Deutsche Forschungsgemeinschaft Microbial Fundamentals of Biotechnology
Microbial Fundamentals of Biotechnology. DFG, Deutsche Forschungsgemeinschaft Copyright © 2001 WILEY-VCH Verlag GmbH, Weinheim ISBN: 3-527-30615-3
Deutsche Forschungsgemeinschaft
Microbial Fundamentals of Biotechnology Final report of the collaborative research centre 323, „Mikrobielle Grundlagen der Biotechnologie: Struktur, Biosynthese und Wirkung mikrobieller Stoffe“, 1986 – 1999 Edited by Volkmar Braun and Friedrich Götz Collaborative Research Centres
Deutsche Forschungsgemeinschaft Kennedyallee 40, D-53175 Bonn, Federal Republic of Germany Postal address: D-53175 Bonn Phone: ++49/228/885-1 Telefax: ++49/228/885-2777 E-Mail: (X.400): S = postmaster, P = dfg, A = d400, C = de E-Mail (Internet RFC 822):
[email protected] Internet: http://www.dfg.de
This book was carefully produced. Nevertheless, editor, authors and publisher do not warrant the information contained therein to be free of errors. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate.
Cover: Crystal structure of E.coli FhuA with bound rifamycin CGP 4832 as determined by Ferguson et al. Structure 9:707-16 (2001)
Library of Congress Card No.: applied for
A catalogue record for this book is available from the British Library.
Die Deutsche Bibliothek ± CIP Cataloguing-in-Publication Data A catalogue record for this publication is available from Die Deutsche Bibliothek ISBN 3-527-30615-3
WILEY-VCH Verlag GmbH, Weinheim (Federal Republic of Germany), 2001 Printed on acid-free and chlorine-free paper. All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form ± by photoprinting, microfilm, or any other means ± nor transmitted or translated into a machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law. Composition: ProSatz Unger, Weinheim Printing: betz-druck gmbh, Darmstadt Bookbindung: J. Schåffer GmbH & Co. KG, Grçnstadt Printed in the Federal Republic of Germany
Contents
Antibiotics and Other Biologically Active Microbial Metabolites 1
1.1 1.2 1.3 1.4 1.5 1.6 2 2.1 2.2 2.3
3 3.1 3.2 3.3 3.4 3.5 3.6 3.7
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Volkmar Braun and Friedrich Götz Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Antibiotic research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The unique features of microbial iron transport . . . . . . . . . . . . . . . . Transport of bacterial proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Membrane components and membrane polarization . . . . . . . . . . . . Chemistry of microbial peptides and proteins . . . . . . . . . . . . . . . . . Summary of short-term projects of the collaborative research centre Screening for New Secondary Metabolites from Microorganisms Hans-Peter Fiedler and Hans Zähner Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Screening methods and novel compounds . . . . . . . . . . . . . . . . . . . . Increasing structural diversity by directed fermentations . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biosynthesis of the Lantibiotics Epidermin and Gallidermin . . . . . Friedrich Götz and Günther Jung History of lantibiotics and lantibiotic research in Tübingen . . . . . . . Primary structure and proposed maturation of epidermin in staphylococci . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Genetic organization and regulation of the epidermin genes . . . . . Isolation and characterization of genetically engineered gallidermin and epidermin analogues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Function of the epidermin immunity genes epiFEG . . . . . . . . . . . . . Inactivation and characterization of the epidermin leader peptidase EpiP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The flavoenzyme EpiD and formation of peptidyl-aminoenethiolates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3 3 3 6 9 10 12 14 16 16 18 41 45 47 52 52 55 56 59 66 72 74 V
Microbial Fundamentals of Biotechnology. DFG, Deutsche Forschungsgemeinschaft Copyright # 2001 WILEY-VCH Verlag GmbH, Weinheim ISBN: 3-527-30615-3
Contents 3.8
3.9
4 4.1 4.2 4.3 4.4 4.5
5 5.1 5.2 5.3 5.4 5.5 5.6
6
6.1 6.2 6.3 6.4
7
7.1 7.2 7.3
VI
Incorporation of d-alanine into S. aureus teichoic acids confers resistance to defensins, protegrins, and other antimicrobial peptides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fermentation of Lantibiotics Epidermin and Gallidermin . Uwe Theobald Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Strains for gallidermin/epidermin production . . . . . . . . . . . . Disadvantages during gallidermin process development . . . Gallidermin – a lantibiotic and its way towards industrial production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
83 86 88
......
93
...... ...... ......
93 94 94
...... ...... ......
95 99 100
Genetics of Nikkomycin Production in Streptomyces tendae Tü 901 Christiane Bormann Introduction: nikkomycins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Isolation of nikkomycin biosynthetic genes . . . . . . . . . . . . . . . . . . . Isolation of the nikkomycin gene cluster and expression in Streptomyces lividans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Organization of the nikkomycin gene cluster . . . . . . . . . . . . . . . . . . Roles of the nik genes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Transcriptional organization and regulation of the nik cluster . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
102
Glycosylated Antibiotics: Studies on Genes Involved in Deoxysugar Formation, Modification and Attachment, and their Use in Combinatorial Biosynthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Andreas Bechthold Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cloning of the avilamycin, landomycin, urdamycin, and granaticin biosynthetic gene clusters . . . . . . . . . . . . . . . . . . . . . Organization of avilamycin, landomycin, urdamycin, and granaticin biosynthetic genes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . New genetically engineered natural compounds . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Analysis of the Biosynthesis of Glycopeptide Antibiotics: Basis for Creating New Structures by Combinatorial Biosynthesis . . . . . . . Stefan Pelzer and Wolfgang Wohlleben Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
102 104 108 109 110 120 121
124 124 127 128 132 136
139 139 141 149 149
Contents 8
8.1 8.2 8.3
Homologous Recombination and the Induction of the SOS-Response in Antibiotic Producing Streptomycetes Günther Muth and Wolfgang Wohlleben Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mutational analysis of the S. lividans recA gene . . . . . . Regulation of RecA activity . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.........
151
. . . .
151 152 156 160
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
Membrane Processes 9
9.1 9.2
9.3
9.4 9.5 9.6
10 10.1 10.2 10.3 10.4
11
11.1 11.2 11.3 11.4
Regulated Transport and Signal Transfer Channels involved in Bacterial Iron Supply . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Volkmar Braun and Helmut Killmann Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Fhu proteins catalyze active transport of ferrichrome and the antibiotic albomycin across the outer membrane and the cytoplasmic membrane of E. coli . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Transduction of energy from the cytoplasmic membrane into the outer membrane for the activation of FhuA as a transporter and phage receptor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Transport of ferrichrome across the cytoplasmic membrane . . . . . . . Ferric-carboxylate transport system of Morganella morganii Volkmar Braun . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Transport of ferric iron ions by the Sfu system of Serratia marcescens Volkmar Braun . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Iron Transport in Gram-negative and Gram-positive Bacteria . Klaus Hantke Ferric iron transport in bacteria . . . . . . . . . . . . . . . . . . . . . . . . . . Ferrous-iron transport systems (Feo) of E. coli . . . . . . . . . . . . . . Regulation of iron transport and metabolism . . . . . . . . . . . . . . . An [2Fe-2S] protein is involved in ferrioxamine B utilization . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
163 163
165
175 178 181
182 183 183
...
188
... ... ... ... ...
188 194 195 198 201
Regulation of the Ferric-Citrate Transport System by a Novel Transmembrane Transcription Control . . . . . . . . . . . . . . . . . . . . . . Volkmar Braun and Sabine Enz Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Transport of Fe3+ is mediated by citrate . . . . . . . . . . . . . . . . . . . . . . Transcription initiation by a signaling cascade from the cell surface into the cytoplasm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Iron regulation of fecIR and fecABCDE transcription . . . . . . . . . . . .
205 205 205 207 209 VII
Contents
12 12.1 12.2 12.3 12.4 12.5 12.6
13
13.1 13.2 13.3
14
14.1 14.2 14.3 14.4 14.5 14.6
15
15.1 15.2 15.3 15.4
VIII
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
209 211
Structure, Function, Import, and Immunity of Colicins . . . . . . . . . Volkmar Braun and Helmut Pilsl Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Colicin M inhibits murein biosynthesis and thus displays a unique activity among the colicins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Colicins 5 and 10 are taken up by a novel mechanism . . . . . . . . . . . Colicins evolved by the exchange of DNA fragments which precisely defined functional domains . . . . . . . . . . . . . . . . . . . . . . . . Pore-forming colicins are inactivated by the cognate immunity proteins shortly before the formation of the transmembrane pores . Pesticin is a muramidase which is inactivated by the immunity protein in the periplasm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
212
Structure, Activity, Activation, and Secretion of the Serratia marcescens Hemolysin/Cytolysin . . . . . . . . . . . . . . . . . . . . . . . Volkmar Braun and Ralf Hertle Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Characterization of the S. marcescens hemolysin (ShlA) . . . . . . Pathogenicity of S. marcescens hemolysin/cytolysin . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
213 214 215 217 218 220 220
...
222
... ... ... ...
222 224 231 235
Staphylococcal Lipases: Molecular Characterization and Use as an Expression and Secretion System . . . . . . . . . . . . . . . . . . . . . . . . Friedrich Götz and Ralf Rosenstein Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Molecular organization of staphylococcal lipases . . . . . . . . . . . . . . . Biochemical characterization of staphylococcal lipases . . . . . . . . . . Role of the pro-peptide region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The use of ShyL as expression and secretion system . . . . . . . . . . . . Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A Multienzyme Complex Involved in Murein Synthesis of Escherichia coli . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Moritz von Rechenberg, Waldemar Vollmer, and Joachim-Volker Höltje The murein sacculus, a “growing” molecule . . . . . . . . . . . . Murein growth is accompanied by massive turnover . . . . . . Enlargement and division of a stress bearing structure . . . . Interaction of murein hydrolases and synthases as indicated by affinity chromatography . . . . . . . . . . . . . . . .
212
238 238 239 241 244 244 246 246
......
249
...... ...... ......
249 252 253
......
254
Contents 15.5 15.6 15.7 15.8
Dimerization of the bifunctional transpeptidase/transglycosylase PBP1B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reconstitution of the core particle of a murein synthesizing machinery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Proposed structure of a hypothetical holoenzyme of murein synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Recent insights in the mechanism of growth of the murein sacculus reveal novel targets for antibiotics . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The Changing Path of Hopanoid Research: From Condensing Lipids to New Membrane Enzymes . . . . . . . . . . . . . . . . . . . . . . Karl Poralla 16.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.2 The cyclization reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.3 Purification of squalene cyclases . . . . . . . . . . . . . . . . . . . . . . . . . 16.4 Properties of purified cyclases . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.5 Cloning of squalene-hopene cyclases . . . . . . . . . . . . . . . . . . . . . 16.6 Properties of SHC sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.7 The structure of squalene-hopene cyclase . . . . . . . . . . . . . . . . . 16.8 Site directed mutagenesis of squalene-hopene cyclase . . . . . . . 16.9 Hopanoid biosynthesis gene clusters . . . . . . . . . . . . . . . . . . . . . 16.10 Miscellaneous results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.11 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
255 256 256 259 260
16
...
263
... ... ... ... ... ... ... ... ... ... ... ... ...
263 265 266 267 270 271 272 274 278 279 280 280 281
17
Genetic and Biochemical Analysis of the Biosynthesis of the Orange Carotenoid Staphyloxanthin of Staphylococcus aureus . . Friedrich Götz 17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.2 Cloning of the carotenoid biosynthetic genes from S. aureus Newman in S. carnosus and E. coli . . . . . . . . . . . . . . . . . . . . . . . . . . 17.3 Function of CrtM and CrtN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.4 Identification of carotenoids in S. carnosus (pOC21), E. coli (pUG1), and S. aureus Newman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.5 Identification of dehydrosqualene in E. coli (pUG1) and E. coli (UG9) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.6 Squalene is very likely no substrate for CrtN, the proposed dehydrosqualene desaturase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.7 The crt operon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.8 Homology of CrtO, CrtP, and CrtQ . . . . . . . . . . . . . . . . . . . . . . . . . . 17.9 Construction of crtM mutants of S. aureus strain Newman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.10 sB-regulated promoter of the crt operon from S. aureus strain Newman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
284 284 285 286 287 287 288 288 289 289 290 IX
Contents 17.11 The carotenoid biosynthesis genes . . . . . . . . . . . . . . . . . . . . . . . . . . 17.12 Function of the pigments in S. aureus strain Newman . . . . . . . . . . . 17.13 Distribution of pigment biosynthesis genes among staphylococcal species . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
290 292
Second Messenger Systems in Paramecium . . . . . . . . . . . . . . . . . . Joachim E. Schultz and Jürgen Linder Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Identification and characterization of cGMP and cAMP second messenger signaling systems in Paramecium . . . . . . . . . . . . . . . . . . Biochemical properties of an adenylyl cyclase . . . . . . . . . . . . . . . . . A guanylyl cyclase disguised as an adenylyl cyclase . . . . . . . . . . . . On the way to an adenylyl cyclase with an intrinsic ion conductance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Downstream of second messengers . . . . . . . . . . . . . . . . . . . . . . . . . In vivo screening of bacterial secondary metabolites . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
295
18 18.1 18.2 18.3 18.4 18.5 18.6 18.7
292 293
295 296 300 302 306 308 311 312 312
Chemical Synthesis and Structure Elucidation 19
19.1 19.2 19.3 19.4
Structure Elucidation and Chemical Synthesis of Microbial Metabolites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Roderich D. Süßmuth, Jörg Metzger, and Günther Jung Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Development of methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Structure elucidation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.....
319
. . . . .
. . . . .
319 320 325 337 339
... ... ... ... ... ... ... ... ... ... ...
345 345 345 348 349 349 355 364 364 365 365
. . . . .
. . . . .
. . . . .
Documentation 20 20.1 20.2 20.3 20.4 20.5 20.6 20.7 20.8 20.9 20.10 X
Documentation of the Collaborative Research Centre 323 List of institutes involved . . . . . . . . . . . . . . . . . . . . . . . . . . . List of supported project areas . . . . . . . . . . . . . . . . . . . . . . Promotion of members of the collaborative research centre Recruitment of new project leaders . . . . . . . . . . . . . . . . . . . Alphabetical list of members and participants . . . . . . . . . . Support of young scientists . . . . . . . . . . . . . . . . . . . . . . . . . Alphabetical list of guests . . . . . . . . . . . . . . . . . . . . . . . . . . International cooperation . . . . . . . . . . . . . . . . . . . . . . . . . . . International conferences . . . . . . . . . . . . . . . . . . . . . . . . . . Funding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . .
... ... ... ... ... ... ... ... ... ... ...
Preface
This book summarizes the scientific contributions of members of the collaborative research centre 323. Tübingen is one of the very few academic places worldwide where microbes have been systematically screened for biologically active metabolites. This research started in 1964 when the first chair of microbiology was created at the University of Tübingen and was led by Hans Zähner. Research on antibiotics in Tübingen continued in the 1980s, when most pharmaceutical companies had abandoned the development of new antibiotics as more than 100 antibiotics were already available to treat seemingly all relevant microbial infections. We now know that this decision was premature. Bacterial antibiotic resistance was already emerging and continued to progress at an increasing pace, resulting in multi-resistant pathogens which now can only be controlled by newly developed antibiotics. The collaborative research centre 323 was an ideal instrument for bringing together scientists of different disciplines and defining common interests. It consisted of chairs and groups of Microbiology/Biotechnology, Microbiology/Membrane Physiology, Microbial Genetics, Pharmaceutical Chemistry, Pharmaceutical Biology, and Organic Chemistry of the University and various groups of the Max-Planck Institutes of Developmental Biology (Biochemistry Department), and Infection Biology. The members of the collaborative research centre met regularly in seminars, which led to very successful co-operations in nearly all scientific projects. Striking examples include the lantibiotic research, the identification of new siderophores in pathogenic microorganisms, the screening/isolation/and subsequent structure elucidation of new antibiotics, and the synthesis of defined substrates for iron transport analysis or hemolysin function. The extremely fruitful cooperation between microbiologists and chemists is documented by co-authorship in numerous publications and is specified in more detail in the “Research Projects”. What is the secret of success of a collaborative research centre? There are several reasons why the scientific outcome of a collaborative research centre is usually more than the sum of the individual projects. First of all, the duration of a collaborative research centre is long-term. Our collaborative research centre 323 was designed to run for 15 years, which allowed the realization of long-term XI Microbial Fundamentals of Biotechnology. DFG, Deutsche Forschungsgemeinschaft Copyright # 2001 WILEY-VCH Verlag GmbH, Weinheim ISBN: 3-527-30615-3
Preface future-oriented projects. Research at universities is normally hampered by short-term grants and fellowships. The collaborative research centre enabled researchers to delve into scientific topics without the persistent fear of premature termination of the project, which would render the field open for competitors to “harvest the fruit”. Secondly, successful scientific cooperation is a matter of trust, competence, and bilateral benefit. At a single university or research institution, it is not easy to find a configuration that meets these prerequisites. A foundation of trust takes time to develop and is largely dependent on individuals. The environment of our collaborative research centre 323 facilitated and stimulated scientific cooperation. The importance of the collaborative research centre for our place at the frontier of science may be best illustrated by the recruitment of the scientific staff to meet the requirements of the collaborative research centre 323. It is no exaggeration to say that through the efforts of the members of the collaborative research centre 323, fundamental and important results were achieved in a number of areas, and many colleagues worldwide were influenced and stimulated by the achievements of the collaborative research centre. We are deeply indebted to the Deutsche Forschungsgemeinschaft for continuous support and to its reviewers for dedicated evaluation of the general concept of the collaborative research centre and the individual research projects. We would also like to thank the University of Tübingen and the Ministry for Science of Baden-Württemberg for their understanding and financial support. Volkmar Braun
XII
Antibiotics and Other Biologically Active Microbial Metabolites
Microbial Fundamentals of Biotechnology. DFG, Deutsche Forschungsgemeinschaft Copyright # 2001 WILEY-VCH Verlag GmbH, Weinheim ISBN: 3-527-30615-3
1 Introduction Volkmar Braun* and Friedrich Götz**
Abstract
A summary of the major scientific achievements in the antibiotic, the membrane-traffic, and chemistry projects of the collaborative research centre 323 from 1986 to 1999 will be presented in the Introduction, which is followed by a more detailed description of the research projects of 1995 to 1999.
1.1 Antibiotic research
Because of the threatening spread of multi-resistant pathogenic microorganisms, the search for and development of new antibiotics (anti-infective drugs) has become more compelling than ever before. Vancomycin-resistant Staphylococcus aureus strains have already been isolated in various countries, and one can foresee that we are facing an increased morbidity and mortality due to treatment failure. Therefore, the search for new anti-infectives and lead compounds and the development of new strategies will always be important. Antibiotics are considered secondary metabolites, which, at least under laboratory conditions, do not participate in the primary metabolism essential for microbial growth. Their role in the natural environment has always been an issue in the collaborative research centre. Antibiotics inhibit microorganisms com-
* Mikrobiologie/Membranphysiologie, Universität Tübingen, Auf der Morgenstelle 28, D-72076 Tübingen ** Mikrobielle Genetik, Universität Tübingen, Waldhäuser Str. 70/8, D-72076 Tübingen
3 Microbial Fundamentals of Biotechnology. DFG, Deutsche Forschungsgemeinschaft Copyright # 2001 WILEY-VCH Verlag GmbH, Weinheim ISBN: 3-527-30615-3
1 Introduction peting for the same ecological niche. It was proposed that secondary metabolism also forms the basis for the evolution of new metabolic pathways. Through elucidation of the mode of action of various antibiotics, a large number of distinct metabolic activities of both prokaryotic and eukaryotic organisms were identified. Because of the beneficial activities of compounds related to antibiotics, such as siderophores, which deliver iron to microorganisms, antibiotics can no longer be regarded as secondary metabolites. Antibiotics with structures similar to those of iron complexes were found to be actively transported into target cells. This finding led to the concept of increasing the efficacy of antibiotics by linking them to actively transported compounds.
1.1.1 Screening and fermentation Antibiotic research in Tübingen involved the screening of many newly isolated microorganisms for novel antibiotics and production of the antibiotics in amounts sufficient for structure elucidation and utility testing. In the course of the collaborative research centre, a state-of-the-art fermentation technology was maintained, and new assay systems, and analytical and synthetic tools were developed. Metabolite production was increased by optimizing the fermentation conditions, aided by a sophisticated on-line analysis of the products released into the culture media. An extensive antibiotic database was created to differentiate new compounds from known compounds at an early stage of the investigation. The spectrum of products usually formed by the microorganisms was modified by directed fermentation and more recently by genetic means, such as mutation and combinatorial pathway recombination. The concept was not a large-scale screening in the sense that thousands of strains and compounds were tested per day. Rather, a selective screening was employed using special growth conditions, novel test systems, and various strains, some of which synthesized a number of different secondary metabolites. The screening was also not target-oriented, and an unbiased selection procedure was used. Some test systems were established within the collaborative research centre, and others were used in cooperation with research groups outside the collaborative research centre. Potential applications were focused not only on antibacterial compounds, antifungal products and therapeutic compounds were also considered. Apart from the biotechnological purpose of searching for new microbial metabolites, the wealth of compounds found demonstrated the extremely high metabolic potential of microorganisms. For example, in the original nikkomycinproducing Streptomyces strain, 14 nikkomycin variants were determined and mutagenesis resulted in 24 additional derivatives. The metabolic diversity of Streptomycetes along with their capability of differentiation is reflected by the size of their genome, which is at the upper limit of bacterial genomes. 4
1.1 Antibiotic research
1.1.2 Marketed compounds developed under the collaborative research centre Among the four commercially most interesting compounds developed under the collaborative research centre, only gallidermin and avilamycin are antibacterial compounds; the latter is employed as a food additive. Desferrioxamine B (trade name Desferral) is used to treat iron-overload diseases, phosphinothricin (trade name BASTA) is a herbicide, and the nikkomycins are potent inhibitors of chitin synthases and thus kill fungi, insects, and acarids without toxic side effects on mammals. Nikkomycin Z is currently in the second stage of clinical trial.
1.1.3 Lantibiotics The lantibiotic era in Tübingen began in 1985 with the determination of the structure of epidermin, a peptide antibiotic isolated from Staphylococcus epidermidis. Four research teams in the collaborative research centre worked to reveal this new class of natural products with a novel biosynthesis pathway. Tübingen, where the name “lantibiotic” was coined, has been a stronghold in this research area over the years. The functions of most of the proteins and enzymes involved in biosynthesis were elucidated, although the mechanism of the key reaction, lanthionine formation, is still unknown. Since chemical synthesis of large amounts of epidermin and gallidermin is currently impossible, great efforts were spent to improve the fermentation process, the scale up, and the downstream processing.
1.1.4 Glycosyl antibiotics and regulation of biosynthesis New derivatives of the glycosylated antibiotics avilamycin, landomycin, urdamycin, and granaticin were generated. The gene clusters involved in the biosynthesis of each of the antibiotics were analyzed, and the functions of most of genes were identified. Especially those genes involved in deoxy-sugar formation, modification, and attachment were used to create novel natural products. A series of new genetically engineered natural compounds were created by inactivation or overexpression of certain genes. Balhimycin is a glycopeptide with properties similar to those of vancomycin, which is often used to combat bacterial infections when no other antibiotic is effective. Balhimycin has the same heptapeptide core as vancomycin and dif5
1 Introduction fers in the glycosylation pattern. A cloning system for the producing strain was developed, and a number of genes were identified by using DNA probes based on consensus sequences of the typical biosynthetic enzymes, such as those encoding peptide synthetases and glycosyltransferases. Most of the balhimycin biosynthesis gene cluster was sequenced, and the function of a number of genes analyzed. The door is now open for the generation of hybrid antibiotics using the combinatorial biosynthesis strategy. Antibiotic production of mycelia-forming Streptomycetes is controlled by a complex regulatory network that allows the cells to sense different growth conditions and to react to these changes by producing antibiotics. Antibiotic production of Streptomyces coelicolor was shown to be affected by cell density, nutritional limitations, nutritional shiftdown, imbalance in metabolism, and different kinds of stress. The key player of the SOS response in Streptomyces lividans, RecA, was investigated.
1.1.5 Antibiotics and transport Membrane studies were important for gaining knowledge on the entry of antibiotics into cells, resistance via permeability barriers, and active export systems, and for identifying novel targets and altering the antibiotics to exert their detrimental function. One antibiotic studied is albomycin, which is actively taken up by cells through an iron siderophore (ferrichrome) transport system. Inside the target cell, the antibiotically active portion is cleaved off the carrier, which is released from the cells. Active transport reduces the minimal inhibitory concentration to the lowest value known for an antibiotic that kills Escherichia coli. Another example is phosphinothricyl-alanyl-alanine, which is transported into cells by an oligopeptide system; the antibiotic is released from the peptide carrier by intracellular proteases to generate phosphinothricin, which then inhibits glutamine synthetase. These findings prompted research by the collaborative research centre and by pharmaceutical companies aimed at developing antimicrobial compounds that are taken up by transporters.
1.2 The unique features of microbial iron transport
It was clear from the beginning that iron transport systems must have special features not shared by transport systems of any other nutrient. Under oxic conditions, iron occurs as the completely insoluble Fe3+. Since iron is an important 6
1.2 The unique features of microbial iron transport cofactor of redox enzymes, iron shortage must be overcome by microorganisms; they handle this by synthesizing iron-complexing compounds of low molecular weight, designated siderophores (originally called sideramines, siderochromes). The antibiotic-screening group of the collaborative research centre also had a long-standing interest in microbial iron complexes because sideromycins, potent antibiotics, belong to this class of compounds. Studies of the uptake of sideromycins, the intracellular metabolism, and their mode of action were essential for understanding antibiotic activity. In addition, it was clear that the iron supply must be carefully balanced since iron overload is toxic due to iron-catalyzed radical formation, which results in the destruction of DNA, proteins, and membrane lipids. Therefore, transport of iron-loaded siderophores and sideromycins and regulation of siderophore synthesis and transport were the focus of the iron projects.
1.2.1 Iron transport through the outer membrane of E. coli and other pathogenic bacteria Novel iron transport and regulatory mechanisms were expected and also found. Transport of Fe3+-siderophores was unique in several respects. Transport across the outer membrane of Gram-negative bacteria consumes energy, which is provided by the proton-motive force of the cytoplasmic membrane. Energy transfer from the cytoplasmic membrane to the outer membrane became an important research topic. A major breakthrough was the identification of the proteins involved in energy transfer: TonB, ExbB, and ExbD (Ton system). The Ton system was extensively characterized at molecular, biochemical, and structural levels. Seven E. coli K-12 Fe3+-siderophore transport systems were identified, and those of ferrichrome and ferric citrate were studied in detail. The receptors undergo conformational changes upon substrate binding and through interaction with the energized Ton system, as supported by the analysis of the crystal structure of the FhuA transporter. Upon binding of ferrichrome to FhuA close to the cell surface, a strong structural transition occurs. The long-range conformational change takes place across most of the molecule and the width of the outer membrane. The link to antibiotics is provided by FhuA, which serves as the active transporter of two antibiotics: albomycin is structurally related to ferrichrome; rifamycin CGP 4832 is structurally unrelated to ferrichrome. Surprisingly, both occupy the same position as ferrichrome on FhuA. Iron siderophores transporters homologous to the E. coli K-12 outer membrane transporters were identified in Yersinia enterocolitica and Morganella morganii. The ferrioxamine B transport system of the highly pathogenic Y. enterocolitica O8 strain explains the occurrence of yersiniosis upon treatment of patients suffering from iron overload with Deferral (mesylate salt of desferri-ferrioxamine B). In addition, a siderophore, designated yersiniabactin, was detected in the culture supernatants of highly pathogenic strains. Yersiniabactin 7
1 Introduction was isolated in amounts sufficient for determination of its novel structure. Furthermore, the genes of the entire heme transport system of Y. enterocolitica were cloned and sequenced, and functions were assigned to the encoded proteins. This was the first characterized heme transport system. Characterization of the iron transport systems of Serratia marcescens was initiated by the finding that transcription of the hemolysin genes is iron-regulated. These studies revealed a plethora of Fe3+-siderophore transport systems, one of which transports Fe3+ across the cytoplasmic membrane without involvement of a siderophore. Other research groups later related this system to the uptake of iron delivered by human transferrin to a variety of human pathogenic bacteria.
1.2.2 Iron transport through the cytoplasmic membrane Transport of Fe3+-siderophores, heme, and Fe3+ across the cytoplasmic membrane is catalyzed by ABC transporters, which consist of a periplasmic binding protein, one or two integral membrane proteins, and a cytoplasmic ATPase. ABC transporters represent the most frequently occurring transport systems in bacteria. Regions of interaction between the periplasmic binding protein and the cytoplasmic membrane transporter were shown for the first time with FhuB/D. The same type of ferrichrome transport system was shown to occur in Bacillus subtilis, a Gram-positive bacterium, which lacks a periplasm. Here, a protein similar to the periplasmic binding protein of Gram-negative bacteria is linked by a lipid anchor of the murein lipoprotein type to the cytoplasmic membrane.
1.2.3 Iron transport regulation Fur was the first iron regulatory gene to be mapped, cloned, and sequenced. Fur functions as an oligomer and binds when loaded with Fe2+ to iron-regulated promoters and inhibits transcription. An assay for the identification of Fur-regulated promoters was developed. The ferric citrate transport system displays the particular property that it is not only repressed by iron, but is induced by ferric citrate. Ferric citrate binds to the outer membrane FecA transport protein; this binding initiates a signal that is transmitted by the FecR protein across the cytoplasmic membrane. In the cytoplasm, FecI is converted to an active sigma factor, which in turn transcribes the fecABCDE transport genes. In this dual stepwise control, first iron limitation is recognized and subsequently the transport system is synthesized only when the cognate substrate is in the culture medium. 8
1.3 Transport of bacterial proteins Under anoxic conditions, bacteria may acquire Fe2+, which is much more soluble than Fe3+ and does not require chelating agents. Feo of E. coli, the only Fe2+ transport system characterized at the molecular level, is encoded by three genes; one gene, feoB, is very likely involved in energizing the transport by nucleotide triphosphate hydrolysis. Mutants in feo were shown to be attenuated in the mouse gut.
1.2.4 Intracellular iron metabolism Very little is known about the intracellular iron metabolism in bacteria. The atypical [2Fe-2S] protein FhuF was characterized; the cysteine residues that bind the iron-sulfur center were identified by amino acid replacement studies. FhuF mutants no longer utilize ferrioxamine B as an iron source, which suggests that FhuF may be involved in iron mobilization from ferrioxamine B. The two ironregulated genes sufS and sufD play a role in utilization of ferrioxamine B as an iron source and possibly in intracellular iron metabolism. Sequence similarities of SufS to NifS suggest that SufS is involved in the formation of the iron-sulfur center of FhuF.
1.3 Transport of bacterial proteins
1.3.1 Transport of colicins and toxins The activities, import, immunity, and evolution of bacterial protein toxins were studied. The genes of eight colicins and pesticin were cloned and sequenced. Cells that synthesize the toxins are protected by immunity proteins with a high specificity for the cognate colicin. Most of the colicins are released from cells by lysis proteins that are encoded downstream of the activity and immunity genes. Colicins can be subdivided into the N-terminal translocation region, the central receptor recognition region, and the C-terminal activity and immunity regions. Comparison of amino acid sequences clearly demonstrated evolution of the pore-forming colicins by exchange of DNA fragments that encode functional domains. How and when the immunity proteins in the cytoplasmic membrane inactivate the pore-forming colicins was a major question. The transmembrane topology of the immunity proteins and the regions of interaction with the colicins in9
1 Introduction dicate that the colicins are inactivated shortly before the pores are opened. In contrast, colicin M, which inhibits murein and O-antigen biosynthesis by interfering with C55-lipid carrier regeneration, and pesticin, which degrades murein by a mechanism similar to that of lysozyme, are inactivated by their immunity proteins in the periplasm before they reach their targets. The hemolysin/cytolysin (ShlA) of Serratia marcescens is activated by a single protein (ShlB) in the outer membrane through a novel mechanism during secretion. The hemolysin is a large protein that remains in a non-hemolytic form in the periplasm of cells that synthesize no ShlB protein. The N-terminal portion of ShlA is important for activation and secretion, the central portion for binding to erythrocytes, and the C-terminus for the formation of small pores in the membrane of erythrocytes, leukocytes, and epithelial cells. ShlA represents one of the very few cases where a major phospholipid of a biomembrane also serves as a cofactor for activity. ShlB has the potential to form pores through which ShlA might be exported.
1.3.2 Transport of staphylococcal (phospho)lipases Five different staphylococcal lipase genes of S. aureus, S. epidermidis, and Staphylococcus hyicus were cloned and sequenced. All corresponding proteins are organized as pre-pro-enzymes in which the pro-region comprises between 207 and 267 amino acids. The pro-region acts as an intramolecular chaperone that facilitates translocation of the native lipase; the pro-peptide can also translocate a number of completely unrelated proteins fused to it. The pro-region protects the proteins from proteolytic degradation. The lipase pro-peptide-based expression and secretion system is used by an increasing number of groups for production of human proteins and peptides in Staphylococcus carnosus, a food-grade microorganism for which a cloning system was developed.
1.4 Membrane components and membrane polarization
1.4.1 Biosynthesis of triterpenes in bacteria There is a tremendous variety of triterpenes in the plant kingdom; a single higher plant always contains several types of triterpenes. The triterpenoic hopanoids found in a large number of Gram-positive and Gram-negative bacteria 10
1.4 Membrane components and membrane polarization show less structural variability. In some bacteria, hopanoid biosynthesis genes are present, but are not expressed under laboratory conditions; therefore, an even wider range of bacteria may synthesize hopanoids. The study of triterpene biosynthesis and the structural variation of triterpenes in nature was approached by investigating the membrane-bound squalene-hopene cyclase of Alicyclobacillus acidocaldarius, which proved to be easier to work with than that of plants. The encoding gene was cloned, sequenced, and expressed, and the gene product was purified and characterized. Comparisons of the amino acid sequence with those of other triterpene cyclases revealed a conserved 16amino acid repeat. Interestingly, the highly purified A. acidocaldarius squalenehopene cyclase forms minor products of mostly tetracyclic structure; this finding was important for a better understanding of the cyclase reaction mechanism. The studies formed the basis for the determination of the crystal structure, which, together with the crystal structures of two sesquiterpene cyclases, are the first three-dimensional structures of terpene cyclases. The squalene-hopene cyclase is a monotopic membrane-bound enzyme. Knowledge of the structure allowed site-directed mutagenesis of specific residues in the catalytic cavity. Some mutant squalene-hopene cyclases significantly increased the synthesis of tetracyclic and bicyclic byproducts; a “new” cyclase in which a leucine in the central cavity is replaced by lysine produced a bicyclic compound. Additional genes involved in hopanoid biosynthesis were detected upstream of the squalene-hopene cyclase genes of Bradyrhizobium japonicum, Zymomonas mobilis, and Methylococcus capsulatus; the first bacterial gene found to encode a squalene synthase is among them. In the aerial mycelium of Streptomyces coelicolor, a differentiation-dependent formation of hopanoids was found; hopanoids are not formed in substrate mycelium and when cultures are grown in liquid.
1.4.2 Biosynthesis of staphyloxanthin The yellow to orange colony color of S. aureus is one of the classical species criteria. The main pigment is staphyloxanthin, a C30-carotenoid that is integrated into the cytoplasmic membrane. The genes involved in the biosynthesis of staphyloxanthin were identified and analyzed. Through the creation of deletion mutants and the analysis of the intermediary compounds formed, a biosynthetic pathway was postulated. The function of staphyloxanthin is still unclear; however, the expression of its gene responds to the stress sigma factor, SigB, which suggests that staphyloxanthin is necessary for survival under certain conditions.
11
1 Introduction
1.4.3 Signal transduction by cAMP and cGMP The initial steps in signal transduction in Paramecium involving the second messengers cAMP and cGMP were characterized. Unlike in metazoans, where hormones as first messengers elicit intracellular second messenger formation, in the ciliate Paramecium abrupt changes in the cell’s membrane potential activated second messenger biosynthesis. Characterized behavioral mutants of the ciliate with defined defects in electrogenesis showed that cAMP generation depends on a K+-outward current, whereas cGMP formation is enhanced by a depolarizing Ca2+-inward current. Analysis of clones carrying genes of the respective protozoan nucleotide triphosphate cyclases demonstrated the presence of an adenylyl cyclase embedded in a protein background that strongly resembles a potassium ion channel. Most surprisingly, the guanylyl cyclase is disguised in a membrane topology identical to that of canonical mammalian adenylyl cyclases and, in addition, carries an extended N-terminus that closely resembles a P-type ATPase unit with a total of ten transmembrane-spanning helices. These findings obtained with the ciliate Paramecium open new vistas on the structural and functional evolution of nucleotide triphosphate cyclases and provide Rosetta Stone sequences to decipher novel binding/regulating partnerships.
1.5 Chemistry of microbial peptides and proteins
During the course of the collaborative research centre, innovative analytical and synthetic methods were introduced. These methods allowed high-level biochemical investigations to solve microbiological research problems in interdisciplinary co-operations. Very often the analytical and synthetic work was even decisive for the success of projects.
1.5.1 Structure determinations The major contributions of the chemistry group were the structure determinations of a large number of antibiotics, and the synthesis of precursors, which allowed the elucidation of the activities of enzymes involved in biosynthesis. The 3-D structures of gallidermin and actagardine are the basis for the model of their mode of action, which very recently became of increased interest due to the inhibitory activity on peptidoglycan biosynthesis. One of the most unusual and in12
1.5 Chemistry of microbial peptides and proteins teresting peptide structures ever found is the 43-peptide antibiotic microcin B17, elucidated by multidimensional NMR of the 13C-, 15N-labeled polypeptide containing eight oxazole and thiazole rings in its backbone. This gyrase (topoisomerase II) inhibitor is ribosomally synthesized as a precursor peptide which is post-translationally modified. To elucidate such structures at that time, innovative instrumental methods, such as greatly improved NMR methods, HPLC-ESI-MS, and the Edman sequencer coupled to an ESI-mass spectrometer, and novel chemical transformations had to be introduced. Unusual peptide structures can now be sequenced using very small amounts of samples. The number and complexity of the elucidated natural products increased considerably, e. g. lipoglycopeptides and other unusual peptides, and new nonpeptidic metabolites, such as siderophores, macrolides, polyols, lactam antibiotics, and steroidal antibiotics. Recently, the complex structure of CDA (calciumdependent peptide antibiotic), a peptide pheromone carrying a thiolactone ring, of intermediates in nikkomycin biosynthesis, and of the first linear glycopeptide precursors in the biosynthesis of the antibiotic balhymicin were elucidated.
1.5.2 Peptide chemistry, peptide libraries, and mass spectrometric analysis The continuous improvements in parallel automated synthesis of peptides, peptide mimetics, and peptide libraries contributed extraordinarily to structure activity studies in various groups of the collaborative research centre. The outstanding synthetic capabilities combined with novel achievements in library analytics by ESI-MS, HPLC-ESI-MS, ICR-MS, and Edman pool sequencing stimulated collaborations in microbiology and immunology, in and outside of Tübingen. The binding regions of the gating loop of FhuA were identified using synthetic peptides. A number of enzyme activities (e. g. oxidative decarboxylase in the epidermin biosynthesis) and binding domains of proteins to the cell wall (e. g. autolysin) could only be studied with the aid of synthetic peptides and peptide libraries. Furthermore, many new proteins were characterized by LCMS and Edman sequencing, circular dichroism, peptide mapping, and antipeptide antibodies, such as the novel antifungal protein from Streptomyces. The recent introduction of a high-resolution, Fourier-transform, ion cyclotron resonance mass spectrometer will provide powerful technologies for further fruitful research between microbiologists and organic chemists.
13
1 Introduction
1.6 Summary of short-term projects of the collaborative research centre
The collaborative research centre was designed to run for 15 years. During this period, the details of the scientific program changed; however, the basic concept was maintained. Detailed descriptions of the results are contained in the research reports of the collaborative research centre from 1986–87, 1988–1990, 1990–1993, and 1993–1995. This book contains the detailed reports of 1995–1999. The following paragraphs describe the contributions made in short-term projects by the groups of the listed project leaders. Karl-Dieter Entian. Major contributions to the cloning and sequencing of the epidermin and the pep5 lantibiotic biosynthesis gene clusters were made. Heterologous proteins and peptides in yeast were synthesized. Bernd Hamprecht. Intercellular communication in the human nerve system was studied. A method was developed and successfully applied to the quantitative determination of adenosine binding to neural adenosine receptors, which paved the way for the isolation of adenosine receptors. The activities of glycogen phosphorylase, creatine kinase, and sorbitol dehydrogenase were determined in an attempt to analyze metabolic processes regulated by cyclic AMP in astroglia-enriched primary cultures. The group was further involved in the study of the role carnosine plays in the brain, taurine transport, and the mode of action of bradykinin. The neuronal cell cultures were used to analyze the activities of products isolated in the microbial screening programs. Thomas F. Meyer. Secretion of the IgA protease by Neisseria gonorrhoeae was investigated, and a novel mechanism was discovered. The mode of action of the translocating b-domain in the outer membrane was studied, and the domain was fused with heterologous proteins that became exposed at the cell surface. The OmpT protease was identified as the enzyme that degrades the fused proteins, which results in decreased yields. The formation of disulfide bonds by oxidation in the periplasm was shown to prevent secretion by locking fusion proteins in a secretion-incompetent conformation. The b-domain is therefore suitable for exposing antigens at the cell surface with the aim to produce antibodies and to stimulate the human immune system. Johannes Pohlner. The a-domain of the IgA protease of N. gonorrhoeae was studied since it contains a sequence of basic amino acids found in proteins that enter the nucleus of eukaryotic cells. The group also studied the post-translational processing of the IgA protease polypeptide to form the protease proper, which is released into the culture medium along with the a-domain and the bdomain that reside in the outer membrane. Rainer Haas. The VacA cytotoxin of Helicobacter pylori was characterized. The growth conditions for the production of the toxin were established, the vacA structural gene was cloned and sequenced, the occurrence of vacA in var14
1.6 Summary of short-term projects of the collaborative research centre ious Helicobacter strains was determined, vacA mutants were isolated and characterized, and the secretion mechanism of VacA was studied. In addition, tools for the genetic analysis of Helicobacter were developed. Susanne Klumpp. The protein phosphatases type 1, 2A, and 2C of Paramecium were studied. The genes of the phosphatases 1 and 2C were cloned and sequenced. The three phosphatases were purified to electrophoretic homogeneity, and functions were ascribed to protein domains. As part of a collaboration, the biochemistry of sensory transduction in Paramecium was studied.
15
2 Screening for New Secondary Metabolites from Microorganisms Hans-Peter Fiedler* and Hans Zähner
2.1 Introduction
Originally screening for secondary metabolites was focused on antibacterial compounds. Later on the screening was extended to antifungal, antiviral and antitumor activity and today it has expanded to human medicine, animal health, and plant protection. The initial idea, using only natural products produced by microorganisms has been replaced by the search for novel lead structures, accompanied by the development of novel targets in all application fields. Still, the most prominent source for novel leads is found in nature and especially in the secondary metabolism of microorganisms [1]. The new lead compounds can be used for derivatisation programs or as platform for chemical synthesis. Therefore, the screening for novel secondary metabolites received an increased interest in the last 15 years. More than hundred new test systems are described till today for applications in pharmaceutical and agricultural fields [2]. These in vitro-assays are mostly based on key enzymes or receptors and differ from classical antibiotic assays by the following aspects: . Proteases in microbial samples or extracts lead to false positive results by degradation of assay enzymes or protein receptors. . Numerous assays are sensitive to metabolites from the intermediary metabolism, which are found in variable concentrations in all microbial cultures. . The assays are sensitive to infections, osmotic conditions, and changes in the pH value. . The assays are selective for a distinct mode of action within a cascade and do not take account to the whole cascade.
* Mikrobiologisches Institut, Universität Tübingen, Auf der Morgenstelle 28, D-72076 Tübingen
16 Microbial Fundamentals of Biotechnology. DFG, Deutsche Forschungsgemeinschaft Copyright # 2001 WILEY-VCH Verlag GmbH, Weinheim ISBN: 3-527-30615-3
2.1 Introduction . Not all new developed assays are suitable for automated robot screening. . Quite a number of assays are sensitive to already known compounds. . None of the target directed assays is suitable to detect chemical diversity in microbial cultures or extracts. Within the collaborative research centre 323 we developed screening strategies which aimed not only on classical antibacterial activity, but also on antifungal activity, on activities which are involved in differentiation processes and on detection of novel siderophores. However, our main screening strategy is based on physico-chemical methods to detect a maximal number of novel secondary metabolites in freshly isolated Actinomycete strains. The so-called “chemical screening” which is based on thin-layer chromatography and staining reagents was first introduced by Hamao Umezawa [3] and continued a few years later by Satoshi Omura, who detected staurosporin by this assay which his most prominent compound found by chemical screening [4]. Hans Zähner modified this method with respect to staining reagents, sample preparation, and variation of the cultivation conditions of the isolated strains [5, 6]. The detection of a biological activity comes only second order. A new dimension of insights into the chemical diversity of produced secondary metabolites was the coupling of high-performance liquid chromatography with computer-assisted diode array detection (HPLC-DAD) and the construction of a database of antibiotics and other natural products based on HPLC and UV-visible absorbance spectral libraries (HPLC-UV-Vis Database) by HansPeter Fiedler [7]. This efficient method allowed the identification of known compounds in raw extracts at a very early stage of investigations or permitted a classification of the compound by comparing UV-visible spectra data. The efficiency of HPLC-DAD screening technique was extended by HPLC-ESI-MS analysis in co-operation with the members of the collaborative research centre Prof. Günther Jung and Prof. Jörg Metzger. The additional information of the molecular mass permitted a more accurate search in commercially available chemical databases. The goal of our strategy was to detect a novel compound which then was isolated and broadly tested for its biological activity, considering that each secondary metabolite will have an activity. A further advantage of chemical screening was achieved by testing pure compounds in the assays. Quite a number of test systems are not compatible with microbial cultures or crude extracts and need the application of pure compounds. Nevertheless, we have not sufficient assay systems available and for many new compounds we have so far not found any biological activity despite intensive co-operation with various pharmaceutical and agrobiological companies. We expect that the “Naturstoffpool” (sponsored since 1996 by BMBF and German pharmaceutical companies) comprising a collection of compounds will be tested by the end of 1999 by a large variety of molecular robot assays. For the analysis of the structure-activity relationship of secondary metabolites we increased the number of original compounds by “directed fermentations” and feeding, by modification of precursors during the production phase, or, by “mutasynthesis” using blocked mutants. 17
2 Screening for New Secondary Metabolites from Microorganisms In the following subchapters all new secondary metabolites are described which were detected in various screening programs during 1986 and 1999 in the groups of Prof. Hans Zähner and Prof. Hans-Peter Fiedler within the collaborative research centre 323.
2.2 Screening methods and novel compounds 2.2.1 Classical screening for antimicrobial activity The classical agar plate diffusion assay for the detection of antimicrobial agents produced by Gram-positive and Gram-negative bacteria, yeasts or filamentous fungi was applied only until 1988. By this assay system we found pyridazomycin [8] and chlorotetain [9]. Both antibiotics show a selective antifungal activity. Pyridazomycin is distinguished by the unusual pyridazine ring that was not described before in microbial secondary metabolites, chlorotetain is a dipeptide containing an unusual chlorinated amino acid. A screening for growth inhibitors against Bacillus subtilis revealed a novel peptide antibiotic named aborycin [10]. The structures of the novel antifungal antibiotics are shown in Fig. 2.1.
Cl+ N H2N
N
O COO+ H3N
Cl
H
H2C
CH3
O +
Pyridazomycin
H3N
Streptomyces violaceoniger sp. griseofuscus Tü 2557
CH
CO
NH
CH
H -
COO
Chlorotetain Bacillus subtilis ATCC 6633
HO
Trp
Phe
Val
Val
Ala
S S Cys
Aborycin Streptomyces griseoflavus Tü 4072
Ala
Tyr
Cys
Asn
Gly
P he
Cys
Asp
S
Ser
Cys
Gly
Leu Ile
Gly
S
Gly
Figure 2.1: Microbial secondary metabolites detected by classical screening for antimicrobial activity.
18
2.2 Screening methods and novel compounds
2.2.2 Screening for antibiotics causing morphological changes of hyphae of Botrytis cinerea This assay is based on both, growth inhibition of Botrytis cinerea and morphological changes of the hyphae, a so-called “bulging effect“. By this assay substances with antifungal action in the presence of polyene antibiotics were found. Nikkomycin Z and X [11, review] were the most prominent antibiotics analysed in the group of Prof. Hans Zähner during the collaborative research centre 76. Nikkomycin Z is a potent inhibitor of chitin synthase and is non-toxic for humans. It was several years under intensive investigation as acaricide for agricultural use at BAYER AG but cancelled in 1984 because of too high costs and its too narrow application in plant protection. From 1994 till 1998 nikkomycin Z was developed as an antimycotic agent for therapy of histoplasmosis, blastomycosis and coccidoidomycosis in human medicine by Shaman Pharmaceuticals in the USA and passed to the second clinical trial. The Botrytis-assay was continued during the beginning of the collaborative research centre 323 and resulted in the detection of galbonolides [12, 13], four new members of antifungal macrolide antibiotics. Their structures are shown in Fig. 2.2. HO
OH
O CH3
R HO
H3C
O
CH2
A B
CH3
O
CH3
: R = OCH3 : R = CH3
OH
H3C H3C
O
CH2
O
CH3
OH CH3
O
CH3
OH
H3CO O
H3C
CH2
C
O
CH3
CH3
O
CH3
D
Galbonolides A-D Streptomyces galbus Tü 2253
Figure 2.2:
Microbial secondary metabolites detected in the Botrytis assay.
2.2.3 Screening for novel siderophores With the exception of lactobacilli all other microorganisms are dependent on the uptake of external iron ions to supply their iron-containing enzymes. Because of the extreme water insolubility of Fe3+-ions all microorganisms have developed very efficient iron-chelating compounds and specific iron-uptake systems. Iron-chelating compounds are for example trihydroxamates, catecholes, tricarboxylates, and other compounds which are able to chelate iron. In the course of the collaborative research centre 323 the following novel chelating metabolites were isolated: 19
2 Screening for New Secondary Metabolites from Microorganisms Maduraferrin was isolated from strain Actinomadura madurae DSM 43067 after detection by an HPLC assay [14]. The complexing centres are a salicylamide moiety, a hydroxamic acid group and an acid hydrazide group. The highly hydrophilic carboxylate-type siderophores staphyloferrins A and B were isolated from Staphylococcus hyicus DSM 20459 grown under strong iron-restricted conditions [15–18]. Both compounds are strictly iron-regulated. Staphyloferrin A consists of two molecules citric acid, each linked to D-ornithine by an amino bond, whereas staphyloferrin B consists of 2,3-diaminopropionic acid, citrate, ethylenediamine and 2-ketoglutaric acid. From Bacillus sp. strain DSM 6940 was isolated besides schizokinen the new dihydroxamate siderophore schizokinen B, in which citrate is replaced by aconitate [19]. Rhizoferrin is a novel carboxylate-type siderophore which was isolated in collaboration with the groups of Prof. Winkelmann and Prof. Jung from Rhizopus microsporus and other fungi of the Mucorales [20, 21]. Rhizoferrin is similar in structure to staphyloferrin A. In case of rhizoferrin, D-ornithine is replaced by putrescin as bridge. From a highly virulent Yersinia enterocolitica strain H1852 a siderophore named yersinibactin was isolated [22]. The novel compound contains a benzene and a thiazolidine ring, as well as two thiazoline rings. It forms stable complexes with trivalent cations such as iron and gallium. While we investigated the fermentation of S,S-ethylenediamine disuccinic acid (EDDS) with Amycolatopsis orientalis strain we found out that EDDS is not an iron but a zinc chelator which opens new biotechnological applications. A novel, not ferrioxamine-type siderophore named amycolachrome was isolated that is similar in structure to the fungal ferrichrome-hexapeptides [23]. The structures of the isolated new siderophores are shown in Fig. 2.3.
2.2.4 Screening for secondary metabolites involved in differentiation processes Actinomycetes are characterised by complex differentiation processes. The search for metabolites which influence these processes is of general importance because they give insight in the tricky sequence and regulation of this dramatic event in the live cycle of these organisms. Prof. Heinz Wolf in the group of Prof. Zähner developed a screening system that allows detection of compounds that stimulate formation of aerial mycelium in Streptomycetes and he isolated hormaomycin from Streptomyces griseoflavus W384, a novel peptide-lactone antibiotic [24, 25]. Hormaomycin induces not only aerial mycelium formation but also antibiotic production in these organisms. Germicidins A and B were isolated from Streptomyces viridochromogenes NRRL B-1551 [26]. Germicidin A is the first known autoregulative inhibitor of spore germination in the genus Streptomyces. Another novel peptide, streptofactin, was isolated from the nikkomycin producer Streptomyces tendae Tü 901/8 c in the group of Prof. Fiedler. This bio20
2.2 Screening methods and novel compounds HO O
H
O
HN
NH N
O
O
OH N
N
H N
O
NH
O
COOH
H CH3
H N H
OH
HOOC
COOH
Maduraferrin
HOOC
Rhizoferrin
Actinomadura madurae DSM 43067
Cunninghamella elegans, Rhizopus microsporus
COOH
H
H N
O
HO
OH
O
O
H N
OH O
N H COOH
H N
O
NH
COOH O
HN
O
HO
OH
HOOC
COOH
H2N
O
COOH
HOOC
COOH
B
A Staphyloferrins
Staphylococcus hyicus DSM 20459
O
O N H
N
CH3
CH3
OH
OH
OH
COOH
OH
H N
N
N
N
CH3
CH3
S
O
Schizokinen B
Yersiniabactin
Bacillus sp. DSM 6940
Yersinia enterocolitica
OH
HO
N
CH2
O C
CH2
CH2
C
O CH
H2C
Figure 2.3:
CH
C
O
NH
O
CH2 CH CH2
NH
C
CH
NH
C
CH3
O
OH
CH2 H2C
N O
O
CH
CH2
C
O
C
OH
CH2
NH
C NH
H2C H3C
CH
NH
COOH S
H3C
S
O
N
OH
CH2 HO
Amycolachrome Amycolatopsis orientalis
N H
Novel iron-chelating compounds.
21
2 Screening for New Secondary Metabolites from Microorganisms surfactant plays a structural role in aerial mycelium development of Streptomycetes and supports the erection of aerial hyphae by lowering the surface tension of water films enclosing the colonies. Mass spectrometry results and amino acid analysis revealed the peptide sequence H2N-Leu-Leu-Ala-Val-Ala-Leu-Lys-Thr and a molecular mass of 1021 Daltons, including a further valine. The missing small part with 94 Daltons of the molecule is bound to the N-terminal end of the peptide [27]. Streptofactin is the first peptide described having structurally and autoregulatory functions. The structures of the isolated secondary metabolites involved in differentiation processes are summarised in Fig. 2.4.
CH3 H3C
CH2
HC
H C
O
C
H N
O HC C
CH3
CH
NH O C HC
C O
C
O
C
CH
C O CH CH
C
NH
O
H3C
C CH2
O2N
NO2
NH
NH CH3
CH2
N H
CH3
O
CH NH C
HO
O
N
Hormaomycin Cl
Streptomyces griseoflavus W-384
OH
OH CH3
H3C
O
O
H3C
CH3 H3C
O
O
H3C
Germicidin A
Germicidin B
Streptomyces viridochromogenes NRRL B-1551
Figure 2.4:
22
New secondary metabolites involved in differentiation processes.
2.2 Screening methods and novel compounds
2.2.5 Chemical screening by TLC, monitoring coloured secondary metabolites Concentrated extracts of culture filtrates and mycelia of Streptomyces strains were separated on silica gel TLC. Such strains were investigated whose extracts showed coloured spots on TLC. The assay lead to the detection of various novel anthraquinone, phenazine and polyene antibiotics. Urdamycins A–F were the most prominent secondary metabolites detected by this method [28–32]. These novel angucycline antibiotics, produced by Streptomyces fradiae Tü 2717, are biologically active against Gram-positive bacteria and show a strong cytotoxic activity against stem cells of murine L1210 leukaemia. For the para-quinone metabolites cinnaquinone and di-cinnaquinone, which were isolated from Streptomyces griseoflavus ssp. thermodiastaticus Tü 2484, no biological activities have been detected so far [33, 34]. Two dark green substances, the esmeraldines A and B, were isolated from Streptomyces antibioticus 2706 [35]. They formally derive by condensation of two phenazine residues of the saphenic acid family. They don’t have any antibacterial activity, but esmeraldine B shows a cytotoxic activity against various tumor cell lines. From Streptomyces violaceus Tü 3556 the new naphthoquinone complex naphthgeranines A–D was isolated [36], from which naphthgeranines A and B show a weak antibacterial and antifungal activity, whereas A, B and C have a moderate cytocidal activity against various tumor cell lines. In addition, strain Tü 3556 produced the new naphthoquinone compounds naphtherythrins D–F. The bright-yellow polyene carboxylic acid serpentene which shows an antibacterial activity against Bacillus subtilis was isolated from Streptomyces sp. Tü 3851 [37]. Remarkable regarding the structure is the benzene ring nearly in the middle of the molecule. The structures of the isolated secondary metabolites screened by this approach are summarised in Fig. 2.5.
2.2.6 Chemical screening by TLC, monitoring fluorescent secondary metabolites Two novel metabolites were detected regarding their blue fluorescence on TLC plates by irradiation with UV light. Pyridindolol glycosides were isolated from Streptomyces parvulus Tü 2480 [38]. No biological activity could be observed of all three compounds. Depsichlorins, isolated from Streptomyces antibioticus ssp. griseorubinosus Tü 1661, represent a group of new cyclopeptide antibiotics which show biological activity against Gram-positive and Gram-negative bacteria [39, 40]. The structures of pyridindolol glycosides and depsichlorins are summarised in Fig. 2.6. 23
2 Screening for New Secondary Metabolites from Microorganisms
O
OH
CH3
Urdamycins Streptomyces fradiae Tü 2717 CH3
O
O
O
OH
CH3 O
HO
OH
O H3C
CH3 O
HO
O
A : R = H
O
HO
R OH
O
E : R = SCH3 O
O
CH3 OH
CH3 O
HO O H3C
CH3 HO
B
O
HO
O
OH
O
O
O
CH3
OH
O R
O
CH3
O O
OH
CH3 O
HO
OH
O H3C
CH3
O
O
HO
O
HO
OH
HO
O
C: R = H N
D: R =
O
O
CH3
CH3
O O
OH
CH3 O
HO
OH
OH
O H3C
CH3 HO HO
O O
OH
O
O
OH
F
Figure 2.5: New secondary metabolites by chemical screening using TLC and monitoring colored compounds.
24
2.2 Screening methods and novel compounds O
O COOH
HO
O H2N
NH2 O
COOH
HO
NH2 OH
HOOC
di-Cinnaquinone
O
Cinnaquinone
O
Streptomyces griseoflavus ssp. thermodiastaticus Tü 2486
H3C CH3 H3C
HN N
HOOC
O
N
R O
CH3 H3C
HN N
HOOC
O
N
OH
O
N
N
COOH
A
COOH
B Esmeraldins
Streptomyces antibioticus Tü 2706
a
R = C13H27
b
R =
(CH2)10
CH CH2 CH3
c
R =
(CH2)12
CH CH3
d
R = C15H31
CH3 CH3 (CH2)13
e
R =
f
R = C16H33
g
R = (CH2)14
CH2R1
CH CH3
O
CH3
HO
R4
O
h R = C17H33 i
H
CH CH3 CH3
OH
R1
R2
R3
R4
A
H
H
H
H
B
OH
H
H
H
C
OH
OH
H
H
D
OH
OH
OH
H
R2
R3 H CH3 CH3
O
R = C17H31
CH2OH OH
O
E
HO
O OH
CH3 R2
C
CH3
Naphthgeranins Streptomyces violaceus Tü 3556
CH3 OH
O R1
O
NH CH3 HOOC
O
CH3
O
R1 D E F
R2
CHO H H H H OH
COOH
CH3
Naphtherythrins D-F Streptomyces violaceus Tü 3556
Serpentene Streptomyces sp. Tü 3851
Fig. 2.5 continued
25
2 Screening for New Secondary Metabolites from Microorganisms
OR1 N H H
N
Pyridindolol glucosides
OR2
Streptomyces parvulus Tü 2480
Tü 2480 F2
H
Tü 2480 F3
H
HO
CH
O
O
C
C
HO
H
OH OH
O
H
OH OH OH
O
H
O CH
H3C
CH
CH
O
C
O
C
N
OH
CH
CH
O HO
CH2
Y
C
O
X
C O H3C
OAc
NH
Y
CH NH
C O
N
C
CH3 CH2 CH3 CH3 CH3 O
O
Leu N
CH3
Depsichlorins Streptomyces antibioticus ssp. griseorubinosus Tü 1661
O OAc
B
H
NH
Cl
A
OH OH
O
CH3
Cl
X
OH
HO
OH
HO
Tü 2480 F4
R3
R2
R1
OR3
Homo-Ile N
CH3
O OAc
C
CH3
Leu N
CH3
O OAc
D
CH3
Homo-Ile N
CH3 O
Figure 2.6: New secondary metabolites by chemical screening using TLC and monitoring fluorescent compounds.
26
2.2 Screening methods and novel compounds
2.2.7 Chemical screening by TLC and Ehrlich reagent Ehrlich reagent reacts mainly with primary amines and the products appear as red-violet zones within few seconds on the TLC. This reagent was successfully applied for detection of pyrrol-3-yl-2-propenoic acid and pyrrol-3-yl-2-propenamide, two further non-active secondary metabolites isolated from Streptomyces parvulus Tü 2480, the producer of pyridindolol glucosides [41]. The group of pyrrolams are four biosynthetically new pyrrolozidinones produced by Streptomyces olivaceus Tü 3082 [42]. They show no antibacterial and antifungal activities, but a weak herbicidal activity against wheat and rice seedlings. Pyrrolam influences the embryonic development of the fish Brachydanio rerio. Obsurolides A2 and A3 produced by Streptomyces viridochromogenes Tü 2580 represent a novel class of phosphodiesterase inhibitors [43]; they have no growth inhibiting potency against bacteria, yeasts and filamentous fungi. Two new phenylpentadienamides were detected in Streptomyces sp. Tü 3946 by orange spots on the TLC stained with Ehrlich reagent, 5-(4-aminophenyl)penta-2,4-dienamide and N2-[5-(4-aminophenyl)penta-2,4-dienoyl]-L-glutamine [44]. Both secondary metabolites show no antibacterial and antifungal activities. The structures of the novel secondary metabolites are summarised in Fig. 2.7.
2.2.8 Chemical screening by TLC and blue tetrazolium staining reagent Blue tetrazolium is a relatively specific derivatisation reagent for steroids and reducing compounds. Blue or violet coloured zones are formed on a light background on the TLC sheet. From Streptomyces aurantiogriseus Tü 3149 a compound was isolated which revealed a yellow-orange colour by staining with blue tetrazolium. Because of its stimulation of aerial mycelium and spore formation of Streptomyces glaucescens, the compound was named differolid [45]. No growth inhibiting activity against bacteria, yeasts and filamentous fungi was observed. A further blue tetrazolium positive compound, (2S,3R,4R,6R)-2,3,4-trihydroxy-6-methylcyclohexanone, was isolated from Streptomyces phaeochromogenes ssp. venezuelae Tü 3154 and Streptomyces albus Tü 3226 [46]. The compound shows no biological activity to bacteria and fungi. A new member of natural compounds having a thiotetronic acid structure was isolated from Streptomyces olivaceus Tü 3010 [47]. The secondary metabolite (2S)-4-ethyl-2,5-dihydro-3-hydroxy-2-[(1E)-2-methyl-1,3-butadienyl]-5-oxo2-thienylacetamide shows antibacterial activity especially against Streptomyces strains. From Streptomyces griseoflavus Tü 2880 the bright yellow colabomycins A–C were isolated which represent new members of the manumycin group [48, 49]. They react with blue tetrazolium as brown spots, with vanillin-sulphuric 27
2 Screening for New Secondary Metabolites from Microorganisms H
COR
R
N
N H
O
Pyrrol-3-yl-2-propenoic acid: R = OH Pyrrol-3-yl-2-propenamide: R = NH2
N O
Pyrrolam
A : R = OH B : R = OCH 3
Streptomyces parvulus Tü 2480
C : R = O
CH O CH2 CH3 CH3
Streptomyces olivaceus Tü 3082 O O
5-(4-Aminophenyl)penta-2,4-dienamide
N H
R
CH3
H2N O
HO
Obscurolides A2:: R = CHO A3:: R = CH2OH Streptomyces viridochromogenes Tü 2580
NH2
N2-(5-(4-aminophenyl)penta-2,4-dienoyl)-L-glutamine H2N O COOH NH
Streptomyces sp. Tü 3946
Figure 2.7: agent.
CONH2
New secondary metabolites by chemical screening using TLC and Ehrlich re-
acid as dark violet and with molybdatophosphoric acid as black spots, indicating their reducing character. The main compound, colabomycin A, is active against Gram-positive bacteria and shows a cytotoxic activity against stem cells of murine L1210 leukaemia. In collaboration with Hoechst AG and Prof. Fiedler, seven musacin compounds were detected in extracts of Streptomyces griseoviridis FH-S 1832 on TLC plates. The compounds were detected by blue tetrazolium chloride (showing a blue-violet colour), by anisaldehyde, orcinol, and Ehrlich’s reagent, respectively. The determination of their structure revealed that six of the seven compounds were new [50]; musacin C shows an anthelmintic activity against Caenorhabditis elegans and Trichostrongylus colubriformis. The structures of the isolated secondary metabolites are summarised in Fig. 2.8.
28
2.2 Screening methods and novel compounds O
H
O
H3C
H
OH
O O
O
OH OH
Differolid
(2S,3R,4R,6R)-2,3,4-trihydroxy6-methylcyclohexanone
Streptomyces aurantiogriseus Tü 3149
Streptomyces phaechromogenes ssp. venezuelae Tü 3154
O
H N
H O
O
H
CH2
OH
CH3 CH3 OH
O C
CH2
H2N
S
CH2 CH3 O
Colabomycin A Thiotetronic acid Tü 3010
Streptomyces griseoflavus Tü 2880
O
Streptomyces olivaceus Tü 3010
NH HO
O OH
A
OH
HO
O
CH3 O
OH
OH
O
OH
B
O
O
CH3 O
OH
OH
OH O
HO
CH3
C O
D
OCH3
H3C
O
OH
O
O
O
O
OH HO
F
H3C OH
Musacins Streptomyces griseoviridis FH-S 1832
Figure 2.8: New secondary metabolites by chemical screening using TLC and blue tetrazolium staining reagent.
29
2 Screening for New Secondary Metabolites from Microorganisms
2.2.9 Chemical screening by TLC and anisaldehyde and orcinol reagent With anisaldehyde-sulphuric acid reagent sugars, steroids, and terpenes can be detected. After heating the stained TLC sheets, a great variety of coloured spots from violet, blue, grey to green were formed on a weakly ochre coloured background. The same strain, Streptomyces griseoviridis FH-S 1832, that showed blue violet musacin spots on the TLC plate when sprayed with blue tetrazolium chloride, showed another pattern of spots with altered Rf values and colour, when the plate was sprayed with anisaldehyde and orcinol reagent, respectively. Besides cineromycin B, three new members of the cineromycin group of macrolide antibiotics were isolated [50]. The cineromycins showed weak activity against Gram-positive bacteria; no further biological activities have yet been observed. The structures of the new cineromycins are summarised in Fig. 2.9. O HOH2C
H3C
CH3
H3C
O
HO
H
H3C
CH3
CH3
OH
OH
CH3
H3C
O
Dehydrocineromycin B
O
H
CH3 OH
CH3
H3C
O
Oxycineromycin B
HO
O
O
2,3-Dihydrocineromycin B
Streptomyces griseoviridis FH-S 1832
Figure 2.9: New secondary metabolites by chemical screening using TLC and anisaldehyde or orcinol staining reagent.
2.2.10 Chemical screening by TLC and vanillin-sulphuric acid staining reagent Vanillin-sulphuric acid reacts relatively specific with higher alcohols, phenols and steroids. Coloured zones are produced on a pale background on the TLC sheet. In the mycelium of the colabomycin producing strain Streptomyces griseoflavus Tü 2880, a further compound, called 2880-II, was detected by vanillin-sulphuric acid staining reagent, resulting in a dark brown spot on the TLC sheet. The compound is related to ferulic acid and shows no antibacterial and antifungal activity [51]. (3S,5R,6E,8E)-Deca-6,8-diene-1,3,5-triol and (3S,6E,8E)-1,3-dihydroxydeca6,8-diene-5-one were isolated from the (3S,8E)-1,3-dihydroxydec-8-en-5-one 30
2.2 Screening methods and novel compounds OH
OH
OCH3
OH OH
H3C
(3S,5R,5E,8E)-Deca-6,8-diene-1,3,5-triol
O
O
NH
HO
O
OH OH
H3C
(3S,6E,8E)-1,3-Dihydroxydeca-6,8-diene-5-one 2880-II Streptomyces griseoflavus Tü 2880
Streptomyces fimbriatus Tü 2335
Figure 2.10: New secondary metabolites by chemical screening using TLC and vanillinsulphuric acid staining reagent.
producer Streptomyces fimbriatus Tü 2335. All compounds are inactive against bacteria and fungi [52]. The structures of the isolated secondary metabolites are summarised in Fig. 2.10.
2.2.11 Screening for new secondary metabolites by polystyrene resin fermentation Addition of the non-polar polystyrene resins Amberlite XAD-16 or XAD-1180 to growing cultures of microorganisms, preferably at the end of the growth phase, enables the absorption of unstable intermediate products or stimulates the producing organism to an altered metabolite pattern. The naphthgeranine and naphtherythrine producer Streptomyces violaceus Tü 3556 (see Section 2.2.5) synthesised the novel series of naphtherythrins A–C, when Amberlite XAD-1180 was added to growing cultures after 36 hours of incubation [53]. The main compounds, naphtherythrins A and B show a biological activity against Gram-positive bacteria and a weak activity against fungi. Under the same conditions Streptomyces exfoliates Tü 1424 produced a group of three new naphthoquinone antibiotics named exfoliamycins [54, 55]. They inhibit growth of Gram-positive bacteria, whereas Gram-negative bacteria and fungi are not sensitive against these antibiotics. The structures of secondary metabolites produced during polystyrene resin fermentations are summarised in Fig. 2.11.
31
2 Screening for New Secondary Metabolites from Microorganisms CH3
H3C
CH3
H3C
OH
O
O
OH
O
O OH
O
O
O
HOOC
O
N H O
HOOC
O
A : R = CH3 B: R=H
O
O
N R
O
CH2OH
O
C Naphtherythrins A-C Streptomyces violaceus Tü 3556
Exfoliamycines Streptomyces exfoliatus Tü 1424 CH3
CH3 HO OH
OR
HO
O
O H3C
OH OH O
CH2OH
OH
O
O
O
CH2OH
H3C O
Exfoliamycin R=H 3-O-Methylexfoliamycin R = CH3
O
Anhydroexfoliamycin
Figure 2.11: New secondary metabolites by screening using polystyrene resin fermentation.
2.2.12 Screening for new secondary metabolites by HPLC and photoconductivity detection Photoconductivity detection has a complete different detection window than UV-Vis spectroscopy and offers the detection of new secondary metabolites in culture filtrates and extracts of microorganisms by HPLC analysis. Streptomyces antibioticus Tü 99, who is known as a producer of chlorothricin, juglomycins A and B, ketomycin, nikkomycins Z and J, as well as nocardamine, was reinvestigated using a HPLC photoconductivity screening system. With this method we could detect four new butenolides [56]. The compounds, which are summarised in Fig. 2.12, show a weak antibiotic activity against Pseudomonas aeruginosa and also a weak inhibition of the chitinase from Serratia marcescens.
32
2.2 Screening methods and novel compounds OH
OH
O
H3C
O
H3C O
H3C H3C
HO
H
Tü 99-1 OH
Tü 99-2 OH
O
H3C
O
H3C H3C H
O
H3C H3C
HO
O
H3C H3C H3C
Tü 99-3
O
Tü 99-4 Tü 99 Butenolides
Streptomyces antibioticus Tü 99
Figure 2.12:
New secondary metabolites by HPLC-photoconductivity screening.
2.2.13 Screening for new secondary metabolites by HPLC and diode-array detection This method represents a modification of the classical chemical screening procedure. TLC and staining reagents were replaced by the more efficient reversedphase HPLC technique coupled with computerised diode-array detection (HPLC-DAD). Commercially available antibiotics and secondary metabolites from our institute pool were analysed with identical standardised HPLC conditions as culture filtrates and raw extracts from freshly isolated Actinomycete strains. Retention times and UV-visible absorbance spectra of references and biological samples were stored in libraries of a HPLC-UV-Vis-Database [7]. Till today more than 600 secondary metabolites, mostly antibiotics, are stored in the database. The technique was first used for detection of minor congeners and for characterisation of blocked mutants and intermediate products of biosynthetic pathways, and since 1990 as screening method for identification of new secondary metabolites in freshly isolated strains. All new secondary metabolites resulting from this screening strategy are summarised in Fig. 2.13. In raw extracts from the elloramycin producer Streptomyces olivaceus Tü 2353 five minor congeners, elloramycins B–F, were detected by HPLC-DAD and determined in structure [57, 58]. All elloramycins are strongly active against Streptomyces strains. As expected, the less methylated elloramycin B shows the best activity against Gram-positive bacteria. The new tetracenomycins B3 and D3 were detected in a blocked mutant of the elloramycin producer, Streptomyces olivaceus Tü 2353-R [59]. The main 33
2 Screening for New Secondary Metabolites from Microorganisms R2 O
O
O
OH
CH3
O OCH3
H3CO OH
OH
O O
R1
CH3
O
OCH3 OR3 O
CH3 O
O
OH
CH3
OR4
R1
R2
R3
R4
B
H
H
H
CH3
C
H
CH3
CH3
H
D F
H
CH3
H
CH3
OH
CH3
CH3
CH3
O OCH3
H3CO O
OH
Elloramycins B-F
O O
Streptomyces olivaceus Tü 2353 CH3
O
E
OCH3 OCH3
OCH3
O RO
OH
COOH OH
O
OH
CH3
B3 R = CH3 D3 R = H Tetracenomycins B3 and D3 H3C
Streptomyces olivaceus Tü 2353-R
O
H3C
N
N
N
N
COOH
COOH
6-Acetylphenazine-1carboxylic acid
H3C N
Saphenic acid methyl ether
O
R O
N
A
Saphenyl fatty acid esters
R = 12-methyltridecanoic acid
B
R = tetradecanoic acid
D
R = 12-methyltetradecanoic acid
E
R = 14-methylpentadecanoic acid
G
R = hexadecanoic acid
I
R = 14-methylhexadecanoic acid
J
R = unsaturated C18 acid
K
R = 16- methylheptadecanoic acid
COOH
Streptomyces antibioticus Tü 2706
Figure 2.13:
34
OCH3
New secondary metabolites by HPLC-diode-array screening.
2.2 Screening methods and novel compounds
A =
HN
B = O
O
O
CHO HN N
C =
O
HN
O
N
R2
R6 R3 R4
C
O
O R1
CH CH CH C NH CH N
O
NH2
OH
R5 OH OH
O
CHO HN
A =
HN
B = O
N
O
N
R2 CH3
O
C
O
CH CH CH C NH CH N
R1 O
NH2
OH
OH OH
O
CHO HN
A =
HN
B = O
N
O
N
OH O HO
C
O
CH2 CH C NH CH
R1 O
NH2
Nikkomycins Streptomyces tendae Tü 901
NH
OH OH
Fig. 2.13 continued
35
2 Screening for New Secondary Metabolites from Microorganisms
R1
R2
O HN
Sz
H
O
N
CHO
HN
Sx O
R2 C
HN O
O
O
OH
OH
OH N
CHO
HN
Sox
O
N O
Soz
R1
OH
H
O
OH
Nikkomycins
N
Streptomyces tendae Tü 901
O
O
O
O
R
R
N H
N H
CH3
A1 : R = COOH A4 : R = CH2OCH3
HO
CH3
HO
B2 : R = CHO B3 : R = CH2OH B4 : R = CH2OCH3 O
COOR2 O
OH R1
N H
N H
OHC CH3 HO
C : R1 = CHO; R2 = H C2 methyl ester: R1 = CHO; R2 = CH3
D2
Obscurolides Streptomyces viridochromogenes Tü 2580
OH
O CH3
Juglomycin Z
O HO
Streptomyces tendae Tü 901
Fig. 2.13 continued
36
H
COOH
CH3 O
2.2 Screening methods and novel compounds CH2OH O
3
O
HO
OH
CH2OH
CH2OH
5
O
CH3
1-(3-Indolyl)-2,3-dihydroxypropan-1-one
Naphthgeranine F Streptomyces violaceus Tü 3556
Streptomyces violaceus Tü 3556
N
H3C C
N
H H N C
O
CH3 O CH3
O C
N C H H
HOCH2
HO O
C
OH
N H
O
H C
C N CH
S
O H C C
C
CH CH3 O H3C CH3
C H
N
N
C H
O
C
OH
SCH3
CH2 N
CH3
CH3 CH
C
CH3 O
H C
H N C
CH3
CH2OH C H
N H
O C
N
O N
Echinoserin Streptomyces tendae Tü 4031
O
O
NH2
OH
O
NH2
O
O
O
O
O H
O O
COOH
H3C
OR
Dioxolid A R = H Dioxolid B R = COCH3
NH2
Dioxolid D
H
OR
OH
Dehydrodioxolid A R = H Dehydrodioxolid B R = COCH3
para-Hydroxybenzamide Streptomyces tendae Tü 4042
Streptomyces tendae Tü 4042
OH
CH3
CH3
CH3
COOH
O O
OCH3
OH
O
1-Hydroxy-4-methoxy-2-naphthoic acid Streptosporangium cinnabarinum ATCC 31213
OH
O
Spirofungin OH
CH3
CH3
Streptomyces violaceusniger Tü 4113
Fig. 2.13 continued
37
2 Screening for New Secondary Metabolites from Microorganisms OH
OH
OH
H3C
OH
10 5
HO
CH3
HO 20
15
O OR2
CH3
O
1
O H3C
O 1'
R1
O 2'
3'
OH
NH
A
35
OH
NH
OH NH
D
NH2
45
CH3
O
R1
40
H NH2+
CH3
CH3
NH2
45
C
30
H
O
OH
O
NH2
45
25
R2
Kanchanamycins Streptomyces olivaceus Tü 4018
O O
H3C OH
HO
(E)-4-oxonon-2-enoic acid
HO O
Streptomyces olivaceus Tü 4018
O
2E,4Z,7Z-decatrienoic acid
2E,4Z-decadienoic acid
Streptomyces viridochromogenes Tü 6105
CH3
O OH O H3C
CH3
O HO H3C
OH O
O
O
OH O
CH3
O
OH O
CH3
OH O
O
CH3
O O O O
H3C
O HO
H3C
O
Tigloside Amycolatopsis sp.
Fig. 2.13 continued
38
CH3
O CH3
OH O
CH3
2.2 Screening methods and novel compounds Simocyclinones
O
Streptomyces antibioticus TuÈ 6040
CH3
O HO
Simocyclinones A
O
A1: R = H A2: R = OH OH
OH
O
Simocyclinones B B1: R1 = H B3: R1 = OH B4: R1 = OH
R2 = H R2 = H R2 = COCH3
R2
H3C
R
CH3
O HO
O
O
OH
O
OH
HO OH
R1
OH
Simocyclinones C C2: R1 = OH C3: R1 = H C4: R1 = OH
R2 = H R2 = COCH3 R2 = COCH3
O R2
OH
H3C
CH3
O HO
O
O
O
OH
O
O
OH
O
R1
OH
R3 O
O
OH
O R2
HN OH
H3C
O
CH3
HO
O
O
O
OH
O
O
OH
O
R1
OH
Simocyclinones D D2: D3: D4: D6: D7: D8:
R1 = OH R1 = H R1 = OH R1 = OH R1 = H R1 = OH
R2 = H R2 = COCH3 R2 = COCH3 R2 = H R2 = COCH3 R2 = COCH3
R3 = H R3 = H R3 = H R3 = Cl R3 = Cl R3 = Cl
O
OH
O
O
CH3 CH3
OH
O
OH
Kyanomycin
HN
O
O C16-18H33-37
O
O OH
OH
O O
Fig. 2.13 continued
P
Nonomuria sp. NN22303
C13-16H27-33 O
39
2 Screening for New Secondary Metabolites from Microorganisms compound B3 is antibiotically inactive against Gram-positive and Gram-negative bacteria, but D3 shows a moderate activity against Bacillus subtilis and Arthrobacter aurescens. The importance of the new compounds is based in their role as key intermediates and in the elucidation of the biosynthetic pathway of elloramycins and tetracenomycins. Seven phenazine compounds were isolated from Streptomyces antibioticus Tü 2706. Besides saphenamycin, saphenic acid and tubermycin B, three new phenazines were detected by HPLC-DAD, 6-acetylphenazine-1-carboxylic acid, saphenic acid methyl ether and a group of eight saphenyl fatty acid esters [60]. A great success of HPLC-DAD screening was the identification of new nikkomycin compounds in mutants of Streptomyces tendae Tü 901. Twenty new nikkomycins were detected by this method [61–67] allowing intensive studies on structure activity relationships [68] and getting new insights in the biosynthetic pathway of nikkomycins. A further new compound from the juglomycin family, juglomycin Z, was detected in the nikkomycin producing strain Streptomyces tendae Tü 901, when the organism was grown under modified nutrition conditions [69]. The naphthoquinone antibiotic shows biological activity against Gram-positive and Gram-negative bacteria and against yeasts. A series of eight new obscurolides was detected besides the main compounds A2 and A3 in Streptomyces viridochromogenes Tü 2580 [43, 70]. B4 is the most active obscurolide in the phosphodiesterase assay. All obscurolides revealed no growth inhibiting potency against bacteria, yeasts and filamentous fungi. In the naphthgeranine producing strain Streptomyces violaceus Tü 3556 besides tubermycin B and 1-phenazinecarboxylate, a new minor congener naphthgeranine F, as well as 1-(3-indolyl)-2,3-dihydroxypropan-1-one, which was not previously described as a natural product, were detected [71]. Naphthgeranine F showed a similar antibacterial activity against Gram-positive bacteria as naphthgeranine C, the main compound in fermentations of strain Tü 3556, whereas 1-(3-indolyl)-2,3-dihydroxypropan-1-one shows no biological activity. A new member of the quinoxaline group antibiotics, echinoserine, was detected in strain Streptomyces tendae Tü 4031 [72]. The new compound is a noncyclic form of echinomycin, but is not a biosynthetic precursor. Echinoserine is less antibiotically active than echinomycin. Dioxolides, a novel class of secondary metabolites, were detected in the culture filtrate of Streptomyces tendae Tü 4042 [73]. Besides dioxolides, which consist of an unusual substituted dioxolane ring, para-hydroxybenzamide was detected, which was not yet described as a natural product. All compounds show no biological activity against Gram-positive and Gram-negative bacteria, yeasts and filamentous fungi. The kanchanamycins, a group of novel 36-membered polyol macrolide antibiotics were detected in Streptomyces olivaceus Tü 4018 [74, 75]. The compounds show antibacterial and antifungal activities, and are especially effective against Pseudomonas. In the same strain the fatty acid (E)-4-oxonon-2-enoic acid was detected, isolated and determined in structure [76]. This new second40
2.3 Increasing structural diversity by directed fermentations ary metabolite shows an antibacterial activity against various Gram-positive and Gram-negative bacteria, especially against Staphylococcus aureus, but not against yeasts and other fungi with the exception of Paecilomyces variotii. Streptosporangium cinnabarinum ATCC 31213 was characterised as a producer of the known antibiotics 43,334 and 43,596, but also as a producer of the new naphthalene compound 1-hydroxy-4-methoxy-2-naphthoic acid, which shows herbicidal activity in the Lemna minor assay [77]. In the culture filtrates and extracts of Streptomyces violaceusniger Tü 4113 the new secondary metabolite spirofungin was detected, a compound having a polyketide-spiroketal structure that shows various antifungal activities, particularly against yeasts [78]. Tigloside, a new tigloylated tetrasaccharide was detected in Amycolatopsis sp. NN0 21702 [79]. This secondary metabolite shows unusual structural elements, which have never before been isolated from Actinomycete strains. Until now, no biological activity could be observed. Two fatty acids, (2E,4Z)-decadienoic acid and (2E,4Z,7Z)-decatrienoic acid, the latter one being described for the first time as a natural product, were detected in the culture filtrate of Streptomyces viridochromogenes Tü 6105 [80]. Both metabolites show strong herbicidal activities against Lemna minor and Lepidium sativum. Simocyclinones, a novel group of angucyclinones, were detected during the fermentation of Streptomyces antibioticus Tü 6040 [81–84]. They are novel natural “hybrid” polyketide antibiotics and consist of an unusual angucyclinone ring with a tetraene side chain and a coumarin ring. Simocyclinones can be subdivided in A-, B-, C- and D-series regarding structures and UV-visible spectra. Compounds of the D-series are distinguished by an activity against Gram-positive bacteria and against several tumor cell lines. Kyanomycin, a blue-coloured secondary metabolite, was detected in the mycelium extract of Nonomuria sp. by HPLC-DAD and HPLC-ESI-MS screening and determined as an unusual anthracycline-phosphatidylethanolamine hybrid that shows weak antibacterial activity [85].
2.3 Increasing structural diversity by directed fermentations 2.3.1 Biomodification of saphenamycin and esmeraldine Streptomyces antibioticus Tü 2706 produces the antibacterial and antitumor active saphenamycin, a phenazine compound, and the cytotoxic active esmeraldine B (see Fig. 2.5). Esmeraldine is the condensation product of one molecule saphenic acid and one molecule saphenamycin. As saphenamycin is a methylsalicylic acid ester of saphenic acid, esmeraldine B shows the same methylsalicyclic acid side-chain than saphenamycin. It was of interest to investigate whether strain Tü 2706 modifies this side chain when derivatives of methylsalicylic acid 41
2 Screening for New Secondary Metabolites from Microorganisms were fed during directed fermentations, and if modified saphenamycins and esmeraldins show an altered antibacterial and antitumor spectrum. Biomodification was achieved by feeding acetylsalicylic acid, 3-methylsalicylic acid, 4-methylsalicylic acid, 5-fluorosalicylic acid, 5-chlorosalicylic acid and 5-bromosalicylic acid, resulting in six new saphenamycin and six new esmeraldine compounds. No incorporation into the molecules was achieved by feeding 5-iodosalicylic acid, 5-bromo-4-hydroxysalicylic acid, 3-hydroxysalicylic acid, 3-methoxysalicylic acid, 3,5-dinitrosalicyclic acid, 4-aminosalicylic acid, 5sulfosalicylic acid, and thiosalicylic acid [86]. The structural modifications are summarised in Fig. 2.14. All saphenamycin derivatives show an antibacterial activity, however, only 4-methylsaphenamycin is as active as saphenamycin. The cytotoxic activity towards an urinary bladder carcinoma cell line was lower in case of the derivatives than with saphenamycin. In comparison to esmeraldine B, 4-methyl- and 5-fluoro-esmeraldine B show an altered antibacterial spectrum and an increased antibacterial activity. 3-Methylesmeraldine B shows a higher cytotoxic activity against the tested tumor cell line than the original esmeraldine B [86].
2.3.2 Biomodification of ferrioxamines Streptomyces olivaceus Tü 2718 is naturally overproducing the iron-chelating compound desferrioxamine E, also known as the antibiotically active nocardamine. Optimisation of the fermentation conditions yielded in amounts of more than 10 g per litre desferrioxamine E. The siderophore consists of three mole succinic acid and three mole L-lysine, forming a trihydroxamate ring. Desferrioxamine E shows the strongest iron(III)-chelating complexing constant related to desferrioxamines A–I. The specificity for the incorporation of diamines containing two to six carbon atoms was investigated by feeding the following diamines: 1,2-diamine ethane, 1,3-diamine propane, 1,4-diamine butane (putrescine), 1,5-diamine pentane (cadaverine), 1,6-diamine hexane, 1,7-diamine heptane, 1,8-diamine octane, 1,5-diamine ethylether, S-2-aminoethyl cysteine, and N-glycin-1,2-ethylenediamine. The incorporation was monitored by HPLC-DAD that allowed the detection of all modified ferrioxamines [87]. Diamines with a space of more than four carbon atoms between the amino groups were not incorporated into the molecule, such as diamines with more than six carbon atoms. All other diamines were incorporated and led to the isolation and identification of 13 new ferrioxamines besides ferrioxamine D2 and E [88], and were determined in their differences of iron-complexation [89]. The new desferrioxamine structures are shown in Fig. 2.15.
42
2.3 Increasing structural diversity by directed fermentations R2 R3
R1 H3C
O
N
R4 O
OR5
N COOH
R2 R3
R1 CH3 H3C
HN HOOC
N
N
O
R4 O
OR5
N COOH
Figure 2.14: Biomodifications of saphenamycin and esmeraldine B by directed fermentations with Streptomyces antibioticus Tü 2706.
43
2 Screening for New Secondary Metabolites from Microorganisms O NH
C
R3
(CH2)2 N
C OH
O
C (CH2)2 C
O
O
HO R4 N
R1 NH
N
R2
NH
O C (CH2)2 C O
Figure 2.15: Desferrioxamine structures produced by directed fermentations with strain Streptomyces olivaceus Tü 2717.
44
Acknowledgments
2.3.3 Biomodification of rhizoferrins Although rhizoferrin represents a fairly simple molecule that consists of two molecules of citric acid linked to 1,4-diaminobutane through two amide bonds (see Fig. 2.3), it may have potential application in biotechnology due to its appreciable metal-binding properties and the ability to be easily degraded by various microorganisms. The specificity of the biosynthetic enzymes of the rhizoferrin producing fungus Cunninghamella elegans was investigated by modifying both, the chain length of the diamine and the citric acid part of the molecule [21]. Variations in chain length of the diamine backbone were very well tolerated. Branching of functionalization in b-position to the amino group of the diamine compounds was also accepted. However, it was not possible to introduce aamino acids. Therefore, the biosynthesis of rhizoferrin must be different to the biosynthesis of staphyloferrin A, produced by staphylococci. Neither by feeding of D-ornithine nor by application of inhibitors of ornithine decarboxylase, with and without simultaneous addition of D-ornithine, it was possible to detect even trace amounts of staphyloferrin A in Cunninghamella elegans. A higher degree of enzyme specificity was involved in the formation of the activated citryl species since analogues of citric acid are more difficult to be incorporated into rhizoferrin analogues than diamines. The structures are summarised in Fig. 2.16 (series A are structures modified in the diamine part, series B are structures modified in the citric acid part). All derivatives obtained by directed fermentations showed similar iron-chelating properties.
Acknowledgments Following scientists from the “Mikrobiologie/Antibiotika” group contributed to the success in search for novel secondary metabolites: Post-docs: J. Bielecki, C. Bormann, Z. Chen, H. Decker, H. Drautz, U. Fauth, M. Harder, T. Hörner, J. Müller, A. Plaga, O. Potterat, T. Schüz, and U. Theobald. Doctoral students: M. Alverado-Kirigin, N. Andres, S. Blum, M. Brandl, D. Braun, K. Burkhardt, I. Cebulla, H.-J. Cullmann, H. Haag, U. Hartjen, H. Hoff, W. Huhn, C. Isselhorst-Scharr, O. Jung,W. Katzer,W. Kuhn, J. Meiwes, F. Petersen, C. Pfefferle, U. Pfefferle, P. Reuschenbach, M. Richter, J. Schimana, P. Schneider, U. Schneider, A. Seiffert, J. Stümpfel, M. Tschierske, B. Wahl, and F. Walz. The excellent collaboration with the groups of Walter Keller-Schierlein (ETH Zürich), Axel Zeeck and Jürgen Rohr (Universität Göttingen), Günther Jung and Jörg W. Metzger (Universität Tübingen), Wilfried König (Universität Hamburg), Urs Séquin (Universität Basel), and Gerhard Bringmann (Universität Würzburg), which were involved in structure elucidation of isolated secondary metabolites, is gratefully acknowledged. 45
2 Screening for New Secondary Metabolites from Microorganisms A-Series
B-Series
H N
O
H N
COO-
C
HO -
OOC
OH
O
O
C
C
-
COO-
H N
H N
COO-
C
HO
OH
COO-
C
-
H N
H N
O
O
HO OOC
C
-
COO-
H N
H N
COOH
H -
OOC
O
C
-
OOC
CH3
O
HO
C
-
OOC
OH
-
COO-
H N
COOCH3
HO -
OOC
O
C
-
OOC
Monomethylmonodesoxyrhizoferrin
H N
COO-
OOC
COO-
OOC
Oxahomorhizoferrin
C
-
H N
O
C
-
OH
O
C
Didesoxyrhizoferrin
COO-
C
OOC
COO-
OOC
Norrhizoferrin
O
HO -
H N
O
C
OOC
H
O
Monodesoxyrhizoferrin O
-
COO-
COO-
OOC
Homorhizoferrin
O
H N
H N
OOC
2-Methylhomorhizoferrin
H N
H N
O
C
COOCH3
COO-
H3C -
OOC
O
C
-
OOC
Dimethyldidesoxyrhizoferrin
O H N
O
C
H N
COOOH
COO-
HO -
OOC
O
C
-
OOC
2-Oxorhizoferrin
Figure 2.16: Rhizoferrin structures produced by directed fermentations with Cunninghamella elegans.
46
References
References
1. Bérdy, J. (1995) Are actinomycetes exhausted as a source of secondary metabolites? Biotekhnologiya 8, 13–34. 2. Hill, D. C., Wrigley, S. K., and Nisbet, L. J. (1998) Novel screen methodologies for identification of new microbial metabolites with pharmaceutical activity. Adv. Biochem. Engin. Biotechnol. 59, 73–121. 3. Umezawa, S., Tsuchiya, T., Tatsuta, K., Horiuchi, Y., Usui, T., Umezawa, H., Hamada, M., and Yagi, A. (1970) A new antibiotic, dienomycin. I. Screening method, isolation and chemical studies. J. Antibiot. 23, 20–27. 4. Omura, S., Iwai, Y., Hirano, A., Nakagawa, A., Awaya, J., Tsuchiya, H., Takahashi, Y., and Masuma, R. (1977) A new alkaloid AM-2282 of Streptomyces origin, taxonomy, fermentation, isolation and preliminary characterization. J. Antibiot. 30, 275–282. 5. Drautz, H. and Zähner, H. (1981) A new method of screening for secondary metabolites in Streptomycetes. In: Actinomycetes, Zbl. Bakt. Suppl., Vol. 11 (Schaal, K. P. and Pulverer, G., eds), Gustav Fischer Verlag, Stuttgart, pp. 515–519. 6. Zähner, H. and Drautz, H. (1985) Methods in the search for new secondary metabolites from microorganisms – a comparison. In: VIII International Symposium on Medicinal Chemistry, Proceedings Vol 2 (Dahlbohm, R. and Nilsson, J. L. G., eds.), Swedish Pharmaceutical Press, Stockholm, pp. 49–68. 7. Fiedler, H.-P. (1993) Screening for secondary metabolites by HPLC and UV-visible absorbance spectral libraries. Nat. Prod. Lett. 2, 119–128. 8. Grote, R., Chen, Y., Zeeck, A., Chen, Z., and Zähner, H. (1988) Pyridazomycin, a new antifungal antibiotic produced by Streptomyces violaceoniger. J. Antibiot. 41, 595–601. 9. Rapp, C., Jung, G., Katzer, W., and Loeffler, W. (1988) Chlorotetain aus Bacillus subtilis, ein antifungisches Dipeptid mit einer ungewöhnlichen chlorierten Aminosäure. Angew. Chem. 100, 1801–1802. 10. Potterat, O., Stephan, H., Metzger, J. W., Gnau, V., Zähner, H., and Jung, G. (1994) Aborycin – a tricyclic 21-peptide antibiotic isolated from Streptomyces griseoflavus. Liebigs Ann. Chem. 1994, 741–743. 11. Fiedler, H.-P., Schüz, T., and Decker, H. (1993) An overview of nikkomycins: history, biochemistry, and applications. In: Cutaneous Antifungal Agents: Compounds in Clinical Practice and Development (Rippon, J. W. and Fromtling, R. A., eds), Marcel Dekker, New York, pp. 325–352. 12. Fauth, U., Zähner, H., Mühlenfeld, A., and Achenbach, H. (1986) Galbonolides A and B – two non-glycosidic antifungal macrolides. J. Antibiot. 39, 1760–1764. 13. Achenbach, H., Mühlenfeld, A., Fauth, U., and Zähner H. (1988) The galbonolids. Novel, powerful antifungal macrolides from Streptomyces galbus ssp. eurothermus. Ann. New York Acad. Sci. 544, 128–140. 14. Keller-Schierlein, W., Hagman, L., Zähner, H., and Huhn, W. (1988) Maduraferrin, ein neuartiger Siderophor aus Actinomadura madurae. Helv. Chim. Acta 71, 1528–1540. 15. Meiwes, J., Fiedler, H.-P., Haag, H., Zähner, H., Konetschny-Rapp, S., and Jung, G. (1990) Isolation and characterization of staphyloferrin A, a compound with siderophore activity from Staphylococcus hyicus DSM 20459. FEMS Microbiol. Lett. 67, 201–206. 16. Konetschny-Rapp, S., Jung, G., Meiwes, J., and Zähner, H. (1990) Staphyloferrin A: a structurally new siderophore from staphylococci. Eur. J. Biochem. 191, 65–74. 17. Drechsel, H., Freund, S., Nicholson, G., Haag, H., Jung, O., Zähner, H., and Jung, G. (1993) Purification and chemical characterization of staphyloferrin B, a hydrophilic siderophore from staphylococci. BioMetals 6, 185–191.
47
2 Screening for New Secondary Metabolites from Microorganisms 18. Haag, H., Fiedler, H.-P., Meiwes, J., Drechsel, H., Jung, G., and Zähner, H. (1994) Isolation and biological characterization of staphyloferrin B, a compound with siderophore activity from staphylococci. FEMS Microbiol. Lett. 115, 125–130. 19. Seiffert, A. and Zähner, H. (1992) Siderophore production in alkalophilic bacilli. In: Dechema Biotechnology Conferences, Vol. 5 (Kreysa, G. and Driesel, A. J., eds.), VCH, Weinheim, pp. 725–728. 20. Drechsel, H., Metzger, J., Freund, S., Jung, G., Boelart, J. R., and Winkelmann, G. (1991) Rhizoferrin – a novel siderophore from the fungus Rhizopus mircrosporus var. rhizopodiformis. BioMetals 4, 238–243. 21. Drechsel, H., Tschierske, M., Thieken, A., Jung, G., Zähner, H., and Winkelmann, G. (1995) The carboxylate type siderophore rhizoferrin and its analogs produced by directed fermentation. J. Industr. Microbiol. 14, 105–112. 22. Drechsel, H., Stephan, H., Lotz, R., Haag, H., Zähner, H., Hantke, K., and Jung, G. (1995) Structure elucidation of yersiniabactin, a siderophore from highly virulent Yersinia strains. Liebigs Ann. 1995, 1727–1733. 23. Cebulla. I. (1995) Gewinnung komplexbildender Substanzen mittels Amycolatopsis orientalis. Doctoral Thesis, Universität Tübingen. 24. Andres, N., Wolf, H., Zähner, H., Rössner, E., and Zeeck, A. (1989) Hormaomycin, ein neues Peptid-lacton mit morphogener Aktivität auf Streptomyceten. Helv. Chim. Acta 72, 426–437. 25. Andres, N., Wolf, H., and Zähner, H. (1990) Hormaomycin, a new peptide lactone antibiotic effective in inducing cytodifferentiation and antibiotic biosynthesis in some Streptomyces species. Z. Naturforsch. 45 c, 851–855. 26. Petersen, F., Zähner, H., Metzger, J. W., Freund, S., and Hummel, R.-P. (1993) Germicidin, an autoregulative germination inhibitor of Streptomyces viridochromogenes NRRL B-1551. J. Antibiot. 46, 1126–1138. 27. Richter, M., Willey, J. M., Süssmuth, R., Jung, G., and Fiedler, H.-P. (1998) Streptofactin, a novel biosurfactant with aerial mycelium inducing acitivity from Streptomyces tendae Tü 901/8 c. FEMS Microbiol. Lett. 163, 165–171. 28. Drautz, H., Zähner, H., Rohr, J., and Zeeck, A. (1986) Urdamycins, new angucycline antibiotics from Streptomyces fradiae. I. Isolation, characterization and biological properties. J. Antibiot. 39, 1657–1669. 29. Zeeck, A., Rohr, J., Sheldrick, G. M., Jones, P. G., and Paulus, E. F. (1986). Structure of the new antibiotic and cytotoxic indicator substance, urdamycin A. J. Chem. Res. (Synopsis) 1986, 104–105. 30. Rohr, J. and Zeeck, A. (1987) Urdamycins, new angucycline antibiotics from Streptomyces fradiae. II. Structural studies of urdamycins B to F. J. Antibiot. 40, 459–467. 31. Rohr, J., Zeeck, A., and Floss, H. G. (1988) Urdamycins, new angucycline antibiotics from Streptomyces fradiae. II. The structures of urdamycins C and D. J. Antibiot. 41, 126–129. 32. Rohr, J., Beale, J. M., and Floss, H. (1989) Urdamycins, new angucycline antibiotics from Streptomyces fradiae. IV. Biosynthetic studies of urdamycins A–D. J. Antibiot. 42, 1151–1157. 33. Korff, U. (1988) Strukturaufklärung, Synthese und Derivatisierung eines Pigments aus dem Stamm Tü 2484. Doctoral Thesis, Universität Tübingen. 34. Alvarado-Kirigin, M. (1990) Monomeres und dimeres Cinnachinon aus Streptomyces griseoflavus (Tü 2484). Doctoral Thesis, Universität Tübingen. 35. Keller-Schierlein, W., Geiger, A., Zähner, H., and Brandl, M. (1988) Die Esmeraldine A und B, tief grüne Farbstoffe aus Streptomyces antibioticus, Stamm TÜ 2706. Helv. Chim. Acta 71, 2058–2070. 36. Wessels, P., Göhrt, A., Zeeck, A., Drautz, H., and Zähner, H. (1991) Naphthgeranines, new naphthoquinone antibiotics from Streptomyces sp. J. Antibiot. 44, 1013–1018. 37. Ritzau, M., Drautz, H., Zähner, H., and Zeeck, A. (1993) Serpentene, a novel polyene carboxylic acid from Streptomyces. Liebigs Ann. Chem. 1993, 433–435.
48
References 38. Hagmann, L., Keller-Schierlein, W., Wahl, B., and Zähner, H. (1988) Pyridindolol glucosides from Streptomyces parvulus. J. Antibiot. 41, 289–295. 39. Gröne, C. (1991) Strukturaufklärung der Depsichlorine: Neue Depsipeptid-Antibiotika aus Streptomyces Tü 1661. Doctoral Thesis, Universität Göttingen. 40. Cullmann, H. J. (1994) Depsichlorine und andere Metabolite aus Streptomyces antibioticus Tü 1661. Doctoral Thesis, Universität Tübingen. 41. Keller-Schierlein, W., Müller, A., Hagmann, L., Schneider, U., and Zähner, H. (1985) (E)-3-(1H-Pyrrol-3-yl)-2-propensäure und (E)-3-(1H-Pyrrol-3-yl)-2-propensäureamid aus Streptomyces parvulus. Helv. Chim. Acta 68, 559–562. 42. Grote, R., Zeeck, A., Stümpfel, J., and Zähner, H. (1990) Pyrrolams, new pyrrolizidinones produced by Streptomyces olivaceus. Liebigs Ann. Chem. 1990, 525–530. 43. Hoff, H., Drautz, H., Fiedler, H.-P., Zähner, H., Schultz, J. E., Keller-Schierlein, W., Philipps, S., Ritzau, M., and Zeeck, A. (1992) Obscurolides, a novel class of phosphodiesterase inhibitors from Streptomyces. I. Production, isolation, structural elucidation and biological activity of obscurolides A1 to A4. J. Antibiot. 45, 1096–1107. 44. Potterat, O., Zähner, H., Metzger, J. W., and Freund, S. (1994) 5-Phenylpentadienoicacid derivatives from Streptomyces sp. Helv. Chim. Acta 77, 569–574. 45. Keller-Schierlein,W., Bahnmüller, U., Dobler, M., Bielecki, J., Stümpfel, J., and Zähner, H. (1986) Isolierung und Strukturaufklärung von Differolid. Helv. Chim. Acta 69, 1833–1836. 46. Müller, A., Keller-Schierlein, W., Bielecki, J., Rak, G., Stümpfel, J., and Zähner, H. (1988) (2S,3R,4R,6R)-2,3,4-Trihydroxy-6-methylcyclohexanon aus zwei ActinomycetenStämmen. Helv. Chim. Acta 69, 1829–1832. 47. Rapp, C., Jung, G., Isselhorst-Scharr, C., and Zähner, H. (1988) A new member of the class of antibiotics with thiotetronic acid structure isolated from Streptomyces olivaceus TÜ 3010. Liebigs Ann. Chem. 1988, 1043–1047. 48. Grote, R., Zeeck, A., Drautz, H., and Zähner, H. (1988) Colabomycins, new antibiotics of the manumycin group from Streptomyces griseoflavus. I. Isolation, characterization and biological properties. J. Antibiot. 41, 1178–1185. 49. Grote, R., Zeeck, A., and Beale Jr., J. M. (1988) Colabomycins, new antibiotics of the manumycin group from Streptomyces griseoflavus. II. Structure of colabomycin A. J. Antibiot. 41, 1186–1195. 50. Burkhardt, K., Fiedler, H.-P., Grabley, S., Thiericke, R., and Zeeck, A. (1996) New cineromycins and musacins obtained by metabolite pattern analysis of Streptomyces griseoviridis (FH-S 1832). I. Taxonomy, fermentation, isolation and biological activity. J. Antibiot. 49, 432–437. 51. Grote, R., Zeeck, A., Drautz, H., and Zähner, H. (1988) 2880-II, a metabolite related to ferulic acid from Streptomyces griseoflavus. J. Antibiot. 41, 1275–1276. 52. Niederer, D., Séquin, U., Drautz, H., and Zähner, H. (1990) (3S,5R,6E,8E)-Deca-6,8dien-1,3,5-triol, ein neues Stoffwechselprodukt von Streptomyces fimbriatus (Millard et Burr 1926). Helv. Chim. Acta 73, 2129–2135. 53. Wessels, P. (1990) Strukturaufklärung neuer Naphthochinon-Antibiotika aus Streptomyces Tü 3556 und Beiträge zum chemischen Screening mit extrem thermophilen Archaebakterien. Doctoral Thesis, Universität Göttingen. 54. Potterat, O., Zähner, H., Volkmann, C., and Zeeck, A. (1993) Exfoliamycin and related metabolites, new naphthoquinone antibiotics from Streptomyces exfoliatus. J. Antibiot. 46, 346–349. 55. Volkmann, C., Zeeck, A., Potterat, O., Zähner, H., Bohnen, F.-M., and Herbst-Irmer, R. (1995) The structures of the exfoliamycins. J. Antibiot. 48, 431–432. 56. Braun, D., Pauli, N., Séquin, U., and Zähner, H. (1995) New butenolides from the photoconductivity screening of Streptomyces antibioticus (Waksman and Woodruff). Waksman and Henrici 1948. FEMS Microbiol. Lett. 126, 37–42. 57. Fiedler, H.-P., Rohr, A., and Zeeck, A. (1986) Elloramycins B, C, D, E and F: minor congeners of the elloramycin producer Streptomyces olivaceus. J. Antibiot. 39, 856–859.
49
2 Screening for New Secondary Metabolites from Microorganisms 58. Fiedler, H.-P. (1986) Identification of new elloramycins, anthracycline-like antibiotics, in biological cultures by high-performance liquid chromatography and diode array detection. J. Chromatogr. 361, 432–436. 59. Rohr, J., Eick, S., Zeeck, A., Reuschenbach, P., Zähner, H., and Fiedler, H.-P. (1988) Tetracenomycins B3 and D3, key intermediates of the elloramycin and tetracenomycin C biosynthesis. J. Antibiot. 41, 1066–1073. 60. Geiger, A., Keller-Schierlein, W., Brandl, M., and Zähner, H. (1988) Phenazines from Streptomyces antibioticus strain TÜ 2706. J. Antibiot. 41, 1542–1551. 61. König, W. A., Hahn, H., Rathmann, R., Hass, W., Keckeisen, A., Hagenmaier, H., Bormann, C., Dehler, W., Kurth, R., and Zähner, H. (1996) Drei neue Aminosäuren aus dem Nikkomycin-Komplex – Strukturaufklärung und Synthese. Liebigs Ann. Chem. 1986, 407–421. 62. Decker, H., Bormann, C., Fiedler, H.-P, Zähner, H., Heitsch, H., and König, W. A. (1989) Isolation of new nikkomycins from Streptomyces tendae. J. Antibiot. 42, 230– 235. 63. Heitsch, H., König, W. A., Decker, H., Bormann, C., Fiedler, H.-P., and Zähner, H. (1989) Structure of the new nikkomycins pseudo-Z and pseudo-J. J. Antibiot. 42, 711–717. 64. Bormann, C., Mattern, S., Schrempf, H., Fiedler, H.-P., and Zähner, H. (1989) Isolation of Streptomyces tendae mutants with an altered nikkomycin spectrum. J. Antibiot. 42, 913–918. 65. Decker, H., Walz, F., Bormann, C., Zähner, H., and Fiedler, H.-P. (1990) Nikkomycins Wz and Wx, new chitin synthetase inhibitors from Streptomyces tendae. J. Antibiot. 43, 43–48. 66. Decker, H., Pfefferle, U., Bormann, C., Zähner, H., Fiedler, H.-P., van Pée, K.-H., Rieck, M., and König, W. A. (1991) Enzymatic bromination of nikkomycin Z. J. Antibiot. 44, 626–634. 67. Schüz, T. C., Fiedler, H.-P., Zähner, H., Rieck, M., and König, W A. (1992) Nikkomycins Sz, Sx, Soz and Sox, new intermediates associated to the nikkomycin biosynthesis. J. Antibiot. 45, 199–206. 68. Decker, H., Zähner, H., Heitsch, H., König, W. A., and Fiedler, H.-P. (1991) Structureactivity relationships of the nikkomycins. J. Gen. Microbiol. 137, 1805–1813. 69. Fiedler, H.-P., Kulik, A., Schüz, T. C., Volkmann, C., and Zeeck, A. (1994) Juglomycin Z, a new naphthoquinone antibiotic from Streptomyces tendae. J. Antibiot. 47, 1166– 1122. 70. Ritzau, M., Philipps, S., Zeeck, A., Hoff, H., and Zähner, H. (1993) Obscurolides, a novel class of phosphodiesterase inhibitors from Streptomyces. II. Minor components belonging to the obscurolide B to D series. J. Antibiot. 46, 1625–1628. 71. Volkmann, C., Hartjen, U., Zeeck, A., and Fiedler, H.-P. (1995) Naphthgeranine F, a minor congener of the naphthgeranine group produced by Streptomyces violaceus. J. Antibiot. 48, 522–524. 72. Blum, S., Fiedler, H.-P., Groth, I., Kempter, C., Stephan, H., Nicholson, G., Metzger, J. W., and Jung, G. (1995) Echinoserine, a new member of the quinoxaline group, produced by Streptomyces tendae. J. Antibiot. 48, 619–625. 73. Blum, S., Groth, I., Rohr, J., and Fiedler, H.-P. (1996) Dioxolides, novel secondary metabolites from Streptomyces tendae. J. Basic Microbiol. 36, 19–25. 74. Fiedler, H.-P., Nega, M., Pfefferle, C., Groth, I., Kempter, C., Stephan, H., and Metzger, J. W. (1996) Kanchanamycins, new polyol macrolide antibiotics produced by Streptomyces olivaceus Tü 4018. I. Taxonomy, fermentation, isolation and biological activities. J. Antibiot. 49, 101–107. 75. Stephan, H., Kempter, C., Metzger, J. W., Jung, G., Potterat, O., Pfefferle, C., and Fiedler, H.-P. (1996) Kanchanamycins, new polyol macrolide antibiotics produced by Streptomyces olivaceus Tü 4018. II. Structure elucidation. J. Antibiot. 49, 109–113.
50
References 76. Pfefferle, C., Kempter, C., Metzger, J. W., and Fiedler, H.-P. (1996) (E)-4-oxonon-2enoic acid, an antibiotically active fatty acid produced by Streptomyces olivaceus Tü 4018. J. Antibiot. 49, 135–137. 77. Pfefferle, C., Breinholt, J., Gürtler, H., and Fiedler, H.-P. (1997) 1-Hydroxy-4-methoxy2-naphthoic acid, a herbicidal compound produced by Streptosporangium cinnabarinum ATCC 31213. J. Antibiot. 50, 1067–1068. 78. Höltzel, A., Kempter, C., Metzger, J. W., Jung, G., Groth, I., Fritz, T., and Fiedler, H.P. (1998) Spirofungin, a new antifungal antibiotic from Streptomyces violaceusniger Tü 4113. J. Antibiot. 51, 487–495. 79. Breinholt, J., Kulik, A., Gürtler, H., and Fiedler, H.-P. (1998) Tigloside: a new tigloylated tetrasaccharide from Amycolatopsis sp. Acta Chem. Scand. 52, 1239–1242. 80. Maier, A., Müller, J., Schneider, P., Fiedler, H.-P., Groth, I., Tayman, F. S. K., Teltschik, F., Günther, C., and Bringmann, G. (1999) (2E,4Z)-Decadienoic acid and (2E,4Z,7Z)decatrienoic acid, two herbicidal metabolites from Streptomyces viridochromogenes Tü 6105. Pesticide Sci. 55, 733–739. 81. Schimana, J., Fiedler, H.-P., Groth, I., Süßmuth, R., Beil, W., Walker, M., and Zeeck, A. (2000) Simocyclinones, novel cytostatic angucyclinone antibiotics produced by Streptomyces antibioticus Tü 6040. I. Taxonomy, fermentation, isolation and biological activities. J. Antibiot. 53, 779–787 82. Walker, M., Zeeck, A., Schimana, J., and Fiedler, H.-P. (2000) Simocyclinones, novel cytostatic angucyclinone antibiotics produced by Streptomyces antibioticus Tü 6040. II. Structure determination and biosynthesis. J. Antibiot., in press. 83. Walker, M., Schimana, J., Süßmuth, R., Beil, W., Fiedler, H.-P., and Zeeck, A. (2000) New simocyclinones of the A-, B-, C- and D-series, novel angucyclinone antibiotics from Streptomyces antibioticus Tü 6040. J. Antibiot., in press. 84. Schimana, J., Walker, M., Zeeck, A., and Fiedler, H.-P. (2000) Simocyclinones: diversity of metabolites is dependent on fermentation conditions. J. Ind. Microbiol. Biotechnol., in press. 85. Pfefferle, C., Breinholt, J., Olsen, C. E., Kroppenstedt, R. M., Gürtler, H., and Fiedler, H.-P. (2000) Kyanomycin, a complex of unusual anthracycline-phospholipid hybrid from Nonomuria species. J. Nat. Prod. 63, 295–298. 86. Kuhn, W. (1993) Untersuchungen zur Produktion der Esmeraldine und der Physiologie von Streptomyces antibioticus. Doctoral Thesis, Universität Tübingen. 87. Fiedler, H.-P., Meiwes, J., Werner, I., Konetschny-Rapp, S., and Jung, G. (1990) Identification of new ferrioxamines by HPLC and diode array detection. J. Chromatogr. 513, 255–262. 88. Meiwes, J., Fiedler, H.-P., Zähner, H., Konetschny-Rapp, S., and Jung, G. (1990) Production of desferrioxamine E and new analogues by directed fermentation and feeding fermentation. Appl. Microbiol. Biotechnol. 32, 505–510. 89. Konetschny-Rapp, S., Jung, G., Raymond, K. N., Meiwes, J., and Zähner, H. (1992) Solution thermodynamics of the ferric complexes of new desferrioxamine siderophores obtained by directed fermentations. J. Am. Chem. Soc. 114, 2224–2230.
51
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin Friedrich Götz* and Günther Jung**
3.1 History of lantibiotics and lantibiotic research in Tübingen
The first lantibiotic structure elucidated was that of nisin [23], and nisin was probably the first lantibiotic identified [65]. Nisin is produced by the cheese starter culture organism Lactococcus lactis subsp. lactis and has an antimicrobial effect on a broad variety of Gram-positive bacteria [23]. The pioneering research of E. Gross and coworkers demonstrated that the peptide antibiotics nisin and subtilin, the latter produced by Bacillus subtilis, actually contain lanthionine (Lan) and 3-methyllanthionine (MeLan) as well as (Dha) and (Dhb), confirming previous hypotheses [6, 21, 22]. The lantibiotic era in Tübingen began in 1985 with the publication of the peptide sequence of epidermin isolated from Staphylococcus epidermidis. With the elucidation of the structure of epidermin, it became clear that it is a heterodet tetracyclic 22-amino acid (2164 Da), amide peptide [1] that contains one residue each of Dhb and MeLan and two residues of Lan. The fourth cyclic structure results from the novel C-terminal mono-carboxy, di-amino acid, AviCys. It was not clear at that time how these amino acid structures arose and how the complex structures of nisin and epidermin are synthesized. Are they non-ribosomally synthesized, as are gramidicin and valinomycin, which are typically synthesized by large multi-enzyme complexes in the cell and for which no structural genes exists [27, 33], or are they synthesized ribosomally as a precursor peptide, which is subsequently posttranslationally modified? In 1987, when Friedrich Götz took over the chair in Mikrobielle Genetik in Tübingen, he and his coworkers started to unravel the biosynthetic principles of
* Mikrobielle Genetik, Universität Tübingen, Waldhäuser Str. 70/8, D-72076 Tübingen ** Institut für Organische Chemie, Universität Tübingen, Auf der Morgenstelle 18, D-72076 Tübingen
52 Microbial Fundamentals of Biotechnology. DFG, Deutsche Forschungsgemeinschaft Copyright # 2001 WILEY-VCH Verlag GmbH, Weinheim ISBN: 3-527-30615-3
3.1 History of lantibiotics and lantibiotic research in Tübingen epidermin. In co-operation with Karl-Dieter Entian’s group, a gene corresponding to the epidermin amino acid sequence was identified on a 54-kb plasmid of the epidermin-producing strain [72]. These pioneering results brought lantibiotic research a big step forward. For the first time it was shown that a lantibiotic is ribosomally synthesized, and clues on the organization of epidermin as a precursor protein and the processing and post-translational modification steps were found. In the following years, the principles of the genetic organization and biosynthesis first found with epidermin were subsequently also found with other lantibiotics – all other lantibiotics studied later were also shown to be ribosomally synthesized. To date, antibiotics have been found in many Gram-positive genera, such as Bacillus, Lactococcus, Pediococcus, Staphylococcus, Streptococcus, and Streptomyces. Lantibiotics have not been found in a Gram-negative species. Subsequent studies of lantibiotics have revealed that they are post-translationally modified at specific positions to give rise to the large number of modified amino acids found in these peptides. In addition, the peptides are produced with a leader peptide, which is removed during maturation, and are transported by specific transport-related proteins out of the cell. Still other lantibiotic-specific proteins are involved in the genetic regulation of biosynthesis and generation of the specific producer-cell self-protection mechanism(s) frequently observed. The name “lantibiotics” was coined in Tübingen and refers to the rapidly expanding group of antibiotic-like peptides that contain the non-protein amino acids lanthionine and 3-methyllanthionine [72]. The discussion whether epidermin and nisin should be regarded as antibiotics or as bacteriocins was intense. The small size and compact structure of the compounds and the wide range of antibacterial activity against most Gram-positive and some Gram-negative bacteria favors consideration as antibiotics – bacteriocins are usually larger proteins with a narrow range of activity, such as colicins. On the other hand, typical amino-acid-derived antibiotics, such as penicillin and gramicidin, are non-ribosomally synthesized. With the knowledge available today, it is clear that lantibiotics fulfill criteria of both, antibiotics and bacteriocins and cannot be placed in one or the other category. The name lantibiotic belies the full extent and complexity of this class of bacterial peptides. A strong and very fruitful collaboration was developed with Hans-Georg Sahl (Bonn) and his coworkers. Since 1983, they had been studying the mode of action of Pep5 and nisin, and later also of gallidermin and other lantibiotics. Earlier studies suggested that nisin could interfere with the biosynthesis of the bacterial cell wall [46, 64]. This appears to be indeed true, as recent studies show [13]. However, the binding to cell wall precursors does not explain the observed bactericidal effect of lantibiotics. Type A lantibiotics form non-specific transmembrane pores in an energy-dependent fashion and allow efflux of preaccumulated intracellular components [66, 69, 73]. In 1984, Friedrich Götz, then at the Technical University in München, isolated an antibiotic compound from a Staphylococcus gallinarum strain, which belongs to a species previously described by Devriese et al. [17]. Preliminary 53
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin studies showed that this antibiotic compound exerts a rather broad activity against Gram-positive bacteria. However, it was not until he moved to Tübingen that the compound was isolated in the group of Hans Zähner and the structure was elucidated in the group of Günther Jung [28]. The compound was named gallidermin after the species name of the producing staphylococcus. Gallidermin is a structural analogue of epidermin, but is slightly more active than epidermin. Today, after long and laborious work to optimize the fermentation and thus increase gallidermin production approximately hundred-fold [29], S. gallinarum Tü 3928 is used for the biotechnological production of gallidermin by the groups of Peter Fiedler (fermentation) and Rolf Werner (downstream processing; Boehringer Ingelheim, Biberach). To say that in the years following 1986, Tübingen became a stronghold in lantibiotic research, by setting the pace of the research, inspiring many other groups, and initiating quite a number of national and international co-operations, is no exaggeration. Three international workshops on lantibiotics have already taken place. Günther Jung (Tübingen) initiated together with HansGeorg Sahl (Bonn) the first lantibiotic workshop, held in April 1991 in Bad Honnef. The contributions were published under the title “Nisin and novel lantibiotics” with Jung and Sahl as editors (Leiden: Escom). The second workshop on “Lantibiotics: a unique group of antibiotic peptides” was organized by Ruud Konigs and Cees Hilbers (Nijmegen, The Netherlands) and held in Arnhem in November 1994. The contributions were published in Antonie van Leeuwenhoek, vol. 69 in 1996. The third workshop on “Lantibiotics and related modified antibiotic peptides” was organized by Friedrich Götz, R. Jack, Günther Jung, and Hans-Georg Sahl and was held in Blaubeuren in April 1998. The ever present international interest in the lantibiotic subject is best documented by the increasing number of applicants and participants from workshop to workshop. A number of potential applications have been found for the mature lantibiotics, including the use as an anti-infective in medical and veterinary areas, in food, beverage and cosmetic preservation, and as regulators of both, human immune function and blood pressure [26, 50, 62, 67]. Lantibiotics represent new lead structures, but their chemical synthesis is too complicated and costly; therefore, the only way to produce large enough quantities for marketing is through biotechnology. In the following sections, more detailed information on lantibiotic research is presented. Since this is a final report on all achievements obtained in the framework of the collaborative research centre 323, we will focus on the results obtained by the groups in Tübingen. However, the picture would be incomplete if major achievements of other groups were not included. The topics on novel structures, mechanism(s) of biosynthesis, genetic organization and regulation, biological activities and mode of actions as well as potential applications for this fascinating, novel class of bacterial-derived, biologically active peptides will be addressed.
54
3.2 Primary structure and proposed maturation of epidermin in staphylococci
3.2 Primary structure and proposed maturation of epidermin in staphylococci
The nucleotide sequence of the epidermin structural gene, epiA, revealed that epidermin is part of a pre-peptide that consists of a 30-amino-acid N-terminal leader peptide and a 22-amino-acid C-terminal propeptide [72]. A comparison of the chemical structure of epidermin [2] with the peptide sequence deduced from epiA indicates that this propeptide undergoes several modifications (Fig. 3.1) before it is transported out of the cell. At each pair of positions where the mature epidermin contains a lanthionine bridge, the precursor peptide contains one serine or threonine and one cysteine. Lanthionine is proposed to be formed in a two-step process involving dehydration of serine and threonine and the subsequent addition of a cysteine thiol group [10, 24].
Figure 3.1: Posttranslational modifications of the epidermin pre-peptide. (A) Model of lanthionine synthesis: dehydration of serine (a) is followed by a nucleophilic addition reaction of a thiol group, thereby forming a thioether bridge (b); the formation of meso-lanthionine from threonine is analogous. (B) Biosynthesis of epidermin: dehydration (a), nucleophilic addition (b), oxidative decarboxylation (c), and removal of the leader peptide (d). Abu, aminobutyric acid; Dhb, dehydrobutyrine.
55
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin
3.3 Genetic organization and regulation of the epidermin genes
3.3.1 Organization and function of epidermin genes The gene cluster for biosynthesis of the lantibiotic epidermin is composed of eleven genes in five transcription units. The genes are involved in 1) synthesis of the epidermin precursor peptide, 2) modification of the precursor peptide, 3) secretion of epidermin, 4) cleavage of the leader peptide, 5) immunity of the producer to epidermin, and 6) regulation of the gene cluster. Four of the transcription units are activated by the DNA-binding regulator EpiQ. The factors or conditions leading to activation of the EpiQ protein or its expression remain unclear. The gene cluster for epidermin biosynthesis, located on plasmid pTü32 of S. epidermidis Tü 3298 [72], contains genes for all activities proposed to be involved in the biosynthetic pathway (Fig. 3.2). Like all lantibiotics, epidermin is ribosomally synthesized and the gene cluster accordingly contains the structural gene epiA, which encodes the precursor peptide [70, 72]. In the same operon, the genes for modification reactions leading to the formation of thioether bridges (epiB and epiC) [4, 41, 58] and to oxidative decarboxylation of the C-terminus (epiD) [42, 43] are encoded. Between epiA and epiB, a weak terminatorlike structure reduces the expression of the subsequent genes [56, 70] whose products are required only in catalytic amounts. In a second transcriptional unit, the extracellular leader peptidase EpiP [20] and the regulator protein EpiQ [56] are encoded. Upstream of epiA, the individually expressed genes epiT and epiH are located [60]. The deduced product of epiT shares sequence similarity with an ABC-type transporter that is involved in secretion of certain peptides and proteins [75]. The coding sequence
Figure 3.2: Organization of the epidermin gene cluster encoded on plasmid pTü32. Epidermin genes are shown as white arrows; flanking genes with putative functions in plasmid replication (rep) and maintenance (resolvase) are indicated in gray. Gene functions: epiA, structural gene; epiB and epiC, dehydration and lanthionine formation; epiD, flavoprotein (oxidative decarboxylation); epiQ, activator; epiP, pro-epidermin processing protease; epiH and epiT, translocation (export) of pro-epidermin; epiFEG, epidermin immunity. Arrows indicate promoters that are activated by EpiQ. The epiT gene is incomplete due to a deletion and frame-shift mutation, but its function can be complemented by the intact gallidermin gene, gdmT.
56
3.3 Genetic organization and regulation of the epidermin genes is disrupted by two frame-shift deletions, and it is very questionable whether epiT has a function in epidermin biosynthesis. Transcriptional reporter gene fusions have demonstrated, however, that epiT is expressed (see below). The homologous gene gdmT from the gene cluster of the closely related lantibiotic gallidermin, has an intact sequence; it mediates an increase of epidermin production in the heterologous host Staphylococcus carnosus [60] and thereby substantiates the proposed capacity of GdmT to secrete epidermin or gallidermin. The gene adjacent to gdmT – gdmH – is also necessary for increased epidermin production; GdmH may be an accessory factor for secretion. EpiH and GdmH are hydrophobic proteins without conspicuous similarities to other proteins; they furthermore exert a limited level of immunity to epidermin [57]. Upstream of epiH, three cotranscribed genes, epiF, epiE, and epiG, are encoded. They mediate resistance (immunity) to epidermin, and their products very likely constitute the subunits of an ABC transporter that expels the harmful epidermin molecules from the cytoplasmic membrane [54, 57]. The epidermin genes are flanked by genes which apparently encode factors for replication and stability of plasmid pTü32 (Fig. 3.2). The res gene product has a high level of identity to plasmid resolvases, whose function is the resolution of cointegrates [58]. The open reading frame upstream of epiP shares significant similarity with plasmid replication genes from various Gram-positive bacteria (M. Hille and A. Peschel, unpublished). The epidermin gene cluster thus seems to be complete; however, several chromosomally encoded genes are also very likely involved in epidermin biosynthesis.
3.3.2 EpiQ is an unusual transcriptional regulator EpiQ shares similarities with DNA-binding proteins of the response regulator family [70] and binds to the epiABCD promoter region to activate expression [56]. Unlike conventional response regulators, EpiQ does not contain the conserved aspartic acid residue that is phosphorylated by a corresponding sensor kinase to activate the regulator [76]. Since a corresponding kinase gene is absent from the epidermin gene cluster, it is very questionable whether EpiQ is phosphorylated at all. Several other lantibiotic gene clusters encode a conventional kinase and regulator pair [75]; therefore, the regulator gene epiQ is unusual. The twocomponent regulatory systems of nisin and subtilin activate the biosynthetic genes in response to the extracellular concentration of the respective lantibiotic [36]. Nisin and subtilin thus act as bacteriocin-like antimicrobial peptides and as quorum-sensing peptide pheromones that control their own expression. Synthesis of epidermin seems to be regulated differently. Regulating agents or conditions have not yet been identified, but EpiQ seems to be active under all laboratory conditions tested so far (A. Peschel, unpublished).
57
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin
3.3.3 Regulation of epidermin genes
Lipase act ivit y ( U/ g cell dry wt .)
Trans-activation experiments with transcriptional reporter gene fusions have demonstrated that EpiQ controls most of the transcriptional units in the epidermin gene cluster; the promoters of the epiFEG, epiH, epiT, and epiABCD gene clusters are activated [56, 58, 60] (Fig. 3.3). All these promoter regions have a palindromic sequence motif (agAaAATTAC – 6 bp – GTAATTtTct) located immediately upstream of the only weakly conserved –35 regions. The motif upstream of the epiABCD promoter is the EpiQ binding site [56]. The EpiQ operator is also present in the gallidermin gene clusters at the same positions, and the promoters are also sensitive to EpiQ. - E p iQ
1000
+ E pi Q 100
10
1
Pep i FE G
Pep i H
Pep i T
Pep i AB C D
Pep i PQ
Figure 3.3: Promoter activities of various epidermin genes. Extracellular lipase activities of S. carnosus strains carrying a promoter test plasmid containing the lipase gene as a reporter under the control of the indicated epidermin gene promoters (Pepi). Lipase expression was determined in the presence and absence of the activator gene, epiQ, encoded on a second plasmid.
The only transcriptional unit not controlled by EpiQ is the epiPQ operon (Fig. 3.2); EpiQ therefore exerts no autoregulation. The epidermin gene cluster contains additional promoters within the operon structures that are not controlled by EpiQ, e. g. in front of epiC, epiD, and epiQ. The epiD promoter has a particularly high activity (Peschel et al., unpublished). The regulation of the epidermin genes remains elusive since no activating agents have been found. Recent literature illustrates the importance of global regulatory relays in Gram-positive bacteria [31]. Quorum-sensing systems control the production of antimicrobial substances in B. subtilis, and it is conceivable that a similar, yet unknown system controls the epidermin genes. A further possibility is the involvement of special sigma factors that couple epidermin production to certain growth phases or environmental conditions. In this respect, it is interesting to note that epidermin production occurs mainly during the exponential growth phase and is obviously switched off in the stationary phase [54]. Since EpiQ con58
3.4 Isolation and characterization of genetically engineered gallidermin trols most of the epidermin genes, the promoters controlling its own expression are likely targets for a chromosomally encoded regulatory system. Chromosomally encoded proteins may also be involved in secretion of epidermin, thereby substituting for the defective epiT, and may be involved in the modification reactions.
3.4 Isolation and characterization of genetically engineered gallidermin and epidermin analogues
Gallidermin (Gdm) and epidermin (Epi) are highly similar. To study the substrate specificity of the modifying enzymes and to find variants of the lantibiotics with altered or new biological activities, we exchanged certain amino acids of the gallidermin and epidermin structural genes, gdmA and epiA, respectively, by sitespecific mutagenesis [53]. No epidermin/gallidermin analogues are found in the supernatant when 1) the hydroxyamino acids involved in thioether amino acid formation are substituted by non-hydroxyamino acids (S3N and S19A); 2) cysteine residues involved in thioether bridging are deleted (C21, C22 and C22); or 3) a ring amino acid is substituted by an amino acid with a completely different character (G10E and Y20G). Production is greatly decreased when serine residues involved in thioether amino acid formation are exchanged by threonine residues (S16T, S19T). A number of conservative exchanges at positions 6, 12, or 14 of the gallidermin backbone are tolerated and lead to analogues with altered biological properties, such as an enhanced antimicrobial activity (L6V) or a remarkable resistance to proteolytic degradation (A12L and Dhb14P). The T14S substitution leads to the simultaneous production of two gallidermin species formed by an incomplete posttranslational modification (dehydration) of the S14 residue. The fully modified Dhb14Dha analogue has antimicrobial activity similar to gallidermin, whereas the Dhb14S analogue is less active. Both peptides are more sensitive to tryptic cleavage than gallidermin. The construction and characterization of the various analogues are described in more detail in the following sections.
3.4.1 Characterization of two Epi – mutants and development of a host-vector system for expression of wild type and mutated gdmA and epiA genes We isolated a series of Epi – mutants of S. epidermidis Tü 3298 by ethyl-methanesulfonate(EMS) mutagenesis; two mutants (EMS5 and 6) carry mutations in the epiA region [4]. Both mutants can be complemented to an Epi+ phenotype by 59
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin transformation with a plasmid carrying epiA, which indicates that other epi biosynthetic genes are functional. Single point mutations in epiA result in an S3N substitution (ring A) in EMS5 and a G10E substitution (ring B) in EMS6 (Fig. 3.1). Analytical HPLC analyses indicated that no epidermin analogue or precursor peptide is present in the culture supernatants of EMS5 and EMS6 [53]. We chose the mutant S. epidermidis EMS6 as an expression host for the synthesis of epidermin and gallidermin analogues. To test this mutant for its ability to synthesize the heterologous gallidermin, a 1.6-kb SalI/EcoRI subfragment (encoding gdmA and a part of gdmB and of gdmT) of the 4.8-kb EcoRI chromosomal fragment of S. gallinarum Tü 3928 [71] was cloned into the polylinker region of the staphylococcal plasmid pT181mcs. S. carnosus TM300 was transformed with the resultant plasmid, pTgdmA [53]. After isolation, the plasmid was transferred into S. epidermidis EMS6 by electroporation. This indirect transformation method is necessary because S. epidermidis Tü 3298 and derivative strains are poorly transformed with ligation products; only ccc plasmids are electroporated efficiently. EMS6 transformants carrying plasmid pTgdmA formed halos on plates containing Micrococcus luteus, and analyses revealed that only fully modified gallidermin is produced.
3.4.2 Characterization of gallidermin and epidermin analogues generated by site-directed mutagenesis The gallidermin and epidermin analogues were isolated from the culture supernatant of various S. epidermidis EMS6 clones, purified, and analyzed by analytical HPLC, electrospray mass spectrometry (ES-MS), and continuous Edman sequencing. MIC values and tryptic sensitivity were determined for all gallidermin analogues, except Gdm L6V/S16T and epidermin S19T because these peptides were only produced in trace amounts. The results are summarized in Table 3.1 [53].
3.4.2.1 Mutations in ring A Since Gdm (L6) is more active than Epi (I6) against various Gram-positive bacteria (23), we generated two more analogues of gallidermin with mutations at position 6: L6V and L6G. The production and modification of these two analogues are not impaired. The L6G analogue is less active than gallidermin, regardless of the indicator strain employed. The L6V analogue is twice as active as gallidermin against Micrococcus luteus and Corynebacterium glutamicum, and just as active against Arthrobacter cristallopoietes and B. subtilis. Antimicrobial activity against S. aureus is reduced fivefold (Table 3.1). The L6G analogue is almost as sensitive to trypsin as gallidermin, whereas the L6V analogue is more resistant. 60
3.4 Isolation and characterization of genetically engineered gallidermin Table 3.1: Antimicrobial activities of gallidermin and its analogues against various Grampositive indicator strains [53]. Gdm + derivatives
Gdm Gdm Dhb14Dha Gdm Dhb14S Gdm Dhb14A Gdm Dhb14P Gdm A12L Gdm L6G Gdm L6V a
Minimal inhibitory concentration (lg/ml)a Micrococcus Arthrobacter luteus cristallopoietes
Corynebacterium glutamicum
Bacillus subtilis
Staphyloc. aureus Cowan I
0.004 0.004 0.008 0.004 0.12 0.15 0.008 0.002
0.065 0.065 0.13 0.065 0.52 0.13 0.13 0.032
3.0 3.0 15.0 >20.0 >20.0 10.0 8.0 3.0
4.0 5.0 25.0 30.0 >30.0 12.0 30.0 20.0
0.005 0.005 0.01 0.02 0.08 0.02 0.04 0.005
MIC values obtained for M. luteus, A. cristallopoietes, and C. glutamicum are located within the nanomolar range (0.002–0.52 mg/ml; *1–240 nM), whereas those observed for B. subtilis and S. aureus are located within the micromolar range (3.0 to 630 mg/ml; 1.4 to 614 mM).
3.4.2.2 Mutations in ring C and D To determine the importance of the C-terminal bicyclic structure (intertwined rings C and D) for epidermin/gallidermin biosynthesis, we exchanged various amino acids in this region. Using the gdmA-L6V gene, we created the double mutant Gdm L6V/S16T. The production of this analogue is only 8% of that observed for Gdm L6V. Molecular mass determinations of Gdm L6V/S16T revealed that the T16 residue is dehydrated. However, it was not possible to verify the formation of the 3-methyllanthionine in ring C by ES-MS. Gdm L6V/S16T has antimicrobial activity; however, the MIC values were not determined because of the very low production of this analogue. The S19T codon of EpiA was replaced in order to create a posttranslationally formed S-(2-aminovinyl)-2-methyl-d-cysteine residue as found in mersacidin. This analogue has antimicrobial activity and is also produced in very low amounts. The production is only 0.4% of that observed for the gallidermin-producing EMS6 (pTgdmA) clone, which produces approximately 12 mg Gdm/l. ES-MS analysis revealed an average molecular mass of 2177 Da, indicating that T19 is dehydrated and that the C-terminus is decarboxylated. It was also not possible to verify the formation of the S-(2-aminovinyl)-2-methyl-d-cysteine residue by ES-MS; however, thioether bridge formation is very likely because of the observed chemical instability of the enethiol structure [42, 43]. To prevent formation of the S-(2-aminovinyl)-d-cysteine residue in ring D, we generated an S19A (pTepiA-S19A) substitution and a C22 deletion (pTepiADC22). These mutations could possibly lead to the formation of alternative thioether bridges. However, the corresponding EMS6 clones form no halos on 61
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin plates of M. luteus, and no epidermin analogue or precursor form was detected by analytical HPLC. In addition no antimicrobial activity and no extracellular product were detected when the last two C residues (pTepiA-DC21, C22) were deleted. An amino acid substitution not directly involved in thioether bridge formation of ring D was created by replacing the bulky Y20 residue with a G (pCUgdmA-Y20G). The corresponding EMS6 clone has no antimicrobial activity, and no gallidermin analogue or precursor peptide was detected by analytical HPLC.
3.4.2.3 Mutations in the flexible middle region (A12 to G15) With the A12L substitution, a more bulky residue was introduced in the flexible middle region. The production of this analogue is comparable to that of gallidermin; however, antimicrobial activity against most of the test strains is reduced threefold (Table 3.1). A striking feature of this analogue is its high resistance to tryptic cleavage. Further amino acid alterations mainly focused on the 2,3-unsaturated Dhb14 residue to investigate whether the reactive C=C double bond is important for biological activity, as suggested for the Dha5 residue of subtilin and nisin [38, 48], or whether this residue plays a role in stabilizing a distinct structure required for pore-forming activity, as reported for the P residue in the channelforming peptides alamethicin and melittin [78]. With the T14S substitution, two analogues were produced in nearly equimolar ratios, as judged from the respective peak areas. By ES-MS and continuous Edman degradation, the two peptides were identified as gallidermin analogues possessing either a dehydrated (Dha14) or an unmodified S14 residue (Fig. 3.1A, B). Antimicrobial activity of Gdm Dhb14Dha is similar to that of gallidermin, whereas the Dhb14S analogue is less active, especially against B. subtilis and S. aureus (Table 3.1). Both analogues were more sensitive to tryptic cleavage than gallidermin; the Dhb14S analogue was one of the most sensitive analogues, being completely cleaved after 30 min. To remove the reactive C=C double bond, a Dhb14A analogue was created. Slightly more of this analogue is produced as compared to gallidermin. Antimicrobial activity of Gdm Dhb14A is similar to that of gallidermin against M. luteus and C. glutamicum and approximately seven-fold less against B. subtilis and S. aureus (Table 3.1). Like Dhb14S, the Dhb14A analogue is more sensitive to tryptic cleavage and is also fully degraded within 30 min. The Dhb14P substitution was expected to cause the strongest conformational change of the middle region. Production of this analogue is comparable to that of gallidermin. The antimicrobial activity is generally reduced 8- to 30-fold (Table 3.1). This analogue has a pronounced resistance to tryptic cleavage.
62
3.4 Isolation and characterization of genetically engineered gallidermin
3.4.3 Overview of the characteristics of gallidermin and epidermin analogues We demonstrated that although epidermin and gallidermin are produced by two different staphylococcal species, gallidermin can be synthesized successfully using the heterologous S. epidermidis EMS6 as expression host and the cloned gdmA gene. The gdmA promoter is controlled by the transcriptional activator EpiQ with an efficiency similar to the epiA promoter [60], and the enzymes involved in epidermin biosynthesis (including maturation and secretion) function efficiently with the gdmA gene product. Mutagenesis of amino acid positions that are directly involved in thioether amino acid formation result in the loss of or a large decrease in production. Loss of production (i. e. no HPLC-detectable extracellular product) is observed with the analogues containing the S19A substitution, the deletion of the C22 residue (ring D, Fig. 3.1), and the S3N substitution (ring A). Thus, it can be hypothesized that biosynthesis and/or secretion of epidermin and gallidermin are severely impaired when formation of only one of the four thioether bridges is prevented. There is also no production observed with the analogues containing the G10E (ring B) and the Y20G (ring D) substitutions, which suggests that these mutations interfere with thioether bridge formation even though the substituted residues are not directly involved. The replacement of the G10 residue for a bulky and negatively charged E residue may impose steric hindrance on thioether bridge formation at ring B. 2D-NMR studies of gallidermin [18] revealed that ring B, which is identical to ring B of nisin and subtilin, adopts a typical b-turn type II conformation. This specific conformation would be disturbed by a G10E exchange and, as a consequence, thioether bridge formation might be affected, regardless whether this reaction is enzyme-catalyzed or occurs spontaneously in a Michael-additionlike reaction [77]. Recent investigations of the substrate specificity of EpiD [43] indicate that a precursor molecule possessing a G residue instead of the Y20 residue is not a substrate of EpiD. This result suggests that at least oxidative decarboxylation of the last C residue of the mutant Y20G precursor peptide is prevented. For all mutations that lead to the loss of production, further investigations will determine whether only secretion or whether also other stages of biosynthesis are blocked. It also cannot entirely be ruled out that partially modified precursor peptides are rapidly degraded by the EMS6 host. The only thioether amino acid mutations that result in the production of antimicrobially active substances are the S19T (ring D) and the S16T (ring C) substitutions. The extremely low production of Epi S19T and Gdm L6V/S16T could be explained by assuming that dehydration of T16 and T19 is inefficient and that only fully modified (dehydrated and subsequently thioether-bridged) precursor peptides are efficiently secreted. The simultaneous production of Gdm Dhb14S and Gdm Dhb14Dha is an indication of different efficiencies of Sand T-dehydration. For the applied use of peptide antibiotics, an advantageous trait would be resistance to proteolytic degradation. Epidermin and gallidermin already natu63
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin rally have a rather high resistance. Among various peptidases tested (pronase, trypsin, pepsin, thermolysin, collagenase, and carboxypeptidases A and B), only pronase and trypsin cleave epidermin [2]. Proteolytic degradation by the other peptidases is sterically hindered by the thioether bridges. In one study, purified gallidermin and analogues were tested for their sensitivity to tryptic cleavage in order to screen for analogues with an increased resistance to proteolytic degradation and, as a consequence, a prolonged period of action. All gallidermin analogues are cleaved by trypsin only in the central part of the molecule between K13 and the residue at position 14, which is in agreement with earlier reports on gallidermin [28] and epidermin [2]. Tryptic cleavage at the second putative tryptic cleavage site, the K4-F5 bond (ring A), is sterically hindered, at least under the conditions employed. 2D-NMR studies of the gallidermin molecule [18, 19] revealed that the rigid rings A/B and C/D are connected by a flexible middle region (A12 to G15). The conformation (e. g. flexibility of the peptide backbone) of this region appears to be important for antimicrobial activity and resistance to tryptic cleavage. Gallidermin analogues with a supposed decreased flexibility of the middle region, such as Gdm Dhb14P and Gdm A12L, are highly resistant to tryptic cleavage, whereas their antimicrobial activity against various Gram-positive indicator bacteria is greatly reduced (Table 3.1). According to modeling studies of the gallidermin/trypsin interaction [26], the gallidermin molecule must adopt a bent structure in order to fit into the catalytic cleft of the trypsin molecule. This bent structure is supported by the flexibility of the peptide backbone within the middle region. Thus, the high resistance to tryptic cleavage observed for Gdm Dhb14P and Gdm A12L may reflect a decreased flexibility of the middle region, impeding the formation of the bent structure. A P residue following a K residue usually has a strong negative influence on trypsin action. This also may explain the high resistance to tryptic cleavage observed for Gdm Dhb14P (tryptic cleavage site: K13–P14). However, a negative influence of a preceding L residue is not known [28], which suggests that the large increase in resistance to tryptic cleavage observed for Gdm A12L (tryptic cleavage site identical to that of gallidermin) is mainly caused by a restricted flexibility of the middle region. The K13–Dhb14 bond of gallidermin is cleaved by trypsin with a strikingly lower efficiency than bonds of K to normal protein amino acids [28]. Thus, an increased substrate affinity of the trypsin molecule and an increased flexibility of the peptide backbone within the middle region, which may lead to a greater accessibility of the tryptic cleavage site, may be responsible for the greatly reduced resistance to tryptic cleavage observed for Gdm Dhb14A and Gdm Dhb14S (tryptic cleavage sites: K13–A14 and K13–S14, respectively). This effect was strikingly less enhanced for Gdm Dhb14Dha (tryptic cleavage site: K13– Dha14). Interestingly, the tryptic sensitivity of Gdm L6V and Gdm L6G differ, although these gallidermin analogues are not altered in the vicinity of the tryptic cleavage site; this observation possibly indicates a slight overall change in conformation of the L6V and the L6G analogue. 64
3.4 Isolation and characterization of genetically engineered gallidermin Of all the gallidermin derivatives altered within the middle region, only the Dhb14Dha analogue has antimicrobial activity similar to that of gallidermin (Table 3.1). This may indicate the importance of the 2,3-didehydroamino acids for maintaining a structure required for efficient pore formation. A critical role of the polypeptide backbone flexibility for biological activity has also been suggested for the channel-forming peptides alamethicin and melittin [78]. With Gdm Dhb14A, a decrease in antimicrobial activity is only observed for three of five indicator strains (Table 3.1). Thus, one can speculate that gallidermin can exert its antimicrobial activity by different mechanisms: a Dhb/Dha-dependent and a Dhb/Dha-independent mechanism. Subtilin exerts its antimicrobial activity against Bacillus by two different mechanisms. The inhibitory effect on Bacillus spore germination is dependent on the Dha5 residue and is most likely caused by a Michael-type reaction of the didehydro residue with nucleophilic membrane sulfhydryl groups [48]. The ability to lyse vegetative Bacillus cells, however, is independent of the Dha5 residue [47]. Thus, our results [53] and the results obtained previously by mutagenesis of other type A lantibiotics strongly indicate that the molecular mechanisms by which these peptides exert their antimicrobial activity may differ and that a general function of the 2,3-didehydroamino acids for biological activity is not yet known. The various test strains varied greatly in their susceptibility to gallidermin and its analogues (Table 3.1), possibly because of differences in the membrane phospholipid composition [25, 35] or different membrane potentials [34, 68]. Both factors are crucial for the formation of voltage-dependent transmembrane pores. Furthermore, the composition of the bacterial cell wall might influence the kinetics of pore formation. Cationic lantibiotics interact with polyanions (e. g. teichoic acids) of the cell wall [11] and with membrane-bound cell wall precursor molecules [64]. Binding of lantibiotics to these components may influence the kinetics of pore formation, and may also be responsible for secondary killing mechanisms observed for Pep5 and nisin, such as activation of autolytic enzymes and inhibition of murein synthesis [64]. Recently it was indeed demonstrated that an increased negatively charged cell wall leads to an increased sensitivity to gallidermin and other positively charged antimicrobial peptides [59]. Since B. subtilis and S. aureus produce a variety of exoproteases, the strikingly lower susceptibility of these test strains (Table 3.1) might be partially due to proteolytic degradation of gallidermin and its analogues. In summary, we successfully engineered gallidermin and epidermin analogues with increased antimicrobial activities and/or resistance to proteolytic degradation. Furthermore, we obtained valuable information about the structural elements required for proper biosynthesis of epidermin and gallidermin. NMR analyses of the generated analogues and black lipid membrane studies may provide further information on the structure/activity relationship of these lantibiotics and may pave the way for target-oriented peptide engineering.
65
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin
3.5 Function of the epidermin immunity genes epiFEG
Epidermin, gallidermin, Pep5, subtilin, and nisin are the best-studied members of type A group of lantibiotics, which are all synthesized by Gram-positive bacteria. Type A lantibiotics act by forming membrane-potential-dependent pores in the cytoplasmic membrane of bacteria [5, 68, 69]. Because of this activity, self-protection against the lantibiotic is of vital importance to the producing organism. Several genes responsible for producer self-protection have been found in the gene clusters of the respective systems: 1) nisI and spaI, which code for lipoproteins [32, 37]; 2) epiF, epiE and epiG [57] which are homologous to nisF, nisE, and nisG of the nisin system [74] and to spaF and spaG of the subtilin system [32], all of which code for ABC transporters; and 3) pepI, which does not share any sequence similarity to any known genes [63]. Analysis of the amino acid sequences suggests that the EpiFEG-type transporters and the lipoproteins NisI and SpaI are membrane-located; this has been experimentally shown for PepI [63]. The mechanism of action has not been elucidated for any of the gene products. The sequence similarity of the EpiFEG proteins to that of ABC transporters led to the proposal that the EpiFEG-type proteins act by transporting the lantibiotic out of the cytoplasmic membrane, thus keeping its concentration below a critical level and preventing pore formation. Two conceivable directions of transport have been discussed: import into the cell for inactivation by proteolytic cleavage and export into the surrounding medium. As shown by heterologous expression experiments in S. carnosus, the selfprotection (immunity) of the epidermin-producing strain S. epidermidis Tü 3298 against the pore-forming lantibiotic epidermin is mediated by an ABC transporter composed of the EpiF, EpiE, and EpiG proteins. We developed a sensitive assay based on HPLC analysis of the substrate gallidermin in cell supernatants to investigate the mechanism of the EpiFEG transporter. Our results indicate that the EpiFEG transporter works by expelling the lantibiotic from the cytoplasmic membrane into the surrounding medium with a high substrate specificity. Thus, the EpiFEG transporter functions according to the “hydrophobic vacuum cleaner” mechanism [12]. Furthermore, we showed that the gallidermin derivative L6G has an EpiE-dependent enhanced activity. These results will be covered in more detail in the following sections. To our knowledge, the EpiFEG transporter is the first of its kind to be investigated that has a pore-forming peptide as substrate. Future research on the EpiFEG transporter will focus on the question whether the EpiFEG transporter can substitute in S. epidermidis Tü 3298 for the non-functional EpiT transporter, whose task is to export completely modified pre-epidermin (with the leader peptide still present) out of the producer cell.
66
3.5 Function of the epidermin immunity genes epiFEG
3.5.1 The epiFEG genes confer immunity to the epidermin producer We characterized a DNA region located upstream of the structural gene epiA that mediates immunity and increased epidermin production. The sequence of a 2.6-kb DNA fragment revealed three ORFs (Fig. 3.2), epiF, E, and G, which form an operon [57]. In the cloning host S. carnosus, the three genes mediate an increased tolerance to epidermin, and the highest level of immunity (sevenfold) is achieved with S. carnosus carrying epiFEG and epiQ (Fig. 3.4). The promoter of the first gene epiF responds to the activator protein EpiQ and contains a palindromic sequence similar to the EpiQ-binding site of the epiA promoter, which is also activated by EpiQ. Inactivation of epiF, E, or G results in the complete loss of the immunity phenotype. An epidermin-sensitive S. epidermidis Tü 3298 mutant is complemented by a DNA fragment containing all three genes. When the epiFEG genes are cloned together with plasmid pTepi14, which contains the biosynthetic genes epiABCDQP, the production of epidermin is approximately fivefold higher. The deduced amino acid sequences of EpiF, E, and G are similar in sequence and proposed structure to the components of various ABC transporter systems. EpiF is a hydrophilic protein with conserved ATP-binding sites, while EpiE and G have six alternating hydrophobic regions and very likely constitute the integral membrane domains. When EpiF is overproduced in S. carnosus, it is at least partially associated with the cytoplasmic membrane.
Figure 3.4: The epiFEG genes confer immunity against gallidermin in the heterologous host S. carnosus. MIC values were determined with S. carnosus TM300 harboring all or only two of the epiFEG genes (in all possible combinations) on plasmid pRB473. Black bars represent MIC values of strains harboring in addition the positive regulator of the epidermin biosynthetic system, epiQ, on plasmid pTepiQ10.
67
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin
3.5.2 Transport mechanism In most cases, it has been assumed that ABC transporters reported to be involved in resistance to non-pore-forming antibiotics have to export the antibiotic from the cytoplasm to the surrounding medium. For an antibiotic whose target is located in the cytoplasm, this mechanism seems self-evident, whereas for a pore former whose target is the cytoplasmic membrane, import into the cell as well as export to the surrounding medium are both conceivable mechanisms for removing the antibiotic from its target (Fig. 3.5).
Figure 3.5: Two alternative models for the mechanism of the EpiFEG transporter. The model shows the EpiE and EpiG proteins situated within the cytoplasmic membrane and an EpiF dimer bound to the complex at the cytoplasmic side of the membrane. EpiF harbors the ATPase binding site, as predicted by hydropathy plots and sequence alignments (Peschel and Götz 1996). Black arrows illustrate the mechanism model 1: the export of gallidermin into the surrounding medium. Every exported molecule in model 1 is likely to be able to reintegrate into the membrane, thereby causing an incessant expulsion process. Gray arrows illustrate model 2: the import of gallidermin into the cytoplasm, where one would expect proteolytic degradation of the lantibiotic.
To investigate the direction of transport mediated by the EpiFEG transporter, it was not possible to use membrane vesicle systems because the cytoplasmic orientation of the ATPase site in the EpiF part of the transporter constitutes the only means to discriminate between right-side-out and inside-out vesicles and because the substrate gallidermin is known to cause efflux of small molecules such as ATP out of the cell. This would, unfortunately, lead to an equal 68
3.5 Function of the epidermin immunity genes epiFEG distribution of ATP in the internal and external fluid. We therefore developed an assay in which the amount of the substrate gallidermin remaining in the supernatant of whole cells incubated with gallidermin was determined. In this transporter assay, a concentration of gallidermin in the supernatant of cells expressing the epiFEG genes higher than that of a control strain would suggest that the EpiFEG transporter works by export; a lower concentration in the supernatant would suggest that the transporter works by import. All experiments were performed with S. carnosus TM300 as a host for heterologous expression of all or some of the epiFEG genes. Gallidermin was quantified by HPLC detection. Our results indicate that model 1 shown in Fig. 3.5 represents the mechanism used by the EpiFEG transporter. It is not known whether the transporter recognizes the substrate in its monomeric or oligomeric form; both possibilities are illustrated in Fig. 3.5. Optimal results were observed with rather low gallidermin concentrations, and HPLC quantification was optimized and resulted in a detection limit of about 20 ng at the very specific wavelength of 266 nm (absorption of C-terminal aminovinylcysteine in gallidermin and all derivatives). At 2 mg gallidermin/ml, the extracellular gallidermin content is about fourfold higher for the epiFEG-expressing strain than for the control strain. At higher and lower concentrations of applied gallidermin, the difference is less pronounced. To exclude the possibility that the observed difference in extracellular gallidermin concentration is caused by non-specific interaction of gallidermin with one of the EpiFEG proteins and not by the functional transporter, the epiFEGexpressing strain was compared with strains harboring the genes encoding only two of the three transporter components (epiF, epiE and epiG) in all three possible combinations. The gallidermin concentration in the assay supernatant of each of the three strains showed a similar basal level, whereas in the epiFEGexpressing strain, this concentration was about fourfold higher, demonstrating that the observed effect is not caused by non-specific interaction with one of the protein components of the EpiFEG transporter (Fig. 3.6).
3.5.2.1 Energy dependence of transport Energy dependence of the EpiFEG transporter could not be directly studied using ATPase inhibitors because destruction of the membrane potential would also result in inhibition of gallidermin activity. Instead, dependence of the EpiFEG-mediated transport on the glucose concentration was determined. Glucose was the only energy source in the incubation buffer. In the absence of glucose, transport is strongly reduced. Increases in the concentration of glucose up to 1% result in increases in the transport efficiency; at concentrations of glucose higher than 1%, transport efficiency does not increase further. These results are in accordance with the expected energy dependence of the ATP-consuming EpiFEG transporter.
69
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin
Figure 3.6: Comparison of the effect of the entire EpiFEG transporter with that of the transporter protein components. The transporter assay was carried out with S. carnosus strains harboring plasmid pTepiQ10, which contains the positive regulator of the epidermin system, and two (“FE”, “EG”, “FG”), none of (“473”), or all of (“FEG”) the epiF, epiE and epiG genes on plasmid pRB473. Gallidermin was used as substrate at a concentration of 2 mg/ml. The values of the determined extracellular gallidermin content from five samples for each strain were averaged.
3.5.2.2 Specificity of the transporter Eight gallidermin derivatives with single amino acid substitutions were used to determine the specificity of the EpiFEG transporter in the transporter assay. In addition to epidermin (gallidermin L6I), seven other derivatives originating from site-specific mutagenesis of the gallidermin structural gene gdmA [53] were investigated. Transporter efficacy was defined as the value obtained with the epiFEG-expressing strain divided by the value obtained with the control strain. This factor is in the range of 3 to 5 for each substance, with the exception of gallidermin Dhb14P and gallidermin L6G, where there is almost no detectable effect. No differences among the test strains were detected with nisin, which suggests that nisin is not a substrate of the EpiFEG transporter. The most prominent feature of gallidermin Dhb14P is that the flexibility of the so-called hinge region is strongly decreased; gallidermin L6G was the most hydrophilic derivative investigated (as concluded from reversed-phase HPLC retention). The three-step model for the mode of action of nisin and type A lantibiotics in general includes adhesion to the outer surface of the cell membrane mediated by electrostatic interaction between the cationic peptide and anionic charges on the surface of the cytoplasmic membrane, integration into the membrane with a DC or DpH present where hydrophobicity and flexibility are considered to be important, and formation of oligomeric pores [51, 52]. Thus, gallidermin Dhb14P and gallidermin L6G appear to be impaired in the integration into the mem70
3.5 Function of the epidermin immunity genes epiFEG brane, and this is presumably also the cause for their low bactericidal activity. Adhesion to the cytoplasmic membrane is not likely to be affected in any of the gallidermin derivatives tested because the charge of the molecule is not changed. Since specifically gallidermin Dhb14P and gallidermin L6G are by far the poorest substrates among all gallidermin derivatives investigated, we assume that integration of the substrate into the membrane is important for EpiFEG activity. This suggests that the substrate binding site of the EpiFEG transporter is within the membrane-spanning part of the protein complex, an assumption that is further supported by the interaction of gallidermin L6G with the internal membrane protein EpiE (see below). MIC values of gallidermin, epidermin, gallidermin derivatives, and nisin of S. carnosus (pRBepiFEG/pTepiQ10) and S. carnosus (pRB473/pTepiQ10) were also determined. No difference in the MIC values of nisin among the test strains was detected. Gallidermin, epidermin, and the gallidermin derivatives Dhb14S, Dhb14Dha, and A12L are more active against the control strain, whereas the activity of gallidermin Dhb14P is only slightly influenced by the presence of the epiFEG genes in the test strain. In contrast, the MICs of gallidermin derivatives L6V and Dhb14A are not influenced by the presence of the epiFEG genes in the strain; however, they are good substrates of the EpiFEG transporter, as shown in the transporter assay. While conditions in the transporter assay were selected to optimize EpiFEG transporter efficacy, this was not the case in the complex medium used for MIC determinations. Conditions in the complex medium seemed to suppress the interaction of EpiFEG with some derivatives.
3.5.2.3 Interaction of EpiE with gallidermin L6G Paradoxically, the activity of gallidermin L6G is clearly higher in the presence of the epiFEG genes in the test strain. Further experiments were designed to investigate this phenomenon. The MIC of gallidermin L6G of S. carnosus (pRBepiFE/pTepiQ10), S. carnosus (pRBepiEG/pTepiQ10), S. carnosus (pRBepiFG/ pTepiQ10), and the strains used before was determined (Table 3.2). The activity of gallidermin L6G is higher whenever the epiE gene is expressed in the strain; the expression of the other two genes (epiF and epiG) is not necessary for this effect. The activity of gallidermin L6G is also higher against S. carnosus (pTXepiE), in which the epiE gene is under the regulation of a xylose-inducible promoter, than against the control strain S. carnosus (pTX16). The integration of gallidermin L6G into the cytoplasmic membrane is usually impaired because of its comparatively low hydrophobicity, yet the activity of gallidermin L6G against strains that expressed the epiE gene was always relatively high (Table 3.2). As an explanation for this phenomenon, we propose a direct protein-protein interaction of gallidermin L6G with EpiE; this interaction would also suggest a participation of EpiE in substrate binding in general. Gallidermin L6G probably binds to the substrate binding site on EpiE and remains bound there, forming a nucleus to which other gallidermin L6G molecules adhere to form a pore. Without this interaction, gallidermin L6G would be too hydrophilic to form a pore. 71
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin Table 3.2: Activity of gallidermin L6G [53]. Strain
MIC a (mg/ml)
S. carnosus (pRB473/pTepiQ10) S. carnosus (pRBepiFEG/pTepiQ10) S. carnosus (pRBepiFE/pTepiQ10) S. carnosus (pRBepiEG/pTepiQ10) S. carnosus (pRBepiFG/pTepiQ10) S. carnosus pTX16 S. carnosus pTXepiE
1.0 0.1 0.1 0.1 1.0 1.0 0.2
a
MIC values of gallidermin L6G (Ottenwälder et al. 1995) of various strains were determined as described in the experimental procedures to investigate the gallidermin L6GEpiE interaction. Strains S. carnosus pTXepiE, in which epiE is under the control of a xylose-inducible promoter, and S. carnosus pTX16 as a control were grown in basic medium without glucose and with 0.5% xylose.
3.6 Inactivation and characterization of the epidermin leader peptidase EpiP
The sequences of lantibiotic leader peptides differ from Sec-dependent protein export signal sequences. Type A lantibiotics have a negatively charged, amphiphilic a-helix that is removed at the characteristic processing-site P–2–Q/R–1 ;X+1 (Fig. 3.7). In nearly all lantibiotic gene clusters, a serine protease is encoded, which is proposed to mediate precursor processing. These proteases are usually located in the cytoplasm. In contrast, NisP, the nisin leader peptidase, and EpiP contain a signal sequence, indicating that they act extracellularly. Therefore, the location of the processing of the various lantibiotics may differ [75]. The serine protease EpiP from S. epidermidis Tü 3298 catalyzes the extracellular processing of the epidermin precursor peptide, as shown in experiments where epiP in the xylose-regulated expression vector pCX15 in S. carnosus is
Figure 3.7: The processing site of modified pre-epidermin. The arrow indicates the processing site of the secreted EpiP protease.
72
3.6 Inactivation and characterization of the epidermin leader peptidase EpiP overexpressed. The cleavage of the unmodified EpiA precursor peptide to leader peptide and pro-epidermin by EpiP-containing culture filtrate of S. carnosus (pCX15epiP) was followed by reversed-phase chromatography and subsequent electrospray mass spectrometry [20]. The epidermin leader peptidase gene epiP was inactivated so that the final intermediates of epidermin biosynthesis could be isolated to characterize the mechanism of precursor processing. The isolated precursor peptides may also serve as natural substrates for EpiP in future experiments.
3.6.1 epiP gene replacement in S. epidermidis Tü 3298 To construct a plasmid for the replacement of epiP, 1 kb upstream and 1 kb downstream of the gene were amplified by PCR and sequenced. The two amplified fragments were ligated into pBT2, a temperature-sensitive shuttle vector [14], flanking an erythromycin resistance cassette. The resulting plasmid was introduced into S. epidermidis Tü 3298 by electroporation. A homologous recombination event deleted the entire epiP gene, leaving the epiP promoter intact to ensure expression of the regulator epiQ [56] downstream of epiP (Fig. 3.2); there is no terminator structure in the erythromycin resistance cassette that could inhibit transcription of epiQ. The antimicrobial activity of S. epidermidis Tü 3298DepiP mutant strains on agar plates containing the epidermin-sensitive M. luteus was strongly decreased. By transforming the mutant with an epiP-expressing plasmid, epidermin production could be reconstituted (Kies and Götz, unpublished).
3.6.2 Detection and isolation of epidermin precursor peptides Epidermin and epidermin precursor peptides were purified by reversed-phase chromatography of the supernatant from S. epidermidis Tü 3298DepiP grown in synthetic medium. A silver-stained nitrocellulose blot of an SDS-polyacrylamide gel with fractions from the reversed-phase chromatography showed peptides with an apparent molecular weight similar to that of mature epidermin as well as peptides with an apparent molecular weight similar to that of the N-terminally cleaved precursor peptides isolated from the S. epidermidis Tü 3298 wild type strain grown in defined medium. These preliminary results indicate that in S. epidermidis Tü 3298DepiP, the same or at least a similar processing at the Nterminus occurs as in the S. epidermidis Tü 3298 wild type strain. To test this hypothesis, the precursor peptides isolated from S. epidermidis Tü 3298DepiP were analyzed by mass spectrometry and Edman degradation. 73
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin From S. carnosus (pTepiABCDQ) grown in synthetic medium, completely modified epidermin precursors could be isolated that were N-terminally processed at different positions within the leader peptide, i. e. before amino acids 17, 19, or 20 (Fig. 3.7). Mature epidermin was also produced. The peptides were characterized by Edman degradation and mass spectrometry. The same peptide mixture was detected in S. epidermidis TÜ 3298 (wild type strain) when grown in synthetic medium, where EpiP shows a significant reduction in activity because of the low buffer capacity of this medium, which becomes rather acidic. The precursor peptides isolated are not antimicrobially active against the epidermin-sensitive M. luteus. Treatment with endoprotease ArgC, a protease that simulates the EpiP reaction in vitro, leads to antimicrobially active, mature epidermin, as shown by mass spectrometry and in the M. luteus bioassay. In conclusion, the serine protease EpiP catalyzes the last modification step in epidermin biosynthesis – the processing of the epidermin precursor peptide. Due to the inactivation of epiP in S. epidermidis Tü 3298DepiP, it was possible to detect the final intermediate of epidermin biosynthesis, i. e. the completely modified precursor peptide. The precursor had undergone proteolytic cleavage within the leader peptide. Similar results have been observed for the S. epidermidis Tü 3298 wild type strain in synthetic medium, where EpiP shows decreased activity, and for S. carnosus (pTepiABCDQ), which indicate a two-step processing of the epidermin precursor peptide.
3.7 The flavoenzyme EpiD and formation of peptidylaminoenethiolates
One of the goals of lantibiotic research is the analysis of the enzymatic mechanisms involved in the modification process from pro-epidermin to pre-epidermin. The key function is certainly associated with the EpiB and EpiC enzymes, which are apparently involved in dehydration of serine and threonine residues and in the thioether bridge formation. Homologous protein sequences that occur in all other lantibiotic genes are referred to as LanB and LanC. Several groups are studying the enzymatic mechanism of these two enzymes; however, to date, no clean in vitro reaction has been achieved. It was speculated that the enzymatic reaction is strictly oxygen sensitive and/or that unusual co-factors are involved. Epidermin and gallidermin modification involves a third enzyme, EpiD, whose enzymatic function was extensively studied by Thomas Kupke (see below). An epiD homologue has so far only been found in the mersacidin biosynthesis gene cluster [9], but not the nisin, pep5, or subtilisin gene clusters.
74
3.7 The flavoenzyme EpiD and formation of peptidyl-aminoenethiolates
3.7.1 Preparation of precursor peptide EpiA To investigate the enzymes involved in epidermin biosynthesis, it is necessary to produce sufficient amounts of pre-epidermin (EpiA) as a substrate and to design methods to detect EpiA. EpiA was therefore expressed in Escherichia coli using a malE-epiA fusion. The malE gene encodes the maltose-binding protein MBP. The fusion protein MBP-EpiA was expressed from pIH902-epiA and purified in one step by amylose affinity chromatography. EpiA was cleaved from MBP-EpiA by factor Xa, and the identity of purified EpiA was confirmed by ES-MS and amino acid sequencing. Upon prolonged incubation, factor Xa not only cleaves EpiA from the fusion protein, but also less efficiently cleaves EpiA internally between R–1 and I+1. The internal factor Xa cleavage site of EpiA was masked by altering the sequence -A–4-E-P-R–1- to -A–4-E-P-Q–1- by site-directed mutagenesis. Since the mutant peptide EpiA R–1Q) has properties different than EpiA (the molecular mass is 28 Da and the isoelectric point is approximately 1.5 pH units lower), it is used as a control peptide in incubation experiments. Anti-EpiA antisera were raised to detect EpiA [44].
3.7.2 Construction and purification of MBP-EpiD fusion proteins According to the nucleotide sequence, epiD encodes a 181-amino acid protein. In order to obtain more information on the function and biochemical properties of EpiD, epiD was expressed in E. coli using the maltose-binding protein (MBP) fusion system. The soluble 67-kDa MBP-EpiD fusion protein was purified in one step by amylose affinity chromatography and was used to raise polyclonal antibodies directed against EpiD. The MBP-EpiD fusion protein was yellow and identified as a flavoprotein by its absorption spectrum with maxima at 277, 378, and 449 nm (compare Fig. 3.8). The coenzyme is very tightly, but not covalently attached. It could only be removed by TCA extraction of the fusion protein and was identified by thin-layer chromatography (TLC) as FMN. S. epidermidis Tü 3298/EMS11 [4] carries a mutation within epiD, designated epiD*, which was PCR-amplified using purified pTü32 from the mutant as a template. This DNA fragment was inserted in the StuI site of the pIH902 polylinker. Plasmids were isolated from four E. coli TB1 (pIH902-epiD*) clones, and the entire epiD* region was sequenced. The epiD* start codon immediately follows the factor Xa cleavage sequence and all four isolated plasmids have a point mutation in codon 93, substituting the wild type GGT (Gly) with GAT (Asp) in epiD*. This G–A transition concurs with the use of EMS as the mutagenizing agent. MBP-EpiD* is not yellow and does not exhibit the characteristic absorption maxima of MBP-EpiD when similar concentrations of fusion proteins are used; it was necessary to dilute MBP-EpiD because of the low amount of soluble MBP75
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin
Figure 3.8:
Absorption spectrum of purified EpiD.
EpiD*. MBP-EpiD* does not bind FMN. Since MBP-EpiD* differs from MBPEpiD only in the substitution of Gly93 by Asp, Gly93 may be important for FMN binding [45].
3.7.3 Characterization of purified EpiD EpiD was expressed using the T7 RNA polymerase promoter system in E. coli and was purified to homogeneity in three column chromatography steps: DEAE-Sepharose Fast Flow, Mono-Q, and Phenylsuperose chromatography. The first 18 N-terminal amino acids of purified EpiD were determined by Edman degradation. The amino acid sequence correlates exactly with the amino acids (Met-Tyr-Gly-Lys-…) deduced from the epiD sequence [70]. Purified EpiD is yellow. The absorption spectrum exhibits maxima at 274, 382 and 453 nm, which are characteristic for flavoproteins in the oxidized state (Fig. 3.8). The liberated flavin component was analyzed by TLC and ES-MS. The mass of the flavin coenzyme was determined to be 455.5 Da (457 Da in a second experiment), which closely agrees with the theoretical value of 456.3 Da and confirms that the flavin component is FMN and not FAD (theoretical mass 785.6 Da). The average molecular mass of EpiD was determined to be 20,827 +/– 5 Da, which is in close agreement with the theoretical value of 20,825 Da calculated for the amino acid sequence derived from the nucleotide sequence, indicating that EpiD is not covalently modified.
76
3.7 The flavoenzyme EpiD and formation of peptidyl-aminoenethiolates
3.7.4 Model of oxidative decarboxylation of peptides Since flavin coenzymes are normally involved in oxidation-reduction reactions, EpiD belongs to the class of oxidoreductases. There is only one obvious reaction involved in epidermin biosynthesis that requires an oxidoreductase activity: The synthesis of S-((Z)-2-aminovinyl)-D-cysteine [1] includes the removal of two reducing equivalents from a -C–C- group to form a -C=C- group. Regarding the proposed order of the modification reactions [72], it was assumed that the four lanthionine rings are formed first, followed by the oxidation of the C-terminal lanthionine by EpiD and the subsequent decarboxylation reaction. In the course of the oxidation, EpiD-FMN becomes reduced. According to this first published hypothesis, EpiD catalyzes the final step in the modification of pre-epidermin, and its reaction product is then processed by the leader peptidase, forming mature epidermin [45]. Thioether formation is only possible if dehydrated serine and threonine residues are present, but in principle, oxidative decarboxylation could be independent of the other modification reactions. Therefore, the following alternative model of the C-terminal oxidative decarboxylation of peptides has been suggested (Fig. 3.9): in the first reaction, EpiD catalyzes the removal of two reducing equivalents from the C-terminal cysteine residue of unmodified EpiA. A double bond is formed, and FMN is reduced to FMNH2. The C-terminal carboxyl group is then removed, catalyzed either by EpiD or occurring spontaneously [42]. The occurrence of a reaction between EpiD and the unmodified precursor peptide EpiA was investigated.
O
EpiD-FMN
HN CH C OH
CO2
EpiD-FMNH2
O HN C C OH
HN C
CH2
CH
CH
SH
SH
SH
H
Figure 3.9: Model of the C-terminal oxidative decarboxylation of unmodified precursor peptide EpiA and the role of flavoprotein EpiD. The C-terminal cysteine residue of the precursor peptide EpiA is shown. After oxidation and decarboxylation, the peptidyl-aminoenethiols are formed.
3.7.5 Interaction between EpiD and EpiA EpiA or EpiAR–1Q was coupled to an NHS-activated HiTrap column in order to purify the enzymes involved in epidermin biosynthesis by affinity chromatography. Binding studies were carried out with purified EpiD or extracts of induced E. coli K38 (pGP1–2, pT7–5epiD) cells. In both cases, the migration of EpiD is re77
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin tarded only under reducing conditions. The applied EpiD focuses as a sharp yellow band on the EpiA-HiTrap column. Other proteins of the cell extract are not retarded, which indicates the specificity of the interaction. Even under reducing conditions, the interaction is weak, and EpiD already elutes with the loading buffer containing no salt. In a control experiment, a column without any coupled peptide was used, and as expected, the migration of EpiD was not retarded [40, 43].
3.7.6 Reaction of EpiD with EpiA and pro-epidermin Under reducing conditions, purified EpiD reacts with unmodified precursor peptide EpiA. Several products were identified after separation of the reaction mixture by reversed-phase chromatography and ES-MS analysis. The product formed depends on the incubation time. The first product formed, product 3 (the products were designated according to increasing hydrophobicity), is more hydrophobic than the unmodified peptide EpiA and has a molecular mass of 5,579–5,585 Da (values obtained from several experiments), i. e. 45–48 Da less than EpiA. In order to determine the mass difference between EpiA and product 3 with greater accuracy, the peptides were measured together, giving a mass difference of 46.5 Da. Furthermore, product 3 has an absorbance at 260 and 280 nm in 0.1% TFA/H2O/acetonitrile, higher than that of EpiA [42]. This enzyme reaction was also detected using crude cell extracts of S. carnosus (pTepi14) and EpiA [40]. This clone has been used for heterologous expression of epidermin [3, 70]. With longer incubation times, two additional, less hydrophobic peptides with molecular masses of 5,524–5,528 Da (product 1; 102–105 Da less than EpiA) and 5,579–5,585 Da (product 2; 45–48 Da less than EpiA) were identified. Product 3 (oxidatively decarboxylated peptide) is unstable and is non-enzymatically converted to products 1 and 2. Product 2 has the same mass as product 3, but its absorbance at 260 and 280 nm is comparable to that of unmodified peptide EpiA. Based on its molecular mass, product 1 is either the peptide EpiA(M1-C51) (lacking the last cysteine residue; 103.2 Da less) or the peptide EpiA(M1-C51)-NH2 (104.1 Da less) [42]. All known lantibiotics are synthesized as pre-peptides with an N-terminal leader peptide. It has been proposed that all the processing signals are in the leader region of the pre-peptides [15]. Thus, the pre-peptides would be recognized as substrates by binding of the leader peptides to the enzymes involved in posttranslational modifications. To test this hypothesis, unmodified pro-epidermin (obtained by factor Xa cleavage of EpiA) was used as a substrate for the flavoprotein EpiD. Even with this substrate, the reaction occurs. Hence, a leader peptide is not required for substrate recognition by EpiD. The primary reaction product was analyzed by Edman degradation to detect unmodified amino acid residues. A pattern of pmol amounts of the PTH-amino acids Ile+1 to Tyr+20 similar to that of unmodified pro-epidermin was obtained, which indicates that amino acids 1 to 78
3.7 The flavoenzyme EpiD and formation of peptidyl-aminoenethiolates 20 are unmodified and that one of the last two cysteine residues (Cys+21, Cys+22) is modified [42]. Recently, a more detailed analysis of the reaction product has become possible using tandem mass spectrometry and NMR methods.
3.7.7 In vivo reaction of affinity-tag-labeled epidermin precursor peptide with the flavoenzyme EpiD The genes encoding the His-tag-labeled epidermin precursor peptide EpiA and the flavoenzyme EpiD or the mutant protein EpiD-G93D, which lacks the coenzyme, were co-expressed and the proteins were synthesized in vivo in E. coli. Only in the presence of EpiD was the precursor peptide converted to a reaction product with a decrease in mass of 44–46 Da. This result confirms the in vitro experiments carried out with purified EpiA and purified EpiD from S. epidermidis. In the presence of EpiD, the amount of purified (modified) peptide EpiA was several-fold higher than in the presence of EpiD-G93D, indicating that the stabilization of EpiA against proteolysis is due to an interaction with EpiD or to the presence of the C-terminal modification [39].
3.7.8 Determination of the substrate specificity of EpiD using mutant precursor peptides and chemically synthesized peptides The precursor peptide EpiAC52S, which is altered by gene mutation and contains a C-terminal serine, was used to test the production of enols by reaction with EpiD [31]. No reaction occurred, which indicates the necessity for a Cterminal cysteine residue. However, not all peptides with a C-terminal cysteine residue, e. g. the peptide EpiA M1-C51, were a substrate of EpiD. EpiAS49A was a substrate for oxidative decarboxylation, providing the first hint that EpiD has no absolute substrate specificity. The leader peptide of the precursor peptide EpiA has no significant influence on the reaction with EpiD. Synthetic peptides with increasingly larger deletions of the amino-terminus were used to determine the minimal size of the substrate. A weak reaction was even observed for the tetrapeptide SYCC. The heptapeptide SFNSYCC was used to study the substrate specificity of EpiD further (Table 3.3). The serine and cysteine residues of this heptapeptide form the Cterminal bicyclic structure of mature epidermin, and it could not be excluded that amino acid exchanges in this peptide have an influence on the reaction with EpiD [31]. Interestingly, the penultimate cysteine residue can be exchanged with at least a serine or threonine residue. Modification or exchange of the C-terminal cysteine residue results in loss of the reaction with EpiD. The peptides SFNSYCS, SFNSYCM, and SFNSYC were no substrates of EpiD, which confirms the results 79
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin Table 3.3: Determination of the substrate specificity of the flavoprotein EpiD [43]. Peptide
Reaction with EpiD
EpiA, EpiAR-1Q, K-EpiA EpiAS+19A EpiAC+22S EpiAdesC+22
+ + – –
proepidermin synthetic proepidermin AKTGSFNSYCC SFNSYCC FNSYCC NSYCC SYCC YCC
+ + + + + + + –(?)
AFNSYCC SANSYCC SFASYCC SFNAYCC SFNSACC SFNSYGC
+ + + + – –
SFNSYCS SFNSYC SFNSYCC-NH2 SFNSYCC(Et) SFNSYCHCy SFNSYCM
– – – – – –
SFNSYSC SFNSYTC
+ +
SFNSFCC SFNSWCC
+ +
SFNYYCC SYNSYCC SWNSYCC
+ + +
TLTSECIC
–
obtained with the mutant precursor peptides. The analogue with a C-terminal ethyl-thioether structure [SFNSYCC(Et)], the amide SFNSYCC-NH2, and the peptide SFNSYCHcy (C-terminal homocysteine residue) were not substrates of EpiD. Mersacidin is another lantibiotic containing a C-terminal -NH–CH=CH–Sgroup [61], probably formed by modification of a C-terminal cysteine residue. The C-terminal peptide TLTSECIC of the mersacidin precursor peptide is not a substrate of EpiD [43]. 80
3.7 The flavoenzyme EpiD and formation of peptidyl-aminoenethiolates
3.7.9 Tandem mass spectra of modified SFNSYTC and SFNSYSC The sequence and structure of peptides can be analyzed with tandem mass spectrometry (collision-induced dissociation). In collision-induced dissociation experiments, peptides are preferentially cleaved at the peptide bonds between NH and CO, resulting in amino-terminal Bn fragments and C-terminal Yn fragments [7, 8]. To verify that the C-terminal cysteine residue of the reaction products is modified, product ion scans of SFNSYTC and SFNSYSC and the corresponding reaction products were recorded. These peptides were used to exclude intramolecular disulfide bridge formation in the peptide SFNSYCC. The B1–B6 fragments of the peptide and its reaction product are identical, proving that the C-terminal amino acid residue is modified. SFNSYTC and SFNSYSC and their reaction products differ from each other by 14 mass units in their B6 fragments, showing the S/T exchange. The mass difference between the modified peptide and its B6 fragment is 75 mass units; the mass difference between the unmodified peptide and its B6 fragment is 121 mass units. It was, therefore, possible to identify the reaction products by neutral loss mass spectrometry [43].
3.7.10 The application of neutral loss mass spectrometry to determine the substrate specificity of EpiD Tandem mass spectrometry methods have already been used to determine the composition of synthetic multicomponent peptide mixtures. For example, O-tertbutylated by-products of peptide libraries were identified by neutral loss scans [49]. In the constant neutral loss scan, the first and second analyzer of the mass spectrometer are scanned together such that there is a constant mass difference between the ions transmitted by the two analyzers. Under these conditions, only ions that lose a neutral fragment with a mass corresponding to the chosen mass difference will be detected. Heptapeptide sublibraries with one variable amino acid residue at positions 1 to 7 of the peptide substrate S1FNSYCC7 were synthesized and incubated with EpiD. Peptides were identified by their masses using product ion scans and neutral loss scans, and by comparison of the mass spectra obtained after various incubation times. The heptapeptides with a single amino acid substitution at positions 1 to 4 are substrates of EpiD. Not all amino acid residue substitutions at positions 5 to 7 of the peptide led to active substrates; therefore, the last three amino acids determine the substrate specificity for EpiD. For the sublibraries SFNSX5CC and SFNSYX6C, the reaction products were identified by neutral loss mass spectrometry of the peptide mixtures. The tyrosine residue at position 5 can be replaced by the hydrophobic amino acid residues V, I/L, (M), F, and W; it is not possible to differentiate between Ile and Leu by ES-MS. The cysteine residue at position 6 can be replaced by A, S, V, (I/L), and T. Pep81
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin tides containing amino acid residues with acidic or basic side chains at position 5 or 6 are not substrates of EpiD. In position 7 of the peptide substrates, only cysteine is accepted [43].
3.7.11 Structure analysis of the reaction products by two-dimensional NMR methods Due to the complexity of the NMR spectra, the introduction of a 13C isotope in the C-terminal cysteine residues was necessary to obtain the interesting signals using heteronuclear correlation experiments. Hence, the peptide KKSFNSYTC was synthesized by solid-phase synthesis using cysteine labeled with 13C at the b-carbon atom. The two lysine residues at the N-terminus were introduced to increase the solubility of the substrate in water. Two crosspeaks occur in the HSQC spectrum of the educt KKSFNSYTC. The first signal is due to the 13C-labeled b-carbon, while the second is due to the methyl groups in Tris[tris(hydroxymethyl)aminomethane]. The NMR data, the UV-VIS spectra, the MS-MS experiments, the observed reaction product with Ellman’s reagent, and the mass difference between educt and product of 46 Da indicate that the product contains a C-terminal thioenol [30].
3.7.12 The pKa value of the enethiol side chain of the reaction products is lower than that of the thiol side chain of peptides The UV-VIS spectra of the reaction products of EpiD are pH-dependent, indicating that the enethiol side chain is converted to an enethiolate anion. The pKa value of the enethiol group was determined to be 6.0 and is substantially lower than the pKa value of the thiol side chain of cysteine residues. This increased acid strength of the enethiol side chain compared to that of the thiol group is attributed to the resonance stabilization of the negative charge of the anion [39].
3.7.13 Overview of the function of EpiD In conclusion, the formation of the lantibiotic epidermin from the precursor peptide EpiA includes the oxidative decarboxylation of the C-terminal cysteine residue of EpiA to a (Z)-enethiol catalyzed by the FMN-containing enzyme EpiD. This oxidative decarboxylation reaction was the first analyzed posttranslational 82
3.8 Incorporation of d-alanine into S. aureus teichoic acids confers resistance modification reaction involved in lantibiotic biosynthesis. Two reducing equivalents from the C-terminal cysteine residue of EpiA are removed, a double bond is formed, and the coenzyme FMN is reduced to FMNH2. The decarboxylation occurs spontaneously or is catalyzed by EpiD. The in vivo conversion of (His)6tag-labeled precursor peptide EpiA by EpiD was demonstrated. The (Z)-enethiol derivative is the intermediate in the formation of the C-terminal S-((Z)-2-aminovinyl)-D-cysteine residue of epidermin. The enethiol structure has been confirmed by UV-VIS spectroscopy, mass spectrometry, tandem mass spectrometry, two-dimensional NMR spectroscopy, and conversion of the enethiol to a mixed disulfide with 5,5'-dithiobis(2-nitrobenzoic acid) (Ellman’s reagent). At physiological pH, the dominant form of the reaction product is the enethiolate anion (“peptidyl-aminoenethiolates”), and the pKa of the enethiol group is 6.0. EpiD has a low substrate specificity, and most of the peptides with the sequence (V/I/ L/(M)/F/Y/W)-(A/S/V/T/C/(I/L))-C at the carboxy terminus are substrates of EpiD, as elucidated by analysis of the reaction of EpiD with single peptides and peptide libraries. Amino acid residues of EpiD involved in FMN binding or substrate binding, or important for the catalytic action are currently identified by analyzing analogues generated by site-directed mutagenesis. In addition, the structure of EpiD has been determined by X-ray crystallography (Kupke et al. unpublished). The overall aim is the detailed molecular analysis of the flavoenzyme EpiD and related enzymes, and a more detailed investigation of the role of EpiD in epidermin biosynthesis. It is still not known how reduced EpiD is re-oxidized in vivo and whether EpiD forms a complex with the enzymes EpiB and EpiC. The chemistry of the peptidyl-aminoenethiolate reaction products has to be studied in order to develop these peptides as enzyme inhibitors.
3.8 Incorporation of d-alanine into S. aureus teichoic acids confers resistance to defensins, protegrins, and other antimicrobial peptides
Why some staphylococcal strains are more tolerant to gallidermin than other strains has always been of interest. In order to answer this question, Andreas Peschel [59] isolated S. aureus and S. xylosus transposon-insertion mutants that were hypersensitive to gallidermin, and found that genes hit by the transposon are involved in teichoic acid modification.
83
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin
3.8.1 Identification and sequence analysis of the S. aureus and S. xylosus dlt operon S. aureus Sa113 and the coagulase-negative S. xylosus C2 a have high innate tolerances towards several antimicrobial peptides. To investigate the resistance mechanism, S. xylosus C2 a was mutagenized with Tn917, and the resulting transposon-insertion mutants were analyzed for reduced growth on agar plates containing the antimicrobial peptide gallidermin. The nucleotide sequence upstream and downstream of the transposon from seven clones whose growth is specifically reduced in the presence of gallidermin was determined. In all seven mutants, the transposon had integrated into the same determinant of 4.6 kb, which encodes four open reading frames arranged in an operon-like structure, followed by a typical terminator (Fig. 3.10). The open reading frames showed sequence similarity to the Lactobacillus casei and B. subtilis dltABCD operons, which are responsible for esterification of teichoic acids with d-alanine [16, 55]. The growth rates in the absence of gallidermin and the microscopic appearance of the mutants were indistinguishable from those of the wild type strain.
Figure 3.10: cocci.
Genetic organization and proposed function of the dlt operon in staphylo-
3.8.2 Disruption of the S. aureus dlt operon and analysis of the d-alanine content in lipoteichoic acids (LTA) and wall teichoic acids (WTA) The dltA gene of S. aureus Sa113 was replaced by a spectinomycin resistance gene (spc) by homologous recombination, thereby producing the galliderminsensitive strain AG1. WTA and LTA of S. aureus and S. xylosus wild type strains and mutants were isolated, and the molar ratios of d-alanine to phosphorus were determined (Table 3.1). In the wild type strains, 75% (S. aureus) and 95% (S. xylosus) of the alditol phosphate residues in LTA are esterified with d-ala84
3.8 Incorporation of d-alanine into S. aureus teichoic acids confers resistance nine, while only 51% (S. aureus) and 15% (S. xylosus) are esterified in WTA. In the dlt mutants, no d-alanine is detected in LTA or WTA, indicating that the pathway for d-alanine incorporation is inactivated by the spectinomycin resistance gene and transposon insertions. When the mutant strains are complemented with plasmid pRBdlt1, which carries the dlt operon, normal or slightly increased amounts of d-alanine are found in LTA and WTA. Transformation of the wild type strains with pRBdlt1 results in an increase of d-alanine in LTA and WTA by 5–18% [59].
3.8.3 Sensitivity towards antimicrobial peptides The minimal inhibitory concentrations of gallidermin and of several other membrane-damaging antimicrobial peptides were determined for the S. aureus and S. xylosus wild type and mutant strains. The mutants are sensitive to a variety of antimicrobial peptides that bear a positive net charge [59]. The sensitivity of the S. aureus mutant to defensin from human neutrophils and to protegrins 3 and 5 from porcine leukocytes is at least 10- to 23-fold higher. Factors of 7–12 were determined with tachyplesin 1 and 3 from hemocytes of the horseshoe crab and to a variant of magainin II from the skin of the clawed frog. The tolerance towards the lanthionine-containing bacterial peptides gallidermin from S. gallinarum and nisin from L. lactis is 8–50-fold lower; very similar results were obtained with the S. xylosus strains. The increased sensitivity of dlt mutants seems to be restricted to cationic peptides since no considerable differences are observed in the inhibitory concentrations of the neutral peptide gramicidin D from Bacillus brevis. Furthermore, the mutants are not sensitive to cationic polylysine, indicating that cationic properties are not sufficient for activity of a peptide against staphylococci lacking d-alanine esters in their teichoic acids [59]. The results indicate that the general surface charge of staphylococci can be severely influenced by the degree of esterification with d-alanine. Loss of dalanine esterification is not deleterious to the bacterium, but has other pleiotropic effects. One such effect is that mutants become hypersensitive to a large variety of cationic peptides. Other effects are currently being studied. One can speculate whether the D-alanination of teichoic acids is a means whereby S. aureus overcomes the human defensin attack to which they are confronted during the first defense regimen.
85
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin
3.9 Conclusions
The collaborative research centre 323 allowed the lantibiotic research to boom, particularly in Tübingen. The fermentation group (H. Zähner and P. Fiedler), the organic chemistry group (G. Jung) and microbial genetics group (F. Götz) complemented each other ideally. The fermentation group optimized the fermentation process of the producing strains and provided sufficient amounts of the compounds to the chemists for structure determination and mass analysis, and the molecular biologists studied the biosynthetic genes and the functions of the corresponding enzymes together with the organic chemists. This concerted action was the secret of success. Quite a few of the Tübingen achievements can be regarded as milestones in lantibiotic research, such as: . Determination of the chemical structure of epidermin (Allgaier et al. 1986) . First finding of an antibiotic substance (later named gallidermin) in Staphylococcus gallinarum (F. Götz 1984) . First identification of the structural gene (epiA) encoding the lantibiotic precursor for epidermin; and the first proof that epidermin, and as we now know, all lantibiotics, are ribosomally synthesized (Schnell et al. 1989) . Determination of the NMR structure of gallidermin (Freund et al. 1991) . First cloning and nucleotide sequence of lantibiotic (epidermin) biosynthesis genes: epiB, C, D, P, and Q (Schnell et al. 1992, Augustin et al. 1992) . First characterization of a novel flavin-mononucleotide-dependent decarboxylase encoded by the epiD gene (Kupke et al. 1992) . Regulation of epidermin biosynthesis by the activator EpiQ (Peschel et al. 1993) . Isolation and characterization of genetically engineered gallidermin and epidermin analogues (Ottenwälder et al. 1995) . Further optimization of gallidermin production leading to a productivity of 300 mg/liter culture supernatant (Kempf et al. 1997) . Secretion mechanism of the lantibiotics epidermin and gallidermin (Peschel et al. 1997) . Unique producer self-protection against the lantibiotic epidermin by the ABC transporter EpiFEG (Peschel et al. 1996; Otto et al. 1998) . Nearly complete X-ray structure of the lantibiotic biosynthesis enzyme, the flavoprotein EpiD (Kupke et al., in progress) In Fig. 3.11 an overview of the biosynthetic pathway of epidermin (gallidermin) is presented. There are still many open questions that deserve attention. For example, the underlying enzyme mechanisms for the key reaction, the lanthionine formation, is still unsolved and it is not known which cofactors are involved. The environmental conditions under which the activator EpiQ is functional are unknown; there are no indications for a two-component system like that of nisin or subtilisin. Finally, the question remains why Gram-positive bac86
3.9 Conclusions
Figure 3.11: Pathway of epidermin biosynthesis.
teria produce such lantibiotics – what is their real function? It is unlikely that the antibiotic activity is the only function since it only becomes obvious when larger amounts of the lantibiotic are produced. Many isolates produce only minute amounts, which might, however, be sufficient for communication with other bacteria. Most of the lantibiotics interact with the cytoplasmic membrane; some, such as mersacidin and gallidermin, bind to the cell wall precursor lipid I and, as recently shown, interact with the teichoic acids to an extent that depends on the degree of negative charges. Perhaps they represent a measure for the fine tuning of cell wall structures. Since genes nearly identical to the epidermin biosynthesis genes are also present in S. aureus, one can ask whether they have an influence on infection and survival in the host. Where the lantibiotic biosynthesis genes originate is not known. The nucleotide sequences of most lantibiotic genes suggest that the genes are either plasmid encoded (e. g. epidermin) or part of a transposon (e. g. nisin). They might be relicts of bacteriophages that used lantibiotics for a better penetration of their genome into the target cells; in this respect, a pore-forming activity could be helpful. Whether any of the lantibiotics were once used as antibiotics or not, the study of these unique peptides sheds light on various areas of microbiology.
87
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin
References
1. Allgaier, H., Jung, G., Werner, R. G., Schneider, U., and Zähner, H. (1985) Elucidation of the structure of epidermin, a ribosomally synthesized, tetracyclic heterodetic polypeptide antibiotic. Angew. Chem. Int. Ed. Engl. 24, 1051–1053. 2. Allgaier, H., Jung, G., Werner, R. G., Schneider, U., and Zähner, H. (1986) Epidermin: sequencing of a heterodet tetracyclic 21-peptide amide antibiotic. Eur. J. Biochem. 160, 9–22. 3. Augustin, J. (1991) Dissertation. Eberhard-Karls-Universität Tübingen. 4. Augustin, J., Rosenstein, R.,Wieland, B., Schneider, U., Schnell, N., Engelke, G., Entian, K. D., and Götz, F. (1992) Genetic analysis of epidermin biosynthetic genes and epidermin-negative mutants of Staphylococcus epidermidis. Eur. J. Biochem. 204, 1149–1154. 5. Benz, R., Jung, G., and Sahl, H.-G. (1991) Mechanism of channel formation by lantibiotics in black lipid membranes. In: Jung, G. and Sahl, H. G. (eds.), Nisin and novel lantibiotics. Leiden: Escom, pp. 359–372. 6. Berridge, N., Newton, N. J., and Abraham, E. P. (1952) Purification and nature of the antibiotic nisin. Biochem. J. 52, 529–535. 7. Biemann, H. P. and Erikson, R. L. (1990) Abnormal protein kinase C down regulation and reduced substrate levels in non-phorbol ester-responsive 3T3-TNR9 cells. Mol. Cell Biol. 10, 2122–2132. 8. Biemann, K. (1992) Mass spectrometry of peptides and proteins. Annu. Rev. Biochem. 61, 977–1010. 9. Bierbaum, G., Brötz, H., Koller, K. P., and Sahl, H. G. (1995) Cloning, sequencing and production of the lantibiotic mersacidin. FEMS Microbiol. Lett. 127, 121–126. 10. Bierbaum, G., Götz, F., Peschel, A., Kupke, T., van de Kamp, M., and Sahl, H.-G. (1996) The biosynthesis of the lantibiotics epidermin, gallidermin, pep5 and epilancin K7. Antonie Van Leeuwenhoek. 69, 119–127. 11. Bierbaum, G. and Sahl, H. G. (1987) Autolytic system of Staphylococcus simulans 22: influence of cationic peptides on activity of N-acetylmuramoyl-L-alanine amidase. J. Bacteriol. 169, 5452–5458. 12. Bolhuis, H., van Veen, H. W., Poolman, B., Driessen, A. J., and Konings, W. N. (1997) Mechanisms of multidrug transporters. FEMS Microbiol. Rev. 21, 55–84. 13. Brötz, H., Josten, M., Wiedemann, I., Schneider, U., Götz, F., Bierbaum, G., and Sahl, H. G. (1998) Role of lipid-bound peptidoglycan precursors in the formation of pores by nisin, epidermin and other lantibiotics. Mol. Microbiol. 30, 317–327. 14. Brückner, R. (1997) Gene replacement in Staphylococcus carnosus and Staphylococcus xylosus. FEMS Microbiol. Lett. 151, 1–8. 15. Buchman, G. W., Banerjee, S., and Hansen, J. N. (1988) Structure, expression and evolution of a gene encoding the precursor of nisin, a small protein antibiotic. J. Biol. Chem. 263, 16260–16266. 16. Debabov, D. V., Heaton, M. P., Zhang, Q., Stewart, K. D., Lambalot, R. H., and Neuhaus, F. C. (1996) The d-alanyl carrier protein in Lactobacillus casei: cloning, sequencing and expression of dltC. J. Bacteriol. 178, 2869–3876. 17. Devriese, L. A., Poutrel, B., Kilpper-Bälz, R., and Schleifer, K. H. (1983) Staphylococcus gallinarum and Staphylococcus caprea, two new species from animals. Int. J. Syst. Bact. 33, 480–486. 18. Freund, S., Jung, G., Gutbrod, O., Folkers, G., and Gibbons, W. A. (1991) The threedimensional solution structure of gallidermin determined by NMR-based molecular graphics. In: Jung, G. and Sahl, H.-G. (eds.), Nisin and novel lantibiotics. Leiden: Escom, pp. 91–102.
88
References 19. Freund, S., Jung, G., Gutbrod, O., Folkers, G., Gibbons, W. A., Allgaier, H., and Werner, R. (1991) The solution structure of the lantibiotic gallidermin. Biopolymers 31, 803–811. 20. Geißler, S., Götz, F., and Kupke, T. (1996) Serine protease EpiP from Staphylococcus epidermidis catalyzes the processing of the epidermin precursor peptide. J. Bacteriol. 178, 284–288. 21. Gross, E., Kiltz, H., and Nebelin, E. (1973) Subtilin. VI. Die Struktur des Subtilins. Hoppe-Seyler’s Z. Physiol. Chem. 354, 810–812. 22. Gross, E. and Morell. J. L. (1971) The structure of nisin. J. Am. Chem. Soc. 93, 4634– 4635. 23. Hurst, A. (1966) Biosynthesis of the antibiotic nisin by Streptococcus lactis organisms. J. Gen. Microbiol. 44, 209–220. 24. Ingram, L. C. (1969) Synthesis of the antibiotic nisin: formation of lanthionine and bmethyl-lanthionine. Biochim. Biophys. Acta. 184, 216–219. 25. Jack, R., Benz, R., Tagg, J., and Sahl, H. G. (1994) The mode of action of SA-FF22, a lantibiotic isolated from Streptococcus pyogenes strain FF22. Eur. J. Biochem. 219, 699–705. 26. Jung, G. (1991) Lantibiotics – ribosomally synthesized biologically active polypeptides containing sulfide bridges and a,b-didehydroamino acids. Angew. Chem. Int. Ed. Engl. 30, 1051–1068. 27. Katz, E. and Demain, A. L. (1977) The peptide antibiotics of Bacillus: chemistry, biogenesis, and possible functions. Bacteriol. Rev. 41, 449–474. 28. Kellner, R., Jung, G., Hörner, T., Zähner, H., Schnell, N., Entian, K. D., and Götz, F. (1988) Gallidermin: a new lanthionine-containing polypeptide antibiotic. Eur. J. Biochem. 177, 53–59. 29. Kempf, M., Theobald, U., and Fiedler, H.-P. (1999) Economic improvement of the fermentative production of gallidermin by Staphylococcus gallinarum. Biotechnol. Lett. 21, 663–667. 30. Kempter, C., Kupke, T., Kaiser, D., Metzger, J. W., and Jung, G. (1996) Thioenols from peptidyl-cysteines: oxidative decarboxylation of a 13C labeled substrate. Angew. Chem. Int. Ed. Engl. 35, 2104–2107. 31. Kleerebezem, M., Quadri, L. E. N., Kuipers, O. P., and de Vos, W. M. (1997) Quorum sensing by peptide pheromones and two-component signal-transducing systems in Gram-positive bacteria. Mol. Microbiol. 24, 895–904. 32. Klein, C. and Entian, K. D. (1994) Genes involved in self-protection against the lantibiotic subtilin produced by Bacillus subtilis ATCC 6633. Appl. Environ. Microbiol. 60, 2793–2801. 33. Kleinkauf, H. and von Döhren, H. (1990) Nonribosomal biosynthesis of peptide antibiotics. Eur. J. Biochem. 192, 1–15. 34. Kordel, M. and Sahl, H. G. (1986) Susceptibility of bacterial, eukaryotic and artificial membranes to the disruptive action of the cationic peptide Pep5 and nisin. FEMS Microbiol. Lett. 34, 139–144. 35. Kordel, M., Schuller, F., and Sahl, H. G. (1989) Interaction of the pore forming-peptide antibiotics Pep 5, nisin and subtilin with non-energized liposomes. FEBS Lett. 244, 99–102. 36. Kuipers, O. P., Beerthuyzen, M. M., de Ruyter, P. G. G. A., Luesink, E. J., and de Vos, W. M. (1995) Autoregulation of nisin biosynthesis in Lactococcus lactis by signal transduction. J. Biol. Chem. 270, 27299–27304. 37. Kuipers, O. P., Beerthuyzen, M. M., Siezen, R. J., and de Vos, W. M. (1993) Characterization of the nisin gene cluster nisABTCIPR of Lactococcus lactis. Eur. J. Biochem. 216, 281–291. 38. Kuipers, O. P., Rollema, H. S., Yap, W. M., Boot, H. J., Siezen, R. J., and de Vos, W. M. (1992) Engineering dehydrated amino acid residues in the antimicrobial peptide nisin. J. Biol. Chem. 267, 24340–2436.
89
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin 39. Kupke, T. and Götz, F. (1997) The enethiolate anion reaction products of EpiD: pKa value of the enethiol side chain is lower than that of the thiol side chain of peptides. J. Biol. Chem. 272, 4759–4762. 40. Kupke, T. and Götz, F. (1996) Expression, purification, and characterization of EpiC, an enzyme involved in the biosynthesis of the lantibiotic epidermin, and sequence analysis of Staphylococcus epidermidis epiC mutants. J. Bacteriol. 178, 1335–1340. 41. Kupke, T. and Götz, F. (1996) Post-translational modifications of lantibiotics. Antonie van Leeuwenhoek. 69, 139–150. 42. Kupke, T., Kempter, C., Gnau, V., Jung, G., and Götz, F. (1994) Mass spectroscopic analysis of a novel enzymatic reaction. J. Biol. Chem. 269, 5653–5659. 43. Kupke, T., Kempter, C., Jung, G., and Götz, F. (1995) Oxidative decarboxylation of peptides catalyzed by flavoprotein EpiD: determination of substrate specificity using peptide libraries and neutral loss mass spectrometry. J. Biol. Chem. 270, 11282– 11289. 44. Kupke, T., Stevanovic, S., Ottenwälder, B., Metzger, J. W., Jung, G., and Götz, F. (1993) Purification and characterization of EpiA, the peptide substrate for posttranslational modifications involved in epidermin biosynthesis. FEMS Microbiol. Lett. 112, 43–48. 45. Kupke, T., Stevanovic, S., Sahl, H. G., and Götz, F. (1992) Purification and characterization of EpiD, a flavoprotein involved in the biosynthesis of the lantibiotic epidermin. J. Bacteriol. 174, 5354–5361. 46. Linnett, P. E. and Strominger, J. L. (1973) Additional antibiotic inhibitors of peptidoglycan synthesis. Antimicrob. Agents Chemother. 4, 231–236. 47. Liu, W. and Hansen, J. N. (1993) The antimicrobial effect of a structural variant of subtilin against outgrowing Bacillus cereus T spores and vegetative cells occurs by different mechanism. Appl. Environ. Microbiol. 59, 648–651. 48. Liu, W. and Hansen, J. N. (1992) Enhancement of the chemical and antimicrobial properties of subtilin by site-directed mutagenesis. J. Biol. Chem. 267, 25078–25085. 49. Metzger, J. W., Kempter, C., Wiesmuller, K. H., and Jung, G. (1994) Electrospray mass spectrometry and tandem mass spectrometry of synthetic multicomponent peptide mixtures: determination of composition and purity. Anal. Biochem. 219, 261–277. 50. Molitor, E. and Sahl, H.-G. (1991) Applications of nisin: a literature survey. In: Jung, G. and Sahl, H.-G. (eds.), Nisin and novel lantibiotics. Leiden: Escom, pp. 434–439. 51. Moll, G. N., Konings, W. N., and Driessen, A. J. (1998) The lantibiotic nisin induces transmembrane movement of a fluorescent phospholipid. J. Bacteriol. 180, 6565–6570. 52. Moll, G. N., Roberts, G. C., Konings, W. N., and Driessen, A. J. (1996) Mechanism of lantibiotic-induced pore-formation. Antonie Van Leeuwenhoek. 69, 185–191. 53. Ottenwälder, B., Kupke, T., Brecht, S., Gnau, V., Metzger, J., Jung, G., and Götz, F. (1995) Isolation and characterization of genetically engineered gallidermin and epidermin analogs. Appl. Environ. Microbiol. 61, 3894–3903. 54. Otto, M., Peschel, A., and Götz, F. (1998) Producer self-protection against the lantibiotic epidermin by the ABC transporter EpiFEG of Staphylococcus epidermidis Tü3298. FEMS Microbiol. Lett. 166, 203–211. 55. Perego, M., Glaser, P., Minutello, A., Strauch, M. A., Leopold, K., and Fischer, W. (1995) Incorporation of d-alanine into lipoteichoic acid and wall teichoic acid in Bacillus subtilis. J. Biol. Chem. 270, 15598–15606. 56. Peschel, A., Augustin, J., Kupke, T., Stevanovic, S., and Götz, F. (1993) Regulation of epidermin biosynthetic genes by EpiQ. Mol. Microbiol. 9, 31–39. 57. Peschel, A. and Götz, F. (1996) Analysis of the Staphylococcus epidermidis genes epiF, E, and G involved in epidermin immunity. J. Bacteriol. 178, 531–536. 58. Peschel, A., Ottenwälder, B., and Götz, F. (1996) Inducible production and cellular location of the epidermin biosynthetic enzyme EpiB using an improved staphylococcal expression system. FEMS Microbiol. Lett. 137, 279–284.
90
References 59. Peschel, A., Otto, M., Jack, R. W., Kalbacher, H., Jung, G., and Götz, F. (1999) Inactivation of the dlt operon in Staphylococcus aureus confers sensitivity to defensins, protegrins, and other antimicrobial peptides. J. Biol. Chem. 274, 8405–8410. 60. Peschel, A., Schnell, N., Hille, M., Entian, K.-D., and Götz, F. (1997) Secretion of the lantibiotics epidermin and gallidermin: sequence analysis of the genes gdmT and gdmH, their influence on epidermin production and their regulation by EpiQ. Mol. Gen. Genet. 254, 312–318. 61. Prasch, T., Naumann, T., Markert, R. L., Sattler, M., Schubert, W., Schaal, S., Bauch, M., Kogler, H., and Griesinger, C. (1997) Constitution and solution conformation of the antibiotic mersacidin determined by NMR and molecular dynamics. Eur. J. Biochem. 244, 501–512. 62. Rayman, M. K., Aris, B., and Hurst, A. (1981) Nisin: a possible alternative or adjunct to nitrite in the preservation of meats. Appl. Environ. Microbiol. 41, 375–380. 63. Reis, M., Eschbach-Bludau, M., Iglesias-Wind, M. I., Kupke, T., and Sahl, H. G. (1994) Producer immunity towards the lantibiotic Pep5: identification of the immunity gene pepI and localization and functional analysis of its gene product. Appl. Environ. Microbiol. 60, 2876–2883. 64. Reisinger, P., Seidel, H., Tschesche, H., and Hammes, W. P. (1980) The effect of nisin on murein synthesis. Arch. Microbiol. 127, 187–193. 65. Rogers, L. A. and Whittier, E. O. (1928) The inhibitory effect of Streptococcus lactis on Lactobacillus bulgarius. J. Bacteriol. 16, 321–325. 66. Ruhr, E. and Sahl, H. G. (1985) Mode of action of the peptide antibiotic nisin and influence on the membrane potential of whole cells and on cytoplasmic and artificial membrane vesicles. Antimicrob. Agents Chemother. 27, 841–845. 67. Sahl, H.-G., Jack, R. W., and Bierbaum, G. (1995) Biosynthesis and biological activities of lantibiotics with unique post-translational modifications. Eur. J. Biochem. 230, 827–853. 68. Sahl, H. G. (1991) Pore formation in bacterial membranes by cationic lantibiotics. In: Jung, G. and Sahl, H. G. (eds.), Nisin and novel lantibiotics. Leiden: Escom, pp. 347– 358. 69. Sahl, H. G. and Brandis, H. (1983) Efflux of low Mr substances from the cytoplasm of sensitive cells caused by the staphylococcin-like agent pep5. FEMS Microbiol. Lett. 16, 75–79. 70. Schnell, N., Engelke, G., Augustin, J., Rosenstein, R., Ungermann, V., Götz, F., and Entian, K. D. (1992) Analysis of genes involved in the biosynthesis of lantibiotic epidermin. Eur. J. Biochem. 204, 57–68. 71. Schnell, N., Entian, K. D., Götz, F., Horner, T., Kellner, R., and Jung, G. (1989) Structural gene isolation and prepeptide sequence of gallidermin, a new lanthionine containing antibiotic. FEMS Microbiol. Lett. 49, 263–267. 72. Schnell, N., Entian, K. D., Schneider, U., Götz, F., Zähner, H., Kellner, R., and Jung, G. (1988) Prepeptide sequence of epidermin, a ribosomally synthesized antibiotic with four sulphide-rings. Nature 333, 276–278. 73. Schüller, F., Benz, R., and Sahl, H. G. (1989) The peptide antibiotic subtilin acts by formation of voltage-dependent multi-state pores in bacterial and artificial membranes. Eur. J. Biochem. 182, 181–186. 74. Siegers, K. and Entian, K.-D. (1995) Genes involved in immunity to the lantibiotic nisin produced by Lactococcus lactis 6F3. Appl. Environ. Microbiol. 61, 1082–1089. 75. Siezen, R. J., Kuipers, O. P., and de Vos, W. M. (1996) Comparison of lantibiotic gene clusters and encoded proteins. Antonie Van Leeuwenhoek. 69, 171–184. 76. Stock, J. B., Ninfa, A. J., and Stock, A. M. (1989) Protein phosphorylation and regulation of adaptive response in bacteria. Microbiol. Rev. 53, 450–490. 77. Vogel, H., Nilsson, L., Rigler, R., Meder, S., Boheim, G., Beck, W., Kurth, H. H., and Jung, G. (1993) Structural fluctuations between two conformational states of a trans-
91
3 Biosynthesis of the Lantibiotics Epidermin and Gallidermin membrane helical peptide are related to its channel-forming properties in planar lipid membranes. Eur. J. Biochem. 212, 305–313. 78. Weil, H. P., Beck-Sickinger, A. G., Metzger, J., Stevanovic, S., Jung, G., Josten, M., and Sahl, H. G. (1990) Biosynthesis of the lantibiotic Pep5: isolation and characterization of a prepeptide containing dehydroamino acids. Eur. J. Biochem. 194, 217–223.
92
4 Fermentation of Lantibiotics Epidermin and Gallidermin Uwe Theobald*
4.1 Introduction
In recent years the research of the lantibiotics epidermin and gallidermin focused on two topics, mainly the more genetically based work on the biosynthesis of those antimicrobial peptides [1–3] and their production in bioreactors, respectively. Since antibiotics serve a vital role in modern medicine, fermentation technology represents a commercially significant part [4, 5]. Additionally, the need for an industrial production of new bioactive compounds become more and more important due to the increasing problem of multi-resistence in bacteria [6]. Both lantibiotics, gallidermin produced by Staphylococcus gallinarum Tü 3928 [7] and epidermin produced by Staphylococcus epidermidis Tü 3298 [8] exhibit strong activity against Gram-positive bacteria, particularly against propioni bacteria which are involved in the acne disease. Other favourable properties for a large-scale production are the comparable biological activity to renowned antibiotics in current clinical practice like erythromycin or fusidin because of the bactericidal impressiveness of both compounds against actively growing and non-dividing bacteria, the advantage for treatment of endocarditis, abscesses or skin infections. Finally, gallidermin or epidermin are eligible alternatives to vancomycin, so that these compounds gained pharmaceutical interest. The remarkable pharmacological properties led to extensive investigations during the last decade to optimise the production process of these drugs for industrial purpose [9–12]. However, each biotechnological production process is an unique operation. So, every step of this process needs optimisation for a promising conversion into an industrial standard. Major crucial points and optimisation steps known in general are the variation of media components or the design of special media compositions [13], the development of a suitable and reproducible process [14], the scale-up problems including oxygen transfer and mixing in the
* LSMW GmbH, Roßbachstraße 38, D-70499 Stuttgart
93 Microbial Fundamentals of Biotechnology. DFG, Deutsche Forschungsgemeinschaft Copyright # 2001 WILEY-VCH Verlag GmbH, Weinheim ISBN: 3-527-30615-3
4 Fermentation of Lantibiotics Epidermin and Gallidermin bioreactor [15], and the implementation of down stream processing steps to the process [16]. Hence, a complete summary of data concerning the fermentative production of gallidermin or epidermin has to reflect all optimisation steps of a fermentation – from strain cultivation to large scale production. Therefore, this chapter will focus on several different topics of an entire antibiotic production process that enabled efficient metabolite production.
4.2 Strains for gallidermin/epidermin production
The fermentative production of epidermin or gallidermin was possible with different microbial systems. Besides the mentioned wild type strains epidermin could be produced by a recombinant strain of Staphylococcus carnosus. In contrast to those wild type strains, Staphylococcus carnosus is a “food-grade” strain which is applied in starter cultures. Beyond that the strain Staphylococcus carnosus TM 300 was successfully used for heterologous expression of epidermin in a two-plasmid system. With the recombinant strain Staphylococcus carnosus TM 300 (pTepiMA+, pRBgepSI), containing the biosynthesis genes for epidermin production and immunity against its own antibiotic, nearly 70 % of the wild type production yield were observed. A problem of the two-plasmid system was the insufficient plasmid stability, so that most work on process development and optimisation was carried out with the wild type strains S. gallinarum Tü 3928 and S. epidermidis Tü 3298. According to the somewhat higher activity against living and non-dividing Gram-positive bacteria, this review will focus on the antibiotic gallidermin solely, although a lot of successful work was done for epidermin, too [17–21].
4.3 Disadvantages during gallidermin process development
Regarding the development of a gallidermin production process over a couple of years a few problems were generally noticed. First of all, a exceedingly strain instability concerning product excretion was observed, though biomass formation remained still stable. This fluctuating lantibiotic production yield was contradictory to a reproducible and standardised process. A second crucial point 94
4.4 Gallidermin – a lantibiotic and its way towards industrial production was the composition of the production medium. Several investigations dealt with the search for a defined and synthetic medium for staphylococci, but the biomass and product concentrations gained with the developed medium [22] could not compete with any of the tested complex media. Best production yield could be observed with a medium based upon meat extract. Meat extract was supposed to be the only suitable complex medium component that led to appropriate product yields [9] but is first of high costs for production or downstream processing and second carried a possible risk of prion infections such as bovine spongiform encephalopathy or Creutzfeld-Jakob-disease [23–25].
4.4 Gallidermin – a lantibiotic and its way towards industrial production
The basis for a microbial process is the stock culture for storage of the talented strain. This first step of the production could be identified as prime cause for the problem of the mentioned instability in product formation during fermentation [26]. A new mathematical parameter (hs-value) based upon the commonly analysed parameters biomass and product concentration was introduced to facilitate the optimisation of the media used for this cultivation of stock cultures on agar slants. The hs-value (high and stable product concentration) reduced the amount of data generated in optimisation experiments to one single value for each medium composition and allowed an assessment of any medium formulation with regard to reproducibility and product formation [26]. Figure 4.1 illustrates the time course of the product concentration over several passages found in the corresponding liquid cultures after 24 hours of incubation (A) and the corresponding hs-values (B) obtained exemplary from four different stock culture media (out of 53 different formulations tested) during several passages. Another indispensable optimisation step was the design of a new medium composition. An analysis of different complex compounds [13] suggested the possibility for a successful use of yeast extract (without risk of prion infection). In contrast to earlier results [9, 27] the investigations revealed that a special yeast extract could efficiently replace meat extract in the production medium of S. gallinarum Tü 3928. In addition to that the special yeast extract (Ohly KAT) applied in a three-fold lower concentration led to approximately 20 % higher product yields compared to the old medium composition [28]. Figure 4.2 shows a comparison of the gallidermin concentrations obtained with these media in batch fermentations. Since the Staphylococcus species are very halotolerant in general, the effect of salt (cation) is very important for growth and production [9–11, 27]. To optimise this parameter, different cations in various concentrations were 95
4 Fermentation of Lantibiotics Epidermin and Gallidermin
Figure 4.1: Gallidermin concentrations (A) and corresponding hs-values (B) obtained from a suitability test shown for four different stock culture media during several passages.
tested [12, 28–30]. Best results were gained with CaCl2 in very high concentrations of 30 to 40 g/l whereas the other cations like NaCl showed only slight influence. Figure 4.3 indicated, that there is an optimum in CaCl2 concentration whereas the NaCl concentration is only of slight influence upon product formation. However, Peschel and co-workers demonstrated that cationic teichonic acids at the staphylococcal cell surface are involved in gallidermin resistance [31]. This may be due to an ionic interaction between the teichonic acids and the gallidermin molecule that also has a positive net charge (3-fold). Although the mechanism of resistance is not yet clear in detail [1] it is possible that Ca2+ 96
4.4 Gallidermin – a lantibiotic and its way towards industrial production
Figure 4.2: Comparison of gallidermin formation by Staphylococcus gallinarum Tü 3928 grown in the previously used medium (open symbols) and the new developed production medium (full symbols). Both cultures were inoculated from the same agar plate.
Figure 4.3: Maximal gallidermin concentrations in the culture broth of Staphylococcus gallinarum Tü 3928 in dependency on different NaCl/CaCl2 proportions in the production medium.
97
4 Fermentation of Lantibiotics Epidermin and Gallidermin probably may be involved in the immunity as well. Hence, the components of the new medium were roughly (qualitatively) adjusted. Final optimisation were carried out with a computer-aided procedure. Genetic algorithms are able to optimise simultaneously a multi-parameter system like a medium composition (quantitatively optimisation) [32]. Several fermentation modes (batch, fed-batch, continuous) were tested. Preliminary experiments in continuous cultivations resulted in lower gallidermin yields compared to those gained in batch or fed-batch cultures. Best results were determined in batch processes or fed-batch processes with pulses of several amino acids or maltose during the antibiotic production phase [33]. A comparison of different types of bioreactor (dialysis fermenter, airlift reactor, stirred tank and loop reactor) showed no significant differences in growth and production. The effect of dissolved oxygen tension seemed to be more critical [11] than the type of reactor. The dialysis reactor consisted of two chambers which are separated by a dialysis membrane with an exclusion limit of 10 000 Dalton [10]. This reactor has the advantage of on-line product separation during the production phase but has the crucial drawback that no scale-up procedure is possible. Hence, a transfer from laboratory scale to industrial scale seemed to be out of question at that time. Scale-up investigations were focused on the batch and fed-batch processes in stirred tank reactors, which enabled a reliable scale-up procedure from 0.2 litre to 200 litre scale [34]. Best results in large scale fermentations were achieved by addition of maltose during the late production phase as shown in Fig. 4.4.
Figure 4.4: Concentrations of gallidermin (full symbols) and cell dry weight (open symbols) observed during a scale-up from 20 to 200 litres bioreactor. The arrow marks the transition from 20 to 200 litres and the time when maltose is pulsed into the 200 litres reactor; the grey area shows the range of the maximal gallidermin concentration (including errors) expected in a 200 litres bioreactor without maltose addition.
98
4.5 Conclusion The downstream processing (product recovery from the fermentation broth) was successfully optimised by product adsorption onto resin (Amberlite XAD-1180) [35]. An integration of the product separation step was performed by direct addition of resin into the bioreactor during fermentation or addition of an cross-flow filtration step between bioreactor and adsorber column. This resulted in a cell retention by the cross-flow module during the production phase. Product containing filtrate was harvested afterwards onto an adsorber column with XAD resin. So, the overall production time could be prolonged.
4.5 Conclusion
Extensive investigations led to a new method for strain storage and cultivation, the design of a new production medium, process development and the ability for scale-up into technical scale. An integration of a filtration step during production permitted an on-line product harvesting. Finally, it should be emphasised that higher production yields were obtained (more than 300 mg/l) and costs for medium components were drastically reduced more than 15-fold from 72 Euro per gram gallidermin to approximately 4.6 Euro per gram. Figure 4.5 summarised the reduction of medium costs and the increase in gallidermin concentration during the three optimisation steps (replacement of ingredients, rough adjustment of salt concentrations and final optimisation using a computer program).
Figure 4.5: Evolution of medium costs (columns) and product yields during three optimisation steps (see text).
99
4 Fermentation of Lantibiotics Epidermin and Gallidermin
References
1. Jack, R. W., Bierbaum, G., and Sahl, H.-G. (1998) Lantibiotics and related peptides. Springer-Verlag, Berlin. 2. Bierbaum, G., Götz, F., Peschel, A., Kupke, T., van de Kamp, M., and Sahl, H.-G. (1996) The biosynthesis of the lantibiotics epidermin, gallidermin, pep5 and epilancin K7. Antonie van Leeuwenhoek. 69, 119–127. 3. Kupke, T. and Götz, F. (1996) Post-translational modifications of lantibiotics. Antonie van Leeuwenhoek. 69, 139–150. 4. Strohl, W. R. (1997) Industrial antibiotics: Today and the future. In: Biotechnology of Antibiotics (ed. W. R. Strohl), Marcel Dekker, New York, pp 1–48. 5. Neijssel, O. M., Teixeira de Mattos, M. J., and Tempest, D. W. (1993) Overproduction of metabolites. In: Biotechnology (eds. H.-J. Rehm and G. Reed), Vol 1 Biological fundamentals (ed. H. Sahm), VCH, Weinheim, pp 163–188. 6. Dennensen, P. J. W., Bonten, M. J. M. and Weinstein, R. A. (1998) Multiresistant bacteria as a hospital epidemic problem. Annals of Medicine 30, 176–185. 7. Kellner, R., Jung, G., Hörner, T., Zähner, H., Schnell, N., Entian, K.-D., and Götz, F. (1988) Gallidermin: a new lanthionine-containing polypeptide antibiotic. Eur. J. Biochem. 177, 53–59. 8. Allgaier, H., Jung, G., Werner, R. G., Schneider, U., and Zähner, H. (1986) Epidermin: a sequencing of a heterodet tetracyclic 21-peptide amide antibiotic. Eur. J. Biochem. 160, 9–22. 9. Hörner, T., Ungermann, V., Zähner, H., Fiedler, H.-P., Utz, R., Kellner, R., and Jung, G. (1990) Comparative studies on the fermentative production of lantibiotics by staphylococci. Appl. Microb. Biotechnol. 32, 511–517. 10. Ungermann, V., Goeke, K., Fiedler, H.-P., and Zähner, H. (1991) Optimisation of fermentation and purification of gallidermin and epidermin. In: Nisin and novel lantibiotics (eds. G. Jung and H.-G. Sahl), Escom, Leiden, pp 410–421. 11. Kempf, M., Theobald, U., and Fiedler, H.-P. (1997) Influence of dissolved O2 on the fermentative production of gallidermin by Staphylococcus gallinarum. Biotech. Lett. 19, 1063–1065. 12. Breckel, A., Harder, M., Fiedler, H.-P., and Zähner, H. (1995) Production of gallidermin by Staphylococcus gallinarum Tü 3928. In: Biochemical Engineering 3 (ed. R. D. Schmid), Kurz & Co, Stuttgart, pp 62–66. 13. Greasham, R. L. (1993) Media for microbial fermentations. In: Biotechnology (eds. H.J. Rehm and G. Reed), Vol 3 Bioprocessing (ed. G. Stephanopoulos), VCH, Weinheim, pp 127–140. 14. Fiechter, A. (1986) Bioprocess development. In: Overproduction of microbial metabolism (eds. Z. Vanek and Z. Hostalek), Butterworths, Boston, pp 231–259. 15. Reuss, M. (1993) Oxygen transfer and mixing: scale-up implications. In: Biotechnology (eds. H.-J. Rehm and G. Reed), Vol 3 Bioprocessing (ed. G. Stephanopoulos), VCH, Weinheim, pp 185–217. 16. Spears, R. (1993) Overview of downstream processing. In: Biotechnology (eds. H.-J. Rehm and G. Reed), Vol 3 Bioprocessing (ed. G. Stephanopoulos), VCH, Weinheim, pp 39–55. 17. Jung, G., Allgaier, H., Kellner, R., Schneider, U., Hörner, T., Zähner, H., and Werner R. G. (1987) Isolation, purification and structure elucidation of epidermin, a ribosomally synthesized polypeptide antibiotic. In: Biochemical Engineering 1 (eds. H. Chmiel, W. P. Hammes and J. E. Bailey), Gustav Fischer, Stuttgart, pp 494–497. 18. Werner, R. G., Zähner, H., Jung, G., Allgaier, H., and Schneider, U. (1985) Antibio-
100
References
19.
20. 21.
22.
23. 24. 25. 26.
27. 28.
29.
30. 31.
32. 33.
34.
35.
tisches Polypeptid, Verfahren zu seiner Herstellung und seine Verwendung. European patent No. 0 181 578 B1. Werner, R. G., Zähner, H., Jung, G., Hörner, T., Kellner, R., and Fiedler, H.-P. (1989) Verfahren zur Gewinnung, Isolierung und Reinigung von Epidermin. European patent No. 0 350 810 B1. Allgaier, H., Hentschel, N., Walter, J. and Werner R. G. (1992) Isolation and purification of lantibiotics. European patent No. 0 508 371 A1. Ungermann, V., Hörner, T., Utz, R., Fiedler, H.-P., and Zähner, H. (1991) Comparative studies on the fermentation of lantibiotics, produced by Staphylococci. In: Biochemical Engineering 2 (eds. M. Reuss, H. Chmiel, E.-D. Gilles, and H.-J. Knackmuss), Gustav Fischer, Stuttgart, pp 301–305. Mollenkopf, F. (1998) Untersuchungen zur Epiderminbiosynthese bei Staphylococcus epidermidis Tü 3298 und zum Wachstum von Staphylokokken. Dissertation Universität Tübingen. Collee, J. G. and Bradley, R. (1997) BSE: A decade on – part 1. Lancet 349, 636–641. Collee, J. G. and Bradley, R. (1997) BSE: A decade on – part 2. Lancet 349, 715–721. Ironside, J. W. (1996) Creutzfeld-Jakob disease. Brain Path 6, 379–388. Theobald, U. and Kempf, M. (1998) A novel tool for medium optimisation and characterization in the early stages of a metabolite production process. Biotechnol. Techniques 12, 893–897. Jung, G., Kellner, R., Zähner, H., Götz, F., Hörner, T., Werner, R. G., and Allgaier, H. (1989) Antibiotic. European patent No. 0 342 486 B1. Kempf, M., Theobald, U., and Fiedler, H.-P. (1999) Economic improvement of the fermentative production of gallidermin by Staphylococcus gallinarum. Biotech. Lett. 21, 663–667. Ungermann, V. (1992) Untersuchungen zur Produktbildung und Scale-up der Produktion von Gallidermin einem Lanthioninhaltigen Peptidantibiotikum aus Staphylococcus gallinarum Tü 3928. Dissertation Universität Tübingen. Hörner, T. (1989) Fermentation und Isolierung Lanthioninhaltiger Polypeptidantibiotika aus Staphylococcen. Dissertation Universität Tübingen. Peschel, A., Otto, M., and Götz, F. (1998) Incorporation of D-alanine into staphylococcal teichonic acids confers resistance to antimicrobial peptides from bacteria, animals and humans. Lecture (No. KA021) held at the VAAM Jahrestagung, (22.–25. 3. 1998), Frankfurt. Holland, J.H. (1975) Adaption in natural and artificial systems. The University of Michigan Press Ann Arbor, Michigan. Kempf, M., Theobald, U., and Fiedler, H.-P. (1999) Correlation between the consumption of amino acids and the production of the antibiotic gallidermin by Staphylococcus gallinarum. Biotech. Lett. (accepted for publication). Kempf, M., Theobald, U., and Fiedler, H.-P. (2000) Production of the antibiotic gallidermin by Staphylococcus gallinarum – Development of a scale-up procedure. Biotech. Lett. (submitted for publication). Allgaier, H., Walter, J., Schlüter, M., and Werner, R. G. (1991) Strategy for purification of lantibiotics. In: Nisin and novel lantibiotics (eds. G. Jung and H.-G. Sahl), Escom, Leiden, pp 422–433.
101
5 Genetics of Nikkomycin Production in Streptomyces tendae Tü 901 Christiane Bormann*
5.1 Introduction: nikkomycins
Nikkomycins are peptidyl nucleoside antibiotics that are potent and specific inhibitors of chitin synthetases. They are structurally similar to UDP-N-acetylglucosamine, the natural substrate of these enzymes, and inhibition occurs in a competitive manner [1, 2]. Structures of nikkomycins are shown in Fig. 5.1. Nikkomycins X and Z, the main compounds produced by Streptomyces tendae Tü 901, exhibit high antifungal, insecticidal, and acaricidal activity [3]. Since their toxicity towards mammals and bees is very low or not detectable, and since nikkomycins are easily degraded in nature, they are potentially useful in agriculture or as therapeutic antifungal agents in humans. Nikkomycin Z shows significant activity towards the highly chitinous, pathogenic, dimorphic fungi Coccidioides immitis and Blastomyces dermatitidis [4]. Polyoxins, antibiotics related to nikkomycins, are commercially produced by Streptomyces cacaoi and applied as agricultural fungicides in Japan (for a review, see [5]). Nikkomycins X and Z are composed of the unusual amino acid hydroxypyridylhomothreonine (HPHT; nikkomycin D) and a peptidically linked nucleoside moiety. The nucleoside moiety comprises an aminohexuronic acid N-glycosidically linked to 4-formyl-4-imidazolin-2-one, forming nikkomycin Cx , or to uracil, forming nikkomycin Cz. Minor components of the culture filtrate of S. tendae Tü 901 are nikkomycins I and J, which are structures analogous to nikkomycins X and Z that contain glutamic acid peptidically bound to the 6'-carboxyl group of the aminohexuronic acid. Nikkomycins Cx , Cz, and D are also detected in culture filtrates and arise by hydrolytic cleavage of nikkomycins X and Z. These three compounds neither act as chitin synthetase inhibitors nor display biological activity. Nikkomycins Kx, Kz, Ox, and Oz (Fig. 5.1), which will be dis-
* Mikrobiologie/Biotechnologie, Universität Tübingen, Auf der Morgenstelle 15, D-72076 Tübingen
102 Microbial Fundamentals of Biotechnology. DFG, Deutsche Forschungsgemeinschaft Copyright # 2001 WILEY-VCH Verlag GmbH, Weinheim ISBN: 3-527-30615-3
5.1 Introduction: nikkomycins
R3 C O R2 NH CH
O R1
OH OH A
O
HN
N
N
O
HO
OH
NH2
OH
R2
CH CH CH C N
C O
CHO
HN
R1
B
N
O
CH3
NH2
CH CH2 CH C
H
O
D
OH
R2
NH2
CH CH2 CH C
HO N
O
nikkomycin
R1
R2
R3
X
A
A
OH
Z
B
A
OH
I
A
A
Glu
J
B
A
Glu
Cx Cz
A B
B B
OH OH
Kx
A
C
OH
Kz
B
C
OH
Ox
A
D
OH
Oz
B
D
OH
D
HO
NH2
OH
CH CH CH C OH N
CH3
O
Figure 5.1: Structures of nikkomycins.
103
5 Genetics of Nikkomycin Production in Streptomyces tendae Tü 901 cussed in a later section, are synthesized by mutants derived from S. tendae Tü 901/8 c by UV treatment and chemical mutagenesis [6]. Biosynthesis of nikkomycins can be divided into two parts: the nucleoside and peptidyl moieties are synthesized in separate pathways and then are linked by peptide bonds [6]. Figure 5.2 summarizes the steps of the nikkomycin biosynthetic pathway based on isotopic labeling and chemical characterization of pathway intermediates and shunt products. It also includes data on the biosynthesis of the polyoxin nucleoside, which contains an aminohexuronic acid moiety identical to that of nikkomycin nucleosides. Isono and coworkers [7] have shown that the polyoxin nucleoside arises from uridine and carbon-3 of phosphoenolpyruvate, and they proposed the reaction mechanism as a condensation of uridine and phosphoenolpyruvate to octofuranuloseuronic acid as the intermediate, followed by oxidative elimination of carbon-7' and carbon-8' and introduction of an amino group on carbon-5'. Isolation of octosyl acids, shunt metabolites derived from the postulated intermediate, supports this hypothesis. Analogues of octosyl acids, nikkomycins Sx and Sz, have been isolated from nikkomycin-producing S. tendae, and therefore the same biosynthetic pathway has been suggested for the nikkomycin nucleosides nikkomycins Cx and Cz [8]. Histidine is the precursor of the imidazolone base [9]; l-lysine is incorporated via picolinic acid into the pyridyl moiety and the attached carbon atom of nikkomycin D [10, 11]. Pyridylhomothreonine (PHT, nikkomycin E) and 4-pyridyl-2-oxo4-hydroxy-isovalerate (POHIV), which have been isolated from S. tendae culture filtrate [12, 13], are potential biosynthetic precursors of HPHT.
5.2 Isolation of nikkomycin biosynthetic genes
A number of approaches have been successfully used to isolate genes of antibiotic biosynthetic pathways from several Streptomyces strains. The strategy of cloning antibiotic-resistance genes, which are usually clustered with antibiotic biosynthetic genes, is not applicable to nikkomycin genes because the producing strain lacks the nikkomycin target site. Complementation of S. tendae mutants blocked in nikkomycin synthesis has led to the isolation of a 9.4-kb fragment that complements a non-producing mutant to nikkomycin Cx, Cz, and Kx synthesis [14]. However, structural genes of the nikkomycin pathway have not been identified. Therefore, the nikkomycin genes were cloned by identifying gene products involved in nikkomycin synthesis, microsequencing the N-termini to design oligonucleotide probes, and cloning the corresponding genes using these probes [15]. Nikkomycin biosynthetic enzymes were identified in gene expression studies using two-dimensional gel electrophoresis to separate cellular proteins. Initially, gene expression was analyzed in S. tendae wild type and mutant strains 104
Figure 5.2: Nikkomycin biosynthetic pathway based on the incorporation of labeled precursors and chemical characterization of pathway intermediates and shunt products; included are data on the biosynthesis of the polyoxin nucleoside (for references, see text). POHIV, 4-pyridyl-2-oxo-4-hydroxyisovalerate; PHT, pyridylhomothreonine.
105
5 Genetics of Nikkomycin Production in Streptomyces tendae Tü 901
Figure 5.3: Two dimensional gels of silver-stained cell extracts prepared from wild type Tü 901/8 c (A) and nikkomycin non-producing mutant Tü 901/NP13 (B). Mycelia were harvested in the stationary phase after 27.5 h of incubation. Proteins P1–P10 observed in the Tü 901/8c extract (A) and their corresponding positions in the Tü 901/NP13 extract (B) are indicated by arrows. These proteins were identified on the basis of three experiments performed with each producing and non-producing S. tendae strain. Positions of protein molecular mass standards are shown on the left. The acidic end of the gel is on the left, and the basic on the right.
blocked in nikkomycin biosynthesis [6]. Mutants which were able to synthesize nikkomycins but which differed from the wild type in the spectrum of nikkomycin structures formed had a protein pattern identical to that of the wild type. In contrast, protein profiles of mutants which did not synthesize any known nikkomycin structure differed from the wild type pattern as well as from each other. Selection of protein spots that were present in the producing strains and absent in each non-producing mutant led to the identification of ten gene products (P1–P10) (Fig. 5.3). As nikkomycin biosynthesis is regulated by the growth phase, the kinetics of the appearance of proteins P1–P10 was investigated during growth of S. tendae Tü 901/8 c in production medium (Fig. 5.4). Growth of S. tendae is characterized by a biphasic growth curve with two phases of rapid growth separated by a 1.5 to 2 h period in which growth slows down. Nikkomycin production begins in the transition to the stationary phase, after approximately 27 h after inoculation. Proteins P1–P6 and P10 were detected in extracts of mycelia harvested in the second exponential growth phase after 22.5 h of inoculation, and the amount of all of proteins P1–P10 increased to maximum levels at the early stationary phase and remained constant throughout the stationary phase. Synthesis of proteins P1–P10 preceded nikkomycin production, as would be expected for nikkomycin biosynthetic enzymes. N-terminal amino acid sequences were obtained for six of the ten identified proteins. Oligonucleotide probes designed from the N-terminal sequences of proteins P4, P5, and P8 gave positive signals of similar intensities with more 106
5.2 Isolation of nikkomycin biosynthetic genes
Figure 5.4: Time course of nikkomycin production (A) and gene expression (B) in S. tendae Tü 901/8 c cultivated in production medium. Nikkomycin Z plus X (g) in the culture filtrate compared to growth as determined by dry weight (&). Arrows indicate time points at which mycelia were harvested to prepare protein extracts for 2-D gel electrophoresis. (B) Enlarged sections of silver-stained 2-D gels of extracts of S. tendae Tü 901/8c harvested in the exponential phase (17.5 h), the late exponential phase (22.5 h), and the stationary phase (27.5 h). The position of proteins P1, P2, P3, P4, P7, P8, and P9 are marked by arrows.
than ten bands of digested genomic DNA and therefore were unsuitable for isolating the corresponding genes. A mixture of oligonucleotide probes designed from proteins P1 and P2, which have identical N-terminal sequences except that protein P1 has three additional amino acid residues at its N-terminus, and probes designed from protein P6 led to the isolation of genomic DNA fragments containing the binding sites for the probes used. The isolated fragments, an 8-kb BamHI fragment and a 6.5-kb PvuII fragment, appeared to overlap by 1.5-kb, and the former contained the binding sites for both oligonucleotide probes used. DNA sequence analysis revealed that the 8-kb BamHI fragment contains two open reading frames (ORFs) whose deduced N-terminal amino acid sequences are identical to that of proteins P1/P2 and that of protein P6. As gene-disruption mutants in the ORF encoding proteins P1/P2 (designated nikJ) and encoding protein P6 (designated nikI) failed to synthesize nikkomycins, it was evident that nikkomycin biosynthesis genes had been isolated. 107
5 Genetics of Nikkomycin Production in Streptomyces tendae Tü 901
5.3 Isolation of the nikkomycin gene cluster and expression in Streptomyces lividans
In bacteria, antibiotic biosynthetic genes are usually clustered. In order to isolate the entire nikkomycin gene cluster, a genomic library of S. tendae Tü 901/8 c was constructed in the Escherichia coli/Streptomyces shuttle cosmid pKC505 [16] and screened with the 8-kb BamHI fragment containing the nikI and nikJ genes. Hybridizing cosmids containing inserts of approximately 30 kb were mapped with restriction enzymes. Cosmids p24/32 and p9/43 carried the recognition sites for oligonucleotide probes designed from the N-terminal amino acid sequences of proteins P1/P2, P4, P5, P6, and P8, and of proteins P1/P2, P5, P6, and P8, respectively. S. lividans TK23, which does not produce nikkomycins, was used as a host to express the cloned nikkomycin biosynthetic genes. The nik cluster was not isolated on a single plasmid, as shown by the lack of synthesis of nikkomycins I, J, X, and Z by S. lividans TK23 transformants containing one of the recombinant plasmids. The nik cluster was cloned on two plasmids, p24/32 and p9/43, which carry inserts of about 31 and 27 kb, respectively, 15 kb of which overlap (Fig. 5.5). To facilitate selection for both plasmids, the apramycin resistance gene of p9/43 was removed by restriction with XhoI, which did not cut within the insert, and replaced with the aphII gene from Tn5. One of the apramycin- and neomycin-resistant S. lividans TK23 transformants synthesized relatively large amounts of nikkomycins I, J, X, and Z, approximately 50 % of that synthesized by S. tendae Tü 901/8 c. This transformant contained a large (>100-kb) plasmid that had formed by recombination between homologous regions of the two plasmids, either the vector pKC505 regions or the homologous regions of the inserts [17]. 27 kb
p9/43 31 kb
p24/32
8 kb
probe
B
S
S B
S
BS
B
S
BB
B
B BS
S
B
S
genome VM5 VM8
VM4
orfR
nikV U T
S R Q P2
P1
A B C
VM6 VM1/2
D
E
F G
I
J
K
LM N
O
Figure 5.5: Organization of the nik gene cluster and restriction map of the cloned chromosomal region containing the nik cluster. The structural genes (nik) and the regulatory gene (orfR) of the nik cluster are indicated by arrows. Wavy arrows indicate transcripts. The boxes below the genome map indicate the recognition sites for oligonucleotide probes VM1/2, VM4, VM5, VM6, and VM8, designed from proteins P1/2, P4, P5, P6, and P8. The hybridization probe used to screen the S. tendae gene library and the inserts of cosmid p24/32 and p9/43 are also indicated. B, BamHI; S, SacI.
108
5.4 Organization of the nikkomycin gene cluster
5.4 Organization of the nikkomycin gene cluster
The nucleotide sequence of an approximately 36-kb genomic region containing the nik cluster has been determined, and a series of 23 ORFs comprising a region of 29 kb identified. The translational start points of the ORFs have been tentatively located using the following criteria: (a) the G+C content in the third position of codons, (b) the location of a potential ribosome binding site at a suitable distance from the putative translation start [18], and (c) the observed similarities of the deduced amino acid sequence with proteins in databases. Each of the deduced proteins of the nikJ, nikS, nikA, nikI, and nikC genes had an N-terminal amino acid sequence identical to that determined for proteins P1/P2, P4, P5, P6, and P8, respectively. The most relevant features of the nik cluster and the encoded proteins deduced from the nucleotide sequence are summarized in Table 5.1, and the organization of the nik cluster is shown in Fig. 5.5. The ORFs of the nik cluster are arranged in three sets of adjacent genes with an intergenic spacing of <92 bp, and the stop codon of several ORFs overlaps with Table 5.1: Relevant features of the nik cluster and the encoded proteins deduced from the DNA sequence. ORF
G+C [mol%]
Residues/MW
Predicted/known function
nikA nikB nikC nikD nikE nikF nikG nikI nikJ nikK nikL nikM nikN
69.3 72.2 72.3 71.0 74.6 71.5 74.5 68.4 70.1 73.9 74.7 70.4 75.8
296/31,429 357/37,146 426/47,072 389/43,055 561/60080 410/45,908 64/6,565 218/24,489 453/50,422 374/40,164 240/25,895 213/24,086 550/57,238
Aldehyde dehydrogenase Synthetase Aminotransferase Oxidase Ligase Cytochrome P450 Ferredoxin
nikO nikP1 nikP2 nikQ nikR nikS nikT
73.4 73.2 75.5 73.1 76.7 70.8 73.6
471/50,345 677/73,061 272/29,514 396/44,479 226/24,991 424/46,873 440/46,227
nikU nikV orfR
63.8 69.2 74.0
155/16,549 423/46,077 1062/114,701
Aminotransferase
Membrane transport protein Enoylpyruvate transferase Peptide synthetase Thioesterase Cytochrome P450 Phosphoribosyltransferase Ligase Acyl carrier protein and aminotransferase Mutase, small subunit Mutase, large subunit Regulatory protein
109
5 Genetics of Nikkomycin Production in Streptomyces tendae Tü 901 the start codon of the subsequent gene, implicating translational coupling. The nikA-nikG gene cluster and the nikI-nikO gene clusters are separated by 944 bp and are oriented in the same direction, while the nikP1-nikV gene cluster is divergently transcribed; the intergenic region between nikP1 and nikA is 273 bp. The orfR gene, which encodes a pathway-specific activator protein for the nik biosynthesis genes is located at the left end of the nik cluster and is divergently oriented relative to the nikP1-nikV genes, with a 447-bp intergenic spacing.
5.5 Roles of the nik genes
The likely roles of the nik genes in nikkomycin biosynthesis as discussed below have been established from: (a) the nikkomycins structures synthesized by strains with inactivated nik genes, (b) sequence comparisons, and (c) enzymatic activities of nik gene products. Except for the nikG and nikU genes, whose roles seemed to be obvious, each gene of the nik cluster was inactivated by inserting a kanamycin resistance cassette via double cross-over homologous recombination. Novel nikkomycin structures accumulated by nik::aphII mutants were isolated from culture broth and their chemical structure (Fig. 5.6) was elucidated. O
R: NH 2
OH
COOH
CH CH CH C NH N
CH3
R
CHO
HN O
N
HN O
N
CH O
(2)
(1)
O
OH OH CHO
HN CHO
HN O
N H
O
HOCH2
N
O
(3) OH OH
(4)
Figure 5.6: Novel nikkomycin structures and biosynthetic intermediates synthesized by nik gene insertion mutants. Nikkomycins Lx (1) and Lz (2) are synthesized by nikF::aphII; 4-formyl-4-imidazolin-2-one (3) by nikK::aphII and fluorouracil-resistant nikR::aphII; and ribofuranosyl-4-formyl-4-imidazolin-2-one (4) by nikO::aphII.
110
5.5 Roles of the nik genes Table 5.2: Nikkomycin structures synthesized by mutants containing an inactivated nik gene. Mutant
Nikkomycin structures
Mutant
Nikkomycin structures
nikA::aphII nikB::aphII nikC::aphII nikD::aphII nikE::aphII nikF::aphII nikI::aphII nikJ::aphII nikK::aphII nikL::aphII nikM::aphII nikN::aphII nikO::aphII
Cx, Cz Cx, Cz Cx, Cz Cx, Cz Cx, Cz Lx , Lz RT 2.7 RT 2.7 Sx, Sz, (3) RT 0.75, Sx, Sz Sx , Sz Sx , Sz (4), RT 2.7
nikP1::aphII nikP2::aphII nikQ::aphII nikR::aphII nikS::aphII nikT::aphII nikV::aphII
Cx*, Cz I*, J, X*, Z RT 2.7, J, Z I*, J, X*, Z Cx, Cz Cx, Cz Cx, Cz, Kx, Kz, Ox, Oz
orfR::aphII
none
Structures of nikkomycins, intermediates, and shunt products are shown in Figs. 5.1, 5.2, and 5.6. Numbers in parentheses refer to structures shown in Fig. 5.6. RT 0.75 and RT 2.7 indicate unknown compounds that might comprise nikkomycin structures; RT refers to retention time in HPLC analyses. * reduced amount; approximately 60 % of wild type nikkomycin X production.
Table 5.2 summarizes the nikkomycin structures synthesized by gene-insertion mutants. Except for the nikJ::aphII and nikP1::aphII mutants, all mutants were able to synthesize the wild type nikkomycin spectrum by genetic complementation or by feeding suitable biosynthetic intermediates, which excluded the possibility that the mutant phenotype was caused by a polar effect of the integrated kanamycin cassette on downstream genes. Table 5.1 summarizes the predicted function of the deduced Nik proteins based on results of database searches. The nikC, nikT, and nikR genes have been expressed in E. coli, and the gene products catalyze the reactions shown in Fig. 5.7.
5.5.1 Biosynthetic pathway of the peptidyl moiety Figure 5.8 shows the proposed biosynthetic pathway for the peptidyl moiety HPHT [19]. The genes needed for its synthesis are all the genes of the nikAnikG operon and the clustered nikT-nikV genes of the nikP1-nikV operon (Fig. 5.5). Except for the nikF::aphII mutant, all mutants inactivated in one of these genes fail to synthesize biologically active nikkomycins, but accumulate the nucleoside structures nikkomycins Cx and Cz. The initial reaction in the HPHT biosynthetic pathway is catalyzed by NikC, which has been characterized as l-lysine-2-aminotransferase [20]. NikC 111
5 Genetics of Nikkomycin Production in Streptomyces tendae Tü 901 OH
OH
R
R O
O
NH2
O OH
(1)
O
NH2
O
OH
(2)
CH CH C C N
CH3
C
O
NikC
N H2
H 2O
OH
PLP
NH 2
O
phenylalanine PLP O
OH
N
O OH
phenylpyruvate
OH
NH2
CH CH CH C
NikT
N
CH3
O OH
O HN
P
(3)
O CH2
O O O
OH OH
P
P
+
O
PPi
HN O
N H
NikR
P
O CH2
N O
OH OH
Figure 5.7: Reactions catalyzed in vitro by recombinant nik gene products. Reactions (1), (2), and (3) are catalyzed by NikC, NikT, and NikR, respectively. PLP, pyridoxalphosphate.
removes the a-amino group from lysine to create piperideine-2-carboxylic acid, the cyclic and dehydrated form of 2-oxo-6-aminohexanoic acid (Fig. 5.7, reaction (1)). NikC belongs to a family of pyridoxamine/pyridoxal-phosphate-dependent enzymes, such as dehydrases involved in the formation of dideoxyhexoses of the Gram-negative cell wall and aminotransferases involved in deoxyamino sugar biosynthesis in secondary metabolism [21, 22]. NikC is a secondary metabolite enzyme that is specific for nikkomycin biosynthesis and is not required for l-lysine catabolism in S. tendae. The next step, the oxidation of piperideine-2-carboxylic acid to picolinic acid, is predicted to be catalyzed by the nikD-encoded protein, as feeding of picolinic acid to nikD mutants restores the ability to synthesize nikkomycins I, J, X, and Z. The NikD protein belongs to the family of monomeric sarcosine oxidases that contain FAD as the coenzyme. Members of this family metabolize sarcosine, l-pipecolic acid, and l-proline [23]. These latter two substrates are oxidized to D-piperideine-6-carboxylic acid and D-pyrrolidine-5-carboxylic acid, respectively. Piperideine-2-carboxylic acid is structurally similar to these compounds. A reaction mechanism similar to that of sarcosine oxidases is proposed 112
NH 2
Figure 5.8:
HO
L-lysine
NH 2
O
N
O
CH
OH
NikC
O
NH2
N
HPHT
CH3
COOH N picolinic acid
COOH
O2+2H+
N
CH
OH
2-oxoglutaric acid 2-oxoglutamic acid
NH2
R
PHT
O
S CoA
O
R
NH2
O
OH
N
CO2
O H
O
NikB
POHIV
CH 3
CH C H C COOH
OH
2-oxo-3-methylsuccinic acid
CH3
N
C
pyridine-2carbaldehyde
CoA-SH NAD+
NikA
+
HC C COOH
HOOC
NikT
O
OH
NikU, NikV
C H CH COOH
CH 3
N
C
NADH + H
picolinic acid-CoA
H 2O AMP + PPi
NikE
CoA-SH ATP
HOOC C H2 CH2 C COOH
NikD
4 [H]
NikF,NikG
H 2O
piperideine-2carboxylic acid
H 2O
C H CH COOH
NH2
OH
Proposed biosynthetic pathway for the peptidyl moiety HPHT.
O
OH
R
O
OH
O
R
5.5 Roles of the nik genes
113
5 Genetics of Nikkomycin Production in Streptomyces tendae Tü 901 for NikD, except that oxidation of piperideine-2-carboxylic acid to picolinic acid requires the removal of two electron pairs. The NikD protein contains an ADPbinding, bab-fold fingerprint sequence at its N-terminus, with an aspartate at position 1, which has been found in several FAD-containing enzymes and seems to be important for the interaction with FAD [24]. This finding implies that NikD might use FAD as a coenzyme. The presumed role of the nikE gene product is the activation of picolinic acid to form picolinic acid-coenzyme A, the predicted substrate for NikA. This proposed reaction rests on the similarity of NikE to adenylate-forming enzymes. NikE is most similar to enzymes that adenylate aromatic acids, such as SnbA, which activates 3-hydroxy-picolinic acid in the pristinamycin biosynthetic pathway [25], and 2,3-dihydroxybenzoate-AMP ligases of enterobactin biosynthesis [26] and other EntE homologs. NikE contains the conserved core sequences needed for ATP-binding and hydrolysis and for adenylate formation [27]. As feeding of HPHT to nikD mutants restores the ability to synthesize nikkomycins I, J, X, and Z, it is clear that NikE functions in the nikkomycin biosynthetic pathway and is not responsible for the formation of peptide bonds. The role of the nikA- and nikB-encoded proteins is deduced from their significant sequence similarity to aldehyde dehydrogenases (acylating) (ADAs) and 4-hydroxy-2-oxovalerate aldolases (HOAs), respectively, of the meta-cleavage pathway for the metabolism of aromatic hydrocarbons. The aldolase releases pyruvate and acetaldehyde, and the acylating dehydrogenase converts acetaldehyde to acetyl-CoA using NAD+ and CoA as cofactors. The mechanistically reverse reactions are postulated for NikA and NikB to form the carbon-carbon bond, thereby yielding the entire skeleton of HPHT. The proposed role of NikA is to catalyze the reduction of picolinic acid-CoA to pyridine 2-carbaldehyde with the release of coenzyme A. The NikA protein has an ADP-binding, bab-fold fingerprint sequence at its N-terminus [28], indicating that NAD(P)H might function as a hydrogen donor. NikB is proposed to catalyze a Claisen condensation of pyridine 2-carbaldehyde and a 2-oxo acid compound that will be discussed below. Two regions of NikB also exhibit similarity with regions of NifV homocitrate synthetases, which catalyze condensation of acetyl-CoA with 2-oxoglutarate to form homocitrate. The reaction mechanism of carbon-carbon bond formation by NifV homologs, also suggested for NikB, is proposed to be similar to that of carbon-carbon-cleavage by HOAs [29]. Interestingly, the order of the nikA and nikB genes in the nikA-nikG operon is the same as that of genes encoding ADA and HOA in the meta-pathway gene clusters. The order of these genes is strictly conserved; the ADA-encoding gene is located directly upstream of the HOA-encoding gene. The two proteins are suggested to form a pre-adapted catabolic module, which is thought to be assembled with other modules of the meta-cleavage operons in Pseudomonas [30]. NikA and NikB represent the first proteins shown to have significant sequence similarities along their entire length with ADAs and HOAs, respectively, that are not involved in meta degradation of aromatic hydrocarbons. These features indicate a common evolutionary origin. The presumed role of nikU and nikV is to catalyze the isomerization of 2oxoglutaric acid to yield 2-oxo-3-methylsuccinic acid, which is the predicted 2114
5.5 Roles of the nik genes oxo acid substrate in the NikB-mediated condensation reaction that yields the skeleton for HPHT with the release of carbon dioxide [31]. This hypothesis is based on the significant sequence similarity of NikU and NikV to methylaspartate mutase S and E chains, respectively, from Corynebacterium. The coenzyme B12-dependent mutase from Corynebacterium is composed of two components (S and E) and converts L-glutamate to threo-b-methyl-L-aspartate [32]. The analogous reaction is proposed for the putative mutase composed of NikV and NikU. The phenotype of nikV mutants is compatible with the proposed role for NikV, as the nikV mutants accumulate nikkomycins Kx , Kz, Ox , and Oz (Fig. 5.1) in addition to the nucleoside moieties nikkomycins Cx and Cz. Compared to HPHT, the peptidyl moiety of these nikkomycins lacks the methyl group at carbon-3. The phenotype of the nikV mutants can be rationalized by assuming that the lack of the nikV gene leads to an increased intracellular concentration of 2oxalacetate or pyruvate, which might be substrates for NikB, yielding the peptidyl moiety of nikkomycins Kx, Kz , Ox, and Oz . The nikT gene encodes a protein which consists of two domains [31]. The N-terminal region (approximately 70 amino acid residues) exhibits 60 % sequence similarity to acyl carrier proteins and contains a putative 4'-phosphopantetheine binding sequence. This domain of NikT is predicted to be involved in peptide bond formation (Section 5.5.3). The other part of NikT (amino acid 113 to 440) reveals significant sequence similarity to aminotransferase class-II and contains a putative pyridoxal-phosphate attachment site. As recombinant NikT transaminates POHIV in vitro to yield PHT (Fig. 5.7, reaction (2); shown in collaboration with K. Liebetrau, University of Münster), the role proposed for NikT is to catalyze this transamination reaction. The final reaction in the HPHT biosynthetic pathway is presumed to be catalyzed by the nikF-encoded cytochrome P450. The NikF protein is responsible for introducing the hydroxy group into the pyridyl ring since disruption of the nikF gene abolishes nikkomycin I, J, X, and Z synthesis, but leads to the formation of nikkomycin Lx and Lz, homologs of nikkomycins X and Z that contain an unhydroxylated pyridyl residue ((1), (2) in Fig. 5.6). PHT is assumed to be the substrate for NikF since PHT has been detected in culture filtrates of S. tendae [12]. NikF is a member of the Class I cytochrome P450 monooxygenases, which are reduced via an iron-sulfur protein by a flavin-containing reductase that accepts electrons from NAD(P)H. The ferredoxin-encoding gene nikG is located immediately downstream of the nikF gene, implying that NikG might be the in vivo electron transport protein for NikF.
5.5.2 Biosynthetic pathway for the nucleoside moiety Figure 5.9 summarizes the information on the nucleoside biosynthetic pathway. The genes nikQ and nikR, arranged in this order within the nikP1-nikV operon, are predicted to be responsible for the formation of 5'-phosphoribofuranosyl-4115
Figure 5.9: and Cz.
Proposed biosynthetic pathway for the nucleoside moieties nikkomycins Cx
5.5 Roles of the nik genes formyl-4-imidazolin-2-one, the precursor for nikkomycin Cx [33]. The genes nikO, nikI, nikJ, nikL, nikM, and nikK, which are located in a single operon, are proposed to be responsible for the formation of the aminohexuronic acid portion of the nucleoside skeletons originating from 5'-phosphoribofuranosyl-4-formyl4-imidazolin-2-one, uridine-monophosphate, and phosphoenolpyruvate [34]. The presumed role of the nikQ-encoded cytochrome P450 monooxygenase, the conversion of L-histidine to 4-formyl-4-imidazolin-2-one, has been deduced from the following findings. The nikQ mutants fail to produce nikkomycins X and I, which contain the imidazolone base, but synthesize nikkomycins Z and J, which contain uracil as the base. Feeding of 4-formyl-4-imidazolin-2-one ((3) in Fig. 5.6) isolated from culture filtrate of a nikK mutant restores the ability of nikQ mutants to synthesize nikkomycins X and J. The nikR gene product is thought to catalyze the transfer of 4-formyl-4imidazolin-2-one to 5-phosphoribosyl-1-pyrophosphate with the release of pyrophosphate, yielding 5'-phospho-ribofuranosyl-4-formyl-4-imidazolin-2-one. The NikR protein has significant sequence similarity to uracil-phosphoribosyl transferases (UPRTases), which catalyze the conversion of uracil and 5-phosphoribosyl-1-pyrophosphate to uridine-monophosphate in the uracil salvage pathway. This reaction is also catalyzed in vitro by recombinant NikR (Fig. 5.7, reaction (3)). The nikR mutants still synthesize nikkomycins containing the imidazolone base (X, I), but at levels of approximately 60 % of wild type levels. Fluorouracilresistant nikR mutants that have no detectable UPRTase activity fail to produce nikkomycins X and I, but synthesize wild type levels of the uracil-containing nikkomycins Z and J. In addition, 4-formyl-4-imidazolin-2-one (Fig. 5.6, structure (3)) is accumulated in culture filtrate. This indicates that the primary metabolism enzyme UPRTase can partly substitute for NikR in nikkomycin biosynthesis in nikR mutants. The proposed role of the nikO gene product is the transfer of phosphoenolpyruvate to the oxygen atom of carbon-5' of the riboside structures, yielding in further steps octofuranuloseuronic acids. NikO has significant sequence similarity to UDP-N-acetylglucosamine enolpyruvyl transferases, which catalyze the first committed step in murein biosynthesis for the bacterial cell wall, and to a minor extent to 5-enol-pyruvylshikimate-3-phosphate synthases of the shikimate pathway. These enzymes catalyze the transfer of phosphoenolpyruvate to a substrate. The nikO mutants accumulate ribofuranosyl-4-formyl-4-imidazolin2-one ((4) in Fig. 5.6), compatible with the role proposed for NikO [35]. The gene products of nikL, nikM, and nikK are involved in the nucleoside biosynthetic pathway after the nikO gene product, as all mutants inactivated in one of these genes accumulate nikkomycins Sx and Sz, which derive from the predicted octofuranuloseuronic acids (Fig. 5.2). While the amino acid sequence of NikL and NikM gave no clue to their role, NikK has significant sequence similarity to histidinol-phosphate aminotransferases. Based on this finding, NikK is proposed to introduce the amino group to 5'-oxohexuronic acid compounds, which are assumed to be the direct precursors for nikkomycins Cx and Cz. Less information is available on the role and order of the nikI and nikJ gene products in the nucleoside biosynthetic pathway. The nikI and nikJ mu117
5 Genetics of Nikkomycin Production in Streptomyces tendae Tü 901 tants do not accumulate nikkomycins Sx and Sz , but accumulate the unknown compound RT 2.7, which might be a biosynthetic intermediate. Since the nikJ mutants could not be complemented genetically, a polar effect of the inserted kanamycin cassette on the downstream located genes cannot be excluded. However, this seems to be unlikely since the nikJ phenotype differs from that of the nikK and nikL mutants. The nikJ gene product displays limited sequence similarity with methyltransferases involved in the biosynthesis of bialaphos and fortimycin, and with magnesium-protoporphyrin monomethylester oxidative cyclases [36]. The former enzymes use methylcobalamin as the methyl donor. Methylcobalamin and protoporphyrin are structurally similar, and the conserved amino acids might participate in binding or recognizing these cofactors. This implies that NikJ might contain cobalamin as a cofactor and might be responsible for the removal of carbon-7' and/or carbon-8' from the predicted octofuranuloseuronic acid intermediates.
5.5.3 Peptide bond formation The genes nikS, nikT, nikP1, and nikP2 are proposed to catalyze the peptide bond formation between the peptidyl moiety HPHT and the nucleoside structures nikkomycins Cx and Cz, yielding nikkomycin X and Z, respectively (Fig. 5.10) [31]. The NikS protein belongs to a superfamily of enzymes that is characterized by a specific ATP-binding structure. Members of this superfamily reveal strong structural similarity, but limited sequence similarity and catalyze carboxylate-amine/thiol ligation reactions. NikS is proposed to transfer POHIV through
Figure 5.10: Predicted reactions for peptide bond formation. The nikT and nikP1 gene products are represented as boxes, and their domains are indicated by bars. Substrates are attached to the 4'-phosphopantetheine prosthetic group by a thioester bond. R, 4-formyl-4-imidazolin-2-on, uracil, nikX, Z, nikkomycins X, Z.
118
5.5 Roles of the nik genes the phosphate to the 4'-phosphopantetheinyl group attached to the acyl carrier domain of NikT. Aminotransferase reaction and hydroxylation reaction (Fig. 5.8) could involve substrates bound to NikT. nikT and nikS mutants accumulate nikkomycins Cx and Cz and were not complemented to nikkomycin X/Z formation by feeding HPHT to the production medium. In addition, the finding that HPHT has never been found to be accumulated as a biosynthetic intermediate in any of the S. tendae Tü 901 mutants, is compatible with the roles proposed for NikS and NikT. The nikP1 gene encodes a peptide synthetase containing all the core sequences necessary for ATP-binding and adenylate-formation, and also displays the conserved binding site for the 4'-phosphopantetheinyl prosthetic group [27]. The nikP1 mutants accumulate nikkomycins Cx and Cz ; HPHT is not detected in culture filtrates. NikP1 is predicted to activate the nucleoside moieties, nikkomycin Cx and Cz , and catalyze peptide bond formation between them and HPHT attached to the acyl carrier domain of NikT. The nikP2 gene product resembles thioesterases required for the termination step in non-ribosomal peptide synthesis, the cyclization reaction in the biosynthesis of cyclic peptides, or the removal of the synthesized peptide from the peptide synthetase. The nikP2 mutants do not have a noteworthy phenotype; they synthesize the entire wild type nikkomycin spectrum. This finding indicates that another cytoplasmic thioesterase might substitute for NikP2 in nikkomycin synthesis. The mechanism of the formation of the second peptide bond between carbon-6' and glutamic acid in the tripeptidyl nikkomycins J and I is still a mystery since there are no more genes available within the nik cluster to encode for a peptide synthetase.
5.5.4 Export The protein deduced from the nikN gene displays limited sequence similarity to some putative membrane transport proteins. The finding that NikN is predicted to have six transmembrane helices and a binding site for ATP/GTP suggest that the nikN gene might be involved in the export of di- and tripeptidyl nikkomycins.
119
5 Genetics of Nikkomycin Production in Streptomyces tendae Tü 901
5.6 Transcriptional organization and regulation of the nik cluster
The 29-kb nik cluster contains three polycistronic operons and the monocistronically transcribed regulatory gene orfR (Fig. 5.5) [19, 31, 34, 37]. Transcriptional analyses have revealed that transcripts of the nikA-nikG genes, the nikI-nikO genes, and the nikP2-nikV genes first become evident at the beginning of the second rapid growth phase (Fig. 5.4), about five hours before nikkomycins become detectable in the culture broth, while transcription of the orfR gene is initiated approximately two hours earlier. Levels of all transcripts increase markedly during the transition to the stationary phase, and high levels are maintained during the stationary phase. Transcription of the nikA-nikG, the nikInikO, and the nikP1-nikV operons is activated by the orfR-encoded protein, as insertion of the kanamycin cassette in the 5'-half of the gene abolishes transcription of the three operons in these mutants. Heptameric repeat sequences detected in the promoter regions of these operons provide potential binding sites for OrfR. OrfR has a molecular mass of 114,701 Da (1062 amino acids), deduced from the nucleotide sequence and determined by SDS-PAGE of recombinant protein expressed in E. coli [37]. OrfR has a potential ATP/GTP-binding site at position 344 to 351, and its N-terminal half (amino acid residues 1 to 275) shows a high level of sequence similarity to members of the family of Streptomyces antibiotic regulatory proteins (SARPs), which are characterized by an OmpRlike DNA-binding fold at their N-terminus [38]. This family includes ActII-Orf4 (225 amino acids) from S. coelicolor A3(2) and DnrI (272 amino acids) from S. peucetius, pathway-specific regulatory proteins that activate transcription of actinorhodin and daunorubicin biosynthesis genes, respectively. The AfsR protein (993 amino acids), a pleiotropic regulatory protein from S. coelicolor A3(2) that effects several biosynthetic pathways, is the only member of the SARPs that has a large size similar to that of OrfR. In addition, AfsR contains a P-loop and has been shown to be phosphorylated by the afsK gene product, a serine/threonine protein kinase. The open question is which signal triggers expression of the orfR gene. In S. tendae Tü 901/8 c, the switch of metabolism to nikkomycin biosynthesis appears to take place during the short phase of reduced growth between the two exponential growth phases.
120
References
References
1. U. Dähn, H. Hagenmaier, H. Höhne, W. A. König, and G. Wolf (1976) Stoffwechselprodukte von Mikroorganismen. 154. Mitteilung. Nikkomycin, ein neuer Hemmstoff der Chitinsynthese bei Pilzen. Arch. Microbiol. 107, 143–160. 2. G. U. Brillinger (1979) Metabolic products of microorganisms. 181. Chitin synthetase from fungi, a test model for substances with insecticidal properties. Arch. Microbiol. 121, 71–74. 3. G. Zoebelein and U. Kniehase (1985) Labor-, Gewächshaus- und Freilanduntersuchungen von Nikkomycinen gegen Insekten und Milben. Pflanzenschutz-Nachrichten Bayer 38, 203–304. 4. R. F. Hector, B. L. Zimmer, and D. Pappagianis (1990) Positive interaction of nikkomycins and azoles against Candida albicans in vitro and in vivo. Antimicrob. Agents Chemother. 34, 587–593. 5. K. Isono and S. Suzuki (1979) The polyoxins, pyrimidine nucleoside peptide antibiotics inhibiting fungal cell wall biosynthesis. Heterocycles 13, 333–351. 6. C. Bormann, S. Mattern, H. Schrempf, H.-P. Fiedler, and H. Zähner (1989) Isolation of Streptomyces tendae mutants with an altered nikkomycin spectrum. J. Antibiot. 42, 913–918. 7. K. Isono, T. Sato, K. Hirasawa, S. Funayama, and S. Suzuki (1978) Biosynthesis of the nucleoside skeleton of polyoxins. Am. Chem. Soc. 100, 3937–3939. 8. T. Schüz, H.-P. Fiedler, H. Zähner, H. Rieck, and W. A. König (1992) Metabolic products of microorganisms. 263. Nikkomycins Sz, Sx, Soz and Sox, new intermediates associated to the nikkomycin biosynthesis of Streptomyces tendae. J. Antibiot. 45, 199–206. 9. R.-M. Schmidt, H. Pape, and M. Junack (1986) Biosynthesis of 4-formyl-4-imidazoline-2-on, the heterocyclic base of nikkomycin X. Naturforsch. 41 c, 135–140. 10. U. Sommer (1983) Untersuchungen zur Biosynthese von Nikkomycin D. PhD thesis, Universität Münster. 11. O. Kellner (1987) Biosynthese von Hydroxypyridylhomothreonin – Aminosäure der Nikkomycine. PhD thesis, Universität Münster. 12. W. A. König, H. Hahn, R. Rathmann, W. Hass, A. Keckeisen, H. Hagenmaier, C. Bormann, W. Dehler, R. Kurth, and H. Zähner (1986) Drei neue Aminosäuren aus dem Nikkomycin-Komplex – Strukturaufklärung und Synthese. Liebigs Ann. Chem. 1986, 407–421. 13. I. Hinterlang (1992) Isolierung, Strukturaufklärung, und Stoffwechsel eines Picolinsäuremetaboliten aus Streptomyces tendae. PhD thesis, Universität Münster. 14. C. Bormann, K. Aberle, H.-P. Fiedler, and H. Schrempf (1990) Genetic complementation of Streptomyces tendae deficient in nikkomycin production. Appl. Microbiol. Biotechnol. 32, 424–430. 15. V. Möhrle, U. Roos, and C. Bormann (1995) Identification of cellular proteins involved in nikkomycin production in Streptomyces tendae Tü901. Mol. Microbiol. 15, 561–571. 16. M. A. Richardson, S. Kuhstoss, P. Solenberg, N. A. Schaus, and R. N. Rao (1987) A new shuttle cosmid vector, pKC505, for streptomycetes: its use in the cloning of three different spiramycin-resistance genes from a Streptomyces ambofaciens library. Gene 61, 231–241. 17. C. Bormann, V. Möhrle, and C. Bruntner (1996) Cloning and heterologous expression of the entire set of structural genes for nikkomycin synthesis from Streptomyces tendae Tü 901 in Streptomyces lividans. J. Bacteriol. 178, 1216–1218. 18. M. J. Bibb (1986) Gene expression in Streptomyces: Nucleotide sequences involved in initiation of transcription and translation. In: G. Szabó, S. Biró, and M. Goodfellow
121
5 Genetics of Nikkomycin Production in Streptomyces tendae Tü 901
19.
20.
21.
22. 23.
24.
25.
26.
27.
28.
29. 30.
31.
32.
33.
34.
(ed.): Biological, biochemical and biomedical aspects of Actinomycetes. Part A, 25–34. Akademiai Kiadó, Budapest, Hungary. C. Bruntner, B. Lauer, W. Schwarz, V. Möhrle, and C. Bormann (1999) Molecular characterization of co-transcribed genes from Streptomyces tendae Tü901 involved in the biosynthesis of the peptidyl moiety of the peptidyl nucleoside antibiotic nikkomycin. Mol. Gen. Genet. 262, 90–101. C. Bruntner and C. Bormann (1998) The Streptomyces tendae Tü901 l-lysine 2-aminotransferase catalyzes the initial reaction in nikkomycin D biosynthesis. Eur. J. Biochem. 254, 347–355. J. S. Thorson, F. L. Stanley, H. Liu, C. R. Hutchinson (1993) Biosynthesis of 3,6-dideoxyhexoses: new mechanistic reflections upon 2,6-dideoxy, 4,6-dideoxy, and amino sugar construction. J. Am. Chem. Soc. 115, 6993–6994. W. Piepersberg (1994) Pathway engineering in secondary metabolite-producing actinomycetes. Crit. Rev. Biotechnol. 14, 251–285. B. E. Reuber, C. Karl, S. A. Reimann, and S. J. Mihalik (1997) Cloning and functional expression of a mammalian gene for a peroxisomal sarcosine oxidase. J. Biol. Chem. 272, 6766–6776. Y. Nishiya and T. Imanaka (1996) Analysis of interaction between the Arthrobacter sarcosine oxidase and the coenzmye flavin adenine dinucleotide by site-directed mutagenesis. Appl. Environ. Microbiol. 62, 2405–2410. V. De Crécy-Lagard, V. Blanc, P. Gil, L. Naudin, S. Lorenzon, A. Famechon, N. Bamas-Jacques, J. Crouzet, and D. Thibaut (1997) Pristinamycin I biosynthesis in Streptomyces pristinaespiralis: molecular characterization of the first two structural peptide synthetase genes. J. Bacteriol. 179, 705–713. F. W. Rusnak, S. Faraci, and C. T. Walsh (1989) Subcloning, expression and purification of the enterobactin biosynthetic enzymes 2,3-dihydroxybenzoate-AMP ligase: demonstration of enzymes-bound (2,3-dihydroxybenzoyl)adenylate product. Biochemistry 28, 6827–6835. T. Stachelhaus and M. A. Marahiel (1995) Modular structure of genes encoding multifunctional peptide synthetases required for non-ribosomal peptide synthesis. FEMS Microbiol. Lett. 125, 3–14. R. K. Wierenga, P. Terpstra, and W. G. J. Hol (1986) Prediction of the occurrence of the ADP-binding bab-fold in proteins using amino acid sequence fingerprint. J. Mol. Biol. 187, 101–107. J. Powlowski and V. Shingler (1994) Genetics and biochemistry of phenol degradation by Pseudomonas sp. CF600. Biodegradation 5, 219–236. S. Aemprapa and P. A. Williams (1998) Implications of the xylQ gene of the TOL plasmid pWW102 for the evolution of aromatic catabolic pathways. Microbiology 144, 1387–1396. B. Lauer, R. Russwurm, W. Schwarz, A. Kálmanánczhelyi, C. Bruntner, A. Rosemeier, and C. Bormann (2001) Molecular characterization of co-transcribed genes from Streptomyces tendae Tü901 involved in the biosynthesis of the peptidyl moiety and assembly of the peptidyl nucleoside antibiotic nikkomycin. Mol. Gen. Genet. 264, 662–673. D. E. Holloway and E. N. G. Marsh (1993) Cloning and sequencing of glutamate mutase component E from Clostridium tetanomorphum. Organization of the mut genes. FEBS Letts. 317, 44–48. B. Lauer, R. Russwurm, and C. Bormann (2000) Molecular characterization of two genes from Streptomyces tendae Tü901 required for the formation of the 4-formyl-4imidazolin-2-one-containing nucleoside moiety of the peptidyl nucleoside antibiotic nikkomycin. Eur. J. Biochem. 267, 1698–1706. C. Bormann, R. Russwurm, W. Schwarz, R. Bräuning, B. Lauer, V. Möhrle, and U. Roos (2001) Molecular characterization of co-transcribed genes from Streptomyces tendae
122
References
35. 36.
37. 38.
Tü901 required for the formation of the nucleoside moiety of the peptidyl nucleoside antibiotic nikkomycin. In preparation. B. Lauer, R. Süßmuth, D. Kaiser, G. Jung, and C. Bormann (2000) A putative enolpyruvyltransferase gene involved in nikkomycin biosynthesis. J. Antibiot. 53, 385–392. T. Kuzuyama, T. Seki, T. Dairi, T. Hidaka, and H. Seto (1995) Nucleotide sequence of fortimicin KL1 methyltransferase gene isolated from Micromonospora olivasterospora, and comparison of its deduced amino acid sequence with those of methyltransferases involved in the biosynthesis of bialaphos and fosfomycin. J. Antibiot. 48, 1191–1193. W. Schwarz and C. Bormann (2001) Transcriptional regulation of the nikkomycin biosynthesis gene cluster by a pathway-specific regulatory gene. In preparation. A. Wietzorrek and M. Bibb (1997) A novel family of proteins that regulates antibiotic production in streptomycetes appears to contain an OmpR-like DNA-binding fold. Mol. Microbiol. 25, 1177–1184.
123
6 Glycosylated Antibiotics: Studies on Genes Involved in Deoxysugar Formation, Modification and Attachment, and their Use in Combinatorial Biosynthesis Andreas Bechthold*
6.1 Introduction
6.1.1 Combinatorial biosynthesis Advances in recombinant DNA technology have enabled the cloning and expression of many biosynthetic gene clusters from different biological sources. Gene clusters encoding proteins catalyzing the biosynthesis of different natural bioactive products have been isolated and characterized from plants, fungi and bacteria. From the first isolation of antibiotic biosynthetic genes from a Streptomycete [1–3] and the cloning and expression of entire biosynthetic pathways of antibiotic producing Streptomycetes [4, 5], a lot of progress has been achieved. Functions have been assigned to many gene products in the biosynthesis of different bioactive compounds. More recently, researchers in this area including molecular biologists and bioorganic chemists have started to use these different genes to alter the structure of natural compounds by genetic engineering or to combine genes from different biosynthetic pathways. This new technology named combinatorial biosynthesis [6, 7] resulted in the formation of novel natural products.
6.1.2 Glycosylated antibiotics Deoxysugars are an important class of carbohydrates occurring in plants, fungi and microorganisms. In microorganisms deoxygenated sugars can be found as elements of lipopolysaccharides (LPS), extracellular polysaccharides (EPS) and antibiotics.
* Pharmazeutische Biologie, Universität Freiburg, Stefan-Meier-Str. 9, D-79104 Freiburg
124 Microbial Fundamentals of Biotechnology. DFG, Deutsche Forschungsgemeinschaft Copyright # 2001 WILEY-VCH Verlag GmbH, Weinheim ISBN: 3-527-30615-3
6.1 Introduction Traditionally, the sugar moieties of antibiotics have been viewed as molecular elements that control the pharmacokinetics of a drug. This view is beginning to change. Recent results clearly show that deoxygenated sugars actively contribute as recognition elements to the mechanism of action of the respective drug [8, 9]. In 1994 the author of this manuscript initiated investigations on the genetic analysis of avilamycin, landomycin and urdamycin biosynthesis and continued a cooperation on granaticin biosynthesis. Our research was motivated by the believe that genes, especially glycosyltransferase genes, are very important tools for combinatorial biosynthesis.
6.1.2.1 Avilamycin The avilamycins which are produced by Streptomyces viridochromogenes Tü 57 (S. viridochromogenes Tü 57) are oligosaccharide antibiotics and belong to the orthosomycin group of antibiotics [10, 11]. Avilamycins as well as other important members of the orthosomycins contain a dichloroisoeverninic acid moiety, as well as one or more orthoester linkages which are associated with carbohydrate residues. Components of the saccharide side chain are D-olivose, 2-deoxyD-evalose, 4-O-methyl-D-fucose, 2,6-Di-O-methyl-D-mannose, L-lyxose and eurekanic acid. Main product of S. viridochromogenes Tü 57 is avilamycin A (Fig. 6.1). The compound SCH27899, another orthosomycin, shows excellent activity against Gram-positive bacteria and is presently being tested for its possible use against human infectious diseases [12, 13].
6.1.2.2 Landomycin S. cyanogenus S136 is the producer of landomycins, consisting of a benz[a]anthraquinone-type aglycone moiety and a varying phenol-glycosidically linked oligosaccharide chain [14]. Landomycin A, the principal metabolite of S. cyanogenus, is the largest angucycline so far described, and the most active landomycin [15]. Its deoxysugar moiety is an unusual hexasaccharide consisting of four D-olivose and two L-rhodinose residues (Fig. 6.1). Besides landomycin A S. cyanogenus S136 produces the related congeners landomycin B and landomycin D. They differ from landomycin A in the length of the sugar side chain. In various tests landomycin A showed interesting antitumor activities, in particular against prostate cancer cell lines. This antitumor activity was discussed to result from interactions with DNA, for which the hexasaccharide chain plays an important role.
125
Figure 6.1:
Structure of avilamycin A, landomycin A, urdamycin A and granaticin.
6 Glycosylated Antibiotics: Studies on Genes
126
6.2 Cloning of the avilamycin, landomycin, urdamycin 6.1.2.3 Urdamycin Urdamycins are produced by S. fradiae. Like landomycins they are polyketides belonging to the angucyline type of antibiotics [16]. In difference to landomycins urdamycins mostly differ in the structure of the polyketide moiety. Main product of S. fradiae is urdamycin A (Fig. 6.1), minor compounds are urdamycin B, C, D, E, F, G and H. Urdamycin A consists of aquayamycin which contains a C-glycosidically linked D-olivose and three additional O-glycosidically linked deoxysugars: two L-rhodinoses, and another D-olivose. Urdamycin A shows only some anticancer and antimicrobial activity.
6.1.2.4 Granaticin Granaticin made by S. violaceoruber Tü 22 is a member of a class of polyketide antibiotics known as benzoisochromanequinones with activity against Gram-positive bacteria and P-388 mouse leukemia [17]. Granaticin contains a D-olivose moiety attached to the polyketide moiety via two carbon-carbon bonds at C-9 and C-10 (Fig. 6.1). A second sugar, L-rhodinose, is attached to D-olivose by a conventional glycosidic bond.
6.2 Cloning of the avilamycin, landomycin, urdamycin, and granaticin biosynthetic gene clusters
6.2.1 Development of a PCR method to clone dNDP-hexose 4,6-dehydratase genes In the past the cloning of biosynthetic gene clusters started with the shotgun cloning of random fragments of DNA from a Streptomyces strain, producing a given antibiotic. Mutants of the strain blocked at a step in the biosynthesis of the antibiotic were transformed with these fragments. Fragments restoring antibiotic production were used for further investigations. In another approach resistance genes of antibiotic producers were cloned in a heterologous host. This led to the identification of linked biosynthetic genes. More recently genes with known functions were used as probes in Southern hybridization to identify homologous genes in other organisms. The most popular gene probes were actI and actIII from S. coelicolor and strD from S. griseus [18, 19]. In our projects a different approach to clone the biosynthetic gene clusters of avilamycin, landomycin, urdamycin and granaticin was performed. The com127
6 Glycosylated Antibiotics: Studies on Genes parison of the sequence of known dTDP-glucose 4,6-dehydratases revealed several regions of high similarity. Based on these sequences two oligonucleotide primers were synthesized taking into account the codon usage of Actinomycetes. These primers were used to amplify DNA fragments from our Actinomycetes. PCR fragments obtained were subcloned and sequenced. The deduced amino acid sequences of the isolated fragments revealed similarity to each other and to the dNDP-glucose dehydratase (StrE) isolated from S. griseus N-2-3-11 [20].
6.2.2 Preparation and screening of cosmid libraries For the construction of gene libraries from our producer strains, DNA fragments of 25–40 kb were ligated into pOJ446. DNA was packed into phages by using the Gigapack Packaging Extract GoldTM from Stratagene (Heidelberg, Germany). The phages were used to transduce E. coli XL1-Blue-MRF '. For screening of the cosmid libraries, DNA fragments obtained by PCR amplification were used as probes. Several cosmid clones hybridizing to the probes were isolated in each case [21].
6.2.3 Identification of the biosynthetic gene clusters by gene inactivation and expression experiments Gene inactivation has been used to show that we in fact had cloned the urdamycin- and avilamycin biosynthetic gene cluster [22, 23], and expression of genes in heterologous strains was used to show that cosmid clones isolated from S. violaceoruber Tü 22 and S. cyanogenus S136 contained the entire granaticinand landomycin biosynthetic gene cluster, respectively [24, 25].
6.3 Organization of avilamycin, landomycin, urdamycin, and granaticin biosynthetic genes
6.3.1 Genes involved in avilamycin biosynthesis More than 42 kb of the avilamycin biosynthetic gene cluster have been sequenced (Fig. 6.2, Table 6.1 a) [26]. So far eight deoxysugar biosynthetic genes (aviQ1 aviQ2, aviQ3, aviD, aviE1, aviE2, aviE3, and aviJ), four glycosyltransfer128
Figure 6.2: Organization of antibiotic biosynthetic gene clusters. Genes are indicated as arrows orientated in the direction of transcription. Genes involved in deoxysugar biosynthesis, modification or in glycosyltransfer are shown as black arrows. A detailed description of the function of these genes is given in the text.
6 Glycosylated Antibiotics: Studies on Genes Table 6.1 a:
Putative functions of genes of the avilamycin biosynthetic gene cluster.
Gene
Deduced function
aviM aviH aviO1, aviO2 aviG1, aviG2, aviG3, aviG4, aviG5 aviB1, aviB2 aviD aviE1, aviE2, aviE3 aviQ1, aviQ2, aviQ3 aviJ1 aviGT1, aviGT2, aviGT3, aviGT4 aviRb, aviRa aviGTP, aviReg1, aviReg2 aviT, aviN, aviABC1+aviABC2 ?
orsellinic acid synthase (PKS) halogenase (tailoring) oxygenase (tailoring) methyltransferase (tailoring) acetoin-dehydrogenase (tailoring) dNDP-hexose synthase (sugar) dNDP-hexose-4,6-dehydratase (sugar) dNDP-hexose-4-epimerase (sugar) oxidoreductase (sugar) glycosyltransferase resistance protein regulatory protein transporter unknown function
ase genes (aviGT1, aviGT2, aviGT3, and aviGT4) and four putative sugar methyltransferase genes (aviG1, aviG2, aviG3, and aviG5) were detected. AviD and AviE1 are probably involved in the biosynthesis of the important key intermediate NDP-4-keto-6-deoxy-D-glucose, which can be converted to the 2,6-dideoxysugar components of avilamycin A, namely ring B (D-olivose), C (D-olivose) and D (D-evalose). AviQ1, AviQ2 and AviQ3 are similar to epimerases (AviQ1, after expression in E. coli, showed UDP-glucose 4-epimerase activity), AviE2 and AviE3 to NDP-D-glucose 4,6-dehydratases and AviJ to oxidoreductases. AviM is an orsellinic acid synthase and AviH seems to be involved in halogenation. AviG4 might code for a methyltransferase involved in the biosynthesis of the dichloroisoeverninic acid moiety. AviO1 and AviO2 resemble oxygenases and might catalyze the formation of the orthoesters. Further genes, aviB1, aviB2, aviRa, aviRb, aviGTP1, aviReg1, aviReg2, aviT, aviN, aviABC1, and aviABC2 are either involved in resistance, transport or regulation. The function of many other genes remains unknown [26, 27].
6.3.2 Genes involved in landomycin A biosynthesis The landomycin gene cluster from S. cyanogenus S136 contained several genes involved in deoxysugar- and polyketide biosynthesis (Fig. 6.2, Table 6.1 b). The formation of dTDP-D-glucose from glucose-1-phosphate is catalyzed by LanG which was shown after expression of lanG in E. coli. dTDP-D-glucose might then be converted to TDP-4-keto-6-deoxy-D-glucose by LanH. Deoxygenation at C-2 is probably catalyzed by LanS leading to a hypothetical intermediate which can be reduced by LanT. TDP-4-keto-2,6-dideoxy-D-glucose might be a 130
6.3 Organization of avilamycin, landomycin, urdamycin Table 6.1 b: Putative functions of genes of the landomycin, urdamycin and granaticin biosynthetic gene cluster. Gene
Deduced function
lanA, urdA, graORF-1 lanB, urdB, graORF-2 lanC, urdC, graORF-3 lan F, lanL, urdF, urdL, graORF-18, graORF-4, graORF-33 lanD, lanN, lanO, lanV, lanZ4, urdD, urdN, urdO, graORF-5, graORF-6, graORF-34 lanE, lanM, lanZ5, urdE, urdM, graORF-21, graORF-29 lanP lanG, lanZ2, urdG, graORF-16 lanH, urdH, graORF-17 lanQ, urdQ, graORF-23
b-ketoacyl-ACP synthase (PKS) chain length factor (PKS) acyl carrier protein (PKS) cyclase (PKS)
lanR, lanZ3, urdR, graORF-22 lanS, urdS, graORF-27 lanT, urdT, graORF-26
reductase (PKS)
oxygenase (tailoring) decarboxylase (PKS) dNDP-hexose-synthetase (sugar) dNDP-hexose-4,6-dehydratase (sugar) dNDP-4-keto-6-deoxy-hexose-3-dehydratase (sugar) dNDP-hexose-4-ketoreductase (sugar) dNDP-4-keto-6-deoxy-hexose-2,3-dehydratase (sugar) oxidoreductase (involved in C-2 deoxygenation) (sugar) dNDP-hexose-3,5-epimerase (sugar) glycosyltransferase
lanZ1, urdZ1, graORF-25 lanGT1, lanGT2, lanGT3, lanGT4, urdGT1 a, urdGT1 b, urdGT1 c, urdGT2, graORF-14 lanK, urdK, graORF-7, graORF-9, regulator graORF-10, graORF-11, graORF-19, graORF-20, graORF-37 lanJ, urdJ, urdJ2, graORF-15 transporter lanU, lanZ6, lanX, graORF-8, graORF-12, ? (unknown) graORF-13, graORF-24, graORF-28, graORF-30, graORF-31, graORF-32, graORF35, graORF-36
central intermediate in the biosynthesis of both TDP-L-rhodinose and TDP-Dolivose. A reduction step at C-4 (LanR) is required to complete the biosynthesis of TDP-D-olivose. 3,5-Epimerization (LanZ1), dehydroxylation at C-3 (LanQ) and reduction at C-4 (LanZ3) would lead to the formation of TDP-L-rhodinose. So far four genes (lanGT1–lanGT4) coding for putative glycosyltransferases have been detected in the biosynthetic gene cluster. Genes involved in the biosynthesis of the polyketide moiety are lanA, lanB, lanC, lanD, lanF, lanL and lanP. Further genes are either involved in tailoring reactions (lanE, lanM, lanZ5, lanN, lanO, lanV, lanZ4) or can be viewed as regulatory (lanK) or transporter (lanJ) genes (Fig. 6.2) [25, 28].
131
6 Glycosylated Antibiotics: Studies on Genes
6.3.3 Genes involved in urdamycin A biosynthesis So far eight genes (urdZ1, urdG, urdH, urdZ3, urdQ, urdR, urdS, urdT) probably involved in deoxysugar formation have been identified in the urdamycin biosynthetic gene cluster (Fig. 6.2, Table 6.1 b). The deduced amino acid sequences of these genes are very similar to the deduced amino acid sequences of genes from the landomycin cluster indicating similar functions. Four putative glycosyltransferase genes (urdGT1 a, urdGT1 b, urdGT1 c, and urdGT2), which have been detected in the cluster, are involved in the formation and attachment of the trisaccharide chain as well as the attachment of L-rhodinose at C12 b. Genes involved in the biosynthesis of the polyketide moiety or in tailoring reactions have also been detected in the urdamycin cluster. Products of these genes (urdA, urdB, urdC, urdD, urdF, urdL, urdE, urdM, urdO) are again very similar to the deduced amino acid sequences of genes from the landomycin cluster, and UrdJ and UrdJ2 resemble antibiotic transporters [26, 27, 29–31].
6.3.4 Genes involved in granaticin biosynthesis As in the landomycin and urdamycin cluster several genes involved in the formation of the polyketide and sugar moieties have been sequenced (Fig. 6.2, Table 6.1 b). The deduced amino acid sequences are very similar to proteins from the landomycin and urdamycin cluster [17, 24].
6.4 New genetically engineered natural compounds
Gene inactivation experiments by introducing in-frame deletions or frame-shift mutations into the genes and combinatorial biosynthetic approaches were carried out. New molecules were created and functions of genes could be determined.
132
6.4 New genetically engineered natural compounds
6.4.1 Avilamycin 6.4.1.1 New avilamycin derivatives through gene inactivation experiments To improve the water solubility of avilamycin A we started a project to inactivate methyltransferase genes of the avilamycin cluster. So far aviG1 and aviG4 have been deleted. Extracts of mutants were analyzed by HPLC-UV and by HPLCMS and were also analyzed for antibiotic activity (bioassay). Experimentally obtained data indicate that aviG1 is encoding a C-methyltransferase and that inactivation of aviG1 is leading to a less antibiotically active derivative. Inactivation of aviG4, which is probably involved in methylation of ring A of avilamycin A was leading to the formation of a more hydrophilic compound which showed high antimicrobial activity.
6.4.1.2 Production of orsellinic acid by expression of aviM The expression of one gene of the avilamycin biosynthetic gene cluster (aviM) in S. lividans was leading to the production of orsellinic acid (Fig. 6.3), a structural component of avilamycin A [22].
6.4.2 Landomycin and urdamycin 6.4.2.1 Accumulation of intermediates of the urdamycin biosynthesis after expression of a cosmid containing genes of the landomycin cluster Cosmid H2-26, containing genes of the landomycin cluster was transformed into the urdamycin producer S. fradiae. Transformants containing the cosmid were accumulating aquayamycin and compound 100-2. The structures of both compounds indicated that both compounds were not produced through combinatorial biosynthesis. Both compounds are likely to be intermediates of urdamycin A biosynthesis and a competition between glycosyltransferases of the urdamycinand landomycin cluster might be responsible for the accumulation of both compounds (Fig. 6.3) [25].
6.4.2.2 New compounds by inactivation of biosynthetic genes Several genes of the urdamycin cluster have been deleted by gene inactivation experiments. Mutants were accumulating new compounds instead of urdamycin A. (Table 6.2, Fig. 6.3) [29–31]. From these experiments the function to several genes can be assigned. 133
6 Glycosylated Antibiotics: Studies on Genes Table 6.2: Production of new antibiotics by gene inactivation- and gene expression experiments. Deleted gene
Gene expressed in the corresponding mutant
Products
urdGT2 urdGT2 urdGT1 a
– urdGT2 –
urdGT1 a urdGT1 b, urdInt and urdGT1 c urdGT1 b, urdInt and urdGT1 c urdGT1 a, urdGT1 b, urdInt and urdGT1 c urdGT1 a, urdGT1 b, urdInt and urdGT1 c urdGT1 a, urdGT1 b, urdInt and urdGT1 c
urdGT1 a
urdamycin I, urdamycin J, urdamycin K urdamycin A 12 b-desrhodinosyl-urdamycin C, 12 b-desrhodinosyl-urdamycin D, 12 b-desrhodinosyl-urdamycin F, urdamycinon D, urdamycin B urdamycin A
–
100-2
urdGT1 c
urdamycin G, urdamycin A
–
aquayamycin
urdGT1 c
12 b-desrhodinosyl-urdamycin G
urdGT1 b and urdGT1 c
12 b-desrhodinosyl-urdamycin G, 12 b-desrhodinosyl-urdamycin A
lanGT3
12 b-desrhodinosyl-urdamycin G
– – – – – – –
urdamycin I, urdamycin J rabelomycin aquayamycin urdamycin I, urdamycin J (no product) aquayamycin aquayamycin
urdGT1 a, urdGT1 b, urdInt and urdGT1 c urdGT1 a, urdGT1 b, urdInt, urdGT1 c and urdGT2 urdM urdR urdZ3 urdS urdZ1 urdQ
UrdGT2 must catalyze the earliest glycosyltransfer step in the urdamycin biosynthetic pathway which is the C-glycosyltransfer of an activated D-olivose. UrdGT1 a is involved in the glycosylation of aquayamycin during urdamycin A biosynthesis and UrdM is an oxygenase involved in oxygenation at position 12 b of urdamycin A.
134
135
Figure 6.3:
Structure of compounds generated during this study.
6 Glycosylated Antibiotics: Studies on Genes 6.4.2.3 New compounds by expression of single genes All glycosyltransferase genes of the urdamycin and landomycin biosynthetic gene cluster have been expressed in glycosyltransferase mutants of S. fradiae. The expression of urdGT1 c and lanGT4 (independent experiments) in a mutant, in which urdGT1 a, urdGT1 b, and urdGT1 c had been deleted and which was accumulating compound 100-2, resulted in the production of 12 b-desrhodinosyl-urdamycin G. After coexpression of urdGT1 c and urdGT1 b the accumulation of 12 b-desrhodinosyl-urdamycin A was detected (Table 6.2, Fig. 6.3). Thus UrdGT1 c is catalyzing the transfer of an activated L-rhodinose to 100–2 during urdamycin biosynthesis and UrdGT1 b is attaching the final sugar [31]. LanGT4 seems to catalyze the attachment of L-rhodinose and might act twice during landomycin biosynthesis.
References
1. Feitelson, J. S. and Hopwood, D. A. (1983) Cloning of a streptomyces gene for an Omethyltransferase involved in antibiotic biosynthesis. Mol. Gen. Genet. 190, 394–398. 2. Gil, J. A. and Hopwood, D. A. (1983) Cloning and expression of a p-aminobenzoic acid synthetase gene of the candicidin-producing Streptomyces griseus. Gene 25, 119–132. 3. Jones, G. and Hopwood, D. A. (1984) Molecular cloning and expression of the phenoxazinone synthase gene from Streptomyces antibioticus. J. Biol. Chem. 259, 2703–2714. 4. Malpartida, F. and Hopwood, D. A. (1984) Molecular cloning of the whole biosynthetic pathway of a Streptomyces antibiotic and its expression in a heterologous host. Nature 309, 462–464. 5. Motamedi, H. and Hutchinson, C. R. (1987) Cloning and heterologous expression of a gene cluster for the biosynthesis of tetracenomycin C, the anthracycline antitumor antibiotic of Streptomyces glaucescens. Proc. Natl. Acad. Sci. USA 84, 4445–4449. 6. Rohr, J. (1995) Kombinatorische Biosynthese – eine Zukunftsstrategie. Angew. Chem. 107, 963–966. 7. Khosla, C. and Zawada, R. J. X. (1996) Generation of polyketide libraries via combinatorial biosynthesis. Trends in Biotechnology 14, 335–341. 8. Kirschning, A., Bechthold, A., and Rohr, J. (1997) Chemical and biochemical aspects of deoxysugars and deoxysugar oligosaccharides. Topics Current Chemistry 188, 1–84. 9. Trefzer, A., Salas, J., and Bechthold, A. (1999) Genes and enzymes involved in deoxysugar biosynthesis. Nat. Prod. Rep. 16, 283–299. 10. Buzzetti, F., Eisenberg, F., Grant, H. N., Keller-Schierlein, W., Voser, W., and Zähner H. (1968) Avilamycin. Experientia 24(4), 320–323. 11. Mertz, J. L., Peloso, J. S., Barker, B. J., Babbitt, G. E., Occolowitz, J. L., Simson, V. L., and Kline, R. M. (1986) Isolation and structural identification of nine avilamycins. J. Antibiot. 39(7), 877–887. 12. Hoffmeister, D. and Bechthold, A. (1999) Halten wir Schritt mit den Mikroben? – Die Suche nach neuen Antibiotika. PZ Prisma 3, 185–194.
136
References 13. Urban, C., Mariano, N., Mosinka-Snipas, K., Wadee, C., Chahrour, T., and Rahal, J. J. (1996) Comparative in-vitro activity of SCH 27899, a novel everninomycin, and vancomycin. J. Antimicrob. Chemother. 37(2), 361–364. 14. Henkel, T., Rohr, J., Beale, J. M., and Schwenen, L. (1990) Landomycins. New angucycline antibiotics from Streptomyces sp. I. Structural studies on landomycins A–D. J. Antibiot. 43, 492–503. 15. Depenbrock, H., Bornschlegl, S., Peter, R., Rohr, J., Schmid, P., Schweighart, P., Block, T., Rastetter, J., and Hanauske A. R. (1996) Assessment of antitumor activity of landomycin A (NSC 6399187-A). Ann. Hematol. 73 (Suppl. II), A80, 316. 16. Drautz, H., Zähner, H., Rohr, J., and Zeeck, A. (1986) Metabolic products of microorganisms. 234. Urdamycins, new angucycline antibiotics from Streptomyces fradiae. 1. Isolation, characterization and biological properties. J. Antibiot. 39, 1657–1669. 17. Bechthold, A., Sohng, J. K., Smith, T. M., and Floss, H. G. (1995) Identification of Streptomyces violaceoruber Tü22 genes involved in the biosynthesis of granaticin. Mol. Gen. Genet. 248(5), 610–620. 18. Bechthold, A. and Salas, J. A. (1999) Combinatorial biosynthesis of microbial metabolites. In: Combinatorial organic chemistry (Jung, G. ed.), Wiley-VCH Weinheim, 381– 407. 19. Hopwood, D. A. (1997) Genetic contribution to understanding polyketide synthases. Chem. Rev. 97, 2465–2497. 20. Decker, H., Gaisser, S., Pelzer, S., Schneider, P., Westrich, L., Wohlleben, W., and Bechthold, A. (1996) A general approach for cloning and characterization of dNDPglucose dehydratase genes from actinomycetes. FEMS Microbiol. Lett. 141, 195–201. 21. Westrich, L., Gaisser, S., Reichenstein, B., and Bechthold, A. (1997) Preparation and screening of cosmid libraries from Streptomyces strains using the rapid DNA ligation kit, digoxigenin labeling and detection kit from Boehringer Mannheim. Biochemica 101, 30–32. 22. Gaisser, S., Trefzer, A., Stockert, S., Kirschning, A., and Bechthold, A. (1997) Cloning of an avilamycin biosynthetic gene cluster from Streptomyces viridochromogenes Tü57. J. Bacteriol. 179, 6271–6278. 23. Decker, H. and Haag, S. (1995) Cloning and characterization of a polyketide synthase gene from Streptomyces fradiae Tü2717, which carries the genes for biosynthesis of the angucycline antibiotic urdamycin A and a gene probably involved in its oxygenation. J. Bacteriol. 177, 6126–6136. 24. Ichinose, K., Bedford, D. J., Tornus, D., Bechthold, A., Bibb, M. J., Revill, W. P., Floss, H. G., and Hopwood, D. A. (1998) The granaticin biosynthetic gene cluster of Streptomyces violaceoruber Tü22: sequence analysis and expression in a heterologous host. Chemistry & Biology 5 (11), 647–659. 25. Westrich, L., Domann, S., Faust, B., Bedford, D., Hopwood, D. A., and Bechthold, A. (1999) Cloning and characterization of the landomycin biosynthetic gene cluster of Streptomyces cyanogenus S136. FEMS Microbiol. Lett. 170, 381–387. 26. Bechthold, A., Domann, S., Faust, B., Hoffmeister, D., Stockert, S., Trefzer, A., Weitnauer, G., and Westrich, L. (1999) Glycosidierte Naturstoffe, Perspektiven für die Kombinatorische Biosynthese. Chemotherapie Journal 4, 130–135. 27. Bechthold, A. and Rohr, J. (1999) Oligosaccharide antibiotics, perspective for combinatorial biosynthesis. In: New aspects in Bioorganic Chemistry (Diederichsen, U., Lindhorst, Th. K., Wessjohann, L., Westermann, B., ed.), Wiley-VCH Weinheim, pp 313– 321. 28. Wohlert, S., Bechthold, A., Beninga, C., Henkel, T., Holzenkämpfer, M., Kirschning, A., Oelkers, C., Weber, S., Weißbach, U., Westrich, L., and Rohr, J. (1999) Investigation on the biosynthesis of landomycin A. In: New aspects in Bioorganic Chemistry (Diederichsen, U., Lindhorst, Th.K., Wessjohann, L. and Westermann, B., ed), Wiley-VCH Weinheim, pp. 305–312.
137
6 Glycosylated Antibiotics: Studies on Genes 29. Künzel, E., Faust, B., Oelkers, C., Bearden, D., Westrich, L., Bechthold, A., and Rohr, J. (1999) The urdGt2 gene encodes a glycosyltransferase responsible for the Cglycosyltransfer of activated D-olivose, the earliest glycosyl transfer step in the urdamycin biosynthesis. J. Am. Chem. Soc., accepted for publication. 30. Faust, B., Hoffmeister, D., Weitnauer, G., Westrich, L., Haag, S., Schneider, P., Decker, H., Künzel, E., Rohr, J., and Bechthold, A. (1999) Two new tailoring enzymes, a glycosyltransferase and an oxygenase, involved in biosynthesis of the angucycline antibiotic urdamycin A in Streptomyces fradiae. Microbiology, accepted for publication. 31. Trefzer, A., Hoffmeister, D., Westrich, L., Weitnauer, G., Stockert, S., Künzel, E., Rohr, J., Fuchser, J., Bindseil, K., and Bechthold, A.(2000) Function of glycosyltransferase genes involved in the biosynthesis of urdamycin A. Chemistry & Biology 7, 13–142.
138
7 Analysis of the Biosynthesis of Glycopeptide Antibiotics: Basis for Creating New Structures by Combinatorial Biosynthesis Stefan Pelzer and Wolfgang Wohlleben*
7.1 Introduction
Together with gentamicin the glycopeptide antibiotic vancomycin (Fig. 7.1) and the structurally related lipoglycopeptide antibiotic teicoplanin are considered to be the last lines of defence against a variety of serious infections caused by Gram-positive organisms, such as enterococci, methicillin-resistant Staphylococcus aureus (MRSA), and Clostridium difficile [1]. Their annual turnover amounts to about one billion US$ [2], which demonstrates their high social and economic value. Glycopeptide antibiotics inhibit the synthesis of the cell wall of Gram-positive organisms. Their primary molecular target is the D-alanyl-D-alanine (D-AlaD-Ala) terminus of growing peptidoglycan. Dimers of vancomycin bind specifically to D-Ala-D-Ala inhibiting simultaneously the two key-steps of the peptidoglycan biosynthesis, the transglycosylase and the transpeptidase reaction [2]. In the last ten years numerous vancomycin resistant enterococci (VRE) emerged in clinical isolates [3] and spread rapidly. These organisms avoid their cell deaths by modifying the drug’s target, specifically by modifying it to the depsipeptide D-alanyl-D-lactat (D-Ala-D-Lac). The affinity of vancomycin to DAla-D-Lac is reduced significantly (factor 1000). The fear that this high-level resistance may eventually spread to methicillin-resistant Staphylococcus aureus (MRSA) has prompted the search for new or modified glycopeptide antibiotics. One way to generate new glycopeptide antibiotics is the manipulation of biosynthetic genes by combinatorial biosynthesis. In order to employ this approach successfully, corresponding gene clusters have to be identified and analysed. Until now only putative genes of the vancomycin producer A. orientalis
* Mikrobiologie/Biotechnologie, Universität Tübingen, Auf der Morgenstelle 28, D-72076 Tübingen
139 Microbial Fundamentals of Biotechnology. DFG, Deutsche Forschungsgemeinschaft Copyright # 2001 WILEY-VCH Verlag GmbH, Weinheim ISBN: 3-527-30615-3
7 Analysis of the Biosynthesis of Glycopeptide Antibiotics
Figure 7.1: Chemical structures of the glycopeptide antibiotics balhimycin, vancomycin and chloroeremomycin [17].
140
7.2 Results C329.4 [4] and the chloroeremomycin producer A. orientalis A82846 were described [5]. However, the role of the sequenced genes was not elucidated. A glycopeptide with similar properties as vancomycin is balhimycin, which is produced by Amycolatopsis mediterranei DSM5908 [6]. Balhimycin shares the heptapeptide core of vancomycin and differs only in the glycosylation pattern (Fig. 7.1). In the following we present the establishment of a transformation protocol and the construction of vectors for manipulation of the balhimycin producer, the isolation of the balhimycin gene cluster, and the functional analyses of different biosynthetic genes.
7.2 Results
7.2.1 Establishment of a gene cloning system for the balhimycin producer Amycolatopsis mediterranei DSM5908 Although transformation techniques were described for the vancomycin producer A. orientalis ATCC 19795 [7] and the producer of the teicoplanin-like antibiotic A47934, S. toyocaensis [8], the application for generating glycopeptide antibiotic mutants was never reported. For the establishment of a gene cloning system for the balhimycin producer A. mediterranei DSM5908 a transformation method was developed first. In order to monitor successful transformation events, DNA of the Amycolatopsis phage W2 [9], which was able to infect the balhimycin producer was employed. Using W2-DNA in transfection experiments, the three different methods described for the transformation of Amycolatopsis strains were tested for their applicability in A. mediterranei DSM5908. Neither various modifications of the PEG induced protoplast transformation method, which was reported as an efficient method for A. orientalis [8], nor electroporation methods described for transformation of A. mediterranei and A. orientalis [10] enabled introduction of W2-phage DNA into A. mediterranei DSM5908. Only after modifying the direct transformation method [11], which was previously described for the rifamycin producer A. mediterranei [12] and for A. methanolica [13], transformants of the balhimycin producer were obtained. Using the optimised protocol it was not possible to establish any of the described Amycolatopsis vectors or broad host range Streptomyces vectors [11]. Therefore, novel non-replicative vectors for integration mutagenesis were developed. Since the balhimycin producer showed a naturally high resistance against the customary actinomycetes resistance markers thiostrepton and kanamycin, the erythromycin-resistance gene ermE [14] and the chloramphenicol-resistance gene cat [15] were used to construct the vectors pSP1 and pSP2 [11]. 141
7 Analysis of the Biosynthesis of Glycopeptide Antibiotics The application of the vectors in gene disruption, gene replacement and in-frame deletion experiments revealed, that the integration frequencies were influenced by several parameters [11]: Highest integration frequencies were observed when the DNA was isolated from the dam/dcm E. coli strain JM110 and PEG 1000 from NBS Biologicals was employed. The efficiency of integration depended directly on the size of the cloned insert. Plasmids with fragments smaller than 1 kb were difficult to integrate. In gene replacement experiments a high double crossover rate (31%) was demonstrated.
7.2.2 Identification of biosynthetic genes by reverse genetic approaches The application of standard strategies for isolation of antibiotic biosynthetic genes like cloning of resistance genes or complementation of non-producing mutants was not possible. Therefore, different reverse genetic approaches were chosen considering characteristic features of glycopeptide antibiotics like the heptapeptide backbone or the sugar residues.
7.2.2.1 Isolation of different peptide synthetase genes using conserved oligonucleotide probes Since the central core heptapeptide of balhimycin includes five non-proteinogenic amino acids, a non-ribosomal synthesis by peptide synthetases is most likely. Peptide synthetases of bacterial and fungal origins are highly conserved and show a modular organisation. Each module activates the cognate amino acid and incorporates it in the peptide product by a stepwise elongation [16]. Comparison of the amino acid sequences of peptide synthetase modules from various microorganisms showed five strongly conserved regions [11]. These conserved motives were used to design oligonucleotide probes, which were adapted to the Streptomyces codon usage. Using these gene probes in plaque hybridisation experiments we identified two different clusters of peptide synthetase genes in a l library of A. mediterranei. Gene disruption experiments revealed, that both different peptide synthetase gene clusters are not involved in balhimycin production [8]. Today we know that genes encoding peptide synthetases are common in Actinomycetes and that the balhimycin producer carries at least four different peptide synthetase gene clusters, two of them with unknown function.
142
7.2 Results 7.2.2.2 Identification of a glycosyltransferase gene fragment involved in balhimycin biosynthesis Another reverse genetic approach considered, that balhimycin includes two sugar groups, a glucose and the rare aminosugar dehydrovancosamine (Fig. 7.1). Such sugars are normally attached to the aglycon by the action of glycosyltransferases. To design PCR primers for the amplification of balhimycin-specific glycosyltransferase gene fragments, sequences of glycosyltransferases probably involved in the biosynthesis of vancomycin and chloroeremomycin [4] were compared and two conserved regions were chosen. Using conserved oligonucleotides we were able to amplify a 900 bp DNA fragment (“bgtfB”) from genomic DNA of the balhimycin producer in an optimised PCR-protocol [17]. DNA sequencing and comparison of the deduced gene product showed significant similarities to various glycosyltransferases. To prove that the glycosyltransferase gene is involved in balhimycin biosynthesis, a gene disruption experiment was performed. An internal fragment of the bgtfB gene was cloned into the gene disruption vector pSP1. This vector was used to inactivate the chromosomal bgtfB gene by homologous recombination (Fig. 7.2). In bioassays, the mutant (HD1) was able to synthesise a substance which was still antibiotically active. HPLC-MS and MS-MS analyses showed that HD1 produced at least four different compounds, which were not glycosylated anymore [17, 18]. Thus the first glycopeptide biosynthetic mutant was created by a targeted approach and the role of bgtfB in balhimycin biosynthesis was proved.
7.2.3 Analysis of the balhimycin biosynthetic gene cluster Since antibiotic biosynthetic genes including resistance and regulatory genes are normally physically linked in the genome of Streptomycetes [19] isolation of the whole balhimycin gene cluster was intended by using a gene fragment of the glycosyltransferase bgtfB as probe.
7.2.3.1 Isolation and sequencing of the major part of the balhimycin biosynthetic gene cluster Hybridisation experiments with the glycosyltransferase PCR fragment (“bgtfB”) led to the identification of twelve cosmids in an A. mediterranei cosmid library. By DNA sequence, one cosmid (16.1) was characterised in detail (manuscript in preparation). An ORF analysis revealed, that a region of 46 kb carries 26 complete and 1 incomplete putative ORFs (Fig. 7.3). Computer-assisted homology searches with the sequences of the deduced gene products showed significant 143
7 Analysis of the Biosynthesis of Glycopeptide Antibiotics
Figure 7.2: Inactivation of the glycosyltransferase gene bgtfB by gene disruption resulting in mutant strain HD1. HPLC and MS-MS analyses showed that HD1 produced four different balhimycin derivatives which are all not glycosylated [17]. The chemical structure of one derivative (HD-1142) is shown in detail.
similarities to many gene products, which may participate in balhimycin biosynthesis like peptide synthetases, glycosyltransferases, cytochrome P450 mono-oxygenases, sugar biosynthetic enzymes, and an enzyme similar to chalcone/stilbene synthases of plants. The organisation of the major part of the balhimycin gene cluster is very similar to the organisation described for the putative chloroeremomycin gene cluster [5]. Differences between the two clusters are on one hand the presence 144
7.2 Results
Figure 7.3: Organisation of the major part of the balhimycin biosynthetic gene cluster and the proposed function of the encoding ORFs.
of an additional ORF (ORF 7, Fig. 7.3) in the balhimycin cluster showing similarities to Na/H antiporter. On the other hand, ORF 9 in the balhimycin cluster (Fig. 7.3), showing significant similarities to NDP-hexose-4-ketoreductases carries an in-frame deletion of 226 amino acids in comparison with the corresponding intact ORF of the chloroeremomycin cluster (manuscript in preparation). This corresponds to the occurrence of the keto-sugar residue dehydrovancosamine in balhimycin and the reduced 4-epi-vancosamine in chloroeremomycin, respectively (Fig. 7.1).
145
7 Analysis of the Biosynthesis of Glycopeptide Antibiotics 7.2.3.2 Characterisation of the function of different balhimycin biosynthetic genes by gene inactivation experiments Cytochrome P450 mono-oxygenases are involved in the coupling of the aromatic side chains of the balhimycin heptapeptide In the major part of the balhimycin cluster four ORFs (OxyA–D, Fig. 7.3) with remarkable similarities to cytochrome P450 mono-oxygenase were identified. Integration of a plasmid between oxyA and oxyB resulted in the balhimycin mutant SP1–1, which was not able to synthesise an antibiotically active compound [17]. Structural analyses by HPLC-MS, fragmentation studies, and amino acid analyses demonstrated, that SP1–1 produced two balhimycin precursors (SP969 and SP1134; Fig. 7.4), which were linear (unbridged) and not glycosylated [18, 20]. These results showed clearly that oxygenases are involved in the coupling of the aromatic side chains, which are essential to hold the peptide backbone in a rigid, cup-shaped form. This structure is important for the antibiotic activity, because only this form interacts with the D-Ala-D-Ala terminus of the cell wall precursor [2]. The halogenase BhaA is responsible for chlorination of balhimycin The product of the bhaA gene (Fig. 7.3) showed significant similarities (42%) to PrnC, a new type of non-heme halogenase of Pseudomonas fluorescens which is involved in chlorination of pyrrolnitrin [21]. Since balhimycin carries two chlorine atoms at the aromatic side chains of amino acid two and six, the participation of BhaA in chlorination of balhimycin is plausible. To prove this, a bhaA in-frame deletion-mutant (PH4) was constructed using a two-step strategy [22]. HPLC-MS analyses and the analysis of the isotopic pattern clearly showed, that the mutant PH4 was not able to synthesise chlorinated balhimycin-derivatives (Fig. 7.5). Thus bhaA is responsible for the incorporation of both chlorine atoms into the glycopeptide antibiotic. Bioassays revealed that dechlorobalhimycin is still antibiotically active (manuscript in preparation). The non-ribosomal biosynthesis of the balhimycin backbone: Eight peptide synthetase modules are involved in the biosynthesis of a heptapeptide For the non-ribosomal biosynthesis of a heptapeptide seven modules were expected. Until now, three peptide synthetase genes (bpsB–D) encoding five peptide synthetase modules were characterised in the balhimycin biosynthetic gene cluster (manuscript in preparation, Fig. 7.3). From comparison with the putative chloroeremomycin biosynthetic gene cluster [5], we conclude that a fourth peptide synthetase gene (“bpsA”) encoding three further modules is located in front of the peptide synthetase bpsB. Surprisingly, altogether eight peptide synthetase modules are expected to be located within the balhimycin biosynthetic gene cluster. 146
7.2 Results
Figure 7.4: Construction of mutant strain SP1–1 by integration mutagenesis between the cytochrome P450 mono-oxygenase genes oxyA and oxyB. HPLC-MS and MS-MS revealed that the mutant produced a linear balhimycin precursor (SP-1134) which is antibiotically inactive showing that oxygenases are involved in coupling of the aromatic rings [17, 18].
The function of three peptide synthetase genes (bpsB–D) was investigated by gene inactivation experiments and by analysing the enzymatic activities of overexpressed peptide synthetase modules (manuscript in preparation). All mutants were unable to synthesise an antibiotically active compound. The successful complementation of the bpsD replacement-mutant, encoding the eighth module, with a copy of the native bpsD gene showed that polar effects of the gene replacement could be excluded. 147
7 Analysis of the Biosynthesis of Glycopeptide Antibiotics
Figure 7.5: Construction of the mutant PH4 which carries an in-frame deletion in the halogenase gene bhaA. HPLC-MS analysis and determination of the isotopic pattern revealed that the mutant produces the antibiotically active compound dechloro-balhimycin indicating that bhaA is responsible for the chlorinating steps [21].
The results of the mutational analyses were confirmed by biochemical experiments showing that the modules 4, 5 and 7 activated the expected amino acid (manuscript in preparation). Moreover, these experiments demonstrated clearly that also the eighth module is involved in the biosynthesis of the heptapeptide core. 148
References
7.3 Outlook
The balhimycin biosynthesis includes further interesting biosynthetic steps. First experiments showed that the css gene, encoding a gene product similar to plant chalcone/stilbene synthases, is presumably involved in the biosynthesis of the seventh amino acid 3,5-dihydroxyphenyl-glycine. Furthermore, the balhimycin producer and the identified genes are suitable tools for the generation of hybrid antibiotics by rational combinatorial biosynthesis.
References
1. Cunha, B. A. (1995) Vancomycin. Med. Clin. N. Am. 79, 817–831. 2. Williams, D. H. and Bardsley, B. (1999) Die Vancomycin-Antibiotika und der Kampf gegen resistente Bakterien. Angew. Chem. 111, 1264–1286. 3. Arthur, M., Depardieu, F., Reynolds, P., and Courvalin, P. (1996) Quantitative analysis of the metabolism of soluble cytoplasmic peptidoglycan precursors of glycopeptideresistant enterococci. Mol. Microbiol. 21, 33–44. 4. Solenberg, P. J., Matsushima, P., Stack, D. R., Wilkie, S. C., Thompson, R. C., and Baltz, R. H. (1997) Production of hybrid glycopeptide antibiotics in vitro and in Streptomyces toyocaensis. Chem. Biol. 4, 195–202. 5. Van Wageningen, A. M. A., Kirkpatrick, P. N., Williams, D. H., Harris, B. R., Kershaw, J. K., Lennard, N. J., Jones, M., Jones, S. J. M., and Solenberg, P. J. (1998) Sequencing and analysis of genes involved in the biosynthesis of a vancomycin group antibiotic. Chem. Biol. 5, 155–162. 6. Nadkarni, S. R., Patel, M. V., Chaterjee, S., Vijayakumar, E. K. S., Desikan, K. R., Blumbach, J., and Ganguli, B. N. (1994) Balhimycin, a new glycopeptide antibiotic produced by Amycolatopsis sp. Y-86,21022. J. Antibiot. 47, 334–341. 7. Matsushima, P., McHenney, M., and Baltz, R. H. (1987) Efficient transformation of Amycolatopsis orientalis (Nocardia orientalis) protoplasts by Streptomyces plasmids. J. Bacteriol. 169, 2298–2300. 8. Matsushima, P. and Baltz, R. H. (1996) A gene cloning system for Streptomyces toyocaensis. Microbiology 142, 261–267. 9. Lechevalier, M. P., Prauser, H., Labeda, D. P., and Ruan, J.-S. (1986) Two new genera of nocardioform actinomycetes: Amycolata gen. nov. and Amycolatopsis gen. nov. Int. J. Syst. Bacteriol. 36, 29–37. 10. Lal, R., Lal, S., Grund, E., and Eichenlaub, R. (1991) Construction of a hybrid plasmid capable of replication in Amycolatopsis mediterranei. Appl. Environ. Microbiol. 57, 665–671. 11. Pelzer, S., Reichert, W., Huppert, M., Heckmann, D., and Wohlleben, W. (1997) Cloning and analysis of a peptide synthetase gene of the balhimycin producer Amycolatopsis mediterranei DSM5908 and development of a gene disruption/replacement system. J. Biotechnol. 56, 115–128.
149
7 Analysis of the Biosynthesis of Glycopeptide Antibiotics 12. Madon, J. and Hütter, R. (1991) Transformation system for Amycolatopsis (Nocardia) mediterranei: Direct transformation of mycelium with plasmid DNA. J. Bacteriol. 173, 6325–6331. 13. Vrijbloed, J. W., Madon, J., and Dijkhuizen, L. (1995) Transformation of the methylotrophic actinomycete Amycolatopsis methanolica with plasmid DNA: Stimulatory effect of a pMEA300-encoded gene. Plasmid 34, 96–104. 14. Uchiyama, H. and Weisblum, B. (1985) N-methyl transferase of Streptomyces erythraeus that confers resistance to the macrolide-lincosamide-streptogramin B antibiotics: Amino acid sequence and its homology to cognate R-factor enzymes from pathogenic bacilli and cocci. Gene 38, 103–110. 15. Gil, J. A., Kieser, H. M., and Hopwood, D. A. (1985) Cloning of a chloramphenicol acetyltransferase gene of Streptomyces acrimycini and its expression in Streptomyces and Escherichia coli. Gene 38, 1–8. 16. Konz, D. and Marahiel, M. A. (1999) How do peptide synthetases generate structural diversity? Chem. Biol. 6, R39–R48. 17. Pelzer, S., Süßmuth, R., Heckmann, D., Recktenwald, J., Huber, P., Jung, G., and Wohlleben, W. (1999) Identification and analysis of the balhimycin biosynthetic gene cluster and its use for manipulating glycopeptide biosynthesis in Amycolatopsis mediterranei DSM5908. Antimicrob. Agents Chemother. 43, 1565–1573. 18. Süßmuth, R., Metzger, J., and Jung, G., Chapter 19 this edition. 19. Chater, K. F. (1990) The improving prospects for yield increase by genetic engineering in antibiotic-producing streptomycetes. Biotechnology 8, 115–121. 20. Süßmuth, R., Pelzer, S., Nicholson, G., Walk, T., Wohlleben, W., and Jung, G. (1999). New advances in the biosynthesis of glycopeptide antibiotics of the vancomycin type from Amycolatopsis mediterranei. Angew. Chem. Int. Ed. Engl. 38, 1976–1979. 21. Kirner, S., Hammer, P. E., Hill, D. S., Altmann, A., Fischer, I., Weislo, L. J., Lanahan, M., van Pée, K.-H., and Ligon, J. M. (1998) Functions encoded by pyrrolnitrin biosynthetic genes from Pseudomonas fluorescens. J. Bacteriol. 180, 1939–1943. 22. Pelzer, S., Huber, P., Süßmuth, R., Recktenwald, J., Heckmann, D., and Wohlleben, W. (1999) Nucleic acid fragment and vector comprising a halogenase, and a process for halogenating chemical compounds. US patent application 19926770.7
150
8 Homologous Recombination and the Induction of the SOS-Response in Antibiotic Producing Streptomycetes Günther Muth and Wolfgang Wohlleben*
8.1 Introduction
Antibiotic production of mycel forming streptomycetes is controlled by a complex regulatory network allowing Streptomyces to sense different growth conditions and to react to changes in the environment by the production of antibiotics [2]. Antibiotic production of S. coelicolor was shown to be affected by cell density, nutritional limitations, nutritional shiftdown, imbalance in metabolism and by different kinds of stress [2]. To sense and to respond to these environmental conditions Streptomyces evolved a set of pleiotropic regulatory factors such as catabolite repression, sigma factor heterogeneity, two component systems, and pleiotropic regulatory proteins which bind small diffusable butyrolacton autoregulators. These regulatory systems were shown to crosstalk to each other, and they might be also linked to the basic cellular stress response mechanisms as stringent response (ppGpp level), N-limitation (NTR) or SOS-response to trigger antibiotic production. On the other site antibiotic production is also affected by the genetic instability of Streptomyces. Genome rearrangements and deletion of chromosomal ends were shown to result in the deletion of antibiotic pathways [8]. Recently it was demonstrated that the amplification process [14] and the deletion of chromosomal ends [4] depended on the RecA protein. RecA is the key enzyme in homologous recombination and in the induction of the SOS-response by supporting autocleavage of the LexA repressor (reviewed in [15], [7]). The recA genes of a variety of bacteria have been characterized. An alignment of the Streptomyces RecA proteins to RecA sequences from other bacteria revealed a high conservation [6]. Only the C-terminus of the Streptomyces RecA protein was about 20 amino acids longer and showed
* Mikrobiologie/Biotechnologie, Universität Tübingen, Auf der Morgenstelle 28, D-72076 Tübingen
151 Microbial Fundamentals of Biotechnology. DFG, Deutsche Forschungsgemeinschaft Copyright # 2001 WILEY-VCH Verlag GmbH, Weinheim ISBN: 3-527-30615-3
8 Homologous Recombination and the Induction of the SOS-Response no similarity to the C-termini of other RecA proteins [8]. RecA seems to play a particular role in Streptomyces. Overexpression of recA is detrimental to Streptomyces. It was only possible to reintroduce the recA gene into S. lividans on a single-copy plasmid, but not when located on a multi-copy plasmid. If recA was placed under control of an inducible promoter, induction resulted in growth impairment (unpublished data). In contrast to other bacteria, where recA could be inactivated without major problems, the recA gene of Streptomyces seemed to be essential for viability. Extensive attempts to generate recA mutants either by UV-mutagenesis [5] or by gene disruption experiments [9, 1] only allowed the isolation of mutants with residual activity. The gene disruption mutant FrecD3 which lacked the very last 86 amino acids was UV-sensitive, its recombination activity was drastically reduced and genetic instability was enhanced [9]. The recA mutant FRECD3 lost the chromosomal end containing the chloramphenicol (CM) resistance gene 70 times more frequently than the S. lividans wild type [14]. Thus it was concluded that the recombination activity of RecA might be required for the repair of single-stranded breaks occurring during the replication of the Streptomyces chromosome. Due to the linearity of the Streptomyces chromosome a single-stranded break would cause the collapse of the replication fork and result in the deletion of the respective chromosomal end. To analyze the role of RecA in Streptomyces we characterized the S. lividans recA gene by site directed mutagenesis and studied the modulation of RecA activity by the LexA repressor and by the RecX protein.
8.2 Mutational analysis of the S. lividans recA gene
8.2.1 Site directed mutagenesis of recA to discriminate the different biochemical activities of RecA A great variety of different E. coli recA mutants have been characterized in great detail. Certain amino acid exchanges were shown to differentially affect the distinct biochemical activities of the RecA protein (reviewed in [7]). Some of these amino acid residues were found to be highly conserved in other RecA proteins [3]. To identify which activity of RecA, recombination or coprotease activity was responsible for the toxic effects of recA overexpression/deficiency, several amino acid residues of the S. lividans RecA protein were exchanged (Pro67 ?Arg, Gly157 ?Asp, Arg169 ?His, Arg243 ?His) by site directed mutagenesis in analogy to well characterized E. coli recA mutants (manuscript in preparation). 152
8.2 Mutational analysis of the S. lividans recA gene A
B
Figure 8.1: 3D-model of the S. lividans RecA protein. The structure was modeled in analogy to the crystal structure of the E. coli RecA protein [11] using the Swissprod DB viewer and insight II (MSI). (A) Binding of a RecA hexamer to the DNA. (B) Location of the amino acid residues that have been mutated by site directed mutagenesis within the DNA binding domain of a RecA hexamer.
153
8 Homologous Recombination and the Induction of the SOS-Response Since the S. lividans recA gene was fully able to complement an E. coli recA deletion mutant, the effect of the different amino acid exchanges was investigated in E. coli and subsequently in the S. lividans recA mutant SV64 (see below). The mutated recA genes were placed under control of the lacZ promoter and introduced into the E. coli recA deletion strain DK1. UV-resistance was analyzed to indicate proficiency for induction of the SOS-response and recombinational repair. The ability to perform homologous recombination was studied in HFR mating assaying the mobilization of a chromosomal tetracycline resistance marker. The mutated S. lividans RecA proteins conferred only in part the expected phenotype. Whereas the E. coli strain containing Pro67/Arg was UV-sensitive, the respective S. lividans strain was UV-resistant as the wild type. Obviously the mutant protein was defective in supporting autocleavage of the E. coli LexA repressor, but proficient to interact with the S. lividans LexA homologue. Exchange of the Gly157/Asp and Arg243/His in the S. lividans RecA protein showed a very similar phenotype as reported for the respective E. coli mutants (constitutive coprotease, reduced recombination activity). Whereas the Arg169/ His substitution in the E. coli RecA protein still allowed homologous recombination at a reduced ratio, the Streptomyces RecA-Arg169/His protein was completely defective in homologous recombination. Thus, the amino acid exchanges in the S. lividans RecA protein had only in part the effects of the respective aa substitutions of the E. coli RecA. This shows that despite the high evolutionary conservation of RecA proteins the enzymatic activity is differentially affected by analogous amino acid substitutions.
8.2.2 3D-modeling of the mutated RecA proteins The X-ray structure of the E. coli RecA protein was elucidated [10]. Due to the high similarity of the S. lividans RecA protein to the E. coli RecA protein it was possible to model the 3D-structure of the S. lividans protein and to simulate the effect of the above described amino acid exchanges on the 3D-structure of the RecA filaments. Computer modeling was done using insight II (MSI) and the SWISSPROD DB viewer3.5b3. Surprisingly we found that all the mutated residues located at the inner site of RecA filaments (manuscript in preparation), probably interacting with DNA (Fig. 8.1, s. p. 153). This suggests, that small differences in the DNA-binding activity of the RecA filaments are responsible for the selective inactivation of the different biochemical functions, as LexA cleavage, recombinational repair and homologous recombination. This also explains why analogous amino acid exchanges in the S. lividans RecA protein conferred only in part the same phenotype as observed for the respective E. coli recA mutants.
154
8.2 Mutational analysis of the S. lividans recA gene
8.2.3 Construction of a recA replacement mutant Since extensive attempts to inactivate recA by gene disruption or gene replacement failed [9], we intended to replace the chromosomal recA gene while providing a second copy of recA (Fig. 8.2). The plasmid-encoded copy of recA was placed under control of the tipA promoter to allow inducible recA expression. Following the replacement of the chromosomal recA gene by the aphII casette, the recA expression vector (pEXrecA) was eliminated by plasmid incompatibil-
Figure 8.2: Construction of the S. lividans recA gene replacement mutant SV64. After elimination of the temperature-sensitive replacement plasmid and selection for the replacement of the chromosomal recA gene by the aphII cassette, the expression plasmid pEXrecA was removed using the incompatible plasmid pIJ920.
155
8 Homologous Recombination and the Induction of the SOS-Response ity: After transformation of the recA mutant with plasmid pIJ920, which carried the same replicon as pEXrecA, several colonies could be selected that had lost pEXrecA. By PCR analysis and Southern hybridization the lack of recA sequences was proved. Thus we were able to delete the chromosomal copy of recA, previously thought to be indispensable [13]. There are two possible explanations why we were able to generate a completely defective recA mutant by this procedure whereas it was not possible by the classical protocol: 1) Due to the lack of chi sequences or other unknown reasons the recA containing DNA-fragment is only poorly recombinogenic. Overexpression of the RecA protein, however, could result in an enhanced recombination activity that allows the recombination of poor substrates. 2) The recA mutant had acquired an additional, suppressing mutation which overcomes the RecA deficiency. Since the plasmid-encoded copy of recA which is under control of the tipA promoter might be just sufficient to override the toxic effect of recA deficiency, but might not be able to complement recA with wild type efficiency, a selection pressure could exist to select for such suppressing mutations. The recA mutant SV64 represents a so-called whi mutant, that is able to erect aerial mycelial, but is deficient in sporulation. Even after complementation with the wild type recA this sporulation defect is still present, whereas UV-sensitivity and recombination activity are fully complemented. This might be a clear indication that SV64 had acquired an additional mutation that could suppress the toxic effects of recA deficiency.
8.3 Regulation of RecA activity 8.3.1 Transcriptional regulation of recA by the LexA repressor In all bacteria recA is regulated by the SOS-repressor LexA. LexA binds to so called SOS Boxes (Cheo Box) in the promoter region of the genes of the SOS-response. In the putative promoter region of recA there is a sequence GAACATCCATTC which resembles the B. subtilis SOS Box GAACNNNNGTTC/T. To analyze transcriptional regulation of recA by LexA, we expressed the S. coelicolor lexA gene in E. coli as a His6-fusionprotein and purified LexA by Ni2+-NTA chromatography. A 109 bp fragment containing the putative SOS Box of the S. lividans recA gene was amplified by PCR and 3'-labeled with digoxigenin. After incubation with LexA the reaction mixture was separated on a 4% Tris-Borat-PAGE and blotted onto a nylon membrane and visualized with Anti-Dig-antibody-conjugate. The retardation of the SOS box containing fragment demonstrated that the Streptomyces LexA is able to bind the proposed SOS box [13]. 156
8.3 Regulation of RecA activity
8.3.2 Interaction of RecA with RecX 8.3.2.1 Construction of a recX mutant Downstream of the S. lividans recA gene lies an orf with significant similarity to bacterial recX genes. The function of RecX which is in many bacteria translationally coupled to recA is unknown. Since in P. aeruginosa overexpression of recA was lethal if recX was not coexpressed a regulatory role on RecA activity was suggested. To analyze the function of RecX, a recX mutant of S. lividans was constructed. A cloned recX gene was disrupted by the insertion of a tsr cassette into the single BclI site. Subsequently, the chromosomal copy of recX was replaced by the inactivated recX copy. The resulting mutant (SVX1) represents the first bacterial recX mutant that is well defined and had no additional defects [12]. SVX1 displayed the following phenotype: Homologous recombination was not affected. UV-resistance and genetic instability (determined by the percentage of CM sensitive segregants) were identical to that of S. lividans WT. However, the recX mutant showed a clearly reduced colony size (Fig. 8.3). An even more drastic phenotype was observed after overexpression of recA. In a S. lividans TK64 culture carrying the recA expression plasmid pEXrecA the colony titer under inducing conditions was about 60 % of that of uninduced cultures, indicating the toxic effect of recA overexpression. In the recX mutant however, no single colony could grow on thiostrepton-containing medium [12]. This showed that induction of the tipA promoter resulting in the overexpression of recA was lethal in the absence of RecX in S. lividans.
Figure 8.3: Morphology of the recX mutant SVX1. The recX gene of S. lividans was disrupted by a thiostrepton resistance cassette. The colony size of the resulting mutant SVX1 was approximately 30 % (area) of that of the wild type strain.
157
8 Homologous Recombination and the Induction of the SOS-Response To confirm that the overexpression of recA is responsible for the toxic effect in the SVX1 mutant, we expressed an inactive RecA protein (Arg169-His) in the mutant. The mutated recA gene was not able to complement an E. coli recA deficient mutant with regard to recombination efficiency and UV-resistance, whereas the wild type RecA protein could. The inactive recA gene was inserted into the expression plasmid under control of the tipA promoter and transferred into the mutant SVX1. After induction with thiostrepton 95% growth in comparison to non-induced conditions was observed. Obviously, overexpression of an inactive RecA protein was tolerated in the recX mutant [12]. This demonstrated that the toxic effect of recA expression is caused by the enzymatic activity of the RecA protein and not by unspecific effects of protein overexpression.
8.3.2.2 Transcriptional analysis of the recX gene In contrast to the situation in other bacteria, where recX overlaps or lies immediately downstream of recA, the recX gene of S. lividans is separated from recA by a fragment of 220 bp. The distance and a potential termination structure (DE –26.2 kJ) 65 bp downstream of recA suggested that these two genes were not cotranscribed. To assess whether both genes were expressed from a single transcript a reverse transcription polymerase chain reaction (RT-PCR) analysis was performed. Since recA is regulated by the SOS-repressor LexA, RNA was isolated at different intervals after induction with the DNA damaging methyl methanesulfonate (MMS, 25 µg ml–1). The RT reaction was performed using random nonamer oligonucleotides. Primers for subsequent PCR were designed for the specific amplification of recA, recX or recA-recX transcripts. The suitability of the primers chosen for RT-PCR was demonstrated by PCR on genomic DNA as template (Fig. 8.4, lanes “DNA”). To confirm the absence of contaminating DNA, a control PCR was performed using RNA as template. From uninduced cultures no recX transcript and only a weak band indicating basal expression of the recA gene were detected (Fig. 8.4, panel A and C, lane 0'). 20 minutes after induction the intensity of the recA-specific band increased, demonstrating induction of the recA gene during the SOS-response. Transcription of the recX gene, however, was not detectable even 20 minutes after induction. After 40 minutes and 60 minutes, expression of recA reached its maximum. Concomitantly a fragment corresponding to the recA-recX cotranscript was amplified. The recA-recX cotranscript, however, was less abundant (~10 %) than the recA transcript alone [12]. The terminator downstream of recA is probably responsible in ensuring the different transcript levels within the recAX operon.
8.3.2.3 Putative role of the RecX protein in modulating RecA activity The phenotype of the recX mutant SVX1 indicated, that RecX is required to overcome the toxic effects of recA overexpression. To analyze whether RecX 158
8.3 Regulation of RecA activity
Figure 8.4: RT-PCR analysis of the recAX operon. Following induction with MMS, RNA was isolated from S. lividans TK64 (panel A, B, C) and SVX1 (panel D) at 0', 20', 40' and 60' minutes and converted to DNA using AMV reverse transcriptase. Subsequently, PCR was performed using different primer combinations to detect the specific transcripts. Panel A and D, internal recA primers; panel B, recA-recX primers; panel C, internal recX primers; lane “DNA”: control PCR using genomic DNA as template.
functions by regulating down recA transcription, RT-PCR analysis was performed with RNA isolated from SVX1. However, recA transcription was not significantly affected in the recX mutant [12], and an induction pattern very similar to that of the wild type was observed following DNA damage (Fig. 8.4 panel D). Because RecX did not seem to influence transcription of recA we propose an interaction of RecX with the RecA protein. Since all of the biochemical functions of RecA are directly affected by the DNA binding an alteration of the binding characteristics of RecA filaments by the interaction with RecX might efficiently interfere with the specific activity of RecA.
159
8 Homologous Recombination and the Induction of the SOS-Response
References
1. Aigle, B., A. C. Holl, J. F. Angulo, P. Leblond, and B. Decaris (1997) Characterization of two Streptomyces ambofaciens recA mutants: Identification of the RecA protein by immunoblotting. FEMS Microbiol. Lett. 149, 181–187. 2. Bibb, M. (1996) The regulation of antibiotic production in Streptomyces coelicolor A3(2). Microbiology 142, 1335–1344. 3. Brendel, V., L. Brocchieri, S. J. Sandler, A. J. Clark, and S. Karlin (1997) Evolutionary comparisons of RecA-like proteins across all major kingdoms of living organisms. J. Mol. Evol. 44, 528–541. 4. Fischer, G., T. Wenner, B. Decaris, and P. Leblond (1998) Chromosomal arm replacement generates a high level of intraspecific polymorphism in the terminal inverted repeats of the linear chromosomal DNA of Streptomyces ambofaciens. Proc. Natl. Acad. Sci. U.S.A. 95, 14296–14301. 5. Harold, R. J. and D. A. Hopwood (1970) Ultraviolet-sensitive mutants of Streptomyces coelicolor: I. Phenotypic characterization. Mutat. Res. 10, 427–438. 6. Karlin, S. and L. Brocchieri (1996) Evolutionary conservation of recA genes in relation to protein structure and function. J. Bacteriol. 178, 1881–1894. 7. Kowalczykowski, S. C., D. A. Dixon, A. K. Eggleston, S. D. Lauder, and W. M. Rehrauer (1994) Biochemistry of homologous recombination in Escherichia coli. Microbiol. Rev. 58, 401–465. 8. Möhrle, V., U. Roos, and C. Bormann (1995) Identification of cellular proteins involved in nikkomycin production in Streptomyces tendae Tü901. Mol. Microbiol. 15, 561– 571. 9. Muth, G., D. Frese, A. Kleber, and W. Wohlleben (1997) Mutational analysis of the Streptomyces lividans recA gene suggests that only mutants with residual activity remain viable. Mol. Gen. Genet. 255, 420–428. 10. Story, R. M. and T. A. Steitz (1992) Structure of the recA protein-ADP complex. Nature 355, 374–376. 11. Story, R. M., I. T. Weber, and T. A. Steitz (1992) The structure of the E. coli recA protein monomer and polymer. Nature 355, 318–325. 12. Vierling, S., Weber, T., Wohlleben, W., and Muth, G. (2000) Transcriptional and mutational analysis of the S.lividans recX gene and its interference with RecA activity. J. Bacteriol. 182, 4005–4011. 13. Vierling, S., Weber, T., Wohlleben, W., and Muth, G. (2001) Evidence that an additional mutation is required to tolerate insertional inactivation of the Streptomyces lividans recA gene. J. Bacteriol. 183, 4374–4381. 14. Volff, J. N. and J. Altenbuchner (1997) Influence of disruption of the recA gene on genetic instability and genome rearrangement in Streptomyces lividans. J. Bacteriol. 179, 2440–2445. 15. West, S. (1992) Enzymes and molecular mechanism of genetic recombination. Annu. Rev. Biochem. 61,603–640.
160
Membrane Processes
Microbial Fundamentals of Biotechnology. DFG, Deutsche Forschungsgemeinschaft Copyright # 2001 WILEY-VCH Verlag GmbH, Weinheim ISBN: 3-527-30615-3
9 Regulated Transport and Signal Transfer Channels involved in Bacterial Iron Supply Volkmar Braun* and Helmut Killmann
9.1 Introduction
During studies on the structure and function of the phage receptor TonA (for phage T1, T one), we discovered in 1975 that the receptor participates in the transport of the iron complex ferrichrome [1]; TonA was renamed FhuA for ferric hydroxamate (ferrichrome) uptake. At that time it was known that the product of an additional gene, termed tonB, is required for the FhuA-dependent infection of Escherichia coli by certain phages, and we were soon able to show that the electrochemical potential of the cytoplasmic membrane is somehow transmitted by TonB to the outer membrane protein FhuA and serves to activate FhuA [2]. The study of energy-consuming active import not only of iron complexes, but also of bacterial protein toxins (colicins), to gain insight into the function of protein transporters and the transfer of energy from one membrane to an adjacent membrane, became a major topic of our laboratory. The Fe2+/Fe3+ pair displays a wide range of redox potentials from –300 to +700 mV, depending on the iron ligands and the protein environment. All organisms, with the exception of certain lactobacilli, take advantage of this unusually wide range of electron transport capacity. However, despite its high abundance in nature, iron is difficult to acquire by most organisms. Under oxic conditions and at the physiological pH of 7.0, the concentration of free ferric ions in equilibrium with the ferric hydroxide polymer is in the order of 10–12 µM. The growth-promoting concentration of iron for microbes is approximately 10–1 µM. Bacteria contain about 105 iron ions per cell. Since bacteria usually reach densities of 109 cells per ml, they require 1014 iron ions per generation, which sharply contrasts the available 103 free ions. Bacteria have solved the iron supply problem by synthesizing iron-chelating compounds, called siderophores, which
* Mikrobiologie/Biotechnologie, Universität Tübingen, Auf der Morgenstelle 28, D-72076 Tübingen
163 Microbial Fundamentals of Biotechnology. DFG, Deutsche Forschungsgemeinschaft Copyright # 2001 WILEY-VCH Verlag GmbH, Weinheim ISBN: 3-527-30615-3
9 Regulated Transport and Signal Transfer Channels they secrete and take up again after loading with Fe3+. The bacteria employ highly specific and efficient transport systems for ferric siderophore uptake. In addition to the siderophores they synthesize themselves, bacteria can use siderophores synthesized and secreted by other bacteria and fungi or iron sources of their hosts (Fig. 9.1). In the human body, iron is bound to proteins: transferrin in the serum, lactoferrin in secretory fluids and in granules of polymorphonuclear leukocytes, and intracellular ferritin. Bacteria take up the free iron in equilibrium with transferrin and lactoferrin via siderophores, or they take up free heme, heme from hemoglobin and hemopexin, and iron from transferrin and lactoferrin without the involvement of siderophores. Heme and iron is released from the host proteins at the bacterial cell surface.
Figure 9.1: Iron transport systems of Gram-negative bacteria. Transport proteins located in the outer membrane (OM), periplasm (PP), and in the cytoplasmic membrane (CC) are symbolized by filled barrels. Siderophores are produced by some species and released as indicated by the dashed arrow. Acquisition of ferric iron from lactoferrin, transferrin, heme, hemoglobin, hemopexin, and ferric siderophores, or of ferrous iron is indicated by solid arrows. The Ton complex composed of TonB, ExbB, ExbD is indicated by empty barrels. Fe2+ is transported by the Feo system consisting of a large and two small transport proteins.
164
9.2 The Fhu proteins catalyze active transport of ferrichrome E. coli K-12, the most widely used laboratory strain, synthesizes six distinct siderophore-mediated Fe3+ transport systems, each of which recognizes a single type of ferric siderophore, and one Fe2+ transport system. However, E. coli K-12 synthesizes only one siderophore, enterobactin (cyclic trimer of 2,3-dihydroxybenzoylserine, also designated enterochelin). Two of the transport systems take up the biosynthetic precursors of enterobactin, 2,3-dihydroxybenzoic acid and 2,3-dihydroxybenzoylserine [3]. One transport system is specific for ferric citrate, although the concentration of citrate released from E. coli K-12 into the growth medium under laboratory conditions (not more than 15 µM) is not sufficient to induce the system (threshold concentration: 100 µM citrate). Citrate is an exogenous siderophore, as are ferrichrome and coprogen of fungal origin, all of which are also used as siderophores by E. coli. Pathogenic E. coli strains frequently synthesize aerobactin and a ferric-aerobactin-specific transport protein, whose genes are encoded on plasmids. The many iron transport systems in E. coli and in other bacteria probably reflect the iron sources to which the bacteria are exposed in their various environments to which they have to be able to adapt. In the following, the systems of iron transport via ferrichrome in E. coli and Bacillus subtilis, citrate in E. coli, and heme in Yersinia enterocolitica will be described as they represent the first systems studied and those studied in most detail and are typical for Fe3+-siderophore transport systems in these organisms and in other bacteria. The reader is referred to recent references for more comprehensive reviews [4–14].
9.2 The Fhu proteins catalyze active transport of ferrichrome and the antibiotic albomycin across the outer membrane and the cytoplasmic membrane of E. coli
9.2.1 Transport across the outer membrane – the need for receptor proteins Gram-negative bacteria such as E. coli are surrounded by an outer membrane, the inner cytoplasmic membrane, and the periplasm between the two membranes (Fig. 9.1). Most substrates diffuse across the outer membrane through the water-filled channels of the porins, which in E. coli have an exclusion limit of approximately 600 Da [15]. In contrast, the Fe3+-siderophores, heme, and vitamin B12 are actively transported across the outer membrane by proteins (Fig. 9.1). The Fe3+-siderophores with molecular masses of 700–1000 Da are too large to diffuse through the porins. The concentration of Fe3+-siderophores is 165
9 Regulated Transport and Signal Transfer Channels also too low for a sufficient iron supply if one considers the strong dilution of the exported siderophores in the environment of the cells. Competition of the siderophores for iron with other strong iron-chelating compounds, such as siderophores produced by bacteria and fungi that cannot be used by a particular strain, or proteins produced by the infected hosts, e. g. transferrin, lactoferrin, and ferritins, further lowers the Fe3+-siderophore concentration. Fe3+-siderophores that make contact with bacterial cells are captured by their cognate receptor proteins. By this means, the Fe3+-siderophores are extracted from the medium and concentrated at the bacterial cell surface. The low KD of the receptors (below 0.1 µM) favors binding of the Fe3+-siderophores. A function similar to that of the receptor proteins at the cell surface of Gram-negative bacteria is displayed by the binding proteins, which are anchored by a lipid moiety of the murein-lipoprotein type to the outer face of the cytoplasmic membrane of Gram-positive bacteria, as demonstrated for the ferrichrome transport in B. subtilis [16]. In Gram-positive bacteria, the Fe3+-siderophores diffuse through the multi-layered murein and teichoic acids and gain direct access to the Fe3+-siderophore transport systems in the cytoplasmic membrane.
9.2.2 The FhuA transport protein forms a regulated channel in the outer membrane of E. coli FhuA was originally chosen and continuously studied since it is a multifunctional protein, and the mechanisms of energy-dependent phage infection, active transport of ferric siderophores and the antibiotics albomycin and rifamycin CGP 4832, and the import of the protein toxin colicin M and of the peptide toxin microcin J25 by FhuA can be studied. At the beginning of the studies, it was known that a function encoded by the tonA gene, later designated as the fhuA gene [17], is required for infection by the phages T1, T5, and f80 and for the sensitivity of cells to colicin M. In 1975, we and the Neilands group independently showed that FhuA is involved in ferrichrome transport [1, 18] and that mutants resistant to albomycin are mostly impaired in albomycin and ferrichrome transport. This provided a convenient way to characterize the transport across the outer membrane and the cytoplasmic membrane. Transport genes and their products were identified, and the transport proteins were localized. Evidence for a channel in the FhuA outer membrane protein involved in ferrichrome transport has been obtained by excision of DNA fragments from the fhuA structural gene. Deletion of residues 322–355 (FhuAD322–355) results in a stable protein that is integrated into the outer membrane and contains an open channel through which ferrichrome enters cells at a rate proportional to the external ferrichrome concentration, without showing saturation at higher ferrichrome concentrations. Transport across the cytoplasmic membrane is not rate limiting. At a low ferrichrome concentration (1 µM), FhuAD322–355 supports iron uptake much less than the FhuA wild type. At higher ferrichrome concen166
9.2 The Fhu proteins catalyze active transport of ferrichrome trations (above 7 µM), iron uptake through FhuAD322–355 exceeds uptake via the FhuA wild type, as one would expect for a diffusion-controlled process. FhuAD322–355 virtually does not bind ferrichrome. FhuAD322–355 incorporated into artificial lipid bilayers (black lipid membranes) forms open channels with a single channel conductance more than three times as high as that of E. coli porins [19]. Cells that synthesize FhuAD322–355 can grow on maltotetraose and maltopentaose in the absence of the LamB protein through which maltodextrins diffuse across the outer membrane, and they are sensitive to SDS and bacitracin [19] because the maltodextrins and noxious agents diffuse through the outer membranes via the open FhuA channels. Segment 322–355 of FhuA is exposed to the cell surface, as revealed by proteolysis of FhuA by added proteases after insertion of a tetrapeptide or a hexadecapeptide after residue 321, which results in ferrichrome-transport-active FhuA derivatives [20]. FhuA contains four cysteine residues that form disulfide bridges [21]. After cleavage of the disulfide bridge with mercaptoethanol between the two cysteines at positions 318 and 329, FhuA in living cells reacts weakly with biotinmaleimide; however, a newly introduced cysteine at position 336 is highly reactive, which demonstrates its exposure near the cell surface [21]. Deletion of a single amino acid, Asp 348, inactivates ferrichrome transport activity of FhuA and therefore is in a region important for FhuA activity [22].
9.2.3 FhuA as phage receptor FhuA not only catalyzes transport of ferrichrome and of the structurally related antibiotic albomycin, but also serves as a receptor for the infection of a number of phages and for killing by toxic proteins (Fig. 9.2). The large variety of ligands that bind to the multifunctional FhuA protein makes FhuA a particularly attractive receptor for studying structure-function relationships. Cells that synthesize FhuAD322–355 and FhuAD335–355 are resistant to the phages T1, T5, and f80 (UC-1 was not tested) and to colicin M (microcin 25 was not tested). This shows that this segment is involved in binding of the phages and in binding and uptake of colicin M. The binding sites of the phages in the region 316–355 were demonstrated by competitive peptide mapping [23]. This method avoids long-range conformational changes, caused by mutations in the gating loop, at regions outside the gating loop that may be involved in phage binding. Synthetic hexapeptides that span the entire region were employed, and those identical to sequences of three segments inhibited infection by phages T1, T5, and f80. Similar results were obtained with pentapeptides comprising residues 316–320, 332–337, 348–351, and with selected tetrapeptides (Killmann, H., unpublished results). The hexapeptides not only interfere with phage binding, but inactivate the phages by triggering DNA release from the phage head [23]. The peptides mimic binding of phage T5 to the entire isolated FhuA protein, which causes DNA release from 167
9 Regulated Transport and Signal Transfer Channels
168
9.2 The Fhu proteins catalyze active transport of ferrichrome the phage as shown by electron microscopy. The peptides also trigger release of DNA from phage f80, in contrast to the entire FhuA molecule, which does not cause DNA release from f80 [23, 24] or f80 inactivation [23]. DNA release caused by the peptides requires input of thermal energy since, under the conditions used, all phages are inactivated at 37 8C and half of them remain intact at 20 8C. After the release of DNA caused by FhuA, phage T5 largely maintains its shape. In contrast, the peptides cause a collapse of the phage head and release of the phage tail. The stronger effect of the peptides compared to that of FhuA can be explained by their flexible conformation, which enables them to adapt to the surface of the phage tail region with which the phages bind to FhuA. One of the peptide conformations is the one that the FhuA segment assumes when energized by the proton-motive force across the cytoplasmic membrane via the Ton complex (see Section 9.3). This energized conformation is required for binding or uptake of all FhuA ligands except phage T5. In contrast, the entire FhuA protein assumes a defined conformation with less flexibility.
9.2.4 Sequence comparison of the FhuA proteins of various Enterobacteriaceae reveal conserved sites important for FhuA activity Comparison of the ferrichrome transporters of E. coli K-12 [25, 26], Salmonella paratyphi, Salmonella typhimurium, and Pantoea agglomerans (formerly Erwinia herbicola) [27] support the division of the E. coli FhuA segment 322–355 into two regions. Mutant FhuAD322–336 still transports ferrichrome, but is resistant to phage T1, and FhuAD335–355 does not transport ferrichrome and is resistant to the phages T1, T5 and f80. FhuA of S. typhimurium lacks residues 321–337 (E. coli FhuA numbering) and P. agglomerans lacks residues 318–331 and 338– 341 [27]. Both strains are resistant to the E. coli phages, but transport ferrichrome. S. paratyphi lacks none of the amino acids, transports ferrichrome, and is sensitive to the E. coli phages. The entire segment 322–355 is therefore required for infection by the phages, but residues 322–336 are dispensable for ferrichrome transport.
3 Figure 9.2: Upper panel (A): Crystal structure of the FhuA protein in the unloaded and albomycin-loaded form [31]. Note the large structural change that occurs at the periplasmic side of FhuA when loaded with albomycin. Lower panel (B): Ligands of the multifunctional FhuA protein in the outer membrane (OM) and transmembrane topology of the TonB, ExbB, and ExbD proteins anchored to the cytoplasmic membrane (CM) and extending into the periplasm (PP). A predicted gating loop that controls the permeability of FhuA is indicated in the cork domain. Interaction of TonB with FhuA is proposed to occur via a region in which residue 160 resides. N designates the N-terminus, and C designates the C-terminus of the proteins. T5, T1, f80, and UC-1 designate phages that use FhuA as a receptor to infect E. coli cells.
169
9 Regulated Transport and Signal Transfer Channels Deletion of fragment 236–248 of the E. coli FhuA [27], which is identical in all four species, abolishes ferrichrome binding and transport. As will be shown below, this region contains a ferrichrome binding site.
9.2.5 Crystal structure of the FhuA transporter in free form and loaded with ferrichrome Initial attempts of the laboratory of Prof. W. Welte of the University of Konstanz to crystallize our cloned FhuA protein yielded only crystals with a diffraction of 8 Å. Later, Andrew Ferguson of the same laboratory obtained crystals with a diffraction of 2.6 Å of a FhuA derivative with a His-tag at residue 405 [28, 29]. Locher et al. [30] succeeded in the determination of the crystal structure of our cloned FhuA protein purified without a His-tag [30]. According to the results of these two groups, FhuA assumes a novel structure (Fig. 9.2, left panel) in which residues 160–714 form a bbarrel composed of 22 antiparallel b-strands. The b-barrel is closed by residues 19– 159, which form a globular structure and enter the b-barrel from the periplasmic side. Since this portion entirely closes the b-barrel, it has been designated the cork or plug. Residues 1–18 are not seen and presumably are flexible. The crystal structure reveals that the above-described segment 322–336 indeed forms a loop (L4), which is the most prominent loop at the cell surface, consistent with the finding that L4 serves as a binding site for the phages and colicin M. In the crystal structure, ferrichrome is bound at a site located outside the outer membrane bilayer (Fig. 9.2, right panel shown with albomycin instead of ferrichrome). Four amino acid residues of the cork and six residues of the b-barrel come close enough to ferrichrome to form hydrogen bonds and van der Waals contacts (Table 9.1). The aromatic residues lining the inner walls of the external pocket of FhuA probably extract ferrichrome from the external medium. Upon binding of ferrichrome, a portion of the cork moves by 1.7 Å towards ferrichrome, accompanied by a large structural transition at the periplasmic side where a short a-helix of the cork is unwound, and Glu19 moves 17.3 Å away from the position it occupies in ferrichrome-unloaded FhuA. Despite this large structural change across the entire FhuA molecule and the thickness of the outer membrane, the b-barrel is not converted to an open channel.
9.2.6 FhuA-albomycin is the first example of an antibiotic-proteintransporter crystal structure Most antibiotics diffuse into bacteria. Their efficiency, as measured by the minimal inhibitory concentration (MIC), is determined by the diffusion rate and the activity at the target sites. Gram-negative bacteria are usually less sensitive to 170
9.2 The Fhu proteins catalyze active transport of ferrichrome Table 9.1:
Interactions of FhuA with its ligands a. Ligand
Residue
Ferricrocin
Albomycin b
Albomycin c
Arg 81
Arg 81
Arg 81
Gly 99 Gln 100
Gly 99 Gln 100
Gly 99 Gln 100
Tyr 116 Tyr 244 Trp 246 Tyr 313 Tyr 315
Tyr 116 Tyr 244 Trp 246 Tyr 313 Tyr 315 Lys 344 Phe 391
Phe 391
Tyr 393
Phe 693 a b c
Phe 557 Phe 558 Phe 693
Phe 115 Tyr 116 Tyr 244 Trp 246 Tyr 313
Phe 391 Tyr 393 Tyr 423 Gln 505
Phe 693
CGP 4832
Glu 98 Gly 99 Gln 100 Ser 101 Phe 115 Tyr 116 Tyr 244 Trp 246 Tyr 313 Tyr 315 Lys 344 Phe 391 Gly 392 Tyr 423 Gln 505
Phe 693 Tyr 696
Distance within 4 Å forming hydrogen bonds and van der Waals contacts; Extended conformation; Compact conformation.
antibiotics than Gram-positive bacteria because they contain an outer membrane that functions as a permeability barrier. However, if antibiotics are actively transported across the outer membrane, their MIC may be lower in Gramnegative bacteria than in Gram-positive bacteria because the antibiotic is accumulated in the periplasm and forms a steep concentration gradient across the cytoplasmic membrane into the cytoplasm, which enhances the diffusion rate, or it may even be actively transported across the cytoplasmic membrane. Both uptake routes have been investigated. The crystal structure of FhuA loaded with albomycin (Fig. 9.2, right panel) reveals that the Fe3+-hydroxamate portion of albomycin occupies the same site on FhuA and is bound by the same amino acid chains as ferrichrome [31] (Table 9.1). In Table 9.1, ferricrocin is shown instead of ferrichrome; ferricrocin has the same structure as ferrichrome except that one of the glycine residues is replaced by a serine residue. Ferricrocin is produced by Aspergillae, functions as an Fe3+-ligand, and is transported as well as ferrichrome. The predominant binding sites of albomycin on FhuA are by far aromatic residues (69%). Binding of the thioribosyl pyrimidine moiety occurs in the external pocket and involves residues Phe 115, Lys 344, Tyr 393, Tyr 423, and Gln 505. These additional bind171
9 Regulated Transport and Signal Transfer Channels ing sites of albomycin as compared to those of ferricrocin do not prevent release of albomycin from FhuA and transport through FhuA. The albomycin transport rate is half the transport rate of ferrichrome, but it is not clear whether the lower rate is caused by transport across the outer membrane or across the cytoplasmic membrane. The structure of the FhuA-albomycin co-crystal also reveals the hitherto unknown conformation of albomycin and the conformation in the transport-competent form. The most unexpected result is the existence of two albomycin conformations in the crystal – an extended and a compact conformation. Both conformations fit into the external cavity of FhuA and occupy different amino acid ligands (Table 9.1). The solvent-exposed external cavity of FhuA is sufficiently large to accommodate the voluminous side chain bound to the Fe3+-hydroxamate moiety of albomycin. With the modular composition of albomycin, in which the iron carrier is linked by a peptide linker to the antibiotically active thioribosyl pyrimidine, nature provides a clue of how to design highly efficient antibiotics that are actively transported into bacteria. Such antibiotics could be synthetically assembled from Fe3+-hydroxamates, which fit into the active center of the transporters, and from an antibiotic that diffuses too slowly into cells to be useful itself as a drug. The FhuA-albomycin structure demonstrates that the water-filled cavities in transporters can tolerate rather large antibiotics that are structurally unrelated to the carrier. The tolerance to the antibiotic structure is not confined to FhuA; albomycin is also transported very well across the cytoplasmic membrane and during this process is recognized by the FhuD and the FhuB proteins.
9.2.7 Crystal structure of FhuA with bound rifamycin CGP 4832 In 1987, a group at Ciba-Geigy reported on a semisynthetic rifamycin derivative, CGP 4832, with an activity against many Gram-negative bacteria 200-fold higher than unmodified rifamycin [32]. It was then shown with our mutants [33] that CGP 4832 is transported by FhuA across the outer membrane of E. coli and that TonB activity is required [34]. Mutants in the fhuBCD genes, which encode the proteins required for active transport of ferrichrome across the cytoplasmic membrane, display unaltered CGP 4832 sensitivity. Our attempts to find additional transport mutants only revealed mutations in fhuA and tonB exbB exbD, which suggests that CGP 4832 crosses the cytoplasmic membrane by diffusion rather than by transport [35]. The use of FhuA as transporter for CGP 4832 is surprising since CGP 4832 does not contain iron and has no structural resemblance to ferrichrome. Therefore, it was particularly of interest to determine the crystal structure of FhuA loaded with CGP 4832 [35]. Analysis of the X-ray diffraction data reveals that CGP 4832 largely occupies the site in FhuA that is also used by ferrichrome (Table 9.1). Interestingly, the amino acid residues Gly 99, Tyr 116, and Tyr 244, which also bind ferri172
9.2 The Fhu proteins catalyze active transport of ferrichrome chrome, recognize those side chains in which CGP 4832 differs from unmodified rifamycin. In contrast to ferrichrome and albomycin, CGP 4832 in the crystal does not cause the large structural change in the periplasmically oriented pocket of FhuA. This does not seem to be caused by restriction of FhuA movements in the crystal since binding of CGP 4832 to FhuA in solution does not result in intrinsic FhuA tryptophan fluorescence quenching [35], as is observed when FhuA binds ferrichrome [35, 36]. This finding has an impact on the concept of how FhuA interacts with TonB since, as discussed above, the large structural transition is thought to facilitate interaction of FhuA with TonB. Since transport of CGP 4932 depends on TonB, interaction of TonB with FhuA may occur in the absence of the structural change. Analysis of a FhuA deletion derivative described below supports this conclusion.
9.2.8 The b-barrel domain of FhuAD5–160 is sufficient for TonB-dependent FhuA activities The FhuA crystal structure reveals that residues 160 to 714 of the mature protein form a b-barrel that is closed from the periplasmic side by the globular N-proximal fragment, residues 1 to 159 designate the cork. We deleted the cork with the idea that the resulting FhuAD5–160 protein might form an open channel [37]. To be active, the remaining b-barrel domain has to be stable and exported across the cytoplasmic membrane into the outer membrane. If the b-barrel lacking the cork does not collapse, one could expect that it would form a permanently open channel larger than the channels of the previously described FhuAD322–355 and FhuAD335–355 deletion derivatives [19, 38]. Such a channel would allow diffusion, but no active transport of ferrichrome, and would confer sensitivity to SDS and antibiotics that are too large to diffuse through the porin channels. One could anticipate that FhuAD5–160 might still function as a receptor for the TonB-independent infection by phage T5, provided loop 4 retains its conformation. In fact, deletion of the cork results in a stable protein, FhuAD5– 160, that is incorporated into the outer membrane. Cells that synthesize FhuAD5–160 display a higher sensitivity to large antibiotics such as erythromycin, rifamycin, bacitracin, and vancomycin, and grow on maltotetraose and maltopentaose in the absence of LamB. High concentrations of ferrichrome support growth of a tonB mutant that synthesizes FhuAD5–160. These results demonstrate non-specific diffusion of compounds across the outer membrane of cells that synthesize FhuAD5–160 [37]. However, growth of a FhuAD5–160 tonB + strain occurs at low ferrichrome concentrations, and ferrichrome is transported at about 45% of the FhuA wild type rate, despite the lack of ferrichrome binding sites provided by the cork. FhuAD5–160 confers sensitivity to phages T1 and f80 and the uptake of colicin M at levels as high or nearly as high through wild type FhuA; FhuAD5–160 also 173
9 Regulated Transport and Signal Transfer Channels confers sensitivity to albomycin and rifamycin CGP 4832. These data confirm the high activity of FhuAD5–160 and its strong dependence on TonB [37], despite the lack of the TonB box (residues 7 to 11) previously implicated in the interaction of FhuA with TonB. FhuAD5–160 still functions as a specific transporter and as a phage receptor, and sites in addition to the TonB box are involved in the TonBmediated response of FhuA to the proton gradient of the cytoplasmic membrane. It is proposed that TonB interacts with the TonB box of FhuA and with the b-barrel to release ferrichrome from the FhuA binding sites and to open the channel in FhuA. For transport of ferrichrome through the open channel of FhuAD5–160, interaction of TonB with the b-barrel is sufficient to release ferrichrome from the residual binding sites at the b-barrel and to induce the active conformation of the L4 loop at the cell surface for infection by the TonB-dependent phages T1 and f80. For the infection by phages T1 and f80, induction of the infection-competent conformation of loop 4 by TonB cannot be mediated by the cork domain, but must be transmitted by the b-barrel to the ferrichrome binding pocket close to the cell surface. This implies that the interaction of FhuA with TonB is propagated from the periplasm through the b-barrel up to loop 4 [37]. The activity of FhuAD5–160 sheds light on the mode of action of wild type FhuA. There is more than a single interaction site between FhuA and TonB. Interaction of TonB with the TonB box of FhuA might impose the structural change in the cork that is required to open a channel; without this interaction, the cork tightly closes the channel lumen of FhuA. This structural change may contribute to the release of ferrichrome from its FhuA binding site. However, the structural change in the cork might only alter the orientation of the cork amino acids that contribute to ferrichrome binding and not affect the amino acids at the b-barrel that bind ferrichrome. For the complete release of ferrichrome, the b-barrel also changes its structure in response to TonB. As our results demonstrate, TonB can activate FhuAD5–160 exclusively through the b-barrel, and we propose that this occurs by a conformational transition that changes the orientation of the aromatic residues such that they no longer bind ferrichrome. These findings imply that the b-barrel does not form such a rigid structure that it cannot change the conformation to alter its activity. Export of FhuAD5–160 across the cytoplasmic membrane and insertion into the outer membrane may also shed light on the assembly of wild type FhuA. Obviously, the FhuA barrel can be assembled in the absence of the cork; this suggests that the barrel is constructed first and then the cork is introduced into the pre-formed barrel. This conclusion is supported by the stability of the FhuAD5–160 b-barrel to intracellular and added proteases as high as the stability of wild type FhuA. FhuA, like many other outer membrane proteins, contains at its C-terminus a phenylalanine; in the outer membrane protein PhoE, this amino acid is important for the formation of an assembly-competent folded monomer. In FhuA, the C-terminal Phe residue is part of the barrel, which may fold in the periplasm and insert into the outer membrane prior to the insertion of the cork.
174
9.3 Transduction of energy from the cytoplasmic membrane
9.3 Transduction of energy from the cytoplasmic membrane into the outer membrane for the activation of FhuA as a transporter and phage receptor
Resistance of cells that synthesize wild type FhuA to SDS and large antibiotics and the failure of the cells to grow on maltodextrins when LamB is lacking indicate that FhuA does not form a permanently open channel with an inner diameter larger than that of the porins. Cells that synthesize the FhuAD5–160, FhuAD322–355, or FhuAD335–355 deletion protein have an increased antibiotic sensitivity, allow diffusion of ferrichrome, and grow on maltodextrins in the absence of LamB, which indicate that the FhuA channels are open. The question arises how the FhuA wild type channel is opened physiologically. The first indication how this may happen and still the best evidence obtained comes from studies on the infection of phage T1 [2]. Phage-resistant cells were mutated in two genes, designated tonA and tonB (ton from T one). tonA was later renamed fhuA [17] to indicate its physiological activity in ferric hydroxamate (ferrichrome) uptake. Early studies at the time when phage genetics opened a field now known as molecular biology revealed that phage T1 adsorbs to tonB mutants reversibly and that cellular energy is required for irreversible adsorption accompanied by infection. Phage T1 does not bind to tonA (fhuA) mutants, suggesting that FhuA determines the primary binding site and TonB defines a later step in infection (cited in [2]). The nature of the energy required for irreversible adsorption of phage T1 (and f80) was then shown to be the electrochemical potential of the cytoplasmic membrane [2]. Correlation of the energy and TonB requirement suggests that TonB is somehow involved in coupling the energy conserved in the transmembrane potential of the cytoplasmic membrane to irreversible adsorption of phage T1 to the outer membrane. Later, the fhuA gene product was shown to be a protein in the outer membrane [39] and that phage T5, which does not require energy and TonB for productive adsorption (infection), is inactivated by the isolated protein, in contrast to phages T1 and f80, which do require TonB in energized cells [40]. Phage T5 inactivation is prevented by colicin M, which suggests a common binding site for phage T5 and colicin M [1]. This has been proven with point mutations and deletions in the region 316–355 and by competitive peptide mapping [19, 22, 23, 38]. The relationship between the adsorption step and the requirements for energy and the TonB activity was shown by isolating phage T1 host mutants that infected tonB mutants. Infection by these mutants demonstrates that transfer of DNA across the outer membrane does not require TonB activity. The energized cytoplasmic membrane induces an infection-competent conformation in FhuA through the activity of TonB. This conformation is recognized by phages T1 and f80, while phage T5 recognizes the unenergized conformation. Evidence for two FhuA conformations has also been obtained by competition studies between phage T5 and ferrichrome at FhuA. Ferrichrome inhibits phage T5 adsorption to FhuA of unenergized cells 175
9 Regulated Transport and Signal Transfer Channels (tonB mutants or energy-deprived tonB+ cells) more strongly than adsorption to FhuA of energized cells [41]. It is likely that all Ton-dependent receptors undergo similar conformational changes upon energization, which results in translocation of the ferric siderophores across the outer membrane. Transport of ferrichrome across the outer membrane into the periplasm of E. coli was determined with a fhuB mutant devoid of ferrichrome transport across the cytoplasmic membrane. The periplasmic FhuD binding protein had to be overproduced to measure accumulation of radioactive [55Fe3+]ferrichrome in the periplasm [42]. About 8000 ferrichrome molecules per cell bind to FhuD and can be chased with non-radioactive ferrichrome. Under the same conditions, 100 000 ferrichrome molecules are taken up into the cytoplasm of a fhuB+ strain and can no longer be chased because iron is released from ferrichrome and incorporated into heme and non-heme iron proteins and into the undefined iron pool of the cell. For energization of outer membrane transport, the electrochemical potential (proton-motive force) of the cytoplasmic membrane is required. Three proteins are known to be involved in the transduction of energy from the cytoplasmic membrane into the outer membrane, where FhuA is activated to transport ferrichrome, albomycin, rifamycin CGP 4832, colicin M, and microcin J25, and to serve in the energized state as receptor for the infection by phages T1 and f80 (Fig. 9.2). The proteins are TonB, ExbB, and ExbD. tonB mutants are devoid of all FhuA activities, except infection by phage T5, while exbB and exbD mutants display residual activities. For this reason, an accessory role was ascribed to the ExbB and ExbD proteins until it was shown that two other proteins, TolQ and TolR, can partially substitute for ExbB and ExbD, respectively [43–45]. exbB tolQ and exbD tolR double mutants are both completely inactive, as are tonB deletion mutants. tonB mutants do not accumulate ferrichrome in the periplasm of a fhuB mutant [42], and uptake in cells that synthesize FhuAD322–355 does not require TonB, which suggests that TonB is required only for active transport across the outer membrane. The transmembrane topology of TonB (Fig. 9.2) further supports its role as an energy transducer between the cytoplasmic and the outer membranes. TonB is anchored by the N-terminal end to the cytoplasmic membrane and extends into the periplasm [46]. Interaction of TonB with outer membrane receptors has been demonstrated by mutations close to the N-terminal end of receptors, in the so-called TonB box, which are suppressed by mutations at residue 160 of TonB [47–49]. Furthermore, overproduced FhuA prevents proteolytic degradation of overproduced TonB in cells only when both proteins functionally interact with each other [50]. Evidence for a structural complex between TonB, ExbB, and ExbD is based on the inhibition by ExbB of TonB and ExbD degradation by cellular proteases [46, 51]. A fragment of TonB, consisting of only 44 N-terminal residues, is still stabilized by ExbB [52]. Since this fragment consists mainly of the portion of TonB that spans the cytoplasmic membrane, it is likely that interaction between the two proteins occurs within or close to the cytoplasmic membrane. Weak suppression of a TonB derivative lacking Val-17 by an Ala-39-to-Glu mutation in the first transmembrane region of ExbB further indicates functional in176
9.3 Transduction of energy from the cytoplasmic membrane teraction of the two proteins in the cytoplasmic membrane [53]. Furthermore, TonB and ExbB can be chemically cross-linked to each other in cells [53]. In vitro evidence for interaction of the three proteins has been obtained by binding isolated ExbB which carried a His-tag at the C-terminal end to a nickel-nitrilotriacetate agarose column. His-tag ExbB specifically retains ExbD and TonB on the column [54]. Although the N-terminal transmembrane segment of TonB and the region around residue 160 are important functional regions of TonB, they are certainly not the only regions that determine TonB activity. Mutations in the C-terminal region beyond residue 160 (E. coli TonB consists of 239 residues) strongly affect uptake of ferrichrome, sensitivity to certain colicins and phages [52, 55], and the activity of certain mutated FhuA proteins with phage T5 [56]. ExbD is, like TonB, anchored by the N