Recognition of anions

129 Structure and Bonding Series Editor: D. M. P. Mingos

Editorial Board: P. Day · X. Duan · L. H. Gade · T. J. Meyer G. Parkin · J.-P. Sauvage

Structure and Bonding Series Editor: D. M. P. Mingos Recently Published and Forthcoming Volumes

Contemporary Metal Boron Chemistry I Volume Editors: Marder, T. B., Lin, Z. Vol. 130, 2008 Recognition of Anions Volume Editor: Vilar, R. Vol. 129, 2008

Non-Covalent Multi-Porphyrin Assemblies Synthesis and Properties Volume Editor: Alessio, E. Vol. 121, 2006 Recent Developments in Mercury Science Volume Editor: Atwood, David A. Vol. 120, 2006

Liquid Crystalline Functional Assemblies and Their Supramolecular Structures Volume Editor: Kato, T. Vol. 128, 2008

Layered Double Hydroxides Volume Editors: Duan, X., Evans, D. G. Vol. 119, 2005

Organometallic and Coordination Chemistry of the Actinides Volume Editor: Albrecht-Schmitt, T. E. Vol. 127, 2008

Semiconductor Nanocrystals and Silicate Nanoparticles Volume Editors: Peng, X., Mingos, D. M. P. Vol. 118, 2005

Halogen Bonding Fundamentals and Applications Volume Editors: Metrangolo, P., Resnati, G. Vol. 126, 2008

Magnetic Functions Beyond the Spin-Hamiltonian Volume Editor: Mingos, D. M. P. Vol. 117, 2005

High Energy Density Materials Volume Editor: Klapötke, T. H. Vol. 125, 2007

Intermolecular Forces and Clusters II Volume Editor: Wales, D. J. Vol. 116, 2005

Ferro- and Antiferroelectricity Volume Editors: Dalal, N. S., Bussmann-Holder, A. Vol. 124, 2007

Intermolecular Forces and Clusters I Volume Editor: Wales, D. J. Vol. 115, 2005

Photofunctional Transition Metal Complexes Volume Editor: V. W. W. Yam Vol. 123, 2007 Single-Molecule Magnets and Related Phenomena Volume Editor: Winpenny, R. Vol. 122, 2006

Superconductivity in Complex Systems Volume Editors: Müller, K. A., Bussmann-Holder, A. Vol. 114, 2005 Principles and Applications of Density Functional Theory in Inorganic Chemistry II Volume Editors: Kaltsoyannis, N., McGrady, J. E. Vol. 113, 2004

Recognition of Anions Volume Editor: Ramón Vilar

With contributions by P. Ballester · G. W. Bates · S. R. Bayly · P. D. Beer · S. L. Ewen P. A. Gale · I. Hamachi · J. H. G. Steinke · S. Tamaru · R. Vilar

123

The series Structure and Bonding publishes critical reviews on topics of research concerned with chemical structure and bonding. The scope of the series spans the entire Periodic Table. It focuses attention on new and developing areas of modern structural and theoretical chemistry such as nanostructures, molecular electronics, designed molecular solids, surfaces, metal clusters and supramolecular structures. Physical and spectroscopic techniques used to determine, examine and model structures fall within the purview of Structure and Bonding to the extent that the focus is on the scientific results obtained and not on specialist information concerning the techniques themselves. Issues associated with the development of bonding models and generalizations that illuminate the reactivity pathways and rates of chemical processes are also relevant. As a rule, contributions are specially commissioned. The editors and publishers will, however, always be pleased to receive suggestions and supplementary information. Papers are accepted for Structure and Bonding in English. In references Structure and Bonding is abbeviated Struct Bond and is cited as a journal. Springer WWW home page: springer.com Visit the Struct Bond content at springerlink.com

ISBN 978-3-540-79091-4 DOI 10.1007/978-3-540-79092-1

e-ISBN 978-3-540-79092-1

Structure and Bonding ISSN 0081-5993 Library of Congress Control Number: 2008923960 c 2008 Springer-Verlag Berlin Heidelberg
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable to prosecution under the German Copyright Law. The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Cover design: WMXDesign GmbH, Heidelberg Typesetting and Production: le-tex publishing services oHG, Leipzig Printed on acid-free paper 9876543210 springer.com

Series Editor Prof. D. Michael P. Mingos Principal St. Edmund Hall Oxford OX1 4AR, UK [email protected]

Volume Editor Dr. Ramón Vilar Imperial College London Department of Chemistry South Kensington London, SW7 2AZ, UK [email protected]

Editorial Board Prof. Peter Day

Prof. Thomas J. Meyer

Director and Fullerian Professor of Chemistry The Royal Institution of Great Britain 21 Albermarle Street London W1X 4BS, UK [email protected]

Department of Chemistry Campus Box 3290 Venable and Kenan Laboratories The University of North Carolina and Chapel Hill Chapel Hill, NC 27599-3290, USA [email protected]

Prof. Xue Duan Director State Key Laboratory of Chemical Resource Engineering Beijing University of Chemical Technology 15 Bei San Huan Dong Lu Beijing 100029, P.R. China [email protected]

Prof. Lutz H. Gade Anorganisch-Chemisches Institut Universität Heidelberg Im Neuenheimer Feld 270 69120 Heidelberg, Germany [email protected]

Prof. Gerard Parkin Department of Chemistry (Box 3115) Columbia University 3000 Broadway New York, New York 10027, USA [email protected]

Prof. Jean-Pierre Sauvage Faculté de Chimie Laboratoires de Chimie Organo-Minérale Université Louis Pasteur 4, rue Blaise Pascal 67070 Strasbourg Cedex, France [email protected]

Structure and Bonding Also Available Electronically

For all customers who have a standing order to Structure and Bonding, we offer the electronic version via SpringerLink free of charge. Please contact your librarian who can receive a password or free access to the full articles by registering at: springerlink.com If you do not have a subscription, you can still view the tables of contents of the volumes and the abstract of each article by going to the SpringerLink Homepage, clicking on “Browse by Online Libraries”, then “Chemical Sciences”, and finally choose Structure and Bonding. You will find information about the – – – –

Editorial Board Aims and Scope Instructions for Authors Sample Contribution

at springer.com using the search function. Color figures are published in full color within the electronic version on SpringerLink.

Preface

A large number of biologically relevant species are negatively charged, therefore it is not surprising that nature has developed sophisticated receptors to recognise, detect and transform anions. For example, complex receptors such as phosphate- and sulphate-binding proteins are employed by living cells to selectively recognise these two geometrically analogous anions. In addition to their roles in biological systems, some anions also have important environmental impacts. For example, cyanide, pertechnetate and chromates pose serious health problems if present in water supplies. Because of their important biological roles and potential environmental impact there is great current interest in developing molecular receptors to selectively recognise anions and in doing so be able to sequester, transform or sense them. The six chapters presented in this volume provide an overview of anion recognition and the most recent advances in this fast-growing area of supramolecular chemistry are highlighted. The first chapter by Bates and Gale provides an overview of the coordination of anions by synthetic organic hosts. The different organic functional groups used to bind anions are presented and this provides an introduction to the structural and electronic properties that hosts must have to recognise anionic guests. On the other hand, Bayly and Beer give a detailed account of the use of metal complexes as anion receptors. Besides the important structural features that metals can confer to receptors, their optical and redox properties make them attractive for the development of anion sensors. Metal-based receptors have found particularly interesting applications in the recognition of phosphorylated species of biological interest (e.g. phosphorylated amino acids and peptides). This area is reviewed in depth by Tamaru and Hamachi with particular emphasis on a series of receptors based on zinc(II) centres which have been shown to bind phosphates with very high binding constants in aqueous media. The applications of this type of receptor for the detection of samples of biological interest are also presented. Ballester provides an interesting account of anion · · · π interactions and their impact in host design. Over the past few years there has been mounting evidence that this type of interaction plays an important role in anion recognition. The chapter starts with a detailed overview of the theoretical aspects of anion · · · π interactions which is followed by a discussion of the existing

X

Preface

experimental evidence for this type of interaction. Both, solution and crystallographic studies are analysed showing the potential impact that this type of interaction could have in the design of new anion receptors. The use of anions as templating agents is discussed by Vilar. The chapter starts with a general overview of the area and a discussion of the applications of anion templates in organic and coordination chemistry. The second part of the chapter deals with examples where anions are employed as templates in dynamic combinatorial libraries. This approach promises to provide an efficient route for the synthesis of better and more selective anion receptors. The last chapter by Ewen and Steinke also deals with the use of anions as templates but in this case in the context of molecular imprinted polymers. The first half of the chapter provides an introduction into molecularly imprinted polymers and this is followed by a detailed discussion of examples where anionic species have been used to imprint this class of polymeric materials. The topics discussed in this volume provide an exciting and stimulating overview of the most recent studies within anion recognition and templation. Although the supramolecular chemistry of anions took a long time to develop, it is now a mature area that provides solutions to challenging problems. There is no doubt that its growth will continue yielding more sophisticated and efficient receptors for the recognition of a wide range of negatively charged species. London, February 2008

Ramón Vilar

Contents

An Introduction to Anion Receptors Based on Organic Frameworks G. W. Bates · P. A. Gale . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

Metal-Based Anion Receptor Systems S. R. Bayly · P. D. Beer . . . . . . . . . . . . . . . . . . . . . . . . . . . .

45

Recent Progress of Phosphate Derivatives Recognition Utilizing Artificial Small Molecular Receptors in Aqueous Media S. Tamaru · I. Hamachi . . . . . . . . . . . . . . . . . . . . . . . . . . .

95

Anions and π-Aromatic Systems. Do They Interact Attractively? P. Ballester . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 Anion Templates in Synthesis and Dynamic Combinatorial Libraries R. Vilar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 Molecularly Imprinted Polymers Using Anions as Templates S. L. Ewen · J. H. G. Steinke . . . . . . . . . . . . . . . . . . . . . . . . . 207 Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249

Struct Bond (2008) 129: 1–44 DOI 10.1007/430_2007_069 © Springer-Verlag Berlin Heidelberg Published online: 10 November 2007

An Introduction to Anion Receptors Based on Organic Frameworks Gareth W. Bates · Philip A. Gale (u) School of Chemistry, University of Southampton, Southampton SO17 1BJ, UK [email protected] 1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 2.1 2.2 2.3

Neutral Receptors . . . . . . . . . . . . . . . . . . Acyclic Amide and Sulfonamide-Based Receptors Macrocyclic Amide Receptors . . . . . . . . . . . Urea and Thiourea-Based Receptors . . . . . . . .

. . . .

2 2 6 12

3 3.1 3.2

Aromatic NH Donor Containing Neutral Receptors . . . . . . . . . . . . . Pyrrole-Based Receptors . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carbazole and Indole-Based Receptors . . . . . . . . . . . . . . . . . . . .

21 21 29

4

Hydroxy (OH) Donors in Neutral Receptors . . . . . . . . . . . . . . . . .

32

5 5.1 5.2 5.3

Charged Receptors . . . . . . . . . . . . . . . Imidazolium and Pyridinium-Based Receptors Guanidinium-Based Receptors . . . . . . . . . Ammonium-Containing Receptors . . . . . . .

. . . .

33 34 37 39

6

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

42

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

42

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

1

Abstract This review article provides a broad overview to the area of anion coordination by synthetic organic receptors and includes examples of different functional groups used to bind anions. The first section examines neutral anion receptors containing amide-, sulfonamide-, urea- and thiourea-based receptors. Then aromatics such as pyrrole, carbazole and indole are discussed before concluding the discussion of neutral systems with examples of hydroxy OH donors. A brief overview of charged systems is also provided. Keywords Anion recognition · Complexation · Crystal structures · Hydrogen bonding · Supramolecular chemistry

1 Introduction The development of new anion receptors based on organic frameworks continues to attract considerable research effort [1–4]. A wide variety of systems have been published in the last 15 years with both macrocyclic and acyclic

2

G.W. Bates · P.A. Gale

systems functioning as effective and selective anion receptors. This review is not comprehensive but provides examples of important classes of anion receptor systems based on organic frameworks.

2 Neutral Receptors 2.1 Acyclic Amide and Sulfonamide-Based Receptors Secondary amides are versatile and highly accessible hydrogen bond donors that have been used in numerous synthetic receptors. In the biological arena, there are many examples of proteins that employ amide NH· · ·anion interactions to bind negatively charged guests [5–9]. The first example of a synthetic amide containing receptor, published in 1986 by Pascal and co-workers, was a crytpand-like tris-amide that was shown to interact with fluoride in DMSOd6 [10]. In 1993, Reinhoudt and co-workers described the synthesis and binding properties of a series of tris-amides and tris-sulfonamides based upon the tren skeleton [11]. These receptors proved to be selective for phosphate in acetonitrile solution and demonstrated, arguably for the first time, that anion receptor systems need not be difficult to make but rather that simple organic compounds could function as very effective receptors. Stability constants were calculated by conductivity experiments and showed that receptor 1f bound dihydrogenphosphate with the highest affinity (14 200 M–1 ) in acetonitrile presumably due to the preorganization of the receptor via π–π interactions between the naphthyl groups.

Four years later, in 1997, Crabtree and co-workers reported that simple isophthalamide receptors e.g. 2 can bind anions in organic solution [12]. These receptors, even simpler than Reinhoudt’s tren-based anion binders, were found to bind smaller halides selectively in dichloromethane solution. The X-ray crystal structure of the bromide complex of 2a shows the receptor adopting the syn–syn conformation with the bromide anion coordinated to the A (Fig. 1). The crystal strucamide NH’s with N· · ·Br distances of 3.44 and 3.64 ˚

An Introduction to Anion Receptors Based on Organic Frameworks

3

ture also reveals that the bromide anion is positioned above the plane of the central aryl ring. Solution studies revealed that receptors 2a and 2b have high affinity for halide anions and form complexes with exclusively 1 : 1 host/guest stoichiometry. Stability constants for 2b were determined by 1 H NMR titration studies and found to be 6.1 × 104 M–1 for chloride, 7.1 × 103 M–1 for bromide and 4.6 × 102 M–1 for iodide in dichloromethane-d2 .

Fig. 1 X-ray crystal structure of a bromide complex of 2a

Almost contemporaneously, B.D. Smith and co-workers reported the use of functionalized isophthalamide receptors for the coordination of anions [13]. Smith appended boronate groups to the peripheral aryl groups in order to form interactions between the Lewis acidic boron and the carbonyl oxygens of the amides therefore “pre-organizing” the receptor into the syn–syn conformation (preferable for anion coordination) and presumably increasing the acidity of the NH group. Proton COSY and NOE difference experiments indicated that the receptor did indeed adopt the desired syn–syn conformation in DMSO-d6 . NMR titration experiments in DMSO-d6 at 295 K showed that re-

4

G.W. Bates · P.A. Gale

ceptor 3 bound acetate with a stability constant of 2.1 × 103 M–1 as compared to 1.1 × 102 M–1 found with the non-preorganized receptor 2a and acetate. Recently J.T. Davis, Gale, Quesada and co-workers have shown that appended hydroxy groups on the central aryl ring of an isophthalamide can pre-organize the receptor into the syn–syn conformation and again presumably increase the acidity of the amide NH groups, which increases the receptor’s ability to bind chloride (Fig. 2) [14]. The preorganization occurs due to intramolecular hydrogen bonds between the hydroxy groups and carbonyl oxygen of the amides. Proton NMR titration experiments in CD3 CN at 298 K revealed that receptor 4 bound chloride most strongly with a stability constant of 5230 M–1 , whereas a stability constant of 195 M–1 was obtained for the unfunctionalized cleft 5. Model compound 6 contains methoxy groups in the 4- and 6-positions and functions as a control. In this system the intramolecular hydrogen bonds form between the amide NH groups and the methoxy oxygens and consequently the compound does not interact with chloride (Fig. 2). Most interestingly, it was shown that compound 4 functions as a highly efficient chloride transport agent across EYPC lipid bilayers whilst the analogous isophthalamide 5, and model compound 6 show no transport ability. Compound 4 seems to be the simplest synthetic lipid bilayer transport agent for chloride studied so far. The origin of this ability is currently being investigated.

Thordarson et al. have studied the aggregation of pyromellitamide 7 and its response to anions. It was found that compound 7 aggregates in nonpolar solvent by the formation of one-dimensional intermolecular hydrogenbonding networks. Upon the introduction of anions the aggregation of 7 is disturbed [15]. Proton NMR titration experiments in d6 -acetone at 300 K re-

Fig. 2 X-ray crystal structures of 4 (left) and 6 (right)

An Introduction to Anion Receptors Based on Organic Frameworks

5

vealed that compound 7 bound a range of anions with 2 : 1 anion/receptor stoichiometry. Although compound 7 has two discrete binding sites the anions were found to bind with negative cooperativity with the strength of anion binding to 7 decreasing in the order Cl– < CH3 CO2 – < Br– < NO3 – ≈ I– .

Prohens and co-workers have synthesized compounds 8a and 8b, simple squaramido-based receptors and investigated their ability to coordinate carboxylate anion in competitive solvents [16]. The amide NH groups of the squaramide form a more open cleft (similar to ureas) than the isophthalamides. Receptors 8a and 8b therefore adopt a more suitable geometry for the coordination of bidentate anions, such as carboxylate anions, through two approximately linear hydrogen bonds. Proton NMR titration experiments revealed association constants of 217 M–1 and 1980 M–1 for the binding of acetate by 8a and 8b respectively in DMSO-d6 at 295 K.

A.P. Davis and co-workers have designed a number of acyclic receptors using the steroid cholic acid as a scaffold upon which they appended sulfonamide and carbamate amide groups [17]. The inflexibility of the fused ring system and the axial conformation of the functional groups in the 7α and 12α positions results in the formation of a convergent hydrogen-bonding array, ideal for anion binding. Both the structural rigidity and the shape of the receptor result in the formation of a relatively small, well-defined binding site that shows selectivity for halide anions with particularly high affinities observed for fluoride with receptor 9a (15 400 M–1 association constants were determined by 1 H NMR titration experiments in CDCl3 at 298 K). In the case of compound 9b the association constant for fluoride was too high to be determined however, as-

6

G.W. Bates · P.A. Gale

sociation constants for chloride and bromide were found to be much higher than for compound 9a [17]. 2.2 Macrocyclic Amide Receptors Macrocyclic receptors often possess a higher degree of selectivity than acyclic systems. Hamilton and Choi have described the synthesis and anion binding properties of a family of cyclic triamides 10a and 10b [18]. These C3 symmetric receptors were found to be selective for oxo-anions such as tosylate with association constants of 2.6 × 105 M–1 and 2.1 × 105 M–1 obtained for compounds 10a and 10b respectively in CDCl3 /2% dimethylsulfoxide at 296 K. Hamilton also studied the binding ability of an acyclic analogue (11) and found significantly lower stability constants for the anion complexes formed. For example, in the case of nitrate, a stability constant of 620 M–1 was calculated for compound 11 whereas a stability constant of 4.6 × 105 M–1 (K2 = 2.1 × 103 M–1 ) was found with compound 10a.

An Introduction to Anion Receptors Based on Organic Frameworks

7

Interestingly, the NMR data led the authors to suggest that receptor 10b forms a 2 : 1 host/anion “sandwich” complex at low concentrations of iodide (as evidenced by an initial up-field shift of the NH resonances until ca. 0.5 equivalents of iodide) then switching to a 1 : 1 complex at higher concentrations of iodide (down-field shift of NH resonances after ca. 0.5 equivalents of iodide) (Fig. 3). Titrations with 10a and 10b in CDCl3 /2% DMSO-d6 mixture displayed complex binding behavior, thus titration experiments were conducted in 100% DMSO-d6 at 296 K to simplify the equilibria occurring in solution. All data for receptor 10b were fitted to a 1 : 1 binding model and again the macrocycle was found to be oxo-anion selective with the highest stability constants being found with dihydrogen phosphate and hydrogen sulfate (1.5 × 104 M–1 and 1.7 × 103 M–1 , respectively).

Fig. 3 Changes in the amide 1 H NMR proton resonance of 10b with increasing [I– ] concentrations. Reprinted in part with permission from [18]

Chmielewski and Jurczak have reported a series of extended tetra-amide macrocycles containing two pyridine-2,6-dicarboxamide “caps” linked via short aliphatic chains. The macrocycles possess a well-defined cavity with all the amide groups directed inwards [19, 20]. Proton NMR titrations in DMSO-d6 at 298 K were conducted in order to determine the stability constants of receptors 12a, 12b and 12c with a range of anions, added as their tetrabutylammonium salts. The strongest association constants were obtained for the 20-membered macrocycle 12b with the most significant increase in affinity between the macrocycles being observed in the binding of chloride. Enlargement from the 18-membered macrocycle

8

G.W. Bates · P.A. Gale

12a to the 20-membered macrocycle 12b results in a 30-fold increase in the association constant for chloride (65 M–1 for 12a against 1930 M–1 ) whereas further enlargement to the 24-membered macrocycle 12c results in the reduction of the association constant by two orders of magnitude (1930 M–1 against 18 M–1 ). This suggests that the 20-membered macrocycle 12b has good size complementarity with the chloride anion. Although the 24-membered macrocycle 12c was designed with a large enough cavity to accommodate two oxygen atoms from oxo-anionic guests, the stability constants obtained were the lowest of the receptors tested. This is evidence that the additional flexibility introduced into the macrocycle via the longer aliphatic chain has a detrimental effect on the anion binding ability of the receptor. More recently the same authors have studied the anion binding ability of similar macrocyclic systems based on isophthalamides [21]. The isophthalamide moieties were introduced as previous studies have shown isophthalamide derivatives bind anions more strongly than the analogous pyridine-2,6dicarboxamides [22].

Stability constants were obtained for receptors 13a, 13b and 13c using analogous conditions to those employed for receptors 12a, 12b and 12c. As with the pyridine-2,6-dicarboxamide macrocycles, the stability constants for the isophthalamide macrocycles appear to be influenced by the size and flexibility of the system with the higher constants observed in the 20-membered receptor 13a with notable decreases in the association constants with the 22and 24-membered receptors 13b and 13c. The greatest decreases where observed in the stability constants obtained with the carboxylate anions. In the case of acetate the constants decreased from 3130 M–1 for 13a to 552 M–1 and 205 M–1 for 13b and 13c, respectively. In the case of benzoate a constant of 601 M–1 was calculated for 13a decreasing to 302 M–1 and 82 M–1 for 13b and 13c, respectively. Unexpectedly lower binding constants were obtained for the isophthalamide macrocycles (13a–13c) compared to the pyridine-

An Introduction to Anion Receptors Based on Organic Frameworks

9

2,6-dicarboxamide macrocycles (12a–12c), a result rationalized in terms of the competition between the formation of intramolecular hydrogen bonds (arising from the preferred syn–anti conformation of the isophthalamide in solution) and complexation of the anion (Scheme 1).

Scheme 1 Intramolecular hydrogen-bonding vs. anion binding in compound 13a

Bowman-James and co-workers have designed polyamide cryptand-type systems based on triamines, such as tren (e.g. 14) and trpn (e.g. 15), and shown that they bind anions [23]. The crystal structure of the hydrochloric acid and fluoride complexes of 14 reveal that the anions are encapsulated within the cavity of the amidocryptand and bound to the six-amide NH groups. In contrast the hydrochloric acid structure of the expanded trpnbased amidocryptand 15 shows the encapsulation of two chloride anions within the cryptand, bridged by a water molecule. Each chloride is bound to the water molecule as well as a protonated bridgehead amine and two hydrogen bonds from the amides groups.

Stability constants for 14 and 15 with different anions (added as their tetrabutylammonium salts) were obtained by 1 H NMR titrations in DMSO-d6 . In

10

G.W. Bates · P.A. Gale

both cases, a slow equilibrium was observed in the titrations with fluoride with stability constants >105 M–1 . The expansion of the cavity from receptor 14 to 15 results in a significant change in the binding and selectivity for anions. In the smaller receptor, 14, chloride is bound much more strongly (3000 M–1 ) as compared to 15 (180 M–1 ) whereas the receptor 15 has a much higher affinity for hydrogen sulfate with a stability constant of 2700 M–1 as compared to 68 M–1 for 14. These findings may be due to the size complementarity between the receptors and guests with 14 being an ideal size to encapsulate chloride and 15 being ideal for hydrogen sulfate, as illustrated by the crystal structures (Fig. 4).

Fig. 4 X-ray crystal structures of the chloride complex of 14 and the sulfate complex of 15

The authors have also synthesized 16, a tricyclic cryptand-like receptor, and have studied its ability to bind anions. Proton NMR titration experiments in DMSO-d6 at 23 ◦ C revealed that compound 16 was selective for bifluoride (FHF– ) with an association constant of 5500 M–1 being calculated. Dihydrogenphosphate, azide and acetate were also found to bind to 16 with stability constants of 740 M–1 , 340 M–1 and 100 M–1 , respectively [24].

An Introduction to Anion Receptors Based on Organic Frameworks

11

Kubik and co-workers have developed a series of highly effective anion receptors based upon cyclic peptides. Cyclic hexapeptide receptors such as 17 consist of alternately linked l-proline and 6-aminopicolinic acid subunits [25]. A 1 : 1 binding stoichiometry for 17 and the sodium salt of benzenesulfonate was confirmed by a Job plot but in the case of the halide and sulfate sodium salts 2 : 1 host/guest complexes were found. This was confirmed by electrospray mass spectrometry and in the case of iodide a crystal structure of the 2 : 1 complex was obtained where the iodide was “sandwiched” between two cyclic hexapeptide receptors.

The formation of the 2 : 1 complexes observed in 17 led the authors to design and synthesize compound 18 where two cyclic hexapeptides are covalently linked. The new receptor binds anions in a 1 : 1 stoichiometry in methanol/water mixtures efficiently, with high affinity and selectivity for sulfate being observed. Both 1 H NMR titrations and ITC experiments were conducted in 50% methanol/water at 298 K and stability constants (log Ka ) of

12

G.W. Bates · P.A. Gale

5.54 and 4.55 ± 0.23 were found by 1 H NMR and ITC respectively for 18 with sulfate (added as Na2 SO4 ) [26]. 2.3 Urea and Thiourea-Based Receptors Ureas and thioureas possess two parallel NH hydrogen bond donor groups and have been shown in a wide variety of receptors to function as highly efficient binding sites for “Y-shaped” anions such as carboxylates. Thioureas are more acidic than analogous ureas and on this basis might be expected to form stronger complexes with anions. However, other effects can often mask or reverse this expected trend. There have been a number of reports of anions triggering the deprotonation of neutral NH groups in anion receptor systems. This is often due, in the case of fluoride, to the formation of the stable HF2 – anion driving the deprotonation process [27–30]. Fabbrizzi and co-workers have shown that this process can occur in urea systems containing electron-withdrawing groups. The interactions between a number of anions and the simple 1,2bis(4-nitrophenyl) urea 19 were investigated. Oxo-anions were found to bind to the receptor with a 1 : 1 host/guest stoichiometry with the strength of the interaction depending on the partial negative charge located on each oxygen atom of the anion [31].

Stability constants were determined for compound 19 by UV-Vis spectrophotometric titrations in acetonitrile at 25 ◦ C and revealed that the association constants increased with the increasing basicity of the anion (CH3 COO– > C6 H5 COO– > H2 PO4 – > NO2 – > HSO4 – > NO3 – ). Addition of fluoride appears to stabilize a strong 1 : 1 complex at low anion concentration, however at higher anion concentrations deprotonation of the urea subunit occurs resulting in the formation of HF2 – (confirmed by 1 H NMR). This process was also characterized by the formation of a new band at 475 nm in the UVVis spectrum upon the additions of the fluoride anion and was clearly present after the addition of two equivalents of fluoride [32]. Gunnlaugsson and co-workers have studied several receptors containing a thiourea group attached to an anthracene moiety [33]. These compounds were designed to behave as fluorescent PET (photo-induced electron transfer) sensors for the detection of anionic species. Proton NMR titration experiments, conducted in DMSO-d6 , confirmed that the anions bind to the

An Introduction to Anion Receptors Based on Organic Frameworks

13

receptors through the two NH protons of the thiourea group and form a 1 : 1 complex. The authors demonstrated that 20a–d act as ideal PET sensors (only fluorescent quantum yield affected upon additions of anions) with quenching of the fluorescence being observed with the addition of fluoride, acetate and dihydrogen phosphate anions. Chloride and bromide did not induce any changes in the fluorescence spectra.

Yoon and co-workers have reported a series of mono- and bis-functionalized anthracenes and described their colorimetric and fluorescent properties for the sensing of both fluoride and pyrophosphate anions. The authors appended either phenylurea or p-nitrophenylurea groups through the 1-position (for the mono-functionalized derivatives 21a and 21b) and the 1- and 8-position (for the bis-functionalized derivative 22a and 22b) of the anthracene [34].

Fluorescent titration experiments with receptors 21b and 22b were carried out in DMSO with a variety of anions, added as the tetrabutylammonium salts, in order to compare the stability constants of the mono- and bisfunctionalized receptors. The strongest anion binding was observed with the bis-functionalized receptor (22b) with stability constants of 108 000, 9700

14

G.W. Bates · P.A. Gale

and 6000 M–1 calculated for fluoride, bromide and pyrophosphate, respectively. For compound 21b a much lower binding constant of 4000 M–1 was found with fluoride as compared to the strong anion complexation observed with 22b, illustrating that there is a cooperative binding effect in operation with the two urea groups in receptors 22a and 22b. Temperature-dependant 1 H NMR experiments in DMF-d also revealed that the anion complex stabil7 ity was enhanced by the formation of a hydrogen bond between the hydrogen atom in the 9-position and both the fluoride and pyrophosphate guests in receptors 22a and 22b. Gale and co-workers have designed 23a, a bis-urea cleft based on ophenylenediamine, to selectively bind carboxylate anions [35]. The geometry of the receptors provides a convergent cleft appropriate for the binding of carboxylate anion through four hydrogen bonds. Stability constants of 3210, 1330 and 732 M–1 were calculated for acetate, benzoate and dihydrogen phosphate, respectively, after analysis of data from 1 H NMR titration experiments conducted in DMSO-d6 /0.5% water at 298 K. The crystal structure of the benzoate complex of compound 23a is shown in Fig. 5 revealing that the receptor binds this carboxylate via four hydrogen bonds in the solid state. The stability constants for compound 23a were found to be greater than constants

Fig. 5 X-ray crystal structure of 23a with benzoate. Chem Commun, p 4696 (2005) reproduced by permission of The Royal Society of Chemistry

An Introduction to Anion Receptors Based on Organic Frameworks

15

obtained with N,N  -diphenylurea with a 2.5-fold increase observed for 23a with acetate compared to the diphenylurea (3210 M–1 for 23a and 1261 M–1 for diphenylurea [36]). The same authors appended electron-withdrawing groups onto both the central aryl ring and peripheral aryl rings of the receptor in order to increase the acidity of the urea NH groups. Titration studies were conducted under the same conditions as for 23a and enhanced stability constants were observed for both receptors 23b and 23c compared to 23a [37]. In the case of acetate the stability constant increased from 3210 M–1 for 23a, to 4020 M–1 and 8080 M–1 for 23b and 23c, respectively. In the case of dihydrogenphosphate there is a decrease in affinity from 732 M–1 for 23a, to 666 M–1 for 23b but a large increase to 4720 M–1 for 23c. The authors proposed that the dihydrogenphosphate anion interacted most strongly with the central NH groups thus with the increased acidity of these NH groups in 23c, due to the presence of the two chloro-groups, stronger complexation is observed. Naphthalene and binaphthalene appended with thiourea groups (24 and 25, respectively) have been synthesized by Kondo and co-workers in order to investigate potential cooperative binding between two thiourea groups in 25 [38]. This group found that 1 : 1 complexes were formed between 25 and fluoride, acetate and dihydrogen phosphate anion, which was confirmed by Job plots in acetonitrile and ESI-MS.

UV-Vis spectroscopy titration experiments in acetonitrile solution were carried out to ascertain the anion binding properties of the receptors with acetate, dihydrogen phosphate, fluoride and chloride. The binding constants revealed that the presence of the second thiourea group in 25 significantly improves the receptor’s affinity for anions with respect to the mono-thiourea 24. The most significant differences were obtained for titration with fluoride and acetate where binding constants of 1.1 × 105 M–1 and 2.1 × 106 M–1 were elucidated for 25 and 3.7 × 103 and 7.7 × 103 for compound 24 with acetate and fluoride, respectively. Pfeffer and co-workers have described the use of a highly rigid[3]polynorbornane as a scaffold on which to append electron-deficient thiourea subunits [39].

16

G.W. Bates · P.A. Gale

The anion binding abilities of 26a and 26b were evaluated by 1 H NMR titration techniques in DMSO-d6 with CH3 COO– , F– , H2 PO4 – and H2 P2 O7 2– (added as their tetrabutylammonium salts). Additions of fluoride to the receptor resulted in a distinctive color change attributed to deprotonation. This process was characterized by the loss of the thiourea NH proton resonances and the appearance of the HF2 – resonance in the 1 H NMR during the titration. Analysis of the binding isotherms of receptors 26a and 26b with acetate revealed that the anions were strongly bound by both 26a and 26b in a 1 : 2 receptor-to-anion complex with each of the thiourea units binding to a single acetate anion. Binding constants (log β1 and log β2 values) of 3.5 (±0.1) and 2.4 (±0.1) were calculated for 26a and 3.5 (±0.1) and 3.0 (±0.1) for 26b with acetate. Titrations with H2 PO4 – were fitted to a 1 : 1 binding model and constants of 3.9 (±0.1) and 3.6 (±0.1) (log β values) were calculated for receptors 26a and 26b, respectively. Pyrophosphate was then investigated to evaluate the binding ability of 26a and 26b with a dianion. Analysis of the titration curves for both 26a and 26b with pyrophosphate revealed the formation of a 2 : 1 receptor-to-anion stoichiometry in which each anion terminus is accommodated by two urea groups of a single receptor.

Extending their work on “cholapods”, A.P. Davis and co-workers have appended urea and thiourea groups from the 7 and 12 positions of the steroid scaffold and evaluated the ability of these receptors to bind chloride and bromide (added as their tetraethylammonium salts) [40]. NMR data was found to be consistent with the formation of predominantly 1 : 1 complexes of the receptors and anions. Stability constants were determined by Cram’s extraction method in water-saturated chloroform at 30 ◦ C [40]. Affinities for both chloride and bromide anions increased through the series 27a–d, reflecting the increase in acidity of the NH groups due to the electron-withdrawing aryl substituents and the change from urea to thiourea in 27d. In the case of chloride the association constant for the “unsubstituted” derivative 27a was calculated to be 1.62 × 107 M–1 , with the

An Introduction to Anion Receptors Based on Organic Frameworks

17

nitrophenyl substituted 27c the constant increased to 4.77 × 108 M–1 and the constant increased further to 1.05 × 109 M–1 with the introduction of the thiourea (27d). The addition of the nitrosulfonamide group in 28a–28c also enhances the anion-binding affinities with the largest constants being observed in the thiourea derivative 28c with constants of 1.03 × 1011 M–1 and 2.59 × 1010 M–1 calculated for chloride and bromide, respectively. A further detailed study of these “cholapod” anion receptors was conducted where the anion binding ability of several receptors with increasing numbers of hydrogen bond donor groups was investigated [41]. It was found that a combination of increasing numbers of hydrogen bonding groups and increasing acidity of the NH groups via electron-withdrawing substituents had a significant effect on the anion stability constants. Receptor 29 was found to have the highest affinities for all the anions investigated except acetate where the previously studied 28b and 28c had higher affinities (2.6 × 1011 M–1 and 2.0 × 1011 M–1 , respectively) when compared to 29 (1.3 × 1011 M–1 ). These steroid-based receptors have also been studied as transport agents for anions across vesicle and cell membranes. Electrochem-

18

G.W. Bates · P.A. Gale

ical, NMR and fluorescence techniques were employed and revealed that the “cholapod” receptors act as mobile carriers and facilitate the transport of chloride ions across vesicle membranes [42]. Reinhoudt and co-workers have synthesized both acyclic and cyclic receptors containing multiple urea-binding sites (e.g. 30 and 31). Anion-binding studies were conducted with these systems and a variety of putative anionic guests (added as their tetrabutylammonium salts) using 1 H NMR titration experiments in DMSO-d6 [43]. In the case of the cleft-like receptors dihydrogen phosphate caused the largest shift in the NH group resonances of all the receptors however an association constant could not be obtained for 30a due to the complexity of the binding processes in solution. Job plot analysis of receptor 30b showed the formation of an exclusive 1 : 2 host/guest complex with dihydrogen phosphate and an association constant of 5 × 107 M–2 was calculated. The thiourea functionalized 30c cleft was also shown to bind dihydrogen phosphate with a 1 : 2 host/guest stoichiometry and chloride with 1 : 1 host/guest stoichiometry.

Macrocyclic receptors 31a and 31b were found to bind both dihydrogen phosphate and chloride in exclusively 1 : 1 host/guest stoichiometries. Binding constants were calculated for 31a and 31b with dihydrogen phosphate and chloride and revealed that dihydrogenphosphate was bound more strongly (2.5 × 103 M–1 for 31a and 4.0 × 103 M–1 for 31b) than chloride (500 M–1 for 31a and <50 M–1 for 31b). Gale and co-workers have combined urea and amide groups into a new macrocyclic motif and studied the anion complexation properties of 32 [44]. Stability constants with a variety of anionic guests were elucidated by 1 H NMR titration techniques in both DMSO-d /0.5% water and DMSO6 d6 /5% water at 298 K. The macrocyclic receptor shows significant selectivity for carboxylate anions over dihydrogen phosphate and chloride. Titrations

An Introduction to Anion Receptors Based on Organic Frameworks

19

in DMSO-d6 /0.5% water resulted in high stability constants for acetate (>104 M–1 ) and benzoate (6430 M–1 ) therefore the titrations were conducted in 5% water, a much more competitive media, and stability constants of 5170 M–1 and 1830 M–1 were calculated for acetate and benzoate, respectively. Interestingly, a crystal structure of a carbonate complex was obtained from a crystallization with tetrabutylammonium fluoride (Fig. 6). It was presumed that carbonate was gained via the fixation of atmospheric CO2 by the fluoride salt-macrocycle solution.

Fig. 6 X-ray crystal structure of a carbonate complex of 32 reproduced by permission of The Royal Society of Chemistry [43]

In 2000 Lee and Hong synthesized tris-thiourea macrocycles 33a and 33b and studied their anion recognition properties by 1 H NMR titration experiments in DMSO-d6 at 25 ◦ C. It was found that macrocycle 33a was selective for dihydrogen phosphate (800 M–1 ) over acetate (320 M–1 ) and chloride (40 M–1 ). In contrast macrocycle 33b was found to be selective

20

G.W. Bates · P.A. Gale

for acetate (5300 M–1 ) over dihydrogen phosphate (1600 M–1 ) and chloride (95 M–1 ) [45]. Recently, Tobe and co-workers have designed cryptand-like macrocycles based on homobenzylic tripodal thiourea and compared their anion-binding properties to a series of acyclic tripod-type receptors [46]. The proton resonances in the 1 H NMR spectra of cryptand-type receptor 34b in various solvents were found to be very broad, possibly due to conformational changes that are slow on the NMR timescale. Therefore, the complexation of 34b with anionic species was evaluated by 1 H NMR titration experiments in CDCl2 CDCl2 at 373 K. Association constants of 116 M–1 and 112 M–1 were calculated for acetate and chloride, respectively, and were found to be much lower than the tripodal receptor 35a under the same condition (3030 M–1 for acetate and 3700 M–1 for chloride). This low binding ability of 34b was attributed to strong intramolecular hydrogen bonds between the thiourea groups. Receptors 35a and 35c were then compared and the stability constants (obtained from 1 H NMR titrations in DMSO-d6 at 303 K) revealed that 35a has poor affinity towards all anionic species in DMSO solutions whereas 35c has high affinity for dihydrogen phosphate and acetate.

An Introduction to Anion Receptors Based on Organic Frameworks

21

3 Aromatic NH Donor Containing Neutral Receptors 3.1 Pyrrole-Based Receptors Sessler and co-workers have pioneered the use of the pyrrole NH hydrogen bond donor group in both charged and neutral anion receptor systems [47]. In 1992 they reported the anion-binding abilities and fluoride selectivity of sapphyrin 36a, a pentapyrrolic macrocycle [48]. Fluorescence titration experiments carried out in methanol revealed that 1 : 1 complexes formed between the diprotonated sapphyrin 36a and halide anions and association constants of 2.8 × 105 , ∼102 and <102 M–1 were calculated with fluoride, chloride and bromide, respectively. Four-years later Sessler reported the effective binding of phosphate by receptor 36b. Phosphorus NMR titration experiments were carried out in methanol-d4 at ambient temperature and revealed that compound 36b bound both phosphoric acid and phenylphosphonic acid with affinity (1.8 × 104 and 1.3 × 104 M–1 for H3 PO4 and C6 H7 PO3 , respectively) [49].

Gale and co-workers have developed a number of receptors based on the 2,5-dicarboxamidopyrrole skeleton where the combination of a pyrrole and amide groups form convergent hydrogen-bonding arrays (e.g. 37a and 37b) [50]. Differences in the solubility of receptors 37a and 37b meant that their ability to bind anions was assessed in different solvents (DMSO-d6 /0.5% water for 37a and CD3 CN for 37b at 298 K). Both receptors proved to be selective for oxo-anions, however 37a bound dihydrogen phosphate most strongly (1450 M–1 ) whereas 37b bound benzoate most strongly (2500 M–1 ) (Fig. 7). Continuing this work, Gale and co-workers have synthesized more acidic diamidopyrrole derivatives 37c and 37d by including electron-withdrawing

22

G.W. Bates · P.A. Gale

Fig. 7 X-ray crystal structure of 37b benzoate complex

nitro groups on the peripheral phenyl rings and assessed their anion-binding ability compared to the unfunctionalized 37a [30]. Proton NMR titrations were carried out in DMSO-d6 /0.5% water solutions at 298 K and revealed that the presence of the electron-withdrawing groups in receptors 37c and 37d improved the systems affinity for anionic guests. In the case of benzoate the association constants increased significantly from 560 M–1 (obtained for 37a under identical conditions) to 4150 M–1 for 37c and 4200 M–1 for 37d. Upon the addition of one equivalent fluoride to 37d the anion appears to coordinate to the receptor. However, upon further additions of fluoride deprotonation occurs (as indicated by the evolution of a blue color in solution due to the deprotonated pyrrole). In the case of compound 37a fluoride was found to bind to the receptor with a stability constant of 1245 M–1 .

An Introduction to Anion Receptors Based on Organic Frameworks

23

Gale, Smith and co-workers have developed prodigiosin mimics based on amidopyrroles in order to co-transport hydrochloride acid across a vesicle membrane. The inclusion of the protonatable imidazole group allows the receptor to carry the proton of the acid and in addition, once protonated, provides an extra hydrogen bond donor group to bind the chloride within the cleft [51].

Sessler, Gale and co-workers have further developed receptors based on the 2,5-diamidopyrrole skeleton by appending 2-aminopyrrole groups (e.g. compound 39) thus increasing the number of hydrogen-bonding groups, which was hoped would increase the selectivity for oxo-anions compared to the previously studied 37a and 37b [52]. Proton NMR titration experiments were conducted in DMSO-d6 /0.5% water at 298 K to investigate the solution phase anion complexation properties of 39 compared to 37a. The stability constants revealed that 39 showed en-

24

G.W. Bates · P.A. Gale

hanced selectivity for both dihydrogen phosphate and benzoate compared to 37a however 39 was found to bind dihydrogen phosphate with higher affinity (10 300 M–1 ) than benzoate (5500 M–1 ), the reverse selectivity observed for 37a. Recently, Sessler and co-workers have appended the 2-aminopyrroles subunit onto a pyridine-2,6-dicarboxamide spacer group to afford two acyclic receptors (compounds 40a and 40b) [53].

Elucidation of the solution phase anion complexation properties of receptors 40a and 40b (determined by UV-Vis spectrophotometric titrations in dichloromethane at 298 K) revealed that 40b bound only acetate with a stability constant of 13 900 M–1 whereas strong binding was observed for benzoate, acetate, NO2 – and CN– with receptor 40a (43 000, 19 000, 13 000 and 5600 M–1 , respectively). Gale and co-workers have also investigated 5,5 -dicarboxamido-dipyrrolylmethanes as anion receptors [54]. These systems demonstrated a remarkable affinity and selectivity for dihydrogen phosphate in highly competitive solvent media with stability constants of 234 M–1 and 20 M–1 being calculated (by 1 H NMR titration experiments at 298 K) for 41a and 41b respectively in DMSO-d6 /25% water. Receptors 41a and 41b were found to be unstable in solution therefore analogous compounds containing two methyl groups attached to the meso-carbon were synthesized (42a and 42b) in the hope they would display increased stability in solution [55].

An Introduction to Anion Receptors Based on Organic Frameworks

25

The meso-substituted derivatives showed lower affinities for anion compared to 41a and 41b however 42a did display selectivity for dihydrogen phosphate over other anions with constants of 1092 M–1 for dihydrogen phosphate, 124 M–1 for fluoride and 41 M–1 for benzoate calculated in DMSOd6 /5% water at 298 K. Sessler, Ustynyuk and co-workers have incorporated dipyrromethane subunits into 2,6-diamidopyridinedipyrromethane hybrid macrocyclic systems [56]. Initially 43 was synthesized and the anion-binding ability was elucidated by UV-Vis spectroscopic titrations in CH3 CN at 23 ◦ C. Weak binding was observed for chloride, bromide, cyanide and nitrate which was rationalized by the receptor forming a deep cavity that favors the formation of well-oriented, directional NH-anion hydrogen bonds. Hydrogensulfate was found to bind strongly to the receptor in a 1 : 1 host/anion fashion in acetonitrile (Ka = 64 000 ± 2600 M–1 ) presumably due to the orientation of the hydrogen-bonding groups within the macrocycle being ideal for tetrahedral anions. The same authors went onto “fine tune” the anion-binding properties of the pyridine-2,6-dicarboxamide-dipyrromethane-hybrid macrocyclic system and designed receptor 44 after DFT calculations suggested it to be more suitable for hydrogensulfate complexation [57]. The receptor’s affinity for a number of anions was determined by UV-Vis spectroscopic titrations in CH3 CN at 23 ◦ C and revealed that bromide, nitrate and chloride are not bound by compound 44. Acetate and dihydrogen phosphate bound with a 1 : 1 stoichiometry with constants of 12 600 ± 450 M–1 and 29 000 ± 1900 M–1 , respectively. An enhanced affinity and selectivity was observed for 44 with hydrogen sulfate (108 000 ± 17 000 M–1 ) compared to 43.

Recently Katayev, Sessler and co-workers have further developed the hybrid macrocycle systems by replacing the pyridine-2,6-dicarboxamide moieties used in receptors 43 and 44 with bipyrrole and dipyrromethane subunits

26

G.W. Bates · P.A. Gale

(compounds 45 and 46) [58]. UV-Vis spectroscopic titrations in acetonitrile at 23 ◦ C showed that the smaller macrocycle 45 bound hydrogensulfate with high affinity (2.7 × 106 M–1 ) similar to the previously studied receptors 43 and 44 whereas the larger bis-dipyrromethane macrocycle 46 exhibited a different selectivity for anions with chloride being selectively bound with high affinity (281 000 M–1 ). Interestingly, titrations with hydrogensulfate resulted in the lowest association constant of all the anions investigated with 46 (2100 M–1 ).

In 1996, Sessler and co-workers reported the anion complexation properties of calix[4]pyrroles e.g. 47. Compound 47 (meso-octamethylcalix[4]pyrrole) was originally synthesized by Baeyer in 1886 and is arguably the simplest anion receptor to synthesize as it is formed in one step via the acid-catalyzed condensation of pyrrole and acetone [59]. The macrocycle forms four hydrogen bonds to anionic guests and binds fluoride and chloride strongly. Recently, it was discovered that this receptor actually functions as an ionpair receptor with large charge diffuse cations sitting in the cup-shaped cavity formed by the pyrrole rings when complexed to an anion [60–62]. A wide variety of functionalized calixpyrroles have been synthesized and are reviewed extensively elsewhere [63, 64].

One approach to dramatically increase the affinity of calixpyrroles for anions is to introduce a “strap” across the macrocycle containing addi-

An Introduction to Anion Receptors Based on Organic Frameworks

27

tional hydrogen bond donor groups. For example, Lee, Sessler and coworkers have recently reported the synthesis of isophthalamide strapped calix[4]pyrroles [65]. The authors studied the effect of varying the length of the strap on the anion complexation properties of the macrocycle and isothermal titration calorimetry was employed to investigate the anion-binding abilities of receptors 48a–c (in CH3 CN at 30 ◦ C). Receptors 48a–c were found to have high binding affinities toward halide anions however they failed to show an appreciable size-dependence selectivity based on the increase of strap length. This is illustrated in the case of chloride (the most strongly bound anion) where association constants of 3.89 × 106 M–1 , 3.35 × 106 M–1 and 3.24 × 106 M–1 were calculated for 48a, 48b and 48c, respectively.

Recently, Lee and co-workers have shown that a binol-strapped calix[4]pyrrole (49) can be used in the enantioselective recognition of carboxylate anions [66]. Both the R- and S-enantiomers of the strapped calixpyrrole were isolated and characterized. Detailed studies of the enantioselectivity of the S enantiomer were carried out by isothermal titration calorimetry experiments in acetonitrile with the chiral anions (R)-2-phenylbutyrate and (S)-2-phenylbutyrate. Stability constants were determined and revealed

28

G.W. Bates · P.A. Gale

that receptor (S)-49 shows selectivity for (S)-2-phenylbutyrate over (R)2-phenylbutyrate with an order of magnitude difference in the constants (Ka (R) = 9.8 × 103 M–1 and Ka (S) = 1.0 × 105 M–1 ). In 2000 Kohnke and co-workers described the synthesis of meso-octamethylcalix[6]pyrrole via the conversion of calix[6]furan into dodecaketone, which was then treated with ammonium acetate to obtain calix[6]pyrrole 50 [67].

Association constants were determined by Cram extraction methods, which revealed that the macrocycle 50 formed a strong complex with chloride (12 800 ± 1300 M–1 ). Proton NMR titration experiments conducted in CD2 Cl2 also revealed the size dependence selectivity of the calix[n]pyrroles where bromide was found to bind approximately seven times stronger to the larger calix[6]pyrrole compared to calix[4]pyrrole (710 ± 25 M–1 for 50 against 10 M–1 for 47) [68]. Recently Cafeo, Kohnke and co-workers have continued work on expanded calixpyrroles and have reported the anion-binding properties of two calix[2]benzo[4]pyrroles 51a and 51b [69]. Elucidation of the stability constants (determined by 1 H NMR titrations in CD2 Cl2 at 20 ◦ C) showed that although chemically similar the two macrocycles have significantly different anion-binding properties. Receptor 51b only bound fluoride and acetate to an appreciable level (2246 ± 132 M–1 for fluoride and 597 ± 236 M–1 for acetate) whereas receptor 51a bound a number

An Introduction to Anion Receptors Based on Organic Frameworks

29

of anions with higher affinities. 51a was found to have selectivity for fluoride with a constant of approximately 20 000 M–1 estimated from competitive binding studies in the presence of 51b. This selectivity is presumably due to good size complementarity between 51b and the fluoride anion. 3.2 Carbazole and Indole-Based Receptors Jurczak and co-workers have described 1,8-diamino-3,6-dichlorocarbazole as a building block for anion receptor construction. Stability constants for receptors 52a and 52b with various anions, added as tetrabutylammonium salts, were calculated by 1 H NMR titration experiments in DMSO-d6 /0.5% water and it was found that compound 52b bound anions more strongly with constants of 115 M–1 and 8340 M–1 for chloride and benzoate, respectively, compared to 13 M–1 and 1230 M–1 with compound 52a with chloride and acetate, respectively [70].

Sessler and co-workers have incorporated two carbazole subunits into expanded calixpyrrole-type macrocycle 53 [71]. Fluorescence titration experiments in dichloromethane at 0.5 µM concentration of host revealed that compound 53 shows selectivity for acetate (Ka = 229 000 M–1 ) over a number of other carboxylate-type anions (benzoate, oxalate and succinate).

Beer and co-workers have reported the anion-binding ability of a number of indolocarbazoles sensors (54a–c) [72]. The highest association constants were obtained with receptor 54c presumably due to the presence of the electron-withdrawing bromide groups increasing the acidity of the NH groups. UV-Vis spectroscopic titrations in acetone at 25 ◦ C revealed that

30

G.W. Bates · P.A. Gale

compound 54c bound benzoate most strongly (log Ka = 5.9) followed by phosphate (log Ka = 5.3) then fluoride (log Ka = 5.0) and chloride (log Ka = 4.9), a trend observed in both 54a and 54b. Recently, indole-based receptors have attracted increasing attention. Sessler and co-workers have described the use of indole subunits for anion recognition within diindolylquinoxalines receptors (55a and 55b) [73]. Stability constants were determined by UV-Vis spectroscopic titrations in dichloromethane at 22 ◦ C and showed that both receptors have appreciable selectivity for dihydrogenphosphate with constants of 6800 M–1 and 20 000 M–1 calculated for compounds 55a and 55b, respectively. Receptor 55b was found to be highly colored and upon the addition of dihydrogenphosphate a visible change in color was observed.

A number of biindoyl-based systems have been developed by Jeong and co-workers for the binding of anionic species. This group have prepared molecular clefts based on the 2,2 -biindoyl scaffold with amide groups attached via alkyne linkers and compared the anion-binding ability of 56 to the more rigid ethyno-bridged receptor 57 [74]. Stability constants were determined by UV-Vis spectroscopy titration experiments in CH3 CN at 22 ◦ C and revealed that receptor 57 did indeed bind the anionic species with higher affinities than 56, the less rigid receptor. Chloride was bound 22-times more strongly by receptor 57 compared to 56 (1.1 × 105 M–1 for 57 against 5.1 × 103 M–1 for 56) whereas bromide was bound 41-times more strongly by 57 compared to 56 (8.7 × 103 M–1 for 57 against 2.1 × 102 M–1 for 56) illustrating the importance of preorganization in the binding of small anionic guests.

An Introduction to Anion Receptors Based on Organic Frameworks

31

Jeong and co-workers have extended the research into biindolyl scaffolds as a building block for anion receptors by synthesizing new macrocyclic systems 58 and 59 [75]. Both compounds 58 and 59 were found to strongly bind various anions (stability constants determined by UV-Vis spectroscopy titrations in CH3 CN at 295 K). In the case of the halide anions a size complementarity was observed for both macrocycles where the smaller fluoride and chloride anions were found to bind more strongly to 58 and 59 than the larger bromide and iodide anions.

Recently, Gale and co-workers have synthesized simple clefts with indole subunits appended to isophthalamide and pyridine-2,6-dicarboxamide spacers and described their anion-binding properties [76]. Proton NMR titrations in DMSO-d6 /0.5 water at 298 K were conducted to elucidate stability constants for a number of anions, added as their terabutylammonium salts. Fluoride affects the largest change on the proton resonance however an association constant was only calculated for 60a (>104 M–1 ) therefore titrations were repeated in more competitive media (DMSO-d6 /5% water). Both receptors showed selectivity for fluoride however 60a bound with a 1 : 1 binding stoichiometry (1360 M–1 ) whereas the data for 60b could only be fitted to a 1 : 2 receptor/anion model (K1 = 940 M–1 and K2 = 21 M–1 ). The crystal structures of 60a with chloride and fluoride are shown in Figs. 8 and 9 respectively.

32

G.W. Bates · P.A. Gale

Fig. 8 X-ray crystal structure of a chloride complex of 60a reproduced by permission of The Royal Society of Chemistry [76]

Fig. 9 X-ray crystal structure of a fluoride complex of 60a reproduced by permission of The Royal Society of Chemistry [76]

4 Hydroxy (OH) Donors in Neutral Receptors In 2003, D.K. Smith showed that simple aromatic hydroxides can complex chloride anions. Smith compared stability constants (obtained from NMR competition experiments in CD3 CN) of phenol, 61, resorcinol, 62 and catechol, 63, with chloride (added as its tetrabutylammonium salt) and found that 63 bound chloride with greater affinity than 61 and 62 (1015 M–1 for 63 against 125 M–1 for 61 and 145 for 62) [77].

An Introduction to Anion Receptors Based on Organic Frameworks

33

Recently, Smith and Winstanley have further explored aromatic hydroxides as anion receptors by studying the effect of ortho-substituents on the chloride binding affinity of catechols 64a, 64b and 65 [78]. Binding constants were elucidated by proton NMR titrations in CD3 CN : DMSO-d6 (9 : 1) solutions and showed that 65 bound chloride with the highest affinity (235 M–1 ) presumably due to the additional hydrogen bonding provided by the amide groups. Although amide groups are present in 64a and 64b it appeared that they were not involved in binding the anion and as a result compounds 64a and 64b bound chloride with lower affinities (110 M–1 for 64a and 115 M–1 for 64b).

Row, Maitra and co-workers have linked two steroid subunits to synthesize macrocycle 66. The fluoride-binding properties of compound 66 were then investigated by a 1 H NMR titration experiment in CDCl3 at 22 ◦ C, which found that the receptor bound fluoride with a 1 : 2 receptor/anion stoichiometry (a result confirmed by Job plot analysis) and stability constants of K1 = 1.8 (±0.1) ×103 M–1 and K2 = 2.5 (±0.35) ×102 M–1 were found [79].

5 Charged Receptors The incorporation of charged groups into receptors designed for anion recognition allows for the receptors to bind the anion with both electrostatic interaction and additional interactions dependent on the group and receptor

34

G.W. Bates · P.A. Gale

design. In the case of imidazolium groups, the cation can stabilize the anion complex with additional CH· · ·A– type hydrogen bonds. 5.1 Imidazolium and Pyridinium-Based Receptors Kang and Kim have synthesized the fluorescent anion receptor compound 67 where two methylene bridged bis-imidazolium subunits are attached to a naphthalene backbone through the 1- and 8-position [80].

Molecular modelling showed that the receptor forms a convergent concave cavity with all the imidazolium C(2)-H’s pointing inwards. Modelling studies led the authors to suggest that the shape of the cavity was predisposed for the binding of halide anions. Fluorescence titration experiments in 90 : 10 CH3 CN : DMSO solutions were carried out with chloride, bromide and iodide anions (added as their tetrabutylammonium salts) and stability constants were calculated that showed that compound 54 had highest affinity for I– (5000 ± 470 M–1 ) followed by Br– (243 ± 15 M–1 ) then Cl– (185 ± 13 M–1 ). Yoon, Kim and co-workers have reported a highly effective fluorescent sensor for dihydrogen phosphate based on a 1,8-disubstituted-anthracene-dimer macrocycle bridged by two imidazolium subunits (68) [81].

Fluorescence titration experiments were conducted in 9 : 1 acetonitrile : DMSO solutions in order to elucidate association constants for 68 with dihydrogenphosphate, fluoride, chloride and bromide. The results confirmed

An Introduction to Anion Receptors Based on Organic Frameworks

35

that compound 68 selectively binds to dihydrogenphosphate over the other anions tested with a stability constant >1 300 000 M–1 . Fluoride also bound to compound 68 with high affinity (340 000 M–1 ) and competition studies of dihydrogenphosphate and fluoride with respect to compound 68 clearly showed that no interference to the dihydrogenphosphate binding occurred in the presence of fluoride. Beer and co-workers have shown how a number of tetrakis(imidazolium) macrocyclic receptors, 69a–d, can be used for anion binding [82]. Proton NMR titration investigations revealed that the macrocycles bind halide anions strongly with fluoride being most strongly bound by 69b and 69c (>104 M–1 for both 69b and 69c). Good size complementarity is seen for iodide with 69d as it gave the highest stability constant (900 M–1 ) compared to the other receptors (370 M, 560 and 470 M–1 for 69a, 69b and 69c, respectively). Benzoate anions were found to bind to the receptor in a 1 : 2 host/anion stoichiometry, a result rationalized by the relative size of the benzoate anion compared with the spherical halides, thus the anion is only partially bound within the cavity allowing a second anion to interact with the cavity.

Alcalde and co-workers have reported that imidazolium-based heterophanes, such as 70, are capable of anion recognition [83]. Proton NMR spectroscopy was employed in order to examine the anionbinding behavior of receptor 70. Upon the addition of a number of anionic

36

G.W. Bates · P.A. Gale

guests (added as their tetrabutylammonium salts) to compound 70, significant changes in the C2 proton resonance of the imidazolium ring were observed in both CD3 CN and DMSO-d6 solutions. Proton NMR titration experiments in DMSO-d6 were then carried out and revealed that 70 binds acetate most strongly and with a 1 : 1 binding stoichiometry (Ka = 359 ± 42 M–1 ). Steed and co-workers have also utilized a tripodal backbone to construct a number of tri-pyridinium “venus flytrap” receptors (71a–c) and investigated their anion binding and sensing properties [84, 85]. Receptors 71a and 71b showed similar anion-binding behavior with both receptors binding chloride most strongly (constants of >100 000 M–1 calculated for both receptors). In the case of compound 71b reduced affinities were observed for both bromide and acetate (3953 M–1 and 2511 M–1 , respectively) compared to 71a (13 800 M–1 and 10 500 M–1 , respectively) which was attributed to the increased steric bulk provided by the benzyl groups. For compound 71c chloride is bound stronger than bromide (similar to 71a and 71b) however the affinities for halides are greatly reduced compared to 71a and 71b (5370 M–1 for chloride and 486 M–1 for bromide). Receptor 71c was highly selective for acetate with the stability constant being almost an order of magnitude higher than the chloride constant (49 000 M–1 against 5370 M–1 ). Variable-temperature 1 H NMR experiments were carried out and

An Introduction to Anion Receptors Based on Organic Frameworks

37

showed that compound 71c selectively binds acetate over the spherical halide anions due to mixtures of conformers being adopted in solution through anthracene-anthracene mutual interactions. Further evidence of the conformational behavior of compound 71c and its selectivity for acetate over other anions was provided by UV spectroscopy and fluorescence studies. Shinoda and co-workers have reported the one-step synthesis and anion binding properties of macrocycle 72. Proton NMR titration experiments (in D2 O) were used to determine the binding properties of 72 for tricarboxylate anions and revealed that the tricarboxylate 72b was bound with the highest affinity (log Ka = 5.1) [86].

5.2 Guanidinium-Based Receptors Guanidinium groups may be regarded as charged analogues of ureas in that they have two parallel NH groups and as a consequence often show high affinities for oxyanionic species such as carboxylates binding these anions by a combination of hydrogen bonding and electrostatic interactions. Guanidinium-carboxylate and phosphate interactions occur in many biological systems as a guanidinium group is present in the amino acid arginine [4] Schmidtchen and co-workers have described the binding of benzoate to the guanidinium-based receptors 73a and 73b [87]. ITC titrations were carried with the iodide salts of 73a and 73b in acetonitrile at 30 ◦ C with benzoate (added as its tetraethylammonium salt) and binding constants of 280 000 M–1 and 203 000 M–1 were calculated for 73a and 73b, respectively. Receptor 73a was investigated further where ITC titration experiments (under identical conditions to the previous titration) were carried with 73a and a variety of counter anions. It was found that the change in counter anion had significant effects upon the binding constant of benzoate observed, for example with

38

G.W. Bates · P.A. Gale

the tetrafluoroborate anion a binding constant of 414 000 M–1 was calculated compared to a binding of 38 000 M–1 with chloride. For several years Schmuck has investigated the binding affinities of guanidinium salts appended to hydrogen bonding pyrrole-containing motifs and has shown how carboxylate anion binding is enhanced by the hybrid receptors. In 1999 Schmuck reported the binding ability of 74a with various carboxylate anions in highly competitive media [88]. Proton NMR titration in DMSO-d6 /40% H2 O at 25 ◦ C revealed that 74a formed stronger complexes with acetate and Ac-L-Phe anions with binding constants of 2790 M–1 and 1700 M–1 , respectively. Schmuck then investigated a series of guanidinium-appended pyrrole receptors and found that 74b bound Ac-l-Ala-O– more strongly than 74a (1610 M–1 and 770 M–1 , respectively) [89]. The binding ability of 74b was then assessed with a range of carboxylate anions by 1 H NMR titrations in DMSO-d6 /40% H2 O at 25 ◦ C and showed that compound 74b formed stronger complexes with the anions than 74a. High affinities were observed for 2pyrrole-COO– and acetate (5275 M–1 and 3380 M–1 ). A notable result was that compound 74b displayed enantioselectivity in the case of Ac-Ala-O– anions where a higher affinity was observed for the l-enantiomer over the d-enantiomer (1610 M–1 vs. 930 M–1 ).

In 2005 Schmuck and Schwegmann reported the study of a tripodal “molecular flytrap” 75 where the pyrrole-guanidinium moieties were appended to a triamide backbone. The receptor was designed to bind tricarboxylate anions and UV and fluorescence titration experiments in water showed that 75 bound citrate and trimesoate with association constants >105 M–1 [90]. Recently, de Mendoza and co-workers have reported the complexation of nitrate to an acyclic cleft and a series of macrocycles based on guanidinium [91]. Association constants were calculated by ITC titrations in acetonitrile at 303 K and it was found that the macrocyclic receptors bound the nitrate an-

An Introduction to Anion Receptors Based on Organic Frameworks

39

ion more strongly than the acyclic system. The constants obtained for the macrocycles 77a–c revealed a size dependence on the binding of nitrate with the largest macrocycle 77c giving rise to the highest association constant of 73.7 × 103 M–1 , an order of magnitude greater than the smallest macrocycle 77a (7.26 × 103 M–1 ). 5.3 Ammonium-Containing Receptors The work of Park and Simmons [92] has inspired many efforts into the research of ammonium- and polyammonium-based anion receptors and are the subject of numerous reviews [93, 94] Here we will only look at receptors containing quaternary ammonium centers. These groups bind anions via electrostatic interactions only. An early pioneer in the area of ammonium-based anion receptors, Schmidtchen synthesized the quaternary ammonium-based macrocyclic receptors 78a–c in 1977 [95].

40

G.W. Bates · P.A. Gale

NMR spectroscopy revealed that upon the addition of 1 equiv. of alkali metal halide salts to receptors 78a–c, 1 : 1 complexes were formed. Stability constants were measured using halide electrodes in water solutions, which revealed that the receptors bound bromide anions (log Ka = 1.8, 2.45 and 2.45 for 78a, 78b and 78c, respectively) and iodide anions (log Ka = 2.2 and 2.4 for 78b and 78c, respectively).

Frontera, Anslyn and co-workers have described the binding of tricarboxylate salts with the tris-ammonium-squaramide-appended tripodal receptor 79 [96]. Isothermal titration calorimetry was used to study the association of a number of tricarboxylate salts and 79 in 1 : 3 water:ethanol solutions at 294 K. It was found that 79 bound the less rigid tricarboxylates citrate and tricaballate more strongly than the rigid benzene-1,3,5-tricarboxylate (1.1 ± 0.1 × 105 M–1 , 1.5 ± 0.2 × 105 M–1 and 4.5 ± 0.5 × 104 M–1 , respectively). Compound 79 was also investigated as a receptor for the biscarboxylates gluterate and succinate. The stability constants revealed that gluterate was bound with significantly higher affinity than succinate (2.2 ± 0.2 × 104 vs. ∼ 2.8 × 102 , respectively). Costa and co-workers have reported 80, a fluorescent squaramidecontaining macrocyclic receptor for monitoring sulfate in water [97]. Isothermal titration calorimetry was employed to characterize the host–guest association of 80 with SO4 2– , PhOPO3 2– and C2 O4 2– dianions (titration carried out in methanol at 294 K). The data was fitted to a 1 : 1 binding model and it was found that 80 bound SO4 2– (4.6 ± 1.0 × 106 M–1 ) with the

An Introduction to Anion Receptors Based on Organic Frameworks

41

strongest affinity followed by C2 O4 2– (3.2 ± 0.3 × 105 M–1 ) then PhOPO3 2– (1.5 ± 0.2 × 104 M–1 ). A fluorescein-80 complex was then synthesized and competitive fluorescent titration experiments in 9:1 methanol:water solutions were conducted with the complex against sodium sulfate and an association constant of 5.2 ± 1.2 × 106 M–1 was calculated.

Bowman-James and co-workers have synthesized a series of amide-based macrocycles containing either tertiary amine spacer groups or quaternized ammonium functionalities and have assessed their ability to bind a number of anions in solution [98]. Stability constants were calculated by proton NMR titration experiments in DMSO-d6 solutions and revealed that the quaternized macrocycles 82a and 82b showed higher affinities for anions compared to the neutral analogues 81a and 81b attributed to the additional electrostatic attraction of the quaternary ammonium group. The pyridine analogues 81b and 82b were also shown to have higher binding constants than the isophthaloyl derivatives 81a and 82a attributed to the pyridine-assisted preorganization of the macrocycle. All the anions binding data was fitted to 1 : 1 isotherms and receptor 82b was found to have the highest affinity for anions with dihydrogenphosphate being most strongly bound (log K = 5.32). Receptor 82b was found to bind halide anions in the order Cl– > Br– > I– > F– whilst receptor 82a was also found to bind chloride more strongly than other halide anions (log K = 3.23 and 2.14 for 82a with Cl– and Br– , respectively), illustrating the good size complementarity of 82a and 82b to chloride.

42

G.W. Bates · P.A. Gale

6 Conclusions The examples discussed in this review provide a broad overview of the variety of synthetic organic receptors used for binding and sensing of anionic species. As the understanding of the processes and factors that influence the effective binding of anions improves there is an increasing impetus to apply this knowledge to solve real-world problems. Areas that are likely to benefit from this knowledge are in the transport of anions, in separation processes and in biological systems leading towards new treatments for disease and cancer. Acknowledgements We would like to thank the EPSRC/Crystal Faraday for a project studentship (GWB).

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.

Gale PA, Quesada R (2006) Coord Chem Rev 250:3219 Gale PA, García-Garrido SE, Garric J (2008) Chem Soc Rev. DOI 10.1039/b715825d Gale PA (2006) Acc Chem Res 39:465 Sessler JL, Gale PA, Cho W-S (2006) Anion Receptor Chemistry. RSC Publishing, Cambridge Xu H, Strater N, Schroder W, Bottcher C, Ludwind K, Saenger W (2003) Acta Cryst D59:815 Pflugrath JW, Quiocho FA (1985) Nature 314:257 Pflugrath JW, Quiocho FA (1988) J Mol Biol 200:163 He JJ, Quiocho FA (1991) Science 251:1497 Luecke H, Quiocho FA (1990) Nature 347:402 Pascal RA, Spergel J, van Engen D (1986) Tetrahedron Lett 27:4099 Valiyaveettil S, Engbersen JFJ, Verboom W, Reinhoudt D (1993) Angew Chem Int Ed 32:900 Kavallieratos K, de Gala SR, Austin DJ, Crabtree RH (1997) J Am Chem Soc 119:2325 Hughes MP, Smith BD (1997) J Org Chem 62:4492 Santacroce PV, Davis JT, Light ME, Gale PA, Iglesias-Sanchez JC, Prados P, Quesada R (2007) J Am Chem Soc 129:1886 Webb JEA, Crossley MJ, Turner P, Thordarson P (2007) J Am Chem Soc 129:7155 Prohens R, Tomas S, Morey J, Deya PM, Ballester P, Costa A (1998) Tetrahedron Lett 39:1063 Davis AP, Perry JJ, Williams RP (1997) J Am Chem Soc 119:1793 Choi K, Hamilton AD (2001) J Am Chem Soc 123:2456 Szumna A, Jurczak J (2001) Eur J Org Chem, p 4031 Chmielewski MJ, Jurczak J (2005) Chem Eur J 11:6080 Chmielewski MJ, Jurczak J (2006) Chem Eur J 12:7652 Kavallieratos K, Bertao CM, Crabtree RH (1999) J Org Chem 64:1675 Kang SO, Powell D, Bowman-James K (2005) J Am Chem Soc 127:13478 Kang SO, Powell D, Day VW, Bowman-James K (2006) Angew Chem Int Ed 45:1921 Kubik S, Goddard R, Kirchner R, Nolting D, Seidel J (2001) Angew Chem Int Ed 40:2648

An Introduction to Anion Receptors Based on Organic Frameworks

43

26. Kubik S, Kirchner R, Nolting D, Seidel J (2002) J Am Chem Soc 124:12752 27. Gunnlaugsson T, Kruger PE, Jensen P, Pfeffer FM, Hussey GM (2003) Tetrahedron Lett 44:8909 28. Camiolo S, Gale PA, Hursthouse MB, Light ME, Shi AJ (2002) Chem Commun, p 758 29. Gale PA, Navakhun K, Camiolo S, Light ME, Hursthouse MB (2002) J Am Chem Soc 124:11228 30. Camiolo S, Gale PA, Hursthouse MB, Light ME (2003) Org Biomol Chem 1:741 31. Boiocchi M, Boca LD, Gomez DE, Fabbrizzi L, Licchelli M, Monzani E (2004) J Am Chem Soc 126:16507 32. Amendola V, Esteban-Gomez D, Fabbrizzi L, Licchelli M (2006) Acc Chem Res 39:343 33. Gunnlaugsson T, Davis AP, Hussey GM, Tierney J, Glynn M (2004) Org Biomol Chem 2:1856 34. Kwon JY, Jang YJ, Kim SK, Lee K-H, Kim JS, Yoon J (2004) J Org Chem 69:5155 35. Brooks SJ, Gale PA, Light ME (2005) Chem Commun, p 4696 36. Brooks SJ, Garcia-Garrido SE, Light ME, Cole PA, Gale PA (2007) Chem Eur J 13:3320 37. Brooks SJ, Edwards PR, Gale PA, Light ME (2006) New J Chem 30:65 38. Kondo S-I, Nagamine M, Yano Y (2003) Tetrahedron Lett 44:8801 39. Pfeffer FM, Gunnlaugsson T, Jensen P, Kruger PE (2005) Org Lett 7:5357 40. Ayling AJ, Perez-Payan N, Davis AP (2001) J Am Chem Soc 123:12716 41. Clare JP, Ayling AJ, Joos J-B, Sisson AL, Magro G, Perez-Payan MN, Lambert TN, Shukla R, Smith BD, Davis AP (2005) J Am Chem Soc 127:10739 42. Koulov AV, Lambert TN, Shukla R, Jain M, Boon JM, Smith BD, Li H, Sheppard DN, Joos J-B, Clare JP, Davis AP (2003) Angew Chem Int Ed 42:4931 43. Snellink-Ruel BHM, Antonisse MMG, Engbersen JFJ, Timmerman P, Reinhoudt DN (2000) Eur J Org Chem, p 165 44. Brooks SJ, Gale PA, Light ME (2006) Chem Commun, p 4344 45. Lee KH, Hong J (2000) Tetrahedron Lett 41:6083 46. Hisaki I, Sasaki S-I, Hirose K, Tobe Y (2007) Eur J Org Chem, p 607 47. Sessler JL, Camiolo S, Gale PA (2003) Coord Chem Rev 240:17 48. Shionoya M, Furuta H, Lynch V, Harriman A, Sessler JL (1992) J Am Chem Soc 114:5714 49. Kral V, Furuta H, Shreder K, Lynch V, Sessler JL (1996) J Am Chem Soc 118:1595 50. Gale PA, Camiolo S, Tizzard GJ, Chapman CP, Light ME, Coles SJ, Hursthouse MB (2001) J Org Chem 66:7849 51. Gale PA, Light ME, McNally B, Navakhun K, Sliwinski KE, Smith BD (2005) Chem Commun, p 3773 52. Sessler JL, Pantos GD, Gale PA, Light ME (2006) Org Lett 8:1593 53. Sessler JL, Barkey NM, Pantos GD, Lynch VM (2007) New J Chem 31:646 54. Vega IED, Camiolo S, Gale PA, Hursthouse MB, Light ME (2003) Chem Commun, p 1686 55. Vega IED, Gale PA, Hursthouse MB, Light ME (2004) Org Biomol Chem 2:2935 56. Sessler JL, Katayev E, Pantos GD, Ustynyuk YA (2004) Chem Commun, p 276 57. Sessler JL, Katayev E, Pantos GD, Scherbakov P, Reshetova MD, Khurstalev VN, Lynch VM, Ustynyuk YA (2005) J Am Chem Soc 127:11442 58. Katayev E, Boev N, Khurstalev VN, Ustynyuk YA, Tananaev IG, Sessler JL (2007) J Org Chem 72:2886 59. Gale PA, Sessler JL, Kral V, Lynch VM (1996) J Am Chem Soc 118:5140 60. Custelcean R, Delmau LH, Moyer BA, Sessler JL, Cho WS, Gross D, Bates GW, Brooks SJ, Light ME, Gale PA (2005) Angew Chem Int Ed 44:2537

44

G.W. Bates · P.A. Gale

61. Sessler JL, Gross D, Cho WS, Lynch VM, Schmidtchen FP, Bates GW, Light ME, Gale PA (2006) J Am Chem Soc 128:12281 62. Bates GW, Gale PA, Light ME (2006) Cryst Eng Comm 8:300 63. Gale PA, Sessler JL, Kral V (1998) Chem Commun, p 1 64. Gale PA, Anzenbacher P Jr, Sessler JL (2001) Coord Chem Rev 222:57 65. Lee C-H, Lee J-S, Na H-K, Yoon D-W, Miyaji H, Cho W-S, Sessler JL (2005) J Org Chem 70:2067 66. Miyaji H, Hong S-J, Jeong S-D, Yoon D-W, Na H-K, Hong J, Ham S, Sessler JL, Lee C-H (2007) Angew Chem Int Ed 46:2508 67. Cafeo G, Kohnke FH, La Torre GL, White AJP, Williams DJ (2000) Angew Chem Int Ed 39:1496 68. Cafeo G, Kohnke FH, La Torre GL, Parisi MF, Nascone RP, White AJP, Williams DJ (2002) Chem Eur J 8:3148 69. Cafeo G, Kohnke FH, White AJP, Garozzo D, Messina A (2007) Chem Eur J 13:649 70. Chmielewski MJ, Charon M, Jurczak J (2004) Org Lett 6:3501 71. Piatek P, Lynch VM, Sessler JL (2004) J Am Chem Soc 126:16073 72. Curiel D, Cowley A, Beer PD (2005) Chem Commun, p 236 73. Sessler JL, Cho D-G, Lynch V (2006) J Am Chem Soc 128:16518 74. Chang K-J, Chae MK, Lee C-H, Lee J-Y, Jeong K-S (2006) Tetrahedron Lett 47:6385 75. Chang K-J, Moon D, Lah MS, Jeong K-S (2005) Angew Chem Int Ed 44:7926 76. Bates GW, Gale PA, Light ME (2007) Chem Commun, p 2121 77. Smith DK (2003) Org Biomol Chem 1:3874 78. Winstanley KJ, Smith DK (2007) J Org Chem 72:2803 79. Ghosh S, Choudhury AR, Row TN, Maitra U (2005) Org Lett 7:1441 80. Kim H, Kang J (2005) Tetrahedron Lett 46:5443 81. Yoon J, Kim SK, Singh J, Lee JW, Yang YJ, Chellappan K, Kim KS (2004) J Org Chem 69:581 82. Wong WWH, Vickers MS, Cowley AR, Paul RL, Beer PD (2005) Org Biomol Chem 3:4201 83. Alcalde E, Mesquida N, Perez-Garcia L (2006) Eur J Org Chem, p 3988 84. Abouderbala LO, Belcher WJ, Boutelle MG, Cragg PJ, Dhaliwal J, Fabre M, Steed JW, Turner DR, Wallace KJ (2002) Chem Commun, p 358 85. Wallace KJ, Belcher WJ, Turner DR, Syed KF, Steed JW (2003) J Am Chem Soc 125:9699 86. Shinoda S, Tadokoro M, Tsukube H, Arakawa R (1998) Chem Commun, p 181 87. Haj-Zaroubi M, Mitzel NW, Schmidtchen FP (2002) Angew Chem Int Ed 41:104 88. Schmuck C (1999) Chem Commun, p 843 89. Schmuck C (2000) Chem Eur J 6:709 90. Schmuck C, Schwegmann M (2005) J Am Chem Soc 127:3373 91. Blondeau P, Benet-Buchholz J, de Mendoza J (2007) New J Chem 31:736 92. Park CH, Simmons HE (1968) J Am Chem Soc 90:2431 93. Garcia-Espana E, Diaz P, Llinares JM, Bianchi A (2006) Coord Chem Rev 250:2952 94. Llinares JM, Powell D, Bowman-James K (2003) Coord Chem Rev 240:57 95. Schmidtchen FP (1977) Angew Chem Int Ed Engl 16:720 96. Frontera A, Morey J, Oliver A, Piña MN, Quiñonero D, Costa A, Ballester P, Deyà PM, Anslyn EV (2006) J Org Chem 71:7185 97. Prohens R, Martorell G, Ballester P, Costa A (2001) Chem Commun, p 1456 98. Hossain MA, Kang SO, Powell D, Bowman-James K (2003) Inorg Chem 42:1397

Struct Bond (2008) 129: 45–94 DOI 10.1007/430_2007_073 © Springer-Verlag Berlin Heidelberg Published online: 19 January 2008

Metal-Based Anion Receptor Systems Simon R. Bayly (u) · Paul D. Beer (u) Department of Chemistry, Inorganic Chemistry Laboratory, University of Oxford, South Parks Road, Oxford OX1 3QR, UK [email protected], [email protected] 1 1.1 1.2

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Metal Complexes in Anion Sensing . . . . . . . . . . . . . . . . . . . . . . Basis of Anion Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

46 46 46

2

Redox-Active Transition Metal-Based Receptors for Electrochemical Anion Sensing . . . . . . . . . . . . . . . . . . . . . Theory of Electrochemical Sensing . . . . . . . . . . . . . . . . . . . . . Metallocene Redox Anion Sensors . . . . . . . . . . . . . . . . . . . . . . Mixed Metal Metallocene-Lewis Acid Anion Receptors . . . . . . . . . . . Transition Metal Polypyridyl Anion Receptors . . . . . . . . . . . . . . . Transition Metal Dithiocarbamates as Receptors for Redox Anion Sensing Dendrimers as Redox Anion Sensors . . . . . . . . . . . . . . . . . . . . Surface Confined Redox Anion-Sensing Systems . . . . . . . . . . . . . .

. . . . . . . .

47 47 48 53 56 59 60 63 68 68 68 72

3.5 3.6 3.7

Optical Anion Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Theory of Optical Anion Sensing . . . . . . . . . . . . . . . . . . . . . . . . Metallocene Optical Anion Sensors . . . . . . . . . . . . . . . . . . . . . . Ruthenium(II) Polypyridyl Complexes as Optical Anion Sensors . . . . . . Optical Anion Sensing Using Reporter Groups Based on Other Transition-Metal Complexes . . . . . . . . . . . . . . . . . Transition Metals as Anion-Binding Groups and/or Structural Components Optical Anions Sensing by Lanthanide(III) Complexes . . . . . . . . . . . . Surface Confined Systems for Optical Anion Sensing . . . . . . . . . . . . .

75 79 86 88

4

Metal-Based Anion Receptors Without Reporter Groups . . . . . . . . . .

89

5

Conclusion/Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

91

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

91

2.1 2.2 2.3 2.4 2.5 2.6 2.7 3 3.1 3.2 3.3 3.4

Abstract Metal complexes play an important role in anion receptor chemistry. In the majority of examples metal centres are used as optical and/or electrochemical reporter groups in anion-sensing applications. Metal centres can also act as Lewis acidic anion-binding sites in their own right, and/or as structural components allowing the self-assembly of anion-binding domains. This review describes the development of metalbased receptors with regard to their anion-sensing properties, and is therefore divided into sections on electrochemical and optical anion sensing. Within these sections coverage has been given to the diverse range of metals and anion-binding groups that have been studied. Emphasis has been placed on recently described novel supramolecular, nanoscale and surface confined anion receptor systems that give added functionality. The last section describes metal-based anion receptors without reporter groups.

46

S.R. Bayly · P.D. Beer

Keywords Anion sensing · Anion recognition · Anion coordination · Anion receptors · Organometallic receptors

1 Introduction 1.1 Metal Complexes in Anion Sensing Metal complexes have played an important role in anion receptor chemistry since its earliest examples. The presence of a metal ion can introduce a range of advantageous physicochemical properties to this class of receptor. In the majority of examples the metal complex is incorporated as a reporter group, whose photochemical or redox response is changed upon proximal binding of an anion. Furthermore, the metal can contribute directly to anion binding, either by using its positive charge to electrostatically attract the anion and/or by acting as a Lewis acidic binding site. The metal complex motif can also be utilised as a structural component in anion receptors, where its coordination geometric properties allow the self-assembly of receptor sites with a wide range of topologies not possible with simple tetrahedral covalent bonds. By exploiting these different properties, often in combination, metal complex anion receptors achieve a range of functionality beyond the scope of purely organic structures. Previous reviews on this topic have included many aspects of anion recognition by metal-based receptors [1–6]. This review does not seek to be comprehensive; instead it is designed to provide an introduction to the area by highlighting notable examples and to bring the reader up to date with significant recent results, especially in the application of metal-based anion receptors in surface fabricated nanoscale sensor systems. 1.2 Basis of Anion Sensing Molecular sensing refers to a remotely detectable change in the properties of a receptor molecule on binding of an analyte. The generic design for an anion sensor utilises a spacer group to covalently link an anion receptor site to a signalling or reporter group. Provided the spacer allows some degree of coupling between the components (through-space, throughbond, or by conformational change), binding of an anion at the receptor site perturbs the electronic properties of the signalling group in a way that can be detected spectroscopically or electrochemically. Thus, when the signalling group is a suitable redox-active metal centre, binding can be probed electrochemically (e.g. by voltammetry). If a suitable chromophore or flu-

Metal-Based Anion Receptor Systems

47

orophore is used, sensing can be accomplished via optical spectroscopy (UV/Vis absorption or luminescence, respectively). The spectroscopic properties of many metal complexes make them particularly amenable for this. NMR spectroscopy can also be used to detect anion binding to diamagnetic metal-based receptors. However, this is not generally regarded as a remote sensing technique since it obviously requires the sample to be placed in an external magnetic field. This review is conveniently divided into sections on metal-based receptors for electrochemical anion sensing, receptors for optical anion sensing and anion receptors without reporter groups.

2 Redox-Active Transition Metal-Based Receptors for Electrochemical Anion Sensing 2.1 Theory of Electrochemical Sensing When a redox-active transition metal is used as the signalling unit of a receptor, anion binding is coupled to electron transfer, i.e. anion binding changes the redox potential (couple) of the transition metal. This electrochemical shift can be represented as ∆E0 , the difference in redox potentials between the receptor : anion complex and the receptor alone. Concomitantly, electron transfer at the redox centre also changes the affinity of the receptor for the guest species. These coupled processes are linked thermodynamically by Eq. 1, where Kred and Kox are the stability constants of the reduced and oxidised forms of the receptor:anion complex respectively [7]. nF(∆E0 ) = RT ln(Kox /Kred )

(1)

From a thermodynamic standpoint, the value of the shift in redox potential is determined by the ratio of Kox /Kred , instead of the absolute value of either Kox or Kred . As a consequence a receptor need not necessarily have a very high binding strength for the anion to be sensed. If electron transfer leads to a sufficiently large change in the stability of the receptor : anion complex, a measurable change in redox potential can be observed. Anion binding stabilises the oxidised form of the receptor, hence Kox / Kred > 1 and the redox potential of the reporter group is shifted to a more negative value (cathodic shift). The Kox /Kred ratio is also a measure of how efficient the coupling is between the metal-based reporter group and the anion-binding site.

48

S.R. Bayly · P.D. Beer

2.2 Metallocene Redox Anion Sensors Anion receptors incorporating the redox-active cobaltocenium group have been studied extensively due to the combination of an accessible redox couple and favourable electrostatic interactions of the cationic organometallic metallocene motif with anions. The first anion receptors based on this species were reported by Beer and co-workers in 1989 [8]. The macrocyclic bis-cobaltocenium receptor 1 was shown to bind bromide in acetonitrile solution (due to electrostatic interaction). Electrochemical anion sensing was also demonstrated, where bromide caused the potential of the cobaltocenium/cobaltocene to undergo a cathodic shift. Augmentation of cobaltocenium-based receptors with hydrogen-bond donor groups, such as amides in receptors 2 and 3, generates both stronger and more selective anion binding [9]. Proton NMR anion titration studies in CD3 CN reveal 2 and 3 to have selectivity for dihydrogenphosphate over chloride by approximately an order of magnitude [10]. This is attributed not only to the greater basicity of the dihydrogenphosphate anion, but also to complementary hydrogen bonding between the receptor and the anion. In these receptors the electrochemical sensing of anions is also enhanced, with chloride giving rise to cathodic shifts in the cobaltocenium/cobaltocene redox couple of 30 and 85 mV for 2 and 3, respectively. The dihydrogenphosphate anion generates cathodic shifts of 200 and 240 mV respectively, confirming that in this class of amide hydrogen-bonding receptors the magnitude of the electrochemical response directly mirrors the strength of the receptor-anion interaction. Ferrocene has also been extensively exploited in redox responsive anion receptor design. One advantage of ferrocene is that its synthetic chemistry

Metal-Based Anion Receptor Systems

49

is highly developed, in particular in terms of its conjugation to organic molecules. From the point of view of anion sensing the most relevant difference between the metallocenes is that ferrocene is neutral in charge and therefore its derivatives have no inherent electrostatic interaction with anions (until oxidised to ferrocenium) and therefore their complexes with anions exhibit relatively lower stability constants.

Molecules 4–8 are a selection of ferrocene-based receptors which incorporate amide groups for the hydrogen-bonding of anions [11, 12]. In acetonitrile solution dihydrogenphosphate induced cathodic shifts of up to 240 mV in the ferrocene/ferrocenium couples of these receptors. Competition experiments demonstrated the same shift even in the presence of a 10-fold excess of chloride or hydrogensulfate. In these receptors it is largely the stability of the electrostatically enhanced anion : ferrocenium complex which determines the magnitude of the redox shift. Receptor 8 has the opposite selectivity, displaying a hydrogensulfate induced shift of 220 mV which does not change in the presence of excess dihydrogenphosphate. It is thought that binding of

50

S.R. Bayly · P.D. Beer

HSO4 – leads to protonation of the amine group of the receptor. The resulting complex is cationic and this has a very high affinity for the residual SO4 2– anion. Electrostatic interactions are particularly important in redox anion sensing, and even very simple anion-binding motifs such as the ammonium cation provide an increased redox response. This has been demonstrated by Moutet and co-workers using 9 [13]. This molecule is able to sense dihydrogenphosphate and ATP2– in a range of solvents, displaying a shift of 470 mV in CH2 Cl2 with dihydrogenphosphate, solely due to a strong ionpairing interaction.

Sensing anions in aqueous conditions is a particular challenge which must be met for molecular anion receptors to become a useful technology in biological or environmental analysis. The high dielectric and competitive hydrogen-bond donor capacity of water diminishes anion-receptor interactions. In general strong electrostatic interactions are required to overcome this. Beer et al. have developed a series of ferrocene-based receptors appended with various open chain and cyclic amine functional groups, e.g. molecules 10 and 11, that bind ATP2– and dihydrogenphosphate in water [14–16]. The selectivity of this class of receptors is pH driven. At pH 6.5 at least two of the amines are protonated and the 1 : 1 anion : receptor complexes formed show cathodic shifts of 60–80 mV in the ferrocene/ferrocenium redox couple. Quantitative determination of phosphate and sulfate in the presence of competitor anions was demonstrated by metallacyclic receptors 12 and 13. The electrochemistry of these receptors was studied in 70 : 30 THF : H2 O over a range of pHs. A maximum selective redox shift of 54 mV for phosphate over sulfate was observed at pH 4 for 12, whereas 13 gave a maximum selective redox shift of 50 mV for phosphate over sulfate at pH 7.

Metal-Based Anion Receptor Systems

51

Recognition of fluoride in aqueous media is particularly difficult due to the strongly hydrated nature of the anion. Shinkai and co-workers have demonstrated that ferrocene-boronic acid 14 acts as a selective redox sensor for fluoride which operates in H2 O [17]. The favourable interaction between boron and fluoride (a hard-acid and hard-base, respectively) generates a stability constant of 700 M–1 for the fluoride-ferrocenium complex. Stability constants for both the bromide and chloride complexes are < 2 M–1 .

In receptor molecules that contain multiple metal centres and anion binding groups the redox sensing properties are dictated by the precise spatial arrangement of these groups. In receptors where the pendant reporter groups are in close proximity to each other an increased redox response to anion binding is often observed. For example, the cyclotriveratrylene amides 15 and 16 include closely spaced multiple amide anion-binding groups with pendant ferrocene reporter units [18]. In CH2 Cl2 or acetonitrile solution cathodic shifts of up to 260 mV were observed in the presence dihydrogenphosphate or ATP2– . On oxidation 15 and 16 gain a triple positive charge. The magnitude of the redox shift can be attributed to the increased electrostatic affinity of this multiply charged oxidised species for a single anion compared to monoferrocenyl receptors which are monopositive on oxidation. Receptor 17 has a similar topology—it comprises four cobaltocenium amide groups attached to a porphyrin backbone as the cis-α,α,α,α-atropisomer [19]. Proton NMR titrations in CD3 CN showed chloride and bromide to be bound in 1 : 1 stoichiometry with stability constants of 860 and 820 M–1 , respectively, whereas nitrate exhibited weaker binding with K = 190 M–1 . Electrochemical studies displayed cathodic shifts in the cobaltocene/cobaltocenium redox couple of 35–75 mV on addition of chloride or hydrogensulfate, and 225 mV for dihydrogenphosphate in acetonitrile solu-

52

S.R. Bayly · P.D. Beer

tion. Smaller shifts were seen in the porphyrin oxidation wave. The overall selectivity trend was Cl– > Br–
NO3 – . The anion recognition properties of cobaltocenium calix[4]arene receptors 18–20 were found to be dependent on the structure of the upper-rim of the calix[4]arene [20]. In 1 H NMR studies in DMSO-d6 solution 18 shows a greater affinity for acetate than for dihydrogenphosphate whereas its isomer 19 displays the opposite trend. In 20 there is only a single cobaltocenium group which bridges the upper rim of the calix[4]arene. This receptor displays a significantly greater affinity for the above anions despite possessing only a single positive charge. For example, the cobaltocene/cobaltocenium redox couple of 20 was found to undergo a cathodic shift of 155 mV in the presence of acetate. It is proposed that the surprising strength of the interaction is due to the topology of the anion-binding cavity, in which the arrangement of the two amide hydrogen bond donors is complementary to bidentate anions such as carboxylates. The same selectivity is seen in the ferrocene-1,1 -bisamide analogues 21–23 [21]. Results of 1 H NMR studies in CD3 CN show that these receptors also preferentially bind carboxylate anions (acetate and benzoate) over dihydrogenphosphate and chloride. Other carboxylate selective redox sensors based on ferrocene include neutral molecule 24, which utilises hydrogen bonding to bind mono and dicar-

Metal-Based Anion Receptor Systems

53

boxylate anions with 2 : 1 and 1 : 1 guest : host stoichiometry [22]. However, it cannot distinguish between these types of anion electrochemically, giving a maximum cathodic shift of 150 mV for both acetate and phthalate. A more recent example shows that selectivity for dicarboxylate anions over monocarboxylates and other simple anions can be achieved. Tetra-ammonium macrocycle 25 binds phthalate, isopthalate and dipicolinate with a 2 : 1 guest : host stoichiometry, giving maximum cathodic shifts in the redox potential of 275, 193 and 168 mV, respectively [23]. In comparison the monoacid 4nitrobenzoate produced a maximum cathodic shift of only 49 mV. The incorporation of crown ether units into a cobaltocenium receptor has been shown to allow the switchable redox sensing of anions. Proton NMR titrations of receptor 26 in CD3 CN solution gave log K values of 3.1 for chloride and 3.0 for bromide [24]. Electrochemical titrations showed cathodic shifts of the cobaltocene/cobaltocenium redox couple of 60 and 30 mV for the two anions, respectively. However, when either the NMR or electrochemical titrations were carried out in the presence of K+ no significant anion induced shifts were observed. It is proposed that the K+ ions form a 1 : 1 intramolecular sandwich complex with the two benzocrown ether units of the receptor causing a concomitant change in conformation of the amide groups which reduces their availability for anion binding. Molina and co-workers have investigated the urea-crown ether functionalised ferrocene 27 [25]. This receptor-produced anion induced cathodic shifts in the ferrocene/ferrocenium redox couple of 52 mV with fluoride and 190 mV with dihydrogenphosphate. In acetonitrile solution on addition of 2 equivalents of K+ ions a dramatic attenuation in the anion-induced cathodic shift was observed, with dihydrogenphosphate giving rise to a shift of only 50 mV.

54

S.R. Bayly · P.D. Beer

2.3 Mixed Metal Metallocene-Lewis Acid Anion Receptors Lewis acidic metal centres can be utilized as anion-binding groups, which in combination with ferrocenyl reporter groups provide enhanced redox anion sensing. Ferrocene-amide receptor 28 utilises pendant phosphine groups to allow the coordination of various transition metals to generate mixed-metal complexes 29–32 [26]. 1 H NMR titrations carried out in CD2 Cl2 solution revealed that the neutral molecules 29–31 bind chloride approximately an order of magnitude more strongly than the parent phosphine. The affinity of these complexes for bromide, iodide and hydrogenphosphate was also found to be increased, but the effect was smaller. Cationic complex 32 was found to bind the same anions an order of magnitude more strongly again, due to the added influence of the electrostatic attraction. With chloride, bromide and hydrogensulfate in acetonitrile/dichloromethane solution a significant anion-induced cathodic redox shift was observed in both the ferrocene/ferrocenium couple and the irreversible oxidation of the secondary metal centre.

Receptor 33 also incorporates a secondary Lewis acid anion binding site. This molecule is the zinc metallated analogue of 17 with the cobaltocenium reporter groups replaced with ferrocenes [27]. The freebase precursor to 33 in dichloromethane solution shows no significant anion induced shifts in the 1 H NMR signals of the amide protons, whereas the metalloporphyrin binds bromide (K = 6200 M–1 ), nitrate (K = 2300 M–1 )

Metal-Based Anion Receptor Systems

55

and hydrogenphosphate (K = 2100 M–1 ). Electrochemical studies in 3 : 2 dichloromethane/acetonitrile revealed anion-induced cathodic shifts in both the porphyrin (∆E = 85–115 mV) and tetraferrocene oxidation (∆E = 20–60 mV) waves. The trend in magnitude of ∆E for the porphyrin oxidation wave is hydrogensulfate > chloride > bromide > nitrate, reflecting the charge density (charge to radius ratio) of the anionic guest species. Atropisomers of 33 (other than the α,α,α,α-atropisomer) were also studied and showed dif-

56

S.R. Bayly · P.D. Beer

ferent anion-sensing properties. For instance the α,β,α,β-atropisomer was found to be selective for nitrate (∆Eporphyrin = 175, ∆Eferrocene = 110). Jurkschat and co-workers have investigated the redox anion-sensing properties of metallomacrocycle 34 [28]. This comprises two ferrocene reporter units linked together covalently by two Lewis acidic organotin spacers. Electrochemical measurements in dichloromethane solution showed anion-induced cathodic shifts in the ferrocene/ferrocenium redox couple of 130 mV for chloride, 210 mV for fluoride and 480 mV for dihydrogenphosphate. Another class of mixed-metal anion receptors has been investigated which possess redox reporter groups based on two different metal complexes. This enables the qualitative comparison of their comparative anion-sensing abilities. Macrocycles 35 and 36 combine the {RuII (bpy)3 } moiety with a bridging ferrocene or cobaltocenium unit [29]. Electrochemical experiments in acetonitrile solution revealed that the RuII /RuIII redox potential was insensitive to anion binding, whereas the ferrocene/ferrocenium (in 35) and cobaltocene/cobaltocenium (in 36) redox couples were shifted cathodically (by 60 mV and 110 mV respectively with chloride). However, the first reduction of {RuII (bpy)3 }, a ligand-centred process based on the amide substituted bipyridyl, was also found to undergo an anion induced cathodic shift (40 mV and 90 mV with chloride for 35 and 36, respectively). 2.4 Transition Metal Polypyridyl Anion Receptors Redox sensing of anions using {RuII (bpy)3 }-amides as a combined receptor/reporter system has also been studied using complexes 37–40 [30–32]. The single crystal X-ray structure of the chloride complex of 37 clearly indicates that the anion is bound tightly within the bipy amide ligand by six hydrogen bonds. It forms hydrogen bonds not only to the two N–H groups but also to four aromatic C–H groups with H–Cl distances ranging from 2.51 A. 1 H NMR titrations in DMSO-d6 revealed strong binding of chloride to 2.71 ˚ and dihydrogenphosphate. Four reversible redox couples (one metal-centred oxidation and three ligand centred reductions are expected for {RuII (bpy)3 } species) were observed in electrochemical studies. Of these only the least cathodic ligand reduction was significantly shifted in the presence of anionic guests. The assignment of this redox process to the relatively electron-poor amide-substituted bipyridyl reaffirms the X-ray structure evidence that anion recognition takes place at this site. It is interesting to note that the calix[4]arene modified molecule 40 which can be compared with the cobaltocenium and ferrocene complexes 7 and 18–23, shows particular selectivity for dihydrogenphosphate and is able to electrochemically sense this anion (giving a cathodic shift of 175 mV) in the presence of a 10-fold excess of chloride or hydrogensulfate [30].

Metal-Based Anion Receptor Systems

57

{RuII (bpy)3 } macrocycles 41–44 demonstrate the influence of the topology of the anion-binding site on redox sensing [29]. The macrocycles with the larger cavities 42–44 were shown by 1 H NMR studies in DMSO-d6 to bind chloride preferentially to acetate and dihydrogenphosphate. This is the opposite trend to 41, which has the smallest macrocyclic cavity and binds acetate more strongly than either chloride or dihydrogenphosphate. Receptor 43 displays outstanding selectivity; with a stability constant for chloride of 40 000 M–1 and no measurable affinity for dihydrogenphosphate. The presence of a second positively charged metal centre in 43 and 44 leads to increased overall anion affinity. The substitution of one Ru(II) centre for Os(II) in 44 results in both an increased stability constant for the chloride complex and a diminished affinity for acetate—in effect an increase in selectivity between the two anions. Results of electrochemical measurements on the complexes in acetonitrile solution confirmed the pattern of anion selectivity. In 41 the cathodic shift in the first bpy redox wave was 30 mV for chloride and 55 mV for acetate. In 43 and 44 cathodic shifts of 110 and 125 mV, respectively, were recorded in the presence of chloride, whereas for acetate the change was 15 and 20 mV, respectively. This class of anion receptor also exhibits optical anion-sensing properties (see Sect. 3.2). {RuII (bpy)3 } has also been used as the basis of a redox sensor for fluoride. Receptors 45 and 46 were studied in acetonitrile solution using differential

58

S.R. Bayly · P.D. Beer

pulse voltammetry. Addition of two equivalents of fluoride was found to produce a peak at 0.73 V vs. Ag/AgCl, ascribed to a ligand-centred redox process. Other anions, including halides, nitrate and hydrogensulfate caused no significant change in the electrochemistry [33]. A receptor based on {CoIII (bpy)3 } has been used by Sessler and coworkers for redox fluoride sensing [34]. In the cyclic voltammetry of 47 in DMSO solution addition of fluoride led to a complete disappearance of the Co(II)/(III) reduction wave. Addition of water to this solution restored this redox process, suggesting that the presence of a strongly bound fluoride anion renders the complex redox inactive. Chloride and dihydrogenphosphate produced cathodic shifts of 160 mV and 70 mV, respectively. It is proposed that

Metal-Based Anion Receptor Systems

59

the pyrrole NH protons of the quinoxaline phenanthroline ligand are made more acidic by the electron-withdrawing effect of the coordinated metal centre, thereby giving the complex an increased affinity for fluoride compared to the free ligand. 2.5 Transition Metal Dithiocarbamates as Receptors for Redox Anion Sensing Another interesting development has been the self-assembly of metallodithiocarbamate macrocyclic receptors for electrochemical anion sensing. The naphthyl-based Cu(II) macrocycle 48 displays a cathodic shift of 85 mV in the Cu(II)/(III) redox couple in the presence of dihydrogenphosphate or perrhenate, but gives no response to halides in acetonitrile solution [35]. It is proposed that this selectivity is controlled by cavity size. A related receptor incorporating thiourea and hydrogen-bond donor groups, 49, revealed

60

S.R. Bayly · P.D. Beer

cathodic shifts in the Cu(II)/(III) couple of up to 160 mV with hydrogenphosphate [36]. Cobalt (III) dithiocarbamate cryptands 50 and 51 also function as redox active anion sensors [37]. In dichloromethane solution the irreversible Co(IV)/Co(III) redox couple of the complexes was found to undergo significant anion-induced cathodic perturbation; up to a maximum of 125 mV for 51 with dihydrogenphosphate. 2.6 Dendrimers as Redox Anion Sensors Dendrimers have been investigated as a platform for enhanced anion sensing. Astruc and co-workers have synthesised dendrimers 52–54, incorporating up to 18 amido-ferrocene units. These multi-metallic multi-binding site receptors are able to electrochemically sense anions in dichloromethane solution [38]. In 54 the selectivity trend is dihydrogenphosphate > hydrogen-

Metal-Based Anion Receptor Systems

61

sulfate > chloride > nitrate. Evidence of a “dendritic effect” was observed in the redox response of the consecutive dendrimer generations in the presence of dihydrogenphosphate or hydrogensulfate. As the number of amido-ferrocene units is increased, the magnitude of the cathodic shift in

62

S.R. Bayly · P.D. Beer

the ferrocene/ferrocenium couple also increases. Stability constants for the hydrogensulfate complexes of 53 and 54 were reported to be 9390 and 216 900 M–1 , respectively. Kaifer and co-workers have studied an analogous series of dendrimers based on a commercial DSM polyamine core with 4, 8, 16 and 32 peripheral ferrocenyl urea groups as the anion-sensing component [39]. No stability constants are reported, but cathodic shifts in the redox response of the ferrocene/ferrocenium couple with various anions in DMSO show a similar selectivity trend to 52–54. In this case the data suggests two ferrocene urea arms are involved in binding a single dihydrogenphosphate anion. The dendritic effect was observed in the change from the first (4 ferrocene units) to second (8 ferrocene units) generation dendrimers. No further increase in response was seen in the third generation and the fourth generation dendrimer showed a decreased response, presumably due to steric crowding.

Metal-Based Anion Receptor Systems

63

Astruc and co-workers have also investigated five generations of pentamethyl-amidoferrocene dendrimers using the DSM polyamine core [40]. The pentamethyl-substituted ferrocene was chosen to overcome the irreversible electrochemistry and electrode adsorption observed with 52–54. In this series the dendritic effect seen in the electrochemistry DMF solution varied according to the anion studied. In changing from lower to higher dendrimer generations modest increases in the anion-induced cathodic shift of the ferrocene/ferrocenium couple were observed with dihydrogenphosphate, whereas with ATP2– anion binding progressed from weak to relatively strong. This is perhaps due to ATP2– adopting a 1 : 2 anion/ferrocene unit binding stoichiometry, whereas dihydrogenphosphate binds 1 : 1. 2.7 Surface Confined Redox Anion-Sensing Systems In a step towards the fabrication of prototype sensory devices organisation of redox-active anion receptors on to electrode surfaces is being exploited. Importantly, self-assembled monolayers or thin polymer films of metal-based receptors can generate an amplified response to anion binding akin to the dendritic effect and could potentially become the basis of robust anionsensing devices.

Beer and co-workers have investigated this concept using self-assembled monolayers of the 1,1 -bis(alkyl-N-amido)ferrocene 55 on gold electrodes [41]. The pendant disulfide groups serve to covalently anchor the receptor to the gold surface. In electrochemical experiments on 55 in acetonitrile/dichloromethane solution anion-induced cathodic shifts of the ferrocene/ferrocenium redox couple were observed for chloride (40 mV),

64

S.R. Bayly · P.D. Beer

bromide (20 mV) and dihydrogenphosphate (210 mV). When confined to a monolayer the anion-induced shifts measured were 100 mV for chloride, 30 mV for bromide and 300 mV for dihydrogenphosphate in the same solvent system—consistently greater than for the solution-phase receptor. This represents a significant “surface sensing amplification”. The modified electrodes were also able to selectively detect dihydrogenphosphate in the presence of a 100-fold excess of halide. In aqueous solution the selectivity of the system was altered, enabling the detection of the poorly hydrated anion perrhenate in the presence of dihydrogenphosphate. A number of groups have been exploring the anion-sensing properties of thin polymer films which incorporate metal-based receptors. Monomer 56 consists of a cobaltocenium amide redox signalling group with a polymerisable pyrrole unit. Thin films of the receptor were prepared by electropolymerisation on a platinum or carbon electrode [42]. In electrochemical experiments on 56 in acetonitrile solution significant anion-induced cathodic shifts of the cobaltocene/cobaltocenium redox couple were observed for dihydrogenphosphate (45 mV) and hydrogensulfate (20 mV) only. When confined to a polymer film the cathodic shifts were amplified: 210 mV for dihydrogenphosphate and 250 mV for hydrogensulfate. Chloride and bromide could also be detected, both giving shifts of 20 mV. Polymerisation of 56 as well as giving rise to surface sensing amplification also resulted in a change in selectivity from dihydrogenphosphate to hydrogensulfate. It was also found that film thickness influences the sensitivity of the sensor. Thin films (Γ = 1.8 × 10–9 mol cm–2 ) exhibited higher sensitivity to dihydrogenphosphate at low concentrations (< 50 µM), whereas thick films (Γ = 2.7 × 10–8 mol cm–2 ) extend the measurable concentration range to higher levels (up to 2 mM). Films of the analogous ferrocene monomer 57 have been studied by Moutet and co-workers [43]. Anion-induced shifts of the ferrocene/ferrocenium couple were measured in acetonitrile for hydrogensulfate (30 mV), ATP2– (180 mV) and dihydrogenphosphate (220 mV). Again this represents a surface sensing amplification. The same group has also explored the anion-sensing properties of viologen 58 in thin polymer films [44]. In aqueous solution poly-58 registered small anion-induced cathodic shifts of the ferrocene/ferrocenium redox couple with hydrogensulfate (20 mV), S2 O4 2– (10 mV), and ATP2– (35 mV). Metal-based receptors that are able to form self-assembled monolayers on planar electrodes can also be used to functionalise the surface of nanoparticles, leading to a surface sensing amplification effect. The very large surface area of nanoparticles may also allow greater overall sensitivity to anions. Astruc and co-workers have prepared the amidoferrocenylalkylthiol (AFAT)-gold nanoparticle system 59 [45]. The proportion of AFAT to dodecanethiol obtained by ligand substitution on different batches of dodecanethiol stabilised nanoparticles ranged from 7–38%, corresponding to an

Metal-Based Anion Receptor Systems

65

average of 8–39 AFAT units per nanoparticle. Electrochemical measurements in dichloromethane solution show a single reversible redox wave for the ferrocene/ferrocenium couple at identical potential in each case. Addition of dihydrogenphosphate led to the appearance of a new redox wave (220 mV cathodically shifted) with the attenuation of the initial wave, which was completely replaced at 1 equivalent of anion per AFAT branch, indicating of 1 : 1 anion/branch binding. The cathodic shift is the same irrespective of the AFAT loading and is considerably larger than observed for the comparable amidoferrocene monomer FcCONHCH2 CH2 OPh (45 mV) or even a representative ferrocene tripod PhC(CH2 CH2 CH2 NHCOFc)3 (110 mV). The same group has also investigated the anion-sensing properties of gold nanoparticles 60 and 61 substituted with dendrons comprising three amidoferrocene or silyl ferrocene branches [40]. The surface loadings of 60 and 61 were 3% and 4.8% respectively, corresponding to an average 3 and 5 dendrons per nanoparticle. Nanoparticles of type 60 show very similar properties to the AFAT-modified nanoparticles 59, with a dihydrogenphosphateinduced cathodic shift of 210 mV in dichloromethane solution. Despite lacking any hydrogen-bonding groups the nanoparticles dendronised with silyl

66

S.R. Bayly · P.D. Beer

ferrocenes, 61, gave a dihydrogenphosphate-induced cathodic shift of 110 mV in the same solvent. A highly ambitious multicomponent surface-anchored anion-sensing rotaxane assembly has recently been described [46]. This comprises two receptor molecules, 62—an isophthalamide macrocycle with an exocyclic ferrocene reporter group, and 63 —a cationic pyridinium amide thread bearing a disulfide tether for SAM formation at one terminus, and a pentaphenylferrocene at the other as a combined redox reporter and bulky stopper group. In low polarity solvents such as dichloromethane chloride is bound simultaneously by both receptors, causing the threading of 63 into the annulus of 62 to form a pseudorotaxane (Fig. 1). Adsorption of the pseudorotaxane onto a clean gold surface to form a SAM causes the components to be locked

Metal-Based Anion Receptor Systems

67

together as a rotaxane. This rotaxane SAM shows a remarkable selective electrochemical response to anions compared to 62 or 63 alone. The addition of molar excesses of chloride, dihydrogenphosphate or hydrogensulfate to 63 in acetonitrile solution or as simple SAMs resulted in only a small cathodic shift (∆E < 10 mV) of the pentaphenylferrocene redox couple. The macrocycle 62 was more responsive to anions, undergoing a cathodic shift in the ferrocene redox couple of 45 mV with dihydrogenphosphate, 15 mV with hydrogensulfate and < 10 mV with chloride. In the rotaxane SAM the pentaphenylferrocene centre of 63 exhibits redox responses to chloride and oxoanions broadly similar to SAMs of this receptor on its own. In contrast, within the surface assembled rotaxane the ferrocene of the macrocyclic component exhibits a markedly greater electrochemical cathodic response to chloride

Fig. 1 Self-assembly of anion templated rotaxane SAM (RC = redox centre). From [46], reproduced by permission of The Royal Society of Chemistry

68

S.R. Bayly · P.D. Beer

(40±5 mV), but gives no significant response to the oxoanions tested. This suggests that chloride binding inside the interlocked cavity of the surface confined rotaxane results in a conformation where the macrocycle’s pendant ferrocene group is in proximity to the complexed halide anion, whereas the oxoanions are too large to penetrate the rotaxane binding pocket. Preliminary electrochemical competition experiments in acetonitrile solutions revealed that these rotaxane SAMs are capable of selectively detecting chloride in the presence of 100-fold excess amounts of dihydrogenphosphate and exhibit an appreciably greater detection sensitivity than that shown by the free macrocycle. The superior electrochemical response of rotaxane SAMs to chloride over dihydrogenphosphate mirrors the high degree of chloride anion selectivity of previous rotaxanes prepared via chloride anion templation.

3 Optical Anion Sensors 3.1 Theory of Optical Anion Sensing Optical reporter groups signal anion binding through a change in their electronic absorption or emission spectra. The precise nature of the response in the UV/vis absorption spectrum will largely depend on the energy differences between the molecular orbitals of the receptor before and after anion binding. For changes of any magnitude to be observed the anion-binding site must be strongly coupled to the metal centre. In the case of metal-based transitions such as d-d transitions this typically requires the anion to bind directly to the metal centre. In the case of MLCT or LMCT transitions it is advantageous if the anion-binding site is π-conjugated to the ligand involved. Luminescence spectroscopy can be a more sensitive technique for probing anion binding to metal-based receptors. In addition to altering the energy of the emission maxima, anion binding often causes significant changes in their intensity. The intrinsic luminescence of a particular reporter group can be “switched off” if the anion : receptor complex provides a more efficient pathway for non-radiative energy loss. Similarly, in systems where the luminophore is quenched by a nearby functional group, anion binding can “switch on” the luminescent emission by blocking this non-radiative decay process. 3.2 Metallocene Optical Anion Sensors The common reporter groups cobaltocenium and ferrocene have not frequently been used in optical anion sensing, since these chromophores are generally insensitive to anion binding. However, metallocene-based receptors

Metal-Based Anion Receptor Systems

69

that incorporate a suitable organic chromophore or luminophore have been shown to operate as combined optical and electrochemical anion sensors. For instance the tetra-cobaltocenium porphyrin 17, exhibits the same selectivity trend (Cl– > Br–
NO3 – ) in UV-vis anion-binding experiments that was observed by electrochemistry [19]. In acetonitrile solution the Soret band (λmax = 425 nm, due to the porphyrin) of 17 was significantly bathochromi-

70

S.R. Bayly · P.D. Beer

cally shifted on addition of dihydrogenphosphate (∆λmax = 15 nm), hypsochromically shifted with C1– (∆λmax = 10 nm) and split into two maxima (λ = 430,440 nm) with HSO4 – . The novel series of ferrocene receptors 64–68, which incorporate thiobarbiturate anion binding/chromophore groups, have been shown to operate as UV-vis anion sensors in acetonitrile solution. Addition of basic anions such as cyanide, acetate and benzoate causes the attenuation of the absorption maximum at around 440 nm (due to a charge transfer transition between the amine and the thiobarbiturate group), with the simultaneous occurrence of a new band at 370 nm. In cases where the titration data could be fitted, 1 : 2 receptor : anion binding stoichiometries were found [47].

Another interesting example of a ferrocene-based optical sensor is 69, which acts as a chromogenic molecular switch [48]. Appended to one cyclopentadienyl ring of the ferrocene of molecule 69 is a p-nitrophenyl urea unit which acts as a combined anion-binding site and chromophore. A crown ether is attached to the other cyclopentadienyl ring for cation binding. Tucker and co-workers reported that on addition of fluoride to a solution of 69 in acetonitrile a significant perturbation of the UV-vis spectrum was observed including the appearance of a new absorption at 472 nm. The Ka for the 1 : 1 anion-receptor complex was determined as 9340 M–1 . The addition of 10 equivalents of KPF6 to the solution of 69 containing 10 equivalents of fluoride caused the complete disappearance of the 475 nm absorption. Ka of the receptor with K+ in the presence of fluoride was calculated as 1460 M–1 . Surprisingly model receptor 70, which lacks the crown ether moiety, exhibited similar switching properties. Ka of this receptor for fluoride is equivalent (9660 M–1 ) but the Ka for K+ in the presence of fluoride is far lower (230 M–1 ). Inhibition by K+ of the response of these receptors to fluoride is therefore thought to be due to the ion-pairing interaction between fluoride and K+ . Aldridge et al. have demonstrated that boryl-ferrocene 71 can be used as a selective colourimetric sensor for fluoride [49]. When fluoride was added to a CH2 Cl2 solution of 71 under aerobic conditions a colour change from orange to pale green was observed. This did not occur with any other anion tested. Spectroscopic and electrochemical measurements suggest that com-

Metal-Based Anion Receptor Systems

71

plexation of fluoride causes the spontaneous formation of a ferrocenium species, i.e. the 150 mV anodic shift in the ferrocene/ferrocenium redox potential caused by fluoride complexation reduces the redox potential enough for the 71 : 2F– complex to be oxidised by atmospheric O2 . The pale green colour is due to the characteristic absorption of ferrocenium. In this case the optical response is the direct result of an anion-induced redox process and does not require an additional chromogenic group.

Luminescence sensing of anions has also been achieved using ferrocene receptors. Example 72 uses amide groups for anion binding in conjunction with naphthalene groups to provide the fluorescence signal [50]. Addition of fluoride to a DMSO solution of 72 led to a 3-fold enhancement (at 5 equivalents) of the intramolecular naphthalene-naphthalene excimer emission at 492 nm. Dihydrogenphosphate also generated a significant response, causing a 2-fold enhancement at 5 equivalents. In electrochemical studies in DMF electrolyte solution fluoride generated a 120 mV cathodic shift in the redox potential. The receptor 73, based on an azaferrocenophane structure bearing two urea groups as linkers between the redox active (ferrocene) and fluorescent (naphthalene) signalling subunits, also shows both fluorescent and electrochemical sensing of fluoride [51]. On addition of excess fluoride it displays an enhancement factor of 13 in the naphthalene emission bands at 362 and 380 nm in DMF solution and a cathodic shift of the ferrocene/ferrocenium couple of 190 mV in DMSO electrolyte solution.

72

S.R. Bayly · P.D. Beer

The unusual new [3,3]ferrocenophane 74·H+ also acts as a selective fluorescent anion sensor—in this case for nitrate [52]. The protonated receptor is weakly fluorescent (Φ = 0.043) in CH2 Cl2 solution and on addition of nitrate the naphthalene-based emission at 354 nm is quenched (to Φ = 0.020). Addition of acetate, hydrogensulfate and dihydrogenphosphate merely induced deprotonation of the receptor. This receptor is also able to act as a redox sensor for other anions and as a fluorescent sensor for group II cations. 3.3 Ruthenium(II) Polypyridyl Complexes as Optical Anion Sensors The spectroscopic and redox properties of {RuII (bpy)3 } have allowed this metal complex to be used for combined optical and electrochemical sensing of anions without the need for additional chromophores or luminophores.

For example Sessler’s complex 75 gives a UV-vis response to fluoride in DMSO solution, with a stability constant for the receptor : fluoride complex of 12 000 M–1 [34]. The new {RuII (bpy)3 }-pyrrole 76 has also been found to selectively sense fluoride in DMSO solution by UV/vis (Ka = 7000 M–1 ) [53].

Metal-Based Anion Receptor Systems

73

The luminescent emission of {RuII (bpy)3 } is also very useful for signalling anion binding. In the emission spectra of 37–40 both a blue shift (of up to 16 nm for 40) and an increase in intensity of the λmax of the MLCT emission band was observed on addition of dihydrogenphosphate. It has been proposed that the conformational flexibility of the receptors is decreased by complexation of the anion guest thus reducing the rate of non-radiative decay through vibrational and rotational relaxation. Similarly, macrocyclic complexes 41–44 and 77–79 were also found to sense chloride by luminescence enhancement.

In other examples anion binding can cause quenching of the {RuII (bpy)3 } MLCT emission. In aqueous solution polyaza receptors 80-82 bind phosphate and ATP anions, producing up to a 15% reduction in the emission intensity of λmax at 605 nm [16]. Similarly, 76 shows up to a 40% reduction in the intensity of the luminescent emission at 630 nm in the presence of dihydrogenphosphate in DMSO solution. The RuII bipyridylcalix[4]diquinone receptor 83 selectively binds and senses acetate anions (from 1 H NMR titrations in DMSO-d6 solution K = 9990 M–1 ) [54]. This receptor is only weakly luminescent because the {RuII (bpy)3 } MLCT emission is partially quenched by oxidative electron transfer to the electron-poor calix[4]diquinone. Addition of acetate to acetonitrile solutions of 83 resulted in a five-fold increase in luminescence intensity (60% for chloride) concomitant with a slight blue shift of the emission maximum. Anion binding causes this increase in emission intensity by interrupting the electron transfer pathway from the {RuII (bpy)3 } to the calix[4]diquinone, thus reducing its quenching effect. A similar effect is seen in macrocycle 36 which incorporates the {RuII (bpy)3 } moiety with a bridging cobaltocenium unit [29]. In acetoni-

74

S.R. Bayly · P.D. Beer

trile solution the quantum yield of the {RuII (bpy)3 } emission of this complex is relatively low. However, in the presence of chloride a 100% increase in emission intensity is observed. Complexes 45 and 46 are capable of selectively sensing fluoride in acetonitrile solution both by UV-vis and luminescence spectroscopy [33]. This anion causes a dramatic reduction in the intensity of the MLCT absorption in the 350–450 nm range with a new absorption appearing in the 500–650 nm range ascribed to a ligand-based CT process. The emission spectra of the complexes (exciting at 465 nm) show no significant peaks, indicating that the characteristic Ru-centred luminescence is quenched by the dinitrophenylhydrazone group. Upon addition of fluoride a strong peak at 625 nm develops, with quantum yields for the 45 : F– and 46 : F– adducts of 8.0 × 10–5 and 4.0 × 10–4 , respectively. Again it is apparent that anion binding interrupts the non-radiative decay pathway.

Deetz and Smith have prepared a heteroditopic {RuII (bpy)3 } receptor 84 incorporating both amide and boronic acid groups which selectively senses certain phosphorylated sugars in aqueous solution [55]. Boronic acids are known to form covalent complexes with the diol groups of saccharides, whereas the adjacent amides are positioned to complex the anionic phosphate component. Sensing was accomplished by measuring luminescence enhancement, with fructose-6-phosphate generating the highest stability constant (log Ka = 3.1). Non-phosphorylated saccharides gave much smaller changes in emission intensity (log Ka < 1.2), showing that the anionic component of the guest is essential for strong binding. A covalent attachment between the anion and the saccharide is not required. In the presence of sodium phosphate buffer non-phosphorylated saccharides are bound with similar strength to their phosphorylated counterparts. It is reported that this apparent cooperativity is a result of favourable hydrogen bonding between the phosphate anion and the saccharide. Watanabe and co-workers have also shown that anionic and neutral phosphodiesters can be sensed in acetone by the imidazole functionalised receptor 85 by both UV-vis and luminescence spectroscopies [56].

Metal-Based Anion Receptor Systems

75

3.4 Optical Anion Sensing Using Reporter Groups Based on Other Transition-Metal Complexes Zn(II) porphyrins are another class of complex which can operate both as UV/vis and luminescent sensors for anions. Example 86 is a picket-fence porphyrin with four imidazolium anion-binding groups [57]. In DMSO solution this receptor undergoes slight (5 nm) bathochromic shifts in the Q-bands of the absorption spectrum on addition of anions—indicative of the anion binding directly to the Zn centre. Using UV/vis titrations in DMSO solution 86 was found to be selective for hydrogensulfate over chloride and dihydrogenphosphate. In water : DMSO (5 : 95) solution the receptor was found to be selective for sulfate and hydrogensulfate over ATP2– and dihydrogenphosphate. Under similar conditions the Q bands in the luminescence emission spectrum of 86 (λex = 424 nm) are also bathochromically shifted with a concomitant decrease in intensity. It should be noted that 86 also functions as an electrochemical anion sensor, where the Zn(II) porphyrin-based oxidation potential is sensitive to

76

S.R. Bayly · P.D. Beer

anion binding. In acetonitrile solution a greater cathodic shift was observed with chloride (175 mV) than hydrogensulfate (140 mV) or nitrate (95 mV). This demonstrates that the anion selectivity of sensor systems is dependent upon the mode of detection used and underlines the fact that optical and electrochemical anion sensing operate by different mechanisms. Bis-terpyridine Iridium(III) has been used as an optical reporter group in anion sensing. In aqueous solution isomers 87 and 88 both exhibit halideinduced luminescence quenching with selectivity for chloride [58]. Receptor 87, although less sensitive than 88, is reported to possess good characteristics for sensing chloride at physiologically relevant concentrations. A related series of novel cyclometallated iridium(III) polypyridine thiourea complexes also display anion-induced luminescence quenching. The representative complex 89 was tested in acetonitrile solution with fluoride, acetate and dihydrogenphosphate, and gave log K values for the 1 : 1 complex of 3.38,

Metal-Based Anion Receptor Systems

77

4.03 and 3.14, respectively. This selectivity trend is ascribed to the combined effect of the basicity and geometry of the guest anions [59]. Complexes of rhenium(I)tricarbonylchloride with pyridyl ligands are luminophores and hence in anion sensing have principally been used as reporter groups. For instance calix[4]diquinone receptor 90 selectively binds and senses acetate in DMSO solution [54] (from 1 H NMR titrations K = 1790 M–1 in DMSO-d6 solution). The receptor exhibits relatively weak luminescence because calix[4]diquinone is an electron acceptor, quenching the Re(I) bipyridyl emission by oxidative electron transfer. Addition of anions to DMSO solutions of 90 resulted in a significant increase in luminescence intensity. It is clear that the presence of the anion in the binding pocket between the {ReI (bpy)} moiety and the quencher interrupts the oxidative electrontransfer process. The mixed Re(I)/Pd(II) molecular square 91 has been found to sense perchlorate in acetone (giving K = 900 M–1 ) by enhancement of the Re(I) luminescent emission [60]. In this case luminescence quenching by oxidative transfer to the Pd(II) ion is inhibited by the bound anion. The Pd(II) ion also plays a role as a structural element and charge carrier. Squares 92–96 are very similar, but incorporate a bis-phosphinylferrocene supporting ligand [61]. Again the normally strong luminescence of the Re(I) component is partially quenched by the bimetallic corners. Binding studies of the squares with different inorganic anions were carried out by luminescence titrations in acetone solution. Of the anions investigated, only hexafluorophosphate and tetrafluoroborate induced significant changes in luminescence. As these anions were added an initial decrease in emission intensity was followed by an increase to a plateau. This is taken to indicate the presence of two competing quenching pathways which are inhibited to different extents by anion binding. Lees and co-workers have investigated Re(I) bipyridyl anion hosts based on aryl bisamide skeletons 97–99 [62]. Measurement of anion-induced luminescence quenching in CH2 Cl2 showed 97 to have strong binding affinities for halides, acetate and cyanide, weaker affinity for dihydrogenphosphate, and even less affinity for nitrate and perchlorate. The iso- and terephthalamide receptors 98 and 99 possess smaller stability constants for all the anions tested. It is proposed that the anion-sensing efficiency of 98 is due to intramolecular hydrogen-bonding of the amido NH proton to the pyridyl nitrogen holding the receptor in a “cleft” conformation. A metal-templated approach has been used by Thomas and co-workers to produce the Re(I) metallomacrocycle 100 [63]. In acetonitrile solution this receptor displayed luminescence enhancement on addition of anions. Stability constants for the 1 : 1 adducts with BF4 – (1575 M–1 ), SO4 2– (7135 M–1 ) and BPh4 – (2895 M–1 ) were determined. Since SO4 2– is of similar size to BF4 – the comparatively high affinity of this anion for 100 is thought to be due to its additional charge. The slightly higher stability of the 100 : BPh4 –

78

S.R. Bayly · P.D. Beer

complex compared to 100 : BF4 – is attributed to π–π interactions with the receptor. Interlocked supramolecular assemblies such as rotaxanes and catenanes have the potential to provide tailor-made binding cavities for guest species. An example of this is 101, a rotaxane formed using anion templated synthesis which incorporates the Re(I) bipyridyl fragment [64]. Addition of chloride, hydrogensulfate or nitrate to the receptor in acetone solution caused an enhancement of the fluorescence emission. Curiously, although the rotaxane was formed using chloride as the template, it was found to be selective for hydrogensulfate with which a stability constant of > 106 M–1 was determined.

Metal-Based Anion Receptor Systems

79

3.5 Transition Metals as Anion-Binding Groups and/or Structural Components Complexes of the late transition metals are well studied in anion sensing, much of the work has been pioneered by Fabbrizzi and co-workers. As well as providing an optical signalling function the Lewis acid metal ion can act as a binding site for the anion. In addition the metal ion often forms an organisational unit designed to create a receptor of a specific shape. In order to harness the metal-anion interaction for anion sensing the binding properties of the metal must be modulated by an ancillary ligand. In this way one or two vacant coordination sites can be made available for anion binding, and other elements appended to allow signalling or modified selectivity. In complexes with simple tripodal amines such as 102 and 103 zinc(II) forms five-coordinate metal complexes of trigonal bipyramidal geometry, leaving one of the axial coordination sites of the metal available for anion binding. The Zn(II) complex of 102 was found to undergo quenching (by photoinduced electron transfer) of the anthracene fluorescent emission in the presence of aromatic carboxylate anions such as 4-N,Ndimethylaminebenzoate in ethanol solution [65]. Complete quenching was observed at the 1 : 1 anion to receptor ratio (log K = 5.45). Likewise the Zn(II) complex of 103 was found to form 1 : 1 adducts with carboxylate anions in methanol solution, with log K values ranging from 4 to 5 [66]. Only aromatic carboxylates induced quenching of the ligand luminescence.

80

S.R. Bayly · P.D. Beer

The bis(boradiazaindacene) substituted bipyridine ligand 104 is highly fluorescent in organic solvents whereas its Zn(II) complex is not [67]. It was found that the complex progressively regained its fluorescent emission when it was titrated with various anions in acetonitrile. Stability constants

Metal-Based Anion Receptor Systems

81

for the anion:receptor complexes were calculated for fluoride (4160 M–1 ), chloride (3230 M–1 ), bromide (2500 M–1 ), acetate (4760 M–1 ), and phosphate (4000 M–1 ). Decomplexation of the chelated Zn(II) ion from 104 by the weakly coordinating anions was ruled out as the sensing mechanism. It is proposed that the quenching of the ligand luminescence by electron transfer to the Zn(II) centre is inhibited by anion coordination. Fabbrizzi and co-workers have demonstrated the use of bis-copper(II) cryptates to sense ambidentate anions [68]. On titrating molecule 105 with NaN3 in aqueous solution the colour changed from pale blue to bright green and an anion-metal LMCT absorption appeared at 400 nm. X-ray diffraction studies have shown that the azide anion is held colinear with the two Cu(II) centres, coordinated through the two terminal sp2 hybridised nitrogen atoms. Stability constants for 105 with a variety of anions in aqueous solution were calculated and the selectivity of this anion sensor for the azide anion was found to be determined by the bite distance between the two copper atoms. Cryptate 106, in which the aryl spacer of 107 is replaced with a furanyl unit, acts as a colourimetric sensor for anions. UV/vis titrations in aqueous solution gave log K values for the 1 : 1 halide/receptor adducts of 3.98 for chloride, 3.01 for bromide and 2.39 for iodide. X-ray diffraction studies confirm that bromide is held between the two copper atoms. Under the same conditions 106 also interacts strongly with azide (log K = 4.7) and thiocyanate (log K = 4.28) anions. This receptor is interesting because of its lack of selectivity compared to 105. The complex appears to be able to expand and contract its bite-length in order to accommodate anions of various dimensions. The use of this class of receptors in practical applications is limited by the small changes in UV-vis absorption which indicate anion binding. To overcome this problem of sensitivity a chemosensing ensemble approach has been applied. The fluorescent indicator coumarine 343 carries a carboxylate group which allows it to be bound by 105 in a 1 : 1 complex (log K = 4.8) with complete quenching of the luminescent emission [69]. Titration of a solution containing 0.2 mM 105 and 0.1 µM coumarine 343 with hydrogencarbonate, azide or cyanate anions resulted in complete recovery of the indicator luminescence. Anions with a lower affinity for the receptor were unable to displace the coumarine 343 and produced only a slight luminescence enhancement. The usefulness of this chemosensing ensemble was demonstrated by the quantitative determination of carbonate in mineral water. Using the same principle Han and Kim have recently reported a chemosensing ensemble made up of the dizinc complex 107 and pyrocatechol violet which is selective for phosphate [70]. Anslyn and co-workers have developed a series of tripodal Cu(II) complexes 108 and 109 in which the metal ion and three cationic organic groups form a tetrahedral cavity designed to host phosphate [71]. Receptor 108 is built from the tris(2-ethylamino)amine skeleton with appended benzylamine

82

S.R. Bayly · P.D. Beer

groups. UV/vis anion titrations were carried out in aqueous solution at pH 7.4 where it can be assumed the terminal amines are all protonated. Hydrogenphosphate and its congener hydrogenarsenate were found to bind strongly in a 1 : 1 anion/host ratio, both with a log K value of 4.40. Perrhenate was bound an order of magnitude less strongly and the affinity for chloride was too small to measure. Model complex 110, which has no ammonium groups, gave a log K value of 2.95 with hydrogenphosphate, indicating that the Cu(II)anion interaction contributes significantly to anion binding. Receptor 109 follows the same design principle, this time incorporating guanidinium binding units with a tris-[(2-pyridyl)methyl]amine skeleton. log K values for hydrogenphosphate and hydrogenarsenate were found to be 4.18 and 4.23, respectively. Other anions, including perrhenate had no significant affinity for 109. It is apparent that the guanidinium groups are responsible for the improved selectivity of this receptor for phosphate. In a separate study the driving force for hydrogenphosphate binding was found to be entropic for receptor 108, but enthalpic for receptor 109 [72]. Partnered with the colourimetric indicator 5-(and 6)-carboxyfluorescein receptor 109 provides an effective chemosensing ensemble for the determination of inorganic phosphate in serum and saliva [73]. Zinc has been used as a binding site for the detection of pyrophosphate in aqueous solution by fluorescence. Complex 111 couples two tridentate Zn centres to a fluorescent naphthalenediimide core. In HEPES buffer the appearance of a new emission band at 490 nm was observed on addition of pyrophosphate, attributed to naphthalenediimide excimer formation. This did not occur with other anions including halides, acetate and ADP. It is proposed that 111 binds pyrophosphate in a 2 + 2 complex, which brings two naphthalenediimides in close enough proximity to give the excimer emission [74]. The recent tripodal Cu(II) complex 112 has intriguing optical anionsensing properties [75]. This receptor has a cavity with two distinct anionbinding sites—the vacant site on the Lewis acidic Cu(II) centre, and the three favourably arranged nitrophenylurea fragments. On titration with up

Metal-Based Anion Receptor Systems

83

to one equivalent of azide or dihydrogen phosphate in DMSO solution the Cu(II) d–d bands in the region 600–900 nm increased markedly in intensity. This indicates that the anion is binding at the metal centre. Upon addition of a second equivalent of the anion the absorption associated with the nitrophenylurea groups (below 500 nm) increased in intensity, showing that these

84

S.R. Bayly · P.D. Beer

hydrogen-bonding units are involved in binding the second anion. On their own halide anions were found to give only 1 : 1 complexes (with the halide bound at the Cu(II) centre). However, on titration of dihydrogenphosphate into a solution of the receptor pre-saturated with chloride (i.e. dissolved in a solution containing a 150-fold excess of chloride) formation of a 1 : 1 adduct of the [112 : Cl] complex with dihydrogenphosphate was observed. The stability constant of this species was found to be approximately 700 times higher than the stability constant of the analogous [112 : H2 PO4 ] complex with a second equivalent of dihydrogenphosphate. Stepwise anion coordination equilibria are also observed in the Cu(II) complexes of ligands 113 and 114 [76]. UV/vis titrations in acetonitrile solution show that each Cu(II) complex binds two anions (chloride, bromide, iodide, nitrate or thiocyanate), the first at the Cu(II) centre and the second in the bis-imidazolium compartment. The Cu(I) complexes of these ligands are able to host only one nitrate anion (in the bis-imidazolium cavity), while other anions induce demetallation. Cyclic voltammetry and spectroelectrochemical experiments showed that in the presence of one equivalent of nitrate the Cu(II)/Cu(I) redox change causes the anion to translocate quickly and reversibly from the metal-based binding site in the Cu(II) complex to the imidazolium binding site in the Cu(I) system. Another noteworthy example in which Cu(I) forms the basis of an optical anion sensor is 115, in which the metal complex acts both as a UV/vis signalling group and as a structural component dictating the topology of the urea anion-binding site [77]. The MLCT band within the CuI (phenanthroline) complex at 282 nm is sensitive to halide ions, acetate and dihydrogenphosphate in 4 : 1 v/v THF/MeCN (a relatively low polarity solvent). However, in DMSO solution, only acetate and dihydrogenphosphate produced a UV/vis re-

Metal-Based Anion Receptor Systems

85

sponse. Crucially, dihydrogenphosphate was found to bind the receptor with 1 : 1 stoichiometry, whereas acetate bound 2 : 1 anion : receptor. It is proposed that dihydrogenphosphate is able to bind simultaneously to the two urea groups to form an overall helical structure. This is not the case with the free ligand—the CuI centre is required to template the formation of the binding cavity, and augments dihydrogenphosphate binding. Acetate can only hydrogen-bond to one urea subunit at a time, and the affinity of 115 for this anion is only slightly higher than that of the free ligand (presumably due to electrostatic effects). Cadmium(II), the heavier congener of zinc(II) can also act as a coordination site for anion binding. Mizukami et al. employ a novel approach

Fig. 2 X-ray crystal structure of bromide encapsulated in the Fe : 117 complex. Reproduced with permission from [79]. © Wiley-VCH Verlag GmbH & Co. KGaA

86

S.R. Bayly · P.D. Beer

with receptor 116, by covalently attaching the coumarin indicator group of a chemosensing ensemble to a macrocyclic Cd(II) anion receptor [78]. It is designed such that anions will displace the nitrogen of the aromatic fluorophore from the coordination sphere of the Cd(II) centre. In aqueous solution as the receptor was titrated with pyrophosphate the luminescent emission of the molecule was observed to shift gradually from 342 nm to 383 nm. Receptor 116 shows a high degree of selectivity for pyrophosphate, citrate, ATP and ADP with log K’s in the range 4–5. It is worthy of note that the Zn(II) analogue of 116 was found to be ineffective for anion sensing. Iron(II) has been used as a supramolecular template for the formation of a tris-imidazolium receptor from ligand 117 [79]. 1 H NMR studies and X-ray crystal structure determination were used to demonstrate the encapsulation of bromide in the cavity of the receptor, with the anion coordinated by six C–H fragments (Fig. 2). Spectrophotometric titrations in acetonitrile solution revealed that this receptor binds halides with selectivity for chloride > bromide > iodide, but has no affinity for dihydrogenphosphate or hydrogensulfate. Presumably the restricted size of the receptor cavity excludes the binding of these larger tetrahedral anions. The linear anions azide, cyanate and thiocyanate also produced a response in the UV/vis spectrum, and azide was found to bind preferentially to 117 in comparison to the nonsymmetrical linear anions. 3.6 Optical Anions Sensing by Lanthanide(III) Complexes Given their favourable luminescence properties and propensity to coordinate oxy-anions it is perhaps surprising that complexes of lanthanide(III) ions have received comparatively little attention as anion sensors. Parker and co-workers have reported a novel method for the selective detection of carboxy anions by time-delayed luminescence using the Eu(III) and Tb(III) complexes of 118 [80]. Luminescence measurements on the coordinatively unsaturated complexes in aqueous solution showed significant increases in lifetime and emission intensity in the presence of anions. This behaviour is consistent with the anions displacing water from the non-ligated coordination sites at the metal centre. Studies allowed the number of water molecules (q) remaining coordinated in the presence of added anions to be estimated. For the triflate salt of the Eu(III) receptor q = 2.14 whereas with hydrogencarbonate q = 0.34 and with hydrogenphosphate q = 0.74. The selectivity for hydrogencarbonate reflects the ability of this anion to chelate the Eu(III) centre whereas hydrogenphosphate prefers to bind in a monodentate fashion. This behaviour is also pH-dependent (pKa HCO3 – /H2 CO3 – = 6.38) so that for a given starting bicarbonate concentration, a pH-dependent lifetime and emission intensity was observed [81].

Metal-Based Anion Receptor Systems

87

Other groups have subsequently reported anion receptors that work on the same principle. For instance an Eu(III) complex of the bis-bipyridinephenylphosphine oxide ligand 119 made by Ziessel and co-workers is able to sense anions by luminescence enhancement in acetonitrile with stability constants which follow the trend fluoride > acetate > chloride > nitrate [82]. Tsukube and co-workers have investigated the properties of the Eu(III) and Tb(III) complexes of the chiral ligand 120 [83]. Anion binding was assessed by profiling luminescence enhancement in acetonitrile and it was found that the different metal centres provided different selectivities. The emission at 548 nm

88

S.R. Bayly · P.D. Beer

of the Tb(III) complex was increased by 5.5 times in the presence of 3 equivalents of chloride compared to 2.2 for nitrate and 1.1 for acetate. Conversely the emission at 618 nm of the Eu(III) complex was increased 8.3 times by 3 equivalents of nitrate, 2.5 times for chloride and 1.0 times for acetate. Stability constants were not reported. The same group most recently reported the use of neutral lanthanide(III) tris-diketonates of type 121 for the determination of chloride [84]. The response in luminescence of the Eu(III) complex for chloride in acetonitrile solution was large enough to be seen by the naked eye. Incorporation of the complexes in PVC membrane electrodes allowed measurement of potentiometric selectivity coefficients. These showed the Eu(III) complex to be the more selective for chloride than the Pr(III), Dy(III) or Yb(III) analogues. Gunnlaugsson and co-workers have developed an anion sensor based on a ternary europium complex [85]. In 122 the naphthalene β-diketonato ligand acts as a sensitising group for Eu(III) emission. Displacement of this antenna group from 122 by competitor anionic species would therefore be expected to decrease the intensity of this emission. In aqueous solution at pH 7.4 it was found that iodide and dihydrogenphosphate reduced the intensity of the emission at 616 nm by 20–40%. More pronounced changes, which could be seen with the naked eye, were observed with highly competitive anions such as tartarate and fluoride. Gunnlaugsson and co-workers have also studied di-europium(III) complex 123, which incorporates two different macrocyclic ligands for Eu(III) in addition to a covalently bound antenna group [86]. Upon titrating 123 with acetate, aspartate and succinate at pH 6.5, each of the Eu(III) emission bands was quenched by up to 50% for acetate and aspartate. Malonate, however, produced a nearly two-fold enhancement in the emission intensity. It is thought that this particular anion is able to displace coordinated water molecules from both the Eu(III) centres thereby reducing the rate of non-radiative energy transfer. 3.7 Surface Confined Systems for Optical Anion Sensing Beer and co-workers have developed a surface-enhanced optical anion sensor based on gold nanoparticles [87]. Dodecanethiol stabilised gold nanoparticles were modified by ligand substitution with a disulfide-substituted zinc porphyrin 124 to provide 30 and 80 receptors per nanoparticle. Titration of both the free receptor and the modified nanoparticles with various anions in dichloromethane or DMSO solution revealed significant changes in the intense porphyrin absorption bands. Calculated stability constants are given in Table 1 and reveal highly enhanced anion-binding affinities (up to two orders of magnitude with chloride and dihydrogenphosphate in DMSO solution) for

Metal-Based Anion Receptor Systems

89

Table 1 Association constant (log K) data of 124 and 124 modified nanoparticles in DMSO solution determined at 293 K, errors ±0.1 Anion

124

124-nanoparticles a

Cl– H2 PO4 –

<2 2.5

4.3 4.1

a Association constant values for the 1 : 1 porphyrin–anion complex on the nanoparticle surface

the surface-bound porphyrin receptor with respect to the free metalloporphyrin.

4 Metal-Based Anion Receptors Without Reporter Groups Described below are a number of interesting examples from the literature of metal-based anion receptors for which anion binding has only been studied by NMR. If one of the cyclopentadienyl rings of ferrocene is replaced with an arene moiety a cationic complex is generated, thereby gaining the potential for electrostatic interaction with anions. Beer and co-workers have exploited this principle in the simple receptor 125 [88]. Similarly, Atwood and co-workers have derivatised cyclotriveratrylene to generate receptor 126 [89]. Qualitative 1 H NMR studies demonstrate that 125 binds chloride and bromide in CH CN, 3 and 126 binds halides in acetone. It is proposed that the strength of the interaction in 126, which does not possess amides or other hydrogen-bonding

90

S.R. Bayly · P.D. Beer

groups, is due to the arrangement of positive charges around the upper-rim of the hydrophobic cavity. The anion-binding properties of a series of highly cationic metallated calix[4]arenes have also been investigated. Host 127, in which the calix[4]arene is coordinated to four Ru(η6 -p-cymene) units was found by 1 H NMR titration to bind chloride, bromide, iodide and nitrate in aqueous solution [90]. Stability constants were determined with Ka = 551, 133, 51, and 49 M–1 , respectively. Cr(CO)3 is an effective electron-withdrawing group [91]. The coordination of this organometallic unit to the arene substituents of the isophthalamide in 128 increases the acidity of the NH protons and hence their affinity for anions. 1 H NMR titrations in CD3 CN solution provided stability constants of the same order of magnitude as those for the non-metallated receptor in the far less competitive solvent CD2 Cl2 . Loeb, Gale and co-workers have used Pt(II) as a structural template for the self-assembly of a series of anion receptors. Receptors 129–132 were found to bind a variety of anions in DMSO-d6 solution [92, 93]. Even the simple tetrapyridine receptor 129 has an affinity for anions due to its electrostatic charge. The addition of pyrrole hydrogen-bond donors increases the stability of the receptor : anion complexes by more than five-fold in 130, but in 131

Metal-Based Anion Receptor Systems

91

they are poorly orientated for anion coordination and the stability constants are only marginally higher than in 129. Receptor 132 which uses urea groups as hydrogen-bond donors binds anions with 1 : 2 receptor : anion stoichiometry, with stability constants an order of magnitude higher than 131.

5 Conclusion/Outlook The geometric, Lewis acidic, optical and electrochemical properties of metals make them ideal for use as multifunctional components in the construction of anion receptors. They are able to combine roles such as reporter groups, anion-binding sites and structural components to form receptor systems with sophisticated anion-sensing properties. In particular metal-containing receptors form the basis of systems in which anion-binding can be switched on and off, or in which the anion guest can even be moved from one binding site to another. Another significant advance has been the development of nanoscale structures such as dendrimers, nanoparticles, thin polymer films and selfassembled monolayers which incorporate redox- or photoactive metal centres as an anion-sensing component. The wide variety of added functionality and sheer adaptability that metals bring to anion receptor chemistry means that metal-based systems are set to continue at the forefront of research in this area.

References 1. 2. 3. 4. 5.

Beer PD, Cadman J (2000) Coord Chem Rev 205:131 Beer PD, Gale PA (2001) Angew Chem Int Ed 40:486 Beer PD, Hayes EJ (2003) Coord Chem Rev 240:167 Sun S-S, Lees AJ (2002) Coord Chem Rev 230:171 Amendola V, Fabbrizzi L, Mangano C, Pallavicini P, Poggi A, Taglietti A (2001) Coord Chem Rev 219:821

92

S.R. Bayly · P.D. Beer

6. 7. 8. 9. 10.

Bayly SR, Beer PD (2005) Anion Sensing. Topics Curr Chem 255:125 Beer PD, Gale PA, Chen GZ (1999) J Chem Soc Dalton Trans, p 1897 Beer PD, Keefe AD, Drew MGB (1989) J Organomet Chem 378:437 Beer PD (1996) Chem Commun, p 689 Beer PD, Hesek D, Kingston JE, Smith DK, Stokes SE, Drew MGB (1995) Organometallics 14:3288 Beer PD, Graydon AR, Johnson AOM, Smith DK (1997) Inorg Chem 36:2112 Beer PD, Chen Z, Goulden AJ, Graydon A, Stokes SE, Wear T (1993) J Chem Soc Chem Commun, p 1834 Reynes O, Moutet J-C, Pecaut J, Royal G, Saint-Aman E (2002) New J Chem 26:9 Beer PD, Chen Z, Drew MGB, Kingston J, Ogden M, Spencer P (1993) J Chem Soc Chem Commun, p 1046 Beer PD, Chen Z, Drew MGB, Johnson AOM, Smith DK, Spencer P (1996) Inorg Chim Acta 246:143 Beer PD, Cadman J, Lloris JM, Martinez-Manez R, Padilla ME, Pardo T, Smith DK, Soto J (1999) J Chem Soc Dalton Trans, p 127 Dusemund C, Sandanayake KRAS, Shinkai S (1995) J Chem Soc Chem Commun, p 333 Reynes O, Maillard F, Moutet J-C, Royal G, Saint-Aman E, Stanciu G, Dutasta J-P, Gosse I, Mulatier J-C (2001) J Organomet Chem 637-639:356 Beer PD, Drew MGB, Hesek D, Jagessar R (1995) J Chem Soc Chem Commun, p 1187 Beer PD, Hesek D, Nam KC, Drew MGB (1999) Organometallics 18:3933 Tomapatanaget B, Tuntulani T, Chailapakul O (2003) Org Lett 5:1539 Kim DS, Miyaji H, Chang BY, Park SM, Ahn KH (2006) Chem Commun, p 3314 Cui XL, Delgado R, Carapuca HM, Drew MGB, Felix V (2005) Dalton Trans, p 3297 Beer PD, Stokes SE (1995) Polyhedron 14:873 Oton F, Tarraga A, Espinosa A, Velasco MD, Molina P (2006) Dalton Trans, p 3685 Kingston JE, Ashford L, Beer PD, Drew MGB (1999) J Chem Soc Dalton Trans, p 251 Beer PD, Drew MGB, Jagessar R (1997) J Chem Soc Dalton Trans, p 881 Altmann R, Gausset O, Horn D, Jurkschat K, Schuermann M, Fontani M, Zanello P (2000) Organometallics 19:430 Beer PD, Szemes F, Balzani V, Sala CM, Drew MGB, Dent SW, Maestri M (1997) J Am Chem Soc 119:11864 Szemes F, Hesek D, Chen Z, Dent SW, Drew MGB, Goulden AJ, Graydon AR, Grieve A, Mortimer RJ, Wear T, Weightman JS, Beer PD (1996) Inorg Chem 35:5868 Beer PD (1998) Acc Chem Res 31:71 Beer PD, Dent SW, Wear TJ (1996) J Chem Soc Dalton Trans, p 2341 Lin ZH, Zhao YG, Duan CY, Zhang BG, Bai ZP (2006) Dalton Trans, p 3678 Mizuno T, Wei WH, Eller LR, Sessler JL (2002) J Am Chem Soc 124:1134 Beer PD, Berry N, Drew MGB, Fox OD, Padilla-Tosta ME, Patell S (2001) Chem Commun, p 199 Berry NG, Pratt MD, Fox OD, Beer PD (2001) Supramol Chem 13:677 Beer PD, Berry NG, Cowley AR, Hayes EJ, Oates EC, Wong WWH (2003) Chem Commun, p 2408 Valerio C, Fillaut J-L, Ruiz J, Guittard J, Blais J-C, Astruc D (1997) J Am Chem Soc 119:2588 Alonso B, Casado CM, Cuadrado I, Moran M, Kaifer AE (2002) Chem Commun, p 1778 Daniel M-C, Ruiz J, Blais J-C, Daro N, Astruc D (2003) Chem Eur J 9:4371

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.

Metal-Based Anion Receptor Systems

93

41. Beer PD, Davis Jason J, Drillsma-Milgrom DA, Szemes F (2002) Chem Commun, p 1716 42. del Peso I, Alonso B, Lobete F, Casado CM, Cuadrado I, Losada del Barrio J (2002) Inorg Chem Commun 5:288 43. Reynes O, Gulon T, Moutet J-C, Royal G, Saint-Aman E (2002) J Organomet Chem 656:116 44. Reynes O, Bucher C, Moutet J-C, Royal G, Saint-Aman E (2004) Chem Commun, p 428 45. Labande A, Astruc D (2000) Chem Commun, p 1007 46. Bayly SR, Gray TM, Chmielewski MJ, Davis JJ, Beer PD (2007) Chem Commun, p 2234 47. Basurto S, Riant O, Moreno D, Rojo J, Torroba T (2007) J Org Chem 72:4673 48. Miyaji H, Collinson SR, Prokes I, Tucker JHR (2003) Chem Commun, p 64 49. Aldridge S, Bresner C, Fallis IA, Coles SJ, Hursthouse MB (2002) Chem Commun, p 740 50. Zhang BG, Xu J, Zhao YG, Duan CY, Cao X, Meng QJ (2006) Dalton Trans, p 1271 51. Oton F, Tarraga A, Espinosa A, Velasco MD, Molina P (2006) J Org Chem 71:4590 52. Oton F, Tarraga A, Molina P (2006) Org Lett 8:2107 53. Plitt P, Gross DE, Lynch VM, Sessler JL (2007) Chem Eur J 13:1374 54. Beer PD, Timoshenko V, Maestri M, Passaniti P, Balzani V (1999) Chem Commun, p 1755 55. Deetz MJ, Smith BD (1998) Tetrahedron Lett 39:6841 56. Watanabe S, Onogawa O, Komatsu Y, Yoshida K (1998) J Am Chem Soc 120:229 57. Cormode DP, Murray SS, Cowley AR, Beer PD (2006) Dalton Trans, p 5135 58. Goodall W, Williams JAG (2000) J Chem Soc Dalton Trans, p 2893 59. Lo KKW, Lau JSY, Lo DKK, Lo LTL (2006) Eur J Inorg Chem, p 4054 60. Slone RV, Yoon DI, Calhoun RM, Hupp JT (1995) J Am Chem Soc 117:11813 61. Sun S-S, Anspach JA, Lees AJ, Zavalij PY (2002) Organometallics 21:685 62. Sun S-S, Lees AJ, Zavalij PY (2003) Inorg Chem 42:3445 63. de Wolf P, Waywell P, Hanson M, Heath SL, Meijer A, Teat SJ, Thomas JA (2006) Chem Eur J 12:2188 64. Curiel D, Beer PD (2005) Chem Commun, p 1909 65. DeSantis G, Fabbrizzi L, Licchelli M, Poggi A, Taglietti A (1996) Angew Chem Int Ed Engl 35:202 66. Fabbrizzi L, Licchelli M, Parodi L, Poggi A, Taglietti A (1999) Eur J Inorg Chem, p 35 67. Coskun A, Baytekin BT, Akkaya EU (2003) Tetrahedron Lett 44:5649 68. Fabbrizzi L, Licchelli M, Rabaioli G, Taglietti A (2000) Coord Chem Rev 205:85 69. Fabbrizzi L, Leone A, Taglietti A (2001) Angew Chem Int Ed 40:3066 70. Han MS, Kim DH (2002) Angew Chem Int Ed 41:3809 71. Tobey SL, Jones BD, Anslyn EV (2003) J Am Chem Soc 125:4026 72. Tobey SL, Anslyn EV (2003) J Am Chem Soc 125:14807 73. Tobey SL, Anslyn EV (2003) Org Lett 5:2029 74. Lee HN (2007) J Am Chem Soc 129:3828 75. Allevi M, Bonizzoni M, Fabbrizzi L (2007) Chem Eur J 13:3787 76. Amendola V, Colasson B, Fabbrizzi L, Douton MJR (2007) Chem Eur J 13:4988 77. Amendola V, Boiocchi M, Colasson B, Fabbrizzi L (2006) Inorg Chem 45:6138 78. Mizukami S, Nagano T, Urano Y, Odani A, Kikuchi K (2002) J Am Chem Soc 124:3920 79. Amendola V, Boiocchi M, Colasson B, Fabbrizzi L, Douton MJR, Ugozzoli F (2006) Angew Chem Int Ed 45:6920 80. Dickins RS, Gunnlaugsson T, Parker D, Peacock RD (1998) Chem Commun, p 1643 81. Bruce JI, Dickins RS, Govenlock LJ, Gunnlaugsson T, Lopinski S, Lowe MP, Parker D, Peacock RD, Perry JJB, Aime S, Botta M (2000) J Am Chem Soc 122:9674

94

S.R. Bayly · P.D. Beer

82. Montalti M, Prodi L, Zaccheroni N, Charbonniere L, Douce L, Ziessel R (2001) J Am Chem Soc 123:12694 83. Yamada T, Shinoda S, Tsukube H (2002) Chem Commun, p 1218 84. Mahajan RK, Kaur I, Kaur R, Uchida S, Onimaru A, Shinoda S, Tsukube H (2003) Chem Commun, p 2238 85. Leonard JP, dos Santos CMG, Plush SE, McCabe T, Gunnlaugsson T (2007) Chem Commun, p 129 86. Plush SE, Gunnlaugsson T (2007) Org Lett 9:1919 87. Beer PD, Cormode DP, Davis JJ (2004) Chem Commun, p 414 88. Beer PD, Dickson CAP, Fletcher N, Goulden AJ, Grieve A, Hodacova J, Wear T (1993) J Chem Soc Chem Commun, p 828 89. Holman KT, Orr GW, Atwood JL, Steed JW (1998) Chem Commun, p 2109 90. Staffilani M, Hancock KSB, Steed JW, Holman KT, Atwood JL, Juneja RK, Burkhalter RS (1997) J Am Chem Soc 119:6324 91. Camiolo S, Coles SJ, Gale PA, Hursthouse MB, Mayer TA, Paver MA (2000) Chem Commun, p 275 92. Bondy CR, Gale PA, Loeb SJ (2004) J Am Chem Soc 126:5030 93. Vega IE, Gale PA, Light ME, Loeb SJ (2005) Chem Commun, p 4913

Struct Bond (2008) 129: 95–125 DOI 10.1007/430_2007_072 © Springer-Verlag Berlin Heidelberg Published online: 28 November 2007

Recent Progress of Phosphate Derivatives Recognition Utilizing Artificial Small Molecular Receptors in Aqueous Media Shun-ichi Tamaru1 (u) · Itaru Hamachi2 1 Department

of Nano-science, Sojo University, 4-22-1 Ikeda, 860-0082 Kumamoto, Japan [email protected] 2 Department of Synthetic Chemistry and Biological Chemistry, Kyoto University, Kyotodaigaku Katsura, Nishikyo-ku, 615-8510 Kyoto, Japan 1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

96

2 2.1 2.2

Phosphate Recognition by Electrostatic Interactions and Hydrogen Bonds Cyclic/Acyclic Polyamine Type Artificial Chemosensors . . . . . . . . . . . Artificial Chemosensors Containing Other Cationic Groups . . . . . . . . .

98 98 102

3 3.1 3.2 3.3

Phosphate Anion Recognition Utilizing Coordination Zn-cyclen Type Artificial Chemosensors . . . . . . . . Zn-Dpa Type Artificial Chemosensors . . . . . . . . . Ratiometric Detection of Phosphate Derivatives . . .

. . . .

104 105 106 109

4

Recognition of Phosphorylated Protein Surfaces . . . . . . . . . . . . . . .

112

5

High Throughput Sensing Systems for Phosphate Derivatives Using Semi-wet Sensor Array . . . . . . . . . . . . . . . . . . . . . . . . . Molecular Recognition of Chemosensors in a Supramolecular Hydrogel . . Use of Hydrophobic Micro-domains of Supramolecular Hydrogel Fibers for Discrimination of Phosphate Derivatives . . . . . . . . . . . . . . . . .

5.1 5.2

Chemistry . . . . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

119 119 120

Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

122

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

123

6

Abstract Phosphate derivatives are ubiquitous in living organisms playing several important roles. Therefore, the development of sophisticated artificial phosphate receptors and chemosensors that can work in aqueous conditions is currently an area of great interest. There is a need to develop new methodologies for detection, separation or transport of biologically relevant phosphate derivatives. In the past two decades, many artificial chemosensors have been reported; these can be broadly classified into (1) chemosensors utilizing electrostatic interactions and hydrogen bonds and (2) chemosensors utilizing coordination chemistry. The development of these receptors is often inspired by the design of substrate-binding centers in natural enzymes such as protein kinases, phosphatases and phospholipases. More recently, the targets of such chemosensors have been broadened to include the recognition of phosphorylated protein surfaces. Here we review the recent progress in the development of molecular receptors and chemosensors that can selectively detect phosphate derivatives in aqueous media.

96

S.-i. Tamaru · I. Hamachi

Keywords Phosphate · Molecular recognition · Phosphoprotein · Host-guest · Fluorescent sensing

Abbreviations ADP adenosine 5
-diphosphate AMP adenosine 5
-monophosphate ATP adenosine 5
-triphosphate AXP adenosine nucleotide cAMP adenosine 3
,5
-cyclic monophosphate CD circular dichroism CTP cytosine 5
-triphosphate CoA coenzyme A 5-CF 5-carboxylfluorescein diMeP dimethyl phosphate GTP guanosine 5
-triphosphate IP3 d-myoinositol 1,4,5-triphosphate MAPK mitogen-activated protein kinase MeP methyl phosphate MLCT metal-ligand charge transfer NPP 4-nitrophenyl phosphate PDGFR-β platelet-derived grows factor receptor-β PET photo-induced electron transfer PhP phenyl phosphate PPi pyrophosphate PV pyrocatechol violet p-Tyr O-phospho-L-tyrosine UDP-Gal uridine 5
-diphospho-α-d-galactose Zn-Dpa zinc(II)-dipicolylamine

1 Introduction In nature, a variety of phosphate anion derivatives are ubiquitously distributed [1]. In addition to inorganic phosphate, there are many phosphates found in biological molecules. For instance, the backbones of DNA and RNA consist of phosphate diesters connecting 3
- and 5
-oxygens of ribose. Nucleoside phosphates (nucleotides) are not only the structural units of RNA and DNA, but also the essential parts of several cofactors such as coenzyme A (CoA), flavin adenine dinucleotide (FAD), and nicotinamide adenine dinucleotide (NAD). They play important roles in metabolic cascades of biological substances and in cellular signaling. Adenosine triphosphate (ATP) is a primary energy currency molecule in all living organisms, which is produced from ADP and inorganic phosphate by ATPase in cellular respiration or photosynthesis. The covalent bond connecting the phosphate groups stores a high energy that is easily broken down by hydrolysis to re-

Phosphate Recognition Utilizing Artificial Small Molecular Receptors

97

lease an amount of energy sufficient to drive many biochemical reactions. For example, the nucleoside pyrophosphate derivatives of saccharides such as uridine 5
-diphospho-α-D-galactose (UDP-Gal) are universal substrates of glycosyltransferase necessary to form a glycosyl bond for the synthesis of glycoprotein and oligo-saccharides. Unlike the molecular elements in bioenergetics, adenosine 3
,5
-cyclic monophosphate (cAMP) and D-myoinositol 1,4,5-triphosphate (IP3) represent important second messengers playing central roles in intra-cellular signal transduction [2, 3]. Phosphorylated proteins are other important phosphate derivatives in biological phenomena, which are produced by protein kinases, a family of enzymes catalyzing phosphorylation. Protein phosphorylation is one of the most common post-translational protein modifications in eucaryotic cells. The vast majority of phosphorylation occurs as a mechanism to regulate the biological activity of a protein [4]. From the genomic sequence, the presence of over 500 kinds of protein kinase in a human body is forecasted [5]. This fact suggests that such post-translational CENOTE phosphorylation is widely used to control diverse protein functions. In particular, phosphorylation on serine, threonine or tyrosine residues by protein kinases is fundamental to cell signaling networks. In these networks protein functions are controlled via the reversible phosphorylation/dephosphorylation processes catalyzed by corresponding kinase/phosphatase coupling with spatio-temporal fashions in response to changes of the cellular environment [6]. In practice, the phosphorylation acts as a molecular switch to turn on the protein activity. A typical example of such allosteric control of protein functions is observed in mitogen-activated protein kinase (MAPK) ERK2. In the deactivated state, two domains of ERK2 (extracellular signal-regulated kinase 2) are separated from each other by the loop strand. The structural consequences of phosphorylation at threonine (Thr 183) and tyrosine (Tyr 185) residues on the loop strand include active site closure, alignment of key catalytic residues that interact with ATP, and remodeling of the activation loop, resulting in recovery of the kinase activity [7]. Protein phosphorylation also plays a crucial role in controlling the reversible assembly of a cellular signaling complex [8, 9]. Phosphorylation of the appropriate peptide sequence generates new interaction sites exposed on a protein surface which are recognized by the downstream proteins containing phosphoprotein-binding domains, forming multibinding domains such as SH2, PTB, FHA, and WW domains involved in many signaling proteins [10]. Due to such significant roles of phosphate derivatives in biological events, it is generally anticipated that development of sophisticated artificial receptors toward phosphates would greatly contribute to the understanding of biological phenomenon involving phosphate derivatives. Such advancement in understanding should afford new methodologies for detection, separation, or transport of biologically important phosphates. In the following we review the recent progress of molecular recognition of phosphate deriva-

98

S.-i. Tamaru · I. Hamachi

tives utilizing artificial small molecular receptors that can work in aqueous media.

2 Phosphate Recognition by Electrostatic Interactions and Hydrogen Bonds 2.1 Cyclic/Acyclic Polyamine Type Artificial Chemosensors Polyamines that exist as polycationic species in water under appropriate pH conditions are suitable scaffolds to capture phosphate derivatives by the formation of multiple-hydrogen bonding and electrostatic interaction. It has previously been shown that naturally occurring polyamines such as spermine and spermidine bind phosphates or DNA [11]. Nakai et al. reported that the apparent affinity constants, logKapp for the 1 : 1 complexation of spermine with AMP, ADP, and ATP were found to be 2.6, 3.1 and 4.0, respectively, in 50 mM tris buffer (pH 7.5) [12]. Based on this moderate affinity, several artificial phosphate receptors containing polyamines were developed. Because of the difficulty of capturing phosphate anions by non-covalent interaction in aqueous media, however, successful examples of artificial small molecular receptors or sensors for phosphates are still relatively scarce.

In a pioneering work, Czarnik and co-workers synthesized the branching polyamine-appended anthracene derivative as a fluorescent chemosensor for phosphate anions. Compound 1, having a branching polyamine chain on the 9-position of the anthracene moiety, formed a 1 : 1 complex with inorganic phosphate or ATP in aqueous solution at pH 6 (logKeq = 0.82 for PO4 2– , 4.2 for ATP) [13]. The complexation caused a fluorescence enhancement that can be ascribed to the cancellation of quenching due to photo-induced electron transfer (PET), as shown in Scheme 1. In the absence of phosphate ions, 1 showed weak fluorescence due to the PET quenching by the free amine group. The complexation of the anionic phosphate oxygens with ammonium ions on 1 localized the remaining phosphate OH group in close proximity to the free amine, and then favorable intracomplex proton transfer occurred. As a result, the intramolecular PET quenching was eliminated and stronger emission was observed.

Phosphate Recognition Utilizing Artificial Small Molecular Receptors

99

Scheme 1 Proposed binding mechanism of Chemosensor 1 with inorganic phosphate. The PET quenching is canceled upon the phosphate binding with 1 causing fluorescence enhancement

The Czarnik group then prepared compound 2, which has two polyamine chains at the 1- and 8-positions of anthracene, as a fluorescence chemosensor toward pyrophosphate (PPi) [14]. To effectively bind both phosphates of PPi simultaneously, the polyamine chains were geometrically preorganized. The dissociation constant (Kd ) for the 2-PPi complex was found to be 2.9 µM, whereas that for 2-phosphate was 6.3 mM. In particular, chemosensor 2 displayed a 2200-fold pyrophosphate/phosphate discrimination at pH 7 (Scheme 2). Such high ion selectivity allows a real-time assay of pyrophosphate hydrolysis catalyzed by inorganic phosphatase. Since macrocyclic polyamines have a much greater charge density within the molecular skeletons in the multi-protonated state as compared to the linear polyamines, they should have an entropic advantage for the complexation of phosphates. On the basis of this advantage of macrocyclic polyamines over the corresponding acyclic polyamines, Kimura et al. reported that a macrocyclic hexaamine 3 showed stronger affinity to AMP, ADP and ATP at pH 8.0

Scheme 2 Proposed binding mechanism of Chemosensor 2 with pyrophosphate. The PET quenching is canceled upon the phosphate binding with 1 causing fluorescence enhancement

100

S.-i. Tamaru · I. Hamachi

(logKs = 3.2, 5.6 and 6.4, respectively) than those for corresponding spermine complexes (mentioned above) [15]. Lehn and his co-workers reported that the oxygen-containing macrocyclic polyamine 4 is a good receptor of nucleotides such as ATP and ADP [16, 17]. Due to the formation of multiple ionic interactions, 4 strongly bound ATP, especially under very acidic conditions (logKs = 4.8 at pH 7.6, 11.0 at pH 4.0). Interestingly, they found that compounds 4 and 5 were capable of catalyzing the hydrolysis of ATP to ADP (Scheme 3). In the complex of 4-ATP, a nitrogen atom attacked the terminal phosphate to yield ADP and thus phosphorylated 4. The phosphorylated 4 was immediately hydrolyzed to regenerate the original 4 in the catalytic process. Compound 5, in which a fluorogenic acridine ring was introduced to the polyamine ring, showed greater selectivity between ATP and ADP than the parent compound 4. Compound 5 bound such nucleotides using two distinct forces, electrostatic interactions between the macrocyclic polycationic moiety and the polyphosphate and π-π stacking interactions between the acridine ring and nucleic base. The latter interactions induced the fluorescence change of the acridine moiety of 5 upon complexation.

Scheme 3 Schematic representation of the catalytic cycle for ATP hydrolysis by the chemosensor 5 following the nucleophilic pathway

Phosphate Recognition Utilizing Artificial Small Molecular Receptors

101

Other types of polycationic macrocyclic compound like 6 ∼ 8 were also reported as water-soluble receptors for nucleotides [18–20]. The apparent complexation constants, logKapp , toward ATP for 6 and 8 at neutral pH were found to be 5.2 and 8.4, respectively, while the logKs for 7 was 4.8. Coupling a suitable signal transducer to the binding scaffold is generally crucial for designing efficient chemosensors. For example, redox-active metal complexes such as metallocenes have been incorporated into various guestbinding scaffolds to yield chemosensors that detect the corresponding guests electrochemically. Beer et al. synthesized many artificial receptors containing redox active metal centers and evaluated their usefulness[21–23] In these studies, they prepared water-soluble phosphate chemosensors. Chemosensors 9 ∼ 11, tethering cyclic polyamines on a ferrocene moiety, bound ATP or phosphate to form a 1 : 1 complex at pH 6.5. By cyclic voltammetric experiments, it was revealed that the receptors were able to detect these phosphate anions using cathodic shifts of the redox potential of ferrocene by 60–80 mV [24, 25].

Ruthenium-(II)-tris-(2,2
-bipyridiyl), Ru(bpy)3 , has proven to be a useful signal transducer by virtue of its ability to change its luminescent emission upon guest binding. Water soluble Ru(bpy)3 -appended polyamine receptors 12 ∼ 14 were prepared by Beer’s group. These receptors detected ATP and phosphate in aqueous media by changing the metal-ligand charge transfer (MLCT) luminescent emission [26]. The intensity of MLCT emission (600–630 nm) was sensitive to pH and phosphate anions. The emission titration studies with phosphate and ATP showed that the MLCT emission of these chemosensors was quenched by 15% upon complexation of the guest.

102

S.-i. Tamaru · I. Hamachi

2.2 Artificial Chemosensors Containing Other Cationic Groups In addition to polyamines, guanidinium groups have also shown to be useful binding groups for phosphate anions. It is well-known that the active site of some phosphatases such as Staphylococcal nuclease [27] and alkaline phosphatase [28] contain guanidinium groups (from Arg residues) that are actively involved in binding phosphate derivatives. Inspired by this approach, several groups have designed guanidinium-based phosphate receptors. Anslyn and co-workers have published several interesting papers on molecular recognition of phosphate by cleft-like receptor molecules containing poly-guanidinium or -ammonium rims [29–31]. Phosphate receptors 15, 16 were designed for detection of D-myoinositol 1,4,5-triphosphate (IP3) [30]. To evaluate the binding affinity, they devised a fluorescent competitive assay. In this system, 5-carboxylfluorescein (5-CF) was initially bound to the receptors (15, 16) to form a 1 : 1 complex. The fluorescence of 5-CF increased as a consequence of the entrapment in the cleft of the receptor from aqueous media. The addition of IP3 to an aqueous solution containing 15-(5-CF) complex led the guest exchange to form the 15-IP3 complex and caused the concurrent release of 5-CF from the cleft to aqueous media, which in turn caused a dramatic decrease of 5-CF emission. From the change in fluorescence it was possible to calculate a logK value of 5.7 for the formation of 15-IP3.

On the basis of this cleft-type scaffold, Anslyn’s group prepared a chemosensor library to discover a sophisticated nucleotide receptor possessing high selectivity toward ATP [32]. The attachment of the scaffold on Wang resin for solid phase synthesis, followed by tethering two tripeptide chains branch-

Phosphate Recognition Utilizing Artificial Small Molecular Receptors

103

ing off the guanidium groups by the split-and-pool method yielded a small library of potential candidates for chemosensors. Several “hit” molecules were selected and picked up by the fluorescence assay with ATP derivative (N-methylanthraniloyl-ATP: MANT-ATP). Among them, chemosensor 17 containing Ser-Tyr-Ser sequences, was finally discovered as an ATP selective chemosensor. The binding constant of 17 for ATP was found to be 3.4 × 103 M–1 , approximately 10 times greater than that of the reference compound 18. Alternatively, Rebek et al. rationally designed and synthesized a watersoluble nucleotide receptor 19 that is comprised of a carbazole backbone, a cyclic guanidinium cation, and two Kemp’s triacidic imides [33]. Compound 19 bound cyclic AMP (cAMP) to form a 1 : 1 complex (logKs = 2.8) in aqueous media by cooperatively accumulated interactions, specifically electrostatic interaction between its guanidinium cation and the anionic phosphate diester, π-π stacking interaction between its carbazole backbone and adenine ring, and hydrogen bonding interactions (Fig. 1).

Fig. 1 Structure of 19 and the predicted lowest energy conformation of the complex between 19 and cAMP

Amidinium cations are used as an alternative to guanidinium cation and as they are more easily accessible in organic synthesis. Gramlich and co-

104

S.-i. Tamaru · I. Hamachi

workers reported the complexation behavior of cationic “cavitands” having four amidinium cations on the peripheral of resorcin[4]arene, toward nucleotides [34]. Cavitand 20 possessed four ethylene glycol chains on the lower rim of resorcin arene that enhance the water-solubility. The binding constants of 20 to nucleotides were estimated to be 6.6 × 105 M–1 for ATP, 4.9 × 104 M–1 for ADP, 1.0 × 104 M–1 for AMP, and 1.4 × 103 M–1 for cAMP. 1 H NMR studies suggested that the adenine ring was encapsulated in the hydrophobic cavity of resorcin[4]arene (Fig. 2). The affinity of 20 to AXP was only slightly greater than that to the corresponding nucleotides having other nucleobases, however, suggesting that electrostatic interactions between amidinium groups and phosphates are predominant in the complexation between the cavitand and nucleotides.

Fig. 2 Structure of 20 and the energy-minimized structure of the complex formed between chemosensor 20 and AMP

Similar to guanidinium, imidazole, a basic group of histidine side chains, might be used as a binding unit for phosphate in artificial receptors. As far as we know, however, there are no reported examples of such receptors that can successfully bind phosphates in aqueous media. Known examples are limited to receptors used in organic media. This again indicates the difficulty in the development of sophisticated phosphate receptors usable in aqueous solutions.

3 Phosphate Anion Recognition Utilizing Coordination Chemistry As mentioned above, many type of phosphate receptors containing polyamines as binding sites have been developed. Because polyamines also have high

Phosphate Recognition Utilizing Artificial Small Molecular Receptors

105

affinity toward various metal ions, however, the metal complexation with such binding sites competes with the desired phosphate capturing under physiological conditions. Therefore, the utilization of the polyamine type phosphate receptors in biological applications may be seriously restricted. In addition, the moderate binding affinity of polyamines toward phosphates under biological conditions is another drawback. On the other hand, coordination chemistry provided by metal complexes is often employed to bind a phosphate unit of substrates in the enzyme active site [35, 36]. This gives one an important clue for the design of phosphate-binding motifs that show the stronger affinity in aqueous media. 3.1 Zn-cyclen Type Artificial Chemosensors Kimura et al. reported a series of zinc(II) complexes of cyclic polyamine ligands as biomimetic models of zinc-containing enzymes such as carbonic anhydrase and alkaline phosphatase A [37]. In these biomimetic model systems, the phosphate group from the substrates coordinate to the zinc(II) centers [38]. Initially, it became clear that the mono-nuclear Zn-cyclen complex 21 formed a 1 : 1 complex with monodentate phosphate dianions such as phosphate, phenyl phosphate (PhP, logKs = 3.5), and 4-nitrophenyl phosphate (NPP, logKs = 3.1), whereas the metal-free cyclen showed considerably weaker affinity with these phosphate dianions under the same conditions. This indicated that the coordination bond between phosphate and metal center was strong enough in aqueous media, thus such the non-covalent interaction should be powerful candidate to capture phosphate anions in water. Ditopic compound 22 bound PhP or NPP as a sandwich-shaped 1 : 1 complex with moderate affinity (logKs = 4.6 for PhP, 4.0 for NPP) [40]. Trinuclear zinc complex 23, a mimic of the active site of zinc enzymes such as phospholipase C, showed the strongest affinity toward these phosphates (logKs = 5.8 for PP, 6.6 for NPP) among the Zn-cyclen type receptors, probably due to capturing the phosphate moiety in its pseudo-cleft like binding site [41]. Three molecules of a bipyridine derivative equipped with two Zn-cyclen complexes 24 were

106

S.-i. Tamaru · I. Hamachi

Scheme 4 Cartoon of the structure of chemosensor (24)3 -Ru

assembled with ruthenium(II) to form another type of phosphate receptor (24)3 -Ru (Scheme 4) [42]. Having two sets of the trinuclear binding site at the top and bottom of the assembly, the receptor selectively bound IP3 or CTP3 in a 1 : 2 stoichiometry. Interestingly, Kimura and coworkers found that receptor 21 selectively formed a 1 : 1 complex (logKs =3.4) with deoxythymidine (dT) or uracil analogues at physiological pH. In these systems, the deprotonated imide anion of dT coordinated to the zinc (II) ion center and two sets of hydrogen bonds were formed between carbonyl oxygens of dT and the amine protons of 21 [43]. This is the first dT (or U)-selective artificial receptor to work in aqueous solution. They subsequently accomplished the selective recognition of thymidine nucleotide by using 22 with relatively high affinity (logKs = 5 ∼ 6). In these complexes, the thymine moiety was bound to Zn-cyclen and the phosphate anion moiety was concurrently coordinated to the other Zn center.

3.2 Zn-Dpa Type Artificial Chemosensors Hamachi and co-workers developed a fluorescent ATP chemosensor consisting of dinuclear zinc(II)-dipicolylamine (Zn-Dpa) appended to 2-acetyl

Phosphate Recognition Utilizing Artificial Small Molecular Receptors

107

anthracene 25 [44]. The apparent complexation constants of 25, Kapp , toward ATP, ADP, and AMP were found to be 2.2 × 106 , 2.2 × 105 , and 6.6 × 104 M–1 , respectively, at neutral aqueous pH, while the considerably weaker binding of mononuclear analogue 26 to AXP was observed. This suggests that the two Zn-Dpa moieties cooperatively captured the phosphate units of the nucleotides. The pattern of the fluorescence change of 25 upon complexation was dependent on the kind of nucleic bases of nucleotides. The binding of ATP to 25 caused a dramatic fluorescence enhancement (I/I0 ≈3), whereas a small increase of fluorescence was observed in the case of CTP. In contrast, the emission was diminished with the addition of GTP, probably due to the electron transfer quenching by the guanine group. Similar nucleic basedependent fluorescence change was also reported in the nucleotides sensing study utilizing a cyclic polyamine-type of receptor 5.

Hong et al. produced binuclear Zn complex type receptors 27, 28 as fluorogenic and chromogenic phosphate sensors, having high selectivity toward pyrophosphate [45, 46]. Kim and his co-workers reported that a complex of 29 with pyrocatechol violet (PV) can colorimetrically detect phosphate derivatives such as inorganic phosphate and AMP through the ligand exchangeinduced color change [47, 48]. In this system, the bound PV was effectively

108

S.-i. Tamaru · I. Hamachi

Scheme 5 Schematic representation of the mechanism of chromatic detection by chemosensor 29-PV toward inorganic phosphate. The effective displacement of PV with phosphate anion at the binding site results in color change of the solution from blue to yellow

displaced with phosphate anion at the binding site so as to cause the remarkable color change of the solution from blue to yellow (Scheme 5). Excimers, excited dimers of two planar fluorophores, often display significantly red-shifted fluorescence relative to the corresponding monomer. Hence, suitable molecules that can form excimers resulting from complexation with the target phosphate would become unique chemosensors that show drastic fluorescence change during phosphate recognition. A chemosensor 30 attaching two Zn-Dpa units to both ends of naphtalenediimide formed a 2 : 2 complex with pyrophosphate resulting in an excimer formation (logKapp = 4.1 × 105 M–1 ), whereas the addition of other anions including AXP did not show any meaningful fluorescence change [49]. This sensor can selectively detect pyrophosphate in the presence of ATP or inorganic phosphate by the characteristic red-shifted emission. In addition to the four Zn binding sites for pyrophosphate, the favorable π-π interaction of the two flat aromatic

Scheme 6 Proposed binding mechanism of chemosensor 30 with PPi. The formation of a 2 : 2 type binding between 30 and pyrophosphate leads to excimer formation

Phosphate Recognition Utilizing Artificial Small Molecular Receptors

109

moieties allowed the formation of a 2 : 2 type binding between 30 and pyrophosphate (Scheme 6). 3.3 Ratiometric Detection of Phosphate Derivatives The detection of a specific anion by using the emission or excitation change at two different wavelengths provides a significant advantage over conventional measurements using a single wavelength change. Such a dual excitation/emission system enables a ratiometric detection of an analyte, thus allowing precise and quantitative analysis and imaging even in complex environments such as that inside living cells. Hamachi et al. successfully developed a bis(Zn-Dpa) appended xanthone type receptor 31 that is capable of ratiometric detection of phosphate derivatives under physiological conditions using excitation fluorescence [50]. As shown in Fig. 3, the addition of ATP to an aqueous solution of 31 resulted in a see-saw type excitation spectral change. The mechanism of such a excitation spectral change was proposed as follows. In the absence of phosphate, the carbonyl oxygen atom of the xanthone fluorophore was coordinated to both Zn centers, as shown in Scheme 7. When a phosphate derivative was bound, the fickle coordination bond between one of the Zn centers and the carbonyl oxygen atom was cleaved by the strong interaction of the metal ion with the phosphate, which caused the excitation change of the xanthone fluorophore. Acridine type chemosensors 32, 33 were also designed by Hamachi and coworkers as dual-emission chemosensors of nucleoside pyrophosphate

Fig. 3 Change in the excitation spectrum of 31 induced by ATP addition

110

S.-i. Tamaru · I. Hamachi

Scheme 7 Phosphate anion-induced coordination rearrangement of 31 upon binding to a phosphate derivative

(Fig. 4) [51]. In the resting state, these sensors are in equilibrium between mono- and binuclear Zn complexes. In this state the sensors predominantly exist as mononuclear Zn complexes, with binuclear Zn complexes as the minor species, as illustrated in Scheme 8. The binding of anionic nucleo-

Fig. 4 Change in the fluorescence emission of 32 (a) and 33 (b) induced by ATP addition

Phosphate Recognition Utilizing Artificial Small Molecular Receptors

111

Scheme 8 Schematic representation of the dual-emission sensing mechanism of the acridine chemosensor for nucleoside PPs

side pyrophosphate with the sensor molecules induced complexation of the second Zn, concurrently cleaving the originally formed coordination bond between Zn and the acridine nitrogen atom. Such a guest-induced rearrangement of the coordination chemistry resulted in the dual-emission change. The chemosensors showed high selectivity for nucleotide di- or tri-phosphate over nucleoside monophosphate, inorganic phosphate, and nucleotide derivatives of monosaccharides such as uridine 5
-diphospho-α-D-galactose (UDPGal). On the basis of such selectivity, these chemosensors were successfully utilized for real-time monitoring of glycosyl transfer catalyzed by β-1,4galactosyltransferase. Since UDP-Gal is converted into UDP in the process of the glycosyltransfer reaction, the detection of UDP by chemosensor 33 implied a monitoring of the reaction progress. It should be noted that UDP-Gal has a lower affinity toward 33 than UDP. This can be done in a dual-emission change manner without any modification of the reactants such as UDP-Gal and the glycosyl acceptors (Scheme 9). Kikuchi et al. reported a cyclen-Cd complex containing 7-amino-4trifluorocumarin 34 as a fluorogenic chemosensor [52]. During the complexation of 34 with phosphates, the coordination bond of the aromatic amino group of the coumarin to Cd was cleaved, causing a dual-change of the excitation spectrum (Scheme 10). The dissociation constant of 34-nucleotide

112

S.-i. Tamaru · I. Hamachi

Scheme 9 Fluorescence sensing of the glycosyl-transfer reaction based on ratiometric detection by 33

Scheme 10 Design concept of sensing phosphates utilizing cyclen-Cd complex 34

complexes was approximately 10–5 M. This chemosensor was utilized for the real-time monitoring of the activity of a phosphodiesterase which cleaves an undetectable cyclic nucleotide (e.g. cAMP) to produce the corresponding detectable nucleotide (e.g. AMP).

4 Recognition of Phosphorylated Protein Surfaces In addition to small molecular phosphate derivatives, phosphate anions having large molecular weight recently emerged as important new subjects in anion recognition research. Hamachi and his co-workers developed

Phosphate Recognition Utilizing Artificial Small Molecular Receptors

113

the first artificial chemosensors that can recognize phosphorylated proteins/peptides [53, 54]. These receptors 35, 36 in which the anthracene derivatives were linked with two Zn-Dpa complexes, bound single phosphate groups of peptides or proteins by the cooperative action of two Zn-Dpa sites, so that the fluorescence intensity increased due to the anthracene fluorophore. From the fluorescence change, the apparent affinity constants for the 1 : 1 complexation with inorganic phosphate or phosphate monoesters such as o-phosphotyrosine (p-Tyr) and relevant species were found to be in the range of 104 ∼105 M–1 in aqueous solution (Table 1). These chemosensors tightly complexed with ATP or ADP with stronger affinity (Kapp > 107 M–1 ), whereas the phosphate diesters such as dimethyl phosphate (diMeP) and cyclic AMP did not cause any fluorescence change up to the millimolar concentration range. No evidence for binding of the chemosensors was obtained when other anions such as sulfate, nitrate, acetate, fluoride, or carboxylate were added, indicating that the chemosensors possessed high selectivity toward phosphate

Table 1 Summary of the apparent binding constants (Kapp , M–1 ) of 35 and 36 to phosphate species by fluorescence change Phosphate derivative a

Chemosensor 35 36

Phosphate derivative a

Chemosensor 35 36

NaH2 PO4 b PhP b p-Tyr b MeP b DiMeP b

4.2 × 105 2.1 × 105 3.1 × 105 1.1 × 105 –d

ATP c ADP c AMP c cAMP c

> 107 > 107 2.3 × 105 –d

a

2.9 × 105 5.1 × 104 6.1 × 105 7.9 × 103 –d

4.0 × 105 1.6 × 105 9.1 × 103 –d

PhP = phenyl phosphate, p-Tyr = O-phospho-L-tyrosine, MeP = methyl phosphate, diMeP = dimethyl phosphate, ATP = adenosine 5
-triphosphate, ADP = adenosine 5
-diphosphate, AMP = adenosine 5
-monophosphate, cAMP = adenosine 3
,5
-cyclic monophosphate. b Measurement conditions: 10 mM HEPES, pH 7.2, 20 ◦ C. c 50 mM HEPES, 50 mM NaCl, pH 7.2, 20 ◦ C. d Since the fluorescence change was scarcely observed, the association constant cannot be obtained.

114

S.-i. Tamaru · I. Hamachi

species among various anions, and can distinguish phosphate and phosphate monoesters from phosphate diesters. Determination of the binding ability of the chemosensors 35, 36 to phosphorylated peptides was conducted as a model study of phosphorylated protein surface recognition. Similar to phosphate sensing, the fluorescence intensity of 35 increased upon addition of a phosphorylated peptide in aqueous solution. In the titration study with peptide-a (EEEI-pY-EEFD), a consensus sequence phosphorylated by a protein kinase v-Src, the fluorescence of 35 was enhanced, with the intensity finally reaching a 2.5 fold increase relative to that in the absence of the peptide. In contrast, the corresponding nonphosphorylated peptide-g (EEEI-Y-EEFD) did not cause any emission change, showing that the chemosensor can distinguish a phosphorylated peptide from a non-phosphorylated one. Interestingly, the affinity of the chemosensors depended on the number of negative charges located on the phosphorylated peptide. Among the tested peptides, both chemosensors showed the strongest binding affinity (Kapp of 106 –107 M–1 ) for peptide-a, which has a larger negative charge (– 8) (Table 2). The significant emission enhancement of these chemosensors and the high selectivity towards phosphorylated peptides enabled the detection of phosphorylated peptides by naked inspection of the emission change. This is illustrated in the photograph shown in Fig. 5. Such fluorescence intensification of the chemosensors is clearly ascribed to the phosphate-assisted binding of the second Zn cation. A schematic illustration of the sensing mechanism toward the phosphorylated peptide is depicted in Scheme 11. In the absence of a phosphorylated peptide, the second Dpa site of the chemosensor is Table 2 Amino acid sequences of peptides containing optimal consensus sequences phosphorylated by different protein kinases and apparent binding constants (Kapp , M–1 ) of 35 and 36 to the peptides as determined by fluorescence change Consensus substrate sequence

Kinase Net 35 charge

peptide-a Glu-Glu-Glu-Ile-pTyr-Glu-Glu-Phe-Asp peptide-b Asp-Glu-Glu-Ile-pTyr-Gly-Glu-Phe-Phe peptide-c Ala-Glu-Glu-Ile-pTyr-Gly-Val-Leu-Phe peptide-d a Lys-Ser-Gly-pTyr-Leu-Ser-Ser-Glu peptide-e Ala-Arg-Arg-Gly-pSer-Ile-Ala-Ala-Phe peptide-f Arg-Arg-Phe-Gly-pSer-Ile-Arg-Arg-Phe

v-Src c-Src Lck1 EGFR PKA Bck1

peptide-g Glu-Glu-Glu-Ile-Tyr-Glu-Glu-Phe-Asp

v-Src

a

–8 –6 –4 –2 0 +2

36

8.9 × 106 1.5 × 106 8.2 × 105 5.8 × 104 –b –b

9.5 × 105 3.6 × 105 1.5 × 105 1.2 × 104 –b –b

–b

–b

Amino acid sequence of ezrin (142–149) phosphorylated by EGFR. Since the fluorescence change was scarcely observed, the association constant cannot be obtained. b

Phosphate Recognition Utilizing Artificial Small Molecular Receptors

115

Fig. 5 Photograph of the increased emission of 36 in the presence of phosphorylated peptide-a (middle) compared to 36 only (left) and 36 with non-phosphorylated peptide-g (right)

Scheme 11 Schematic representation of the sensing mechanism of the chemosensors 35, 36 toward the phosphorylated peptide

116

S.-i. Tamaru · I. Hamachi

partially free from the Zn-complexation, so that the PET quenching by the benzylic amine of Dpa lessened the fluorescence intensity of anthracene. In the presence of phosphorylated peptide, on the other hand, the binding of the second Zn cation to the free Dpa site was facilitated, and as a result the PET quenching was canceled so as to recover the fluorescence intensity. The careful thermodynamic study of this molecular recognition using isothermal titration calorimetry (ITC) undoubtedly demonstrated that the binding was an endothermic and entropy-driven event in the aqueous buffer solution. These chemosensors were applied to two biological assays. Firstly, the realtime fluorescence monitoring of phosphatase-catalyzed dephosphorylation reactions was demonstrated. A phosphopeptide DADE-pY-LIPNNG (a fragment (988–998) of EGFR) was dephosphorylated by phosphatase PTP1B to yield the non-phosphorylated peptide (Fig. 6) [54]. The pre-binding of the substrate peptide with 35 enhanced the fluorescence intensity of the chemosensor. After the addition of PTP1B, the fluorescence declined in a time-dependent manner during the progress of enzymatic dephosphorylation of the peptide. This method is much simpler than the conventional one using a radio-active phosphorylated peptide as the enzyme substrate. Secondly, the selective staining of phosphoprotein in SDS-PAGE was carried out [55]. The chemosensor 36 was shown to work well as a fluorescent staining reagent in the conventional gel electrophoresis of the protein mixtures. As shown in Fig. 7, only two distinct bands were fluorescently observed under a UV trans-illuminator, those corresponding to phospho-ovalbumin (MW = 45.0 kDa) and phospho-α-casein (MW = 23.6 kDa). In contrast, very slight or no emission was observed in other bands corresponding to the

Fig. 6 Time trace of PTP1B-catalyzed dephosphorylation monitored by the emission of 35 with (circle) or without (square) PTP1B using a fluorescence plate reader

Phosphate Recognition Utilizing Artificial Small Molecular Receptors

117

Fig. 7 Selective phosphoprotein detection in SDS-PAGE using 36. Each lane includes two phosphoproteins (ovalbumin (45.0 kDa) and α-casein (24.0 kDa)), and four nonphosphorylated proteins (β-galactosidase) (MW = 116.0 kDa), bovine serum albumin (MW = 66.2 kDa), avidin (MW = 18.0 kDa), and lysozyme (MW = 14.4 kDa). Lane 1: CBB staining of the six proteins. Lane 2, 3: Detection of the phosphoproteins with UV transilluminator after staining with 36. The amount of each protein in lane 2 and lane 3 is 5.0 and 2.5 µg, respectively

four non-phosphorylated proteins β-galactosidase (MW = 116.0 kDa), bovine serum albumin (MW = 66.2 kDa), avidin (MW = 18.0 kDa), and lysozyme (MW = 14.4 kDa). Interestingly, brighter fluorescence was observed at the band of phospho-α-casein which has eight to nine phosphorylated amino acid residues, compared to that of less phosphorylated phospho-ovalbumin containing one to two phosphate units. This suggested that the chemosensors can distinguish the degree of protein phosphorylation. Recent advances in understanding of post-translational modification revealed that multisite phosphorylation, so-called hyper-phosphorylation, is a common mechanism for regulating protein functions in cell signaling pathways. For example, platelet-derived growth factor receptor-β (PDGFR-β), a membrane-bound cytokine receptor, exposes multiple tyrosine residues in the cytoplasmic domain. The binding of PDGF to the extracellular domain induces auto-phosphorylation at these tyrosine residues, followed by recruiting of specific signaling proteins containing SH-2 domains, consequently triggering multiple signaling pathways. Thus, the development of artificial receptors that can sense hyperphosphorylated proteins is currently another important topic in biology and biochemistry. Toward this end Hamachi et al. subsequently designed chemosensors 37, 38, in which two Zn-Dpa units functioning as phosphate binding sites were connected with bipyridine spacers [56, 57]. These chemosensors were able to bind a multiply-phosphorylated protein surface. The Zn-Dpa units were juxtaposed at an appropriately distal position to enable the cross-linking interaction with two distinct phosphate groups on a protein surface (Scheme 12). The binding abilities of the chemosensors with a series of bis-phosphorylated model peptides were evalu-

118

S.-i. Tamaru · I. Hamachi

Scheme 12 Strategy for phosphorylated protein/peptide recognition by chemosensors 37, 38 applying a cross-linking interaction with two phosphate groups on a protein surface

ated by circular dichroism (CD) spectral studies (Fig. 8), in which the α-helix content of the peptide was measured upon addition of the chemosensors. Sensors 37, 38 induced the α-helix formation of peptides having two phosphoserine residues (pS-5,16, – 9,16, and – 12.16), whereas they did not affect the secondary structure of the mono-phosphorylated peptide (pS-16). The complexation process was also monitored by changes in the emission intensity. The fluorescence titration of 38 with pS-9,16 gave the affinity constant (Kapp = 2.0 × 106 M–1 ), the value of which is over 40-fold higher than that of 38 with the mono-phosphorylated pS-16 peptide (Kapp = 4.8 × 104 M–1 ). This indicated that 38 can discriminate the number of phosphorylation sites

Phosphate Recognition Utilizing Artificial Small Molecular Receptors

119

Fig. 8 CD (θ) value change of phosphopeptides upon addition of bipyridiyl-type chemosensor 37 ∼ 41

of peptides. More recently, it was found that similar binuclear type artificial receptors could strongly bind phosphorylated CTD peptide by tight twopoint interactions between Zn-Dpa sites and phosphates on the peptide. Such a complexation disrupted phosphoprotein/protein interactions in a phosphorylated CTD peptide and the Pin1 WW domain, a phosphoprotein-binding domain [57]. The strategy based on a small molecular disruptor that directly interacts with phosphoprotein is unique and should be promising in developing a designer inhibitor for phosphoprotein-protein interaction.

5 High Throughput Sensing System for Phosphate Derivatives Using Semi-wet Sensor Arrays 5.1 Molecular Recognition of Chemosensors in a Supramolecular Hydrogel In order to achieve a complete understanding of the functions and roles of phosphate derivatives in biological systems, it is necessary to collect data simultaneously for a large range of different phosphorylation events by exhaustive analysis. Rapid and high throughput methods based on chemosensors are expected to significantly contribute to determining the phosphorylation level of biological samples. Similar to DNA arrays [58], a chemosensor array immobilizing a number of chemosensors on a surface of solid substrates is a promising candidate to carry out convenient and high-throughput sensing. To establish the chemosensor array, the effective immobilization of sensor molecules on a solid support is regarded to be of primary im-

120

S.-i. Tamaru · I. Hamachi

portance. Hamachi et al. recently proposed a unique chemosensor array utilizing a supramolecular hydrogel as a matrix in which the chemosensors were non-covalently immobilized in the water-rich environment of the gel, while practically retaining the original molecular recognition functions [59, 60]. An artificial glycolipid mimic GalNAc-suc-glu-(O-methyl-cyc-hexyl)2 spontaneously formed into supramolecular fibers based on a bimolecular layer structure in water (Fig. 9) [61–64]. The entanglement of the fibers gave a transparent hydrogel consisting largely of aqueous media and welldeveloped hydrophobic domains. Chemosensor 36 was non-covalently embedded in the hydrogel and showed binding affinity/selectivity toward phosphate derivatives and consequent fluorescence responses comparable to those observed in aqueous solution [59].

Fig. 9 Schematic representation of the hierarchal molecular assembly used to form a supramolecular hydrogel. The artificial glycolipid mimic GalNAc-suc-glu-(O-metyl-cychexyl)2 forms incipient nano-fibers based on a bimolecular layer structure. Such fibers contain extensive hydrophobic domains in their cores with oriented saccharide arrays exposed at the interfaces. The incipient nano-fibers are bundled to give thicker fibrils whose entangling results in the formation of a hydrogel

5.2 Use of Hydrophobic Micro-domains of Supramolecular Hydrogel Fibers for Discrimination of Phosphate Derivatives In addition to the uses above, the elaborate utilization of hydrophobic nanofiber domains in the supramolecular hydrogel allowed more selective discrimination of the Zn-Dpa-based chemosensors among various phosphate

Phosphate Recognition Utilizing Artificial Small Molecular Receptors

121

derivatives [60]. The chemosensor 42, newly synthesized by Hamachi and coworkers, had two Zn-Dpa binding sites and an environmentally sensitive fluorophore (dansyl). After the chemosensor was embedded in the supramolecular hydrogel, solutions including various anions were placed on it. As shown in Fig. 10, roughly three patterns of the fluorescence spectral change were observed: (1) the emission intensity of 42 (at 512 nm) increased with a blue shift in the case of the relatively hydrophobic PhP, (2) the intensity decreased with a red shift of the emission for the relatively hydrophilic ATP, phosphate and phospho-Tyr, (3) addition of phospho-diesters and nonphosphate anions caused no fluorescence changes. Fluorescence titration with ATP (Fig. 10c) in the gel spot showed that the emission gradually decreased with the concurrent red-shift of the emission maximum. In contrast, the reverse type of spectral change was observed when PhP was added to the hydrogel spot containing 42 (Fig. 10d). From these spectral changes, the binding constants were calculated to be 1.8 × 105 M–1 for ATP and 7.2 × 103 M–1 for PhP. Interestingly, such fluorescence change never occurred in homo-

Fig. 10 a Schematic illustration of chemosensor redistribution upon binding to a hydrophobic or hydrophilic phosphate derivative between the hydrophobic hydrogel nanofiber and the hydrophilic cavity. The three patterns of change observed in the fluorescence spectra of the hydrogel containing the dansyl-appended receptor 42. b Fluorescence spectral change of 42 (60 µM, red line) embedded in the hydrogel upon addition of PhP, ATP, phosphate, and phospho-Tyr. The emission intensity of 42 increases with the blue shift for PhP, whereas the intensity decreases with the red shift for ATP, phosphate, or phospho-Tyr. (c–d) Fluorescence spectral change and the fluorescence titration plots (inset) of 42 embedded in the hydrogel upon addition of ATP or PhP, respectively: [42] = 2 µM, [ATP] = 0–60 µM for (e), [42] = 6 µM, [PhP] = 0–600 µM for (f) at λex = 322 nm

122

S.-i. Tamaru · I. Hamachi

geneous aqueous solutions, indicating that the hydrophobic domain of the present supramolecular hydrogel is crucial for the phosphate-induced change. Most significantly, the pattern of fluorescence response depended on the hydrophilicity of the phosphate derivatives. Specifically, the strongly hydrophilic ATP induced a red-shift in the emission of 42 with reduced intensity, whereas the rather hydrophobic PhP caused an emission increase with a blue shift. These spectral changes imply that the microenvironment of the 42-ATP complex is more hydrophilic than that of 42 itself, whereas the 42-PhP complex locates in the more hydrophobic microenvironment relative to the original 42. Using the unique semi-wet hydrogel array, a variety of phosphate anion species may be rapidly and conveniently discriminated from each other by both the emission intensity change and the wavelength shift as shown in Fig. 11.

Fig. 11 Photo images of the sensing patterns of semi-wet molecular recognition (MR) chips of the hydrogel of GalNAc-suc-glu(O-metyl-cyc-hexyl)2 containing (a) 36 (40 µM), (b) 42 (60 µM)

6 Summary Phosphates and their derivatives are abundant in biological systems including inorganic phosphates, small molecular organic phosphates such as ATP, phospholipids and phosphorylated organic intermediates, and phosphate derivatives having large molecular weight such as DNA and phospho-peptides and proteins. It is now clear that these derivatives play distinct roles in biologi-

Phosphate Recognition Utilizing Artificial Small Molecular Receptors

123

Fig. 12 a Construction of the hybrid biosensor and fluorescence sensing for the doubly phosphorylated peptide. b Structure of the chemosensor unit on the hybrid biosensor. c Amino acid sequences of the WW domain and its mutant

cal phenomena under aqueous conditions, depending on their corresponding forms. Therefore, development of sophisticated artificial chemosensors that possess high binding affinity and selectivity toward the target phosphate is keenly desired in order to significantly contribute the understanding of the roles of phosphorylation in complex biological systems. Although many artificial chemosensors have been developed in the past two decades, several problems still remain to be overcome. In particular, clear discrimination among the phosphate anion families, such as phosphates, phosphate esters, and ATP derivatives, etc., is thus far quite difficult and remains challenging. In addition, molecular recognition in aqueous systems is generally difficult to manage. Fundamental research in many areas is required for the rational design of efficient artificial chemosensors. Most recently, Hamachi et al. proposed a hybrid system of artificial sensors with a biological receptor scaffold to achieve improved selectivity among various phosphates. They successfully constructed a unique biosensor in which phosphoprotein-binding peptide WW domains were attached to Zn-Dpa units on the side chains (Fig. 12) [65]. Such pattern recognition and detection may also be promising for the precise discrimination between various phosphates as proposed by Anslyn [66] and Hamachi [67]. More creative ideas and concepts are required for progress in this growing area because the design and synthesis of novel chemosensors is envisioned to have a great impact on both basic research and practical, diverse applications in the biological, diagnostic and medicinal fields.

References 1. Fraústo da Silva JJR, Williams RJP (1991) The Biological Chemistry of the Elements. Clarendon Press, Oxford 2. Reitz AB (1990) Inositol Phosphates and Derivatives. American Chemical Society, Washington DC

124

S.-i. Tamaru · I. Hamachi

3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.

Woodgett J (ed) (2000) Protein Kinase Function. Oxford University Press, New York Sefton BM, Hunter T (1998) Protein Phosphorylation. Academic Press, New York Lander SE et al. (2001) Nature 409:860 Johnson LN, Lewis RJ (2001) Chem Rev 101:2209 Canagarajah BL, Khokhlatchev A, Cobb MH, Goldsmith EJ (1997) Cell 90:859 Yaffe MB (2002) Nat Rev 3:177 Yaffe MB, Elia AEH (2001) Curr Opin Cell Biol 13:131 Holmberg CJ, Tran SEF, Eriksson JE, Sistonen L (2002) Trends Biochem Sci 27:619 Labadi I, Jenei E, Lahti R, Lönnberg H (1991) Acta Chem Scand 45:1055 Nakai C, Glinsmann W (1979) Biochemistry 16:5636 Huston ME, Akkaya EU, Czarnik AW (1989) J Am Chem Soc 111:8735 Vance DH, Czarnik AW (1994) J Am Chem Soc 116:9397 Kimura E (1985) Top Curr Chem 128:113 Hosseini MW, Lehn JM (1987) Helv Chim Acta 70:1312 Hosseini MW, Blacker AJ, Lehn JM (1990) J Am Chem Soc 112:3896 Aguilar JA, García-España E, Guerrero JA, Luis SV, Llinares JM, Miravet JF, Ramirez JA, Soriano C (1995) J Chem Soc Chem Commun 21:2237 Bazzicalupi C, Bencini A, Bianchi A, Fusi V, Giorgi C, Granchi A, Paoletti P, Valtancoli B (1997) J Chem Soc Perkin Trans 2 4:775 Baudoin O, Gbonnet F, Teulade-Fichou MP Vigneron JP, Tabet JC, Lehn JM (1999) Chem Eur J 5:2762 Beer PD, Gale PA, Chen Z (1999) Coord Chem Rev 185–186:3 Beer PD, Cadman J (2000) Coord Chem Rev 205:131 Beer PD, Gale PA (2001) Angew Chem Int Ed 40:486 Beer PD, Chen Z, Drew MGB, Kingston J, Ogden M, Spencer P (1993) J Chem Soc Chem Commun 24:1046 Beer PD, Chen Z, Drew MGB, Johnson AOM, Smith DK, Spencer P (1996) Inorg Chim Acta 246:143 Beer PD, Cadman J (1999) New J Chem 23:347 Cotton FA, Hazen EE Jr, Legg MJ (1979) Proc Natl Acad Sci USA 76:2551 Kim EE, Wyckoff HW (1991) Clin Chim Acta 186:175 Metzger A, Anslyn EV (1998) Angew Chem Int Ed 37:649 Niikura K, Metzger A, Anslyn EV (1998) J Am Chem Soc 120:8533 Suzanne LT, Anslyn EV (2003) J Am Chem Soc 125:14807 Schneider SE, O’Neil SN, Anslyn EV (2000) J Am Chem Soc 122:542 Kato Y, Conn MM, Rebek J Jr (1994) J Am Chem Soc 116:3279 Sabo L, Diederich F, Gramlich V (2000) Helv Chim Acta 83:93 Wilcox DE (1996) Chem Rev 96:2435 Lipscomb WN, Sträter N (1996) Chem Rev 96:2375 Kimura E (2000) Curr Opin Chem Biol 4:207 Aoki S, Kimura E (2002) Rev Mol Biotechnol 90:129 Koike T, Kimura E (1991) J Am Chem Soc 113:8935 Fujioka H, Koike T, Yamada N, Kimura E (1996) Heterocycles 42:775 Kimura E, Aoki S, Koike T, Shiro M (1997) J Am Chem Soc 119:3068 Aoki S, Zulkefeli M, Shiro M, Kohsako M, Takeda K, Kimura E (2005) J Am Chem Soc 127:9129 Aoki S, Kimura E (2000) J Am Chem Soc 122:4542 Ojida A, Park S, Mito-oka Y, Hamachi I (2002) Tetrahedron Lett 43:6193 Lee DH, Kim SY, Hong JI (2004) Angew Chem Int Ed 43:4777 Lee DH, Im JH, Son SU, Chung YK, Hong JI (2003) J Am Chem Soc 125:7752

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.

Phosphate Recognition Utilizing Artificial Small Molecular Receptors

125

47. Han MS, Kim DH (2003) Bioorg Med Chem Lett 13:1079 48. Han MS, Kim DH (2002) Angew Chem Int Ed 41:3809 49. Lee HN, Xu Z, Kim SK, Swamy KMK, Kim Y, Kim SJ, Yoon J (2007) J Am Chem Soc 129:3828 50. Ojida A, Nonaka H, Miyahata Y, Tamaru S, Sada K, Hamachi I (2006) Angew Chem Int Ed 45:5518 51. Ojida A, Miyahara Y, Wongkongkatep J, Tamaru S, Sada K, Hamachi I (2006) Chem Asian J 1:555 52. Mizukami S, Nagano T, Urano Y, Odani A, Kikuchi K (2002) J Am Chem Soc 124:3920 53. Ojida A, Mito-oka Y, Inoue M, Hamachi I (2002) J Am Chem Soc 124:6256 54. Ojida A, Mito-oka Y, Sada K, Hamachi I (2004) J Am Chem Soc 126:2454 55. Ojida A, Kohira T, Hamachi I (2004) Chem Lett 33:1024 56. Ojida A, Inoue M, Mito-oka Y, Hamachi I (2003) J Am Chem Soc 125:10184 57. Ojida A, Inoue M, Mito-oka Y, Tsutsumi H, Sada K, Hamachi I (2006) J Am Chem Soc 128:2052 58. Shena M, Shalon D, Davis RW, Brown PO (1995) Science 270:467 59. Yoshimura I, Miyahara Y, Kasagi N, Yamane H, Ojida A, Hamachi I (2004) J Am Chem Soc 126:12204 60. Yamaguchi S, Yoshimura I, Kohira T, Tamaru S, Hamachi I (2005) J Am Chem Soc 127:11835 61. Kiyonaka S, Snikai S, Hamahci I (2003) Chem Eur J 9:976 62. Kiyonaka S, Sugiyasu K, Shinkai S, Hamachi I (2002) J Am Chem Soc 124:10954 63. Tamaru S, Kiyonaka S, Hamachi I (2005) Chem Eur J 11:7294 64. Kiyonaka S, Sada K, Yoshimura I, Shinkai S, Kato N, Hamachi I (2004) Nat Mater 3:58 65. Anai T, Nakata E, Koshi Y, Ojida A, Hamachi I (2007) J Am Chem Soc 129:6232 66. Lavigne JL, Anslyn EV (2001) Angew Chem Int Ed 40:3118 67. Koshi Y, Nakata E, Yamane H, Hamachi I (2006) J Am Chem Soc 128:10413

Struct Bond (2008) 129: 127–174 DOI 10.1007/430_2007_070 © Springer-Verlag Berlin Heidelberg Published online: 6 November 2007

Anions and π-Aromatic Systems. Do They Interact Attractively? Pablo Ballester ICREA, Pg. Lluís Companys 23, 08010 Barcelona, Spain [email protected] Present address: Institute of Chemical Research of Catalonia (ICIQ), Avda. Països Catalans 16, 43007 Tarragona, Spain 1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

128

2

Theoretical Investigations of the Interaction of Anions (Halides) with π Aromatic Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . .

132

3 3.1 3.2

Experimental Evidence of the Anion-π Interaction . . . . . . . . . . . . . Solution and Related Crystallographic Studies . . . . . . . . . . . . . . . . Further Crystallographic Evidence of Anion-π Interactions . . . . . . . . .

152 152 163

4

Summary and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

170

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

172

Abstract Hydrogen bonds and charge–charge interactions have been widely used (either alone or in combination) in the design of efficient and selective synthetic receptors for anions. Intuitively, the interaction between anions and π-aromatic systems is associated with a repulsive force. Consequently, for many years, anion-π interactions have been completely neglected as favorable non-covalent interactions for the construction of efficient anion receptors. Recently, however, theoretical studies indicate the existence of an attractive interaction between anions and certain type of π-acidic aromatic systems. These theoretical studies together with the observation of supramolecular complexes in the solid state in which anions are included in deep aromatic cavities have encouraged an in depth study of the anion-π interaction. Nowadays, the anion-π interaction is considered by several researchers a potential non-covalent interaction for the design of anion receptors. This chapter will provide an overview of recent theoretical investigations, performed since 2002, on anion binding involving six-membered aromatic rings. A series of experimental studies, carried out since 2004, also evidencing the existence of a possible attractive interactions betweens anions and six-membered aromatic moieties of host-guest systems in the solid state and in solution is also discussed. Keywords Anion-π interactions · Anion coordination · Anion receptors · Anions · Host–guest systems · Supramolecular chemistry Abbreviations HB Hydrogen bond RHF Restricted Hartree-Fock

128 BSSE MEP MP2 DFT ZPE SI PCM AIM GMIPp MIPp EPS SAPT NICS RI-MP2 DFT TCB CSD bptz bppn

P. Ballester Basis set superposition error Molecular electrostatic potential Møller Plesset Density functional theory Zero point energy International system of units Polarized continuum solvent model Atoms in molecules General molecular interaction potential with polarization Molecular interaction potential with polarization Electrostatic potential surfaces Symmetry-adapted perturbation theory Nucleus-independent chemical shift Resolution identity MP2 Density functional theory 1,2,4,5-Tetracyanobenzene Cambridge Structural Database 3,6-bis(2
-pyridyl)-1,2,4,5-teatrazine 3,6-bis(2
-pyridyl)-1,2-pyridazine

1 Introduction The driving forces controlling the formation of non-covalent complexes between molecular or ionic species are quite varied and include the so-called hydrogen-bonding, stacking, hydrophobic, charge-transfer, van der Waals and ionic interactions. There is good experimental evidence of the dominant role of electrostatic in many of those intermolecular interactions [1]. An electrostatic or Coulombic interaction by definition requires no adjustment of the electronic properties around the ion or the molecule. Consequently, ion– ion, ion–dipole, dipole–dipole, ion–quadrupole and quadrupole–quadrupole interactions can be considered as mainly electrostatic non-covalent interactions. A “conventional” hydrogen bond (HB) may be represented as D – H· · ·A whereby D (donor) and A (acceptor) are both electronegative atoms (usually N and O). Hydrogen bonds are a prominent and typical representation of a dipole–dipole interaction. However, HBs are not necessarily restricted to N and O, but may involve other electronegative donor and acceptor functionalities. For example, a hydrogen bond of the type N – H· · ·X– , where X– is an anion, is termed an ionic hydrogen bond. In fact, molecules containing polarized N – H functionalities which behave as H-bond donors towards anions are widely used as receptors for recognition and sensing purposes in aprotic solvents [2]. The coordination of the anion by the N – H functionality is an ion–dipole interaction clearly dominated by electrostatics. The covalent bonding contribution to the interaction can be considered as insignificant. Equation 1 represents the electrostatic potential created by a point

Anions and π-Aromatic Systems. Do They Interact Attractively?

129

charge q, at a distance r from the charge. Equation 2 corresponds to the electrostatic potential at a distance r from the center of a dipole of q charges having a 2d separation in space. θ is the value of the angle defined by the lines that connect the center of the dipole with the point at r distance and the two point charges. As deduced from Eqs. 1 and 2, the ion–dipole interaction has the same dependence on the dielectric environment as the interaction between fully charged partners but is appreciably weaker on an absolute scale falling off with distance more steeply. 1 q 4πξ0 r 1 2dq cos θ V(r, θ) = 4πξ0 r2 V(r) =

(1) (2)

The main virtue of the ion–dipole compared to the ion–ion interaction, from the viewpoint of host–guest chemistry, is its directionality (i.e., the energy of the interaction of an ion and the dipole depends on their mutual orientation defined by the θ angle). This translates into a fundamental property to achieve selective recognition of anions using hydrogen-bonding receptors, that is, the topology of the targeted anion should be considered when deciding the placement of the hydrogen bond donors in the receptor’s structure. Other virtues of hydrogen bonding in the design of abiotic receptors for anions derive from its electroneutrality, from its capacity to form simultaneously several bonding interactions with the anion and from the rich chemistry available to incorporate this functionality into the molecular scaffold.

Scheme 1 Preparation of the diazabicycloalkanes and ion pairing in acidic aqueous media

130

P. Ballester

A clear analogy exists in the supramolecular chemistry of anions and the coordination chemistry of main group cations and transition metals [3, 4]1 . Fundamental design principles of cationic receptors (i.e., macrocyclic effect and preorganization) have been directly applied to the supramolecular chemistry of anions. Many hosts capable of forming ion-pairs with anions have been prepared by simply inverting the electronic nature of macrocyclic and macrobicyclic receptors by protonation of a suitable Lewis-basic center. In principle at least, the non-protonated receptor could be suitable for cation binding. This was the case of the serendipitous discovery of the first receptor of a guest chloride emerged during the preparation of a macrobicyclic compound having two converging tertiary amines (Lewis-basic centers). The protonation of the Lewis-basic centers afforded two protonated ammonium groups (Lewis-acid centers) capable of maintaining the spherical chloride anion within the receptor’s cavity through the formation of two ion-pair-reinforced hydrogen-bonding interactions (salt-bridge) [5–7]. In the case at hand, the protonation of the amine groups has a twofold effect: 1) adds positive charge to the receptor and 2) converts a tertiary amine into an excellent hydrogen bond donor (ammonium). Nevertheless, the addition of positive charge to an anionic receptor, even in the absence of additional hydrogenbonding influences, enhances anion binding. The great majority of positively charged organic receptors for anions are based on nitrogen compounds. The incorporation of cationic metal centers into the molecular scaffold of the receptor represents an interesting alternative to the introduction of positive charge by protonation or quaternization of the amine functionality. Furthermore, when the coordination features of the metal are partially fulfilled they can be exploited to assist the electrostatic binding with metal-anion covalent bonding. The major drawback of using positively charged anion receptors is the internal competition established between the anionic target and the counteranion of the cationic host. This is an unavoidable disadvantage of positively charged receptors when compared with electroneutral hosts for anions. The “anticrown chemistry” term has been coined to describe the way in which many electroneutral hosts were built to recognize anions. That is the incorporation of multiple and convergent Lewis-acid centers into preorganized cyclic molecular scaffolds. The Lewis-acid centers, usually transition metals or elements like B, Si, Sn or Hg, should expose their electron-deficient sites for interaction with the lone electron pairs of the anions. The hydrogen-bonding interaction of an anion with an amide N – H mentioned above has also been widely used in the preparation of neutral receptors for anionic guests [8]. While the above-mentioned non-covalent interactions (hydrogen bonds, charge–charge) have been widely used in the design of efficient and selective synthetic receptors for anions either alone or combined, the use of attrac1

The researchers working in the area of anion recognition having an inorganic chemistry background usually refer to the field as anion coordination chemistry

Anions and π-Aromatic Systems. Do They Interact Attractively?

131

tive interactions between negative charges and aryl rings towards the same end has remained dormant in the literature since they were first noticed by Scheneider in 1993 [9]. Recently, the studies into the anion-π interaction have intensified and gained interest among the scientific community. This is due, on the one hand, to the publication in 2002 of several papers dedicated to exploring the physical nature of the interaction of anions with aromatic compounds using electronic structure methods and, on the other hand, to the observation since 2004 of supramolecular complexes in the solid state in which anions are included in deep aromatic cavities. Furthermore, the recently noticed fact that the orientation of negatively charged groups [10] and lone pairs [11] above the plane of an aromatic is a frequently occurring structural motif in biopolymers has added additional interest to the subject. At first sight, a stabilizing effect for the interaction between a pair of electrons or a negative charge and the face of a π system seems to be counterintuitive. Not surprisingly, simple modelling studies indicated a repulsive interaction between anions and a benzene ring. It is now clear that noncovalent interaction between aromatics and positively charged cations, the cation-π interaction, are prominent in a wide variety of systems and should be considered as an important and general binding force [12]. The cationπ interactions are expected simply from electrostatic arguments. In fact, a simple electrostatic model based on the visual inspection of electrostatic potential surface of the aromatic ring has been used to rationalize the major binding trends of the cation-π interaction. In line with the principles outlined above for the preparation of anion receptors through the inversion of the electronic character of suitable Lewis-basic centers (amine to ammonium), a possible explanation for the observed geometry that points lone pairs or places negative charges into the face of the π system has to do with an inversion of the electronic character of the aromatic ring [13]. Theoretical investigations have clearly established the existence of binding interactions between a negatively charged species and an electron-poor π system. The theoretical results have encouraged the experimental obser-

Fig. 1 Inversion of the electronic character of Lewis-basic centers: a the lone pair of an amine is a good binding site for a cation while the protonated ammonium group is complementary to anions; b the π-system of an electronically rich aromatic compound establishes attractive interactions with cations (cation-π interaction) the introduction of electro-withdrawing substituents produces the depletion of electronic density of the π-system giving rise to attractive interaction with anions

132

P. Ballester

vation and evaluation of this new non-covalent interaction. Several authors have already claimed the potential use of anion-π interactions in the preparation of selective receptors for anions. Solid-state structural examples of anions included in deep aromatic cavities of supramolecular complexes have been used as evidences for the existence of a substantial anion-π attractive interaction. Few studies, however, have attempted to quantify the strength of the interaction in solution. Since the anion-π interaction seems to be a directional non-covalent interaction it has potential and unexplored applications in the design of selective neutral receptors for anion recognition. This chapter will provide a short overview of recent theoretical investigations on anion binding involving six-membered aromatic rings followed by a summary of experimental studies evidencing or not the existence of attractive interactions between anions and six-membered aromatic moieties of host–guest systems in the solid state and in solution2 .

2 Theoretical Investigations of the Interaction of Anions (Halides) with π Aromatic Systems Since 2002 several groups have studied in detail the interaction of anions with electron deficient aromatic rings using theoretical methods. These studies can be considered to derive from earlier fundamental work carried out by Dougherty [14], Besnard [15] and Alkorta [16] on the interaction between hexafluorobenzene and the heteroatom in molecules such as H2 O, HCN and HF wherein the negative end of the dipole is directed toward the π-system and aligned with the C6 axis of the ring. In particular, Alkorta et al. [16] studied the interaction of C6 F6 with HF using both MP2 and hybrid DFT methods (B3LYP) with 6-31-G∗∗ and 6-311++G∗∗ basis sets. The authors considered that these methods are more accurate for weak complexes than methods without electron correlation. A shortening effect in the calculated distance value between the interacting atom and the centroid of the aromatic ring is clearly observed by the inclusion of electron correlation (MP2 and B3LYP methods). Thus, for the minimum A when structure of the C6 F6 · · ·FH complex the computed distance is 3.076 ˚ A and 2.814 ˚ A, using the RHF method and the 6-31G∗∗ basis set and 2.858 ˚ respectively, when the B3LYP/6-31G∗∗ and MP2/6-31G∗∗ methodologies are applied. The use of a more extended basis set B3LYP/6-311++G∗∗ produces A while the calculated binda larger elongation of the same distance to 3.127 ˚ ing energy at this level of theory and basis set for the C6 F6 · · ·FH complex is 2

Complexes in which the anion is one component of an aromatic ring sandwich complex are outside of the scope of this chapter. Likewise, examples for the interaction of anions with five and seven membered aromatic rings will not be presented

Anions and π-Aromatic Systems. Do They Interact Attractively?

133

Fig. 2 Schematic representation of the geometry of the complexes of water, HCN, and HF with C6 H6

Fig. 3 Calculated MP2/6-31+G∗ MEP surfaces for a 1,3,5-triazine, b trifluoro-1,3,5-triazine, and c hexafluorobenzene. Electrostatic potential surfaces energies range from –19 (red) to +19 (blue) kcal/mol for 1,3,5-triazine, –39 (red) to +39 (blue) kcal/mol for trifluoro-1,3,5-triazine, and –24 (red) to +24 (blue) kcal/mol for hexafluorobenzene

∆EBSSE = 1.23 kcal mol–13 . The interaction energy of the C6 F6 · · ·FH complex is similar to the one observed in the formation of weak hydrogen bonds. Several years later, in 2002, Mascal et al. [17], Alkorta et al. [18] and Frontera, Deyà et al. [19] reported almost simultaneously the theoretical demonstration of the existence of an attractive interaction between a formally charged negative species and the π-system of an aromatic ring. Mascal et al. [17] performed ab initio orbital calculations at the MP2 level of theory with the 6-31+G∗ basis set, including counterpoise corrections for the basis set superposition error (BSSE), for the interaction of both 1,3,5-triazine and trifluoro-1,3,5-triazine with chloride and fluoride. The molecular electrostatic potential (MEP)4 maps of 1,3,5-triazine and trifluoro1,3,5-triazine clearly indicate an area of positive density concentrated on the C3 rotational axis passing through the center of the hexagonal ring and being perpendicular to the plane of the aromatic ring, similar to that observed for C6 H6 on the C6 rotational axis. Optimization from geometries that place the chloride near the area of positive density found on the C3 rotational axis of 1,3,5-triazine demon3

BSSE stands for Basis Set Superposition Error and the interaction energies are corrected for this inherent error 4 The MEP maps have been used for long time as a tool to identify both nucleophilic and electrophilic regions in a molecule

134

P. Ballester

strated the existence of an attractive interaction between the chloride and the π-system. The optimized structure obtained for the chloride-1,3,5-triazine complex positioned the chloride anion exactly on the C3 rotational axis of the A of the aryl centroid. The fact 1,3,5-triazine (Φ = 90) and at a distance of 3.2 ˚ that the structure of the complex was a minimum was confirmed by calculating the corresponding frequencies (no negative vibrational frequencies were found). The MP2/6-31+G∗ energy calculated for the geometry of the chloridearyl-centroid complex optimized using the same level of theory and basis set –1 was ∆EBSSE 0 K =– 4.8 kcal mol . Mascal et al. [17] also located another geometry for the complexation of the chloride and bromide anion with 1,3,5-triazine. This geometry is also a minimum and involves a C – H· · ·Cl– hydrogen bond. The formed hydrogen bond shows good geometrical characteristics, having a C· · ·Cl– distance A and a C – H· · ·Cl– angle of 180◦ . The calculated MP2/6-31+G∗ enof 3.4 ˚ ergy of this interaction geometry optimized at the same level of theory and –1 –1 basis set was ∆EBSSE 0 K =– 7.4 kcal mol , that is 2.6 kcal mol more stable than chloride-aryl centroid complex discussed above. The authors clearly state that although the participation of ions in gas phase chemistry yields interaction energies which seem to be exaggerated when compared to solution values, the reported energies of these interactions may still be relevant in the interior of a receptor. The authors used the trifluoro-1,3,5-triazine as a model to evaluate the extent to which further electron withdrawal in the triazine

Fig. 4 Side and top views of the minimized structures at the MP2/6-31+G∗ level of a the triazine chloride aryl centroid complex and b the triazine chloride hydrogen-bonding complex. The triazine and the chloride are shown in a scaled ball-and-stick representation

Anions and π-Aromatic Systems. Do They Interact Attractively?

135

Fig. 5 Side view and top view of the “attack” structure for the fluoride 1,3,5-triazine complex

Fig. 6 Molecular structures of the perfluoroaromatic compounds studied by Alkorta et al. [18]

system would enhance complexation. Higher interaction energy and even stronger chloride-aryl centroid complexes were located using the trifluoro–1 ◦ ˚ 1,3,5-triazine (∆EBSSE 0 K =– 14.8 kcal mol , r = 3.0 A and θ = 90 ). In fact, this high energy and close interaction distance computed for the Cl– -trifluoro1,3,5-triazine complex is comparable to those of the potassium cation-π complexes of benzene calculated at the same level of theory [20, 21]5 . The corresponding fluoride-aryl centroid complexes with 1,3,5-triazine and trifluoro-1,3,5-triazine are characterized by the presence of one negative vibrational frequency and represent a shallow inflection point on a surface connecting to a second geometry, the so-called “attack” structure which is a real minimum structure. This structure was suggestive of a reactant complex with the nucleophilic fluoride anion “attacking” one of the carbon atoms with close to the Bürgi-Dunitz trajectory [22, 23]. The considerable stabiliza–1 ˚ tion energy of the “attack” structure (∆EBSSE 0 K =– 18 kcal mol , r = 1.5 A and ◦ θ = 106.6 ) together with the distance values for the C· · ·F interaction and the lengthening of the adjacent C – N bonds points to a strong σ -complex. 5

Na+ · · ·C6 H6 interaction E0 = –15.0 kcal mol–1 , r=2.84 ˚ A. K+ · · ·C6 H6 interaction E0 =–18.3 kcal mol–1

136

P. Ballester

The calculated energies for these interactions fall off sharply with increasing polarity of the medium. The simple comparison of uncorrected MP2 single point energies of chloride-aryl centroid complex with triazine, using Tomasi’s Polarized Continuum solvent Model (PCM), gas phase (–8.5), heptane (–0.4), chloroform (2.1), ethanol (2.8), and finally water (3.2 kcal mol–1 ) suggest that the practical manifestation of these forces will be most likely in the context of anion containment. Alkorta et al. [18] studied the complexes formed by a variety of anions with perfluoro derivatives of benzene, naphthalene, tiophene and furan using DFT (B3LYP/6-31++G∗∗ ) and MP2 (MP2/6-31++G∗ and MP2/6-311++G∗∗ ) ab initio methods. The minimum structures of hexafluorobenzene with chloride and bromide show the anion interacting with the π-system with an anion-aryl centroid geometry. In both cases, a C6v symmetry was assumed during the optimization procedure. However, the C6 H6 · · ·F– complex structure that locates the A) anion on the C6 symmetry axis and on top of the aromatic ring (r = 2.5 ˚ shows two degenerate imaginary frequencies. In this case, the minimum structure corresponds to an “attack” geometry similarly to the interaction of fluoride with 1,3,5-triazine discussed above. The calculated interaction energy values for the anion-aryl centroid geometry range between –18.7 and –12.19 kcal mol–1 , which is comparable, as mentioned above, to the interaction energy of some benzene-cation complexes [20, 21]. A (see Table 1). The centroid-to-anion distance varies from 2.554 to 3.230 ˚ The effects of the complexation of the anion to the C6 F6 molecule are shortA) and a lengthening of the C – F bonds ening of the C – C bonds (0.004 ˚ A). On complexation, the fluorine atoms move slightly towards the (0.006 ˚ A further anion and the plane formed by the carbon atoms is about 0.02 ˚ away from the anion than that formed by the fluorine atoms. The hexafluorobenzene molecule adopts a “cup” conformation. Non-covalent interactions have been characterized using Bader’s theory of “Atoms In Molecules” (AIM) [24] which has been used successfully to under-

Table 1 Corrected interaction energies (kcal mol–1 ) and C6 F6 centroid-anion distance (r, ˚ A) calculated at the MP2/6-31++G∗ level together with the electron densities and their Laplacian (au) for the calculated bond critical point Complex

∆EBSSE MP2/6-31++G∗

r

ρ

∇2ρ

C6 F6 · · · F– C6 F6 · · · Cl– C6 F6 · · · Br–

–18.63 –12.76 –12.19

2.554 3.159 3.230

0.0118 0.0080 0.0087

0.0470 0.0238 0.0240

Anions and π-Aromatic Systems. Do They Interact Attractively?

137

Fig. 7 Schematic representation of the location of a bond (red), a ring (blue) and the cage critical point (black) originated from the interaction of hexafluorobenzene with fluoride

stand conventional [25] hydrogen bonds and cation-π interactions [26]. The AIM analysis of the electron density of these complexes (C6 F6 · · ·X– , X = F, Cl, Br) carried out by Alkorta et al. [18] indicated the formation of six degenerate bond critical points (bcp) between the anion and each of the carbon atoms of the C6 F6 molecule in a similar way to what was observed for the C6 F6 · · ·X – Y neutral complexes [16] and the C6 F6 · · ·Na+ complex [20]. In addition, sixring critical points (rcp) and one cage critical point (ccp) are generated. The rcp connect the anion with the middle of the C – C bond. The ccp is located over the hexafluorobenzene molecules along the C6 axis, connecting the anion with the center of the ring. The quantitative values for ρ and ∇ 2 ρ at the cps give a hint on the character and strength of the interaction. The values of electron density of the new bcp and its corresponding rcp are almost the same values. The positive and small value of the Laplacian of the bcps indicates a depletion of the electron density, as is common in closed shell interactions like those found in hydrogen bonds, ionic and cation-π interactions. Table 2 Corrected interaction energy at the MP2/6-31++G∗ level (kcal mol–1 ) and electrostatic, polarization and van der Waals contribution (kcal mol–1 ) to the interaction energy calculated using GMIPp method Complex

∆EBSSE

Eele

Epol

EvdW

Et

C6 F6 · · · Cl–

–12.7

–11.8

–6.6

5.2

–13.2

138

P. Ballester

It is worth mentioning that the interaction energies calculated at the MP2/6-31++G∗∗ level of the perfluoronaphtalene complexes with anions are the largest of all the study (∆EBSSE =– 17.31 (F– ), –16.70 (Cl– ) and –16.51 (Br– ) kcal mol–1 ) approximately 4.5 kcal mol–1 more than the energies of the corresponding C6 F6 · · ·X– complexes (see Table 1). However, the anions are located over the C(4a)–C(8a) bond, that is, the anion interacts with one bond, not with one aromatic ring. In the optimized structure, the naphthalene ring bent away from the anion for the complexes with chloride and bromide. In this case, instead of the “cup” observed for hexafluorobenzene, the perfluoronaphtalene adopts a “book” conformation pointing toward the anion. Alkorta et al. also analyzed the interaction energies using the general molecular interaction potential with polarization partition (GMIPp) [27, 28] of all the complexes of the aromatic compounds with chloride in order to calculate the contribution of the electrostatic, polarization and van der Waals terms to the overall interaction energy. The obtained results indicate that the polarization energy is 50 to 100% of the electrostatic energy. The sum of the three terms calculated with the GMIPp method provides an energetic value very close to the corrected interaction energy obtained at the MP2 level. Again, the obtained results for the contribution of the polarization and the electrostatic term to the overall interaction are analogous to those found for the cation-π interaction. Frontera, Deyà et al. [19] also studied the interactions of several anions with C6 F6 using HF/6-31++G∗∗ and MP2/6-31++G∗∗ ab initio methods. In all the complexes the anion is positioned over the aromatic ring along the C6 rotational axis passing through the center of the hexagonal ring and being perpendicular to the plane of the aromatic ring, that is, having an anionaryl centroid geometry. Initially the geometries of all complexes were fully optimized at the HF/6-31++G∗∗ level. The authors reported the corresponding binding energies with and without basis set superposition error (BSSE) and zero-point vibrational energy (ZPE) corrections. Frequency calculations at the same level confirmed that the structures are at their energy minimum. When they extended the calculations to the MP2/6-31++G∗∗ level, assuming C6v symmetry for the complexes C6 F6 · · ·X– , X = F, Cl, Br, the frequency calculations gave either one or more imaginary frequencies. This problem was solved by performing the geometry optimizations without imposing any symmetry constrains. In complete agreement with Alkorta’s report, the minimum energy complex found for the interaction C6 F6 · · ·F– corresponds to the nucleophilic attack of the anion at one carbon atom. In general, they conclude that the MP2-computed binding energies are more negative than the HF ones and the equilibrium distances are shorter. Frontera, Deyà et al. [19] also performed a topological analysis of the charge-density distribution and properties of critical points in the complexes using the AIM method providing an unambiguous definition of chemical bonding between the anion and the π-system. These calculations were per-

Anions and π-Aromatic Systems. Do They Interact Attractively?

139

Table 3 Interaction energies with (∆EBSSE , ∆EBSSE kcal mol–1 ) and without (∆E, 0K kcal mol–1 ) the basis set superposition error (BSSE) and zero-point vibrational energy correction (ZPE, 0 K) and equilibrium distances (r/˚ A) at HF/6-31G++G∗∗ and MP2/6∗∗ 31++G (italics) levels of theory and selected electron-density topological properties for the complexes of hexafluorobenzene with anions from reference [19] Complex

∆E

∆EBSSE ∆EBSSE r 0K

C6 F6 · · · F–

–18.8 –18.1

–17.8

2.669

C6 F6 · · · Cl–

–11.0 –10.8 –18.0 –13.2

–10.6 –12.9

3.404 3.155

C6 F6 · · · Br–

–13.2 –9.3 –20.7 –12.4

–9.5 –11.9

3.479 3.214

CPa

n

ρ

∇2ρ

bond ring cage bond ring cage bond ring cage

6 6 1 6 6 1 6 6 1

0.01001 0.00996 0.00702 0.00572 0.00571 0.00450 0.00618 0.00616 0.00480

0.04200 0.04226 0.04106 0.01637 0.01637 0.01820 0.01612 0.01613 0.01887

The electron density (ρ) and its Laplacian (∇ 2 ρ) in atomic units at the critical points (CP) originated upon complexation are given, as well as, the number (n) of each type of critical points a

formed by means of the program AIMPAC using HF/6-31++G∗∗ wavefunction and the obtained results are completely coincident with the ones we have previously discussed based on Alkorta’s report. The physical nature of the anion-π interaction and the importance of the polarization were analyzed by computing its contribution to the total interaction energy using the molecular interaction potential with polarization (MIPp) [29]. The MIPp is an improved generalization of the molecular electrostatic potential (MEP) and was also used by Alkorta in the energetic analysis of the C6 F6 · · ·Cl– . In the MIPp calculation three terms contribute to the total interaction energy: 1) an electrostatic term identical to MEP, 2) a classical dispersion-repulsion term, and 3) a polarization term derived from perturbation theory. In the calculation F– ion was considered as a classical nonpolarizable particle. The electro-

Table 4 Contribution to the total interaction energy (kcal mol) calculated with MIPp for hexafluorobenzene interacting with F– at several distances (˚ A) from the center of the ring Distance

Eele

Epol

Evdw

Et

1.5 2.0 2.5 3.0 3.5

–36.59 –22.40 –16.39 –12.66 –9.90

–43.84 –24.89 –13.93 –7.96 –4.74

1047.82 119.05 13.44 –0.50 –0.83

967.38 71.72 –16.89 –20.12 –15.48

140

P. Ballester

Fig. 8 Schematic representation (top and side view) of the positive molecular quadrupole moment of C6 F6 as cylinders with positive charges on the ends and negative charges in the center

static (Eele ), polarization (Epol ), van der Waals (Evdw ), and total interacting energies were calculated when a fluorine ion approaches the hexafluorobenzene molecule perpendicular to the center of the aromatic ring. The obtained results point out the importance of the polarization component, which is A range, the equilibrium dissimilar to the electrostatic term in the 2.0 to 3.0 ˚ A. The authors mention the tance for the fluoride aryl centroid complex is 2.6 ˚ importance of the quadrupole moment for understanding intermolecular interactions of aromatic system but they do not elaborate on this issue in this work. To date, most of the work reported on anion-π interactions using theoretical methods is between anions and electron deficient aromatics rings, i.e., hexaflurorobenzene, 1,3,5-trinitrobenzene, and 1,3,5-triazine. The electrondeficient aromatic rings are also called π-acidic aromatic rings and are characterized by having a permanent positive quadrupole moment value Qzz . The value of the quadrupole moment is a measure of the distribution of charge within a molecule, relative to a particular axis. In six-membered aromatic rings, the Qzz quadrupole measures the distribution of charge relative to the C6 rotational axis passing through the center of the hexagonal ring and being perpendicular to the plane of the aromatic ring. When the ring presents a high degree of symmetry, one may relate the distribution of charge with respect to the axis perpendicular to the aromatic plain to that along the main rotational axis and gain a quick characterization of the charge distribution in the molecule [30]. The SI value of the electric quadrupole moment of hexafluorobenzene is 31.7 × 10–40 C m2 (Qzz =+ 9.5 B); 1 B (Buckingham)6 . The schematic representation of such a molecule as two like positive charges, through which the main rotational axis (C6 ), passes, separated by the opposite, balancing negative charges lying perpendicular to the main rotational axis gives a clear picture of its molecular positive quadrupole value. Topologically, the quadrupoles can be considered equivalent to d orbitals, as dipoles are to p orbital [12]. In particular, the Qzz quadrupole of six-membered 6 Debye suggested in the 1960s that the quantity 1 unit of charge distributed over 1 ˚ A2 be termed the Buckingham

Anions and π-Aromatic Systems. Do They Interact Attractively?

141

aromatic rings can be considered equivalent to a dz2 orbital, having a nonspherical charge distribution with regions of relative negative and positive charges. Molecules like C6 F6 having no net dipole have a defined charge distribution, that is, the molecular quadrupole described above. Consequently, the Qzz molecular quadrupole of aromatics has the adequate spatial orientation to be involved in the formation of non-covalent anion-π complexes in which the halide is located above the arene centroid (anion-aryl centroid geometry). The great majority of the theoretical studies on the anion-π interaction deal with complexes displaying this type of geometry in the gas phase. It has been shown conclusively and elegantly by Alkorta [18], and Frontera and Deyà [19] using the molecular interaction potential with polarization (MIPp) energetic partition scheme as discussed above that the two fundamental components contributing to the stabilization energy of non-covalent anion-π complexes in the gas phase are electrostatic and anion-induced polarization. The electrostatic component correlates well with the magnitude of the Qzz of the aromatic ring [31]. In molecules having a very positive Qzz , the electrostatic contribution dominates the anion-π interaction but the polarization energy is not negligible. For example, 27% of the total energy (Et =– 16.5 kcal mol–1 ) calculated using MIPp at the MP2/6-31++G∗ for the interaction of trifluoro-1,3,5-triazine (Qzz =+ 8.23 B) with Cl– when the anion approaches perpendicular to the center of the aromatic ring is due to polarization energy (Epol =– 4.5 kcal mol–1 ). The electrostatic term for this interaction is the major component Eele =– 12.9 kcal mol–1 while the van der Waals term is almost negligible Evdw = 0.9 kcal mol–1 . The computed molecular polarizability of the trifluoro-1,3,5-triazine aromatic ring is α|| = 30.26 a.u. As the Qzz value diminishes, the term of the electrostatic energy (Eele ) becomes less important. Thus, the contributions to the total interaction energy (Et =– 6.6 kcal mol–1 ) of 1,3,5-triazine (Qzz =+ 0.9 B) with chloride are: Eele =– 2.2 kcal mol–1 , Epol =– 4.1 kcal mol–1 and Evdw =– 0.3 kcal mol–1 . It is worth to note that the computed molecular polarizability of the 1,3,5-triazine aromatic ring is α|| = 30.34 a.u. very similar to the fluoride derivative. This relationship corroborates the central role of the quadrupole moment value in anion-π interactions when the anion approaches the aromatic compound perpendicular to the center of the aromatic ring. Since the molecular quadrupole moment describes the electron density of the aromatics, and because a direct correlation between the interaction energy of the anion binding of positive Qzz value and the aromatic Qzz value has been observed, it is reasonable to suggest that the anion-π interaction is governed by electrostatics. Frontera and Deyà have also demonstrated that the contribution due to polarization (Epol ) to the total energy calculated with MIPp increases linearly with the computed molecular polarizabilities of the compounds [31]. This becomes the predominant term of the total interaction energy for compounds having Qzz values lower than 1 B. Overall, however, the contribution due to polarization (Epol ) can be considered as almost constant with a value of 4–6 kcal/mol.

142

P. Ballester

Based on the results of the theoretical calculations, the binding energy of the anion-π interaction is not 100% electrostatic. In fact, as shown in Fig. 9 the fraction of the total binding energy that is electrostatic varies considerably depending on the aromatic. However, the variation of the anion-π binding energies is faithfully mirrored by the electrostatic term (plots a and d). To predict the trend in an anion-π interaction across a series of similar aromatics, all with the same anion, i.e., chloride, it is enough to consider the electrostatic term. It is for this reason that the visual inspection of electrostatic potential surfaces (EPS), which are a good way to visualize the charge distribution of aromatics and consequently the quadrupole moment, provides a simple and reliable guide to the relative strength of anion-π interaction across a series of aromatics interacting with the same anion. The EPSs are also useful to locate the computed minimum geometry of anion-aryl centroid complex stabilized by anion-π interaction, although this geometry is not always observed in the experimental studies (vide infra). The simple consideration of the electrostatic term explains many trends derived from the theoret-

Fig. 9 Plot of the regressions between the quadrupole moments and molecular polarizabilities to: a the electrostatic contribution, b polarization contribution and c van der Waals contribution of the total interaction energy calculated with MIPp for the compounds interacting with chloride at the minimum. d Plot of the regression between the quadrupole moments to the interaction energy at the MP2/6-31++G∗∗ level of theory for the compounds interacting with chloride at the minimum. From reference [31]

Anions and π-Aromatic Systems. Do They Interact Attractively?

143

ical studies although sometimes the major component of the binding energy could be “non-electrostatic”. It is worth to note that across a series of aromatics compound having quadrupole moment values from Qzz = 8.23 to 0.57 B the “non-electrostatic” component of the interaction energy (Epol + EvdW ) makes a contribution of –3 to –7 kcal mol–1 . The binding energy of the anionπ interaction decreases slightly with the increase of the ionic radius of the anions which is also consistent with an electrostatic model. Kim et al. [32] also studied the nature of the anion-π interactions using the symmetry-adapted perturbation theory (SAPT) [33] to obtain a physical interpretation of the interaction energy. In this method, the interaction energy is expressed as a sum of perturbative corrections in which each correction results from a different physical effect. The different intermolecular terms obtained from this method can be summarized in electrostatic, exchange-repulsion, induction, and dispersion contribution. The authors show that for different halogen complexes with several π systems the electrostatic term follows the same trend as the total interaction energy. This tendency can be explained by taking in consideration the fact that both dispersion and induction energies can be ascribed, to a large extent, to the interaction of the molecular orbitals of the anion and the π system. The attractive dispersion and induction energies increase as the diffuse electron cloud of the anion interacts with the substrate. The repulsive exchange interaction also depends on the molecular orbitals overlap. This has the effect of establishing a balance between dispersion-induction and exchange-repulsion energies resulting in a good correlation between the electrostatic energy and the total interaction energy. One of the main conclusions of Kim’s work is that the total interaction energies calculated for the anion-π complexes are comparable to those obtained for the cation-π complexes. This is a fundamental statement also mentioned in other theoretical studies, if one wants to hypothesise on the strength and importance of the anion-π interaction in solution. It has been possible to estimate an energy value in the range of –2 to –0.5 kcal mol–1 for a single cation-π interaction using supramolecular model systems [34–36] and protein engineering studies [37]. As complement to these experimental studies, other experimental gas-phase measurements and high-level theoretical studies have been used to assign a binding energy of approximately –10 kcal mol–1 for the interaction of tetramethylammonium with benzene in the absence of solvent [12]. Frontera and Deyà [38] warn that although it is true that the interaction energies of benzene with cations and hexafluorobenzene with anions are similar, it is not possible to generalize that the interaction energies calculated for the anion-π complexes are comparable to those obtained for the cation-π complexes. These authors also indicate that the same is applicable to Kim’s conclusion stating that the largest contribution in anion-π complexes are electrostatic and induction, because as we have seen before, these contributions sharply depend of the Qzz and α|| values of the aromatic system.

144

P. Ballester

Molecules with negligible Qzz values are expected to interact favorably with either anions or cations, and the strength of the interaction would be comparable, especially if the ionic van der Waals radii of the charged species are similar. Other conclusions from the work by Frontera and Deyà [38] comparing different aspects of the cation-π and anion-π interactions are: a) the contribution of dispersion and correlation terms to the total interaction energy are small, but they are more important in anion-π complexes, b) the density at the cage critical point generated upon complexation of the ion is a useful parameter for measuring the strength of the interaction, even when comparing anion-π to cation-π complexes, and c) a gain in aromaticity of the ring is observed, based on the nucleus-independent chemical shift (NICS) [39] criterion at the center of the ring, upon complexation of the anion, and the contrary is observed for the cation. Many authors agree that the contribution from the dispersion energy is more important in the anion-π complexes than in the cation-π interaction. The induction energy emerges from the interaction of the occupied p orbital of the halide anion and the LUMO of the π-system. The inductive type of the MO interaction can also be correlated to the extent of charge transfer from the anion to the π system. In fact, several experimental studies have established a charge-transfer character to anion-π complexes (see below). The degree of charge transfer in several anion-π complexes has been evaluated using different quantum chemical approaches. The computed charge transfer reported in the work of Frontera and Deyà from the anion (F– ) to the π-system is in the range of 0.1 to 0.2|e| and 0.005 to 0.14|e| in Kim’s study. The results based on AIM methodology indicate that the charge transfer is almost negligible for all complexes studied. Table 5 Binding energies MP2(full)/6-31++G∗∗ //RI-MP2(full)/6-31++G∗∗ (kcal mol–1 ) with (∆EBSSE ) and without (∆E) basis set superposition error correction and anion-aryl centroid distance (r, ˚ A) Complex

∆E

∆EBSSE

r

TFZ· · ·Cl– (TFZ)2 · · ·Cl– (TFZ)3 · · ·Cl– TFZ· · ·Br– (TFZ)2 · · ·Br– (TFZ)3 · · ·Br– TAZ· · ·Cl– (TAZ)2 · · ·Cl– (TAZ)3 · · ·Cl–

–20.3 –38.5 –65.6 –21.8 –41.7 –75.3 –9.0 –17.4 –39.6

–15.0 –28.5 –41.0 –14.2 –26.8 –38.6 –5.2 –10.4 –22.2

3.008 3.006 3.019 a 3.176 3.170 3.172 a 3.220 3.213 3.015 a

a

Mean distance

Anions and π-Aromatic Systems. Do They Interact Attractively?

145

Fig. 10 Schematic geometries of the anion-π complexes studied by Frontera and Deyà [38]

Fig. 11 RI-MP2(full)/6-31++G∗∗ fully optimized structure of the trimeric complex of TAZ with chloride. The distances between the N atom and the C – H bonds of the 1,3,5triazine, as well as, between the anion and the centroid of the aromatic ring are shown

The additivity of the anion-π interactions has also been explored by Frontera and Deyà using high-level ab initio calculations [38]. They optimized chloride and bromide complexes with one, two, and three aromatic units, such as trifluoro-1,3,5-triazine (TFZ) and 1,3,5-triazine (TAZ) – and analyzed the interaction using the AIM theory and studied the charge transfer using several methods for deriving atomic charges. The results revealed additivities in the binding energies and complex geometries that are almost insensitive Table 6 Computed binding energies (∆E, kcal mol–1 ) for TFZ complexes with chloride simulating two solvents and in the gas phase at the MP2(full)/6-31++G∗∗ /RI-MP2(full)/ 6-31++G∗∗ level of theory Complex

∆E (CHCl3 )

∆E (H2 O)

∆EBSSE (gas phase)

TFZ· · ·Cl– TFZ2 · · ·Cl– TFZ3 · · ·Cl– receptor

–5.9 –11.1 –31.7 –15.8

–3.5 –7.0 –24.2 –11.9

–15.0 –28.5 –41.0 –31.0

146

P. Ballester

Fig. 12 CAChe optimized structure of the chloride complex based on the receptor proposed by Frontera and Deyà [38]. The receptor binds the anion via four anion-π interactions

to its stoichiometry (geometry additivity). To speed up the calculations process, the authors used the resolution identity MP2 method (RI-MP2) [40, 41]. They state that the interaction energies and equilibrium distances obtained with the RI-MP2 method in the study of anion-π and cation-π interactions are almost identical to those obtained with the time consuming MP2 calculations. Calculations simulating two solvents systems (CHCl3 and H2 O) within the self-consisting reaction field PCM model using the RI-MP2(full)/6-31++G∗∗ geometries were also performed. The additivity for a set of complexes (TFZ with chloride) is maintained in both solvents, i.e., the binding energy of the 1 : 2 complex is twice the value of the 1 : 1 complex. There is, however, a significant reduction of energy in comparison with the gas phase although this is less important in the 1 : 3 complex. As we discussed in the introduction, the anion-π interaction has potential application in the field of molecular recognition of anions. In this sense, several theoretical studies have proposed structures for novel receptors based

Fig. 13 Molecular structures of the anion receptors proposed by Mascal [13]. The anion is bound by a combination of three ion-pair reinforced hydrogen-bonding and two anion-π interactions

Anions and π-Aromatic Systems. Do They Interact Attractively?

147

on multiple anion-π interactions. Thus, Frontera and Deyà propose a neutral receptor for the binding of chloride that is composed of a neutral platform of 1,3,5-triazine substituted by three 2,4-difluoro-6-methylene triazine groups resulting in a tripodal architecture. The receptor adopts a cup-like cavity that includes and binds the chloride via four anion-π interactions. The theoretical study simulating water and chloroform has been extended to the receptor indicating that this type of receptor could be adequate for the binding of chloride in organic solvents. Mascal [13] has also proposed the synthesis of three novel receptors for the binding of anions that take advantage of the high conformational stability of cylindrophane, which can effectively discriminate guests based on size. The receptors are designed to bind the anion through a combination of ion-pair reinforced hydrogen bonds and two anion-π interactions. Mascal performed a detailed theoretical study of the association of the triazine cage and its cyanuric acid and boroxine analogues with the fluoride and chloride anions. Since the principal motivation of the work is the selective complexation of anions within the cages, the author compared the A average) for the empty rebisector distance (Ar – Ar distance/2 = 2.30 ˚ ceptors with the anion-aryl centroid distance of halide sandwich complexes A and raverage Ar···F– = 2.45 ˚ A), which are analogous to (raverage Ar···Cl– = 3.21 ˚ the 1 : 2 complex studied by Frontera and Deyà. The energies calculated for the sandwich complexes are approximately additive, which is in agreement with the work of Frontera and Deyà. The comparison of the bisector distances with the anion-aryl centroid distance of the sandwich complexes suggest that both chloride and fluoride are too large for all cage-like receptors. In the three receptors, however, the optimal fluoride distance is better accommodated for the empty cavity. Table 7 Interaction energies (kcal mol–1 ) for the complexes of the triazine cages (TAZ) and its analogues cyanuric acid (CNA) and boroxine (BOX) with fluoride and chloride ion Complex

∆EBSSE

TAZcage· · ·Cl– CNAcage· · ·Cl– BOXcage· · ·Cl– TAZcage· · ·F– CNAcage· · ·F– BOXcage· · ·F–

–237.5 –245.0 –245.8 –292.6 –296.3 –295.6

a

a

∆EF–Cl b

∆EBSSE H2 O

–55.1 –51.4 –40.7

–1.1 –13.4 –21.8 –30.0 –36.9 –35.7

a

∆EF–Cl(H2 O) b

–28.9 –23.5 –13.9

BSSE corrected B3LYP/6-31+G(d, p) interaction energies in the gas phase (∆EBSSE ) and in an aqueous solvent model (∆EBSSE H2 O ) b Differences between the energy values of the fluoride and chloride complexes

148

P. Ballester

The binding properties of the halide complexes (F– , Cl– ) were evaluated in the gas phase and solution. Energies were also determined using the conductor-like polarisable continuum model for water, with the molecular cavity specified by the United Atom Topological Model applied on radii optimized for the PBE0/6-31g(d) level of theory. The complexation of fluoride with the three receptors is calculated to be more energetically favorable than chloride by >40 kcal mol–1 in the gas phase and >13 kcal mol–1 in water. It is worth noting that the complexation energies calculated for F– ion are about three times the experimental enthalpy of F– hydration. The author indicates that the raw comparison of these values is of limited predictive significance in regard to the potential of these receptors to extract F– from water, given the absence of activation and entropy values, as well as, the assumptions inherent to nonexplicit, polarized continuum models of water. The comparison of relative energetics of complexation, however, clearly indicates a preference for fluoride binding over chloride, from the gas phase to water, in all three receptors (see Table 7). The principal interaction force in these complexes is the ion-pair reinforced hydrogen bond of the ammonium groups with the anions. It may be supposed that the intrinsic stronger ammonium· · ·F– interaction could actually form the basis for the difference in binding energy. The isolation of the relative binding contributions (ion-pair reinforced hydrogen bond and anionπ interactions) was achieved by modelling the association of three NH4 + ions in a trigonal plane around a chloride and fluoride ions. The obtained results are shown in Table 8. The high differences in binding energies calculated for the cage complexes and the modelled NH4 + trigonal systems are clearly due to the fact that the three ammonium groups are enforced in close proximity in the free receptor. The point to note is that although the stabilization of three NH· · ·F– bonds in the trigonal (NH4 + )3 · · ·F– complex is about 11 kcal mol–1 greater than the corresponding (NH4 + )3 · · ·Cl– complex, in the triazine and cyanuric acid cage complexes there is still >10 kcal mol–1 “additional” stability in the fluoride complexes in aqueous solution. This observation suggests that the discrimination has to do with a better fit of the anion at least in these two Table 8 Interaction energies (kcal mol–1 ) for NH4 + · · ·X– complexes Aggregate

∆EBSSE

(NH4 + )3 · · · Cl– (NH4 + )3 · · · F–

–130.0 –161.9

a

a

∆EF–Cl b

∆EBSSE H2 O

–31.9

–10.8 –21.6

a

∆EF–Cl(H2 O)

–10.8

BSSE corrected B3LYP/6-31+G(d, p) interaction energies in the gas phase (∆EBSSE ) and in an aqueous solvent model (∆EBSSE H2 O ) b Difference between the energy values of the fluoride and chloride complexes

Anions and π-Aromatic Systems. Do They Interact Attractively?

149

Table 9 Interaction energies (kcal mol–1 ) and noncovalent bonded distances for selected complexes of 1,3,5-triazine (TAZ) and boroxine (BOX) with chloride and fluoride anions Complex

∆E a

∆EBSSE

TAZ· · ·Cl– TAZ· · ·F– BOX· · ·Cl– BOX· · ·F– TAZ· · ·Cl– · · ·TAZ TAZ· · ·F– · · ·TAZ BOX· · ·Cl– · · ·BOX BOX· · ·F– · · ·BOX TAZcage BOXcage

–4.2 –10.4 –12.9 –31.9 –8.1 –18.9 –23.1 –47.2

–4.0 –9.0 –12.5 –29.6 –7.6 –16.5 –22.1 –43.9

a

∆EF–Cl b

–5.0 –17.1 –8.9 –21.8 –55.1 –40.7

rc 3.48 2.66 2.94 1.93 3.49 2.79 3.02 2.23

a B3LYP/6-31+G(d, p) interaction energies without (∆E) and with (∆EBSSE ) BSSE correction b Difference between the ∆EBSSE values of the fluoride and chloride complexes c Noncovalent bond distance (r) for simple (Ar· · ·X– ), sandwich (Ar· · ·X– · · ·Ar). Corresponding data for the cage complex can be found in Table 7

Fig. 14 Structure of the tweezers studied in [42]

Table 10 Calculated interaction energies (kcal mol–1 ) of the investigated tweezer-ion complexes Complex

∆Ecomplex a

∆Einter-MP2 b

6FT 10FT 14FT

5.20 –0.28 –5.95

–4.06 –9.72 –15.82

a b

Obtained by using B3LYP functional ∆Ecomplex = ∆Einter + ∆Edeform Obtained by performing single point MP2 calculations on the optimized geometries

150

P. Ballester

cages. Surprisingly, the order of complementarity or selectivity for F– vs Cl– is TZNcage > CNAcage > BOXcage, which is completely reverse to the discrimination of the halide ions observed in simple binding complexes. Hermida-Ramón and Estévez [42] have recently conducted a theoretical study showing that fluorinated tweezers, that derive from those synthesized by Kläner [43] were able to bind anions. Kläner tweezers are generally used for the molecular recognition of eletrodeficient aromatic and aliphatic substrates as well as organic cations. Several complexes formed between the fluorinated tweezers derivatives and an iodide anion were characterized. The nature of the interaction was analyzed using the SAPT method while the energetics for the complexation process were computed using calculations at the MP2 level on complex geometries that were optimized using density functional theory (DFT) and the B3LYP functional. The differences between MP2 and B3LYP interacting energies can be considered as a rough approximation of the importance of the dispersive interactions in the stability of the tweezeranion complex. The values of the molecular electrostatic potential in the center of the cavity are –10.25 (6 FT), –5.79 (10 FT) and –0.64 (14 FT) kcal mol–1 . Clearly, the best receptor to accommodate iodine in its cavity is 14 FT, in which the MEP is almost neutral. An increase in the fluoride substitution produces a stabilizing trend in the energetics of the complexes. A comparison between the MEP value and the binding energies indicates that the stability comes from the depletion of the electrostatic repulsion between the anion and the π-clouds of the aromatic rings, which have a lower electron density owing to the influence of fluorine. The decrease in electrostatic repulsion is accompanied by a decrease in the distance between anion and the tweezer together with an increase in attractive energies (induction and dispersion), which gives more stability to the complexes. Calculations that include a polarizable continuum model of the solvent (H2 O) indicate that the complex is not stable. It must be noted that for the I– @14 FT complex, a large attractive binding energy is calculated even though the electrostatic potential inside the cavity is slightly negative. Lewis and Clements [44] have performed quantum mechanical computations that show that negative Qzz aromatics bind anions in the gas phase, dispelling the idea that the π electron density of negative Qzz aromatics is only appropriate for cation binding. No correlation was observed, however, between anion binding enthalpies and the Qzz values for negative Qzz aromatics. This observation is in striking contrast to what has been reported and mentioned above for positive Qzz aromatics binding anions. The authors do indicate that the Morokuma-Kitaura decomposition calculations show that the major contribution to binding in the case of negative Qzz aromatics with anions is the energy due to polarizability of the aromatic ring. They also mention that it may be not the polarizability of the aromatic ring density that is responsible for the binding but rather it may be the polarizability of

Anions and π-Aromatic Systems. Do They Interact Attractively?

151

Fig. 15 Schematic representations of the geometries for Cl– complexes with TCBm

the aromatics substituents. The work concludes indicating that even electron rich, negative Qzz aromatics should be considered experimentally tenable for anion-π interactions. We believe that this statement should be taken with serious caution. Johnson, Hay et al. [45] have refined the nature of the interactions between electron deficient arenes and halide anions. In particular, they have performed calculations at the MP2/aug-cc-pVDZ level of theory of 1 : 1 complex formed between 1,2,4,5-tetracyanobenzene (TCB) and F– , Cl– , Br– anions. Four geometries were evaluated for each halide. The well-known anionaryl centroid (A) (termed non-covalent anion-π complex in this study), two charge transfer (CT) complexes in which the halide is positioned above a C – H bond (B) or above a C – CN bond (C), and a C – H hydrogen bond complex (D) similar to the one already discussed by Mascal in his seminal study of the interaction of triazine with halide anions [17]. The theoretical analysis of TCB halide complexes revealed the first example in which noncovalent anion-aryl centroid complexes are not stable for Cl– and Br– . These halides do interact with the π system, but the interaction involves CT, characterized by a high second-order stabilization energy ∆E(2) . The resulting optimized geometries of the CT complexes locate the anion over the periphery ring rather than over the center of the ring. Since as we have presented above prior computational studies on Cl– and – Br have focused almost exclusively on the anion-aryl centroid complex, there has been no investigation to determine the existence of such CT complexes for other arenes. Consequently, the authors expand the electronic structure calculations to halide complexes with triazine, hexafluorobenzene and 1,3,5-

152

P. Ballester

Fig. 16 Calculated EPS at the HF-6-31G level for TCB. The range of energy values is –39 (red) to +39 (blue) kcal mol–1

tricyanobenzene. The results with F– and triazine are fully consistent with earlier work [17]. Although the anion-aryl centroid and hydrogen-bonding geometries have been previously identified as minima for Cl– and Br– [17], none of the previous studies report the existence of an “attack” geometry for either Cl– or Br– only for F– . Two geometries were located for hexafluorobenzene, the anion-aryl centroid which is not a stable point for F– and the “attack” geometry that is the global minimum for F– but was not located as a stable point for Cl– or Br– . These results are in complete agreement with prior theoretical studies [18, 19]. Results obtained for 1,3,5-tricyanobenzene are similar to those obtained for TCB. The anion-aryl centroid geometry is only stable for Br– . The CT complex that locates the anion above a C – H bond is the global minimum for the three halides. The hydrogen bond geometry is also a stable structure for all the halides. All anion-aryl centroid complexes have a low extend of charge transfer. Alternate geometries do have a high extend of charge transfer. It is worth noting that for the interaction of F– with triazine in the “attack” geometry, the computed values for the parameters used to gauge the extent of charge transfer, indicate the formation of a strong σ bond. In conclusion, triazine, TCB and 1,3,5-tricyanobenzene form stable off-center CT complexes with Cl– and Br– . Except for the CT complex of Br– with triazine, the CT complexes are more stable than the anion-aryl centroid complexes. The differences in energies, however, are very small (<1 kcal mol–1 ), indicating a relative flat potential surface for positioning the halide above the arene plane. In fact, the EPS of TCB shows a widespread distribution of positive electrostatic potential values all over the π-system.

3 Experimental Evidence of the Anion-π Interaction 3.1 Solution and Related Crystallographic Studies To the best of our knowledge, the first report suggesting the experimental existence of attractive interactions between anionic negative charges and π-electron aromatic systems was published by Schneider et al. in 1993 [9]. The 1 H NMR measurement of the association constant between a simple

Anions and π-Aromatic Systems. Do They Interact Attractively?

153

Fig. 17 Schematic representation of the complex formed with the cleft-like para-sulfonato receptor and diphenylamine placing the ionic part above the aromatic guest center

diphenylmethane organic host bearing negatively charged para-sulfonato groups and an electroneutral diphenylamine show in aqueous solution weak but attractive interactions that were quantified to reach approximately 0.5 kcal mol–1 per X– /arene unit7 . The small complexation induced shift observed for the proton Hp (0.09 ppm) is interpreted by the authors as the existence of a displacement of the aryl rings in the complex. The authors state “such a slight displacement will replace repulsions between permanent negative charges of the π-cloud and of the sulfonato substituent by attractions, as the π-cloud can still be polarized by the charge of the sulfonato even if the charge is not exactly above the arene center”. In other words, although anion-π interactions are claimed to be responsible for the stability of the complex, the proposed geometry does not coincide with the one mainly investigated by the theoretical calculations that locates the anion above the π-aromatic systems. The nature of the attraction is primarily assigned to the interaction between the anion and the dipole that it induces in the arene (polarization). While the theoretical studies have revealed that this “nonelectrostatic” term maybe important, it seems that it’s not the defining feature of the anion-π interactions. In this and another report [46], Schneider claims that the anion-π interaction can have almost the same size as the cation-π interaction. In 2004, an experimental study on the solid-state and solution behavior of halide complexes with a series of highly electron-deficient arenes (Fig. 18) also suggests that the complex geometry investigated theoretically, wherein the anion is located along the principal axes of the π-system, may not be operative in these highly electron deficient arene systems [47]. In this study, the recognition of halide anions X– (X = Cl, Br, I) is clearly established by the isolation and X-ray structure determination of a series of well-defined complexes containing the halide salt and the admixed aromatic 7 This study will suggest anion-arene interactions involving aromatics with negative Q zz values. A note in reference [44] states based: “on the systems studied and figures presented in the publication it appears that the binding energy was probably due to arene-arene face to face interactions”.

154

P. Ballester

Fig. 18 Molecular structures of the neutral organic π-acceptors investigated in the study [47]

compounds in different stoichiometries, as well as by the spectral assignment of diagnostic charge transfer absorption bands to the formation of the anion-π complex. Intense colorations were observed upon the addition of the anions as the corresponding tetraalkylammonium salts to acetonitrile solutions containing the aromatic π acceptor, i.e., tetracyanopyrazine TCP (calculated Qzz = 18.53 B) [48]. A new absorption band appears and grows with increasing halide concentration. Close inspection of the new absorption band reveals that it consists of two Gaussian components. The stability con-

Fig. 19 Mulliken dependence of the energy of the low-energy absorption band (νCT ) in the TCP/Hal– complexes and the oxidation potential of the anion

Anions and π-Aromatic Systems. Do They Interact Attractively?

155

Table 11 Solid-state characteristics of the halide complexes obtained by slow diffusion of hexanes into 1 : 1 mixtures of neutral organic π-acceptors with the corresponding halides as tetraalkylammonium salts in CH3 CN/CH2 Cl2

TCP/Br– TCP/I– TCP/Cl– o-CA/Br–

Molar ration

Counterion

X– · · · C [˚ A] a

3:2 4:1 2:1 1:1 4:1 1:1

Et4 N+ Pr4 N+ Et4 N+ Bu4 N+ Bu4 N+ Pr4 N+

3.16 3.15 3.52 3.49 b 3.07 2.93

The X– · · · C distance in closest contacts; note that the sums of the van der Waals radii A (Br– · · · C), and 3.65 ˚ A (I– · · · C) are 3.45 ˚ A (Cl– · · · C), 3.55 ˚ b The average of the distances to the two or three neighbouring acceptors is given a

stant values calculated for the complexes formed in acetonitrile with TCP and Br– as different tetraalkylammonium salts are in the range of 7–9 M–1 . Furthermore, a job plot indicates a 1 : 1 stoichiometry for the TCP/Br– complex formed in solution. A clear correlation between the energy of the low-energy band and the oxidation potential of the anion was also observed and used to establish a charge-transfer character for the complexes. Furthermore, the study of the interaction of Br– with the series of neutral organic π-acceptor reveals that the increase of acceptor strength of the aromatic (characterized by a positive shift of the reduction potential) is accompanied by a bathochromic shift of the new absorption band (from 355 to 465 nm). This observation further confirms the charge-transfer (CT) character of these complexes. The isolation and X-ray characterization of single crystals containing the halide salt and the aromatic compound allows the identification in the solidstate of the non-covalent anion-π complexes. The anion is located in the space

Fig. 20 Local packing of the solid-state structures of the Br– complexes with TCP. Left: Bromide used as tetraethylammonium salt. Right: Bromide used as tetrabutylammonium salt. Only molecules that are in short contact (< sum of vdw radii) with respect to the Br– atom are shown. In the case of the [(TCP)3(Br– )2 ](Et4 N+ )2 complex, the two different Br– ions located within the asymmetric unit are represented

156

P. Ballester

close to the electron deficient aromatic ring explaining the observed elecA tronic (charge-transfer) transitions. The bromide ion lies approximately 3 ˚ over the periphery of the aromatic ring. The observed complex geometry differs from those derived from quantum-mechanical calculations that usually show the anion placed along an axis perpendicular to the aromatic ring and passing through its centroid. The existence of multiple halide-π interactions becomes evident either directly in the asymmetric unit or after the packing of the lattice. The complex stoichiometries that can be derived from the X-ray molecular formula of the asymmetric unit i.e., [(TCP)3 (Br– )2 ](Et4 N+ )2 (1.5 : 1), [(TCP)4 Br– ]Pr4 N+ (4 : 1) and [(TCP)I– ]Bu4 N+ (1 : 1), [(TCP)2 I– ]Et4 N+ (2 : 1), as well as, the number of contacts established between the anion and the aromatic rings seems to be controlled not only by the anion itself but by the size of its countercation, which should influence the three-dimensional solid structure. Johnson, Hay et al. [45] have used 1,2,4,5-tetracyanobenzene (TCB) to gain further structural information of the anion-π interaction. TCP, 18crown-6 and alkali halide were dissolved in acetonitrile (KBr) or 9 : 1 dichloromethane:acetonitrile (KI, NaI) mixture. The purpose of the crown ether is to enhance the solubility of the salt. It is important to note that the solution of the crown ether and TCB is colorless, however, the addition of the halide salt produces a color change consistent with the formation of CT complexes and in complete agreement with Kochi’s previous observation using TCP instead. Slow evaporation at room temperature yielded single crystals suitable for X-ray analysis. Despite packing differences, the local environment about each halide is remarkably similar in all crystals. As shown in Fig. 21, four TCB molecules surround the anion at interaction distances. Three different orientations can be distinguished for the anion: above the arene plane nearest to a C atom bearing a CN group (a and b), above the arene plane nearest to a C atom bearing a H atom (c), and nearly within the plane of the arene, forming a C – H hydrogen bond (d). This latter orientation was not observed in the study of Kochi. In conclusion, Kochi’s and Johnson’s results clearly

Fig. 21 Local packing of four TCB molecules around the anion. The closest contact to each arene is indicated by a dotted black line

Anions and π-Aromatic Systems. Do They Interact Attractively?

157

show that the anions are not located over the center of the arene (anion-aryl centroid geometry) in the solid state structures of these complexes. As discussed above, theoretical studies indicate that TCB and 1,3,5-tricyanobenzene are capable of forming stable off-center CT complexes with Cl– and Br– . In 2003, Hoffmann et al. [49] introduced an uncharged host for the selective binding of Cl– as the tetrabutylammonium salt in CDCl3 : DMSO 95 : 5 – /K – = 105). This host features a triazine-trione platform with three short (KCl Br arms that are conformationally preorganized through the introduction of methyl groups. The three arms are equipped with p-nitrophenylsulfonamide groups capable of hydrogen-bonding to the anion. In this study, they also used an analogous receptor with slightly longer side chains in order to evaluate how the relative position of the anionic host with respect to the triazine-trione platform (anion-π interaction) could affect the stability of the complex. The binding parameters of both hosts with chloride were determined using isothermal titration calorimetry. The microcalorimetry experiments re-

Fig. 22 Molecular structures of the receptors based on the triazine-trione platform having three arms of different length terminated with p-nitrosulfonamide hydrogen-bonding groups Table 12 Thermodynamic binding parameters determined using ITC for the complexation of tetrabutylammonium chloride by the tris-sulfonamides in CHCl3 at 298 K

K/L mol–1 ∆G/kcal mol–1 ∆H/kcal mol–1 – T∆S/kcal mol–1

Host long arm

Host short arm

81 000 ± 10 000 –6.7±0.1 –4.7±0.1 –2.1±0.1

155 000 ± 11 000 –7.1±0.1 –2.2±0.1 –4.8±0.1

158

P. Ballester

vealed that the host with the longer arms binds chloride slightly stronger that the one with the shorter arms (∆∆G ≈ 0.4 kcal mol–1 ). There are, however, more important differences in the binding processes of the hosts to the tetrabutylammonium chloride. Although both complexation processes of chloride by the hosts are entropically favored, due to the release upon binding of ordered solvent molecules from the surface of the chloride and the host to the bulk, the entropic change (– T∆S) for the host with longer arms is far less negative. This result may be the consequence of a higher reduction in the rotational entropy of the host with the long arms on complex formation. In terms of enthalpy, the binding of chloride by the host with short arms is clearly less exothermic. In order to explain the observed result, the authors speculate about the relationship between complex geometry and enthalpy. The more flexible host could adopt a conformation that is more optimal for coordination of chloride giving rise to higher exothermicity. Furthermore, they conclude that from the measured enthalpy values of complexation a potential electrostatic attraction of chloride by the triazinetrione platform in the receptor with the short arms, an anion-π interaction, can be discarded. We believe that the thermodynamic parameters measured for the chloride complexation with these systems are highly contaminated with solvent effects. If the number of solvent molecules released to the bulk on complexation is different for each host, then the comparison of the data is completely inappropriate. In terms of free energy of binding, the experimental results indicate that both chloride complexes formed with the two tridentate p-nitrophenylsulfonamide hosts have approximately the same binding energy and the potential electrostatic attraction of chloride by the triazinetrione (cyanuric) platform is not reflected. In 2005, Frontera, Saczewski, Deyà et al. [50] performed a combined crystallographic and computational study of the anion-π interactions with

Fig. 23 Minimized structures (Maestro, MMFFF) of the chloride complexes formed with the tridentate p-nitrophenylsulfonamide having long and short arms. The distance of the chloride atoms to the centroid of the cyanuric acid ring is shown in each case

Anions and π-Aromatic Systems. Do They Interact Attractively?

159

Fig. 24 Synthesis of the halide salts of the 2-ethyleneamine derivatives of thio- and dithiocyanuric acid

cyanuric acids [50]. These authors interpreted the experimental results previously obtained by Hoffmann et al. making use of MIPp calculations. The MIPp value for the interaction of cyanuric acid with a chloride anion placed A apart from the ring centroid is almost the same to that computed at 2.5 ˚ A (≈–16 kcal mol–1 ). Consequently, a likely explanation for a distance of 3.5 ˚ the similarity of free energies of complexation measured experimentally with long or short arms tridentate p-nitrophenylsulfonamide hosts is that the effect of the cyanuric platform on the anion binding is similar and does not allow to differentiate one with respect to the other. To obtain experimental evidences of the ability of cyanuric acid to interact favorable with anions, Frontera, Saczewski, Deyà et al. [50] synthesized several derivatives of thiocyanuric and dithiocyanuric acid with a flexible 2-ethyleneamine arm attached to the s-triazine ring. The one-step synthesis of these compounds requires a 1–4 h reflux in aqueous 12% HX (X = Cl, Br,

Fig. 25 Top view and side view of the local packing of the chloride in the crystal structure of the hydrochloride salt of the 2-ethyleneamine thiocyanuric acid derivative

160

P. Ballester

Fig. 26 Top view and side view of the local packing of the chloride in the crystal structure of the hydrochloride salt of the 2-ethyleneamine dithiocyanuric acid derivative

Fig. 27 Top view and side view of the local packing of the bromide in the crystal structure of the hydrochloride salt of the 2-ethyleneamine dithiocyanuric acid derivative

Fig. 28 Top view and side view of the local packing of the iodide in the crystal structure of the hydrochloride salt of the 2-ethyleneamine dithiocyanuric acid derivative

I) acid. After cooling the reaction mixture to ambient temperature, colorless crystals suitable for X-ray analysis of the corresponding hydrohalide salts of the amine were obtained.

Anions and π-Aromatic Systems. Do They Interact Attractively?

161

The X-ray crystal structures of all the hydrohalide salts except the hydrobromide of the thiocyanuric acid derivative revealed the existence of anion-π interactions. The organic cation, which has several hydrogen-bond donor and acceptor functionalities, displays an electrostatic interaction between the N atom of the ammonium moiety and an anion (X– ) located nearly above the center of the electron-deficient ring. The packing of the lattice reveals that the anion (X– ) placed on top of the aromatic rings also interacts electrostatically with at least one ammonium group of an adjacent cyanuric acid derivative. Additionally, one of the NH groups of the s-triazine ring of another adjacent organic cation binds the anion (X– ) through a hydrogen-bond interaction. Finally, two other molecules of the organic cation are also surrounding the anion. One of them has the 2-ethyleneamine arm positioned at short contact of the anion (distance < sum of vdW radii). A The distance between the ring centroid and the chloride anion is 0.1 ˚ longer in the dithione than in the monothione derivative. In conclusion, the X-ray structures of the halide salts of mono- and dithiocyanuric acid derivatives having an ethylenenammonium arm attached to one of the nitrogen atoms of the s-triazine do show the existence of anion-π accompanied by saltbridge and hydrogen-bonding interactions as a robust structural motif of the solid-state packing. In an attempt to probe the efficacy of the non-covalent anion-π interaction, Johnson et al. [51] have prepared two sulfonamide-derived receptors (A and B in Fig. 29). The design of the receptor is based on a convergent two-point recognition motif utilizing both a hydrogen bond and an aromatic ring that can be involved in the formation of anion-π interactions. The prepared receptors differ in the electronic properties of one of the rings of the biphenylamino moiety due to the substitution with five fluorine atoms. The quadrupole moment of such an aromatic ring should be highly positive while the simple phenyl should have a negative quadrupole moment. Receptor B has a pentafluoro substituted aromatic ring that is electrondeficient and more appropriate to engage in an attractive anion-π interaction.

Fig. 29 Molecular structures of the receptors used by Johnson et al. to evaluate the anionπ interaction in solution

162

P. Ballester

Fig. 30 CPK representation of the single-crystal X-ray structure of the pentafluorophenyl receptor. The receptor adopts in the solid-state a conformation that is adequate to interact simultaneously through hydrogen-bonding (sulfonamide NH) and anion-π interaction with an halide located on top of the pentafluorophenyl ring Table 13 Ka (M–1 ) for the sulfonamide receptors with Cl– , Br– and I– Anion

Phenyl receptor (A)

Pentafluorophenyl receptor (B)

Cl– Br– I–

<1 <1 <1

30 ± 3 20 ± 2 34 ± 6

The tetraethylammonium salts of each anion were used

The stability constants of the complex formed with receptor B and a series of anions (Cl– , Br– , I– ) were evaluated in CDCl3 using NMR titration techniques. Receptor B binds all the screened halides with a measurable but modest association constant. The association constants measured were 20 M–1 for

Fig. 31 X-ray crystal structure of the complex formed between the pentafluorophenyl receptor B and tetra-n-butylammonium bromide. The sulfonamide receptor and the organic cation are shown in stick representation while the bromide is represented with van der Waals surface

Anions and π-Aromatic Systems. Do They Interact Attractively?

163

Br– , 30 M–1 for Cl– and 34 M–1 for I– . On the other hand, the association constant for receptor A lacking the electron-deficient aromatic ring required for the anion-π interaction, were too small to be determined using the same titration methodology. The enhanced affinity exhibited by receptor B over receptor A with the array of halides is attributed to the existence of anion-π interaction in the anion complex. The experiments presented by Johnson et al. support further studies on the possible use of the anion-π interaction as an emerging no covalent interaction of the selective targeting of anions in solution. The difference in the reported stability constants (∆∆G) can be used to estimate a value for the anion-π interaction in the range of approximately –2 kcal mol–1 . This value almost doubles the value estimated for the cation-π interaction using protein engineering [37] (–2.75 kcal mol–1 for a cavity lined by three π-systems – 2.75/3 = 0.9 kcal mol–1 ) and supramolecular chemical model systems [35] (–2.4 kcal mol–1 for a receptor with four π-systems – 2.4/4 = 0.6 kcal mol–1 ). Probably, the anion-π interaction is over-evaluated when using the chemical model based on sulfonamides. Johnson et al. performed the 1 H NMR titrations experiments in CHCl3 , and the literature indicates that tetraalkylammonium salts form tight ion pairs with small anions, i.e., Cl– in CHCl3 solutions [34, 52]. Consequently, the sulfonamide receptors bind the tight ion-pair and the observed enhancement of the association constants for the pentafluorophenyl receptors could be partially caused due to the existence of electrostatic interactions between the fluorine atoms and the tetraalkylammonium salt. The X-ray crystal structure of the pentafluorophenyl receptor and tetra-n-butylammonium bromide, in which the ion-pairing in the solid state is held responsible to force the sulfonamide NH away from the preferred conformation, shows close contacts between the fluorine atoms and the organic cation (Fig. 21). 3.2 Further Crystallographic Evidence of Anion-π Interactions Examples of crystal structures of supramolecular complexes in which anions are nested in the interior of aromatic cavities have been used as evidence for the existence of an attractive anion-π interaction. In 2004, the first crystallographic evidence of anion recognition by aromatic receptors was described by Meyer et al. [53], and Gamez, Reedijk et al. [54]. While investigating the structural and magnetic properties of metal complexes with the hexakis(pyridine-2-yl)-[1,3,5]triazine-2,4,6-triamine ligand (L), Meyer synthesized a copper (II) chloride complex that showed chloride-triazine binding with geometrical parameters almost exactly to those calculated computationally for the 1,3,5-triazine· · ·Cl– complex. The cationic moiety of [L2 (CuCl)3 )]3+ consists of two ligands L that are stacked in a parallel fashion and held together by three copper(II) ions. The four pyridine-N groups and

164

P. Ballester

Fig. 32 Left: molecular structure of the ligand (L). Right: side view and top view of the copper(II) chloride complex, few solvent molecules, some hydrogen and chloride atoms are omitted for clarity. The atoms Cl(8) and Cl(5) are shown in CPK representation. Special emphasis is placed on the location of these two atoms in relation to the triazine ring of the cationic moiety of the complex [L2 (CuCl)3]3+

an apical chloride constitute a square pyramidal coordination of each metal center. Cl atoms of Cl2 CH2 are placed between the paddles of the structure. Both triazine rings are arranged in almost perfect face-to-face arrangement. The most interesting structural feature related to the anion-π interaction is the position of the chloride atoms 8 and 5 belonging to [CuCl4 ]2– ions. Cl(8) is located above one of the triazine rings in an anion-aryl centroid geometry. A while The distance between the centroid of the ring and the anion is 3.17 ˚ the angle Cl– · · ·centroid axis to the plane of the ring is 87◦ , that is, the chloride is almost located on the C3 axis above the ring. The Cl(5) chloride shows a similar geometry.

Fig. 33 Side view and top view of a fragment of the crystal structure of the Cl– complex reported by Meyer. The six hydrogen-bonding interactions with the aryl C – H groups of the Cu(II) coordinated pyridine rings are indicated. C – H· · ·Cl– angles ≥150◦ and C to A Cl– distances range from 3.9 to 4.3 ˚

Anions and π-Aromatic Systems. Do They Interact Attractively?

165

Fig. 34 Left: Molecular structure of azadendritz. Right: side views and top view of the crystal structure of the Cu(II) complex of azadendritz reported by Gamez, Reedijk et al. [54]. Solvent molecules and hydrogen atoms are omitted for clarity. The structure shows the four pentacoordinate Cu(II) ions coordinated two apical chloride atoms (scaled ball-and-stick representation) and two encapsulated chloride anions (CPK representation)

The authors state that a “Cl– · · ·triazine complex due to electrostatic anionπ interaction is present in the complex”. Johnson, Hay et al. [45] have recently noticed that the role of the C – H groups of deficient arenes as potent hydrogen bonds should not be overlooked, in fact the structure of Meyer nicely illustrates this observation. The Cl(8) anion positioned above the center of the melamine is interacting with the C – H groups of the Cu(II)-coordinated pyridines. The hydrogen bonds formed may play a dominant role in determining the position of the anion within this cavity. Almost at the same time, Gamez, Reedijk et al. [54] described the first coordination compound of the ligand azadendritz. The supramolecular Cu(II) complex shows intramolecular π–π interactions between two 1,3,5-triazine rings. Similar to the structure reported by Meyer, four pyridine-N groups and an apical chloride constitute a square pyramidal coordination of each metal center. In this case, however, the chloride atoms are not located on top of the 1,3,5-triazine rings, which are perfectly stacked, instead the anion is nestled against four aromatic pyridyl residues. In Meyer’s structure, this same position was occupied by Cl atoms of Cl2 CH2 molecules. Each encapsulated chloride is in close contact with the four pyridine rings with A. It has to be noted that centroid· · ·Cl– distances ranging from 3.5 to 3.7 ˚ in both examples the electron-poor character of the pyridine moieties is probably enhanced by their coordination to the Cu(II) metals. In addition to the π interaction with the pyridine rings the authors noted that the chloride ions are in close proximity to the triazine rings, suggesting the existence of some additional electrostatic interactions with the electron deficient triazine moiety. The angles of the Cl– · · ·centroid axis to the plane of the different pyridine rings ranges from 74◦ to 82◦ . This result encouraged the authors to investigate further the anion · · · triazine interaction

166

P. Ballester

using other triazine-based ligands and anions [55, 56]. Thus, the reaction of Zn(NO3 ) · 6H2 O or Cu(NO3 ) · 3H2 O with the ligand dipicatriz in acetonitrile afforded mono- or trinuclear coordination complexes depending on the metal-to-ligand ratio used during the crystallization process [56]. In particular, the trinuclear Zn(II) complex shows anion-π interactions between two nitrate anions coordinated to two different zinc atoms and the same triazine ring, one in each face of the aromatic ring. On the other hand, the corresponding trinuclear Cu(II) complex exhibits an even more remarkable double anion-π interaction between the triazine ring and two nitrate anions. In this case, the two nitrate anions are now uncoordinated. One of the nitrates interacts via a somewhat longer distances, which is most likely due to the fact that the triazine is already involved in a nitrate-centroid contact. These examples constitute one of the first reports on anion-π-anion interaction. It is also worth noting that the observed spatial arrangement of the nitrate anion with respect to the triazine ring is distinct to the one proposed by Kim et al. [32] based on theoretical calculations. Gamez, Reedijk et al. [56] performed ab initio calculations demonstrating that the interaction energies calculated for this peculiar positioning of the nitrate is comparable to the energy of the binding geometry proposed by Kim et al. in his theoretical studies of the interaction of triazine with nitrate [32]. In the model proposed by Kim et al., the anion is parallel to the triazine ring and the oxygen atoms are located on top of the electropositive carbon atoms of the ring. Related to the previous example Gamez, Reedijk et al. [55] also found crystallographic evidences of another geometry for the nitrate-π interaction in the tetranuclear complex [Cu4 (dpatta)(NO3 )4 ](NO3 )4 obtained under hydrothermal conditions from the reaction of copper(II) nitrate and the ligand dpatta [55]. Each two triazine units are perfectly π–π stacked but A. Both triazine rings staggered with a centroid-to-centroid distance of 3.45 ˚

Fig. 35 Left: Molecular structure of dipicatriz. Right: side views of the crystal structures of the Zn(II) and the Cu(II) trinuclear complexes of dipicatriz. The structures show the three coordinate Zn(II) and Cu(II) ions in scaled ball-and-stick representation and the distances between the centroids of the triazine ring and the closest oxygen atoms of the interacting nitrate anions

Anions and π-Aromatic Systems. Do They Interact Attractively?

167

are alternatively involved in an interaction with a lattice nitrate but with a different geometry. A DFT analysis was performed with a model complex triazine · · · nitrate to analyze the anion-π interaction. In particular, the preference for the coordination of the nitrate ion observed in the X-ray structure with one oxygen pointing to one nitrogen of the triazine ring. A DFT-B3LYP geometry optimization indicates that a local minimum exists with a geometry very similar to that observed in the X-ray crystal structure. The plane of the nitrate is not completely parallel to the triazine ring and one of the oxygen atoms points in the direction of one of the N atoms of the ring. This calculated structure is 0.7 kcal mol–1 more stable that Kim’s proposed geometry. Furthermore, the dpatta Cu(II) complex represents one of the first crystallographic evidences of anion-π–π interactions (nitrate-triazine-triazine). The computed energy for the optimized nitrate-triazine-triazine complex is only 0.46 kcal mol–1 higher than the complex including one ring. Due to this low energy difference, it is difficult to draw any conclusions about the stabilization of the ternary complex induced by the extra π–π interaction. Limitations in the functionals available to describe the π–π interactions together with unconsidered coordination forces may account for the impossibility to reproduce this experimental observation.

Fig. 36 Molecular structure of dpatta. Stick representation of the crystal structure of the tetranuclear complex [Cu4 (dpatta)(NO3)4 ](NO3 )4 complex. Hydrogen atoms, noncoordinated nitrate anions and lattice molecules are omitted for clarity. In box: partial stick view of the structure showing the anion-π–π interactions

168

P. Ballester

Other beautiful examples of anion-π interactions in which the aromatic system is a N-containing arene have been described in recent years [57–59]. For example, Dunbar et al. [60] undertook a tandem crystallographic and computational investigation of the effect of the anion-π interaction on the preferred structural motifs of Ag(I) complexes with tetrazine and pyridazinebased ligands. The investigated anions are polynuclear like PF6 – , AsF6 – , SbF6 – , and BF4 – . The ligands used are 3,6-bis(2
-pyridyl)-1,2,4,5-tetrazine (bptz) and 3,6-bis(2
-pyridyl)-1,2-pyridazine (bppn). The MEP of the free ligands indicated that the bptz tetrazine ring has a higher π-acidic character compared to the bppn pyridazine ring. Consequently, the bptz ring is more likely to participate in anion-π interactions. This preference is in complete agreement with the observed structural motifs of the complexes of Ag(I)-bptz, regardless of the anion. In the Ag(I)-bptz compounds, multiple and shorter anion-π interactions are established between the anion and the central tetrazine rings, whereas in the Ag(I)-bppn complexes intermolecular interactions are maximized at the expenses of anion-π interactions. The solid-state evidences are also supported by DFT calculations. The preceding theoretical evaluation of the anion-π interaction, especially the results of Johnson, Hay et al. [45], together with the experimental evidences derived from the work of Kochi et al. [47], seem to suggest that the anion-aryl centroid complex geometry may not be the more common geometry in the interaction of halides with π-deficient electron systems. The Cambridge Structural Database (CSD) is a convenient and reliable storehouse for geometrical information. Furthermore, it is well established the utility of the solid-state structure of small molecules in analyzing the geometric parameters for intermolecular interactions. In order to confirm the experimental existence of the anion-π interaction in a general way, a sur-

Fig. 37 Molecular structure of the ligands bptz and bppn. Middle: portion of the packing diagram of [Ag2 (bptz)3 ][SbF6 ]2 depicting the anion and cation arrangement. Right: portion of the grid-type structure of [Ag2 (bppn)4 ][PF6 ]4 depicting π–π and anion-π interactions

Anions and π-Aromatic Systems. Do They Interact Attractively?

169

vey of the CSD has been carried out by several authors [10, 19, 45]. It is clear that the result of a search of the CSD will be highly dependent on the search criteria used to store a hit. Johnson, Hay et al. [45] carried A of the centroid of sixout a search for halide anions located within 4.0 ˚ membered ring π systems yielded 600 examples. In most cases, however, the π-system was either positively charged or bonded directly to a positively charged atom. If only charge-neutral π systems are considered, a much smaller set of 19 different structures is retrieved. Within these structures, 30 halide arene complexes were found. They performed an analysis of several distances of the complexes, i.e., distance anion-to-centroid, distance anion-to-plane, etc., concluding that 84% of the anions in the data set are closer to the ring carbons than to the centroid. Thus, the available structural data of the CSD indicates that the CT-binding motif of a halide interacting with a π systems is more prevalent that the anion-aryl centroid motif. The authors explained that the incongruence of their results with prior analysis of the CSD—which reported evidences of a marked preference of charge-neutral atoms bearing lone pairs to position themselves over the center of pentafluoroarenes [19] and trinitrobenzene derivatives [61]—is due to the use of a misleading search criterion. Finally, based on the statistical analysis of the hits obtained for a search criteria, which is that the A of the pentafluorophenyl centroid, they electronegative atoms must be 4 ˚ established a complete absence of any preferred location over the π system. The use of solid-state structures to analyze the geometric parameters of weak intermolecular forces such as anion-π interactions, has a general drawback, that is, other intermolecular interactions that are stronger, i.e., charge–charge interactions may determine the final crystal packing arrangement. Consequently, the geometry observed in the crystal for the weak interaction could be not the preferred one but the one resulting of the balance of intermolecular forces that controls the packing of the lattice. Probably, this is one of the causes that produces the complete absence of a preferred location of the anion over the π system of perfluorobenzenes. However, several searches carried out in the CSD do show that many of the geometries theoretically calculated for the interaction of anions with π systems are in fact present in the solid state. Consequently, this type of non-covalent interaction is worth to be seriously investigated by the practitioners of supramolecular chemistry. A recent account by Matile et al. [62] can be considered as the first experimental example of the use of anion-π interactions for the design of a synthetic anion channel. Matile and co-workers report the design, synthesis, and evaluation of π-acidic, shape-persistent oligo-(p-phenylene)-N,Nnaphthalenediimide (O-NDI) rods that can transport anions across lipid bilayer membranes with a rare selectivity Cl– > F– > Br– > I– and a substantial anomalous mole fraction effect. DFT calculations revealed a global quadrupole moment Qzz =+ 19.4 B for a model NDI. By comparison with

170

P. Ballester

Fig. 38 a The concept of the anion-π slide in lipid bilayer; b MEP for model NDI; red: electron-rich, blue: electron-poor

rigid p-oligophenyl rods, the authors deduced that the alignment of three NDI acceptors separated by phenyl spacers would afford rods with the appropriate length for hydrophobic matching with common lipid bilayer membranes. The results obtained in this study are in agreement with operational dynamic anion-π interactions and the existence of multiple anion-π sites for transmembrane anion hopping, that is, anion-π slide as shown in Fig. 38. A final caveat of the authors indicates that further studies are necessary to corroborate insights on the novel and complex system introduced in the study.

4 Summary and Outlook Theoretical studies indicate the existence of a counter-intuitive attractive interaction between anions and π-systems. The strength of the interaction

Anions and π-Aromatic Systems. Do They Interact Attractively?

171

(binding energies) seems to be drastically reduced by the inclusion of solvents effects in the calculations, although in some cases the interaction is still favorable. Calculations of F– , Cl– , Br– and NO3 – complexes with electrondeficient aromatic systems (quadrupole moment Qzz > 0) like, triazine, hexafluorobenzene and polycyanobenzenes establish the existence of different binding geometries. When the halide lies above the plane of the π-system strongly covalent sigma complexes, weakly covalent donor-π acceptor (CT) complexes and anion-aryl centroid complexes may be located as stable geometries. The halides can also form stable hydrogen-bonded complexes with π-aromatic systems having C – H donors groups. In fact, the presence of electro-withdrawing substituents increases the acidity of the C – H donor and strengthens this type of interaction. The most stable binding geometry (global minimum) depends on the anion and the π-system. Also, some binding geometries which are not even stable points (minimum) for certain anions could be global minimum for others. In many cases, the off-center CT complexes are more stable that the anion-aryl centroid binding geometry. The physical nature of the anion-π interaction has also been investigated in great detail from a theoretical point of view. The Qzz molecular quadrupole of aromatics has the adequate spatial orientation to be involved in the formation of non-covalent anion-π complexes in which the halide is located above the arene centroid (anion-aryl centroid geometry). The two fundamental components contributing to the stabilization energy of non-covalent anion-π or anion-aryl centroid complexes in the gas phase are electrostatic and anion-induced polarization. The electrostatic component correlates well with the magnitude of the Qzz of the aromatic ring and the contribution due to polarization can be considered as almost constant. The contributions of the dispersion and correlation terms to the total interaction energy are small, but they are more important in anion-π complexes, than in cation-π complexes. In recent years, numerous experimental studies have shown crystallographic evidences of anions encapsulated in aromatic cavities formed by electron-deficient arenes. These observations clearly hint to the existence of an attractive interaction between anions and π-systems. Nevertheless, many of the reported crystal structures do not show the anion located above the arene centroid (anion-aryl centroid geometry). On the contrary, the great majority of the theoretical studies on the anion-π interaction deal with complexes displaying the anion-aryl centroid geometry in the gas phase. Although several theoretical studies have proposed structures for novel receptors based on multiple anion-π interactions, to date, and to the best of our knowledge, the experimental realization of such type of receptors and their anion-binding properties have not been reported. The few experimental studies that have attempted the experimental quantification of the anionπ interaction in solution have produced very different and even opposite results.

172

P. Ballester

The last paragraph of a chapter like this one asks the writer to predict if the non-covalent interaction of anions with π-system will be as important and useful as the nowadays well-established cation-π interaction. We are cautious in answering this question affirmatively, but we have to confess that we are currently working in the construction of synthetic receptors for anions that do incorporate electron deficient π-aromatic systems. We hope that our designs will be valuable for the experimental evaluation of the anionπ interaction in solution and we will report shortly our findings. Once the strength of the anion-π interaction in solution is determined, it will be easier to evaluate its possible use in the construction of selective receptors for anions making use of the directionality properties that we have discussed in the chapter. Acknowledgements I want to thank Dr. Antonio Frontera, Dr. David Quiñonero, and Prof. Pere M. Deyà from the University of the Balearic Islands for sharing with me his interest and results of their studies on the anion-π interaction. As a consequence, my group is now involved in trying to quantify experimentally the strength of this new, counterintuitive and rather unnoticed non-covalent interaction, and I find myself writing a book chapter on the topic. Generous financial support from MEC (CTQ2005-08989-C01-02/BQU and CSD20060003), ICIQ Foundation, ICREA Foundation and Generalitat de Catalunya (2005SGR00108) is gratefully acknowledged.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19.

Hunter CA (2004) Angew Chem Int Ed 43:5310 Amendola V, Esteban-Gomez D, Fabbrizzi L, Licchelli M (2006) Acc Chem Res 39:343 Bowman-James K (2005) Acc Chem Res 38:671 Beer PD, Schmitt P (1997) Curr Opin Chem Biol 1:475 Simmons HE, Park CH (1968) J Am Chem Soc 90:2428 Park CH, Simmons HE (1968) J Am Chem Soc 90:2429 Park CH, Simmons HE (1968) J Am Chem Soc 90:2431 Schmidtchen FP, Berger M (1997) Chem Rev 97:1609 Schneider HJ, Werner F, Blatter T (1993) J Phys Org Chem 6:590 Gamez P, Mooibroek T, Teat S, Reedijk J (2007) Acc Chem Res 40:435 Egli M, Sarkhel S (2007) Acc Chem Res 40:197 Ma JC, Dougherty DA (1997) Chem Rev 97:1303 Mascal M (2006) Angew Chem Int Ed 45:2890 Gallivan JP, Dougherty DA (1999) Org Lett 1:103 Danten Y, Tassaing T, Besnard M (1999) J Phys Chem A 103:3530 Alkorta I, Rozas I, Elguero J (1997) J Org Chem 62:4687 Mascal M, Armstrong A, Bartberger MD (2002) J Am Chem Soc 124:6274 Alkorta I, Rozas I, Elguero J (2002) J Am Chem Soc 124:8593 Quiñonero D, Garau C, Rotger C, Frontera A, Ballester P, Costa A, Deyà PM (2002) Angew Chem Int Ed 41:3389 20. Caldwell JW, Kollman PA (1995) J Am Chem Soc 117:4177 21. Sunner J, Nishizawa K, Kebarle P (1981) J Phys Chem 85:1814 22. Burgi HB, Dunitz JD, Shefter E (1973) J Am Chem Soc 95:5065

Anions and π-Aromatic Systems. Do They Interact Attractively? 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.

173

Bürgi HB, Lehn JM, Wipff G (1974) J Am Chem Soc 96:1956 Bader RFW (1991) Chem Rev 91:893 Cheeseman JR, Carroll MT, Bader RFW (1988) Chem Phys Lett 143:450 Koch U, Popelier PLA (1995) J Phys Chem 99:9747 Cubero E, Luque FJ, Orozco M (1998) Proc Natl Acad Sci USA 95:5976 Hernandez B, Luque FJ, Orozco M (1999) J Comput Chem 20:937 Luque FJ, Orozco M (1998) J Comput Chem 19:866 Williams JH (1993) Acc Chem Res 26:593 Garau C, Frontera A, Quinonero D, Ballester P, Costa A, Deya PM (2003) Chem Phys Chem 4:1344 Kim D, Tarakeshwar P, Kim KS (2004) J Phys Chem A 108:1250 Jeziorski B, Moszynski R, Szalewicz K (1994) Chem Rev 94:1887 Hunter CA, Low CMR, Rotger C, Vinter JG, Zonta C (2002) Proc Natl Acad Sci USA 99:4873 Kearney PC, Mizoue LS, Kumpf RA, Forman JE, McCurdy A, Dougherty DA (1993) J Am Chem Soc 115:9907 Hans-Jörg Schneider TB, Patrick Zimmermann (1990) Angew Chem Int Ed Engl 29:1161 Ting AY, Shin I, Lucero C, Schultz PG (1998) J Am Chem Soc 120:7135 Garau C, Quinonero D, Frontera A, Ballester P, Costa A, Deya PM (2005) J Phys Chem A 109:9341 Schleyer PvR, Maerker C, Dransfeld A, Jiao H, Hommes NJRvE (1996) J Am Chem Soc 118:6317 Vahtras O, Almloef J, Feyereisen MW (1993) Chem Phys Lett 213:514 Feyereisen M, Fitzgerald G, Komornicki A (1993) Chem Phys Lett 208:359 Hermida-Ramon JM, Estevez CM (2007) Chem Eur J 13:4743 Klarner F-G, Kahlert B (2003) Acc Chem Res 36:919 Clements A, Lewis M (2006) J Phys Chem A 110:12705 Berryman OB, Bryantsev VS, Stay DP, Johnson DW, Hay BP (2007) J Am Chem Soc 129:48 Schneider H-J, Yatsimirsky AK (2000) Principles and Methods in Supramolecular Chemistry. Wiley, Chichester Rosokha YS, Lindeman SV, Rosokha SV, Kochi JK (2004) Angew Chem Int Ed 43: 4650 Gaussian 03W Version 6.1. DFT B3LYP/6-31G(d,p) Hettche F, Hoffmann RW (2003) New J Chem 27:172 Frontera A, Saczewski F, Gdaniec M, Dziemidowicz-Borys E, Kurland A, Deyà PM, Quiñonero D, Garau C (2005) Chem Eur J 11:6560 Berryman OB, Hof F, Hynes MJ, Johnson DW (2006) Chem Commun, p 506 Mo H, Wang A, Wilkinson PS, Pochapsky TC (1997) J Am Chem Soc 119:11666 Demeshko S, Dechert S, Meyer F (2004) J Am Chem Soc 126:4508 de Hoog P, Gamez P, Mutikainen I, Turpeinen U, Reedijk J (2004) Angew Chem Int Ed 43:5815 Casellas H, Massera C, Buda F, Gamez P, Reedijk J (2006) New J Chem 30:1561 Maheswari PU, Modec B, Pevec A, Kozlevcar B, Massera C, Gamez P, Reedijk J (2006) Inorg Chem 45:6637 Dorn T, Janiak C, Abu-Shandi K (2005) Cryst Eng Comm 7:633 Gural’skiy IyA, Solntsev PV, Krautscheid H, Domasevitch KV (2006) Chem Commun, p 4808 Zhou X-P, Zhang X, Lin S-H, Li D (2007) Crystal Growth & Design 7:485

174

P. Ballester

60. Schottel BL, Chifotides HT, Shatruk M, Chouai A, Perez LM, Bacsa J, Dunbar KR (2006) J Am Chem Soc 128:5895 61. Quinonero D, Garau C, Frontera A, Ballester P, Costa A, Deya PM (2002) Chem Phys Lett 359:486 62. Gorteau V, Bollot G, Mareda J, Perez-Velasco A, Matile S (2006) J Am Chem Soc 128:14788

Struct Bond (2008) 129: 175–206 DOI 10.1007/430_2008_083 © Springer-Verlag Berlin Heidelberg Published online: 12 March 2008

Anion Templates in Synthesis and Dynamic Combinatorial Libraries Ramon Vilar Department of Chemistry, Imperial College London, London SW7 2AZ, UK [email protected] 1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

175

2

Templating Effects in Chemistry . . . . . . . . . . . . . . . . . . . . . . . .

177

3 3.1 3.2 3.3

Recent Examples of Anion-Templated Processes Macrocycles . . . . . . . . . . . . . . . . . . . . Cages and Capsules . . . . . . . . . . . . . . . . Interlocked Species . . . . . . . . . . . . . . . .

. . . .

178 178 182 186

4 4.1 4.2

Anions as Templates in Dynamic Combinatorial Chemistry . . . . . . . . Using Metal–Ligand Coordination Bonds . . . . . . . . . . . . . . . . . . . Using Reversible Covalent Bonds . . . . . . . . . . . . . . . . . . . . . . . .

191 192 201

5

Conclusions and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . .

204

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

205

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

Abstract This review presents an overview of the area of anion-templated synthesis of molecules and supramolecular assemblies. The review is divided into two main sections: the first part deals with anion-templated systems where the final products are linked by bonds that are not reversible under the conditions of the experiment. Several recent examples of macrocycles, cages and interlocked species are presented in this section. The second part of the chapter, presents a discussion of anion-templation in systems containing reversible bonds that give rise to dynamic combinatorial libraries (either by formation of coordination metal–ligand bonds or by reversible covalent bonds). Keywords Anion template · Dynamic combinatorial library · Molecular recognition · Self assembly

1 Introduction The use of templates in chemistry is nowadays a widespread strategy for the synthesis of complex molecules and supramolecular assemblies [1–4]. A plethora of species ranging from macrocycles and cages to interlocked molecules and imprinted polymers have been efficiently synthesised using this approach. In a reaction where several potential products can be formed,

176

R. Vilar

chemical templates pre-organize the molecular building blocks to favour the formation of products with specific nuclearity, geometry and overall structure. The interactions between the building blocks and the corresponding template are generally of a non-covalent nature (e.g., hydrogen bonding, electrostatic interactions, coordination bonds or π-π stacking). In order for a templated process to successfully yield the targeted product, the structural and electrostatic properties of the template need to be carefully selected. From a structural point of view, both the size and geometry of the template have to be considered, while electrostatically the choice is restricted to neutral, positive or negatively charged species. While cations have been widely used as templates in synthetic chemistry, the role of anion templates did not start to be exploited until relatively recently. In spite of the initial reservations regarding the high solvation energies, sizes and pH dependency of anions, the past few years have shown the great potential these species have as templates in a wide range of synthetic routes [5–10]. In a templated reaction, once the building blocks have been pre-organised by the template, they can be linked together by irreversible covalent bonds or by reversible interactions (either covalent or supramolecular). Initially, most templated reactions aimed at forming irreversible covalent bonds between the building blocks being brought together by the template. However, more recently, a different approach to templated processes has been developed in which the building blocks are linked together by reversible bonds which can rapidly form and break. If in this reaction mixture there are several different building blocks that can react with each other in a reversible fashion, then a dynamic equilibrium between all the possible combinations of the molecular components can be established. This approach, known as dynamic combinatorial chemistry (DCC), can then lead to the formation of virtual libraries of compounds – named dynamic combinatorial libraries (DCL) – that are thermodynamically controlled [11–15]. As with any other chemical equilibrium, different external stimula (e.g., temperature, concentrations, addition of a template) can modify the equilibrium between the components of a DCL. Amongst the most efficient ways of modifying the composition of these virtual combinatorial libraries, is by adding a template which can shift the equilibrium towards the formation of one particular assembly or molecule over others. This process is known as amplification. The concept of dynamic combinatorial chemistry was pioneered by Sanders [16, 17] and Lehn [15, 18, 19] in the 1990s. Since then, a wide range of reports have appeared in which different templates have been used to amplify specific components of a dynamic library. In most cases, the templates used are either cationic or neutral species leading to the amplification of the corresponding receptors for these types of species. Considering that anion templates have only started to be used recently in synthesis, it is not surprising that there are only a handful of examples where DCLs are amplified by anionic species.

Anion Templates in Synthesis and Dynamic Combinatorial Libraries

177

This chapter aims to provide an update on the role of anions as templates. The review is divided in two main sections: (a) anion-templated synthesis of assemblies linked together by irreversible bonds (or bonds that are inert under mild experimental conditions); (b) anion templates in systems where the bonds linking the components are reversible and lead to anion-controlled dynamic combinatorial libraries. Since some comprehensive reviews in the area of anion templation have appeared over the past few years [5–7], this chapter will mainly focus on papers published recently and will aim to show the principles of anion templation rather than being a comprehensive account of the literature. In addition, the scope of the chapter will be restricted to finite assemblies (molecular or supramolecular) and not polymeric (for a review on molecularly imprinted polymers using anions see Steinke’s chapter in this volume).

2 Templating Effects in Chemistry In the 1960s, Busch carried out pioneering work on metal-directed syntheses of macrocycles establishing the concept of chemical template. As defined by him “a chemical template organizes an assembly of atoms, with respect to one or more geometric loci, in order to achieve a particular linking of atoms” [20–22]. This provides an efficient route to prepare a specific molecular assembly when several others can be potentially formed. Ideally templates should be removed from the final product once the reaction has reached completion; however, templates often form an integral part of the final product; hence, they cannot always be removed from it. A templated process can be driven thermodynamically or kinetically [23]. In the former case, the template binds more strongly to one of the products present in an equilibrium (i.e., a mixture under thermodynamic control) shifting the reaction towards the formation of this specific product which is then obtained in higher yields. As will be discussed later in this review, this is particularly important in the generation of dynamic combinatorial libraries. On the other hand, kinetic templates operate under irreversible conditions by stabilising the transition state leading to the final product. In a templated process the interactions between the directing group and the building blocks can be either covalent or non-covalent. The latter can make use of a wide range of supramolecular interactions, such as hydrogen bonding, π-π stacking, electrostatic interactions and hydrophobic effects. From these, hydrogen bonding interactions are particularly useful since they are relatively strong and, more importantly, are directional. In fact, hydrogen bonding interactions are employed as the main driving force in many of Nature’s templated processes. One of the most elegant examples of this is the replication and transcription of nucleic acids.

178

R. Vilar

3 Recent Examples of Anion-Templated Processes Since the first examples of anion-directed assemblies were reported in the 1990s, a wide range of molecular and supramolecular systems have been successfully prepared using anions as templates. This section has been divided by the type of molecular or supramolecular species formed and, as mentioned before, it will mainly focus on systems that have been recently reported in the literature (making reference to previous examples when important for the sake of clarity and completion). 3.1 Macrocycles Böhmer has recently shown that the presence of chloride can dictate the size of macrocyclic poly-urea systems. The reaction between diamine 1 and diisocyanate 2 yield the macrocyclic species 3 and 4 in a 5:1 ratio (see Scheme 1) [24]. When this reaction was repeated in the presence of two equivalents of tetrabutylammonium chloride, the formation of the larger macrocycle 4 was favoured over the trimeric species 3. In fact, the ratio reverted to 1:5 in favour of the larger macrocycle. Crystals of macrocycle 4 were obtained from the crude of the first reaction when grown in the presence of tetrabutylammonium chloride. The structure (Fig. 1) revealed an interesting host-guest complex in which two chlorides are bound by the macrocycle (which explains the need of two equivalents of chloride to favour the formation of the hexameric compound). Alfonso and Luis have recently reported an example of anion-templated synthesis of a pseudopeptidic macrocycle. The reaction between diamine 5 and dialdehyde 6 was carried out (see Scheme 2) [25]. 1 H NMR spectroscopy showed a complex patter of signals together with protons associated with aldehyde and methoxyamine suggesting the presence of a range of different macrocyclic and acyclic products. Addition of different anions to the reaction mixture had little effect to the distribution of products shown by 1 H NMR spectroscopy. However, when the reaction was repeated in the presence terephthalate, the almost quantitative formation of a single product as revealed by 1 H NMR spectroscopy was observed. This intermediate product has been postulated to be host-guest complex 7· Terephthalate). This imine-containing macrocycle was then reduced to the corresponding amine to yield macrocycle 8, which was isolated and fully characterised. In the first step of this reaction, terephthalate is expected to form hydrogen bonding interactions between its anionic carboxylates and the amides of the macrocycle. In addition, π-π interactions between the aromatic ring of the terephthalate and that one of the di-aldehyde are likely to also play a role in the templating process.

Anion Templates in Synthesis and Dynamic Combinatorial Libraries

179

Scheme 1 Synthetic procedure for the preparation of macrocycles 3 and 4

Fig. 1 X-ray crystal structure of macrocycle 4 highlighting the two chloride anions (solid spheres) bound to the macrocycle

180

R. Vilar

Scheme 2 Terephthalate-templated synthesis of macrocycle 8

It is worth noting that this process could in principle lead to dynamic combinatorial libraries of products since the type of bond that brings together the different components in the reaction is a reversible one (namely imine formation). In fact, the initial mixture of products observed by the authors has been postulated to be a mixture of different-sized macrocycles and linear species (from which one of them is amplified upon addition of terephthalate). More detailed studies would be needed to determine whether this system indeed leads to the formation of a DCL of receptors (see Sect. 4 for examples of anion-directed DCLs). A widely used approach to the synthesis of large macrocycles and cages (see Sect. 3.2) is by self-assembly of metal centres and polydentate ligands. Careful choice of the geometry around the metal and the number and relative position of the coordinating groups on the bridging ligands can generate large 2D and 3D assemblies from a one-pot reaction. Often, the addition of a template to the reaction mixture provides a more efficient path for the formation of one specific assembly. Some recent examples of anion-templated metallo-macrocycles and -cages are discussed in this and the following sections. Lippert has shown that the assembly between platinum(II) centres and purine bases can be controlled by specific anions. In analogy to the hydrogenbonded tetrads that guanine bases can form, this group synthesised a metallosquare in which the purine bases are interconnected by coordination to platinum(II) centres rather than hydrogen bonding interactions (see Fig. 2) [26].

Anion Templates in Synthesis and Dynamic Combinatorial Libraries

181

Fig. 2 Schematic representation of: a guanine quartet with a cationic guest and b metallosquare 9 based on platinum(II) and methylpurine with an anionic guest

Square 9 formed over a period of 5 days by self-assembly process of four units of [(NH3 )2 Pt(Pur)(H2 O)]2+ (where Pur = 9-methylpurine). Interestingly, in this process the formation of a second species with a triangular geometry was observed (see Scheme 3). NMR experiments showed that the triangular metallo-macrocycle 10 is favoured (in a 0.6:1 ratio) to square 9. However, if the self-assembly process is carried out in the presence of SO4 2– , the proportion of the two species changes and the square is favoured over the triangle in a 2.5:1 ratio. This change in the preference for the square over the triangle has been attributed to the templating properties of the sulfate anion.

Scheme 3 Reaction scheme for the formation of metallo-triangle 10

Maekawa and Kitagawa have recently reported an interesting aniondirected approach to the synthesis of bowl-shaped metallo-macrocycles [27]. Initially, they observed that the reaction between [Cu(MeCN)4 ](PF6 ) and 4(2-pyridyl)pyrimidine (pprd) under an atmosphere of C2 H4 and in acetone yielded the coordination polymer {[Cu(pprd)(C2 H4 )](PF6 )}n (see Scheme 4).

182

R. Vilar

Scheme 4 Formation of a coordination polymer from the reaction between [Cu(MeCN)4] (PF6 ) and 4-(2-pyridyl)pyrimidine (pprd)

When the same reaction was carried out in methanol the macrocyclic species [Cu4 (pprd)4 (C2 H4 )4 ](PF6 )4 (11) was formed instead. The crystal structure of this metallo-assembly revealed a bowl-shaped structure with the PF6 – anion positioned at its centre (see Fig. 3).

Fig. 3 X-ray crystal structures of a the tetra-copper assembly [Cu4 (pprd)4 (C2 H4 )4 ](PF6 )4 (11) and b the tri-copper assembly {[Cu3(pprd)3 (C2 H4 )3 ](ClO4 )3 }3 (12)

Interestingly, the reaction between [Cu(C2 H4 )n ]ClO4 and pprd in Me2 CO under a C2 H4 atmosphere yielded the trinuclear metallo-macrocycle {[Cu3 (pprd)3 (C2 H4 )3 ](ClO4 )3 }3 (12). Although the reactions to obtain the two differently-sized macrocycles are not identical, it is plausible to suggest that the volume of the anions dictates the size of the macrocycle. With PF6 , the tetranuclear metallo-macrocycle forms while the smaller tetrahedral ClO4 anion directs the formation of the tri-copper macrocycle. 3.2 Cages and Capsules In the late 1990s we reported one of the first examples of anion-templated metallo-cage [28–30]. The system was based on the chloride-directed assem-

Anion Templates in Synthesis and Dynamic Combinatorial Libraries

183

bly of eight units of amidinothiourea (H-atu, see Scheme 5) and six nickel(II) centres to yield [Ni6 (atu)8 ⊂Cl]Cl3 (13). The resulting assembly showed to contain a completely encapsulated chloride interacting via hydrogen bonding and weak metal-anion interactions with the cage.

Scheme 5 Reaction scheme for the synthesis of hexanickel cage 13

The approach described above was expanded afterwards to incorporate a second type of metal within the framework yielding the mixed Ni-Pd and Ni-Pt complexes [Ni4 M2 (atu)8 ⊂Cl]Cl3 (M = Pd, 15; Pt, 16) (see Fig. 4). This was carried out by first preparing [Ni(atu)2 ] (14) and then reacting it with the corresponding palladium(II) or platinum(II) salt. As in the case of the hexanickel cage, it was found that the mixed-metal cage would only form in the presence of the appropriate halide, namely chloride or bromide. The assembly of the nickel(II)/H-atu cage 13, is accompanied by a dramatic colour change, from orange to green. Considering that in methanol only chloride acts as a template for the formation of cage 13, we have recently employed this anion-templated process to develop a colorimetric chemical sensor for chlorides [31]. Bidentate pyrazolyl-based ligands have shown to be versatile building blocks for the synthesis of a plethora of coordination assemblies. Initial work by McCleverty and Ward in the late 1990s showed that the synthesis of cages [Co4 (Ln )6 ⊂X](X)7 (where L1 and L2 = bidentate pyrazolyl-pyridine ligands;

184

R. Vilar

Fig. 4 X-ray crystal structure of mixed-metal cage [Ni4 Pt2 (atu)8⊂Cl]Cl3 showing the encapsulated chloride anion

X = BF4 – , ClO4 – ; see Figs. 5 and 6) was dependant on the presence of specific anions [32]. Detailed investigations (by NMR spectroscopy) of the assembly process in solution demonstrated that the tetrahedral BF4 – and ClO4 – anions indeed act as templating agents for the formation of these metalloassemblies [33]. More recently, Ward has explored the effect that the length of bidentate pyrazolyl-based ligands has on the formation of the metallo-cages [34]. When

Fig. 5 A selection of bidentate pyrazolyl-based ligands employed for the synthesis of metallo assemblies

Anion Templates in Synthesis and Dynamic Combinatorial Libraries

185

Fig. 6 X-ray crystal structure of [Co4 (L1 )6 ⊂(ClO4 )](ClO4)7 highlighting the encapsulated tetrahedral anion (in space-fill representation)

L3 was used as ligand, the expected [Co4 (L3 )6 ⊂X](X)7 (X = BF4 – , ClO4 – , PF6 – , I– ) cages were formed, but in this case the anion did not seem to play a determinant role in defining the geometry of the final assembly. As indicated by the authors, the resulting cage is sufficiently large to leave gaps in the centre of the faces through which the encapsulated anion can easily exchange with the external anions.

186

R. Vilar

Amouri has reported the anion-directed synthesis of a series of coordinatively unsaturated metallo-cages with general formula [Co2 (L4 )4 (RCN)2 ⊂ (BF4 )](BF4 )3 (R = Me, Et, Ph) [35]. The X-ray crystal structures of some of these assemblies (see Fig. 7 for an example) have revealed the presence of an encapsulated BF4 – anion which interacts with the coordinatively unsaturated cobalt(II) centres. Interestingly when analogous reactions were performed in the presence of other anions (such as Cl and NO3 ) different metal-organic assemblies were formed [36].

Fig. 7 X-ray crystal structure of [Co2(L4 )4 (MeCN)2 ⊂(BF4 )](BF4 )3 showing the encapsulated anion and its interactions with the metal centres

3.3 Interlocked Species Molecular interlocked systems such as catenanes and rotaxanes can also be prepared using anion-directed approaches. Although anionic templates did not make their way into this area until relatively recently, nowadays there are several examples that demonstrate the utility of this approach for the synthesis of this topologically interesting species. The first examples of interlocked species synthesized by anion templated approaches were the pseudorotaxanes and rotaxanes reported in the late 1990s by Stoddart and Vögtle. Stoddart reported that by mixing four equivalents of [NH2 (CH2 Ph)2 ][PF6 ] with one equivalent of the macrocycle tetrakis-p-phenylene[68]crown-20, the quadruply-stranded pseudorotaxane (17) formed (see Fig. 8) [37]. This assembly was structurally characterized revealing the presence of a PF6 – anion at its center forming multiple C–H· · ·F hydrogen bonds with the hydroquinone methine and the benzylic methylene hydrogen atoms.

Anion Templates in Synthesis and Dynamic Combinatorial Libraries

187

Fig. 8 Schematic representation of quadruply-stranded pseudorotaxane 17

Using a different approach, Vögtle successfully showed that organic anions can induce the formation of rotaxanes [38–41]. In this approach a strong host-guest complex between a tetralactam macrocycle and a phenolate anion is first formed (see Scheme 6). In this assembly the phenolate anion is positioned at the center of the ring to further react with a second component (e.g., an alkyl bromide or acyl chloride) yielding a rotaxane. The negatively charged phenolic functionality can be located either at the stopper component or at the axle precursor providing a range of different possibilities for the synthesis of the interlocked species. More recently, Beer has developed a series of synthetic procedures for the halide-templated syntheses of pseudorotaxanes, rotaxanes and catenanes. This methodology is based on combining the recognition of halides by a hydrogenbonding host (e.g., a macrocycle) with ion-pairing between the corresponding halide and a cationic species (see Fig. 9). These two interactions allow for the interpenetration of the two molecular components to yield the interlocked molecules. Some examples of the macrocycles and axels employed to generate the corresponding pseudorotaxanes are shown in Fig. 10 [38–41]. A similar approach was later employed to prepare rotaxanes by “clipping” (via ruthenium-catalysed olefin metathesis) the acyclic hydrogen bonding

188

R. Vilar

Scheme 6 Schematic representation of the anion templated synthesis of rotaxanes based on tetralactam macrocycles and a phenolate anions

Fig. 9 Schematic representation of the anion-templated synthesis of interlocked species based on combining halides recognition by a hydrogen-bonding host with ion-pairing

species 18 with the corresponding axel 19 in the presence of chloride (see Scheme 7) [42]. The resulting rotaxane (20) has been structurally characterised showing the strong binding between the interlocked species and the templating chloride (see Fig. 11). Using the same general strategy described above, Beer reported the first anion-templated synthesis of catenanes [43, 44]. Mixing macrocycle 21 with

Anion Templates in Synthesis and Dynamic Combinatorial Libraries

189

Fig. 10 Selection of macrocycles and axels employed to synthesise pseudorotaxanes and rotaxanes

Scheme 7 Reaction scheme for the chloride-templated synthesis of rotaxane 20

190

R. Vilar

Fig. 11 X-ray crystal structure of rotaxane 20 showing the chloride anion bound to the interlocked species

Scheme 8 Reaction scheme for the chloride-templated synthesis of catenane 23

Fig. 12 X-ray crystal structure of catenane 23 showing the chloride anion bound to the interlocked species

Anion Templates in Synthesis and Dynamic Combinatorial Libraries

191

the acyclic hydrogen-bonding molecule 22 in the presence of chloride as a template, led to the formation of [2]catenane 23 (see Scheme 8). As for rotaxane 20, the crystal structure of this interlocked species revealed the strong interaction between the templating chloride and the two macrocyclic components of the catenane (Fig. 12). For a comprehensive review of this area, see the recent reviews published by Beer [7, 9, 10].

4 Anions as Templates in Dynamic Combinatorial Chemistry One of the seminal papers in defining the concept of dynamic combinatorial chemistry was published by Lehn in the 1990s [45] In this work it was shown

Scheme 9 Chloride-templated synthesis of circular helicate 24

192

R. Vilar

that the assembly of iron(II) salts and a tris-bipy ligand (L5 ) is a dynamic process that can yield a range of different metal helicates. Which assembly is formed, is highly dependent on the nature of the counter-anions present in solution. With FeCl2 , the pentanuclear circular helicate [Fe5 (L5 )5 Cl]9+ (24) was formed in high yields (see Scheme 9), while a mixture of the pentaand hexa-nuclear helicates were obtained in the presence of bromide (and only the hexa-nuclear helicate [Fe6 (L5 )6 (SO4 )]10+ (25) was formed with sulfate). Further studies by the same authors demonstrated that in the reaction with FeCl2 (and also in the analogous one with NiCl2 ) a linear helicate is formed first (i.e., the kinetic product) which progressively converts into the thermodynamic circular helicate product [Fe5 (L5 )5 Cl]9+ [46] Structural characterization of 25 confirmed the assembly to be a circular double helix with a chloride ion located in the central cavity. Although this anion-directed system was one of the first examples of DCCs reported in the literature, there are in fact very few known systems to date where negatively charged species are used to amplify the formation of a specific assembly from a dynamic combinatorial library. Such examples will be reviewed in Sects. 4.1 and 4.2. 4.1 Using Metal–Ligand Coordination Bonds Dunbar has elegantly demonstrated the use of anionic templates for the syntheses of a range of nickel(II) and zinc(II) metalla-cyclophanes using 3,6bis(2-pyridyl)-1,2,4,5-tetrazine (bptz) as bridging ligand (see Scheme 10) [47, 48]. Anions such as BF4 – and ClO4 – induce the formation of the tetra-metallic square assemblies [{M4 (bptz)4 (CH3 CN)8 }X](X)7 , (M = ZnII , NiII ; X = BF4 – , ClO4 – ), while the larger octahedral anion SbF6 – templates the formation of the molecular pentagon [{Ni5 (bptz)5 (CH3 CN)10 }SbF6 ](SbF6 )9 . The X-ray crystal structures of these species (see Figs. 13 and 14) have shown that in both the squares and pentagon one anion is encapsulated at the centre of the corresponding metalla-cyclophane (displaying anion-π interactions between the O and F atoms of the anions and the tetrazine rings of bptz). Further studies by the same authors showed that the molecular pentagon can be easily converted into the corresponding molecular square in the presence of excess BF4 – and ClO4 – . The conversion of the molecular pentagon to the square is also observed upon addition of iodide (which due to its large size and polarizability it can adopt the directionality of a tetrahedral anion). In contrast to the above, the conversion of the nickel square to the corresponding pentagon upon addition of excess SbF6 – is not readily observed (only partial conversion takes place). This apparent higher stability of the molecular square in comparison to the molecular pentagon has been attributed to more strain in the pentagon. This is supported by the X-ray crystal structure of the metallo-pentagon in which the btpz ligands are considerably bent.

Anion Templates in Synthesis and Dynamic Combinatorial Libraries

193

Scheme 10 Reaction scheme showing the anion-directed synthesis of metallo-macrocycles and their interconversion

Fig. 13 X-ray crystal structure of metallo-square [{Ni4 (bptz)4 (CH3 CN)8 }BF4 ](BF4 )7

More recently, Fujita has reported an interesting example of anioncontrolled dynamic equilibrium of palladium-containing assemblies [49]. The reaction of ligands L6 and L7 with palladium(II) was investigated under different experimental conditions, namely in different solvents, at different concentrations of the ligand and in the presence of different counteranions.

194

R. Vilar

Fig. 14 X-ray crystal structure of metallo-square [{Ni5 (bptz)5 (CH3 CN)10 }SbF6 ](SbF6 )9

In this study it was shown that the nature of the resulting metalloassemblies was highly dependant on the conditions employed. For example, the reaction between L6 and [Pd(en)(NO3 )2 ] in DMSO was found to yield two macrocycles (26 and 27) in roughly 60:40 proportions when the concentration of the ligand was 5 mM. By reducing the concentration of the ligand to 1 mM, the equilibrium was shifted to the simpler [2+2] assembly 26 which was present in nearly 90%. Increasing the concentration of ligand to 20 mM or above, yielded yet another product, metallo-assembly 28 which at 500 mM concentration of ligand was practically the only product observed by 1 H NMR spectroscopy. In the same paper, the reactions between these two ligands and “naked” palladium(II) cations were also discussed. Interestingly, the structure and nuclearity of the resulting assemblies was shown to depend on the counteranion of the palladium salt used. Thus, the reaction between Pd(NO3 )2 and ligand

Anion Templates in Synthesis and Dynamic Combinatorial Libraries

195

196

R. Vilar

Fig. 15 X-ray crystal structure of metallo-tetrahedron 29

L6 yielded mainly the tetrahedral assembly 29 (see Fig. 15). An analogous assembly was obtained in the presence of BF4 – . However, when the reaction was carried out in the presence of triflate a double-walled triangle (30) was obtained as the major product (Scheme 11).

Scheme 11 Schematic representation of double-walled metallo-triangle

When carrying out a similar reaction with ligand L7 (which is longer than the structures of the resulting metallo-assemblies were again found to be anion-dependant. In this case, dynamic equilibrium between the two assemblies 31 and 32 was observed. With nitrate as the counter-anion, a roughly 1:1 mixture of the two assemblies was observed; however, with triflate the main L6 )

Anion Templates in Synthesis and Dynamic Combinatorial Libraries

197

Scheme 12 Schematic representation of the equilibrium between the double-walled metallo-triangle 31 and metallo-tetrahedron 32

product formed was the double-walled triangle. In contrast in the presence of the aromatic p-tosylate anion, the formation of the M4 L8 assembly was favored. A similar approach to that reported by Fujita for the generation of dynamic combinatorial libraries of metallo-assemblies has been developed in our own group [50]. In contrast to most of the bipyridyl-based ligands employed for the synthesis of metallo-macrocycles reported so far, we were interested in using bipyridyl ligands containing spacers with hydrogen bonding functionalities such as ligands L8 and L9 . It was rationalized that having hydrogen bonding donor groups would aid in the interaction with potential anionic guests/templates.

The reactions between each of these two ligands and [Pd(dppp)(OTf)2 ] (dppp = 1,3-bis(diphenylphosphino)propane) were studied aiming at pro-

198

R. Vilar

Scheme 13 Reaction scheme for the synthesis of metallo-macrocycles 33 and 34

ducing cyclic metallo-assemblies (see Scheme 13). The combination of these ligands and the metal complex (in 1:1 ratios) could in principle give a range of different cyclic and acyclic materials. Since palladium-pyridine bonds are relatively labile, it was expected that a dynamic equilibrium of different assemblies would be established. Crystals were grown from the corresponding reaction mixtures. The X-ray crystal structures obtained revealed that the crystallised products correspond to [2+2] assemblies with general formula [Pd(dppp)(L)]2 (OTf)4 (L = L8 , 33; L = L9 , 34) and a “bowl-type” structure (see Fig. 16).

Fig. 16 X-ray crystal structure of assembly 34 showing the triflate anion at the centre of the bowl-type assembly

Anion Templates in Synthesis and Dynamic Combinatorial Libraries

199

Although the solid-state structures obtained for these systems indicate the formation of the [2+2] assembly, 1 H NMR studies showed that more than one species co-existed in solution. As indicated above, in principle a 1:1 mixture of the bis-pyridyl ligands and cis-[Pd(dppp)]2+ centres could yield macrocycles of different sizes or a range of different acyclic products. Several 1 H NMR spectroscopic and ESI-mass spectrometric studies were carried out (using [Pd(dppp)(OTf)2 ] and L8 ) to establish the behaviour of the system in solution. These studies suggested that there is an equilibrium between two species which have been assigned to the [2+2] and a [3+3] metallo-assemblies (see Scheme 14).

Scheme 14 Reaction scheme showing the equilibrium between the [2+2] and [3+3] metallo-macrocycles

Interestingly, this equilibrium can be shifted by modify the experimental conditions such as the solvent, concentration of ligand and temperature. Furthermore, the equilibrium between the different metallo-assemblies present in solution can also be shifted by the presence of different anions. For example, addition of several equivalents of H2 PO4 – to a DMSO solution containing a mixture of [2+2] and [3+3] assemblies, shifted the equilibrium to the formation of only the [2+2] assembly. Surprisingly, addition of HSO4 – to the 1:1 mixture in DMSO of [Pd(dppp)(OTf)2 ] and L8 did not modify the equilibrium between the [2+2] and [3+3]. Williams has reported another type of system in which a dynamic combinatorial library of metal complexes is generated by metal-ligand interactions and controlled by anionic templates [51]. In this work it was shown that chloride can modify the distribution of products in an equilibrated solution containing cobalt(II) salts and 2,2 -bipyridyl ligands (bipy or the chiral ligand L10 – see Scheme 15). More specifically, mixing Co(NO3 )2 with bipy and L10 generated a library of complexes with general formula [Co(bipy)X (L10 )3–X ]2+ . This was shown by electrospray mass spectrometry which revealed that all the possible combinations of products were indeed present in solution:

200

R. Vilar

Scheme 15 Reaction scheme showing the different possible complexes that can be formed by mixing Co(NO3 )2 with L10 and bipy

[Co(L10 )3 ]2+ (80%), [Co(bipy)(L10 )2 ]2+ (100%), [Co(bipy)2 (L10 )]2+ (91%) and [Co(bipy)3 ]2+ (11%). Furthermore, 1 H NMR spectroscopy indicated that for each of the complexes – except for [Co(bipy)3 ]2+ which gave an enantiomeric pair – the corresponding ∆- and Λ-diastereomers were present. The equilibrium of the above mixture was shown to change upon addition of CF3 COOH. Addition of the acid led to protonation of the amines groups on L8 inducing diasteroselectivity and, as a consequence, some of the complexes initially present in the mixture disappeared. Interestingly, when DCl rather than CF3 COOH was added to the reaction mixture, only the two homoleptic complexes {Cl2 ⊂∆-[Co(L10 H2 )3 ]6+ } and [Co(bipy)3 ]2+ could be detected suggesting that chloride acts as a template amplifying the formation of these species. These observations have been rationalised on the grounds that coulombic repulsion between the protonated amines is minimized by the presence of chloride. A similar approach to the one discussed above has been recently reported by Rice [52, 53]. In these investigations it was shown that nitrate can modify the distribution of products in a mixture of cobalt(II) and two different N,N  chelating ligands. First, the reaction between Co(ClO4 )2 and ligand L11 was studied showing that a triple helicate with formula [Co2 (L11 )3 ](ClO4 )4

Scheme 16 Reaction scheme showing the different possible complexes that can be formed by mixing Co(ClO4)2 with L11 and L12

Anion Templates in Synthesis and Dynamic Combinatorial Libraries

201

formed (see Scheme 16). The ligands around the cobalt(II) centres generate “pockets” of the right size to bind perchlorate anions via hydrogen bonding (see Fig. 17).

Fig. 17 X-ray crystal structure of the triple helicate [Co2 (L11 )3 ](ClO4 )4 showing the perchlorate anions bound to the “pockets” formed by the three ligands

The authors then studied the possibility of anion exchange in this complex. Thus, upon addition of two equivalents of [Bu4 N][NO3 ] to [Co2 (L11 )3 ](ClO4 )4 a new host-guest complex with formula [Co2 (L11 )3 ](ClO4 )2 (NO3 )2 was obtained. The X-ray crystal structure of this mixed-anion helicate showed that the perchlorates initially bound to the binding pockets of the helicate, had been replaced by nitrates. Having established the basic host-guest chemistry between the helicate and the two anions, the authors then investigated the reaction between [Co(ClO4 )2 ]·6H2 O, L11 and L12 (see Scheme 16 for the chemical structure of the ligands) in a 2 to 1.5 to 1.5 ratio. This reaction resulted in the formation of four complexes: [Co2 (L11 )3 ]4+ , [Co2 (L11 )2 (L12 )]4+ , [Co2 (L11 ) (L12 )2 ]4+ and [Co2 (L12 )3 ]4+ , with a 1:3:3:1 statistical distribution. This was confirmed by both 1 H NMR spectroscopy and ES mass spectrometry. Interestingly, upon addition of KNO3 to this mixture a dramatic change in product distribution was observed. The two homoleptic complexes [Co2 (L11 )3 ]4+ and [Co2 (L12 )3 ]4+ were found to be the main components of the mixture (with only 5% of the heteroleptic compounds being present). This behaviour has been attributed to the strong binding of nitrate (which acts as a template) to the anion-binding pockets present in [Co2 (L11 )3 ]4+ . 4.2 Using Reversible Covalent Bonds A large number of dynamic combinatorial libraries reported in the literature make use of reversible covalent chemistry to generate the required dynamic equilibria [12]. In spite of this, there are very few examples of anion-directed

202

R. Vilar

DCLs where this type of reversible bond-formation is used to generate the virtual library of receptors. One of these examples has been reported by Otto and Kubic who developed a DCL of anion receptors [54]. The library is based on disulfide exchange reactions between a dimeric cyclic peptide (35) – where the two peptidic cycles are linked by a disulfide group – and a range of different thiol-substituted spacers a–f (see Scheme 17). Mixing of all these components yields a range of dimeric cyclic peptides in different proportions.

Scheme 17 Components of a DCL of dimeric cyclic peptides linked by disulfide bonds

Interestingly, addition of K2 SO4 or KI to the DCL amplified the formation of three receptors (35a, 35b and 35c), while addition of other anions such as chloride or fluoride did not shift the equilibrium. Two of these receptors (35b and 35c) were amplified more than 35a; these were then chosen for further studies and therefore prepared and isolated using a second generation biased library. Isothermal titration calorimetry studies revealed the binding constants between the selected dimeric-receptors and sulfate or iodide to be around 106 (in a mixture of MeCN/H2 O). This example nicely demonstrates how dynamic combinatorial chemistry can be employed to improve the selectivity and binding properties of a specific receptor; in this case this has been done by simply modifying the length and flexibility of the spacer that joins the two macrocycles. A different type of reversible covalent bond has been employed by Sessler to generate anion-directed DCLs. More specifically, the synthesis of new

Anion Templates in Synthesis and Dynamic Combinatorial Libraries

203

bipyrrole-based macrocyclic receptors has been achieved by reacting diamine 36 with diformylbipyrrole (37) in acidic media (see Scheme 18) [55]. Interestingly, depending on the acid used (namely HCl, HBr, CH3 CO2 H, CF3 CO2 H, H3 PO4 , H2 SO4 or HNO3 ) different distributions of oligomeric species and macrocycles were obtained. While several products were formed with most acids, in the presence of sulfuric acid the [2+2] macrocycle 38·2H2 SO4 formed in nearly quantitative yield.

Scheme 18 Synthesis of macrocycle 38·2H2 SO4 by reacting 36 and 37 in the presence of H2 SO4

The anion-free macrocycle 38 was isolated upon addition of triethylamine to 38·2H2 SO4 . This macrocycle was structurally characterised and its anionbinding properties studied, revealing high association constants between the receptor and HSO4 – (1:1; Ka = 63 500±3000 M–1 ) and H2 PO4 – (2:1; Ka1 = 191 000±15 400 M–1 and Ka2 = 60 200±6000 M–1 ). Interestingly, when 38 was allowed to stand in acetonitrile for 5 days in the presence of HSO4 – or H2 PO4 – (as tetrabutylammonium salts) a rearrangement of the poly-imine compound took place. More specifically, the [2+2] macrocycle 38 expanded into the [3+3] macrocycle 39 (see Fig. 18) quantitatively in the presence of H2 PO4 – (and in 47% yield in the presence of HSO4 – ). More recently, the same authors have published a comprehensive study in which they show a similar behaviour for other related di-amino and di-aldehyde building blocks [56, 57]. The nature of the anions (from the corresponding acid) present in solution, dictates the size of the macrocycle formed. Under specific circumstances, some of the macrocycles undergo anion-induced ring-expansion.

204

R. Vilar

Fig. 18 Schematic representation of the [3+3] macrocycle 39

The results discussed above, strongly suggest the presence of a dynamic equilibrium between various macrocyclic species in solution. Upon addition of the appropriate anionic template, amplification of one of them is then observed.

5 Conclusions and Outlook The first examples of anion-templated processes were reported in the early 1990s. Since then, this area of supramolecular chemistry has grown steadily showing the important role that anionic templates can play in directing the synthesis of specific molecules and supramolecular assemblies. A wide range of organic and metal-organic macrocycles have now been prepared by anion-templated processes. Similarly, the syntheses of molecules with complex topologies (such as those of pseudorotaxanes, rotaxanes and catenanes) can now be achieved by anion-directed assembly of simple building blocks. Another area where anion templates have had an important impact is in the synthesis of coordination cages and capsules. More recently, the use of anions as templates in dynamic combinatorial libraries has been realized showing

Anion Templates in Synthesis and Dynamic Combinatorial Libraries

205

that structurally challenging anion-receptors can be developed using this dynamic approach. This promises to be an area of important future developments with the potential to generate receptors that can not be easily achieved using more “classical” synthetic methodologies.

References 1. Schalley CA, Voegtle F, Doetz KH (eds) (2005) Templates in Chemistry II. Top Curr Chem, vol 249 2. Schalley CA, Voegtle F, Doetz KH (eds) (2005) Templates in Chemistry I. Top Curr Chem, vol 248 3. Vilar R (2004) Struct Bond 111:85 4. Diederich F, Stang PJ (eds) (2000) Templated Organic Synthesis. Wiley VCH, Weinheim 5. Gimeno N, Vilar R (2006) Coord Chem Rev 250:3161 6. Vilar R (2003) Angew Chem Int Ed 42:1460 7. Beer PD, Sambrook MR, Curiel D (2006) Chem Commun, p 2105 8. Beer PD, Wong WWH (2005) Macrocyclic Chem, p 105 9. Lankshear MD, Beer PD (2006) Coord Chem Rev 250:3142 10. Lankshear MD, Beer PD (2007) Acc Chem Res 40:657 11. Corbett PT, Leclaire J, Vial L, West KR, Wietor J-L, Sanders JKM, Otto S (2006) Chem Rev 106:3652 12. Rowan SJ, Cantrill SJ, Cousins GRL, Sanders JKM, Stoddart JF (2002) Angew Chem Int Ed 41:899 13. Otto S (2003) Curr Opin Drug Discov Dev 6:509 14. Lehn J-M (2007) Chem Soc Rev 36:151 15. Lehn J-M, Eliseev AV (2001) Science 291:2331 16. Otto S, Furlan RLE, Sanders JKM (2000) J Am Chem Soc 122:12063 17. Rowan SJ, Lukeman PS, Reynolds DJ, Sanders JKM (1998) New J Chem 22:1015 18. Eliseev AV, Lehn JM (1999) Curr Topics Microbiol Immunol 243:159 19. Lehn J-M (1999) Chem Eur J 5:2455 20. Melson GA, Busch DH (1964) J Am Chem Soc 86:4834 21. Thompson MC, Busch DH (1964) J Am Chem Soc 86:3651 22. Busch DH (1992) J Inclusion Phenom 12:389 23. Anderson S, Anderson HL, Sanders JKM (1993) Acc Chem Res 26:469 24. Meshcheryakov D, Boehmer V, Bolte M, Hubscher-Bruder V, Arnaud-Neu F, Herschbach H, Van Dorsselaer A, Thondorf I, Moegelin W (2006) Angew Chem Int Ed 45:1648 25. Bru M, Alfonso I, Burguete MI, Luis SV (2006) Angew Chem Int Ed 45:6155 26. Roitzsch M, Lippert B (2006) Angew Chem Int Ed 45:147 27. Maekawa M, Konaka H, Minematsu T, Kuroda-Sowa T, Munakata M, Kitagawa S (2007) Chem Commun, p 5179 28. Vilar R, Mingos DMP, White AJP, Williams DJ (1998) Angew Chem Int Ed 37:1258 29. Vilar R, Mingos DMP, White AJP, Williams DJ (1999) Chem Commun, p 229 30. Cheng S-T, Doxiadi E, Vilar R, White AJP, Williams DJ (2001) J Chem Soc Dalton Trans, p 2239 31. Diaz P, Mingos DMP, Vilar R, White AJP, Williams DJ (2004) Inorg Chem 43:7597

206

R. Vilar

32. Fleming JS, Mann KLV, Carraz C-A, Psillakis E, Jeffery JC, McCleverty JA, Ward MD (1998) Angew Chem Int Ed 37:1279 33. Paul RL, Bell ZR, Jeffery JC, McCleverty JA, Ward MD (2002) Proc Nat Acad Sci USA 99:4883 34. Paul RL, Argent SP, Jeffery JC, Harding LP, Lynam JM, Ward MD (2004) Dalton Trans, p 3453 35. Amouri H, Mimassi L, Rager MN, Mann BE, Guyard-Duhayon C, Raehm L (2005) Angew Chem Int Ed 44:4543 36. Amouri H, Desmarets C, Bettoschi A, Rager MN, Boubekeur K, Rabu P, Drillon M (2007) Chem Eur J 13:5401 37. Fyfe MCT, Glink PT, Menzer S, Stoddart JF, White AJP, Williams DJ (1997) Angew Chem Int Ed Engl 36:2068 38. Hubner GM, Glaser J, Seel C, Vogtle F (1999) Angew Chem Int Ed 38:383 39. Reuter C, Schmieder R, Vogtle F (2000) Pure Appl Chem 72:2233 40. Reuter C, Wienand W, Hubner GM, Seel C, Vogtle F (1999) Chem Eur J 5:2692 41. Seel C, Vogtle F (2000) Chem Eur J 6:21 42. Wisner JA, Beer PD, Drew MGB, Sambrook MR (2002) J Am Chem Soc 124:12469 43. Sambrook MR, Beer PD, Wisner JA, Paul RL, Cowley AR (2004) J Am Chem Soc 126:15364 44. Ng K-Y, Cowley AR, Beer PD (2006) Chem Commun, p 3676 45. Hasenknopf B, Lehn J-M, Kneisel BO, Baum G, Fenske D (1996) Angew Chem Int Ed Engl 35:1838 46. Hasenknopf B, Lehn J-M, Boumediene N, Dupont-Gervais A, Van Dorsselaer A, Kneisel B, Fenske D (1997) J Am Chem Soc 119:10956 47. Campos-Fernandez CS, Schottel BL, Chifotides HT, Bera JK, Bacsa J, Koomen JM, Russell DH, Dunbar KR (2005) J Am Chem Soc 127:12909 48. Campos-Fernandez CS, Clerac R, Dunbar KR (1999) Angew Chem Int Ed 38:3477 49. Chand DK, Biradha K, Kawano M, Sakamoto S, Yamaguchi K, Fujita M (2006) Chem Asian J 1:82 50. Diaz P, Tovilla JA, Ballester P, Benet-Buchholz J, Vilar R (2007) Dalton Trans, p 3516 51. Telfer SG, Yang X-J, Williams AF (2004) Dalton Trans, p 699 52. Harding LP, Jeffery JC, Riis-Johannessen T, Rice CR, Zeng Z (2004) Dalton Trans, p 2396 53. Harding LP, Jeffery JC, Riis-Johannessen T, Rice CR, Zeng Z (2004) Chem Commun, p 654 54. Otto S, Kubik S (2003) J Am Chem Soc 125:7804 55. Katayev EA, Pantos GD, Reshetova MD, Khrustalev VN, Lynch VM, Ustynyuk YA, Sessler JL (2005) Angew Chem Int Ed 44:7386 56. Katayev EA, Boev NV, Khrustalev VN, Ustynyuk YA, Tananaev IG, Sessler JL (2007) J Org Chem 72:2886 57. Katayev EA, Sessler JL, Khrustalev VN, Ustynyuk YA (2007) J Org Chem 72:7244

Struct Bond (2008) 129: 207–248 DOI 10.1007/430_2008_084 © Springer-Verlag Berlin Heidelberg Published online: 25 April 2008

Molecularly Imprinted Polymers Using Anions as Templates Sally L. Ewen · Joachim H. G. Steinke (u) Department of Chemistry, Imperial College London, South Kensington Campus, London SW7 2AZ, UK [email protected] 1 1.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

208 208

2 2.1 2.2 2.3 2.4 2.5 2.6

Molecularly Imprinted Polymers . . . . . . The Concept . . . . . . . . . . . . . . . . . . Inspiration, Incorporation and Assimilation A Brief “Developmental” History . . . . . . . Scope and Limitations . . . . . . . . . . . . . Formats . . . . . . . . . . . . . . . . . . . . Design Criteria . . . . . . . . . . . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

209 209 210 211 215 217 219

3 3.1 3.2 3.3 3.4 3.5

Molecular Imprinting Using Anionic Templates Anionic Phosphate Derivatives . . . . . . . . . . Carboxylates . . . . . . . . . . . . . . . . . . . . Anionic Sulfate Derivatives . . . . . . . . . . . . Other Anions . . . . . . . . . . . . . . . . . . . Concluding Comments . . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

222 225 232 238 240 242

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

244

. . . . . . .

Abstract Molecularly imprinted polymers (MIPs) are a class of solid-phase artificial receptors that are prepared using templates during the polymer network forming step; these are subsequently removed to generate the selective receptor sites. The ease of synthesis, the possibility of nanomolar binding constants and high levels of molecular discrimination, as well as environmental stability and ability to be reused, have led to a dramatic boost in research interest in MIPs. One particularly promising area of study is the use of anionic templates in the synthesis of MIPs and the targeting of substrates that carry biologically important anionic functionalities. Benefiting from concurrent developments in supramolecular receptor design and synthesis, it has become clear that MIPs for anion recognition will impact biological screening, diagnosis, point-of-care devices (including online sensing) and read-out. Keywords Anionic templates · Molecularly imprinted polymers · Self-assembly · Biomimicry · Synthetic receptors Abbreviations FIP Functional group imprinted polymer ISFET Ion-sensitive field effect transistor

208 ITC ITO MIP MPA NIP NMR ODS PMP SDS SPE TSA

S.L. Ewen · J.H.G. Steinke Isothermal titration calorimetry Indium tin oxide Molecularly imprinted polymer Methylphosphonic acid Non-molecularly imprinted polymer Nuclear magnetic resonance Octadecylsiloxane Pinacolyl methylphosphonate Sodium dodecyl sulfate Solid phase extraction Transition state analogue

1 Introduction Molecular templates are involved universally in relaying structure and function into both biological and synthetic molecular assemblies. Information on shape, size and electronic spatial configuration of a template, i.e. the molecular state of a molecule, is used to guide and control the formation, structure and function of new self-assembled molecular entities. The template either becomes part of the new structure or is separated from it and reused, or involved in other templated processes. Messenger RNA in the polymerase chain reaction functions as a reusable template for new complementary polynucleotide strands [1]. Diblock copolymers can be used as templates for mineralisation of non-trivial inorganic phosphate structures [2, 3]. Patterned surfaces of phosphate groups template the mineralisation and resulting complex morphology of inorganic salts such as calcium carbonate [4]. In liquid crystals the addition of dopants can cause a switch to a different liquid crystalline phase [5]. The formation of a molecular capsule can be triggered through the presence of the correct solvent which occupies the interior [6]. Micelles have been used for many years as templating agents in the formation of zeolites [7]. In this review we will be discussing synthetic polymer receptors formed as crosslinked networks in the presence of molecular templates. The focus will be on anionic templates and the performance of the resulting solid-phase receptor structures in terms of chemical selectivity and molecular recognition. 1.1 Context Anion templation as a means of producing molecularly imprinted polymers has to deal with the issues that one encounters designing and synthesising molecular receptors for anions in general [8–10]. The increasing understanding of the factors involved in anion recognition, such as how to design receptor sites and ligand arrays that are appropriate for the larger and more diffuse

Molecularly Imprinted Polymers Using Anions as Templates

209

charge distribution in anions, is one reason for the recent surge in interest in preparing MIPs for anion recognition. Another reason is the impact of proteomics and the challenges and opportunities that accompany it with regard to sensing and detection of charged species (simple anions as well as carboxylate, phosphate and sulfate derivatives) in the interrogation and analysis of complex biological systems. A third reason relates to advances in the ability of chemists to apply supramolecular and self-assembly approaches to the design of complex structures and materials, from which MIP synthesis has already begun to benefit.

2 Molecularly Imprinted Polymers 2.1 The Concept The synthesis of a molecularly imprinted polymer is at first sight a straightforward affair. A template molecule, which can essentially be freely chosen, is mixed together with one or several polymerisable receptor molecules (“functional monomers”) [11–14] (Fig. 1). The latter, so-called binding sites, are selected on the basis of preferably strong and directed interactions with the template molecule to maximise the molecular fidelity of the imprinting process (covalent and non-covalent interactions are possible). This step

Fig. 1 Pictorial representation of molecular imprinting, showing a formation of prepolymerisation complex; b polymerisation; c removal of template molecule to generate molecularly imprinted recognition site

210

S.L. Ewen · J.H.G. Steinke

takes place once template and binding sites have self-assembled, typically in solution, and crosslinking molecules have been added which produce a crosslinked polymer network once polymerisation has been initiated, in most instances thermally or photochemically although electrochemical or redox initiation has also been demonstrated successfully. The template is then removed from the polymer, most commonly through liquid extraction, leaving behind a polymer matrix made up of supramolecular receptor sites which have been formed throughout the polymer network as a result of the preorganisation of the binding site/template complex [15]. These receptor sites are often referred to as molecularly imprinted cavities as a polymer matrix has formed around the template, becoming part of the receptor structure. The complementarity of shape and electron density distribution transferred into the polymer matrix has obvious analogies with the substrate recognition sites of enzymes or antibodies where the spatial organisation of the functional groups of amino acid side chains define catalytic sites and substrate selectivity [12]. As the ability of conformational adaptation to the substrate is a key feature of enzyme recognition and catalysis, so is a certain level of conformational freedom in a MIP [16]. Regarding the molecular recognition mechanism operational in MIPs, a number of recent studies illustrate the decade-old controversy between the imprint mechanism (cavity generated by template) and the association mechanism (trapped templates act as binding/nucleation sites) [17, 18]. 2.2 Inspiration, Incorporation and Assimilation The concept of molecularly imprinted polymers is strikingly simple and generic. Very often MIPs are referred to as synthetic enzyme or antibody mimics due to the analogy of having spatially defined functional groups positioned inside a molecular pocket. Once this simplistic view took hold, most MIP research revolved around identifying the level to which MIPs can mimic the structure and function of protein receptors and catalyst active sites [11–14]. However, just as inspirational as the likening of MIPs to antibodies and enzymes has been, as incomplete is the appreciation of their potential under conditions where proteins are mostly unsuitable or inadequate; examples are their use in organic solvents and in variable and extreme temperature environments. In recent years, MIPs have been discussed more generally as a special type and format of receptor design and structure, within the wider chemical world of supramolecular chemistry [19–21]. Consequently more of the advances in self-assembly and molecular recognition have been explored and incorporated into MIP design and synthesis. This trend will be even more beneficial to the development of MIPs in years to come, during which many concepts and approaches will have been assimilated, causing a step change in MIP performance. Some instruc-

Molecularly Imprinted Polymers Using Anions as Templates

211

tive examples of this progress include stoichiometric non-covalent binding sites (vide infra Sects. 2.3 and 3), high-throughput and parallel synthesis experimentation [22–24], integration of sensing modules into MIPs [25, 26], new polymerisation methods [27], modelling approaches of templating complexes [28, 29] and single site MIPs as dendrimer [30] or microgel [31]. Future activities will include: improvements of recognition in water [32]; the use of (dynamic) combinatorial chemistry [10, 33] and thermodynamically controlled approaches for MIP synthesis; the use of MIPs as the sensing component for medical diagnostics [34] and environmental analysis [35], also as polymeric drugs, but, more importantly, as drug delivery vehicles (including molecular tags and labels) [36]; the application of MIPs as identification and separation tools for biological screening [37]; the preparation and processing of MIPs with smaller feature sizes and more complex shapes [38, 39]; the successful use of more challenging template molecules (proteins, complex drug molecules). On the whole, MIPs are relatively easily prepared multidentate receptors, in contrast with small molecule receptor analogues, and have shown nanomolar binding affinities [40]. Their solid-state nature is in itself a useful format for screening strategies, with further benefits arising from long-term stability, recyclability and modularity. 2.3 A Brief “Developmental” History Polyakov et al. were the first to report the effect of generating selectivity in a polymer matrix through addition of a template molecule (for a more detailed account see [14]). In 1931 they prepared silica particles from sodium silicate and (NH4 )2 CO3 in water, adding benzene, toluene or xylene in the process. After prolonged drying and exhaustive extraction with hot water, the polymer showed higher capacity for the additive (template) than for structurally related compounds. A contemporary debate ensued about the origin of antibody selectivity in the immune system with contributions from Breinl and Haurowitz [41], Mudd [42] and Linus Pauling [43]. The latter favoured the view that antibodies are generated in the presence of an intruding antigen which would determine the antibody conformation (as he attempted to demonstrate in 1942 [14]). Similar work to that of Polyakov was reported in 1949 by Dickey, this time using alkyl orange dyes as the templates. The imprinted silica gel showed pronounced selectivity for the template, which was present throughout the silica network-forming step [44]. Curti et al. showed the first examples of enantioselectivity with both mandelic acid and camphor sulfonic acid (also in silica) and their application as stationary phases for chromatography (Fig. 2) [45, 46]. A particularly imaginative example was contributed by Patrikeev et al. in 1960, in which they resorted to bacteria as templates. Curiously the imprinted

212

S.L. Ewen · J.H.G. Steinke

Fig. 2 A selection of important anionic template molecules and their corresponding binding sites reflecting the development of MIPs

Molecularly Imprinted Polymers Using Anions as Templates

213

silica promoted certain bacterial growth over that of control samples imprinted with other varieties [47]. In a related study either the levo or dextro form of the same bacillus (Bacillus cereus var. mycoides) was incorporated into a silica gel matrix showing discrimination for L- and D-linanool vapour respectively [48]. The same group provided perhaps the earliest example of a catalytic MIP templating amino acid condensation reactions, though rates were enhanced modestly by a factor of two and experimental details are scarce [49]. Two decades later, in 1972, Wulff et al. [50] and Klotz et al. were independently pioneering the use of organic polymers as tailor-made receptors. Wulff et al. were the first to harness the versatility of radical vinyl chemistry as polymerisation methodology and exemplified the covalent imprinting approach (Fig. 2), whereas Klotz and coworkers harnessed the benefits of ring-opening polymerisation combined with reversible disulfide crosslinking employing non-covalent interactions [51]. In the following years Wulff et al. elaborated the covalent approach with the addition of non-covalent (predominantly electrostatic) interactions [52, 53]. A method to imprint on a surface (“surface imprinting”) was first developed by Sagiv et al. in 1979 using silica particles as surfaces with polymerisable siloxane as surface modifiers [54]. This methodology is transferable to other surfaces as long as the template absorbs onto the chosen surface [55]. The work in the group of Mosbach in 1981 revolutionised molecular imprinting through the non-covalent approach [56] (Fig. 2). The impact derived from the much simpler synthetic route for making MIPs. Rather than having to prepare templates connected to polymerisable binding sites via reversible covalent bonds, non-covalent interactions allow the self-assembly of template and binding sites in solution, affording a pre-polymerisation complex prior to vinyl monomer network formation. In further development, Mosbach et al. demonstrated that MIPs with high selectivity could be obtained even without the necessity to invoke covalent or ionic bonds [57]. With time, more and more groups have taken up the non-covalent approach and today it is the most widely used methodology to prepare MIPs [14]. However, comparative studies of covalent and non-covalent imprinting are rare, and even those reported lack the confidence in which optimised conditions have been employed for both approaches, so that clear conclusions from which to select the better approach could not be drawn [58, 59]. Although the notion of combining the advantages of covalent bonds during the imprinting step with those of non-covalent interactions for rebinding was present in earlier work by Wulff et al. [50, 52, 60, 61], the first example of using exclusively covalent interactions for the imprinting step and noncovalent ones for rebinding was executed by Sellergren and Andersson in 1990 [62]. Polymerisable groups were linked by ester bonds, though the target molecule, p-aminophenylalanine ethyl ester, was necessarily different to the template. This changed completely in 1994, when Whitcombe et al. significantly refined the semi-covalent approach, outlining a concept based on a small covalent

214

S.L. Ewen · J.H.G. Steinke

fragment as space holder (“sacrificial spacer approach”) [63]. With this strategy, template and a binding site are covalently linked with a cleavable spacer which is designed to reveal complementary functional groups upon removal, closely following the geometry of the covalent juncture for improved complementarity [64]. The semi-covalent approach is an attempt to synergise the advantages of the covalent methodology (strict control of functional group location, more uniform distribution) with that of the non-covalent one (reduced kinetic restriction upon rebinding). As a “natural” coalescence of the covalent and the original non-covalent approach, a stoichiometric non-covalent approach was developed [65]. If the non-covalent interactions are strong enough to produce association constants of at least 103 M–1 (or preferably higher) the equilibrium will lie well on the side of the template-functional monomer complex. On the way to stoichiometric non-covalent interactions (discussed in detail in Sect. 3) a representative example of carboxylic acid templates and various N-base binding sites is found in Kempe et al. (1993) [66]. Polymerisable amidines were first used for stoichiometric imprinting of a transition-state analogue (TSA) by Wulff et al. in 1997 (Fig. 2) [67]. Other noteworthy developments were valine-derived binding sites for dipeptide imprinting by Yano et al. in the same year [68] and bidentate dioxoborolanyl recognition sites for carboxylates exploited by Lübke et al. in 2000 [69]. In 1994 Sellergren was already successful in complementing the standard MAA binding site monomer with pentamidine, offering association constants close to those typically considered to be required for stoichiometric imprinting strategies (Fig. 2) [70]. With time and improved synthetic protocols, larger templates (fullerenes, dendrimers, nanoparticles, colloids, micelles, lipid bilayers, self-assembled block copolymers, oligonucleotides, DNA and proteins) have been imprinted [14] and the choice of matrices has expanded to liquid crystal polysiloxanes, carbon networks, zeolites, layered aluminophosphates and colloidal crystals, though organic polymer networks remain the dominant imprint casting medium [14]. Further important developments related to templates emerged in the midand late 1990’s. In 1999, Sreenivasan et al. demonstrated that it is possible to imprint with two different template molecules simultaneously [71]. A year later, Rachkov and Minoura introduced the epitope imprinting methodology in which a structurally unique 3-amino acid fragment of a peptide chain was shown to be a sufficient template to produce MIPs that are selective for the entire nonapeptide [72]. Another year later, Sellergren et al. used a template analogue rather than the target molecule itself for generating MIPs [73], further suggesting that complex molecules can be imprinted successfully on the basis of a molecular information-rich substructure or derivative. Also, manipulation of the binding site chemistry has widened the opportunities for MIP applications. Useful synthetic post-polymerisation transformations were elaborated on disulfide templates by Mukawa et al. in 2002, as their reduction

Molecularly Imprinted Polymers Using Anions as Templates

215

was found to yield thiols as binding sites (semi-covalent approach), which can also be oxidised up to sulfonic acids, changing the chemical nature of the non-covalent binding site in a controlled manner [74]. The first MIPs designed for influencing molecular reactivity came from the laboratories of Shea et al. in 1978 [75], closely followed by Damen and Neckers in 1980 [76]. Cycloadditions leading to cyclopropane and cyclobutane dicarboxylic acids respectively were performed, with significant regio- and diastereoselectivity in the latter case. Also in 1980, Belokon et al. organised amino acid templates via Schiff base formation inside a MIP cavity, and showed that upon deprotonation the amino acid carbanion retained its original configuration unlike the same reaction in solution [77]. A few years later, in 1987, Sarhan et al. showed the possibility of stereospecific reversal of configuration rather than retention with mandelic acid [78]. Andersson et al. achieved the catalytic deprotonation of bound amino acids (vide supra Belokon et al.) electing a pyridoxal-coenzyme analogue as a template in a non-covalent imprinting system [79]. In this particularly eventful year Wulff et al. were the first to carry out an asymmetric reaction in a chiral cavity designed to generate optically active amino acids, which was achieved with an ee of 36% and used glycine as a template [80]. The observed chiral induction is solely the result of enantioselective induction by the chiral cavity. The above examples give a taste of the potential of MIP cavities as nanoreactor sites. MIPs that are catalytically active (apart from some early work in silica matrices [81]) and which thereby influence reaction outcomes, were only embarked upon in 1987, by Leonhardt et al. [82]. Initially the product of a reaction was pursued, and not the better mimicry of using a transition state analogue. It produced esterolytic activity with a 2–3 fold rate enhancement and substrate selectivity of MIPs. Two years later however Robinson et al. prepared a MIP imprinted with a TSA. Poly(vinylimidazole) and p-nitrophenylphosphonate TSAs were coordinated to a CoII ion and the resulting template assembly was crosslinked with dibromobutane. Despite the more rational design of the template only a small enhancement of 1.6-fold was observed [83]. As the development of catalytic MIPs is strongly intertwined with the evolution of stoichiometric non-covalent binding sites as oxyanionic receptor sites, we will return to this connection in one of the later sections (Sect. 3). 2.4 Scope and Limitations Discussions about the current and general scope and limitations associated with MIPs are ongoing. For a while, a good number of publications were of two minds in their assessment of the potential of MIPs, where small advances in performance were hailed as stepping stones to rival, if not excel, antibody and enzyme methods. This was followed by a healthy sobering period. Over the last ten years a much more realistic assessment of the potential of

216

S.L. Ewen · J.H.G. Steinke

MIPs has emerged, accompanied by a substantial increase in the number of researchers involved in MIP science and technology, creating a rise in publication numbers. Strengths of MIPs include: • Robustness/longevity – MIPs have been shown to keep their molecular recognition performance over months and even years; this is a stark contrast to enzymes and antibodies. Depending on the chosen polymer matrix, MIPs are chemically and physically robust and not limited to an aqueous environment or well-controlled physiological conditions (pH, temperature, solvent). MIPs can be recycled and reused [84]. • Solid-state format – Most MIPs are solid-phase which makes them amenable to automated processes requiring only extraction and reloading cycles, with substrate isolation being a simple filtration step [85]. • Relative ease of synthesis – The concept of molecular imprinting is simple and elegant, which is also reflected in the relatively straightforward manner in which they can be synthesised [14]. This is certainly true if one compares MIP synthesis with other strategies available to produce synthetic receptors. Many of the latter require a large number of synthetic steps, rather than self-assembly, to achieve the spatial positioning of functional groups identified as being key for mimicking enzyme catalysis or antibody performance. • Relative ease of design – MIPs rely entirely on self-assembly (cf. Pauling’s theory of antibody formation) and present a modular system, which also allows MIPs to be designed more easily in contrast to approaches in which the geometric requirements of the receptor molecule have to be carefully calculated or estimated during the design phase [86] (though the originally chosen template may not be the best “template” for the desired performance) [87, 88]. • Unrestricted choice of substrate – As long as it is possible, either in solution or in bulk, to self-assemble template with binding sites, and as long as the polymerisation chemistry is compatible with such a complex, the choice of template molecule would appear to be unrestricted [14]. • Can be made to give a response (e.g. QCM, fluorescence). – Through the addition of multi-functional binding sites, additional function for sensing, or some other form of responsive behaviour, can be incorporated without having to redesign template or receptor sites [25, 26, 34, 35].

Molecularly Imprinted Polymers Using Anions as Templates

217

Limitations of MIPs: • There are some common limitations associated with MIPs regardless of the chosen template, binding site, solvent, etc.: – Binding sites are heterogeneous (polyclonal rather than monoclonal as in enzymes and monoclonal antibodies) as a consequence of the statistical nature of the polymer network forming process, leading to undesirable band broadening in chromatographic separations and to a distribution of different activities of catalytic sites [89]. – Similarly not all template molecules may be accessible within the polymer matrix, or will require very long extraction times leading to problems of template leaching especially when employing MIPs for trace analysis [35]. – Diffusion to and from the recognition sites within the MIP matrix is typically slow, depending on the solid-phase format and phase structure. High pore volume and interconnectivity can reduce the diffusion problem but generating such a hierarchical pore structure may not be compatible with the requirement of forming a strongly associated template complex. Thin film, surface imprinting and membrane approaches are alternative means of addressing diffusion problems but are in many cases accompanied by lower selectivity due to interface effects or changes in the crosslinking stoichiometry, leading to a reduction of the molecular fidelity of the imprint [14]. – For applications such as enantiopolishing or chiral separations, MIPs are said to offer low capacity. This is true when comparing MIPs with sorbents that rely on an interaction with a surface or a surface modified with a chiral selector (e.g. Pirkle phases). On the other hand, MIPs are at least competitive if one compares their atom economy with that of enzymes or antibodies adding up the molar mass of crosslinker and receptor sites per template (between ∼5–100 kDa depending on synthesis recipe) [14]. – Dealing with polar and particularly water soluble template molecules and recognition in aqueous milieu have been major issues for MIPs for many years. Progress across the field however indicates that generic strategies are available to overcome this limitation [34]. • Major issues regarding deficits in technology for anion MIPs will be discussed in more detail in Sects. 3.1–3.4. 2.5 Formats Molecularly imprinted polymers are implicated for a wide range of applications. Their solid-phase nature necessitates that the synthesis step includes the processing step. As is the case with any type of covalently crosslinked ma-

218

S.L. Ewen · J.H.G. Steinke

terial, once polymerisation has taken place the molecular structure is fixed and cannot be reconfigured. The most common formats encountered for MIPs are: • Irregular particles – The most popular format prepared by grinding the bulk-polymerised MIP into smaller particles. For chromatographic applications this is usually followed by sizing through sieving [90]. • Monoliths – Upon crosslinking, the polymerisation mixture retains the shape of the reaction vessel. Separating the inner wall of the reaction vessel from the polymer reveals a single piece of MIP, a monolith. The monoliths could be used for MIP applications that have been shown to be useful as a direct means of packing chromatographic columns (HPLC, CEC) [90, 91]. If the monolith is exposed to large variations in solvent polarity without space constraints it will slowly disintegrate into smaller particles due to the high density of crosslinks in connection with a glass transition temperature (Tg ) that is usually well above room temperature. • Beads – This third format is particularly useful to pack chromatography columns and generally to manipulate MIPs in automated processes [90]. As it requires a two- or multi-phase solvent system during polymerisation to impart the spherical shape, until recently it has proven quite challenging to obtain narrow disperse bead sizes with high yield at the desired size. Protocols are now available that offer generic solutions for beading a wide range of MIP formulations [92]. Beads are particularly desirable for catalytic MIP applications as selectivity improves (compared to monolith synthesis) and catalytic activity increases due to faster mass transfer [93]. • Films – Especially for sensing (UV-Vis, luminescence, QCM), film formats, i.e. surfaces coated with a MIP layer, are attractive as a simple means of device fabrication. This does not differ greatly from the monolith format, and film thicknesses can be controlled through an appropriately shaped substrate or through spin coating. The changes in solvent concentration caused by evaporation of solvent and precipitation of the MIP during the film forming processes will alter the stoichiometry of the template/binding site complex set in the starting solution [94]. • Membranes – Free standing films of MIPs have been prepared in various ways. Mechanical robustness has been imparted through lower crosslink

Molecularly Imprinted Polymers Using Anions as Templates

219

densities or monomers generating lower Tg polymer backbones compared to monoliths. Reducing crosslink density or rigidity generally reduces selectivity and long-term stability of MIPs. Membrane composites offer a means of accommodating brittle MIP particles into an overall flexible matrix. Phase inversion techniques with polar highperformance polymers have also been successfully applied to form MIP membranes [95]. • (Hydro)gels – Instead of high crosslink levels and thus polymer networks with high Tg , MIP gels and hydrogels can be produced in the same way as monoliths, by selecting monomers leading to lower Tg polymer backbones and/or by choosing polar monomers with high affinity to water and protic solvents, so that the solvent causes the polymer network to swell and to become more flexible, i.e. gel-like. Interesting applications are antibody mimics with MIPs becoming single site catalysts [31] and, more recently, controlled drug delivery [96]. • Single molecule species – Zimmerman et al. were the first to have synthesised a dendrimer imprinted with a porphyrin derivative using the sacrificial spacer methodology: a single cavity within a single polymer confinement structure [30]. • Spun fibres – Electrospinning has been applied to MIP synthesis employing membrane formation techniques with the help of a support polymer, promising higher surface area, faster diffusion and more efficient washing out of the template molecule [97].

2.6 Design Criteria MIPs are complex synthetic receptors prepared by self-assembly, followed by network formation. They are by their very nature amorphous solids, and those are inherently difficult to analyse at the molecular level. Most analyses of MIPs are indirect, assessing the fidelity of the molecular recognition events through binding assays of various kinds. Nuclear magnetic resonance (NMR) studies on the self-assembled complex formed prior to polymerisation as well as molecular modelling have been shown to be useful in arriving at a more rational design of MIPs [98], though pre-polymerisation analysis and modelling [99–102] are still in stages too early to extrapolate trend patterns and reach more general conclusions on reaction conditions, binding strength and template/binding site complex formation. The vastness of the

220

S.L. Ewen · J.H.G. Steinke

parameter space involved when synthesising MIPs has prompted the use of parallel synthesis combined with statistical data analysis to more rapidly analyse data sets and identify trends. These activities will mature and, with time, will have wider impact on our ability to design MIPs for any given template molecule with tunable thermodynamic and kinetic performance parameters. Although there are always exceptions, the following design criteria have been selected as being useful and reliable guides that help one arrive at a reasonable starting point for developing a new MIP: • Polymerisation temperature – Lower temperatures improve selectivity and capacity. The exothermic nature of radical vinyl polymerisation compounded by the gelling of the imprint mixture is likely to cause higher internal temperatures than those used to control the environment externally. • Binding site design – Statistical copolymerisation kinetics with the crosslinker lessens the heterogeneity of receptor sites. – Too much conformational freedom between binding site and polymer backbone reduces selectivity. – Strong, ideally close to stoichiometric ratios of binding interactions improve the quality of the imprint. – Faster equilibration kinetics increase performance in chromatographybased application. • Crosslinker – There is an optimum level of crosslinking with typical values to be around 70–95 mol %. Conformational flexibility of the crosslinker has to be balanced with the overall level of crosslinks. The chemical nature of the crosslinker determines the useful solvent range for a MIP. The size of the crosslinker ideally matches the dimensions of the template, to avoid conformational frustrations within the polymer matrix as a result of a structural misfit, which can lead to a loss in capacity and/or selectivity. • Solvent – The best choice is a solvent that maximises template binding site interactions while producing a highly porous MIP matrix to minimise diffusion limitations. • Template – Imprinted cavities can be viewed as multi-point receptor sites plus the added shape of the cavity as an additional recognition feature. A template offering many points for binding and where stoichiometric binding sites for the exhibited functional groups exist offers the best conditions for high selectivity. On the other hand, large templates

Molecularly Imprinted Polymers Using Anions as Templates

221

with many binding positions become more difficult to imprint with as the differences between related molecules become relatively smaller. This is also the case for templates with only one functional group, where the shape of the cavity becomes a more important contributor. Larger templates pose the problem of being difficult to extract from the polymer as they more readily become entrapped. Degradable templates or template fragments (epitopes) have been shown to be helpful in overcoming this limitation. Conformationally dynamic and demanding templates such as proteins are a particular case in point. Presenting proteins in native form on a degradable nanocrystal surface, for example, is an ingenious way to address both shape stability of the template and access to the imprinted cavity. • Template to binding site stoichiometry – A good starting point is a ratio that reflects the number of sites available on the template for interaction with the chosen polymerisable receptor. For covalently bound templates or stoichiometrically coordinating binding sites the ratio is obvious. • Crosslinker to self-assembled complex stoichiometry – The minimum number of “atoms” that are required to form the desired binding pocket and typical stoichiometries that give high selectivity are 20 : 1 in equimolar terms. Larger ratios can lead to enhanced selectivity but this is traded against a loss in capacity. • Covalent versus non-covalent imprinting – The covalent approach is synthetically more labour-intensive but produces MIPs that have better defined affinity distributions. It is still undecided as to which situation one of these two strategies will lead to the higher performing MIP. The polymerisation chemistry and other network forming processes applied to MIP generation (e.g. membrane phase inversion) lack a thermodynamic means of controlling the formation of receptor sites to render them less polyclonal. Some means of increased control such as living [103] and thermodynamically controllable polymerisation chemistry [27, 104, 105] (in analogy to dynamic combinatorial library synthesis [106]), are at a very early stage. Improved binding site design, especially in water or buffered system, would benefit from hydrophobic effects for recognition cf. enzyme pocket; for this, however, one needs to incorporate the necessary conformational/environmental change associated with it. Also, the area of modelling for selecting optimised stoichiometries for MIP manufacture would become a more powerful means of guiding the synthesis, with more precisely defined interactions that generate an imprint with increased 3D/shape fidelity.

222

S.L. Ewen · J.H.G. Steinke

3 Molecular Imprinting Using Anionic Templates One of the most attractive features of molecular imprinting is its applicability to such a diversity of analytes. Thus the literature boasts MIPs made using a striking array of templates, which range from small molecules (<100 Da) to biological macromolecules (several kDa), and include sugars, steroids, dyes, herbicides, pharmaceuticals and biochemicals amongst others (Table 1). Nevertheless, upon inspection it is apparent that anionic templates have seldom been considered throughout the history of imprinting. Indeed, up until the last 7 or 8 years, there has been a surprising paucity of MIPs that are designed either to recognise “naked” anions (such as oxyanions, halides or cyanide) or to exploit the predominating anionic character of a particular analyte. Indirectly, however, anionic templates have played a role as early as the 1940s up to the present time. For example, Dickey prepared silica gels in the presence of methyl orange (p-dimethylamino-azobenzenesulfonic acid) and then later on using a sulfonamide analogue as template [44, 107]. In a subsequent communication it was disclosed that the two MIPs exhibited very similar selectivity for their target molecule, indicating that the negative charge on the p-substituent was nonessential to the recognition mechanism [107]. Curti et al. imprinted in silica using enantiomers of mandelic acid and camphorsulfonic acid. While the conditions for their MIP preparation (pH 4) suggest that the sulfonic acid, and to some extent the carboxylic acid, were dissociated throughout the imprinting process, the aforesaid functional groups were not deliberately targeted in the MIP synthesis [45]. In a more recent example, Rajkumar et al. generated a MIP for fructosyl valine recognition, using boronic acid-functionalised monomers to covalently interact with the diol moieties [108]. The carboxyl group, which would have been fully dissociated during the rebinding event (pH 11.4), was not targeted by any functional monomer. The relative scarcity of anion-imprinted polymers is especially remarkable given the prevalence of anionic molecules of importance. It is widely recognised that the design of anion hosts is a challenging task, especially in contrast to cation receptor design. The main reason for this is the diffuse nature of anions. Since Anions are significantly larger than the equivalent isoelectronic cations, they have a lower charge to radius ratio; consequently, electrostatic binding interactions are markedly diminished. For the same reason, anions are generally more polarisable than cations, and it is therefore more difficult to develop specific binding sites for their recognition. With regard to molecular imprinting, a significant hindrance to the development of anion-recognition MIPs has been the fact that anionic species are very often incompatible with apolar media. The molecular imprinting process involves the formation of a pre-polymerisation complex between the template molecule and functional monomers. This is typically based upon H-bonding

Molecularly Imprinted Polymers Using Anions as Templates

223

Table 1 Examples of molecules that have been used as templates for molecular imprinting Template molecule

Selected Refs.

Cyclobarbital

[162, 163]

Naproxen

[164–166]

(S)-Propranolol

[167]

L-Menthol

[27]

Glucose

[168, 169]

Cholesterol

[64, 170, 171]

Nicotine

[172, 173]

Bisphenol A

[174]

224

S.L. Ewen · J.H.G. Steinke

and electrostatic interactions (though it might also include dipole–dipole attractions or metal ion coordination). In general, such interactions are maximised when a non-competitive, apolar solvent is employed during the imprinting step, thereby affording a well-defined binding site. Similarly, the use of such a solvent for subsequent rebinding of the target molecule tends to promote specific binding as well as minimise levels of undesirable nonspecific interactions. The last 20 years have witnessed notable advances pertaining to the compatibility of molecular imprinting with polar/aqueous media. While much of that is outside of the scope of this review, one particular issue – the influence of functional monomers used – is of prime importance to our discussion. Traditionally, the majority of MIPs have been made using a limited repertoire of simple, commercially available functional monomers (for example, methacrylic acid, methacrylamide, 1-vinylimidazole), in accordance with the non-covalent imprinting strategy. In this regard, Simon and Spivak recently investigated a selection of commercially available functional monomers deemed to be particularly suitable for oxyanionic templates (in particular, for carboxylates and phosphonates) [109]. They prepared a series of MIPs, incorporating various functional monomers including pyridine-derivatives, aliphatic amines, and 1-vinylimidazole. A chiral template molecule, t-Boc-Lphenylalanine, was employed to study enantioselective discrimination. The MIP particles were slurry-packed into HPLC columns, and capacity factors and separation factors of the L- and D-enantiomers were calculated to provide values for binding affinity and enantioselectivity. Interestingly, the authors observed that, while the highest binding affinity was exhibited by the more basic aliphatic amine derivatives, the best selectivity was obtained by functional monomers bearing aromatic amine groups. Directional binding, as afforded by these latter functional monomers, is evidently vital for the production of selective binding sites. However, while the use of simple, commercially available functional monomers is relatively undemanding from a synthetic chemistry perspective, the drawback is that such monomers generally tend to afford only relatively weak binding interactions. This is all the more evident in aqueous/polar media. As a consequence, it is usually necessary to imprint using a large excess of functional monomer relative to the amount of template molecule, in order to push the equilibrium and ensure a high degree of complexation. Unfortunately, this approach results in undesirable binding site heterogeneity and subsequent non-specific rebinding of the analyte. The introduction of a novel imprinting methodology, which has attracted considerable attention within the field, serves to explain much of the recent progress with anion-imprinted polymers. Termed “stoichiometric noncovalent imprinting”, this approach uses carefully selected (often custommade) functional monomers which bind with relatively high affinity to particular motifs on the template molecule [110]. Since the vast majority of

Molecularly Imprinted Polymers Using Anions as Templates

225

functional monomer is associated with the template molecule, it is possible to use an equimolar mixture of template and functional monomer during imprinting. In theory, this not only reduces non-specific interactions but also affords binding sites that are more defined, since the functional monomers are more precisely arranged by the template molecule in a definite spatial orientation within the imprint cavity. Furthermore, the strong binding interactions afforded by these functional monomers means they are often strong enough to work in solvent systems more suited to anionic molecules, as well as apolar systems. Wulff et al. were the first to propose the concept of stoichiometric non-covalent imprinting [111] (this can be thought of as a logical extension to the covalent imprinting strategy, also pioneered by Wulff). They reported the use of custom-made polymerisable amidinium species for targeting oxyanions, in line with the prevalence of guanidinium-based arginine residues in oxyanion binding sites in biological systems. More recently, other authors have promoted functional monomers based upon urea [112, 113], thiourea [114] and guanidinium groups [115, 116]. Regarding the scope of this review, in the following subsections, the use of molecular imprinting for the recognition of selected classes of anions will be discussed. In defining “anions”, we refer not only to small “typical” anions, such as nitrate, phosphate, sulfate, etc., but also to larger, increasingly more complex substrates, which bear an anionic moiety. Moreover, we shall limit our discussion to MIPs in which the anionic part of the template molecule is deliberately targeted or used to advantage. In this way, we disregard any reports of molecular imprinting where the template molecule might happen to feature an anionic component (for example, the C-terminus of peptides) but where this is not deliberately accounted for in the design of the functional binding site. 3.1 Anionic Phosphate Derivatives Phosphate derivatives are ubiquitous in biological systems, rendering them a very attractive target for anion recognition chemistry. For example, many coenzymes and metabolic intermediates are esters of phosphoric and pyrophosphoric acid; the nucleic acids DNA and RNA are based upon a phosphate diester backbone; and phospholipids, the major component of all biological membranes, contain a phosphate diester moiety. Protein phosphorylation is a reversible post-translational modification involving the kinase-catalysed transfer of a phosphate group from ATP to a protein (typically to serine, threonine or tyrosine residues). This highly important process can be thought of as a means of “switching on” proteins, and underlies the regulation of virtually all cellular events. In addition, many man-made molecules of interest are phosphate derivatives. This includes environmental pollutants and degradation products of chemical warfare agents.

226

S.L. Ewen · J.H.G. Steinke

One of the earliest examples of a polymer imprinted with a phosphate derivative came from Turkewitsch et al., who constructed a MIP for the selective recognition of cAMP in aqueous media (Fig. 3) [117]. The customdesigned functional monomer 1 was used to target the nucleotide via both aryl stacking and electrostatic interactions between the positively-charged nitrogen and the phosphate diester anion (Fig. 4). Rather cleverly, this func-

Fig. 3 Schematic representation of cAMP inside MIP, showing possible binding interactions

Fig. 4 Functional monomer custom-designed for cAMP-imprinted polymer

Molecularly Imprinted Polymers Using Anions as Templates

227

tional monomer also served as a transducer element: binding of cAMP effected fluorescence quenching within the imprinted polymer, enabling it to be used as a chemosensor. Further recognition points within the imprinted cavity were afforded by the use of a large excess of HEMA, which was assumed to bind to the purine moiety via a network of H-bonds. The fluorescent nature of the MIP was used to calculate an association constant, Ka , for binding of cAMP by the MIP in aqueous media: this was determined to be in the order of 105 M–1 . With regard to selectivity, the MIP was shown to discriminate between the sodium salts of cAMP and the potentially competitive analogue, cGMP, thus highlighting the role of the adenosine moiety in the recognition process. From the standpoint of our discussion, an investigation into selectivity over other oxyanions would also have been of interest. A steadily increasing number of publications report the use of phosphatederived template molecules to prepare MIPs for solid-phase extraction (SPE). SPE involves the selective pre-concentration of trace-level compounds from complex mixtures. The high selectivities and affinities obtainable by molecular imprinting render these polymers ideal for use as SPE sorbents. For example, Möller et al. prepared a MIP for the selective isolation of diphenyl phosphate (a flame retardant hydrolysis product) from urine [118]. The MIP was prepared using an excess of 2-vinylpyridine as the functional monomer, with ditolyl phosphate as the template molecule (imprinting with a structural analogue of the target molecule eliminates complications caused by possible leaching of the template during subsequent rebinding analysis). In this study, the influence of matrix components normally present in human urine (urea, NaCl) was investigated. While urea did not compete with diphenyl phosphate for the basic binding sites, it was observed that NaCl affected both the recovery and repeatability of diphenyl phosphate binding. This was explained by the assumed complexation between Na+ ions and the anionic analyte, which would render the molecule more neutral and thereby hinder ionic interaction with the polymer. The last decade has seen considerable interest in the application of molecular imprinting to nerve agent detection. The general objective is to develop a sensitive and specific portable sensor device that boasts faster response times than existing nerve agent analysis techniques (gas chromatography, HPLC, immunoassay). In this regard, Jenkins et al. used pinacolyl methylphosphonate (PMP), the hydrolysis product of Soman, as a template in the preparation of an imprinted polymeric luminescent sensor [119, 120]. PMP reversibly bound to ligand-coordinated Eu3+ , which was incorporated into the recognition site, and this binding event corresponded to the appearance of a narrow luminescence band in the 610 nm region of the Eu3+ spectrum. The presence of organophosphorous pesticides and herbicides was determined not to interfere with PMP recognition (although none of the compounds screened bore an anionic moiety).

228

S.L. Ewen · J.H.G. Steinke

Fig. 5 Schematic representation of MPA-imprinted polymer

Zhou et al. developed a potentiometric chemosensor for the detection of methylphosphonic acid (MPA), another degradation product of nerve agents [121]. In the presence of MPA, an octadecylsiloxane (ODS) thin layer was covalently bound to the surface of an indium tin oxide (ITO) transducer (Fig. 5). Subsequent removal of the physisorbed MPA afforded what the authors claim to be imprinted recognition sites, with functional and steric complementarity. It was demonstrated that rebinding of MPA by the “imprinted” film generated a greater potentiometric response than binding of homologous substrates (ethylphosphonic acid, propylphosphonic acid, tert-butylphosphonic acid). However, binding of MPA by the control, non-imprinted film, afforded a greater potentiometric response than binding of MPA homologues by the “imprinted” film. It is thus probable that recognition in this case is governed by a size-exclusion phenomenon. Further experiments should be carried out to eliminate this ambiguity. Prathish et al. also generated a potentiometric sensor for MPA detection, although using an altogether different methodology [122]. MIP particles, prepared via the copolymerisation of EDMA and either methacrylic acid or 4-vinylpyridine in the presence of MPA (which was afterwards removed using 1:1 (v/v) methanol:acetic acid) were dispersed in 2-nitrophenyloctyl ether (plasticizer) and polyvinyl chloride, and this solution was used to cast a membrane of thickness ∼0.45 mm. Potentiometric analysis of ground water samples, which were spiked with MPA and adjusted to pH 10 in the presence of Tris buffer, yielded a greater response by the MIPs than by the corresponding NIPs, thus indicating an imprinting effect. However, the high level of binding by the NIPs suggests that a substantial degree of nonspecific, presumably hydrophobic interactions is involved in, or even dominates, the recognition event. It should also be noted that since the amount of template recovered from the imprinted polymers was not reported, the possibility of residual template “bleeding” from the MIPs and affecting the responses cannot be wholly discounted.

Molecularly Imprinted Polymers Using Anions as Templates

229

Imprinting with phosphorylated/phosphonylated template molecules has been particularly valuable in the synthesis of catalytically-active enzyme mimicking MIPs. In this approach, the tetrahedral geometry of the phosphate/phosphonate ester functional group is likened to the tetrahedral transition state of carbonate or ester hydrolysis reactions and the template molecule is thus referred to as a TSA. The recognition sites in the resulting imprinted polymer are designed to stabilise the high energy transition state of the hydrolysis reaction, thereby leading to enhanced rates of reaction. This strategy, pioneered by Mosbach [83], emulates a method of producing catalytically-active polyclonal antibodies, viz. immunisation with stable TSAs [123]; however, in contrast to polyclonal antibodies, MIPs are considerably more robust and cheaper to produce. The work of Kawanami et al., who produced a MIP to catalyse p-nitrophenyl acetate hydrolysis, serves to illustrate the methodology (Fig. 6) [124]. p-Nitrophenyl phosphate, employed as a template molecule, was observed to form an insoluble complex when mixed in solution with a 10-fold excess of the chosen functional monomer, 1-vinylimidazole; this was then copolymerised with a 20-fold excess of divinylbenzene. Following polymerisation,

Fig. 6 Schematic representation of: a molecular imprinting with template (TSA); b imprinted recognition site following template removal; c,d catalysis

230

S.L. Ewen · J.H.G. Steinke

the template molecule was removed by means of continuous extraction with acetonitrile, phosphate buffer and methanol, affording an insoluble macroporous polymer. MIP-catalysed hydrolysis of p-nitrophenyl acetate was observed to be 85 times faster than uncatalysed hydrolysis. This was two times faster than the catalytic activity of a NIP, thus indicating that catalysis occurs within the imprinted sites. Further evidence for this was provided by the fact that addition of the template molecule, p-nitrophenyl phosphate, clearly inhibits hydrolytic activity. However, the authors conceded that the catalytic performance of the MIP does not yet compare with that of catalytic antibodies for the same hydrolysis reaction. Although no explanation has been given for this shortcoming, the use of an excess of functional monomer (cf. stoichiometric non-covalent imprinting strategy), which would have afforded a heterogeneity of binding sites encompassing a broad spectrum of affinities and selectivities, is likely to have been one contributing factor. Also using a phosphate-derived TSA template, Wulff et al. have demonstrated significant progress in the synthesis of catalytic MIPs modelled on carboxypeptidase A [115]. The active site of this hydrolytic enzyme features two guanidinium groups (arginine residues) and a Zn2+ cation. One of the guanidinium moieties is thought to bind the oxyanion formed during hydrolysis, while the metal ion promotes catalysis via polarisation of the substrate; the second guanidinium and a hydrophobic pocket influence substrate specificity. Following this, the authors prepared a MIP with a phosphate diester template molecule employed as a TSA of carbonate hydrolysis (Fig. 7). A functional monomer was custom-designed, incorporating an amidinium group as an H-bond donor (cf. guanidinium groups) and a triamine to coordinate Zn2+ . This monomer bound to the template molecule through stoichiometric non-covalent interaction.

Fig. 7 Schematic representation of: a molecular imprinting with template (TSA); b catalysis of diphenylcarbonate hydrolysis

Molecularly Imprinted Polymers Using Anions as Templates

231

The catalytic performance of the MIP and appropriate controls were investigated with the hydrolysis of diphenylcarbonate. Reaction kinetics, followed by HPLC analysis of aliquots, were calculated as pseudo first-order rate constants. Rather encouragingly, the imprinted polymer showed typical Michaelis-Menten kinetics, in line with natural enzymes. The catalytic activity of the MIP (expressed as kcat /kuncat , the ratio of turnover number of the catalysed reaction to turnover number in the absence of catalyst) was calculated to be 6900. This is markedly higher than has been reported for catalytic antibodies for carbonate hydrolysis (kcat /kuncat = 810), clearly demonstrating the great potential of MIPs for use as artificial enzymes. All of the imprinted polymers discussed above involve phosphorylated/phosphonylated organic molecules, where the recognition process relies equally upon binding to, and accommodation of, both the organic backbone and the anionic functional group of the template. In fact, we are aware of only one report of a MIP designed to target inorganic phosphate, H2 PO4 – . In this one example, Kugimiya et al. present a series of MIPs prepared using thioureaderived functional monomers [114]. The thiourea group is known to be a good host for anions, particularly for the phosphate anion, to which the N – H moieties bind through a set of coplanar H-bonds. The authors’ initial experiments indicated that 1-allyl-2-thiourea is particularly suitable for use in aqueous media: this was in direct contrast to a related functional monomer, N-methylN  -(4-vinylphenyl)-thiourea, which showed almost no binding ability under the conditions studied. Such results were explained by the greater hydrophilicity of 1-allyl-2-thiourea. Thus a polymer was imprinted with phenylphosphonic acid as the template molecule, using EGDMA as the crosslinker, acetonitrile as the porogen and four equivalents (to template molecule) of 1-allyl-2-thiourea as the functional monomer. Incidentally, Kugimiya et al. omit explaination as to why a phosphonic acid template molecule was chosen for a MIP designed for NaH2 PO4 recognition. Presumably they had anticipated that the difference in geometry between these two functional groups would be negligible. To evaluate the ability of the MIP to selectively bind NaH2 PO4 , MIP and NIP particles were stirred in distilled water with a mixture of various anions (NaH2 PO4 , KNO3 , NaCl, Na2 SO4 , CH3 COONa). After 1 h, the supernatant was removed and analysed by ion chromatography. Under these conditions, both the MIP and the NIP preferentially bound phosphate over the other anions measured, and the MIP bound 58% more template than the NIP, thus demonstrating a clear imprinting effect and selectivity for the target anion. Unfortunately, due to the insufficient level of information provided, it is difficult to fully appreciate the data. For example, while the amount of each anion bound is presented as % binding activity, it is unclear how this relates to the estimated number of available binding sites. Furthermore, it was belatedly discovered that CH3 COO– could not actually be detected via ion chromatography, and the influence of this potentially-competitive anion was therefore not determined.

232

S.L. Ewen · J.H.G. Steinke

3.2 Carboxylates The prevalence of the carboxylate moiety in both biogenic and man-made molecules of interest makes this functional group a popular target for anion host chemistry. Needless to say, carboxylates are a major constituent of proteins, peptides and amino acids, and the expansion of proteomics begets increasing requirements for means of specific detection of such biomolecules. Other relevant examples of carboxylates include fatty acids, while many small molecule di- and tricarboxylates are implicated in key metabolic pathways such as the citric acid cycle (e.g. citrate, succinate, fumarate and malonate). Carboxylated anthropogenic molecules include trichloroacetic acids, anionic surfactants and β-lactam antibiotics. While a small number of early examples of molecular imprinting with carboxylate-bearing templates exist (e.g. Curti et al. imprinting with mandelic acid [45]) in many of these cases the carboxylate moiety was not deliberately accounted for. In 1982, Sarhan and Wulff imprinted with D-glyceric acid, using monomers bearing a boronic acid (for covalent interactions) and an amino functionality [125]. In 1990, Andersson and Mosbach prepared MIPs using N-protected amino acid derivatives as templates with a 4-fold excess of methacrylic acid as the functional monomer, demonstrating that imprinting (and enantiomeric resolution) is possible in the absence of both covalent and ionic interactions (binding was predominantly due to H-bond interactions) [126]. The imprinting of sialic acid reported in 1996 by Kugimiya et al. is worthy of mention as it used a combination of covalent (boronate esterification) and ionic interactions for MIP formation, illustrating that this is possible in aqueous media [127]. This brief synopsis illustrates the initial underrepresentation of carboxylate-targeted MIPs. The situation began to change with a few groups exploring the use of polymerisable nitrogen bases as potential functional monomers for (non-stoichiometric) non-covalent binding of carboxylic acids [66, 128, 129]. Today, a number of groups focus their efforts on constructing MIPs which deliberately target the carboxylate moiety. In particular, various tailor-made functional monomers have been reported, many adapted from naturally-occurring binding motifs, which are designed to bind to the template with affinity sufficient to incorporate them stoichiometrically into the polymer matrix. In one of the earliest examples of stoichiometric non-covalent imprinting, Lübke et al. used a combination of custom-designed functional monomers to imprint with ampicillin, a β-lactam antibiotic which bears both an amine and a carboxylate functional group [69]. Rather elegantly, a polymerisable electron-deficient quinone 2 was used to interact with the amine (via formation of an n–π complex), while the carboxylate was targeted using a bis(boronate-amide)-functionalised monomer 3 (Figs. 8–10). This latter species, adapted from synthetic receptors described by Hughes and

Molecularly Imprinted Polymers Using Anions as Templates

233

Fig. 8 Functional monomer custom-designed to bind the carboxylate of ampicillin

Fig. 9 Functional monomer custom-designed to bind the amine of ampicillin

Smith [130], exploits internal Lewis acid coordination between the boron and the carbonyl oxygen, which serves to polarise the amide, thereby increasing affinity for the oxyanion. Imprinting was carried out by allowing stoichiometric amounts of the monomers and the tetrabutylammonium salt of ampicillin to complex in DMSO over 24 h, prior to incorporation into a highly crosslinked polymer. Subsequent extraction with acetonitrile afforded a 92% recovery of the template. The resulting MIP showed good potential, performing well in aqueous media with optimal binding at pH 8 (where the carboxylate is present as the anion) and with selectivity over structural analogues. Zhang et al. reported the design and four-step synthesis of a novel guanidinium-functionalised monomer which was built upon a vinylanthracene scaffold to effect fluorescence [116]. NMR titration experiments in deuteriomethanol indicated the formation of a 1 : 1 complex with acetate, with estimated association constants of the order of 105 M–1 . Although synthesis of this particular monomer does demand more expertise than monomers typically used for imprinting – with somewhat uninspiring recoveries reported by the authors – it nonetheless appears to be a promising candidate for stoichiometric non-covalent imprinting with carboxylates. Hall et al. are amongst the most prolific authors in the field of carboxylate imprinting. Making use of expertise in the field of supramolecular

234

S.L. Ewen · J.H.G. Steinke

Fig. 10 Schematic representation of ampicillin inside MIP, showing possible binding interactions

chemistry, viz. Hamilton et al. [131, 132], they have developed a series of novel monotopic and ditopic functional monomers based upon a urea core (Fig. 11) [112, 133]. These synthetically accessible host species can usually be prepared in a one-step procedure, either from 1-isopropenyl-4-isopropyl-2isocyanate and the corresponding amine or by reaction of 4-vinylaniline with the corresponding isocyanate. Ureas bind to oxyanions via donation of two H-bonds, and it was shown that varying the substitution of the monomers – and hence modifying the acidity of the urea core – can result in quite dramatic changes in affinity for the model substrate, tetrabutylammonium benzoate. For example, NMR titration experiments in DMSO-d6 afforded an association constant of 1322 ± 48 M–1 for 4 where R1 = R2 = H, while the corresponding value for 4 where R1 = R2 = CF3 was 8820 ± 1600 M–1 . The authors have explored a number of potential applications for these urea-based monomers. These include a class-selective MIP for β-lactam antibiotics, for which it was hypothesised that the mono-urea functionalised binding site directly targets the β-lactam carboxylate moiety while providing sufficient

Molecularly Imprinted Polymers Using Anions as Templates

235

Fig. 11 Functional monomers based upon urea core, designed by Hall et al.

space to accommodate pendant substituents [88, 134]. In another example, a nitrophenyl-functionalised analogue was used to prepare a chromogenic MIP for the enantioselective recognition of Z-glutamate and the related drug, methotrexate; this monolithic polymer exhibited a change in colour intensity, visible to the naked eye, upon binding the target analytes in basic, aqueous media [135]. Along similar lines, Schmitzer and Gomy have used NMR titrations, isothermal titration calorimetry (ITC) and molecular modelling to design, synthesise and screen a library of functional monomers that are also based upon the urea motif [113]. Out of 16 candidates, the two monomers 5 and 6 (Fig. 12) demonstrated particularly high affinity for the model substrate, bis(tetrabutylammonium)-N-Z-L-glutamate. Association constants between this latter species and both 5 and 6 were calculated to be 3370 ± 260 M–1 and 4000 ± 400 M–1 respectively (in DMSO-d6 ). These monomers should be promising receptors for future stoichiometric incorporation into an imprinted polymer. While most of the aforementioned MIPs were constructed using functionalised, highly crosslinked organic polymers, other workers have demonstrated that inorganic matrices, assembled by a surface sol–gel technique, are also valid candidates for anion receptor chemistry [136, 137]. With this

236

S.L. Ewen · J.H.G. Steinke

Fig. 12 Functional monomers based upon a urea core, designed by Schmitzer and Gomy

approach, molecularly imprinted ultrathin TiO2 films were prepared by first allowing titanium butoxide to covalently complex (as confirmed by FT-IR measurements) with a carboxylated template molecule in a mixture of toluene and ethanol (Fig. 13). Water was subsequently added, and the complex was allowed to chemisorb onto a hydroxylated solid substrate, such as the gate surface of an ion-sensitive field effect transistor (ISFET) device [138]. Follow-

Fig. 13 Hydrolysis of Ti(IV)-carboxylate template and subsequent rebinding

Molecularly Imprinted Polymers Using Anions as Templates

237

ing complete hydrolysis, aqueous ammonia was used to remove the template, affording a sol–gel film with imprinted recognition sites. It was suggested that rebinding of the template molecule occurs via a combination of covalent bonding, H-bonding, metal coordination, and hydrophobic interactions with Ti–O moieties. Such molecularly imprinted films are particularly suited to use as transducer elements in chemosensor devices. This method was demonstrated to afford molecularly imprinted TiO2 films selective for azobenzene carboxylic acids [136, 139], chloroaromatic acids [138], chiral carboxylic acids including amino acid derivatives [140, 141] and anthracenecarboxylic acids [139]. The methodology was also adapted for use with water-soluble di- and tri-peptides [142]. In this latter case, even though NaOH solution was used as an alternative to aqueous ammonia, complete template removal proved impracticable, with 30–40% of the template trapped in the film: nonetheless, the TiO2 films were said to show reproducible binding for guest molecules. More recently, TiO2 films were also used for imprinting thiolate- and phosphonate-functionalised templates [143]. The inherent simplicity of metal oxide sol–gel films, in particular the lack of requirement for carefully chosen functional monomers, makes this approach an attractive alternative to the use of highly crosslinked organic polymers. The technique is optimised for – indeed, necessitates – an anionic functional group on the template, and results in imprinted films with a notable degree of regioselectivity, structural selectivity and enantioselectivity. To date no studies have been published demonstrating the degree of selectivity between substrates with differing anionic functional groups. Finally, we wish to mention an interesting and somewhat more unusual example of molecular imprinting, albeit it is debatable as to whether this particular application should rightly be included within a discussion on anionic templates. D’Souza et al. copolymerised 6-methacrylamidohexanoic acid and divinylbenzene in the presence of calcite (calcium carbonate polymorph) crystals and, after exhaustive removal of the template, showed that the imprinted polymer promoted nucleation of calcite from an aqueous supersaturated calcium carbonate solution [144, 145]. Using several methods of analysis, it was demonstrated that no template remained in the MIP after the wash steps; subsequent crystal-formation was therefore attributed to heterogeneous nucleation as opposed to seeding by residual crystals. Regarding the functional monomer used, though the authors do not explicitly state their reasoning behind choosing 6-methacrylamidohexanoic acid, they report that the use of other functional monomers (acrylic or methacrylic acids) in place of 6-methacrylamidohexanoic acid afforded MIPs with inferior nucleating abilities. It was suggested that the presence of the flexible spacer allowed the acidic head group of the monomer to more easily match the spacing of ions on the surface of the crystal. More recently, Egan et al. applied this same methodology to imprinting with calcium oxalate crystals, using the resulting MIP to model nucleation of renal calculi from artificial and real urine [146].

238

S.L. Ewen · J.H.G. Steinke

3.3 Anionic Sulfate Derivatives While perhaps not as ubiquitous as phosphate derivatives and carboxylates, sulfate derivatives are nonetheless a frequently encountered class of oxyanions and are therefore another attractive target for anion receptor chemistry. Important examples of such molecules include sulfated sugars and anionic surfactants. Intriguingly, there are actually relatively few accounts of imprinting with sulfates/sulfonates and, with the exception of aforementioned historical reports where the anionic component was either not directly targeted [45] or found to be unnecessary for the recognition event [107], most examples come from the last 3–4 years. That said, looking at the binding motifs that target sulfate derivatives in biological systems and in small molecule receptor chemistry, it is apparent that many of the functional monomers designed for phosphate and carboxylate recognition could eventually be adapted for use with sulfates/sulfonates. In 2004, Caro et al. imprinted a polymer with 1-naphthalene sulfonic acid, using a 4-fold excess of 4-vinylpyridine as functional monomer (cf. stoichiometric non-covalent imprinting), EGDMA as crosslinker and AIBN as free radical initiator [147]. A 4 : 1 v/v mixture of methanol and water was used as the porogen, due to the high polarity of the template, and it was assumed that molecular recognition would occur via a combination of hydrophobic effect and ionic interactions. Following extraction of the template, the polymer particles were packed into a polyethylene cartridge, and a mixture of different polar compounds (including eight naphthalene sulfonates) was percolated through at pH 2.3. All compounds, except oxamyl and methomyl, were retained on the MIP in the loading step. The MIP was then washed with MeOH to remove non-selectively bound species. Somewhat unexpectedly, all eight of the naphthalene sulfonates were retained by the imprinted polymer in spite of the wash step; these included various mono- and disulfonated species, some of which were hydroxyl- or amine-substituted, as well as the original template, 1-naphthalene sulfonic acid. However, all other polar compounds (phenol, nitrophenols, bentazone and 4-chloro-2-methylphenoxy acetic acid) were eluted. The explanation offered by the authors is that the MIP was cross-reactive for naphthalene sulfonates. However, at pH 2.3, only the naphthalene sulfonates were anionic (pKa 0.5–0.6). It is quite plausible that in this example, molecular recognition was predominantly an anion-exchange mechanism, and that this overshadowed the influence of any molecularly imprinted cavities. In this way, the MeOH wash step could be seen to have disrupted non-covalent, non-ionic interactions, leaving behind the anionic naphthalene sulfonates, which were retained much more strongly via ionic interactions with the protonated pyridyl moieties. Unfortunately, this is always a poten-

Molecularly Imprinted Polymers Using Anions as Templates

239

tial hazard with constructing such hosts: there is often a fine balance between obtaining high affinity binding interactions while still retaining the selectivity afforded by the act of imprinting. The recent work of Albano et al. illustrates a particularly innovative application of molecular imprinting [148]. In the presence of sodium dodecyl sulfate (SDS), a thin layer of polypyrrole was electrodeposited onto the surface of a quartz crystal. Removal of the SDS by washing with water afforded a molecularly imprinted sensor element which could be used to monitor levels of pollutant surfactants in river water. Thus, the piezoelectric quartz sensor was contained within a flow cell so that the MIP-coated side of the crystal remained in contact with analyte solution while the electrodes of the sensor were connected to an oscillator circuit. A significant drop in the quartz crystal oscillation frequency occurred when the sensor was in contact with SDS. This is consistent with an increase in mass on the surface of the sensor, suggesting that SDS molecules had bound to the layer of imprinted polypyrrole. A reference polypyrrole-coated sensor, prepared in parallel to the MIP-sensor but in the absence of SDS, showed only a minimal response to the target analyte. The influence of pH on sensor performance was investigated: MIP-coated sensors exhibiting the highest response and sensitivity were those buffered at pH 9 during the polymerisation step; likewise, an alkaline measurement solution (pH 8) was found to afford the best results. It is evident that the anionic form of SDS, rather than the free acid, is required for optimal binding to the pyrrole moieties. With regard to selectivity, the sensor was also applied to solutions of both sodium dodecanoate (structurally identical to SDS except bearing a carboxylate group in place of the sulfate ester) and sodium dodecylbenzenesulfonate. While the sensor showed low sensitivity to the carboxylate-functionalised analyte, a response equal to that afforded by SDS was attained with the benzenesulfonate analogue. Nonetheless, the degree to which molecular recognition by the MIP was governed by ionic strength, as opposed to a definite molecular imprinting effect, is as yet unclear. Recently, Sineriz et al. reported the first example of a MIP imprinted with a sulfated sugar [149]. Sulfated sugars, such as heparan sulfate and chondroitin sulfate, are a class of highly abundant biological molecules with important intercellular roles. Their activities are directly related to patterns and degree of sulfation. Understanding their biological functions therefore necessitates structure elucidation, yet antibodies developed for this purpose have generally been found to be insufficiently selective. In an initial study, various amine-functionalised monomers (Fig. 14) were incorporated into polymers templated around glucose-6-O-sulfate. Relative binding strengths of the resulting MIPs were determined by equilibration with glucose-6-O-sulfate solution followed by HPLC analysis of the supernatant. In DMSO, it was found that the MIP made using quaternary aminefunctionalised monomer 7 exhibited high affinity towards the analyte but no selectivity (the MIP and corresponding NIP performing equally). In contrast,

240

S.L. Ewen · J.H.G. Steinke

Fig. 14 Functional monomers screened by Sineriz et al.

the MIP made using a primary amine (8) was observed to bind up to 80% of a given concentration of glucose-6-O-sulfate in DMSO, while the corresponding NIP showed negligible binding. Consistent with the aforementioned work of Simon and Spivak [109], it was apparent that the directional nature of Hbonds from the primary amine, coupled with the inherent pre-organisation of the imprinted binding site, were crucial for selectivity. Further experiments showed that the MIP prepared using the primary amine-functionalised monomer preferentially bound glucose-6-O-sulfate over other sulfated saccharides (galactose-6-sulfate, glucose-3-sulfate and N-acetyl-glucosamine-6sulfate). The degree of selectivity over other anionic functional groups was not investigated. 3.4 Other Anions While relatively few MIPs have been made to recognise the oxyanionic functional groups phosphate/phosphonate, carboxylate and sulfate/sulfonate, MIPs designed to target inorganic anions are even more scarce. The following paragraphs list examples of this mode of imprinting. Fujiwara and co-workers attempted to surface imprint pyridine-bearing microspheres using tetravalent ferrocyanide anions as the template [150]. Adsorption studies indicated that, while the imprinted microspheres did show greater affinity for the template anion in comparison to non-imprinted microspheres, they lacked selectivity and bound all other polyvalent anions tested. This suggests that, although the pyridine groups were apparently preorganised by the imprinting process, negligible selectivity for the target anions may be due to the absence of imprinted cavities on the surface of the microspheres. In 1988, Dong et al. reported a chloride-ion selective electrode, prepared via electropolymerisation of pyrrole in the presence of LiCl [151]. The resultant polymer film was observed to give a Nernstian response to chloride, with a limit of detection of 3.5 × 10–5 M (said to be comparable to that of most other contemporary chloride-selective electrodes). Later, the same methodology was adapted to produce a chloride-selective microsensor, used to detect chloride in serum [152]. Both the ion-selective electrode and the microsensor

Molecularly Imprinted Polymers Using Anions as Templates

241

showed poor selectivity for chloride over other anions. For example, potential selectivity coefficients of the electrode were of the order 10–1 for other halides, IO4 – and ClO4 – , and >1 for NO3 – , HCOO– [151]. It was postulated that selectivity was related to ionic radius. A rare example of electropolymerisation employed for the formation of an anion selective crosslinked polymer matrix was introduced by Kamata et al. [153, 154]. Reductive coupling between trifunctional p-cyanopyridinium crosslinkers (Fig. 15) in water containing the required counteranion led to stable polymer networks which showed thermodynamic and kinetic anion selectivity. A size exclusion effect was observed whereby only the imprint halogen (and any smaller halogen anion) was electrochemically recognised. Diffusion of the counterion was reduced in the case of the same template anion MIP, but increased with the size of the templating anion. Experiments clearly demonstrating anion selectivity have yet to be carried out.

Fig. 15 Trifunctional p-cyanopyridinium crosslinker used to prepare halide-imprinted polymer

In the mid-1990s, the molecular imprinting principle was adopted for the preparation of a nitrate-selective electrode [155]. Ion-selective electrodes, i.e. sensors which convert the activity of a specific ion into an electrical potential, have widespread application in biochemical and biophysical analysis; however, their major limitation is poor selectivity leading to interference from other ions. Hutchins and Bachas electropolymerised pyrrole onto a glass-carbon electrode in the presence of an aqueous solution of NaNO3 . The polymer-coated electrode was demonstrated to detect aqueous nitrate, giving a near-Nernstian response, with response times ranging from 24 to <6 s. Recognition appeared to be based on size-exclusion phenomena. Therefore, while the polymer was able to discriminate over traditional interferents whose radii is larger than that of NO3 – , such as ClO4 – and I– , the much smaller SCN– still effected a response. It was hypothesised that the hydrophobicity of the polypyrrole films induced the superimposition of a Hofmeistertype selectivity (i.e. based upon lipophilicity) upon the nitrate imprinting selectivity. Thus lipophilic anions not large enough to be sterically hindered

242

S.L. Ewen · J.H.G. Steinke

from the film, such as thiocyanate, still interfered with nitrate recognition. Much more recently, this technology has been translated into the production of nanolitre-scale electrochemical cells [156]. 3.5 Concluding Comments Tailoring molecular imprinting for anion recognition is a recent development in a relatively young art. To date, the large majority of anion-imprinted polymers have been designed to recognise organic molecules which bear one or more oxyanionic functional group, and we see few examples where the target analyte is a simple, inorganic anion. A plausible explanation for this bias may be that there is simply less demand for such receptors: after all, simple anions can be resolved with relative ease using ion chromatography. Yet this contrasts sharply with the rest of the synthetic receptor scene, where the focus is very much upon developing host molecules for anions such as phosphate, sulfate, acetate and the halides [10, 157, 158]. As an alternative explanation, it has been suggested that small and simple species are much poorer “imprintogens” than larger, multi-functionalised molecules [159], and that it is generally quite difficult to create a well-defined recognition site when less chemical information is available during the imprinting step. For example, Li and Wu reported an inability to imprint with formate, acetate and propionate [160]. Following computer simulation studies, they suggested that the crosslinker (EDMA) had been unable to form adequate cavities for such small, noncomplex molecules. The extent to which this applies to the use of other crosslinkers has not been investigated. It is apparent that the current emphasis is towards MIPs that can discriminate by both the anionic moiety and the organic residue of a substrate. Progress in this direction is likely to be particularly beneficial to the recent efforts to imprint with proteins [20, 161], or indeed to any application where the target species is adorned with multiple anionic functionalities. To separately target each functional group on a complex template molecule should naturally give rise to better defined binding sites with greater affinity and selectivity for the target species. Equally beneficial will be progress towards compatibility with aqueous media and pH independent binding. An interesting topic of discussion concerns the different levels of selectivity required by anion-recognition MIPs; such requirements are dictated by the intended final application of the imprinted polymer. To date, most MIPs have been designed to be specific for one particular molecule, and in this regard the aim has been to optimise imprinting technology so as to make binding sites as uniform and “monoclonal” as possible. Still, there are a number of instances where imprinted polymers benefit from a degree of cross-selectivity. This is illustrated by the aforementioned MIP from Hall et al., which recognises compounds bearing the anionic structural motif particular to β-lactam

Molecularly Imprinted Polymers Using Anions as Templates

243

antibiotics [134]. However, while recognition of a particular (anionic) substructure is one notion of cross-selectivity, a possible extension of this would be to develop imprinted polymers that are specific for one particular (anionic) functional group, irrespective of the type, class, substructure or size of the rest of the molecule (Fig. 16). In a way, this could be thought of as “turning the technology on its head”. Whereas with classic molecular imprinting the intention is to eradicate all substrate promiscuity, this novel mode of imprinting would afford binding sites that selectively recognise a particular functional group present on the substrate, while discrimination by compound type, class, substructure or size is deliberately suppressed. One might theorise that such “functional group imprinted polymers” (“FIPs”) could be prepared using an adaptation of the “epitope approach” to imprinting, whereby a substructure of the target molecule(s) – in this case the functional group – is used to imprint. Certainly the very act of imprinting, where functional binding sites are pre-organised into an exact spatial and geometric arrangement, lends itself (conceptually, at least) to discrimination between even near-isosteric functional groups. While molecular imprinting, both in general and with regard to anionic templates, has seen significant development since the first, exploratory experiments of Polyakov and Dickey, there is still vast room for improvement. The technology will continue to benefit from concomitant advances in a breadth of disciplines, from supramolecular chemistry to polymer science to organic synthesis. And yet, the possibilities of molecular imprinting are manifest. We hope that this review will serve to promote the potential of

Fig. 16 Pictorial representation of functional group imprinted polymer (FIP) concept

244

S.L. Ewen · J.H.G. Steinke

MIPs, as an extremely versatile class of potent synthetic receptors, where the benefits for anion recognitions are beginning to emerge. Acknowledgements Support for S.L.E. through a Chemical Biology Centre studentship at Imperial College London is gratefully acknowledged.

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.

Mullis KB (1994) Angew Chem Int Edit 33:1209 Cheng YJ, Gutmann JS (2006) J Am Chem Soc 128:4658 Smarsly B, Antonietti M (2006) Eur J Inorg Chem: 1111 Antonietti M, Breulmann M, Goltner CG, Colfen H, Wong KKW, Walsh D, Mann S (1998) Chem Eur J 4:2493 Shibaev V, Bobrovsky A, Boiko N (2003) Prog Polym Sci 28:729 Conn MM, Rebek J (1997) Chem Rev 97:1647 Tanev PT, Pinnavaia TJ (1995) Science 267:865 Blondeau P, Segura M, Perez-Fernandez R, de Mendoza J (2007) Chem Soc Rev 36:198 Hay BP, Firman TK, Moyer BA (2005) J Am Chem Soc 127:1810 Kubik S, Reyheller C, Stuwe S (2005) J Incl Phenom Macrocycl Chem 52:137 Steinke J, Sherrington DC, Dunkin IR (1995) Adv Polym Sci 123:81 Wulff G (1995) Angew Chem Int Edit 34:1812 Zhang HQ, Ye L, Mosbach K (2006) J Mol Recognit 19:248 Alexander C, Andersson HS, Andersson LI, Ansell RJ, Kirsch N, Nicholls IA, O’Mahony J, Whitcombe MJ (2006) J Mol Recognit 19:106 Andersson HS, Nicholls IA (1997) Bioorganic Chem 25:203 Wulff G (2002) Chem Rev 102:1 Maier WF, Benmustapha W (1997) Catal Lett 46:137 Davis ME, Katz A, Ahmad WR (1996) Chem Mat 8:1820 Liu ZS, Zheng C, Yan C, Ga RY (2007) Electrophoresis 28:127 Bossi A, Bonini F, Turner APF, Piletsky SA (2007) Biosens Bioelectron 22:1131 Qiao FX, Sun HW, Yan HY, Row KH (2006) Chromatographia 64:625 Dirion B, Cobb Z, Schillinger E, Andersson LI, Sellergren B (2003) J Am Chem Soc 125:15101 Lanza F, Sellergren B (2004) Macromol Rapid Commun 25:59 Martin-Esteban A, Tadeo JL (2006) Comb Chem High Throughput Screen 9:747 Southard GE, Van Houten KA, Ott EW, Murray GM (2007) Macromolecules 581:202 Rachkov A, McNiven S, El’skaya A, Yano K, Karube I (2000) Anal Chim Acta 405:23 Patel A, Fouace S, Steinke JHG (2004) Anal Chim Acta 504:53 Karim K, Breton F, Rouillon R, Piletska EV, Guerreiro A, Chianella I, Piletsky SA (2005) Adv Drug Deliv Rev 57:1795 Wei ST, Jakusch M, Mizaikoff B (2006) Anal Chim Acta 578:50 Zimmerman SC, Wendland MS, Rakow NA, Zharov I, Suslick KS (2002) Nature 418:399 Wulff G, Chong BO, Kolb U (2006) Angew Chem Int Edit 45:2955 Wang XJ, Xu ZL, Yang ZG, Bing NC (2007) Prog Chem 19:805 Batra D, Shea KJ (2003) Curr Opin Chem Biol 7:434 Piletsky SA, Turner NW, Laitenberger P (2006) Med Eng Phys 28:971 Caro E, Marce RM, Borrull F, Cormack PAG, Sherrington DC (2006) Trends Anal Chem 25:143

Molecularly Imprinted Polymers Using Anions as Templates

245

36. Sellergren B, Allender CJ (2005) Adv Drug Deliv Rev 57:1733 37. Theodoridis GA, Papadoyannis LN (2006) Curr Pharm Anal 2:385 38. Voicu R, Faid K, Farah AA, Bensebaa F, Barjovanu R, Py C, Tao Y (2007) Langmuir 23:5452 39. Belmont AS, Sokuler M, Haupt K, Gheber LA (2007) Appl Phys Lett 90:3 40. Berglund J, Nicholls IA, Lindbladh C, Mosbach K (1996) Bioorg Med Chem Lett 6:2237 41. Breinl F, Haurowitz F (1930) Z Physiol Chem 192:45 42. Mudd S (1932) J Immunol 23:423 43. Pauling L (1940) J Am Chem Soc 62:2643 44. Dickey FH (1949) Proc Nat Acad Sci USA 35:227 45. Curti R, Colombo U (1952) J Amer Chem Soc 74:3961 46. Curti R, Colombo U, Clerici F (1952) Gazz Chim Ital 82:491 47. Patrikeev VV, Balandin AA, Klabunovskii EI, Mardashev YS, Maksimova GI (1960) Dokl Akad Nauk SSSR 132:850 48. Patrikeev VV, Smirnova ZS, Maksimova GI (1962) Dokl Akad Nauk SSSR 146:707 49. Patrikeev VV, Kozarenko TD, Balandin AA (1962) Izvest Akad Nauk SSSR, Ser Khimi, p 170 50. Wulff G, Sarhan A (1972) Angew Chem Int Edit 11:341 51. Takagishi T, Klotz IM (1972) Biopolymers 11:483 52. Wulff G, Sarhan A, Zabrocki K (1973) Tetrahedron Lett, p 4329 53. Wulff G, Sarhan A, Gimpel J, Lohmar E (1974) Chem Ber 107:3364 54. Sagiv J (1979) Isr J Chem 18:346 55. Rubinstein I, Steinberg S, Tor Y, Shanzer A, Sagiv J (1988) Nature 332:426 56. Arshady R, Mosbach K (1981) Macromol Chem Phys 182:687 57. Andersson L, Sellergren B, Mosbach K (1984) Tetrahedron Lett 25:5211 58. Kugimiya A, Matsui J, Takeuchi T (1997) Mater Sci Eng C 4:263 59. Piletsky SA, Piletskaya EV, Panasyuk TL, El’skaya AV, Levi R, Karube I, Wulff G (1998) Macromolecules 31:2137 60. Wulff G, Sarhan A, Gimpel J, Lohmar E (1974) Chem Ber 107:3364 61. Lauer M, Boehnke H, Grotstollen R, Salehnia M, Wulff G (1985) Chem Ber 118:246 62. Sellergren B, Andersson L (1990) J Org Chem 55:3381 63. Whitcombe MJ, Rodriguez ME, Vulfson EN (1994) Sep Biotech III 158:565 64. Whitcombe MJ, Rodriguez ME, Villar P, Vulfson EN (1995) J Am Chem Soc 117:7105 65. Wulff G, Knorr K (2001) Bioseparations 10:257 66. Kempe M, Fischer L, Mosbach K (1993) J Mol Recogn 6:25 67. Wulff G, Gross T, Schonfeld R (1997) Angew Chem Int Edit 36:1962 68. Yano K, Nakagiri T, Takeuchi T, Matsui J, Ikebukuro K, Karube I (1997) Anal Chim Acta 357:91 69. Luebke C, Luebke M, Whitcombe MJ, Vulfson EN (2000) Macromolecules 33:5098 70. Sellergren B (1994) Anal Chem 66:1578 71. Sreenivasan K, Sivakumar R (1999) J Appl Polym Sci 71:1823 72. Rachkov A, Minoura N (2000) J Chromatogr A 889:111 73. Quaglia M, Chenon K, Hall AJ, De Lorenzi E, Sellergren B (2001) J Am Chem Soc 123:2146 74. Mukawa T, Goto T, Takeuchi T (2002) Analyst 127:1407 75. Shea KJ, Thompson E (1978) J Org Chem 43:4253 76. Damen J, Neckers DC (1980) J Am Chem Soc 102:3265 77. Belokon YN, Tararov VI, Savel’eva TF, Vitt SV, Bakhmutov VI, Belikov VM (1980) Makromol Chem 181:89

246

S.L. Ewen · J.H.G. Steinke

78. 79. 80. 81. 82. 83. 84. 85. 86. 87.

Sarhan A, El-Zahab MA (1987) Makromol Chem Rapid Commun 8:555 Andersson LI, Mosbach K (1989) Makromol Chem Rapid Commun 10:491 Wulff G, Vietmeier J (1989) Makromol Chem 190:1727 Wulff G (2002) Nanoporous Mater III 141:35 Leonhardt A, Mosbach K (1987) React Polym 6:285 Robinson DK, Mosbach K (1989) J Chem Soc Chem Commun, p 969 Owens PK, Karlsson L, Lutz ESM, Andersson LI (1999) Trends Anal Chem 18:146 Turiel E, Tadeo JL, Martin-Esteban A (2007) Anal Chem 79:3099 Mayes AG, Whitcombe MJ (2005) Adv Drug Deliv Rev 57:1742 Caro E, Marce RM, Cormack PAG, Sherrington DC, Borrull F (2005) Anal Chim Acta 552:81 Urraca JL, Moreno-Bondi MAC, Hall AJ, Sellergren B (2007) Anal Chem 79:695 Toth B, Pap T, Horvath V, Horvai G (2007) Anal Chim Acta 591:17 Liu HY, Row KH, Yan GL (2005) Chromatographia 61:429 Lanza F, Hall AJ, Sellergren B, Bereczki A, Horvai G, Bayoudh S, Cormack PAG, Sherrington DC (2001) Anal Chim Acta 435:91 Yoshimatsu K, Reimhult K, Krozer A, Mosbach K, Sode K, Ye L (2007) Anal Chim Acta 584:112 Strikovsky AG, Kasper D, Grun M, Green BS, Hradil J, Wulff G (2000) J Am Chem Soc 122:6295 Schmidt RH, Mosbach K, Haupt K (2004) Adv Mater 16:719 Ulbricht M (2004) J Chrom B 804:113 Siemoneit U, Schmitt C, Alvarez-Lorenzo C, Luzardo A, Otero-Espinar F, Concheiro A, Blanco-Mendez J (2006) Int J Pharm 312:66 Chronakis IS, Milosevic B, Frenot A, Ye L (2006) Macromolecules 39:357 Svenson J, Ning Z, Fohrman U, Nicholls IA (2005) Anal Lett 38:57 Liu Y, Wang F, Tan TW, Lei M (2007) Anal Chim Acta 581:137 Dineiro Y, Menendez MI, Blanco-Lopez MC, Lobo-Castanon MJ, Miranda-Ordieres AJ, Tunon-Blanco P (2006) Biosens Bioelectron 22:364 Nantasenamat C, Naenna T, Ayudhya CIN, Prachayasittikul V (2005) J Comput Aided Mol Des 19:509 Dong WG, Yan M, Zhang ML, Liu Z, Li YM (2005) Anal Chim Acta 542:186 Vaughan AD, Sizemore SP, Byrne ME (2007) Polymer 48:74 Buchmeiser MR (2001) J Chrom A 918:233 Enholm EJ, Allais F, Martin RT, Mohamed R (2006) Macromolecules 39:7859 Corbett PT, Leclaire J, Vial L, West KR, Wietor JL, Sanders JKM, Otto S (2006) Chem Rev 106:3652 Bernhard SA (1952) J Amer Chem Soc 74:4946 Rajkumar R, Warsinke A, Moehwald H, Scheller FW, Katterle M (2007) Biosens Bioelectr 22:3318 Simon RL, Spivak DA (2004) J Chrom B 804:203 Wulff G, Knorr K (2002) Bioseparation 10:257 Wulff G, Schonfeld R (1998) Adv Mater 10:957 Hall AJ, Manesiotis P, Emgenbroich M, Quaglia M, De Lorenzi E, Sellergren B (2005) J Org Chem 70:1732 Gomy C, Schmitzer AR (2006) J Org Chem 71:3121 Kugimiya A, Takei H (2006) Anal Chim Acta 564:179 Liu JQ, Wulff G (2004) Angew Chem Int Edit 43:1287 Zhang H, Verboom W, Reinhoudt DN (2001) Tetrahedron Lett 42:4413 Turkewitsch P, Wandelt B, Darling GD, Powell WS (1998) Anal Chem 70:2025

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.

Molecularly Imprinted Polymers Using Anions as Templates 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.

247

Moller K, Nilsson U, Crescenzi C (2004) J Chromatogr B 811:171 Jenkins AL, Uy OM, Murray GM (1997) Anal Commun 34:221 Jenkins AL, Uy OM, Murray GM (1999) Anal Chem 71:373 Zhou YX, Yu B, Shiu E, Levon K (2004) Anal Chem 76:2689 Prathish KP, Prasad K, Rao TP, Suryanarayana MVS (2007) Talanta 71:1976 Lerner RA, Benkovic SJ, Schultz PG (1991) Science 252:659 Kawanami Y, Yunoki T, Nakamura A, Fujii K, Umano K, Yamauchi H, Masuda K (1999) J Mol Catal A 145:107 Sarhan A, Wulff G (1982) Makromol Chem 183:85 Andersson LI, Mosbach K (1990) J Chrom 516:313 Kugimiya A, Takeuchi T, Matsui J, Ikebukuro K, Yano K, Karube I (1996) Anal Lett 29:1099 Steinke JHG, Dunkin IR, Sherrington DC (1999) Trends Anal Chem 18:159 Spivak D, Shea KJ (1999) J Org Chem 64:4627 Hughes MP, Smith BD (1997) J Org Chem 62:4492 Fan E, Van Arman SA, Kincaid S, Hamilton AD (1993) J Am Chem Soc 115:369 Linton BR, Goodman MS, Fan E, Van Arman SA, Hamilton AD (2001) J Org Chem 66:7313 Hall AJ, Achilli L, Manesiotis P, Quaglia M, De Lorenzi E, Sellergren B (2003) J Org Chem 68:9132 Urraca JL, Hall AJ, Moreno-Bondi MC, Sellergren B (2006) Angew Chem Int Edit 45:5158 Manesiotis P, Hall AJ, Emgenbroich M, Quaglia M, De Lorenzi E, Sellergren B (2004) Chem Commun, p 2278 Lee SW, Ichinose I, Kunitake T (1998) Langmuir 14:2857 Kunitake T, Lee SW (2004) Anal Chim Acta 504:1 Lahav M, Kharitonov AB, Katz O, Kunitake T, Willner I (2001) Anal Chem 73:720 Lee SW, Yang DH, Kunitake T (2005) Sens Actuators B 104:35 Lahav M, Kharitonov AB, Willner I (2001) Chem Eur J 7:3992 Lee SW, Ichinose I, Kunitake T (2002) Chem Lett, p 678 Ichinose I, Kikuchi T, Lee SW, Kunitake T (2002) Chem Lett, p 104 Pogorelova SP, Kharitonov AB, Willner I, Sukenik CN, Pizem H, Bayer T (2004) Anal Chim Acta 504:113 D’Souza SM, Alexander C, Carr SW, Waller AM, Whitcombe MJ, Vulfson EN (1999) Nature 398:312 D’Souza SM, Alexander C, Whitcombe MJ, Waller AM, Vulfson EN (2001) Polym Int 50:429 Egan TJ, Rodgers AL, Siele T (2004) J Biol Inorg Chem 9:195 Caro E, Marce RM, Cormack PAG, Sherrington DC, Borrull F (2004) J Chrom A 1047:175 Albano DR, Sevilla F (2007) Sens Actuators B 121:129 Sineriz F, Ikeda Y, Petit E, Bultel L, Haupt K, Kovensky J, Papy-Garcia D (2007) Tetrahedron 63:1857 Fujiwara I, Maeda M, Takagi M (2003) Anal Sci 19:617 Dong S, Sun Z, Lu Z (1988) Analyst 113:1525 Dong SJ, Che GL (1991) Talanta 38:111 Kamata K, Suzuki T, Kawai T, Iyoda T (1999) J Electroanal Chem 473:145 Kamata K, Kawai T, Iyoda T (2001) Langmuir 17:155 Hutchins RS, Bachas LG (1995) Anal Chem 67:1654 Lenihan JS, Ball JC, Gavalas VG, Lumpp JK, Hines J, Daunert S, Bachas LG (2007) Anal Bioanal Chem 387:259

248

S.L. Ewen · J.H.G. Steinke

157. 158. 159. 160. 161.

Schmidtchen FP (2005) Anion Sensing (Topics in Current Chemistry) 255:1 Katayev EA, Ustynyuk YA, Sessler JL (2006) Coord Chem Rev 250:3004 Sun BW, Yang ML, Li YZ, Chang WB (2003) Acta Chim Sin 61:878 Wu LQ, Li YZ (2004) Anal Chim Acta 517:145 Turner NW, Jeans CW, Brain KR, Allender CJ, Hlady V, Britt DW (2006) Biotech Prog 22:1474 Tanabe K, Takeuchi T, Matsui J, Ikebukuro K, Yano K, Karube I (1995) J Chem Soc Chem Commun, p 2303 Kubo H, Yoshioka N, Takeuchi T (2005) Org Lett 7:359 Kempe M, Mosbach K (1994) J Chrom A 664:276 Caro E, Marce RM, Cormack PAG, Sherrington DC, Borrull F (2004) J Chrom B 813:137 Haginaka J, Sanbe H (2001) J Chrom A 913:141 Andersson LI (1996) Anal Chem 68:111 Oral E, Peppas NA (2006) J Biomed Mater Res A 78A:205 Seong H, Lee HB, Park K (2002) J Biomater Sci – Polym Edit 13:637 Asanuma H, Kakazu M, Shibata M, Hishiya T, Komiyama M (1997) Chem Commun, p 1971 Hwang CC, Lee WC (2002) J Chrom A 962:69 Matsui J, Takeuchi T (1997) Anal Commun 34:199 Zander A, Findlay P, Renner T, Sellergren B, Swietlow A (1998) Anal Chem 70:3304 Sanbe H, Hosoya K, Haginaka J (2003) Anal Sci 19:715

162. 163. 164. 165. 166. 167. 168. 169. 170. 171. 172. 173. 174.

Subject Index

Adenosine triphosphate (ATP) 96 Alkaline phosphatase 102 Aluminophosphates 214 Amides 2 Amidinothiourea 183 Amidoferrocenylalkylthiol (AFAT)-gold 64 7-Amino-4-trifluorocumarin 111 6-Aminopicolinic acid 11 Ammonium-containing receptors 39 Ampicillin 232 Anionic templates 175, 207 Anion-π interactions 127, 152 Anthracene 13, 99 Anthracenecarboxylic acids 237 Anticrown chemistry 130 Aryl bisamide 77 ATPase 96 Azadendritz 165 Azaferrocenophane 71 Azobenzene carboxylic acids 237 Bidentate pyrazolyl-based ligands 183 Binaphthalene 15 Biomimicry 207 3,6-Bis(2
-pyridyl)-1,2-pyridazine (bppn) 168 3,6-Bis(2-pyridyl)-1,2,4,5-tetrazine (bptz) 192 1,2-Bis(4-nitrophenyl) urea 12 1,1
-Bis(alkyl-N-amido)ferrocene, gold electrodes 63 Bis-bipyridinephenylphosphine oxide 87 Bis-cobaltocenium receptor 48 Bis-copper(II) cryptates 81 Bis-phosphinylferrocene 77 Bis-terpyridine Iridium(III) 76 Boroxine (BOX) 147 Bowl-type structure 198

Cages 182 Calix[4]arenes 90 –, receptors 52 Calix[4]diquinone receptor 77 Calix[6]furan 28 Calix[n]pyrroles 26–28 cAMP 97, 103, 226 Capsules 182 Carbazole-based receptors 29 Carboxylates 232 Carboxypeptidase A 230 Catechol 32 Catenanes 78, 186 Cd(II) 85 CENOTE phosphorylation 97 Charged receptors 33 Chemosensors, artificial 98 Chloride-selective microsensor 240 Cholapod anion receptors 17 Chondroitin sulfate 239 Cobalt (III) dithiocarbamate cryptands 60 Cobaltocenium/cobaltocene 48 Coenzyme A (CoA) 96 Complexation 1 Coordination polymer, bowl-shaped metallo-macrocycles 182 Covalent bonds, reversible 201 Cryptates 81 Crystal structures 1 Crystallographic studies 152 p-Cyanopyridinium crosslinker 241 Cyanuric acid (CNA) 147 Cyclen-Cd complex, 7-amino-4-trifluorocumarin 111 Cyclic peptides 202 Cyclobarbital 223 Cyclotriveratrylene amides 51 Cytokine receptor 117

250 Dendrimers, redox anion sensors 60 Deoxythymidine 106 2,5-Diamidopyrrole skeleton 23 1,8-Diamino-3,6-dichlorocarbazole 29 Diazabicycloalkanes 129 5,5
-Dicarboxamido-dipyrrolylmethanes 24 2,5-Dicarboxamidopyrrole skeleton 21 Diindolylquinoxalines receptors 30 Diphenyl phosphate 227 Diphenylcarbonate hydrolysis 230 Diphenylurea 15 Dipicatriz 166 Dithiocarbamates, redox anion sensing 59 Dithiocyanuric acid, 2-ethyleneamine derivatives 159 DNA 96, 98, 225 Dodecanethiol 88 Dynamic combinatorial chemistry (DCC) 176, 191 Dynamic combinatorial library 175 Electrochemical sensing 47 ERK2 (extracellular signal-regulated kinase 2) 97 Eu(III) 86 Excimers 108 EYPC lipid bilayers 4

Subject Index Hexapeptide receptors 11 Host/anion “sandwich” 7 Host–guest systems 95, 127 Hydrogel, supramolecular 119 Hydrogel fibers, hydrophobic microdomains, phosphate derivatives 120 Hydrogen bond donors 2 Hydrogen bonding 1 Hydroxy (OH) donors, neutral receptors 32 Hyper-phosphorylation 117 Imidazolium-based receptors 34 Indium tin oxide (ITO) 228 Indole-based receptors 29 Indolocarbazoles sensors 29 Interlocked species 186 Ion-sensitive field effect transistor (ISFET) device 236 IP3 97, 102 Iridium(III) polypyridine thiourea, cyclometallated 76 Isophthalamide receptors 2 1-Isopropenyl-4-isopropyl-2-isocyanate 234 Kemp’s triacidic imides 103 Kinase-catalysed transfer 225 Lanthanide(III) complexes

Ferrocene 48 Ferrocene-boronic acid 51 [3,3]Ferrocenophane 72 Flavin adenine dinucleotide (FAD) 96 Fluorescent sensing 95 Fluoride-ferrocenium 51 Glucose-6-O-sulfate 239 Guanidinium-based receptors 37, 102, 225 H donor, aromatic, neutral receptors 21 Halides 2 –, π aromatic systems 132 Helicates 191 Heparan sulfate 239 Hexafluorobenzene 132 Hexakis(pyridine-2-yl)-[1,3,5]triazine2,4,6-triamine 163 Hexanickel cage 183

86

Macrocycles 178 Macrocyclic amide receptors 6 Metal complexes, anion sensing 46 Metal–ligand coordination bonds 192 Metalla-cyclophanes 192 Metallocene optical anion sensors 68 Metallocene redox anion sensors 48 Metallocene-Lewis acid anion receptors 53 Metallodithiocarbamate macrocyclic receptors 59 Metallo-squares 193 Metallo-tetrahedron 196 Metallo-triangle 196 6-Methacrylamidohexanoic acid 237 Methotrexate 235 Methyl orange (p-dimethylaminoazobenzenesulfonic acid) 222 Methylphosphonic acid (MPA) 228

Subject Index Methylpurine 181 Micelles 208 MIPs 207 Mitogen-activated protein kinase (MAPK) 97 Molecular interaction potential with polarization (MIPp) 139 Molecular recognition 95, 175 Molecularly imprinted polymers (MIPs) 207, 209 Multisite phosphorylation 117 Myoinositol-1,4,5-triphosphate 97, 102 Naphthalene 15 Naphthalene sulfonates 238 Nerve agent detection, MIPs 227 Neutral receptors 2 Nicotinamide adenine dinucleotide (NAD) 96 p-Nitrophenyl acetate hydrolysis 229 p-Nitrophenylphosphonate 215 Nitrophenylurea 82 Nucleoside phosphates (nucleotides) 96 Octadecylsiloxane (ODS) 228 Octamethylcalix[4]pyrrole 26 Oligo-(p-phenylene)-N,N-naphthalenediimide (O-NDI) 169 Online sensing 207 Optical anion sensors 68 Organometallic receptors 45 Organophosphorous pesticides/herbicides 227 PDGFR-β 117 Pentafluorophenyl receptor 162 Pentamethyl-amidoferrocene dendrimers 63 Perfluoroaromatic compounds 135 Pesticides/herbicides, organophosphorous 227 PET sensors 12 Phenylbutyrate 27 Phosphate 95 Phosphate anion recognition, coordination chemistry 104 Phosphate derivatives 225 –, ratiometric detection 109 Phosphate recognition, electrostatic interactions/hydrogen bonds 98

251 Phosphoprotein 95 Photo-induced electron transfer (PET) 98 Pinacolyl methylphosphonate (PMP), Soman 227 Platinum(II), methylpurine 181 Point-of-care devices 207 Poly(vinylimidazole) 215 Polyamine, cyclic/acyclic 98 Polyammonium-based anion receptors 39 Polyaza receptors 73 Poly-guanidinium 102 Polynucleotides 208 Polypyridyl anion receptors 56 Polypyrrole 239 Poly-urea 178 Prodigiosin mimics 23 Protein phosphorylation 225 Protein surfaces, phosphorylated 112 Pseudopeptidic macrocycle 178 Pseudorotaxanes 66, 186 Pt(II) 90 Pyridine-2,6-dicarboxamide “caps” 7 Pyridinium-based receptors 34 Pyromellitamide 4 Pyrophosphate 99 Pyrrole-based receptors 21 Quinoxaline phenanthroline 59 Read-out 207 Redox anion-sensing systems, surface confined 63 Resorcin arene 104 Resorcinol 32 Rhenium(I)tricarbonylchloride 77 RNA 96, 225 Rotaxanes 67, 78, 186 Ruthenium(II)bipyridylcalix[4]diquinone receptor 73 Ruthenium(II) polypyridyl 72 Ruthenium-(II)-tris-(2,2
-bipyridiyl), Ru(bpy)3 101 Sacrificial spacer approach 214 Sapphyrin 21 Self-assembly 175, 207 Semi-wet sensor array, phosphate derivatives, high throughput sensing 119 Sialic acid 232

252 Solid-phase extraction (SPE), MIPs 227 Soman 227 Spermine/spermidine 98 Squaramide-containing macrocyclic receptor, fluorescent 40 Squaramido-based receptors 5 Staphylococcus nuclease 102 Steroids 33 Sulfate derivatives 238 Sulfate in water 40 Sulfonamide receptors 162 Sulfonamide-based receptors 2 Supramolecular chemistry 1, 127 Surface confined systems, optical anion sensing 88 Surface imprinting 213 Symmetry-adapted perturbation theory (SAPT) 143 syn–syn conformation 3 Synthetic receptors 207 Tb(III) 86 Templating effects 177 Terephthalate 178 Tetracyanobenzene (TCB) 151, 156 Tetrakis(imidazolium) macrocyclic receptors 35 Tetrakis-p-phenylene[68]crown-20 186 Tetralactam macrocycles 188 Thiocyanuric acid, 2-ethyleneamine derivatives 159 Thiourea 225 Thiourea-based receptors 12

Subject Index Transition metal dithiocarbamates, redox anion sensing, receptors 59 Transition metal polypyridyl anion receptors 56 Transition metal-based receptors, redox-active 47 Transition metals, anion-binding groups 79 Tren skeleton 2 Triazine 133 Tricarboxylate anions 37 Trichloroacetic acids 232 Trifluoro-1,3,5-triazine 133 Tris-[(2-pyridyl)methyl]amine 82 Tris-amides 2 Tris-sulfonamides 2 Tweezers 149 UDP-Gal 97, 111 Urea 225, 234 Urea-based receptors 12 Urea-crown ether functionalised ferrocene 53 Vesicle membranes, transport of chloride ions 18 Vinylanthracene 233 Zeolites 208, 214 Zinc(II)-dipicolylamine (Zn-Dpa) type artificial chemosensors 106 Zn-cyclen type artificial chemosensors 105