1884 cm−1 in CHCl3 ) would be in accord with a linear metal nitrosyl bond as expected for the FeII (NO+ ) formalism. The crystal structure of Fe(TPP)(NO2 )(NO) itself showed severe disorder, but the structure of an analogous complex Fe(TpivPP)(NO2 )(NO) shows an Fe N O bond angle of 169.3 in accord with these expectations [92]. The kinetics of reactions leading to the formation of III were studied by a combination of laser flash photolysis and stopped flow spectroscopic techniques [66]. For example, flash photolysis (irr = 355 nm) of III in toluene solution competitively labilizes NO and NO2 to form Fe(TPP)(NO2 ) (IV) and Fe(TPP)(NO) (I), respectively (Scheme 3).

Table 2. NO values for various ferrous porphyrin nitrosyl complexes Fe(Por)(NO) and equilibrium constants K determined for the reaction of Fe(Por)(NO) with N2 O3 [93] K

FePNO + N2 O3  FePNO2 NO + NO Pa

NO (cm−1 b

Kc

1662 1673 1676 1678 1679 1690 1702

400 138 151 160 122 24 1.0

OEP TMeOPP TTP TPP TClPP TCl2 PP TF5 PP

TMeOPP2− = Tetra(4-methoxyphenyl)porphyrinato; TClPP2− = Tetra (4-chlorophenyl)porphyrinato; TCl2 PP = Tetra(2,6-dichlorophenyl)porphyrinato; TF5 PP = Tetra(perfluoro-phenyl)porphyrinato. b in toluene solution. c 273 K in toluene. a

NO

N2O3

kN O 2 3 k –NO

NO 2

Fe –NO2

NO Fe

hν –NO

NO2

Fe k –NO

NO2

k NO

NO

Scheme 3. Laser flash photolysis of Fe(TPP)(NO2 )(NO) and related reactions.

NO and NOx Interactions with Hemes

77

Under excess NO, the NO2 is trapped to give N2 O3 so the kinetics observed were the back reactions of Fe(TPP)(NO) with N2 O3 and the reaction of IV with NO. The respective second-order rate constants for these reactions were kN2 O3 = 18 × 106 M−1 s−1 and kNO = 42 × 105 M−1 s−1 in 298 K toluene solution. Solution-phase flash photolysis studies [93–95] of nitro nitrosyl complexes of related porphyrins Fe(Por)(NO2 )(NO) demonstrated photochemistry similar to III but with some modification of the rate constants KN2 O3 and kNO . As will be seen below, the reactions of FeIII (Por) complexes with NO are generally several orders of magnitude slower than the reactions of FeII (Por) species under analogous conditions. When solid Fe(TPP)(NO2 )(NO) in a KBr pellet was irradiated (irr = 300 − 500 nm) at 200 K, nitro to nitrito linkage isomerization occurred to give Fe(TPP)(ONO)(NO), which is stable at this temperature [96,97] (nb: “nitro” is N-coordinated, while “nitrite” is O-coordinated NO− 2 ). Such photoinduced nitro ←−−→ nitrito linkage isomerization of the coordinated NO2 group is known for a number of systems [98]. Photolysis of III at lower temperatures (11 K) demonstrated linkage isomerization of both the nitrosyl and nitro groups to form a species with IR spectral properties consistent with formation of the isonitrosyl-nitrito analog, Fe(TPP)(ON)(ONO); however, the latter is only stable below 50 K [96,97]. The formation of this double linkage isomer species is supported by DFT calculations. A likely mechanism for these isomerizations would be the dissociation of either NO or NO2 as seen in Scheme 3, followed by relatively nonselective trapping of these by the metal. When either radical is trapped to form a less stable linkage isomer, an activation barrier for isomerization to the thermodynamically favored form is sufficient to stabilize such a species at low temperature. Notably, solutions of Fe(TPP)(NO2 )(NO) are stable only under an NO atmosphere [81,66]. For example, vacuum removal of NO from toluene solution of III results in conversion to an equimolar mixture of the nitrato complex Fe(TPP)(NO3 ) (V) and Fe(TPP)(NO) [66,72]. The optical spectral changes accompanying this transformation could be monitored by the rapid dilution of a toluene solution of III, NO, and excess N2 O3 using the sequential mixing accessory of a stopped-flow spectrophotometer. These were resolved in terms of relatively fast [NO] independent pathway interpreted as NO dissociation to give IV (kNO = 26 s−1 at 298 K). This was followed by a much slower step, which was concluded to be the oxidation of IV to the nitrato complex V. The rate of the slow step proved dependent on the NO2 /N2 O3 concentration, so it appears likely that V is formed via oxygen atom transfer from NO2 . Although there was significant experimental uncertainty under these conditions (e.g., kslow = 002 ± 001 s−1 for [NO2 /N2 O3 ] = 20 M), the rate constant kO can be estimated as ∼103 M−1 s−1 . It seems likely that the oxygen transfer step to give V (Equation 11) is preceded by isomerization of the nitro complex formed by NO dissociation from III to its nitrito analog as illustrated. Since the nitrato ligand of Fe(TPP)(NO3 ) coordinates to FeIII in a bidentate fashion [67], the nitrito complex Fe(TPP)(ONO) would be a logical intermediate. DFT computations [93–97,] of optimized structures for III, Fe(TPP)(ONO)(NO), IV, and its linkage isomer Fe(TPP)(ONO) indicate that while III is significantly lower energy than Fe(TPP)(ONO)(NO), Fe(TPP)(ONO) is much lower energy than Fe(TPP)(N O2 ), lending credence to Equation 11 as a possible channel to formation of V.

78

P.C. Ford et al.

O N

O

NO 2

NO

Fe

Fe

Fe

N

N

O

O

O

O

Fe N O

NO

O

O N O

(11) Fe(Por)(NO2 ). Studies into the reactivity of IV and related species were derived from attempts to model nitrite-heme interactions that occur with deoxyhemoglobin [17,18,99] and nitrite reductase [100]. IV is expected to be a low-spin (S = 1/2) nitro complex [101]. Castro has proposed that the powerful oxygen donor formed upon mixing of substrate [K(18-crown-6)](NO2 ) and Fe(OEP)Cl in mildly acidic N -methylpyrolidinone is the nitro complex Fe(OEP)(NO2 ), and this was claimed to be active for the oxidation of PPh3 , several alkenes, and other substrates [102,103]. (It should be noted, however, that the acidic medium would give HNO2 , which is in equilibrium with N2 O3 , another strong oxidant.) So far, attempts to prepare the TPP analog IV from the reaction of nitrite salts with Fe(TPP)Cl have been unsuccessful, resulting in the isolation of the disproportionation nitrosyl and nitrato complexes I and V as products [104]. Use of the more sterically demanding picket fence porphyrin protects coordinated nitrite from disproportionation. The five-coordinate Fe(TpivPP)(NO2 ) has been prepared in two ways, by direct reaction of KNO2 with Fe(SO3 CF3 )(H2 O)(TpivPP) [101] and by BF3 extraction of NO− 2 from [Fe(TpivPP)(NO2 2 ], as observed by electron paramagnetic resonance (EPR) [105]. Several six-coordinate analogs, Fe(TpivPP)(NO2 )(L) (L = py [80]), CO and tert-butyl isocyanide [106] have all been prepared and structurally characterized. Recently, Richter-Addo and coworkers determined the 1.20 Å–resolution crystal structure of the nitrite ion adduct of horse heart met-myoglobin. The ligand is bound to iron in the nitrito form, and the complex is formulated as hh-metMb(ONO) with an Fe ONO bond length of 1.94 Å, and O N O angle of 113 [70]. As noted above, IV was generated transiently by the 355-nm flash photolysis of Fe(TPP)(NO2 )(NO) (Scheme 3) [66]. However, the exact structure of the species directly observed is somewhat in doubt, in part because DFT computations suggest that IV is a higher energy species than the nitrito linkage isomer Fe(TPP)(ONO). This conclusion derives support from a recent study [107] where it was shown that introduction of small increments of NO2 gas into a cryostat that contained sublimed layers of Fe(TTP) led to the appearance of three new IR spectroscopic bands at 1528, 901, and 751 cm−1 that have their isotope counterparts at 1495, 878, and 747 cm−1 when 15 NO2 was used. The intensity growth of these bands correlates directly with the addition of each portion of NO2 (15 NO2 ) and demonstrates the formation of a species that includes a coordinated NO2 . The new complex is the five-coordinate Fe(TTP)( 1 -ONO) species and is fairly stable in the solid state even at room temperature, but decomposes under vacuum within a few days to give mostly the nitrosyl complex Fe(TTP)(NO) ( (NO)=1677 cm−1 ). Low-temperature reaction of NO with Fe(TTP)(ONO) gave the nitrite nitrosyl complex Fe(TPP)( 1 -ONO)(NO), which upon warming converted to more stable nitro nitrosyl isomer (Fe(TPP)(NO2 )(NO)) (Scheme 4).

NO and NOx Interactions with Hemes

O

79

O N

Fe

O + NO

100 –150 K

Fe

O N

O

O N

R.T.

Fe

N

N O

O

Scheme 4. Fe(TPP)(NO3 ) and Fe(TPP)(NO3 )(NO).

The nitrato complex V was characterized structurally by Goff and coworkers to have a bidentate coordinated nitrate and was shown to have a high-spin electronic configuration (S = 5/2) [67]. It can be synthesized by reaction of the -oxo dimer [Fe(TPP)]2 O with excess N2 O3 or nitric acid [67,91]. For high-spin FeIII porphyrins, the 1 H NMR resonances for the pyrrole hydrogens are characteristically ∼80 ppm downfield from tetramethylsilane [108]. For V, this was about ∼73 ppm [66]. The structures of several other nitrato complexes analogous to V have been determined by X-ray diffraction. The picket fence porphyrin complex Fe(TpivPP)(NO3 ) also has a bidentate nitrate ion that is bound in a less symmetric fashion [105]. On the other hand, the solid state structure for Fe(OEP)(NO3 ) was found to have a monodentate nitrate, the difference attributed to the steric interaction with the OEP ethyl groups [109]. When sublimed layers of solid V in a cryostat were exposed to gaseous NO, the result was the formation of the nitrosyl nitrato complex Fe(TPP)(NO3 )(NO) (Equation 12) ( NO = 1901, 1863 cm−1 for 14 N16 O, 15 N16 O respectively) [68,110]. IR bands at 1505, 1266, and 978 cm−1 were attributed to the presence of a monodentate nitrate. Solutionphase IR measurements of a rapidly mixed CH2 Cl2 solution of V and NO demonstrated the rapid formation of a similar species ( NO = 1909, 1873 cm−1 for 14 N16 O and 15 N18 O, respectively). This was attributed to formation of Fe(TPP)( 1 -ONO2 )(NO) (VI) [110]. In the presence of excess NO, spectral changes in the solution indicate that VI undergoes further reaction to give III. On the basis of isotopic labeling experiments, it was proposed that the most likely scenario involved direct attack by excess NO on the

1 -coordinated nitrate ligand (Equation 13) [111] to give N2 O4 plus the ferrous nitrosyl complex Fe(TPP)(NO). Reaction of the latter with N2 O4 , NO2 (from dissociation of the dimer) or N2 O3 (from reaction of NO2 with NO) would lead to formation of the final Fe(TPP)(NO2 )(NO) product. FeTPP2 − O2 NO + NO  FeTPP1 − ONO2 NO

(12)

O

NO O

N O Fe

Fe N O

N O

+

ONONO2

NO2 + NO2

(13)

80

P.C. Ford et al.

These observations illustrate the plasticity of the Fe(Por)(Nx Oy ) systems. Moderate changes in the reaction conditions, principally the presence of excess NO or other nitrogen oxides, can shift the direction of the reactions between various species.

4. REACTIONS OF NO AND NOX WITH HEME MODELS AND PROTEINS IN AQUEOUS MEDIA NO substitution reactions: Although rates of NO reactions with various iron porphyrins and heme proteins in aqueous media have been studied for several decades [111–113], systematic mechanistic investigations are more recent. These were initiated by carrying out the flash photolysis of aqueous FeII (Por)(NO) and FeIII (Por)(NO) solutions in the presence of excess NO, where Por is a water-soluble porphyrin such as TPPS [29,113,114,115]. Photoexcitation results in NO labilization from Fe(Por)(L)(NO) followed by exponential relaxation of the system to equilibrium (Equation 14). The observed rate constant kobs could be extracted for different [NO] and other variables. According to this model, kobs = kon [NO] +koff , and a plot of kobs versus [NO] should be linear with a slope equal to kon and an intercept equal to koff . The slopes of such plots are inherently more accurate than the intercepts, so koff values so determined will have a higher relative uncertainty. Fe(Por)(L) + NO

k on k off

Fe(Por)(L)NO

(14)

hν

Laverman and coworkers carried out such studies with the iron(II) and iron(III) complexes of the water-soluble porphyrins TPPS and TMPS as well as the ferriheme protein metMb [29,114,115]. These studies involved systematic measurements of kon and koff as functions of temperature (298–318 K) and hydrostatic pressure (0.1–250 MPa) to determine values of H‡ , S‡ , and V‡ (Table 3). Table 3. Kinetics data for NO “on” and “off” reactions of Fe(Por) centers in aqueous mediaa “On” reactions

kon (M−1 s−1 )

H‡ on kJ mol−1

S‡ on J mol−1 K−1

V‡ on cm3 mol−1

FeIII (TPPS)+NO FeIII (TMPS)+NO metMb+NO FeII (TPPS)+NO FeII (TMPS)+NO

45 × 105 28 × 106 48 × 104 15 × 109 10 × 109

69 ± 3 57 ± 3 63 ± 2 24 ± 3 26 ± 6

95 ± 10 69 ± 11 55 ± 8 12 ± 10 16 ± 21

9±1 13 ± 1 20 ± 6 5±1 2±2

“Off” reactions

koff (s−1 )

H‡ off

S‡ off

V‡ off

05 × 10 09 × 103 42 64 × 10−4

76 ± 6 84 ± 3 68 ± 4 B

60 ± 11 94 ± 10 14 ± 13 B

18 ± 2 17 ± 3 18 ± 3 B

III

Fe (TPPS)(NO) FeIII (TMPS)(NO) metMb(NO) FeII (TPPS)(NO) a

3

From references [29] and [114,115]; TMPS = (tetra(4-sulfonato-mesityl)porphyrinato).

NO and NOx Interactions with Hemes

81

For the ferriheme models, which are present as the diaquo complexes FeIII (Por)(H2 O)2 , the large and positive S‡ on and V‡ on values (Table 3) indicate an “on” substitution mechanism dominated by ligand dissociation as illustrated in Scheme 5. For P = TPPS, the exchange between coordinated and solvent water for FeIII (Por)(H2 O)2 was shown by van Eldik and coworkers to display activation parameters (H‡ ex = 67 kJ mol−1 , S‡ ex = 99 J mol−1 K−1 , and V‡ ex = 79 cm3 mol−1 , for FeIII (TPPS)(H2 O)2 ) [116], nearly identical to those measured for the NO substitution pathway (Table 3). The coinciding values of the activation parameters are in full agreement with the dissociative mechanism for the reaction shown in Equation 14.

FeIII(Por)(H2O)2

FeIII(Por)(H2O)2 + NO

k1 k –1 k2 k –2

FeIII(Por)(H2O) + H2O

FeIII(Por)(H2O)(NO)

Scheme 5. Limiting dissociative mechanism for NO substitution onto FeIII (Por)(H2 O)2 .

On the basis of microscopic reversibility, the intermediate(s) in the “off” step should be the same as those generated during the kon pathway, thus Fe NO bond breaking (k−2 ) would be the energetically dominant step. Coordination of NO to FeIII (Por) is accompanied by considerable charge transfer to give a linearly bonded, diamagnetic complex that can be formally represented as FeII (Por)(NO+ ). Thus, the “off” reaction must reflect intrinsic entropy and volume changes associated with the spin change and solvent reorganization as the charge localizes on the metal. Table 3 also lists the rate constants and activation parameters for the ferriheme protein met-myoglobin. These are sufficiently close to the model complexes to suggest that the mechanisms for the “on” and “off” reactions for the protein are related [115]. The water-soluble ferrous complex FeII (TPPS) reacts with NO about 103 times faster than does the ferric analog [29]. The small values of the activation parameters are consistent with rates largely defined by diffusional factors, although the kon values reported are about an order of magnitude less than diffusion limits in water. High-spin ferroheme proteins complexes tend to be considerably more reactive toward ligands than are the ferriheme analogs, and a likely explanation would be that the former are often five-coordinate; therefore, the rates are not limited by the lability of the metal center. The “off” reaction rates for FeII (TPPS) were obtained by using Ru(edta)− as an NO scavenger. Addition of excess Ru(edta)− to an aqueous solution of FeII (TPPS)NO allowed measurements of the koff value (63 × 10−4 s−1 , 298 K) [29]. The equilibrium constant for formation of the FeII (TPPS)(NO) complex can be determined as >1012 M−1 from the kon /koff ratio, about 9 orders of magnitude larger than that for the Fe(III) analog. The above conclusion that the faster “on” rate for ferro-heme models relative to the ferrihemes is due to the five-coordinate nature of the former is challenged by a recent result involving the flash photolysis kinetics of a sterically crowded watersoluble ferriheme complex FeIII (TMSPP)(NO) (TMSPP = meso-tetrakis(2,4,6-trimethyl3-sulfanatophenyl)-porphyrinato) [117,118]. The steric features of the TMSPP ligand

82

P.C. Ford et al.

prevent the complex from forming oxo-bridged dimers; therefore, its kinetics could be investigated at a wide range of solution pHs including alkaline conditions. Van Eldik et al. [117,118] concluded that at low pH, the predominate form of the nonnitrosated complex is the hexacoordinate diaquo complex FeIII (TMSPP)(H2 O)2 , but at high pH it is the five-coordinate hydroxo complex FeIII (TMSPP)(OH). However, surprisingly, the latter reacts with NO with a rate much slower than the former (kNO = 74 × 103 and 96 × 105 M−1 s−1 , respectively). They proposed that the reaction of FeIII (TMSPP)(H2 O)2 with NO exhibits a typical dissociative substitution mechanism at low pH as seen above for FeIII (TPPS)(H2 O)2 but that the rate-limiting step for the hydroxo analog is associative in character. Although an associative pathway makes sense for the reaction of NO with five-coordinate hydroxo complex, it is still puzzling that this would be slower. This effect was attributed to an increase in the activation barrier related to spin reorganization and structural rearrangements accompanying NO coordination to the high-spin FeIII (TMSPP)(OH). Reductive nitrosylations: The water-soluble ferrous porphyrin complex FeII (TPPS)(NO) (Fig. 1) is formed via the reductive nitrosylation of the ferric analog FeIII (TPPS) in aqueous solution under excess NO [50–52,119]. Similar reactions have been demonstrated for various ferriheme proteins and studied quantitatively for ferri-cytochrome C, metMb and metHb by Hoshino et al. [52]. For FeIII (TPPS), initial formation of the ferric nitrosyl complex (Equation 15, K = 132 × 103 M−1 ) activates the NO to nucleophilic attack by solvent H2 O (or OH− ), leading to net reduction of the metal center and oxidation of NO to nitrite ion (Equation 16). The FeII (TPPS) initially formed is trapped by the very rapid reaction with the excess NO (Equation 17) [29,113–115]. In near-neutral conditions, the driving force for the overall transformation is the very large equilibrium constant for the formation of FeII (TPPS)(NO) (K>1012 M−1 ) [29]. In moderately acidic solution (pH 4–6), the reductive nitrosylation of aqueous FeIII (TPPS) occurs slowly with a pH-independent pathway (kH2 O = 27 × 10−4 s−1 at 298 K) and is also subject to general base catalysis by the buffer [119]. FeIII TPPS + NO  FeIII TPPSNO

(15)

+ FeIII TPPSNO + H2 O  FeII TPPS + NO− 2 + 2H

(16)

FeII TPPS + NO  FeII TPPSNO

(17)

A particularly interesting twist to the FeIII (TPPS) reductive nitrosylation mechanism in aqueous solution was the discovery that the nitrite ion is not only the product but also a catalyst for this reaction [119]. Nitrite catalysis of reductive nitrosylation has now been shown for another ferriheme model as well as for metHb and metMb [50]. Two mechanistic explanations have been offerred. The first is an inner sphere pathway proceeding via nucleophilic attack of NO− 2 to the ferric nitrosyl in a manner analogous to the apparent reaction with other nucleophiles such as water. However, it is not clear whether the nucleophilicity of nitrite is sufficient to explain this special reactivity. An alternative is the outer sphere electron transfer mechanism proposed in Scheme 6, whereby NO− 2 is oxidized to NO2 by the ferric nitrosyl complex and the latter is trapped by excess NO to give N2 O3 , which is rapidly hydrolyzed by water to nitrous acid. Notably, both of these hypothetical pathways would have N2 O3 as a likely intermediate, and this possibility

NO and NOx Interactions with Hemes

OH2 FeIII OH2

+ NO

83 O N+

N

FeII

FeII

OH2

NO2–

O + H2O

.

NO2

NO

N2O3 H2O 2 NO2– + 2 H+

Scheme 6. Proposed outer sphere electron transfer pathway for reductive nitrosylation of ferriheme models and proteins.

may have biological consequences if this is generated in the hydrophobic pocket of a protein [50,51]. N2 O3 is a nitrosating agent and its formation might be one source of nitrosated proteins such as SNO-Hb. Nitrosation of a ligand: Ligands other than water (hydroxide) may also be nitrosated by the transfer of an NO+ from the FeIII (Por)(NO) of a ferriheme nitrosyl model or protein. For example, the reaction of the biological antioxidant glutathione (GSH) with metMb(NO) has been reported to give S-nitrosoglutathione. Reichenbach et al. [120] reported that NO reduction of metMb in pH 7.4 phosphate buffer solution by GSH gave spectral changes indicating the formation of Mb(NO) as one product, while amperometric sensor experiments were interpreted in terms of the nitrosoglutathione (GSNO) being the other product (Equation 18). The second-order rate constant for reaction of GSH with metMb(NO) was determined to be 47 M−1 s−1 (298 K). This is somewhat surprising given that kOH for hydroxide ion is only an order of magnitude higher (32 × 102 M−1 s−1 ) [51]. metMb + 2NO + GSH → MbNO + GSNO + H+

(18)

A similar reductive nitrosylation pathway [121] has been invoked as a possible mechanism for the nitrosation of the -cys-93 of hemoglobin to form S-nitroso-hemoglobin (SNO-Hb), the subject of a highly debated proposal as an NO carrier in the cardiovascular system [122,123]. One concern, however, is that crystal structural data indicate the distance between iron center and the -cysteine-93 is quite large (>10 Å) [124], thus, direct reaction of the cysteine with Fe(III) coordinated NO would be an unlikely step in the transfer of NO+ to the -cys-93. Others have also demonstrated that S-nitrosylthiols result from bolus addition of NO solution to normoxic solutions of Hb or red blood cells [125–127]. The major product under these conditions is metHb owing to the rapid reaction of NO with oxyhemoglobin [23,128], and it was argued that such formation of SNO-Hb may result from mixing artifacts especially when dioxygen is present [125–127,129]. As noted above, partitioning between hydrophilic and hydrophobic regions may lead to the compartmentalization of NO inside the protein and thus may impact the selectivity for S-nitrosothiol formation under physiological conditions [23,130]. •

84

P.C. Ford et al.

In this context, Angelo et al. [49] suggested that specific micropopulations of nitrosyl Hb could support the chemistry of SNO-Hb formation. By using nitrite as the source of NO, they report that a T state micropopulation of a heme-NO species, with spectral properties of Fe(III)NO, acts as a precursor to SNO-Hb formation, accompanying the allosteric transition of Hb to the R state. They also reported that an S-nitrosothiol precursor is formed within seconds at physiological concentrations of nitrite and deoxyHb, and produces SNO-Hb in high yield upon its prompt exposure to O2 or CO. Deoxygenation/reoxygenation cycling of oxyHb in the presence of physiological amounts of nitrite also produces SNO-Hb. It remains unclear how the system might overcome the separation between the cys-93 and ferriheme sites for transfer of an equivalent of NO+ , but again as suggested above [51,59], if N2 O3 were generated inside the hydrophobic pocket, it should be sufficiently long lived to accomplish this task. A different mode of ligand nitrosation has been observed by Montfort and coworkers [131]. The crystal structure of the salivary nitrophorin from the bed bug Cimex lectularius (a ferriheme protein that serves as a NO carrier) has been solved and shown to have a cysteinate ligand in the axial site of the heme, trans to the nitrosyl ligand. After the crystal was soaked in a NO saturated solution, the structure changed; NO remained bound to the heme in the distal pocket, but with a bent Fe N O bond consistent with reduction to an {Fe NO}7 center. In addition, a second NO was found on the proximal cysteine, giving an S-nitroso-cysteine (cys SNO) that is no longer coordinated. In this case, reductive nitrosylation of the ferriheme complex occurs at the coordinated cysteinate sulfur of the FeIII (Por)(Cys)(NO) complex to give FeII (Por)(NO) plus cysSNO (where Por = the heme of the nitrophorin protein).

O

O

N Fe III R

N Fe II

+ NO

S

(19)

S NO R

A similar ligand nitrosation has been described by van Eldik and coworkers, who studied the mechanism of NO binding to the synthetic heme thiolate complex(SR1 ) shown in Fig. 3 [132]. The reaction of SR1 complex with excess NO to form SR1 (NO) in methanol (kon = 27 × 106 M−1 s−1 at 298 K) follows a limiting dissociative mechanism consistent with that seen for other hexacoordinate ferriheme models such as FeIII (TPPS) [29]. However, formation of SR1 (NO) is followed by a slower reaction, perhaps involving attack of a second NO molecule on the thiolate ligand accompanied by the homolytic cleavage of the Fe S and formation of the five-coordinate SR1 (FeII ) nitrosyl complex in analogy to Equation 18. Moreover, they found that the attack of N2 O3 (present in the saturated NO solutions as the result of NO oxidation by trace oxygen impurities) on this five-coordinate SR1 (FeII ) nitrosyl complex gives a nitrosyl nitrito complex.

NO and NOx Interactions with Hemes

85

R N N

III

Fe N N HN

NH S

O

O

O

NH O

R = NHCOC(CH3)3

Fig. 3. The synthetic heme-thiolate complex (SR1 ) (Figure provided by R. van Eldik).

These observations find analogy in the previously reported base catalyzed N-nitrosation of a coordinated ligand seen for the reaction of NO with the copper (II) complex Cu(DAC)2+ (where DAC = dimethylanthrancenyl-cyclam) giving Cu(I) and the now luminescent N-nitroso analog of DAC [133] (Equation 20). Furthermore, metalcatalyzed decomposition of S-nitroso thiols is the reverse of ligand nitrosation [134,135], and analogous intermediates/transition states are likely to be playing a role in these mechanisms. For example, the ruthenium complex RuII (OEP)(CO) reacts with S-nitrosothiols to form the respective RuII (OEP)(NO)(thiolate). Stopped flow spectrophotometric studies have shown that this reaction occurs via an S-coordinated RuII (OEP)(RSNO)(CO) intermediate (Scheme 7) [136]. The latter readily undergoes cleavage of the S NO bond to release NO and form RuIII (OEP)(RS)(CO) followed by NO replacement of the coordinated CO to give the final product. The cleavage of the RS NO bond of RuII (OEP)(RSNO)(CO) would be the microscopic reverse of the NO reaction with S-coordinated thiolate and is clearly analogous to the reaction depicted by Equation 18.

2+ HN

N Cu N

N NH

+ NO

N

HN

NO

+ Cu+ + H+

(20)

N

I

II

Reactions with peroxynitrite: It has been proposed that iron porphyrins act as antioxidants to protect the organism from ONOO− [137]. In this context, the reactions of several ferric water-soluble porphyrins with peroxynitrite were studied using stopped

86

P.C. Ford et al. O

O

C

C

+RSNO

RuII

II

Ru

S NO

R O

O C

N

–CO

RuIII S: R

RuII +NO

S: R

= octaethylporphyrin

Scheme 7. Reactions with peroxynitrite.

flow techniques. Stern et al. found that at pH 7.4 (310 K), these porphyrins catalyzed the isomerization of peroxynitrite to nitrate through the reversible oxidation to a ferryl (FeIV O) state [44]. This ferryl intermediate was proposed to react through a variety of pathways that were dependent on the conditions (concentration of iron, ONOO− , and other biological substrates such as antioxidants) to decompose the peroxynitrite and return the porphyrin to its ferric state [138,139]. Herold and coworkers studied the reaction of peroxynitrite with several mutated metmyoglobin proteins, where the distal histidine was substituted with various amino acids that could not form hydrogen bonds with the coordinated water molecule. They observed that the mutated proteins catalyzed the decomposition of peroxynitrite more efficiently than the histidine containing metMb, and proposed that the role of the hydrogen bond between water and histidine was to regulate the rate of catalysis [140,45]. The same group also reported that peroxynitrite reacts with nitrosyl hemoglobin following an outer sphere electron transfer mechanism to release NO through the formation of a nitrosyl met-hemoglobin intermediate. They proposed that this reaction may be relevant for elucidating the mechanism of protein modification observed on the onset of various immune-related disease states. Peroxynitrite has also been shown to react rapidly with oxyhemoglobin both in the presence and in the absence of CO2 [141]. The reactions proceed via the intermediate HbFeIV O, which is reduced to metHb by its reaction with NO2 . In the presence of physiological relevant amounts of CO2 , HbFeIV O can also be generated by reaction of NO2 with oxyhemoglobin, via formation of a peroxynitrato-metHb complex. In this context, pulse radiolysis studies by Goldstein et al. of the kinetics and mechanism of NO2 reacting with oxymyoglobin are particularly relevant. The reaction is quite fast (45 × 107 M−1 s−1 ) and the product is MbFeIII OONO2 , which undergoes homolysis of the peroxide bond to form MbFeV O. Decay of the latter by reaction with another oxyMb leads eventually to metMb and nitrate. Thus heme proteins can detoxify NO2 [142]. •

•

•

•

•

NO and NOx Interactions with Hemes

87

Autoxidation of heme-nitrosyls: As noted above, reactions of NO with oxyhemoglobin and oxymyoglobin to form nitrate plus metHb and metMb, respectively, (e.g., Equation 1) are very fast. Herold and coworkers examined the time-resolved spectroscopy of such reactions and concluded that NO reacts with the FeII (O2 ) species to give peroxynitrito intermediates FeIII (OONO) [22] Under neutral or acidic conditions, the latter rapidly decays to the FeIII forms of the proteins with quantitative formation of nitrate. Thus, the metal mediates the isomerization of peroxynitrite to nitrate (Equation 21). HbO2  + NO → metHbOONO−  → NO− 3 + metHb

(21)

Reaction of the analogous nitrosyl complexes Mb(NO) and Hb(NO) with dioxygen to give metMb plus nitrate, a reaction of very great importance regarding the stability of cured meats, is dramatically slower. The kinetics of Mb(NO) autoxidation were studied by Skibsted et al. [143,144], who reported that even at low O2 concentrations, the rate displayed limiting first-order behavior with a kobs of 23 × 10−4 s−1 in 298 K. As has been pointed out previously [56], the similarity of the limiting rate constant to the rate of NO dissociation (2 × 10−4 s−1 ) from Mb(NO) suggests a mechanism for which NO dissociation is rate limiting. Regardless, since NO− 3 is the nitrogen product, the metal center must be involved in the eventual oxidation step, since uncatalyzed NO autoxidation in aqueous media gives nitrite not nitrate (Equation 4). A related study [145] of the autoxidation of nitrosyl hemoglobin also demonstrated that the kinetics do not depend on the O2 concentration and concluded that the mechanisms proceeds in three steps: rate-limiting NO dissociation from Hb(NO), rapid binding of O2 to Hb, and reaction of NO with oxyHb to give metHb and nitrate. •

5. SUMMARY As an overview, it is clear that the apparent plasticity of the ferrous and ferric heme models in their reactions with NO and other NOx is in part the result of the general ligand lability of these particularly for {Fe(NO)(X)}6 complexes. Understanding NOx interconversions as mediated by various Fe(Por) continues to provide a fertile and dynamic area for research. Novel interactions of NO2 , nitrite ion, and NO with heme model complexes and heme proteins have been characterized, and nitrite has been shown to participate in human blood pressure regulation. New found physiological effects of the higher nitrogen oxides have been and will continue to be the source of recent discussion and speculation. Even coordinated nitrate may participate in certain redox reactions. For example, ferrous hemes may be oxidized by nitrite to ferric hemes to give NO, while nitrite may in turn function as reducing agent with higher oxidation states of heme (forming NO2 ). It is clear that further study will be necessary to fully elucidate the biochemistry and physiology of these systems.

ACKNOWLEDGMENTS Research in these laboratories related to the reactions of metalloporphyrin complexes with nitrogen oxides was supported by grants from the U.S. National Science Foundation,

88

P.C. Ford et al.

the U.S. Department of Energy, the Petroleum Research Fund of the American Chemical Society, and the Civilian Research and Development Foundation (CRDF #AC2-2520TB-03). We are especially appreciative of the fruitful collaboration with Dr. Tigran Kurtikyan of the Armenian Academy of Sciences.

REFERENCES [1] Ignarro, L., Ed. (2000) Nitric Oxide: Biology and Pathobiology. San Diego: Academic Press. [2] Miranda, K.M. (2005) Coord. Chem. Rev., 249, 433–455. [3] Radi, R. (2004) Proc. Nat. Acad. Sci. USA, 101, 4003–4008. [4] Dedon, P.C. and Tannenbaum, S.R. (2004) Arch. Biochem. Biophys., 423, 12–22. [5] Mancardi, D., Ridnour, L.A., Thomas, D.D. et al. (2004) Curr. Mol. Med., 4, 723–740. [6] Gladwin, M.T., Schechter, A.N., Kim-Shapiro, D.B. et al. (2005) Nat. Chem. Biol., 1, 308–314. [7] Beckman, J.S. (2003) Critical Reviews of Oxidative Stress and Aging, 1, 54–83. [8] Bryan, N.S., Rassaf, T., Maloney, R.E. et al. (2004) Proc. Nat. Acad. Sci. USA, 101, 4308–4313. [9] Ford, P.C. and Lorkovic, I.M. (2002) Chem. Rev., 102, 993–1017. [10] Ignarro, L.J., Lippton, H., Edwards, J.C. et al. (1981) J. Pharmacol. Exp. Ther., 218, 739–749. [11] Brown, G.C. (2001) Biochim. Biophys. Acta, 1504, 46–57. [12] Carr, G.J. and Ferguson, S.J. (1990) Biochim. Biophys. Acta, 1017, 57–62. [13] Andersen, J.F., Ding, X.D., Balfour, C. et al. (2000) Biochemistry, 39, 10118–10131. [14] Feelisch, M., Rassaf, T., Mnaimneh, S. et al. (2002) FASEB J., 16, 1775–1785. [15] Rodriguez, J., Maloney, R.E., Rassaf, T. et al. (2003) Proc. Nat. Acad. Sci. USA, 100, 336–341. [16] Cosby, K., Partovi, K.S., Crawford, J.H. et al. (2003) Nat. Med., 9, 1498–1505. [17] Huang, K.T., Keszler, A., Patel, N. et al. (2005) J. Biol. Chem., 280, 31126. [18] Rifkind, J.M., Nagababu, E., and Ramasamy, S. (2006) Antioxid. Redox Signal., 8, 1193–1203. [19] Bellamy, T.C., Griffiths, C., and Garthwaite, J. (2002) J. Biol. Chem., 277, 31801–31807. [20] Malinski, T., Bailey, F., Zhang, Z.G., and Chopp, M. (1993) J. Cereb. Blood Flow Metab., 13, 355–358. [21] Doyle, M.P. and Hoekstra, J.W. (1981) J. Inorg. Biochem., 14, 351–358. [22] Herold, S., Exner, M., and Nauser, T. (2001) Biochemistry, 40, 3385–3395. [23] Shiva, S., Wang, X., Ringwood, L.A. et al. (2006) Nat. Chem. Biol., 2, 486–493. [24] Liu, X., Miller, M.J.S., Joshi, M.S. et al. (1998) Proc. Nat. Acad. Sci. USA., 95, 2175–2179. [25] Dierks, E.A. and Burstyn, J.N. (1998) Arch. Biochem. Biophys., 351, 1–7. [26] Kharitonov, V.G., Sharma, V.S., Magde, D., and Koesling, D. (1997) Biochemistry, 36, 6814–6818. [27] Sharma, V.S. and Magde, D. (1999) Methods, 19, 494–505. [28] Laverman, L.E. and Ford, P.C. (2001) J. Am. Chem. Soc., 123, 11614–11622. [29] Olson, J.S. and Phillips, G.N. (1996) J. Biol. Chem., 271, 17593. [30] Ikeda-Saito, M., Dou, Y., Yonetani, T. et al. (1993) J. Biol. Chem., 268, 6855. [31] Boon, E.M., Huang, S.H., and Marletta, M.A. (2005) Nat. Chem. Biol., 1, 53–59 [32] Boon, E.M., Huang, S.H., and Marletta, M.A. (2005) Curr. Opin. Chem. Biol., 9, 441–446. [33] Boon, E.M. and Marletta, M.A. (2006) J. Am. Chem. Soc., 128, 10022–10023. [34] Nioche, P., Berka, V., Vipond, J. et al. (2004) Science, 306, 1550–1553.

NO and NOx Interactions with Hemes [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75] [76]

89

Uchida, Y., Tsukahara, F., Ohba, K.-I. et al. (1997) Eur. J. Pharmacol., 335, 235–243. Garthwaite, J., Southam, E., Boulton, C. et al. (1995) Mol. Pharmacol., 48, 184–188. Ford, P.C., Wink, D.A., and Stanbury, D.M. (1993) FEBS Lett., 326, 1–3. Nottingham, W.C. and Sutter, J.R. (1986) Int. J. Chem. Kinet., 18, 1289–1302. Wink, D.A., Darbyshire, J.F., Nims, R.W. et al. (1993) Chem. Res. Toxicol., 6, 23–27. Huie, R.E. and Padmaja, S. (1993) Free Rad. Res. Commun., 18, 195–199. Kissner, R., Nauser, T., Bugnon, P. et al. (1997) Chem. Res. Toxicol., 10, 1285–1292. Pfeiffer, S., Gorren, A.C.F., Schmidt, K. et al. (1997) J. Biol. Chem., 272, 3465–3470. Goldstein, S., Czapski, G., Lind, J., and Merényi, G. (2001) Chem. Res. Toxicol., 14, 1273–1276. Stern, M.K., Jensen, M.P., and Kramer, K. (1996) J. Am. Chem. Soc., 118, 8735–8736. Herold, S., Matsui, T., and Watanabe, Y. (2001) J. Am. Chem. Soc., 123, 4085–4086. Gladwin, M.T., Shelhamer, J.H., Schechter, A.N. et al. (2000) Proc. Nat. Acad. Sci. USA, 97, 11482–11487. Crawford, J.H., Isbell, T.S., Huang, Z. et al. (2006) Blood, 107, 566–574. Doyle, M.P., Pickering, R.A., Deweert, T.M. et al. (1981) J. Biol. Chem., 256, 12393. Angelo, M., Singel, D.J., and Stamler, J.S. (2006) Proc. Nat. Acad. Sci. USA, 103, 8366–8371. Fernandez, B.O. and Ford, P.C. (2003) J. Am. Chem. Soc., 125, 10510–10511. Fernandez, B.O., Lorkovic, I.M., and Ford, P.C. (2004) Inorg. Chem., 43, 5393—5402. Hoshino, M., Maeda, M., Konishi, R. et al. (1996) J. Am. Chem. Soc., 118, 5702–5707. Cheng, L. and Richter-Addo, G.B. (2000). Porphyrin Handbook (K.M. Kadish, K.M. Smith, and R. Guilard, eds) San Diego, CA: Academic Press. Richter-Addo, G.B. and Legzdins, P. (1992) Metal Nitrosyls. New York: Oxford University Press. Richter-Addo, G.B. (2000) J. Porphyrins Phthalocyanines, 4, 354–357. Hoshino, M., Laverman, L., and Ford, P.C. (1999) Coord. Chem. Rev., 187, 75–102. Wyllie, G.R.A. and Scheidt, W.R. (2002) Chem. Rev., 102, 1067–1090. Moeller, J.K.S. and Skibsted, L.H. (2002) Chem. Revs., 102, 1167–1178. Ford, P.C., Fernandez, B.O., and Lim, M.D. (2005) Chem. Revs., 105, 2439–2455. Ford, P.C. and Wecksler, S. (2005) Coord. Chem. Rev., 249, 1382–1395 Enemark, J.H. and Feltham, R.D. (1974) Coord. Chem. Rev., 13, 339–406. Wayland, B.B. and Olson, L.W. (1974) J. Am. Chem. Soc., 96, 6037–6041. Scheidt, W.R. and Frisse, M.E. (1975) J. Am. Chem. Soc., 97, 17–21. Lorkovic, I.M. and Ford, P.C. (2000) J. Am. Chem. Soc., 122, 6516–6517. Patterson, J.C., Lorkovic, I.M., and Ford, P.C. (2003) Inorg. Chem., 42, 4902–4908. Lim, M.D., Lorkovic, I.M., Wedeking, K. et al. (2002) J. Am. Chem. Soc., 124, 9737–9743. Phillippi, M.A., Baenziger, N., and Goff, H.M. (1981) Inorg. Chem., 20, 3904–3911. Kurtikyan, T.S., Martirosyan, G.G., Hakobyan, M.E., and Ford, P.C. (2003) Chem. Commun., 13, 1706–1707. Ghosh, A. and Wondimagegn, T. (2000) J. Am. Chem. Soc., 122, 8101–8102. Copeland, D.M., Soares, A.S., West, A.H., and Richter-Addo, G.B. (2006) J. Inorg. Biochem., 100, 1413–1425. Leu, B.M., Zgierski, M.Z., Wyllie, G.R.A. et al. (2004) J. Am. Chem. Soc., 126, 4211–4227. Kurtikyan, T.S., Martirosyan, G.G., Lorkovic, I.M., and Ford, P.C. (2002) J. Am. Chem. Soc., 124, 10124–10129. Bari, S.E., Martí, M.A., Amorebieta, V.T. et al. (2003) J. Am. Chem. Soc., 125, 15272–15684. Miranda, K.M., Nims, R.W., Thomas, D.D. et al. (2003) J. Inorg. Biochem., 93, 52–60. Bazylinski, D.A. and Hollocher, T.C. (1986) J. Am. Chem. Soc., 107, 7982–7986. Hoshino, M. and Kogure, M. (1989) J. Phys. Chem., 93, 5478–5484.

90 [77] [78] [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95] [96] [97] [98] [99] [100] [101] [102] [103] [104] [105] [106] [107] [108] [109] [110] [111] [112] [113] [114] [115] [116] [117] [118]

P.C. Ford et al. Cheng, L., Novozhilova, I., Kim, C. et al. (2000) J. Am. Chem. Soc., 122, 7142–7143. Wyllie, G.R.A., Schulz, C.E., and Scheidt, W.R. (2003) Inorg. Chem., 42, 5722–5734. Bohle, D.S. and Hung, C.-H. (1995) J. Am. Chem. Soc., 117, 9584–9585. Cheng, L., Powell, D.R., Khan, M.A., and Richter-Addo, G.B. (2000) Chem. Commun., 23, 2301–2302. Yoshimura, T. (1984) Inorg. Chim. Acta, 83, 17–21. Lin, R. and Farmer, P.J. (2001) J. Am. Chem. Soc., 123, 1143–1150. Miranda, K.M., Bu, X., Lorkovic, I., and Ford, P.C. (1997) Inorg. Chem., 36, 4838–4848. Lorkovic, I. and Ford, P.C. (1999) Inorg. Chem., 38, 1467–1473. Kadish, K.M., Adamian, V.A., Caemelbecke, E.V. et al. (1996) Inorg. Chem., 35, 1343–1348. Leal, F.A., Lorkovic, I.M., Ford, P.C. et al. (2003) Can. J. Chem., 81, 872–881. Martirosyan, G.G., Azizyan, A.S., Kurtikyan, T.S., and Ford, P.C. (2004) Chem. Commun., 13, 1488–1489. Lorkovic, I.M. and Ford, P.C. (2000) Inorg. Chem., 39, 632–633. Kurtikyan, T.S., Stephanyan, T.G., and Akopyan, M.E. (1999) Russ. J. Coord. Chem., 25, 721–725. Conradie, J., Wondimagegn, T., and Ghosh, A. (2003) J. Am. Chem. Soc., 125, 4968–4969. Settin, M.F. and Fanning, J.C. (1988) Inorg. Chem., 27, 1431–1435. Ellison, M.K., Schulz, C.E., and Scheidt, W.R. (1999) Inorg. Chem., 38, 100–108. Lim, M.D. (2004) Ph.D. Dissertation. Santa Barbara: University of California. Patterson, J. (2004) Ph.D. Dissertation. Santa Barbara: University of California. Kurtikyan, T.S., Hovhannisyan, A.A., Hakobyan, M.E. et al. (2007) J. Am. Chem. Soc. 129, 3576–3585. Lee, J., Kovalevsky, A.Y., Novozhilova, I.V. et al. (2004) J. Am. Chem. Soc., 126, 7180–7181. Novozhilova, I.V., Coppens, P., Lee, J. et al. (2006) J. Am. Chem. Soc., 128, 2093–2104. De Leo, M. and Ford, P.C. (2000) Coord. Chem. Revs., 208, 47–59. Kosaka, H., Tyuma, I., and Imaizumi, K. (1983) Biomed. Biochim. Acta, 42, 144–148. Firestone, M.K. and Davidson, E.A. (1989) Life Sci. Res. Rep., 47, 7–21. Nasri, H., Wang, Y., Hanh, H.B., and Scheidt, W.R. (1991) J. Am. Chem. Soc., 113, 717–719. Castro, C.E. (1996) J. Am. Chem. Soc., 118, 3984–3985. O’Shea, S.K., Wang, W., Wade, R.S., and Castro, C.E. (1996) J. Org. Chem., 61, 6388–6395. Finnegan, M.G., Lappin, A.G., and Scheidt, W.R. (1990) Inorg. Chem., 29, 181–185. Munro, O.Q. and Scheidt, W.R. (1998) Inorg. Chem., 37, 2308–2316. Nasri, H., Ellison, M.K., Shang, M. et al. (2004) Inorg. Chem., 43, 2932–2942 Kurtikyan, T.S. and Ford, P.C. (2006) Angew. Chem. Int. Ed., 45, 492–496. Reed, C.A. and Guiset, F. (1996) J. Am. Chem. Soc., 118, 3281–3282. Ellison, M.K., Shang, M., Kim, J., and Scheidt, W.R. (1996) Acta Cryst., C52, 3040–3043. Kurtikyan, T.S., Gulyan, G.M., Martirosyan, G.G. et al. (2005) J. Am. Chem. Soc., 127, 6216–6224. Moore, E.G. and Gibson, Q.H. (1976) J. Biol. Chem., 251, 2788. Rose, E.J. and Hoffman, E.B.M. (1983) J. Am. Chem. Soc., 105, 2866. Hoshino, M., Ozawa, K., Seki, H., and Ford, P.C. (1993) J. Am. Chem. Soc., 115, 9568–9575. Laverman, L.E., Hoshino, M., and Ford, P.C. (1997) J. Am. Chem. Soc., 119, 12663. Laverman, L.E., Wanat, A., Oszajca, J. et al. (2001) J. Am. Chem. Soc., 123, 285. Schneppensieper, T., Zahl, A., and van Eldik, R. (2001) Angew. Chemie. Int. Ed. Eng., 40, 1678. Wolak, M. and van Eldik, R. (2005) J. Am. Chem. Soc., 127, 13312–13315. Jee, J., Eigler, S., Hampel, F. et al. (2005) Inorg. Chem., 44, 7717–7731.

NO and NOx Interactions with Hemes

91

[119] Fernandez, B.O., Lorkovic, I.M., and Ford, P.C. (2003) Inorg. Chem., 42, 2–4. [120] Reichenbach, G., Sabatini, S., Palombari, R., and Palmerini, C.A. (2001) Nitric Oxide, 5, 395. [121] Luchsinger, B.P., Rich, E.N., Gow, A.J. et al. (2003) Proc. Nat. Acad. Sci. USA, 100, 461–465. [122] Stamler, J.S., Jaraki, O., Osborne, J. et al. (1992) Proc. Nat. Acad. Sci. USA, 89, 7674–7677. [123] Gladwin, M.T., Ognibene, F.P., Pannell, L.K. et al. (2000) Proc. Natl. Acad. Sci. USA., 97, 9943–9948. [124] Chan, N.-L., Rogers, P.H., and Arnone, A. (1998) Biochemistry, 37, 16459. [125] Han, T.H., Hyduke, D.R., Vaughn, M.W. et al. (2002) Proc. Nat. Acad. Sci. USA, 99, 7763. [126] Herold, S. and Röck, G. (2003) J. Biol. Chem., 278, 6623–6634. [127] Han, T.H., Fukuto, J.M., and Liao, J.C. (2004) Nitric Oxide Biol. Chem., 10, 74. [128] Feelisch, M. (1991) Cardiovas. Pharmacol., 17, S25. [129] Zhang, Y. and Hogg, N. (2002/2004) Free Rad. Biol. Med., 32, 1212; 36, 947. [130] Herold, S. and Röck, G. (2005) Arch. Biochem. Biophys., 436, 386–396. [131] Weichsel, A., Maes, E.M., Andersen, J.F. et al. (2005) Proc. Nat. Acad. Sci. USA, 102, 594–599. [132] Franke, A., Stochel, G., Suzuki, N. et al. (2005) J. Am. Chem. Soc., 127, 5360–5375. [133] Tsuge, K., DeRosa, F., Lim, M.D., and Ford, P.C. (2004) J. Am. Chem. Soc., 126, 6564–6565. [134] Williams, D.L. (1996) Meth. in Enzymol., 268, 299. [135] Toubin, C., Yeung, D.Y.H., English, A.M., and Peslherbe, G.H. (2002) J. Am. Chem. Soc., 124, 14816. [136] Andreasen, L.V., Lorkovic, I. M., Richter-Addo, G.B., and Ford, P.C. (2002) Nitric Oxide, 6, 228–235. [137] Jensen, M.P. and Riley, D.P. (2002) Inorg. Chem., 41, 4788–4797. [138] Lee, J., Hunt, J.A., and Groves, J.T. (1998) J. Am. Chem. Soc., 120, 7493–7501. [139] Herold, S., Kalinga, S., Matsui, T., and Watanabe, Y. (2004) J. Am. Chem. Soc., 126, 6945–6955. [140] Herold, S. (2004) Inorg. Chem., 43, 3783–3785. [141] Boccini, F. and Herold, S. (2004) Biochemistry, 43, 16393–16404. [142] Goldstein, S., Merenyi, G., and Samuni, A. (2004) J. Am. Chem. Soc., 126, 15694–15701. [143] Andersen, H.J. and Skibsted, L.H. (1992) J. Agricul. Food Chem., 40, 1741. [144] Moeller, J.K.S. and Skibsted, L.H. (2004) Chemistry-A European Journal, 10, 2291–2300. [145] Herold, S. and Röck, G. (2005) Biochemistry, 44, 6223–6231.

This page intentionally left blank

Part II Electronic Structure and Spectroscopy

This page intentionally left blank

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 4

CO, NO, and O2 as Vibrational Probes of Heme Protein Active Sites Thomas G. Spiro, Mohammed Ibrahim, and Ingar H. Wasbottena Department of Chemistry, Princeton University, Princeton, NJ 08544, USA a Department of Chemistry, University of Tromsø, N-9013 Tromsø, NORWAY

Abstract Carbon monoxide is a useful vibrational probe of heme-binding sites in proteins, because FeCO backbonding is modulated by polar interactions with protein residues, and by variations in the donor strength of the trans ligand. This modulation is sensitively monitored by the CO and FeC stretching frequencies, which are readily detectable in infrared and resonance Raman spectra. The two frequencies are negatively correlated, and the FeC/CO position along the correlation line reflects the type and strength of distal polar interactions. Changes in the trans ligand donor strength shift the correlation to higher or lower positions. Illustrative applications of the FeC/CO diagram are reviewed for proteins bearing histidine and thiolate axial ligands. Steric crowding has not been found to affect the FeC/CO correlations significantly, except in the special case of cytochrome oxidase, where the heme-bound CO may interact with the nearby CuB center. NO adducts of Fe(II) heme proteins also show variations in the FeN and NO frequencies, but the data show considerable scatter, in contrast to the CO adduct data. However, protein-free Fe(II)porphyrin NO adducts give well-behaved backbonding correlations; new data show this to be true of 6-coordinate (6-c) as well as 5-coordinate (5-c) adducts. The scatter in the protein data is suggested to reflect changes in the FeNO angle induced by distal polar groups, especially histidine. The few data available for NO adducts of Fe(III) heme proteins suggest a weak negative FeN/NO correlation when the proximal ligand is histidine, but a positive correlation when the proximal ligand is thiolate. This behavior is markedly different from that of the isoelectronic Fe(II)CO. DFT modeling indicates that the altered response reflects a change in frontier orbitals resulting from the lowered dz2 energy in Fe(III). A similar pattern is suggested for Fe(II)O2 adducts from the limited available data. Keywords: Carbon monoxide, Nitric oxide, Dioxygen, Heme, Protein, Vibration, Raman.

ABBREVIATIONS IR RR DFT cyt P450cam

infrared Resonance Raman Density functional theory cytochrome P450 camphor hydroxylase

96

Mb Hb NOS CPO HRP TPP PPDME CCP cyt ox N-MeIm

T.G. Spiro et al.

myoglobin hemoglobin nitric oxide synthase chloroperoxidase horseradish peroxidase tetraphenylporphyrin protoporphyrin IX dimethylester cytochrome c peroxidase cytochrome c oxidase N -methylimidazole

1. INTRODUCTION The C O stretching frequency has long been used as a monitor of structure and bonding in transition metal carbonyl complexes, including CO adducts of the heme group. When bound to heme, the CO frequency diminishes by ∼200 cm−1 from its gas phase value, 2143 cm−1 , and falls in a region of the vibrational spectrum that is relatively free of interferences from other molecular vibrations. At the same time, the CO infrared intensity is greatly augmented, as is the frequency shift in an applied electric field [1]. These effects are due to backbonding. Fe[II] d electrons are donated to the empty CO  ∗ orbitals, thereby diminishing the CO bond order. The transition dipole moment is increased because compressing and expanding the CO bond shifts electrons back and forth between the CO and Fe[II] orbitals. The CO frequency is sensitive to the molecular environment of the bound CO because the extent of backbonding is readily altered. In particular, the presence of polar molecules or protein residues near the CO has a marked effect [2]. Consequently, there has been considerable interest in using CO as a probe of the protein groups in the heme-binding pocket; these groups can determine the reactivity of the heme toward other ligands, particularly O2 and NO. In addition to the CO frequency, the IR spectrum can provide the direction of the transition moment via polarization measurements. These have been carried out for the CO adduct of myoglobin, MbCO, in oriented single crystals [3], or via photoselection of partially photolyzed molecules, either in frozen solution [4] or by using picosecond laser pulses [5,6], which are shorter than the protein tumbling time in liquid solution. It was hoped that the polarization measurements would establish the degree of FeCO bending, a historically contentious issue [7]. However, the assumption that the transition moment lies along the CO bond vector is incorrect; due to the backbonding, CO stretching induces electrons to move throughout the  system, and the transition moment lies essentially along the FeC bond vector, as shown by DFT calculations [8,9]. Thus, the observation that the transition moment is within 7 of the heme normal [3,6] is not inconsistent with modest FeCO distortion, as seen in recent crystal structures [7]. The CO vibration is also influenced by the nature of the trans axial ligand of the heme-CO adduct [10,11]. In heme proteins, the proximal ligand is usually histidine, but other ligands are possible. In addition, there can be variations in the strength of the proximal ligand bond, due to H-bonding or mechanical strain. Measurement of CO alone cannot distinguish proximal from distal effects, but these effects can be unraveled if

CO, NO, and O2 as Vibrational Probes

97

the FeC vibration, near 500 cm−1 , is also monitored [10]. For a given axial ligand, a plot of FeC against CO forms a line with negative slope, as expected from backbonding, but altering the axial donor strength displaces the line. FeC is difficult to detect in the IR spectrum, but it is readily observed in the resonance Raman (RR) spectrum [12], which can simultaneously reveal CO. Thus, RR spectroscopy has emerged as a widely used technique for studying CO adducts of heme proteins. In addition to the FeC and CO vibrations, the RR spectrum sometimes reveals a band near 570 cm−1 , assignable to Fe C O bending, FeCO [12]. Actually, this mode is an out-of-phase combination of Fe C O bending and Fe C tilting coordinates, as described by Ghosh and Bocian [13], whose DFT calculation revealed a large bend-tilt interaction constant. This interaction accounts for the elevated FeCO frequency (which had been a source of controversy over the assignment [14]). The in-phase combination is predicted [13] at a correspondingly low frequency (84 cm−1 ), and has not been observed. Both modes break the fourfold approximate symmetry of the heme group, and are expected to be RR-active only to the extent that this symmetry is lowered by the heme environment [10]. Indeed, the band is not detectable in protein-free heme adducts (unless they have covalent superstructures that can interact with the bound CO [15]), or in many heme proteins. RR intensity is an indicator of off-axis interactions with nearby groups, which could be steric in character, but are more likely to be electrostatic [15]. The FeCO values have been found to correlate with FeC [16], but the variation in FeCO is small. The vibrational frequencies of NO adducts can provide additional information about the heme environment [17]. However, the NO and FeN frequencies are not obviously correlated among heme protein NO adducts, leading Boxer and coworkers to question whether there is significant backbonding to NO [18,19]. The issue is complicated by mixing between Fe N stretching and Fe N O bending coordinates, resulting from the naturally bent FeNO angle, raising the question whether the ∼550 cm−1 15 N-sensitive RR band, which is generally assigned to FeN, really represents Fe N stretching [20]. However, the data on protein-free adducts establish that this band is indeed negatively correlated with NO [21,22]. The poor correlations among heme protein adducts can be traced to the character of the distal residue, and probably reflect the ease with which the FeNO angle can be altered by steric and/or polar forces [21,22]. Limited vibrational data have also been reported for Fe[III]NO and Fe[II]O2 adducts. These data are difficult to obtain for technical reasons. Fe[III]NO adducts readily autoreduce to Fe[II]NO adducts, while the OO band of Fe[II]O2 adducts falls in a crowded region of the vibrational spectrum. In both cases, it appears that FeX/XO correlations are negative, as expected for nitrogenous axial ligands, but they become positive when the axial ligand is thiolate. However, more data are needed to clarify this interesting dichotomy. It is possible to detect the ligand vibrational frequencies for Fe[III](CN− ) adducts of heme proteins (see [23] and references therein). Although isoelectronic with Fe[II]CO adducts, Fe[III](CN− ) adducts experience much less backbonding, and the spread of frequencies is small [24]. However, the Fe[III](CN− ) is easier to bend, and bent forms can be detected, giving indications of steric and/or electrostatic distal interactions [23]. This chapter focuses on the uses of FeCO and FeNO vibrational data to elucidate binding-site interactions in heme proteins, with special attention to the separation of distal and proximal effects. Reports of such data are already voluminous, and we have

98

T.G. Spiro et al.

not attempted a comprehensive compilation. Rather, prototypical examples have been chosen to illustrate the reasoning behind structural interpretation of the spectroscopic data. Regularities in the available data for Fe[III]NO and Fe[II]O2 adducts are also examined briefly.

2. FeCO VIBRATIONS 2.1. The FeC Versus CO Backbonding Correlations A negative correlation between FeC and CO was noted early by several workers [16,25], and has been revisited many times. To a good approximation, the frequencies correlate linearly (as do their squares [16], which more properly represent the force constants), provided that the proximal ligand is unchanged [10]. Most of the available data is for heme proteins with proximal histidine and for protein-free hemes with imidazole or pyridine trans to the CO. All of these data are more or less on the same line, which covers a range of ∼100 cm−1 in CO and ∼70 cm−1 in FeC [10,15]; even hemes with thioether ligands fall approximately on this line [26]. Essentially all neutral trans ligands seem to behave similarly. An extensive series of myoglobin variants with distal pocket mutations [2,25,27–42], which together with protein-free adducts, form the most comprehensive available set of FeC/CO data (Table 1) (Figs. 1–3 and 5–7) at constant trans ligation. The Mb and Table 1. FeC /CO and FeN /NO data for Fe(II) heme proteins and models (Figs. 1–3 and 5–7) Sample

(Fe C) (cm−1 )

(C O) (cm−1 )

Reference

(Fe N) (cm−1 )

(N O) (cm−1 )

530 — 527 524 524 523 518 515 514

1663 — 1666 1675 1678 1686 1689 1695 1703

[17] — [17] [17] [17] [17] [17] [17] [17]

— 587 586 582 576 574 577 —

— 1621 1622 1623 1630 1632 1632 —

— [22] [22] [22] [22] [22] [22] —

Reference

Five-coordinated XO adducts of Fe(II)TPP-Y OH NH2 OCH3 CH3 H CN 2,6-dichloro 2,6-difluoro Pentafluoro

527 527 526 — 525 521 516 521 —

1947 1948 1949 — 1951 1957 1966 1963 —

[11] [11] [11] — [11] [11] [11] [11] —

Six-coordinated XO adducts of (N-MeIm) Fe(II)TPP-Y OH OCH3 CH3 H 2,6-dichloro 2,6-difluoro Pentafluoro NO2

501 498 498 495 — — — 492

1953 1960 1961 1962 — — — 1966

[22] [22] [22] [22] — — — [22]

CO, NO, and O2 as Vibrational Probes

99

Table 1. (Continued) Sample

(Fe C) (C O) (cm−1 ) (cm−1 )

Reference

(Fe N) (cm−1 )

(N O) Reference (cm−1 )

MbII XO variants (subscripts: E = elephant, sw = sperm whale, p = pig, h = human) WTE WTSW (pH 8.4) WTSW (pH 7.0)

515 512 508

1937 1944 1945

WTSW (pH 2.6–4) WTh WTp H64Qsw H64Gsw H64ASW (H64Ah ) H64VSW (H64Vh ) H64ISW (H64Ih ) H64LSW (H64Lh ) V68Tp V68Fsw V68Nsw F46Asw F46Lsw F46Vsw L29Fsw L29F/H64Qsw H64V/V68Tp HbASC

489 508 508 507 492 490 488 490 490 493 — 526 — — 489 525 513 479 543

1966 1941 1944 1944 1965 1966 1967 1968 1965 1961 1940 1922 1963 1962 1962 1932 1938 1984 1909

[31,40] [35] [2,25,27,28, 36,37,40,42] [28,29,36,38] [2,32,39] [2,33,37] [40,42] [41] [2,32,37,39] [2,32,39,42] [2,39] [2,27,32,37,39] [2,37,42] [2] [2,37,42] [2] [2] [2,37,42] [2,37,40,42] [40] [33,34,37,42] [72]

— — 550

— — 1614

— — [21]

(pH 4) 524 552 551 555 554 555 557 558 563 548 570 551 551 550 552 551 558 553 —

1668 1612 1612 1619 1633 1631 1634 1638 1635 1632 1605 1595 1636 1619 1623 1601 1614 1631 —

[43] [21] [21] [21] [21] [21] [21] [21] [21] [21] [21] [21] [21] [21] [21] [21] [21] [21] —

Cyt P450cam adducts (with substrate, as indicated; ns = no substrate) P450cam + TMCH P450cam + camphor P450cam + CPRQ P450cam + fenchone P450cam + norcamphor P450cam + adamantanone P450cam (ns)

485 481 476 480 473

1934 1940 1941 1945 1947

[44,45] [44–50] [44,45] [44,45] [44,45]

— 553 — — 545

— — — — —

— [50,52] — — [52]

474

1942

[44,51]

554

1591

[52]

464

1963

[44–48]

547

—

[52]

Nitric oxide synthase and chloroperoxidase CPO, pH 6.1 nNOS, l-Arginine-free,  nNOS, l-Arginine-free,  nNOS, l-Arginine-free nNOS + l-Arginine

485 487

1957.5 [48,53,54] 1949 [54,55]

543 —

— —

[53] —

501

1930

[54,55]

—

—

—

491

1936

[56]

—

—

—

503

1929

[54]

—

—

— (Continued)

100

T.G. Spiro et al.

Table 1. (Continued) Sample nNOS + N /OH lArginine iNOS, l-Arginine-free iNOS + l-Arginine iNOSoxy , l-Arginine-free iNOSoxy + lArginine P450cam , L358P + camphor

(Fe C) (cm−1 )

(C O) (cm−1 )

(Fe N) (cm−1 )

(N O) (cm−1 )

502

1928

[54]

—

—

487

1945

[56]

—

—

—

512 491

1906 1946

[56] [57]

540 —

— —

[57] —

512

1907

[57]

540

—

[57]

489

1936

[81]

—

—

—

495 530 507 503 531 490 539 516 530 547 526

1922 1922 1948 1922 1933 1932 1906 1933 1933 1911 1952

[58] [58] [59] [59] [59] [60] [16,60,61] [16,61] [16,61] [16] [62]

490 495

1942 1960

[60] [60]

487 500

1982 1964

[63] [63]

1966 1966 1989 1967 1973 1969 1972 1973 1971 1972

[64] [66] [67] [68] [68] [68] [68] [69] [71] [71]

579 569 553 568 560 563 567 539 569 573

1624 1625 1655 1634 1637 1632 1639 1620 1604 1600

Reference

Reference —

Peroxidases CCP(I) CCP(II) CCPMI(alk) CCPMI CCPMI(D235N) HRP(I) HRP(II) HRP(III) HRP(alk) HRP(BHA) HRPEG(Bz)

Models (in methylene chloride) PPDMe(Im− ) PPMeImH CooA CooA CooA(alk)

Other His-ligated heme proteins AXCP RCCP TtTar4H BjFixLH AxPDEA1H EcDosH MtDosH Cyt ox (T.t.) Cytoglobin (Cgb) Neuroglobin (Ngb)

491 494 490 497 493 487 494 507 492 494

[65] [66] [67] [68] [68] [68] [68] [70] [71] [71]

CO, NO, and O2 as Vibrational Probes

101 FeCO Backbonding Correlations O

HbASC

540

C Fe

530

C Fe

WTE

νFeC (cm–1)

N

2,6-F2

OCH3

CN

5-c TPP-Y

WTSW

H O

H(CH2Cl2)

2,6-Cl2

N

510

H

OH

V68NSW

520

500

NH2

O

H

OCH3

WTp

O

OH

490

O

Fee Fe F S

470

NO2

TMCH camphor

C

480

CH3 H V68TP

H64Vh fenchone WTSW(pH 2.6)

CPRQ adamantanone norcamphor

H64V/68Tp

Mb and 6-c TPP-Y ns

P450 460 1900

1910

1920

1930

1940

1950

1960

1970

1980

νCO (cm–1)

Fig. 1. Canonical FeCO backbonding correlations. Data and references are in Table 1. The top line is for 5-c CO adducts of Fe(II)TPP, with the indicated phenyl substituents, Y, in DMF and CH2 Cl2 solvents. The middle line is for the indicated variants of myoglobin, and for (N-MeIm)Fe(II)TPPY(CO) adducts in CH2 Cl2 . Also shown is Ascaris hemoglobin (HbASC ). The bottom line is for cyt P450cam with the indicated substrates. Inset figures show the trans ligand, with H-bonding indicated by dotted lines.

imidazole adduct data are plotted in Fig. 1 and form the canonical Mb line, against which other adducts can be compared. The line can be expressed by the following equation [11] FeC =   FeC − s  CO −   CO

(1)

where  CO is the standard triple-bonded value in the gas phase, 2145 cm−1 , and  FeC is the corresponding single-bonded value for Fe C. A least squares fit gives  FeC = 346 cm−1 for Mb, and s = −081 (Table 2). The slope represents the backbonding sensitivity of FeC. For imidazole adducts, including Mb, FeC changes by 4/5 of the CO change for a given increment of backbonding. A large variation in backbonding is evident for the Mb variants and can be understood in terms of altered polar interactions with distal residues [2,15]. In the middle of the line [CO ∼ 1945 cm−1 ] are the wild-type proteins, for which moderate backbonding is induced by weak H-bonding from the distal histidine (H64) residue. This H-bond is evidenced by a 2 cm−1 CO shift in D2 O [27], although it is worth only about 0.5 kcal/mol in energy [7]. When the distal histidine is replaced by nonpolar residues (e.g., H64V), the FeC/CO point slides down the line [CO ∼ 1965 cm−1 ], reflecting the expected decrease in backbonding. The wild-type protein also occupies this location

102

T.G. Spiro et al.

540

Mb

Proximal Thiolate

530

520

(cm–1)

500

NOS O

nNOS + Arg

C

νFe

–

510

C

iNOS + Arg

Fe Fe Fe

490

480

nNOS

–

L358P + cam

S

iNOS

CPO

TrpNH P450

470

460 1900

1910

1920

1930

1940

1950

1960

νCO (cm–1)

Fig. 2. Backbonding correlation (data in Table 1) for CO adducts of nitric oxide synthase (NOS) isoforms with and without bound arginine, in comparison with cyt P450cam . Also plotted are chloroperoxidase (CPO) and the L358P variant of cyt P450cam . The inset emphasizes the Trp side chain H-bond to the proximal thiolate ligand in NOS.

if the distal histidine is protonated at low pH [28,29], because the positively charged side chain swings out into solution [30], leaving the bound CO in a hydrophobic pocket. Elephant Mb has a glutamine in place of H64 [31,40], which evidently forms a somewhat stronger H-bond, moving the point up the line relative to the distal histidinecontaining species. Further, H-bonding can be induced by introducing additional polar groups, as in V68N; Val68 is adjacent to the bound CO, and the Asn replacement interacts strongly, pushing CO down to 1922 cm−1 [2,42]. Also falling near the Mb line, at its top end, is a hemoglobin from the Ascaris nematode [72]. It has distal glutamine and tyrosine residues, both of which are positioned to H-bond with the CO, leading to CO = 1909 cm−1 and a correspondingly high FeC. The low end of the Mb line is occupied by the double mutant H64V/V68T [33,34,37,42], in which the introduced threonine residue is oriented (via H-bonding to a backbone carbonyl) so that the O atom lone pairs point at the CO, providing negative polarity, and diminished backbonding. (Similar frequencies have been reported for superstructured model porphyrins, in which naphtholic hydroxyl groups are positioned over the bound CO [73].) Thus, a complete range of backbonding, from strongly positive to distinctly negative polarity is represented in the Mb correlation. This dependence on distal polarity has been put on a quantitative basis by Phillips et al. [42], who calculated the electrostatic potential at the bound CO for 20 Mb variants using crystallographic coordinates and a linearized Poisson–Boltzman method. The potentials all correlated directly with FeC and inversely with CO. The sensitivity of

CO, NO, and O2 as Vibrational Probes

103 Proximal Histidine

550

540

HRP(BHA) HRP(II) CCP(II) CCPMI(D235N)

530

HRPEG(Bz)

HRP(alk)

O

νFe – C (cm–1)

520

C

5c

HRP(III) 510

Fe Fe

CCPMI(Alk)

N

CooA(alk) N

CCPMI

500

CCP(I)

H

HRP(I)

PPDMeImH

490

?

PPDMe(Im–) CooA

480

Mb 470 1900

P450 1910

1920

1930

1940

1950

1960

1970

1980

νC – O (cm–1)

Fig. 3. FeC/CO data for peroxidase and CooA forms (Table 1), showing divergences from the Mb line, which indicate strengthening or weakening of the Fe His bond. Table 2. Backbonding parametersa for Fe(II) porphyrin CO and NO adducts Compound Five-coordinate Six-coordinate modelsb Mb P450 NOS a b

 (Fe CO) (cm−1 )

Slope (s)

 (Fe NO) (cm−1 )

Slope (s)

435 — 346 339 369

−0.46 — −0.81 −0.68 −0.60

445 329 — — —

−0.40 −1.0 — — —

   FeX = FeX − s XO − XO

XO (gas phase) = 2145 cm−1 for CO and 1876 cm−1 for NO. For CO, the range of model values is small (Fig. 5) and the points are within experimental error of the Mb line (Fig. 1).

these frequencies to distal electrostatic interactions has also been investigated in model computations by Kushkuley and Stavrov [74,75], using semiempirical methods, and by Franzen, using DFT [76]. An interesting finding of the latter study is that H-bond donors are expected to interact preferentially on the side, rather than the top, of the CO ligand. This is indeed the orientation seen for the distal histidine in high-resolution crystal structures of MbCO [77,78], and is reproduced in a full quantum computation of the MbCO heme site [79]. This preference for side on H-bonding may account for the observation that the FeCO mode is usually RR-detectable only for adducts with elevated positions on the backbonding line, indicating positive polar interactions that are off-axis [15].

104

T.G. Spiro et al.

In the case of protein-free adducts, variations in backbonding can be traced to different peripheral substituents. For example, the alkyl groups in C -substituted octa-alkyl porphyrins, including heme itself, are more electron-donating than the phenyl groups in Cm -substituted tetraphenylporphyrin (TPP), and augment the backbonding. This property has been used to deliberately tune backbonding via electron-donating and withdrawing substituents, Y, in a series of TPP-Y porphines (Table 1), in order to establish protein-free correlations for both CO and NO (see below) adducts. Data on the class of cytochrome P450cam adducts fall below the Mb line, and form a line of their own (Fig. 1). The slope is lower, −0.68, although the intercept,  FeC = 339 cm−1 , is similar to Mb (Table 2). The proximal ligand in cyt P450cam is a cysteine thiolate, which is a stronger donor than imidazole. Increased electron donation is important for the enzyme mechanism, which involves heterolytic cleavage of bound O2 [80,81]. The increased electron density on the heme Fe enhances backdonation, lowering CO, but the accompanying increase in FeC is countered by the trans thiolate ligand, which competes with the CO for sigma overlap with the Fe dz2 orbital [10]. For a given degree of backbonding, FeC is lower when the trans ligand is thiolate than when it is imidazole (although the similar intercepts suggest that in the absence of backbonding, there would be little difference). The cyt P450cam points are spread out along the lower line because they include complexes with a series of substrates and inhibitors [44–48,50–52] that bind in a cavity next to the heme, forcing water out, and inducing an interaction of the bound CO with the H-bonding residues, Thr252 and Asp251, which are believed to act as a proton relay in the O2 cleavage reaction [82,83]. This interaction is induced to different extents in the different adducts, thereby varying the backdonation that defines the cyt P450cam correlation. On the same reasoning, CO adducts with trans ligands that are weaker donors than imidazole should lie above the Mb backbonding line. The ultimate in weak donation is no ligand at all, i.e., a 5-c CO adduct. In order to examine the backbonding correlation for 5-c adducts, Vogel et al. [11] recorded RR spectra for CO adducts of a series of tetraphenylporphyrins bearing substituents of variable electron donating and withdrawing ability. These data established the 5-c line shown in Fig. 1. Although variations in peripheral substituents are not the same thing as variations in distal polar interactions, these substituents do influence the  electron density on the Fe, via the porphyrin  orbitals, and therefore modulate the extent of backdonation to CO. The line in Fig. 1 has a slope of −0.46 and an intercept  FeC = 435 cm−1 . This intercept is much higher than those of 6-c hemes (Table 2), suggesting a substantially strengthened Fe C bond when a trans ligand is absent, regardless of backbonding. The effect of trans ligand variation on the backbonding correlations has been examined computationally via DFT calculations on substituted Fe-porphine CO adducts with imidazole or thiolate or no trans ligands [11,22]. The hierarchy of negative straight lines was reproduced, as was the relative variation in the slopes.

2.2. Thiolate Modulation: H-Bonding in NOS and CPO The family of nitric oxide synthases (NOS) also have thiolate axial ligation of the heme, but the FeC/CO data [54–57] (Table 1) fall on a separate line, below Mb, but distinctly higher than cyt P450cam (Fig. 2). The slope is similar as for cyt P450cam , but the intercept

CO, NO, and O2 as Vibrational Probes

105

is higher. The NOS data are spread even more widely along the CO axis, because the substrate, arginine, can interact strongly with the CO via its positively charged guanidinium group. Also, the distal pocket itself is more polar than in cyt P450cam . There are three distinct isoforms of the human enzyme, nNOS, which is expressed in neurons, eNOS, from endothelial cells, and iNOS, the form inducible in macrophages [84]. These display different interactions with the bound CO. Why do NOS and cyt P450cam fall on separate FeC/CO lines? Although both have cysteinate ligands trans to the CO, the negatively charged sulfur atom is the recipient of different kinds of H-bonds. In cyt P450cam , there are H-bonds from three backbone amide NH groups [85], while in NOS there are two such H-bonds plus an additional H-bond from a tryptophan side chain [86–88]. This H-bond has been shown via mutagenesis to weaken the Fe NO bond (lower FeN) in the NO adduct [89], as would be expected from the thiolate becoming a stronger donor in the absence of a stabilizing H-bond. Thus, it is possible that the tryptophan H-bond weakens the thiolate donor enough to raise the NOS backbonding correlation above that of cyt P450cam . There is a fly in this particular ointment, however, since chloroperoxidase (CPO), another thiolate-heme protein shows FeC/CO frequencies [48,53,54] that place it on the NOS line, and not on the cyt P450cam line (Fig. 2), even though it has no sidechain H-bond to the thiolate, and only two backbone amide H-bonds [90]. Thus, it is unclear why the thiolate ligand is a weaker donor in CPO than in cyt P450cam . Furthermore, when one of the three backbone H-bonds in cyt P450cam is eliminated by the substitution of a proline residue [81] (making it similar to CPO), the FeC/CO point moves upward from the cyt P450 line (see L358P-camphor in Fig. 2), implying weaker donation. Yet, the enzymatic properties of the L358P variant are consistent with stronger donation [81]. Thus, the picture with respect to thiolate H-bonding and the FeC/CO frequencies is not as clear as one would like.

2.3. Imidazole Modulation: Peroxidases and CooA The histidine side chain is also subject to modulation of its donor properties by H-bonding, because it can donate an H-bond from the NH group across the imidazole ring from the heme Fe. In Mb, the H-bond is donated to neutral acceptors, a backbone carbonyl and a serine OH group (see diagram in Fig. 1). However, in the peroxidase class of enzymes, there is a much stronger acceptor, the anionic carboxylate group of an aspartate side chain. The effect of this H-bond is seen most dramatically [91] in the Fe–histidine stretching vibration, which is detectable in high-spin 5-c Fe[II] hemes (but not in 6-c adducts). In Mb, this occurs at 220 cm−1 [92], but in the peroxidases there is a broad band with components at ∼230 and ∼245 cm−1 , which have been suggested to reflect tautomerism between forms in which the proton alternates between the histidine and the aspartate [92]. When the aspartate is replaced by an asparagine residue in cytochrome c peroxidase (CCPMI – the MI refers to residue replacements distant from the heme in a recombinant form of the enzyme), the broad band disappears and is replaced by a narrow band at 205 cm−1 [91]. This is substantially lower than in Mb, and suggests that the histidine is no longer H-bonded at all. This change is mirrored in the FeC/CO data for the CO adducts [58,59] (Fig. 3). The point for CCPMI [91] falls far below the Mb line, consistent with strong donation

106

T.G. Spiro et al.

from what is essentially an imidazolate anion. Likewise, a model heme adduct in which a trans imidazole has been deprotonated chemically, PPDME(Im− ), lies nearby, also below the Mb line, whereas the complex with neutral imidazole, PPDME(ImH), lies near the Mb line [60]. However, the point for the mutated protein, CCPMI(D235N), lies well above the Mb line, consistent with the absence of histidine H-bonding. The PPDME(Im− ) and PPDME(ImH) data indicate little change in FeC, but a large reduction in CO, upon imidazole deprotonation. This is consistent with the view (discussed above) that the effect of stronger trans ligand donation is to increase backbonding, thereby lowering CO, but that the corresponding increase in FeC is compensated by enhanced competition for -bonding. Likewise, model DFT calculations by Franzen [93] find CO lowering but little change in FeC when a trans imidazole donates an H-bond; however, the calculations uncovered an additional compensating factor, namely an expansion of the porphyrin ring associated with the H-bond donation. In any event, these considerations suggest that the CCPMI FeC/CO point should be translated horizontally to the Mb line when comparing distal polarity with the Mb variants. This would place it at about the position occupied by wild-type Mb, suggesting a weak H-bond interaction with a distal histidine, which is indeed present in the CCP binding pocket [94]. At this same position in the FeC/CO plot one finds the alkaline form, CCPMI(alk), which evidently interacts with the distal histidine, but has a Mb-like proximal histidine, indicating that the H-bond to the proximal aspartate residue is attenuated. When baker’s yeast CCP was examined [58], a similarly low-lying FeC/CO point was obtained (CCP(I) in Fig. 3) at low CO levels, but when the CO pressure was raised, another form was observed with FeC/CO on the Mb line (CCP(II) – Fig. 3), and at a high position, indicating a strong positive polar interaction. Such an interaction is consistent with the crystal structure of the CCP-CO adduct [95], which shows the CO to be in contact with a water molecule that is associated with a distal arginine residue. However, the fact that this form is on the Mb line indicates again that the proximal histidine no longer interacts strongly with the adjacent aspartate residue. There seems to be a balance between a strong proximal or a strong distal H-bond, and this balance is shifted by raising the CO pressure, suggesting a secondary binding site for CO in the vicinity of the heme [59]. In the case of horseradish peroxidase (HRP), a low-CO form I is again detected [60], with a low-lying FeC/CO point, consistent with the expected strong His-Asp proximal H-bond. At high CO, there are two forms, II [16,60,61] and III [16,61], both of which are on the Mb line, implying His-Asp attenuation. Form II is very high on the line (CO = 1906 cm−1 ), suggesting a direct interaction with the distal arginine, instead of one mediated by a water molecule, as in CCP-CO. Form III lies toward the middle of the line, perhaps indicating a distal histidine interaction instead. The peroxidases all have adjacent distal arginine and histidine residues that play critical roles in peroxide activation. The bound CO acts as a molecular probe, interacting with different residues under different conditions. This probe has been used in conjunction with site-directed mutagenesis to trace the distal and proximal residue interactions in CCPMI [58,59] and HRP [61]. Curiously, HRP–CO has an alkaline form whose FeC/CO point [16,61] is at the same position as the CCPMI(D235N) variant (Fig. 3), implying that the His-Asp interaction is broken. Moreover, when the peroxidase substrate benzhydroxamic acid (BHA) is bound to HRP [16], the FeC/CO point is again elevated from the Mb line, implying a broken His-Asp interaction, but the CO is very low (1911 cm−1 ), again suggesting direct interaction with the distal arginine. Finally, when HRP that has been

CO, NO, and O2 as Vibrational Probes

107

derivatized with polyethylene glycol (HRPPEG) is examined in benzene solution, the FeC/CO point lies on the 5-c line, suggesting that in this case the proximal histidine has dissociated from the heme [62]. Thus, the CCP and HRP data encompass a wide range of proximal and distal interactions, as evidenced by the FeC/CO plot. Similar interactions have been documented for other peroxidases [96–99]. Another interesting instance of imidazole modulation is found in the CooA protein, which regulates enzyme expression in CO-metabolizing bacteria [100]. CooA is activated to bind its target DNA sequence when CO binds to its heme. The heme is ligated by a histidine, and, unusually, to a proline amine, which is the N-terminus of the opposite subunit in the homodimeric protein [101]. It is the proline that is displaced by CO [102], setting in motion a large-scale conformation change that repositions the DNA-binding domains on the two subunits. A crystal structure is available only for the inactive, resting form of the protein [101], but the activity-inducing conformation change can be deduced from a homologous regulatory protein, CAP, whose crystal structure is available in the effector-bound (cAMP) form [103]. Resonance Raman spectroscopy on a series of site mutants led to the proposal that the initiating event in the conformation change was displacement of the heme into a nearby hydrophobic cavity formed by the C-helices of the two subunits [63,104]. Important evidence in support of this proposal was the finding that the FeC/CO point for the CO adduct was elevated from the Mb line (Fig. 3), implying a weak Fe–histidine bond. The unusually high CO, 1982 cm−1 , would translate to a more normal value (1969 cm−1 ) for a Mb-like hydrophobic binding pocket, if the FeC/CO point were translated horizontally to the Mb line. Additional evidence for a weak Fe–histidine bond comes from Aono and coworker’s picosecond RR study [105,106] in which the Fe His stretching vibration of 5-c photoproduct was located at 216 cm−1 , well below the 220 cm−1 characteristic of deoxyMb. This weakening was attributed [63,104] to the heme displacement, which would break an H-bond from the proximal histidine to an adjacent asparagine residue (2.7 Å O     N distance [101]). Evidence in support of this proposal was the absence of change in the FeC and CO frequencies when the asparagine was replaced by non-H-bonding residues, although they should have been shifted had the His Asn H-bond been intact [63]. At high pH, another form of CooA CO was detected [63], with significantly lower CO (1964 cm−1 ), although the FeC/CO point remained well above the Mb line (Fig. 3). It was proposed that the bound CO is subject to weak H-bonding from the displaced proline residue in this form, whereas at lower pH, the proline amine would be protonated and expelled from the hydrophobic pocket. Consistent with this view, a variant in which the N-terminal chain was shortened by deletion of the two penultimate residues failed to show any change at high pH [63]. The pKa for the alkaline transition of the wild-type protein was 8.6, two units lower than the aqueous solution pKa of the proline amine. This lowering represents the energy cost of burying the positively charged proline in the hydrophobic binding pocket.

2.4. Steric Hindrance and Distal Compression Early attempts to understand the variation in heme-CO vibrations wrestled with the possible effects of steric hindrance to upright binding of CO [8–10,12,13,15], since the

108

T.G. Spiro et al.

prevailing view was that the distal histidine in Mb imposed significant FeCO bending. This view has been overthrown by subsequent experiment and computation [see ref. [7] for details], although a small deviation from linearity in MbCO is documented [77,78] (and is actually attributable to a distal valine residue and not to the distal histidine [79]). Nevertheless, the question remains whether steric hindrance might in some cases influence the FeC/CO correlation. In principle, it could do so, since FeCO bending is expected to lower both FeC and CO because of lowered bond orders. This effect has been explored via DFT computation on a sterically constrained heme-CO model [9]. A steady progression of the FeC/CO pairs away from the backbonding correlation with increasing FeCO distortion was found. However, for modest distortions, up to 0.4 Å displacement of the O atom from the heme normal, the deviation was within the scatter of the experimental data for Mb variants. Thus, small deviations from linearity are unlikely to be detected by vibrational analysis. It now seems unlikely that strongly distorted FeCO can occur from steric hindrance. Although the energy cost of such distortion is not as large as once thought [9,13], it still is likely to exceed the energy required for local conformation changes that move sterically intrusive protein residues out of the way. Even in highly constrained porphyrins with covalent superstructures, the crystal structures show small FeCO distortions, but large displacements of the superstructure, as well as distortion of the porphyrin ring [107–109]. However, there is one documented large deviation from the FeC/CO backbonding correlation for a constrained porphyrin, C2 Cap [15,108] (Fig. 4) (Table 3). In C2 Cap, a benzene ring is strapped laterally to a porphyrin ring via short covalent tethers. Distal Compression 530 cyt ox (α)

520

Cu

B

510

O

νFe – C (cm–1)

C

500

Fe

5c

N

490

cyt ox (β)

C2Cap(NMeIm) N H

O C

480

Fe Fe N

470 N

Me

460

Mb 450 1940

1950

1960

1970

1980

1990

2000

2010

νCO (cm–1)

Fig. 4. Examples of displacements from the Mb line attributable to distal compression of the Fe C bond (see Table 3).

CO, NO, and O2 as Vibrational Probes

109

Table 3. Examples of Distal Compression of FeCO (Fig. 4) Sample C2 Cap(N-MeIm) cyt ox() cyt ox()

(Fe C) (cm−1 )

(C O) (cm−1 )

Reference

497 519 493

2002 1966 1955

[15] [110,111] [110,111]

With N -methylimidazole as trans ligand, the CO adduct gives a very high CO (2002 cm−1 ), which is attributed to the close contact between the benzene  electrons and the bound CO; only 2.8 Å separate the O atom from the center of the benzene ring [108]. The FeCO angular distortion is small (∼5 each of bend and tilt [108]). But the 497 cm−1 FeC frequency places the FeC/ CO point on the 5-c line (Fig. 4), and far above the Mb line. Since the adduct is clearly 6-coordinate, this elevation was attributed [15] to Fe C bond compression, due to the interaction with the constrained benzene ring. Similar bond compression has been proposed to account for the anomalously high FeC/CO position for the dominant form, , of cytochrome oxidases [110,111] (Fig. 4). The heme-binding site in this case is part of a binuclear site with a nearby Cu+ complex, CuB . Photolysis of the heme-CO adduct leads to ultrafast (<1 ps) formation of a transient Cu CO complex [112], and it seems likely that the Cu is close enough (3.5–5.0 Å Cu      Fe separation in the resting enzyme [113–115]) to interact directly with the bound CO, and compress the Fe C bond. The much lower CO, relative to the C2 Cap adduct (1966 versus 2002 cm−1 ) is consistent with a positive polar interaction with Cu+ , versus interaction with the benzene  cloud. A minor form of the cytochrome oxidase adduct, , which becomes prevalent at low and high pH, has FeC/CO frequencies falling close to the Mb line (Fig. 4), suggesting movement of the Cu away from contact with the CO. Consistent with this interpretation is the observation that mutations of the CuB ligands shift FeC/CO toward the Mb line [110,111]. Earlier, it had been suggested [116,117] that the anomalously high FeC/CO position of the -adduct might reflect weakening or actual displacement of the proximal histidine, but this seems a less likely explanation than bond compression. In summary, although steric distortion of the FeCO unit seems unlikely to affect FeC/CO values significantly, it is possible that that a tight distal constraint, as in the case of the cytochrome oxidase binuclear site, can produce FeC/CO elevation via Fe C bond compression.

3. Fe(II)NO VIBRATIONS FeN and NO frequencies have been recorded via 15 N-edited RR spectra for many heme protein NO adducts, but unlike the corresponding data for CO adducts, well-behaved correlations have not been observed [18,19,21]. Noting an apparent insensitivity of FeN to variations in NO, Park and Boxer [19] inferred that the Stark shift (variation with electric field) of NO results not from backbonding, as in CO adducts, but rather to anharmonicity of the vibration. However, backbonding had clearly been demonstrated for 5-c NO adducts of Fe(II)TPP-Y, with variably donating peripheral substituents [17].

110

T.G. Spiro et al.

A negative linear FeN/NO plot was obtained, with essentially the same slope as the FeC/CO plot for the corresponding CO adducts. Could the inconsistency lie in the nature of 6-c NO adducts? NO has a strong trans labilizing effect when bound to Fe(II) porphyrins. The bond to the trans ligand is unusually weak. Could the trans effect somehow obscure the expected backbonding correlation? Data for protein-free 6-c NO adducts have not been available for comparison because the trans effect renders RR spectra of 6-c adducts elusive; the Raman laser beam readily dissociates the weakly bound trans ligand, and only 5-c spectra are detectable in solution [17]. We have recently overcome this problem by using frozen samples, in which the dissociated trans ligand recombines efficiently. Data for a series of (N-MeIm)Fe(II)TPPY(NO) adducts have been obtained, and a well-behaved FeN/NO anticorrelation is indeed observed, demonstrating that backbonding does determine the vibrational frequencies, in the absence of additional protein effects [22]. Sage and coworkers have raised the question whether FeN has been properly assigned, since another lower-frequency band in the nuclear resonance vibrational spectrum (NRVS) actually displays a larger Fe displacement [20]. Because the FeNO equilibrium geometry is bent, the Fe N stretching and Fe N O bending coordinates are mixed in the normal modes, none of which are pure stretching or bending. However, since the 15 NO-sensitive RR band at ∼550 cm−1 is well correlated with NO, we take the pragmatic view that the conventional “FeN” label should be retained [22].

3.1. Protein-Free Adducts The new 6-c adduct FeN/NO data (Table 1) are plotted in Fig. 5, along with the previously reported 5-c adduct data. Data for CO adducts are shown for comparison [22]. The slopes are nearly the same for 5-c adducts of CO and NO, while for both ligands, the 6-c slopes are larger than the 5-c slopes. The increase in slope is even higher for NO than for CO, indicating that the NO trans effect is manifested in a higher backbonding sensitivity for NO than CO when a trans ligand is bound. The reason for this difference can be traced to the overlap of the Fe dz2 orbital with the NO  ∗ orbital bearing the unpaired electron [118], an interaction that does not exist for CO. The trans ligand raises the energy of the dz2 orbital, augmenting this interaction, which contributes to the backbonding sensitivity. However, the 6-c line lies below the 5-c line for CO but above the 5-c line for NO. Adding an axial ligand depresses the FeC frequency but elevates the FeN frequency. There are two contributing factors to this difference [22]: 1. The trans ligand lengthens the Fe X bond for both Fe C and Fe N, but less so for Fe N, due to the strong NO trans effect. 2. The trans ligand increases the mixing of Fe N stretching and Fe N O bending coordinates in the “FeN” mode, an effect that is absent for the (linear) FeCO unit. The enhanced mixing lowers the effective mass, thereby raising the frequency.

CO, NO, and O2 as Vibrational Probes

111 νC — O (cm–1)

1900

1920

1940

1960 5-c, (Slope = – 0.46)

II

Fe [TPP-Y] — CO

H

2,6-F2

OH

600

OCH3

520

CN 2,6-Cl2

OCH3 H

580

6-c, (Slope = – 1.0)

2,6-Cl2

νFe — NO (cm–1)

2,6-F2

OH

6-c, (Slope = – 0.68)

F5

CH3

OCH3

CH3 H

480

FeII[TPP-Y] — NO

XO FeII XO

OH

N N

Fe

6-c

1620

H

II

520

500

500

NO2

560

540

540

NH2

νFe — CO (cm–1)

1880

5-c

1640

OCH3

CN

CH3

2,6-Cl2 5-c, (Slope = – 0.40)

1660

1680

F5

2,6-F2

1700

νN — O (cm–1)

Fig. 5. FeXO backbonding correlations for 5-c and 6-c (N-MeIm axial ligand) NO and CO adducts of Fe(II)TPP-Y with the indicated phenyl substituents, Y, in organic solvents (CH2 Cl2 , DMF, Bz). For the 5-c adducts, NO data are from reference [17] and CO data are from reference [11]. The 6-c adduct data are from reference [22].

3.2. Heme Proteins Data are compared in Fig. 6 for CO and NO adducts of a series of heme proteins. In most cases, the points fall fairly close to the experimental lines for protein-free 6-c adducts, indicating that there is no necessary protein effect, for either CO or NO. Cyt P450 falls below the line for NO as well as CO, reflecting the stronger thiolate donor effect in both cases. The other proteins all have histidine proximal ligands. However, there are marked deviations below the NO line for Mb, neuroglobin, and cytoglobin, although all three fall on the CO line. The globins have distal histidine residues, and when this is replaced in Mb by hydrophobic residues (H64L, H64I), the points are much closer to the line. The rest of the points falling near the proteinfree line for NO as well as CO belong to heme sensor proteins, which have mainly hydrophobic pockets. Cytochrome oxidase, which lies above the CO line, probably due to a compressive effect from the distal Cu+ (see above), falls far below the NO line. Thus, the presence of a distal polar group, or of a positive charge (Cu+ ), appears to produce strong negative deviations from the line in the case of NO. This view of what differentiates NO from CO is reinforced by data for Mb variants with specific distal residue replacements, plotted in Fig. 7. All the CO data fall on the

112

T.G. Spiro et al. νC — O (cm–1) 1930

1940

1950

1960

1970

1980

1990

2000 O

6-c CO

2010

520

C

Cyt ox (T.t.)

WT-Mbh

BjFixLH RCCP H64L-Mb

AxPDEA1H Cgb TtTar4H EcDosH AXCP H64I-Mb

P450

620

FeII N

Ngb

500

NH

480

O

νFe — NO (cm–1)

600

N

460

FeII N

580

6-c NO

NH

BjFixLH

560 P450

440

AXCP Ngb Cgb WT-Mbh

540 1580

νFe — CO (cm–1)

1920

RCCP MtDosH H64L-Mb EcDosH AxPDEA1H H64I-Mb

TtTar4H

Cyt ox (T.t.)

1590

1600

1610

1620

1630

1640

1650

1660

1670

νN — O (cm–1)

Fig. 6. FeX/XO plot for available data from heme protein Fe(II)XO adducts (see Table 1). The filled circles are H64L and H64I variants of Mb. Open diamonds are proteins with distal histidine (Ngb, Cgb, and WT-Mb) or, in the case of cyt ox, with a distal CuB center. The filled square is an adamantanone–bound cyt P450cam . Cytchrome P450 has a cysteine axial ligand; the other proteins all have histidine ligands. The backbonding correlations (solid lines) are for 6-c protein-free adducts (from Fig. 5).

protein-free 6-c line; as noted above, their positions along the line reflect the polarity of the pocket. In marked contrast, the NO data show large and variable deviations from the line. These deviations can, however, be related to the likely strength of positive polar interactions with the bound NO. The displacement seen for WT Mb is diminished when His64 is replaced by the more weakly H-bonding glutamine (H64Q) and essentially disappears for the hydrophobic Leu and Ile replacements, as noted above. Likewise, the displacement is diminished when Phe46, whose bulky side chain buttresses the His64 H-bond, is replaced by smaller side chains from Leu, Val, or Ala. But the displacement is increased when Leu29, at the back of the pocket, is replaced by Phe, reinforcing the His64 H-bond. The largest displacement occurs for V68N; substitution of Asn for Val68, which is in direct contact with the bound ligand [78] introduces a strong H-bond. The deviations from the protein-free line for NO adducts are in the same order as displacements along the line for the CO adducts. Why does H-bonding produce the expected backbonding trend for CO, but anomalous deviation from the backbonding line for NO? The answer appears to be that H-bonding alters the FeNO angle, but not the FeCO angle. The bent FeNO unit can be viewed as an equilibrium among valence isomers (Fig. 8): Fe(I)(NO+ ), Fe(II)(NO), and Fe(III)(NO− ).

CO, NO, and O2 as Vibrational Probes

113 νC — O (cm–1)

1910

1920

1940

1930

1950

1960

1970

L29

620

6-c NO

540

F46

6-c CO V68N

H64

L29F

V68

520 L29F/H64Q

νFe — NO (cm–1)

H64Q

WT

600

H93

H64A

500

νFe — CO (cm–1)

1900

H64I

580 H64L

560

L29F/H64Q

WT

V68N L29F

F46V

480

V68F

H64L H64Q F46V

H64I F46A

F46L

540 1590

1600

1610

1620

1630

1640

1650

1660

νN — O (cm–1)

Fig. 7. FeX/XO plot for the indicated variants of Mb: WT (stars), H64X (circles), L29F (circles), V68X (squares), and F46X (diamonds). The open symbols represent Fe(II)NO adducts; closed symbols are Fe(II)CO adducts. The solid lines are the backbonding correlations for 6-c protein-free XO adducts (from Fig. 5). The inset shows the positions of mutated distal residues in the Mb binding pocket [121]. O N Fe

L

(I)

O

180°

N Fe

L

(II)

140°

XH

O N

120°

Fe L

(III)

Fig. 8. Valence isomers of LFeP(NO). Although formal charges on NO are + for (I) and − for (III), these are compensated by greater backbonding in (I). However, negative charge builds up on N in (III), and is stabilized by appropriately oriented H-bonds, as in MbNO.

H-bonding should favor Fe(III)(NO− ), via electrostatic stabilization, thereby decreasing the FeNO angle. This effect is absent in FeCO, for which H-bonding simply polarizes the orbitals. Because the FeCO unit lacks the antibonding electron that shifts back and forth in FeNO with changing angle, the FeCO geometry is indifferent to polarization. However, the directionality of the H-bond is important, because increased FeNO bending builds up negative charge on the N, rather than the O atom. This is indicated by

114

T.G. Spiro et al. Table 4. Angle dependence of DFT-computed frequencies (cm−1 ), distances (A ), energies (kcal mol−1 ), and atomic charges (z) for (ImH)FeII P(NO) [22] Angle ( ) 129 FeN NO FeNO d(Fe–N) d(N–O) d(Fe–Im) E zN zO

576 1634 467 1775 1202 2230 142 −0019 −0178

139 602 1669 471 1749 1199 2236 0 −0005 −0195

149 600 1697 474 1734 1196 2247 123 +0007 −0212

DFT computations on (imidazole)Fe(II)porphine(NO) [22], in which the FeNO angle is constrained to be 10 larger or smaller than the equilibrium angle (Table 4). As the angle diminishes, negative charge increases on N but decreases on O, leaving the total NO charge nearly constant. The reason is that the formal charges in the valence isomers are compensated by backbonding, which is greater for Fe(I)(NO+ ) than for Fe(III)(NO− ); the former has two d − ∗ interactions while the latter has one. Thus, increased FeNO bending is favored if a distal H-bond donor is oriented toward the N atom of the bound NO (Fig. 8). This is in fact the orientation of the distal His64 in MbNO (Fig. 7, inset). In contrast, a distal tyrosine H-bond donor in the H–NOX protein TtTar4H points toward the O atom (judging from the crystal structure of the O2 adduct [119]), and the NO adduct of TtTar4H does fall on the protein-free NO line (Fig. 6). For MbNO, however, the structural data indicate that the FeNO angle is highly variable. There are now four crystal structures, with reported angles of 112 [120], 147 [121], 144 [122], and 120 [122]. The first two of these were obtained at moderate resolution, 1.7 and 1.9 Å, while the last two were at higher resolution, 1.3 Å. The first three were from crystals prepared by treating Mb(III) crystals with nitrite/dithionite, while the fourth was from crystals prepared by treating Mb(II) crystals with NO. A frozen solution EXAFS study [123] gave an FeNO angle of 147 , while early EPR studies indicated large changes in the FeNO angle with changing temperature in frozen solution [124,125]. It appears that the FeNO unit is mobile in MbNO, its angle changing with sample history. The RR data represented in Fig. 7 were obtained in fluid solution, and presumably represent the equilibrium geometry of MbNO and its variants. The DFT calculations (Table 4) indicate that modest changes in FeNO angle would be sufficient to account for the large observed deviations from the backbonding line [22]. Closing the angle by 10 is predicted to decrease FeN by 26 cm−1 and NO by 35 cm−1 . These are about twice as large as the differences between WT MbNO and the H64L variant, which lies near the protein-free line (Fig. 7). Thus, an angle closing of ∼5 would be sufficient to account for the effect of the His64 H-bond.

CO, NO, and O2 as Vibrational Probes

115

It remains, however, to explain why the spread of points in Fig. 7 is mainly horizontal; the variation in FeN is small. (This is what led Park and Boxer [19] to infer that backbonding is unimportant in NO adducts.) In addition to closing the FeNO angle, H-bonding is expected to polarize the FeNO orbitals, just as for FeCO. The resulting increase in backbonding is expected to decrease NO, reinforcing the angle effect, but to increase FeN, counteracting the angle effect (provided that the H-bond is directed at N rather than O). Thus, variations in H-bonding in MbNO are expected be seen more strongly in NO than in FeN. Nevertheless, some of the MbNO variants do show appreciable FeN variation. Intriguingly, the points for F46V, H64Q, L29F/H64Q, and V68F are ranged along a line (dotted in Fig. 7) parallel to the protein-free line. This is the behavior expected for variable backbonding at constant (but reduced) FeNO angle. Thus, the vibrational data indicate that FeNO is a more sensitive probe of heme pocket influences than is FeCO.

4. Fe(III)NO VIBRATIONS Unlike CO and O2 , NO can bind to Fe(III) heme, producing an adduct which is isolectronic with Fe(II)CO. This adduct is easily reduced to the Fe(II)NO adduct, and vibrational data are therefore sparse. However, several pairs of FeN/NO frequencies are available for Fe(III)NO adducts of thiolate heme proteins (Table 5), since thiolate Table 5. Six-coordinate Fe(III) NO proteins (Fig. 9) Proteins

(Fe NO) (cm−1 )

(N O) (cm−1 )

Reference

594 604 595 591 591 594

1904 1903 1918 1917 1914 1921

[126] [127,128] [129] [130,131] [132] A

528 522 524 520 538 529 530 532 533 510 515

1806 1806 1818 1818 1868 1851 1853 1852 1851 1828 1837

[51,133] [51,133] [51,133] [51,133] [51,133] [133,134] [135] [135] [135] [136] [136]

His-Fe-NO hemes NOR HRP hHO-1 NP1 HbN WT-Mb Cys-Fe-NO hemes P450cam (ns) P450cam + camphor P450cam + norcamphor P450cam + adamantanone Chloroperoxidase (CPO) P450nor (Fusarium oxysporum) P450nor (proto) P450nor (meso) P450nor (deuteron) SR (model complex) SR-HB (model complex) A: this work.

116

T.G. Spiro et al. 5-c OEP.CHCl3 5-c OEP 600

HRP

O N

HbN

FeIII

νFe — NO (cm–1)

580

hHO-1 WT-Mb

NOR

NP1

Slope = – 0.38

N NH

560

Slope = 0.22 O

540

CPO P450cam(substrate free) P450cam + camphor

520

N

P450nor FeIII

P450cam + norcamphor P450cam + adamantanone SR-HB SR

S

500 1780

1800

1820

1840

1860

1880

1900

1920

νN — O (cm–1)

Fig. 9. FeN/NO plot for available data from heme protein Fe(III)NO adducts (Table 5). Also shown are data for crystalline forms of Fe(III)OEP(NO)+ and a CHCl3 solvate.

ligation stabilizes Fe(III). Linder et al. [137] noticed that these points describe a positive correlation, instead of the familiar negative correlations for Fe(II)CO; FeN and NO increase in concert. This positive correlation is shown in the lower part of Fig. 9. In addition, the frequencies Linder et al. [137] determined for two crystalline forms of 5-c Fe(octaethylporphyrin)(NO)(ClO4 ), one with and one without chloroform of solvation, also correlated positively (open diamonds in Fig. 9). DFT calculations [137,138] on model porphines with electron or acceptor substituents supported the existence of a positive correlation, and this effect was explained as resulting from the nature of the second highest occupied molecular orbital (HOMO-1), which was found to be antibonding with respect to the Fe X as well as the X O bond for Fe(III)NO, but nonbonding for Fe(II)CO. The underlying reason for this difference is that the dz2 orbital energy is lower for Fe(III) than it is for Fe(II). However, these orbital interactions are subtle, and Linder et al. [137,138] found that the antibonding character of the HOMO-1 was limited to model adducts in which substituents were attached to the porphine meso carbon atoms. When the substituents were instead attached to the C atoms, the orbital was nonbonding, and a negative FeN/NO correlation was expected. (It was unclear why the crystalline Fe(III)NO adducts of octaethylporphine, which is C -substituted, nevertheless seemed to be positively correlated.) We find that the available data on Fe(III) adducts of histidine-ligated heme proteins does indeed describe a negative correlation (Fig. 9). Only the thiolate-ligated hemes give a positive correlation. However, the slope of the negative Fe(III)NO correlation, −0.38, is only half that observed for the Fe(II)CO correlation of histidine-ligated proteins.

CO, NO, and O2 as Vibrational Probes

117

Clearly, the electronic interactions for Fe(III)NO are different and more subtle than for Fe(II)CO. Further experimental data are needed to clarify the interpretation of FeN/NO data for these adducts.

5. Fe(II)O2 VIBRATIONS Vibrational data on O2 adducts of heme proteins are fragmentary, but interesting trends can nevertheless be observed when FeO is plotted against OO (Fig. 10) (Table 6). As reported previously [17], the data for 5-c protein-free adducts of Fe(II)porphyrins fall on a well-behaved backbonding correlation, with a slope that is distinctly higher than that of 5-c CO and NO adducts (∼ − 09 versus ∼ − 05). Data have also been reported for three 6-c protein-free adducts with imidazole or piperidine axial ligands (Table 6), and establish that FeO is higher than for 5-c adducts, as they are for NO, but not CO. They also appear to describe a steeper backbonding line, as is the case for 6-c NO and CO adducts. If the indicated line is correct, then it can be seen that the available data for four different hemoglobins (Table 6) all fall well below the line. The pattern is similar to that shown by NO adducts of the globins. It is possible that a similar explanation applies, namely that distal H-bond donors that are oriented toward the O atom which is bound to the Fe induce further bending of the FeOO unit, decreasing both FeO and OO. 580

Slope = –4.5 (6-c) O2

560

FeII N

νFe — O (cm–1)

540

520

Slope = –0.87 (5-c) Slope = 0.70 (6-c)

O2

O2

500

FeII

FeII S

480

1120

1140

1160

1180

1200

1220

1240

νO — O (cm–1)

Fig. 10. FeO/OO plot for available data (Table 6) from O2 adducts of Fe(II) porphyrins (5-coordinate  and with thiolate or nitrogenous ligand ) and of heme proteins (with histidine or thiolate ligands).

118

T.G. Spiro et al.

Table 6. O2 adducts (Fig. 10) (O O) (cm−1 )

(Fe O) (cm−1 )

Reference

Proximal histidine 5-c models TPP TMP TMP OEP TPFPP PC

1195 1171 1188 1192 1223 1207

508 522 516 509 486 488

[17] [17] [17] [17] [17] [17]

1133 1136 1140 1135

554 554 556 568

[139] [139] [139] [140–142]

1157 1159 1159

575 564 568

[143,144] [145–147] [147]

1135 1123 1140 1139 1147 1128 1136 1129 1137

517 517 541 537 537 536 536 521 536

[148] [148] [149,150] [149]

1140 1147

527 536

Proteins (His ligation) Synechocystis Hb Chlamydomonas WT Hb Chlamydomonas K(E10)A Hb Human adult HbA 6-c models TPP(pip) TPivP(N-MeIm) TPivP(1,2-Me2 Im)

Proximal thiolate Proteins (Cys ligation) saNOS saNOS + l-Arginine P450cam + camphor P450cam + adamantanone D251N P450cam D251N P450cam /putidaredoxin

[151] [151]

6-c models TPivP(C6 HF4 S− ) TpivP(C6 F5 S− )

[149,152] [149]

Data are also available for O2 adducts with thiolate axial ligands (Table 6). In contrast to the nitrogenous axial ligands, the thiolate ligands appear to define a FeO/OO line with positive slope. Moreover, the protein adducts (of NOS and cyt P450) fall on the same line as protein-free adducts. This pattern is similar to that seen for Fe(III)NO adducts (Fig. 9), which also show a positive correlation when the axial ligand is thiolate, albeit with a smaller slope. It is possible that the orbital interactions described by Linder et al. [137,138] for the Fe(III)NO adducts apply with even greater force to the Fe(II)O2 adducts. Thus, the O2 adduct appears to show behaviors that are similar to those shown by both Fe(II)NO and Fe(III)NO adducts. However, more data on O2 adducts, both protein-free and protein-bound are needed to test these hypotheses.

CO, NO, and O2 as Vibrational Probes

119

ACKNOWLEDGMENT This work was supported by NIH grant GM 33576 from the National Institute of General Medical Sciences. I. W. was the recipient of Norwegian Research Council. We thank Radhika Rajendran for helpful assistance.

REFERENCES [1] Park, E.S., Andrews, S.S., Hu, R.B. and Boxer, S.G. (1999) J. Phys. Chem. B 103, 9813–9817. [2] Li, T.S., Quillin, M.L., Phillips, G.N. and Olson, J.S. (1994) Biochemistry 33, 1433–1446. [3] Ivanov, D., Sage, J.T., Keim, M. et al. (1994) J. Am. Chem. Soc. 116, 4139–4140. [4] Ormos, P., Braunstein, D., Frauenfelder, H. et al. (1988) Proc. Nat. Acad. Sci. U.S.A. 85, 8492–8496. [5] Moore, J.N., Hansen, P.A. and Hochstrasser, R.M. (1988) Proc. Nat. Acad. Sci. U.S.A. 85, 5062–5066. [6] Lim, M., Jackson, T.A. and Anfinrud, P.A. (1995) Science 269, 962–966. [7] Spiro, T.G. and Kozlowski, P.M. (2001) Acc. Chem. Res. 34, 137–144. [8] Spiro, T.G. and Kozlowski, P.M. (1998) J. Am. Chem. Soc. 120, 4524–4525. [9] Kozlowski, P.M., Vogel, K.M., Zgierski, M.Z. and Spiro, T.G. (2001) J. Porphyrins Phthalocyanines 5, 312–322. [10] Li, X.Y. and Spiro, T.G. (1988) J. Am. Chem. Soc. 110, 6024–6033. [11] Vogel, K.M., Kozlowski, P.M., Zgierski, M.Z. and Spiro, T.G. (2000) Inorg. Chim. Acta 297, 11–17. [12] Kerr, E.A. and Yu, N.T. (1988) In Biological Applications of Raman Spectroscopy, Vol. III (T.G. Spiro, ed.) New York: Wiley-Interscience, pp. 39–95. [13] Ghosh, A. and Bocian, D.F. (1996) J. Phys. Chem. 100, 6363–6367. [14] Kincaid, J.R. (2000) In The Porphyrin Handbook, Vol. 7 (K.M. Kadish, K.M. Smith, and R. Guilard, eds.) San Diego, CA: Academic Press, pp. 227–291. [15] Ray, G.B., Li, X.Y., Ibers, J.A. et al. (1994) J. Am. Chem. Soc. 116, 162–176. [16] Uno, T., Nishimura, Y., Tsuboi, M. et al. (1987) J. Biol. Chem. 262, 4549–4556. [17] Vogel, K.M., Kozlowski, P.M., Zgierski, M.Z.and Spiro, T.G. (1999) J. Am. Chem. Soc. 121, 9915–9921. [18] Thomas, M.R., Brown, D., Franzen, S. and Boxer, S.G. (2001) Biochemistry 40, 15047– 15056. [19] Park E.S. and Boxer, S.G. (2002) J. Phys. Chem. B 106, 5800–5806. [20] Zeng, W., Silvernail, N.J.,Wharton, D.C. et al. (2005) J. Am. Chem. Soc. 127, 11200–11201. [21] Coyle, C.M., Vogel, K.M., Rush, T.S., III, et al. (2003) Biochemistry 42, 4896–4903. [22] Ibrahim, M., Xu, C. and Spiro, T.G. (2006) J. Am. Chem. Soc. 128, 16834–16845. [23] Deng, T.J., Macdonald, I.D.G., Simianu, M.C. et al. (2001) J. Am. Chem. Soc. 123, 269–278. [24] Kushkuley, B. and Stavrov, S.S. (1997) Biochim. Biophys. Acta—Prot. Str. Mol. Enzym. 1341, 238–250. [25] Tsubaki, M. and Ichikawa, Y. (1985) Biochim. Biophys. Acta 827, 268–274. [26] Vogel, K.M., Spiro, T.G., Shelver, D. Thorsteinsson, M.V. and Roberts, G.P. (1999) Biochemistry 38, 2679–2687. [27] Unno, M., Christian, J.F., Olson, J.S. et al. (1998) J. Am. Chem. Soc. 120, 2670–2671. [28] Ramsden, J. and Spiro, T.G. (1989) Biochemistry 28, 3125–3128. [29] Zhu, L., Sage, J.T., Rigos, A.A.et al. (1992) J. Mol. Biol. 224, 207–215.

120

T.G. Spiro et al.

[30] Quillin, M.L., Brantley, R.E., Johnson, K. A. et al. (1992) Faseb J. 6, A466–A466. [31] Kerr, E.A., Yu, N.T., Bartnicki, D.E. and Mizukami, H. (1985) J. Biol. Chem. 260, 8360–8365. [32] Balasubramanian, S., Lambright, D.G., Marden, M.C. and Boxer, S.G. (1993) Biochemistry 32, 2202–2212. [33] Biram, D., Garratt, C.J. and Hester, R.E. (1991) In Spectroscopy of Biological Molecules (R.E. Hester and R.B. Girling, eds.) Cambridge, U.K.: Royal Society of Chemistry, pp. 433–434. [34] Cameron, A.D., Smerdon, S.J., Wilkinson, A.J. et al. (1993) Biochemistry 32, 13061–13070. [35] Tsubaki, M., Srivastava, R.B. and Yu, N.T. (1982) Biochemistry 21, 1132–1140. [36] Fuchsman, W.H. and Appleby, C.A. (1979) Biochemistry 18, 1309–1321. [37] Anderton, C.L. Hester, R.E. and Moore, J.N. (1997) Biochim. Biophys. Acta—Prot. Str. Mol. Enzym. 1338, 107–120. [38] Han, S., Rousseau, D.L., Giacometti, G. and M. Brunori, (1990) Proc. Natl. Acad. Sci. U.S.A. 87, 205–209. [39] Ling, J.H., Li, T.S., Olson, J.S. and Bocian, D.F. (1994) Biochim. Biophys. Acta—Bioenerg. 1188, 417–421. [40] Zhao, X., Vyas, K., Nguyen, B.D. et al. (1995) J. Biol. Chem. 270, 20763–20774. [41] Morikis, D., Champion, P.M., Springer, B.A. and Sligar, S.G. (1989) Biochemistry 28, 4791–4800. [42] Phillips, G.N., Teodoro, M.L., Li, T.S. and Olson, J.S. (1999) J. Phys. Chem. B 103, 8817–8829. [43] Tomita, T., Hirota, S., Ogura, T. et al. (1999) J. Phys. Chem. B 103, 7044–7054. [44] Jung, C., Hoa, G.H.B., Schroder, K.L., et al. (1992) Biochemistry 31, 12855–12862. [45] Bangcharoenpaurpong, O. (1987) Resonance Raman Studies of Myoglobin, Cytochromes P450 and chloroperoxidase, Ph. D. dissertation, Northeastern University: Boston, MA. [46] Uno, T., Nishimura, Y., Makino, R. et al. (1985) J. Biol. Chem. 260, 2023–2026. [47] Wells, A.V., Li, P.S. Champion, et al. (1992) Biochemistry 31, 4384–4393. [48] O’keefe, D.H., Ebel, R.E., Peterson, J.A. et al. (1978) Biochemistry 17, 5845–5852. [49] Nagano, S., Shimada, H., Tarumi, A. et al. (2003) Biochemistry 42, 14507–14514. [50] Unno, M., Christian, J.F., Sjodin, T. et al. (2002) J. Biol. Chem. 277, 2547–2553. [51] Hu, S. and Kincaid, J.R. (1991) J. Am. Chem. Soc. 113, 2843–2850. [52] Hu, S. and Kincaid, J.R. (1991) J. Am. Chem. Soc. 113, 9760–9766. [53] Hu, S. and Kincaid, J.R. (1993) J. Biol. Chem. 268, 6189–6193. [54] Wang, J., Stuehr, D.J. and Rousseau, D.L. (1997) Biochemistry 36, 4595–4606. [55] Wang, J.L., Stuehr, D.J., Ikedasaito, M. and Rousseau, D.L. (1993) J. Biol. Chem. 268, 22255–22258. [56] Fan, B., Wang, J., Stuehr, D.J and Rousseau, D.L. (1997) Biochemistry 36, 12660–12665. [57] Li, D., Stuehr, D.J., Yeh, S.-R. and Rousseau, D.L. (2004) J. Biol. Chem. 279, 26489–26499. [58] Smulevich, G., Evangelistakirkup, R., English, A. and Spiro, T.G. (1986) Biochemistry 25, 4426–4430. [59] Smulevich, G., Mauro, J.M., Fishel, L.A. et al. (1988) Biochemistry 27, 5486–5492. [60] Evangelistakirkup, R., Smulevich, G. and Spiro, T.G. (1986) Biochemistry 25, 4420–4425. [61] Feis, A., Rodriguez-Lopez, J.N., Thorneley, R.N.F. and Smulevich, G. (1998) Biochemistry 37, 13575–13581. [62] Mabrouk, P.A. and Spiro, T.G. J. (1998) J. Am. Chem. Soc. 120, 10303–10309. [63] Coyle, C.M., Puranik, M., Youn, H. et al. (2003) J. Biol. Chem. 278, 35384–35393. [64] Andrew, C.R., Green, L.E., Lawson, D.M. and Eady, R.R. (2001) Biochemistry 40, 4115–4122. [65] Andrew, C.R., George, S. J., Lawson, D.M. and Eady, R.R. (2002) Biochemistry 41, 2353–2360.

CO, NO, and O2 as Vibrational Probes [66] [67] [68] [69] [70] [71] [72] [73] [74] [75] [76] [77] [78] [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95] [96] [97] [98] [99] [100] [101] [102] [103] [104] [105] [106] [107] [108]

121

Andrew, C.R., Kemper, L.J., Busche, T.L., et al. (2005) Biochemistry 44, 8664–8672. Karow, D.S., Pan, D., Tran, R., et al. (2004) Biochemistry 43, 10203–10211. Tomita, T., Gonzalez, G., Chang, A.L. et al. (2002) Biochemistry 41, 4819–4826. Pinakoulaki, E., Ohta, T., Soulimane, T. et al. (2004) J. Biol. Chem. 279, 22791–22794. Pinakoulaki, E., Ohta, T., Soulimane, T. et al. (2005) J. Am. Chem. Soc. 127, 15161–15167. Sawai, H., Makino, M., Mizutani, Y. et al. (2005) Biochemistry 44, 13257–13265. Das, T.K., Friedman, J.M., Kloek, A.P. et al. (2000) Biochemistry 39, 837–842. Tani, F., Matsu-ura, M., Ariyama, K. et al. (2003) Chem. Eur. J. 9, 862–870. Kushkuley, B. and Stavrov, S.S. (1996) Biophysical J. 70, 1214–1229. Kushkuley, B. and Stavrov, S.S. (1997) Biophysical J. 72, 899–912. Franzen, S. (2002) J. Am. Chem. Soc. 124, 13271–13281. Kachalova, G.S., Popov, A.N. and Bartunik, H.D. (1999) Science 284, 473–476. Krzywda, S., Murshudov, G.N., Brzozowski, A.M. et al. (1998) Biochemistry 37, 15896– 15907. De Angelis, F., Jarzecki, A.A., Car, R. and Spiro, T.G. (2005) J. Phys. Chem. B 109, 3065–3070. Sono, M.P., Roach, M., Coulter, E.D. and Dawson, J.H. (1996) Chem. Rev. 96, 2841–2887. Yoshioka, S., Takahashi, S., Ishimori, K. and Morishima, I. (2000) J. Inorg. Biochem. 81, 141–151. Imai, M., Shimada, H., Watanabe, Y. et al. (1989) Proc. Natl. Acad. Sci. U.S.A. 86, 7823– 7827. Martinis, S.A., Atkins, W.M., Stayton, P.S. and Sligar, S.G. (1989) J. Am. Chem. Soc. 111, 9252–9253. Griffith, O.W. and Stuehr, D.J. (1995) Ann. Rev. Physiol. 57, 707–736. Schlichting, I., Jung, C. and Schulze, H. (1997) FEBS Lett. 415, 253–257. Crane, B.R., Arvai, A.S., Ghosh, D.K. et al. (1998) Science 279, 2121–2126. Fischmann, T.O., Hruza, A., Niu, X.D. et al. (1999) Nat. Str. Biol. 6, 233–242. Raman, C.S., Li, H.Y., Martasek, P. et al. (1998) Cell 95, 939–950. Couture, M., Adak, S., Stuehr, D.J. and Rousseau, D.L. (2001) J. Biol. Chem. 276, 38280– 38288. Sundaramoorthy, M., Terner, J. and Poulos, T.L. (1995) Structure 3, 1367–1377. Smulevich, G., Mauro, J.M., Fishel, L.A. et al. (1988) Biochemistry 27, 5477–5485. Kitagawa, T., Nagai, K. and Tsubaki, M. (1979) FEBS Lett. 104, 376–378. Franzen, S. (2001) J. Am. Chem. Soc. 123, 12578–12589. Finzel, B.C., Poulos, T.L. and Kraut, J. (1984) J. Biol. Chem. 259, 13027–13036. Edwards, S.L.and Poulos, T.L. (1990) J. Biol. Chem. 265, 2588–2595. Smulevich, G., Miller, M.A., Kraut, J. and Spiro, T.G. (1991) Biochemistry 30, 9546–9558. Feis, A., Santoni, E., Neri, F. et al. (2002) Biochemistry 41, 13264–13273. Heering, H.A., Jansen, M.A.K., Thorneley, R.N.F. and Smulevich, G. (2001) Biochemistry 40, 10360–10370. Heering, H.A., Smith, A.T. and Smulevich, G. (2002) Biochem. J. 363, 571–579. Shelver, D., Kerby, R.L., He, Y.P. and Roberts, G.P. (1997) Proc. Natl. Acad. Sci. U.S.A. 94, 11216–11220. Lanzilotta, W.N., Schuller, D.J., Thorsteinsson, M.V. et al. (2000) Nat. Str. Biol. 7, 876–880. Yamamoto, K., Ishikawa, H., Takahashi, S. et al. (2001) J. Biol. Chem. 276, 11473–11476. Passner, J.M., Schultz, S.C. and Steitz, T.A. (2000) J. Mol. Biol. 304, 847–859. Ibrahim, M., Kerby, R.L., Puranik, M. et al. (2006) J. Biol. Chem. 281, 29165–29173. Uchida, T., Ishikawa, H., Ishimori, K. et al. (2000) Biochemistry 39, 12747–12752. Kubo, M., Inagaki, S., Yoshioka, S. et al. (2006) J. Biol. Chem. 281, 11271–11278. Kim, K., Fettinger, J., Sessler, J.L. et al. (1989) J. Am. Chem. Soc. 111, 403–405. Kim, K., and Ibers, J.A. (1991) J. Am. Chem. Soc. 113, 6077–6081.

122

T.G. Spiro et al.

[109] Slebodnick, C., Fettinger, J.C., Peterson, H.B. and Ibers, J.A. (1996) J. Am. Chem. Soc. 118, 3216–3224. [110] Wang, J.L., Takahashi, S., Hosler, J.P. et al. (1995) Biochemistry 34, 9819–9825. [111] Das, T.K., Tomson, F.L., Gennis, R.B. et al. (2001) Biophysical J. 80, 2039–2045. [112] Dyer, R.B., Peterson, K.A., Stoutland, P.O. and Woodruff, W.H. (1991) J. Am. Chem. Soc. 113, 6276–6277. [113] Powers, L., Chance, B., Ching, Y. and Angiolillo, P. (1981) Biophysical J. 34, 465–498. [114] Thomson, A.J., Greenwood, C., Gadsby, P.M.A. et al. (1985) J. Inorg. Biochem. 23, 187–197. [115] Scott, R.A. (1989) Ann. Rev. Biophys. Biophys. Chem. 18, 137–158. [116] Argade, P.V., Ching, Y.C. and Rousseau, D.L. (1984) Science 225, 329–331. [117] Rousseau, D.L., Ching, Y.C. and Wang, J.L. (1993) J. Bioenerg. Biomem. 25, 165–176. [118] Tangen, E., Svadberg, A. and Ghosh, A. (2005) Inorg. Chem. 44, 7802–7805. [119] Pellicena, P. Karow, D.S., Boon, E.M. and Marletta, M.A. (2004) Proc. Natl. Acad. Sci. U.S.A. 101, 12854–12859. [120] Brucker, E.A., Olson, J.S. Ikeda-Saito, M. and Phillips, G.N. (1998) Proteins 30, 352–356. [121] Copeland, D.M., West, A.H. and Richter-Addo, G.B. (2003) Proteins 53, 182–192. [122] Copeland, D.M., Soares, A.S., West, A.H. and Richter-Addo, G.B. (2006) J. Inorg. Biochem. 100, 1413–1425. [123] Rich, A.M., Armstrong, R.S. Ellis, P.J. and Lay, P.A. (1998) J. Am. Chem. Soc. 120, 10827–10836. [124] Hori, H., Ikeda-Saito, M. and Yonetani, T. (1981) J. Biol. Chem. 256, 7849–7855. [125] Hüttermann, J., Burgard, C. and Kappl, R. (1994) J. Chem. Soc. Faraday Trans. 90, 3077–3087. [126] Pinakoulaki, E., Gemeinhardt, S., Saraste, M. and Varotsis, C. (2002) J. Biol. Chem. 277, 23407–23413. [127] Benko, B., and Yu, N.-T. (1983) Proc. Natl. Acad. Sci. U.S.A. 80, 7042–7046. [128] Tomita, T., Haruta, N., Aki, M., Kitagawa, T. and Ikeda-Saito, M. (2001) J. Am. Chem. Soc. 123, 2666–2667. [129] Wang, J., Lu, S., Möenne-Loccoz, P. and Ortiz de Montellano, P. (2003) J. Biol. Chem. 278, 2341–2347. [130] Ding, X.-D., Weichsel, A., Andersen, J.F. et al. (1999) J. Am. Chem. Soc. 121, 128–138. [131] Maes, E.M., Walker, F.A., Montfort, W.R. and Czernuszewicz, R.S. (2001) J. Am. Chem. Soc. 123, 11664–11672. [132] Mukai, M., Ouellet, Y., Ouellet, H. et al. (2004) Biochemistry 43, 2764–2770. [133] Obayashi, E., Tsukamoto, K., Adachi, S. et al. (1997) J. Am. Chem. Soc. 119, 7807–7816. [134] Shimizu, H., Obayashi, E., Gomi, Y. et al. (2000) J. Biol. Chem. 275, 4816–4826. [135] Singh, U.P., Obayashi, E., Takahashi, S. et al. (1998) Biochim. Biophys. Acta 1384, 103–111. [136] Suzuki, N., Higuchi, T., Urano, Y. et al. (2000) J. Am. Chem. Soc. 122, 12059–12060. [137] Linder, D.P., Rodgers, K.R., Banister, J. et al. (2004) J. Am. Chem. Soc. 126, 14136–14148. [138] Linder, D.P. and Rodgers, K.R. (2005) Inorg. Chem. 44, 1367–1380. [139] Das, T.K. Couture, M. Ouellet, Y. et al. (2001) Proc. Natl. Acad. Sci. U.S.A. 98, 479–484. [140] Hirota, S., Ogura, T., Appleman, E.H. et al. (1994) J. Am. Chem. Soc. 116, 10564–10570. [141] Potter, W.T., Tucker, M.T., Houtchens, R.A. and Caughey, W.S. (1987) Biochemistry 26, 4699–4707. [142] Bruha, A. and Kincaid, J.R. (1989) J. Am. Chem. Soc. 110, 6006–6014. [143] Nakamoto, K., Paeng, I.R. and Kuroi, T. (1988) J. Mol. Struct. 189, 293–300. [144] Wagner, W.D., Paeng, I.R. and Nakamoto, K. (1988) J. Am. Chem. Soc. 110, 5565–5567. [145] Burke, J.M., Kincaid, J.R., Peters, S. et al. (1978) J. Am. Chem. Soc. 100, 6083–6088. [146] Walters, M.A., Spiro, T.G. Suslick, K.S. and Collman, J.P. (1980) J. Am. Chem. Soc. 102, 6857–6858.

CO, NO, and O2 as Vibrational Probes

123

[147] Collman, J.P. Brauman, J.I. Halbert, T.R. and Suslick, K.S. (1976) Proc. Natl. Acad. Sci. U.S.A. 73, 3333–3337. [148] Chartier, F.J.M., Blais, S.P. and Couture, M. (2006) J. Biol. Chem. 281, 9953–9962. [149] Hu, S. and Kincaid, J.R. (1991) J. Am. Chem. Soc. 113, 4815–4822. [150] Bangcharoenpaurpong, O., Rizos, A.K., Champion, P.M. et al. (1986) J. Biol. Chem. 261, 8089–8092. [151] Sjodin, T., Christian, J.F., MacDonald, I.D. et al. (2001) Biochemistry 40, 6852–6859. [152] Chottard, G., Schappacher, M., Ricard, L. and Weiss, R. (1984) Inorg. Chem. 23, 4557–4561.

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 5

Nuclear Resonance Vibrational Spectroscopy — NRVS W. Robert Scheidta , Stephen M. Durbinb , and J. Timothy Sagec a

The Department of Chemistry and Biochemistry, 251 Nieuwland Science Hall, University of Notre Dame, Notre Dame, Indiana 46556 b Department of Physics, Purdue University, West Lafayette, Indiana 47907 c Department of Physics and Center for Interdisciplinary Research on Complex Systems, Northeastern University, Boston, MA 02115

ABBREVIATIONS Ligands: Porph DPIXDME MPIXDME PPIXDME TPP OEP TpivPP HIm 1-MeIm 2-MeHIm Other: Np Mb DFT NRVS VDOS KED B3LYP BP86

a generalized porphyrin dianion dianion of deuteroporphyrin IX dimethyl ester dianion of mesoporphyrin IX dimethyl ester dianion of protoporphyrin IX dimethyl ester dianion of meso-tetraphenylporphyrin the dianion of octaethylporphyrin dianion of picket fence porphyrin imidazole 1-methylimidazole 2-methylimidazole pyrrole nitrogen atom myoglobin density functional theory nuclear resonance vibrational spectroscopy vibrational density of states kinetic energy distribution Becke–Lee–Young–Parr hybrid functional with 20% Hartree–Fock exchange correlation functional Becke–Perdew 1986 exchange–correlation functional

1. INTRODUCTION Nuclear Resonance Vibrational Spectroscopy (NRVS) is a recent synchrotron-based vibrational spectroscopy technique. What are the particularly useful features of this vibrational spectroscopy technique? This chapter will attempt to answer that question

Nuclear Resonance Vibrational Spectroscopy

125

by showing the vibrational information the NRVS experiment uniquely makes available. The general utility of vibrational spectroscopy is well understood and need not be given here. However, the molecular complexity of many biological systems means that site-selective techniques are needed to obtain vibrational information for active sites. A certain degree of site selectivity is provided by resonance Raman [1], difference infrared [2,3], and femtosecond coherence spectroscopies [4,5]. Nonetheless, important active-site vibrational information is not available with these techniques. The selection rules for both infrared and Raman spectroscopies restrict the observation of many important vibrations. An important example of effectively forbidden vibrations are the in-plane Fe modes of heme derivatives, which could be expected to provide significant information on the strength of the bonds. Other heme vibrations have simply not been identified even though not forbidden by selection rules. Reactive modes, which could provide information on reaction energetics or conformational changes, are at low frequency and rarely identified by other vibrational probes. NRVS (sometimes called nuclear (resonant) inelastic X-ray scattering) is selective for vibrations involving displacement of Mössbauer-active nuclei, these include 83 Kr, 119 Sn, 157 Eu, and 57 Fe [6–23]. The isotope 57 Fe has yielded most of the results to date, and of course iron is at the center of hemes and heme proteins as well as many other biologically important centers. While NRVS provides sensitivity only to the Mössbauer-active probe nuclei, it does yield the complete vibrational spectrum of the probe nucleus. Thus, the NRVS experiment for iron selectively yields the complete set of modes that involve motion of the iron atom. The method has a selectivity that is reminiscent of that of resonance Raman spectroscopy, but again note the significant advantage that NRVS is not subject to the optical selection rules of Raman or infrared spectroscopy. Indeed, NRVS provides the ultimate limit in selectivity because only the vibrational dynamics of the probe nucleus contribute to the observed signal. Moreover, the NRVS intensity is directly related to the magnitude and direction of the motion; hence, the method has a unique quantitative component in the measured vibrational spectrum. A special feature of NRVS is that it will allow the detection and measurement of all iron–ligand modes. For hemes, these include in-plane iron vibrations that have not yet been reported by resonance Raman studies and the iron–imidazole stretch that has not been identified in six-coordinate porphyrins. Other modes that can be investigated include those of heme doming, which is expected to be a low-frequency mode. As discussed subsequently, the wealth of vibrational data leads to the (pleasant) problem of making a complete assignment of the vibrational data. NRVS has significant potential for probing the dynamics of Fe-containing molecules of biological interest. Although this chapter will emphasize applications to heme derivatives, NRVS has also been applied to iron cluster species [14,24–26]. In simplest terms, NRVS results from the combination of nuclear excitation (classical Mössbauer effect) and molecular vibrations. As shown by Mössbauer [27], the eponymous effect results from the absorption of a photon to excite the nucleus to an excited nuclear state without a change in the vibrational state of the nucleus, that is, a recoil-free absorption of a photon of energy exactly equal to the energy difference between the ground and excited nuclear states. This occurs when the nucleus is in a bound state, and its motion is thus quantized; the normal modes involving motion of the nucleus can only change their vibrational energies by discrete amounts that are characteristic of the molecular mass and the bound potential. As was then pointed out by Visscher [28] and Singwi [29], there should also be peaks displaced from the recoil-free

126

W.R. Scheidt et al.

resonance by energies corresponding to vibrational quanta. Despite this early prediction, the actual observation of such vibrationally shifted nuclear absorptions (absorption with recoil) requires a suitable synchrotron source. Mössbauer spectra with sidebands cannot be observed in a classical Mössbauer experiment, which records the transmission of Doppler-shifted photons emitted by nuclei in a moving source. Although the vibrational energy shifts are small (a few hundred wave numbers or tens of meV), the Doppler velocities needed to provide such shifts amount to hundreds of meters/sec [30], clearly beyond the usual instrumental velocities. A second, more fundamental, limitation stems from the fact that the resonance is now broadened by vibrational lifetimes, typically of the order of 1 meV. Thus, a strong vibrational sideband with an integrated cross section of the order of 1% of the recoilless resonance will have a peak cross section 6–7 orders of magnitude smaller than the recoilless resonance which has a linewidth of a few neV. With rare exceptions [30], there is little possibility of observing such a weak feature in a conventional transmission experiment. NRVS overcomes these difficulties by using a tunable X-ray synchrotron source. The incident X-ray beam must have an energy width that is narrow compared to vibrational bandwidths in order to obtain well-resolved spectra and the source must be of unusually high brightness. Many technical and engineering difficulties also had to be resolved, but these hurdles have now been overcome to the point where NRVS is a useful vibrational spectroscopy for heme compounds and proteins. Details of the experimental requirements are presented below, followed by examples of NRVS studies of several heme systems.

2. EXPERIMENTAL METHODOLOGY A third-generation synchrotron source is the key development, but not the only development, that has made NRVS a viable experimental technique [6,7]. Radiation from such a third-generation source enables NRVS measurements to be made in a time-resolved fluorescence excitation mode. The basic experimental setup is depicted schematically in Fig. 1. X-rays are generated from the charged particle beam by an undulator; the high beam brightness allows the use of a tunable, ultra-high-resolution monochromator to give Resolution = 7–8 cm–1 heat-load monochromator undulator

high resolution monochromator avalanche photodiode detector

slits

APD

1 eV

14.413 keV

1 meV 0

sample 0

Fig. 1. Schematic diagram illustrating the experimental setup for NRVS including the highresolution monochromator. The avalanche photodiode (APD) is disabled during the X-ray pulse in order to gate out elastically scattered 14.4 keV photons and background fluorescence, so that only delayed photons corresponding to nuclear absorption are detected.

Nuclear Resonance Vibrational Spectroscopy

127

a beam with an extremely small energy dispersion and acceptable photon fluxes [31,32]. The astonishingly narrow energy width of the beam (1.0 meV or 0.001 eV) can be contrasted with the energy of the beam (14,413 eV for iron); the experimental resolution of the vibrational spectrum is 7–8 cm−1 . The sample is irradiated with photons selected around the resonance of the Mössbauer isotope under investigation; the resonance for 57 Fe is 14.413 keV and all further experimental information is given for this specific isotope. A typical experimental scan would range from −20 meV below the recoilfree resonance energy to +100 meV above the resonance energy (−160 cm−1 below to 800 cm−2 above). With current instrumentation (at Sector 3-ID-D of the Advanced Photon Source at Argonne National Laboratory), counts are recorded for 5 s every 0.2 meV while scanning over the selected incident X-ray energy range. The number of scans will depend on the amount of the Mössbauer-active iron nuclei (57 Fe) in the sample; recent model compound studies have required 4–10 scans to achieve adequate signal, while protein samples have required measurements extending over a greater than 24-hr period. Photons that are absorbed by the sample at any given wavelength are determined by measuring the resulting 6.4 keV atomic fluorescence emitted from the excited 57 Fe atoms with an avalanche photodiode detector (APD) [33]. Thus, while detection is by fluorescence, the experiment is truly an absorption phenomenon. Fig. 2 illustrates the absorption and fluorescence phenomena. Current APDs are about 1 cm2 ; the development of larger-area APDs or other detector developments would allow detection of a larger fraction of the total emission. Pulses from the APD were recorded by a counter that was enabled after a delay in order to discriminate the desired fluorescence photons (nuclear resonance signal), which arrive with a delay on the order of the 141 ns excitedstate lifetime of 57 Fe, from the large background of electronically scattered 14.4 keV photons, which arrive in coincidence with the X-ray pulse [34]. The importance of a subnanosecond, synchrotron-pulsed photon beam, rather than a continuous beam of photons, to the success of the measurement is thus clear. The background count rate, due to electronic noise and timing errors, is typically 0.03 Hz or less in the presence of a beam with an X-ray flux of 109 Hz. Solid samples are typically polycrystalline samples mulled with a minimum amount of Apiezon grease and placed in a milled polystyrene sample holder transparent to the X-ray beam. The dimensions of the sample are fixed by the size of the usable X-ray beam and by X-ray absorption. The original size of the NRVS sample holders was approximately 2 × 4 × 15 mm3 . A recent development is the use of focusing mirrors on the incoming beam so that the size of the beam at the sample is now 05 × 05 mm2 . This allows the use of a sample holder of size 1 × 2 × 10 mm3 , which allows for the use of significantly less sample. The same sample holder can be used for measurements on frozen solutions. The sample holder is mounted on the cold finger of a He flow cryostat with X-ray access through either a Mylar window or a beryllium dome. For NRVS measurements on oriented single crystals, a single-crystal orienter with its rotation axis perpendicular to the incident X-ray beam is used; crystals are mounted on a eucentric goniometer head. A gas stream from the boil-off of liquid nitrogen cools the crystalline sample to 80–100 K. A single-crystal specimen for NRVS measurements should be large enough to contain about 1017 atoms of 57 Fe; clearly, 95% enrichment must be used for the single crystals. An ideal oriented single crystal for NRVS has molecules with a single orientation of porphyrin planes in the crystal lattice. Crystals must be independently aligned to have the goniometer head rotation axis parallel to the porphyrin plane; the

128

W.R. Scheidt et al.

EX Evib

Fluorescence x-ray

Conversion electron

E0

x-ray photon

Energy

0 EL

Energy

L

0 Nuclear Absorption

K

EK Atomic Emission

Fig. 2. Diagram illustrating absorption and fluorescence X-rays in the NRVS experiment. The left-hand panel illustrates the basic NRVS absorption. The Mössbauer resonant condition is the excitation between the nuclear ground state (0) and the nuclear excited state (E0 ). However, additional resonances can occur when when the excitation energy corresponds to the addition of a vibrational energy quantum to state EX . The right-hand panel shows the decay of the excited state, with a lifetime of ∼141 ns, to yield a fluorescence X-ray photon with energy of 6.4 keV.

rotation angles that yield porphyrin plane orientations either parallel or perpendicular to incident X-ray beam are readily available. The resulting spectrum is dominated by a central resonance due to the nuclear excited state of 57 Fe at E0 = 14413 keV (the recoilless absorption line). Peaks at energies higher than E0 indicate excitation of discrete vibrational modes coincident with X-ray absorption by the 57 Fe nucleus. Peaks at energies lower than E0 correspond to equivalent vibrational modes with annihilation of vibrational quanta. The peaks are frequently, but not always, relatively sharp. An example of such a spectrum is given in Fig. 3. The vibrational frequency  is determined by the energy shift hc = E − E0 of a one-quantum transition from the recoilless nuclear resonance. A constant background level, estimated on the basis of the observed signal at the high or low (negative) energy extremes, is subtracted from the observed spectrum. Normalization of the spectrum according to Lipkin’s sum rule [7,35] then yields an excitation probability S , with peak areas for each mode representing the mean square amplitude of the Fe. Removal of multiphonon

Nuclear Resonance Vibrational Spectroscopy

129 E–E0 (meV)

–20 25 × 103

0

20

40

60

recoilless absorption (Mössbauer effect)

80

100

57Fe(TPP)(NO)

T = 80 K

Counts

20 NRVS = absorption with (quantized) recoil

15

10

×30

5

0 0

200

400

600

800

Frequency (cm–1)

Fig. 3. Diagram illustrating the raw NRVS data obtained from a powder sample of [Fe(TPP)(NO)]. The classical Mössbauer line is the energy zero of the experiment.

contributions, temperature dependence, and an overall factor proportional to inverse frequency yields the Fe-weighted vibrational density of states, D  (VDOS) [36]. The excitation probability S  and density of states D  are illustrated in Fig. 4 for [Fe(TPP)(NO)]. The excitation probability S  has peak areas for each mode representing the mean square amplitude of the iron. The VDOS obtained from NRVS provides direct 2 . In NRVS, a mode composition information on the iron mode composition factors eFe factor is given by the area under the curve for unique frequencies, as illustrated in Fig. 4. Peak deconvolution may sometimes be required in order to completely resolve all peaks. 2 for any given vibrational The physical significance of the mode composition factor eFe mode  is (i) that it gives the fraction of the kinetic energy associated with the motion of the iron atom in mode  and (ii) that it is proportional to the contribution of mode  to the mean square displacement of the iron atom. NRVS thus samples the kinetic energy distribution (KED) of the vibrational modes, which may be contrasted with the common theoretical description of normal modes in terms of a potential energy distribution (PED). A more complete description of these issues may be found in reference [18]. Measurements are typically made at cryogenic temperatures to minimize multiphonon contributions. The requirement that the ratio S  /S − equal the Boltzmann factor exphc/kB T yields the sample temperature. Temperature measurements made in this fashion are more reliable than readings from a sensor attached to a sample holder, which are typically 15 K lower. Most polycrystalline porphyrin samples are measured near 30 K. The absorption length is much shorter for the 6.4 keV photons that constitute most of the experimental signal than for the incident 14.4 keV photons, and all measurements were recorded with the X-ray beam at grazing incidence. The photon flux from the ultra-high-resolution monochromator is necessarily relatively low. In order to maximize signal with minimum scanning time, samples are

130

W.R. Scheidt et al.

S(ν) (cm)

6 × 10–4

57 Fe(TPP)(NO) T = 80 K

4

2

5 × 10–2

0.8

D(ν) (cm)

4

0.6 acoustic modes 2 ΣeFe = 0.25

3 2

2

0.4 eFe 0.2

1 0

0

100

200

300

400

500

600

0

Frequency (cm–1)

Fig. 4. Diagram showing the excitation probability S (top panel) and the vibrational density of states D (bottom panel) for [Fe(TPP(NO)]. The top panel shows the NRVS experimental values (gray line), the individual Lorentzian fitting components (dotted line), and the overall fit (solid line). The hatched area in the top panel corresponds to the recoil fraction for the mode at 538 cm−1 . The bars in the bottom panel show the e2Fe values, which give the fraction of the kinetic energy in each mode that results from motion of the iron. The crosshatched area at very low frequency approximately represents the area fraction attributable to the motion of the entire molecule (acoustic modes).

enriched in the Mössbauer active 57 Fe isotope. We have been enriching all samples to ∼95%. For porphyrin complexes using the sample holder described above, a sample of up to ∼10 mg can be used. This is roughly 8 × 1018 atoms of 57 Fe, or somewhat higher than the minimum required. At the present time, protein samples need to have solution concentrations in the range of 10 mM. Information on the direction of the iron motion (parallel or perpendicular to the porphyrin plane) can be obtained from appropriate, oriented single-crystal measurements. In the ideal crystalline case, the porphyrin planes are all parallel (coplanar). In this case, the crystal can be oriented so that the NRVS measurement can be made with the incident X-ray beam either parallel to the porphyrin planes or perpendicular. Such orientations lead to intensity enhancements in the in-plane or out-of-plane modes respectively. This condition can always be obtained for crystalline samples that are found in the triclinic crystal system. A crystal orientation for which all porphyrin planes are parallel to the incident beam can be obtained for all monoclinic crystal systems with no more than a single molecule in the asymmetric unit of the crystal, whereas the perpendicular orientation is likely to be inaccessible. Finally, the imposition of crystallographically required symmetry can yield solid-state systems where the desired parallel and perpendicular plane orientations can be achieved.

Nuclear Resonance Vibrational Spectroscopy

131 Fe(TPP)(1-MeIm)(NO)

powder

0.03

Density of States (cm–1)

0.02

0.01

0 perpendicular parallel

0.03

0.02

0.01

0

0

100

200

300

400

500

600

Frequency (cm–1)

Fig. 5. Diagrams illustrating the in-plane and out-of-plane intensity enhancements, relative to the powder spectrum, for the six-coordinate complex [Fe(TPP)(NO)(1-MeIm)]. In this case, the crystallographic symmetry of the unit cell leads to porphyrin planes that are within 13.8 of being parallel to a common plane rather than the ideal value of 0 . Nonetheless, useful intensity enhancement differences sufficient for spectral assignment are still evident.

Fortunately, some relaxation in these conditions can still lead to useful information on mode assignment. Fig. 5 illustrates the in-plane or out-of-plane intensity enhancements obtained from single-crystal measurements for [Fe(TPP)(1-MeIm)(NO)] compared with its powder spectrum. In this crystal system [19,20], there are four distinct orientations of porphyrin planes that coincidentally are all within 13.8 of being parallel to a common plane so that in the nominal “out-of-plane” orientation and resulting spectrum, the X-ray beam was ∼13.8 from exact orthogonality to all of the porphyrin planes. A similarly small “misorientation” pertains to the in-plane orientation. As shown in Fig. 5, this is sufficient to clearly identify in-plane/out-of-plane modes and regions where in-plane and out-of-plane modes overlap. Slightly less 57 Fe is required for a single-crystal measurement than a powder sample; the approximately 1 × 1017 atoms of 57 Fe needed means that crystals greater than ∼2–5 times the volume of a crystal needed for an X-ray diffraction study will be sufficient for an oriented single-crystal measurement. The final methodology issue concerns the important matter of the assignment of the NRVS-observed modes. As has been noted above, the experimentally obtained vibrational density of states is the complete set of modes that have an iron motion component in the mode. Frequency assignments obtained from infrared (IR) and/or resonance Raman (rR) experiments can be compared with the observed frequencies of modes obtained from NRVS. However, even if there is good agreement between NRVS and the other vibrational technique, which does not always seem to be the case, this will lead to only a limited number of assignments of the NRVS modes. Thus, while comparisons with prior results may lead to some useful assignments, if a complete set

132

W.R. Scheidt et al.

of assignments is desired, then additional efforts need to be made. The use of selective isotopic labeling appears to be of limited value for the low-frequency modes, because these modes are often delocalized with contributions to the energy distribution from many atoms, i.e., the modes involve the motion of a relatively large number of atoms. Moreover, the expected shifts from such isotopic substitutions are nearly always below the current resolution limit (7–8 cm−1 ) of the NRVS experiment. In principle, a complete set of calculated mode assignments can be obtained either by an empirical approach (normal coordinate analysis) or by de novo calculations (density functional theory, DFT). Both computational approaches begin with an isolated porphyrin molecule, while the NRVS data are obtained from ordered structures where intermolecular coupling may be important. Normal-mode calculations are an intrinsically underdetermined problem; however, two features of NRVS data provide constraints in the fitting of the spectra and provide confidence in the reliability of the final results. The first important feature of NRVS data is that not only are all frequencies of the iron modes observed but that the iron amplitudes of each mode are known quantitatively. Secondly, a single-crystal measurement allows the unambiguous identification of many modes as either having in-plane or out-of-plane character. The experimentally determined vibrational density of states from NRVS can be compared with an Fe VDOS obtained from the normal coordinate analysis. The procedure has been described in more detail in reference [22]. The use of DFT calculations for mode assignments requires the determination of an energy-minimized structure of the molecule in question, a computationally intensive process. Theoretical calculations on porphyrin systems have been traditionally simplified by truncating the porphyrin skeleton; DFT calculations have typically used the porphine nucleus, replacing all peripheral groups with hydrogen atoms. However, it has become clear that many features of the vibrational spectrum are strongly dependent on peripheral groups of the porphyrin moiety. The importance of the peripheral groups on the vibrational spectrum leads to the necessity of including the complete molecule in the DFT calculation and thereby substantially increasing the computational time required. The importance of the peripheral groups on the vibrational spectroscopy of the iron atom can be readily seen from Fig. 6 which displays the 57 Fe excitation probabilities observed for five different five-coordinate nitrosyl complexes. Although significant differences might be expected for meso-substituted compared with -substituted systems, i.e., tetraphenylporphyrin derivatives compared to octaethylporphyrin derivatives, even the three species in the series [Fe(MPIXDME)(NO)], [Fe(DPIXDME)(NO)], and [Fe(PPIXDME)(NO)], where only two (of the eight total) peripheral groups change, show noticeable variation in the in-plane region (300–400 cm). Given a calculated energy-minimized structure, both the frequencies and relative Cartesian displacements of each atom for the 3N -6 modes of an N -atom molecule can be calculated. As was noted previously [18], a convenient comparison between 2 ; values measurements and calculations is given by the mode composition factors eFe 2 of the mode composition factors ej for all other atoms are available for the theoretical calculations. For a given mode, the sum of the mode composition factors ej2 must 2 from both theory and experiment are on the equal 1. Furthermore, the values of eFe same quantitative scale. Comparisons between theory and experiment can thus provide additional assurance of the quality of the vibrational assignments.

Nuclear Resonance Vibrational Spectroscopy (A)

Excitation Probability (10–4 cm)

6 4 2 0 6 4 2 0 6 4 2 0 6 4 2

Fe(TPP)(NO)

(B)

Fe(OEP)(NO)

(C)

0 6 4 2 0 6 4 2 0 6 4 2 0

133

Fe(MPIXDME)(NO)

(D)

Fe(DPIXDME)(NO)

(E)

Fe(PPIXDME)(NO)

(F)

Fe(OEP)

(G)

100

200

Fe(OEP)(Cl)

300

400

500

600

Frequency (cm–1)

Fig. 6. Diagram illustrating the NRVS excitation probabilities for five differing [Fe(Por)(NO)] derivatives. Also shown are the excitation probabilities for four-coordinate [Fe(OEP)] and fivecoordinate [Fe(OEP)(Cl)].

3. APPLICATIONS In this section, we present a limited number of results that exemplify the vibrational information that NRVS can provide; information from heme systems with the diatomic ligands NO and CO are given. We first describe analyses of NRVS data obtained for the five-coordinate nitrosyl complex [Fe(TPP)(NO)] and observations on related nitrosyl species. In retrospect, the analysis of the vibrational modes of the nitrosyl systems probably presents a significantly more difficult challenge than most iron porphyrinate systems. NRVS data were obtained for [Fe(TPP)(NO)] both as a powder sample and as an oriented single-crystal array. Both samples were enriched to 95% 57 Fe. Because we were unable to obtain one single crystal large enough for measurements, we resorted to a 5 × 5 array of smaller crystals with a common orientation; data were obtained with all porphyrin planes at an angle of 6 to the incident X-ray beam. This orientation ensures that all in-plane modes are enhanced in intensity relative to the powder (random orientation) sample while the out-of-plane mode intensities are reduced relative to the powder sample. The spectra are shown in Fig. 7. A total of 14 frequencies are observed,

134

W.R. Scheidt et al.

powder

Fe VDOS (cm)

Fe(TPP)(NO)

oriented crystals 0.06 0.04

in-plane

NRVS

out-of-plane

out-of-plane

0.02 0 200

400

600

–1)

Frequency (cm

Fig. 7. NRVS spectra for five-coordinate [Fe(TPP)(NO)]. Powder sample and oriented crystal array measurements are illustrated.

with the lowest-frequency feature most likely dominated by acoustic modes, which involve overall translation of the entire molecule. The data of Fig. 7 show that the modes at 74, 128, and 538 cm−1 of the oriented crystal spectrum are less intense than the equivalent peaks in the powder spectrum. The data thus indicate that these frequencies represent modes with out-of-plane character. Conversely, the remaining modes, in particular those at 313, 333, and 470 cm−1 , clearly have strong in-plane character. The first mode assignments for [Fe(TPP)(NO)] utilized an empirically derived force field to carry out the normal mode analysis. Rai et al. [22] started normal coordinate analysis for [Fe(TPP)(NO)] using force constants developed for nickel porphyrins [37–39]. These force fields adequately described the high-frequency modes seen by resonance Raman and commonly known as marker bands. However, the iron VDOS did not match those observed in the NRVS experiment. An improvement in the force field for iron was obtained by the Jacobian Determinant method [40] along with trial and error adjustments to force constants in order to match the experimental frequencies and amplitudes. These adjustments were constrained by keeping the frequencies of 10 marker lines (these are at frequencies of 1200–1600 cm−1 , well above the iron VDOS frequencies) to within 20 cm−1 of the published values. This procedure resulted in very good agreement between the experimental iron VDOS and the best-fit normal mode calculation [41]. Experimentally observed modes were characterized by the classification scheme developed for porphyrin derivatives with D4h symmetry [37,42,43]. Although assignments for almost all modes could be made, it is also clear that the observed modes are very delocalized with significant mixing of the classification scheme modes. Four categories of modes are noted: (i) in-plane modes between 237 and 406 cm−1 , (ii) ligand modes at 538 and 470 cm−1 , (iii) modes involving vibrations of the peripheral phenyl groups and (iv) low-frequency out-of-plane modes at 74 and 128 cm−1 . The in-plane modes in D4h symmetry are Raman inactive, and were previously unobserved. Although in the case of [Fe(TPP)(NO)] the removal of D4h symmetry caused by the axial NO ligand might allow Raman observation of these in-plane vibrations, they have not been reported. On the other hand, NRVS will always allow observation of these in-plane modes. From the normal coordinate analysis, near-degenerate pairs of bands at 237 and 247 cm−1 , 312 and 333 cm−1 , and 400 and 406 cm−1 are assigned as modes with in-plane motion of iron and in-plane core ring vibrations. Obviously, the frequency and

Nuclear Resonance Vibrational Spectroscopy

135

amplitudes of these in-plane modes should display sensible variation as a function of spin state, oxidation state, and coordination number of iron. Such correlations should be most informative, but are still somewhat in the future. Additionally, the band in the region of 538 cm−1 was assigned to the Fe NO stretch. As can be seen in Fig. 6, there is a band between 520 and 540 cm−1 for all five-coordinate nitrosyl complexes. This band is not present in four-coordinate Fe(OEP) (Fig. 6), a convenient standard of reference, and is clearly seen as an out-of-plane mode in the single-crystal array of [Fe(TPP)(NO)]. However, the observed frequency of the Fe N stretch is slightly different than that observed in solution by Raman spectroscopy. The band at 470 cm−1 was characterized as an Fe N O bending mode from the normal coordinate analysis of [Fe(TPP)(NO)]. Both of these modes merit further discussion that is given in the section following on DFT assignment results. Two low-frequency modes at 75 and 128 cm−1 that are shown by the oriented crystal measurements to be out-of-plane modes are expected to be related to the doming mode, and are shown so by the normal coordinate analysis. The doming mode is one in which the movement of the iron atom out of plane is countered by the opposite motion of pyrrole atoms, especially those at the periphery of the porphyrin core. A detailed understanding of doming modes is important in understanding the reactive modes of heme proteins that bind and dissociate diatomic ligands [4,21,44]. However, although the (pure) doming mode makes a substantial contribution to these low-frequency modes, the modes are delocalized with contributions, inter alia, from both Np Fe NO ligand tilting and Fe N O bending modes. A complete description of the results from the normal coordinate analysis fitting and mode assignment can be found in the original publication [22]. A concise summary might be that the low-frequency modes are, to a much larger extent than might have been initially expected, highly delocalized. The vibrational density of states in [Fe(TPP)(NO)] was further studied by the use of DFT to predict and assign modes [45]. As noted earlier, the observed frequencies depend on the identity of the peripheral groups, and thus, the DFT calculation requires the inclusion of all peripheral groups. The 361 electrons of [Fe(TPP)(NO)] thus present a significant computing problem. DFT mode prediction requires an energy-minimized structure, which can be compared with the experimentally determined structure as an indication of the reliability of the calculations. Two different DFT calculations for [Fe(TPP)(NO)] were made for mode assignments: one based on the B3LYP functional and a second based on a BP86 functional. There is overall general agreement on the nature of the mode assignments and predicted frequencies with one major exception: the prediction of (Fe NO). These DFT predicted results are shown in Fig. 8 for both oriented and randomly oriented molecules. The B3YLP calculation predicts the frequency of (Fe NO) to be at 386 cm−1 , much lower than the 540 cm−1 actually observed; the BP86 calculation predicts this band to be at 623 cm−1 , higher than actually observed. A calculation for the simpler species [Fe(porphine)(NO)] predicts a value of 565 cm−1 ; the experimental value is not known but is likely to be in the 520–540 cm−1 range seen for other nitrosyls. The Fe NO bond distance calculated in the B3LYP functional is significantly longer than the observed value while the BP86 calculation yielded a distance a bit shorter than the experiment value. The energy minimized Fe NO distance in [Fe(porphine)(NO)] is closest to the canonical experimental value of 1.728 Å [46,47]. All of the DFT calculations correctly

W.R. Scheidt et al.

Fe VDOS (cm)

Fe VDOS (cm)

136

powder oriented crystals

Fe(TPP)(NO) 0.06

B3LYP 540

0.04 0.02

386

0 200

0.06

400

600

BP86

0.04

540

0.02 0

623 0

100

200

300

400

500

600

700

Frequency (cm–1)

Fig. 8. Diagram illustrating the DFT-predicted NRVS signal. Each panel shows the predicted powder and oriented crystal spectrum. Top panel B3LYP functional, bottom panel BP86 functional. The peak for the Fe NO stretch has been artificially shifted as indicated and the predicted frequencies are convoluted with a Gaussian line shape function.

predict the Fe N O angle, the NO ligand tilt, and the in-plane asymmetry in the Fe Np bonds that are experimentally observed [46]. All three DFT calculations suggest that the observed Fe NO stretch mode has significant Fe N O bend character as well as Fe NO stretch character. The richness and complexity of the vibrational spectra of the five-coordinate nitrosyls are further shown by the fact that the mode composition factors for the Fe NO stretch for the five complexes shown in Fig. 6 vary significantly. Values of e2 Fe (fraction of the mode kinetic energy carried by iron motion) range from 0.33 to 0.23 (0.30 for [Fe(TPP)(NO)]); 0.34 is the value expected for a two-body 57 Fe NO oscillator. A final feature of the richness of the iron modes is that the three-atom FeNO group, which classically should have three localized vibrational modes: (i) the N O stretch, (ii) the Fe NO stretch, and (iii) the Fe N O bend, is vibrationally more complex. Only the first two modes are found to be significantly localized in the DFT predictions. However, the Fe N O bend is delocalized, with contributions found in a number of different in-plane modes. The band assigned as the Fe N O bend mode in the normal coordinate analysis is differently assigned by the DFT calculations as a pair of in-planes modes at 474 and 476 cm−1 . These near-degenerate modes have relatively small amounts of iron motion along with rotation of the two pyrrole rings orthogonal to the iron motion direction that lead to effective Fe Np stretches. This pair of modes is closely related to 50 calculated for four-coordinate Fe(porphine) with D4h symmetry [48]. As shown by the DFT predictions and the oriented crystal measurement, a number of additional in-plane modes are found in the 200–500 cm−1 region. In D4h symmetry, all in-plane modes should be found as degenerate (orthogonal) pairs with equal e2 Fe values. The presence of a bent Fe N O linkage will break the fourfold symmetry; near-degenerate mode pairs are observed. In addition to the 474 and 476 cm−1 pair described above that are not resolved in the NRVS experiment, a resolved pair observed at 241 and 253 cm−1 corresponds to Fe Np stretches. An experimentally unresolved pair (predicted by DFT at 410 and 413 cm−1 ) is a second set of Fe Np stretches in which

Nuclear Resonance Vibrational Spectroscopy

137

the bond stretch is at a maximum, the prior set has the Fe Np bond stretch minimized. The relative e2 Fe magnitudes of modes in this region reveal that these have relatively small amplitudes of iron motion compared to the bands in the 312–333 cm−1 region. This is consistent with a large fraction of the kinetic energy of the mode associated with porphyrin ring motion. The bands observed at 312 and 333 cm−1 were originally assigned to a pair of Fe Np stretching modes, with the presence of two bands and amplitude asymmetry being attributed to different amounts of Fe N O bending character [18,22]. The DFT predictions suggest a significantly more complex picture of the modes in this region. Rather than a pair of modes, the DFT calculation suggests that this region of the vibrational spectrum has contributions from two pairs of modes at 308/318 cm−1 and 317/334 cm−1 . The modes at 308, 318, and 317 cm−1 are not resolved, and appear in the experimental spectrum as a single feature; thus, the DFT prediction neatly reproduces the distinct amplitude differences in the 312 and 333 cm−1 features. One mode from each pair (those at 317 and 318 cm−1 ) exhibits in-plane iron motion approximately perpendicular to the FeNO plane. The remaining modes from each pair (those at 308 and 334 cm−1 ) exhibit iron motion approximately parallel to the FeNO plane. These latter two modes have, in addition, oppositely directed out-of-plane iron motion and significant motion of the nitrosyl nitrogen atom. Thus, the NO ligand contributes to the complexity of the spectrum in this region by imparting Fe NO stretch and Fe N O bend character to these two modes. A comparison of the DFT-predicted modes given in Fig. 8 with the experimental data presented in Fig. 7 shows that there is close correspondence in the average frequency, frequency splitting, and iron amplitudes. This suggests that the nature of the experimental modes observed have been reliably described [45]. Note, however, that the spectra given in Fig. 6 clearly show that the in-plane modes are strongly dependent on the peripheral substituents. DFT calculations for [Fe(porphine)(NO)] show that that the band at 470 cm−1 is related to the phenyl substituents and the detailed character of other bands are modified as well. Further study on predicting the nature of in-plane modes is clearly needed. Nevertheless, the importance of the peripheral groups on porphyrin vibrational dynamics is clearly demonstrated. The last group of observed frequencies are at very low frequency and have substantial out-of-plane iron motion. Observed modes (those that are clearly higher than the acoustic mode) are found at 74 and 128 cm−1 and their out-of-plane character confirmed by oriented crystal measurements. As noted earlier, knowledge of these low-frequency modes is essential to understanding the reactive modes of heme proteins, especially those heme proteins that bind (and dissociate) diatomic ligands [4,21,24]. Control of the out-of-plane displacement of iron in hemoglobin is believed to be essential to control reactivity and to communicate the ligation state of iron to the other subunits [49–52]. The B3LYP calculation predicts a frequency at 109 cm−1 for the dominant out-ofplane mode. The iron moves out of the plane while the pyrrole rings, but especially the -carbon atoms, all move in the direction opposite the iron motion. The atoms of the NO ligand move in near concert with the iron atom in this mode. The difference between the predicted frequency is somewhat larger than most of the modes discussed earlier. Two other modes (at 54 and 77 cm−1 ) are predicted to have out-of-plane iron motion, but the majority of the kinetic energy of these modes is in NO motion. The 77 cm−1 mode has a majority of its kinetic energy associated with movement of the oxygen atom. A more

138

W.R. Scheidt et al.

complete description of the modes can be found in reference [45]. Differences in the intensity of modes believed to be the doming modes have been observed in myoglobin [21]. The signals for these equivalent modes in myoglobin are much smaller, and it has been hypothesized that the mode acquires a more global character in the protein, reducing the iron amplitude [21]. Importantly, NRVS provides a quantitative method to explore such issues. The predicted modes stated in the foregoing have been based on the results of the B3LYP calculations, but the results from the BP86 calculations are similar, with the exception of the predicted Fe NO stretch frequency. The DFT predictions support and extend the earlier conclusion that the low-frequency modes are quite delocalized; even modes that one might expect to be localized (for example, the Fe N O bend) are mixed with other modes. The vibrational spectrum is found to be quite complex from the perspective that the modes, even those of the FeNO group, are quite mixed and delocalized. Preliminary NRVS analyses for nitrosylmyoglobin (MbNO) and a related sixcoordinate nitrosyl derivative [Fe(TPP)(1-MeIm)(NO)] have been completed and demonstrate the unique capabilities of NRVS in identifying the character of vibrational modes [53]. A key issue that has been clarified in this study is the assignment of the Fe NO stretching frequency. Resonance Raman studies had assigned this band in sixcoordinate MbNO to be at 556 cm−1 , based on, inter alia, 15 NO isotope substitution. This band is higher in frequency than the 521 cm−1 value assigned for five-coordinate species in which the trans Fe–histidine bond is absent, a somewhat surprising trend. However, NRVS measurements showed that the band at 547 cm−1 (NRVS, 21 K) had a relatively 2 = 011), whereas a band at 443 cm−1 had a small contribution from iron motion (eFe 2 = 025). The analogous band in the resonance larger contribution from iron motion (eFe Raman at 451 cm−1 shows a significant 54 Fe/57 Fe isotope shift, while no significant shift is seen for the 556 cm−1 band. These two sets of observations are consistent with the lower-frequency band (443 cm−1 , NRVS; 451 cm−1 , resonance Raman) being the band with predominant Fe NO stretching character. The NRVS data for oriented crystals of [Fe(TPP(1-MeIm)(NO)] (displayed in Fig. 5) further confirm the vibrational mode character described above. In samples oriented to maximize the out-of-plane contributions of iron (i.e., those perpendicular to the heme plane), only the band at 440 cm−1 shows large signal intensity in the region where the Fe N stretch could occur. The variation (smaller magnitude) in [Fe(TPP)(1-MeIm)(NO)] may reflect differing contributions from FeNO bending. That consideration and any possible correlation between the Fe N and N O stretches (correlation or anticorrelation?) analogous to those in carbonyl heme complexes [54,55] are currently under study with additional six-coordinate nitrosyl complexes. The ideal studies require both powder and oriented single-crystal measurements and should identify additional Fe–ligand vibrations. Another heme species with a coordinated diatomic ligand studied by NRVS is the carbonyl complex [Fe(TPP)(1-MeIm)(CO)]. The initial analysis was made from the powder spectrum and assignment of the modes from a normal coordinate analysis [15]. The observed and normal-coordinate-analysis-predicted spectra are shown in Fig. 9. The most exciting feature of the spectrum was that the long-sought Fe Im stretch has now been identified as expected from the unique capabilities of NRVS. The character from the normal coordinate analysis is illustrated in Fig. 10. While this important vibration had been identified in five-coordinate heme complexes by resonance Raman spectroscopy,

Nuclear Resonance Vibrational Spectroscopy

Vibrational Density of States (1/eV)

400

139

ν49,50,53

calculation NRVS

FeTPP(CO)(1-MeIm) 300

νFe—C

200

δFe—CO νFe-imid 100

0

γ 6,9

0

δFe-imid

200

400

600

Frequency (cm–1)

Fig. 9. NRVS-observed (powder) and predicted (normal coordinate analysis) vibrational density of states for the complex [Fe(TPP)(1-MeIm)(CO)].

νFe — Im, 226 cm–1

Fig. 10. The out-of-plane iron–imidazole stretch mode at 226 cm−1 . The lengths of the arrows indicate relative magnitudes of atom displacements perpendicular to the heme plane.

the Fe Im vibration had not been identified in any six-coordinate heme derivative. An understanding of the variation in this particular vibration as the ligand trans to histidine (imidazole) is changed, and how such variation can otherwise be accomplished has broad implications for understanding function of heme proteins. This is especially pertinent for heme proteins involved in sensing, storing, and transporting diatomic molecules.

140

W.R. Scheidt et al.

The assigned frequency for the Fe Im stretch is 226 cm−1 , perhaps a bit closer than might have been expected to the values found for five-coordinate high-spin iron(II) imidazole complexes (215 cm−1 by NRVS for [Fe(TPP)(2-MeHIm)] [56] and 209 cm−1 by resonance Raman for [Fe(TpivPP)(2-MeHIm)] [57]). Clearly, spectra of additional six-coordinate species, both carbonyl as well as other ligands trans to the imidazole are needed to thoroughly explore the vibrational nature of the Fe Im bond. NRVS measurements on appropriate, oriented single-crystal specimens to thoroughly confirm assignments are needed (see below). Another significant mode involving the imidazole identified from the NRVS data and the normal coordinate calculation is a band at 172 cm−1 that has been assigned as the bending motion of the imidazole. This mode has not been previously identified in resonance Raman or IR spectra and again emphasizes the power of NRVS in identifying all modes with significant iron motion. A number of bands in the range of 250–475 cm−1 are all assigned as in-plane modes for this carbonyl complex. The values of the frequencies form a pattern similar to that observed for [Fe(TPP)(NO)]. As in the VDOS of [Fe(TPP)(NO)], the most prominent feature in the VDOS of [Fe(TPP)(1-MeIm)(CO)] (Fig. 9) is the doublet at 325 and 338 cm−1 . The normal coordinate analysis fit suggests that the degeneracy of these two in-plane modes are lifted as a result of coupling to the Fe C O bending mode, which also leads to the asymmetry in amplitudes. This pair of bands has the largest iron amplitudes of all the in-plane modes. It will be interesting to learn if a DFT study suggests that the modes of this apparent doublet are more complicated than a single near-degenerate pair of in-plane modes, as was found for the nitrosyl. In addition to the Fe Im stretch already mentioned, bands at 506, 561, and 587 cm−1 are associated with the axial Fe C O. The band at 506 cm−1 is assigned as the Fe C stretch and the bands at 561 and 587 cm−1 are assigned as the Fe C O bend doublet. Similar bands are seen in carbonmonoxymyoglobin and -hemoglobin by resonance Raman spectroscopy [58,59]. MbCO has also been studied by NRVS and comparable bands have been observed [21,23]. The spectra of the porphyrin derivative and the myoglobin sample are compared in Fig. 11. A DFT calculation [60] for [Fe(porphine)(HIm)(CO)] predicted bands at 501, 572, and 578 cm−1 . Two bands at low frequency are clearly seen in NRVS data as well as a still lower frequency that must contain acoustic modes as well as a possible molecular mode. The modes near 69–76 cm−1 (calculated at 69, 75, and 76 cm−1 ) and at 124 cm−1 are attributed to out-of-plane iron motion; doming and pyrrole tilting are significantly overlapped. A final important feature of the NRVS spectra of [Fe(TPP)(1-MeIm)(CO)] is that in adjusting the force constants to get good fits at both 124 cm−1 and the cluster of frequencies near the calculated lines at 69, 75, and 76 cm−1 , it was necessary to adjust the interaction between the FeCO bending and Np FeC tilting force constants. The adjustment requires a negative interaction force constant between the tilting and bending modes. The magnitude of this interaction force constant (−0.34 mdyn-Å/rad2 ) is similar to values previously reported (i.e., −0.40 [61] and −0.26 [62] mdyn-Å/rad2 ). Importantly, the necessity of this negative force constant to fit our NRVS data provides confirmation of the suggestion of Ghosh and Bocian [61] that the combination of tilting and bending provides a soft mode for Fe C O distortion. In work in progress, a combination of single-crystal measurements and calculations on heme CO complexes has provided additional insight into one of the more interesting

Nuclear Resonance Vibrational Spectroscopy

141

CO

Fe Vibrational DOS (arb. units)

MbCO

CO FeTPP(CO)(MeIm)

Fe-imid

0

0

200

400

600

Frequency (cm–1)

Fig. 11. Comparison of the NRVS data from the porphyrin derivative [Fe(TPP)(1-MeIm)(CO)] and MbCO. Data for MbCO were reported in reference [21].

features of their vibrational spectra, namely additional information on the nature of the out-of-plane iron frequencies and the Fe–imidazole stretch [63]. Bands in the NRVS spectra at 168–172, 194–209, and 221–224 cm−1 (three distinct complexes) are clearly shown to have iron out-of-plane character; DFT calculations have shown these to have both iron and imidazole motions. None of the three bands has the character of a pure Fe Im stretching mode. Calculations suggest that the two lower-frequency modes involve motions characteristic of Fe Im bending. The third mode is closest in character to the Fe Im stretch, but has a substantially smaller observed e2 Fe than would be expected for a pure Fe Im stretching mode. A mode at 331 cm−1 in [Fe(TPP)(1-MeIm)(CO)] is observed in the single-crystal out-of-plane data; this is unresolvable in the powder data because of the large in-plane modes in that region. In the calculations, the character of this mode is a FeCO translational mode normal to the heme plane in which the imidazole has no appreciable motion. The Fe, C, and O move in unison while the Im is motionless, making this a nontraditional Fe Im stretch (similar to a spring that is fixed at one end). As additional information for six-coordinate hemes on the nature and magnitude of their observed frequencies with Fe Im character becomes available, it should be possible to make correlations on the nature of this bond in terms of chemically noteworthy changes. Such changes must include the variation in the ligand trans to the Fe Im bond, modifications of the character of the imidazole ligand and perturbations on the Fe N distance. Will such correlations be multidimensional, requiring consideration

142

W.R. Scheidt et al.

of both frequency correlations and iron kinetic energy contributions? The future for investigations of these issues remains exciting. Experimental NRVS data agree well with the frequencies, amplitudes, and directions of Fe motion predicted by DFT calculations on [Fe(TPP)(1-MeIm)(CO)] [64]. Intermolecular interactions absent from the gas-phase calculations appear to be the primary factor responsible for a 30-cm−1 discrepancy between observed and predicted Fe CO stretching frequencies but the amplitudes and direction is not significantly affected. Moreover, the calculations reproduce the behavior of the high-frequency FeCO bending modes. This contrasts with simplified empirical three- and four-body models, which have commonly been used to model these vibrations, but which greatly underestimate the mean squared amplitude of the Fe motion. These results further reinforce the proposed interaction between FeCO bending and tilting motion [61,62], and thus indirectly support quantum chemical descriptions of the FeCO distortion energy. NRVS data have also been measured for the high-spin five-coordinate complex [Fe(TPP)(2-MeHIm)]. Spectra have been measured on a powder sample and a large (17 × 07 × 06 mm3 ) single-crystal specimen [56]. Since [Fe(TPP)(2-MeHIm)] is in the triclinic crystal system [65], obtaining measurements with all porphyrin planes either parallel or perpendicular to the incident beam was straightforward. Fig. 12 shows

polycrystalline single crystal, in-plane

Fe(TPP)(2-MeHIm) 400

Density of States (1/eV)

200

0 polycrystalline single crystal, out-of-plane

400

200

0

0

100

200

300

400

Frequency (cm–1)

Fig. 12. Experimental NRVS spectra of the polycrystalline and single crystal [Fe(TPP)(2-MeIm)]. Crystal oriented with heme planes parallel to the incident beam (top panel). Crystal oriented with heme planes perpendicular to the incident beam (bottom panel).

Nuclear Resonance Vibrational Spectroscopy

143

NRVS, in-plane Calculated, in-plane

[Fe(TPP)(2-MeHIm)] 400

Density of States (1/eV)

200

0 300

NRVS, out-of-plane Calculated, out-of-plane

200

100

0

0

100

200

300

400

Frequency (cm–1)

Fig. 13. Experimental and calculated Fe vibrational density of states for (top panel), in-plane (bottom panel), and out-of-plane iron modes. The calculated VDOS have been scaled by an arbitrary scale factor.

comparisons of the experimental powder and single-crystal NRVS spectra. Although overlap of in-plane and out-of-plane modes in the 190–300 cm−1 range is evident, the availability of single-crystal data does provide substantial assistance in assigning the modes. Fig. 13 displays a comparison of the experimental and normal-coordinateanalysis-assigned in-plane and out-of-plane modes. Perhaps the most unexpected feature is the assignment of two distinct Fe Im out-of-plane modes at 215 and 248 cm−1 . The complete description of mode assignments from normal coordinate analysis is given in reference [56]. Additional study on related derivatives is in prospect. The high signal-to-noise level available from polycrystalline powders allowed the identification of vibrational overtones and combinations for both [Fe(TPP) (1-MeIm)(CO)] and [Fe(TPP)(2-MeHIm)] [64]. An analysis confirms the strong dependence of the signal on the relative direction of Fe motion of the component vibrations for vibrational combinations. This suggests an alternative method for obtaining information on directional motion in molecules for which single crystals are not available.

144

W.R. Scheidt et al.

SUMMARY Although much work remains to be done, this chapter shows that the vibrational spectroscopy method NRVS is an important new tool to investigate heme systems. It clearly provides a singularly sensitive way of examining all modes with iron motion in a molecule of any complexity. NRVS makes accessible vibrational information not easily attainable by other methods. Although the assignment of the modes observed in the NRVS experiment is a substantial challenge, perhaps much more difficult than might have been anticipated, further measurements are expected to provide a general basis for assignments. NRVS experiments provide a unique avenue into studying the effects of varying protein environments on the low-frequency modes and the possible effects on reactive modes.

ACKNOWLEDGMENTS WRS thanks the National Institutes of Health for the long-term support of his porphyrin research under Grant GM-38401-33. JTS acknowledges grants from NIH GM-52002 and NSF PHY-0545787. SMD acknowledges support by NSF PHY-9983180. Use of the Advanced Photon Source was supported by the U.S. Department of Energy, Basic Energy Sciences, Office of Science, under Contract No. W-31-109-Eng-38. We thank the staff of the Sector 3-ID-D beamline for their unstinting efforts in assisting us in making measurements.

REFERENCES [1] Spiro, T.G. (Ed.) (1988) Biological Applications of Raman Spectroscopy, New York: WileyInterscience,. [2] Dong, A. and Caughey, W.S. (1994) Methods Enzymology 232, 139–175. [3] Barth, A. and Zscherp, C. (2002) Q. Rev. Biophys. 35, 369–430. [4] Zhu, L., Sage, J.T. and Champion, P.M. (1994) Science 226, 629–632. [5] Rosca, F., Kumar, A., Ionascu, D. et al. (2001) J. Phys. Chem. 114, 10884–10898. [6] Seto, M., Yoda, Y., Kikuta, S. et al. (1995) Phys. Rev. Lett. 74, 3828–3831. [7] Sturhahn, W., Toellner, T.S., Alp, E.E. et al., (1995) Phys. Rev. Lett. 74, 3832–3835. [8] Keppler, C., Achterhold, K., Ostermann, A. et al., Eur. Biophys. J. 25, 221–224. [9] Paulsen, H., Winkler, H., Trautwein, A.X. et al. (1999) Phys. Rev. B 59, 975–984. [10] Paulsen, H. Benda, R. Herta, C. et al. (2001) Phys. Rev. Lett. 86, 1351–1354. [11] Mao, H.K., Xu, J., Struzhkin, V.V. et al. (2001) Science 292, 914–916. [12] Paulsen, H., Rusanov, V., Benda, R. et al. (2002) J. Am. Chem. Soc. 124, 3007–3011. [13] Parak, F. and Achterhold, K. (1999) Hyp. Int. 123, 825–840. [14] Bergmann, U., Sturhahn, W., Linn, D.E., Jr. et al. (2003) J. Am. Chem. Soc. 125, 4016–4017. [15] Rai, B.K., Durbin, S.M., Prohofsky, E.W. et al. (2003) J. Am. Chem. Soc. 125, 6927–6936. [16] Achterhold, K. and Parak, F.G. (2003) J. Phys. Condens. Matter 15, S1683–S1692. [17] Budarz, T.E., Prohofsky, E.W., Durbin, S.M. et al. (2003) J. Phys. Chem. 107, 11170–11177. [18] Sage, J.T., Paxson, C., Wyllie, G.R.A. et al. (2001) J. Phys: Condens. Matter 13, 7707–7722. [19] Scheidt, W.R., Piciulo, P.L. (1976) J. Am. Chem. Soc. 98, 1913–1919. [20] Wyllie, G.R.A., Schulz, C.E. and Scheidt, W.R. (2003) Inorg. Chem. 42, 5722–5734. [21] Sage, J.T., Durbin, S.M., Sturhahn, W. et al. (2001) Phys. Rev. Lett. 86, 4966–4969.

Nuclear Resonance Vibrational Spectroscopy [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58]

[59] [60]

145

Rai, B.K., Durbin, S.M., Prohofsky, E.W. et al. (2002) Biophys. J. 82, 2951–2963. Achterhold, K., Keppler, C., Ostermann, A. et al. (2002) Phys. Rev. E 65, 051916. Oganesyan, V.S., Barclay, J.E. Hardy, S.M. et al. (2004) Chem. Commun. 215–216. Smith, M.C., Xiao, Y., Wang, H. et al. (2005) Inorg. Chem. 44, 5562–5570. Sage, J.T., Barabanschikov, A., Barbiellini, B. et al., in preparation. Mössbauer, R.L. (1958) Z. Phys. 151, 124–143. Visscher, W.M. (1960) Ann. Phys. NY 9, 194–210. Singwi, K.S. and Sjölander A. (1960) Phys. Rev. 120, 1093–1102. Weiss, H. and Langhoff, H. (1979) Phys. Lett. A 69, 448–450. Toellner, T.S., Hu, M.Y., Sturhahn, W. et al. (1997) Appl. Phys. Lett. 71, 2112–2115. Toellner, T.S. (2000) Hyperfine Interact. 125, 3–28. Baron, A.Q.R. (1995) Nuclear Instruments and Methods in Physics Research A 352, 665–667. Sturhahn, W., Toellner, T.S., Quast, K.W. et al. (1996) Nucl. Instrum. Methods A 372, 455–457. Lipkin, H.J. (1962) Ann. Phys. (NY) 18, 182–197. Sturhahn, W. (2000) Hyp. Int. 125, 149–172. Abe, M. Kitagawa, T. and Kyogoku, Y. (1978) J. Chem. Phys. 69, 4526–4534. Li, X.-Y., Czernuszewicz, R.S., Kincaid, J.R. and Spiro, T.G. (1989) J. Am. Chem. Soc. 111, 7012–7023. Rush, T.S., III, Kozlowski, P.M., Piffat, C.A. et al. (2000) J. Phys. Chem. B. 104, 5020–5034. Levin, I.W. and Pearce, R.A.R. (1975) In Vibrational Spectra and Structure, Chap. 4. (J.R. Durig, ed.) Amsterdam: Elsevier, pp. 102–186. The empirically derived normal coordinate analysis fit is illustrated in Figure 3 of reference 21. Spiro, T.G. and Li, X.-Y. In Biological Applications of Raman Spectroscopy, New York: Wiley-Interscience, pp. 1–37. Procyk, A.D. and Bocian, D.F. (1992) Annu. Rev. Phys. Chem. 43, 465–496. Klug, D.D., Zgierski, M.Z., Tse, J.S. et al. (2002) Proc. Natl. Acad. Sci. U.S.A. 99, 12526–12530. Leu, B., Zgierski, M., Wyllie, G.R.A. et al. (2004) J. Am. Chem. Soc. 126, 4211–4227. Scheidt, W.R., Duval, H.F., Neal, T.J. and Ellison, M.K. (2000) J. Am. Chem. Soc. 122, 4651–4659. Wyllie, G.R.A. Schulz, C.E. and Scheidt, W.R. (2003) Inorg. Chem. 42, 5722–5734. Kozlowski, P.M., Spiro, T.G., Bérces, A. and Zgierski, M.Z. (1998) J. Phys. Chem. B 102, 2603–2608. Perutz, M.F. (1970) Nature 228, 726–739. Srajer, V., Reinisch, L. and Champion, P.M. (1988) J. Am. Chem. Soc. 110, 6656–6670. Perutz, M.F., Wilkinson, A.J., Paoli, M. and Dodson, G.G. (1998) Annu. Rev. Biophys. Biomol. Struct. 27, 1–34. Hoard, J.L. and Scheidt, W.R. (1973) Proc. Natl. Acad. Sci. U.S.A. 70, 3919–3922. Zeng, W. Silvernail, N.J. Wharton, D.C. et al. (2005) J. Am. Chem. Soc. 127, 11200–11201. Ray, G.B., Li, X.-Y., Ibers, J.A. et al. (1994) J. Am. Chem. Soc. 116, 162–176. Silvernail, N.J., Roth, A., Schulz, C.E. et al. (2005) J. Am. Chem. Soc. 127, 14422–14433. Rai, B.K. Durbin, S.M. Prohofsky, E.W. et al. (2002) Phys. Rev. E 66, 051904-1–12. Hori, H. and Kitagawa, T. (1980) J. Am. Chem. Soc. 102, 3608–3613. Kincaid, J.R. (2000) The Porphyrin Handbook, In Resonance Raman Spectra of Heme Proteins and Model Compounds, Vol. 7 (K.M. Kadish, K. Smith and R. Guilard,) New York: Academic Press. Unno, M., Christian, J.F. Olson, J.S. et al. (1998) J. Am. Chem. Soc. 120, 2670–2671. Kozlowski, P.M., Vogel, K.M. Zgierski M.Z. and Spiro, T.G. (2001) J. Porph. Phthal. 5, 312–322.

146

W.R. Scheidt et al.

[61] [62] [63] [64] [65]

Ghosh, A. and Bocian, D.F. (1996) J. Phys. Chem. 100, 6363–6367. Spiro, T.G. and Kozlowski, P.M. (1998) J. Am. Chem. Soc. 120, 4524–4525. Leu, B.M., Silvernail, N.J., Zgierski, M.Z. et al. (2007) J. Biophys. 92, 3764–3783. Leu, B.M., Zgierski, M.Z., Wyllie, G.R.A. et al. (2005) J. Phys. Chem. Solids 66, 2250–2256. Ellison, M.K., Schulz, C.E. and Scheidt, W.R. (2002) Inorg. Chem. 41, 2173–2181.

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 6

EPR and Low-temperature MCD Spectroscopy of Ferrous Heme Nitrosyls Nicolai Lehnert Department of Chemistry, The University of Michigan, 930 N. University, Ann Arbor, Michigan 48109-1055, USA

1. INTRODUCTION Ferrous heme NO adducts are of enormous biological significance and are involved in many of the known physiological functions of NO, which include nerve signal transduction, vasodilation, blood clotting, and immune response by white blood cells [1–8]. New biological functions of NO and of the corresponding oxidized and reduced NOx species are still being discovered [9–20]. NO is produced in vivo by the nitric oxide synthase (NOS) family of enzymes [21–24]. The cardiovascular regulation is then mediated by soluble-guanylate cyclase (sGC), which is activated for production of the messenger cGMP by coordination of NO to its ferrous heme active site [12,25–28]. Besides these physiological functions, NO also occurs in other biological processes, for example as an intermediate in denitrification [29–31]. Historically, nitric oxide has been widely used as a “spin-label probe” to study the interaction of ferrous heme centers with dioxygen [32–41]. Among the different hemes investigated, the NO adducts of hemoglobin and myoglobin are probably some of the most studied systems in biology [42]. Recently, the NO chemistry of myoglobin has been revived, since evidence has been presented that oxymyoglobin efficiently catalyzes the oxidation of nitric oxide to nitrate, and this process could be involved in the regulation of the concentration of the neurotransmitter NO in vivo [42]. Since the ferrous heme NO adducts are paramagnetic with an S = 1/2 ground state, EPR has been the key technique for the investigation of these species as evident from the fact that many review articles and even whole books have been dedicated to this topic (see, for example, reference [43]). Stimulated by this immense interest in ferrous heme NO adducts, much research has also been directed toward the synthesis and EPR spectroscopic characterization of either five-coordinate (5C) model compounds [Fe(porphyrin)(NO)] or corresponding six-coordinate (6C) complexes with bound N-donor ligands L (L = 1-methylimidazole (MI), pyridine (Py), piperidine (Pip), etc.) resembling the observed biological species [44–47]. Most of these compounds use the synthetic ligands tetraphenylporphyrin (TPP), octaethylporphyrin (OEP) and corresponding derivatives, and protoporphyrin IX diester (PPDE), which are shown

148

Nicolai Lehnert

m N

α

H N

N β

N

N N

H N

N

N

ROOC

Tetraphenylporphyrin (TPP)

N H

H

H N

N H

Octaethylporphyrin (OEP)

COOR

Protoporphyrin IX Diester (PPDE)

Scheme 1.

in Scheme 1. A large body of EPR data is available for corresponding ferrous nitrosyls with these ligands [48–53]. Much less research has been directed toward the corresponding six-coordinate complexes with trans S-donor ligands. Nevertheless, these systems are of great biological importance. Ferrous heme NO complexes with axial cysteinate coordination have been characterized by EPR for a number of P450-type enzymes [38,54]. A few model complex studies are also available [55–57]. Single-crystal EPR data have been obtained for the 5C model complex [Fe(OEP)(NO)] [58] and for the Fe(II) NO adducts of hemoglobin and myoglobin, which are 6C with axial histidine ligation [59–61]. These data allow to experimentally establish the orientation of the g-tensor in the molecular frame, and to determine the complete hyperfine tensor A of the nitrogen of the bound NO. Importantly, the EPR data on proteins and model complexes indicate interesting differences in the electronic structures of the corresponding 5C and 6C complexes [62], the nature of which has only been elucidated recently in detail using magnetic circular dichroism (MCD), vibrational spectroscopy, and quantum chemical calculations [63,64]. Density functional theory (DFT) calculations on simplified model systems [Fe(P)(L)(NO)] (P = Porphine) have also successfully been applied to calculate the EPR spectroscopic parameters of ferrous heme nitrosyls [63–66]. From these results, a quantitative understanding of the EPR spectra of ferrous heme nitrosyls and their electronic structures has now evolved. Magnetic circular dichroism (MCD) spectroscopy is a very powerful tool for the investigation of the ground and excited states of paramagnetic compounds [67–72]. Therefore, this technique has also been widely applied to heme proteins and corresponding model complexes [73–75]. However, the application of this method to paramagnetic ferrous heme nitrosyl complexes is very limited, and until recently, only room temperature spectra of protein NO adducts and model complexes have been published (see, for example, reference [76,77]). However, information about the electronic structures of these complexes is mostly available from the low-temperature spectra, where the temperature-dependent C-term spectrum representing the paramagnetic contribution to the total intensity can be determined. Recently, corresponding spectra for 5C and 6C heme nitrosyl model complexes have been published, which shed some light on the differences in ground states of these complexes.

EPR and MCD Spectroscopy of Ferrous Heme Nitrosyls

149

2. GENERAL CONSIDERATIONS Ferrous heme nitrosyls with the general formula [Fe(porphyrin)(L)(NO)] in proteins and model complexes all show low-spin (S = 1/2) ground states regardless of whether the iron is 5C or 6C, and independent of the nature of the axial ligand L trans to NO (L = N-donor, thiolate, etc.). A number of crystal structures have been determined for 5C model complexes, and corresponding 6C species with axial N-donor ligands bound [45]. On the basis of these structural data, Scheidt and coworkers have defined the general structural motifs of these compounds. As shown in Fig. 1, these complexes exhibit short Fe NO bonds between 1.73 and 1.75 Å and Fe N O angles of ∼140 [62]. A strongly simplified representation of the electronic structures of these complexes is shown in Scheme 2. Since iron is in the +II oxidation state and low-spin, it has a dxz  dyz  dx2 −y2 6 = t2 6 electron configuration in the given coordinate system shown in Scheme 2, top. Here, the z-axis is oriented approximately along the Fe N(O) bond, and x and y are located in the porphyrin plane such that y is perpendicular to the Fe N O unit. Free nitric oxide is a radical with S = 1/2 ground state, with the unpaired electron being located in the (singly occupied)  ∗ orbitals. Since the t2 orbitals of iron are fully occupied, donation from this  ∗ orbital of NO is only possible into the dz2 orbital of iron. This leads to a partial occupation of the dz2 orbital depending on the strength of this interaction. Besides this -donor bond,  backdonation from the occupied dxz and dyz orbitals of iron into the empty  ∗ orbitals of NO is also present. Using a restricted open shell approach, the simplified MO diagram shown in Scheme 2 is obtained.a On the basis of this electronic structure description, it can be expected that the EPR spectra of ferrous heme NO adducts show similar g-values as low-spin d7 complexes with the electron configuration dxz  dyz  dx2 −y2 6 dz2 1 , as observed, for example, for 1.168 Å O

N

1.174 Å O

N

143.4°

1.728 Å

long

Np Np

138.5°

1.753 Å

Fe 0.27 Å

short

long

Np Np

Fe

short

Np 0.05 Å

Np

Np Np

L

Fig. 1. General structural motifs of five-coordinate [Fe(porphyrin)(NO)] and six-coordinate [Fe(porphyrin)(L)(NO)] (L = N-donor ligand) complexes (Reprinted with permission from reference [62]. Copyright 2003 American Chemical Society).

Note that a1u , a2u , and eg refer to porphyrin / ∗ orbitals, whereas v∗ (v = vertical) and h∗ (h = horizontal) are the two  ∗ orbitals of NO located within (h) and perpendicular to (v) the Fe N O plane.

a

150

Nicolai Lehnert

z y

x

Energy

dz2_πh∗ dxy_b1g π v∗_dyz LUMO

eg

π h∗ _dz2 dx2–y2 a2u a1u dxz_π h∗ dyz_π v∗

σ

π

Scheme 2.

Co(II)-porphyrin complexes. For an S = 1/2 system with a nondegenerate ground state, perturbation theory yields the following expression for the g-values [78]: gi = ge − 2

 0 Li n  n Li 0  n=0

En − E 0

i = x y z

Here, ge is the Landé factor of the free electron, 0 is the ground state, and the sum runs over all excited states n (n = 0). Applying a pure ligand field model, the following states need to be considered in a first-order treatment (cf. Scheme 2):    2 2 2 dxz dx2 −y2 dz12 0  =  passive shells dyz    2 2 1 dxz dx2 −y2 dz22 1  =  passive shells dyz    2 1 2 dxz dx2 −y2 dz22 2  =  passive shells dyz    1 2 2 dxz dx2 −y2 dz22 3  =  passive shells dyz Using these ligand field states, but including orbital reduction factors for covalency, the g-values for the d7 electron configuration are (cf. also reference [79,80]): gz = ge = 2002 gx = g e − gy = g e −

6 2 2 Eyz→z2 2

2

6  Exz→z2

(1)

EPR and MCD Spectroscopy of Ferrous Heme Nitrosyls

151

Here,  is the spin-orbit coupling constant for iron(II) (0 ≈ − 390 cm−1 )a , and , ,  are the coefficients of the dz2 , dyz , and dxz functions, respectively, in the corresponding molecular orbitals. The denominator E corresponds to the configurational excitation energy from dyz or dxz to dz2 , respectively. Since the unpaired electron occupies a pure dz2 orbital in this model, there are no additional contributions to gz since Lz dz2 = 0. In addition, the excited state 1 > does not contribute. On the basis of Eq. (1), it can be expected that the EPR spectra of ferrous nitrosyls are centered around g = 200, and since dyz or dxz are not degenerate due to their anisotropic interaction with NO, the EPR spectra of these species should in general be rhombic with gx = gy > gz . The g-shifts for gx and gy should be small and positive and hence, all g-values should be larger or equal to ge . Since the unpaired electron density on iron in ferrous heme nitrosyls resides mostly in the dz2 orbital, this ligand field model also suggests that well resolved hyperfine lines of the coordinated NO should occur on gz .

3. EPR SPECTRA OF FIVE- AND SIX-COORDINATE Fe(II)–PORPHYRIN NO ADDUCTS Figure 2 shows typical EPR spectra of 5C ferrous heme nitrosyls. On the left, the spectrum of the hemoglobin-NO adduct in the presence of inositol hexaphosphate is shown [81], whereas on the right the data for the model complex [Fe(TPP)(NO)] are presented [57]. Typical g-values in these cases are about 2.1, 2.06, and 2.01 for gmax , gmid , and gmin , respectively, as listed in Table 1. The observed g-shifts >0 are in agreement with the ligand field model presented above. Here, gmin then corresponds to gz . From single-crystal EPR experiments on [Fe(OEP)NO)], it has also been deduced that gmax is equivalent to gy and gmid corresponds to gx for this complex [58], which probably holds

2.106

0

dχ“/dB

2.066

[Fe(TPP)(NO)]

Hb-NO/IHP 3100

3200

2.013

3300

3400

3000 3100 3200 3300 3400 3500 3600 3700 3800

H (GAUSS)

B[Gauss]

Fig. 2. EPR spectra of five-coordinate ferrous heme nitrosyls. Left: Hb NO in the presence of inositol hexaphosphate, taken from ref. [81]. Right: model complex [Fe(TPP)(NO], taken from reference [57].

a

< 0 for a more than half filled d-electron configuration.

152

Table 1. Comparison of experimental and calculated g-values and g-tensor orientations for [Fe(porphyrin)(L)(NO)]n− (n = 0 1) centers Moleculea

g-values g(mid)

g(min)

2102 2106 2106 2105 2070

2064 2066 2057 2059 204

2010∗ 2013∗ 2015∗ 2010∗ 2008∗

[Fe(TPP)(Pip)(NO)] [Fe(PPDE)(MI)(NO)] [Fe(PPDE)(Pip)(NO)] [Fe(PPDE)(Py)(NO)] Mb NO Hb NO Hb NO

2079 2074 208 2074 2078 2076 2076 2060 2082

2004∗ 2005∗ 2003∗ 2004∗ 2004∗ 2004∗ 2002∗ 2000∗ 2000∗

[Fe(TPP)(SPh)(NO)]− [Fe(TPP)(H4 Tp)(NO)] [Fe(PPDE)(EtSH)(NO)]

2108 2104 2096

2068 2054 205

[Fe(TPP)(NO)] [Fe(OEP)(NO)] [Fe(PPDE)(NO)] Hb NO + inositol–P6 [Fe(TPP)(MI)(NO)]

1972 1978 <200 1971 197 1978 1979 1965 1978 2013∗ 2010∗ 2011∗

Conditions (solvent, etc.)

Reference

g(min)/Fe NO: 8

Toluene Toluene Single crystal Toluene pH = 7

[48] [57] [58] [52] [81]

CHCl3 Toluene Toluene CHCl3 CHCl3 Acetone Single crystal Single crystal Single crystal

[39] [57] [48] [51] [51] [51] [60] [59] [59]

CHCl3 Toluene Acetone

[57] [57] [55]

g(mid)/heme⊥: 30 g(mid)/heme⊥: 10 g(mid)/heme⊥: 8

Nicolai Lehnert

g(max)

Orientationb

2094 2092 2082 2073 207

2049 2047 2007∗ 2009∗ 200∗

[Fe(P)(NO)]d [Fe(P)(IM)(NO)]d [Fe(P)(NO)]e [Fe(P)(IM)(NO)]e [Fe(P)(SCH3 )(NO)]−f [Fe(P)(SPh)(NO)]−f

2049 2024 2063 2034 2019 2026

2025 1991 ∗

2005 ∗

1995 ∗

1994 1998

2010∗ 2009∗ 1970 1976 <200 2004 ∗

1955 1994 1955 1974 ∗

1972 ∗

g(min)/Fe g(mid)/Fe g(mid)/Fe g(mid)/Fe g(min)/Fe g(min)/Fe

NO: 20 NO: 29 NO: 20 NO: 32 NO: 26 NO: 36

Acetone Acetone Buffer Buffer Toluene

[55] [55] [54] [38] [56]

From DFT calculations

[63] [63] [65] [65] [57] [57]

EPR and MCD Spectroscopy of Ferrous Heme Nitrosyls

[Fe(PPDE)(H4 Tp)(NO)] [Fe(PPDE)(Me2 S)(NO)] P450nor: Fe(II) NO P450cam: Fe(II) NO [Fe(SR)(NO)]−c

P = Porphine2− ligand used for the calculations (cf. Scheme 2); ∗ = shows well resolved 14 N hyperfine lines of the coordinated NO ligand; (∗ ) = principal axis in the calculations closest to the Fe N(O) bond. b Relative orientation of the principal axis of the given g-value with respect to the Fe N(O) bond or the heme normal (heme⊥). c Model complex with axial benzylthiolate coordination. d Calculated with B3LYP and the following basis sets: Fe: CP(PPP); N (EPR-II); C, O: TZVP; H: TZV. e˙ Calculated with VWN/triple- basis set of Slater-type functions. f Calculated with BP86/TZVP. a

153

154

Nicolai Lehnert

true for all 5C ferrous heme nitrosyls. This means that the principal axis of gmax = gy is perpendicular to the Fe N O plane (vide supra). Importantly, the smallest g-value shows very strong hyperfine lines of the 14 N of the coordinated NO ligand. Since 14 N has a nuclear spin of I = 1, this leads to the appearance of three lines in the spectrum. The fact that these hyperfine lines belong to NO and not the porphyrin nitrogens is proven by using 15 N-labeled NO, which, due to the nuclear spin of I = 1/2 for this isotope, leads to the appearance of only two hyperfine lines in the spectrum [53]. The observed hyperfine couplings for gmin are in the range of 40–50 MHz, as shown in Table 2. The hyperfine tensor itself is actually quite isotropic with all three coupling constants occurring in this range. In summary, these results are in very good agreement with the ligand field model presented above, which indicates that in the case of 5C ferrous heme nitrosyls, a substantial electronic population of the dz2 orbital is observed. This is due to strong mixing of dz2 with the singly occupied  ∗ orbital of NO, as has been proposed in several studies (see, for example, reference [48,58,65,82]).

3.1. Effect of Binding an axial N-donor Ligand Figures 3 and 4 present the EPR spectra of 6C ferrous heme nitrosyls. In Fig. 3 (left), the spectra of horseradish peroxidase NO adducts with different isotopes are shown [34]. Figure 4 presents the EPR data of the model complex [Fe(TPP)(MI)(NO)] [57] that is obtained upon addition of excess N-donor ligand to solutions of [Fe(TPP)(NO)]. As evident by comparison with Fig. 2, these data show large differences compared to the 5C case. The g-shifts are in general smaller for the 6C compounds, and are observed at about 2.08, 2.00, and 1.98 for gmax , gmid , and gmin , respectively, as presented in Table 1. An additional g-value, often designated as g? (cf. Fig. 4), is usually present in 6C protein and model complex species. This feature has been investigated in some detail in the literature, and is believed to be related to a second (less populated) conformation of the complexes [39–41,49,65]. Since no corresponding feature is observed in the 5C case, this signal probably relates to two different orientations of NO relative to the trans N-donor ligand. In this respect, it is interesting to note that crystal structures of corresponding model complexes with bound imidazole ligands frequently show two different orientations of NO relatively to the molecular frame [62,64]. From single-crystal EPR and ENDOR experiments, different explanations for the two conformations have also been offered. In the original publication by Morse and Chan, it was proposed that a topological isomer exists where the iron is located below the porphyrin plane toward the imidazole ligand [49]. Other explanations for the second species consider a linear Fe NO unit [60], or a conformation with the ON Fe N(Im) axis exactly orthogonal to the porphyrin plane [39,40]. Another striking difference between the EPR spectra of the 5C and 6C complexes relates to the observed hyperfine splittings. Compared to the 5C case, the well resolved hyperfine lines of the coordinated NO are now observed on the medium g-value gmid . High-resolution data identify a total of nine hyperfine lines as shown in Fig. 3 (left), which is due to the fact that both the 14 N nitrogen of NO and of the trans N-donor ligand contribute as shown in Fig. 3 (right). The origin of the hyperfine lines can be proven by using different isotopes, which lead to different coupling patterns as demonstrated for the horseradish peroxidase NO adducts in Fig. 3. The observed hyperfine coupling of 14 NO is about 60 MHz and, hence, distinctively stronger than in the 5C complexes.

14

Molecule

A[g(max)] [Fe(TPP)(NO)] [Fe(OEP)(NO)] [Fe(PPDE)(NO)] [Fe(TPP)(Pip)(NO)] [Fe(PPDE)(MI)(NO)] [Fe(PPDE)(Pip)(NO)] [Fe(PPDE)(Py)(NO)] Hb NO Hb NO

371 409 383

442

[Fe(PPDE)(EtSH)(NO)] [Fe(PPDE)(H4 Tp)(NO)] [Fe(PPDE)(Me2 S)(NO)] P450nor–NO P450cam–NO [Fe(P)(NO)]d [Fe(P)(MI)(NO)]d [Fe(P)(MI)(NO)]e

N Hyperfine A[g] [MHz]a A[g(mid)] 49.7 49.7 40.3 60.8 60.3 58.9 60.0 29.6/32.9/63.6c 62.3 43.0 40.1 56.2 54.0

283 152

62.0 84.0 29.0/31.5/73.9

Orientationb

Conditions (solvent, etc.)

Reference

g/A: 30

Toluene Single crystal Toluene

[48] [58] [52]

Toluene CHCl3 CHCl3 Acetone Single crystal Single crystal

[48] [51] [51] [51] [59] [59]

Acetone Acetone Acetone Buffer Buffer

[55] [55] [55] [54] [38]

From DFT calculations

[64] [64] [66]

A[g(min)] 487 427 464

269

g(mid)/A(max): 37 g(mid)/A(max): 43

512 551 554

499 132

g(min)/A(mid): 20 g(mid)/A(max): 30 Not provided

For the solution data, the A-values in MHz have been calculated with the formula: A[MHz] = 1.39916 · Ai [G] · gi . Strictly speaking, this formula is only valid if the g- and A-tensors are aligned. Hence, these values have to be treated with some caution. b Angle between the principal axes of g- and A-tensor components as indicated. c Since the A- and g-tensors show large differences in their relative orientation, no classification of A with respect to g is useful. d Calculated with B3LYP and the following basis sets: Fe: CP(PPP); N (EPR-II); C, O: TZVP; H: TZV. e Calculated with BPW91 and the following basis sets: Fe: large Wachter’s basis; C, O, N: 6-311G∗ ; H: 6-31G∗ . In the model system, an Fe N(NO) bond length of 1.74 Å and an Fe N O angle of 145 were used.

EPR and MCD Spectroscopy of Ferrous Heme Nitrosyls

Table 2. Comparison of experimental and calculated 14 N hyperfine (A) tensors of the nitrosyl nitrogen for [Fe(porphyrin)(L)(NO)]n− (n = 0 1) centers. The values of the A-tensor [MHz] are given relative to the g-tensor, i.e., a given A-value is paired with the g-value whose principal axis is closest to the principal axis of this A

a

155

156

Nicolai Lehnert gx = 2.08 gz = 2.004

15

gy = 1.95

NO Fe

15

14

HRP– NO

N

(A) 14

NO

Fe

57

HRP–15NO

(B)

14

N

HRP–14NO

15

NO

(C)

57

Fe

14

N

57

HRP–14NO

(D)

14

NO

57

Fe

14

N

3.05

3.10

3.15 3.20 Ho (k Oe)

3.25

3.30

Fig. 3. Left: EPR spectra of six-coordinate horseradish peroxidase ferrous NO adducts using different isotopes. Right: coupling scheme explaining the observed hyperfine splittings of the medium g-value. Taken from reference [34].

g?

dχ“/dB

2.074

0 2.005

1.978

[Fe(TPP)(MI)(NO)] 3000

3100

3200

3300

3400

3500

3600

3700

3800

B[Gauss]

Fig. 4. EPR spectrum of the six-coordinate model complex [Fe(TPP)(MI)(NO)]. Adapted from reference [57].

EPR and MCD Spectroscopy of Ferrous Heme Nitrosyls

157

In comparison, the coupling constant for the 14 N of the N-donor ligand is 1 order of magnitude smaller [48]. In addition, single-crystal EPR data for hemoglobin-NO show that the hyperfine tensor is much more anisotropic in the 6C case as presented in Table 2 [59]. These distinct differences between the 5C and 6C complexes are surprising. From a comparison of the structural motifs of these complexes as presented in Fig. 1, and based on the observed small binding constants of the N-donor ligands, one would anticipate that the electronic structures of the 5C and 6C compounds are similar. However, the large differences in the EPR spectra demonstrate that this is not the case and that these complexes differ in the nature of their ground states [62].

3.2. Electronic Structural Differences of Five- and Six-coordinate Ferrous Heme Nitrosyls In recent studies, the differences in the electronic structures of 5C versus 6C complexes as described above have been analyzed in detail using MCD, NMR, and vibrational spectroscopies coupled to density functional (DFT) calculations [63,64]. Importantly, the obtained MCD spectra (vide infra) indicate that upon binding of an axial N-donor ligand to a 5C [Fe(porphyrin)(NO)] complex, the spin density on iron is greatly reduced. In addition, comparison of the N O and Fe NO force constants of these complexes available from detailed vibrational analysis shows that the coordination of the axial N-donor weakens the Fe NO -bond. These experimental results can be understood using density functional calculations. Figure 5 (top) shows the singly occupied molecular [Fe(P)(MI)(NO)] [Fe(P)(NO)]

+0.78

+0.53

SOMO:

+0.47

SOMO: πh∗_dz2 /dxz

+0.21

dz2_πh∗

g(min) α

g(min) α γ

β

g(mid)

β g(mid)

γ g(max)

g(max)

N(imidazole)

Fig. 5. Top: singly occupied molecular orbitals (SOMOs) of five- and six-coordinate ferrous heme nitrosyls. Bottom: calculated (B3LYP) g-tensor orientations. Adapted from reference [64] (Reprinted with permission from ref. [64]. Copyright 2006 American Chemical Society).

158

Nicolai Lehnert

orbitals (SOMOs, cf. Scheme 2) of the corresponding model systems [Fe(P)(NO)] and [Fe(P)(MI)(NO)] (P = porphine2− ) together with calculated spin populations [64]. In the 5C case, strong mixing between the singly occupied  ∗ orbital of NO and a pure dz2 of iron occurs, leading to a full delocalization of the unpaired electron of NO over the Fe N O subunit. Correspondingly, the calculated spin populations are about +0.5 on iron and +0.5 on NO. Upon coordination of the axial N-donor ligand, the spin density is pushed back from the iron toward the NO ligand, leading to spin populations of about +0.8 on NO and only +0.2 on iron in agreement with the MCD result (see Section 4). Correspondingly, mixing of the singly occupied  ∗ orbital of NO with dz2 of iron is reduced, which weakens the Fe NO -bond as reflected by the vibrational force constants. The SOMO of the 6C complex also shows an admixture of a -donor orbital of MI, which is antibonding with respect to iron. This antibonding interaction explains why the N-donor has such a profound effect on the electronic structure of the Fe N O subunit, but only shows weak binding. On the basis of these descriptions, the 6C complex corresponds to the prototype of an Fe(II) NO(radical) adduct, whereas the 5C compound has noticeable Fe(I) NO+ character relative to the 6C case [63]. Having elaborated these differences in the electronic structures of the complexes, it is now possible to explore how this might relate to the different appearance of their EPR spectra. First, the decrease of spin density on iron in the 6C complex is in agreement with the observed smaller g-shifts in this case. This is due to the fact that a decrease of unpaired electron character in dz2 reduces the magnitude of spin-orbit coupling matrix elements, which causes smaller deviations from the free-electron g-value, as has been noted before [65]. The occurrence of the strong hyperfine splittings on gmin in 5C compared to gmid in 6C is due to a rotation of the g-tensor in these complexes with respect to the molecular frame [63,65]. The calculated g-tensor orientations are shown in Fig. 5 (bottom) [63,64]. In the 5C complex, the principal axis of the smallest g-value gmin is closest to the Fe NO axis (deviation: 20 ), which is also in agreement with the single-crystal EPR data on [Fe(OEP)(NO)] [58], where the angle between the gmin principal axis and the Fe NO bond has been determined to be 8 (cf. Table 1). Correspondingly, gmin shows well resolved hyperfine splittings. Upon coordination of the sixth ligand, the g-tensor rotates such that the principal axis of the medium g-value gmid is now closest to the Fe NO axis. As shown in Fig. 5, the calculated deviation is 29 . This compares well with the experimentally determined angle between the gmid principal axis and the Fe NO bond for myoglobin-NO of ∼30 [60,61]. Hence, the strong hyperfine lines of the coordinated NO are then observed on gmid in the 6C complex. Since the singly occupied MO in this case also contains a contribution from the imidazole nitrogen, additional small hyperfine splittings due to the occurrence of a small amount of unpaired spin density on this N atom are also observed as illustrated in Fig. 3 (right). In summary, the calculated g-shifts (g) from DFT are obtained somewhat too small for both the 5C and the 6C complexes as shown in Table 1, but the obtained g-tensor orientations show very good agreement with experiment. The rotation of the g-tensor in the 5C and 6C complexes is probably related to the different “magnetic orbitals” of iron in these systems. In the 5C case, the spin density on iron is located in an almost pure dz2 orbital that is oriented along the z-axis. On coordination of the sixth ligand, the orbital becomes mixed with dxz and hence, is rotated off the z-axis (cf. Fig. 5, top). Presumably, the g-tensor then follows the rotation of the spin density at the iron center. As an alternative to this phenomenological interpretation of the change in hyperfine pattern, the ligand field model presented above

EPR and MCD Spectroscopy of Ferrous Heme Nitrosyls

159

can be used to analyze the changes. Note that the model in Eq. (1) predicts that all g-values are larger than ge , which is violated in the 6C case. Hence, the applicability of this model might be limited. As described above, the single-crystal EPR data indicate that in the 5C case, gz corresponds to the minimum g-value gmin (∼2.01) [58]. In the 6C case, however, gz is equal to gmid at ∼2.00, whereas gy corresponds to gmax (∼2.08) and, importantly, gx is shifted below ge to approximately ∼1.98, becoming gmin in the 6C complexes. This interpretation leads to the conclusion that the largest change actually occurs for gx , which shows a shift from ∼2.06 (5C) to ∼1.98 (6C). This observation cannot be explained on the basis of the simple “dz2 model” in Eq. (1). In a classic study, Ziegler and coworker have analyzed the individual contributions to the g-shifts (g) in 5C and 6C ferrous heme nitrosyls using DFT calculations [65]. These results show that the g-shifts for gx and gz in general arise from a sensitive numerical balance of a number of small contributions, where spin polarization effects seem to play a major role for the obtained g values. From these results, the change of the character of the SOMO as described above is less important than the actual shift of this orbital to higher energy in the 6C case, which opens up additional pathways for spin-orbit coupling. Hence, assuming that the model in Eq. (1) holds for low-symmetry ferrous heme nitrosyls, then gz , which shows the well resolved hyperfine splittings, is identified with gmin in the 5C and gmid in the 6C case, and the observed change of the EPR spectra actually relates to a shift of the SOMO to higher energy in the 6C complex, which mostly influences gx . In addition to the magnitude of the g-values and the g-tensor orientation in 5C and 6C compounds, it is interesting to compare the 14 N hyperfine tensors A listed in Table 2. For the 5C complexes, the experimental 14 N hyperfine tensor is quite isotropic with A[g(mid)] ≈ 50 MHz being somewhat larger than the other two values; although the experimental results for [Fe(TPP)(NO)] and [Fe(OEP)(NO)] are somewhat different in this respect. On the other hand, from the Fe(II) NO adducts of hemoglobin and myoglobin, it is known that the hyperfine tensor becomes anisotropic in the 6C case where A[g(mid)] ≈ 60 MHz is much larger than the other two values (cf. Table 2). In agreement with this, the A[g(mid)] value of [Fe(TPP)(Pip)(NO)] has been determined from solution EPR to 61 MHz [48]. This general trend with the A-tensor being more anisotropic for 6C compared to 5C Fe(II) NO complexes is reproduced well by the calculations as shown in Table 2. The calculated values of A[g(min)] = 49.9 MHz for [Fe(P)(NO)] and A[g(mid)] = 84.0 MHz for [Fe(P)(MI)(NO)] can directly be compared to the hyperfine coupling constants determined from solution EPR spectra and show very good agreement. The other A-values show some deviations (cf. Table 2), but the overall agreement between theory and experiment is satisfactory. Similar results for a model equivalent to [Fe(P)(MI)(NO)] have recently been published [66]. The changes of the A-tensor in going form 5C to 6C indicate that (i) the s orbital character on the nitrogen of NO is reduced in the SOMO of the 6C complex (increase in anisotropy) and (ii) that a significant amount of spin density resides on the nitrogen of NO (increase in hyperfine coupling) for 6C, which has been concluded before [82]. These implications are in perfect agreement with the electronic structure descriptions elaborated for the 5C and 6C complexes as described above. Extensive theoretical work of Patchkovskii and Ziegler, and Oldfield and coworkers has shown that the g- and A-values of the 5C and 6C compounds are very sensitive to the actual structures of the complexes, i.e., the Fe NO bond angle, the orientation of NO relative to the porphyrin ligand, and the Fe N(Im) bond length (6C) [65,66].

160

Nicolai Lehnert

Hence, these parameters reflect small structural changes of the complexes much better than, for example, vibrational frequencies or electronic absorption spectra. Note that these structural variations correspond to total energy changes of the systems of only a few kcal/mol and, therefore, these are also thermally accessible [83]. On the basis of these findings, these authors have then used the g- and A-values as structural descriptors to refine the geometries of the corresponding [Fe(P)(NO)] and [Fe(P)(MI)(NO)] model complexes. The obtained g- and A-values from these studies are also included in Tables 1 and 2, respectively.

3.3. Effect of Binding an axial Thiolate Ligand For ferrous heme nitrosyls with axial thiolate coordination, some spectroscopic data are available for corresponding species in P450-type enzymes. Resonance Raman spectroscopy shows the N O stretching vibration at 1591 cm−1 in P450cam-NO [84], which corresponds to a shift of about 30 cm−1 to lower energy of this mode compared to 6C ferrous heme complexes with axial N-donor coordination. This indicates a change of the electronic structure in the thiolate-bound complexes. From DFT calculations on the corresponding model system [Fe(P)(SCH3 )(NO)]− , it was shown that thiolate coordination leads to further weakening of the Fe NO bond and that these complexes have unusual spin populations, which are >1.0 on NO (+1.14) and negative on iron (−0.27) [57]. Using S2 -expansion techniques first successfully applied to ferrous nitrosyls by Zilberberg and coworkers [85,86], it was shown that the axial thiolate ligation in ferrous heme-NO adducts leads to a distinct admixture of Fe(III) NO− character into the Fe(II) NO(radical) ground state of these species [87]. This is due to the stabilization of Fe(III) by the strong donor thiolate and explains the change in spin populations and N O stretching frequency in this case compared to the complexes with trans N-donor coordination. Interestingly, the EPR spectra of ferrous heme nitrosyls with axial thiolate ligation do not seem to reflect this difference in electronic structures [38,54]. Figure 6 shows the

P450cam-NO

A.

gy

gz

gx

Fig. 6. EPR spectrum of the NO adduct of ferrous P450cam, taken from reference [38].

EPR and MCD Spectroscopy of Ferrous Heme Nitrosyls

161

2.108

2.104

0

0

dχ“/dB

2.068

2.054

2.010

S

2.013

[Fe(TPP)(SPh)(NO)] [Fe(TPP)(SPh)(NO)]–

S

[Fe(TPP)(H44Tp)(NO)] Tp)(NO)] [Fe(TPP)(H

3000 3100 3200 3300 3400 3500 3600 3700 3800

3000 3100 3200 3300 3400 3500 3600 3700 3800

B [G]

B [G]

Fig. 7. EPR spectra of model complexes [Fe(TPP)(L)(NO)] − , where L = thiophenolate (left) and tetrahydrothiophene (right), taken from reference [57].

EPR spectrum obtained for P450cam-NO, which shows g-values of 2.073, 2.009, and 1.976 (cf. Table 1), which are very similar to the values obtained with axial N-donor coordination. In addition, the strong hyperfine splittings of the coordinated nitrogen of NO are observed on the medium g-value (gmid ), which is again similar. This indicates similar g-tensor orientations in 6C ferrous heme-NO adducts with both axial N-donor and thiolate ligation. The calculated g-values for model system [Fe(P)(SCH3 )(NO)]− listed in Table 1 show moderate agreement with experiment and are obtained too small, as has been observed for the other models presented above. Importantly, however, in this case, the g-tensor orientation is incorrectly predicted by DFT, showing the principal axis of the minimum g-value gmin closest to the Fe N(O) axis, which is not in agreement with experiment. Hence, the DFT description of [Fe(P)(SCH3 )(NO)]− is of lower quality compared to those obtained for the 5C complex and the corresponding 6C species with axial N-donor ligation. Model complex studies on heme nitrosyls with axial thiolate coordination are rare, because these complexes are unstable; especially due to the fact that NO is able to react with (coordinated) thiolates. Complexes with bound thiophenolates and thioethers have been prepared in situ and investigated using EPR spectroscopy [55,57]. Figure 7 shows the corresponding spectra obtained by addition of thiophenolate and tetrahydrothiophene, respectively, to the 5C complex [Fe(TPP)(NO)] as examples [57]. In both cases, similar g-values are obtained as for the starting (5C) material and in addition, the well resolved hyperfine splittings of the coordinated nitrogen of NO are observed on the minimum g-value (gmin ) as in the 5C case (cf. Table 1). This indicates that the Fe S interaction is weak in these cases and hence, these compounds are not suitable models for the NO adducts of the P450 enzymes.

4. MCD SPECTROSCOPY ON FERROUS HEME NITROSYLS MCD spectroscopic investigations on ferrous heme nitrosyls have been conducted by a number of researchers (cf. reference [76,77] for examples). However, the spectra have

162

Nicolai Lehnert

usually been recorded at room temperature. The general formula for MCD intensity, I, derived by P. J. Stephens [67,68]  I ∼ A1



    −f E

C0 f E

+ B0 + kT E

(2)

shows that in this case, the MCD spectra are dominated by the A- and B-term signals, which are temperature independent and do not require a paramagnetic ground state. However, detailed information about the ground state of a paramagnetic molecule is mostly available from the C-term signal, the intensity of which is inversely proportional to the temperature, as can be seen from Eq. (2). Hence, the C-term spectrum has to be determined at very low temperature (usually in the range of 1.8–30 K). Temperature and field dependent measurements of the C-term intensity then reveal detailed insight into the properties of the ground state (zero field splitting, etc.) as well as the polarizations of electronic transitions as has been demonstrated in the literature [67–75]. Recently, the low-temperature MCD spectra of 5C [Fe(TPP)(NO)] and 6C [Fe(TPP)(MI)(NO)] have been determined [63]. As shown in Fig. 8, these data show characteristic differences. In the 5C case, the calculated C-term intensity (in grey) is identical to the total MCD spectrum (in black). This is the normal case for paramagnetic compounds like ferrous heme nitrosyls, which have an S = 1/2 ground state (vide supra). In comparison, the total MCD intensity (black) for the 6C complex (cf. Fig. 8, right) is different from the C-term spectrum (grey) and shows large diamagnetic (A- and/or B-term) contributions. This means that the iron starts to behave like a diamagnetic lowspin iron(II) in this complex, which indicates that the binding of the axial N-donor ligand leads to a distinct reduction of the spin density on iron. The most likely explanation for this finding is that the coordination of the N-donor ligand pushes the spin density from the iron back to the NO ligand, which is in agreement with the electronic structure descriptions for the 5C and 6C ferrous heme nitrosyls elaborated above. Besides the evaluation of the total MCD and C-term intensities of 5C and 6C Fe(II) NO adducts, another important aspect is the actual assignment of the C-term spectra and the question of whether any charge-transfer (CT) transitions between iron(II) and NO are observed in these data. In general, the absorption as well as the MCD spectra of metalloporhyrins are dominated by intense  to  ∗ transitions of the porphyrin dianion. These have been analyzed in detail by Gouterman using his famous four-orbital model shown in Scheme 3 [88]. Two  orbitals of a2u and a1u symmetry constitute the HOMO and HOMO-1 of the porphyrin ligand, respectively, whereas the LUMO is a  ∗ orbital of eg symmetry. This leads to two  to  ∗ excited states of Eu symmetry (in D4h ): 1  a1u −→ eg

and

2  a2u −→ eg

These show strong configuration interaction (CI) leading to the intense Soret- or B-band at higher energy ( ≥ 100000 M−1 cm−1 ) and the Q-band at lower energy, which has low to zero intensity. The corresponding wave functions for these excited states are defined as:  Soret 1  = √ 1 + 2  2

and

 Q  = √1 1 − 2  2

(3)

EPR and MCD Spectroscopy of Ferrous Heme Nitrosyls 2000

163

[Fe(TPP)(NO)] (I )

Δε [M–1cm–1T–1]

1500 1000

x10

500 0

–500 –1000 MCD C-Term

–1500 –2000 –2500 30000

27500

25000

22500

20000

17500

15000

Δε [M–1cm–1T–1]

Wavenumbers (cm–1) 120 100 80 60 40 20 0 –20 –40 –60 –80 –100 –120 –140

[Fe(TPP)(MI)(NO)] (II)

x5

MCD C-Term

30000

27500

25000

22500

20000

17500

15000

Wavenumbers (cm–1)

Fig. 8. MCD (black) and C-term (grey) spectra of five-coordinate [Fe(TPP)(NO)] (top) and sixcoordinate [Fe(TPP)(MI)(NO)] (bottom) (Reprinted with permission from ref. [63]. Copyright 2005 American Chemical Society).

eg

(2)

D4h

eg

C1

(1)

Soret, Q, Qv

perturbation

eg

a2u a1u

a2u a1u Four orbital model of Gouterman

Scheme 3.

Soret, Q, Qv

164

Nicolai Lehnert 120000

15000 [Fe(TPP)(NO)] 405 nm

100000

8

10000

ε [M–1 cm–1]

ε [M–1 cm–1]

80000 4

60000

40000

7

20000

5000

5

10

2

6 5

Δε [M–1 cm–1]

0 40000 1200 1000 800 600 400 200 0 –200 –400 –600 –800 –1000 –1200 –1400 –1600 40000

35000

30000

25000

1

3

x8

0 10000

15000

20000

8

10

6

9

2 1

5

x 10

C-Term 4 3

7

35000

30000

25000

20000

15000

10000

Wavenumbers (cm–1)

Fig. 9. Absorption (top) and MCD C-term (bottom) spectra of five-coordinate [Fe(TPP)(NO)] including Gaussians from a correlated fit of the data (Reprinted with permission from ref. [64]. Copyright 2006 American Chemical Society).

Vibronic mixing of the Q- with the Soret-state leads to a third band, designated as Qv , about 1250 cm−1 to higher energy from Q. The vibronic nature of this quite intense band ( ≈ 10000 M−1 cm−1 ) has been proven by Resonance Raman spectroscopy [89–92]. Figure 9 shows the absorption and MCD spectra of 5C [Fe(TPP)(NO)] as an example [64]. As one can see, the absorption spectra of ferrous heme nitrosyls are dominated by the  to  ∗ transitions of the porphyrin ligand, in agreement with the findings of Gouterman: the Soret-band appears around 400 nm and the Qv -band is found at about 540 nm. On the other hand, the MCD C-term spectrum shown in Fig. 9 (bottom) allows for the identification of about 10 individual transitions below 30,000 cm−1 , which again demonstrates the enormous potential of this method for the investigation of the electronic spectra of paramagnetic compounds.

EPR and MCD Spectroscopy of Ferrous Heme Nitrosyls

165

4.1. Assignment of the C-term Spectra of Ferrous Heme Nitrosyls Owing to the presence of the phenyl substituents, the out-of-plane distortions of the porphyrin core [45] (cf. also the crystal structures of [Fe(TPP)(MI)(NO)] [62] and [Fe(To-F2 PP)(MI)(NO)] [64]) and the bent NO ligand, the effective symmetry of the porphyrin ligand in ferrous heme nitrosyls is lower than D4h . Hence, the twofold degeneracy of the LUMO is lifted (cf. Scheme 3, right), which leads to a splitting of the individual excited states 1 and 2 into two components: 1

2a  a2u → eg

2

2a  a2u → eg

1a  a1u → eg

1b  a1u → eg

1

2

Since the effective perturbation of the fourfold symmetry can be expected to be small, the orbitals will still be labeled within the D4h group. CI coupling of these transitions then leads via cross-coupling to the Soret- and Q-states:  Soret 1  = √ 1a + 2b  1 2  Soret 1  = √ 2a + 1b  2 2

 Q 1 = √1 1a − 2b  2  Q 2 = √1 2a − 1b  2

(4)

As shown by Neese and Solomon, the MCD C-term intensity C0 for an S = 1/2 (orbitally nondegenerate) ground state is defined [70]: C0 = −



 −1  KA 1 −1  vAJ · L¯ KJ u ×D  AJ  JK · L¯ KA uvw gw KJ D W + KA Du × Dv W 6 uvw K=AJ

(5)

for a transition from the ground state A> to the excited state J>. Here, uvw is the Levi-Civita symbol with u v w = x y z ; gw are the g-values of the molecule; the sum over K runs over all other excited states K = J, which are intermediate states;  uKA , for example, is the transition dipole KJ = EK − EJ is the energy denominator; D moment for a transition between the states A> and K> in direction u; and L¯ KJ W = w ImKHSOC (SOC) matrix element between J , for example, is the spin-orbit coupling  w states K> and J> in direction w, where HSOC =  rA lAW (the integrals over the A

spin functions have already been taken into account). Eq. (5) describes two different mechanisms for C-term intensity as shown in Scheme 4. In order for C-term intensity to arise, two transitions with orthogonal transition dipole moments are needed and, in addition, either the two excited states J> and K> (mechanism 1: Scheme 4, middle), or the ground and the intermediate state A> and K> (mechanism 2: Scheme 4, right) have to spin-orbit couple in the direction orthogonal to the plane formed by the two  u and D  v . Importantly, as discussed above, the ferrous heme transition dipole moments D nitrosyl complexes have low-spin d6 configurations of the metal (cf. Scheme 2), where the three “t2 ” orbitals of iron(II) are fully occupied. Hence, no low-energy ligand-field transitions and porphyrin () to metal (t2 ) charge-transfer (CT) transitions are possible. Considering the Soret transition, this leads to the simple energy diagram shown in Scheme 4 (left). Since the two transitions from the ground state A> to the two Soret

166

Nicolai Lehnert

|K> |J>

|A>

E

|J>

|K> |A>

|A> −1

AJ

JK

I MCD ∼ ΔKA(Du × Dv

)⋅LwKA

−1

KA

SOC

E

MCD C-Term Mechanisms 2 SOC

Soret (2) Soret (1)

1

AJ

I MCD ∼ ΔKJ (Du × Dv

)⋅LwKJ

Scheme 4.

excited states have orthogonal transition dipole moments, the electronic structure of the ferrous heme nitrosyls is clearly consistent with the C-term mechanism 1, shown in Scheme 4 (middle). Using the wave functions in Eq. (4), this leads to the expression [64]:   KA  AJ · L¯ KJ I MCD ∼ −1 w = KJ Du × Dv

=

  1  Soret  u A × A v 2Soret 1 Soret  W  Soret   · Im 1Soret  HSOC 2 1  Soret 1

 vSoret 2

u ×D D 2Soret W · Im 1a + 2b  HSOC 1b + 2a 

considering 1Soret as the intermediate state K> and 2Soret as the excited state J>.  uSoret 1 and D  vSoret 2 are the transition dipole moments of the two individual Soret Here, D  uSoret 1 × D  Soret 2

components, which are large. Hence, their vector product D  large  vSoret  isw also and the MCD intensity then depends on the SOC matrix element 1  HSOC 2Soret .  uSoret 1 and D  vSoret 2 are oriented within the porphyrin (xy) plane (cf. Scheme 2), Since D SOC of these states must be effective along the z-axis (w = z). The SOC matrix element can then be expanded to:   1  z Im 1Soret  H zSOC 2Soret = Im 1a + 2b  HSOC 1b + 2a  2 1 z z = Im 1a  HSOC 1b  + 1a  HSOC 2a  2 z z + 2b  HSOC 1b  + 2b  HSOC 2a     z e 2 + Im a1u  H z a2u  = Im eg 1  HSOC g SOC

(6)

leading to a sum of two SOC matrix elements, one over the two components of the former eg (LUMO) orbital and the other one over the a1u and a2u orbitals. The latter contribution  1

vanishes. Using the LCAO approximation for molecular orbitals, i.e., eg = cr r and r

EPR and MCD Spectroscopy of Ferrous Heme Nitrosyls 2

eg =

 s

167

cs s , and neglecting two- and three-center integrals, the first contribution can

be written as: Im







z e 2

eg 1  HSOC g



= Im

  A

   A  A  cr r lAz cs s  A

(7)

rs

where the sum over A runs over all atoms in the complex, and A is the spin-orbit coupling constant of atom A. Importantly, since lAz pzA = 0, the integrals in Eq. (7) should all vanish for pure porphyrin  or  ∗ orbitals leading to zero C-term intensity. However, as shown in reference [64], the eg orbitals of the porphyrin undergo a small backbonding interaction with the dxz and dyz orbitals of iron(II). Assuming identical 1

2

mixing coefficients c of the d orbitals for eg and eg , one obtains:  Im

 A

  A   A     r lAz s = Im −i Fe c2 = −c2  Fe  A

(8)

rs

Therefore, the MCD intensity for the Soret-band is obtained as [64]: 1  Soret 1

w  vSoret 2 · Im 1a + 2b  HSOC u ×D D 1b + 2a  2Soret c2  Fe  Soret 1

 vSoret 2

u =− ×D D Soret

I MCD ∼

(9)

Although the SOC matrix element is quite small due to the small amount of d orbital admixture into the porphyrin eg orbitals, this is compensated by the large transition dipole moments of the Soret band. Hence, this should give rise to relatively large MCD C-term intensity in the Soret region. In the above derivation, the electronic transition to 2Soret is treated as the excited state J> and 1Soret serves as intermediate state K>. If one alternatively considers the transition to 1Soret as J> and 2Soret as intermediate state K>, the same expression for the C-term intensity in Eq. (9) is obtained, but due to the sign change in the denominator −1 that occurs when K> and J> are exchanged, the resulting C-term signal changes its sign. Hence, this mechanism gives rise to a so-called pseudo-A term in the MCD spectrum [73–75], where two adjacent bands are observed with opposite signs due to SOC of the corresponding excited states. The same applies to the Q- and Qv -bands in MCD, but these features will be of lower C-term intensity due to their smaller transition dipole moments. Note that the C-term mechanism 2 in Scheme 4 (right) would, in principle, allow to directly utilize the unpaired electron density in dz2 for the generation of MCD intensity. However, the in-plane (xy) polarized transitions require SOC in z-direction, and, since lAz dAz2 = 0, this mechanism cannot contribute. With this information at hands, the MCD C-term spectrum of [Fe(TPP)(NO)] can easily be assigned. As shown in Fig. 9, bands 7 and 8 belong to the two Soret components (split by approximately 1000–1500 cm−1 ). This is a very important result, because the energy splitting of the two Soret components serves as a measure for the low-symmetry perturbation of the porphyrin ring. This information is not accessible from the absorption spectra. The splitting of the two components of Q is large (∼1800 cm−1 ), which leads 1

to a pattern where Q 1 (band 1) is followed by its vibronic band Qv (band 2; both

168

Nicolai Lehnert

with positive sign) before the second component Q 2 (band 3; negative sign) appears. Time-dependent (TD-) DFT and semiempirical INDO/S-CI calculations show that the remaining bands mostly belong to other porphyrin-centered transitions, or porphyrin() to iron(d) CT transitions, which steal some intensity from the Soretband [64]. Unfortunately, there is no indication for the appearance of CT transitions between iron and NO in the MCD spectra of ferrous heme nitrosyls.

5. CONCLUSIONS Both EPR and MCD spectroscopies are very powerful tools for the investigation of ferrous heme nitrosyls. The results presented in this review show that these methods are very sensitive to changes in the ground state (i.e., metal-ligand covalency, spindensity distribution, etc.) of these species. From EPR, both the absolute magnitude of the g-shifts, and the magnitude and observed pattern of the 14 N hyperfine splittings of NO are directly linked to the coordination number of the iron center. Relatively strong ligands like N-donors or cysteinate (in P450-type enzymes) lead to small g-shifts and well resolved 14 N hyperfine splittings on the medium g-value gmid , whereas 5C complexes or compounds with weak axial ligands (such as alcohols, (thio)ethers, thiophenolates, etc.) show larger g-shifts and strong hyperfine splittings on the minimum g-value gmin . From MCD, the overall paramagnetic and diamagnetic contributions to the spectral intensity are linked to the spin-density distribution in these complexes. The detailed assignments of the C-term spectra provide access to the energy splittings of the Soret- and Q-bands, which are a measure for the (electronic) low-symmetry distortions of the porphyrin ring in the 5C and 6C complexes. The results show that this splitting is distinctively larger for the 5C species [64]. Considering the Fe NO interaction, the spectral assignments do not provide further insight as no charge-transfer transition between iron and NO can be identified in the spectra.

ACKNOWLEDGMENT This work was supported by the Deutsche Forschungsgemeinschaft (DFG; grant LE 1393/1-2) and the Fonds der Chemischen Industrie (FCI).

REFERENCES [1] [2] [3] [4] [5] [6]

Moncada, S., Palmer, R.M. and Higgs, E.A. (1991) Pharmacol. Rev. 43, 109–142. Snyder, S.H. (1992) Science. 257, 494–496. Culotta, E. and Koshland, D.E. (1992) Science 258, 1862–1865. Butler A.R. and Williams, D.L.H. (1993) Chem. Soc. Revs. 22, 233–241. Bredt D.S. and Snyder, S.H. (1994) Annu. Rev. Biochem. 63, 175–195. Lancaster, J.R., Jr. (1994) In Encyclopedia of Inorganic Chemistry, (R.B Bruce., ed.). Chichester: Wiley. [7] Feelisch, M. and Stamler, J.S. (eds.) (1996) Methods in Nitric Oxide Research, Chichester; Wiley.

EPR and MCD Spectroscopy of Ferrous Heme Nitrosyls [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41]

[42] [43]

[44] [45] [46] [47]

169

Cooper, C.E. (1999) Biochim. Biophys. Acta 1411, 290–309. Stamler, J.S., Singel, D.J. and Loscalzo, J. (1992) Science 258, 1898–1902. Hughes, M.N. (1999) Biochim. Biophys. Acta 1411, 263–272. Ignarro. L. (ed.) (2000) Nitric Oxide: Biology and Pathobiology, San Diego: Academic Press. Ballou, D.P., Zhao, Y., Brandish, P.E. and Marletta, M.A. (2002) Proc. Natl. Acad. Sci. U. S. A. 99, 12097–12101. Feelisch, M., Rassaf, T., Mnaimneh, S. et al. (2002) FASEB J. 16, 1775–1785. McCleverty, J.A. (2004) Chem. Rev. 104, 403–418. Farmer, P.J. and Sulc, F.J. (2005) Inorg. Biochem. 99, 166–184. Boon, E.M. and Marletta, M.A.J. (2005) Inorg. Biochem. 99, 892–902. Luchsinger, B.P., Rich, E.N., Yan, Y. et al. (2005) Inorg. Biochem. 99, 912–921. Bryan, N.S., Fernandez, B.O., Bauer, S.M. et al. (2005) Nature Chem. Biol. 1, 290. Gladwin, M.T. (2005) Nature Chem. Biol. 1, 245. Gladwin, M.T., Schechter, A.N., Kim-Shapiro, D.B. et al. (2005) Nature Chem. Biol. 1, 308. Stuehr, D. (1997) J. Annu. Rev. Pharmacol. Toxicol. 37, 339–359. Poulos, T.L., Li, H. and Raman, C.S. (1999) Curr. Opin. Chem. Biol. 3, 131–137. Li, H. and Poulos, T.L. (2005) J. Inorg. Biochem. 99, 293–305. Rousseau, D.L., Li, D., Couture, M. and Yeh, S.-R. (2005) J. Inorg. Biochem. 99, 306–323. Garbers, D.L. and Lowe, D.G. (1994) J. Biol. Chem. 269, 30741–30744. Zhao, Y., Hoganson, C., Babcock, G.T. and Marletta, M.A. (1998) Biochemistry 37, 12458–12464. Karow, D.S., Pan, D., Tran, R. et al. (2004) Biochemistry 43, 10203–10211. Gilles-Gonzalez, M.-A. and Gonzales, G. (2005) J. Inorg. Biochem. 99, 1–22. Ferguson, S.J. (1998) Curr. Opin. Chem. Biol. 2, 182–193. Richardson, D.J. and Watmough, N.J. (1999) Curr. Opin. Chem. Biol. 3, 207–219. Moura, I. and Moura, J.J.G. (2001) Curr. Opin. Chem. Biol. 5, 168–175. Kon, H. (1968) J. Biol. Chem. 243, 4350–4357. Kon, H. and Kataoka, N. (1969) Biochemistry 8, 4757–4762. Yonetani, T., Yamamoto, H., Erman, J.E. et al. (1972) J. Biol. Chem. 247, 2447–2455. Brunori, M., Falcioni, G. and Rotilio, G. (1974) Proc. Natl. Acad. Sci. U.S.A. 71, 2470–2472. Perutz, M.F., Kilmartin, J.V., Nagai, K. et al. (1976) Biochemistry 15, 378. Doetschman, D.C., Schwartz, S.A. and Utterback, S.G. (1980) Chem. Phys. 49, 1–8. O’Keeffe, D.H., Ebel, R.E. and Peterson, J.A. (1978) J. Biol. Chem. 253, 3509–3516. Hüttermann, J., Burgard, C. and Kappl, R. (1994) J. Chem. Soc. Faraday Trans. 90, 3077–3087. Tyryshkin, A.M., Dikanov, S.A., Reijerse, E.J. et al. (1999) J. Am. Chem. Soc. 121, 3396–3406. Henry, Y.A. (1997) EPR Characterization of Nitric Oxide Binding To Hemoglobin” (Chapter 4) and ‘Utilization of Nitric Oxide as a Paramagnetic Probe of the Molecular Oxygen Binding Site of Metalloenzymes’ (Chapter 6). In Nitric Oxide Research from Chemistry to Biology: EPR Spectroscopy of Nitrosylated Compounds (Y.A. Henry, A. Guissani and B. Ducastel, eds.), Austin: R.G. Landes Company, and references therein. Møller, J.K.S. and Skibsted, L.H. (2002) Chem. Rev. 102, 1167–1178. Henry, Y.A., Guissani, A. and Ducastel B., (eds.) (1997) Nitric Oxide Research from Chemistry to Biology: EPR Spectroscopy of Nitrosylated Compounds Austin: R. G. Landes Company. Ford, P.C. and Lorkovic, I.M. (2002) Chem. Rev. 102, 993–1017. Wyllie, G.R.A. and Scheidt, R. (2002) Chem. Rev. 102, 1067–1089. Wasser, I.M., de Vries, S., Moënne-Loccoz, P. et al. (2002) Chem. Rev. 102, 1201–1234. Walker, F.A. and Simonis, U. (2005) Iron porphyrin chemistry, In Encyclopedia of Inorganic Chemistry (R.B. King ed.), Chichester: Wiley.

170

Nicolai Lehnert

[48] [49] [50] [51] [52] [53] [54] [55] [56] [57]

Wayland, B.B. and Olson, L.W. (1974) J. Am. Chem. Soc. 96, 6037–6041. Morse, R.H. and Chan, S.I. (1980) J. Biol. Chem. 255, 7876–7882. Yoshimura, T. (1978) Bull. Chem. Soc. Jpn. 51, 1237–1238. Yoshimura, T. (1980) Inorg. Chim. Acta 46, 69–76. Yoshimura, T. (1986) Inorg. Chim. Acta 125, L27–L29. Yoshimura, T. (1991) Bull. Chem. Soc. Jpn. 64, 2819–2828. Shiro, Y., Fujii, M., Isogai, Y. et al. (1995) Biochemistry 34, 9052–9058. Yoshimura, T. (1982) Inorg. Chim. Acta 57, 99–105. Suzuki, N., Higuchi, T., Urano, Y. et al. (2000) J. Am. Chem. Soc. 122, 12059–12060. Praneeth, V.K.K., Haupt, E. and Lehnert, N. (2005) J. Inorg. Biochem. 99, 940–948. Erratum: ibid., 1744. Hayes, R.G., Ellison, M.K. and Scheidt, W.R. (2000) Inorg. Chem. 39, 3665–3668. Utterback, S.G., Doetschman, D.C., Szumowski, J. and Rizos, A.K. (1983) J. Chem. Phys. 78, 5874–5880. Hori, H., Ikeda-Saito, M. and Yonetani, T. (1981) J. Biol. Chem. 256, 7849–7855. Kappl, R. and Hüttermann, J. (1989) Isr. J. Chem. 29, 73–84. Wyllie, G.R.A., Schulz, C.E. and Scheidt, W.R. (2003) Inorg. Chem. 42, 5722–5734. Praneeth, V.K.K., Neese, F. and Lehnert N. (2005) Inorg. Chem. 44, 2570–2572. Praneeth, V.K.K., Näther, C., Peters, G. and Lehnert, N. (2006) Inorg. Chem. 45, 2795–2811. Patchkovskii, S. and Ziegler, T. (2000) Inorg. Chem. 39, 5354–5364. Zhang, Y., Gossman, W. and Oldfield, E.J. (2003) Am. Chem. Soc. 125, 16387–16396. Stephens, P.J. (1974) Ann. Rev. Phys. Chem. 25, 201–232. Stephens, P.J. (1975) Adv. Chem. Phys. 35, 197–264. Solomon, E.I., Pavel, E.G., Loeb, K.E. and Campochiaro, C. (1995) Coord. Chem. Rev. 144, 369–460. Neese, F. and Solomon, E.I. (1999) Inorg. Chem. 38, 1847–1865. Oganesyan, V.S., George, S.J., Cheesman, M.R. and Thomson, A.J. (1999) J. Chem. Phys. 110, 762–777. Lehnert, N., DeBeer George, S. and Solomon, E.I. (2001) Curr. Opin. Chem. Biol. 5, 176–187. Cheesman, M.R., Greenwood, C. and Thomson, A.J. (1991) Adv. Inorg. Chem. 36, 201–255. Walker, F.A. (1999) Coord. Chem. Rev. 185–186, 471–534. Cheek, J. and Dawson, J.H. (2000) Magnetic circular dichroism spectroscopy of heme proteins and model systems, In The Porphyrin Handbook, Vol. 7, (K.M. Kadish, K.M. Smith and R. Guilard eds.), New York: Academic Press. Yoshimura, T., Kamada, H., Toi, H. et al. (1993) Inorg. Chim. Acta 208, 9–15. Suzuki, S., Yoshimura, T., Nakahara, A. et al. (1987) Inorg. Chem. 26, 1006–1008. Lever, A.B.P. and Solomon, E.I. (1997) Ligand field theory and the properties of transition metal complexes, In Inorganic Electronic Structure and Spectroscopy, Vol. 1, (E.I. Solomon and A.B.P. Lever, eds.), New York: Wiley. Maki, A.H., Edelstein, N., Davidson, A. and Holm, R.H. (1964) J. Am. Chem. Soc. 86, 4580–4587. McGarvey, B.R. (1975) Can. J. Chem. 53, 2498. Hille, R., Olson, J.S. and Palmer, G. (1979) J. Biol. Chem. 254, 12110–12120. Palmer, G. (1983) Electron paramagnetic resonance of hemoproteins, In Iron Porphyrins, Part 2 (A.B.P. Lever and H.B. Gray, eds.), Massachusetts: Addison-Wesley. Zemojtel, T., Rini, M., Heyne, K. et al. (2004) J. Am. Chem. Soc. 126, 1930–1931. Hu, S., Kincaid, J.R.J. (1991) Am. Chem. Soc. 113, 9760–9766. Zilberberg, I., Ruzankin, S., Ph., Malykhin, S. and Zhidomirov, G.M. (2004) Chem. Phys. Lett. 394, 392–396. Zilberberg, I. and Ruzankin, S., Ph. (2004) Chem. Phys. Lett. 394, 165–170. Lehnert, N., Praneeth, V.K.K. and Paulat, F. (2006) J. Comp. Chem. 27, 1338–1351.

[58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75]

[76] [77] [78]

[79] [80] [81] [82] [83] [84] [85] [86] [87]

EPR and MCD Spectroscopy of Ferrous Heme Nitrosyls

171

[88] Gouterman, M. (1978) Optical Spectra and Electronic Structure of Porphyrins and Related Rings, In The Porphyrins, Vol. III (D. Dolphin, ed.), New York: Academic Press. [89] Spiro, T.G. and Strekas, T.C. (1972) Proc. Nat. Acad. Sci. U.S.A. 69, 2622–2626. [90] Spiro, T.G. (1983) In Iron Porphyrins, Part 2 (A.B.P. Lever and H.B. Gray, eds.), Massachusetts: Addison-Wesley. [91] Egawa, T., Suzuki, N., Dokoh, T. et al. (2004) J. Phys. Chem. A 108, 568–577. [92] Paulat, N., Praneeth, V.K.K., Näther, C. and Lehnert, N. (2006) Inorg. Chem. 45, 2835–2856.

This page intentionally left blank

Part III Aspects of Hemoglobins (Except Heme

NOx Interactions)

This page intentionally left blank

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 7

Protoglobin and Globin-coupled Sensors Tracey Allen K. Freitasabc , Jennifer A. Saitoa , Xuehua Wana , Shaobin Houc , and Maqsudul Alamabc a

Department of Microbiology, Snyder Hall 207, 2538 McCarthy Mall, University of Hawaii, Honolulu, HI 96822, USA b Maui High Performance Computing Center, 550 Lipoa Parkway, Kihei, Maui, HI 96753 c Advanced Studies in Genomics, Proteomics and Bioinformatics, Keller Hall 319, 2565 McCarthy Mall, University of Hawaii, Honolulu, HI 96822, USA

Abstract The strategy for detecting oxygen, carbon monoxide, nitric oxide, and sulfides is predominantly through heme-based sensors utilizing either a globin domain or a PAS domain. Whereas PAS domains bind various cofactors, globins bind only heme. Globincoupled sensors (GCSs) couple an N-terminal sensor globin domain to varied C-terminal signaling domains to effect aerotaxis and gene regulation. Having descended from an ancient protoglobin, GCSs are now ubiquitous and are encoded in the genome of several extremophiles (temperature, salt, and pH). We postulate that their role in regulating gene expression governs microbial processes critical to elemental recycling, bioremediation, and cellulose degradation. Functional and evolutionary analyses of the GCSs, their protoglobin ancestor, and their relationship to the Last Universal Common Ancestor (LUCA) are also discussed in the context of globin-based signal transduction.

1. BACKGROUND ON THE GLOBINS Globins are ubiquitous heme-binding -helical proteins whose function is principally thought to regulate oxygen homoeostasis [1,2]. Initially described as an eight-helix globin fold labeled A through H [3], other members of the globin family have been observed that are less helical, with the “truncated” globins maintaining the overall globin fold with as few as four helices [4]. In contrast to PAS domains that can bind an assortment of cofactors, globins are known to bind only heme. The hydrophobic core of the globin fold favors heme binding, but it is the proximal F8 (helix F, position 8) histidine that actually binds the heme cofactor to the protein by coordinating to the heme iron’s fifth coordinate. Gaseous ligands bind to the iron’s sixth coordinate at the opposite side of the heme plane. Ligand binding and discrimination in this distal region is usually regulated by at least one distal residue, and in many mammalian globins, a distal E7 histidine accomplishes this [5]. In the Ascaris hemoglobin, however, an E7 Gln and a B10 Tyr regulate ligand binding by forming a tight hydrogenbond network with bound oxygen, thus imparting a very high O2 affinity [6]. The heme iron in globins cycles between a reduced and an oxidized state, with the oxidation state capable of binding different ligands. Known physiological ligands of

176

T.A.K. Freitas et al.

the reduced form include O2 , CO, H2 S, and NO, although NO is known to bind to the oxidized form as well [7]. The first globin-coupled sensors to be characterized were the heme-based aerotactic transducers, or HemATs [1]. HemAT-Hs, HemAT-Bs, and HemAT-Bh all possess C-terminal signaling/transmitter domains similar to that of the chemotactic transducers in Escherichia coli and were shown to bind O2 and CO at the N-terminus [1, 8]. All globin-coupled sensors thus far purify in the ferrous O2 -bound form [1,8–11], giving a characteristic heme-protein spectrum with the Soret band at ∼410 nm.

2. HISTORY OF THE GLOBIN-COUPLED SENSORS AND THE PROTOGLOBIN In February of 2000, the first heme-based aerotactic transducers (HemAT) in Archaea and Bacteria were discovered [1]. Identified initially by their absorption spectra and heme staining, HemAT-Hs from Halobacterium salinarum and HemAT-Bs from Bacillus subtilis were shown by time-lapsed capillary assay [12] to elicit aerophobic and aerophilic responses, respectively. While their C-termini were clearly identified as signaling domains analogous to those of bacterial methyl-accepting chemotaxis proteins (MCP), the N-termini possessed only a weak similarity (∼10% similarity) to the globin fold. Molecular modeling, site-directed mutagenesis, and truncation experiments were used to determine the minimal protein length required to bind heme and the histidine residue of heme-iron coordination [8]. Soon after, the structure of the HemAT-Bs globin domain was determined [13,14] by X-ray crystallography, and biophysical, kinetic, and equilibrium experiments ensued with the aim of elucidating its binding properties and signaling mechanism [10,11,15]. Similar HemATs were found in other microorganisms [8,9], and it became apparent that such globin “sensors” were prevalent in bacteria with functions surpassing aerotaxis [16]; thus, the term globin-coupled sensor (GCS) was coined. Currently, over 80 GCSs have been identified in approximately 60 bacteria and three archaea, with each sensor tentatively classified as either aerotactic or gene regulating [16,17]. As genomesequencing projects generate more sequence data, additional GCSs will emerge and this classification scheme may have to be expanded. A phylogenetic tree based on the GCS globin domains indicated that each globin domain evolved independently with its particular signaling partner. It was thought that the ancestor globin, or protoglobin, could still exist within more primitive organisms, perhaps the Archaea or deeply branching photosynthetic bacteria and indeed, such protoglobins were discovered in two Archaea and two Bacteria and were shown to be heme proteins with very unusual properties [18].

3. GCS FUNCTIONAL CLASSIFICATION As the number and domain variability of the GCSs identified increased, it became increasingly difficult to classify them on their own. Hence, all biological heme-based sensors with known functions were collected and classified as either aerotactic or gene regulating [16]. The gene-regulating group was further subdivided into three subgroups: protein-DNA [19–23], protein–protein [1,24–26], and 2nd messenger [27–32] pathways. Table 1 summarizes the ubiquitous nature of the globin-coupled sensors and protoglobins, many (>30) of which, due to their specialized niches critical to the process

No. 1 2 3 4 5 6 7 8 9 10 11 12 13

Name

NCBI accession #

Classification

Pfam

Taxonomic ID

Protein length

Other name

Magnetococcus sp. MC-1 Silicibacter sp. TM1040 Silicibacter sp. TM1040 Caulobacter crescentus str. CB15 Caulobacter crescentus str. CB15 Agrobacterium tumefaciens str. C58 Rhizobium etli str. CFN 42 Bradyrhizobium sp. BTAi1 Nitrobacter hamburgensis str. X14 Rhodopseudomonas palustris str. BisB18 Oceanicaulis alexandrii str. HTCC2633 Oceanicaulis alexandrii str. HTCC2633 Rhodobacter sphaeroides str. ATCC 17029

HemAT-Mg

EAN27973.1

Aerotactic

HAMP:MCP

-Proteobaceria

515

Mmc10749

HemAT-SiB

ZP_00623162.1

Aerotactic

HAMP:MCP

-Proteobaceria

485

–

HemAT-SiA

ZP_00623428.1

Aerotactic

MCP

-Proteobaceria

487

–

McpB

NP_419247

Aerotactic

HAMP:MCP

-Proteobacteria

537

McpB

McpM

NP_421120

Aerotactic

HAMP:MCP

-Proteobacteria

555

McpM

HemAT-At

NP_354049

Aerotactic

HAMP:MCP

-Proteobacteria

499

AGR_C_1888

HemAT-Re

YP_467960.1

Aerotactic

HAMP:MCP

-Proteobacteria

523

–

HemAT-Bra

ZP_00859922.1

Aerotactic

MCP

-Proteobacteria

450

–

HemAT-Nh

ZP_00625186.1

Aerotactic

MCP

-Proteobacteria

422

–

HemAT-Rp

ZP_00848742.1

Aerotactic

MCP

-Proteobacteria

550

–

HemATOaA HemATOaB HemAT-Rs

ZP_00952054.1

Aerotactic

MCP

-Proteobacteria

467

–

ZP_00952053.1

Aerotactic

MCP

-Proteobacteria

452

McpK

ZP_00918112.1

Aerotactic

MCP

-Proteobacteria

370

TlpL

(Continued)

177

Organism

Protoglobin and Globin-coupled Sensors

Table 1. Domain organization, source and accession numbers for the globin-coupled sensors. Trx = transcription; TM = transmembrane

178

Table 1. (Continued) Organism

Name

NCBI accession #

Classification

Pfam

Taxonomic ID

14

Acidiphilium cryptum str. JF-5 Gluconobacter oxydans str. 621H Magnetospirillum magnetotacticum MS-1 Magnetospirillum magnetotacticum MS-1 Rhodospirillum rubrum str. ATCC 11170 Novosphingobium aromaticivorans str. DSM 12444 Zymomonas mobilis subsp. mobilis str. ZM4 environmental sample Haloarcula marismoriui str. ATCC 43049 Halobacterium sp. NRC-1 Natronomonas pharaonis DSM 2160 Natronomonas pharaonis DSM 2160

HemAT-Ac

ZP_01146785.1

Aerotactic

HAMP:MCP

-Proteobacteria

481

–

HemAT-Go

AAW60540.1

Aerotactic

MCP

-Proteobacteria

458

–

HemAT-MmA

ZP_00054774

Aerotactic

MCP

-Proteobacteria

443

Magn7582

HemAT-MmB

ZP_00054075

Aerotactic

MCP

-Proteobacteria

721

Magn6867

HemAT-Rr

YP_426265.1

Aerotactic

MCP

-Proteobacteria

469

Rrub1164

HemAT-Na

YP_496708.1

Aerotactic

HAMP:MCP

-Proteobacteria

481

Saro2089

HemAT-Zm

AAV89506.1

Aerotactic

MCP

-Proteobacteria

467

–

EAC21812 HemAT-Hm

EAC21812.1 YP_134953.1

Aerotactic Aerotactic

MCP HAMP:MCP

environmental Euryarchaeota (Archaea)

304 497

– –

HemAT-Hs

NP_280321

Aerotactic

MCP

489

HemAT-NpB

YP_325693.1

Aerotactic

MCP

492

HtrX, HtB, Htr10 –

HemAT-NpA

YP_326586.1

Aerotactic

PAS:PAS:MA

Euryarchaeota (Archaea) Euryarchaeota (Archaea) Euryarchaeota (Archaea)

661

Htr26

15 16 17 18 19

20 21 22

23 24 25

Protein length

Other name

T.A.K. Freitas et al.

No.

27 28 29 30 31 32 33

34 35

36 37 38

Bacillus anthracis str. Ames Ancestor Bacillus cereus str. ATCC 14579 Bacillus clausii KSM-K16 Bacillus haiodurans str. C-125 Bacillus licheniformis str. ATCC 14580 Bacillus sp. NRRL B-14911 Bacillus subtilis subsp. subtilis str. 168 Bacillus thuringiensis serovar israelensisstr. ATCC 35646 Bacillus thuringiensis serovar konkukianstr. 97-27 Bacillus weihenstephanensisstr. KBAB4 Exiguobacterium sibiricum str. 255-15 Exiguobacterium sibiricum str. 255-15 Desulfitobacterium hafniense DCB-2

HemAT-Ba

AAT34830.1

Aerotactic

MCP

Firmicutes

433

BA_0532

HemAT-Bc

NP_835085

Aerotactic

MCP

Firmicutes

434

Tfu_0727

HemAT-Bcl HemAT-Bh

BAD66012.1 NP_241371

Aerotactic Aerotactic

MCP MCP

Firmicutes Firmicutes

449 439

– BH505

HemAT-Bli

AAU22675.1

Aerotactic

MCP

Firmicutes

430

–

HemAT-NRRL HemAT-Bs

ZP_01170456.1 NP_388919

Aerotactic Aerotactic

MCP MCP

Firmicutes Firmicutes

434 432

– YhfV

HemAT-Bti

ZP_00739958.1

Aerotactic

MCP

Firmicutes

434

–

HemAT-Btk

YP_039413.1

Aerotactic

MCP

Firmicutes

433

–

HemAT-Bwe

EAR75232.1

Aerotactic

MCP

Firmicutes

432

–

HemAT-EsA

ZP_00537959.1

Aerotactic

MCP

Firmicutes

430

–

HemAT-EsB

ZP_00540109.1

Aerotactic

MCP

Firmicutes

435

–

HemAT-Dh

ZP_00558369.1

Aerotactic

MCP

Firmicutes

415

–

Protoglobin and Globin-coupled Sensors

26

(Continued)

179

180

Table 1. (Continued) Organism

Name

NCBI accession #

Classification

Pfam

Taxonomic ID

39

Carboxydothermus hydrogenoformans str. Z-2901 Desulfotomaculum reducens str. Ml-1 Moorella thermoacetica str. ATCC 39073 Magnetococcus sp. MC-1 Silicibacter sp. TM1040 Acidiphilium cryptum str. JF-5 Bordetella bronchiseptica RB50 Bordetella parapertussis str. 12822 Bordetella pertussis Tohama I Burkholderia fungorum

HemAT-Ch

YP_359006.1

Aerotactic (truncated)

MCP

Firmicutes

250

–

HemAT-Dr

ZP_01147508.1

MCP

Firmicutes

251

–

HemAT-Mt

YP_429445.1

MCP

Firmicutes

245

–

MgGReg

EAN28412.1

ERERQR:GGDEF

-Proteobaceria

453

Mmc10355

SiGRegB

ZP_00620647.1

STAS

-Proteobaceria

308

–

AcGReg

ZP_01147165.1

881

–

NP_888505

UNK:GAF:EAL: UNK ERERQR:GGDEF

-Proteobaceria

BbGReg

-Proteobacteria

475

–

BpaGReg

NP_884745

Aerotactic (truncated) Aerotactic (truncated) Gene Reg (2nd Msgr) Gene Reg (Trx) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr)

ERERQR:GGDEF

-Proteobacteria

475

–

BpeGReg

NP_882025

ERERQR:GGDEF

-Proteobacteria

475

–

BfGReg

ZP_00277651.1

GAF:EAL

-Proteobacteria

723

Bcep2859

CvGRegA

NP_899909

Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr)

GGDEF

-Proteobacteria

375

–

CvGRegB

NP_900548

Gene Reg (Trx)

STAS

-Proteobacteria

295

–

40 41 42 43 44 45 46

47 48 49

50

Chromobacterium violaceum str. ATCC 12472 Chromobacterium violaceum str. ATCC 12472

Protein length

Other name

T.A.K. Freitas et al.

No.

Azoarcus sp. EbN1

AzoGReg

CAl07755.1

52

DpGReg

CAG34930.1

AdGReg

YP_464540.1

54

Desulfotalea psychrophila LSv54 Anaeromyxobacter dehalogenans str. 2CP-C environmental sample

EAJ19547

EAJ19547

55

environmental sample

EAK35357

EAK35357

56

environmental sample

EAK60420

EAK60420

57

EcGReg

NP_287665

AfGReg

TIGR contig 10428

SbaGReg

ZP_00582721.1

60

Escherichia coli O157:H7 EDL933 Acidithiobacillus ferrooxidans str. ATCC 23270 Shewanella baltica str. OS155 Shewanella sp. ANA-3

Sh3GReg

ZP_00852332.1

61

Shewanella sp. MR-4

Sh4GReg

ZP_00883268.1

62

Shewanella sp. MR-7

Sh7GReg

ZP_00854887.1

63

Shewanella putrefaciens str. CN-32 Shewanella sp. W3-18-1

SpGReg

ZP_00814636.1

ShGReg

ZP_00904891.1

53

58

59

64

Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr)

PAS:GGDEF: EAL GGDEF

-Proteobacteria

698

–

-Proteobacteria

363

–

HisKA: HATPase_c GGDEF

-Proteobacteria

379

–

environmental

353

–

GGDEF

environmental

403

–

GGDEF

environmental

360

–

ERERQR: GGDEF GGDEF:EAL

-Proteobacteria

460

-Proteobacteria

880

YddV EcDos –

GGDEF

-Proteobacteria

303

–

GGDEF

-Proteobacteria

403

–

GGDEF

-Proteobacteria

397

–

GGDEF

-Proteobacteria

397

–

GGDEF

-Proteobacteria

454

–

UNK:GGDEF

-Proteobacteria

454

– 181

(Continued)

Protoglobin and Globin-coupled Sensors

51

182

Table 1. (Continued) Organism

Name

NCBI accession #

Classification

Pfam

Taxonomic ID

65

Erwinia carotovora subsp. atroseptica SCRI1043 Shigella dysenteriae str. 1012 Shigella sonnei str. Ss046 Shigella flexneri serotype 2a str. 301 Escherichia coli str. E110019 Azotobacter vinelandii AvOP Reinekea sp. MED297

ErwGReg

CAG74587.1

Gene Reg (2nd Msgr)

GGDEF

-Proteobacteria

442

–

SdGReg

ZP_00922316.1

GGDEF

-Proteobacteria

336

–

SsGReg

YP_310564.1

GGDEF

-Proteobacteria

460

–

SfGReg

NP_707605

ERERQR:GGDEF

-Proteobacteria

381

YddV

EcGReg

ZP_00723852.1

GGDEF

-Proteobacteria

453

–

AvGReg

ZP_00415257.1

ERERQR:GGDEF

-Proteobacteria

464

–

ReGReg

ZP_01113979.1

Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr) Gene Reg (2nd Msgr)

-Proteobacteria

1090

–

Vibrio vulnificus str. CMCP6 Pelobacter propionicus DSM 2379 Desulfuromonas acetoxidans str. DSM 684 Geobacter metallireducens str. GS-15

VvGReg

NP_762059

-Proteobacteria

306

VV20073

PpGCS

ZP_00677957.1

Gene Reg (Trx) Unclassified

HAMP:GAF:HisKA: HATPase_c: REC:REC STAS four TM helices

-Proteobacteria

299

–

DaGCS

ZP_00550559.1

Unclassified

four TM helices

-Proteobacteria

301

–

GmGCS

ABB30581.1

Unclassified

four TM helices

-Proteobacteria

300

Gmet3020

66 67 68 69 70 71

72 73 74

75

Protein length

Other name

T.A.K. Freitas et al.

No.

77

78 79 80 81 82 83 84 85 86

87 88

Geobacter sulfurreducens str. PCA Thiomicrospira denitrificans str. ATCC 33889 Natronomonas pharaonis DSM 2160 Shigella boydii str. Sb227 Thermosynechococcus elongatusstr. BP-1 Thermus thermophilus str. HB27 Nitrosococcus oceani str. ATCC 19707 Methanosarcina acetivorans str. C2A Methanosarcina barkeri str. fusaro Chloroflexus aurantiacus str. J-10-fl Rubrobacter xylanophilus str. DSM 9941 Thermobifida fusca str. YX Aeropyrum pernix str. K1

GsGCS

NP_954351

Unclassified

four TM helices four TM helices

-Proteobacteria

300

GSU3311

TdGCS

YP_393641.1

Unclassified

-Proteobacteria

300

Tmden_1128

NpGCS

YP_326321.1

Unclassified

–

268

–

YP_408006.1

Unclassified

–

Euryarchaeota (Archaea) -Proteobacteria

SboGCS

240

–

TeGlb

NP_682779

–

Cyanobacteria

194

–

TtGlb

YP_005074.1

–

TTC1105

ABA57594.1

DeinococcusThermus -Proteobacteria

203

NoPgb

sensor globin sensor globin protoglobin

196

–

MaPgb

NP_617780

protoglobin

–

195

MA2883

MbPgb

AAZ71822.1

protoglobin

–

195

–

CaPgb

ZP_00768902.1

protoglobin

–

Euryarchaeota (Archaea) Euryarchaeota (Archaea) Chloroflexi

195

–

RxPgb

ZP_00600437.1

protoglobin

–

Actinobacteria

196

–

TfPgb

AAZ54765.1

protoglobin

–

Actinobacteria

197

–

ApPgb

NP_147118

protoglobin

–

Crenarchaeota (Archaea)

195

APE0287

–

Protoglobin and Globin-coupled Sensors

76

183

184

90% Consensus HemAT-Bs HemAT-Bh HemAT-Ba HemAT-Bc HemAT-Mg GmGCS GsGCS SfGReg EcGReg Bpe GReg BpaGReg BbGReg AvGReg HemAT-Hs EAJ19547 EAC21812 EAK60420 EAK35357 CaPgb MaPgb TfPgb ApPgb HemAT-MmB HemAT-MmA HemAT-At McpB McpM HemAT-Rs HemAT-Na VvGReg CvGRegB CvGRegA T.elongatus HemAT-Ch HemAT-Rr MgGReg AfGReg BfGReg

90% Consensus HemAT-Bs HemAT-Bh HemAT-Ba HemAT-Bc HemAT-Mg GmGCS GsGCS SfGReg EcGReg Bpe GReg BpaGReg BbGReg AvGReg HemAT-Hs EAJ19547 EAC21812 EAK60420 EAK35357 CaPgb MaPgb TfPgb ApPgb HemAT-MmB HemAT-MmA HemAT-At McpB McpM HemAT-Rs HemAT-Na VvGReg CvGRegB CvGRegA T.elongatus HemAT-Ch HemAT-Rr MgGReg AfGReg BfGReg

T.A.K. Freitas et al.

..........htt.t.t.ht.htthh.tt.t.h.t.FYt.h.t.tt...hhtt...........tt.ptp.tta.....tt.......tt |--Z---| |------A------| |-----B----| |----C---| |------E------| |-DVKKQLKMVRLGDAELYVLEQLQPLIQENIVNIVDAFYKNLDHESSLMDIINDHS-------SVDRLKQTLKRHIQEMFAGVI----DDE ELSAQLRMIHLTLDDLKRMKALQPLVEENMEVLADAFYSNIIKQPNLNEIIETHS-------SVERLKETLKQHILEMFNGEI----DQA ELKVQMDMLHISKEDLQIVKVLQPFIYEEIDWITEKFYANITKQPNLITIIERYS-------SIPKLKQTLKTHIKELFSGDM----HED ELKIQMDMLHISKEDLQIVKVLQPFIYEEIDWITEKFYSNITKQPNLITIIERYS-------SIPKLKQTLKTHIKELFSGDM----HEN HIEQMKRFVGFTEKDASILKKLRPVAAKHATAVVNTFYTRLSGFAHLEKIIGGAGS------SVERLKRTQEEYLVQLFDGEY----GRD SMQEIKAHYLFGDEDAETLKSLLSIAQANRELMIEDFYDYLLGIPETAAFLQDDT-------VLQRLKLSHGGWFVNLFRGVY----DNQ TMQEIKAHYRFTDEDAELLGSLFPLAETNKERLADQFYDYLLGIPETAEFLKEDL-------VLQKLKQTHQDWFVSLFAGSY----DNR KRMKDEWTGLVEQADPLIRAKAAEIALAHAHYLSIEFYRIVRIDPHAEEFLSNEQ-------VERQLKSAMERWIINVLSAQV--DDVER KRMKDEWTGLVEQADPPIRAKAAEIAVAHAHYLSIEFYRIVRIDPHAEEFLSNEQ-------VERQLKSAMERWIINVLSAQV--DDVER EILALRWKDTCAHYSPHEWVAARNVVTANKAALADYFYECMLADPNAAFFLSDQL-------VKTKLHASMQDWLESVYAAAP-TEEYER EILALRWKDTCAHYSPHEWVAARNVVTANKAALADYFYECMLADPNAAFFLSDQL-------VKTKLHASMQDWLESVYAAAP-TEEYER EILALRWKDTCAHYSPHEWAAARNVVTANKAALADYFYECMLADPNAAFFLSDQL-------VKTKLHASMQDWLESVYAAAP-TEEYER EQQAAEWKLLLGQFPAPVVAQIRELATTHQSELPGYFYEQMLQDEQAMLFLTHEQ-------VKSRLHGTLRQWIVSVFSMSDDDAALQA EIAWRLSFTGIDDDTMAALAAEQPLFEATADALVTDFYDHLESYERTQDLFANSTK------TVEQLKETQAEYLLGLGRGEY----DTE ***************************KRNRQIVDDFYGLQTSVSEIALLIGDSD-------TLARLRTAQRRYVLDLFNGVY----DLE *********RSGRSNQARLQKASKMVMALLPEVLDHFYDRVGREPEMAAFFKSDK-------MLERAKGEQLKHWSRLFSGEY----CED EIAHRKELLLLDERDFALLASYRPKIEPHIDALVDKFYTLQTGITEIALLIGDAD-------TLTRLRAAQRRYILDLFSGLY----DLE EIDFRKSLFSFTLADVRALQSFKPVIEENIDKIVDDFYGLQTSVSEIALLIGDSD-------TLARLRTAQRRYVLDLFNGVY----DLE EWELLKQTVLWTAEDEQYLRMAGEVLGDQVDAILDLWYGFVASHAHLVYYFTGSD-GQPIADYLSRVRQRFGQWILDTCRRPY----DQD DLKLLKEAVMFTAEDEEYIQKAGEVLEDQVEEILDTWYGFVGSHPHLLYYFTSPD-GTPNEKYLAAVRKRFSRWILDTCNRSY----DQA DLNKLKTTVMFTSADEEALRMAGDVLEDQVEDVLDVWYGFVADHPHLLAYFSTPD-GHPIQEYLDRVRERFGQWILDTCRRPY----NQE EFDLLKKTVMLGEKDVMYLKKACDVLKDQVDEILDLWYGWVASNEHLIYYFSNPDTGEPIKEYLERVRARFGAWILDTTCRDY----NRE KNRERLRFLRLDDDAISTVKSVRQMVESSLPGIADGFYAHLMQWPALKALLGGGA------KIGHLKETQQAHWAS-LFSGRF----DDD DRTSSIAFLQIDEDTKRALREFREVLSRHIDGVLDTFYRHVSNNPATAKMFANPD------RMAHARSMQKKHWMESVFLGQF----DDR QLDERLNFLGLGHGERQNLSDMKGVITGSLDASLDRFYTKVRAVPETAKFFSSEA------HIHHAKSMQLKHWSR-IASGTF----NED AIGERTAFMGIDDKARSALRDLRPVIRAEIGKALDNFYGKVRATPETRKFFSDDR------HMNAASSRQQAHWGV-IAEGQF----SDD KLDQRMAFMRFDERSRAHLRAIKPVIDAEIGAALGQFYSQVRLFPDTRVKFRDDG------HMAGAERAQAAHWRR-IAEAGY----GES LLDERLRFLGIG-RDDALDDATRALLNEAVGRALDRFHERMRQTSAAG-FFADAT------HMDSAKSRQARHWAR-LASGEI----DAA DVARKLAFFNIDHKDFERFPHIAKVLENYAPPALDKLYDQIATTPETASFFGSRQ------AMRHARDKQIEHWAG-MFSGRA----DRS DADELLKLHDLTEADLALIRKFGQIMVPKLDEYVKHFYDWLRNTPEYEQYFGDAQ------KLQRVQDSQVRYWKT-FFDARI----DSA EAGEFLHTFAIQEDDLKRVRAMGEAVLPRLDEAMDRFYEWLPSLPEYEGLFARPS------ALRNAREAQAAYWRS-FFSGVV----DAA EIEQRKHLFALAPRDEMLLRAAGNLVESHLEELVTRFYELQTSTPEIALLIGDAD------TLQRLRSAQRRYVVD-LFSGIY----DLE FMATMVRRVQLTDEDKSLLAEAAPWGKEIAPQMADTFYDYLGRDEEMNAILNATEG------RIHRLHQTFVDWFYEMFTGMDS--WGKA RYQETLSFLNLTAEDLQLMAEFKELFIQKAQEFVNKFYQHLTKFPYLQELIKKHS-------TVEKLSKTQAEYFISLTSEKI----DAD MPNPTLSLLKADARLTEDLNEIHPLMVSMIDDLLGEFYDTVSRTPELYAMFGSAQS------VERARLAQRRHWVEVLFKGDW-----KA ARDEQRLKDIYLGVDAEKVNFIGDLIKDRLNQTVERFYIELLEVESARFFLDSAL-------VKERLHGSLTEWLQMLFSHKD------D NSGTLPAFLGLQDSDFQVIDRYRDALDKEASALAHAFYDYLLSHPATAAVFRDFSS-----ARLDALIQKQTEHAKGLLVSRL----DRP NDLMDDGSHLYSQARASALTSLTEVLRHNAVEIVKRFYDGLIRLPKSKHTLAALS--------EHELQHLKTQQIQNLYALASPDLTAMD ψ #

h.t...thG..H.p.t.t.p...tuh..h.......h..t...................tt.........c....t..h..t.a.....t ------F------| |---------G----------| |---------H----------| FIEKRNRIASIHLRIGLLPKWYMGAFQELLLSMIDIYEASITN---------------QQELLKAIKATTKILNLEQQLVLEAFQSEYNQ FLQKRLQIAQAHVRIGLQTKWYVSAFQQLTDSLIQLLEQHLQS---------------PSDIVLATRSLLKLLNLEQQLVLEAYENENKR FIEQRVKIAKRHVQIGLHRKWYTAAYQELFRSIMKILKTKITT---------------IDDFSYSINVINKLFTLEQELVIAAYESKYER FIEQRVRIAKRHVQIGLHRKWYTAAYQELFRSIMKILKTKITT---------------IDDFSYSINVINKLLTLEQELVIASYESEYER Y---FWRIGQIHNKIGLEPDWYLGGYSLYRQLLLPILLDVFDNK--------------PKKVQRAMAAIDKILTLDSELAIGSYIDAVMA YLHDLQRVGHVHVKIGLNAHFVNAAMQKVRRFAVGMIRENFPD---------------RDERRKKTEAVEKILDINLDIMTASYIEEELK YIHNLQKIGHAHVRVGLNAHYVNVAMNVVRQFTLSIIQDNFPD---------------PEERRQRREAVEKILDINLDIMSASYREEEMR LIQIQHTVAEVHARIGIPVEIVEMGFRVLKKILYPVIFSSDYS---------------AAEKLQVYHFSINSIDIAMEVMTRAFTFSDSS LIQIQHTVAEVHARIGIPVEIVEMGFRVLKKILYPVIFSSDYS---------------AAEKLQVYHFSINSIDIAMEVMTRAFTFSDSS TVAFQRKVGEVHARIDIPVHLVTRGACALIRRICELLDRDASLS--------------AAQAAATCRYVADVTMTAVEMMCHAYSVSHDR TVAFQRKVGEVHARIDIPVHLVMRGACALIRRICELLDRDASLS--------------AAQAAATCRYVADVTMTAVEMMCHAYSVSHDR TVAFQRKVGEVHARIDIPVHLVMRGACALIRRICELLDRDASLS--------------AAQAAATCRYVADVTMTAVEMMCHAYSVSHDR LIAQQKQIGEIHARIKIPIHLVLRGARHLRERLFVLL-RQRPLD--------------PEHKLFGQRLISETVDLAMEIMSRAFSDAYDR YAAQRARIGKIHDVLGLGPDVYLGAYTRYYTGLLDALADDVVADRGEEAAAA------VDELVARFLPMLKLLTFDQQIAMDTYIDSYAQ YVNNRLRIGLVHKRIGVEPKLYLSAVHTLKELIYAEINNSVKD---------------AAQNERIRIAIDKLVLFDVTLVFDTYIRSLVS YQSSARNIGQVHTRIGLPFAFFNAGYAHANAQIQALILKRQTGGLFRR----------ASETHILLGILSRAMALDIQLIFDAHAEAVQE YVNNRLRIGLVHKRIGVEPKLYLAAINTLKGLLIEDIFTQIDH---------------EPDRITMLTALDKLFLFDITLVFETYIRSLVS YVNNRLRIGLVHKRIGVEPKLYLSAVHTLKELIYAEINNSVKD---------------AAQNERIRIAIDKLVLFDVTLVFDTYIRSLVS WLNYQMEIALRHYRTKKNQTDGVQSVPMIPLRYMIAFIYPITATIREFLARKGHS---AAEVDRMHQAWFKSIVLQVTLWSYPYTREGDF WLDYQYEIGLRHHRTKKNQTDNVESVPNIGYRYLVAFIYPITATMKPFLARKGHT---PEEVEKMYQAWFKATTLQVALWSYPYVKYGDF WLDYQQEIALRHTPEKKNVTDNANSVDNIPLRYVIAFIYPVTATLRPFLAKKGHS---ADQVEAMYQAWFKSVTMQIALWSQPYTRDGYW WLDYQYEVGLRHHRSKKGVTDGVRTVPHIPLRYLIAFIYPITATIKPFLAKKGGS---PEDIEGMYNAWFKSVVLQVAIWSHPYTKENDW YFTRAVAIGAAHERIGLEVNWYLGGYCFVLEKLMAELHAKCE----------------KARFPQMAGAVLRAAFLDMDLAISTYIEHGEA YFAQVTEIGKVHQRIGLDPKWYTAGYCFVLNMVIGVAVEHYRKD--------------PKRLTQVLAAVNKAAFLDMDLATSVYIETNTA YTNAVTAIGRTHARLGLEPRWYIGGYALMLDGIVKAVIESELKGLFMEKK--------AKKVKDALSATIKAALLDMDYSISVYLDVLAT YVQAVRAIGQTHARIGLEPRWYIGGYAVVGDHLVRAVIDSMWPRGLLAKGG-------SDRAGEAVAALMKAIFLDMDFAISIYLETLEN YVRDVERIGRSHADADIAPQWYIGGYAVVVEEVMRALVAKRAKGLFNSAKS-------DAELADGLSALIKAAFLDMDLSVSTYIDVLLE YVEEAVRVGRTHARIGLEPRWYLGGYALILEEIVQTMLPRMAGRGFFGRRR-------ATRAAHALGYIVKVALLDMDYGVSTYFDAVQS YFESAERIGNVHARIGLEPGWYIGGYAMVLEQVINAMFSGIGILG-------------AKRTARSVGSLVKMALLDMEVALSTYFRAEEA YLKERRDVGEIHARVGLPLPTYFAGMNISMVIFTKRMYDGSLY---------------SDEYSSLVTAFTKLLHLDTTIVVDTYSRLINK YLAERVCAGETHARIGLPLSSYFAGVNYAFTLFCGYLKSGS-----------------RETASQTLLSTAKLLHMDTALVVETYSRLLHE YVNIRLRIGLVHKRIGVEPKLYLAAVDSLKFLLAEKLTELIPD---------------AEVRLHTLQALDKLMMFDVALVFETYIRSLVA YAERRWKIGLVHVRIGIGPQHVVPAMAVVVNAVRQKLREANKSEALSDALGKICMIDLAFIEQAYFEVSSQAVLKETGWTQALFQRLIAT YIKNRLAVGKKHMEIALYPNWYIGAYRLYYEVVGELVARKYSPG--------------TELYFKAVNAFYKRINFDIQLAIENYIAEQLK HASQAQRIGKAHVDRGITPSIYFAAYSHVLCGLTGRMAQAKGLR--------------SEALARGLRAAIRAVYIDMLAVLDVYFAEERD DTLEQKNIGNVHARINIPMHLVVEGMRILRREIICFLSESDIPRQR------------LVDLVVLVGEVLDHNLSLINESYVRMSASYER WRESMRKIGALHHHLGIGPSWIAGAYILYWRHWQKILQVQVP----------------ESDRDLLRDALFRLLVGDLMVQLEGYAHASRE HRTMALRVGRIHAIVGLEWEDLIRSRGILSAAIHDTLDTTVHGIALAVLGRRLTQDLAWQTEAYQRLQTSRQDVLMRVTQL-AWEVESYT †

Protoglobin and Globin-coupled Sensors

185

Fig. 1. Sequence and evolutionary relationship amongst the globin-coupled sensors and protoglobins. (A) Sequence alignment of representative GCS globin domains and the protoglobins. The 90% consensus sequence is represented in the following letter code: hydrophobic = “h” (ACLIVMHYFW); turn-like = “t” (ACDEGHKNPQRST); aromatic = “a” (FHWY); polar = “p” (KRHEDQNST); charged = “c” (KRED); and tiny = “u” (AGS). Regions of incomplete sequence information are marked with an asterisk (∗ ). The proximal histidine is indicated with a dagger (†), the E19 cysteine residue with the pound sign (#), and the distal B10 tyrosine with psi (). The globin secondary structure indicated at the top follows that of the HemAT-Bs sensor domain (PDB: 1OR4). Environmental sequences EAJ19547, EAC21812, EAK60420, and EAK35357 were identified from genome shotgun sequencing of the Sargasso Sea.

of elemental recycling on our planet, are currently being sequenced by the Department of Energy. In particular, many organisms are involved in carbon and nitrogen sequestration, bioremediation, and cellulose degradation, in addition to promising roles in biotechnological applications such as magnetite production and biomineralization (see http://doegenomes.org/ for more information). While some of the organisms in Table 1 live in temperate environments, others have adapted to more extreme environments. Accordingly, some GCSs and protoglobins are hyperthermophilic (95 C; Aeropyrum pernix), psychrophilic (−25 C; Exiguobacterium sibiricum, Desulfotalea psychrophila), halophilic (4 M salt; H. salinarum, Natronomonas pharaonis, Haloarcula marismortui), alkalophilic (pH 11; N. pharaonis), and acidophilic (pH 1.2; Acidithiobacillus ferrooxidans), and will likely have broad academic and biotechnological applications. The protoglobin alignment with various globins is shown in Fig. 1 and Fig. 3.

3.1. Aerotactic The HemATs are the only GCSs whose physiological effects have been experimentally studied and are the sole members populating the aerotactic group. HemATs are thus far characterized by an N-terminal sensor globin with or without a HAMP domain, followed by a C-terminal MCP domain (Fig. 5A). The latest searches have turned up three new archaeal HemATs in two organisms. The first is a HemAT from H. marismortui, a close relative of H. salinarum [33] exhibiting HAMP:MCP domains C-terminal to the sensor globin. The other HemATs were found in the archaeon N. pharaonis DSM 2160 and one is very unique – it is the first HemAT discovered that couples a sensor globin with a PAS:PAS domain combination. PAS domains are known to bind various cofactors, including heme, and sense a variety of ligands [34]. This extra sensing capability may aid in this archaeon’s task of dealing with both high osmolality (3.5 M NaCl) and high pH (lake pH ∼11). Characterized by time-lapsed capillary assays (Fig. 5B), HemATs were shown to elicit either positive or negative aerotaxis [1] by interacting with the downstream chemotaxis machinery (Fig. 2A). Whether this positively or negatively biased response originates from the N-terminal, C-terminal, or interdomain region is not yet clear. In E. coli, a clockwise (CW) rotation of the flagella results in a tumbling event associated with negative aerotaxis. It is possible to determine the tumbling frequency and hence, aerotactic response of a population of E. coli or B. subtilis resulting from the photorelease of O2 from a molecular cage [35] by noting their rate of change of direction (rcd) calculated

Gene Regulating

Aerotactic Methylation

HemAT-Si A HemAT-Bra HemAT-Nh HemAT-Rp HemAT-Oa A HemAT-Oa B HemAT-Rs HemAT-Go HemAT-Mm A HemAT-Mm B HemAT-Rr HemAT-Zm EAC21812 HemAT-Hs HemAT-Np B HemAT-Ba

nd 2 Messenger

DNA-binding

HemAT-Bc HemAT-Bcl HemAT-Bh HemAT-Bli HemAT-NRRL HemAT-Bs HemAT-Bti HemAT-Btk HemAT-Bwe HemAT-Es A HemAT-Es B HemAT-Dh HemAT-Ch HemAT-Dr HemAT-Mt

Transcription Regulator

FixL

sGC

CooA

Cv GRegB Si GRegB Vv GReg

AxPDEA1 NPAS2 Dos

Ad GReg Re GReg

Mg GReg Bb GReg Bpa GReg Bpe GReg Sf GReg Ec GReg

Af GReg HemAT-Mg HemAT-Si B McpB McpM HemAT-At

186

Biological Heme-based Sensors

HemAT-Re HemAT-Na HemAT-Hm HemAT-Ac

Dp GReg Cv GRegA EAJ19547 EAK35357 EAK60420

HemAT-Np A

Av GReg Sp GReg Sh GReg Erw GReg Sd GReg Ss GReg

Membrane-bound/ Unknown function:

AzoGReg

Pp GCS Gs GCS Da GCS Td GCS Gm GCS

Ac GReg Bf GReg

SbaGReg Sh3GReg Sh4GReg Sh7GReg

SMART domains:

cNMP

GAF

HLH

CYCc

GLOBIN*

HTH CRP

DUF1 (GGDEF)

HAMP

MA

REC

HATPase_c

DUF2 (EAL)

transmembrane helix

PAC

PFAM: STAS

PAS

ERERQR*

HisKA

T.A.K. Freitas et al.

Fig. 2. Classification schema of biological heme-based sensors. Heme-based sensors and their domain organization are illustrated. Individual globin-coupled sensors are assigned to their respective class on the basis of the known/putative functions of their signaling domains. The name ERERQR is a name given to the domain between the globin and GGDEF (DUF1) domain and based on the ERERQR motif it contains [16]. See Table 1 for the source organisms of each GCS.

Protoglobin and Globin-coupled Sensors

187

with the use of motion tracking software. This method, in conjunction with molecular methods, has the potential to isolate regions of the HemAT responsible for the rotational bias simply by measuring the increase or decrease in the rcd. More than one bacterium, however, has been identified with multiple HemATs encoded in the genome. In particular, Magnetospirillum magnetotacticum possesses two HemATs and is known to exhibit temporal O2 sensing capabilities, whereas Magnetococcus MC-1 cells, possessing only one HemAT, do not [36]. HemATs and their role, if any, in magneto-aerotaxis have yet to be investigated.

3.2. 2nd Messenger Thirty GCSs were included in this category on the basis of the functions identified by their individual domains (Table 1, Fig. 2). Homologous to the adenylyl cyclase catalytic domain, the GGDEF domain has been found in proteins modulating cyclic diguanosine monophosphate (c-diGMP) turnover and phosphodiesterase (PDE) activity [37], while the EAL domain has been implicated in diguanylate phosphodiesterase function [38]. In Gluconacetobacter xylinus (formerly Acetobacter xylinum), the AxPDEA1 protein (GGDEF:EAL domains) is a heme-binding protein that regulates cellulose production in response to cellular O2 levels by linearizing c-diGMP, while the GGDEF protein PleD regulates cellular morphology in Caulobacter crescentus. The GCS from A. ferrooxidans has the exact signaling domain organization as heme-PAS proteins EcDos (aka YddV) and AxPDEA1 [31] and may possess a similar function. Expression of EcDos has been shown to be oxygen dependent, with little protein expressed in anaerobic conditions. Aerobic, but not anaerobic growth of an EcDos knockout strain increases intracellular cAMP levels in vivo with a concomitant change in cell morphology (filamentation) [39]. Overexpression of EcDos, however, reduces the cAMP levels, forming mini cells. In vitro cAMP turnover rates are physiologically low [31,32], suggesting that the in vivo activity of EcDos requires additional enhancement factors [39]. Recently, reverse transcriptase-PCR experiments have provided strong evidence supporting the cotranscription of the genes for EcDos and the E. coli generegulating GCS, EcGReg (YddV), in a single, bicistronic mRNA [40]. Additionally, IPTG-induced expression of EcGReg in wild-type E. coli K-12 bearing an EcGReg plasmid resulted in an increased production of c-diGMP with a concomitant upregulation and downregulation of more than 50 and 27 genes, respectively, followed by a change in cellular morphology (filamentation). Whereas EcDos degrades c-diGMP [31], EcGReg is believed to increase its production in vivo [40]. This may explain the extremely low rates of EcDos-dependent cAMP degradation measured in vitro [32] and would imply that the effects observed on cAMP are downstream effects stemming from c-diGMP metabolism. The GCS from Burkholderia fungorum exhibits a GAF:EAL domain organization, where GAF domains are nucleotide-specific cAMP- and cGMP-regulating domains [38,41,42] with broad cellular roles as far reaching as in the human rod photoreceptors [43]. Perhaps the most interesting additions to this category are the GCSs from the -Proteobacterium Azoarcus sp. EbN1 (AzoGReg) and the -Proteobacterium Reinekea sp. MED297 (ReGReg). Azoarcus sp. EbN1, like the HemAT from N. pharaonis, expresses a PAS:PAS domain just C-terminal to the sensor globin, but unlike N. pharaonis, Azoarcus possesses the GGDEF and EAL domains. As mentioned earlier, EAL

Physter catodon Aquifex aeolicus Tgb Nitrosococcus oceani Pgb Methanosarcina acetivorans Pgb Methanosarcina barkeri Pgb Chloroflexus aurantiacus Pgb Rubrobacter xylanophilus Pgb Thermobifida fusca Pgb Aeropyrum pernix Pgb Thermus thermophilus Thermosynechococcus elongatus

1 1 1 1 1 1 1 1 1 1 1

50% Consensus Physter catodon Aquifex aeolicus Tgb Nitrosococcus oceani Pgb Methanosarcina acetivorans Pgb Methanosarcina barkeri Pgb Chloroflexus aurantiacus Pgb Rubrobacter xylanophilus Pgb Thermobifida fusca Pgb Aeropyrum pernix Pgb Thermus thermophilus Pgb Thermosynechococcus elongatus

38 38 70 69 69 69 70 71 68 71 63

50% Consensus Physter catodon Aquifex aeolicus Tgb Nitrosococcus oceani Pgb Methanosarcina acetivorans Pgb Methanosarcina barkeri Pgb Chloroflexus aurantiacus Pgb Rubrobacter xylanophilus Pgb Thermobifida fusca Pgb Aeropyrum pernix Pgb Thermus thermophilus Pgb Thermosynechococcus elongatus

99 99 138 137 137 137 138 139 137 134 128

.M....IPGYTYG...V.+SPh..LE-hcLLK.pVMFTEEDEcYL+.AGEVLcDQV-EhLD.WYGFV..SHP |--Z---| |------A------| |-----B----| ...................................MLSEGEWQLVLHVWAKVEADVAGHGQDILIRLFKSHP ...................................MLSEETIRVIKSTVPLLKEHGTEITARMYELLFSKYP .MGEKEIPGYTYGTQAVAKSPVS-LEDFDLLKKTVLFTEEDEKYLRLAGEVLGDQVEEVLDLWYGFV-GSHP .MSVEKIPGYTYG-ETENRAPFN-LEDLKLLKEAVMFTAEDEEYIQKAGEVLEDQVEEILDTWYGFV-GSHP .MSIEKIPGYTYG-KTESMSPLN-LEDLKLLKDSVMFTEEDEKYLKKAGEVLEDQVEEILDTWYGFV-GSHP .MSEA-IPGYTYGTAQVAQSPVS-LEEWELLKQTVLWTAEDEQYLRMAGEVLGDQVDAILDLWYGFV-ASHA .MAEAGIPGYAYGAREVARSPVS-LEELDLLRQTVLFTGEDERYLRMAGEVLEGRLDELLDVWYGFV-ADHS MATKTLIPGYTYGTEQVAKSPIG-LEDLNKLKTTVMFTSADEEALRMAGDVLEDQVEDVLDVWYGFV-ADHP .MTPSDIPGYDYG--RVEKSPITDLEFDLLKK-TVMLGEKDVMYLKKACDVLKDQVDEILDLWYGWV-ASNE .MRGPASPAWTGRARDRYICPVDPGELLDLLKRRTGFTEAHAALLRELGEVMVPIAHEVALAFYDYL-GRDP .........MVIQSFEVKKMTIEPINFMATMVRRVQLTDEDKSLLAEAAPWGKEIAPQMADTFYDYL-GRDE † HLhhYFp..c-GpP..cYLcRVRpRF.pWILD...TCpR.YDQcWLpYQ.EIGLRHaRpKKN.TD.VppV.. |----C---| |------E------| |--------F------| |------G-ETLEKFDRFKHLKTEAEMKASEDLKKHGVTVLTALGAILKKKGHHEAELKPLAQSHATKHK----------KTKELFAGASEEQP-----KKLANAIIAYATY------IDRLEELDNAISTIARSHVRRNVKPEHYPLVKEC HLVRYFSDLQ-GEPDSSYLAAVRKRFAQWILD---TCNRTYDQDWLNYQHEIGLRHYHTKKNKTDNVQSVPI HLLYYFTSPD-GTPNEKYLAAVRKRFSRWILD---TCNRSYDQAWLDYQYEIGLRHHRTKKNQTDNVESVPN HLLYYFTSPD-GTPNEEYLAAVRKRFSKWILD---TCNRNYDQAWLDYQYEIGLRHHRTKKNRTDNVESVPN HLVYYFTGSD-GQPIADYLSRVRQRFGQWILD---TCRRPYDQDWLNYQMEIALRHYRTKKNQTDGVQSVPM HLVYYFSSPE-GEPIQEYLERVRERFKRWVLD---ACRRPYDQEWLDYQQEIALRHTREKKNRTDGVEAPEE HLLAYFSTPD-GHPIQEYLDRVRERFGQWILD---TCRRPYNQEWLDYQQEIALRHTPEKKNVTDNANSVDN HLIYYFSNPDTGEPIKEYLERVRARFGAWILD---TTCRDYNREWLDYQYEVGLRHHRSKKGVTDGVRTVPH ELGALLHAEP-GRV-----ERLYRTFARWYGE---LFSGVYDRAYAERRRRIGLVHARLGIGPRAMIPAMGI EMNAILNATE-GRI-----HRLHQTFVDWFYE-MFTGMDSWGKAYAERRWKIGLVHVRIGIGPQHVVPAMAV * I.hRYLhhFIYPITATh+PFLAcKGHp.c-VccMaQAWFKhhhLQVhLWS.PY.cpG.a........... ----G------| |---------H----------| IPIKYLEFISEAIIHVLHS-----RHPGDFGADAQGAMNKALELQRKDIAAKYKELGYQG.......... LLQAIEEVLNPG---------------EEVLKAWEEAYDFLAKTLITLEKKLYSQP.............. ISYRYLITFIYPITATIKPFLEKKGHSAEEVEKMHQAWFKSLLLQVTLWTNPYLRQEDY........... IGYRYLVAFIYPITATMKPFLARKGHTPEEVEKMYQAWFKATTLQVALWSYPYVKYGDF........... INYRYLVAFIYPITATIKPFLARKGHTSEEVEKMHQAWFKATVLQVALWSYPYVKQGDF........... IPLRYMIAFIYPITATIREFLARKGHSAAEVDRMHQAWFKSIVLQVTLWSYPYTREGDF........... VSLRYMISFIYPITATVRPFLEEGGRPAEDVEKMHQAWFKAVVLHVTLWSQPYAREGSF........... IPLRYVIAFIYPVTATLRPFLAKKGHSADQVEAMYQAWFKSVTMQIALWSQPYTRDGYW........... IPLRYLIAFIYPITATIKPFLAKKGGSPEDIEGMYNAWFKSVVLQVAIWSHPYTKENDW........... VQELSLEHMRMA-------------LRGHEVYSAVEAFEKLVAMEVALIEESYLEALSLGLSLGHRDLTQ VVNAVRQKLREANKSEA-LSDALGKICMIDLAFIEQAYFEVSSQAVLKETGWTQALFQRLIATGAAAM..

37 37 69 68 68 68 69 70 67 70 62

188

50% Consensus

98 98 137 136 136 136 137 138 136 133 127

153 139 196 195 195 195 196 197 195 191 194

T.A.K. Freitas et al.

Fig. 3. Protoglobin alignment. Identical residues (>50%) are shaded black, whereas similar residues are shaded gray. In addition to the protoglobins, the thermoglobin (Tgb) from Aquifex aeolicus and the Physter catodon (sperm whale) myoglobin are included for comparison. The structural designation is according to the HemAT-Bs structure (PDB ID: 1OR4). The key to the 50% consensus sequence is as follows: “h” = hydrophobic, “a” = aromatic, “+” = positively charged, “−” = negatively charged, “c” = charged, and “p” = polar.

Protoglobin and Globin-coupled Sensors

189

domains linearize cyclic di-GMP into 5 -GMP via its PDE activity, whereas the GGDEF domain either cyclizes two GTP molecules into one c-diGMP or regulates the PDE activity of the EAL domain itself [44,45]. It is not known whether coupling of the GCS to a PAS:PAS domain adds another level of regulation to the protein or whether additional functions are conferred. Within the 1090-aa long Reinekea GCS are seven identifiable domains (sensor_ globin:HAMP:GAF:HisKA:HATPase_c:REC:REC). HATPase_c is a histidine kinaselike ATPase, which binds and hydrolyzes ATP. Coincidentally, REC (receiver) domains contain a phosphoacceptor site that is phosphorylated by histidine kinase homologs (HisKA) and forms homodimers. Although the Reinekea GCS exhibits the HisKA and HATPase_c domains like FixL, it is classified in this category solely on the presence of its GAF domain.

3.3. Protein–Protein Interactions Four proteins currently populate this subdivision of GCSs. The -Proteobacterium Anaeromyxobacter dehalogenans is a metabolically versatile facultative anaerobe with efficient mechanisms for bioremediation: it reduces important oxidize metals such as Fe(III) and U(VI) as well as halogenated compounds by using them as terminal electron acceptors [46–50]. In addition, these activities are not dependent on the redox conditions and A. dehalogenans. In fact, A. dehalogenans is capable of using acetate and hydrogen as a source of electrons; however, the myriad of genes involved in regulating such processes is currently unknown. The domain organization of the A. dehalogenans GCS (AdGReg) resembles that of the ubiquitous two-component signal transduction system; its signaling domain consists of a catalytic HATPase_c domain and a HisKA histidine kinase domain and probably functions as a sensory histidine kinase like FixL. Two-component sensory histidine kinases typically autophosphorylate at a conserved His residue, followed by a transfer of that phosphate to the Asp residue on the response regulator responsible for effector functions. In FixL, the deoxy form autophosphorylates and the phosphate is then transferred to its response regulator, FixJ. FixJ in turn acts as a transcriptional activator, upregulating the expression of key nitrogen fixation genes [28] whose products are particularly sensitive to O2 . Oxygen binding to FixL at the heme-PAS N-terminal domain thus prevents autophosphorylation and hence, unphosphorylated FixJ cannot initiate transcription of the nitrogen fixation genes [28,51]. Currently, neither the genes regulated by AdGReg nor its response regulator are known. Three GCSs have been identified with a C-terminal STAS domain (sulfate transporter and antisigma-factor antagonist). Found in Vibrio vulnificus, Chromobacterium violaceum, and Silicibacter sp. TM1040, these proteins may regulate gene expression by functioning as an anti-antisigma factor (antisigma-factor antagonist or ASA) similar to SpoIIAA in B. subtilis spore formation [52]. ASAs, in their unphosphorylated form, positively regulate sigma factors (and hence, gene transcription) by directly binding the antisigma-factor (protein kinase) and are, in turn, phosphorylation inactivated at a conserved serine residue by the kinase activity of the antisigma-factor; active ASA is then regenerated by dephosphorylation by a phosphatase [52,53]. STAS domains are also known to function in anion gating [52,54]. Ko et al. [54] have shown that the regulatory (R) domain of the cystic fibrosis transmembrane conductance regulator

190

T.A.K. Freitas et al.

(CFTR), enhanced by PKA-mediated phosphorylation, interacts with the STAS domain of the SLC26 transporter in epithelial cells to enhance chloride-bicarbonate exchange primarily by increasing the open probability of the channel. Loss-of-function STAS domain mutants prevented activation of CFTR, resulting in congenital chloride diarrhea (CLD) [54,55]. In both cases, the STAS domain elicits its regulatory function by a direct protein–protein interaction.

3.4. Unclassified All GCSs identified thus far are soluble except for the small group of membrane-spanning (∼four transmembrane helices) GCSs found in the anaerobic -Proteobacteria Geobacter sulfurreducens, Geobacter metallireducens, Pelobacter propionicus DSM 2379, and Desulfuromonas acetoxidans, and the -Proteobacterium Thiomicrospira denitrificans [16]. The 3D-PSSM fold-recognition server [56] has identified the transmembrane region as a possible S-nitrosoglutathione (GNSO) reductase domain, which offers protection from nitrosative stress; however, their true function is currently unknown. The recent sequencing of the G. sulfurreducens genome [57] has revealed evidence of aerobic metabolism; however, any involvement of the GCS in aerobic metabolism has yet to be shown. Other GCSs have been labeled unclassified either due to what appears to be partial C-terminal signaling domains or lengthy sequences with no recognizable domain. With the appearance of more and more domain variance within the GCS family and the realization of c-diGMP-dependent signal transduction in bacteria, it is becoming apparent that the GCSs play a significant role in the normal functioning of cells.

4. BIOPHYSICAL AND KINETIC CHARACTERISTICS 4.1. Sequence and Structure The globin fold is but one three-dimensional structure successful at preventing rapid oxidation of the heme iron; however, it is currently the only other biological structure, in addition to the PAS domain, that is capable of reversibly binding O2 to the heme iron. Like the PAS domain, globin sequences are very divergent, and similarities can be as low as 10% when comparing mammalian globins with their bacterial relatives, presenting a problem for protein domain–recognition algorithms like SMART and Pfam [16]; NCBI’s conserved domain database (CDD), however, now has the capacity to distinguish these sensor globins (cd01068) from the globin (cd01040) and truncated hemoglobin (cd00454) domains within the globin-like (cd01067) superfamily [58]. Ultimately, the only residue conserved throughout the globins is the heme-binding proximal His, although potential homologs have been identified with Glu, Asn, and Tyr substitutions at the proximal His site (unpublished results). With the wealth of sequences available, identification of the proximal histidine in newly discovered globins should be relatively straightforward and provide a starting point for sequence and modeling studies. Distal residues vary [5], but most microbial globins retain a tyrosine residue at the B10 position that has been shown to be vital to their oxygen-binding properties [4,16,59–65].

Protoglobin and Globin-coupled Sensors

191

A resonance Raman study [10] on HemAT-Bs indicates that a strong hydrogenbonding interaction exists between the proximal oxygen atom of the bound O2 and a distal residue, and the subsequent 2.15 Å structure of the HemAT-Bs sensor domain (residues 1–178) shows that the B10 tyrosine may be responsible. The ferric cyanoliganded structure shows the B10 tyrosine directed toward the heme and hydrogen bonded to the ligand, whereas in the unliganded structure, the tyrosine is directed away from the heme, more disordered, and devoid of any ligand-stabilizing interactions [14]. Whether the structural changes observed in the ferric-liganded structure are similar to that of the physiologically relevant ferrous O2 -liganded form [1] is not yet known.

4.2. Aggregation State Light scattering (unpublished results) and gel filtration [10] results have previously indicated that recombinant full-length HemAT-Hs and HemAT-Bs are homotetramers, whereas the globin domain alone purifies as a homodimer. More recently, however, Zhang et al. [11] have indicated potential solubility and pH-dependent oligomerization problems during purification of the HemAT-Bs C-terminal signaling domain, a behavior seen in the purification of the cytoplasmic portion of the E. coli Tar receptor [66]. Analytical ultracentrifugation experiments have shown that in the range of 2–20 M, full-length HemAT-Bs purifies as a homodimer (98.292 kDa) and from its large frictional coefficient, has a highly asymmetric rod-like shape not unlike other bacterial chemotaxis receptors [11].

4.3. Stability Thermal, ionic, and pH tolerance ranges of three full- and minimal-length GCSs have been investigated (unpublished results) by monitoring their ability to retain bound O2 and the heme cofactor itself. In all experiments, the aerotaxis transducers HemAT-Bs and HemAT-Bh behaved similarly while HemAT-Hs behaved differently from them. In addition, the single-domain sensor globins alone were, in general, more stable under experimental conditions than their full-length counterparts. HemAT-Bs and HemAT-Bh display a higher degree of pH and thermotolerance at low ionic strength (0–0.2 M NaCl) as opposed to a high ionic strength (3.2 M NaCl). Aggregation and heme loss occurs as the pH drops to 4.0 and 4.5 for these low and high ionic strengths, respectively, presumably due to protonation of the proximal histidine. At the other extreme, significant denaturation is evident after pH 11.5. HemAT-Hs, on the other hand, shows lower pH and thermotolerance at these low ionic strengths and a higher tolerance at the higher ionic strength, as expected of a halophilic protein. Full-length HemAT-Hs aggregates at pH 5 and 3.5 for low and high ionic strength solutions, compared to pH 3.5 and 3.0 for low and high ionic strength truncated protein solutions. Although O2 binding is essentially prohibitive after 80 C, at high ionic strength and neutral pH, there is no significant change in the absorption spectrum of full-length HemAT-Hs as the temperature exceeds 95 C. Experiments relating sequence to structure and functional range in this class of highly diverse globins could have a tremendous impact in the ability to create customized globin-based biosensors.

192

T.A.K. Freitas et al.

5. MECHANISM OF SIGNALING Equilibrium and kinetic experiments involving UV-vis spectrophotometry, resonance Raman spectroscopy, stopped flow and laser flash photolysis methods [10,11] have shown that the aerotaxis sensor, HemAT-Bs (full-length and globin domain), possesses both high-affinity and low-affinity O2 -binding components [11], affording B. subtilis the sensory capacity to respond to both low and high oxygen concentrations in the environment.

5.1. The Open (Low Affinity) and Closed (High Affinity) Conformers Resonance Raman has shown that HemAT-Bs exhibits an oxygen isotope–sensitive band at ∼560 cm−1 [10]. Others have shown that the same low Fe−O2 (560 cm−1 ) in a Mycobacterium tuberculosis Hb is characteristic of a novel hydrogen-bonding network between the proximal oxygen of the bound O2 and the distal side of the heme pocket [67]. Recent analyses identified the centers of the 560 cm−1 oxygen-sensitive band at 554, 566, and 572 cm−1 . Examination in D2 O shows a frequency shift of the two longer wave numbers, indicating that a deuterium exchange has taken place with a protein residue resulting in a longer D-bond to the bound O2 . Ohta et al. [15] have tentatively identified these different conformers as open (weak H-bonding with moderate O2 affinity; Tyr70 facing solvent; 566 and 572 cm−1 ) and closed (strong H-bonding with greater O2 affinity; Tyr70 facing bound O2 ; 554 cm−1 ).

5.2. The Up (Low Affinity) and Down (High Affinity) Conformers Zhang et al. [11] arrived upon a similar multiconformational state after observing biphasic O2 -binding equilibria and kinetics arising from two different dissociation rates, presumably due to nonidentical binding sites among the HemAT-Bs monomers within the crystal dimer [14]. Physiologically, the difference in affinities would allow B. subtilis to respond to a wide range of O2 concentrations: the low-affinity conformer for hypoxic conditions and the high-affinity conformer for aerobic conditions. Carbon monoxide association and dissociation rates, however, are purely monophasic and are not significantly altered when Tyr70 is mutated to Phe, Leu, or Trp [11]. Using the ferrous unliganded crystal dimer as reference (PDB ID: 1OR6), Zhang et al. have designated the low-affinity HemAT-Bs monomer (molecule B) with Tyr70 facing the solvent as the “up” conformer and the high-affinity monomer (molecule A) with a low O2 dissociation rate with Tyr70 poised to interact with a bound ligand as the “down” conformer. The “up” conformer and the “down” conformer are equal to the open and closed forms, respectively.

5.3. Tyr70 Mutants Decrease the O2 Affinity in HemAT-Bs Mutational analyses of the HemAT-Bs full-length and sensor globin have been explored to identify any O2 -binding dependencies on the hydrogen-bonding potential of Tyr70 [11]. A significant influence of Tyr70 on oxygen-binding equilibria and kinetics has been demonstrated in Y70F, Y70L, and Y70W mutants, with the overall effect of

Protoglobin and Globin-coupled Sensors

193

lowering the O2 affinity (3–10-fold) by predominantly increasing the rate of O2 dissociation (10–30-fold) over that of association (three- to fivefold). The relative absence of this effect in similar experiments with CO has been attributed to the apolarity of the FeCO complex, where the addition or loss of a hydrogen bond with such a complex would only have minimal effects. Compared to the biphasic nature of wild-type HemAT-Bs (full-length and sensor globin), O2 dissociation time courses of the Tyr70 mutants are almost CO-like in that they are practically monophasic. The association rates, however, still maintain their biphasic nature [11].

5.4. Tyr70 Does Not Stabilize Heme-bound O2 by a Direct Hydrogen Bond Surprisingly, the loss of H-bond potential in the Y70F mutant was not observed within the oxygen isotope–sensitive region of the HemAT-Bs resonance Raman spectrum. The 560 cm−1 band is thought to reflect a complex H-bonding network between the proximal oxygen of the bound O2 and that of the distal pocket, like that of the M. tuberculosis Hb [66], and it was expected that Tyr70 provided this interaction. Site-directed mutagenesis of Tyr70 should be reflected in the oxygen-sensitive region, however, the Y70F mutant spectrum in this region (555, 566, and 573 cm−1 ) is almost indistinguishable from the wild type (554, 566, 572 cm−1 ). Ohta et al. thus concluded that Tyr70 does not provide a hydrogen bond directly to the heme-bound O2 [15].

5.5. Thr95 is Essential to the Closed (High Affinity) Conformer Examination of the HemAT-Bs sensor globin crystal structure indicates that the H-bond to the proximal O of the bound O2 may be provided by Thr95 and a nearby crystallized water molecule [15]. The 554 cm−1 band (closed form) did not produce a H2 O/D2 O shift and, therefore, is not H-bonded by the HemAT protein, suggesting that water is the donor. In addition, a T95A mutant (which should lack the water molecule) abolished the closed form band altogether and only a single O2 -bound open band remained at 569 cm−1 , demonstrating that Thr95 is essential for the closed, high-affinity (“Down”) form of HemAT-Bs. Ohta et al. [15] illustrate this scheme as Thr95 donating an H-bond to a nearby water molecule that, in turn, H-bonds the heme-bound O2 . Resonance Raman data on Fe CO stretching and Fe C O bending frequencies indicate that heme-bound CO does not interact with residues of the distal pocket [10]. FTIR data [11] suggest that Tyr70 may even switch to the “up” conformation upon CO binding. An initial report [15] of the Fe His bond strain upon NO binding indicates that NO may be discriminated on the proximal side of the heme pocket.

5.6. Interaction of the Sensor Globin Domain with the C-terminal Domain Given that the Y70F mutation affects O2 and not CO equilibrium binding and kinetics, how is it that these effects do not correlate with changes in the Fe−O2 bands? Truncation

194

T.A.K. Freitas et al.

of HemAT-Bs into the sensor globin domain only abolishes the closed form but retains the bands of the open conformer (563 and 571 cm−1 ) that are relatively unchanged from the wild type (566 and 572 cm−1 ), implying a structural linkage of the closed conformer with the C-terminal signaling domain. Ohta et al. [15] describe a scheme whereby Thr95 regulates O2 binding and that the orientation of the Tyr70 phenyl ring distinguishes between the closed and open forms and relays this information to the C-terminal signaling domain. Thr95 is essential to maintaining the closed form, and therefore, supports its role in sensing O2 directly. The biphasic O2 dissociation rates of the full-length HemAT and the sensor globin, however, were nearly identical [11], suggesting that the ligand-binding properties of HemAT-Bs are independent of the C-terminal signaling domain. Examination of the affinities of the Azotobacter vinelandii GCS, however, yield different results; that is, the sensor domain alone is pentacoordinated with myoglobin-like behavior, whereas the full protein is hexacoordinated with completely different affinities (Luc Moens, personal communication). These apparently disparate results may be due to differences among the various classes of GCSs or may simply be due to variations in the construction of the sensor domain. Whether or not the affinities of the globin sensor are influenced by the various C-terminal signaling domains will likely depend on the signaling mechanism and therefore, the degree of interaction between the signaling domain with those sensor domain residues that influence the environment of the heme pocket.

6. GCS DIVERSITY AND EVOLUTION Examination of the GCS phylogenetic tree [17] results in two possible interpretations: (i) certain bacterial species have a preference for specific signal transducing elements and/or (ii) the globin domains have evolved simultaneously with their signal transducing partners. Since the tree created was based on the globin sensor domain only, the branching-by-function observed demonstrates that aerotactic-type globins evolved from a common ancestral globin, or protoglobin, as did the gene-regulating type of globins. This is further emphasized by the protein stability data discussed earlier, where the globin domain itself is more stable than the protein as a whole. Assembly of the various globin-coupled sensors from the individual globin and transmitter domains was not necessarily a one-step evolutionary process like that in Fig. 4B, and may have resulted from cycles of fusions and divisions, combining the pathways of Fig. 4A and Fig. 4B. Novel functions imparted by these globin-coupled sensors could allow the host organism to thrive in environments normally unsuitable for survival. Regulation of oxygensensitive metabolic processes and taxis resulting in the emergence of new species is made possible because of a common sensory globin, evolved from the protoglobin, which is integral to the core function of oxygen homeostasis. Attempts at decoding the origin of life through molecular phylogeny has led to the concept of the Last Universal Common Ancestor, or LUCA, a community of cells whereby genetic information was shared freely [68] prior to selective pressures leading to the distinction between Bacteria and Archaea. As the information used in such analyses is constantly changing, LUCA’s origin is constantly under debate. LUCA is believed to have been a metabolically “flexible” single-celled organism with the ability to utilize oxygen for energy before free oxygen even existed in the air, thus preceding oxygenic photosynthesizers. The idea that an organism existed with the capacity to “breathe” O2

Protoglobin and Globin-coupled Sensors

195 (B)

(A) Globin-coupled sensor Flavohemoglobin

Protoglobin

? Bacterial hemoglobin Thermoglobin Truncated hemoglobin Neuroglobin Cytoglobin Myoglobin Hemoglobin

Globin-coupled sensor Flavohemoglobin

Bacterial hemoglobin Thermoglobin Truncated hemoglobin Neuroglobin Cytoglobin Myoglobin Hemoglobin where

= globin

= transmitter domain module

Fig. 4. GCS and protoglobin evolutionary pathways. (A) Division of a multidomain protein into components. Illustration depicts the liberation of the individual domains of the globin-coupled sensors where the liberated domains are free to evolve on their own. (B) Fusion of the individual domains to form a multidomain protein. The once separate globin and transmitter domains are now fused together into a globin-coupled sensor where both domains evolve together.

before there was a real need to, however, goes against the textbook viewpoint [69]. In his recent book [2], Nick Lane argues that LUCA likely made use of a hemoglobinlike protein to manage oxygen homeostasis and an antioxidant enzyme like superoxide dismutase (SOD) to protect itself. This hemoglobin would not have to deal with much oxygen at all, but rather very low levels of oxygen, perhaps similar to the role of leghemoglobin in nitrogen-fixing bacteria. Uncovering a molecular fossil of this ancient globin by sequence identity alone would be unlikely without a more in-depth profile of the expected characteristics. Using the determined globin domain length, positioning of the proximal histidine and distal residues, and the chemical nature of the heme pocket [16], protoglobins were identified in both the Archaea and Bacteria [18]. Preliminary analyses of the two archaeal protoglobins ApPgb and MaPgb from A. pernix and Methanosarcina acetivorans, respectively, indicate that these proteins are very oxygen sensitive, as predicted for the LUCA hemoglobin. As the O2 affinity of the protoglobins studied thus far is quite low, possibly generating O2 radicals in the process, antioxidants like SOD would be essential in protecting LUCA from intracellular damage.

196

T.A.K. Freitas et al.

The current universal tree of life, clouded by horizontal gene transfer (HGT) [70] and sampling bias [71], makes it difficult to draw conclusions about the true identity of LUCA. Extremophiles – organisms populating areas of the planet previously thought inhabitable – are thought by some, to most closely resemble LUCA’s habitat, whereas others believe that LUCA may have existed in a more temperate climate. Whether or not LUCA was a (hyper)thermophile is still not certain despite evidence put forward both in support of [72–74] and against [75–77] this hypothesis. The most recent attempt at resolving the tree of life by circumventing the aforementioned issues above involved the analysis of 31 orthologous genes universally distributed amongst 191 species with completely annotated genomes [74]. Although the tree of life tends to be rooted along the bacterial lineage, these results come from an unrooted tree and support a Gram-positive thermophilic LUCA of likely bacterial origin rather than one of archaeal origin. The data indicate that the Firmicutes branch from LUCA sooner than the Archaea. Of the Firmicutes, almost all Bacilli sequenced thus far possess a HemAT (see Table 1), one of the protoglobin descendents. Unless each species of Bacillus received the HemAT via HGT, the Bacilli predecessor must have possessed either the HemAT or the protoglobin itself. Either the predecessor(s) have not been identified or this tree of life cannot yet account for the evolution of the GCS from the protoglobin.

7. PROTOGLOBINS IN THE ARCHAEA Seven proteins have been identified that conform to the protoglobin criteria set forth above and have been found in the Bacteria (Actinobacteria Thermobifida fusca str. YX and Rubrobacter xylanophilus str. DSM 9941, the green nonsulfur bacterium Chloroflexus aurantiacus, and the -Proteobacterium Nitrosococcus oceani ATCC 19707) and the Archaea (aerobic hyperthermophile Aeropyrum pernix and strictly anaerobic methanogens Methanosarcina acetivorans str. C2A and Methanosarcina barkeri str. fusaro). Of these, only two of the three archaeal proteins have been experimentally analyzed [18]. The proximal histidine was initially identified by sequence analysis (Fig. 1) and later confirmed in the Archaeal protoglobins by site-directed mutagenesis and molecular modeling [18]. Both archaeal proteins purify in the oxidized state, and reduction with concentrated sodium dithionite can be a particularly time-consuming process (>45 minutes) unless facilitated with an electron shuttle like methyl viologen (∼15 minutes). Other reducing agents such as ascorbate, -mercaptoethanol, and dithiothreitol (DTT) have little to no effect on these protoglobins. Generation of reduced protoglobin requires an environment free of O2 , as they autoxidize rapidly at rates of 0.0032 and 0.0027 s−1 , corresponding to half-lives of 3.6 and 4.3 minutes for oxy-MaPgb and oxy-ApPgb, respectively [18]. Protoglobins are, on average, 196 (±1) amino acids in length, roughly the same as the minimum heme-binding length experimentally determined (195-aa) for the archaeal HemAT from H. salinarum [9]. The HemAT-Hs sensor globin domain, however, retains bound oxygen for much longer periods of time [9] than the protoglobins. This evidence indicates that the protoglobins are not simple equivalents of the liberated GCS globin domains and that there is a fundamental structural and/or functional difference between

Protoglobin and Globin-coupled Sensors (A)

197 (B)

SENSOR DOMAIN WITH HEME

VARIABLE SIGNALING DOMAIN

Flx15 (HemAT-Hs and HtrVIII)

OI1085 Wild type

ΔHemAT-Hs + HtrVIII

OI3545 ΔTen

+ HemAT-Hs ΔHtrVIII

OI3555 ΔTen + HemAT-Bs

SIGNALING PATHWAYS

Fig. 5. Aerotaxis in the HemATs. (A) Basic model of aerotactic responses in the HemATs. Oxygen is bound at the N-terminus and this signal is relayed to the C-terminus by an unknown mechanism. (B) Aerotactic responses of HemAT-Hs and HemAT-Bs. Left panel: Effects of HemAT-Hs on aerotaxis in H. salinarum. Flx15 is wild type and HtrVIII is an aerophilic oxygen transducer in H. salinarum. Right panel: Effects of HemAT-Bs on aerotaxis in B. subtilis. Ten corresponds to a deletion mutant of all 10 chemotaxis transducers in B. subtilis.

the two. It is doubtful that globins with such high oxidation rates are suitable for functioning within high oxygen concentration environments. One of the characteristic features of the Pgbs is the presence of the W59/Y60 residue pair, where Tyr60 aligns with the B10 Tyr in most other bacterial globins. Molecular modeling and dynamics simulations suggested that the position of Trp59 may facilitate interaction with the incoming gaseous ligand; subsequent mutational studies supported this hypothesis (Newhouse et al., in progress). In addition, the autoxidation rate of the archaeal Pgbs was drastically altered by a single W59A mutation. Reduction of the W59A mutant was thus possible with dithiothreitol (DTT) and proceeded in under 5 minutes as compared to the wild type (>2 hours), indicating the importance of W59 in the redox properties of the protoglobin. Interestingly, the oxygen requirements for these two Archaea are prohibitively small. Although A. pernix is an obligate aerobe, it is also a hyperthermophile, and only ∼25 M of O2 can dissolve in water (1 atm pressure) at its optimal growth temperature of 95 C. Methanosarcina acetivorans, on the other hand, is a strict anaerobe, and although it possesses some components of aerobic metabolism in its genome [78], it is unlikely that this organism can deal with oxygen concentrations exceeding the nanomolar level. Another strict anaerobe, Bacteroides fragilis, is known to grow in and benefit from nanomolar concentrations of oxygen [79]; however, this has not been shown in M. acetivorans. Sequence identity among the protoglobins range from 57 to almost 90%, and the archaeal protoglobins MaPgb and ApPgb, thus far, bind the same range of ligands: O2 , CO, and NO in the ferrous form, and cyanide, azide, and imidazole in the ferric form [18]. The seven protoglobins identified in Bacteria and Archaea all possess a cysteine residue near the end of the E-helix (E19 Cys); however, the hyperthermophilic ApPgb

198

T.A.K. Freitas et al.

is the only one with a second cysteine near the end of the A-helix. Molecular modeling of ApPgb based on the HemAT-Bs sensor domain suggests that these two cysteines are in positions suitable for forming a disulfide bond [18], likely contributing to ApPgb’s thermotolerance; ApPgb can reversibly bind CO at near-boiling temperatures in vitro (data not shown). The Cys residue of A. pernix (C102) is analagous to the E19 Cys of both the Ascaris suum hemoglobin and the H2 S-binding annelid hemoglobin from Riftia pachyptila. In the A. suum hemoglobin, the E19 Cys near the distal pocket has been shown to play a role in its NADPH-dependent NO-activated deoxygenase function [80]. In the annelid hemoglobin from R. pachyptila, however, Cys E19 has been shown to be critical in H2 S binding [81], serving as the site for sulfide addition. The high reactivity of cysteine thiols predisposes the parent protein to binding atypical ligands and bringing about diseased states. This scenario has been suggested as a possible driving force for the evolutionary loss of H2 S-binding in hemoglobins from organisms living in sulfide-free habitats [82]. This would be consistent with ancient globins working to detoxify sulfide and nitric oxide, and the E19 Cys being absent from modern globins that have adapted to highly oxic environments. Protoglobins, with their broad ligand range, thermostability, and oxygen sensitivity predisposes them to functioning in low-oxygen environments. Of all the ancient extant globins, the protoglobins may be the molecular fossil that closest resembles the LUCA globin. As the atmospheric contents began to shift and the oxygen levels rose, the protoglobin was exposed to environmental pressures selecting for the capacity to transport and store this oxygen more efficiently to maintain oxygen homeostasis, eventually becoming the hemoglobin in our blood and the myoglobin in our muscles. Toxic molecules like nitric oxide and sulfides can compete with oxygen binding and affect oxygen homeostasis. A consequence of these changes, therefore, was a protein less efficient at binding the toxic molecules of the time. Those that have not lost this capability, however, now have a great impact in present-day biology – organisms with the capacity to deal with reactive molecules like NO and H2 S should be able to survive in very inhospitable environments. Prevalent in many bacterial and fungal pathogens, flavohemoglobin, also known as a “fungal defense” enzyme, has been shown to be closely tied to virulence [83]. Upon invasion, macrophages engulf pathogens and produce oxidants like nitric oxide to combat infection. de Jesus-Berrios et al. [83] have shown that inactivation of flavohemoglobin led to a significant loss of virulence both in vitro and in vivo. Mycobacterium tuberculosis, another intracellular pathogen that infects macrophages, produces two hemoglobins, HbN and HbO. In vitro experiments have shown that HbN has the capability to scavenge and detoxify nitric oxide, thus assisting the pathogen in infection and pathogenesis by circumventing the deleterious effects of the host immune system’s “oxidative burst.” Here are but two examples of globins that have retained and perhaps improved their ability to combat toxic molecules and exploit them for survival in hostile environments today. It has been suggested that the single-domain hemoglobin from the hyperthermophile Aquifex aeolicus (Thermoglobin – Tgb), due to its high O2 affinity and thermostability, resembles the globin in LUCA [84]. The extant protoglobins from Bacteria and Archaea are also highly thermostable, and a recent analysis of the tree of life [74] identifies the M. acetivorans and A. pernix as very slowly evolving species. Although these Pgb-encoding organisms have slightly longer branch lengths (from root) than that of

Protoglobin and Globin-coupled Sensors

199

A. aeolicus, many of the organisms encoding GCSs have shorter branch lengths than A. aeolicus [74]. Since the Pgb logically must have come before the GCS, the current tree cannot account for diversification of the LUCA globin. Pgbs are ∼195-aa in length with low oxygen affinities [18], whereas Tgb is only 139-aa long and has a very high O2 affinity [84]. The origin of Tgb and Pgb cannot be answered conclusively yet with the data currently available; however, application of the metagenomic approach to extremophile genome analysis may fill in the evolutionary “gaps.”

8. ANCIENT OXYGEN SIGNALING AND THE FUTURE The globin fold is successful at protecting the heme iron from rapid oxidation and yet allows reversible O2 binding. PAS domains are capable of accepting various input stimuli and participate in proteins regulating a whole slew of functions [85]. Some PAS domains are heme-binding (heme-PAS) domains as well, also capable of reversibly binding oxygen. A computer simulation study based on known PAS structures [86] suggests that a common conformational flexibility exists between these structures that is essential for signaling. A given polypeptide can possess multiple PAS domains (up to six) [85] and these domains are not restricted to the N-terminus. Globins, on the other hand, are known to bind heme only and those polypeptides with globin domains (GCSs and flavohemoglobins) express this domain at the N-mostterminal region only. In addition, only one globin domain is found per polypeptide. This implies that globin domains may not be as modular as PAS domains, leading to one of three conclusions: either (i) a signaling mechanism exists such that the globin must be positioned at the N-terminus of the polypeptide for proper signal relay; or (ii) the transmitter requires positioning at the C-terminus to interact and function properly with downstream components within the cell; or (iii) there is currently an inadequate representation of the GCS population. Out of the 645 genomes (512 bacterial, 29 archaeal, 104 eukaryotic) that have been or are in the process of being sequenced, the sensor globin domain of all GCSs, thus far, is at the N-most-terminal portion of the protein. Communication between the HemAT signaling domain and the proteins of the chemotaxis pathway is the likely reason for this organization schema in the aerotactic GCSs. Future structure-function studies will hopefully shed light on the rules of topological restrictions in assembling a properly functioning globin-coupled sensor. Ongoing genome-sequencing projects will undoubtedly help to clear this up. In recent years, neuroglobin and cytoglobin have been implicated in signal transduction and gene regulation [87,88]. In mice, they have been found in the brain, retina, liver, heart, striated muscle, lung, kidney, and small bowel [88–90]. Traditionally, the reduced form of globins has been considered to be the physiologically active form. However, a recent finding has shown that the rapidly autoxidizing neuroglobin (expressed in vertebrate brain and retina) is active in the ferric form in preventing GDP-GTP exchange in G proteins by sequestering the GDP-bound G subunit [87]. By crossing over into the G protein signal transduction cascade, it is evident that globins have the capacity to be signal transducing elements all on their own. Cytoglobin has even been discovered in the nucleus of vertebrate cells [88], is upregulated in all tissues upon hypoxia [91], prevents ischemic cell death [92], and may possibly function as a transcription regulator [88].

200

T.A.K. Freitas et al.

Interestingly, the heme-regulated -subunit of eukaryotic initiation factor-2 (eIF2) kinase (HRI), only recently described as having a globin fold at its N-terminus [93], also prohibits GTP-GDP exchange but this time by interaction with the -subunit of eIF-2 in reticulocytes. When heme is deficient in reticulocytes, HRI prohibits GTP-GDP exchange in eIF-2 by phosphorylating eIF-2 [94] and therefore prevents eIF-2 from interacting with eIF-2. The 43 S initiation complex is thus unable to form and protein synthesis is shut down. For comparison, we tested the mammalian HRI N-terminal sequence with those of the (bacterial) GCSs. Indeed, the HRI N-terminal sequence does fit a profile-based alignment with the GCSs (unpublished data). As the protoglobin may have given life to LUCA, its globin descendents allowed higher organisms to evolve by maintaining their core function of oxygen homeostasis. This collective evolution of the globin family as a whole made life possible; whether by aerotaxis, gene regulation, detoxification, sequestration, or transport, the inter- and intracellular balance of oxygen was key to the evolution of humans. Thus, it only seems logical that globins be found in most mammalian tissues and in the blood that bathes them.

ACKNOWLEDGMENTS This investigation was supported by a National Science Foundation (MCB0080125) grant, UH intramural Bioinformatics grant, and by US Army Award TATRC # W81XWH-05-2-0013.

REFERENCES [1] Hou, S., Larsen, R.W., Boudko, D. et al. (2000) Nature 403, 540–544. [2] Lane, N. (2002) Oxygen. The Molecule that Made the World. New York: Oxford University Press Inc. [3] Moens, L., Vanfleteren, J., Van de Peer Y. et al. (1996) Mol. Biol. Evol. 13, 324–333. [4] Pesce, A., Couture, M., Dewilde, S. et al. (2000) EMBO J. 19, 2424–2434. [5] Perutz, M.F. (1989) Trends Biochem. Sci. 14, 42–44. [6] Peterson, E.S., Huang, S.,Wang, J. et al. (1997) Biochemistry 36, 13110–13121. [7] Sharma, V.S., Isaacson, R.A., John, M.E. et al. (1983) Biochemistry 22, 3897–3902. [8] Hou, S., Belisle, C., Lam, S. et al. (2001) Extremophiles 5, 351–354. [9] Hou, S., Freitas, T., Larsen, R.W. et al. (2001) Proc. Natl. Acad. Sci. U.S.A. 98, 9353–9358. [10] Aono, S., Kato, T., Matsuki, M. et al. (2002) J. Biol. Chem. 277, 13528–13538. [11] Zhang, W., Olson, J.S., Phillips, G.N., Jr. (2005) Biophys. J. 88, 2801–2814. [12] Boudko, D., Yu, H.S., Ruiz, M. et al. (2003) J. Microbiol. Methods 53, 123–126. [13] Zhang, W. and Phillips, G.N., Jr. (2003) Acta. Crystallogr. D. Biol. Crystallogr. 59, (Pt 4) 749–751. [14] Zhang, W. and Phillips, J.N., Jr. (2003) Structure (Camb). 11, 1097–1110. [15] Ohta, T., Yoshimura, H., Yoshioka, S. et al. (2004) J Am Chem Soc. 126, 15000–15001. [16] Freitas, T.A., Hou, S. and Alam, M. FEBS. Lett. (2003) 552, 99–104. [17] Freitas, T.A., Saito, J.A., Hou, S. and Alam M. (2005) J. Inorg. Biochem. 99, 23–33. [18] Freitas, T.A.K., Hou, S., Dioum, E.M. et al. (2004) Proc. Natl. Acad. Sci. U.S.A. 101, 6675–6680.

Protoglobin and Globin-coupled Sensors

201

[19] Gekakis, N., Staknis, D., Nguyen, H.B. et al. (1998) Science 280, 1564–1569. [20] Hogenesch, J.B., Gu, Y.Z., Jain, S. and Bradfield, C.A. (1998) Proc. Natl. Acad. Sci. U.S.A. 95, 5474–5479. [21] Lanzilotta, W.N., Schuller, D.J., Thorsteinsson, M.V. et al. (2000) Nat. Struct. Biol. 7, 876–880. [22] Reick, M., Garcia, J.A., Dudley, C. and McKnight, S.L. (2001) Science 293, 506–509. [23] Dioum, E.M., Rutter, J., Tuckerman, J.R. et al. (2002) Science 298, 2385–2387. [24] David, M., Daveran, M.L., Batut, J. et al. (1988) Cell 54, 671–683. [25] Virts, E.L., Stanfield, S.W., Helinski, D.R. and Ditta, G.S. (1988) Proc. Natl. Acad. Sci. U.S.A. 85, 3062–3065. [26] Gilles-Gonzalez, M.A. and Gonzalez, G. (1993) J. Biol. Chem. 268, 16293–16297. [27] Ross, P., Mayer, R. and Benziman, M. (1991) Microbiol. Rev. 55, 35–58. [28] Gilles-Gonzalez, M.A., Ditta, G.S. and Helinski, D.R. (1991) Nature 350, 170–172. [29] Zhao, Y., Brandish, P.E., Ballou, D.P. and Marletta, M.A. (1999) Proc. Natl. Acad. Sci. U.S.A. 96, 14753–14758. [30] Delgado-Nixon, V.M., Gonzalez, G. and Gilles-Gonzalez, M.A. (2000) Biochemistry 39, 2685–2691. [31] Chang, A.L., Tuckerman, J.R., Gonzalez, G. et al. (2001) Biochemistry 40, 3420–3426. [32] Sasakura, Y., Hirata, S., Sugiyama, S. et al. (2002) J. Biol. Chem. 277, 23821–23827. [33] Baliga, N.S., Bonneau, R., Facciotti, M.T. et al. (2004) Genome Res. 14, 2221–2234. [34] Taylor, B.L. and Zhulin, I.B. (1999) Mol. Biol. Rev. 63, 479–506. [35] Yu, H.S., Saw, J.H., Hou, S. et al. (2002) FEMS Microbiol. Lett. 217, 237–242. [36] Frankel, R.B., Bazylinski, D.A., Johnson, M.S. and Taylor. B.L. (1997) Biophys. J. 73, 994–1000. [37] Tal, H.C., Wong, R., Calhoon, D. et al. (1998) J. Bacteriol. 180, 4416–4425. [38] Galperin, M.Y., Nikolskaya, A.N. and Koonin, E.V. FEMS Microbiol. Lett. 203, 11–21. [39] Yoshimura-Suzuki, T., Sagami, I., Yokota, N. et al. (2005) J. Bacteriol. 187, 6678–6682. [40] Mendez-Ortiz, M.M., Hyodo, M., Hayakawa, Y. and Membrillo-Hernandez. J. (2006) J. Biol. Chem. 281, 8090–8099. [41] Kanacher, T., Schultz, A., Linder, J.U. and Schultz, J.E. (2002) EMBO J. 21, 3672–3680. [42] Hurley, J.H. (2003) Sci. STKE. 164, PE1. [43] Paglia, M.J., Mou, H. and Cote. R.H. (2002) J. Biol. Chem. 277, 5017–5023. [44] D’Argenio, D.A. and Miller, S.I. (2004) Microbiology. 150, 2497–2502. [45] Christen, M., Christen, B. and Folcher, M. et al. (2005) J. Biol. Chem. 280, 30829–30837. [46] Cole, J.R., Cascarelli, A.L., Mohn, W.W. and Tiedje. J.M. (1994) Appl. Environ. Microbiol. 60, 3536–3542. [47] Loffler, F.E., Tiedje, J.M. and Sanford. R.A. (1999) Appl. Environ. Microbiol. 65, 4049–4056. [48] Sanford, R.A., Cole, J.R. and Tiedje, J.M. (2002) Appl. Environ. Microbiol. 68, 893–900. [49] He, Q. and Sanford, R.A. (2002) Biodegradation 13, 307–316. [50] He, Q. and Sanford, R.A. (2003) Appl. Environ. Microbiol. 69, 2712–2718. [51] Lois, A.F., Weinstein, M., Ditta, G.S. and Helinski. D.R. (1993) J. Biol. Chem. 268, 4370–4375. [52] Aravind, L. and Koonin, E.V. (2000) Curr. Biol. 10, R53–R55. [53] Duncan, L., Alper, S. and Losick, R. (1996) J. Mol. Biol. 260, 147–164. [54] Ko, S.B., Zeng, W. and Dorwart, M.R. et al. (2004) Nat. Cell Biol. 6, 343–350. [55] Gray M.A. (2004) Nat. Cell Biol. 6, 292–294. [56] Kelley, L.A., MacCallum, R.M. and Sternberg, M.J.E. (2000) J. Mol. Biol. 299, 499–520. [57] Methe, B.A., Nelson, K.E. and Eisen, J.A. et al. (2003) Science 302, 1967–1969. [58] Marchler-Bauer, A., Anderson, J.B., Cherukuri, P.F. et al. (2005 Database issue) Nucleic Acids Res. 33, D192–D196. [59] Ermler, U. Siddiqui, R.A., Cramm, R. and Friedrich, B. (1995) EMBO J. 14, 6067–6077.

202

T.A.K. Freitas et al.

[60] Tarricone, C., Galizzi, A., Coda, A. et al. (1997) Structure 5, 497–507. [61] Couture, M., Yeh, S.R., Wittenberg B.A. et al. (1999) Proc. Natl. Acad. Sci. U.S.A. 96, 11223–11228. [62] Mukai, M. Mills, C.E. Poole, R.K. and Yeh, S.R. (2001) J. Biol. Chem. 276, 7272–7277. [63] Ilari, A., Bonamore, A., Farina, A. et al. (2002) J. Biol. Chem. 277, 23725–23732. [64] Mukai, M., Savard, P.Y., Ouellet, H. et al. (2002) Biochemistry 41, 3897–3905. [65] Ouellet, H., Juszczak, L. and Dantsker, D. et al. (2003) Biochemistry 42, 5764–5774. [66] Seeley, S.K., Wittrock, G.K., Thompson, L.K. and Weis, R.M. (1996) Biochemistry 35, 16336–16345. [67] Yeh, S.R., Couture, M. and Ouellet, Y. (2000) J. Biol. Chem. 275, 1679–1684. [68] Woese, C. (1998) Proc. Natl. Acad. Sci. U.S.A. 95, 6854–6859. [69] Castresana, J. and Saraste, M. (1995) Trends Biochem. Sci. 20, 443–448. [70] Philippe, H. and Douady, C.J. (2003) Curr. Opin. Microbiol. 6, 498–505. [71] Hillis, D.M., Pollock, D.D., McGuire, J.A. and Zwickl, D.J. (2003) Syst. Biol. 52, 124–126. [72] Di Giulio, M. (2003) J. Mol. Evol. 57, 721–730. [73] Atomi, H., Matsumi, R. and Imanaka. T. (2004) J. Bacteriol. 186, 4829–4833. [74] Ciccarelli, F.D., Doerks, T. and von Mering, C. et al. Science. 311, 1283–1287. [75] Forterre, P. (1996) Cell. 85, 789–792. [76] Galtier, N., Tourasse, N. and Gouy, M. (1999) Science 283, 220–221. [77] Xu, Y. and Glansdorff, N. (2002) Comp. Biochem. Physiol. A Mol. Integr. Physiol. 133, 677–688. [78] Galagan, J.E., Nusbaum, C., Roy A. et al. (2002) Genome Res. 12, 532–542. [79] Baughn, A.D. and Malamy, M.H. (2004) Nature 427, 441–444. [80] Minning, D.M., Gow, A.J., Bonaventura, J. et al. (1999) Nature 401, 497–502. [81] Bailly, X., Jollivet, D., Vanin S. et al. (2002) Mol. Biol. Evol. 19, 1421–1433. [82] Bailly, X., Leroy, R., Carney, S. et al. (2003) Proc. Natl. Acad. Sci. U.S.A. 100, 5885–5890. [83] de Jesus-Berrios, M., Liu, L., Nussbaum, J.C. et al. (2003) Curr. Biol. 13, 1963–1968. [84] Miranda, J.J., Maillett, D.H., Soman, J. and Olson, J.S. (2005) J. Biol. Chem. 280, 36754– 36761. [85] Gilles-Gonzalez, M.A. and Gonzalez, G. (2004) J. Appl. Physiol. 96, 774–783. [86] Vreede, J., van der Horst, M.A. and Hellingwerf, K.J. et al. (2003) J. Biol. Chem. 278, 18434–18439. [87] Wakasugi, K., Nakano, T. and Morishima, I. (2003) J. Biol. Chem. 278, 36505–36412. [88] Geuens, E., Brouns, I. and Flamez, D. et al. (2003) J. Biol. Chem. 278, 30417–30420. [89] Burmester, T., Weich, B., Reinhardt, S. and Hankeln, T. (2000) Nature 407, 520–523. [90] Schmidt, M., Giessl, A., Laufs, T. et al. (2003) J. Biol. Chem. 278, 1932–1935. [91] Fordel, E., Geuens, E., Dewilde, S. et al. (2004) Biochem. Biophys. Res. Commun. 319, 342–348. [92] Stagner, J.I., Parthasarathy, S.N., Wyler, K. and Parthasarathy, R.N. (2005) Transplant Proc. 37, 3452–3453. [93] Uma, S., Matts, R.L., Guo, Y. et al. (2000) Eur. J. Biochem. 267, 498–506. [94] Chen, J.J. and London, I.M. (1995) Trends Biochem. Sci. 20, 105–108.

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 8

Neuroglobin and Cytoglobin Thomas Hankelna and Thorsten Burmesterb a

Institute of Molecular Genetics, Johannes Gutenberg University of Mainz, J. J. Becherweg 32, D-55099 Mainz, Germany b Institute of Zoology, Johannes Gutenberg University of Mainz, D-55099 Mainz, Germany

Abstract Neuroglobin and cytoglobin are two recent additions to the family of hemecontaining respiratory proteins of man and other vertebrates. Here, we summarize the current state of knowledge of the structures, ligand-binding kinetics, evolution, and expression patterns of these two proteins. These data provide working hypotheses with regard to the possible physiological roles of these globins in the animal’s metabolism. Both neuroglobin and cytoglobin are structurally similar to myoglobin, but they contain distinct features like extraordinarily high temperature resistances and unusual cavities inside the molecules. Kinetic and structural studies show that neuroglobin and cytoglobin belong to the class of hexacoordinated globins with a biphasic ligand-binding kinetics. Nevertheless, their oxygen affinities resemble that of myoglobin. While neuroglobin is evolutionarily related to one lineage of invertebrate nerve globins, cytoglobin shares a more recent common ancestry with myoglobin. Neuroglobin expression is confined to neurons of the central and peripheral nervous system and to endocrine tissues, with the highest expression observed in the retina. Present evidence points to an important role of neuroglobin in neuronal oxygen homeostasis and hypoxia protection, although other or additional functions are conceivable. Cytoglobin is predominantly expressed in fibroblasts and related cell types, but also in distinct nerve cell populations. Much less is known about its function; cytoglobin may be involved in oxygen transfer to enzymes like collagen prolylhydroxylase or NO synthase, ROS protection, or signaling.

1. GLOBINS: THE ANCIENT PROTEIN SUPERFAMILY CONTAINS TWO NOVICES Globins are small globular metalloproteins consisting of about 150 amino acids. Typically, globins comprise eight -helical segments (named A through H) that display a characteristical 3-over-3 -helical sandwich structure. This conserved “globin fold” identifies them as members of a large protein superfamily [1–3], which also includes truncated versions whose globin fold consists of only four -helices [4]. Globins contain a heme prosthetic group (Fe-protoporphyrin IX), by which they can reversibly bind gaseous ligands like O2 , CO, and NO. Most known globins fulfill respiratory functions, supplying the cell with adequate amounts of O2 for aerobic energy production via the respiratory chain in the mitochondria [5–7]. They are phylogenetically ancient molecules whose intricate adaptive evolution is demonstrated by their widespread occurrence

204

T. Hankeln and T. Burmester

in bacteria, fungi, plants, invertebrate, and vertebrate animals [8,9]. In man and most other vertebrates, the heterotetrameric hemoglobin (Hb), which is located in red blood cells (erythrocytes), transports O2 from the respiratory organs to the tissues via the circulatory system, and CO2 (at the N-terminal amino groups of the protein chains) in the reverse direction [5]. The monomeric vertebrate myoglobin (Mb) is present in cardiac and striated muscles, where it functions as a local O2 store and probably facilitates intracellular diffusion of O2 to the mitochondria [7,10]. In addition, Mb acts as a dioxygenase, converting potentially harmful nitric oxide (NO) radicals into innocuous nitrate [11]. With such exciting data, globins continue to be amongst the best-studied proteins on the levels of structure, function, and evolution. A few years ago, intrigued by the discovery of unexpected globins in the insect model organism Drosophila [12–14], we commenced to systematically search human and mouse genome project sequence databases for the presence of novel, additional globins. First found was neuroglobin (Ngb), a globin predominantly expressed in nervous tissue [15]. Shortly after that, a fourth vertebrate globin type was described independently by three groups [16–18] and is now officially named cytoglobin (Cygb), based on its widespread expression in many mammalian tissues and organs. These findings add considerable complexity to our view on O2 metabolism in the vertebrate cell and may have substantial biomedical implications. Here, we summarize the structural, biochemical, gene expressional, and functional data on Ngb and Cygb available until February 2006 (Fig. 1).

hemoglobin

myoglobin

cytoglobin

red blood cells

skeletal muscle heart smooth muscle

fibroblast cell lineage liver stellate cells CNS/PNS

neurons (CNS, PNS) retina endocrine tissue fish gills

Fe-atom coordination

penta

penta

hexa

hexa

oxygen affinity (P50(O2) in torr)

26

1

1

1

expression sites

neuroglobin

gene location (human)

α-cluster 16p13 β-cluster 11p15

22q13

17q25

14q24

gene ID

(α)-cluster 16p13 (β)-cluster 11p15

4151

114757

58157

phylogeny

HBA

MB

CYGB

NGB

HBB

400

million years

0

800

Fig. 1. Characteristics of vertebrate globins. The graphic summarizes selected expressional, biochemical, and phylogenetic features of vertebrate globin types.

Neuroglobin and Cytoglobin

205

2. NEUROGLOBIN: THE DISTANTLY RELATED COUSIN “ON OUR NERVES” 2.1. Neuroglobin Structure and Ligand Binding Ngb is a substantially divergent member of the globin family, displaying only 20–25% amino acid sequence identity to Mbs and Hbs ([15]; Fig. 2). Ngb represents a typical Mb-type monomeric globin, which can bind O2 reversibly [15,19,20]. In spite of its sequence differences, Ngb features the conserved globin fold consisting of the eight -helices A–H, albeit with some peculiarities that reflect a pronounced adaptive potential

Fig. 2. Amino acid sequence alignment of Ngb and Cygb sequences from various vertebrate species, compared with human hemoglobin - and -chains and human myoglobin. Amino acid positions are shaded to indicate conserved residues. Cysteine residues in Ngb and Cygb, possibly engaging in disulfide bond formation (see text), are linked by brackets. The intron positions found in Ngb and Cygb genes are indicated by arrows (B12.2, e.g., means that the intron is found between positions 2 and 3 in the 12th codon of helix B). The species abbreviations are: Hsa, Homo sapiens; Ptr, Pan troglodytes; Mmu, Mus musculus; Rno, Rattus norvegicus; Cfa, Canis familiaris; Bta, Bos taurus; Ssc, Sus scrofa; Tni, Tetraodon nigroviridis; Tru, Takifugu rubripes; Dre, Danio rerio; Omy, Onchorhynchus mykiss.

206

T. Hankeln and T. Burmester (A)

C F H

D

E B G A

myoglobin

neuroglobin

cytoglobin

(B)

hexa

penta

oxy

Fig. 3. (A) 3-D Protein structure of human MB (PDB accession number: 2MM1), NGB (1OJ6), and CYGB (1UMO). In MB, the eight -helices are designated A through H. Note that the globin fold is conserved in all three proteins. (B) Scheme of globin hexacoordination. The equilibrium of the hexacoordinated and pentacoordinated form is the rate-limiting step in ligand binding for NGB and CYGB. Colors: red, heme group; green, interacting histidines; blue, oxygen ligand.

of this basic globin structure (Fig. 3). The crystal structures of human and mouse Ngb have been solved [21,22], revealing the presence of unusual protein cavities that are not found as such in Hb and Mb and that may influence ligand storage and diffusion paths inside the molecule. The most peculiar structural characteristic of Ngb is the so-called “hexacoordinated” binding scheme of the heme Fe atom in the ferrous (Fe2+ ) deoxy and in the ferric (Fe3+ ) states (Fig. 3). The crystallographic data have ultimately confirmed several types of spectroscopic analyses [15,19,23–27], showing that in the absence of external ligand, the histidine at position 7 of the E-helix (HisE7) binds the heme iron at its sixth, distal position. Thereby, any external gaseous ligand such as O2 or CO has to compete with the internal His(E7) ligand for Fe binding. This produces a biphasic ligandbinding kinetics for Ngb: the displacement of the His(E7) is the rate-limiting, slow step, while the inherent affinity of the Fe atom after His(E7) displacement is high and makes the gaseous ligand-binding step fast [19,28]. Heme hexacoordination has previously been reported in plant, bacteria, and invertebrate globins [29], and although this widespread occurrence may suggest a conserved function, its physiological significance is not yet understood. Recent kinetic studies show that even slight variations in pH may cause pronounced changes in the association rates of exogenous ligands in Ngb [30]. On the other hand, hexacoordination in Ngb and other globins may render the process of external ligand binding relatively independent of temperature variations [31], which

Neuroglobin and Cytoglobin

207

might be physiologically relevant under conditions of fluctuating body temperatures, e.g., in poikilothermic animals. Hexacoordination may also protect the Fe3+ iron atom against various oxidizing molecules, which seems to suppress the formation of cytotoxic ferryl (Fe4+ ) heme [32]. Ngb protein is an extremely thermostable protein, resisting temperatures close to 100 C in vitro [33]. Another notable structural feature of Ngb, which may discriminate it from other known globins, is a small sliding of the heme upon ligand binding [34]. Despite the complex ligand-binding scheme, the overall O2 affinity (P50 ) of mammalian and of fish Ngb is in the range of 1–2 Torr [15,19,28,35], which is similar to the O2 affinity of Mb. These values strongly depend on buffer conditions used, and lower P50 affinity values of 5–7 Torr have also been measured for Ngb [36]. As expected for a monomeric globin, neither mammalian nor zebrafish Ngb display any cooperativity [36]. Human NGB is able to form an internal disulfide bridge at cysteines CD7 and D5 in vitro (see Fig. 2) [20], which may break up under reducing conditions in the cell, e.g., when NADH+ reduction equivalents accumulate under hypoxia. Reduction of the disulfide bond in turn lowers the O2 affinity of Ngb by a factor of 10, which would lead to a release of O2 and thus, possibly, an attenuation of hypoxic stress. It is not yet clear if this mechanism is acting in vivo: while fish Ngbs possess two cysteines at roughly equivalent positions, rodent Ngbs lack the CD7 cysteine (c.f., Fig. 2). To fully understand Ngb function, it is essential to investigate the possibility of binding to ligands other than O2 , namely the noxious reactive oxygen and nitrogen species, which accumulate in the cell, e.g., after ischemic insult and subsequent reperfusion of the tissue [37]. EPR (electron paramagnetic resonance) and kinetic studies revealed that the binding affinity of the reactive nitric oxide (NO) to Ngb Fe2+ is low compared to pentacoordinate Hbs and Mbs, which is due to protection of the Fe2+ by the internal His(E7) ligand [38,39]. Under excess of NO applied in vitro, however, Ngb Fe2+ NO readily forms and decompose peroxynitrite. Recently, it has been demonstrated that Ngb Fe2+ O2 reacts in vitro with NO, yielding Ngb Fe3+ and NO− 3 via a peroxynitrite intermediate [40].

2.2. Ancient Phylogenetic Origin of Neuroglobin Ngb sequences are now known for many mammalian and fish species (Fig. 2) [15,35, 41–43]. Our recent finding of Ngb orthologs in frogs and chicken show its conserved presence in other vertebrate taxa [44,45]. Phylogenetic reconstructions show that Ngb resembles nerve globins that have been found in some invertebrate species [15], while other invertebrate taxa appear to have recruited “normal” Hb genes for nerve cell function [46]. Recent results have shown the existence of a fifth globin type in lower vertebrates, which we named globin X [47]. Together with Ngb, invertebrate nerve and other intracellular globins, and the Ciona intestinalis globins [48], this novel globin of presently unknown function defines a distinct branch of the globin phylogenetic tree. Thus, Ngb is representative of an old globin lineage, which already existed before the separation of Protostomia and Deuterostomia more than 600 million years ago (Fig. 1). Ngb sequence conservation during mammalian evolution has been high, with an evolutionary rate approximately threefold slower than in Mb and Hb [49]. This observation suggests a strongly selected, important function of Ngb in vertebrates.

208

T. Hankeln and T. Burmester

2.3. Neuroglobin Expression Patterns Since the initial discovery of Ngb mRNA in the mammalian brain [15], several studies have confirmed the widespread expression of both Ngb mRNA and protein in nerve cells of the central and peripheral nervous system in mammals [41,50–54]. Most studies agreed that Ngb is exclusively expressed in the cytoplasm of neurons, but not in glia [52,54–57], which may be explained by the presence of candidate neuron-restrictive silencer elements in the Ngb gene region [56,58]. Reported trace amounts of Ngb in cultured astrocytes is a surprising finding that needs to be confirmed. In some studies [50,53], virtually all neurons appear to be Ngb-positive, albeit at regionally substantially different expression intensities. These varying expression levels reconcile these data with other studies that have reported a more focal expression pattern of Ngb [51,52]. In situ hybridization studies in the zebrafish (Danio rerio) showed a global Ngb expression in fish brain neurons, which is largely consistent with the pattern found in mammals [35]. In addition to the central nervous system (CNS) and the peripheral nervous system (PNS), mammalian Ngb is expressed in endocrine tissues such as the adenohypophysis, adrenal gland, testes [50], and the pancreatic islets of Langerhans [52]. Like neurons, these cell types are known to be metabolically highly active, which appears to be a general feature of Ngb expression sites. While the average Ngb protein content in total mouse brain has initially been estimated to be rather low (ca. 1 M; [15]), there are currently no quantitative data on regional and intracellular variations of Ngb expression. We found that the retina of the mammalian eye is a major site of Ngb expression [59]. Ngb concentrations in mouse total retina extracts amount to an estimated 100 M, and may be higher in the distinct cell layers that contain the Ngb protein. In the retina, Ngb levels thus almost approach Mb content in muscle cells, which usually range from 100 to 350 M [60]. In the vascular mammalian retina of rat and mice, Ngb is found in the inner and outer plexiform cell layers, the ganglion cell layer, and in the inner segments of the photoreceptors, which again coincides with regions of high metabolism, mitochondria content, and O2 demand [59,61]. The avascular retina of other mammalian species like the guinea pig, however, lacks the deep retinal and inner capillaries, such that only the inner segments of the photoreceptors adjacent to the choroidal capillaries display oxidative metabolism. Correspondingly, Ngb expression and the presence of mitochondria in avascular retinae are both restricted to the inner segments [62]. High retinal Ngb expression has recently been confirmed in zebrafish [35]. In Danio, we also noticed an additional Ngb expression site, namely the chloride cells of the gills, which are known to sustain high metabolic rates during regulation of osmolarity.

2.4. Neuroglobin Regulation and Medical Implications Given that Ngb may be instrumental in sustaining cellular O2 levels, it is consequent to study its regulation under conditions of hypoxic and ischemic stress. In cultured rat cortical neurons, Ngb mRNA and protein were shown to be upregulated maximally 2.5-fold after 16 hours of anoxia [55]. Cell culture hypoxia experiments may however be interpreted critically, since the O2 conditions achieved in the cell incubator may in reality

Neuroglobin and Cytoglobin

209

merely reflect the rather low physiological O2 tensions normally present in tissues like the brain [63]. In contrast, other authors [51,64] did not find any upregulation of Ngb mRNA in brains in vivo after prolonged (up to 14 days) exposure of mice to moderate hypoxia (10% O2 ) and after short-term treatment (2 hours, 7.6% O2 ). Our own hypoxia and ischemia experiments confirm a lack of Ngb upregulation in brain tissue. However, most mammals are not adapted to cope with hypoxic conditions, and, therefore, a physiological response may not be expected. By contrast, in the brain of zebrafish, which naturally live in tropical waters with low and fluctuating oxygen concentrations, Ngb is significantly upregulated (more than fivefold on the protein level) under severe hypoxia (48 hours, ∼4% O2 ) (Anja Roesner, Thomas Hankeln, Thorsten Burmester, unpublished data). Interestingly, mammalian Ngb genes lack conserved hypoxia-responsive sequence elements [49], which argues against a direct transcriptional hypoxia response of Ngb mediated by the hypoxia-inducible “master” transcription factor HIF-1 (hypoxia-inducible factor-1). However, the moderate hypoxia response of Ngb in cell culture was reported to be dependent on the mitogen-activated protein kinase (MAPK) signal transduction pathway [65], which is known to interact with the HIF-pathway via the recruitment of p300/CREB transcriptional coactivator. In addition to hypoxia, Ngb seems to be moderately upregulated in cell cultures by the addition of hemin, the ferric chloride salt of heme, which is already known to transcriptionally activate Mb and Hb [65]. This response of Ngb seems to proceed via the soluble guanylate cyclase–protein kinase G (sGC–PKG) signal transduction pathway. In the mouse immortalized hippocampal cell line HN33, antisense-mediated downregulation of Ngb leads to decreased levels of cell survival under hypoxia, while overexpression of Ngb in the same cells improves cell survival. This finding suggests that Ngb may exert some protective effect under hypoxic stress in the nervous system [55]. It is not clear, however, if neuroprotection is due to an O2 supply function of Ngb or some other function, like the binding of noxious reactive oxygen species (see below). Recently, it was reported that Ngb is also able to promote neurogenic survival in vivo [66]: in mice, intracerebral administration of an Ngb antisense oligodeoxynucleotide increases infarct size by twofold and worsens neurological outcome after an induced focal ischemia. In turn, an adeno-associated-virus (AAV)-mediated Ngb overexpression ameliorates ischemic pathology [66]. In a rat model of transient global brain ischemia, we could not observe an upregulation of Ngb (Rainald Schmidt-Kastner, Mark Haberkamp, Thomas Hankeln and Thorsten Burmester, unpublished). This, however, may not be surprising in mammals, which are not adapted to efficiently fight ischemic conditions. Nevertheless, the studies indicate that Ngb represents a candidate target for diagnosis and, possibly, therapy of stroke and of neurodegenerative disorders, which are known to be associated with hypoxia or increased levels of reactive oxygen species. Ngb was also reported to decrease with age in several cortical brain regions of 12–24-months-old mice, opening up the possibility that this increases the susceptibility of the aging brain toward stroke and neurodegeneration [57]. A recent publication [67] reports the presence of Ngb in the cerebrospinal fluid from three of nine females (but not males) with chronic pain, which clearly requires more data for explanation. Ngb has also been delivered to Langerhans’ islet cells by protein transduction and enhanced their survival, thus potentially increasing the quality of islets during transplantation for type 1 diabetes treatment [68]. The cellular protection mediated by Ngb thus also pertains to endocrine cells.

210

T. Hankeln and T. Burmester

3. CYTOGLOBIN: MUSCLE MYOGLOBIN’S BROTHER IN FIBROBLASTS AND NEURONS 3.1. Cytoglobin Structure and Ligand Binding Cygb shares about 30% amino acid sequence identity with Mb, pointing at a common recent evolutionary ancestry [16–18]. Compared to Mb, mammalian Cygb is unusually long containing 190 amino acids, owing to extensions of about 20 amino acids at both the N- and the C-terminus (Fig. 2). Part of the N-terminal extension may be explained by sequence motif duplication, while the C-terminal extension partly derives from a small additional exon, which has been recruited during mammalian evolution and is lacking in fish Cygb sequences (Fig. 2). Irrespective of these functionally elusive terminal extensions, which are also observed in some invertebrate globins, Cygb features the sequence hallmarks of a standard globin, e.g., the key residues Phe(CD1), His(E7), and His(F8). The crystal structure [69,70] ultimately proves that the Cygb core folds as a classic globin. Unfortunately, however, no interpretable electron density data could be obtained so far from the extended Cygb termini. In agreement with spectroscopic data [18,71], the crystal structure proves that Cygb is also a hexacoordinated globin (Fig. 3). Moreover, Cygb displays an unprecedented large apolar protein matrix cavity next to the heme, which is connected to the exterior and may provide a special “ligand tunneling” pathway [69,72]. Ligand-binding kinetics of Cygb are – as in the case of Ngb – determined by the comparatively slow phase of displacing the internal His(E7) ligand, before an external gaseous ligand can rapidly bind to the iron atom [28,29]. The resultant O2 affinity of Cygb is also in the range of 1 Torr [18,20,36]. Cygb also shares other kinetic features with Ngb, such as high thermal stability [32]. Like Ngb, Cygb may form an internal disulfide bridge (albeit at different positions, between Cys(B2) and Cys(E9); cf. Fig. 2). Reduction of this bond lowers O2 affinity of Cygb only moderately, by about twofold [20]. While there is no evidence for intermolecular disulfide bonds, biochemical and crystallographic data suggest that full-length Cygb might act as a homodimeric protein, while the truncated form is monomeric [20,69]. This is in agreement with the cooperativity in O2 binding, as measured for Cygb by equilibrium methods [73].

3.2. Cytoglobin Relationships Reconstructions of globin phylogeny confirmed that Cygb is distantly related to vertebrate Mbs (Fig. 1), with which it may have shared a common ancestor before the split of jawless and jawed vertebrates about 450 million years ago [17]. Independent evidence for this proposed “relationship by gene duplication” comes from human genome data, showing that Cygb on chromosome 17q25 and Mb on chromosome 22q12 are both parts of paralogous gene groups that have been formed by an ancient large-scale duplication event [74]. Cygb sequences are known from various vertebrates including man, mouse, rat, several fish species, chicken, and frog [43–45]. In bony fishes, we have obtained evidence for duplicated paralogous Cygb genes, making the evolution of this globin type

Neuroglobin and Cytoglobin

211

rather complex [75]. Cygb is very slowly evolving in mammals, and is even slightly more conservative than Ngb [49], which indirectly points toward a strongly selected function.

3.3. Gene Expression Patterns Cytoglobin mRNA was originally detected by Northern blot hybridization in essentially all tissues of the mammalian body, indicating a very widespread expression pattern [17,18]. At the same time, the rat ortholog of Cygb was independently isolated by a proteomics approach from the fibroblast-related stellate cells of the liver (and it was therefore dubbed “stellate cell activation–associated protein” STAP) [16,76]. Subsequently, it was reported that Cygb protein is localized exclusively within the cell nuclei in a wide variety of tissues [52]. Two recent studies have reinvestigated the expression pattern of Cygb [77,78]. Using independently derived antibodies, both report that Cygb is cytoplasmatically expressed in fibroblasts and fibroblast-related cell types in a broad variety of splanchnic organs like liver, heart, muscle, gut, kidney, lung, and pancreas. The earlier publications on the presence of Cygb (synonym STAP) in the fibroblast-like hepatic stellate cells were therefore confirmed. In addition, Cygb expression was also detected in bone osteoblasts and in tracheal chondroblasts, but not in mature osteocytes and chondrocytes [77]. In summary, the data by the majority support the specific expression of Cygb in the cytoplasm of cells that are actively engaged in the production of extracellular matrix components in visceral organs. Adding complexity to the Cygb expression pattern, our results suggest that Cygb is also expressed in specific, but ill-defined neuronal cell populations in the brain, as well as in peripheral nerve cells and in retinal neurons [77,61]. Here, Cygb immunostaining yielded signals in both the cytoplasm and the nucleus, possibly pointing at a specific role of Cygb in nervous tissues.

3.4. Regulation of Cytoglobin Expression and Medical Implications Cygb was originally described as a protein that is upregulated in activated, fibroblast-like hepatic stellate cells during liver fibrosis, and recent data demonstrate Cygb expression in stellate cells of fibrotic pancreas tissue during pancreatitis and in fibroblast-like cells from necrotic regions in the kidney after diet-induced chronic nephropathy in rats [16,78]. In primary cultures of rat hepatic stellate cells, Cygb expression is slightly augmented by addition of recombinant transforming growth factor  (TGF) and platelet-derived growth factor-B (PDGF-B), serum factors, which accelerate stellate cell activation [78]. Addition of protein kinase inhibitors suggests that Cygb may be regulated via a protein kinase C (PKC)-dependent signal transduction pathway. When NIH 3T3 fibroblast cells were transfected with a Cygb expression construct and, subsequently, collagen 1 (I) synthesis was induced by TGF, a substantial enhancement of collagen production was observed in the Cygb-transfected cells as compared to non-Cygb-expressing wild-type 3T3 cells [78]. This result suggests a stimulatory, yet undefined role of Cygb in collagen expression, a finding that is corroborated by the shutdown of Cygb expression during osteoblast and chondroblast maturation [77].

212

T. Hankeln and T. Burmester

By exposing mice to hypoxic conditions, we have shown that Cygb is upregulated twoto threefold in heart and liver [77], which is in good agreement with the presence of conserved hypoxia-responsive sequence elements in the Cygb gene region [49]. Acute tissue hypoxia is a stimulatory signal in processes like osteogenesis, chondrogenesis, and wound healing [79,80], in which collagens are massively produced, thereby possibly creating a link between the above observations on Cygb regulation. Recently, Cygb was reported to also exert a protective effect on islet beta-cells by an unknown mechanism [81]. In summary, Cygb may have substantial biomedical impact due to its involvement in organ fibrosis and in the production of extracellular matrix collagens during normal tissue development and fibrotic pathogenesis.

4. TWO GLOBINS IN SEARCH OF THEIR ROLES IN THE FAMILY (AND IN THE CELL) Theoretically, and partly in analogy to other globins, we can consider several possible cellular functions for Ngb and Cygb (Fig. 4) and discuss them in light of the currently available data: a. As with Mb and many other Mb-type molecules, both novel globins could either store O2 for long or short time, or assist in the diffusion of O2 within the cell toward the mitochondria [7]. b. Both globins could function as oxygen sensor proteins, which have been well studied in bacteria [82]. Alternatively, they could be involved in other intracellular signaling pathways. O2

O2 (B)

(A)

(D) (C)

O2 [O·]

+

(F)

(E)

NO⋅ + O2

2 NADH/H + O2

2 H2O + 2 NAD

H2O +

–

NO3

O2 O2

OH

Fig. 4. Hypothetical functions of intracellular globins. See text for further explanations. Colors: green, globin molecule; brown, heme group; red, oxygen ligand; pink, mitochondrium; yellow, oxygen-needing enzyme.

Neuroglobin and Cytoglobin

213

c. Ngb and Cygb might act as terminal oxidases, regenerating NAD+ to support glycolysis and sustain ATP production under hypoxic conditions, as proposed for maize hemoglobin [83]. d. Both globins could be instrumental as scavengers of reactive oxygen or nitrogen species, which are produced, e.g., after reperfusion/reoxygenation following ischemia. e. As proven for Mb in mammalian muscle cells [11], they could possess dioxygenase activity, converting harmful excess NO into innocuous nitrate. f. Several cytoplasmatic enzymes use molecular O2 for chemical reactions, and globins like Ngb or Cygb could supply these other enzymes with adequate amounts of O2 . In the case of Ngb, indirect evidence favors scenario a. As initially pointed out [15], it makes sense for highly O2 -demanding and metabolically most active cells like neurons to possess a specialized respiratory protein, which helps to sustain aerobic metabolism, possibly by acting as a short-term O2 store on the encounter of acute fluctuations of O2 levels. The correlation between Ngb expression levels and metabolic activity is intriguing, most notably in the case of the vertebrate retina, where Ngb localization perfectly matches the layers of strong O2 consumption and is strictly associated with the mitochondria [59,62]. Thus, there is little doubt that Ngb is linked to the oxidative metabolism. While the overall amount of Ngb protein in the mammalian brain may be low, there are pronounced differences in expression levels. For certain highly active brain regions, and of course the most strongly Ngb-expressing retina, Ngb levels may in fact be sufficient for an Mb-like O2 supply function. Mathematical calculations in a simplified model of the retina suggest that Ngb levels may be too low to significantly facilitate the diffusion of O2 to the mitochondria, but are high enough to support a short-term O2 storage function of Ngb [73]. The somewhat lower O2 affinity values measured for Ngb (P50 between 2 and 7 Torr; [15,25,36]), as compared to the P50 of 2.5 Torr for myoglobin [7], have been interpreted to contradict an O2 storage function of Ngb in neuronal tissues, which were presumed to have O2 tensions close to 0 [64]. However, literature values of O2 partial pressure range between 5 and 40 Torr in regions of vertebrate brain [63] and in the vertebrate retina [84]. These partial pressures well enable the loading of Ngb with O2 . It also must be clearly pointed out, however, that Ngb rapidly autoxidizes to metNgb(Fe3+ ) [19]. To function as an O2 supply, we therefore have to postulate a yet unidentified Ngb-reducing enzymatic activity in neurons, in analogy to Mb reductase in muscle. The neuroprotective effects of Ngb (over)expression [55,66] are certainly in line with an O2 supply function. Also, a short-term storage role does not necessarily require a pronounced upregulation of the globin under hypoxia, as also shown for Mb in hypoxic muscle during athletic training [85]. One may even doubt that any gross physiological hypoxia response on the globin level can be expected in those mammalian species, which are not adapted to regularly encounter hypoxic conditions in their environment. We therefore studied Ngb levels in the subterranean blind mole rat Spalax ehrenbergi, a mammal that dwells in underground burrows and is able to survive extended periods of extreme hypoxia without neuronal damage [86]. Spalax brains reveal a constitutively higher expression of Ngb mRNA and protein as compared to laboratory rats, arguing for the adaptive involvement of Ngb in the unique hypoxia tolerance of this animal (Aaron Avivi, Frank Gerlach, Thorsten Burmester, Stefan Reuss, Eviatar Nevo and Thomas Hankeln, unpublished data).

214

T. Hankeln and T. Burmester

We regard the current evidence for Ngb as a signaling protein (scenario b) as rather weak. Globin-coupled O2 sensors [82] have until now only been found in archaea and bacteria and usually require signal-transducing protein domains in addition to the globin part. A role of both Ngb (and Cygb) in O2 sensing also appears unlikely, because at least in vitro, their observed ligand affinities [15,19] are substantially higher than those of functional sensors like the HIF prolyl-4-hydroxylases, which are known to work under natural cellular O2 concentrations [87]. It has been proposed on the basis of in vitro studies using surface plasmon resonance that oxidized Ngb(Fe3+ ) is involved in intracellular signaling under oxidative stress by inhibiting release of GDP from G proteins and triggering release of the G complex, thereby enhancing cell survival [88,89]. The interaction of Ngb with G proteins was postulated on the basis of a proposed sequence similarity between Ngb and regulators of G protein signaling (RGS) and RGS domains of G protein–coupled receptor kinases. The evidence given for this similarity, however, is weak, and it will require further studies to substantiate this proposed Ngb involvement in signal transduction in vivo. A recent paper by the same authors has shown that the GDP dissociation–inhibiting feature is not conserved for zebrafish Ngb [90], shedding additional doubt on the biological relevance of this interaction. A number of studies report yet other, functionally very diverse, potential interacting partners of Ngb, e.g., the membrane-bound 2 subunit of the Na, K-ATPase ion pump [91], the lipid raft protein flotillin-1 [92] and the extracellular cysteine proteinase inhibitor cystatin C [93]. Since these data are difficult to reconcile with the known cellular and intracellular distribution of Ngb, and have not been confirmed by independent evidence, it must remain rather uncertain whether these interactions found in vitro help to explain the function of Ngb in vivo. The role of Ngb as a terminal oxidase for sustaining glycolysis under hypoxia (scenario c) has not yet been investigated. According to the “lactate shuttle” hypothesis, it is currently believed that in normoxia and even under functional activation of a brain region, glycolysis occurs predominantly in the astroglia, which produces substantial amounts of lactate [94]. This lactate is taken up by neurons, which appear to have a preference to oxidize imported lactate instead of producing lactate/pyruvate by their own glycolysis. Under this scenario, Ngb as a purely neuronal protein should not play a substantial role in glycolytic energy production. Energy depletion under hypoxia, however, may stimulate enhanced glucose oxidation in neurons, compensating for a reduction in lactate supply by astroglia. A function of Ngb as a scavenger of reactive oxygen species (ROS) and nitrogen species would be consistent with the neuroprotective effect of Ngb after ischemia and reperfusion of brain tissue [55,66] when such harmful molecules are known to form. A possible chemistry showing how Ngb(Fe2+ ) might react first with NO and then with peroxynitrite has been published [32]. It remains to be shown whether this chemistry works in vivo at physiological Ngb and NO concentrations. Our own data (Tilmann Laufs, Heidrun Witan, Sigrid Saaler-Reinhardt, Thorsten Burmester and Thomas Hankeln, unpublished) reveal that Ngb expression levels in the developing mouse nervous system do not peak around birth time, when ROS increase due to the rise in O2 from intrauterine to external conditions and when known ROS defense proteins are reported to show peak performance [95]. In addition, exposure of cultured neurons to very severe ROS stress induced by the herbicide paraquat did not trigger an upregulation of Ngb (Tilmann Laufs, Gabriele Schmuck, Thorsten Burmester and Thomas Hankeln,

Neuroglobin and Cytoglobin

215

unpublished). If ROS scavenging is the predominant function of Ngb, we would expect it to be acting primarily on ROS species, which are more or less constitutively produced by mitochondria. An Mb-style NO dioxygenase activity has recently been proposed for Ngb based on kinetical measurements for the O2 and NO ligands [40]. As for the O2 supply function, this NO dioxygenase activity crucially depends on a yet unidentified metNgb(Fe3+ )reductase activity in neurons. The authors [40] preferentially discuss the potential dioxygenase function of Ngb in the context of ischemic insults, when O2 is low and NO levels are increased. However, most current expression data argue for a housekeeping function of Ngb, rather than a stress-induced role. In this respect, we note that Ngb and NOproducing synthases do not seem to strictly colocalize in neurons under nonpathological conditions [50]. In addition, Ngb expression was not increased by the NO donor sodium nitroprusside in cultured cells [55], although an upregulation is not necessarily required for a housekeeping dioxygenase function. Less data are currently available for Cygb, and several possible cellular roles can still only be hypothesized for this protein. On the one hand, Cygb shows a globin fold and an O2 affinity reminiscent of Mb, to which it is phylogenetically related. On the other hand, Cygb features peculiarities like its N- and C-terminal extensions, which might be mediating special protein–protein interactions, plus the heme hexacoordination, a redox-dependent O2 affinity, and special cavities for ligand diffusion. On the basis of the expression pattern, which is not as global as suggested by its name, we envisage that Cygb will perform distinct functions in the cytoplasm of fibroblast-like cells, and in the nuclear and cytoplasmic compartments of yet to be defined nerve cell populations [77]. This hypothesis is supported by the presence of two paralogous Cygb genes in fish, one of which is predominantly expressed in the brain [75]. Cygb certainly has a most prominent role in cells of the fibroblast lineage. Any proposed function here must take into account the following: (i) fibroblast-like cells are not known to be metabolically active in general, but engage in the massive production of extracellular matrix proteins like collagen; (ii) collagen synthesis consumes molecular O2 during hydroxylation at proline residues, performed by dedicated prolyl-hydroxylases; (iii) Cygb (like collagen production) is upregulated by hypoxia [77]; and (iv) Cygb overexpression is able to somehow enhance collagen expression [78]. We therefore hypothesize that Cygb could be involved in collagen production. One formal possibility is that Cygb provides O2 directly to the collagen prolyl-hydroxylase (scenario f in Fig. 4), although the lower O2 affinity of collagen prolyl-hydroxylase [87] appears to be at odds with this mechanism. Alternatively, Cygb might participate in some unknown signaling pathway, ultimately augmenting collagen synthesis. While Ngb expression patterns are clearly related to oxidative metabolism (see above), no such correlation could be found for Cygb in the mammalian retina [61]. This observation argues against an important role of Cygb in cellular respiration. Interestingly, the neuronal cells that harbor Cygb are also positive for the expression of neuronal NO synthase (nNOS) and thus produce NO (Stefan Reuss, Sylvia Wystub, Thorsten Burmester and Thomas Hankeln, unpublished). We therefore hypothesize that Cygb in these cells either provides O2 to nNOS for making of NO, or detoxifies NO as a dioxygenase (just like its “brother” Mb). Although vertebrate Ngb and Cygb were discovered only recently, many laboratories have since contributed significantly to our knowledge on the biochemistry, structure,

216

T. Hankeln and T. Burmester

comparative physiology, and molecular genetics of these proteins. Future studies will focus on the ultimate understanding of the function(s) of intracellular globins in the metabolism of eukaryotic cells and on their biomedical impact.

ACKNOWLEDGMENTS The authors are especially grateful to all their colleagues from the mutual EU project “Neuroglobin and the survival of the neuron,” namely Martino Bolognesi and Daniele De Sanctis (Milano), Michael C. Marden and Laurent Kiger (Paris), Luc Moens and Sylvia Dewilde (Antwerp), Eviatar Nevo and Aaron Avivi (Haifa), Roy E. Weber and Angela Fago (Aarhus), Bettina Ebner, Christine Fuchs, Frank Gerlach, Mark Haberkamp, Tilmann L. Laufs, Stephanie Mitz, Anja Roesner, Marc Schmidt, Bettina Weich, Sylvia Wystub, Sigrid Saaler-Reinhardt and Stefan Reuss (Mainz), who contributed a wealth of data on neuro- and cytoglobin. Our work has been supported by grants from DFG (Ha2103/3 and Bu956/5), the European Union (QLG3-CT2002-01548), the Stiftung für Innovation Rheinland-Pfalz (695), and the Fonds der Chemischen Industrie.

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24]

Kendrew, J.C. (1963) Science 139, 1259–1266. Perutz, M.F. (1979) Ann. Rev. Biochem. 48, 327–386. Bolognesi, M., Bordo, D., Rizzi, M. et al. (1997) Prog. Biophys. Mol. Biol. 68, 29–68. Milani, M., Pesce, A., Nardini, M. et al. (2005) J. Inorg. Biochem. 99, 97–109. Dickerson, R.E. and Geis, I. (1983) Hemoglobin: Structure, Function, Evolution, and Pathology. Menlo Park, CA: Benjamin/Cummings. Weber, R.E. and Vinogradov, S.N. (2001) Physiol. Rev. 81, 569–628. Wittenberg, J.B. and Wittenberg, B.A. (2003) J. Exp. Biol. 206, 2011–2020. Hardison, R.C. (1996) Proc. Natl. Acad. Sci. U.S.A. 93, 5675–5679. Vinogradov, S.N., Hoogewijs, D., Bailly, X. et al. (2005) Proc. Natl. Acad. Sci. U.S.A. 102, 11385–11389. Merx, M.W., Flögel, U., Stumpe, T. et al. (2001) FASEB J. 15, 1077–1079. Flögel, U., Merx, M.W., Gödecke, A. et al. (2001) Proc. Natl. Acad. Sci. U.S.A. 98, 735–740. Burmester, T. and Hankeln, T. (1999) Mol. Biol. Evol. 16, 1809–1811. Hankeln, T., Jaenicke, V., Kiger, L. et al. (2002) J. Biol. Chem. 277, 29012–290127. Burmester, T., Storf, J., Hasenjager, A. et al. (2006) FEBS J. 273, 468–480. Burmester, T., Weich, B., Reinhardt, S. and Hankeln, T. (2000) Nature 407, 520–523. Kawada, N., Kristensen, D.B., Asahina, K. et al. (2001) J. Biol. Chem. 276, 25318–25323. Burmester, T., Ebner, B., Weich, B. and Hankeln, T. (2002) Mol. Biol. Evol. 19, 416–421. Trent, J.T., III and Hargrove, M.S. (2002) J. Biol. Chem. 277, 19538–19545. Dewilde, S., Kiger, L., Burmester, T. et al. (2001) J. Biol. Chem. 276, 38949–38955. Hamdane, D., Kiger, L., Dewilde, S. et al. (2003) J. Biol. Chem. 278, 51713–51721. Pesce, A., Dewilde, S., Nardini, M. et al. (2003) Structure (Camb) 11, 1087–1095. Vallone, B., Nienhaus, K., Brunori, M. and Nienhaus, G.U. (2004) Proteins 56, 85–92. Couture, M., Burmester, D.L., Hankeln, T. and Rousseau, D.L. (2001) J. Biol. Chem. 276, 36377–36382. Trent, J.T., III, Watts, R.A. and Hargrove, M.S. (2001) J. Biol. Chem. 276, 30106–30110.

Neuroglobin and Cytoglobin

217

[25] Kriegl, J.M., Bhattacharyya, A.J., Nienhaus, K. et al. (2002) Proc. Natl. Acad. Sci. U.S.A. 99, 7992–7997. [26] Du, W., Syvitski, R., Dewilde, S. et al. (2003) J. Am. Chem. Soc. 125, 8080–8081. [27] Uno, T. Ryu, D., Tsutsumi, H. et al. (2004) J. Biol. Chem. 279, 5886–5893. [28] Pesce, A. Bolognesi, M. Bocedi, A. et al. (2002) EMBO Rep. 3, 1146–1451. [29] Kundu, S., Trent J.T. III and Hargrove, M.S. (2003) Trends Plant Sci. 8, 387–393. [30] Nienhaus, K., Kriegl, J.M. and Nienhaus, G.U. (2004) J. Biol. Chem. 279, 22944–22952. [31] Uzan, J., Dewilde, S., Burmester, T. et al. (2004) Biophys. J. 87, 1196–1204. [32] Herold, S., Fago, A., Weber, R.E. et al. (2004) J. Biol. Chem. 279, 22841–22847. [33] Hamdane, D., Kiger, L., Dewilde, S. et al. (2005) FEBS J. 2076–2084. [34] Vallone, B., Nienhaus, K., Matthes, A. et al. (2004) Proc. Natl Acad. Sci. U.S.A. 101, 17351–17356. [35] Fuchs, C., Heib, V., Kiger, L. et al. (2004) J. Biol. Chem. 279, 24116–24122. [36] Fago, A., Hundahl, C., Dewilde, S. et al. (2004) J. Biol. Chem. 44417–44426. [37] Becker, L.B. (2004) Cardiovasc. Res. 61, 461–470. [38] Van Doorslaer, S., Dewilde, S., Kiger, L. et al. (2003) J. Biol. Chem. 278, 4919–4925. [39] Trandafir, F., Van Doorslaer, S., Dewilde, S. and Moens, L. (2004) Biochim. Biophys. Acta. 1702, 153–161. [40] Brunori, M., Giuffre, A., Nienhaus, K. et al. (2005) Proc. Natl. Acid. Sci. U.S.A. 8483–8488. [41] Zhang, C., Wang, C., Deng, M. et al. (2002) Biochem. Biophys. Res. Commun. 290, 1411–1419. [42] Awenius, C., Hankeln, T. and Burmester, T. (2001) Biochem. Biophys. Res. Commun. 287, 418–421. [43] Burmester, T., Haberkamp, M., Mitz, S. et al. (2004) IUBMB Life 703–707. [44] Kugelstadt, D., Haberkamp, M., Hankeln, T. and Burmester, T. (2004) Biochem. Biophys. Res. Commun. 325, 719–725. [45] Fuchs, C., Burmester, T. and Hankeln, T. (2006) Cytogenet. Genome Res. 112, 296–306. [46] Dewilde, S., Ebner, B., Vinck, E. et al. (2005) J. Biol. Chem. 281, 5364–5372. [47] Roesner, A., Fuchs, C., Hankeln, T. and Burmester, T. (2005) Mol. Biol. Evol. 22, 12–20. [48] Ebner, B., Burmester, T. and Hankeln, T. (2003) Mol. Biol. Evol. 20, 1521–1523. [49] Wystub, S., Ebner, B., Fuchs, C. et al. (2004) Cytogenet. Genome Res. 105, 65–78. [50] Reuss, S., Saaler-Reinhardt, S., Weich, B. et al. (2002) Neuroscience 115, 645–656. [51] Mammen, P.P., Shelton, J.M., Goetsch, S.C. et al. (2002) J. Histochem. Cytochem. 50, 1591–1598. [52] Geuens, E., Brouns, I., Flamez, D. et al. (2003) J. Biol. Chem. 278, 30417–30420. [53] Wystub, S., Laufs, T., Schmidt, M. et al. (2003) Neurosci. Lett. 346, 114–116. [54] Hankeln, T., Wystub, S., Laufs, T. et al. (2004) IUBMB Life 56, 671–679. [55] Sun, Y., Jin, K., Mao, X.O. et al. (2001) Proc. Natl. Acad. Sci. U.S.A. 98, 15306–15311. [56] Laufs, T.L., Wystub, S., Reuss, S. et al. (2004) Neurosci. Lett. 362, 83–86. [57] Sun, Y., Jin, K., Mao, X.O. et al. (2005) Neurobiol. Aging. 26, 275–278. [58] Chen, X.Q., Qin, L.Y., Zhang, C.G. et al. (2005) Glia 50, 182–186. [59] Schmidt, M., Giessl, A., Laufs, T. et al. (2003) J. Biol. Chem. 278, 1932–1935. [60] Wittenberg, J.B. and Wittenberg, B.A. (1990) Ann. Rev. Biophys. Biophys. Chem. 217–241. [61] Schmidt, M. Laufs, T. Reuss, S. et al. (2005) Neurosci Lett. 207–211. [62] Bentmann, A., Schmidt, M., Reuss, S. et al. (2005) J. Biol. Chem. 280, 20660–20665. [63] Erecinska, M. and Silver, I.A. (2001) Resp. Physiol. 128, 263–276. [64] Hundahl, C., Stoltenberg, M., Fago A. et al. (2005) Neuropathol. Appl. Neurobiol. 31, 610–617. [65] Zhu, Y., Sun, Y., Jin, K. and Greenberg, D.A. (2002) Blood 100, 2494–2498. [66] Sun, Y., Jin, K., Peel, A. et al. (2003) Proc. Natl. Acad. Sci. U.S.A. 100, 3497–3500. [67] Casado, B., Pannell, L.K., Whalen, G. et al. (2005) Proteome Sci. 2.

218

T. Hankeln and T. Burmester

[68] [69] [70] [71] [72]

Mendoza, V., Klein, D., Ichii, H. et al. (2005) Transplant Proc. 37, 237–240. de Sanctis, D., Dewilde, S., Pesce, A. et al. (2004) J. Mol. Biol. 336, 917–927. Sugimoto, H., Makino, M., Sawai, H. et al. (2004) J. Mol. Biol. 339, 873–885. Sawai, H., Kawada, N., Yoshizato, K. et al. (2003) Biochemistry 42, 5133–5142. De Sanctis, D., Dewilde, S., Pesce, A. et al. (2004) Biochem. Biophys. Res. Commun. 316, 1217–1221. Fago, A., Hundahl, C., Malte, H. and Weber, R.E. (2004) IUBMB Life 56, 689–696. McLysaght, A., Hokamp, K. and Wolfe, K.H. (2002) Nat. Genet. 31, 200–204. Fuchs, C., Luckhardt, A., Gerlach, F. et al. (2005) Biochem. Biophys. Res. Commun. 337, 216–223. Asahina, K., Kawada, N., Kristensen, D.B. et al. (2002) Biochim. Biophys. Acta 1577, 471–475. Schmidt, M., Gerlach, F., Avivi, A. et al. (2004) J. Biol. Chem. 279, 8063–8069. Nakatani, K., Okuyama, H., Shimahara, Y. et al. (2004) Lab. Invest. 84, 91–101. Schipani, E., Ryan, H.E., Didrickson, S. et al. (2001) Genes Dev. 15, 2865–2876. Tandara, A.A. and Mustoe, T.A. (2004) World J. Surg. 28, 294–300. Stagner, J.I., Parthasarathy, S.N., Wyler, K. and Parthasarathy, R.N. (2005) Transplant Proc. 37, 3452–3453. Freitas, T.A.K., Saito, J.A., Hou, S. and Alam, M. (2005) J. Inorg. Biochem. 99, 23–33. Sowa, A.W., Duff, S.M.G., Guy, P.A. and Hill, R.D. (1998) Proc. Natl. Acad. Sci. U.S.A. 17, 10317–10321. Yu, D.Y. and Cringle, S.J. (2001) Prog. Retin. Eye Res. 20, 175–208. Hoppeler, H. and Vogt, M. (2001) J. Exp. Biol. 204, 3133–3139. E. Nevo, E. Ivanitskaya and A. Beiles. (2001) Adaptive Radiation of Blind Subterranean Mole Rats. Backhuys, Leiden. Hirsilä, M., Koivunen, P., Günzler, V. et al. (2003) J. Biol. Chem. 278, 30772–30780. Wakasugi, K., Nakano, T. and Morishima, I. (2003) J. Biol. Chem. 278, 36505–36512. Wakasugi, K., Kitatsuji, C. and Morishima, I. (2005) Ann. NY Acad. Sci. 1053, 200–230. Wakasugi, K. and Morishima, I. (2005) Biochemistry 44, 2943–2948. Xu, W.L., Wang, C.L., Liao, Z.Y. et al. (2003) Sheng Wu Hua Xue Yu Sheng Wu Wu Li Xue Bao (Shanghai) 35, 823–828. Wakasugi, K., Nakano, T., Kitatsuji, C. and Morishima, I. (2004) Biochem. Biophys. Res. Commun. 318, 453–460. Wakasugi, K., Nakano, T. and Morishima, I. (2004) Biochemistry 43, 5119–5125. Peng, L., Arystarkhova, E. and Sweadner, K.J. (1998) Glia 24, 257–271. Khan, J.Y. and Black, S.M. (2003) Pediatr. Res. 54, 77–82.

[73] [74] [75] [76] [77] [78] [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95]

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 9

Extreme pH Sensitivity in the Binding of Oxygen to Some Fish Hemoglobins: The Root Effect T. Brittain School of Biological Sciences, University of Auckland, Auckland, New Zealand

Fish species constitute over half of all vertebrate species. They inhabit a vast range of very differently oxygenated environments from polar to equatorial regions, from fresh water streams to highly saline waters of deep oceanic trenches. Individuals not only occupy a range of environments but may also experience a range of changing environments. Unlike land vertebrates, which normally experience an essentially unchanging oxygen environment and hence utilize a single form of hemoglobin, optimized for oxygen transport under this condition, fish typically possess multiple hemoglobins, each with particular oxygen-binding properties. One particular form of hemoglobin found in many fish species is named after its discoverer; the Root effect hemoglobin [1], which has been a source of considerable scientific interest for the past 80 years. The Root effect hemoglobins are notable for their extreme sensitivity to pH. Although the physiological significance of this unusual property now seems to be generally agreed upon, the molecular mechanism and in particular the identity of the particular amino acids within the hemoglobin molecule that provide this phenomenon are still matters of scientific contention. This long search for the “molecular holy grail” may well turn out to be just that, as the Root effect is in fact displayed to different extents by different hemoglobins and may well arise not from a unique set of amino acids but from a number of different amino acid combinations that exhibit similar fundamental characteristics. It is the goal of this chapter to put into context the occurrence, nature, and characteristics of Root effect hemoglobins and to give an up-to-date perspective on the ongoing search for the molecular origins of the Root effect. In its simplest form, the Root effect can be described as an extreme pH sensitivity in oxygen binding to the hemoglobins of some fish. In these cases, oxygen affinity drastically and continually decreases as the pH is lowered, in contrast to the case of the hemoglobins of higher vertebrates. Furthermore, this dramatic decrease in oxygen affinity, at acidic pH, is accompanied by a loss of cooperative behavior. In order to better understand this phenomenon and its likely molecular origins, it is necessary to consider this process within the context of the detailed framework that has been derived to describe the oxygen-binding process in mammalian hemoglobins.

220

T. Brittain

1. BACKGROUND The hemoglobin protein has an ancestry probably beginning with an ancient b-type cytochrome. The molecule’s origins can be traced back to at least 450 million years ago [2,3]. In the blood of present-day vertebrates, the hemoglobin molecule always consists of a 22 tetrameric structure in which two pairs of very similar globin proteins ( and ) combine in a noncovalent quaternary structure. Each protein subunit exhibits the characteristic globin fold and binds a protoheme IX prosthetic group within the protein cleft. In all cases, the heme group is held in place within the protein by a number of weak interactions with a range of amino acids. The only exception to this is that each globin subunit provides a heme iron–bound proximal histidine ligand residue that serves to further anchor the heme group to the protein. The normally six-coordinate ferrous iron atom contained within the heme group is restrained by the protein to be five-coordinate and high spin in the absence of bound oxygen (the deoxygenated form of the protein). Oxygen binds reversibly, in a nonlinear configuration, at the vacant sixth coordination site of the iron atom. The nonlinear configuration prevents oxygen oxidation of the ferrous iron atom and is mandated by the presence of another, distal, histidine residue provided by the protein. The strength of the binding of oxygen to the heme iron atom is further modulated by hydrogen bonding of the distal histidine to the bound oxygen molecule [4] and by the electrostatic field within the binding pocket produced by the surrounding protein amino acids [5–9]. These structures provide reversible oxygen binding to each subunit. However, hemoglobin is a tetrameric protein and shows a reasonably high level of subunit interactions via ion pairing and salt bridge formation between the subunits. Changes in the pattern of these interactions during the oxygenation process leads to a cooperative, sigmoidal, rather than hyperbolic oxygenbinding curve and reflects the fact that oxygen binding at each step in the binding process is sensitive to the state of ligation of the other subunits within the hemoglobin tetramer (the homotropic effect).

2. QUANTITATIVE ANALYSIS OF OXYGEN BINDING 2.1. The Homotropic Effect At the phenomenological level, the homotropic effect is evident in the observation of sigmoidal binding curves (Fig. 1A). These curves can themselves be characterized by two empirical parameters: (i) p50 – the oxygen partial pressure corresponding to half saturation of the oxygen-binding sites (for historical reasons, usually given in units of mm Hg partial pressure) and (ii) h – the Hill coefficient obtained from the plot originally derived by Hill [76] to describe oxygen binding to the multiple sites of hemoglobin: Hb + nO2 ←→ HbO2 n    = h∗ logpO2  + log K Hill plot  log 1− where K is the equilibrium constant for the overall oxygen-binding process and  is the fractional saturation at a particular oxygen concentration pO2 . With regard to the Hill

Root Effect Hemoglobins

221

(A) 1.0

Fractional Saturation

0.8

0.6

0.4

0.2

0.0

0

10

20

30

40

50

60

70

80

90

100

pO2 (mm Hg) (B) 3

2

Log (Y/1–Y)

1

KR 0

KT –1

–2

–3 0.0

0.5

1.0

1.5

2.0

2.5

3.0

Log (pO2)

Fig. 1. Graphical representations of the homotropic effect associated with oxygen binding to hemoglobin. The oxygen-binding curve for human hemoglobin is shown (A) together with a Hill plot (B) of the same data. Note that the lower unitary slope component of the Hill plot correlates with binding to the T state and the upper unitary slope component correlates with binding to the R state.

analysis of oxygen binding to hemoglobin, the p50 value describes the positioning of the curve on the [O2 ] axis whilst the value of h relates to the steepness of the sigmoidal curve at the p50 value. The mechanistic origins of the homotropic effect have been investigated from two distinct but complimentary perspectives. A qualitative structural explanation based on the three-dimensional structures of both the fully deoxygenated and oxygenated forms of the hemoglobin protein was proposed by Perutz [10–12]. This model

222

T. Brittain

envisages the process of oxygenation as a sequence of oxygen-binding events that are associated with subunit tertiary structural changes that break intersubunit charge–charge interactions, leading to a quaternary structural switch from the deoxygenated (T, tense state) to the oxygenated (R, relaxed state). As this proposal is based solely on the initial and final structures, it can give no indication of the sequence of events or characteristics of intermediate states. Nevertheless, this structural model has been very important in informing other quantitative models of homotropic oxygen binding to hemoglobin. A number of different quantitative models have been derived to describe the homotropic effect in hemoglobin. Each of these models proposes a different detailed mechanism for the production of oxygenated intermediates, with more or less emphasis on the roles of tertiary and quaternary structural effects [13–15]. At our present level of knowledge and the accuracy of the available experimental data, it is not possible to conclusively determine which model best fits the experimental data relating to the homotropic effect. Mainly for historical reasons, the model first proposed by Monod et al. [13] has been most often applied to studies of the homotropic effect in hemoglobin. Within this model, it is normally assumed that the hemoglobin molecule can exist in an equilibrium between two quaternary structures, which are identified as the oxygenated and deoxygenated structures determined by Perutz. This model is thus often referred to as the two-state model. Core to this model is the assumption that the subunits within any particular tetramer have identical, fixed oxygen affinities determined solely by the quaternary structure of the molecule as a whole. The R state has a high oxygen affinity (KR ) and the T state a low oxygen affinity (KT ). This model accounts for the observed sigmoidal oxygen-binding characteristics of hemoglobin, if, in the absence of oxygen, the T state is overwhelmingly favored and in the presence of saturating amounts of oxygen, the R state is favored. Under these conditions, the progress of saturation (Y ) of hemoglobin with oxygen can be expressed by the equation: Y=

LK T pO2 1 + KT pO2 3 + KR pO2 1 + KR pO2 3 L1 + KT pO2 4 + 1 + KR pO2 4

where L represents the equilibrium constant for the T/R equilibrium in the absence of oxygen. Using this model, it is possible to calculate the population of partially liganded intermediates. Using the parameters derived for human adult hemoglobin (L, KT , KR ), the model predicts that a concerted switch from predominantly T state to R state occurs during the binding of the four molecules of oxygen to any particular hemoglobin molecule occurs at around half saturation. In the case of other hemoglobins, however, the switch point can occur much earlier or later depending on the relative magnitudes of the oxygen affinities of the two states and the equilibrium constant (L). The parameters associated with this model can be obtained by curve-fitting of the oxygen-binding curve and can be visualized in the form of a Hill plot (Fig. 1B). In the Hill plot, the upper limb, of unitary slope, present at high oxygen concentrations is related to KR , whilst the lower, unitary slope, limb observed at low oxygen concentration is related to KT . Both the values can be assessed by simple extrapolation as shown. A measure of the interaction between the subunits of the protein is expressed by the slope of the central portion of the curve, which is referred to as the Hill coefficient (h). The Hill coefficient varies from a value of 1 when the subunits do not interact up to a value of 4 if the subunit interaction

Root Effect Hemoglobins

223

is infinite and all the subunits show identical oxygen affinity In human hemoglobin, under physiological conditions, h usually has a value of approximately 3. Although the homotropic effect describes the fundamental operation of vertebrate hemoglobins, the capacity of the oxygen transport system to respond to changing physiological demands requires a further level of modulation of the oxygen-binding properties of hemoglobin beyond that possible simply by the operation of the homotropic effect.

2.2. The Heterotropic Effect During short-term adaptation to such conditions as exercise stress, an immediate increase in oxygen off-loading to the muscles is obviously advantageous. Longer-term changes to oxygen affinity of hemoglobin, in situations such as lower oxygen availability or reduced circulatory flow, could also be advantageous to an organism. Hemoglobin oxygen-binding affinity does in fact respond to such situation by employing not the homotropic effect but rather the heterotropic effect. In the heterotropic effect, the binding of oxygen to the hemoglobin tetramer is affected by the presence of a third substance. In the case of exercise stress, both carbon dioxide and protons, produced by the onset of anaerobic metabolism, lead to a lowering of oxygen affinity in hemoglobin. In the case of longer-term adaptations, in general, the intracellular concentration of organic phosphates such as 2,3-diphosphoglycerate increase, and these too lead to a lowering of oxygen-binding affinity. These substances that modulate the hemoglobin protein affinity for oxygen are known as allosteric effectors [16–19]. Most vertebrate hemoglobins alter their oxygen-binding properties in response to changes in the concentration of the allosteric effector, the proton (Fig. 2). This pH sensitivity was recognized many years ago, and is now known as the Bohr effect, in

1.0

Fractional Saturation

0.8 1

0.6

2

0.4

0.2

0.0

3

0

20

40

60

80

100

120

pO2 (mm Hg)

Fig. 2. The effect of pH on the binding curve of oxygen to hemoglobin. The oxygen-binding curve is shown for human hemoglobin at alkaline (1) and slightly acid pH (2). The binding of oxygen to a Root effect hemoglobin at alkaline pH closely matches (1), whilst curve 3 shows an oxygen-binding curve for a Root effect hemoglobin at low pH.

224

T. Brittain

honor of one of the earliest investigators of this phenomenon. Although changes in proton concentration alter the oxygen-binding affinity, as apparent in changes in the value of p50 , it should be noted however that in higher vertebrates, this Bohr effect is not accompanied by any change in the level of cooperativity expressed by these proteins (h is approximately 3 over all the relevant physiological pH range). The Bohr effect can simply be considered as the release or uptake of protons from deoxygenated hemoglobin as the protein becomes oxygenated: Hb − Hn + O2 ←→ Hb − O2 + nH+ In this process, the protons released as a consequence of oxygen binding are known as the Bohr protons. For human hemoglobin, approximately two Bohr protons are released at physiological pH during oxygenation. From Fig. 3, it is clear however that for human hemoglobin two different processes occur, one at pH values less than approximately 6.0, known as the acid Bohr effect, and one above pH 6.0, known as the alkaline Bohr effect. The structural origins of the Bohr effect were first proposed by Perutz and his colleagues [20,21] on the basis of comparisons of their X-ray structures of the fully oxygenated and fully deoxygenated forms of adult human hemoglobin. Within this structural context, the Bohr effect is seen to arise from the fact that the protein structural changes associated with the oxygenation process alter the local microenvironments of a number of amino acids, with consequent alterations in the pKa values of various ionizable groups. In the alkaline region, the changes in pKa values associated with the oxygenation process are such that oxygenation of hemoglobin leads to the release of protons from the protein in solution. These structural studies have identified a number of candidate amino acids that contribute to the Bohr effect, and these have been verified by either comparison of the characteristics of native and naturally occurring point mutations or else, more recently,

3.5 3.0

log p50

2.5 2

2.0 1.5 1.0

1

0.5 6.0

6.5

7.0

7.5

8.0

8.5

pH

Fig. 3. The effect of pH on the oxygen affinity of hemoglobin. The pH dependence of the oxygen affinity for human (1) and a Root effect (2) hemoglobin is shown.

Root Effect Hemoglobins

225

by site-directed mutation of the appropriate sites in genetically engineered forms of hemoglobin [22–25,9]. Three particular amino acids have been identified as playing a significant role in the alkaline Bohr effect, namely, Val1, His122, and His146. In the deoxygenated structure, His146 interacts with the negative charge provided by Asp94. This bond is broken on oxygenation and causes the release of approximately 40% of the Bohr protons released at pH 7.4 [20,26,27]. In the deoxy state, the N-terminal group of Val1 interacts with Arg141, and the process of oxygenation breaks this interaction releasing approximately 20% of the Bohr protons from this site [28]. His122 is generally accepted as contributing approximately 20% of the Bohr protons released during oxygenation but the structural origins of this contribution are not understood [29]. The origins of the remaining 20% of the Bohr protons is still controversial, but comparative studies on mutant proteins suggest a potential role for His143 and Lys144 [30,31]. This uncertainty, at least in part, arises from the fact that in solution, there is an interplay between the various allosteric effectors, which means the exact contribution of any particular amino acid, to the Bohr effect, will also be sensitive to the presence and concentration of the other allsoteric effectors, i.e., sensitive to the exact solution conditions under which the measurements were made. According to the original formulation of the two-state model of cooperativity, the presence of allosteric effectors should only change the magnitude of the equilibrium constant (L) governing the distribution between the quaternary structural states and should not affect the intrinsic oxygen-binding affinities of the subunits within the two quaternary states, i.e., KR and KT are constant. Experimental observations clearly show that the unitary slope of the Hill plot at low oxygen concentration, which reflect the binding of oxygen to the T state, shifts its position on the oxygen concentration axis in response to changes in solution pH. That is, the apparent value of KT alters significantly as a function of pH (Fig. 3). As early as 1974, this shortcoming of the two-state model was identified by Imai, and prompted the extension of the two-state model to include, explicitly, another form of the deoxygenated protein, which was labeled the S state [32]. In this three-state model, it is assumed that the deoxygenated form of hemoglobin exists in two quaternary states, namely T and S, in which allosteric effector is respectively unbound and bound. The two-state model also ignores the recently discovered second structural R state (R2) [33,34]; the mechanistic significance of which is still a topic of debate [35,36]. Nevertheless, this model adequately describes the Bohr effect in human hemoglobin at pH values around 7.4 but did not advance any explanation for the structure of the S state. More recently, the same approach of extending the two-state model to include altered T forms has been put forward in different form, namely, explicitly within a framework in which the altered oxygen affinity within the T state arises from tertiary structural changes [37]. However, even this extension is inadequate to explain the binding of oxygen to human hemoglobin at lower pH or, particularly, in the presence of strong allosteric effectors, which have recently been shown to bind to the R as well as the T state of the protein [38–41]. In summary, at present, it appears that although the original two-state model is not able to account for heterotropic effects in detail, it nevertheless provides a useful framework, which when extended to include effector-induced changes in KR and KT [39,40] gives an adequate quantitative description of the response of hemoglobin oxygen binding to changes in solution pH.

226

T. Brittain

3. THE ROOT EFFECT HEMOGLOBINS Although the majority of the studies of hemoglobin structure and function originally centered on the hemoglobins of mammals, in the 1930s and 1940s, R.W Root and his colleagues observed that the hemoglobins of a number of fish species showed an extreme response to solution pH. At low pH, these hemoglobins not only lost their sigmoidal oxygen-binding curves, but reached less than half saturation with oxygen when equilibrated in air [42–45]. In the following years, the Root effect, as it has come to be known, has been identified in the hemoglobins of innumerable teleost fish [45–47], most often in those species displaying multiple hemoglobins in their circulations. However, not all teleost species possess a Root effect hemoglobin. It should be noted moreover that just as in the case of the Bohr effect, any particular Root effect hemoglobin does not exhibit a uniquely defined set of oxygen-binding parameters, but rather a set that falls within a range, designated as indicative of the Root effect. Additionally, it should be pointed out that although at first sight the Root effect might simply appear to be an exaggerated form of the Bohr effect, the Root effect hemoglobins display two distinctive characteristics not seen in Bohr effect proteins. Firstly, in Root effect hemoglobins, the decrease in oxygen-binding affinity seen at low pH is accompanied by a complete loss of cooperativity (in fact, many Root effect hemoglobins exhibit Hill coefficients <1 at low pH), unlike Bohr systems in which cooperativity is maintained to low pH. Secondly, Root effect hemoglobins normally do not exhibit any detectable reverse (acid) Bohr effect (increase in oxygen affinity as pH decreases) as is seen in the acid Bohr effect of other hemoglobins (Fig. 3). This absence of an acid Bohr effect must contribute to the continual drop in affinity at lower pH seen in Root effect hemoglobins, although its contribution is almost universally ignored.

4. EVOLUTION AND PHYSIOLOGICAL ROLE OF ROOT EFFECT HEMOGLOBINS The extreme pH sensitivity of Root effect hemoglobins coupled to their sporadic occurrence in numerous fish species meant that for some time, their physiological role was a point of contention [48,49]. It was initially proposed that Root effect hemoglobin was uniquely associated with the existence of a swim bladder – a structure consisting of a gas-containing sac within the fish that is variously inflated and deflated to alter the natural buoyancy of the fish to optimize energy use during swimming at various depths. It was soon recognized however that herring and some other species, which have primitive swim bladders that are filled by “gulping” air [50], lack Root effect hemoglobins. Closer inspection did however show that Root effect hemoglobins are only found in species that posses an anatomically advanced swim bladder. It is accepted, however, that Root effect hemoglobins are found universally in fish that have poorly vascularized retinas [51] that need to be oxygenated to maintain function even at low ambient oxygen concentrations. In both the case of swim bladder inflation and retinal oxygenation, Nature has provided the same organ system that drives the Root effect hemoglobin gas exchange. Both the swim bladder and the retina of fish are provided with a small acid-producing organ known as a gas gland. The gland produces lactic acid

Root Effect Hemoglobins

227

by normal metabolic means but then excretes this acid into a local counter current circulatory system known as a rete. Within the counter current loops, the received oxygenated Root effect hemoglobin experiences a local drop in pH. As shown above, under these conditions, the oxygen affinity of the Root effect hemoglobin drops precipitously and oxygen is delivered into the organ. The change in oxygen affinity is so great that oxygen can be passed into the swim bladder or retina against opposing pressures of hundreds of atmospheres [52,53]! In this way, Root effect hemoglobin can be used to inflate swim bladders at considerable depths or maintain oxygenation of poorly vascularized retinal tissues even in low ambient oxygen situations. The evolutionary connections between Root effect hemoglobin, swim bladder, and poorly vascularized retina has recently been untangled [54]. It appears that all Root effect hemoglobins have a common ancestry first evolving about 420 million years ago. Around 250 million years ago, a capillary network supplying the fish eye first evolved, which utilized the pH-dependent properties of the previously created Root effect hemoglobin. Then, around 100 million years ago, in four different fish groups, the unique blood supply to the swim bladder appeared, allowing the Root effect hemoglobin to take on the role of inflating the swim bladder in these fishes. Although the Root effect is obviously crucial to the survival of many species, its extreme pH sensitivity is also potentially harmful. Exercise stress of necessity is associated with a decrease of blood pH as anaerobic metabolism begins to acidify the blood. Exercise stress, however, does not lead to deoxygenation of circulating hemoglobins for two reasons. Firstly, Root effect hemoglobins are found in species that show multiple hemoglobins, some of which, of necessity, are not significantly pH sensitive. Secondly, universally, the red blood cells of species using Root effect hemoglobins contain a membrane-bound Na+ /H+ exchanger, which is under  adrenergic control. Thus, under stress conditions, adrenaline stimulates excretion of red blood cell protons and so, whilst the blood may become acidified, during stress the intracellular pH of the red blood cells is maintained [55].

5. MECHANISTIC ORIGINS OF THE ROOT EFFECT The operation of the Root effect in a hemoglobin molecule has most often been described within the framework of the two-state model of cooperativity, which is outlined above. The most common interpretation of the phenomenon has been to describe the Root effect as a superstabilization of the low affinity oxygen-binding T state of the protein, at low pH. However, this rather simplistic view ignores a number of important points. Firstly, the two-state model, as stated above, may adequately explain the action of hemoglobin under near-neutral pH conditions, but fails to explain hemoglobin action as a function of pH. Secondly, even if the two-state model is fitted to Root effect hemoglobin action, at low pH, the apparent oxygen affinity is often lower than the value of KT measured at higher pH, i.e., there must be at least multiple T states. Thirdly, a question until recently, rarely if ever, raised in the context of the Root effect is the nature of the R state in these systems. There is now convincing evidence that indicates that the Root effect not only involves stabilization of the deoxy quaternary T state, but also involves the destabilization of the oxy quaternary R state, together with intrastate tertiary structural changes [56]. Finally, it is found that the Root effect is typically associated with the

228

T. Brittain

release of four protons per tetramer on oxygenation, whilst the Bohr effect is associated with the release of only two protons per tetramer.

6. STRUCTURAL INTERPRETATIONS OF THE ORIGINS OF THE ROOT EFFECT Much of the early investigations of the Root effect centered on experimental studies of the hemoglobins of trout (component IV) and carp and although a great deal of functional information was obtained [57–62], in the absence of structures of these hemoglobins, little progress was made in terms of identifying the molecular origins of the Root effect. The search for the molecular origins of the Root effect has undergone a shift in emphasis over the past 20 years or so. In the 1980s, a combination of sequence comparisons, chemical modification studies [63,64], and very limited structural data was used in the first attempts to identify particular amino acids within Root effect hemoglobins, the presence of which confer this unique functional property. The studies were based on the approach previously used in identifying the amino acids responsible for the Bohr effect in which comparison of normal and naturally occurring point mutations of human hemoglobin were proven so powerful. However, in the case of the Root effect hemoglobins, few sequences were available and even less structural data. Additionally, attempts were made to identify uniquely important amino acids within sequences that otherwise showed only low levels of homology. Most notably, in this era, Perutz and Brunori [64] proposed, on the basis of sequence comparisons, that the Cys93 to Ser replacement seen in species that display the Root effect might be responsible for the extreme pH sensitivity. This argument was based on the prediction that the Cys93 residue in human hemoglobin when replaced by Ser would produce strong additional hydrogen bonds to the terminal His146 residue. These new bonds were envisaged as stabilizing the salt bridge already present between His146 and Asp94, producing a superstabilized T state and hence eliciting the Root effect. Additional support for this idea was the finding that in trout hemoglobin I, which lacks the Root effect, the terminal His is replaced by Phe [65]. Early chemical modification studies on carp hemoglobin also suggested that the terminal His147 residue was responsible for a significant fraction of the observed Root effect, as enzymic cleavage of this residue halves the pH sensitivity of the protein [63]. In 1985, one of the earliest genetically engineered forms of human hemoglobin was produced that contained this Cys93 to Ser mutation [10]. This mutation did not produce the expected hydrogen bond and failed to induce any Root effect in the otherwise human hemoglobin. Moreover, the introduction of the Ser residue was found to partially replace the normal His93–Asp94 interaction and, in fact, led to an increased oxygen affinity and reduced Bohr effect in the mutant hemoglobin – the exact opposite of what was predicted [66]. Furthermore, recent structural studies have shown that in a number of well-characterized structures of Root effect hemoglobins [67,68], in the T state, the terminal His is displaced into solution and does not take part even in those interactions that normally produce the Bohr effect, let alone the Root effect! Following these early attempts to identify a common molecular origin for the Root effect, the field languished somewhat. Over the past decade or so, it has become clear that in order to identify pertinent molecular structures, which govern the expression of the

Root Effect Hemoglobins

229

Root effect, the minimum data set will consist of the detailed atomic level structures of a Root effect hemoglobin from a single species in both the deoxygenated and oxygenated forms, each obtained at both low and high pH. In the past 10 years, the study of the origins of the Root effect has undergone something of a renaissance thanks largely to attention being focused on the Root effect hemoglobins of Antarctic fish [69]. The required data set has not yet been obtained for any single species but three of the four required structures have appeared for tuna [70]. The investigation of the origins of the Root effect in Antarctic fish began in 1992, with the determination of the detailed structure of the major hemoglobin, Hb1, of Trematomus (previously Pagothenia) bernachii in the carbon monoxide form (equivalent to the oxygenated R state – but more stable) at pH 8.0. Comparison of this protein structure with that of oxygenated human hemoglobin gave no indications of any structures that might give rise to the observed Root effect in this hemoglobin [68]. Three years later, the structure of this same hemoglobin, in the deoxygenated T state, was determined at pH 6.2 [67]. Comparison of the deoxygenated T state structure with that of the previously determined T state structure identified a significant interaction that may well contribute up to half of the observed Root effect. On oxygenation, the  dimers of hemoglobin undergo a relative rotation and approach each other. This motion is accompanied by a significant shift at the 1 2 interface. This structural change, which accompanies the T to R transition, brings Asp95 and Asp101 into very close proximity in the T state. The quaternary structural change produced on oxygenation of this hemoglobin would thus produce, in addition to the normal Bohr protons, protons associated with this interacting pair. It was noted that these residues are conserved in the Root effect hemoglobins of trout IV, carp, and tuna and also in the non–Root effect hemoglobins of Trematomus newnesii, trout I, and humans, although in these species they are spatially separated (Fig. 4). The authors of this work also note that their detailed studies could not identify any other interactions that could possibly account for the remaining half of the Root effect during the T to R state transition.

Tuna

Asp101 Asp96

Asp94

Asp99 Human

Fig. 4. The structure of the Asp–Asp pair in different hemoglobins. The local structure of the Asp–Asp pair is shown for human hemoglobin and the Root effect hemoglobin tuna.

230

T. Brittain

The field was further advanced by the analysis of the detailed structure of the carbon monoxide R state form of an unrelated Root effect hemoglobin from the teleost fish Leiostomus xanthurus, commonly known as Spot, at pH 7.5 [71,72]. The authors identified seven key amino acid residues specific to Root effect hemoglobins and four additional residues conserved in most hemoglobins, which serve to assemble positively charged clusters across the 1 2 interface in the R state of the protein. At low pH, it is speculated that the positive charge induced in these clusters would destabilize the R state and hence favor deoxygenation. Each of these clusters consists of the  N-terminal residue and its associated Lys82, within the same -chain, and the C-terminal His and Arg 144 of the other -chain. These clusters are anchored into position by segments of the F-helix and the C-terminal region. Further stabilization of the clusters is ascribed to the effects of seven additional amino acids (Ser1, Glu140, Trp3, Ser93, Glu94, Arg144, and Gln145). This work is also of significance in that it explains the earlier inability of researchers to induce the Root effect in recombinant mutant proteins when only one or two residues were introduced into a non–Root effect hemoglobin [66,73] – the Root effect arising from positively charged clusters requires the additional interaction provided by distal amino acids for its full expression. The analysis of the Spot hemoglobin structure also showed that the mere presence of specific amino acids cannot be assumed to define function. In the case of the Spot hemoglobin, the two ionizable groups at the -chain termini (which are also present in the human protein) derive their function by being forced into positions, not taken up by these residues in the human protein, as a consequence of the presence of other amino acids in the fish protein, which do not themselves directly take part in the ionization process. In more recent years, further light has been thrown into the origins of the Root effect by two very different approaches. In the first approach, researchers at the University of Naples have, in the absence of further structures of the hemoglobin from T. bernachii, compared the structure of the Root effect hemoglobin from T. bernachii with that of the very closely related non–Root effect hemoglobin from T. newnesii [74]. The hemoglobins of these two fishes differ by only four amino acids in their -chains and 10 amino acids in their -chains. A comparison of the structures of these hemoglobins in the R state at pH 8.0 identified, in both, essentially identical positive clusters to those found in the Spot structure. This led the authors to suggest that “all the differences in proton uptake, among these three hemoglobins seem to be uniquely associated with the T state,” although they do not completely reject the possibility of long-range interactions from other amino acids as being important in the expression of the Root effect. The role of positive clusters in the destabilization of the R state in Root effect hemoglobins is thus still somewhat inconclusive. In a second approach, in the absence of an experimentally determined structure for the deoxygenated form of the hemoglobin from T. newnesii, homologous theoretical structural models have been built using the T. bernachii deoxygenated structure as a template [75]. Comparison of the derived structures suggests that the Ile-Thr substitution at position 41, which is part of the 1 2 interface, may well have a significant role to play in the expression of the Root effect. However, in the light of the recognized effects of distal residues in the functioning of the Root effect hemoglobins, such model structures must be considered with caution. Thus, at the turn of the century, there was general agreement that the Asp–Asp pair, under appropriate local structural constraints, was a major contributor to the Root effect.

Root Effect Hemoglobins

231

The possible role of positive clusters in the destabilization of the R state at low pH was less clear. Nevertheless, in all cases, there was a growing awareness that the Root effect appeared to require not only residues of direct functional significance, but also amino acids that provide the correct microenvironment for the expression of these functional effects. In all cases, the interpretation of the Root effect and its origins were firmly embedded in the context of the quaternary structural changes accompanying the R to T transition on hemoglobin deoxygenation. It was not until recently that structures of identical quaternary states were obtained, at different pH values, for a single hemoglobin species. For the first time, Tame and colleagues [70] have obtained detailed structures for T state deoxygenated tuna hemoglobin at pH 7.5 and at pH 5.0, together with an R state structure obtained at pH 7.5. This unique data set has not only significantly advanced our understanding of the molecular origins of the Root effect, but has further highlighted the need to consider the Root effect within a tertiary as well as quaternary structural context. The important findings are that: a. The quaternary-linked protonation of the Asp–Asp pair is confirmed in tuna hemoglobin (present in both the pH 5.0 and pH 7.5 deoxygenated T state structures). Indeed, the two residues move close enough together that they can share a proton (Fig. 4). Comparison with other hemoglobin structures identifies the requirement for a “permissive” amino acid at position 96 of the globin chain in order for the Root effect to be expressed by this pair (Fig. 4). The tuna hemoglobin has Ser at this position and T. bernachii an Ala, both of which allow the formation of a functional Asp–Asp pair, whilst human hemoglobin has a Val residue and trout I a Gly at position 96, both of which prevent the formation of an active pair. b. Comparison of the T state structure obtained at pH 5.0 with that at pH 7.5 identifies a dramatic tertiary structural change as the pH is lowered. At pH 5.0, the distal histidine (His60) (see Fig. 1) of the -chain swings out of the heme pocket and hydrogen bonds with the heme propionate side chain! Reference to previous data, obtained from distal histidine mutations, suggests that this structural change may well be responsible for the characteristic drop in oxygen affinity of the -chain between pH 7 and pH 5. What is not clear, however, is how the tuna hemoglobin molecule prevents the rapid autoxidation seen in mutant forms of hemoglobin when the distal His is replaced by Gly or Gln [5,70]. It should be emphasized that this change is only seen in the -chain where loss of the hydrogen-bonding potential, to the bound oxygen, by the distal histidine would have a major impact on oxygen affinity. c. In the T state, a novel salt bridge is formed between His69 and Asp72, which is dependent on a ligand binding–induced tertiary structural change. In the case of the tuna protein, deoxygenation of the -heme produces a shift in Leu71 and Tyr85. The phenoxyl group of Tyr85 then moves into and disrupts the E-helix. The reverse movement associated with oxygenation breaks the Asp72–His69 interaction and releases Root effect protons. Trout I hemoglobin has three of the amino acids necessary for this structural change but does not display a Root effect. In the human protein, all of these amino acids are different.

232

T. Brittain

7. CONCLUSIONS Although significant progress has recently been made toward the identification of the molecular mechanism producing the Root effect in tuna, we do not as yet have a complete picture for even a single hemoglobin species. Nevertheless, there now appears to be a convergence of views relating to the probable origins of the Root effect in many species. In general terms, it appears that in Root effect hemoglobins, not only is the deoxy T state stabilized but the oxy R state is destabilized. Further tertiary structural changes associated with a shift in pH are also important. A number of amino acids have been identified as crucial in a number of species, including Asp–Asp pairs and positive clusters at subunit interfaces. It is also clear that the operation of the Root effect, in any particular species, requires the permissive actions of a number of distal residues that allow the important functional amino acids to take up the required positions in space. Thus, rather than the original interpretation of the Root effect as requiring the presence of one or two specific amino acids in a hemoglobin sequence, we now see the Root effect rather as arising from a synergistic interaction of many amino acids. This key finding not only solves the many riddles originating from simplistic sequence homology studies, but also provides a clear explanation of why the Root effect hemoglobins, as a set of proteins, exhibit a range of sensitivity toward pH, which we have labeled as the Root effect.

REFERENCES [1] Root, R.W. (1931) Biol. Bull. Mar. Biol. Lab. Woods Hole. 61, 427–456. [2] Dickerson, R.E. (1971) J. Mol. Evol. 1, 26–45. [3] Dickerson, R.E. and Geiss, I. (1983) Hemoglobin: Structure, Function, Evolution and Pathology. Benjamin Cummings, U.S.A. [4] Lukin, J.A., Simplaceanu, V., Zou, M. et al. (2000) Proc. Natl. Acad. Sci. U.S.A. 97, 10345–10358. [5] Colman, J.P., Brauman, J.I., Iverson, B.L. et al. (1983) J. Amer. Chem. Soc. 105, 3052–3056. [6] Olson, J.S., Mathews, A.J., Rohlfs, R.J. et al. (1988) Nature. 336, 256–266. [7] Slebodnick, C. and Ibers, J.A. (1997) J. Bio. Inorg. Chem. 2, 521–525. [8] Olson, J.S. and Phillips, G.N. (1997) J. Bio. Inorg. Chem. 2, 544–552. [9] Sigfridson, E. and Ryde, U. (1999) J. Bio. Inorg. Chem. 4, 99–110. [10] Nagai, K., Perutz, M.F. and Poyart, C. (1985) Proc. Natl. Acad. Sci., U.S.A. 82, 7252–7255. [11] Perutz, M.F. (1970) Nature. 228, 726–739. [12] Perutz, M.F. (1990) Mechanisms of Cooperativity and Allosteric Regulation in Proteins. Cambridge University Press, Cambridge, U.K. [13] Monod, J., Wyman, J. and Changeaux, J.P. (1965) J. Mol. Biol. 12, 88–118. [14] Koshland, D.E., Nemethy, G. and Filmer, D. (1966) Biochem. 5, 365–385. [15] Szabo, A. and Karplus, M. (1972) J. Mol. Biol. 72, 163–197. [16] Benesch, R. and Benesch, R.E. (1967) Biochem. Biophys. Res. Com. 26, 162–167. [17] Imaizumi, K., Imai, K. and Tyuma, I. (1979) J. Biochem. (Tokyo) 86, 1829–1840. [18] Kilmartin, J.V. and Rossi-Bernardi, L. (1973) Physiol. Rev. 53, 836–890. [19] Bohr, C., Hasselbach, K. and Krogh, A. (1904) Skand. Arch. Physiol. 16, 402–412. [20] Perutz, M.F., Muirhead, H., Mazzarella, L. et al. (1969) Nature. 222, 1240–1243. [21] Perutz, M.F., Kilmartin, J.V., Nishikura, K. et al. (1980) J. Mol. Biol. 138, 649–670. [22] Perutz, M.F., Pulsinella, P.D., Ten, L. et al. (1971) Nature. 232, 147–149. [23] Kilmartin, J.V., Hewitt, J.A. and Wooton, J.F. (1975) J. Mol. Biol. 93, 203–218.

Root Effect Hemoglobins

233

[24] Shih, D.T.B., Luisi, B.F., Miyazaki, G. et al. (1993) J. Mol. Biol. 230, 1291–1296. [25] Perutz, M.F., Fermi, G. and Shih, D.T.B. (1984) Proc. Natl. Acad. Sci, U.S.A. 81, 4781–4784. [26] Kilmartin, J.V., Breen, J.J. and Roberts, G.C.K. (1973) Proc. Natl. Acad. Sci, U.S.A. 70, 1246–1249. [27] Kilmartin, J.V. and Wooton, J.F. (1970) Nature. 228, 766–767. [28] Kilmartin, J.V. and Rossi-Bernardi, L. (1969) Nature. 222, 1243–1246. [29] Nishikura, K. (1978) Biochem. J. 173, 651–657. [30] Russu, I.M., Ho, N.T. and Ho, C. (1980) Biochemistry. 19, 1043–1052. [31] Mathews, J.B., Hanania, G.F.H. and Gurd, F.R.N. (1979) Biochemistry. 18, 1928–1936. [32] Minton, A.P. and Imai, K. (1974) Proc. Natl. Acad. Sci. U.S.A. 71, 1418–1421. [33] Smith, F.R., Lattman, E.E. and Carter, C.W. (1991) Proteins. 10, 81–91. [34] Silva, M.M., Rogers, P.H. and Arnone, A. (1992) J. Biol. Chem. 267, 17248–17256. [35] Janin, J. and Wodack, S.J. (1993) Proteins. 15, 1–4. [36] Srinivasen, R. and Rose, G.D. (1994) Proc. Natl. Acad. Sci. U.S.A. 91, 11113–11117. [37] Bettati, E.R.S., Hofrichter, J. and Eaton, W.A. (2002) Biophys. Chem. 98, 149–164. [38] Shibayama, N., Muira, S., Tame, J.R.H. et al. (2002) J. Biol. Chem. 277, 38791–38796. [39] Yonetani, T. Park, S. Tsuneshige, A. et al. (2002) J. Biol. Chem. 277, 34508–34520. [40] Tsuneshige, A., Park, S. and Yonetani, T. (2002) Biophys. Chem. 98, 49–63. [41] Imai, K., Tsuneshige, A. and Yonetani, T. (2002) Biophys. Chem. 98, 79–91. [42] Root, R.W. and Irving, L. (1943) Biol. Bull. Mar. Biol. Lab. Woods Hole. 84, 207–212. [43] Root, R.W. and Irving, L. (1941) Biol. Bull. Mar. Biol. Lab. Woods Hole. 81, 307–323. [44] Green, A.A. and Root, R.W. (1933) Biol. Bull. Mar. Biol. Lab. Woods Hole. 64, 383–404. [45] Brittain, T. (1987) Comp. Biochem. Physiol. 86, 473–481. [46] Brittain, T. (1986) Comp. Biochem. Physiol. 85, 241–243. [47] Brittain, T. (1985) Biochem. J. 228, 409–414. [48] Binotti, I., Giovenco, S., Giardina, B. et al. (1971) Arch. Biochem. Biophys. 142, 274–280. [49] Brunori, M., Falconi, G., Fortuna, G. and Giardina, B. (1975) Arch. Biochem. Biophys. 168, 512–519. [50] Ingerman, R.L. and Terwilliger, R.C. (1982) J. Exp. Biol. 220, 171–177. [51] Waser, W.P. and Heisler, N. (2004) Fish Physiol. Biochem. 30, 77–85. [52] Scholander, V.V.P. and van Dam, L. (1954) Biol. Bull. Mar. Biol. Lab. Woods Hole. 107, 247–259. [53] Pelster, B. (2001) News Physiol. Sci. 16, 287–291. [54] Berenbrink, M., Koldkjaer, P., Kepp, O. and Cossins, A.R. (2005) Science. 307, 1752–1757. [55] Pelster, B. and Decker, H. (2004) Micron. 35, 73–74. [56] Bonaventura, C., Crumbliss, A.L. and Weber, R.E. (2004) Acta. Physiol. Scand. 182, 245–258. [57] Noble, R.W., Parkhurst, L.J. and Gibson, Q.H. (1970) J. Biol. Chem. 245, 6628–6633. [58] Tan, A.L., De Young, A. and Noble, R.W. (1972) J. Biol. Chem. 247, 2493–2498. [59] Tan, A.L., Noble, R. W. and Gibson, Q.H. (1973) J. Biol. Chem. 248, 2880–2888. [60] Pennelly, R.R., Tan-Wilson, A.L. and Noble, R.W. (1975) J. Biol. Chem. 250, 7239–7244. [61] Wyman, J., Gill, S.J., Gaud, H.T. et al. (1978) J. Mol. Biol. 124, 161–175. [62] Brunori, M., Coletta, M., Giardina, B. and Wyman, J. (1978) Proc. Natl. Acad. Sci. U.S.A. 75, 4310–4312. [63] Parkhurst, L.J., Goss, D.J. and Perutz, M.F. (1983) Biochem. 22, 5401–5409. [64] Perutz, M.F. and Brunori, M. (1982) Nature. 299, 421–426. [65] Bossa, F., Barra, D., Petruzelli, R. et al. (1978) Biochim. Biophys. Acta. 536, 298–305. [66] Luisi, B.F. and Nagai, K. (1986) Nature. 320, 555–556. [67] Ito, N., Komiyama, N.H. and Fermi, G. (1995) J. Mol. Biol. 250, 648–658. [68] Camardella, L., Caruso, C., D’Avino, R. et al. (1992) J. Mol. Biol. 224, 449–460. [69] Verde, C., Parisi, E. and di Prisco, G. (2003) J. Mol. Evol. 57, 5258–5267.

234

T. Brittain

[70] Yokoyama, T., Chong, K.T., Miyazaki, G. et al. (2004) J. Biol. Chem. 279, 28632–28640. [71] Perutz, M.F. (1996) Nat. Struct. Biol. 3, 211–212. [72] Mylvaganam, S.E., Bonaventura, C., Bonaventura, J. and Getzoff, E.D. (1996) Nat. Struct. Biol. 3, 275–283. [73] Luisi, B.F., Nagai, K. and Perutz, M.F. (1987) Acta Haemat. 78, 85–89. [74] Mazzarella, L., D’Avino, R. Di Prisco, G. et al. (1999) J. Mol. Biol. 287, 897–906. [75] D’Avino, R. and DeLuca, R. (2000) Proteins 39, 155–165. [76] Hill, A.V. (1910) J. Physiol. (London), 40, 4–7.

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 10

Microbial Hemoglobins: Structure, Function, and Folding Changyuan Lua , Tsuyoshi Egawaa , Dipanwita Batabyala , Masahiro Mukaib , and Syun-Ru Yeha a

Department of Physiology and Biophysics, Albert Einstein College of Medicine, Bronx, NY 10461, b Mitsubishi Kagaku Institute of Life Sciences, Minamiooya, Machida, Tokyo 194-8511, Japan

ABBREVIATIONS Hb RR swMb hhMb HbA CcP HRP trHb FHb sdHb trHbN trHbO trHbC trHbS trHbP trCtb Cgb Vgb Hmp EPR FTIR NMR NOD

hemoglobin resonance Raman spectroscopy sperm whale myoglobin horse heart myoglobin human hemoglobin cytochrome c peroxidase horseradish peroxidase truncated hemoglobin flavohemoglobin single domain hemoglobin the truncated hemoglobin I from Mycobacterium tuberculosis the truncated hemoglobin II from Mycobacterium tuberculosis the truncated hemoglobin from green algae Chlamydomonas eugametos the truncated hemoglobin from cyanobacterium Synechocystis sp. PCC 6803 the truncated hemoglobin from protozoan Paramecium caudatum the truncated hemoglobin III from Campylobacter jejuni the single domain hemoglobin from Campylobacter jejuni the single domain hemoglobin from Vitreoscilla sp. the flavohemoglobin from E. coli Electron Paramagnetic Resonance Spectroscopy Fourier Transform Infrared Spectroscopy Nuclear Magnetic Resonance Spectroscopy NO dioxygenase

1. HEMOGLOBIN SUPER FAMILY: AN OVERVIEW Hemoglobins (Hbs) have been discovered in organisms from virtually all kingdoms [1–10]. Their presence in unicellular organisms suggests that the gene for Hb is very ancient, and it has functions other than oxygen transport and storage. For example, the

236

Changyuan Lu et al.

flavohemoglobin from Escherichia coli or yeast was found to detoxify NO through an oxygen-dependent NO dioxygenation reaction or an oxygen-independent NO reduction reaction, thereby protecting the cells against nitrosative stress [1,11–14]. On the other hand, the Hb found in the perienteric fluid of the Ascaris worm was shown to use endogenously produced NO to detoxify oxygen, which is considered poisonous to the anaerobic mitochondrial oxidation pathway in the worm [15,16]. A similar NO-dependent deoxygenase activity was proposed for leghemoglobin found in root nodules of legumes, for keeping the symbiotic nitrogen-fixing bacteria anaerobic and protecting the nitrogenfixing enzyme system from oxidation [17,18]. In addition to the deoxygenase activity, it has also been proposed that leghemoglobin facilitates oxygen diffusion to terminal oxidases of symbiotic bacteria [17–19]. A Hb-like dehaloperoxidase found in a marine worm, Amphitrite ornata, dehalogenates bromophenols to protect itself against the toxic halocompounds produced by other worms [20,21]. Recently, a new class of globincoupled sensors discovered in bacteria and archaea has a hemoglobin-like sensing domain and a catalytic domain covalently linked to it [22]. They function as regulators for aerotactic response or gene expression. Furthermore, two novel globins, cytoglobin and neuroglobin, newly discovered in vertebrates have been implicated in signal transduction and gene regulation [9]. These new findings have changed our common perception of globins as oxygen carriers and added an exciting new twist to the structure-and-function relationship of Hb. The individual subunit in every Hb discovered to date consists of a polypeptide chain with 6–8 -helical segments that fold around a heme group (Fig. 1). The helices building up the globin fold are conventionally labeled A to H according to the linear sequence order as illustrated in Fig. 1(A); in addition, the various topological positions within each helix are numbered sequentially. The quaternary assemblage of the Hb subunits into its functional structure is remarkably diverse, ranging from the monomeric Hbs of bacteria to the gigantic 144-subunit Hb of annelid [2,23]. The quaternary interactions play an important role in modulating the ligand-binding affinity of Hb. The most renowned example is the tetrameric human Hb (HbA). In HbA, ligand binding to the

(A)

(C)

(B)

D B

trHbN

A

B G

E

E

C

E11

G H

F8

B10

H

trHbN

E7 E10

F

swMb

Fig. 1. The crystal structures of (A) sperm whale oxy-myoglobin (PDB:1MBO) and (B–C) oxytrHbN from M. tuberculosis (PDB:1IDR). The nomenclatures of the eight helices in (A) are labeled A to H as indicated. The B- and E-helices are labeled in green and yellow, respectively. (see Plate 6.)

Microbial Hemoglobins

237

heme iron in one subunit induces structural changes to the other subunits through a cascade of allosteric structural transition events, forming the basis for its oxygen-binding cooperativity [24–26]. In all Hbs, the proximal heme ligand is a histidine at the F8 position. The distal ligand-binding pocket is typically constructed from the B-, E-, and part of the G-helices. The identity of the B10 residue on the B-helix and the E7, E10, and E11 residues on one unique topological surface of the E-helix (Fig. 1C) are important for the regulation of ligand binding and discrimination, as will be discussed in this review. In mammalian globins, the E7 position is almost invariably occupied by His. The E7His is believed to stabilize the heme-bound dioxygen by H-bonding to it. On the other hand, the B10, E10, and E11 positions are typically occupied by hydrophobic residues to ensure that the chemical integrity of the heme-bound dioxygen is preserved. In contrast, the distal residues in other heme proteins, such as peroxidases and oxygenases, which perform oxygen chemistry, are much more polar. The polar nature of the distal heme environment in these proteins plays an important role in activating the O O bond of heme-bound peroxide. Intriguingly, the distal residues of microbial Hbs are also more polar than those of mammalian globins, suggesting that the structures of these Hbs are tailored to perform functions other than oxygen delivery.

2. MICROBIAL HEMOGLOBINS Since the first recognition of Hb in microorganisms in 1930s by Warburg, three groups of Hbs have been characterized in unicellular organisms (Fig. 2) [5–7,27,28]. The first group, termed truncated Hb (trHb), consists of proteins with 110–140 amino acid residues and a novel 2-over-2 -helical structure (Fig. 1B) [29–37]. The second group consists of flavohemoglobins (FHb) from bacteria and fungi [11,38–43], which comprises a Hb domain with a classical 3-over-3 -helical structure and a covalently attached flavincontaining reductase domain [11,38–43]. The third group is made up of single-domain Hbs (sdHb) that exhibits high sequence homology and structural similarity to the Hb domain of FHb [4,45]. On the basis of phylogenetic analysis, the trHbs can be further divided into three subgroups: TrHb-I, TrHb-II, and TrHb-III.4 Intriguingly, the various classes of microbial Hbs may coexist in the same organism. For example, Mycobacterium tuberculosis contains a TrHb-I (trHbN) and a TrHb-II (trHbO); Campylobacter jejuni contains a TrHb-III (trCtb) and a group-three Hb (Cgb); and Mycobacterium avium contains three trHbs, one from each subgroup. These findings suggest distinct functions for each class of Hb. Microbial Microbial Hb

TrHb-I

TrHb

FHb

TrHb-II

TrHb-III

sdHb

Fig. 2. The classification of microbial Hbs as discussed in the text.

238

Changyuan Lu et al.

Table 1. Amino acid residues of trHbs, Hmp, and Vgb at various topological positions as compared to those of swMb

trHb

I

II III FHb sdHb swMb

trHbN trHbP trHbC trHbS trHbO trCtb Hmp Vgb

F8

B10

E7

E10

E11

G8

CD1

PDB

H81 H68 H68 H70 H75 H72 H85 H85 H93

Y33 Y20 Y20 Y22 Y23 Y19 Y29 Y29 Y29

L54 Q41 Q41 Q43 A44 H46 Q53 Q53 H64

K57 K44 K44 H46 R47 K49 A56 A56 T67

Q58 T45 Q45 Q47 L48 I50 L57 L57 V68

— — — — W88 W86 — — —

F46 F33 F33 F35 Y36 F32 F43 F43 F43

1IDR 1DLW 1DLY 1MWB 1NGK 2IG3 1GVH 1VHB 1MBO

Despite their structural and functional diversity, all microbial Hbs discovered to date contain a highly conserved tyrosine residue at the B10 position in the distal pocket. The distal histidine at the E7 position, which is important in stabilizing heme-bound dioxygen in mammalian globins, may be replaced by a variety of different polar or nonpolar residues (Table 1). The crystal structures of at least one member of each group/subgroup of the microbial Hbs are available in the protein data bank. Although the amino acid sequence identity between the three trHb subgroups is very low, all trHbs discovered so far exhibit a 2-over-2 -helical structure, made up by the B-, E-, G-, and H-helices (Fig. 1B). The F-helix, hosting the F8 His, is mostly replaced by an extended loop; in addition, the pre-EF loop is shorter, as compared to the classical globin structure. The exogenous ligand-binding site, like mammalian globins, is formed by the B-, E-, and part of the G-helices. The CD-D region linking the B- and E-helices is very short, forcing the E-helix to be very close to the heme distal face. The functional diversity of the trHbs is in general dictated by the H-bonding network linking the highly conserved B10Tyr and the neighboring E7, E11, CD1, and/or G8 residues (vide infra). The tertiary fold of the heme domain of FHb and sdHb is similar to the classical 3-over-3 -helical structure of mammalian globins. However, their distal heme pocket resembles that of trHbs, in which heme ligands are stabilized by a sophisticated H-bonding interaction involving the B10Tyr and/or E7Gln residues. In addition, their proximal His ligand, like that in peroxidases, exhibits an imidazolate character due to an H-bonding network involving a “catalytic trade,” composed of the His and nearby Glu and Tyr residues. The imidazolate character of the proximal His and the polar distal environment are plausibly important for the physiological functions of these Hbs (vide infra).

3. STRUCTURES AND FUNCTIONS OF MICROBIAL HEMOGLOBINS Although crystallographic studies play an indispensable role in advancing our knowledge of the structure-function relationships of microbial Hbs, it is sometimes a concern as to whether a crystal structure reflects the true solution structure, not a metastable structure

Microbial Hemoglobins

239

trapped in the crystal lattice. In addition, it remains a nontrivial task to obtain the crystal structures of a Hb of interest in all desired oxidation and coordination states, despite the fact that the crystallographic techniques have been substantially improved over the years. In this regard, it is important to compliment crystallographic studies with other spectroscopic approaches, such as resonance Raman (RR), UV-VIS, CD, FTIR, EPR, or NMR, that are applicable for samples of aqueous nature. Among these techniques, RR spectroscopy is especially versatile and informative. In the following section, we review the structural, functional, and folding properties of the following microbial Hbs, mostly based on RR spectroscopic studies. 1. The TrHb-I from: a. Mycobacterium tuberculosis (trHbN) b. Paramecium caudatum (trHbP) c. Chlamydomonas eugametos (trHbC) d. Synechocystis sp. PCC 6803 (trHbS) 2. The TrHb-II from Mycobacterium tuberculosis (trHbO) 3. The TrHb-III from Campylobacter jejuni (trCtb) 4. The FHb from Escherichia coli (Hmp) 5. The sdHb from Vitreoscilla sp. (Vgb)

3.1. The TrHb-I from Mycobacterium tuberculosis (trHbN) Mycobacterium tuberculosis is a Gram-positive obligate aerobic bacterium that grows most successfully in tissues with high oxygen content, such as the lungs. It is estimated that about one-third of the human population is latently infected by M. tuberculosis. In most healthy individuals, the initial infection by the tubercule bacilli is contained by the immune system, forcing the bacteria to enter latency for decades with possible reactivation later in life. Mycobacterium tuberculosis possesses two Hbs, trHbN and trHbO, which belong to the TrHb-I and TrHb-II family of proteins, respectively [33,34,46]. The extent of amino acid identity between these two proteins is only 18%. In aerobic cultures of M. bovis BCG, the nonpathogenic model for M. tuberculosis, a steady level of trHbO is detected throughout the growth phase, whereas trHbN is only detected after cells have reached the stationary phase (Fig. 3) [33,34]. Moreover, when expressed in E. coli, trHbO is mainly associated with membranes, while trHbN remains mostly cytoplasmic [47]. The distinctive temporal and spatial distributions of these two Hbs suggest that they perform different functions. TrHbN is a homodimer with a molecular mass of 14.4 kDa per monomer [33]. It contains 136 amino acids. The B10, E7, E10, and E11 positions in trHbN are occupied by Tyr-33, Leu-54, Lys-57, and Gln-58, respectively (Table 1) [36]. In the crystal structure of the O2 -bound trHbN, the heme-bound O2 accepts H-bonds from the B10Tyr, which in turn accepts an H-bond from the E11Gln (Fig. 4A). The ligand-free ferrous derivative of trHbN exhibits a five-coordinate high-spin configuration as judged by the 3 and 4 at 1471 cm−1 and 1356 cm−1 , respectively (Fig. 5) [46]. The Fe His mode is at 226 cm−1 . It is significantly higher than that of HbA (214 cm−1 ) or sperm whale myoglobin (swMb, 220 cm−1 ), consistent with a staggered orientation of the imidazole ring of the proximal histidine with respect to the pyrrole

240

Changyuan Lu et al. 10

A580

1

0.1 trHbO

0.01 trHbN

0.001 0

50

100

150

200

Time (hr)

Fig. 3. The growth curve of M. bovis BCG cells and the associated temporal expression pattern of trHbN and trHbO. The data are adapted from Mukai et al. [34].

(A)

(B)

(C)

(D)

B10 E11

E11 B10

E11 E10

B10

E10

B10 E10

E14

trHbN-O2 B10 3.13 E11 O-O

3.86

dE7-E11 = 5.17 Å

E10

E14

E14

3.14

E11 E7 E14

trHbP-H2O 2.69

E7 2.73

B10 3.28 2.26 E11

H2O 4.33

dE7-E11 = 3.99 Å

trHbS

trHbC-CN– 3.29

E7 2.92

B10

2.95

2.36 E11

2.82

E7

B10

8.31

E11

3.69

CN H2O dE7-E11 = 4.83 Å

dE7-E11 = 7.84 Å

Fig. 4. The distal heme pockets of (A) the oxy derivative of trHbN (PDB:1IDR), (B) the waterbound ferric derivative of trHbP (PDB:1DLW), (C) the ferric cyanide–bound derivative of trHbC (PDB:1DLY), and (D) the exogenous ligand-free ferric derivatives of trHbS (PDB:1MWB). The B-and E-helices are shown in a ribbon representation. The Phe at the E14 position, which is believed to play an important role in shielding the ligand-binding site from the solvent, is labeled as a reference. The side chain groups of the distal E10, E14, E7, and E11 residues appear in a clockwise order along the ring representing the E-helix. The exogenous ligands are labeled in a ball-and-stick presentation. The heme group is displayed as thin sticks. The bottom cartoons show the distances between the various distal residues and heme ligands as indicated.

nitrogen atoms, in contrast to the eclipsed orientation in HbA or swMb [48]. In the CO derivative, two Fe CO modes were identified at 500 and 535 cm−1 on the basis of the 12 C16 O −13 C18 O isotope difference spectrum shown in Fig. 6(A) [46]. Two C O modes associated with the two Fe CO modes were found at 1960 and 1916 cm−1 ,

Microbial Hemoglobins

241 ν4

trHbN

1356

ν3 ν2 1471

1622 1564

Deoxy pH 7.5

1561 1579 1501 1479

Ferric pH 7.5

pH 10.5

900

x

1100

1300

1500

1700

Raman Shift (cm–1)

Fig. 5. The high-frequency resonance Raman spectra of the ferrous deoxy derivative (pH 7.5) and ferric derivative (pH 7.5 and 10.5) of wild-type trHbN as indicated. The peak indicated as “x” is a laser plasma line. The data are adapted from Yeh et al. [46]. (A) νFe — CO 500

trHbN-CO

534

(B) wt

560

Fe––C

O+ ↔ Fe

C O

550

νFe — CO (cm–1)

B10 Y→L wt

489

12 16

C O

—13 18

C O

L518 B10 Y→L

540 530 520 510 500

12 16

C O —13C18O

200

500

800

Raman Shift (cm–1)

VgbC HmpC

trHbNC trHbO

trCtb trHbNO

swMb HmpO

490 1100

VgbO

1890 1905 1920 1935 1950 1965

νC — O (cm–1)

Fig. 6. The low-frequency resonance Raman spectra of the CO derivative of the wild type and the B10Tyr → Leu mutant of trHbN at pH 7.5 and the corresponding 12 C16 O–13 C18 O difference spectra (A), and the Fe CO versus C O inverse correlation line for the various microbial Hbs and sperm whale myoglobin (B). The subscripts “O” and “C” stand for open and closed conformations, respectively. The data in (A) are adapted from Yeh et al. [46].

respectively [46]. The 535/1916 cm−1 modes are assigned to a “closed” conformation, in which the heme-bound CO is stabilized by distal H-bonding interactions; whereas the 500/1960 cm−1 modes are assigned to an “open” conformation, in which the H-bonding interactions are absent.

242

Changyuan Lu et al.

In general, when CO is coordinated to the ferrous heme iron, the C O frequency shifts from 2143 to ∼1900–1970 cm−1 . The -bond formed by the donation of two electrons from CO to the heme iron greatly increases the electron density on the heme iron; to stabilize the Fe C O moiety, the excess electron density on the heme iron is donated from its d orbital back to the  ∗ orbital of the CO. As a result of this so-called “-backbonding” effect, the Fe C O moiety can be presented by the following two extreme structures. Fe−

C

I

O+ ←→ Fe

C

II

O

(1)

The degree of -backbonding in heme proteins is modulated by the protein matrix surrounding the CO moiety. As a general rule, a positive polar environment destabilizes form (I) and facilitates the -backbonding interaction. Consequently, the bond order of the Fe CO is increased, which is concurrent with a decrease in the bond order of the C O. On this basis, a well-known inverse correlation between the Fe CO and C O frequencies is established as illustrated in Fig. 6(B) [49–52]. The data points associated with the closed and open conformations of trHbN are located at the upper left corner and lower right corner of the C O −Fe CO inverse correlation line, respectively. Mutation of the B10Tyr to Phe converts trHbN to a single open structure, as indicated in the inverse correlation curve shown in Fig. 7(A), suggesting that in the wild-type protein, the B10Tyr forms an H-bond with the CO. Intriguingly, the mutation of the E11Gln to Leu also converts trHbN to a single open conformation, indicting that without the E11Gln, the B10Tyr by itself cannot form an H-bond with the heme-bound CO. The data hence suggest that the B10Tyr in wild-type trHbN is positioned by E11Gln in a proper stereo-orientation to stabilize the heme-bound ligand via an H-bond.

(A)

(B) 560

520

WT B10 CD1

500

trHbN

WT C

540

G8

swMb B10

WTO E11 CD1/G8 G8

trHbO

480

ν(Fe–CO) (cm–1)

ν(Fe–CO) (cm–1)

560

E7C

trCtb

540 520

B10 B10/E7

E7O WT

500 480

1900

1920

1940

νC–O) (cm–1)

1960

1980

1900

1920

1940

1960

1980

νC–O) (cm–1)

Fig. 7. The Fe CO versus C O inverse correlation line for the wild type and mutants of trHbN (squares)/trHbO (circles) (A) and trCtb (squares) (B). The shaded square in (A) is a reference data point from sperm whale myoglobin. The corresponding solid and open symbols are from the wild-type proteins and mutants, respectively. The arrows indicate the direction along which the data shift in response to mutations in the distal residues from polar to apolar substituents. The B10, E7, E11, CD1, and G8 stand for the B10Tyr → Phe, E7His → Leu, E11Gln → Leu, CD1Tyr → Phe, and G8Trp → Phe mutants, respectively.

Microbial Hemoglobins

243 νFe – O2

trHbN-O2

560

wt 570

B10 Y→L 542

wt 16

545

B10 Y→L 16

200

500

18

O2– O2

18

O2– O2

800

1100

Raman Shift (cm–1)

Fig. 8. The low-frequency resonance Raman spectra of the O2 derivative of the wild type and the B10Tyr → Leu mutant of trHbN at pH 7.5 and the corresponding 16 O2 –18 O2 difference spectra. The data are adapted from Yeh et al. [33,46].

In the oxy derivative, the Fe−O2 mode was found at 560 cm−1 based on the 16 O2 −18 O2 isotope difference spectrum shown in Fig. 8 [33]. The O O mode, although rarely seen in RR spectra of heme proteins with a proximal histidine ligand, was identified at 1139 cm−1 , indicating superoxide character of the heme-bound dioxygen (unpublished data). When the B10Tyr is mutated to Leu, the Fe O2 frequency shifts to 570 cm−1 , a frequency similar to that of swMb (569 cm−1 ) [46]. In swMb, the heme-bound dioxygen is stabilized by the E7His, via an H-bond with its terminal oxygen atom; in addition, its O O bond is highly polarized by the heme iron. Accordingly, the Fe O2 frequency is relatively insensitive to the E7His mutations [53]. The 10 cm−1 shift in the Fe O2 frequency in the B10Tyr → Leu mutant of trHbN suggests that the B10Tyr interacts with the proximal oxygen atom (instead of the terminal oxygen atom) of the hemebound O2 in the wild-type trHbN. This proposal is supported by the observation that the mutation of the B10Tyr to Phe increases the O2 off-rate by a factor of ∼200 [46]. The mutation of the E11Gln to Leu, on the other hand, increases the off-rate by a factor of only ∼20 (unpublished data), consistent with the scenario that the E11Gln does not directly interact with the heme-bound ligand; instead, it positions the B10Tyr in a more favorable stereo-orientation to form H-bond(s) with the ligand, as that proposed for the CO derivative. This unique H-bonding network in trHbN leads an unusually high O2 affinity, as shown in Table 3. At neutral pH, the ferric derivative of trHbN exhibits a mixture of six-coordinate high- and low-spin configurations as indicated by the 2 modes at 1561 and 1579 cm−1 , respectively (Fig. 5). The data suggest a typical aquo-met species with a water ligand bound to the distal site of the heme [46]. When the pH is raised to pH 10.5, the contribution from the six-coordinate low-spin increases, due to the deprotonation of the heme-bound water. On the basis of the H2 16 O − D2 18 O isotope substitution experiments,

244

Changyuan Lu et al. νFe – NO

TrHbN-NO δFe – NO

594

578

568 588

591 579

wt

B10 Y→F

595 587

250

350

450

550

650

Raman shift (cm–1)

Fig. 9. The low-frequency resonance Raman spectra of the NO-bound ferric derivative of the wild type and the B10 Tyr → Phe mutant of trHbN at pH 7.5, and the corresponding 14 N16 O–15 N16 O difference spectra shown at the top and the bottom of the traces. The data are adapted from Mukai et al. [54].

two Fe OH modes were identified at 454 and 561 cm−1 [46]. When the B10Tyr is mutated to Phe, the line at 454 cm−1 totally disappeared, leaving only a high-frequency line at 552 cm−1 . Accordingly, the 454 cm−1 mode was assigned to a closed conformation in which the hydroxide is stabilized by an H-bond donated from the B10Tyr, whereas the 561 cm−1 mode was identified as an open conformation in which the H-bond is absent [46]. The presence of two alternative conformations, like that in the CO derivative, reflects the plasticity of the distal heme pocket of trHbN. In the NO-bound ferric derivative of trHbN, the Fe NO and Fe N O modes were identified at 591 and 579 cm−1 , respectively (Fig. 9) [54]. When the B10Tyr is mutated to Phe, the intensity of the Fe NO mode is slightly increased and that of the bending mode is diminished. In addition, the frequency of the N O stretching mode (N O ) shifts from 1914 to 1908 cm−1 [54]. These data are consistent with the presence of H-bond(s) between the B10Tyr and the heme-bound NO in the wild-type protein [54]. UV Raman data show that upon NO binding to the ferric protein, the frequency of the Y8a mode of the B10Tyr shifts from 1616 to 1622 cm−1 , confirming a direct interaction between the B10Tyr and the heme-bound NO [54]. Intriguingly, the Y8a mode of the other two Tyr residues at positions 16 and 72, which are remote from the heme active site, are also affected by NO binding, suggesting that NO binding to the heme iron triggers a largescale conformational change that propagates throughout the B- and E-helical region to the pre-F-helix loop region [54]. This large-scale conformational change triggered by NO binding in trHbN is consistent with that associated with the allosteric structural transition underlying its O2 -binding cooperativity [55]. Since the environment that the bacteria experience in the stationary phase in vitro is very similar to that in the latent phase found in vivo and the expression of trHbN is greatly enhanced during the stationary phase in aerobic cultures, we proposed that the

Microbial Hemoglobins

245

physiological role of trHbN is to protect the bacilli against NO produced by the host macrophage during latency [33], via the NO dioxygenase (NOD) reaction [56]. Fe2+

O

O+ N •

O −→ Fe3+

OONO− −→ Fe3+ + NO− 3

(2)

The NOD activity of trHbN is confirmed by in vitro experiments, which show that titration of the oxygenated trHbN with NO results in stoichiometric oxidation of the protein along with nitrate formation; in addition, the second-order rate constant between the oxy derivative and NO is 745 M−1 s−1 , 20-fold faster than swMb [57]. These in vitro data are supported by in vivo results, which show that disruption of the glbN gene encoding trHbN in M. bovis BCG causes a dramatic reduction in the NO-consuming activity of the stationary phase cells, and this activity could be fully restored by complementing the knockout cells with wild-type glbN, but not the B10 mutant of glbN [57]. Taken together, the data suggest that the NOD reaction of trHbN is facilitated by two factors: (i) the H-bonding network involving the heme-bound O2 and the B10Tyr-E11Gln, by promoting the O O bond cleavage of the heme-bound peroxynitrite intermediate (Fe3+ OONO− ) and (ii) the plasticity of the distal heme pocket, by facilitating the release of the product nitrate.

3.2. The TrHb-I from Paramecium caudatum (trHbP) The monomeric Hb from unicellular protozoan Paramecium caudatum contains only 116 amino acid residues [29,58]. The B10, E7, and E10 positions in trHbP are occupied by Tyr-20, Gln-41, and Lys-44, respectively (Table 1). The E11 position is occupied by Thr-45, instead of a typical Gln found in other members of the TrHb-I family of proteins. The side chain group of Thr is significantly shorter than that of Gln, which limits its role as an H-bond donor or acceptor for heme-bound ligands. In the crystal structure of the ferric derivative of trHbP (Fig. 4B), the OH group of the E11Thr is within an H-bonding distance from the B10Tyr and the heme-bound water is stabilized by an H-bonding network involving the B10Tyr and the E7Gln. RR studies show that the ferric derivative of trHbP exhibits a mixture of six-coordinate high- and low-spin configurations, which is consistent with an aquo-met heme [58]. The ferrous derivative displays a five-coordinate high-spin configuration with a single histidine as the axial ligand. The Fe His mode is at 220 cm−1 [58], which is similar to that of swMb (Table 2), indicating an eclipsed orientation of the imidazole ring of the proximal histidine with respect to the pyrrole nitrogen atoms of the porphyrin ring. In contrast to the two conformations in trHbN, the CO derivative of trHbP exhibits only an open conformation as indicated by Fe CO and C O at 493 and 1974 cm−1 , respectively [58]. Consistent with the open conformation, the nanosecond laser flash photolysis studies show a low CO geminate recombination yield [48]. In contrast to the CO derivative, the Fe−O2 of the oxy derivative is at 563 cm−1 , which is similar to that of trHbN [58], suggesting that the proximal oxygen atom of the heme-bound O2 accepts H-bond(s) from distal residue(s) in its vicinity, possibly the B10Tyr and E7Gln residues, as suggested by the crystallographic data shown in Fig. 4(B). Intriguingly, despite of the presence of the H-bonding interaction, the oxygen off-rate of trHbP is relatively fast (25 s−1 as shown in Table 3) [58], indicating that the

246

Table 2. The coordination and spin states of the various derivatives of the differing Hbs and their ligand-associated vibrational modes Ferrica trHb

I

II III sdHb FHb swMbc HRPc

c

trHbN trHbPe trHbCf trHbSg trHbOc trCtbh Vgbh Hmpc

6CHS + LS 6CHS + LS 6CLS 6CLS 6CHS + LS 6CHS + LS 5CHS 5CHS 6CHS + LS 5CHS

Ligandsa H H H H H H H H H H

H2 O H2 O YB10 HE10 H2 O H2 O

H2 O

Deoxya 5CHS 5CHS 5CHS 6CLS 5CHS 5CHS 5CHS 5CHS 5CHS 5CHS

Fe

His

226 220 232 nd 226 226 252 244 220 ∼240

Fe−O2 562 563 554 554 559 542 557 — 569 562

O

O d

1139 nd 1136 1133 1140 1133 nd — nd nd

b

Fe 500, 493 491 492 525 515 489, 494, 512 531,

CO

535

536 535 541

a

Changyuan Lu et al.

The “6C,” “5C,” “HS,” and “LS” stand for six-coordinate, five-coordinate, high-spin, and low-spin heme iron, respectively; H and Y stand for histidine and tyrosine, respectively. b “nd” stands for not detected. c [3] d To be published. e [58] f [62] g [63] h [73,78]

Microbial Hemoglobins

247

Table 3. The O2 and CO on- and off-rate constants of the various Hbs, as well as the associated equilibrium constants (KO2 and KCO ) calculated from the rate constants kon (O2 ) (M−1 s−1 )

koff (O2 ) (s−1 )

KO2 (M−1 )

kon (CO) (M−1 s−1 )

koff (CO) (s−1 )

KCO (M−1 )

KO2 / KCO

trHbN trHbP trHbC trHbS trHbO

71a 30c >10d — 0.13a

sdHb

0.91h 200i

0.005b 0.328c 0.0022d — 0.0015g 0.0041g 0.0048a —

1100 84 — — 6.7 2.5 13.3 —

0.4 0.01 0.20e — 14 8.8 17 —

FHb

Hmp

38j

444 1.2 — — 93 22 222 48i 1333i 86

5.5a 27.7c >10d — 0.01a

trCtb Vgb

0.16a 25.2c 0.014d 0.011f 0.0014g 0.0058g 0.0041h 4.2i 0.15i 0.44j

14k

12k

1.2

0.057j 0.018j 0.019k

386 77.8 26.8

0.22 1.1 0.04

trHb

I

II III

swMb

0.064a 31i 0.66i 22j 1.4j 0.51k

a

To be published. [33] c [58] d [62] e This number was measured on the basis of direct competition experiments [62]. f [68] g [69] h [78] i [114] j [88] k [115,116] b

H-bonding interaction may be further regulated by an additional H-bond between the E11Thr and the B10Tyr to avoid the overstabilization of the heme ligand. It is important to note that a similar B10Tyr-E7Gln pair has been found in an Hb (AscHb) from the nematode parasite Ascaris suum. In AscHb, the B10Tyr-E7Gln pair forms an interlaced H-bonding network with the heme-bound O2 , which leads to an extremely low oxygen dissociation rate (0.004 s−1 ) and high oxygen affinity [59]. In contrast, the B10TyrE7Gln pair found in CerHb, a Hb from Cerebratulus lacteus nerve tissue, does not seem to play the same stabilizing role, due to an additional H-bond between the E11Thr and the B10Tyr, which forces the phenolic oxygen of the latter to face the heme-bound O2 , thereby creating a destabilizing negative electrostatic force around the heme-bound O2 [60]. As a result, the oxygen dissociation rate of CerHb is more than four orders of magnitude faster (180 s−1 ) with respect to AscHb [60]. Similarly, in leghemoglobin from the roots of leguminous plants, the H-bond between the E7His and B10Tyr has been proposed to prevent the overstabilization of heme-bound oxygen [61]. Along these lines, we hypothesize that the E11Thr in trHbP regulates its O2 -binding properties by forming an H-bond with the B10Tyr. It is noteworthy in Table 3 that all the trHbs show unusually slow O2 off-rate and high O2 affinity, except trHbP. Furthermore, trHbP is the only trHb that does not exhibit a O O mode in its RR spectrum (Table 2). These unique properties of trHbP may be

248

Changyuan Lu et al.

attributed to the regulatory effect of the E11Thr residue. In view of the fact that almost all the trHbs discovered to date exhibit at least three polar residues at critical topological positions in the distal heme pocket (Table 1), this type of regulatory mechanism may be more general than has been recognized in the past.

3.3. The TrHb-I from Chlamydomonas eugametos (trHbC) The first chloroplast Hb was discovered in the unicellular green algae Chlamydomonas eugametos [32]. The expression of this monomeric Hb is induced by light during active photosynthesis [62]. The intracellular localization of trHbC is predominantly in the pyrenoid and the thylakoid regions of the chloroplast. TrHbC exhibits high sequence homology (43%) with trHbP [63]. The B10, E7, and E10 positions of trHbC are occupied by Tyr-20, Gln-41, and Lys-44, respectively (Table 1). The E11 position in trHbC is occupied by Gln-45, instead of a Thr as that found in trHbP. The topological positions of these residues are shown in Fig. 4(C) in the cyanide-bound ferric derivative [35]. RR studies show that the heme iron in the exogenous ligand-free ferric derivative is six-coordinate low-spin with an intrinsic amino acid coordinated to it, as indicated by the 3 and 4 modes at 1502 and 1374 cm−1 , respectively [64]. The distal ligand was identified as the B10Tyr, because the EPR signal associated with the wild-type protein is characteristic for a tyrosinate-bound heme, and the EPR signal disappears when the B10Tyr is mutated to Leu [64]. In the B10Tyr → Leu mutant, another six-coordinate low-spin heme associated with a new EPR signal, signifying a His Lys-bound heme, was observed. When the E10Lys is mutated to Ala, the protein converts to an aquo-met form [64], suggesting that the E10Lys is the ligand coordinate to the heme iron in the B10Tyr → Leu mutant. More importantly, the data suggest that without the assistance from the E10Lys, the B10Tyr by itself cannot coordinate to the heme iron. Despite that the spectroscopic studies suggest that both the B10 Tyr and E10Lys are nearby the heme iron, in the crystal structure of the cyanide-bound ferric derivative, the E10Lys protrudes out to the solvent and forms an H-bond with a propionate group of the heme; in addition, instead of the E10Lys, the E7Gln and E11Gln locate within H-bonding distance from the B10Tyr (Fig. 4C). The data suggest that cyanide binding to the heme iron must induce significant conformational change to trHbC. In response to reduction, the B10Tyr in the wild-type protein dissociates from the distal heme-binding site, as indicated by the 3 and 4 modes at 1468 and 1355 cm−1 , respectively [62], which are typical for a five-coordinate high-spin heme. The Fe His frequency of the deoxy derivative is 232 cm−1 [62], which is 12 cm−1 higher than that of swMb, consistent with a staggered orientation of the imidazole ring of the proximal histidine with respect to the four pyrrole nitrogen atoms of the porphyrin ring. The Fe−O2 frequency of the oxy derivative is 554 cm−1 [65], which is similar to that of trHbN and trHbP; in addition, the typically Raman inactive O O stretching mode (O O ) was observed at 1136 cm−1 [65]. The oxygen off-rate of trHbC is very slow (∼1000-fold slower than that of swMb), as shown in Table 3. It increases by a factor of 70 and 30, when the B10Tyr and E7Gln are mutated to Leu and Gly, respectively [62], suggesting that both the B10Tyr and the E7Gln form H-bonds with the heme-bound dioxygen. The interaction between the heme-bound dioxygen and the E7Gln is further supported by the

Microbial Hemoglobins

249

observation that the mutation of the E7Gln to Gly causes the shift of the Fe−O2 mode to 569 cm−1 and the diminishment of the O O mode [65]. The presence of the H-bonding interaction stabilizing the heme-bound ligand in trHbC is consistent with its high geminate recombination yield [48]. It is important to note that in most Hbs studied to date, CO is more likely to escape out of the protein into the solvent than rebind to the heme iron in a geminate manner. Despite that these data suggest the presence of H-bonding interaction between the CO and the surrounding protein matrix, only a single open conformation associated with Fe CO at 491 cm−1 was observed under equilibrium conditions in trHbC [62]. The E11Gln in trHbC, like the E11Thr residue in trHbP, hence may regulate the H-bonding interaction between the B10Tyr-E7Gln pair and the heme-bound ligand by forming an H-bond with the B10Tyr.

3.4. The TrHb-I from Synechocystis sp. PCC 6803 (trHbS) Synechocystis is a unicellular non-nitrogen-fixing cyanobacterium, which is capable of growing heterotrophically at the expense of glucose. The trHbS from Synechocystis sp. PCC 6803 is monomeric with 123 amino acid residues [66]. The B10, E7, E10, and E11 positions of trHbS are occupied by Tyr-22, Gln-43, His-46, and Gln-47, respectively (Table 1). All these residues are identical to those of trHbC except that the E10 residue is a His instead of a Lys. This single amino acid replacement introduces an interesting twist to the structural properties of trHbS (vide infra). RR studies show that the ferric derivative of trHbS is six-coordinate low-spin, based on the 3 and 4 modes at 1496 and 1372 cm−1 , respectively [63]. Upon reduction, the protein remains as six-coordinate low-spin as indicated by 3 and 4 modes at 1489 and 1359 cm−1 , respectively [63]. When the E10His is mutated to Ala, the ferrous protein converts to a five-coordinate high-spin configuration, whereas the ferric protein becomes a mixture six-coordinate high- and low-spin, characteristic for a water bound heme. The data suggest that in the wild-type protein, the E10His coordinates to the heme iron to form the six-coordinate low-spin species [63]. The coordination of the E10His to the heme iron in the ferric derivative is confirmed by the NMR solution structure, as shown in Fig. 4(D) [67]. The ability of the E10His to coordinate to the heme iron in trHbS, reminiscent the B10Tyr in trHbC, suggesting that the B/E helix bundle in these two Hbs is flexible enough to allow the two intrinsic amino acids to coordinate to the heme iron. The movement of the B-/E-helix bundle may be facilitated by the flexible hinge region linking the B-/E-helix bundle and the F-helix, which contains a GGP sequence motif conserved in trHbS and trHbC [4]. This proposal is consistent with the fact that this hinge region is the most dynamic structural element observed in the solution NMR structure of trHbS [67]. The CO and O2 -bound derivatives of trHbS exhibit structural properties analogous to those of trHbC. In the CO derivative, only an open conformation was observed, which is associated with Fe CO and C O at 492 and 1955 cm−1 , respectively [63]; in addition, a high geminate CO rebinding yield was observed following laser flash photolysis [68]. The Fe O2 and O O modes of the oxy derivative are at 554 and 1133 cm−1 , respectively [65]. The frequencies of these two modes are almost identical to those of trHbC (Table 2), suggesting that the heme-bound dioxygen is stabilized by the B10Tyr and the E7Gln via H-bonding interactions as that suggested for trHbC. Likewise,

250

Changyuan Lu et al.

the oxygen off-rate of trHbS, 0.011 s−1 [63], is similar to that of trHbC (Table 3). When the B10Tyr or the E7Gln is mutated to a nonpolar residue, the O2 off-rate increases by a factor of ∼100, confirming the important roles of these residues in stabilizing the heme-bound dioxygen [68]. In the solution NMR structure of the ferric derivative (Fig. 4D) [67], the coordination of the E10His to the heme iron forces the side chain groups of the B10Tyr, the E7Gln, and the E11Gln out of the vicinity of the heme iron. In this structure, the E7Gln is within H-bonding distance from the B10Tyr, whereas the E11 side chain group is too far to form an H-bond with the B10Tyr. To satisfy the proposed H-bonding interaction between the heme-bound dioxygen and the B10Tyr-E7Gln pair, the E10His has to dissociate from the heme iron, in addition, the B-/E-helix bundle has to move toward the heme plane. Although there is a great deal of similarity between trHbS and trHbC, several structural features of trHbS set it apart from trHbC: (i) the intrinsic sixth heme ligand in trHbS is the E10His, in contrast to the B10Tyr in trHbC, (ii) upon reduction, the E10His in trHbS stays coordinated to the heme iron; whereas the B10Tyr in trHbC dissociates from the heme iron (possibly due to the fact that His is a better ligand for ferrous heme iron), and (iii) the F-helix in trHbS contains more than three -helical turns, in contrast to a single turn in trHbC. It remains to be determined as to how these structural features influence the functional properties of trHbS and trHbC. Nevertheless, it is important to note that the F-helix is one of the most dynamic regions in the NMR structure of trHbS [67], and it is one of the most diversified structural elements among the various trHbs; its flexibility, along with the polar nature of the distal heme pocket, may play important roles in controlling the functional properties of the trHbs.

3.5. The TrHb-II from Mycobacterium tuberculosis (trHbO) The TrHb-II from M. tuberculosis, trHbO, has 128 amino acid residues and a molecular mass of 14.9 kDa [34]. Under solution conditions, trHbO exists as a mixture of monomer and dimer [69]. The monomer and dimer equilibrium is sensitive to the ionic strength of the solution. In the presence of high concentration of salt, the protein is mostly in a monomeric state, suggesting that the dimer interface is stabilized by salt bridges. Under the solution conditions used in the RR studies summarized here, trHbO exists primarily as a dimer [69]. In contrast, in the crystalline state, trHbO displays as a dodecamer with six pairs of asymmetric dimeric units [37]. The B10 and E10 positions in trHbO are occupied by Tyr-23 and Arg-47, respectively. The E7 and E11 positions are both occupied by apolar amino acid residues, Ala-44 and Leu-48, respectively, in sharp contrast to the TrHb-I family of proteins, precluding ligand stabilization by theses two residues. The most intriguing feature of trHbO is that the CD1 residue is a Tyr, instead of Phe, which is highly conserved in most other Hbs discovered to date. Furthermore, a covalent bond between the phenyl oxygen of the B10Tyr and the C 2 of the CD1Tyr was observed in six subunits of the dodecamer (Fig. 10A), but not in the other six subunits, although in the latter, the aromatic side chain groups of the B10Tyr and the CD1Tyr are in very close contact and in a similar orthogonal orientation [37]. A Trp residue at the G8 position, which is highly conserved in the TrHb-II and TrHb-III family of proteins, also plays a critical role in ligand stabilization. In the crystal structure of the ferric cyanide–bound derivative of trHbO, the heme-bound

Microbial Hemoglobins

251 (A)

(B) B10 B10

B9 CD1

CD1

G8

G8

trHbO-CN (C)

(D) B10 E7

B10

E7

B9 G8

CD1

G8

CD1

trCtb-CN

Fig. 10. The distal heme pockets of the cyanide-bound ferric derivatives of trHbO (A–B) and trCtb (C–D) in differing views.

cyanide accepts H-bonds from the CD1Tyr as well as the G8Trp, but not the B10Tyr (Fig. 10A). RR studies show that the ferric derivative of trHbO is in an aquo-met form, with the heme in a mixture of six-coordinate high- and low-spin configurations [34]. The ferrous derivative is five-coordinate high-spin, with the Fe His mode at 226 cm−1 , indicating a staggered geometry of the proximal histidine with respect to the pyrrole nitrogen atoms of the porphyrin ring [70]. The CO derivative of trHbO exhibits only one Fe CO mode at 525 cm−1 (Table 2), which is assigned to a closed conformation (Fig. 6B) [34]. It represents the only trHb, which exhibits a single conformation locked in the closed state. The mutation of the B10Tyr to Phe does not affect the position of the data point (Fig. 7A), confirming that the B10Tyr does not play any significant role in ligand stabilization. On the other hand, the mutation of CD1Tyr or G8Trp to Phe causes the data point to shift down to the middle of the correlation line, whereas the CD1/G8 double mutation causes it to further shift to the lower right corner of the correlation line (Fig. 7A), indicating that in the wild-type protein, the heme ligand is stabilized by the CD1Tyr and G8Trp in a synergetic fashion (unpublished results). It is important to note that the G8Trp → Phe mutant exhibits an additional open conformation (Fig. 7A), suggesting that the H-bond between the G8Trp and heme-bound ligand helps to position the ligand to accept an additional H-bond from the CD1Tyr. The Fe O2 mode of the oxy derivative of trHbO was identified at 559 cm−1 and the typically RR-silent O O mode was detected at 1140 cm−1 [34]. The mutation of the B10Tyr to Phe does not affect these two modes [34,69]. In contrast, the mutation of the CD1Tyr to Phe causes a slight downshift in the Fe O2 frequency to 556 cm−1 , which is accompanied by the disappearance of the O O mode [34]. The data is consistent with the scenario that the heme-bound ligand in trHbO is stabilized by the H-bonds donated by both the CD1Tyr and the G8Trp as that suggested for the CO derivatives. This unique distal interaction results in an extremely slow O2 off-rate (∼0.0014 s−1 ),

252

Changyuan Lu et al.

which is ∼10,000-fold slower than that of swMb, making it one of the slowest oxygenreleasing Hbs reported to date (Table 3). The on-rates for O2 and CO are also very slow [69], ∼100- and 50-fold slower than those of swMb, respectively (Table 3). When the CD1Tyr is mutated to Phe, the on-rates for O2 and CO increase by a factor of 25 and 77, respectively [69], whereas the mutation of the G8Trp to Phe increases the on-rates by a factor of 160 and 530 (unpublished data), respectively [69], demonstrating the importance of the G8Trp and CD1Tyr residues in controlling the ligand entry into the heme distal site. TrHbO and trCtb (vide infra) represent the only two Hbs discovered to date that preferentially bind O2 over CO. The intrinsic affinity of the heme prosthetic group for O2 in free solution is roughly 20,000-fold weaker than that for CO [71]. In swMb, this ratio is reduced to 25 because the electrostatic environment created by the distal E7His encourages the binding of O2 with respect to CO [72]. In trHbO, the affinity for O2 is 14 times stronger than that of CO, presumably due to its novel distal environment congested with the aromatic side chain groups of the CD1Tyr, G8Trp, B10Tyr, and the B9 Phe (Fig. 10B), which may disfavor the linear Fe C O moiety. In contrast, the heme-bound O2 is stabilized by two H-bonds provided by the CD1Tyr and the G8Trp in an optimized geometry. The structurally confined distal heme pocket of trHbO may also account for its high geminate CO recombination yield (∼86%) observed in the nanosecond flash photolysis experiments (unpublished data). The physiological function of trHbO remains to be investigated. In vivo studies show that overexpression of trHbO in E. coli recombinant cells stimulates cellular respiration and oxygen uptake in the wild-type cells, but not in terminal oxidase-deficient mutant cells, suggesting a direct interaction between trHbO and terminal oxidases [47]. TrHbO, hence, may function as an oxygen sequester in M. tuberculosis to sustain aerobic metabolism. Although trHbO has a relatively low oxygen affinity and slow O2 onand off-rates (Table 3), trHbO does have the advantage for O2 -sequestering due to its localization in the cell membrane. In addition, membrane binding may modulate the ligand-binding properties of trHbO, making it a better oxygen sequester. The oxygen sequestering function of trHbO may also be important for facilitating the NOD reaction carried out by trHbN by providing O2 required for the reaction.

3.6. The TrHb-III from Campylobacter jejuni (trCtb) Campylobacter jejuni is a Gram-negative bacterium present in the gut of many foodsupply animals and birds. It is an obligate microaerophile, meaning oxygen is necessary for growth yet also toxic when at atmospheric concentrations. Like M. tuberculosis, C. jejuni comprises two Hbs, trCtb and Cgb, which belong to the TrHb-III and sdHb groups of Hbs, respectively [73–75]. It has been demonstrated that both Hbs are not required for the survival of the bacterium in air [73,74]. In addition, the expression of Cgb was found to be strongly and specifically induced by nitrosative stress [75]. Along the same lines, a Cgb knockout mutant of C. jejuni was shown to be hypersensitive to reactive nitrogen species [75]. As such, Cgb, like trHbN in M. tuberculosis, has been proposed to protect the bacterium from the toxic effects of NO by means of an NOD reaction. Although the trCtb knockout mutant of C. jejuni does not display any sensitivity to nitrosative stress, the expression of trCtb can be induced by NO donors [74].

Microbial Hemoglobins

253

More importantly, the O2 consumption rate of the trCtb knockout mutant cells showed a 50% reduction as compared to the wild-type cells, suggesting its involvement in regulating the flux of O2 into and within the cell [76]. TrCtb is a monomeric protein with 127 amino acid residues and a molecular mass of 14.1 kDa. The B10, E7, E10, and E11 positions of trCtb are occupied by Tyr-19, His-46, Lys-49, and Ilu-50, respectively. Like trHbO, the G8 position is occupied by a Trp residue. In the crystal structure of the cyanide-bound ferric derivative, the cyanide accepts H-bonds donated from the B10Tyr and G8Trp (Fig. 10C) [77]. The ferric state of trCtb is in an aquo-met form, with the heme in a mixture of six-coordinate high- and low-spin configurations [73]. The ferrous derivative has a five-coordinate high-spin configuration with a histidine as the sole axial ligand. The Fe His mode was found at 226 cm−1 [73], which is similar to that of most other trHbs (Table 2), indicating a staggered orientation of the proximal histidine with respect to the pyrrole nitrogen atoms of the porphyrin ring. In the CO derivative, the Fe CO and C O modes are at 515 and 1936 cm−1 , respectively [73]. The data point sits in the middle of the correlation line as shown in Fig. 6(B). In trHbN and trHbO, the mutation in the key distal polar residues causes the data point to shift along the correlation line to the lower right corner (Fig. 7A), due to the reduction of the electrostatic potential surrounding the heme-bound CO. In contrast, the mutation of either the B10Tyr to Phe or the E7His to Leu in trCtb causes the data point to shift toward the upper left corner of the inverse correlation line (Fig. 7B), indicating a higher electrostatic potential of the protein environment surrounding the CO [78]. How can the removal of a positive distal residues result in an increase in the electrostatic potential? We hypothesize that in the wild-type trCtb, the E7His forms an H-bond with the B10Tyr, thereby preventing either residue from forming an H-bond with the CO ligand. The mutation in one of the two residues releases the structural constraint on the other and allows it to form an H-bond with the CO. In addition to the closed form with the extremely high Fe CO , the E7His mutant can also adopt a wild-type-like conformation, suggesting that the B10Tyr exists in two alternative conformations in the absence of the H-bond donated from the E7His. It is noticeable that in the crystallographic structure of the ferric cyanide–bound derivative, the imidazole side chain of the E7 residue is not in a correct orientation for forming an H-bond with the B10 Tyr (Fig. 10C), suggesting that reduction of the heme iron may induce the rotation and repositioning of the His side chain. The Fe O2 and O O modes of the oxy derivative of trCtb are at 542 and 1133 cm−1 , respectively [78]. Like other trHbs, the appearance of the O O mode in the RR spectrum is attributed to an H-bonding network between the heme-bound O2 and the distal residues in its proximity [78]. In the B10Tyr → Phe and E7His → Leu single mutants, the Fe O2 /O O shift to 552/1139 cm−1 and 550/1139 cm−1 , respectively, whereas those in the B10/E7 double mutant shift to 557/1144 cm−1 . Markedly, all the Fe O2 modes exhibit frequencies lower than 560 cm−1 , suggesting that the G8Trp plays an important role in stabilizing the heme-bound dioxygen. Moreover, the Fe O2 appears to correlate well with the O O in a positive linear fashion (Fig. 11A), which came as a complete surprise. As discussed above, the Fe CO and C O frequencies of the Fe C O moiety in heme proteins are inversely correlated in a linear fashion as a result of the -backbonding effect. In contrast, due to the nonlinear nature and high polarizability of the Fe O O moiety, there is typically no clear correlation between Fe O2 and O O in six-coordinate (6C) O2 -bound heme proteins [65,79]. We propose

Changyuan Lu et al.

νFe – O2 (cm–1)

254

560

(B)

(A)

555

B10

B10/E7

550

E7

G8

B10 E7

N

545

N WT

540 1130

1135

N

N

trCtb

1140

1145

F8

νO – O (cm–1)

Fig. 11. The Fe−O2 versus O O correlation line of the wild type and mutants of trCtb and the pictorial illustration of the O2 -protein interactions. The mutants in (A) are as defined in Fig. 7. This figure is adapted from Lu et al. [78].

that the positive correlation shown in Fig. 11(A) is a result of the unique structural feature of trCtb as illustrated in Fig. 11(B). In this structure, the G8Trp anchors the proximal oxygen atom of the heme-bound dioxygen in a fixed position via an H-bond, whereas the B10Tyr, positioned by the E7His, donates additional H-bonds to the dioxygen to further stabilize it. The mutation in the B10Tyr and/or E7His perturbs the -bonding system, without affecting the -bonding interaction, thereby accounting for the positive Fe O2 −O O correlation. This unique distal H-bonding interaction leads to a slow O2 off-rate (∼0.0041 s−1 ) [78]. The mutation of the B10Tyr to Phe cause the off-rate to increase to 0.0088 s−1 , whereas the single mutation of E7His to Leu or double mutation of the E7His/B10Tyr leads to a decrease in the off-rate to 0.0003 and 0.0028 s−1 , respectively [78]. The data confirms that the heme-bound dioxygen is stabilized by accepting H-bonds from the B10Tyr, in addition to the G8Trp. They also suggest that the additional H-bond between the E7His and B10Tyr plays an important role in preventing the overstabilization of the heme-bound dioxygen. Like trHbO, the distal heme pocket of trCtb is congested with bulky aromatic residues, including the G8Trp, B9 Phe, B10Tyr, CD1Phe, and E7His (Fig. 10D), which may account for the slow on-rate of O2 and CO, as well as the preferential binding of O2 versus CO (Table 3). The extremely high oxygen affinity of trCtb, mostly resulting from the unusually slow off-rate, makes it unlikely to function as an oxygen transporter. On the other hand, the distal heme environment of trCtb is surprisingly similar to that of cytochrome c peroxidase (CcP), suggesting a role of trCtb in performing a peroxidase or P450-type of oxygen chemistry [78].

3.7. The FHb from Escherichia coli (Hmp) Escherichia coli is a Gram-negative facultative anaerobic bacterium that normally lives in the intestines of humans and animals. Escherichia coli has only one Hb, Hmp, which belongs to the FHb group of proteins. The expression of Hmp in E. coli is upregulated by NO and nitrosating agents. In addition, the lack of the hmp gene in E. coli causes it to be hypersensitive to NO and nitrosative stress [5,80], suggesting that at least part of the physiological function of Hmp is to detoxify NO, possibly via the NOD reaction [13].

Microbial Hemoglobins

255

This proposal is supported by the observation that purified Hmp binds O2 strongly and converts NO to nitrate [5,80]. In E. coli, maximal protection against nitrosative challenge was provided only by the full-length Hmp, not the heme domain (HmpHD ) alone [81], demonstrating the importance of the flavin-containing reductase in catalyzing the NOD reaction. Intriguingly, overexpression of HmpHD alone in mutant E. coli cells lacking the full-length Hmp resulted in an improvement in cell growth [81], suggesting that HmpHD , like trCtb in C. jejuni, may be able to control O2 flux in cells. Hmp has a molecular mass of 44 kDa [41]. The B10, CD1, E7, and the E11 positions in Hmp are occupied by Tyr-29, Phe-43, Gln-53, and Leu-57, respectively (Table 1), which are highly conserved in the FHb family of proteins. RR studies of the ferric derivative of Hmp show a five-coordinate high-spin heme, as indicated by the 2 and 3 frequencies at 1570 and 1491 cm−1 , respectively [82]. The data indicate that the distal heme ligand binding site of Hmp is empty, in contrast to the water or intrinsic amino acid bound heme found in most of other Hbs discovered to date. The ferrous derivative of Hmp also displays a five-coordinate high-spin configuration as indicated by the 3 and 4 frequencies at 1470 and 1353 cm−1 , respectively [82]. The Fe His mode is at 244 cm−1 (Table 2), which is higher than that of swMb (∼220 cm−1 ), but similar to those of heme peroxidases, e.g., 244 cm−1 for horseradish peroxidase (HRP) and 248 cm−1 for CcP [83,84]. The high frequency of the Fe His mode in the peroxidases has been attributed to the imidazolate character of the proximal histidine due to the presence of a strong H-bond between the histidine and a nearby negatively charged amino acid side chain [83]. Intriguingly, the ferric derivative of CcP, like Hmp, has a five-coordinate high-spin configuration [85–87]. It is believed that in CcP, the strong proximal iron– histidine bond pulls the iron out of the heme plane, and the repulsive force exerted by the pyrrole nitrogen atoms of the porphyrin ring on the water prevents it from binding to the heme iron [85–87]. Accordingly, we attribute the five-coordinate high-spin nature of the ferric heme in Hmp to the same origin. Along this line, an extended H-bonding network, involving the proximal His Glu Tyr is recognized in the crystallographic structure of Hmp (Fig. 12A). This H-bonding network is highly conserved in FHbs and sdHbs, suggesting that it may play important structural and functional roles in these Hbs. (A)

(B)

(C)

B10 B10

E7

E7 B10 E7

Hmp

E11

44-52 E10 E14 Vgb

Fig. 12. The crystal structures of Hmp (A) and Vgb (B–C). The B- and E-helices are labeled in a ribbon representation. The dotted line labeled as “44–52” in (C) indicates the missing fragment (residues 44–52) in Vgb. The other dotted lines indicate the H-bonding interactions.

256

Changyuan Lu et al.

In the CO derivative, two pairs of Fe CO /C O modes assigned to an open and a closed conformations were found at 494/1960 cm−1 and 535/1907 cm−1 , respectively (Table 1) [82]. The presence of the two alternative conformations in Hmp demonstrates the plasticity of its ligand-binding pocket. In the Fe CO /C O inverse correlation curve, the data associated with the two conformations fall at similar locations as those of trHbN, which, like Hmp, also performs a NOD reaction physiologically (Fig. 6B), suggesting that the protein plasticity is important for the execution of the NOD reaction. When the B10Tyr is mutated to Phe, the off-rate of dioxygen increases by a factor of ∼80, suggesting that in the closed conformation, the heme-bound CO accepts an H-bond from the B10Tyr [88]. When the E7Gln is mutated to Leu, the two conformations converts to a single open conformation associated with Fe CO at 499 cm−1 (unpublished data), indicating that the B10Tyr by itself cannot form an H-bond with CO, and that the E7Gln plays an important role in positioning the B10Tyr for ligand stabilization in the wild-type protein. Despite the solid evidence supporting the critical role of the B10Tyr and the E7Gln in ligand stabilization, in the crystal structure of the ligand-free ferric derivative of Hmp, the side chain group of the E7Gln protrudes out into the solvent and forms an H-bond with one of the two propionate groups of the heme (Fig. 12A). In addition, the B10Tyr is remote from the heme group because of the steric hindrance exerted by the hydrophobic side chain of the E11Leu, which sits directly on top of the heme iron. Since both the B10Tyr and E7Gln directly interact with the heme-bound ligands, the E-helix has to rotate and both the B- and the E-helices must move toward the heme plane. This anticipated ligand-induced structural transition in Hmp is similar to that proposed for trHbC and trHbS, as discussed above. Taken together, the data suggest that the proximal imidazolate ligand offers an electronic push and the distal B10Tyr-E7Gln pair provides an electronic pull to activate the O O bond of the heme-bound peroxynitrite intermediate formed during the NOD reaction [82]. Furthermore, the plasticity of the distal pocket may facilitate the release of the product nitrate. The mutation of the B10Tyr to Phe causes ∼30-fold reduction in the rate of the NOD reaction, confirming its critical role in the NOD reaction of Hmp [88]. In addition to its high NOD activity, Hmp also exhibits high affinity toward phospholipids [89]. In the crystallographic structure of the FHb from Ralstonia eutropha, a large phospholipid was found in the distal ligand-binding site [90]. The capability for the B- and E-helices to move away from the heme vicinity to accommodate a large hydrophobic phospholipid in the structure of the FHb reflects the flexible nature of its protein matrix. In addition, the conservation of the residues involved in the interaction with lipids in the FHb family of proteins [91] suggests additional physiological functions of Hmp related to membrane phospholipids.

3.8. The sdHb from Vitreoscilla sp. (Vgb) Vitreoscilla sp. is a Gram-negative obligatory aerobic bacterium. It expresses a soluble sdHb, Vgb, in response to hypoxic environments [44,92,93]. Vgb has attracted significant attention in recent years, because its expression in a variety of host cells used in biotechnology industry has been shown to improve the growth of the host cells and the productivity of the proteins of interest that are coexpressed with Vgb [6,44,94–103].

Microbial Hemoglobins

257

Vgb is a homodimer. It has 146 amino acid residues and a molecular mass of 15.7 kDa [44,104]. The amino acid sequence shares 51% identity with the N-terminal globin domain of Hmp [103]. RR studies show that similar to Hmp, Vgb exhibits a fivecoordinate high-spin configuration in both the ferric and ferrous states and a high Fe His frequency at 252 cm−1 (unpublished data). Like Hmp, two pairs of Fe CO /C O modes at 489/1965 and 536/1911 cm−1 were observed for the open and closed conformations, respectively (Fig. 6B). The RR properties of the various oxidation/ligation states of Vgb are analogous to those of Hmp, but the crystal structure of Vgb exhibits structural features distinct from that of Hmp (Fig. 12A versus 12C), for example: (i) the peptide segment connecting the B- and E-helices (amino acid #44–52) is disordered and unresolved in the crystal structure; (ii) the E7–E11 region of the peptide adopts a coil-like structure, instead of a helical turn as that observed in Hmp; and (iii) the side chain group of the E7Gln protrudes into the solvent, instead of pointing toward the heme propionate group as observed in Hmp (Fig. 12C versus 12A). On the basis of the structures shown in Fig. 12(A) and (C), a structural transition must occur for both Vgb and Hmp to enable the interaction between the heme-bound ligands and the distal amino acid residues, including the B10Tyr and E7Gln, as suggested by the RR data. The physiological function of Vgb remains unclear. It has been shown that Vgb is concentrated near the periphery of the cytosolic side of the cell membrane and it directly binds to subunit I of cytochrome bo ubiquinol oxidases [99,100]. Consequently, it was proposed that the physiological function of Vgb is to sequester oxygen from the environment and transfer it to the respiratory terminal oxidase, thereby facilitating respiration under hypoxic conditions [99]. Recently, it was shown that a chimeric protein carrying Vgb and a flavoreductase relieves nitrosative stress in E. coli cells, suggesting that Vgb, like Hmp, can also function as an NOD in the presence of the partner reductase [105]. On the basis of these observations, it has been proposed that the expression of the reductase domain as a separate protein may provide advantage for Vgb to perform two discrete functions in vivo: (i) in the single-domain homodimeric state, an oxygen sequester that facilitates oxygen transfer and (ii) in a two-domain heterodimeric state associated with its partner reductase, an NO dioxygenase to protect cells against nitrosative stress. It is not an easy task for a protein to perform two mechanistically distinct functions. In Vgb, this may be facilitated by the structural changes induced by the reductase-binding and/or the membrane-binding events.

4. FOLDING STABILITIES OF MICROBIAL Hbs It has been shown by Olson and coworkers that the mutation of the distal histidine to apolar substituents increases the folding stability of swMb by a factor of 10–30 [106]. However, the enhanced folding stability increases the autooxidation rate and decreases the oxygen-binding affinity. On the other hand, by analyzing 13 mammalian Mbs with 80–99% amino acid identity, it was found that Mb from deep-diving mammals exhibit higher folding stability than those from terrestrial mammals, although they exhibit similar oxygen-binding properties [107]. On this basis, it was concluded that the effects of mutations along the evolutionary pathway should be compensatory unless there is a strong physiological need for resistance to unfolding; in addition, under selective

258

Changyuan Lu et al.

pressure, an accumulation of small stabilizing or destabilizing mutations will not alter the functional character of the protein [107]. Considering the fact that trHbs are evolutionarily distant and functionally distinct from conventional globins with 3-over-3 -helical structure, it is important to understand the folding properties of this new group of Hbs. Figure 13 shows the guanidine hydrochloride (GHCl) unfolding curves of trHbN, trHbO, and trCtb with respect to Vgb and horse heart Mb (hhMb). The free energy of folding at any given GHCl concentration can be estimated by assuming the folding as a two-state transition from the native state (N) to the unfolded state (U). As shown in Fig. 14, the free energy of folding linearly correlates with the GHCl concentration applied. The standard free energies of folding ( GH2 O ) obtained by extrapolating the linear free energy line back to zero GHCl concentration are listed in Table 4, along with the corresponding Cm and m-value for each Hb, where Cm is the transition midpoint of the titration curve (i.e., the GHCl concentration at which the G is zero) and the m-value is the slope of the linear free energy line. In general,

(A)

0.8

trCtb

trHbO

0.6 trHbN

0.4 0.2

(B)

1.0

[F] / [F] + [U]

[F] / [F] + [U]

1.0

0.8 trCtb

hhMb

0.6 0.4 0.2

Vgb

0.0

0.0 0

1

2

3

4

5

6

7

1

0

2

GHCl [M]

3

4

5

6

7

GHCl [M]

Fig. 13. The equilibrium unfolding curves of trHbN, trHbO, trCtb, Vgb, and horse heart myoglobin. The unfolding reaction of each Hb was monitored by Circular Dichroism (CD) at 222 nm (unpublished results).

Vgb

ΔGFolding (kcal/mol)

0.8 0.4 0

trCtb –0.4

trHbO

–0.8

trHbN Mb 1

2

3

4

5

6

GHCl [M]

Fig. 14. The folding free energies ( GFolding ) calculated from the data shown in Fig. 13 as a function of GHCl concentration for trHbN, trHbO, trCtb, Vgb, and hhMb (A).

Microbial Hemoglobins

259

Table 4. The Cm , m-value, and GH2 O obtained from the equilibrium unfolding studies of the various Hbs hhMb trHbN (TrHb-I) trHbO (TrHb-II) trCtb (TrHb-III) Vgb (sdHb) Cm (M) 15 m-value (kcal/mol/M) 50

GH2 O (kcal/mol) −74

25 10 −26

26 13 −36

52 45 −23

19 09 −18

Cm is determined by the resistance of the protein to the denaturant, whereas the m-value is defined by the solvent accessibility of the protein matrix (a small m-value indicates a high solvent accessibility, hence a more loosely packed protein matrix) [108,109]. As shown in Fig. 13, the two Hbs from M. tuberculosis, trHbN and trHbO, show very similar transition curves, whereas trCtb from C. jejuni has much higher stability as compared to Vgb, which is a structural analog of Cgb, the sdHb from C. jejuni. In addition, the Cm of all trHbs with 2-over-2 -helical structures, trHbN, trHbO, and trCtb, appear to be higher than those of the globins with 3-over-3 -helical structures (i.e., Vgb and hhMb). On the basis of Table 4, the folding stability, as defined by GH2 O , is of the following order: trCtb (−23 kcal/mol) >> hhMb (−7.4 kcal/mol) > trHbO (−3.6 kcal/mol) > trHbN (−2.6 kcal/mol) > Vgb (−1.8 kcal/mol). The low folding stability of trHbN and trHbO with respect to hhMb may be attributed to the lack of the A-helix since it is generally believed that in Mb, the hydrophobic AGH core made up by the A-, G-, and H-helices is important for maintaing its folding stability [110–113]. In addition, several large hydrophobic cavities/tunnels are present in trHbN and trHbO, but not in hhMb, which may further destabilize their proetin fold. Although the A-helix is also missing in trCtb, it exhibits extremely high Cm and high folding stability. The extreme folding stability of trCtb may be attributed to the unusually large number of aromatic residues surrounding the heme group, which is important in holding the overall structure together via hydrophobic interactions, and/or the absence of the two Gly Gly motifs located at the AB and EF interhelical corners, which are highly conserved in TrHb-I and TrHb-II, but are not present in TrHb-III. The m-value of trCtb is similar to that of hhMb, which is much higher than those of the other two trHbs, suggesting that trCtb, like hhMb, has a protein matrix with very low solvent accessibility. Vgb, on the other hand, has the lowest m-value, although it has a global structure similar to hhMb. It remains to be investigated as to how the folding stabilities of the microbial Hbs evolved during the evolution and how they impact on their physiological functions. However, it is clear on the basis of the data listed in Tables 3 and 4 that the folding stability of the Hbs does not correlate with their ligand-binding properties; for example, the folding properties of trHbN and trHbO are very similar but their ligand-binding properties are distinct; on the other hand, the ligand-binding properties of trHbO and trCtb are similar but the folding stability is very different.

5. CLOSING REMARKS The microbial Hbs show a diverse spectrum of structural and ligand-binding properties, as well as folding stabilities. In mammalian globins, the oxygen transport function is

260

Changyuan Lu et al.

believed to be regulated by the motion of the E7His, the so-called “histidine gate,” which swings in and out of the distal pocket in response to ligand loading and release. In contrast, the oxygen binding and release in the microbial Hbs seem to rely on the intricate H-bonding interactions between the distal residues, including the B10Tyr, E7Gln/His, E11Gln/Thr, CD1Tyr, and/or G8Trp. In TrHb-I group of Hbs, the B10Tyr plays the dominant role in ligand stabilization by directly forming H-bonds with the heme-bound ligand. In trHbN, the ligand stabilization is further enhanced by an additional H-bond donated from the E11Gln to the B10Tyr to properly position the B10Tyr for ligand stabilization. Conversely, in trHbP, the additional H-bond between the E11Thr and the B10Tyr forces the phenolic oxygen of the B10Tyr to face the heme-bound ligand, thereby destabilizing it. In contrast, in TrHb-II and TrHb-III groups of Hbs, the ligand is primarily stabilized by H-bonds donated from the G8Trp to the ligand. In trHbO (a TrHb-II), the ligand stabilization is further enhanced by an H-bond donated from the CD1Tyr to the ligand; whereas in trCtb (a TrHb-III), an additional H-bond between the B10Tyr and E7His is used to prevent overstabilization of the ligand by the B10Tyr. These sophisticated regulatory mechanisms for ligand binding may be further modulated by the coordination of an intrinsic amino acid, e.g., the B10Tyr in trHbC or the E10His in trHbS, to the distal ligand-binding site. On the other hand, in Hmp (a FHb) and Vgb (a sdHb), the heme-bound ligand is stabilized by the B10Tyr and E7Gln, synergistically. With CO as the structural probe, the electrostatic potential of the heme ligand-binding pocket of the microbial Hbs can be estimated. Intriguingly, as shown in Fig. 15, the electrostatic potential of the heme ligand-binding pocket of the microbial Hbs (as indicated by the Fe CO ) appear to correlate well with their O2 and CO binding rate constants. It is noticeable that all the Hbs implicated in NOD functionality, including trHbN, Cgb, and Hmp, exhibit low electrostatic potential (here, only the open conformation is considered, assuming that the open and closed conformations are interconverting and only the open conformation can uptake ligands) and fast on-rates; whereas trHbO and trCtb, which comprise rigid and congested distal pocket exhibit high electrostatic potential and slow on-rates. Although the physiological functions of these microbial Hbs remain to be further explored, it is clear that the unique structural features of the microbial Hbs point to functions other than oxygen transport.

8.0

log(k on)

7.0 6.0 5.0

k on(O2)

Cgb Hmp trHbN k on(CO)

Mb

4.0

trCtb trHbO

3.0 490

500

νFe – CO

510

520

530

(cm–1)

Fig. 15. A plot of log(kon ) of the various microbial globins versus their associated Fe kon is the O2 or CO on-rates, as indicated. For trHbN, Cgb, and Hmp, only the Fe corresponding open conformations are plotted.

CO , CO

where of the

Microbial Hemoglobins

261

ACKNOWLEDGMENT This work was supported by the National Institute of Health Research Grant HL65465 to S.-R.Y. We would like to thank Dr. Denis L. Rousseau for many invaluable discussions.

REFERENCES [1] Poole, R.K. and Hughes, M.N. (2000) New functions for the ancient globin family: Bacterial responses to nitric oxide and nitrosative stress. Mol. Microbiol., 36, 775–783. [2] Weber, R.E. and Vinogradov, S.N. (2001) Nonvertebrate hemoglobins: Functions and molecular adaptations. Physiol. Rev., 81, 569–628. [3] Egawa, T. and Yeh, S.R. (2005) Structural and functional properties of hemoglobins from unicellular organisms as revealed by resonance Raman spectroscopy. J. Inorg. Biochem., 99, 72–96. [4] Wittenberg, J.B., Bolognesi, M., Wittenberg, B.A. and Guertin, M. (2002) Truncated hemoglobins: A new family of hemoglobins widely distributed in bacteria, unicellular eukaryotes, and plants. J. Biol. Chem., 277, 871–874. [5] Wu, G., Wainwright, L.M. and Poole, R.K. (2003) Microbial globins. Adv. Microb. Physiol., 47, 255–310. [6] Frey, A.D. and Kallio, P.T. (2003) Bacterial hemoglobins and flavohemoglobins: Versatile proteins and their impact on microbiology and biotechnology. FEMS Microbiol. Rev., 27, 525–545. [7] Vinogradov, S.N., Walz, D.A. and Pohajdak, B. (1992) Organization of non-vertebrate globin genes. Comp. Biochem. Physiol. B, 103, 759–773. [8] Burmester, T., Ebner, B., Weich, B. and Hankeln, T. (2002) Cytoglobin: A novel globin type ubiquitously expressed in vertebrate tissues. Mol. Biol. Evol., 19, 416–421. [9] Hankeln, T., Ebner, B., Fuchs, C. et al. (2005) Neuroglobin and cytoglobin in search of their role in the vertebrate globin family. J. Inorg. Biochem., 99, 110–119. [10] Burmester, T., Weich, B., Reinhardt, S. and Hankeln, T. (2000) A vertebrate globin expressed in the brain. Nature, 407, 520–523. [11] Cramm, R., Siddiqui, R.A. and Friedrich, B. (1994) Primary sequence and evidence for a physiological function of the flavohemoprotein of Alcaligenes eutrophus. J. Biol. Chem., 269, 7349–7354. [12] Crawford, M.J. and Goldberg, D.E. (1998) Role for the Salmonella flavohemoglobin in protection from nitric oxide. J. Biol. Chem., 273, 12543–12547. [13] Gardner, P.R., Gardner, A.M., Martin, L.A. and Salzman, A.L. (1998) Nitric oxide dioxygenase: An enzymic function for flavohemoglobin. Proc. Natl. Acad. Sci. USA, 95, 10378– 10383. [14] Gardner, P.R., Gardner, A.M., Martin, L.A. et al. (2000) Nitric oxide dioxygenase activity and function of flavohemoglobins: Sensitivity to nitric oxide and carbon monoxide inhibition. J. Biol. Chem., 275, 31581–31589. [15] Minning, D.M., Gow, A.J., Bonaventura, J. et al. (1999) Ascaris haemoglobin is a nitric oxide-activated ‘deoxygenase’. Nature, 401, 497–502. [16] Goldberg, D.E. (1999) Oxygen-avid hemoglobin of ascaris. Chem. Rev., 99, 3371–3378. [17] Appleby, C.A. (1992) The origin and functions of hemoglobin in plants. Sci. Prog., 76, 365–398. [18] Appleby, C.A. (1984) Leghemoglobin and rhizobium respiration. Annu. Rev. Plant. Physiol., 35, 443–478.

262

Changyuan Lu et al.

[19] Wittenberg, J.B. (1974) Facilitated oxygen diffusion. The role of leghemoglobin in nitrogen fixation by bacteroids isolated from soybean root nodules. J. Biol. Chem., 249, 4057–4066. [20] LaCount, M.W., Zhang, E., Chen, Y.P. et al. (2000) The crystal structure and amino acid sequence of dehaloperoxidase from amphitrite ornata indicate common ancestry with globins. J. Biol. Chem., 275, 18712–18716. [21] Franzen, S., Roach, M.P., Chen, Y.-P. et al. (1998) The unusual reactivities of amphitrite ornata dehaloperoxidase and notomastus lobatus chloroperoxidase do not arise from a histidine imidazolate proximal heme iron ligand. J. Am. Chem. Soc., 120, 4658–4661. [22] Freitas, T.A., Saito, J.A., Hou, S. and Alam, M. (2005) Globin-coupled sensors, protoglobins, and the last universal common ancestor. J. Inorg. Biochem., 99, 23–33. [23] Riggs, A.F. (1998) Self-association, cooperativity and supercooperativity of oxygen binding by hemoglobins. J. Exp. Biol., 201, 1073–1084. [24] Perutz, M.F., Wilkinson, A.J., Paoli, M. and Dodson, G.G. (1998) The stereochemical mechanism of the cooperative effects in hemoglobin revisited. Annu. Rev. Biophys. Biomol. Struct., 27, 1–34. [25] Nagai, K. and Kitagawa, T. (1980) Differences in Fe(II) N epsilon(His-F8) stretching frequencies between deoxyhemoglobins in the two alternative quaternary structures. Proc. Natl. Acad. Sci. USA, 77, 2033–2037. [26] Friedman, J.M., Rousseau, D.L., Ondrias, M.R. and Stepnoski, R.A. (1982) Transient Raman study of hemoglobin: Structural dependence of the iron-histidine linkage. Science, 218, 1244–1246. [27] Wajcman, H. and Kiger, L. (2002) Hemoglobin, from microorganisms to man: a single structural motif, multiple functions. C. R. Biol., 325, 1159–1174. [28] Moens, L., Vanfleteren, J., Van de Peer, Y. et al. (1996) Globins in nonvertebrate species: Dispersal by horizontal gene transfer and evolution of the structure-function relationships. Mol. Biol. Evol., 13, 324–333. [29] Iwaasa, H., Takagi, T. and Shikama, K. (1989) Protozoan myoglobin from Paramecium caudatum. Its unusual amino acid sequence. J. Mol. Biol., 208, 355–358. [30] Potts, M., Angeloni, S.V., Ebel, R.E. and Bassam, D. (1992) Myoglobin in a cyanobacterium. Science, 256, 1690–1691. [31] Watts, R.A., Hunt, P.W., Hvitved, A.N. et al. (2001) A hemoglobin from plants homologous to truncated hemoglobins of microorganisms. Proc. Natl. Acad. Sci. USA, 98, 10119–10124. [32] Couture, M., Chamberland, H., St-Pierre, B. et al. (1994) Nuclear genes encoding chloroplast hemoglobins in the unicellular green alga Chlamydomonas eugametos. Mol. Gen. Genet., 243, 185–197. [33] Couture, M., Yeh, S.R., Wittenberg, B.A. et al. (1999) A cooperative oxygen-binding hemoglobin from Mycobacterium tuberculosis. Proc. Natl. Acad. Sci. USA, 96, 11223–11228. [34] Mukai, M., Savard, P.Y., Ouellet, H. et al. (2002) Unique ligand-protein interactions in a new truncated hemoglobin from Mycobacterium tuberculosis. Biochemistry, 41, 3897–3905. [35] Pesce, A., Couture, M., Dewilde, S., Guertin, M., Yamauchi, K., Ascenzi, P., Moens, L. and Bolognesi, M. (2000) A novel two-over-two alpha-helical sandwich fold is characteristic of the truncated hemoglobin family. EMBO J., 19, 2424–2434. [36] Milani, M., Pesce, A., Ouellet, Y. et al. (2001) Mycobacterium tuberculosis hemoglobin N displays a protein tunnel suited for O2 diffusion to the heme. EMBO J., 20, 3902–3909. [37] Milani, M., Savard, P.Y., Ouellet, H. et al. (2003) A TyrCD1/TrpG8 hydrogen bond network and a TyrB10TyrCD1 covalent link shape the heme distal site of Mycobacterium tuberculosis hemoglobin O. Proc. Natl. Acad. Sci. USA, 100, 5766–5771. [38] Probst, I., Wolf, G. and Schlegel, H.G. (1979) An oxygen-binding flavohemoprotein from Alcaligenes eutrophus. Biochim. Biophys. Acta, 576, 471–478. [39] Iwaasa, H., Takagi, T. and Shikama, K. (1992) Amino acid sequence of yeast hemoglobin. A two-domain structure. J. Mol. Biol., 227, 948–954.

Microbial Hemoglobins

263

[40] Zhu, H. and Riggs, A.F. (1992) Yeast flavohemoglobin is an ancient protein related to globins and a reductase family. Proc. Natl. Acad. Sci. USA, 89, 5015–5019. [41] Vasudevan, S.G., Armarego, W.L., Shaw, D.C. et al. (1991) Isolation and nucleotide sequence of the hmp gene that encodes a haemoglobin-like protein in Escherichia coli K-12. Mol. Gen. Genet., 226, 49–58. [42] Iijima, M., Shimizu, H., Tanaka, Y. and Urushihara, H. (2000) Identification and characterization of two flavohemoglobin genes in Dictyostelium discoideum. Cell Struct. Funct., 25, 47–55. [43] Takaya, N., Suzuki, S., Matsuo, M. and Shoun, H. (1997) Purification and characterization of a flavohemoglobin from the denitrifying fungus Fusarium oxysporum. FEBS Lett., 414, 545–548. [44] Wakabayashi, S., Matsubara, H. and Webster, D.A. (1986) Primary sequence of a dimeric bacterial haemoglobin from Vitreoscilla. Nature, 322, 481–483. [45] Tarricone, C., Calogero, S., Galizzi, A. et al. (1997) Expression, purification, crystallization, and preliminary X-ray diffraction analysis of the homodimeric bacterial hemoglobin from Vitreoscilla stercoraria. Proteins, 27, 154–156. [46] Yeh, S.R., Couture, M., Ouellet, Y. et al. (2000) A cooperative oxygen binding hemoglobin from Mycobacterium tuberculosis. Stabilization of heme ligands by a distal tyrosine residue. J. Biol. Chem., 275, 1679–1684. [47] Pathania, R., Navani, N.K., Rajamohan, G. and Dikshit, K.L. (2002) Mycobacterium tuberculosis hemoglobin HbO associates with membranes and stimulates cellular respiration of recombinant Escherichia coli. J. Biol. Chem., 277, 15293–15302. [48] Samuni, U., Dantsker, D., Ray, A. et al. (2003) Kinetic modulation in carbonmonoxy derivatives of truncated hemoglobins: The role of distal heme pocket residues and extended apolar tunnel. J. Biol. Chem., 278, 27241–27250. [49] Yu, N.-T. and Kerr, E.A. (1988) In Biological Applications of Raman Spectroscopy: Resonance Raman Spectra of Hemes and Metalloproteins (T.G. Spiro, ed.). New York: John Wiley & Sons, pp. 39–95. [50] Vogel, K.M., Kozlowski, P.M., Zgierski, M.Z. and Spiro, T.G. (2000) Role of the axial ligand in heme-CO backbonding; DFT analysis of vibrational data. Inorg. Chim. Acta, 297, 11–17. [51] Feis, A., Rodriguez-Lopez, J.N., Thorneley, R.N.F. and Smulevich, G. (1998) The distal cavity structure of carbonyl horseradish peroxidase as probed by the resonance Raman spectra of His 42 Leu and Arg 38 Leu mutants. Biochemistry, 37, 13575–13581. [52] Li, T., Quillin, M.L., Phillips, G.N., Jr. and Olson, J.S. (1994) Structural determinants of the stretching frequency of CO bound to myoglobin. Biochemistry, 33, 1433–1446. [53] Hirota, S., Li, T., Phillips, G.N. et al. (1996) Perturbation of the Fe O2 bond by nearby residues in heme pocket: Observation of Fe O2 Raman bands for oxymyoglobin mutants. J. Am. Chem. Soc., 118, 7845–7846. [54] Mukai, M., Ouellet, Y., Ouellet, H. et al. (2004) NO binding induced conformational changes in a truncated hemoglobin from Mycobacterium tuberculosis. Biochemistry, 43, 2764–2770. [55] Yeh, S.R. (2004) Novel inter-subunit communication mechanism in a truncated hemoglobin from M. Tuberculosis. J. Phys. Chem. B., 108, 1478–1484. [56] Gardner, P.R. (2005) Nitric oxide dioxygenase function and mechanism of flavohemoglobin, hemoglobin, myoglobin and their associated reductases. J. Inorg. Biochem., 99, 247–266. [57] Ouellet, H., Ouellet, Y., Richard, C. et al. (2002) Truncated hemoglobin HbN protects Mycobacterium bovis from nitric oxide. Proc. Natl. Acad. Sci. USA, 99, 5902–5907. [58] Das, T.K., Weber, R.E., Dewilde, S. et al. (2000) Ligand binding in the ferric and ferrous states of Paramecium hemoglobin. Biochemistry, 39, 14330–14340. [59] Gibson, Q.H. and Smith, M.H. (1965) Rates of reaction of Ascaris haemoglobins with ligands. Proc. R. Soc. Lond. B. Biol. Sci., 163, 206–214.

264

Changyuan Lu et al.

[60] Pesce, A., Nardini, M., Ascenzi, P. et al. (2004) Thr-E11 regulates O2 affinity in Cerebratulus lacteus mini-hemoglobin. J. Biol. Chem., 279, 33662–33672. [61] Kundu, S., Blouin, G.C., Premer, S.A. et al. (2004) Tyrosine B10 inhibits stabilization of bound carbon monoxide and oxygen in soybean leghemoglobin. Biochemistry, 43, 6241–6252. [62] Couture, M., Das, T.K., Lee, H.C. et al. (1999) Chlamydomonas chloroplast ferrous hemoglobin. Heme pocket structure and reactions with ligands. J. Biol. Chem., 274, 6898–6910. [63] Couture, M., Das, T.K., Savard, P.Y. et al. (2000) Structural investigations of the hemoglobin of the cyanobacterium Synechocystis PCC6803 reveal a unique distal heme pocket. Eur. J. Biochem., 267, 4770–4780. [64] Das, T.K., Couture, M., Lee, H.C. et al. (1999) Identification of the ligands to the ferric heme of Chlamydomonas chloroplast hemoglobin: Evidence for ligation of tyrosine-63 (B10) to the heme. Biochemistry, 38, 15360–15368. [65] Das, T.K., Couture, M., Ouellet, Y. et al. (2001) Simultaneous observation of the O O and Fe O2 stretching modes in oxyhemoglobins. Proc. Natl. Acad. Sci. USA, 98, 479–484. [66] Scott, N.L. and Lecomte, J.T. (2000) Cloning, expression, purification, and preliminary characterization of a putative hemoglobin from the cyanobacterium Synechocystis sp. PCC 6803. Protein Sci., 9, 587–597. [67] Falzone, C.J., Christie Vu, B., Scott, N.L. and Lecomte, J.T. (2002) The solution structure of the recombinant hemoglobin from the cyanobacterium Synechocystis sp. PCC 6803 in its hemichrome state. J. Mol. Biol., 324, 1015–1029. [68] Hvitved, A.N., Trent, J.T., III, Premer, S.A. and Hargrove, M.S. (2001) Ligand binding and hexacoordination in synechocystis hemoglobin. J. Biol. Chem., 276, 34714–34721. [69] Ouellet, H., Juszczak, L., Dantsker, D. et al. (2003) Reactions of Mycobacterium tuberculosis truncated hemoglobin O with ligands reveal a novel ligand-inclusive hydrogen bond network. Biochemistry, 42, 5764–5774. [70] Samuni, U., Ouellet, Y., Guertin, M. et al. (2004) The absence of proximal strain in the truncated hemoglobins from Mycobacterium tuberculosis. J. Am. Chem. Soc., 126, 2682–2683. [71] Springer, B.A., Sligar, S.G., Olson, J.S. and Phillips, G.N., Jr. (1994) Mechanisms of ligand recognition in myoglobin. Chem. Rev., 94, 699–714. [72] Spiro, T.G. and Kozlowski, P.M. (2001) Is the CO adduct of myoglobin bent, and does it matter? Acc. Chem. Res., 34, 137–144. [73] Wainwright, L.M., Wang, Y., Park, S.F. et al. (2006) Purification and spectroscopic characterization of Ctb, a group III truncated hemoglobin implicated in oxygen metabolism in the food-borne pathogen Campylobacter jejuni. Biochemistry, 45, 6003–6011. [74] Wainwright, L.M., Elvers, K.T., Park, S.F. and Poole, R.K. (2005) A truncated haemoglobin implicated in oxygen metabolism by the microaerophilic food-borne pathogen Campylobacter jejuni. Microbiology, 151, 4079–4091. [75] Elvers, K.T., Wu, G., Gilberthorpe, N.J. et al. (2004) Role of an inducible single-domain hemoglobin in mediating resistance to nitric oxide and nitrosative stress in Campylobacter jejuni and Campylobacter coli. J. Bacteriol., 186, 5332–5341. [76] Elvers, K.T., Turner, S.M., Wainwright, L.M. et al. (2005) NssR, a member of the Crp-Fnr superfamily from Campylobacter jejuni, regulates a nitrosative stress-responsive regulon that includes both a single-domain and a truncated haemoglobin. Mol. Microbiol., 57, 735–750. [77] Nardini, M., Pesce, A., Labarre, M. et al. (2006) Structural determinants in the group III truncated hemoglobin from Campylobacter jejuni. J. Biol. Chem., 281, 37803–37812. [78] Lu, C., Egawa, T., Wainwright, L.M., Poole, R.K. and Yeh, S.R. (2007) Structural and functional properties of a truncated hemoglobin from a food-borne pathogen Campylobacter jejuni. J. Biol. Chem., 282, 13627–13636.

Microbial Hemoglobins

265

[79] Vogel, K.M., Kozlowski, P.M., Zgierski, M.Z. and Spiro, T.G. (1999) Determinants of the FeXO (X ) C, N, O) vibrational frequencies in heme adducts from experiment and density functional theory. J. Am. Chem. Soc., 121, 9915–9921. [80] Poole, R.K. (2005) Nitric oxide and nitrosative stress tolerance in bacteria. Biochem. Soc. Trans., 33, 176–180. [81] Poole, R.K., Anjum, M.F., Membrillo-Hernandez, J. et al. (1996) Nitric oxide, nitrite, and Fnr regulation of hmp (flavohemoglobin) gene expression in Escherichia coli K-12. J. Bacteriol., 178, 5487–5492. [82] Mukai, M., Mills, C.E., Poole, R.K. and Yeh, S.R. (2001) Flavohemoglobin: A globin with a peroxidase-like catalytic site. J. Biol. Chem., 276, 7272–7277. [83] Smulevich, G., Hu, S., Rodgers, K.R. et al. (1996) Heme-protein interactions in cytochrome c peroxidase revealed by site-directed mutagenesis and resonance Raman spectra of isotopically labeled hemes. Biospectroscopy, 2, 365–376. [84] Dasgupta, S., Rousseau, D.L., Anni, H. and Yonetani, T. (1989) Structural characterization of cytochrome c peroxidase by resonance Raman scattering. J. Biol. Chem., 264, 654–662. [85] Smulevich, G. (1998) Understanding heme cavity structure of peroxidases: Comparison of electronic absorption and resonance Raman spectra with crystallographic results. Biospectroscopy, 4, S3–S17. [86] Anni, H. and Yonetani, T. (1988) The coordination and spin states of yeast cytochrome c peroxidase and their implication to peroxidase mechanism. Prog. Clin. Biol. Res., 274, 437–449. [87] Yonetani, T. and Anni, H. (1987) Yeast cytochrome c peroxidase. Coordination and spin states of heme prosthetic group. J. Biol. Chem., 262, 9547–9554. [88] Gardner, A.M., Martin, L.A., Gardner, P.R. et al. (2000) Steady-state and transient kinetics of Escherichia coli nitric-oxide dioxygenase (flavohemoglobin). The B10 tyrosine hydroxyl is essential for dioxygen binding and catalysis. J. Biol. Chem., 275, 12581–12587. [89] Ilari, A., Bonamore, A., Farina, A. et al. (2002) The X-ray structure of ferric Escherichia coli flavohemoglobin reveals an unexpected geometry of the distal heme pocket. J. Biol. Chem., 277, 23725–23732. [90] Ermler, U., Siddiqui, R.A., Cramm, R. and Friedrich, B. (1995) Crystal structure of the flavohemoglobin from Alcaligenes eutrophus at 1.75 A resolution. EMBO J., 14, 6067–6077. [91] Ollesch, G., Kaunzinger, A., Juchelka, D. et al. (1999) Phospholipid bound to the flavohemoprotein from Alcaligenes eutrophus. Eur. J. Biochem., 262, 396–405. [92] Dikshit, K.L., Dikshit, R.P. and Webster, D.A. (1990) Study of Vitreoscilla globin (vgb) gene expression and promoter activity in E. coli through transcriptional fusion. Nucleic Acids Res., 18, 4149–4155. [93] Khosla, C. and Bailey, J.E. (1989) Characterization of the oxygen-dependent promoter of the Vitreoscilla hemoglobin gene in Escherichia coli. J. Bacteriol., 171, 5995–6004. [94] Khosravi, M., Webster, D.A. and Stark, B.C. (1990) Presence of the bacterial hemoglobin gene improves alpha-amylase production of a recombinant Escherichia coli strain. Plasmid., 24, 190–194. [95] Wei, M.L., Webster, D.A. and Stark, B.C. (1998) Metabolic engineering of Serratia marcescens with the bacterial hemoglobin gene: Alterations in fermentation pathways. Biotechnol. Bioeng., 59, 640–646. [96] Khosla, C., Curtis, J.E., DeModena, J., et al. (1990) Expression of intracellular hemoglobin improves protein synthesis in oxygen-limited Escherichia coli. Biotechnology (NY), 8, 849–853. [97] Webster, D.A. (1988) Structure and function of bacterial hemoglobin and related proteins. Adv. Inorg. Biochem., 7, 245–265. [98] Webster, D.A. and Hackett, D.P. (1966) The purification and properties of cytochrome o from Vitreoscilla. J. Biol. Chem., 241, 3308–3315.

266

Changyuan Lu et al.

[99] Park, K.W., Kim, K.J., Howard, A.J. (2002) Vitreoscilla hemoglobin binds to subunit I of cytochrome bo ubiquinol oxidases. J. Biol. Chem., 277, 33334–33337. [100] Ramandeep, Hwang, K.W., Raje, M. et al. (2001) Vitreoscilla hemoglobin. Intracellular localization and binding to membranes. J. Biol. Chem., 276, 24781–24789. [101] Dikshit, K.L., Orii, Y., Navani, N. et al. (1998) Site-directed mutagenesis of bacterial hemoglobin: The role of glutamine (E7) in oxygen-binding in the distal heme pocket. Arch. Biochem. Biophys., 349, 161–166. [102] Bolognesi, M., Boffi, A., Coletta, M. et al. (1999) Anticooperative ligand binding properties of recombinant ferric Vitreoscilla homodimeric hemoglobin: A thermodynamic, kinetic and X-ray crystallographic study. J. Mol. Biol., 291, 637–650. [103] Tarricone, C., Galizzi, A., Coda, A. et al. (1997) Unusual structure of the oxygen-binding site in the dimeric bacterial hemoglobin from Vitreoscilla sp. Structure, 5, 497–507. [104] Dikshit, K.L. and Webster, D.A. (1988) Cloning, characterization and expression of the bacterial globin gene from Vitreoscilla in Escherichia coli. Gene, 70, 377–386. [105] Kaur, R., Pathania, R., Sharma, V. et al. (2002) Chimeric Vitreoscilla hemoglobin (VHb) carrying a flavoreductase domain relieves nitrosative stress in Escherichia coli: New insight into the functional role of VHb. Appl. Environ. Microbiol., 68, 152–160. [106] Hargrove, M.S., Krzywda, S., Wilkinson, A.J. et al. (1994) Stability of myoglobin: A model for the folding of heme proteins. Biochemistry, 33, 11767–11775. [107] Scott, E.E., Paster, E.V. and Olson, J.S. (2000) The stabilities of mammalian apomyoglobins vary over a 600-fold range and can be enhanced by comparative mutagenesis. J. Biol. Chem., 275, 27129–27136. [108] Pace, C.N. (1986) Determination and analysis of urea and guanidine hydrochloride denaturation curves. Methods Enzymol., 131, 266–280. [109] Alonso, D.O. and Dill, K.A. (1991) Solvent denaturation and stabilization of globular proteins. Biochemistry, 30, 5974–5985. [110] Hughson, F.M., Wright, P.E. and Baldwin, R.L. (1990) Structural characterization of a partly folded apomyoglobin intermediate. Science, 249, 1544–1548. [111] Barrick, D., Hughson, F.M. and Baldwin, R.L. (1994) Molecular mechanisms of acid denaturation. The role of histidine residues in the partial unfolding of apomyoglobin. J. Mol. Biol., 237, 588–601. [112] Eliezer, D., Jennings, P.A., Dyson, H.J. and Wright, P.E. (1997) Populating the equilibrium molten globule state of apomyoglobin under conditions suitable for structural characterization by NMR. FEBS Lett., 417, 92–96. [113] Gilmanshin, R., Williams, S., Callender, R.H. et al. (1997) Fast events in protein folding: Relaxation dynamics of secondary and tertiary structure in native apomyoglobin. Proc. Natl. Acad. Sci. USA, 94, 3709–3713. [114] Giangiacomo, L., Mattu, M., Arcovito, A. et al. (2001) Monomer-dimer equilibrium and oxygen binding properties of ferrous Vitreoscilla hemoglobin. Biochemistry, 40, 9311–9316. [115] Springer, B.A., Egeberg, K.D., Sligar, S.G. et al. (1989) Discrimination between oxygen and carbon monoxide and inhibition of autooxidation by myoglobin. Site-directed mutagenesis of the distal histidine. J. Biol. Chem., 264, 3057–3060. [116] Quillin, M.L., Li, T., Olson, J.S. et al. (1995) Structural and functional effects of apolar mutations of the distal valine in myoglobin. J. Mol. Biol., 245, 416–436.

Part IV Heme NOx Interactions

This page intentionally left blank

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 11

The Reaction between Nitrite and Hemoglobin: The Role of Nitrite in Hemoglobin-mediated Hypoxic Vasodilation Daniel B. Kim-Shapiroa , Mark T. Gladwinb , Rakesh P. Patelc , and Neil Hoggd a

Department of Physics, Wake Forest University, Winston-Salem, NC Vascular Medicine Branch, NHLBI and Critical Care Medicine Department, Clinical Center, National Institutes of Health, Bethesda, MD c Department of Pathology and Center for Free Radical Biology, University of Alabama, Birmingham, AL d Department of Biophysics and Free Radical Research Center, Medical College of Wisconsin, Milwaukee, WI b

1. INTRODUCTION Recent evidence suggests that plasma nitrite anion represents a latent substance that can be activated by hemoglobin in areas of hypoxia to elicit vasodilation [1]. The mechanisms by which activation and vasodilation occur are currently uncertain and are under intense study. Although the reaction between nitrite and hemoglobin has been appreciated since at least the middle 1800s, a definitive mechanistic understanding of these reactions is lacking. In this chapter, we survey published mechanisms and highlight how such mechanisms either complement or are at odds with the recent physiological findings. In addition, we place the nitrite/hemoglobin reaction in its physiological and pharmacological context.

2. THE CHEMISTRY OF THE NITRITE/HEMOGLOBIN REACTION 2.1. The Reaction between Nitrite and Oxyhemoglobin (oxyHb) Ask most researchers in the nitric oxide or hemoglobin fields “what happens when you mix nitrite with oxyHb” and they will say the same thing: nitrite gets oxidized to nitrate and the hemoglobin gets oxidized to the ferric form (methemoglobin or metHb).

Much of this chapter appeared in Journal form and is reprinted from Journal of Inorganic Biochemistry, Vol. 99, Daniel B. Kim-Shapiro, Mark T. Gladwin, Rakesh P. Patel, and Neil Hogg. The reaction between nitrite and hemoglobin: the role of nitrite in hemoglobin-mediated hypoxic vasodilation, 237–246. Copyright (2005), with permission from Elsevier.

270

D.B. Kim-Shapiro et al.

Ask these same researchers for the mechanism and you will likely also get the same answer: “It’s complicated!”. Nevertheless, there are often underlying assumptions that (i) this reaction must be responsible for the oxidation of nitrite to nitrate in the blood stream, that (ii) this reaction is perhaps the most important factor in determining why nitrate and not nitrite is the predominant metabolite of nitric oxide in vivo, and that (iii) this reaction determines why plasma nitrite is maintained at low- to submicromolar levels. In this section, we will detail what is known about this reaction, suggest that many of these assumptions are not entirely accurate, and then contrast it to the reaction between nitrite and hemoglobin in the deoxygenated state to highlight the oxygen-sensing capability of the nitrite/hemoglobin reaction. The first reports of a reaction between nitrite and hemoglobin came from Arthur Gamgee in 1868 [2]. Dr. Gamgee wrote “My attention was directed to the peculiar action of nitrites on the blood-colouring-matter by observing that the blood of mice poisoned by exposure to an atmosphere impregnated with nitrite of amyl presented a chocolate colour.” As milk chocolate was not invented until 1876, one is led to conclude that the color referred to was plain or dark chocolate. Spectroscopic analysis of this solution revealed the decrease of the - and -bands of oxyHb and the increase in a faint band between the sodium and lithium emission lines – which corresponds to a wavelength of around 625 nm. Very similar results were obtained from the reaction of oxyHb with sodium nitrite. Despite some early discussions (see ref. [3]), the product of this reaction was firmly identified as methemoglobin (metHb). Although the stoichiometry of this reaction has been an area of significant debate, the careful examination by Kosaka et al. [4] gave a stoichiometry of 4 oxyHb:4 nitrite giving 4 metHb:4 nitrate:1 O2 . Since this time, although some reports indicated that nitrosylhemoglobin (HbNO) was formed during this reaction [5], it has been established that the only end product is metHb. However, at high nitrite concentrations and low pH, an additional product is formed from the reaction of nitrite with metHb (see below). The complexity of the reaction between nitrite and oxyHb arises from kinetic and allosteric considerations. Firstly, the kinetic profile of the oxidation reaction is sigmoidal, and secondly, organic phosphates (e.g., inositol hexaphosphate) inhibit the rate of oxidation. This latter issue is of interest as oxidation by most other oxidants (e.g., ferricyanide, hydrogen peroxide, etc.) is accelerated by IHP [6]. The time course exhibits a slow initial phase, often referred to as a “lag” phase that accelerates to a rapid rate of oxidation (Fig. 1). As this type of kinetic profile is reminiscent of autocatalysis, this latter phase is sometimes referred to as the autocatalytic phase. The work of Marshall and Marshall [3] in 1945 demonstrated that the length of the lag phase depended inversely on the concentration of nitrite and was acutely sensitive to pH, with 1 mM nitrite resulting in full oxidation within the dead-time of their experiments (20 s) at pH 5.2, but causing little if any oxidation for 1 hour at pH 9.2. Clues to the mechanism of oxyHb oxidation came in 1964 when Cohen et al. indirectly observed the formation of hydrogen peroxide by observing catalase compound I through the inhibitory action of aminotriazole [7]. In 1977, F. Lee Rodkey [8] observed that the length of the “lag” period of oxyHb oxidation by nitrite was inversely proportional to the concentration of metHb in the original oxyHb sample, and concluded that the acceleration in rate occurred as a consequence of metHb formation. In support of this, it was observed that the fast phase, but not the slow phase of the reaction was inhibited by cyanide. Specifically, the reaction was represented as occurring by an initial phase that

The Reaction between Nitrite and Hemoglobin

271

25

[oxyHb] (μM)

20

15

10

5

0 0

100

200

300

400

500

Time (s)

Fig. 1. The autocatalytic time course of the reaction between nitrite and oxyHb. OxyHb was incubated with nitrite (600 M) at 37 C. The decay of oxyHb was followed spectrophotometrically.

was proportional to nitrite and proton concentration and a second phase that involved a reaction between oxyHb and metHbNO− 2 , a form of ferric hemoglobin in which nitrite is bound to the heme iron. Although the mechanism of the latter step was not elucidated, this study introduced the idea that autocatalysis was related to the formation of metHb. Interestingly, Rodkey also observed the inhibition of the rapid phase of oxidation by sodium iodide, which he attributed to the ability of iodide to disrupt Hb tetramers into dimers. However, Wallace and Caughey [9] had shown that iodide is converted to iodine during oxidative reactions of hemoglobin, and this was used as additional evidence for the intermediacy of hydrogen peroxide. These authors proposed a mechanism for the oxidation of oxyHb by nitrite and phenolic compounds that involved two simultaneous + 3+ HbO2 2+ + NO− + H2 O2 + NO2 2 + 2H → Hb

HbO2 

2+

−

+ Cl → Hb · Cl

2+

+ O− 2

(1) (2)

reductions of bound oxygen by both the heme iron and the reducing agent (Equation 1). This mechanism is significantly different from the nucleophilic displacement of superoxide by, for example, chloride ion (Equation 2) that has been reported to be responsible for hemoglobin autoxidation. It is tempting to think that nitrite could potentially oxidize Hb by both these mechanisms, the former most likely predominates as high millimolar concentrations of nucleophilic anions are usually required to drive Equation 2 [10]. In 1982, Doyle et al. [11] proposed a mechanism to explain the autocatalytic kinetics of this reaction according to the following scheme. The initial reaction was proposed to be the two electron oxidation of bound oxygen as shown in Equation 1. HbO2 2+ + NO2 → Hb3+ O2 NOO− −

O2 NOO + H2 O2 →

− NO− 2 + 2O2

− + O− 2 + NO2 + 2H → H2 O2 + NO2

(3) (4) (5)

272

D.B. Kim-Shapiro et al.

In a chain of reactions, the nitrogen dioxide reacts with oxyHb to generate metHb and peroxynitrate, the latter of which reacts with hydrogen peroxide to form nitrite and superoxide (Equations 3 and 4). Once formed, the superoxide reacts with nitrite to form hydrogen peroxide and nitrogen dioxide (Equation 5). The net result of reactions 3–5 is that nitrogen dioxide catalyzes the dissociation of superoxide from oxyHb to form metHb, i.e., the so-called hemoglobin autoxidation reaction. Evidence for this scheme comes from experiments that show that the reaction can be accelerated by hydrogen peroxide and inhibited by catalase and by superoxide dismutase. It is stated that nitrate, the observed final product of nitrite oxidation derives from the hydrolysis of N2 O4 (after NO2 dimerization) or from the reaction between peroxynitrate and nitrite, presumably giving two molecules of nitrate. One major problem with this mechanism is that as nitrogen dioxide is the catalyst, it should stoichiometrically uncouple heme oxidation from nitrate formation and the 1:1 stoichiometry observed would therefore be a coincidental result of the downstream reactions of peroxynitrate and nitrogen dioxide. In addition, there is little experimental evidence for reactions 4 and 5. Kosaka et al. [12] invoked the mechanism shown in Equations 6–10, mainly on the basis of their observation that a protein radical is formed during the oxidation reaction. HX − HbO2 2+ + NO2− + 2H+ → HX − Hb3+ + NO2 + H2 O2 HX − Hb

3+

+ H2 O2 → •X − HbO

2+

+ H2 O + H

+

+ 2+ X − HbO2+ + NO− + NO2 2 + H → HX − HbO

•

HX − HbO

2+

HX

+ + NO− 2 +H

→ HX − Hb

3+

+ NO2 + H2 O

HbO2 2+ + NO2 → HX − Hb3+ + O2 + NO− 2 2NO2 + H2 O →

− NO− 2 + NO3

(6) (7) (8) (9) (10) (11)

In this scheme, HX represents an oxidizable globin amino acid residue. The initial oxidation step (Equation 6) is identical to that proposed by Doyle et al. [11], forming hydrogen peroxide and nitrogen dioxide. The hydrogen peroxide then reacts with metHb to form a peroxidase compound I–like species that consists of an oxoferryl heme and a protein radical (Equation 7). These products have been previously observed upon incubation of metHb with hydrogen peroxide [13], and the radical is likely located on a tyrosine residue, although electron density could be spread over a number of amino acids [14]. It is then proposed that the protein radical is reduced by nitrite to generate nitrogen dioxide (Equation 8), and subsequently the oxoferryl species is reduced by nitrite to again generate nitrogen dioxide and reforming metHb (Equation 9). The nitrogen dioxide formed in steps 6, 8, and 9 then oxidizes oxyHb to form metHb and regenerate nitrite (Equation 10). Finally any nitrogen dioxide that is not reduced by the reaction shown in Equation 10 will dimerize and hydrolyze to give nitrite and the final product nitrate (Equation 11). This mechanism introduces oxoferryl and protein radical intermediates into the mechanism and gives a clear mechanistic rationale for the inhibitory effects of catalase. However, the most obvious objection to this mechanism, as pointed out by Lissi [15], is that it is not autocatalytic. The flux through the reaction will be limited by the formation of hydrogen peroxide, which is only produced from the first reaction step.

The Reaction between Nitrite and Hemoglobin

273

In order to alleviate this problem, Lissi [15] suggested the inclusion of an additional step shown in reaction 12. HbO2 + NO2 → Hb3+ + H2 O2 + NO− 3

(12)

Recently, Goldstein et al. [16] have investigated the reaction of NO2 with oxymyoglobin and show that the reaction is rapid, with a rate constant of 45 × 107 M−1 s−1 and occurs via a multistep process but, importantly, without the intermediacy of hydrogen peroxide. This process involves the initial formation of a bound peroxynitrate (similar to equation 3, but without dissociation of the peroxynitrate), followed by hetrolysis of the O O bond to form nitrate and the ferryl radical form of the heme protein. This is effectively the sum of reactions 12 and 7. The importance of this is that the autocatalytic process is no longer rate limited by the relatively slow oxidation of metHb by hydrogen peroxide (Equation 7), allowing a more rapid acceleration of the rate during the time course of the reaction. Using myoglobin, Wade and Castro [17] demonstrated the interesting observation that solution oxygen plays a large role in the kinetics of oxyMb oxidation. If the experiment was performed under argon, but with enough oxygen to maintain oxyMb in the oxygenated state, the oxidation by nitrite was pseudo first-order with sharp isosbestic points indicating conversion of oxyMb to metMb. In this case, the reaction was limited by a second-order rate constant of 0.21 M−1 s−1 . However, if oxygen was present at atmospheric levels, the reaction proceeded via an autocatalytic mechanism, and isosbestic points were not sharp, indicating that at least one additional species was present. We have recently confirmed this data and conclude, using multiple regression analysis, that the additional species is spectrally identical to oxoferrylMb [18]. Wade and Castro proposed the involvement of ozone from the addition of an oxygen atom to oxygen by a putative ferric/nitrite intermediate as a way in which atmospheric oxygen could be activated. However, this highly speculative series of reactions is unlikely as the addition of nitrite to ferric myoglobin does not form MbNO as would be predicted from their scheme. Regardless of the mechanism, these observations suggest that the formation of the ferryl oxidation state is influenced by the presence of oxygen in solution and not by oxygen bound to the hemeprotein, an observation that is inconsistent with all other proposed mechanisms. Our recent studies suggest that with hemoglobin, it is the low levels of deoxyHb, present when solution oxygen is low, that inhibit the autocatalytic acceleration of the nitrite/oxyHb reaction [19]. The fact that NO2 is an intermediate in the nitrite-mediated oxidation of oxyHb raises the possibility that this reaction can damage proteins and lipids within the red cell via the oxidative chemistry of this reactive free radical. We have recently demonstrated, using immuno spin-trapping techniques, that the incubation of nitrite and oxyHb with bovine serum albumin (BSA) results in the formation of a BSA protein radical [18], indicating that the oxidative propensity of this reaction can be transferred to bystander molecules and may result in the oxidation of cellular components. It has been stated that nitrite-dependent oxyHb oxidation is peculiar in that it is facilitated by stabilization of the R state and slowed if the protein is T-state stabilized with agents such as IHP – the opposite of auto-oxidation and ferricyanide oxidation. There is an appreciable degree of confusion concerning factors that increase or decrease hemoglobin oxidation. The confusion stems in large part from lack of clarity regarding

274

D.B. Kim-Shapiro et al.

the phenomena under study. Most often, researchers are concerned about the redox stability of oxygenated hemoglobin, and the parameter of interest is the rate of autooxidation. The most striking factor affecting the auto-oxidation rate is the oxygen tension. Partially deoxygenated samples oxidize much more quickly than fully saturated samples under high oxygen tension. Although the mechanism underlying this phenomena has been debated, the data are clear [20]. Because of the greater rate of auto-oxidation of partially deoxygenated samples, the rate of auto-oxidation reactions that promote the oxidation of oxygenated Hb are typically facilitated by agents such as IHP and decreasing pH (the Bohr effect) that lower the percent occupancy of the active site. Underlying the oxidation process is the intrinsic redox potential of the heme iron. This is most easily studied under anaerobic conditions, where oxygen-linked processes are not part of the picture [10]. Unlike auto-oxidation events, measured in the presence of oxygen, CPA-Hb (the classic prototype of R-state Hb, generated by removal of the C-terminal salt bridges that allow formation of the T state) has an ease of oxidation close to that of myoglobin, its oxidation is noncooperatiave and it is relatively insensitive to pH or anionic effectors. Although the redox potentials of - and -chains are nonequivalent, with the -chain about 60 mV more reducing that the -chain, the reductions are not independent in the intact tetramer and show apparent cooperativity in an analogous manner to oxygen binding [21]. In this regard, the oxidation of oxyHb (and deoxyHb, see below) by nitrite follow more closely the ease of oxidation under anaerobic conditions, which is more closely linked to the redox potential of the heme iron. This strongly suggests that the nitrite-dependent oxidation is more closely linked to the intrinsic redox potential of the heme iron, rather than the ease of superoxide disassociation from oxyHb. While this was once thought to be a unique property of nitrite, a similar observation has been recently made for S-nitrosoglutathione, a nitrite thioester [22]. The fact that there is chain nonequivalence and cooperativity in the redox potential points to not only differential reactivity with nitrite as a function of ligand binding, but also to the fact that oxidation of one heme may affect the reactivity of other hemes in the same tetramer. The major conclusion of the above discussion is that despite the fact that the reaction between oxyHb and nitrite has been known for over 100 years, its mechanism still remains elusive. However, the fact that the transition from the slow phase to the autocatlytic phase of this reaction is inhibited by superoxide dismutase and catalase (as well as by electron-donating antioxidants such as ascorbate and glutathione) indicates that in the intraerythrocitic environment it is highly unlikely that the reaction will enter the autocatalytic phase. In addition, our recent studies suggest that low levels of deoxyHb inhibit the autocatalytic acceleration of the nitrite/oxyHb reaction, presumably by scavenging an intermediate [19]. Consequently, physiological submicromolar levels of nitrite may coexist with oxyHb and the conversion of nitrite to nitrate via this mechanism may be an extremely slow process. Although accurate rate constants have not been established, if only the slow phase is taken into account, the half-time of nitrite in the presence of 20 mM oxyHb will be measured in hours. Deoxygenation of the red cell will then allow the more facile reaction between deoxyHb and nitrite to occur, as discussed in the next section. The dominance of the reaction of nitrite with deoxygenated hemoglobin over that with oxygenated hemoglobin under physiological conditions (where the concentration of nitrite is low and one has mixtures of both oxygenated and deoxygenated Hb) has been pointed out and demonstrated recently [23,24].

The Reaction between Nitrite and Hemoglobin

275

2.2. Reaction of Nitrite with Deoxyhemoglobin In 1937, J. Brooks studied the reaction of nitrite with deoxygenated hemoglobin [25]. Using spectrophotometric methods he found that each nitrite molecule yielded one Fe(II)NO Hb and one Fe(III) Hb. He added sodium dithionite to the reaction products to yield a sample that is a pure mixture of deoxyHb and HbNO and then fit the resultant spectrum. He also studied this reaction using gas capacity of the Hb solutions and confirmed the same stoichiometry. In 1981, Michael Doyle and colleagues published an extensive study of the reaction of nitrite with deoxyhemoglobin [26]. They used absorption spectroscopy to study both the kinetics and stoichiometry of the reaction of hemoglobin with excess nitrite. The reaction products were quantified by fitting to HbNO and MetHb at a few wavelengths. No evidence for other species was observed. Addition of sodium dithionite yielded a mixture of deoxyHb and HbNO, and the percentage of HbNO determined in fitting that spectrum was the same as that calculated before sodium dithionite addition. The between-trial deviation in determining the percentage of HbNO was 2%. The kinetics of the reaction were examined by studying changes in absorbance as a function of time. The time dependence of the reactions was exponential and the observed rate was directly proportional to the nitrite concentration. Thus, Doyle and colleagues concluded that the reaction is second-order in Hb and nitrite. They reported a bimolecular rate constant of 2.69 M−1 s−1 at 25 C in pH 7.0 phosphate buffer. The kinetics of the reaction were found to depend linearly on the concentration of protons (The log of the observed rate was linearly dependent on the pH.). This increase in the observed rate constant as a function of time confirmed their hypothesis that nitrous acid (HONO) was involved. Doyle et al. proposed that the reaction of nitrite with deoxyHb begins with protonation of the nitrite to nitrous acid (possibly by an internal proton donor), followed by oxidation of the heme to form MetHb and release NO, which would then rapidly bind another ferrous heme, Equations 13–15 [26]. In addition, nitrite could directly oxidize the heme to form NO2 2− , which would then decompose into NO and water (Equation 16). K

H+ + NO− 2 ←→ HONO k0

Hb + HONO −→ MetHb + NO + OH− ka

Hb + NO −→ HbNO k

 2−  Hb + NO− 2 −→ MetHb + NO2

(13) (14) (15) (16)

The kinetics of the reaction are then given by −

dHb = k0 HbHONO dt

(17)

which leads to, using Equation 13, −

  dHb k0 H+   + k  = HbNO− 2 T Ka + H+  dt

(18)

276

D.B. Kim-Shapiro et al.

−3148 = where NO2− T = HONO + NO− 2 . So, for example, with pKa = 3148, Ka = 10 −4 + 71 × 10 M, so the [H ] in the denominator can be ignored. The fraction is approximately, with [H+ ] = 10−7 M, equal to 14 × 10−4 k0 . Doyle then gets k0 = 123 × 103 M−1 s−1 and k = 010 × M−1 s−1 . The fraction on the right-hand side is then (with [H+ ] = 10−7 M) = 1.7 M−1 s−1 , so that the term involving k can be mostly ignored. The total is 1.8 M−1 s−1 , a little less than his number of 2.69 M−1 s−1 . The reaction of NO with deoxyHb (Equation 15) is extremely rapid, ka ∼5 × 107 M−1 s−1 [27]. The presence of isosbestic points during the reactions indicated that the ratio of MetHb to HbNO made was constant in time. However, although the equations above predict that this ratio would be 1:1, that is not what Doyle et al. observed [26]. Rather, they found that the ratio of MetHb/HbNO was 0.72/0.28, which was independent of pH in the range from 6.0 to 8.0. The authors proposed that the reduced HbNO yield was due to oxidation of the heme via an (NO)2 intermediate. However, we find this explanation implausible since it requires the dimerization of NO to be competitive with iron nitrosylation of the heme. Since the latter reaction occurs at a rate on the order of 107 M−1 s−1 , and hemes are likely to always be in great excess to NO during the reaction of nitrite with Hb, we don’t see how (NO)2 could form. In 2003, Nagababu and colleagues published a paper using chemiluminescence and electron paramagnetic resonance spectroscopy (EPR) to study the reaction of deoxyHb and nitrite [28]. Like earlier studies, these authors found that the reaction of nitrite with Hb produces NO and hence HbNO. However, they also claimed that the majority of the NO-bound hemoglobin was in the form of Fe(III)NO Hb (where NO is bound to the ferric heme) rather than Fe(II)NO Hb (or HbNO – where the NO is bound to a ferrous heme). The authors claimed that as Fe(III)NO Hb is EPR-silent and chemiluminescence techniques potentially lack specificity, this species had been hitherto unnoticed. In fact, these authors claimed that 75% of NO bound to Hb in vivo is actually of the Fe(III)NO Hb form rather than Fe(II)NO Hb. They suggested that the NO bound as Fe(III)NO Hb is considerably more labile than Fe(II)NO Hb, and thus serves as a way to deliver NO formed from nitrite in the vasculature under hypoxic conditions. We find the idea that there is a stable Fe(III)NO Hb involved difficult to accept. Our skepticism over this idea stems from the facts that (i) the equilibrium binding constant of NO to Fe(II)NO Hb is at least 1 million times stronger than to Fe(III)NO Hb [29], (ii) the dissociation rate of NO from Fe(III)NO Hb is about 1/s [29], and (iii) there is usually about 100 times more free ferrous than ferric hemes in red blood cells (RBC). Thus, since the production of NO from nitrite occurs on the order of seconds to minutes, any Fe(III)NO Hb that forms as an intermediate will dissociate on the order of seconds and released NO will bind preferentially to free ferrous hemes, or oxidize oxyHb. The evidence [28] for an Fe(III)NO Hb intermediate is inconclusive. Specifically, detection of Fe(III)NO Hb by chemiluminesence was absolutely dependent on an increase in EPR-detected MetHb and Fe(II)NO Hb upon exposure to argon, and a change in the chemiluminescent signal when the sample is exposed to oxygen. Neither of these directly imply a Fe(III)NO Hb species. Fe(III)NO Hb has a distinct absorption spectrum and should be observable by deconvoluting kinetic spectra. Two separate labs have failed to see spectroscopic evidence for this intermediate [24,30]. In addition, we have recently employed a novel, sensitive, chemiluminescence-based assay to demonstrate that no Fe(III)NOHb is detectable in the nitrite/deoxyHb reaction [31].

The Reaction between Nitrite and Hemoglobin

277

The reaction of deoxyHb and nitrite has recently been revisited [23,30]. In these studies, the original stoichiometry proposed by Brooks (one nitrite molecule plus two deoxyHb molecules yield one ferric heme and one NO-bound heme) was confirmed. Results leading to other stoichiometries were attributed to either oxygen or Ferric heme contamination [23,30]. In addition, it was found that the kinetics of the reaction appeared to be of zero-order rather than first-order when nitrite was in excess to deoxyHb [23,30]. Closer inspection (Fig. 2) revealed that the kinetics has a sigmoidal character, and could be explained by R-state Hb reacting with nitrite faster than T-state Hb [23,30]. Specifically, during the anaerobic reaction of excess nitrite with deoxyhemoglobin, the reaction starts between nitrite and T-state hemoglobin tetramer; as the reaction proceeds, the formation of methemoglobin and iron-nitrosyl-hemoglobin changes the hemoglobin conformation to the R state. The lower redox potential of R-state hemoglobin increases the bimolecular rate constant for nitrite reduction and thus accelerates the reaction rate. We consider this a novel chemical mechanism of “allosteric autocatalysis.” Later in the course of the reaction, the deoxyhemes are depleted and the reaction rate decreases once again, completing the sigmoidal reaction rate process. Consistent with this model, Fig. 2(A) and (B) shows that myoglobin (which has no allosteric behavior) obeys first-order kinetics, whereas hemoglobin does not. Here, the deoxygenated heme in R3 (R-state Hb with three ligands bound or ferric hemes – so that one heme is still free) is the most reactive heme group. Similar kinetics are observed when measuring NO gas release using a chemiluminescent NO analyzer (Fig. 2C). In addition, since the ligation of CO to heme produces the R-state conformation of hemoglobin, deoxyhemoglobin with varying concentrations of CO reacted with nitrite should effect the rate of NO generation. As shown in Fig. 2(D), the time to peak NO production decreased as carboxyhemoglobin saturation increased from 0 to 75%, corresponding with increasing R-state character at the beginning of the reaction. Moreover, other Hb forms locked in a particular allosteric state (using inositol hexaphosphate or beta tetramers) behave as expected (Fig. 2E and F) with R-state Hb reacting quickly and T-state Hb reacting slowly. While increasing Hb oxygenation decreases the number of hemes available to react with nitrite (and thus slows the reaction down), increasing Hb oxygenation also produces more R3 (and thus speeds up the reaction (Fig. 3)). Figure 3 shows the initial rate of the reaction with increasing oxygen saturation. It is seen that the initial rate increases as the oxygen saturation increases up to 50% oxygen saturation after which increasing oxygen saturation leads to a decrease in the oxygen saturation of Hb. Importantly, this is also observed with intact red cells, and it was thus proposed that Hb acts as an allosterically controlled nitrite reductase whose activity is maximal at the Hb oxygen p50 [23]. Two aspects of the allosteric state of the protein may contribute to modulating the rate of the reaction: (i) conformation and (ii) the intrinsic redox potential of the heme iron. Evidence has been presented supporting a direct role of the redox potential (Table 1, [23]). The reaction rate is seen to correlate with heme redox potential with the lower E1/2 (such as in R-state Hb compared to T-state Hb) reacting fastest. These data do not rule out a role for the conformation of the protein as well. The position of the iron in the heme pocket, the entry rate of nitrite in to the pocket, and other factors may also contribute to the different rates observed for R-state and T-state Hb. The rate of the reaction of Hb and nitrite depends on the allosteric state of the protein as well as the pH [23,26]. Lower pH increases the rate of the reaction according to

278

D.B. Kim-Shapiro et al.

(A)

(B) 4

Mb Hb

0.02

50 μM Hb 10 mM nitrite

0.01

50 μM Mb 2.5 mM nitrite

3

Hb Mb

2 1

Hb R2 = 0.7903

0

Mb R2 = 0.9985

0.00

–1 0

100

200

0

300

100

200

Time (s)

(D) –d [deoxyheme]/d t (μM/s)

Hb

7

Mb

6 5

Mb

4

225 0.5 0.4

Hb

200

Mb

175

0.3 0.2 0.1 0.0 0

3

100 200 300

Time (s)

2

Time to Peak (s)

8

No Signal (mV)

300

Time (s)

(C)

Hb

1

150 125 100 75 50 25 0

0 0

100

200

0%

300

50%

75%

%HbCO

Time (s)

(F) 4

beta chains of HbA R2 = 0.9971

3

HbA + IHP at pH 6.4 R2 = 0.9996

–d [deoxyheme]/d t (μM/s)

(E)

Ln [deoxyHeme]

A630

0.03

Ln [deoxyHeme]

0.04

2

Hb

1

beta

0 –1

1.6 Beta Chains of HbA HbA + IHP at pH 6.4

1.4 1.2 1.0 0.8 0.6

beta

0.4

Hb

0.2 0.0

0

100

200

300

Time (s)

400

500

600

0

100

200

300

400

Time (s)

500

600

The Reaction between Nitrite and Hemoglobin

279

Fig. 2. Rate of nitrite reductase reaction and NO gas formation is under allosteric control. (A) Progress of the anaerobic reaction of Mb (50 M heme) and Hb (50 M heme) with nitrite (10 mM with Hb and 2.5 mM with Mb), monitored spectrophotometrically by metHb formation at 630 nm. (B) First-order fits for Mb and non-first-order behavior of tetrameric Hb (fits of natural log of deoxyheme concentration for the same reactions shown in A). (C) Simultaneous measurement of NO gas by chemiluminescence during the course of the reaction shown in (A). Inset shows the instantaneous rate of deoxyheme consumption over the course of the reaction, obtained from spectral deconvolution. (D) The time to peak NO production measured by chemiluminescence for the reaction of nitrite (10 mM) with Hb (50 M heme) with varying saturation (0–75%) of carbon monoxide; %HbCO, percentage of Hb that is saturated with carbon monoxide. (E) Reaction progress for -chains of HbA (locked in R-state tetramer; 35 M heme reacted with 0.5 mM nitrite at pH 7.0) and IHP-treated Hb (locked into T-state tetramer; 50 M heme reacted with 2.5 mM nitrite at pH 6.4) was monitored by the rate of deoxyheme consumption. (F) First-order fits for -chains of HbA at pH 7.0 and for IHP-treated Hb at pH 6.4 for conditions in (E), showing that the deviation from first-order requires an allosteric structural transition of the Hb tetramer (fits of natural log of deoxyheme concentration for the same reactions shown in E). Republished with permission of American Society for Clinical Investigation, from Journal of Clinical Investigation, Zhi Huang, Sruti Shiva, Daniel B. Kim-Shapiro, Rakesh P. Patel, Lorna A. Ringwood, Cythia E. Irby, Kris T. Huang, Chien Ho, Alan N. Schechter, Neil Hogg, and Mark T. Gladwin, volume 115, number 8, 2005; permission conveyed through Copyright Clearance Center, Inc.

(A)

(B) 11% Oxy

Δ – [Deoxyheme] (μM)

Deoxy

4 3

40% Oxy

2 1 0 0

10

20

30

Time (s)

40

50

Initial Reaction Rate (μM/s)

6% Oxy 18% Oxy

5

0.25 0.20 0.15 0.10 0.05 0.00 0

20

40

60

80

100

% Oxyheme

Fig. 3. Maximal rates for nitrite reduction to NO occurs around the Hb p50. (A) The negative of the change in deoxyheme concentration (obtained by spectral deconvolution) over time during the reaction of nitrite (10 mM) with partially oxygenated (0–40%) Hb (50 M total heme). (B) Initial rate of reaction for the conditions described above plotted as a function of initial oxygen saturation. Figure 3(A) and (B) republished with permission of American Society for Clinical Investigation, from Journal of Clinical Investigation, Zhi Huang, Sruti Shiva, Daniel B. Kim-Shapiro, Rakesh P. Patel, Lorna A. Ringwood, Cythia E. Irby, Kris T. Huang, Chien Ho, Alan N. Schechter, Neil Hogg, and Mark T. Gladwin, volume 115, number 8, pages 2099–2107, 2005; permission conveyed through Copyright Clearance Center, Inc.

280

D.B. Kim-Shapiro et al. Table 1. Initial rate of reaction corresponds to heme redox potential (E1/2 ) of various heme proteins (50 M heme, 2.5 mM nitrite) Mutant

Reaction Rate (M/s)

horseMb HbA-NEM HbA HbA+IHP

E1/2 (mV)

0.1884 0.0915 0.0161 0.0054

25 45 85 135

R2 = 1000, fitting to equation, y = 2927 × 10−5 + 04844e−x/2662 , where y is the reaction rate and x is E1/2 .

(A)

(B) Direct effect of [H+] Combined effects of pH Original Redox Bohr effect alone

Reaction Rate

3.5 3.0

Hb pH 7.6 × 12.59 Hb pH 6.5 Hb pH 7.6

1.2

–d [deoxyheme]/d t (μM/s)

4.0

2.5 2.0 1.5 1.0 0.5

1.0 0.8 0.6 0.4 0.2 0.0

0.0 0.0

0.2

0.4

0.6

0.8

Fraction Ligated or Ferric Hemes

1.0

0.0

0.2

0.4

0.6

0.8

1.0

Fraction Ligated or Ferric Hemes

Fig. 4. Effect of pH on the nitrite reductase reaction. (A) Modeling the rate of the reaction of nitrite with deoxyhemoglobin over a full range of hemoglobin ligand states from T state to R state at pH 7.6 compared with the rate of the reaction at pH 6.8 based only on the redox Bohr effect that stabilizes T state, compared with the rate at pH 6.8 based only on nitrite protonation, and the combined effects of redox Bohr and nitrite protonation (physiological effect) at pH 6.8. The rate was calculated as R3 kR + 4T0  + 3T1  kT , where capital R and T represent the quaternary states, the subscripts give the number of hemes that are ferric or ligand-bound (so R3 is R-state Hb with one deoxygenated heme), and kR and kT are the rates for the nitrite reaction of each quaternary state. The concentrations of each species (indicated by brackets) was calculated using an MWC model [27,34]. The value of c, the ratio of equilibrium binding constants for T (taken as 1/77 torr) and R states was set at 0.015. The R-state rate, kR , was set at 120 times kT . (B) Experimental data are consistent with a model of direct proton effect increasing the reaction rate and an opposing lesser dampening effect caused by the redox Bohr stabilizing T state. The instantaneous rates of the reaction of nitrite with deoxyhemoglobin is shown over a full range of hemoglobin ligand states at pH 7.6 compared with its normalization by multiplication factor 12.59 given by 1076−65 to account for the direct effect of increasing proton concentration on reaction rate as pH is lowered from 7.6 to 6.5. This is compared with the actual observed instantaneous rates of the reaction at pH 6.5 (reflecting both proton effect and redox Bohr effect).

The Reaction between Nitrite and Hemoglobin

281

Equations 14–18 (the Bohr effect [32,33]), while it also decreases the rate of the reaction by stabilizing the T state. Figure 4 shows that of these competing factors, the direct effect of the pH (as governed by the concentration of HONO) dominates. This is shown using kinetic modeling of the reaction and using experimental data. The modeling results also confirm that one may obtain a maximal rate of nitrite reduction at the p50 by assuming that R-state Hb reacts faster than T-state Hb. From a physiological standpoint, the intravascular infusion of low concentrations of nitrite (2 M) into the human forearm artery increases forearm blood flow. These data are discussed in more detail later in this chapter. During the infusion of nitrite, HbNO and to a lesser extent S-nitrosohemoglobin (SNO Hb) form during one half-circulatory time; the formation of both of these products was found to be inversely proportional to oxygen tension [1]. In addition, the kinetics of HbNO formation observed in vivo were consistent with those measured separately in whole blood and hemoglobin. The rate of the reaction in whole blood was found to be slightly slower than that for pure hemoglobin, indicating that the reaction is rate limited by nitrite uptake of the red blood cells. That the reaction of deoxyHb with nitrite is bimolecular was confirmed by obtaining the same observed rate constants when the reaction was performed with excess nitrite or excess deoxyHb [1].

2.3. Reactions of Nitrite with Methemoglobin High concentrations of nitrite, at low pH, will react with metmyoglobin to form a green product referred to as nitrimyoglobin. For example, this product can be generated in high yield from the incubation of 53 mM nitrite with 530 M metmyoglobin at pH 5.5 [35]. Analysis of this compound reveals a nitro substitution of the heme vinyl group, which causes little change in the ligand-binding properties of the heme protein [35]. This reaction appears to be largely responsible for the “greening” of cured meats, and can also be observed with hemoglobin. Nitrite has also been reported to bind to metHb to form a reversible complex with a dissociation constant of between 1 and 3 mM [8]. Recently we have discovered that nitrite bound metHb has a radical (FeII -NO•2 ) character that reacts with NO to form N2 O3 providing a potential mechanism for export of NO activity from the red blood cell [31].

2.4. Reactions of Nitrite During Deoxygenation/Reoxygenation Most, if not all, of the experimental data concerning the reaction of nitrite with hemoglobin has been obtained in either the fully oxygenated or the fully deoxygenated state. While this facilitates understanding, it does not reflect the situation in vivo, where the red cell is constantly traversing through areas of varying oxygen tension. In the oxygenated (arterial) blood, the major reaction will be with oxyHb. As we have seen, this is a complex autocatalytic reaction that likely never reaches the autocatalytic phase in the presence of intracellular reducing agents and antioxidant enzymes. Upon deoxygenation, the dissociation of oxygen from hemoglobin exposes free ferrous-binding sites that can mediate different chemistry, giving rise to HbNO via the intermediacy of NO or some

282

D.B. Kim-Shapiro et al.

NO-generating intermediate. We observed that in partially deoxygenated Hb, as compared to fully deoxyHb, the ratio of metHb:HbNO increases, consistent with the liberated NO reacting with oxyHb to form metHb in lieu of HbNO [23]. Recently, we have shown that under partially oxygenated conditions, the reaction of nitrite with deoxyHb runs simultaneously with that with oxyHb [19]. Interestingly, when deoxyHb is oxygenated in the presence of nitrite, intermediates from the oxyHb/nitrite reaction oxidize the heme and release NO from HbNO, thereby constituting a mechanism for fast oxidative denitrosylation. Although a full understanding of these mechanisms is wanting, the chemical data give a rationale for the oxygen-dependent control of nitrite-mediated vasodilation [1].

3. PHYSIOLOGICAL CONSEQUENCES OF THE NITRITE/HEMOGLOBIN REACTION 3.1. Role of the Nitrite-Deoxyhemoglobin Reaction in Vasodilation As outlined above, understanding the interactions of Hb and NO or nitrite poses an intriguing problem to biochemists, biophysicists, vascular biologists, physiologists, and hematologists alike. The emergence of the concept that these reactions are playing roles in as fundamental a process as blood flow underscores the importance of understanding reaction mechanisms. We now shift the focus of this chapter to discuss our current understanding of how Hb and RBCs modulate blood flow through interactions with NO and nitrite. Firstly, it is important to stress the word “modulate” in the previous sentence. Since the discovery that NO is produced in the vascular endothelium and serves multiple functions, including the regulation of 25% of basal blood flow in humans, the general notion has been that Hb and by extrapolation the RBC, inhibit NO bioactivity. This view is supported by the rapid reactions of NO with oxyHb and deoxyHb and the significantly (about 6 orders of magnitude) higher concentrations of Hb relative to NO. However, work from a variety of groups has all but dismissed the notion that Hb is a bottomless sink for NO bioactivity in vivo [36–38], and show that it is an active player in processes that serve to regulate vascular homeostasis. The potential for nitrite to reduce the oxoferryl (compounds I and II) species of peroxidases forming NO2 has been described in the context of inflammation and may result in oxidative modifications of biological molecules. For the most part, however, nitrite has been regarded as a relatively inert end product of NO metabolism. However, this view is changing rapidly as a direct consequence of the developments of methodologies that allow more accurate detection of nitrite in biological samples. These methodologies have revealed that plasma nitrite is present at a concentration of 0.15–1 M, and reflects endothelial NOS activity more closely than that of nitrate levels [39]. While it has been long appreciated that nitrite has vasodilator activity at high concentrations in aortic ring bioassay systems [40,41], its vasoactivity at physiological concentrations under hypoxia has only recently been appreciated [1]. On the basis of human studies revealing arterial-to-venous gradients in nitrite across the human circulation, increased nitrite consumption during exercise [42], and enhanced rates of nitrite consumption by deoxygenated erythrocytes [43–45], we speculated that nitrite might be activated in vivo and modulate an “endocrine” or blood-transported bioactivity [46–48]. Indeed, the potency of nitrite increases in aortic ring preparations at lower pH values

The Reaction between Nitrite and Hemoglobin

P = 0.0006

Forearm blood flow (mL /min/100 mL tissue)

Forearm blood flow (mL /min/100 mL tissue)

6

(C)

5 4 3 2

22.5

3

20.0 17.5 15.0 5.0

* *

2.5 0.0

1 Baseline

Nitrite

e e A A A A lin trit M M M M se Ni NM NM NM NM a L- L- /L- /LB & x X te E & E i r t te Ni tri Ni

P = 0.05

4

*

25.0

(D)

*

2

1

*

0 e e e lin trit itrit arm n ite se Ni a & B os ise pp rc O e Ex

NO-modified hemoglobin (% NO per heme subunit)

(B)

Nitrite (μM)

(A)

283

0.0005 0.0004 0.0003 P = 0.19 0.0002 0.0001 0.0000 e

Ba

s

in el

NO NO NO NO S- S- Hb- Hbte ti ri line trite N se Ni Ba

Fig. 5. Effects of infusion of low-dose sodium nitrite in bicarbonate-buffered normal saline into the brachial arteries of 10 healthy subjects at baseline and during exercise, without and with inhibition of NO synthesis. (A) Forearm blood flow at baseline and following a 5-minute infusion of NaNO2 (0.36 mol/ml in 0.9% saline, infused at 1 ml/min). (B) Forearm blood flow with and without low-dose nitrite infusion at baseline and during l-NMMA infusion with and without exercise stress. (C) Venous levels of nitrite from the forearm circulation at the time of blood flow measurements. (D) Venous levels of S-nitrosohemoglobin (S NO) and iron-nitrosyl-hemoglobin (Hb NO) at baseline and following nitrite infusion during exercise stress. Reproduced from Nature Medicine, Cosby et al., Volume 9, number 12, pages 1498–1505, 2005.

[49]. Infusion of nitrite into the forearm circulation of 28 normal human volunteers at pharmacologic (200 microM levels in forearm) and near-physiological (0.9–2.5 microM) concentrations resulted in a robust vasodilation (170% and 20% increase in blood flow, respectively) [1]. The increase in blood flow was associated with the formation of NO-modified hemoglobin across the forearm circulation within one half-circulatory time (i.e., from artery to vein during the infusion) (Fig. 5) [1]. A strong inverse correlation between iron-nitrosylation and hemoglobin oxygen saturation in vitro and in vivo suggested a reaction of nitrite with deoxyhemoglobin. Interestingly, SNO Hb was also formed, albeit to a lesser extent than HbNO. Additional in vitro studies of nitrite with deoxygenated hemoglobin solutions and erythrocytes supported a novel model that the nitrite reductase activity of deoxyhemoglobin and deoxygenated erythrocytes produces vasodilation along the physiological oxygen gradient. Additional mechanistic insights were gleaned utilizing modified vessel bioassay chambers that allowed simultaneous monitoring of oxygen concentrations and tension of isolated vessel segments [1,49]. Using such an approach allowed the construction of vessel tension versus oxygen concentration relationship curves. Under control conditions, vessels spontaneously relax upon reaching 10–15 mm Hg oxygen. This “threshold” for relaxation was dramatically shifted in the presence of RBC and low (0.5–2 M) nitrite to approximately 30 mm Hg for human RBC and 40 mm Hg for rat RBC. Interestingly, these oxygen tensions correlate directly with the reported p50s for human and rat blood, respectively, consistent with a maximal nitrite reduction at Y = 05 (Fig. 6) [1,49]. These experiments were also performed using the classical pharmacology approach of adding increasing concentrations of nitrite to vessels in the presence of RBC or cell-free

D.B. Kim-Shapiro et al. P = 0.004

45

P = 0.03

40

30 25 20 15 10

P50 ~ 28 mmHg

35

P50 ~ 40 mmHg

pO2 at which relaxation starts (mmHg)

284

5 0 Control

NO2–

rat RBC’s

rat RBC human human +NO2– RBC RBC + NO2–

Fig. 6. The oxygen tensions at which nitrite and red cells initiated dilatation of isolated rat thoracic aorta were determined, and show that, alone, neither red cells (0.3% Hct) nor nitrite (2 M) are able to stimulate vasodilatation. However, in combination, vasodilatation is initiated at significantly higher oxygen tensions and correlate with the respective hemoglobin p50s, as indicated by use of rat and human red cells. Note that under control conditions, vessels spontaneously begin to dilate at ∼20 mmHg oxygen due to oxygen substrate limitation and compromised ATP production. These data were obtained under conditions where both endogenous NO production and ATP-dependent mechanisms of dilation are inhibited. This figure is a composite of data published previously in (1) Nature Medicine, Cosby et al., Volume 9, number 12, pages 1498–1505, 2005 and (2) in Blood. Jack H. Crawford, T. Scott Isbell, Zhi Huang, Sruti Shiva, Balu K. Chacko, Alan N. Schechter, Victor M. Darley-Usmar, Jeffrey D. Kerby, John D. Lang Jr, David Kraus, Chien Ho, Mark T. Gladwin MD, Rakesh P. Patel. Hypoxia, Red Blood Cells and Nitrite Regulate NO-dependent Hypoxic Vasodilatation, 15, 566–574, 2006, © the American Society of Hematology.

Hb at different oxygen concentrations [49]. With RBC, nitrite stimulated vasodilation that became more efficient as the oxygen tension was decreased. Importantly, dilation was still observed at oxygen tensions above the RBC p50 [49], suggesting that oxygen regulates nitrite-RBC interactions in a graded manner. These data are also consistent with: (i) the earlier described biochemical studies that demonstrate a preferential reaction of nitrite with deoxyHb relative to oxyHb, (ii) the surprising stability of low concentrations of nitrite in oxygenated red cells where the reductive environment inhibits autocatalysis, and (iii) the maximal nitrite reductase rate observed around the hemoglobin p50, an effect determined by a balance between deoxyheme availability for nitrite binding and the increasing bimolecular rate constant of oxygenated R-state tetramer. The current model being proposed is that under high oxygen conditions, red cell nitrite levels are maintained at a steady state level of 0.3–1 M through limited oxidation by oxyHb to nitrate and perhaps other metabolic processes. However, upon RBC sensing local decreases in oxygen tensions (which is manifested by formation of deoxyHb), nitrite is converted into a vasodilatory stimulus, thereby increasing blood flow. Indeed, this appears to be a general function that can be attributed to the RBC. RBCs have been shown to release ATP upon deoxygenation and/or mechanical deformation. ATP

The Reaction between Nitrite and Hemoglobin

285

then activates purinergic signaling to elicit dilation in specific vascular beds [49,50]. Whether these serve as overlapping or redundant mechanisms through which RBCs increase blood flow to hypoxic regions is not known, but underscores the importance of the RBC as an oxygen sensor. An immediate challenge to this area is to understand how the process of transforming nitrite into a vasodilator is regulated. This can include nitrite entry into the RBC, which has been shown to be accelerated upon deoxygenation with carp RBC [44] and human RBC (Gladwin et al., preliminary observations). Interestingly, RBC nitrite has been found to be in higher concentrations than in the plasma, but a mechanism for this is concentration gradient is not yet known [51]. As discussed above, the direct reaction between Hb and nitrite is also complex and likely to regulate biological responses. Further insights into this process can be obtained from investigating the vasodilatory mechanisms of nitrite in the presence of cell-free Hb. Under conditions where Hb is principally in the oxygenated state (∼90%), Hb does not stimulate nitrite-dependent dilation. However, if the oxygen affinity of Hb is decreased with the allosteric modulator IHP, so that deoxyHb is the primary state (∼90%), nitrite-dependent vasodilation is enhanced by approximately 3–4 orders of magnitude [1]. Moreover, this effect is directly proportional to the amount of deoxyHb present, strongly suggesting that direct reactions of nitrite with deoxyHb are central in mediating the relaxation response. For the reader who is familiar with the recent concepts, the data discussed above are similar to the concepts proposed for S-nitrosohemoglobin [24,38,52–55]. This is an important issue that impacts upon mechanisms and potential therapeutic strategies for affecting blood flow. Importantly then, using similar experimental approaches as those discussed above, recombinant cell-free Hb in which the 93cys residue was replaced with an alanine residue also stimulated nitrite-dependent vasodilation that was proportional to the content of deoxyheme (peak effect at Y = 05) [49]. These data rule out any possible role for S-nitrosohemoglobin in mediating the nitrite-dependent vasodilation response and are consistent with studies demonstrating that this S-nitrosothiol does not play a role in mediating blood flow under physiological conditions. Moreover, given that low levels of nitrite (0.9–2 M) has been shown to stimulate vasodilation in the human circulation and levels of 0.2 microM in vitro [49], it is possible that effects previously attributed to S-nitrosohemoglobin may be explained by the low levels of nitrite that contaminate many NO-containing solutions and biological buffers (see ref. [56]) in addition to enhanced vessel responsiveness during hypoxia. Interestingly, an important distinction between the vasoactivity of native and 93cys-ala Hb was observed. At any given oxygen tension, the mutant Hb stimulated nitrite mediated dilation to a greater extent than native Hb. In other words, whereas the 93cys residue is not a direct participant in converting nitrite to a vasoactive species, it appears to regulate how nitrite reacts with deoxyheme. Consistent with this concept, alkylation of the 93cys modulates the reaction of nitrite with either deoxyHb or oxyHb. Furthermore, using a variety of mutants in which this residue has been replaced with different amino acids, the rate of nitrite-deoxyHb reactions is increased [23]. This occurs as a consequence of the effect of cys93 alkylation on decreasing the heme redox potential (Table 1). Collectively, these data suggest a model in which the 93cys is not absolutely required, and SNO Hb is not a necessary

286

D.B. Kim-Shapiro et al.

intermediate, but this residue regulates deoxyheme-mediated conversion of nitrite into vasodilator. A number of factors support the notion that nitrite is a stable biological storage form of NO: (i) The relative stability of nitrite in the presence of oxygenated red blood cells and tissue with an increased rate of uptake and utilization by deoxygenated red blood cells is ideal for selective conversion to NO under deoxygenated conditions [43–45,57]. (ii) Nitrite is relatively stable under cellular reducing conditions, unlike S-nitrosothiols [58]. (iii) The rate of reaction of nitrite with hemoglobin is 10,000 times slower than that of authentic NO [26]. (iv) Nitrite reactions with deoxyhemoglobin and deoxymyoglobin ultimately generate NO and vasodilation, allowing for oxygen-linked and pH-dependent bioactivation [1].

3.2. Therapeutic Exploitation of Nitrite-Hemoglobin Chemistry Current studies are evaluating the therapeutic application of nitrite in a number of disease states. A selective reaction of nitrite with deoxyhemoglobin to form NO would be ideal for the treatment of hypoxic conditions or hemolytic conditions such as sickle cell disease and cardiopulmonary bypass, disease states characterized by peripheral NO consumption by cell-free plasma hemoglobin. Inhaled nebulized nitrite has been shown to effectively reduce pulmonary hypertension in a sheep model, and thus shows promise as a new treatment for neonatal pulmonary hypertension [59]. Using a primate model, nitrite infusions were shown to reduce cerebral vasospasm after subarachnoid aneurismal hemorrhage, a vasoconstrictive complication that occurs secondary to blood in the cerebral spinal fluid [60]. Administration of nitrite has also revealed particular promise for the amelioration of ischemia-reperfusion injury, targeting NO to tissue under the greatest anoxic stress [61,62]. We expect ongoing preclinical and clinical work to help define the therapeutic promise of the nitrite-hemoglobin reaction.

4. SUMMARY AND CONCLUSIONS Figure 7 summarizes the essential features of the nitrite/hemoglobin hypothesis. The interaction of nitrite with deoxyHb and not oxyHb generates a diffusible vasodilator with the properties of nitric oxide. The barrier to diffusion that exists at the red-cell membrane will limit the ability of red cells to destroy NO generated in the extracellular space and allow diffusion of red-cell generated NO to the smooth muscle tissue. There are many unanswered questions necessary for a full understanding of the mechanism and role of nitrite-hemoglobin biochemistry in the regulation of blood flow and vascular NO homeostasis. The complex mechanisms of reaction are not fully understood, even in simple chemical systems devoid of the complexity of oxygen gradients, allosteric effectors, and additional cellular constituents. Until these mechanisms have been established, several puzzles will remain – the major one being that if NO is made by hemoglobin, how does it escape the huge NO-scavenging potential of the interior of the red cell? If it is not NO that is made but another diffusible intermediate species [63], then what is this − intermediate species (i.e., NO2 , ONOO− , HNO− 2 , NO2 , N2 O3 )? we have recently found

The Reaction between Nitrite and Hemoglobin

287

β-cys93 S NO NO+ T

Diffusion barrier

FeII

GSH?

NO2– NO NO–

Arginine

Nitric Oxide Citrulline Synthase NO –

sGC

Fig. 7. The role of nitrite in the physiological and pathological regulation of blood flow. A barrier to diffusion exists that prevents the red cell from rapidly destroying endothelial derived nitrite oxide. This barrier also has the potential to allow NO, generated within the lumen of blood vessel enough time to diffuse to the smooth muscle layer. There is a growing body of evidence to suggest that the red cell can participate in the mechanism of hypoxic vasodilation though an NO-dependent mechanism. The original mechanism for this effect involved the formation of S-nitrosohemoglobin and oxygen-dependent release of NO from the hemoglobin thiol. We present here an alternative mechanism involving the activation of nitrite by deoxygenated hemoglobin. While the details of this mechanism are still under investigation, the oxygen dependency of the NO-generating ability of hemoglobin is dictated not by conformational changes of hemoglobin, but by the differential reactivity of hemoglobin in the oxygenated and deoxygenated state. Source: Reprinted from Free Radicals In Biology And Medicine, Vol 36, Mark T. Gladwin, Jack H. Crawford, and Rakesh P. Patel, The Biochemistry of Nitric Oxide, Nitrite, And Hemoglobin: Role in Blood Flow Regulation, (707–717) Copyright (2004), with permission from Elsevier.

evidence that the intermediate is N2 O3 [31]. However, the physiological in vitro and in vivo studies all point to the fact that this interaction generates a diffusible vasodilator that may have a crucial role in the physiology, pathology, and therapy of the vascular system and responses to regional hypoxia and ischemia.

ACKNOWLEDGMENTS This work was supported by NIH grants HL58091 (DK-S), GM55792 (NH), HL70146 (RPP), and the intramural research divisions of the NHLBI, NIH. DBK-S is grateful for support from K02 078076.

288

D.B. Kim-Shapiro et al.

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44]

Cosby, K., Partovi, K.S., Crawford, J.H. et al. (2003) Nat. Med., 9, 1498. Gamgee, A. (1868) Philos. Trans. R. Soc. Lond., 158, 589. Marshall, W. and Marshall, C.R. (1945) J. Biol. Chem., 158, 187. Kosaka, H., Imaizumi, K., Imai, K., and Tyuma, I. (1979) Biochim. Biophys. Acta, 581, 184. Haldane, J., Makgill, R.H., and Mavrogordano, A.E. (1897) J. Physiol., 21, 160. Doyle, M.P., Lepoire, D.M., and Pickering, R.A. (1981) J. Biol. Chem., 256, 2399. Cohen, G., Martinez, M., and Hochstein, P. (1964) Biochemistry-US, 3, 901. Rodkey, F.L. (1976) Clin. Chem., 22, 1986. Wallace, W.J. and Caughey, W.S. (1975) Biochem. Biophys. Res. Commun., 62, 561. Taboy, C.H., Faulkner, K.M., Kraiter, D. et al. (2000) J. Biol. Chem., 275, 39048. Doyle, M.P., Pickering, R.A., Dykstra, R.L. et al. (1982) Biochem. Biophys. Res. Commun., 105, 127. Kosaka, H., Imaizumi, K., and Tyuma, I. (1982) Biochim. Biophys. Acta, 702, 237. Yusa, K. and Shikama, K. (1987) Biochemistry-US, 26, 6684. Gunther, M.R., Tschirret-Guth, R.A., Witkowska, H.E. et al. (1998) Biochem. J., 330, 1293. Lissi, E. (1998) Free Radic. Biol. Med., 24, 1535. Goldstein, S., Merenyi, G., and Samuni, A. (2004) J. Am. Chem. Soc., 126, 15694. Wade, R.S. and Castro, C.E. (1996) Chem. Res. Toxicol., 9, 1382. Keszler, A., Mason, R.P., and Hogg, N. (2006) Free Radic. Biol. Med., 40, 507. Grubina, R., Huang, Z., Shiva, S. et al. (2007) J. Biol. Chem., 282, 12916. Brantley, R.E., Smerdon, S.J., Wilkinson, A.J. et al. (1993) J. Biol. Chem., 268, 6995. Antonini, E. and Brunori, M. (1971) Hemoglobin and Myoglobin in their Reactions with Ligands, North Holland Publishing Company, Amesterdam. Spencer, N.Y., Zeng, H., Patel, R.P., and Hogg, N. (2000) J. Biol. Chem., 275, 36562. Huang, Z., Shiva, S., Kim-Shapiro, D.B. et al. (2005) J. Clin. Invest., 115, 2099. Luchsinger, B.P., Rich, E.N., Yan, Y. et al. (2005) J. Inorg. Biochem., 99, 912. Brooks, J. (1937) Proc. Royal Soc. London — Series B, Biol. Sci., 123, 368. Doyle, M.P., Pickering, R.A., Deweert, T.M. et al. (1981) J. Biol. Chem., 256, 12393. Kim-Shapiro, D.B. (2004) Free Radic. Biol., Med. 36, 402. Nagababu, E., Ramasamy, S., Abernethy, D.R., and Rifkind, J.M. (2003) J. Biol. Chem., 278, 46349. Cooper, C.E. (1999) Biochim. Biophys. Acta-Bioenerg., 1411, 290. Huang, K.T., Keszler, A., Patel, N. et al. (2005) J. Biol. Chem., 280, 31126. Basu, S., Grubina, R., Huang, J., et al. (2007) Nat. Chem. Biol., In Press. Bohr, C., Hasselbalch, K., and Krogh, A. (1904) Skand. Arch. Physiol., 16, 402–412. Perutz, M.F. (1970) Nature, 228, 726–739. Monod, J., Wyman, J., and Changeux, J.-P. (1965) J. Mol. Biol., 12, 88. Bondoc, L.L. and Timkovich, R. (1989) J. Biol. Chem., 264, 6134. Liu, X.P., Miller, M.J.S., Joshi, M.S. et al. (1998) J. Biol. Chem., 273, 18709. Liao, J.C., Hein, T.W., Vaughn, M.W. et al. (1999) Proc. Natl. Acad. Sci. USA, 96, 8757. Jia, L., Bonaventura, C., Bonaventura, J., and Stamler, J.S. (1996) Nature, 380, 221. Kleinbongard, P., Dejam, A., Laver, T., et al. (2003) Free Radic. Biol. Med., 35, 790. Furchgott, R.F. and Bhadrakom, S. (1953) J. Pharmacol. Exp. Ther., 108, 129. Ignarro, L.J., Lippton, H., Edwards, J.C. et al. (1981) J. Pharmacol. Exp. Ther., 218, 739. Gladwin, M.T., Shelhamer, J.H., Schechter, A.N. et al. (2000) Proc. Natl. Acad. Sci. USA, 97, 11482. Jensen, F.B. (2003) Comp. Biochem. Physiol. A-Mol. Integr. Physiol., 135, 9. Jensen, F.B., Koldkjaer, P., and Bach, A. (2000) J. Comp. Physiol. B-Biochem. Syst. Environ. Physiol., 170, 489.

The Reaction between Nitrite and Hemoglobin

289

[45] Jensen, F.B., Andersen, N.A., and Heisler, N. (1987) J. Comp. Physiol. B-Biochem. Syst. Environ. Physiol., 157, 533. [46] Gladwin, M.T., Lancaster, J.R., Freeman, B.A., and Schechter, A.N. (2003) Nat. Med., 9, 496. [47] Schechter, A.N. and Gladwin, M.T. (2003) New Engl. J. Med., 348, 1483. [48] Modin, A., Bjorne, H., Herulf, M. et al. (2001) Acta Physiologica. Scandinavica., 171, 9. [49] Crawford, J.H., Isbell, T.S., Huang, Z. et al. (2006) Blood, 15, 566. [50] Sprague, R.S., Olearczyk, J.J., Spence, D.M. et al. (2003) Am. J. Physiol.-Heart Circul. Physiol., 285, H693. [51] Dejam, A., Hunter, C.J., Pelletier, M.M. et al. (2005) Blood, 106, 734. [52] Singel, D.J. and Stamler, J.S. (2005) Annu. Rev. Physiol., 67, 99. [53] Gow, A.J. and Stamler, J.S. (1998) Nature, 391, 169. [54] McMahon, T.J., Moon, R.E., Luschinger, B.P. et al. (2002) Nat. Med., 8, 711. [55] Luchsinger, B.P., Rich, E.N., Gow, A.J. et al. (2003) Proc. Natl. Acad. Sci. USA, 100, 461. [56] Xu, X.L., Cho, M., Spencer, N.Y. et al. (2003) Proc. Natl. Acad. Sci. USA, 100, 11303. [57] Bryan, N.S., Rassaf, T., Maloney, R.E. et al. (2004) Proc. Natl. Acad. Sci. USA, 101, 4308. [58] Gladwin, M.T., Wang, X.D., Reiter, C.D. et al. (2002) J. Biol. Chem., 277, 27818. [59] Hunter, C.J., Dejam, A., Blood, A.B. et al. (2004) Nat. Med., 10, 1122. [60] Pluta, R.M., Dejam, A., Grimes, G. et al. (2005) Jama-J. Am. Med. Assoc., 293, 1477. [61] Duranski, M.R., Greer, J.J.M., Dejam, A. et al. (2005) J. Clin. Invest., 115, 1232. [62] Webb, A., Bond, R., McLean, P. et al. (2004) PNAS., 101, 13683. [63] Jeffers, A., Xu, X., Huang, K.T. et al. (2005) Comp. Biochem. Physiol. A-Mol. Integr. Physiol., 142, 130.

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 12

Nitric Oxide Dioxygenase: An Ancient Enzymic Function of Hemoglobin Paul R. Gardnerab and Anne M. Gardnerb a b

Department of Chemistry, University of Dayton, Dayton, OH 45469 Cincinnati Children’s Hospital Research Foundation, Cincinnati, OH 45229 USA

Summary Structurally diverse members of the ancient hemoglobin (Hb) superfamily show NO dioxygenase (NOD) activity (EC 1.14.12.17), suggesting a common and primal function for Hbs and myoglobins (Mbs). NO reacts rapidly with the oxy complexes of Hbs, Mbs, flavoHb, truncated Hb, legume Hb, and neuroglobin, generating stoichiometric ferric heme and nitrate, and incorporating both O-atoms of the bound O2 . Associated flavincontaining reductases, cytochrome b5 , or ascorbate reduce ferric heme to allow O2 rebinding and catalytic turnover. Hb functions as a true enzyme by controlling O2 binding and electrochemistry, by guiding NO diffusion and the dioxygenation reaction, and by shielding reactive FeIII− O2 •, FeIII OONO, FeIV O, and NO2 intermediates from solvent water and biomolecules. The activity protects against NO toxicity and modulates NO signaling in a variety of life forms. NODs and NOD inhibitors are finding applications in medicine, agriculture, and biotechnology. Keywords: Hemoglobin; Nitric oxide; Dioxygen; Dioxygenase; Heme

1. Hb FUNCTIONS 1.1. Unraveling Hb Functions and Evolution Discoveries of new members of the Hb superfamily in diverse and ancient life forms [1–16] continue to stimulate investigations of Hb functions and evolution. For the majority of Hbs and Mbs, an O2 storage and transport function has been doubtful. Most notably, Hbs and Mbs are either expressed at levels too low to support O2 transport, bind O2 with extremely high affinity precluding O2 transfer, or are found in unicellular organisms in which O2 is transported by simple diffusion. Biochemical and genetic evidence [15–17] is mounting for a common NO dioxygenase (NOD) function that may have guided the evolution of Hb.

Nitric Oxide Dioxygenase

291

1.2. Early Investigations of the Reactions of NO and Hb David Keilin was first to envision oxidative enzymic functions for low-abundance Hbs [18], following his seminal discovery of Hbs in primitive unicellular life forms [19]. As early as 1937, Keilin and others had described the nitrosylation of the ferrous red blood cell Hb (Equation 1) and pondered upon the oxidative reaction of NO with Hb [20,21]. In 1981, Doyle and Hoekstra demonstrated the rapid reaction between NO and HbFeII O2 and MbFeII O2 to form nitrate and proposed a two-step mechanism (Equations 2a and 2b) [22]. Large rate constants for Hb oxidation were estimated that could account for the oxidation of circulating red blood cell Hb by inhaled NO, a common atmospheric pollutant [23]. At the time, investigators could not appreciate the connection between these seemingly obscure reactions and the biological function and evolution of Hb. HbFeII + NO → HbFeII NO HbFeII O2 + NO → HbFeIII +−OONO −

OONO →

NO− 3

(1) (2a) (2b)

1.3. Physiology of the Reactions of NO with Hb and Mb Greater understanding of the physiology of the reactions of NO with Hb quickly followed the discoveries of functions for NO as a vascular relaxing factor and as an immune cell–derived antibiotic and antitumor agent [24–29]. Tissue Hb and Mb were soon viewed as abundant sinks for NO that could inhibit signaling [30–36]. The reaction (Equation 2a) was also thought to impair Mb and Hb functions in O2 transport and storage [22] and potentially release toxic peroxynitrite (− OONO) from the hydrophobic pocket (Equation 2a) [37,38].

1.4. Hb Functions as a NO Dioxygenase (NOD) Realization of the enzymatic nature of the NO + HbO2 reaction (Equation 3) provided greater insights into the function and evolution of Hbs. By 1998, investigations of the resistance of Escherichia coli to NO poisoning and aconitase inactivation [39,40] led us to the discovery of an inducible, protective, O2 , NAD(P)H and FAD-dependent, and cyanide-sensitive NO metabolic enzyme that yielded nitrate that was flavoHb [40,41] (Equation 4). This finding, coupled with Doyle’s 1981 report [22] and a wealth of structural and chemical data, led us to hypothesize an enzymic NOD function (EC 1.14.12.17) for the two-domain flavoHbs (Fig. 1). Importantly, a protective NOD function necessarily precluded the formation and release of toxic intermediates such as − OONO. A mechanism involving − OONO release, which had been allowed for by Doyle’s mechanism (Equation 2a) [22,36–38], and reportedly confirmed by Wade and Castro using horse heart Mb [37], was inconsistent with (i) the protection flavoHb afforded bacteria against NO [40,41] and (ii) the near-quantitative nitrate yields [22]. Thus, a mechanism for isomerization of the − OONO intermediate to nitrate, facilitated by the ferric heme

292

P.R. Gardner and A.M. Gardner

Hemoglobin

Reductase

Fig. 1. FlavoHb structure. FlavoHbs are composed of an N-terminal Hb domain linked to a C-terminal FAD-containing reductase domain. FAD and heme cofactors are highlighted with Tyr29(B10) in distal heme pocket. Structure shown is that for the A. eutrophus flavoHb monomer (1CQX) [42].

iron atom, was suggested (Equation 3) [41]. Furthermore, a primal NOD function for the Hb superfamily, including muscle Mb and red blood cell Hb, was inferred since modern Hbs evolved from the ∼1.8-billion-year-old microbial flavoHb (Fig. 1) [1,15,16].   HbFeII O2 + NO → HbFeIII− OONO → HbFeIII + NO− 3

(3)

+ + 2NO + 2O2 + NADPH → 2NO− 3 + NADP + H

(4)

The work of Hausladen et al. [43] confirmed our conclusions and supported the proposed dioxygenase mechanism. Indeed, several groups were already on distinct paths toward identifying a role for primordial flavoHbs in NO biology. Most notably, the nitrite and NO inducibility of flavoHbs in Bacillus subtilis and E. coli [44,45] led Poole and coworkers to surmise a role for flavoHb in protecting against NO toxicity in 1996. Indeed, Crawford and Goldberg reported protection of aerobic or anaerobic Salmonella typhimurium against acidified nitrite and GSNO by its inducible flavoHb, suggesting a role in O2 -independent NO detoxification [46]. A molecular function for Alcaligenese eutrophus flavoHb as an O2 -independent NO reductase had been hypothesized earlier by B. Friedrich’s group [47], but no activity was observed. Here, we discuss the evidence for a primal and common function of diverse flavoHbs, Hbs, and Mbs as NODs. Next, we focus on the role of Hb structure in the chemistry and enzymology of the NOD reaction. Key structures associated with the evolution of Hb from an enzyme to an O2 transport-storage protein are discussed. Finally, NODs and inhibitors are considered for their potential medical, agricultural, and biotechnological applications. Other enzymic functions for flavoHbs and Hbs are considered.

Nitric Oxide Dioxygenase

293

2. ROLE FOR Hbs AS NO-METABOLIZING ENZYMES 2.1. Diverse Organisms Metabolize NO to Nitrate via Hbs Conrad and coworkers first reported that various microorganisms metabolize NO to nitrate via nitrate-inducible enzymes [48]. Dioxygen-dependent NO metabolism can now be attributed to the nitrate-yielding reaction of flavoHb and Hb [41,43] (Equations 3 and 4). Oxygenated red blood cell Hb, muscle Mb, neuronal neuroglobin (Ngb), and cytoglobin (Cygb) can also metabolize NO to nitrate [22,32,49–52]. In rats, greater than 95% of inhaled NO is converted to serum nitrate [23]. In plants, the oxygenated symbiotic Lba and nonsymbiotic (ns) Hb catalytically convert NO to nitrate [53–56]. Enzymes with flavoHb-like properties also appear to fulfill a NOD function in other organisms and tissues. For example, various mammalian cell types metabolize NO to nitrate via a dioxygen and heme-dependent, cyanide and CO-sensitive, NADPH-cytochrome P450 oxidoreductase-coupled NOD activity [52,57,58]. Nitrate can also form through the diffusion-limited reaction of NO with the ubiquitous superoxide radical (− O2 •) (6700 M−1 s−1 ) [59,60], but the reaction forms − OONO, which either isomerizes to nitrate (Equation 2b), forms nitrite, or nitrosylates and nitrates proteins in the cellular milieu [60,61]. Indeed, − OONO formation has been consistently considered a major path for NO decomposition within cells and tissues [62]. Yet, at the low steady state − O2 • levels found within cells (∼10−10 M), rapid reaction of NO with more abundant and highly reactive flavoHbs, single-domain Hbs, trHbs, Cygb, Ngb, Lba, ns Hbs, and Mb (≥10−7 M) is overwhelmingly favored, as most clearly evidenced by diminished NO metabolism by (flavo)Hb-deficient cells [41,63,64] (vida infra).

2.2. NO is ubiquitous and requires detoxification A wealth of evidence now supports the view that NO is a ubiquitous poison and that all life forms require NO-metabolizing enzymes and other systems to defend against NO poisoning (Fig. 2) [65]. The immune system of animals or plants produce NO via inducible NO synthases to inhibit or kill infectious organisms and neoplasms (reviewed in refs. [66–70]). NO is also abundantly formed within tissues and cells by enzymic or nonenzymic nitrite reduction or by disproportionation under acid conditions [71–73]. Nitrite reductase generates copious NO in soils during microbial nitrite dissimilation [47,74,75]. Combustion and electrical discharges also generate NO from the reaction of O2 and N2 . Furthermore, NO is toxic to most cells. At <100 nM, NO inactivates sensitive and critical [4Fe-4S]-containing (de)hydratases including the Krebs cycle aconitase [39,40,56,76,77] and Entner-Doudoroff pathway 6-phosphogluconate dehydratase [78]. NO also potently poisons respiration by reversibly binding the heme a3 /CuB site in terminal oxidases and competing with O2 binding [57,79–84]. Indeed, the list of NOsensitive metalloenzymes and other targets is long and growing [85–88]. NO exposures can also form myriad secondary reactive nitrogen species including NO2 , nitrous anhydride (N2 O3 ), GSNO, nitroxyl (− NO), and − OONO, displaying varied reactivities and toxicities (reviewed in refs. [61,62,69,88]). NO toxicity has often been attributed to the more lethal reactions of − OONO [62,89,90] and other secondary reactive nitrogen species including GSNO [91–94], NO2 ,

294

P.R. Gardner and A.M. Gardner •NO Sensor

Signaling

–

NO3 +O2

NOD POISONING

•NO NOR

N2O

•NO Sensor

Signaling

Fig. 2. Role of NODs and NORs in NO biology. NODs and NORs are major NO metabolic pathways in the biosphere producing nitrate and N2 O, respectively. Functioning either together or independently, NODs and NORs prevent NO poisoning of cellular targets and modulate various NO signaling functions including the feedback regulation of NOD and NOR expression.

and N2 O3 rather than NO per se. Following this view, sensitive and critical targets of NO poisoning and protective NO-metabolizing enzymes were overlooked or ignored. Indeed, ignorance of NO toxicity most likely delayed the recognition of Hbs as NODs and the discernment of the critical role of aerobic and anaerobic NO-metabolizing enzymes in NO biology (Fig. 2). Enzymes metabolizing secondary reactive nitrogen species such as − OONO and GSNO reductases have been identified [43,67,95–97], but these would necessarily serve a secondary role in the defense against NO toxicity.

2.3. Microbial FlavoHbs and Hbs Metabolize and Detoxify NO Many organisms employ flavoHb or Hb to metabolize NO and serve as a first line of defense against NO toxicity. FlavoHb-deficient mutants of E. coli (hmp) [41,43,98], Saccharomyces cerevisiae (yhb1) [99], Cryptococcus neoformans (fhb1) [97], Candida albicans (yhb1) [100], Salmonella enterica (hmp) [101], and Staphylococcus aureus (hmp) [102] lack significant aerobic NO metabolic activity despite the potential for reactions of NO with − O2 • and target molecules [62]. Mycobacterium tuberculosis mutants lacking trHbN (glbN) also show decreased aerobic NO metabolic activity [64]. Further, NOD activity, flavoHb protein, or flavoHb gene (hmp) transcription has been shown to be induced by NO, or NO-generating agents, in a variety of microbes including E. coli, C. albicans and B. subtilis, S. typhimurium, and Pseudomonas aeruginosa [40,41,43,45,46,93,100,103–107]. Salmonella enterica induce hmp transcription when engulfed by macrophages, presumably via NO [108], and Yersinia pestis induce hmp transcription within the infected bubo [109]. Many Hbs, such as the Vitreoscilla Hb [110,111], some ns Hbs and Ngb are apparently not induced by NO, but by hypoxia, suggesting functional benefits for increased Hb at low O2 . FlavoHbs protect a variety of microbes against the toxicity of NO and NO generators [41,43,46,70,97,99,100,102,104,106,107,112]. The NO metabolic activities of

Nitric Oxide Dioxygenase

295

(flavo)Hbs effectively protect aconitases [40,98] and terminal oxidases [81,113] from inhibition and prevent growth impairment at nanomolar to micromolar NO concentrations. Campylobacter jejuni and C. coli contain a novel NO-inducible single-domain Hb (Cgb) that fulfills a NOD function [114]. The inducible flavoHb (HmpX) from the plant pathogen Erwinia chrysanthemi serves a similar function by protecting the bacterium from NO toxicity and interfering with NO signaling in the hypersensitive response, both of which are necessary for effective plant immunity [70].

2.4. Worm Hbs Metabolize and Detoxify NO Ascaris suum Hb also metabolizes NO and O2 through a dioxygenase or “deoxygenase” mechanism in vitro, protecting the O2 -sensitive intestinal worm from both O2 and NO toxicity [115]. An unusually low turnover number of ∼0.07 NO heme−1 s−1 was recorded for A. suum Hb, but this may be due to the fact that only NADPH, and no reductase, was included in the assay. Alternatively, the Ascaris Hb may function in the nematode as an NO-scavenging enzyme with a low turnover rate, low apparent KM (NO) and KM (O2 ) similar to that anticipated for muscle Mb, red blood cell Hb, neuronal Ngb, Cygb, and other highly expressed single-domain Hbs.

2.5. Plant Hbs Metabolize and Detoxify NO Near-stoichiometric nitrate has also been shown to form in the rapid reaction of NO with the ns maize HbFeII O2 [53]. Moreover, evidence for roles of the hypoxiainduced ns Hb, or heterologously expressed ns barley Hb, in protecting cellular energetics and maintaining low endogenous NO levels in alfalfa root or maize cultures under hypoxic stress supports a NOD function [53,116–118]. The benefit of hypoxic induction of ns Hbs can now be explained by the decreased efficiency of the NOD activity at low O2 concentrations. Indeed, the hypoxia-inducible ns Hb (AHb1) from Arabidopsis thaliana limits NO emission in transgenic Arabidopsis plants under hypoxic stress [119,120]. In further support of a NO metabolic function, ns Hbs are strongly induced by NO and NO-generating growth conditions in the model legume Lotus japonicus [121], in cultured rice cells [122], and in Arabidopsis seedlings [123]. The ability of the root nodule oxy-Lba (legume Hb) to rapidly convert NO to nitrate in vitro [56] suggests a function for the abundant Lba ancillary to O2 transport sequestration [124,125]. Symbiotic Lbas are essential for nitrogen fixation, bacteroid nitrogenase expression and energetics in root nodules [126], observations reminiscent of the effect of externally added oxy-Lba on respiration and energetics within isolated Rhizobia [127]. Acting as a NOD, Lba would be expected to protect the NO-sensitive bacteroid nitrogenase [128–130] and terminal respiratory oxidase, as well as other sensitive targets. NOD activity would also scavenge NO produced by bacteroids and soil microbes and oxidatively “fix” nitrogen to nitrate. Lba may also facilitate O2 diffusion for bacteroid respiration while limiting O2 -mediated nitrogenase inactivation. Further investigation of

296

P.R. Gardner and A.M. Gardner

a NOD function may help clarify the enigmatic plant (leg)Hb functions [124,125,131]. The potential significance of Hb modulation of NO bioactivity in complex plant physiology has been reviewed [55,132].

2.6. Heterologously Expressed Hbs Metabolize and Detoxify NO A variety of single-domain Hbs and flavoHbs show O2 -dependent NO metabolic (NOD) activity when expressed in heterologous hosts including plants, and these engineered strains show increased growth and resistance to NO poisoning [133–139]. Surprisingly, Hbs can increase bacterial growth yields in the absence of an exogenous NO source presumably by preventing the toxicity of endogenously produced NO [140]. A list of flavoHbs and single-domain Hbs including truncated Hbs and mammalian Hbs showing NOD activity in vitro or in cells is provided in Table 1. It should be noted that some rate constants were determined under suboptimal conditions such as within cells engineered to overproduce the protein or with inhibitory concentrations of NO, whereas other rate constants were determined under defined in vitro conditions. Furthermore, in the case of single-domain Hbs and Mb, maximal turnover numbers remain to be determined with optimal reductase coupling. Nevertheless, the data show the range of NO metabolic activities and NO dioxygenation rate constants that have been measured. The availability of simple methods for measuring NOD steady state and transient kinetic constants [38,141,142,149] should facilitate future comparisons. Table 1. NO dioxygenation by flavoHbs, Hbs, truncated Hbs, and Mb Source

Type

kNOD  M−1 s−1 or (kcat , NO heme−1 s−1 )

References

Escherichia coli Alcaligenese eutrophus Saccharomyces cerevisiae Candida albicans (YHB1) Bacillus subtilus Klebsiella pneumoniae Pseudomonas aeruginosa Deinococcus radiodurans Slamonella typhimurium Vitreoscilla sp. Campylobacter jejuni Mycobacterium tuberculosis Mycobacterium tuberculosis Mycobacterium leprae Ascaris suum Human Sperm whale Human Plant-legume

flavoHb flavoHb flavoHb flavoHb flavoHb flavoHb flavoHb flavoHb flavoHb Hb Hb trHbN trHbO trHbO Hb Hb Mb Ngb Lba

2400 (10–670) 2900 (3.4–290) 860 (112) (17) (68) (95) (26) (3) (76) (0.3–0.9) (0.4) a 745 (10) a 0.6 a 2.1 (0.07) a 50–89 a 34–43 b >70 a >100

[134,135,141,142,173] [135,142] [142] [148] [135] [135] [135] [135] [135] [134,135] [135] [64,134] [143] [147] [115] [35,38,144,145,146] [22,38,145] [51] [56]

ab

Rate constants reported for 20 C and 5 C, respectively.

Nitric Oxide Dioxygenase

297

2.7. Roles for (flavo)Hbs and Mb in Microbial Virulence Evidence for a protective role of microbial flavoHbs against immune system-derived NO is rapidly gathering. Lethal mucoid P. aeruginosa strains isolated from cystic fibrosis patient lungs show flavoHb and NOR mRNA expression elevated manyfold, suggesting a role for NO metabolism in virulence [150,151]. FlavoHb-deficient mutants (hmpX) of the bacterium E. chrysanthemi lack virulence toward African violets (Saintpaulia ionantha) [152], presumably because of deficient metabolism of NO released by leaf cells [70]. However, the role of a polar effect of the hmpX mutation on expression of the proximal virulence gene pectate lyase (pelA) has not been excluded. FlavoHb-deficient S. enterica [101,153] and Y. pestis [109] show greater sensitivity to NO-related killing by aerated macrophages, suggesting a similar role in infections. Modest decreases in the virulence of flavoHb-deficient C. neoformans [97] and C. albicans [100,154] have been observed in systemic fungal infections in mice. In the case of C. albicans, decreased virulence may be due to filamented growth or other defects rather than an increased sensitivity to NO per se since the inducible NO synthase was reportedly not required for decreased virulence [154]. More recently, flavoHb has been shown to be critical to the virulence of S. enterica [101], S. aureus [102], and Y. pestis [109] in systemic infections of mice or rats. Respective roles for Hb, Mb, and Ngb in sheltering infectious Plasmodium falciparum in red blood cells [155], Trypanosoma cruzi and Toxoplasma gondii in muscle [156,157], and T. gondii in neurons and retina [157] have also been hypothesized. Other abundant Hbs such as the root nodule Lba may similarly shield symbiotic bacteroids from the hypersensitive reaction of plants and the copious NO produced by soil microorganisms during nitrite dissimilation [48,74,75]. Further, the adjuvant effect of red blood cells in bacterial peritonitis is thought to be due at least in part to the NO scavenging reactions of Hb [158].

2.8. NORs Complement NODs in Microbial NO Detoxification O2 -independent NO reductases (NORs) (EC 1.7.99.7) can supplant or complement the NOD function of flavo(Hbs) in some organisms (Fig. 2). Functional complementation or replacement may be especially significant in the microaerobic and anaerobic environments of infected tissues [150,159,160], possibly explaining the limited roles of flavoHbs in the virulence of some pathogens. NORs also serve important energetic and NO detoxification functions in anaerobic soil microorganisms that reduce nitrate to nitrogen in the energy-yielding process of denitrification [74,161–163]. Inducible NORs reduce NO (Equation 5) and maintain low steady state NO concentrations [164,165]. Nondenitrifying microbes also induce robust NO-metabolizing NORs in response to anaerobic exposure to NO, nitrite, nitrate, and other NO sources [48,74,78,166–169] including macrophage infection [108]. 2NO + NADPH + H+ → N2 O + NADP+ + H2 O

(5)

298

P.R. Gardner and A.M. Gardner

2.9. A Dual Function for FlavoHbs as NODs and NORs? FlavoHbs have been shown to protect the anaerobic growth of S. typhimurium and S. cerevisae against GSNO or synthetic NO generators [46,99], thus suggesting an O2 -independent NO detoxification mechanism. Poole and coworkers have repeatedly suggested that E. coli flavoHb, and other microbial flavoHbs, function as anaerobic NORs by avidly binding NO (at the ferrous heme) and univalently reducing NO to − NO, which spontaneously protonates, dimerizes, and dehydrates to form N2 O in solution [170–172]. However, flavoHbs show relatively poor anaerobic NO metabolic activity in cells or in vitro [40,98,141,142], making a NOR function unlikely. Thus, the NOD activity of E. coli flavoHb under normoxia at 37 C (∼240 NO heme−1 s−1 ) is roughly 1000-fold greater than the highest activity reported for NO reduction by E. coli flavoHb (0.24 NO heme−1 s−1 ) [98,141,142,170]. Much lower NOD activities (<10 NO heme−1 s−1 ) have been reported under suboptimal assay conditions [43,149,173] employing excess inhibitory [NO], low [O2 ], or FAD-deficient enzyme, which suggest a more significant role for the NOR activity. However, elevated flavoHb expression fails to protect E. coli against NOmediated growth inhibition and aconitase inactivation in the absence of O2 , but protects both in its presence [98]. Moreover, E. coli, and S. typhimurium express an authentic NOinducible NOR from the norRVW operon [78,169,174]. E. coli NorV (flavorubredoxin) shows far greater NOR activity in NO-exposed cells [78] and in vitro (∼4 NO monomer−1 s−1 ) [167] than the flavoHb (0.07–0.24 NO heme−1 s−1 ). Other flavoHbs show negligible NOR activity [142], suggesting poor conservation of a NOR function. In addition, authentic NORs do not release − NO, which can be toxic [74,175]. Using GSNO as a NO generator, Arai and coworkers have reported results clearly showing protection of P. aeruginosa by flavoHb under aerobic, but not anaerobic, growth conditions [106]. On the other hand, Justino et al. reported flavoHb protection in anaerobic E. coli exposed to NO [107], suggesting an anaerobic function; however, bacteria were exposed to large nonphysiological NO boluses, thus making the results difficult to interpret. E. coli flavoHb shows modest GSNO reductase (0.03 GSNO heme−1 s−1 ) and NADHnitrite oxidoreductase (0.04 NADH heme−1 s−1 ) activities [98], which may explain the protection flavoHb affords against GSNO, nitrite, large NO boluses, and other NO generators under anaerobic conditions [46,99,107]. Similarly, flavoHb from the Actinomycete Streptomyces antibioticus inhibits NO formation from the NO-releasing diazenium diolate (NONOate) NOC-7 without nitrate formation [176], suggesting novel mechanisms for diazenium diolate metabolism. Caution is thus warranted when deducing a NO detoxification function from protection afforded against the toxicity of nitrite, GSNO, or nonphysiological NO-generating agents or exposures.

2.10. Hb, Mb, and FlavoHb Modulate NO Signaling In addition to NO detoxification, NO-metabolizing enzymes also control NO signaling (Fig. 2). In mammals, low NO fluxes are produced by constitutive NO synthases for signaling functions, and NO metabolism or decomposition diminishes steady state NO levels, lowering soluble guanylate cyclase activation [177], blood flow, and O2

Nitric Oxide Dioxygenase

299

delivery to tissues, and affects many other physiologic processes. The potentially huge effect of red blood cell Hb and muscle Mb on NO signaling is well known [30,31], although the significance of Hb and Mb reactions have only recently been illuminated [178]. Rapid and catalytic dioxygenation of NO by HbFeII O2 , MbFeII O2 , NgbFeII O2 , or CygbFeII O2 provides potential paths for NO removal in mammalian cells and tissues [49,51,52,82,179,180]. In the case of Mb, mouse hearts genetically deficient in Mb show altered hemodynamics, depressed energetics, hypertrophy, and interstitial fibrosis following elevated NO synthesis or infusion, suggesting critical roles for Mb in NO metabolism, signal modulation, and detoxification in cardiac muscle [49,181–183]. NMR evidence for MbFeIII formation was obtained within cardiac muscle infused with NO or exposed to increased NO production, supporting a mechanism for NO metabolism involving NO dioxygenation (Equation 3) [49,182]. Curiously, nitrosylation of MbFeII (Equation 1) was not measured in these investigations. Thus, at low tissue pO2 s and high NO levels, formation of stable MbFeII NO, and inhibition of NO dioxygenation, is expected given the comparable rate constants for nitrosylation and dioxygenation (Equations 1 and 2a) [82]. Also, the accumulation of MbFeIII in response to NO production reveals a limited capacity of myocytes for MbFeIII reduction and NO dioxygenation. Indeed, contrary to the proposed NO detoxification function, Li et al. observed remarkably similar NOmediated inhibition of respiration in bradykinin-stimulated heart and skeletal muscle tissue from wild-type and Mb knockout mice, indicating a limited contribution of Mb to NO metabolism and detoxification under these conditions [184]. Overall, the evidence suggests that Mb can act as an effective NOD, an ancillary function to facilitating O2 diffusion. But, the results also suggest that Mb only possesses the capacity for a low NO flux. An efficient, robust, O2 -dependent, cyanide and CO-sensitive flavoHb-like NOD activity has been measured in a variety of mammalian cell types [57,58] that may function in NO metabolism. The absence of immunologically detectable Mb, Hb, or Cygb in NO-metabolizing epithelial cells [52,57] and the enrichment of the activity in low-density microsomal membranes [58] strongly suggest that this activity functions independent of Mb, Hb, or Cygb. However, the relationship of this heme-dependent NOD to the Hb superfamily remains to be fully elucidated. In addition, microbial NODs and NORs feedback regulate their own expression by modulating NO levels (Fig. 2) [185,186]. Experimental and in silico genomic evidence [174] suggests roles for NO-sensing transcription factors such as the iron-dependent [187] NorR [105,166,185,188,189], NsrR [101,190,191], DnrD [186], ResDE [192], and [4Fe-4S]-containing Fnr [193] in controlling expression of NOD (hmp), NOR (norVW and norBC), trHbN (glbN by nsrR in Legionella pneumophila [174]), and other nitrosative stress genes in a variety of bacteria. Ancillary roles for the iron-containing Fur [194,195], [2Fe-2S]-containing SoxR [196,197] and OxyR [92,198] as global NO sensor-regulators in bacteria have also been considered, but the supportive evidence is minimal or lacking [105,174,199]. Transcription regulators in yeast and fungi are similarly thought to sense NO and regulate flavoHb and NO levels [154,200]. FlavoHb expression also affects D. discoideum [112] and C. albicans [154] differentiation and morphologies, suggesting a role for NO signaling in fungal development. The significance of NO signaling and Hbs in the context of plant physiology has been recently reviewed [55,132,201].

300

P.R. Gardner and A.M. Gardner

3. Hb STRUCTURE AND THE NOD MECHANISM 3.1. Conserved FlavoHb, Hb, and Mb Structure and Function Representative X-ray structures of the flavoHb Hb domain (A. eutrophus), trHbN (M. tuberculosis HbN), single-domain Hb (Vitreoscilla), red blood cell Hb (human), muscle Mb (sperm whale), and nonsymbiotic plant Hb (rice ns Hb) (Fig. 3) show the remarkable preservation of the overall globin fold and orientation of heme during the ∼1.8 billion years of Hb evolution within bacteria, fungi, plants, protozoa, and mammals [1–3]. Helices A–G and the CD-region are present in each isoform, albeit trHbN displays significant structural deviation with a two-over-two rather than classic three-over-three -helical sandwich [9]. Despite the conservation of the globin fold, a wealth of physical and biological evidence has suggested that these Hbs evolved for very diverse functions [6–10,125,207]. A survey of 15 flavoHbs for homology in the N-terminal Hb domain reveals a primary structure consensus (Fig. 4, top). This consensus is also preserved in the single-domain

A. eutrophus flavoHb

Sperm Whale Mb

F

H

A E G

CD B Vitreoscilla Hb

M. tuberculosis HbN

Human Hbα

Rice ns Hb

Fig. 3. Conservation of globin structure. A. eutrophus flavoHb monomer (1CQX) [42], Vitreoscilla sp. single-domain Hb monomer (2VHB) [202], M. tuberculosis trHbN monomer (1IDR) [205], sperm whale MbO2 (1MBO) [203], human HbO2 alpha subunit monomer (1HHO) [206], and rice nonsymbiotic (ns) Hb monomer (1D8U) [204] structures showing orientations of heme and distal and proximal ligands. A–H labels on A. eutrophus globin domain follow classical Hb structure nomenclature.

Nitric Oxide Dioxygenase

301

A •#

M

#∗ #

∗ #

#

#

∗ •#

∗ #

•

∗ •#

D

∗ •#

L DA QT I A I I K A T I P L L EE TG PK L T AH F Y DR MF GH H P E L K N I F N MS N

1

Consensus

C

B

10

20

30

40

F

E •

#

#

∗ ##

∗

∗ • •#

•

• #

#

#

•

QR N GD Q R E A L A N A V L A Y A S N I D N L P A L L P A V E R I A H K H V S L Q I K

50

∗#

60

∗• •

70

80

G •# ∗ •#

H

∗∗

•

••

•

• P E H YP I V GT H L L A A I KE V L

100

GD A

90

∗

#

∗ •

#

A T Q E V L D A W G K A YG V L AD V F I G R E K E I Y EQ S A E

110

120

130

140

A

1 2 119 74

H

128 93 135

95

139 85

F 84 e

120 124

15

102 17 9798 28 21 57 32 39 43 29 53 38 B 37

G

E CD

FAD

48

Fig. 4. Conservations within the Hb domains of flavoHbs and single-domain Hbs. (Top) Consensus amino acid sequence and scores determined from a random set of flavoHb Hb domains from bacteria and fungi. The set includes flavoHbs from E. coli, A. eutrophus, S. cerevisiae, S. aureus, Y. pestis, V. cholera, B. subtilis, E. chrysanthemi, S. typhimurium, P. aeruginosa, F. oxysporum, K. pneumoniae, D. radiodurans, D. discoideum, S. pombe. Boxes designate region of helix structure and letters A–H follow classical Hb structure nomenclature. Sequences underlined in the Hb domain are identities with the Vitreoscilla sp. Hb. Number symbol (#) indicates identities with human Hb. Solid dot (•) indicates residue identities with the mouse and human neuroglobins. (∗ ) Asterisks indicate identities with mouse and human cytoglobins. (Bottom) Plot of high-consensus amino acids on E. coli ferric-flavoHb (1GVH) [208] ribbon backbone structure.

302

P.R. Gardner and A.M. Gardner

Vitreoscilla Hb, as revealed by the sequences underlined in Fig. 4 (top). However, other Hbs including the M. tuberculosis trHbN, human red blood cell Hb, the human Ngb, and Cygb show very weak homology with the flavoHb consensus as shown by the symbols (#, •, and ∗ , respectively). Key residues including proximal His(F8) and Phe(CD1) are invariant in these diverse Hbs. Yet, many of the unique residues in key functional positions in flavoHb and Vitreoscilla Hb are absent in other Hbs, suggesting adaptation for other functions. Foremost, the Tyr(B10) hydroxyl is positioned in the A. eutrophus, Vitreoscilla, and E. coli ferric flavoHb distal pockets to hydrogen bond and stabilize O2 binding (see Figs. 1, 3, and 4) [42,202], serving the same general role as His(E7) in mammalian Hbs, Mbs, and the rice ns Hb [203,204,209–211]. Indeed, replacement of Tyr(B10) with Phe in E. coli flavoHb results in a ∼60-fold lower O2 affinity [141]. A similar stabilization of O2 by Tyr(B10) is seen in Ascaris Hb [212–214], Aquifex aeolicus thermoglobin [15], M. tuberculosis HbN [143,215–218], and other Hbs, suggesting similar adaptation for a NOD function. In addition, highly conserved Gln(E7) in the flexible CD-loop may form a hydrogen-bond network stabilizing O2 binding [42,202,219,220], although evidence for O2 binding by E7 mutants of Vitreoscilla Hb suggests otherwise [220]. A similar role has been previously described for Gln(E7) in Ascaris Hb and other Hbs [212–214]. Other unique conservations within the Hb domains of flavoHbs include the cluster of hydrophobic residues in the distal heme pocket, the site of the NO dioxygenation reaction as illustrated by the residues highlighted in the E. coli flavoHb structure (Fig. 4, bottom). Conserved Phe28(B9), Met32(B13), Phe43(CD1), Leu57(E11), and Val98(G8) are in position to influence (i) O2 binding and reactivity, (ii) NO entry and reactivity with bound O2 , and (iii) intermediates of the NOD reaction. A unique positioning of the isobutyl side chain of Leu57(E11) within 3.6 Å of iron in the ferric E. coli flavoHb structure led Ilari et al. to suggest a unique functional role for this side chain in shielding ferric iron from ligands and propagating an O2 -linked conformational change that brings Tyr(B10) inward in the distal pocket to hydrogen-bond and stabilize O2 [208]. A much different hydrophobic distal pocket is observed in the modern O2 transport and storage red blood cell Hb and muscle Mb. The distal pockets are smaller and show different residues in most of the key conserved positions. In myoglobin, the corresponding conserved distal pocket residues are Ile28(B9), Leu29(B10), Leu32(B13), Phe43(CD1), His64(E7), Val68(E11), and Ile107(G8) [36]. The influences of conserved Mb and Hb distal pocket residues and replacements on O2 and NO binding and the NO dioxygenation activity have been well documented [35,36,38,144]. For example, bulky hydrophobic aromatic replacements at positions E11 and B10 hinder NO accessibility and the NO dioxygenation reaction, while replacements of His(E7) with nonpolar residues increase bimolecular entry of NO and NO oxidation (dioxygenation) [35,38]. Clearly, many Hbs including red blood cell Hb, muscle Mb, Ngb, Cygb, and trHbN lack structural homology within conserved functional regions of the flavoHb and Vitreoscilla Hb structures (Fig. 4, top), suggesting a loss of the NOD function during the ∼1.8 billion years of Hb evolution. Nevertheless, trHbN, trHbO, red blood cell Hb, muscle Mb, and Ngb do show NO oxidation (dioxygenation) activity (Table 1). Further, biological studies support a NOD function for these distantly related Hbs (vide supra) and further suggest an intrinsic capacity for all Hbs to act as NODs [221]. Thus, NO dioxygenation would appear to be a reaction catalyzed by all Hbs, not dependent upon

Nitric Oxide Dioxygenase

303

an absolute structure. On the other hand, NOD activities may be vestigial and function poorly, if at all, in some cells. Thus, the lack of a suitable reductase, significant release of − OONO (Equation 2a) or other intermediates, or the presence of other NO metabolic enzymes within cells may diminish the capacity of various Hbs and Mbs to serve as NODs. In addition, recent studies of Candida albicans suggest that not all flavoHbs function as NODs. Thus, C. albicans genome contains three genes encoding flavoHbs, but only one of these, YHB1, has been shown to possess NOD activity and function within nondifferentiated cells [100]. The two other flavoHbs, YHB4 and YHB5, lack key conserved amino acids in the distal pocket, suggesting a different function. The YHB4 sequence shows Phe(B10), Leu(B13), Phe(E11), and Phe(G8) in place of the highly conserved Tyr(B10), Met(B13), Leu(E11), and Val(G8), respectively. YHB5 contains the conserved Tyr(B10), but shows unique Leu(B13), Met(E11), and Leu(G8) residues. Studies are needed to determine the role of these and other conserved residues (Fig. 4, top) in the NOD function of flavoHb. It remains possible that YHB4 and YHB5 function as NODs in special aspects of fungal differentiation and NO signaling.

3.2. Ligand Binding to (flavo)Hbs and Mbs and the NOD Function O2 binding to ferrous heme is an intrinsic property of all Hbs [10], and is essential for the NOD reaction and function. FlavoHbs, Hbs, and Mb show different rate constants for O2 association and dissociation (Table 2) that are largely determined by dielectric and steric influences within the distal heme pocket structure [10,211]. Ligand binding is also under the influence of proximal His(F8) and neighboring interactions [212,226,227]. The heme pocket structures and O2 -binding properties of flavoHbs, Hbs, and Mbs are adapted for different functions, with the O2 storage and transport Hbs and Mbs showing lower O2 affinity due to larger O2 dissociation rate constants (kO2 ) and the enzymatic flavoHbs showing higher O2 affinity due to smaller O2 dissociation rate constants (Table 2). In contrast, flavoHb, Hb, and Mb show similar NO association (kNO ) and dissociation (kNO ) rate constants, indicating little adaptation of these heme pockets for NO binding. As already mentioned, distal heme pocket structure and hydrogenbonding residues, including Tyr(B10) in flavoHbs and His(E7) in red blood cell Hb and muscle Mb, play key roles in stabilizing the bound O2 and in determining O2 affinity. For example, loss of hydrogen bonding by Tyr(B10) in E. coli flavoHb, thermoglobin, Ascaris Hb and His(E7) in red blood cell Hb and muscle Mb increases O2 dissociation rate constants by >60-fold, while showing lesser, albeit beneficial, effects on O2 association [15,141,210,228,222]. Moreover, a 60-fold lower O2 affinity of a Phe(B10) mutant of E. coli NOD (flavoHb) results in a profound susceptibility to NO inhibition [141]. Similar, albeit lesser, effects of Gln(E7) mutations on O2 affinity and NOD activity have also been observed [219]. These results demonstrate the critical role of O2 affinity in the NOD function of flavoHb and suggest an inherent limit on the NOD function of Hbs with relatively large kO2 values and small polar distal pockets such as the muscle Mb and red blood cell Hb. Rate constants for O2 association-dissociation are routinely measured and compared [13,15,229–231], and suggest capacity for a NOD function. Thus, studies of E. coli flavoHb suggest that Hbs with large kO2 values will only function at low NO levels (relative to O2 ) and low turnover rates, being limited by competitive

304

P.R. Gardner and A.M. Gardner

Table 2. Comparison of kinetic constants for diverse hemoglobins. Elementary rate contants for ligand binding and steady state rate constants are given for E. coli, S. serevisiae, and A. eutrophus flavohemoglobins (flavoHbs), human red blood cell hemoglobin (RBC Hb) (R state), sperm whale myoglobin (SW Mb), M. tuberculosis trHbN, and Vitreoscilla sp. Hb (VHb) Rate Constant

FlavoHbs

kO 2 , M−1 s−1 kO 2 , s−1

17–50 [141,142] 14–17 [142,210] 0.2–0.6 [141,142] 12–15 [142,210]

Kd (O2 ), nM  kNO , M−1 s−1

4–36 10–26 [141,142]

kNO , s−1 Kd (NO), pM  kNOD , M−1 s−1

kcat (NOD), NO heme−1 s−1 KM (O2 ), M KM (NO), nM kcat (NOR), NO heme−1 s−1 a

SW Mb

RBC Hb

HbN

10–20 [223] 9–24 [222]

25 [215] 0.20 [215]

710–880 >1000 17–22 25 [225] [38,142,225] 0.0002 [141,142] 0.0001 [142,225] 0.00005 [225] 8–20 5 2 900–2900a 34–43 [38,145] 50–89 [141,142] [35,38, 144–146] 112–670a – – [141,142] 60–90a – – [141,142] 100–250a – – [141,142] 0.01–0.24 – – 0.02–0.12a [141,142,170]

VHb

8 –

200 [224] 0.15–4.2 [224] 8–21 –

–

–

– 745 [64]

– –

–

–

–

–

–

–

–

–

Rate constants were determined at 37 C. All other rate constants are reported for room temperature.

NO inhibition and slow NO dissociation (kNO ). Nevertheless, these Hbs may act as very effective NODs when expressed at micromolar to millimolar levels within cells and tissues.

3.3. The NO Dioxygenation Mechanism As initially proposed by J. J. Weiss in 1964 [232,233], O2 is bound to most Hbs with significant ferric-iron-superoxide (FeIII − O2 •) character [209,234–236]. Moreover, the NO dioxygenation function and mechanism of flavoHbs, Hbs, and Mbs appears to depend entirely upon this unique character. Doyle and Hoekstra [22] were the first to suggest a rapid reaction of NO with HbFeII O2 to form − OONO (Equation 2a). The rate constants for the reactions of MbFeII O2 and HbFeII O2 with NO are indeed large at ∼40 M−1 s−1 and ∼70 M−1 s−1 , respectively (Table 2) [22,38,145]. Furthermore, the rate constants for the reaction of NO with flavoHbFeII O2 or trHbNFeII O2 are even larger at 750 to 2900 M−1 s−1 [141] (Table 2), and approach the diffusion-limited rate constant for the radical–radical coupling reaction of NO (•NO) with − O2 •, or HO2 •, in aqueous

Nitric Oxide Dioxygenase

305

solution (6700 M−1 s−1 ) [59,229]. Intermediary values for the oxy-Ngb and oxy-Lba reactions have been recently estimated at >70 M−1 s−1 and >100 M−1 s−1 at 5 C and 20 C, respectively [51,56]. The mechanism of Hb-catalyzed NO dioxygenation can be compared to the reaction of NO with − O2 • to form − OONO and the isomerization of HOONO to form nitrate in solution. In the first reaction, the unpaired electron in the  ∗ antibonding orbital of NO couples with the  ∗ antibonding orbital of − O2 • to form − OONO [59,237]. − OONO is relatively stable; however, HOONO (pKa = 6 8) isomerizes spontaneously to nitrate with a first-order rate constant of ∼1.3 s−1 [60,238–240]. In 1952, Halfpenny and Robinson proposed that ONOOH decayed via ONO OH peroxo bond homolysis [239,241]. •OH and NO2 (•NO2 ) then recombined to form nitrate (Fig. 5, reactions 9 and 10). Anbar and Taube further suggested that the ONO OH bond is polarized, which weakens it to rupture [242]. Bending and vibration of ONO OH bond brings the terminal O-atom in contact with the electron pair on the nitrogen, allowing minimal exchange of solvent O-atoms and a concerted internal O-atom rearrangement with the retention of peroxide O-atoms as demonstrated by isotope studies [242]. Later, Beckman and coworkers concluded that a vibrationally activated trans-ONOO− isomer allows the terminal peroxide oxygen to approach the N-atom by a slight lengthening of the peroxo bond and contraction of the − O O N bond angle, causing a direct rearrangement to nitrate [243–245]. Other rearrangement mechanisms have been suggested [246,247], but do not adequately explain the retention of O-atoms (∼87%) reported in nitrate [242,248]. In the proposed flavoHb-catalyzed mechanism, univalent reduction of the ferric heme by the reductase domain NADH/FADH2 cofactors initiates the NO dioxygenase reaction cycle (reaction 1, Fig. 5) [141]. O2 with two antibonding  ∗ orbital electrons binds to the reduced ferrous heme of flavoHb (reaction 2), forming an O2 complex with significant superoxo bond character. A d orbital electron from ferrous iron pairs with the proximal O2 O-atom  ∗ orbital electron to form the FeIII− OO• complex. The FeIII−OO• complex, like free − O2 • and its protonated form HOO•, reacts rapidly with  = ≤ 2 9 × 109 M−1 s−1 [141,142] to form a transient FeIII− OONO intermeNO kNOD diate (reaction 3). Stretching and homolysis of the peroxide bond forms caged •NO2 and FeIII− O• or the peroxidase Compound II-like resonance form FeIV O (reaction 4) that rapidly combine to produce FeIII (nitrate) (reaction 5). Nitrate is then released from the heme iron (reaction 6). Another possibile reaction path includes a concerted internal O-atom rearrangement (reaction 4a) like that proposed for HOONO [242–245], in which bending and vibration of the peroxo bond and contraction of the FeIII − OO–NO bond brings the free electron pair on nitrogen in close contact with the iron-bonded O-atom, and the peroxo bond is simultaneously ruptured as the iron-bonded O-atom bonds nitrogen to form FeIII (nitrate) (reaction 4a). Regardless of the mechanism of O-atom rearrangement, both O-atoms are conserved quantitatively, forming nitrate from NO. Thus, reaction of NO with E. coli flavoHb18 O2 , sperm whale Mb18 O2 , or human Hb18 O2 results in >99% double O-atom incorporation into nitrate [221], whereas O-atom retention during isomerization of HOONO in aqueous solution yields ∼87% O-atom retention [248], indicating water O-atom incorporation and a heterogeneous nonenzymatic mechanism. Low O-atom retention is also observed in reactions of NO with the structurally dissimilar indoleamine dioxygenase-like Mb18 O2 (∼94%) or free −18 O2 • (∼79%) [221], suggesting a unique capacity of Hbs for NO dioxygenation. Moreover, the rate constant for isomerization (kIS ) catalyzed by flavoHb

306

P.R. Gardner and A.M. Gardner

H2C

(1) e

HO

N FeII

HN

(2) O=O

CH2

H2C HN

N FeIII

HO

H2C HN

CH2

H+ O + –O N O– (10)

(6)

•N

HO N FeIII O • O

(9)

O

HO • O

N O O H

O •N=O

H+ (8) H2C

O + N FeIII O N HO O–

HN

(3) N O –O

CH2

(7)

H2C N Fe

III

O •N O• HO O

H2C HN CH2

H2C HN

O

(4a)

(5)

HN

CH2

IV

N Fe

(4)

N FeIII O

N

O HO

O CH2

O •N HO O O CH2

Fig. 5. Proposed mechanism for Hb-catalyzed NO dioxygenation-peroxynitrite isomerization reaction sequence (reactions 1–5) and nitrate release (reaction 6), and the mechanism for spontaneous isomerization of peroxynitrite expelled from the distal heme pocket (reactions 7–10).

is estimated to be >>670 s−1 from the Vmax at 37 C [141], whereas the solution rate constant for − OONO isomerization at a similar physiological pH value of 7.5 is ∼4000fold lower at 0.17 s−1 [145,238]. Furthermore, high-fidelity O-atom incorporation did not support a mechanism involving significant − OONO intermediate release (Equation 2a) [22,38] as previously evidenced by the absorbance of − OONO at 302 nm at alkaline pH and by the formation of •OH [37].

Nitric Oxide Dioxygenase

307

The mechanism of nitrate formation by flavoHb (and other Hbs) differs from the spontaneous mechanism in that the distal heme pocket has capacity to sterically and electrostatically facilitate and shield intermediates from reaction with water, NO, or biomolecules. In this regard, conserved flavoHb distal pocket amino acids including Leu(E11), Val(G8), Phe(CD1), Phe(B9), and Tyr(B10) may guide the NO reaction with FeIII− OO•, decrease the activation barrier for a trans-peroxynitrite transition state [243–245,247], and cage the isomerization intermediates for rapid rearrangement, thus simultaneously eliminating NO and shielding − OONO and its isomerization intermediates (FeIV O and •NO2 ) from reaction with NO, water, and other biomolecules. A similar high-fidelity NOD reaction and function has been proposed for HbFeII O2 and MbFeII O2 [36,144,221]. Herold and others have provided evidence for a bound and transient HbFeIII− OONO intermediate by measuring changes in the Hb spectra at 406, 504, 577, and 636 nm under alkaline conditions [145,146,244]. A similar spectral signature was reported for MbFeIII− OONO [145] and, more recently, for Mycobacterium leprae trHbOIII− OONO [147]. In addition, Olson and coworkers have observed a transient high spin ferric g = 6 feature in EPR spectra taken at alkaline pH and tentatively assigned it to the HbFeIII− OONO intermediate [144], although the identity of the signal with Hb FeIII (nitrate) has not been excluded [249]. The FeIII− OONO intermediate decays rapidly with high first-order rate constants of 58 s−1 and 341 s−1 for Hb at pH 7.5 and Mb at pH 8.3, respectively [145,146,250], indicating a 350–2000-fold catalytic enhancement of − OONO isomerization approaching that measured for flavoHb. In addition, His(E7) mutants of MbFeIII with open active sites were shown to catalyze − OONO isomerization, with rate constants approaching those determined from the decay of the MbFeIII− OONO intermediate [251]. These results strongly support a catalytic dioxygenation mechanism (Equation 3) over a mechanism involving − ONOO release (Equation 2a) and isomerization (Equation 2b). We should make clear, however, that the identity of the bound intermediate detected by Herold et al. has been questioned [252]. The remarkably similar heme signature and decay rate (∼190 s−1 ) for the FeIII (nitrate) intermediate formed in reactions of NO2 with oxo-ferryl Mb (FeIV O) suggest that the transient spectral signature reported by Herold and others is in fact FeIII (nitrate). Further, on the basis of the thermokinetic calculations, Goldstein et al. have argued that the FeIII− OONO intermediate would have a submicrosecond half-life that would be too short-lived to observe [252]. Moreover, they argue that FeIII− OONO decay should not be so strongly influenced by alkaline pH as reported by Herold et al. Nevertheless, Blomberg et al. provided possible explanations for the alkaline stabilization of the peroxynitrite intermediate in their models of the Mb reaction [253]. An alkaline pH is expected to lead to the deprotonation of the proximal His(F8) imidazole (pKa = 10 5), an increased imidazole bond strength to iron, a decreased Lewis acid character of heme iron, a weakening of the iron peroxynitrite bond, and a decreased O O bond homolysis rate. Alternatively, alkalinization may cause an altered orientation of the distal His(E7), resulting in increased hydrogen bonding to the peroxynitrite intermediate at an alkaline pH, or decreased protonation of the imidazole that may hinder the isomerization mechanism. Surprisingly, Blomberg et al. independently concluded that none of these effects could fully account for the large alkaline stabilization of the peroxynitrite intermediate reported by Herold et al. and that the peroxynitrite intermediate should have been too short-lived to observe. Experimental reconciliation of these critical findings is clearly required. Preliminary results suggest

308

P.R. Gardner and A.M. Gardner

that alkaline pH conditions alter the Mb reaction pathway, resulting in significant peroxynitrite release and nitrite formation (Gardner and Olson et al., unpublished results). Two possible mechanisms for Hb–FeIII− OONO isomerization to nitrate have been considered and investigated. These include (i) an oxo-ferryl mechanism in which a caged [FeIV O •NO2 ] intermediate is formed through O O bond homolysis (reaction 4) followed by ferryl oxygen attack of the nitrogen (reaction 5) [36,144,145,252–254] and (ii) a concerted internal rearrangement mechanism in which O O bond homolysis and O-atom rearrangement to nitrogen occur simultaneously (reaction 4a) [145,253,254]. Evidence for a ferryl intermediate, in the absence of H2 O2 formation [255], would indicate a mechanism for − OONO isomerization involving FeIV O (or FeIII− O•) and NO2 radical intermediates similar to that described for metalloporphyrins [240,256,257] and different from an internal rearrangement mechanism (reaction 4a). The relatively slow reaction of •+ HbFeIV = O with nitrite (16 M−1 s−1 ) and the rapid reaction of HbFeIV O, or its resonance species HbFeIII− O•, with NO2 radical (∼107 M−1 s−1 ) [252,258,259] favors a mechanism involving homolysis of the peroxide bond to produce NO2 over the nitrite-producing heterolytic mechanism. While Herold et al. were unable to detect an oxo-ferryl intermediate by monitoring visible heme spectra for ferryl in reactions of NO and HbO2 [145,146,250], Olson and coworkers have observed a transient low spin signal (g = 2) showing maxima at 539 and 575 nm in reactions of NO and a Mb FeII O2 mutant containing glutamine in place of His(E7) suggestive of FeIV O formation, although the assignment of these spectra to a ferric peroxo- or hydroxide Mb complex has not been excluded [144,249]. Together, these experimental results suggest that the oxo-ferryl intermediate is too short-lived to observe in reactions of wild-type Hb or Mb with current methods and that the His(E7) imidazole is important for the Mb-catalyzed isomerization mechanism. In their models of the Mb reaction employing hybrid density functional theory, Blomberg et al. reportedly failed to find a concerted mechanism yielding plausible activation energies [253]. Subject to several important assumptions including dielectric effects, metal-ligand bond strengths, and an unexplained low-spin assignment to the ferric peroxynitrite intermediate (S = 1), they found the oxo-ferryl mechanism to be supported by a low-energy barrier for O O bond homolysis (G = ∼10 kcal/mol) and a highly exergonic overall reaction sequence from NO and MbO2 to form nitrate and ferric Mb (G = −29 kcal/mol) [252,253]. It is noteworthy that hybrid density functional theory analysis of models of the M. tuberculosis trHbN NOD reaction by Crespo et al. suggested a similar highly exergonic oxo-ferryl mechanism [218]. Moreover, in contrast to the analysis of the Mb reaction, models of the trHbN dioxygenation reaction were treated in more appropriate low-spin doublet and high-spin quartet states for the oxo-ferryl and peroxynitrite intermediates, respectively. Nevertheless, density functional theory calculations are limited in their reliability especially when dealing with low-lying and high-valent electronic states and noninnocent ligand states [260]. Thus, high-level ab initio methods such as CASPT2 and CCSD(T) that attempt to incorporate electron orbital energies and perturbations offer greater promise for resolving the NO dioxygenation mechanism. In summary, contrary to the early results of Wade and Castro [37] demonstrating − OONO release from Hb (reaction 7) and •OH formation during the spontaneous isomerization of HOONO (reaction 9), evidence for (i) high-fidelity of O-atom retention [221], (ii) HbFeIII− OONO intermediate formation [144,145,146], (iii) Hb(Mb)FeIII -catalyzed

Nitric Oxide Dioxygenase

309

−

OONO isomerization [56,251], and (iv) no − OONO release [43,145,146] demonstrates a common and efficient (flavo)Hb-catalyzed NO dioxygenation mechanism. Groves and coworkers have shown evidence for NO2 release (∼20%) and Mb nitration in the bimolecular reaction of − OONO with MbFeIII , suggesting inefficient isomerization of − OONO involving FeIV O and diffusible NO2 [261]. NO2 release is difficult to reconcile with the proposed high-fidelity protein caged NO dioxygenation mechanism (Fig. 5) (vide supra), but could be explained by NO2 formation independent of MbFeIII catalysis. Indeed, Herold et al. reported similar nitration of MbFeIII and apoMb by − OONO, and have argued that nitration occurs independent of heme-catalyzed isomerization [262].

3.4. Catalytic Reaction Cycle of the flavoHb NOD FlavoHbs utilize 1 NAD(P)H and 2 O2 to catalytically convert 2 NO to 2 nitrate molecules (Equation 4) [41,142]. High rates of catalytic nitrate formation by flavoHbs (112–670 s−1 ) are achieved in a multistep reaction cycle (Fig. 6). Rapid turnover is achieved with (i) a large association rate constant for flavoHbFeII and O2 kO 2 , (ii) a rapid reaction of •NO with the FeIII− OO• intermediate to form a FeIII− OONO intermediate k  NOD , (iii) rapid peroxynitrite intermediate isomerization (kIS ), (iv) rapid nitrate release (kP ), and (v) rapid reduction of the ferric flavoHb by 2e transfer from NAD(P)H to FAD kH  and a sequential 1e transfer from FADH2 to the ferric heme (kET ). Many of the elementary rate constants for the reaction cycle have been directly measured at 20 C and derived from the steady state kinetic parameters and equations determined at 20 and 37 C (Table 2) [141]. Missing from Table 2 are rate constants determined or estimated for hydride transfer from NADH kH , electron transfer (kET ) and product release (kP ) catalyzed by E. coli flavoHb. The respective values determined for 20 C are 15 M−1 s−1 , 150 s−1 , and >200 s−1 [141]. FlavoHbs show varying specificities and capacities for NADH and NADPH utilization, but NADH appears preferred showing >10-fold lower KM values [142]. From these constants, it is apparent that electron transfer (kET ) is near-limiting for NOD turnover (90 s−1 at 20 C) with O2 , NADH, and NO saturation. Another factor potentially limiting maximal turnover is NO inhibition as NO saturates. For deducing a function for other Hbs as NODs, it is informative to compare the elementary rate constants of flavoHbs, Hbs, and Mb with those influencing the multistep reaction cycle (Fig. 6). O2 association rate constants and NO affinities are similar for the various Hbs (Table 2), suggesting similar capacities for the NOD reaction cycle when coupled to a reductase. However, O2 dissociation rate constants of sperm whale Mb and human red blood cell Hb are much larger than those of the flavoHb and HbN, and produce much lower O2 affinities. The consequence of larger O2 off-rates for the flavoHb-catalyzed reaction is NO inhibition because NO competes with O2 for the ferrous heme, producing an inactive enzyme (Fig. 6). Thus, a comparable 60-fold increase in the kO2 for E. coli flavoHb causes profound NO inhibition at physiologically relevant NO and O2 concentrations [141]. Thus, enzymatic NOD activities of Hbs and Mbs with large O2 off-rates are expected to be highly susceptible to NO binding and inhibition. Only high O2 concentrations and low NO fluxes would circumvent this inherent limit on a NOD function, and this may be achieved by simple mass action of abundant MbFeII O2 or HbFeII O2 in fully oxygenated red blood cells or myocytes. Dissociation of NO from flavoHbFeII NO, HbFeII NO or MbFeII NO, albeit slow (kNO = <0 0002 s−1 ), ultimately

310

P.R. Gardner and A.M. Gardner INACTIVE

(FADH•)FeII•NO k'•NO

NADH

•NO k'H

(FAD)FeIII

(FADH2)FeIII

kET

k•NO •NO

(FADH•)FeII

–

NO3

O2

kP O2

–

(FAD)FeIII NO3

k'O2

kO2 [(FAD)FeIII – O• •NO2]

(FADH•)FeIII – OO• •NO

kIS

k'NOD

(FAD)FeIII – OONO (FADH•)FeIII – OONO •NO

k'NOD

kIS [(FADH•)FeIII – O• •NO2]

(FAD)FeIII – OO• k'O2 O2

kO2 O2

(FAD)FeII •NO k•NO

–

(FADH•)FeIII NO3 kP kET

–

NO3

(FADH•)FeIII

•NO k'•NO

(FAD)FeII •NO INACTIVE

Fig. 6. FlavoHb-catalyzed NOD reaction cycle. NOD catalysis can be broken down into the fundamental reactions and rate constants. For the forward reaction, these are hydride transfer (kH ),  electron transfer (kET ), O2 association (kO 2 ), NO dioxygenation (kNOD ), peroxynitrite isomerization  (kIS ), and nitrate release (kP ). NO binds ferrous heme with a large association rate constant (kNO ), competes with O2 binding, and inhibits the NOD reaction cycle.

results in NO decomposition via NO dioxygenation [263]. Alternatively, the NOR activity of flavoHbs (Vmax >0 02 s−1 ) and Mb [264] may effectively release NO and the inhibition of the NO dioxygenation reaction.  and low kNO values relative to kO 2 and kO2 determined for flavoHb The high kNO (Table 2) [141,142] led Hausladen et al. to later refute the proposed NOD mechanism for NO metabolism and suggest an alternative denitrosylase mechanism [265]. In this mechanism, NO binds ferrous heme and O2 reacts with the flavoHbFeII NO intermediate

Nitric Oxide Dioxygenase

311

or resonance species flavoHbFeIII− NO to form nitrate. There are several reasons why a denitrosylase mechanism is untenable. First, NO inhibition was observed at low O2 levels [141,142]. Second, inducible NORs supplant flavoHbs under these unfavorable conditions (vide supra) [78,98]. Third, the partially inhibited NO dioxygenation reaction with bimolecular rate constants of 0.8–2.9 × 109 M−1 s−1 would far exceed the low denitrosylase reaction rate that can be expected from the millionfold smaller rate constant estimated for the reaction of O2 with MbFeII NO (<0.5 M−1 s−1 ) [266]. In fact, Herold et al. have convincingly argued that in the reaction of nitrosyl-Hb with O2 , NO first dissociates from HbFeII NO to allow O2 binding and NO dioxygenation [263]. Fourth, various flavoHbs and Hbs are induced by hypoxia [44,110,116,267] and NO, which mitigates the inherent problem of NO inhibition at low [O2 ] and high NO fluxes. Lastly, rate constants for NO dissociation are routinely determined under conditions in which dithionite reduction must compete with NO rebinding, which may produce artifactually low dissociation constants.

3.5. Conservations for O2 Activation and Heme Reduction In addition to conservation in structure for O2 binding and a NOD function, flavoHbs and Hbs show remarkable conservations in proximal histidine architecture and in the structure for electron transfer that appear important for a NOD function. A comparison of the A. eutrophus (panel A, Fig. 7), E. coli (panel B), and Vitreoscilla sp. (panel C) proximal heme structures reveals a hydrogen bond network with proximal His85(F8) formed by highly conserved glutamate 137/135 and tyrosine 95 residues (Fig. 4, top). It has been suggested that this triad uniquely controls the O2 bond, and O2 affinity, by providing an electronic push from iron to O2 and other ligands [42,208,268]. Indeed, Mukai et al. and Bonamore et al. have reported peroxidase-like infrared stretching frequencies for His(F8)–Fe, Fe–CO, and C–O for E. coli flavoHb indicative of electron enrichment and have argued for an O2 -activated peroxidase-like function [226,227]. In light of the evidence for a NOD function for these two flavoHbs and the Vitreoscilla Hb, it is more likely that this novel structure adapted for optimal FeIII− OO• formation or peroxynitrite isomerization (Fig. 5). Indeed, an infrared signature band for FeIII− OO• has been reported for the Vitreoscilla Hb at 1134 cm−1 [269] within the range of that for ferric-superoxide complexes and other HbO2 s [235]. Furthermore, Vitreoscilla Hb shows v C–O frequencies remarkably similar to those seen with E. coli flavoHb, with a strong band at 1964 cm−1 and weaker band at ∼1906 cm−1 [269]. In the truncated Hb subfamily, hydrogen bond interactions of Met77 with the proximal His81(F8) may provide a similar, albeit smaller, electron enrichment to iron and iron-bound O2 [218]. In muscle Mb, interactions of the proximal heme ligand His(F8) are also thought to be important for the peroxynitrite isomerization mechanism. Blomberg et al. have suggested that the hydrogen bond chain formed between the protonated His(F8) imidazole, Ser92 hydroxyl, and heme propionate A carboxyl decreases the strength of the imidazole-iron bond, strengthens the iron-O2 bond, and weakens the O O bond to rupture, thus facilitating the peroxynitrite isomerization mechanism [253]. Unique proximal pocket effects on the His(F8)-heme-O2 interaction are likewise envisaged for Lba [270], suggesting

312

P.R. Gardner and A.M. Gardner (B)

(A)

Heme Heme

C8

H2O

His85

C8

H2O

His85

Tyr95 Lys84

Tyr95

Lys84

FAD

FAD

H2O Lys80

Glu388 Glu394 Tyr190

Asn80

Glu137

Glu135

Tyr188

(C) Heme H2O

His85

Lys84

Tyr95

H2O Lys80

Glu137

Fig. 7. Conserved proximal heme structure in (A) A. eutrophus flavoHb (1CQX) [42], (B) E. coli flavoHb (1GVH) [208], and in (C) the single-domain Vitreoscilla sp. Hb-azide (2VHB) [202] X-ray structures. The respective O, N, C, and Fe atom colors are red, blue, grey, and yellow. The adenine nucleoside portion of FAD is colored pink and the P-atom in phosphate orange. (see Plate 7.)

the evolution of multiple mechanisms for achieving proper heme-O2 electrochemistry for the dioxygenation reaction. Ermler et al. suggested two possible paths for electron transfer from the FAD isoalloxazine ring to the heme propionate involving water [42]. The ammonium group of Lys84 interacts with a water molecule between C8 methyl group of the isoalloxazine ring and the heme propionate in the A. eutrophus flavoHb structure (Fig. 7A) that may serve in electron transfer. A similar Lys84 bonded water is observed in the other structures, although there are clear differences in the orientation of the ammonium group (Fig. 7, compare panels A–C). Lys84 and Glu394 form a salt bridge in A. eutrophus flavoHb that connects the isoalloxazine ring to a more distant water bonded to Tyr190 and Asn80 that may serve as an alternate path for electron flow to the heme proprionate. However, there is less conservation in this potential path with Lys80 substituting for Asn80 in the E. coli flavoHb and Vitreoscilla Hb structures. Moreover, E. coli flavoHb Glu388 assumes a different orientation than the salt bridging Glu394 in A. eutrophus flavoHb and is entirely lacking in the Vitreoscilla Hb presumably being supplied by a NADH-dependent FAD-containing reductase [271].

Nitric Oxide Dioxygenase

313

3.6. Hb and Mb Reduction As previously noted, the lack of a suitable reductase or reductant to drive the NOD reaction cycle (Fig. 5, reaction 1) would diminish the capacity of various Hbs and Mbs to serve as NODs. Moreover, the various reductases and reductants for the various oxidized Hbs and Mbs remain poorly defined. Within red blood cells, NADH-dependent cytochrome b5 reductase and cytochrome b5 clearly serve an important role in Hb reduction. Within muscle, several mechanisms for MbFeIII reduction including a cytochrome b5 pathway have been suggested [272,273]. Within legume root nodules, ascorbate (vitamin C) and flavin-containing reductases participate in ferric Lba reduction [274,275]. A role for ascorbate in ns plant Hb reduction is also suggested [276]. A similar role for abundant ascorbate in Mb, Ngb, and Cygb reduction in mammalian tissues is likely. It is noteworthy that within mammalian cells, the microsomal membrane-bound NADPHcytochrome P450 reductase functions as a reductase for a NOD activity with flavoHb-like properties [58]. Reductases for trHbN, Vitreoscilla Hb [271] and other Hbs remain largely undefined. The ferredoxin reductase-ferredoxin system acts as a fairly efficient HbFeIII and MbFeIII reductase in vitro [277], is well distributed in nature, and is thus a good candidate reductase for single-domain Hbs. Moreover, ferredoxin reductase is structurally related to the reductase domain of flavoHb and the NADPH-cytochrome P450 reductase (Fig. 1) [1], thus suggesting a conserved HbFeIII reductase function.

4. EVOLUTION OF Hb FUNCTION A primal NOD function for the Hb superfamily is suggested given that the Hb domain of flavoHb and other Hbs originated in unicellular organisms ∼1.8 billion years ago [1,15,16] around the time of oxygenic photosynthesis and elevated atmospheric O2 . In this situation, there was arguably little advantage for an O2 transport or storage function [18,207,278], but an apparent need for a NOD function because of the muiltiple sources for NO formation and targets for NO poisoning. In this regard, it is noteworthy that Hb is not the only O2 transport and storage protein to have apparently evolved from a dioxygenase, oxygenase, or oxidase function [279]. The unrelated heme-containing abalone “myoglobins” apparently formed from indoleamine 2,3-dioxygenase [280]. The binuclear nonheme iron-containing invertebrate O2 storage and transport protein hemerythrin likely evolved from related O2 -binding oxygenases [281]. Similarly, the copper-containing O2 storage and transport arthropod hemocyanin evolved from a phenoloxidase or related O2 -activating enzyme [279,282,283]. Significant changes in the structure, chemistry, and expression of Hbs have accompanied the adaptation of a NOD function to an O2 transport and storage function. Foremost, structural features influencing O2 binding, release, and activation have adapted for these diverse, but complementary functions. O2 storage and transport Hbs and Mbs have fine-tuned O2 affinity for uptake, storage, and release in tissues for respiration [209,211,223,284], and moreover, achieve high levels of expression for high capacity O2 storage and delivery. The O2 dissociation rate constant (kO2 ) for muscle Mb is 20–75-fold larger than those of NODs (flavoHbs) (Table 2). A lower O2 affinity facilitates O2 dissociation and delivery for mitochondrial respiration. On the other hand, the NOD function of microbial flavoHbs benefits from a high O2 affinity and low O2

314

P.R. Gardner and A.M. Gardner

dissociation rate constant. Low O2 dissociation rate constants increase the fraction of O2 bound to the ferrous heme and decrease the susceptibility of the NOD reaction cycle to NO binding and inhibition (Fig. 6) [141]. The distal His(E7) found in O2 transport and storage Hbs and the distal Tyr(B10) found in NOD-like flavoHbs and Hbs play key roles in controlling O2 affinity and function by hydrogen bonding O2 in the distal pocket, and thus appear to differentiate these two primary functions. However, ns plant Hbs, which are hexacoordinate Hbs with distal His(E7) (Fig. 3) showing a high O2 affinity [125], clearly provide an exception to this corollary [116,119]. More subtle features including the distribution of electrons in the heme-O2 complex, size, and dielectric effects of residues occupying the distal pocket, and the presence of channels may influence the   accessibility of NO kNO , the reactivity of NO kNOD , the stability and isomerization of the FeIII− OONO intermediate (kIS ), the release of the product nitrate (kP ), and ultimately the NOD function. These properties were inevitably altered during the adaptation of Hb to an O2 storage and transport function. In addition, coupling of heme reduction to NO-mediated oxidation is critical for maximal turnover of the NOD catalytic cycle (Fig. 6). In this regard, Hb within red blood cells is a relatively poor catalyst of NO dioxygenation. Sustained NO metabolic rates of red blood cells with 1 M NO are remarkably low [57], presumably because of limiting NADH-cytochrome b5 (metHb) reductase activity and the susceptibility of Hb to NO binding and inhibition. Clearly, NO dioxygenation is a remnant activity that acts ancillary to the O2 transport and storage function of the modern red blood cell Hb and muscle Mb [284]. In some cases, NO-induced oxidation or nitrosylation of Hb or Mb, as is frequently observed in nitrite-induced methemoglobinemia or following elevated NO production in sepsis or shock, may impair the primary function of Mb and Hb in O2 transport.

5. INHIBITORS AND APPLICATIONS OF THE NOD REACTION Inhibitors of NOD continue to serve as valuable tools for both the characterization of NO metabolic enzymes in various organisms and the interogation of mechanism. In addition, given the ability of NODs (flavoHbs and Hbs) to defend microbes from NO toxicity (vide supra) and the existence of a similar activity in various mammalian cells [57,58], mechanistic inhibitors of NODs are being investigated for their potential application as (i) antibiotics, (ii) antitumor agents, and (iii) vasorelaxants. Several heme ligands have been shown to be particularly effective in inhibiting the flavoHb-NOD and the mammalian NOD activity. Cyanide inhibits/inactivates microbial (flavoHb) NODs and the mammalian microsomal NOD at low micromolar concentrations [40,41,57], suggesting a reversible inhibition involving the high-affinity binding of cyanide to the ferric heme [208]. In addition, inactivation may occur through covalent modifation of the oxidation-, halogenation-, and nitration-sensitive  (or   ) meso carbon in the iron-protoporhyrin [285] by an activated cyanyl radical (unpublished results) through a mechanism similar to that described for peroxidases and cytochrome c oxidase [286]. Characteristic cyanide inactivation has been used to decipher involvement of heme-dependent NODs in cellular NO metabolism [149] and may be used to identify short-lived intermediates in the NOD mechanism.

Nitric Oxide Dioxygenase

315

0.30

4

(A)

0.20

(B) 4

0.25 0.15 0.20

1/ v (s)

3 0.15

3

0.10

2

2 0.10

1

0.05 0.05

1

0.00

0.00 0

0.05

1/[O2] (μ M–1)

0.1

0

5

1/[• NO]

10

(μ M–1)

Fig. 8. Effect of CO on the O2 and NO dependence of the E. coli flavoHb-NOD. NOD activity of flavoHb was measured at various O2 concentrations (Panel A) with 0.4 M NO and 0 M CO (line 1), 0.5 M CO (line 2), 1.0 M CO (line 3), and 2.0 M CO (line 4) or at various NO concentrations (Panel B) with 20 M O2 and 0 M CO (line 1), 0.5 M CO (line 2), 1.0 M CO (line 3), and 2.0 M CO (line 4). Reactions were at 37 C in 100 mM NaPi buffer, pH 7.0 with 100 M NADH and 1 M FAD [141,142].

In addition, CO shows high affinity for the ferrous hemes of flavoHbs with dissociation equillibrium constants of less than 0.7 M and shows strong competitive inhibition of NOD activities with respect to O2 (Ki = ∼1 M) [142] (Fig. 8A). CO similarly inhibits the NOD activity in mammalian cells (Ki = ∼3 M) [57,58], suggesting a flavoHb-like mechanism for that activity. The competition of CO with O2 also provides incontrovertible evidence for a dioxygenase mechanism and is inconsistent with a denitrosylase mechanism [265] (Fig. 8A). Thus, in a denitrosylase mechanism, CO would be expected to show competitive inhibition with respect to NO especially at lower [O2 ], but this is not observed. CO is uncompetitive with respect to [NO] over a range of [O2 ], including the physiologically relevant concentration of 20 M (Fig. 8B). Heme-binding antifungal imidazoles were also found to potently inhibit flavoHbNODs. Miconazole, econazole, clotrimazole, and ketoconazole (Fig. 9) show respective apparent Ki values of 80, 550, 1300, and 5000 nM with the E. coli NOD (flavoHb) [287,148]. A similar pattern of uncompetitive inhibition is observed for A. eutrophus, S. cerevisiae, and C. albicans flavoHbs. Interestingly, a single chlorine atom in miconazole increases inhibition ∼sevenfold over that observed with econazole, suggesting specific interactions of phenyl group substituents within the conserved hydrophobic distal heme pocket. Antigungal triazoles, such as itraconazole (Fig. 9E) and fluconazole, were found to be much less effective. Evidence supports a mechanism involving imidazole binding and trapping of the ferric heme intermediate in the NOD reaction cycle (Fig. 6). Organisms lacking a NOR pathway and preferentially utilizing a NOD pathway for survival and virulence (Fig. 2) such as S. aureus [102], S. enterica [101], and Y. pestis

316

P.R. Gardner and A.M. Gardner Cl

Cl

(A)

Cl

(E)

Cl

Cl

N–CH2CHOCH2

N

(D)

O

N N

Cl

N

Cl

Cl

O

N O

N

O O

Cl O

(B)

N

N–CH2CHOCH2

N

Cl

N N N

Cl

C H3C

(C)

O N

O N

N

N

C

N

H3C Cl

CH3

Fig. 9. Structures of inhibitory heme-binding imidazoles miconazole (A), econazole (B), clotrimazole (C), ketoconazole (D), and the triazole itraconazole (E).

[109] serve as targets for NOD inhibition and antibiotic design. Imidazoles also inhibit the mammalian cell NOD [58], albeit less potently, suggesting that heme-binding imidazoles may also be engineered to specifically target NO metabolism and modulate NO functions in a variety of organisms substituting for NO modulation therapies employing NO delivery agents [288]. NO-scavenging red blood cell Hbs have long been considered potentially valuable therapeutics for disease conditions involving excesses of NO including sepsis and shock [289,290]. Improved modified Hbs are showing promise as therapeutics for these life-threatening conditions [291]. Modifications of Hb that support high NOD turnover rates (Equation 3) over simple stoichiometric nitrosylation (Equation 1) and single turnover NO dioxygenation may increase the utility and efficacy of these therapeutics.

Nitric Oxide Dioxygenase

317

In addition, Hb expression increases growth yields of bacteria, and strains with elevated (flavo)Hb-NOD activity are being considered for increasing yields of recombinant proteins and other biotechnology products [137,140]. Most notably, Monsanto Corp. recently applied for international patent protection for the use of transgenic plants expressing heterologous flavoHb to increase seedling and crop yields (PCT/US2006/017161).

6. OTHER ENZYMIC FUNCTIONS FOR (FLAVO)Hbs AND Mbs? E. coli flavoHb can reduce various alkyl hydroperoxides to alcohols with maximal turnover rates of 0.3–1.8 s−1 , suggesting an ancillary role in the detoxification of alkyl or lipid hydroperoxides [292]. Tightly bound and large phospholipids, diacylglycerophospho-ethanolamine and diacylglycerophospho-glycerol, within the distal heme pocket of A. eutrophus flavoHb [42,293], and the affinity of other flavoHbs for phospholipids [293,294], have also suggested roles for flavoHbs in lipid metabolism [292,294]. Nevertheless, it remains to be demonstrated that alkyl hydroperoxide reduction by flavoHb is important for the viability of E. coli or any other organism especially given the existence of more efficient and inducible alkylhydroperoxide reductases. Moreover, as pointed out by Ollesch et al. [293], structurally homologous flavoHbs do not bind phospholipids equally, suggesting that the binding of phospholipids is adventitious rather than functional. Evidence also suggests that some microbial Hbs, like their modern descendants, may be suited for O2 delivery or transfer [147,295]. Given their propensity to autooxidize [296], microbial flavoHbs and Hbs have also been proposed to function as alternative respiratory oxidases [297] or − O2 • generators [298–300]. The potential for myriad dioxygen-dependent functions for flavoHbs and Hbs is further suggested by the many nonspecific oxidative enzymatic activities of red cell Hb and muscle Mb and the peroxidative dehalogenation of phenols by the polychaete Amphitite ornata Hb [301]. Adventitious catalytic activities include the oxidation of phenols, catechols, hydroquinones, hydrazines, and nitrite [302], the peroxide-dependent epoxidation of styrenes [303,304], the dealkylation of phenoxazone ethers [305], the oxidation of arachadonic acid to form prostaglandins [306], the production or scavenging of − O2 • [296,307], and the peroxide-dependent hydroxylation, dealkylation, and oxidation of a variety of substrates [123,296,308,309]. Intriguingly, a single substitution of the conserved Mb His(E7) residue with alanine or aspartate produces a very efficient peroxynitrite isomerase [251], suggesting a simple route for the evolution of this function. Indeed, a peroxynitrite isomerase function has been recently proposed for Lba [56,310]. The abundant red blood cell Hb can also act, albeit inefficiently, as a nitrite reductase that yields NO [311,312]. Moreover, evidence supports a role for nitrite reduction in NOmediated hypoxic vasorelaxation [73,313]. Thus, in the absence of a NOD function, other oxidative enzymatic functions for low-abundance Hbs require a critical consideration. Identifying new functions for new and old members of the Hb superfamily is a challenging goal that is uniquely primed by an unsurpassed wealth of structural and chemical knowledge.

318

P.R. Gardner and A.M. Gardner

7. CONCLUSIONS AND PROSPECTIVE With greater knowledge of NO toxicity, signaling, and the many sources of NO exposure, the biological necessity and significance of inducible oxidative and reductive NO metabolic pathways in diverse organisms are being elucidated. Diverse Hbs, flavoHbs, and Mbs play a pivotal role in metabolizing NO and regulating NO homeostasis throughout the aerobic biosphere, efficiently catalyzing the dioxygenation of NO to nitrate without the release of toxic intermediates. The NOD mechanism and function of flavoHb and various members of the Hb superfamily in diverse organisms has been subsequently tested and largely confirmed, albeit, there has been an understandable reluctance to call flavoHbs and other Hbs NODs. Indeed, myriad potential functions and a lingering uncertainty of the mechanism of NO detoxification perpetuate a functional ambiguity for many Hbs. Several key chemical and structural properties appear to correlate with the NOD function of flavoHbs and Hbs. These include a high O2 affinity, a superoxide-like character of the bound dioxygen that provides a large NO dioxygenation rate constant, and an accessible hydrophobic heme pocket that allows for rapid NO entry and that facilitates and shields reactive intermediates from water and biomolecules. In addition, rapid reduction, essential for rapid catalysis and the release of (flavo)Hbs from NO inhibition, is achieved by an associated flavin-containing reductase showing a rapid and efficient electron transfer pathway. Knowledge of the enzymology of the Hb-catalyzed NO dioxygenation reaction is also providing insights into the development of inhibitors and therapeutics, and the discovery of similar NODs in humans and other organisms. Knowledge of structural-functional conservations will play an increasingly important role in predicting a NOD function from genomic information especially as flavoHbs and Hbs reveal adaptations to other enzymic functions. Finally, knowledge of the NOD reaction is providing insights into the oxidative enzymatic capacities of low-abundance Hbs foreseen by Keilin over 50 years ago [18]. FlavoHbs, Hbs, and Mbs can indeed be viewed as enzymes finely-tuned by structure, energy landscape, dynamics, and allostery [50]. Hbs, like the cytochrome P450s, nonheme iron enzymes, and copper enzymes [281,282,314] have a capacity for dioxygen binding, activation, and catalysis of reactions involving FeIII− O2 •, FeIII− OOH, FeIII− O•/FeIV O, or •+ FeIV O/FeV O. Hydroxylations, oxidations, peroxidations, epoxidations, peroxidative dealkylations are some of the oxidative reactions catalyzed by Hbs. These reactions may have evolved into catalytic functions for some Hbs. The unique capacity of various Hbs for stabilization of a − O2 •-like intermediate suggests that enzymic functions utilizing the superoxo-like reactivity of the bound O2 will be common for the Hb superfamily. The bound − O2 • may be particularly suited to terminating free radicals, or generating radicals [302], with a similar capacity as the radical-radical terminating and radical-generating reactions of free − O2 • [315,316]. However, in cells higher concentrations of − O2 • can be effectively achieved with HbFeIII− O2 • without the attendant cellular toxicity of free − O2 •. Moreover, reactions can be accomplished in the Hb pocket without the release of reactive and toxic intermediates. The NO dioxygenation reaction provides a prototype for this radical-scavenging enzymatic function. Another example is the scavenging of − O2 • by HbFeIII − O2 • in red blood cells forming O2 , H2 O2 and HbFeIII kcat = 4000 M−1 s−1  [307,317].

Nitric Oxide Dioxygenase

319

In addition, nonenzymatic functions for Hbs such as O2 -sensing are also suggested given the homology of various Hbs to the DNA-binding globin sensors found in various microbes [318] and the binding of metNgb to regulatory G-proteins [319,320].

ACKNOWLEDGMENTS This work was supported in part by grants from National Institutes of Health (R01 GM65090) and the American Heart Association (9730193N). We gratefully acknowledge our colleagues for their contributions, discussions, and suggestions during the course of this work. We are especially grateful to Professors John Olson and Irwin Fridovich for their tutelage and insightful discussions. We thank Arin Fletcher for her skillful technical assistance in generating the data for Fig. 8.

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26]

Zhu, H. and Riggs, A.F. (1992) Proc. Natl. Acad. Sci. USA, 89, 5015–5019. Hardison, R.C. (1996) Proc. Natl. Acad. Sci. USA, 93, 5675–5679. Hardison, R. (1998) J. Expt. Biol., 201, 1099–1117. Hardison, R. (1999) Am. Scientist., 87, 126–137. Moens, L., Vanfleteren, J., Van de Peer, Y. et al. (1996) Mol. Biol. Evol., 13, 324–333. Wittenberg, J.B. and Wittenberg, B.A. (1990) Ann. Rev. Biophys. Chem., 19, 217–241. Riggs, A.F. (1991) Am. Zoologist., 31, 535–545. Weber, R.E. and Vinogradov, S.N. (2001) Physiol. Rev., 81, 569–628. Wittenberg, J.B., Bolognesi, M., Wittenberg, B.A., and Guertin, M. (2002) J. Biol. Chem., 277, 871–874. Shikama, K. and Matsuoka, A. (2004) Crit. Rev. Biochem. Mol. Biol., 39, 217–259. Burmester, T., Weich, B., Reinhardt, S., and Hankeln, T. (2000) Nature, 407, 520–522. Burmester, T., Ebner, B., Weich, B., and Hankeln, T. (2002) Mol. Biol. Evol., 19, 416–421. Trent, J.T., III and Hargrove, M.S. (2002) J. Biol. Chem., 277, 19538–19545. Hankeln, T., Ebner, B., Fuchs, C. et al. (2005) J. Inorg. Biochem., 99, 110–119. Miranda, J.L., Maillett, D.H., Soman, J., and Olson, J.S. (2005) J. Biol. Chem., 280, 36754–36761. Vinogradov, S.N., Hoogewijs, D., Bailly, X. et al. (2005) Proc. Natl. Acad. Sci. USA, 102, 11385–11389. Vinogradov, S.N., Hoogewijs, D., Bailly, X. et al. (2006) BMC Evol. Biol., 6, 31. Keilin, D. (1953) Nature, 172, 390–393. Keilin, D. (1966) The History of Cell Respiration and Cytochrome. Cambridge: Cambridge University Press. Keilin, D. and Hartree, E.F. (1937) Nature, 139, 548. Barnard, R.D. (1937) J. Biol. Chem., 120, 177–191. Doyle, M.P. and Hoekstra, J.W. (1981) J. Inorg. Biochem., 14, 351–358. Yoshida, K., Kasama, K., Kitabatake, M. et al. (1980) Int. Arch. Occup. Environ. Health., 46, 71–77. Furchgott, R.F. and Zawadski, J.V. (1980) Nature, 288, 373–376. Arnold, W.P., Mittal, C.K., Katsuki, S., and Murad, F. (1977) Proc. Natl. Acad. Sci. USA, 74, 3203–3207. Ignarro, L.J., Byrns, R.E., Buga, G.M., and Wood, K.S. (1987) Circ. Res., 61, 866–879.

320

P.R. Gardner and A.M. Gardner

[27] Moncada, S., Herman, A.G., and Vanhoutte, P.M. (1987) Trends Pharmacol. Sci., 8, 365–368. [28] Hibbs, J.B., Jr., Vavrin, Z. and Taintor, R.R. (1987) J. Immunol., 138, 550–565. [29] Marletta, M.A., Yoon, P.S., Iyengar, R. et al. (1988) Biochemistry, 27, 8706–8711. [30] Kelm, M. and Schrader, J. (1990) Circ. Res., 66, 1561–1575. [31] Lancaster, J.R., Jr. (1994) Proc. Natl. Acad. Sci. USA, 91, 8137–8141. [32] Kelm, M. (1999) Biochim. Biophys. Acta, 1411, 273–289. [33] Deem, S. (2004) Free Rad. Biol. Med., 36, 698–706. [34] Rassaf, T., Feelisch, M., and Kelm, M. (2004) Free Rad. Biol. Med., 36, 413–422. [35] Doherty, D.H., Doyle, M.P., Curry, S.R. et al. (1998) Nature Biotech., 16, 672–676. [36] Dou, Y., Maillett, D.H., Eich, R.F., and Olson, J.S. (2002) Biophys. Chem., 98, 127–148. [37] Wade, R.S. and Castro, C.E. (1996) Chem. Res. Tox., 9, 1382–1390. [38] Eich, R.F., Li, T., Lemon, D.D. et al. (1996) Biochemistry, 35, 6976–6983. [39] Gardner, P.R., Costantino, G., Szabó, C., and Salzman, A.L. (1997) J. Biol. Chem., 272, 25071–25076. [40] Gardner, P.R., Costantino, G., and Salzman, A.L. (1998) J. Biol. Chem., 273, 26528–26533. [41] Gardner, P.R., Gardner, A.M., Martin, L.A., and Salzman, A.L. (1998) Proc. Natl. Acad. Sci. USA, 95, 10378–10383. [42] Ermler, U., Siddiqui, R.A., Cramm, R., and Friedrich, B. (1995) EMBO J., 14, 6067–6077. [43] Hausladen, A., Gow, A.J., and Stamler, J.S. (1998) Proc. Natl. Acad. Sci. USA, 95, 14100–14105. [44] LaCelle, M., Kumano, M., Kurita, K. et al. (1996) J. Bacteriol., 178, 3803–3808. [45] Poole, R.K., Anjum, M.F., Membrillo-Hernández, J. et al. (1996) J. Bacteriol., 178, 5487–5492. [46] Crawford, M.J. and Goldberg, D.E. (1998) J. Biol. Chem., 273, 12543–12547. [47] Cramm, R., Siddiqui, R.A., and Friedrich, B. (1994) J. Biol. Chem., 269, 7349–7354. [48] Baumgärtner, M., Koschorreck, M., and Conrad, R. (1996) FEMS Microbiol. Ecol., 19, 165–170. [49] Flögel, U., Merx, M.W., Gödecke, A. et al. (2001) Proc. Natl. Acad. Sci. USA, 98, 735–740. [50] Frauenfelder, H., McMahon, B.H., Austin, R.H. et al. (2001) Proc. Natl. Acad. Sci. USA, 98, 2370–2374. [51] Brunori, M., Giuffré, A., Nienhaus, K. et al. (2005) Proc. Natl. Acad. Sci. USA, 102, 8483–8488. [52] Gardner, P.R. and Gardner, A.M. (2005) 12th Annual Meeting of the Society for Free Radical Biology and Medicine, Austin, TX 2005, Free Rad. Biol. Med., 39, S100. [53] Dordas, C., Hasinoff, B.B., Rivoal, J., and Hill, R.D. (2004) Planta, 219, 66–72. [54] Igamberdiev, A.U. and Hill, R.D. (2004) J. Exp. Bot., 55, 2473–2482. [55] Igamberdiev, A.U., Baron, K., Mana’H-Little, N. et al. (2005) Ann. Bot., 96, 557–564. [56] Herold, S. and Puppo, A. (2005) J. Biol. Inorg. Chem., 11, 1–11. [57] Gardner, P.R., Martin, L.A., Hall, D., and Gardner, A.M. (2001) Free Rad. Biol. Med., 31, 191–204. [58] Hallstrom, C.K., Gardner, A.M., and Gardner, P.R. (2004) Free Rad. Biol. Med., 37, 216–228. [59] Huie, R.E. and Padmaja, S. (1993) Free Rad. Res. Commun., 18, 195–199. [60] Pryor, W.A. and Squadrito, G.L. (1995) Am. J. Physiol., 268, L699–L722. [61] Patel, R.P., McAndrew, J., Sellak, H. et al. (1999) Biochim. Biophys. Acta, 1411, 385–400. [62] Radi, R. (2004) Proc. Natl. Acad. Sci. USA, 101, 4003–4008. [63] Liochev, S.I. and Fridovich, I. (2002) Free Rad. Biol. Med., 33, 137–141. [64] Ouellet, H., Ouellet, Y., Richard, C. et al. (2002) Proc. Natl. Acad. Sci. USA, 99, 5902–5907. [65] Poole, R.K. (2005) Biochem. Soc. Trans., 33, 176–180. [66] Hibbs, J.B., Jr. (2002) J. Infect. Dis., 185, S9–17.

Nitric Oxide Dioxygenase

321

[67] Nathan, C. and Shiloh, M.U. (2000) Proc. Natl. Acad. Sci. USA, 97, 8841–8848. [68] Durner, J., Wendehenne, D., and Klessig, D.F. (1998) Proc. Natl. Acad. Sci. USA, 95, 10328–10333. [69] Dedon, P.C. and Tannenbaum, S.R. (2004) Arch. Biochem. Biophys., 423, 12–22. [70] Boccara, M., Mills, C.E., Zeier, J. et al. (2005) Plant J., 43, 226–237. [71] Rockel, P., Strube, F., Rockel, A. et al. (2002) J. Exp. Bot., 53, 103–110. [72] Zweier, J.L., Samouilov, A., and Kuppusamy, P. (1999) Biochim. Biophys. Acta, 1411, 250–262. [73] Gladwin, M.T., Schechter, A.N., Kim-Shapiro, D.B. et al. (2005) Nature Chem. Biol., 1, 308–314. [74] Zumft, W. (1997) Microbiol. Mol. Biol. Rev., 61, 533–616. [75] Cutruzzolà, F. (1999) Biochim. Biophys. Acta, 1411, 231–249. [76] Drapier, J.-C. and Hibbs, J.B., Jr. (1996) Meth. Enzymol., 269, 26–36. [77] Navarre, D.A., Wendehenne, D., Durner, J. et al. (2000) Plant Physiol., 122, 573–582. [78] Gardner, A.M., Helmick, R.A., and Gardner, P.R. (2002) J. Biol. Chem., 277, 8172–8177. [79] Brown, G.C. and Cooper, C.E. (1994) FEBS Lett., 356, 295–298. [80] Brunori, M., Guiuffre, A., Sarti, P. et al. (1999) Cell. Mol. Life Sci., 56, 549–557. [81] Stevanin, T.M., Ioannidis, N., Mills, C.E. et al. (2000) J. Biol. Chem., 275, 35868–35875. [82] Brunori, M. (2001) Trends Biochem. Sci., 26, 21–23. [83] Brunori, M., Giuffré, A., and Sarti, P. (2001) J. Inorg. Biochem., 99, 324–336. [84] Borisov, V.B., Forte, E., Konstantinov, A.A. et al. (2004) FEBS Lett., 576, 201–204. [85] Cooper, C.E. (1999) Biochim. Biophys. Acta, 1411, 290–309. [86] Henry, Y. and Guissani, A. (1999) Cell. Mol. Life Sci., 55, 1003–1014. [87] Radi, R. (1996) Chem. Res. Tox., 9, 828–835. [88] Miranda, K.M., Espey, M.G., and Wink, D.A. (2000) J. Inorg. Biochem., 79, 237–240. [89] Brunelli, L., Crow, J.P., and Beckman, J.S. (1995) Arch. Biochem. Biophys., 316, 327–334. [90] Beckman, J.S. and Koppenol, W.H. (1996) Am. J. Physiol., 271, C1424–C1437. [91] De Groote, M.A., Granger, D., Xu, Y. et al. (1995) Proc. Natl. Acad. Sci. USA, 92, 6399–6403. [92] Hausladen, A., Privalle, C.T., Keng, T. et al. (1996) Cell, 86, 719–729. [93] Membrillo-Hernández, J., Coopamah, M.D., Channa, A. et al. (1998) Mol. Microbiol., 29, 1101–1112. [94] Fang, F.C. (1997) J. Clin. Invest., 99, 2818–2825. [95] Liu, L., Hausladen, A., Zeng, M. et al. (2001) Nature, 410, 490–493. [96] Bryk, R., Griffin, P., and Nathan, C. (2000) Nature, 407, 211–215. [97] de Jesús-Berrîos, M., Liu, L., Nussbaum, J.C. et al. (2003) Curr. Biol., 13, 1963–1968. [98] Gardner, A.M. and Gardner, P.R. (2002) J. Biol. Chem., 277, 8166–8171. [99] Liu, L., Zeng, M., Hausladen, A. et al. (2000) Proc. Natl. Acad. Sci. USA, 97, 4672–4676. [100] Ullmann, B.D., Myers, H., Chiranand, W. et al. (2004) Euk. Cell, 3, 715–723. [101] Bang, I.-S., Liu, L., Vazquez-Torres, A. et al. (2006) J. Biol. Chem., 281, 28039–28047. [102] Richardson, A.R., Dunman, P.M., and Fang, F.C. (2006) Mol. Microbiol., 61, 927–939. [103] Moore, C.M., Nakano, M.M., Wang, T. et al. (2004) J. Bacteriol., 186, 4655–4664. [104] Membrillo-Hernández, J., Coopamah, M.D., Anjum, M.F. et al. (1999) J. Biol. Chem., 274, 748–754. [105] Mukhopadhyay, P., Zheng, M., Bedzyk, L.A. et al. (2004) Proc. Natl. Acad. Sci. USA, 101, 745–750. [106] Arai, H., Hayashi, M., Kuroi, A. et al. (2005) J. Bacteriol., 187, 3960–3968. [107] Justino, M.C., Vicente, J.B., Teixeira, M., and Saraiva, L.M. (2005) J. Biol. Chem., 280, 2636–2643. [108] Eriksson, S., Lucchini, S., Thompson, A. et al. (2003) Mol. Microbiol., 47, 103–118. [109] Sebbane, F., Lemaitre, N., Sturdevant, D.E. et al. (2006) Proc. Natl. Acad. Sci. USA, 103, 11766–11771.

322

P.R. Gardner and A.M. Gardner

[110] Dikshit, K.L., Spaulding, D., Braun, A., and Webster, D.A. (1989) J. Gen. Microbiol., 135, 2601–2609. [111] Frey, A.D., Koskenkorva, T., and Kallio, P.T. (2003) FEMS Microbiol. Lett., 224, 127–132. [112] Iijima, M., Shimizu, H., Tanaka, Y., and Urushihara, H. (2000) Cell Struct. Funct., 25, 47–55. [113] Hernández-Urzúa, E., Mills, C.E., White, G.P. et al. (2003) J. Biol. Chem., 278, 34975–34982. [114] Elvers, K.T., Wu, G., Gilberthorpe, N.J. et al. (2004) J. Bacteriol., 186, 5332–5341. [115] Minning, D.M., Gow, A.J., Bonaventura, J. et al. (1999) Nature, 401, 497–502. [116] Dordas, C., Hasinoff, B.B., Igamberdiev, A.U. et al. (2003) Plant J., 35, 763–770. [117] Igamberdiev, A.U., Seregélyes, C., Manac’h, N., and Hill, R.D. (2004) Planta, 219, 95–102. [118] Dordas, C., Rivoal, J., and Hill, R.D. (2003) Ann. Bot., 91, 173–178. [119] Delledonne, M., Perazzolli, M., Dominici, P. et al. (2004) 12th Biennial Meeting of the Society for Free Radical Research, Buenos Aires, Argentina, Free Rad. Biol. Med., 36, S41. [120] Perazzolli, M., Dominici, P., Romero-Puertas, M.C. et al. (2004) Plant Cell, 16, 2785–2794. [121] Shimoda, Y., Nagata, M., Suzuki, A. et al. (2005) Plant Cell Physiol., 46, 99–107. [122] Ohwaki, Y., Kawaishi-Kobayashi, M., Wakasa, K. et al. (2005) Plant Cell Physiol., 46, 324–331. [123] Sakamoto, A., Sakurao, S., Fukunaga, K. et al. (2004) FEBS Lett., 572, 27–32. [124] Appleby, C.A. (1992) Sci. Prog., 76, 365–398. [125] Kundu, S., Trent, J.T., III, and Hargrove, M.S. (2003) Trends Plant Sci., 8, 387–393. [126] Ott, T., van Dongen, J.T., Günther, C. et al. (2005) Curr. Biol., 15, 531–535. [127] Appleby, C.A., Turner, G.L., and Macnicol, P.K. (1975) Biochim. Biophys. Act, 387, 461–474. [128] Meyer, J. (1981) Arch. Biochem. Biophys., 210, 246–256. [129] Trinchant, J.C. and Rigaud, J. (1982) Appl. Environ. Microbiol., 44, 1385–1388. [130] Hyman, M.R., Seefeldt, L.C., Morgen, T.V. et al. (1992) Biochemistry, 31, 2947–2955. [131] Andersson, C.R., Ostergaard, E., Llewellyn, D.J. et al. (1996) Proc. Natl. Acad. Sci. USA, 93, 5682–5687. [132] Perazzolli, M., Romero-Puertas, M.C., and Delledonne, M. (2006) J. Exp. Bot., 57, 479–488. [133] Kaur, R., Pathania, R., Sharma, V. et al. (2002) Appl. Environ. Microbiol., 68, 152–160. [134] Pathania, R., Navani, N.K., Gardner, A.M. et al. (2002) Mol. Microbiol., 45, 1303–1314. [135] Frey, A.D., Farrés, J., Bollinger, C.J.T., and Kallio, P.T. (2002) Appl. Environ. Microbiol., 68, 4835–4840. [136] Frey, A.D., Oberle, B.T., Farrés, J., and Kallio, P.T. (2004) Plant Biotech. J., 2, 221–231. [137] Frey, A.D. and Kallio, P.T. (2003) FEMS Microbiol. Rev., 27, 525–545. [138] Fabozzi, G., Ascenzi, P., Di Renzi, S., and Visca, P. (2006) Microb. Pathog., 40, 211–220. [139] Lama, A., Pawaria, S. and Dikshit, K.L. (2006) FEBS Lett., 580, 4031–4041. [140] Frey, A.D. and Kallio, P.T. (2005) Trends Biotech., 23, 69–73. [141] Gardner, A.M., Martin, L.A., Gardner, P.R. et al. (2000) J. Biol. Chem., 275, 12581–12589. [142] Gardner, P.R., Gardner, A.M., Martin, L.A. et al. (2000) J. Biol. Chem., 275, 31581–31587. [143] Ouellet, H., Juszczak, L., Dantsker, D. et al. (2003) Biochemistry, 42, 5764–5774. [144] Olson, J.S., Foley, E.W., Rogge, C. et al. (2004) Free Rad. Biol. Med., 36, 685–697. [145] Herold, S., Exner, M., and Nauser, T. (2001) Biochemistry, 40, 3385–3395. [146] Herold, S. (1999) FEBS Lett., 443, 81–84. [147] Ascenzi, P., Bocedi, A., Bolognesi, M. et al. (2006) Biochem. Biophys. Res. Commun., 339, 450–456. [148] Helmick, R.A., Fletcher, A.E., Gardner, A.M. et al. (2005) Antimicrob. Agents and Chemother., 49, 1837–1843. [149] Gardner, P.R., Gardner, A.M., and Hallstrom, C.K. (2004) Nitric Oxide Protocols. In Dioxygen-Dependent Metabolism of Nitric Oxide, Vol. 279 (A. Hassid, ed.) Totowa, NJ: Humana Press, pp. 133–150.

Nitric Oxide Dioxygenase [150] [151] [152] [153] [154] [155] [156] [157] [158] [159] [160] [161] [162] [163] [164] [165] [166] [167] [168] [169] [170] [171] [172] [173] [174] [175] [176] [177] [178] [179] [180] [181] [182] [183] [184] [185] [186] [187] [188] [189] [190]

323

Firoved, A.M. and Deretic, V. (2003) J. Bacteriol., 185, 1071–1081. Firoved, A.M., Wood, S.R., Ornatowski, W. et al. (2004) J. Bacteriol., 186, 4046–4050. Favey, S., Labesse, G., Vouille, V., and Boccara, M. (1995) Microbiology, 141, 863–871. Stevanin, T.M., Poole, R.K., Demoncheaux, E.A.G., and Read, R.C. (2002) Infect. Immun., 70, 4399–4405. Hromatka, B.S., Noble, S.M., and Johnson, A.D. (2005) Molec. Biol. Cell, 16, 4814–4826. Taylor-Robinson, A.W. and Looker, M. (1998) Lancet, 351, 1630. Ascenzi, P., Salvati, L., and Brunori, M. (2001) FEBS Lett., 501, 103–105. Ascenzi, P., Bocedi, A., and Gradoni, L. (2005) IUBMB Life, 57, 689–691. Kim, Y.-M., Hong, S.-J., Billiar, T.R., and Simmons, R.L. (1996) Infect. Immun., 1996, 3074–3080. Householder, T.C., Fozo, E.M., Cardinale, J.A., and Clark, V.L. (2000) Infect. Immun., 68, 5241–5246. Anjum, M.F., Stevanin, T.M., Read, R.C., and Moir, J.W.B. (2002) J. Bacteriol., 184, 2987–2993. Nakahara, K., Tanimoto, T., Hatano, K.-I. et al. (1993) J. Biol. Chem.,268, 8350–8355. Zumft, W.G. (2005) J. Inorg. Biochem., 99, 194–215. Daiber, A., Shoun, H., and Ullrich, V. (2005) J. Inorg. Biochem., 99, 185–193. Zumft, W.G. (2002) J. Mol. Microbiol. Biotechnol., 4, 277–286. Goretski, J., Zafiriou, O.C. and Hollocher, T.C. (1990) J. Biol. Chem., 265, 11535–11538. Hutchings, M.I., Mandhana, N. and Spiro, S. (2002) J. Bacteriol., 184, 4640–4643. Gomes, C.M., Giuffrè, A., Forte, E. et al. (2002) J. Biol. Chem., 277, 25273–25276. Bleakley, B.H. and Tiedje, J.M. (1982) Appl. Environ. Microbiol., 44, 1342–1348. Mills, P.C., Richardson, D.J., Hinton, J.C.D., and Spiro, S. (2005) Biochem. Soc. Trans., 33, 198–199. Kim, S.O., Orii, Y., Lloyd, D. et al. (1999) FEBS Lett., 445, 389–394. Wu, G., Wainwright, L.M., and Poole, R.K. (2003) Adv. Microb. Physiol., 47, 255–310. Poole, R.K. and Hughes, M.N. (2000) Mol. Microbiol., 36, 775–783. Mills, C.E., Sedelnikova, S., Søballe, B. et al. (2001) Biochem. J., 353, 207–213. Rodionov, D.A., Dubchak, I.L., Arkin, A.P. et al. (2005) PLoS Comput. Biol., 1, 415–431. Hughes, M.N. (1999) Biochim. Biophys. Acta, 1411, 263–272. Sasaki, Y., Takaya, N., Nakamura, A., and Shoun, H. (2004) Biosc. Biotechnol. Biochem., 68, 1106–1112. Boon, E.M. and Marletta, M.A. (2005) J. Inorg. Biochem., 99, 892–902. Gödecke, A. (2006) Cardiovasc. Res., 69, 309–317. Thomas, D.D., Liu, X., Kantrow, S.P., and Lancaster, J.R., Jr. (2001) Proc. Natl. Acad. Sci. USA, 98, 355–360. Brunori, M. (2001) Trends Biochem. Sci., 26, 209–210. Gödecke, A., Molojavyi, A., Heger, J. et al. (2003) J. Biol. Chem., 278, 21761–21766. Wunderlich, C., Flögel, U., Gödecke, A. et al. (2003) Circ. Res., 92, 1352–1358. Wegener, J.W., Gödecke, A., Schrader, J., and Nawrath, H. (2002) Br. J. Pharmacol., 136, 415–420. Li, W., Jue, T., Edwards, J. et al. (2004) Am. J. Physiol., 286, H47–54. Gardner, A.M., Gessner, C.R., and Gardner, P.R. (2003) J. Biol. Chem., 278, 10081–10086. Vollack, K.-U. and Zumft, W.G. (2001) J. Bacteriol., 183, 2516–2526. D’Autréaux, B., Tucker, N.P., Dixon, R., and Spiro, S. (2005) Nature, 437, 769–772. Pohlmann, A., Cramm, R., Schmelz, K., and Friedrich, B. (2000) Mol. Microbiol., 38, 626–638. Justino, M.C., Gonçalves, V.M.M., and Saraiva, L.M. (2005) Biochem. Biophys. Res. Commun., 328, 540–544. Bodenmiller, D.M. and Spiro, S. (2006) J. Bacteriol., 188, 874–881.

324

P.R. Gardner and A.M. Gardner

[191] Nakano, M.M., Geng, H., Nakano, S., and Kobayashi, K. (2006) J. Bacteriol., 188, 5878–5887. [192] Nakano, M.M. (2002) J. Bacteriol., 184, 1783–1787. [193] Cruz-Ramos, H., Crack, J., Wu, G. et al. (2002) EMBO J., 21, 3235–3244. [194] Crawford, M.J. and Goldberg, D.E. (1998) J. Biol. Chem., 273, 34028–34032. [195] D’Autréaux, B., Touati, D., Bersch, B. et al. (2002) Proc. Natl. Acad. Sci. USA, 99, 16619–16624. [196] Nunoshiba, T., DeRojas-Walker, T., Wishnok, J.S. et al. (1993) Proc. Natl. Acad. Sci. USA, 90, 9993–9997. [197] Ding, H. and Demple, B. (2000) Proc. Natl. Acad. Sci. USA, 97, 5146–5150. [198] Kim, S.O., Merchant, K., Nudelman, R. et al. (2002) Cell, 109, 383–396. [199] Crawford, M.J. and Goldberg, D.E. (2006) J. Biol. Chem., 281, 3752. [200] Sarver, A., DeRisi, J. (2005) Molec. Biol. Cell, 16, 4781–4791. [201] Neill, S.J., Desikan, R., and Hancock, J.T. (2003) New Phytol., 159, 11–35. [202] Tarricone, C., Galizzi, A., Coda, A. et al. (1997) Structure, 5, 497–507. [203] Phillips, S.E.V. and Schoenborn, B.P. (1981) Nature, 292, 81–82. [204] Hargrove, M.S., Brucker, E.A., Stec, B. et al. (2000) Structure, 8, 1005–1014. [205] Milani, M., Pesce, A., Ouellet, Y. et al. (2001) EMBO J., 20, 3902–3909. [206] Shaanan, B. (1983) J. Mol. Biol., 171, 31–59. [207] Riggs, A. (1965) Physiol. Rev., 45, 619–673. [208] Ilari, A., Bonamore, A., Farina, A. et al. (2002) J. Biol. Chem., 277, 23735–23732. [209] Perutz, M.F. (1979) Ann. Rev. Biochem., 48, 327–386. [210] Rohlfs, R.J., Mathews, A.J., Carver, T.E. et al. (1990) J. Biol. Chem., 265, 3168–3176. [211] Olson, J.S. and Phillips, G.N., Jr. (1996) J. Biol. Chem., 271, 17593–17596. [212] Peterson, E.S., Huang, S., Wang, J. et al. (1997) Biochemistry, 36, 13110–13121. [213] De Baere, I., Perutz, M.F., Kiger, L. et al. (1994) Proc. Natl. Acad. Sci. USA, 91, 1594–1597. [214] Yang, J., Kloek, A.P., Goldberg, D.E., and Matthews, F.S. (1995) Proc. Natl. Acad. Sci. USA, 92, 4224–4228. [215] Couture, M., Yeh, S.-R., Wittenberg, B.A. et al. (1999) Proc. Natl. Acad. Sci. USA, 96, 11223–11228. [216] Mukai, M., Savard, P.-Y., Ouellet, H. et al. (2002) Biochemistry, 41, 3897–3905. [217] Yeh, S.-R., Couture, M., Ouellet, Y. et al. (2000) J. Biol. Chem., 275, 1679–1684. [218] Crespo, A., Marti, M.A., Kaslko, S.G. et al. (2005) J. Am. Chem. Soc., 127, 4433–4444. [219] Gardner, P.R., Gardner, A.M., Dou, Y., and Olson, J.S. (2001) 8th Annual Meeting of the Oxygen Society, Research Triangle Park, NC, Free Rad. Biol. Med., 31, S79. [220] Dikshit, K.L., Orii, Y., Navani, N. et al. (1998) Arch. Biochem. Biophys., 349, 161–166. [221] Gardner, P.R., Gardner, A.M., Brashear, W. et al. (2006) J. Inorg. Biochem., 100, 542–550. [222] Olson, J.S., Mathews, A.J., Rohlfs, R.J. et al. (1988) Nature, 336, 265–266. [223] Perutz, M.F. (1990) Ann. Rev. Physiol., 52, 1–25. [224] Giangiacomo, L., Mattu, M., Aercovito, A. et al. (2001) Biochemistry, 40, 9311–9316. [225] Gibson, Q.H., Regan, R., Elber, R. et al. (1992) J. Biol. Chem., 267, 22022–22034. [226] Mukai, M., Mills, C.E., Poole, R.K., and Yeh, S.-Y. (2001) J. Biol. Chem., 275, 7272–7277. [227] Bonamore, A., Chiancone, E., and Boffi, A. (2001) Biochim. Biophys. Acta., 1549, 174–178. [228] Kloek, A.P., Yang, J., Mathews, F.S. et al. (1994) J. Biol. Chem., 269, 2377–2379. [229] Farrés, J., Rechsteiner, M.P., Herold, S. et al. (2005) Biochemistry, 44, 4125–4134. [230] Trent, J.T., III, Watts, R.A., and Hargrove, M.S. (2001) J. Biol. Chem., 276, 30106–30110. [231] Hvitved, A.N., Trent, J.T., III, Premer, S.A., and Hargrove, M.S. (2001) J. Biol. Chem., 276, 34714–34721. [232] Weiss, J.J. (1964) Nature, 202, 83–84. [233] Weiss, J.J. (1964) Nature, 203, 182–183. [234] Potter, W.T., Tucker, M.P., Houtchens, R.A., and Caughey, W.S. (1987) Biochemistry, 26, 4699–4707.

Nitric Oxide Dioxygenase [235] [236] [237] [238] [239] [240] [241] [242] [243] [244] [245] [246] [247] [248] [249] [250] [251] [252] [253] [254] [255] [256] [257] [258] [259] [260] [261] [262] [263] [264] [265] [266] [267] [268] [269] [270] [271] [272] [273] [274] [275] [276] [277] [278] [279] [280]

325

Das, T.K., Couture, M., Ouellet, Y. et al. (2001) Proc. Natl. Acad. Sci. USA, 98, 479–484. Jensen, K.P., Roos, B.O., and Ryde, U. (2005) J. Inorg. Biochem., 99, 45–54. Blough, N.V. and Zafiriou, O.C. (1985) Inorg. Chem., 24, 3502–3504. Keith, W.G. and Powell, R.E. (1969) J. Chem. Soc. A, 90. Halfpenny, E. and Robinson, J. (1952) J. Chem. Soc., 928–938. Balavoine, G.G.A., Geletii, Y.V., and Bejan, D. (1997) Nitric Oxide, 1, 507–521. Halfpenny, E. and Robinson, J. (1952) J. Chem. Soc., 939–946. Anbar, M. and Taube, H. (1954) J. Am. Chem. Soc., 76, 6243–6247. Tsai, J.-H.M., Harrison, J.G., Martin, J.C. et al. (1994) J. Am. Chem. Soc., 116, 4115–4116. Beckman, J.S. (1996) Chem. Res. Tox., 9, 836–844. Koppenol, W.H., Moreno, J.J., Pryor, W.A. et al. (1992) Chem. Res. Tox., 5. Hughes, M.N. and Nicklin, H.G. (1968) J. Chem. Soc. A, 450–452. Jursic, B.S., Klasnic, L., Pecur, S., and Pryor, W.A. (1997) Nitric Oxide 1, 494–501. Bohle, D.S. and Hansert, B. (1997) Nitric Oxide 1, 502–506. Hvitved, A.N., Foley, E.W., Rogge, C. et al. (2005) 12th Annual Meeting of the Society for Free Radical Biology and Medicine, Austin, TX, Free Rad. Biol. Med., 39, S101. Herold, S. (1999) Arch. Biochem. Biophys., 372, 393–398. Herold, S., Matsui, T., and Watanabe, Y. (2001) J. Am. Chem. Soc., 123, 4085–4086. Goldstein, S., Merenyi, G., and Samuni, A. (2004) J. Am. Chem. Soc., 126, 15694–15701. Blomberg, L.M., Blomberg, M.R.A., and Siegbahn, P.E.M. (2004) J. Biol. Inorg. Chem., 9, 923–935. Gardner, P.R. (2005) J. Inorg. Biochem., 99, 247–266. Gunther, M.R. (2004) Free Rad. Biol. Med., 36, 1345–1354. Lee, J., Hunt, J.A., and Groves, J.T. (1998) J. Am. Chem. Soc., 120, 7493–7501. Stern, M.K., Jensen, M.P., and Kramer, K. (1996) J. Am. Chem. Soc., 118, 8735–8737. Herold, S. and Rehmann, F.-J.K. (2001) J. Biol. Inorg. Chem., 6, 543–555. Herold, S. and Rehmann, F.-J.K. (2003) Free Rad. Biol. Med., 34, 531–545. Ghosh, A. and Taylor, P.R. (2003) Curr. Opin. Chem. Biol., 7, 113–124. Bourassa, J.L., Ives, E.P., Marqueling, A.L. et al. (2001) J. Am. Chem. Soc., 123, 5142–5143. Herold, S., Shivashankar, K., and Mehl, M. (2002) Biochemistry, 41, 13460–13472. Herold, S. and Röck, G. (2005) Biochemistry, 44, 6223–6231. Immoos, C.E., Chou, J., Bayachou, M. et al. (2004) J. Am. Chem. Soc., 126, 4934–4942. Hausladen, A., Gow, A., and Stamler, J.S. (2001) Proc. Natl. Acad. Sci. USA, 98, 10108–10112. Arnold, E.V. and Bohle, D.S. (1996) Methods Enzymol., 269, 41–55. Probst, I., Wolf, G., and Schlegel, H.G. (1979) Biochim. Biophys. Acta, 576, 471–478. Labesse, G., Craescu, C.T., Mispelter, J. et al. (1998) Eur. J. Biochem., 253, 751–759. Choc, M.G., Webster, D.A., and Caughey, W.S. (1982) J. Biol. Chem., 257, 865–869. Kundu, S., Snyder, B., Das, K. et al. (2002) Proteins, 46, 268–277. Jakob, W., Webster, D.A., and Kroneck, P.M. (1992) Arch. Biochem. Biophys., 292, 29–33. Livingston, D.J., McLachlan, S.J., La Mar, G.N., and Brown, W.D. (1985) J. Biol. Chem., 260, 15699–15707. Liang, Z.-X., Jiang, M., Ning, Q., and Hoffman, B.M. (2002) J. Biol. Inorg. Chem., 7, 580–588. Becana, M., and Klucas, R.V. (1990) Proc. Natl. Acad. Sci. USA, 87, 7295–7299. Moreau, S., Puppo, A., and Davies, M.J. (1995) Phytochemistry, 39, 1281–1286. Igamberdiev, A.U., Bykova, N.V., and Hill, R.D. (2006) Planta, 223, 1033–1040. Hayashi, A., Suzuki, T., and Shin, M. (1973) Biochim. Biophys. Acta, 310, 309–316. Oshino, R., Oshino, N., and Chance, B. (1973) Eur. J. Biochem., 35, 23–33. Terwilliger, N.B. (1998) J. Exp. Biol., 201, 1085–1098. Suzuki, T. (1994) J. Prot. Chem., 13, 9–13.

326 [281] [282] [283] [284] [285] [286] [287]

[288] [289] [290] [291] [292] [293] [294] [295] [296] [297] [298] [299] [300] [301] [302] [303] [304] [305] [306] [307] [308] [309] [310] [311] [312] [313] [314] [315] [316] [317] [318] [319] [320]

P.R. Gardner and A.M. Gardner Solomon, E.I., Brunold, T.C., Davis, M.I. et al. (2000) Chem. Revs., 100, 235–349. Solomon, E.I., Chen, P., Metz, M. et al. (2001) Angew. Chem. Int. Ed., 40, 4570–4590. Burmester, T. (2002) J. Comp. Physiol. [B], 172, 95–107. Wittenberg, J.B. and Wittenberg, B. (2003) J. Exp. Biol., 206, 2011–2020. Treibs, A. (1971) Das Leben und Wirken von Hans Fischer. Munich: Hans FischerGesellschaft. Chen, Y.-R., Deterding, L.J., Tomer, K.B., and Mason, R.P. (2000) Biochemistry, 39, 4415–4422. Helmick, R.A., Fletcher, A.E., Gessner, C.R. et al. (2003) 10th Annual Meeting of the Society for Free Radical Biology and Medicine, Seattle, WA, Free Rad. Biol. Med., 35, S94. Janero, D.R. (2000) Free Rad. Biol. Med., 28, 1495–1506. De Angelo, J. (1999) Expert Opinion Pharmacotherapy, 1, 19–29. Privalle, C., Talarico, T., Keng, T., and DeAngelo, J. (2000) Free Rad. Biol. Med., 28, 1507–1517. Murakami, K., Privalle, C., Enkhbaatar, P. et al. (2003) Clin. Sci., 105, 629–635. Bonamore, A., Gentili, P., Ilari, A. et al. (2003) J. Biol. Chem., 278, 22272–22277. Ollesch, G., Kaunzinger, A., Juchelka, D. et al. (1999) Eur. J. Biochem. 262, 396–405. Bonamore, A., Farina, A., Gattoni, M., et al. (2003) Biochemistry, 42, 5792–5801. Pathania, R., Navani, N.K., Rajamohan, G., and Dikshit, K.L. (2002) J. Biol. Chem., 277, 15293–15302. Misra, H.P. and Fridovich, I. (1972) J. Biol. Chem., 247, 6960–6962. Dikshit, R.P., Dikshit, K.L., Liu, Y.X., and Webster, D.A. (1992) Arch. Biochem. Biophys., 293, 241–245. Anjum, M.F., Ioannidis, N., and Poole, R.K. (1998) FEMS Microbiol. Lett., 166, 219–223. Poole, R.K. (1994) Antonie Van Leeuwenhoek, 65, 289–310. Poole, R.K., Ioannidis, N., and Orii, Y. (1996) Microbiology, 142, 1141–1148. Chen, Y.P., Woodin, S.A., Lincoln, D.E., and Lovell, C.R. (1996) J. Biol. Chem., 271, 4609–4612. Wallace, W.J. and Caughey, W.S. (1975) Biochem. Biophys. Res. Commun., 62, 561–567. Belvedere, G. and Samaja, M. (1994) Methods Enzymol., 231, 598–621. Rao, S.I., Wilks, A., and Oritz de Montellano, P.R. (1993) J. Biol. Chem., 268, 803–809. Juchau, M.R., Chapman, D.E., Yang, H.Y. et al. (1996) Drug Metab. Dispos., 24, 1362–1368. Zilletti, L., Ciuffi, M., Franchi-Micheli, S. et al. (1994) Methods Enzymol., 231, 562–573. Sutton, H.C., Roberts, P.B., and Winterbourn, C.C. (1976) Biochem. J., 155, 503–510. Mieyal, J.J. and Starke, D.W. (1994) Methods Enzymol., 231, 573–598. Everse, J., Johnson, M.C., and Marini, M.A. (1994) Methods Enzymol., 231, 547–561. Herold, S., and Puppo, A. (2005) J. Biol. Inorg. Chem., 10, 946–957. Huang, Z., Shiva, S., Kim-Shapiro, D.B. et al. (2005) J. Clin. Invest., 115, 2099–2107. Huang, K.T., Keszler, A., Patel, N. et al. (2005) J. Biol. Chem., 280, 31126–31131. Kim-Shapiro, D.B., Gladwin, M.T., Patel, R.P., and Hogg, N. (2005) J. Inorg. Biochem., 99, 237–246. Coon, M.J. (2002) J. Biol. Chem., 277, 28351–28363. Fridovich, I. (1986) Arch. Biochem. Biophys., 247, 1–11. Fridovich, I. (1995) Ann. Rev. Biochem., 64, 97–112. Gabbianelli, R., Santroni, S.M., Fedeli, D. et al. (1998) Biochem. Biophys. Res. Commun., 242, 560–564. Freitas, T.A.K., Hou, S., and Alam, M. (2003) FEBS Lett., 552, 99–104. Wakasugi, K., Nakano, T., and Morishima, I. (2003) J. Biol. Chem., 278, 36505–36512. Wakasugi, K. and Morishima, I. (2005) Biochemistry, 44, 2943–2948.

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 13

Respiratory Nitric Oxide Reductases, NorB and NorZ, of the Heme–Copper Oxidase Type Walter G. Zumft Department of Applied Biosciences, Division of Molecular Microbiology, University Karlsruhe, D-76128 Karlsruhe, Germany

1. INTRODUCTION Respiratory NO metabolism by prokaryotes is tied to the properties of two variants of nitric oxide reductases. The membrane-bound activities catalyze the reduction of NO to nitrous oxide: 2NO + 2H+ + 2e− → N2 O + H2 O E0 pH 70 = +1177 V G0 = −3063 kJ/mol. The study of the respiratory transformation of nitrite to N2 O by bacteria led to the discovery of NO as a biomolecule and of its specific role in bacterial energy conservation, which is embedded in the prokaryotic process of denitrification [1]. NO exerts its function as a respiratory substrate at concentrations intermediate to that of acting as a signal molecule and as toxic agent. The range of proteins known to metabolize NO has significantly broadened in recent years, and respiratory NO reductase (NOR) is not the only means by which the bacterial cell deals reductively with NO. The list of various heme and nonheme iron proteins with NO reductase activity comprises now cytochrome oxidases, cytochrome c , cytochrome c nitrite reductase, flavohemoglobin, flavorubredoxin, flavodiiron protein, and the tetraheme cytochrome c554 as the most recent addition [2,3]. Pseudomonas stutzeri served as the source organism for the first isolation of a respiratory type NOR [4], elucidation of its primary structure [5], first spectroscopic and redox studies [6], discovery of the heme-nonheme iron binuclear catalytic site [7,8], and proposal of NO as the signal molecule for NOR regulation [5,9]. Together, these data provided the paradigm of a respiratory NOR with which to compare enzymes and variants from other bacteria. This chapter will give a synopsis of the biochemistry and genetics of respiratory NO metabolism, represented in two homologous types of NOR, NorB and NorZ. The role of NO as signaling molecule will be considered in as much as it pertains to the regulation of the respiratory function.

328

W.G. Zumft

2. GENETIC ORGANIZATION AND FUNCTIONAL PROPERTIES OF nor GENE PRODUCTS 2.1. Genetic Organization A comparison of primary structures shows that NOR proteins separate into two groups: the short-chain NO reductases (scNOR) of about 450 amino acids and the long-chain NO reductases (lcNOR) of about 760 amino acids. This division parallels the electron donor specificity insofar as scNORs are in a complex with a cytochrome c, whereas lcNORs derive electrons from quinol. Thus, one finds also the designations cNOR and qNOR for the two enzyme forms in the literature to emphasize this distinction [10]. Since electron donor specificity is not a systematic element for classification as illustrated with the heme– copper oxidases [11], this feature will be used only in a descriptive way. The recommended designation for both types of NOR is based on phylogenetic considerations and uses the genetic terminology NorB for a scNOR (or cNOR) and NorZ for a lcNOR (or qNOR). The first nor gene, belonging to a scNOR, was identified by reverse genetics based on the purified NOR complex from P. stutzeri. The nor locus was found to be linked to nir genes for respiratory nitrite reductase [12,13]. This information and by extending the sequence of a nir locus to flanking regions, helped to identify the nor genes of the principal model organisms used in denitrification research (Table 1). The cytochrome c-dependent NOR complex is encoded from the structural genes norC and norB but only a single gene is required for the quinol-dependent enzyme. The first gene sequenced of a lcNOR [14], and purification of the corresponding NOR [15], was that of Ralstonia Table 1. Properties of nor genes encoding respiratory NO reductases and ancillary functions Gene

Mass of gene product (kDa)a

norB

527

Catalytic subunit of short-chain NOR

norC

167

Cytochrome c subunit of short-chain NOR

norD

696

Presumed cytoplasmic protein; affects NirS and NorCB expressions and functions; carries C-terminally von Willebrand factor type A metal-binding domain

norE=nirO

190

5-span membrane protein; affects anaerobic growth yield and growth rate; similarity with cytochrome c oxidase subunit III

norF=nirP

86

2- or 3-span membrane protein; affects NO and nitrite reduction

norQ=nirQ

296

Affects NirS and NorCB functions and cellular growth characteristics; similarity with CbbQ proteins

norZ=norB2

845b

Structural gene of the long-chain NOR with an N-terminally fused quinol oxidase domain

a

Functions of gene product or observations

Average mass of unprocessed proteins without cofactors from the model organisms P. aeruginosa, P. stutzeri, and Pa. denitrificans. b Value for R. eutropha.

NorB and NorZ

329 nor E F C

B

Q

D A. tumefaciens

nir P O

Q S

7 kb nir

P. stutzeri P. aeruginosa

nir S Q

O P Pa. denitrificans Magnetococcus sp. napH

T. denitrificans napH

N. europaea R. eutropha norZ

Fig. 1. Organizational patterns of nor genes of denitrifying bacteria. Homologous genes are shown in identical patterns. The coding reading frames are scaled to relative size; the arrow boxes indicate transcriptional directions. nirS genes are drawn truncated.

eutropha (now taxonomically transferred to Cupriavidus necator). This bacterium represents the unique case where two copies of the structural gene are distributed between the chromosome 2 (the smaller one of two chromosomes in this organism) and a 0.45-Mb plasmid. Figure 1 illustrates organizational patterns of nor genes observed in different NOutilizing bacteria as deduced from individual sequence analyses or in silico from genomes. The norCB genes are associated in various ways with the ancillary genes, norD, norE (= nirO, in parentheses, the original nomenclature used for Pseudomonas genes), norF (= nirP), and norQ (= nirQ). The most common gene organization for a scNOR is the six-gene cluster norEFCBQD. Genes norQ and norD are linked with norCB and are of predictive value for a scNOR in a genome. norD is found downstream of norCB and often norQ is intercalated between norB and norD. The location of the ancillary genes norQ, −E, and −F is variable; norE and norF may even be absent (Fig. 1). Clear roles for the products of norE/nirO and norF/nirP are not established. Bacteria harboring norZ have no ancillary nor genes. Two copies of norB are present in the single, 2.9-Mb circular chromosome of Thiobacillus denitrificans. The distinct arrangement of nor genes in cluster Tden1 is shown in Fig. 1; a gene with similarity to napH of the periplasmic nitrate reductase system is intercalated in the gene cluster, a situation also found in Magnetococcus sp. The second nor cluster, Tden2, is organized like that of Nitrosomonas europaea or Nitrosospira multiformis. Two norZ genes are distributed over the two chromosomes in Burkholderia thailandensis. Besides a regular norZ gene (Susi1), an incomplete gene (Susi2) is present in Solibacter usitatus.

330

W.G. Zumft

2.2. Functional Properties of nor Gene Products The organization of the structural genes for P. stutzeri NOR suggests a two-subunit core complex from the norCB operon with a single transcript of 2.2 kb [5]. Purified NORs from P. stutzeri and other bacteria are consistently composed of two subunits, and activity assays show that a minimal composition of two subunits is sufficient for catalysis. Additional bands occasionally observed in electrophoretic patterns of the purified enzyme are aggregates from the two reductase subunits [6,16]. The use of different isolation procedures has not challenged the heterodimeric subunit structure of NOR thus far [17,18]. Genes nirQ (norQ) and norD are constant features of nor clusters. The nirQ gene was the very first accessory gene identified for NO reduction. Because of its location next to nirS and pleiotropic function for both nitrite and NO reduction, it was given a nir designation. nirQ encodes a ≈30-kDa protein that carries the Walker nucleotidebinding motifs A and B. A knockout mutation in nirQ results in an inactive NOR and affects nitrite respiration in vivo [19]. Deletion of nirQ together with nirO and nirP abolishes denitrifying growth. Both defects are complemented in trans by nirQ alone, which indicates nonessentiality both for the nirO and nirP products (and their homologs norE and norF probably as well) [20]. The nirQ homolog of Paracoccus denitrificans was termed norQ because of its association with nor genes. A norQ strain loses nitrite reductase and about 80% of NOR together with all NOR activity. The strain also has no N2 O reductase activity [21]. NirQ is similar to CbbQ, required for CO2 fixation in the Calvin–Benson–Bassham cycle, or the gas vesicle protein, GvpN. An intriguing aspect is the membership of NirQ and NorQ together with CbbQ and GvpN in a subfamily of the large MoxR family of AAA+ ATPases [22]. It has been noted that members of this family are frequently associated with proteins exhibiting the von Willebrand factor A domain, and it is hypothesized that the two proteins act concertedly in an energy-dependent function. The C-terminal region of NorD of about 170 amino acids correspond indeed to the type A domain of the von Willebrand factor. The nor gene clusters encoding the NorB-type reductase carry consistently the norD gene downstream of norCB (Fig. 1). Mutation of norD results for Pa. denitrificans in the loss both of NirS and activity, together with a nearly complete loss of NOR protein and all NOR activity [21]. A mutation in Rhodobacter sphaeroides affecting both norQ and norD because of the underlying operon structure results in lack of anaerobic growth on nitrate or nitrite, although norB is expressed [23]. Mutation of the monocistronic norD gene of P. stutzeri results in strongly decreased NO-reducing activity, but a NOR protein can still be isolated, which points to an ancillary function of NorD for the NorCB complex (A. Freiberg and W. G. Zumft, unpublished data). NorD of Brucella suis is a virulence factor; norD mutants lose their capability to multiply within the mouse macrophage [24]. An apparent interpretation of this phenotype is that the loss of NO reduction results in susceptibility to macrophage-produced NO. Since the ≈70-kDa protein NorD has no signal peptide and no extended hydrophobic region, it is probably a cytoplasmic component. The type A domain of the von Willebrand factor is a metal-binding fold that serves in eukaryotic proteins as adhesion site and is also involved in membrane trafficking [25,26]. Since the catalytic domains of both NorB and NorZ are structurally similar, it is unlikely that NorD would be required for catalysis

NorB and NorZ

331

or domain biogenesis. Also, it does not seem plausible that NorD would insert Fe into the heme groups of NOR, given that it is associated with the NorB-type reductase but not with NorZ. NorD may function in concert with NirQ(NorQ) in the formation of the NorCB complex and assist in membrane assembly. Both NorC and NorB carry no signal sequence; posttranslational processing of either protein is limited to the removal of the N-terminal methionine residue, and it is not clear how insertion is achieved of the NOR complex into the membrane.

3. RESPIRATORY NITRIC OXIDE REDUCTASES ARE MEMBERS OF THE HEME–COPPER OXIDASE SUPERFAMILY The catalytic domain of NorZ and the NorB subunit of respiratory NORs are homologous to each other and, in turn, are homologous to the catalytic subunit of the heme– copper oxidase family. The structural similarity comprises positional identity of metal center–ligating residues in the primary structure and the binuclear active site. Main differences are the lack of proton translocation by NOR and substitution of CuB with Fe. Heme–copper oxidases have a characteristic sequence of topologically conserved histidine residues for coordination of the metal centers as determined for Pa. denitrificans [27,28]: His94(60) (in parentheses, positions of homologous NOR residues of P. stutzeri) and His413(349) for the low-spin heme, His411(347) for the high-spin heme, and His267(207), −325(258), and −326(259) for CuB . The catalytic subunit of NOR has these six conserved histidines coordinating two heme groups and a nonheme Fe instead of the Cu atom as depicted schematically in Fig. 2. In analogy to CuB , the nonheme Fe of NOR is referred to as FeB and is part of the antiferromagnetically coupled binuclear catalytic site. From the amino acid sequence of the NorB protein from P. stutzeri, a topology of a 12-span membrane protein was deduced [5,31], which also provides the best topological fit for other NORs, including the catalytic domain of NorZ.

FeB

E 135

H60

E

H207

138

E

211

E

I

II

III

IV

V

H

H

S 258

H259

274

H347

E280 P

H349

281

215

VI

VII

VIII

IX R

N

339

heme b

X

XI

301

heme b 3

XII C

Fig. 2. Structural elements of NorB. The 12-span membrane part of the catalytic subunit is shown. Residues whose functions were addressed by site-directed mutagenesis are circled; for explanation, see text. Data are from P. stutzeri [7]; underlined glutamate residues were probed in Pa. denitrificans (counting there E135→122, E138→125, E211→198, and E215→202) [29,30]. The four helices carrying the histidine residues for heme and FeB ligation, defining the greater heme–copper family, are indicated.

332

W.G. Zumft

Cytochrome cbb3

Cytochrome c and quinol oxidases

ba

e

pn

γL

2 ha Bt vol H ar r a Hm n c ar Pa ac

ac

β

0.1

δ

M av i m fi S et au fi r fi G Gste kau fi Blic fa Su β Nmensi2 β Ngon γ Msuc γ Phal γ lloi β • CnecJ β • CnecHc p ne β • C cHet β • Cme h γ • Hc i1 s fa Su ip d ac C ol s R β Rfer β ap Pn β mal B a1 β Bth vio β C fri β γS m fu δ S eh d δ A nec Sy cy

•P

No

δG

e

ar

rB

o Ss

No

•γ •γ ro sy Pp Cpr • γ e γ PaPflu • • γ ac β tu Bb l β Ps u Nm en2 γ Td ap q Mc the a • ε n H de Tm

l

ar

typ eN

Or

•α ag α M m tl • α Re G179α Pscyc α A el α m S m Atu e α Afa α Bmel Bjap α Rpal α Pden α Rden α Spom α Rsph Rbac αα Hha Neu l γ Noc r β e γ S M den γ D agM Az aro • c1 β T o M den ar • β aq 1 • u• β γ

edu ctases

–

rZ t y

pe N O

reductases

Fig. 3. Phylogenetic relationships among NO reductases in the context of the heme–copper oxidase superfamily. Genome sequence projects and data banks were searched by January 2006 with the bioinformatics tools described in [32] with four query sequences, consisting of the NorB proteins of P. stutzeri and Pa. denitrificans and the NorZ proteins of Neisseria meningitidis and R. eutropha H16. Sequence alignment and tree construction were done with Clustal X, and the tree displayed with TreeView. Incomplete sequences and those of unidentified bacteria were excluded. For each entry, the taxonomic affiliation is given: ac, Actinobacteria; aq, Aquificae; ar, Archaea; cy, Cyanobacteria; fa, Fibrobacteres/Acidobacteria group; fi, Firmicutes; greek letters denote the respective groups of Proteobacteria. Symbols:  denotes presence of Cu-containing nitrite reductase, NirK, in the respective species; • denotes presence of cytochrome cd1 nitrite reductase, NirS; no symbol, absence of either nitrite reductase. Abbreviations: Acyc, “Achromobacter cycloclastes” IAM1013 (the 16S rRNA sequence reveals this as an alphaproteobacterium of the genus Ensifer); Adeh, Anaeromyxobacter dehalogenans 2CP-C; Afae, Alcaligenes faecalis S-6, 6978960; Atum, Agrobacterium tumefaciens C58; Azoar, Azoarcus sp. EbN1; Blic, Bacillus licheniformis ATCC14580; Bbac, Bdellovibrio bacterivorus HD100; Bjap, Bradyrhizobium

NorB and NorZ

333

Figure 3 shows an unrooted phylogenetic protein tree of NOR proteins in the context of cytochrome oxidase sequences, including NORs identified by bioinformatic tools. The cbb3 -type oxidases are structurally related closest to respiratory NOR. Notably, catalytic activities with respect to NO or oxygen reduction of NORs and heme–copper oxidases overlap in distinct cases. Whether the NO-reducing activity of heme–copper oxidases [33] can be considered an evolutionary consequence of a common origin or is an adjunct property of certain active site structures has to be resolved. The concept of an evolutionary relationship underlying the members of this tree seems cogent, given the similarity in primary structure and other shared structural and functional elements. It has been known for some time that NO interacts with and inhibits cytochrome aa3 ; this type of oxidase has very little catalytic activity toward NO or catalyzes a single or few turnovers only [34,35]. However, certain cytochrome ba3 -, caa3 -, cbb3 -, and cbo3 -type

Fig. 3. (Continued) japonicum USDA110; Bmel, Brucella melitensis 16M, positionally identical with Brucella abortus 9-941 and Brucella suis 1330; Bmal, Burkholderia mallei ATCC23344, positionally identical with Burkholderia pseudomallei K96243; Btha, Burkholderia thailandensis E264; 1 and 2 refer to location of the norZ gene on chrosome I or II, respectively; Cvio, Chromobacterium violaceum ATCC12472; Cdip, Corynebacterium diphtheriae NCTC13129; Cmet, Cupriavidus metallidurans CH34; CnecHc, Cupriavidus necator (formerly Ralstonia eutropha) H16, chromosomally encoded; CnecHp, Cupriavidus necator H16, plasmid encoded; CnecJ, Cupriavidus necator JMP43; Cpsy, Colwellia psychrerythraea 34H; Daro, Dechloromonas aromatica RCB; Gkau, Geobacillus kaustophilus HTA426 (positionally nearly identical with Geobacillus thermodenitrificans NG80-2); Gmet, Geobacter metallireducens GS15; Gste, Geobacillus stearothermophilus 10; Hche, Hahella chejuensis KCTC2396; Hhal, Halomonas halodenitrificans IFO14912; Hmar, Haloarcula marismortui ATCC43049; Hthe, Hydrogenobacter thermophilus TK-6; Hvol, Haloferax volcanii DS2; Iloi, Idiomarina loihiensis L2TR; Lpne, Legionella pneumophila Philadelphia-1; MagMC1, Magnetococcus sp. MC-1 (taxonomically unclassified); Maqu, Marinobacter aquaeolei VT8; Mavi, Mycobacterium avium ssp. paratuberculosis K-10; Mcap, Methylococcus capsulatus Bath; Mmag, Magnetospirillum magneticum AMB-1; Msuc, Mannheimia succiniproducens MBEL55E; Neur, Nitrosomonas europaea ATCC 19718; Ngon, Neisseria gonorrhoeae FA1090; Nmen, Neisseria meningitidis Z2491, serogroup A, positionally identical with strains MC58, serotype B and FEM18, serogroup C; Nmul, Nitrosospira multiformis ATCC 25196; Noce, Nitrosococcus oceani ATCC 19707; Pacn, Propionibacterium acnes KPA171202; Paer, Pseudomonas aeruginosa PAO1; Pbae, Pyrobaculum aerophilum IM2; Pden, Paracoccus denitrificans PD1222; Pflu, Pseudomonas fluorescens C7R12; PsG179, Pseudomonas sp. G-179; Phal, Pseudoalteromonas haloplanktis TAC125; Pnap, Polaromonas naphthalenivorans CJ2; Ppro, Photobacterium profundum SS9; Pstu, Pseudomonas stutzeri ATCC14405; Rbac, bacterium strain HTCC2654 of Rhodobacterales; Retl, Rhizobium etli CFN 42; Rfer, Rhodoferax ferrireducens DSM 15236; Rden, Roseobacter denitrificans ATCC33942; Rpal, Rhodopseudomonas palustris CGA009; Rsph, Rhodobacter sphaeroides, strains 2.4.1 (lacks nirK) and 2.4.3; Rsol, Ralstonia solanacearum GMI1000; Saur, Staphylococcus aureus EMRSA-16; Sden, Shewanella denitrificans OS-217; Sfri, Shewanella frigidimarina NCIMB 400; Sfum, Syntrophobacter fumaroxidans MPOB; Smel, Sinorhizobium meliloti 1021; Spom, Silicibacter pomeroyi DSS-3; Ssol, Sulfolobus solfataricus P2; Susi, Solibacter usitatus Ellin6076; for explanation of Susi1 and 2, see the text; Synec, Synechocystis sp. PCC6803; Tden, Thiobacillus denitrificans ATCC 25259; for explanation of Tden1 and 2, see the text; Tmden, Thiomicrospira denitrificans ATCC 33889. Sequences were obtained from the NCBI data bank; COMB at the University of Maryland, and The Wellcome Trust Sanger Institute.

334

W.G. Zumft

oxidases have NO-reducing activity [36–38]. A hyponitrite intermediate is suggested in the reduction of NO by a heme–copper oxidase from density functional theory (DFT) calculations [39,40]. Reflections on a tenable evolutionary link of oxygen and nitrate respiration [41] and the relation of nitrous oxide reductase to cytochrome c oxidase [42,43] fostered the hypothesis of a common ancestry of NOR with heme–copper oxidases [44,45]. A feasible evolutionary scenario has been suggested for the heme–copper oxidases [11]. On the basis of key residues for proton translocation and positionally conserved amino acids of the catalytic subunit, the mitochondrial cytochrome aa3 oxidases (themselves subdivided) and cytochrome ba3 oxidases (here referred to globally as cytochrome and quinol oxidases) are distinguished from cytochrome cbb3 oxidases. A discussion of the phylogeny of the heme–copper oxidase superfamily must include NOR as an explicit element (Fig. 3). Four archaeal NORs cluster with the lcNORs comprising the NorZ clade. The enzyme from Pyrobaculum aerophilum has been characterized as a quinol-dependent NOR and a two-subunit reductase has been suggested for Sulfolobus solfataricus [46]. Newer genome data not included in Fig. 3 suggest now a separate archaeal NorZ branch. It is notable in the protein tree that the taxonomic affiliations obtained with the 16S rRNA and phylogenetic reconstructions of both types of NO reductases, NorZ, and NorB depart considerably (Fig. 3). Although this can have a number of reasons, it may suggest a high lateral mobility of genes encoding these enzymes independent of bacterial speciation. The average mass of a lcNOR is 84.7 kDa. NorZ proteins trunctated N-terminally by about 250 amino acids, and possibly lacking the quinol-binding domain, are observed for Rhodoferax ferrireducens and Mycobacterium avium. A shortened NOR version is also present in Mannheimia succiniproducens, but other putative cases, pointed out previously [2], were not consolidated upon sequence completion. Of the two norB copies in the circular chromosome of T. denitrificans, Tden1 (Fig. 3) is highly expressed in nitrate-denitrifying, thiosulfate-oxidizing cells [47]. Interestingly, the second copy (Tden2), only lowly expressed under the same conditions, has a putative diheme NorC of about twice the size of canonical NorC. Such modified NorC proteins are found also in the parasitic bacterium Bdellovibrio bacterivorus and the ammoniaoxidizing bacterium N. multiformis. The NorC modification may hint at a different electron transfer pathway in these bacteria, which together form a small separate lineage. Of another small (and taxonomically diverse) group, the norCB genes of Hydrogenobacter thermophilus have been studied. It seems that the ancillary genes are absent from this member of the Aquificales and other members of the group (Fig. 3) [48]. Since the Hydrogenobacter clade represents evolutionary deep-branching members, it seems feasible that they harbor norCB genes that have not yet acquired the ancillary genes otherwise found within the NorB family (Fig. 3). The main enzymatic NO generators of nitrite-denitrifying bacteria are two types of respiratory nitrite reductases (reviewed in ref. [49]). The active site of the NirK protein is a type 2 copper atom, whereas in the NirS protein (synonymous with cytochrome cd1 ), it is heme d1 . The Nir designations are derived from the respective gene names. No distributional pattern links NorB and NorZ enzymes with either type of respiratory nitrite reductase (Fig. 3). A phylogenetic analysis of NirK proteins reveals an Achromobacter cluster, an AniA cluster, an Haloferax cluster, and several dissimilar outlying members. The AniA group frequently comprises NirK proteins with a C-terminally fused putative heme c-binding domain, even though AniA itself is devoid of this feature [50]. NorZ

NorB and NorZ

335

is usually associated with the AniA-type of NirK, whereas NorB is nearly exclusively associated with the Achromobacter-type NirK. Overall, the distribution of the two respiratory nitrite reductase enzymes is such that lateral gene transfer and a high mobility of the nirK and nirS is suggested rather than a coevolution of the nitrite reductase genes with norB or norZ. Complete genomes reveal also cases where a NOR is present but respiratory nitrite metabolism is not encoded in the genome (Fig. 3). Here, the lack of nirK or nirS seems inherent rather than the result of a secondary evolutionary event. This again underlines independent evolutionary trajectories for NOR and respiratory nitrite reductases. The absence of a respiratory nitrite reductase is more frequently observed in bacteria harboring the NorZ enzyme, whose scavening or detoxifying function seems therefore at least as important as its respiratory function. Detoxification may be seen as a primitive trait, seemingly giving norZ evolutionary precedence [51], but one can also argue that only a more complex metabolic network would require a protective function.

4. SHORT-CHAIN RESPIRATORY NITRIC OXIDE REDUCTASE, NorB, IS A COMPLEX WITH CYTOCHROME c 4.1. Properties of NorB scNOR is a complex of the catalytic subunit NorB with the c-type cytochrome, NorC. The two subunits have sequence-deduced masses of about 17 kDa (NorC) and 53 kDa (NorB). Electrospray mass spectrometry matches the predicted mass of 17,477 for NorC from Pa. denitrificans within one unit, indicating absence of protein or cofactor modifications [18]. The mass obtained by SDS electrophoresis coincides only for NorC with the theoretical value, whereas NorB is a highly hydrophobic b-type cytochrome that deviates notably in its apparent mass of 34–38 kDa from the sequence-deduced value. A stoichiometry of 1:1 is usually assumed for an overall tetrameric enzyme structure. Two-dimensional crystals show a dimeric (NorCB)2 structure [18]. NOR is solubilized from the cytoplasmic membrane by detergents such as Triton X-100 [4,6], dodecyl maltoside [17,18,52], octyl glucoside [16], or sucrose monocaprate [53]. NOR is probably complexed with a lipid. The environment of the high-spin heme of NorB is sensitive to hydrophobic interactions as seen in alterations in the electron paramagnetic resonance (EPR) spectrum in different detergents [6]. Chemical analysis of the first purified NOR yielded more Fe than could be accounted for by the heme content [4]. The extra Fe showed no EPR signal for an iron-sulfur species and was considered a nonheme Fe that was subsequently proposed to be bound by the conserved histidines from helixes VI and VII [6,7]. The stoichiometry of nonheme Fe to heme b and heme c Fe in a minimal composition of the reductase complex of one cytochrome b and one cytochrome c is 1:2:1 [17,18]. The enzyme does not contain a significant amount of Cu to form a bimetallic center as in cytochrome c oxidase. However, as an exception, the scNOR from Roseobacter denitrificans (formerly Erythrobacter sp. OCh114) has a heme–copper center [54,55]. The absorption spectrum of NOR shows the spectral features of heme c and heme b [4,6,17,18,52,53]. The isolated NorB subunit in the reduced form has absorption maxima

336

W.G. Zumft

at 428, 531, and 560 nm. On electrophoretic separation of the reductase complex, NorC retains heme c [6]. The isolated, reduced NorC subunit has absorption maxima at 418, 523, and 551.5 nm. NorC is a membrane-bound monoheme c-type cytochrome with the N-terminus oriented toward the cytoplasm and a large heme c-binding domain of ≈120 amino acids residing in the periplasm. A single sequence motif, CXXCH, for covalent heme c attachment is located in the periplasmic domain following the single transmembrane helix for anchoring the protein in the membrane [5,21,56]. The exposure of the heme c-binding domain toward the periplasm is shown by a topologically sensitive reporter gene fusion into Leu67 in the immediate vicinity of the heme attachment site, 61-CIGCH-65, of P. stutzeri NOR [7]. NorC orientation toward the outside allows this protein to interact with a periplasmic cytochrome or (pseudo)azurin and supply electrons to the membrane-bound NorB. A distinguishing feature of NOR versus oxygen reductases is the lack of the conserved amino acids for vectorial proton transport. The reaction of NO to N2 O has a reduction potential of E0 (pH 7) of +1177 mV and a free energy of G0 of −306.3 kJ/mol. It is energetically comparable to the respiration of N2 O to N2 E0 pH7 = +1352 mV G0 = −3395 kJ/mol, and both reactions are thermodynamically more favorable than the respiration of nitrate to nitrite. Although the overall electron flow to NO is accompanied by proton extrusion, this is not due to the enzyme per se but the activity of upstream coupling sites. Electron transfer proceeds through cytochrome bc1 [57,58], and energy conservation is associated with a dehydrogenase and/or the cytochrome bc1 complex. A growth yield study has shown that the overall reaction of nitrite to N2 O reduction is energy conserving [59]. Protons for NO reduction are taken up from the periplasmic side [57,60,61]. Glu125 at the periplasmic side of helix IV (equivalent to Glu138 in Fig. 2) has been found by sitedirected mutagenesis to be essential for activity of the Pa. denitrificans enzyme. The residue is strictly conserved; a feasible role is proton uptake from the periplasm [29,62]. A putative proton-transfer pathway with protonable residues has been proposed [63]. Glu122 and Glu125 are the likely candidates to initiate the proton uptake process [30]. NOR incorporated into liposomes does not generate an electric potential during steady state turnover. This implies that electrons and protons are taken up during catalysis from the same side of the membrane and indirectly confirm that this side is the periplasm [64].

4.2. Electron Transfer Ranking the metal centers by their reduction potential suggests NorC as the port of entry for electrons delivered to the catalytic subunit NorB. Its Em is about +280 mV [6,65]. NorC and NorB interact such that NorB affects the Em of NorC. Thus, Em of the isolated periplasmic NorC domain is downshifted to +183 mV [65]. NO reduction proceeds at a cytochrome c–coated gold electrode at a reduction potential of +108 mV, pH 6, involving a one-electron transfer step [66]. The suggested electron flow from NorC is supported from the midpoint potential for heme b of NorB of about +322 to 345 mV [6,65]. The Em values for FeB and heme b3 of NorB are +320 and +60 mV, respectively. This generates a thermodynamic barrier for the two-electron reduction of

NorB and NorZ

337

the binuclear center, but may also avoid the formation of a stable inhibitory Fe(II)–heme b3 –NO species during enzyme turnover [67]. The low reduction potential of the highspin heme b3 allows to study cyanide binding to NOR in three different redox states, viz., oxidized, fully reduced, and three-electron reduced. It was suggested from this that there are open and closed forms of the ferric heme dinuclear center [68]. Electron transfer rates have been measured by optical spectroscopy and electrometry. The transfer rate between heme c from cytochrome c and heme b in the NorB subunit is 3 ± 2 × 104 s−1 . The value corresponds to that of cytochrome c oxidase. The distance of the two centers is estimated at 20 Å. The electron transfer rate between heme b and heme b3 is too fast to be resolved (>106 s−1 ) [64]. Besides these similarities, there are also notable differences in ligand binding and ligand dynamics between both enzymes [69]. Mutational inactivation of cytochrome c550 of Pa. denitrificans PD1222 does not affect nitrate denitrification, but NO reduction of such a mutant becomes sensitive to the Cu chelator diethyl dithiocarbamate (30% inhibition). This behavior is expected if scNOR activity in this bacterium depends on cytochrome c550 (Em = +256 mV [70]) and pseudoazurin (Em = +230 mV [71]) as interchangeable electron donors [72]. The cytochrome bc1 complex and cytochrome c2 are active in the electron transfer to scNOR in photodenitrifiers [60,73]. The interaction with the electron donor is thought to be electrostatic as the activity decreases rapidly with increasing ionic strength [58]. The involvement of the proton translocating cytochrome bc1 complex in electron transfer to a scNOR should make this pathway energetically more favorable versus that of the lcNOR. Studies toward this aspect, however, are not available. The quinol-dependent NOR of Bacillus azotoformans reacts also with a membrane-bound cytochrome c551 , making it feasible that within the same organism, both energetically differing pathways are being used [74].

5. THE ACTIVE SITE 5.1. Active Site Model The strong hydrophobicity of NorB finds its explanation in 12 hydrophobic segments of the primary structure. Precise information about the transmembrane helixes and cofactor arrangement of a heme–copper oxidase came from the crystal structure of the Pa. denitrificans cytochrome c oxidase [27], and these data were used to generate a three-dimensional model of NorB from P. stutzeri to elaborate the structural relation of both enzymes and reveal the possibility to accomodate the binuclear Fe center in the NorB subunit. [7]. In the NorB model, the metal-binding ligands can be positioned at the corresponding residues of the subunit I of cytochrome c oxidase within an overall conserved array of membrane-spanning segments (Fig. 4). The predicted -helices of NorB are nearly superimposable on that of the oxidase. In modeling NorB on the oxidase structure, it places the conserved residues His207 (helix VI), and His258 and His259 (helix VII) in the position of the FeB center [7]. A model with the same features was generated for NorB from Pa. denitrificans [63].

338 (A)

W.G. Zumft outside

(B)

His-60

II

low-spin heme His-349

VI His-259

X His-207 His-347

inside

His-258

high-spin heme

VII

Fig. 4. Structural model of NorB. (A) Folding and heme location modeled onto the crystal structure coordinates of cytochrome c oxidase [7]. (B) Cofactor-binding sites with ligating histidines conserved in the heme–copper oxidase superfamily. Residue positions are based on the primary structure of P. stutzeri [5]. Reproduced from reference [1] with permission.

5.2. Active Site Properties The optical features of NOR indicate the presence of low-spin hemes b and c and of a high-spin heme b, in the absorbance properties of the CO-reacted material [53]. The magnetic circular dichroism spectrum of NOR shows bands originating from a low-spin heme c (with histidine and methionine coordination) attributed to NorC, and a magnetically isolated low-spin heme b (with bis-histidine coordination) and a highspin heme b3 magnetically coupled with nonheme Fe; the two heme b species are both attributed to NorB [8]. The EPR spectrum shows two low-spin heme signals in the form of two rhombic trios, one with gz = 35 and unknown gx and gy , and a second rhombic trio with g values 2.9, 2.3, and 1.4. The assignment of this spectrum to the different low-spin heme species has been ambiguous for long [8,17,67,75]. Expression of the periplasmic domain of Pa. denitrificans NorC and purification of the recombinant protein, however, assured the discrimination of the EPR signals arising from NorC [65]. The recombinant protein shows the g = 35 resonance, referred to as the “high gmax ” signal, which thus must originate from heme c. The presence of the second rhombic trio is thought to arise from two conformation states due to flexibility of the methionine ligand. The heme b3 –nonheme Fe binuclear site of NOR (Fig. 4) represents the catalytic site homologous to the heme a3 –CuB bimetallic center of cytochrome c oxidase. Supportive evidence for this comes from mutagenesis of the predicted FeB ligand, His258; substitution shows that it is necessary for activity [7]. The nonheme Fe site in NorB is likely to involve, in addition to the three histidines, further ligands to provide a suitable octahedral environment rather than the distorted tetrahedral environment seen with CuB [67].

NorB and NorZ

339

Other than the three conserved residues His207, −258, and −259, the NorB model yields no clues as to the nature of the additional ligands as no carboxylate or hydroxy amino acid is positioned in sufficient proximity [7]. The conserved glutamate 280 of helix VIII, adjacent to the FeB -binding helix VII, could be close enough for binding, providing an oxgen ligand, but in the model, this glutamate stabilizes His207 rather than to ligate FeB . Glu211 is a conserved residue of NOR that is in the immediate vicinity of the FeB -ligating His207, and site-direct mutagenesis in Pa. denitrificans (there Glu198) has revealed its essentiality. It may serve as a ligand to FeB or provide exchangeable protons for the protonation step at the active site [29,62]. The conserved Glu215 of NorB has also been targeted in Pa. denitrificans (there Glu202) and resulted in 70% loss of wild-type activity. Note also that the NorB structure harbors no conserved tyrosine that would be equivalent to the covalent tyrosine–histidine link for one of the histidine ligands of CuB as in the oxidase [76]. In oxidized NOR, heme b3 is not coordinated to the protein by its proximal histidine residue [77]. A single oxygen isotope–sensitive ligand in the resonance Raman spectrum observed at 811 cm−1 has been attributed to the bridged Fe O Fe structure of the binuclear diiron center [78]. The insensitivity of the resonance Raman frequency to D2 O suggests that the oxo-bridge is not protonated. Metal complexes have been synthesized to mimick the binuclear Fe site of NorB by a porphyrin-coordinated Fe connected via a

-oxo bridge to a nonheme Fe [79,80], and recently also by a reduced diiron complex that reacts with NO to a dinitrosyl adduct [81]. Another model mimicks the active site by a heme Fe and a trisimidazole- and glutaric acid–bound FeB . The carboxylic acid increases the stability of FeB retention in the distal site and is important for modulation of the reduction potential of both iron sites [82]. Since FeB of NOR seems to be ligated by histidine and glutamate residues, the latter model may come close in mimicking the active site. The interaction of CO with reduced heme iron can be used as a probe for the nature of the heme distal pocket. CO binding to the reduced heme b3 follows a diffusion-limited bimolecular rate constant [83]. This indicates that the active site has an open structure, which may account for the high affinity of NOR to the substrate and keep NO at a low concentration during denitrification. A fraction of the reduced enzyme displays a barrier for initial CO binding, which can be interpreted as indicating the presence of a sixth ligand on heme b3 . This ligand will have to be displaced to allow NO binding to reduced heme b3 , as envisaged in two mechanisms (see below), or alternatively, it may be required for avoiding the formation of an inhibitory Fe(II) nitrosyl [83,84]. The (CO) bands observed in FTIR with dithionite-reduced NOR from B. azotoformans are assignable to heme–CO and FeB CO stretching frequencies. A high chloride concentration displaces the CO from the nonheme Fe [85]. The possibility to demonstrate ligand binding to both Fe centers is of importance for the so-called trans mechanism of NOR (see following section). Other than cytochrome oxidase, NOR of Halomonas (formerly Paracoccus) halodenitrificans does not easily accommodate small anions such as cyanide or azide. The channel for NO access is therefore thought to be highly hydrophobic. Ion access increases in the reduced form at low pH, at which the enzyme seems to change conformation [86]. At the same time, the intensity of the high-spin EPR signal increases, which would be expected on leaving of the bridging group. However, no nonheme Fe signal appears concomitantly. Probing the binuclear site further by cyanide-binding revealed that this anion

340

W.G. Zumft

binds several orders of magnitude tighter to the ferric heme b3 of the three-electronreduced state than to the fully oxidized enzyme, suggesting open and closed forms of the catalytic site [68]. The -oxo bridged form of NOR is seen as a closed state with respect to ligand binding that opens by the input of one electron into the dinuclear site. A mixed valence Fe(III)–heme b3 –FeB (II) is therefore suggested as the substrate-binding state for the catalytic cycle [67,68,87].

5.3. Active Site Reactivity Much interest is currently directed at the details of the reduction of NO by respiratory NOR. Mechanistic variatiations, discussed below, are colloquially referred to as cis (i.e., binding of two NO molecules to only one of the metals) and trans mechanisms (i.e., binding of NO to each one of the metal centers of the binuclear site) (Fig. 5). The trans mechanism was deduced from steady state kinetics of NOR, which indicated the sequential binding of two NO molecules [17]. It starts with oxidized resting NOR as a -oxo-bridged system [Fe3+ O Fe3+ ] with a five-coordinate high-spin heme [78]. The site is antiferromagnetically coupled. The distance between the Fe atoms and the Fe O Fe angle were estimated at 3.5 Å and 145 , respectively. Two-electron reduction of the binuclear site weakens the -oxo bridge and leads to the formation of a His Fe(II) bond; the oxo-atom leaves as water on protonation. The ferrous heme is now bonded only to the proximal histidine. The fully reduced site binds two NO molecules to the ferrous heme b3 and the nonheme Fe(II) (Fig. 5). The heme-b3 Fe is again five-coordinate in

His

NHis

3+

Fe

O

3+ Fe B

His

NHis

2+

Fe

resting

3+

N O

His

NHis

O N

2+

FeB

trans

NO

Fe OH NO

2+

FeB

NHis

2+

Fe

–

NO

2+

FeB

NO

cis:FeB

cis:b3

Fig. 5. States of the binuclear NOR center for cis and trans mechanisms of NO reduction. For description of these states and their position in the reaction sequence, see the text. The species of the respective mechanisms do not display equivalence of electron transfer steps or protonation. The resting species of oxidized NOR (top left) is thought to be common to all mechanisms.

NorB and NorZ

341

this state [77]. EPR signals from FeB (II) NO as well as Fe(II)–heme b3 –NO have been reported for the enzyme under turnover [18,88,89]. A two-electron reduction releases N2 O and reestablishes the -oxo-bridged situation of the resting state. Problematic for the trans mechanism are observations that upon reduction, the sixth ligand on heme b3 continues to be a hydroxide ion, and the high stability of an Fe(II)–heme b3 –NO complex [90–92]. To overcome the latter difficulty, either the environment of the binuclear center weakens the binding of NO to Fe(II)–heme b3 or the low reduction potential of heme b3 keeps this center oxidized [67]. Kumita et al. [89] propose a trans mechanism that starts with the activation of the resting state by opening the -oxo bridge by proton transfer and a four-electron reduction to the fully reduced NOR state. NO binding breaks the proximal His–heme b3 bond. NO reduction leaves the enzyme in the half-reduced state, when the oxidized heme b3 and FeB have been reduced from heme c of the NorC subunit and the low-spin heme b of the NorB subunit. The rate constant for these electrons transfers is estimated at <102 s−1 . The ferric low-spin state of hemes c and b is detected in time-resolved EPR at 10 ms. The half-reduced state is proposed to undergo a second step of subtrate reduction, i.e., two N2 O molecules are released in a single turnover that leaves NOR in the fully oxidzed state. Reduction by four electrons leads to the next turnover cycle. Alternative reaction or electron transfer pathways of a trans mechanism have been deduced for NOR of Halomonas denitrificans [93]. The enzyme at pH 8 may either proceed along the fully reduced route or the mixed valence form Fe(III)–heme b3 –FeB (II), with the low-spin hemes b and c both reduced. In addition, a fully oxidized NOR at pH 5.5 (the pH optimum for enzyme activity) is also competent in NO binding and subsequent two-electron reduction. Rather than reduction, the first reaction step of Pa. denitrificans NOR is proposed to be the opening or at least weakening of the bridged resting structure, which is thought to occur at low pH. The mixed valence state is also proposed as the activating species for NO reduction to bind a single molecule of NO to Fe(III)–heme b3 . Addition of two electrons yields an Fe N O− species that is attacked by a second NO molecule and has to be protonated to yield transiently hyponitrite (HONNO− ). A second protonation step releases N2 O and water [94]. Because of the extremely high affinity of ferrous heme complexes for NO, it has been argued that the reduction potential of the heme Fe2+ − NO0 /Fe2+ NO− couple is at −0.9 V, too negative to be accessible by a physiological reductant [95], and the formation of a nonheme Fe dinitrosyl complex at FeB (cis:FeB mechanism, Fig. 5) has been suggested for N,N bond formation. However, a Fe–dinitrosyl complex might also stabilize two NO molecules in a geometry not apt for interaction. Escherichia coli cytochrome bo3 is a member of the heme–copper family and has been used as a model for NOR. Two molecules of NO bind to CuB with different affinity, but NO never binds to the Fe(III) heme group to form a low-spin Fe(III) NO species [84]. The mode of NO-binding to cytochrome bo3 was suggested to be applicable to NOR. A cis:b3 mechanism (Fig. 5), as derived from the situation proposed for NO reduction by cytochrome cbb3 of P. stutzeri [96,97], has also been hypothesized but reinterpreted more recently in favor of a trans mechanism [94]. The former mechanism is sometimes referred to as the P450 mechanism, in analogy to the monoheme-based mechanism operative in fungal NOR. Here, an intermediate Fe(II)HNO species is attacked by a second NO molecule to break the Fe N bond [98]. Data for NO-binding both to the

342

W.G. Zumft

nonheme iron and high-spin heme b are in apparent contrast to a pure cis mechanism [18,88,93]. A monoheme-based one-electron reduction of NO to NO− was proposed prior to knowledge about the binuclear structural element in NOR [99,100]. Reduction of NO was thought to occur at the heme with N2 O formation proceeding nonenzymatically by dimerization of the nitroxyl (HNO) and dehydration: 2HNO = N2 O + H2 O. DFT calculations were applied to develop a feasible mechanism of NO reduction at the heme b3 –FeB binuclear center [101]. They provide for NOR a detailed framework of reaction steps and alternatives to consider (Fig. 6). The energetically most favored mechanism is formally a cis:b3 -type mechanism and consists of initial binding of NO to the reduced Fe(II)–heme b3 –FeB (II) binuclear site. The binding of NO oxidizes the ferrous high-spin heme b3 to a low-spin heme in between. As a result, NO is partly reduced to a species with nitroxyl anion character. NO is thus activated toward attack by a second NO molecule. As the next step, a hyponitrite dianion intermediate is formed. The exergonic formation of the -oxo bridge between the two ferric irons drives the reaction to release N2 O. Reduction and protonation of the -oxo bridge and expulsion of water are energetically undemanding. The calculation is based on an octahedrally coordinated FeB , whereas a tetrahedral site is seen energetically as less favorable [101]. It seems that during the catalytic cycle, irrespective of the proposed mechanism, a ferric heme b3 with a bound water molecule will have to be generated. Spectral signals have been identified of distinct pH forms of NOR, which suggest such an intermediate in the catalytic cycle [102]. The oxidized form as the -oxo-bridged Fe(III) O Fe(III) species (Fig. 5) has a ligand to metal charge transfer (LMCT) band at 595 nm. Upon

III

FeB O

III

+H+

FeIII

FeB HO

II

+H+ +e–

FeIII

FeB H2 O FeIII –H2O

–N2O IV

FeB

II

O N O 2

FeB

FeII

FeIII +e–

III

FeB

O O N N FeIII

II

+NO

FeB NO FeII

II

+NO

FeB FeII

Fig. 6. Proposed catalytic mechanism of NOR based on favorable free-energy profiles and DFT calculations assuming an octahedrally coordinated FeB [101]. The heme b3 plane is represented by bars. For experimental evidence of particular species, see the text. Critical steps in this mechanism are formation of the hyponitrite dianion and the ferrylic FeB species. The latter could be a reason for iron at the binuclear site instead of Cu as in cytochrome oxidase.

NorB and NorZ

343

a three-electron reduction of NOR and reduction of FeB (whose Em is much more positive than that of heme b3 ), a species with an LMCT band at 635 nm is detected at pH 6, whereas at pH 8.5, the band is at 605 nm [102]. Ferric hemes with His H2 O coordination usually have LMCT bands around 630 nm. The different species are thus taken as a water molecule bound to ferric heme b3 at low pH (when the enzyme is in its most active form), and a hydroxide at high pH. The low pH 635-nm form may arise from a simple protonation of the high-pH form [83]. Turnover of NOR has been investigated by optical spectroscopy of liposomes, and has resolved at least five phases during the catalytic cycle, as well as the minimal uptake of one proton on reduction of the active site. The reaction is started by laser-flash photolysis of the fully reduced and CO-inhibited enzyme in the presence of NO, and is followed by time-resolved measurement of the membrane potential [64]. Overall, NOR turnover is quite complex. The first phase assumes the formation of a ferrous NO adduct of heme b3 within 2 s, as has been advanced in the trans model. It involves the detachment of the proximal histidine from heme b3 upon NO binding [77]. This would be expected to induce optical changes in the Soret region; however, no such changes are observed in the initial phase, which indicates an unperturbed heme b3 . The first phase is followed by a charge-separation phase, possibly involving charged side chains near heme b3 or FeB , and may be electron transfer from FeB to heme b3 –NO. Other phases can be interpreted as -oxo bridge formation and oxidation of the low-spin hemes. The data are compatible with a trans mechanism, rather than a cis model involving FeB , but would also allow for a P450-type mechanism [64]. In as much as observations with NO reductase activity of terminal oxidases may serve to understand bona fide NOR, the oxygen reductase activity of Pa. denitrificans NOR can serve to study ligand binding, and electron and proton transfer. The flow–flash technique applied to NOR, reacting with oxygen, was used to resolve these aspects kinetically [62]. Fully reduced NOR and O2 , after laser-flash dissociation of CO from heme b3 , yields two kinetic phases observed at 430 nm: a rapid phase ( = 40 s) of O2 binding to heme b3 and a slow phase ( = 25 ms) of proton-coupled electron transfer from the low-spin hemes b and c to the binuclear site. The rate constants for the slow phase are pH dependent with a pKa of ≈6.6. This behavior is thought to arise from a rate-limiting internal proton transfer from a donor group (possible one of the conserved glutamates), followed by reprotonation from the bulk solution. The reaction product of oxygen reduction by NOR is water.

6. LONG-CHAIN RESPIRATORY NITRIC OXIDE REDUCTASE, NorZ, IS ALSO A QUINOL OXIDASE The two genomic nor elements of R. eutropha encode orthologous genes with 90% positional amino acid identity. The chromosomally encoded gene was originally named norZ (later changed to norB2) but, as explained above, NorZ is the preferred label for this type of NOR. Both norZ genes encode functional NORs; only a double mutant is negative for NO consumption. At the same time, such a mutant is not viable under nitrate-denitrifying growth conditions, because of NO accumulation [14]. On sequencing norZ of Ralstonia, the absence of a homolog encoding the cytochrome c subunit (norC) was noted. The predicted NorZ protein is an 84.5-kDa polypeptide, possibly with 14

344

W.G. Zumft

transmembrane helices [14]. The extra helices versus NorB form an N-terminal extension of the protein and delimit a hydrophilic domain of about 200 amino acids. NorZ depends on quinol as electron donor, and it is assumed that the N-terminal extension forms the quinol oxidase domain. It has no sequence similarity to respiratory quinol oxidases. A weak relation to NorC has been postulated on the basis of similarity of short isolated sequence segments [10]. The optical spectrum of NorZ is that of a b-type cytochrome. EPR features of the protein show a low-spin heme with bis-histidine coordination and an EPR-silent, antiferromagnetically coupled high-spin heme–nonheme iron center [15,46], as shown for the NorB-type enzyme. Nitrate-denitrifying bacteria are found in a sizable number among the gram-positive bacteria [42]. The quinol-dependent NOR from B. azotoformans is a two-subunit enzyme, with the small subunit of 16 kDa thought to be a CuA protein. The dual electron donor specificity is suggested to alternatively promote respiratory or detoxifying NOR activity [74,103]. The EPR spectrum of the enzyme in the g = 2 region shows the seven-line feature of CuA , with a hyperfine splitting constant of 3.5 mT. DeVries and coworkers noted that the N-terminal sequence of the putative CuA subunit is similar to the subunit II of the bo3 -type cytochrome c oxidase of Bacillus stearothermophilus (now transferred to the genus Geobacillus [104]). Whether the NOR from B. azotoformans is a phylogenetic member of the NorZ group remains to be seen. The primary structure of the two-subunit enzyme from B. azotoformans is not known; genome-deduced NorZ proteins of grampositive relatives of B. azotoformans such as Geobacillus kaustophilus or Geobacillus thermodenitrificans are members of the NorZ clade (Fig. 3). Their masses of a single subunit enzyme of about 90 kDa and primary structures are that of a canonical NorZ. A CuA protein is not encoded as part of the norZ region, nor is a CuA domain evident in these NOR proteins. Several archaea are known with complete denitrification or the truncated version terminating with N2 O (reviewed in ref. [43]). A quinol-dependent NorZ is found in the archaeon Py. aerophilum [46]. Most of its properties are similar to Ralstonia NorZ. The novel feature of NorZ from Pyrobaculum is the presence of heme o, the ethenylgeranylgeranyl derivative of heme b. A second heme o with a hydroxyethenylgeranylgeranyl side chain is present in a stoichiometry of 1:1, and suggests that both modified hemes are equivalent to the heme b and heme b3 positions of NorB.

7. STRUCTURAL AND FUNCTIONAL VARIATIONS AMONG RESPIRATORY NITRIC OXIDE REDUCTASES Several modifications of the typical features of NOR have been observed (Table 2). Roseobacter denitrificans has an scNOR [54] that is structurally a member of the NorB group (Fig. 3), and shares with the Pa. denitrificans and Rh. sphaeroides enzymes 69 and 79% identity, respectively. The Roseobacter enzyme exhibits the structural elements of a NOR but has the heme b3 –CuB center and activity of a cytochrome c oxidase. The protein exhibits cytochrome c oxidase activity but has hardly any NOR activity. The resonance Raman spectrum of the oxidized protein indicates a six-coordinate high-spin species for heme b3 with the other heme six-coordinate low-spin species. In the reduced enzyme, all hemes are ferrous, six-coordinate low-spin species. The (Fe CO) stretching modes of the CO-reacted enzyme are closer to aa3 - and bo3 -type oxidases rather than NOR [55].

NorB and NorZ

345

Table 2. Structural modifications among nitric oxide reductases of the heme–copper oxidase family Electron donor for catalytic domain

Type of binuclear center

Substrate

Representative organism

Cytochrome c

Heme b3 –FeB

NO NO, O2 NO, O2 O2 NO NO NO

Pseudomonas stutzeri Paracoccus denitrificans Paracoccus pantotrophus Roseobacter denitrificans Ralstonia eutrophaa Pyrobaculum aerophilumb Bacillus azotoformans

Quinol CuA or quinol a b

Heme Heme Heme Heme

b3 –CuB b3 –FeB o3 –FeB b3 –FeB

Reference 8 18 53 54 15 46 103

Enzyme has not been studied spectroscopically. Archaeon.

The presence in this bacterium of an oxidase-type binuclear center within the NOR scaffold seems to be a convergent process since the Ro. denitrificans NOR branches from the Rh. sphaeroides and Pa. denitrificans scNORs, rather than representing some ancestral form (see Fig. 3). The NOR of Paracoccus pantotrophus is a cytochrome bc complex with properties of the Pa. denitrificans or P. stutzeri enzymes [53]. The enzyme not only reduces NO (84 mol NO • min−1 • mg protein−1 , Km 0.25 M), but has significant oxygen reductase activity with Pa. pantotrophus ferrocytochrome c550 as the electron donor. Low cytochrome oxidase activity is also observed of the recombinant Pa. denitrificans enzyme expressed in E. coli membranes. The activity is lost in the Glu125 and −198 mutants [29].

8. NITRIC OXIDE SIGNALING AND nor GENE REGULATION 8.1. Transcription of nor Genes The nor genes of P. stutzeri are organized in three transcriptional units consisting of the norCB and nirQOP operons and the monocistronic norD transcript. Downstream of norD follows the dnrN operon carrying the gene for the nor regulator DnrD [105]. The nor genes of P. aeruginosa are activated at the nirQ and norC promoters. nirQ is activated in a norCBD background, as well as norC is activated in a nirQOP background, indicating independent transcriptional units. nirQOP is organized as an operon [106,107]. Two transcripts comprising norCBQ and norCBQDEF (ca. 2.7 and 5.4 kb, respectively) have been found in Pa. denitrificans [108]. The norC promoter exhibits two transcriptional start sites; the distal one is active under both aerobic and anaerobic conditions [109]. Transcription start sites have been determined for norC of Rh. sphaeroides [23] and B. japonicum [110]; tetracistronic transcripts of norCBQD are suggested in either case. The nor gene cluster of A. tumefaciens is transcribed from the norC promoter, which indicates norEF and norCBQD transcripts [111]. The norA promoter of the norAB operon of R. eutropha has been characterized with respect to three binding sites for its cognate regulator NorR (see below) [112].

346

W.G. Zumft

8.2. NO as the Signal Molecule Denitrification is a facultative process, usually expressed in the absence or at a lowered tension of oxygen, when an N-oxide, for example nitrate, is present. Nitrate as a signaling molecule is sensed in pseudomonads by the NarXL two-component sensor-regulator system that switches on nitrate respiration [113]. However, nitrate itself is not the signaling molecule for the expression of NOR, but rather NO as one of the downstream reduction products of nitrate. The control of nitrite reduction and NO reduction in denitrifying bacteria is regulated coordinately to assure removal of NO by NOR or, if this is not possible, by downregulation of nitrite reduction, relying on a common regulatory mechanism. The steady state concentration of free NO during denitrification is in the nanomolar range [114,115]. A key observation for the nature of the N-oxide signal was the finding that a nirS mutant, lacking cytochrome cd1 nitrite reductase, can be complemented with the nirK gene for copper–nitrite reductase. Inactivation of nirS results simultaneously in decreased NO reduction and a low level expression of NOR in P. aeruginosa, P. stutzeri, and P. fluorescens [9,107,116]. Reestablishing the NO generator by nirK expression causes the nirS mutant to revert to wild-type phenotype. The observation implied that NO is the signal molecule for the induction of NOR [5,9]. Direct activation of gene expression by NO was shown for the nirK promoter of Rh. sphaeroides [117,118] and subsequently confirmed for a number of other NO responsive promoters. NO is now being viewed as the central signaling molecule for the expression of the nitrite-denitrifying apparatus (Fig. 7). The effective concentration of free NO as an inducer for the transcription of the nirSTB and norCB operons of P. stutzeri is in the range of 5–50 nM [105], a concentration that builds up during steady state nitrate denitrification.

8.3. Crp–Fnr Factors as Master Regulators for the Expression of NO Reductase NOR is not formed under aerobic culture conditions [119–121], and is purified in high yield only from the O2 -limited or anaerobically grown cell. Anaerobic expression of NOR is consistent with the presence of a recognition motif for an Crp–Fnr-type regulator in the promoters of norC operons and other nor genes. The study of anaerobic arginine degradation of P. aeruginosa led to the identification of Anr of the Crp–Fnr family as the positive regulator [122], which affected anaerobic nitrate denitrification also. The existence of recognition sequences specific for a Crp–Fnr regulator thus found its satisfactory explanation [123]. The picture became more complicated subsequently when regulatory genes were found on sequencing the regions downstream of the nor genes of various denitrifiers [1,21,124–127]. The encoded regulators (termed variably Dnr, DnrD, Nnr, or NnrR for their involvement in nitrite-, nitric oxide–reduction, and denitrification) are also structural and functional homologs of the Crp–Fnr family. As their dinstinct feature versus Anr, however, they lack the Fe S cluster for oxygen sensing and transcriptional activation. Phylogenetically, they belong within the Crp–Fnr family to two separate regulator branches of which Dnr and NnrR are lead members [32]. A bioinformatic study of possible target genes of Dnr and NnrR shows several ones participating in NO

NorB and NorZ

347 O2

NO

O2

Anr

?

PnirS P. stutzeri PnorC (Pa. denitrificans)

DnrD (Nnr)

PnirS PnorC P. aeruginosa

dnr/Dnr

NO Redox/O2

NO

O2

FixLJ

fixK2/FixK2

PrrAB

PnirK R. sphaeroides PnorC

NnrR

nirK B. japonicum PnorC

nnrR/NnrR ?

e.g. fix, hem

NO

?

RpoN

NO

NorR

NO

NsrR

PnorA R. eutropha

–

PaniA N. gonorrhoeae PnorB

Fig. 7. Signal transduction pathways and regulators for the expression of core components of bacterial NO respiration. The mode of interaction of the NO signal with the regulator remains to be established. P denotes the target promoter of an identified transcriptional unit. A question mark indicates an uncertain correlation or unknown element.

homeostasis other than those of denitrification, and predicts a considerable overlap of cognate regulons [128]. Signal transduction pathways clarified in several nitrite-denitrifying bacteria by knockout mutagenesis are shown in Fig. 7. The regulatory interactions achieve the coordinate expression of the principal enzymes for nitrite and NO metabolism under anaerobic conditions. The Dnr and NnrR target promoters carry partially palindromic motifs, which are similar to the so-called Fnr box of E. coli: TTGAT-N4 -ATCAA (Dnr box) and TTGCG-N4 -CGCAA (NnrR box). The presence of such motifs in the promoters of the dnr and nnrR genes themselves suggests autoregulation, but in specific cases, this is also the basis of hierarchical control structures (Fig. 7). Direct binding of a Dnr or NnrR factor to target DNA is suggested from mutational evidence obtained with the nirS and norC promoters of Pa. denitrificans [109,129]. How Dnr and NnrR regulators are converted by NO to active transcription factors is not known (developments reviewed in ref. [130]). Currently, a heme-based mechanism finds the most support. The pseudomonadal regulators have been isolated as hemeassociated proteins [105,131], and Pa. denitrificans Nnr responds in a hemA mutant

348

W.G. Zumft

with defective heme biosynthesis to the heme precursor 5-aminolaevulinic acid [132]. A critical region for transcription activity around Tyr93 has been identified by sitedirected mutagenesis in NnrR [133]. The same position had been previously pinpointed in Nnr of Pa. denitrificans [134]. Substitution of Arg80 of Paracoccus Nnr renders the protein insensitive to NO. There is preliminary evidence that Nnr senses both NO and oxygen and that the Arg80 mutant protein has both processes uncoupled [132]. The nor genes of Pa. denitrificans are activated in response to anaerobiosis and NO as the signal [135,136], but a minor, aerobically active norC promoter is not dependent on Nnr [109]. In B. japonicum, the signal transduction pathway starts with the heme-based oxygen sensor-regulator system FixLJ. The pathway is branched at the level of FixK2 (also a Crp–Fnr family member), and involves an NnrR-type regulator further downstream in the regulatory cascade (Fig. 7) [137,138].

8.4. The NorR Regulator The nor genes of R. eutropha of both the chromosome 2 and the megaplasmid HG1 encode functional NORs. They are transcribed in each case from the promoter of the upstream-located norA1 and norA2 genes whose functions are not known. The norA promoter is dependent on the sigma factor RpoN ( 54 ) and is activated by the NOresponsive regulator NorR [139,140]. NorR is encoded on the complementary strand immediately upstream of norA. NorR shows similarity to NtrC-like regulators. It has a central AAA + ATPase nucleotide-binding domain, a C-terminal DNA-binding domain, and an N-terminal GAF domain typical of signal-transduction proteins. Given the presence of this domain, it is assumed that NorR is regulated directly by the signal and targeted in sensitive promoters to the partially palindromic sequence GGT-(N7 )-ACC [112,140]. Data in silico show that norR homologs are found in other bacteria where they are associated often with the hmp gene for flavohemoglobin, demonstrating an overlap of nor and other NO homeostasis genes. NorR of E. coli has recently been shown to carry a mononuclear nonheme-Fe center within the GAF domain [141]. NO binding results in an {Fe(NO)}7 (S = 3/2) mononitrosyl complex, which causes activation of the nucleotide-binding domain. ATP hydrolysis drives open complex formation and thus NO-dependent transcription by 54 –RNA polymerase.

8.5. The NsrR Regulator A third type of NO responsive regulator, NsrR, is active in NO reduction in the causative agents of gonorrhea and meningitis. Neisseria gonorrhoeae and N. meningitidis harbor NorZ-type NORs (Fig. 3). Neisserias exhibit a two-step denitrification from nitrite to nitrous oxide. The role of NsrR is to relieve repression of NOR in the presence of NO during the bacterial pathogenic life cycle [142,143]. The NsrR regulator has a broad action spectrum in bacterial inorganic nitrogen metabolism. Putative regulons have been identified in a bioinformatic study [128]. NsrR belongs to the Rrf2 family of transcriptional repressors, which are putative [2Fe 2S] proteins and whose bestcharacterized member is the IscR protein that acts on the isc genes for iron–sulfur cluster biogenesis.

NorB and NorZ

349

9. CONCLUSIONS It is two decades that respiratory NOR was discovered in P. stutzeri. At that time, NO was seemingly an odd inorganic metabolite of anaerobic respiration by denitrifying bacteria, and the existence of NOR had even been seriously questioned in the years before. Besides bacterial bioenergetics, there was only the aspect of NO toxicity from the macrophage defense, directed against bacteria. We have now progressed to a much broader picture and can define an entire NO metabolome of oxidative and reductive NO transformations contributing to NO homeostasis in which respiratory NOR has a prominent place but shares it with several players in bacterial metabolism. The aspect most neglected had been the active-site mechanism of NO reduction. Fortunately, there are now clearly formulated alternatives that can be probed and subjected to experimental scrutiny. This overview has pointed out, however, that the reaction chemistry of NOR is only one aspect of an enzyme that is embedded in biosynthetic processing and intricate regulation. Continuing challenges are the three-dimensional structure of NOR, elucidation of the function of the ancillary genes for the NorB-type reductase, and the mechanistical basis of NO signaling. Thus, it is likely that NOR will remain an interesting topic for some time to come.

ACKNOWLEDGMENTS Work from the author’s laboratory was supported by the Deutsche Forschungsgemeinschaft and Fonds der Chemischen Industrie.

REFERENCES [1] [2] [3] [4] [5] [6] [7]

[8] [9] [10] [11] [12] [13] [14] [15] [16]

Zumft, W.G. (1997) Microbiol. Mol. Biol. Rev., 61, 533. Zumft, W.G. (2005) J. Inorg. Biochem., 99, 194. Upadhyay, A., Hooper, A.B., and Hendrich, M.P. (2006) J. Am. Chem. Soc., 128, 4330. Heiss, B., Frunzke, K., and Zumft, W.G. (1989) J. Bacteriol., 171, 3288. Zumft, W.G., Braun, C., and Cuypers, H. (1994) Eur. J. Biochem., 219, 481. Kastrau, D.H.W., Heiss, B., Kroneck, P.M.H., and Zumft, W.G. (1994) Eur. J. Biochem., 222, 293. Kannt, A., Michel, H., Cheesman, M.R. et al. (1998) In Biological Electron-Transfer Chains: Genetics, Composition and Mode of Operation, (G.W. Canters and E. Vijgenboom, eds.) NATO ASI series, Ser. C, Vol. 512, Dordrecht, The Netherlands: Kluwer Academic Publishers, p. 279. Cheesman, M.R., Zumft, W.G., and Thomson, A.J. (1998) Biochemistry, 37, 3994. Palmedo, G., Seither, P., Körner, H. et al. (1995) Eur. J . Biochem., 232, 737. Hendriks, J., Oubrie, A., and Castresana, J. et al. (2000) Biochim. Biophys. Acta, 1459, 266. Pereira, M.M., Santana, M., and Teixeira, M. (2001) Biochim. Biophys. Acta, 1505, 185. Braun, C. and Zumft, W.G. (1991) J . Biol. Chem., 266, 22785. Braun, C. and Zumft, W.G. (1992) J . Bacteriol., 174, 2394. Cramm, R., Siddiqui, R.A., and Friedrich, B. (1997) J . Bacteriol., 179, 6769. Cramm, R., Pohlmann, A., and Friedrich, B. (1999) FEBS Lett., 460, 6. Dermastia, M., Turk, T., and Hollocher, T.C. (1991) J . Biol. Chem., 266, 10899.

350 [17] [18] [19] [20]

[21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42]

[43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54]

W.G. Zumft Girsch, P., and de Vries, S. (1997) Biochim. Biophys. Acta, 1318, 202. Hendriks, J., Warne, A., Gohlke, U. et al. (1998) Biochemistry, 37, 13102. Jüngst, A. and Zumft, W.G. (1992) FEBS Lett., 314, 308. Arai, H., Kawasaki, S., Igarashi, Y. and Kodama, T. (1996) In Molecular Biology of Pseudomonads (T. Nakazawa, K. Furukawa, D. Haas, and S. Silver, eds.) Washington, DC, American Society of Microbiology, p. 298. de Boer, A.P.N., van der Oost, J., Reijnders, W.N.M. et al. (1996) Eur. J . Biochem., 242, 592. Snider, J. and Houry, W.A. (2006) J . Struct. Biol., 156, 200. Bartnikas, T.B., Tosques, I.E., Laratta, W.P. et al. (1997) J. Bacteriol., 179, 3534. Loisel-Meyer, S., Jiménez de Bagüés, M.P., Bassères, E. et al. (2006) Infect. Immun., 74, 1973. Ponting, C.P., Aravind, L., Schultz, J. et al. (1999) J . Mol. Biol., 289, 729. Letunic, I., Goodstadt, L., Dickens, N.J. et al. (2002) Nucleic Acids Res., 30, 242. Iwata, S., Ostermeier, C., Ludwig, B., and Michel, H. (1995) Nature, 376, 660. Michel, H., Behr, J., Harrenga, A., and Kannt, A. (1998) Annu. Rev. Biophys. Biomol. Struct., 27, 329. Butland, G., Spiro, S., Watmough, N.J., and Richardson, D.J. (2001) J . Bacteriol., 183, 189. Thorndycroft, F.H., Butland, G., Richardson, D.J., and Watmough, N.J. (2006) Biochem. J., 401, 111. Zumft, W.G. (1993) Arch. Microbiol., 160, 253. Körner, H., Sofia, H.J., and Zumft, W.G. (2003) FEMS Microbiol. Rev., 27, 559. Brunori, M., Giuffrè, A., and Sarti, P. (2005) J . Inorg. Biochem., 99, 324. Brudvig, G.W., Stevens, T.H., and Chan, S.I. (1980) Biochemistry, 19, 5275. Giuffrè, A., Barone, M.C., Brunori, M. et al. (2002) J . Biol. Chem., 277, 22402. Giuffrè, A., Stubauer, G., Sarti, P. et al. (1999) Proc. Natl. Acad. Sci. USA, 96, 14718. Forte, E., Urbani, A., Saraste, M. et al. (2001) Eur. J . Biochem., 268, 6486. Butler, C.S., Forte, E., Scandurra, F.M. et al. (2002) Biochem. Biophys. Res. Commun., 296, 1272. Blomberg, L.M., Blomberg, M.R.A., and Siegbahn, P.E.M. (2006) Biochim. Biophys. Acta, 1757, 31. Ohta, T., Kitagawa, T., and Varotsis, C. (2006) Inorg. Chem., 45, 3187. Broda, E. (1975), The Evolution of the Bioenergetic Processes, Oxford, UK, Pergamon Press. Zumft, W.G. (1992) In The prokaryotes. A Handbook on the Biology of Bacteria: Ecophysiology, Isolation, Identification, Applications, Vol. 1 (A. Balows, H. G. Trüper, M. Dworkin, W. Harder and K.-H. Schleifer, eds) New York, NY: Springer-Verlag, p. 554. Zumft, W.G. and Kroneck, P.M.H. (2007) Adv. Microb. Physiol., 52, 107. van der Oost, J., de Boer, A.P.N., de Gier, J.-W.L. et al. (1994) FEMS Microbiol. Lett., 121, 1. Saraste, M. and Castresana, J. (1994) FEBS Lett., 341, 1. de Vries, S., Strampraad, M.J.F., Lu, S. et al. (2003) J . Biol. Chem., 278, 35861. Beller, H.R., Chain, P.S.G., Letain, T.E. et al. (2006) J . Bacteriol., 188, 1473. Suzuki, M., Arai, H., Ishii, M., and Igarashi, Y. (2006) Biosci. Biotechnol. Biochem., 70, 1666. Cutruzzolà, F. (1999) Biochim. Biophys. Acta, 1411, 231. Boulanger, M.J. and Murphy, M.E.P. (2002) J . Mol. Biol., 315, 1111. de Vries, S. and Schröder, I. (2002) Biochem. Soc. Trans., 30, 662. Carr, G.J. and Ferguson, S.J. (1990) Biochem. J ., 269, 423. Fujiwara, T. and Fukumori, Y. (1996) J . Bacteriol., 178, 1866. Matsuda, Y., Inamori, K., Osaki, T. et al. (2002) J Biochem. (Tokyo), 131, 791.

NorB and NorZ [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75] [76] [77] [78] [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94]

351

Matsuda, Y., Uchida, T., Hori, H. et al. (2004) Biochim. Biophys. Acta, 1656, 37. Arai, H., Igarashi, Y., and Kodama, T. (1995) Biochim. Biophys. Acta, 1261, 279. Carr, G.J., Page, M.D., and Ferguson, S.J. (1989) Eur. J . Biochem., 179, 683. Itoh, M., Mizukami, S., Matsuura, K., and Satoh, T. (1989) FEBS Lett., 244, 81. Koike, I. and Hattori, A. (1975) J . Gen. Microbiol., 88, 11. Bell, L.C., Richardson, D.J., and Ferguson, S.J. (1992) J . Gen. Microbiol., 138, 437. Shapleigh, J.P. and Payne, W.J. (1985) J . Bacteriol., 163, 837. Flock, U., Watmough, N., and Ädelroth, P. (2005) Biochemistry, 44, 10711. Reimann, J., Flock, U., Lepp, H. et al. (2007) Biochim. Biophys. Acta, 1767, 362. Hendriks, J.H.M., Jasaitis, A., Saraste, M., and Verkhovsky, M.I. (2002) Biochemistry, 41, 2331. Oubrie, A., Gemeinhardt, S., Field, S. et al. (2002) Biochemistry, 41, 10858. Heering, H.A., Wiertz, F.G.M., Dekker, C., and de Vries, S. (2004) J . Am. Chem. Soc., 126, 11103. Grönberg, K.L.C., Roldán, M.D., Prior, L. et al. (1999) Biochemistry, 38, 13780. Grönberg, K.L., Watmough, N.J., Thomson, A.J. et al. (2004) J . Biol. Chem., 279, 17120. Pinakoulaki, E. and Varotsis, C. (2003) Biochemistry, 42, 14856. Samyn, B., Berks, B.C., Page, M.D. et al. (1994) Eur. J . Biochem., 219, 585. Martinkus, K., Kennelly, P.J., Rea, T., and Timkovich, R. (1980) Arch. Biochem. Biophys., 199, 465. Moir, J.W.B. and Ferguson, S.J. (1994) Microbiology (UK), 140, 389. Satoh, T. (1984) Arch. Microbiol., 139, 179. Suharti, Heering, H.A. and de Vries, S. (2004) Biochemistry, 43, 13487. Sakurai, N. and Sakurai, T. (1997) Biochemistry, 36, 13809. Buse, G., Soulimane, T., Dewor, M. et al. (1999) Protein Sci., 8, 985. Moënne-Loccoz, P. and de Vries, S. (1998) J . Am. Chem. Soc., 120, 5147. Moënne-Loccoz, P., Richter, O.-M.H., Huang, H.-W. et al. (2000) J . Am. Chem. Soc., 122, 9344. Martens, C.F., Murthy, N.N., Obias, H.V., and Karlin, K.D. (1996) J . Chem. Soc. Chem. Commun., 1996, 629. Wasser, I.M., Martens, C.F., Verani, C.N. et al. (2004) Inorg. Chem., 43, 651. Wasser, I.M., Huang, H.-W., Moënne-Locoz, P., and Karlin, K.D. (2005) J . Am. Chem. Soc., 127, 3310. Collman, J.P., Yan, Y.-L., Lei, J., and Dinolfo, P.H. (2006) Inorg. Chem., 45, 7581. Hendriks, J.H.M., Prior, L., Baker, A.R. et al. (2001) Biochemistry, 40, 13361. Butler, C.S., Sewar, H.E., Greenwood, C., and Thomson, A.J. (1997) Biochemistry, 36, 16259. Lu, S., de Vries, S., and Moënne-Loccoz, P. (2004) J . Am. Chem. Soc., 126, 15332. Kurose, S., Sakurai, N., and Sakurai, T. (2001) J . Inorg. Biochem., 83, 281. Pinakoulaki, E., Gemeinhardt, S., Saraste, M., and Varotsis, C. (2002) J . Biol. Chem., 277, 23407. Sakurai, T., Sakurai, N., Matsumoto, H. et al. (1998) Biochem. Biophys. Res. Commun., 251, 248. Kumita, H., Matsuura, K., Hino, T. et al. (2004) J . Biol. Chem., 279, 55247. Romberg, R.W. and Kassner, R.J. (1979) Biochemistry, 18, 5387. Shimizu, M., Basolo, F., Vallejo, M.N., and Baldwin, J.E. (1984) Inorg. Chim. Acta, 91, 251. Brucker, E.A., Olson, J.S., Ikeda-Saito, M., and Phillips, Jr., G.N. (1998) Proteins, 30, 352. Sakurai, T., Nakashima, S., Kataoka, K. et al. (2005) Biochem. Biophys. Res. Commun., 333, 483. Pinakoulaki, E., Ohta, T., Soulimane, T., et al. (2005) J . Am. Chem. Soc., 127, 15161.

352

W.G. Zumft

[95] Ye, R.W., Averill, B.A., and Tiedje, J.M. (1994) Appl. Environ. Microbiol., 60, 1053. [96] Pinakoulaki, E., Stavrakis, S., Urbani, A., and Varotsis, C. (2002) J . Am. Chem. Soc., 124, 9378. [97] Stavrakis, S., Pinakoulaki, S., Urbani, A., and Varotsis, C. (2002) J . Phys. Chem. B, 106, 12860. [98] Daiber, A., Shoun, H., and Ullrich, V. (2004) J . Inorg. Biochem., 99, 185. [99] Goretski, J. and Hollocher, T.C. (1988) J . Biol. Chem., 263, 2316. [100] Garber, E.A.E., Wehrli, S., and Hollocher, T.C. (1983) J . Biol. Chem., 258, 3587. [101] Blomberg, L.M., Blomberg, M.R.A., and Siegbahn, P.E.M. (2006) Biochim. Biophys. Acta, 1757, 240. [102] Field, S.J., Prior, L., Roldán, M.D. et al. (2002) J . Biol. Chem., 277, 20146. [103] Suharti, Strampraad, M.J.F., Schröder, I., and de Vries, S. (2001) Biochemistry, 40, 2632. [104] Nazina, T.N., Tourova, T.P., Poltaraus, A.B. et al. (2001) Int. J . Syst. Evol. Microbiol., 51, 433. [105] Vollack, K.-U. and Zumft, W.G. (2001) J . Bacteriol., 183, 2516. [106] Arai, H., Igarashi, Y., and Kodama, T. (1994) Biosci. Biotechnol. Biochem., 58, 1286. [107] Arai, H., Kodama, T., and Igarashi, Y. (1999) FEMS Microbiol. Lett., 170, 19. [108] Murai, K., Miyake, K., Andoh, J., and Iijima, S. (2000) J Biosci. Bioeng., 89, 384. [109] Hutchings, M.I. and Spiro, S. (2000) Microbiology (UK), 146, 2635. [110] Mesa, S., Velasco, L., Manzanera, M.E. et al. (2002) Microbiology (UK), 148, 3553. [111] Baek, S.-H. and Shapleigh, J.P. (2005) Appl. Environ. Microbiol., 71, 4427. [112] Büsch, A., Pohlmann, A., Friedrich, B., and Cramm, R. (2004) J . Bacteriol., 186, 7980. [113] Härtig, E., Schiek, U., Vollack, K.-U., and Zumft, W.G. (1999) J . Bacteriol., 181, 3658. [114] Goretski, J., Zafiriou, O.C., and Hollocher, T.C. (1990) J . Biol. Chem., 265, 11535. [115] Kalkowski, I. and Conrad, R. (1991) FEMS Microbiol. Lett., 82, 107. [116] Ye, R.W., Arunakumari, A., Averill, B.A., and Tiedje, J.M. (1992) J . Bacteriol., 174, 2560. [117] Kwiatkowski, A.V. and Shapleigh, J.P. (1996) J . Biol. Chem., 271, 24382. [118] Tosques, I.E., Kwiatkowski, A.V., Shi, J., and Shapleigh, J.P. (1997) J . Bacteriol., 179, 1090. [119] Remde, A. and Conrad, R. (1991) FEMS Microbiol. Lett., 80, 329. [120] Körner, H. (1993) Arch. Microbiol., 159, 410. [121] Jones, A.M. and Hollocher, T.C. (1993) Biochim. Biophys. Acta, 1144, 359. [122] Zimmermann, A., Reimmann, C., Galimand, M., and Haas, D. (1991) Mol. Microbiol., 5, 1483. [123] Ye, R.W., Haas, D., Ka, J.-O. et al. (1995) J . Bacteriol., 177, 3606. [124] Arai, H., Igarashi, Y., and Kodama, T. (1995) FEBS Lett., 371, 73. [125] Tosques, I.E., Shi, J., and Shapleigh, J.P. (1996) J . Bacteriol., 178, 4958. [126] Vollack, K.-U., Härtig, E., Körner, H. and Zumft, W.G. (1999) Mol. Microbiol., 31, 1681. [127] Bedzyk, L., Wang, T., and Ye, R.W. (1999) J . Bacteriol., 181, 2802. [128] Rodionov, D.A., Dubchak, I.L., Arkin, A.P. et al. (2005) PLoS Comput. Biol., 1, e55. [129] Saunders, N.F.W., Houben, E.N.G., Koefoed, S. et al. (1999) Mol. Microbiol., 34, 24. [130] Zumft, W.G. (2002) J. Mol. Microbiol. Biotechnol., 4, 277. [131] Rinaldo, S., Giardina, G., Brunori, M., and Cutruzzola, F. (2005) Biochem. Soc. Trans., 33, 184. [132] Lee, Y.-L., Shearer, N., and Spiro, S. (2006) Microbiology (UK), 152, 1461. [133] Laratta, W.P. and Shapleigh, J.P. (2003) FEMS Microbiol. Lett., 229, 173. [134] Hutchings, M.I., Shearer, N., Wastell, S. et al. (2000) J . Bacteriol., 182, 6434. [135] van Spanning, R.J.M., de Boer, A.P.N., Reijnders, W.N.M. et al. (1995) FEBS Lett., 360, 151. [136] van Spanning, R.J.M., Houben, E., Reijnders, W.N.M. et al. (1999) J . Bacteriol., 181, 4129. [137] Mesa, S., Bedmar, E.J., Chanfon, A. et al. (2003) J . Bacteriol., 185, 3978.

NorB and NorZ

353

[138] Mesa, S., Ucurum, Z., Hennecke, H., and Fischer, H.-M. (2005) J . Bacteriol., 187, 3329. [139] Römermann, D., Warrelmann, J., Bender, R.A., and Friedrich, B. (1989) J . Bacteriol., 171, 1093. [140] Pohlmann, A., Cramm, R., Schmelz, K., and Friedrich, B. (2000) Mol. Microbiol., 38, 626. [141] D’Autréaux, B., Tucker, N.P., Dixon, R., and Spiro, S. (2005) Nature, 437, 769. [142] Overton, T.W., Whitehead, R., Li, Y. et al. (2006) J . Biol. Chem., 281, 33115. [143] Rock, J.D., Thomson, M.J., Read, R.C., and Moir, J.W.B. (2007) J . Bacteriol., 189, 1138.

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 14

Nitric Oxide Reductase (P450nor ) from Fusarium oxysporum Andreas Daibera , Hirofumi Shounb and Volker Ullrichc a

Klinikum der Johannes Gutenberg-Universität Mainz, Medizinische Klinik II — Molekulare Kardiologie, Obere Zahlbacher Str. 63, 55101 Mainz, Germany. b Dept. of Biotechnology, University of Tokyo, Tokyo 113–8657, Japan. c Faculty of Biology, University of Konstanz, 78457 Konstanz, Germany.

1. INTRODUCTION 1.1. The Diversity of Heme-thiolate Proteins It was almost 50 years ago that spectroscopic investigations led to the discovery of a pigment that by difference spectroscopy of its carbon monoxide (CO) complex under reducing conditions revealed an unusual absorption band at 450 nm [1]. Using Warburg’s technique of the photochemical action spectrum Estabrook, Cooper and Rosenthal [2] identified this pigment as the “terminal oxidase” of the steroid-21-monooxygenase. In liver, the unspecific monooxygenase system for xenobiotics showed the same spectral properties, but when the “P450” chromophore was isolated, it turned out to be converted to a “P420” form without monooxygenase activity [3]. Although it did not fulfill the definition of a cytochrome as a pure electron carrier, the pigment was called “cytochrome” P450, for which model studies allowed scientists to conclude that a thiolate ligand was the characteristic feature of “P450” proteins [4,5]. Many membrane-bound monooxygenases were found to activate dioxygen at their heme-thiolate active site, but the chemical link between this structure and the insertion of an oxygen atom into the substrate was the subject of speculation [6–8]. Today, there seems to be consensus that the thiolate ligand, due to its electron density, donates negative charge to the iron and at the same time prevents electron donation from ligands at the sixth ligand position of the iron [9,10]. This redox-active nature of the thiolate seems to determine the specific catalytic activity of heme-thiolate proteins, and we comment this property on the following pages. The surprisingly different properties of heme-thiolate proteins became obvious when thromboxane synthase and prostacyclin synthase, two endoperoxide isomerases, were characterized as P450 proteins [11], and also an allyl peroxide isomerase (allene oxide synthase) was identified as a heme-thiolate protein [12]. Another challenge arose in 1991, when Shoun and coworkers described a P450 or heme-thiolate protein in the fungus Fusarium oxysporum that was involved in the

P450nor Enzymology and Biochemistry

355

reduction of nitrogen monoxide (NO) to dinitrogen oxide (N2 O) [13]. Even more surprising, the enzyme did not require a flavoprotein for electron transfer from NADH to the heme as in other bacterial cytochrome bc–containing NO-reductases [14], but used NADH directly for the reduction process according to the sum equation: 2NO + NADH + H+ → N2 O + NAD+ + H2 O There was little doubt that this NO-reductase, also called “P450nor ” or CYP55 [15], participated in the reduction of nitrate/nitrite leading to N2 O, since P450nor could be induced by nitrate and nitrite [16], thus completing the nitrate reduction pathway where nar, nir, and nor stand for nitrate, nitrite, and nitric oxide reductase, respectively: nar

nir

nor

− NO− 3 −→ NO2 −→ ·NO −→ N2 O

In 1996, two other isoenzymes of the P450nor family were isolated and cloned from the fungus Cylindrocarpon tonkinense [17,18] and another one from Trichosporum cutaneum [19,20]. Recent genome analyses have revealed the general presence of P450nor enzymes in fungi [21]. Since X-ray data are known for P450nor from Fusarium oxysporum, we shall concentrate on this isoform in our review. Emphasis will be put on the P450 nature of this enzyme and the requirement of the heme-thiolate bond in the mechanistic details of the NO-reduction process.

2. ISOLATION OF P450NOR AND MOLECULAR PROPERTIES P450nor was purified from a culture of Fusarium oxysporum (MT-811) and remained in the 1900 xg supernatant. It is therefore the first and so far the only soluble eukaryotic P450 protein [22]. The pure form of the 46-kD monomeric protein was obtained with a high recovery and showed an extremely high molecular activity of 31,500, which was much higher than those known from the bacterial respiratory NO-reductases. The enzyme was peculiar in requiring only NADH and no electron transport system for the reduction of NO. NADPH was much less active as the cosubstrate. The optical absorption spectra of the oxidized, reduced, and reduced-CO enzyme were typical for P450 proteins (Fig. 1A). Again, NADH could not serve as a reductant for the ferric enzyme, indicating that the reduction to N2 O occurred in a different way than in the cytochrome bc–containing bacterial NO-reductases. In agreement with the close similarity to other heme-thiolate proteins, the electron paramagnetic resonance (EPR) spectrum showed characteristic g-values of 2.44, 2.26, and 1.91 [23], which indicate a low-spin configuration at 60 K. At 5 K, a small high-spin signal at g = 797, 4.12 and 1.75 could also be detected (Fig. 1B). Titration of the ferric enzyme by photoreduction resulted in a midpoint potential of −307 mV, which is lower than that of the camphor monooxygenase P450CAM (−270 mV) [23]. P450CAM , as the best-described P450 monooxygenase system, also served as a reference with regard to the ligand-binding properties of P450nor . At the sixth ligand position, water seemed to be present in P450nor as well as in P450CAM , but nitrogenous ligands derived from imidazole or pyridine, which form stable complexes with reduced P450CAM , do not interact with

356

A. Daiber et al.

reduced P450nor [24]. In the oxidized ferric state, imidazole derivatives show spectral changes only in the mM concentration range. In contrast, good affinities were observed with alkylated isocyanides [24,25]. Such compounds, when bound to P450CAM , resulted in Soret bands at 455 nm, but some other P450s exhibit bands at 430 and 455 nm in a pH-dependent equilibrium. P450nor only showed a Soret band at 430 nm and therefore behaved differently from P450CAM . The two absorbance maxima, at 430 and 455 nm, were interpreted as indicating the binding of the isocyanides in two conformational states, but one could also consider a protonation of the thiolate ligand, leading to a sulfhydryl complex (430 nm) and the original 455-nm thiolate complex. The kinetics of carbon monoxide binding to ferrous P450nor showed higher kon rates for P450nor than for P450CAM , which was interpreted as a facilitated access for CO at P450nor [26]. This was a first indication of a quite large open space at the sixth coordination position, which was later confirmed by X-ray crystallography [27]. Of special interest was the binding of NO to the oxidized P450nor , since it could be expected that as a substrate, it

Absorbance (arbitrary unit)

406 (Fe2+)

(A)

400

414 (Fe3+)

540

431 (Fe3+NO) 2+

539

(Fe )

572 533 (Fe3+) 565

(Fe3+NO)

450

500

550

600

Wavelength (nm)

Fig. 1. (A) Absorption spectra of P450nor in various oxidation ligation states measured at 10 C with the stopped-flow rapid scan method: A ligand-free Fe3+ , an NO-bound Fe3+ , and a ligandfree Fe2+ spectrum. Reproduced from Shiro et al., J. Biol. Chem. 270, 1617–23 (1995). (B) EPR spectra of the ferric low-spin complexes of P450nor at 35 K. (1) Spectrum of the ferric resting enzyme dissolved in 0.1 M phosphate buffer at pH 7.2. (2) Spectrum of the cyanide (10 mM) complex dissolved in 0.1 M phosphate buffer containing 10% glycerol. (3) Spectrum of the metyrapone complex. This species gave the same spectrum independently of the buffer composition. Reproduced from Shiro et al., Biochemistry 34, 9052–58 (1995).

P450nor Enzymology and Biochemistry

357

(B) 2.442 2.260 1.911

1

2.307 2.592 1.821

2

2.264 2.561 1.846

3

300

400

Magnetic Field/mT

Fig. 1. (Continued)

would interact in a first step with the ferric enzyme. From titrations with NO-saturated solutions, a Soret band formed at 436 nm with kd of 20 M [28]. For the overall NO affinity, determined as the Km for N2 O formation, a value of 0.6 mM was observed [22], indicating that binding of the first NO molecule to the ferric iron was not the determinant of the rate of N2 O formation. This was later confirmed by mechanistic studies, which assigned the low affinity for NO to the second NO molecule that enters the reaction cycle [29]. When the stable ferric NO complex was subjected to infrared, resonance Raman [30], or X-ray absorption fine structure spectroscopy (EXAFS) [31,32], the N O stretching frequency was found at 1851 cm−1 and that of the Fe N frequency occurred at 530 cm−1 [33,34]. Corresponding values for P450CAM were measured as 1806 and 522 cm−1 , respectively. From these data, a stronger NO binding to P450nor was postulated and confirmed by the EXAFS data, which showed an FeN bond distance of 1.66 ± 0.02 Å compared to 1.76 ± 0.02 Å for P450CAM [35]. By this smaller distance, a facilitated 2p∗ electron transfer to the iron 3d orbital can be assumed, which would reduce the electron density at the NO ligand, which would produce, in the limit, Fe(II) NO+ . This unusual feature may provide the chemical basis for NO reduction, especially if it requires hydride transfer, as suggested from the absence of an electron transfer system. It is at least tempting to link the unusual properties of the P450nor NO complex to the crucial reduction step performed by NADH directly. The fact that NO serves as a substrate for the enzyme brought up the question of how, under in vivo conditions, the radical

358

A. Daiber et al.

process of NO oxidation by dioxygen could influence enzyme activity. It was then found that the enzyme was localized in the mitochondrial fraction [36], and therefore, it was concluded that the enzyme must be working under low oxygen tension. In addition, the fungus seems to use nitrate and nitrite reduction to NO as an additional source of energy when oxygen becomes limiting [37]. This largely prevents a toxifying autoxidation of NO under conditions of enzymatic activity. At the same time, the enzyme avoids the accumulation of NO in aerobic mitochondria and therefore fulfills an important task in the energy metabolism of this organism, since it protects the respiratory chain from NO inhibition [38].

3. GENE STRUCTURE OF P450NOR Deeper insight into the evolutionary and physiological aspects of P450nor came from work on the gene sequence. A cDNA clone was isolated by an immunoscreening method that revealed a polypeptide of 403 amino acid residues, corresponding to a molecular weight of 44,371 [15]. The full-length sequence later derived from the isolated protein contained an additional threonine at the N-terminus, but a second isoform was also detected with threonine and the adjacent methionine lacking [36]. This P450norB contained the N-terminus blocked by an acetylated alanine, which usually is seen after post- or cotranslational elimination of a methionine residue. This latter isoform was mainly found in the cytosol, whereas the longer form (P450norA ) was located to the particulate and mainly mitochondrial fraction. It was assumed that P450norA was synthesized with a presequence of 26 amino acids that targets it to mitochondria. This localization of the two isoforms to different compartments was mirrored in the fungus Cylindrocarpon tonkinense, but here two separate genes (CYP 55A2 and CYP 55A3) were found to code for the mitochondrial and cytosolic enzymes, respectively [18]. The cytosolic form was also more selective for NADPH, thus confirming the hypothesis that the toxicity of NO required its reduction and elimination from the mitochondrial as well as the cytosolic compartment. An additonal function could be that of an electron sink to allow NADP+ formation as a substrate of the pentose–phosphate cycle [39]. All known genes of the NOR family exhibited higher homologies against those of bacterial P450s than to those of eukaryotic origin. In particular, the bacterial protein SU2, which is a herbicide-inducible monooxygenase P450, has a 40% similarity to P450norA [40] (Fig. 2). Therefore, horizontal gene transfer from bacteria to fungi could have occurred very early in evolution. In P450nor , a highly conserved region around Cys 352 was identified as the heme-binding stretch forming the catalytic center [15,27]. Also, the I-helix motif −G/A-A-X-D/E-T- was found to be conserved which was interesting, since this is usually part of the oxygen- and substrate-binding pocket in the monooxygenase P450s [41]. Considering the very different function of P450nor , this similarity is quite surprising. With regard to the specific interaction of P450nor with NADH, a search for the NAD binding motif G-X-G-X-X-G/A revealed a similar sequence A-X-G-X-X-A found as consensus sequence in three P450nor species [15]. Structural data on the crystallized NADH-P450nor complex will soon prove or disprove the involvement of this motif in NADH binding (H. Shoun, work in progress).

P450nor Enzymology and Biochemistry –5

359 0

+5

+10

b

F S T G K R I C L G E G I A R N E L F L F

VIA1

F G D G P R N C I G M R F G K M Q S R L G

a1k

F N G G P R I C L G Q Q F A L T E A S Y V

BM–3

F G N G Q R A C I G Q Q F A L H E A T L V

14DM

F G G G R H R C I G E H F A Y C Q L G V L

CAM

F G H G S H L C L G Q H L A R R E I

chop

F G H G M H Q C L G R Q L A R I

SU1

F G F G V H Q C L G Q N L A R L E L E V I

SU2

F G F G V H Q C L G Q P L A R V E L Q I A

dNIR

F G F G D H R C I A E H L A K A E L T T V

I V T

E L R V A

Fig. 2. Comparison of the possible heme-binding domain in representatives of distinct P450 families. Standard one-letter symbols for amino acids are used. The sequence was numbered by centering the invariant cysteines at position 0, and signed plus and minus for the downstream and upstream directions, respectively. Invariant positions are shaded. Reproduced from Kizawa et al., J. Biol. Chem. 266, 10632–37 (1991).

Of great value with regard to the metabolic aspects of the NO-reductase pathway in eukaryotes has been the sequence analysis. P450nor constitutes an essential part of the nitrate reduction pathway, since the promotor region between nucleotides −526 and −515 contains the consensus sequence for nitrate responsiveness. A similar region has been found for the expression of the nitrate assimilation protein genes of Aspergillus nidulans by the transcription factor nirA [40]. Further downstream, the sequence between −118 and −107 is responsible for suppression of the CYP55 gene under aerobic conditions [42]. This consensus is also found as Rox1p responsive element in yeast, where it represses the anoxic protein genes under aerobic conditions. Such data allow the conclusion that the fungus adapts to the denitrifying conditions by a combination of nirA/Rox1p-like transcription factors. It was of great interest to find a fourth gene of the P450nor family in the yeast Trichosporum cutaneum (CYP55A4), which showed 65% similarity to the other three genes [20]. In contrast to P450nor , the yeast enzyme seemed to be induced independently of the presence of nitrate and nitrite, but rather dioxygen could act as an inducer. Thus, it may turn out that the presence and absence of dioxygen may even induce different P450nor isoenzymes in yeast.

4. MECHANISTIC STUDIES Very early after the discovery of its P450 nature, the mechanism of NO reduction to N2 O became a challenge. The absence of an electron-transferring prosthetic group or cofactor was puzzling. In view of the bacterial respiratory chain-associated NO-reductases [14],

360

A. Daiber et al.

it appeared logical to postulate a transfer of two electrons from NADH [28], although the biochemical properties of the reduced pyridine nucleotides as hydride donors had to be taken into consideration. A first indication of a possible mechanism was obtained by the spectral change that occurs after addition of NADH to the ferric P450 NO complex [28] (Fig. 3A). The new absorption band at 444 nm was assigned to an intermediate (I), which became the key to the mechanism. It could be excluded that I was the reduced Fe2+ NO complex. At +10 C, the 444-nm band decayed with a concomitant increase around 413 nm corresponding to the ferric enzyme (Fig. 3B). It is important to note that for these experiments, only a slight excess of NO over enzyme was used. If NO at 1 mM was added, the intermediate disappeared and the Fe3+ NO spectrum appeared at steady state. This suggested that NADH could react to I, allowing the following reaction sequence to be set up: NADH+H+

NO

Fe3+ NO −−−−−−→ Fe3+ NOH− + H+ −−→ Fe3+ + N2 O + H2 O The kinetics of Fe3+ NO formation could be determined by flash photolysis with kon = 26 × 107 M−1 s−1 at 10 C [28]. This is fast, and since with an excess of NO the intermediate was converted to product without showing significant steady state concentrations, the rate-limiting step should reside in the formation of I by reaction of NADH with the ferric NO complex. Assuming that NADH would react by hydride transfer, we synthesized 4,4-2 H,2 H-NADH and established a kinetic isotope effect of 3.8 ± 0.2 for NADH oxidation, indicating that indeed, at high NO levels, the removal of the hydrogen is rate limiting [29]. There was even some selectivity for the prochiral 4R hydrogen, since the 4R-2 H-NADH resulted in an isotope effect for NADH oxidation of 2.3 ± 0.2, whereas the 4S-2 H-NADH derivative showed a ratio of 1.7 ± 0.1 [29]. This could be confirmed by kinetic measurements for the formation of the 444-nm intermediate, which resulted in ratios of 2.7 ± 0.4 and 1.1 ± 0.1 for 4R-2 H and 4S-2 HNADH, respectively [29]. Such data would predict that an excess of NO would show the Fe3+ NO complex in the steady state, which indeed is the case. Conversely, low stationary concentrations of NO (<1 M/min) as can be generated by a decomposition of diethylamine NONOate (DENO), which first shifts the ferric enzyme to the Fe3+ NO complex, and after adding NADH, a constant amount of the enzyme is kept in the 444-nm intermediate state [29]. In a chemical approach, NADH could be replaced by the hydride donor sodium borohydride that, with the ferric NO complex, also formed an intermediate absorbing around 444 nm [29]. Concerning the nature of the 444-nm species, such experiments were in favor of hydride transfer to the NO group bound to the ferric heme (Fig. 4). This species was also formed when hydroxylamine radicals were generated by pulse radiolysis in the presence of the ferric enzyme [29]. On the basis of this observation, I could be a hydroxylamine species since the negative thiolate ligand could increase the pKa of the hydride adduct to allow simultaneous protonation. The radical resonance structure could explain the subsequent addition of ·NO to the intermediate. For this crucial reaction, the so-called “rebound” process would result in breaking of the Fe N bond after attack of the NO radical. After elimination of water, N2 O would be the stable product (Fig. 5) [43].

P450nor Enzymology and Biochemistry

361

(A) 0.25 1 1

431

Absorbance (arbitrary unit)

0.20

2

6

3 4

444

6

5 6

0.15

1

8 ms 13 ms 16 ms 21 ms 26 ms 36 ms

0.10

0.05

0.06 400

450

500

550

Wavelength (nm) (B) 0.25 412

7

1 2

Absorbance (arbitrary unit)

0.20

3 4 1

444

5 1

0.15

6 7

0.4 sec 7.6 sec 19.7 sec 39.7 sec 59.8 sec 79.8 sec 204.9 sec

7

0.10

0.05

0.12 400

450

500

550

Wavelength (nm)

Fig. 3. (A) Absorption spectral change of P450nor (Fe3+ NO) 8∼36 ms after mixing the enzyme solution with NADH (100 M) in the stopped-flow experiment at 10 C. The spectra measured at 8, 13, 16, 21, 26, and 36 ms after the NADH mixing are shown. The Soret band for the Fe3+ NO complex at 431 nm decreases in intensity, while that for the intermediate, I, at 444 nm concomitantly increases in intensity. An isosbestic point is located at 440 nm. The gate time for the measurement was 1 ms. (B) Absorption spectral change 0.4–204.9 s after the NADH mixing. A new isosbestic point is observed at 427 nm. The Soret absorption for I at 444 nm decreases in intensity, while that for the Fe3+ enzyme at 413 nm concomitantly increases in intensity. Reproduced from Shiro et al., J. Biol. Chem. 270, 1617–23 (1995).

362

A. Daiber et al. NH2 NH2

O

O O

N+ R

H

H

N R

H

O

N+

H N

Hydride

Fe(II)

Fe(II)

Transfer S–

S–

Fig. 4. Proposed mechanism of hydride transfer from NADH to the ferric NO complex (Fe(II)NO+ ) yielding the intermediate I, a ferrous HNO complex. The Fe N O group in the ferric NO complex is not linear, which may be a special feature of the P450nor (Fe3+ NO) complex as compared to other heme or porphyrin NO complexes. 415 nm (III)

Fe

432 nm H

•

NO

O

–H2O

H

(II)

Fe

N

(II)

+

Fe

O

+

N

O

NADH + H+ –N2O NAD+ H (III)

Fe

N NO

OH

•

NO

(III) •

Fe

H

H (II)

Fe

N

+

N

OH

OH 444 nm

Fig. 5. Proposed reaction mechanism for the catalytic cycle of the NADH-dependent reduction of NO by P450nor . The mesomeric structures describe realistic electronic configurations for the Fe(III) NO complex and the 444-nm intermediate (as proposed by Dr. Ann Walker and based on preliminary data of Dr. Frank Neese). The rate-limiting step is either the hydride transfer or the addition of the second NO molecule depending on the actual concentrations of NO and NADH. This scheme does not account for a role of the thiolate ligand as considerd in Fig. 10.

As outlined before, the rate limitation at low NO levels will shift to the last step, and this will be the case at physiological NO levels. Effective removal of NO, therefore, has to rely on a high turnover of the enzyme, since the affinity for NO in this last step (kd ∼600 M) is so low that formation of N2 O will occur linearly with the NO levels found in cells. Although the above-postulated mechanism satisfies all experimental data, there are still many details to be explored. Without doubt, the transfer of hydride from NADH to the Fe3+ NO complex is the most spectacular step, and certainly must require a strict steric coordination of both partners. The crystallization of the NADH-P450nor complex

P450nor Enzymology and Biochemistry

363

has been achieved, and the results from such X-ray data will greatly improve our understanding. It may be expected that NADH binds close to the Fe3+ NO position at the distal side with respect to the thiolate. In order to lower the activation energy for the hydride transfer, it would be feasible to form a hydrogen bridge from an amino acid to the NO-oxygen. Indeed, available data on the structure of P450nor can be interpreted this way, since the usually linear Fe NO group is tilted. Moreover, a network of water and hydrogen bridges seems to connect the entire active site, making protonation an important issue for the hydride/proton transfer step [33]. This will become clearer after taking a look at the so far available structural data.

5. REACTIONS WITH PEROXYNITRITE Peroxynitrite (PN) originates from a combination of nitric oxide (·NO) with superoxide (·O− 2 ), and therefore has to be considered a physiologically occurring but unstable intermediate under conditions of NO synthesis. It is anticipated that superoxide is always present at low levels from autoxidations, but the presence of Cu, Zn–, and Mn-superoxide dismutase (SOD) will tightly control the basal levels, allowing NO to regulate the activity of guanylyl cyclase as its main effector enzyme. Upon enzymatic generation of ·O− 2 by NADPH-oxidases, xanthine oxidase, mitochondria or uncoupled P450-enzymes (including NO synthases), the almost diffusion-controlled reaction with NO will generate PN and will limit the availability of NO. This Yin-Yang behavior of NO and ·O− 2 has been greatly substantiated by the observation that PN in submicromolar concentrations can inhibit prostacyclin (PGI2 ) synthase [44,45]. Since PGI2 synthase is a cytochrome P450 (heme-thiolate) enzyme, we had suggested a role of heme in this reaction, which has led to the discovery of a heme-catalyzed nitration of a tyrosine residue as a likely mechanism of inhibition [46]. Using model investigations, we were able to prove that heme proteins react with PN as shown in Fig. 6. The primary ferryl complex can be reduced to its ferric form by an endogenous tyrosine or by exogenously added phenols, which then as phenoxy radicals add the ·NO2 radical and form nitrated phenols as products. This reaction has been

–OONO + FeIII

OO heterolytic

cleavage homolytic

FeV = O + NO2–

FeIV = O + • NO2 ArOH

FeIV = O + • NO2

ArOH

FeIII + O2N–ArOH

Fig. 6. Proposed mechanism of Tyr-nitration by peroxynitrite catalyzed by heme proteins. Reproduced from Morgan et al., Drug Metabolism and Disposition 29, 1366–1376 (2001).

364

A. Daiber et al.

shown for cytochrome P450CAM to nitrate tyrosine residues at the active site [47], and also, the heme portion of P450BM-3 was nitrated in the vicinity of the heme [48]. The nitration was even higher when the active site located F87 in P450BM-3 was replaced by Tyr, indicating that the spatial factor is important for an effective nitration. This allows the conclusion that a Tyr residue occurs in the direct neighborhood of the hemeiron of PGI2 synthase, which is in agreement with a block of nitration by a substrate analog [44]. In view of the important regulatory role of PN for this P450 enzyme, we have further extended our studies on PN interaction to other heme-thiolate proteins. Of special interest was the recently crystallized P450NOR [22,49] for which NO is a natural substrate, being reduced to N2 O by NADH. The organism Fusarium oxysporum may also become exposed to superoxide and hence peroxynitrite may be a second substrate for the enzyme. We therefore investigated its reaction with PN and compared the data to those of other P450 enzymes and heme proteins. Nitration of tyrosine residues in a protein can be conveniently followed by Western blots with staining by anti-nitrotyrosine antibodies [50]. Such monoclonal or polyclonal antibodies are available, but they differ in their specificities and sensitivities depending on the probed protein and the antigen used for their generation. By this technique, PN-treated P450 proteins were found to be tyrosine-nitrated [46,51]. Surprisingly, we found that P450NOR only reacted positively at very high PN concentrations (Fig. 7, left side). If NO and ·O− 2 were released together from SIN-1 [52], generating a steady state concentration of PN, neither the polyclonal nor the monoclonal antibodies to 3-nitrotyrosine (3-NT) detected Tyr-nitrated P450NOR , but quite readily detected the nitrated forms of P450CAM and chloroperoxidase (CPO) (Fig. 7, right side).

P450cam (monoclonal)

P450cam (polyclonal)

CPO (monoclonal)

CPO (polyclonal)

P450nor (monoclonal)

P450nor (polyclonal)

0

25

50

100

PN [μM]

250

100

250

500

1000

Sin-1 [μM]

Fig. 7. Anti-nitrotyrosine Western blots of P450 proteins. Proteins (10 M) were nitrated at pH 7.4 by peroxynitrite or SIN-1, as indicated. After separation by SDS-PAGE, the proteins were blotted to a membrane and stained with a monoclonal or a polyclonal antibody to 3-NT from Upstate Biotechnologies Inc (1:1000). Complexes were detected using peroxidase conjugated goat antimouse (1:7500) or goat anti-rabbit (1:3000) IgG and an ECL kit from Pierce. Reproduced from Morgan et al., Drug Metabolism and Disposition 29, 1366–1376 (2001).

P450nor Enzymology and Biochemistry

365

In order to determine the exact quantity of 3-NT formed, complete pronase digests of PN-treated P450CAM and P450NOR were separated by HPLC and detected at 360 nm (Fig. 8). From such chromatograms, 3-NT could be detected in both P450 proteins, but at comparable bolus PN concentrations, P450NOR was three- to fourfold less nitrated. Using SIN-1 at a concentration of 500 M in the presence of 5 M P450 protein, the quantitation for P450CAM was 1.51 ± 0.03 M compared to 0.26 ± 0.11 M for

(A)

3-NT 4,000

+ 25 μM PN + 50 μM PN + 100 μM PN + 250 μM PN + 500 μM PN

3,500

Response [μV]

3,000 2,500 2,000 1,500 1,000 500 0 –500 2.5

5.0

7.5

10.0

12.5

15.0

17.5

20.0

Time [min] (B) + 50 μM PN + 100 μM PN + 250 μM PN + 500 μM PN + 1000 μM PN

3-NT 5,000

Response [μV]

4,000 3,000 2,000 1,000 0 2.5

5.0

7.5

10.0

12.5

15.0

17.5

20.0

Time [min]

Fig. 8. Pronase digestion of Tyr-nitrated P450 proteins. 5 M P450CAM (A) or P450NOR (B) were nitrated by increasing concentrations of peroxynitrite at pH 7.4 and digested with 2 mg/ml pronase for 12–14 h at 37 C. 100 l aliquots were subjected to isocratic separation on a C18 Nucleosil (250 × 4.6) 100-5 column followed by a washing gradient. Detection was at 365 nm. Reproduced from Morgan et al., Drug Metabolism and Disposition 29, 1366–1376 (2001).

366

A. Daiber et al.

P450NOR (2-h incubations in 0.1 M NaPi ; pH 7.4). In the mutant of P450BM-3 in which F87 was exchanged for Y, the extent of 3-NT reached 6.5 ± 0.5 M. The primary conclusion from the data is that P450NOR is less sensitive to PN with regard to an autocatalytic nitration of endogenous tyrosines. Reasons for this could be a lack of accessible Tyr residues, or a decreased reactivity with PN. The latter possibility could be ruled out since previous measurements had shown the highest rate of PN decomposition for P450NOR compared to P450BM-3 , P450CAM , or CPO [53]. We then checked the possibility that the enhanced turnover with PN involved a different route of decomposition, or a preferred isomerization of PN to nitrate [46]. Rapid mixing experiments, however, yielded the same transient ferryl intermediate (Compound II) as had been observed with P450BM-3 or CPO [46,48]. A final experiment was designed to confirm that P450NOR reacted faster with a second molecule of PN than other P450 proteins as judged from its rapid degradation of PN. When the decomposition of PN was carried out in the presence of phenol, the yields of 2- and 4-nitrophenols were highest with CPO and lower, but comparable with P450NOR and P450BM-3 [53]. This again confirmed that P450NOR behaved qualitatively similar to other P450 proteins, but reacted with faster kinetics. It can be assumed that the easy access of PN to the active site is the main reason for its enhanced turnover of PN compared with other P450 and heme proteins. In view of the high stability of Compound II of peroxidase in the presence of PN, it is likely that the reactivity of the ferryl complexes of P450 enzymes is high, and possibly due to the special properties of the S Fe O entity. In summary, nitric oxide and superoxide anion form PN, which can react with heme proteins and features especially high turnovers with heme-thiolate (P450) proteins. We studied the reactions of PN with various P450 enzymes as models for the Tyrnitration of prostacyclin synthase, which was found to be Tyr-nitrated at very low levels of PN. We report that of all P450 proteins tested, the P450-dependent NOreductase (P450NOR ) shows the highest rate of PN decomposition with a very low rate of auto-Tyr-nitration. The catalytic cycle involves a ferryl species that can either react with a second molecule of PN or with exogenous phenol. The open active site favors rapid kinetics, and the obvious absence of active site–located Tyr residues keeps autonitration low. This reaction of P450NOR may be of physiological significance. Thus, in addition to its catalysis of NO reduction by NADH, P450NOR may also be able to lower PN concentrations in Fusarium oxysporum when superoxide causes the extremely fast combination with the natural substrate NO to yield PN.

6. CRYSTALLOGRAPHY, X-RAY STRUCTURE, AND ENZYMOLOGY After overexpression and purification, the enzyme was crystallized by vapor diffusion using the sitting drop technique. The first attempt to resolve the structure of P450nor was started in 1994, but X-ray crystallographic studies did not result in suitable structural data [54]. Three years later, preliminary diffraction and electron paramagnetic resonance (EPR) studies were performed with a single crystal of P450nor using native and multiplewavelength anomalous diffraction data collection [49]. The system was determined to be orthorhombic with cell dimensions of a = 5499 Å, b = 8266 Å, c = 8721 Å

P450nor Enzymology and Biochemistry

367

and to belong to the P21 21 21 space group. According to these preliminary results, the crystal diffracted synchroton radiation at higher than 2.0 Å, and therefore was suitable for X-ray crystal structure analysis at atomic resolution. The first structure of P450nor was published in 1997, and this structure provided direct proof that, as predicted from sequence homologies, cysteine352 served as the axial ligand of the heme-iron, forming the heme-thiolate moiety [27]. In 2002, the structure of P450nor at atomic resolution was published [55]. The distance found between sulfur and iron was 2.48 Å, which is at variance with the calculated value of 2.26 Å, which was derived by X-ray absorption fine structure spectroscopy (EXAFS) spectroscopy. Similarly, the heme-thiolate-forming cysteine in cytochrome P450 camphor-5-monooxygenase (P450CAM ) is cysteine357 [56]. In contrast to P450CAM , the structure of P450nor revealed that the distal heme pocket is wide open to solvent, implicating this region as a possible NADH binding site. Ser286, Thr243, and Asp393, together with a water molecule adjacent to the iron form a hydrogen-bonding network to the protein surface of the distal pocket. This network was proposed to be essential for the proton delivery during the NO-reduction reaction, which requires charge compensation of the transferred NADH-hydride [33]. The most reasonable NADH binding site, as well as the heme-thiolate moiety and the most frequently mutated amino acids Ser286 and Thr243, are highlighted in the structure of P450nor (Fig. 9).

F,G-loop

B′-helix I-helix Thr243

Ser286

Cys352

Fig. 9. Crystal structure of P450nor in its resting ferric state. The residues Ser286 and Thr243 are highlighted in red since these amino acids were subject to numerous mutation experiments to prove their involvement in the proton delivery being essential for the enzyme activity. The proximal ligand of the iron, Cys352, is highlighted in yellow. Moreover, the distal I- and B -helices as well as the F,G-loop are marked. The structure was rendered from the protein database file “1ROM” using the PyMol Molecular Graphics System (version 0.93) from DeLano Scientific LLC. (see Plate 8.)

368

A. Daiber et al.

In parallel, much effort has been spent on site-directed mutagenesis at the distal region of P450nor , especially of the conserved threonine (Thr243) of the distal helix [31,57]. A structural comparison of P450nor , P450CAM , and cytochrome P450 bacterial monooxygenase 3 (P450BM-3 ) revealed that this threonine in the distal I-helix is conserved in all three proteins. Replacement of threonine243 by Asn, Val, and Leu moderately decreased the NO-binding rate constant, whereas Gln, Ile, Tyr, Trp, and Lys dramatically decreased it. The rate of NADH consumption, N2 O release, as well as the 444-nm intermediate formation was significantly reduced upon replacement of threonine243 by Asn, Gly, and Ser and almost completely suppressed upon replacement by any other amino acid [34]. Since only the Thr243 to Lys mutant exhibited an absorption spectrum characteristic of an inhibitory nitrogenous ligand-bound form, another explanation for the importance of Thr243 for NO reduction had to be postulated. The already-discussed hydrogen-bonding network, which was suggested to be essential for proton delivery, requires the presence of Thr243 as well as Ser286, and seems to be an attractive explanation [33,34]. This hypothesis could be further substantiated by the finding that replacement of Ser286 by Val or Thr resulted in a significant decrease in the formation rate of the 444-nm intermediate, but not in the rate constant for the formation of the Fe(III) NO complex [58]. However, steric aspects could also play an important role in determining the rate constant of NO reduction. The replacement of Thr243, especially by amino acids with longer side chains, could have negative effects on NADH-binding and hydride transfer to the NO-hemin complex. The impact of Thr243 replacement by Asn, Val, or Ala on the Fe NO and Fe CO stretching and the Fe C O bending frequencies is small, and the differences range from 0 to 3 cm−1 [34]. The replacement of Ser286 by Val caused a marked shift in the the proximal NO Fe distance from 2.31 to 2.37 Å, and also of the Fe N O angle from 161 to 165 , whereas replacement by Thr caused only marginal changes [33]. Additional insight into the binding properties of nitrogenous ligands to P450nor was obtained by the structural characterization of n-butyl-isocyanide complexes of P450nor [32]. A detailed structural comparison of P450nor with P450CAM , P450BM-3 , cytochrome P450-terp (cytochrome P450 108, P450terp ), and 6-deoxyerythronolide B hydroxylase (cytochrome P450 107A1, P450eryF ) revealed that the secondary structural elements were very similar, and showed the same overall fold [58]. To quantify this similarity, the root mean square (r.m.s.) deviations of the helical C atoms were estimated and the resulting values were 1.301 for P450CAM , 1.158 for P450BM-3 , 1.069 for P450terp , and 1.202 for P450eryF . These relatively small values quantitatively substantiated the structural similarity in the overall fold pattern among the P450s. Some structural features are conserved in the P450 enzymes, while some are significantly different, possibly reflecting the functional differences between the enzymes. The iron–thiolate distance for P450nor is 2.26 Å, which is comparable to those of other P450s: 2.25 Å for P450CAM , 2.05 Å for P450BM-3 , 2.15 Å for P450terp , and 2.17 Å for P450eryF [58]. In all P450s, the iron-coordinating thiolate is surrounded by the N-terminal end of the L-helix, an antiparallel -pair called the cysteine ligand loop or -bulge. The F/G-helices of P450nor are characteristically different from other P450s. The hydrophilic region of the I-helix pushes the hydrophilic region of the F/G-helices. Consequently, the G-helix is moved away from the heme site, and the loop between the F/G-helices becomes positioned just above the I-helix, making the heme pocket more open than those of the other P450s.

P450nor Enzymology and Biochemistry

369

The B -helix, which is thought to be important in controlling substrate specificity, is located about 22 Å above the heme iron, and its axis is oriented approximately parallel to both the G-helical axis and the heme plane. Beneath the B -helix, some positively charged residues (Lys and Arg) are present [59]. The charge distribution on the distal and proximal surface is thus completely different in P450nor as compared to P450CAM and P450BM-3 [33]. A unique accumulation of positively charged amino acids (Lys62, Arg64, Lys291, and Arg392) is present beneath the B -helix in the heme distal pocket of P450nor , whereas the corresponding area in P450CAM and P450BM−3 are occupied by side chains of some hydrophobic residues [58]. Since NO reduction by P450nor requires no additional redox protein mediators, but relies on the presence of NADH as an electron donor, it may be suggested that P450nor itself has an NADH binding site. Therefore, the above-described positively charged cluster was suggested to represent the NADH binding site, since no classical NADH binding motif was found. Although the structure of P450nor with bound NADH is not yet available at this point in time, the crystal structure of P450nor with bound bromide ions is in accordance with the suggested NADH binding site beneath the B -helix [60]. Support for this hypothesis came from recent mutagenesis work on amino acids of the B -helix, which clearly showed that NADH binding depends on the charge distribution and steric characteristics in this region [59]. Only a high-resolution structure of the NADH-P450nor complex will definitely prove that the 4H of NADH, most likely in the prochiral R state, will bind close to the bound NO group in order to allow hydride transfer, with concomitant proton transfer along the Ser-water pathway. This situation of such a suitable enzyme-substrate complex leading to the 444-nm intermediate I is depicted in Fig. 5. In a recent study, Oshima et al. succeeded in determining the crystal structure of P450nor in a complex with an NADH analog, nicotinic acid adenine dinucleotide, which provides conclusive evidence for the mechanism of the unprecedented electron transfer [61]. Comparison of the structure with those of dinucleotide-free forms revealed a global conformational change accompanied by intriguing local movements caused by the binding of the pyridine nucleotide. Arg64 and Arg174 fix the pyrophosphate moiety upon the dinucleotide binding. Stereo-selective hydride transfer from NADH to NObound heme was suggested from the structure, the nicotinic acid ring being fixed near the heme by the conserved Thr residue in the I-helix and the upward-shifted propionate side chain of the heme. A proton channel near the NADH channel is formed upon the dinucleotide binding, which should direct continuous transfer of the hydride and proton. A salt-bridge network (Glu71-Arg64-Asp88) was shown to be crucial for a high catalytic turnover. The nicotinic acid ring in the NAAD complex binds in close vicinity to the heme in a stereoselective manner, the pro-R side of its C4 atom facing the NO molecule bound to the heme. The results are consistent with our previous observations concerning the significant kinetic isotope effect of [4R-2 H]-NADH on the reduction step yielding form I as well as the successful formation of I upon reduction of the Fe3+ NO complex with borohydride [29]. The present and previous results strongly support the direct transfer of a hydride ion from the pro-R side of NADH to the Fe3+ NO complex to form intermediate I. The side chain of Ser286 is located closest to the C4 atom of nicotinic acid or the oxygen atom of NO bound to the heme (Fig. 10).

370

A. Daiber et al. solvent

O Asp393

P

O–

124

2.7 Å

NAAD P

2.8 Å 30

O 287 2.8 Å

2.8 Å

O

261 Ser286

2.8 Å

15

N

3.4 Å

OH 3.3 Å

Heme

solvent

2.7 Å

C4

1.9 Å O N Fe

2.9 Å

63

3.1 Å

3.1 Å HO Thr243

Fig. 10. NADH and proton channels. A representation of the hydrogen-bonding network of the NAAD complex. Reproduced from Oshima et al., J. Mol. Biol. 342, 207–217 (2004).

7. IMPLICATIONS AND OUTLOOK The finding of P450nor and its homologs in a broad spectrum of eukaryotes certainly is of interest to those biologists working on nitrate/nitrite reduction pathways and their regulation. N2 O is released in considerable quantities from P450nor by nitrate-metabolizing organisms, and it was even speculated that the 200-fold higher greenhouse effect of N2 O versus CO2 may contribute to global warming [21]. Taking into consideration the wide use of nitrates as fertilizers, the global release of N2 O may continuously increase. On a cellular level, NO is a toxic agent as higher concentrations accumulate. Therefore, P450nor acts as a detoxifying enzyme for NO, but it may also react and detoxify peroxynitrite as the very aggressive product of NO and O− 2 [46,53]. Its unfavorable affinity for NO is compensated by its high turnover, which requires very fast kinetics at all steps of the cycle. Since rate limitation under physiological NO levels is at the step of NO attack on the intermediate I, the details of its catalysis, which allow a minimization of the activation energy, will be of utmost interest. Answers are expected to come not only from the structural data of the NADH-P450nor complex, but also from spectroscopic studies on the intermediate I, which are underway using Mössbauer, magnetic circular dicroism (MCD), and EPR spectroscopy [Neese, in preparation]. If the structure of the proposed hydroxylamine ligand can be confirmed, the interest will not only concentrate on the hydride transfer, but also on proton transfer. Although this seems to be no obvious difficulty, it may be of crucial importance to have the proton present during hydride transfer for a low activation energy. The slightly tilted configuration of the Fe N O axis (as observed by IR and Raman spectroscopy) may indicate hydrogen bonding by Thr286, allowing the nitrogen at the iron to increase its electrophilic nature. In the context of P450 catalysis, we would like to emphasize an apparent analogy of the P450nor mechanism to the classical P450-dependent monooxygenase or isomerase reactions [54].

P450nor Enzymology and Biochemistry

371

Intermediate I at the first sight looks different from the Compound II species postulated for monooxygenases and thromboxane and prostacyclin synthase [54]. However, for intermediate I also an electron configuration can be formulated, which represents a formally Fe(IV) species (Fig. 11). At the limit, a thiyl radical can arise, allowing electron transfer from the second ·NO to the iron–sulfur entity. This completes the analogy to the other heme-thiolate mechanisms (Fig. 11). Common to all heme-thiolate catalytic mechanisms would be an extremely fast electron transfer from the adjacent radical to the Fe(IV)–thiolate ( Fe(III)–thiyl) redox center. This could be formulated as a concerted attack of the radical at the Fe O/Fe N bond, but also as fast electron transfer to the thiyl radical–ferric iron–porphyrin redox system [62]. Assuming that the latter reacts as a semiconductor similar to other iron–sulfur

P450nor

•NO OH (IV) S Fe N

•NO OH (III)

S• Fe

H

N H

O • TxA2synthase

(IV)

S Fe

• R O

(III)

S• Fe

R2

R1

O

•

R1

O H

PGI2synthase

(IV)

S Fe

• R O

(III)

S• Fe

R2

• CR3

• CR3 Monooxygenases

O

(IV)

S Fe

OH

(III)

S• Fe

OH

Fig. 11. Postulated structures of Compound II analogs of several P450 enzymes. The iron-thiolate binding is presented as a covalent bond in order to indicate the electron transfer from the thiolate to the ferryl heme. At the limit, the electron is transferred completely generating the thiyl resonance structure. A third structure with ferrous iron is not shown, although this seems to be most realistic for P450nor (Dr. Ann Walker and Dr. Frank Neese, personal communications).

372

A. Daiber et al.

centers in biology, the activation energy of electron transfer could be lower than that for cleavage of the Fe O/Fe N bond. For the P450nor reaction mechanism, the attacking NO radical certainly is less reactive than the carbon radicals of the isomerases and the monooxygenases, thus favoring the electron transfer hypothesis for the last step in P450nor catalysis. To prove this detail of the mechanism may appear of minor importance, but it may be a clue to the role of the thiolate ligand in P450 catalysis.

8. LATEST RESULTS ON THE MOLECULAR MECHANISM OF P450NOR BASED ON COMPUTATIONAL CALCULATIONS Recent analysis from Lehnert and coworkers could mainly confirm the aforepostulated mechanism [63]: From the work of Shoun and coworkers, it is known that P450nor is catalytically active in its ferric oxidation state, which binds one molecule of NO. The next step is the reaction with NADH leading to the formation of intermediate I as presented above. As shown by our analysis presented above and kinetic studies by Daiber et al., this step corresponds to a direct hydride transfer initially generating an Fe(II) HNO species as shown in Fig. 12. This species is N-protonated, which is energetically strongly

S Fe3+

NO

NO

S

P450nor

Fe

NADH

2 e–

NAD+

O S Fe

NO

N 5:

S

Fe

+

N O

N O

H+

2: +46.6 kcal/mol

+21.2 kcal/mol

2+

S Fe

H

2+

N

S Fe

N O

H+

O 3+

O

3+ +

S Fe

H

N O

N O 6a: –21.3 kcal/mol

N

H

H

3b: +26.2 kcal/mol H+

H+

H

S Fe N

O

3a: 0.0 kcal/mol

NO

4+

H

S Fe N NO

H

O 4: –8.6 kcal/mol (Intermediate ‘I’)

7a: –20.4 kcal/mol

H+ Transition state

H H O– 3+ S Fe N

3+

S Fe

O N

H

N O N O

O

H

7b: –36.9 kcal/mol

S Fe

N

3+

H

7c N O

S Fe 8

+ N2O + H2O

–89.6 kcal/mol

6b: +3.7 kcal/mol

Fig. 12. Calculated intermediates in the catalytic cycle of P450nor . Reproduced from Lehnert et al., J. Comput. Chem. 27, 1338–1351 (2006).

P450nor Enzymology and Biochemistry

373

favored over the O-protonated form. Hence, in the following mechanistic discussion, all free energies will be related to this species 3a, which is set to 0.0 kcal/mol in Fig. 12. This species is subsequently protonated, leading to the fast formation of the corresponding Fe NHOH species 4, which therefore can be identified with intermediate I as observed spectroscopically in agreement with the proposal of Daiber et al. It is stabilized by approximately 9 kcal/mol relative to 3a. The electronic structure of this intermediate is very interesting as described above, and formally corresponds to an Fe(IV) NHOH− complex. Intermediate 4 could be envisioned to react with NO by an outer sphere electron transfer to the iron center creating Fe(III) and NO+ , where the NO+ then subsequently or concertedly attacks the lone pair on the nitrogen of the coordinated NHOH ligand. This process corresponds to a spin-allowed reaction. Importantly, the calculations show that the N-protonated Fe(III) N2 O2 H2 species formed this way is unstable and easily loses the generated N2 O2 H2 ligand. However, proton transfer from the nitrogen atom of the former NHOH ligand to the oxygen of the incoming NO molecule leads to the generation of the stable O,O -protonated Fe(III) N2 O2 H2 complex 7b as shown in Fig. 12. The formation of 7b from 4 and NO is highly exergonic by about 28 kcal/mol. This is a very important result, because it shows that the attack of NO on intermediate 4 is a three-step process including (i) electron transfer to the formal iron(IV), (ii) attachment of the resulting NO+ forming an N N bond, and (iii) simultaneous proton transfer from the N atom of the former NHOH ligand to the O atom of the incoming NO. This mechanism explains why the anticipated radical reaction of NO with the coordinated HNO species is efficiently catalyzed by the heme center. On the basis of this reaction sequence, the N-protonated Fe(III) N2 O2 H2 species is an intermediate on the reaction coordinate leading to 7b. To estimate the energy of this species, we have optimized its structure using a fixed Fe N distance. This leads to intermediate 7a, where NO is loosely attached to the NHOH ligand. The energy of this species is −12 kcal/mol relative to 4 and NO. Therefore, the subsequent proton transfer generating the tautomer 7b is highly exergonic by −16 kcal/mol, and hence, should occur extremely fast in the actual enzymatic reaction. This ensures that the formed N2 O2 H2 ligand stays bound to the iron(III) center. In addition, the structure of 7b as shown in Fig. 12 is perfectly set up for proton transfer leading to the tautomer 7c, where the oxygen of the former NHOH ligand is doubly protonated. Upon geometry optimization, this species relaxes back to 7b, which therefore corresponds to the most stable tautomer of the Fe(III) N2 O2 H2 complex. However, fixing the two O H bonds in 7c at 0.97 Å and geometry optimization leads to the spontaneous cleavage of the N OH2 bond and the formation of a five-coordinate Fe(III) complex, nitrous oxide, and water. This reaction is highly exergonic, which shows that 7c must corresponds to a transition state, because this species is higher in energy than both the reactant 7b that it is formed from and the products obtained after its decomposition. Therefore, we tried to optimize the structure of this transition state, but these calculations were not successful. To estimate the energy for proton transfer leading from 7b to 7c, a geometry optimization was performed where both O H and the N O bond distances were fixed at 0.97 and 1.41 Å, respectively. As shown, this structure is 35 kcal/mol higher in energy than 7b. This indicates that the transition state must have a significantly elongated N O bond, which would lead to a substantial lowering of the barrier as evident from the strong thermodynamic driving force of the reaction. Because it is known from experiments that the five-coordinate ferric heme in P450nor is high spin, we therefore used the corresponding high-spin complex 8 to calculate the free energy

374

A. Daiber et al.

for the reaction leading from 7b to 8, nitrous oxide, and water. This process is exergonic by about 53 kcal/mol, and therefore, thermodynamically highly favorable. Note that the decay of 7c initially generates an Fe(III) N2 O complex, which loses the N2 O ligand in the calculations regardless of whether the high-spin or low-spin state of the ferric heme is considered. This dissociation of nitrous oxide closes the catalytic cycle.

ACKNOWLEDGMENTS We thank Drs. Ann Walker and Frank Neese for helpful contributions and corrections. We are also indebted to Dr. Walter G. Zumft for proofreading of the manuscript. The mechanistic contributions by A.D. and V. U. were supported by a grant from the German Research Foundation (DFG), priority program “Radicals in Enzyme Catalysis” (Ul36/25-1) and the “Fonds der Chemischen Industrie.”

REFERENCES [1] Klingenberg, M. (1958) Pigments of rat liver microsomes. Arch. Biochem. Biophys., 75, 376–386. [2] Estabrook, R.W., Cooper, D.Y., and Rosenthal, O. (1963) The light reversible carbon monoxide inhibition of the steroid C21-Hydroxylase system of the adrenal cortex. Biochem. Z, 338, 741–755. [3] Omura, T. and Sato, R. (1964) The Carbon Monoxide-Binding Pigment of Liver Microsomes. I. Evidence for Its Hemoprotein Nature. J. Biol. Chem., 239, 2370–2378. [4] Hill, R., Röder, A., and Williams, R.J.P. (1970) Structure Bonding (Berlin), 8, 123. [5] Ullrich, V., Ruf, H.H., and Wende, P. (1977) Croat. Chem. Acta, 49, 213–222. [6] Newcomb, M., Shen, R., Choi, S.-Y. et al. (2000) J. Am. Chem. Soc., 122, 2677–2686. [7] Guengerich, F.P. and MacDonald, T.L. (1990) Mechanisms of cytochrome P-450 catalysis. Faseb. J., 4, 2453–2459. [8] Ullrich, V. (1980) J. Mol. Catal., 7, 159–167. [9] Ullrich, V. and Brugger, R. (1994) Angew. Chemie Int. Ed., 33, 1911–1919. [10] Ogliaro, F., Harris, N., Cohen, S. et al. (2000) J . Am. Chem. Soc., 122, 8977–8989. [11] Hecker, M. and Ullrich, V. (1989) On the mechanism of prostacyclin and thromboxane A2 biosynthesis. J. Biol. Chem., 264, 141–150. [12] Song, W.C. and Brash, A.R. (1991) Purification of an allene oxide synthase and identification of the enzyme as a cytochrome P-450. Science, 253, 781–784. [13] Shoun, H. and Tanimoto, T. (1991) Denitrification by the fungus Fusarium oxysporum and involvement of cytochrome P-450 in the respiratory nitrite reduction. J. Biol. Chem., 266, 11078–11082. [14] Zumft, W.G. (1997) Cell biology and molecular basis of denitrification. Microbiol. Mol. Biol. Rev., 61, 533–616. [15] Kizawa, H., Tomura, D., Oda, M. et al. (1991) Nucleotide sequence of the unique nitrate/nitrite-inducible cytochrome P-450 cDNA from Fusarium oxysporum. J. Biol. Chem., 266, 10632–10637. [16] Shoun, H., Suyama, W., and Yasui, T. (1989) Soluble, nitrate/nitrite-inducible cytochrome P-450 of the fungus, Fusarium oxysporum. FEBS Lett., 244, 11–14. [17] Usuda, K., Toritsuka, N., Matsuo, Y. et al. (1995) Denitrification by the fungus Cylindrocarpon tonkinense: Anaerobic cell growth and two isozyme forms of cytochrome P-450nor. Appl. Environ. Microbiol. 61, 883–889.

P450nor Enzymology and Biochemistry

375

[18] Kudo, T., Tomura, D., Liu, D.L. et al. (1996) Two isozymes of P450nor of Cylindrocarpon tonkinense: Molecular cloning of the cDNAs and genes, expressions in the yeast, and the putative NAD(P)H-binding site. Biochimie., 78, 792–799. [19] Tsuruta, S., Takaya, N., Zhang, L. et al. (1998) Denitrification by yeasts and occurrence of cytochrome P450nor in Trichosporon cutaneum. FEMS Microbiol. Lett., 168, 105–110. [20] Zhang, L., Takaya, N., Kitazume, T. et al. (2001) Purification and cDNA cloning of nitric oxide reductase cytochrome P450nor (CYP55A4) from Trichosporon cutaneum. Eur. J. Biochem., 268, 3198–3204. [21] Shoun, H., Kano, M., Baba, I. et al. (1998) Denitrification by actinomycetes and purification of dissimilatory nitrite reductase and azurin from Streptomyces thioluteus. J. Bacteriol., 180, 4413–4415. [22] Nakahara, K., Tanimoto, T., Hatano, K. et al. (1993) Cytochrome P-450 55A1 (P-450dNIR) acts as nitric oxide reductase employing NADH as the direct electron donor. J. Biol. Chem., 268, 8350–8355. [23] Shiro, Y., Fujii, M., Isogai, Y. et al. (1995) Iron-ligand structure and iron redox property of nitric oxide reductase cytochrome P450nor from Fusarium oxysporum: Relevance to its NO reduction activity. Biochemistry, 34, 9052–9058. [24] Imai, Y., Okamoto, N., Nakahara, K., and Shoun, H. (1997) Absorption spectral studies on heme ligand interactions of P-450nor. Biochim. Biophys. Acta, 1337, 66–74. [25] Tomita, T., Ogo, S., Egawa, T. et al. (2001) Elucidation of the differences between the 430and 455-nm absorbing forms of P450-isocyanide adducts by resonance Raman spectroscopy. J. Biol. Chem., 276, 36261–36267. [26] Shiro, Y., Kato, M., Iizuka, T. et al. (1994) Kinetics and thermodynamics of CO binding to cytochrome P450nor. Biochemistry, 33, 8673–8677. [27] Park, S.Y., Shimizu, H., Adachi, S. et al. (1997) Crystal structure of nitric oxide reductase from denitrifying fungus Fusarium oxysporum. Nat. Struct. Biol., 4, 827–832. [28] Shiro, Y., Fujii, M., Iizuka, T. et al. (1995) Spectroscopic and kinetic studies on reaction of cytochrome P450nor with nitric oxide. Implication for its nitric oxide reduction mechanism. J. Biol. Chem., 270, 1617–1623. [29] Daiber, A., Nauser, T., Takaya, N. et al. (2002) Isotope effects and intermediates in the reduction of NO by P450(NOR). J. Inorg. Biochem., 88, 343–352. [30] Singh, U.P., Obayashi, E., Takahashi, S. et al. (1998) The effects of heme modification on reactivity, ligand binding properties and iron-coordination structures of cytochrome P450nor. Biochim. Biophys. Acta, 1384, 103–111. [31] Okamoto, N., Imai, Y., Shoun, H., and Shiro, Y. (1998) Site-directed mutagenesis of the conserved threonine (Thr243) of the distal helix of fungal cytochrome P450nor. Biochemistry, 37, 8839–8847. [32] Lee, D.S., Park, S.Y., Yamane, K. et al. (2001) Structural characterization of n-butylisocyanide complexes of cytochromes P450nor and P450cam. Biochemistry, 40, 2669–2677. [33] Shimizu, H., Obayashi, E., Gomi, Y. et al. (2000) Proton delivery in NO reduction by fungal nitric-oxide reductase. Cryogenic crystallography, spectroscopy, and kinetics of ferric-NO complexes of wild-type and mutant enzymes. J. Biol. Chem., 275, 4816–4826. [34] Obayashi, E., Shimizu, H., Park, S.Y. et al. (2000) Mutation effects of a conserved threonine (Thr243) of cytochrome P450nor on its structure and function. J. Inorg. Biochem., 82, 103–111. [35] Singh, U.P., Obayashi, E., Takahashi, S. et al. (1998) Biochim. Biophys. Acta, 1384, 103–111. [36] Nakahara, K., and Shoun, H. (1996) N-terminal processing and amino acid sequence of two isoforms of nitric oxide reductase cytochrome P450nor from Fusarium oxysporum. J. Biochem. (Tokyo), 120, 1082–1087. [37] Kobayashi, M., Matsuo, Y., Takimoto, A. et al. (1996) Denitrification, a novel type of respiratory metabolism in fungal mitochondrion. J. Biol. Chem., 271, 16263–16267.

376

A. Daiber et al.

[38] Tanimoto, T., Nakahara, K., and Shoun, H. (1992) Biosci. Biotech. Biochem., 56, 2058–2059. [39] Watsuji, T.O., Takaya, N., Nakamura, A., and Shoun, H. (2003) A possible role of NADPHdependent cytochrome P450nor isozyme in glycolysis under denitrifying conditions. Biosci. Biotechnol. Biochem., 67, 1109–1114. [40] Tomura, D., Obika, K., Fukamizu, A., and Shoun, H. (1994) Nitric oxide reductase cytochrome P-450 gene, CYP 55, of the fungus Fusarium oxysporum containing a potential binding-site for FNR, the transcription factor involved in the regulation of anaerobic growth of Escherichia coli. J. Biochem. (Tokyo), 116, 88–94. [41] Poulos, T.L., Finzel, B.C., and Howard, A.J. (1987) High-resolution crystal structure of cytochrome P450cam. J. Mol. Biol., 195, 687–700. [42] Zitomer, R.S. and Lowry, C.V. (1992) Regulation of gene expression by oxygen in Saccharomyces cerevisiae. Microbiol. Rev., 56, 1–11. [43] Silaghi-Dumitrescu, R. (2003) Nitric oxide reduction by heme-thiolate enzymes (P450nor): A reevaluation of the mechanism. Eur. J. Inorg. Chem., 6, 1048–1052. [44] Zou, M., Martin, C., and Ullrich, V. (1997) Tyrosine nitration as a mechanism of selective inactivation of prostacyclin synthase by peroxynitrite. Biol. Chem., 378, 707–713. [45] Zou, M., Jendral, M., and Ullrich, V. (1999) Prostaglandin endoperoxide-dependent vasospasm in bovine coronary arteries after nitration of prostacyclin synthase. Br. J. Pharmacol., 126, 1283–1292. [46] Mehl, M., Daiber, A., Herold, S. et al. (1999) Peroxynitrite reaction with heme proteins. Nitric Oxide, 3, 142–152. [47] Daiber, A., Schoneich, C., Schmidt, P. et al. (2000) Autocatalytic nitration of P450CAM by peroxynitrite. J. Inorg. Biochem., 81, 213–220. [48] Daiber, A., Herold, S., Schoneich, C. et al. (2000) Nitration and inactivation of cytochrome P450BM-3 by peroxynitrite. Stopped-flow measurements prove ferryl intermediates. Eur. J. Biochem., 267, 6729–6739. [49] Park, S.Y., Shimizu, H., Adachi, S. et al. (1997) Crystallization, preliminary diffraction and electron paramagnetic resonance studies of a single crystal of cytochrome P450nor. FEBS Lett., 412, 346–350. [50] Ye, Y.Z., Strong, M., Huang, Z.Q., and Beckman, J.S. (1996) Antibodies that recognize nitrotyrosine. Methods Enzymol., 269, 201–209. [51] Daiber, A., and Ullrich, V. (2002) Peroxynitrite reactions with heme and heme-thiolate (P450) proteins. Methods Enzymol., 359, 379–389. [52] Darley-Usmar, V.M., Hogg, N., O’Leary, V.J. et al. (1992) The simultaneous generation of superoxide and nitric oxide can initiate lipid peroxidation in human low density lipoprotein. Free Radic. Res. Commun., 17, 9–20. [53] Zou, M.H., Daiber, A., Peterson, J.A. et al. (2000) Rapid reactions of peroxynitrite with heme-thiolate proteins as the basis for protection of prostacyclin synthase from inactivation by nitration. Arch. Biochem. Biophys., 376, 149–155. [54] Nakahara, K., Shoun, H., Adachi, S. et al. (1994) Crystallization and preliminary X-ray diffraction studies of nitric oxide reductase cytochrome P450nor from Fusarium oxysporum. J. Mol. Biol., 239, 158–159. [55] Shimizu, H., Park, S.Y., Shiro, Y., and Adachi, S. (2002) X-ray structure of nitric oxide reductase (cytochrome P450nor) at atomic resolution. Acta Crystallogr. D Biol. Crystallogr., 58, 81–89. [56] Schlichting, I., Berendzen, J., Chu, K. et al. (2000) The catalytic pathway of cytochrome p450cam at atomic resolution. Science 287, 1615–1622. [57] Okamoto, N., Tsuruta, K., Imai, Y. et al. (1997) Fungal P450nor: Expression in Escherichia coli and site-directed mutageneses at the putative distal region. Arch. Biochem. Biophys., 337, 338–344.

P450nor Enzymology and Biochemistry

377

[58] Shimizu, H., Park, S., Lee, D. et al. (2000) Crystal structures of cytochrome P450nor and its mutants (Ser286→Val, Thr) in the ferric resting state at cryogenic temperature: A comparative analysis with monooxygenase cytochrome P450s. J. Inorg. Biochem., 81, 191–205. [59] Zhang, L., Kudo, T., Takaya, N., and Shoun, H. (2002) The B helix determines cytochrome P450nor specificity for the electron donors NADH and NADPH. J. Biol. Chem., 277, 33842–33847. [60] Kudo, T., Takaya, N., Park, S.Y. et al. (2001) A positively charged cluster formed in the heme-distal pocket of cytochrome P450nor is essential for interaction with NADH. J. Biol. Chem., 276, 5020–5026. [61] Oshima, R., Fushinobu, S., Su, F. et al. (2004) Structural evidence for direct hydride transfer from NADH to cytochrome P450nor. J. Mol. Biol., 342, 207–217. [62] Ullrich, V. (2003) Thoughts on thiolate tethering. Tribute and thanks to a teacher. Arch. Biochem. Biophys., 409, 45–51. [63] Lehnert, N., Praneeth, V.K., and Paulat, F. (2006) Electronic structure of iron(II)–porphyrin nitroxyl complexes: Molecular mechanism of fungal nitric oxide reductase (P450nor). J. Comput. Chem., 27, 1338–1351.

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 15

Nitric Oxide Interaction with Insect Nitrophorins and Possibilities for the Electron Configuration of the {FeNO}6 Complex F. Ann Walker Department of Chemistry, University of Arizona, Tucson, AZ 85721–0041

Abstract The nitrophorins are NO-carrying heme proteins that are found in the saliva of two species of blood-sucking insects, the kissing bug (Rhodnius prolixus) and the bedbug (Cimex lectularius). In both insects NO is bound to the ferric form of the protein, which gives rise to Kd s in the micromolar to nanomolar range, and thus upon injection of the saliva into the tissues of the victim the NO can dissociate to cause vasodilation and inhibition of platelet aggregation. The structures of the proteins from each of these insects are unique, and each has a large component of -sheet structure, which is unusual for heme proteins. While the R. prolixus nitrophorins increase the effectiveness of their NO-heme proteins by also binding histamine, secreted by the victim in response to the bite, to the heme, the Cimex nitrophorin does not bind histamine but rather binds two molecules of NO reversibly, one to the heme and the other to the cysteine thiolate which serves as the heme ligand in the absence of NO. This requires homolytic cleavage of the Fe S Cys bond, which produces an EPR-active Fe(II) NO complex having the {FeNO}7 electron configuration and a Cys SNO. This reaction is reversible at low pH. Investigations of the R. prolixus nitrophorins and their axial ligand complexes by NMR and vibrational spectroscopies have allowed detailed characterization of the complexes, and investigation of site-directed mutants by spectroelectrochemistry and NMR spectroscopy has allowed pinpointing the importance of certain charged, distal pocket and belt residues in stabilizing the Fe(III) form of the protein that allows NO to be released into the tissues of the victim. For the Rhodnius nitrophorins, the heme of the {FeNO}6 stable NO complex could have the limiting electron configurations Fe(III) NO• or Fe(II) NO+ . While vibrational spectroscopy suggests the latter and Mössbauer spectroscopy cannot differentiate between a purely diamagnetic Fe(II) center and a strongly antiferromagnetically coupled Fe(III) NO• center, the strong ruffling of the heme (with alternate meso-carbons shifted significantly above and below the mean plane of the porphyrin, and concomitant shifts of the -pyrrole carbons above and below the mean plane of the ring, to produce a very nonplanar porphyrin macrocycle) may suggest at least an important contribution of the latter. The strong ruffling would help to stabilize the (dxz , dyz )4 dxy 1 electron configuration of low-spin Fe(III) (but not low-spin Fe(II)), and the dxy orbital does not have correct symmetry for overlap with the half-filled  ∗ orbital of NO. This Fe(III) NO• electron configuration would facilitate reversible dissociation of NO. Keywords: Nitrophorins, nitric oxide, vasodilation, {FeNO}6 , antiferromagnetic coupling, ruffled heme, dxy orbital, a2u () porphyrin orbital.

Interaction of NO with Insect Nitrophorins

379

1. BACKGROUND Nitric oxide has been shown to serve as an autacoid and neurotransmitter in vertebrates, as well as a toxic defense substance for eliminating invading organisms [1–7]. The 1998 Nobel Prize in Medicine was awarded to Ignarro, Furchgott and Murad, whose papers are referenced herein, for their work in showing how NO, a gaseous molecule produced by many cells, interacts with various tissues. NO is produced by three forms of the enzyme NO synthase (NOS), which was originally thought to include a fusion of the heme enzyme cytochrome P450 with its own reductase [1–18]. This complex enzyme contains FAD, FMN and NADPH binding sites as well as a heme and non-heme iron, and a cofactor not present in cytochromes P450, tetrahydrobiopterin. In addition, the enzyme either includes a calmodulin-like subunit that is activated by binding Ca2+ or, as in neuronal NOS, is activated by association with Ca2+ -bound calmodulin [1–19]. The crystal structure of the domain that contains the heme and tetrahydrobiopterin centers has been reported, and although the heme is bound axially to a cysteine, as in all cytochromes P450, the protein folding pattern is quite different; P450s are largely -helical [20], while the NOS heme/tetrahydrobiopterin domain has a number of -sheets as well as -helices in a structure known as a curved  −  domain [21]. Thus, the heme/biopterin domain of NOS and the heme domains of cytochromes P450 probably achieved similar catalytic activities through convergent evolution [21]. To aid in defense against microbes, immune cells such as neutrophils and activated macrophages produce large amounts of NO [22,23], which helps to kill phagocytosed bacteria and parasites [24]. Detailed studies of inhibition of these enzymes by l-arginine analogs [25–27], and of subunit organization and conditions for dimer dissociation [28,29] have been reported. In some cases, activated macrophages induce production of NO synthase, which then produces NO. In some instances, production of large amounts of NO can result in metabolic dysfunction of target cells, leading to several autoimmune diseases, including insulin-dependent diabetes mellitus [30–52], allergic encephalomyelitis [53], tissue injury and various inflammatory disorders [54,55], neuronal disease [56,57], cancer or tumor cell proliferation [58,59], angiogenesis and related pathological changes [60,61] and possibly multiple sclerosis [53]. eNOS can produce oxidant stress due to uncoupling, which generates reactive oxygen species [62]. Inducible NOS also may be stimulated by pro-inflammatory cytokines [63,64], possibly leading to myocardial depression following cardiopulmonary bypass. The excess NO is believed to complex with the iron-sulfur centers in proteins such as aconitase, which leads to inactivation of the tricarboxylic acid pathway cycle [65] of invading bacteria and parasites. The induction of NO synthase by interferon- also appears to inhibit viral replication [66]. As a physiological regulatory molecule, NO is produced by a variety of tissues, such as endothelium and neural tissue; hence, NO is believed to be synonymous with the endothelial-derived relaxing factor (EDRF) that was described by Furchgott in the early 1980s [67]. It acts by activating the soluble form of the heme protein guanylyl cyclase [68]. It has been proposed that the binding of NO to the heme moiety of guanylyl cyclase triggers dissociation of the proximal histidine, causing a conformational change in the enzyme that leads to its activation [69–73]. Increased cyclic 2 ,3 -guanosine monophosphate (cGMP) then activates phosphorylation events that change the state of the target tissue. In smooth muscle, relaxation occurs, with effects including lowered blood pressure [4–7] and mediation of penile erection [74,75]. In platelets, inhibition of

380

F. Ann Walker

aggregation occurs [68]. A role for NO in mediating neurotransmitter release in cerebral cortex has also been found [76], and it has been suggested that NO, in conjunction with light, may be involved in resetting the human circadian clock [77–79]. It is interesting to note that both the synthesis and many of the physiological actions of NO involve heme proteins. Indeed, the accepted NO synthase model proposes that the arginine binding site resides in the distal heme pocket [6], and the action of NO in guanylyl cyclase also requires NO binding to heme [6]. Additionally, because NO is a very reactive substance, it is believed that in biological systems NO reacts with superoxide anions, including those produced by autoxidation of hemoglobin, to produce nitrite and nitrate, or reacts with hemoglobin to form complexes that are very stable to dissociation, and can decay to methemoglobin and nitrite. Cellular NO synthesis has been found to cause the loss and degradation of certain enzyme-bound hemes, and to impede the action of the ferrochelatase enzyme involved in heme synthesis [80]. The interaction of nitrite with myoglobin during the curing of meat has been shown to produce a red pigment that has identical optical spectral characteristics to those of the mononitrosyl derivative of heme [81], and treatment of metmyoglobin with high concentrations of nitrite at pH < 7, as in improper curing of meat, produces a green pigment known as nitrimyoglobin [82], in which the 2-vinyl group of the heme has been nitrated in the -trans position. Thus, biochemically speaking, the synthesis, signaling and destruction of NO all involve heme proteins.

2. STRATEGIES USED BY BLOOD-SUCKING INSECTS TO INSURE THAT THEY OBTAIN A SUFFICIENT MEAL There are approximately 15,000 species of blood-sucking insects in the world, each of which has various components in its saliva that aid the insect in obtaining a sufficient blood meal. These substances help to cause vasodilation and to counteract blood coagulation, platelet aggregation and/or swelling and the beginning of the immune response in the victim [83,84]. They include a wide range of anticoagulants, anti-platelet aggregation compounds, substances that cause vasodilation and/or create antihistaminic agents, and multiple substances are typically found in the saliva of a given insect, hence providing redundancy in aiding it to feed successfully. The presence of these substances in the saliva allows blood-sucking arthropods to minimize the time taken for obtaining a blood meal [85,86]. Because the evolution of blood feeding crosses a number of classes of arthropods (is polyphyletic), it is believed that these organisms developed a large variety of anti-hemostatic compounds via convergent evolution [87]. Within the class Insecta and order Hemiptera, the blood-sucking bug R. prolixus, a member of the Kissing Bug family that is native to the Amazon river basin but is found in all temperate areas of South, Central and even southern North America, has a group of salivary nitrovasodilators [88]. These nitrovasodilators are composed of unique heme proteins that act as storage and delivery systems for NO [89,90]. This new class of heme proteins has accordingly been named nitrophorins (nitro = NO, phorin = carrier), and abbreviated NP [91]; there are a total of four nitrophorins in the adult R. prolixus insect, denoted NP1-NP4, or in recent papers rNP1-rNP4, to differentiate them from the Cimex nitrophorin, cNP (see below). The nitrophorins provide a means of stabilizing NO for long periods of time in the insect salivary glands, and then releasing it upon dilution

Interaction of NO with Insect Nitrophorins

381

and increase in pH, when the saliva is injected into a victim. NO can dissociate from the carrier proteins by dilution at neutral or alkaline pH, but less readily at pH 5.0, the approximate pH of the insect’s saliva. Indeed, the titration curve indicates an ionizable group with a pKa of 6.5 [89,92]. In the notation of Feltham and Enemark [93], these are {FeNO}6 centers, which are expected to be EPR-silent. In fact, the appearance of the high-spin Fe(III) EPR (electron paramaagnetic resonance) signal when argon was blown over a sample of homogenized R. prolixus salivary glands, and its disappearance when NO was blown over the same sample (Fig. 1) [89], provided the initial clue as to how NO could be released upon dilution of the protein, for Fe(II) NO centers, having the {FeNO}7 electron configurations, have dissociation constants in the pico- to femto-molar range [69] and would thus not release NO upon dilution by a factor of ∼100 or so. We have also shown that these proteins provide an additional means of assuring the insect a sufficient blood meal by binding histamine [94,95], which is produced by the

Homogenate

A

Homogenate + Ar

B Homogenate + Ar + NO

C g=2

(B – C)

g=6

D

Homogenate + DT

E

0

1000

2000 3000 4000 Magnetic field (G)

5000

Fig. 1. EPR spectra of 100 pairs of R salivary glands obtained in 125 L of PBS (A) before argon equilibration (B) after equilibration in an argon atmosphere for 4 h and (C) after equilibration of (B) with NO for 2 min. (D) is the difference spectrum, that is, (B–C). (E) is the homogenate as in (A) treated with dithionite (DT) to reduce Fe(III) to Fe(II). All spectra are on the same scale, except (E), which is shown reduced 3x in amplitude. The spectra were recorded on a Bruker ESP-300E X-band spectrometer equipped with a helium cryostat (Oxford Instruments). Conditions were as follows: Power attenuation, 30 dB; modulation frequency, 100 kHz; modulation amplitude, 3.2 G; receiver gain, 1 25 × 105 ; resolution 1024 points; time constant, 82 ms; and sweep width, 5000 G. Reprinted from [95] with permission from the American Association for the Advancement of Science.

382

F. Ann Walker KdIII(NP–NO) >> KdII(NP–NO) 0.02–1 μM 60–90 μM at pH 7.5 Facilitates NO release

•NO NP1-4- •NO

NP1-4- Histamine

Histamine KdIII(NP–Hm) < KdIII(NP–NO) 0.01–0.02 μM 0.02–1 μM at pH 7.5 Facilitates histamine binding and NO release

Fig. 2. Role of the R. prolixus salivary nitrosyl heme proteins NP1-NP4 in feeding by the insect. The NP NO complex is stored in the insect saliva and is delivered to the victim’s tissue while the insect probes for a blood vessel. Dilution, binding of histamine and an increase in pH facilitate the release of NO. Histamine is released by the victim in response to tissue damage and the detection of foreign antigens. Histamine binding to the nitrophorins aids in the transmission of Trypanosoma cruzi, the causative agent of Chagas’ disease, to the victim, by preventing detection of the insect for a period of time (see text). Modified from [95] with permission from Nature.

victim in response to the wound, and thereby at least slowing the defense mechanisms of the victim, as diagrammed in Fig. 2. The histamine complex dissociation constant of rNP1 is 19 nM at pH 7.0 [95] and 11 nM at pH 8.0 [92], while the NO complex dissociation constant is 950 nM, about 50 times larger, at pH 7.0 [96]; the Kd at pH 8.0 is 850 nM [92], a factor of 80 times larger. For rNP4 and rNP2 the histamine Kd is 10 and 9 nM, respectively, at pH 8.0 and 25 C, while the NO Kd is 540 and 20 nm, respectively (NP3 has Kd s for both ligands that are too small to measure at 25 C) [92]. Hence, histamine can displace NO from the Fe(III) form of the nitrophorins, and the tight binding of histamine provides yet another means of insuring a successful meal for the insect. The structural basis for this tight binding is discussed below. rNP2 is also able to interfere with the activation of Factor Xa in the blood coagulation cascade [97,98], and thus possesses a third antihemostatic function. The binding of histamine to the nitrophorins, mentioned above, which leads to the insect’s not being detected for long periods of time during feeding, is clearly a contributing factor in the potential infection of the victim with the protozoan T. cruzi, which is carried by R. prolixus, because the trypanosome is left at the site of the bite in the feces of the insect after feeding [99]. Later, secretion by the victim of more histamine than can be bound by the nitrophorins injected by the insect causes itching and the beginning of the immune response, which may lead the victim to scratch the bite, thus possibly introducing the trypanosome into the blood stream. T. cruzi causes Chagas’ disease, a debilitating disease that leads to weakening of the heart and gut muscles on a very slow timescale, and in some cases to death, after periods of tens of years [100–103]. Both the insect and the trypanosome are found in tropical areas of the New World, and now as far north as southern Texas and Arizona [102,104]. This, combined with migration of large numbers of Latin Americans, some of whom may be infected with Chagas’ disease, now puts the U.S. blood supply at risk, for while the blood supplies in Latin America are

Interaction of NO with Insect Nitrophorins

383

screened with antibodies to T. cruzi, such tests are currently only beginning to be used in this country [102]. There is only one drug approved for use against T. cruzi in the U.S. (Nifurtimox, a nitrofuran), while a second drug (Benznidazole, a nitroimidazole) is also available in Latin American countries [100,101]. Neither is very effective. Much [105–119], though certainly not all [120–131], of the research currently being carried out on T. cruzi and Chagas’ disease is being done in Latin America, where so many people are infected with the disease. It must also be pointed out that R. prolixus is not the only carrier of T.cruzi in the New World. In fact, there are a number of triatomine bugs that carry the trypanasome, including Triatoma infestans as the major disease vector in Argentina, Bolivia, parts of Brazil, Chile, Paraguay, Peru and Uruguay [132,133]. However, only the genus Rhodnius has red saliva and thus has the nitrophorins.

3. PROTEIN SEQUENCES AND STRUCTURES OF THE NITROPHORINS FROM R. prolixus. 3.1. Protein Sequences Following the discovery of the salivary nitrophorins in the early 1990s [88,89], further work in Dr Ribeiro’s laboratory led to the purification of four salivary heme proteins from R. prolixus, the predominant one, NP1, being 50% of the total heme protein content and 25% of the total gland protein [91]. The amino terminal sequence for all four heme proteins and partial sequences from proteolytic digests of the two most abundant proteins were obtained [91]. A cDNA library was also produced from R. prolixus salivary glands and the gene for the most abundant nitrophorin, NP1, was cloned [91] and expressed [96]. While the work in our laboratory first concentrated on NP1, the genes for NP2, NP3 and NP4 have also been cloned, sequenced and found to be similar in protein sequence to NP1 [67]. In fact, the four nitrophorins fall into two pairs of nearly homologous protein sequences: NP1 and NP4 as one pair and NP2 and NP3 as the other, as shown in Fig. 3. NP1 and NP4 are 91% identical in sequence, while NP2 and NP3 have 78% sequence identity. Overall, the four proteins have only 34% sequence identity. In line with the two pairs of similar protein sequences, the rates of NO release also fall into two groups, with NP1 and NP4 having much larger Kd s than NP2 and NP3 [92]. More recently, additional nitrophorin genes have been found to be expressed in earlier life stages of R. prolixus, NP5 [134], NP6 [134] and NP7 [135]. Because the sequence of NP5 is very similar to that of NP4 and the sequence of NP6 is very similar to that of NP2 [136], we have not investigated these two proteins. However, NP7 is very different in sequence from NP1 to NP4. It shows most similarity to NP2 and NP3 (62 and 60% identity, respectively); many residues conserved in NP1-NP4 are very different in NP7, as shown in Fig. 3. This is in some part due to the different role of NP7, which binds to anionic phospholipid membranes [135]. Our characterization of NP7 thus far is discussed in the next section. Overall, the four proteins have only 34% sequence identity, while when NP7 is added to this mix the sequence identity drops to 27%. Database searches showed little similarity between the sequences of NP1-NP4 and other proteins [91]. The presence of the non-covalently bound protoheme, the size of the protein and the interaction of histidine with the heme (indicated by the pH

384

F. Ann Walker

1 NP1 KCTKNALAQT NP4 ACTKNAIAQT NP2 DCSTNISPKQ NP3 DCSTNISPKK NP7 LPGECSVNVIPKK NP1 NP4 NP2 NP3 NP7 NP1 NP4 NP2 NP3 NP7

71_____ VSELQEESPG VSELQVESLG IGEGKLESSG IGEGKLGSSG TGTGPLESNG

__α1__ 21 _β A__ |--A-B loop--|41__βB ___ GFNKDKYFNG DVWYVTDYLD LEPDDVPKRY CAALAAGTAS GFNKDKYFNG DVWYVTDYLD LEPDDVPKRY CAALAAGTAS GLDKAKYFSG .KWYVTHFLD KDP.QVTDQY CSSFTPRESD GLDKAKYFSG .TWYVTHYLD KDP.QVTDPY CSSFTPKESG NLDKAKFFSG .TWYETHYLD MDP.QATEKF CFSFAPRESG ___βE_____91 .KYTANFKKV EKNGNVKVDV .KYTANFKKV DKNGNVKVAV LQYTAKYKTV DKKKAVLKEA VQYTAKYNTV DKKRKEIEPA AKYTAKFNTV DKKGKEIKPA

161 141 _____ α2 __ NKDTNAGDKV KGAVTAASLK FSDFISTKDN NKDAAAGDKV KSAVSAATLE FSKFISTKEN QKDAEPSAKV KSAVTQAGLQ LSQFVGTKDL QKTGEPSATV KNAVAQAGLK LNDFVDTKTL NKNALPNKKI KKALNKVSLV LTKFVVTKDL

__βC____61 ____βD GKLKEALYHY DPKTQDTFYD GKLKEALYHY DPKTQDTFYD GTVKEALYHY NANKKTSFYN GTVKEALYHF NSKKKTSFYN GTVKEALYHF NVDSKVSFYN

___βF_111_ ____βG___ 131 __βH__ TSGNYYTFTV MYADDSSALI HTCLHKGNKD LGDLYAVLNR TAGNYYTFTV MYADDSSALI HTCLHKGNKD LGDLYAVLNR DEKNSYTLTV LEADDSSALV HICLREGSKD LGDLYTVLTH DPKDSYTLTV LEADDSSALV HICLREGPKD LGDLYTVLSH DEKYSYTVTV IEAAKQSALI HICLQEDGKD IGDLYSVLNR ___α3___ KCEYDNVSLK SLLTK NCAYDNDSLK SLLTK GCQYD.DQFT SL~~~ SCTYD.DQFT SM~~~ DCKYD.DKFL SSWQK

IDENTICAL RESIDUES CONSERVATIVE REPLACEMENTS CHARGE MUTANTS BELT AND DISTAL POCKET MUTANTS RESIDUES UNIQUE TO NP7

Fig. 3. Sequences of five of the nitrophorins from R. prolixus, arranged to show sequence identities within the two groups; NP7 has greatest homology to NP2 and NP3. Helices and -sheet strands are labeled  and , respectively. Definition of secondary structure is based on the structure of NP4. Residues mutated in each of the nitrophorins by the author’s research group are shown in red and cyan. (see Plate 9.)

dependence of NO binding and the EPR spectral behavior of salivary homogenates [89]) first suggested a possible relationship to hemoglobins or cytochromes b. Alignment of the sequences of NP1NP4 with hemoglobin sequences from the insect Chironomus [137], the annelids Lumbricus and Tylorrhynchus [138,139], the mollusc Glycera [140], the parasitic nematode Ascaris [141], human beta chain hemoglobin, leghemoglobin and vertebrate cytochromes [142] indicated an overall sequence similarity of only 38 to 45%, with no pronounced regions of sequence identity. No other higher overall sequence similarities could be found with other proteins, and thus we initially assumed that NP1NP4 might be found to have globin folds.

3.2. Protein Structure Determination of the three-dimensional structures by Montfort and coworkers of various ligand complexes of recombinant NP1 [95], NP4 [143–145] and NP2 [146,147] by X-ray crystallography show that the protein fold of the nitrophorins belongs to a diverse class of proteins called “lipocalins” [148,149], a family of relatively small secreted proteins that typically bind small, principally hydrophobic molecules such as pheromones [150– 152], retinol [153,154], prostaglandins [153,155], retinoic acid [156] and biliverdin [153–161], although a recently discovered lipocalin binds two molecules of histamine [162,163], another binds ADP and other adenine nucleotides [164], another binds a Fe(III)-bound tris-catecholate siderophore [165], one is an enzyme [166], one, C8 , is a subunit of the eighth component of complement that interact to form the cytolytic membrane attack complex, MAC [167], and two, lactoglobulin and 1 -microglobulin,

Interaction of NO with Insect Nitrophorins

385

play important, but as yet incompletely understood roles in the processing of fatty acids in milk [168,169] and free hemoglobin and heme in the circulatory system [170–172], respectively. Preliminary evidence suggests that 1 -microglobulin degrades hemin in the bloodstream and in tissues [172]. The lipocalins have little sequence homology and are almost entirely -sheet proteins that are folded into eight-stranded -barrels that are closed at one end and bind the small molecule within the other end of the barrel [148,173]. R. prolixus also utilizes lipocalins that do not bind heme for several other antihemostatic purposes [164,176]. The ribbon structure of the cyanide complex and the histamine-bound heme pocket of NP1 [95] are shown in Fig. 4. (A)

(B)

(C) Leu 130

Leu 123

Glu 32

Asp 30 Leu 133

His 59 Asp 70

Fig. 4. (A,B) Ribbon and ball-and-stick diagram of the NP1 CN− structure. CN− and iron are shown in a space-filling representation. Disulfide bonds linking Cys-2 to Cys-122 and Cys-41 to Cys-171 are shown. The view in (B) is rotated approximately 90 about the vertical axis from the view (A) on the left. The mobile loops A-B and G-H discussed in the text are above and next to the heme in this view. (C) Histamine binding to NP1. Hydrogen bonds are shown as dashed lines, nitrogen-Fe bonds as solid lines. Shown are hydrogen bonds between histamine and Asp-30 (2.7 Å), Glu-32 (3.1 Å), Leu-130 (2.7 Å), and an ordered water molecule (2.8 Å), which further hydrogen bonds to Thr-121 (3.2 Å), Leu-123 (3.0 Å) and Gly-131 (2.7 Å). Van der Waals contacts are made to Leu-123 (3.7 Å), Leu-130 (4.2 Å), and Leu-133 (3.7 Å). Also shown are hydrogen bonds between an ordered water molecule and residues His-59 (2.7 Å) and Asp-70 (2.6 Å), and between Asp-70 and a heme propionate (2.5 Å). The other propionate has been omitted for clarity. Reprinted from [95] with permission from Nature.

386

F. Ann Walker

Our first structure of an NO complex, NP1 NO, had a highly bent Fe N O unit, with angles of 123 and 135 for molecules I and II in the unit cell [177], values consistent with a Fe(II) NO (Enemark/Feltham notation {FeNO}7 [93]) center, which is not the stable oxidation state of the protein in solution, as shown by the EPR spectra of Fig. 1. At the time of the determination of that structure, we suspected that the Fe NO center might have been photoreduced by the X-rays used for structure determination. Thus, when the first structure of NP4 NO was determined, the X-ray data were collected at low temperatures [144]. For one of two resolved conformations of the NP4 NO complex, the Fe N O unit appeared to be linear (174–178 ), as shown in Fig. 5; however, in the same structure there were molecules that had a strongly bent Fe N O unit, with an angle of 138–144 [144]. With further study, Dr Montfort and his coworkers confirmed that the nitrophorin-NO complexes are readily photoreduced by the X-rays used for single crystal X-ray diffraction [178]. In order to prevent photoreduction it was found necessary to collect the X-ray diffraction data rapidly, preferably using synchrotron radiation, and at very low temperatures, as well as using short wavelength X-rays [178], and as the techniques were improved, the Fe N O unit has become more and more linear [145]. As shown in a close-up view (Fig. 6) the NP4 NO unit is still somewhat bent, with a Fe N O angle of 156 [145], and it is now believed that the previous resolution of two very different angles [144] was due to partial reduction of the FeNO complex to {FeNO}7 [145]. However, whether the 156 angle may also contain some contribution from photoreduction is not known. Structures of small molecule ferro- and ferriheme(L)-NO and their five-coordinate analogue complexes have been reported [179–181], and calculations aimed at explaining the bent and off-axis NO binding have also been reported [182]. However, the lower precision of the bond angles and atom positions in a protein structure make it difficult to discuss meaningfully the apparent similarities or differences between the model heme

Fig. 5. Ribbon drawing of the NP4 NO structure. Except for the loops that move on NO binding (loop A-B, residues 31–37, and loop G-H, residues 125–132); the linear NO orientation is in ball-and-stick representation. Reprinted from [144] with permission from Nature. The color version of this figure, and the caption describing the color key, can be found at the end of this volume. (see Plate 10.)

Interaction of NO with Insect Nitrophorins

387 Leu 133

Leu 123

ϕ

3.35 Å

3.47 Å

φ

His 59

Fig. 6. Near-linearity of the Fe N O unit of NP4 NO, and interaction of the two distal leucines with the highly ruffled heme of this complex. The methyl groups of both leucines are within van der Waals contact of the heme  system. Reprinted from [145] with permission from the American Chemical Society.

metrics and those of the nitrophorins themselves. Undoubtedly, if it were possible to obtain crystal structures of heme proteins to the same resolution as those of model heme complexes, it would turn out that the structural features observed in the model hemes would also be observed for the proteins – if photoreduction of the Fe NO centers of the protein crystals could be completely prevented. Other antihemostatic proteins that have been found in the saliva of R. prolixus include an inositol polyphosphate 5-phosphatase (IPP) that appears to facilitate the ingestion of blood [183]. The amino acid sequence of this protein is very similar to that of the IPP5C domain from SP synaptojanin [184] and the Cimex nitrophorin [185] discussed below.

4. THE SPECIAL PROPERTIES OF NP7 After Drs Ribeiro and Andersen had cloned the gene for NP7, expressed it and investigated it preliminarily [135], they contacted us and offered the gene to us for more detailed characterization of the protein’s molecular properties. From their initial findings, NP7 appeared to have high affinity for anionic phospholipid membranes, and was apparently targeted to the negatively charged surfaces of activated platelets and other cells, where it can serve as a vasodilator, antihistaminic agent, platelet aggregation inhibitor and anticoagulant [135]. They found, using sedimentation and surface plasmon resonance experiments, that there were two classes of phospholipid binding sites that have Kd values of 4.8 and 755 nM, and that NP7 inhibits prothrombin activation by blocking phospholipid binding sites for the prothrombinase complex on the surface of vesicles

388

F. Ann Walker

and activated platelets [135]. As an NO complex, NP7 inhibits collagen and ADPinduced platelet aggregation and induces disaggregation of ADP-stimulated platelets by an NO-mediated mechanism. Molecular modeling of NP7 revealed a putative, positively charged membrane interaction surface comprised mainly of a helix lying outside of the lipocalin -barrel structure [186]. (This is the same helix identified by multinuclear NMR spectroscopy, mentioned below, as the element that provides structural rigidity by hydrophobic interactions with one side of the -barrel.) The cDNA sequence coding for NP7 was kindly provided by Dr J.F.Andersen (NIH) in a pET-17b expression vector. NcoI (5 end) and an XhoI restriction site (3 end) were inserted using standard PCR techniques and the DNA was ligated into a pET-28a(+) plasmid. First attempts to produce NP7 similarly to the other four NPs resulted in only marginal amounts of protein, probably due to the much higher pI (8.97 calc.) as compared to the first four characterized NPs, which range from 6.11 (NP2) to 6.47 (NP3). Thus, new methods had to be developed to refold and reconstitute the protein from inclusion bodies. The most significant difference was that all procedures needed to be carried out at much lower pH (∼6.6) [186]. Inclusion of 0.5 M l-arginine and 10% sucrose in the initial refolding buffer helped to prevent aggregation, and addition of 1 mM oxidized glutathione during the refolding helped to form the necessary Cys Cys bridges. Furthermore, the usual KPi buffer causes precipitation of NP7 if the pH is above 6.0. In its place 3-(N-morpholino)propanesulfonic acid (MOPS) was found to be a useful buffer. Holoprotein formation was performed by hemin-chloride (∼1 mM in DMSO) titration [186]. Attempts to obtain crystals of NP7 for X-ray crystallography have so far been frustrated by aggregation and precipitation of the protein. Further purification of the protein using chelating Sepharose chromatography loaded with Ca2+ and eluted with a CaCl2 gradient was developed and may help the process of obtaining crystals in the future. NP7 has been characterized by spectroelectrochemical techniques in the absence and presence of NO, histamine and imidazole at pH 5.5 and 7.5. Overall, the reduction potentials of NP7 without ligand and in complex with NO or histamine are quite similar to those of the other NPs. However, at pH 7.5 NP7-histamine has a significantly higher reduction potential, −319 mV, 84 mV more positive than NP1-Hm and 133 mV more positive than NP2-Hm. The E27V mutant of NP7 (making it more NP1- or NP2-like) was prepared, and its histamine complex was found to have E = −396 mV, very similar to that of WT NP1 [186]. The NO and histamine binding constants have also been measured as a function of pH. Surprisingly, NP7 exhibits a much lower Hm affinity at pH 7.5 (log Keq III = 5 02 as compared to 8.00 for NP2), although the ImH binding constants are somewhat larger (log Keq III = 6 04 for NP7, 7.43 for NP2). The E27V mutant of NP7 shows no pH dependence for log Keq III over the pH range 5.5–7.5 (8.00 vs. 8.05). Thus E27 is essential for the function of this protein. Another NP7 mutant prepared lacks the 3N-terminal residues LPG (Fig. 3). Because of our previous results on the significant contribution of the N-terminus to the properties of the NP2 heme [187], discussed below, we have begun characterization of NP7 1-3 in comparison to the WT NP7. In line with the marked influence of the D1A mutant of NP2 [187], we found that NP7 1-3-Hm complex has a reduction potential of −254 mV at pH 7.5, the highest determined among all the NPs. Furthermore, the binding constant at pH 7.5 increased to log Keq = 7 10. Thus the N-terminal sequence has a strong influence on both reduction potential and binding affinity of the heme center.

Interaction of NO with Insect Nitrophorins

389

Interaction of NP7 with liposomes has been studied carefully in our laboratory by Dr Knipp [188]. Besides phosphatidylcholine (PC) and phosphatidylserine (PS), which had been investigated previously [135], he investigated the interaction of liposomes composed of phosphatidylinositol (PI), phosphatidic acid (PA), phosphatidylglycerol (PG), lysophosphatidylcholine (lysoPC), sphingomyelin (SM) and cardiolipin (CL) with NP1-NP4 and NP7. It was found that while NP1-NP4 did not bind to any of the liposomes, NP7 bound not only to PS, but also other phospholipids having an overall negative charge (PA, PG, PI and CL). This extends the findings of Andersen, Ribeiro and coworkers [135] to a number of other important components of cell membranes, and provides additional understanding of the binding of this NP to phospholipids and thus recognition of cells with negatively charged surfaces, for example activated platelets. This would allow NP7 to transport more NO to such cells than can the other NPs and, consequently, prevent platelet aggregation. The log Keq for NO binding to NP7 dropped from 6.6 to 5.2 (or equivalently, the Kd increased from 4 M to 150 mM) upon binding to PS containing liposomes, which is actually the smallest log Keq or largest Kd value found among all NPs. The existence of a positively charged patch on the surface of NP7 explains Dr Knipp’s observation of aggregation of negatively charged liposomes in the presence of this protein. Creation of a model of the structure of NP7 based on the NP2 crystal structure (62% identity) using the Swiss-Model server and manual refinement (rmsd = 0.39 Å between backbone atoms of NP7 and NP2) indicates not just one, but also a second positively charged site on the protein, that one being near the mouth to the heme-binding cavity [188].

5. NMR SPECTROSCOPIC STUDIES OF THE NITROPHORINS 5.1. Heme Resonances of Paramagnetic Forms of the Nitrophorins The heme resonances of WT NP2 and its ImH and NMeIm complexes were reported and a theory was developed for predicting the chemical shifts of the heme resonances of high-spin heme proteins [189]. More recently the NMR spectra of the four main nitrophorins in the high-spin (S = 5/2) ferriheme states have been reported [190], and are shown in Fig. 7. The majority of the ferriheme resonances of high-spin NP1 and NP4 were assigned and compared to those of NP2 [190]. It was found that the structure of the ferriheme complexes of NP1 and NP4, in terms of the orientation of the histidine imidazole ligand can be described with good accuracy by NMR techniques, and that the angle plot proposed previously for the high-spin form of the nitrophorins [190] describes the angle of the effective nodal plane of the axial histidine imidazole in solution. There is an equilibrium between the two heme orientations (A and B) that arise from incorporation of the unsymmetrical protohemin IX molecule into a chiral protein environment (Scheme 1) that depends on the heme cavity shape, which can be altered by mutation of amino acids with side chains (phenyl vs. tyrosyl) near the potential position where a heme vinyl group would be in one of the isomers. It was found that the A:B ratio can be much more accurately measured by NMR spectroscopy than by X-ray crystallography. For NP3, for which there are no crystal structures available, the heme pocket structure is found by NMR techniques to be very similar to that of NP2 [191].

390

F. Ann Walker

NP 1

70

65

60

55

50

45

40

35

30

25

ppm

3M, 7Hα

5,8M

NP 2

1M

4VHα 2VHα

6H′α 7H′α 6Hα

70

65

60

55

50

45

40

35

30

25

ppm

NP 3

70

65

60

55

50

45

40

35

30

25

ppm

NP 4

70

65

60

55

50

45

40

35

30

25

ppm

Fig. 7. 1D 1 H NMR spectra of the high-frequency (16–73 ppm) region of the four nitrophorins in the absence of added ligand at 25  C, recorded at 500 MHz. The more intense resonances in the 55–68 ppm region for NP1 and NP4 and the 49–67 ppm region for NP2 and NP3 are those of the eight heme methyls of the two heme-rotational isomeric proteins (Scheme 1). Resonance assignments for NP2 [189] are given. Reproduced from [190] with permission from the American Chemical Society.

Interaction of NO with Insect Nitrophorins

391 x

y I120

II N

NP1,3,4: Both A & B

δ H3C

3

1

H3C

N

N

CH2

II

L132

CH2

N

CH2

CH2

CH2

COO–

COO–

COO–

NP2: Mainly B

δ

Fe

N 5 III H3C L122 CH2

CH3

CH3

I

N

β

5

III

2

1

β N

α

3

4

4

γ

L122

I120

L132

Fe

8 IV

CH3

CH3

α

2

y

x

N IV

8

γ

CH3

CH2 CH2 COO–

Ligand plane angle = 145°

Ligand plane angle = 125°

A

B y

x CH3 I120 3 H3C 4 II N

α

β

Fe

1

CH3

I N δ N

N

5 III H3C L122 CH2

CH3 2 L132

IV

8

γ

CH3

CH2

CH2

CH2

COO–

COO–

"Symmetrical Hemin": 2,4-dimethyldeuterohemin B numbering

Scheme 1.

The majority of the ferriheme resonances of low-spin nitrophorins (NP) 1 and 4 have also been assigned for the histamine, imidazole and cyanide complexes, and have been compared to those of NP2 [192]. It was found that the structure of the ferriheme complexes of NP1 and NP4, in terms of the orientation of the ligand(s) can be determined with good accuracy by NMR techniques in the low-spin forms, and that angle plots proposed previously [193] describe the angle of the effective nodal plane of the axial ligands in solution. The effective nodal plane of low-spin NP1, NP4 and NP2 complexes is in all cases of imidazole and histamine complexes quite similar to the average of the His-59 or -57 and the exogenous ligand angles seen in the X-ray crystal structures [192]. For the cyanide complexes of the nitrophorins, however, the effective nodal plane of the axial histidine imidazole does not coincide with the actual imidazole plane orientation. This appears to be a result of the contribution of an additional source of asymmetry, the orientation of one of the zero-ruffling lines of the heme. Probably this

392

F. Ann Walker

effect exists for the imidazole and histamine complexes as well, but because the effect of asymmetry that occurs from planar exogenous axial ligands is much larger than the effect of heme ruffling, the effect of the zero-ruffling line can only be detected for the cyanide complexes, where the only ligand plane is that of the proximal histidine. The three-dimensional structures of three of the NP CN complexes, including that of NP2 CN, confirm the high degree of ruffling of these complexes [192]. As for the high-spin forms of the nitrophorins, there is an equilibrium between the two heme orientations (A and B), Scheme 1, that depends on the heme cavity shape, and changes somewhat with exogenous axial ligand. And also as for the high-spin nitrophorins, the A:B ratio can be much more accurately measured by NMR spectroscopy than by X-ray crystallography. The effect of the L122, L132 and double L122,132V mutations on the NMR spectra of NP2 in the absence and presence of ligands has been reported (Fig. 8) [194]. NOESY (A)

(B)

5,3 M

5M 1M 3M

1M

LCOORD

ppm

0

8 M* 5 10

5 M* 1 M* 15 20 25 25

20

15

10

5

0

ppm

20

15

10

5

0

ppm

Fig. 8. 1D and NOESY spectra of the imidazole (A) and 2-methylimidazole (B) complexes of the L122V mutant of NP2. Spectra (A) were recorded at 30 C, pH 7 at 500 MHz; spectra (B) were recorded at 15 C, pH 7, also at 500 MHz. From these spectra, together with the HMQC spectra of the same complexes, the assignment of all heme resonances could be made [194]. In spectra (A) chemical exchange cross peaks are also seen at 23.4, 2.9 ppm (8-Me), 22.7, 13.6 ppm (5-Me), 19.1, 12.7 ppm (1-Me) for the major  minor exchange process and are marked with an asterisk (∗ ). For NP2-L122V-ImH, from the volume intensities of the diagonal and cross peaks and the mixing time utilized (m = 30 ms), the rate constants calculated are kf = 54 s−1 and kr = 0 5 s−1 at 30 C, with Keq = kf /kr ∼100. The order of heme methyl resonances for the second isomer (8>5>1>3) is reproduced if the orientation of the ImH nodal plane is ∼51 . In spectra (B) chemical exchange cross peaks are seen at 12.6, 2.4 ppm for the free ligand/bound ligand exchange process. Reproduced from [194] with permission from Proc. Natl. Acad. Sci. USA.

Interaction of NO with Insect Nitrophorins

393

spectra of the ImH complexes of most of the mutants exhibit chemical exchange cross peaks, while WT NP2-ImH shows no chemical exchange. Chemical exchange in the case of the distal L → V mutants is due to imidazole ligand orientational dynamics [194]. The two angular orientations of the imidazole ligand could be determined from the 1 H chemical shifts of the heme methyls, and the rate of interconversion of the two forms could be estimated from the NOESY/EXSY diagonal and cross peak intensities [194]. For NP2-L122V, Keq is about 100, and favors an orientation similar to that found for the wild-type NP2 ImH complex [194]. For the other two mutants, NP2-L132V and NP2-L122,132V, Keq is even larger [194], which makes the quantitative determination of rate and equilibrium constants from chemical exchange cross peak intensities impossible. The NP2-L→A single and double mutants prepared more recently show more rapid chemical exchange dynamics between the two ImH orientations, which are in rapid exchange on the NMR timescale. Isoleucine-120 is another large residue in the distal pocket of NP2, NP3 and NP7; in comparison, the corresponding residue for NP1 and NP4 is threonine (Fig. 3). The I120V mutant of NP2 and its imidazole complex were fully characterized by 1 H and 13 C NMR spectroscopy [195]. The effect of the I120V mutation on the A:B ratio (changes A:B from 1:8 to 1:1.1) can be explained by comparison of the structure of wild-type NP2-ImH with wild-type NP4-ImH, where it can be seen that the distance between carbons 3M and the -CH3 of Ile is only 3.28 Å (PDB file 1PEE), which is probably not large enough to allow a vinyl group to be stable at this position for the A isomer. The bulky I120 side chain makes the A orientation less favorable in NP2, and thus its removal makes the A isomer much more favorable [195]. The E of the imidazole complex of the I120V mutant of NP2 is similar to that of wild-type NP2-ImH.

5.2. Three-Dimensional Structure Determination of apo- and High-spin holo-NP2 In addition to these studies of the heme resonances of the paramagnetic forms of the nitrophorins, we have also undertaken an investigation of the NMR spectrum and dynamics of apo- and high-spin holo-NP2. A near-complete (>97%) assignment of backbone spin systems (1 HN , 15 N, 13 CO, 13 CA , 13 CB , 1 HA , 1 HB ) of apo-NP2 has been achieved by following independent connectivity pathways using the standard suite of 3D triple-resonance experiments: HNCACB, CBCA(CO)NH, HNCA, HN(CO)CA, HNCO, HBHA(CBCACO)NH, HA(CA)NH, and HA(CACO)NH [196]. An additional group of 2 H-decoupled datasets, which included constant-time HNCA, HN(CO)CA, HN(CA)CB and HN(COCA)CB, was also collected on a 2 H,13 C,15 N triple-labeled sample. These experiments were required to resolve a number of sequential connectivity ambiguities along the 13 C dimension of apo-NP2 data sets acquired on the non-perdeuterated sample. These ambiguities resulted from the high degree of crowdedness within several regions of the apo-NP2 spectra due to resonance overlap. This overlap was especially severe in the 13 C dimension for those residues that belong to the regions of relatively high mobility, resulting in random-coil type averaging of chemical shifts, and thus their poor dispersion [197]. An initial model structure of the apoprotein was generated using an approach previously reported [198]. Briefly, the cross peak volumes of those 1 HN -1 HN , 1 HN -1 HA

394

F. Ann Walker

and 1 HN -1 HB NOE interactions, which can be unambiguously identified in the 15 N-edited 3D NOESY-HSQC (120 ms mixing time) using sequential assignments and 3D HA(CA)NH, HA(CACO)NH, HAHB(CACBCO)NH and HNHB correlation experiments, were extracted and converted to 444 sequential, 248 medium-range and 308 long-range distance restraints. These restraints were calibrated against the known inter-H distances in the -helical region of the protein (i.e. dNNii+1 , 2.8 Å and dNNii+2 , 4.2 Å). The restraint lower bound was set to the minimal allowed van der Waals distance (1.8 Å), and the upper bounds were set to the estimated distance +15%. For those regions of the protein sequence where both the CSI predictions and the measured 3 JHNHA coupling constants were in agreement, experimental restraints were supplemented with conservative generalized distance and torsion restraints corresponding to allowed regions of the Ramachandran map. Additionally, all measured 3 JHNHA were included as direct coupling-constant restraints. Although we also measured the H/D exchange rates with D2 O for all of the backbone N H using SOFAST-HMQC [199], these measurements were not included in the initial model calculations since the H-bonding partners cannot be unambiguously identified for slowly exchanging N H at this point. A set of 100 structures was calculated in CNS [200]. Since the calculated structures are greatly under-restrained, only one lowest energy structure was selected (violations: NOE, 1; dihedral angle restraints, 8; J-coupling, 33). This structure was further gently refined by calculating 50 structures using low-temperature TAD (300 K) for 15 ps, followed by 1000 steps of the conjugate gradient energy minimization. Ten lowest-energy structures (all of which showed only a single NOE and no J-coupling violations) were selected and averaged (Fig. 9A and B). Considering that the known crystal structure of holo-NP2 was not used as input during the NMR analysis, it is gratifying that the overall structure of this preliminary model (Fig. 9B) is very similar to the fold of the holo-NP2 protein seen in the solid state (Fig. 9C; for greater clarity, the structures of both the apo and holo forms are shown with the -barrel rotated 90 about its axis). Detailed analysis of the 1 H-1 H NOEs and structure calculations leading to a high-resolution structure of apo-NP2 are in progress. Work has begun on the 3D structure determination of two NP2 holoprotein complexes, the high-spin (S = 5/2) NO-off form of NP2, and its NO complex. The high-spin form of NP2 shows a well-resolved 15 N/1 H HSQC spectrum, where all observed resonances have been assigned; eight resonances are missing because of their proximity to the paramagnetic iron center, but we should be able to find them using special experiments, including those created by Bertini [201,202] and by Rivera [203]. And to our delight, the NO complex of NP2 is stable for extended periods of time (months) in sealed NMR tubes, and behaves as a diamagnetic species in the presence of 0.5 equivalents excess NO. It also gives a better-resolved 15 N/1 H HSQC spectrum than does the apoprotein, and assignment of the resonances is in progress.

5.3. First Studies of the Dynamics of apo-NP2 We also investigated the H/D exchange rates of the amide-H of the apoprotein. As shown in Fig. 10, the residues that show the slowest H/D exchange rates are localized on the 2 helix and F, G, H and A strands of the -barrel. The slow exchange rates are indicative of the presence of a compact hydrophobic core involving the FGHA part of

Interaction of NO with Insect Nitrophorins

395

(A)

(B)

(C)

Fig. 9. (A) An ensemble of the ten lowest energy structures obtained by gentle refinement of the model calculated from the initial unambiguously assigned NOE connectivities, and (B) an average structure calculated from this ensemble [196]. (C) X-ray structure of holo-NP2, PDB access code 2ACP.

396

F. Ann Walker

Tex (min) 90°

>1000 500–1000 100–500 50–100 10–50 5–10 <5

Fig. 10. The backbone amide H/D exchange rates in H2 O/D2 O (25/75). The rates were obtained at 30 C by dissolving a lyophilized sample of 15 N-labeled apo-NP2 in 100% H2O-based buffer (50 mM phosphate, pH 6.5) followed by a rapid 1:3 dilution into 100% D2 O-based buffer (final protein concentration ∼1 mM, 75% D2 O). Progress of the exchange process was followed by collecting a series of 1 H-15 N SOFAST-HMQC spectra at 600 MHz on a spectrometer equipped with a cryoprobe. The dead time was 2 min, and spectra were taken at 5, 10 and 20 min intervals for the first 2, 4 and 6 h, respectively, and every 40 min thereafter for a total of 30 h. Note that the calculated rates are shown here mapped onto an X-ray structure of holo-NP2. (see Plate 11.)

the -barrel with the opposing side of the barrel exhibiting a higher degree of plasticity. Higher plasticity may account for the fact that heme ruffling (and therefore the iron reduction potential) is insensitive to the bulkiness of the hydrophobic side chain residues located in the part of the protein (shown in yellow) generally away from the 2 helix. However, the steric requirements of the residues located in the rigid part of the -barrel (shown in red) do disrupt the planarity of the heme, thus significantly modulating the iron reduction potential as has been shown by site-directed mutagenesis studies discussed below. Relative rigidity of the -barrel fold makes it uniquely suitable for heme proteins such as NO carriers, where a very strong stabilization of the oxidized state of the metal center is required.

6. SOURCE OF NO IN THE INSECT SALIVA: A SALIVARY GLAND NO SYNTHASE FROM R. prolixus Ribeiro and coworkers have shown that the salivary gland homogenates of R. prolixus contain NO synthase activity, which is activated by tetrahydrobiopterin, Ca2+ , calmodulin, FAD and requires NADPH (but not NADH) to convert arginine to citrulline and NO [204]. Furthermore, similarly to vertebrate enzymes, NO synthase activity co-elutes

Interaction of NO with Insect Nitrophorins

397

with a diaphorase activity (a two-electron transfer from NADPH to tetrazolium dyes) in a molecular sieving column. Taken together, the data show that the activity is thus similar to the vertebrate soluble constitutive NOS enzyme [204]. The salivary cDNA from R. prolixus salivary NOS was cloned and expressed, and shown to have homology to the vertebrate constitutive enzymes, including binding sites for flavins, NADPH and calmodulin [205]. Ultrastructural localization of the associated diaphorase activity of the NO synthase indicates that it is present within cellular vacuoles, similarly to the vertebrate neural NOS enzyme [206], with which it has highest sequence homology [207].

7. A NITROPHORIN FROM ANOTHER INSECT: STRUCTURE, SPECTROSCOPIC AND REDOX PROPERTIES OF cNP We have shown that the bedbug, Cimex lectularius, also has a salivary nitrophorin that shows similar, yet unique, pH-dependent reversible NO binding behavior, and optical and EPR spectral properties [208]. The size (∼30 kD), amino acid sequence [208] and even the heme ligand (Cys) of the Cimex nitrophorin, cNP, are all quite different from those of the Rhodnius nitrophorins. The Cimex nitrophorin gene was expressed, and a purification method for the protein was developed [184]. The three-dimensional structure of this protein was solved by the Montfort group [184], and the heme is indeed bound to a cysteinate ligand, one of two Cys in the protein. The structure, shown in Fig. 11, is

(A)

(B)

(C)

Val 42 Thr 87 Asn 78

Ile 80

Val 44 Phe 49

Gln 56

Fig. 11. (A,B) Two views of the structure of cNP, showing the thiolate coordination of the heme and the -sandwich structure of the protein. (C) Closeup of the surroundings of the heme and cysteine thiolate as viewed from the opposite side of the protein as shown in (A), showing the hydrophobic nature of the NO binding site. Reprinted from [185] with permission from Proc. Natl. Acad. Sci. USA. The color version of this figure, and the caption describing the color key, can be found at the end of this volume. (see Plate 12.)

398

F. Ann Walker

that of a -sandwich with the heme on the outside of the sandwich, and covered by only one helix, which contains the Cys ligand. A close-up of the heme and its cysteine ligand showing the hydrophobic nature of the protein side chains surrounding the distal pocket is shown in Fig. 11C. The second Cys is inside the -sandwich and is isolated from solvent and other potential reactive protein side chains. The heme is thus fairly exposed on the surface of the protein. Despite this exposure, neither histamine nor imidazole forms stable complexes with the heme at ambient temperatures [209], probably because of the tight packing of the distal side of the heme against the -sandwich (Fig. 11C). The only sequence homology found for this protein is with an exonuclease [210] and with some inositol polyphosphate binding proteins and phosphatase enzymes [211,212]. The structure of one of the latter has been reported, and its protein backbone overlays very well with that of the Cimex nitrophorin [185], as shown in Fig. 12. The fact that Rhodnius and Cimex come from different families of Hemiptera suggests that occurrence of salivary nitrovasodilators in blood-sucking insects may be fairly widespread, and may serve as a common strategy for such insects, to assure them a sufficient blood meal. The fact that other members of each family do not have salivary nitrophorins suggests that these proteins developed in the particular species in which they are found via convergent evolution. A fascinating finding during the structural investigations of this protein is that when crystals were soaked in an argon-saturated solution for 1 h, followed by moving it to a similar solution saturated with NO for 20 min and then flash-frozen in liquid nitrogen, the diffraction data obtained for this crystal revealed a structure having NO bound to the heme in the distal pocket, as expected, but also with a modification to the proximal cysteine consistent with formation of a S-nitrosyl (SNO) conjugate, as shown in Fig. 13B,

Diphosphoinositol

Heme

Fig. 12. Overlay of the protein backbone structures of cNP and IPP5P [185,210] showing the similarity in protein fold. The protein backbone of cNP is shown in dark grey and that of IPP5P in light grey; the visible atoms of heme, the proximal cysteine and diphosphoinositol are shown as ball and stick. The color version of this figure, and the caption describing the color key, can be found at the end of this volume. (see Plate 13.)

Interaction of NO with Insect Nitrophorins (A)

399 (B)

Cys 60 S-nitroso Cys 60

(D)

(C)

0

3300

3400

EPR Intensity

EPR Intensity

3200

1000

2000

3000

4000

Magnetic field/G

5000

6000

0

1000

2000

3000

4000

5000

6000

Magnetic field/G

Fig. 13. Electron density maps of the heme in the absence (A) and presence (B) of NO. Electron density is shown as wires, the heme and cysteine 60 are represented as sticks. Note that in the presence of NO there is no electron density between the cysteine sulfur and the heme iron, but that there is electron density to the side of the cysteine sulfur that is consistent with formation of a thionitrosyl (Cys SNO). EPR spectra in the absence (C) and presence (D) of NO. Reprinted from [185] with permission from Proc. Natl. Acad. Sci. USA.

as compared to the heme-bound cysteine present in the absence of NO (Fig. 13A). We have investigated the protein in the absence of NO by EPR spectroscopy and found that it indeed has similar EPR spectra to those of cytochromes P450 in both the high-spin Fe(III) state [213], as shown in Fig. 9C, and the low-spin Fe(III) complex present as a minor species in Fig. 13C, or formed by addition of imidazole at the low temperatures used for EPR spectroscopy [214], indicating the presence of a cysteinate proximal ligand to the heme [214]. The very bent Fe N O bond, 119 , shown in Fig. 13B is typical of a {FeNO}7 center [93], and indeed, an EPR spectrum is observed for this complex in frozen solution, as shown in Fig. 13D. The shape of the EPR signal is indicative of a Fe(II) NO complex that has no sixth ligand [215], which is consistent with the electron density map of the complex shown in Fig. 13B. Thus, formation of the Cys SNO moiety is accomplished by homolytic cleavage of the Fe S bond to produce Fe(II) NO and a

400

F. Ann Walker O O

N + NO

+ NO Fe(III)

Fe(III)

Fe(II) – NO

– NO S–1

N

S–1

S

N O

Fig. 14. The reaction of NO with cNP first forms an EPR-silent Fe(III) NO complex with the Fe N O bond expected to be linear; homolytic cleavage of the Fe S bond and reaction of the thiyl radical with the second NO forms the Cys SNO, and at the same time the Fe NO moiety has become a Fe(II) NO complex, which is EPR active. Reprinted from [185] with permission from Proc. Natl. Acad. Sci. USA.

thiyl radical, which can readily react with a second molecule of NO, as diagrammed in Fig. 14. Both steps of the reaction of NO with cNP shown in the Fig. 14 are reversible at pH 5.6, and the spectral changes that occur at each step exhibit excellent isosbestic behavior [184]. Cys SNO formation from S− and NO requires a one-electron oxidation, which is readily facilitated by metal centers. In cNP this is accomplished through reversible reduction of the heme iron [184]. In this mechanism, NO first binds to the ferric heme, giving rise to a complex that remains vulnerable to undesirable side reactions. Although not yet observed crystallographically in the protein, the loss of EPR signal and absorption spectral changes confirm the existence of the Fe(III) NO complex [184]. A model compound having a Fe(III) NO center with Fe N O bond angle of 159.6 and a thiolate ligand trans to the NO has been reported [216]. This complex has a Fe S bond length of 2.359 Å. In cNP, a second NO molecule binds at the Cys-60 thiolate, leading to reduction of the heme iron and formation of a neutral SNO conjugate. NO release, in response to lowered NO concentrations, involves simple reversal of these steps: Decomposition of Cys SNO is induced by Fe(II) [217], and release of NO from the now-ferric heme readily takes place, more so at higher pH. These data represent the first observation of reversible Cys SNO conjugation in a protein [184]. It is possible that Cys SNO formation at metal centers may be common. For example, inhibition of the ferrous state of inducible NOS (iNOS) by NO is known to occur, but apparently does not involve release of proximal Cys-415; however, if a hydrogen bond to the proximal cysteine is removed through mutation in endothelial NOS (eNOS), a five-coordinate Fe(II) NO complex is formed [218], much like that of cNP. Whether SNO formation also occurs in this case, or in any of the many proteins with cysteines linked to redox active metal centers is not known, but this would be difficult to discover without freeze-trapping in the crystal, since the typically employed spectroscopic signals do not readily detect sulfur adducts. We have worked out a method for expressing soluble Cimex holoprotein in E. coli at 20 C, supplemented with iron and -aminolevulinic acid, which is purified by an affinity His6 -tag that is subsequently cleaved (thrombin digest) before further purification to obtain very high yields of pure Cimex nitrophorin as the holoprotein [219]. We have

Interaction of NO with Insect Nitrophorins

401

used this purification method to prepare samples for NMR, Mössbauer and electrochemical investigations. Furthermore, if expressed in the presence of 57 Fe, the heme can be extracted and used to prepare other 57 Fe-labeled heme proteins from their corresponding apoproteins [219]. The protein gives a well-resolved high-spin Fe(III) 1 H NMR spectrum with proton resonances to 75 ppm similar to those found for NP2 (Fig. 7). We have also found, from NMR investigations, that the Cimex nitrophorin does not bind histamine or imidazole at high ligand concentrations (500 mM) at ambient temperatures, even though at 4.2 K we can obtain the EPR spectrum of the low-spin imidazole adduct, which has identical g- values to those of imidazole adducts of cytochromes P450 [214]. We have determined the reduction potentials of the Cimex nitrophorin in the absence and presence of NO (Table 1), and found them to be more positive than those of the Rhodnius NPs (Table 1), and thus only the ferric form is able to release NO by dilution inside the victim. Unlike the Rhodnius NPs, at pH 5.5 (the estimated pH inside the salivary gland) the reduction potential is significantly more positive and shifts further positive (117 mV over the 2 pH units) [220]. This result is consistent with the finding that the Cimex nitrophorin can bind two molecules of NO by homolysis of the Fe S bond to form the Fe(II) NO complex so that it can then react with another NO in the salivary gland to form the −SNO moiety seen in the crystal structure [184]. Because these E values are measured by equilibrium methods, we do not observe the substantial rearrangement that would be necessary to accomplish this binding and release of the second NO, with concomitant change in oxidation state of Fe(II) NO to Fe(III) NO. The spectroelectrochemical titrations were carried out with small excess NO concentration (and free NO is reduced during the titration) [220]; thus we do not observe the formation of the −SNO under these conditions; nevertheless, the positive E is consistent with that mechanism [184].

7.1. Mutants of cNP We have prepared the H18A mutant of cNP and several heme ligand mutants of cNP, including C60H, A and G [220]. The mutant genes of cNP were produced by quickchange mutagenesis, expressed in E. coli and purified as described above to obtain pure soluble protein. The H18A mutant of cNP has identical optical and EPR spectra to those of wild-type. Thus we find no evidence that H18 is an alternate ligand for the heme of this protein. Of the heme ligand mutants of cNP, we find that despite the longer side chain of H than C, the C60H mutant of cNP has very similar optical and EPR spectra to the Rhodnius NPs, and similar reactivity with NO to form an EPR-silent FeIII NO complex. Although it binds NO with similar affinity as do the Rhodnius NPs, this affinity is a factor of 10 lower than that of wild-type cNP (Table 1). The E of cNP-C60H at pH 7.5 is more than 100 mV more positive than wild-type cNP, and clearly unlike that of the Rhodnius NPs [220]. In like manner, the gene for the H57C mutant of R. prolixus NP2 was prepared, expressed and the protein characterized. It was found that the Cys thiolate does not bind to the heme, undoubtedly because the side chain of Cys is too short. The corresponding H59C mutant of NP1 was prepared by Vetter and Goodin, and the structure of its mono-histamine and bis-ImH complexes are available in the PDB (files 1U18 and 1U17, respectively).

402

Table 1. Reduction potentials and binding constants for NO to Cimex nitrophorin.a Nitrophorin

Rhodnius (NP1) Rhodnius (NP2) Cimex Cimex C60H a

E (5.5)

−274 ± 2 −287 ± 5 −99 ± 2

E (6.5)

−166 ± 4

E (7.5)

−303 ± 4 −310 ± 5 −216 ± 4 −113 ± 2

ENO (7.5)

+127 ± 4 +8 ± 3 +155 ± 2

Log KfIII NO

Log KfII NO

Log KfIII ImH

(7.5)

(7.5)

(7.5)

6 0 ± 0 1 8 3 ± 0 1 7 4 ± 0 3 6 5 ± 0 1

13 3 ± 0 1 13 6 ± 0 1 13 7 ± 0 3

6 9 ± 0 1 7 4 ± 0 1 v. small 4 2 ± 0 1

Reference

[177,194] [177,194] [222] [222]

Reduction potentials in mV vs. SHE, measured at 27 C; binding constants in log10 M−1 , measured at 27 C.

F. Ann Walker

Interaction of NO with Insect Nitrophorins

403

8. NITRIC OXIDE REACTIVITY WITH HEME CENTERS Nitric oxide is unique among diatomic molecules in being able to interact with both iron(II) and iron(III) heme proteins [69]. In the notation of Feltham and Enemark [93], these are {FeNO}7 and {FeNO}6 centers, respectively. NO reacts rapidly according to second order kinetics with the heme center in hemoglobins [221–224], myoglobins [225,226], catalase [226] and less rapidly with cytochrome c [226] in both their ferrous and ferric forms. The association rate constants are of the order of 107 –108 M−1 s−1 for ferrous Hb [221–223] and Mb and 103 –104 M−1 s−1 for most species of ferric Hb [225] and Mb [225,226] (except elephant metMb, 107 M−1 s−1 [225], for which the ferric form has no coordinated water). The dissociation rates are vastly different for the two oxidation states: For Fe(II), dissociation rate constants are of the order of 4 × 10−4 sec−1 [222], leading to very large equilibrium constants for NO binding (Keq = kf /kr = 1011 −1012 M−1 [69]); for Fe(III), dissociation rate constants range from 0.65 to 40 s−1 [225], leading to relatively small binding constants for NO (Keq = kf /kr = 103 −105 M−1 ) [69,225,226]. And unlike the binding of CO or O2 , the equilibrium constant for binding NO to unligated ferrous hemes is 103 –104 larger than when the heme carries a ligand (histidine or, presumably, cysteinate) [69,227]. This has led to the hypothesis that the loss of protein axial ligand upon binding NO to a ferrous heme protein could begin the activation of soluble guanylyl cyclase (sGC) [69]. Guanylyl cyclase itself has also been shown to have the largest known NO dissociation rate from the Fe(II) form of the protein, kobs = 6 × 10−4 s−1 , leading to a predicted half-life of the NO complex of about 2 min at 37 C [227]. This value may have implications for the mechanism of regulation of the activity of guanylyl cyclase, since this enzyme releases NO relatively quickly. Resonance Raman spectroscopy has lent strength to this hypothesis [71]. Marletta and coworkers have recently shown that guanylyl cyclase is activated by reaction of NO with the heme of sGC in a two-step reaction, the first of which has a very large rate constant (kon >1 4 × 108 M−1 s−1 ) and is NO-concentration dependent. Formation of this Fe(II) NO complex weakens the proximal histidine-Fe bond, which is then followed by a second NO-concentration-dependent reaction with a rate constant k6c-5c = 2 4 × 105 M−1 s−1 , where the second NO is believed to bind to the proximal side of the heme after the histidine has dissociated [228,229]. A similar reaction path is followed by cytochromes c [230–233], whose main function is now believed to be NO scavenging [234,235]. More recently, however, Marletta and coworkers have reported that a series of bacterial analogues of sGC that they call H-NOX domains show NO dissociation from the distal side of the heme in both five- and six-coordinate complexes (protein His ligand off and on, respectively) [236]. They have also shown that discrimination between O2 and NO by these H-NOX domains is accomplished by the presence of a tyrosine residue in the distal pocket of those that are O2 -sensors that greatly stabilizes the O2 complex [236–239]. This work is discussed in detail elsewhere in this volume. Spectroscopic techniques that have been utilized to characterize nitrosylheme proteins include magnetic circular dichroism (MCD) [240,241], infrared (IR) [177,242–246], resonance Raman (RR) [71,228–233], Mössbauer [234–238] and EPR spectroscopies [89,177,196,239–253]. MCD has been used to characterize the protein-provided axial ligand bound to both Fe(II) and Fe(III) forms of various heme proteins, using the NO, CO, phosphine and other adducts to develop MCD spectra that are unique for the fifth (protein-provided) ligand [221,222]. MCD information particularly definitive for the

404

F. Ann Walker

paramagnetic forms of heme proteins is obtained at very low temperatures (4.2 K) in the near-IR region of the optical spectrum [254,255]. IR spectroscopy has been used to characterize the N O stretching frequencies of ferrous (1675 and 1618–1635 cm−1 for five- and six-coordinate model hemes, respectively [224], 1615–1617 cm−1 for HbA [223,225] and 1611 cm−1 for NP1 [177]) and ferric (1925 cm−1 for HbA [225,227] 1917 and 1904 cm−1 for NP1 [177] and 1865–1910 cm−1 for related heme-based Fe(III) NO centers [227]), while RR spectroscopy has been used to characterize not only this stretch [228,229], but also the low-energy Fe NO stretch for ferrous (522–527 cm−1 for model hemes [232] and 551–554 cm−1 for HbA [228]) and ferric (601–603 cm−1 for models [232], 595 cm−1 for HbA [229] and 591 cm−1 for NP1 [256]), as well as porphyrin ring vibrations. It should be noted with respect to the N O stretching frequencies that those of the nitrophorin-NO complexes (and other ferric-NO complexes, 1865–1917 cm−1 ) are similar to those of ferrous CO stretching frequencies (1936, 1960 cm−1 ) [177], which lends support to the suggestion that the electron configuration is Fe(II) NO+ .

9. REDOX CHEMISTRY OF NO-HEME SYSTEMS INCLUDING THE NITROPHORINS OF R. prolixus 9.1. Background Information Since NO binds to both Fe(III) and Fe(II) hemes, the redox chemistry that links these two oxidation states of the nitrosylheme complexes is of major importance to the functioning of these proteins. Ferric nitrosylhemoglobin undergoes rapid autoreduction to the ferrous protein under 1 atm NO gas; the autoreduction process is pseudo-first-order in Fe(III) NO, the reaction rate is enhanced by medium-intensity broad-band optical illumination, and excess NO is clearly the reductant [257]. Recent studies have shown that the rate law contains not only NO, due to the pre-equilibrium between metHb and NO, but also hydroxide ion; attack of hydroxide on coordinated NO+ (Fe(III) NO ↔ Fe(II) NO+ ) is the rate determining step in the autoreduction reaction [258]. This rate determining step is followed by rapid dissociation of HNO2 and rapid binding of a second NO to the Fe(II) formed by hydroxide attack on coordinated NO+ [258]. Methemoglobin-NO autoreduces more than ten times faster than metmyoglobin-NO (k1 = 9 67 × 10−4 s−1 and 0 64 × 10−4 s−1 , respectively), and good isosbestic points are not observed [257]. However, upon “recycling” the protein, the pseudo-first-order rate constant decreases markedly (k1 = 0 18×10−4 s−1 ) and slightly different isosbestic points are observed, suggesting that some chemical modification of the protein may be involved. The authors point out that both lysine and histidine could be nitrosylated [257]. Free cysteine residues can also be nitrosylated, and Stamler and coworkers have shown evidence for a physiological role in lowering blood pressure for the nitrosylation of the Cys 93 thiols of hemoglobin [259–262]. Electrochemical investigations of several synthetic nitrosylheme complexes have been reported [263–266]. Fe(II) NO complexes can be reversibly oxidized by one electron to Fe(III) NO in the case of octaethylporphyrin (OEP) and tetraphenylporphyrin (TPP) complexes. Based upon vibrational spectroscopic data mentioned above, Fe(III) NO heme complexes are commonly believed to have the electron configuration Fe(II) NO+ ,

Interaction of NO with Insect Nitrophorins

405

and from this standpoint it is not surprising that they readily act as electrophilic nitrosating agents [267–277]. In aqueous solution, nitrosation can occur at –S, –N, –O and –C in organic molecules [267,268]. With particular relevance to biological systems, primary amines are readily deaminated and secondary and tertiary amines are readily N-nitrosated by NO+ [3]. Nitrosation of the N atoms of DNA bases can lead to carcinogenesis [274,276]. Thiols have a particularly high propensity for nitrosation under physiological conditions [259,274–277], but this requires one-electron oxidation of the thiol (or NO), as discussed above for the case of cNP. Inhibition of the catalytic activity of alcohol dehydrogenase by NO is believed to be associated with S-nitrosylation of the zinc-bound cysteine, followed by release of zinc [278].

9.2. Redox Properties of the Nitrophorins Our investigations of the electrochemistry of NP1 and NP1 NO [177] as well as the other three nitrophorins [92,279] have pointed out striking differences from the results obtained for metmyoglobin and its NO complex [92,177,257,258] and model hemes [263–265]. We have measured and reported the reduction potentials of NP1-NP4 in the absence of ligand and in the presence of NO [92,177], histamine [92,279], imidazole [279] and 4-iodopyrazole [279]. We find that in marked contrast to methemoglobin and metmyoglobin [207,257,258], the nitrosylheme protein of R. prolixus autoreduces slowly only after prolonged treatment with gaseous NO [89,102]. The reduction potentials of the four nitrophorins are about 300 mV more negative than that of metMb: The metMb reduction potential ranges from +28 to 0 mV at pH 5.5 to 7.5, while those of the nitrophorins range from −274 to −303 mV (NP1), −259 to −278 mV (NP4), −287 to −310 mV (NP2) and −321 to −335 mV (NP3) vs. the standard hydrogen electrode (SHE) over the pH range 5.5–7.5 [92,177]. The negative shifts in potential relative to metMb are consistent both with the sluggishness of autoreduction of NP1 NO by excess NO and with the presence of several buried potentially negatively charged residues in the heme pocket of NP1 [96,177]. Such negative charges have previously been shown to stabilize the Fe(III) state of mutant Mbs, thus shifting the reduction potential in the negative direction [280]. However, the very large difference in the rate of autoreduction of metMb NO (seconds) [257,258] and NP1 NO (hours) is likely mainly a result of the difficulty of OH− attack on the Fe(II) NO+ form of the protein [258] in the presence of the potential negative charges in and near the heme pocket [95,143–145,177]. The electrochemistry of MbNO and the original four NP NO complexes has also been investigated [92,177]. We find that reversible Fe(III) NO/Fe(II) NO reduction is not observed for myoglobin because of rapid dissociation of NO from the Fe(III) state of MbNO in the absence of excess NO [177]. In contrast, NP1 shows this reversible reduction at +154 mV vs. SHE at pH 5.5, and at +127 mV at pH 7.5, although NO also readily dissociates from the Fe(III) form at the latter pH [177]. NP2-NP4 exhibit similar redox behavior in the presence of NO [92]. From the shift in reduction potential upon complexation with NO and the Kd for Fe(III) NO for each of the nitrophorins, we have calculated the Kd for Fe(II) NO from the Nernst equation. ENO − E = −2 303 RT/nFlog10 Keq III /Keq II 

(1)

406

F. Ann Walker

We find that Kd = 8 fM at pH 5.5 and 50 fM at pH 7.5, or the NO binding constants at the two pH values are 1 3 × 1014 and 2 × 1013 M−1 , respectively for the Fe(II) NO complex [92,279]. These NO binding constants are larger than for most other heme proteins [69,202–207]. Such large binding constants for the Fe(II) NO state makes it all the more important that these nitrophorins are set up so as to have potentially negatively charged residues nearby and a very ruffled heme (with alternate meso-carbons shifted strongly above and below the mean plane of the porphyrin, and concomitant shifts of the -pyrrole carbons above and below the mean plane of the porphyrin ring, to produce a very nonplanar porphyrin macrocycle), to remain in the Fe(III) state, in order that NO can be released upon dilution, to do its jobs of vasodilation and preventing platelet aggregation (Fig. 2); if the Fe(II) NO oxidation state were the stable form, NO would not dissociate upon ∼100-fold dilution and pH change.

9.3. Redox Properties of Mutants of NP2: NP2-D1A The D1A mutation of recombinant NP2 was prepared to see if the Met-0 would be processed off, as it is for recombinant NP4, where Ala is the first amino acid for the recombinant and for the mature native protein. For NP4, the N-terminal protonated amino group has been shown to participate strongly in hydrogen-bonding between the amino acid side chains of the A-B and G-H loops (between the side chain carboxylates of D32 and D129) in the closed loop structure of NP4 NO [144]. The triple mutant NP4V36A/D129A/L130A was shown not to form the closed loop conformation [281]. In our work on the D1A mutant of recombinant NP2 it was found that M0 is processed off for NP2-D1A as well [187]. A new crystal structure of the NH3 complex of NP2-D1A (PDB file 2EU7) shows a partially closed loop structure that is somewhat different than that of NP4 NO, but with similar H-bonding between the A-B and G-H loops. The solution properties of NP2-D1A and its NO, histamine, imidazole and cyanide complexes are found to be consistent with formation of the closed loop form in solution. Comparison of recombinant WT NP2 and its D1A mutant shows that the FeIII /FeII reduction potentials and ligand binding constants are similar for the two, while the heme substituent chemical shifts are rather different, suggesting a difference in the shape of the distal pocket for the two proteins. Most notably, the ratio of the two heme rotational isomers A and B (Scheme 1) is very different for the two proteins, and the rate at which the A:B ratio reaches equilibrium is strikingly different for the two, with NP2-M0D1 having a half-life for heme rotation of about 2 hours while NP2-D1A has a half-life of 1.8 days [187]. The thermodynamic properties observed for NO and histamine binding and release are thus the result of a large equilibrium preference for the B heme rotational isomer for NP2-D1A (A:B = 1:22), and to a lesser extent NP3 (A:B = 1:4) and NP4 (A:B = 1.1:1) [187], and the change in this ratio on ligand binding gives us the difference in binding constants for these two rotational isomers. These differences are too small to be observed in ligand equilibrium titration experiments, but are probably responsible for the previously observed biphasic kinetics of NO dissociation for these proteins [92], because the magnitude of contribution from each phase agrees with the expected A:B ratio. The closed loop form of NP2-D1A (as for NP4) is consistent with the slower histamine binding and release kinetics observed for NP2-D1A [187]. The implications of the slow A/B hemin reorientation kinetics on the behavior of the native nitrophorin

Interaction of NO with Insect Nitrophorins

407

proteins in the salivary glands of the insect, and following injection into the tissues of the victim are that since none of the native proteins has an N-terminal methionine (because the mature proteins are produced by proteolysis of a leader sequence) their NO complexes are probably kinetically trapped at near 1:1 ratios of unstable and stable heme orientations, rather than the equilibrium A:B ratios studied in vitro [187]. Thus the NO release profile for the four nitrophorins is probably different than found previously for the equilibrium A:B ratio samples [92]. Thus, mutations to remove the Met-0 from NP1 and NP3 should also be made so that the behavior of each of the mature nativelike nitrophorins can be studied. This project has made us very aware of the dangers of making conclusions based on study of recombinant proteins. Because of the large influence on the kinetic properties of the D1A mutant of NP2, we included this mutation in all of the carboxylate mutants discussed below.

9.4. Redox Properties of Mutants of NP2: The Role of Carboxylates Near the Heme Based upon the structures of the nitrophorins, we expected that the carboxyl groups of the A-B loop plus the buried glutamic acid (E53 of NP2) would have the greatest effect on the reduction potential at pH 7.5, which is close to the pH of most tissues. We have prepared 11 carboxylate mutants of NP2-D1A, the E and ENO values at pH 7.5 of which are presented in Table 2, columns 3 and 4, and the difference (ENO –E ) in column 5. We have then subtracted the (ENO –E ) of the standard (NP2-D1A) from that of each of the other mutants, all of which include the D1A mutation, and tabulated those values in column 6 of Table 2. By doing this, we can hopefully cancel out differences in water occupation of the NO binding pockets of the various mutants in order to see how the difference in E of the NO complex changes from one mutant to another, and thus be able to evaluate the importance of the charge mutant in each case. What is apparent is that when the buried charge, E53, is removed by mutation to Q, a large change in E occurs, +94 mV at pH 7.5.j The COO− of E53 is about 6 Å from the heme edge or 10 Å from the iron, and it is buried in the mainly hydrophobic heme binding pocket. Surprisingly, the COO− of D128 has almost as large an effect, for when it is removed the potential shifts +78 mV at pH 7.5 (Table 2, column 5). This COO− is 13.25 Å from the heme edge and 13.67 Å from the Fe. (For these distance measurements we use the structure of NP2 NO, PDB file 1T68, which is the structure of the FeII NO complex. The distances quoted are the smallest of all NP2 structures available at present, and, by comparison of other Fe(II) NO and Fe(III) NO structures of NPs, especially NP4, the distances for the two oxidation states are almost identical.) The effect of a charge extrinsic to the heme center itself is a result of its Coulombic interaction with the oxidized heme (the Fe of which carries a formal + charge); thus the effect should be proportional to q1 q2 /dD (D = effective dielectric constant within the protein, d = distance between Fe and the charged atom, q12 = +/ − 1); this is true of a wide range of NP2 mutants, including those with charges exposed to the aqueous medium. A plot of the data from columns 2 and 6 of Table 2 is shown in Fig. 15, where the value of D that fits the dependence is 20 [282]. This assumes no S contribution, so that ECoul ≈ G. While D = 20 is not the value expected for a charge totally buried in a hydrophobic protein matrix (D ∼ 4) nor that expected for a charge exposed to

408

Table 2. Reduction potentials and binding constants for NO to NP2 WT and charged mutants at pH 7.5.a Distance from O or N to Fe, Å NP2 WT NP2-D1A NP2-D1A,AB loopb NP2-D1A,D29Ac NP2-D1A,D31Ac NP2-D1A,D36A NP2-D1A,E53Qc NP2-D1A,D89Ac NP2-D1A,D99A NP2-D1A,E100A NP2-D1A,K127Ac NP2-D1A,D128Ac NP2-D1A,E53Dc NP2-K30A

7.66 12.74 11.63 9.96 18.41 15.74 18.12 (15.73) 13.67 9.96 17.01

E

ENO

 (ENO − E )



ENO − E 

Log KfIII NO

Log KfII NO

Reference

−310 ± 5 −325 ± 4 −288 ± 4 −313 ± 2 −366 ± 3 −307 ± 4 −316 ± 4 −329 ± 3 −337 ± 4 −357 ± 5 −379 ± 3 −382 ± 3 −315 ± 2 −307 ± 3

8±3 −20 ± 2 35 ± 2 2±2 −8 ± 3 8±3 84 ± 3 17 ± 2 −9 ± 2 0±2 −20 ± 4 2±2 10 ± 2 41 ± 2

318 ± 6 305 ± 5 323 ± 5 315 ± 3 358 ± 4 315 ± 5 400 ± 4 336 ± 4 328 ± 5 357 ± 6 359 ± 5 384 ± 4 325 ± 3 348 ± 3

– Standard 17 ± 7 10 ± 5 52 ± 6 10 ± 7 94 ± 6 31 ± 6 23 ± 7 51 ± 7 54 ± 7 78 ± 6 20 ± 5 30 ± 6

8 3 ± 0 1 8 3 ± 0 1 – 8 0 ± 0 2 8 0 ± 0 1 8 3 ± 0 1 7 3 ± 0 1 8 3 ± 0 1 8 0 ± 0 1 8 3 ± 0 1 8 4 ± 0 1 8 1 ± 0 1 8 0 ± 0 1 8 8 ± 0 3

13 6 ± 0 1 13 4 ± 0 1 – 13 3 ± 0 2 14 0 ± 0 1 13 6 ± 0 1 14 1 ± 0 1 14 0 ± 0 1 13 6 ± 0 1 14 3 ± 0 1 14 5 ± 0 1 14 6 ± 0 1 13 5 ± 0 1 14 7 ± 0 3

[177,187] [187] [303] [303] [303] [303] [303] [303] [303] [303] [303] [303] [303] [303]

Reduction potentials in mV vs. SHE, measured at 27 C; binding constants in log10 M−1 , measured at 27 C. The AB loop of NP2 (YLDKDPQVTDQY) was replaced with that of NP1 (YLDLEPDDVPKRY). This makes no difference in overall charge but increases the number of charges in the loop by two, and amino acid residues by one. c Underlined amino acids are conserved among all five NPs.

a

b

F. Ann Walker

Interaction of NO with Insect Nitrophorins

409 350

0.5

300 0.4

0.3

200

1

150

0.2

2 0.1

100

3

5

4 8

6 12

0.0 0

2

4

6

10

12

7 9

13 8

ΔE° (mV)

1/d (Å–1)

250

10 11

50 0

14

16

18

20

d (Å)

Fig. 15. Summary of E shifts from those of NP2–D1A as a function of the distance d between the redox-active heme and the carboxylate. The distance d is measured from the iron center to the closest O atom of the COO− . Data points 1 to 4 (filled triangles) are those for which the COO− is buried inside the protein 1: V63E or D [301]; 2: NP2–D1A,E53Q (this work); 3: NP4– E55Q; 4: Rhodobacter sphaeroides bc1 complex cyt b E295Q [302]. The remaining data point are surface COO− 5: NP4–D30N; 6: NP4–D30A; and 7: NP2–D1A,D31A; 8: –D128A; 9: –D99A; 10: –E100A; 11: –D89A (this work). From these data it appears that there is no difference in the effect of buried charges as compared to surface charges on the reduction potential of the heme. Open circles are those of NP2–D1A,D29A (12) and –D36A (13), and, to some extent, –D99A (9), which do not follow the correlation (see text).

the aqueous medium (D = 78.5), it provides a reasonable correlation not only for the NP2-D1A mutants, but also several examples of supposedly buried charges, including the E55Q mutant of NP4 [282], the V63E and D mutants of recombinant sperm whale Mb [280] and also for the cytochrome b E295Q mutant of the Rhodobacter sphaeroides bc1 complex [283]. What is constant in all these systems is the buried heme, which is always found to have several water molecules associated with it, and not only with the propionate COO− s. Furthermore, after first trying a dielectric constant of 12 for the protein interior in the vicinity of a charged group [284], a dielectric constant of 20 was finally found to fit the ionization equilibria of histidines in Mb [285] and staphylococcal nuclease [286]. The authors explain this finding as being a result of “weaker than expected electrostatic effects, even when the protein interior is treated as a medium with high dielectric constant” [285]. Others have also found D = 20 for E s of heme or chlorophyll-containing proteins [287,288]. Two charge mutants do not obey the correlation shown in Fig. 15, the D29A and D36A mutants of NP2-D1A (points 12 and 13). Both of these COO− side chains are in the A-B loop and both are very much on the surface of the protein. D36 COO− NH3 complex (2.89 Å, is an H-bond acceptor from K62 NH+ 3 in the pH 7.7 NP2 PDB file 1EUO), the pH 6.5 ImH complex (2.74 Å, PDB file 1PEE) and one pH 6.5 NP2 FeII H2 O complex (2.91 Å, PDB file 2AL0), but is somewhat further away

410

F. Ann Walker

for the pH 6.5 NP2 FeIII H2 O structure (3.33 Å, PDB file 2AH7). These structures represent two different crystal forms, with 1EUO having unit cell P21 21 2 and 1PEE, 2AH7 and 2AL0 having unit cell P41 21 2, so the good H-bond between D36 COO− and K62 –NH+ 3 side chains is not limited to one crystal form. Because K62 is on the surface of the protein it should be able to move freely, and it is likely that this H-bond exists in homogeneous solution. This provides a good explanation for the small change in (ENO –E ) (10 mV) for this residue, column 6, Table 2. Likewise, K101 makes an H-bond to D99 in the three P41 21 2 structures, but not in the P21 21 2 structures. The smaller effect of this COO− on (ENO –E ) (23 mV) for its distance from Fe (15.74 Å) may be due to this H-bond [282]. However, no such H-bond interactions exist in any of the above-mentioned NP2 crystal structures for D29, and the D29A mutant of NP-D1A exhibits such a small

(ENO –E ) value of 10 mV when, considering the distance of the COO− from Fe, a value of 75–80 mV would have been expected. Furthermore, for NP4 the corresponding COO− , D30, the two mutants studied and included in the plot of Fig. 15 do show much larger (ENO –E ), as expected. However, it may be that D29 in its protonated form H-bonds to backbone N H and C O atoms as seen for NP4. Another puzzle is why K127A changes the E by 54 mV, but in the same direction as the D128A mutant of NP2-D1A [282]. Both of these residues are on the surface of the protein and K127 does not appear to interact with any other protein side chains other than in one NP4 crystal structure (PDB file 1D2U) in which the equivalent K128 amino group is only 2.8 Å from a heme carboxylate. K127 and D128 are both conserved amino acids amongst the five NPs, and thus they must be particularly important to the redox chemistry of the heme iron. Further investigations must be carried out to probe the reason for the opposite effect of K127.

9.5. Redox Properties of Mutants of NP2: Belt Mutants The heme group in all of the wild-type (WT) nitrophorins is ruffled [95,99,143–146], and is even more ruffled for the NO complexes [143–145,289]. The structure of NP4 NO indicates that close contacts of both a series of mainly aromatic residues that form a “catcher’s mitt” around the heme (Fig. 16), as well as the two distal pocket leucines (L123 and L133) probably contribute significantly to this ruffling (Fig. 6), and may help to stabilize the Fe(III) NO state [145,309], and recent structures of NP2 NO also show the close contacts of the two distal leucines (L122 and L132 in the case of NP2); the NP2 heme is even more ruffled than the NP4 heme [289]. Hence, we have investigated the role of both the “belt” residues [195] and the distal leucines [194] of NP2 on the reduction potential of the protein. The so-called “belt” residues, which are mainly hydrophobic, and mainly aromatic residues positioned around the heme are shown in Fig. 16. These include, for NP2, V24, F27, Y38, F42, E53, Y104 and Y85, although under the heme and parallel to H57, F66 was also considered important. We felt that the packing of these residues was likely important in helping to cause heme ruffling, and thus stabilization of the FeIII form, as well as in determining the orientation of the heme. We have thus prepared the X→A mutant of all of these, except for E53, where we made the E53Q and E53D mutants, the E53D of which is smaller but preserves the negative charge. We find that all “belt”

Interaction of NO with Insect Nitrophorins

411

Val24 Phe42

Phe27

Glu53

Tyr104

Tyr38

Tyr85

Fig. 16. Residues surrounding the heme, as viewed from above (left) and from the front of the heme pocket (right). The color version of this figure, and the caption describing the color key, can be found at the end of this volume.

mutants except Y85A at pH 7.5 shift the reduction potential more positive than the protein without the “belt” mutation. The Y38A, S40A, F42A, E53D, F66A and Y104A each shift the heme reduction potential positive by 42–70 mV, while F27A results in only a 19 mV positive shift and Y85A actually results in essentially no shift in potential at pH 7.5 [309]. The positive shift in E difference between the standard (NP2 WT or NP2-D1A) upon binding NO and the mutant produced by removing the large, usually aromatic residue indicates that the Fe(II) form of the protein is significantly stabilized by this removal, and thus the majority of the residues investigated are involved in helping to stabilize the Fe(III) form that is able to release NO into the tissues of the victim at the expected physiological concentrations.

9.6. Redox Properties of Mutants of NP2: Distal Pocket Mutants The strong ruffling of the heme in the structure of NP4 NO, for which the heme binding center of the protein is shown in Fig. 6, suggested to us that the Fe(III) NO• valence isomer of the complex might be stabilized by this strong ruffling. We thus suspected that the two distal leucines of that protein might play an important role in heme ruffling and hence the stability of the Fe(III) form of the complex, and that removal of the pressure of one or both of the distal Leu by mutation to smaller residues (Val, Ala) might relieve the strain and allow the heme to flatten. We thus produced, first, the L122V, L132V single mutants, and the L122,132V double mutant of NP2 and investigated them by spectroelectrochemistry, NMR and resonance Raman spectroscopy, and part of this work has been published [194]. Since then we have prepared the L122A and L132A single and L122,132A double mutants, as well as the I120V and I120T mutants and studied them by the same techniques. We find that replacing L132 with the smaller side chain-containing Val or Ala causes large changes in E over the pH range 5.5–7.5; the buried distal L132 has

412

F. Ann Walker

a significant effect on reduction potential, whereas the distal L122 near the opening of the binding pocket appears to have minimal effect, and thus solvent accessibility is not necessarily a factor in the redox reaction [195]. The fact that additional water molecules in the binding pocket have a significant effect is most apparent for the pH dependence, which for the wild-type protein shifted 23 mV negative between pH 5.5 and 7.5 (ligand-free E ), 21 mV for the L122V mutant, 65 mV for the L132V mutant, 140 mV for the L122V,L132V double mutant [194] and 115 mV for the L122A,L132A double mutant of our most recent work [194]. In fact, for the NP2-L122V,L132V double mutant in particular it was found that there was an extremely large change in potential over the pH range 5.5 to 6.5 (114 mV), and a much smaller change between pH 6.5 and 7.5 (26 mV) that is fairly typical among all other nitrophorin reduction potentials measured previously [92,177,195]. The NP2-L122A, L132A double mutant also shows a large change in potential between pH 5.5 and 7.5, but in this case the change is small between pH 5.5 and 6.5 (24 mV) while that between pH 6.5 and 7.5 is large (91 mV) [195]. The 114 mV potential shift between pH 5.5 and 6.5 has been interpreted as being indicative of a proton-coupled reduction reaction with loss of two protons over the single pH unit (114/59 = 1.93 H+ ) [194], while the L→A change is somewhat smaller than expected for two protons E /(59 mV) = 1.54 H+ [195]. Since the mutations do not involve protonatable residues, the change must be due to additional water molecules that can fit into the cavity created by the mutations that are involved in hydrogen bonding close to the Fe center and help to stabilize the FeIII state relative to the FeII state due to a charge-dipole effect. The water molecules require charged or polar groups (including other water molecules) with which to form H-bonds inside the hydrophobic pocket. Indeed, since that time several structures of the NP2-L122V, L132V double mutant have been solved, and that protein, to which no ligand was added, was shown to have at least one additional water molecule in the distal pocket [147]. This pH dependence is most apparent for the ligand-free, and less apparent for the NO- and ImH-bound nitrophorins, because the presence of the ligand partially fills the cavity. With respect to the single mutants, the fact that NP2-L132V and A single mutants behave very similarly to the double mutant while NP2-L122V and A single mutants have little effect on the properties of the protein is consistent with our NMR findings mentioned above that the H/D exchange rates of the part of the -barrel on the side of the protein where L122 resides are much faster than those on the side where L132 resides; the slower H/D exchange rates for the “right side” of the -barrel in the view of Fig. 5 are believed to be due to the hydrophobic interaction between helix 2 (also on the right side in picture of Fig. 5), which also has slow H/D exchange rates (Fig. 10). Thus we postulate that the “right side” of the -barrel is much more rigid than the “left side”.

10. POSSIBLE ROLE OF HEME RUFFLING IN STABILIZING THE {FENO}6 CENTER OF THE NITROPHORIN–NO COMPLEXES The hemes of all R prolixus nitrophorin complexes are somewhat ruffled, but those of NP4-NO [145] and NP2 NO [289], NP2 H2 O [289], NP4 CN− [145] and NP2 CN− [289] are highly ruffled, much more ruffled than the NP4 NH3 complex [143]. Furthermore, the resolutions of these structures are higher than those of structures

Interaction of NO with Insect Nitrophorins

413

of NP1 [95] and the first structure of NP2 [146], thus making it possible to measure accurately the deviations of the 25 atoms of the heme core from the mean plane, a measure that has frequently been used for model heme complexes [310]. More recent structures of a number of NP2 ligand compexes have been obtained to high resolution [289], and show that in all cases the NP2 complexes are more ruffled than the corresponding NP4 complexes (Table 3). Although these structures are highly ruffled, the most highly nonplanar hemes are found in the heme-NO and oxygen binding (H-NOX) domain of the O2 -binding protein of the obligate anaerobe Thermoanaerobacter tengcongensis, where the hemes in two crystalline forms have rms deviations of 0.33, 0.44, 0.45 and 0.46 Å, which normal coordinate structural decomposition shows to be composed of close to equal contributions from saddled and ruffled conformations [311]. This protein domain has a sequence very similar to that of human soluble guanylyl cyclase (sGC) and is believed to have a structure similar to that of sGC [312]. Table 3. RMSD from 25-atom mean plane for various Fe(III) and Fe(II) heme complexes Complex Fe(III)–L Complexes

Fe electron Configuration

RMSD, 0.01 Å

Type of Distortion

Reference

NP4-NH3 NP4-NO NP4-CN− NP4-t-BuNCa NP4-Histamine NP4-ImHa NP4-4IPzHa

d5 d5 d5 d5 d5 d5 d5

13.8 18.9 21.4 14.5 13.1 11.8 8.8

[143] [145] [145] [347] [145] [298] [298]

NP2-H2 O NP2-NO L122,132V NP2-H2 O L122,132V NP2-NO [(OEP)Fe (1-MeIm)(NO)]ClOa4 [(OEP)Fe((PzH)(NO)]ClOa4 [(OEP)Fe(Iz)(NO)]ClOa4 [(OEP)Fe(Prz)(NO)]ClOa4 [(OEP)Fe(PzH)(NO)]ClOa4 {[(OEP)Fe(NO)]2 (Prz)}ClOa4

d5 d5 d5 d5 d5

25.0 28.5 20.8 24.6 10.0

Ruffled Ruffled Ruffled Ruffled Ruffled ∼Ruffled ∼Planar (Ruffled) Ruffled Ruffled Ruffled Ruffled Ruffled

d5 d5 d5 d5 d5

9.5 20.9b 10.5 8.0 4.9

[313] [313] [313] [313] [313]

[(TMP)Fe(1-MeIm)2 ]ClOa4 [(TMP)Fe(1-MeIm)2 ]ClOa4 [(OEP)Fe(4-NMe2 Py)2 ]ClOa4 [(TMP)Fe(4-NMe2 Py)2 ]ClOa4 [(TMP)Fe(4-CNPy)2 ]ClOa4 [(TPP)Fe(4-CNPy)2 ]ClOa4 [(TPP)Fe(t-BuNC)2 ]ClOa4 [(p-TTP)Fe(2,6XylylNC)2 ]ClOa4 [(OEP)Fe(t-BuNC)2 ]ClOa4

d5 d5 d5 d5 d5 d5 d5 d5

1.8 5.4 6.0 31.7 24.4 35.6 36.8 33.8

∼Ruffled Ruffled Ruffled Ruffled ∼Planar (Saddled) Planar ∼Planar ∼Planar Ruffledc Ruffledc Ruffledc Ruffledc Ruffledc

[314] [314] [314] [314] [326] [315] [317] [316]

d5

23.2

Ruffled

[317]

[308] [308] [308] [308] [313]

(Continued)

414

F. Ann Walker

Table 3. (Continued) Complex Fe(III)–L Complexes

Fe electron Configuration

RMSD, 0.01 Å

Type of Distortion

[(OETPP)Fe(1-MeIm)2 ]ClOa4

d5

Saddled

[328]

perp-[(OMTPP)Fe (1-MeIm)2 ]ClOa4 paral-[(OMTPP)Fe (1-MeIm)2 ]ClOa4 [(TC6 TPP)Fe(1-MeIm)2 ]ClOa4

d5

Saddled

[328]

Saddled

[328]

Saddled

[328]

[(OETPP)Fe (4-NMe2 Py)2 ]ClOa4 [(OMTPP)Fe (4-NMe2 Py)2 ]ClOa4 [(OMTPP)Fe (4-NMe2 Py)2 ]ClOa4 [(OMTPP)Fe(t-BuNC)2 ]ClOa4

d5

Saddled

[327]

Saddled

[328]

Saddled

[328]

Saddled

[329]

[(OMTPP)Fe(t-BuNC)2 ]ClOa4

d5

(mean meso displ. = 3) (mean meso displ. = 10) (mean meso displ. = 1) (mean meso displ. = 26) (mean meso displ. = 28) (mean meso displ. = 35) (mean meso displ. = 7) (mean meso displ. = 3) (mean meso displ. = 6)

Saddled

[329]

Fe(II)–L Complexes [(OEP)Fe(t-BuNC)2 ]a

d6

Saddled

[318]

d6 d6

42.1 (mean meso displ. = 1) 5 10

[319] [320]

d6

8

∼Planar ∼Ruffled/ Planarcd ∼Planarcd

d6

7

∼Planarcd

[321]

d6

(mean meso displ = 15.5) (mean meso displ = 29) (mean meso displ = 44) 2.5 2.6 3.8

Ruffledce

[322]

Ruffledce

[322]

Ruffledce

[322]

Planar Planar Planar

[323] [323] [323]

[(TPP)Fe(Py)(CO)]a [(C2 -capTPP)Fe (1-MeIm)(CO)]a [(OC3 O-capTPP)Fe (1-MeIm)(CO)]a [(OC3 O-capTPP)Fe (1,2-Me2 Im)(CO)]a [(10-strapTPP)Fe (1-MeIm)(CO)]a [(8-strapTPP)Fe (1-MeIm)(CO)]a [(6-strapTPP)Fe (1-MeIm)(CO)]a [(TMP)Fe(4-CNPy)2 ]a [(TMP)Fe(3-CNPy)2 ]a [(TMP)Fe(4-MePy)2 ]a

d5 d5

d5 d5 d5

d6 d6 d6 d6 d6

Reference

[321]

a Ligand abbreviations: t-BuNC = t-butylisocyanide, ImH = imidazole, 4IPzH = 4-iodopyrazole, 1-MeIm = 1 methylimidazole, PzH = pyrazole, Iz = indazole, Prz = pyrazine, 4-NMe2 Py = 4-dimethylaminopyridine, 4-CNPy = 4-cyanopyridine, 2,6-XylylNC = 2,6-xylylisocyanide, Py = pyridine, 1,2-Me2 Im = 1,2dimethylimidazole, 3-CNPy = 3-cyanopyridine, 4-MePy = 4-methylpyridine. b Fused six-membered ligand ring may cause crystal packing effects that induce ruffling. c TPPs are always more ruffled. d Superstructure may cause ruffling. e Strap definitely causes ruffling.

Interaction of NO with Insect Nitrophorins

415

Formal core diagrams of the nitrophorin complexes of NP4 [143,144] and NP2 [289] have been constructed, and root-mean-square deviations (RMSD) from the mean plane, in units of 0.01 Å, are listed in Table 3, where they are compared to the values obtained for relevant model heme complexes [313–329]. In analyzing these data, two factors must be noted: (1) Among synthetic hemes, it is known that meso-substituted metalloporphyrinates are in general more ruffled than -pyrrole-only-substituted metalloporphyrinates [310] and (2) Fe(III) porphyrinates are usually much more ruffled than Fe(II) porphyrinates [310,317,323,324]. From consideration of the data of Table 3, it appears that RMSD values of less than 0.10 Å may be considered essentially planar, while as RMSD values increase above 0.10 Å, a moderately to significantly ruffled heme is indicated. Of the reported structures of six-coordinate Fe(III) NO synthetic porphyrinates [313], only the complex in which the sixth ligand is indazole is ruffled to an extent similar to that of NP4 NO and NP2 NO. No comment is made about any difference in crystal packing of this complex as compared to the others [313], but it is possible that the larger ligand, with a six-membered ring fused to the five-membered pyrazole ring, takes up enough space in the crystal so as to contribute to this ruffling. Neglecting this complex, all other [(OEP)Fe(L)(NO)]+ complexes have RMSD values less than or close to 0.10 Å, while NP4 NO has a value nearly twice that, 0.189 Å, and the cyanide complex has a larger yet RMSD (0.214 Å). As mentioned above, all complexes of NP2 are more ruffled than the corresponding NP4 complexes, with the NP2 NO complex having a RMSD (0.285 Å) [221], which approaches three times that of the model hemes of the same coordination. Although all of the Fe(III) NO complexes may be classified according to the notation of Feltham and Enemark as {FeNO}6 complexes [93], we choose in this discussion and in Table 3 to classify them according to their metal d-electron count, as d5 systems, for reasons that will soon become apparent. In this classification, we can compare the -NO and -CN− complexes directly, and can also compare them to Fe(III) bis-ligand complexes of other types. There are two possible electron configurations of low-spin Fe(III) porphyrinates, the more commonly observed (dxy )2 (dxz , dyz )3 , or d , electron configuration, which is usually observed for the elecron-transferring ferricytochromes a, b, c, f and o [324–328], and the “novel” (dxz ,dyz 4 (dxy 1 , or dxy , electron configuration, which is often observed in model hemes having axial ligands that are good -acceptors, such as isocyanides [316,317,329], low-basicity pyridines [315], and under certain conditions, cyanide ions [330–333]. The dxy ground state is also believed to exist for some reduced hemes in biological systems, including heme d and d1 and siroheme, but this appears to depend upon the nature of the axial ligands [334–336]. A highly ruffled heme macrocycle is one of the hallmarks of dxy ground state complexes, although many d heme centers are also ruffled. Rousseau and coworkers havve noted that the NOS heme is also quite ruffled, and that this may indicate a major role for the dxy electron configuration in that heme protein [337,338]. Probably the most relevant comparison of porphyrin ring ruffling that can be made, among model heme complexes, is to the bis-(tert-butylisocyanide) complex of (OEP)Fe(III). This complex is highly ruffled, with a RMSD of 0.232 Å [317], while the Fe(II) analog, although highly saddled, has RMSD meso-carbon displacements of only 0.01 Å [318], and thus has no ruffling component. Meso-only substituted metalloporphyrins having six-membered ring axial ligands in perpendicular planes, as is the case for all of the (TMP)Fe(III) bis-pyridine complexes, are highly ruffled [314,326],

416

F. Ann Walker

whereas the corresponding Fe(II) complexes, all of which have the axial pyridine ligands in parallel planes [323], are quite planar; Fe(III) porphyrinates having axial ligands in parallel planes, such as [(TMP)Fe(1-MeIm)2 ]+ and [(OEP)Fe(4-NMe2 Py)2 ]+ , are also quite planar [314]. Fe(III) dodecasubstituted porphyrin complexes such as the octaalkyltetraphenylporphyrins, which have both meso- and -pyrrole substituents, are highly saddled, irrespective of axial ligand(s) or coordination number, but often have a ruffling component as well [327,328,339]. However, the bis-t-butylisocyanide complex of octamethyltetraphenylporphyrinatoiron(III) is purely saddled in two crystalline forms [329], showing that a low-spin Fe(III) complex with an electron configuration of dxz  dyz 4 dxy 1 can exist without ruffling of the porphyrinate ring. Thus, apart from model hemes having planar axial ligands in nearly perpendicular planes over the meso positions, most of the cases for which the porphyrinate ring is highly ruffled are those for which the Fe(III) center has the electron configuration dxz  dyz 4 dxy 1 , that is, the bis-isocyanide complexes of both (OEP)Fe(III) [317] and various (TPP)Fe(III) derivatives [316,317], and the bis-4-cyanopyridine complex of (TPP)Fe(III) [315], as well as the bis-cyanide complexes of meso-alkyl-substituted hemins [327,332–334] and other modified porphyrin ring systems [333]. Complexes of this electron configuration have EPR g-values close to 2.0 [315–317,325], as well as unique NMR, Mössbauer and MCD spectra [325], that readily differentiate them from the more typical low-spin Fe(III) complexes that have the electron configuration dxy 2 dxz  dyz 3 , such as the vast majority of the cytochromes a, b, c, f and o [324,325]. It has been suggested [315–317,325] that the reason for the strong departure of the porphyrin ring conformation from planarity in the bis-isocyanide and related complexes of Fe(III) porphyrins is electronic, because the dxy unpaired electron does not have proper symmetry to engage in porphyrin → Fe  donation unless the porphyrin ring ruffles so that the nitrogen pz orbitals are twisted away from the normal to the mean plane, as shown in Fig. 17, and thus have a component in the xy plane that has proper

–

+ +

– +

+ – –

– +

–

+ –

+

+ Py Sin 15°

Px Sin 15° –

X

Y

Fig. 17. Possible interactions of the dxy metal orbital with the porphyrin nitrogens in the case of strong S4 ruffling. The remaining nitrogen p projections have the proper symmetry to allow delocalization via the 3a2u () porphyrin orbital. Reprinted [317] with permission from the American Chemical Society.

Interaction of NO with Insect Nitrophorins

417

symmetry to overlap with the dxy orbital. This type of overlap allows significant spin delocalization via  donation from the a2u () orbital of the porphyrin ring to the hole in the dxy orbital, and gives rise to very large contact shifts at the meso positions and very small contact shifts at the -pyrrole positions of the porphyrin ring [315–317,325], thus yielding unique NMR and EPR spectra for all metalloporphyrins with dxz  dyz 4 dxy 1 electron configurations. Since the nitrophorin-NO complexes are even-electron systems, no EPR signal is observed, and it is thus not possible to determine what the electron configuration of the metal actually is. If it were dxz  dyz 4 dxy 1 , then the unpaired electron of NO would be in an orbital that is orthogonal to the metal dxy orbital, and there could therefore be no direct bonding interaction between the two unpaired electrons. Thus the NO unpaired electron would be isolated on the NO unit in the ruffled heme complex, which would allow facile bending of the Fe NO bond, possibly contributing to the nonlinearity of the NP4 NO unit in the latest structure (Fig. 6) [145]. The Fe(dxy ) and NO• unpaired electrons would therefore, because of their proximity in space, simply have to be magnetically coupled, either antiferromagnetically (antiparallel spins) or ferromagnetically (parallel spins). Mössbauer spectroscopic investigations show that for both model ferriheme–NO complexes that undoubtedly have quite planar porphyrin cores in frozen solution [340] and the highly ruffled NP2 NO and NP4 NO complexes, also in frozen solution [341], the spectra are those of fully electron-paired “diamagnetic” species, and whether those electrons are truly paired, as in a Fe(II) NO+ electron configuration, or strongly antiferromagnetically coupled, as in a Fe(III)(dxy ) NO• configuration, cannot be differentiated from the Mössbauer data. If the latter were the case, then this antiferromagnetically coupled configuration of Fe(III) NO would allow more facile departure of NO upon dilution of the protein into the tissues of the victim than would the valence tautomer, Fe(II) NO+ , which has long been believed to be the likely electron configuration of the {FeNO}6 centers, based on infrared spectroscopic data [177], as mentioned above. However, both the original review of Feltham and Enemark [93], as well as a recent chapter by Westcott and Enemark [342], emphasize the fact that the {FeNO}n triatomic fragment is highly covalent, and thus assigning electrons to the metal and NO may be meaningless, except for those die-hard scientists such as the author who want to know the exact orbital occupation and possible coupling mechanism involved. DFT calculations on the {FeNO}6 nitrophorin centers do not allow “visualization” of the possibility of the existence of the antiferromagnetically coupled Fe(III) NO• electron configuration, because the even number of electrons and apparent diamagnetism of the system leads to the calculations assuming complete pairing of the electrons. In all calculations, whether for the ruffled porphyrin core observed for the NP4 NO complex, an optimized planar core, or an optimized planar core that is “relaxed” to allow ruffling, the 3a2u  porphyrin orbital is the HOMO, but it contains no significant amount of metal character in any of these cases [341]. However, the larger Mössbauer quadrupole splitting observed for NP2 NO (1.84 mm s−1 ) [341] as compared to that of NP4 NO (1.61 mm s−1 ) [341] and a model heme complex [OEPFeIII (NO)(1-MeIm)]+ (1.61 mm s−1 ) [340] can only be calculated if it is assumed that the heme is very ruffled [341]. Since the NP4 NH3 complex, which has the open conformation for the loops surrounding the distal heme pocket, is also somewhat ruffled (RMSD = 0.138 Å, Table 1), it is tempting to speculate that the leucines pointing toward the heme in the distal pocket predispose all nitrophorins to be ruffled, thus stabilizing the dxz  dyz 4 dxy 1 ground state for Fe(III). However, both the NP4 NH3 and the NP4 CN− complexes have totally

418

F. Ann Walker

normal “large gmax ” EPR signals at g = 3 41 and 3.28, respectively [343], indicating that both of these complexes have dxy 2 dxz  dyz 3 ground states, yet the cyanide complex is more ruffled and the NH3 complex is much less ruffled than the NO complex, and both have open loop conformations. But these facts do not tell us about the NP4 NO or NP2 NO complex electron configuration. The NP4–histamine complex, which also has the open loop conformation, is significantly ruffled, considering the fact that the dihedral angle between the ligand planes of the proximal histidine and histamine imidazoles is approximately 29 , which is much closer to parallel than to perpendicular orientation of the axial ligands, and its EPR spectrum is that of the normal rhombic type (g = 2 92, 2.24, 1.52) [344]. Thus, one might have expected this complex to have a nearly planar heme, as do those of [(TMP)Fe(1-MeIm)2 ]+ and [(OEP)Fe(4-NMe2 Py)2 ]+ [314]. The imidazole complex, which has the closed loop conformation, is among the least ruffled complexes of NP4, and the 4-iodopyrazole complex, which has a halfclosed loop conformation, has the very least ruffled ring conformation among all NP4 complexes (Table 3). Perhaps the most enigmatic NP4 L complex is that where L = tert-butylisocyanide, which shows an RMSD (0.145 Å) [345] not much different from the NH3 , ImH and histamine complexes, in spite of the large ruffling of the bis-t-BuNC complexes of OEPFe(III) and TPPFe(III) [317]. However, in the protein-bound complex, the axial t-butylisocyanide ligand binds very far off the heme normal [345], and it is clear that the binding of this ligand to the heme of NP4 is highly hindered. Thus no major conclusions can be made with regard to the much smaller RMSD of this complex than expected. Nevertheless, both the NMR and the EPR spectra of this complex are consistent with a dxz  dyz 4 dxy 1 electron configuration [345]. For NP2 L complexes, as mentioned above, all are more ruffled than the NP4 L counterparts (Table 3). NP2 H2 O, for example, is more ruffled (RMSD = 0.250 Å) than NP4 NO, and NP2 NO is yet more ruffled (RMSD = 0.285 Å); both L122,132V mutant complexes are less ruffled (0.208 and 0.246 Å, respectively) than the wild-type protein counterparts [308]. Clearly, the ruffled nature of the heme of NP2 and NP4 exists for all ligand complexes, and mutating the two leucines that are within van der Waals contact of the heme to valines causes only a modest decrease in ruffling [308]. It has been said that nonplanar porphyrins are easier to oxidize and more difficult to reduce than planar porphyrins, but the data on which this statement is based involves only (-bromo)1−8 tetraphenylporphyrin complexes of Fe(III)Cl [346–348] and Co(II) [349]. While the proper controls appear to have been done to separate the inductive effects of the halogens from the steric crowding effects, leaving still a large residual difference between predicted and observed oxidation potentials [346] that is consistent with the conclusion stated above, the type of nonplanarity of the porphyrinate complexes having 5-8 bromo substituents on the -pyrrole positions is saddled rather than ruffled. Unfortunately, it is the ruffled conformation which would be more appropriate for comparison to the current work on NP2 NO and NP4 NO, and only one recent study of the reduction potentials of ruffled vs. planar hemes has been carried out that has been able to factor out the effects of heme substituents [350]. In that study, the reduction potentials of imidazole or N-methylimidazole complexes of model ferrihemes with varying documented tendencies to have ruffled, planar or saddled porphyrinate ring conformations were compared to those of the corresponding 2-methylimidazole complexes. The rationale for that study was that while the imidazole or N-methylimidazole ligands could bind to all of the iron porphyrins studied in parallel planes in both oxidation states, thus encouraging

Interaction of NO with Insect Nitrophorins

419

Table 4. Reduction potentials for iron porphyrinates with imidazole ligandsa Ligand → Porphyrin ↓

Core Type

FeOETPP FeOMTPP FeTC6 TPP

Saddled Saddled Saddled (Ruffled) Ruffled or Planar Ruffled or Planar Ruffled or Planar Ruffled or Planar

FeTiPrP FeTMP Fe(2,6-Cl2 )TPP Fe(2,6-Br2 )TPP a b c d e

ImHb E1/2 (FeIII /FeII ), mV

2-MeImHb E1/2 (FeIII /FeII ), mV

“2-MeImH-ImH” E1/2 (FeIII /FeII ), mV

−430c −396c −387c

−400c −290c −300c

+30c +106c +87c

−368c −343e −169c −130de −3de

−417c

−115d

−59c −74ce −43c −82d −118d

−40de

−131d

−91d

−212d

Measured in dimethylformamide. Potentials tabulated vs. SCE, mV. ImH = imidazole, 2-MeImH = 2-methylimidazole Reference [350]. Reference [353]. 1-MeIm instead of ImH.

a planar ring conformation [314,323], the 2-methylimidazole ligands would be required to bind in perpendicular planes in both oxidation states, which would encourage ruffling of the porphyrin ring [350–353]. It was found, as summarized in Table 4, that all Fe(III) porphyrinates that could readily ruffle when the hindered 2-methyimidazole ligands were bound, including three [(2,6-X2 )4 TPPFeL2 ]+/0 complexes (X=Cl, Br or CH3 ) [353], have half-wave potentials (E1/2 ) for the Fe(III)/Fe(II) redox step that are more negative when the ligands L are 2-methylimidazole than when they are imidazole or N-methylimidazole. Since G = −nFE, a negative shift in potential indicates a stabilization of the oxidized form. The same was found to be true of the iron complex of tetra-isopropylporphyrin [350], which is known to be very ruffled [354,355]. However, for three iron octaalkyltetraphenylporphyrin complexes, where the dominant nonplanar distortion is saddled rather than ruffled, it was found that the 2-methylimidazole complexes have more positive reduction potentials for the Fe(III)/Fe(II) couple than do the imidazole complexes [350], thus indicating that for porphyrins that cannot adopt ruffled ring conformations, the binding of axial ligands to Fe(II) in perpendicular planes is not so unfavorable as it is for those that readily adopt ruffled ring conformations. Thus, in all cases, the oxidized state of ruffled heme complexes is favored over those of planar heme complexes, but by different amounts, depending on the nature of the substituents. This is probably because Fe(II) does not appear to favor a ruffled heme core [323,324]. Beyond this correlation, however, studies of additional carefully designed model heme systems, specifically designed to test the hypothesis of the effect of heme ruffling on stabilization of the oxidized (ferric) state, should be carried out. Nevertheless, it appears possible that provision of a ruffled heme at least aids in maintaining the iron in the ferric state and preventing autoreduction of the Fe(III) NO complex. Steric contacts with protein residues force the heme to be ruffled, thus favoring a ferric state for the iron

420

F. Ann Walker

in all ligand complexes, most importantly the NO complex. The closure of the mobile loops that form a hydrophobic pocket around the NO create a tightly constrained and desolvated NO binding site [144], which may further discourage reduction, since water can assist in this reaction [77].

ACKNOWLEDGEMENTS The support of National Institutes of Health grant HL54826 is gratefully acknowledged. The author also wishes to thank all of her coworkers and colleagues who appear as coauthors on the nitrophorin papers for their important contributions to this work.

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29]

Moncada, S., Palmer, R.M.J. and Higgs, E.A. (1991) Pharmacol. Rev., 43, 109–142. Snyder, S.H. (1992) Science, 257, 494–496. Stamler, J.S., Singel, D.J. and Loscalzo J. (1992) Science, 258, 1898–1902. Lancaster J.R., Jr. (1992) Amer. Scientist., 80, 248–259. Butler, A.R. and Williams, D.L.H. (1993) Chem. Soc. Revs., 233–241. Feldman, P.L., Griffith, O.W. and Stuehr, D.J. (1993) Chem. Eng. News., 26–38. Lancaster, J.R., Jr. (1994) In Encyclopedia of Inorganic Chemistry (King, R.B., ed.) Chichester: John Wiley & Sons. Nathan, C. (1992) FASEB J., 6, 3051. Lamas, S., Marsden, P.A., Li, G.K. et al. (1992) Proc. Natl. Acad. Sci. USA., 89, 6348–6352. Janssens, S.P., Shimouchi, A., Quertermous, T. et al. (1992) J. Biol. Chem., 267, 14519– 14522. McMillan, K. and Masters, B.S.S. (1995) Biochemistry, 34, 3686–3693. Mayer, B., Wu, C., Gorren, A.C.F. et al. (1997) Biochemistry, 36, 8422–8427. Rodriguez-Crespo, P., Moënne-Loccoz, T.M., Loehr, P.R. and Ortiz de Montellano (1997) Biochemistry, 36, 8530–8538. Ghosh, D.K., Wu, C., Pitters, E. et al., (1997) Biochemistry, 36, 10609–10619. Salerno, J.C., Martásek, P., Williams, R.F. and Masters, B.S.S. (1997) Biochemistry, 36, 11821–11827. Ghosh, D.K., Abu-Soud, H.M. and Stuehr, D.J. (1995) Biochemistry, 34, 11316–11320. Hall, A.V., Antoniou, H., Wang, Y. et al., (1994) J. Biol. Chem., 269, 33082–33090. Hibbs, J.B., Jr. and Bastian, N.R. (1999) Chapter 2. In Nitric Oxide and Infection (Fang, F.C., ed.) New York: Kluwer/Plenum, pp. 13–35. Abu-Soud, H.M., Wang, J., Rousseau, D.L. et al. (1995) J. Biol. Chem., 270, 22997–23006. Bernstein, F.C., Koetzle, T.F., Williams, G.J.B. et al. (1977) J. Mol. Biol., 112, 535–542. Crane, B.R., Arvai, A.S., Gachhui, R. et al. (1997) Science, 278, 425–431. Drapier, J.-C., Pellat, C. and Henry, Y. (1991) J. Biol. Chem., 266, 10162–10167. Stadler, J., Bergonia, H.A., Di Silvio, M. et al. (1993) Arch. Biochem. Biophys., 302, 4–11. Terenzi, F., Diaz-Guerra, J.M., Casado, M. et al. (1995) J. Biol. Chem., 270, 6017–6021. Komori, Y., Wallace, G.C. and Fukuto, J.M. (1994) Arch. Biochem. Biophys., 315, 213–218. Abu-Soud, H.M., Feldman, P.L., Clark, P. and Stuehr, D.J. (1994) J. Biol. Chem., 269, 32318–32326. Moali, C., Boucher, J.-L., Sari, M.-A. et al. (1998) Biochemistry, 37, 10453–10460. Ghosh, D.K. and Stuehr, D.J. (1995) Biochemistry, 34, 801–807. Abu-Soud, H.M., Loftus, M. and Stuehr, D.J. (1995) Biochemistry, 34, 11167–11175.

Interaction of NO with Insect Nitrophorins

421

[30] Corbett, J.A., Lancaster, J.R., Jr., M.A. Sweetland, M.A. and McDaniel, M.L. (1991) J. Biol. Chem., 266, 1351–21354. [31] Corbett, J.A., Wang, J.L., Hughes, J.H. et al. (1992) Biochem. J., 287, 229–235. [32] Corbett, J.A., Tilton, R.G., Chang, K. et al. (1992) Diabetes, 41, 552–556. [33] Corbett, J.A., Wang, J.L., Sweetland, M.A. et al. (1992) J. Clin. Invest., 90, 2384–2389. [34] B. Morris, J., Markus, A., Glenn, C.L. et al. (2002) J. Mol. Med., 80, 96–104. [35] Johannesen, L., Pie, A., Pociot, F. et al. (2001) J. Clin. Endocrinol. Metab., 86, 2792–2796. [36] Maritim, A.C., Sanders, R.A. and Watkins III, J.B. (2003) J. Biochem. Molec. Toxicol., 17, 24–38. [37] Brands, M.W., Bell, T.D. and Gibson, B. (2004) Hypertension, 43, 57–63. [38] Awata, T., Neda, T., Iizuka, H. et al. (2004) Diabetes Care, 27, 2184–2190. [39] Abu El-Asrar, A.M., Meersschaert, A., Dralands, L. et al. (2004) Eye, 18, 306–313. [40] Berkowitz, B.A., Luan, H., Gupta, R.R. et al. (2004) Diabetes, 53, 173–178. [41] Cellek, S., Qu, W., Schmidt, A.M. and Moncada, S. (2004) Diabetologia, 47, 331–339. [42] Dandona, P., Aljada, A., Chaudhuri, A. et al. (2005) Circulation, 111, 1448–1454. [43] Storling, J., Binzer, J., Andersson, A.K. et al. (2005) Diabetologia, 48, 2039–2050. [44] Palm, F., Buerk, D.G., Carlsson, P.-O. et al. (2005) Diabetes, 54, 3282–3287. [45] Noronha, B.T., Li, J-M., Wheatcroft, S.B. et al. (2005) Diabetes, 54, 1082–1089. [46] Veluthakal, R., Chvyrkova, I., Tannous, M. et al. (2005) Diabetes, 54, 2576–2585. [47] Liu, Y., Burdon, K.P., Langefeld, C.D. et al. (2005) J. Am. Soc. Nephrol., 16, 1085–1090. [48] Kashyap, S.R., Roman, L.J., Lamont, J. et al. (2005) J. Clin. Endocrinol. Metab., 90, 1100–1105. [49] Green, D.J., Maiorana, A.J., Tschakovsky, M.E. et al. (2006) Am. J. Physiol. Heart Circ. Physiol., 291, H1193–H1191. [50] Park, J.W., Park, S.J., Park, S.H. et al. (2006) Neurobiol. Dis., 21, 43–49. [51] Khamaisi, M., Keynan, S., Bursztyn, M. et al. (2006) Nephron Physiol., 102, 72–80. [52] Wang, X., Desai, K., Juurlink, B.H.J. et al. (2006) Kidney Internat., 69, 281–287. [53] Lin, R.F., Lin, T.-S., Tilton, R.G. and Cross, A.H. (1993) J. Exp. Med., 178, 643–648. [54] Calatayud, S., Barrachina, D. and Esplugues, J.V. (2001) Microsc. Res. Tech., 53, 325–335. [55] Abramson, S.B., Amin, A.R., Clancy, R.M. and M. Attur (2005) Best Pract. Res. Clin. Rheumatol., 15, 831–845. [56] Dawson, V.L. and Dawson, T.M. (1998) Prog. Brain Res., 118, 215–229. [57] Heneka, M.T. and Feinstein, D.L. (2001) J. Neuroimmunol., 114, 8–18. [58] Kondo, S., Toyokuni, S., Tsuruyama, T. et al. (2002) Cancer Lett., 179, 87–93. [59] Kagoura, M., Matsui, C., Toyoda, M. and Morohashi, M. (2001) J. Cutan. Pathol., 28, 476–481. [60] Wilson, K.T. (2002) Dis. Esophagus., 15, 16–21. [61] Franchi, A., Gallo, O., Paglierani, M. et al. (2002) Head Neck, 24, 16–23. [62] Takimoto, E., Champion, H.C., Li, M. et al. (2005) J. Clin. Invest., 115, 1221–1231. [63] Finkel, M.S., Oddis, C.V., Jacob, T.D. et al. (1992) Science, 257, 387–389. [64] Geng, Y.-J., Petersson, A.-S., Wennmalm, A. and Hansson, G.K. (1994) Exper. Cell Res., 214, 418–428. [65] Henry, Y., Durocq, C., Servent, D. et al. (1990) J Eur. Biophys., 20, 1–15. [66] Karupiah, G., Xie, Q., Buller, R.M.L. et al. (1993) Science, 261, 1445–1448. [67] Furchgott, R.F. and Zawadzki, J.V. (1980) Nature, 288, 373–376. [68] Mollace, V., Salvemini, D., Anggard, E. and Vane, J. (1992) Br. J. Pharmacol., 104, 633–638; Yuen, P.S.T. and Garbers, D.L. (1992) Ann. Rev. Neurosci., 15, 193–225. [69] Traylor, T.G. and Sharma, V.S. (1992) Biochemistry, 31, 2847–2849. [70] Tsai, A.-L. (1994) FEBS Lett., 341, 141–145. [71] Yu, A.E., Hu, S., Spiro, T.G. and Burstyn, J.N. (1994) J. Am. Chem. Soc., 116, 4117–4118. [72] Dierks, E.A., Hu, S., Vogel, K.M. et al. (1997) J. Am. Chem. Soc., 119, 7316–7323.

422

F. Ann Walker

[73] Zhao, Y., Hoganson, C., Babcock, G.T. and Marletta, M.A. (1998) Biochemistry, 37, 12458–12464. [74] Ignarro, L.J. (1992) J. NIH Res., 4, 59–62. [75] Burnett, A.L., Lowenstein, C.J., Bredt, D.S. et al. (1992) Science, 257, 401–403. [76] Montague, P.R., Gancayco, C.D., Winn, M.J. et al. (1994) Science, 263, 973–977. [77] Ding, J.M., Chen, D., Weber, E.T. et al. (1994) Science, 266, 1713–1717. [78] Oren, D.A. and Terman, M. (1998) Science, 279, 333–334. [79] Campbell, S.S. and Murphy, P.J. (1998) Science, 279, 396–399. [80] Kim, Y.-M., Bergonia, H.A., Müller, C. et al. (1995) J. Biol. Chem., 270, 5710–5713. [81] Jankiewicz, L., Kwaëny, M., Wasylik, K. and Graczyk, A. (1994) J. Food Sci., 59, 57–59. [82] Bondoc, L.L. and Timkovich, R. (1989) J. Biol. Chem., 264, 6134–6145. [83] Ribeiro, J.M.C. (1987) Ann. Rev. Entomol., 32, 463–478. [84] Law, J., Ribeiro, J.M.C. and Wells, M. (1992) Ann. Rev. Biochem., 61, 87–112. [85] Ribeiro, J.M.C. and Garcia, E.S. (1981) J. Exp. Biol., 94, 219–230. [86] Ribeiro, J.M.C., Rossignol, P.A. and Spielman, A. (1984) J. Exp. Biol., 108, 1–7. [87] Ribeiro, J.M.C., Nussenzveig, R.H. and Tortorella, G. (1994) J. Med. Entomol., 31, 747–752. [88] Ribeiro, J.M.C., Gonzales, R. and Marinotti, O. (1990) Br. J. Pharmacol., 101, 932–936. [89] Ribeiro, J.M.C., Hazzard, J.M.H., Nussenzveig, R.H. et al. (1993) Science, 260, 539–541. [90] Moncada, S. and Martin, J.F. (1993) Lancet, 341, 1511. [91] Champagne, D.E., Nussenzveig, R. and Ribeiro, J.M.C. (1995) J. Biol. Chem., 270, 8691–8695. [92] Andersen, J.F., Ding, X.D., Balfour, C. et al. (2000) Biochemistry, 39, 10118–10131. [93] Feltham, R.D. and Enemark, J.H. (1974) Coord. Chem. Rev., 13, 339–406. [94] Ribeiro, J.M.C. and Walker, F.A. (1994) J. Exp. Med., 180, 2251–2257. [95] Weichsel, A., Andersen, J.F., Champagne, D.E. et al. (1998) Nature Struct. Biol., 5, 304–309. [96] Andersen, J.F., Champagne, D.E., Weichsel, A. et al. (1997) Biochemistry, 36, 4423–4428. [97] Ribeiro, J.M.C., Schneider, M. and Guimaraes, J.A. (1995) Biochem. J., 308, 243–249. [98] Sun, J., Yamaguchi, M., Yuda, M. et al. (1996) Thrombosis & Hemostasis, 75, 573–577. [99] Walker, F.A. and Montfort, W.R. (2001) In Chapter 5. Advances in Inorganic Chemistry, Vol. 51 (G. Mauk and A.G. Sykes, eds) San Diego: Academic Press, pp. 295–358. [100] Neva, F.A. (1988) In Cecil Textbook of Medicine, 18th ed., sect. 383 (J.B. Wyngaarden and L.H. Smith, eds) Philadelphia: Saunders, Hartcourt Brace Jovanovich, Inc., pp. 1865–1869. [101] Kirchhoff, L.V. (1991) In Harrison’s Principles of Internal Medicine, 12th ed. (J.D. Wilson, E. Braunwald, K.J. Isselbacher, R.G. Petersdorf, J.B. Martin, A.S. Fauci and R.K. Root, eds) New York: McGraw-Hill, pp. 791–793. [102] Kirchhoff, L.V. (1993) N. Engl. J. Med., 329, 639. [103] http://www.cdc.gov/ncidod/dpd/parasites/chagasdisease/default.htm [104] McNeil, D.G., Jr. (2003) Rare infection threatens to spread in blood supply. New York Times, 18 November. [105] Almeida, I.C., Camargo, M.M., Procópio, D.O. et al. (2000) EMBO J., 19, 1476–1485. [106] Pereira, C.A., Alonso, G.D., Paveto, M.C. et al. (2000) J. Biol. Chem., 275, 1495–1501. [107] Judice, W.A.S., Cezari, M.H.S., Lima, A.P.C.A. et al. (2001) Eur. J. Biochem., 268, 6578– 6586. [108] Navarro, M., Cisneros-Fajardo, E.J., Lehmann, T. et al. (2001) Inorg. Chem., 40, 6879–6884. [109] Reyes-Vivas, H., Hernández-Alcantura, G., López-Velazquez, G. et al. (2001) Biochemistry, 40, 3134–3140. [110] Villalta, F., Smith, C.M., Ruiz-Ruano, A. and Lima, M.F. (2001) FEBS Lett., 505, 383–388. [111] Rigden, D.J., Monteiro, A.C.S. and Gossi de Sá, M.F. (2001) FEBS Lett., 504, 41–44. [112] Pereira, C.A., Alonso, G.D., Ivaldi, S. et al. (2002) FEBS Lett., 526, 111–114. [113] Pavão, F., Castilho, M.S., Pupo, M.T. et al. (2002) FEBS Lett., 520, 15–17. [114] Thomson, L., Denicola, A. and Radi, R. (2003) Arch. Biochem. Biophys., 412, 55–64.

Interaction of NO with Insect Nitrophorins [115] [116] [117] [118] [119] [120] [121] [122] [123] [124] [125] [126] [127] [128] [129] [130] [131] [132] [133] [134] [135] [136] [137] [138] [139] [140] [141] [142] [143] [144] [145] [146] [147] [148] [149] [150] [151] [152] [153] [154] [155]

423

Otero, L., Noblia, P., Gambino, D. et al. (2003) Inorg. Chim. Acta., 344, 85–94. Watts, A.G., Damager, I., Amaya, M.L. et al. (2003) J. Am. Chem. Soc., 125, 7532–7533. Chamond, N., Gregoire, C., Coatnoan, N. et al. (2003) J. Biol. Chem., 278, 15484–15494. Marroquin-Quelopana, M., Oyama, S., Jr., Pertinhez, A. et al. (2004) Biochem. Biophys. Res. Commun., 325, 612–618. Pinto, A.P.A., Campana, P.T., Beltramini, L.M. et al. (2003) Biochim. Biophys. Acta., 1652, 107–114. Oza, S.L., Tetaud, E., Ariyanayagam, M.R. et al. (2000) J. Biol. Chem., 277, 35853–35861. Wilkinson, S.R., Temperton, N.J., Mondragon, A. and Kelly, J.M. (2000) J. Biol. Chem., 278, 8220–8225. Wilkinson, S.R., Meyer, D.J., Taylor, M.C. et al. (2002) J. Biol. Chem., 277, 17062–17071. Nozaki, T., Shigeta, Y., Saito-Nakano, Y. et al. (2001) J. Biol. Chem., 276, 6516–6523. Vaena de Avalos, S., Blader, I.J., Fisher, M. et al. (2002) J. Biol. Chem., 277, 639–644. Morenos, B., Urbina, J.A., Oldfield, E. et al. (2000) J. Biol. Chem., 275, 28356–28362. Vanhamme, L., Paturiaux-Hanocq, F., Poelvoorde, P., Nolan, D.P. et al. (2003) Nature, 422, 83–87. Sarkar, M., Hamilton, C.J. and Fairlamb, A.H. (2003) J. Biol. Chem., 278, 22703–22708. Leon, J.S., Wang, K. and Engman, D.M. (2003) J. Immunol., 171, 4271–4277. Leon, J.S., Wang, K. and Engman, D.M. (2003) Circulation, 107, 2264–2269. Leon, J.S. and Engman, D.M. (2003) Front Biosci., 8, 315–322. Perez-Morga, D., Vanhollebeke, B., Paturiaux-Hancocq, F. et al. (2005) Science, 309, 469–472. http://en.wikipedia.org/wiki/Triatoma_infestans http://www.arose.net/triatoma/ Moreira, M.F., Coelho, H.S.L., Zingali, R.B. et al. (2003) Insect Biochem. Mol. Biol., 33, 23–28. Andersen, J.F., Gudderra, N.P., Francischetti, I. M.B. et al. (2004) Biochemistry, 43, 6987– 6994. Ribeiro, J.M.C., personal communication. Antoine, M., Erbil, C., Munch, E. and S. Schnell (1987) Niessing, J., Gene, 56, 41–51. Garlick, R.L. and Riggs, A.F. (1982) J. Biol. Chem., 275, 9005–9015. Suzuki, T. and Gotoh, T. (1986) J. Biol. Chem., 261, 9257–9267. Imamura, T., Baldwin, T.O. and Riggs, A.F. (1972) J. Biol. Chem., 247, 2785–2797. Sherman, D.R., Kloek, A.P., Krishnan, B.R. et al. (1992) Proc. Natl. Acad. Sci. USA, 89, 11696–11700. Runnegar, B. (1984) J. Mol. Evol., 2, 33–41. Andersen, J.F., Weichsel, A., Balfour, C. et al. (1998) Structure, 6, 1315–1327. Weichsel, A., Andersen, J.F., Roberts, S.A. and Montfort, W.R. (2000) Nature Struct. Biol., 7, 551–554. Roberts, S.A., Weichsel, A., Qiu, Y. et al. (2001) Biochemistry, 40, 11327–11337. Andersen, J.F. and Montfort, W.R. (2000) J. Biol. Chem., 275, 30496–30503. Weichsel, A., Berry, R.E., Zhang, H. et al., to be submitted. Flower, D.R. (1996) Biochem. J., 318, 1–14. Flower, D.R. (2000) Biochim. Biophys. Acta., 1482, 44–56. Marie, A.D., Veggerby, C., Robertson, D.H.L. et al. (2001) Protein Science, 10, 411–417. Spinelli, S., Vincent, F., Pelosi, P. et al. (2002) Eur. J. Biochem., 269, 2449–2456. Lazar, J., Greenwood, D.R., Rasmussen, L.E.L. and Prestwich, G.D. (2002) Biochemistry, 41, 11786–11794. Newcomer, M.E., Jones, T.A., Aqvist, J. et al. (1984) EMBO J., 3, 1451–1454. Newcomer, M.E., Ong, D.E. (2000) Biochim. Biophys. Acta., 1482, 57–64. Nagata, A., Suzuki, Y., Igarashi, M. et al. (1991) Proc. Natl. Acad. Sci. USA, 88, 4020–4024.

424

F. Ann Walker

[156] Newcomer, M.E. (1993) Structure, 1, 7–18. [157] Riley, C.T., Barbeau, B.K., Keim, P.S. et al. (1984) J. Biol. Chem., 259, 13159–13165. [158] Holden, H.M., Rypniewski, W.R., Law, J.H. and Rayment, I. (1987) EMBO J., 6, 1565–1570. [159] Huber, R., Schneider, M., Epp, O. et al. (1987) J. Mol. Biol., 195, 423–434. [160] Huber, R., Schneider, M., Mayr, I. et al. (1987) J. Mol. Biol., 198, 499–513. [161] Valenzuela, J.G., Walker, F.A. and Ribeiro, J. M.C. (1995) J. Exper. Med., 198, 1519–1526. [162] Paesen, G.C., Adams, P.L., Harlos, K. et al. (1999) Molec. Cell., 3, 661–671. [163] Paesen, G.C., Adams, P.L., Nuttall, P.A. and Stuart, D.L. (2000) Biochim. Biophys. Acta., 1482, 92–101. [164] Francischetti, I.M.B., Andersen, J.F. and Ribeiro, J.M.C. (2002) Biochemistry, 41, 3810–3818. [165] Goetz, D.H., Holmes, M.A., Borregaard, N. et al. (2002) Molec. Cell, 10, 1033–1043. [166] Urade, Y., Nagata, A., Suzuki, Y. et al. (1989) J. Biol. Chem., 264, 1041–1045. [167] Ortlund, E., Parker, C.L., Schreck, S.F. et al. (2002) Biochemistry, 41, 7030–7037. [168] Brownlow, S., Morais Cabral, J.H., Cooper, R. et al. (1997) Structure, 5, 481–495. [169] Uhrinová, S., Uhrín, D., Denton, H. et al. (1998) J. Biomol. NMR., 12, 89–107. [170] Berggård, T., Cohen, A., Persson, P. et al. (1999) Protein Sci., 8, 2611–2620. [171] Åkerström, B., Lögdberg, L., Berggård, T. et al. (2000) Biochim. Biophys. Acta., 1482, 172–184. [172] Alhorn, M., Berggård, T., Nordberg, J. et al. (2002) Blood, 99, 1894–1901. [173] Åkerström, B., Flower, D.R. and J.-P. Salier (2000) Biochim. Biophys. Acta., 1482, 1–8. [174] Flower, D.R., North, A.C.T. and Sansom, C.E. (2000) Biochim. Biophys. Acta., 1482, 9–24. [175] Gutiérrez, G., Ganfornina, M.D. and Sánchez, D. (2000) Biochim. Biophys. Acta., 1482, 35–45. [176] Montfort, W.R., Weichsel, A. and Andersen, J.F. (2000) Biochim. Biophys. Acta., 1482, 110–118. [177] Ding, X.D., Weichsel, A., Andersen, J.F. et al. (1999) J. Am. Chem. Soc., 121, 128–138. [178] Maes, E.M., Weichsel, A., Andersen, J.F. et al. (2004) Biochemistry, 43, 6679–6690. [179] Scheidt, W.R., Duval, H.F., Neal, T.J. and Ellison, M.K. (2000) J. Am. Chem. Soc., 122, 4651–4659. [180] Richter-Addo G.B, Wheeler, R.A., Hixson, C.A. et al. (2001) J. Am. Chem. Soc., 123, 6314–6326. [181] Wyllie, G.R.A. and Scheidt, W.R. (2002) Chem. Rev., 102, 1067–1089. [182] Ghosh, A. (2005) Acc. Chem. Res., 38, 943–954, and references therein. [183] Andersen, J.F. and Ribeiro, J.M.C. (2006) Biochemistry, 45, 5450–5457. [184] Tsujishita, Y., Guo, S., Stolz, L.E. et al. (2001) Cell, 105, 379–389. [185] Weichsel, A., Maes, E.M., Andersen, J.F. et al. (2005) Proc. Natl. Acad. Sci. USA., 102, 594–599. [186] Knipp, M., Zhang, H., Berry, R.E. et al. (2007) Prot. Expr. Purif., 54, 183–191. [187] Shokhireva, T.Kh., Berry, R.E., Filippov, I. et al. (2007) Biochemistry, 46, 6830–6843 (in press). [188] Knipp, M. and Walker, F.A., to be submitted. [189] Shokhireva, T.Kh., Shokhirev, N.V. and Walker, F.A. (2003) Biochemistry, 42, 679–693. [190] Shokhireva, T.Kh., Smith, K.M., Berry, R.E. et al. (2007) Inorg. Chem., 46, 170–178. [191] Shokhireva, T.Kh., Berry, R.E., Zhang, H. et al. (2007) Inorg. Chim. Acta 360, in press. [192] Shokhireva, T.Kh., Weichsel, A., Smith, K.M. et al. (2007) Inorg. Chem., 46, 2041–2056. [193] Shokhirev, N.V. and Walker, F.A. (1998) J. Biol. Inorg. Chem., 3, 581–594. [194] Shokhireva, T.Kh., Berry, R.E., Uno, E. et al. (2003) Proc. Natl. Acad. Sci. USA, 100, 3778–3783. [195] Berry, R.E., Shokhireva, T.Kh., Shokhirev, M.N. et al., to be submitted.

Interaction of NO with Insect Nitrophorins [196] [197] [198] [199] [200] [201] [202] [203] [204] [205] [206] [207] [208] [209] [210] [211] [212] [213] [214] [215] [216] [217] [218] [219] [220] [221] [222] [223] [224] [225] [226] [227] [228] [229] [230] [231] [232] [233] [234] [235] [236]

425

Filippov, I. and Walker, F.A., to be submitted. Wishart, D.S., Bigam, C.G., Holm, A. et al. (1995) J. Biomol. NMR., 5, 67–81. Coles, M., Diercks, T., Muehlenweg, B. et al. (1999) J. Mol. Biol., 289, 139–157. Schanda, P., Kupce, E. and Brutscher, B. (2005) J. Biomol. NMR, 33, 199–211. Brunger, A.T., Adams, P.D., Clore, G.M. et al. (1998) Acta Cryst., D54, 905–921. Bermel, W., Bertini, I., Felli, I.C. et al. (2003) J. Am. Chem. Soc., 125, 16423–16429. Bertini, I., Felli, I.C., Kuemmerle, R. et al. (2004) J. Am. Chem. Soc., 126, 464–465. Rodriguez, J.C., Wilks, A. and Rivera, M. (2006) Biochemistry, 45, 4578–4592. Ribeiro, J.M.C. and Nussenzveig, R.H. (1993) FEBS Lett., 330, 165–168. Yuda, M., Hirai, M., Miura, K. et al. (1996) Eur. J. Biochem., 242, 807–812. Nussenzveig, R.H., Bentley, D.L. and Ribeiro, J.M.C. (1995) J. Exp. Biol., 198, 1093. Luckhart, S. and Rosenberg, R. (1999) Gene, 232, 25–34. Valenzuela, J.G. and Ribeiro, J.M.C. (1998) J. Exper. Biol., 201, 2659–2664. Shokhireva, T.Kh. and Walker, F.A., unpublished results. Mol, C.D., Kuo, C.-F., Thayer, M.M. et al. (1995) Nature, 374, 381–386. Ross, T.S., Jefferson, A.B., Mitchell, C.A. and Majerus, P.W. (1991) J. Biol. Chem., 266, 20283–20289. Laxminarayan, K.M., Chan, B.K., Tetaz, T. et al. (1994) J. Biol. Chem., 269, 17305–17310. Peisach, J., Blumberg, W.E., Ogawa, S. et al. (1971) J. Biol. Chem., 246, 3342–3355. Chevion, M., Peisach, J. and Blumberg, W.E. (1977) J. Biol. Chem., 252, 3637–3645. Wayland, B.B. and Olson, L.W. (1974) J. Am. Chem. Soc., 96, 6037–6041. Xu, N., Powell, D.R., Cheng, L. and Richter-Addo, G.B. (2006) Chem. Comm., 2030–2032. Vanin, A.F., Malenkova, I.V. and Serezhenkov, V.A. (1997) Nitric Oxide, 1, 191–203. Couture, M., Adak, S., Stuehr, D.J. and Rousseau, D.L. (2001) J. Biol. Chem., 276, 38280–38288. Filippov, I., Berry, R.E., Zhang, H. and Walker, F.A., unpublished work. Filippov, I., Ho, A.Y.W., Berry, R.E. et al., F.A., to be submitted. Gibson, Q.H. and Roughton, F.J.W. (1957) J. Physiol. (London), 136, 507–526. Antonini, E., Brunori, M., Wyman, J. and Noble, R.W. (1966) J. Biol. Chem., 241, 3236–3238. Cassoly, R. and Gibson, Q.H. (1975) J. Mol. Biol., 91, 301–313. Lim, M.D., Lorcovic, I.M., Ford, P.C. (2004) J . Inorg. Biochem., 99, 151–165. Sharma, V.S., Traylor, T.G., Gardiner, R. and Mizukami, H. (1987) Biochemistry, 26, 3837–3843. Hoshino, M., Ozawa, K., Seki, H. and Ford, P.C. (1993) J. Am. Chem. Soc., 115, 9568–9575. Kharitonov, V.G., Sharma, V.S., Magde, D. and Koesling, D. (1997) Biochemistry, 36, 6814–6818. Zhao, Y., Brandish, P.E., Ballou, D.P. and Marletta, M.A. (1999) Proc. Natl. Acad. Sci. USA, 96, 14753–14758. Ballou, D.P., Zhao, Y., Brandish, P.E. and Marletta, M.A. (2002) Proc. Natl. Acad. Sci. USA, 99, 12097–101. Lawson, D.M., Stevenson, C.E.M., Andrew, C.R. and Eady, R.R. (2000) EMBO J., 19, 5661–5671. George, S.J., Andrew, C.R., Lawson, D.M. et al. (2001) J. Am. Chem. Soc., 123, 9683–9684. Andrew, C.R., George, S.J., Lawson, D.M. and Eady, R.R. (2002) Biochemistry, 41, 2353–2360. Mayburd, A.L. and Kassner, R.J. (2002) Biochemistry, 41, 11582–11591. Yoshimura, T., Iwasaki, H., Shidara, S. et al. (1988) J. Biochem., 103, 1016–1019. Yoshimura, T., Shidara, S., Ozaki, T. and Kamada, H. (1993) Arch. Microbiol., 160, 498–500. Boon, E.M., Davis, J.H., Tran, R. et al. (2006) J. Biol. Chem., 281, 21892–21902.

426 [237] [238] [239] [240] [241] [242] [243] [244] [245] [246] [247] [248] [249] [250] [251] [252] [253] [254] [255] [256] [257] [258] [259] [260] [261] [262] [263] [264] [265] [266] [267] [268] [269] [270] [271] [272] [273] [274] [275] [276] [277] [278] [279] [280] [281]

F. Ann Walker Boon, E.M. and Marletta, M.A. (2005) J. Inorg. Biochem., 99, 892–902. Boon, E.M., Huang, S.H. and Marletta, M.A. (2005) Nature Chem. Biol., 1, 53–59. Boon, E.M. and Marletta, M.A. (2006) J. Am. Chem. Soc., 128, 10222–10023. Dawson, J.H., Andersson, L.A. and Sono, M. (1983) J. Biol. Chem., 258, 13637–13645. Sono, M., Eble, K.S., Dawson, J.H. and Hager, L.P. (1985) J. Biol. Chem., 260, 15530–15535. Maxwell, J.C. and Caughey, W.S. (1976) Biochemistry, 15, 388–396. Yoshimura, T., (1983) Bull. Chem. Soc. Jpn., 56, 2527–2528. Sampath, V., Zhao, X.-J. and Caughey, W.S. (1994) Biochem. Biophys. Res. Comm., 198, 281–286. Miller, L.M., Pedraza, A.J. and Chance, M.R. (1997) Biochemistry, 36, 12199–12207. Wang, Y. and Averill, B.A. (1996) J. Am. Chem. Soc., 118, 3972–3973. Chottard, G. and Mansuy, D. (1977) Biochem. Biophys. Res. Commun., 77, 1333–1338. Tsubaki, M. and Yu, N.-T. (1982) Biochemistry, 21, 1140–1144. Walters, M. and Spiro, T.G. (1982) Biochemistry, 21, 6989–6995. Benko, B. and Yu, N.-T. (1983) Proc. Natl. Acad. Sci. USA, 80, 7042–7046. Lipscomb, L.A., Lee, B.-S. and Yu, N.-T. (1983) Inorg. Chem., 32, 281–286. Hu, S. and Kincaid, J.R. (1993) J. Am. Chem. Soc., 113, 2843–2850. Lang, G. and Marshall, W. (1966) Proc. Phys. Soc., 87, 3–34. Oosterhuis, W.T. and Lang, G. (1969) J. Chem. Phys., 50, 4381–4387. Liu, M.-C., Huynh, B.-H., Payne, W.J. et al. (1987) Eur. J. Biochem., 169, 253–258. Christner, J.A., Münck, E. Nanick, P.A. and Siegel, L.M. (1983) J. Biol. Chem., 258, 11147–11156. Pfannes, H.D., Bemski, G., Wajnberg, E. et al. (1994) Hyperfine Int., 91, 797–803. Hori, H., Ikeda-Saito, M., Lang, G. and Yonetani, T. (1990) J. Biol. Chem., 265, 15028–15033. Kon, H. (1968) J. Biol. Chem., 243, 4350–4357. Henry, Y. and Banerjee, R. (1973) J. Mol. Biol., 73, 469–482. LoBrutto, R., Wei, Y.-H., Mascarenhas, R. et al. (1983) J. Biol. Chem., 258, 7437–448. Ebel, R.E., O’Keeffe, D.H. and Peterson, J.A. (1975) FEBS Lett., 55, 198–201. Henry, Y., Lepoivre, M., Drapier, J.-C. et al. (1993) FASEB J., 7, 1124–1134. Szabo, A. and Perutz, M.F. (1976) Biochemistry, 15, 4427–4428. Nagai, K., Hori, H., Morimoto, H. et al. (1979) Biochemistry, 18, 1304–1308. Magliozzo, R.S., McCracken, J. and Peisach, J. (1987) Biochemistry, 26, 7923–7931. Rein, H., Ristau, O. and Scheler, W. (1972) FEBS Lett., 24, 24–26. Yoshimura, T. (1983) J. Inorg. Biochem., 18, 263–277. Brunori, M., Falcioni, G. and Rotilio, G. (1974) Proc. Nat. Acad. Sci. USA, 71, 2470–2472. Karthein, R., Nastainczyk, W. and Ruf, H.H. (1987) Eur. J. Biochem., 166, 173–180. Stevens, T.H. and Chan, S.I. (1981) J. Biol. Chem., 256, 1069–1071. Henry, Y., Ishimura, Y. and Peisach, J. (1976) J. Biol. Chem., 251, 1578–1581. Cheesman, M.R., Thomson, A.J., Greenwood, C. et al. (1990) Nature, 346, 771. Berry, M.J., George, S.J., Thomson, A.J. et al. (1990) Biochem. J., 270, 413–417. Maes, E.M., Walker, F.A., Montfort, W.R. and Czernuszewicz, R.S. (2001) J. Am. Chem. Soc., 123, 11664–11672. Addison, A.W. and Stephanos, J.J. (1986) Biochemistry, 25, 4104–4113. Hoshino, M., Maeda, M., Konishi, R. et al. (1996) J. Am. Chem. Soc., 118, 5702–5707. Jia, L., Bonaventura, C., Bonaventura, J. and Stamler, J.S. (1996) Nature, 380, 221–226. Stamler, J.S., Jia, L., Eu, J.P. et al. (1997) Science, 276, 2034–2037. Luchsinger, B.P., Rich, E.N., Gow, A.J. et al. (2003) Proc. Natl. Acad. Sci. USA., 100, 461–466. Angelo, M., Singel, D.J. and Stamler, J.S. (2006) Proc. Natl. Acad. Sci. USA., 103, 8366–8371.

Interaction of NO with Insect Nitrophorins

427

[282] Fujita, E. and Fajer, J. (1983) J. Am. Chem. Soc., 105, 6743–6745. [283] Olson, L., Schaeper, D., Lancon, D. and Kadish, K.M. (1982) J. Am. Chem. Soc., 104, 2042–2044. [284] Choi, I.-K., Liu, Y., Feng, D. et al. (1991) Inorg. Chem., 30, 1832–1839. [285] Lancon, D. and Kadish, K.M. (1983) J. Am. Chem. Soc., 105, 5610–5617. [286] Ridd, J.H. (1978) Adv. Phys. Org. Chem., 16, 1–49. [287] Williams, D.C., ed. (1988) Nitrosation. New York: Cambridge Univ. Press, pp. 1–214. [288] Wade, R.S. and Castro, C.E. (1990) Chem. Res. Toxicol., 3, 289–291. [289] Mirvish, S.S. (1975) Toxicol. Appl. Pharmacol., 31, 325–351. [290] Miwa, M., Stuehr, D.J., Marletta, M.A. et al. (1987) Carcinogenesis, 8, 955–958. [291] Wink, D.A., Kasprzak, K.S., Maragos, C.M. et al. (1991) Science, 254, 1001–1003. [292] Challis, B.C., Fernandes, M.H.R., Glover, B.R. and Latif, F. (1987) IARC Sci. Publ., 84, 308–314. [293] Tannenbaum, S.R. (1987) IARC Sci. Publ., 84, 292–296. [294] Stamler, J.S., Simon, D.I., Osborne, J.A. et al. (1992) Proc. Nat. Acad. Sci. USA, 89, 444–448. [295] Ignarro, L.J. (1989) Circ. Res., 65, 1–21. [296] Stamler, J.S., Jaraki, O., Osborne, J. et al. (1992) Proc. Natl. Acad. Sci. USA, 89, 7674–7677. [297] Gergel, D. and Cederbaum, A.I. (1996) Biochemistry, 35, 16186–16194. [298] Berry, R.E., Ding, X.D., Weichsel, A. et al. (2004) J. Biol. Inorg. Chem., 9, 135–144. [299] Varadarajan, R., Zewert, T.E., Gray, H.B. and Boxer, S.G. (1989) Science, 243, 69–72. [300] Amoia, A.M. (2006) Ph.D. Dissertation. University of Arizona. [301] Berry, R.E., Shokhirev, M.N., Ho, A.Y.W. et al., submitted for publication. [302] Crofts, A., Hacker, B., Barquera, B. et al. (1992) Biochim. Biophys. Acta., 1101, 162–165. [303] Dwyer, J.J., Gittis, A.G., Karp, D.A. et al. (2000) Biophys. J., 79, 1610–1620. [304] Kao, Y-H., Fitch, C.A., Bhattacharya, S. et al. (2000) Biophys. J., 79, 1637–1654. [305] Lee, K.K., Fitch, C.A., Lecomte, J.T.J. and Garcia-Moreno, E.B. (2002) Biochemistry, 41, 5656–5667. [306] Williams, J.C., Haffa, L.M., McCulley, J.L. et al. (2001) Biochemistry, 40, 15403–15407. [307] Louro, R.O., Catarino, T., Paquete, C.M. and Turner, D.L. (2004) FEBS Lett., 576, 77–80. [308] Weichsel, A. and Montfort, W.R., unpublished results. [309] Shokhireva, T.Kh., Berry, R.E., Zhang, H. et al., to be submitted. [310] Scheidt, W.R. (2000) Chapter 16. In The Porphyrin Handbook I, Vol. 3 (K.M. Kadish, K.M. Smith and R. Guilard, eds) San Diego, CA: Academic Press, pp. 50–112. [311] Pellicena, P., Karow, D.S., Boon, E.M. et al. (2004) Proc. Natl. Acad. Sci. USA, 101, 12854–12859. [312] Karow, D.S., Pan, D., Pellicena, P. et al. (2004) Biochemistry, 43, 10203–10211. [313] Ellison, M.K. and Scheidt, W.R. (1999) J. Am. Chem. Soc., 121, 5210–5219 and Supporting information. [314] Safo, M.K., Gupta, G.P., Walker, F.A. and Scheidt, W.R. (1991) J. Am. Chem. Soc., 113, 5497–5510. [315] Safo, M.K., Walker, F.A., Raitsimring, A.M. et al. (1994) J. Am. Chem. Soc., 116, C7760–C7770. [316] Simonneaux, G., Schünemann, V. Morice, C. et al. (2000) J. Am. Chem. Soc., 122, 4366–4377. [317] Walker, F.A., Nasri, H., Turowska-Tyrk, I. et al. (1996) J. Am. Chem. Soc., 118, 12109–12118. [318] Jameson, G.B. and Ibers, J.A. (1979) Inorg. Chem., 18, 1200–1208. [319] Peng, S.-M. and Ibers, J.A. (1976) J. Am. Chem. Soc., 98, 8032–8036. [320] Kim, K. and Ibers, J.A. (1991) Soc., 113, 6077–6081. [321] Slebodnick, C., Dubal, M.L. and Ibers, J.A. (1996) Inorg. Chem., 35, 3607–3613.

428

F. Ann Walker

[322] Tetreau, C., Lavalette, D., Momenteau, M. et al. (1994) J. Am. Chem. Soc., 116, 11840–11848. [323] Safo, M.K., Nesset, M.J.M., Walker, F.A. et al. (1997) J. Am. Chem. Soc., 119, 9438–9448. [324] Walker, F.A. (2004) In Biomimetic inorganic chemistry. Chem. Rev. (R.H. Holm and Solomon, E.I., eds), 104, 589–615. [325] Walker, F.A. (1999) Coord. Chem. Rev. 185–186; 471–534. [326] Safo, M.K., Gupta, G.P., Watson, C.T. et al. (1992) J. Am. Chem. Soc., 114, 7066–7075. [327] Ogura, H., Yatsunyk, L., Medforth, C.J. et al. (2001) J. Am. Chem. Soc., 123, 6564–6578. [328] Yatsunyk, L.A., Carducci, M.D. and Walker, F.A. (2003) J. Am. Chem. Soc., 125, 15986–16005. [329] Yatsunyk, L.A. and Walker, F.A. (2004) Inorg. Chem., 43, 4341–4352. [330] Nakamura, M., Ikeue, T., Fujii, H. and T. Yoshimura (1997) J. Am. Chem. Soc., 119, 6284–6291. [331] Wolowiec, S., Latos-Grazynski, L., Mazzanti, M. and Marchon, J.-C. (1997) Inorg. Chem., 36, 5761–5771. [332] Wolowiec, S., Latos-Grazynski, L., Toronto, D. and Marchon, J.-C. (1998) Inorg. Chem., 37, 7732. [333] Wojaczynski, J., Latos-Grazynski, L. and Glowiak, T. (1997) Inorg. Chem., 36, 6299–6306. [334] Astashkin, A.V., Raitsimring, A.M. and Walker, F.A. (2001) J. Am. Chem. Soc., 123, 1905–1913. [335] Cai, S., Belikova, E., Yatsunyk, L.A. et al. (2004) Inorg. Chem., 43, 1882–1889. [336] Cai, S., Lichtenberger, D.L. and Walker, F.A. (2004) Inorg. Chem., 43, 1890–1903. [337] Li, D., Stuehr, D.J., Yeh, S.R. and Rousseau, D.L. (2004) J. Biol. Chem., 279, 26489–26499. [338] Rousseau, D.L., Li, D., Couture, M. and Yeh, S-R. (2005). J. Inorg. Biochem., 99, 306–323. [339] Yatsunyk, L.A. and Walker, F.A. (2004) Inorg. Chem., 43, 757–777. [340] Schünemann, V., Benda, R., Trautwein, A.X. and Walker, F.A. (2000) Isr. J. Chem., 40, 9–14. [341] Wegner, P., Benda, R., Schünemann, V. et al., to be submitted. [342] Westcott, B.L. and Enemark, J.H. (1999) In Inorganic Electronic Structure and Spectroscopy, Volume II: Applications and Case Studies (E.I. Solomon and A.B.P. Lever, eds) New York: John Wiley & Sons, pp. 403–450. [343] Shokhireva, T.Kh. and Walker, F.A., unpublished work. [344] Astashkin, A.V., Raitsimring, A.M. and Walker, F.A. (1999) Chem. Phys. Lett., 306, 9–17. [345] Shokhireva, T.Kh., Weichsel, A., Montfort, W.R. and Walker, F.A., manuscript in preparation. [346] Kadish, K.M., D’Souza, F., Villard, A. et al. (1994) Inorg. Chem., 33, 5169–5170. [347] Tagliatesta, P., Li, J., Autret, M. et al. (1996) Inorg. Chem., 35, 5570–5576. [348] Duval, H., Bulach, V., Fischer, J. et al. (1997) J. Biol. Inorg. Chem., 2, 662–666. [349] Kadish, K.M., Li, J., Van Caemelbecke, E. et al. (1997) Inorg. Chem., 36, 6292–6298. [350] Hall, R.S., Shokhirev, M.N., Yatsunyk, L.A. et al., to be submitted for publication. [351] Grodzicki, M., Flint, H., Winkler, H. et al. (1997) J. Phys. Chem. A, 101, 4202–4207 [352] Polam, J.R., Shokhireva, T.Kh., Raffii, K. et al. (1997) Inorg. Chim. Acta., 263, 109–117. [353] Nessett, M.J.M., Shokhirev, N.V., Enemark, P.D. et al. (1996) Inorg. Chem., 35, 5188–5200. [354] Ema, T., Senge, M.O., Nelson, N.Y. et al. (1994) Chem. Int. Ed. Engl., 33, 1879. [355] Jentzen, W., Simpson, M.C., Hobbs, J.D. et al. (1995) J. Am. Chem. Soc., 117, 11085–11097.

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 16

Bioinorganic Chemistry of the HNO Ligand Filip Sulc and Patrick J. Farmer Department of Chemistry, University of California, Irvine, 92697–2025

Abstract The chemistry of the one-electron reduced form of nitric oxide, termed a nitroxyl or nitrosyl hydride (NO− or HNO), is described with special focus on its coordination chemistry.∗ Many fundamental characteristics of HNO as a molecule or ligand are not yet fully understood, and widely cited values of its oxidation potential and pKa have recently been significantly revised. The physiological effects of HNO are dominated by competitive reactivity with heme proteins and cysteine-containing proteins such as glutathione. Nitroxyl intermediates have long been proposed in the catalytic cycles of several heme-based enzymes integral to the biological nitrogen cycle; but only in fungal cytochrome P450nor has a shortlived nitroxyl-adduct been spectroscopically characterized. Small molecule HNO complexes of transition metals are rare, and the several reported species are presented with descriptions of their synthesis, and a comparison of available spectroscopic data. Ferrous-nitroxyl adducts were first identified in electrochemical reductions of nitrosyl porphyrins and heme proteins, and recently have these species been characterized in solution. Emphasis is given to the HNO adduct of myoglobin, including its synthesis by various routes and characterization by 1 H NMR, resonance Raman and X-ray absorption spectroscopy. HNO is isoelectronic with 1 O2 , and as with oxymyoglobin, there are several possible descriptions for its bonding with a ferrous heme; an analogy to the -bonding interactions of a Fischer carbene is presented. A survey of the reactivity associated with the characterizable HNO complexes is made, including redox and protonation equilibrium, reactivity with small molecules and dissociation or displacement reactions.

ABBREVIATIONS AS COD Cy Das ddab deoxyMb

∗

Angeli’s salt cycloocta-1,5-diene cyclohexyl o-phenylenebis(diniethylarsine) dimethyldidodecylammonium bromide Mb FeII

This chapter represents an update and expansion of a previous review: Farmer, P.J. and Sulc, F. (2005) J. Inorg. Biochem., 99, 166–184.

430

DMOEiBC EPR Irrev 1-MeIm metMb MLCT MSHA NOESY NHE OEP Ph i Pr ‘pybu S4 ’ PzH SCE THF TOCSY TPP TPPS ttp XAFS/exafs XANES XYL

F. Sulc and P.J. Farmer

dimethyloctaethylisobacteriochlorin Electron paramagnetic resonance irreversible 1-methylimidazole Mb FeIII metal ligand charge transfer methylsulfonylhydroxylamine acid nuclear overhauser effect spectroscopy normal hydrogen electrode octaethylporphyrin phenyl isopropyl 2,6-Bis(2-mercapto-3,5-di-tertbutylphenylthio)dimethylpyridine(2-) pyrazole saturated calomel electrode Tetrahydrofuran total correlation spectroscopy tetraphenylporphyrin tetra(4-sulfonatophenyl)porphyrin tetratolylporphyrinatodianion X-ray Absorption Fine-Structure X-Ray Absorption Near Edge (Fe K-edge) Structure Phenol, 2,6-bis[[bis[2-(2-pyridinyl)ethyl]amino]methyl]−

1. NO AND HNO The physiological importance of nitric oxide has generated tremendous interest in the chemistry of heme-nitrosyls, and justifiably so, as both the formation and activity of NO is dependent on the heme cofactors in nitric oxide synthase and soluble guanyl cyclase [1]. There has also been considerable interest over the last decade in the possible biological activity of the one-electron reduced form of nitric oxide, termed nitroxyl or nitrosyl hydride (NO− or HNO) [2–4]. Free nitroxyl has been proposed as the immediate biological precursor of NO during enzymatic oxidation of l-arginine [5], and nitroxyl intermediates have long been proposed in denitrification processes in plants, bacteria and fungi which are catalyzed by a variety of metalloenzymes [6]. Several metalloproteins have been suggested to mediate the physiological activity of HNO, for instance, Cu, Zn superoxide dismutase [7–9], and the ferric heme proteins such as the peroxidases and the cytochromes P450 [10]. Despite the obvious importance, there is relatively little substantive knowledge about the bonding of nitroxyl to transition metal ions. Only a few examples of nitroxyl-metal complexes have been characterized, typically air- and water-sensitive second- or thirdrow transition metal complexes. We recently reported the synthesis and characterization of the HNO adduct of myoglobin, Mb HNO, which displays unusual stability in aqueous solution [11,12]. The formation of this nitroxyl adduct in a heme protein provides a unique opportunity to understand the properties and fate of nitroxyl intermediates

Bioinorganic Chemistry of the HNO Ligand

431

in enzymes involved in NO synthesis or conversion, as well as the physiology of nitroxyl-releasing drugs that have shown recent promise. This review will focus on heme-bound nitroxyl adducts in such biological systems and in related Fe-porphyrin models, with a subsequent discussion of the structure, bonding and reactivity of wellcharacterized nitroxyl adducts of other transition metals.

2. PRODUCTION AND DETECTION OF HNO HNO does not accumulate in aqueous solutions due its rapid dimerization producing N2 O between pH 2 and 11 (Eq. 1) [13]. Thus HNO cannot be stored as reactant or as a stable product for detection; it is generated in situ via precursor compounds that decompose to release HNO. By far the most widely used precursor is Angeli’s salt (sodium trioxodinitrate/Na2 N2 O3 ), AS, which decomposes to give HNO at pH 4–8 (Eq. 2) [14,15]. The decomposition of AS also produces stoichiometric NO2 − , also biologically reactive, which is a complicating factor. Another family of precursors are various alkylsulfohydroxamic acids, for example, benzylsulfohydroxamic acid or Piloty’s acid, which generate HNO upon deprotonation at a high pH (Eq. 3) [16]. 2HNO → H2 N2 O2  → N2 Og + H2 O HN2 O− 3



HNO + NO− 2

RSO2 NHO−  RSO− 2 + HNO

pH 4 to 8 pH 8 to 13

(1) (2) (3)

As it is highly reactive, the direct detection of HNO remains difficult, most assays rely on quantification of more stable byproducts or on the action of specific inhibitors that are known to react rapidly with HNO. The most common assay relies on detection by gas chromatography of N2 O in the head gas of a reaction mixture, formed by HNO dimerization via hyponitrous acid (H2 N2 O2 ) (Eq. 1) [17]. 15 N-labeling has been used to confirm that the N2 O detected originates from HNO [17]. Reactivity assays are the alternative detection technique; for example, nitrosobenzene reacts with HNO to give cupferron, which complexes with Cu(II) to yield a colored product that can be detected spectrophotometrically [18]. Likewise, the rapid reaction of HNO with thiols is used as a test of the biological activity specific to HNO: such activity should diminish or disappear in the presence of excess thiol such as glutathione [19].

3. NITROXYL REVISITED Nitroxyl, either as NO- or HNO, has been speculated as an intermediate in many processes but its verification remains difficult [20,21]. The existence of HNO was first postulated by Angeli in the early 1900s as resulting from the decomposition of AS [14]. In the 1950s nitroxyl was characterized in the gas phase during the combustion of nitrogen-containing compounds; the first spectroscopic observations were of HNO generated by a microwave discharge in a mixture of nitric oxide and hydrogen gas [22,23], and by photolysis of methyl nitrite [24,25]. During the same period, HNO was

432

F. Sulc and P.J. Farmer H 1.026 Å ν(NH)str 2685 cm–1

108.5° N

O 1.211 Å ν(NO)bnd 1500 cm–1 ν(NO)str 1565 cm–1

Fig. 1. Structural and vibrational data for gas HNO [26–28].

2NO + e–

E1o – 0.81 Va,b (+0.30 V)d

2

NO + e–

E1o – 1.7 Vb,c

+ H+ 2NO–

106 mol–1sec–1 f 1HNO

pKa 11.6b,c (4.7)e

(109 mol–1sec–1)g

N2O + H2O

1NO–

(– 0.30 V)e

Fig. 2. Re-evaluated reduction potentials, pKa and dimerization constants. Previously reported values given in parenthesis. a [30]; b [29]; c [13]; d [31]; e [32]; f [13]; g [37].

characterized in condensed phase by isolation in an Ar matrix [24]. Some of the relevant physical parameters for HNO obtained from such studies are given in Fig. 1 [26–28]. Recently, much of the conventional wisdom concerning nitroxyl has been dramatically revised, as illustrated in Fig. 2. Theoretical calculations [13,29] as well as physical measurements [30] obtain a low reduction potential ca. −0.8 V (at 1 M vs. NHE) for the reduction of NO to nitroxyl anion (3 NO− ), a substantial revision of the widely cited values of +0.39 V and −0.35 V [2–4,31]. Nitric oxide is much harder to reduce than molecular oxygen, and consequently nitroxyl generation is much less likely in a physiological setting. Likewise, the pKa of 3 NO− has been re-evaluated from a previously quoted value of 4.7 [32] to above 11.5 [13,29]. Thus, nitroxyl exists in aqueous solution almost exclusively in its protonated form, HNO. The neutral nitrosyl hydride has a singlet ground state, 1 HNO while the nitroxyl anion is a ground state triplet, 3 NO− isoelectronic with molecular oxygen. Therefore, upon deprotonation of HNO it must undergo a spin flip to avoid a high energy intermediate (Fig. 3) [33]. The spin state barrier slows the proton exchange considerably, such that the triplet state anion, if generated directly, would have a lifetime of seconds in pH 7 buffer. The slow acid–base kinetics also influences the pH dependant redox reactivity of nitroxyl. At pH 13, decomposition of the nitroxyl donor methylsulfonylhydroxylamine  acid (MSHA) readily reduces a low potential viologen derivative, with E = − 690 mV (NHE). At pH 7, decomposition of the HNO donor Angeli’s salt does not reduce methyl  viologen, with E =−440 mV (NHE) [29]. Interpretation of such behavior is complicated by the rapid dimerization of nitroxyl, which limits the concentration and lifetime of nitroxyl in solution. The dimerization reaction (Eq. 1) has been the topic of several theoretical and experimental publications [34]. Widely conflicting estimates for the rate of this reaction are found in the literature, including several in the gas phase, with the majority of the rate constants reported between 106 and 109 M−1 s−1 [35,36]. Bazylinski and Hollocher estimated the rate for the dimerization/dehydration of nitroxyl by gas

Bioinorganic Chemistry of the HNO Ligand

433

3 {O

3HNO + OH–

…O …H N

2O

3NO– + H

2O

1

1HNO + OH–

{O

N… H… O

H

–

Δ Go

}#

– }#

H

1NO– + H

0

1

Δ d(N–H) (Å)

Fig. 3. Energy diagram for the spin-forbidden deprotonation of HNO.

chromatography/mass spectrometry analysis of the trapping of nitroxyl (from AS) by two 15 NO molecules to make N3 O3 − , which decomposed rapidly into N2 O (1415 (N)2 O and 15 (N)2 O) and nitrite. The minor 14 N2 O product was attributed to the HNO dimerization, and by further calculation a dimerization rate was estimated between 1.8 and 72 × 109 M−1 s−1 [37]. In a more recent determination by Lymar, nitroxyl was generated from AS by UV photolysis in air-equilibrated alkaline solutions, and its reaction with O2 followed. The oxygen-trapped product, peroxynitrite anion (ONOO− ) was monitored by absorption band at 248 nm; an observed decrease in signal at pH values below 13 was interpreted as due to competitive nitroxyl dimerization. The rate obtained by modeling these reaction sequences was decidedly slower than that of Hollocher, at 8 ± 3 × 106 M−1 s−1 [13].

4. PHYSIOLOGICAL EFFECTS OF FREE NITROXYL While pathways for generation of NO and NO+ (as thionitrites) within the body are well known, no direct evidence of HNO generation in vivo has yet been reported. It has been suggested that nitroxyl may be generated in the body by several non-enzymatic pathways, such as the reaction nitrosothiols (RSNO) with thiols (RSH) (Eq. 4) [38]. The lack of in vivo evidence may be due to an inadequate detection or the rapid reactivity of HNO itself, but many doubt it has a fundamental biological role. But that said, there are notable effects of HNO demonstrated in cell and animal model studies [39]. Much of the interest in the chemistry of HNO stems from recent investigations of the use of HNO- releasing compounds as drugs. RSH + R SNO → HNO + RSSR

(4)

RSH + HNO → RSNHOH

(5)

R SH + RSNHOH → H2 NO2 + RSSR

(6)

434

F. Sulc and P.J. Farmer

HNO reacts readily with cysteines, ultimately generating disulfides and hydroxylamine, Eqs 5 and 6, and thus may affect or inhibit numerous enzymes [40]. In early 1990s Nagasawa and coworkers demonstrated the intermediacy of HNO in the inhibition of aldehyde dehydrogenase (an enzyme involved in the metabolism of ethanol to acetate) by cyanamide, an anti-alcoholism drug [41–43]. Oxidation of cyanamide by catalase/H2 O2 generates an N-hydroxy intermediate that further decomposes to yield HNO and cyanide. The nitroxyl was postulated to react with a critical cysteine within the enzymes active site. Wink and coworkers showed that exposure of fibroblasts to HNO, from Angeli’s salt, results in substantial oxidation of intracellular glutathione (GSH), and thus has a great effect on the overall redox state of the cells [44,45]. These results are consistent with competitive kinetic measurements of HNO with other cellular species that imply reaction of HNO with GSH as one of the most likely fates of HNO within a cell [10] Similarly, Fukuto looked at the effect of HNO on a Cu-sensing transcription factor, Ace1, that protects against this metal’s toxicity in yeast. They found that HNO (from AS) selectively reacts with specific Cys sites on Ace1 involved in Cu binding; thus HNO-treated Ace1 remains inactivated and genes responsible for Cu protection are not transcribed [46,47]. Nitroxyl was also shown to alter the activity of the glutaminergic N-methyl-d-aspartate (NMDA) receptors, with the observed effect of exogenous HNO from Angeli’s salt highly dependent on the experimental conditions. Glutamate-based exocitotoxicity, a neuronal damage, is caused by overstimulation of the NMDA receptors by Ca2+ influx. In one study, HNO was shown to protect against the overstimulation by reacting with a critical Cys residue and decreasing the Ca2+ influx [48]. Another study, however, indicated damaging effects of HNO caused by the blocking of a desensitization pathway [49]. The possible use of nitroxyl-generating drugs in the treatment of heart disease remains the most important medical application to date. Nitric oxide has long be known to induce vasorelaxation in the cardiovascular system, and similar effects have been observed for AS. Initial studies showed HNO to be a powerful vasodilator, analogous to NO, but these results were later questioned based on in vivo experiments [50,51]. For example, activation of the enzyme guanylate cyclase (sGC), responsible for vasodilation, by HNO was later shown to be exclusively due to NO co-production during the HNO generation [52,53]. This contention is supported by amplification of vasorelaxation upon addition of copper-zinc superoxide dismutase (CuZnSOD), which is known to oxidize HNO to NO [50,54]. However, Paolocci and coworkers showed that AS showed a distinctive response from NO on the efficiency of the heart. When HNO was administered, an increase in left ventricular contractility was observed, with lower cardiac preload and diastolic pressure, without an increase in arterial resistance [55]. These effects HNO on the cardiovascular system seem to be involved in the cAMP signal transduction pathway, and this research shows great promise in the treatment of heart failure. Nitroxyl also has a distinct effect on ischemia-reperfusion injury, where oxygendepleted tissue (after an ischemic loss of blood flow) is damaged when oxygenated blood is reintroduced, or reperfused, into it. When NO or HNO (from AS) were administered before the ischemia both showed protection against the reperfusion damage, with HNO affording a higher degree of protection. Nitric oxide showed similar protection when administered between the ischemia and the reperfusion, but HNO actually worsened the reperfusion injury under these conditions [56,57].

Bioinorganic Chemistry of the HNO Ligand

435

Recently, Motterlini has shown that NO, NO+ and HNO donors all augment the uptake of hemin into endothelial cells in vascular model systems, resulting in amplified HO-1 expression and production of bilirubin [58]. The effect of HNO (from AS) in HO-1 amplification was almost double that of NO (from NONOates) or NO+ (from RSNO), and the enhanced activity was suggested to be due to the co-production of nitrite (NO2 − ), an NO+ equivalent, during the decomposition of AS. As will be shown, nitrite reacts with deoxy-hemoglobin or myoglobin to generate NO, and has itself considerable physiological effect [59]. Motterlini’s work suggests that NO-based species modulate heme transport, and may play a fundamental role in certain hematological disorders. But this work highlights the importance of the fundamental affinity of hemes for NO and HNO as well as the problem of defining the NOx species responsible for any physiologic effect.

5. THE ENEMAR/FELTHAM BONDING EXTREME As much of this review will focus on the coordination of nitroxyl to transition metals, its bonding modes are best understood by comparison to other diatomic ligands such as carbon monoxide, dioxygen and its oxidized congener, nitric oxide. Nitric oxide is the prototypical “non-innocent” ligand, so-called because in a metal complex it may exist in three different oxidation states (NO+ , NO or NO− ) that are intimately linked to the metal ion’s oxidation state. The nitroxyl adduct of a ferrous heme represents the limiting extreme in the generalized description of the metal-NO bonding proposed by Enemark and Feltham [60,61]. In this description, the redox states of the metal and NO are tightly coupled via the -backbonding interaction between the metal t2g and NO  ∗ orbitals, such that structural parameters of any complex may be predicted by the term {M NO}n , where n is the sum of the metal d-electrons and the nitrosyl  ∗ electrons. Octahedral complexes with n ≤ 6 are described as bound to a nitrosyl cation (NO+ ) with a predicted M NO angle of ∼180 , as the single  ∗ electron of NO will be shifted to the lower-energy metal t2g orbitals upon binding. Complexes with n = 8 will have two electrons occupying -backbonding orbitals and are formally described as d6 metal with a nitroxyl anion (NO− ) and a bond angle of ∼120 is predicted. Complexes with n = 7 are described a nitrosyl (NO) and have bond angles, intermediate of ∼150 (Fig. 4) [62]. These predictions are well supported in a series of well-characterized complexes: MnII (TPP)(NO+ ), {Mn NO}6 nitrosyl cation (NO+ ) with an angle of 176.2 , CoIII (TPP)(NO− ), {Co NO}8 nitroxyl (NO− ) with an angle of 128.5 and FeII (TPP)(NO), {Fe NO}7 nitrosyl (NO) radical with an angle of 142.1 [63,64]. The first is consistent with the cationic nitrosonium adduct, MnII (TPP)(NO+ ), the latter

O N ∼180°

O

•

N ∼150°

O N ∼120°

M

M

M

NO+

NO

NO –

{M–NO}6

{M–NO}7

{M–NO}8

Fig. 4. Enemark and Feltham description of metal-NO bonding.

436

F. Sulc and P.J. Farmer

with that of the nitroxyl complex CoIII (TPP)(NO− ), while the adduct with FeII (TPP) is intermediate and has substantial radical character on the NO ligand. Single-electron oxidation of FeII (TPP)(NO) gives a {Fe NO}6 system, predicted as a linear nitrosonium adduct [65]; likewise, ferrous heme nitroxyl adducts are {FeNO}8 and predicted to be highly bent, ca. 120 .

6. BIOLOGICAL ROLES OF NITROXYL INTERMEDIATES A variety of metalloproteins have been suggested to mediate the physiological activity of HNO, for instance, Cu and Zn superoxide dismutase [7–9,66], and ferric heme proteins, such as the peroxidases and the cytochromes P450 [10]. Free nitroxyl has been proposed as the immediate biological precursor of NO during enzymatic oxidation of l-arginine at the P450 active site of nitric oxide synthase, and it has indeed been identified as a major product during turnover of the enzyme in the absence of biopterin [5]. Likewise, nitroxyl intermediates have long been proposed in biological denitrification processes in plants, bacteria and fungi, which are catalyzed by a variety of heme-containing enzymes (Fig. 5) [6].

Nitric Oxide (NO)

nitrification

multiheme HAO

Hydroxylamine (NH2OH)

(NO–/HNO) siroheme NiR

Nitrite assimilation (NO – ) 2

Ammonia (NH3)

hexaheme

NoR

denitrification

Nitric (heme bc, P450) Nitrous N2OR Dinitrogen Oxide Oxide (N2) (heme cd1) (NO) (N2O) (Cu enzymes)

oxidized N (+3)

NiR

0

1

2

3

4

5

most reduced 6 N (– 3)

number of electrons

Fig. 5. Steps in the nitrogen cycle catalyzed by heme-containing enzymes.

7. NITRITE REDUCTASES Transient nitroxyl intermediates have long been proposed in the catalytic cycles of the heme-based nitrite reductases, NiR. In plants, nitrite is reduced with six electron equivalents to ammonium by an assimilatory NiR which contains a 4Fe–4S/siroheme active site [67]. Nitrite binds to the siroheme, an isobacteriochlorin or doubly reduced heme derivative [68] (Fig. 6), which is tightly coupled to a 4Fe–4S cluster. An identical

Bioinorganic Chemistry of the HNO Ligand

437

S Cys S

Fe

Cys S Fe Fe S

Cys S

Me

S S

Fe S (Cys)

HO2C H

HO2C

CO2H Me N

H

N

CO2H

N

CO2H

Fe N

HO2C

HO2C

HO2C

Fig. 6. The 4Fe–4S siroheme reaction center of spinach nitrite reductase.

active site is found in the siroheme sulfite reductases (SiR) that catalyze the six-electron reduction of SO3 2− to H2 S [69]. The similarity of the enzymes extends to their activity, SiR will catalyze nitrite reduction and NiR will catalyze sulfite reduction, the main difference being a greater affinity (by several orders of magnitude) for each enzyme’s specific substrate [70]. Nitrogen-based substrates bind to the Fe(siroheme), and that the 4Fe–4S cluster serves as a reservoir of reducing equivalents [71]; how the overall six-electron reduction is accomplished remains unclear. The required electrons are delivered from the singleelectron donor ferredoxin, leading to the necessity of long-lived, non-labile intermediates during catalysis. Both the siroheme and the 4Fe–4S cluster are redox active, and the electronic coupling of these two redox active cofactors in both NiR and SiR is well established by EPR and Mossbauer studies [70,72–74]. In a study of the reduction of nitrite by SiR from Dracunculus vulgaris, Cowan and coworkers proposed a mechanism entailing sequential two-electron cleavages of the N O bonds (Fig. 7) [75–77]. One reducing equivalent is envisioned to come from the FeIII/II couple of the Fe-heme itself, and one from the Fe4 S4 2+/1+ cluster (shown as [ ]). Thus the active site is “primed” in the reduced state to perform two-electron reductions of substrates and this formalism is maintained, perhaps unrealistically, in the proposed mechanism (Fig. 7). For example, expected ferric-nitrosyl intermediate (FeIII NO) typically obtained from dehydration of the ferrous nitrite adduct is further reduced by the Fe4 S4 1+ cluster to a ferric-nitroxyl, FeIII NO− , which may more reasonably be formulated as a ferrous-nitrosyl, FeII NO. A second proton-coupled two-electron reduction generates a nitrosyl hydride intermediate, FeII HNO; further protonation of this species engenders an electronic reorganization, yielding a ferric complex of a hydroxylamine anion, formally the product of a two-electron reduction of HNO. Several crystal structures of substrate-bound SiR forms have been obtained, and in one, a nitroxyl intermediate has been tentatively identified [78]. Photo-reduction and subsequent treatment with nitrite within a crystal of SiR yielded an NO adduct with

438

F. Sulc and P.J. Farmer O

OH

O

O

OH–

N

N ••

II

H+ 2 e–

III

Fe

Fe

[e–]

[ ]

H

N II

Fe

[e–] H+

H+, NO2– NH3 H H

H

H

N

••

H+ 2 e–

II

Fe

[e–]

H

N III

II

Fe OH–

HO

H

N

Fe [ ]

H

HO

H

••

N H+ 2 e–

[e–]

III

Fe [ ]

Fig. 7. Proposed reaction mechanism for nitrite reduction by spinach NiR.

an Fe N O angle of 125 , much lower than expected for ferrous-NO adducts [78]. The authors suggest it to be a bound nitroxyl, stabilized by the extensive H-bonding within the active pocket. But in general, the Fe N O angles of heme-protein nitrosyls are quite variable [79], and a poor guide to the oxidation state of the heme. It has been suggested that the relative energy difference for Fe N O angle variation between 130 and 150 is less than 1 kcal/mol [80]. Another ferrous nitroxyl intermediate has been proposed in the catalytic cycle of the cytochrome c nitrite reductases, ccNiR, which serve as terminal electron acceptors in certain anaerobic bacteria [81]. In contrast to the siroheme NiR, ccNiR have five type c hemes per monomer and an unusual lysine-coordinated heme at the active site. Several substrate-bound intermediates, including the nitrite and hydroxylamine adducts, were structurally characterized for the ccNiR from W . succinogenes [82–84]. The proposed mechanism [82] point out an inherent problem associated with generating a ferrous nitroxyl intermediate, FeII NO− , by sequential electron transfers: if transiently formed, the thermostable ferrous nitrosyl FeII NO may likely act as a kinetic trap, inhibiting catalytic turnover. The authors suggest that the FeII NO− state may be attained by two rapid single-electron reductions of the linear FeIII NO, such that both occur before the nuclear rearrangement to the stable bent form of FeII NO. A more reasonable alternative was suggested in a computational study which found that the amino ligation in ccNiR destabilizes the nitric oxide adduct relative to the nitroxyl, making the second reduction more kinetically feasible [85].

8. NITRIC OXIDE REDUCTASES Hollocher first advocated the possibility of a nitroxyl intermediate in the heme bc nitric oxide reductases, NorBC, as a result of N-labeling experiments in denitrifying bacterial cell cultures [86,87]. The bacterial NorBC are integral membrane proteins that catalyze the reduction of nitric oxide to nitrous oxide and water, within a larger denitrification complex containing both nitrite and nitrous oxide reductases. The nitroxyl adduct was

Bioinorganic Chemistry of the HNO Ligand

439

suggested as a branching point in these N-oxide reductases between pathways leading to N2 O by N N coupling and NH3 by further reduction and protonation [88,89]. Recent work has shown NorBC to contain an Fe heme b/non-heme Fe binuclear center, termed NorB [90], which is quite similar, both structurally and genomically, to the Fe Cu center of cytochrome c oxidases [91]. A number of spectroscopic experiments suggest that the di-ferrous form is required for NO reduction [90,92,93]. But, the reactivity of a model binuclear heme/non-heme Fe complex suggests that a simple coupling of FeII NO species at the two active site metals can yield N2 O and an FeIII O FeIII bridging species [94], and a similar bridging oxo species has been observed in the inactive enzyme [95]. More conclusive evidence for a nitroxyl-heme intermediate is found for fungal cytochrome P450nor, an unusual enzyme which does not exhibit the usual monooxygenation activity of other P450 enzymes. In the proposed catalytic cycle, the ferricnitrosyl complex of P450nor is reduced by NADH to generate an identifiable intermediate ( max at 444 nm) that reacts rapidly with nitric oxide to give nitrous oxide (Fig. 8) [96–98]. Resonance Raman of isotopically labeled samples identified the Fe NO stretch for the intermediate at 596 cm−1 , significantly higher than that for the FeIII NO at 550 cm−1 . It was reasoned that the addition of two electrons to  ∗ orbitals of FeIII NO increases metal-ligand back-donation, thus strengthening the Fe NO bond and shifting the Fe NO to a higher frequency. It was also demonstrated that the intermediate does not undergo deuterium exchange with D2 O, which was used as evidence against an FeII HNO adduct (vide supra). Subsequent kinetic and spectroscopic experiments show that the 444 nm intermediate may be formed from reaction of NaBH4 with the ferric nitrosyl adduct of P450nor at −10 C, and also by reaction of the ferric P450nor with hydroxylamine radicals generated by pulse radiolysis oxidation of hydroxylamine [99]. Thus the intermediate was formulated as a ferric hydroxylamine radical adduct, FeIII NHOH, which would then react with NO directly to generate N2 O and the FeIII OH2 resting state. The sequential mechanism proposed, as adapted in Fig. 8, includes rather unusual formulations of Fe-based oxidation states and its coupling with the proximal cysteine ligand. The direct

H

III

S– Fe

O H

•N

O

IV

S– Fe

– H2O

N

O

+

N

H

H

H S– Fe

+ + N

–NAD+ +NADH +H+

– N2O

III

II

S– Fe

O

OH

N O

–

•N

IV

S Fe

O

N

O

e–

Fig. 8. Proposed reaction mechanism for P450nor.

S

III

Fe

N •

OH

440

F. Sulc and P.J. Farmer

addition of a hydride to the FeIII NO avoids the thermodynamically stable FeII NO and its potentially unfavorable reduction. The authors failed to note that the proposed FeIII NHOH intermediate is inconsistent with the lack of deuterium exchange. But the formation of long-lived nitroxyl adducts and their coupling reactions with free NO have been recently confirmed in both heme protein and small molecule models.

9. ELECTROCHEMICAL INVESTIGATIONS 9.1. Electrochemical Generation of Nitroxyl Adducts Denitrification reduces nitrate to a mixture of nitrogen containing compounds. Nitrate decomposition by supercritical water oxidation (SCWO) reactors, thermal processes and by electrolytic and chemical methods is a topic of 20 U.S. patents issued since 1976 [100]. The most efficient use of electrolysis for possible industrial removal of nitrate with cadmium cathodes yielded 70–75% N2 , ∼25% NH3 , ∼2% N2 O with an overall current efficiency of 55%, but the exact mechanism of the reaction remains unknown [101]. In nature the denitrification is achieved enzymatically by various bacteria, including genera Pseudomonas, Micrococus, Archromobacter, Thiobacillus and Bacillus. Biological denitrification utilizes metalloproteins to accomplish the whole or part of the denitrification process, but the mechanism of the reduction of NOx during denitrification is still disputed. Electrochemical analysis of NOx adducts and electrocatalysis of NOx by metal complexes allow for the mechanistic understanding of this process. The catalysis of NOx by proteins specifically designed for this function, like NiR, allow for the study of the turnover numbers and other aspects of the overall catalytic process, but in order to understand mechanistic steps and possible intermediates a less efficient system for catalysis is needed. Simple metal complexes, like hemes, allow for the electrochemical study of NOx adducts and catalysis under much more variable conditions. Likewise, proteins that are not specifically designed for NOx catalysis but can still perform it, albeit less efficiently, are myoglobin (Mb) and hemoglobin (Hb) [102,103]. The metal-catalyzed reductions of NOx are readily studied electrochemically, and the obtained results can result in the isolation and characterization of NOx adducts and intermediates, ultimately leading to a more intimate understanding of the mechanisms involved. The overall investigations of NOx electrocatalysis has been covered in more detail in a recent review [104], the following section emphasizes HNO in particular.

9.2. Nitroxyl Fe Porphyrin Complexes One of the more straightforward methods of making nitroxyl adducts electrochemically is by a one-electron reduction of a stable nitrosyl adduct [105,106]. Initial examinations of the redox chemistry of the nitroso ferrous(tetraphenylporphyrin), FeII (TPP)(NO) by the Kadish group, showed an electrochemically reversible reduction of the adduct in non-protic solvents. Transient spectra of the product FeII (TPP)(NO)− were observed, but attempts to isolate it yielded only the more stable starting material, FeII (TPP)(NO) [107,108].

Bioinorganic Chemistry of the HNO Ligand

441

Later, Ryan characterized the diamagnetic product Fe(TPP)(NO)− by 1 H-NMR, UV/visible and resonance Raman spectroscopy, which was stable in the absence of excess NO [109]. The NO band decreased from 1681 cm−1 (15 N 1647 cm−1 ) in the nitrosyl adduct to 1496 cm−1 (15 N 1475 cm−1 ) in the nitroxyl; likewise the Fe N at 525 cm−1 (15 N 517 cm−1 ) of the nitrosyl adduct increased to 549 cm−1 (15 N 538 cm−1 ) upon reduction, suggesting a strengthening of the Fe N and a weakening of the N O bond due to the addition of an electron to a half-filled -bonding orbital. The nitroxyl slowly reacted with proton sources, such as phenols, to convert back to the stable nitrosyl adduct [110]. In contrast, the two-electron reduced product Fe(TPP)(NO)2− reacts rapidly with phenols to yield ammonia. The proton dependence of these reductions was studied using para-substituted phenols as proton sources in non-aqueous solutions [110]. An initial reversible reduction of Fe(TPP)(NO) was followed by two reversible protonations before an irreversible protonation and a three-electron reduction formed hydroxylamine (Eqs 7–10). FeTPPNO + e−  FeTPPNO− −

+

(7)

FeTPPNO + H  FeTPPHNO

(8)

FeTPPHNO + H+  FeTPPH2 NO+

(9)

+

+

−

−

FeTPPH2 NO + H + 3e → FeTPP + NH2 OH

(10)

Ryan and coworkers also compared electrochemical reduction of different porphyrin derivatives and correlated them to apparent basicity of the reduced Fe-nitroxyl [108–113]. The reduction potential (vs. NHE) of the nitrosyl adducts was −0.69 V for Fe(TPP)(NO) [108], −0.78 V for Fe(OEP)(NO) [113], −0.63 V (irr.) for Fe(TPPS)(NO) [115], −0.93 V for Fe(2,4-DMOEiBC)(NO) [113] and −0.45 V for Fe(2,4-OEPdione)(NO) [113]. Iron-isobacteriochlorins, for example 2,4-dimethyloctaethyl isobacteriochlorin (2,4-DMOEiBC), which model the assimilatory siroheme enzymes NiR, were the most basic. The oxo-derivatives, for example 2,4-octaethylporphinedione (2,4-OEPdione), which model the green cytochrome d1 , NiR, of the denitrifying enzymes were the least. These results mirror their enzymatic activity, in that the former produces NH3 and the latter NO and N2 O. Meyer and coworkers examined the electrocatalytic reduction of nitrite using water soluble iron porphyrins (e.g. FeII -tetra(4-sulfonatophenyl)porphyrin)4− or Fe(TPPS)4− ) which yielded ammonia, nitrous oxide and smaller amounts of dinitrogen in aqueous solutions [114–117]. The initial reduction of the ferrous nitrosyl to the nitroxyl was assigned to a potential of ca. −600 mV vs. NHE with subsequent bimolecular coupling of nitroxyl-heme intermediates to form the observed N2 O. The half-life of the nitroxyl product, generated by flash photolysis in aqueous pH 6 solution, was only two seconds [118].

9.3. Nitroxyl Heme Proteins Several important model systems for the reactivity of NOx have been developed using simple heme proteins such as myoglobin and hemoglobin in surfactant film–modified

442

F. Sulc and P.J. Farmer (A)

(B)

Blank Mb electrode

(C)

10 μA

+Nitric Oxide (NO)

+Nitrite (NO2– )

0.2 – 0.1 – 0.3 – 0.5 – 0.7 –1.0 –1.2

0.2 – 0.1 – 0.3 – 0.5 – 0.7 –1.0 –1.2

0.2 – 0.1 – 0.3 – 0.5 – 0.7 –1.0 –1.2

V/SCE

V/SCE

V/SCE

Fig. 9. Voltammograms of Mb/DDAB at 100 mV/s demonstrating catalytic activity: (A) alone in pH 7 buffer; (B) in presence of 5 mM NaNO2 ; (C) in saturated NO solution (ca. 3 mM) [89,120].

electrodes. Rusling first reported that myoglobin contained within dimethyldidodecylammonium bromide (DDAB) films on an electrode showed greatly enhanced electrochemical response, indicative of diffusion controlled currents within the thin film [118]. Such an electrode, termed Mb/DDAB, is stable to repeated cycling, can be moved from solution to solution, and will remain active for several weeks with proper storage. Coulometry of Mb/DDAB modified electrodes consistently showed that a large percentage of the deposited Mb remain electrochemically active. The temperature dependence for the FeIII/II current-response follows that of the solid-to-liquid crystal phase transition for DDAB films [120]. Importantly, these Mb/DDAB films allow access to a wide potential range, with both the FeIII/II (−0.22 V vs. SCE) and FeII/I (−1.09 V vs. SCE) couples observable at pH 7, as illustrated in Fig. 9 and Eqs 11 and 12. Mb

FeIII + e−  Mb

FeII

(11)

Mb

FeII + e−  Mb

FeI

(12)

Lin first showed that when nitrite was introduced to the solution above, two new catalytic reduction waves appeared close to the FeII/I couple, both of which displayed complex pH dependence indicative of multiple reaction pathways [89]. Online mass spectral analysis of the head gas over electrolyzed solutions of 15 NO2 − demonstrated that production of 15 N2 O was initiated at the first catalytic wave, and tentatively assigned to the reduction of the ferrous NO adduct [121]. The second wave showed pH dependence suggestive of a multi-electron (ca. 3e− /H+ ) limiting step associated with the production of NH3 (Eq. 13). − + + + NO− 2 + ne + mH → NH4 + NH3 OH + N2 O + NO + N2

(13)

In a subsequent study by Bayachou, the ability of Mb to catalyze the reduction of NO was investigated [122]. Voltammograms of Mb/DDAB under an NO solution, C in Fig. 9, show catalytic reduction of NO characterized by the loss of the FeIII/II couple and an introduction of a large cathodic current at ca. −600 mV NHE. The reduction of

Bioinorganic Chemistry of the HNO Ligand

443

NO by Mb/DDAB was distinct from the reduction of NO2 − , since it occurred at more positive potentials and remained catalytic at all pH tested and differed in the catalytic current. Bulk electrolysis of saturated 15 NO solutions produced 15 N2 O, as identified by mass spectroscopy, with no 15 NH3 formation observed. This indicates that an N N coupling reaction, analogous to that of P450nor, was the primary reaction. Fast scan voltammetry at rates >1 V/s, under NO gas, showed the fingerprint of reversibility in the form of anodic currents associated with the re-oxidation of the FeII nitroxyl state for the nitroxyl adduct, which implied that a nitroxyl adduct had a measurable lifetime during the catalytic reaction. The same fingerprint was apparent after successive reductive scans in a nitrite solution. The apparent reversibility decreased with increasing concentrations of H+ or NO, implying that both these species react with the Fe-bound nitroxyl, as in Eqs 14–18. FeII

Mb

Mb FeII Mb Fe

NO + e−  Mb

FeII

NO−

NO− + H+ → Mb

FeII

HNO

−

+

NO + NO + H → Mb

II

Mb FeII Fe

II

HNO  Mb −

NONO  ↔ Mb

(14)

Fe + N2 O + HO III

(15) −

(16)

FeII + HNO Fe

III

−

(17) −

NO NO 

(18)

Based on the fingerprint of reversibility and in order to determine the lifetime of the nitroxyl intermediate, modified DDAB electrodes were fabricated directly from Mb NO with the hopes of observing a single-turnover reduction in the absence of excess NO. A reversible single-electron reduction was indeed observed. The reduction potential of this couple (Eq. 14) was pH-independent, but short-lived at pH 7, as demonstrated by the low reversibility at faster scan rates in Fig. 10, with the lifetime on the order of a tenth of a second (determined by digital simulation). The two pathways for N2 O formation are thus suggested (Fig. 11) in that nitroxyl may decompose by loss of HNO, path A (Eq. 17), as was proposed by Hollocher [88] and others [111], or by the coupling of heme-bound nitroxyl with exogenous NO, path B (Eq. 16), as proposed for P450nor (Fig. 8) [96]. Digital simulations of the catalytic voltammograms gave estimates for the N N coupling reaction (Eq. 16)

pH 7 3 V/s

–0.6

pH 10 20 mV/s

pH 8.6 1 V/s

–0.8

V vs. SCE

–1.0 –0.6

–0.8

V vs. SCE

–1.0 –0.6

–0.8

–1.0

V vs. SCE

Fig. 10. Reversibility of Mb NO reduction in different pH solutions, circles in top represent simulation used to derive lifetime of Mb NO− [120].

444

F. Sulc and P.J. Farmer path A _

O

H+

FeII + HNO

N2O + H2O

N FeII

O–

O– NO path B

N

O

N

H+ O–

N

N

FeII

FeIII

Fig. 11. Two suggested mechanisms for N2 O formation.

as 108 M−1 sec−1 [122]. Subsequently, it was found that the protonation of Mb NO− does not induce the release of free HNO but gives the stable nitrosyl hydride adduct, Mb HNO [11]. The effect of proximal ligation on the NoR activity was addressed by comparing the electrocatalytic activity of Mb with that of a thermophilic cytochrome P450, CYP119 [123]. Because the NO adduct of CYP119 is not isolable, a method to generate it in situ was developed via pulsed electrochemical reduction in a nitrite solution. Both Mb and CYP119 have comparable reduction potentials and lifetimes for the ferrous nitroxyl intermediate FeII NO− , but the overall catalytic current is markedly reduced for CYP119. The comparatively lower reactivity of a cytochrome P450 for NO reduction had in fact been predicted in a theoretical paper by Silaghi-Dumitrescu, but a decrease in the reduction potential of the FeII NO adduct due to thiolate ligation was not supported experimentally [85].

10. NITROXYL ADDUCTS OF HEME PROTEINS During 1980s, the Doyle and Hollocher groups published a series of papers exploring the reaction of AS with heme proteins [124–127]; that have been widely referenced to explain the biological effect of HNO [128]. Hollocher showed that the reaction of Mb FeIII (whale) with AS was first-order in AS and zero-order in ferric heme protein [125]; the rate of Mb FeIII disappearance was slower than the rate of AS decomposition, with the trapping rate by Mb FeIII estimated at ∼80%. This was held as proof that AS decomposition produces HNO and NO2 − (Eq. 19). But in the reaction of AS with Hb FeIII (human), the accumulation of Hb NO was sigmodial over time, as opposed to the expected simple exponential binding curves seen for Mb NO production in equivalent experiments. Originally the behavior was attributed to allosteric interactions between hemes in Hb, but in 1988 Doyle and coworkers demonstrate that the sigmodial binding curves were due to competitive reactivity of HNO with -93 cysteine; more typical binding curves were obtained upon blocking the -93 cysteines with iodoacetamide prior to the reaction [127]. The oxidation of thiophenol by AS to form phenyl disulfide, NH2 OH and NO2 − was used to illustrate the reaction of HNO with RSH. The proposed

Bioinorganic Chemistry of the HNO Ligand

445

mechanism based on the observations involves Eqs 20 and 21. In order to explain the low H2 NOH yield, Eq. 22 was proposed. AS → HNO + NO− 2 HNO + C6 H5 SH → C6 H5 SNHOH

(19) (20)

C6 H5 SNHOH → C6 H5 SSC6 H5 + H2 NOH

(21)

HNO + H2 NOH → N2 + 2H2 O

(22)

The reaction of ferrous hemoglobin, Hb FeII , with AS was also source of confusion: both groups observed an accumulation of Hb FeIII and Mb NO when Hb FeII was reacted with AS at pH 7 and 25 C. Various explanations were proposed to account for production of Hb FeIII : Doyle initially proposed that AS decomposes to give NO and the nitrous acid radical anion, HONO− ; Hollocher’s group proposed one-electron reduction of AS by Hb FeII to give Hb FeIII + NO + HNO + OH− or, alternatively, a two-electron trimolecular reaction (2 Hb FeII + AS) that would produce 2 Hb FeIII + − 2HNO + OH− . The previously accepted products from AS, HNO/NO− 2 or NO /HNO2 III were dismissed based on the observation of Hb Fe . In a latter report, Doyle attributed the accumulation of Hb FeIII to reaction of the HNO adduct with NO− 2 , which generates both Hb FeIII and Hb NO, and also to a new reactivity that included an HNO adduct as an intermediate (Eqs 23 and 24). Analogous reactivity is seen for the reaction of AS with Mb FeII , and will be discussed subsequently. Hb + HNO → Hb NO2 − + H + + Hb

HNO → Hb

HNO

(23)

FeIII + NO + HNO + OH−

(24)

More recently, Gow and Stamler have suggested generation of NO− upon anaerobic reaction of NO with hemoglobin (Hb) [129]. Excess NO reacts with Hb to form surface S-nitrosothiols and four ferrous NO adducts that are relatively unreactive toward O2 or CO. However, at substoichiometric NO concentrations, complicated kinetics were observed. It was proposed that nitroxyl anion (NO− ) was liberated under conditions when the NO/Hb ratio was intermediate (0.3 to 0.6), generating oxidized ferric hemes (i.e., Hb FeIII ). At lower ratios only Hb NO was detected, at higher ratios the generated Hb FeIII further reacted with NO to form the ferrous nitrosyl, Hb NO and nitrite.

10.1. The HNO Adduct of Myoglobin: Synthesis and Characterization The reversible couple of Mb NO/Mb NO− observed electrochemically suggested a stable nitroxyl adduct may be obtained by stoichiometric reduction of Mb NO; subsequently this was confirmed by bulk electrolysis of Mb NO using N-methyl-4,4 bipyridine as a redox mediator and by bulk chemical reduction of Mb NO with CrII salts (Eq. 25) [11]. Once generated and isolated, the nitroxyl adduct is stable in anaerobic solutions from pH 6 to 10 for over six months. Its synthesis is pH dependent, with highest

446

F. Sulc and P.J. Farmer 1.2

NO HNO

Absorbance

1.0

Mb Mb

0.8

x 10 0.6 0.4 0.2 0 250

350

450

550

650

Wavelength (nm)

Fig. 12. Comparison of the absorbance spectra of Mb NO and Mb HNO.

yields obtained at pH 10; the maximum yield is ca. 60–80%, due to over-reduction of the product to Mb FeII . Mb

NO + Cr II + H+ → Mb

HNO + Cr III

(25)

The reduction of Mb NO to Mb HNO produces only a subtle change in electronic absorbance spectra, the Soret band shifts from 421 to 423 nm with similar absorbitivity, and the Q-bands shift to slightly higher energy with a slight drop in intensity of the lower energy feature (Fig. 12). As a result, the HNO adduct spectra is easily misinterpreted as that of the well known Mb NO, especially in mixtures of products. The 1 H NMR provided definitive identification of the nitroxyl adduct, as it is diamagnetic with a unique nitroxyl peak at 14.8 ppm indicated that the Mb NO− was actually protonated in the active site as Mb HNO (Fig. 13) This peak is split into a doublet (72 Hz) in samples made from Mb 15 NO, consistent with protonation at the nitrogen, Mb H15 NO. The stability of the nitroxyl adduct with time was observed based on the intensity of the HNO peak which remained unchanged from pH 6 to 10 relative to peaks assigned to non-ionizable protons, such as those of the methyl groups of Val 68 at −0.9 and −2.7 ppm. The adduct may also be formed by trapping of free HNO by deoxyMb (Mb FeII ) produced from the decomposition of methylsulfonylhydroxylamine (MSHA) or Angeli’s salt (AS), to directly yield Mb HNO (Eq. 26) [130]. The generation and quantification of Mb HNO was analyzed by several spectroscopies: 1 H NMR for the formation of diamagnetic Mb HNO, UV/visible to follow the loss of Mb FeII and EPR to quantify persistent Mb NO impurities. The maximum Mb HNO yield obtained was ca. 80%; competitive side reactions with byproducts as well as the further reactivity of the Mb HNO combine to decrease the overall yield. At pH 10, the observed rate of Mb HNO generation by trapping HNO from MSHA is close to that for MSHA

Bioinorganic Chemistry of the HNO Ligand

447 Val 68

(A) HNO

Mb

(B) H15 NO

15

14

Mb

13

12

11

10

–1

–2

–3

–4

–5

ppm

Fig. 13. Water-decoupled 1 H NMR spectra of: (A) natural abundance Mb HNO, (B) Mb H15 NO. Broad peaks are due to paramagnetic Mb impurities, modified from reference [11].

decomposition; kinetic simulations give a lower limit to the bimolecular rate of trapping as 14 × 104 M−1 sec−1 . Mb Mb

FeII + HNO → Mb

FeII + NO2 − + H+ → Mb

HNO

(26)

FeIII + NO + OH−

(27)

As was seen for Hb FeII , the reaction of stoichiometric AS with Mb FeII at neutral pH generates a complex mixture of products which includes transient formation of metMb (Mb FeIII ), but ultimately gives Mb NO as the major product. The transient formation and accumulation of the ferric protein at the same time that the ferrous form is rapidly diminishing suggests that HNO reacts faster with Mb FeII than Mb FeIII , contrary to several published theories [128]. Time-based analysis of the reaction by 1 H NMR showed the disappearance of Mb FeII (t1/2 ≈ 4 min.) was ca. 4 times faster than the known decomposition rate of AS (t1/2 = 17 min.). Angeli’s salt decomposes to give both HNO and NO2 − , therefore the competitive reaction of nitrite with Mb FeII was investigated. The rate of reaction of Mb FeII with nitrite was found to be zero order in protein, and first order in nitrite and protons (Eq. 27) [131]. Good yields of Mb HNO were obtained for stoichiometric reactions by simply raising the pH and diluting the nitrite concentration to inhibit the secondary reaction (Fig. 14). Thus the reaction of AS with deoxyMb consists of a series of sequential reactions (Eqs 28–34) some of which are dependent on pH and concentrations. At 1 mM, the half-life for the reaction of Mb FeII and nitrite is ∼1 min, producing both free NO and Mb FeIII (Eq. 30). NO readily reacts with Mb FeII to form Mb NO (Eq. 31). Likewise, the reaction of Mb FeIII with HNO gives Mb NO (Eq. 32). The HNOadduct itself further reacts with both nitrite and NO, ultimately forming Mb NO by as yet unknown mechanisms [132]. These results help to explain the previous ambiguities concerning the reaction of Hb with AS, and likewise illustrates a major complication

448

F. Sulc and P.J. Farmer 1

4

14.7

11.0

2

4 1,2

3 1123

(A)

(B)

15.0

10.5

10.0

9.5 ppm

Fig. 14. 1 H NMR spectra of (A) 1:1 reaction of AS with Mb FeII at 4 mM concentration, pH 7 after 4 min; and (B) at a 1.5:1 stoichiometry, 5 M Mb FeII concentration, and pH 8, after 30 min. Numbers above the spectra indicate peaks due to Mb HNO (1) Mb NO (2) Mb FeII (3) and Mb FeIII (4) Combined 1 H NMR, UV/visible and EPR analysis give estimated concentrations for reaction mixture in (B) as 70% Mb HNO and 30% Mb NO.

inherent in the use of AS as an HNO donor for medical applications – the additional reactivity of nitrite. HN2 O3 ↔ HNO + NO2 − Mb FeII + HNO → Mb −

HNO

+

Mb Fe + NO2 +H → NO + Mb II

(29) Fe + OH III

−

(30)

Mb FeII + NO → Mb

NO

(31)

Mb Fe + HNO → Mb

NO

(32)

III

3Mb

(28)

FeII + HNO + NO2 − + H+ → Mb

2Mb FeII + HNO + NO2 − + H+ → 2Mb

HNO + Mb NO + OH−

11. BONDING PARAMETERS IN Mb

NO + Mb

FeIII + OH− (33) (34)

HNO

Isolated solutions of Mb HNO are stable for many months anaerobically, which has allowed the adduct to be investigated by several spectroscopic methods (Table 1). X-ray absorption (Fe K-edge) spectroscopy generated a best-fit XAFS model that shows the expected bending of the Fe N O angle in Mb HNO, as well as a lengthening of both the N O and Fe N bonds as compared with those of Mb NO [133]. The XAFS analysis was based on previously optimized MS XAFS calculations for the structure of Mb NO (R = 138) [135]; for the analysis of Mb HNO a somewhat higher goodnessof-fit value was obtained (R = 155), likely because of deoxy-Mb and Mb NO impurities. The XANES edge energies for MbII HNO, MbII NO and MbIII NO are also listed in Table 1. The edge energy decreases in the order MbII HNO > MbII NO > MbIII NO, correspondingly to the increase in the oxidation state from {FeNO}8 to {FeNO}7 to {FeNO}6 , but the difference between Mb HNO and Mb NO is much larger than that between MbII NO and MbIII NO. Such a large increase in the edge energy suggests an increase in the negative charge, that is reduction, at the metal center.

Bioinorganic Chemistry of the HNO Ligand

449

Table 1. Bonding parameters for Mb HNO obtained from X-ray absorbance and resonance Raman spectroscopies XAFS/Raman Fe N (por), Å Fe N (His), Å Fe N (xNO), Å Edge energy, eV N O, Å Fe N O, deg

N O cm−1

Fe N cm−1 a

MbII HNOa

MbII NOb

MbIII NOb

200 209 182 71225 124 131 1385 651

199 205 176 71247 112 150 1613 551

200 204 168 71254 113 180 1927 595

data from [133]; b XAFS data from [134], IR data from [135].

Resonance Raman spectra were obtained by Soret-excitation ( ex = 413 nm) of Mb HNO and Mb H15 NO in the high- (1300–1700 cm−1 ) and low-frequency (500–800 cm−1 ) regions and isotope-difference features were seen in both. The lowfrequency data displayed a single band that downshifts from 651 cm−1 to 636 cm−1 upon 15 N substitution, tentatively assigned to the FeII N stretching mode. The assigned frequency is significantly higher than that of either the ferric or ferrous adducts of Mb NO, and suggestive of a stronger FeII N bond in the nitroxyl than in either of the nitrosyl complexes. However, the bending of the HNO ligand effectively raises the frequency independent of change in bond strength, and in highly bent geometries the Fe N stretching and Fe NO bending modes can be highly mixed [136]. Likewise, the interpretation of the high-frequency isotope-difference data is complicated by overlap with the very strong 4 heme mode. Spectral simulations were used to identify a band of comparable intensity that downshifts from 1385 to 1355 cm−1 upon 15 N substitution; this band has been assigned as the N O stretching mode of the bound nitroxyl group. The decrease in N O stretching frequency from 1617 cm−1 in Mb NO to 1385 cm−1 in Mb HNO is consistent with the formally ligand-based reduction; the corresponding decrease as compared to free HNO at 1510 cm−1 demonstrates that the HNO ligand acts as a strong -acceptor in complex with FeII . Less expected was the observed increase in Fe N stretching frequency from 554 cm−1 to 651 cm−1 . An increase in the Fe N bond strength may explain the unusual stability of the HNO adduct, but is difficult to correlate with the EXAFS data showing a lengthened Fe N bond. As nitroxyl adducts of model Fe-porphyrin complexes (e.g. Fe-TPP and Fe-TPPS) are short-lived, the stability of Mb HNO must be due to a stabilizing influence within the protein itself. To address this, 1 H NMR, NOESY and TOCSY data were obtained on HNO adduct of horse heart Mb and compared with that of the CO adduct [136]. The 2D NOESY maps showed over 20 NOE connectivities with the nitroxyl absorbance, indicative protons in the distal heme pocket. NOESY spectra were used to verify the spatial orientation of residues with respect to the heme and TOCSY spectra verified assignments of protons belonging to specific residues. A computer model of Mb HNO was generated by replacing CO with HNO in the crystal structure of CO Mb and by adjusting position of residues and nitroxyl based of any differences in the position or the NOEs of the two adducts. The major difference between the residues of the two Mb

450

F. Sulc and P.J. Farmer (B)

(A) α

Val68

3

2

4

His64

1 N

N

V68

H

β

Fe O

δ

H93

N

N H64

5 6

8 7

γ P

His93 P

Fig. 15. The orientation of the HNO ligand in Mb adduct. (A) axial overview of HNO orientation to the heme in Mb HNO; circles indicate the position of particular protein residues. (B) stick model of Mb HNO, showing non-bonding interactions with Val68 and His64.

adducts was the movement of the distal histidine 64, based on NOESY cross peak and chemical shift changes, by ca. 20 rotation to move within hydrogen bonding distance of the nitroxyl oxygen (Fig. 15). The spatial orientation of the HNO ligand was determined by the cross peak intensities to two heme meso-protons, and further supported by NOE intensities to other residues. Contrary to expectations, a single orientation of the HNO was found, ca. −104 to the N Fe N vector between - and - meso-protons (Fig. 15). This orientation situates the HNO almost perpendicular to proximal His93 (at ∼0 ), suggestive of a -backbonding competition between the ligands; similar orientations have been observed for RNO complexes as will be described below. Steric hindrance also seems to play a role, as a similar single orientation was obtained by simple energy minimization.

11.1. RNO and HNO Complexes of Heme Proteins and Metalloporphyrin HNO is the simplest analogue of aryl and alkylnitroso compounds, RNO, whose coordination chemistry has recently been reviewed [137,138]. Mansuy and coworkers were the first to describe the binding of RNO compounds to ferrous cytochrome P450s [139] and the oxygen-binders Mb and Hb [140]. They subsequently reported the first crystallographic characterization of an RNO model heme complex, (TTP)Fe(i PrNO)(i PrNH2 ) [141]. More recently, the group of Richter-Addo has published a series of papers concerning Fe- [142,143] Ru- [144,145] and Os- [146] porphyrin adducts of alkylnitroso compounds. The Richter-Addo group has also crystallographically characterized the nitrosyl and ethylnitroso adducts of Mb [79]. The Mb EtNO structure (pdb 1NPG) places the nitroso

Bioinorganic Chemistry of the HNO Ligand

451

oxygen at almost the same position as that of the HNO in the 1 H NMR structure of Mb HNO, within a comparable hydrogen bonding distance to His64 and a similar perpendicular orientation to the trans His93. Preferential orientations of an axial RNO to its trans ligand is commonly observed in the metalloporphyrin adduct structures and has been explained by an energetic preference to maximize -bonding interactions and thus implies a strong -bonding component to the RNO-metal bonds. The Mb EtNO structure also shows a steric bending of EtNO group by ∼30 (from perpendicular) away from His64 and Val68, apparently due to the inability of the Et group to fit between the two Val68 methyl groups, as in the 1 H NMR structure for Mb HNO. The Richter-Addo group has also recently generated the first characterizable metalloporphyrin HNO adduct, [HNO-Ru(ttp)(1-MeIm)] (ttp = tetratolylporphyrinatodianion, 1-MeIm = 1-methylimidazole) generated by hydride addition to the stable cationic nitrosyl [Ru(ttp)(NO)(1-MeIm)]BF4 [147]. This HNO adduct is moderately stable as a solid, but decomposes in solution over 6 h to give unidentified nitrosyl containing species. The spectral characteristics of HNO-Ru(TPP) match well with that found for Mb HNO, with an N O at 1380 cm−1 and a characteristic 1 H NMR peak at 13.64 ppm.

12. NON-HEME NITROXYL COMPLEXES 12.1. NO− Complexes Unprotonated nitroxyl complexes, as predicted by the Enemark–Feltham rules, have low NO stretching frequencies indicative of occupation of the N O antibonding  ∗ orbitals. Enemark and Feltham showed that the addition of an anionic ligand to a fivecoordinate linear nitrosyl metal complex, [Co(das)2 NO]2+ (Co N O 178 , Co N 1.68 Å, N O 1.16 Å, NO = 1852 cm−1 ), resulted in two different six-coordinate complexes: [Co(das)2 (NCS)(NO)]1+ (Co N O 132 , Co N 1.85 Å, N O 1.18 Å,

NO = 1587, 1561 Å) from reaction with NCS− and [Co(das)2 (NO)Br]+ ( NO = 1565, 1550 cm−1 ) from reaction with Br− . These transformations were described as formal two-electron reduction of a coordinated NO+ to NO− . The latter complex could be protonated to tentatively form [Co(das)2 (HNO)Br]2+ as characterized by a less intense

NO at 1560 cm−1 and nitrosyl hydride signal (sic) at 2.1 ppm in the 1 H NMR [148]. Bridging NO− may be obtained by the oxidative addition of NO+ to reduced binuclear metal centers, for example, [(Mn L)2 ]+m to [(Mn+1 L)2 (NO− )]+m+1 . Lippard and coworkers found that NO+ reacted with dimeric FeIII and CoIII complexes with N2 S2 ligands to form cationic, bridging nitroxyl adducts with NO at 1553 cm−1 (Fe) and 1545 cm−1 (Co). Crystallographic characterization demonstrated that both complexes had long N O bonds (1.193 Å for the Fe complex and 1.211 Å for the Co) comparable to that in free HNO [25,149]. Similarly, a bridging NO− was obtained by reaction of NO+ with a binuclear compound with CuI in N3 O ligation, used as models for the active site of Cu nitrous oxide reductases; the cationic product [CuII2 (XYL O− )(NO− )]+2 had a lower energy NO at 1536 cm−1 but a much shorter N O bond at 1.176 Å. [150]. Another example of a bridging nitroxyl is found in [(Ir(COD)(Cl)( -pz))2 ](NO− ), formed by several routes from IrI precursors, and has a NO of 1630 cm−1 [151].

452

F. Sulc and P.J. Farmer

12.2. Characterized HNO-Adducts The few well-characterized examples of protonated nitroxyl complexes, formally nitrosyl hydrides, are air- and water-sensitive organometallic compounds A metal-bound HNO is isoelectronic with 1 O2 and all characterized complexes are with low spin d6 metals and are diamagnetic. Roper (Ir and Os) [152] and La Monica (Re) [153] synthesized the first true HNO complexes by addition of HCl to five-coordinate nitrosyl complexes. Several other methods of obtaining an HNO-metal complex have been recently demonstrated, as illustrated in Fig. 16. These include a two-electron oxidations of a hydroxylamine adduct, an insertion of NO into an M H bond, and an addition of hydride to a metal-bound NO. At the heart of these transformations is the complex redox chemistry of NO itself. In 1970, Roper and coworkers suggested an HNO adduct as a likely intermediate in the conversion of Ir(NO)(PPh3 )3 to IrCl3 (NH2 OH)(PPh3 )2 upon addition of excess HCl (Table 2) [152]. Using stoichiometric HCl addition to OsCl(CO)(NO)(PPh3 )2 , a product was isolated and assigned to OsCl2 (CO)(HNO)(PPh3 )2 based on the low energy

NO at 1410 cm−1 and elemental analysis [152]. Subsequently, Ibers and coworkers confirmed this formulation crystallographically [154], and the nitrosyl hydride identified by its unique absorbance at 21.1 ppm in the 1 H NMR, which splits when isotopically labeled with 15 N. The 15 N–1 H (1 JNH ) coupling of 75 Hz is indicative of metal bound sp2 -hybridized HNO and the NO band at 1410 cm−1 was significantly lower than those of unprotonated nitroxyl complexes (Table 2). The crystal structure shows a long N O bond at 1.193 Å, as well as a short N H bond at 0.94 Å as compared to that in free HNO at 1.026 Å [154].

H

OH H2N

reduced –3 NH3

Pb(OAc)4

MLx

MLx

Mb FeII

HNO

Mb NO

e−

H

O N Mb

N2O

H

NO HNO

O N

HCl

MLx

MLxCL

NO

H NO

+3

–

NO2

+

N

NH2OH 0

O

H

–

O N MLx

MLx

oxidized Oxidation States of NO

H MLx

NO

+

H

O

+

N

MLx

Fig. 16. Synthetic routes for preparation of metal nitroxyl complexes.

1970

Year O

N OC R3P

Os

1970

1974

1974

1997 O

H

N

PR3 Cl

Re(X)(Cl)(NO) (HNO)(PR3)2

OC R3P

Cl

2000 H

Re

+

O N H R3P

O

H

Ir

N

PR3 Cl

OC R3P

Cl

Re

N

PR3 CO

O

H

2002 O

H

N

S S

OC R3P

O N

Fe

OC R3P

N

CO

2002

N

H

N

PR3 Cl

Ru

S S

CO

structures

2001 O

H

Os

M

2004 H PR3 Cl

H O

+

Ru N

CO

NH

complex R

2

1 Ph

Ph

M/X 1H

NMR (ppm) 21.21

Ph

Ph

Cl

I

4a Ph

20.662 66.22

757

77.68

IR νNO cm–1 (ν15NO)

1410s7 (1393)s7

1493s8 (1465s)8

Reference

3b

22.752

JNH (Hz)

νNH cm–1 (ν15NO)

3a

1370 1720

2810w8 (2801w)8 [152,154]14 [152,155]14 [153]

1380 1745

1376s8

4b Cy

21.352

1335s8

3059w8

[153] [153, 156]

7

Ph

21.662

14.83

72.52

72.53

1391m8

13859

19.564

[153, 156]

[153,155]

8

9a

9b

Ph

iPr

iPr

Ru

Os

20.95

20.91

20.72

10

13.646 71.16

1358t10 137811

13658

139212

138813

138011 (1348)11

[159, 160]

[159, 160]

[147]

3056w8

[11,133, 136]

[158]14

[155]

453

characterized.

6

N

3056w8

NMR: 1 C6D6, 2 CD2Cl2, 3 20% D2O, 4 [D8]THF, 5 C7D8, 6CDCl3 IR: 7 Nujol mulls, 8 fluorolube/CaF2, 9 H2O, 10 KBr, 11 DFT calculation,12 C6D6, 13 C6H6 14 Crystallographically

5

H

N

PR3 Br

Bioinorganic Chemistry of the HNO Ligand

Table 2. Spectroscopic data for known metallo-HNO complexes

454

F. Sulc and P.J. Farmer

The complex IrHCl2 (HNO)(PPh3 )2 , first identified by Roper, has only recently been structurally characterized by Hillhouse and coworkers [155]. The structure displays pseudo-octahedral symmetry with trans phosphines and cis chlorides. The nitroxyl proton points syn toward the chloride, while the oxygen points toward the hydride. A solvent THF molecule is within hydrogen bonding distance to HNO. The orientation of the HNO allows for -backbonding with a filled metal d-orbital. The Hillhouse group has also investigated the mechanism of its formation. An initial slow oxidative addition of HCl reaction to four-coordinate Ir(NO)(PPh3 )3 gives a fivecoordinate IrHCl(NO)(PPh3 )2 , followed by a rapid second addition of HCl to generate the six-coordinate HNO adduct [152,155]. In the presence of ethanol, further reaction with HCl results in reduction of the metal-bound HNO to hydroxylamine, cis,transIrCl3 (NH2 OH)(PPh3 )2 . The addition of HCl was shown to be reversible by addition of KOH to the hydroxylamine adduct to reform the four-coordinate Ir(NO)(PPh3 )3 ; single dehydrohalogenation of the HNO complex was effected by addition of NaN(SiMe3 )2 to give the IrHCl(NO)(PPh3 )2 . The Ir HNO complex could be regenerated upon addition of HCl [155]. A Re HNO adduct was first described by La Monica, and subsequently characterized by the Hillhouse group using IR and NMR spectroscopies and theoretical modeling [153,156]. As with the Ir HNO complex, the addition of HCl to a five-coordinate Re nitrosyl, Re(CO)2 (NO)(PR3 )2 , yielded the HNO complex Re(Cl)(CO)2 (HNO)(PPh3 )2 . In this case, the reaction most likely proceeds through a 1,2 addition of HCl across Re N bond of a slightly bent nitrosyl complex. The syn position of Cl to the nitroxyl proton is predicted by this mechanism, and may stabilize the complex by electronic interaction [155]. The cationic [Re(CO)3 (HNO)(PPh3 )2 ][SO3 CF3 ] nitroxyl complex has been synthesized by two different routes: the first, by oxidation of a hydroxylamine adduct [Re(CO)3 (NH2 OH)(PPh3 )2 ]+ , with Pb(OAc)4 [161]; the second involved 1,1-insertion of NO+ [NO][SO3 CF3 ], into the metal-hydride bond in ReH(CO)3 (PPh3 )2 , to form [Re(HNO)(CO)3 (PPh3 )2 ][SO3 CF3 ]. When [PF6 ] or [BF4 ] were used instead of [SO3 CF6 ] the [Re(HNO)(CO)3 (PPh3 )2 ]+ complex was unstable and decomposed to a neutral fluoride complex ReF(CO)3 (PPh3 )2 [157]. Sellman and coworkers reported the first ruthenium nitroxyl complex [Ru(HNO)(‘pybu S4 ’)] [158], prepared by addition of NaBH4 to a cationic octahedral nitrosyl complex [Ru(NO)-(‘pybu S4 ’)]+ , a route analogous to hydride reduction of a metallocarbonyl to give a formyl (HCO) complex. In the solid form, [Ru(HNO)-(‘pybu S4 ’)] is present as a dimer with hydrogen bonds between the nitroxyl proton and the CH3 OH solvate and nitroxyl O and the thiolate of the other molecule; the HNO is oriented parallel to the trans pyridine, arguing against -backbonding competition, but this may be due to the multiple hydrogen bonding interactions. Solvent-dependent differences in (NO), as well as DFT modeling, support the importance of hydrogen bonding in the structure. As a solid at −20 C the complex is stable for several weeks, but in room temperature it decomposes within 24 h. The most recent HNO complexes reported were identified during a reaction of NO with five-coordinate metal hydride, MHCl(CO)(Pi Pr3 )2 (M = Ru, Os). Initially, an intermediate nitrosyl hydride is formed, MHCl(NO)(CO)(Pi Pr3 )2 , which undergoes a bimolecular H-atom transfer to give equal molar mixtures of the sixcoordinate HNO adduct, MHCl(HNO)(CO)(Pi Pr3 )2 , and the five-coordinate nitrosyl,

Bioinorganic Chemistry of the HNO Ligand

455

MCl(NO)(CO)(Pi Pr3 )2 . The 1 H NMR of RuHCl(HNO)(CO)(Pi Pr3 )2 shows the nitroxyl proton at 20.9 ppm, but at −20 C the peak appears as a doublet (JHH 9.8 Hz), due to coupling between the trans nitroxyl and hydride protons [159,160].

12.3. Bonding in Metallo-HNO Complexes Only three HNO complexes have been crystallographically characterized, OsCl2 (CO)(HNO)(PPh3 )2 [154], [Ru(HNO)-(‘pybu S4 ’)] [158] and IrHCl2 (HNO)(PPh3 )2 [157]. A comparison of their bonding parameters are given in Table 3. In these terminal HNO adducts, a bent M N O angle is seen between 130–137 and, as expected, the N O bond of metal-bound HNO is longer and the N H bond shorter than free HNO. The low energy NO (1358–1463 cm−1 ) and a diamagnetic 1 H NMR signal (ca. 20 ppm) are the most diagnostic features of such complexes (Table 2). In several cases, the propensity of the nitroxyl oxygen to hydrogen bond is seen and this may affect the hybridization of the nitroxyl N. Several series of Re complexes allow for direct comparisons between nitroxyls and related compounds. Two compounds, of the form Re(X)(Cl)(NO)(HNO)(PPh3 )2 (X = Cl, I), have both NO and HNO on the same metal center; these exhibits (NO) stretches for the HNO and NO at 1370, 1720 (X = Cl) and 1380, 1745 (X = I) cm−1 respectively [153]. Further comparison can be made between [Re(CO)3 (HNO)(PPh3 )2 ]+ , [Re(CO)3 (NH2 OH)(PPh3 )2 ]+ and [Re(CO)3 (MeNO)(PPh3 )2 ]+ . The HNO adduct has

NO at 1391 cm−1 , which is comparable to that of the MeNO adduct at 1378 cm−1 [157]. The comparative strength of HNO as a -acceptor is demonstrated in the observed increases in the CO stretching from the pure  donor NH2 OH (2061, 1966 and 1926 cm−1 ), to HNO (2078, 2001 and 1967 cm−1 ), to MeNO (2082, 2007 and 1977 cm−1 ). Strong -backbonding is further supported by a MLCT at 418 nm in Re(Cl)(CO)2 (HNO)(PPh3 )2 as well as by theoretical modeling [156,157,161]. The effect of spectator ligands is seen by a comparison to Re(Cl)(CO)2 (HNO)(PR)2 where R equals phenyl (Ph) or cyclohexyl (Cy). The stronger sigma donor PCy3 , resulted in a 40 cm−1 lowering of the NO energy compared to that of the PPh3 substituted compound, while only a ca. 7 cm−1 lowering was observed for the CO; thus HNO dominates the backbonding on the more electron rich metal center. Both complexes were synthesized by HCl addition, but only the Cy3 was isolated as a pure complex. Both complexes were only moderately stable over time in chloromethane [153,161].

Table 3. Structural parameters of free and complexed HNO Compound Modeling free HNOa Free HNO (gas)b OsCl2 (CO)(HNO)(PPh3 )2 c cis,trans-IrHCl2 (HNO)(PPh3 )2 d RuHNO − ‘ pybu S4’ e a

[161]; b [25]; c [154]; d [157]; e [158].

<MNO

136.9(6) 129.8(7) 130.0(6)



d(MN) Å

d(NO) Å

d(NH) Å

1.915(6) 1.879(7) 1.875(7)

1.199 1.211(6) 1.193(7) 1.235(11) 1.242(9)

1.069 1.062(8) 0.94(11)

456

F. Sulc and P.J. Farmer

13. SURVEY OF REACTIVITY 13.1. Redox and Protonation Equilibia The importance of protonation in the reductive transformations of metal-bound nitrosyls is apparent in the initial reports of the conversion of Ir(NO)(PPh3 )3 to IrCl3 (NH2 OH)(PPh3 )2 upon addition of excess HCl [152]. As subsequently described, reversible conversions may be effected between the NO, HNO and NH2 OH adducts, with corresponding oxidations of the metal center and a hydride ligand (Fig. 17) [155]. Thus sequential protonation induces rehybridization and reduction of the coordinated ligand, from sp (2 NO) to sp2 (HNO) to sp3 (H2 NOH). There is a strong dependence on the counterion in the formation of HNO adducts that are formed by addition of HX to five-coordinate nitrosyl complexes. For example, the substitution of HO3 SCF3 for HCl in the reaction with Re(CO)2 (NO)(PR)2 results in protonation of the metal center, giving HRe(CO)2 (NO)(PR)2 with the hydride trans to the nitrosyl; subsequent addition of chloride did generate the HNO complex [161]. Similarly, reaction of Re(Cl)(CO)2 (HNO)(PPh3 )2 with strong BrØnsted bases (pKb < 69) resulted in dehydrohalogenation, loss of HX, not deprotonation. In contrast, HNO adducts formed from six-coordinate nitrosyl complexes are resistant to deprotonation, suggesting strong N H bonding. The characteristic 1 H NMR signal of Mb HNO in purified samples does not lose intensity from pH 6 to 10 [130]; and attempts to deprotonate [Ru(HNO)-(‘pybu S4 ’)] with BrØnsted bases were unsuccessful [158]. Both compounds do undergo D+ /H+ exchange, but the mechanism is not directly apparent. In these latter complexes, the HNO ligand may best be considered as hydride or hydrogen atom sources rather than proton donors. For instance, two are generated from hydridic reactions, for example, reaction of NaBH4 with [Ru(NO)(‘pybu S4 ’)]+ [158], or with the FeIII NO of P450nor [99]. The hydridic character of the HNO ligand is apparent in the reaction of [Ru(HNO)-(‘pybu S4 ’)] with BrØnsted acids HX (X = H2 PO− 3, bu + Br− , CF3 SO− ) which results in a two-electron oxidation to [Ru(NO)(‘py S ’)] , pre4 3 sumably with H2 evolution. Similarly, electrochemically or photochemically generated Fe(porphyrin)(NO− ) complexes typically react with protic impurities to regenerate the ferrous nitrosyl adducts, an overall one-electron oxidation [106,113,117]. Mb HNO does not decompose via protonation, but it is a powerful one-electron reductant and may be oxidized to Mb NO by stoichiometric reaction with methyl viologen [11]. Thus, the binding of the HNO to a ferrous heme may allow for its more

H O

O N

N H Ph3P

Ir

PPh3 Cl

+HCl –HCl

H Ph3P

Ir Cl

HO

H PPh3 Cl

N +HCl –HCl

Cl Ph3P

Ir

H PPh3 Cl

Cl

Fig. 17. Addition of excess HCl to Ir(NO)(PPh3 )3 showing conversion between NO, HNO and NH2 OH adducts.

Bioinorganic Chemistry of the HNO Ligand

457

rapid oxidation to NO. This conversion has been suggested as an alternative reason de vive for Cu, Zn superoxide reductases, which can quite efficiently oxidize free HNO generated in biological systems [8]. This is noteworthy, as free HNO is a kinetically poor reductant, and cannot reduce methyl viologen under equivalent conditions [29]. The interconversion of [Ru(HNO)-(‘pybu S4 ’)] ( (NO) at 1379 cm−1 ) to [Ru(NO)(‘pybu S4 ’)] (1610 cm−1 ) at −430 mV NHE, and then to [Ru(NO) (‘pybu S4 ’)]+ (1877 cm−1 ) at 515 mV NHE are electrochemically reversible, and could be effected in bulk solution by sequential oxidations of the HNO adduct by ferrocenium [158]. Likewise, Mb HNO can be obtained by either bulk or electrochemical reduction. Surprisingly, the reduction of Mb NO at −630 mV NHE is pH-independent; the slow, pH-dependent reaction that follows affects only reversibility, as seen in Fig. 10 [121]. Thus, the reduction and protonation steps must be well separated kinetically, perhaps due to electronic reorganizations induced by protonation.

13.2. Reactivity with Small Molecules As free HNO is a highly reactive and short-lived in solution, metal-HNO complexes are likewise highly reactive. All known metal-HNO complexes are air-sensitive, for instance, solutions of Mb HNO are stable for months under N2 atmosphere, but decompose upon exposure to air over a period of minutes to generate Mb FeIII . The proposed nitroxyl intermediate of P450nor reacts rapidly with NO. Mb HNO reacts under an NO atmosphere to give N2 O and Mb NO over a period of minutes [12]; nitrite also reacts with Mb HNO to generate Mb NO at a pH dependent rate [130,132]. RuHCl(HNO)(CO)(Pi Pr3 )2 is also reactive with NO; when excess NO is used during the synthesis from RuHCl(CO)(Pi Pr3 )2 , the only observed product is the RuCl(NO)(CO)(Pi Pr3 )2 [159]. When substoichiometric NO is used, equal molar amounts of RuHCl(HNO)(CO)(Pi Pr3 )2 and RuCl(NO)(CO)(Pi Pr3 )2 are produced [160]. The HNO adduct is stable for ∼12 h, but this time is dramatically shortened by the presence of NO, yielding the same products as generated from N N coupling reactions of free HNO with NO in solution. Products of the reaction of the HNO complex with NO are the same as those for HNO in solution (Eq. 35) [32,33]. HNO + 2NO → HONO + N2 O

(35)

13.3. Dissociation or Displacement Reactions There is evidence that metal-bound nitroxyls may be sources of free HNO by dissociation or displacement reactions. The initial reports of Roper demonstrated that CO can displace the HNO ligand in OsCl2 (CO)(HNO)(PPh3 )2 , giving the dicarbonyl OsCl2 (CO)2 (PPh3 )2 [152]. Likewise, the cationic complex [Re(CO)3 (HNO)(PPh3 )2 ]+ reacted with (t-But)4 NBr to give ReBr(CO)3 (PPh3 )2 [156], and when NOPF6 or NOBF4 were used in the synthesis of [Re(HNO)(CO)3 (PPh3 )2 ]+ , the adduct was unstable and decomposed to the neutral fluoride complex ReF(CO)3 (PPh3 )2 [157]. Likewise, the HNO ligand in [HNO-Ru(ttp)(1-MeIm)] may be displaced by nitrosobenzene to give [Ru(ttp)(PhNO)(1-MeIm)] in high yield [147].

458

F. Sulc and P.J. Farmer

In contrast, the stability of Mb HNO is unaffected by CO and as yet no evidence suggests Mb HNO as a source of free HNO. The binding of HNO to deoxy Mb is essentially irreversible; based on the long-lived stability of purified Mb HNO, the affinity of Mb FeII for HNO is much greater than that of its native substrate dioxygen, and is likely on the same order as that for NO.

CONCLUSIONS The unusual characteristics of HNO as a molecule and a ligand are as yet not fully understood, and as seen there remains uncertainty about many aspects of its generation and reactivity. There remains a desperate need for better HNO donors, for example, precursors that produce HNO at neutral pH without reactive byproducts, as well as sensitive and specific detection techniques for nitroxyl both in vitro and in vivo. Likewise, interference from other NOx species remains to be a major obstacle in defining the effect of HNO in physiology. Several questions remain about the possible formation of HNO or NO− adducts in the NiR and NoR enzymes: Is the single electron reduction of a ferrous nitrosyls a viable route to a ferrous nitroxyl? If only by hydridic mechanisms as proposed (but disputed) [162] in P450nor, how are such oxidation states achieved in the NiR enzymes? Are the resulting species best described as HNO FeII or H2 NO FeIII ? Intimately linked to such questions are the many different ways that a protein may control the flux of electrons and protons to the metal-bound substrate. Likewise, the physiological effects of HNO-precursors are tied to the reactions of HNO with metalloproteins, and therefore further study of the synthesis and reactivity of metal-HNO complexes will be of high significance. In this regard, Mb HNO is unique as an isolable heme-HNO complex, and also as the only complex formed by direct trapping of free HNO. With the renewed interest in HNO as well as several new small molecule complexes in hand, the complex reactivity of metallo-nitroxyl complexes will be revealed which, if comparison holds to that of the Fischer carbenes, will be rich indeed.

REFERENCES [1] Thomas, D.D., Miranda, K.M., Colton, C.A. et al. (2003) Antioxidants & Redox Signaling, 5, 307–317. [2] Stamler, J.S., Singel, D.J. and Loscalzo, J. (1992) Science, 258, 1898–1902. [3] Bonner, F.T. and Stedman, G. (1996) The chemistry of nitric oxide and redox-related species. In Methods Nitric Oxide Res (M. Feelisch and J.S. Stamler, eds). Chichester (UK): Wiley, pp. 3–8. [4] Hughes, M.N. (1992) Biochim. Biophys. Acta., 1411, 263–272. [5] Rusche, K.M., Spiering, M.M. and Marletta, M.A. (1998) Biochem., 37, 15503–15512. [6] Averill, B.A. (1996) Chem. Rev., 96, 2951–2964. [7] Murphy, M.E. and Sies, H. (1991) Proc. Natl. Acad. Sci. USA., 88, 10860–10864. [8] Liochev, S.I. and Fridovich, I. (2002) Arch. Biochem. Biophys., 402, 166–171. [9] Liochev, S.I. and Fridovich, I. (2003) Free Rad. Biol. Med., 34, 1399–1404.

Bioinorganic Chemistry of the HNO Ligand

459

[10] Miranda, K.M., Paolocci, N., Katori, T. et al. (2003) Proc. Natl. Acad. Sci. USA., 100, 9196–9201. [11] Lin, R. and Farmer, P.J. (2000) J. Am. Chem. Soc., 122, 2393–2394. [12] Lin, R. (2001) Dissertation. Irvine: University of California. [13] Shafirovich, V. and Lymar, S.V. (2002) Proc. Natl. Acad. Sci. USA., 99, 7340–7345. [14] Angeli, A. (1903) Gazz. Chim. Ital., 33, 245. [15] Hughes, M.N. and Cammack, R. (1999) Methods Enzymol., 301, 279–287. [16] King, S.B. and Nagasawa, T. (1999) Meth. Enzym., 301, 211–220. [17] Wink, D.A. and Feelisch, M. (1996) Formation and detection of nitroxyl and nitrous oxide. In Methods Nitric Oxide Res. (M. Feelisch and J.S. Stamler, eds). Chichester (UK): Wiley, pp. 403–412. [18] Shoeman, D.W. and Nagasawa, H.T. (1998) Nitric Oxide Biol. Chem., 2, 66–72. [19] Pino, R.Z. and Feelisch, M. (1994) Biochem. Biophys. Res. Commun., 201, 54–62. [20] Raschig, F. (1928) Angew. Chem., 41, 207. [21] Andrussov, L. (1928) Angew. Chem., 41, 262–263. [22] Kohout, F.C. and Lampe, F.W. (1965) J. Am. Chem. Soc., 87, 5795–5796. [23] Jacox, M.E. and Milligan, D.E.J. (1973) Mol. Spectrosc., 48, 536–559. [24] Brown, H.W. and Pimentel, G.C. (1958) J. Chem. Phys., 29, 883–888. [25] Dalby, F.W. (1958) Can. J. Phys., 36, 1336–1371. [26] Clough, P.N., Thrush, B.A., Ramsay, D.A. and Stamper, J.G. (1973) Chem. Phys. Lett., 23, 155–156. [27] Johns, J.W.C. and McKellar, A.R.W. (1977) J. Chem. Phys., 66, 1217–1224. [28] Anderson, W.R. (1999) Combust. Flame, 117, 394–403. [29] Bartberger, M.D., Liu, W., Ford, E., et al. (2002) Proc. Natl. Acad. Sci. USA, 99, 10958–10963. [30] Benderskii, V.A., Krivenko, A.G. and Ponomarev, E.A. (1990) Elektrokhimiya, 26, 318–325. [31] Stanbury, D.M. (1989) Reduction potentials involving inorganic free radicals in aqueous solution. In Advances in Inorganic Chemistry, Vol. 33 (A.G. Sykes, ed.). Orlando: Academic Press, Inc., pp. 69–138. [32] Graetzel, M., Taniguchi, S. and Henglein, A. (1970) Berichte der Bunsen-Gesellschaft, 74, 1003–1010. [33] Shafirovich, V. and Lymar, S.V. (2003) J. Am. Chem. Soc., 125, 6547–6552. [34] Ruud, K., Helgaker, T. and Uggerud, E. (1997) J. Mol. Struc (Theochem), 393, 59–71. [35] He, Y., Sanders, W.A. and Lin, M.C. (1988) J. Phys. Chem., 92, 5474–5481. [36] Bryukov, M.G., Kachanov, A.A., Timonnen, R., Seetula, J. et al. (1993) Chem. Phys. Lett., 208, 392–398. [37] Bazylinski, D.A. and Hollocher, T.C. (1985) Inorg. Chem., 24, 4285–4288. [38] Wong, P.S.Y., Hyun, J., Fukuto, J.M., Shiroda, F.N. et al. (1998) Biochemistry, 37, 5362–5371. [39] Arnelle, D.R. and Stamler, J.S. (1995) Arch. Biochem. Biophys., 318, 279–285. [40] Fukuto, J.M., Bartberger, M.D., Dutton, A.S., Paolocci, N. et al. (2005) Chem. Res. Toxicol., 18, 790–801. [41] Lee, M.J.C., Nagasawa, H.T., Elberling, J.A. and DeMaster, E.G. (1992) J. Med. Chem., 35, 3648–3652. [42] DeMaster, E.G., Redfern, B. and Nagasawa, H.T. (1998) Biochem. Pharmacol., 56, 2007–2015. [43] Nagasawa, H.T., DeMaster, E.G., Redfern, B. et al. (1990) J. Med. Chem., 33, 3120–3122. [44] Wink, D.A., Feelisch, M., Fukuto, J.M. et al. (1998) Arch. Biochem. Biophys., 351, 66–74. [45] Schafer, F.Q. and Buettner, G.R. (2001) Free Radic. Biol. Med., 30, 1191–1212. [46] Cook, N.M., Shinyashiki, M., Jackson, M.I. et al. (2003) Ace1. Arch. Biochem. Biophys., 410, 89–95.

460

F. Sulc and P.J. Farmer

[47] Shinyashiki, M., Pan, C.-J.P., Switzer, C.H. and Fukuto, J.M. (2001) Chem. Res. Toxicol., 14, 1584–1589. [48] Kim, W.-K., Choi, Y.-B., Rayudu, P.V., Das, P. et al. (1999) Neuron, 24, 461–469. [49] Colton, C.A., Gbadegesin, M., Wink, D.A., et al. (2001) J. Neurochem., 78, 1126–1134. [50] Fukuto, J.M., Chiang, K., Hszieh, R. et al. (1992) J. Pharmacol. Exp. Ther., 263, 546–551. [51] Fukuto, J.M., Switzer, C.H., Miranda, K.M. and Wink, D.A. (2005) Annu. Rev. Pharmacol. Toxicol., 45, 335–355. [52] Hobbs, A.J. (1997) Trends Pharmacol. Sci., 18, 484–491. [53] Dierks, E.A. and Burstyn, J.N. (1996) Biochem. Pharmacol., 51, 1593–1600. [54] Fukuto, J.M., Hobbs, A.J. and Ignarro, L.J. (1993) Biochem. Biophys. Res. Commun., 196, 707–713. [55] Paolocci, N., Saavedra, W.F., Miranda, K.M. et al. (2001) Proc. Natl. Acad. Sci. USA, 98, 10463–10468. [56] Ma, X.L., Cao, F., Liu, G.L. et al. (1999) Proc. Natl. Acad. Sci. USA, 96, 14617–14622. [57] Pagliaro, P., Mancardi, D., Rastaldo, R. et al. (2003) Free Radic. Biol. Med, 34, 33–43. [58] Foresti, R., Bains, S., Sulc, F., Farmer, P.J., Green, C.J., Motterlini, R. (2006) J. Pharmacol. Exp. Ther. 317, 1125–1133. [59] Cosby, K., Partovi, K.S., Crawford, J.H. et al. (2003) Nature Med., 9, 1498–1505. [60] Enemark, J.H. and Feltham, R.D. (1974) J. Am. Chem. Soc., 96, 5002–5004. [61] Westcott, B.L. and Enemark, J.H. (1999) Transition metal nitrosyls. In Inorganic Electronic Structure and Spectroscopy, Vol. II (E.I. Solomon and A.B.P. Lever, eds). New York: John Wiley & Sons, Inc. [62] Ford, P.C. and Lorkovic, I.M. (2002) Chem. Rev., 102, 993–1017. [63] Piciulo, P.L., Rupprecht, G. and Scheidt, W.R. (1974) J. Am. Chem. Soc., 96, 5293–5295. [64] Scheidt, W.R. and Hoard, J.L. (1973) J. Am. Chem. Soc., 95, 8281–8288. [65] Scheidt, W.R., Lee, Y.J. and Hatano, K. (1984) J. Am. Chem. Soc., 106, 3191–3198. [66] Liochev, S.I. and Fridovich, I. (2001) J. Biol. Chem., 276, 35253–35257. [67] Vega, J.M., Cardenas, J. and Losada, M. (1980) Meth. Enzym., 69, 255–270. [68] Fujita, E. and Fajer, J. (1983) J. Am. Chem. Soc., 105, 6743–6745. [69] Krueger, R.J. and Siegel, L.M. (1982) Amer. Chem. Soc., 21, 2892–2904. [70] Lancaster, J.R., Vega, J.M., Kamin, H. et al. (1979) J. Biol. Chem., 254, 1268–1272. [71] Mikami, B. and Ida, S. (1989) J. Biochem., 105, 47–50. [72] Christner, J.A., Munck, E., Kent, T.A. et al. (1984) J. Am. Chem. Soc., 106, 6786–6794. [73] Christner, J.A., Munck, E., Janick, P.A. and Siegel, L.M. (1981) J. Biol. Chem., 256, 2098–2101. [74] Wilkerson, J.O., Janick, P.A. and Siegel, L.M. (1983) Amer. Chem. Soc., 22, 5048–5054. [75] Lui, S.M., Soriano, A. and Cowan, J.A. (1993) J. Amer. Chem. Soc., 115, 10483–10486. [76] Lui, S.M., Liang, W., Soriano, A. and Cowan, J.A. (1994) J. Amer. Chem. Soc., 116, 4531–4536. [77] Lui, S.M. and Cowan, J.A. (1994) J. Amer. Chem. Soc., 116, 11538–11549. [78] Crane, B.R., Siegel, L.M. and Getzoff, E.D. (1997) Biochemistry, 36, 12101–12137. [79] Copeland, D.M., West, A.H. and Richter-Addo, G.B. (2003) Proteins: Structure, Function, and Genetics, 53, 182–192. [80] Zemojtel, T., Rini, M., Heyne, K. et al. (2004) J. Amer. Chem. Soc., 126, 1930–1931. [81] Simon, J. (2002) FEMS Microbiol. Rev., 26, 285–309. [82] Einsle, O., Messerschmidt, A., Huber, R. et al. (2002) J. Amer. Chem. Soc., 124, 11737–11745. [83] Einsle, O., Messerschmidt, A., Stach, P. et al. (1999) Nature (London), 400, 476–480. [84] Rudolf, M., Einsle, O., Neese, F. and Kroneck, P.M.H. (2002) Biochem. Soc. Trans., 30, 649–653. [85] Silaghi-Dumitrescu, R. (2003) Eur. J. Inorg. Chem., 6, 1048–1052.

Bioinorganic Chemistry of the HNO Ligand [86] [87] [88] [89] [90] [91] [92] [93] [94] [95] [96] [97] [98] [99] [100] [101] [102] [103] [104]

[105] [106] [107] [108] [109] [110] [111] [112] [113] [114] [115] [116] [117] [118] [119] [120] [121] [122] [123] [124] [125] [126]

461

Garber, E.A.E. and Hollocher, T.C. (1982) J. Biol. Chem., 257, 4705–4708. Garber, E.A.E., Wehrli, S. and Hollocher, T.C. (1983) J. Biol. Chem., 258, 3587–3591. Kim, C.H. and Hollocher, T.C. (1983) J. Biol. Chem., 258, 4861–4863. Farmer, P.J., Lin, R. and Bayachou, M. (1998) Comm. Inorg. Chem., 20, 101–120. Hendriks, J., Warne, A., Gohlke, U. et al. (1998) Biochemistry, 37, 13102–13109. Zumft, W.G. (1997) Microbiol. Mol. Biol. Rev., 61, 533–616. Moënne-Loccoz, P. and de Vries, S. (1998) J. Am. Chem. Soc., 120, 5147–5152. Cheesman, M.R., Zumft, W.G. and Thomson, A.J. (1998) Biochemistry, 37, 3994–4000. Ju, T.D., Woods, A.S., Cotter, R.J. et al. (2000) Inorganica Chimica Acta., 297, 362–372. Moeenne-Loccoz, P., Richter, O.-M.H., Huang, H.-W. et al. (2000) J. Am. Chem. Soc., 122, 9344–9345. Shiro, Y., Fujii, M., Iizuka, T. et al. (1995) J. Biol. Chem., 270, 1617–1623. Shiro, Y., Fujii, M., Isogai, Y. et al. (1995) Biochemistry, 34, 9052–9058. Nakahara, K., Tanimoto, T., Hatano, K. et al. (1993) J. Biol. Chem., 268, 8350–8355. Daiber, A., Nauser, T., Takaya, N. et al. (2002) J. Inorg. Biol. Chem., 88, 343–352. http://www.lanl.gov/projects/nitrate/Other.htm Kalu, E.E., White, R.E. and Hobbs, D.T. (1996) J. Electrochm. Soc., 143, 3094–3099. Immoos, C.E. (2003) Dissertation. Irvine: University of California. Kim-Shapiro, D., Schechter, A.N., Gladwin, M.T. (2006) Arteriosclerosis, Thrombosis, and Vascular Biology 26, 697–705. Blair, E., Sulc, F. and Farmer, P.J. Biomimetic NOx reductions by Heme models and proteins. In N4 Macrocyclic Metal Complexes, J.H. Zagal, F. Bedioui, J.P. Dodelet eds. Springer. 2006, Chp.4. pp. 149–190. Callahan, R.W. and Meyer, T.J. (1977) Inorg. Chem., 16, 574–581. Murphy, W.R. Jr., Takeuchi, K.J. and Meyer, T.J. (1982) J. Am. Chem. Soc., 104, 5817–5819. Olson, L.W., Schaeper, D., Lancon, D. and Kadish, K.M. (1982) J. Am. Chem. Soc., 104, 2042–2044. Lancon, D. and Kadish, K.M. (1983) J. Am. Chem. Soc., 105, 5610–5617. Liu, Y. and Ryan, M.D. (1994) Inorg. Chim. Acta, 225, 57–66. Liu, Y. and Ryan, M.D. (1994) J. Electroanal. Chem., 368, 209–219. Choi, I.-K., Liu, Y., Feng, D. et al. (1991) Inorg. Chem., 30, 1832–1839. Choi, I.K. and Ryan, M.D. (1992) New J. Chem., 16, 591–597. Liu, Y., DeSilva, C. and Ryan, M.D. (1997) Inorg. Chim. Acta, 258, 247–255. Barley, M.H. and Meyer, T.J. (1986) J. Am. Chem. Soc., 108, 5876–5885. Barley, M.H., Takeuchi, K.J., Murphy, W.R. and Meyer, T.J.J. (1985) Chem. Soc. Chem. Comm., 8, 507–508. Barley, M.H., Rhodes, M.R. and Meyer, T.J. (1987) Inorg. Chem., 26, 1746–1750. Younathan, J.N., Wood, K.S. and Meyer, T.J. (1992) Inorg. Chem., 31, 3280–3285. Seki, H., Hoshino, M. and Kounose, S. (1996) J. Chem. Soc., Faraday Trans., 92, 2579–2583. Rusling, J.F. and Nassar, A.E.F. (1993) J. Am. Chem. Soc., 115, 11891–11897. Nassar, A.-E., Zhang, Z., Chynwat, V. et al. (1995) J. Phys. Chem., 99, 11013–11017. Lin, R., Bayachou, M. and Farmer, P.J. (1997) J. Am. Chem. Soc., 119, 12689–12690. Bayachou, M., Lin, R., Cho, W. and Farmer, P.J. (1998) J. Am. Chem. Soc., 120, 9888–9893. Immoos, C.E., Chou, J., Bayachou, M., Blair, E. and Farmer, P.J. (2004) J. Am. Chem. Soc., 126, 4934–4942. Doyle, M.P. and Mahapatro, S.N. (1984) J. Am. Chem. Soc., 106, 3678–3679. Bazylinski, D.A. and Hollocher, T.C. (1985) J. Am. Chem. Soc., 107, 7982–7986. Bazylinski, D.A., Goretshi, J. and Hollocher, T.C. (1985) J. Am. Chem. Soc., 107, 7986–7989.

462

F. Sulc and P.J. Farmer

[127] Doyle, M.P., Mahapatro, S.N., Broene, R.D. and Guy, J.K. (1988) J. Am. Chem. Soc., 110, 593–599. [128] Miranda, K.M., Nims, R.W., Thomas, D.D. et al. (2003) J. Inorg. Biochem., 93, 52–60. [129] Gow, A.J. and Stamler, J.S. (1998) Nature, 391, 169–173. [130] Sulc, F., Immoos, C.E., Pervitsky, D. and Farmer, P.J. (2004) J. Am. Chem. Soc., 126, 1096–1101. [131] Chad, C.E. (2002) Dissertation. Irvine: University of California. [132] Sulc, F. (2006) Dissertation. Irvine: University of California. [133] Immoos, C.E., Sulc, F., Farmer, P.J. et al. (2005) J. Am. Chem. Soc., 127, 814–815. [134] Rich, A.M., Armstrong, R.S., Ellis, P.J. and Lay, P.A. (1998) J. Am. Chem. Soc., 120, 10827–10836. [135] Vogel, K.M., Kozlowski, P.M., Zgierski, M.Z. and Spiro, T.G. (1999) J. Am. Chem. Soc., 121, 9915–9921. [136] Sulc, F., Ma, D., Fleisher et al. (2003) J. Biol. Inorg. Chem., 8, 348–352. [137] Lee, J., Chen, L., West, A.H. and Richter-Addo, G.B. (2002) Chem. Rev., 102, 1019–1065. [138] Richter-Addo, G.B. (1999) Acct. of Chem. Res. 32, 529–536. [139] Mansuy, D., Gans, P., Chottard, J.C. and Bartoli, J.F. (1977) Euro. J. Biochem., 76, 607–615. [140] Mansuy, D., Gans, P., Chottard, J.C. and Chottard, G. (1977) Euro. J. Biochem., 76, 617–623. [141] Mansuy, D., Battioni, P., Chottard, J.C. et al. (1983) J. Amer. Chem. Soc., 105, 455–463. [142] Wang, L.-S., Chen, L., Khan, M.A. and Richter-Addo, G.B. (1996) Chem. Comm., 323–324. [143] Sohl, C.D., Lee, J., Alguindigue, S.S. et al. (2004) J. Inorg. Biochem., 98, 1238–1246. [144] Lee, J., Twamley, B. and Richter-Addo, G.B. (2002) Can. J. Chem., 80, 1252–1258. [145] Lee, J., Twamley, B. and Richter-Addo, G.B. (2004) Dalton Trans., 98, 189–196. [146] Chen, L., Khan, M.A., Richter-Addo et al. (1998) Inorg. Chem., 37, 4689–4696. [147] Lee, J. and Richter-Addo, G.B. (2004) J. Inorg. Biochem., 98, 1247–1250. [148] Enemark, J.H., Feltham, R.D., Riker-Nappier, J. and Bizot, K.F. (1975) Inorg. Chem., 14, 624–632. [149] Rabinowitz, H.N., Karlin, K.D. and Lippard, S.J. (1977) J. Am. Chem. Soc., 99, 1420–1426. [150] Paul, P.P., Tyeklár, Z., Farooq, A. et al. (1990) J. Am. Chem., 112, 2430–2432. [151] Fjeldsted, D.O.K., Stobart, S.R. and Zavorotko, M.K. (1985) J. Am. Chem. Soc., 107, 8258–8259. [152] Roper, W.R., Grundy, K.R. and Reed, C.A. (1970) J. Chem. Soc. [Section] D: Chem. Comm., 22, 1501–1502. [153] LaMonica, G., Freni, M. and Cenini, S.J. (1974) Organomet. Chem., 71, 57–64. [154] Wilson, R.D. and Ibers, J.A. (1979) Inorg. Chem., 18, 336–343. [155] Melenkivitz, R. and Hillhouse, G.L. (2002) Chem. Comm., 6, 660–661. [156] Southern, J.S., Green, M.T., Hillhouse, G.L. et al. (2001) Inorg. Chem., 40, 6039–6046. [157] Melenkivitz, R., Southern, J.S., Hillhouse, G.L. et al. (2002) J. Am. Chem. Soc., 124, 12068–12069. [158] Sellmann, D., Gottschalk-Gaudig, T., Haussinger, D. et al. (2001) Chem. Eur. J., 7, 2099–2103. [159] Marchenko, A.V., Vedernikov, A.N., Dye, D.F. et al. (2002) Inorg. Chem., 41, 4087–4089. [160] Marchenko, A.V., Vedernikov, A.N., Dye, D.F. et al. (2004) Inorg. Chem., 43, 351–360. [161] Southern, J.S., Hillhouse, G.L. and Rheingold, A.L. (1997) J. Am. Chem. Soc., 119, 12406–12407. [162] Miller, L.M., Pedraza, A.J. and Chance, M.R. (1997) Biochemistry, 36, 12199–12207.

Part V Selected Enzymes and Sensors

This page intentionally left blank

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 17

Ligand-Protein Interactions in Mammalian Nitric Oxide Synthase Denis L. Rousseaua , David Lia , Eric Y. Haydena , Haiteng Dengb and Syun-Ru Yeha a

Department of Physiology and Biophysics, Albert Einstein College of Medicine, Bronx, NY 10461 b Proteomics Resource Center, Rockefeller University, New York, NY 10021

Abstract Nitric oxide synthases (NOSs) are heme proteins that catalyze the formation of nitric oxide (NO) from l-arginine and oxygen in a sequential two-step process. Three structurally similar isoforms have been identified that deliver NO to different tissues for specific functions. An understanding of the interactions of ligands with the protein is essential to determine the mechanism of catalysis, the design of inhibitors and the differential auto-inhibitory regulation of the enzymatic activity of the isoforms due to the binding of NO to the heme. Ligand-protein interactions in the NOS enzymes revealed by resonance Raman scattering studies are reviewed in this article. For consistency, most of the examples are drawn from studies of the iNOS oxygenase domain (iNOSoxy ). The CO-related modes in the CO-bound ferrous enzyme are sensitive to the presence of substrate, either l-arginine or N-hydroxy-l-arginine, in the distal pocket, but insensitive to the presence of the tetrahydrobiopterin (H4B) cofactor. In contrast, when NO is coordinated to the ferric heme, the NO is sensitive to the substrate only when H4B is present. Furthermore, in the NO-bound ferric enzyme, the addition of H4B induces a large heme distortion that may modulate heme reduction and thereby regulate the NO auto-inhibitory process. It has also been found that the exposure of the H4B-free enzyme to NO induces the dissociation of the dimer to monomers followed by the reduction of the heme iron with the concurrent breakage of the proximal iron–thiolate bond, leading to the formation of a five-coordinate NO-bound ferrous species. Mass spectrometry revealed that NO-induced monomerization of the enzyme is associated with an intramolecular disulfide bond formation between Cys104 and Cys109 located in the Zn-binding motif. Resonance Raman studies of ligand interactions of NOSs have expanded our understanding of the mechanistic features of this important family of enzymes.

ABBREVIATIONS Fe C Fe Fe Fe

CO O C

O

NO N

O

The The The The The

iron–carbon monoxide stretching mode. carbon–oxygen stretching mode. iron–carbon–oxygen bending mode. iron–nitric oxide stretching mode. iron–nitrogen–oxygen bending mode.

466

Cam CPO FAD FMN FTIR H4B HbA l-Arg NADPH NOHA NOS iNOS, eNOS, nNOS NOSoxy iNOSoxy , eNOSoxy , nNOSoxy iNOSFL eNOSFL nNOSFL NSD P450cam PDB PAR

D.L. Rousseau et al.

Camphor. Chloroperoxidase. Flavin adenine dinucleotide. Flavin mononucleotide. Fourier Transform Infrared. Tetrahydrobiopterin. Human adult hemoglobin. l-arginine. Nicotinamide adenine dinucleotide phosphate. N -hydroxy-l-arginine. Nitric oxide synthase. Inducible, endothelial & neuronal nitric oxide synthase, respectively. The oxygenase domain of nitric oxide synthase. The oxygenase domain of inducible, endothelial & neuronal nitric oxide synthase, respectively. The full-length enzymes of inducible, endothelial & neuronal nitric oxide synthase, respectively. Normal-coordinate structural decomposition. Cytochrome P450cam . Protein Data Bank. 4(2-Pyridylazo Resorcinol).

1. INTRODUCTION Nitric oxide synthases (NOSs) catalyze the formation of NO from oxygen and l-arginine (l-Arg). Three major isoforms, iNOS, eNOS and nNOS have been found in macrophages, endothelial cells and neuronal tissues, respectively. The NO that is produced by iNOS, eNOS and nNOS are used as a cytotoxic agent for immune responses, a regulator for vascular function and a neurotransmitter for signal transduction, respectively [1–5]. All three NOS isoforms are dimeric. Each subunit of the dimer contains two domains: a reductase domain that binds FMN, FAD and NADPH and an oxygenase domain that contains heme and tetrahydrobiopterin (H4B) [6–11]. The catalytic process occurs in the heme active site of the oxygenase domain. It involves two consecutive oxygenase reactions as illustrated in Fig. 1. In the first step, l-Arg is hydroxylated to N -hydroxyl-arginine (NOHA) and in the second step, NOHA is oxidized to citrulline and NO [12–14]. It is an oxygen atom from the molecule in the first step that is incorporated into the NO. In each reaction, the NADPH in the reductase domain is utilized as the electron source. The electron transfer from the reductase domain to the oxygenase domain, which is essential for the catalytic activity, is regulated by binding of a calcium–calmodulin complex [15–18]. When the calcium–calmodulin complex is present, electrons flow from NADPH through FMN and FAD in one subunit to the oxygenase domain of the other subunit [19].

Ligand Interactions in NOS H2N

467

NH2+

H2N

NH

H2N

N–OH NH

O=O

O NH

H2O

O=O

H2O

+ N=O ½NADPH ½NADP+

NADPH NADP+

NH3+

C O

–

O

L-arginine

C O

NH3+ O–

N-hydroxy-L-arginine

NH3+

C O

–

O

L-citrulline

Fig. 1. The two-step reaction catalyzed by nitric oxide synthases. In the first step l-Arg is hydroxylated to form N-hydroxy l-Arg with the utilization of one molecule of oxygen and two electrons from NADPH. In the second step, with one electron from NADPH and another molecule of oxygen, N-hydroxy l-Arg is converted to citrulline and nitric oxide (NO).

The crystal structures of the oxygenase domains (NOSoxy ) of all three isoforms have been determined and shown to be almost identical [20–25]. Like cytochrome P450 and chloroperoxidase (CPO), NOSs belong to the family of cysteine-coordinated heme proteins in which the proximal ligand to the heme-Fe is the sulfur atom of an intrinsic cysteine residue [26–29]. The sulfur atom of the proximal cysteine ligand accepts an H-bond from a proximal Trp residue (Fig. 2), which is very important for its function. The substrate, l-Arg, binds directly above the heme iron atom in the distal pocket, while the cofactor, H4B, binds along the side of the heme. The l-Arg and H4B are linked together through an extended H-bonding network mediated by one of the two propionate groups of the heme. In addition to these H-bonding interactions, l-Arg is stabilized by a sophisticated H-bonding network involving Trp366, Tyr367 and Glu371; and H4B is further stabilized by a -stacking interaction with the side chain group of Trp457 and an H-bond with the peptide carbonyl group of the same residue. The NO generated in NOS at the end of the catalytic cycle can either diffuse out of the heme pocket to the medium or it can bind to the heme iron thereby inhibiting the turnover of the enzyme [30,31]. The balance between these two fates for the catalytically generated NO determines the temporal release of the NO during multiple turnovers. Despite the fact that the crystal structures of the three isoforms are almost identical, each of the three NOS isoforms displays a very different degree of NO auto-inhibition [32], providing an exquisite control mechanism for the differential temporal release of NO that is critical for each unique cellular environment in which the NO is produced [30–33]. The molecular mechanism controlling the degree of NO auto-inhibition remains unclear. Thus, the understanding of the interactions between the NO, the heme-iron and the protein matrix surrounding the heme is crucial for uncovering the factors determining the isoform-specific NO auto-inhibition. In addition, non-physiological heme ligands, such as CO, are commonly used as probes for the structural as well as electronic properties of the heme active sites [34,35]. In this report, the distal ligand-protein interactions involving CO and NO in NOSs are reviewed and the implications on NOS physiological functions revealed by these studies are discussed. Most of the examples are drawn from studies of iNOSoxy .

468

D.L. Rousseau et al.

(A)

L-arginine

Tyr367

Glu371 H4B

Trp366 Heme Cys194

Trp457

Trp188

(B)

[393-404]

Heme

[393-404]

C109

C109 Heme

[77-118] C104

“swapped” conformation

[77-118] C104

“unswapped” conformation

Fig. 2. (A) The catalytic site of NOS, consisting of the heme, the tetrahydrobiopterin (H4B) cofactor and the l-Arg substrate. The proximal ligand to the heme is a cysteine. The various hydrogen bonds in the active site are indicated by the dotted lines. The structure is that of iNOSoxy and is taken from the protein data bank (PDB code:1NOD) [20]. The catalytic sites of all three isoforms are nearly identical. (B) Crystal structures of the dimeric H4B-bound iNOSoxy . (left) The swapped structure (PDB code: 1QOM) with an intermolecular disulfide bond between the C109 residue from each subunit. (right) The “unswapped” structure (PDB code: 1DF1) with a zinc (shown as the orange sphere) coordinated by C104 and C109 from each subunit. The C104 and C109 residues are labeled with ball-and-stick representation. The important peptide segments of the two subunits located in the dimer interface are shown in yellow and green separately. Those of the yellow subunit and its associate prosthetic heme group are labeled as indicated. (see Plate 14.)

Ligand Interactions in NOS

469

2. SPECTROSCOPIC PROPERTIES OF NOS 2.1. Optical Absorption Spectra The Soret transition associated with the  to  ∗ electronic transition of the porphyrin macrocycle is sensitive to the oxidation, coordination and spin states of the heme iron. In the absence of exogenous ligands, substrates and cofactor, the heme iron in the ferric derivative of NOS adopts primarily a six-coordinate low-spin configuration, possibly with a solvent water molecule bound to the distal binding site of the heme, in analogy to that observed in P450 type of proteins [36]. The Soret transition of this ferric species is located at 420 nm as shown in Fig. 3A. The addition of either l-Arg or H4B destabilizes the putative heme-bound water molecule through steric interference and structural changes to the protein matrix, respectively. As a result, the Soret transition shifts to ∼400 nm, due to a partial conversion of the heme iron to a five-coordinated high-spin configuration. When both l-Arg and H4B are present, the Soret transition further shifts to 395 nm, indicating a full conversion to the five-coordinated high-spin heme [15,34,37–39]. In the ferric oxidation state, NOS forms stable complexes with several exogenous heme ligands, such as NO and CN− . When NO or CN− is coordinated to the ferric heme iron, the Soret transition shifts to the red to ∼439 nm (Fig. 3B), which is typical for a six-coordinated low-spin ferric heme with a proximal cysteine axial ligand [37]. In contrast to the ligand-free protein, the Soret transitions of the ligand-bound derivatives are unaffected by the addition of H4B and/or l-Arg [37]. The heme iron in the ferrous derivative of NOS exhibits a five-coordinated high-spin configuration with the proximal Cys as the sole axial ligand, regardless the binding of H4B and/or l-Arg. The Soret transition of the exogenous ligand-free ferrous species (A)

(B) 0.6 400

420

Absorbance (A.U.)

iNOSoxy

439

395

437

445

0.4 Fe+2– CO (–, –) Fe+3–NO (–, –) Fe+3– CN–(–, –) Fe+2–NO (+, +)

0.2 (–, –) (+, –) (–, +) (+, +) 0.0 380

420

460

Wavelength (nm)

400

450

500

550

600

Wavelength (nm)

Fig. 3. (A) Optical absorption spectra of the ligand-free ferric form of iNOSoxy under the various substrate/cofactor bound states as indicated. The (− −), (+ −), (− +) and (+ +) stand for −Arg/−H4B, +Arg/−H4B, −Arg/+H4B and +Arg/+H4B, respectively. (B) Optical spectra of various ligand-bound complexes of iNOSoxy as indicated.

470

D.L. Rousseau et al.

is found in the 410–420 nm region [34]. The binding of CO and NO to the ferrous protein shifts the Soret transition to ∼445 nm and ∼437 nm, respectively (Fig. 3B) [33,35,37], indicating the formation of the six-coordinate low spin species. Again, the Soret transitions of the ligand-bound derivatives are unaltered by the addition of H4B and/or l-Arg [37]. However, in the absence of l-Arg and H4B, the six-coordinate NO-bound ferrous species is not stable; it rapidly converts to a five-coordinate NO-bound form due to the breakage of the proximal Fe Cys bond but this process does not occur in the presence of H4B and/or l-Arg [33,40,41].

2.2. Resonance Raman Spectra Resonance Raman spectroscopy with Soret excitation has been successfully applied to study the structural and functional relationships of heme proteins for several decades and has been demonstrated to be especially useful for studying NOSs. The high frequency region (1000–1700 cm−1 ) of the Raman spectrum is very sensitive to the oxidation and coordination states of the heme iron. In particular, the 4 vibrational mode in the 1340–1380 cm−1 region is very sensitive to the electron density on the heme macrocycle and hence is a good indicator of the oxidation state of the heme iron. The 3 vibrational mode in the 1475–1520 cm−1 region is sensitive to both the coordination and spin state of the heme iron, whereas the 2 vibrational mode in the 1560–1590 cm−1 region is sensitive to the heme spin state. On the other hand, in the low frequency region of the spectrum (200–800 cm−1 ), in addition to the modes of the heme macrocycle, iron-ligand stretching and/or bending modes can be detected. Accordingly, the specific axial ligand coordinated to the prosthetic heme group can be identified and studied. In addition, several heme out-of-plane modes may be strongly enhanced in this region of the spectrum when the prosthetic heme group is deformed from the regular planar structure [37,42,43]. The frequencies and intensities of these Raman lines are further modulated by the protein environment surrounding the heme and, therefore, provide useful structural information on heme proteins.

3. CARBON MONOXIDE AS A PROBE OF THE CATALYTIC SITE 3.1. The Fe

C

O Vibrational Modes

The use of CO as a probe for the structure and dynamics of heme proteins has been well established [44]. In CO-bound heme proteins, the Fe CO stretching mode (Fe CO ) and the Fe C O bending mode (Fe C O ) are in general detected in the 450–550 and the 560–590 cm−1 regions, respectively [45–47]. The C O stretching mode (C O ), on the other hand, is typically detected in the 1900–1970 cm−1 range. The frequencies of these modes are sensitive to the nature of the proximal ligand and to polar and/or steric interactions between the CO and intrinsic residues or exogenous substrates present in the distal pocket. In iNOSoxy , in the absence of l-Arg and H4B, the Fe CO stretching (Fe CO ) and the Fe C O bending (Fe C O ) modes are detected at 491 and 562 cm−1 , respectively (Fig. 4A) [37]. These assignments were confirmed in the isotope difference spectrum

Ligand Interactions in NOS

471

(A)

(B)

Fe+2–CO iNOSoxy

346 (a)

(C)

676 1946

508 562

491

(–, –)

693 718 803 752

(a)

563

492 469 514

(a)

556

1907

512 569

(b)

(b)

500

1897

570 579

(b) 1864

(+, –)

1944

564

(c) 491 (c)

490

508

562 (–, +)

515

(d)

569

(d)

600

1899

558 1905

497

(+, +)

500

(c)

570 579

512

400

565

468

700

Raman shift (cm–1)

(d)

1863

563

800

400

450

500

550

600

1850

1950

Raman shift (cm–1)

Fig. 4. Panel A: Resonance Raman spectra of the ferrous-CO-bound complexes of iNOSoxy in the presence and/or absence of l-Arg and H4B. The (− −), (+ −), (− +) and (+ +) stand for −Arg/−H4B, +Arg/−H4B, −Arg/+H4B and +Arg/+H4B, respectively; Panel B: Isotopic difference spectra (12 C16 O–12 C18 O) in the Fe CO spectral region; Panel C: Isotope difference spectrum (12 C16 O–12 C18 O) in the C O spectral region. Adapted from [37].

shown in Fig. 4B [37]. The broad width of the Fe CO mode, which is also evident in the difference spectrum, is attributed to structural inhomogeneity of the Fe CO moiety. It can be deconvoluted into two Gaussian peaks centered at 482 and 502 cm−1 with equal widths of 26 cm−1 as shown in Fig. 5A (top left) [48]. The data suggest the presence of two distinct conformations. This is consistent with the width of the C O stretching mode (C O ) identified at ∼1946 cm−1 , as shown in Fig. 4C, which is too broad to be deconvoluted reliably. Nonetheless, the C O mode may be roughly accounted for by a combination of two peaks at ∼1962 and 1930 cm−1 , associated with the 482 and 502 cm−1 conformers, respectively, as predicted by the Fe CO and CO inverse correlation curve discussed below. Table 1 summarizes the Fe CO , Fe C O , and C O modes of the various NOS complexes. The similarity of these modes between the oxygenase domain and the full-length enzyme indicates that the reductase domain does not significantly modify the heme environment in the oxygenase domain and, hence, the oxygenase domain serves as a valid model for the native enzyme. Furthermore, the CO-related vibrational modes and their broad width found in iNOSoxy are similar to those found in nNOSoxy [34,35,49,50]. In nNOSoxy , the Fe CO stretching mode was deconvoluted into two major components, one at 489 cm−1 and the other at 502 cm−1 as well as a very small component at 514 cm−1 (Fig. 5B, top right) [48], which are attributed to distinct protein conformations [34].

472

D.L. Rousseau et al.

(A)

(B)

iNOSoxy

nNOSoxy

502

502 489

482 512

514

No substrate

L-Arginine

460

500

540

Raman shift

460

500

540

(cm–1)

Fig. 5. Deconvolution of the Fe CO stretching modes of iNOSoxy and nNOSoxy in the presence and absence of the l-Arginine substrate.

Table 1. The Fe CO , Fe C O and C O modes of the CO-bound ferrous complexes of iNOSoxy , full length iNOS (iNOSFL ), nNOSoxy , full length nNOS (nNOSFL ) and full length eNOS (eNOSFL ) Complex (Fe+2 –CO)

Substrate Cofactor

Fe−CO (cm−1 )

iNOSoxy

− l-Arg − H4B

562

+ l-Arg − H4B − l-Arg + H4B + l-Arg + H4B

491 (Broad) (482,502) 512 490 (Broad) 512

iNOSFL

− l-Arg + H4B + l-Arg + H4B

487 (Broad) 512

nNOSoxy

− l-Arg − H4B + l-Arg − H4B + l-Arg + H4B

495 (Broad) (489,501) 502 502

nNOSFL

− l-Arg + H4B + l-Arg + H4B + NOHA + H4B

eNOSFL

+ l-Arg + H4B

Fe C (cm−1 )

C O (cm−1 )

Reference [37,48]

569 562 569

1946 (1930,1962) 1907 1944 1905

560 567

1945 1906

[35] [35]

N.D.

[48]

565 565

1932 N.D.

[49] [49]

498 (Broad) 503 502

562 565 563

1936 1929 1928

[34,50] [50] [50]

512

567

N.D.

[35]

O

[37] [37] [37]

Note: N. D. – Not Determined. The contributions to the broad lines are shown in parentheses below the relevant entry.

Ligand Interactions in NOS

473

3.2. The Effect of Substrates and Cofactor In iNOSoxy , the addition of l-Arg causes a change in the distribution of the components of the Fe CO mode seen in the absence of substrate and the appearance of a new line at 512 cm−1 ; in addition, the intensity of the Fe C O bending mode is significantly enhanced (see spectrum (b) in Fig. 4A [37] and deconvolution shown in Fig. 5A, lower left [48]). The appearance of the sharper line at 512 cm−1 suggests a decreased conformational freedom for the Fe C O moiety due to H-bonding to the CO due to the presence of l-Arg. A direct H-bonding interaction with the CO ligand is supported by FTIR studies of the ferrous-CO derivative of iNOSoxy , showing a 0.8 cm−1 shift in C O when the solvent H2 O was replaced with D2 O [51]. A similar isotope effect on the C O mode in the FTIR spectrum was observed for nNOS [52]. The crystal structures of CO-bound nNOSoxy was recently reported by Li et al. [53] and it was found that the CO was H-bonded by both the l-Arg and a water molecule consistent with the spectroscopic observations. In the absence of l-Arg, the addition of H4B does not affect the spectral shapes or frequencies of the CO-related vibrational modes. Likewise, addition of H4B to the l-Arg-bound protein does not introduce additional changes to the CO-related vibrational modes. For nNOSoxy , addition of l-Arg changes the distribution of the lines seen in its absence as may be seen for the deconvolution of the Fe CO mode shown in Fig. 5B, lower right [48]. Thus, significant differences exist between nNOS and the other two isoforms. For nNOS, the dominant component of the Fe CO mode is found at 502 cm−1 whereas that for both iNOS and eNOS is located at ∼512 cm−1 (Table 1) [35,50]. Similarly, in the presence of both l-Arg and H4B the C O mode is found at 1929 cm−1 in nNOS [50] and 1906 cm−1 in iNOS (that in eNOS has not been reported). The lower Fe CO frequency and the correspondingly higher C O frequency in nNOS indicate a weaker H-bonding interaction to the heme-bound CO in nNOS. The data suggest that nNOS has a distinct distal pocket with respect to the other two isoforms resulting in different H-bonding strengths. Interestingly, this type of isoform-specific difference was not detected for the NO-bound ferric derivative as will be discussed below, suggesting that the response of the enzyme to ligand binding is sensitive to the identity of the ligand.

3.3. The Fe

CO − C

O

Inverse Correlation Curve

There is a well known inverse correlation between the frequencies of the Fe CO and C O modes as illustrated in Fig. 6, due to the backbonding from the d orbital of the heme iron to the empty  ∗ -orbital of the CO [46]. The changes in the polarity of the distal heme environment affect the degree of backbonding, thus shifting the data points along the correlation curve. On the other hand, changes in the electronic properties of the proximal iron-ligand bond affect the offset of the Fe CO and C O correlation curve, due to the competition between the proximal and distal ligands for -bonding to the dz2 orbital of the heme iron. As a result, the correlation line for cysteine-ligated heme proteins, like P450s, is distinct from that of the correlation line for the histidine-ligated heme proteins, for example hemoglobins and that for the five-coordinate CO-bound model compounds (Fig. 6).

D.L. Rousseau et al.

Fe–C–O Stretching mode (cm–1)

474

L–Fe+2–CO

520 510

xxxx

500

L=His

490

x 480 iNOS 470

P450

x nNOS

CPO 460 1900

1920

1940

1960

1980

C–O Stretching mode (cm–1)

Fig. 6. Inverse correlation curve between C O and Fe CO in the CO derivatives of hemeproteins coordinated with a histidine proximal ligand, the nitric oxide synthase enzyme family and the P450 enzyme family. Points on the nitric oxide synthase curve were compiled from Table 1. All other points were taken from correlation curves determined by Yu et al. [47].

Figure 6 shows a plot of Fe CO versus C O frequencies for the CO-bound NOS complexes in comparison to histidine-ligated heme proteins and thiolate-ligated P450 types of proteins. The NOS data follow the well-known inverse correlation as discussed above. The NOS correlation line is located lower than that of the histidine-ligated heme proteins but is higher than that of the P450s, despite the fact that both the NOS and the P450 groups of enzymes have a cysteine as the proximal ligand [46]. Nonetheless, this observation is consistent with the fact that the Fe Cys stretching frequency of NOS is ∼ 13 cm−1 lower than that of P450, due to the weaker electron donating capability of the proximal Cys in NOS with respect to P450 [54]. Intriguingly, the frequency of the Fe Cys stretching mode of CPO is 9 cm−1 higher than that of NOS in the ferric ligand-free state, but the data point for CPO lies on the same correlation curve as the NOS family. This inconsistency may reflect the difference in the Fe Cys bond strength in the ligand-free ferric state with respect to that in the CO-bound ferrous state in NOS and/or CPO. The shift of the iNOSoxy data points toward the upper left corner of the correlation line upon l-Arg binding reflects the strong H-bonding interactions with the CO ligand induced by the presence of l-Arg.

3.4. Heme Deformation Induced by Substrate and Cofactor Binding The changes in the Fe C O related modes upon the addition of l-Arg and/or H4B are associated with a small enhancement of the out-of-plane heme modes, indicating a slight deformation of the heme induced by substrate/cofactor binding [37]. Most noticeable is the increase in the intensity of a heme mode at 693 cm−1 upon the addition of l-Arg and/or H4B. The mode is strongest in the presence of both l-Arg and H4B, indicating an additive effect of each. An enhancement was also observed at 752 and 803 cm−1 ,

Ligand Interactions in NOS

475

suggesting that they are of similar origin. Analogous spectral changes were also observed in the reported data for the nNOSoxy domain and for the full-length enzymes of all three isoforms, indicating that this effect is not isoform-specific [34,35,49,50]. Similarly, but more significant, heme-distortion was observed in the NO-bound derivatives of NOS in response to substrate and/or cofactor binding (vide infra). Interestingly, although the heme distortion is further enhanced upon the addition of H4B to the l-Arg-bound protein, no additional changes are observed in the Fe C O related vibrational modes, suggesting that the changes in heme deformation due to the presence of H4B do not affect the H-bonding interactions between the l-Arg and the heme-bound CO.

4. THE NITRIC OXIDE-BOUND COMPLEXES At the completion of the catalytic cycle, the NO that is produced in the distal pocket of NOS may bind geminately to the ferric heme and thereby inhibit further turnover of the enzyme. Hence NO plays an important role in regulating the enzymatic activity of NOS [30,31,55]. Santolini et al. demonstrated that the degree of self-inhibition by NO depends on the following factors: (1) the off-rate of NO from the ferric heme, (2) the ease of reduction of the ferric NO-bound form to the ferrous derivative, because the ferrous heme iron has a much stronger NO affinity and (3) the ease of oxidation by oxygen of the NO-bound ferrous form to the ferric form plus nitrate [32]. Due to isoform-specific rates, the degree of self-inhibition differs substantially from one isoform to the other, ranging from 70 to 90% in nNOS, to 25% in iNOS and to a negligible amount in eNOS. In addition to self-inhibition, NO also influences the NOS activity by modulating the stability of the dimeric interface. This process is discussed in depth in Section 5. In order to obtain an integrated molecular picture of the auto-inhibition process, it is first necessary to understand the interactions between NO, the heme and the surrounding protein matrix.

4.1. The NO-bound Ferric Species The NO-bound ferric heme iron and the CO-bound ferrous heme iron are isoelectronic; in addition, both ligands typically bind in a preferentially perpendicular orientation with respect to the porphyrin plane. However, the interaction between l-Arg and the hemebound NO in the ferric protein is distinct from that between l-Arg and CO in the ferrous protein. The interactions of NO with the ferric heme in nNOS, nNOSoxy and iNOSoxy have been reported [33,37,41]. The low frequency resonance Raman spectra of the NO derivatives of iNOSoxy and nNOSoxy are shown in Fig. 7. The Fe N O related modes and the out-of-plane heme modes, which reflect the degree of heme distortion, are sensitive to the addition of substrate or cofactor as will be discussed below. In the absence of l-Arg and H4B, the Fe NO stretching mode (Fe NO ) is identified at 537 cm−1 (spectrum (a) in Fig. 7A). It shifts to 533 cm−1 upon isotope substitution of 14 N16 O with 15 N16 O. The addition of l-Arg does not introduce any changes to the spectrum as shown in spectrum (b). Unfortunately, it is unclear if l-Arg binds in a site too far from the heme-bound NO to interact with it or if the l-Arg does not bind at all. In contrast, significant changes are seen upon the addition of H4B. In the presence of

476

D.L. Rousseau et al. (B)

(A) ν7

Fe+3–NO iNOSoxy

533

ν6 344

390

15N16 16O

N O

537

γ15

676 685

540 551

ν15

ν47 752 718

(–, –)

(a) (a)

(a) 533

527 15

N O 15N16O

537

(b)

(+, –)

541 550

550

710

γ5

15 15N16 16O

352 352

N O

528

729 (–, +)

(c) (c)

(c) 537

545

533 15N16 16O

N O

(+, +)

(d)

547

(d) (d)

746

γ1

300

551

(b) (b)

γ11

537

γγ6 6

541

16

400

500

600

700

534

800

480

520

560

600

Raman shift (cm–1)

Fig. 7. Panel A: Resonance Raman spectra of the NO-bound ferric complexes of iNOSoxy in the presence and/or absence of l-Arg and H4B. The isotope sensitive peaks determined from the 15 16 N O-bound ferric-iNOSoxy complexes are shown in the inset above their corresponding peaks associated with the natural abundance species. Panel B: The resulting isotope difference spectra (14 N16 O–15 N16 O) are shown. The (− −), (+ −), (− +) and (+ +) stand for −Arg/−H4B, +Arg/−H4B, −Arg/+H4B and +Arg/+H4B, respectively. Adapted from [37].

H4B, two isotope-sensitive lines are detected at 541 and 550 cm−1 for the 14 N16 O that merge into a single line at 537 cm−1 for the 15 N16 O adduct as shown in spectrum (c). Although in the absence of H4B, the addition of l-Arg does not affect the spectrum, in the presence of H4B, the addition of l-Arg causes these modes to shift to a single mode at 545 cm−1 for 14 N16 O and 537 cm−1 for 15 N16 O. The larger isotope shift of 8 cm−1 with respect to the 4 cm−1 shift found in spectrum (a), suggests that the 545 cm−1 mode originates from a Fe N O bending mode (Fe N O ), instead of a stretching mode (Fe NO ). Accordingly, the 541 and 550 cm−1 lines in spectrum (c) are assigned to the Fe NO and Fe N O modes, respectively. The enhancement of the Fe N O mode in the presence of H4B indicates that the Fe N O moiety adopts a bent conformation. This interpretation is substantiated by the fact that there are no reported cases showing the presence of the bending mode when the Fe N O assumes a linear structure that is perpendicular to the heme plane. Furthermore, similar Fe N O stretching and bending modes have been reported at 522 and 546 cm−1 , respectively, for cytochrome P450cam by Hu and Kincaid [56,57].

Ligand Interactions in NOS

477

Similar modes were also observed in CPO [56,57] and it was found that, in P450cam , the bending mode was enhanced in the presence of substrate just as that observed here for the iNOSoxy complex. However, in P450cam the substrate binds directly over the NO and the enhancement of the bending mode is accounted for by a direct steric interaction between the substrate and NO, whereas in NOS, the H4B binding site is remote from the distal ligand binding site (Fig. 2). To examine if there is a second H4B binding site on the distal side of the heme, H4B was titrated into NO-bound iNOSoxy . It was found that the binding of H4B to iNOSoxy is stoichiometric with one H4B molecule per protein molecule, thereby excluding the possibility of a second binding site for H4B and a direct interaction between H4B and the Fe N O moiety [37]. It was postulated by Li et al. that the bent Fe N O conformation is a result of an electronic effect introduced by heme distortion as evident from the enhancement of the heme out-of-plane modes [37], as will be discussed below. Although it is uncertain if l-Arg binds to the ferric NO-bound iNOSoxy in the absence of H4B, the distinct spectrum (d) in Fig. 7 with respect to spectrum (c) demonstrates that l-Arg does bind to the protein in the presence of H4B. These data show that the Fe N O mode is further enhanced and shifted upon the addition of l-Arg to the H4B-bound protein, although the heme out-of-plane modes are unaffected. Hence, in the presence of H4B, a direct steric or H-bonding interaction between heme-bound NO and l-Arg must be present, which causes a further increase in the tilting and/or bending of the Fe N O moiety. Similar spectral changes induced by l-Arg and/or H4B are observed in nNOSoxy [41,58]. Furthermore, in the presence of H4B, the degree of heme distortion induced by NO binding is similar in these two isoforms as will be discussed below.

4.2. The NO-bound Ferrous Species The resonance Raman spectra of the ferrous-NO complexes of iNOSoxy and nNOSoxy and full-length nNOS have been reported [33,37,41]. The spectroscopic behavior of each of these complexes shows distinct features. In the ferrous-NO complex of the nNOSoxy , the six-coordinate NO-bound form is unstable in the absence of l-Arg and H4B. It quickly converts to a five-coordinate NO-bound species. In the presence of H4B, an NO isotopesensitive line was detected at 543 cm−1 , which shifts to 523 cm−1 in the 15 N16 O-bound derivative [33,41]. The isotopic shift of ∼20 cm−1 is much greater than that observed for the ferric NO adducts but it is similar to that reported for the NO adducts of the ferrous P450 and CPO [56,57]. Hu and Kincaid attributed the large isotopic shift in P450 and CPO to the bent Fe N O geometry and the partial mixing of the stretching and bending modes [57]. Accordingly, the line in nNOSoxy was assigned as the Fe NO stretching mode with some bending character. Interestingly, the Fe NO mode, characterized at 536 cm−1 in the full-length ferrous nNOS with H4B, is different from that observed in nNOSoxy (543 cm−1 ) obtained under identical conditions. This suggests a structural change to the oxygenase domain induced by its interaction with the reductase domain. In contrast, when CO is used as the structural probe, structural differences between the full length and the oxygenase domain of nNOS are not observable (Table 1). The origin of these differences remains to be determined. In nNOSoxy , as well as in the full-length enzyme, the Fe NO mode shifts to 549 cm−1

478

D.L. Rousseau et al.

upon the addition of l-Arg to the H4B-bound protein. This spectral change is associated with small enhancements of several out-of-plane heme modes, most notably the line at 692 cm−1 which is assigned to the heme distortion-sensitive 15 mode (vide infra). The enhancement of the out-of-plane heme modes, as well as the sensitivity of the ligand related modes to the presence of l-Arg, is very similar to the behavior observed in the CO-derivative. In iNOSoxy , the response of the ferrous NO-bound complex to the addition of substrate and cofactor is somewhat different from that of the nNOS complexes. In the absence of l-Arg and H4B, the ferrous NO complex of iNOSoxy is unstable and converts to a five-coordinate NO-bound species as observed for nNOSoxy . The addition of NO to the ferrous complex of iNOSoxy in the presence of H4B causes a rapid auto-oxidation and leads to the six-coordinate NO-bound ferric species [37]. In contrast, the NO adduct of ferrous nNOSoxy in the presence of H4B alone forms a stable six-coordinate complex. Accordingly, the resonance Raman spectrum of the NO-bound ferrous derivative of iNOSoxy can only be measured when l-Arg is present as shown in Fig. 8 [37]. The Fe NO mode is located at 540 cm−1 in the presence of l-Arg alone, on the basis of the 15 16 N O isotopic substitution experiment. Addition of H4B to the l-Arg-bound protein causes the Fe NO mode to shift to 550 cm−1 (spectrum (b) in Fig. 8), a frequency similar to that of nNOSoxy in the presence of both l-Arg and H4B. Furthermore, several outof-plane heme modes, such as the 692, 715, 734 752 and 800 cm−1 lines, are enhanced due to heme deformation.

γ5 γ15 734

676

Fe+2–NO iNOSoxy

692

540

γ11 752

(+, –)

800

715 533

(a)

15

(+, +)

550

(b)

N16O

550

(c) 527 350

400

450

500

550

600

650

700

750

800

Raman shift (cm–1)

Fig. 8. Resonance Raman spectra of the ferrous-NO complexes of iNOSoxy . The (+ −) and (+ +) stand for +Arg/−H4B and +Arg/+H4B, respectively. The 14 N16 O–15 N16 O isotope difference spectrum in the presence of both l-Arg and H4B are shown below the (+ +) spectrum. Adapted from [37].

Ligand Interactions in NOS

479

Since at the end of the catalytic cycle the l-Arg is totally consumed, the facile conversion of the ferrous NO-bound form to the ferric derivative in the absence of l-Arg in iNOS may prevent the enzyme from being trapped in the NO-bound ferrous state. The inhibitory NO-bound ferric complex can release its NO efficiently, since the NO off-rate is much faster for the ferric heme, and iNOS can be poised for further enzymatic activity when additional l-Arg becomes available. The instability of the ferrous NO-bound form in the absence of l-Arg in iNOS is not found in nNOS, hence it may partially account for the smaller degree of NO auto-inhibition in iNOS with respect to nNOS.

4.3. Heme Distortion Significant changes are seen in the low frequency region of the Raman spectrum upon the addition of H4B to the ferric NO-bound complexes of both iNOSoxy and nNOSoxy , in addition to the changes in the ligand-related modes. Changes in low frequency vibrational modes in the resonance Raman spectrum have been seen in other heme protein systems when the heme becomes distorted [43,59]. When the porphyrin macrocycle has a planar structure with D4h symmetry it has a relatively simple spectrum; however, when it is non-planar several out-of-plane heme modes become active and the low frequency resonance Raman spectrum displays a much more complicated pattern with respect to that of the undistorted heme [59]. Heme distortion may be quantitatively assessed, if the structure of the heme is known, with the normal-coordinate structural decomposition method (NSD) developed by Shelnutt and coworkers [60]. In the NSD method, the heme distortion is broken down into low frequency normal coordinates, including ruffling (B1u ), saddling (B2u ), doming (A2u ), waving (Eg ) and pyrrole propellering (A1u ) deformations as illustrated in Fig. 9. With this method, the out-of-plane displacement associated with each distortion coordinate can be calculated. NSD analysis of ferrochelatase indicates that the distortion follows a doming coordinate. In contrast, the mesoporphyrin-bound antibody 7G12 shows strong saddling (B2u ) and ruffling (B1u ) deformations, as well as a moderate doming (A2u ) deformation [42]. These assignments are consistent with the symmetries of the out-of-plane heme modes that are enhanced in these two proteins [42,61]. Li et al. postulated that the enhancement of the low frequency Raman modes of NOS complexes induced by l-Arg and/or H4B binding is a result of a distorted heme [37]. The most dramatic changes in the low frequency Raman modes were observed in the ferric NO derivatives of both iNOSoxy and nNOSoxy induced by H4B binding (Fig. 7). Based on the assignments for ferrochelatase [42] and other normal mode analyses, several of the modes in the resonance Raman spectra of the NO-bound ferric complexes of iNOSoxy were assigned (Fig. 7). In the absence of l-Arg and H4B (spectrum a), the lines at 344, 676 and 752 cm−1 are assigned as the in-plane modes, 6 , 7 and 15 , respectively. The weak 685 cm−1 line is assigned as an out-of-plane mode, 15 , with B2u symmetry, suggesting a slightly saddled heme. In the presence of H4B, the new lines at 352 and 729 are assigned to the out-of-plane A2u modes, 6 and 5 , respectively, indicating a doming distortion. The lines at 710 and 746 cm−1 are assigned to a B1u mode (11 ) indicating ruffling and an A1u mode (1 ), indicating a propeller deformation, respectively. In addition, the large increase in the intensity of the B2u mode (15 ) at 685 cm−1 upon H4B binding indicates a saddling distortion. The presence

480

D.L. Rousseau et al.

sad (B2u)

wav(x) (Egx)

ruf (B1u)

wav(y) (Egy)

dom (A2u)

pro (A1u)

Fig. 9. Various heme distortion modes. Symmetry types for non-planar distortion of the porphyrin macrocycle used for the normal coordinate structural decomposition (NSD) from Shelnutt et al. [60]. Here sad, ruf, dom, wav and pro stand for saddling, ruffling, doming, waving and propellering, respectively.

of these out-of-plane modes with differing symmetry types indicates that the heme is distorted along several coordinates. In addition to heme deformation, the enhancement of the 390 cm−1 line, which is assigned to a propionate mode, is attributed to the direct H-bonding interaction between the propionate and the H4B. Similar heme deformations were observed in nNOSoxy upon the addition of H4B. The absence of any changes in the low frequency modes upon l-Arg binding indicates that it does not introduce significant distortion to the NO-bound ferric heme. Since the crystal structures of the ferric NO-bound complexes of iNOSoxy and nNOSoxy have not been reported, the two structures [62] of the ferrous NO-bound eNOSoxy complexes (1FOO and 1FOP) that are available in the PDB were examined and analyzed by the NSD method [37]. It was found that the heme is highly distorted, especially in the presence of H4B, consistent with the conclusions drawn from the Raman measurements on the NO-bound ferric complexes of iNOSoxy and nNOSoxy . In the ferrous NO-bound derivative of iNOSoxy (Fig. 8), H4B binding significantly enhances the 692 cm−1 line (15 ) with B2u symmetry and brings about small increases in the 715 (11 ) and 734 cm−1 (5 ,) lines with B1u and A2u symmetries, respectively. These changes indicate a large change in the saddling (B2u ) deformation and small changes in the doming (A2u ) and ruffling (B1u ) deformations. This conclusion is consistent with the NSD analysis of the NO-bound ferrous derivative of eNOSoxy in which the addition of H4B in the presence of l-Arg generated a large change in the saddling deformation and small changes in the doming and ruffling coordinates [37]. A closely related example of another distorted heme was reported for the NO-bound nitrophorin-4 by Roberts et al. [63]. Nitrophorins are a family of proteins present in blood-sucking insects that release NO to bring about vasodilation and reduction of

Ligand Interactions in NOS

481

blood coagulation [64]. The heme in nitrophorin-4 is highly ruffled and the NO is bent. The low spin configuration of the ferric iron d-orbitals in heme proteins is usually dxy 2 dxz dyz 3 . However, Walker and collaborators noted that ruffling of the heme in nitrophorin-4 is associated with a change in the electronic structure to dxz  dyz 4 dxy 1 as the unpaired electron in the dxy orbital can not mix with the porphyrin -system if the heme is planar [63]. When the heme is ruffled, the -orbitals of the porphyrin have in-plane components that can overlap with the dxy orbitals thereby increasing their electron density and stabilizing a distorted structure [63]. This additional electron density in the dxy orbital donated from the porphyrin -system lowers the redox potential of the heme iron because upon reduction and formation of the Fe2+ state another electron must be transferred to this orbital which is no longer empty. In most of the heme proteins discovered to date the NO off-rate from the ferrous protein is extremely slow due to the high stability of the reduced state [65–67]. Thus, the lower redox potential of the heme iron in nitrophorins, due to the distortion of the heme, is very important for its function as an NO transporter [63]. On the same basis, we postulate that the heme distortion observed here in NOS also serves to regulate the NO auto-inhibition by modulating the redox potential of the heme iron. Moreover, the differing degree of NO auto-inhibition in the various NOS isoforms may be a consequence of variations in the heme distortion.

5. THE EFFECT OF NO ON THE MONOMER/DIMER EQUILIBRIUM It is well accepted that dimerization is essential for NOS function [6,68]. However, it has been found that NO produced from the catalytic reaction in iNOS not only can rebind to the heme iron, thereby directly inhibiting the turnover of the enzyme [30,31] as discussed above, but in the absence of H4B it can also induce monomerization of the functional dimers [55]. To evaluate the effect of NO on the dimeric interactions in iNOSoxy , the substrate and cofactor-free ferric enzyme were subjected to NO and the reactions were monitored with optical absorption spectroscopy as a function of time [69].

5.1. Formation of a Five-coordinate NO-bound Species As shown in Fig. 10A, immediately after the addition of NO to iNOSoxy in the absence of l-Arg and H4B, a species with a Soret absorption maximum at 439 nm and visible absorption bands at 549 and 580 nm was produced. It is assigned as a six-coordinate (6C) NO-bound ferric iNOSoxy complex, since it is analogous to the spectra of other reported 6C NO-bound NOS complexes [33]. The 6C NO-bound ferric enzyme gradually converted to a species with a Soret maximum at ∼390 nm over a ∼300 min time period with a clear isosbestic point at 411 nm. The new species is assigned as a five-coordinate (5C) derivative of iNOSoxy , since its spectral properties are similar to those of other 5C derivatives of NOS [33,37]. The properties of the 5C species are discussed below. To further evaluate the mechanism of the 6C to 5C conversion,

482

D.L. Rousseau et al.

439

0.6

439

Population

Absorbance

0.8

0.8 τ1 = 21 min τ2 = 287 min

390 0

0.4

200 400 Time (min)

τ1 = 23 min τ2 = 396 min

0.6

600

0

200 400 600 Time (min)

0.4 0.2

0.2

0.0

390

Population

(B)

(A)

Wild type 400

0.0 450

500

550

Wavelength (nm)

600

Wild type + L-Arg 400

450

500

550

600

Wavelength (nm)

Fig. 10. Optical absorption spectra of the H4B and DTT-free ferric form of iNOSoxy in the absence (A) and presence (B) of l-Arg (10 mM) as a function of time following the exposure to ∼1 mM NO. The arrows indicate the direction of absorbance changes with increasing time. Inset: The relative populations of the 5C NO species as a function of time on the basis of spectral deconvolution of the time-dependent optical absorption data. The dotted lines show double exponential fits of the data. The relative amplitudes of the fast to the slow phases are 51/49 in (A) and 33/67 in (B). The concentrations of the iNOSoxy samples were 100 and 90 M for (A) and (B), respectively.

each time-dependent spectrum was deconvoluted into a linear combination of the spectrum of the 6C NO-bound ferric species and that of the 5C species. Typical examples demonstrating the reliability of the deconvolution process are shown in Fig 11A). The resulting population of the 5C species is plotted as a function of time in the inset in Fig. 10A and the associated kinetic trace was best-fitted with a double exponential function with lifetimes of 21 and 287 min. A similar reaction was observed for iNOSoxy in the presence of l-Arg as shown in Fig. 10B. Although the kinetic lifetimes, 23 and 396 min, are only slightly altered upon the addition of l-Arg, the relative amplitude of the slow phase increases from 49% in the absence of l-Arg to 67% in its presence. The conversion of the 6C NO-bound ferric derivative to the 5C species was inhibited by the binding of H4B, either with or without l-Arg. On the basis of gel-filtration analysis, Panda and coworkers have reported that, in the absence of H4B, iNOSoxy is in equilibrium between a monomeric and a “loose” dimeric form and l-Arg binding shifts the equilibrium toward the loose-dimer, whereas H4B binding generates a “tight”-dimer [70]. Accordingly, it was postulated that the production of the 5C species in the reactions shown in Fig. 10 is a consequence of the lack of strong dimeric interactions in the absence of H4B [69]. The slow phase was attributed to the reaction of the loose-dimer because the amplitude of the slow phase increased from 49% to 67% when l-Arg was added, and the fast phase to the reaction of the monomeric fraction of the iNOSoxy samples, as it decreased correspondingly in the presence of l-Arg. To test this hypothesis the reaction between the monomeric form of iNOSoxy and NO was examined by using a urea-induced monomer as a model [69]. It has been reported

Ligand Interactions in NOS

483

(A)

(B) 0.8

390

Data Best-fit 6C (54%) 5C (46%)

439

Absorbance

0.6

Data Best-fit 6C (25%) Hemin (75%)

365

0.5

0.4

0.2

0.0

420 0.8

0.2 –0.1

Residual

400

450

500

550

Wavelength (nm)

600

Residual

400

450

500

550

600

650

Wavelength (nm)

Fig. 11. Typical examples of spectral deconvolution. (A) Deconvolution of the spectrum of the ferric form of iNOSoxy in the presence of 4 M urea following exposure to 1 mM of NO for 25 min. The spectrum was deconvoluted into 54% of the 6C species and 46% of the 5C species. The negligible residuals from the fitting confirm the reliability of the fitting procedure. (B) Deconvolution of the spectrum of the partially denatured protein in the presence of 7.0 M urea. The spectrum was deconvoluted into 25% 6C ferric ligand-free species and 75% free hemin; again the negligible residuals from the fitting confirm the reliability of the procedure.

that 5 M urea induces 100% conversion of the dimeric enzyme into monomers, but it is accompanied by significant loss of the heme group due to denaturation; on the other hand, reducing the urea to 3 M can only induce ∼94% of the dimer to convert to its monomeric form [70]. In order to find the best conditions for generating the monomeric enzyme without denaturation, the iNOSoxy was titrated with urea and it was found that 4 M urea was an optimum condition for generating the monomeric enzyme without heme loss (Fig. 12A). The monomeric state of the 4 M urea-treated iNOSoxy sample was confirmed by MALDI-TOF mass spectrometric measurements (data not shown) and by gel filtration analysis (Fig. 13). As shown in Fig. 12B, exposure of the 4 M urea-treated iNOSoxy to NO instantaneously produced a 6C NO-bound ferric species with a Soret maximum at 439 nm, just as that observed in the urea-free samples shown in Fig. 10; in addition, an analogous spectral transition from the 6C NO-bound ferric derivative to a 5C species was observed, although with altered kinetics. To gain quantitative information, the population of the 5C species was estimated by spectral deconvolution of the optical absorption data and was plotted as a function of reaction time in the inset in Fig. 12B. The resulting kinetic trace was best fitted with a single exponential function with a lifetime of ∼30 min. This lifetime is similar to that of the fast phase (21–23 min) obtained in the absence of urea (Fig. 10), consistent with the scenario that the fast phase originates from the monomeric enzyme. It should be noted that in the gel filtration measurements in 4 M urea, a small amount of dimer was detected but its elution volume was spread out in comparison to the absence of urea. It was attributed to the presence of some very loose-dimer, which reacted with NO as rapidly as the monomer.

484

D.L. Rousseau et al.

(A) Population

419

Absorbance

1.2 1.0

365

419 nm

1.0

365 nm

0.8

439 390

0.8

0

2 4 6 Urea (M)

8

Population

(B)

τ = 30 min 0

0.6

40 80 120 Time (min)

0.6 0.4 0.4 0.2

0.2 0.0

Wild type (Urea titration)

Wild type + 4 M Urea 0.0

400

450

500

550

400

600

Wavelength (nm)

450

500

550

600

Wavelength (nm)

Absorbance (280 nm)

Fig. 12. (A) Optical absorption spectra of the H4B and DTT-free ferric derivative of the wild-type iNOSoxy in the presence of 0–8 M urea. (B) The optical absorption spectra of the 4 M ureatreated sample as a function of time following the exposure to ∼1 mM NO. The arrows indicate direction of absorbance change with increasing urea concentrations or reaction time. In (A) as the concentration of urea was increased, the 6C ferric derivative with a Soret maximum at 419 nm progressively converted to a new species with a Soret band at 365 nm, which is characteristic of free hemin (-oxo dimer) in aqueous solution [71] indicating the release of the hemin from the protein matrix due to unfolding of the polypeptide chain. The populations of the 419 nm and 365 nm species are plotted as a function of the concentration of urea in the inset. In (B) the relative populations of the 5C NO species as a function of time on the basis of spectral deconvolution of the time-dependent optical absorption data are plotted in the inset. The dotted lines show single exponential fits of the data. The protein concentrations were 13–3 M in (A) with increasing dilution by urea and 100 M in (B).

Dimer Monomer 0 M urea

4 M urea

6

8

10

12

14

16

Elution volume (ml)

Fig. 13. Gel filtration of iNOSoxy in the absence and presence of 4 M urea. A large increase in the amount of monomer may be seen in the presence of urea. These are two of a series of data obtained as a function of urea concentration. The species at an elution volume of ∼8 ml is attributed to aggregation. The shift in the scales in the two data sets is a consequence of the higher viscosity of the urea.

Ligand Interactions in NOS

485

390

439

365

0.3

439

τ = 28 min 0

0.2

390

40 80 120 Time (min)

0.1

Population

0.3

(B) Population

(A)

τ = 25 min 0

0.2

40 80 120 Time (min)

0.1

D92A mutant

K82A mutant

0.0

0.0 400

450

500

550

600

400

450

500

550

600

Wavelength (nm)

Wavelength (nm)

Fig. 14. Optical absorption spectra of the D92A (A) and the K82A (B) mutants of the H4B- and DTT-free ferric derivative of the wild-type iNOSoxy as a function of time following the exposure to ∼1 mM NO. The arrows indicate the direction of absorbance changes with increasing time. Inset: The relative populations of the 5C NO species as a function of time on the basis of spectral deconvolution of the time-dependent optical absorption data. The dotted lines show single exponential fits of the data. The concentrations of the iNOSoxy samples were 30 M.

To confirm that the fast phase indeed originates from the monomeric-derivative and to eliminate any possible side effects caused by the addition of urea, the NO reaction was examined with two iNOSoxy mutants, D92A and K82A, which adopt a pure monomeric conformation in the absence of H4B [72]. Figure 14A shows the time-dependent optical absorption spectra of the D92A mutant of iNOSoxy following exposure to NO. Again, the instantaneously formed 6C NO-bound enzyme with a Soret band at 439 nm converted to the 5C species with a Soret maximum at 390 nm with a single exponential decay rate of ∼28 min, similar to that observed in the urea-stabilized monomeric wild-type enzyme sample. Similar kinetic behavior was observed in the K82A mutant (Fig. 14B), confirming that the 20–30 min kinetic phase originates from the monomeric form of the enzyme. On the basis of these data, it was concluded that NO binding to the ferric iNOSoxy protein, in either the monomeric or dimeric state, instantaneously produces a 6C NObound ferric derivative with indistinguishable optical absorption spectra with a Soret transition maximum at 439 nm. Subsequently, the fast phase is attributed to the transition from the 6C NO-bound state of the monomeric protein ([M NO]6C ) to the 5C species ([M]5C NO ), whereas the slow phase is ascribed to the same reaction originating from the loose-dimer ([D NO]6C ) as described below. D

NO 6C → M

NO 6C → M 5C

NO

(1)

Here, the formation of the 5C species is rate-limited by the monomerization of the dimer, with an apparent lifetime of ∼300–400 min.

486

D.L. Rousseau et al.

5.2. Identification of the Five-coordinate Species To gain insights into the nature of the 5C species, the NO-treated samples were examined with resonance Raman spectroscopy [69]. As shown in Fig. 15, the resonance Raman spectra of the 5C species with a Soret maximum at 390 nm generated either in the presence or absence of urea are very similar (top and middle traces), indicating that the two 5C species are the same. Since these spectra are virtually identical to those of 5C NO-bound ferrous derivatives of a variety of heme proteins, as characterized by the heme modes located at 349, 677 and 756 cm−1 and a broad Fe NO stretching mode (Fe NO ) in the 520–526 cm−1 region [73–77], the resonance Raman spectrum of the 5C NO-bound ferrous derivative of the iNOSoxy complex was also obtained (the bottom trace in Fig. 15). Here, the 5C NO-bound ferrous derivative was formed by directly adding NO to the substrate-free and cofactor-free ferrous enzyme, since the conversion of the six-coordinate ferrous NO complex to its five-coordinate form has been previously demonstrated in both iNOSoxy and nNOSoxy by Stuehr and coworkers [40,78]. The small differences in the 378 and 524 cm−1 regions are attributed to differences in the contributions of the laser plasma lines. The identical features in the three traces shown in Fig. 15 indicate that exposure of the ferric derivative of H4B-free iNOSoxy to NO leads to the reduction of the ferric heme iron to the ferrous form and the breakage of the proximal iron–thiolate bond. According to Eq. 1, the dimer to monomer conversion occurs prior to the reduction to the ferrous 5C form. This is consistent with prior reports of monomerization induced by the presence of NO. It is also consistent with our observation, as well as that of others, that the addition of l-Arg to the ferrous enzyme results in a stable dimer that does not form a five-coordinate species upon the addition of NO.

* 349 423 378

677

756

524

*

*

* ** * *

*

Urea

Fe+3 + NO

Fe+3 + NO

588

*

(Urea)

*

Fe+2 – NO

* 300

400

500

600

700

800

Raman shift (cm–1)

Fig. 15. The resonance Raman spectra of the 5C NO-bound species generated from the ferric derivative of iNOSoxy in the absence and presence of 4 M urea and the 5C ferrous-NO boundderivative. All the spectra were taken with an excitation wavelength at 406.7 nm. The lines marked with an asterisk (∗ ) denote the plasma lines from the laser. The vibrational mode at 588 cm−1 is associated with urea.

Ligand Interactions in NOS

487

5.3. Autoreduction Mechanism One possible mechanism to account for the conversion of the 6C NO-bound ferric derivative to the five-coordinate NO-bound ferrous form is a heterolytic cleavage of the proximal iron–thiolate bond: Cys− − Fe+3

NO → Cys• + Fe+2

NO

(2)

To test this mechanism, we re-examined the NO reaction with the 4 M urea-treated iNOSoxy as a function of the NO concentration. We found that the formation rate of the 5C species increases, approximately linearly, as the NO concentration increases (data not shown). Since the SN 1-type heterolytic cleavage reaction predicts an NO concentrationindependent kinetic process, this mechanism is excluded. NO mediated conversion of a 6C NO-bound ferric protein to a 6C NO-bound ferrous protein has been well-documented for histidine-ligated heme proteins, such as hemoglobin or myoglobin [65,67,68]. In these cases, the exposure of the ferric protein to NO first produces a 6C NO-bound ferric heme, via the displacement of the distal water ligand by NO: His

Fe+3

H2 O + NO → His

Fe+3

NO + H2 O

(3)

The formation of the 6C NO-bound ferric heme is followed by autoreduction of the heme iron, leading to the formation of a 6C NO-bound ferrous heme. Three different mechanisms have been proposed to account for the NO-induced heme iron reduction reaction. In the first mechanism proposed by Chien [79] and by Hoshino et al. [80], a base catalyzed conversion of the 6C NO-bound ferric form to a nitrite-bound ferrous derivative (His Fe2+ NO2 − ) is followed by the displacement of the heme-bound nitrite with a second NO molecule to form the NO-bound reduced heme (His Fe2+ NO). In the second mechanism, postulated by Ehrenberg and Szczepkowski [81], a rate-limiting heterolytic cleavage of the distal iron NO bond, generating His Fe2+ and NO+ , is followed by coordination of a second NO molecule. In the third mechanism, proposed by Addison and Stephanos [67], the reduction of the ferric NO-bound heme is induced by a nucleophilic attack by a second NO, resulting in a metastable species [NO Fe3+ NO], which rapidly converts to a ferrous form with the release of NO+ and rebinding of the proximal His ligand. In all three mechanisms, the reductant is the exogenous NO, which is oxidized to either a nitrite or a nitrosonium ion (NO+ ). In a recent study, NO-induced auto-reduction was observed in a hemoglobin from a clam, Scapharca inaequivalvis [65]. In that work, a 6C NO-bound ferrous species was observed instantaneously following the addition of NO to the ferric species. The authors proposed that the reaction follows the Addison and Stephanos type of mechanism, and the absence of any detectable 6C NO-bound ferric derivative suggests that the overall reaction is rate-limited by the binding of NO to the ferric protein (Eq. 3), instead of the following autoreduction reaction. More interestingly, it was shown that the nitrosonium ion (NO+ ) released from the autoreduction reaction was able to nitrosylate a Cys residue to form an S-nitrosylated species (SNO). On the basis of the above observations, it was postulated [69] that the monomeric 6C NO-bound ferric iNOSoxy is first reduced to the 6C NO-bound ferrous derivative

488

D.L. Rousseau et al.

by a mechanism similar to that proposed by Addison and Stephanos. Studies of the pH dependence of the reactions are needed to determine the contribution from the mechanism described by Chein [78] and by Hoshino et al. [79]. The formation of the ferrous derivative is followed by a homolytic cleavage of the proximal iron–thiolate bond to produce the 5C NO-bound ferrous species: Cys− − Fe3+

NO + NO → Cys− − Fe2+

Cys− − Fe2+

NO → Cys− + Fe2+

NO + NO+

(4)

NO

(5)

Since the 6C NO-bound ferrous derivative was not observed during the reaction, the reaction must be rate-limited by the autoreduction process described in Eq. 4. This is consistent with the result reported by Abu-Soud et al., who found that at 10 C the conversion of the 6C NO-bound ferrous iNOSoxy , in the absence of substrate and cofactor, to its 5C NO-bound derivative exhibited a lifetime of 4.6 min [40,78]. This is much shorter than that of the overall formation rate of the 5C species observed here. In the presence of substrate and/or the H4B cofactor, the ferrous NO derivative is stable, as described in Section 4.1, presumably owing to the formation of a tight-dimer. Thus, the dimer to monomer conversion must occur in the oxidized state thereby facilitating the five-coordinate formation. The NO+ released from the autoreduction process can diffuse to the Zn-binding site and nitrosylate C104 or C109 leading to an intramolecular disulfide formation between these two residues as will be discussed in more detail below. As discussed above (Section 4.3), H4B binding to the NO-bound ferric iNOSoxy brings about a significant out-of-plane distortion of the heme [37] that perturbs the electronic properties of the heme iron, making it more difficult to reduce. The resistance of the H4B-bound iNOSoxy to NO-induced autoreduction reported here can in part be attributed to the same origin. Although it is well known that the 6C NO-bound ferric derivatives of myoglobin or hemoglobin are susceptible to autoreduction, the resulting 6C NO-bound ferrous species can be quite stable. On the other hand, the 6C NO-bound ferrous species found in several other heme protein systems, especially those involved in NO sensing and binding functions such as soluble guanylate cyclase [82], cytochrome c’ [83] and other hemebased sensors [84], are labile. They readily convert to 5C NO-bound ferrous derivatives due to a weakened proximal Fe His bond. For thiolate ligated heme protein systems, such as P450 or NOS (the H4B-bound form), the 6C NO-bound ferric complexes are typically more resistant to autoreduction owing to the electron donating capability of the proximal cysteine thiolate that stabilizes the higher oxidation state of the heme iron. However, autoreduction and the associated breakage of the proximal iron–thiolate bond have been reported in these protein systems due to the changes in the electronic properties of the proximal thiolate ligand or the electrostatic environment of the distal NO binding site. As an example, in P450 1A2, the D318 residue in the distal pocket stabilizes the ferric NO complex by forming an H-bond to the heme-bound NO. Exposure of the ferric derivative of the D318A mutant to NO results in a 5C NO-bound ferrous species [85]. In NOS, a Trp residue on the proximal side of the heme, which forms an H-bond with the sulfur atom of the proximal thiolate heme ligand, is very important in tuning the electron density on the thiolate ligand. Exposure of the ferrous derivative of nNOS mutants, in which this Trp residue is mutated to a Tyr or Phe, to NO leads to the formation a 5C ferrous NO species [86]. The iNOSoxy data presented here provides an

Ligand Interactions in NOS

489

additional example in which NO binding to ferric heme iron induces autoreduction of the heme iron as well the cleavage of the proximal iron–thiolate bond.

5.4. The Mechanism of Monomerization In iNOS, the N-terminal region, between residues 76 and 111, comprising a -hairpin hook and a CXXXC Zn-binding motif, also is believed to be important for stabilizing the dimeric structure. Crane et al. reported that the N-terminal region of iNOSoxy can be in either a “swapped” or an “unswapped” conformation as illustrated in Fig. 2B [87]. In the “unswapped” conformation, Cys104 and Cys109 in the Zn-binding motif of each subunit of the dimer are tetrahedrally coordinated to a single zinc ion at the dimer interface and the -hairpin hook interacts primarily with its own subunit; whereas in the “swapped” conformation Cys109 forms a self-symmetric disulfide bond across the dimer interface and the -hairpin hook in one subunit of the dimer interacts primarily with the other subunit across the interface [87]. Similar, Zn-binding sites are present in the other isoforms. Recently, Ravi et al. reported that exposure of eNOS to NO causes S-nitrosylation of a Cys residue at the Zn-binding site, leading to monomerization of the enzyme [88]. S-nitrosylation has also been reported in iNOS by Mitchell et al. [89]. To determine if NO caused any chemical modifications on the polypeptide chain of iNOSoxy , mass spectrometric measurements on the NO treated samples were carried out. All the samples examined were first subjected to trypsin digestion prior to the mass spectrometric analysis. The major modification in the mass spectra of the NO-treated samples versus the control sample without NO treatment was the enhancement of the three fragment ions with an m/z of 581.94, 640.27 and 743.85. The charge states of the three fragments were determined to be +3, +3 and +2, respectively, on the basis of their characteristic isotopic distributions. The parent masses of the 581.94 and 743.85 ion peaks (M = 1742.82 and M = 1485.70) calculated based on the charges are exact matches to two expected trypsin cleavage products of iNOSoxy , corresponding to the [82–97] and [393–404] peptide fragments, respectively. These assignments were confirmed by the tandem mass spectrometric data (data not shown). Intriguingly, all observed fragment ion peaks in the mass spectra can be accounted for by the expected trypsin cleavage products, except the triply charged ion at 640.27 with a parent mass of 1917.81. This ion peak is an exact match of a disulfide bond-linked [98–105]–[108–117] peptide fragment through C104 and C109. This assignment was confirmed by the tandem mass data shown in Fig. 16. To further verify the disulfide bonded peptide fragments, the trypsin-digested fragments of the NO treated sample (in the presence of 4M urea) were reduced by DTT (to reduce the disulfide bond) and alkylated by iodoacetamide (to alkylate the reduced free cysteine residues). This treatment resulted in the appearance of a doubly charged ion with an m/z of 553.76, whose parent mass (M=1105.5) is an exact match for the [108–117] fragment with a carbamidomethylated cysteine residue, at the expense of the fragment ion peak at m/z of 640.27. (see Fig. 17). The modified [98–105] fragment was not observed, plausibly due to its low ionization propensity. Other than a very small contribution from a C109-C109 disulfide bonded fragment (data not shown), no other disulfide linked trypsin-digested iNOSoxy fragments were observed; furthermore, no fragments were found to contain

490

D.L. Rousseau et al. 874.32(2) yA6*

145.09 aA1 Fragment A : ATSDFTCK (residues 98-105) Fragment B: SCLGSIMNPK (residues 108-117)

746.34 yB7 244.15

yB2 173.08

Relative intensity

bA1

689.34 yB6 147.10 yB1

358.19 yB3

602.31 yB5

489.22 yB4

859.40 yB8

773.32(2) yA4*

830.82(2) yA5*

~ ~

341.17 yB3-NH3

120 160 200 240 280 320 360 400 440 480

500

550

600

650

700

750

800

850

900

950 1000 1050

m/z, amu

Fig. 16. ESI MS/MS spectrum of the triply charged fragment ion at 640.27. The triply charged fragment ion at 640.27 contains [98–105] and [108–117] fragments that are disulfide bonded, as labeled as A and B, respectively. The coverage of the b and y ions, representing the cleavage products resulting from the N-terminus and C-terminus, respectively, is indicated above each corresponding peak, with the subscript denoting their fragment of origin and the number of residues remaining in the fragment. The charge state for each ion is +1, unless otherwise indicated in parentheses. The ions labeled as yA4∗ and yA5∗ stand for the modified yA4 and yA5 ion that are still disulfide bonded to the B-fragment. The yB3-NH3 ion is the modified yB3 ion, in which an NH3 side chain group of an asparagines residue at the N-terminus is lost. The aA1 ion is a modified bA1 ion, in which the CO group is lost.

any NO-derivatized amino acids. This is in contrast to the results of NO treatment of nitrophorins in which the proximal cysteine bond becomes ruptured and nitrosylated [90,91]. Since the enzyme is in a monomeric state in the presence of 4 M urea, the formation of the disulfide linked [98–105]–[108–117] peptide fragment must be a result of intramolecular rather than intermolecular interactions. Taken together these data indicate that the NO-induced monomerization of the loose-dimer is coupled to an intramolecular disulfide bond formation between cysteine residues at position 104 and 109, and the monomerization process exposes the [82–97] and [393–404] peptide fragments to solvent, making them more accessible to trypsin digestion as reflected by the enhancement of the corresponding fragment ion peaks shown in Fig. 18.

Ligand Interactions in NOS

491 606.81

100

(A)

80 60

Relative intensity (%)

40

555.89 581.94

552.63

640.29 (+2)

20 0

(B)

60

553.76 (+3) 606.82

40

20

569.76

581.94

0 560

580

600

620

640

m/z, amu

Fig. 17. The mass spectra of the trypsin digested products of the NO-treated iNOSoxy sample pretreated with 4 M urea before (A) and after (B) reduction and alkylation. The arrow indicates the disappearance of the triply charged ion peak at 640.27 and the concurrent appearance of the doubly charged ion peak at 553.76 upon reduction and alkylation. (A)

(B)

Relative intensity (%)

581.94 (+3)

[82-97]

100

(C) 640.27 (+3)

20

[98-105] [108-117]

743.85 (+2)

[393-404]

20

80 15

15

10

10

5

5

60

(a) 40 20 0

(b) (c)

581.0

0 582.0

583.0

640.0

641.0

0 743.0

744.0

745.0

746.0

m/z(amu)

Fig. 18. The mass spectra of the NO-treated iNOSoxy samples with (A) and without (B) pretreatment with 4 M urea and the control sample without NO treatment (C). All the samples were digested with porcine trypsin for 12 h prior to the mass spectral measurements. The data show the three fragment ions [82–97], disulfide-linked [98–105]–[108–117] and [393–404], which undergo the largest change in intensity in response to the NO treatment. In each panel the zero positions were displaced for clarity. The equally spaced peaks associated with each fragment are a result of H/D isotopic substitution; based on the spacing the charge state of each fragment ion was determined as indicated in parentheses. The identity of each fragment as indicated was confirmed by tandem mass spectrometry.

492

D.L. Rousseau et al.

On the basis of the mass spectrometric data, the autoreduction of the heme iron and the breakage of the proximal iron–thiolate bond in iNOSoxy are associated with intramolecular disulfide bond formation between C104 and C109. Although, no evidence of nitrosylated cysteine residues in our NO-treated iNOSoxy samples by either optical absorption (as an increase in absorbance at ∼320 nm) or mass spectrometry was found, it is most likely that the disulfide bond formation is triggered by the nitrosylation of C104 and/or C109 by the NO+ released from the autoreduction reaction as illustrated below [69]. R1 S− + R2 S− + NO+ → R1 SNO + R2 S− → R1 S

SR2 + NO−

(6)

In this model one of the residues is S-nitrosylated and the other thiolate attacks the thiol through an SN 2-type of reaction resulting in the loss of nitroxyl and the formation of the disulfide bond. Although NO+ is a much better nitrosylation reagent than NO especially under the strictly anaerobic conditions applied, the possibility of a direct reaction between neutral NO and the cysteine residues as have been reported for several other protein systems could not be excluded [92–94]. Nonetheless, based on either scenario, the disulfide bond is only formed in the monomeric or loose-dimeric state when the two cysteine residues at position 104 and 109 are accessible to NO+ /NO. Additional reactions may occur under aerobic conditions that also lead to disulfide bond formation [92–94], however they are unlikely under the reported experimental conditions [69]. (B) PAR

.4 .3

(A) .6

PAR +10 µM Zn

.2

min tt ==00min (100% (100% 6C) 6C)

NO A: iNOS ++NO

.5 tt ==300 300min min (~26% (~26% 5C) 5C)

.4

.1 0 300

400

500

(a) PAR + iNOS (Fe3+) No Zn released

.3 .3 .2

.2

(b) PAR + iNOS+NO ~2.6 µM 5C ~2.5 µM Zn released

.1

.1

600

(C)

0

0 300

400

500

600

300

400

500

600

Fig. 19. Zinc release by the absorbance change of the PAR dye. (A) Absorbance spectra of the treatment of iNOSoxy with NO. In this experiment 26% of 10 M enzyme converted to the 5C species indicating that in the resulting solution ∼2.6 mM of the dimer was dissociated. (B) Spectra of the PAR in the absence and the presence of 10 M Zn. (C). Addition of PAR to iNOSoxy . In both spectra, the contribution from the enzyme was subtracted out to more clearly show the PAR spectrum. In (a) the PAR was added to the ferric enzyme in the absence of NO. In (b) the PAR was added to the 300-minute sample shown in Panel A. The increase in the absorbance near 500 nm in (b) indicates that ∼25 M of Zn is formed in the reaction consistent with the amount of monomer that was produced.

Ligand Interactions in NOS

493

Based on the crystal structures shown in Fig. 2A, the iNOS dimer is stabilized through the tetrahedral coordination of C104 and C109 to a Zn atom in an “unswapped” conformation or by forming a disulfide linkage between C109 residues in a “swapped” conformation in which the N-terminal -hook interacts with peptide segment from the opposite subunit. To determine if the samples contained Zn, the amount of Zn released by a 4(2-Pyridylazo Resorcinol) (PAR) absorbance assay was carried out [88]. Upon the conversion of iNOSoxy dimers to monomers, a quantitative release of Zn was found confirming the presence of the “unswapped” configuration of the enzyme (Fig. 19). It appears that the nitrosylation reaction of the cysteine residues and the consequent formation of the intramolecular C104-C109 disulfide bonds trigger the dissociation of the NO-bound loose-dimer into monomers leading to the formation of the 5C NO-bound ferrous protein with the release of NO+ , which may further catalyze the monomerization reaction of the loose-dimer. A similar disulfide bond formation reaction can also occur in the NO-bound monomeric state resulting in the same 5C NO-bound ferrous product.

6. OVERVIEW OF NO INTERACTIONS The overall regulatory role of NO in iNOSoxy is illustrated in Fig. 20. When apo-iNOS is produced and released from the ribosome, it recruits a prosthetic heme group to

Monomer Inhibitory Pathway [M+2 – NO]*5c NO+ NO [M+3 – NO]*6C

[D′+3 – NO]6C NO

NO [M+3]

NO +3] [D’+3 [D′ 6C 6C

6C

H4B e– [D+2] 5C

e–

[D+3] 6C

NO3–

NO O2

NO

[D+2 – O2 ]6C

[D+3 – NO] 6C

NO Producing Catalytic Cycle

[D+2] 5C

O2

NO

e–

NO [D+2 – NO] 6C

NO Autoinhibitory Pathway

Fig. 20. Proposed NO-mediated regulatory mechanism in iNOS. The M, D and D represent the monomeric, the “tight-” dimeric and the “loose-” dimeric forms of the enzyme, respectively. The asterisk (∗ ) indicates the presence of an intramolecular C104-C109 disulfide bond. See text for full description.

494

D.L. Rousseau et al.

form a monomeric holo-protein ([M+3 ]6C ), which can assemble into a “loose”-dimer ([D’+3 ]6C ). In the absence of the substrate (l-Arg or NOHA) and cofactor (H4B), the loose-dimer is in equilibrium with the monomeric state. l-Arg or NOHA binding shifts the equilibrium from the monomer toward the loose-dimer; whereas H4B binding introduces conformational changes that convert the enzyme to the functional tight-dimer ([D+3 ]6C ). NO is generated in the tight-dimer via the NO producing catalytic cycle illustrated in the lower left section of the figure. The detailed mechanism of this process has not been delineated but many aspects of the process are being revealed by various studies of ligand interactions with the enzyme. The NO produced in the distal pocket of the tight-dimer can rebind to the heme iron to produce a 6C NO-bound ferric heme ([D+3 NO]6C ) as indicated by the NO autoinhibitory pathway illustrated in the lower right in Fig. 20. If the 6C NO-bound ferric heme is reduced to the 6C NO-bound ferrous heme ([D+2 –NO]6C ) by receiving an electron from the reductase domain, the enzyme is trapped in this inactive state, because the NO dissociation rate is very slow from the ferrous enzyme (the dissociation reaction is thus ignored in this figure). When NO is overproduced, the [D+2 NO]6C species can also be generated by NO binding to the ligand-free ferrous protein ([D+2 ]5C ), although NO binding to the ferric heme is expected to be the dominant pathway. The [D+2 NO]6C species can be converted back to the active [D+3 ]6C state by reacting with O2 to produce nitrate. It has been shown that the NO autoinhibitory pathway is regulated by a delicate balance between the dissociation rate of the NO, the reduction rate of the [D+3 NO]6C species and the nitration rate of [D+2 NO]6C with O2 [32,95,96]. Catalytically generated NO may also bind to the freshly produced monomeric species ([M+3 ]6C ) or the loose-dimer ([D’+3 ]6C ) to generate the 6C NO-bound species, [M+3 NO]∗ 6C and ([D’+3 NO]6C , respectively (shown in the top of Fig. 20). The 6C NO-bound monomeric species is not stable and readily converts to a 5C NO-bound ferrous species ([M+2 NO]∗ 5C ) through autoreduction and the associated proximal iron– thiolate bond cleavage reaction. The autoreduction reaction produces a nitrosonium ion (NO+ ), which may react with the C104 and C109 residues in the monomer to form an S-nitrosylated product that subsequently leads to an intramolecular disulfide bond between these two residues. The NO+ can also react with the Zn-binding site of the loose-dimer ([D’+3 NO]6C ) via the same reaction, thereby inducing the monomerization of the dimer. A similar monomerization reaction may also be directly induced by the NO molecule as indicated (although perhaps to a lesser extent).

7. CONCLUSIONS Interactions between heme ligands and catalytic center of nitric oxide synthase are important for the enzymatic activity of this ubiquitous enzyme. Studies of these interactions have expanded our understanding of many aspects of the enzymatic mechanism but much remains to be resolved. It has been found that NO can regulate iNOS function in several ways. Through the interactions described above it can modify the monomer– dimer equilibrium by inducing the formation of a non-native disulfide linkage between C104 and C109. However, the role of this process under in vivo conditions has not been determined. NO formed by the enzyme can geminately bind to the heme iron atom and thereby autoinhibit the enzyme. Interestingly, the degree of autoinhibition varies widely

Ligand Interactions in NOS

495

among the three mammalian isoforms by mechanisms that can be understood kinetically, but the molecular mechanisms underlying the isoforms-specific rates of heme oxidation/reduction and NO release, as well as the related auto-inhibition efficiency, are yet to be determined.

ACKNOWLEDGMENTS This work was supported by the National Institute of Health Research Grant GM54806 to D.L.R. and HL65465 to S.-R.Y. Both E.Y.H. and D.L. are supported by the Molecular Biophysics Training Grant (GM08572) at Albert Einstein College of Medicine and D.L. is also supported by the Medical Scientist Training Program (GM07288). We would like to thank Dr Tsuyoshi Egawa and Dr Jack Peisach of Albert Einstein College of Medicine for many useful discussions.

REFERENCES [1] Garthwaite, J., Charles, S.L. and Chess-Williams, R. (1988) Nature, 336, 385–388. [2] Furchgott, R.F. (1988) In Vasodilation: Vascular Smooth Muscle, Peptides, Autonomic Nerves, and Endothelium (P.M. Vanhoutte, ed.), pp. 401–414, New York: Raven. [3] Hibbs, J.B., Jr., Taintor, R.R., Vavrin, Z. and Rachlin, E.M. (1988) Biochem. Biophys. Res. Commun., 157, 87–94. [4] Ignarro, L.J., Buga, G.M., Wood, K.S., Byrns, R.E., and Chaudhuri, G. (1987) Proc. Natl. Acad. Sci. USA., 84, 9265–9269. [5] Stuehr, D.J. and Nathan, C.F. (1989) J. Exp. Med., 169, 1543–1555. [6] Stuehr, D.J. (1999) Biochim. Biophys. Acta., 1411, 217–230. [7] Alderton, W.K., Cooper, C.E. and Knowles, R.G. (2001) Biochem. J., 357, 593–615. [8] Masters, B.S., McMillan, K., Sheta, E.A. et al. (1996) FASEB J., 10, 552–558. [9] Mayer, B. and Andrew, P. (1998) Naunyn Schmiedebergs Arch. Pharmacol., 358, 127–133. [10] Xie, Q.W., Cho, H., Kashiwabara, Y. et al. (1994) J. Biol. Chem., 269, 28500–28505. [11] Marletta, M.A. (1993) Adv. Exp. Med. Biol., 338, 281–284. [12] Stuehr, D.J., Kwon, N.S., Nathan, C.F. et al. (1991) J. Biol. Chem., 266, 6259–6263. [13] Kwon, N.S., Nathan, C.F., Gilker, C. et al. (1990) J. Biol. Chem., 265, 13442–13445. [14] Leone, A.M., Palmer, R.M., Knowles, R.G. et al. (1991) J. Biol. Chem., 266, 23790–23795. [15] Wu, C., Zhang, J., Abu-Soud, H. et al. (1996) Biochem. Biophys. Res. Commun., 222, 439–444. [16] Stuehr, D.J., Abu-Soud, H.M., Rousseau, D.L. et al. (1995) Adv Pharmacol., 34, 207–213. [17] Adachi, S., Nagano, S., Ishimori, K. et al. (1993) Biochemistry, 32, 241–252. [18] Abu-Soud, H.M., Yoho, L.L. and Stuehr, D.J. (1994) J. Biol. Chem., 269, 32047–32050. [19] Siddhanta, U., Presta, A., Fan, B. et al. (1998) J. Biol. Chem., 273, 18950–18958. [20] Crane, B.R., Arvai, A.S., Ghosh, D.K. et al. (1998) Science, 279, 2121–2126. [21] Crane, B.R., Arvai, A.S., Gachhui, R. et al. (1997) Science, 278, 425–431. [22] Li, H., Raman, C.S., Glaser, C.B. et al. (1999) J. Biol. Chem., 274, 21276–21284. [23] Raman, C.S., Li, H., Martasek, P. et al. (1998) Cell, 95, 939–950. [24] Fedorov, R., Hartmann, E., Ghosh, D.K. and Schlichting, I. (2003) J. Biol. Chem., 278, 45818–45825. [25] Li, H., Shimizu, H., Flinspach, M. et al. (2002) Biochemistry, 41, 13868–13875. [26] White, K.A. and Marletta, M.A. (1992) Biochemistry, 31, 6627–6631.

496

D.L. Rousseau et al.

[27] [28] [29] [30] [31]

McMillan, K. and Masters, B.S. (1995) Biochemistry, 34, 3686–3693. Richards, M.K., Clague, M.J. and Marletta, M.A. (1996) Biochemistry, 35, 7772–7780. McMillan, K., Salerno, J.C. and Masters, B.S. (1996) Methods Enzymol., 268, 460–472. Abu-Soud, H.M., Wang, J., Rousseau, D.L. et al. (1995) J. Biol. Chem., 270, 22997–23006. Abu-Soud, H.M., Ichimori, K., Nakazawa, H. and Stuehr, D.J. (2001) Biochemistry, 40, 6876–6881. Santolini, J., Meade, A.L. and Stuehr, D.J. (2001) J. Biol. Chem., 276, 48887–48898. Wang, J., Rousseau, D.L., Abu-Soud, H.M. and Stuehr, D.J. (1994) Proc. Natl. Acad. Sci. USA, 91, 10512–10516. Wang, J., Stuehr, D.J., Ikeda-Saito, M. and Rousseau, D.L. (1993) J. Biol. Chem., 268, 22255–22258. Fan, B., Wang, J., Stuehr, D.J. and Rousseau, D.L. (1997) Biochemistry, 36, 12660–12665. Poulos, T.L., Finzel, B.C. and Howard, A.J. (1986) Biochemistry, 25, 5314–5322. Li, D., Stuehr, D.J., Yeh, S.R. and Rousseau, D.L. (2004) J. Biol. Chem., 279, 26489–26499. Martasek, P., Liu, Q., Liu, J. et al. (1996) Biochem. Biophys. Res. Commun., 219, 359–365. Roman, L.J., Sheta, E.A., Martasek, P. et al. (1995) Proc. Natl. Acad. Sci. USA, 92, 8428–8432. Abu-Soud, H.M., Wu, C., Ghosh, D.K. and Stuehr, D.J. (1998) Biochemistry, 37, 3777–3786. Couture, M., Adak, S., Stuehr, D.J. and Rousseau, D.L. (2001) J. Biol. Chem., 276, 38280–38288. Blackwood, M.E., Jr., Romesberg, F., Rush, T.S. et al. (1998) Biochemistry, 37, 779–782. Yeh, S.R., Takahashi, S., Fan, B. and Rousseau, D.L. (1997) Nat. Struct. Biol., 4, 51–56. Spiro, T.G. and Wasbotten, I.H. (2005) J. Inorg. Biochem., 99, 34–44. Yu, N.T. (1986) Methods Enzymol., 130, 350–409. Yu, N.-T. and Kerr, E.A. (1987) In Biological Applications of Raman Spectroscopy (3 vols), Vol. 3 (T.G. Spiro, ed.) New York: John Wiley & Sons, pp. 39–96. Yu, N.T., Benko, B., Kerr, E.A. and Gersonde, K. (1984) Proc. Natl. Acad. Sci. USA, 81, 5106–5110. Li, D., Kabir, M., Stuehr, D.J. et al. (2007) J. Am. Chem. Soc.. Couture, M., Stuehr, D.J. and Rousseau, D.L. (2000) J. Biol. Chem., 275, 3201–3205. Wang, J., Stuehr, D.J. and Rousseau, D.L. (1997) Biochemistry, 36, 4595–4606. Jung, C., Stuehr, D.J. and Ghosh, D.K. (2000) Biochemistry, 39, 10163–10171. Ingledew, W.J., Smith, S.M., Salerno, J.C. and Rich, P.R. (2002) Biochemistry, 41, 8377–8384. Li, H., Igarashi, J., Jamal, J. et al. (2006) J. Biol. Inorg. Chem., 11, 753–768. Schelvis, J.P., Berka, V., Babcock, G.T. and Tsai, A.L. (2002) Biochemistry, 41, 5695–5701. Chen, Y., Panda, K. and Stuehr, D.J. (2002) Biochemistry, 41, 4618–4625. Hu, S. and Kincaid, J.R. (1991) J. Am. Chem. Soc., 113, 2843–2850. Hu, S. and Kincaid, J.R. (1993) J. Biol. Chem., 268, 6189–6193. Rousseau, D.L., Li, D., Couture, M. and Yeh, S.R. (2005) J. Inorg. Biochem., 99, 306–323. Jordan, T., Eads, J.C. and Spiro, T.G. (1995) Protein Sci., 4, 716–728. Shelnutt, J.A. (2000) In The Porphyrin Handbook, Vol. 7 (K.M. Kadish, K.M. Smith and R. Guilard, eds) New York: Academic Press, New York, pp. 168–223. Lu, Y., Sousa, A., Franco, R. et al. (2002) Biochemistry, 41, 8253–8262. Li, H., Raman, C.S., Martasek, P. et al. (2001) Biochemistry, 40, 5399–5406. Roberts, S.A., Weichsel, A., Qiu, Y. et al. (2001) Biochemistry, 40, 11327–11337. Valenzuela, J.G., Walker, F.A. and Ribeiro, J.M. (1995) J. Exp. Biol., 198 (Pt. 7), 1519–1526. Boffi, A., Sarti, P., Amiconi, G. and Chiancone, E. (2002) Biophys. Chem., 98, 209–216. Keilin, D. and Hartree, E.F. (1937) Nature, 139, 548. Addison, A.W. and Stephanos, J.J. (1986) Biochemistry, 25, 4104–4113. Marletta, M.A. (1994) Cell, 78, 927–930.

[32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68]

Ligand Interactions in NOS [69] [70] [71] [72] [73] [74] [75] [76] [77] [78] [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95] [96]

497

Li, D., Hayden, E.Y., Panda, K. et al. (2006) J. Biol. Chem., 281, 8197–8204. Panda, K., Rosenfeld, R.J., Ghosh, S. et al. (2002) J. Biol. Chem., 277, 31020–31030. Smalley, J.W., Birss, A.J., Withnall, R. and Silver, J. (2002) Biochem. J., 362, 239–245 Ghosh, D.K., Crane, B.R., Ghosh, S., Wolan, D., Gachhui, R., Crooks, C., Presta, A., Tainer, J.A., Getzoff, E.D., and Stuehr, D.J. (1999) EMBO J., 18, 6260–6270. Deinum, G., Stone, J.R., Babcock, G.T. and Marletta, M.A. (1996) Biochemistry, 35, 1540–1547. Vogel, K.M., Hu, S., Spiro, T.G. et al. (1999) J. Biol. Inorg. Chem., 4, 804–813. Andrew, C.R., Green, E.L., Lawson, D.M. and Eady, R.R. (2001) Biochemistry, 40, 4115–4122. Reynolds, M.F., Parks, R.B., Burstyn, J.N. et al. (2000) Biochemistry, 39, 388–396. Lukat-Rodgers, G.S. and Rodgers, K.R. (1997) Biochemistry, 36, 4178–4187. Huang, L., Abu-Soud, H.M., Hille, R. and Stuehr, D.J. (1999) Biochemistry, 38, 1912–1920. Chien, J.C. (1969) J. Am. Chem. Soc., 91, 2166–2168. Hoshino, M., Maeda, M., Konishi, R. et al. (1996) J. Am. Chem. Soc., 118, 5702–5707. Ehrenberg, A. and Szczepkowski, T.W. 1684–1692. (1960) Acta Chem. Scand. 14, 1684–1692. Stone, J.R. and Marletta, M.A. (1996) Biochemistry, 35, 1093–1099. Lawson, D.M., Stevenson, C.E., Andrew, C.R. and Eady, R.R. (2000) EMBO J., 19, 5661–5671. Karow, D.S., Pan, D., Tran, R. et al. (2004) Biochemistry, 43, 10203–10211. Nakano, R., Sato, H., Watanabe, A. et al. (1996) J. Biol. Chem., 271, 8570–8574. Voegtle, H.L., Sono, M., Adak, S. et al. (2003) Biochemistry, 42, 2475–2484. Crane, B.R., Rosenfeld, R.J., Arvai, A.S. et al. (1999) EMBO J., 18, 6271–6281. Ravi, K., Brennan, L.A., Levic, S. et al. (2004) Proc. Natl. Acad. Sci. U S A., 101, 2619–2624. Mitchell, D.A., Erwin, P.A., Michel, T. and Marletta, M.A. (2005) Biochemistry, 44, 4636–4647. Weichsel, A., Maes, E.M., Andersen, J.F. et al. (2005) Proc. Natl. Acad. Sci. U S A., 102, 594–599. Walker, F.A. (2005) J. Inorg. Biochem., 99, 216–236. Pryor, W.A., Church, D.F., Govindan, C.K. and Crank, G. (1982) J.Org. Chem. 47, 156–159. Gow, A.J., Buerk, D.G. and Ischiropoulos, H. (1997) J. Biol. Chem., 272, 2841–2845. Aravindakumar, C.T., Ceulemans, J. and De Ley, M. (1999) Biochem. J., 344 (Pt 1), 253–258. Stuehr, D.J., Santolini, J., Wang, Z.Q. et al. (2004) J. Biol. Chem., 279, 36167–36170. Gautier, C., Negrerie, M., Wang, Z.Q. et al. (2004) J. Biol. Chem., 279, 4358–4365.

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 18

CooA: A Paradigm for Gas-sensing Regulatory Proteins Gary P. Roberts, Robert L. Kerby, Hwan Youn and Mary Conrad Department of Bacteriology, University of Wisconsin – Madison, Madison WI 53706, USA

Abstract The heme-containing transcriptional factor CooA (CO-Oxidation Activator protein) regulates the expression of genes involved in the anaerobic oxidation of carbon monoxide (CO) in different prokaryotes, but has primarily been studied in the bacterium Rhodospirillum rubrum. In R. rubrum, CooA is both a redox sensor and a specific CO sensor, a combination of properties that is uncommon among heme proteins. Extensive biochemical and genetic analyses, interpreted in the context of structural information, have allowed the creation of hypotheses concerning the mechanism of CooA activation by CO as well as the basis for its CO specificity. This analysis has been enriched by comparison of CooA homologs from other prokaryotes, which typically lack the redox-sensing property and display some differences in their specificity for CO. We provide a hypothesis for the important properties of CooA, detail the experimental support for that hypothesis, and highlight the important areas for future research. Keywords: CO-sensing, transcriptional activation, heme, CRP.

ABBREVIATIONS -CTD -NTD  AR1,2,3 CRP PcooF FNR WT cAMP

the carboxyl terminal domain of the alpha subunit of RNA polymerase the amino terminal domain of the alpha subunit of RNA polymerase the sigma subunit of RNA polymerase the activating regions of CooA and CRP that interact with RNA polymerase cAMP receptor protein the normal promoter of the cooF gene fumarate and nitrate reductase activator protein wild-type cyclic adenosine mono-phosphate

1. INTRODUCTION CooA is a homodimeric heme-containing protein that senses the specific presence of carbon monoxide (CO) in the cytoplasm and responds by binding a specific DNA sequence. This DNA binding allows a precise interaction between CooA and RNA

CooA: A Paradigm for Gas-sensing Regulatory Proteins

499

polymerase, which stimulates the transcription of two operons (where the genes are termed coo for CO oxidation) that encode proteins that oxidize CO to CO2 and reduce protons to H2 . This system of CO sensing and oxidation has been found in a limited set of diverse prokaryotes [1], but has been primarily studied in the photosynthetic bacterium R. rubrum and in the remainder of this chapter we will refer to CooA from this organism unless otherwise stated. At least in R. rubrum, this biochemical system not only oxidizes CO, but generates sufficient energy from that process so that CO serves as a rather good energy source [2,3]. Aside from organisms that inhabit CO-rich environments, the environmental role for such a system is unclear, because we have a poor sense of CO levels in the microenvironments inhabited by most of these organisms. One hypothesis is that R. rubrum, which is primarily aquatic, uses photosynthesis during the day, but must rely on other energy-conservation mechanisms at night. If there is a sufficient flux of CO in the water column, the presence of a CO-oxidation system would provide that alternative mechanism. Though CooA homologs are found in other organisms and these display substantial sequence similarity to CooA of R. rubrum, it has become clear that they also have important differences [1]. CooA of R. rubrum has the additional property that it is also a redox sensor, as discussed immediately below, but this property seems to be lacking in the other homologs that have been found. In some cases the organisms are strict anaerobes and so apparently never encounter oxidizing conditions, and in others, it is possible that the cytoplasmic environment is maintained at a low redox potential irrespective of the external environment. The other very interesting difference is that at least one CooA homolog has the ability to respond to NO as well as CO, albeit under conditions that do not match those encountered by the microbe [4]. The biochemical bases for these differences are discussed in this chapter. As with all biological regulatory processes, the biochemical properties of CooA make sense in terms of its physiological role. The proteins that are expressed under CooA control in R. rubrum are specific for the oxidation of CO, so it is reasonable that CooA should only activate transcription in response to the presence of this molecule. While this specificity of CooA is not surprising, it happens to be rather unusual in its stringency. Many other heme-containing biosensors bind to a variety of small molecule ligands, and in some cases, these ligands serve as physiologically inappropriate effector molecules by activating the sensor [5–8]. One must suppose that the inappropriate ligands are simply not routinely encountered at the necessary levels by the organisms with these sensors. The extreme specificity of CooA from R. rubrum for CO implies that such specificity is relevant for this organism, at least, and the biochemical basis for that is discussed later in this review. R. rubrum CooA also exists in both oxidized (Fe(III) heme) and reduced (Fe(II) heme) forms, but only the reduced form is competent to bind and respond to CO [9]. This makes biological sense because R. rubrum is a facultative anaerobe and CO oxidation only occurs in that organism under reducing conditions. Indeed, the midpoint potential for the reduction of CooA is approximately −300 mV [10], and the CO dehydrogenase that is regulated by CooA is only catalytically active below that redox potential [11]. Although the molecular basis for the redox switch within CooA has been studied and is described below, the actual redox signal that is sensed by CooA in the cell remains unknown. Again, the absence of redox regulation of CooA in other organisms suggests either that the issue of redox sensitivity of their CO

500

G.P. Roberts et al.

dehydrogenase is addressed in a different way or that oxidizing environments are not encountered. CooA of R. rubrum has a number of features that make it biochemically interesting: (i) It displays dramatic selectivity toward a single ligand, CO. (ii) It senses redox by a mechanism that is apparently distinct from other studied redox- and O2 -sensing proteins. During this redox sensing, CooA undergoes an unusual switch of heme ligands: Cys75 is a ligand in the Fe(III) state, but is replaced by His77 in the Fe(II) state [12,13]. (iii) The ligand trans to Cys75 (in the Fe(III) form) and His77 (in the Fe(II) form) is Pro2. Proline has never before been seen as a heme ligand for steric reasons, but in CooA the steric issue is obviated because Pro2 is the N-terminus of each protein monomer [14]. (iv) CooA is a member of the very large family of CRP/FNR transcriptional activators that contains distinct subfamilies that respond to very different effector molecules. In only a few cases are the effectors known [15], but there is reason to believe that there are at least some commonalities in the response mechanisms across the family. Analysis of CooA has provided some of that insight. Additionally, there are a number of technical reasons why CooA provides an outstanding model system for the analysis of the sensing of, and response to, small molecules: (i) The protein is biochemically well-behaved, easily purified and readily assayed both in vivo and in vitro [16]. (ii) The structure of CO-free CooA (inactive for DNA binding) has been solved and a comparison of that structure with the active form of a close homolog, the cAMP receptor protein (termed CRP, though also known as CAP), has allowed testable hypotheses to be formed about the mechanism by which CO binding leads to a transcriptionally active form [14,17,18]. (iii) Because CooA is a regulatory protein, it is technically easy to set up an assay system within the cell that allows the ready detection of CooA variants with biochemically interesting properties. The expression of -galactosidase has been placed under the control of CooA, such that strains with wild-type (WT) R. rubrum CooA produce this easily assayable protein only under strongly reducing conditions and in the presence of CO. However, following a variety of mutageneses of the cooA gene, CooA variants can readily be identified that are functional under other conditions or show aberrant function under normal activating conditions. The subsequent biochemical and spectral analyses of these have provided insights into the function of normal CooA, as well as the rules that govern its interesting properties [19–22]. A recent review of the biochemical and biophysical properties of CooA has been published [23].

2. OVERVIEW OF THE SENSING MECHANISMS OF R. rubrum CooA 2.1. The CooA Response to Changes in Redox The redox state of the cell in general and the presence of O2 specifically are of great importance to cells, because so many aspects of cell metabolism are potentially affected. Not surprisingly, a variety of sensors of both redox state and O2 have evolved [15,24,25]. Perhaps coincidentally, some of these happen to fall in the large CRP superfamily of proteins, though only in the case of FNR is there a good biochemical understanding of the sensing process. The issue is further clouded by the careless assignment of gene names

CooA: A Paradigm for Gas-sensing Regulatory Proteins

501

based only on sequence, so that many of the proteins designated as FNR homologs lack a number of the critical residues found in Escherichia coli FNR [15], which is by far the best characterized. Nevertheless, it is quite possible that some of these mis-designated proteins actually are redox or O2 sensors, albeit by a mechanism unlike that of FNR of E. coli, and there are other members of the CRP protein family that are also likely to have such a physiological role [15]. The response of CooA to redox changes is significantly different from those of other characterized regulatory proteins that sense a similar environmental stimulus. For example, E. coli FNR exists in two forms, a monomeric form that is incapable of highaffinity DNA binding and a dimeric form that binds DNA and activates transcription at appropriate sites [26,27]. FNR only dimerizes when reducing conditions allow the synthesis of Fe4 S4 clusters in each monomer [28–30]. (In contrast, CRP is a dimer both in the presence and absence of cAMP [31,32].) Similarly, CooA is a dimer regardless of redox potential or the presence of CO [9]. Though not a CooA/CRP/FNR homolog, FixL is a heme-containing protein that senses O2 directly and functions as part of a protein complex to activate gene expression only in the absence of O2 [33]. O2 binding to the heme of FixL converts it from five-coordinate high-spin to six-coordinate low-spin [34]. This spin-state change correlates with a conformational change that transmits a signal to its regulatory protein partner, but the actual mechanism of response remains controversial [35]. However, de-activation of FixL does not appear to be specific to O2 binding, though this issue also remains unresolved [25,35–38]. In contrast, the heme of CooA is six-coordinate and low-spin in the Fe(III), Fe(II) and Fe(II) CO forms so neither a spin-state nor a coordination-number change is involved in its activation. Rather, the displacement of Pro2, one of the two endogenous protein ligands, by CO triggers the conformational change that leads to activation [39,40]. Importantly, CooA fails to become active in response to any tested small molecule other than CO. The presence of O2 oxidizes the heme, though in contrast to FixL, it cannot form a stable heme adduct. NO binding to the CooA of R. rubrum yields a five-coordinate species that appears to be inactive [41]. Other examined small molecules fail to bind to the heme of CooA [16].

2.2. The Response of R. rubrum CooA to the Presence of CO The structure of the Fe(II) form of CooA, as determined by X-ray crystallography, is shown in Fig. 1A [14]. One of the axial ligands is His77, which is consistent with a variety of spectroscopic and mutagenic analyses [12,13,42,43]. The presence of Pro2 as a ligand was one of the surprises of the structure for the reasons mentioned above. It is able to serve this role because it is the N-terminus in the mature protein; the terminal Met on any newly synthesized protein is removed by protein processing when the second residue is small. As described below, His77 is critical for CO response, while the primary role of Pro2 appears to be to stabilize the inactive forms when CO is absent [44]. While the structure of the Fe(III) CooA has not been determined, mutational and spectroscopic analyses have shown that Cys75 replaces His77 in this form [12,13]. Similar analysis of variants altered at the N-terminus strongly suggests that Pro2 is the ligand in the Fe(III) form as well [40,45,46].

502

G.P. Roberts et al. (A)

(B)

AR1 Hinge region

D N A

D-helix

F-helix

F-helix F-helix AR1 AR3 β-4/5 loop

AR3

F-helix

E f f e c t o r

AR2

AR2

C-helices

C-helices

Fig. 1. (A) The structure of inactive CooA (PDB ID#1FT9). The two protein chains are shaded differentially. The DNA-binding domain of monomer is noted at the top, but in the other monomer it is folded down near the effector-binding domain. Note that in the case of both monomers, the F-helix, which makes specific DNA contacts, is buried from solvent. The long helices indicated at the dimer interface are the C-helices. The D-helix is not indicated (but see the CRP structure), but is actually the extended portion of the C-helix in the dark protein monomer. A loop that projects upward from the effector-binding domain is termed the 4/5 loop as indicated; this loop also serves as AR3. AR1 is not well defined in CooA and is only generally indicated on the more extended of the two monomers. Finally, the AR2 region occupies one surface of the effector-binding domain as shown. Fig. 3 clarifies the fact that only one of the monomers in each dimer actually presents each AR region to RNA polymerase. The hemes are indicated by the ball-and-stick figures. (B) The structure of active CRP (PDB ID# 1G6N). The structure is rotationally symmetric, but has been turned slightly to align with CooA. The notation system is similar to that shown in Panel A. The hinge region between the effector- and DNA-binding domains is shown, as are the D-helices that are immediately adjacent. The AR1, 2 and 3 regions are indicated, but the AR3 regions are obscured by the DNA-binding domains and are difficult to identify in this view. The ball-and-stick structure represents bound cAMP.

In the presence of CO, the Pro2 ligand of Fe(II) CooA is displaced [39] and the protein undergoes a substantial conformational change detected both on native gels [13], through activity measurements [9], and by analogy to CRP [47] as described in the next section. By analogy with other proteins, it is our working hypothesis that Fe(II) CooA exists in at least two metastable forms, active and inactive. In the absence of CO, that equilibrium is strongly shifted toward the inactive form, while CO binding shifts that equilibrium toward the active form. The key questions concerning CooA are the molecular basis for CO specificity and the related issue of how CO binding triggers a conformational change that activates the protein.

CooA: A Paradigm for Gas-sensing Regulatory Proteins

503

3. COMPARISON OF R. rubrum CooA STRUCTURE AND EFFECTOR RESPONSE TO THOSE OF CRP An understanding of the process of CO activation must proceed from a structural perspective, so we will consider what is known about the active and inactive forms of CooA. The structure of inactive Fe(II) CooA is known, but the structure of the active Fe(II) CO form has not been determined. However, the active structure of the CooA homolog CRP has been determined, both in the presence and the absence of DNA [17,18,48]. The comparison of inactive CooA and active CRP is obviously a bit problematic. The two proteins share only 28% sequence identity, CRP lacks a heme and CRP is activated by binding cAMP. Differences in these two structures will therefore reflect primary sequence differences between the two proteins, and their relative states of activation. However, as indicated in Fig. 1, CooA and CRP nevertheless display remarkable global structural similarity. Both proteins exist as homodimers and the protein monomers of each have two distinct domains: an effector-binding domain that includes the N-terminal two-thirds of each monomer and a DNA-binding domain that involves the C-terminal one-third. These two domains are linked by the hinge region indicated in Fig. 1B. The effector-binding domains in CRP and CooA have important differences, reflecting their different effectors. CRP binds its effector, cAMP, directly, while CooA uses the heme moiety as a prosthetic group. The presence of this heme necessarily demands some changes in the protein structure and CooA lacks six amino acid residues in the heme region relative to CRP. The “isolated” DNA-binding domains of each protein are nearly superimposable on each other, indicating that activation must involve the re-orientation of these domains, rather than a change in their tertiary domain structure. Also, because CooA binds a palindromic DNA sequence reminiscent of that bound by CRP [49], it must also be true that the DNA-binding regions of Fe(II) CO CooA achieve a position reminiscent of that seen in the known CRP structures. As described below, binding of the appropriate effector to either protein must cause a conformational change that shifts the DNA-binding domains from a position where interaction with target DNA sequences is poor to one where it is highly effective and specific [50].

3.1. The Role of Effector Binding on Protein Population Dynamics The above text has implied that both CRP and CooA exist in two forms: an inactive form in the absence of the effector and an active form in its presence. However, this simplistic view cannot be strictly correct both because the proteins are represented by complex populations in the cell and because the two ligand-binding sites within any single protein may interact with varying degrees of cooperativity. Concerning the issue of protein populations, each protein must exist in a dynamic equilibrium between active and inactive forms, with effector binding shifting that equilibrium. However, we currently have little insight into the populations of each protein in either form, because there are no easy quantitative assays of solution conformation. Several disparate observations might be instructive. (i) The crystal structure of the effector-bound form of CRP has been solved suggesting that this species is fairly homogeneous. In contrast, the effector-free form has not been solved, consistent with

504

G.P. Roberts et al.

the notion that it is less homogeneous. (ii) Conversely, the CooA structure has only been solved for the effector-free form and efforts to solve the effector-bound form have been unsuccessful to this point. (iii) CooA variants have been created that increase the variants’ affinity for DNA in the presence of CO over that of CO-bound WT CooA, even though the substitutions lie nowhere near the DNA-binding surfaces (R.L. Kerby, H. Youn and G.P. Roberts, unpublished results). Indeed, such variants have been created by two completely separate sets of changes, yet yield comparable affinity values that are approximately fourfold higher than that of WT CooA. This result is most easily rationalized by a model in which only a fraction of CO-bound WT CooA is in the DNAbinding form, but that these substitutions shift the equilibrium to a more homogenously active state. Taken together, these results suggest a fairly homogenous population of effector-free CooA, consistent with the very low in vivo activity of this form (<2% of the CO-bound form). CO binding apparently causes an incomplete shift in the equilibrium to a dynamic mixture of active and inactive forms, which might explain the difficulty in obtaining crystals of that form. Though not supported by compelling data, the model further suggests that the opposite might be the case for CRP: the active form is homogeneous, yet the inactive form is not. This hypothesis can be rationalized biologically. Under conditions of glucose limitation cAMP accumulates and binds to CRP, which must then activate transcription of a large number of promoters in the cell. As a consequence, a large population of active CRP is necessary. The presence of a low level of CRP activity even under conditions where cAMP levels are negligible happens also to make sense, since a low level of CRP activity has been shown to be important for optimal cell growth under all conditions [31,51]. Finally, the operons activated by CRP also typically have operon-specific regulatory factors, so that substantial aberrant gene expression would not necessarily result from the sub-population of active CRP in the absence of cAMP. The situation with CooA is significantly different. In contrast to the CRP case, there is absolutely no utility in expressing the genes encoding CO oxidation functions in the absence of CO, so CooA function should be extremely low in the absence of effector (and the accumulation of coo gene products is undetectable in R. rubrum in the absence of CO [52]). In the presence of CO, the cell needs only enough active CooA to saturate the two coo promoters, so that relatively little active CooA should be sufficient. The second reason that the simple active-inactive model is not correct is because each of these proteins is a dimer, so there is the additional complexity of heterogeneity in effector binding by each monomer. In the case of CRP, there has been disagreement about whether cAMP binding is positively or negatively cooperative. The disagreement flowed from confusion about the precise number of cAMP molecules bound per dimer and the issue is still under debate [53–55]. There was also a long-standing belief that a CRP dimer bound with a single cAMP had higher DNA affinity than did the form with cAMP bound to each monomer. Again, the problem is a technical one and in part reflects differences in the conditions of the assays employed. It is now clear that the dimer with two cAMP bound has the highest affinity, but the affinity of the singly bound form is unclear. In the case of CooA, kinetic and spectral analysis has shown that CooA is moderately positively cooperative for CO binding (see Section 6), though the activity of the singly bound form is again unknown.

CooA: A Paradigm for Gas-sensing Regulatory Proteins

505

3.2. Insights into CooA Structure and Function through Physical Methods While there are substantial differences between CooA and CRP, there is no doubt that there are underlying similarities in both structure and response to their respective effectors. The ongoing analysis of CooA, informed as it is by the long history of analysis of CRP, should prove fruitful for elucidating the functions of both proteins. The most obvious difference between the X-ray crystal structures of the two proteins is in the placement of the DNA-binding domains. The individual DNA-binding domains from either CooA monomer can be aligned very closely with the analogous domains from either CRP monomer, indicating that neither protein differences nor the different activation states significantly affect the structure within these domains. Rather, it is the position of these domains with respect to the effector-binding domains that is dramatically different in the two proteins (Fig. 1). It is also apparent that the CooA dimer is asymmetric in the solved structure and it is unclear whether this is true for the solution form of the protein. Because the unshaded monomer, with the more “foldeddown” structure, makes other contacts in the crystal lattice, while the shaded monomer does not, it was our original prejudice that a homodimer of the latter form was more likely to represent the major species in solution. A completely different analysis of CooA structure has recently been reported. The Aono and Morishima groups have analyzed the structure of CooA in the presence and absence of CO using small angle X-ray scattering [56]. This approach provides information about the overall protein structure and has the advantage of examining that structure in solution. However, it has the disadvantages of measuring an average of the population of molecules, relying heavily on modeling and being unable to distinguish among multiple plausible forms. Given the arguments above on population dynamics, this is a serious handicap for the application of this method to certain forms of CooA and CRP. The authors’ interpretation of this analysis of CooA is that CO binding has only a very modest effect on the average overall structure, with a “slight swing of the DNA-binding domains away from the heme domains coupled with their rotation by about 8 around the axis of two-fold symmetry” [56]. These results certainly argue that the extended form (shaded monomer) of CooA is not more than a minor component in solution. It is less clear if the modeling actually precludes a dimer of the “folded-down” form, and an independent structural analysis is probably necessary to resolve the matter. Another implication of the model proposed by Akiyama et al. is that the F-helices, which make specific contact with DNA, are solvent exposed in the presence and absence of CO [56]. This is a particularly important claim for several reasons. First, it has biological implications because one supposes that the exposure of the F-helix is essential for non-specific, as well as specific, DNA interaction [50]. Second, both subunits of inactive CooA in the X-ray crystal analysis share the striking feature that the F-helix is rotated away from the solvent (Fig. 1A). This latter view of the CooA structure is consistent with the NMR analysis of CRP, where it was found that the inactive form of CRP also has buried F helices [53,57]. While it is true that CRP and CooA need not share the same inactive structure, it is certainly a reasonable hypothesis that it might be the case and this would be inconsistent with the hypothesis of Akiyama et al. [58]. It should be noted, moreover, that the overall structure of inactive CRP remains a matter of dispute. A small angle X-ray analysis of cAMP-free CRP was published some years ago,

506

G.P. Roberts et al.

but the data were not sufficient to support detailed analysis [59], and a chromatographic analysis of the Stokes radius of CRP showed little structural change until well above the cAMP level where both monomers should be bound [47]. Lastly, knowing the nature of the inactive form of each protein is critical if we want to claim an understanding of the activation process caused by effector binding, since that form defines the starting material. The issue of the activation process of CooA is discussed below, but it is clear that the issue of the solution structures of the various forms of CooA remains a technically daunting challenge.

3.3. Other Approaches to Understanding CooA Structure and Function Other approaches to understand these biological processes involve spectroscopy, which provides valuable information about the heme vicinity, and mutational analysis, in which protein variants have been constructed and analyzed. Another potentially useful approach, which has been rarely used for this specific pair of proteins, is the physical tagging of sites on the protein, and the subsequent analysis of the behavior of those tags in a population in solution. We will very briefly touch on the advantages and disadvantages of each approach and then summarize the current insights that have been gained from their use. Spectroscopic methods that probe protein structure, such as circular dichroism, are of little utility because such methods are primarily sensitive to gross changes in the degree of -helices and -sheets, neither of which appears to differ in inactive CooA when ompared to active CRP, or between active and inactive CRP [57]. NMR spectroscopy has technical challenges with proteins the size of CooA and CRP, but has yielded some very important insights for CRP [56,57]. Spectroscopies of the heme region have also been informative. Electron paramagnetic resonance spectroscopy provides insights concerning the heme ligands in the paramagnetic Fe(III) state and similar information is obtained for the Fe(II) state with magnetic circular dichroism. The disadvantage of these tools is that their interpretation is largely limited to the ligands themselves and relies on comparison to model compounds. As a consequence, the Pro ligand of Fe(II) CooA was originally hypothesized to be a histidine based on these methods [42], because that was the only neutral nitrogen compound studied to that point. Resonance Raman has been very useful in identifying heme ligand candidates and, when used in conjunction with mutational analysis, residues in the vicinity of the heme [60–62]. The challenge in interpretation of mutant data is to know if observed effects are direct or indirect, a concern that is sometimes ignored [63]. Because these methods are largely insensitive to changes beyond the ligands themselves, they have been of only modest value in understanding the movement of the heme and ligands within CooA, which is a key to understanding the activation of CooA by CO. The mutational analysis of proteins has been more successful, but this approach is also complicated. One approach has been to make educated guesses about which residues are functionally interesting and make site-directed changes. This approach has been somewhat useful, but only to the extent that the hypotheses guiding the selection of targets and substitutions have been insightful. More powerful has been the use of direct selections for variants with a novel function, such as effector-independence, since this avoids the problem of simply damaging the protein. In either case, there are challenges in the analysis. One challenge is whether the effect of the substitutions is examined

CooA: A Paradigm for Gas-sensing Regulatory Proteins

507

in vivo or in vitro. In the former case, there are potential concerns with protein processing and stability, which would provide negative results but which might be interpreted as defects in protein function. There can also be problems with adventitious ligands and protein overexpression, which can lead to confusing results. In vitro analysis requires at least partial purification of the protein, which might itself affect its function, and no in vitro functional analysis completely matches the complex environment within the cell. Additionally, analysis of variants in which functional effects are seen must carefully consider whether these effects are direct or indirect. Finally, any results with protein variants must be extrapolated to the WT protein and the confidence in this extrapolation can sometimes be low. Despite these concerns, the careful use of mutational approaches, coupled with in vivo and in vitro analyses of functional interesting proteins, has been extremely informative and provides the bulk of the information on CooA function. The general insights from these various approaches is briefly summarized below and detailed in the subsequent sections of this chapter. The long  helices at the CooA (and CRP) dimer interface (Fig. 1), termed the “C-helices”, are organized in a leucine zipper motif. This is a helix:helix (coiled-coil) interaction where the “a” and “d” residues of a heptad repeat interact with pockets formed in the other helix, reminiscent of a zipper [64,65]. However, the precise relative positions of these  helices are different in active CRP and inactive CooA, which led to the hypothesis that C-helix repositioning upon CO binding might be a signal transduction pathway in CooA [14]. Because of the expected rigidity of these helices and the fact that they form the dual distal heme pockets of CooA, a repositioning in the vicinity of the heme might be transmitted to the hinge region at the boundary between the effector- and DNA-binding domains. Experimental support for this pathway in CooA is detailed in Section 4.3. The notion that cAMP binding in CRP can also affect C-helix repositioning has also been proposed [18], so it might be that this is a general mechanism for many members of the protein family. When active, both CooA and CRP bind specific DNA sequences with high affinity. Their biological effect of transcription activation results from their interactions with RNA polymerase. While the interactions of the two proteins with RNA polymerase do not appear to be identical, they are certainly rather similar in general outline and this is detailed in Section 4.5 below.

4. HYPOTHESIS FOR ACTIVATION OF CooA BY CO The analysis of CooA by a number of labs has led to a wealth of experimental data, which is difficult to appreciate unless it is presented in the context of an overall model. For that reason, we start with the following hypothesis for the overall behavior of CooA. Some aspects of this hypothesis are strongly supported by data, while others are not, and the nature of experimental evidence is described in subsequent sections. Despite these uncertainties, the model should provide the reader with the necessary context for thinking about detailed results. In the R. rubrum cell, CooA presumably exists in three general heme states [16]. Under oxidizing conditions, it is in a six-coordinate, low-spin Fe(III) form that is unable to bind CO, and is also incompetent to bind its specific DNA sequence with high affinity. When conditions become strongly reducing, an Fe(II) form is created, which

508

G.P. Roberts et al.

involves a conformational change accompanying a switch of one of the heme ligands (Section 4.1). This Fe(II) form is also six-coordinate and low-spin and displays poor DNA affinity. When Fe(II) CooA is exposed to CO, transient de-ligation of Pro2 allows binding of CO. This displacement of the Pro2 ligand by CO allows heme movement with respect to the protein, although the heme is still tethered to the His77 ligand trans to CO (Section 4.2). The distal pocket of CooA (on the Pro2 side of the heme) is formed by the C-helices and is quite hydrophobic, so that Pro2 deligation leads to its protonation and apparent expulsion from the hydrophobic heme pocket. This Pro2 expulsion exposes that hydrophobic pocket to solvent. We imagine that the C-helices are repositioned in part to address this pocket exposure. The movement of the CO-ligated heme within the protein results in a repositioning of the C-helices at the dimer interface with respect to each other (Section 4.3). The rigidity of the C-helices transmits this repositioning to the hinge regions between the effector- and DNA-binding domains. The movement helps to destabilize the inactive form of CooA and stabilize the active form, though other signal pathways are likely present between the heme and the DNA-binding domain. This active form is then able to bind specific DNA sequences adjacent to appropriate promoters (Section 4.4) and, through protein–protein interactions with RNA polymerase, activate transcription (Section 4.5).

4.1. R. rubrum CooA as a Redox Sensor: Ligand Switch, Conformational Change and Redox Poise As mentioned earlier, Cys75 is a ligand to the heme in the Fe(III) form of CooA [66], but is replaced by His77 upon reduction to the Fe(II) form [12,13]. Apart from the fact that a ligand switch occurs and must affect the redox poise, little is known about the mechanism of the Fe(III) to Fe(II) transition or, indeed, of the Fe(III) state itself. This unusual ligand switch must involve a significant conformational change within the protein. In the structure of the Fe(II) form of CooA, the ligand to the Fe(III) form, Cys75, lies toward the interior of the protein (and away from solvent) relative to the position of His77 (Fig. 2). This implies that the heme and protein must move approximately 2.5 Å relative to each other for this ligand switch to occur (Fig. 2A). This movement is presumably made possible by the apparent flexibility of the protein arm attached to the Pro2 ligand on the other side of the heme, since Pro2 appears to be the trans ligand in both forms. The midpoint potential of the heme in CooA has been determined to be about −300 mV using an optically transparent thin-layer electrochemical cell [10]. Curiously, those authors report a hysteresis in the titration curve dependent upon oxidation/reduction direction, which might suggest the existence of a very slow step, probably the unusual ligand exchange upon oxidation/reduction. However, the same authors found a rather fast millisecond time frame for this ligand exchange [67]. One possibility for this discrepancy is that the protein was not coming to equilibrium in the time frame of the experiment because of the relatively thick (1 mm) cell used but this paradox has not been further investigated. Other than WT CooA, only a H77G substitution has been examined for its effects on R. rubrum CooA redox behavior [67]. It would therefore be useful to systematically examine the redox properties of CooA variants altered in the heme environment in order to better understand this property of CooA. Because

CooA: A Paradigm for Gas-sensing Regulatory Proteins (A)

509

(B)

(b) Leu120

Cys75

(a) Cys75

(a) His77

(a) IIe95

(b) (b) Gly117 Pro2 (a) Leu116

Pro2 His77

(b) IIe113

Fig. 2. (A) The Fe(II) CooA heme-region residues. This representation expands the area around the left-hand heme. For each residue shown, the prefix “(a)” or “(b)” refers to the protein monomer to which it belongs, with chain “a” lightly shaded and forming the majority of the heme environment. Residues 113, 116, 117 and 120 lie along the C-helices shown on the right side of this panel. (B) The Fe(II) CooA heme pocket surface. This representation illustrates the same heme region as Panel (A), adjusted to show only the potential heme Fe ligands (Pro2, Cys75 and His77) and the calculated surface of the remaining residues. The white line demarcates the surface contributions of the protein monomers, shaded as in Panel (A). This view illustrates the protein cleft into which the heme is inserted, and indicates that the Cys75 ligand of the Fe(III) heme iron is more interior than the His77 ligand of the Fe(II) form. As discussed in the text, one model of CooA activation envisions movement of the CO-ligated heme into a void (not explicitly shown here) formed by the “back” edge of the heme and the “upper” interior surface of the heme pocket.

R. rubrum is fully capable of an aerobic lifestyle, such redox regulation is likely to be of biological significance environmentally. The conformational change during the transition from the Fe(III) to the Fe(II) form has the potential to create problems for the protein’s subsequent activation. Specifically, it has been found that some CooA variants with a perturbed Pro2 ligand can be activated if CO is added to the Fe(III) form and then reduced, but that adding CO to the Fe(II) form itself results in poor activation [46]. This has been interpreted to mean that without the normal Pro2 ligand in the Fe(II) form, the protein achieves a conformation that is capable of CO binding, but no longer can be efficiently converted to the active form. Presumably the presence of CO at the time of reduction makes the transition through the Fe(II) state sufficiently brief that this inactive conformation can be avoided. The authors attributed this to a collapse of the distal heme pocket upon reduction when Pro2 is absent, though other hypotheses cannot be excluded [46]. This result highlights not only the complexity of the redox transition, but also implies another role for the proper Pro2 ligand in CooA function, namely precluding the creation of functionally inactive CO-bound Fe(II) species. The redox properties of the other identified CooA homologs are strikingly different from those of CooA of R. rubrum, though they have not been evaluated in the organisms

510

G.P. Roberts et al.

that naturally produce them. By far the best studied of these is from Carbodxydothermus hydrogenoformans since it accumulates well when expressed in E. coli [1]. This homolog does not appear to undergo a ligand switch upon oxidation, even though it has a Cys residue at the analogous position of Cys75 of R. rubrum: a Ser substitution at that Cys has little effect on UV/Visible spectra [1]. This CooA homolog also has a very different redox poise than does CooA from R. rubrum, initially evidenced by its mixture of Fe(II) and Fe(III) forms when isolated from E. coli (under conditions where CooA of R. rubrum would be in the Fe(III) form). This homolog has been further characterized by the Aono group where they showed that His82 (the His77 analog) is the heme ligand in both Fe(II) and Fe(III) forms and that the midpoint potential of this homolog is approximately +200 mV, which is roughly 500 mV higher than that of R. rubrum [58]. Though the other homologs were not studied in this way due to poor accumulation in E. coli, they all lack an analogous Cys residue, so there is no reason to suppose that they undergo a redox-mediated ligand switch either. The general conclusion is that the ligand switch and the very low midpoint potential of R. rubrum CooA are probably unique in the current set of CooA homologs. It therefore appears unlikely that redox regulation is an important aspect of the function of the other CooA homologs. Why should R. rubrum have redox regulation of CooA, where the other organisms do not? In some cases the homologs are found in organisms that are strict anaerobes, so O2 should not be found in the environment in any event. However, Azotobacter vinelandii is an obligate aerobe, so a different explanation is necessary. The primary explanation appears to be that this organism has an efficient respiratory system that reduces cytoplasmic O2 levels, a mechanism that has been termed “respirator protection” when the effects on O2 -sensitive nitrogenase have been considered [68]. Nevertheless, the regulation of the nif genes, encoding the nitrogenase system, does show a level of O2 sensing, implying that the respiratory system is not completely efficient [69]. It might simply be that respiration is adequate for the needs of the CO-oxidation apparatus, but not for that of nitrogenase, or it might be that additional factors protect the CO-oxidation system of this organism. In any event, it appears that other identified CooA homologs exist in organisms for which redox sensing is not important for CooA function. In R. rubrum, a facultative anaerobe that certainly faces fluctuating redox conditions in its aquatic environment, such a sensing system in CooA is apparently an advantage.

4.2. The Immediate Effect of CO Binding to the Heme of CooA When Fe(II) CooA is exposed to CO, a distinct UV/Visible spectral shift occurs, consistent with CO binding to the heme and the creation of a six-coordinate low-spin CO adduct [9]. The identity of the displaced ligand was unclear for some time, but was conclusively shown to be Pro2 by the Aono group using NMR spectroscopy [39]. This is consistent with the characterization of CooA variants showing that Pro2 was relatively unimportant for that response [40], while the presence of His77 was essential for any significant response of CooA to CO [12,13]. What can we presently say about the nature of that CO-bound active form of CooA? The fact that Pro2, and indeed the general nature of the N-terminus of the protein, is unimportant for the response of CooA to the presence of CO disproves the hypothesis that the displaced N-terminus is itself a critical trigger for activation. This leads to

CooA: A Paradigm for Gas-sensing Regulatory Proteins

511

the alternative view that displacement of Pro2 allows a protein conformational change because of a repositioning of the heme itself. The notion that heme positioning might be important is consistent with the absolute requirement for His77, which presumably serves as a tether to the heme in that repositioning. Another insight into the active form is provided by the geminate recombination results, which indicate that the heme-bound CO resides in a tight small pocket [70]. This pocket cannot be formed by the N-terminus, however, as resonance Raman analysis has shown that this region is not close to the bound CO [60]. Similar spectroscopic analysis has been performed on a variety of CooA variants altered in residues predicted to be in the vicinity of the bound CO (Fig. 2A), based on the structure of Fe(II) CooA, though the precise relevance of this structure to that of the CO-bound form is unknown. Some substitutions of residues 116, 117 and 120 on the C-helices do perturb CO-stretching frequencies, which suggests a certain proximity of those residues to the CO [60,63]. However, one cannot rule out indirect effects of these substitutions on the CO vicinity through perturbations of other portions of the protein structure. The indirect effects of substitutions are a particular issue in this region of the protein, since this region is involved in a coiled-coil dimer interface through which protein information is transmitted (see Section 4.3). It is therefore difficult to know how to interpret spectral analyses of the CO-bound form of CooA variants for which no effort has been made to separate direct and indirect effects, or to correlate spectral perturbation with DNA-binding activity [63,71]. Part of the difficulty in predicting the location of the CO-bound heme is that CooA has already demonstrated a remarkable flexibility of heme positioning with respect to the protein. First there is the heme movement that must be taking place during the redox switch, as noted above. Second, there is the fact that the mutational alteration of the known axial ligands – Cys75 in the Fe(III) form, His77 in the Fe(II) form or Pro2 in either form – often yields variants that are predominately six-coordinate even in the redox condition that must be perturbed [13,21,43]. This result is also obtained when the introduced residues cannot possibly serve as heme ligands and in some cases, water and hydroxyl ion have been ruled out as possibilities as well [46]. This suggests that there are other protein ligands that can serve as adventitious ligands. However, an examination of the known Fe(II) structure reveals no obvious candidates for these adventitious ligands. We therefore believe that more distant residues can become ligands through substantial movement of the heme with respect to the protein. Despite these technical challenges, several interesting observations have been made that provide some insight into the position of the heme in the active form. The first is that CooA variants altered at the Pro2 ligand often display more of a five-coordinate population in the Fe(III) form than in the Fe(II) form [40]. This is consistent with there being a less stable N-terminal ligand in the Fe(III) form because the heme must remain close to the Cys75 residue, which lies farther than His77 from the surface of the protein and therefore from the N-terminus of the other subunit. The second observation concerns the results of in vivo screens for CooA variants with effector-independent activity. These screens are done under both aerobic and anoxic growth conditions and more variants with activity are typically found with aerobic screening (R.L. Kerby, H. Youn and G.P. Roberts, unpublished results). These results, confirmed by more quantitative assays, might mean that the heme position in CO-bound CooA is more similar to that of Fe(III) CooA than of Fe(II) CooA, though other explanations are possible. The final observation

512

G.P. Roberts et al.

concerns the effect of an I95W substitution, which is a residue near the bottom of the protein and extends toward the “bottom” of the heme from the His77 side (Fig. 2A). This substitution has little effect on the CO activation of otherwise WT CooA, but significantly increases the activity in the presence of CO in many variants that otherwise respond poorly [22]. This is most consistent with a hypothesis that the steric interaction created by the bulky Trp95 forces the heme to a new position similar to that found in the active form of WT CooA. Where is the “new position” of the heme-CO moiety in active WT CooA? Our activation model posits a modest C-helix repositioning and the shift of the CO-ligated heme toward the protein interior. This shift requires an expanded interior pocket to accept the heme-CO moiety, and we noted the presence of internal cavities adjacent to the heme in the inactive Fe(II) CooA structure. While these cavities were too small to accommodate the postulated heme movement, their expansion in the active state was envisioned to create the necessary void into which the heme-CO moiety could insert [60]. Such internal cavities are now seen to be crucial for the activity and ligandbinding characteristics of other hemoproteins [72], and it is reasonable to expect that their presence in CooA, adjacent to the heme, is likewise associated with its function.

4.3. The Role of the CO-bound Heme in Activation The following section describes the importance of helix repositioning in response to CO for activation of CooA, but how might CO binding to the heme effect this repositioning? To some extent, the answer to this can only be known when we have a firmer understanding of the active form of CooA, but we have some insights. The most important one came about through the analysis of the functional requirements of the residues that lie near the heme in Fe(II) CooA on the Pro2 side where CO binds. Specifically, we randomized C-helix residues 113 and 116 (Fig. 2A) and demanded CooA variants that displayed substantial activity in the presence of CO. The clear rule was that a variety of residues were acceptable at these positions (with position 116 being somewhat more stringent), but that hydrophilic residues were not. Indeed the specific creation of hydrophilic residues at two positions in this pocket created CooA variants that responded poorly to CO [22]. We believe that this requirement for hydrophobicity is most easily rationalized by positing that CO binding exposes this hydrophobic pocket to solvent by the expulsion of the N-terminus. This then would create a driving force that helps reposition the C-helices as detailed below. Similar experiments were performed on positions 117 and 120, with the result that the normal residue at each position, Gly117 and Leu120, is critical for CO-activation of CooA [43,73]. It appears likely that these residues have a role in C-helix repositioning, rather than in specific CO recognition, for reasons discussed in Section 4.5 below. In summary, we know some of the critical residues for the heme-CO-induced helix repositioning, but the precise basis for the heme effect in the process remains to be elucidated. How might CO binding and release of the Pro2 ligand lead to this helix repositioning? We currently favor two possibilities that are not exclusive of each other. The first involves the hydrophobic distal heme pocket (Fig. 2B), which becomes exposed to solvent upon CO binding and expulsion of the Pro2 arm from that region. It seems reasonable that the repositioning of the C-helices is actually driven by the favorable

CooA: A Paradigm for Gas-sensing Regulatory Proteins

513

energetics of collapsing that hydrophobic pocket to exclude solvent. The other possibility involves the cavities described above. It might be that the positioning of the CO-bound heme in these, together with its interactions with the C-helices, helps stabilize their repositioned conformation. This hypothesis is consistent with recent observations with a variety of CooA variants altered in the C-helices [60]. It is also attractive that such an active heme position would substantially overlap the position occupied by cAMP in the active form of CRP. Obviously there is a great need for experimental evidence to test these and other hypotheses. As will be discussed in a bit more detail below, CooA from the thermophile Carboxydothermus hydrogenoformans not only is activated in response to CO, but is also able to form a six-coordinate NO adduct that is active both in vivo and in vitro at 37 C [4]. This result is consistent with the above hypothesis for the requirement of an apolar heme ligand for proper activation. This NO responsiveness is almost certainly physiologically irrelevant, since raising the temperature of the NO-bound species causes a five-coordinate heme species to appear, which is presumably inactive for DNA binding [4].

4.4. Helix Repositioning is a Major Pathway of Signal Transduction within CooA In any signal-sensing protein, there must be one or more pathways within the protein through which the binding of the effector signal is transmitted to the portion of the protein that responds. In the case of CooA the effector signal is the binding of CO to the heme, and in the case of CRP it is the binding of cAMP. In both cases, however, these binding sites are far from the F helices that interact with DNA. Indeed, in neither case are they immediately adjacent to any part of the DNA-binding domain, which is the portion of the protein whose movement is the actual response. So how is the signal transmitted? More specifically, Is there more than one signaling pathway within CooA and are any of the pathways conserved in the CRP superfamily of proteins? As noted previously, the comparison of inactive CooA with active CRP revealed a subtle difference in the relative positioning of the two long C-helices at the dimer interfaces. In each case, these helices interact with each other through a leucine zipper motif. It is notable, however, that the heptad repeat in the vicinity of the cAMPbinding site of CRP is poor, as is the homologous region of CooA (positions 121–126); where a Leu would be optimal for a leucine zipper at the critical “d” position, CRP has Thr and CooA has Cys. This is consistent with the notion that the sub-optimal leucine zipper residues allow protein flexibility, and that binding of the effector shifts this structure toward helix positioning appropriate for activation. This view of helix flexibility is further supported by the observation that the C-helices in CRP and CooA lack the stabilizing interhelical salt bridges that are often found in such leucine zipper motifs [64,65]. We therefore reasoned that a deliberate mutational shifting of the helix positioning should “short circuit” the signal pathway and provide CooA activity in the absence of CO. Positions 121–126 of CooA were completely randomized using synthetic oligos and mutated cooA alleles were screened for those rare variants that were active without effector [44]. Without exception, these variants with effector-independent activity possessed a 121–126 region that was closer to the optimal consensus leucine

514

G.P. Roberts et al.

zipper motif. The correlation between an improvement in the leucine zipper and effectorindependent activity strongly suggests that repositioning of the C-helices is an aspect of CooA activation. It was also clear from the analysis of even the best of these leucine zipper variants that CO addition always caused a further increase in their DNA binding activity [44]. This observation might simply reflect a technical matter that the helices have not been repositioned in precisely the same way as in the CO-bound WT CooA. Alternatively, it might suggest that helix repositioning alone is not sufficient to fully activate CooA. We currently believe that CO binding and Pro2 release might allow the protein loop that contains the heme-bound His77 to move as well. This movement would affect the 4/5 loop through the peptide backbone loop (Fig. 1A) and might serve as the second signal pathway. This 4/5 loop occupies a very different position in active CRP compared to inactive CooA, so that a direct repositioning of the 4/5 loop might simultaneously destabilize the inactive form and stabilize the active form. It is reasonable that a CooA variant that bypasses only one of the normal CO-signaling pathways within the protein would display an incomplete shift in the CooA population to the active form, resulting in lower detected activity. Though not the subject of this review, we should comment briefly on the situation in CRP. We performed a randomization of the two CRP residues on the C-helix that make contact with the bound cAMP, and one of these is the critical “d” position residue analogous to position 123 of CooA. This library was analyzed and all effectorindependent variants had an improved leucine zipper motif, consistent with repositioning about the C-helix as a more general signal pathway within the protein superfamily [51]. That analysis also revealed a secondary signal pathway within CRP in which the bound cAMP appeared to directly affect the position of the 4/5 loop, a notion previously predicted by structural analysis [18]. The mechanism for 4/5 loop repositioning is rather different than that envisioned for CooA, indicating that there are both generalities in signaling, as well as protein- and effector-specific aspects to it.

4.5. Nature of the Active Form of CooA Ultimately, the signal of CO binding to the heme must lead to the massive rearrangement of the DNA-binding domains with respect to the rest of the protein as shown in Fig.1. We assume that both the active and inactive forms of CooA have a number of intraand inter-subunit interactions that stabilize each form. CO binding provides sufficient energy to shift that equilibrium toward the active form. We currently imagine that at least two sorts of alterations are involved in this equilibrium shift. The first is alteration of the hinge between the DNA- and effector-binding domains that results from the C-helix repositioning. In CooA, this hinge is at Phe132, while in CRP, it is at the analogous Phe136. Structural analysis has suggested an importance for this residue in CRP because of necessary interactions with the 4/5 loop in the active form [18], and mutational studies have confirmed its importance in CooA (H. Youn and G.P. Roberts, unpublished results). No doubt there are other local effects on the D-helix, which exists immediately beyond the hinge, and its interaction with both domains (Fig. 1B). In any event, the repositioning of the hinge positions must lead to

CooA: A Paradigm for Gas-sensing Regulatory Proteins

515

a destabilization of certain interactions that stabilize the inactive form and allow new interactions that stabilize the active form. The other mechanism of response to CO almost certainly involves movements within the effector-binding domain itself. In part this view reflects the very different position of the 4/5 loop in active CRP as compared to inactive CooA. These different positions each allow completely different sets of interactions with other portions of the protein that must be important for the stability of each form. As a working hypothesis for this movement, we note that the release of the heme from the Pro2 ligation might allow the movement of the His77 region as well. This region connects rather directly to the base of the 4/5 loop, as suggested by the breakage of the His77–Asn42 hydrogen bond upon CO binding [60]. Because Asn42 is several residues from the base of the 4/5 loop, this provides a plausible mechanism for communication between the 4/5 loop and CO binding. Finally, the active form of CooA must bind DNA, in order to promote transcriptional activation as described in the following section. The DNA sequence bound by CooA is reminiscent of those bound by CRP and FNR. This fact is reflected in the rather similar sequences of the F helices in these proteins, which is the region that makes the base-specific contacts in the known CRP structure and other examined helix-turn-helix proteins [16]. With CRP, a consensus sequence was determined through the analysis of the many sequences bound by that protein in E. coli [74]. However, CooA only has two known binding sites in R. rubrum and they are sufficiently similar that we cannot make predictions about the nature of the DNA sequence that would support the highest CooA affinity.

4.6. Basis for Transcription Activation by CooA The biological function of CooA (and CRP) is to bind a specific DNA sequence immediately adjacent to a promoter with relatively poor affinity for RNA polymerase and then to recruit RNA polymerase to that promoter through protein–protein interactions. Effectively, CooA binding converts a poor promoter into a strong promoter by this mechanism. These contacts between the regulatory protein and RNA polymerase have been studied extensively with CRP and with the related protein FNR. CRP and FNR actually happen to have two rather different classes of promoters at which they function, where the distinction concerns their precise binding site with respect to RNA polymerase [75–78]. Because both CooA sites in R. rubrum are of the “class II” type, where the protein-binding site is immediately adjacent to that of RNA polymerase, we will restrict our discussion to that type [49]. The sites on the activator proteins that make specific contacts with different portions of RNA polymerase are called Activating Regions (AR). These contacts are positioned in three general regions of these proteins. Their regions of contact on the several subunits of RNA polymerase have also been defined to varying degrees [76,77,79–81]. The general model for CooA interaction with RNA polymerase (based largely on that of CRP) is depicted in Fig. 3 and is supplemented with related information on the action of FNR at similar sites. The first interaction site, termed AR1, occurs in the DNA-binding domain (Figs 1 and 3), though there is evidence for some contacts in the proximal region of the effector-binding domain in FNR [82]. This region makes contact

516

G.P. Roberts et al. CooA AR2

αCTD AR1

AR3

–43.5

β′

α-NTD σ

β

–10

Fig. 3. Schematic of the interaction of CooA with RNA polymerase. The cartoon shows the CooA dimer binding at its target sequence immediately 5 of a promoter. As indicated, the AR1 of the 5 CooA monomer interacts with the carboxyl-terminal domain (CTD) of the -subunit of RNA polymerase. The AR2 and the AR3 of the 3 monomer of CooA interact with the amino-terminal domain (NTD) of  and with , respectively. Transcription will proceed to the right in this figure.

with the C-terminal domain of the alpha (-CTD) subunit of RNA polymerase, which lies on the end of a flexible linker region that allows this domain to extend away from the rest of the holoenzyme and actually make contact exclusively with the promoter distal monomer of the DNA-bound activator protein. The second set of contacts is termed AR2 and represents an interaction between the N-terminal domain of alpha (-NTD) and the effector-binding domain of the activator, typically on the portion of that domain that lies farthest from the DNA. Finally, AR3 is found in a loop (the 4/5 loop of CRP and CooA) and contacts the sigma subunit of RNA polymerase. The AR2 and AR3 regions are only functional in the promoter proximal monomer [83], because of the proximity of this monomer to the appropriate subunits of RNA polymerase (Fig. 3). The analysis of the AR regions of CooA has been performed with the normal PcooF DNA sequence and with RNA polymerase from E. coli. Because the analyses used the same RNA polymerase as used in the CRP and FNR analyses, the results demonstrate the general functionality of the protein and therefore reveal a more general set of rules about the protein family. The results indicate that CooA also has three activation regions and, at least for the interaction with the major form of RNA polymerase of E. coli, all three ARs appear to be of functional significance. The presence of an AR1 was revealed by an analysis in which the presence of -CTD was shown to be critical for CooA-dependent transcription in vitro [84] (in contrast, -CTD is not essential for strong promoters that do not involve positive activators). Several substitutions on the surface of -CTD were specifically defective in CooAmediated activation in vitro and these residues occur in a patch that is near to one involved in CRP interaction [85]. The presence of AR2 and AR3 of CooA has been revealed by a mutational analysis, coupled with structural comparisons [84]. A genetic screen was performed starting with a CooA variant with a low level of DNA-binding activity in the effector-free form and variants with enhanced activity were sought after random mutagenesis of the portion of cooA that encoded the effector-binding domain. Mutants with improved activity were found, sequenced, moved to a WT background, purified and analyzed in in vitro DNA binding assays. The variants displayed normal affinities for DNA, as well as normal accumulation in the cell, consistent with the hypothesis that the basis of their altered activity was through enhanced interaction with RNA polymerase. This was confirmed

CooA: A Paradigm for Gas-sensing Regulatory Proteins

517

in in vitro transcription assays, where they displayed transcriptional activity higher than WT CooA. Appropriate loss-of-function variants were also created and examined with expected results [84]. A final curiosity should be noted with R. rubrum CooA. It has a requirement for mM levels of divalent cations in order to bind its target DNA sequence [86]. Such a requirement is highly unusual, since the presence of divalent cations typically reduces the affinity of DNA-binding proteins for their target DNA [87,88]. The source of this requirement has been localized to a specific residue, Glu167, in the E-helix that lies immediately adjacent to the DNA and apparently lowers DNA affinity through charge repulsion in the absence of divalent ions. The divalent ions apparently bridge Glu167 and other negatively charged regions of the protein to draw Glu167 away from the DNA. Because the other CooA homologs have the same residue, they should also have such an effect, though this has not been examined. In the active form, the role of this residue is to provide an improved contact with RNA polymerase and it appears to serve as an AR3 surface [86]. CRP has Gln at the homologous position and apparently creates the appropriate interaction with RNA polymerase through completely different AR contacts. Because of the presence of mM levels of divalent cations in the cell, this requirement probably does not have a physiological impact on CooA in the cell. However, it does mean that the results in any paper where DNA interaction was supposed, but where cations were not supplied, must be viewed with skepticism [62].

5. BASIS FOR THE SPECIFICITY FOR CO IN CooA ACTIVATION Biological sensors must provide the appropriate specificity for a given effector molecule, but that does not mean that they must be highly specific. Rather, they must discriminate between correct and incorrect effectors at the levels in which they find them in the cell. For that reason, there is no particular reason for a sensor protein to discriminate against an effector that it never encounters in nature. It is striking, then, that CooA of R. rubrum displays such exquisite selectivity for CO. As noted before, the protein ligands of CooA appear to have been chosen to provide a major part of this selectivity. Relatively weak small molecule ligands are unable to displace either protein ligand and cannot bind the heme. O2 can presumably bind the heme, but this leads to heme oxidation and the accompanying ligand switch, which in turn keeps CooA in an inactive form. In contrast, the very strong NO ligand displaces both protein ligands and fails to activate CooA [41]. CO not only displaces a single ligand, but apparently only the correct ligand, Pro2. A variety of mutagenic analyses have strongly suggested that continued tethering of the heme to the His77 ligand is critical, at least for proper response to CO. Thus the specificity of CooA for CO could be explained simply by positing that only CO is able to displace Pro2 without displacing His77. The following arguments show that this view is too simplistic, however. In the course of analyzing the role of the unusual Pro2 ligand in CooA function, a variety of substitutions were made in that vicinity of the protein, one of which was a two-codon deletion, removing residues Pro3 and Arg4. Not surprisingly, CooA variants with this P3R4 change have difficulty in stably connecting Pro2 to the heme, as revealed spectroscopically by the presence of some five-coordinate high-spin species

518

G.P. Roberts et al.

in both Fe(III) and Fe(II) forms of P3R4 CooA [46]. Given this population of open coordination sites, it was also not surprising that imidazole and CN− were able to bind to this protein, forming what appear to be fully six-coordinate Fe(II) species [22]. We assume that these small molecules are binding on the same side as that normally bound by CO in WT CooA, as it is the perturbation of Pro2 that has allowed their ligation. Given the simple model above that the specificity of CO activation is based only on its ability to create such a species, we expected to see a significant amount of DNA-binding ability of P3R4 CooA in response to imidazole and CN− . However, DNA binding as determined by fluorescence anisotropy was negligible. P3R4 CooA does have significant activity in the presence of CO, showing that the variant is capable of activation with the proper small molecule ligand. This result implies that, while the ability of a small molecule to displace Pro2 is important, there is some additional mechanism of specificity in CooA for CO [22]. If the precise nature of the bound small molecule ligand is important to activity, one model for this level of specificity could be that there should be some sort of specific contact between the protein and that ligand. This is a bit problematic, because at present we do not know exactly where the heme-bound CO is positioned with respect to the protein backbone. We initially supposed that the heme might be in the same general vicinity of the protein as it is in the Fe(II) form and analyzed the nearby residues. This was the original rationalization for the randomization experiments with positions 113 and 116 described above. As explained already, however, a variety of differently sized hydrophobic residues at these positions were acceptable for normal CO-sensing function, which rules out these residues as specificity determinants by direct interaction with CO. We also found that Gly117 [45] and Leu120 [73] are critical residues for CO activation. However, the notion that one or both might serve as a specificity determinant for CO is undercut by the following results. CooA variants have been found that bind and are activated by imidazole, and both Gly117 and Leu120 are critical for this imidazole responsiveness [73]. Because Gly117 and Leu120 are necessary for response to both CO and imidazole, it appears that their role is in a common conformation change during activation, rather than a specific interaction with CO. Finally, although the precise positioning of the heme-bound CO remains unclear, resonance Raman analysis of a variety of CooA variants does strongly suggest that it falls somewhere in this region [60,63]. We therefore have a paradox concerning the absolute requirement for CO as the activating ligand. The inability of small molecules to activate P3R4 CooA, though they certainly bind, implies that some residues in the heme environment sense the nature of the bound ligand. However, changing the obvious candidate residues in the heme vicinity did not appear to cause a substantial effect on the ability of P3R4 CooA to sense CO, as if these residues are not making specific contacts with the bound CO. Nevertheless, changing these same residues did allow variants of P3R4 CooA to become active in response to imidazole binding [22,74]. We currently favor the following hypothesis for these results. If activation of CooA involves movement of the heme within a hydrophobic cleft, then there actually might be no residues that provide a specific interaction with the bound CO. Rather, imidazole and CN− might fail to activate for a different reason. The presence of a bulky group such as imidazole might simply preclude that heme movement. Similarly, the failure of CN− to activate might reflect its negative charge, which might attract a water molecule and would again preclude the necessary heme movement.

CooA: A Paradigm for Gas-sensing Regulatory Proteins

519

The ability of CooA variants with changes at 113 and 116 (in a P3R4 CooA background) to respond to imidazole would therefore reflect a very different site of interaction between the heme and the C-helices, which coincidentally leads to a similar repositioning of the C-helices. By this hypothesis, CO binding allows the heme to move to a position that is necessary for the conformational change leading to activation. The role of CO is presumably to provide a small, uncharged ligand on the heme, and also to affect heme positioning through a perturbation of the trans Fe His77 bond, as has been shown by resonance Raman spectroscopy [34,62]. The specificity for CO is therefore the result of a combination of size, ligand strength and absence of charge that apparently affects proper heme positioning through direct effects on both distal and proximal sides of the heme [73]. This hypothesis has been reinforced in a surprising way by the recent observation that a CooA homolog from C. hydrogenoformans is activated, both in vivo and in vitro, by both CO and NO at 37 C [4]. This result is consistent with the model, since NO is also a small apolar ligand. The curiosity here is that this CooA homolog remains six-coordinate upon NO binding, where R. rubrum CooA does not. A clue to the important difference between the two proteins is the fact that C. hydrogenoformans is a thermophile with an optimal growth temperature above 70 C [89]. When CooA of C. hydrogenoformans is raised to 70 C in vitro the NO-bound form becomes five-coordinate, which is presumably inactive for DNA binding [4]. It therefore appears that it is the relatively high stability of the CooA from the thermophile that happens to stabilize the six-coordinate NObound form at ambient temperatures, which in turn allows DNA binding. However, under normal (thermophilic) growth conditions, the CooA of C. hydrogenoformans would be expected to be as CO-specific as is that of R. rubrum at its normal growth temperature.

6. COOPERATIVITY OF CO BINDING In the course of the analysis of CooA, we have noticed that CO binding to the CooA dimer occurs with positive cooperativity [90]. This is interesting for several reasons: (i) It is of physiological significance, since CooA should turn on gene expression when reasonable levels of CO are in the cell, yet turn off that expression in a timely fashion as the CO is depleted through oxidation. This response would be greatly affected by not only the fact of cooperativity but also perhaps by its exact mechanism. (ii) The matter is biochemically interesting because there are a variety of ways in which the two monomers might communicate to support cooperativity and an understanding of that would speak to the underlying process of activation. (iii) The cooperativity of cAMP binding by CRP has been a subject of dispute for a number of years, with some groups arguing that it is positively cooperative and other suggesting that it is negatively cooperative. Many of the differences no doubt are a result of differences in assay conditions and it appears to be the case that CRP is actually positively cooperative under physiological conditions [55]. The physiological implication of this remains unclear because there is also a disagreement about whether or not CRP with a single cAMP bound is active [54,55]. Thus, understanding of CooA cooperativity and its mechanism might be instructive for

520

G.P. Roberts et al.

the entire CRP protein family. Finally, we should note that it remains unclear whether or not CooA with a single CO bound is predominately in the active or inactive form. The basis for the positive cooperativity of CooA is complex, since CO binding is itself a product of the relative on- and off-rates of both CO and the Pro2 ligands. Kinetic and equilibrium analyses appear to show that binding of CO to one heme in the dimer lowers the off-rate of the Pro2 ligand to the other heme (which by itself would yield negative CO cooperativity), but also lowers the on-rate of the Pro2 to the other ligand by a much greater degree. The resultant decrease in strength of Pro2 coordination has the net effect of positive cooperativity because this other heme is now more accessible to CO [90]. The biological implication of this cooperativity is that the regulatory system would respond to micromolar levels of CO by activating transcription, even if the CO levels were insufficient to saturate the entire population of CooA hemes. The kinetic analysis also revealed that there are two distinct kinetic forms of CooA, with distinctly different properties, that are in rather slow interconversion [90]. While the precise nature of these two forms remains to be determined, one biological implication of the slow interconversion might be that it acts as a buffer against rapid fluctuations in CO levels.

7. SUMMARY AND FUTURE DIRECTIONS CooA is a particularly powerful system for analyzing the impact of the heme environment on the functionality of a protein. This is in part because it is readily expressed and purified and because there is a substantial body of structural and biochemical information in hand. More importantly, however, CooA is a regulatory protein that is able to “report its own functionality” both through very simple color screens of bacterial colonies on agar plates and in in vitro DNA-binding assays. As a consequence, millions of different variants can be produced in mutagenic protocols and readily screened for extremely rare but interesting functions, which may be carefully analyzed in vitro. This system of in vivo screens for interesting variants is simply not available for most heme proteins. The in vitro analysis permits a correlation of spectral features and biological utility, avoiding the confusion caused by the analysis of functionally dead proteins. These tools have allowed us to both probe the critical residues for a normal response to CO and identify variants active under aerobic [19,44] or anaerobic [20,44] conditions, as well as those that can respond to NO [4] and imidazole [22,73]. Screens for variants sensing other effectors are ongoing. The results of these analyses not only inform us of the behavior of WT CooA to CO, but more importantly indicate the biochemical properties that underlie ligand specificity. While we have a broad outline of the activation pathway of CooA, there are many details to be confirmed and hypotheses to be tested. Particularly important is the analysis of the commonalities between the activation mechanism of CooA and CRP, since this will have implications for the much larger family of CRP homologs. We also need to learn more about the Fe(II) CO and Fe(III) forms of CooA. Crystal structures would be hugely valuable, but the population distribution of different forms in solution is a separate and interesting question.

CooA: A Paradigm for Gas-sensing Regulatory Proteins

521

ACKNOWLEDGMENTS This work was supported by the College of Agricultural and Life Sciences at the University of Wisconsin, Madison, and National Institutes of Health Grant GM53228 (to G.P.R.). The authors wish to thank Jose Serate for technical assistance throughout the work. We also thank colleagues in the labs of Judith Burstyn, Thomas Poulos, Thomas Spiro and John Olson for their intellectual input and support and for permission to communicate unpublished information.

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26]

[27] [28] [29] [30] [31] [32] [33] [34]

Youn, H., Thorsteinsson, M.V., Kerby, R.L. et al. (2005) J. Bacteriol., 187, 2573–2581. Kerby, R.L., Ludden, P.W. and Roberts, G.P. (1995) J. Bacteriol., 177, 2241–2244. Ensign, S.A., Bonam, D. and Ludden, P.W. (1989) Biochemistry, 28, 4968–4973. Clark, R.W., Lanz, N.D., Lee, A.J. et al. (2006) Proc. Natl. Acad Sci. USA, 103, 891–896. Hao, B., Isaza, C., Arndt, J. et al. (2002) Biochemistry, 41, 12952–12958. Dunham, C.M., Dioum, E.M., Tuckerman, J.R. et al. (2003) Biochemistry, 42, 7701–7708. Aono, S., Kato, T., Matsuki, M. et al. (2002) J. Biol. Chem., 277, 13528–13538. Dioum, E.M., Rutter, J., Tuckerman, J.R. et al. (2002) Science, 298, 2385–2387. Shelver, D., Kerby, R.L., He, Y. and Roberts, G.P. (1997) Proc. Natl. Acad. Sci. USA, 94, 11216–11220. Nakajima, H. and Aono, S. (1999) Chem. Lett., 1233–1234. Heo, J., Halbleib, C.M. and Ludden, P.W. (2001) Proc. Natl. Acad. Sci. USA, 98, 7690–7693. Aono, S., Ohkubo, K., Matsuo, T. and Nakajima, H. (1998) J. Biol. Chem., 273, 25757–25764. Shelver, D., Reynolds, M.F., Thorsteinsson, M.V. et al. (1999) Biochemistry 38, 2669–2678. Lanzilotta, W.N., Schuller, D.J., Thorsteinsson, M.V. et al. (2000) Nat. Struc. Biol., 7, 876–880. Körner, H., Sofia, H.J. and Zumft, W.G. (2003) FEMS Microbiol. Rev., 27, 559–592. Roberts, G.P., Youn, H. and Kerby, R.L. (2004) Microbiol. Mol. Biol. Rev., 68, 453–473. Weber, I.T. and Steitz, T.A. (1987) J. Mol. Biol., 198, 311–326. Passner, J.M., Schultz, S.C. and Steitz, T.A. (2000) J. Mol. Biol., 304, 847–859. Thorsteinsson, M.V., Kerby, R.L., Youn, H. et al. (2001) J. Biol. Chem., 276, 26807–26813. Youn, H., Kerby, R.L., Thorsteinsson, M. et al. (2002) J. Biol. Chem., 277, 33616–33623. Thorsteinsson, M.V., Kerby, R.L. and Roberts, G.P. (2000) Biochemistry 39, 8284–8290. Youn, H., Kerby, R.L. and Roberts, G.P. (2003) J. Biol. Chem., 278, 2333–2340. Aono, S. (2003) Acc. Chem. Res., 36, 825–831. Jain, R. and Chan, M.K. (2003) J. Biol. Inorg. Chem., 8, 1–11. Rodgers, K.R. and Lukat-Rodgers, G.S. (2005) J. Inorg. Biochem., 99, 963–977. Patschkowski, T., Bates, D.M. and Kiley, P.J. (2000) In Bacterial Stress Response (G. Storz and R. Hengee-Aronis, eds), Washington, DC: American Society for Microbiology, pp. 61–78. Lazazzera, B.A., Bates, D.M. and Kiley, P.J. (1993) Genes & Dev., 7, 1993–2005. Lazazzera, B.A., Beinert, H., Khoroshilova, N. et al. (1996) J. Biol. Chem., 271, 2762–2768. Unden, G. and Schirawski, J. (1997) Mol. Microbiol., 25, 205–210. Sutton, V.R., Mettert, E.L., Beinert, H. and Kiley, P.J. (2004) J. Bacteriol., 186, 8018–8025. Kolb, A., Busby, S., Buc, H. et al. (1993) Annu. Rev. Biochem., 62, 749–795. Harman, J.G. (2001) Biochim. Biophys. Acta., 1547, 1–17. Monson, E.K., Ditta, G.S. and Helinski, D.R. (1995) J. Biol. Chem., 270, 5243–5250. Gilles-Gonzalez, M.-A., Gonzalez, G. and Perutz, M.P. (1995) Biochemistry, 34, 232–236.

522

G.P. Roberts et al.

[35] Key, J. and Moffat, K. (2005) Biochemistry, 44, 4627–4635. [36] Gong, W., Hao, B. and Chan, M.K. (2000) Biochemistry, 39, 3955–3962. [37] Tuckerman, J.R., Gonzalez, G., Dioum, E.M. and Gilles-Gonzalez, M.-A. (2002) Biochemistry, 41, 6170–6177. [38] Rodgers, K.R., Lukat-Rodgers, G.S. and Brown, J.A. (1996) Biochemistry 35, 9539–9548. [39] Yamamoto, K., Ishikawa, H., Takahashi, S. et al. (2001) J. Biol. Chem., 276, 11473–11476. [40] Thorsteinsson, M.V., Kerby, R.L., Conrad, M. et al. (2000) J. Biol. Chem., 276, 39332–39338. [41] Reynolds, M.F., Shelver, D., Vogel, K.M. et al. (2000) Biochemistry, 39, 388–396. [42] Dhawan, I.K., Shelver, D., Thorsteinsson, M.V. et al. (1999) Biochemistry, 38, 12805–12813. [43] Vogel, K.M., Shelver, D., Kerby, R.L. et al. (1999) Biochemistry, 38, 2679–2687. [44] Kerby, R.L., Youn, H., Thorsteinsson, M.V. and Roberts, G.P. (2003) J. Mol. Biol., 325, 809–823. [45] Youn, H., Kerby, R.L., Thorsteinsson, M.V et al. (2001) J. Biol. Chem., 276, 41603–41610. [46] Clark, R.W., Youn, H., Parks, R.B. et al. (2004) Biochemistry, 43, 14149–14160. [47] Heyduk, E., Heyduk, T. and Lee, J.C. (1992) J. Biol. Chem., 267, 3200–3204. [48] Parkinson, G., Wilson, C., Gunasekera, A. et al. (1996) J. Mol. Biol., 260, 395–408. [49] He, Y., Shelver, D., Kerby, R.L. and Roberts, G.P. (1996) J. Biol. Chem., 271, 120–123. [50] Kalodimos, C.G., Biris, N., Bonvin, A.M. et al. (2004) Science 305, 386–389. [51] Youn, H., Kerby, R.L., Conrad, M. and Roberts, G.P. (2006) J. Biol. Chem., 281, 1119–1127. [52] Bonam, D. and Ludden, P.W. (1987) J . Biol. Chem., 262, 2980–2987. [53] Won, H.S., Lee, T.W., Park, S.H. and Lee, B.J. (2002) J . Biol. Chem., 277, 11450–11455. [54] Lin, S.-H. and Lee, J.C. (2002) Biochemistry, 41, 11857–11867. [55] Tomlinson, S.R., Tuitar, Y. and Harman, J.G. (2003) Biochemistry, 42, 3759–3765. [56] Akiyama, S., Fujisawa, T., Ishimori, K. et al. (2004) J . Molec. Biol., 341, 651–668. [57] Won, H.S., Yamazaki, T., Lee, T.W. et al. (2000) Biochemistry, 39, 13953–13962. [58] Inagaki, S., Masuda, C., Akaishi, T. et al. (2005) J . Biol. Chem., 280, 3269–3274. [59] Kumar, S.A., Murthy, N.S. and Krakow, J.S. (1980) FEBS Lett., 109, 121–123. [60] Coyle, C.M., Puranik, M., Youn, H. et al. (2003) J . Biol. Chem., 278, 35384–35393. [61] Uchida, T., Ishikawa, H., Ishimori, K. et al. (2000) Biochemistry, 39, 12747–12752. [62] Uchida, T., Ishikawa, H., Takahashi, S. et al. (1998) J . Biol. Chem., 273, 19988–19992. [63] Yamashita, T., Hoashi, Y., Tomisugi, Y. et al. (2004) J . Biol. Chem., 279, 47320–47325. [64] Lupas, A. (1996) Trends Biochem. Sci., 21, 1123–1134. [65] Burkhard, P., Stetefeld, J. and Strelkov, S.V. (2001) Trends Cell Biol., 11, 82–88. [66] Reynolds, M.F., Shelver, D., Kerby, R.L. et al. (1998) J . Am. Chem. Soc., 120, 9080–9081. [67] Nakajima, H., Nakagawa, E., Kobayashi, K. et al. (2001) J . Biol. Chem., 276, 37895–37899. [68] Poole, R.K. and Hill, S. (1997) Biosci. Rep., 17, 303–317. [69] Martinez-Argudo, I., Little, R. and Dixon, R. (2004) Molec. Microbiol., 52, 1731–1744. [70] Rubtsov, I.V., Zhang, T., Nakajima, H. et al. (2001) J . Am. Chem. Soc., 123, 10056–10062. [71] Yamashita, T., Hoashi, Y., Watanabe, K. et al. (2004) J . Biol. Chem., 279, 21394–21400. [72] Brunori, M., Bourgeois, D. and Vallone, B. (2004) J . Structural Biol., 147, 223–234. [73] Youn, H., Kerby, R.L. and Roberts, G.P. (2004) J . Biol. Chem., 279, 45744–45752. [74] Gunasekera, A., Ebright, Y.W. and Ebright, R.H. (1992) J . Biol. Chem., 267, 14713–14720. [75] Busby, S. and Ebright, R.H. (1999) J . Mol. Biol., 293, 199–213. [76] Niu, W., Kim, Y., Tau, G., Heyduk, T., Ebright, R.H. (1996) Cell, 87, 1123–1134. [77] Bell, A. and Busby, S. (1994) Mol. Microbiol., 11, 383–390. [78] Lawson, C.L., Swigdon, D., Murkami, K.S. et al. (2004) Curr. Opin. Struct. Biol., 14, 10–20. [79] Rhodius, V.A. and Busby, S.J.W. (2000) J . Mol. Biol., 299, 295–310. [80] Niu, W., Zhou, Y., Dong, Q. et al. (1994) J . Mol. Biol., 243, 595–602. [81] Williams, R., Bell, A., Sims, G. and Busby, S. (1991) Nucl. Acids Res., 19, 6705–6712. [82] Williams, S.M., Savery, N.J., Busby, S.J.W. and Wing, H.J. (1997) Nucleic Acids Res., 25, 4028–4034.

CooA: A Paradigm for Gas-sensing Regulatory Proteins

523

[83] Zhou, Y., Busby, S. and Ebright, R.H. (1993) Cell, 73, 375–379. [84] Leduc, J., Thorsteinsson, M.V., Gaal, T. and Roberts, G.P. (2001) J . Biol. Chem., 276, 39968–39973. [85] He, Y., Gaal, T., Karls, R. et al. (1999) J . Biol. Chem., 274, 10840–10846. [86] Youn, H., Kerby, R.L., Conrad, M. and Roberts, G.P. (2004) J . Bacteriol., 186, 1320–1329. [87] Lohman, T.M., deHaseth, P.L. and Record, M.T., Jr. (1980) Biochemistry, 19, 3522–3530. [88] Suh, W.C., Liermo, S. and Record, M.T., Jr. (1992) Biochemistry, 31, 7815–7825. [89] Svetlichny, V.A., Sokolova, T.G., Gerhardt, M. et al. (1991) System. Appl. Microbiol., 14, 254–260. [90] Puranik, M., Nielsen, S.B., Youn, H. et al. (2004) J . Biol. Chem., 279, 21096–21108.

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 19

Soluble Guanylyl Cyclase and Its Evolutionary Relatives Eduardo Henrique Silva Sousa, Gonzalo Gonzalez, and Marie-Alda Gilles-Gonzalez Department of Biochemistry, University of Texas Southwestern Medical Center, 5323 Harry Hines Blvd, Dallas, Texas, USA 75390-9038

Abstract Bioinformatic analysis of the heme-binding fragment of mammalian soluble guanylyl cyclase (sGC) has uncovered a domain called HNOB (Heme NO Binding) that is widespread in bacteria and eukarya. Some heme-containing HNOB domains can bind oxygen, by contrast to mammalian sGC, which binds nitric oxide but not oxygen. Additionally, the Drosophila melanogaster proteins Gyc-88E and Gyc-89Da and Gyc-89Db were found to govern hypoxia-activated cGMP synthesis in an in vivo assay, and the nematode Caenorhabditis elegans was observed to display aerotaxis behaviors that require the sGC-like protein GCY-35. A crystal structure of an oxygen-bound HNOB domain from the Thermoanaerobacter tengcongensis methyl-accepting chemotaxis receptor Tar4 (also called Tt-HNOX or Tt-SONO) has led to various proposals about motifs for heme binding and the occurrence of oxygen-binding HNOB domains. Overall, the data suggest a broad role for HNOB-coupled sensors in responding to diatomic gases. We examine these results, taking care to note some of the overstatements arising from the excitement generated by these novel heme-based sensors and to consider the outstanding questions. Recent advances on mammalian sGC research are also discussed, and a sGC switching mechanism is proposed that is consistent with the structural results and available data on sGC activators such as YC-1 and BAY58-2667. Keywords: Aerotaxis, cGMP, DosS, DevS, DosT, DevR, FixL, heme-based sensor, HNOB, HNOX, hypoxia, nitric oxide sensor, oxygen sensor, Rv2027c, SONO, YC-1

1. INTRODUCTION Guanylyl cyclases are signal-transducing enzymes that respond to specific stimuli by synthesizing the second messenger cGMP. The cGMP in turn directly regulates many activities, including kinases, diesterases, and cyclic-nucleotide gated ion channels. In the early 1980s, mammalian soluble guanylyl cyclase (sGC) was demonstrated to contain heme, and the ligation of nitric oxide (NO) to the heme was shown to enhance the cyclase activity 200- to 400-fold [1–3]. Nitric-oxide signaling via mammalian sGC triggers

sGC relatives

525

many important physiological events, including smooth muscle relaxation, inhibition of platelet aggregation, and neurotransmission [4–10]. Guanylyl cyclases are usually classified as being either soluble (sGCs) or transmembrane receptors (rGCs), sometimes called transmembrane particulate (pGC) [11,12]. Soluble GCs are activated by NO only when they are heterodimeric and comprised of an - and a - subunit. By contrast, the rGCs function as homodimers [12]. Each rGC polypeptide typically contains a single transmembrane region and an extracellular ligand-binding domain. The rGCs are activated either by their interaction with peptide ligands or with activating proteins [12]. This also appears to be a mode of sGC regulation, since several protein-binding partners are now known to control the cellular localization and activity of sGC [13–18]. For example, the 2-subunit of sGC, which is predominantly expressed in the brain, can bind to the post-synaptic density-95 protein (PDS-95) while other regions of this protein interact with the NMDA receptor and neuronal NOS [13]. The physiological relevance of this complex has not been explored, though it has been postulated to maximize NO signaling. Interestingly, another sGC interacting protein, CCTeta (chaperonin containing t-complex polypeptide) has been shown to depress the activity of NO-sGC by 30–50% without altering the sGC basal activity, suggesting that the interaction of CCTeta with sGC is specific to the NO-bound state [16]. An intense biochemical and pharmacological interest in mammalian sGC over the last three decades has led to a large body of literature on this protein, despite the considerable difficulties encountered in expressing and purifying it in its active and fully hemed state [19,20]. Several recent reviews cover sGC biochemistry, focusing on mammalian sGCs [21–30]. Here we discuss some recently discovered sGC homologs that provide an exciting new perspective to our understanding of these proteins. Although this discussion follows some essential background on methods for studying sGC biochemistry and previous findings about sGC regulation by small molecules.

2. BIOINFORMATIC ANALYSIS OF sGC FAMILY AND GENEALOGY In 2003, Aravind and colleagues identified a domain related to the heme-binding region of sGC and named it HNOB (Heme NO Binding) [31]. A phylogenetic-tree of the sGC family supported the original suggestion that the HNOB domains in eukaryotic organisms were acquired by horizontal transfer from bacteria, and particularly cyanobacteria [31,32]. The HNOB domains in eukaryotic, including mammalian, proteins usually occur in association with guanylyl cyclase regions, whereas those in bacterial proteins typically occur as stand-alone proteins or in combination with methyl-accepting chemotaxis receptor domains [31] (Fig. 1). A sub-group of HNOB domains contain a conserved histidine residue, a conserved proline residue, and a YxS/TxR motif [33–35]. These residues are proposed to represent a motif that participates in various heme contacts, including coordination to the heme iron (the histidine residue), based on mutagenesis studies of mammalian sGC and the first X-ray crystallographic structures for a sGC-like heme-binding domain [33–35]. Although the acronym “HNOB” suggests heme

526

Eukaryotes

Prokaryotes

E.H.S. Sousa et al.

[ [

HNOB

HNOB

HNOB

HNOB

HNOBA

HNOBA

MA

Guan_cyc

Guan_cyc

MA

Homo sapiens

Drosophila melanogaster

Thermoanaerobacter tengcongensis

Anabaena sp

Fig. 1. Common domain organizations of HNOB-coupled sensors. Domain organizations are shown for eukaryotic and prokaryotic sensors, as predicted by Pfam analysis: heme-binding HNOB domain (red); HNOBA domain that associates with HNOB but does not bind heme (green); guanylyl cyclase domain (Guan_cyc, yellow); methyl accepting domain of chemeotaxis receptors (MA, gray/black). (see Plate 15.)

binding by all members of this class, it might have many of these domains that do not bind heme. For example, the HNOB domains in the -subunit of mammalian sGC, the Rhodobacter sphaeroides uncharacterized protein Rshp2043, and the Magnetoccocus sp. uncharacterized protein Mmc10739, all lack the proposed heme-binding motif, including the conserved histidine for axial coordination to the heme iron [31]. Indeed, heme binding does not appear to be obligatory for any class of domain so far discovered to contain heme in some cases. For example, most PAS (Per-ARNT-Sim) domains do not bind heme, although several well-known PAS-domain containing proteins are heme-based sensors [36,37]. The same fold that binds heme in CooA was initially known for functioning as a cyclic-nucleotide-binding site in the cAMP-receptor proteins [38]. Likewise, the GAF fold, recently found to bind heme in the DevS protein, was first discovered for contributing an allosteric site for modulation by cyclic nucleotides in some phosphodiesterases (GAF = cGMP-regulated cyclic nucleotide phosphodiesterases, adenylate cyclases, and bacterial transcriptional regulator FhlA) [39,40]. Thus many domains that can bind heme alternatively occur entirely unoccupied, associated with another cofactor, or associated with a ligand. This is the case even for one of the two HNOB domains in mammalian sGC. An observation of avid ligation of O2 to the heme-binding HNOB domain (Tt-HNOX) from a Thermoanaerobacter tengcongensis methyl-accepting chemotaxis receptor called Tar4, together with indirect evidence associating some sGCs with in vivo responses to O2 , has led some to speculate that an entire O2 -binding class of HNOB-coupled sensors exist, and even that the sGCs initially evolved to sense O2 and later switched to sense NO [32,41–44]. While this may yet prove to be true, only two HNOB domains have so far been shown to yield recognizable oxy-heme spectra on addition of O2 to the deoxy-forms: T. tengcongensis HNOX and C. elegans GCY-35 [42,45] (Table 1). The great majority of these putative sGC-related O2 sensors have not been examined for signal transduction simply because they have not been purified in full-length form (Table 1).

O2 binding

Linked to O2 response

NO binding

Organism

Protein

Effect of O2

Notes

Reference

Thermoanaerobacter tengcongensis

Tt-HNOX, Tt-SONO

Absorption spectrum Kd = 90 nM

Strict Anaerobe.

[44,45]

Caenorhabditis elegans

GCY-35

Absorption spectrum

No distal Tyr at 140. No response to NO.

[46]

Caenorhabditis elegans

GCY-35

Aerotaxis

No response to NO (cell extract).

[42]

Caenorhabditis elegans

GCY-36

Aerotaxis

No distal Tyr at 140. No response to NO (cell extract).

[47]

Drosophila melanogaster

Gyc-88E/ Gyc-89Da

GC activity 8-65x inhibition (in vivo assay)

ODQ inhibited. Y140 in Gyc-88E not in other subunit.

[43,87,88]

Drosophila melanogaster

Gyc-88E/ Gyc-89Db

GC activity 3-39x inhibition (in vivo assay)

ODQ inhibited. Distal Tyr at 140 in Gyc-88 but not in other subunit.

[43,48,49]

Organism

Protein

NO coordination

Notes

Reference

Thermoanaerobacter tengcongensis

Tt-HNOX, Tt-SONO

6c at 25 C/5c at 70 C

Caenorhabditis elegans

GCY-35

5c/6c mix at 25 C

Legionella punctinforme

L1-HNOX

5c at 25 C

[44,45] [42] Facultative aerobe.



Legionella punctinforme

L2-HNOX

5c/6c mix at 25 C

Mammals

sGC

5c at 25 C

sGC relatives

Table 1. Oxygen and NO binding HNOB-coupled sensors

Facultative aerobe.

[50] [50] [1,51] 527

(Continued)

528

Table 1. (Continued) Organism

Protein

Effect of NO

Notes

Reference

Drosophila melanogaster

Gyc-88E/ Gyc-89Da

GC activity 10x inhibition (in vivo assay)

ODQ inhibited Distal Tyr at 140 in Gyc-88 but not in other subunit

[43,48,49]

Drosophila melanogaster

Gyc-88E/ Gyc-89Db

GC activity 10x inhibition (in vivo assay)

ODQ inhibited Distal Tyr at 140 in Gyc-88 but not in other subunit

[43,48,49]

Drosophila melanogaster

Gyc-99/ Gyc-100

GC activity >300 activation (in vivo assay)

Mammals

sGC

GC activity 200–400x activation

NO sensing

[1,51]

E.H.S. Sousa et al.

sGC relatives

529

3. MEASUREMENT OF sGC ACTIVITY AND REGULATION 3.1. In Vitro Assays Fully functioning and sensing sGCs are notoriously difficult to purify. Of all the sGCs that have been identified from sequence comparisons or genetics, the only ones so far assayed in vitro for cyclase activity are the mammalian sGCs. Gerber and colleagues have developed an assay that is a less expensive and more rapid alternative to the enzyme-linked immunosorbent assays (ELISA) commonly used, although it is also appropriate for automated screening [52]. The ELISA assays detect the cGMP product with antibodies against it; these assays are sensitive but expensive (∼$295/plate) and time consuming (6–7 hours) [52]. They are commercially designed such that the entire microplate is used per experiment even when very few assays are needed. In the less expensive protocol, the pyrophosphate produced from the cyclization of GTP to cGMP is converted to two free phosphate molecules, and these are quantified colorimetrically from their reaction with molybdate and malachite green [52]. The molybdate reaction is very selective, and the background of free phosphate can be completely accounted for and minimized. In vitro assays such as this are preferable to in vivo assays because they allow for a complete set of controls and conditions to be established, and for the state of the sGC heme to be verified at the beginning and end of the experiments. To our knowledge, there currently exist no in vitro studies of the cyclase activity of nonmammalian sGCs. This is because biochemical examinations of these proteins have so far been limited to the HNOB fragments, unaccompanied by the corresponding enzymatic regions [41,44,45,50]. Given the strong homology between nucleotide cyclases, it is quite possible that many of the proteins presumed to be guanylate cyclases are in fact adenylate cyclases.

3.2. In Vivo Assays Since all conclusions regarding cyclase activity and its regulation in non-mammalian sGCs are based on in vivo studies, an understanding of the design and limitations of in vivo cyclase assays is essential for a critical assessment of these results. An assay that has been used to assess sGC activity in vivo is one intended by Stasch and colleagues for an automated screen of drug candidates and employed as a photometric steady state and real-time measure of sGC activity [53]. In this assay, cGMP modulates the intracellular Ca2+ flux through a cGMP-gated channel, and the Ca2+ flux is in turn detected by the visible light produced from a Ca2+ -dependent, aequorin-catalyzed reaction of coelenterazine [54–56]. For this method to work in vivo, a CHO (Chinese hamster ovary) cell line is stably transfected with the cDNAs encoding the 1- and 1subunits of sGC, and this cell line is also made to express apo-aequorin and a gene encoding an olfactory cGMP-gated cation channel. The basal activity is measured by comparing the light output of sGC-transfected cells to that of control cells transfected with an empty vector. The effects of NO and other stimulatory compounds are measured from the increased aequorin chemiluminescence due to the rise in cGMP and Ca2+ levels. Although this assay was successfully used to demonstrate a sensitive sGC dose response to the NO donor DEA-NO and to identify the compound BAY58-2667 as an

530

E.H.S. Sousa et al.

activator of sGC from a library of 90 000 chemicals, some of the limitations of the assay for standard in vivo cyclase measurements are worth noting. The expression of the sGC subunits is quite variable from one transfection to the next, and the heme content of the sGC in expressing cells cannot be determined. Additional concerns are the following: the cation channel has a calmodulin/Ca2+ -dependent activity that leads to its repression; a variety of physiological conditions affect cGMP levels; NO and its many side products, the screened compounds, and hypoxic conditions may affect living cells in unpredictable ways that influence cGMP levels and cannot be directly controlled.

4. STRUCTURAL HIGHLIGHTS 4.1. Conformational Changes in Mammalian sGC Though the structure of mammalian sGC is not yet known, each one of its two subunits can be dissected into three recognizable regions of predictable structure: an N-terminal HNOB domain, a middle PAS-like domain involved in dimerization, and a C-terminal catalytic domain resembling adenylate cyclase [29,33,34,57]. Spectroscopic studies have been used to test the validity of sGC structural models and probe the conformational changes that accompany activation of this protein. Gerber and colleagues have focused on the changes triggered by NO, YC-1 (3-(5 -hydroxymethyl)-(2 -furyl)-1-benzylindazole), or ODQ (1H-[1,2,4]-oxadiazolo-[4,3-a]-quinoxalin-1-one) in the circular dichroism spectra, small angle X-ray scattering, and tryptophan fluorescence of sGC [58,59]. Curiously, NO, YC-1, or ODQ promoted significant changes in secondary structure only when used in combination, and those changes suggested that sGC is more compact in the on- than in the off-state. Work such as theirs has offered new insights into the global conformational changes that sGC undergoes during signal transduction. Selective tryptophan excitation along with the use of catalytic-site probes might provide a clearer view of such events.

4.2. Structure of the T. tengcongensis HNOB Domain X-ray crystallographic structures are available for a fragment comprising the HNOB domain of the Tar4 protein from the thermophilic strict anaerobe T. tengcongensis. The two groups who independently solved the structures of the FeII O2 form at 1.77 Å resolution and 2.5 Å resolution, respectively, have alternately called them Tt-HNOX [Tar41−188 , heme-NO-oxygen sensor) and Tt-SONO [Tar41−191 , sensor of NO] [33,34]. A structure is also available for the FeIII form of Tt-HNOX at 1.90 Å resolution [33]. Experience has shown that structures for evolutionarily related domains are often similar in their broad outlines, and so Tt-HNOX has been used to model the structures of other HNOB domains. It is reasonable to expect that, as in Tt-HNOX, the heme of HNOB-coupled sensors will be sandwiched between an -helix on the distal side and an  motif on the proximal side (Fig. 2). The contacts between the heme and protein in Tt-HNOX lead to far more robust binding than observed in the notoriously unstable mammalian sGC. Indeed, this difference is likely to be one reason why the Tt-HNOX structure is available but not the mammalian sGC structure. Though it is clear that specific details of Tt-HNOX radically differ, at least in their effects, from corresponding

sGC relatives (A)

531 (B)

W9

N74

Y140 R135

Y131

H102

S133

Fig. 2. Structure of the T. tengcongensis. Tar4 HNOB domain bound to O2 [33,34]. Part (A) shows the overall X-ray crystallographic structure of the heme-domain of oxy-TtHNOX (PDB 1U55). Part (B) shows the heme and its interactions, with key residues involved in hydrogen bonding indicated by dashed lines. (see Plate 16.)

features of sGC, there is some reason to believe that the Yx(S/T)xR motif probably serves as a region of heme contact in mammalian sGC and diverse HNOB domains. The Y135, S137, and R139 residues, comprising the Yx(S/T)xR motif of mammalian sGC, have been implicated in heme binding based on the in vivo responses of wild-type sGC and sGC mutants toward NO donors, ODQ inhibition, BAY 41-2272 (NO independent and heme-dependent stimulator) and BAY 58-2667 (heme-independent activator) [35,60]. In particular, BAY 58-2667, which is known to displace heme and occupy its binding site, could activate sGC with a proximal-histidine (H105F) substitution but with substitutions at Y135 and R139 [60]. In Tt-HNOX, the coordinating protein side chain is supplied by the H102 residue from the proximal helix (alternatively called F in Tt-HNOX or 5 in Tt-SONO), Y131, and S133 form hydrogen bonds to the heme propionate 6, and R135 forms hydrogen bonds to both heme propionates [33,34]. The structure of the iron porphyrin in O2 -bound Tt-HNOX is quite distorted, mostly due to the “saddling” and “ruffling” that results from bending of the pyrrole rings with respect to each other [33]. Consequently, heme distortions have been suggested as a possible cause of the Fe His bond weakening that leads to formation of pentacoordinate nitrosyl species on exposure of mammalian sGCs and some other HNOB-coupled sensors to NO. Interestingly, NO binding breaks the Fe His bond in mammalian sGC at room temperature, whereas this ruptures the Fe His bond in Tt-HNOX at 70 C but not at room temperature [34]. If the heme distortions constitute a major source of the bond strain, then the heme in mammalian sGC or that in Tt-HNOX warmed to 70 C would

532

E.H.S. Sousa et al. Low Activity

Z

CO

High Activity

X X

CO

Z

YC-1

Fe2+

Fe2+

X X

His

His

CO

Z

Z

NO Fe2+

X X

His

Fe 2+ NO

X X

His

Basal

BAY 58-2667

High Activity

Z X X

His High Activity

Fig. 3. Proposed Modes of sGC Activation by NO, CO, YC-1, and BAY58-2667. A model accounting for an sGC activation event upon binding to NO, CO, CO/YC-1, and BAY 58-2667 is proposed and finds support from recent data [60,64]. In every case, activation is induced by changing the conformation of protein “region Z”. Nitric oxide binding to the heme proximal side might shift the porphyrin ring toward the distal side and into position for electrostatic interactions of the heme edge with that region. The YC-1 heme-dependent stimulant is thought to shift the protein toward the heme to facilitate the interaction of “region Z” with the heme edge, as illustrated for YC-1 activation of the CO-bound species. The BAY58-2667 heme-independent stimulant displaces the porphyrin ring from the heme pocket and directly accesses “region Z” by electrostatic interactions reminiscent of those of the heme edge.

have to be even more distorted than the heme seen in the O2 -bound Tt-HNOX crystal structure. Perhaps the biggest question is whether HNOB domains control the activities of their accompanying transmitters by similar mechanisms. The first point to note is that the heme-binding domains in sGC and in Tar4 control different activities; specifically,

sGC relatives

533

Tt-HNOX is not coupled to a guanylyl cyclase, but rather to a methyl-accepting chemotaxis receptor (Fig. 1). Still, it is possible that a similar conformational change at the heme controls two quite different activities. Unfortunately, we do not know which states of the Tt-HNOX heme activate or inhibit the chemotaxis activity of Tar4. In the O2 sensor FixL, two arginine residues near the heme periphery form and break hydrogen bonds to the heme propionates as the heme distorts, and these bond rearrangements couple to the inhibition of this protein’s kinase activity [37,61–63]. A similar mechanism has been proposed for HNOB-coupled sensors, based on the observation in Tt-HNOX of contacts between the Yx(S/T)xR motif with the heme periphery and other polar residues such as D45 [33,35,64] (Fig. 3). Inspection of the heme propionate contacts in the available crystal structures of the FeII O2 and FeIII forms of Tt-HNOX show no significant differences, but this could be because both structures represent the activating (or the inhibiting) conformation.

5. OXYGEN BINDING TO Tt-HNOX To detect NO in a large background of O2 , the mammalian sGCs have evolved a heme pocket that discriminates against O2 binding so effectively that essentially no oxy-sGC is formed in air. This very remarkable discrimination against O2 binding that exists in sGC is absent from Tt-HNOX; in fact, some structural features in Tt-HNOX reinforce rather than prevent O2 binding. Hemoglobins and myoglobins employ a variety of polar residues, including histidine, glutamine, and tyrosine to stabilize bound O2 by hydrogen-bond donation [65,66]. Some higher affinity O2 carriers, such as the trematode hemoglobins, supplement the Fe O2 bond with multiple hydrogen bonds to the bound ligand [67]. At room temperature, every 1.36 kcal/mole of O2 -binding energy contributed by hydrogen bonding increases the O2 affinity by a factor of 10. A network of hydrogen bonds, in particular a hydrogen bond to bound O2 from a distal tyrosine (Y140), reinforce the Fe O2 bond in Tt-HNOX. This distal tyrosine has been singled out as the reason for an extraordinarily low equilibrium dissociation constant for binding of O2 to Tt-HNOX (Kd ∼ 90 nM) [44]. Mutation of this tyrosine did raise the Kd , though only to about the same value as that of myoglobin (about 1.4 M) [44]. Surprisingly, this modest 1.6 kcal contribution of a tyrosine to O2 binding has led to a great overstatement of the importance of this residue [41,44]. Some authors go as far as to predict that a sGC will or will not bind O2 solely on the basis of whether this distal tyrosine is present or absent, and claims of an entire O2 -binding class of sGC have been made from sequence homology [44]. In fact, at least one sGC with a distal tyrosine, the one from Clostridium botulinum, does not bind O2 at all, and there is as yet no evidence of O2 binding to most of the recently discovered HNOB-coupled sensors that are postulated to sense O2 [34]. Currently, the most that can reasonably be said is that for some sGCs that bind O2 , a distal tyrosine can enhance binding. In the complete absence of any hydrogen-bond stabilization of bound O2 , that is, if the free energy for O2 binding came exclusively from formation of the Fe O2 bond, O2 would bind to heme with a Kd of about 100 M. This is roughly the Kd value for O2 binding to heme dissolved in benzene or soap micelles, or to myoglobins with their stabilizing distal residues replaced by aliphatic side chains [65,66]. A heme with this affinity would be about 72% saturated with O2 in aerated buffer. Even this level

534

E.H.S. Sousa et al.

of O2 binding, however, is avid compared to that of mammalian sGC. In addition to binding O2 relatively weakly, ferrous heme iron is usually rapidly oxidized on exposure to O2 in heme pockets that cannot supply hydrogen-bonding distal residues [68]. By contrast, sGC is quite stable. Thus the mere absence of hydrogen-bond stabilization is insufficient to explain the inertness of sGC to O2 binding or oxidation in air. In fact, the failure to ligate O2 implies that the binding energy for O2 is much lower than that of an unsupplemented Fe O2 bond. To achieve such a low affinity, a protein has to not only refrain from adding to the binding energy of O2 , but actually subtract binding energy from the Fe O2 bond. Dunham and colleagues have suggested that binding energy might be subtracted by distorting the porphyrin [61]. Such strain might also explain the unusual rupture of the Fe His bond in some GCs upon binding of NO.

6. REGULATION BY NO, CO, AND OTHER REGULATORY COMPOUNDS The mechanism by which NO activates sGC is not well understood. When electron paramagnetic resonance spectroscopy revealed that in fact nitrosyl-sGC contains pentacoordinate heme iron, it was widely assumed that the Fe His bond rupture triggers the enzymatic activation [69,70]. Nitric oxide exerts a strong trans-effect on the heme iron that weakens the Fe His bond on the opposite side of the heme and can rupture this bond if it is weak or strained. By contrast, O2 has no significant trans-effect, and CO has the reverse effect of strengthening the Fe His bond [71]. Thus a Fe His bond-rupture mechanism quickly gained favor because it discriminated well against ligands other than NO that could potentially bind to the heme. A suggested pathway for sGC activation, inspired by the cytochrome c from Alkaligenes xylosoxidans (AXCP), involves two NO molecules [72,73]. This model is supported by an X-ray crystallographic structure of the NO-bound form of AXCP showing NO bound to the heme iron on proximal side of the heme, and by theoretical DFT calculations [73,74]. Like sGC, AXCP is unusual in not binding O2 at all and forming a pentacoordinate nitrosyl form when exposed to NO. A perennial source of controversy about sGC activation is whether there is single NO-binding, two-step NO-binding, a requirement for second NO-binding event for full activation, and partially active conformational states [75–79]. In the above two-step NO-binding model, formation of Fe NO on the distal side occurs first and leads to partial activation, and subsequent formation of Fe NO on the proximal side leads to full activation [76]. Two even more fundamental questions are this: What does NO do to the protein to initiate the regulatory conformation change, and can a similar event be brought about without rupturing the Fe His bond? It is possible that binding of NO to the heme proximal side leads to interactions of the heme propionates with distal residues that trigger the regulatory conformational change and result in full activation (Fig. 3). The Fe His bond rupture is not strictly required for sGC activation. For example, CO activates sGC, although only three to fivefold [80–82]. In the presence of YC-1, a species with hexacoordinate CO-bound heme iron is produced with activity comparable to the pentacoordinate NO-bound form [82,83]. A number of synthetic drugs, most of which do not affect Fe His coordination, also activate sGC [22]. The drugs that can activate the cyclase activity of sGC fall into two groups: the heme-dependent stimulants that cannot

sGC relatives

535

activate sGC if the heme is absent or the heme iron is oxidized (e.g. YC-1), and the hemeindependent compounds (e.g. BAY58-2667) [22]. Is there a common mode of activation that is consistent with all these different stimulants? All of the synthetic drugs are much too large to bind to the heme or coexist with it inside of the heme pocket, so these compounds must either displace the heme or bind at another site. Let us suppose that the activation involves movement of the heme peripheral groups relative to residues of the heme pocket (Fig. 3). Since motion is relative, displacing either the heme, by severing the Fe His bond, or some heme-pocket residues, by initiating a conformational change with a drug outside the heme pocket, might result in the same final relative positions (Fig. 3). By itself, YC-1 stimulates sGC only about tenfold, but in combination with CO it brings about full sGC activation [82,83]; YC-1 also greatly enhances the sensitivity of sGC to NO and appears to slow the dissociation of NO [84]. These synergistic effects of YC-1 and the ligand would require that YC-1, despite binding outside the heme pocket, nevertheless alter the heme-pocket conformation [22,84] (Fig. 3). We can account for the activation of sGC by heme-independent stimulants by regarding these compounds as heme analogs that mimic the NO-bound heme. In apo-sGC these compounds would insert into the heme pocket, whereas in holo-sGC or met-sGC they would displace the heme from the pocket. Support for this hypothesis comes from the observation that protoporphyrin IX, which is simply heme without its central iron atom, can stimulate sGC 20- to 47-fold [82,85,86]. BAY 58–2667, which in nanomolar levels activates ferric sGC about 200-fold, has been shown in micromolar concentrations to displace the heme [55,87]. A space-filling model of this compound is remarkably similar to a heme and features carboxylate residues at the edges that might interact with key residues of the heme pocket.

7. OXYGEN-SENSING sGC PROTEINS? 7.1. Atypical sGCs from D. melanogaster For the more recently discovered non-mammalian sGCs, the effects of ligands have been inferred from genetic studies and in vivo (or crude extract) measurements of cyclase activity. Morton and colleagues identified some sGCs in invertebrates that they initially called “atypical sGCs” [43,88,89]. The Drosophila genome encodes five sGClike proteins: Gyc-88E, Gyc-89Da, Gyc-89Db, Gyc-99B, and Gyc-100B [46,90]. Except for Gyc-99B/Gyc-100B heterodimer, all the other proteins (Gyc-88E, Gyc89Da, Gyc-89Db) were poorly responsive or entirely unresponsive to NO in transfected COS-7 cells [43,48,90,]. In cells expressing Gyc-88E, Gyc-89Da, Gyc-89Db, the levels of cGMP decreased in the presence of O2 , and this “hypoxic response” was abolished by the heme-dependent sGC inhibitor ODQ [43]. A similar response was observed for crude lysates [49]. The O2 inhibition continued to increase in up to 50% pure O2 , suggesting the implied O2 affinity to be quite low and comparable to that of the proven O2 sensor FixL [36,91]. Interesting and exciting as these results are, it must be remembered that O2 , especially at elevated concentrations, has many effects on live cells and crude extracts, for example raising the concentrations of reactive-oxygen species. In addition, ligation of O2 to the Gyc proteins remains to be proven and is so far being inferred from sequence motifs that are anticipated to interact with bound heme. Hopefully, these proteins will

536

E.H.S. Sousa et al.

soon be purified and probed in vitro for O2 -regulated cyclase activity. If the observed inhibitory effect is indeed due to O2 binding to these sGCs, then the coupling between their HNOB and enzymatic domains would be reversed from that of “typical” sGCs. That is, binding of ligand would stabilize the inactive rather than the active conformation of these enzymes. This difference alone would be far more remarkable than the mere ability to bind O2 . Some physiological data linking hypoxia to cGMP production have come from studies of Drosophila larvae, which were observed, at low oxygen tensions, to rapidly stop eating, back away from their food, and begin to wander about and escape [88,92]. Experiments where larvae neuronal cells containing Gyc-89Da or Gcy-89Db were made to overexpress a bovine cGMP phosphodiesterase 5 (PDE5), showed that the hypoxic response of the larvae was slowed such that they escaped rapidly at 1% O2 but slower at 5% O2 , and slower still at 10% O2 [49]. The expression of PDE5 appeared to disrupt a normally strong attraction of the larvae to glucose, and for this reason, the Drosophila “atypical” sGCs have been postulated to be O2 sensors that might modulate movement toward sweets [49].

7.2. O2 -Sensing by Nematodes Caenorhabditis elegans can tolerate reasonably large fluctuations in O2 tension even though it relies entirely on diffusion for its gaseous exchange and contains no specialized respiratory or circulatory systems [47,93]. The metabolic rate of C. elegans remains unchanged as the O2 tension drops from 21 to 3.6% O2 and is only halved when the O2 level drops to 1% O2 [93]. They enter a suspended animation as the O2 tension declines from 1 to 0.25%, and they only begin to die between 0.1 and 0.01% [94,95]. It is likely that, as suggested for these nematodes, O2 levels serve as migratory indicators of proximity to the soil surface, food, and water-logged soil interstices [42,93,96]. Gray and colleagues have shown that C. elegans show a distinct chemotactic behavioral preference for 5–12% O2 and avoidance of higher or lower O2 levels [42]. The best evidence thus far for an O2 -sensing sGC comes from GCY-35: one of seven sGCs predicted in C. elegans. The GCY-35 protein is unlikely to be an NO sensor since no NO synthase is encoded by the C. elegans genome and NO does not stimulate cGMP synthesis by C. elegans homogenates [46]. A heme-binding HNOB domain in GCY-35 demonstrably binds O2 , as determined from the absorption spectrum of a recombinant fragment containing this domain [42]. C. elegans mutants with a GCY-35 gene disruption can be restored to a normal aerotactic response by complementation with the wildtype GCY-35. The GCY-36 protein also appears to be important for this aerotactic response, and it is expressed together with GCY-35 in neuronal cells [47]. These genetic data, together with the demonstration of O2 binding to the GCY-35 HNOB domain make these sGCs quite promising candidates for heme-based O2 sensors. Biochemical characterization of the active enzyme remains the only outstanding challenge. Future research should soon help to resolve questions such as the following: Are GCY-35 and GCY-36 homodimeric or heterodimeric enzymes? What is their specific cyclicnucleotide product? What are the roles of the other sGCs in C. elegans? Does production of cyclic nucleotide respond to the O2 concentration, and if so, Is the O2 dose response consistent with the observed aerotactic response of C. elegans?

sGC relatives

537

8. FUTURE DIRECTIONS Clearly, a most urgent goal is to demonstrate in vitro ligand regulation of an activity for one of the novel HNOB-coupled sensors. If an O2 response can be shown, it will be worthwhile to compare the in vitro responses of some methyl-accepting chemotaxis receptors and nucleotide cyclases to the in vivo dose responses involving these sensors. It will be of particular interest to investigate if CO-bound ferrous proteins or liganded ferric states can also regulate, since the response to these heme states would provide additional clues to the regulatory mechanisms. Mammalian sGC activation is still a matter of great debate and a number of mechanisms, of varying validity, have emerged to describe how NO activates sGC. The postulated coupling of the heme distortions and protein-propionate contacts to signal transduction are very exciting and offer many new avenues for exploration. Many important questions remain to be answered. For example, do all these HNOB-coupled sensors regulate their transmitters by a similar mechanism? Might this be explained by a global model including heme distorsions? Which evolutionary adaptations have led some HNOB domains to respond to NO and others to O2 ? Both heme-binding HNOB domains so far shown to bind O2 also bind NO and CO. Might they respond to more than one of these three ligands? Do some bacteria naturally possess heme- or non-heme dependent stimulants? If so, this might provide clues about possible natural stimulants of mammalian sGC. What factors govern the ligand affinities of HNOB-coupled sensors? The distal tyrosine proposed to assist O2 -binding by HNOB domains does not satisfactorily explain the extraordinarily high O2 affinity of Tt-HNOX. It is likely that a broad comparison of HNOB heme pockets will be needed to address these questions, along with careful studies of the activities of full-length proteins, and the structures of on- and off-states.

ACKNOWLEDGMENTS We thank Jason Tuckerman for his comments on the manuscript, and the National Research Initiative of the USDA Cooperative State Research, Education and Extension Service, grant number 2002-35318-14039 and a Welch Foundation Grant No. I-1575, for financial support.

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9]

Gerzer, R., Bohme, E., Hofmann, F., and Schultz, G. (1981) FEBS Lett., 132, 71–74. Craven, P.A. and DeRubertis, F.R. (1978) J. Biol. Chem., 253, 8433–8443. Ignarro, L. (1990) J. Pharmacol Toxicol., 67, 1–7. Schwarz, U.R., Walter, U., and Eigenthaler, M. (2001) Biochem. Pharmacol., 62, 1153–1161. Lincoln, T.M. (1989) Pharmacol. Ther., 41, 479–502. Rapoport, R.M. and Murad, F. (1983) J. Cyclic Nucleotide Protein Phosphor Res., 9, 281–296. Mellion, B.T., Ignarro, L.J., Ohlstein, E.H. et al. (1981) Blood, 57, 946–955. Garthwaite, J., Charles, S.L., and Chess-Williams, R. (1988) Nature, 336, 385–388. O’Dell, T.J., Hawkins, R.D., Kandel, E.R., and Arancio, O. (1991) Proc. Natl. Acad. Sci. USA, 88, 11285–11289.

538

E.H.S. Sousa et al.

[10] [11] [12] [13] [14]

Zhuo, M. and Hawkins, R.D. (1995) Rev. Neurosci., 6, 259–277. Waldman, S.A. and Murad, F. (1987) Pharmacol. Rev., 39, 163–196. Wedel, B. and Garbers, D. (2001) Annu. Rev. Physiol., 63, 215–233. Russwurm, M., Wittau, N., and Koesling, D. (2001) J. Biol. Chem., 276, 44647–44652. Nedvetsky, P.I., Sessa, W.C., and Schmidt, H.H. (2002) Proc. Natl. Acad. Sci. USA, 99 (2002), 16510–16512. Meurer, S., Pioch, S., Wagner, K. et al. (2004), J. Biol Chem., 279, 49346–49354. Hanafy, K.A., Martin, E., and Murad, F. (2004) J. Biol. Chem., 279, 46946–46953. Balashova, N., Chang, F.J., Lamothe, M. et al. (2005) J. Biol. Chem., 280, 2186–2196. Venema, R.C., Venema, V.J., Ju, H. et al. (2003) Am. J. Physiol. Heart Circ. Physiol., 285, H669–H678. Russwurm, M. and Koesling, D. (2005) Methods Enzymol., 396, 492–501. Lee, Y.C., Martin, E., and Murad, F. (2000) Proc. Natl. Acad. Sci. USA, 97, 10763–10768. Murad, F.N. (2006) Engl. J. Med., 355, 2003–2011. Evgenov, O.V., Pacher, P., Schmidt, P.M. et al. (2006) Nat. Rev. Drug. Discov., 5, 755–768. Cary, S.P., Winger, J.A., Derbyshire, E.R., and Marletta, M.A. (2006) Trends Biochem. Sci., 31, 231–239. Poulos, T.L. (2006) Curr. Opin. Struct. Biol., 16, 736–743. Boon, E.M. and Marletta, M.A. (2005) J. Inorg. Biochem., 99, 892–902. Koesling, D., Russwurm, M., Mergia, E. et al. (2004) Neurochem. Int., 45, 813–819. Krumenacker, J.S., Hanafy, K.A., and Murad, F. (2004) Brain Res. Bull., 62, 505–515. Friebe, A. and Koesling, D. (2003) Circ. Res., 93, 96–105. Andreopoulos, S. and Papapetropoulos, A. (2000) Gen. Pharmacol., 34, 147–157. Pyriochou, A. and Papapetropoulos, A. (2005) Cell Signal, 17, 407–13. Iyer, L.M., Anantharaman, V., and Aravind, L. (2003) BMC Genomics, 4, 5. Fitzpatrick, D.A., O’Halloran, D.M., and Burnell, A.M. (2006) BMC Evol. Biol., 6, 26. Pellicena, P., Karow, D.S., Boon, E.M. et al. (2004) Proc. Natl. Acad. Sci. USA, 101, 12854–12859. Nioche, P., Berka, V., Vipond, J. et al. (2004) Science, 306, 1550–1553. Schmidt, P.M., Rothkegel, C., Wunder, F. et al. (2005) Eur. J. Pharmacol., 513, 67–74. Gilles-Gonzalez, M.A. and Gonzalez, G. (2004) J. Appl. Physiol., 96, 774–783. Gilles-Gonzalez, M.A. and Gonzalez, G. (2005) J. Inorg. Biochem., 99, 1–22. Lanzilotta, W.N., Schuller, D.J., Thorsteinsson, M.V. et al. (2000) Nat. Struct. Biol., 7, 876–880. Sardiwal, S., Kendall, S.L., Movahedzadeh, F. et al. (2005) J. Mol. Biol., 353, 929–936. Martinez, S.E., Beavo, J.A., and Hol, W.G. (2002) Mol. Interv., 2, 317–323. Boon, E.M. and Marletta, M.A. (2005) Cur. Opin Chem Biol., 9, 441–446. Gray, J.M., Karow, D.S., Lu, H. et al. (2004) Nature, 430, 317–322. Morton, D.B. (2004) J. Biol. Chem., 279, 50651–50653. Boon, E.M., Huang, S.H., and Marletta, M.A. (2005) Nat. Chem. Biol., 1, 53–59. Karow, D.S., Pan, D., Tran, R. et al. (2004) Biochemistry, 43, 10203–10211. Morton, D.B., Hudson, M.L., Waters, E., and O’Shea, M. (1999) Curr Biol., 9, R546–R547. Cheung, B.H., Cohen, M., Rogers, C. et al. (2005) Curr Biol., 15, 905–917. Morton, D.B., Langlais, K.K., Stewart, J.A., and Vermehren, A. (2005) J. Insect Sci., 5, 12. Vermehren, A., Langlais, K.K., and Morton, D.B. (2006) J. Insect Physiol., 52, 340–348. Boon, E.M., Davis, J.H., Tran, R. et al. (2006) J. Biol. Chem., 281, 21892–21902. Stone, J.R. and Marletta, M.A. (1994) Biochemistry, 33, 5636–5640. Sousa, E.H., Garay, P.A., Tinianow, J.N., and Gerber, N.C. (2006) Anal Biochem., 348, 57–63. Wunder, F., Stasch, J.P., Hutter, J. et al. (2005) Anal Biochem., 339, 104–112. Rich, T.C., Tse, T.E., Rohan, J.G. et al. (2001) J. Gen. Physiol., 118, 63–78.

[15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54]

sGC relatives [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75] [76] [77] [78] [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95] [96]

539

Shimomura, O. and Johnson, F.H. (1978) Proc. Natl. Acad. Sci. USA, 75, 2611–2615. Reinscheid, R.K., Kim, J., Zeng, J., and Civelli, O. (2003) Eur. J. Pharmacol., 478, 27–34. Hobbs, A.J. (1997) Trends Pharmacol. Sci., 18, 484–491. Kosarikov, D.N., Young, P., Uversky, V.N., and Gerber, N.C. (2001) Arch Biochem. Biophys., 388, 185–197. Kosarikov, D.N., Lee, J.M., Uversky, V.N., and Counts Gerber, N. (2001) J. Inorg. Biochem., 87, 267–76. Schmidt, P.M., Schramm, M., Schroder, H. et al. (2004) J. Biol. Chem., 279, 3025–3032. Dunham, C.M., Dioum, E.M., Tuckerman, J.R. et al. (2003) Biochemistry, 42, 7701–7708. Gong, W., Hao, B., Mansy, S.S. et al. (1998) Proc. Natl. Acad. Sci. USA, 95, 15177–15182. Gilles-Gonzalez, M.A., Caceres, A.I., Sousa, E.H. et al. (2006) J. Mol. Biol., 360, 80–89. Rothkegel, C., Schmidt, P.M., Stoll, F. et al. (2006) FEBS Lett., 580, 4205–4213. Olson, J.S. and Phillips, G.N. (1997) J. Biol. Inorg. Chem., 2, 544–552. Rohlfs, R.J., Mathews, A.J., Carver, T.E. et al. (1990) J. Biol. Chem., 265, 3168–3176. Kiger, L., Rashid, A.K., Griffon, N. et al. (1998) Biophys J., 75, 990–998. Brantley, R.E., Jr., Smerdon, S.J., Wilkinson, A.J. et al. (1993) J. Biol. Chem., 268, 6995–7010. Stone, J.R., Sands, R.H., Dunham, W.R., and Marletta, M.A. (1995) Biochem. Biophys. Res. Commun., 207, 572–577. Makino, R., Matsuda, H., Obayashi, E. et al. (1999) J. Biol. Chem., 274, 7714–7723. Coe, B.J. and Glenwright, S.J. (2000) Coordination Chem. Rev., 203, 5–80. Lawson, D.M., Stevenson, C.E., Andrew, C.R. et al.. (2003) Biochem. Soc. Trans., 31, 553–7. Lawson, D.M., Stevenson, C.E., Andrew, C.R., and Eady, R.R. (2000) EMBO J., 19, 5661–5671. Marti, M.A., Capece, L., Crespo, A. et al. (2005) J. Am. Chem. Soc., 127, 7721–7728. Ballou, D.P., Zhao, Y., Brandish, P.E., and Marletta, M.A. (2002) Proc. Natl. Acad. Sci. USA, 99, 12097–12101. Russwurm, M. and Koesling, D. (2004) EMBO J., 23, 4443–4450. Cary, S.P., Winger, J.A., and Marletta, M.A. (2005) Proc. Natl. Acad. Sci. USA, 102, 13064–13069. Roy, B. and Garthwaite, J. (2006) Proc. Natl. Acad. Sci. USA, 103, (2006), 12185–12190. Bellamy, T.C., Wood, J., and Garthwaite, J. (2002) Proc. Natl. Acad. Sci. USA, 99, 507–510. Brune, B. and Ullrich, V. (1987) Mol. Pharmacol., 32, 497–504. Brune, B., Schmidt, K.U., and Ullrich, V. (1990) Eur. J. Biochem., 192, 683–688. Friebe, A., Schultz, G., and Koesling, D. (1996) EMBO J., 15, 6863–6868. Stone, J.R. and Marletta, M.A. (1998) Chem. Biol., 5, 255–261. Russwurm, M., Mergia, E., Mullershausen, F., and Koesling, D. (2002) J. Biol. Chem., 277, 24883–24888. Wolin, M.S., Wood, K.S., and Ignarro, L.J. (1982) J. Biol. Chem., 257, 13312–13320. Ignarro, L.J., Wood, K.S., and Wolin, M.S. (1982) Proc. Natl. Acad. Sci. USA, 79, 2870–2873. Stasch, J.P., Schmidt, P., Alonso-Alija, C. et al. (2002) Br. J. Pharmacol., 136, 773–783. Langlais, K.K., Stewart, J.A., and Morton, D.B. J. Exp. Biol., 207, 2323–2338. Morton, D.B. (2004) Mol. Neurobiol., 29, 97–116. Shah, S. and Hyde, D.R. J. Biol. Chem., 270, 15368–15376. Gilles-Gonzalez, M.A., Gonzalez, G., Perutz, M.F. et al. (1994) Biochemistry, 33, 8067–8073. Wingrove, J.A. and O’Farrell, P.H. (1999) Cell, 98, 105–114. Van Voorhies, W.A. and Ward, S. (2000) J. Exp. Biol., 203, 2467–2478. Cheung, B.H., Cohen, M., Rogers, C. et al. (2005) Curr Biol., 15, 905–917. Nystul, T.G. and Roth, M.B. (2004) Proc. Natl. Acad. Sci. USA, 101, 9133–9136. Stone, J.R. and Marletta, M.A. (1994) Biochemistry, 33, 5636–5640.

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Published by Elsevier B.V.

Chapter 20

Resonance Raman Studies of the Activation Mechanism of Soluble Guanylate Cyclase Biswajit Pala and Teizo Kitagawab a b

Centre for Cellular and Molecular Biology, Uppal Road, Hyderabad – 500007, India. Okazaki Institute for Integrative Bioscience, National Institutes of Natural Sciences, Higashiyama 5-1, Okazaki 444-8787, Japan.

Abstract Soluble guanylate cyclase (sGC, EC 4.6.1.2) acts as a sensor for nitric oxide (NO), but is also activated by carbon monoxide in the presence of an allosteric modulator. Resonance Raman studies on the structure–function relations of sGC are reviewed with a focus on the CO-adduct in the presence and absence of allosteric modulator, YC-1/BAY, and substrate analogues. It is demonstrated that the sGC isolated from bovine lung contains one species with a five-coordinate (5c) ferrous high-spin heme with the Fe His stretching mode at 204 cm−1 , but its CO-adduct yields two species with different conformations about the heme pocket with the Fe CO stretching (Fe CO ) mode at 473 and 489 cm−1 , both of which are His- and CO-coordinated 6c ferrous adducts. Addition of YC-1/BAY to it changes their population and further addition of GTP yields one kind of 6c (Fe CO = 489 cm−1 ) in addition to 5c CO-adduct (Fe CO = 521cm−1 ). Under this condition the enzymatic activity becomes nearly the same level as that of NO adduct. Addition of -S-GTP yields the same effect as GTP does but cGMP and GDP give much less effects. Unexpectedly, ATP cancels the effects of GTP. The structural meaning of these spectroscopic observations is discussed in detail. Keywords: sGC; Resonance Raman; YC-1/BAY; NO; CO; Activation

1. INTRODUCTION Heme proteins constitute a large class of macromolecules and play a significant role in diverse and distinct biological functions. Among these, heme-based sensors for diatomic gases are of special interest. Several sensory proteins for diatomic molecules are known. CooA [1] and NPAS2 [2] have been identified as CO sensors, while HemAT [3], DOS [4], PDEA1 [5] and FixL [6] detect molecular oxygen. Soluble guanylate cyclase (sGC, EC 4.6.1.2) acts as a sensor for nitric oxide (NO) [7,8]. Recently, prokaryotic NO sensory proteins named SONO [9] and HNOX [10] and eukaryotic NO sensitive initiation factor 2  kinase (HRI) [11] have also been reported. In general, sensory proteins have distinct regulatory and catalytic domains and the information is conveyed

Activation Mechanism of sGC

541

from the regulatory domain to the catalytic domain upon sensing the signal. This sensing is further shrouded in mystery regarding the ligand discrimination by these proteins, although they share a similar type of heme. It’s important to unveil the mechanism how the sensed signal is communicated to the catalytic site not only for the understanding of the basic mechanism, but also for application. One of the extensively studied sensory heme proteins is sGC, which is involved in the cyclic GMP (cGMP)-mediated cell signaling pathways [7,8]. Among sensory proteins, sGC is the physiological receptor of nitric oxide [7,8] and is expressed in the cytoplasm of almost all mammalian cells. This enzyme catalyzes the conversion of GTP to cGMP in the presence of Mg2+ or Mn2+ . Nitric oxide, synthesized from l-arginine by NO synthase in physiological conditions or provided by an exogenous source like nitroglycerin, binds to the heme iron of sGC, and raises the conversion up to 400 times compared with its basal activity [12]. As illustrated in Fig. 1, produced cGMP acts as a second messenger in many cell signaling

Signal

Ca2+ CaM

L-Arg

NOS

NADPH

cGMP-gated Ion channels Citrulin + NO

NO

E

PD

NADP+ cGMP sGC-XO

GM

P

GTP Heme

HO

NADPH

Biliverdin + Fe +

NADP+ BVR

Bilirubin

CO

PKG

CO

NADPH NADP+

Ca2+ Cellular response Target Cell

Fig. 1. Activation of sGC by NO and CO: A general model. NO and CO are synthesized by their respective synthetic enzymes, nitric oxide synthase (NOS) and heme oxygenase (HO). The NO biosynthesis is regulated by Ca2+ via calmodulin (CaM). Upon receiving external signal, cellular concentration of Ca2+ is upregulated and NOS forms an activated complex with CaM. On the other hand, HO degrades heme (iron protoporphyrin IX) bringing about free iron, biliverdin and gaseous CO. Both NO and CO diffuse into the neighboring target cell, for example, smooth muscle cell in cardiovascular system, and bind to sGC, which converts GTP into cGMP. cGMP acts as a second messenger of cells and activates a cascade of events. The produced cGMP activates a cGMP dependent kinase, PKG, and triggers numerous cellular responses. However, the CO dependent activation is very weak compared with that of NO dependent one. Reproduced from [74].

542

B. Pal and T. Kitagawa

pathways, including cardiovascular systems, nervous systems, platelets, macrophages and apoptosis [7,8,13,14]. Carbon monoxide (CO), also identified as a neurotransmitter [15], also binds to sGC, resulting in a marginal elevation of activity [12]. The sGC is generally a heterodimer of two similar but non-identical subunits and simultaneous expression of both subunits are required for catalytic activity [16]. A new isoform of sGC has also been identified from the nervous system of Manduca sexta (MsGC-3) that does not require the formation of heterodimer for catalytic activity [17]. An atypical sGC, orthologous to MsGC-3, have recently been reported for Drosophila melanogaster that can sense oxygen [18]. Several proteins with sequence homology to sGC have been identified on the basis of genomic searching [19] and some of them have recently been cloned and characterized [9,10,20]. SONO from Clostridium botulinum, H-NOXs from Vibrio cholerae (VCA0720, 181 residues) and heme domain of Tar4 protein from the Thermoanaerobacter tengcongenesis (TtTar4H, 188 residues) are well characterized. These proteins share some sequence homology, spectroscopic properties and key conserved residues with sGC, and are placed in the same family. VCA0720 has essentially identical ligand binding properties with sGC, whereas TtTar4H can bind O2 and has very different ligand binding properties. Both the proteins form stable six-coordinated (6c) low-spin (LS) CO-adducts. VCA0720 forms five-coordinated (5c) high-spin (HS) NO complex whereas TtTar4H forms a 6c LS NO complex with His as the proximal ligand. Structure of TtTar4H has also been solved independently by two different groups [9,20]. Structure of TtTar4H suggests a new fold containing seven -helices and one four-stranded antiparallel -sheet [9,20]. The heme group is tightly packed within a central cavity where proximal side is surrounded by the -sheet and two helices, whereas the distal pocket is lined by residues from another helix. Interestingly, the heme group in TtTar4H was found to be significantly distorted. The structure of TtTar4H also indicates the reason for ligand discrimination and it has been argued that a Tyr (at position 140) in the distal pocket may contribute to the high affinity of the protein for oxygen. In case of sGC, the residue is more hydrophobic in nature [20]. A number of excellent reviews are available on many different aspects of sGC [16,21–27]. In this chapter we focus on the resonance Raman (RR) studies to unveil the structure–function relationship in sGC and its mechanism of activation with an emphasis on CO-adduct of sGC.

2. STRUCTURAL CHARACTERISTICS OF sGC The sGC is a heterodimer with molecular masses between 73 and 82 kDa for -subunits and about 70 kDa for -subunits [28,29]. The crystal structure of sGC is not available yet. In the absence of crystal structure, spectroscopic methods have extensively been applied along with biochemical and modeling studies to correlate the structure of this enzyme with its function. The spectroscopic studies include UV–visible, electron paramagnetic resonance (EPR) [30–32], CD [33], fluorescence [33], RR [34–39], and FT-IR [40] spectroscopy. This enzyme contains an iron–protoporphyrin IX in an N-terminal region

Activation Mechanism of sGC

543

of the -subunit with a stoichiometry of one heme per heterodimer [28,41], while the catalytic activity lies in the C-terminal region that has sequence similarity to the catalytic domain of particulate guanylate and adenylate cyclases [42–43]. The native sGC as isolated from bovine lung stays in the reduced state under aerobic conditions and contains a 5cHS heme, ligated by a conserved His residue of -subunit [34,37]. However, two other groups independently found sGC as a mixture of 5c HS and 6c LS species [38,39]. The latter enzyme is categorized as sGC1 to distinguish from the former, which is categorized as sGC2 . Both the sGC1 and sGC2 enzymes are activated by NO [34,44,45]. Each subunit of sGC can be divided into three functional domains: heme binding, dimerization and catalytic domains. The heme binding domain (HBD) of sGC has been mapped [46–48]. It was shown that the 1–385 is the necessary and sufficient region to be treated as the heme-binding region of sGC [46], although more recent work pointed out that 1–194 residues are minimum requirement for 1 isoform [47]. For 2 isoform of rat sGC, 1–217 residues are required for a stable heme-binding region [49]. It has been shown in a detailed mapping study [47] that 80–120 residues are essential for heme binding and 341–385 residues play an auxiliary role in heme binding. It has also been shown that 195–230 residues act as an inhibitory domain of heme binding [47]. Another interesting observation was that the two conserved cysteines on 1 subunit at positions 78 and 214 are also necessary for the binding of heme [48]. Based on the crystal structure of sGC-related protein, it has been proposed that conserved Tyr131 of YXSXR motif interacts with the propionate of heme [9,20,50]. Actually it was predicted on the basis of mutational studies that Tyr and Arg of YXSXR motif interacts with heme propionate in sGC [51]. The role of the HBD in the -subunit is not clear. The dimerization domain has very important role in the catalytic activity of sGC and co-expression of both the subunits is indispensable for production of an active enzyme. The dimerization region in sGC has been mapped and it has been shown that 204–244 and 397–408 are in contact with the -subunit [52]. On the other hand, 61–128 in the very N-terminus and a part of the central domain (367–462) in the -subunit is indispensable for dimerization [53]. The catalytic site of sGC is situated in the C-terminal region and both the subunits are involved in the binding of GTP and catalysis [42–43]. Residues 467–690 from the -subunit and residues 414–619 from the -subunit of rat sGC were co-expressed and shown that they are functionally active. Based on mutational and modeling studies, it is proposed that GTP binds to a single binding site per the – dimer as illustrated in Fig. 2 [42], where the -subunit binds the metal ion and triphosphate moieties whereas the -subunit binds the guanosine moiety. Residues interacting with the ribose, triphosphate and metal ions are conserved between adenylate and guanylate cyclases, but interactions with purine ring differ between them. To identify the amino acid residues critically responsible for GTP binding and catalysis, several residues were mutated and the substrate specificity was changed to ATP when three residues, R592, E473 and C541 of sGC were changed to their equivalent residues of an adenylate cyclase [43]. It has been proposed that E473 and C541 interact with N-2 and O-6 atoms of a guanine ring, respectively, and E473 may be the most important residue for the recognition of GTP while R592 may act as a second shell ligand for a guanine ring.

544

B. Pal and T. Kitagawa

Fig. 2. Model structure of catalytic domain of sGC. Model is prepared using SETOR from PDB ID 1AWN. Light grey represents the -subunit whereas dark gray represents the -subunit. Structure of GTP is shown as stick whereas Mg2+ is represented as sphere. (see Plate 17.)

3. ACTIVATORS OF sGC Being the physiological sensor of NO, sGC is activated by NO and marginally by CO. Quest for the synergistic activator along with CO resulted in the syntheses of 3-(5 -Hydroxymethyl-2 -furyl)-1-benzyl indazole (YC-1) and related compounds [54,55]. Also, a series of NO independent activators of sGC besides YC-1 have been synthesized and characterized [56–60]. Their structures are illustrated in Fig. 3. The BAY [BAY 41-2272: 3-(4-Amino-5-cyclopropylpyrimidin-2-yl)-1-(2-fluorobenzyl)-1Hpyrazolo[3,4-b]pyridine] group of compounds synthesized by BAYER AG is quite promising [56–57]. These compounds are classified into two groups: heme-dependent (BAY 41-2272 and BAY 41-8543) and -independent ones (BAY 58-2667). They are more potent and do not appear to inhibit the phosphodiesterase activity, at least at therapeutic doses [58]. These compounds have been successfully used in experimental animals and are of immense clinical promise. In addition, A-350619, synthesized at Abbot Laboratories, also activates sGC and induces penile erection in a conscious rat model [59]. Two compounds, S-3448 and HMR-1766, synthesized in Aventis Pharma, have been shown to activate oxidized sGC [60]. The identification of these novel non-NO-based sGC activators is important both as pharmacological tools and in the development of new therapeutics. These compounds have revealed the presence of previously unknown regulatory sites in the enzyme, which may be physiologically important and may serve as target sites for endogenous molecules modulating sGC activity [58]. Recently, we have shown that YC-1 effect becomes more prominent in the presence of imidazole [61]. Some inhibitors of sGC have also been identified. Among these, 1H-[1,2,4]oxadiazolo[4,3-a] quinoxalin-1-one (ODQ) is non-specific compound that inhibits sGC by oxidizing heme iron [62]. Isatin, an endogenous indole, also inhibited NO-stimulated sGC from platelets [63].

Activation Mechanism of sGC

545

O N

HO O

N

H2N

N

N

N

N N

N

O

O

S

F

YC-1 Cl

BAY 41-2272

H N

NH

O

Na

S

N S

H2N

O

S N

N

N Cl

A-350619

Cl

N

N

O

N

N

HMR-1766

F

H2N

O

O

HN

BAY 41-8543

N

Imidazole

O

HO

S

N

O O N

O

O

S

N O

N

N

ODQ

OH

O

NH

O

O

Cl

H N S

O

O

MeO N H

Isatin

O

OMe

BAY 58-2667

S-3448

Fig. 3. Structures of allosteric modulators. YC-1 was initially shown to activate platelet. BAY group of compounds are synthesized by Bayer AG, Germany. A-350619 is synthesized and tested in Abbot Laboratory, USA. A new series of Anthranilic acid derivatives, S-3448 and HMR-1766, are synthesized by Aventis Pharma, Germany, and are shown to activate sGC. ODQ is potent inhibitor of sGC. Isatin, an endogenous compound, inhibited NO-stimulated sGC activity in human platelet.

546

B. Pal and T. Kitagawa

4. SPECTROSCOPIC CHARACTERIZATION OF sGC 4.1. UV–VIS Characterization of sGC Native sGC purified from bovine lung shows an absorption maximum at 431 nm and a single broad peak in the / region at 555 nm, indicative of the presence of a 5c Fe(II)heme. However, the Soret bands of reconstituted sGC, reported by different groups, are somewhat different. Initially, the Soret peak of the heme-reconstituted ferrous enzyme reported by Ignarro et al. [64] was located around 425 nm and it was further verified by others [39,65]. However, other groups reported it at around 431 nm [66]. A recent report shows that a recombinant sGC mutant (1 H105C), in which the heme-coordinating His is replaced by Cys, has a Soret peak at 426 nm in the reduced state. When NO binds to the heme of native sGC, the Soret maximum shifts from 431 to 399 nm with the appearance of a shoulder at 485 nm and peaks at 537 and 572 nm, indicating the formation of a 5c nitrosyl-heme complex. On the other hand, Soret peak appears at 424 nm and the single broad / peak is replaced by peaks at 567 and 541 nm for CO bound sGC, indicating the formation of a 6c species (as shown in Fig. 4). The Soret band of sGC-CO at 424 nm shifts to 423 nm upon addition of YC-1. When YC-1 and GTP are further added to it, 4 nm blue shift is observed as reported [67]. BAY 41-2272 (will be abbreviated as BAY for rest of the text) also shifts Soret band of sGC-CO from 424 to 419 nm. Imidazole binds to sGC at high concentration and Soret maxima shifts at 414 nm [61].

1.6

424

1.4

431

Absorbance

1.2 1.0

567

541 399

0.8 0.6 0.4

X5 0.2 0.0 350

400

450

500

550

600

650

Wavelength (nm)

Fig. 4. Electronic absorption spectra of sGC. Solid line: Native; Dotted line: sGC CO; and Broken line: sGC NO.

Activation Mechanism of sGC

547

4.2. Resonance Raman Studies of sGC Resonance Raman spectroscopy enables us selective observation of vibrational frequencies of a chromophore by tuning the excitation wavelength into one of its absorption bands. Application of this technique to heme proteins has provided unique and important information of a heme group on dynamical as well as static structures [68]. For example, the presence or absence of the Fe His stretching band (Fe His ) directly indicates the presence of 5c or 6c-heme and a state of the Fe His bond, if present [69]. Furthermore, frequencies of some macrocycle modes including 2 , 3 , 4 and 10 vibrations can directly be correlated with the spin state, core-size and coordination number of the heme [70]. Similarly, the change in the Fe CO stretching (Fe CO ) frequency would reflect the polarity of heme pocket [71]. In the absence of crystal structure of sGC, resonance Raman spectroscopy is the method of choice to explain the underlying complexity of the structure, function and their correlation for this extremely important enzyme.

4.3. Structure of Native sGC in the Reduced State Several groups have independently studied sGC using RR spectroscopy and a rich literature is available [34–39]. Native sGC purified from bovine lung has a stable ferrous heme in its resting state, yielding a Soret peak at 431 nm. The heme structure marker bands including 2 , 3 , 4 and 10 were observed at 1563, 1472, 1356 and 1607 cm−1 , respectively, for native sGC as shown in Fig. 5A. The band at 204 cm−1 (Fig. 5B), observed for native sGC, is assigned to the Fe His mode and 7 appears at 676 cm−1 . These results indicate the presence of 5cHS heme where conserved histidine 105 is the proximal ligand. This structure is also supported by Deinum et al. [37]. The two recently found

1627 1607

1563

1524

1472 1426 1442

1390

1283 1308

1222

1185

1116 1140

Raman intensity

1356

(A)

Raman shift /cm–1

Fig. 5A. RR spectra of 1000–1700 cm−1 region of native sGC purified from bovine lung. Excitation wavelength used was 413 nm. Laser power was 5 mW. Reproduced from [74].

548

B. Pal and T. Kitagawa

788

321

995

420

372

815

258

758

204

Raman intensity

345

676

(B)

Raman shift /cm–1

Fig. 5B. RR spectra of 160–1000 cm−1 region of native sGC purified from bovine lung. Excitation wavelength used was 422 nm. Laser power was 2 mW. Reproduced from [74].

proteins with sequence homology to sGC, VCA0720 and TtTar4H give the Fe His band at 224 and 218 cm−1 , respectively [9], frequencies of which are distinctly higher than that of bovine sGC but are closer to that of deoxyMb (220 cm−1 ). On the other hand, two independent studies reported contradictory results [38,39]. In one case, sGC was purified from bovine lung as a heme depleted apo-protein and then reconstituted with exogenous heme [38]. In the other case, after sGC was expressed in baculovirus, the recombinant enzyme was purified as apo-protein and heme was incorporated during reconstitution [39]. In both cases RR spectra indicated the mixture of 5cHS and 6cLS ferrous heme. The Fe His mode was not observed while two 3 bands were observed around 1469 and 1491 cm−1 for both preparations. Interestingly, 4 was observed at 1359 cm−1 for the reconstituted sGC [39] compared with at 1356 cm−1 for enzyme purified as the heme-bound form [34–37]. The presence of the 6cLS component in the reconstituted enzyme (sGC1 ) was attributed to coordination of an unidentified histidine residue from the distal side [44]. It was also shown that the 6cLS sGC was converted to 5cHS by laser irradiation at a higher power [44]. On the contrary, native sGC purified as a holo protein (sGC2 ) was purely 5cHS and did not exhibit such photosensitivity [34,36]. The difference in the spectra and thus in the structure, was attributed to different methods of purification. In a recent work, we showed that 1.2 M imidazole treatment to the sGC2 oxidized the heme and reduction of the ferric enzyme with sodium dithionite produced enzyme similar to sGC1 [61]. This observation may help to explain the contradictory features. Presumably, the native enzyme has the 5cHS heme, but partial denaturation during the purification causes the release of heme and its reconstitution with exogenous heme brings about sGC1 .

Activation Mechanism of sGC

549

4.4. Structure of sGC–NO The RR spectra of NO-bound sGC (sGC NO), which are presented in Fig. 6, have been reported by several groups and their spectra are basically identical [34,37–39] despite the fact that the native sGC was reported to contain two different conformations.

1681

1634

1583 1609

1509

1375

1646

(A)

(a)

(b)

1648

Raman intensity

(c) ×5

(d)

(f) ×5

1650 1668

1684 1700

(e)

Raman shift /cm–1

Fig. 6A. RR spectra in the 1250–1800 cm−1 region of NO sGC in the absence (a–c) and presence (d–f) of GTP and MgCl2 . (a) 14 NO sGC; (b) 15 NO sGC; (c) difference, spectrum (a)–spectrum (b); (d) 14 NO sGC; (e) 15 NO sGC; and (f) difference, spectrum (d)–spectrum (e). The ordinate scales of spectra (c) and (f) are expanded 5 times. Solvent, 25 mM TEA buffer, pH 7.8 [containing 10 mM GTP and 4 mM MgCl2 for (d) and (e)]. Reproduced from [34].

550

B. Pal and T. Kitagawa

795

719

521

(a)

512

423

365

324

343

755

678

(B)

693

(c) ×2

521

Raman intensity

(b)

512

(d)

523

(e)

(f)

Raman shift /cm–1

Fig. 6B. RR spectra in the 250–850 cm−1 region of NO sGC in the absence (a–c) and presence (d–f) of GTP and MgCl2 . (a) 14 NO sGC; (b) 15 NO sGC; (c) difference, spectrum (a)– spectrum (b); (d) 14 NO sGC; (e) 15 NO sGC; and (f) difference, spectrum (d)–spectrum (e). The ordinate scale of spectrum (c) is expanded 2 times, but that of spectrum (f) is not expanded. Reproduced from [34].

The marker bands of heme, 2 , 3 , 4 and 10 , were observed at 1583, 1509, 1375 and 1646 cm−1 , respectively Fig. 6A (a). The frequencies of heme skeletal vibrations of ferric sGC NO were strikingly similar to those of ferrous sGC NO [34]. The Fe NO stretching (Fe NO ) band appeared at 521 cm−1 in Fig. 6B (a), consistent with other

Activation Mechanism of sGC

551

reports [37], although the band was not identified by Vogel et al. [37] due to strong background scattering. The N O stretching (N O ) band appeared at 1681 cm−1 in Fig. 6A (a) (1676 cm−1 in [37]) that is shifted to 1646 cm−1 with 15 NO (Fig. 6A (b)). However, an interesting observation was the appearance of two N O bands at 1700 and 1681 cm−1 in the presence of GTP as demonstrated by Fig. 6A (f) [34]. A similar splitting of the N O band was also observed in the presence of cGMP [34]. It is noted that the O2 -bindable sGC-like protein (TtTar4H) gives Fe NO at 553 cm−1 and N O at 1655 cm−1 , whereas the O2 -unboundable sGC-like protein (VCA0720) yields them at different frequencies (Fe NO = 523 cm−1 , N O = 1674 cm−1 ) [9], which are close to those of bovine sGC.

4.5. Structure of sGC–CO

1579

1615 1630

(B')

1307

(C')

1230

(a)

1138

751

1499

1582

(A')

1370

1600

1558

1470 1496

1422

676 473 487 521 565

399 423

Raman intensiy

1627

The RR spectra of CO-bound sGC (sGC CO) are shown in Fig. 7. In general, the heme skeletal vibrations, 2 , 3 , 4 and 10 were observed around 1579, 1496, 1370 and 1627 cm−1 , respectively. The Fe CO stretching (Fe CO ) Raman band of sGC CO appears at 473 cm−1 along with a second component at 487 cm−1 [34–36], and its frequency is in fact lowest among the known Fe CO bands of heme proteins. The low frequency can be attributed to the negative polarity of the heme pocket in sGC [37]. Xenobiotics like YC-1 and BAY have been studied using RR as these compounds synergistically activate sGC in conjugation with CO [72–73]. In the presence of YC-1, intensity of the 487 cm−1 component increases along with the appearance of a new weak band at 521 cm−1 [35,36], although the Fe CO in the presence of YC-1 was

(b)

(c) Raman shift /cm–1

Fig. 7. RR spectra in the 380–1400 cm−1 region of CO-bound sGC in the absence and presence of 200 M YC-1 and 1 mM GTP. (a) sGC CO; (b) sGC CO+YC-1; (c) sGC CO+YC-1+GTP. Inset shows the 1400–1700 cm−1 regions of spectra (a), (b) and (c). Excitation wavelength used was 422 nm. Laser power was <200 W. Reproduced from [74].

552

B. Pal and T. Kitagawa

previously reported at 492 cm−1 [72]. This YC-1 effect becomes much more prominent in the presence of GTP as delineated by Fig. 7(c) [36]. BAY also has same effect as that of YC-1 and two Fe CO corresponding to 6c species appears at 473 and 487 cm−1 whereas 5c species appear at 521 cm−1 [73]. The 521 and 487 cm−1 bands arise from the 5c heme [35,36] and His-coordinated 6c heme, respectively. However, the assignment of 487/492 cm−1 is self-evident. The weak intensity of the 521 cm−1 band, that is, the small amount of 5c sGC CO in the presence of YC-1 may not be able to explain the high level of activation similar to that of sGC NO. However, we cannot estimate quantitatively how much percent of heme adopt the 5c state from RR spectra. Intensity of the band at 521 cm−1 shows an excitation wavelength dependence and in fact, the intensity ratio, I521 /I487 is <0.2 in the case of 422 nm excitation whereas it goes up to 1.6 at 406.7 nm excitation; 0.15 for 428.7 nm, and 0.8 for 413 nm excitation [74]. Thus, a possibility cannot be ruled out that only the 5c species is responsible for the raised activity in the presence of YC-1 and GTP. Sharma et al. proposed that the proximal histidine might be replaced by YC-1 [75]; one of the two nitrogen atoms of YC-1 may donate a non-bonding pair of electrons to coordinate to the heme iron. This may result in only a small spectral shift of the Soret absorption band, because one nitrogenous base is simply replaced by another with no change in the coordination number. This hypothesis may explain the appearance of the Fe CO at 487 cm−1 in the presence of YC-1/BAY and can explain the level of activation. GTP may further facilitate this process. Other possible candidates for proximal ligations are cysteine or methionine, as proposed by Makino et al. [36]. It is worth to note that both VCA0720 and TtTar4H proteins give the Fe CO band at similar frequencies (490–491 cm−1 ), despite the fact that the latter can bind O2 but the former cannot [10]. The C O stretching (C O ) RR band appears as a weak band around 1985 cm−1 as shown in Fig. 8(a). The frequency is very close to that observed with FTIR (1987 cm−1 ) [35,40]. In the presence of YC-1 several components have been identified as a function of temperature [35]. We identified the corresponding band at 1969 cm−1 in RR measurements (Fig. 8(g), which was shifted to 1882 cm−1 with 13 C18 O Fig. 8(h). In the case of BAY a broad band was observed at 1970 cm−1 using FT-IR, which was resolved in two different components [73]. Intensities of these two components were dependent on temperature. In the presence of imidazole C O was observed at 1960 cm−1 in sGC CO [63]. The C O band becomes relatively sharp and intense, indicating more homogenous geometry of Fe C O. On the other hand, the Fe C O bending mode ( Fe C O ) appears at 565 cm−1 in sGC CO and becomes too weak to be detected in the presence of YC-1 and GTP as shown by Fig. 7(c). Although the bending mode in the presence of YC-1 has been reported at 589 cm−1 [35], we failed to detect a CO isotope sensitive band in that region even with higher resolution (0.5 cm−1 ) using the second order reflection of grating of the monochromator as illustrated by Fig. 9. The intensity increase in C O may have a relation with the weakening of the Fe C O mode in the presence of YC-1 and GTP. The -backbonding correlation of the Fe CO vs. C O frequencies, plotted in Fig. 10, suggests that both 473 and 487 cm−1 bands appear from a 6c sGC CO complex with an imidazole-type ligand. We also observed from the static RR measurements with a spinning cell that the bound CO is apparently photoinert in the presence of YC-1 and GTP as demonstrated in Fig. 11(A, B) and a band appears at 223 cm−1 [36,74,76]. However, complete photodissociation of CO in the presence of YC-1 and GTP using comparatively high

Activation Mechanism of sGC

553

(a)

1985

1969

1893

1882

(b) (c)

Raman intensity

(d) (e)

(f) (g)

(h)

(i)

Raman shift /cm

–1

Fig. 8. RR spectra of CO-bound sGC in the absence and presence of 200 M YC-1 and 1 mM GTP. (a) and (b): sGC CO; (d) and (e): sGC CO+YC-1; (g) and (h): sGC CO+YC-1+GTP. (a), (d) and (g): sGC 12 C16 O; (b), (e) and (h): sGC 13 C18 O. (c): difference (a–b), (f): difference, (d–e) and (i): difference (g–h). Excitation wavelength used was 422 nm. Laser power was <200 W. Reproduced from [74].

590

548 565

463 472 489 507 521

399

345 363

313

(b) 213 227

Raman intensity

(a)

(c)

Raman shift /cm–1

Fig. 9. High resolution RR spectra of CO-bound sGC in the presence of 200 M YC-1 and 1 mM GTP. (a): sGC 12 C16 O; (b): sGC 13 C18 O and (c): difference (a–b). Excitation wavelength used was 422 nm. Laser power was 2 mW. Reproduced from [74].

554

B. Pal and T. Kitagawa 540 530

νFe–CO /cm–1

520 510 500 490 480 470 1900

1940

1920

νC–O /cm–1

1960

1980

Fig. 10. The Fe−CO vs. C−O plot. Two closed circles denote sGC1 CO and sGC2 CO and an open circle represents sGC CO in the presence of 200 mM imidazole. A closed square denotes the five-coordinate sGC CO. Reproduced from [74].

(d)

988

951

752

788

712

577

487 521

321 345 369 399 420

204 223

Raman intensity

312

676

(A)

(c)

(b)

(a)

Raman shift /cm–1

Fig. 11A. Laser power dependent RR spectra in the 160–1000 cm−1 region of sGC and sGC CO in the absence and presence of 200 M YC-1 and 1 mM GTP. (a): native sGC; (b): sGC CO (photodissociated); (c): sGC CO+YC-1 (partially photodissociated); (d): sGC CO+YC-1+GTP (apparently not photodissociated). Laser power and cell spinning are the same for the four measurements. Excitation wavelength used was 422 nm. Laser power was 2.5 mW. Reproduced from [74].

Activation Mechanism of sGC

555

1562 1584 1607 1627

1501 1523

1472

1615 1632

1371 1393

1422

1356 1308

1184 1192 1222 1230

Raman intensity

1117 1139

(B)

(d)

(c)

(b)

(a)

Raman shift /cm–1

Fig. 11B. Laser power dependent RR spectra in the 1000–1700 cm−1 region of sGC and sGC CO in the absence and presence of 200 M YC-1 and 1 mM GTP. (a): native sGC; (b): sGC CO (photodissociated); (c): sGC CO+YC-1 (partially photodissociated); (d): sGC CO+YC-1+GTP (apparently not photodissociated). Experimental conditions are the same as those of Fig. 13. Reproduced from [74].

power and non-spinning cell resulted in disappearance of 223 cm−1 and instead the Fe His band appeared at 204 cm−1 [74]. One may argue that the 223 cm−1 band observed in sGC CO in the presence of YC-1 and GTP may originate from the Fe His mode of photodissociated transient species, was once considered as a plausible explanation. If such a replacement of the proximal ligand takes place upon addition of YC-1 and GTP, the CO-photodissociated transient species should exhibit the Fe-ligand stretching band at a unique frequency. However, the transient species gave rise to the Fe-ligand stretching mode at 204 cm−1 , indicating that 105 His is coordinated to the heme in the presence of YC-1 and GTP. Thus, the observation of the apparent photoinertness in static RR studies with a spinning cell is probably due to fast recombination of photodissociated CO, supporting the earlier hypothesis by Sharma et al. [75]. We expect some difference in photodissociation kinetics between the 5c and 6c sGC CO generated in the presence of YC-1 and GTP. It had been deduced that the photodissociation rate itself is different between the 5c and 6c sGC CO; in the absence of a trans base CO photodissociates much faster compared with that in its presence [77,78]. To resolve this issue, we performed timedependent density functional calculations for isolated heme in gaseous state and tried to explain the apparent photoinert nature of sGC CO [79]. We have shown that the geometry of 5c Fe CO in sGC may be responsible for this phenomena. We also showed that the 223 cm−1 may appear from the heme deformation. However, to get more detailed picture for proteins a molecular dynamics simulation may be required.

556

B. Pal and T. Kitagawa

5. EFFECTS OF SUBSTRATE AND ANALOGUES

565

522

473

489

424

400

372

349

333

304 313

223

Occurrence of substrate-induced structural changes is well known for many enzymes, particularly for G-proteins; there are distinct structural differences between GTP- and GDP-bound forms [80]. Possible interactions of GTP in heme domain as reported with sGC NO may also play a role in this process [34]. However, it is noted that GTP alone produced no effect. Synergistic effect of YC-1 with GTP is important. Therefore, we examined the effects of substrate analogues on the RR spectra of sGC CO in the presence of YC-1 and the results are given in Fig. 12. In the presence of YC-1 and GTP, several new bands appear or the intensities of existing bands are changed compared with those in their absence. The Fe CO band at 473 cm−1 in the absence of YC-1 and GTP (Fig. 12(a)) is replaced by two bands at 489 and 522 cm−1 in their presence (Fig. 12(b)) as mentioned above. Since the excitation wavelength is shifted to 413 nm where the intensity of the 522 cm−1 band is fairly enhanced, the relative intensity of the 522/489 cm−1 bands is different from that in Fig. 7(c). Even with this excitation wavelength, intensities of the 522 and 489 cm−1 bands upon addition of YC-1 (Fig. 12(b)) are not so large and seem to be insignificantly increased by further addition of GDP (Fig. 12(c)) or cGMP (Fig. 12(d)). However, addition of -S-GTP which is a GTP analogue but is not hydrolyzed, increased their intensities significantly (Fig. 12(e)), almost similar to the case of GTP (Fig. 12(f)). Very interestingly, the addition of ATP canceled the changes induced by YC-1 and GTP as delineated by Fig. 12(g). It is emphasized that the effects are less intense in the presence of only YC-1 or only GTP (or

(g)

Raman intensity

(f) (e)

(d) (c) (b)

(a)

Raman shift /cm–1

Fig. 12. RR spectra in the 200–600 cm−1 region of sGC CO in the absence and presence of 200 M YC-1 and 1 mM substrate/analogues. (a): sGC CO; (b): sGC CO + YC-1; and (c–g): sGC CO+YC-1+substrate/analogue [(c): GDP; (d): cGMP; (e): -S-GTP; (f): GTP; and (g): ATP]. Excitation wavelength used was 413.1 nm. Laser power was <200 W. Reproduced from [74].

Activation Mechanism of sGC

557

-S-GTP). Simultaneous presence of a substrate or substrate analogue with YC-1 yields a very specific effect on RR spectra. Since cGMP and GDP have less effects both guanine and triphosphate groups are considered to interact with the protein simultaneously. Since ATP has a reverse effect of GTP, triphosphate group is not essential in this effect and guanine residue is important. It has been shown that ATP inhibits sGC and the inhibition is also dependent on the number of phosphate groups present in the substrate/analogues [81]. Another study also shows that ATP inhibits sGC by binding to a regulatory site that prefers ATP compared with GTP [82]. Although it is not clear where the regulatory site is located, it is established in this study that the conformational changes induced by the regulatory site and by the heme moiety are of similar type and they are along the same direction for GTP but along the opposite direction for ATP.

6. CHANGES IN HEME VIBRATIONS The changes described above are concerned with the Fe C O moiety of the heme. In addition to them some noticeable changes are seen for heme vibrations. The intensity of the band that appears at 424 cm−1 for sGC CO (Fig. 12(a)) decreases and a new band appears at 400 cm−1 in the presence of YC-1 and GTP (Fig. 12(f)). This change is accompanied by the change in intensity of other bands at 1600 and 1615 cm−1 (Fig. 7). The band pair at 424/400 cm−1 may be assigned to the vinyl bending mode, whereas that at 1615/1600 cm−1 is assigned to the vinyl C C stretching mode at positions −2 and −4 [84]. The intensity change suggests an interchange of intensities between the two vinyl groups at positions −2 and −4, as if the heme was turned over [83]. The intensity of another band at 372 cm−1 , which could be attributed to the propionate bending mode [84], also changes. We also observed a new band at 223 cm−1 for sGC CO in the presence of YC-1 and GTP. This might be assigned to the Fe CO tilting vibration. Similar RR spectral changes of the vinyl and propionate groups were also observed for sGC NO [34]. This inclines us to speculate that these changes may arise from direct interactions of vinyl and propionate side chains of heme with YC-1, and thus the binding of YC-1 to the heme pocket. The other possibility is that YC-1 may interrupt the hydrogen-bonding network between heme and protein, resulting in a geometrical change of the vinyl and propionate groups. Recent proposal of hydrogen bonding between two propionate groups and 139-Arg/135-Tyr also supports this explanation [51]. The apparent rotation of heme mentioned above may practically indicate the rotation of the axis of non-planar distortion of heme due to the change of hydrogen bonding network. In the presence of YC-1 and BAY the 424 cm−1 band that normally appears in sGC NO also disappears and the band at 399 cm−1 appears. Intensity of propionate bending mode at 367 cm−1 also increases in the presence of YC-1 and BAY. GTP also has similar effect although the effect is less compared to YC-1 and BAY.

7. MECHANISM OF ACTIVATION The activation of sGC could be divided into two distinct but related phenomena. First, the binding of NO or CO to the heme and accompanied changes in the heme geometry. This triggers the second event, which is a structural change at the polypeptide level

558

B. Pal and T. Kitagawa NO NO

Fe N

NO

NO

NO

Fe

Fe

N NH

OR

NO

N NH

Fe

NH

N NH

5c sGC

6c sGC-NO

5c sGC-NO

Fig. 13. NO-dependent axial coordination of heme iron in sGC. Adapted from [86].

responsible for the phosphate group transfer and cyclization producing cGMP from GTP. Neither of these two processes is well understood. The uncharged radical NO among its three redox forms of NO− , NO· and NO+ significantly activates sGC [85]. As illustrated in Fig. 13, the NO-dependent activation of sGC requires the binding of NO to the heme prosthetic group, probably resulting in a transient 6c His Fe NO species [30,31], which in turn forms the 5c Fe NO complex with cleavage of the Fe His bond by either NO dependent or independent manner. There are contradictory hypothesis about how the Fe NO bond of the first NO adduct is broken and the role of binding of the second NO molecule [86,87]. Nevertheless, it is in general agreement that the cleavage of the Fe His bond probably results in a conformational change of the protein, which renders an elevated level of activity in the C-terminal region, although exact mechanism of this structural change leading to activation is not clear yet. It was demonstrated that a microbial cytochrome c (cyt c ) possesses similar spectroscopic properties to that of sGC [88]. Cyt c contains a His-ligated 5c heme in some pH region and forms stable adducts with CO and NO, although it does not bind with O2 . It was noted that NO binding to the cyt c heme cleaves the Fe–His bond like sGC, while CO does not. The crystallographic analysis of cyt c from Alcaligenes xyloxidans (Axcyt c ) shows that NO binds to the proximal side whereas CO binds to the distal side of the heme [88]. It has been reported for Axcyt c that the conversion of 6c heme-NO to 5c heme-NO exhibits a concentration dependence on NO [88]. On the other hand, a similar NO-concentration dependence was observed for sGC [41]. Another study with the simulation technique also predicted that NO might bind to both sides of the heme in sGC [86]. It has been proposed that NO binds to the heme resulting in two different populations of sGC NO [41]. In the minor population (about 28%) the 6c to 5c conversion is rapid (with a rate of about 20 s−1 ), whereas in the remaining 72% population this conversion is slow (with a rate between 0.1 and 1.0 s−1 ). It has been argued that this slow conversion in the second population originates from the reassociation of 105-His to the ferrous heme [41]. It has been argued that the sGC is activated by NO only in the presence of cGMP, the product of enzymatic reaction. In the absence of cGMP, a non-active 5c sGC NO is produced with Soret band of 399 nm, same as that of active form [89]. However, all these models explain structural changes at the heme proximity only, while the details at the polypeptide level are still speculative. Nevertheless, the cleavage of Fe His bond results in an initiation of

Activation Mechanism of sGC

559

conformational change leading to elevated levels of activity. Activation of NO-sensitive sGC by metal-free protoporphyrin IX also supports this hypothesis [90]. However, the cleavage of Fe His bond is not sufficient to activate sGC, because a mutant lacking the proximal histidine did not show an increased catalytic rate [27,91]. This suggests that the coordination of the proximal histidine to the heme group is not only indispensable to prepare the active conformation of protein but also to mediate its delegation to the catalytic site inducing the stimulatory action. In fact, it was proposed by Senter et al. [92] that the reaction catalyzed by sGC proceeds via a single displacement reaction involving the configuration inversion of the phosphorous group. The 105-His, after release from heme group, is thought to accept a proton to complete the conversion of GTP to cGMP. This hypothesis was supported by Koesling and coworkers [77]. Recently, it has been proposed that the heme group acts as an endogenous inhibitor of sGC through the coordination of 105-His [93]. Disruption of Fe His bond releases the restrictions imposed by this bond and allows the formation of an optimally organized catalytic center in the heterodimer [93], although such an interpretation postulates that the catalytic site is located near the heme pocket. Figure 14 outlines the probable activation mechanism of sGC by CO and YC-1 on the basis of the present knowledge, although the mechanism of activation of sGC by CO in the absence and presence of YC-1 is also not always clear. Recent studies suggest the formation of 5c heme for CO-bound sGC in the presence of YC-1 [35,36]. The amount of 5c

Specific Activity: 8000 nmol.min–1.mg prot–1 P3 His

Fe–CO

&

(X)

His

Fe–CO

GTP

YC-1

β

α

α

β

His His

Fe–CO

&

His

X

Fe–CO

β

P3

α

Fe–CO P3

GTP

β

YC-1

α

Specific Activity: 250 nmol.min–1.mg prot–1

His

β

P3

Fe–CO

β

P3

α

α

Fig. 14. Schematic illustration of our proposed model for synergistic activation of sGC by CO and YC-1/BAY. X would be His residue other than 105-His. A gray triangle and P3 mean YC-1/BAY and triphosphate, respectively. Reproduced from [74]. (see Plate 18.)

560

B. Pal and T. Kitagawa

species generated may not be sufficient to interpret the raised activity. Regarding the binding site of YC-1, the studies with photo-affinity labels indicated that 3H-meta-PAL, an NO-independent sGC stimulator similar to YC-1, binds to two Cys residues (Cys-238 and Cys-243) of the -subunit [94] and this is used in Fig. 14. The presence of multiple binding sites for YC-1 is supported by other studies [95]. Its binding to the heme pocket is proposed from our study and others [72,75], although it is not explicitly represented in Fig. 14. While many models postulated the formation of a 5c-heme due to the cleavage of Fe His bond in the presence of CO and YC-1 [77,92], it is always a mixture with 6c-heme and there is a study reporting that the cleavage of the Fe His bond is not necessary [96]. The role of side chains of heme group in the activation of sGC has been ignored for a long time, although its importance was noted by Ignarro et al. [90] nearly 20 years ago. It was shown recently that porphyrin conformation and the side chains may play an important role in sGC activation [51,97]. The study with sGC NO in the presence of GTP indicated the intensity enhancement of the propionate bending mode at 365 cm−1 [34]. Our recent experiments showed that YC-1 also affects the vinyl as well as propionate bending modes in sGC NO. We also detected some changes in the vinyl stretching (1600–1615 cm−1 ) and bending modes (424 to 400 cm−1 ) along with some change in the propionate bending mode for sGC CO in the presence of YC-1 and GTP [36]. Marzocchi and Smulevich [98] pointed out that the vinyl bending frequency depends on the dihedral angle(s) of the vinyl group against the plane of pyrrole ring, and the appearance of vinyl stretching mode at 1615 cm−1 is indicative of nearly trans conformation with  = 135 . This may further suggest that YC-1 probably sits in the heme cavity and the steric hindrance/other unknown interaction(s) force the vinyl group to adopt a specific conformation.

8. PROSPECTS NO is already established as a signaling molecule, but there is still some controversy about the role of CO, although more and more evidences suggest its role as a signaling molecule [99–102]. “Cross-talk” between NO and CO is also considered [103]. It has been suggested that CO may regulate the action of NO and thus the CO mediated regulation of sGC. In a recent study it has been suggested that CO-induced vasodilation may require a permissive enabling action of cGMP that can be produced by basal NO, but not a CO-induced elevation of cGMP [104]. However, the concentration of physiologically produced CO may not be enough to activate sGC and being gaseous, it should be synthesized and used rapidly as it lacks vesicular storage [105]. This opens up the possibility of existence of a “sensitizer” in physiological condition equivalent to the xenobiotic YC-1. CO is produced by the degradation of heme by a heme oxygenase (HO) and regulates cell signaling mostly by activating sGC although this enzyme may not be the only target of CO as evident from the recent discovery of another CO receptor protein, NPAS2 that regulates circadian rhythm [2]. In fact, the expression of sGC and that of constitutive HO isoforms are found to be co-localized in many parts of the brain [106]. Some products or intermediates of heme metabolism might act as sensitizers of sGC. It has been reported that biliverdin and bilirubin enhanced the rate of the CO-induced relaxation in pig gastric fundus, although the mechanism is not clear [107].

Activation Mechanism of sGC

561

On the other hand, the corresponding in vitro assays suggest that biliverdin inhibits and bilirubin has no significant effect on sGC activity [108]. It was also reported that the inhibitory effect of biliverdin was different from ODQ, which is a potent inhibitor of sGC. Discovery and identification of the new compound may resolve this issue in future.

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37]

Aono, S. and Nakajima, H. (1999) Coord. Chem. Rev., 192, 267–282. Dioum, E.M., Rutter, J., Tuckerman, J.R. et al. (2002) Science, 298, 2385–2387. Hou, S., Larsen, R.W., Boudko, D. et al. (2000) Nature, 403, 540–544. Delgado-Nixon, V.M., Gonzalez, G. and Gilles-Gonzalez, M.A. (2000) Biochemistry, 39, 2685–2691. Chang, A.L., Tuckerman, J.R., Gonzalez, G. et al. (2001) Biochemistry, 40, 3420–3426. Gilles-Gonzalez, M.A., Ditta, G.S. and Helinski, D.R., (1991) Nature, 350, 170–172. Bredt, D.S. and Snyder, S.H. (1994) Annu. Rev. Biochem., 63, 175–195. Nathan, C. (1992) FASEB J., 6, 3051–3064. Nioche, P., Berka, V., Vipond, J. et al. (2004) Science, 306, 1550–1553. Karow, D.S., Pan, D., Tran, R. et al. (2004) Biochemistry, 43, 10203–10211. Uma, S., Yun, B-G. and Matts, R.L. (2001) J. Biol. Chem., 276, 14875–14883. Stone, J.R. and Marletta, M.A. (1996) Biochemistry, 35, 1093–1099. Waldman, S.A. and Murad, F. (1987) Pharmacol. Rev., 39, 163–196. Heinloth, A., Brune, B., Fischer, B. and Galle, J., (2002) Atherosclerosis, 162, 93–101. Baranano, D.E., Ferris, C.D. and Snyder, S.H. (2001) Trends Neurosci., 24, 99–106. Lucas, K.A., Pitari, G.M., Kazerounian, S. et al. (2000) Pharmacol. Rev., 52, 375–414. Nighorn, A., Byrnes, K.A. and Morton, D.B. (1999) J. Biol. Chem., 274, 2525–2531. Vermehren, A., Langlais, K.K. and Morton, D.B. (2006) J. Insect Physiol., 52, 340–348. Iyer, L.M., Anantharaman, V. and Arvind, L. (2003) BMC Genomics, 6, 490–497. Pellicena, P., Karow, D.S., Boon, E.M. et al. (2004) Proc. Natl. Acad. Sci. USA., 101, 12854–12859. Denninger, J.W. and Marletta, M.A. (1999) Biochim. Biophys. Acta, 1411, 334–350. Russwurm, M. and Koesling, D. (2002) Mol. Cell. Biochem., 230, 159–164. Andreopoulos, S. and Papapetropoulos, A. (2000) Gen. Pharmacol., 34, 147–157. Bellamy, T.C. and Garthwaite, J. (2002) Mol. Cell. Biochem., 230, 165–176. Brioni, J.D., Nakane, M., Hsieh, G.C. et al. (2002) Int. J. Impot Res., 14, 8–14. Friebe, A. and Koesling, D. (2003) Circ. Res., 93, 96–105. Pyriochou, A. and Papapetropoulos, A. (2005) Cell. Signal., 17, 407–413. Stone, J.R. and Marletta, M.A. (1994) Biochemistry, 33, 5636–5640. Kamisaki, Y., Saheki, S., Nakane, M. et al., (1986) J. Biol. Chem., 261, 7236–7241. Stone, J.R., Sands, R.H., Dunham, W.R. and Marletta, M.A. (1995) Biochem. Biophys. Res. Commun., 207, 572–577. Stone, J.R., Sands, R.H., Dunham, W.R. and Marletta, M.A. (1996) Biochemistry, 35, 3258–3262. Makino, R., Matsuda, H., Obayashi, E. et al. (1999) J. Biol. Chem., 274, 7714–7723. Kosarikov, D.N., Lee, J.M., Uversky, V.N. and Counts Gerber, N. (2001) J. Inorg. Biochem., 87, 267–276. Tomita, T., Ogura, T., Tsuyama, S. et al. (1997) Biochemistry, 36, 10155–10160. Makino, R., Obayashi, E., Homma, N. et al. (2003) J. Biol. Chem., 278, 11130–11137. Li, Z., Pal, B., Takenaka, S. et al. (2005) Biochemistry, 44, 939–946. Deinum, G., Stone, J.R., Babcock, G.T. and Marletta, M.A. (1996) Biochemistry, 35, 1540–1547.

562 [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75] [76] [77] [78]

B. Pal and T. Kitagawa Fan, B., Gupta, G., Danziger, R.S. et al. (1998) Biochemistry, 37, 1178–1184. Vogel, K.M., Hu, S., Spiro, T.G. et al. (1999) J. Biol. Inorg. Chem., 4, 804–813. Kim, S., Deinum, G., Gardner, M.T. (1996) J. Am. Chem. Soc., 118, 8769–8770. Stone, J.R. and Marletta, M.A. (1996) Biochemistry, 35, 1093–1099. Liu, Y., Ruoho, A.E., Rao, V.D. and Hurley, J.H. (1997) Proc. Natl. Acad. Sci. USA., 94, 13414–13419. Sunahara, R.K., Beuve, A., Tesmer, J.J. et al. (1998) J. Biol. Chem., 273, 16332–16338. Yu, A.E., Hu, S., Spiro, T.G. and Burstyn, J.N. (1994) J. Am. Chem. Soc., 116, 4117–4118. Traylor, T.G. and Sharma, V.S. (1992) Biochemistry, 31, 2847–2849. Zhao, Y. and Marletta, M.A. (1997) Biochemistry, 36, 15959–15964. Namiki, S., Hirose, K. and Iino, M. (2001) Biochem. Biophys. Res. Commun., 288, 798–804. Friebe, A., Wedel, B., Harteneck, C. et al. (1997) Biochemistry, 36, 1194–1198. Karow, D.S., Pan, D., Davis, J.H. et al. (2005) Biochemistry, 44, 16266–16274. Boon, E.M., Huang, S.H. and Marletta, M.A. (2005) Nat. Chem. Biol., 1, 53–59. Schmidt, P.M., Schramm, M., Schroder, H. et al. (2004) J. Biol. Chem., 279, 3025–3032. Zhou, Z., Gross, S., Roussos, C. et al. (2004) J. Biol. Chem., 279, 24935–24943. Wagner, C., Russwurm, M., Jäger, R. et al. (2005) J. Biol. Chem., 280, 17687–17693. Ko, F.N., Wu, C.C., Kuo, S.C. et al. (1994) Blood, 84, 4226–4233. Hering, K.W., Artz, J.D., Pearson, W.H. and Marletta, M.A. (2006) Bioorg. Med. Chem. Lett., 16, 618–621. Stasch, J.P., Alonso-Alija, C., Apeler, H. et al. Br. (2002) Br. J. Pharmacol., 135, 333–343. Schmidt, P., Schramm, M., Schroder, H. and Stasch, J.P. (2003) Eur. J. Pharmacol., 468, 167–174. Hobbs, A.J. (2002) Br. J. Pharmacol., 136, 637–640. Miller, L.N., Nakane, M., Hsieh, G.C. et al. (2003) Life Sci., 72, 1015–1025. Schindler, U., Strobel, H., Schonafinger, K. et al. (2006) Mol. Pharmacol., 69, 1260–1268. Pal, B., Li, Z., Ohta, T. et al. (2004) J. Inorg. Biochem., 98, 824–832. Zhao, Y., Brandish, P.E., DiValentin, M. et al. (2000) Biochemistry, 39, 10848–10854. Medvedev, A., Bussygyna, O., Pyatakova, N. et al. (2002) Biochem. Pharmacol., 63, 763–766. Ignarro, L.J., Degnan, J.N., Baricos, W.H. et al. (1982) Biochim. Biophys. Acta, 718, 49–59. Gupta, G., Kim, J., Yang, L. et al. (1997) Protein Expr. Purif., 10, 325–330. Foerster, J., Harteneck, C., Malkewitz, J. et al. (1996) Eur. J. Biochem., 240, 380–386. Kharitonov, V.G., Sharma, V.S., Magde, D. and Koesling, D. (1999) Biochemistry, 38, 10699–10706. Spiro, T.G., ed. (1988) Biological Applications of Raman Spectroscopy, Resonance Raman Spectra of Heme and Metalloproteins, Vol. 3. New York: Wiley. Kitagawa, T. In Biological Applications of Raman Spectroscopy, Resonance Raman Spectra of Heme and Metalloproteins, Vol. 3 (T.G. Spiro, ed.) New York: Wiley, pp. 97–131. Kitagawa, T. and Ozaki, Y. (1987) Struct. Bond., 64, 71–114. Phillips, G.N., Jr., Teodoro, M.L., Li, T. et al. (1999) J. Phys. Chem., B103, 8817–8829. Denninger, J.W., Schelvis, J.P.M., Brandish, P.E. et al. (2000) Biochemistry, 39, 4191–4198. Martin, E., Czarnecki, K., Jayaraman, V. et al. (2005) J. Am. Chem. Soc., 127, 4625–4631. Pal, B. and Kitagawa, T. (2005) J. Inorg. Biochem., 99, 267–279. Sharma, V.S., Magde, D., Kharitonov, V.G. and Koesling, D. (1999) Biochem. Biophys. Res. Commun., 254, 188–191. Kitagawa, T., Pal, B., Ohta, T. et al. (2003) J. Inorg. Biochem., 96, 23 (Abstract of 11th ICBIC meeting). Kharitonov, V.G., Sharma, V.S., Pilz, R.B. et al. (1995) Proc. Natl. Acad. Sci. USA., 92, 2568–2571. White, D.K., Cannon, J.B. and Traylor, T.G. (1979) J. Am. Chem. Soc., 101, 2443–2454.

Activation Mechanism of sGC [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95] [96] [97] [98] [99] [100] [101] [102] [103] [104] [105] [106] [107] [108]

563

Ohta, T., Pal, B. and Kitagawa, T. (2005) J. Phys. Chem., B.199, 21110–21117. Sprang, S.R. (1997) Annu. Rev. Biochem., 66, 639–678. Suzuki, T., Suematsu, M. and Makino, R. (2001) FEBS Lett., 507, 49–53. Ruiz-Stewart, I., Tiyyagura, S.R., Lin, J.E. et al. (2004) Proc. Natl. Acad. Sci. USA., 101, 37–42. Mak, P.J., Podstawka, E., Kincaid, J.R. and Proniewics, L.M. (2004) Biopolymer, 75, 217–228. Hu, S., Smith, K.M. and Spiro, T.G. (1996) J. Am. Chem. Soc., 118, 12638–12646. Dierks, E.A. and Burstyn, J.N. (1996) Biochem. Pharmacol., 51, 1593–1600. Ballou, D.P., Zhao, Y., Brandish, P.E. and Marletta, M.A. (2002) Proc. Natl. Acad. Sci. USA., 99, 12097–12101. Bellamy, T.C., Wood, J. and Garthwaite, J. (2002) Proc. Natl. Acad. Sci. USA., 99, 507–510. Lawson, D.M., Stevenson, C.E.M., Andrew, C.R. and Eady, R.R. (2000) EMBO J., 19, 5661–5671. Russwurm, M. and Koesling, D. (2004) EMBO J., 23, 4443–4450. Ignarro, L.J., Ballot, B. and Wood, K.S. (1984) J. Biol. Chem., 259, 6201–6207. Wedel, B., Humbert, P., Harteneck, C. et al. (1994) Proc. Natl. Acad. Sci. USA., 91, 2592–2596. Senter, P.D., Eckstein, F., Mulsch, A. and Bohme, E. (1983) J. Biol. Chem., 258, 6741–6745. Martin, E., Sharina, I., Kots, A. and Murad, F. (2003) Proc. Natl. Acad. Sci. USA., 100, 9208–9213. Becker, E.M., Alonso-Alija, C., Apeler, H. et al. (2001) BMC Pharmacol., 1, 13. Stasch, J.P., Schmidt, P., Alonso-Alija, C. et al. (2002) Br. J. Pharmacol., 136, 773–783. Stone, J.R. and Marletta, M.A. (1998) Chem. Biol., 5, 255–261. Carr, H.S., Tran, D., Reynolds, M.F. et al. (2002) Biochemistry, 41, 10149–10157. Marzocchi, M.P. and Smulevich, G. (2003) J. Raman Spectrosc., 34, 725–736. Boehning, D. and Snyder, S.H., (2002) Science, 298, 2339–2340. Ingi, T., Chiang, G. and Ronnett, G.V. (1996) J. Neurosci., 16, 5621–5628. Zakhary, R., Poss, K.D., Jeffrey, S.R. et al. (1997) Proc. Natl. Acad. Sci. USA., 94, 14848– 14853. Cary, S.P. and Marletta, M. (2001) J. Clin. Invest., 107, 1071–1073. Hartsfield, C.L. (2002) Antioxid. Redox. Signal, 4, 301–307. Koneru, P. and Leffler, C.W. (2004) Am. Physiol. J. Heart Circ. Physiol., 286, H304–H309. Boehning, D., Moon, C., Sharma, S. et al. (2003) Neuron, 40, 129–137. Verma, A., Hirsch, D.J., Glatt, C.E. et al. (1993) Science, 259, 381–384. Colpaert, E.E., Timmermans, J.P. and Lefebvre, R.A. (2002) Eur. J. Pharmacol., 457, 177–186. Koglin, M. and Behrends, S. (2002) Biochem. Pharmacol., 64, 109–116.

The Smallest Biomolecules Diatomics and their Interactions with Heme Proteins Edited by A. Ghosh © 2008 Elsevier B.V. All rights reserved.

Chapter 21

Insights into Heme-based O2 Sensing from Structure–Function Relationships in the FixL Proteins Kenton R. Rodgers, Graeme R.A. Wyllie, and Gudrun S. Lukat-Rodgers Department of Chemistry, Biochemistry and Molecular Biology, North Dakota State University, Fargo, ND 58105-5516

Abstract FixL proteins are bacterial heme-containing signal transduction proteins responsible for sensing the O2 concentration in the organism’s environment. In Sinorhizobium meliloti, FixL is a protein histidine kinase that, together with its response regulator FixJ, constitute an oxygen-sensitive switch for regulation of the organism’s nitrogen fixation and microaerobic respiration genes. The O2 sensitivity of the switch is such that it transitions during the process of symbiosis in alfalfa roots. Bradyrhizobium japonicum FixL similarly regulates microaerobic and anaerobic respiration genes during symbiosis in soybean roots. FixLs responds to low oxygen concentrations with increased autophosphorylation activity of their kinase domains. The phosphorylated FixL provides a phosphoryl group to FixJ within a FixLJ complex. The phosphorylated FixJs are transcriptionally active toward their target genes. The FixL kinase domain is inhibited when the heme in FixL is oxygenated. Kinetic and thermodynamic studies of ligand binding to both ferrous and ferric FixLs have shown a generally low affinity for ligands relative to myoglobins. These relatively low ligand affinities are attributable almost completely to diminished rates of ligand binding. The heme and its environment in liganded and unliganded FixLs have been characterized by UV–visible spectroscopy, resonance Raman spectroscopy, EXAFS, and X-ray crystallography. These studies have revealed that in the purified proteins, the heme is converted from a six-coordinate low spin state to a five-coordinate high spin state upon O2 release. Comparisons of spectroscopic and structural characteristics of deoxyFixL with oxy-FixL, met-FixL CN, FixL CO, and FixL NO complexes indicate that distal affects in the heme pocket are, at least in part, responsible for communicating the ligation state of the heme to the kinase domain. The mechanisms by which ligand binding events are communicated from the heme to the kinase domain involve propagation and/or amplification of the ligation-coupled conformational transitions of the heme and its immediate protein environment. More recently, time-resolved experiments examining the non-equilibrium, ligand-coupled dynamics initiated by O2 , CO, and NO photolysis from the corresponding FixL complexes have begun to shed light on the landscape of the switching coordinate. Site specific mutation of a number of amino-acid residues in the region of the heme environment have also provided valuable insight into the initial stages of the signal transduction event. Current thinking and understanding of the mechanism for signal transduction in the FixLJ systems are discussed in the context of these physical investigations.

FixL

565

ABBREVIATIONS 5c 6c Å ADP Arg ATP BjFixL BjFixLH CO cm−1 CN− Ct EPR EXAFS F− {FeNO}6 {FeNO}7 fs FTIR His HPH HS Ile ImH IR kon Leu LS ms N–MeIm NMR N− 3 NO ns OH− O2 PAC PHD P(O2 ) ps rR SCN− SmFixL SmFixLN SmFixLH

five-coordinate six-coordinate Angstrom (1 Å = 10−10 m) adenosine diphosphate arginine adenosine triphosphate FixL from Bradyrhizobium japonicum heme domain construct from BjFixL carbon monoxide wavenumber (reciprocal wavelength) cyanide ion center of the heme pyrrole nitrogen plane electron paramagnetic resonance extended x-ray absorption fine structure fluoride ion “ferric” FeNO complex in which FeNO harbors 6 valence electrons “ferrous” FeNO complex in which FeNO harbors 7 valence electrons femtoseconds (1 fs = 10−15 s) Fourier transform infrared histidine HIF-prolyl hydroxylase high spin isoleucine imidazole infrared rate constant for ligand binding leucine low spin milliseconds (1 ms = 10−3 s) N–methyl imidazole nuclear magnetic resonance azide ion nitric oxide nanoseconds (1 ns = 10−9 s) hydroxide ion molecular oxygen or dioxygen photoacoustic calorimetry prolyl hydroxylase domain O2 pressure picoseconds (1 ps = 10−12 s) resonance Raman thiocyanate ion FixL from Sinorhizobium meliloti a heme domain construct from SmFixL a different heme domain construct from SmFixL

566

SmFixL∗ SmFixLT XO s X−Y

K.R. Rodgers et al.

a soluble heme-kinase construct from SmFixL a different soluble heme-kinase construct from SmFixL neutral diatomic ligand in which X = C, N, or O microseconds (1 s = 10−6 s) a normal vibrational mode having substantial X Y bond stretching character

1. INTRODUCTION The past 20 years or so have borne witness to a myriad of studies focused on proteinbased sensing of the diatomic gas molecules, CO, NO, and O2 . Oxygen-sensing mechanisms fall into two principal classes. One takes advantage of the thermodynamic oxidizing power of O2 to affect oxygen atom transfer (i.e., -ketoglutarate-dependent prolyl hydroxylases, PHDs/HPHs [1–3]) or multi-electron transfer (i.e., flavin binding PAS proteins [4], Aer [5–7] and NifL [8,9], the iron sulfur-containing protein FNR [10,11]) dependent changes in protein structure and/or conformation. The other relies on reversible O2 binding and the ability to transduce the free energy change associated with making or breaking of the Fe O2 bond to drive structural reorganization of the protein. The O2 -binding sensors identified to date are heme-b containing proteins, some of which comprise PAS domains (i.e., FixLs [12,13], EcDos [14–16], AxPDEA1 [17,18]), and others do not (i.e., the globin-coupled sensor BsHemAT [19–21]). This review describes the biophysical characterization of FixLs and the contributions of these studies to the current understanding of their signal transduction mechanism. Plant growth is essentially limited by the availability of nitrogen which is high in the atmosphere but limited in the earth’s crust. The legumes are able to thrive in nitrogendeficient soils by virtue of their ability to enter into symbiotic relationships with Gramnegative soil-dwelling bacteria from the genus Rhizobium, which carry genetic code for the enzymatic machinery necessary to fix nitrogen and to survive under the conditions required for fixation [22]. The symbiotic relationships are initiated upon infection of the plant roots by the bacteria. During symbiosis, the bacterium enters the plant cell and resides therein as an endosymbiont having a plant-derived membrane. Among the plant responses to infection is its expression of the O2 -binding heme protein leghemoglobin, which sequesters the O2 pool and provides for its controlled delivery to the endosymbiont. Under these conditions, the symbiotic bacteriod begins to express high-affinity oxidases and/or denitrification enzymes that support its aerobic respiration under the microaerobic conditions or anaerobic conditions of the root nodule, respectively. It also begins to express nitrogenase along with its associated proteins to catalyze the conversion of N2 to ammonia. However, the catalytic turnover of N2 by nitrogenase is rather slow and is facilitated by considerable expenditure of ATP [23]. Moreover, activated (reduced) nitrogenase [24] and the N N bonded intermediates are unstable with respect to oxidation by O2 . Given the material and energy commitments to producing and running the nitrogenfixing machinery, there would be little survival value in constitutive expression of the O2 -sensitive nitrogenase enzyme. Thus it is tightly regulated at the transcriptional level in inverse proportion to P(O2 ), to which Rhizobia must respond in such a way as to generate signals that can directly or indirectly regulate (a) the cell’s ability to respire under conditions of low O2 tension and (b) its commitment to the energy and materials

FixL

567

intensive enterprise of N2 fixation. Although FixLs are known from a number of other organisms, two have been substantially characterized by multiple biophysical methods; they are the FixLs from Sinorhizobium meliloti (SmFixL) and Bradyrhizobium japonicum (BjFixL). Thus, the following discussion will focus on these two FixL proteins. The regulatory roles of SmFixL. Sinorhizobium meliloti targets alfalfa for symbiosis. In S. meliloti, both fixK and nifA transcription are regulated by the SmFixLJ system [12,13]. FixK regulates transcription of genes whose products, which include a highaffinity cytochrome oxidase, facilitate microaerobic respiration as P(O2 ) drops during symbiosis. NifA regulates transcription of the nitrogen fixation genes. Therefore the genes for both microaerobic respiration and nitrogen fixation are under FixLJ control. The regulatory roles of BjFixL. Bradyrhizobium japonicum targets soybean for infection and subsequent symbiosis. Only fixK2 is regulated by BjFixLJ [25,26]. FixK2 regulates the transcription of genes for both microaerobic respiration and anaerobic respiration wherein NO− 3 is the terminal electron acceptor [25]. Transcription of these genes provides the ability of the organism to transform its respiration mechanism as O2 tension drops during symbiosis. Since transcription of nifA does not appear to be regulated by FixLJ in B. japonicum N2 fixation is not under FixLJ control in this organism. The FixL protein is not found exclusively in nitrogen-fixing bacteria as demonstrated by the recent report of the oxygen-sensing CcFixL from the bacterium Caulobacter cresentus [27]. Here, CcFixL and the subsequent downstream regulatory proteins FixJ and FixK regulate expression of a number of genes which encode for both multiple anaerobic terminal oxidases and a number of carbon and nitrogen metabolic enzymes. The current understanding of the regulatory roles played by SmFixL and BjFixL facilitating the symbiotic lifestyles of their respective organisms stands on results from a variety of experimental approaches including microbial genetics, heme ligand binding kinetics, autophosphorylation and phosphotransfer assays, steady state and time-resolved spectroscopies, and solution and solid state structural methods. Several recent studies have also focused on the effects of a number of site-specific mutations in the vicinity of the heme environment. Characterization of these mutants, their reactivity, and ligand affinity has provided additional information upon the specific roles played by these residues in the early stages of the signal transduction mechanism. This review focuses primarily on the biophysical studies of the FixLs performed to date and their contribution to our understanding of the mechanism by which FixL transduces changes in the O2 chemical potential to modulate the autophosphorylation rate of its kinase domain and the transfer of the phosphoryl group to FixJ.

2. FixL PROTEINS The FixLs are multidomain, cytoplasmic proteins. BjFixL is soluble [28] and SmFixL contains an N-terminal membrane-spanning domain that localizes it to the cytoplasmic side of the inner membrane [26]. Both proteins comprise a heme-binding sensor domain and a kinase domain. Both the sensor and kinase domains of the FixLs are required for P(O2 ) responsive transcriptional regulation. Together with their corresponding FixJs, the FixLs constitute two-component signal transduction systems wherein FixL is kinase active under microaerobic or anaerobic conditions [28]. The phosphorylated histidine kinase of FixL is competent to phosphorylate a conserved aspartic acid in the

568

K.R. Rodgers et al. His 285 (phosphorylation site)

FixL

55.0 kDa

FixL*

42.8 kDa

FixLN

16.1 kDa

505

1

127

505

127

260 His 194 N

O

OH

N

Fe N N

HO

O

Fig. 1. Schematic diagram of SmFixL sequences [29,31].

N-terminal signaling domain of its response regulator, FixJ. Phosphorylation derepresses the C-terminal domain of FixJ, at which point it becomes transcriptionally active toward binding to the promoters of its target gene(s). Two soluble SmFixL truncations were generated [29–31], an 18 kDa construct, SmFixLN, contains the heme domain alone and a 42 kDa protein, SmFixL∗ comprises both the heme and kinase domains. Both of these proteins are soluble and SmFixL∗ is a functioning sensor-kinase; that is, its kinase activity is inversely responsive to P(O2 ). Figure 1 shows a schematic diagram of these constructs, indicating the domains [29–31], the proximal heme ligand [32], and the site of kinase phosphorylation [32]. Although SmFixLN binds O2 reversibly, it exhibits neither kinase nor phosphotransferase activity. This work clearly showed that regulation of kinase activity was intimately coupled to the level of heme oxygenation. More recently, it has been shown that autophosphorylation of FixLs and phosphoryl transfer to FixJ occurs in a FixLJ complex [33]. These early results of SmFixL characterization were exciting for two principal reasons. First, SmFixL was the first bonafide O2 -sensing protein that involved reversible O2 binding. This introduced a theretofore unknown role for heme proteins and presented the intriguing prospect that heme-based sensors could be ubiquitous. Secondly, the FixLJ system was shown to be a member of the broadly conserved bacterial two component systems, which regulate a host of functions [34]. Given the extensive knowledge and databases for the sensitivities of heme structure, reactivity, and spectroscopic signatures to protein environment, the FixLJ system presented an opportunity to exploit those heme signatures as reporters of the intra-domain, inter-domain, and possibly the intermolecular events that culminate in kinase activation and transcriptional activation of FixJ. The availability of the soluble SmFixLN and SmFixL∗ proteins opened the door to general biophysical characterization of the sensor domain and to the design of experiments to probe the mechanism of kinase activation and phosphoryl transfer to SmFixJ. Subsequently, other heme (SmFixLH) and heme-kinase (SmFixLT) truncations were produced [35]. The soluble FixL from B. japonicum, BjFixL, was also cloned and expressed in Escherichia coli [35]. BjFixL exhibits kinase activity that is inversely responsive to P(O2 ). Analogous to SmFixLN, a deletion derivative of BjFixL comprising only the heme domain (BjFixLH) was also constructed. BjFixLH was shown to bind O2 reversibly but, like SmFixLN, it exhibits neither kinase nor phosphoryl transfer activity.

FixL

569

A further truncated version of BjFixLH has also been reported [36] which revealed differences in ligand binding compared to the parent BjFixLH following deletion of multiple residues from both N- and C-terminal ends. A recent work also describes improvements in the production and purification of a number of FixL truncations [37]. In addition, the first high-resolution MALDI-TOF mass spectra for the FixL proteins are reported which provide accurate correlations to the predicted molecular weights. With the development of this improved method of preparation, production of large quantities of high-purity FixL constructs should become more straightforward, opening up the possibility of many studies which perhaps had been material limited to date.

3. EARLY PHYSICAL CHARACTERIZATION OF FixLs AND THEIR HEME LIGAND COMPLEXES 3.1. Coordination Environment Early characterization of the FixLs focused on the thermodynamic stabilities of their exogenous ligand adducts and the rates of their formation and dissociation. In some cases, characterization of the heme was carried out on both the heme domain and the functional heme kinase in an effort to gain insight into how changes in interactions between the heme and kinase might be coupled to the ligation state of the heme [35–42]. As is the case with all known O2 -binding heme proteins, the oxygenated hemes are six-coordinate (6c) and low-spin (LS) [43]. Release of O2 leaves a 5c HS ferrous heme which has been shown by resonance Raman (rR) spectroscopy [39,43], site-directed mutagenesis [32], and X-ray crystallography [44,45] to be coordinatively linked to the protein through the imidazole (ImH) side chain of a proximal histidine. SmFixL is known to self associate and to form a heterotetramer with FixJ. While there is reason to suggest that there is transphosphorylation of the kinase domains within the heterotetramer [46], this functional intimacy does not appear to support cooperative O2 binding [35,47]. Based on the early recognition of correlations between level of kinase inhibition and the spin states of exogenous ligand complexes, a spin state model for switching of the kinase activity was set forth. This model was based on (a) the observation that met-BjFixL F exhibits nearly the same kinase activity as deoxy-BjFixL and (b) a correlation between kinase activity and the met-BjFixL:met-BjFixL CN ratio in a series of solutions of varying [CN− ] [28]. This was a reasonable hypothesis in the context of structure–function relationships in other O2 -binding proteins such as the cooperative hemoglobins. However, there were also early indications that although a change in spin state was likely to be necessary to trigger the switch of kinase activity, it was unlikely to be sufficient. The Fe ImH bond, as judged by the relatively low Fe ImH frequency of 210 ± 1 cm−1 [39–41,43] in 5c HS deoxy-FixL∗ (Fig. 2), is rather weak, consistent with conformational strain on this bond in the deoxy state. Such strain is well recognized as a means by which energy yields due to formation of 6c LS hemes are used to affect conformational transitions that can be propagated out of the heme pocket. Interestingly, the Fe ImH frequency is virtually identical in the heme kinase, SmFixLN.

K.R. Rodgers et al.

300

400

587

473 489

ν(Fe–NO) 557 ν(Fe–O2) 571 587 587 587

476 491 498

491

(B)

ν(Fe–CO)

342 360 386 414 310

275

200

(A)

388 413

388 412

346, ν8 367 383 409 344

292 308 310

341 349

254 274 251 274

216

ν(Fe–His)

211

570

500

(C)

(D)

600

Raman shift (cm–1)

Fig. 2. Low frequency rR spectra of SmFixL∗ complexes showing the Fe-ligand stretching regions. (A) deoxy-SmFixL∗ in 50 mM Tris/HCl pH 7.8, 441.6 nm excitation, (B) SmFixL∗ O2 in 50 mM Tris/HCl pH 7.8, 413.1 nm excitation, (C) SmFixL∗ NO in 80 mM sodium phosphate pH 7.8, 5% glycerol, 406.7 nm excitation, and (D) SmFixL∗ CO in 50 mM Tris/HCl pH 7.8, 413.1 nm excitation.

This somewhat surprising result is consistent with the heme-kinase interactions in kinaseactive deoxy-SmFixL∗ having little influence on proximal bonding or conformation in kinase-active SmFixL at equilibrium. Given the insensitivity of the Fe ImH bond to the heme-kinase interactions, a direct conformational switching “connection” through which the spin state transition could be communicated from the proximal heme pocket to the kinase seemed tenuous at the time. An early comparative spectroscopic and kinetic study suggested that kinase influence on the equilibrium coordination chemistry of the heme with exogenous ligands was probably exerted through modulation of distal non-bonded interactions [39]. Subsequent to solution of the first BjFixLH crystal structures [48], it was suggested that the distal heme pocket structure supported a strictly distal steric model for communication between the heme and kinase domains [49]. Other subsequent studies (vide infra) have shown distal non-bonded interactions to play crucial roles in ligand discrimination and ligand specific signal transmission [45,50–52]. The heme in deoxy-SmFixL is quite susceptible to autoxidation [53], yielding metSmFixL by what is apparently a unique mechanism. Like deoxy-FixLs, the autoxidized ferric heme in the met-FixLs is also 5c and HS [35–39]. Exogenous ligand adducts of both ferrous and ferric FixLs have been made and studied. In addition to O2 , the ferrous adducts include the FeII CO [35] and {FeNO}7 [38] complexes. Adducts of the ferric

FixL

571

− hemes include OH− , F− , CN− , N− 3 , SCN , ImH, and NO [38,39,42]. In general these studies reveal relatively low affinities for exogenous ligands, relative to Mb, with small rate constants for ligand association accounting largely for the diminished stabilities. Moreover, heme-kinase interactions generally decrease the stabilities of these adducts by further decreasing their ligand association rates. While conformational strain on the proximal Fe ImH bond likely contributes to the relatively low ligand association rates and stabilities of the corresponding ligand complexes, the insensitivity of its stretching frequency suggests that it does not account for the negative influence of the kinase domain on the ligand-binding rates. This negative control has been suggested to derive from influence of the kinase on structure and/or dynamics in the distal heme pocket [39,54].

3.2. Redox Chemistry of FixL The redox potential of deoxy-BjFixLH has been determined to be reversible with E1/2 = 68 mV vs. NHE [55], a more positive value than that of myoglobin (0.05 mV vs. NHE) [56]. Under atmospheric conditions, the rate constant for autoxidation of SmFixLH is ∼60 times that of sperm whale Mb [53]. Mechanistic work on this reaction revealed a two term rate law consistent with two classes of [O2 ] dependent autoxidation pathways with the second term being linear in both [SmFixLH O2 ] and [deoxy-SmFixLH]. The second term accounts for the kinetic behavior characteristic of the oxygen protection effect, wherein O2 competes with a nucleophile such as water for the ferrous heme center at high P(O2 ). This was an unexpected result, given that the ferric heme of met-SmFixL is pentacoordinate, suggesting that spontaneity of this autoxidation cannot be driven by formation of a hexacoordinate heme. The original interpretation of this seemingly contradictory result was that the kinetics were consistent with electron transfer between deoxy-SmFixL and SmFixL O2 , with this outer sphere electron transfer rate slowing as the heme is oxidized to the ferric form. A subsequent study has shown oxygen dependent dissociation of dimeric HS SmFixLN [57]. Thus, the autoxidation kinetics may rely on the intimate contact provided by dimerization of the heme domain, which could facilitate rapid electron transfer (ET) from deoxy-SmFixL to SmFixL O2 . As P(O2 ) is increased, the fraction of dimeric SmFixL decreases, thereby protecting it against efficient oxidation by SmFixL O2 . This competition between O2 binding and SmFixL dimerization may be the behavior that mimics the competition between O2 and solvent in the oxygen protection mechanism observed in the autoxidation of sperm whale Mb. Ligand rebinding studies of O2 , NO, and CO complexes of FixLs are proving useful in investigations of the non-equilibrium dynamics of heme-ligation coupled structural and conformational transitions [57–59]. The properties of these complexes are summarized in the following paragraphs. Figure 2 shows rR spectra of SmFixL∗ O2 , SmFixL∗ NO, SmFixL∗ CO, and deoxy-SmFixL∗ . The Fe XO and Fe ImH bands are indicated and will be discussed in the relevant sections below. FixL O2  O2 is the most inhibiting of the neutral diatomic ligands for the wild type FixLs [51,60,61]. Consequently, investigation of the FixL O2 complexes has been widely pursued. The rR spectrum in Fig. 2 and that reported for SmFixL O2 reveal the Fe O2 band at 570 cm−1 [43]. This band has been assigned by isotope substitution, which elicits a 24-cm−1 shift of this band to lower frequency. This is typical of oxy-Mb

572

K.R. Rodgers et al.

and oxy-Hb. Although the details of a distal H-bonding interaction involving the bound O2 were not known at the time of this study, the similarity between Fe O2 frequencies of FixL∗ O2 and the oxy-globins suggested that it was present. X-ray crystallography later revealed Arg220 as the H-bond donor in oxy-BjFixL [52]. Moreover, and like the globins, the rR signature of SmFixL O2 is consistent with a 6c LS ferric heme containing an axial superoxide ligand [43]. Rate constants for FixL O2 association and dissociation have been reported [35]. The dissociation rates are similar to those for Mb but the association rates for FixLs (kon = 22 × 105 M−1 s−1 ) are roughly two orders of magnitude slower than those of Mb. Thus, FixL O2 complexes are considerably less stable than oxy-Mb due to slow association rates. As is the case for other heme–O2 complexes, the O O band has not been observed by rR spectroscopy due to poor resonance enhancement. FixL CO complexes. Carbonyl complexes of heme proteins have been widely studied because of their high thermodynamic stabilities, the sensitivity of FeCO vibrational signatures to heme environment, and their high quantum efficiencies for Fe CO photolysis [62]. As for O2 , the thermodynamic stabilities of the CO adducts are roughly 102 -fold lower for FixLs than for SW Mb and these diminished stabilities are entirely attributable to decreased rates of CO association (kon ≈ 104 M−1 s−1 ) [35,57]. Additionally, the affinities of FixLs for CO are only about 10-fold higher than for O2 , as compared to the ∼200-fold difference in SW Mb. Thus FixLs are able to diminish the thermodynamic advantage of CO over O2 by a factor of ∼20 compared with SW Mb. It has been argued based on crystal structures of BjFixL O2 and BjFixL CO that the energetic basis of this discrimination is attributable to a distal H-bonding network involving the coordinated O2 ligand, the guanadinium side chain of Arg220, and a distal water molecule that is not observed when CO is bound [45,52]. It is possible that this lack of non-bonded interaction is linked to COs roughly twofold smaller inhibition of kinase activity relative to O2 [59]. In contrast to the heme–O2 complexes, both the Fe CO and C O bands are well enhanced in the Soret-excited rR spectrum. Thus the frequencies of these vibrations can be used as diagnostic probes of the distal heme environment because of their sensitivity to the nature of non-bonded interactions between bound CO and the distal heme pocket. The Fe CO and C O frequencies (Fe CO = 498 cm−1 , C O = 1962 cm−1 ) from solution rR studies of SmFixLT CO show that it exhibits less distal H-bond donation to bound CO than does MbCO [54]. The first indication that interdomain interactions were manifested in the distal heme pocket came from rR spectra of SmFixLN CO and SmFixL∗ CO [39]. The spectra revealed a 2.7 cm−1 difference in the frequencies of their porphyrin  ∗ electron density marker bands (4 ) with SmFixLN CO exhibiting the higher frequency. This frequency difference is consistent with more heme → CO -backbonding due to a less distorted or more H-bond accepting FeCO moiety. As the crystal structure of BjFixL CO suggests that bound CO is not H-bonded [45], it is reasonable to suggest that the frequency difference is attributable to less off-axis distortion of the FeCO moiety in the heme domain–CO complexes. A subsequent report of the corresponding Fe CO and C O frequencies for SmFixLH CO and SmFixLT CO revealed kinase-dependent shifts that are also consistent with increased backbonding in SmFixLH CO [54]. The same report provided evidence from solution EXAFS measurements indicating that ∠FeCO is indeed smaller when the kinase is present (154 vs. 171 ). These results were interpreted as an indication that the kinase exerts steric influence on the bound ligand structure,

FixL

573

suggesting that release of the exogenous heme ligand would be sensed by the kinase domain through relaxation of those same steric interactions. FixL NO complexes. Both ferrous {FeNO}7 and ferric {FeNO}6 FixLs have been prepared and characterized. The FeNO vibrational frequencies, as determined by rR spectroscopy, are typical of {FeNO}7 [63] and {FeNO}6 [42] hemes. The rate constant for association of NO with met-BjFixL has been determined to be 10 × 106 M−1 s−1 , roughly 10- and 102 -fold greater than those for O2 and CO, respectively [38]. Despite the clear thermodynamic advantage of NO over O2 for binding to FixLs, its ability to inhibit kinase activity is only about half that of O2 [61]. The {FeNO}7 complexes have been examined by X-band EPR spectroscopy as a means of examining both the coordination number and identifying the proximal His ligand. Reports of varying heme–NO coordination number have appeared in the literature. Three reports of rR and EPR spectra are consistent with temperature-dependent mixtures of 5c and 6c heme nitrosyls [38,42,64] in both SmFixL NO and BjFixL NO. Even in the most recent study claiming complete absence of rR bands attributable to 5c BjFixL NO [64], the high-frequency spectrum exhibits a weak band at the correct 3 frequency for a pentacoordinate heme–NO. This shoulder is observed in spite of the investigators’ having used 413.1-nm Raman excitation, which discriminates against resonance enhancement of 5c {FeNO}7 hemes. In both the Bj and Sm proteins, the low-temperature limit of the pentacoordinate form is ∼50%. Another report only shows a 20 K EPR spectrum of SmFixLN NO, which appears to indicate virtually 100% pentacoordinate {FeNO}7 heme [43]. This temperature dependence of the 5c:6c {FeNO}7 heme population ratio was exploited in a variable temperature rR study to reveal a specific protein–protein interaction between the SmFixL heme domain and FixJ [42]. This was the first study to suggest that the interaction of the sensor kinase and the response regulating phosphoryl receiver protein might be energetically coupled to the ligation state of the heme. The reactivity of {FeNO}6 SmFixL toward reductive nitrosylation has been investigated by UV–visible spectrophotometry [42]. The reductive nitrosylation reactions of {FeNO}6 SmFixLN proceeds via two pathways while only one pathway appears available to {FeNO}6 SmFixL∗ at pH 9.5. The multiple pathways in SmFixLN were attributed to conformational heterogeneity in the heme pocket of SmFixLN that is eliminated by the presence of the kinase domain. However, the possibility of differences in heterogeneity due to corresponding differences in extents of possible NO-modulated self association cannot be discounted as the reason for differences in reductive nitrosylation kinetics of ferric SmFixLN NO and SmFixL∗ NO. Like the ferrous CO complexes of the FixLs, their {FeNO}7 complexes are photolabile. However, in contrast to the CO adducts, the heme NO recombination reaction is dominated by recombination of the geminate heme/NO pair [59]. In other words, recombination of nearly all of the heme and NO occurs before NO diffuses out of the heme pocket. This reaction typically occurs on the ps timescale, which facilitates probing of the local conformational dynamics on the timescale of recombination lifetime. This provides a view of the conformational reorganizations within the heme pocket that are coupled to ligand release. Such ultrafast laser photolysis studies have been carried out on BjFixL [59] and will be further discussed below in the context of signal transmission to the interdomain interface.

574

K.R. Rodgers et al.

4. STRUCTURAL STUDIES 4.1. X-ray Crystallographic Studies of Equilibrium Structures Perhaps it is an effect of the increased conformational flexibility inherent in the hemekinase construct but, without exception, all FixL structures reported to date are of the heme domain only. While the majority of these structures are of BjFixLH in varying states of ligation and oxidation [44,45,48,51,52,65,66], only five-coordinate high spin structures of SmFixL have been reported [44,65]. Unfortunately, in spite of the efforts of multiple groups, the structure of a functional heme-kinase remains elusive. Nonetheless, structural comparisons of the unliganded and liganded heme domains, which at least correspond in ligation to the kinase active and kinase inhibited heme domain components of the larger construct, illustrate some of the structural changes which occur upon the coordination of exogenous ligands. The structure of the five-coordinate high-spin complex deoxy-BjFixLH, reported to 2.4 Å resolution [45] as illustrated in Fig. 3 reveals the heme located in a pocket form by five antiparallel -sheets. Three of these strands, G, H, and I form the distal heme environment with the long (F) helix located on the proximal side. These are linked by the flexible FG loop, comprising residues Thr209 to Arg220 in BjFixL, a characteristic feature of PAS domains which appears to provide an impressively versatile binding environment for a number of small molecules capable of inducing protein conformational changes in response to a physical or chemical change in the surroundings of the protein [67,68].

(A) H

H I

A F

C

G

I

A

B E

D

F

C

G B E

D

(B)

Fig. 3. Cross-eyed stereo views of ribbon diagrams of (A) deoxy-BjFixLH (PDB code: 1LSW) and (B) deoxy-SmFixLH (PDB code: 1EW0) showing the striking structural similarities of the two sensor domains. The heme is bound to the protein through the ImH side chains of His200 and His194, respectively. Labeling scheme in (A) is that set forth in [48]. N-terminal and C-terminal helices have been omitted from (A) and (B), respectively, for clarity of viewing and ease of comparison.

FixL

575

The FG loop lies close to the heme in the five-coordinate deoxy-BjFixLH adopting the so-called closed conformation and it is this loop which then undergoes significant changes upon proximal ligand coordination. Indeed, some slight changes in the orientation of the FG loop are observed between the structure at 2.4 Å resolution [45] and more recent higher resolution structures [66]. The latter reports two 1.9 Å resolution structures differing in crystallographic space group. The rhombohedral form (R32) which contains one monomer per asymmetric unit corresponds with that of the previously reported structure, whilst a monoclinic form (C2), containing two distinct monomers per asymmetric unit, is unique to this new work [66]. All structures display variations in both the position of the FG loop and that of the flexible HI turn. The difference in the latter feature is in part justified by intermolecular interactions since, in the monoclinic form, this turn lies directly between the two monomers. The fact that the most significant differences are observed in both the FG loop and the HI turn suggests these regions tend to be more conformationally flexible and in the solid state at least, more easily influenced by environmental effects such as crystalline lattice contacts. The earliest FixL structures reported were those of the ferric complexes met-BjFixLH and met-BjFixLH CN [48], with resolutions of 2.4 Å and 2.7 Å, respectively and the latter being a kinase-inhibiting form of the heme domain. Again with the exception of the FG loop, overall there is extensive correlation between the backbone alignments of these two structures. The position of the proximal histidine ligand is essentially unchanged between the two structures which were interpreted as indicating that proximal forces are invariant between the kinase-active and kinase-inhibited forms of the protein. This suggests that the proximal coordination environment plays little or no role in the signal transduction mechanism. Coordination of the cyanide ligand causes a displacement in the both Leu and Ile side chains which line the distal pocket in order to accommodate this new ligand. This is accompanied by a flattening of the porphyrin ligand with these ligation coupled changes in heme planarity suggested by the authors as the possible driving force of the kinase regulatory conformation transformation of the FG loop. For the ferrous state, structures comparing the five-coordinate kinase-active deoxyBjFixLH [45] and the kinase-inhibited BjFixLH O2 [52] are illustrated in Fig. 4. The specific residues illustrated are those of the FG loop which clearly show noticeable changes in both conformation and H-bonding status upon ligand release. In the 2.3 Å resolution structure of BjFixLH O2 , the guanidinium group of Arg220 is found to form an H-bond with the coordinated dioxygen. Upon ligand release, this is replaced following movement of the residue with a new H-bond between Arg220 and one of the heme propionate groups. This is also accompanied by dissolution of an H-bonding network found in BjFixLH O2 which employs a water molecule to act as a structural template anchoring the flexible carboxy-terminal end of the FG loop. The increased flexibility of the FG loop in the deoxy form is accompanied by a loss of heme planarity, an analogous result to that seen between the ferric met-form and its cyanide coordinated derivative. Thus, in both oxidation states, the increase in coordination number is accompanied both by a flattening of the heme and by an apparent increase in the extent of kinase inhibition. The structures of deoxy-BjFixLH and BjFixLH O2 would thus appear to represent the two endpoints in the series of conformational changes and non-bonding interaction reorganizations. This series, associated with the ligation coupled changes of the heme then drives the increase in catalytic activity of the kinase domain upon ligand release.

576

K.R. Rodgers et al. (A)

(B)

Fig. 4. Cross-eyed stereo projections of (A) 5c HS BjFixLH (PDB code: 1DRM) and (B) BjFixLH O2 (PDB code: 1DP6). The FG loop conformation appears to be largely dictated by H-bond interactions with the heme propionate groups. Major changes in protein conformation are localized to the FG loop. Note the H-bonded water molecule templating the loop conformation at the top of the distal pocket and the stabilization of Arg220 in the distal pocket by an H-bond interaction with the bound O2 in (B) [52].

Strong kinase inhibition is observed for the CN− , ImH, and N-MeIm ligated BjFixLH, all of which have been structurally characterized. Yet of these, only the structure of the cyanide-coordinated species reveals an H-bond network incorporating the guanidinium group of Arg220. Since neither imidazole or the N-methyl imidazole forms these H-bonds, it would appear that it is not the H-bond which is critical to kinase inhibition but rather the presence of increased stearic bulk. In the case of the cyanide derivative, the H-bonding pulls in the bulk of the guanidinium group whilst both the ImH and N-MeIm complexes provide their own bulk and kinase inhibition occurs even in the absence of interactions with Arg220. The structure of the ferrous nitrosyl complex BjFixLH NO [52] was reported at 2.5 Å resolution. NO coordination also results in kinase inhibition, albeit to a lesser extent than that observed for BjFixLH O2 . The Fe N O angle (∠FeNO = 146 ) is comparable with that observed in a series of NO-coordinated proteins [69] with the bent NO supporting the ferrous oxidation state. Yet, for this diatomic ligand, the distal H-bonding interactions observed for BjFixLH O2 and BjFixLH CN are not present. Neither are they present in the structures of BjFixLH CO reported as with the deoxyBjFixLH in two distinct crystalline space groups [45,66]. The original structure [45] reports a 2.4 Å resolution structure in the ubiquitous rhombohedral (C32) space group. CO ligation was achieved by treatment with high pressures of CO following reduction by dithionite. However, the more recent report [66] which reports both rhombohedral (2.0 Å resolution) and monoclinic (1.8 Å resolution) structures similar to those discussed previously for deoxy-BjFixLH, described a more benign preparative method which requires simply equilibration under an atmosphere of CO. Perhaps the most noticeable difference in these structures lies in the Fe C O moiety with the earlier structure

FixL

577 (A)

(B)

Fig. 5. Stereo views of (A) 5c HS BjFixLH (PDB code: 1DRM) and (B) BjFixLH CO (PDB code: 1LSV), illustrating the similarities in the conformations of their FG loops and the H-bond interactions between the FG loop and the heme propionates. By contrast with the O2 complex, both Arg side chains maintain their deoxy H-bond partners and the bound CO ligand does not appear to be invovled in H-bonding [45].

exhibiting ∠FeCO = 157 . In the higher resolution structures, a more linear Fe C O geometry is observed with angles ranging from 169 to 171 and the Fe C O moiety lying perpendicular to the porphyrin plane. The more linear structure is also supported by the fact that there are no distal polar side chains present in the hydrophobic pocket which are able to interact with the CO which would have resulted in a more bent geometry. Figure 5 illustrates a comparison of the rhombohedral structure of BjFixLH CO with that of the previously described deoxy-BjFixLH [45]. Overall, there is a high backbone correlation with the only major deviations occurring in both the FG loop and the HI turn. Leu236 of the H strand is displaced by increased steric interactions upon CO coordination. Yet CO binding does not break the H-bond interaction between Arg220 and heme propionate 7. CO binding is known to cause kinase inhibition, although like NO, to a lesser extent than O2 . So what is the structural basis of this kinase inhibition? The increase in steric bulk whether from the coordination ligand such as HIm or by movement of the guanadinium group of Arg220 is not occurring for either CO or NO. These diatomic ligands do not appear themselves to possess sufficient steric bulk since in the case of O2 and CN− , the bulk proposed to result in inhibition arises from movement of the guanadinium group which is clearly not occurring in the CO or NO structures. The H-bonding network remains intact for these latter two ligands as well, eliminating this as a possible cause for the inhibitory effect. But from the crystal structures of the CO and NO derivatives, what if any changes are responsible for the increase in kinase inhibition relative to the deoxy-form of the protein? The authors of the most recent high resolution crystal structures [66] emphasize the changes in the FG loop and HI turn following CO binding as the most significant result. They postulate that it is through these regions of conformational flexibility and specifically through residues Leu236 and Arg220 which lie on the distal heme side that

578

K.R. Rodgers et al.

the ligand-coupled changes are correlated with the kinase-inhibiting ability of the protein This may occur via structural changes occurring upon the protein surface particularly in the interdomain region. Figure 6 shows the RMSD of the main chain atoms for all of the structurally characterized BjFixL complexes [45]. This plot reveals the differences between the backbone conformations of the most and least kinase-inhibiting BjFixL complexes. The most pronounced differences lie in the FG loop sequence. It has been suggested that the apparent

(A)

Main-chain RMSD for each residue with respect to met-BjFixLH 3 CO NO cyanomet imidazole Melm oxy deoxy

2.5

RMSD (A)

2 1.5 1 0.5 0 154

164

174

184

194

204

214

224

234

244

254

264

Residue numbers (B)

Overall RMSD with respect to met-BjFixLH 1.2 All the atoms Main-chain

1

Heme

RMSD (A)

0.8

0.6

0.4

0.2

0

deoxy

CO

NO

cyanomet imidazole

Melm

Oxy

Ligand-bound forms

Fig. 6. Plots of RMSD obtained from least-squares fitting of deoxy-BjFixLH and each of the six ligand-bound structures, CO (red), NO (violet), cyanomet (blue), ImH (green), N–MeIm (cyan), and O2 (orange) to met-BjFixLH. (A) The RMSD for each residue based on its main chain atoms only. (B) The overall RMSD for each structure based on all the atoms (blue), the main chain atoms (red), and the heme (ivory). Reproduced with permission from [45].

FixL

579

inabilities of NO and CO to induce these changes could be a means by which nature avoids abhorrent kinase inhibition by these strong-field -acid ligands. Yet, until a structure of a heme-kinase construct is determined, the exact effects of ligand binding and release upon quaternary structures are likely to remain elusive. Further studies on the effects of both mutation of the Arg220 and other groups which may play a role in the H-bonding network are discussed in the section “Site Specific Mutation Studies”. It would appear that from a simple comparison of the unligated structures of both BjFixLH [45] and SmFixLN [44] in both the ferric-met and ferrous-deoxy states that there are major structural correlations between the heme domains of the two proteins. Based upon the structural similarities, one would perhaps expect analogous changes to occur upon ligand coordination. Yet, attempts to diffuse a variety of exogenous ligands into the crystals of deoxy-SmFixL have been uniformly unsuccessful. Whilst this lack of results may simply arise from the delicate nature of these crystals, there is perhaps a second more interesting possibility. Could it be that ligand coordination in SmFixL cause much more significant structural changes than those observed in BjFixLH. Perhaps, these are of a sufficient magnitude so as to destabilize the crystalline lattice. As to the nature of the ligand-coupled changes, it is not clear whether ligand-induced conformational reorganization is inter-molecular or intra-molecular. But it may be worth mentioning in this context, the observation that O2 coordination was found to cause dissociation of homodimer (SmFixLN)2 in solution [57]. Whether ligand coordination causes a change of similar magnitude to occur in the solid state alas at this times remains unobservable.

4.2. Solution Studies of Equilibrium Structure High resolution structures from multidimensional NMR data remain elusive. At least for SmFixL, this is likely a consequence of the propensity for purified FixLs to self associate into complexes having sufficiently long correlation times to scuttle attempts at high-resolution NMR spectroscopy. However, several studies of the hyperfine shifted heme spectra of both HS and LS met-FixLs have been published [70,71]. These spectra are well resolved and most of the hyperfine shifted heme resonances have been made. The proximal ImH proton resonances have also been assigned [71]. Assignment of these resonances was accomplished in a more recent study [72] using a further truncation of SmFixLH2 CN, and several mutants thereof. Through studies utilizing labelled 13 C15 N, the resonances for the H-bonding arginine side chain were also assigned. These results are discussed in more depth in the section on “Mutants”. Assignments of other proximal and distal protein resonances remain to be made. One EXAFS study has been reported for a series of heme ligand adducts of the heterodimeric (SmFixLT)2 (SmFixJ)2 complex [54]. The results of these experiments revealed two interesting relationships. The first is that in solution both the Fe NHis bond length and the equilibrium distance between the center of the pyrrole nitrogen plane (Ct) and the coordinated nitrogen atom of the proximal His, R(Ct NHis ), does depend upon whether an exogenous ligand is bound and upon the nature of that ligand. Second, R(Ct NImH ) is directly correlated with kinase activity, as illustrated in Fig. 7, which shows kinase activity plotted vs. R(Ct NImH ) [54]. This ligand-dependent variability in Fe NHis bond length suggests that ligand-coupled changes in proximal conformational strain could play a role in the signal transmission between the heme and kinase domains of SmFixL.

K.R. Rodgers et al. Kinase activity (pmol P-FixL /mmol FixL /5 min)

580 30

deoxy

met-F 25 20

met

15 10 5

met-CN

0 oxy 2.0

2.1

2.2

2.3

2.4

2.5

2.6

2.7

Ct -NImH (Å)

Fig. 7. Plot of FixL autophosphorylation activity vs. distance between the heme center and the N atom of the proximal ImH, as determined by solution EXAFS measurements. A correlation between activity and distance is clearly apparent. Figure adapted from [54].

5. NON-EQUILIBRIUM, LIGAND-COUPLED DYNAMICS AS A PROBE OF SIGNAL TRANSDUCTION 5.1. Methods The spectroscopic and X-ray scattering experiments described thus far have been used to probe thermodynamically stable states of the FixL proteins. While these experiments provide insight into the structural bases of the stabilities of these states, they do not necessarily yield insight into the structural dynamics that occur in conjunction with the ligand-coupled switching of enzyme activity. Because of the photolability of heme–XO complexes, where XO CO, NO, and O2 , the activity switching that is driven by change in exogenous ligand availability is amenable to study by ligand photolysis experiments. These experiments are generally initiated by pulsed laser-induced photolysis of the Fe XO bond. In this experiment, photolysis by laser flashes that range in duration from fs to ns is intended to mimic the change in chemical potential that occurs as P(O2 ) drops in the protein solution. The photoinduced release of the XO ligand triggers the sequence of changes in protein structure and conformation that are energetically coupled to Fe XO bond scission and which, in the case of FixLs, culminates in the de-inhibition of enzymatic activity in its kinase domain. This sequence of timedependent conformational transitions can be probed by any one of a number of optical spectroscopic methods, including UV and/or visible absorbance, IR absorbance, Raman scattering, and fluorescence. These probes can yield information on rates of change, the number of intermediates along the switching coordinate, and the nature of the change(s) in structure and conformation that characterize the intermediates. The changes that influence the rate of XO ligand rebinding can be tracked by measuring the timedependent changes in the heme absorbance spectrum that occur as the XO complex reforms. While this experiment may not reveal, in a straight forward fashion, the nature of remote conformational changes, it provides a convenient and sensitive means of tracking the corresponding relaxations that slow the ligand rebinding rate. Alternatively,

FixL

581

spectroscopic probes specific to chromophores in the region of conformational change can be employed to (a) track relaxations as they propagate through the protein and between domains or subunits and (b) probe the changes in structure that characterize the intermediates. For example, time-resolved UV and visible rR spectroscopy, time-resolved fluorescence spectroscopy, and step-scan FTIR spectroscopy can facilitate tracking the propagation of conformational relaxations that occur in response to photolytic release of the heme XO ligand from a detectable ensemble of protein molecules. In addition to monitoring such changes spectroscopically, Photoacoustic calorimetric (PAC) methods [73] allow measurements of both the magnitude and timescale of changes in both sample volume and enthalpy following the photolysis event. This has the advantage of being able to discriminate between species that would otherwise be spectroscopically indistinguishable. Both the transient absorbance spectroscopy and the PAC experiments lend themselves naturally to cycling, which facilitates signal averaging. It is important to be vigilant in monitoring the integrity of protein samples over the course of these experiments, as photolysis pulses carry large amounts of energy with peak powers ranging from kilowatts to gigawatts. These methods are most fully exploited with knowledge of atomic-level structure and site-directed mutagenesis so that time-dependent spectral features can be associated with specific regions of the protein. Several published time-resolved spectroscopic studies based on FixL ligand photolysis and recombination are described below in order of increasing timescale, even though the results were not published in this order. The reasoning behind this presentation is that by developing the course of events from the ultrafast to the millisecond timescales, we track the time-dependent propagation of the free energy change originated by photolysis of the heme ligand.

5.2. Tracking Recombination of the Geminate Heme/XO Pair by Ultrafast Transient Absorbance Spectroscopy A recent study of ultrafast ligand rebinding to BjFixLH has revealed several interesting properties of its CO, NO, and O2 complexes [59]. The geminate pair produced by fs photolysis of BjFixLH O2 recombines in a single phase with an unprecedented lifetime of <5 ps. Additionally, the Soret absorbance band maximum for the geminate pair was observed at 424 nm, as shown in Fig. 8. Although the structural and electronic bases of this blue-shifted Soret band is not clear at this time, both the kinetic and spectroscopic behavior of the geminate pair have been attributed to the O2 ligand being tethered to the vicinity of the 6th coordination site via its H-bonding interaction with Arg220. Rebinding of NO and CO occur at slower rates than O2 with NO rebinding in three geminate phases and CO exhibiting essentially no recombination up to a few ns after photolysis (Table 1). The three kinetic phases of geminate NO recombination are attributed to different geminate states over which the heme/NO pair can distribute because, unlike O2 , NO is not tethered by an H-bond interaction with any distal heme pocket residues [64]. The lack of geminate CO recombination suggests that the CO ligand cannot access a distal binding site after photolysis and diffuses out of the heme pocket at a rate competitive with geminate recombination. The Soret maxima of the geminate heme/NO and heme/CO pairs are also blue-shifted relative to the equilibrium

582

K.R. Rodgers et al. 1

Absorption

0.8

unliganded CO dissociated

0.6

NO dissociated 0.4

O2 dissociated 0.2

0 400

420

440

460

480

500

Wavelength (nm)

Fig. 8. Visible spectra of the BjFixLH/CO, NO, and O2 geminate pairs determined by femtosecond transient absorbance spectroscopy. Reproduced with permission from [59].

Table 1. Kinetic parameters for ligand rebinding reactions following laser flash photolysis. Millisecond lifetimes were calculated from rate constants reported in [35] and [57] Protein

Effector Ligand CO

Reference

NO

O2

CO

gem. total

NO

gem. total

O2

gem. total

BjFixLH

 4 ns

0.0

5.3 ps (0.62) 20 ps (0.24) 220 ps (0.08)

0.94

4.7 ps

0.90

BjFixL SmFixLN SmFixLH SmFixL∗

130 ms 42 ms 37 ms 53 ms 150 ms 53 ms

SmFixLT

5.7 ms 3.8 ms

3.8 ms

[59]

[35] [57] [35] [57] [35]

deoxy-BjFixL spectrum, 427 and 431 nm, respectively. These Soret maxima are closer to that in the spectrum of equilibrium deoxy-BjFixLH than that of the heme/O2 pair. Thus, the extent of the blue shift of the Soret maximum immediately after photolysis appears to be inversely related to the geminate recombination rate. These seemingly smaller perturbations relative to the spectrum of equilibrium deoxy-BjFixLH were attributed to the lack of distal H-bond interactions involving the bound NO ligand. At the time of this study, there was solution evidence from rR spectroscopy of SmFixLH CO that the CO ligand is not strongly H-bonded [39,54]. However, since the crystal structure of BjFixLH CO had not yet been reported, it was not known that this structure is also

FixL

583

ν(7)

675

498 ν(Fe – CO), geminate ν(Fe–CO)

475 488

415 δ(CβCaCb ) (vinyl)

384 δ(CβCcCd ) (propionate) 365

288

(B)

408

(C)

344

340 ν(8)

218 ν(Fe–His)hv

210 ν(Fe–His)deoxy

consistent with a lack of H-bond donation to the CO ligand from Arg220. In the context of the crystallographic results [45], these transient absorbance data are consistent with a defining role for the distal H-bonding network in the kinetics of ligand binding and conformational reorganization. The observation of electronically distinct heme intermediates identified in the aforementioned transient absorbance study is also consistent with an earlier time-resolved rR study in which a single 3 ns laser flash was used for both CO photolysis and excitation of Raman scattering by the photolysis product [58]. This study revealed that immediately after photolysis of bound CO, the proximal Fe NHis bond is left in a non-equilibrium state. Figure 9 shows that this transient state is marked by an 8 cm−1 upshift in the intense Fe His band relative to its position in the spectra of equilibrium deoxy-SmFixLN and deoxy-SmFixL∗ . The existence of this conformational intermediate suggests that ligand-coupled changes in proximal conformational strain are also likely to play a role in the signal transmission between the heme and kinase domains of SmFixL,

250

300

350

400

415

387

200

342

(A)

450

500

650

700

Raman shift (cm–1)

Fig. 9. Single color, single pulse transient resonance Raman (rR) spectra showing (A) steady state rR spectrum of SmFixL∗ CO with 200 W of cw excitation at 441.6 nm from an HeCd laser, (B) authentic deoxyFixL∗ recorded with 435.7-nm pulsed excitation, pulse repetition rate = 50 Hz, pulse energy = 0.3 mJ/pulse, and (C) difference spectrum showing the features of non-equilibrium deoxyFixL∗ that persists during the 3 ns laser pulse after CO photolysis. Adapted from [58].

584

K.R. Rodgers et al.

despite the suggestion of BjFixL crystal structures to the contrary. This result suggests that there is more to FixL’s signal transduction mechanism than is apparent from the crystal structures of the ligand-bound and ligand-free states of BjFixL at equilibrium. Specifically, it is consistent with signal transduction involving conformational dynamics in the proximal as well as in the distal heme pockets. These dynamical processes described above are confined to the immediate surroundings of the heme. They are the motions that direct conformational free energy out of the heme pocket toward the interdomain interface. However, these relaxations constitute only the earliest in the sequence of conformational relaxations that culminate in upregulation of FixL’s kinase activity. Given that NO and the physiological ligand, O2 , tend to recombine almost completely on the ps timescale, these reactions are not optimal for probing the long-range conformational relaxations that occur on the s to ms timescale of tertiary and quaternary (or interdomain) conformational transitions. In order to investigate the kinetics of such relaxations by ligand recombination, a slower ligand-binding reaction was necessary. Carbon monoxide fulfills this requirement, as its second order recombination occurs on the ms timescale. SmFixL CO recombination was used to identify and track slower processes that could characterize interdomain transitions [57]. A second study examined second order CO recombination to both BjFixLH and a further truncation thereof [36]. This study also reported the first example of PAC studies upon a heme sensor protein. While it appears that the level of kinase inhibition by CO and NO are only about half of that imposed by O2 [61], the premise of this CO rebinding study was that ligand-coupled conformational changes occurring in response to Fe CO photolysis would mimic those that occur in response to thermal dissociation of the physiological O2 ligand. The study of second order ligand recombination for SmFixLN CO and SmFixL∗ CO were investigated [57] and determined with a quantum yield of ∼80%. Second order recombination of the sensor domain, SmFixLN-CO, occurs in a single kinetic phase with a rate constant of 15×104 M−1 s−1 . Interestingly, the functional sensor kinase, SmFixL∗ CO, recombines in two kinetic phases having second order rate constants of 12 × 104 M−1 s−1 and 42 × 103 M−1 s−1 . The initial interpretation of this result was that recombination occurred via two parallel pathways corresponding to distinct but interconvertible conformers, which was followed by a subsequent relaxation of the mixture to the single slowly recombining conformation with a rate constant of 13.5 s−1 . However, more recent results [74] have demonstrated that for SmFixL∗ CO, both the fast and slowly rebinding conformer exist in equillibrium in solution as illustrated in Fig. 10. Hence the presence of the kinase domain results in the formation of a mixture of two spectroscopically indistinguishable SmFixL∗ CO conformers which possess different CO rebinding rates. These were assigned by the authors as the kinase-active and kinase-inhibited forms of the protein. This then suggests that the lower kinase inhibition from the CO actually results from there being only a fraction of the protein in solution being in a kinase-inhibiting conformation unlike the oxygen-bound system where only one conformer, a kinase-inhibiting form, is observed. Transient absorbance measurements on the heme domain BjFixLH CO recombination kinetics [36] revealed comparable rate constants (Table 1), albeit with a markedly lower quantum yield of ∼50% for the second order recombination. Studies on a further truncation of the heme domain in which a number of residues were removed from both ends of the protein sequence (residues 151–256 compared to 140–270 as found

FixL

585

(FixL*-CO)2 Conformation I

k –6

k 1 or h νslow

k2

(FixL*/CO)2 k –2

k –1

k6

(FixL*-CO)2 Conformation II

(FixL*)2 + 2CO Conformation I

k –3

k – 5 or h νfast k5

k–4

(FixL*/CO)2

k4

k3

(FixL*)2 + 2CO Conformation II

Fig. 10. Scheme showing the parallel second order recombination of CO with SmFixL∗ . Current evidence suggests that conformers I and II are kinase active and inhibited forms, respectively [74].

in BjFixLH). This truncation exhibited an increase in rate constant for second order recombination, even though the absorbance spectra do not reveal any detectable disturbances of the heme environment relative to the parent truncation. It would therefore appear that the effect of the truncation is to facilitate both ligand release to the solvent and subsequent ligand reentry to the heme pocket, a suggestion supported by the results of the PAC study reported therein. PAC measurements on the full-length BjFixLH revealed two decay phases, the first faster than the detection limits of the instrument ( < 50 ns), and the other, slower phase with a lifetime of 150 ns. In the truncated form described above this latter phase is not detected. Through comparison with the well-characterized PAC studies on Mb CO [75,76], the authors assigned the two processes with the fast phase resulting from the initial volume contraction following CO dissociation from the heme and disruption of a salt bridge. The longer ( ≤ 150 ns) lifetime process is associated with escape of CO from the geminate pair arrangement found in the heme pocket. While this phase appears absent in the truncated BjFixLH, the magnitude of both the volume and enthalpy changes of the single phase observed therein approximate to the sum of the two separate phases in the full-length heme domain. The additional truncation therefore causes accelerated release of CO from the geminate pair relationship to the solvent environment in addition the increased rate in second order ligand rebinding already described. But what then is the nature of the reorganization tentatively assigned as a salt bridge disruption? The authors suggest this either involves a surface salt bridge between Glu182 and Arg227 or based upon changes already described in the high resolution crystal structures, which show increased disorder around the Arg206 residue following CO coordination, that the interaction may be a reorientation of this residue. While PAC may be a relatively new technique, particularly in the study of heme sensor proteins, the fact that two transitions undetectable by conventional transient spectroscopy [74] are clearly observed demonstrates the value of this technique. While not reported at the time of writing, PAC measurements on a sensor/kinase construct are eagerly anticipated particularly when one considers that such measurements may allow observation of solely kinase domain based conformational changes that would otherwise be undetectable by visible absorbance or resonance Raman spectroscopy.

586

K.R. Rodgers et al.

6. ROLE OF FixL ASSOCIATION It has been known for some time and shown in multiple experimental studies that SmFixL–SmFixL and SmFixL–SmFixJ associations are coupled to the state of heme ligation in SmFixL [42,44,60,65]. A recent study has provided compelling evidence (B)

(A)

O2

Subunit B

ATP

Low affinity O2

ADP Fe

Fe

FixJ~P His His FixJ Fe ~P

ATP

His His~P Allosteric effect

Fe ADP

High affinity

High affinity

Subunit A

O2 FixJ~P

(D)

(C)

O2 High affinity Allosteric effect

ADP Fe His~P

High affinity ATP Fe

His His FixJ~P

His

Fe

Fe ADP ATP

~P

O2

Low affinity O2

Fig. 11. Illustration of the reciprocating engine model of the FixLJ phosphorylation reactions. To follow the reactions, one cycle in subunit A is gray. (A) ATP is catalyzed at the nucleotide binding site of the kinase domain because of deoxygenation of the sensor domain in subunit A, and the phosphoryl group is transferred to the histidine in subunit B. (B) ADP produced at the binding site reduces the oxygen-binding affinity of subunit B in a transacting manner. The phosphoryl group of subunit B is transferred to FixJ. (C) Phospho-FixJ and ADP are then released from subunit A. In turn, the ATP-phosphoryl transfer reaction in subunit B is enhanced because of the allosteric effect in B. (D) FixJ and ATP are reloaded into subunit A. The oxygen-binding affinity of subunit A is decreased by the ADP that is produced in subunit B, leading to the ATP-phosphoryl transfer reaction in A. On overall phosphoryl transferring reactions, two phospho-FixJ are liberated from FixL to form a regulatory active dimer (center). Oxygen binding at the high-affinity hemes (subunit (A) in A and subunit (B) in C) determines the content of kinase-active FixL in the turnover cycle, and the allosteric effect facilitates the velocity of the transitions from B to C and from D to A, resulting in the amplification of the phospho-FixJ production. If the oxygen-bound sensor domain inhibits the histidine phosphorylation of the same subunit (cis-acting repression), the model implies that the ADP allosteric effect would be exerted in the sensor domain of the same subunit (cis-acting manner). Reproduced with permission from [47].

FixL

587

that the heterotetramer (FixL)2 (FixJ)2 is mechanistically relevant due to intermolecular allosteric effects [47]. It was shown that ADP is an allosteric effector for negative feedback control of the O2 affinity of the heme domain, which translates to positive allosteric regulation of kinase activity. It has been proposed that, in place of using cooperativity to overcome the intrinsically low O2 affinity of the monomeric protein, the sensor forms a dimer wherein the affinity for the inhibiting effector is diminished by an allosteric effect of ADP. These aspects of the FixLJ phosphotransfer cascade are illustrated in Fig. 11. It has been suggested that the allosteric effect of ADP cannot be correct because if it were, the heme would have such a low affinity for O2 that the kinase could never be sufficiently inhibited to prevent transcriptional activation of FixJ [77]. This argument assumes that the level of transcriptionally active FixJ is solely determined by the kinase activity of FixL. There could be other enzymatic activities, phosphatase activity for example, that are external to the FixLJ complex and contributes to regulation of phospho– FixJ in a fashion that compliments the P(O2 ) responsive regulation. Therefore, while the logic in this argument is reasonable, the assumptions on which it is based may not be. Investigation of the kinetic effects of oxygen upon the upon the various stages of the FixL phosphorylation reaction [78] revealed specifically that O2 inhibits autophosphorylation of the (SmFixL)2 (FixJ)2 tetramer. Neither formation of the tetramer nor ATP binding shows any inhibition from the presence of oxygen.

7. SITE-DIRECTED MUTAGENESIS STUDIES Up to this point, this review has focused on the characterization of the naturally occurring protein in both B. japonicum and S. meliloti. Through both spectroscopic and crystallographic studies, the importance of a number of residues, located in the immediate heme environment, to the beginning stages of the signal transduction mechanism has become apparent. The publication of a recent series of work [55,72,79] have systematically examined the effects of site specific mutations of these residues and the effects that altering these residues has on the fundamental properties of the protein. These have included examination of ligand binding, kinase inhibition, redox potential, and structural changes as deduced by NMR and rR spectroscopy. While these are not the first reported studies on FixL mutants, these works do report a thorough study, not only on the effects of the loss of the specific residue but also the intermediate effects, where relevant, from the substituting residue. Given that the preceding sections of review, particularly the crystallographic studies have stressed the importance of the H-bonding network in the kinase inhibition process, these works examine the effects upon this in a systematic manner. As mentioned above, there have been previous studies on a series of FixL mutants. The evidence that the allosteric effect on oxygen binding, as discussed in the previous section, was categorically caused by ADP binding and not the parent ATP was carried out using a site specific mutation of the residue responsible in part for the phosphorylation of FixL [47]. However, most of the recent work has focused on studies of residues which are located in the immediate heme environment. The studies of these mutants have been carried out on FixL derived from both of the two organisms

588

K.R. Rodgers et al.

B. japonicum and S. meliloti. As mentioned previously there is a high degree of structural similarity between the crystal structures of the met form of each protein and much has also been said of the importance of the distal arginine residue which interacts with a heme propionate in the kinase-active form and with the bound ligand in the kinaseinhibited form. Due to sequence differences this residue is numbered Arg214 in SmFixL and Arg220 in BjFixL yet clearly has similar roles in both. Although the mutation studies have focused predominantly upon the changes arising from alteration of this residue, several other residues have also been selectively mutated. Since a number of these mutations cause a marked decrease in oxygen affinity and increased autoxidation rates, kinase inhibitory effects were commonly studied for both cyanide and imidazole. Whilst the specific comparability of such results to the biological effector ligand (oxygen) may be questioned, the conclusions reached from these studies do at least appear applicable. Two isoleucine residues have been identified, lying on the distal side of the heme. One of these (Ile209 in SmFixL and Ile215 in BjFixL) has been identified as a member of a hydrophobic triad of residues responsible in part for the high hydrophobic character of the heme pocket [50]. In particular this residue also appears to serve as a barrier preventing entry of water molecules into the heme pocket. An earlier study on SmFixL had shown that mutation of this residue and its neighbor Ile210 resulted in a loss of regulatory ability of the protein in air [50]. This selective mutation also results in an increase in the rate of the autoxidation reaction in addition to a reduced affinity for O2 . The recent high-resolution crystal structures of deoxyBjFixLH and BjFixLH CO [66] also revealed a slight displacement of the analogous Ile215 upon ligand binding and it could be suggested that this is in accordance with the work of Perutz [49] who suggested that the kinase inhibition may arise partially from the interaction of a bound ligand with the hydrophobic triad. Indeed the reduced inhibition observed with O2 in SmFixL following mutation of Ile209 and 210 would further support this. However, no such downregulation is observed with either CN− or imidazole [72]. Mutation of either residue results in an increased autoxidation rate, a reduced O2 affinity, yet little change in CN− affinity and an increase in imidazole affinity, the latter effect likely arising from the removal of steric bulk which would normally hinder the entrance of this larger ligand to the heme pocket. The authors [72] reviewing the earlier results of the mutations of Ile209 and Ile210 propose that the reduced kinase inhibition of the oxygen bound complex does not arise due to changes in the kinase regulation mechanism as evinced by normal regulation for both the CN− and ImH complexes but rather from a reduction in stability of the oxygen complex. Indeed mutation of these residues would appear to increase accessibility of the heme pocket to a number of nucleophilic species justifying both the larger koff and the increased rate of autoxidation. Mutations of a number of arginine residues, namely Arg200 and Arg208 in SmFixL have also been reported [72]. From the crystal structure of met-SmFixLH, these residues appear to form a salt bridge in the unligated form of the enzyme and it has been suggested that these salt bridges may indeed play a role in the kinase inhibitory process [54]. Replacement of these residues by alanine resulted in comparable kinase inhibition to the wild type protein. Similar to the isoleucine mutations there was a measurable decrease in oxygen affinity and an increase in autoxidation rate. It was therefore suggested that these residues in fact stabilize the closed conformation of the protein through salt bridge formation preventing nucleophile entrance. Hence it appears that both the arginine and

FixL

589

isoleucine residues, while playing no direct role in the signal transduction mechanism instead serve by stabilizing the oxygen ligated protein. Their mutation results in a more accessible open conformation of the heme pocket which coupled with the already low affinity of FixL for O2 results in the lower stability of the oxygen ligated form and what would appear to manifest as a loss in regulatory capability. Since neither of the preceding mutations have revealed the residue responsible for the ligand-coupled kinase inhibition, the next logical suspect in keeping with the crystallographic studies would be the arginine of the FG loop (Arg214 in SmFixL and Arg220 in BjFixL). Indeed, a number of recent studies have then focused on the effects of mutating this residue [55,72,79] determining both the changes in regulatory ability and structural effects of the mutation of this residue in particular considering the effects of the substituting residue both steric and electronic. While the majority of the mutants are spectroscopically indistinguishable from the wild type, the substitution of this arginine by histidine in both BjFixL and SmFixL results in the unexpected formation of a six coordinate ferric species [55,72]. Ordinarily, the metFixL is pentacoordinate with the axial ligation comprising only the proximal histidine residue. In the histidine substituted mutant, the ferric species exhibits spectroscopic properties consistent with the increased coordination number. This hexacoordination is temperature independent as determined by NMR and EPR spectroscopy [55] but does display a pH dependence reverting to pentacoordination at lower pHs with spectra comparable to both the wild type and other mutants. The increased coordination number and the resulting spectra are consistent with an axially coordinated water molecule which is protonated at lower pHs resulting in dissociation. While the ligation of the histidine ligand is possible, the resulting bis-histidine species would be low-spin, a result inconsistent with the high-spin character of the histidine mutant. In the ferrous state, the deoxy-FixL is for the most part spectroscopically indistinguishable from the wild type protein being five coordinate as illustrated in Fig. 12. The only significant difference spectroscopically is an increase in both frequency and intensity of the Fe His stretching band, which is consistent with an increase in the proximal Fe ImH bond strength. Although these results from both the ferric and ferrous states may appear simply as an unexpected oddity, the six-coordinate ferric state mimics the coordination environment reported for another heme-oxygen sensor, EcDos [80]. Both the FixL histidine mutant and EcDos display reversible redox behavior although the ferrous EcDos complex remains six-coordinate with a methionine ligand instead of water. Taking into account the different species formed in the histidine mutant, it is still possible to compare the effects of mutation of the arginine residue both spectroscopically and regulatory. Mutation caused an increase in autoxidation rate, most dramatically for the histidine mutant while O2 affinity again was reduced often times below the level of detection [72]. In the ferrous state, imidazole affinity increased whilst cyanide affinity was dependent upon the nature of the substituting residue. Autophosphorylation rates of CN− and imdazole varied from levels similar to those seen in the wild type to almost no regulation depending on the nature of the substituting residue. Examination of the NMR spectra of the Arg214 mutants of SmFixLH2 CN [72] resulted in a failure to observe CN− resonances for several of the mutations, suggesting that multiple conformers may coexist in solution, a conclusion borne out by rR studies. More interestingly, there appeared to be a correlation between these observations and the kinase regulatory abilities of the mutant. The R214K and R214Q mutants exhibit

590

K.R. Rodgers et al. FG loop I215

R220 I238

HN

Ferric /Ferrous Bj FixLH(WT)

HN O

NH N

O

N

O

Fe

N N

N O HN

H200

H220 FG loop

I215

I238

HN

Ferric Bj FixLH(R220H)

N H

O

H O

N

O

N

O

Fe

N N

N O HN

H200

FG loop

R220

I215

I238

Ferrous Bj FixLH(R220H)

HN N O N

O

N

HO

Fe

N N

N O HN

H200

Fig. 12. Schematic comparison of the heme pocket structures in ferric and ferrous BjFixLH (R220H) with that of the wild-type protein at neutral pH. Adapted from [55].

FixL

591

comparable inhibition to the wild type protein and display sharp NMR features for the CN− ligand. The former lysine-containing mutant with its positively charged side chain may interact directly with the negatively charged cyanide, whilst the glutamine may interact via a water molecule. Both cases support the requirement of a single non-fluctuating conformer for ligand-coupled regulation. Mutation of the equivalent residue (Arg220) in BjFixLH [55,79] results in significant spectroscopic changes. While the heme dependent bands in the rR spectra of BjFixLH CO including core size and oxidation state markers show no effects through mutation, the CO bands are not unaffected by these changes [79]. As described previously, the wild type displays a single Fe CO at ∼500 cm−1 and C O at 1968 cm−1 , both bands indicative of an open conformation. While mutation causes some shifts in these frequencies, in all cases the values remain consistent with the open conformation. It is however in the case of the glutamine and histidine mutant that a second series of such bands is observed with Fe CO at 515/517 cm−1 and C O at 1938/1943 cm−1 , values more in keeping with a closed conformation. In the case of BjFixLH O2 , the Fe O2 stretch is observed at 573 cm−1 for the wild type with a 25 cm−1 downshift following 18 O2 substitution. This is in keeping with the ∠Fe O O of 124 from the crystal structure [52]. Mutation to either histidine or isoleucine has neglible effect on these frequencies whilst mutation to glutamine results in a Fe O2 of 563 cm−1 with an isotope downshift of only 19 cm−1 . While the latter would be consistent with a ∠Fe O O of 155 [81] this is inconsistent with the observed Fe O2 and it is more likely that the changes result from formation of a new H-bond. Figure 13 illustrates the important interactions occurring between residue 220 and the heme-bound O2 ligand. In several cases, the residue serves to stabilize the oxy-FixL via interaction with the bound ligand as reflected by changes in O2 affinity.The same work also reports a correlation between a number of heme marker bands particularly those associated with core size. In general, the smaller heme core size is observed in the wild type BjFixLH O2 compared to the mutants, similarly, the largest degree of ligand-coupled inhibition is also observed for the wild type. The authors conclude that, while the Arg220 plays an important role in stabilizing the oxy-form of the complex through formation of the H-bonding network, it is not merely coordination of a strong ligand nor the displacement of residue 220 which results in the structural conformation changes that lead to signal inhibition. Rather it is the interaction between residue 220 and specifically the bound oxygen via the H-bond network which leads to enhanced ligand -acidity and subsequent heme core size changes. This is supported by the mutation studies which track these changes in heme core size with changes in back-bonding resulting from increasing interactions between residue 220 and the bound oxygen. A further study on mutation of Arg220 in BjFixL [55] looked in detail at the five coordinate ferric and ferrous forms. Similar to SmFixL [72], substitution of histidine for the arginine results in a six-coordinate ferric heme, likely with a coordinated water ligand occupying the vacant coordination site. The redox potential of the wild type, determined as 68 mV vs. NHE is decreased upon mutation with the extent of the decrease consistent with the increasing negative charge on the side chain of the substituting residue. Resonance Raman spectroscopy shows a disruption in the interactions of the heme propionate groups with residue 220 upon mutation in deoxy-BjFixLH. It is determined that

592

K.R. Rodgers et al. R220

NH

H214 O

NH2

NH H2N N

O δ–

O

H

δ+ O

O O

N N

NH

N

O

Fe

N

Bj FixLH-O2(WT)

N

N

NH2

H220

HN H2N

N H

H214

H200

O

NH

N

206R

H N

N

O

O

H

O

O O

Bj FixLH-O2(R220H)

N N

NH

N Q220

H214

N N

N

O

H 2N

HN N H

H2N

O

Fe

H200

206R

N H N

O

NH2 O

O

H

O

O O

N N

N

NH

N

Bj FixLH-O2(R220Q)

N

N

O

H2N

H2N

Fe

I220

HN N H

206R

H214

H200

O N H N

O

O

H

O

O O

N N

Bj FixLH-O2(R220I)

N

NH

N N

N

O

H 2N

H2N

Fe

HN N H

206R

H200

Fig. 13. Schematic representation of heme–O2 –protein interactions thought, based on the resonance Raman studies, to be mechanistically important in WT BjFixLH O2 . Analogous interactions in the R220H, R220Q, and R220I mutants are also shown. Adapted from [79].

FixL

593

the optimal interaction occurs with arginine based both on length of the side chain and the positive charged nature. Disruption of this interaction also results in an increase in heme planarity relative to the wild type, an effect normally observed upon coordination of a sixth ligand as is the observed weakening of the Fe–His200 bond. These changes which have been previously associated with the coordination of O2 may instead arise solely from the disruption of the propionate interactions. Furthermore, since both crystallographic [66] and spectroscopic [79] results reveal that Arg220 does not interact directly with bound CO, it is only upon binding of O2 that the Arg220 residue moves from interacting with the heme propionate to interacting with the bound ligand that maximum kinase inhibition occurs. Thus it appears that this interaction also plays an important role in the ligand discrimination process. Finally, the authors report a correlation between the hydrophobicity of residue 220 and ligand affinity of BjFixL. The correlation with the flexibility of the heme propionates as determined by rR spectroscopy leads to the suggestion that both residue 220 and propionate 7 are located in the solvent entrance channel with the increasing hydrophobicity resulting in increased ligand accessibility.

8. PERSPECTIVES The mechanistic view of FixL function has evolved continually since the first reports of their biophysical characterization appeared in the mid-1990s. The majority of the mechanistic models (spin state [28], distal steric [49], and heme flattening [48]) have tended to focus on conformational motion induced along a particular coordinate as a driving force for ligation coupled signal transduction. Moreover, most of the models have been deduced from contrasting structural or electronic properties between the kinaseactive and kinase-inhibited forms of the FixLs. However, it is becoming increasingly clear that conformational relaxations occur along multiple coordinates in response to ligand loss, some of which are transient and unlikely to be seen in crystal structures of FixLs in equilibrium states. Hence, it appears that the hypothesis of multi-coordinate ligand-coupled signaling [39,42] is gathering support as more mechanistic data appear. Although the details remain to be elucidated, the signal transduction mechanism appears to involve dynamics in both the distal and proximal pockets. The spectroscopic, structural, and kinetic results discussed herein have provided considerable insight into the mechanism of signal transduction in FixLs. The current understanding of this particular two component system (FixLJ) is largely derived from studies that, in some fashion, have exploited the spectroscopic signatures of one or more states of its heme. However, the high degree of sequence and structural homology between the protein modules of the two component systems [34] argues that much of the mechanistic insight we have gained and have yet to gain into FixL-initiated signal transduction will be transferable to other systems. Indeed, recent studies provide compelling evidence that the PAS domains comprise a set of structurally and dynamically homologous modules whose output has been tuned for species-specific purposes [27,82]. Thus, continued exploitation of the “spectroscopic handle” provided by FixL’s heme is

594

K.R. Rodgers et al.

expected to pay multiple dividends in understanding of this and other PAS-based sensing and signal transduction systems.

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36]

Epstein, A. C., Gleadle, J. M., McNeill, L. A., et al. (2001) Cell 107, 43–54. Bruick, R. K. and McNight, S. L. (2001) Science 294, 1337–1340. Wenger, R. H. (2002) FASEB J. 16, 1151–1162. Zhulin, I. B., Taylor, B. L., Dixon, R. (1997) Trends Biochem. Sci. 22, 331–333. Lux, R., Jahres, K., Bettenbrock, K. et al. (1995) Proc. Natl. Acad. Sci. USA 92, 11583–11587. Rebbapragada, A., Johnson, M. S., Harding, G. P. et al. (1997) Proc. Natl. Acad. Sci. USA 94, 10541–10546. Bibikov, S. I., Biran, R., Rudd, K. E., Parkinson, J. S. (1997) J. Bacteriol. 179, 4075–4079. Hill, S., Austin, S., Eydmann, T., Jones, T. et al. (1996) Proc. Natl. Acad. Sci. USA 93, 2143–2148. Martinez-Argudo I., Little, R., Shearer, N., Johnson, P. et al. (2005) Biochem. Soc. Trans. 33, 152–156. Khoroshilova, N., Beinert, H., Kiley, P. J. (1995) Proc. Natl. Acad. Sci. USA 92, 2499–2503. Kiley, P. J. and Beinert, H. (1999) FEMS Microbiol. Rev. 22, 341–352. David, M., Daveran, M. L., Batut, J., et al. (1988) Cell 54, 671–683. de Philip, P., Batut, J., Boistard, P. (1990) J. Bact. 172, 4255–4262. Blattner, F. R., Plunkett, G., III, Bloch, C. A. et al. (1997) Science 277, 1453–1474. Delgado-Nixon, V. M., Gonzalez, G., Gilles-Gonzalez, M. A. (2000) Biochemistry 39, 2685–2691. Sasakura, Y., Yoshimura-Suzuki, T., Kurokawa, H., Shimizu, T. (2006) Acc. Chem. Res. 39, 37–43. Tal, R., Wong, H. C., Calhoon, R. et al. (1998) J. Bacteriol. 180, 4416–4425. Chang, A. L., Tuckerman, J. R., Gonzalez, G. et al. (2001) Biochemistry 40, 3420–3426. Hou, S., Larsen, R. W., Boudko, D. et al. (2000) Nature 403, 540–544. Hou, S., Freitas, T., Larsen, R. W. et al. (2001) Proc. Natl. Acad. Sci. USA 98, 9353–9358. Aono, S., Kato, T., Matsuki, M. et al. (2002) Biol. Chem. 277, 13528–13538. Agron, P. G., Helinski, D. R. (1995). Two-Component Signal Transduction (J. A. Hoch, T. J. Silhavy, eds.) Washington, D.C.: ASM Press, pp. 275–287. Haaker, H. and Klugkist, J. (1987) FEMS Microbiol. Rev. 46, 57–71. Hill, S. (1988) FEMS Microbiol. Rev. 54, 111–130. Mesa, S., Bedmar, E. J., Chanfon, A. et al. (2003) J. Bacteriol. 185, 3978–3982. Sciotti, M. A., Chanfon, A., Hennecke, H., Fischer, H. M. (2003) J. Bacteriol. 185, 5639–5642. Crosson, S., McGrath, P. T., Stephens, C. et al. (2005) Proc. Natl. Acad. Sci. USA 102, 8018–8023. Gilles-Gonzalez, M. A., Gonzalez, G., Perutz, M. F. (1995) Biochemistry 34, 232–236. Gilles-Gonzalez, M. A., Ditta, G. S., and Helinski, D. R. (1991) Nature 350, 170–172. Agron, P. G., Ditta, G. S., Helinski, D. R. (1993) Proc. Natl. Acad. Sci. USA 90, 3506–3510. Monson, E. K., Weinstein, M., Ditta, G. S., Helinski, D. R. (1992) Proc. Natl. Acad. Sci. USA 89, 4280–4284. Monson, E. K., Ditta, G. S., Helinski, D. R. (1995) J. Biol. Chem. 270, 5243–5250. Tuckerman, J. R., Gonzalez, G., Gilles-Gonzalez, M. A. (2001) J. Mol. Biol. 308, 449–455. Stock, A. M., Robinson, V. L., Goudreau, P. N. (2000) Annu. Rev. Biochem. 69, 183–215. Gilles-Gonzalez, M. A., Gonzalez, G., Perutz, M. F. et al. (1994) Biochemistry 33, 8067–8073. Mikšovská, J., Suquet, C., Satterlee, J. D., Larsen, R. W. (2005) Biochemistry 44, 10028–10036.

FixL

595

[37] [38] [39] [40] [41] [42] [43] [44] [45] [46]

Suquet, C., Savenkova, M., Satterlee, J. D. (2005) Protein Expr. Purif. 42, 182–193. Winkler, W. C., Gonzalez, G., Wittenberg, J. B. et al. (1996) Chem. Biol. 3, 841–850. Rodgers, K. R., Lukat-Rodgers, G. S., Barron, J. A. (1996) Biochemistry 35, 9539–9548. Lukat-Rodgers, G. S., Rexine, J. L., Rodgers, K. R. (1998) Biochemistry 37, 13543–13552. Lukat-Rodgers, G. S. and Rodgers, K. R. (1998) J. Biol. Inorg. Chem. 3, 274–281. Rodgers, K. R., Lukat-Rodgers, G. S., Tang, L. (2000) J. Biol. Inorg. Chem. 5, 642–654. Tamura, K., Makamura, H., Tanaka, Y. et al. (1996) J. Am. Chem. Soc. 118, 9434–9435. Miyatake, H., Mukai, M., Park, S. Y. et al. (2000) J. Mol. Biol. 301, 415–431. Hao, B., Isaza, C., Arndt, J. et al. (2002) Biochemistry 41, 12952–12958. Ninfa, E. G., Atkinson, M. R., Kamberof, E. S., Ninfa, A. J. (1993) J. Bacteriol. 175, 7024–7032. Nakamura, H., Kumita, H., Imai, K., Iizuka, T. et al. (2004) Proc. Natl. Acad. Sci. USA 101, 2742–2746. Gong, W., Hao, B., Mansy, S. S. et al. (1998) Proc. Natl. Acad. Sci. USA 95, 15177–15182. Perutz, M. F., Paoli, M., Lesk, A. M. (1999) Chem. Biol. 6, R291–R297. Mukai, M., Nakamura, K., Nakamura, H. et al. (2000) Biochemistry 39, 13810–13816. Dunham, C. M., Dioum, E. M., Tuckerman, J. R. et al. (2003) Biochemistry 42, 7701–7708. Gong, W., Hao, B., Chan, M. K. (2000) Biochemistry 39, 3955–3962. Gonzalez, G., Gilles-Gonzalez, M. A., Rybak-Akimova, E. V. et al. (1998) Biochemistry 37, 10188–10194. Miyatake, H., Mukai, M., Adachi, S. I. et al. (1999) J. Biol. Chem. 274, 23176–23184. Balland, V., Bouzhir-Sima, L., Anxolabéhère-Mallart, E. et al. (2006) Biochemistry 45, 2072–2084. Blankman, J.I., Shahzad, N., Miller, C.J., Guiles, R.D. (2000) Biochemistry 39, 14806–14812. Rodgers, K.R., Tang, L., Lukat-Rodgers, G.S., Wengenack, N.L. (2001) Biochemistry 40, 12932–12942. Rodgers, K.R., Lukat-Rodgers, G.S., Tang, L. (1999) J. Am. Chem. Soc. 121, 11241–11242. Liebl, U., Bouzhir-Sima, L., Négrerie, M. et al. (2002) Proc. Natl. Acad. Sci. USA 99, 12771–12776. Tuckerman, J.R., Gonzalez, G., Dioum, E.M., Gilles-Gonzalez, M.A. (2002) Biochemistry 41, 6170–6177. Akimoto, S., Tanaka, A., Nakamura, K., Shiro, Y. et al. (2003) Biochem. Biophys. Res. Comm. 304, 136–142. Spiro, T.G., Zgierski, M.Z., Kozlowski, P.M. (2001) Coord. Chem. Rev. 219–221, 923–936. Lukat-Rodgers, G.S. and Rodgers, K.R. (1997) Biochemistry 36, 4178–4187. Tomita, T., Gonzalez, G., Chang, A.L., Ikeda-Saito, M. et al. (2002) Biochemistry 41, 4819–4826. Miyatake, H., Kanai, M., Adachi, S.I. et al. (1999) Acta Cryst. D55, 1215–1218. Key, J. and Moffat, K. (2005) Biochemistry 44, 4627–4635. Pellequer, J.L., Brudler, R., Getzoff, E.D. (1999) Curr. Biol. 9, R416–R418. Vreede, J., van der Horst, M.A., Hellingwerf, K.J. et al. (2003) J. Biol. Chem. 278, 18434–18439. Wyllie, G.R.A. and Scheidt, W.R. (2002) Chem. Rev. 102, 1067–1089. Satterlee, J.D., Suquet, C.M., Savenkova, M.I., Lian, C. (2003) Paramagnetic Resonance in Metallobiomolecules 858, 244–257. Bertolucci, C., Ming, L.J., Gonzalez, G., Gilles-Gonzalez, M.A. (1996) Chem. Biol. 3, 561–566. Tanaka, A., Nakamura, H., Shiro, Y., Fuji, H. (2006) Biochemistry 45, 2515–2523. Braslavsky, S.E. and Heibel, G.E. (1992) Chem. Rev. 92, 1381–1410. Wyllie, G.R.A. and Rodgers, K.R. unpublished results. Westrick, J.A., Peters, K.S., Ropp, J.D., Sligar, S.G. (1990) Biochemistry 29, 6741–6746.

[47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75]

596 [76] [77] [78] [79]

K.R. Rodgers et al.

Westrick, J.A. and Peters, K.S. (1990) Biophys. Chem. 37, 73–79. Gilles-Gonzalez, M.A. and Gonzalez, G.J. (2005) J. Inorg. Biochem. 99, 1–22. Sousa, E.H.S., Gonzalez, G., Gilles-Gonzalez, M.A. (2005) Biochemistry 44, 15359–15365. Balland, V., Bouzhir-Sima, L., Kiger, L., Marden, M.C., Vos, M.H., Liebl, U., Mattioli, T.A. (2005) J. Biol.Chem. 280, 15279–15288. [80] Kurokawa, H., Lee, D.-S., Watanabe, M. et al. (2004) J. Biol. Chem. 279, 20186–20193. [81] Takahashi, S., Ishikawa, K., Takeuchi, N. et al. (1995) J. Am. Chem. Soc. 117, 6002–6006. [82] Pandini, A. and Bonati, L. (2005) Protein Eng. Des. Sel. 18, 127–137.

Index

A:B heme ratio, 389, 392, 393, 406, 407 Alkylnitroso compounds, 450 Allosteric effector, 43, 223, 225, 286, 587 Angeli’s salt, 72, 429, 431–2, 434, 446–7 Antiferromagnetic, 331, 340, 344, 378, 417 Antihemostatic, 382, 385, 387 Asp-Asp pair, 230–2 Auto-inhibition, 467, 475, 479, 481, 495 Autophosphorylation, 26, 28, 52, 189, 567–8, 587, 589 Autoreduction, 404, 405, 419, 487–9, 492, 494 Autoxidation, 68, 69, 74, 87, 197, 231, 271–2, 358, 363, 380, 570, 571, 588–9 AxPDEA1, 18, 30–3, 38, 40, 41, 100, 122, 186, 187, 566 BAY 41–2272, 531, 544, 546 BAY 41–8543, 544 BAY-58–2667, 545 BAY activators, 535, 544 Bedbug, 397 Bending: FeCO, 4, 96, 108, 140, 142 propionate, 557, 560 vinyl, 557, 560 Bent nitrosyl, 70, 454 -barrel, 385, 388, 394, 396, 412 -sandwich, 398 bHLH-PAS, 35–7, 42, 43, 56 Bioinorganic chemistry, 429–58 B. japonicum, 25, 28, 40, 58, 345, 347, 348, 567, 568, 587, 588 BjFixL, 24–30, 38–43, 58, 567–9, 572–4, 578, 582–4, 588, 589, 591–3 BjFixLH, 24, 27, 29–30, 38–43, 100, 112, 568–71, 574–9, 581–5, 588, 590–2 Bradyrhizobium japonicum, 24, 564, 567 Calmodulin, 379, 396–7, 466, 530 cAMP, 31–4, 46, 107, 187, 434, 498, 500–7, 513–14, 519, 526

CAMP receptor protein, 45, 498, 500, 526 carbon monoxide, 18, 28, 36, 37, 46, 95, 175, 192, 229, 230, 279, 354, 356, 435, 498, 540, 542, 584 C2 Cap: distal compression, 107–9 FeC/CO backbonding correlations, 98–104 c-di-GMP, 31–5 Cgb, 100, 235, 237, 252, 259–60, 295 CGMP, 20, 31, 54, 67, 68, 147, 187, 379, 524, 526, 529–30, 535–6, 541, 551, 556–60 Chagas’ disease, 382–3 ChCooA, 47–8 Chemotaxis, 20, 49–51, 55, 58, 176, 185, 191, 199, 525–6, 533, 537 Chloroperoxidase (CPO), 96, 99, 104–5, 115, 364, 366, 466, 467, 474, 477 proximal H-bonding, 104–5 FeC and CO frequencies, 105, 107, 109 FeN and NO frequencies of Fe(III)NO adduct, 115 FeN frequency of Fe(II)NO adduct, 109 Cimex lectularius, 84, 397 Citrulline, 396, 466 Closed loop structure or conformation, 406, 418 CN− , 27–9, 39, 41, 48, 181, 412, 413, 415, 417, 469, 518, 565, 588–9, 591 CO, 95–118 CO recombination, 252, 581, 584 CO sensitizer, 560 Coiled-coil motif, 507, 511 Compound II formation in P450BM-3 , 364, 366, 368, 369 Compound II in, catalysis, 371 Computational calculations on P450nor catalysis, 372–4 Conformation, 4, 23–4, 41–3, 45, 59, 419, 508–10, 530

598 Conformational equilibrium, 41 Continuous photolysis, 73 CooA, 6, 13, 14, 18, 19, 22, 24, 45–8, 56, 58, 100, 103, 105, 107, 186, 498–521, 526, 540 distal polar interactions, 104 proximal H-bonding, 105 FeC/CO backbonding correlations, 98–104 Cooperativity, 24, 207, 210, 224–7, 237, 244, 274, 503, 519–20, 587 Copper-Dimethylanthrancenyl-cyclam (DAC) complex, 85 Cyanide, 28, 41, 47–8, 197, 248, 250–1, 253, 270, 291, 293, 299, 314, 337, 339, 385, 391–2, 406, 415–16, 418, 434, 465, 575–6, 588–9, 591 Cyclase Assays, 529 Cyclic guanylyl monophosphate (cGMP), 68 cyclic di-GMP, 18, 189 CYP, 55 family, 358 Cytochrome c peroxidase (CCP), 96, 105, 235, 254 distal polar interactions, 104 proximal H-bonding, 105 FeC/CO backbonding correlations, 98–104 FeN/NO backbonding correlations for Fe(II)NO adduct, 113, 115 Cytochrome, 220, 257, 328, 379, 384, 466, 467, 476, 567 Cytochrome c´ , 327, 534, 558 Cytochrome oxidase (cyt ox) distal polar interactions, 113, 115 Fe-C bond compression, 139 FeC/CO backbonding correlations, 98–104 FeN/NO backbonding correlations for Fe(II)NO adducts, 113, 115 Cytochrome P450cam (cyt P450cam) distal polar interactions, 104 proximal H-bonding, 104 FeC/CO backbonding correlations, 98–104 FeN and NO frequencies of Fe(II)NO adduct, 103, 108 FeN/NO backbonding correlations of Fe(III)NO adducts, 113, 115 FeO/OO backbonding correlations, 117–18

Index Cytoglobin/Cygb, 204, 293 FeC/CO backbonding correlations, 101, 111 FeN/NO backbonding correlations, 105, 111 DevR (DosR), 57, 524 DevS (DosS), 18, 19, 57–8, 524, 526 DFT calculations, 4, 5, 70, 77, 96, 104, 106, 114, 116, 132, 135–7, 141–2, 148, 153, 155, 157, 159–60, 334, 342, 417, 534 Dielectric constant, 407, 409 Dimerization, 43–5, 51–2, 54, 272, 276, 342, 431–3, 481, 530, 543, 571 Disulfide bond, 198, 207, 210, 465, 489, 490, 492–4 DOS, 32–5 DosT, 18, 19, 58, 524 DR, 39, 41–3, 48, 56 E75, 18, 19, 56–7 EAL domain, 18, 31–3, 35, 187, 189 EcDosH, 24, 39, 41–2, 44, 100, 112 E. coli Dos, 31, 40 Electrocatalytic, 441, 444 Electrochemical, 401, 404, 440–4, 457, 508 Electrochemical/electrochemistry, 312, 405 Electrochemically, 440, 442, 445, 456–7 Electronic structure, 148–9, 157–60, 162, 166, 373, 481 Endothelial cells, 68, 105, 435, 466 ENOS, 66, 68, 105, 379, 400, 466, 472, 473, 475, 480, 489 EPR, 74, 78, 114, 147–9, 151, 153–61, 168, 207, 235, 239, 248, 276, 307, 335, 338–9, 341, 344, 355, 366, 370, 381, 384, 386, 397, 399–401, 403, 416–18, 430, 437, 446, 542, 565, 573, 589 EPR spectra of horseradish peroxidase (HRP), 156 EPR spectrum of Hb NO (Hb = Hemoglobin), 151 EPR spectrum of P450cam, 160 EPR spectroscopy, 161, 370, 399, 573, 589 ESI MS/MS, 490 EXAFS, 114, 357, 367, 449, 564–5, 572, 579 {FeNO}6 , 75, 378, 381, 403, 412, 415, 417, 448, 565, 573 {FeNO}7 , 70, 381, 386, 399, 403, 448, 565, 570, 573 Fe(OEP)(NO2), 78

Index Fe(Por)(NO) further reactions, 79 photochemical studies, 73 Ferric heme, 27, 87, 255, 276, 277, 291, 305, 309, 314, 315, 337, 340, 342, 343, 360, 373, 374, 400, 430, 436, 444, 445, 469, 475, 479, 480, 486, 487, 489, 494, 570, 571, 572, 591 Ferric heme nitrosyls, 160 Ferrous heme, 14, 33, 45, 53, 68, 87, 147–9, 151, 154, 157, 159–62, 164, 165–6, 168, 242, 250, 275–6, 298, 303, 305, 309–10, 314–15, 340–1, 403, 435–6, 456, 475, 487, 494, 534, 547–8, 558, 569, 571 Ferrous heme nitrosyls, 148, 149, 151, 153, 154, 157, 159, 160, 161–168 Fe(TPP)(NO), 70, 72, 73, 74, 75, 76, 77, 441 Fe(TPP)(NO2), 76 Fe(TPP)(NO2)(NO), 73–9 Fe(TPP)(NO3), 77, 79 FixJ, 18, 35, 36, 539, 561 FixL, 26, 27, 39–41, 564–9 FixLJ complex, 26, 564, 568, 587 Flavohemoglobin, 198–9, 235–7, 304, 327, 348 Fluorescence, 26, 43, 126–7, 518, 530, 542, 580, 581 FNR, 498, 501, 515–16, 566 Folding, 235, 239, 257–9, 379, 388 FTIR, 6, 8, 74, 193, 235, 239, 339, 466, 473, 552, 565, 581 Fusarium oxysporum, 115, 354–5, 364, 366 GAF domain, 57–8, 187, 189, 348 GDP, 51, 199–200, 214, 556–7 Gel filtration, 43, 191, 482–3 GTP, 51, 54, 189, 199–200, 529, 541, 543, 546, 551–2, 555–60 analogues, 556 GGDEF, 18, 25, 31–3, 35, 51, 180, 182, 187, 189 Globin-coupled sensor, 18, 24, 51, 176, 195, 199, 566 Gluconoacetobacter xylinum, 30 G. xylinum, 30, 31, 35, 40 Guanylyl cyclase, 18, 22, 54, 55, 67, 363, 379, 380, 403, 413, 524, 525–6, 533 H-bond, 101–3, 105–7, 112–15, 117, 193, 239, 242–5, 247–51, 253–6, 260, 409–10, 467, 488, 572, 575–7, 581–3, 591

599 H-bond(ing), 96, 101, 103, 104–105, 193 H-NOX domains, 403 H/D exchange rate, 394, 412 HemAT, 48–51, 192–3 Heme deformation, 474–5, 478, 480, 555 Heme distortion, 475, 477, 478, 479–81, 531, 537 Heme-HNOB motif, 18, 19, 24, 55, 525, 526 Heme ligand switch, 506, 514–16 Heme orientation, 389, 392, 407 Heme oxygenase, 37, 560 heme-PAS, 24–8, 30–2, 35–44, 58, 59, 187, 189, 199 Hemoglobin: EPR parameters of NO complex, 151–2, 157 FeC/CO backbonding correlations, 98–104 FeO/OO backbonding correlations, 117–18 Heterolytic cleavage, 104, 487 Heterotetramer, 569, 587 Heterotropic effect, 223–5 Hill analysis, 46, 220 Histamine, 381–2, 384–5, 388, 391–2, 398, 401, 405, 406, 413, 418 Histidine kinase, 20, 55, 57, 189, 567 Hmp, 235, 238, 247, 254–7, 260, 294, 299, 348 HMR-1766, 544 HNO-adduct characterization: electrochemical, 401, 457 IR, 454 NMR, 38, 43, 451 Raman, 107, 125, 138, 193, 449 vibrational, vNO, 110, 573 X-ray absorbance, XAFS, EXAFS, XANES, 449 HNOB-coupled sensor, 24 HNOB domain, 526, 530, 536 HNO donors: Angeli’s salt, AS, 72, 431, 434, 447 methylesulfonylhydroxylamine acid, MSHA, 430, 432, 446 Piloty’s acid, PA, 431 HNOX, 540 H-NOX domain, 18, 413 HO-2, 37 Homolytic cleavage, 84, 399, 488 Homotropic effect, 220–3 Horse heart myoglobin nitrosyl complex, hh-Mb(NO), 70

600 Horseradish peroxidase (HRP) distal polar interactions, 104 proximal H-bonding, 106 FeC/CO backbonding correlations, 98–104 FeN and NO frequencies of Fe(III)NO adduct, 115–16 HRI, 52–3, 59, 200, 540 Hydride transfer in P450nor , 354–7 Hydrophobic triad, 588 Hypoxia, 30, 35, 57, 199, 207, 209, 212, 213, 214, 215, 269, 282, 285, 287, 294–5, 311, 536 Imidazole, 7, 12, 27, 28, 38, 46–7, 98, 101, 104, 105–7, 114, 117, 124–5, 139–41, 154, 158, 197, 239, 245, 253, 307–8, 311, 315, 355, 388–9, 391–3, 398–9, 401, 405–6, 418–19, 518–20, 544, 546, 548, 552, 565, 569, 576, 588–9 Implications for 450nor catalysis, 372 Inducible nitric oxide synthase iNOS, 66, 68 IR, 4, 71, 74–5, 77–9, 95–7, 131, 140, 370, 403, 404, 449, 453–4, 542, 552, 565, 580 Iron porphyrin, 67, 80, 133, 371, 418–19, 441, 531 Isatin, 544 Isotope effect in P450nor -NADH/NADD reaction, 370 Kissing bug, 380 Rhodnius prolixus, 378 Triatoma infestans, 383 Laser flash photolysis, 72, 76, 192, 245, 249, 343, 582 LBD, 18, 19, 20, 24, 56 Leucine zipper, 507, 513–14 Ligand nitrosation, 84–5 Ligand rebinding, 571, 580–2, 585 Ligand specificity, 520 Linkage isomerization, 73, 77 Lipocalin, 384–5, 388 Liposomes, 336, 343, 389 Loose-dimer, 482, 483, 485, 490, 492, 493, 494 Macrophage, 68, 105, 198, 245, 294, 297, 330, 349, 379, 466, 542 MALDI-TOF, 483, 569 Mass spectra, 442, 489, 569

Index Mass spectrometry, 335, 433, 492 MCD spectra of [Fe(TPP)(NO)], 163 MCD spectroscopy, 147–68 Mechanism of sGC Activation, 55, 534–5, 537, 560 Methylsulfonylhydroxylamine acid, 430, 432 Model complexes, 70, 81, 87, 148, 149, 154, 160 Monomerization, 43–4, 481, 485, 486, 489, 490, 493, 494 Mössbauer, 75, 125–7, 130, 370, 401, 403, 416–17, 437 MSHA, 430, 432, 446 M. tuberculosis, 19, 57, 58, 193, 236, 239, 250, 252, 259, 300, 302, 304, 308 Mutation, 9, 12, 14, 228, 407 Mycobacterium tuberculosis DevR, 18, 67 Myoglobin, 3, 5, 12, 18, 32, 49, 68, 70, 78, 81, 96, 98, 101, 138, 140, 147–9, 194, 195, 198, 203–6, 213, 273, 274, 277, 302, 380, 405, 429, 430, 435, 440–2, 445, 533, 571 crystal structures of Fe(II)NO adducts, 114 distal polar interactions, 96–7, 101–104, 111–15 FeC/CO backbonding correlations, 98–104 FeN and NO frequencies of Fe(III)NO adducts, 115 FeN/NO backbonding correlations for Fe(II)NO adducts, 111–12, 114–15 FeN/NO backbonding correlations for Fe(III)NO adducts, 115 NADH-binding in P450nor , 358, 367–9 N-hydroxyl-arginine, 466 N2 O formation, 342, 357, 443 Neuoroglobin: proximal H-bonding, 104 FeC/CO backbonding correlations, 98–104 FeN/NO backbonding correlations, 98–104 NiR, 436, 437, 440, 441 NO, 66–87, 95–118 Nitrate reduction, 355, 359 Nitric oxide, 331, 335, 343, 344, 345, 355, 363 Nitric oxide dioxygenase, 245, 252, 254, 255, 256, 257, 260, 290–319 Nitric oxide photolabilization, 73

Index Nitric oxide reductase, 331, 335, 343, 344, 354, 438–40 Nitric oxide reduction, 346, 438–40 Nitric oxide synthase (NOS) FeC/CO backbonding correlations, 101–105 FeO/OO backbonding correlations, 117–18 Nitrite catalysis, 82 Nitrite-heme interaction, 78 Nitrite in biology, 67 Nitrite reductase, 78, 277, 283, 284, 293, 327, 328, 334, 335, 436–8 Nitrite reduction, 277, 281, 283, 293, 317, 346, 358, 370, 437 Nitro nitrosyl complex, 73, 74, 75, 77 Nitrophorin, 67, 389, 392, 397, 404, 405, 406, 407, 412, 417, 480, 481, 490 Nitroso compounds, 450 Nitrosyl nitrato complex, 79 3–nitrotyrosine detection by Western blotting, 364 3–nitrotyrosine quantification by HPLC anaylsis, 365 3–nitrotyrosine specific antibodies, 364 NMR, 38, 43, 74, 79, 157, 239, 249, 250, 299, 388, 389, 391, 393, 394, 398–401, 416, 417, 421–3 NNOS, 475, 477, 105, 215, 466, 473, 478, 479, 488 (NO), 71, 453, 455, 457, 551 NO, 66–87, 95–117 NO− complex in P450nor , 359–63 NO concentrations in blood plasma, 68 NO sensing, 55, 299, 488 NO synthase, 215, 293, 297, 298, 363, 379, 380, 396, 397, 536, 541 NoR, 444, 458 NorB, 327, 328, 330, 331, 334, 335, 336, 337, 338, 341, 348, 349, 439 NorBC, 438, 439 Normal-coordinate structural decomposition, 479 NOS, 67, 104, 105, 118, 147, 282, 379, 397, 400, 415, 466, 467, 469, 471, 475, 479, 481, 488, 525 NPAS2, 35–8, 42, 43, 540, 560 nuclear receptor, 18, 20, 56 O2 , 95–118 Octaethylporphyrin: EPR parameters of NO complex, 152–3

601 ODQ, 530, 531, 535, 544, 561 Oxygen atom transfer, 77 Oxygen sensor, 32, 55, 212, 285, 348, 530, 589 Oxygenase domain, 467, 471, 477 P420 form, 354 P450, 105, 111, 160, 161, 254, 293, 313, 341, 354, 355, 358, 359, 360, 363, 364, 365, 366, 367, 368, 370, 379, 399, 436, 439, 444, 467, 474, 488 P450–catalyzed protein tyrosine nitration, 363–6 P450nor , 354–7 characterization, 355, 358–9, 360–3, 368 gene structure, 358–9 isolation, 355–8 reduction mechanism, 359–63 reduction, 444 nm intermediate, 360, 368, 369, 439 spectral properties, 366–70 PAS, 24–43 PAS domain, 39 PDEA1, 540 Peroxynitrato-metHb complex, 86 Peroxynitrite reactions with proteins, 363–6 Photoacoustic, 30, 581 Photolysis, 72, 73, 76, 77, 78, 80, 81, 198, 242, 249, 252, 343, 360, 431, 433, 441, 564, 572, 573, 580, 581, 582, 583, 584 -bonding, 4, 254, 429, 451 Picket fence porphyrin, 78, 79 Piloty’s acid, 431 Platelet aggregation, 380, 387, 388, 389, 406, 525 Porphyrin, 67, 70, 72, 74, 78, 79, 80, 82, 104, 106, 108, 127, 128, 129, 130–4, 137, 140, 142, 149, 154, 159, 162, 164, 165–7, 245, 255, 258, 339, 378, 404, 406, 415, 416, 417, 431, 441, 450, 469, 479, 481, 534, 560, 575, 577, 756 Prostacyclin synthase nitration, 363–6 Proteins, 24–5, 32–3, 45–9, 55, 56, 80–87, 95–118, 203–204 Proton delivery in P450nor , 367–70 Protoporphyrin IX dianion, 68 Proximal histidine, 30, 38, 39, 47, 48, 53, 106, 107, 109, 190, 191, 195, 196, 220, 239, 243, 245, 248, 251, 253, 255, 311, 339, 340, 343, 379, 392, 403, 418, 531, 552, 559, 569, 575, 589

602 Quantum-chemical calculations, 148 Quantum yield, 72, 584 R206A, 28, 29, 39, 41 R220A, 28–30, 39, 41 R state, 84, 221–2, 225, 227, 229–30 Raman, 20, 30, 42, 45, 46, 49, 465, 470, 471, 472, 475, 476, 477, 478, 479, 480, 486 Reactive oxygen species/ROS, 52, 214, 379, 535 Reciprocating engine model, 586–7 Redox potential, 22, 32. 47, 274, 277, 285, 481, 499, 501, 571, 587, 591 Redox sensing, 500, 510 Reductase domain, 190, 237, 257, 305, 313, 466, 471, 477, 494 Reductive nitrosylation, 72, 82–5 Resonance Raman spectroscopy, 20, 107, 125, 138, 140, 160, 164, 192, 239, 240, 241, 242, 243, 244, 245, 247, 248, 249, 250, 251, 253, 255, 256, 257, 403, 411, 441, 470, 486, 519, 547, 564, 585, 591 Response regulator, 18, 57 Rhizobia, 24, 2 RmFixL, 21, 24–8, 39–40 R state, 84, 221–2, 225, 227, 229–30 RrCooA, 47–8 Response regulator FixJ, 189, 568 Rhodospirillum rubrum, 22, 45 Ruffled, ruffling (of heme), 392, 396, 399, 410, 411, 412–20 Ruthenium models, 75, 85, 454 S-3448, 544 Salt-bridge network in P450nor , 41–2, 369 Scapharca inaequivalvis, 487 Second order ligand recombination, 584 Self-inhibition, 475 Sensor kinase, 18, 568, 573, 584, 585 Sensors, 18–60 SGC, 22, 54, 67, 403, 413, 524 Single domain hemoglobin, 198 Sinorhizobium meliloti, 21, 564, 567 SiR, 437 Site-directed mutagenesis in P450nor , 368 Site-Directed Mutagenesis, 106, 176, 193, 196, 368, 587–93 S. meliloti, 24, 28, 40, 567, 587, 588 SN 1, 487 S-nitrosothiols, 67, 70, 85, 286 S-nitrosylation, 405, 489

Index SNO, 398, 399, 400, 401, 487 Soluble guanylate cyclase, sGC, 22, 54, 67, 147, 209, 298, 403, 413, 488, 524–37, 540–61 activation, 557–60 model, 512 activators, 544–5 catalytic domain, 541, 543 model, 544 dimerization domain, 543 heme binding domain, 543 high spin, 542, 564, 574 low spin, 542, 564 native, 547–8 Raman spectra, 449, 475, 477, 479, 486 sGC-CO, 551–5 sGC-NO, 549–51 structure, 542–4 UV-Vis spectra, 546 SONO, 540, 542 Soret, 53, 162, 163, 167, 176, 343, 356, 357, 446, 469, 470, 481, 483, 484, 485, 486, 546, 552, 558, 572, 581, 582 Spectroscopic properties, 542, 558, 589 Spin density, 157, 158, 159, 162, 168 Stopped-flow kinetics with P450nor , 359–63 Stretching: C-O vs. Fe-CO, 551–4 C-O, 473–4 C=C (vinyl), 557 Fe-CO, 311, 557–60 Fe-NO, 138, 475, 477, 486, 550 N-O, 75, 357, 451 Structure alignment of P450nor with other P450s, 366–70 Swapped, 489, 493 Synergistic activation, 559 model, 559 Tar4, 18, 524, 526, 530–3, 542 Tetrahydrobiopterin, 379, 396, 465, 466 Tetraphenylporphyrin (TPP): EPR parameters of NO complex, 152–3 FeC/CO backbonding correlations for five-coordinate Fe(II)CO adduct, 98, 101, 104 FeC/CO backbonding correlations for six-coordinate Fe(II)CO adduct, 98, 101 FeN/NO backbonding correlations for five-coordinate Fe(II)NO adduct, 109, 111

Index FeN/NO backbonding correlations for six-coordinate Fe(II)NO adduct, 110–11 FeO/OO backbonding correlations for five-coordinate Fe(II)O2 adduct, 117–18 FeO/OO backbonding correlations for six-coordinate Fe(II)O2 adduct, 117–18 Tight-dimer, 488, 494 Time-resolved, 87, 126, 341, 343, 564, 567, 581, 583 TMSPP, 81, 82 TOB-1, 34–5 TPP2− , 70 TPPS, 68, 80, 81, 82 Transcriptional activation, 46, 346, 515, 568, 587 Transcriptional regulation, 567 Translabilizing effect, 110 TrCtb, 237, 239, 252, 253, 254, 255, 258, 259, 260 TrHbC, 239, 248, 249, 250, 256, 260 TrHbN, 237, 239, 240, 241, 242, 243, 244, 245, 248, 252, 253, 256, 258, 259, 260, 294, 299, 300, 302, 304, 308, 313 TrHbO, 237, 239, 250, 251, 252, 253, 258, 259, 260, 302 TrHbP, 239, 245, 247, 248, 260 TrHbS, 239, 247, 249, 250, 253, 256, 259, 260 Truncated hemoglobin, 190, 237, 239

603 Trypanosoma cruzi, 297 T state, 13, 84, 221, 225, 227, 228, 230, 231, 273, 277, 279, 281 TtTar4H, 114, 542, 548, 552 sGC-CO, 546, 551–5 sGC-NO, 549–51 Ultrafast, 109, 573, 581 Unswapped, 489, 493 Urea, 482, 483, 485, 486, 487, 489, 490 Vgb, 239, 256, 257, 258, 259, 260 Voltammetry, 443 W3110, 34 X-ray Absorption Fine-Structure, 430 X-Ray Absorption Near Edge (Fe K-edge) Structure, 430 X-ray structure of HNOB domain, 530–3 X-ray structure of NADH-P450nor -complex, 358, 362, 369, 370 X-ray structure of P450nor , 366–70 XAFS, 114, 357, 448, 449, 564, 572, 579 XAFS/exafs, 114, 357, 448, 449, 564, 572, 579 XANES, 448 YC-1, 530, 534, 535, 544, 546, 551, 552, 555, 556, 557, 559, 560 YC-1 structure, 551–555

This page intentionally left blank

Mb

deoxyMb(H2O) Kentry

KH2O H2O

Mb

MbXs

X Kbond Kstabilization MbX

Plate 1. Mechanism of ligand binding to mammalian myoglobin. The four-step mechanism originally described by Olson and Phillips [2] was based on the crystal structures of sperm whale deoxyMb (2mgl, [24]), the photoproduct of low temperature MbCO crystals (1ABS, [48,49]), and room temperature sperm whale MbCO (2mgk, [24]). Similar photoproduct structures for Mb•••X have now been observed at room temperature in time-resolved X-ray crystallography studies ([50,51,61–63]). The lower states, deoxymyoglobin with an “empty” distal pocket, Mb, and liganded myoglobin without any stabilization of the bound ligand, MbX cannot been “seen” by crystallography for the wild-type protein, but can be simulated by mutation of the distal histidine to an apolar amino acid that cannot stabilize either internal water or bound ligands [24]. (see Fig. 2, p. 5.)

LeuB10

TyrB10 HisE7

SW MbO2

SW Mb

LbaO2

ValE11

LeuE11 ValG8

IIeG8

HisE7

eclipsed

HisF8

HisF8

(A)

Cerebratulus HbO2 GlnE7

HisF8

Ascaris Hb

Cerebratulus Hb

staggered

staggered

TyrB10

IIeE11

PheG8

staggered

(B)

TyrB10

Ascaris HbO2

Soybean Lba

ThrE11

GlnE7

AlaG8

HisF8

Plate 2. Proximal geometries and electrostatic interactions in four globins. Panel A, Distal pockets of SW MbO2 (2mgm); postulated structure of soybean LbO2 , (1bin, [73]). Domain 1 of Ascaris HbO2 (1ash, [83]); and Cerebratulus HbO2 (1kr7,[84]). Panel B, Proximal His(F8) plane orientations of the globins shown in Panel A. Only SW Mb shows an eclipsed orientation with the edge of the imidazole ring directly beneath two of the pyrrole nitrogens, a conformation that inhibits in-plane movement of the iron atom. (see Fig. 4, p. 10.)

Switching by true signal:

Activity (e.g. kinase)

Activity (e.g. kinase)

Response to alternative signal:

(A) No discrimination. False signal can bind and switch off.

(B) Discrimination at switching step. False signal binds but does not switch off; it also prevents true ligand from binding.

(C) Discrimination at binding step. False signal does not bind. Sensor stays in unliganded, on state.

Plate 3. Switching and discrimination by heme-based sensors. This cartoon illustrates the possible responses to true signal (closed circles) and to alternative ligands (open circles) by a hypothetical sensor that is normally active in the unliganded state. The top of the figure shows switching by the true signal. The three lower panels show the likely responses to alternative ligands: absence of discrimination (A), binding without switching (B), exclusion of the ligand (C). To apply the same figure to a hypothetical sensor that is normally inactive in the unliganded state, reverse all the switches. (see Fig. 2, p. 21.)

RmFixL

Histidine Kinases

BjFixL

PAS

PAS

PAC

MtDos

PAC

PAS

PAS

EcDos

PAS

PDEA1

PAS

PAC

HATPase

HisKA

PAC

PAS

PAC

PAS

HisKA

HATPase

PAC

PAS

PAC

HATPase

DUF2

DUF1

PAC

Phosphodiesterases

bHLH DNA Binding

NPAS2

HLH

PAC

GAF

DUF1

A

B

PAS

PAS

DUF2

PAC

Plate 4. Domain organization of known heme-binding PAS proteins [37,38]. PAS domains are depicted as purple rectangles, and the heme-PAS domains are highlighted with red circles. Note that each protein-histidine kinase or phosphodiesterase contains only one heme-PAS domain. The enzymatic subdomains of the protein-histidine kinases (HisKA/HATPase) are shown in green and those of the cyclic dinucleotide phosphodiesterases (DUF1 and DUF2, corresponding to GGDEF and EAL, respectively) in brown. In NPAS2, an N-terminal basic-helix-loop-helix DNAbinding region (HLH) is followed by two heme-binding PAS domains, PAS-A and PAS-B, and a C-terminal region of unknown function. Domain nomenclatures, symbols, and protein organizations are according to the simple modular architecture research tool (SMART) from the European Molecular Biology Laboratory [57,58]. (see Fig. 4, p. 25.)

(A)

(B) G

G

H

H D212

R220

H214

I

D212

C

A

I

H214 R220

B

B H200 R206

H200 R206

D

E

(C)

BjFixL RmFixL Mt Dos AxPDEA1 EcDos

F alpha helix

proximal

197 191 181 64 74

220 214 201 87 97

D

E F

F

BjFixL RmFixL Mt Dos AxPDEA1 EcDos

C

A

FG loop

0

1

2

3

4

5

6

7

8

9

10

11

12

13

R R L R R

S H E H P

R E K E A

H H H H H

D D D D P

S G R E

Y Y F Y Y

I L I I

S Q N D R

R R R R H

Y Y I N N

R M I R R

T A D E E

T T R T G

S G G G

D E E H K

–2

–1

G beta 0 1

2

3

–1

0

1

2

3

R R R R

I V E E

V V V V L

T S E E Q

G G G F L

K Q V T E

T T R Y K

F F V I I

P P P C W

M M L G T

H K E E R

R R R R K

R K K A K

D D D D D

G G G G G

T S T E S

Distal P K I N A

H R S R R

H beta 4 5 Distal L L I L F

S A S S A

I I I I V

I I M V E

G G V G G

I I T T M

G D D S S

6

7

8

9

10

I V V L L

G G F S S

E E S K K

M M W V V

Q R N Q S

Plate 5. Structural elements implicated in conformational switching by the BjFixL heme-PAS domain, and their occurrence in proven heme-PAS proteins. In unliganded “on-state” BjFixLH (A), the G -2 arginine (Arg 220) on the distal side of the heme forms a hydrogen bond to the heme propionate 7, and the F 9 arginine (Arg 206) on the proximal side interacts with the FG loop [29,51,52]. In liganded “off-state” BjFixLH (B), the G -2 and F 9 arginines switch their hydrogen-bonding interactions to the bound O2 and the heme propionate 6, respectively [29,51]. The structures of the known on-state deoxy and met forms are very similar; likewise, the structures the known off-state oxy and cyanomet forms closely resemble each other. Structures are compared for PDB files 1XJ3 and 1LT0; on versus off, respectively. An alignment is shown (C) for the F-helix and FG-loop sequences of B. japonicum FixL, S. meliloti FixL, Methanobacterium thermoautotrophicum Dos, G. xylinum PDEA1, and E. coli Dos. The absolutely conserved F 3 residue (H200 in BjFixL), or proximal histidine, coordinates the protein to the heme iron. The conserved G -2 arginine (R220 in BjFixL) and usually basic F 9 residue (R206 in BjFixL) strongly influence affinity and regulation [4,5]. The G -2 arginine alternately interacts with the heme propionate 7 in the unliganded on-state or with bound O2 or CN− in the off-state; the F 9 residue (R206 in BjFixL) alternately interacts with the FG-loop in the BjFixL on-state or the heme propionate 6 in the off-state. (see Fig. 6, p. 40.)

(A)

(C)

(B)

D B

trHbN

A

B G

E

E

E11

G

C

B10

H

H

E7

trHbN

F8

E10

F

swMb

Plate 6. The crystal structures of (A) sperm whale oxy-myoglobin (PDB:1MBO) and (B–C) oxy-trHbN from M. tuberculosis (PDB:1IDR). The nomenclatures of the eight helices in (A) are labeled A to H as indicated. The B- and E-helices are labeled in green and yellow, respectively. (see Fig. 1, p. 236.) (A)

(B) Heme Heme

C8

H2O

His85

Tyr95 Lys84

Tyr95

Lys84

FAD

FAD

H2O Lys80

Glu388 Glu394 Tyr190

His85

C8

H2O

Asn80

Glu137

Glu135

Tyr188

(C) Heme H2O

His85

Lys84

Tyr95

H2O Lys80

Glu137

Plate 7. Conserved proximal heme structure in (A) A. eutrophus flavoHb (1CQX) [42], (B) E. coli flavoHb (1GVH) [208], and in (C) the single-domain Vitreoscilla sp. Hb-azide (2VHB) [202] X-ray structures. The respective O, N, C, and Fe atom colors are red, blue, grey, and yellow. The adenine nucleoside portion of FAD is colored pink and the P-atom in phosphate orange. (see Fig. 7, p. 312.)

F,G-loop

B′-helix I-helix Thr243

Ser286

Cys352

Plate 8. Crystal structure of P450nor in its resting ferric state. The residues Ser286 and Thr243 are highlighted in red since these amino acids were subject to numerous mutation experiments to prove their involvement in the proton delivery being essential for the enzyme activity. The proximal ligand of the iron, Cys352, is highlighted in yellow. Moreover, the distal I- and B -helices as well as the F,G-loop are marked. The structure was rendered from the protein database file “1ROM” using the PyMol Molecular Graphics System (version 0.93) from DeLano Scientific LLC. (see Fig. 9, p. 367.) 1 NP1 KCTKNALAQT NP4 ACTKNAIAQT NP2 DCSTNISPKQ NP3 DCSTNISPKK NP7 LPGECSVNVIPKK NP1 NP4 NP2 NP3 NP7 NP1 NP4 NP2 NP3 NP7

71_____ VSELQEESPG VSELQVESLG IGEGKLESSG IGEGKLGSSG TGTGPLESNG

__α1__ 21 _βA__ |--A-B loop--|41__βB___ GFNKDKYFNG DVWYVTDYLD LEPDDVPKRY CAALAAGTAS GFNKDKYFNG DVWYVTDYLD LEPDDVPKRY CAALAAGTAS GLDKAKYFSG .KWYVTHFLD KDP.QVTDQY CSSFTPRESD GLDKAKYFSG .TWYVTHYLD KDP.QVTDPY CSSFTPKESG NLDKAKFFSG .TWYETHYLD MDP.QATEKF CFSFAPRESG ___βE_____91 .KYTANFKKV EKNGNVKVDV .KYTANFKKV DKNGNVKVAV LQYTAKYKTV DKKKAVLKEA VQYTAKYNTV DKKRKEIEPA AKYTAKFNTV DKKGKEIKPA

161 141 _____ α2 __ NKDTNAGDKV KGAVTAASLK FSDFISTKDN NKDAAAGDKV KSAVSAATLE FSKFISTKEN QKDAEPSAKV KSAVTQAGLQ LSQFVGTKDL QKTGEPSATV KNAVAQAGLK LNDFVDTKTL NKNALPNKKI KKALNKVSLV LTKFVVTKDL

__βC____61 ____βD GKLKEALYHY DPKTQDTFYD GKLKEALYHY DPKTQDTFYD GTVKEALYHY NANKKTSFYN GTVKEALYHF NSKKKTSFYN GTVKEALYHF NVDSKVSFYN

___βF_111_ ____βG___ 131 __βH__ TSGNYYTFTV MYADDSSALI HTCLHKGNKD LGDLYAVLNR TAGNYYTFTV MYADDSSALI HTCLHKGNKD LGDLYAVLNR DEKNSYTLTV LEADDSSALV HICLREGSKD LGDLYTVLTH DPKDSYTLTV LEADDSSALV HICLREGPKD LGDLYTVLSH DEKYSYTVTV IEAAKQSALI HICLQEDGKD IGDLYSVLNR ___α3___ KCEYDNVSLK SLLTK NCAYDNDSLK SLLTK GCQYD.DQFT SL~~~ SCTYD.DQFT SM~~~ DCKYD.DKFL SSWQK

IDENTICAL RESIDUES CONSERVATIVE REPLACEMENTS CHARGE MUTANTS BELT AND DISTAL POCKET MUTANTS RESIDUES UNIQUE TO NP7

Plate 9. Sequences of five of the nitrophorins from R. prolixus, arranged to show sequence identities within the two groups; NP7 has greatest homology to NP2 and NP3. Helices and -sheet strands are labeled  and , respectively. Definition of secondary structure is based on the structure of NP4. Residues mutated in each of the nitrophorins by the author’s research group are shown in red and cyan. (see Fig. 3, p. 384.)

Plate 10. Ribbon drawing of the NP4 NO structure. Except for the loops that move on NO binding (loop A-B, residues 31–37, and loop G-H, residues 125–132); the linear NO orientation is in ball-and-stick representation. Reprinted from [144] with permission from Nature. The color version of this figure, and the caption describing the color key, can be found at the end of this volume. (see Fig. 5, p. 386.)

Tex (min) 90°

>1000 500–1000 100–500 50–100 10–50 5–10 <5

Plate 11. The backbone amide H/D exchange rates in H2 O/D2 O (25/75). The rates were obtained at 30 C by dissolving a lyophilized sample of 15 N-labeled apo-NP2 in 100% H2O-based buffer (50 mM phosphate, pH 6.5) followed by a rapid 1:3 dilution into 100% D2 O-based buffer (final protein concentration ∼1 mM, 75% D2 O). Progress of the exchange process was followed by collecting a series of 1 H-15 N SOFAST-HMQC spectra at 600 MHz on a spectrometer equipped with a cryoprobe. The dead time was 2 min, and spectra were taken at 5, 10 and 20 min intervals for the first 2, 4 and 6 h, respectively, and every 40 min thereafter for a total of 30 h. Note that the calculated rates are shown here mapped onto an X-ray structure of holo-NP2. (see Fig. 10, p. 396.)

(A)

(B)

(C)

Val 42 Thr 87 Asn 78

Ile 80

Val 44 Phe 49

Gln 56

Plate 12. (A,B) Two views of the structure of cNP, showing the thiolate coordination of the heme and the -sandwich structure of the protein. (C) Closeup of the surroundings of the heme and cysteine thiolate as viewed from the opposite side of the protein as shown in (A), showing the hydrophobic nature of the NO binding site. Reprinted from [185] with permission from Proc. Natl. Acad. Sci. USA. The color version of this figure, and the caption describing the color key, can be found at the end of this volume. (see Fig. 11, p. 397.)

Diphosphoinositol

Heme

Plate 13. Overlay of the protein backbone structures of cNP and IPP5P [185,210] showing the similarity in protein fold. The protein backbone of cNP is shown in dark grey and that of IPP5P in light grey; the visible atoms of heme, the proximal cysteine and diphosphoinositol are shown as ball and stick. The color version of this figure, and the caption describing the color key, can be found at the end of this volume. (see Fig. 12, p. 398.)

(A)

L-arginine

Tyr367

Glu371 H4B

Trp366 Heme Cys194

Trp457

Trp188

(B)

[393-404]

Heme

[393-404]

C109

C109 Heme

[77-118] C104

“swapped” conformation

[77-118] C104

“unswapped” conformation

Plate 14. (A) The catalytic site of NOS, consisting of the heme, the tetrahydrobiopterin (H4B) cofactor and the l-Arg substrate. The proximal ligand to the heme is a cysteine. The various hydrogen bonds in the active site are indicated by the dotted lines. The structure is that of iNOSoxy and is taken from the protein data bank (PDB code:1NOD) [20]. The catalytic sites of all three isoforms are nearly identical. (B) Crystal structures of the dimeric H4B-bound iNOSoxy . (left) The swapped structure (PDB code: 1QOM) with an intermolecular disulfide bond between the C109 residue from each subunit. (right) The “unswapped” structure (PDB code: 1DF1) with a zinc (shown as the orange sphere) coordinated by C104 and C109 from each subunit. The C104 and C109 residues are labeled with ball-and-stick representation. The important peptide segments of the two subunits located in the dimer interface are shown in yellow and green separately. Those of the yellow subunit and its associate prosthetic heme group are labeled as indicated. (see Fig. 2, p. 468.)

Eukaryotes

Prokaryotes

[ [

HNOB

HNOB

HNOB

HNOBA

HNOBA

MA

Homo sapiens

Guan_cyc

Guan_cyc

Drosophila melanogaster

MA

Thermoanaerobacter tengcongensis

HNOB

Anabaena sp

Plate 15. Common domain organizations of HNOB-coupled sensors. Domain organizations are shown for eukaryotic and prokaryotic sensors, as predicted by Pfam analysis: heme-binding HNOB domain (red); HNOBA domain that associates with HNOB but does not bind heme (green); guanylyl cyclase domain (Guan_cyc, yellow); methyl accepting domain of chemeotaxis receptors (MA, gray/black). (see Fig. 1, p. 526.)

(A)

(B)

W9

N74

Y140 R135

Y131

H102

S133

Plate 16. Structure of the T. tengcongensis. Tar4 HNOB domain bound to O2 [33,34]. Part (A) shows the overall X-ray crystallographic structure of the heme-domain of oxy-TtHNOX (PDB 1U55). Part (B) shows the heme and its interactions, with key residues involved in hydrogen bonding indicated by dashed lines. (see Fig. 2, p. 531.)

Plate 17. Model structure of catalytic domain of sGC. Model is prepared using SETOR from PDB ID 1AWN. Cyan and green represents the -subunit whereas blue and purple represents the -subunit. Structure of GTP is shown as stick whereas Mg2+ is represented as magenta sphere. (see Fig. 2, p. 544.)

Specific Activity: 8000 nmol.min–1.mg prot–1 P3 His

Fe–CO

&

(X)

His

Fe–CO

GTP

YC-1

β

α

α

β

His His

Fe–CO

&

His

X

Fe–CO

β

P3

α

Fe–CO P3

GTP

β

YC-1

α

Specific Activity: 250 nmol.min–1.mg prot–1

His

β

P3

Fe–CO

β

P3

α

α

Plate 18. Schematic illustration of our proposed model for synergistic activation of sGC by CO and YC-1/BAY. X would be His residue other than 105-His. A green triangle and P3 mean YC-1/BAY and triphosphate, respectively. Reproduced from [74]. (see Fig. 14, p. 559.)