Myxobacteria: Multicellularity and Differentiation

Citation preview

-Myxobacteria Multicellularity and Differentiation

EDITED BY

DAVID E. WHITWORTH

Department of Biological Sciences, University of Warwick, Coventry, United Kingdom

ASM PRESS Washington, DC

Copyright 0 2008

ASM Press American Society for Microbiology 1752 N Street, N.W. Washington, DC 20036-2904

Library of Congress Cataloging-in-Publication Data Myxobacteria :multicellularity and differentiation / edited by David E. Whitworth. p. ;cm. Includes index. ISBN 978-1-5558 1-420-5 1. Myxobacterales. 2. Cell differentiation. I. Whitworth, David E. 11. American Society for Microbiology. [DNLM: 1. Myxococcales. 2. Cell Differentiation. QW 150 M999 20081 QR82.M95M98 2008 579.3’2-dc22 2007038056

All Rights Reserved Printed in the United States of America 1 0 9 8 7 6 5 4 3 2 1 Address editorial correspondence to: ASM Press, 1752 N St., N.W., Washington, DC 20036-2904, U.S.A. Send orders to: ASM Press, P.O. Box 605, Herndon, VA 20172, U.S.A. Phone: 800-546-2416; 703-661-1593 Fax: 703-66 1- 1501 Email: [email protected] Online: estore.asm.org

Contents

Contributors Preface xu

...

vaaa

I Myxobacterial Biology

1

1 From Glycerol to the Genome 3 DALEKAISERAND MARTINDWORKIN 2 Why Cooperate? The Ecology and Evolution of Myxobacteria 17 GREGORY J. VELICER AND KRISTINAL. HILLESLAND

II Development and Motility

41

43 3 Initiation and Early Developmental Events MICHELLE E. DIODATI,RONALD E. GILL,LYNDAPLAMANN, AND MITCHELL SINGER 4

Contact-Dependent Signaling in Myxococcus xanthus: the Function 77 of the C-Signal in Fruiting Body Morphogenesis LOTTES0GAARD-ANDERSEN

5

Reversing Myxococcus xanthus Polarity DALEKAISER

6

103 Gliding Motility of Myxococcus xanthus PATRICIA HARTZELL, WENYUAN SHI, AND PHILIPYOUDERIAN

93

7 The Frz Chemosensory System of Myxococcus xanthus DAVIDR. ZUSMAN, YUKIF. INCLAN, AND TAMMIGNOT

123 V

CONTENTS

vi

III Regulatory Mechanisms

133

8 Chemosensory Signal Transduction Systems in Myxococcus xantbus 135 JOHNR. KIRBY,JAMESE. BERLEMAN,SUSANNEMULLER, DI LI, JODIEC. SCOTT,AND JANETM. WILSON 9 Transcriptional Regulatory Mechanisms during Myxococcus xantbus Development 149 LEEKROOSAND SUMIKOINOUYE 10 Two-Component Signal Transduction Systems of the Myxobacteria 169 AND PETER J. A. COCK DAVIDE. WHITWORTH

11 Protein Ser/Thr Kinases and Phosphatases in Myxococcus xanthus 191 SUMIKOINOUYE, HIROFUMI NARIYA, AND JOSEMUROZ-DORADO 12 Carotenogenesis in Myxococcus xantbus: a Complex Regulatory Network 211 MONTSERRAT EL~AS-ARNANZ, MARTAFONTES,AND S. PADMANABHAN

N

Structure and Metabolism

227

13 Composition, Structure, and Function of the Myxococcus xanthus Cell Envelope 229 ZHAOMIN YANG,XUE-YAN DUAN,MEHDIESMAEILIYAN, AND HEIDI B. KAPLAN 14 Metabolic Pathways Relevant to Predation, Signaling, and Development 241 J. SHIMKETS PATRICK D. CURTISAND LAWRENCE 15 Secondary Metabolism in Myxobacteria HELGE B. BODE AND ROLFMULLER

259

V Myxobacterial Genomics and Postgenomics

283

16 The Genomes of Myxococcus xantbus and Stigmatella aurantiaca 285 AND WILLIAM C. NIERMAN CATHERINE M. RONNING 299 17 A Postgenomic Overview of the Myxobacteria SUEN,BARRYS. GOLDMAN, AND ROYD. WELCH GARRET

VI Stigmatella and Sorangium

313

18 The Challenge of Structural Complexity: Stigmatella aurantiaca as an Alternative Myxobacterial Model 315 WULFPLAGA 329 19 Sorangium cellulosum KLAUSGERTH,OLENAPERLOVA, AND ROLFMULLER

CONTENTS

VII Analogous Systems

vii

349

20 Bdellovibrio: Lone Hunter “Cousin” of the “Pack Hunting” 351 Myxobacteria K. J. EVANS,L. HOBLEY, C. LAMBERT, AND R. E. SOCKETT 21 Bacillus subtilis Sporulation and Other Multicellular Behaviors 363 P. MORAN, JR. LEEKROOS,PATRICK J. PIGGOT,AND CHARLES 22 Developmental Control in Caulobacter crescentus: Strategies for Survival in Oligotrophic Environments 3 85 DEANNE L. PIERCEAND YVESV. BRUN 23 Developmental Biology of Heterocysts, 2006 JINDONG ZHAOAND C. PETERWOLK

397

24 Multicellular Development in Streptomyces 4 19 MARIEA. ELLIOT,MARK J. BUTTNER,AND JUSTINR. NODWELL 25 A Eukaryotic Neighbor: Dictyostelium discoideum DERRICK BRAZILLAND RICHARDH. GOMER

439

26 Multispecies Interactions and Biofilm Community Development 453 S. JAKUBOVICS, AND PAULE. KOLENBRANDER, NICHOLAS NATALIA I. CHALMERS

VIII Myxobacterial Methods

463

27 Myxococcus xanthus: Cultivation, Motility, and Development 465 PENELOPE I. HIGGS AND JOHNP. MERLIE, JR. 28 Myxococcus xanthus: Expression Analysis 479 FRANK-DIETRICH MULLER AND JIMMY SCHOUV JAKOBSEN 29 Genetic Tools for Studying Myxococcus xanthus Biology A. MURPHY AND ANTHONY G. GARZA KIMBERLY

503 30 Sorangium cellulosum Methods ANKETREUNER-LANGE, SABRINA Dog, AND TINAKNAUBER Index

513

491

Contributors

E. BERLEMAN Dept. of Microbiology, University of Iowa, 51 Newton Rd., Iowa City, IA 52242

JAMES

HELGE B. BODE Institut fur Pharmazeutische Biotechnologie, Universitat des Saarlandes, Postfach 15 11 50, 66041 Saarbrucken, Germany DERRICK BRAZILL Dept. of Biology, Hunter College, 695 Park Ave., New York, NY 10021

YVESV. BRUN Dept. of Biology, Indiana University, Bloomington, IN 47405

MARK J. BUTTNER Dept. of Molecular Microbiology, John Innes Centre, Colney Lane, Norwich, NR4 7UH, United Kingdom

NATALIA I. CHALMERS Oral Biofilm Communication Unit, Oral Infection and Immunity Branch, National Institute of Dental and Craniofacial Research, National Institutes of Health, Bethesda, MD 20892, and University of Maryland School of Dentistry, Baltimore, MD

PETERJ. A. COCK MOAC Doctoral Training Centre, University of Warwick, Coventry CV4 7AL, United Kingdom

PATRICK D. CURTIS Dept. of Microbiology, University of Georgia, Athens, GA 30602

...

VZZZ

CONTRIBUTORS MICHELLE E. DIODATI Section of Microbiology, University of California-Davis, Davis, CA 95616 SABRINA DoB Dept. of Microbiology and Molecular Biology, Justus-Liebig-University Giessen, Heinrich-Buff-Ring 26-32, D-35392 Giessen, Germany XUE-YANDUAN Dept. of Molecular and Cellular Biology, Baylor College of Medicine, Houston, TX 77030

MARTIN DWORKIN Dept. of Microbiology, University of Minnesota, Minneapolis, M N 55455-0312 MONTSERRAT EL~AS-ARNANZ Departamento de Genetica y Microbiologia (Unidad Asociada a1 IQFR-CSIC), Facultad de Biologia, Universidad de Murcia, 30100 Murcia, Spain

MARIE A. ELLIOT Dept. of Biology, McMaster University, 1280 Main St. West, Hamilton, ON, Canada L8S 4K1 MEHDIESMAEILIYAN Dept. of Natural Sciences, University of HoustodDowntown, Houston, TX 77002 K. J. EVANS Institute of Genetics, School of Biology, University of Nottingham Medical School, Nottingham NG7 2UH, United Kingdom MARTAFONTES Departamento de Genetica y Microbiologia (Unidad Asociada a1 IQFRCSIC), Facultad de Biologia, Universidad de Murcia, 30100 Murcia, Spain

ANTHONYG. GARZA Dept. of Biology, Syracuse University, Syracuse, NY 13244 KLAUSGERTH Helmholtz-Zentrum fur Infektionsforschung GmbH, InhoffenstraBe 7, 3 8 124 Braunschweig, Germany

RONALDE. GILL Dept. of Microbiology, University of Colorado Health Sciences Center, Denver, CO 80262 BARRYS. GOLDMAN Monsanto Company, St. Louis, M O 63167 RICHARDH. COMER Howard Hughes Medical Institute and Dept. of Biochemistry and Cell Biology, MS-140, Rice University, 6100 S. Main St., Houston, TX 77005-1892

ix

CONTRIBUTORS

X

PATRICIA HARTZELL Dept. of Microbiology, Molecular Biology and Biochemistry, University of Idaho, Moscow, ID 83844

PENELOPE I. HIGGS Dept. of Ecophysiology, Max Planck Institute for Terrestrial Microbiology, Marburg, Germany 35043

KRISTINAL. HILLESLAND Dept. of Civil and Environmental Engineering, University of Washington, Seattle, WA 98195-2700 L. HOBLEY Institute of Genetics, School of Biology, University of Nottingham Medical School, Nottingham NG7 2UH, United Kingdom YUKIF. INCLAN Dept. of Molecular and Cell Biology, University of California, Berkeley, CA 94720-3204 SUMIKOINOUYE Dept. of Biochemistry, Robert Wood Johnson Medical School, Piscataway, NJ 08854 SCHOUVJAKOBSEN 194 Chemin du Siege, Residence le Mirabaou, F-06140 Vence, France

JIMMY

NICHOLAS S. JAKUBOVICS Oral Biofilm Communication Unit, Oral Infection and Immunity Branch, National Institute of Dental and Craniofacial Research, National Institutes of Health, Bethesda, MD 20892

DALEKAISER Dept. of Biochemistry and Dept. of Developmental Biology, Stanford University Medical School, Stanford, CA 94305

HEIDIB. KAPLAN Dept. of Microbiology and Molecular Genetics, University of Texas Medical School, Houston, TX 77030 R. KIRBY Dept. of Microbiology, University of Iowa, 51 Newton Rd., Iowa City, IA 52242

JOHN

TINAKNAUBER Dept. of Microbiology and Molecular Biology, Justus-Liebig-University Giessen, Heinrich-Buff-Ring 26-32, D-35392 Giessen, Germany

PAULE. KOLENBRANDER Oral Biofilm Communication Unit, Oral Infection and Immunity Branch, National Institute of Dental and Craniofacial Research, National Institutes of Health, Bethesda, MD 20892 LEEKROOS Dept. of Biochemistry and Molecular Biology, Michigan State University, East Lansing, MI 48 824

CONTRIBUTORS C. LAMBERT Institute of Genetics, School of Biology, University of Nottingham Medical School, Nottingham NG7 2UH, United Kingdom

DI LI Dept. of Microbiology, University of Iowa, 51 Newton Rd., Iowa City, IA 52242 JR. JOHNP. MERLIE, Dept. of Molecular and Cell Biology, University of California-Berkeley, Berkeley, CA 94720

TAMMIGNOT Laboratoire de Chimie Bacttrienne, 31, Chemin Joseph Aiguier, 13009 Marseille, France

P. MORAN, JR. CHARLES Dept. of Microbiology and Immunology, Emory University School of Medicine, Atlanta, GA 30322 MULLER FRANK-DIETRICH Dept. for Ecophysiology, Max Planck Institute for Terrestrial Microbiology, Karl-von-Frisch-Strage, D-35043 Marburg, Germany ROLFMULLER Institut fur Pharmazeutische Biotechnologie, Universitat des Saarlandes, Postfach 15 11 50,66041 Saarbriicken, Germany SUSANNE MULLER Dept. of Microbiology, University of Iowa, 51 Newton Rd., Iowa City, IA 52242

JosB MUAOZ-DORADO Departamento de Microbiologia, Facultad de Ciencias, Universidad de Granada, E-18071 Granada, Spain KIMBERLY A. MURPHY Dept. of Biology, Syracuse University, Syracuse, NY 13244

HIROFUMI NARIYA Dept. of Biochemistry, Robert Wood Johnson Medical School, Piscataway, NJ 08854 WILLIAMC. NIERMAN J. Craig Venter Institute, 9712 Medical Center Dr., Rockville, MD 20850 R. NODWELL Dept. of Biochemistry and Biomedical Sciences, Health Sciences Centre, McMaster University, Hamilton, ON, Canada L8N 325

JUSTIN

S. PADMANABHAN Instituto de Quimica-Fisica “Rocasolano,” CSIC, 28006 Madrid, Spain OLENAPERLOVA Institut fur Pharmazeutische Biotechnologie, Universitat des Saarlandes, Postfach 15 11 50,66041 Saarbrucken, Germany

xi

CoNTRIB UTORS

xii

DEANNE L. PIERCE Dept. of Biology, Indiana University, Bloomington, IN 47405

J. PIGGOT PATRICK Dept. of Microbiology and Immunology, Temple University School of Medicine, Philadelphia, PA 19140

WULFPLAGA Zentrum fur Molekulare Biologie der Universitat Heidelberg (ZMBH), University of Heidelberg, 69120 Heidelberg, Germany LYNDAPLAMANN School of Biological Sciences, Cell Biology and Biophysics, University of Missouri-Kansas City, Kansas City, M O 64110

CATHERINE M. RONNING J. Craig Venter Institute, 9712 Medical Center Dr., Rockville, MD 20850

C. SCOTT Dept. of Microbiology, University of Iowa, 51 Newton Rd., Iowa City, IA 52242

JODIE

WENYUAN SHI Dept. of Oral Biology, School of Dentistry, and Dept. of Microbiology, Immunology and Molecular Genetics, School of Medicine, University of California, Los Angeles, Los Angeles, CA 90095 LAWRENCE J. SHIMKETS Dept. of Microbiology, University of Georgia, Athens, GA 30602

MITCHELL SINGER Section of Microbiology, University of California-Davis, Davis, CA 95616 R. E. SOCKETT Institute of Genetics, School of Biology, University of Nottingham Medical School, Nottingham NG7 2UH, United Kingdom LOTTES0GAARD-ANDERSEN Max Planck Institute for Terrestrial Microbiology, Karl-von-Frisch Str., 35043 Marburg, Germany

GARRET SUEN Dept. of Biology, Syracuse University, Syracuse, NY 13244 ANKETREUNER-LANGE Dept. of Microbiology and Molecular Biology, Justus-Liebig-University Giessen, Heinrich-Buff-Ring 26-32, D-35392 Giessen, Germany GREGORY J. VELICER Dept. of Biology, Indiana University, Bloomington, IN 47405, and Max Planck Institute for Developmental Biology, 72076 Tiibingen, Germany ROY D. WELCH Dept. of Biology, Syracuse University, Syracuse, NY 13244

CONTRIBUTORS DAVIDE. WHITWORTH Dept. of Biological Sciences, University of Warwick, Coventry CV4 7AL, United Kingdom M.WILSON Dept. of Microbiology, University of Iowa, 51 Newton Rd., Iowa City, IA 52242

JANET

C. PETERWOLK MSU-DOE Plant Research Laboratory and Dept. of Plant Biology, Michigan State University, E. Lansing, MI 48824

ZHAOMIN YANG Dept. of Biology, Virginia Polytechnic Institute and State University, Blacksburg, VA 24060

PHILIPYOUDERIAN Dept. of Biology, Texas A & M University, College Station, TX 83843 JINDONG

ZHAO

State Key Laboratory of Protein and Plant Genetic Engineering, College of Life Sciences, Peking University, Beijing 100871, China DAVIDR. ZUSMAN Dept. of Molecular and Cell Biology, University of California, Berkeley, CA 94720-3204

...

xzzz

Preface

Since their discovery, the myxobacteria have proven to be enduring sources of wonder and inspiration for microbiologists. Myxobacteria exhibit several behaviors that are rare within the bacterial world but commonplace in eukaryotes, including multicellular development and cellular differentiation. They have consequently been used as amenable model organisms for studies into the general principles of such behaviors-studies that have addressed the evolutionary, ecological, and molecular mechanisms involved. This book focuses on myxobacterial multicellularity and differentiation, but attempts to paint a broader canvas by also describing analogous behaviors seen in a wide range of microbiological systems. In recent decades biology has been revolutionized by the emergence of techniques capable of probing the molecular basis of cellular behavior. The ability to manipulate an organism’s genes, to characterize their protein products, and to extract information from molecular sequences has led to an ever-increasing understanding of cellular processes. In 1984, the first book on myxobacteria was published (edited by E. Rosenberg), and this was followed in 1993 by a second volume (edited by M. Dworkin and D. Kaiser; American Society for Microbiology). In these works it is apparent that the appropriate genetic tools had been developed to enable research into fundamental features of myxobacterial behavior, with particular attention being paid to the regulation of motility and multicellular development. With the advent of a new millennium, further technological advances have continued to revolutionize biology. The ability to routinely determine the entire genome sequence of an organism, to simultaneously assess expression of every gene within that genome, and to identify changes in that organism’s global pool of proteins has resulted in a flood of molecular biological data. These large sets of data are now being actively generated and exploited by researchers of myxobacterial biology. The complete genome sequence for the model myxobacterium Myxococcus xanthus has recently become available, along with those of three xv

xvi

PREFACE less well-characterized myxobacteria (Sorangium cellulosum, Stigmatella aurantzaca, and Anaeromyxobacter dehalogenans). The historical voyage of discovery from glycerol induction of sporulation to the advent of the Myxococcus xanthus genome sequence provides a plot for the introductory chapter of this volume (chapter l),kindly provided by the editors of the last collected volume on myxobacterial biology. The picture that emerges is that the myxobacteria are extremely intricate organisms, with complexity rivaling that of many eukaryotes. Dworkin and Kaiser’s preface to the 1993 book stated that “we are thus hopeful and cautiously optimistic that the next edition of this book will see the emergence of insights about fruiting body morphogenesis, the mechanisms of multicellular communication and coordination, the mechanism and function of rippling, and the mechanism of gliding motility, to name only a few of the fascinating aspects of myxobacterial biology.” This optimism was not misplaced. Fourteen years later, chapters in this book review major progress in our understanding of many features of myxobacterial biology. Indeed, some topics that were single chapters in the 1993 book are now entire sections in this volume. For example, descriptions of the mechanisms of motility span four chapters in this book (chapters 5 through 8). Similarly, our understanding of multicellular development has progressed significantly (chapters 3 and 4), as has knowledge of regulatory mechanisms (chapters 9, 10, 11, and 15). Myxobacteria continue to attract significant pharmaceutical interest through their production of bioactive secondary metabolites, and myxobacterial metabolism forms the topic of two chapters (chapters 13 and 14). The genome sequence of M . xanthus ushered myxobacterial research into the post-genomic era (chapter 17). Since then, comparative genomic analyses have continued to provide insights into the molecular biology of other myxobacteria, particularly Sorangium cellulosum (chapter 19) and Stigmatella aurantiaca (chapters 16 and 18). The first chapters in the book underpin the others by providing historical and ecological/evolutionary contexts for contemporary myxobacterial research (chapters 1 and 2). In their preface to the 1993 book, Dworkin and Kaiser correctly warned about the dangers of focusing on a single model organism. In addition to descriptions of M . xanthus, chapters in this volume present the biology of two other myxobacteria: Sorangium cellulosum and Stigmatella aurantiaca (chapters 18 and 19). Behaviors exhibited by the myxobacteria can also be found in other (often very different) organisms. I am therefore delighted to include seven diverse examples of microbial multicellularity and differentiation (chapters 20 through 26), enabling myxobacterial biology to be set in a much broader context. I am especially grateful to the authors for chapters on proteobacterial predation and differentiation, cyanobacterial differentiation, development and sporulation in gram-positive bacteria, eukaryotic multicellularity and differentiation, and multispecies biofilm development. The myxobacterial research community currently numbers around 40 laboratories across the globe, and further expansion is to be encouraged. To aid researchers who are unfamiliar with the myxobacteria but wish to start working with these organisms, chapters have also been included that describe the most commonly used methods for cultivating, manipulating, and characterizing myxobacteria (chapters 27 to 30). I hope that these chapters will also act as a compendium of techniques for existing myxobacteria researchers. At the start of a new millennium it is tempting to think that we are finally beginning to comprehend the complex behavior of the myxobacteria. While huge leaps have indeed been made in our knowledge of molecular mechanisms (particularly those governing motility and fruiting body formation), the significance of

PREFACE

xvii

those mechanisms for the physiology of the myxobacteria in their natural environment is still largely unappreciated. Molecular ecology and evolutionary analyses are starting to address such gaps in our understanding, but there is still much to learn. I hope this volume stimulates seasoned myobacteriologists and interested amateurs alike, and it is to be hoped that the next book on the myxobacteria will be able to claim a true understanding of this most complex of prokaryotes. Particular thanks must go to Heidi Kaplan (University of Texas, Houston), who helped greatly during the conception and early stages of this project. I would also like to thank Carolyn Love (also at the University of Texas, Houston) for clerical assistance, and Greg Payne and Ellie Tupper at ASM Press for their invaluable and humane support. All chapter authors deserve special thanks for timely submission of manuscripts. Every chapter in this volume has been rigorously peer reviewed, and I would like to thank all the reviewers for assisting with this process. Finally, I would very much like to thank all the members of the myxobacteria research community, and my family, for their encouragement and support for this project and for the stimulating environment that they provide.

David E. Whitworth W a r w i c k , 2007

References Rosenberg, E. (ed). 1984. Myxobacteria. Development and Cell Interactions. SpringerVerlag, New York, NY. Dworkin, M., and D. Kaiser (ed.). 1993. Myxobacteria 11. American Society for Microbiology, Washington, DC.

Mvxobacteria 1 siblogy

1

Myxobacteria: Multiceiiularity and Differentiation Edited by David E. Whitworth 0 2008 ASM Press, Washington, D.C.

Dale Kaiser Martin Dworkin

From Glycerol to the Genome

GLYCEROL INDUCTION OF MYXOSPORES Dorothy Powelson’s laboratory at Purdue University had been studying the nature of the cell surface of Myxococcus xanthus and had been attempting to compare the chemical composition of the cell walls of vegetative cells and myxospores. Seeking a more convenient method of collecting myxospores than harvesting fruiting bodies, her group had explored techniques for converting vegetative cells into myxospores in liquid culture. The method they settled on involved exposing the cells to high concentrations of sucrose, which induced the conversion of vegetative cells to round, optically refractile, somewhat resistant cells, which they concluded were spores (Adye and Powelson, 1961).Dworkin’s laboratory at the University of Minnesota was interested in the processes that were involved in the cellular morphogenesis of the rod-shaped vegetative cells to the round myxospores. However, when they examined the sucrose induction method more closely they became convinced, based on the sequence of morphological events during the conversion, that the cells were converting to osmotically resistant spheroplasts rather than to myxospores. Nevertheless, the conversion of a rod-shaped cell to a sphere, albeit not a bona fide myxospore, was sufficiently interesting

1

that they decided to try to understand the mechanism of the conversion to spheroplasts. They looked at the effect of a variety of other polyhydroxy compounds and found that glycerol at an optimal concentration of 0.5 M was able to induce the conversion of vegetative cells to myxospores rapidly (within 120 min), synchronously, and quantitatively (Dworkin and Gibson, 1964).A more detailed description of the process followed, which demonstrated that glycerol-induced myxospores were able to germinate, were resistant to elevated temperature, W irradiation, and sonication, and mimicked the sequence of morphological stages during the formation of fruiting body myxospores (Dworkin and Sadler, 1966; Ramsey and Dworkin, 1968; Sadler and Dworkin, 1966; Sudo and Dworkin, 1969). While David Zusman (Zusman, 1984) pointed out some important differences between glycerol-induced and fruiting body myxospores, glycerol induction quickly became a favorite vehicle for comparing the properties and processes of vegetative cells and myxospores, albeit with careful qualifications. Following the pioneering work of Powelson, White et al. (White et al., 1968) characterized the peptidoglycan of M. xanthus and showed that unlike the murein of Escherichia coli and other gram-negative bacteria, which

Dale Kaiser, Departments of Biochemistry and Developmental Biology, Stanford University Medical School, Stanford, CA 94305. Martin Dworkin, Department of Microbiology, University of Minnesota, Minneapolis, MN 55455-03 12.

3

4 existed as a continuous bag-shaped macromolecule (Weidel and Pelzer, 1965), the sacculus of M. xanthus existed as discrete patches of peptidoglycan held together by trypsin-sensitive material. Moreover, the peptidoglycan of vegetative cells contained substantial amounts of covalently bound glucose. Bacon et al. (Bacon et al., 1975) showed that during glycerol-induced myxospore formation there was a substantial shift in carbon flow to polysaccharide synthesis, which was accompanied by an increased amount of cross-linking via diaminopimelic acid and a substantial decrease in the amount of peptidoglycan-linked glucose in the myxospore cell wall. White (White, 1984) suggested that the patchy quality of the peptidoglycan and the above changes during glycerol induction were causally related to the shape change during myxospore formation. Despite structural differences between fruiting body and glycerol-induced myxospores, the TnSlac insertion mutation, Q7536, described below, simultaneously blocked the development of glycerol-induced spores as well as fruiting body spores as they changed their shape from rod to sphere (Licking et al., 2000). The discovery of a common step encouraged the study of glycerolinduced conversion of the cylindrical vegetative rods to the spherical myxospores as a simple model of cellular morphogenesis.

MYXO FILMS A series of momentous events in the history of myxobacterial research took place between 1965 and 1974. During this time Hans Reichenbach (Fig. 1)was working on his Ph.D. dissertation research in the laboratory of Hans Kiihlwein (Fig. 2) at the University of Gottingen. Reichenbach, with the collaboration of the Institut fur den Wissenschaftlichen Film in Gottingen, created a series of masterful time-lapse photomicrographic films demonstrating the behavior of a variety of different myxobacteria. These films illustrated their cellular morphology, cell division, gliding motility, swarming behavior, aggregation, fruiting body formation, myxospore and sporangiole formation and germination, feeding on prey bacteria, and pervasive cell-cell interactions (Kuhlwein and Reichenbach, 1968; Reichenbach, 1965, 1966, 1968, 1974). One of the films illustrated a unique and mysterious rippling wave behavior that Reichenbach referred to as rhythmic oscillations (Reichenbach, 1965). In 1974, excerpts from these films were edited into a 13-min-long mosaic by Dworkin, which has since served to introduce the myxobacteria to large numbers of fascinated microbiologists, some of whom were influenced to study these organisms in their laboratories. (This film can be

MYXOBACTERIAL BIOLOGY

Figure 1 Hans Reichenbach in 1980, collecting samples of soil in the Loire Valley during the Myxo meeting in Poitiers, France.

viewed as part of the chapter “The Myxobacteria” in the online edition of The Prokaryotes). The “myxo movies,” as they have come to be known, document the unique myxobacterial grade of multicellularity, one of Nature’s many explorations of that state. They challenged myxobacteriologists to explain it, and some of the challenges have been taken up.

DENSITY-DEPENDENT GROWTH ON CASEIN By 1976 it had already become clear that a defining feature of myxobacterial behavior was the pervasive tendency of cells to maintain a high cell density. An examination of Reichenbach’s films revealed that even though individual cells could momentarily leave the swarm, the tendency was for the swarm to remain intact throughout feeding and development (A- and S-motility were yet to be distinguished). The demonstration that experimental induction of fruiting body formation required a high cell density inoculum (Wireman and Dworkin, 1975) was consistent with the broader notion that “. . . the life cycle of the myxobacteria is directed at all times to preserving the existence or the potential for the swarm” (Dworkin, 1972). Furthermore, the ability of the myxobacteria to hydrolyze a wide variety of

TO THE GENOME 1. FROMGLYCEROL

macromolecules as a nutrient source had led to the proposal that “. . . generation of optimal concentrations of lower molecular weight subunits of these polymers will depend on a certain optimal density of cells excreting the hydrolases. In other words, a wolf-pack effect” (Dworkin, 1973). Experimental support for this idea was absent until Eugene Rosenberg began thinking about the problem. Rosenberg had begun work on the myxobacteria while a Professor of Microbiology at UCLA and continued that work after his emigration to Tel Aviv University in Israel. On his sabbatical in Minneapolis in 1976 he formulated an experimental strategy for addressing the problem. We called on our colleague Ken Keller, a chemical engineer, to help guide us through the mathematical analyses; from that collaboration there emerged clear and definitive proof that while a single cell of M. xanthus could grow perfectly well on an enzymatic hydrolyzate of casein, alternatively, when that cell was presented with the intact casein molecule as a substrate, growth was clearly cell density-dependent (Rosenberg et al., 1977). This was an obvious reflection of the fact that the cells, when growing on a macromolecular substrate, were at the mercy of diffusion of their hydrolytic enzymes away from the cell and diffusion of the low-molecular-weight products of the hydrolysis toward the cell. Slowly, the modus vivendi of the myxobacteria began to make sense-their predilection for insoluble macromolecular substrates, their ability to move by gliding on a solid surface, the density dependence of their growth and fruiting body formation, and the collection of myxospores in fruiting bodies in what were probably optimally sized packages of resistant resting cells poised to form a swarm upon germination-which helps put the entire episode of density-dependent growth into a larger biological perspective.

CAROTENOIDS Myxobacteria are well known for their carotenoid production, and the exciting story of its regulation is told in chapter 12. Experiments on carotenogenesis began in 1964 in Dworkin’s group. Robert Burchard, then a graduate student, was trying to isolate a bacteriophage for M. xanthus. At that time, no phages for any of the myxobacteria had been isolated. Lawns of M. xanthus were exposed to soil extracts and, after incubation, examined for characteristic plaques. After a series of unsuccessful attempts, one series of plates was left on a windowsill after having been incubated for about 1 week. When preparing to discard the plates, Burchard noticed that the bacterial lawn on the uppermost plate in the

5

Figure 2 Martin Dworkin (left) and Hans Kiihlwein (right), during the Myxo meeting in Poitiers, France. pile had pockmarks suggestive of phage plaques. However, repeated attempts to transfer the plaque material to fresh lawns produced no new plaques. Eventually, we realized that the top plate in the pile had been exposed to the sunlight shining through the window. Subsequent experiments also revealed that the plates that had been incubated in the dark did not produce the characteristic myxobacterial carotenoids, which we later showed served a photoprotective purpose and were photoinduced. Furthermore, the photosensitizing pigment was identified as protoporphyrin IX, a cytochrome precursor, which accumulated in the cells only as they entered the stationary phase (Burchard and Dworkin, 1966).

FRUITING BODY DEVELOPMENT The most famous segments of Reichenbach’s movies deal with aggregation and fruiting body formation of several different genera of myxobacteria. How does the order evident in the shape of a species-specific fruiting body arise from the apparent disorder of a swarm? Although each species built a different structure, all were initiated by starvation. How does starvation induce fruiting body development? To investigate starvation, it was first necessary to find which nutrients were required. Dworkin had shown

6 that although M . xanthus grew well in Casitone and preferred peptides, it could be grown in a chemically defined medium that contained 17 amino acids with a generation time of 8 to 10 h (Dworkin, 1962). Despite improvements that shortened the generation time to 6.5 h (Witkin and Rosenberg, 1970), it was not clear which amino acids in the medium were essential and which were serving as sources of carbon. In 1978, Anthony Bretscher identified leucine, isoleucine, and valine as essential amino acids and found that vitamin B,, was essential for the synthesis of methionine (Bretscher and Kaiser, 1978). Those requirements were confirmed in 2005 by the absence in the M. xanthus genome of genes for the biosynthesis of the essential branched-chain amino acids (Goldman et al., 2006). Dworkin (Dworkin, 1962) observed that phenylalanine addition stimulated growth, and Bretscher verified its low rate of synthesis that severely limited growth. Using Bretscher’s minimal synthetic medium, Colin Manoil found that limitation for any amino acid, whether essential or nonessential, induced fruiting body development (Manoil and Kaiser, 1980). He found that starvation for carbon, energy, or phosphorus also induced development, while Kimsey showed that neither purine nor pyrimidine starvation would induce development (Kimsey and Kaiser, 1991). These observations suggested how starvation might be recognized. Since a complete set of aminoacyl tRNAs is essential for protein synthesis, the absence of one or more aminoacyl tRNAs could readily be perceived by a halt in protein synthesis. In M. xanthus, as in many other bacteria, the absence or shortage of any one of the charged tRNAs causes a ribosome to synthesize guanosine tetraphosphate (and pentaphosphate), (p)ppGpp, by transferring P-P from ATP to GTP, to trigger a stringent response. Mitchell Singer showed that M. xanthus has a relA gene (ppGpp synthetase) and that (p)ppGpp was necessary and sufficient to initiate fruiting body development (Singer and Kaiser, 1995). It soon was recognized that this was a multicellular response.

MOLECULAR GENETICS It was clear in 1972 that genetic experiments would be necessary to complement the ongoing biochemical and cellular studies, if the program of fruiting body development were to be found. Mutations that blocked a single step in development would need to be analyzed, and this would require a means for gene transfer. In a search for transduction, we found that E. coli phage PlCM both adsorbed to and injected DNA into M. xanthus (Kaiser and Dworkin, 1975). Subsequently, transposon Tn5,

MYXOBACTERIAL BIOLOGY which conferred kanamycin resistance, was found to tag interesting mutations (Kuner et al., 1981; Sodergren and Kaiser, 1983).Meanwhile, several generalized transducing phages, Mx4, Mx8, and Mx9, were found for Myxococcus (Campos et al., 1978; Martin et al., 1978). Lee Kroos constructed TnSlac, which carried a promoterless trp-lac fusion fragment inserted near the end of the transposable element. Transposition of Tn5lac into M. xanthus transcriptionally fused lacZ to an adjacent promoter, creating a reporter for that promoter (Kroos and Kaiser, 1984). Kroos and Adam Kuspa created a library of TnSlac insertions that were expressed at different times in development (Kroos et al., 1986).Leon Avery developed transposon replacement in situ, enabling construction of doubly marked mutant strains with transposons at two different sites in the same strain that brought resistance to two different antibiotics (Avery and Kaiser, 1983). The transposon library made it possible to analyze signaling mutants (Kuspa et al., 1986). Electroporation is effective for introducing M. xanthus DNA into M. xanthus (Ramaswamy et al., 1997) and has been systematically improved (Youderian et al., 2003).

CELLS SIGNAL EACH OTHER Two groups (Hagen et al., 1978, and LaRossa et al., 1983) analyzed the same set of M. xanthus mutants that were conditionally defective in the formation of sporefilled fruiting bodies. Interestingly, these mutants were unable to develop on their own, but when mixed with wild-type cells or with certain other mutants, they were able to develop. Pairwise testing of 57 mutants divided them into four groups (A to D), and a fifth group was discovered by John Downard (Downard, 1993). Complementation did not result from cross-feeding of essential intermediary metabolites that were diffusible because the mutants, with the exception of group E, could grow on a minimal defined medium (LaRossa et al., 1983). It now appears that groups A and C define extracellular signals that are exchanged between cells, while the ByD, and E groups are more likely consequences of cell contact-dependent exchanges of materials between cells, like stimulation (Kaiser, 2004; Nudleman et al., 2005).

A-Signal and Responses to A-Signal Medium conditioned by Myxococcus development was found to contain a heat-stable and a heat-labile form of A-signal activity. Lynda Plamann found that heat-labile A-signal was a mixture of proteases and proteins that were sensitive to those proteases (Plamann et al., 1992). Then, Adam Kuspa showed that heat-stable A-signal is a set of amino acids and small peptides containing those

1. FROMGLYCEROL TO THE GENOME amino acids (Kuspa et al., 1992). Most likely amino acids are the primary A-signal molecules, while the extracellular release of proteases and proteins generates first peptides and then A-signal amino acids. Because M. xanthus can develop fruiting bodies in a medium devoid of external amino acids, developmental proteins are synthesized at the expense of cellular reserves. Mitchell Singer found that A-signal helps M . xanthus assess the nutrient available to it for development so it can complete the program (Singer and Kaiser, 1995). As Myxococcus faces starvation, it must choose between initiating fruiting body development with differentiation of spores and slow growth at a rate compatible with whatever level of nutrient happens to be available. Spore counts indicate that fewer than 1% of cells undertaking fruiting body development eventually become spores. If nutrient is on its way to eventual exhaustion, then slowing growth to match the level of residual nutrient will lead to slower and slower growth, until death ensues. Since either option kills the majority of cells, the better choice from the cell’s perspective depends on its projection of nutrient availability in its near future. M . xanthus appears to use a stringent response and the cell-density-dependent A-signal to predict the nutrient available to it. Heidi Kaplan, analyzing suppressors of asg mutants, called sas mutants (Kaplan et al., 1991), found that a sensor histidine kinase, Sass, and a response regulator, SasR, constituted a two-component system that senses the extracellular level of A-signal. If the level is adequate, sasR protein triggers the expression of certain A-signaldependent genes, such as a 4 5 2 1 (Kaplan and Plamann, 1996; Keseler and Kaiser, 1995).

C-Signal Protein Active C-signal was purified from detergent-extracted cell membranes. A bioassay of restoring aggregation and sporulation, in vitro, to a mutant lacking a csgA gene was employed to monitor purification (Kim and Kaiser, 1990a).Because csgA mutants arrest fruiting body development having only formed traffic jams (Kaiser, 2003) and very few viable spores, the sporulation assay was quite sensitive. A 17-kDa protein that could restore development was purified from starved wild-type cells (Kim and Kaiser, 1990a). No C-signal activity was recovered from extracts of cells that were not starved, or from csgA mutant cells (Kim and Kaiser, 1990a). Nevertheless, the genomic sequence of the csgA gene predicted a 25-kDa protein homologous to the short-chain alcohol dehydrogenase family of enzymes (Lee et al., 1995). The discrepancy in size was resolved by Sune Lobedanz, then a graduate student working with Lotte Sargaard-Andersen.

7 Using antibodies to fragments of p25, Lobedanz showed that p17 corresponded to a C-terminal fragment of p25, in agreement with Kim’s amino acid sequence data (Kim and Kaiser, 1990b). Lobedanz also detected serine protease activity in an M. xanthus cell surface fraction that was capable of cleaving p25, removing an N-terminal peptide with the NAD+ binding site, leaving p17 adhering to the surface (Lobedanz and Sargaard-Andersen, 2003). Evidently, p17 is the signal, and processing by a cell-surface protease ensured that the signal is transmitted to another cell; a cell never signals itself.

C-Signal Transmission An unexpected observation that nonmotile mutants of M. xanthus arrested fruiting body development at the same morphological stage as a csgA mutant (Kroos et al., 1988) suggested that motility might be required for C-signaling because it took place between cells in end-to-end contact. Seung Kim tested this hypothesis by mechanically forcing nonmotile cells into end-to-end alignment (Kim and Kaiser, 1 9 9 0 ~ )He . used the asymmetry of the long rod-shaped M. xanthus cells to orient them lengthwise as they fell into the narrow grooves produced by scoring agar with a fine-grained aluminum oxide abrasive paper. Phase-contrast microscopy revealed that cells, which had settled into the grooves, were indeed oriented with their long axes parallel to the axis of the groove (Kim and Kaiser, 1 9 9 0 ~ )Then, . a second independent line of experiments on traveling waves also led to the conclusion that C-signal transmission was specific for end-to-end contact, as opposed to contact with the side of a cell. Traveling wave crests, colliding at any angle, were never seen to interfere with each other (Sager and Kaiser, 1994). Instead, colliding wave crests reflected from one another, providing additional evidence that the C-signal is transmitted through the ends of two cells in contact. Moreover, signaling restricted to cell ends was in quantitative agreement with the mathematical analysis of traveling waves (Igoshin et al., 2001; Welch and Kaiser, 2001). Thus, C-signal carries information about the local cell density as well as the orientation with respect to neighboring cells.

C-Signal Transduction Work from several laboratories has shown that C-signal is a morphogen that, in a dose-dependent manner, manages cell movement, initiates the expression of many developmentally regulated genes, and triggers sporulation (Kim and Kaiser, 1991; Kruse et al., 2001; Li et al., 1992; Sargaard-Andersen et al., 1996). Thomas Gronewold discovered a positive-feedback loop in the C-signal response circuit that is controlled by the act operon of

MYXO BACTERIAL BIOLOGY

8

five cotranscribed genes (Gronewold and Kaiser, 2001, 2002). This feedback is responsible for raising the number of C-signal molecules per cell from a few at 3 h poststarvation to several hundred by 18 h (Gronewold and Kaiser, 2001; Kim and Kaiser, 1991). The rise serves to time C-signal-dependent gene expression and to restrict that expression to the fruiting body (Julien et al., 2000; Kroos and Kaiser, 1987). Small aggregates enlarge when the number of Csignal molecules per cell rises above a moderate threshold. Above that threshold, a responding cell decreases its reversal frequency and increases its speed (Jelsbak and Ssgaard-Andersen, 1999, 2002; Ssgaard-Andersen et al., 2003). Responding cells tend to form a chain or stream whose cells are in frequent end-to-end contact with each other. Upon contact, they signal each other (Jelsbak and Ssgaard-Andersen, 2000). Streaming was observed by Brian Sager, when he tracked individual cells inside nascent fruiting bodies. He observed one-half of the cells to circulate clockwise and one-half counterclockwise (Sager and Kaiser, 1993).None of the tracked cells reversed during the experiment. As cells stream, they have more opportunities to C-signal each other and there is still more positive feedback. Finally, when the number of C-signal molecules per cell rises to the higher threshold for sporulation, the dev operon is expressed (Ellehauge et al., 1998; Kroos et al., 1986). dev Operon The last three genes of the dev operon, devT, devR, and devS, have been characterized (Boysen et al., 2002; Kroos et al., 1990; Thony-Meyer and Kaiser, 1993). Linda Thony-Meyer showed that devR devS double mutants are able to aggregate, but fail to sporulate (Thony-Meyer and Kaiser, 1993). Bryan Julien showed that dev expression was spatially restricted to the fruiting body (Julien et al., 2000). Ellen Licking and Lisa Gorski showed that since the Tn51ac::R7536 mutant aggregates normally and fails to sporulate, this reporter gene could be placed in a sporulation pathway downstream of dev (Licking et al., 2000). A consequence of dev action is that the differentiation of myxospores occurs only after aggregation is complete.

MOTILITY Many years ago myxobacterial motility was described by Jahn (Jahn, 1924), and by Schmidt-Lorenz and Iciihlwein (Schmidt-Lorenz and Kuhlwein, 1968) in terms of the microscopic structure of cells and their appendages. Henrichsen distinguished gliding by myxobacteria from the flagellar swimming motility observed in many

other bacteria as movement on a surface usually in the direction of a cell’s long axis (Henrichsen, 1972).A large stride forward was taken by Hans Reichenbach, working with Hans Kiihlwein, who made movies that showed cell movement during cell division, swarming, and the building of fruiting bodies by several different species (Kuhlwein and Reichenbach, 1968; Reichenbach, 1966, 1968,1974). Experimental genetic investigations of myxobacterial motility began with the isolation and examination of mutants by Burchard (1970), and by MacRae and McCurdy (1976). Then, Jonathan Hodgkin investigated many motility mutants all derived from the same genetically characterized strain. Comparisons between mutants revealed two different swarm patterns, indicative of two different gliding engines, referred to as engine A and engine S (Hodgkin and Kaiser, 1979a, 1979b). Hodgkin found roughly equal numbers of mutants that lacked either engine A or engine S, which he distinguished by their different swarm patterns. Normally the two engines cooperate with each other, but when there is only engine A (A+,?-),the cell clusters in swarms are long and strung out; when there is only engine S (A-S’), the clusters are short and stubby. In both types of swarms, cells move singly and in groups. A-S- colonies have sharp edges, much like those of E. coli, and the cells are unable to swarm or to form fruiting bodies.

S-Motility The S-engines are composed of type IV pili (TFP), long thin retractile hairs that extend from the front end of a cell and have the ability to retract, pulling the cell forward. M. xanthus TFP share at least 15 proteins that are involved in extending and retracting pili with motile Neisseria and Pseudomonas (whose motility is traditionally called twitching despite its relation to gliding), and with Synechocystis (Wall and Kaiser, 1999). The role played by a number of the Pi1 proteins has been worked out (Nudleman and Kaiser, 2004). For instance, PilT, an inner membrane protein, is an AAA motor protein necessary for pilus retraction. PilT mutants have pili, but because those pili are unable to retract, the cells lack S-motility (Li et al., 2003; Wu et al., 1997).Like the PilT mutants, another group of S- mutants that was discovered by David Morandi-the dsp (dispersed growth) mutants-have pili but nevertheless lack S-motility. The Dsp mutants lack fibrils.

Fibrils Stimulated by a report on “filaments” given at the 1978 Myxo meeting at Spring Hill, MN, M. Dworkin

TO THE GENOME 1. FROMGLYCEROL

saw their similarity to “myxonemata” described 15 years earlier by Walter Fluegel, who had viewed them by staining living cells with India ink. Dworkin recalls that both findings were received with mild interest and then quietly ignored until 1988, when Arnold and Shimkets showed that these filaments, which they termed “fibrils,” were responsible for cell-cell cohesion of M. xanthus (Arnold and Shimkets, 1988a, 198813). The late Rich Behmlander morphologically characterized the fibrils by means of low-voltage scanning electron microscopy. He showed that fibril formation required cells to be at a high cell density on a solid surface and that they consisted of a polysaccharide matrix with associated proteins (Behmlander and Dworkin, 1991, 1994a, 199413).He demonstrated that the so-called fibrils were not preparational artifacts by showing that when fibrils were decorated with carbon particles, examination of the cells by phase-contrast microscopy revealed the presence of fibrils which had not been subjected to fixation or dehydration (Behmlander and Dworkin, 1994a, 1994b). Arnold and Shimkets (Arnold and Shimkets, 1988a, 1988b) provided unambiguous evidence that the fibrils were required for the effective social and developmental behavior of M . xanthus. This was supported by the experiments of Chang and Dworkin (Chang and Dworkin, 1994) showing that isolated fibrils, added to a fibril-minus dsp mutant, were able to rescue cohesion and development of a dsp mutant. Subsequently, Yang et al. (Yang et al., 2000) showed that isolated fibrils could also partially rescue cohesion and development in dif mutants, one class of dsp mutants. Li et al. (Li et al., 2003) provided evidence that the TFP at a leading end of a cell of M. xanthus attached to amine-containing polysaccharide from fibrillar material deposited on the agar and triggered pilus retraction. This result was consistent with an earlier observation that glucosamine, one of the components of fibril polysaccharide, blocked cellcell cohesion (Behmlander and Dworkin, 1994b).In sum, S-motility results when a pilus from one cell attaches to a network of fibrils that encloses a group of cells ahead. As the attached pilus retracts, the piliated cell pulls up to the leading cells.

A-Motility Gliding M . xanthus leaves a phase-bright trail of “slime” behind on the agar surface, which has the same width as the cell. Reichenbach’s movies showed examples, and Lars Jelsbak photographed three isolated cells at higher magnification laying trails as they moved (online movie available in Kaiser, 2003). Wolgemuth et al. (2002) observed the extrusion of ribbons of slime-like material uniquely from one end of the cell and proposed that

9 polar slime secretion from the more than 100 pores visible at the back end of a cell pushes the cell forward. This proposal found support in the work of Rosa Yu and Kaiser (2007),who showed that A-motility is perfectly correlated with unipolar slime secretion. Bipolar slime secretion in the mglA mutants results in loss of motility (Kaiser and Yu, 2005).

Elasticotaxis It must be a source of great satisfaction for an author to find that a paper written over 50 years ago continues to provoke interest and to generate new experiments. Roger Stanier was one of the most outstanding microbiologists of our era, who had made major contributions to our understanding of the phototrophic bacteria, the pseudomonads, the cyanobacteria, and Caulobacter, but whose sole published contribution to the myxobacteria was one little note. As a young master’s student at UCLA, Stanier characterized the gliding motility, nutritional physiology, and taxonomy of the marine cytophagas and concluded casually, following an earlier suggestion by Krzemieniewska, that they be classified as nonfruiting myxobacteria in the order Myxobacteriales (sic) (Stanier, 1940). Stanier then joined Kees van Niel’s laboratory at the Hopkins Marine Station in Pacific Grove, CA, where his Ph.D. dissertation resulted in a classic monograph on the CytophagalSporocytophaga group (Stanier, 1942b). He characterized the group thoroughly, provided a much more detailed rationale for classifying them among the myxobacteria, and mentioned extensive, unpublished experience with the so-called “higher” myxobacteria while in van Niel’s laboratory. It was during this period that he noted that cultures of members of the Myxococcaceae when streaked on agar slants oriented themselves and their subsequent fruiting bodies in a consistently ordered fashion relative to the streak lines on the agar. He demonstrated by means of a simple and elegant experiment that this behavior was reproducible and that the gliding cells of Chondrococcus (Myxococcus) exiguus (now classified by Reichenbach as Corallococcus exiguus) oriented themselves parallel to lines of stress in the agar substrate and subsequently formed fruiting bodies at right angles to their direction of movement. He called the phenomenon “elasticotaxis” (Stanier, 1942a).It is interesting that the later Cytophagu monograph contained numerous photographs showing a similar tactic behavior as the cellulose-decomposing cytophagas oriented themselves to the cellulose fibers. A segment in one of Reichenbach’s films showing myxobacterial lysis of prey bacteria, in which the myxobacteria seemed to head directly for the clusters of Sarcina lutea prey (Reichenbach, 1968) particularly

MYXOBACTERIAL BIOLOGY

10 intrigued Dworkin. It seemed to him a good system for demonstrating myxobacterial chemotaxis, and he set about intending to do so. As a first control, to eliminate the possibility that the myxobacteria were responding to some nonspecific physical presence of the cells, Dworkin substituted 10-pm-diameter polystyrene latex beads for the clumps of prey bacteria. To his astonishment some of the myxobacterial swarms seemed to head directly to the plastic beads. To eliminate the possibility that the beads contained a diffusible chemical perceived by the cells, Dworkin substituted washed, incinerated glass beads and watched while the cells repeated their directed movement. After many hours of watching in disbelief as the swarms moved toward the beads, and after subjecting the process to a statistical analysis, Dworkin concluded that the cells were indeed detecting the physical presence of the beads on the agar (Dworkin, 1983). A careful rereading of Stanier’s elasticotaxis paper revealed that in one of his experiments the cells had also responded in an elasticotactic fashion to sterile glass beads which he had scattered over the agar surface. This made Dworkin comfortable in suggesting that the directed movement was an elasticotactic response. More recently, Marta Fontes and Kaiser (Fontes and Kaiser, 1999) quantified the assay for elasticotaxis and, drawing on a collection of A- and S- mutants, showed that the elasticotactic response in M. xanthus requires A-motility but not S-motility. Indeed, the tactic response was enhanced by the absence of S-motility. This suggested to them that S-motility, which requires that the cells move toward other cells ahead of them, competes with elasticotaxis for setting the direction of cell movement. Since in most cases the stress direction would be different from the pilus direction, competition would result. They also observed the progressive reorientation of cells that happened to start parallel to lines of stress in agar. Almost all those cells became perpendicular within 15 min, the time required for gliding a couple of cell lengths. Kaiser (Kaiser, 2003) suggested that elasticotaxis was like following a slime trail and that oriented agarose chains in compressed agar or slime polymer chains in a trail would be expected to have similar orientating effects on an A-motile cell. It seems reasonable to suggest that in nature, the ability of myxobacteria to perceive the presence of a colony of potential prey bacteria resting on their own deformable pad of slime would be useful for predation. It would allow the myxobacteria to move directly to the prey by following a stress line, just as Dworkin’s photographs record (Dworkin, 1983). Dworkin also observed that movement toward a bead depended on A-motility but not on S-motility. Since so little is known about the

specific niches occupied by myxobacteria in the soil, this suggestion is tentative.

Stimulation Early in his investigation of the genetics of gliding motility, Hodgkin made a startling discovery concerning contacts between 111. xanthus cells. He observed that several nonmotile (A-S-) mutants could be stimulated to move transiently by contact with wild-type cells or with cells of a different mutant type. Most of his motility mutants were not rescuable, but tgl, cglB, cglC, cglD, cglE, and cglF mutants could be stimulated (Hodgkin and Kaiser, 1977). Because stimulation did not occur when contact between cells was prevented, he also learned that stimulation depends on close apposition of interacting cells (Hodgkin and Kaiser, 1977). Subsequent progress in understanding stimulation required knowing the molecular function and the cellular location of stimulatable proteins. Jorge Rodriguez-Soto showed that Tgl protein was essential for S-motility, not for A-motility, and that it is a 27-kDa lipoprotein to be found in the outer membrane of M. xanthus (Rodriguez-Soto and Kaiser, 1997a, 1997b). Dan Wall, Sam Wu, and Kaiser (1998) showed that tgl mutants make ample quantities of PilA pilin, but fail to assemble it into pili. Nudleman, Wall, and Kaiser (Nudleman et al., 2006) demonstrated that Tgl is required for the assembly of PilQ monomers into a multimeric secretin channel in the outer membrane. The pilus elongates through the assembled PilQ channel as it extends outside the cell (Nudleman et al., 2006). These authors also showed that Tgl for PilQ assembly could be provided by stimulation from another cell. Finally, by separating the stimulated recipient cells from donor cells, Nudleman et al. (2005) demonstrated that Tgl protein was transferred from donor to recipient cell. They also demonstrated the transfer from donor to recipient of CglB protein, which is essential for A-motility (Nudleman et al., 2005). CglB had been found to be a lipoprotein (Rodriguez and Spormann, 1999) that also localized to the outer membrane of M. xanthus (Simunovic et al., 2003). The concentration of Tgl and CglB proteins in stimulated cells was found to be similar to the concentration in donor cells, as if the donor and recipient cells shared their mobile outer membrane proteins equally, which thereby created a primitive tissue (Nudleman et al., 2005).

Reversing the Engines Reichenbach’s movies of myxobacterial swarms show individual cells moving alternately along both directions of their long pole-to-pole axis (Kuhlwein and Reichenbach, 1968).Jelsbak’s movie, noted above, also

TO THE GENOME 1. FROMGLYCEROL

illustrates frequent alternation of gliding direction. No cell has either a permanent head or a permanent tail. Studies of the frizzy mutants of M. xanthus have shown that cells have a well-defined average frequency of gliding reversal that is inherited and controlled by the frz genes (Blackhart and Zusman, 1985). Tracking cells in the traveling waves of M. xanthus has shown very regular reversals at 8-min intervals (Welch and Kaiser, 2001). M. xanthus cells clearly do not reverse randomly in time; instead, each cell appears to have its own reversal clock (Igoshin et al., 2004). Finally, reversal is not coupled to the cell cycle because M. xanthus cells can reverse their gliding direction 20 or more times in traveling waves in a single cell cycle. Wild-type M. xanthus colonies on the surface of agar are flat disk-shaped swarms that taper down to a monolayer of cells at their edge. That edge is observed to spread outward symmetrically at a constant rate for many days (Burchard, 1974; Kaiser and Crosby, 1983).The colony of an A-S- strain, which lacks motility, measures the colony expansion that is due to growth. Since an A-Scolony expands at about one-eighth the rate at which the motile (A+S+)swarm expands, the outward spreading of the swarm can be attributed largely to motility, although growth is essential (Kaiser and Crosby, 1983). Since the swarm rate is observed to increase with the cell density of the inoculum, individual cells appear to help each other move outwards (Kaiser and Crosby, 1983). The A-engines cooperate through slime trail following; the S-engines employ pili that bind fibrils on another cell. The advantage of swarming is evident for colonies on the surface of nutrient agar: by spreading the cells out, swarming decreases competition between feeding cells for nutrient that comes from below the agar surface. The selective advantage of swarming appears to be the enhancement of nutrient absorption by the swarm. As described above, both engines are polar. There is evidence that the two engines are located at opposite poles of the cell, and while the S-engine pulls at one end, the A-engine pushes at the other. In addition, the maximum swarming rate of an A+S+strain is 1.6 pm min-' while the sum of the maximum rates of two strains, each having one of the two engines, is 1.0 pm min-l (Kaiser and Crosby, 1983). This synergism implies that the Sand A-engines occupy opposite poles. It also indicates that for each cell, reversal of one engine is highly correlated with reversal of the other engine. Finally, there is evidence that reversal is necessary for swarming. A mutant with Tn5 inserted in the frzE gene reverses only once every 2 h (Shi and Zusman, 1995) compared to the wild type with a reversal every 7 min, and the frzE mutant fails to swarm (Shi et al., 1993). Qualitatively, outward

11 swarming might be explained by periodic reversals that need not be coordinated between cells. A quantitative theory of swarming is needed to relate the reversal frequency, the speed, and the direction of individual cells to the overall rate of swarm expansion. A quantitative theory could be used to test whether periodic reversals that are independent from cell to cell are sufficient to explain swarming, or whether reversals of different cells in the same area must be coordinated by a gradient of nutrient availability. Although the frequency of reversals depends on the genotype and whether cells are growing or are starving (Jelsbak and Sargaard-Andersen, 2000,2002), directional bias in response to an attractant has yet to be demonstrated (Ward and Zusman, 1997). Nevertheless, it has been proposed that 16:l phosphatidylethanolamine, a lipid found in the outer membrane of M. xanthus, may serve as an attractant during fruiting body aggregation because it appears to change the reversal frequency (Kearns et al., 2001).

THE GENOME M . xanthus exhibits multicellular behavior: it feeds as a coordinated group of cells, and when its food supply nears exhaustion, many thousands of cells cooperate to build a fruiting body and to differentiate spores within it. The recently released genome sequence of M. xanthus by Monsanto and The Institute for Genomic Research sheds new light on the origin and regulation of myxobacterial multicellularity (GenBank assession no. CPOOOll3). The sequence revealed a large genome of 9.14 Mb, 7,388 predicted coding sequences, and a gene density comparable to that of E. coli. The sequence confirms the membership of M. xanthus in the delta subgroup of proteobacteria, but the six other sequenced deltaproteobacteria range from 3.6 to 3.9 Mb in size. This raises the question-how did the M. xanthus genome grow to 9.1 Mb from the size of 3.9 Mb or below that is found in all other deltaproteobacteria? A myxobacterium-specific genome expansion is suggested by the similarity in size of M. xanthus and two species of Stigmatella (a rough draft of one Stigmatella sequence is available at The Institute for Genomic Research). A substantial part of the expansion can be traced to the duplication of individual genes followed by differentiation of function between duplicates. About one-half (48%)of the 7,388 predicted coding sequences in M. xanthus are members of families of closely related sequences. Moreover, 16% of the M. xanthus genome constitutes families of paralogous proteins. Paralogs are more closely related to one another than they are to any protein from any other organism whose genome has been sequenced. Since it appears that many of these paralogs

MYXOBACTERIAL BIOLOGY

12

arose by gene duplication in the ancestors of M. xanthus, they most likely represent myxobacterial-lineage-specific duplications. The lineage-specific duplications found in 111.xanthus are far from a random sample of the genome; rather, they involve particular functions, and because the duplications survived, those functions must have been important to M. xanthus. Among the duplications are genes for sensing and signaling, proteolysis, predation, and development. For sensing, the lineage-specific duplications include Ser/Thr protein kinases (a total of 99 proteins; at least 20 are lineage specific), sigma 54 enhancer-binding proteins (about 50), 137 sensor and hybrid histidine protein kinases and a roughly equal number of response regulators, and about 40 extracytoplasmic sigma factors that respond to extracellular signals (Helmann, 2002). In combination these proteins may create complex sensory circuits for regulating transcription in M. xanthus. There is experimental evidence for a Ser/Thr protein kinase, sigma 54 enhancer-binding protein combination (Jelsbak et al., 2005). Such multistep regulators would resemble those that control embryonic development in multicellular eukaryotes in having sensory input at multiple steps of the circuit. Apparently the multistep regulators were gained by M. xanthus at the expense of one-component regulators that are common in bacteria. The total number of transcriptional regulatory proteins in M. xanthus lies in the range expected for a genome of 9.1 Mb, but the fraction devoted to multistep regulators is increased compared to other soil organisms with large genomes, like Streptomyces coelicolor. For predation and scavenging, the lack of the ZlvC and ilvD genes, which are necessary for branched-chain amino acid biosynthesis, results in the observed requirement for leucine, isoleucine, and valine in the laboratory and suggests that their natural diet consists mainly of protein. In addition, the genome includes many genes whose products may be used for predation. Among the duplicated genes are copies of chaperones and related proteases-DnaK (15copies), DnaJ, GrpE, ClpX, ClpAB, ClpP, HslU, HslV, HtpX, GroEL (2 copies), and Lon (2 copies). Almost 9 % of the M. xanthus genome encodes enzymes that produce secondary metabolites. This amounts to twice the capacity of S. coelicolor or Streptomyces avermitilis, organisms that are well-recognized producers of secondary metabolites. Considering the large variety of duplicated and differentiated genes, the proposed increase in genome size from an ancestral deltaproteobacterium is likely to have occurred over a long stretch of time. It seems likely that predation involving cooperative cell interactions evolved first, allowing M. xanthus to hunt like a pack of wolves

(Dworkin, 1973). Complex sensory modules, such as those now found in M. xanthus, may have enabled scavenging and predation of whole surface colonies of bacteria found in nature. Given the capacity to hunt, fruiting body development could have evolved to enhance survival of the colony when food is exhausted. Many fruiting body cysts are elevated above the surface, sticky, and attached by a thin stalk-properties that would permit a package of spores to be broken off and then to stick to a small animal that is hunting for food. Thousands of spores would thereby be transported together, and it is likely that fruiting bodies are optimally sized and positioned packages of spores. Insects can move long distances in soil and have sophisticated senses for finding food. When the animal encounters nutrients and feeds, the package of myxospores may be deposited in the organic matter. The spores would germinate together and instantly create a feeding swarm of myxobacteria. This scenario, obviously speculative, calls out for experimental tests and refinements. Those experiments would constitute no more than one step in the overall task of understanding the whole organism that is revealed to us in the M. xanthus genome.

References Adye, J. C., and D. M. Powelson. 1961. Microcyst of Myxococcus xanthus: chemical composition of the wall. J. Bacterial. 81:780-785. Arnold, J. W., and L. Shimkets. 1988a. Inhibition of cell-cell interactions in Myxococcus xanthus by Congo red. J. Bacterial. 1705765-5770. Arnold, J. W., and L. J. Shimkets. 1988b. Cell surface properties correlated with cohesion in Myxococcus xanthus. J. Bacteriol. 1705771-5777. Avery, L., and D. Kaiser. 1983. In situ transposon replacement and isolation of a spontaneous tandem genetic duplication. Mol. Gen. Genet. 191:99-109. Bacon, K. D., R. H. Clutter, M. Kottel, M. Orlowski, and D. White. 1975. Carbohydrate accumulation during myxospore formation in Myxococcus xanthus. J. Bacteriol. 124:16351636. Behmlander, R. M., and M. Dworkin. 1991. Extracellular fibrils and contact-mediated cell interactions in Myxococcus xanthus. J. Bacteriol. 173:7810-7821. Behmlander, R. M., and M. Dworkin. 1994a. Integral proteins of the extracellular matrix fibrils of Myxococcus xantbus. J. Bacteriol. 176:6304-6311. Behmlander, R. M., and M. Dworkin. 1994b. Biochemical and structural analyses of the extracellular matrix fibrils of Myxococcus xanthus. J. Bacteriol. 176:6295-6303. Blackhart, B. D., and D. Zusman. 1985. Frizzy genes of Myxococcus xanthus are involved in control of frequency of reversal of gliding motility. Proc. Natl. Acad. Sci. USA 82~8767-8770.

TO THE GENOME 1. FROMGLYCEROL

Boysen, A., E. Ellehauge, B. Julien, and L. Ssgaard-Andersen. 2002. The DevTprotein stimulates synthesis of FruA, a signal transduction protein required for fruiting body morphogenesis in Myxococcus xanthus. J. Bacteriol. 184:1540-1546. Bretscher, A. P., and D. Kaiser. 1978. Nutrition of Myxococcus xanthus, a fruiting myxobacterium. J. Bacteriol. 133:763768. Burchard, R. P. 1970. Gliding motility mutants of Myxococcus xanthus. J. Bacteriol. 104:940-947. Burchard, R. P. 1974. Growth of surface colonies of the gliding bacterium Myxococcus xanthus. Arch. Microbiol. 96:247254. Burchard, R. P., and M. Dworkin. 1996. Light-induced lysis and carotenogenesis in Myxococcus xanthus. J. Bacteriol. 180535-545. Campos, J., J. Geisselsoder, and D. Zusman. 1978. Isolation of bacteriophage MX4, a generalized transducing phage for Myxococcus xanthus. 1.Mol. Biol. 119:167-178. Chang, B. Y., and M. Dworkin. 1994. Isolated fibrils rescue cohesion and development in the Dsp mutant of Myxococcus xanthus. J. Bacteriol. 176:7190-7196. Downard, J. 1993. Identification of esg, a genetic locus involved in cell-cell signaling during Myxococcus xanthus development. ]. Bacteriol. 175:7762-7770. Dworkin, M. 1962. Nutritional requirements for vegetative growth of Myxococcus xanthus. J. Bacteriol. 84:250-257. Dworkin, M., and S. Gibson. 1964. A system for studying microbial morphogenesis: rapid formation of microcysts in Myxococcus xanthus. Science 146:243-244. Dworkin, M., and W. Sadler. 1966. Induction of cellular morphogenesis in Myxococcus xanthus. I. General description. J. Bacteriol. 91:15 16-151 9. Dworkin, M. 1972. Myxobacteria: new directions in studies of prokaryotic development. Crit. Rev. Microbiol. 1:435-452. Dworkin, M. 1973. Cell-cell interactions in the Myxobacteria. Symp. Soc. Gen. Microbiol. 23:125-147. Dworkin, M. 1983. Tactic behavior of Myxococcus xanthus. J. Bacteriol. 154:452-459. Ellehauge, E., M. Norregaard-Madsen, and L. Ssgaard-Andersen. 1998. The FruA signal transduction protein provides a checkpoint for the temporal coordination of intercellular signals in M . xanthus development. Mol. Microbiol. 30:807-813. Fontes, M., and D. Kaiser. 1999. Myxococcus cells respond to elastic forces in their substrate. Proc. Natl. Acad. Sci. USA 96~8052-8057. Goldman, B. S., W. C. Nierman, D. Kaiser, S. C. Slater, A. S. Durkin, J. A. Eisen, C. M. Ronning, W. B. Barbazuk, M. Blanchard, C. Field, C. Halling, G. Hinkle, 0. Iartchuk, H. S. Kim, C. Mackenzie, R. Madupu, N. Miller, A. Shvartsbeyn, S. A. Sullivan, M. Vaudin, R. Wiegand, and H. B. Kaplan. 2006. Evolution of sensory complexity recorded in a myxobacterial genome. Proc. Natl. Acad. Sci. USA 103: 15200-15205. Gronewold, T. M. A., and D. Kaiser. 2001. The act operon controls the level and time of C-signal production for M. xanthus development. Mol. Microbiol. 40:744-756. Gronewold, T. M. A., and D. Kaiser. 2002. act operon control of developmental gene expression in Myxococcus xanthus. J. Bacteriol. 184:1172-1179.

13 Hagen, D. C., A. P. Bretscher, and D. Kaiser. 1978. Synergism between morphogenetic mutants of Myxococcus xanthus. Dev. Biol. 64:284-296. Helmann, J. D. 2002. The extracytoplasmic function (ECF) sigma factors. Adv. Microb. Physiol. 46:47-110. Henrichsen, J. 1972. Bacterial surface translocation: a survey and a classification. Bacteriol. Rev. 36:478-503. Hodgkin, J., and D. Kaiser. 1977. Cell-to-cell stimulation of movement in nonmotile mutants of Myxococcus. Proc. Natl. Acad. Sci. USA 74:2938-2942. Hodgkin, J., and D. Kaiser. 1979a. Genetics of gliding motility in M . xanthus (Myxobacterales): genes controlling movement of single cells. Mol. Gen. Genet. 171:167-176. Hodgkin, J., and D. Kaiser. 197913. Genetics of gliding motility in M . xanthus (Myxobacterales):two gene systems control movement. Mol. Gen. Genet. 171:177-191. Igoshin, O., A. Mogilner, R. Welch, D. Kaiser, and G. Oster. 2001. Pattern formation and traveling waves in myxobacteria: theory and modeling. Proc. Natl. Acad. Sci. USA 98~14913-14918. Igoshin, O., A. Goldbetter, D. Kaiser, and G. Oster. 2004. A biochemical oscillator explains the developmental progression of myxobacteria. Proc. Natl. Acad. Sci. USA 101:15760-15765. Jahn, E. 1924. Beitrage zur botanischen Protistologie. I. Die Polyangiden. Gebruder Borntraeger, Leipzig, Germany. Jelsbak, L., and L. Ssgaard-Andersen. 1999. The cell-surface associated C-signal induces behavioral changes in individual M . xanthus cells during fruiting body morphogenesis. Proc. Natl. Acad. Sci. USA 965031-5036. Jelsbak, L., and L. Ssgaard-Andersen. 2000. Pattern formation: fruiting body morphogenesis in Myxococcus xanthus. Curr. Opin. Microbiol. 3:637-642. Jelsbak, L., and L. Ssgaard-Andersen. 2002. Pattern formation by a cell-surface associated morphogen in M . xanthus. Proc. Natl. Acad. Sci. USA 99:2032-2037. Jelsbak, L., M. Givskov, and D. Kaiser. 2005. Enhancer-binding proteins with a forkhead-associated domain and the sigma54 regulon in Myxococcus xanthus fruiting body development. Proc. Natl. Acad. Sci. USA 102:3010-3015. Julien, B., A. D. Kaiser, and A. Garza. 2000. Spatial control of cell differentiation in Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 97:9098-9103. Kaiser, A. D., and C. Crosby. 1983. Cell movement and its coordination in swarms of Myxococcus xanthus. Cell Motil. 3:227-245. Kaiser, D., and M. Dworkin. 1975. Gene transfer to myxobacterium by Escherichia coli phage P1. Science 187:653-654. Kaiser, D. 2003. Coupling cell movement to multicellular development in myxobacteria. Nut. Rev. Microbiol. 1:45-54. Kaiser, D. 2004. Signaling in myxobacteria. Annu. Rev. Microbiol. 58:75-98. Kaiser, D., and R. Yu. 2005. Reversing cell polarity: evidence and hypothesis. Curr. Opin. Microbiol. 8:216-221. Kaplan, H. B., A. Kuspa, and D. Kaiser. 1991. Suppressors that permit A signal-independent developmental gene expression in Myxococcus xanthus. J. Bacteriol. 173:1460-1470. Kaplan, H. B., and L. Plamann. 1996. A Myxococcus xanthus cell density-sensing system required for multicellular development. FEMS Microbiol. Lett. 139539-95.

14

MYXOBACTERIAL BIOLOGY

LaRossa, R., J. Kuner, D. Hagen, C. Manoil, and D. Kaiser. Kearns, D. B., A. Venot, J. T. Bonner, B. Stevens, G.-J. Boons, 1983. Developmental cell interactions in Myxococcus: analand L. J. Shimkets. 2001. Identification of a developmental ysis of mutants. J. Bacteriol. 153:1394-1404. chemoattractant in Myxococcus xanthus through metabolic engineering. Proc. Natl. Acad. Sci. USA 98:13990-13994. Lee, B.-U., K. Lee, J. Mendez, and L. J. Shimkets. 1995. A tactile sensory system of Myxococcus xanthus involves an Keseler, I. M., and D. Kaiser. 1995. An early A-signalextracellular NAD(P)+-containingprotein. Genes Dev. 9: dependent gene in Myxococcus xanthus has a sigma-54-like promoter. J. Bacteriol. 177:4638-4644. 2964-2973. Li, S., B. U. Lee, and L. Shimkets. 1992. csgA expression Kim, S. K., and D. Kaiser. 1990a. Purification and properties of Myxococcus xanthus C-factor, an intercellular signaling entrains Myxococcus xanthus development. Genes Dev. 6:401-410. protein. Proc. Natl. Acad. Sci. USA 873635-3639. Li, Y., H. Sun, X. Ma, A. Lu, R. Lux, D. Zusman, and W. Shi. Kim, S. K., and D. Kaiser. 1990b. C-factor: a cell-cell signal2003. Extracellular polysaccharides mediate pilus retraction ling protein required for fruiting body morphogenesis of during social motility of Myxococcus xanthus. Proc. Natl. M. xanthus. Cell 61:19-26. Acad. Sci. USA 1005443-5448. Kim, S. K., and D. Kaiser. 1990c. Cell alignment required in difLicking, E., L. Gorski, and D. Kaiser. 2000. A common step ferentiation of Myxococcus xanthus. Science 249:926-928. for changing the cell shape in fruiting body and starvationKim, S. K., and D. Kaiser. 1991. C-factor has distinct aggregaindependent sporulation of Myxococcus xanthus. J. Bactetion and sporulation thresholds during Myxococcus develrial. 182:3553-3558. opment. J. Bacteriol. 173:1722-1728. Lobedanz, S., and L. Ssgaard-Andersen. 2003. Identification of Kimsey, H. H., and D. Kaiser. 1991. Targeted disruption of the the C-signal, a contact-dependent morphogen coordinating Myxococcus xanthus orotidine 5’-monophosphate decarmultiple developmental responses in Myxococcus xanthus. boxylase gene: effects on growth and fruiting-body developGenes Dev. 17:2151-2161. ment. /. Bacteriol. 173:6790-6797. T. H., and H. D. McCurdy. 1976. Gliding motilMacRae, Kroos, L., and D. Kaiser. 1984. Construction of TnSlac, a ity mutants of Myxococcus xanthus. Can. J. Microbiol. transposon that fuses lacz expression to exogenous promot22: 1282-1 292. ers, and its introduction into Myxococcus xanthus. Proc. Manoil, C., and D. Kaiser. 1980. Accumulation of guanosine Natl. Acad. Sci. USA 81:5816-5820. tetraphosphate and guanosine pentaphosphate in MyxococKroos, L., A. Kuspa, and D. Kaiser. 1986. A global analysis cus xanthus during starvation and myxospore formation. of developmentally regulated genes in Myxococcus xanthus. J. Bacteriol. 141:297-304. Dev. Biol. 117:252-266. Martin, S., E. Sodergren, T. Masuda, and D. Kaiser. 1978. Kroos, L., and D. Kaiser. 1987. Expression of many develSystematic isolation of transducing phages for Myxococcus opmentally regulated genes in Myxococcus depends on a xanthus. Virology 88:44-53. sequence of cell interactions. Genes Dev. 15340-854. Nudleman, E., and D. Kaiser. 2004. Pulling together with type Kroos, L., P. Hartzell, K. Stephens, and D. Kaiser. 1988. A IV pili. J. Mol. Microbiol. Biotechnol. 752-62. link between cell movement and gene expression argues that Nudleman, E., D. Wall, and D. Kaiser. 2005. Cell-to-cell motility is required for cell-cell signalling during fruiting transfer of bacterial outer-membrane lipoproteins. Science body development. Genes Dev. 2:1677-1685. 309~125-127. Kroos, L., A. Kuspa, and D. Kaiser. 1990. Defects in fruiting Nudleman, E., D. Wall, and D. Kaiser. 2006. Polar assembly body development caused by Tn5lac insertions in M. xanof the type IV pilus secretin in Myxococcus xanthus. Mol. thus. J. Bacteriol. 172:484-487. Microbiol. 60:16-29. Kruse, T., S. Lobendanz, N. M. S. Bertheleson, and L. SsgaardPlamann, L., A. Kuspa, and D. Kaiser. 1992. Proteins that Andersen. 2001. C-signal: a cell surface-associated morphorescue A-signal-defective mutants of Myxococcus xanthus. gen that induces and coordinates multicellular fruiting body J. Bacteriol. 174:3311-3318. morphogenesis and sporulation in M. xanthus. Mol. Microbiol. 40: 156-1 68. Ramaswamy, S., M. Dworkin, and J. Downard. 1997. Identification and characterization of Myxococcus xanthus mutants Kuhlwein, H., and H. Reichenbach. 1968. Swarming and Morphogenesis in Myxobacteria, Archangium, M~XOCOCCUS, deficient in calcofluor white binding.]. Bacteriol. 179:28632871. Chondrococcus, Chondromyces. Film C893/1965. Institut fur den Wissenschaftlichen. Film, Gottingen, Germany. Ramsey, W. S., and M. Dworkin. 1968. Microcyst germination in Myxococcus xanthus. J. Bacteriol. 95:2249-2257. Kuner, J., L. Avery, D. E. Berg, and D. Kaiser. 1981. Uses of transposon Tn5 in the genetic analysis of Myxococcus xanReichenbach, H. 1965. Rhythmic motion in swarms of Myxothus, p. 128-132. In D. Schlessinger (ed.), Microbiologybacteria. Ber. Dtsch. Bot. Ges. 78:102-105. 2 981. American Society for Microbiology, Washington, DC. Reichenbach, H. 1966. Myxococcus spp. (Myxobacterales) Kuspa, A., L. Kroos, and D. Kaiser. 1986. Intercellular signalSchwarmentwicklung und Bildung uon Protocysten. Institut ing is required for developmental gene expression in Myxofur den Wissenschaftlichen Film, Gottingen, Germany. coccus xanthus. Dev. Biol. 117:267-276. Reichenbach, H. 1968. Archangium violaceum (Myxobacteriales) Schwarmentwicklung und Bildung von Protocysten. Kuspa, A., L. Plamann, and D. Kaiser. 1992. Identification of heat-stable A-factor from Myxococcus xanthus. J . Bacteriol. Film E 777/1965. Institut fur den Wissenschaftlichen Film, 1 7 4 ~ 319-3326. 3 Gottingen, Germany.

1. FROMGLYCEROL TO THE GENOME Reichenbach, H. 1974. Chondromyces apiculatus (Myxobacteriales) Schwarmentwicklung und Morphogenese. Film E 779/1965. Institut fur den Wissenschaftlichen Film, Gottingen, Germany. Rodriguez, A. M., and A. M. Spormann. 1999. Genetic and molecular analysis of cglB, a gene essential for single-cell gliding in Myxococcus xanthus. J. Bacteriol. 181:438 1-4390. Rodriguez-Soto, J. P., and D. Kaiser. 1997a. The tgl gene: social motility and stimulation in Myxococcus xanthus. J. Bacteriol. 179:4361-4371. Rodriguez-Soto, J. P., and D. Kaiser. 1997b. Identification and localization of the tgl protein, which is required for Myxococcus xanthus social motility. J. Bacteriol. 179:4372-4381. Rosenberg, E., K. H. Keller, and M. Dworkin. 1977. Cell density-dependent growth of Myxococcus xanthus on casein. J. Bacteriol. 129:770-777. Sadler, W., and M. Dworkin. 1966. Induction of cellular morphogenesis in Myxococcus xanthus. 11. Macromolecular synthesis and mechanism of inducer action. J. Bacteriol. 91:1520-1525. Sager, B., and D. Kaiser. 1993. Two cell-density domains within the Myxococcus xanthus fruiting body. Proc. Natl. Acad. Sci. USA 90:3690-3694. Sager, B., and D. Kaiser. 1994. Intercellular C-signaling and the traveling waves of Myxococcus. Genes Dev. 8:2793-2804. Schmidt-Lorenz, W., and H. Kuhlwein. 1968. Intracellulare Bewegungsorganellen der Myxobakterien. Arch. Mikrobiol. 60:95-98. Shi, W., T. Kohler, and D. R. Zusman. 1993. Chemotaxis plays a role in the social behaviour of Myxococcus xanthus. Mol. Microbiol. 9:601-611. Shi, W., and D. R. Zusman. 1995. The frz signal transduction system controls multicellular behavior in Myxococcus xanthus, p. 419-430. In J. A. Hoch and T. J. Silhavy (ed.), TwoComponent Signal Transduction. ASM Press, Washington, DC. Simunovic, V., F. C . Gherardini, and L. J. Shimkets. 2003. Membrane localization of motility, signaling, and polyketide synthase proteins in Myxococcus xanthus. J. Bacteriol. 185:5066-5075. Singer, M., and D. Kaiser. 1995. Ectopic production of guanosine penta- and tetra-phosphate can initiate early developmental gene expression in Myxococcus xanthus. Genes Dev. 9:1633-1644. Sodergren, E., and D. Kaiser. 1983. Insertions of Tn5 near genes that govern stimulatable cell motility in Myxococcus. J. Mol. Biol. 167:295-310. Ssgaard-Andersen, L., F. Slack, H. Kimsey, and D. Kaiser. 1996. Intercellular C-signaling in Myxococcus xanthus involves a branched signal transduction pathway. Genes Dev. 10:740-754. Ssgaard-Andersen, L., M. Overgaard, S. Lobedanz, E. Ellehauge, L. Jelsbak, and A. A. Rasmussen. 2003. Coupling gene expression and multicellular morphogenesis during fruiting body formation in Myxococcus xanthus. Mol. Microbiol. 48:l-8. Stanier, R. Y. 1940. Studies on the cytophagas. J . Bacteriol. 40:619-635.

15 Stanier, R. Y. 1942a. Elasticotaxis in myxobacteria. J . Bacterial. 44:405-412. Stanier, R. Y. 194213. The cytophaga group: contributions to the biology of the myxobacteria. Bacteriol. Rev. 6:143196. Sudo, S. Z., and M. Dworkin. 1969. Resistance of vegetative cells and microcysts of Myxococcus xanthus. J. Bacteriol. 98:8 83-887. Thony-Meyer, L., and D. Kaiser. 1993. devRS, an autoregulated and essential genetic locus for fruiting body development in Myxococcusxanthus. J. Bacteriol. 175:74507462. Wall, D., and D. Kaiser. 1999. Type IV pili and cell motility (MicroReview). Mol. Microbiol. 32:l-10. Wall, D., S. S. Wu, and D. Kaiser. 1998. Contact stimulations of Tgl and type IV pili in Myxococcus xanthus. J. Bacteriol. 180:759-761. Ward, M. J., and D. R. Zusman. 1997. Regulation of directed motility in Myxococcus xanthus. Mol. Microbiol. 245385893. Weidel, W., and H. Pelzer. 1964. Bagshaped macromoleculesa new outlook on bacterial cell walls. Adv. Enzymol. 26: 193-232. Welch, R., and D. Kaiser. 2001. Cell behavior in traveling wave patterns of myxobacteria. Proc. Natl. Acad. Sci. USA 98:14907-14912. White, D., M. Dworkin, and D. J. Tipper. 1968. Peptidoglycan of Myxococcus xanthus: structure and relation to morphogenesis. J. Bacteriol. 95:2186-2197. White, D. 1984. Structure and function of myxobacteria cells and fruiting bodies, p. 51-67. In E. Rosenberg (ed.), Myxobacteria, Development and Cell Interactions. SpringerVerlag, New York, NY. Wireman, J., and M. Dworkin. 1975. Morphogenesis and developmental interactions in the Myxobacteria. Science 189~516-523. Witkin, S., and E. Rosenberg. 1970. Induction of morphogenesis by methionine starvation in Myxococcus xanthus: polyamine control. J. Bacteriol. 103:641-649. Wolgemuth, C., E. Hoiczyk, D. Kaiser, and G. Oster. 2002. How myxobacteria glide. Curr. Biol. 12:369-377. Wu, S. S., J. Wu, and D. Kaiser. 1997. The Myxococcus xanthus pilT locus is required for social gliding motility although pili are still produced. Mol. Microbiol. 23:109-121. Yang, Z., X. Ma, L. Tong, H. B. Kaplan, L. J. Shimkets, and W. Shi. 2000. The Myxococcus xanthus dif genes are required for the biogenesis of cell surface fibrils essential for social gliding motility. J. Bacteriol. 1825793-5798. Youderian, P., N. Burke, D. J. White, and P. L. Hartzell. 2003. Identification of genes required for adventurous gliding motility in Myxococcus xanthus with the transposable element mariner. Mol. Microbiol. 49555-570. Yu, R., and D. Kaiser. 2007. Gliding motility and polarized slime secretion. Mol. Microbiol. 63:454-467. Zusman, D. 1984. Developmental program of Myxococcus xanthus, p. 185-213. In E. Rosenberg (ed.), Myxobacteria. Springer, New York, NY.

Myxobacteria: Multicellularity and Differentiation Edited by David E. Whitworth 0 2008 ASM Press, Washington, D.C.

Gregory J. Velicer Kristina L. Hillesland

Why Cooperate? The Ecology and Evolution of Myxobacteria

The myxobacteria fascinate through their ability to swarm and hunt as packs and to cooperatively crawl together into complex fruiting bodies. But when starvation strikes, why should tens of thousands of individuals aggregate to sporulate in social structures rather than staying put and sporulating as individuals? What are the selective forces that have led myxobacteria to move, hunt, and develop as social units rather than as dispersed individuals? What were the first evolutionary steps in myxobacterial sociality upon which later innovations were built and integrated? What other microbes do myxobacteria eat, and what can eat the myxobacteria? Within the myxobacteria, who meets whom and who cooperates with whom? Molecular biologists have made great strides in understanding the genetics and molecular mechanisms that underlie social behavior in Myxococcus xanthus and a few other species. Much territory remains to be explored on the front of M. xanthus social genetics, with new genes that affect social traits and their combinatorial interactions being identified and analyzed at an increasing rate. However, as M. xanthus and other species become increasingly well characterized at the genetic and biochemical levels, the challenges of understanding

3

the natural ecology, evolutionary histories, and diversity of myxobacteria also loom large on the research horizon. Many fundamental ecological and evolutionary questions remain to be satisfactorily answered. Ecological and evolutionary research seeks to understand the many forces that shape the diversity and distribution of organisms, a complex endeavor that remains in its exciting infancy for the myxobacteria. In this chapter, we first describe research on general ecological and evolutionary issues with the myxobacteria such as their diversity and distribution, population structure, and issues relating to their predatory behavior. Then, we discuss what is known about the sociobiology of these organisms.

ECOLOGY AND EVOLUTION OF THE MYXOBACTERIA Investigating the Abundance, Distribution, and Diversity of Species Where do myxobacteria live and why? Why are they so abundant in some places but not others? What is their relationship to the communities in which they live? These questions form the basis of myxobacterial ecology. More generally, the science of ecology is concerned ~

~~~

Gregory J. Velicer, Department of Biology, Indiana University, Bloomington, IN 47405. Kristina L. Hillesland, University of Washington, Department of Civil and Environmental Engineering, Seattle, WA 98195-2700.

17

18 with identifying the factors that define the distribution and abundance of species in nature and the processes responsible. These factors can be abiotic, such as temperature or humidity, or biotic, such as the abundance of species that are competitors or predators. Such ecological variables may mediate the action of natural selection across genetic variants within populations and thereby differentially affect their relative frequencies. Our understanding of ecology therefore also informs core issues in evolutionary biology. Evolutionary biology seeks not only to illuminate the historical relationships of species, lineages, genomes, and genes, but also to understand the forces and processes that generated the full diversity of organismic traits (from genomes to social behaviors) in the past and those that continue to shape them in the present. Investigation of the myxobacteria from an evolutionary perspective therefore involves answering challenging questions such as the following. How do myxobacteria benefit from forming fruiting bodies, maintaining multiple motility systems, or hunting in groups? How did the tendency to produce various complex fruiting bodies arise, and under what conditions might this ability be lost? Are some myxobacterial traits more likely to be found among populations living in soil than those living in sand or water? There are two complementary research strategies to address such a wide range of ecological and evolutionary questions. The first approach is to observe, describe, and analyze features of natural populations and communities. This approach has greatly inspired and informed our current understanding of ecology and evolution. For example, the most common approach to evolutionary studies is the comparative method, in which phylogenetic relationships or past evolutionary forces are inferred from patterns of trait variation among extant organisms. Evolutionary biologists might also follow changes in genotype frequencies and other traits over time in natural populations to better understand evolutionary forces at play in the present (Grant and Grant, 2002; Hanslti and Saccheri, 2006). An ecologist may measure the abundance of multiple populations across space and time and relate that to physiological properties of the organism and other measured properties of the environment or community. Most studies of natural myxobacteria populations have involved isolating species from diverse environments across the globe and looking for relationships between particular environmental features and the ability to isolate particular species. While these approaches with natural populations are of unquestionable value and are ultimately necessary for understanding natural communities of myxobacteria,

MYXOBACTERIAL BIOLOGY they also have inherent limitations. First, as with all microorganisms, it is difficult to observe and manipulate them directly in their natural environment. Modern molecular approaches employed in environmental microbiology such as sequencing DNA and RNA from environmental samples and fluorescence in situ hybridization improve our ability to link the presence or absence of species and genes to particular communities and environments (Amann et al., 1995; Olsen et al., 1986), but it is still difficult to measure phenotypes such as the rate of swarming across a surface or the rate of prey killing in natural conditions. Second, because natural environments are so complex, it is difficult to discern which features of the environment are responsible for changes in population abundance or mediate the action of natural selection (Endler, 1986).This issue is especially relevant in evolutionary studies employing the comparative method for which there may be poor knowledge of the selective environments and other forces that shaped lineages of interest in the past. Third, rigorous tests of some hypotheses may require combinations of ecological variables that do not occur naturally. Thus, complete reliance on studies of natural populations to inform our understanding of myxobacterial ecology and evolution would limit the range of questions that can be addressed. These limitations can be overcome by using model species such as 211. xanthus to test ecological and evolutionary hypotheses under controlled laboratory conditions. Ecological questions can be addressed by constructing alternative laboratory environments with only one or a few variables that differ between them. This approach allows the researcher to rigorously test the role of spatial structure or prey abundance on the performance of a species, or stability of a simple community (Bohannan and Lenski, 2000; Elena and Lenski, 2003; Jessup et al., 2004; Kassen and Rainey, 2004). Another powerful technique is experimental evolution, in which an organism is propagated in one or more simple, controlled environments for many generations. Phenotypic characteristics of the evolved populations, particularly competitive fitness, can then be measured and compared to those of the ancestral genotype to test a hypothesis of interest (Lenski et al., 1991). Experimental evolution has been used to address evolution of aging (Rose, 1984), host-parasite coevolution (Bohannan and Lenski, 2000; Elena, 2002), the role of chance and history in shaping evolutionary trajectories (Travisano et al., 1995), the evolution of social behavior (Rainey and Rainey, 2003; Turner and Chao, 1999; Velicer and Yu, 2003),and a variety of other topics that have been reviewed by Elena and Lenski (2003) and Feldgarden et al. (2003).

AND EVOLUTION OF MYXOBACTERIA 2. ECOLOGY

As described in this chapter, such laboratory-based approaches to testing ecological and evolutionary hypotheses have been applied to the myxobacteria in recent years. These approaches not only are enhancing our understanding of the myxobacteria per se, but also have the potential to inform our understanding of ecology and evolution more generally. Microorganisms such as the myxobacteria have several unique features which increase the range and power of experimental evolution for addressing hypotheses about adaptation relative to what is possible with many eukaryotic model organisms (Elena and Lenski, 2003). First, they can be stored frozen for long periods and then resuscitated. This property allows for direct comparisons of ancestors and evolved clones if both are stored in the freezer as they are generated. Second, many microbes have very short generation times and thus allow evolution experiments thousands of generations long to be conducted within the course of a Ph.D. degree. Third, asexual reproduction allows independent populations derived from a single ancestral organism to be evolved under the same conditions. Such repetition of evolutionary trials allows the researcher to distinguish between changes caused by natural selection and those that were likely due to random forces. Since Roland Thaxter first formally described the myxobacteria late in the nineteenth century (Thaxter, 1892), most ecological and evolutionary research has focused on the isolation, description, and classification of species from a variety of environments and the definition of their phylogenetic relationships (Shimkets et al., 2005). In this section we briefly summarize this research as well as newer cultivation and molecular studies that have advanced our understanding of the diversity of the myxobacteria, their habitats, and the structure of their natural populations. We then proceed to describe some laboratory studies addressing the effects of abiotic variables on ecologically relevant phenotypes and the interactions of myxobacteria with prey species and how such interactions affect predator evolution.

Natural Diversity and Distributions

Who Are the Myxobacteria? The myxobacteria form a monophyletic group in the order Myxococcales in the delta subgroup of the proteobacteria. They are most closely related to sulfate-reducing bacteria and Bdellovibrio species, which are also predators of bacteria (Shimkets et al., 2005). For a century, taxonomic classification of myxobacterial species was based almost exclusively on morphological traits such as fruiting body shape, size, and color, swarm patterns,

19

and cell shape. However, morphological similarity does not necessarily represent genetic similarity, such that morphology-based phylogenies may fail to accurately model ancestral relationships among species and phenotypes. The advent of DNA-sequence-based classification has opened the possibility of fully understanding not only the patterns of ancestral relatedness among myxobacterial species but also the degree to which social phenotypes reflect those patterns. Sproer et al. (1999) first classified 54 myxobacterial strains representing 10 previously named genera by traditional morphological criteria such as fruiting body phenotype. Subsequently, near-complete 16s rRNA gene sequences obtained for each strain were used to construct a molecular phylogeny. Strains assigned to the same genus by morphological classification tended to cluster tightly in the 16s rRNA phylogeny, providing strong evidence that myxobacterial phenotypes reflect overall genetic relatedness at this level, at least for this one essential gene. The order Myxococcales has traditionally been subdivided into the two suborders Cystobacterineae and Sorangiineae, and the 16s rRNA gene analysis of Sproer et al. (1999) confirms a deep phylogenetic bifurcation between these suborders (Fig. 1).Within the Sorangiineae, the family Nannocystaceae (genus Nannocystis) is deeply divergent from that of Polyangiaceae (genera Chondromyces, Polyangium, and Sorangium). Some have classified the Nannocystaceae as a distinct suborder (Shimkets et al., 2005). The characteristics that define these taxonomic groups are thoroughly described by Shimkets et al. (2005) and Reichenbach (1993), but we briefly summarize the information here. The first suborder, Cystobacterineae, includes the two most thoroughly studied species, Myxococcus xanthus and Stigmatella aurantiaca. The suborder Cystobacterineae contains the families Myxococcaceae (genera M ~ X O C O C CArchangium, US, and Corallococcus) and Cystobacteriaceae (genera Cystobacter, Hyalangium, Melittangium, and Stigmatella). Vegetative cells in this suborder tend to be long rods with tapered ends. Myxospores are much shorter and rounder and tend to have capsules. Swarms of vegetative cells usually remain on the surface of the agar and tend to make striking patterns. In contrast, swarms of the suborder Sorangiineae are often embedded in the agar, sometimes forming pits in the agar surface, and they may not produce pronounced swarming patterns on surfaces. The morphology of vegetative cells of the Sorangiineae differ from those in suborder Cystobacteriaceae in that they are stout with bluntly shaped ends. Myxospores of this suborder tend to be similar in shape to vegetative cells and do not have visible capsules but are always grouped

A

\

Cystobacterineae

AND EVOLUTION OF MYXOBACTERIA 2. ECOLOGY

into sporangioles, which are thick-walled casings for groups of myxospores. All species in the Cystobacteriacede are bacteriolytic, but some species of Sorangiineae are cellulose decomposers. The suborder Sorangiineae includes the family Polyangium, and the genera Sorangium (all cellulose decomposers), Haploangium, Chondromyces (which contain some of the most elaborate fruiting bodies), Byssophaga, and Jahnia. The Nunnocystineae are closely related to the Sorangiineae. They differ morphologically from species in the other two suborders of the Myxobacteria in that species do not produce fruiting bodies, although they may produce sporangioles or spores (Shimkets et al., 2005). Images of fruiting body morphologies representative of the various taxonomic groups can be found in Reichenbach (1993) and Shimkets et al. (2005). Below the genus level, the Sproer et al. (1999)phylogeny does not clearly follow the morphologically based species definitions. This lack of resolution highlights the perennial problem of choosing species definition criteria across bacteria more generally (Gevers et al., 2005) and the question of what constitutes a myxobacterial species in particular. Recent advances in sequencing technology (Margulies et al., 2005; Velicer et al., 2006) should soon allow the definition of entire genome sequences for multiple isolates of each classified species and thus a nearoptimal data set for mapping evolutionary relationships among strains and comprehensive sequence-based criteria for species level classification.

Where Do Myxobacteria Live? Myxobacteria have been isolated all over the globe from a variety of substrates, but the most common environment for most species is soil in tropical to temperate regions (Reichenbach, 1993, 1999; Shimkets et al., 2005). Nutrient-rich soils tend to harbor more myxobacterial

21

species, but they can also be found on rocky surfaces and in pure sand (Reichenbach, 1999). Myxobacteria have also been isolated from animal dung, decaying plant material, animal bark, marine and freshwater environments, and uranium-contaminated U.S. Department of Energy sites (Petrie et al., 2003; Reichenbach, 1999). Most cultured species prefer mild temperatures (20 to 30"C), neutral pH, and high concentrations of organic matter but low ionic concentrations (Reichenbach, 1993, 1999; Shimkets et al., 2005). Thus, there is a higher abundance of myxobacteria and greater density of species in tropical and temperate soils than in locations with more extreme climatic conditions such as Antarctica or highly acidic or alkaline soils. Although myxobacteria have been cultured from these more extreme environments, it has been unclear in many cases whether the cultured organisms were actually capable of growth there (Reichenbach, 1993,1999; Shimkets et al., 2005). In recent years, the range of environmental conditions that may support active growth of myxobacteria has been shown to be greater than expected. Dawid et al. (1988) documented Polyangium and Nunnocysti5 species isolated from Antarctica that grow at 4°C but are unable to grow at moderate temperatures, while Gerth and Miiller (2005) readily isolated strains from warm arid climates with optimum growth temperatures much higher than those of previous myxobacterial isolates (42 to 44°C). Zhang et al. (2005) reported that a marine isolate of Huliangium ochraceum fails to grow in the absence of sodium chloride. Two additional halotolerant strains in the Nannocystineae suborder have also been isolated from marine environments (Iizuka et al., 2003, 1998; Zhang et al., 2005). In their vegetative state, most myxobacterial species require significant aeration, yet a new strain has been isolated from anaerobic soil enrichments. Initial phylogenetic analyses

Figure 1 (A) Neighbor-joining tree of 16s rRNAs showing the phylogenetic position of the type strains of different genera of the order Myxococcales and isolates that were assigned to myxobacterial species on the basis of morphological characteristics (e.g., fruiting bodies, myxospores, and color). The sequences of gram-negative, sulfate-reducing bacteria were used to root the dendrogram. Numbers within the dendrogram indicate the percentages of occurrence of the branching order in 100 bootstrapped trees. The bar represents 10 nucleotide substitutions per 100 nucleotides. Reprinted with permission from Sproer et al. (1999). (B) Images of fruiting bodies of various species of Myxobacteria. In the Sorangiineae, from left to right: (i) Chondromyces spp. isolated during the Microbial Diversity course at Woods Hole (copyright 1995, D. E. Graham); (ii) Chondromyces crocatus fruiting bodies (photo courtesy of Hans Reichenbach);(iii) Polyungium fumosum fruiting bodies (used with permission from Shimkets et al., 2005). In the Nannocystaceae: (i)Koflerza flava swarm on agar (not fruiting bodies). Used with permission from Shimkets et al., 2005. In the Cystobacterineae: (i) Cystobacter badius fruiting bodies on agar (used with permission from Shimkets et al., 2005); (ii) M. xalzthus fruiting bodies in soil and a fruiting body of M. xanthus on agar (photos courtesy of M. Vos and S. Kadam, respectively).

22 indicate that this strain, called “Anaeromyxobacter,” branches deeply within the Myxococcales and is capable of reducing metals and other pollutants during anaerobic growth on electron donors such as acetate (He and Sanford, 2003; Sanford et al., 2002). That myxobacteria with such diverse optimal growth conditions have been cultivated in these and other studies (Neil et al., 2005; Watve et al., 1999) suggests that our knowledge of the distribution of these organisms is limited by our methods of cultivation and scale of sampling. It has been estimated by Watve et al. (1999) that finer-scale biogeographical sampling will reveal far more species of myxobacteria than have previously been classified, perhaps severalfold more numerous. Varying the temperature, pH, and other parameters in incubation of enrichments and greater utilization of molecular methods may reveal an even greater diversity and broader geographic range of myxobacteria than has been previously known. Modern molecular approaches promise to revolutionize our ability to define myxobacterial communities and populations and are beginning to be used and developed (Vos and Velicer, 2006; Wu et al., 2005; Jiang et al., 2007). Given that the preferred growth conditions of myxobacterial isolates vary substantially, it is reasonable to expect that the distribution of species and strains across environments is nonrandom, i.e., that some species may be more prosperous in certain environments than in others. This issue has been addressed in part by perhaps the most comprehensive study of species distribution patterns to date. Dawid (2000) examined 1,398 soil samples from 64 nations or states on all continents by using standard protocols (including three different isolation methods) throughout the long-term study. Strikingly, one or more myxobacterial species could be isolated from the vast majority (91%) of soil samples worldwide, demonstrating the ubiquitous success of this order in occupying a wide spectrum of terrestrial soils. Acidic soils and soils from low-temperature zones yielded fewer species than samples with more moderate parameters, and the highest average species diversity was found in samples from tropical climate zones. Four species (Corallococcuscoralloides, Archangium gephyra, Myxococcus fulvus, and a Polyangium species) were found in more than 40% of samples, 6 species in 10 to 35% of samples, 8 species in 5 to 10% of samples, and 21 species appeared in fewer than 5 % of samples. The species distribution reported by Dawid differed from that of previous studies, highlighting the limitations of comparisons across independently designed isolation studies and the need to develop sophisticated molecular tools for rapid quantification of myxobacterial species types and distributions within soil samples.

MYXOBACTERIAL BIOLOGY Spatial Structure in the Global M. xanthus Population Genetic diversity in natural populations of higher organisms is highly structured due t o both limited migration and adaptation to local conditions, such that spatially distinct populations are often genetically differentiated at numerous genetic loci. However, the degree to which genetic variation in microbial species is nonrandomly distributed is a long-standing problem in microbial ecology (Baas Becking, 1934; Fenchel, 2003; Martiny et al., 2006). Geographically distant populations of thermophiles (Whitaker et al., 2003) and pathogens that occupy specialized, noncontiguous habitats (Linz et al., 2007) are known to show distinct patterns of genetic variation. However, a recent study has demonstrated that genotypes of the soil bacterium M. xanthus, which has a relatively contiguous distribution across terrestrial soil ecosystems, are also nonrandomly distributed across large spatial scales (Vos, 2006). In a survey of M. xanthus populations sampled at the meter scale within each of 1 0 globally distributed sites, it was found that most populations (38 of 45 pairwise comparisons) show significantly distinct patterns of genetic variation across several highly conserved genes (Fig. 2). Moreover, the degree of genetic divergence between populations was found to correlate significantly with distance between the populations. Several European populations separated by only hundreds of kilometers showed no significant divergence, but all populations separated by >1,700 km were found to be distinct. The distribution of variation below the 100-km scale in Germany was found to be largely homogeneous, indicating thorough dispersal of genotypes at this scale. M. xanthus populations tend to be genetically distinct when separated by large distances, but what evolutionary mechanisms drive this nonrandom distribution? One possibility is that populations are adapting to a wide range of ecological habitats across broad spatial scales and that genetic differences between populations reflect such adaptation. Alternatively, populations might be limited by dispersal such that they diverge by randomly accumulating different sets of selectively neutral nucleotide changes (a process termed “genetic drift”). Baas Becking (1934) famously speculated that dispersal does not limit the distribution of microbes due to their immense numbers and small size, but rather local adaptation to heterogeneous environments determines what species and genotypes are found across different habitats. Although Whitaker et al. (2003) presented evidence that limited dispersal can indeed lead to genetic divergence between hot-spring thermophile populations, soil bacteria

2. ECOLOGY AND EVOLUTION OF MYXOBACTERIA

A

O0..?8 0.7

23

&

1

o = 5

A

rn I

5.5

6.5

6

7

7.5

a

distance (log km) Figure 2 F,, values between 10 different meter scale populations from around the globe plotted against distance between populations ( y 2 = 0.38, P > 0.05, n = 45). F,, values provide a measure of the degree of genetic diversity between populations relative to within populations, in this case based on the sequences of several highly conserved genes (Vos, 2006). Significantly differentiated population pairs are indicated by triangles, and nonsignificantly differentiated population pairs are indicated by squares.

that occupy large swaths of terrestrial soils should disperse more freely. Formation of stress-resistant spores should facilitate survival during dispersal events. Therefore, M. xanthus provides a conservative test of whether such broadly distributed soil bacteria can be limited by dispersal. As was found for several housekeeping genes (Fig. 2), genetic differentiation at the highly variable pilA gene among the 10 global populations of M. xanthus described above was found to increase significantly with distance. However, the average degree of differentiation between populations at pilA was found to be substantially lower than at the housekeeping genes (M. Vos, unpublished data). This result indicates that some piZA genes spread across populations via recombination and are maintained by selection in multiple locations. Nonetheless, both diversity and population differentiation at piZA still increase with distance, thus indicating that dispersal is limited. Furthermore, the increase of differentiation at pilA with spatial scale is caused as much by differences at synonymous sites as at nonsynonymous sites (Vos, unpublished data), as is expected if dispersal is limiting and populations are differentiating by genetic drift. The degree to which genetic differentiation of myxobacterial populations is caused by limited dispersal and subsequent genetic drift versus local adaptation (or some combination of these mechanisms) is a fundamental problem for future research seeking to understand the forces that drive myxobacterial biogeography.

Defining a Local Population The isolation studies described above provide a general picture of how species are distributed across diverse environments, but what does any particular local population of a myxobacterial species look like? How genetically similar are isolates that are close together in the same patch of soil? Which genes underlying social traits are under selection? These questions were addressed in a recent isolation study in which substantial genetic diversity was found among 78 isolates of M. xanthus within a 16- by 16-cm patch of soil in Tubingen, Germany, an area within which encounters among resident genotypes due to local migration are likely (Vos and Velicer, 2006). One hundred soil samples separated by 1.6 cm in a grid design yielded 78 successful clone isolations (one clone per sample for successful isolations), with isolates classified as M. xanthus based on morphology and 16s rRNA gene sequences. Fragments of the csgA, fibA, and pilA genes were sequenced for each, and concatemers of the three fragments revealed 21 distinct genotypes among the 78 clones, generating a prediction that a total of -26 genotypes would have been found in the plot with large increases in sampling effort. Thus, the genetic variation sampled largely reflected the total variation actually present at this spatial scale that was accessible from the utilized isolation protocol. Importantly, there was no statistically significant clustering of identical genotypes within the plot, indicating that the average clonal patch size in this location was smaller than 1.6 cm.

MYXOBACTERIAL BIOLOGY

24

A H o . 0010

/"

A4

99.2

\ 95'2

\

DK1622 '

97.4

A1 7 A5,h12

A75

\ 8e\o

A1

B -0.01

A12, A17 A53,A75

C

A66, A75

-0.1

A2,A3 A9, A53

57.i

A12

A25 A0 A98

The 21 detected genotypes formed six distinct phylogenetic branches, but the structure of the concatemer phylogeny was largely determined by variation in the pilA fragment due to its far greater diversity among isolates than csgA or fibA. Multiple tests provided strong evidence that diversifying natural selection is maintaining greater diversity in a highly variable region of the pilA gene than would be predicted by chance. In contrast, purifying natural selection appears to be purging more variation in the csgA gene than is expected by chance, indicating that differences in the CsgA signal are not responsible for developmental or motility incompatibilities observed among these isolates (see below). There was no evidence that any form of selection was acting on the sequenced portion of fibA. The degree of genetic exchange among individuals of the same species is a crucial determinant of how genomes evolve, and the evolutionary effects of sexual versus asexual modes of reproduction have been extensively modeled, investigated, and debated (Xu, 2004). Among microbes, the degree of genetic exchange (horizontal gene transfer [HGT])might range from levels approaching linkage equilibrium (when there is no statistical correlation between the presence of particular alleles across genes) to the complete absence of recombination among genomes. There have been no reports of spontaneous genetic exchange between M. xanthus cells under laboratory conditions, and tests for linkage disequilibrium among genes suggest that M. xanthus genome evolution is predominantly clonal (Vos and Velicer, 2006). Nonetheless, standard tests for recombination within genes provided evidence that recombination is likely to have occurred within six of nine genes examined among the lineages of 20 randomly chosen isolates. Moreover, comparison of the structure of the csgA, fibA, and pilA gene phylogenies provides additional evidence that HGT does occur in M. xanthus. The relative positions of several strains on the three gene trees were found to be radically incongruent (Fig. 3 ) , indicating transfer of alleles across evolving genome lineages rather than purely clonal Figure 3 Neighbor-joining trees of the csgA (A), fibA (B), and pilA (C) gene fragments. One or more clones were selected as representatives of each major clade in a csgA-fibApilA concatemer phylogeny generated for 78 local isolates from Tubingen, Germany. The laboratory strain DK1622 is included for comparison. Note the highly incongruent positions of strains A12, A17, and A75 across the three gene trees. The bootstrap value (1,000 replicates) is given at each node. Trees are not drawn to the same scale, and values in the upper left corner are genetic distances calculated with the Kimura two-parameter distance model. Reprinted with permission from Vos and Velicer (2006).

2. ECOLOGY AND EVOLUTION OF MYXOBACTERIA diversification, which would have resulted in congruent phylogenetic structures. Natural HGT in M. xanthus might occur by uptake of environmental DNA, ingestion of DNA from prey cells (which might include cannibalized cells of other myxobacteria), phage transduction, or via an undiscovered mode of conjugative transfer. Evidence suggesting HGT events from eukaryotes to myxobacteria has been reported (Porta and Rocha-Sosa, 2001; Quillet et al., 1995) but the frequency of such gene transfer across major taxonomic divisions or among strains of the same species remains unclear.

Ecological Determinants of Genotypic Performance Which myxobacterial phenotypes do we expect to find in particular environments? Answers to this question depend on how genotypes and environments interact to produce particular phenotypes and the relationship between those phenotypes and fitness. Although there have been significant advances in understanding the genetic mechanisms responsible for swarming and fruiting body morphogenesis, little systematic study of phenotypic variability in these traits across genotypes and environments has been conducted. Recently researchers have begun to compare phenotypic variability in swarming (Hillesland and Velicer, 2005; Shi and Zusman, 1993), fruiting body development at different population densities (Kadam and Velicer, 2006), and predatory ability of diverse genotypes (Bull et al., 2002; Pham et al., 2005) in the laboratory. The reproductive rate and survival of myxobacteria genotypes depend on their ability to survive in the absence of resources and on how quickly they can gain access to new ones. For myxobacteria living on surfaces, the latter parameter is tied to their rate of movement in search of food relative to other genotypes. In M. xanthus, swarming toward new food sources is accomplished by two genetically and physiologically distinct motility systems (A-motility and S-motility; see chapter 6 for details [Hodgkin and Kaiser, 1979a, 1979bl). Shi and Zusman (1993) showed that genotypes harboring either A-motility, S-motility, or both vary significantly in their swarming rate on different surfaces. Dually motile genotypes were able to produce large swarms on a range of surfaces, but on soft (0.3%) agar solely A-motile genotypes swarmed quite slowly while solely S-motile genotypes swarmed as fast as dually motile genotypes. On hard (1.5%) agar, solely A-motile genotypes swarmed much faster than solely S-motile genotypes. Further research by Hillesland and Velicer (2005) showed that the differential performance of A- and S-motility on hard versus soft surfaces is qualitatively

25

maintained across a wide range of nutrient concentrations. However, each motility system was found to respond differently to nutrient concentration, and this affected the relative performance of genotypes on the two surface types. The swarming rates of a dually motile genotype, a solely A-motile genotype, and a solely Smotile genotype were measured on hard and soft agar at varying nutrient concentrations. Genotypes with Amotility swarmed proficiently on hard agar across most nutrient concentrations, but S-motility swarming was almost undetectable at most low nutrient concentrations. Above 0.32% Casitone concentrations, the rate of swarming by S-motility on soft agar increased dramatically and was faster than A-motility swarming on hard agar. Figure 4 shows the relationships between the swarming rates of these genotypes across nutrient concentrations. At almost all nutrient concentrations, swarming by the solely A-motile genotype was faster than swarming by the solely S-motile genotype on hard agar (Fig. 4a). This ranking was reversed on soft agar, but the dominance of S-motility swarming was not significant except at high nutrient concentrations. At high nutrient concentrations, the dominance of S-motility swarming over A-motility swarming was very high (Fig. 4b). Finally, the genotype with both motility systems was superior to the solely A-motile genotype across nutrient concentrations on hard agar, indicating that S-motility enhances swarming by A-motility (Fig. 4c). However, A-motility does not seem to enhance S-motility swarming on soft agar (Fig. 4d). These results suggest that in a population containing multiple motility genotypes, solely A-motile genotypes would be at a disadvantage relative to genotypes harboring S-motility provided that nutrients were abundant and the swarming surface resembled soft agar. On hard agar, or at low nutrient concentrations, or both, solely S-motile genotypes should be at a significant disadvantage relative to genotypes that have A-motility. Another feature of 211. xanthus that may be important to fitness in nature and also exhibit variation across genotypes is the relationship between sporulation efficiency and population density. Recently, Kadam and Velicer (2006) documented considerable variation between nine genetically distinct natural isolates in their ability to form fruiting bodies and sporulate efficiently at low population densities. Several strains failed to form fruiting bodies or sporulate effectively at low densities, whereas other strains performed dramatically better at the same low densities (Fig. 5). All strains showed a threshold density below which sporulation efficiency rapidly decreased, but this threshold for one unusual strain was severalfoldlower than that of the strain with the second-lowest threshold. Why

MYXOBACTERIAL BIOLOGY

26

b

4 h

(u c

CJ 3 2 a, c

P

CO C -

-2

0

0.001 0.01

0.1

1

10

% casitone

0

0.001 0.01

0.1

1

10

% casitone

Figure 4 Relative A-motility, S-motility, and dual-motility swarming rates on hard and soft agar across a range of nutrient concentrations. Shown are the natural logs of the ratios of absolute swarming rates for solely A-motile versus solely S-motile genotypes on hard agar (a), solely S-motile versus solely A-motile genotypes on soft agar (b), dually motile versus solely A-motile genotypes on hard agar (c), and dually motile versus solely S-motile genotypes on soft agar (d). Shaded boxes indicate the half of the graph where data points should fall if the indicated genotype swarms comparatively faster than the alternative strain. Closed symbols indicate ratios that were significantly different from zero in a one-sample, one-tailed t test after sequential Bonferroni's correction for multiple comparisons ( P < 0.05 for eight comparisons in the same surface type; some comparisons not shown here). Error bars indicate bounds of the 95% confidence interval about the mean. Used with permission from Hillesland and Velicer (2005)

would there be phenotypic diversity in the ability to form fruiting bodies and spores at low population densities? It is possible that this variability is simply a by-product of variation across natural habitats on another, genetically linked phenotypic trait. However, it may be advantageous in some (but not all) habitats to form fruiting bodies and spores at low population densities. For example, populations evolving in typically nutrient-poor soils may tend to have lower population densities than populations living in habitats that fluctuate between high and low nutrient conditions. Thus, it may be advantageous to form fruiting bodies at low population densities in consistently poor nutrient environments.

Predation and Its Evolution Predators can affect the structure of entire food webs and the diversity of communities depending on whether they prey on competitively dominant or inferior species (Bohannan and Lenski, 2000; Lubchenco, 1978; Spiller and Schoener, 1998).Myxobacteria prey upon a diversity

of prokaryotic species, including cyanobacteria (Daft et al., 1985), various gram-positive organisms, Escherichiu coli, Rhizobiurn, and other gram-negative species (Beebe, 1941; Rosenberg and Varon, 1984). Many species of myxobacteria that have been tested are capable of utilizing multiple species as prey, although distinct species and isolates vary in the range of prey species they are capable of killing (Beebe, 1941; Bull et al., 2002; Rosenberg and Varon, 1984). A given population of myxobacteria may therefore significantly affect the density of a variety of organisms within a community and hence may be important players in the soil food web. To the extent that they prey upon organisms tied to animal or plant health, myxobacteria may also influence communities of macroorganisms. Bull and colleagues (2002) isolated several clones of Myxococcus spp. from organic and conventionally tilled strawberry fields. Assays of predatory capabilities of these myxobacteria indicated that they are capable of enhancing strawberry plant health by inhibiting the growth of fungal plant pathogens.

AND EVOLUTION OF MYXOBACTERIA 2. ECOLOGY

8.0

O

Log (initial cell density) 9.0 9.5

8.5

10.0

t

-1 h

$ -2

.-a, 0 E

;

-3 0 .c m 2 -4 0

Q v)

v

0 3 -5

9

J

-6 -7

I

Figure 5 Log-transformed sporulation efficiencies (spores produced/initial population size) of nine isolates of M. xanthus at five cell densities. Solid lines are used for isolates that sporulate efficiently at -3 x lo8cells/ml, whereas dashed lines indicate isolates that sporulate poorly at this density. Data points show the grand mean of all three replicate estimates, and error bars indicate 95% confidence intervals about the mean. Reprinted with permission from Kadam and Velicer (2006).

However, the Myxococcus spp. isolates were also capable of preying on fungi that may be used to control plant pathogens, so they could have negative effects on plant health as well. Pseudomonads, another potential biocontrol agent, were also susceptible to attack by Myxococcus spp. but could protect themselves by producing antibiotics. This study demonstrates the complexity of possible predator-prey interactions between myxobacteria and the diversity of soil microorganisms. It also shows that natural isolates of myxobacteria vary in their predatory ability. Different Myxococcus isolates exhibited different extents of predation on some species of fungi. Why would there be variation in predatory ability between different isolates from the same genus? To attack prey, myxobacteria must first search them out, a process that requires at least one functional motility system and might be enhanced by elasticotaxis (Fontes and Kaiser, 1999) and one or more chemotaxis systems (Kearns and Shimkets, 2001). Once prey have been found, myxobacteria use bacteriolytic enzymes, proteases, and possibly antibiotics (Reichenbach and Hofle, 1993; Rosenberg and Varon, 1984) to lyse them open and move across

27

the patch of prey and may further employ chemotaxis systems to ensure that they remain within the vicinity of prey populations. The genome contains multiple copies of a variety of proteases and chaperones that may be used in “digesting” prey (Goldman et al., 2006). Other genes, such as those involved in motility and development, may also affect predation. Pham et al. (2005) surveyed the effects of several mutations on the rate of lysis of lawns of prey on nitrocellulose membranes and streaks of prey on agar surfaces. They found that when Serratia marcescens lawns on nitrocellulose membranes were the only available resource for M . xanthus, functional A-motility was necessary for predation, but mutations affecting the S-motility system did not typically affect the rate of lysis. Mutations in the frz chemotaxis system that regulates cell reversal frequencies also negatively affected predation. The frz chemotactic system has also been implicated in predation by studies of single cells devouring microcolonies of E. coli (McBride and Zusman, 1996). In addition to these genes involved in swarm movement, prey lysis was reduced relative to the wild type when each of three genes (the sdeK, asgA, and csgA genes) necessary for early stages of fruiting body formation was mutated. The effects of these mutations on lysis varied significantly depending on the type of prey (Pham et al., 2005). These results show that predation involves many loci, and mutations in these loci may result in diverse predatory phenotypes on different prey species. Thus, significant genetic variation in predatory ability within species is likely, yet largely unexplored. Predatory rates by myxobacteria may also be significantly affected by features of the environment in which predation is taking place. This possibility has been examined for three variables: surface type, prey type, and the density of prey patches (Hillesland et al., 2007). In these experiments, patches were deposited onto a buffered agar surface in a grid configuration, M. xanthus was added to the center patch of the plate, and the plate was incubated for 2 weeks at 32°C (Fig. 6). During this incubation, M. xanthus swarms expanded radially outward across the agar surface in search of patches and then consumed them, as shown in Fig. 6c and d. All three variables influenced the percentage of total patches that was encountered by the swarm in this incubation period. A higher proportion of patches was encountered at high patch density than at low patch density, on hard agar than on soft agar, and on E. coli than on Micrococcus luteus (unless E. coli and M . luteus were on soft agar, in which case the ranking was reversed). Most of the prey within a patch were killed within a matter of a few hours if M. xanthus was equally distributed across the patch regardless of the surface type or which prey was used.

Next Page

28

Figure 6 Growth of M. xanthus on plates covered in patches of prey. Each plate consisted of buffered agar which was overlaid with thick patches of E. coli. A clone of M. xanthus was added to a central patch of E. coli, and photos were taken after 1 day of swarming at high (a) and low (b) patch density and again after 14 days of swarming at high (c) and low (d) patch density.

However, prey were more likely to be recovered from the patch after 24 h of incubation if the patches were distributed on soft agar, indicating that some environments offer greater protection to prey cells than others. Given the possibility of variation in predatory phenotypes due to both genetic and ecological causes, what predatory phenotypes might be expected to evolve in different environments? This question can be addressed through experimental evolution. The density of prey or prey patches affects many predators in the same manner that it affects M. xanthus (Holling, 1959). As shown experimentally above, more prey patches are consumed per unit time when they are densely distributed. At lower densities, prey patches are farther apart and the predator therefore has to spend relatively more time searching for prey at low densities than high densities. Thus, a reasonable hypothesis is that genotypes with faster search rates will have a competitive advantage at low patch densities. At high patch densities, the effect of searching rate on predation will be diminished and the competitive advantage of genotypes with higher searching rates will be less significant. This hypothesis was tested by allowing several replicate populations to evolve at high or low patch density for 2 1 0 0 generations

MYXOBACTERIAL BIOLOGY (Fig. 6). The searching rates of these evolved populations and the ancestor were then estimated by measuring the rate of swarm expansion on buffered agar in the absence of prey. All 16 populations evolved faster swarming on this surface, but as expected the increase in searching ability was much greater for the 8 populations that evolved at low density (-7-fold improvement) than it was for the 8 populations that evolved at high density (-2-fold improvement) (Hillesland, 2005). In another experiment, the fate of predatory ability after evolution in a prey-free environment was tested. Biologists have documented losses in traits that were unnecessary in a variety of species, including E . coli (Cooper et al., 2001; Velicer et al., 1998). Given the complex combination of genes potentially involved in predation and the likely metabolic cost of producing lytic enzymes, we tested whether predatory ability would be easily lost if it was not necessary for competitive success. The predatory abilities of eight populations that had evolved in batch culture in the absence of prey for 1,000 generations and eight populations that had evolved in the absence of prey on hard agar plates were measured in multiple environments along with their common ancestor. The populations that had evolved in the batch culture environment exhibited significant declines in their ability to encounter patches of prey in the grid environment shown in Fig. 6. They were also worse than the ancestor at killing prey in assay environments where searching for patches was not required (Hillesland, 2005; Velicer and Stredwick, 2002). This result suggests that predatory ability is readily lost during evolution in an environment where resources are abundant but prey are unavailable. However, most of the eight populations that had evolved on hard agar plates under selection for improved motility did not exhibit significant changes in predatory ability (Hillesland, 2005; Velicer and Stredwick, 2002).

SOCIOBIOLOGY OF THE MYXOBACTERIA Sophisticated cooperation is the hallmark of myxobacterial behavior, yet it can be argued that the precise evolutionary benefits of cooperation have not been clearly demonstrated for any myxobacterial trait, at least in natural habitats or relevant approximations thereof. While plausible speculation about such benefits has been appropriately abundant, the lack of relevant empirical data highlights the enduring difficultly of rigorous in situ study of microbial behavioral ecology. Beyond defining the precise manner in which cooperation benefits those organisms that engage in it, the evolutionary stability

Previous Page

2. ECOLOGYAND EVOLUTIONOF MYXOBACTERIA of cooperation remains one of the biggest problems in evolutionary biology. This is because cooperative behavior, by definition, benefits the evolutionary fitness of recipient individuals other than a cooperative actor and therefore can potentially be exploited by selfish cheaters. How can cooperation be stable if some individuals increase the representation of their genes in future generations by noncooperation (or “defection”) at the expense of cooperators? The combined challenges of (i) identifying traits that confer fitness benefits to organisms other than a cooperative actor; (ii) defining precisely what those benefits are and who receives them and in what degree; and (iii) understanding the evolutionary, behavioral, and molecular mechanisms that allow cooperative traits to be stably maintained are core themes in sociobiology. Although these challenges were first defined and explored in the context of metazoan cooperation, many microbiologists and evolutionary biologists have recognized that they apply with equal force to a wide variety of behavioral traits expressed by microbes, including the myxobacteria (Axelrod and Hamilton, 1981; Crespi, 2001; Keller and Surette, 2006; Parsek and Greenberg, 2005; Sachs et al., 2004; Smith and Szathmary, 1995; Velicer, 2003; West et al., 2006; Zahavi and Ralt, 1984).

Why Live in Groups or Cooperate in the First Place? Lions live in prides, migratory geese fly together, and newborn mice sleep in piles rather than alone. Ducklings swim in group formation, wolves, African wild dogs, and humans hunt in packs, and Hymenopteran insects construct complex breeding societies in which labor tasks are divided. The pervasiveness of group living across the biological spectrum indicates that organisms often enjoy greater reproductive success as members of social groups than they would with relatively individualistic life histories. However, the precise nature and quantitative degree of evolutionary benefits derived from group living remain unclear for many cooperative organisms, perhaps especially for microbes (Keller and Surette, 2006; Redfield, 2002; Travisano and Velicer, 2004; West et al., 2006). When nonmotile bacteria divide by fission in viscous habitats, physical proximity to colony mates in highdensity groups is unavoidable. Under such forced social proximity, natural selection will favor behavioral traits that succeed best when group life is the only option, and such traits are likely to involve cooperation between cells. However, cells with the capacity to “opt out” of sociality by actively migrating away from neighbors will maintain grouping behavior only if it is advantageous relative to a dispersal strategy (Shimkets, 1999). The

29

A-motility system of M. xanthus readily allows the movement of single cells (Hodgkin and Kaiser, 1979a, 1979b), yet the vast majority of M. xanthus natural isolates produce adhesions and slime tracks that chemically and physically hinder cells from migrating alone (Reichenbach, 1999).Why are such constraints on individualistic dispersal maintained? Here we revisit basic questions about why myxobacterial group behaviors (particularly as represented by M. xanthus) might be evolutionarily advantageous.

Communal Digestion Is there a benefit to myxobacterial grouping specific to food acquisition that requires extracellular catabolism? It has been proposed that individuals in high-density groups of myxobacteria enjoy a higher effective local concentration of extracellular hydrolytic enzymes during predation than single cells or low-density groups and therefore reproduce at a faster rate (Rosenberg et al., 1977). Rosenberg et al. (1977)demonstrated such a benefit of high density in the spatially extreme habitat of thoroughly shaken liquid medium. Cell density had no effect on growth rate when M. xanthus was provided with free amino acids and short peptides that can be transported directly into the cell. In contrast, populations fed with complex, nonhydrolyzed peptides that required extracellular catabolism prior to ingestion grew faster at high density than at low density, presumably due to a higher concentration of catabolic enzymes in the liquid. However, shaken liquid is a radical departure from the dining forum that terrestrial myxobacteria normally experience in the soil. In spatially structured soil habitats, enzyme diffusion should be very limited relative to shaken liquid and therefore density may have a much less pronounced effect on growth rate than in the Rosenberg et al. (1977) experiments. Although the Rosenberg et al. (1977) results are strongly suggestive, it has yet to be demonstrated that cooperative feeding provides a selective advantage in contexts more relevant to most natural myxobacterial habitats. Numerous myxobacteria are cellulolytic, and it will also be of interest to test whether population growth rate is density dependent when extracellular breakdown of cellulose is necessary for energy acquisition in spatially structured habitats. It has been intriguingly proposed that density-enhanced growth during predation may have been foundational to other forms of myxobacterial cooperation (Shimkets, 1990). Stress Support Groups? It has been reported that individual cells of some species are unable to grow in isolation, even in luxurious complex medium (Reichenbach, 1999). However, it is

30 unclear whether this phenomenon is due to actual synergistic interactions among cells or rather to high rates of cell death or inviability upon exposure to laboratory media. If the latter hypothesis is correct, inoculation populations of a minimum size would be necessary to statistically ensure obtaining one or more cells that proceed to grow. Even if synergism between cells is involved, the inability of single cells to grow is not likely to be an inherent trait of these genotypes across all environments. Rather, laboratory media may have parameters that are physiologically stressful for many strains relative to the low-nutrient soil habitats that myxobacteria normally inhabit, much the same way that moderate temperatures are stressful to genotypes adapted to cold habitats (Dawid et al., 1988).Group life may buffer against environmental stress even during growth. Addition of spent medium has been found to sometimes improve the growth of low-density populations, leading to speculation that a quorum-sensing mechanism may be in place to “prevent the futile growth of individual cells” (Reichenbach, 1999).However, it is difficult to imagine the ability of single cells to grow as being futile in an evolutionary sense, because successful population growth (from any starting density) is central to the definition of evolutionary success. Even if the competitive success of cells is greater at high densities during growth, growth from a single cell inherently moves a population toward the advantageous state of high cell density and therefore the retention of this ability should be favored by natural selection. It is therefore expected that the ability to grow as single cells is the norm in environments to which particular genotypes are well adapted. In contrast, some forms of physiological stress might be mitigated by cell-cell interactions within groups and therefore cause growth rate to positively correlate with population density when stress is high. Initial research indicates that pH stress causes the growth rate of M. xanthus strain DK1622 to be highly density dependent on complex, prehydrolyzed medium to a degree that is not observed in the absence of pH stress (H. Peitz and G. Velicer, unpublished results). Stress mitigation during vegetative growth may be a foundational benefit of group living in the myxobacteria. Group Motility Pilus-mediated S-motility in M. xanthus is costly to maintain when it is not important for fitness (Velicer et al., 1998; Velicer et al., 2002), so its maintenance in natural isolates indicates that it confers a significant overall fitness benefit. S-motility (in addition to A-motility) may confer several benefits that favor its evolutionary maintenance. First, studies of swarming by motility mutants on hard

MYXOBACTERIAL BIOLOGY agar surfaces suggest that the A- and S-motility systems are likely to be synergistic over a wide range of natural surfaces (Hillesland and Velicer, 2005; Shi and Zusman, 1993). Second, S-motility may facilitate the action of kin selection in M. xanthus (Hillesland and Velicer, 2005; West et al., 2006). The type IV pili that drive S-motility promote group cohesion and a kin-clustered population structure, thus ensuring that recipients of other forms of cooperative behavior (e.g., developmental signaling) are likely to be close relatives. Third, S-motility is necessary for tight packing of spores within fruiting bodies (Wu et al., 1998), which may itself provide several benefits to M. xanthus, discussed below. Finally, at least in the M. xanthus lab strain DK1622, the A-motility and S-motility systems appear to exhibit distinct swarming “specializations” that are specific to different surface types (hard agar for A-motility and soft agar for S-motility) (Hillesland and Velicer, 2005; Shi and Zusman, 1993). However, the faster swarming of S-motility on soft laboratory surfaces is observed only under abundant resource conditions, suggesting that such environmental specificity of performance may be only a minor component of the total fitness advantage conferred by the maintenance of S-motility. Some natural habitats may harbor ecologically successful strains of M. xanthus (or other species) that lack pilus-driven S-motility. Finally, individual prey and prey microcolonies may more frequently be “bumped into” by a group of several cells moving as a raft than they would by an isolated cell, which would cover a much smaller area. Sporulating Together Why do many myxobacteria sporulate in fruiting bodies rather than alone? A compelling answer to this question is still lacking. Laboratory mutants that sporulate effectively without normal fruiting body formation exist (Velicer et al., 1998), and some species of Nannocystineae do not make fruiting bodies at all (Shimkets et al., 2005). This raises the possibility that numerous myxobacteria in the wild may sporulate without fruiting body construction rather than bothering to pack themselves into a very tight spot with thousands of neighbors and a high probability of not differentiating into spores. Many methods of myxobacterial isolation are biased toward obtaining fruiting genotypes, and there may be many natural nonfruiting strains of species known for their fruiting ability in the laboratory (Velicer et al., 2002; Jiang et al., 2007). An enhanced probability of dispersal to a new, foodrich habitat is a commonly proposed benefit of group sporulation (Kaiser, 2001). Although plausible for some species that make tall fruiting bodies or exhibit

AND EVOLUTION OF MYXOBACTERIA 2. ECOLOGY

light-induced development (White et al., 1980),the status of this hypothesis remains uncertain for any myxobacterial species. The lack of prominent fruiting body stalks or chemotaxis toward warmth and light in some species gives pause when considering the dispersal hypothesis. There are several additional (although nonexclusive) potential benefits to group fruiting as well. Fruiting body sporulation facilitates germination and growth in highdensity groups, which may begin sooner and occur at a faster rate, respectively, than for isolated cells or lowdensity groups (Kaiser, 2001; Rosenberg et al., 1977). There may be some aspect of differentiation within fruiting bodies (e.g., autolytic release of compounds beneficial to spores [Wireman and Dworkin, 19771) that generates individual spores of higher quality than can be produced by asocial or low-density sporulation, thus resulting in greater longevity under stress. Further, the extracellular matrix surrounding spores in simple fruiting bodies (such as those of M. xanthus) and the spore-bearing sporangioles of the suborder Sorangiineae may provide a degree of protection against environmental stresses that would not be available to relatively isolated spores. Such stresses include starvation duration, temperature and pH extremes, caustic biotic (e.g., antibiotics produced by competitors) and abiotic molecules, and ingestion or digestion by would-be predators. The ultimate evolutionary causes and benefits of myxobacterial cooperation remain open to much exploration.

The Physiological Cost of Cooperation All behaviors, whether cooperative or not, are produced at a short-term physiological cost and can be maintained by natural selection only when they yield a net gain in evolutionary fitness across generational time scales. Some cooperative microbial behaviors, such as intercellular signal production, may generate a fitness benefit that exceeds the cost of production only in a group context (Travisano and Velicer, 2004), such that natural selection will favor nonexpression or loss during evolution under conditions where social interaction is limited. Rapid loss of fruiting behavior in myxobacteria and complex multicellular phenotypes in other bacteria during laboratory cultivation is a widespread observation among microbiologists (Branda et al., 2001; Dawid, 2000; Sproer et al., 1999; Velicer et al., 1998).Such losses during domestication may often cause large fitness gains under luxurious laboratory conditions due to reallocation of metabolic resources away from unnecessary functions. The evolutionary fate of unnecessary social functions was examined systematically by allowing 12 independent lineages of M. xanthus all derived from a

31

single clone of strain DIC1622 to evolve for 1,000 generations in nutrient-rich liquid medium (Velicer et al., 1998).This selective regime did not favor proficiency at motility, predation, or development. The evolved populations showed significant adaptation to this “asocial” regime by achieving an average growth rate increase of -28 %. However, this adaptation was associated with major losses in social capacities, with most populations showing complete loss of S-motility and/or fruiting body development. The rapidity (many losses occurred between 200 and 500 generations [our unpublished data]) and convergence of losses across populations indicated that the mutations responsible for inactivation of social capacities rose to high frequency via natural selection rather than chance. In a subsequent study, mutations that caused the loss of S-motility in several lines were localized to the pi1 genes responsible for the generation and function of type IV pili necessary for S-motility (Velicer et al., 2002). It was shown that genetic inactivation of essential pi1 genes in the ancestral strain conferred a fitness advantage to the mutants during growth in the liquid evolution regime. Reciprocally, partial restoration of S-motility in deficient evolved genotypes was detrimental to fitness in liquid medium. These results demonstrated the advantage of eliminating social capacities that are costly to maintain when they do not enhance fitness.

Genetic Conflict over Public Goods Microbial public goods, such as intercellular developmental signals in the myxobacteria, are made accessible to other cells at an immediate cost to the producer. To be maintained in populations over evolutionary time, therefore, the cost of public-good production must be outweighed by direct or indirect fitness benefits, or both, for the producing genotype. Enter cheaters, the perennial bane of social harmony and productivity. Public-good contributions by cooperative actors need not be reciprocated by all recipients. Because such contributions are costly, failure by an individual to proportionately contribute to a public good (“defection”)may be advantageous for fitness in the short run, In this case, natural selection favors selfish cheaters rather than cooperative contributors. If mutational pathways (e.g., in social microbes) or behavioral plasticity (e.g., in social metazoans) readily allows the appearance of cheating behavioral strategies in the presence of cooperators, evolutionary game theory predicts that such cheaters should be a persistent feature in the long-term evolution of cooperative biological systems. When cheating can appear, it will spread unless there exist mechanisms to stop it (Travisano and Velicer, 2004; West et al., 2006).

MYXOBACTERIAL BIOLOGY

32 Various forms of cheating are common in social animals and insects. For example, entire species of social ants and wasps are obligate social parasites in which the queens do not generate their own worker offspring but rather usurp colonies of their social hosts and utilize host workers to raise their own reproductive brood (Bourke and Franks, 1995; Lorenzi et al., 2004). In the myxobacteria, social cheating might occur during social swarming, group predation, or multicellular development, and in fact developmental cheaters can be readily obtained. Velicer et al. (Velicer et al., 2000; Velicer and Stredwick, 2002) examined several genotypes derived from DK1622 that are defective at spore production to varying degrees in pure culture. Some of these defectors were defined mutants defective at A- or C-signal production, whereas others carried undefined defect mutations that had accumulated during evolution in liquid culture (Velicer et al., 1998). Strikingly, more than one-half of the defective genotypes sporulated more efficiently than the fully proficient strain DK1622 when mixed as a small minority (1%) with DK1622, despite their defects in pure culture (Fig. 7). These cheaters included two defined mutants that presumably differ from DK1622 by only their mutations in either asgB or csgA. This result demonstrates the ease of crossing the thin evolutionary line between cooperation and cheating by simple mutational pathways. The full range of M. xanthus genes in which single mutations can cause cheating has yet to be determined. Theory predicts the short-term evolutionary success of cheaters whenever there exists a cooperatively produced public good and accessible mutational or neurological pathways to cheating behavior. Consistent with this expectation, cheating has been documented in several microbial public-good scenarios other than that of M. xanthus development (Greig and Travisano, 2004; Griffin et al., 2004; MacLean and Gudelj, 2006; Rainey and Rainey, 2003; Turner and Chao, 1999; Vulic and Kolter, 2001). Most published examples of microbial social exploitation share the common theme of defectors gaining an advantage by failing to contribute to a group-generated public good in a genetically determined manner, a mode of exploitation that has been termed “obligate cheating.” Obligate cheaters are inherently poor performers under clonal conditions in which the exploited public good is important for fitness. Numerous other microbial public-good scenarios may also be subject to cheating, including quorum sensing and biofilm formation in many species and group predation and social motility in the myxobacteria. Obligate cheating constitutes a potentially enormous problem for microbial cooperation, and mechanisms

-3

-2

-1 Log,, (initial mixing ratio)

0

1

Figure 7 Spore production of an evolved cheater genotype when mixed with its wild-type progenitor (DK1622)at nine initial ratios (a) and the corresponding relative sporulation efficiencies (b). The cheater produces no detectable spores during development in pure culture (data not shown). (a) Squares, triangles, and circles indicate total, DK1622, and cheater spore production, respectively. The expected production of the evolved clones under the hypothesis that DK1622 does not improve the defective strain’s sporulation efficiency ( H l ) is represented by the solid line. The expected production of evolved clones under the hypothesis that the defective strain is rescued to the same efficiency as DK1622 (H2)is represented by the dotted line. The spore production of DK1622 in independent pure cultures is represented by the dashed line. Error bars indicate 95% confidence intervals. (b) Sporulation efficiency of the cheater relative to that of DK1622 for these same initial mixing ratios. The dashed line indicates a relative efficiency of 1. Reprinted with permission from Velicer et al. (2006).Copyright (2006)National Academy of Sciences, United States.

therefore must exist to limit its appearance or effects (Travisano and Velicer, 2004; West et al., 2006). Such mechanisms may include negative pleiotropic effects of defector mutations, targeting of benefits to cooperators, targeted punishment of cheaters, and group-level selection in which individuals in groups without cheaters outperform individuals in distinct groups burdened by heavy cheating loads.

2. ECOLOGY AND EVOLUTION OF MYXOBACTERIA An alternative mode of microbial exploitation does not involve an inherent social defect in the exploiter. In this case, a genotype that is fully proficient by itself at the exploited social task is able to perform that task more efficiently in the presence of a particular competitor (and to that competitor’s expense) than in clonal isolation. Such facultative exploitation has been documented in natural isolates of the eukaryotic slime mold Dictyostelium discoideum and of M. xanthus during painvise development competition mixes with other natural isolates of the same respective species (Fiegna and Velicer, 2005; Strassmann et al., 2000). That obligate cheating can be so readily generated in the lab and that facultative exploitation is so easily detected among natural isolates suggest that both forms of exploitation may be common in natural microbial populations.

The Group Perils of Cheating Obligate cheaters that make no spores in isolation can invade cooperative populations when cheats are rare, but they inherently diminish group productivity when they reach high frequencies. When the exploited cooperative trait is important for survival, obligate cheaters may drive populations to outright extinction. To examine the competitive fates and population-level effects of obligate developmental cheaters, three distinct cheater genotypes were allowed to compete against marked variants of DK1622 through several successive rounds of starvation and growth (Fiegna and Velicer, 2003). Although all three cheaters fail to make any spores in isolation, they had very different effects on population dynamics. One cheater with a defect in CsgA production was maintained at high levels in the population without causing major population crashes. A second cheater was also maintained but caused large decreases in spore production upon reaching high frequencies. A third and particularly virulent cheater caused outright extinction events due to the short-term success of its own “selfish” behavior during development. In some replicate competitions with the latter cheater the entire population went extinct due to cheater-induced sporulation crashes, whereas in other replicates only the cheater (or neither competitor) was driven to extinction. Such cheater-induced extinctions demonstrate the concept of evolutionary suicide, in which selfish behaviors that improve the fitness of some genotypes in the short run are disastrous for populations or species in the long run (Rankin and Lopez-Sepulcre, 2005). The observed variation in cheater effects at the population level is presumably due to differences in cheating intensity and the frequency thresholds at which the cheaters lose their advantage and cause large drops in spore production. Cheaters with high short-term fitness

33

in nature are expected to hinder the success of others in the groups that they exploit.

Evolutionary Escape from Social Defects Re-Evolution of Development Myxobacteria can rapidly lose their social functions during evolution under favorable growth conditions, suggesting that socially defective strains may be common in nature. This raises the question of whether such social degeneration is an evolutionary dead end or whether accessible mutational pathways exist that can allow an evolutionary escape from a defect under conditions in which restored cooperation would be advantageous. Two recent studies demonstrate that M. xanthus has a remarkable capacity to reevolve social functions from a socially defective genomic background and does so through novel genetic pathways relative to the original genomic state prior to mutational deterioration of social functions. First, the extinction-inducing cheater described above (hereafter abbreviated OC for “obligate cheater”) did not always cause total self-destruction. In one replicate population, OC did drive the population to the brink of extinction, but a spontaneous mutant of OC not only survived the population crash but did so by reevolving developmental proficiency (Fiegna et al., 2006). The newly evolved genotype (PX for “Phoenix”) sporulated even more efficiently than a marked variant of the proficient ancestor of OC (GJV2) in pure culture and showed higher developmental fitness than both GJV2 and OC in mixed populations over a wide range of mixing frequencies. The appearance of PX represents not only an evolutionary escape from a state of obligate social cheating but also the emergence of a competitively superior cooperative strategy and the spontaneous evolution of resistance to cheating. In this transition, obligate cheating served as a stepping-stone to a novel and superior form of cooperation rather than being an evolutionarily terminal fate. Using two sequencing technologies, one new (“sequencing-by-synthesis”) and one traditional ( “Sanger” sequencing), the entire PX genome was sequenced in search of the mutational basis of the evolutionary transitions to OC and PX (Velicer et al., 2006). The resulting data are likely to have revealed all mutational differences between PX, OC, and GJVl. Such comprehensive mutation identification is a crucial beginning to the complex task of understanding the genetic basis of evolutionary change and adaptation in laboratory evolutionary experiments. While 15 mutational differences between PX and GJVl were discovered, only one was unique to PX. The others were all present in OC and

MYXOBACTERIAL BIOLOGY

34 distinguish it from its cooperative ancestor (Fig. 8). Strikingly, PX was shown to have arisen from just this single regulatory mutation. This mutation occurred upstream of an uncharacterized GNAT-acetyltransferase enzyme. It increases the production of this enzyme early in development and causes many other changes in PX gene expression that appear to drive a novel mechanistic route to social development in 111. xanthus (Fiegna et al., 2006; Kadam, 2006). That a social trait can be restored by evolutionary conversion of a cheater into a cooperator expands our view of how social systems can evolve. That a new genetic basis for social cooperation can arise with “builtin” resistance to cheating increases our understanding of how the effects of cheaters on microbial social systems might be limited over evolutionary time scales. That such a dramatic behavioral restoration can be accomplished

D K I622

5 mutations

I

Tn5

? gen.

by a single spontaneous mutation highlights how much we have yet to learn about existing and potential genetic networks that underlie social traits in M. xanthus and other social organisms.

Reevolution of Social Swarming Another study shows that the ability of 211. xanthus to reevolve lost social functions extends to group motility as well as development (Velicer and Yu, 2003). In this case, S-motility was intentionally eliminated by deletion of a large portion of the pilA gene that encodes the pilin protein used to construct type IV pili necessary for S-motility. Multiple populations derived from the ApilA ancestors were allowed to undergo multiple 2-week cycles of growth on a soft agar surface on which S-motility is required to drive effective swarming in the lab strain DK1622. Thus, the ApilA populations initially

\ “‘+ioOs

PX

14 mutations

GJVl

b

GVB207.3

1000 generations

b

growth in rich liquid medium

b

four cycles of alternating starvation and growth

Figure 8 Mutational history of the PX mutant. The previously sequenced strain DK1622 and its derivative clone GJVl are separated by five mutations and an unknown number of generations of lab stock cultivation. The lineage from GJVl to GVB207.3 incurred 1 4 mutations over 1,000 generations of growth in liquid medium (Velicer et al., 1998), one or more of which eliminated the ability to undergo multicellular development. OC was generated by integration of a Tn5 transposon (which confers resistance to kanamycin) into the GVB207.3 genome. OC evolved into PX by regaining the ability to sporulate via social development during an extended developmental competition against a marked variant of GJV1. Only one mutation was found to distinguish PX from OC, and this mutation was subsequently shown to cause the restoration of developmental proficiency in PX. Reprinted with permission from Fiegna et al. (2006).

2. ECOLOGYAND EVOLUTION OF MYXOBACTERIA expanded very little during the early transfer cycles. At the end of each cycle, however, cells from the point on the population perimeter furthest from the initial inoculation point were transferred to a fresh plate, thus favoring any mutants capable of swarming outward more effectively than the ancestral defective population. After extended evolution, six populations showed significant but relatively small increases in soft-agar swarming rates relative to their ApiZA ancestor, whereas two other populations gained much larger increases (Fig. 9). The evolved mutants showing the largest gains in swarming performance did not regain soft-agar swarming by restoring their ancestral capacity for pilin-mediated S-motility. Rather, they compensated by evolving a new mechanism to drive swarming on soft agar that does not involve pili (Velicerand Yu, 2003). The evolved swarmers were shown to lack pilin just like their ancestors, but they had reevolved the ability to produce cohesive extracellular fibril material that had been eliminated by the ApilA deletion. This fibril material and the intact A-motility system were both necessary for the reevolved swarming phenotypes. Thus, soft-agar swarming was restored by

35

the evolution of a novel (and as yet undefined) relationship between the M. xanthus extracellular matrix and the A-motility system. Moreover, it was shown that social interactions between cells that are mediated by fibrils, and not mere fibril production per se, were involved in generating the evolved swarming phenotypes. Therefore, M . xanthus reevolved a socially mediated behavior (swarming on soft agar) but found a novel molecular mechanism for how to drive this behavior.

Social Divergence in Natural Populations A fundamental question in myxobacterial sociobiology is the following: Who cooperates with whom? Selfpropelled motility and external migration vectors should frequently lead to encounters between distinct species and strains of myxobacteria. In such cases of external encounter of different colony types, how distinct can two genotypes be at various genetic loci and still swarm, aggregate, or sporulate together? Suboptimal cooperators (or even outright antagonists) might also arise within groups, since every bacterial colony above a certain population size contains some degree of genetic

Figure 9 Swarming phenotypes of ancestral ApilA genotypes (third and fifth positions clockwise from top) and their evolved descendants (second and fourth positions clockwise from top, respectively) relative to DK1622 (top position) on soft agar. The DK1622 swarm was 3 days old, and the ApilA, and evolved strain swarms were 7 days old. Reprinted with permission from Velicer and Yu (2003).

MYXOBACTERIAL BIOLOGY

36 diversity. For example, the normal spontaneous mutation rate of M. xanthus may be similar to that of E. coZi (-1 to 5 x 10-lo per base pair per generation [Lenski et al., 2003; Velicer et al., 2006]), such that one in every several hundred cell divisions may result in a new genetic variant within colonies of M. xanthus. Because single spontaneous mutations can dramatically affect cell surface molecules, intercellular signal production, or cell behavior (Fiegna et al., 2006; Velicer et al., 2006), the appearance of genetically distinct neighbors by simple mutational steps that reduce compatibility to the majority genotype may occur frequently. In the first study to examine developmental compatibility across myxobacterial genotypes, Smith and Dworkin (1994) mixed two strains classified as Myxococcus virescens and M. xanthus at the onset of starvation. These two genotypes appear to have fully segregated into species-specific fruiting bodies, and the presence of the M. virescens strain strongly inhibited spore production by the M . xanthus isolate. Subsequently, Fiegna and Velicer (2005) quantified developmental compatibility across all possible pairwise combinations of nine M. xanthus strains isolated from distant global locations. Most pairings resulted in greatly reduced sporulation for at least one competitor and reduced total spore production (relative to expectations from pure-culture controls). Strain fitness relationships were predominately hierarchical rather than circular (i.e., nontransitive), with three strains dominating over six inferior strains, which in turn exhibited a linear rank hierarchy. Such competitive asymmetries are likely to be due to differences in one or more components of cell surface or diffusible “secretomes” across competitors. In several instances, superior competitors actually sporulated more efficiently in the presence of their inferior partner than they did as clonal cultures, thus demonstrating the possibility of social exploitation among socially proficient myxobacteria in the wild (Fig. 10). That isolates from across the globe have inadvertently evolved social incompatibility in isolation from one another is perhaps not surprising. But what happens when closely related neighbors encounter one another? In a study parallel in design to that of Fiegna and Velicer (2005), pervasive developmental antagonism has been observed among nine isolates from the 16- by 16-cm Tubingen sample plot (see above), including between isolates that have identical csgA, fibA, and pilA alleles (Vos, 2006). Moreover, a swarm compatibility test was conducted for all possible pairs of distinct isolates from the plot that share an identical csgA-fi6A-piZA concatemer sequence. In most cases, the paired swarms (inoculated near one another on an agar plate) failed

D

”j -7

I

I

G

1

Figure 10. Facultative and antagonistic exploitation by two natural isolates of M . xanthus (D against I and I against G) during mixed development. The log-scale effect of mixing strains i and j on the sporulation efficiency of strain i is termed Ci(j). Open bars show the effect of mixing on sporulation efficiency for the dominant, exploitative competitor in each pair (D and I, respectively) in response to its inferior competitor. Shaded bars indicate the effect of mixing on the inferior strain (I and G, respectively). Error bars indicate 95 % confidence intervals. Reprinted with permission from Fiegna and Velicer (2005).

to merge, whereas control pairs of two swarms of the same isolate always merged (Vos, 2006). These results indicate that divergence into incompatible social genotypes occurs at a very fine scale of genomic divergence that is not reflected by variation in the highly variable piZA gene. Extrapolating these results across whatever portion of the -150 million km2 of terrestrial earth surface that M. xanthus has colonized (which is certainly a substantial fraction), it appears that M. xanthus has diverged into many trillions of distinct social genotypes that are in most cases socially incompatible with one another to some degree. Distinct species of myxobacteria can often be isolated from the same soil sample (Dawid, 2000), prompting the question of how such coexisting species might interact. N. Dahanukar performed a study of swarm interactions between seven distinct myxobacterial isolates that appeared to represent both classified species ( M . xanthus, M . fulvus, and StigmateZla aurantiaca) and previously unclassified species. The results suggested that some species strongly dominate over others in pairwise swarm encounters on agar plates, even to the point of invading competitor territory and decimating competitor populations (Dahanukar, 2003). Intriguingly, the average degree of dominance over competitors during swarm mixes appeared to positively correlate with

2.

ECOLOGY AND EVOLUTION OF MYXOBACTERIA

the stalk length that each isolate exhibited during development. Further research is needed to determine whether there are general patterns of dominance across species, or whether intraspecific social diversification is so great that competitive dominance between any two random isolates cannot be predicted by species classification.

CONCLUSION Laboratory-based approaches to the ecology and evolution of myxobacteria have begun to open up a new range of largely unexplored questions in this field. Moreover, we stand on the verge of great increases in our power to ask fundamental questions about the distribution, ecology, and evolution of myxobacteria in their native habitats by using current and future molecular tools. These research trends promise to tell us much more not only about the myxobacteria themselves but also about broader biological issues such as the evolution of group living, cooperation, conflict, and multicellularity.

References Amann, R. I., W. Ludwig, and K. H. Schleifer. 1995. Phylogenetic identification and in-situ detection of individual microbial cells without cultivation. Microbiol. Rev. 59:143-169. Axelrod, R., and W. D. Hamilton. 1981. The evolution of cooperation. Science 211:1390-1396. Baas Becking, L. G. M. 1934. Geobiologie: of inleiding tot de milieukunde. W. P. Van Stockum & Zoon N. V., The Hague, The Netherlands. Beebe, J. M. 1941. Studies on the Myxobacteria. Iowa St. Coll. J. Sci. 15:307-337. Bohannan, B. J. M., and R. E. Lenski. 2000. Linking genetic change to community evolution: insights from studies of bacteria and bacteriophage. Ecol. Lett. 3:362-377. Bourke, A. F. G., and N. R. Franks. 1995. Social Evolution in Ants. Princeton University Press, Princeton, NJ. Branda, S. S., J. E. Gonzalez-Pastor, S. Ben-Yehuda, R. Losick, and R. Kolter. 2001. Fruiting body formation by Bacillus subtilis. Proc. Natl. Acad. Sci. USA 98:11621-11626. Bull, C. T., K. G. Shetty, and K. V. Subbarao. 2002. Interactions between Myxobacteria, plant pathogenic fungi, and biocontrol agents. Plant Dis. 869894396. Cooper, V. S., D. Schneider, M. Blot, and R. E. Lenski. 2001. Mechanisms causing rapid and parallel losses of ribose catabolism in evolving populations of Escherichia coli. B. J. Bucteriol. 183:2834-2841. Crespi, B. J. 2001. The evolution of social behavior in microorganisms. Trends Ecol. Evol. 16:178-183. Daft, M. J., J. C. Burnham, and Y. Yamamoto. 1985. Lysis of Phormidium luridum by Myxococcus fulvus in continuous flow cultures. J. Appl. Bacteriol. 59:73-80. Dahanukar, N. 2003. Species, morphological and behavioral diversity in myxobacteria. Abasaheb Garware College, Pune, India.

37

Dawid, W. 2000. Biology and global distribution of Myxobacteria in soils. FEMS Microbiol. Rev. 24:403-427. Dawid, W., C. A. Gallikowski, and P. Hirsch. 1988. Psychrophilic myxobacteria from antarctic soils. Polarforschung 58~217-278. Elena, S. F. 2002. Restrictions to RNA virus adaptation: an experimental approach. Antonie Leeuwenhoek 81:135-142. Elena, S. F., and R. E. Lenski. 2003. Evolution experiments with microorganisms: the dynamics and genetic bases of adaptation. Nut. Rev. Genet. 4:457-469. Endler, J. A. 1986. Natural Selection in the Wild. Princeton University Press, Princeton, NJ. Feldgarden, M., D. M. Stoebel, D. Brisson, and D. E. Dykhuizen. 2003. Size doesn’t matter: microbial selection experiments address ecological phenomena. Ecology 84:1679-1687. Fenchel, T. 2003. Biogeography for bacteria. Science 301:925926. Fiegna, F., and G. J. Velicer. 2003. Competitive fates of bacterial social parasites: persistence and self-induced extinction of Myxococcus xanthus cheaters. Proc. R. SOC. London B 270~1527-1534. Fiegna, F., and G. J. Velicer. 2005. Exploitative and hierarchical antagonism in a cooperative bacterium. PLoS Biol. 3 :1980-1 987. Fiegna, F., Y. T. N. Yu, S. V. Kadam, and G. J. Velicer. 2006. Evolution of an obligate social cheater to a superior cooperator. Nature 441:310-314. Fontes, M., and D. Kaiser. 1999. Myxococcus cells respond to elastic forces in their substrate. Proc. Nut. Acad. Sci. USA 96:8052-8057. Gerth, K., and R. Muller. 2005. Moderately thermophilic Myxobacteria: novel potential for the production of natural products isolation and characterization. Environ. Microbiol. 72374-8 80. Gevers, D., F. M. Cohan, J. G. Lawrence, B. G. Spratt, T. Coenye, E. J. Fed, E. Stackebrandt, Y. Van de Peer, P. Vandamme, F. L. Thompson, and J. Swings. 2005. Re-evaluating prokaryotic species. Nut. Rev. Microbiol. 3:733-739. Goldman, B., W. Nierman, D. Kaiser, S. Slater, A. Durkin, J. Eisen, C. Ronning, W. Barbazuk, M. Blanchard, C. Field, C. Halling, G. Hinkle, 0. Lartchuk, H. Kim, C. Mackenzie, R. Madupu, N. Miller, A. Shvartsbeyn, S. Sullivan, M. Vaudin, R. Wiegand, and H. Kaplan. 2006. Evolution of sensory complexity recorded in a myxobacterial genome. Proc. Nutl. Acad. Sci. USA 103:15200-15205. Grant, P. R., and B. R. Grant. 2002. Unpredictable evolution in a 30-year study of Darwin’s finches. Science 296:707-711. Greig, D., and M. Travisano. 2004. The Prisoner’s Dilemma and polymorphism in yeast SUC genes. Proc. R. Soc. London 271:S25-S26. Griffin, A. S., S. A. West, and A. Buckling. 2004. Cooperation and competition in pathogenic bacteria. Nature 430:10241027. Hanski, I., and I. Saccheri. 25 April 2006. Molecular-level variation affects population growth in a butterfly metapopulation. PLoS Biol. 4:e129. [Epub ahead of print.] He, Q., and R. A. Sanford. 2003. Characterization of Fe(II1) reduction by chlororespiring Anaeromxyobacter dehalogenuns. Appl. Environ. Microbiol. 69:2712-2718.

38 Hillesland, K. L. 2005. Evolutionary ecology of predation by the soil bacterium, Myxococcus xanthus. Ph.D. thesis. Michigan State University, East Lansing. Hillesland, K. L., R. E. Lenski, and G. J. Velicer. 2007. Ecological variables affecting predatory success in Myxococcus xanthus. Microb. Ecol. 53571-578. Hillesland, K. L., and G. J. Velicer. 2005. Resource level affects relative performance of the two motility systems of Myxococcus xanthus. Microb. Ecol. 49558-566. Hodgkin, J., and D. Kaiser. 1979a. Genetics of gliding motility in Myxococcus xanthus (Myxobacterales): two gene systems control movement. Mol. Gen. Genet. 171:177-191. Hodgkin, J., and D. Kaiser. 1979b. Genetics of gliding motility in Myxococcus xanthus (Myxobacterales):genes controlling movement of single cells. Mol. Gen. Genet. 171:167-176. Holling, C. S. 1959. Some characteristics of simple types of predation and parasitism. Can. Entomol. 91:385-398. Iizuka, T., Y. Jojima, R. Fudou, M. Tokura, A. Hiraishi, and S. Yamanaka. 2003. Enhygromyxa salina gen. nov., sp nov., a slightly halophilic myxobacterium isolated from the coastal areas of Japan. Syst. Appl. Microbiol. 26:189-196. Iizuka, T., Y. Jojima, R. Fudou, and S. Yamanaka. 1998. Isolation of myxobacteria from the marine environment. FEMS Microbiol. Lett. 169:317-322. Jessup, C. M., R. Kassen, S. E. Forde, B. Kerr, A. Buckling, P. B. Rainey, and B. J. M. Bohannan. 2004. Big questions, small worlds: microbial model systems in ecology. Trends Ecol. Evol. 19~189-197. Jiang, D.-M., Z.-H. Wu, J.-Y. Zhao, and Y.-Z. Li. 2007. Fruiting and non-fruiting myxobacteria: a phylogenetic perspective of cultured and uncultured members of this group. Mol. Phylogenet. Evol. 44545-552. Kadam, S. 2006. Ecology and evolution of multicellular development in the social bacterium Myxococcus xanthus. Doctoral thesis, Universitat Tubingen, Tubingen, Germany. Kadam, S. V., and G. J. Velicer. 2006. Variable patterns of density-dependent survival in social bacteria. Behav. Ecol. 17:833-838. Kaiser, D. 2001. Building a multicellular organism. Annu. Rev. Genet. 35:103-123. Kassen, R., and P. B. Rainey. 2004. The ecology and genetics of microbial diversity. Annu. Rev. Microbiol. 58:207-231. Kearns, D.B., and L. J. Shimkets. 2001. Lipid chemotaxis and signal transduction in Myxococcus xanthus. Trends Microbiol. 9:126-129. Keller, L., and M. G. Surette. 2006. Communication in bacteria: an ecological and evolutionary perspective. Nut. Rev. Microbiol. 4:249-258. Lenski, R. E., M. R. Rose, S. C. Simpson, and S. C. Tadler. 1991. Long-term experimental evolution in Escherichia coli. I. Adaptation and divergence during 2,000 generations. Am. Nut. 138~1315-1341. Lenski, R. E., C. L. Winkworth, and M. A. Riley. 2003. Rates of DNA sequence evolution in experimental populations of Escherichia coli during 20,000 generations. J. Mol. Evol. 56~498-508. Linz, B., F. Balloux, Y. Moodley, A. Manica, H. Liu, P. Roumagnac, D. Falush, C. Stamer, F. Prugnolle, S. W. van der Menve, Y. Yamaoka, D. Y. Graham, E. Perez-Trallero, T.

MYXOBACTERIAL BIOLOGY Wadstrom, S. Suerbaum, and M. Achtman. 2007. An African origin for the intimate association between humans and Helicobacter pylori. Nature 445:915-918. Lorenzi, M. C., R. Cervo, F. Zacchi, S. Turillazzi, and A. G. Bagneres. 2004. Dynamics of chemical mimicry in the social parasite wasp Polistes semenowi (Hymenoptera: Vespidae). Parasitology 129:643-651. Lubchenco, J. 1978. Plant species diversity in a marine intertidal community: importance of herbivore food preference and algal competitive abilities. Am. Nut. 112:23-39. MacLean, R. C.,and I. Gudelj. 2006. Resource competition in bacteria: an ecological and evolutionary perspective. Nut. Rev. Microbiol. 4:249-258. Margulies, M., M. Egholm, W. E. Altman, S. Attiya, J. S. Bader, L. A. Bemben, J. Berka, M. S. Braverman, Y. J. Chen, Z. Chen, S. B. Dewell, L. Du, J. M. Fierro, X. V. Gornes, B. C. Godwin, W. He, S. Helgesen, C. H. Ho, G. P. Irzyk, S. C. Jando, M. L. Alenquer, T. P. Jarvie, K. B. Jirage, J. B. Kim, J. R. Knight, J. R. Lanza, J. H. Leamon, S. M. Lefkowitz, M. Lei, J. Li, K. L. Lohman, H. Lu, V. B. Makhijani, K. E. McDade, M. P. McKenna, E. W. Myers, E. Nickerson, J. R. Nobile, R. Plant, B. P. PUC,M. T. Ronan, G. T. Roth, G. J. Sarkis, J. F. Simons, J. W. Simpson, M. Srinivasan, K. R. Tartaro, A. Tomasz, K. A. Vogt, G. A. Volkmer, S. H. Wang, Y. Wang, M. P. Weiner, P. Yu, R. F. Begley, and J. M. Rothberg. 2005. Genome sequencing in microfabricated high-density picolitre reactors. Nature 437:376-380. Martiny, J. B., B. J. Bohannan, J. H. Brown, R. K. Colwell, J. A. Fuhrman, J. L. Green, M. C. Homer-Devine, M. Kane, J. A. Krumins, C. R. Kuske, P. J. Morin, S. Naeem, L. Ovreas, A. L. Reysenbach, V. H. Smith, and J. T. Staley. 2006. Microbial biogeography: putting microorganisms on the map. Nut. Rev. Microbiol. 4:102-112. McBride, M. J., and D. R. Zusman. 1996. Behavioral analysis of single cells of Myxococcus xanthus in response to prey cells of Escherichia coli. FEMS Microbiol. Lett. 137:227231. Neil, R. B., D. Hite, M. I. Kelrick, M. L. Lockhart, and K. Lee. 2005. Myxobacterial biodiversity in an established oak-hickory forest and a savanna restoration site. Curr. Microbiol. 50:88-95. Olsen, G. J., D. J. Lane, S. J. Giovannoni, N. R. Pace, and D. A. Stahl. 1986. Microbial ecology and evolution: a ribosomal RNA approach. Annu. Rev. Microbiol. 40:337-365. Parsek, M. R., and E. P. Greenberg. 2005. Sociomicrobiology: the connections between quorum sensing and biofilms. Trends Microbiol. 13:27-33. Petrie, L., N.N. North, S. L. Dollhopf, D. L. Balkwill, and J. E. Kostka. 2003. Enumeration and characterization of iron(II1)reducing microbial communities from acidic subsurface sediments contaminated with uranium(V1). Appl. Environ. Microbiol. 69:7467-7479. Pham, V. D., C. W. Shebelut, M. E. Diodati, C. T. Bull, and M. Singer. 2005. Mutations affecting predation ability of the soil bacterium Myxococcus xanthus. Microbiology 1 51:1865-1 874. Porta, H., and M. Rocha-Sosa. 2001. Lipoxygenase in bacteria: a horizontal transfer event? Microbiology 147:3199-3200. Quillet, L., S. Barray, B. Labedan, F. Petit, and J. GuespinMichel. 1995. The gene encoding the beta-1,4-endoglucanase

AND EVOLUTION OF MYXOBACTERIA 2. ECOLOGY

(CelA) from Myxococcus xanthus: evidence for independent acquisition by horizontal transfer of binding and catalytic domains from actinomycetes. Gene 158:23-29. Rainey, P. B., and K. Rainey. 2003. Evolution of cooperation and conflict in experimental bacterial populations. Nature 425 ~72-74. Rankin, D. J., and A. Lopez-Sepulcre. 2005. Can adaptation lead to extinction? Oikos 111:616-619. Redfield, R. J. 2002. Is quorum sensing a side effect of diffusion sensing? Trends Microbiol. 10:365-370. Reichenbach, H. 1993. Biology of the Myxobacteria: ecology and taxonomy, p. 13-62. In M. Dworkin and D. Kaiser (ed.), Myxobacteria II. American Society for Microbiology, Washington, DC. Reichenbach, H. 1999. The ecology of the Myxobacteria. Environ. Microbiol. 1:15-21. Reichenbach, H., and G. Hofle. 1993. Biologically active secondary metabolites from Myxobacteria. Biotechnol. Adv. 11~219-277. Rose, M. R. 1984. Laboratory evolution of postponed senescence in Drosophila melanogaster. Evolution 38:10041010. Rosenberg, E., K. H. Keller, and M. Dworkin. 1977. Cell density-dependent growth of Myxococcus xanthus on casein. J. Bacteriol. 129:770-777. Rosenberg, E., and M. Varon. 1984. Antibiotics and lytic enzymes, p. 109-125. In E. Rosenberg (ed.),Myxobacteria: Development and Cell Interactions. Springer-Verlag, New York, NY. Sachs, J. L., U. G. Mueller, T. P. Wilcox, and J. J. Bull. 2004. The evolution of cooperation. Q. Rev. Biol. 79:135-160. Sanford, R. A., J. R. Cole, and J. M. Tiedje. 2002. Characterization and description of Anaeromyxobacter dehalogenans gen. nov., sp nov., an aryl-halorespiring facultative anaerobic myxobacterium. Appl. Environ. Microbiol. 689393900. Shi, W., and D. R. Zusman. 1993. The two motility systems of Myxococcus xanthus show different selective advantages on various surfaces. Proc. Natl. Acad. Sci. USA 90:33783382. Shimkets, L. J. 1999. Intercellular signaling during fruiting-body development of Myxococcus xanthus. Annu. Rev. Microbiol. 53:525-549. Shimkets, L. J. 1990. Social and developmental biology of the myxobacteria. Microbiol. Rev. 54:473-501. Shimkets, L. J., H. Reichenbach, and M. Dworkin. 2005. The Myxobacteria. In M. Dworkin (ed.), The Prokaryotes, 3rd ed., vol. 7, p. 31-115. Springer, New York, NY. Smith, D. R., and M. Dworkin. 1994. Territorial interactions between two Myxococcus species. J. Bacteriol. 176:12011205. Smith, J. M., and E. Szathmary. 1995. The Major Transitions in Evolution. W. H. Freeman Spektrum, New York, NY. Spiller, D. A,, and T. W. Schoener. 1998. Lizards reduce spider species richness by excluding rare species. Ecology 79503516. Sproer, C., H. Reichenbach, and E. Stackebrandt. 1999. The correlation between morphological and phylogenetic

39

classification of myxobacteria. Int. J. Syst. Bacteriol. 49~1255-1262. Strassmann, J. E., Y. Zhu, and D. C. Queller. 2000. Altruism and social cheating in the social amoeba Dictyostelium discoideum. Nature 408:965-967. Thaxter, R. 1892. On the Myxobacteriaceae, a new order of Schizomycetes. Bot. Gaz. 17:389-406. Travisano, M., J. A. Mongold, A. F. Bennett, and R. E. Lenski. 1995. Experimental tests of the roles of adaptation, chance, and history in evolution. Science 26787-90. Travisano, M., and G. J. Velicer. 2004. Strategies of microbial cheater control. Trends Microbiol. 12:72-78. Turner, P. E., and L. Chao. 1999. Prisoner’s dilemma in an RNA virus. Nature 398:441-443. Velicer, G. J. 2003. Social strife in the microbial world. Trends Microbiol. 11:330-337. Velicer, G. J., L. Kroos, and R. E. Lenski. 2000. Developmental cheating in the social bacterium Myxococcus xanthus. Nature 404598-601. Velicer, G. J., L. Kroos, and R. E. Lenski. 1998. Loss of social behaviors by Myxococcus xanthus during evolution in an unstructured habitat. Proc. Natl. Acad. Sci. USA 95:1237612380. Velicer, G. J., R. E. Lenski, and L. Kroos. 2002. Rescue of social motility lost during evolution of Myxococcus xanthus in an asocial environment. J. Bacteriol. 184:2719-2727. Velicer, G. J., G. Raddatz, H. Keller, S. Deiss, C. Lanz, I. Dinkelacker, and s. C. Schuster. 2006. Comprehensive mutation identificationin an evolved bacterial cooperator and its cheating ancestor. Proc. Natl. Acad. Sci. USA 103:8107-8112. Velicer, G. J., and K. L. Stredwick. 2002. Experimental social evolution with Myxococcus xanthus. Antonie Leeuwenhoek 81:155-164. Velicer, G. J., and Y. N. Yu. 2003. Evolution of novel cooperative swarming in the bacterium Myxococcus xanthus. Nature 42575-78. Vos, M. 2006. Natural variation in the social bacterium Myxococcus xanthus. Doctoral thesis. Universitat Tiibingen, Tiibingen, Germany. Vos, M., and G. J. Velicer. 2006. Genetic population structure of the soil bacterium Myxococcus xanthus at the centimeter scale. Appl. Environ. Microbiol. 72:3615-3625. Vulic, M., and R. Kolter. 2001. Evolutionary cheating in Escherichia coli stationary phase cultures. Genetics 158:519526. Watve, M. G., A. M. Shete, N. Jadhav, S. A. Wagh, S. P. Shelar, S. S. Chakraborti, A. P. Botre, and A. A. Kulkarni. 1999. Myxobacterial diversity in Indian soils: how many species do we have? Curr. Sci. 77:1089-1095. West, S. A., A. S. Griffin, A. Gardner, and S. P. Diggle. 2006. Social evolution theory for microbes. Nat. Rev. Microbiol. 4597-607. Whitaker, R. J., D. W. Grogan, J. W. Taylor. 2003. Geographic barriers isolate endemic populations of hyperthermophilic Archaea. Science 301:976-978. White, D., W. Shropshire, and K. Stephens. 1980. Photocontrol of development by Stigmatella aurantiaca. J. Bacteriol. 142:1023-1024.

40 Wireman, J. W., and M. Dworkin. 1977. Developmentally induced autolysis during fruiting body formation by Myxococcus xanthus. J. Bacteriol. 129:796-802. Wu, S. S., J. Wu, Y. L. Cheng, and D. Kaiser. 1998. The pilH gene encodes an ABC transporter homologue required for type IV pilus biogenesis and social gliding motility in Myxococcus xanthus. Mol. Microbiol. 29:1249-1261. Wu, Z. H., D. M. Jiang, P. Li, and Y. Z. Li. 2005. Exploring the diversity of myxobacteria in a soil niche by myxobacteriaspecific primers and probes. Environ. Microbiol. 71602-1610.

MYXOBACTERIAL BIOLOGY Xu, J. P. 2004. The prevalence and evolution of sex in microorganisms. Genome 47:775-780. Zahavi, A., and D. Ralt. 1984. Social adaptations in myxobacteria, p. 215-220. In E. Rosenberg (ed.), Myxobacteria: Development and Cell Interactions. Springer-Verlag, New York, NY. Zhang, Y. Q., Y. Z. Li, B. Wang, Z . H. Wu, C. Y. Zhang, X. Gong, Z. J. Qiu, and Y. Zhang. 2005. Characteristics and living patterns of marine myxobacterial isolates. Appl. Enviyon. Micro biol. 71:33 3 1-3 336.

DeueloDment

Myxobacteria: Multicellulurity and Differentiation Edited by David E. Whitworth 02008 ASM Press, Washington, D.C.

Michelle E. Diodati, Ronald E. Gill, Lynda Plamann, Mitchell Singer

Initiation and Early Developmental Events

Myxococcus xanthus is a rod-shaped, gram-negative soil bacterium that, when subjected to nutrient deprivation, undergoes a developmental process culminating in the formation of a multicellular fruiting body filled with spores. These predatory bacteria utilize proteins, peptides, and amino acids as their primary source of carbon, nitrogen, and energy and acquire these essential nutrients by cooperatively pooling extracellular hydrolytic enzymes to degrade other bacteria in soil (Bretscher and Kaiser, 1978; Dworkin, 1962). The colonial association of M . xanthus provides a competitive advantage over single, dispersed cells, and upon nutrient depletion, the formation of fruiting bodies cluster spores together, ensuring that new microcolonies arise after germination of myxospores under favorable environmental conditions. Elucidating the relationship between nutrient limitation and the mechanisms that cells employ to recognize and respond to it is fundamental for understanding how these organisms adapt to their environment. In order for M. xanthus cells to initiate development, three criteria must be met: (i) cells must be on a solid surface to allow gliding motility to occur (Kroos et al.,

3

1988; Wireman and Dworkin, 1975); (ii) cells must be at an appropriate density (Shimkets and Dworkin, 1981; Wireman and Dworkin, 1975);and finally, (iii)cells must be able to perceive a nutrient downshift, such that some energy capacity for protein synthesis remains (Dworkin, 1963; Hemphill and Zahler, 1968). The process of M . xanthus development requires approximately la5cells, so it is necessary that individual cells monitor not only their own nutritional status at the cellular level but also the nutritional status at the population level. Once the developmental program is initiated by starvation, it proceeds in a predictable manner with aggregation, mound formation, fruiting body formation, and sporulation within the fruiting bodies in a 24- to 48-h timeline (depicted in Fig. 1).For the purposes of this chapter, early development can be defined as events occurring from initiation to the start of aggregation at approximately the first 6 h poststarvation. In 1995, Singer and Kaiser proposed a dual model of starvation recognition for developing 111. xanthus cells (Singer and Kaiser, 1995). First, individual cells need to recognize starvation at the cellular level, and second, cells

~

Michelle E. Diodati and Mitchell Singer, Section of Microbiology, University of California-Davis, Davis, CA 95616. Ronald E. Gill, Department of Microbiology, University of Colorado Health Sciences Center, Denver, CO 80262. Lynda Plamann, School of Biological Sciences, Cell Biology and Biophysics, University of Missouri-Kansas City, Kansas City, MO 641 10.

43

DEVELOPMENT AND MOTILITY

44 Starvation

I

0 hours

Aggregation

1

6 hour$

FB and spore maturation

Mounds

t

I

i

"

12 hours

24 hours

TPM

Figure 1 Pictorial and photographic representations of the developmental process in M. xunthus DK1622. The diagram shows approximate times for each step in the process: starvation (0 h), aggregation (6 to 8 h), mound formation (12 h), fruiting body formation and sporulation (24 to 48 h). The first row represents development in an MC7 submerged culture system (ICuner and Kaiser, 1982), and the second row represents development on TPM starvation agar plates at a magnification of X40. This figure is adapted from Tzeng et al., 2006.

need to know that the population as a whole is starving. These two pieces of information, cellular starvation and population starvation, are perceived and integrated by these individual cells to activate the developmental program. In retrospect, this model as first described is relatively simplistic, yet the three basic tenets remain the same. Over the last 10 years, new experimental evidence pertaining to early development and sequence information obtained from the M. xanthus genome has increased our understanding of how M. xanthus cells recognize and respond to nutrient limitation. In this chapter, the following tenets are addressed: how individual M. xanthus cells recognize starvation; how these cells perceive population starvation; and how individual cells integrate this information to ultimately initiate fruiting body formation and cellular differentiation.

INITIATION OF DEVELOPMENT: CELLULAR STARVATION RECOGNITION M . xanthus is a representative of the proteolytic myxobacteria, relying primarily on proteins, peptides, fatty acids, and tricarboxylic acid cycle intermediates for their carbon, nitrogen, and energy needs (Bretscher and Kaiser, 1978; Dworkin, 1962). Like most developing microbes, the trigger for the developmental process in M. xanthus is nutrient deprivation, in conjunction with high cell density and a solid surface, as described above. Understanding how cells recognize a nutritional downshift

at the molecular level is critical for discerning the transition from vegetative growth to development.

Conditions That Induce Development and Other Starvation Conditions There are various types of starvation, not all of which induce fruiting body formation or differentiation in M. xanthus. These are listed in Table 1.Starvation for amino acids, carbon (such as pyruvate), or phosphate induces the developmental response (Dworkin, 1996; Manoil

Table 1 Conditions known to initiate development and

induce a stringent response" Condition Carbon starvation Amino acid starvation Essential amino acids Nonessential amino acids (auxotrophs) Amino acid analogues Phosphate starvation Purine starvation Decoyinine

Aggregation Sporulation (p)ppGpp

+

+

t

+ + + +

+ +

t 1'

+ +

t t

-

-

NC

Mycophenolic acid

-

-

NC

Pyrimidine starvation

-

-

ND

aAbbreviations and symbols: +, induces the condition; -, does not induce the condition; t,increases the intracellular concentration of (p)ppGpp; NC, no change in (p)ppGpplevels; ND, not determined.

EVENTS 3. INITIATION AND EARLYDEVELOPMENTAL

45 initiate the sporulation process (Ochi et al., 1982).In M . xanthus, there is a small decrease in GTP levels during the stringent response, but this decrease alone is unable to initiate the developmental program based on studies with inhibitors of in situ GTP synthesis decoyinine and mycophenolic acid (Singer and Kaiser, 1995).These studies led to the hypothesis that M. xanthus uses (p)ppGpp as an internal starvation signal, and the level of this molecule either by itself or in concert with other factors leads to the initiation of fruiting body formation.

and Kaiser, 1980b; Shimkets, 1984, 1987). However, unlike what is observed for Bacillus subtilis, starvation for guanine nucleotides does not instigate the developmental process (Singer and Kaiser, 1995); and while the addition of excess purines can lead to development (Campos and Zusman, 1975), this has been shown to be an artifact that indirectly causes a nutritional imbalance (Manoil and Kaiser, 1 9 8 0 ~ )Furthermore, . Kimsey and Kaiser have shown that pyrimidine starvation hinders growth but does not trigger fruiting body development (Kimsey and Kaiser, 1991). A common feature of all conditions that induce development is that they also induce a stringent response in M . xanthus (Manoil and Kaiser, 1980b).It has been previously demonstrated that one of the earliest responses in M. xanthus development is the increase in the intracellular concentration of (p)ppGpp (Manoil and Kaiser, 1980a, 1980b; Singer and Kaiser, 1995; for reviews see Cashel et al., 1996, and Chatterji and Ojha, 2001). This alarmone accumulates during amino acid starvation and acts as an intracellular starvation signal that is both necessary and sufficient for the initiation of early developmental gene expression in M . xanthus (Singer and Kaiser, 1995).This is in contrast to B. subtilis sporulation, where a dramatic decrease in the GTP pool has been shown to

The Stringent Response: a General Model Coordinating Metabolism to Global Gene Expression The stringent response was first observed in Escherichia coli more than 50 years ago (Cashel and Gallant, 1969; Pardee and Prestidge, 1956; Sands and Roberts, 1952; Stent and Brenner, 1961). Since then, components of the system have been found in virtually every bacterium examined to date (Chatterji and Ojha, 2001; Ogata et al., 2001; Sun et al., 2001; M. E. Diodati and M. Singer, personal communication), and the E. coli stringent response remains the paradigm. A model of the E. coli stringent response is shown in Fig. 2. Interestingly, an analogous system with RelNSpoT homologues has also

p+ i T p p k

= ADP ATP

PPPG (GTP)

Ndk

1;:- I

PPG(GW

DksA+

1~ p ' " + ~ p i

~

SpoT PPPGPP

~

-

7

0

Positively controlled genes

PPGPP

PPi

Negatively controlled genes

Figure 2 Diagram of the E. coli stringent response. The enzymes involved in the (p)ppGpp metabolism are shown in bold. Ribosome-associated RelA or SpoT catalyzes the synthesis of pppGpp from ATP and GTP upon amino acid or carbon starvation, respectively. Gpp (or Ppx) dephosphorylates pppGpp to make ppGpp. (p)ppGpp accumulates in the cell and interacts with RNAP and DksA [which has been shown to play an important role in (p)ppGppdependent transcriptional regulation (Paul et al., 2004, 2005; Perederina et al., 2004)] to positively and negatively control transcription to respond to starvation. (p)ppGpp levels are modulated by SpoT, and when nutrient conditions change, SpoT hydrolyzes ppGpp to GDP (ppG). Ppk is involved in ATP synthesis, and Ndk forms GTP from ATP and GDP (ppG).For reviews, consult Cashel et al., 1996, and Chatterji and Ojha, 2001.

46

been found in plants (van der Biezen et al., 2000; Givens et al., 2004). Comparisons of the Stringent Response in Bacteria The stringent response links amino acid availability to the rate of protein synthesis through the signaling mol[(p)ppGpp]. ecules guanosine-Sr-(tri)di-3’-diphosphate The response directly inhibits stable RNA synthesis and protein elongation and activates the transcription of amino acid biosynthetic operons. There is a plethora of secondary or indirect effects which include the inhibition of ribosomal protein, phospholipid, and cell wall constituents synthesis, inhibition of DNA replication, and an increase in the production of stress proteins (for reviews see Cashel et al., 1996, and Chatterji and Ojha, 2001). Therefore, the stringent response allows cells to rapidly respond to nutrient limitation by modulation of metabolic pathways, rRNA synthesis, and stressadaptive genes, thus allowing cells to respond and adapt to starvation. In E. coli, two related proteins, RelA and SpoT, modulate the intracellular levels of (p)ppGpp either by the transfer of pyrophosphate from ATP to the 3’ hydroxyl of GTP to form pppGpp or by hydrolyzing ppGpp to pyrophosphate and GDP, respectively (Cashel and Gallant, 1969; Fiil et al., 1977; Laffler and Gallant, 1974). RelA is a ribosome-associated protein required for the synthesis of (p)ppGpp in response to stalled ribosomes due to a decrease in charged tRNAs binding to the acceptor site during amino acid starvation (Block and Haseltine, 1974; Cochran and Byrne, 1974; Haseltine and Block, 1973; Pedersen et al., 1973) (Fig. 2). Thus, (p)ppGpp acts as a monitor of amino acid availability for translation. In contrast, SpoT does not appear to be associated with the ribosomes, and it is a bifunctional enzyme that has both biosynthetic and degradative properties. The activity of SpoT, whether it is biosynthetic or hydrolytic, is influenced by a variety of environmental signals, although it is unclear how SpoT’s activity is regulated at a molecular level (Murray and Bremer, 1996). As mentioned above, E. coli has two distinct and separable enzymes for (p)ppGpp metabolism, RelA and SpoT (Cashel and Gallant, 1969; Laffler and Gallant, 1974). Although many enterics and other species have homologues of these two proteins, recent studies have demonstrated that there is divergence in the strict conservation of these homologues. Many bacteria, both gram negative and gram positive, have only a single bifunctional ribosome-associated protein that is both SpoT- and RelA-like, including B. subtilis (Wendrich and Marahiel, 1997), Sinorhizobium meliloti (Wells and

DEVELOPMENT AND MOTILITY Long, 2002), and Rhodobacter capsulatus (Masuda and Bauer, 2004). Sequence analysis of the M. xanthus RelA protein (MXAN3204) revealed that it contains both the hydrolytic and biosynthetic domains found in E. coli SpoT (Sun et al., 2001; Diodati and Singer, personal communication), implying that the M. xanthus RelA protein is a bifunctional enzyme similar to B. subtilis RelA (Wendrich and Marahiel, 1997),Streptomyces coelicolor RelA and RshA (Chakraburtty et al., 1996; Sun et al., 2001), Streptococcus equisimilis Re1 (Mechold and Malke, 1997), and RelA in the gram-negative bacterium Rhodobacter capsulatus (Masuda and Bauer, 2004). This is very intriguing because it has been suggested that the relA and spoT genes in gram-negative organisms evolved from a duplicated gram-positive re1 (rsh)-like gene (Mittenhuber, 2001) and the majority of relAlspoT hybrid (rsh)genes are found in gram-positive organisms (Jain et al., 2006). Interestingly, Harris et al. (Harris et al., 1998) found two bands with a 311-bp relA probe that mapped to two different regions of the M. xanthus chromosome. This suggested the possibility of two relAlspoT-like genes in 111. xanthus. This initial observation is supported by the identification of a second putative relAlspoT-like gene, in addition to relA, in the M. xanthus genome (Goldman et al., 2006). Stringent Response-Related Homologues in M. xanthus Sequence analysis revealed that M. xanthus encodes all the known components of the (p)ppGpp cycle. It has a single copy of ndk (nucleoside diphosphate kinase), gpp (guanosine pentaphosphatase), ppx (exopolyphosphatase), and ppk (polyphosphate kinase), with the last two being involved in polyphosphate metabolism, as well (for a review, see Cashel et al., 1996).

Two relA/spoT-like homologues in M. xanthus The M. xanthus genome (Goldman et al., 2006) has a second putative relAlspoT-like gene (MXAN1364).The product of MXAN1364 has strong sequence similarity to the N-terminal hydrolase domain of SpoT; therefore, it has been designated “shd” for SpoT hydrolase domain. shd has been previously described by Diodati et al. as mx-1.594 (Diodati et al., 2006). shd is a small 420-bp gene and, based on sequence homology, is predicted to have a partial SpoT motif HD domain (Aravind and Koonin, 1998). Shd has the conserved HD domain with the substrate binding pocket and the HD doublet motifs found in the superfamily of metal-dependent hydrolases (Aravind and Koonin,

EVENTS 3 . INITIATIONAND EARLYDEVELOPMENTAL 1998; Hogg et al., 2004). Notably, E. coli RelA, which is solely a synthetase, has substitutions in these regions of the protein. The downstream RelA/SpoT domains that are necessary for (p)ppGppsynthesis are missing in Shd. Therefore, Shd may have hydrolytic properties and may possibly be involved in regulating (p)ppGpplevels. Although Shd has not yet been examined for SpoT activity, of particular interest is that the shd gene was previously implicated in M. xanthus development by O’Connor and Zusman (O’Connor and Zusman, 1990; O’Connor and Zusman, personal communication) as a temperature-sensitive aggregation (Tag) mutant. However, the exact nature of the Tag mutation in shd is not known.

M. xanthus has four DksA homologues Annotation of the M . xanthus genome has identified several homologues to dksA, recently shown in E . coli to play a critical and synergistic role in (p)ppGpp-dependent transcriptional regulation (Paul et al., 2004, 2005; Perederina et al., 2004). DksA binds to RNA polymerase and enhances ppGpp’s direct negative effect on rRNA promoters by reducing the open complex lifetime of RNA polymerase (RNAP) and inhibiting rRNA promoter activity (Paul et al., 2004; Perederina et al., 2004). DksA also directly and indirectly affects activation of amino acid promoters in concert with ppGpp (Paul et al., 2005). Furthermore, DksA has been shown to be involved in cell division, quorum sensing, expression of virulence factors, and the suppression of temperaturesensitive growth in dnaK mutants (Branny et al., 2001; Ishii et al., 2000; Kang and Craig, 1990; Turner et al., 1998).Although AdksA mutants have pleiotropic effects on cells including misregulation of numerous genes, mild UV sensitivity (Clifton et al., 1994), and filamentation (Ishii et al., ZOOO), it has been suggested that these are indirect effects of alterations in rRNA transcription and subsequent RNAP titration in the mutant (Paul et al., 2004). In d k s A mutant cells, rRNA promoters are unresponsive to changes in amino acid availability, growth rate, or growth phase (Paul et al., 2004). Interestingly, M . xanthus has four DksA homologues in its genome (Goldman et al., 2006) designated DksA (MXAN3200), DksB (MXAN3006), DksC (MXAN5718), and DksD (MXAN7086) (Table 2), instead of the typical single copy that exists in other organisms. These homologues were found by performing a BLAST analysis of the M. xanthus genome with E. coli DksA. MXAN3200 has the highest homology to E. coli DksA and therefore was named “DksA.” The M. xanthus homologues are slightly shorter than E. coli or B. subtilis DksA, but have the full-length

47

C-terminal 4-cysteine zinc finger suggestive of a transcriptional regulator.

EshA, PgpH, and HvrA Recently, three additional proteins, EshA, PgpH, and HvrA, have been shown to modulate (p)ppGpplevels or (p)ppGpp-dependent gene expression during vegetative growth in a variety of bacterial species (Liu et al., 2006; Masuda and Bauer, 2004; Saito et al., 2006). EshA is a cyclic AMP (CAMP)-binding protein that when disrupted, results in lower levels of (p)ppGppaccumulation during early to late growth phase in S. coelicolor (Saito et al., 2006). Based on the Saito et al. work, EshA is proposed to fine-tune and maintain a specific ppGpp level during stationary phase for antibiotic production via its nucleotide-binding domain. M. xanthus has two homologues to EshA (Table 2). In Listeria monocytogenes, PgpH is a putative integral membrane protein with an HD domain at its C terminus (Liu et al., 2006). The HD domain suggests that it may act as a metal-dependent phosphohydrolase (Aravind and Koonin, 1998). A cold-sensitive L. monocytogenes mutant with a transposon insertion in pgpH accumulates higher levels of (p)ppGppthan the wild-type cells. Therefore, PgpH is hypothesized to play a role in the cold-induced stress response by directly or indirectly modulating ppGpp levels, which are increased during low-temperature growth, and restoring them to normal vegetative levels (Liu et al., 2006). M . xanthus has a PgpH homologue defined by MXAN4737 (Table 2). Lastly, a third stringent-response-related protein, HvrA, was recently identified. In R. capsulatus, an hvrA mutation suppresses the lethality of a SPOTmutant (Masuda and Bauer, 2004). HvrA is a nucleoid-like protein, and in E. coli similar proteins coregulate (p)ppGppdependent genes by influencing the supercoiled state of the promoters (Johansson et al., 2000). Masuda and Bauer (2004)suggest that HvrA may be acting in concert with (p)ppGpp to regulate the transcription of specific promoters during growth. Although M. xanthus has homologues of the eshA and pgpH genes, no homologues have been found for hvrA (Table 2).

The Role of (p)ppGppin M. xanthus Development The initial starvation response occurs at the level of the individual cell, with each cell required to evaluate its own nutritional status. Because M. xanthus is unable to utilize carbohydrates, cells primarily rely on amino acids and a-keto acids to serve as carbon and energy sources, as well as substrates for protein synthesis. Physiological studies support the hypothesis that M. xanthus initially senses starvation by monitoring its translational capacity

DEVELOPMENT AND MOTILITY

48 Table 2

List of stringent-response-related homologues in M. xanthus“

Homologue name (organism)

Gene name (derivation) in M. xantbus

M x no.

MXAN no.

RelA (E. coli)

relA

4330

3204

SpoT HD domain (E. coli)

Shd (SpoT H D domain )

1594

1364

DksA (E. coli)

dksA

2229

3200

DksA ( E . coli)

dksB (DksA homologue B)

6139

3006

DksA (E. coli)

dksC (DksA homologue C)

0673

5718

DksA ( E . coli)

dksD (DksA homologue D)

5736

7086

EshA ( S . coelicolor)

MXAN6248

6739

6248

EshA ( S . coelicolor)

MXAN6249

6738

6249

PgpH (L. monocytogertes)

MXAN4737

3629

4737

HvrA (R. sphaeroides)

NA

NA

NA

Role in original organism

Comments

Actually an rsh (relAlspoT) gene; annotated as “stringentrespon” in M. xanthus Expression is modified in (p)ppGpp degradation (Diodati and Singer, nla4 mutant; incorrectly personal communicaannotated as “relA” in tion) M. xanthus Originally identified as (p)ppGpp-regulated transcription DnaK suppressor ( E . coli) (Paul et al., 2004) Partial DksA homologue, (p)ppGpp-regulated transcription also known as DksA3006 (Paul et al., 2004) (p)ppGpp-regulated Partial DksA homologue, transcription also known as DksA5718 (Paul et al., 2004) (p)ppGpp-regulated Partial DksA homologue, transcription also known as DksA7086 (Paul et al., 2004) Sustains (p)ppGpp dur- CAMP binding protein; ing late growth phase finely tunes ppGpp (Saito et al., 2006) threshold for antibiotic production Sustains (p)ppGpp dur- Overlap or duplication? ing late growth phase (Saito et al., 2006) Adjusts (p)ppGpp levels Putative integral membrane protein; has H D domain during low temp growth (Liu et al., 2006) Coregulates (p)ppGpp- Mutations suppress spoT lethality (Masuda and dependent genes durBauer, 2004); no homoing growth (Masuda logue in M. xanthus and Bauer, 2004) (p)ppGpp synthesis (Cashel et al., 1996)

“Mx and MXAN numbers refer to gene names from the original M 1 genome (Jakobsen et al., 2004) and the completed TIGR (The Institute of Genomic Research)/ Monsanto versions of the M. xanthus genome (Goldman et al., 2006),respectively. N o homologues of HvrA were found in M. xanthus. The corresponding references for determining the roles of the above genes are included in parentheses. NA, not applicable.

via the intracellular levels of (p)ppGpp (Manoil and Kaiser, 1980a, 1980b; Singer and Kaiser, 1995). This model is very attractive, as it provides a molecular link between metabolism and development of M. xanthus, and remains a starting point to understand this complex sensory pathway. Previously, Singer and Kaiser (1995) demonstrated that the M. xanthus RelA protein also functions as a ribosome-dependent (p)ppGpp synthetase and is required for the earliest aspects of development (Harris

et al., 1998; Manoil and Kaiser, 1980a, 1980b; Singer and Kaiser, 1995).Where it has been specifically examined, most developmentally regulated genes require an increase in the intracellular levels of (p)ppGpp for their expression, includingproduction of the extracellular A- and C- signals. relA mutants do not aggregate or form spores, and other genes that are known to decrease ppGpp levels within the cells, such as the csgA (Crawford and Shimkets, 2000b), nlal8 (Diodati et al., 2006), and nla4 (Diodati and Singer, personal communication; F. Ossa,

3. INITIATION AND EARLYDEVELOPMENTAL EVENTS M. E. Diodati, N. B. Caberoy, M. Singer, and A. G. Garza, unpublished data) mutants, also fail to sporulate. Interestingly, todK (MXAN6955), a gene that affects the timing of development, requires starvation for its expression but not (p)ppGppaccumulation (Rasmussen and Sngaard-Andersen, 2003). This suggests that todK may represent a member of a new class of genes that are starvation dependent and RelA independent. This is supported by preliminary DNA microarray data comparing global patterns of expression from reZA mutant and wildtype cells. Jose and Singer (I. R. Jose and M. Singer, personal communication) have identified a subset of genes whose expression is starvation induced and (p)ppGpp independent. In addition to relA, other genes have been implicated in (p)ppGpp regulation in M . xanthus, including two genes involved in C-signal regulation, socE and csgA (MXAN1294) (Crawford and Shimkets, 2000; see chapter 4 for a discussion on C-signal), and the nutrient sensor, Nsd (Brenner et al., 2004). Nsd (MXAN7402) is described in more detail in the nutrient sensors portion of this chapter. More recently, we found that the inactivation of two genes that encode os4-transcriptional activators, nZal8 (MXAN3692) (Caberoy et al., 2003; Diodati et al., 2006) and nZu4 (MXAN2516) (Caberoy et al., 2003; Ossa et al., unpublished), result in ppGpp accumulation defects, suggesting that at least two d4 promoters are regulating the expression of unknown genes whose gene products affect ppGpp levels. The exact mechanism of action for each of these gene products is not known and may be through RelA activity or stability, interaction with the ribosome, influencing the levels of (p)ppGppprecursors, or (p)ppGppmetabolism. The identification of these additional components that affect (p)ppGpp levels suggest that the regulation of the stringent response is more complex in M . xanthus. A list of genes and their (p)ppGppeffects upon inactivation is given in Table 3 .

49 levels necessary to support growth. Once the M . xanthus developmental process is under way, starvation must be monitored and available nutrients must be diverted toward completion of development. Crawford and Shimkets (2000b) have shown that when SocE is depleted, cell growth is arrested, and cells start to accumulate (p)ppGpp and initiate a stringent response even in the presence of sufficient nutrients to support growth. In addition, DNA and stable RNA synthesis is inhibited [which correlates with increased (p)ppGpplevels], and sporulation is induced upon socE depletion. Their data are consistent with SocE acting as a repressor of development. The C-signaling protein, CsgA, appears to induce the stringent response. It has been shown, through amino acid substitution studies of the CsgA protein and mutant analysis, that CsgA maintains growth arrest during development, stimulates (p)ppGpp synthesis in the absence of SocE, and mediates sporulation. Moreover, upon starvation, the csgA mutant can initiate a stringent response but accumulates only one-half of the (p)ppGppof wild-type cells and fails to sustain these higher levels during development (Crawford and Shimkets, 2000b). Therefore, it appears that although CsgA is not essential for the stringent response when it is initiated by amino acid starvation, it is important for maintaining the response during A-signaling and beyond. Interestingly, not only is the M. xunthus stringent response regulated by the SocE and CsgA proteins, but also their transcription is dependent on the stringent response. The transcription of socE is inhibited by increased (p)ppGpp levels, as may be expected for

Table 3 List of genes affecting (p)ppGpp accumulation in M. xanthus Gene name

SocE and CsgA The stringent response, the recognition of starvation and the subsequent increase in (p)ppGpp levels that accompany the limitation of amino acids, mediates multiple physiological and metabolic changes in the cell (Cashel and Rudd, 1989). A unique aspect of the M . xanthus stringent response is that it is regulated, in part, by the SocE and CsgA proteins. Maintenance of the response is important because the stringent response is coupled to the formation of the multicellular fruiting body. Within 2 h of development, A-signal, which is composed mostly of amino acids, is produced and cells need to decipher the extracellular signal levels of amino acids from the

relA socE csgA

nsd nlal8 nla4

(PjPPGPP effect when inactivated"

No accumulation Increased 2.5- to 5-fold in 1% CYE Decreased 2-fold after starvation Increased 2-fold in 0.5% CTT Decreased 2-fold (veg), 5.5-fold (dev) Decreased 2-fold (veg), 2.6-fold (dev)

Reference(sj Harris et al., 2001 Crawford and Shimkets, 2000b Crawford and Shimkets, 2000b Brenner et al., 2004 Diodati et al., 2006 Ossa et al., unpublished; Diodati and Singer, personal communication

"dev, development; veg, vegetative growth.

50 a repressor of development. In contrast, increased (p)ppGpplevels stimulate csgA transcription (Crawford and Shimkets, 2000a). This RelA-dependent transcriptional regulation of both csgA and socE RNA levels creates an increase in the ratio of CsgA to SocE and results in cessation of growth and the redirection of resources towards development (Crawford and Shimkets, 2000a, 2000b). In summary, the balance of SocE and CsgA proteins in the cell is critical for sustaining the developmental program past initiation and is just one example of the unique aspects of the stringent response in this organism.

Nla18 and Nla4 When nZal8 is inactivated, it results in pleiotropic effects in vegetative and developmental gene expression (Caberoy et al., 2003; Diodati et al., 2006). These mutants grow two to three times slower than wild-type cells, exhibit a mild temperature-sensitive phenotype, and are very prone to lysis due to disruptions in membrane permeability and overall integrity (Diodati et al., 2006). These mutants accumulate 18 to 50% less ppGpp than the wild type upon nutrient downshift and have similar developmental defects to relA mutants, albeit not as severe. The phenotype of the nla28 mutant cannot simply be explained by its ppGpp defect; the membrane protein defect is not seen with relA mutants, and vegetative microarray data show that nlal8 mutants affect the regulation of a variety of translation-related genes and genes encoding transcriptional regulators (Diodati et al., 2006). Therefore, Nlal8’s role in the accumulation of ppGpp appears to be indirect, because Nlal8 is required for overall balanced growth. Nla4 is a second d4transcriptional activator that is required for normal vegetative growth and development in M . xanthus. nla4 mutant cells fail to aggregate and sporulate normally, and when codeveloped with wildtype cells, the sporulation deficiency fails to be rescued (Caberoy et al., 2003). Further analysis of this mutant reveals multiple defects in developmental gene expression and signal production (Ossa et al., unpublished) and ppGpp accumulation (Diodati and Singer, personal communication; Ossa et al., unpublished). RelA regulation Although recent work (Harris et al., 1998; Manoil and Kaiser, 1980a, 1980b; Singer and Kaiser, 1995)has demonstrated the importance of the stringent response in starvation recognition and development in M. xanthus, little is known about the regulation and control of the key regulator RelA. Pertinent questions such as how the RelA protein modulates its biosynthetic and degradative

DEVELOPMENT AND MOTILITY properties in response to environmental cues still remain. It has recently been shown that the bifunctional RelM SpoT homologue in Streptococcus dysgalactiae subsp. equisimilis catalyzes opposing synthetase and hydrolase reactions at distinct active sites (Hogg et al., 2004). The two different conformations of the enzyme, in which either the synthetase or the hydrolase activity is turned on while the reciprocal activity is repressed, appear to involve transmission of a ligand-induced signal between the two active sites (Hogg et al., 2004). This reciprocal regulation of the two catalytic activities is governed by the C-terminal half of the Re1 protein (Mechold et al., 2002). Similar regulation has been found for the RelN SpoT homologue in Mycobacterium tuberculosis, as well (Avarbock et al., 2000). The regulatory process controlling M. xanthus RelA enzyme activities may be similar to these other bifunctional proteins. In vitro biochemical and kinetic tests with purified RelA have yet to be performed. One approach to begin to answer the question of RelA regulation is to first identify and characterize the components of the M . xanthus stringent response and then elucidate the mechanism that controls the levels and activities of the secondary messenger, (p)ppGpp.A model summarizing the M . xanthus (p)ppGppresponse is shown in Fig. 3.

Other Early Developmental Effectors: Nutrient Sensors versus Developmental Timers Over the last 10 to 15years, data from several laboratories have identified two general classes of genes that alter the timing of development: developmental timers and nutrient sensors. A partial list of genes involved in these two processes is provided in Table 4. Developmental timers are genes defined by mutations that either speed up or slow down the developmental process but still require starvation for activation of development. Nutrient sensors, on the other hand, are genes defined by mutations that cause a premature initiation of development under inappropriate conditions, i.e., conditions that would normally not initiate the developmental pathway by wild-type cells. Mutations in genes encoding developmental timers, such as espAB (MXAN0931 and -0932) (Cho and Zusman, 1999a), espC (MXAN6855) (Lee et al., 2005), redCDEF (MXAN0459 through -0462) (Higgs et al., 2005), todK (MXAN6955) (Rasmussen and Sogaard-Andersen, 2003), and rodK (MXAN0733) (Rasmussen et al., ZOOS), have been shown to either accelerate or delay the developmental program. However, in all cases nutrient limitation was still required to initiate the developmental response in these mutants. Intriguingly, many of the developmental timers uncouple the spatial requirement of aggregation and

3. INITIATION AND EARLYDEVELOPMENTAL EVENTS

51

P+ i T P p k

j:zp 1:c.(= ADP ATP

PPPG (GTP)

Ndk

+ ppi

DksA + + Positively controlled

-r'

GPPf Ppx P P P G P P . 7

Nla18

CsgA

SocE Nsd

Development

PPG(GDP)

PPi

PPGPP

E d

genes

1

Negatively controlled genes

Figure 3 Diagram of the M. xanthus (p)ppGpp response and genes involved in its activation. The enzymes involved in (p)ppGpp metabolism are shown in bold. The asterisk (") represents uncharged tRNA in the acceptor site of the ribosome that triggers the associated RelA to catalyze the synthesis of (p)ppGpp from ATP and GTP. See Figure 2 legend for more details. In M . xanthus, (p)ppGpp levels are maintained by balancing the hydrolase activity of RelA with its biosynthetic activity. Diodati and Singer have postulated that Shd, a gene product with homology to the hydrolase domain of E. coli SPOT, may play a role in (p)ppGpp degradation. In addition, five proteins (SocE [Crawford and Shimkets, 200Ob1, Nsd [Brenner et al., 20041, Nla18 [Diodati et al., 20061, Nla4 [Diodati and Singer, personal communication; Ossa et al., unpublished], and CsgA [Crawford and Shimkets, 2000bJ) have been shown to either inhibit or stimulate ppGpp accumulation in M. xanthus through as yet unknown mechanisms. For more details, consult text. With elevated (p)ppGpp levels, RNA polymerase (Eo*) is predicted to interact with DksA, to modulate changes in RNA polymerase activity to alter gene expression. To date no secondary (p)ppGpp biosynthetic pathway has been identified.

sporulation. Inactivation of espA, espC, or rodK results in sporulation outside the fruiting body (Cho and Zusman, 1999a; Lee et al., 2005; Rasmussen et al., 2005). In addition, it has been shown that overexpression of CsgA results in this same phenotype (Kruse et al., 2001), implying that a lack of inhibition of the C-signaling pathway leads to induction of sporulation independent of the high cell density requirement to reach the appropriate C-signal threshold. While increased C-signal production results upon loss of RodK and EspC function (Lee et al., 2005; Rasmussen et al., 2005), csgA and espA mutant mixing experiments suggest that the espA mutant produces less C-signal than wild-type cells (Lee et al., 2005). These data imply that EspA may have a function other than inhibiting C-signal production causing a mutant in

espA to sporulate outside the fruiting body. Elucidating the mechanisms of spatial coupling of fruiting bodies and sporulation can lead to fascinating insights into the intricate regulation of this process. Also, there are many other developmental mutants that affect the timing of development. The overall defect is usually a delay, which may be due to numerous factors, including defects in motility, outer membrane components, or poor growth (i.e., poor nutrient stores).

Nutrient Sensors Nutrient sensing encompasses at least four important elements: (i) the detection of nutrients in the environment, (ii) the uptake of nutrients into the cell, (iii) metabolism

DEVELOPMENT AND MOTILITY

52

Table 4 Partial list of genes implicated in nutrient sensing and developmental timing Gene or locus

Sensor or timer

asgD

Sensor

bcsA

Sensor

Che3 operon

Sensor

MXAN2 9 02 (mx-332 0) nsd

Sensor Sensor

sigC

Sensor

socE

Sensor

spdR

Sensor

espA espB esp C redCDEF rodK todK

Timer Timer Timer Timer Timer Timer

Null phenotype Hypersensitive to nutrient levels Forms fruiting bodies on rich media Forms fruiting bodies on rich media Hypersensitive to nutrient levels Forms fruiting bodies on rich media Forms fruiting bodies on rich media Forms fruiting bodies on rich media Forms fruiting bodies on rich media Speeds up development Slows down development Speeds up development Speeds up development Speeds up development Speeds up development

and utilization of those nutrients by the cell, and (iv) the regulation of developmentally specific signal transduction pathways necessary to respond to the abundance or lack of nutrients in the cell’s environment. We have previously described how (p)ppGpp can act as an internal starvation signal, using the cell’s translational capacity as a monitor of the cell’s nutritional status. However, it has become evident that (p)ppGpp is not the only participant in starvation recognition. Work from several labs has identified genes that have been implicated in affecting nutrient sensing. These genes include the previously described socE (Crawford and Shimkets, 2000a, 2000b), nsd (Brenner et al., 2004), the genes of the che3 cluster (crdB, mcp3A, mcp3B, cheA3 [MXAN5147 through 51521) (Kirby and Zusman, 2003), a component of the A-signaling generation complex, asgD (MXAN6996) (Cho and Zusman, 1999a), and a developmentally regulated os4-transcriptional activator, MXAN2902 (previously described as Mx-3320) (Jakobsen et al., 2004). Nsd Nsd (nutrient-sensing/utilizing defective) (Brenner et al., 2004) was originally identified as the gene controlled by the developmentally regulated promoter known as

04469 (IR nomenclature represents TnSlac insertions) (Kroos et al., 1986). Nsd appears to be important for sensing nutrients in the environment and acts as an inhibitor of development in the presence of nutrients. Brenner et al. (2004)have shown that, under low nutrient conditions, growth is decreased 2- to 2.5-fold in nsd mutants compared to wild-type cells. Upon further characterization in 0.5% Casitone-Tris (CTT) broth, nsd mutants accumulate twofold more (p)ppGppthan wild-type cells. Moreover, nsd mutant cells initiate development on nutrient agar, but development is compromised; viable spore count is reduced (Brenner et al., 2004). Interestingly, on higher-nutrient plates (>0.5% CTT), nsd mutants can go through development but produce heat- and sonicationresistant phase-dark cells instead of spores (Brenner et al., 2004). These data demonstrate that the nsd mutants can initiate development and aggregate in the presence of nutrients, yet appear to require additional factor(s) for wild-type sporulation under these conditions. T h e Che3 operon Che3 is a chemosensory cluster (which includes crdB, mcp3A, mcp3B, cheA3, and others) that has been shown to expedite development when the genes are inactivated. crdB, mcp3A, mcpSB, and cheA3 mutants form fruiting bodies earlier than wild-type cells in a density-independent manner (Kirby and Zusman, 2003). This is corroborated by the observation that developmental reporter fusions 04403, IR4411, and 04521 are expressed earlier and at higher levels in mutant cells during development, as well as during vegetative growth in the Amcp3A mutant (Kirby and Zusman, 2003). In addition, these mutants form distorted fruiting bodies with spores on nutrient agar. These data suggest that the proteins encoded by the che3 cluster genes are nutrient sensors that act by blocking developmental gene expression during growth. There is a divergently transcribed os4-transcriptionalactivator gene, crdA, upstream of the gene cluster. It is proposed, based on mutational and yeast two-hybrid analyses, that the Che3 system modulates the activity of CrdA, which controls expression of specific developmental genes (Kirby and Zusman, 2003). It is important to note that the rapidly formed fruiting bodies of the che3 mutants on clone fruiting (CF) media look similar to wild-type fruiting bodies but the mutants are unable to sporulate normally. The vegetative expression of development-specific genes, the bypassing of the highdensity requirement for development, and the general premature entry into development phenotype of the che3 mutants suggest that these temporal and/or signaling checkpoints of fruiting body formation are critical for wild-type sporulation.

3 . INITIATIONAND EARLYDEVELOPMENTAL EVENTS As gD The inactivation of the asgD gene results in cells that are hypersensitive to nutrients; their development is inhibited by limited amounts of nutrients that are low enough to induce development in wild-type cells. On CF medium (10 mM MOPS [3-4-morpholine propanesulfonic acid], 0.015% Casitone [Difco], 8 mMMgSO,, 1 mMKH,PO,,0.2% sodiumcitrate,0.02% (NH,),SO,, 0.1% pyruvate, and 1.5% agar), which contains asmall amount of nutrients for cells to undergo a gradual starvation, AasgD mutants form loose aggregates and do not sporulate (Cho and Zusman, 1999a). In contrast, on the more-stringent starvation medium MMC (10 mM MOPS buffer, 4 mM MgSO,, 2 mM CaCl,, and 1.5% agar), AasgD cells are able to form slightly irregular fruiting bodies and have 35% of wild-type sporulation (Cho and Zusman, 1999a).In order to decipher the component(s) in the CF media that were causing the developmental inhibition, the authors removed components and added them back individually. Eliminating citrate, pyruvate, or Casitone from the media increased the asgD mutant’s spore count to 0.25, 0.96, and 44% of wild-type numbers, respectively (Cho and Zusman, 1999a). Taken together, these data suggest that AsgD may be involved in nutrient sensing. When asgD is inactivated, the cells do not perceive starvation until nutrient concentrations are minuscule. Once these nutrients are consumed, the mutant cells may not have enough energy to sustain and complete development. Therefore, as opposed t o having a function of inhibiting development, like Nsd and the proteins of the Che3 cluster, AsgD appears to be acting by promoting cells to develop when there are still sufficient nutrients in the environment to support the developmental process. Furthermore, the AasgD mutant can be extracellularly complemented and appears to be a member of the “asg mutant” group (Cho and Zusman, 1999a). Since mutations in asgD also affect A-signal production, AsgD may serve as a link between starvation sensing and activation of the extracellular A-signal generation pathway described in the latter sections of this chapter. MXAN2 902 MXAN2902 (previously described as Mx-3320) is a crs4-transcriptional activator protein that appears to be involved in sensing nitrogen-related nutrients in the environment at about 12 h of development (Jakobsen et al., 2004). When the MXAN2902 gene is disrupted, mutants are defective in mound formation (Jakobsen et al., 2004). The mounds of the mutant are flatter than the wild type, and they have projecting tails of cells at

53 one end. Spores are found within the mound but also at the periphery and are concentrated in the protruding tail of lower-density cells. This density-independent manner of sporulation is similar to what is seen with the che3 cluster mutants and many developmental timer mutants described above. Like the asgD mutant, these cells had defective development on CF agar but show wild-type development on more-stringent starvation agar. When particular components were removed from the CF medium or added to the TPM medium (10 mM Tris-HC1 [pH 8.01, 1 mM KH,PO,, 8 mM MgSO,, and 1.5% agar) to interpret the nutrient-sensing defect, the mutants were hypersensitive to the nitrogen sources in the media (Jakobsen et al., 2004). With these results, and the phenotypic similarities that the MXAN2902 mutant shares with the asgD mutant in regards to nutrient sensing, MXAN2902 may act as a positive effector of development as well. Nutrient sensors can work by the promotion or inhibition of development, as described above. The method by which these nutrient sensors elicit their actions to affect entry into development is not yet known. Possible direct or indirect mechanisms could include detection, transport, or metabolism of nutrients as well as the regulation of signal transduction pathways that may or may not control these above-mentioned functions. As demonstrated by the characterization of the nutrient sensor genes, the overall decision to initiate development requires a highly regulated system with the input and cooperation of many factors in order to proceed.

Global Gene Regulation during the Cellular Response to Starvation Once nutrient limitation is detected, cells respond by redirecting transcription to prepare for the new environmental condition. M. xanthus cells must meet density requirements as well as have solid support. It has been shown that overall protein synthesis patterns are dramatically different during the early stationary phase and initiation of development, even though under both conditions, cells are entering a nutrient-limiting environment (Ueki and Inouye, 1998). It is evident that there are additional factors that influence the cells to initiate development, which is an expensive, energy-demanding process. Therefore, cells need to distinguish between a gradual, an immediate, or an absolute starvation state with enough nutrients to carry out development but not enough to sustain vegetative growth, coupled with a strong community consensus. For example, wild-type cells grown on rich agar media, such as CTT (Hodgkin and Kaiser, 1979) or CTTYE (CTT supplemented with

54 0.5% yeast extract), enter stationary phase but do not undergo fruiting body development or differentiation. These data suggest that M. xanthus cells and populations evaluate multiple facets of their environment and then respond accordingly. Strict control of gene expression is needed for the appropriate survival response to be evoked. Sigma factors are important regulatory elements that bacteria utilize to increase the specificity of largescale transcriptional responses. Control of induction of development and/or early developmental gene expression appears to rely, in part, on three global regulators: RpoN, SigD (the M . xanthus RpoS homologue), and SigC.

RpoN and Associated Activator Proteins If developmental criteria are met, M . xanthus cells respond by initiating a complex pathway that causes changes in behavior and culminates in cellular differentiation. Many of the known early developmentally regulated genes in M . xanthus are transcribed from d4like promoters. These include the previously characterized genes spi (MXAN4276) (Keseler and Kaiser, 1995), mbhA (MXAN7061) (Romeo and Zusman, 1991), asgE (MXAN1010) (Garza et al., 2000a, 2000b), and sdeK (MXAN1014) (Garza et al., 1998; Pollack and Singer, 2001), all of which are expressed within the first few hours of development. rpoN (MXANlOGl), which encodes d4, is usually associated with specialized metabolic functions, such as nitrogen regulation in E . coli and Salmonella (Kustu et al., 1989; Ninfa et al., 1995) or motility in Pseudomonas (Hobbs et al., 1993), Caulobacter (Brun and Shapiro, 1992), and M. xanthus (Wu and Kaiser, 1997). There is a single copy of rpoN in M. xanthus (Goldman et al., 2006). Uniquely, rpoN is essential for growth (Keseler and Kaiser, 1997). as4-dependent promoters require a positive activator for transcription. The identification of as4-likepromoter elements (based on their -12 and -24 sequences) 5’ to many of the earliest developmental genes suggests a role for rpoN and these associated transcriptional activators in early development. Enhancer binding proteins (EBPs) are transcriptional activators for d4promoters and are named after the eukaryotic enhancer binding proteins they resemble. These proteins typically bind 100 to 200 bp upstream of d4promoter elements (Morett and Buck, 1988) and interact with RNAP through a DNA looping mechanism that allows the isomerization of the RNAP from the closed to the transcriptionally active open complex in an ATP-dependent manner. (For reviews, see Morett and Segovia, 1993; Studholme and Dixon, 2003; and Xu and Hoover, 2001.)

DEVELOPMENT AND MOTILITY The NtrC protein of E . coli is one of the best characterized of these activator proteins, and the term “NtrClike activators” has been used interchangeably with “enhancer binding proteins” to describe these activators in M. xanthus. NtrC-like activators are usually associated with a subset of activators containing an N-terminal response regulator domain. For the purposes of this general discussion and to prevent confusion, the term “enhancer binding protein” or “EBP” will be used to describe these os4-transcriptional activators. Fifty-two potential EBPs have been identified by sequence analysis in M . xanthus (Caberoy et al., 2003; Gorski and Kaiser, 1998; Jakobsen et al., 2004; Jelsbak et al., 2005) and are listed in Table 5. EBPs typically have a three-domain structure that consists of a regulatory domain(s)at the N terminus, a highly conserved central ATPase domain, and a helix-turn-helix DNA binding motif at the C terminus (Jelsbak et al., 2005; Studholme and Dixon, 2003). These activators were identified in the M . xanthus genome sequence, using the conserved central ATPase domain. Based on the essential nature of rpoN in M. xanthus and the fact that many of the earliest known developmental genes have predicted RpoN-dependent promoters, Keseler and Kaiser (1997) suggested that early development may be a function of a succession or cascade of EBPs that drives progression through the early stages of development; analogous to the sigma factor cascade that drives B . subtilis sporulation (Kroos et al., 1999). The collective efforts of several labs to identify EBPs and to characterize knockout and insertion mutants have revealed 18 that have vegetative and/or developmental phenotypes (Caberoy et al., 2003; Gorski and Kaiser, 1998; Gronewold and Kaiser, 2001, 2002; Guo et al., 2000; Hager et al., 2001; Jakobsen et al., 2004; Jelsbak et al., 2005; Kaplan, 2003; Tse and Gill, 2002). Interestingly, most of these activators with developmental phenotypes are important for early developmental gene expression. Therefore, RpoN and associated EBPs may play an important role in coordinating the starvation recognition process and the early steps of development. Eleven EBPs are described in detail below. The seven remaining EBPs with vegetative and/or developmental defects have been attributed to motility or late developmental defects (Table 5 ) and are beyond the scope of this chapter.

EBPs with vegetative growth defects As previously described, nlal8 and nla4 mutants have severe vegetative defects including slow growth and decreased ppGpp accumulation (Diodati et al., 2006; Diodati and Singer, personal communication; Ossa

Table 5 Names, MXAN and M x numbers, N-terminal regulatory domains, gene knockout constructions, mutant phenotypes, and related references for all EBPs in M . xanthusa Regulatory domain(s)b

KO?

Phenotype and comments

Name(s) of EBP

MXAN no.

M x no.

ActB, Mxa259

3214

4338

RR

Yes: I, D

Late devdef; adjacent to RR

CrdA, Nla26, Mxa227

5153

1467

RR

Yes: I, D

Early devdef, nutrient sensor (see text); HK nearby

FrgC, Nla25

1128

223 8

RR

Yes: I

WT; adjacent to HK

HsfA

5364

1035

RR

No

MrpB, Mxa189

5124

5602

RR

Yes: I, D

Mx-1288

4339

1288

FHA

Yes: I

Mx-3098, Mxa198

5041

3098

GAF

Yes: I

Interacts with sigma 70 RNAP and activates lonD (bsgA)in vitro; adjacent to HK Early devdef (see text); adjacent to HK WT; adjacent to STK (divergently transcribed) WT; up at 12 h of development

Mx-3320

2902

3320

Not identified

Yes: I

Mx-3 725

3333

3725

FHA, GAF

Yes: I

Mx-48 85

4899

4885

FHA

Yes: I, D

Mxal91 Mxa211

0353 0172

1757 3558

FHA RR

Yes: I No

Mxa213 Mxa221 Mxa249 Mxa264 Mxa296, Mxa2lO

4020 4196 5672 0116 0180

1598 3656 2469 5079 5565

FHA, GAF RR FHA FHA Not identified

Yes: I No No Yes: I Yes: I

Nla 1 Nla2 Nla3 Nla4

5853 3381 4785 2516

3336 3471 3814 0839,0840

RR Not identified RR RR

Yes: I Yes: I Yes: I Yes: I

WT WT; adjacent to HK

Nla5

2501

1502

FHA, FHA

Yes: I

WT

Devdef, nutrient sensor MXAN 2902 (see text) WT; lacks GAFTGA motif; STK nearby Late devdef; near STK-associated KapB Slow growth (see text) HK nearby (divergently transcribed) Early dev; STK nearby Adjacent to HK

WT; adjacent to STK Early devdef (see text); HK nearby

S-motility defect; adjacent to HK

Slow growth, devdef (see text)

Reference(s) Gorski et al., 2000; Gorski and Kaiser, 1998; Gronewold and Kaiser, 2001,2002 Caberoy et al., 2003; Gorski and Kaiser, 1998; Kirby and Zusman, 2003 Caberoy et al., 2003; Cho et al., 2000 Ueki and Inouye, 2002

Gorski and Kaiser, 1998; Sun and Shi, 2001a, 2001b Jelsbak et al., 2005 Gorski and Kaiser, 1998; Jakobsen et al., 2004 Jakobsen et al., 2004 Jelsbak et al., 2005 Jelsbak et al., 2005 Gorski and Kaiser, 1998 Gorski and Kaiser, 1998; Kaufman and Nixon, 1996 Gorski and Kaiser, 1998 Kaufman and Nixon, 1996 Kaufman and Nixon, 1996 Gorski and Kaiser, 1998 Gorski and Kaiser, 1998; Diodati and Singer, personal communication Caberoy et al., 2003 Caberoy et al., 2003 Caberoy et al., 2003 Caberoy et al., 2003; Ossa et al., unpublished Caberoy et al., 2003

(Continued)

Table 5 Names, MXAN and M x numbers, N-terminal regulatory domains, gene knockout constructions, mutant phenotypes, and related references for all EBPs in M . xanthusa(Continued) Name(s) of EBP

Regulatory domain(s)*

KO?

Phenotype and comments

Reference(s)

MXAN no.

Mx no.

NlaG

4042

2063

RR

Yes: I

Nla7 Nla8 Nla9, TaR3

0937 4580 3952

2140 2840 2942

RR RR GAF

Yes: I Yes: I Yes: I

NlalO

5048

4193

Not identified

Yes: I

Nlall

6426

4170

Not identified

Yes: I

Nla12 Nla13 Nla14

2159 3811 3095

4562 6755 4901

FHA, GAF RR FHA

Yes: I Yes: I Yes: I

NlalS, Nla16 Nla17 Nla18

5680 3418 3692

0033, 0233 0517 0888,0889

RR' RR FHA

Yes: I Yes: I Yes: I

Nla20 Nla21

4252 4983

2594 4341

RR RtcR

Yes: I Yes: I

Nla22, Mx-4756

4240

4756

RR

Yes: I

Nla23, PilR2

5777

1973

RR

Yes: I

Nla24

7440

2057

RR

Yes: I

Nla27 Nla2 8

1345 1167

2176 1617

FHA RR

Yes: I Yes: I

Nla34

1565

4965

GAF

No

A- and S-motility defect; adjacent to HK WT Early devdef (see text); adjacent to HK Down regulated in nlal8 mutant

PilR, Mxa15

5784

3013

RR

Yes: I, D

S-motility defect; adjacent to HK

Gorski and Kaiser, 1998; Kaufman and Nixon, 1996; Wu and Kaiser, 1995, 1997

SasR, Mxa287

1245

0124

RR

Yes: I, D

Early devdef (see text); adjacent to HK

Gorski and Kaiser, 1998; Guo et al., 1996; Yang and Kaplan, 1997

Early devdef (see text); adjacent to HK WT; adjacent to HK WT; adjacent to HK WT, regulates antibiotic TA synthesis WT; adjacent to STK (divergently transcribed) WT; adjacent to STK (divergently transcribed) WT; adjacent to STK WT; adjacent to HK WT; adjacent to STK (divergently transcribed) WT WT, adjacent to HK Slow growth, devdef (see text); adjacent to STK (divergently transcribed) WT, adjacent to HK WT; upstream RNA 3' terminal phosphate cyclase (divergently transcribed) WT; up at 12 h of development; HK nearby S-motility defect; adjacent to HK

Caberoy et al., 2003 Caberoy et al., 2003 Caberoy et al., 2003 Caberoy et al., 2003; Y. Paitan, E. Orr, E. Z. Ron, and E. Rosenberg, unpublished data Caberoy et al., 2003 Caberoy et al., 2003 Caberoy et al., 2003 Caberoy et al., 2003 Caberoy et al., 2003 Caberoy et al., 2003 Caberoy et al., 2003 Caberoy et al., 2003; Diodati et al., 2006 Caberoy et al., 2003 Caberoy et al., 2003

Caberoy et al., 2003; Jakobsen et al., 2004 Caberoy et al., 2003; Jelsbak and Kaiser, 2005 Caberoy et al., 2003; Lancero et al., 2004 Caberoy et al., 2003 Caberoy et al., 2003 Diodati et al., 2006

3. INITIATION AND EARLYDEVELOPMENTAL EVENTS

3 0

v 4-2

8V

3

c

e?

*

d p:

Z 2 2 2

00

m c n m 3 r . c n

I\ 0

CA 3

m

z

!xa

"J

v)

0

W

Y

W

r

d d

.

2

r

.

m

f

--

57 et al., unpublished). Nla18 (MXAN3692) is an atypical EBP that has a forkhead-associated (FHA) domain. as its N-terminal regulatory module, instead of the more-common response regulator domain that Nla4 (MXAN2516) contains (Diodati et al., 2006; Jelsbak et al., 2005). Also, nlul8 is adjacent to pskB5, a gene that encodes a serine threonine kinase that is important for timely development (S. Inouye, personal communication), which is down regulated in an nlu18 mutant during vegetative growth (Diodati et al., 2006). Proteins with FHA domains have been shown to interact with threonine-phosphorylated substrates of serinekhreonine protein kinases in M. tuberculosis(Alderwick et al., 2006; Molle et al., 2003). Such activators may act as regulatory links to various signal transduction pathways mediated by serinehhreonine kinases (Alderwick et al., 2006; Jelsbak et al., 2005; Kroos, 2005; Molle et al., 2003). Twelve EBPs with FHA domains have been found in M. xanthus (Jelsbak et al., 2005), and four of these activators, including Nlal8, have vegetative or developmental phenotypes when their genes are disrupted. Inactivation of mxa213 (MXAN4020) (Gorski and Kaiser, 1998) or mx-4885 (MXAN4899) (Jelsbak et al., 2005) results in cells with developmental defects and is discussed below. Disruption of mxul91 (MXAN0353) (Gorski and Kaiser, 1998) results in a mutant with a slow growth rate similar to nlul8 or nlu4 mutants but, interestingly, displays wild-type development.

EBPs with early developmental defects Eight of the 18 EBPs with vegetative or developmental phenotypes affect early developmental events. These activators are SpdR (MXAN1078),CrdA (MXAN5153), SasR (MXAN1245), MrpB (MXAN5124), Mxa296 (MXAN0180), Mxa213 (MXAN4020), Nla6 (MXAN4092), and Nla28 (MXAN1167) (Caberoy et al., 2003; Gorski and Kaiser, 1998; Guo et al., 2000; Hager et al., 2001; Kirby and Zusman, 2003; Sun and Shi, 2001a, 2001b; Tse and Gill, 2002). SpdR is a typical EBP with a response regulator domain that is in an operon with its putative cognate histidine kinase, SpdS. It was found as a bypass suppressor for bsgA mutants (Hager et al., 2001; Tse and Gill, 2002) and is described in more detail in the cell signaling portion of this chapter. SpdR is expressed during vegetative growth and acts as a nutrient sensor that inhibits development in the presence of nutrients. Mutations in spdR alter a cell's vegetative patterns of gene expression, and spdR mutants express developmental genes on rich media. Also, an early A-signal-dependent developmental gene with a oS4promoter, spi (R4521), requires SpdR for its expression (Hager et al., 2001). Based on these data, SpdR may

Next Page

58 act between the assessment of nutrient conditions and signal-dependent gene expression. CrdA is an EBP that is negatively regulated by the nutrient-sensing genes of the Che3 gene cluster described in the previous section. The crdA gene is divergently transcribed from the che3 promoter region, and CrdA and CheA3 (a histidine kinase) interact very strongly with each other in yeast two-hybrid studies (Kirby and Zusman, 2003). Cells with mutations in crdA are delayed 12 to 24 h in development and reduce and delay the expression of mbhA during development. Furthermore, CrdA may be directly or indirectly controlling a wide variety of developmental genes because mutations in che3 cluster genes, of which CrdA is epistatic, affect a large number of developmental genes (Kirby and Zusman, 2003). The CheA3 kinase appears to modulate the activity of its cognate response regulator, CrdA, by inhibiting activation of the EBP to prevent developmental gene expression during vegetative growth. SasR was first identified and characterized by Gorski and Kaiser (1998) as Mxa287. The gene corresponding to Mxa287 was found, along with 1 2 other putative d4 transcriptional activators, with degenerate PCR probes to the conserved central domain (Gorski and Kaiser, 1998; Kaufman and Nixon, 1989). Mutants in mxa287 have both vegetative and developmental defects. On nutrient agar, the colonies are smaller and less cohesive, and on starvation media the cells are blocked early in development, appear flat, and are unable to express 04521. When complementation studies were carried out, mxa287 mutants could produce both A- and C-signal but failed to respond to the extracellular signals provided by the wild-type cells (Gorski and Kaiser, 1998). This led Gorski and Kaiser to hypothesize that Mxa287 may be acting as part of a signal reception pathway. These observations and conclusions corroborated very nicely with the sasR characterization described by Kaplan and colleagues (Guo et al., 2000; Yang and Kaplan, 1997). SasR is part of a signal transduction pathway involved in A-signal sensing (Guo et al., 2000). The sasR gene encodes a d4response regulator that is in an operon with the histidine kinase, sass. SasR is predicted to function downstream of Sass, as a positive regulator of spi (a4521) expression (Yang and Kaplan, 1997).SasR is discussed in more detail in the A-signaling section of this chapter. The mrp locus consists of a histidine kinase homologue (mrpA)and an NtrC-like response regulator gene ( m r p B )in an operon adjacent to an independently transcribed CAMP receptor protein-like transcriptional regulator (mrpc) (Sun and Shi, 2001b). MrpB is induced upon starvation, up-regulated during development, and

DEVELOPMENT AND MOTILITY required for both aggregation and sporulation. MrpB is also known as Mxa189, which was unable to be characterized due to difficulties in obtaining insertion mutants with the targeted PCR probe method (Gorski and Kaiser, 1998). The mrpB deletion mutants are completely flat on starvation media. Based on site-directed mutagenesis studies of the conserved aspartate of MrpB, the phosphorylated form of MrpB is required for aggregation while dephosphorylation of MrpB is needed for sporulation (Sun and Shi, 2001b). Sun and Shi (2001a, 2001b) proposed that MrpB functions after (p)ppGpp and A-signaling, but before C-signaling. The expression of mrpB is reduced in both a relA and an asgA mutant background, but not in a csgA mutant. In addition, mrpB mutants produce A-signal to 80% of wild-type levels and very little C-signal. Expression of six TnSluc fusions, including A- and C-signal-dependent markers, were reduced in the mrpB mutant (Sun and Shi, 2001a, 2001b). Although these data do support the claim that mrpB is acting after starvation initiation, the predicted function of MrpB after A-signaling does not account for the 41% reduction in the expression of the A- and Csignal-independent a4408 (sdeK) in the mrpB mutant. The role of MrpB in sdeK regulation requires further characterization. MrpB function prior to C-signaling is supported by data showing that MrpC is essential for the expression of fruA, a key transcription factor required for C-signaling, and that MrpB regulates mrpC expression (Nariya and Inouye, 2006). Taken together, the role of MrpB is complex and is an essential part of the development process in M. xanthus. Two EBPs with unique domain structures that affect early development are Mxa296 and Mxa213. Mxa296 is also known as Mxa210 because DNA sequence analysis of the mxa210 pLAGl (Gorski and Kaiser, 1998) plasmid insert reveals that it is identical to the central region of the mxa296 gene (Diodati and Singer, personal communication). mxa296 mutant cells are able to develop normally on TPM agar but fail to develop in submerged culture in polystyrene microtiter plates (Gorski and Kaiser, 1998). Consistent with this phenotype is the observation that 04521 is expressed on TPM agar but not in submerged culture. Interestingly, the mutant is able to express 04521 in TPM or MC7 buffered suspension (Kuner and Kaiser, 1982), suggesting that the polystyrene is somehow inhibiting development at the preaggregation stage (Gorski and Kaiser, 1998). Further investigation is needed to decipher the subtleties of this fascinating phenotype. Based on sequence analysis, Mxa296 has an unusual EBP domain structure in that it lacks a readily identifiable N-terminal sensory domain (Jelsbak et al., 2005). Only 6 of 52 activators have this

Previous Page

3. INITIATIONAND EARLYDEVELOPMENTAL EVENTS characteristic (Jelsbak et al., 2005; Diodati and Singer, personal communication; Table 5 ) . In contrast, Mxa213 is an EBP with two regulatory domains; it has both an FHA and a GAF domain at its N terminus (Jelsbak et al., 2005). As described previously, FHA domains potentially interact with threonine-phosphorylated substrates of serinehhreonine kinases (STKs) and there is a gene encoding a putative STK close to mxa213 (Jelsbak et al., 2005). GAF domains are found in cyclic GMP-specific and stimulated phosphodiesterases, adenylate cyclases, and E. coli FhlA and NifA related proteins (Studholme and Dixon, 2003). It is predicted that GAF domains may regulate signaling events via the binding of ligands such as nucleotides and small molecules (Aravind and Ponting, 1997). Insertional disruption of mxa213 causes cells to be blocked in development. On TPM agar, these mutants form large aggregates that have irregular shapes and patterns of darkening. The mxa223 mutant cells fail to sporulate, and the sporulation defect is not corrected by mixing with wild-type cells, indicating a cell autonomous defect in sporulation. Interestingly, these mutants are defective in the expression of the C-signal-dependent a 4 4 1 4 (Gorski and Kaiser, 1998). A closer examination of C-signal production in this mutant is needed to determine if it is lower than wild-type levels. Nla6 and Nla28 are EBPs that are NtrC-like response regulators with similar mutant phenotypes. In addition, the nla28 gene is downstream of a histidine kinase gene. Inactivation of nla28 or nla6 results in a short delay in aggregation, reduced production of A- and C-signals, and 50- to 500-fold fewer spores than wild-type cells, respectively (Caberoy et al., 2003). Even though these mutants are defective in the production of the early extracellular A-signal and show a defect in aggregation, more-detailed developmental characterization is needed to temporally pinpoint where the mutations first manifest themselves. RpoN and the associated EBPs that activate transcription at u54promoters are fundamental to the overall physiology of M. xanthus. Elucidating the roles of these activators can uncover the secrets to the unique essentiality of RpoN in M . xanthus. With the data from detailed mutational analyses of these activators and prospective genetic and direct binding studies, we can begin to build models of ordered networks of these regulators in M. xanthus in the near future.

SigD, the M. xanthus RpoS Homologue In E. coli, RpoS functions as a master stress-related sigma factor and is induced under a variety of general stress conditions including starvation, transition into stationary phase, and osmotic shock (Brown et al., 2002; Hengge-Aronis, 2002). The RpoS homologue,

59 SigD (MXAN2957), appears to have roles in early and late developmental gene expression (Viswanathan et al., 2006; Yoder and Kroos, 2004a, 2004b), as well as in stationary phase (Ueki and Inouye, 1998). SigD is essential for M . xanthus survival during stationary phase and for the expression of a large number of proteins that are produced or degraded under conditions of nutrient deprivation. During vegetative growth, AsigD mutants cease to grow past late exponential phase and lose viability when the cultures are plated on nutrient media after entry into stationary phase (Ueki and Inouye, 1998). In addition, two-dimensional gel electrophoresis of wild-type and AsigD mutants reveals that protein patterns are significantly different during late exponential phase in the two strains (Ueki and Inouye, 1998). Furthermore, AsigD mutants lack or have limited resistance to a variety of stresses. Mutants fail to grow after heat shock, h.ave reduced growth upon cold shock, are more sensitive to H,O,-induced oxidative stress, and fail to accumulate osmoprotective trehalose in response to osmotic shock (Ueki and Inouye, 1998). These data demonstrate the importance of SigD to overall cell viability in M. xanthus. In addition to the multiple growth phenotypes, the sigD mutant is delayed in development and defective in sporulation (Ueki and Inouye, 1998). Recently, a more detailed developmental characterization of the sigD mutant was performed, and SigD appears to play a largely positive role in regulating aspects of the developmental program (Viswanathan et al., 2006). Regulation of RpoS in E. coli is very complex and multifaceted; regulation can occur at the level of transcription, mRNA turnover, translation initiation, and proteolysis (Brown et al., 2002). In M. xanthus, based on studies with lac2 transcriptional and translational fusions, there is differential regulation of SigD at the level of transcription and translation during exponential growth and transition into stationary phase (Ueki and Inouye, 1998). The transcription of the sigD gene increases upon entry into stationary phase and then decreases during stationary phase. Conversely, the activity of the SigD translational fusion is increased during stationary phase and remains constant. This pattern of regulation appears to depend on whether cells are exposed to more-gradual nutrient-limiting conditions with accumulation of waste products (stationary phase), gradual nutrient-limiting conditions to initiate development (CF plates), or abrupt starvation conditions such as a shift to TPM liquid media (Ueki and Inouye, 1998; Viswanathan et al., 2006). When cells were plated on CF agar to induce development, the activity of both the transcriptional and translational fusions increased until

60 12 h and then decreased (Ueki and Inouye, 1998). Upon stringent starvation of cells in TPM media, sigD transcript levels fall after 20 min and stay low after 40 min (Viswanathan et al., 2006). These dissimilar patterns of expression under the three starvation conditions exemplify the fundamental physiological differences between these nutrient-restricted environments. Interestingly, in relA mutants, sigD transcript levels are substantially lower during exponential growth than in wild-type cells, but the sigD mRNA levels rise above wild-type levels after 40 min poststarvation (Viswanathan et al., 2006). This suggests that (p)ppGpp levels in the cell directly or indirectly regulate sigD transcription. SigC, a Sigma Factor and Nutrient Sensor SigC (MXAN6209)is a third sigma factor that is important for the cell’s decision to undergo development. Deletion mutants of sigC have normal vegetative growth and form fruiting bodies on CF agar, albeit slightly deformed and elongated, with wild-type sporulation (Apelian and Inouye, 1993). The sequence of SigC has homology to the heat shock sigma factor, 032, but when sigC is inactivated, heat shock proteins are still induced to wild-type levels (Ueki and Inouye, 2001). Also, in contrast to sigD mutants, AsigC mutants can produce normal amounts of trehalose upon osmotic shock and during sporulation (Apelian and Inouye, 1993). Interestingly, AsigC mutants form fruiting bodies on 0.5% CTT after 15 h and sporulate. Under these same conditions, wild-type DZFl cells formed few (9 s, and then reverse direction (Spormann and Kaiser, 1999). Among wild-type cells, only 5% of the cells reverse direction after pausing for l o s M. xanthus cells cooperate to form a three-dimensional fruiting body with an inner core that supports spore differentiation (Shimkets, 1999). During the first 10 h of development, cells exchange extracellular signals, including amino and fatty acids, which help coordinate cell movement and aggregation to produce a mound-shaped structure. Within this time period, large numbers of M. xanthus cells aggregate to form multicellular fruiting bodies, within which a subset of cells differentiate into refractile, heat-resistant spores. Motility is critical for fruiting body formation and sporulation. Nonmotile mutants, whether the result of a double mutation (one mutation in an A gene and one mutation in an S gene) or a single mutation in the mglA gene, fail to produce fruiting bodies and a full complement of mature spores and show reduced expression of developmentally regulated genes (Kroos et al., 1988). The production of heat-resistant spores is reduced about 105-foldin these mutants (Kroos et al., 1988). The role of motility in development is less clear when one examines the phenotypes of strains with single mutations in A or S genes. Mutations in most A genes, including aglZ and cglB, do not appear to affect fruiting body formation or sporulation (Yang et al., 2004). In contrast, mutations in the A-gliding genes that comprise the Tollike genes do not affect the timing or the formation of fruiting bodies but do affect the ability of M . xanthus to form heat-resistant spores (MacNeil et al., 1994b; White and Hartzell, 2000; Youderian et al., 2003). This subset of A genes encodes proteins that are predicted to form multisubunit transport complexes. Some of these proteins, including AglU and AglW, are predicted to be lipoproteins anchored to the outer membrane, whereas

DEVELOPMENT AND MOTILITY

120 others, including AglX, AglS, and AglR, are predicted to reside in the inner membrane. The sporulation phenotype of one mutant defective in a gene belonging to this group, aglU, has been studied in detail. The expression of agluincreases significantlyupon starvation-induced development (MacNeil et al., 1994b; Srinivasan et al., 2005; White and Hartzell, 2000). Vegetative rod-shaped cells of the aglU mutant differentiate into ovoid cells within the same time frame as the wildtype strain but fail to mature into refractile spheres, even with prolonged incubation, and yield only 0.01 % of the number of heat-resistant spores made by the wild-type strain. Microscopic examination of thin sections reveals that the spores formed within wild-type fruiting bodies have a thick, dense layer immediately surrounding the cytoplasmic core and an outermost layer composed of a loose, fibrous material. Although the outer, fibrous material is present in the spores formed by the aglU mutant, they have a less dense layer around the cytoplasmic core, which appears less organized than in wild-type spores (White and Harzell, 2000). This layer may contain the major spore coat proteins, proteins S and C, which are produced by the aglU mutant, but may not be localized properly. A subset of the genes required for S-motility are also required for fruiting body formation. Mutants listed in Table 2, particularly the ones defective in EPS production, are defective in fruiting body formation and sporulation, underscoring the importance of S motility for development. The role of TFP in the formation of fruiting bodies is complicated. A pilA mutant is still able to form fruiting bodies, but pilH mutants are defective in fruiting body formation (Bonner et al., 2006). Recently Lee et al. (2006)engineered a ApilA agiA double mutant, which is A-S- genetically, yet it retains social motility. Under certain conditions, M. xanthus, like P. aeruginosa (Durand et al., ZOOS), may form “pseudopili,” or alternative pili, that polymerize from a prepilin paralogue different from PilA. Like the P. aeruginosa genome, which contains multiple (four),potential prepilin genes, the M. xanthus genome has six, one of which may substitute for PilA during development.

THERE IS MUCH TO BE LEARNED ABOUT THE GENETIC BASIS AND MECHANISMS OF GLIDING Extensive genetic and genomic studies allow the identification and characterization of a large number of A and S motility genes. Significant advances in the past several years have shown that S gliding is powered by the extension and retraction of TFP. Retraction is likely

stimulated by components present in the extracellular matrix. The requirement of the extracellular matrix in retraction may explain why S-gliding is restricted to cells that are within one cell length of one another. The mechanism that allows movement of isolated (A) cells is still unclear. One model proposes that polar nozzles are used to generate thrust by secreting a polyelectrolyte from the rear pole of the cell. The wild-type cell appears to use both S and A gliding systems simultaneously, and MglA, a protein required for the function of both gliding systems, may help to coordinate the two motility systems so that the systems generate thrust in the same direction. We are just beginning to delve into the complexity of gliding motility in M. xanthus. The sequence of the M. xanthus genome reveals that many of the individual genes known to be required for gliding motility, such as pilA, have multiple paralogs. From a genetic point of view, only a few genes in the families of response regulators (28 of at least 148), TPR repeat proteins, and Tol/Ton exporters, to name a few of the gene families with known paralogs involved in gliding and very large numbers of paralogs predicted to be encoded by M. xanthus, have been analyzed. Indeed, 211. xanthus provides a unique model system in which we can dissect the roles of what appear to be partially redundant gene functions in the complex processes of motility. From a physiological point of view, M. xanthus is certain to yield novel insights into the mechanisms of cellular motility and the complex interplay between these mechanisms during the execution of its elegant program of multicellular development. We were supported by grants from the National Institutes of Health (GM.54666 to T S . and GM07.5242 to P.L.H.) and the National Science Foundation (MCB0242191 to P.L.H.).

References Aldridge, P., and K. T. Hughes. 2002. Regulation of flagellar assembly. Curr. Opin. Microbiol. 5:160-165. Behmlander, R. M., and M. Dworkin. 1991. Extracellular fibrils and contact-mediated cell interactions in Myxococcus xanthus. 1.Bacteriol. 173:7810-7821. Behmlander, R. M., and M. Dworkin. 1994. Integral proteins of the extracellular matrix fibrils of Myxococcus xanthus. J . Bacteriol. 176:6304-6311. Berg, H. C. 2005. Swarming motility: it better be wet. Curr. Biol. 15 :R599-R600. Bonner, P. J., W. P. Black, Z. Yang, and L. J. Shimkets. 2006. FibA and PilA act cooperatively during fruiting body formation of Myxococcus xanthus. Mol. Microbiol. 61: 1283-1 293. Bouveret, E., H. Benedetti, A. Rigal, E. Loret, and C. Lazdunski. 1999. In vitro characterization of peptidoglycanassociated lipoprotein (PAL)-peptidoglycan and PAL-TolB interactions. 1.Bacteriol. 15:6306-6311.

MOTILITY OF M. XANTHUS 6. GLIDING Bouveret, E., R. Derouiche, A. Rigal, R. Lloubes, C. Lazdunski, and H. Benedetti. 1995. Peptidoglycan-associated lipoprotein-TolB interaction. A possible key to explaining the formation of contact sites between the inner and outer membranes of Escherichia coli. J. Biol. Chem. 270: 11071-1 1077. Bowden, M. G., and H. B. Kaplan. 1998. The Myxococcus xanthus lipopolysaccharide 0-antigen is required for social motility and multicellular development. Mol. Microbiol. 30~275-284. Burchard, R. P. 1981. Gliding motility of prokaryotes: ultrastructure, physiology, and genetics. Annu. Rev. Microbiol. 35:497-529. Burrows, L. L. 2005. Weapons of mass retraction. Mol. Microbiol. 57: 8 78-8 8 8. Caberoy, N. B., R. D. Welch, J. S. Jakobsen, S. C. Slater, and A. G. Garza. 2003. Global mutational analysis of NtrC-like activators in Myxococcus xanthus: identifying activator mutants defective for motility and fruiting body development. J. Bacteriol. 185:6083-6094. Durand, E., G. Michel, R. Voulhoux, J. Kurner, A. Bernadac, and A. Filloux. 2005. XcpX controls biogenesis of the Pseudomonas aeruginosa XcpT-containing pseudopilus. J. Biol. Chem. 280:3 1378-3 1 389. Fontes, M., and D. Kaiser. 1999. Myxococcus cells respond to elastic forces in their substrate. Proc. Natl. Acad. Sci. USA 96~8052-8057. Guo, D., M. G. Bowden, R. Pershad, and H. B. Kaplan. 1996. The Myxococcus xanthus rfbABC operon encodes an ATPbinding cassette transporter homolog required for O-antigen biosynthesis and multicellular development. J. Bacteriol. 178~1631-1639. Hartzell, P., and D. Kaiser. 1991a. Function of MglA, a 22kilodalton protein essential for gliding in Myxococcus xanthus. J. Bacteriol. 173:7615-7624. Hartzell, P. L. 1997. Complementation of Myxococcus xanthus sporulation and motility defects by a eukaryotic RAS homolog. Proc. Natl. Acad. Sci. USA 979881-9886. Hartzell, l?L., and D. Kaiser. 1991b. Upstream gene of the mgl operon controls the level of MglA protein in M. xanthus. J. Bacteriol. 172:7625-7635. Hodgkin, J., and D. Kaiser. 1977. Cell-to-cell stimulation of movements in non-motile mutants of Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 74:2938-2942. Hodgkin, J., and D. Kaiser. 1979a. Genetics of gliding motility in Myxococcus xanthus (Myxobacterales): genes controlling movement of single cells. Mol. Gen. Genet. 171:167-171. Hodgkin, J., and D. Kaiser. 1979b. Genetics of gliding motility in Myxococcus xanthus (Myxobactererales):two gene systems control movement. Mol. Gen. Genet. 171:177-191. Hoiczyk, E. 2000. Gliding motility in cyanobacteria: observations and possible explanations. Arch. Microbiol. 174: 11-17. Hoiczyk, E. 1998. Structural and biochemical analysis of the sheath of Phormidium uncinatum. J. Bacteriol. 180:39233932. Hoiczyk, E., and W. Baumeister. 1997. Oscillin, an extracellular, Ca2+-bindingglycoprotein essential for the gliding motility of cyanobacteria. Mol. Microbiol. 26:699-708.

121 Kaiser, D., and M. Dworkin. 1975. Gene transfer to myxobacterium by Escherichia coli phage P1. Science 182653-654. Kearns, D. B., P. J. Bonner, D. R. Smith, and L. J. Shimkets. 2002. An extracellular matrix-associated zinc metalloprotease is required for dilauroyl phosphatidylethanolamine chemotactic excitation in Myxococcus xanthus. J. Bacteriol. 184:1678-1684. Kim, S. H., S. Ramaswamy, and J. Downard. 1999. Regulated exopolysaccharide production in Myxococcus xanthus. J . Bacteriol. 181:1496-1 507. Kroos, L., P. Hartzell, K. Stephens, and D. Kaiser. 1988. A link between cell movement and gene expression argues that motility is required for cell-cell signaling during fruiting body development. Genes Dev. 2:1677-1685. Kroos, L., and D. Kaiser. 1984. Construction of Tn5-lac, a transposon that fuses lacZ expression to exogenous promoters, and its introduction into Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 815816-5820. Kroos, L., A. Kuspa, and D. Kaiser. 1986. A global analysis of developmentally regulated genes in Myxococcus xanthus. Dev. Biol. 117252-266. Lancero, H., N. B. Caberoy, S. Castaneda, Y. Li, A. Lu, D. Dutton, X. Y. Duan, H. B. Kaplan, W. Shi, and A. G. Garza. 2004. Characterization of a Myxococcus xanthus mutant that is defective for adventurous motility and social motility. Microbiology 150:4085-4093. Lazzaroni, J. C., P. Germon, M. C. Ray, and A. Vianney. 1999. The To1 proteins of Escherichia coli and their involvement in the uptake of biomolecules and outer membrane stability. FEMS Microbiol. Lett. 15:191-197. Lee, C., J. Chung, J. Kim, and K. Cho. 2006. Identification of a gene required for gliding motility in Myxococcus xanthus. J. Microbiol. Biotechnol. 16:771-777. Li, Y., R. Lux, A. E. Pelling, J. K. Gimzewski, and W. Shi. 2005. Analysis of type IV pilus and its associated motility in Myxococcus xanthus using an antibody reactive with native pilin and pili. Microbiology 151:353-360. Li, Y., H. Sun, X. Ma, A. Lu, R. Lux, D. Zusman, and W. Shi. 2003. Extracellular polysaccharides mediate pilus retraction during social motility of Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 100:5443-5448. Limberger, R. J. 2004. The periplasmic flagellum of spirochetes. J. Mol. Microbiol. Biotechnol. E30-40. Lu, A., K. Cho, W. P. Black, X. Y. Duan, R. Lux, Z. Yang, H. B. Kaplan, D. R. Zusman, and W. Shi. 2005. Exopolysaccharide biosynthesis genes required for social motility in Myxococcus xanthus. Mol. Microbiol. 55:206-220. MacNeil, S. D., F. Calara, and P. L. Hartzell. 1994a. New clusters of genes required for gliding motility in Myxococcus xanthus. Mol. Microbiol. 14:61-71. MacNeil, S. D., A. Mouzeyan, and l? L. Hartzell. 1994b. Genes required for both gliding motility and development in Myxococcus xanthus. Mol. Microbiol. 14:785-795. Mignot, T., J. P. Merlie, Jr., and D. R. Zusman. 2005. Regulated pole-to-pole oscillations of a bacterial gliding motility protein. Science 310:855-85 7. Nudleman, E., and D. Kaiser. 2004. Pulling together with type IV pili. J. Mol. Microbiol. Biotechnol. 752-62.

122 Nudleman, E., D. Wall, and D. Kaiser. 2005. Cell-to-cell transfer of bacterial outer membrane lipoproteins. Science 309:125-127. Nudleman, E., D. Wall, and D. Kaiser. 2006. Polar assembly of the type IV pilus secretin in Myxococcus xanthus. Mol. Microbiol. 60:16-29. Peabody, C. R., Y. J. Chung, M. R. Yen, D. Vidal-Ingigliardi, A. P. Pugsley, and M. H. Saier, Jr. 2003. Type I1 protein secretion and its relationship to bacterial type IV pili and archaeal flagella. Microbiology 149:3051-3072. Pham, V. D., C. W. Shebelut, B. Mukherjee, and M. Singer. 2005. RasA is required for Myxococcus xanthus development and social motility. J. Bacteriol. 187:6845-6848. Ponting, C. P., and M. J. Pallen. 1999. A p-propeller domain within TolB. Mol. Microbiol. 31:739-740. Rodriguez, A. M., and A. M. Spormann. 1999. Genetic and molecular analysis of cglB, a gene essential for single-cellgliding in Myxococcus xanthus. J. Bacteriol. 181:4381-4390. Rodriguez-Soto, J. P., and D. Kaiser. 1997. The tgl gene: social motility and stimulation in Myxococcus xanthus. J. Bacteriol. 179:4361-4371. Rosenbluh, A., and M. Eisenbach. 1992. Effect of mechanical removal of pili on gliding motility of Myxococcus xanthus. J. Bacteriol. 1745406-54 13. Rubin, E. J., B. J. Akerley, V. N. No&, D. J. Lampe, R. N. Husson, and J. J. Mekalanos. 1999. In vivo transposition of marinerbased elements in enteric bacteria and mycobacteria. Proc. Natl. Acad. Sci. USA 96:1645-1650. Shi, W., and D. R. Zusman. 1993. The two motility systems of Myxococcus xanthus show different selective advantages on various surfaces. Proc. Natl. Acad. Sci. USA 90:33783382. Shimkets, L. J. 1999. Intercellular signaling during fruitingbody development of Myxococcus xanthus. Annu. Rev. Microbiol. 53:525-549. Spormann, A. M., and A. D. Kaiser. 1995. Gliding movements in Myxococcus xanthus. J. Bacteriol. 1775846-5852. Spormann, A. M., and D. Kaiser. 1999. Gliding mutants of Myxococcus xanthus with high reversal frequencies and small displacements. J. Bacteriol. 181:2593-2601. Srinivasan, B. S., N. B. Caberoy, G. Suen, R. G. Taylor, R. Shah, F. Tengra, B. S. Goldman, A. G. Garza, and R. D. Welch. 2005. Functional genome annotation through phylogenomic mapping. Nat. Biotechnol. 23:691-698. Stephens, K., P. L. Hartzell, and D. Kaiser. 1989. Gliding motility in Myxococcus xanthus: mgl locus, RNA, and predicted protein products. J. Bacteriol. 171:819-830. Stephens, K., and D. Kaiser. 1987. Genetics of gliding in Myxococcus xanthus: molecular cloning of the mgl locus. Mol. Gen. Genet. 207:256-266. Sun, H., D. R. Zusman, and W. Shi. 2000. Type IV pilus of Myxococcus xanthus is a motility apparatus controlled by the frz chemosensory system. Curr. Biol. 10:1143-1146. Thomasson, B., J. Link, A. G. Stassinopoulos, N. Burke, L. Plamann, and P. L. Hartzell. 2002. The GTPase, MglA, interacts with a tyrosine kinase to control type-IV

DEVELOPMENT AND MOTILITY pili-mediated motility of Myxococcus xanthus. Mol. Microbiol. 46: 1399-14 13. Ueki, T., and S. Inouye. 1998. A new sigma factor, SigD, essential for stationary phase is also required for multicellular differentiation in Myxococcus xanthus. Genes Cells 3:371-3 85. Wall, D., P. E. Kolenbrander, and D. Kaiser. 1999. The Myxococcus xanthus pilQ (sglA) gene encodes a secretin homolog required for type IV pilus biogenesis, social motility and development. J. Bacteriol. 181:24-33. Ward, M. J., H. Lew, and D. R. Zusman. 2000. Social motility in Myxococcus xanthus requires FrzS, a protein with an extensive coiled-coil domain. Mol. Microbiol. 37:13571371. Weimer, R. M., C. Creighton, A. Stassinopoulos, P. Youderian, and P. L. Hartzell. 1998. A chaperone in the HSP70 family controls production of extracellular fibrils in Myxococcus xanthus. J. Bacteriol. 1805357-5368. White, D. J., and P. L. Hartzell. 2000. AglU, a protein required for gliding motility and spore maturation of Myxococcus xanthus, is related to WD-repeat proteins. Mol. Microbiol. 36:662-678. Wolgemuth, C., E. Hoiczyk, D. Kaiser, and G. Oster. 2002. How myxobacteria glide. Curr. Biol. 12:369-377. Wu, S. S., and D. Kaiser. 1995. Genetic and functional evidence that Type IV pili are required for social gliding motility in Myxococcus xanthus. Mol. Microbiol. 18547-558. Wu, S. S., and D. Kaiser. 1996. Markerless deletions of pi1 genes in Myxococcus xanthus generated by counterselection with the Bacillus subtilis sacB gene. J. Bacteriol. 17858175821. Wu, S. S., and D. Kaiser. 1997. Regulation of expression of the pilA gene in Myxococcus xanthus. J. Bacteriol. 179:77487758. Wu, S. S., J. Wu, and D. Kaiser. 1997. The Myxococcus xanthus pilT locus is required for social gliding motility although pili are still produced. Mol. Microbiol. 23:109-121. Yang, R., S. Bartle, R. Otto, A. Stassinopoulos, M. Rogers, L. Plamann, and P. Hartzell. 2004. AglZ is a filament-forming coiled-coil protein required for adventurous gliding motility of Myxococcus xanthus. J. Bacteriol. 186:6168-6178. Yang, Z., Y. Geng, D. Xu, H. B. Kaplan, and W. Shi. 1998. A new set of chemotaxis homologues is essential for Myxococcus xanthus social motility. Mol. Microbiol. 30:1123-1130. Yang, Z., X. Ma, L. Tong, H. B. Kaplan, L. J. Shimkets, and W. Shi. 2000. Myxococcus xanthus dif genes are required for biogenesis of cell surface fibrils essential for social gliding motility. J. Bacteriol. 1825793-5798. Youderian, P., N. Burke, D. J. White, and P. L. Hartzell. 2003. Identification of genes required for adventurous gliding motility in Myxococcus xanthus with the transposable element mariner. Mol. Microbiol. 49:555570. Youderian, P., and P. L. Hartzell. 2006. Transposon insertions of magellan-4 that impair social gliding motility in Myxococcus xanthus. Genetics 172:1 397-14 10.

Myxobacteriu: Multicellularity and Differentiation Edited by David E. Whitworth 0 ZOOS ASM Press, Washington, D.C.

David R. Zusman Yuki F. I n c h Tiim Mignot

7

The Frz Chemosensory System of Myxococcus xanthus

MYXOCOCCUS XANTHUS EXHIBITS MANY SOCIAL BEHAVIORS Myxococcus xanthus has attracted much scientific interest because of its complex life cycle and morphogenetic potential. The bacteria grow in nature on complex organic material or prey upon other microorganisms. 111. xanthus cells are generally found in large groups (swarms) or biofilms. This social behavior facilitates predation and food gathering as large numbers of bacteria cooperate by producing antibiotics and digestive enzymes. When M. xanthus swarms are unable to find sufficient nutrients, they enter a developmental pathway in which they aggregate in a coordinated manner, forming raised pigmented mounds, 0.1 to 0.2 mm in height. Within the mounds, termed fruiting bodies, the cells differentiate to form spores. While the large majority of cells (80 to 90%)aggregate to form fruiting bodies, some cells follow a different developmental fate. These cells, called peripheral rods, remain as a monolayer of rod-shaped cells around and between fruiting bodies (O’Connor and Zusman, 1991). These cells do not aggregate or sporulate unless they are harvested and resuspended at high cell concentration on a fresh substrate. The peripheral rods move backwards and forwards in a rhythmic manner, forming “accordion waves” (Sliusarenko et al., 2006). The peripheral rods

/

have been hypothesized to be resting vegetative cells or scout cells, ready to feed if food or prey becomes available; spores are resting cells that cannot search for prey. The complex life cycle of 111. xantbus makes it an excellent bacterial system to study directed cell movements. The social behaviors of M . xanthus depend on cell motility. Individual bacterial cells move very slowly by gliding motility, about 2 to 4 p d m i n . While this slow rate of movement may be a disadvantage for cell dispersal, it may be advantageous for cell feeding, as it ensures that cells do not outrun their extracellular enzymes or their intercellular signals. Indeed, cells at the leading edge of a vegetative swarm venture outward but very quickly reverse direction, returning to the swarm (Reichenbach, 1999). Gliding motility is traditionally described as movement in the direction of the long axis of the cell at a solid-liquid or a solid-air interface without the aid of flagella (McBride, 2001). M . xanthus has two systems for gliding (Hodgkin and Kaiser, 1979).The first system is called adventurous (A)-motility and involves the movement of individual cells. A-motility is still not well understood, although many A-motility mutants have been isolated, and it is thought to require slime secretion (Youderian et al., 2003; Yu and Kaiser, 2007). Wolgemuth et al. hypothesized that directed extrusion of the slime, a polyelectrolyte gel, could

David R. Zusman, Yuki F. I n c h , and Tiim Mignot, Department of Molecular and Cell Biology, University of California, Berkeley, CA 94720-3204.

123

DEVELOPMENT AND MOTILITY

124 generate enough force to move cells forward (Wolgemuth et al., 2002),although recently Mignot et al. (2007)found evidence that suggests that A-motility is powered by the movement of specific adhesion complexes that track along helical cytoskeletal filaments. The second system is called social (S)-motility and involves the movement of cells in groups. S-motility requires type IV pili (Sun et al., 2000), lipopolysaccharide (LPS) O-antigen (Bowden and Kaplan, 1998), and extracellular matrix polysaccharide (EPS), which is a component of fibrils (Arnold and Shimkets, 1988). S-motility is similar to twitching motility in Pseudomonas aeruginosa and is powered by the extension and retraction of type IV pili. The pili are long, thin fibers that may be as long as 10 cell lengths (40 to 50 pm). They are extruded from one cell pole, where they adhere

to EPS or complex polysaccharides on another cell, the slime trail, or a prey microorganism. Retraction of the pili then pulls the cell in the direction of the adhering pili (Li et al., 2003; Sun et al., 2000).

THE frx GENES CONTROL CELLULAR REVERSALS The frz (frizzy) genes were discovered as part of a search for new mutants defective in cellular aggregation (Zusman, 1982). These mutants were defective at vegetative swarming and failed to aggregate into discrete mounds but instead formed “frizzy” filaments on starvation agar (Fig. 1).The first clue to the function of the frz genes came from observations of the motility pattern of single

A. Vegetative swarming on CYE agar (0.3%) Wild type, strain DZ2

Afrz (Z, A, B, CD, E or F)

B. Aggregation and development on CF starvation agar (1.5%) Wild type, strain DZ2

Afrz (Z, A, B, CD, E or F)

Figure 1 Developmental aggregation and vegetative swarming phenotypes of wild-type and frzCD mutants. Cells from M . xanthus wild-type strain DZ2 and Afrz ( Z ,A, B , CD, E , or F ) mutants were concentrated to 4 x l o 9 CFU ml-’ and spotted on CYE media containing 0.3% agar to analyze swarming or on CF agar (1.5%)to analyze fruiting body formation. The plates were examined under a dissecting microscope after 72 h of incubation at 32°C. The figure is modified from photographs published by Bustamante et al., 2004.

7. FRZ CHEMOSENSORY SYSTEMOF M.XANTHUS cells. When M. xanthus cells glide on an agar surface, cells reverse their direction of movement approximately every 7 to 8 min; net movement occurs since the interval between reversals can vary widely. The frz genes control the frequency at which cells reverse their direction of movement. For example, most frz mutants very rarely reverse direction; in contrast, some mutants in the receptor frzCD reverse much more frequently than the wild type, approximately every 2 min, and individual cells show no net movement (Blackhart and Zusman, 1985). These behavioral patterns suggested that the frz mutants might be similar to enteric chemotaxis mutants, as control of cellular reversals could affect directional movements. For example, enteric bacteria are known to direct their movements by undergoing a biased random walk. When the flagella rotate counterclockwise, the flagella form lefthanded helical bundles and the cells are pushed forward in a “run.” In contrast, when the flagella rotate clockwise, the flagellar bundles disperse and cells “tumble.” Tumbling results in randomly reorienting the bacteria. Thus, directed movements are achieved in these bacteria by controlling the run and tumble intervals. M. xanthus, in contrast, does not contain flagella and is nonmotile in a liquid medium. On a solid surface, cells predominantly move in existing slime trails, mostly in two dimensions. Since cells are flexible (individual cells can be seen bending), they require periodic directional corrections. Regulated cell reversals are proposed to be required for M. xanthus cells to undergo directed motility.

Frz PROTEINS SHOW SEQUENCE SIMILARITIESTO CHEMOTAXIS PROTEINS When the frz genes were sequenced, strong similaritieswere found between the Frz proteins and the major chemotaxis proteins of enteric bacteria (Fig. 2) (McBride et al., 1989).

~,

CheY-CheY

regulator

Chew

Chew

MCP

~~~~

125 In these bacteria, signals are recognized by chemoreceptors termed methyl-accepting chemotaxis proteins (MCPs). Receptors stimulate the activity of a histidine protein kinase (CheA)through interaction with a coupling protein, Chew (West and Stock, 2001). Typically an active CheA autophosphorylates and transfers a phosphoryl group to a single domain response regulator protein called CheY. In most flagellated bacteria, a change in direction is induced when phosphorylated CheY interacts with a switch component of the flagellar motor. The Frz system consists of FrzCD, a cytoplasmic chemoreceptor; FrzA and FrzB, Chew homologues; FrzE, a CheA-CheY-like fusion protein; FrzF, a methyltransferase; and FrzG, a methylesterase (McBride et al., 1989). FrzZ consists of two CheY-like domains connected by a linker region. frzZ is located 5’ to the frz operon but transcribed in the opposite orientation (Trudeau et al., 1996). Analysis of mutants containing inframe deletions showed that FrzCD (MCP), FrzA (Chew), and the CheA domain of FrzE constitute the core components of the Frz pathway, as they are essential for vegetative swarming, responses to repellents, and directed movement during development. FrzB (Chew), FrzF (CheR), FrzG (CheB),the CheY domain of FrzE, and FrzZ (CheY-CheY) are required for some but not all responses. Based on the Escherzchza coli paradigm, active FrzE should stimulate cellular reversals and inactive FrzE should inhibit cellular reversals (Ward and Zusman, 1990).

METHYLATION OF FrzCD IS CORRELATED WITH FACTORS AFFECTING CELL BEHAVIOR In enteric bacteria, methylation of receptors is required for adaptation to stimuli. Thus, the level of methylation of an MCP increases following the addition of an attractant and decreases following the addition of a

CheA-CheY

Response-

CheB

CheR

transferase

regulator

Figure 2 The frz operon contains genes that are homologous to proteins encoded by che genes from the enteric bacteria.

126 repellent. Methylated FrzCD, like the enteric MCPs, migrates as a ladder of bands that varies with the level of methylation during polyacrylamide gel electrophoresis. These MCP bands can be detected by Western immunoblot analysis using anti-FrzCD antibodies; FrzCD appears as multiple bands corresponding to the unmethylated (amidated and deamidated) and methylated forms of the receptor. Western blot analysis showed that vegetative cells are highly methylated (about 50% of FrzCD is methylated) but cells that are starved are relatively unmethylated (McBride et al., 1989). Developmental cells show an initial loss in methylation followed by an increase in the level of methylation so that by 72 h of development, when cells are mostly in fruiting bodies, FrzCD is about 70% methylated. The methylation changes in FrzCD suggest that during aggregation, FrzCD senses a chemical(s) produced by other cells that promotes cell movements towards aggregation centers (Geng et al., 1998). Geng et al. (1998) found that developmental mutants could be divided into two groups based on the level of FrzCD methylation: nonaggregating or abnormally aggregating mutants, including the asg, bsg, csg, and esg mutants, showed poor FrzCD methylation. Mutants blocked in late development and sporulation showed normal FrzCD methylation. Thus, the methylation of FrzCD defines a discrete step in the developmental program of M . xanthus. Indeed, Ssgaard-Andersen and Kaiser (1996) found that the csgA mutant, which does not show FrzCD methylation during development, can be stimulated to methylate FrzCD when cells are treated with C-factor and rescued for development. In an attempt to identify chemicals or nutrients that might be potential attractants or repellents for M. xanthus, numerous substances were tested to determine whether they affected the methylation state of FrzCD. Although the methylation of FrzCD was stimulated by the addition of rich media containing peptides (Casitone yeast extract [CYE] medium), yeast extract, or several defined chemicals including lauric acid and lauryl alcohol, individual amino acids, sugars, or nucleotides had no effect. Surprisingly, some phospholipids such as phosphatidyl ethanol did stimulate methylation of FrzCD (McBride et al., 1992). In contrast, several short-chain alcohols and dimethyl sulfoxide caused the demethylation of FrzCD.

DOES M. XANTHUS EXHIBIT CHEMOTAXIS ? The complex movements and behavior of M. xanthus suggested chemotaxis, but demonstrating chemotaxis and defining chemoeffectors were difficult to

DEVELOPMENT AND MOTILITY establish experimentally. For example, Dworkin and Eide (1983)tried but were unable to show chemotaxis of M . xanthus to a wide variety of potential chemoeffectors and suggested that chemotaxis may not be possible in M. xanthus since cells move so slowly, slower than the rate of diffusion of some small molecules. However, the rate of diffusion of peptides in agar is much slower than in buffer, even slower than the rate of movement of M. xanthus cells (R. Welch, personal communication). Shi et al. (1993) employed chambered petri dishes to establish sharp chemical gradients and found directed movement of M. xanthus cells towards yeast extract and Casitone and away from dimethyl sulfoxide and isoamyl alcohol. Furthermore, these movements were completely dependent on the Frz system, supporting the hypothesis that the Frz chemosensory system controls directed movements in this organism. Chemotaxis in M. xanthus was also documented by Kearns and Shimkets, who found that dilauroyl and dioleoyl phosphatidylethanolamine (PE) stimulated directed cell movement in a gradient on an agar surface (Kearns et al., 2002). They used changes in reversal period to show that M . xanthus responds to and adapts to these lipids. Unexpectedly, stimulation was not dependent on the Frz pathway although adaptation was dependent onthe pathway. They noted that directed movements towards these lipids were indeed chemotaxis since (i) biased movements were correlated with the suppression of reversals to achieve longer runs when exposed to PE, (ii)the responses were specific to PE molecules with particular fatty acids, and (iii) cells showed adaptation to the lipid attractants.

THE frz GENES REGULATE BOTH THE A- AND S-MOTILITY SYSTEMS Although it was known for many years that the f y z genes were required for normal aggregation during fruiting body formation, it was not known if they controlled the Amotility system, the S-motility system, or both. To investigate this point, the swarming ability of cells containing a single motility system was studied (V. H. Bustamante and D. R. Zusman, unpublished data). An A-S+ mutant (aglBI),which can move only by S-motility, swarms as well as the wild type on 0.3% CYE agar, but swarming is reduced and disorganized in an aglB2 frzCD double mutant. Thus, the Frz system is required for the organized spreading of M. xanthus colonies. In contrast, an A+S-mutant (pilA),which can move only by A-motility, fails to swarms at all on 1.5% CYE agar. However, a pilA fyzCD double mutant shows restored movement and spreads at about the same rate as an frzCD mutant. Thus, the Frz system inhibits the movement of A-motile

SYSTEMOF 111. XANTHUS 7. FRZ CHEMOSENSORY cells on rich media; it should be noted that this inhibition is nutrient dependent. The suppression of A-motility by the Frz system remains to be investigated.

THE f k z GENES REGULATE S-MOTILITY-DEPENDENT REVERSALS S-motility in M . xanthus, like twitching motility in P. aeruginosa, has been shown to involve the extension and retraction of type IV pili localized at the leading cell pole. The pili are extended from the cell pole, where they adhere to exopolysaccharides on the surface of another cell or the cell surface (slime trails) (Li et al., 2003). This interaction triggers pilus retraction, which pulls the cells forward in the direction of the adhering pili. Cellular reversals must therefore result from the sites of pilus extension switching from one cell pole to another. This switching is controlled by the f ~ chemosensory z system. Sun et al. (2000) developed an assay allowing single cells to move using only S-motility. Cells were placed on microscope slides and overlaid with 1% methylcellulose medium. Under these conditions, the pili presumably bind to the surface of slides coated with methylcellulose. The binding of pili to the slides “tether” the cells (they appear to be spherical instead of rod-shaped, from an end-on view). These tethered cells were followed by timelapse videomicroscopy. The cells remained bound to the surface for about 8 min, after which they were released, which corresponds to the reversal interval of wild-type cells when gliding on a solid surface. frzA-F mutants, which rarely reverse on an agar surface, remained tethered for extended periods of time; in contrast, constitutive signaling mutants, like the frzCD::TnSa224 mutant, remained tethered for only 2 min, which corresponds to the reversal period for these cells on a solid surface. The behavior of the tethered cells is consistent with the pili being extruded from one cell pole, adhering to a surface, and then retracting, pulling the cell in the direction of the adhering pili. This process is controlled by the frz chemosensory system.

ANALYSIS OF THE CYTOPLASMIC RECEPTOR f.xCD FrzCD, the Frz system chemoreceptor, contains a conserved C-terminal module present in MCPs; but in contrast to most MCPs, FrzCD is localized in the cytoplasm. Immunofluorescence and deconvolution microscopy showed that it is localized as an array of discrete clusters (D. P. Astling, E. M. F. Mauriello, and D. R. Zusman, unpublished data). Since the N-terminal region of FrzCD does not contain the canonical periplasmic

127

domain, a series of in-frame deletion mutants were constructed in fyzCD to determine the function of the various domains of FrzCD. Surprisingly, deletion of the N-terminal region of FrzCD (codons 6 to 130) showed only minor defects in swarming, and development was normal. Thus, the N-terminal region of FrzCD probably is not directly involved in sensing signals: signal input to the Frz system must be sensed by the conserved Cterminal module of FrzCD (Bowden and Kaplan, 1998). Perhaps the methyltransferase (FrzF) may be regulated to recognize different methylation sites in the C-terminal module of FrzCD. Alternatively, FrzCD may interact with another unidentified protein or MCP that could transduce a signal. Interestingly, deletion of about 25 amino acids from either end of the conserved C-terminal region of FrzCD resulted in a constitutive signaling state of FrzCD, which induces hyperreversals with no net cell movement. These deletions may result in FrzCD locked in a constitutively signaling conformation or deleted regions may contain potential interaction sites. Since the mechanisms governing methylation of a chemoreceptor can differ and could potentially provide a mechanism for differential responses by the bacteria to different stimuli, methylation site mutants of FrzCD were constructed and analyzed (Astling et al., 2006). For this study, potential methylation sites of FrzCD were systematically modified by site-directed mutagenesis, changing glutamine/glutamate pairs to alanines. Two of the sites, when mutated, had a stimulatory effect on the pathway (causing constitutive signaling), as evidenced by cells hyperreversing. In contrast, two other sites, when mutated, had an inhibitory effect on the pathway, causing cells to rarely reverse. This indicates that the methyltransferase can both activate and inhibit the Frz pathway, depending on which sites are modified by methylation. The stimulatory mutations blocked both vegetative swarming and developmental aggregation. The inhibitory mutations blocked developmental aggregation at low cell density, but not at high cell density, suggesting that specific methylation sites may be required for sensing low concentrations of developmental signals. The different phenotypes of the mutants observed in this study suggest that differential methylation could provide a potential signal input to the Frz chemosensory pathway.

ROLE OF THE METHYLTRANSFERASE FrzF I N FrzCD METHYLATION If the N-terminal domain of FrzCD is not required for sensing signals, how are these signals detected? Is regulated methylation/demethylation important for responses? To

128 address these questions, in-frame deletion mutants of frzG, which encodes a methylesterase, and frzF, which encodes a methyltransferase, were constructed. The frzG deletion mutant had only a minor impact on swarming and development. In contrast, the frzF mutant was defective in development (forming frizzy filaments) and in vegetative swarming. FrzF, unlike CheR from E. coli, is a large protein that contains three tetratricopeptide repeats (TPR) motifs, which are typically involved in protein-protein interactions. To see if the repeats were important for the regulation of methyltransferase activity, an in-frame deletion mutant lacking the TPR domains was constructed (I. Martinez-Flores, V. H. Bustamante, A. E. Scott, and D. R. Zusman, unpublished data). The mutant retained the ability to methylate FrzCD, albeit at a reduced level, but was unable to form fruiting bodies. However, the frzF,,, mutant was able to swarm on rich media. This indicates the importance of the TPR motifs for regulating FrzCD methylation during developmental aggregation. Proteins that might interact with the TPR domains are currently being investigated.

REGULATION OF THE Frz PATHWAY BY A NOVEL CheW-LIKE PROTEIN, FrzB In E. coli, the main apparent role of Chew is to facilitate the interactions of the receptor with the kinase, CheA. The Frz pathway has two Chew homologues, FrzA and FrzB. FrzB is a CheW-like protein with a novel regulatory role. Mutations in frzB give a phenotype similar to that of frzA-E mutants with respect to vegetative swarming and developmental aggregation; however, frzB mutants can still respond to repellents. This suggests that FrzB is not a part of the core Frz signal transduction pathway but acts as an accessory factor or regulator of the pathway, like FrzF and FrzG. We examined the interaction of both FrzA and FrzB with FrzCD and FrzE using the yeast two-hybrid interaction assay and pull-down assays with purified proteins (Astling, 2003). Using these assays, we found that FrzA can interact with both the receptor, FrzCD, and the CheA, FrzE. In addition, FrzA and FrzCD stimulate the kinase activity of FrzE in vitro. Thus, FrzA is a true Chew homolog. FrzB, on the other hand, interacted with FrzCD but not with FrzE, which is consistent with the observation that it lacks the CheAbinding domain. We used surface plasmon resonance spectroscopy to examine the kinetics of the interaction of FrzA and FrzB with FrzCD (D. P. Astling and D. R. Zusman, unpublished data). From this analysis, we obtained the kinetic parameters of both binding events. The data best fit a bivalent model where one receptor dimer can bind two molecules of FrzA or FrzB. FrzA has a greater

DEVELOPMENT AND MOTILITY affinity for FrzCD than does FrzB. We hypothesize that FrzB plays a role in receptor clustering or alternatively, in coupling the receptor to additional proteins. To test the latter hypothesis, we used the yeast two-hybrid interaction assay to search for proteins that interact with FrzB. We found several interacting proteins, two of which were MCPs. This is relevant because M. xanthus has 21 putative MCPs and eight CheA proteins. The significance of these interactions is currently being investigated.

ANALYSIS OF FrzE, A CheA-CheY HYBRID The Frz system regulates reversal frequency in both the A- and S-motility systems. How does a single chemotaxis-like pathway regulate coordinated signaling to two disparate motility systems? FrzE, a CheA-CheY fusion, plays a large role in this regulation. FrzE is essential for Frz signaling as shown by genetic studies (Blackhart and Zusman, 1985). The CheA domain of FrzE ( FrzEcheA)autophosphorylates and transfers the phosphate to downstream components, ultimately causing a reversal. Recent studies revealed that the activity of FrZECheAis regulated by the CheY domain of FrzE. Genetic analyses showed that the state of the CheY domain can independently influence the reversal period of the A- and S-motility systems (Bonner et al., 2005), and in vitro studies showed that this regulation occurs by inhibiting autophosphorylation activity of FrzECheA (Y. F. I n c h and D. R. Zusman, unpublished data) as described below. A deletion mutant of frzE, AfrzE, is unable to coordinate cellular movements and results in the typical Frz phenotype: reduced cellular reversals, decreased swarming, and frizzy aggregates on developmental media (Bustamante et al., 2004). However, a deletion of the CheY domain of FrzE, frzEAChcY,displayed differential cellular reversal periods with respect to A- and S-motility, defective vegetative swarming, and surprisingly, the formation of fruiting bodies on developmental media (Bonner et al., 2005). To further investigate the function of the CheY domain, point mutations of the conserved phospho-accepting aspartate residue in the CheY domain of FrzE (FrzEchey)were constructed such that the encoded proteins potentially mimic the constitutively active (FrzE-D709E) or constitutively inactive (FrzE-D709A)conformation observed for some response regulators (Klose et al., 1993). Single-cell motility analysis revealed that both frzE-D709A and frzE-D709E hyperreverse with respect to the A-motility system (Bonner et al., 2005). Because FrzECheycannot regulate FrzE autophosphorylation activity in these mutants, FrZEcheA

7. FRZ CHEMOSENSORY SYSTEMOF M.XANTHUS is most likely phosphorylated at increased levels compared to the wild type. Thus, phospho-signaling to the A-motility system is elevated, resulting in the observed hyperreversing phenotypes. Conversely, the frzE-D709A mutant displayed hyperreversals with respect to the Smotility system in contrast to the frzE-D709E mutant, which displayed hyporeversals. It was hypothesized that FrzECheydirectly regulates reversals of the S-modity system by a physical interaction that is enhanced in the frzE-D709A mutant. However, the double mutant frzEAChey AfrzZ hyporeversed with respect to both motility systems, suggesting that FrzZ regulates both systems downstream of FrzE. Further information as to how the CheY domain affects activity of the CheA domain came from in vitro analyses. To complement the genetic analyses, the Frz system was reconstituted in vitro to test biochemical activities by purifying FrzCD, FrzA, FrzE, and the independent domains of FrzE: FrzECheA and FrZECheY (Inclhn et al., 2007). FrzECheA autophosphorylated in the presence of FrzCD, FrzA, and ATP. FrzCD and FrzA were required for significant autophosphorylation. Surprisingly, the full-length FrzE protein did not autophosphorylate in the presence or absence of FrzCD and FrzA, suggesting that the CheY domain of FrzE inhibits autophosphorylation activity of the CheA domain. These genetic and biochemical data suggest that FrzE has divergent signaling effects on both the A- and S-motility systems and that FrzECheynegatively regulates FrzE autophosphorylation. It is unclear how this regulation is accomplished in vivo, and this is currently under investigation.

THE Frz PATHWAY OUTPUT The output pathway from the Frz system to the motility engines is still unknown. We predicted that additional unidentified proteins interact with FrzE and relay signals from the FrzCD, FrzA, and FrzE complex to the downstream motility components, but these downstream proteins remain elusive. To search for these downstream genes, I n c h et al. (2007) used a genetic screen for suppressors of a constitutively active FrzCD‘ mutant. Since the frzCDc colonies are very compact (cells hyperreverse and show no net translocation), suppressor mutants could easily be identified by their spreading phenotype. Presumably, only mutations in the frz genes or in genes that function downstream of the Frz chemosensory pathway would allow cells to escape the hyperreversing FrzCD‘ phenotype. Using this strategy, frzZ was found to suppress frzCDc,revealing that FrzZ is a downstream component of the Frz

129

pathway. FrzZ is a CheY-CheY fusion protein. frzZ deletion mutants show the typical Frz phenotype, including decreased cellular reversal frequencies in both the Aand S-motility systems (Bustamante et al., 2004). AfrzZ is also epistatic to frzEACheY, again suggesting that FrzZ acts downstream of FrzE. Both CheY domains of FrzZ contain the highly conserved Asp residues necessary for phosphotransfer from FrZECheA (Trudeau et al., 1996). In vitro experiments with [32P]FrzECheA demonstrated that both CheY domains of FrzZ can rapidly accept phosphate from FrzECheAand that phosphorylation occurs on the predicted aspartate residues (Inclhn et al., 2007). Based on these data, we propose that FrzZ mediates signals from FrzE to the downstream motility systems (Color Plate 2).

A DOWNSTREAM OSCILLATOR CONTROLS REVERSAL OF THE A- AND S-ENGINES Genetic analysis has shown that reversals for both the A- and S-motility engines are controlled by Frz signaling (Li et al., 2003). However, the link between the Frz pathway and the two motility engines and how it regulates reversal frequency remain major unsolved questions. The Frz system could be a branched pathway that signals each motility engine independently. This model would involve complex cross talk to integrate signaling output for two motility engines. Alternatively, the Frz system may communicate with a single protein that triggers coordinated reversals. The latter model predicts that disruption of the common output protein should disrupt both A- and S-motility. Furthermore, this model suggests that the output protein should interact with coupling proteins from each of the two motility systems. We have recently obtained evidence from the study of FrzS, an S-motility protein, and AglZ, an A-motility protein, that suggests that these similarly constructed proteins may be engine-specific output proteins for the two motility systems and that they regulate cell reversals. FrzS and AglZ have remarkably similar modular structures: a pseudoreceiver domain at their N terminus that is connected to an extended “coiled-coil’’ domain by a flexible alanine-proline-rich linker region. Interestingly, the function of each protein is related to a single motility system: FrzS and AglZ are specifically required for S- and A-motility, respectively (Ward et al., 2000). Microscopy and genetic studies have shown that FrzS is not a structural component of the S-engine, but more likely a regulator of S-motility (Ward et al., 2000). AglZ is structurally related to FrzS and may therefore also be a regulator of the A-engine rather than one of its structural components.

130

A strain that expressed a functional FrzS-green fluorescent protein (GFP) fusion was constructed and monitored in moving cells by fluorescence microscopy (Mignot et al., 2005). FrzS was observed to oscillate from pole to pole as cells reverse. Immediately after a reversal, FrzS accumulated at the leading, piliated pole; as cells moved forward, FrzS began to also accumulate at the lagging cell pole, until equivalent amounts of the FrzS protein were found at both poles. The cell then reversed, and FrzS rapidly relocalized from the old leading pole to the new leading pole. Importantly, the oscillations of FrzS were regulated by the signaling activity of the Frz system: pole-to-pole switching of FrzS was rarely observed in the frzE mutant, whereas a constitutive frzCDc mutation induced hyperreversals and concomitant hyperoscillations of FrzS (Mignot et al., 2005). The proposed mechanism of FrzS pole-to-pole trafficking is not discussed here, as it is described in detail in chapter 6. FrzS was also studied by constructing in-frame deletion mutants that express stable cryptic proteins; these mutants showed loss of function and/or distinct localization defects. Specifically, in-frame deletions of the pseudoreceiver domain, the coiled-coil domain, and a motif located at the very C terminus of the protein each resulted in aberrant localization and loss of function defects (Mignot et al., 2007a). Removal of the pseudoreceiver domain caused preferential targeting of the protein to the lagging end of the cells (normal localization is to the leading cell pole), whereas deletion of the C-terminal tail led to weak localization to the leading cell pole. Deletion of both the pseudoreceiver domain and the C-terminal tail resulted in complete loss of polar localization, a phenotype also observed when the coiled-coil domain was deleted. This analysis showed that pole-to-pole oscillations of FrzS result from complementary roles for the protein domains: the pseudoreceiver domain is essential for accumulation of FrzS at the leading pole, whereas the C-terminal tail is a polar anchoring factor that affects localization at both the leading pole and the lagging pole. Based on these experiments, we hypothesize that (i) signaling to the pseudoreceiver domain results in loss of affinity of FrzS for the leading pole, targeting it to the lagging pole; (ii) this targeting is facilitated by the C-terminal tail; and (iii) FrzS is transported from pole to pole via the coiled-coil domain. Since AglZ is similar in structure to FrzS, we studied the dynamic localization of AglZ, anticipating that it might provide information about cellular reversals mediated by the A-engine. A strain that expressed a functional AglZ-yellow fluorescent protein (YFP) fusion protein was constructed, and its localization was monitored by fluorescence microscopy (Mignot et al.,

DEVELOPMENT AND MOTILITY 2007b). We observed that the localization of AglZ was dependent on the activity of the A-engine: when the cells were actively moving, AglZ-YFP was localized in a series of clusters that spanned the cell length; in contrast, nonmoving cells showed AglZ-YFP localized at one cell pole or diffused in the cytoplasm. When moving cells reversed, the clusters dispersed and AglZ was rapidly relocalized to the new leading pole. As cell movement resumed, AglZ redistributed as ordered clusters from the leading pole. These AglZ oscillations were also regulated by the Frz system: AglZ never switched poles in the frzE mutant and the frzCD“ mutant displayed hyperoscillations of AglZ. These studies show that in M. xanthus, cellular reversals involve regulated oscillations of A- and S-motility proteins that are targeted to the new leading cell pole at the time of reversal.

HYPOTHESES AND FUTURE PERSPECTIVES The fact that proteins from both motility systems oscillate together from pole to pole during reversals suggests that a common regulator controls their dynamics. It is unlikely that FrzS and AglZ are directly phosphorylated by FrzE because FrzS and AglZ lack the critical aspartate residue that is phosphorylated in canonical receiver domains (Fraser et al., 2007) and FrzZ is known to be the cognate response regulator for phospho-FrzE. MglA is an ideal candidate for the common output regulator of the Frz pathway, as it is essential for both A- and S-motility (Spormann, 1999; see also chapter 6). Interestingly, MglA is homologous to small GTPases of the Ras family such as the yeast Sarl protein. In eukaryotic cells, small GTPases act as regulatory proteins often by recruiting factors to their site of action. It is therefore possible that MglA acts to recruit factors that are important for reversal of the two motility systems. Consistent with this hypothesis, MglA is essential for the localization of both FrzS and AglZ: in an mglA mutant, FrzS could localize to only one cell pole and AglZ was almost completely diffuse in the cytoplasm. This is likely a direct effect because protein interaction studies have shown that MglA can interact directly with FrzS and AglZ and MglA colocalizes with FrzS and AglZ. We speculate that the Frz pathway acts on MglA, which in turns recruits motility proteins to their sites of action to regulate cellular reversals. Cellular reversals are regulated by two molecular oscillators. The activator of the upstream oscillator consisting of the Frz chemosensory proteins and referred to as the “Frzilator” (Igoshin et al., 2004), is essential for operating the second oscillator, composed of structural components of the A- and S-motility proteins that periodically

7. FRZ CHEMOSENSORY SYSTEMOF M.

XANTHUS

oscillate from pole to pole (Color Plate 2). Indeed, core frz mutants are defective at cell reversals and pole-to-pole oscillations, showing that the downstream oscillator is not operational in these mutants. The periodicity of the downstream oscillator is probably regulated by signal flow from the Frz system: indeed, constitutive signaling from Frz leads to hyperoscillations of the polar proteins. How then does the Frz system regulate the oscillations of the downstream proteins? We hypothesize that Frz signaling somehow activates MglA, the common regulator of both motility systems (Color Plate 2). To validate this hypothesis, it will be critical to elucidate how the Frz system acts on MglA and how MglA then acts to regulate the downstream oscillator. For example, one could imagine a scenario where FrzZ acts as a shuttle to transduce signals directly from FrzE to MglA or MglB, a putative regulator of the GTPase activity of MglA. The other proteins recruited by MglA and the dynamics of their interactions are unknown. These proteins are the likely missing links that couple the Frz chemosensory system with the two engines of motility that propel M. xanthus. We are grateful to members of the Zusman laboratory, past and present, for many helpful discussions. The research in our laboratory was supported by a grant from the National Institutes of Health (GM20509).

References Arnold, J. W., and L. J. Shimkets. 1988. Cell-surface properties correlated with cohesion in Myxococcus xanthus. J. Bacterial. 170:5771-5777. Astling, D. P. 2003. Novel Regulatory Mechanisms of a Chemotaxis Pathway in the Gliding Bacterium Myxococcus xanthus. Ph.D. thesis. University of California, Berkeley. Astling, D. P., J. Y. Lee, and D. R. Zusman. 2006. Differential effects of chemoreceptor methylation-domain mutations on swarming and development in the social bacterium Myxococcus xanthus. Mol. Microbiol. 59:45-55. Blackhart, B. D., and D. R. Zusman. 1985. “Frizzy” genes of Myxococcus xanthus are involved in control of frequency of reversal of gliding motility. Proc. Natl. Acad. Sci. USA 82:8767-8770. Bonner, P. J., Q. Xu, W. P. Black, Z. Li, Z. Yang, and L. J. Shimkets. 2005. The Dif chemosensory pathway is directly involved in phosphatidylethanolamine sensory transduction in Myxococcus xanthus. Mol. Microbiol. 571499-1508. Bowden, M. G., and H. B. Kaplan. 1998. The Myxococcus xanthus lipopolysaccharide 0-antigen is required for social motility and multicellular development. Mol. Microbiol. 30~275-284. Bustamante, V. H., I. Martinez-Flores, H. C. Vlamakis, and D. R. Zusman. 2004. Analysis of the Frz signal transduction system of Myxococcus xanthus shows the importance of the conserved C-terminal region of the cytoplasmic chemoreceptor FrzCD in sensing signals. Mol. Microbiol. 53:15011513.

131 Dworkin, M., and D. Eide. 1983. Myxococcus xanthus does not respond chemotactically to moderate concentration gradients. J. Bacteriol. 154:437442. Fraser, J. S., J. P. Merlie, Jr., N. Nichols, S. R. Westfield, T. Mignot, D. E. Wemmer, D. R. Zusman, and T. Alber. 2007. An atypical receiver domain controls the dynamic polar localization of the Myxococcus xanthus social motility protein FrzS. Mol. Microbiol. 65:317-332. Geng, Y., Z. Yang, J. Downard, D. Zusman, and W. Shi. 1998. Methylation of FrzCD defines a discrete step in the developmental program of Myxococcus xanthus. J. Bacteriol. 180:5765-5768. Hodgkin, J., and D. Kaiser. 1979. Genetics of gliding motility in Myxococcus xanthus (Myxobactera1es)-two gene systems control movement. Mol. Gen. Genet. 171:177-191. Igoshin, 0. A., A. Goldbeter, D. Kaiser, and G. Oster. 2004. A biochemical oscillator explains several aspects of Myxococcus xanthus behavior during development. Proc. Natl. Acad. Sci. USA 101:15760-1 5 765. Inclh, Y. F., H. C. Vlamakis, and D. R. Zusman. 2007. FrzZ, a dual CheY-like response regulator, fuctions as an output for the Frz chemosensory pathway of Myxococcus xanthus. Mol. Microbiol. 65:90-102. Kearns,D.B.,P. J.Bonner,D.R. Smith, andL. J. Shimkets. 2002. An extracellular matrix-associated zinc metalloprotease is required for dilauroyl phosphatidylethanolamine chemotactic excitation in Myxococcus xanthus. J. Bacteriol. 184~1678-1684. Klose, K. E., D. S. Weiss, and S. Kustu. 1993. Glutamate at the site of phosphorylation of nitrogen-regulatory protein NTRC mimics aspartyl-phosphate and activates the protein. J. Mol. Biol. 232:67-78. Li, Y., H. Sun, X. Ma, A. Lu, R. Lux, D. Zusman, and W. Shi. 2003. Extracellular polysaccharides mediate pilus retraction during social motility of Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 1005443-5448. McBride, M. J. 2001. Bacterial gliding motility: multiple mechanisms for cell movement over surfaces. Annu. Rev. Microbiol. 55:49-75. McBride, M. J., T. Kohler, and D. R. Zusman. 1992. Methylation of FrzCD, a methyl-accepting taxis protein of Myxococcus xanthus, is correlated with factors affecting cell behavior. J. Bacteriol. 174:4246-4257. McBride, M. J., R. A. Weinberg, and D. R. Zusman. 1989. “Frizzy” aggregation genes of the gliding bacterium Myxococcus xanthus show sequence similaritiesto the chemotaxis genes of enteric bacteria. Proc. Natl. Acad. Sci. USA 86:424-428. Mignot, T., J. P. Merlie, and D. R. Zusman. 2005. Regulated pole-to-pole oscillations of a bacterial gliding motility protein. Science 310:855-857. Mignot, T., J. P. Merlie, Jr., and D. R. Zusman. 2007a. Two localization motifs mediate polar residence of FrzS during cell movement and reversals of Myxococcus xanthus. Mol. Microbiol. 65:363-372. Mignot, T., J. W. Shaevitz, P. L. Hartzell, and D. R. Zusman. 2007b. Evidence that focal adhesion complexes power bacterial gliding motility. Science 315:853-856. O’Connor, K. A., and D. R. Zusman. 1991. Development in Myxococcus xanthus involves differentiation into two cell

132 types, peripheral rods and spores. J. Bacteriol. 173:33183333. Reichenbach, H. 1999. The ecology of the myxobacteria. Enviyon. Microbiol. 1:15-21. Shi, W., T. Kohler, and D. R. Zusman. 1993. Chemotaxis plays a role in the social behaviour of Myxococcus xanthus. Mol. Microbiol. 9:601-611. Sliusarenko, O., J. Neu, D. R. Zusman, and G. Oster. 2006. Accordion waves in Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 103:1534-1539. Ssgaard-Andersen, L., and D. Kaiser. 1996. C factor, a cellsurface-associated intercellular signaling protein, stimulates the cytoplasmic Frz signal transduction system in Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 93:2675-2679. Spormann, A. M. 1999. Gliding motility in bacteria: insights from studies of Myxococcus xanthus. Microbiol. Mol. Biol. Rev. 63:621-641. Sun, H., D. R. Zusman, and W. Y. Shi. 2000. Type IV pilus of Myxococcus xanthus is a motility apparatus controlled by the frz chemosensory system. Curr. Biol. 10:1143-1146. Trudeau, K. G., M. J. Ward, and D. R. Zusman. 1996. Identification and characterization of FrzZ, a novel response regulator necessary for swarming and fruiting-body

DEVELOPMENT AND MOTILITY formation in Myxococcus xanthus. Mol. Microbiol. 20:645655. Ward, M. J., H. Lew, and D. R. Zusman. 2000. Social motility in Myxococcus xanthus requires FrzS, a protein with an extensive coiled-coil domain. Mol. Microbiol. 37:13571371. Ward, M. J., and D. R. Zusman. 1999. Motility in Myxococcus xanthus and its role in developmental aggregation. Curr. Opin. Microbiol. 2:624-629. West, A. H., and A. M. Stock. 2001. Histidine kinases and response regulator proteins in two-component signaling systems. Trends Biochem. Sci. 26:369-376. Wolgemuth, C., E. Hoiczyk, D. Kaiser, and G . Oster. 2002. How myxobacteria glide. Curr. Biol. 12:369-377. Youderian, P., N. Burke, D. J. White, and P. L. Hartzell. 2003. Identification of genes required for adventurous gliding motility in Myxococcus xanthus with the transposable element mariner. Mol. Microbiol. 4955.5-570. Yu, R., and D. Kaiser. 2007. Gliding motility and polarized slime secretion. Mol. Microbiol. 63:454-467. Zusman, D. R. 1982. “Frizzy” mutants: a new class of aggregation-defective developmental mutants of Myxococcus xanthus. J. Bacteriol. 150:1430-1437.

Regulatory Mechanisms

Myxobacteria: Multicellularity and Differentiation Edited by David E. Whitworth 0 2008 ASM Press, Washington, D.C.

John R. Kirby, James E. Berleman, Susanne Muller, Di Li, Jodie C. Scott, Janet M. Wilson

Chemosensory Signal Transduction Systems in Myxococcus xanthus

Many microbes have been analyzed for their capacity to utilize signal transduction in order to navigate their environment by responding to a multitude of stimuli including physical, chemical, and biological cues. The majority of the analyses have focused on two-component signal transduction (TCST) systems, even though these are likely to compose only a fraction of the total systems that have evolved to translate information derived from the environment (Ulrich et al., 2005). The prototype for the TCST system is composed of a sensor kinase (or phosphatase) and a response regulator. The sensor kinase is phosphorylated on a conserved histidine residue using ATP as the donor and subsequently passes the phosphoryl group to a conserved aspartate residue within the response regulator (Fig. 1).The domain architecture for TCST systems is highly variable but typically consists of an input sensor domain that is periplasmic (or extracellular in gram-positive bacteria), responds directly to environmental signals, and is covalently linked to the kinase domain. The prototypical response regulator contains the aspartyl receiver domain covalently linked to an output domain which binds DNA to affect transcription (Parkinson and Kofoid, 1992; Parkinson, 1993; Hoch and Silhavy, 1995; Stock et al., 2000; Hoch,

8

2000). Variable inputs and outputs allow these systems to respond to a great number of stimuli and subsequently regulate specific subsets of genes or operons. TCST systems in Myxococcus xanthus are discussed in more detail in chapter 10. The best-studied TCST system is the chemotaxis system (Fig. 1)that regulates flagellar motility in Escherichia coli. Forty years of research in the field of chemotaxis has led to a great understanding of the overall mechanism governing the behavioral responses to chemical stimuli. The most detailed description to date has been given for the response by E. coli to the chemoattractant aspartate that is mediated by Tar, a methyl-accepting chemotaxis protein (MCP) that also mediates the response to maltose (via the maltose binding protein) and two repellents, Ni2+ and Co2+.When the periplasmic ligand binding site is titrated with aspartate, the Tar homodimeric receptor undergoes a conformational change that influences the cytoplasmic Chew-CheA ternary signaling complex. Aspartate-bound Tar influences phosphotransfer from CheA to CheY such that the concentration of the phosphorylated CheY response regulator transiently decreases. Diminished levels of phospho-CheY (the tumble regulator) lead to prolonged duration of

John R. Kirby, James E. Berleman, Susanne Miiller, Di Li, Jodie C. Scott, and Janet M. Wilson, The University of Iowa, Department of Microbiology, 51 Newton Rd., Iowa City, IA 52242.

135

REGULATORY MECHANI sM s

136

B

C

Figure 1 Domain topology of TCST systems. (A) A prototypical TCST system is shown. A variable input (sensor) domain is covalently bound to the histidine kinase domain. Phosphorylation occurs on a conserved histidine residue. The phosphoryl group is transferred to a conserved aspartate residue within the receiver domain (Rec) in the response regulator. The prototypical RR output is a DNA-binding domain capable of influencing gene expression. The vertical bar represents the cytoplasmic membrane. (B) The specialized TCST system that controls chemotaxis is shown. The MCP transducer is depicted as transmembrane and is coupled by Chew to the CheA kinase. Only two methyl groups are shown to represent methylation of the receptor by CheR. Phosphorylated CheB can remove these methyl groups. Methylation is a hallmark feature of the chemotaxis TCST systems. Phosphotransfer to the response regulator CheY influences its ability to bind the FliM switch component at the flagellar motor. (C) A chemosensory system such as the Che3 system found in M . xanthus is depicted. Chemosensory systems represent a composite of the prototypical TCST and the specialized chemotaxis TSCT systems.

counterclockwise rotation of the flagellar motor, thereby producing a swimming event in response to the attractant aspartate. The flux of phosphoryl groups (or diminution) is attenuated by a second posttranslational modification, methylation, of the MCP chemoreceptor. CheR acts

constitutively as a methyltransferase (utilizing S-adenosylmethionine as the methyl donor) to methylate specific glutamate residues (generating glutamate-0-methyl esters) located within two cytoplasmic domains of the Tar receptor. CheB acts as a methylesterase to demethylate

8.

CHEMOSENSORY SIGNALTRANSDUCTION SYSTEMSIN M. XANTHUS

the glutamate methyl esters regenerating the glutamate residues within the Tar receptor. CheB is a response regulator and is regulated by phosphorylation by the CheA kinase. Therefore, CheA regulates two response regulators, CheY and CheB. Because CheB requires phosphorylation by CheA prior to becoming an active methylesterase, demethylation lags temporally with respect to the flux of phosphoryl groups transferred to CheY. Methylating the Tar receptor serves to upregulate the CheA kinase, thereby restoring the flux of phosphoryl group transfer to CheY. This results in adaptation to prestimulus levels even in the presence of the stimulating ligand, aspartate, and constitutes a feedback loop. Adaptation allows the cell to compare the chemical environment over time and adjust its behavior such that the cell migrates toward more favorable conditions (Falke et al., 1997; Falke and Hazelbauer, 2001; Bourret and Stock, 2002). Many other TCST systems have been analyzed including those that regulate responses to quorum sensing (Bassler, 2002), virulence (Krukonis and DiRita, 2003), protein folding (DiGiuseppe and Silhavy, 2003), osmolarity (Qin et al., 2003), competence (Tortosa and Dubnau, 1999), DNA uptake (Brencic and Winans, 2005), sporulation (Piggot and Hilbert, 2004), and others. Importantly, the vast majority of these systems have been shown to directly regulate transcription via a response regulator possessing a DNA-binding domain. Thus, the TCST system controlling chemotaxis whereby CheY interacts directly with switch components (FliM)at the base of the flagellar motor is an exception to the rule (Fig. 1). The majority of all work on chemotaxis systems up until the last decade focused on the control of flagellum-based motility. Recently several organisms that do not exclusively utilize flagellar motility but are known to encode multiple homologs for the chemotaxis proteins have been the focus of much work. The chemotaxis genes in these organisms are usually organized into multiple, discrete operons randomly dispersed on one or more chromosomes. Investigation into the roles of the various chemotaxis gene clusters in these organisms has revealed that not all chemotaxis systems regulate flagellar motility. Indeed, some of these chemotaxis-like systems do not regulate motility at all and have been termed “chemosensory” systems. Three important examples include the Che3 system in M. xanthus (Kirbyand Zusman, 2003),the Che3 system in Rhodospirillum centenum (Berleman and Bauer, 2005), and the Che4 (Wsp) system in Pseudomonas aeruginosa (Hickman et al., 2005). In these cases, it is clear that a chemotaxis-like system comprising MCPs, Chew, CheA, CheR, and CheB has been co-opted to regulate other outputs. In the case of the M. xanthus Che3 system, the data indicate that CrdA, a homolog of NtrC, directly regulates

137

genes that are necessary for the control of development. Similarly, the R. centenum Che3 system affects the timing and level of cyst formation. The Che4 (Wsp) system in P. aeruginosa affects biofilm formation by directly regulating the concentration of cyclic diguanylate (c-di-GMP) via the response regulator, WspR, that possesses a GGDEF domain. The conclusion drawn from these examples is that chemosensory systems are specialized TCST systems with built-in adaptation modules by virtue of utilizing an MCP chemoreceptor to process the stimulus (Fig. 1).Because the MCP chemoreceptors are not covalently linked to the kinase domain (as is the case with prototypical TCST systems), multiple chemoreceptors can interact with a given CheA kinase, thereby effectively increasing the repertoire for ligand binding and stimulus processing. Integration of mltiple, varied inputs therefore occurs at the level of the CheA kinase. CheA kinases dictate the flow of information from the receptors to the outputs governed by a given system. Therefore, the existence of a gene encoding a CheA homolog defines a chemosensory system. Several hundred genomes are now fully sequenced and publicly available, and it is apparent that most motile organisms utilize multiple chemosensory systems, with the average being either two or three (I. B. Zhulin, personal communication). It is worth mentioning that nonmotile organisms typically lack chemotaxis-like genes altogether. M . xanthus is unique in that is has eight full chemosensory systems and is currently the only species known to possess such a large number of these specialized TCST systems. Because these systems are classified as homologs based on sequence similarity but appear to carry out alternate functions, they are by definition paralogs and not orthologs. M. xanthus is therefore an ideal organism for analysis of chemosensory signal transduction systems in bacteria. Several major biological questions arise from the above observations: (i) What is the role of each chemosensory system in M. xanthus? (ii)How do these systems maintain molecular insulation in order to prevent cross talk? (iii) Do these systems display cross-regulation? (iv) Is there a relationship between the diversity of chemosensory signal transduction systems and the complex lifestyle displayed by M. xanthus in particular or for bacteria in general? Discussion of the M. xanthus chemosensory systems and the role of multiple paralogous systems follows.

THE EIGHT CHEMOSENSORY SYSTEMS IN M. XANTHUS Analysis of the completed 9.14-Mb 111. xanthus genome indicates that of nearly 7,400 putative open reading frames (ORFs), 605 genes encode putative

REGULATORY MECHANISMS

138 signal transduction proteins. Three hundred twentytwo of these genes are predicted to encode homologs that contain either histidine kinase domains or aspartyl receiver domains that compose 125 TCST systems to process environmentally derived stimuli (Ulrich and Zhulin, 2007; see chapter 10). Within this group of TCST systems are the eight CheA homologs that define eight chemosensory systems. Importantly, each cheA gene is located with a cluster that also encodes homologs to known chemotaxis genes, as well as some nonchemotaxis genes (Fig. 2 ) . Based on the paradigm for E . coli chemotaxis, we can predict with a high degree of

confidence that the chemotaxis homologs will behave in specific ways within each chemosensory system. Nevertheless, recent examples (mentioned above) have demonstrated that many systems deviate from the E. coli paradigm and therefore warrant full experimental investigation.

The M. xanthus Frz (Chel) and Dif (Che2) Systems The M . xanthus Frz (Chel) and Dif (Che2) systems are described in detail in chapters 7 and 13, respectively, and are not discussed here.

che3 crdA

crdB crdC cheW3 mcp3A

mcp3B

cheA3

cheY4cheW4b

cheA4

cheB3

cheR3

che4 cheW4a cheR4

mcp4

che5 cheYS cheVS

che6

V

cheRS

chew5

V

cheW6a cheR6

cheAS

cheBS

cheA6

cheB6

Y

cheW6b

mcp6

mcp5

V

V

socD

V

kefC

che7

che8

ybaD

ribD

ribE

nbH nusB

cheY7

cheA7

cheY8a

cheA8

cheW7

cpc7

mcp 7

cheW8a cheW8b

cpc8

cheR7

cheR8

cheB7

cheB8

Figure 2 Genetic organization of the eight chemosensory systems in M. xanthus. Each cluster or operon is defined by the existence of a gene encoding a homolog to CheA (black). Genes encoding homologs to Chew (dotted), MCPs (white), CheB (squares), CheR (circles), CheC (gray), and CheY-like response regulators (shaded) are shown. ORFs that do not display homology to any known chemotaxis gene are also shown (diagonal stripes).

des 7

cheY8b

8. CHEMOSENSORY SIGNALTRANSDUCTION SYSTEMSIN 211. XANTHUS The M. xanthus Che3 Chemosensory System Previous work demonstrated that the M. xanthus Che3 system affects the timing of development (Kirby and Zusman, 2003). Mutations generated in several of the che3 genes led to premature development on starvation (complement fixation [CF]) media and also appeared to influence the vegetative component of the life cycle. For example, the mcp3B mutant displayed a rippling phenotype on rich (charcoal-yeast extract [CYE]) media and was observed to form aggregates, indicative of starvation, at the outer edge of the growing colony (Fig. 3). Likewise the cheA3 mutant was shown to be premature for development. Although no obvious motility defects were identified through direct observation of individual cells, elevated gene expression was observed for known developmental markers including spi, tps, and mbhA. Because premature developmental gene expression was shown to correlate with premature aggregation and fruiting body formation for these mutants, the Che3 system was predicted to regulate developmental gene expression. Analysis of the che3 gene cluster indicated that a divergently transcribed NtrC homolog (CrdA [chemosensory regulator of development]) was the likely regulator for this system. A mutation in the crdA gene led to delayed development consistent with the prediction that CrdA is a homolog of NtrC, a known transcriptional activator. Because the cheA3 and crdA mutants have opposite phenotypes with respect to the timing of development, epistasis analysis was possible. The cheA3crdA double mutant was generated, analyzed, and found to be delayed for development. Thus, crdA is epistatic

DZ2 (wild type)

139

to cheA3 and allows us to generate a model in which CheA3 processes information via CrdA as its primary output (Fig. 4). Because cheA3 and crdA have opposite phenotypes, it is likely that CheA3 acts as a phosphatase (or an inhibitor of CrdA phosphorylation) during vegetative growth. The CheR3 methyltransferase and CheB3 methylesterase were shown to be critical for control of development as well. Based on the paradigm chemotaxis system in E. coli and the premature developmental phenotype displayed by the cheA3 mutant, we predicted that the cheR3 and cheB3 mutant phenotypes would show premature and delayed phenotypes, respectively. The cheR3 mutant cells displayed premature aggregation, as expected, but the cheB3 mutant also displayed prematurc‘aggregation. However, the timing of maturation of the cheB3 fruiting bodies and sporulation was found to be delayed. Thus, the cheB3 mutant displays a complex phenotype suggesting that the role of methylation may not follow the E. coli paradigm and is likely to be more complex, like that observed for Bacillus subtilis (Kirby et al., 1999; see “Summary and Conclusions” below). Another factor that may relate to the complex phenotype displayed by the cheB3 mutant is the presence of two MCPs, Mcp3A and Mcp3B, within the Che3 system. All MCPs studied to date are thought to form homodimers. The homodimers are thought to be highly organized within functional arrays (Maddock et al., 1993) and possibly form trimers of dimers that affect signal integration and processing (Ames et al., 2002). Although direct evidence of such higher-order structure

mcp3B

Figure 3 The che3 mutants display premature development on rich medium. Both the parent (DZ2) and mcp3B mutant were grown at 32°C on rich medium (CYE) for 3 days and photographed (Kirby and Zusman, 2003); both colonies are approximately 2 cm in diameter. The mcp3B mutant displays a rippling phenotype characteristic of development. Additionally, aggregates that resemble fruiting bodies are visible at the colony edge under higher magnification.

REGULATORY MECHANI sM s

140

as part of the mechanism for regulating CheA3 signal transduction to CrdA. CrdB Is a n Essential Lipoprotein That Senses Envelope Stress

CrdA

r+* 054

cheBR3

7

Figure 4 Model for Che3 signal transduction. The current model for Che3 chemosensory signal transduction is shown. CrdB is an outer membrane lipoprotein with a Germinal OmpA-like peptidoglycan binding domain and is predicted to sense envelope stress. A conformational change in CrdB transmits a signal to the MCP receptor complex affecting CheA kinase levels in a manner similar to that observed during chemotaxis. CrdC is homologous to Chew and may affect coupling of the receptors to the CheA3 kinase. Results indicate that CrdA is autoregulatory and regulates expression of che3, cheBR3, and other genes, thereby affecting development as described previously (Kirby and Zusman, 2003). The cytoplasmic and outer membranes (black lines) and the peptidoglycan layer (crosshatching) demarcate the periplasm.

has only been shown for E. coli, there is experimental evidence indicating that MCPs are organized in arrays in Caulobacter crescentus (Alley et al., 1992), Rhodobacter sphaeroides (Wadhams et al., ZOOO), B. subtilis (Kirby et al., 2000), and others. Because the ligands for MCP receptors in species other than E. coli are largely unknown, the precise function of receptor arrays in other bacteria has not been elucidated. We predict that Mcp3A and Mcp3B will form homodimers and function within an array of chemoreceptors. We also predict that the homodimers will display some level of interaction

The first gene in the che3 cluster is crdB (Fig. 2). This gene is predicted to encode a lipoprotein that also binds to peptidoglycan via its carboxy-terminal OmpA-like domain. A crdB-phoA fusion was created and assayed for activity on medium containing 5-bromo-4-chloro3-indolylphosphate (BCIP). The phoA gene encodes an alkaline phosphatase which is only active when exported. Subsequently, PhoA converts BCIP to a blue product within the medium. The CrdB-PhoA fusion construct was active both in E. coli and in M . xanthus. This analysisdlows us to conclude that CrdB is exported to the periplasm. Further analysis of the N-terminal sequence of CrdB indicates that the protein contains a signal peptide sequence and a lipobox (Sierakowska et al., 2003). CrdB contains a threonine residue immediately following the +1 cysteine anchor which is known to direct lipoproteins to the inner leaflet of the outer membrane (Seydel et al., 1999). The processing of lipoproteins occurs via the Signal Peptidase I1 pathway (Sankaran and Wu, 1995) and is specifically inhibited by the antibiotic globomycin. Globomycin was able to prevent the processing of the CrdB proprotein to the mature form, indicating that CrdB is exported via Signal Peptidase I1 to the periplasm, where it functions as a lipoprotein in M. xanthus. An insertion mutation previously generated in crdB was found to affect expression of downstream che3 genes, and thus, a new construct was created. Based on several failed attempts to delete this gene, we now believe that crdB is essential. Previous reports have indicated that certain lipoproteins affect outer membrane stability and are therefore critical for cell viability (Cascales et al., 2002). Multiple attempts to generate new insertions in crdB also failed. However, one insertion mutant was finally generated (crdB2303)but was found to be a merodiploid that allows for low level of expression of crdB and the downstream che3 genes. The crdB1303 mutant is delayed for development in contrast to the other che3 mutants, consistent with an overall membrane defect. Further analysis of the crdB1303 mutant indicated that it is sensitive to EDTA or ampicillin, both of which have the potential to affect membrane integrity. Both EDTA and ampicillin were able to lyse the crdB2303 mutant cells when present in the medium at concentrations that had no effect on the wild type.

8. CHEMOSENSORY SIGNALTRANSDUCTION SYSTEMSI N M. XANTHUS Based on these data, we hypothesize that CrdB senses membrane integrity or periplasmic intermediates that reflect the presence of environmental stress to initiate the developmental program. Our current working model is that CrdB transduces signals through one of the two MCPs in the Che3 signaling complex to either inhibit or stimulate CheA3 activity (Fig. 4).

141

information via Frz to affect TFP-based motility. It is worth noting that the data for the che4 mutants were obtained for single cells and not for cells within a population. Therefore it is not clear how these results apply to individuals within larger groups. Assessment of green fluorescent protein-labeled cells within populations is possible and will allow for more thorough analysis of the Che4 system.

CrdC is a Homolog of Chew The gene encoding CrdC lies between the chew3 gene and the first of the two MCP homologs (Fig. 2). Analysis of the ORF for CrdC indicates that crdC and mcp3A are translationally coupled, suggesting that these two proteins interact within the signal transduction complex. Recent sequence analysis suggests that CrdC is likely to be a highly diverged homolog of Chew. Chew3 and CrdC have been shown to interact in the yeast twohybrid assay while only CheW3 is able to bind CheA3. Therefore, our current model places CrdC in a position to affect Chew3 within the signaling complex. It is also possible that CrdC may act as an inhibitor of Chew3 to affecting coupling between the MCPs and CheA3.

The M. xanthus Che4 Chemosensory System The genes in the che4 operon were shown to compose an operon and encode homologs to one MCP, two Chews, CheA, a CheY-like response regulator, and CheR, but no CheB. Analysis of the Che4 system led to the conclusion that the system affects type IV pilus (TFP)-based motility. The link between the Che4 system and regulation of the TFP motor is not known. Phenotypes for the che4 mutants were only discernible in the uglBl mutant background which lacks adventurous (A) motility and were only apparent under specific developmental conditions. The observations are consistent with a model whereby the Che4 system perceives physical stimuli such as surface hydration and transduces information to the pilus-based machinery. Analysis of the parent and che4 system mutants led to the conclusion that cellular velocity is inversely correlated with reversal frequency for wild-type cells. This inverse correlation was eliminated in mcp4 and cheY4 mutants. The structural components required for TFP were not disrupted by any of the mutations in the che4 operon. Together the results allow us to conclude that the Che4 system affects reversal frequency of cells by modulating the function of the TFP (Vlamakis et al., 2004). Consistent with this model is the observation that CheA4 was found to interact with the N terminus of FrzCD, the cytoplasmic MCP receptor for the Frz system (D. R. Zusman, unpublished data). Because the Frz system is known to regulate cellular reversal frequency, it is possible that Che4 communicates specific

The M. xanthus CheS Chemosensory System The genes in the che.5 cluster encode homologs to one MCP, two Chews (one is CheV-like), CheA, a CheY-like response regulator, CheR, and CheB. A thorough analysis of the Che.5 system has not yet been performed and thus is not discussed in detail here. We do not yet know if tke che5 genes shown in Fig. 2 compose an operon. However, we have constructed a mutation in the cheA.5 gene and have observed premature development in the cheA5 mutant, similar to what was observed for cheA3 and cheA6. Sequence analysis indicates that the Che.5 system possesses a unique chemotaxis-like protein that is a receiver domain-Chew fusion, which we refer to as “CheV-like.” CheV proteins are currently described as Chew-receiver domain fusions (Fredrick and Helmann, 1994). Thus, the CheV-like homolog in CheS possesses both domains but in the opposite orientation relative to CheV. The domain orientation should have little impact in the overall function of the protein, as it is predicted to function within the signaling complex as a Chew homolog by coupling Mcp.5 to the CheA.5 kinase. No further details are available at this time.

The M. xanthus Che6 Chemosensory System There are two primary reasons for our investigation into the role of the Che6 chemosensory system. First, the gene order is consistent with that found in other organisms known to utilize TFP-based motility (Zhulin, personal communication) and we therefore hypothesized that the Che6 system would regulate TFP-based motility. Second, a mutant allele known to suppress the csgA mutant developmental defect was mapped to SOCD(suppressor of csgA) (Rhie and Shimkets, 1991),which we now know is cotranscribed as part of the che6 operon. In order to verify that the nonchemotaxis genes, especially socD, were part of the che6 operon, we performed reverse transcriptase PCR. The results indicated that the genes shown compose an operon (Fig. 2). As with the other che operons, the che6 cluster contains a full complement of chemotaxis genes encoding homologs to CheA, two Chew proteins, CheR, CheB, and one MCP. These chemotaxis genes, as well as SOCDand kefC (encoding a potassium efflux pump), are cotranscribed

142 as part of the che6 operon. Analysis of the start and stop codons and putative ribosome binding sites for each of the ORFs within the che6 cluster leads us to predict that each pair of these genes within the che6 operon (except socD-KefC) is translationally coupled and therefore likely to function within the same pathway. As is the case with all known chemosensory systems studied thus far, the CheA kinase is the central processor through which MCP-generated signals are integrated and processed. Multiple MCP homologs can affect one CheA kinase, as is the case for E. coli, B. subtilis, and Halobacterium salinarum. Likewise, each CheA typically has multiple targets including CheY and CheB. Thus, CheA is the central processor for each system and mutations made within CheA block its function and usually give distinct phenotypes. The cheA6 mutant was created and displays an obvious motility defect on all agar surfaces. However, the mutant will eventually form very small flares on 0.3% agar after several days, indicating that TFP-based motility is dramatically reduced. This TFP motility defect led to the hypothesis that Che6 affects pilA gene expression or pilus production. We measured the level of PilA monomer and production of surface pili by immunoblot analysis using anti-PilA antibody (Wu and Kaiser, 1997; Wall et al., 1998). Both the DZ2 parent and cheA6 mutant produced equal amounts of PilA monomer indicating that gene expression from the pi1 cluster is not affected in the cheA6 mutant. However, the cheA6 mutant cells were found to lack PilA normally detectable in cell surface preparations of TFP. Recent results from transmission electron microscopy confirm these results. In contrast to the wild type, the cheA6 mutant cells do not produce detectable pili under the conditions tested. Because the PilA monomer is produced at or near wild-type levels, we conclude that CheA6 is required for TFP assembly. Similar results were obtained for the mcp6 mutant, indicating that Mcp6 processes information via CheA6. The mechanism by which Che6 affects pilus assembly remains to be determined. The roles for SocD and KefC also remain unknown.

The M. xanthus Che7 Chemosensory System Similar to the case described above, the M . xanthus che7 gene cluster contains a full set of chemotaxis genes and additional ORFs that show no homology to chemotaxis proteins. In order to verify that these genes compose an operon, we performed reverse transcriptase-PCR and were able to show that the che7 genes are cotranscribed (Fig. 2). The che7 operon encodes homologs to CheA, Chew, CheY, CheR, CheB, and one cytoplasmic MCP. In

REGULATORY MECHANISMS addition, genes encoding a phycocyanobilin lyase (cpc) and a fatty acid desaturase (des)are expressed as part of the che7 operon. Analysis of the che7 ORFs allows us to predict that Mcp7 and Cpc7 are translationally coupled and are therefore thought to interact within the Che7 system. It has been demonstrated that Cpc phycocyanobilin lyase proteins carry out specific covalent modifications whereby phycobilins are attached to apophycobiliprotein subunits ( Fairchild et al., 1992). The phycobiliprotein subunits are part of the macromolecular phycobilisome light-harvesting complex. Because M . xanthus does not possess phycobilisomes, Cpc7 is not likely to function as a phycocyanobilin lyase per se but might function in another light-dependent adaptation process. Previous work has demonstrated that the 211. xanthus car locus regulates carotenoid production in a light-dependent fashion such that colonies become bright orange when exposed to light (Fontes et al., 1993). We therefore tested the che7 mutants for their ability to produce carotenoids with peak absorbance at 475 nm in response to light. The mcp7, cheB7, and cpc7 mutants each displayed enhanced production of carotenoids (475 nm) relative to the parent, while the cheA7 mutant was completely incapable of producing this carotenoid. The mechanism by which the Che7 system affects carotenoid production is unknown. However, it is worth noting that the cheA7 mutant displays a severe swarming defect when assayed on 0.3% agar surfaces, does not aggregate (Fig. 5 ) ,and does not produce viable spores under the conditions tested. The mcp7 and cpc7 mutants also display reduced motility, no aggregation, and no viable spore production. Together, these observations lead us to speculate that the Che7 system affects membrane composition.

The M. xanthus Che8 Chemosensory System The genes in the che8 cluster encode homologs to CheA, two Chews, two CheY-like receiver domain proteins, CheR, and CheB, but no MCP. There are 1 3 orphan MCP homologs encoded by the M. xanthus genome, and it is likely that one or more of these MCPs will transduce signals via the Che8 complex. Interestingly, the che8 cluster also encodes a homolog of the cpc phycocyanobilin lyase gene similar to that seen in the che7 operon. Even more striking is the presence of several genes that constitute a major portion of the riboflavin biosynthetic operon. Whether these genes are cotranscribed with the che homologs and compose an operon has not been clearly established, although preliminary evidence suggests that they are cotranscribed as depicted (Fig. 2).

8. CHEMOSENSORY SIGNALTRANSDUCTION SYSTEMSIN M. XANTHUS

143

Figure 5 The cheA7 mutant displays defective motility and aggregation. The DZFl parent and cheA7 mutant are shown. Cells were grown in rich medium (CYE), washed in buffer (MMC), and plated on either 0.3% agar containing CYE or on 1.5% agar with low nutrients (CF).DZFl cells swarm on rich medium (A) and develop on low-nutrient medium to produce fruiting bodies (B). The cheA7 mutant cells display major defects in swarming (C) and development (D) relative to the parent.

SUMMARY AND CONCLUSIONS The presence of multiple paralogous chemosensory systems in M. xanthus presents a unique opportunity to explore variations on a common theme for signal transduction, network integration, and genome evolution in a model organism for prokaryotic development. Analysis of each chemosensory system thus far indicates that each system plays a unique role during the vegetative and/or developmental aspects of the complex multicellular life cycle of M. xanthus (Table 1).Analysis of the timing and level of expression of each operon is currently under way. However, it is worth noting that mutations in all eight che systems display phenotypes during vegetative growth as well as during development, indicating that expression of all eight che systems is required for both vegetative and developmental aspects of the M. xanthus life cycle. The Frz chemosensory system appears to directly affect motility and most likely does so by regulating cell reversals via FrzS (Mignot et al., 2005; see chapter 7). Each

of the other chemosensory systems appears to have been co-opted to regulate functions other than chemotaxis per se, although the Dif system indirectly affects motility by regulating EPS production (Yang et al., 2000; Xu et al., 2005). Additionally, it is likely that Che4 affects cell reversal frequency but may do so indirectly by interacting with Frz. Furthermore, Che6 affects TFP assembly by an unknown mechanism. Thus, motility is affected by four of the eight chemosensory systems either directly or indirectly. Analysis of the other four chemosensory systems suggests that these specialized TCST systems have different outputs ranging from gene expression to biosynthesis of riboflavin (Table 1). One of the major conclusions from our analysis of the M. xanthus chemosensory systems is that the physiological role of each chemosensory system is specified by its nonchemotaxis genes. The genes in question encode proteins that may provide alternative inputs, alternative outputs, or regulators that impinge on the basic design of the chemosensory system. In several cases, more than

REGULATORY MECHANISMS

144 Table 1 Chemosensory systems in M . xanthus Che system

Proposed function

Reference(s)

Frz (Chel) Dif (Che2) Che3

Gliding reversal frequency EPS production and PE detection Developmental gene expression

Che4 CheS Che6

TFP-based motility TFP assembly

Che7

Carotenoid production

Che8

Riboflavin biosynthesis

Zusman lab; chapter 7 Yang and Shimkets; chapter 13 Kirby and Zusman, 2003; S. Mueller and J. R. Kirby, unpublished data; D. Li and J. R. Kirby, unpublished data Vlamakis et al., 2004 Kirby lab, unpublished data J. C. Scott and J. R. Kirby, unpublished data J. M. Wilson and J. R. Kirby, unpublished data Leclerc and Kirby, unpublished data

?

one permutation on the basic theme is evident (Fig. 4). There is at least one gene that is not homologous to any known chemotaxis gene in each operon (Fig. 2). The relationship between those gene products and the associated chemosensory system is the primary focus of our ongoing investigation. Another major conclusion that results from our analysis is that many of the chemosensory systems maintain molecular insulation (each system gives phenotypes specific to its system). However, it is possible and even probable that cross-regulation will emerge as part of the network design (e.g., Che4 and Frz). The level and specificity of integration remains largely unexplored. Several factors are likely to directly influence cross-regulation including the similarity and nature of specific functions such as receptor clustering and methylation systems. As mentioned above, C. CYescentus, E. coli, and R. sphaeroides MCP receptors are capable of forming large clusters either at the cell poles or in the cytoplasm. The topology of the receptor dictates localization, and the clustering is thought to affect the overall function of a given MCP within the receptor array (Bray et al., 1998). It is therefore likely that several of the M. xanthus MCP homologs will form an array and affect the mechanism of signal transduction for a given receptor. There are 13 orphan mcp chemoreceptor genes on the M. xanthus genome. The orphan MCPs are likely to influence the composition of a putative receptor array. Moreover, these MCPs are predicted to interact with at least one Chew homolog encoded within one of the eight che clusters. The sequence of each MCP and its gene neighborhood have not yet been correlated with the chemosensory systems, stimuli, or outputs. One important example of cross-regulation in chemotaxis has been demonstrated for the methylation system

in E. coli. In E. coli, the CheR methyltransferase and the CheB methylesterase have both been shown to interact with the carboxy-terminal pentapeptide NWETF motif on either Tar or Tsr, the serine receptor (Feng et al., 1999; Li and Hazelbauer, 2005). However, neither Tap (dipeptide receptor) nor Trg (ribose and galactose binding protein receptor) possesses the CheRB pentapeptide NWETF docking sequence. Nevertheless, both Tap and Trg require methylation and demethylation for proper functioning. The evidence indicates that Tap and Trg are methylated and demethylated much more efficiently if Tar and Tsr are present. The conclusion from these data is that CheR and CheB target the docking sites on Tar and Tsr and carry out modification of the neighboring Trg and Tap receptors. Thus, the receptor array allows for cross-regulation by CheR and CheB. Although there is no direct evidence suggesting this is the case in M. xunthus, it is likely that CheR and CheB from any given chemosensory system would be capable of modifying neighboring receptors if those receptors are found to lie within a clustered receptor array. Several observations support the notion that M. xunthus receptors exist within a signaling array and that methylation and demethylation reactions occur via one of many CheR and CheB homologs. First, the cheB3 mutant displayed a complex phenotype (premature aggregation and delayed sporulation) relative to the mcp3A and mcp3B mutants (see above). Furthermore, normal DifA function appears to require methylation by CheR and CheB homologs (Bonner et al., 2005) yet no genes for cheR or cheB are present in the difcluster. Finally, the Che4 system requires the CheR4 methyltransferase for proper functioning but lacks a cheB methylesterase gene (Vlamakis et al., 2004). The most likely scenario is that a CheB homolog from another

8. CHEMOSENSORY SIGNALTRANSDUCTION SYSTEMSIN M. XANTHUS system functions to modulate the overall methylation level of Mcp4. It is worth noting that even though the Dif system lacks CheR and CheB homologs, it does possess a CheC homolog likely to act as a phosphatase (Black and Yang, 2004; Szurmant et al., 2004). Therefore, the Dif chemosensory signal transduction system is regulated in a unique way relative to the other seven Che systems in M. xanthus. Another important point is that there is an orphan cheR-cheB gene cluster with associated ORFs that do not encode chemotaxis-like proteins. The role of this cluster is not yet known. In summary, the chemosensory systems found in M. xanthus appear to have evolved in order to regulate functions that need temporal control mechanisms. Temporal control is the hallmark feature of chemotaxis-like systems where methylation and demethylation of the input MCP chemoreceptor affects the rate of phosphotransfer from the CheA kinase to a response regulator. The role of methylation and demethylation for each of the eight M . xanthus chemosensory systems remains to be fully analyzed. The kinetics of modification have not been assessed for any specific residue in any MCP homolog in M. xanthus, although the overall rate of methylation changes for FrzCD was shown to occur over a 30-min to 2-h window (McBride and Zusman, 1993) consistent with the time domain in which M. xanthus makes decisions regarding cell reversals and commitment to sporulation. Lastly, it has been demonstrated that there is a relationship between the number of signal transduction systems, genome size, and the complexity of the environmental niche and corresponding lifestyle for a given microbe (Van Nimwegen, 2003; Konstantinidis and Tiedje, 2004; Ulrich et al., 2005). The M . xanthus genome does not represent a major deviation from this relationship with respect to the total number of signal transduction proteins. However, M. xanthus does possess a relatively large number of TCST systems (see chapter 10) and is unparalleled with respect to its eight chemosensory systems. It therefore appears that the eight chemosensory systems provide a particular advantage for M. xanthus. The most logical conclusion derived from these observations is that the chemosensory systems are being utilized for temporal regulation of a variety of functions necessary for survival by M. xanthus. Those systems listed in Table 1 are therefore predicted to be crucial for survival and temporally regulated during all aspects of the M. xanthus life cycle. It is well known that a variety of organisms use multiple paralogous chemosensory systems including P. aeruginosa, R. sphaeroides, R. centenum, and M . xanthus. Evolution of paralogous chemosensory systems

145

is strictly correlated with motility. For each organism with multiple chemosensory systems, at least one system has been shown to regulate motility. Whether or not additional chemosensory systems affect motility varies from one organism to another. For example, it appears that all chemosensory systems in R. sphaeroides affect flagellum-based motility (Armitage, 2003) while those in P. aeruginosa and R. centenum regulate distinct functions. In that regard M. xanthus may represent a case study encompassing both extremes where several systems regulate motility and several systems regulate alternative functions. Analysis of chemosensory signal transduction systems in each model organism is an excellent example of the modular nature of signal transduction and the evolution of bacterial genomes. We thank Aaron Buldoc, Carolin Groeger, Marion Leclerc, and Hsu-Ming Wen for work on the various Che systems. We also thank Larry Shimkets, Zhaomin Yang, Heidi Kaplan, Mitch Singer, David Zusman, Trish Hartzell, and Dale Kaiser for strains, plasmids, and antibodies. Support for this work was provided by Grant A1059682 from the National Institutes of Health to1.K.

References Alley, M. R., J. R. Maddock, and L. Shapiro. 1992. Polar localization of a bacterial chenioreceptor. Genes Dev. 6532.5836. Ames, P., C. A. Studdert, R. H. Reiser, and J. S. Parkinson. 2002. Collaborative signaling by mixed chemoreceptor teams in Escherichia coli. Proc. Natl. Acad. Sci. USA 99: 7060-7065. Armitage, J. P. 2003. Taxing questions in development. Trends Microbiol. 11:239-242. Bassler, B. L. 2002. Small talk. Cell-to-cell communication in bacteria. Cell 109:421-442. Berleman, J. E., and C. E. Bauer. 2005. Involvement of a Che-like signal transduction cascade in regulating cyst cell development in Rhodospirillum centenum. Mol. Microbiol. 56A4.57-1466. Black, W. P., and Z. Yang. 2004. Myxococcus xanthus chemotaxis homologs DifD and DifG negatively regulate fibril polysaccharide production. /. Bacteriol. 186: 1001-1008. Bonner, P. J., Q. Xu, W. P. Black, Z. Li, Z. Yang, and L. J. Shimkets. 2005. The Dif chemosensory pathway is directly involved in phosphatidylethanolamine sensory transduction in Myxococcus xanthus. Mol. Microbiol. 57: 1499-1508. Bourret, R. B., and A. M. Stock. 2002. Molecular information processing: lessons from bacterial chemotaxis. J. Biol. Chem. 27E9625-9628. Bray, D., M. D. Levin, and C. J. Morton-Firth. 1998. Receptor clustering as a cellular mechanism to control sensitivity. Nature 393:85-88.

146 Brencic, A., and S. C. Winans. 2005. Detection of and response to signals involved in host-microbe interactions by plantassociated bacteria. Microbiol. Mol. Biol. Rev. 69:155-194. Cascales, E., A. Bernadac, M. Gavioli, J.-C. Lazzaroni, and R. Lloubes. 2002. Pal lipoprotein of Escherichia coli plays a major role in outer membrane integrity. 1. Bacteriol. 184~754-759. DiGiuseppe, P. A., and T. J. Silhavy. 2003. Signal detection and target gene induction by the CpxRA two-component system. J. Bacteriol. 185:2432-2440. Fairchild, C. D., J. Zhao, J. Zhou, S. E. Colson, D. A. Bryant, and A. N. Glazer. 1992. Phycocyanin alpha-subunit phycocyanobilin lyase. Proc. Natl. Acad. Sci. USA 89:7017-7021. Falke, J. J., and G. L. Hazelbauer. 2001. Transmembrane signaling in bacterial chemoreceptors. Trends Biochem. Sci. 26~257-265. Falke, J. J., R. B. Bass, S. L. Butler, S. A. Chervitz, and M. A. Danielson. 1997. The two-component signaling pathway of bacterial chemotaxis: a molecular view of signal transduction by receptors, kinases, and adaptation enzymes. Annu. Rev. Cell Dev. Biol. 13:457-512. Feng, X., A. A. Lilly, and G. L. Hazelbauer. 1999. Enhanced function conferred on low-abundance chemoreceptor Trg by a methyltransferase-docking site.]. Bacteriol. 181:31643171. Fontes, M., R. Ruiz-Vazquez, and F. J. Murillo. 1993. Growth phase dependence of the activation of a bacterial gene for carotenoids synthesis by blue light. EMBO J. 12:12651275. Fredrick, K. L., and J. D. Helmann. 1994. Dual chemotaxis signaling pathways in Bacillus subtilis: a sigma D-dependent gene encodes a novel protein with both Chew and CheY homologous domains.]. Bacteriol. 176:2727-2735. Hickman, J. W., D. F. Tifrea, and C. S. Harwood. 2005. A chemosensory system that regulates biofilm formation through modulation of cyclic diguanylate levels. Proc. Natl. Acad. Sci. USA 102:14422-14427. Hoch, J. A. 2000. Two-component and phosphorelay signal transduction. Curr. Opin. Microbiol. 3:165-170. Hoch, J. A., and T. J. Silhavy (ed.). 1995. Two-Component Signal Transduction. ASM Press, Washington, DC. Kirby, J. R., M. M. Saulmon, C. J. Kristich, and G. W. Ordal. 1999. CheY-dependent methylation of the asparagine receptor, McpB, during chemotaxis in Bacillus subtilis. 1. Biol. Chem. 274:11092-11100. Kirby, J. R., T. B. Niewold, S. Maloy, and G. W. Ordal. 2000. CheB is required for behavioural responses to negative stimuli during chemotaxis in Bacillus subtilis. Mol. Microbiol. 35:44-57. Kirby, J. R., and D. R. Zusman. 2003. Chemosensory regulation of developmental gene expression in Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 100:2008-2013. Konstantinidis, K. T., and J. M. Tiedje. 2004. Trends between gene content and genome size in prokaryotic species with larger genomes. Proc. Natl. Acad. Sci. U S A 101:31603165. Krukonis, E. S., and V. J. DiRita. 2003. From motility to virulence: sensing and responding to environmental signals in Vibrio cholerae. Curr. Opin. Microbiol. 6:186-190.

REGULATORY MECHANISMS Li, M., and G. L. Hazelbauer. 2005. Adaptational assistance in clusters of bacterial chemoreceptors. Mol. Microbiol. 56:1617-1626. Maddock, J. R., M. R. Alley, and L. Shapiro. 1993. Polarized cells, polar actions. J. Bacteriol. 175:7125-7129. McBride, M. J., and D. R. Zusman. 1993. FrzCD, a methylaccepting taxis protein from Myxococcus xanthus, shows modulated methylation during fruiting body formation. J. Bacteriol. 175:4936-4940. Mignot, T., J. P. Merlie, Jr., and D. R. Zusman. 2005. Regulated pole-to-pole oscillations of a bacterial gliding motility protein. Science 310:855-857. Parkinson, J. S., and E. C. Kofoid. 1992. Communication modules in bacterial signaling proteins. Annu. Rev. Genet. 26~71-112. Parkinson, J. S. 1993. Signal transduction schemes of bacteria. Cell 73:857-871. Piggw P. J., and D. W. Hilbert. 2004. Sporulation of Bacillus subtilis. Curr. Opin. Microbiol. 7579-586. Qin, L., S. Cai, Y. Zhu, and M. Inouye. 2003. Cysteinescanning analysis of the dimerization domain of EnvZ, an osmosensing histidine kinase. J. Bacteriol. 185:34293435. Rhie, H., and L. J. Shimkets. 1991. Low-temperature induction of Myxococcus xanthus developmental gene expression in wild-type and csgA suppressor cells. J . Bacteriol. 173:2206-2211. Sankaran, K., and H. C. Wu. 1995. Bacterial prolipoprotein signal peptidase. Methods Enzymol. 248:169-180. Seydel, A., P. Gounon, and A. P. Pugsley. 1999. Testing the “ + 2 rule” for lipoprotein sorting in the Escherichia coli cell envelope with a new genetic selection. Mol. Microbiol. 34:810-821. Sierakowska, A., H. Willenbrock, G. von Heijne, H. Nielsen, S. Brunak, and A. Krogh. 2003. Prediction of lipoprotein signal peptides in Gram-negative bacteria. Protein Sci. 12:1652-1662. Stock, A. M., V. L. Robinson, and P. N. Goudreau. 2000. Two-component signal transduction. Annu. Rev. Biochem. 69: 183-2 15. Szurmant, H., T. J. Muff, and G. W. Ordal. 2004. Bacillus subtilis CheC and FliY are members of a novel class of CheY-Phydrolyzing proteins in the chemotactic signal transduction cascade.]. Biol. Chem. 279:21787-21792. Tortosa, P., and D. Dubnau. 1999. Competence for transformation: a matter of taste. Curr. Opin. Microbiol. 2588-592. Ulrich, L., E. V. Koonin, and I. B. Zhulin. 2005. Onecomponent systems dominate signal transduction in prokaryotes. Trends Microbiol. 1352-56. Ulrich, L. E., and I. B. Zhulin. 2007. MIST: a microbial signal transduction database. Nucleic Acids Res. 35:D386-D390. Van Nimwegen, E. 2003. Scaling laws in the functional content of genomes. Trends Genet. 19:479-484. Vlamakis, H. C., J. R. Kirby, and D. R. Zusman. 2004. The Che4 pathway of Myxococcus xanthus regulates type IV pilus-mediated motility. Mol. Microbiol. 52:17991811. Wadhams, G. H., A. C. Martin, and J. P. Armitage. 2000. Identification and localization of a methyl-accepting

8. CHEMOSENSORY SIGNALTRANSDUCTION SYSTEMS IN M. XANTHUS chemotaxis protein in Rhodobacter sphaeroides. Mol. Micro biol. 36:1222-1233. Wall, D., S. S. Wu, and D. Kaiser. 1998. Contact stimulation of Tgl and Type IV pili in Myxococcus xanthus.]. Bacteriol. 180:759-76 1. Wu, S. S., and D. Kaiser. 1997. Regulation of expression of the PilA gene in Myxococcus xanthus. J. Bacteriol. 179:77487758.

147

Xu, Q., W. P. Black, S. M. Ward, and Z. Yang. 2005. Nitratedependent activation of the Dif signaling pathway of Myxococcus xanthus mediated by a NarX-DifA interspecies chimera. J. Bacteriol. 187:6410-6418. Yang, Z., X. Ma, L. Tong, H. B. Kaplan, L. J. Shimkets, and W. Shi. 2000. Myxococcus xanthus dif genes are required for biogenesis of cell surface fibrils essential for social gliding motility. J. Bacteriol. 1825793-579 8.

Myxobacteria: Multicellularity and Differentiation Edited by David E. Whitworth 0 2008 ASM Press, Washington, D.C.

Lee ICroos Sumiko Inouye

Transcriptional Regulatory Mechanisms during Myxococcus xanthus Development'

Since the writing of Myxobacteria I1 there has been tremendous progress in understanding transcriptional regulation of gene expression in Myxococcus xanthus. Most of the progress has involved studies of genes induced during development or during light-induced carotenoid biosynthesis. The latter is described in chapter 12. In this chapter, we focus primarily on transcriptional regulation of developmental genes. While the identification of developmentally regulated M. xanthus genes continues, now on a comprehensive genome-wide scale with the use of DNA microarray expression profiling (see chapter 28), an understanding of the cis-acting DNA elements (promoters and transcription factor binding sites) and trans-acting proteins (RNApolymerase [RNAP] with particular sigma factors, activators, and repressors) has emerged for a handful of developmental genes. Much more work will be necessary in order to achieve an understanding of the transcriptional regulatory network comparable to that learned from studies of Bacillus subtilis sporulation (see chapter 21) and the Caulobacter cell cycle (see chapter 22). The task seems daunting because analysis of the M. xanthus genome sequence reveals copious potential for transcriptional regulation (see chapter 16).On the other hand, the

9

potential for discovery of novel regulatory mechanisms and strategies makes the task irresistible, not only to satisfy curiosity but also to provide paradigms for less tractable organisms and communities (i.e., biofilms) of practical importance.

SIGMA FACTORS-d4

AND THE u7OFAMILY

In prokaryotes, sigma factors of RNAP play a key role in the regulation of gene expression by recognizing specific promoters and initiating transcription. There are two structurally unrelated families of sigma factors, the as4 and a 7 0 families (Helmann and Chamberlin, 1988). As in many bacteria, M. xanthus has a single as4(Goldman et al., 2006), but unlike in other bacteria, as4is ' essential for M . xanthus growth (Keseler and Kaiser, 1997). This has made it difficult to determine the role of d4 in development. Nevertheless, a number of genes, some crucial for development, have been predicted to be transcribed by aS4-RNAP(Table 1).In one case (spi), mutational analysis of the promoter supports the prediction (Keseler and Kaiser, 1995). In the other cases, the prediction is supported by the sequence of the promoter and/or genetic dependence of expression on a predicted

Lee Kroos, Department of Biochemistry and Molecular Biology, Michigan State University, East Lansing, MI 48824. Sumiko Inouye, Department of Biochemistry, Robert Wood Johnson Medical School, Piscataway, NJ 08854.

149

REGULATORY MECHANISMS

150 Table 1 Putative d4-RNAP-transcribed genes Gene or operon actAB CDE orf2 asgE sdeK spi mbhA pilA

mrpc

Promoter sequence"

Induction time (h)b

-29 T a C A C AACCA T m T -24 TCGCGGAGTCCGGEA -28 C E T G C A C A A G G G E T -26 C E C G C T C T C A G C E G -28 GAGCACGCGTCTmT -27 T E C A C G C CA T C m T -29 T E C A T G C G T A G m T -29 T a C A C G A A C C T E G A

2-4 2-4 0-2 2 10 veg 0-2

Inferred EBP

SasR PilR MrpB

Reference(s) Gronewold and Kaiser, 2001 Garza et al., 2000 Garza et al., 2000 Garza et al., 1998 Keseler and Kaiser, 1995; Guo et al., 2000 Romeo and Zusman, 1991 Wu and Kaiser, 1997 Sun and Shi, 2001a; Nariya and Inouye, 2005

OEach putative promoter sequence is shown with the number to the left indicating the position relative to the predicted start site of transcription. Bold nucleotides match consensus sequence, TGGYRYRNNNNTTSA, and underlined nucleotides match those most highly conserved in the consensus (Thony and Hennecke, the E. coli us4 1989). bThe approximate time during development when expression begins to increase is listed. This has not been determined for the actABCDE operon. The pilA gene is expressed during vegetative growth (Veg), and its expression does not increase during development.

enhancer-binding protein (EBP) (see below), though in no case has transcription been reconstituted in vitro. d4RNAP appears to play a critical role primarily early in development, though it is also believed to transcribe at least one gene ( m b h A )later in development (Table 1). The a 7 0 family is predicted to include 47 members in M. xanthus, an unusually large number even considering the large genome size (Goldman et al., 2006). This includes the primary sigma factor, SigA, presumed to be responsible for the majority of transcription in growing cells; five smaller sigma factors, SigB through SigF (SigB-F), closely related in sequence to each other and to SigA (Fig. 1);SigG, which is less similar to SigA-F but is similar to Escherichia coli FIiA; 38 sigma factors that appear to belong to the extracytoplasmic function (ECF) subfamily; and 2 sigma factors with limited similarity to known sigma factors.

SigA-G SigA shows high sequence similarity to E. coli u70,except that SigA has an extra -100 residues at its amino-terminal end (Inouye, 1990). SigA was originally described as 01 (Rudd and Zusman, 1982) and later shown to be the most abundant sigma factor in RNAP holoenzyme purified from growing M. xanthus cells (Biran and Kroos, 1997). Biran and Kroos (Biran and Kroos, 1997) further demonstrated that reconstituted oA-RNAP could initiate transcription accurately in vitro from the vegA promoter, which is active in growing M. xanthus (Komano et al., 1987). Two other promoters active in growing cells, ugh11 (Biran and Kroos, 1997) and ZonD (Ueki and Inouye, 2002), also appear to be utilized by oARNAP in vitro. The lonD gene is also called bsgA, and this heat shock inducible gene (Ueki and Inouye, 2002) encodes an ATP-dependent protease that is essential for

development (Gill et al., 1993; Tojo et al., 1993). The only developmentally regulated promoter that has been shown to be recognized by a*-RNAP in vitro is the promoter of an operon identified by Tn5 lac insertion R4.514 (Ha0 et al., 2002). Expression of this operon is induced at about 9 h into development (Kroos et al., 1986) and appears to involve both negative autoregulation and positive control (Ha0 et al., 2002). Another indication that SigA is active during development comes from the finding that a missense mutation in the sigA gene causes a defect in A-signal production (Davis et al., 1995). SigA is highly similar to E. coli 0 7 0 and B. subtilis 043in the regions expected to recognize -10 and -35 promoter sequences (Inouye, 1990), and promoters recognized by M. xanthus oA-RNAPin vitro match the 070/u43 consensus quite well in their -35 regions (Table 2). However, their -10 regions match the consensus poorly and are more GC rich, leading to speculation that 111. xanthus a*-RNAP better tolerates GC-rich - 10 sequences (Biran and Kroos, 1997; Hao et al., 2002). By using a synthetic oligonucleotide corresponding to the highly conserved region 2.2 of SigA as a probe for Southern blot analysis with chromosomal DNA digests, genes sigB-E were identified and subsequently isolated and characterized (Apelian and Inouye, 1990,1993; Ueki and Inouye, 1998, 2001). Of these, SigD is most similar to SigA (Fig. 1).SigD shares 35 and 34% identity with M. xanthus SigA and E. coli RpoS, respectively (Ueki and Inouye, 1998). SigD lacks region 1.1,which is found only in primary sigma factors and RpoS homologues (Gruber and Bryant, 1997), and most of region 3.1, which is typically conserved in all sigma factors except those in the ECF subfamily (Lonetto et al., 1994). By analyzing protein expression patterns and cell viability of a sigD deletion mutant during the late log and stationary phases, as

9. 211. XANTHUS TRANSCRIPTIONAL REGULATORY MECHANISMS SigA S igD SigE SigC S igB SiqF

151

................ MEAINLNVSFESPELWP

18

RpoH box SigA SigD SigE SigC SigB SigF SigG

IETINIQIRTSRYLVQE ... I ~ R E P .T. . .P E E I ~ E K M E L P L D ~ K V .

DSHLG.. ETTFL . .

.................................................

:."QAHG

DSRTTRP DATHL . . GNSHV.. EATRL . .

RTRRE E G D A "... A EPE IK~ R L L KASE ~ R ~EME T E Q ~ TRRE EKF SGDA.AVNVDDI~R KPGE EME . . . . .WVL_Q- RRERS EARWGEGHPEVE.KRL,E GKREDE LAM

s

. . . . . . . . . . . .N A W G ~ ~ Y L G N S L D R E A G A ~ N R ~ S S F D D D ~ D I S D A V T G L I E G T ~ D T A G Y T

. . . . . . . .D E S L P ~ I R M E Q

630 167 202 216 216 185 161

(core binding) Sign SigD SigE SigC SigB SigF SigG

PSRSKRLRSFVES !GVSGHPGPFP LINRMRDFMREQIPDFDLVASPKA LMAEAGVDESTLNA LMAEVDPEAVAAOO

------4

708 246 280 295 295 264 239

I------(-35 recognition)

Figure 1 Alignment of parts of the amino acid sequences of M. xanthus SigA-G. Amino acids identical in more than 50% of the sequences are indicated by a black background. Conserved subregions of sigma factors and their functions are denoted under the sequences (Helmann and Chamberlin, 1988; Lonetto et al., 1992). well as its response to various stresses (osmotic, oxidative, heat, and cold), it was concluded that SigD shows characteristic features of stationary-phase sigma factors (Ueki and Inouye, 1998). The deletion of sigD also affects protein synthesis patterns during early development, resulting in a 24-h delay in the initiation of fruiting body formation and a reduced spore yield (26% of the parent strain). Therefore, SigD is a stationary-phase sigma factor that is also required for multicellular development. The sigD mutant responds to starvation by inducing (p)ppGppsynthesis normally but is impaired for production of A- and C-signals (Viswanathan et al., 2006b). SigD is needed for cellular responses to A-signal, but the sigD mutant can respond to C-signal from codeveloping wild-type cells by inducing a subset of late developmental genes. Other late genes require oD-RNAPor a gene under its control cell autonomously. M. xanthus appears to have expanded

the repertoire of its stationary-phase sigma factor to include a role in the decision whether to initiate fruiting body formation in response to starvation. As shown in Fig. 1, SigB, SigC, and SigE exhibit high similarity to M. xanthus SigA and SigD and have Table 2 oA-RNAP-transcribedgenes Gene or operon

-35 region"

a4514 vegA

TTGACA TAGACA TTGCCA TTGCCA

aphII lonD "old

Spacer (bp)b

-10 regionc

18

TACCTA TAAGGG TAAGGT TACGTT

17 17 16

nucleotides match the E. coli c7'O-B. strbtilis u+3consensus sequence of

TTGACA. 'Distance between the -35 and -10 regions. R> Orphan

T

Ntr

redCDEF

Yes

Tu, Chew, R Tu, Chew Tu, Chew, R

Mono Mono Mono

Orphan R>H> R>K> H>R>

Yes

Tu, Chew, R

Mono

H>R> Orphan

Development Development A-motility Development Development S-motility S-motility A- and S-motility Development Development

actAB

Development

CheY

R, coiled coil R, coiled coil R

TodK

PhoR

PAS, PAS, T

RedF

CheY

R

RedE

NtrB

MXAN2671 MXAN0732 FrzZ RedD FruA CheY4 DifD FrzG

CheY CheY FrzZ FrzZ NarL CheY CheY CheB

R R R, R R, R R, -HTH-LUXR R R R, Me esterase

AsgA RodK

? ?

RedC

NtrB

CheA4 DifE FrzE

CheA CheA CheA

Yes

CheB3 CrdA

CheB NtrC

R, Me esterase R, AAA, HTH-8

CheA3

CheA

ActB

NtrC

R, AAA, HTH-8

FrgC MrpB Nlal Nla4 Nla6 Nla23 Nla24

NtrC NtrC NtrC NtrC NtrC NtrC NtrC

R, AAA, H T H 8 R, AAA, H T H 8 R, AAA, HTH-8 R, AAA, HTH-8 R, AAA, HTH-8 R, AAA, H T H 8 R, AAA, HTH-8

FrgB MrpA MXAN5852

NtrB NtrB NtrB

TM

MXAN4043 MXANS778 MXAN7439

NtrB NtrB NtrB

TM TM

Nla28 PilR SasR SpdR HsfA

NtrC NtrC NtrC NtrC NtrC

R, AAA, HTH-8 R, AAA, H T H 8 R, AAA, HTH-8 R, AAA, H T H 8 R, AAA, HTH-8

MXANll66 PilS Sass SpdS HsfB

PhoR NtrB NtrB NtrB NtrB

MXAN5366

PleD

R, GGDEF

K>R> K>R> K>R> K>R> mxan5366> hsfBA> mxan53 66>

R, GGDEF

hsfBA> actAB

ActA

PleD

Locus phenotype

Genes

Yes Yes

TM

TM TM Yes

T T PutP, PAS, T

Ntr Ntr Bi

T T HAMP, T

Bi Bi Ntr

T PAS, T HAMP, T PAS, GAF, T R, T

Bi Ntr Ntr Ntr Bi

K>R> K>R> K>R> Orphan K>R> K>R> K>R>

?

Development S-motility Development Development S-motility A- and S-motility Development S-motility Development Development Heat shock

Reference Yang et al., 2004 Ward et al., 2000 Rasmussen and S.-Andersen, 2003 Higgs et al., 2005 Li and Plamann, 1996 Rasmussen et al., 2005 Trudeau et al., 1996 Higgs et al., 200.5 Ellehauge et al., 1998 Vlamakis et al., 2004 Yang et al., 1998 McCleary and Zusman, 1990 Kirby and Zusman, 2003 Kirby and Zusman, 2003 Gronewold and Kaiser, 2001 Cho et al., 2000 Sun and Shi, 2001 Caberoy et al., 2003 Caberoy et al., 2003 Caberoy et al., 2003 Caberoy et al., 2003 Lancer0 et al., 2004 Caberoy et al., 2003 Wu and Kaiser, 1997 Guo et al., 2000 Hager and Gill, 2001 Ueki and Inouye, 2002

?

Development

Gronewold and Kaiser, 2001

(Continued)

Table 1 TCSs of M. xanthusa (Continued)

RR

Family

RR domains

HK

Family

Hybrid?

HK Domains

TM?

P’ase

Genes

Locus phenotype

Reference

Ntr

Orphan

Development

Cho and Zusman, 1999a

Ntr

Orphan

Development

Lee et al., 2005

TM

FHA, PAS, PAS, T, R MASE1, PAS, T, R LBD,T, R

Ntr

Orphan

Kimura et al., 2003

EspA

NtrB

Yes

EspC

NtrB

Yes

TM

MokA

NtrB

Yes

PhoPl

PhoB

R, Trans-regC

PhoRl

PhoR

TM

T

Ntr

R>K>

Osmotic shock Development

PhoP2

PhoB

R, Trans-regC

PhoR2

PhoR

TM

T

Bi

K>R>

Development

PhoP3

PhoB

R, Trans-reLC

PhoR3

PhoR

TM

T

Bi

K>R>

Development

PhoP4

PhoB

R, Trans-reLC SdeK AsgD

PhoR

Bi Bi

Orphan Orphan asgD >

Development Development Development

MXAN6994

PhoR

PAS, T R,GAF, GAF, T T

Bi

asgD >

?

Yes

Carrero-Lerida et al., 2005 Moraleda-Muiioz et al., 2003 Moraleda-Muiioz et al., 2003 Pham et al., 2006 Pollack and Singer, 2001 Cho and Zusman, 1999b

mxan6994>

?

mxan6994>

SocD CyaB

R.CYCc

PhoR

PAS, GAF, T

Bi

Complex

Development

S. aurantiaca

?

Shimkets, 1999 Coudart-Cavalli et al., 1997

dExperimentallycharacterized TCS proteins of M. xanthus, grouped according to response regulator family (based on domain organization). Partner histidine kinases were assigned to response regulators on the basis of gene adjacency, unless additional evidence was available. Histidine kinase family membership is based on phylogenetic analyses, and the presence of transmembrane helices was determined using TMHMM v2.0 (Sonnhammer et al., 1998). The ability of a histidine kinase to act as a phosphatase (P’ase) was predicted using the method of Alves and Savageau (2003) as described in the text. In descriptions of domain architecture, some domain names were abbreviated as follows: R, receiver; T, orthodox transmitter; Tu, unorthodox transmitter; LBD, ligand binding domain. Arrowheads are used in descriptions of gene organization to represent relative directions of transcription. H, hybrid kinase; R, response regulator; K, histidine kinase.

10. TWO-COMPONENT SIGNALTRANSDUCTION SYSTEMS during fruiting body development (including regulation of motility), although several have been shown to act during vegetative growth. This situation contrasts with other bacteria, where most TCSs maintain homeostasis, and a minority are involved in complex regulated processes such as sporulation or cell cycle progression (for a review see Hoch and Silhavy, 1995). Such a distinction may reflect a bias in myxobacterial research (which largely focuses on development). However, myxobacterial TCSs are known which have roles in both vegetative growth and in development (Ueki and Inouye, 2002; Kimura et al., 2001). In a recent study a large number of M. xanthus NtrC-like response regulator genes were inactivated and their phenotypes were assessed (Caberoy et al., 2003). Of the 28 genes mutated, only 8 gave developmental and/or motility phenotypes, suggesting that most TCSs of M. xanthus either are not involved in development/motility or are redundant under laboratory conditions. The recent availability of genome sequences of four myxobacteria has enabled a comparative genomic analysis of TCS genes in myxobacterial genomes. According to an automated search protocol (Cock and Whitworth, 2007), the M. xanthus genome encodes 276 TCS proteins, and a similar number can be found in the genomes of Sorangium cellulosum and Stigmatella aurantiaca, with nearly 200 TCS proteins encoded by the smaller Anaeromyxobacter dehalogenans genome. For comparison, Escherichia coli and Bacillus subtilis possess 62 and 70 TCS proteins, respectively (Mizuno, 1997; Fabret et al., 1999).Figure 2 shows the number of TCS genes as a function of genome size for 228 bacterial genomes, including four myxobacteria. All four myxobacterial genomes possess significantly more TCS genes than would be predicted on the basis of their genome size alone and thus possess a high IQ as defined by Galperin (2005). Table 1summarizes the TCSs of M. xanthus that have been characterized experimentally. Pertinent features of each system are described, including their genomic organization, function, domain architecture, and predicted localization. Partnerships between histidine kinase and response regulators in Table 1 are described on the basis of gene adjacency unless further evidence is available in the literature. The large number of characterized M. xanthus TCS proteins precludes an exhaustive description of each system; however, general features of myxobacterial TCS properties can now be discerned and their description forms the basis for the remainder of this chapter.

Receiver and Transmitter Domains Amino acid sequences of transmitter and receiver domains are highly conserved, and several critical motifs

173 Sorangium cellulosum Stigmatella aurantiaca . . Myxococcus xanthus ....... . .

....

400

.

... ... ...... .. ,.

...

Anaeromyxobacter dehalogenans :,

.... ...

... ..

1

350

.. ...

..

. '... ..... ... ... ..... .... ... ... . .

300

d In 2

250

Nostoc

200

Geobactee

*-

2

n

E

150 100

3

50 0

r

0

1

2

4

6

8

10

12

14

Genome Size (Mbp)

Figure 2 Numbers of TCS genes found in different genomes as a function of genome size. Trend lines are shown for all bacteria (gray) and for four myxobacteria (black). The myxobacteria all possess an exceptionally large complement of TCS genes, given their genome size. Nostoc (PCC7120) and Geobacter sulfurreducens (PCA) also have relatively large numbers of TCS genes.

and residues within them have been identified (Parkinson and Kofoid, 1992). Receiver domains are characterized by a series of acidic residues and a lysine residue (D12, D13, D57, and K109 of E. coli CheY), while orthodox transmitter domains possess a series of sequence motifs called the H, N, D, F, and G boxes. The H box is also referred to as a HisKA domain (histidine kinase phosphoacceptor), while the N, D, F, and G boxes are together described as an HATPase domain (histidine kinase-type ATPase). Transmitter domains operate as dimers, with dimerization mediated by the H-box/HisKA region and autophosphorylation occurring in trans (for a review see Wolanin et al., 2002). Nearly all M. xanthus receiver domains (described in Table 1)contain the defining series of amino acids as found in E. coli CheY (D12/13, D57, and K109). However, ActA, AglZ, FruA, and FrzS contain only a single acidic residue equivalent to CheY D12/13 (FrzG lacks both acidic residues), while AglZ, FrzS, and MXAN5366 lack the phosphorylated aspartate equivalent to CheY D.57 (AglZ and MXAN5366 also lack a basic residue equivalent to CheY K109). Lack of a phospho-accepting aspartate residue suggests that the activity of AglZ, FrzS, and MXAN5366 might not be regulated by phosphorylation,

REGULATORY MECHANISMS

174 while the lack of conserved residues in ActA, FrzG, and FruA suggests that they might not be able to become phosphorylated either. None of these proteins have been shown to phosphorylate in vitro, though genetic evidence suggests that the putative phospho-accepting aspartate residue of FruA is required for function (Ellehauge et al., 1998).AglZ and FrzS share the same domain architecture, are both isolated in the genome, have no known partner kinase (although FrzS is probably regulated by the histidine kinase FrzE), and are both implicated in regulating directed motility. FrzS has recently been shown to dynamically relocalize from pole to pole during cell reversals (Mignot et al., 2005), and it would be interesting to determine whether AglZ exhibits similar dynamic properties. It has been proposed (Yang et al., 2004) that an alternative aspartate residue found in both FrzS and AglZ might be phosphorylated instead of the normal residue, corresponding to position 52 of CheY, instead of position 57. However, position 52 does not encode an aspartate residue in MXAN5366 (the other M . xanthus receiver domain lacking an aspartate at position 57), and by superimposing on known structures of receiver domains (for example, Feher et al., 1997),residue 52 would be predicted to lie on the face of the receiver domain opposite to position 57. It is tempting to speculate on possible alternative mechanisms responsible for regulating the activity of AglZ, FrzS, and MXAN5366. Phosphotransfer between histidine kinases and response regulators requires appropriate proteinprotein interactions, and it is possible that such interactions could be regulatory even in the absence of phosphotransfer (altering the conformation of the receiver domain to an "active" form equivalent to that typically resulting from phosphorylation). Various lines of evidence have allowed suggestions of potential partners for all three proteins; FrzE for FrzS (Bustamante et al., 2004), MglA for AglZ (Yang et al., 2004), and HsfB for MXAN.5366 (predicted by gene adjacency), two of which are histidine kinases (MglA is a GTPase). An alternative possibility is that these unusual proteins have evolved from response regulators and are no longer regulated by phosphorylation in the fashion suggested by the TCS paradigm. Phylogenetic analysis of receiver domains suggests that AglZ and FrzS are paralogues and that S. aurantiaca, S. cellulosum, and A. dehalogenans each possess orthologues of AglZ and FrzS. It is interesting that aglZ of S. cellulosum lacks a coiled-coil domain but is fused to a gene encoding a transmitter domain, which is particularly similar to SocD and MXAN5990 of M. xanthus. In orthodox transmitter domains, the H-box/HisI200 complete bacterial genomes sequenced, 74% of bacterial histidine kinases are found to contain transmembrane helices and would be predicted to possess regions extending into the periplasmic space. The percentage of transmembrane hybrid kinases is slightly lower (52%). Of the TCSs described in Table 1, only 45% of histidine kinases and 20% of hybrid kinases are predicted or known to be transmembrane (or 54 and 8%, respectively, if we consider all histidine and hybrid kinases encoded in the genome of M. xanthus). Figure 6 shows the numbers of histidine kinases in different genomes, compared with the number of those histidine kinases predicted to be transmembrane. Most myxobacteria (with the exception of A. dehalogenans) possess a surprisingly small proportion of transmembrane histidine kinases, implying an unusual degree of intracellular sensing, matched only by members of the cyanobacteria (Galperin, 2005). The presence of transmembrane helices in a histidine kinase suggests that the signal being sensed is extracellular or periplasmic, while those histidine kinases lacking transmembrane regions would most likely be sensing a cytoplasmic signal. There are well-known exceptions to this rule of thumb. CheA homologues are known to associate with membranebound receptor complexes that detect extracellular signals and lack their own transmembrane helices. Similarly, the transmembrane helices of the histidine kinase KdpD have been shown not to be involved in sensing (Heermann et al., 2003). AsgA, EspA, FrgB, HsfB, MrpA, RedE, RodK, SdeK, SocD, SpdS, and TodK do not possess transmembrane helices, implying that they respond to internal signals (Hsfl3responds indirectly to heat shock). Conversely, EspC, MokA, PhoR1, PhoR2, PhoR3, PilS, RedC, and Sass all contain transmembrane helices, implying that they respond to extracytoplasmic stimuli.

Input Functions Many histidine kinases possess conserved N-terminal domains, which appear to be important for signal perception, acting as input domains. Some input domains are commonly found in large numbers of histidine kinases, whereas others are relatively rare. The commonest input domain is the PAS domain, which usually

REGULATORY MECHANISMS

182 120 100 -

80 In 8 In

m K

60

E

E

40 20 0

0

50

100

150

200

Total kinases Figure 6 Transmembrane histidine kinases (includinghybrid kinases) found in different genomes as a function of total numbers of histidine kinases (including hybrids). Trend lines are shown for all bacteria (gray) and for four myxobacteria (black).A d , Anaeromyxobacter dehalogenans; M x , Myxococcus xanthus; Sa, Stigmatella aurantiaca; Sc, Sorangium cellulosum. Most myxobacteria (with the exception of A. dehalogenans) have exceptionally low proportions of TM histidine kinases, implying an unusual degree of sensing of intracellular conditions. Two genomes of Nostoc sp. also exhibit evidence of significant intracellular sensing (see Galperin, 2005).

senses redox potential (Taylor and Zhulin, 1999). Two PAS domains are found in EspA and TodK, while EspC, MXAN5852, PilS, SdeK, and SocD each possess a single PAS domain. Other frequently found input domains are HAMP domains, which act as signal transducing linker domains (found in MXAN7439 and Sass) and GAF domains, which bind cyclic GMP (Galperin et al., 2001). GAF domains are found in AsgD (2 copies), SocD, and S/pdS. CheA homologues frequently possess CheW-like input domains, including the four CheA homologues of M. xanthus described in Table 1. Several histidine kinases of M. xanthus also possess relatively rare input domains. EspA contains an N-terminal FHA domain and is one of only four TCS proteins in GenBank to do so. The presence of an FHA domain in EspA and the lack of transmembrane helices suggest that EspA is activated by binding to specific phosphothreonine residues of a protein(s) within the cytoplasm. EspC also contains an unusual input domain (MASE1), an integral membrane sensor domain, which is found only 13 times in nonmyxobacterial TCS proteins in GenBank. MXAN5852 possesses an N-terminal Na+/proline symporter domain (COG0591j, while MokA contains a predicted periplasmic ligand-binding sensor domain (COG3292j. Several histidine kinases (andhybrid kinases) identified in the M . xanthus genome sequence possess uncommon

input domains. In one protein (MXAN4053) a Ser/Thr kinase domain, an ATPase domain, and a GAF domain precede the transmitter domain, which is a relatively unusual domain architecture (around 50 cases in GenBank). Other input domains found among the histidine kinases include a CHASE domain in MXAN6941 (predicted to be a ligand-binding domain) and a short coiled-coil domain in MXAN2606 (an unusual protein containing two transmitter domains separated by a receiver domain). One histidine kinase (MXAN0168) possesses a KdpD input domain and is preceded by a series of kdp gene orthologues (kdpFABC) implying a role in osmosensitive K+ channel regulation. Intriguingly, the putative kdpD orthologue is not adjacent to the usual partner response regulator gene (KdpEj. The M . xanthus genome also encodes two hybrid kinases, MXAN0712 and MXAN673.5 (each containing a single transmitter domain, upstream of three receiver domains), with a large number of tandemly repeated HAMP/Tar domains (10 and 14 HAMP domains, respectively), a geometry which again is relatively unusual (around 50 cases in GenBank).

Output Functions Most response regulator output domains are DNAbinding domains, and the proteins typically exhibit

10. TWO-COMPONENT SIGNALTRANSDUCTION SYSTEMS phosphorylation-dependent regulation of gene expression. The PhoB, NtrC, and NarL families of response regulators each possess different types of DNA-binding domains (Trans-reg-C, HTH-8, and HTH-LUXR, respectively), and several examples of each are found in the M. xanthus genome. However, despite the large number of myxobacterial DNA-binding response regulators described in the literature (Table l),the genes directly regulated by such proteins are known in very few cases. The NarL-like response regulator FruA has been shown to be required for the production of at least seven developmental proteins, including protein S and DofA (Horiuchi et al., 2002). The DNA-binding domain of FruA bound in vitro to the dofA and fdgA promoters, indicating that they are direct targets of FruA regulation (Ueki and Inouye, 2005a, 2005b). The NtrC homologue HsfA binds to the promoter region of the lonD heat shock gene and is able to initiate transcription from that promoter in vitro (Ueki and Inouye, 2002). As a final example, the pilA gene, encoding pilin, requires the NtrC homologue PilR for expression, and a d4-like promoter sequence has been identified upstream of pilA, suggesting direct regulation by PilR (Wu and Kaiser, 1997). Three non-DNA-binding output domains have also been identified in myxococcal TCS proteins: GGDEF, methyl esterase, and coiled-coil domains (Table 1). GGDEF domains are found in ActB and MXAN5366 and typically have diguanylate cyclase activity, producing cyclic-di-GMP as a signaling molecule. The genome sequence of M. xanthus encodes at least 11TCS proteins containing GGDEF domains, suggesting an extensive integration of TCS and cyclic-di-GMP signaling networks in the myxobacteria. However, none of the predicted TCS proteins of M. xanthus contain EAL domains which possess cyclic-di-GMP phosphodiesterase activity ( Galperin et al., 2001). CheB homologues are characterized by a Cterminal methyl esterase domain and are responsible for modulating chemotactic receptor sensitivity by reversible demethylation. Two orthologues of CheB are known in M. xanthus. One operates within the Frz system (FrzG), while the other is encoded within the che3 chemotactic gene cluster (CheB3). Coiled-coil output domains are found within the paralogous proteins FrzS and AglZ, which regulate motility. It has been suggested that the coiled-coil domain allows interaction with components of the motility systems (Yang et al., 2004). A survey of the TCS genes in M. xanthus identified several response regulators with unusual output domains. Two proteins, MXAN4257 and MXAN5053, contained GAF and GGDEF domains C-terminal to a receiver domain. Four cases with such domain architecture were found in GenBank, all from members

183

of the Deltaproteobacteria. One response regulator (MXAN5592) contained a putative C-terminal HTHXre domain, while another (MXAN4717) contained C-terminal DnaJ and TPR domains, both domain architectures without examples in GenBank, although with orthologues in other myxobacteria. One protein (MXAN6032) contains an N-terminal receiver domain and a Germinal Chew domain. While the combination of single Chew and receiver domains is common, the domains are usually found in the opposite orientation. In fact the case with an N-terminal receiver domain does not appear in GenBank. In one response regulator (MXAN2807),GSPII-E-N and HDc domains are found N-terminal to a receiver domain. GSPII-E-N domains are found in proteins typically involved with type IV pilus biogenesis, suggesting a role in motility, while HDc domains are thought to act as metal-dependent phosphohydrolases. Seven examples of proteins with similar domain architectures appear in GenBank, all from members of the Deltaproteobacteria.

COMPLEX SYSTEMS A typical TCS comprises a single histidine kinase that brings about phosphorylation of a partner response regulator through a His-to-Asp phosphotransfer reaction. However, well-characterized systems suggest that TCS pathways do not always act in such a straightforward and simple manner. The TCSs shown in Fig. 1 exhibit phosphorylation of multiple response regulators by a single histidine kinase, phosphorylation of a single response regulator by multiple histidine kinases, and flow of phosphoryl groups from Asp to His as well as from His to Asp. Such phenomena are possible because of the structural similarity of all receiver domains and all transmitter domains. Consequently TCS proteins can interact in a multitude of ways to create a diverse set of signaling pathway structures, of which the typical TCS is the simplest case. There are many factors which suggest that M. xanthus has complex TCS pathways. First, it possesses an exceptionally large number of response regulators that consist solely of a receiver domain. These response regulators most probably act as intermediaries in phosphotransfer schemes, shuttling phosphoryl groups between different histidine phospho-accepting TCS proteins. Second, an extremely large number of TCS genes are orphaned or in complex TCS gene clusters in the M. xanthus genome. Genes for histidine kinases and response regulators that have single partners are typically found to be adjacent, so the lack of partner genes suggests that orphan TCSs are likely to have multiple partners. By a similar argument,

184 it might be expected that the proteins encoded in complex gene clusters act together, as exemplified by the RedCDEF system (Higgs et al., 2005), necessarily forming complex pathways. A third phenomenon that is suggestive of complex pathways is the failure to find partner proteins for TCS proteins with major phenotypes. For instance, even with much effort partner proteins have not been identified for the response regulator FruA or the histidine kinase SdeK, despite both proteins being critical for multicellular development. If each protein had a single partner, it would be expected that the partner would be found relatively easily in screens of randomly generated mutations. However, if each protein had multiple “redundant” partners, each of which contributed to signaling, then deletion of a single partner might be insufficient to generate the expected phenotype. It seems plausible that proteins such as FruA and SdeK, which regulate key stages of development, might have multiple partners, by analogy to the sporulation phosphorelay of B. subtilis, where multiple kinases affect the phosphorylation state of the master regulator SpoOA (Fig. 1B). A further argument for complex TCSs can be obtained by addressing the frequency with which TCS gene deletions give rise to phenotypes. If we are correct in our assumption that the relative abundance of TCSs found in 111.xanthus is due to requirements for the regulation of motility, development, and/or sporulation, then we might expect the majority of TCS gene disruptions to generate a developmental phenotype. During a systematic disruption of NtrC homologues, Caberoy et al. (2003) found that less than one-third of gene disruptions gave an apparent phenotype. The low proportion of phenotypes observed could be indicative of redundancy in those TCS signaling pathways and thus of multiple partnerships between the TCS proteins. Finally, supporting evidence for complex TCS pathways is also provided by an enrichment for paired TCSs in the list of characterized systems (63%), in comparison to those identified in the genome sequence (34%). This implies that paired systems generally give stronger phenotypes than those TCS proteins encoded by orphan genes or in TCS gene clusters, which would be a likely consequence of genetic redundancy in the pathways encoded by orphans and gene clusters. The domain architecture of individual TCS proteins can also be suggestive of complexity beyond that of the typical TCS. For instance the hybrid histidine kinase RodK has a single transmitter domain and three receiver domains. The most plausible model for RodK action is that the three receiver domains compete as substrates for phosphoryl group transfer from the RodK transmitter domain and that the effector activity of RodK is determined by the phosphorylation state of its receiver

REGULATORY MECHANISMS domains (Rasmussen et al., 2005). An example where a similar mechanism is known to operate is the Vir system of Agrobacterium tumefaciens. VirA is a hybrid kinase containing single transmitter and receiver domains. The response-eliciting component of the Vir system is VirG, a PhoB-like response regulator. The receiver domain of VirA inhibits phosphotransfer from VirA to VirG; however, phosphorylation of the VirA receiver domain relieves the inhibition, allowing phosphorylation of VirG (Chang et al., 1996). Competition between multiple receiver domains to indirectly modulate the phosphorylation state of a response-eliciting receiver domain has also been observed in the chemotactic system of Helicobacter pylori. In this system, retrophosphorylation of CheAY2 occurs by phosphotransfer from the CheY1 receiver domain, and it is consequently thought that the CheY2 receiver domain acts as a phosphate sink to modulate the half-life of phosphorylated CheY1 ( CheYl-P) (J‘1menezPearson et al., 2005). Similar mechanisms can be invoked to provide a simple model of signal transduction through the RedCDEF system, as described above, which contains a response regulator with two receiver domains. There are 16 TCS genes in the M. xanthus genome that are predicted to encode multiple receiver domains (13 in S. cellulosum and 19 in S. aurantiaca), while one gene encodes multiple transmitter domains ( 5 in S. cellulosum and 7 in S. aurantiaca).It seems probable that the signaling pathways in which these proteins act will operate by mechanisms that are dependent upon competition between multiple transmitter-receiver domain interactions and presumably will exhibit dynamic behaviors very different from those of typical TCSs.

DOWN-REGULATION OF PHOSPHORYLATION The phosphorylation state of the proteins within a TCS dictates the strength of the signal passing through that pathway. Generally TCSs become activated, upon receipt of appropriate environmental cues, by a regulated increase in kinase activity. However, in many TCSs regulated phosphatase activities and/or the regulated inhibition of kinase activity is just as important in modulating signal strength. TCS phosphatase activities can act upon phosphorylated histidine or aspartate residues and are found to occur within histidine kinases and response regulators and as separate modulatory proteins. An inhibitor of kinase activity, KipI, has been identified in the B. subtilis sporulation phosphorelay. KipI inhibits KinA kinase activity, preventing sporulation (Wang et al., 1997). A gene encoding a putative KipI homologue is

10. TWO-COMPONENT SIGNALTRANSDUCTION SYSTEMS present in myxobacterial genomes, though it is not found in the vicinity of any TCS genes.

Phosphoaspartate Phosphatases Phosphorylated response regulators typically possess intrinsic autophosphatase activity. In some cases this can be very fast, but it is often slow, depending on the dynamic requirements of the process being regulated. For example, E. coli CheB-P has a half-life of 2 s (Stewart, 1997), while KdpD of E. coli has no significant autophosphatase activity (Puppe et al., 1996). Large variations in phosphatase activity have been observed for different response regulators using in vitro phosphorylation assays (Skerker et al., 2005). This suggests that in some cases, failure to observe phosphorylation of TCS proteins in vitro might be due to high intrinsic phosphatase activities of the proteins rather than an inability to phosphorylate. It is intriguing that for most M. xanthus TCS proteins (with the exception of FrzE and HsfA [Acuna et al., 1995; Ueki and Inouye, 2002]),there is no evidence of aspartate phosphorylation in vitro, though genetic and sequence analyses often suggest the importance of phosphorylation site residues (see, for example, Li and Plamann, 1996, and Ellehauge et al., 1998). Phosphoaspartate residues in response regulators can also be hydrolyzed by extrinsic phosphatases. For example the phosphorelay governing sporulation in B. subtilis contains two receiver domains, which can each be dephosphorylated by three phosphatases: SpoOE, YisI, and YnzD act on SpoOA-P, while RapA, RapB, and RapE act upon SpoOF-P (Perego, 1998). In addition to sporulation pathways, response regulator phosphatases are also important components of chemotactic signaling pathways in enteric bacteria (CheZ orthologues), and a putative CheY-P phosphatase has been identified in the 211. xanthus Dif pathway (DifG [Yang and Li, 2005]), though not in the Frz pathway (McBride et al., 1992). Surprisingly, there are no homologues of the Rap, SpoOE, YisI, YnzD, or CheZ phosphatases encoded in the M. xanthus genome, suggesting that modulation of signal flow by regulated phosphoaspartate phosphatase activity is not generally adopted by the myxobacteria.

Phosphohistidine Phosphatases In contrast to response regulators, phosphorylated histidine kinases appear to be relatively long-lived species in isolation; however, they typically each have at least one specific phosphatase in vivo-their partner response regulator(s).However, phosphohistidine phosphatase proteins are known, the best-studied example being SixA of E. coli. SixA is known to dephosphorylate an Hpt domain within the Arc phosphorelay (Matsubara and Mizuno,

185

2000). There are two homologues of SixA encoded in each available myxobacterial genome (except for a single homologue in A. dehalogenans), but they are typically found to be separated from any TCS genes. The exception to this generalization is M. xanthus, where one of the SixA homologues is encoded close to the phoRPl genes.

Bifunctional Histidine Kinases In a typical histidine kinase, the presence/absence of appropriate environmental signals acts as an on/off switch regulating kinase activity of the transmitter domain. However, some histidine kinases also possess phosphatase activity towards their partner response regulators when they are inactive as kinases. In these bifunctional proteins, both kinase and phosphatase activities reside within the transmitter domain, but phosphatase activity exists at a discrete site from kinase activity, indicating that phosphatase activity is more than just a consequence of retrophosphorylation (Hsing and Silhavy, 1997; Kramer and Weiss, 1999).Phosphorylation of response regulators can occur independently of the partner histidine kinase by interactions with noncognate histidine kinases or small molecule phospho-donors such as acetylphosphate. It is thought that phosphatase activity in bifunctional transmitter domains acts to reduce the phosphorylation of the partner response regulator by noncognate sources, while the partner response regulators of monofunctional kinases are efficient at integrating signals from multiple sources (Alves and Savageau, 2003). Alves and Savageau (2003) have devised an approach to predict whether a transmitter domain is monofunctional or bifunctional by modeling transmitter domain sequences onto known monofunctional (CheA) or bifunctional (EnvZ) transmitter domain structures. The results of such an analysis are shown in Table 1 for experimentally characterized histidine kinases of M . xanthus, which are consequently described as Mono, Bi, or Ntr. Ntr-type transmitter domains would be expected to be bifunctional in the presence of a PI1 protein, but otherwise monofunctional (Alvesand Savageau, 2003). As M. xanthus encodes no obvious PI1 homologue, it is likely that the Ntr-type proteins are monofunctional. Nevertheless, one example suggests that this assumption is invalid. Expression of pilA is dependent on the PilRS TCS, and pilA expression is increased in a pilS mutant, despite being reduced in a pilR background (Wu and Kaiser, 1997).Such behavior suggests that PilS has a bifunctional transmitter domain, despite being predicted to be Ntr-type (Table l),which may in turn imply that there is a PI1 analogue for the Pi1 system in M . xanthus. It would be extremely interesting to know definitively whether particular histidine kinases do have phosphatase

REGULATORY MECHANISMS

186 activity, as such properties have a fundamental impact on the dynamic properties of a TCS, and such knowledge would enable inferences to be drawn about the integration of multiple signals by different TCS pathways.

TCSs OF OTHER MYXOBACTERIA Only one myxobacterial TCS protein has been described from a species other than M. xantbus. The CyaB protein of S. aurantiaca contains an N-terminal receiver domain and a C-terminal CyaA adenylate cyclase domain (CoudartCavalli et al., 1997). Site-directed mutagenesis of the phospho-accepting aspartate residue in the CyaB receiver domain significantly reduced adenylate cyclase activity of the protein, implying that phosphorylation of the receiver domain stimulates cyclic AMP production. CyclicAMP is known to transiently accumulate during early development in S. aurantiaca (Coudart-Cavalli et al., 1997) and M . xanthus (Yajko and Zusman, 1978). An orthologue of cyaB exists in the M . xanthus genome alongside a histidine kinase gene, while S. cellulosum possesses two cyaB orthologues (one orphan gene and one adjacent to a hybrid kinase). The myxobacteria seem to possess multiple TCS proteins that contain cyclic nucleotide metabolism domains. In addition to a cyaB orthologue, the M . xantbus genome includes the gene for a response regulator with an N-terminal Ser/Thr kinase domain, a central receiver domain, and a C-terminal CyaA domain. Orthologues of this gene are apparent in the S. cellulosum genome and in S. aurantiaca (three paralogues). Additionally, M . xantbus possesses the gene for a histidine kinase which has an N-terminal cyclic nucleotide binding domain, implying that TCSs of M . xanthus are responding to changes in cyclic nucleotide levels, as well as regulating them. With the sequencing of multiple myxobacterial genomes it has become possible to use comparative genomics to gain novel insights into the TCSs of myxobacteria. Genomes can be assessed for the presence or absence of specific TCS homologues, lineage-specific changes in TCS properties can be identified, and in some cases changes in gene organization can guide searches for partner proteins. The sequences of multiple myxobacterial genomes can also provide important insights into the evolution of TCSs, from which inferences can be made regarding evolutionary changes in TCS network connectivity and the properties of contemporary TCSs.

PERSPECTIVES The myxobacteria possess exceptionally large numbers of TCS genes, which is not merely due to their large genome sizes (Fig. 2). Many TCS proteins of M . xantbus

have been characterized experimentally, with most being involved with the regulation of motility and/or development (Table 1).There is accumulating evidence that the TCS pathways of M. xantbus are relatively complex. The large number of response regulators that lack output domains, an absence of conventional phosphorelays, large numbers of TCS genes found as orphans or in complex gene clusters (Table 2; Fig. 4), and the complex domain architectures of myxobacterial TCS proteins all suggest that the TCSs of myxobacteria operate in significantly different ways from most other bacteria (with the notable exception of the cyanobacteria). One particularly unusual feature of myxobacterial TCSs is the adoption of chemotactic signaling pathways to regulate gene expression, a behavior which so far has been observed only in M. xanthus. The low proportion of sensor kinases with transmembrane helices implies that myxobacteria are unusually sensitive to changes in their internal state. This observation and their large numbers of TCS proteins suggests that the myxobacteria are complex introverts with high IQs, making them excellent role models as well as model organisms for cellular signaling studies.

References Acuna, G., W. Shi, K. Trudeau, and D. R. Zusman. 1995. The ‘CheA’ and ‘CheY’ domains of Myxococcus xanthus FrzE function independently in vitro as an autokinase and a phosphate acceptor respectively. FEBS Lett. 16:3 1-33. Alves, R., and M. A. Savageau. 2003. Comparative analysis of prototype two-component systems with either bifunctional or monofunctional sensors: differences in molecular structure and physiological function. Mol. Microbiol. 48:25-51. Burbulys, D., K. A. Trach, and J. A. Hoch. 1991. Initiation of sporulation in B. subtilis is controlled by a multicomponent phosphorelay. Cell 64545-552. Bustamante, V. H., I. Martinez-Flores, H. C. Vlamakis, and D. R. Zusman. 2004. Analysis of the Frz signal transduction system of Myxococcus xanthus shows the importance of the conserved C-terminal region of the cytoplasmic chemoreceptor FrzCD in sensing signals. Mol. Microbiol. 53:15011513. Caberoy, N. B., R. D. Welch, J. S. Jakobsen, S. C. Slater, and A. G. Garza. 2003. Global mutational analysis of NtrC-like activators in Myxococcus xanthus: identifying activator mutants defective for motility and fruiting body development. ]. Bacteriol. 185:6083-6094. Carrero-LCrida, J., A. Moraleda-Muiioz, R. Garcia-Herniindez, J. PCrez, and J. Muiioz-Dorado. 2005. PhoR1-PhoP1, a third two-component system of the family PhoRP from Myxococcus xanthus: role in development. ]. Bacteriol. 187:4976-4983. Chang, C. H., J. Zhu, and S. C. Winans. 1996. Pleiotropic phenotypes caused by genetic ablation of the receiver module of the Agrobacterium tumefaciens VirA protein. ]. Bacteriol. 178:4710-4716.

10. TWO-COMPONENT SIGNALTRANSDUCTION SYSTEMS Cho, K., and D. R. Zusman. 1999a. Sporulation timing in Myxococcus xanthus is controlled by the espAB locus. Mol. Bacteriol. 34:7 14-725. Cho, K., and D. R. Zusman. 1999b. AsgD, a new two-component regulator required for A-signalling and nutrient sensing during early development of Myxococcus xanthus. Mol. Microbiol. 34:268-281. Cho, K., A. Treuner-Lange, K. A. O’Connor, and D. R. Zusman. 2000. Developmental aggregation of Myxococcus xanthus requires frgA, an frz-related gene. J. Bacteriol. 182: 6614-6621. Cock. P. J., and D. E. Whitworth. 2007. Evolution of gene overlaps: relative reading frame bias in prokaryotic twocomponent system genes. J. Mol. Evol. 64:457-462. Coudart-Cavelli, M. P., 0. Sismeiro, and A. Danchin. 1997. Bifunctional structure of two adenylyl cyclases from the myxobacterium Stigmatella aurantiaca. Biochimie 79:757-767. Ellehauge, E., M. Norregaard-Madsen, and L. SsgaardAndersen. 1998. The FruA signal transduction protein provides a checkpoint for the temporal co-ordination of intercellular signals in Myxococcus xanthus development. Mol. Microbiol. 30:807-817. Fabret, C., V. A. Feher, and J. A. Hoch. 1999. Two-component signal transduction in Bacillus subtilis: how one organism sees its world. J. Bacteriol. 181:1975-1983. Feher, V. A., J. W. Zapf, J. A. Hoch, J. M. Whiteley, L. P. McIntosh, M. Rance, N. J. Skelton., F. W. Dahlquist, and J. Cavanagh. 1997. High-resolution NMR structure and backbone dynamics of the Bacillus subtilis response regulator, SpoOF: implications for phosphorylation and molecular recognition. Biochemistry 36: 10015-1 0025. Galperin, M. Y., A. N. Nikolskaya, and E. V. Koonin. 2001. Novel domains of the prokaryotic two-component signal transduction systems. FEMS Microbiol. Lett. 203:ll-21. Galperin, M. Y. 2005. A census of membrane-bound and intracellular signal transduction proteins in bacteria: bacterial IQ, extroverts and introverts. BMC Microbiol. 5:35. Gronewold, T. M. A., and D. Kaiser. 2001. The act operon controls the level and time of C-signal production for Myxococcus xanthus development. Mol. Microbiol. 40:744-756. Guo, D., Y. Wu, and H. B. Kaplan. 2000. Identification and characterization of genes required for early Myxococcus xanthus developmental gene expression. J. Bacteriol. 182:4564-4571. Hager, E., and R. E. Gill. 2001. Identification and characterization of spdR mutations that bypass protease-dependent regulation of developmental gene expression in Myxococcus xanthus. Mol. Microbiol. 39:765-780. Heermann, R., A. Fohrmann, K. Altendorf, and K. Jung. 2003. The transmembrane domains of the sensor kinase KdpD of Escherichia coli are not essential for sensing K+ limitation. Mol. Micro biol. 47: 83 9-84 8. Higgs, P. I., K. Cho, D. E. Whitworth, L. S. Evans, and D. R. Zusman. 2005. Four unusual two-component signal transduction homologs, RedC to RedF, are necessary for timely development in Myxococcus xanthus. J. Bacteriol. 187~8191-8195. Hoch, J. A. 2000. Two-component and phosphorelay signal transduction. Curr. Opin. Microbiol. 3:165-170.

187

Hoch, J. A., and T. J. Silhavy. 1995. Two-Component Signal Transduction. American Society for Microbiology, Washington, DC. Horiuchi, T., M. Taoka, T. Isobe, T. Komano, and S. Inouye. 2002. Role of fruA and csgA genes in gene expression during development of Myxococcus xanthus. Analysis by two-dimensional gene electrophoresis. J. Biol. Chem. 277~26753-26760. Hsing, W., and T. J. Silhavy. 1997. Function of conserved histidine-243 in phosphatase activity of EnvZ, the sensor for porin osmoregulation in Escherichia coli. J. Bacteriol. 179:3729-373 5. Hutchings, M. I., P. A. Hoskisson, G. Chandra, and M. J. Buttner. 2004. Sensing and responding to diverse extracellular signals? Analysis of sensor kinases and response regulators of Streptomyces coelicolor A3(2). Microbiology 150:2795-2806. Jakobsen, J. S., L. Jelsbak, L. Jelsbak, R. D. Welch, C. Cummings, B. Goldman, E. Stark, S. Slater, and D. Kaiser. 2004. Sigma54 enhancer binding proteins and Myxococcus xanthus fruiting body development. J. Bacteriol. 186:4361-4368. Jelsbak, L., M. Givskov, and D. Kaiser. 2005. Enhancer-binding proteins with a forkhead-associated domain and the d4regulon in Myxococcus xanthus fruiting body development. Proc. Natl. Acad. Sci. USA 102:3010-3015. Jimenez-Pearson., M. A., I. Delany, V. Scarlato, and D. Beier. 2005. Phosphate flow in the chemotactic response system of Helicobacter pylori. Microbiology 151:3299-3311. Keseler, I. M., and D. Kaiser. 1997. Sigma54, a vital protein for Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 94:19791984. Kimura, Y., H. Nakano, H. Terasaka, and K. Takegawa. 2001. Myxococcus xanthus m o k A encodes a histidine kinaseresponse regulator hybrid sensor required for development and osmotic tolerance. J. Bacteriol. 183:1140-1146. Kirby, J. R., and D. R. Zusman. 2003. Chemosensory regulation of developmental gene expression in Myxococcus xanthus. Proc. Natl. Acad. Sci USA 100:2008-2013. Kofoid, E. C., and J. S. Parkinson. 1988. Transmitter and receiver modules in bacterial signaling proteins. Proc. Natl. Acad. Sci. USA 85:4981-4985. Kramer, G., and V. Weiss. 1999. Functional dissection of the transmitter module of the histidine kinase NtrB in Escherichia coli. Proc. Natl. Acad. Sci. USA 96604-609. Lancero, H., N. B. Caberoy, S. Castaneda, Y. Li, A. Lu, D. Dutton, X.-Y. Duan, H. B. Kaplan, W. Shi, and A. G. Garza. 2004. Characterization of a Myxococcus xanthus mutant that is defective for adventurous motility and social motility. Microbiology 150:4085-4093. Lancero,H., S. Castaaneda,N. B. Caberoy,X. Ma,A. G. Garza, and W. Shi. 2005. Analysing protein-protein interactions of the Myxococcus xanthus Dif signalling pathway using the yeast two-hybrid system. Microbiology 151:1535-1541. Lee, B., P. I. Higgs, D. R. Zusman, and K. Cho. 2005. EspC is involved in controlling the timing of development in Myxococcus xanthus. J. Bacteriol. 1875029-5031. Li, Y., and L. Plamann. 1996. Purification and in vitro phosphorylation of Myxococcus xanthus AsgA protein. J. Bacteriol. 178:289-292.

188 Matsubara, M., and T. Mizuno. 2000. The SixA phosphohistidine phosphatase modulates the ArcBA phosphorelay signal transduction in Escherichia coli. FEBS Lett. 470:118124. McBride, M. J., T. Kohler, and D. R. Zusman. 1992. Methylation of FrzCD, a methyl-accepting taxis protein of Myxococcus xanthus, is correlated with factors affecting cell behavior. J. Bacteriol. 174:4246-4257. McBride, M. J., R. A. Weinberg, and D. R. Zusman. 1989. “Frizzy” aggregation genes of the gliding bacterium Myxococcus xanthus show sequence similarities to the chemotaxis genes of enteric bacteria. Proc. Natl. Acad. Sci. USA 86~424-428. McCleary, W. R., and D. R. Zusman. 1990. FrzE of Myxococcus xanthus is homologous to both CheA and CheY of Salmonella typhimurium. Proc. Natl. Acad. Sci. USA 875898-5902. Mignot, T., J. P. Merlie, Jr., and D. R. Zusman. 2005. Regulated pole-to-pole oscillations of a bacterial gliding motility protein. Science 310:855-857. Mizuno, T., T. Kaneko, and S. Tabata. 1996. Compilation of all genes encoding bacterial two-component signal transducers in the genome of the cyanobacterium, Synechocystis sp. strain PCC 6803. D N A Res. 3:407-414. Mizuno, T. 1997. Compilation of all genes encoding twocomponent phosphotransfer signal transducers in the genome of Escherichia coli. D N A Res. 4:161-168. Moraleda-Muiioz, A., J. Carrero-LCrida, J. PCrez, and J. Muiioz-Dorado. 2003. Role of two novel two-component regulatory systems in development and phosphatase expression in Myxococcus xanthus. J. Bacteriol. 185:1376-1383. Ogawa, M., S. Fujitani, X. Mao, S. Inouye, and T. Komano. 1996. FruA, a putative transcription factor essential for the development of Myxococcus xanthus. Mol. Microbiol. 22~757-767. Parkinson, J. S., and E. C. Kofoid. 1992. Communication modules in bacterial signalling proteins. Annu. Rev. Genet. 26:71-112. Perego, M. 1998. Kinase-phosphatase competition regulates Bacillus subtilis development. Trends Microbiol. 6366-370. Pham, V. D., C. W. Shebelut, I. R. Jose, D. A. Hodgson, D. E. Whitworth, and M. Singer. 2006. The response regulator PhoP4 is required for late developmental events in Myxococcus xanthus. Microbiology 152:1609-1620. Pollack, J. S., and M. Singer. 2001. SdeK, a histidine kinase required for Myxococcus xanthus development. J. Bacteriol. 183:35 89-3 596. Pritchard, L., J. A. White, P. R. J. Birch, and I. K. Toth. 2006. GenomeDiagram: a python package for the visualisation of large-scale genomic data. Bioinformatics 22:616-617. Puppe, W., M. Jung, M. Lucassen, and K. Altendorf. 1996. Characterization of truncated forms of the KdpD protein, the sensor kinase of the K+-translocating Kdp system of Escherichia coli. J. Biol. Chem. 271:25027-25034. Rasmussen, A. A., and L. Ssgaard-Andersen. 2003. TodK, a putative histidine protein kinase, regulates timing of fruiting body morphogenesis in Myxococcus xanthus. J. Bacteriol. 185~5452-5464. Rasmussen, A. A., S. L. Porter, J. P. Armitage, and L. SsgaardAndersen. 2005. Coupling of multicellular morphogenesis

REGULATORY MECHANI sM s and cellular differentiation by an unusual hybrid histidine protein kinase in Myxococcus xanthus. Mol. Microbiol. 56:1358-1372. Sasse-Dwight, S., and J. D. Gralla. 1990. Role of eukaryotictype functional domains found in the prokaryotic enhancer receptor factor sigma 54. Cell 62:945-954. Shimkets, L. J. 1999. Intercellular signalling during fruitingbody development of Myxococcus xanthus. Annu. Rev. Microbiol. 53525-549. Skerker, J. M., M. S. Prasol, B. S. Perchuk, E. G. Biondi, and M. T. Laub. 2005. Two-component signal transduction pathways regulating growth and cell cycle progression in a bacterium: a system-level analysis. PLoS Biol. 3:e334. Sonnhammer, E. L., G . von Heijne, and A. Krogh. 1998. A hidden Markov model for predicting transmembrane helices in protein sequences. Proc. Int. Conf. Intell. Syst. Mol. Biol. 6~175-182. Stewart, R. C. 1997. Activating and inhibitory mutations in the regulatory domain of CheB, the methylesterase in bacterial chemotaxis. J. Biol. Chem. 268:1921-1930. Stock, J. B., A. J. Ninfa, and A. M. Stock. 1989. Protein phosphorylation and regulation of adaptive responses in bacteria. Microbiol. Rev. 53:450-490. Sun, H., and W. Shi. 2001. Genetic studies of mrp, a locus essential for cellular aggregation and sporulation of Myxococcus xanthus. J. Bacteriol. 183:4786-4795. Taylor, B. L., and I. B. Zhulin. 1999. PAS domains: internal sensors of oxygen, redox potential, and light. Microbiol. Mol. Biol. Rev. 63:479-506. Trudeau, K. G., M. J. Ward, and D. R. Zusman. 1996. Identification and characterization of FrzZ, a novel response regulator necessary for swarming and fruiting-body formation in Myxococcus xanthus. Mol. Microbiol. 20:645-655. Ueki, T., and S. Inouye. 2005a. Activation of a developmentspecific gene, dofA, by FruA, an essential transcription factor for development of Myxococcus xanthus. J. Bacteriol. 187~8504-8506. Ueki, T., and S. Inouye. 2005b. Identification of a gene involved in polysaccharide export as a transcription target of FruA, an essential factor for Myxococcus xanthus development. J. Biol. Chem. 280:32279-32284. Ueki, T., and S. Inouye. 2003. Identification of an activator protein required for the induction of fruA, a gene essential for fruiting body development in Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 10023782-8787. Ueki, T., and S. Inouye. 2002. Transcriptional activation of a heat-shock gene, lonD, of Myxococcus xanthus by a two component histidine-aspartate phosphorelay system. J. Biol. Chem. 2 7 7 6 170-61 77. Vlamakis, H. C., J. R. Kirby, and D. R. Zusman. 2004. The Che4 pathway of Myxococcus xanthus regulates type IV pilus-mediated motility. Mol. Microbiol. 52:17991811. Wang, L., R. Grau, M. Perego, and J. A. Hoch. 1997. A novel histidine kinase inhibitor regulating development in Bacillus subtilis. Genes Deu. 11:2569-2579. Ward, M. J., K. C. Mok, D. P. Astling, H. Lew, and D. R. Zusman, 1998a. An ABC transporter plays a developmental aggregation role in Myxococcus xanthus. J. Bacteriol. 1805697-5703.

10. TWO-COMPONENT SIGNALTRANSDUCTION SYSTEMS Ward, M. J., H. Lew, A. Treuner-Lange, and D. R. Zusman. 1998b. Regulation of motility behavior in Myxococcus xanthus may require an extracytoplasmic-function sigma factor. J. Bacteriol. 180:5668-5675. Ward, M. J., H. Lew, and D. R. Zusman. 2000. Social motility in Myxococcus xanthus requires FrzS, a protein with an extensive coiled-coil domain. Mol. Microbiol. 37:1357-1371. Wolanin, P. M., E A. Thomason, and J. B. Stock. 2002. Histidine protein kinases: key signal transducers outside the animal kingdom. Genome Biol. 3:REVIEWS3013. Wu, S. S., and D. Kaiser. 1995. Genetic and functional evidence that Type IV pili are required for social gliding motility in Myxococcus xanthus. Mol. Microbiol. 18547-55 8. Wu, S. S., and D. Kaiser. 1997. Regulation of expression of the pilA gene in Myxococcus xanthus. 1. Bacteriol. 179:77487758.

189

Yajko, D. M., and D. R. Zusman. 1978. Changes in cyclic AMP levels during development in Myxococcus xanthus. J. Bacteriol. 133:1540-1542. Yang, R., S. Bartle, R. Otto, A. Stassinopoulos, M. Rogers, L. Plamann, and P. Hartzell. 2004. AglZ is a filament-forming coiled-coil protein required for adventurous gliding motility of Myxococcus xanthus. 1. Bacteriol. 186:61686178. Yang, Z., and Z. Li. 2005. Demonstration of interactions among Myxococcus xanthus Dif chemotaxis-like proteins by the yeast two-hybrid system. Arch. Microbiol. 183:243252. Yang, Z., Y. Geng, D. Xu, H. B. Kaplan, and W. Shi. 1998. A new set of chemotaxis homologues is essential for Myxococcus xanthus social motility. Mol. Microbiol. 30:11231130.

Myxobacteria: Mlrlticelltrlarity and Differentiation Edited by David E. Whitworth 02008 ASM Press, Washington, D.C.

Sumiko Inouye Hirofumi Nariya Jos6 Mufioz-Dorado

Protein Ser/Thr Kinases and Phosphatases in

11

Myxococcus xanthus

Protein Ser/Thr/Tyr kinases and protein phosphatases function as biological switches that turn on and off signal transduction pathways, where they participate by phosphorylation and dephosphorylation. Recently, whole bacterial genome sequencing has revealed that they exist in a wide variety of pathogenic and developmental bacteria ( Av-Gay and Everett, 2000; Wang et al., 2002; Petrickova and Petricek, 2003).The physiological roles of the protein Ser/Thr kinase (PSTK) signaling systems in prokaryotes are beginning to be understood at the molecular level. PSTKs are now known to play important roles in the secondary metabolism of Streptomyces coelicolorA3(2)(Lee et al., 2002), the survival of Mycobacterium tuberculosis within macrophages (Walburger et al., 2004), activation of a myxobacterial enzyme for the consumption of glycogen (Nariya and Inouye, 2002, 2003), and regulation of gene expression during Myxococcus xanthus development (Nariya and Inouye, 2006). The first PSTK in prokaryotes was identified and physiologically characterized because of its role in M. xanthus development (Mufioz-Dorado et al., 1991). During the past 15 years since their discovery, several PSTICs have been cloned and investigated for their roles in the M . xanthus life cycle.

In the course of these investigations, three multikinase-associated proteins, MkapA, MkapB, and MkapC, each of which contains well-known protein-protein interaction domains, have been identified and studied for their roles in the PSTK signaling systems in M. xanthus (Nariya and Inouye, 2005a). Recently, the first functional PSTK cascade in prokaryotes was discovered (Nariya and Inouye, 2005b). This regulatory cascade regulates the expression of two transcription activators, MrpC and FruA, which are essential for multicellular fruiting body formation and sporulation (Nariya and Inouye, 2006). Based on analyzing the recently complete M. xanthus genome sequence database (Goldman et al., 2006), 102 genes that encode putative PSTKs and 34 genes of putative protein phosphatases (PPs) have been identified. In contrast to PSTKs, only one protein Ser/ Thr phosphatase has been characterized (Treuner-Lange et al., 2001). The PSTK signal transduction systems function differently from the two-component His-Asp phosphorelay system (TCST) that plays predominant roles in prokaryotic signal transduction. There are two important differences. In response to a signal, PSTICs use ATP to phosphorylate

Sumiko Inouye, Department of Biochemistry, Robert Wood Johnson Medical School, Piscataway, NJ 08854. Hirofumi Nariya, Department of Biochemistry, Robert Wood Johnson Medical School, Piscataway, NJ 08854. Jose Mufioz-Dorado, Departamento de Microbiologia, Facultad de Ciencias, Universidad de Granada, E-18071 Granada, Spain.

191

REGULATORY MECHANISMS

192 their own Ser/Thr residues for activation of their kinase activities. An activated PSTK phosphorylates its substrate using ATP, in contrast to TCST, where that high-energy phosphogroup is transferred from His residues of kinases to Asp residues of response regulators (Stock et al., 2000). Thus, the signal received by a PSTK system can be greatly amplified and maintained for long time periods, because the phosphogroups of Ser/Thr/Tyr are much more stable than those of His/Asp, and a specific phosphatase is required for the inactivation of PSTKs. In this chapter, we first describe PSTKs and then PPs in M. xanthus.

EUKARYOTIC-LIKE PROTEIN Ser/Thr KINASES Identification and Classification of PSTKs in the Genome M. xanthus contains 102 PSTKs, the largest number so far found in a bacterial genome, based on an analysis of the genome sequence database (http://www.tigr.org) (Goldman et al., 2006), and the M. xanthus kinome is roughly equivalent to that of Saccharomyces cerevisiae, which has 130 PSTKs (Manning et al., 2002). The catalytic domains of PSTKs consist of 11 subdomains, each of which contains highly conserved sequences (Fig. 1) (Hanks et al., 1988). All but three myxobacterial PSTKs are typical eukaryotic protein kinases (ePKs) containing most of the conserved essential residues in subdomains I through XI of the kinase catalytic domain (CD) (Hanks and Hunter, 1995).In addition to ePKs, M . xanthus contains three atypical protein kinases (aPKs) that belong to the RIO and ABCl (RIO-1 and ABC1-1 and 1-2, respectively) families, and have weak similarity to ePKs (reviewed by Laronde-LeBlanc and Wlodawer, 2005). I

II

Ill IV

V

Vla

Although phosphorylation of Tyr has been shown with cell extracts from M. xanthus (Frasch and Dworkin, 1996), as in the case of yeast, no typical protein Tyr kinases have been detected in the Myxococcus kinome, based on sequence analysis. Classification of PSTKs can be made based upon amino acid substitutions in the CD, and five groups of ePKs have been identified due to substitutions in subdomains VIb (HRDxK) and VIII (TxxxxAPE) as shown in Fig. 1. Most ePKs fall into one of two groups, 56 Pkt and 27 Psk, with all three remaining groups making up less than 20% of the ePKs (7 Pab, 5 Psd, and 4 Pdd). Classification of ePK from other bacteria shows that the Psk, Psd, and Pdd groups are almost exclusively found in M . xanthus. Phylogenetic analysis, using the neighborjoining method on the catalytic domain sequences, allows additional classification consistent with the structural classification- As shown in Fig. 2, the Pkt group is further divided into six subgroups, 14 PktA, 10 PktB, 9 PktC, 12 PktD, 4 PktE, and 7 PktF, while the Psk group is divided into two subgroups, 17 PskA and 10 PskB. Among the 11 ePK-branches, PktE and Pdd seem to be recently diverged from PktF and Psd, respectively. The 102 PSTKs identified were named according to the classification described above, disregarding their previous characterization (Table 1). The classification and catalytic domain substitution patterns of some ePKs are particularly interesting. In the Psk group, the conserved K residue in subdomain VIb that forms the catalytic loop facilitating the phosphotransfer reaction by its positive charge, is replaced by a Ser residue. In addition, the Thr residue in subdomain VIII known as the activation site of ePK (Adams et al., 1995), is replaced by a Lys residue. PskA2/Pkn6 and PskASIPknl4 have been demonstrated to be active Vlb

VII

Vlll

IX

x

XI

Group

Pkt (56) Psk(27) PSd (5) Pdd(4) Pab(7)

G G A G E R

G T G G

G V A K E A V V K E G L V/L K E G S L K E G V/LV# #

ATP binding

Catalytic loop

- -

Active Acti- Interact Stabilize Interact center vation with catalytic with loop R(XI) loop E(VIII)

Figure 1 Classification of 99 eukaryotic-like PSTKs. The amino acid residues used for classification are highlighted by gray boxes. Invariant residues in the catalytic domain (Hanks and Hunter, 1995) are in boldface. Symbols: #, at least one residue is substituted or deleted at the position of each Pab; x, the residue is not well conserved.

11. PROTEINSER/THR KINASESAND PHOSPHATASES IN M . XANTHUS

193

Figure 2 Phylogenetic tree of 99 eukaryotic-like PSTKs. The phylogenetic tree was built by the neighbor-joining method (CLUSTALW [http://align.genome.jp/] and NJplot [http://pbil. univ-lyon.fr/software/njplot.html])using manually aligned kinase catalytic domains (subdomain from I to XI [Hanks et al., 19881) and illustrated using Tree View programs (http:// taxonomy.zoology.gla.ac.uk/rod/treeview.html). The Pkt group is shown by black lines in a medium-gray background except for the PktC subgroup, which is shown with dotted lines. The Psk group is shown by gray lines in a light-gray background, and the Pab group is indicated by dotted lines. Psd and Pdd groups are shown with dotted and large dotted lines in light- and dark-gray backgrounds, respectively. PSTKs grouped by open ovals have homologies not only in the catalytic domain but also in the regulatory domain.

ePKs, phosphorylating at Ser and Thr residues, implying that the Ser residue in subdomain VIb may function as the activation site in the Psk group (Zhang et al., 1996; Nariya and Inouye, 2005b). The three rarer ePK groups have striking substitutions. Pab are abnormal ePKs having substitutions at essential residues in subdomains I1 (AxK), VIb (HRD), and VII (DFG) in addition to substitutions in VIb and VIII (Fig. 1). Therefore, Pab PSTKs may be catalytically inactive ePKs that act as modulators, substrates for active kinases, and scaffolds for assembly of signaling complexes as in eukaryotes (reviewed by Manning et al.,

2002). The Psd group has a negative-charged Asp residue at the Lys residue position in subdomain VIII of the Psk group, and Pdd has another Asp residue substitution at the Ser residue in subdomain VIb of the Psd group. In contrast, the Asp residue in the HRD sequence of subdomain VIb is replaced with an Ala residue. However, an Asp residue two residues downstream (HRAxD)may compensate for the role of the Asp residue in HRD.

Analysis of Domain Structure of 102 PSTKs Most M . xanthus PSTKs have CDs at the N-terminal end and regulatory domains (RDs) at the C-terminal end,

REGULATORY MECHANISMS

194 Table 1 List of PSTK genes" Gene

MXAN

Type

pktAl A2/pknl A3 A4/pkn9 AS/p knD 1 A6 A7 A8Ipkn3 A9 A10 All A12 A13 A14

6500 1467 614 755 930 4338 3094 3202 4591 3338 6043 2680 6317 1297

s2 s2 s2 M1 s2 s2 s2 M1 M1 M1 M1 M1 M3

pktBl B2 B3 B4 BS B6 B7 B8IpknD2 B9 B10

642 8 5045 882 2255 6183 871 5176 933 4337 3092

M1 M1 M1 M1 M1 M3 s2 s2 s2 s2

pktCl C2/pkn8 c3 c4

70 1710 273 8 1480 7162 117 6561 6009 7082

M1 M2 M2 M1 M1 M1 M1 M1 M2

4017 2156

M1 M1

cs C6 c7 C8 c9 pktDl/pkn4 D2

s1

Gene

MXAN

Type

906 525 1577 5886 2549 7251 6420 3955 1163 4049

M1 M1 s1 s2

pktEl E2 E3 E4

3182 2176 1233 4842

M1 M1 M1 M1

pktFl F2 F3 F4 FS F6 F7

4373 4482 1896 7370 2840 7269 2399

M1 M1 M1 M1 M1

pskAl A21pkn6 A3/masK A4 AS/p knl4 A6 A7 A8 A9 A10 A1 1/pkn7 A12 A13/pknll A14

2059 2550 1929 2596 5116 5696 4700 5976 552 2586 2910 4557 291 1 6545

M1 M1 M1 M1 s2 M1 M1 s1 M1 M1 M1 M1 M1 M3

D3 04 DS D6 D7/pknS DUpkn12 D9/pkn2 D1 OIpknlO Dll D12

s1 s2 M1 M1 M1 s2

s1 M1

Gene

MXAN

Type

23 18 3099

M1 M1 M1

pskBl B2 B3 B4 BS B6 B7 B8 B9 B10

396 7208 6570 265 3693 5620 6669 6312 960 2980

M3 M1 M1 M1 M1 M1 51 M1 51 51

psdl 2 3 4

3272 7371 1892 4479 4371

51 51 51 51 51

pddl 2 3 4

3183 2177 1234 4841

51 51 51 51

pub1 2 3 4 6 7

5517 40.53 2077 1088 3710 4434 724

M3 M2 M2 51 52 M1 51

riol abcl-1 abcl-2

2315 725 3899

52 51 M3

A1 5lpknl3 A16 A17

S

S

"Abbreviations: M1, type I receptor type kinase; M2, type I1 receptor type kinase; M3, membrane embedded kinase; S1, small cytoplasmic kinase; S2, large cytoplasmic kinase.

like PSTKs in other bacteria, according to CDD-BLAST searches at http://www.ncbi.nlm.nih.gov/Structure/cdd/ cddshtml. Prediction of hydrophobic sequences using the TMHMM program (http://www.cbs.dtu.dk/services/ TMHMM-2.0/) indicates diverse PSTK subcellular localization (Table 1).The largest group consists of 55 type I receptor kinases with an exposed RD in the periplasmic space (Ml). Two moderate-sized groups are made up of cytoplasmic kinases, 19 small PSTKs with fewer

than 150 residues in the C-terminal domain (Sl) and 17 PSTKs with a longer RD at the C-terminal end (S2).The two smallest groups are of five type I1 receptor kinases that have two hydrophobic sequences resulting in exposure of both CD and RD in the cytoplasmic space (M2) and a group of six membrane-anchored kinases (M3). Despite the diverse localization, over one-half of the PSTKs are receptor-type kinases. The frequency of receptor-type kinases suggests that these PSTKs are involved in

11. PROTEINSER/THRKINASESAND PHOSPHATASES IN 211. XANTHUS signaling systems required to regulate the complex M. xanthus life cycles together with the sensor histidine kinases in TCST. The majority of RDs have no similarity to other proteiddomains based upon the CDD-BLAST search. However, some RDs have recognizable functional domains. PktD9/Pkn2 has a Class I11 adenylyl cyclase (AC) domain (PF00211) and PktD12 consists of a Class I11 AC domain and a typical receiver domain of TCST. These domains suggest that M. xanthus may have unique regulatory modes for signal transduction by cyclic AMP (cAMP)/cGMP via both PSTK and TCST systems. Six PSTKs, PktBWPktD2, PktB9, PktB10, PktD6, PktD8/ Pknl2, and Pab2, are predicted to contain an ATPase domain with P-loop (COG3899). Pab2 contains a histidine kinase domain with a cNMP binding domain. Thirteen homologues of Pab2 are found in the cyanobacterium Nostoc PCC7120, a filamentous heterocyst former, but not in the unicellular cyanobacterium Synechocystis PCC6803 (Wang et al., 2002). Three PSTKs, PktD3, PktD4, and Pab3, contain a conserved domain with unknown function (DUF323)in their RDs. The RDs of PktB1, B2, B8, B9, and B10, and PskB5 and B8 contain a TPR (tetratricopeptide repeat), a repetitive domain that is generally responsible for protein-protein interactions and assembly of multiprotein complexes. The RDs of all PSTKs in the PktC subgroup except PktCl contain a kinesin light chain-like domain similar to TPR. The RD of PktA12 contains the C-terminal domain of TonB that is known to function as a dimerization domain and to contact with TonB receptor (Chang et al., 2001), implying that PktA12 may form a homodimer or a heterodimer with TonB to regulate iron transportation. The diverse properties of PSTK RDs indicate that they have a broad importance for M. xanthus physiological activity.

195

pktE-pdd and pktF-psd Duplications Nested gene duplication, divergence, and rearrangements have occurred with p k t E , pktF, pdd, and psd PSTKs. Analysis of the PSTK phylogenetic tree indicates that the PktE and PktF groups recently diverged, as did the Pdd and Psd groups (Fig. 2). Each pair of groups retains weak similarity not only in their CDs but also in their RDs. We first focus on p k t E and pdd. Based on genome linkage analysis, every pdd is located downstream of a pktE: p k t E l - p d d l , pktE2-pdd2, pktE3-pdd3, and pktE4-pdd4 (Fig. 3a). Upstream of each pktE-pdd, there

a PkE-Pdd 1

ll"_".." ." . I.*_ . __ _"""._._ ". .I"

2 3 4

....."_". _.. . -.

orf309 -like

pktE

b

Multiple-Gene Duplications of PSTK Gene duplication is a potent mode of evolutionary adaptation (Ohno, 1999).Numerous examples of gene duplication followed by evolutionary modification of one or both of the duplicates are known throughout eubacteria, archaea, and eukaryotes (reviewed by Pao et al., 1998; Gilsdorf et al., 2004; and Irish and Litt, 2005). However, the PSTK family in M. xanthus is a unique example of diversity and size for a recognizable functional family of duplicated genes within a single genome. Determination of the diverse PSTK functions in M. xanthus is providing fundamental information on cell function and will be important in understanding how M. xanthus gained a large genome, with many genes involved in PSTK signaling systems together with TCST signaling systems. Below we describe the best-investigated duplication events.

pktF4 psdl Figure 3 Multiple gene duplication of pktE-pdd and psd-pktF. (a) Highly conserved regions in pktE-pdd duplications are shown by bars. A gray bar in pdd2 indicates a local duplication. Black, dark-gray, and light-gray arrows are p k t E , pdd, and an orf309-like ORF (having homology with orf309 in psd4-pktF2 in panel b), respectively. (b) Thick black, dark-gray, and light-gray arrows represent genes for PktF, Psd, and ORF309 homologues, respectively. Thin black and light-gray arrows indicate the genes for a putative phagerelated transcriptional regulator with H T H 3 type DNA binding domain (PF01381) and an ORF with unknown function, respectively. Putative hydrophobic sequences are shown by open ovals.

196 is another gene with similarity to orf309 (Akiyama and Komano, 2004) having weak similarity to the C-terminal half of bacterial small Ras-like GTP-binding proteins based on BLAST searches. The four trios of orf309-like open reading frame (ORF), PktE, and Pdd have over 95% amino acid and DNA identity, and each appears to form an operon. Based on the sequence analysis of the pktE-pdd regions, the region containing pktE3pdd3 appears to retain the most ancestral structure of the pktE-pdd duplications. Although we do not know the mechanism(s) by which these duplications occurred, the upstream region of pktE2-pdd2 contains a transposase gene while integrase and transposase genes are upstream of pktE4-pdd4. Similar duplication was also observed in the pktF and psd groups, although the gene organization of pktF-psd duplicates is more complex than the pktE-pdd duplication (Fig. 3b). Four of seven pktFs exist as a pair with four psds, while three pktFs are found alone. Interestingly, whereas orf309 identified previously by Akiyama and Komano (2004) was found downstream of psd4pktF2, its homologue, orf309-like ORF, was also found downstream of pktF7 and two pktF-psd pairs, psd5pktFl and psd3-pktF3, and upstream of pktF5. In addition to the orf309-like ORF, a putative transcriptional regulator with helix-turn-helix type 3 of DNA-binding domain (PF01381) is present in these regions with local duplications. In contrast to the pktE-pdd pairs, the RDs of PktF4 and PktF6 have apparently been lost by deletion and the other PktF RDs share less than 30% similarity at the amino acid level. The distribution of ORF309 homol o p e indicates that the pktE-pdd duplication family is likely derived from one of the pktF-psd duplicates. As in the case of pktE-pdd duplications, an integrase gene upstream of psd4-pktF2, and transposase and resolvase genes downstream of psd3-pktF3 were observed. Structurally, PktE and PktF are type I receptor PSTKs ( M l ) while Pdd and Psd are cytoplasmic PSTKs with short RDs ( S l ) . These PSTK pairs likely form kinase cascades, in a manner similar to that of PktC2/ Pkn8 to PskASIPknl4 (Nariya and Inouye, 2005b). In addition to the nine ORF309 homologues associated with PktE-Pdd and PktF-Psd, no other ORF309 homologue has been detected in M. xanthus or other representative bacterial genomes. ORF309 may play a specific role in M. xanthus physiology.

pktC Duplication Similar duplication events are also evident in the PktC group, which consists of 9 PSTKs (Fig. 2). PSTKs belonging to the PktC group have 8 to 10 tandem repeats of KLC-type TPR domain in their RDs downstream of

REGULATORY MECHANISMS the hydrophobic sequence, except for PktC1, in which the TPR domain seems to be truncated. Gene organization analysis shows that each member of the pktC gene family forms a gene cluster with a gene encoding an extracytoplasmic-function-typesigma factor, RpoE-like homologue (Erickson and Gross, 1989). Homologues of rpoE are located downstream of pktC2 and C3, upstream of pktC5, C8, and C9, and two genes upstream of C1. The rpoE homologue associated with pktC5 seems to be a pseudogene having a frame shift mutation by addition of a guanine residue in the middle, and there is no rpoE homologue associated with pktC4, C6, and C7. Thus, PktC may form a signaling pathway with an extracytoplasmic function sigma factor to regulate gene expression in specific circumstances. As in the pktE-pdd and pktF-psd duplications, transposase and integrase genes are also found in the upstream region of pktC2. One of the nine PktC has been shown to form a PSTK-PSTK cascade with a cytoplasmic PSTK. PktC2/ Pkn8, in conjunction with PskAS/Pknl4, regulates MrpC function, an essential transcriptional regulator for fruiting body formation and sporulation (Nariya and Inouye, 2006; see following section). The other PktCs may also form specific PSTK-PSTK cascades and/or modulate the PktC2-PskA5 cascade by forming heterodimer complexes with PktC2 via their TPR domains.

Gene Organization Adjacent to pstk Genes The role of PSTK genes and their importance can be investigated by assessing adjacent genes. We have observed that genes with diverse functions are present immediately adjacent to or near pstk genes, forming gene clusters. p s t k s Adjacent to p s t k Genes We previously found that pskA13/pknll is only 20 bp downstream of pskAl llpkn7 in the same orientation, forming an operon (Inouye et al., 2000). Similarly, pub7 and pktB9 are 76 and 78 bp upstream of abcl-1 and pktA6, respectively, in the same orientation and hence may form operons. In contrast, pktD7/pkn.5 and pskA2/pkn6 are found with opposite orientations and share a 128-bp common promoter region (Zhang et al., 1996). Several pstk pairs are separated by one to three genes; pktBl0 and pktB8 are found one and two genes upstream of pktA7 and pktA5, respectively. Another example is pab2, which has a histidine kinase domain and is three genes away from pktDl2, which contains the receiver domain of a TCST. In addition, we have already described sets of PktF-Psd and PktE-Pdd PSTK pairs arising from multiple-gene duplications.

11. PROTEINSER/THR KINASESAND PHOSPHATASESIN M . XANTHUS

pstks Adjacent to Phosphatase ( p p ) Genes Four p p genes that encode protein phosphatases in the phosphoprotein phosphatase (PPP) superfamily (see “Protein Phosphatases” below) are located close to pstk genes. One of them, MXAN7163, is located immediately downstream of pktC.5, forming an operon. Two genes, MXAN6S43 and MXANOSSS, are separated from pskB5 and pskA9 by one and two genes, respectively, and may form operons. Two genes located between MXANO.555 and pskA9 encode ATPase and ABC permease activities of a putative ABC transporter, activities that are possibly regulated by pskA9 and MXAN0.555 by reversible phosphorylation and dephosphorylation. Another gene, MXAN0267, is located in the opposite orientation to pskB4, separated by one unknown gene. In addition, pskA12 is located in the same orientation and probably in the same operon as a SpoIIE-like protein phosphatase (see below).

pstks Adjacent to oS4-DependentEBPs with FHA Domains As described in chapter 9, 12 enhancer-binding proteins (EBPs) have been predicted to contain the forkheadassociated (FHA) domain at their N terminus, and 6 of them are located adjacent to or near PSTKs (Jelsbak et al., 2005). Based on a genome analysis, a new gene ( M X A N 0 9 0 7 ) that encodes FHA-EBP was identified (Fig. 4), and this gene’s termination codon overlaps with the initiation codon of pktD3. The regulatory mechanism of a PSTK-EBP mediated by its FHA domain is described in chapter 9.

197

pstks Adjacent to Genes for Carbohydrate Metabolism Several genes required for carbohydrate metabolism are adjacent to or near the pstks. The pfk gene ( M X A N 4 0 16 ) that encodes 6-phosphofructokinase (PFK), a key enzyme for glycolysis, is located 18 bp upstream of pktDllpklz4, forming an operon. PFK activity was activated upon phosphorylation by PktD 1/Pkn4, and effective sporulation of M . xanthus requires glycogen consumption by activated PFK during development (see below) (Nariya and Inouye, 2002, 2003, 2 0 0 5 ~ )Simi. larly, glgA (glycogen synthase; M X A N 1 2 9 6 ) is located 35 bp upstream of pktA14, while the poly-3-hydroxybutyrate (PHB) depolymerase gene ( M X A N 6 3 1 3 ) is terminated by overlapping 4 bp with the initiation codon of pskB8. Another PHB depolymerase gene ( M X A N 0 0 1 6 ) is also located upstream of pskA16 in the same orientation, separated by one gene. PHB is another reservoir carbohydrate in prokaryotes (reviewed by Steinbiichel et al., 1992). Moreover, a gene for a-lY4-glucanase( M X A N 3694) for glycogen consumption is located 91 bp downstream of pskB.5. pktA1 is located between the mannose-1-phosphate guanylyltransferase (MXAN6SO1) and phosphomannomutase ( M X A N 6 4 9 9 ) genes for lipopolysaccharide synthesis in the same orientation, probably forming an operon. Since M . xanthus cannot utilize carbohydrates as a carbon source, sugar metabolites and their reservoir carbohydrates seem to be synthesized via gluconeogenesis using glucogenic amino acids and lipids (Bretscher and Kaiser, 1978). Thus, M. xanthus appears to uniquely develop the regulation of carbohydrate metabolism through PSTK signaling pathways.

Roles of PSTKs in the M. xunthus Life Cycle pktA6 Mx1288 pktBl0

pktA7

1 kP

-

Mx4901

MAXN0907 pktD3

Figure 4 FHA-EBP genes adjacent to or near pstks. Thick black and dark-gray arrows represent pstk and the gene for FHA-EBP, respectively. Thin light-gray arrows represent the genes for PFK, a-l,4-glucanase, and unknown functions.

After PktA2/Pknl was discovered and characterized for its role in the M. xanthus life cycle (Mufioz-Dorado et al., 1991), eight additional PSTKs were studied among the 13 PSTKs described by Inouye et al. (2000), PskA3/ MasK (Thomasson et al., 2002), and PktASIPknDl and PktBWPknD2 (Stein et al., 2006). Eight eukaryote Ser/ Thr/Tyr protein kinase inhibitors whose specificities are well documented have been tested for growth, motility, and developmental effects (Jain and Inouye, 1998).None of them had any effect on vegetative growth, while several inhibited development and sporulation to various degrees. The protein kinase C inhibitors, staurosporine, K-252c, and chelerythrin, and a Tyr kinase inhibitor, tyrphostin B52, are found to inhibit fruiting body development. Furthermore, sporulation is completely inhibited by K-252c, chelerythrin, and tyrphostin B52, but not by staurosporine, suggesting that staurosporine specifically

REGULATORY MECHANISMS

198 inhibits one branch required for the formation of the fruiting bodies but does not affect other branches leading to sporulation.

PktA2Pknl The first PSTK identified not only in M. xanthus but in all prokaryotes was pktA2/pknl (Mufioz-Dorado et al., 1991). The expression of pktA2/pkn1 is developmentally regulated and starts immediately prior to spore formation. Strains in which pktA2/pknl have been deleted show premature fruiting body formation and a 35% spore yield relative to the parent strain. PktA2/Pknl purified from E. coli is a Mg2+-dependentkinase and is autophosphorylated at both Ser and Thr.

PktD9mkn2 PktD9/Pkn2 is a transmembrane receptor-type PSTK ( M l ) with a cytoplasmic CD and a 207-residue C-terminal domain outside the cytoplasmic membrane. Disruption of pktD9/pkn2 has no effect on vegetative growth but causes faster development and reduces spore yield to 50 to 70% of that of the parent strain (Udo et al., 1995). In contrast to the pktD9/pkn2 disruption strain, a pktD9/pkn2 overexpression strain forms fruiting bodies more slowly than the parent strain but with a similar reduction in spore yield (Udo et al., 1996). PktD9/Pkn2 has a 28-fold-higher activity in the presence of Mn2+than Mg2+,and its activity was inhibited by H-7, a eukaryotic PSTK inhibitor, but not by genistein (Udo et al., 1997). Interestingly, when PktD9/Pkn2 was produced in E. coli, p-lactamase was found to serve as an effective substrate for PktD9/Pkn2 (Udo et al., 1995). Furthermore, a novel method to identify substrates using the toxic effect of PktD9/Pkn2 expression in E. coli was developed and found that E. coli histonelike proteins HUa and HUP act as suppressors that are phosphorylated at Thr-59 (Udo et al., 2000). As mentioned above, PktD9/Pkn2 contains a Class I11 AC domain (PF00211) in the cytoplasmic region followed by the CD. Class I11 ACs are commonly found in eukaryotes and some developmental and pathogenic bacteria. Class I11 ACs are evolutionally different from Class I ACs, exclusively found in enterobacteria (Linder and Schultz, 2003). AC synthesizes the most prevalent signaling molecule, CAMP, which plays important roles in regulating various cellular processes under the influence of diverse environmental stimuli in both prokaryotes and eukaryotes. Although the regulation and cellular functions of mammalian ACs have been well studied (reviewed by Linder and Schultz, 2003), those of prokaryotic Class I11 ACs are still unclear, especially their cellular functions. In M. xanthus, the intercellular level

of CAMPis known to rapidly increase during early development (Yajko and Zusman, 1978; Kimura et al., 2002). Biochemical studies revealed that AC activity of PktD9/ Pkn2 is greatly activated upon autophosphorylation at Thr residues in the Ala-Ser-Gly-Thr-rich region between the CD and AC domains. The activation of AC by phosphorylation occurs during development and is required for normal fruiting body formation with effective sporulation (Nariya and Inouye, unpublished results). M. xanthus contains another PSTK with a Class I11 AC domain, PktD12, which has a unique domain architecture found only in M . xanthus. PktD12 is located in the cytoplasm. It has a CD at the N terminus followed by the receiver domain of a TCST and then by a Class I11 AC domain. The receiver domain of PktD12 was found to associate with the histidine kinase domain of Pab2 by a genomic yeast two-hybrid screen (Nariya and Inouye, unpublished results). This suggests the intriguing possibility that the regulation of AC occurs by TCST and PSTK. If the receptor-type histidine kinase in Pab2 receives an environmental signal, it may activate AC by phosphorylating the receiver domain of PktD12, and Pab2 activity may also be regulated positively or negatively by the Ser/Thr kinase domain of PktD12.

PktD7/PknS and PskA2/Pkn6 The two pstk genes, pktD7/pknS and pskA2/pkn6, are oriented in opposite directions but share a 128-bp promoter region between their transcription initiation sites (Zhang et al., 1996). PktD7mkn5 consists of 380 amino acid residues with the insertion of 113 amino acid residues between subdomain V and VIa. It is a soluble PSTK in the cytoplasm, while PskA2/Pkn6 has 710 amino acid residues and is a transmembrane receptor-type PSTK. Purified from Escherichia coli, PktD7/Pkn5 is autophosphorylated at Ser while PskA2/Pkn6 is autophosphorylated at both Ser and Thr. Both genes are expressed constitutively throughout the M. xanthus life cycle, with slight increases early in development. PktD7Pkn5 and PskA2Pkn6 have reciprocal roles in M. xanthus growth and development, because a pktD7/pkn5 deletion strain forms fruiting bodies much faster than the parent strain, while a pskA2/kn6 deletion strain develops slower than the parent strain. The pktD7/pknS deletion strain is able to form fruiting bodies on semirich media, suggesting that PktD7/PknS negatively regulates M. xanthus development.

PktA4/Pkn9 PktA4/Pkn9 is a transmembrane receptor-type PSTK exposing a CD in the cytoplasmic space and a C-terminal RD in the periplasmic space. It is constitutively expressed during vegetative growth and upregulated during the

11. PROTEINSER/THRKINASES AND PHOSPHATASES IN M. XANTHUS aggregation stage of early development. The deletion of pktA4/pkn9 causes severely reduced development progression and spore formation. Two-dimensional gel analysis reveals that the deletion of pktA4/pkn9 prevents the expression of four unidentified membrane proteins, KREP9-1-4 (Hanlon et al., 1997).

PskAYMasK Isolation of pskA3ImasK was as a suppressor gene for an mglA mutant (Thomasson et al., 2002). The mglA gene encodes a 22-kDa GTPase critical for single-cell (A)gliding, type IV pilus-mediated (S) gliding, and M. xanthus development. S-motility and starvation-induced development can be restored by mas81 5, an allele-specific extragenic suppressor of mglA8, but it is unable to restore A-motility. pskA3/masK encodes a protein Ser/Thr/Tyr kinase and is in an operon immediately upstream of the mglBA operon. The interaction between PskA3lMasK and MglA was observed using a genomic yeast twohybrid screen. Tyr phosphorylation of PskA3lMasK in E. coli extracts has been shown by Western blot analysis using the anti-PskA3/MasK antibody and phosphoTyr antibody. However, PskA3lMasK purified from E. coli has autokinase activity to its own Ser/Thr residue(s), but not to its Tyr residue (Nariya and Inouye, unpublished results).

PktDl/Pkn4-PFK Cascade PktDUPkn4 is a PFK kinase, and the pfk gene forms an operon with pktDZlpkn4 (Nariya and Inouye, 2002). PFK is a key enzyme for glycogen metabolism in prokaryotes and eukaryotes, and its activity in eukaryotes is regulated by phosphorylation (Kemp et al., 1981; Foe and Kemp, 1982). By mutational analysis, PktDUPkn4 purified from E. coli is found to phosphorylate M. xanthus PFK at Thr-226, located in the putative allosteric effector site. Phosphorylation of PFK by PktDllPkn4 enhanced its activity 2.7-fold, and phosphorylation of PFK by PktDl/Pkn4 is almost completely inhibited by phosphoenolpyruvate, an allosteric inhibitor for PFK. The association of PFK with the regulatory domain of Pkn4 was demonstrated by immunoprecipitation using anti-Pkn4 immunoglobulin G, and this association is inhibited by phosphoenolpyruvate. M . xanthus accumulates glycogen during stationary phase and early in development (Nariya and Inouye, 2003). A pfk-pktDllpkn4 deletion strain accumulates glycogen at a higher level than the parent strain but is unable to consume glycogen during developmental progression and exhibits a poor spore yield. Based on a genetic complementation analysis of the pfk-pktD1l pkn4 deletion strain with the pfk and pktDllpkn4

199

genes, glycogen consumption and a high spore yield require not only the pfk gene but also the pktDllpkn4 gene. Furthermore, phosphorylation is critical for glycogen consumption because the pfk (Thr226Ala) gene mutated at the phosphorylation site did not complement a pfk mutant, suggesting that glycogen metabolism in M . xanthus is regulated in a manner similar to that in eukaryotes, requiring a PSTK (Nariya and Inouye, 2003).

p ktAS/pk n D 1 and p ktB 8/pknD2 Two pstk genes, pktASlpknD1 and pktB81pknD2, were identified during the course of sequencing the espAB region (Cho and Zusman, 1999). pktAUpknD2 is located upstream of the espAB operon and is oriented in the opposite direction. pktBWpknD2 is located immediately downstream of the espAB operon and is also oriented in the opposite direction. Based on hydrophobic sequence analysis, both PSTKs appear to be cytoplasmic kinases belonging to the S2 group (Table 1).The espAB operon regulates the timing of aggregation and sporulation. espA encodes a sensor histidine kinase with an FHA domain at the N terminus, and espB encodes a putative peptide transporter with 12 or 13 hydrophobic sequences (Cho and Zusman, 1999). The espA deletion mutant aggregates and sporulates much earlier than the wild type (DZ2)and forms a large number of small-sized fruiting bodies containing mature spores. Interestingly, numerous individual spores are found outside the fruiting bodies. espB deletion causes delays in aggregation and translucent mound formation, indicating defective sporulation. The espA-espB deletion mutant reveals that EspA is epistatic to EspB. The pktASlpknD2- or pktB8/ pknD2-deficient mutant forms translucent mounds and produces low spore yields, similar to the espB mutants (Stein et al, 2006). Double-mutant analysis reveals that espA is epistatic to pktA.5 and pktB8 to aggregation and fruiting body morphology, while pktA.5 and pktB8 are epistatic to sporulation efficiency. Expression of both pktASlpknD2 and pktB8lpknD2 is observed in vegetative growth and development. PktASIPknDl is phosphorylated at the Thr residue(s) in vitro, but PktB8lPknD2 is not autophosphorylated. However, PktB8/PknD2(Aspl62Asn), a phosphorylation-defective mutant, reveals a phenotype similar to that of the pktB8l pknD2-deficient mutant, indicating that PktB8lPknD2 functions as an active PSTK in vivo. The interaction of Tap-tagged PktASIPknDl or PktB8lPknD2 with EspA has been demonstrated by immunoprecipitation during development, suggesting that both PSTKs regulate fruiting body morphology and sporulation by interacting with EspA and EspB.

REGULATORY MECHANISMS

200

Mkaps and PktDl/Pkn4-PFK Cascade Three new factors, multikinase associated protein A (MkapA), MkapB, and MkapC, each of which contains well-known protein-protein interaction domains, and their associated PSTKs have been identified using a genomic yeast two-hybrid screen (Nariya and Inouye, 2005a). MkapA contains a Zn-finger-like domain that consists of CX,CX,HX,H with nine residues in a finger loop. Zn-fingers function as molecular recognition elements and are extremely common protein domains; perhaps 1% of all mammalian genes encode Zn-finger proteins (Mackay and Crossley, 1998). Interestingly, no MkapA homologues are found in the M. xanthus genome database or in any prokaryote genome sequence database so far determined. MkapB contains eight tandem repeats of a TPR domain. MkapB inhibits PFK activation in a phosphorylation-dependent manner by interfering with the association between MkapB and PFK. MkapC contains three repeats of a fibronectin type 3-like domain that is commonly found in mammalian proteins. A schematic diagram of the complex PSTK networks including the common modulating factors, MkapA, MkapB, and MkapC, is shown in Fig. 5. MkapA associates not only with PktDl/Pkn4 but also with other membrane-associated PSTKs, PktA2Rkn1, PktD9/Pkn2, PktC2/PltnSY and PktA4/Pkn9. MkapA may function as a regulator for the autokinase activity of PSTKs or as an adapter molecule for recruiting downstream factors in their signaling pathways. MkapB also associates with PktC2/PknS containing a TPR domain and PktA4/Pkn9

without a TPR domain; therefore, MkapB also functions as an adapter molecule for recruiting factors in the PktC2/Pkn8 and PktA4IPkn9 signaling pathway. MkapC associates with PktC2/Pkn8 in addition to PktDUPkn4. K9apl is an FHA protein with unknown function. Thus, environmental signals transmitted by membrane receptor-type PSTKs form complex networks with common modulating factors and regulate physiological functions of M. xanthus.

PktC2/PknS-PskAS/Pkn14 Cascade and Regulation of Essential Transcription Activators MrpC and FruA In addition to the Mkaps, PktC2IPkn8 associates with PskA5/Pknl4, a cytoplasmic PSTK. PktC2/Pkn8 phosphorylates PskA5IPknl4, suggesting that PktC2IPkn8 forms a kinase cascade with PskASIPknl4. In addition, PskASIPknl4 associates with MrpC (Nariya and Inouye, 2005a), an essential transcription factor for fruA expression (Ueki and Inouye, 2003). PskASIPknl4 is an MrpC kinase, with Thr-21 and/or Thr-22 as the likely site(s) of MrpC phosphorylation (Nariya and Inouye, 2005b). Both MrpC and FruA are transcriptional activators that are essential for multicellular fruiting body formation and sporulation (see chapter 9). Importantly, MrpC binding activity is greatly reduced upon its phosphorylation by PskASIPknl4. MrpC binds to at least eight sites in the upstream region of its promoter region (Nariya and Inouye, 2006). Based on analysis of MrpC binding sites in the mrpC and fruA promoter regions,

PSTK cascade

Figure 5 A signaling network of PSTKs sharing Mkaps in M. xanthus. Bars with arrowheads at both ends indicate interactions identified by the yeast two-hybrid screens. The T-bar is the inhibition of PFK phosphorylation by MkapB. Gray arrows are characterized phosphorylation pathways.

11. PROTEINSER/THR KINASESAND PHOSPHATASES IN M. XANTHUS there are two types of MrpC-specific binding sequences, A/GTTTC/GAA/G and GTGTCNNNNNNNGACAC. We have proposed a model for regulation of mrpC and fruA expression by a eukaryotic-like protein Ser/ Thr kinase cascade and prokaryotic two-component His-Asp phosphorelay system (Fig. 6 ) (Nariya and Inouye, 2006). As described by Sun and Shi (2001), mrpC is located downstream of mrpAB, encoding MrpA (histidine kinase) and MrpB (response regulator belonging to the NtrC family). A typical -24/-12 box for crS4is present in the upstream region of the mrpC promoter. During vegetative growth, the mrpC gene is likely transcribed at low levels by RNA polymerase with oS4 in the presence of basal levels of phosphorylated MrpB. The MrpC that is produced is phosphorylated by PskASIPknl4, activated by the PktC2IPkn8. The PktC2/ Pkn8-PskASIPknl4 kinase cascade negatively regulates mrpC expression by phosphorylation to prevent the

201

untimely initiation of development during vegetative growth. In early development, mrpAB expression is activated by environmental signals and produces MrpA and MrpB. MrpB, essential for mrpC expression, is activated by phosphorylation. Activated MrpB induces mrpC expression, and MrpC autoregulates its own expression. Since PskASIPkn14 expression is observed mainly during vegetative growth, the newly synthesized MrpC is likely not phosphorylated but instead processed to MrpC2, which has a higher affinity for the mrpC and fruA promoter regions. MrpC2 lacks the N-terminal 25 residues of MrpC and exhibits four- and eightfold-greater binding promoter regions, respecactivity to the mrpC and f r ~ A tively. Importantly, PskASIPknl4 is not able to phosphorylate MrpC2. Since MrpC2 was not detected in lonD mutant cells, LonD may play a role in the proteolytic processing of

I Vegetative growth I

I+

Developmental gene expression

I

I

Development

I

Fruiting body formation Sporulation

Figure 6 Regulation of mrpC and fruA expression in M . xanthus. PktC2JPkn8 and PskASJ Pknl4, forming a kinase cascade, and MrpA, a TCST histidine kinase, are highlighted in dark and light gray, respectively. See the text for details.

REGULATORY MECHANISMS

202 MrpC. LonD is the only protease known at present to be essential for fruiting body formation (Gill et al., 1993; Tojo et al., 1993). The accumulation of MrpC2 activates the expression of fruA, and FruA activates downstream genes involved in the regulation of aggregation, fruiting body formation, and sporulation. Therefore, fruiting body development and sporulation of 111. xanthus are achieved not only by the PSTK signaling system and a TCST system but also by PSTK networks and Mkaps.

Mutational Analysis of 102 PSTKs for Motility and Development In addition to the PSTKs described above, four PSTKs, PktA8mkn3, PskAl 1lPkn7, PskAl3lPkn11, and PskAlSI Pknl3, have been also characterized (S. Inouye, W. Zhang, M. Y. Hsu, R. Jain, and E. Farez-Vidal, unpublished results). To investigate the roles of the remaining 89 PSTKs, their plasmid-insertion mutants were constructed using plasmids carrying DNA fragments of the kinase catalytic core domain (subdomains I11 to IX) in Fig. 1. The disruption mutants were made in DZF1, which is proficient at fruiting body formation but has a partially impaired or leaky S-motility (+/- in Table 2). All 89 disruption mutants could be isolated, suggesting that all 102 PSTKs are not required during vegetative growth at 30°C in Casitone-yeast extract (CYE) medium. All mutants were examined for A- and S-motilities, fruiting body formation, and sporulation and were found to have A-motility, while three mutants, pktF4, pskA9, and pskB7, were defective in S-motility (-), forming a smooth edge at the end of colonies on 0.3% CYE agar plates (Table 2). Whereas the parent strain completes fruiting body formation in 2 days, the majority of the mutants showed defective phenotypes to various degrees including delayed fruiting body formation. pskA12 and abcl-2 mutants are stopped at the aggregation stage, and pktASlpknD1, pktB9, pktD6, pktE2, and pdd3 mutants are able to aggregate but cannot form mature fruiting bodies. pktA.5, pktD6, pskAl2, pdd3, and abcl-2 mutants are also defective in sporulation. The fruiting body morphology and sporulation of the pktA.5 mutant differ from those of the pktA.5-deficient mutant in DZ2 (A+S+)as described by Stein et al. (2006). These differences appear to be caused by the partially impaired Smotility of DZFl (A+S+’-).Six mutants, (pktB4, pktB6, pktD8/pkn12, pktF4, pktF6, and pab4) are able to form fruiting bodies with abnormal patterns. While the majority of mutants delay fruiting body formation, four mutants, pktA2/pknl, pktC2lPkn8, pktD7lpkn.5, and pskASIPknl4, accelerate fruiting body formation (Mufioz-Dorado et al., 1991; Zhang et al., 1996; Nariya and Inouye, 2005b). Since the pktD7lpkn.5 mutant is

Table 2 Developmental phenotypes and motility of pstk mutants Strain DZFl PktAS PktD6 PskA12 Pdd3 ABC1-2 PktA7 PktAl PktBl PktB4 PktB6 PktB8 PktB9 PktC7 PktC8 PktD8 PktE2 PktF4 PktF6 PskA9 PskA15 PskB7 PskB8 Pab4

FB formation (days)”

Sporulation (%)

2 def def def def def 5 4 4 def def 7 def 4 2 def def def def 7 7 4 7 def

100 0 0 0 0 0 10 7 3 2 2 1 0.1 2 6 1 0.3 7 10 80 40 85 40 2

S-motility +/-‘2

+/f/-

+/+/+/+/+/ff-

+/-

+/+/+/+/+/+/+/-

+/-

f/-

+/+/-

“Fruiting body formation. bSee text. Spore numbers were counted a t 7 days of development. ‘Deficient in FB formation.

able to form fruiting bodies on semirich medium, PktD71 PknS has been proposed to negatively regulate fruiting body development (Zhang et al., 1996).

PROTEIN PHOSPHATASES The finding of a large family of PSTKs in 111. xanthus raised the question of the existence of PPs. PPs are the counterpart of PSTKs, since they are required to dephosphorylate the substrates phosphorylated by the kinases. Identification of PPs in myxobacteria has been difficult; the first gene encoding a phosphatase was not cloned until 2001 (Treuner-Lange et al., 2001), 1 0 years after the report of the cloning of the first PSTK in M. xanthus. However, five patterns of phosphatase activity on p-nitrophenyl phosphate (PNPP) were detected in 1990 in this bacterium: two during vegetative growth and three during development. In addition to their expression profiles, the five activities could be differentiated by their dependency on magnesium, optimum pH, and inhibition by dithiothreitol (Weinberg and Zusman, 1990). None of these activities have been correlated with a

11. PROTEINSER/THR KINASESAND PHOSPHATASES IN 211. XANTHUS specific protein, so far. Neither is it known how many proteins are responsible for each activity. In an attempt to clone genes that encode proteins with phosphatase activity, three two-component regulatory systems of the family PhoRP have been identified in M. xanthus ( Carrero-Lirida et al., 2005; Moraleda-Muiioz et al., 2003). These three systems seem to be involved in the regulation of the expression of putative Mg2+-independent acid and neutral phosphatases, which are induced during development. Single- and double-deletion mutants for these systems exhibit reduced levels of these phosphatase activities and severe defects in both aggregation and sporulation in phosphate-free media (Carrero-LCrida et al., 2005; Moraleda-Muiioz et al., 2003), as described below.

The Protein Phosphatase Pphl Pphl is the only PP from M. xanthus that has been characterized to date (Treuner-Lange et al., 2001). It is a protein of 254 residues with a molecular weight of 28,308. It corresponds to MXAN2044 in the genome. This phosphatase belongs to the PP2C family and exhibits all the characteristics of this type of protein, such as dependency on Mn2+ and MgZ+,resistance to okadaic acid, and an ability to dephosphorylate phosphothreonine and phosphoserine. The gene that encodes this phosphatase was cloned in a search of proteins that interact with FrzZ by using the yeast two-hybrid system. The Frz system regulates the directed movement of cells during growth and development (seechapter 7 of this book). It consists of several proteins that are homologues of chemotaxis proteins of enteric bacteria. FrzZ contains two CheY-like domains connected by a linker rich in alanine and proline (Trudeau et al., 1996). As a result of the interaction between FrzZ and Pphl, a pphl mutant exhibits defects in cell motility as it spreads 10 to 30% less than the wild type on soft agar. This reduction in swarming seems to be originated by a defect in pilin transport, low levels of methylation of FrzCD, and low frequency of cell reversals. The pphl mutation also causes defects in late exponential growth and aggregation. Thus, mutant cells enter into stationary phase at low cell densities and originate abnormal fruiting bodies with a reduced number of myxospores. In addition to FrzZ, Pphl interacts with the protein kinase Pkn5l PktD7 in the yeast two-hybrid system, and this opens the door to the elucidation of the signal transduction pathways in which this kinase and phosphatase participate. To date, although there have been reported in M. xanthus several signal transduction pathways for kinases and phosphatases, in none of them have the two partners of the reversible phosphorylation been identified. Even in the case of Pphl, in spite of the reported interaction between

203

Pphl and FrzZ and PknSIPktD7, it remains unknown whether this interaction is mediated through the addition or removal of a phosphate group.

The Two-Component Systems of the Family PhoRP Three two-component systems of the family PhoRP were cloned while searching for genes that encode PPs in M . xanthus (Carrero-LCrida et al., 2005; Moraleda-Muiioz et al., 2003). These three systems are partially responsible for the expression of Mg2+-independent acid and neutral phosphatase activities, which are detected only during development. In addition, a fourth response regulator (PhoP4) has been found in the M. xanthus genome, which regulates the expression of the three developmentspecific phosphatase activities. This response regulator is orphan in the genome, but it has been demonstrated by using the yeast two-hybrid system that it interacts with the histidine kinase PhoR2 (Pham et al., 2006). It is so far unknown which genes encode the proteins responsible for phosphatase activities in M. xanthus extracts during vegetative growth and development. However, analysis of the genome seems to indicate that at least part of the activity on PNPP must be originated by specific PPs. Only one putative nonspecific acid phosphatase and two alkaline phosphatases are encoded by M. xanthus. If acid and neutral phosphatase activities on PNPP are partially due to PPs, the PhoRP systems would regulate the expression of these types of proteins. Nonetheless, it is known that pphl is not under control of any PhoRP system (CarreroLCrida et al., 2005; Moraleda-Muiioz et al., 2003). Single- and double-deletion mutants have been constructed in the PhoRP systems, and analyses of the mutants have revealed no effect on vegetative growth. On the contrary, these mutations cause severe defects in both aggregation and/or sporulation. Thus, single phoRP2 and phoRP3 mutants and the double-deletion mutant in these two systems originate flat fruiting bodies in phosphate-free media. However, cells are able to sporulate in these flat aggregates almost at the same level as the wild-type strain, although some myxospores fail to reshape completely, remaining as rod cells instead of becoming coccoids (Moraleda-Muiioz et al., 2003). A phoRPl deletion also causes defects in development, although they are not so dramatic. Thus, the mutant originates lower numbers of fruiting bodies than the wild-type strain, although this smaller number is correlated with a larger size. As a consequence, the number of spores is similar in both strains (Carrero-LCrida et al., 2005). On the contrary, a phoP4 mutant only originates a reduction in spore viability (Pham et al., 2006). It has still to be elucidated whether these phenotypes are the

REGULATORY MECHANISMS

204

result of the lower levels of expression of phosphatase activities or they are originated through defects in the expression levels of unknown genes that could be under the control of these systems.

PPs in the Genome PPs are more diverse in sequence than kinases. In fact, four major superfamilies of phosphatases exist (Shi et al., 1998): phosphoprotein phosphatases (PPP), Mn2+-or Mg2+-dependentprotein phosphatases (PPM), conventional phosphotyrosine phosphatases (CPTP), and low-molecular-mass phosphotyrosine phosphatases (LMMPTP).While PPP and PPM dephosphorylate phosphoserine and phosphothreonine residues, PTP dephosphorylate mainly phosphotyrosine residues. A group of PTP can dephosphorylate the three phosphoamino acids, and they have been named dual-specificity phosphatases. In addition, the four families of phosphatases exhibit differences in catalytic mechanism. Members of the PPP and PPM families, in spite of their sequence differences, are metalloenzymes which dephosphorylate their substrate in a single step through the activation of a water molecule by the metal. On the contrary, phosphatases belonging to the PTP families do not require metal, and dephosphorylation is performed in two steps, with the formation of a cysteinyl-phosphate enzyme intermediate (Barford et al., 1998). An inspection of the M . xanthus genome has revealed the presence of the four superfamilies of PPs, a fact that is observed only in a few prokaryotes, members of the PPP family being the most abundant. The total number of genes that encode PPs is 34, which corresponds to approximately one-third of the genes for kinases. However, this lower number of phosphatases than kinases seems to be a general rule among living organisms, prokaryotes and eukaryotes, rather than an exception. Among the 34 PPs, only four seem to be forming operons with protein kinases: three are PPPs, and the fourth one belongs to the PPM superfamily. The rest are either far from genes that encode kinases or encoded by complementary strands (see above).

The PPM Superfamily The PPM superfamily includes PP2C phosphatases, which are abundant in eukaryotes, and SpoIIE-like phosphatases, which have been reported only in the prokaryotes, with the exception of three found in Saccharomycetes (http://www.sanger.ac.uk/cgi-bin/Pfam/

speciesdist.pl?acc=PFO7228&id=SpoIIE&depth=all). The proteins of this superfamily consist of 11 subdomains, which contain the residues that are involved in the coordination of the metal ions of the active site (Shi, 2004). There are four genes in the M . xanthus genome

containing the Pfam for PP2C protein phosphatases (PF00481), including Pphl (Treuner-Lange et al., 2001). Three of these proteins (Table 3 ) have a number of amino acids that ranges from 247 to 265, approximately the size of the PP2C domain. On the contrary, M X A N 4 3 9 8 encodes a multidomain protein, with a cNMP-binding motif located in the C-terminal portion of the protein, and the PP2C consensus sequences in the amino terminus. This architecture has not been found in any other phosphatase and points to M X A N 4 3 9 8 participating in a signal transduction pathway where adenylyll guanylyl cyclases are implicated, responding to changes in the levels of cyclic nucleotides. PP2C phosphatases are not found in all the bacteria whose genomes have been sequenced. In those bacteria where PP2C phosphatases are encoded, their number ranges from one to three, with Anabaena sp. PCC 7120 (Kaneko et al., 2001) being the only bacterium, along with M . xanthus, where four PP2C phosphatases are present. In contrast, only one gene ( M X A N 4 5 6 2 ) has been found to encode a SpoIIE-like phosphatase (PF07228), a number that is extremely low if we compare M. xanthus with other bacteria that undergo developmental cycles, such as Bacillus subtilis, which carries 4 such genes (Duncan et al., 1995; Vijay et al., 2000; Yang et al., 1996), or S. coelicolor and Streptomyces avermitilis, which carry 48 and 4 7 such genes, respectively (Zhang and Shi, 2004b). The M . xanthus SpoIIE phosphatase exhibits a multidomain architecture, since it contains a HAMP domain in the middle portion of the protein and a SpoIIE domain in the carboxyl terminus (Table 3). This domain organization has also been reported in the phosphatase IcfG from Synechocystis (Beuf et al., 1994). M . xanthus SpoIIE seems to be in the same operon as the protein kinase, PskA12, so it is possible that they participate in the same signal transduction pathway, controlling the same process.

The PPP Superfamily The PPP superfamily of phosphatases has been defined by three domains, with the consensus sequences GDXHG, GDXXDRG, and GNHDIE, where X can be any amino acid (Zhang and Shi, 2004a; Shi et al., 1998). The search for these signature motifs in M . xanthus has revealed the existence of only two genes ( M X A N Z 5 0 9 and M X A N 5 4 6 7 ) . However, all phosphatases belonging to the PPP family contain the Pfam for metallophosphatases (PF00149, also named calcineurin-like phosphoesterases), which exhibits a logo that is not exactly identical to the sequence signatures defined by Shi and coworkers (http://www.sanger.ac.uk/cgi-bid Pfam/getacc?PF00149). The residues that are conserved in metallophosphatases are involved in metal binding,

11. PROTEINSEF~THR KINASESAND PHOSPHATASES IN M. XANTHUS

205

Table 3 List of protein phosphatases Superfamily PPM

Family and gene PP2C M X AN1412 MXAN2044 MXAN43 98 MXAN5349 SpoIIE MXAN4562

Size (aa)

Pfam

Location

247 254 442 265

PP2C PP2C PP2C PP2C

Cytoplasm Cytoplasm Cytoplasm Cytoplasm

521

SpoIIE

Membrane

PPP

MXANO149 M X ANO2 67 M X ANO344 MXAN0414 MXANO.555 MXAN0888 MXAN12 72 MXAN1509 MXAN2 613 MXAN3577 M X AN3 722 M X AN4086 MXAN4207 MXAN4.514 M X AN4 779 MXAN4 8 82 MXAN513 1 MXAN5467 MXAN6076 MXAN6105 MXAN6383 MXAN6.543 MXAN6890 MXAN6972 MXAN7163

417 367 245 3 79 276 465 290 235 310 485 260 398 399 3 75 316 234 256 339 217 276 426 336 389 300 230

Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase Metallophosphatase

Cytoplasm Cytoplasm Cytoplasm Membrane Cytoplasm Membrane Cytoplasm Cytoplasm Cytoplasm Cytoplasm Cytoplasm Membrane Membrane Cytoplasm Cytoplasm Cytoplasm Cytoplasm Cytoplasm Cytoplasm Cytoplasm Membrane Cytoplasm Cytoplasm Cytoplasm Cytoplasm

CPTP

MXAN0419 MXAN0448 MXANl665

214 193 237

Dual-specificity PTP Dual-specificity PTP Dual-specificity PTP

Cytoplasm Cytoplasm Membrane

LMMPTP

M X ANO575

135

Low-molecular-mass PTP

Cytoplasm

which is required for dephosphorylation (Egloff et al., 1995). A search of the M. xanthus genome has revealed 33 genes encoding proteins that contain the Pfam for metallophosphatases. As the signature for this family of proteins is also found in proteins such as nucleotidases, sphingomyelin phosphodiesterases, and 2'-3' CAMP phosphodiesterases, as well as nucleases, it is difficult to estimate the exact number of PPP-type phosphatases in M. xanthus. However, as 6 of the 33 proteins exhibit

an architecture where the metallophosphatase domain is found with a nucleotidase domain, at least these six genes must be considered to encode proteins with nucleotidase activity (MXAN2236, 2661, 4313, 5325, 5361, and 6266). Another one (MXAN2579) encodes a multidomain protein, with a PKD signature sequence in the C-terminal region. PKD domains received their name because they were first identified in the polycystic kidney disease protein PKDl (Bycroft et al., 1999).

REGULATORY MECHANISMS

206 They are involved in protein-protein and proteincarbohydrate interactions and are usually located extracellularly. This location matches with the fact that the 111. xanthus protein exhibits a signal peptide sequence, which indicates that it most likely will be secreted to the periplasmic space. For this reason we have also excluded this protein as a PPP. Finally, gene MXAN09.58 contains only the metallophosphatase domain, but it shows a great similarity with the C subunit of the exonuclease Sbc. In addition, the gene is located before another one that encodes a protein with similarities to the D subunit of the same type of exonuclease. For these two reasons, MXANO958 has not been considered to be a PPP either. The other 25 genes can be considered to encode putative serinehhreonine phosphatases (Table 3 ) . The fact that the 25 proteins possess most of the residues that are involved in binding the catalytic metals also suggests that

they will most likely exhibit phosphoprotein phosphatase activity. A logo of the three domains defined for PPPs obtained with the sequences of the 25 M . xanthus proteins is shown in Fig. 7 (Schuster-Bockler et al., 2004). Only experimental characterization of these proteins will shed light on their activity and function. If all these genes really encode PPPs, this number will be the highest so far reported in a bacterium. Even if we consider the density of genes, expressed as the number of PPPs per megabase of genome, M . xanthus will exhibit the highest density in all living organisms, prokaryotes and eukaryotes. Only 5 of the 25 proteins exhibit transmembrane domains in the N-terminal portion of the protein, which indicates that they must be anchored to the membrane, with the catalytic domain located in the cytoplasm. This topology makes these proteins very interesting, as they may somehow sense signals and function as the first

Domain I dq

Domain I 1

Domain 111 A4

Figure 7 Logo of the three domains of the 25 putative PPP superfamily protein phosphatases from M. xanthus.

11. PROTEINSER/THRKINASESAND PHOSPHATASES IN M. XANTHUS component of the signal transduction pathways in which they participate. The rest of the proteins are all located in the cytoplasm. T h e PTPs As mentioned above there are two superfamilies of proteins with the ability to dephosphorylate phosphotyrosine residues, CPTP and LMMPTP. These two superfamilies share the sequence signature CX,R. However, they are distinguished by the position of a catalytic aspartate, which is found 25 to 50 residues in the N-terminal direction from the active site cysteine in CPTP and 80 to 110 residues in the C-terminal direction from the catalytic cysteine in LMMPTP (Shi et al., 1998). There are three members in the M. xanthus genome with the consensus sequences of the CPTP family (Table 3 ) that exhibit the Pfam for dual-specificity phosphatases (PF00782). Therefore, they will most likely dephosphorylate not only phosphotyrosine residues, but also phosphoserine and phosphothreonine. On the contrary, there is only one gene that encodes a protein with the signature sequence for LMMPTPs (PF01451).As a protein with the same signature sequence has been experimentally shown to confer arsenate resistance (Bennett et al., 2001), the capacity of the M . xanthus protein to desphosphorylate phosphotyrosine remains questionable. Two of the CPTPs and the LMMPTP are located in the cytoplasm, while the third CPTP is anchored to the membrane through two transmembrane domains located in the N-terminal region of the protein. The four PTPs are not found scattered along the chromosome, as it can be observed with the PPs of the other families. On the contrary, all of them are clustered in the first quarter of the genome. Characterization of these PPs will undoubtedly shed some light on the significance of tyrosine phosphorylation in M . xanthus.

CONCLUDING REMARKS The physiological roles of the PSTK signaling systems in M . xanthus are beginning to be understood at the molecular level. Their roles appear to be similar to those of the protein Ser/Thr and Tyr kinases in eukaryotes, known to regulate diverse cellular functions by forming kinase cascades with scaffold and adapter proteins. In M . xanthus, a receptor-type PSTK, PktC2/Pkn8, forms a kinase cascade with a cytoplasmic PSTK, PskASIPknl4, that negatively regulates mrpC expression during vegetative growth. This cascade prevents the untimely initiation of development (Nariya and Inouye, 2005b). Expression of mrpC during development is activated by MrpA and MrpB, a histidine kinase and a response regulator of a two-component His-Asp phosphorelay system (Sun and

207

Shi, 2001). Accumulation of MrpC induces fruA expression, producing FruA, a key transcription factor for both C-signal-independent and -dependent gene expression (Fig. 6; see also Fig. 4 in chapter 9). FruA is likely to activate directly dofA and fdgA expression and, indirectly, tps and sasA expression. However, they are not essential for fruiting body formation and sporulation, suggesting that FruA possibly regulates the expression of other essential genes for development (see chapter 9). Thus, a PSTK signaling cascade, together with a twocomponent His-Asp phosphorelay system, regulates the timely expression of developmentally essential genes, mrpC and fmA, to achieve a fruitful development. The PSTK signaling pathways in M. xanthus are also modulated by the formation of PSTK networks mediated by MkapA, B, and c, which associate with multiple PSTKs (Fig. 5 ) . The activation of PFK by PktDl/Pkn4 is inhibited by MkapB, which also associates with PktC21 Pkn8 and PktA4/Pkn9 (Nariya and Inouye, 2005a). MkapA associates with the kinase catalytic domain of PktDl/Pkn4 and PktC2/Pkn8, suggesting that PFK activation by PktDl/Pkn4 can be modulated at a specific period of vegetative growth and in late development, during which mkapA expression is observed (Nariya and Inouye, unpublished results). Therefore, M . xanthus PSTKs appear to regulate cellular functions by forming complex signaling networks with adapter proteins via protein-protein interaction domains. As described in chapter 9, M . xanthus contains 1 3 FHA-EBPs. In addition, 3 1proteins containing the FHA domain have been identified in the genome database (Nariya and Inouye, unpublished results). A majority of them have no known function, but two are associated with Class I11 adenylyl/guanylyl cyclases (MXAN3956 and 6571) and one with EspA (MXAN0931 [Cho and Zusman, 19991). This is another clue that PSTKs regulate protein function by forming complexes mediated by well-known protein-protein interaction domains. To understand how PSTKs regulate the function of FHAproteins, identification of their upstream PSTKs is essential. Various methods are now available to detect these protein-protein interactions. Using PSTKs as baits in genomic yeast two-hybrid screens has been well established as PknSapl, a FHA-protein, was identified using PktA4/Pkn9 as bait (Nariya and Inouye, 2005a) (see also Fig. 5 ) . A genomic yeast two-hybrid screen can be used to identify the upstream PSTKs of FHA-proteins. Proteomic and phosphoproteomic approaches are other choices that show promise but are thus far unproven with M . xanthus PSTKs. Interestingly, a putative physiological substrate with the FHA domain, GarA, of PknB of M. tuberculosis was identified using the

208 proteomic approach (Villarino et al., 2005). Moreover, three physiological substrates of PknA and PknB of 111. tuberculosis were identified based on peptide library screening for their preferred phosphorylation motives (Kang et al., 2005). The roles of the upstream PSTKs can be addressed by identifying the downstream genes regulated by FHA-EBPs. The microarray expression profiling of nlal8, an FHA-EBP mutant (see Table 3 in chapter 9), has been demonstrated (Diodati et al., 2006). Analysis of the downstream genes’ expression and gene products will elucidate how PSTKs regulate FHA-EBP by phosphorylation and modulate cellular functions. Although mutational analysis showed that the majority of PSTKs were not essential for fruiting body formation and sporulation under the conditions used, they must be important in their natural habitats. The uniquely large number of PSTKs containing various functional domains in M . xanthus suggests an unusually flexible regulatory capacity in their signaling systems. Furthermore, the pattern and level of pstk gene duplications suggests that PSTKs have acquired many new functions, which have allowed refinement of preexisting functional capacities. Although a large number of PPs also exist in M . xanthus and they are known to play critical roles in PSTK signaling pathways, the interactions between PPs and PSTKs have not been well established in M. xanthus and their reversible phosphorylation remains to be elucidated. Further study of the PSTK signaling networks with Mkaps, PSTK-associated proteins, and PPs will lead us to understand the complicated signaling pathways, mediated not only by PSTKs but also by His-Asp phosphorelay systems, which govern the unique physiology of M . xanthus. We are grateful to M. Travisano for comments on the first part the manuscript. This work has been supported by grants from the Foundation of University of Medicine and Dentistry of New Jersey for S.I. and from the Ministerio de Ciencia y Tecnologia, Spain, for J.M.-D. (grant BMC2003-02038). of

References Adams, J. A., M. L. McGlone, R. Gibson, and S. S. Taylor. 1995. Phosphorylation modulates catalytic function and regulation in the CAMP-dependent protein kinase. Biochemistry 34:2447-2454. Akiyama, T., and T. Komano. 2004. Analysis of fruE, a novel developmental gene of Myxococcus xanthus. J. Mol. Microbiol. Biotechnol. 6:164-173. Av-Gay, Y., and M. Everett. 2000. The eukaryotic-like Serl Thr protein kinases of Mycobacterium tuberculosis. Trends Microbiol. 8:238-244. Barford, D., A. K. Das, and M.-P. Egloff. 1998. The structure and mechanism of protein phosphatases: insights into catalysis and regulation. Annu. Rev. Biophys. Biomol. Struct. 27:133-164.

REGULATORY MECHANISMS Bennett, M. S., Z. Guan,M. Laurber, andX.-D. Su. 2001. Bacillus subtilis arsenate reductase is structurally and functionally similar to low molecular weight protein tyrosine phosphatases. Proc. Natl. Acad. Sci. USA 98:13577-13582. Beuf, L., S. Bedu, M. C. Durand, and F. Joset. 1994. A protein involved in co-ordinated regulation of inorganic carbon and glucose metabolism in the facultative photoautotrophic cyanobacterium Synechocystis PCC6803. Plant Mol. Biol. 25:855-864. Bretscher, A. P., and D. Kaiser. 1978. Nutrition of Myxococcus xanthus, a fruiting myxobacterium. J. Bacteriol. 133:763768. Bycroft, M., A. Bateman, J. Clarke, S. J. Hamill, R. Sandford, R. L. Thomas, and C. Chothia. 1999. The structure of a PKD domain from polycystin-1: implications for polycystic kidney disease. EMBO J. 18:297-305. Carrero-Lkrida, J., A. Moraleda-Muiioz, R. Garcia-Hernandez, J. Perez, and J. Muiioz-Dorado. 2005. PhoR1-PhoP1, a third two-component system of the family PhoRP from Myxococcus xanthus: role in development. J. Bacteriol. 187:4976-49 83. Chang, C., A. Mooser, A. Pluckthun, and A. Wlodawer. 2001. Crystal structure of the dimeric C-terminal domain of TonB reveals a novel fold. J. Biol. Chem. 276:27535-27540. Cho, K., and D. R. Zusman. 1999. AsgD, a new twocomponent regulator required for A-signalling and nutrient sensing during early development of Myxococcus xanthus. Mol. Microbiol. 34:268-281. Diodati, M. E., F. Ossa, N. B. Caberoy, I. R. Jose, W. Hiraiwa, M. M. Igo, M. Singer, and A. G. Garza. 2006. Nla18, a key regulatory protein required for normal growth and development of Myxococcus xanthus. J. Bacteriol. 188:1733-1743. Duncan, L., S. Alper, F. Arigoni, R. Losick, and P. Stragier. 1995. Activation of cell-specific transcription by a serine phosphatase at the site of asymmetric division. Science 270:641-644. Egloff, M. P., P. T. W. Cohen, P. Reinemer, and D. Barford. 1995. Crystal structure of the catalytic subunit of human protein phosphatase 1 and its complex with tungstate. J. Mol. Biol. 254~942-959. Erickson, J. W., and C. A. Gross. 1989. Identification of the sigma E subunit of Escherichia coli RNA polymerase: a second alternate sigma factor involved in high-temperature gene expression. Genes Dev. 3:1462-1471. Foe, L. G., and R. G. Kemp. 1982. Properties of phospho and dephospho forms of muscle phosphofructokinase. J. Biol. Chem. 257:6368-6372. Frasch, S. C., and M. Dworkin. 1996. Tyrosine phosphorylation in Myxococcus xanthus, a multicellular prokaryote. J. Bacteriol. 178:4084-4088. Gill, R. E., M. Karlok, and D. Benton. 1993. Myxococcus xanthus encodes an ATP-dependent protease which is required for developmental gene transcription and intercellular signaling. J. Bacteriol. 175:4538-4544. Gilsdorf, J. R.,C. F. Mars, and B. Foxman. 2004. Haemophilus influenzae: genetic variability and natural selection to identify virulence factors. Infect. Immun. 72:2457-2461. Goldman, B. S., W. C. Nierman, D. Kaiser, S. C. Slater, A. S. Durkin, J. A. Eisen, C. M. Ronning, W. B. Barbazuk, M. Blanchard, C. Field, C. Halling, G. Hinkle, 0. Iartchuk,

11. PROTEINSEF~THR KINASESAND PHOSPHATASES IN M. XANTHUS H. S. Kim, C. Mackenzie, R. Madupu, N. Miller, A. Shvartsbeyn, S. A. Sullivan, M. Vaudin, R. Wiegand, and H. B. Kaplan. 2006. Evolution of sensory complexity recorded in a myxobacterial genome. Proc. Natl. Acad. Sci. USA 103:15200-15205. Hanks, S. K., A. M. Quinn, and T. Hunter. 1988. The protein kinase family: conserved features and deduced phylogeny of the catalytic domains. Science 241:42-52. Hanks, S. K., and T. Hunter. 1995. The eukaryotic protein kinase superfamily: kinase domain structure and classification. FASEBJ. 9576-596. Hanlon, W. A., M. Inouye, and S. Inouye. 1997. Pkn9, a Ser/ Thr protein kinase involved in the development of Myxococcus xanthus. Mol. Microbiol. 23:459-471. Inouye, S., R. Jain, T. Ueki, H. Nariya, C. Y. Xu, M. Y. Hsu, B. A. Fernandez-Luque, J. Muiioz-Dorado, E. Farez-Vidal, and M. Inouye. 2000. A large family of eukaryotic-like protein Ser/Thr kinases of Myxococcus xanthus, a developmental bacterium. Microb. Comp. Genomics 5:103-120. Irish, V. F., and A. Litt. 2005. Flower development and evolution: gene duplication, diversification and redeployment. Curr. Opin. Genet. Dev. 15:454-460. Jain, R., and S. Inouye. 1998. Inhibition of development of M~XOCOCCUS xanthus by eukaryotic protein kinase inhibitors. J. Bacteriol. 180:6544-6550. Jelsbak, L., M. Givskov, and D. Kaiser. 2005. Enhancer-binding proteins with a forkhead-associated domain and the sigma54 regulon in Myxococcus xanthus fruiting body development. Proc. Natl. Acad. Sci. USA 102:3010-3015. Kaneko, T., Y. Nakamura, C. P. Wolk, T. Kuritz, S. Sasamoto, A. Watanabe, M. Iriguchi, A. Ishikawa, K. Kawashima, T. Kimura, Y. Kishida, M. Kohara, M. Matsumoto, A. Matsuno, A. Muraki, N. Nakazaki, S. Shimpo, M. Sugimoto, M. Takazawa, M. Yamada, M. Yasuda, and S. Tabata. 2001. Complete genomic sequence of the filamentous nitrogenfixing cyanobacterium Anabaena sp. strain PCC 7120. D N A Res. 8:205-213. Kang, C. M., D. W. Abbott, S. T. Park, C. C. Dascher, L. C. Cantley, and R. N. Husson. 2005. The Mycobacterium tuberculosis serinelthreonine kinases PknA and PknB: substrate identification and regulation of cell shape. Genes Dev. 19:1692-1704. Kemp, R.G., L. G . Foe, S. P. Latshaw, R. A. Poorman, and R. L. Heinrikson. 1981. Studies on the phosphorylation of muscle phosphofructokinase. J. Biol. Chem. 256:7282-7286. Kimura, Y., Y. Mishima, H. Nakano, and K. Takegawa. 2002. An adenylyl cyclase, CyaA, of Myxococcus xanthus functions in signal transduction during osmotic stress. J. Bacterial. 184:3578-3585. Laronde-LeBlanc, N., and A. Wlodawer. 2005. The RIO kinases: an atypical protein kinase family required for ribosome biogenesis and cell cycle progression. Biochim. Biophys. Acta 1754:14-24. Lee, P. C., T. Umeyama, and S. Horinouchi. 2002. afsS is a target of AfsR, a transcriptional factor with ATPase activity that globally controls secondary metabolism in Streptomyces coelicolor A3(2).Mol. Microbiol. 43:1413-1430. Linder, J. U., and J. E. Schultz. 2003. The class I11 adenylyl cyclases: multi-purpose signalling modules. Cell. Signal. 15~1081-1089.

209

Mackay, J. P., and M. Crossley. 1998. Zinc fingers are sticking together. Trends Biochem. Sci. 23:l-4. Manning, G., D. B. Whyte, R. Martinez, T. Hunter, and S. Sudarsanam. 2002. The protein kinase complement of the human genome. Science 298:1912-1934. Moraleda-Muiioz, A., J. Carrero-Ltrida, J. Perez, and J. Muiioz-Dorado. 2003. Role of two novel two-component regulatory systems in development and phosphatase expression in Myxococcus xanthus. J. Bacteriol. 185:1376-1383. Muiioz-Dorado, J., S. Inouye, and M. Inouye. 1991. A gene encoding a protein serinekhreonine kinase is required for normal development of M . xanthus, a gram-negative bacterium. Cell 67:995-1006. Nariya, H., and S. Inouye. 2002. Activation of 6-phosphofructokinase via phosphorylation by Pkn4, a protein Ser/Thr kinase of Myxococcus xanthus. Mol. Microbiol. 46:13531366. Nariya, H., and S. Inouye. 2003. An effective sporulation of Myxococcus xanthus requires glycogen consumption via Pkn4-activated 6-phosphofructokinase. Mol. Microbiol. 49:5 17-528. Nariya, H., and S. Inouye. 2005a. Modulating factors for the Pkn4 kinase cascade in regulating 6-phosphofructokinase in Myxococcus xanthus. Mol. Microbiol. 56:1314-1328. Nariya, H., and S. Inouye. 2005b. Identification of a protein SeriThr kinase cascade that regulates essential transcriptional activators in Myxococcus xanthus development. Mol. Microbiol. 5 8:3 67-3 79. . that modulate the Nariya, H., and S. Inouye. 2 0 0 5 ~ Factors Pkn4 kinase cascade in Myxococcus xanthus. J. Mol. Microbiol. Biotechnol. 9:147-153. Nariya, H., and S. Inouye. 2006. A protein SerKhr kinase cascade negatively regulates the DNA-binding activity of MrpC, a smaller form of which may be necessary for the Myxococcus xanthus development. Mol. Microbiol. 60:1205-1217. Ohno, S. 1999. Gene duplication and the uniqueness of vertebrate genomes circa 1970-1999. Semin. Cell Dev. Biol. 10~517-522. Pao, S. S., I. T. Paulsen, and M. H. Saier, Jr. 1998. Major facilitator superfamily. Microbiol. Mol. Biol. Rev. 62:l-34. Petrickova, K., and M. Petricek. 2003. Eukaryotic-type protein kinases in Streptomyces coelicolor: variation on a common theme. Microbiology 1491609-1621. Pham, V. D., C. W. Shebelut, I. R. Jose, D. A. Hodgson, D. E. Whitworth, and M. Singer. 2006. The response regulator PhoP4 is required for late developmental events in Myxococcus xanthus. Microbiology 152:1609-1620. Schuster-Bockler, B., J. Schultz, and S. Rahmann. 2004. HMM Logos for visualization of protein families. BMC Bioinformatics 5:7-15. Shi, L., M. Potts, and P. J. Kennelly. 1998. The serine, threonine, and/or tyrosine-specific protein kinases and protein phosphatases of prokaryotic organisms: a family portrait. FEMS Microbiol. Rev. 22:229-253. Shi, L. 2004. Manganese-dependent protein 0-phosphatases in prokaryotes and their biological functions. Front. Biosci. 9: 1382-1 397. Stein, E. A., K. Cho, P. I. Higgs, and D. R. Zusman. 2006. Two Ser/Thr protein kinases essential for efficient

210 aggregation and spore morphogenesis in Myxococcus xanthus. Mol. Microbiol. 60:1414-1431. Steinbiichel, A., E. Hustede, M. Liebergesell, U. Pieper, A. Timm, and H. Valentin. 1992. Molecular basis for biosynthesis and accumulation of polyhydroxyalkanoic acids in bacteria. FEMS Microbiol. Rev. 9:217-230. Stock, A. M., V. L. Robinson, and P. N. Goudreau. 2000. Two-component signal transduction. Annu. Rev. Biochem. 69~183-215. Sun, H., and W. Shi. 2001. Analyses of mrp genes during Myxococcus xanthus development. J. Bacteriol. 183:67336739. Thomasson, B., J. Link, A. G. Stassinopoulos, N. Burke, L. Plamann, and P. L. Hartzell. 2002. MglA, a small GTPase, interacts with a tyrosine kinase t o control type IV pilimediated motility and development of Myxococcus xanthus. Mol. Microbiol. 46:1399-1413. Tojo, N., S. Inouye, and T. Komano. 1993. The lonD gene is homologous to the lon gene encoding an ATP-dependent protease and is essential for the development of Myxococcus xanthus.J. Bacteriol. 175:4545-4549. Treuner-Lange, A., M. J. Ward, and D. R. Zusman. 2001. Pphl from Myxococcus xanthus is a protein phosphatase involved in vegetative growth and development. Mol. Microbiol. 40~126-140. Trudeau, K. G., M. J. Ward, and D. R. Zusman. 1996. Identification and characterization of FrzZ, a novel response regulator necessary for swarming and fruiting-body formation in Myxococcus xanthus. Mol. Microbiol. 20:645-655. Udo, H., C. K. Lam, S. Mori, M. Inouye, and S. Inouye. 2000. Identification of a substrate for Pkn2, a protein Ser/ Thr kinase from Myxococcus xanthus by a novel method for substrate identification. J. Mol. Microbiol. Biotechnol. 2557-563. Udo, H., J. Muiioz-Dorado, M. Inouye, and S. Inouye. 1995. Myxococcus xanthus,a gram-negative bacterium, contains a transmembrane protein serinekhreonine kinase that blocks the secretion of beta-lactamase by phosphorylation. Genes Dev. 9:972-983. Udo, H., M. Inouye, and S. Inouye. 1996. Effects of overexpression of Pkn2, a transmembrane protein serinekhreonine kinase, on development of Myxococcus xanthus. J. Bacterial. 178:6647-6649. Udo, H., M. Inouye, and S. Inouye. 1997. Biochemical characterization of Pkn2, a protein Ser/Thr kinase from Myxococcus xanthus, a Gram-negative developmental bacterium. FEBS Lett. 400:188-192.

REGULATORY MECHANISMS Ueki, T., and S. Inouye. 2003. Identification of an activator protein required for the induction of fruA, a gene essential for fruiting body development in Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 100:8782-8787. Vijay, K., M. S. Brody, E. Fredlund, and C. W. Price. 2000. A PP2C phosphatase containing a PAS domain is required to convey signals of energy stress to the sigmaB transcription factor of Bacillus subtilis. Mol. Microbiol. 35:180-188. Villarino, A., R. Duran, A. Wehenkel, P. Fernandez, P. England, P. Brodin, S. T. Cole, U. Zimny-Arndt, P. R. Jungblut, C. Cervenansky, and P. M. Alzari. 2005. Proteomic identification of M. tuberculosis protein kinase substrates: PknB recruits GarA, a FHA domain-containing protein, through activation loop-mediated interactions. J . Mol. Biol. 350:953-963. Walburger, A., A. Koul, G. Ferrari, L. Nguyen, C. PrescianottoBaschong, K. Huygen, B. Klebl, C. Thompson, G. Bacher, and J. Pieters. 2004. Protein kinase G from pathogenic mycobacteria promotes survival within macrophages. Science 304:1800-1804. Wang, L., Y. P. Sun, W. L. Chen, J. H. Li, and C. C. Zhang. 2002. Genomic analysis of protein kinases, protein phosphatases and two-component regulatory systems of the cyanobacterium Anabaena sp. strain PCC 7120. FEMS Microbiol. Lett. 21E155-165. Weinberg, R. A., and D. R. Zusman. 1990. Alkaline, acid, and neutral phosphatase activities are induced during development in Myxococcus xanthus. J. Bacteriol. 172:2294-2302. Yajko, D. M., and D. R. Zusman. 1978. Changes in cyclic AMP levels during development in Myxococcus xanthus. J. Bacteriol. 133:1540-1542. Yang, X., C. M. Kang, M. S. Brody, and C. W. Price. 1996. Opposing pairs of serine protein kinases and phosphatases transmit signals of environmental stress to activate a bacterial transcription factor. Genes Dev. 10:22652275. Zhang, W., and L. Shi. 2004a. Comparative analysis of eukaryotic-type protein phosphatases in two streptomycete genomes. Microbiology 150:2247-2256. Zhang, W., and L. Shi. 2004b. Evolution of the PPM-family protein phosphatases in Streptomyces: duplication of catalytic domain and lateral recruitment of additional sensory domains. Microbiology 150:4189-4197. Zhang, W., M. Inouye, and S. Inouye. 1996. Reciprocal regulation of the differentiation of Myxococcus xanthus by Pkn5 and Pkn6, eukaryotic-like Ser/Thr protein kinases. Mol. Microbiol. 20:435-447.

Myxobacteria: Mtrlticellularity and Differentiation Edited by David E. Whitworth 02008 ASM Press, Washington, D.C.

Montserrat Elias-Arnanz Marta Fontes S. Padmanabhan

Carotenogenesis in Myxococcus xanthus:a Complex Regulatory Network

CAROTENOIDS Myxobacteria frequently occur as brightly colored colonies and sporangioles, due to the presence of carotenoids and/or other pigments (Reichenbach and Kleinig, 1984; Hodgson and Murillo, 1993; Hodgson and Berry, 1998). The carotenoid pigments, which range in color from light yellow to deep red, form a major class of lipophilic isoprenoids that include the hydrocarbon carotenes with acyclic, monocyclic, or bicyclic ends and their oxygenated (hydroxy, aldehydic, keto, carboxyl, methoxy, oxy, epoxy, and glycosidic) derivatives known as xanthophylls. Most natural carotenoids are C,, terpenes derived from eight isoprenoid units (although some C3,, C45, and C,, have been reported), the colorless C40 phytoene being the universal progenitor in their biosynthesis. The color of a given carotenoid is determined by the number of conjugated double bonds, which exist mostly in the all-trans conformation except for phytoene, which often occurs as the 15,15’-cis isomer. The carotenoids serve important biological roles and are widely distributed in nature, being synthesized de novo in anoxygenic photosynthetic bacteria, cyanobacteria, some nonphotosynthetic bacteria, some fungi, and all algae and plants, but need to be provided as dietary supplements in animals

12

( Goodwin, 1980). Being relatively hydrophobic, carotenoids are typically associated with membranes and may or may not be specifically bound to proteins. Their primary function is to protect cells against photo-oxidative damage by quenching the triplet excited states of photosensitizer molecules such as chlorophyll and porphyrins and the singlet excited state of oxygen that result from absorption of light energy. They also play roles in light harvesting and as redox intermediates in the shuttling of electrons in photosynthesis (Frank and Brudvig, 2004). Carotenoids also serve as precursors for molecules required for photoreception and hormonal action, the classic example being the production of retinoids from carotenes. The chemistry and biosynthesis of carotenoids, their species-specific distribution, and their cellular functions and localization have been extensively discussed in various other reviews, including those with an emphasis on carotenoids in eubacteria in general (Armstrong, 1997) or specifically on myxobacteria (Reichenbach and Kleinig, 1984; Hodgson and Murillo, 1993; Hodgson and Berry, 1998). Almost 1 decade has passed since the last review on carotenogenesis in Myxococcus xanthus was published, undoubtedly one of the most important

Montserrat Elias-Arnanz, Marta Fontes, and S. Padmanabhan, Departamento de GenCtica y Microbiologia, Facultad de Biologia, Universidad de Murcia, 30100 Murcia, Spain.

211

212

model systems for the genetic analysis of light-induced carotenogenesis. The aim of the present review is to provide an update on the considerable amount of new information that has since been acquired with respect to the inducing signals, their reception and transduction, and the transcriptional regulation of the structural genes involved in carotenoid biosynthesis in M. xanthus.

CAROTENOID SYNTHESIS IN M . XANTHUS-AN INDUCIBLE RESPONSE A Brief Overview of the Known Genetic Elements Involved Early work showed that carotenoid production in M. xanthus is a response induced on exposure to blue light (Burchard and Dworkin, 1966). No carotenoids or their precursors are found in dark-grown M. xanthus cultures (Martinez-Laborda et al., 1990), which appear yellow due to noncarotenoid pigments, named DKxanthenes, that have only recently been characterized (Meiser et al., 2006). In the light, cells turn red due to production of carotenoids, and this wild-type, light-inducible phenotype has been designated Car+.Loss of the ability to synthesize carotenoids would maintain the yellow color under all conditions (Car- phenotype), while their constitutive production would keep cells always red ( Carcphenotype). By means of this conspicuous color change, spontaneous as well as chemical-, W-,and transposon-induced mutants were isolated. While Car' mutants would be affected in the regulation of light-induced carotenogenesis, Carmutants could result either from the loss or from the inactivation of an enzyme essential for a step leading to the synthesis of a colored carotene or of a positive regulator of carotenogenesis. Analysis of Car- and Carc mutants obtained primarily by transposon insertions allowed the early mapping and identification of structural and regulatory genes, and their epistatic analysis. These were later extended by chemical analysis of the carotenoid content of the mutants, gene expression assays, and cloning and sequencing of the affected loci. More recently, in silico analysis of the M. xanthus genome or the study of proteins that physically interact with those previously identified has allowed the identification of new elements involved in the light induction of carotenoid synthesis. Also, a number of the corresponding gene products have been subjected to detailed molecular analysis of their domain organization, structures, and modes of action. The designation crt has been employed for structural genes that encode the enzymes with known or putative functions in the carotenoid biosynthetic pathway, and the cur notation has been used for regulatory genes and for all operons, including those for structural genes. Figure 1

REGULATORY MECHANISMS summarizes our present knowledge of the genetic elements involved and their interactions. Structural genes for carotenoid synthesis are located at two unlinked loci: crtlb and carBA, the latter of which is constituted by two contiguous operons (curl3 and curA). The carB operon groups six structural genes, and its expression is controlled by a light-inducible promoter (Martinez-Laborda et al., 1986, 1989, 1990; Balsalobre et al., 1987; Ruiz-VAzquez et al., 1993; Botella et al., 1995). Immediately downstream of carB lies the carA operon, made up of structural and regulatory genes which are expressed in a light-independent manner (Martinez-Laborda et al., 1986; Hodgson, 1993; Botella et al., 1995). The unlinked crtIb gene is under the control of a light-inducible promoter but is fully activated only in stationary-phase cells (Balsalobre et al., 1987; Fontes et al., 1993). The light-dependent expression of carB and crtIb is controlled by the products of the carQRS operon, constituted by three translationally coupled genes whose expression is also induced by light (Hodgson, 1993; McGowan et al., 1993). CarQ and CarR form an ECF (extracytoplasmic function) sigma-antisigma pair that acts in regulating the expression of carQRS itself and crtlb (Gorham et al., 1996; Martinez-Argudo et al., 1998; Browning et al., 2003; Whitworth et al., 2004). Cars activates carB expression in the light by counteracting the action of CarA, which is a repressor of carB in the dark and is encoded by a gene in the carA operon (Whitworth and Hodgson, 2001; Cervantes and Murillo, 2002; L6pez-Rubio et al., 2002, 2004; Perez-Marin et al., 2004; Navarro-AvilCs et al., 2007). Other regulatory factors, all expressed independent of light, include (i)CarF, which may be a receiver or transmitter of the light signal (Fontes et al., 2003); (ii) CarD, a global regulator of transcription similar to the eukaryotic high-mobility group A proteins (Nicolh et al., 1994; 1996; Padmanabhan et al., 2001; Cayuela et al., 2003); (iii) CarG, a recently identified protein that acts in concert with CarD (Pefialver et al., 2006); and (iv) the product of ihfA, the ortholog of the Escherichia coli integration host factor (Moreno et al., 2001). Our present understanding of each of these structural and regulatory genes and the specific functions of their products will be discussed in the following sections.

The Structural Genes and Their Functions As mentioned earlier, structural genes for carotenoid synthesis cluster at the carBA locus, with the exception of crtlb. Their activitieswere assigned by analyzing the carotenoids accumulated in different mutants, from sequence homology to previously characterized genes, and by heterologous expression in E. coli (Martinez-Laborda et al.,

12. CAROTENOGENESIS IN 211. XANTHUS

213 copper

blue-light

h crtE-la-6-D-C orf6 )IcrtYc-Yd orf9 carA orfll car6operon

Y

carA operon

Figure 1 Known genetic elements involved in carotenogenesis and their interactions. Squat arrows show photoinducible genes (crtlb)or operons (carQRS and the carB-carA cluster with the constituent genes indicated) implicated in the carotenogenic pathway. Genes for other proteins involved whose expression is not light dependent are not shown. Labeled ovals are the regulatory factors that have been identified and characterized, with continuous arrows indicating positive regulation, and blunt-ended lines indicating negative regulation. Carotenoid biosynthesis enzymes are encoded by crtlb, all of the genes in the carB operon, and crtYc and crtYd in the carA operon. The latter operon also codes for the regulatory factors CarA and O r f l l . See the text for further explanation.

1990; Ruiz-VBzquez et al., 1993; Botella et al., 1995; Cervantes, 2000; Iniesta, 2003). Genes at the carBA region are organized as crtE-crtla-crtB-crtD-crtC-orf6-crtYccrtYd-orf9-carA-orfl l, where the ones designated as orf are genes whose functions remain to be fully assigned, and the two most downstream are regulatory genes (Fig. 1). Figure 2 shows the proposed pathway for carotenoid biosynthesis in M . xanthus and the genes involved in each step. The pathway represents the basic steps for the synthesis of myxobacton, the primary carotenoid end product in lightgrown M . xanthus (large amounts of phytoene, the first C,, precursor in the biosynthetic pathway, are also found in light-grown M. xanthus, as well as lower amounts of other end products not shown in Fig. 2). Myxobacton is a monocyclic carotenoid with a keto group on the ring at one end of the molecule and a glycosyl group usually esterified to a straight-chain fatty acid at the other (for a detailed review on the structures and biosynthesis of myxobacterial carotenoids see Reichenbach and Kleinig, 1984). CrtE, the product of the first gene of the carB operon, catalyzes the conversion of farnesyl pyrophosphate into geranylgeranyl pyrophosphate (GGPP).Condensation of

two GGPP molecules to yield the colorless C,, phytoene is carried out by CrtB. Four successive dehydrogenation steps introduce conjugated double bounds into the chain to produce lycopene, a colored intermediate. Unlike other bacteria, M. xanthus requires two dehydrogenases (CrtIa and CrtIb) to transform phytoene into lycopene, as has been shown by heterologous expression of CrtIa and/or crtlb in E. coli strains that accumulate phytoene (Iniesta, 2003). Among bacteria, only cyanobacteria are known to use two different enzymes for this conversion, but their action is sequential and not synergetic as in M . xanthus. The cyclization at one end of lycopene to produce y-carotene is catalyzed by the combined action of CrtYc and CrtYd, so denoted from their amino acid sequence homology to heterodimeric lycopene cyclases from gram-positive bacteria, and to the N-terminal domain of bifunctional fungal lycopene cyclases, whose C-terminal regions also show phytoene synthase activity (Krubasik and Sandmann, 2000). Heterologous gene expression of crtYc or crtYd, or the two together, in E. coli strains overproducing lycopene showed that their products act in concert to catalyze cyclization of lycopene but at only one end of the molecule, unlike their

REGULATORY MECHANISMS

214

*cH20pP FP P m

C

4 H

CrtE 2

O

P

P

GGPP

CrtB

1

dehydrogenases, catalyzes an additional desaturation reaction. Gene orf6, located at the 3‘ end of the carB operon, codes for a protein similar to bacterial glycosyltransferases. So, Orf6 may catalyze the addition of the sugar moiety to the chain. At present, it is not known if the product of orf9 plays a structural or a regulatory role. While some data point to Orf9 participating in the regulation of the light-inducible promoters (Cervantes and Murillo, 2002), its sequence similarity to bacterial acyltransferases makes it tempting to assign a role for Orf9 in myxobacton esterification.

Crtla + Crtlb Molecular Aspects of Signal Reception and Transduction in Carotenogenesis The Signals-Blue Light and Copper

Lycopene

1

CrtYc + CrtYd

y-carotene

crtc

1 1-OH-1’2’dihydroy-carotene

1 I-OH-3’4’dehydro1‘2’d ihyd roy-carote ne

Orf6? Orf9?

Fatty acid

Figure 2 Carotenoid biosynthetic pathway and the genes involved in each step. The proposed pathway for the biosynthesis of myxobacton, the primary carotenoid in M. xanthus, is shown. The roles of Orf6 and Orf9 remain to be defined but may be linked to the glycosyl- and acyltransferase activities, respectively, that are essential for production of myxobacton. counterparts mentioned above that do so at both ends (Iniesta, 2003). CrtC catalyzes the hydroxylation of y-carotene, while CrtD, a protein similar to carotene

Burchard and Dworkin (1966) demonstrated that accumulation of carotenoids in M. xanthus was dependent on the extent of illumination and the growth phase. Carotenoid levels, whose maximum build-up occurred for stationaryphase cells irradiated with blue light (405to 410 nm), were found to correlate directly with increasedprotection against photolysis (Burchard and Hendricks, 1969). The observation that the action spectra for photolysis and carotenogenesis in M. xanthus were similar and corresponded closely to the absorption spectrum of the iron-containing protoporphyrin IX, led to the proposal that this compound was the photosensitizer that linked the processes of photolysis and carotenogenesis (Burchard et al., 1966; Burchard and Hendricks, 1969). This highly hydrophobic compound accumulates in cell membranes, with an almost 16-fold increase in its levels as cell cultures approach stationary phase. Protoporphyrin IX absorbs strongly at 410 nm to form the relatively long-lived yet highly reactive “excited triplet state” species, which can directly cause cell damage and photolysis by modifying cellular components such as lipids, proteins, and nucleic acids. The excited triplet-state protoporphyrin IX can also react with oxygen to generate another highly reactive and diffusible singlet oxygen (‘02), species, which is significantly long-lived in hydrophobic environments such as membranes. Both of these reactive species are quenched by carotenoids. The major site for accumulation of the hydrophobic carotenoids, as with protoporphyrin IX and lo2,is the cytoplasmic membrane (Kleinig, 1972), which is thus the cellular location where the initial signal reception and transmission, as well as the final quenching by carotenoids, take place. Blue light is the principal environmental agent required for the induction of carotenogenesis in M. xanthus. It has been reported that the need for light can be partially bypassed by hyperoxygenation in the dark, suggesting that an oxygen-related species could be the actual

12. CAROTENOGENESIS IN 211. XANTHUS switch that triggers the cascade culminating in carotenogenesis (Robson, 1992; Hodgson and Murillo, 1993). Indications that this effector was '0, came from the observation that expression of the carQRS operon could be induced by agents capable of generating lo2,such as the dye toluidine blue in the presence of both oxygen and red light (Robson, 1992). On their own, neither the dye nor red light was capable of inducing carQRS expression. Moreover, quenchers of '0,other than carotenoids also greatly diminished the effect of light. It has very recently been found that carotenoid synthesis in wild-type M . xanthus in the absence of light can be induced by copper (Moraleda-Muiioz et al., 2005). Like many essential transition elements, copper must be maintained at specific levels in cells but is toxic at elevated concentrations. At metal concentrations below the lethal threshold, copper-induced carotenogenesis is observed under conditions suboptimal for growth like lower temperatures and diminished nutrient levels. Every regulatory element implicated in the light-triggered induction of carotenoid synthesis is found to also respond to copper, with the one exception discussed below. While the chemistry of the copper response and whether its action is direct or indirect remain to be worked out, it is noteworthy that can be generated from the membranelocated phosphatidylcholine hydroperoxide if copper is present (Takayama et al., 2001). Thus, lo2could provide the common link between the light- and the copperinduced pathways to carotenogenesis in M . xanthus.

Role of CarF and CarR in Signal Reception or Transduction Available data suggest that induction by light and copper converge to a common route very early in carotenoid synthesis in M . xanthus, as all but one of the known regulatory elements involved respond to both light and copper, except for the factor CarF (Moraleda-Muiioz et al., 2005). CarF, whose expression is not affected by illumination, is the factor that comes into play earliest in the response to light (Fontes et al., 2003). Its precise mechanism of action remains to be elucidated, but two alternatives have been considered. Either CarF is the primary photoreceptor working directly or through an associated chromophore, or it is a transmitter of the light signal from the primary to-be-identified photoreceptor to the next junction in the cascade. Sequence analysis failed to reveal any similarity of CarF to known blue light receptors, prokaryotic or eukaryotic. Furthermore, the only proteins showing any resemblance to the CarF sequence are members of a family of proteins of undefined functions named Kua (see Fontes et al., 2003). Whatever may be the exact function of CarF, its role is in the response to light and not in

215

that to copper. If the responses to both light and copper are, as suggested, dependent on the same inducer (singlet oxygen), then CarF would be specifically required in the photosensitizing mechanism to generate that inducer. Epistatic analysis established that CarF is followed in the light-induced cascade by CarR. Mutations in carR were among the earliest to be mapped in this light-induced process (Martinez-Laborda et al., 1986; Balsalobre et al., 1987). These cause constitutive synthesis of carotenoids, indicating that CarR is a negative regulator of light-induced carotenogenesis. carR is the second of three translationally coupled genes of the light-inducible carQRS operon, which plays a central role in the induction of carotenogenesis (McGowan et al., 1993).Its expression increases almost 80-fold when exposed to light (Hodgson, 1993). CarR is an inner membrane protein that is unstable when exposed to light, particularly when cells enter the stationary phase (Gorham et al., 1996; Browning et al., 2003). It is worth noting that maximal levels of carotenoids are also produced in the early stationary phase. These, however, do not appear to afford any protection to the lightmediated elimination of CarR (Browning et al., 2003). These observations have led to the proposal that light induces carotenogenesis by bringing about degradation of CarR, to thereby relieve the CarR-mediated repression of the process. CarF would then be expected to be involved in channeling this light-driven inactivation of CarR. The functional link between CarR and CarF appears to be via direct physical interactions between them, suggesting that CarF is an anti-anti-sigma factor (Fontes et al., 2003; Galbis-Martinez, 2005; also see below). The molecular basis for light- or copper-induced CarR destruction is, however, unknown. Whether these agents chemically modify CarR to lower its stability or whether it is modifications of the proteins associated with CarR that cause the release and the consequent destruction of an unstable CarR remains an intriguing question. Besides CarF, at least one other key factor is known to be physically associated with CarR, as will be discussed next.

Specific and Global Transcriptional Factors in the Regulation of the curQRS Operon

CarQ Cells bearing lack-of-function mutations in carQ or cars, the first and third genes of the carQRS operon, are Car- (McGowan et al., 1993; Gorham et al., 1996).Thus, CarQ and Cars, in contrast to CarR, are positive regulators of carotenogenesis. Mutations at carQ are epistatic over those at carR and block the activation of the crtIb structural gene and of the carQRS operon itself (Fontes et al., 1993; McGowan et al., 1993; Gorham et al., 1996).

REGULATORY MECHANISMS

216 The CarQ sequence suggested it to be a member of the extracytoplasmic function (ECF) subfamily of bacterial ~'O-likefactors (Lonetto et al., 1994). ECF cr factors act in cellular responses to different extracytoplasmic stimuli, and their activity is typically modulated by their association with membrane-bound anti-a factors (Hughes and Mathee, 1998). In vitro transcription runoff assays using E. coli core RNA polymerase indicated that CarQ could specifically initiate transcription at PQRS,the curQRS promoter (but not at PI, the crtIb promoter-see below), and that it physically interacted with CarR (Browning et al., 2003). CarR would then be an anti-o factor. As with many other ECF a-anti-cr factor pairs the genes for CarQ and CarR are translationally coupled. The membraneassociated CarR sequesters CarQ in a transcriptionally inactive complex. Inactivation of CarR, possibly via degradation, in response to specific environmental signals (light and copper) and the consequent release of CarQ appear to be the mechanism by which carotenogenesis is induced. The -10 region of PQRs (Fig. 3 ) , like other ECF crfactor-dependent promoters in M. xanthus, shares little

P,S

or no sequence identity with the consensus sequences for known cr factors (Whitworth et al., 2004). On the other hand, the -35 region of PQRsshows a greater resemblance to the consensus for ECF cr-factor-dependent promoters (Lonetto et al., 1994). Using a series of nested deletions, the minimal carQRS promoter was mapped to a 145-bp stretch extending upstream from the transcriptional start point, which also includes the promoter for the divergent gufA (for gene of unknown function). Mutational analysis of this minimal promoter identified several positions important for promoter activity and suggested transcriptional coupling between the gufA and carQRS promoters (Whitworth et al., 2004). It was also proposed that DNA between positions +40 and +66 may influence mRNA stability or have enhancer activity, just as reported earlier with P, (Martinez-Argudo et al., 1998; see below). As mentioned previously, a minimal heterologous in vitro transcription system with CarQ and E. coli core RNA polymerase alone could be made to initiate transcription at Pqns (Browning et al., 2003). However, the rather large size of the promoter region for

-35 I

-10 I

TCACCGAACCTTGAG~GCGCGAGCGCCG~~CTTTCGCAGGTGGCCCGTAGAGGAGTCG AGTGGCTTGGAACTCTTCGCGCTCGCGGCCTTTGTG~GCGTC~CCGGGCATCTCCTCAGC

CarD-binding

PI3

-35

-10

I

I

-35

-10 I

TGGACGCAAACGCTACCTCTAGGAAA ACCTGCGTTTGCGATGGAGATCCTTT

CarA-binding

Figure 3 Design of the light-inducible promoters identified in M. xanthus. Sequences show the promoter regions of PQRS,P,, and P,, the three M. xanthus light-inducible promoters. The -10 and -35 bases are marked for all three promoters. In PQRS,the CarD-binding site is boxed (with the two AT-rich tracts underlined). Two bases at the -35 region of PQRS,which on mutation remove light-induced promoter activity, are underlined (Whitworth et al., 2004). The bases underlined in the P, promoter sequence are critical for promoter activity (MartinezArgudo et al., 1998). In the P, promoter region, the inverted repeats of the bipartite CarA operator, PI and pII, are boxed and marked by arrows.

12. CAROTENOGENESIS IN 211. XANTHUS

217

carQRS suggests that accessory transcriptional regulators come into play in activating this promoter and, as described next, several factors have indeed been found.

The CarD-CarG complex The carD gene was identified in a screen for Carmutants among a large collection of strains bearing Tn.5 insertions (Nicolb et al., 1994). carD is actively and continuously transcribed during vegetative growth in a light-independent manner. It was established that the critical role of CarD in carotenogenesis was in the lightinduced activation of the carQRS operon and of crtIb (see below). The sequence of carD revealed it to code for a very novel transcriptional factor in prokaryotes, with attributes more akin to eukaryotic transcriptional factors (Nicolhs et al., 1996).Thus, the 136-residue C-terminal segment of CarD contains four repeats of the so-called "AT-hook" motif (a conserved RGRP sequence embedded in a cluster of basic residues and proline) linked to a highly acidic region, an arrangement very similar to that in eukaryotic high-mobility group A (HMGA) proteins (Fig. 4). Eukaryotic HMGA, of which the most extensively studied are the mammalian proteins, are small (I, 107 residues), relatively abundant nonhistone components of chromatin that function as DNA architectural factors to remodel chromatin and prime it for the assembly of specific nucleoprotein complexes essential in transcription, replication, recombination, and repair (Reeves, 2001,2003). The HMGA-like module in CarD is autonomously stable and monomeric, with an intrinsically random structure that exhibits specific minor-groove DNAbinding to appropriately spaced AT-rich tracts characteristic of eukaryotic HMGA (Padmanabhan et al., 2001).

1

180 183

Pfam 02559 4

228229

316

HMGA-like - b

Interaction with CarG

Interaction with DNA

Figure 4 Domain organization of CarD. The independent domain formed by the first 180 N-terminal residues of CarD, which defines the family Pfam 02559 of CarD-like proteins, is schematically represented by an unfilled rectangle. The Cterminal HMGA-like domain is made up of two stretches: a highly acidic one represented by the filled rectangle (residues 183 to 228) and a basic one which contains the four AT-hook repeats represented by the four stippled boxes (residues 229 to 316). The specific CarD domains that interact with CarG and DNA are as indicated.

Moreover, like the latter, CarD plays multifunctional roles in vivo. Besides being required for light-induced carotenogenesis, CarD is necessary for multicellular development and for the regulation of various vegetatively expressed genes (Nicolhs et al., 1994; Galbis-Martinez et al., 2004; see below). CarD has an N-terminal stretch of around 180 amino acids that is absent in eukaryotic HMGA (Fig. 4). This segment of CarD defines a family of proteins (Pfam 02559) each of whose members exist as a stand-alone module in a diverse array of bacterial species, including 211. xanthus. This domain of well-defined structure is essential for CarD function in vivo, as its absence results in a phenotype identical to that caused by a complete deletion of carD. Acquisition of the eukaryotic HMGA segment via lateral gene transfer, followed by fusion to a module existing only in bacteria, has been invoked to account for the unique domain architecture of CarD (Cayuela et al., 2003). Activation of the light-inducible PQRs promoter depends on a DNA segment which includes two appropriately spaced AT-rich tracts at -63 to -77 bp relative to the transcription start site (Fig. 3; Nicolhs et al., 1996). In vitro gel shift assays show that CarD binds to DNA probes containing these tracts via its HMGAlike C-terminal domain with the expected minor-groove binding specificity (Padmanabhan et al., 2001; Cayuela et al., 2003), and DNase I footprinting maps CarDbinding to the above two AT-rich tracts (Peiialver-Mellado et al., 2006). Intriguingly, it has been reported that only mutation of the more upstream AT-rich tract completely inactivates PQRS,and it was thus suggested that the two AT-rich tracts have different roles in vivo (Whitworth et al., 2004). Recent studies have identified a new factor, CarG (the product of the gene directly downstream of carD), which is required in every CarD-dependent process analyzed (Pefialver-Mellado et al., 2006). Although CarG shows no significant overall similarity to any protein in the database, it does contain a conserved H/C-rich segment, HEx,Hx,Gx,HCX,CXMX~~CX~C (where x is any amino acid) that is found in predicted zinc-dependent metalloproteases found in Archaea (hence named archaemetzincins), the hyperthermophilic bacterium Aquifex aeolicus, and two recently reported cases in humans (Diaz-Perales et al., 2005). The motif in CarG has the conserved G replaced by E, and even more significantly a Q replaces the invariant E, a change known to eliminate protease activity but not zinc binding. Purified CarG has been shown to lack protease activity and to contain two equivalents of zinc; also, a crucial structural role for the conserved Cs has been demonstrated by sitedirected mutagenesis. CarG thus appears to be a novel

218 transcriptional factor in which a metalloprotease motif has evolved to fulfill a purely structural role. Although no DNA-binding activity has been detected for the zincbound CarG, it associates with DNA through physical interaction with the N-terminal domain of CarD (Peiialver-Mellado et al., 2006). That CarD and CarG form a complex rationalizes not only how they simultaneously regulate different processes but also why they always coexist in the bacteria in which they occur. Analysis of microbial genomes for the presence of proteins similar to CarD and CarG indicates that these exist only in myxobacteria and that when one is present in a myxobacterium, so is the other: carD and carG exist in Stigmatella aurantiaca and Anaeromyxobacter dehalogenans, but both are absent in Sorangium cellulosum (Cayuela et al., 2003; Peiialver et al., 2006). What role CarG provides to the complex in transcriptional regulation is currently under study. One working hypothesis is that CarG serves as an adaptor to bridge the DNA-bound CarD with the basal transcriptional machinery. A direct interaction of CarG, CarD, or the CarD-CarG complex with CarQ, or M . xanthus RNA polymerase holoenzyme, has so far not been observed. It is therefore likely that other factors also participate in transcriptional regulation orchestrated by CarD and CarG, and these remain to be identified.

IhfA One other factor that has been experimentally shown to play a role in the initiation of transcription at PQRs is the nucleoid-associated integration host factor (IHF) (Moreno et al., 2001). It was identified in a screen for Car- mutants as a mutation unlinked to any of the known genes involved in carotenogenesis. The mutation mapped to ihfA, which encodes the or-subunit of the IHF heterodimer, a histone-like global transcriptional factor. Mutations in ihfA are epistatic over those in carR, and genetic data suggest that its gene product participates directly in activating PQRS.The IHF heterodimer and the related HU (heat unstable nucleoid protein) homodimer are DNAbinding architectural factors whose ability to bend DNA facilitates the assembly of higher-order nucleoprotein complexes essential in diverse processes including recombination, transcription, and replication (Nash, 1996).In contrast to nonspecific DNA-binding for HU, IHF binds preferentially to the consensus WATCAANNNNTTR (Friedman, 1988). Although no exact match to this consensus has been found at the PQRs promoter region, we cannot discard the possibility that IHF does interact with this region, as the strong “HU-like” character of M. xanthus IHF could confer it with the ability to bind DNA less specifically than, or differently from, E. co2iIHF (Moreno

REGULATORY MECHANISMS et al., 2001). Thus, IHF may activate PQRsby directly participating in the assembly of an essential transcriptionally competent complex together with CarD and CarG or by controlling the expression of an as yet unknown gene.

Transcriptional Regulation of the Structural Gene crtIb As mentioned earlier, the monocistronic crtlb codes for one of the two dehydrogenases involved in the conversion of phytoene to lycopene. Activity at the crtlb promoter, PI, is stimulated by light, but maximal induction (-400-fold) occurs only when cells reach stationary phase or are starved of a carbon source. Genetic evidence indicated that expression of PI requires CarQ (Fontes et al., 1993). As with PQRs and other ECF o-factordependent promoters in M. xanthus, the -35 region of PI but not the -10 region resembles the consensus (Fig. 3; Martinez-Argudo et al., 1998; Whitworth et al., 2004). Mutational analysis of P, revealed that promoter activity is critically dependent on a 5-bp stretch around position -31, of which four are conserved in PQRS,and three contiguous base pairs at - 10, which are also present in PQRs (Fig. 3; Martinez-Argudo et al., 1998).However, in contrast to PQns, no transcription was observed at PI with a minimal heterologous in vitro transcription system employing CarQ and E. coli RNA polymerase (Browning et al., 2003), suggesting the involvement of additional factors. As with PQRS,the minimal crtIb promoter region is rather large and spans positions -59 and +120 relative to the transcription start, with a putative enhancer-like element between +30 and +120 (Martinez-Argudo et al., 1998). This could also reflect the need for accessory factors in regulating transcription from P,. Although CarD is also absolutely required for the activation of crtlb (Nicolis et al., 1994), the PI promoter region does not reveal a sequence that could be a CarD-binding site, nor is there any evidence yet for CarD binding to this region (Martinez-Argudo et al., 1998). Thus, while CarD is critical for the activation of PI, its regulatory role may be indirect. Mutations at ihfA also impair expression of crtlb, but this is overcome if carQ is expressed from a heterologous, ihfA-independent promoter (Moreno et al., 2001). Thus, the role of IHF in activating PI appears to be indirect and linked to PQRs activation and the consequent production of CarQ.

Transcriptional Regulation of the Structural Genes in the curl3 Operon Other than crtlb, all of the structural genes for carotenoid synthesis in M. xanthus are grouped at the carBcarA cluster. The carB locus was identified by Tn5 insertion mutations which generated a Car- phenotype

12. CAROTENOGENESIS IN M. XANTHUS (Martinez-Laborda et al., 1986; Balsalobre et al., 1987; Ruiz-Vizquez et al., 1993). Subsequent sequencing revealed the presence of six structural genes at this locus, and the enzymes they encode were deduced by analyzing predicted amino acid sequences (Botella et al., 1995) and characterizing the carotenoids accumulated on heterologous expression in E. coli (Cervantes, 2000; Iniesta, 2003). Gene expression analyses indicate that most or all of the genes in the carB cluster are transcribed from its light-inducible promoter, P,. The carA locus was originally defined by the point mutation carA 2 that resulted in constitutive, lightindependent expression of carotenoids, just as with mutations in the unlinked carR locus (Balsalobre et al., 1987; Martinez-Laborda and Murillo, 1989). However, the carAl mutant cells were orange and so less intensely pigmented than the deep red of cells with carR mutations, because the latter also activate crtlb (Martinez-Laborda and Murillo, 1989: Fontes et al., 1993).There is evidence for a light-independent promoter, PA,located within orf6, which would drive expression of the five genes that are further downstream and encompass the carA operon. Nevertheless, expression of the genes in the carA operon is also increased in the light presumably due to transcriptional readthrough from P, (Ruiz-Vizquez et al., 1993; Botella, 1996). The mechanism of light-induced expression at P, differs from those at PQRs and PI. Light enhances P, activity about 20-fold (monitored by the expression of reporter lacZ fusions [Balsalobre et al., 1987]), compared to 400-fold for PI (Fontes et al., 1993) or 80-fold for PqRs (Hodgson, 1993). Whereas maximal light activation of P, required cells to reach stationary phase, P, could be photoinduced at any time during the growth cycle. In contrast to PQRs and PI, the -35 region of P, perfectly matches the TTGACA of E. coli promoters dependent on the major vegetative u70-RNA polymerase holoenzyme but, as is often the case in GC-rich M. xanthus, the - 10 hexamer in P, (TACCTC) diverges considerably from the AT-rich consensus (TATAAT)in E. coli (Fig. 3) (Botella et al., 1995).Purified M . xanthus containing the major vegetative uA(MxRNAP) does indeed bind specifically to P, to form stable, open complexes capable of transcription in vitro (L6pez-Rubio et al., 2004). P,, but not PQRs or PI, is also affected by mutations a t two unlinked loci. One of these is c a d , the third gene of the carQRS operon, and the other is carA, the fourth gene of the carA operon. Absence of cars blocks activation of P, by light, while expressing cars from a lightindependent promoter strongly activates P, even in the dark (McGowan et al., 1993), indicating that Cars is a positive regulator of P,. The first known CUTSmutation

219

identified (carS1)was nonetheless a gain-of-function one that led to a light-independent, constitutive expression at P, (Balsalobre, 1989). It corresponded to a stop codon that produced CarS1, a truncated version of Cars lacking the last 25 amino acids (McGowan et al., 1993).As discussed later, the molecular basis of Cars action can rationalize this gain-of-function carS2 phenotype. Constitutive expression at P, was observed for the carAl mutation that originally defined the carA locus, which suggested that this locus houses elements for the negative regulation of P, in the dark (MartinezLaborda and Murillo, 1989). DNA sequence analysis of the carA2 clone revealed two closely linked nucleotide changes: one was a single nucleotide deletion at the 3’ end of orf9, which changes the last six amino acids of Orf9 and adds 76 residues at its C terminus; the second mutation was an A-to-T transversion at the fifth codon of carA that causes an I-to-F substitution (Botella et al., 1995). Nonpolar deletions in each of orf9, carA, and the downstream o r f l l showed that orf9 may play a role in positively regulating PQRs and thereby crtlb, while the negative regulator of P, is encoded by carA (Cervantes and Murillo, 2002). It was also shown that CarA, which does not participate in the light activation of PI or of PQRS, is a transcriptional repressor that binds specifically to the DNA region encompassing P,, while Cars interacts specifically with CarA and not DNA to derepress P, (Whitworth and Hodgson, 2001; Lopez-Rubio et al., 2002). CarA and Cars thus form a repressorantirepressor pair in regulating P,.

CarA Operator Design and Model for Regulation of P, The CarA operator design and the mechanism underlying the repression-antirepression switch of P, have been subjected to detailed molecular analysis (Fig. 5 ) (LopezRubio et al., 2004). The operator has a bipartite design and is made up of a high-affinity CarA-binding site (PI), which is a perfect interrupted palindrome located between positions -46 and -63 relative to the transcription start site, and a low-affinity one (pII)corresponding to an imperfect interrupted palindrome spanning positions -25 to -40 (Fig. 3). This is manifested in vitro by the stepwise binding of dimeric CarA, first to PI and then cooperatively to pII. Binding of CarA to pII, which overlaps partially with the -35 promoter region, effectively occludes a*-RNA polymerase from P, and thereby downregulates expression of the carB cluster. On the other hand, CarA bound to pII is efficiently dislodged by Cars, when present, to activate P,. The bipartite operator design thereby enables a rapid and effective response to light and provides the operative mechanism for the

REGULATORY MECHANISMS

220

f 2arS I )

LIGHT

CarA dimer

-35

e c=acsQ

PI

-10

car5operon

PI1

Figure 5 Model for the action of CarA and Cars in the regulation of the P, promoter. The dimeric CarA repressor is shown with its N- and C-terminal domains represented by the small and large spheres, respectively. In the dark (left panel), two CarA dimers bind via their Nterminal domains and in a cooperative fashion to the bipartite CarA operator. The two sites in the operator, palindromes PI and pII, are each shown as a pair of convergent arrows. Occupancy of pII by CarA blocks promoter access to the RNA polymerase holoenzyme (shown by the object labeled RNAP) leading to repression of curB. On exposure to light (right panel) CarS, shown by the dark ellipsoids, is produced, and its interaction with the N-terminal domain of CarA readily dismantles CarA-pII complexes. The RNA polymerase holoenzyme thereby gains access to the promoter, leading to the derepression of carB. repression of the carB operon by CarA and the derepression of P, by CarS. T h e CarA-Cars Repressor-Antirepressor Pair While Cars has no known sequence homologs, CarA has an N terminus which shows sequence similarity to the Nterminal, DNA-binding domains of bacterial MerR transcriptional factors (Botella et al., 1995). MerR factors act in different stress responses and, depending on the union of specific cofactors like heavy metals or drugs to their Cterminal domains, they repress or activate transcription by inducing DNA conformational changes (Brown et al., 2003). Their DNA-binding sites, in contrast to the CarA operator, fall entirely within a suboptimal 19-bp (not the usual 17-bp) spacer region that separates the -35 and -10 regions. As with MerR proteins, the CarA C-terminal domain was predicted to bind a specific cofactor by sequence analysis. However, the putative ligand, vitamin B,, or a related cobalamin derivative, is a novel one for a transcriptional cofactor (Cervantes and Murillo, 2002). Recent studies confirm the predicted structuralfunctional domain organization for CarA. Its 78-residue N-terminal segment is indeed a stable, autonomously folded unit that contains the binding determinants not only for the operator DNA but also for the antirepressor Cars (PCrez-Marin et al., 2004). That the domain adopts the winged-helix topology of MerR family DNAbinding domains has been confirmed by nuclear magnetic resonance. These studies, together with site-directed

mutagenesis, show that common structural elements in Car.A mediate specific interactions with DNA as well as with Cars (Navarro-AvilCset al., 2007). The findings also suggest that the CarA-binding site on Cars may mimic operator DNA in size, shape, and electrostatic complementarity, so that this could form the physical-structural basis for antirepression by CarS. Like reported DNA mimics, Cars is a highly acidic protein. Significantly, the truncated form of CarS, CarS1, is even more acidic and its greater affinity for CarA would explain the gainof-function carSl phenotype mentioned earlier (L6pezRubio et al., 2002). That the CarA C-terminal domain is an independently stable domain that binds vitamin B,, and mediates CarA dimerization has been shown experimentally (PCrezMarin et al., 2004; M. C. P6rez-Marin, S. Padmanabhan, M. C. Polanco, F. J. Murillo, and M. Elias-Arnanz, unpublished data). Moreover, B,, appears to be involved in the regulation of P, (Cervantes and Murillo, 2002). However, the findings that the N-terminal domain of CarA is capable of repressing P, in vivo when expressed at sufficiently high levels (Ptrez-Marin et al., 2004) and that mutating key H residues of the B,,-binding motif does not affect CarA function in vivo (Pkrez-Marin et al., unpublished), suggest that it may not be through CarA that B,, acts in regulating P,. In fact, detailed genetic analyses have indicated that it is O r f l l , the product of the gene directly downstream of CMA,that is involved in repressing P, in a B12-dependentmanner. O r f l l , which

12. CAROTENOGENESIS IN 211. XANTHUS

22 1

shares significant sequence similarity and domain organization with CarA (Botella et al., 1995; Cervantes and Murillo, 2002), would thus be yet another transcriptional factor in the gene regulation cascade for M. xanthus carotenogenesis. Its binding to the CarA operator and to Cars via the N-terminal domain has been experimentally confirmed, and the molecular details of how its activity is modulated by B,, are being worked out (Pkrez-Marin et al., unpublished data). The reasons for B,, recruitment in this light response is still an open question. A possible link to photoreception is suggested by the fact that at least one protein-bound form of BI2, cob(I)alamin, exhibits strong blue-light absorbance with a maximum near 400 nm (Jarrett et al., 1997), the same blue-light region where maximal photoinduction of carotenogenesis occurs. Whether this or some other mode of action underlies the role of B,, in M. xunthus carotenogenesis, however, remains to be experimentally examined.

THE CarD-CarG COMPLEX LINKS MULTICELLULAR DEVELOPMENT AND CAROTENOGENESIS The most studied phenomenon in M . xanthus is undoubtedly its striking ability to form multicellular fruiting bodies on starvation, a process that has served as a prokaryotic model for the study of cell-cell interactions and cellular differentiation (Kaiser, 2003, 2004). A genetic link between fruiting body development and the blue-light response in M. xanthus was first provided by the identification of gene carD. The carD2 mutation, a Tn5 insertion at curD originally isolated because it impaired light-induced carotenogenesis, was found to affect fruiting body formation as well. The mutant exhibited only weak signs of aggregation, as if blocked at an early stage of development (Nicolis et al., 1994).The same defective phenotype was observed for a strain bearing a complete in-frame deletion of curD

WT

(Fig. 6) (Cayuela et al., 2003). This developmental phenotype of carD mutants was not due to lack of carotenogenesis, since normal differentiation and development was observed for M . xanthus bearing other Car- mutations such as those in curB, crtlb, or carQ (Nicol6s et al., 1994). CarD was shown to be essential for activating the expression of genes involved in both the early and late stages of development mediated by the A-factor and C-factor intercellular signals, respectively (Nicolis et al., 1994). As mentioned earlier, CarG is a recently identified factor that works with CarD in the regulation of multicellular development and carotenogenesis in M. xanthus. Strains with in-frame deletions at carG are Car- and are blocked at an early stage of fruiting body formation, just as with carD mutant strains (Fig. 6). Like CarD, CarG appears to be required for expression of a set of A- and C-signaldependent developmental markers (Pefialver et al., 2006). Thus, CarD and CarG might be required in the production of the C- and A- signals or in the pathways that these signals mediate. Light and lack of nutrients, respectively, trigger carotenogenesis and multicellular development in M. xanthus, and both entail the transcriptional activation of specific genes. But of the carotenoid genes whose expression is stimulated by light, only that for crtlb is also dependent on starvation, the cue for activating the developmental process. Thus, two distinct environmental signals converge in their requirement for the CarD-CarG pair but diverge in the set of genes that are ultimately triggered.

LIGHT-REGULATED CAROTENOGENESIS IN OTHER BACTERIA Carotenoids are present in all photosynthetic and some nonphotosynthetic bacteria. Most prokaryotes known to produce carotenoids do so in a constitutive manner. However, besides M. xunthus, some other species

carDA

carGA

Figure 6 Developmental phenotype of the carD and carG mutants. Photographs were taken after 5-day incubation of 10-pl droplets of cells (1.25 X lo8 cells/ml) spotted on CF agar. The wild-type control strain is DK1622.

222 of nonphotosynthetic bacteria such as Mycobacterium marinum, Flavobacterium dehydrogenans, Brevibacterium sulfureum, and Streptomyces, have been reported to produce carotenoids in a light-dependent manner (Bramley and Mackenzie, 1988; Armstrong, 1997). As with algae, fungi, and plants, light is the most influential environmental agent in bacterial carotenoid synthesis. Here, we focus on those bacteria for which the effect of blue light on carotenogenesis has been best characterized at a genetic and molecular level, highlighting the parallels with M . xanthus. In anoxygenic photosynthetic bacteria, light and oxygen have long been known to control accumulation of carotenoids (Armstrong, 1997). The seven structural carotenogenic genes (crtA through crtF and crtl) in Rhodobacter capsulatus and Rhodobacter sphaeroides are clustered with those for bacteriochlorophyll and light-harvesting proteins. Carotenoids accumulate preferentially under low-oxygen and low-light conditions. Regulation includes repression by PpsWCrtJ, a soluble tetrameric protein with a helix-turn-helix DNA-binding motif that recognizes a conserved 18-bp interrupted palindrome sequence. Repressed genes or gene clusters always contain two binding sites, one located within the promoter region and another either straddling the -35 promoter region or lying 100 to 150 bp upstream of the transcription start site (Bauer et al., 2002; Movskin et al., 2005). The two sites and the requirement for cooperative binding between them for effective repression by PpsR are reminiscent of the operator design and mode of action of M . xanthus CarA. Moreover, at least in R. sphaeroides, PpsR-DNA binding is inhibited by its antirepressor AppA, a flavin-containing blue-light photoreceptor, which both mediates breakage of a disulfide bond in PpsR and forms a stable AppA-PpsR, complex (Bauer et al., 2002; Masuda and Bauer, 2002). In this case, blue-light excitation of AppA deters its ability to complex with PpsR,, thereby permitting it to bind DNA and to downregulate carotenogenesis under high-light conditions. In the nonphotosynthetic Streptomyces, several species have been found to synthesize carotenoids on illumination, but this ability is lost at a relatively high frequency by many strains (see Kato et al., 1995).Genetic studies on carotenogenesis in Streptomyces had been reported only for S. griseus and S. setonii (Kato et al., 1995; Schumann et al., 1996; Lee et al., 2001). In these, however, carotenogenesis occurs in a cryptic manner: a crt gene cluster is found, but the condition under which carotenoid synthesis takes place is unknown. Involvement of the stress response sigma factor CrtS has been suggested, though it is unclear how its expression or activation is controlled

REGULATORY MECHANISMS (Kato et al., 1995; Lee et al., 2001). The recent construction of a strain that did not produce two pigmented antibiotics revealed light-induced carotenogenesis in Streptomyces coelicolor A(2) and prompted its genetic analysis (Takano et al., 2005). The structural genes for carotenogenesis in S. coelicolor A(2)exist as two convergent operons, crtEIBV and crtYTU, whose expression is induced on illumination. Their photodependent transcription has been linked to at least two genes, litS and litR, located in two divergent operons that flank crtYTU: litRQ and litSAB. Genetic analysis suggested an essential role for litS in the activation of the crt genes by light. A complete deletion of litS impaired carotenoid synthesis, while its overexpression off a light-independent promoter led to constitutive carotenogenesis. Moreover, the activities of the photoinducible PcrtE, PcrtY, and Pli, promoters were completely abolished in the litS mutant. LitS, a 22-kDa protein, has 21% sequence identity to 211. xanthus ECF sigma factor CarQ. By in vitro runoff transcription analysis, LitS was indeed shown to be a (T factor necessary for transcriptional activation of PcrtE and PcrtY. However, specific transcriptional initiation could not be reproduced in vitro with the P,, DNA probe, suggesting that factors other than LitS and core RNA polymerase may be required in this case. As exemplified by the CarQ-CarR pair, many ECF sigma factors use membrane-bound antisigma factors to recognize the extracytoplasmic inducing cues. Of the two other genes in the litSAB operon, LitB shows weak similarity to RsrR, an S. coelicolor antisigma factor. However, inactivating LitB (or the putative lipoprotein LitA) did not affect carotenoid synthesis. LitR was originally proposed to work as an essential positive regulator in the light induction of Plits, since all photoinducible promoters were inactivated in a litR mutant background (Takano et al., 2005). However, adding purified LitR to the in vitro runoff reactions with the LitS-RNA polymerase holoenzyme did not result in production of specific transcripts from Plits. Nor did LitR, a 35-kDa protein with a MerR-like DNA-binding domain, show in vitro binding to the litS promoter region (which does contain an interrupted palindrome). Interestingly, sequence analysis suggested a domain organization for LitR (Takano et al., 2006a) that resembles that for M . xanthus CarA (Cervantes and Murillo, 2002): a predicted DNA-binding motif linked to a putative BIZbinding C-terminal domain. In their first study (Takano et al., ZOOS), a positive regulatory role was attributed to LitR. More recently, photoinduction of PcrtYon heterologous expression of litS and litR in S. griseus and constitutive expression from this promoter on deleting litR from the plasmid led to the conclusion that LitR, like CarA,

12. CAROTENOGENESIS IN 111. XANTHUS acts as a repressor in the dark, rather than as an activator (Takano et al., 2006a). Moreover, the authors speculated that B,, may play a role in the light-dependent regulation by LitR, as was proposed earlier for M. xanthus CarA (Cervantes and Murillo, 2002). Nonetheless, LitR binding to DNA and/or B,, remains to be demonstrated. Although in a divergent orientation, rather than convergent as in s. coelicolor A(2),the carotenogenic crtEIBV and crtYTUoperons are also found in S. avermitilis,which appears to produce carotenoids in a light-dependent manner (Takano et al., 2005). Furthermore, sequence conservation of the litQRS gene cluster in S. avermitilis could be indicative of a regulatory mechanism that is similar to the one described for S. coelicolor A(2). The increasing number of newly available microbial genome sequences continues to provide new insights into the existence and arrangement of crt and possible regulatory genes in different bacteria. Thus, until recently CarA and O r f l l in M. xanthus were the only known transcriptional regulators made up of a MerR-type DNA-binding domain linked to a vitamin B,,-binding domain. Together with the recently reported S . coelicolor and S . avermitilis LitR, microbial genome data reveal new examples of proteins with a domain organization similar to that of C a r N Orfl 1 in several other nonphotosynthetic bacteria such as Bdellovibrio bacteriovorus, Dechloromonas aromatica, Exiguobacterium sp., Magnetococcus sp., Nocardia farcinica, Pseudomonas putida, Ralstonia eutropha, and Thermus thermophilus. Of these, only T. thermophilus has been reported to produce carotenoids (Hoshino et al., 1993),possibly in a light-dependent manner (Takano et al., 2006b). The genome sequence of T. thermophilus reveals some noteworthy features with regard to carotenogenesis (Henne et al., 2004). First, the crt genes involved in the early steps of the biosynthetic pathway are found in the chromosome, while those involved in the subsequent steps are plasmid encoded (though not organized in a single cluster, they are found in close proximity to one another). The T. thermophilus CarA analog (TtCarA) gene is also located in the megaplasmid and lies immediately upstream of the phytoene synthase gene but is transcribed in the opposite sense. This suggests a link between TtCarA and carotenogenesis in T. thermophilus, perhaps as in M. xanthus. Interestingly, genes involved in the early steps of cobalamin biosynthesis are located in the chromosome, whereas those participating in the later steps of the pathway occur in the megaplasmid. This genomic organization mirrors that for the crt genes, and its significance, if any, is an open question. It is nonetheless tempting to speculate that such an arrangement might reflect a relationship between cobalamin synthesis, carotenogenesis, and CarA function.

223

CONCLUDING REMARKS Studies with M . xanthus continue to provide insights into the general principles underlying the complexity of the biosynthetic pathways and regulatory mechanisms involved in eubacterial carotenogenesis. They reveal how an ingenious circuitry for receiving the inducing signal and transmitting it meshes with the gene regulatory machinery. This machinery involves, among others, transcriptional factors with novel domain architectures, and a design of cis-regulatory elements that together lead to an exquisite control of gene expression. A rapid and an effective response to the environmental cues that stimulate the process is thereby accomplished. W e thank Francisco Murillo for continual support, and him as well as all past and present members of our group for their many contributions. Our current understanding of lightinduced carotenogenesis in M. xanthus also owes considerably t o work from David Hodgson and his group at the University of Warwick (United Kingdom). W e benefited from sequence data provided to us by The Institute for Genomic Research (http://www.tigr.org) for M. xanthus and S. aurantiaca; Rolf Miiller (Universitat des Saarlandes) for S. cellulosum; and the U.S. Department of Energy Joint Genome Institute (http:// www.jgi.doe.gov) for A. dehalogenans. We are grateful to the Ministerio de Educacidn y Ciencia (Spain) for uninterrupted research funding.

References Armstrong, G. A. 1997. Genetics of eubacterial carotenoid synthesis: a colourful tale. Annu. Rev. Microbiol. 51:629-659. Balsalobre, J. M. 1989. Induccidn por la luz de la expresi6n ge'nica y la carotenoginesis en Myxococcus xanthus. Ph.D. thesis. University of Murcia, Murcia, Spain. Balsalobre, J. M., R. M. Ruiz-Vazquez, and F. J. Murillo. 1987. Light induction of gene expression in Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 84:2359-2362. Bauer, C., S. Essen, L. R. Swem, D. L. Swem, and S. Masuda. 2002. Redox and light regulation of gene expression in photosynthetic prokaryotes. Philos. Trans. R. SOC. Lond. B 358:147-154. Botella, J. A., R. M. Ruiz-Vazquez, and F. J. Murillo. 1995. A cluster of structural and regulatory genes for light-induced carotenogenesis in Myscococcus xanthus. Eur. J. Biochem. 2 2 3 ~ 28-248. 3 Bramley, P. M., and A. Mackenzie. 1988. Regulation of carotenoid biosynthesis. Curr. Top. Cell. Regul. 29:291-343. Brown, N. L., J. V. Stoyanov, S. P. Kidd, and J. L. Hobman. 2003. The MerR family of transcriptional regulators. FEMS Microbiol. Rev. 27:145-163. Browning, D. F., D. E. Whitworth, and D. A. Hodgson. 2003. Light-induced carotenogenesis in Myxococcus xanthus: functional characterization of the ECF sigma factor CarQ and antisigma factor CarR. Mol. Microbiol. 48:237-251. Burchard, R. P., and M. Dworkin. 1966. Light-induced lysis and carotenogenesis in Myxococcus xanthus. J. Bacteriol. 9 1535-545.

224 Burchard, R. P., S. A. Gordon, and M. Dworkin. 1966. Action spectrum for the photolysis of Myxococcus xanthus. J. Bacteriol. 9 1:896-8 97. Burchard, R. P., and S. P. Hendricks. 1969. Action spectrum for carotenogenesis in Myxococcus xanthus. J. Bacteriol. 971665-1668. Cayuela, M. L., M. Elias-Arnanz, M. Peiialver-Mellado, S. Padmanabhan, and F. J. Murillo. 2003. The Stigmatella aurantiaca homolog of Myxococcus xanthus high-mobilitygroup A-type transcription factor CarD: insights into the functional modules of CarD and their distribution in bacteria. J. Bacteriol. 185:3527-3537. Cervantes, M. 2000. Accidn de genes estructurales y reguladores en la respuesta a la luz azul de Myxococcus xanthus. Ph.D. thesis. University of Murcia, Murcia, Spain. Cervantes, M., and F. J. Murillo. 2002. Role for vitamin B,, in light induction of gene expression in the bacterium Myxococcus xanthus. J. Bacteriol. 184:2215-2224. Diaz-Perales, A., V. Quesada, J. R. Peinado, A. P. Ugalde, J. Alvarez, M. F. SuLrez, F. X. Gomis-Ruth, and C. Lopez0 t h . 2005. Identification and characterization of human archaemetzincin-1 and -2, two novel members of a family of metalloproteases widely distributed in Archaea. J. Biol. Chem. 280:3 0367-3 0375. Fontes, M., L. Galbis-Martinez, and F. J. Murillo. 2003. A novel regulatory gene for light-induced carotenoid synthesis in the bacterium Myxococcus xanthus. Mol. Microbiol. 47~561-571. Fontes, M., R. M. Ruiz-Vazquez, and F. J. Murillo. 1993. Growth-phase dependence of the activation of a bacterial gene for carotenoid synthesis by blue light. E M B O J. 12~1265-1275. Frank, H. A,, and G. W. Brudvig. 2004. Redox functions of carotenoids in photosynthesis. Biochemistry 43536078615. Friedman, D. I. 1988. Integration host factor: a protein for all reasons. Cell 5554.5-554. Galbis-Martinez, L. 2005. Analisis genitico y molecular de un transductor de la seiial luminosa en la bacteria Myxococcus xanthus. Ph.D. thesis. University of Murcia, Murcia, Spain. Galbis-Martinez, M., M. Fontes, and F. J. Murillo. 2004. The high-mobility group A-type protein of the bacterium Myxococcus xanthus as a transcription factor for several distinct vegetative genes. Genetics 167:15 85-1 595. Goodwin, T. W. 1980. The Biochemistry of Carotenoids, vol. 1. Plants. Chapman and Hall, Ltd., London, United Kingdom. Gorham, H. C., S. J. McGowan, P. R. H. Robson, and D. A. Hodgson. 1996. Light-induced carotenogenesis in Myxococcus xanthus: light-dependent membrane sequestration of ECF sigma factor CarQ by anti-sigma factor CarR. Mol. Microbiol. 48:237-251. Henne, A., et al. 2004. The genome sequence of the extreme thermophile Thermus thermophilus. Nut. Biotechnol. 22: 547-553. Hodgson, D. A. 1993. Light-induced carotenogenesis in Myxococcus xanthus: genetic analysis of the carR region. Mol. Microbiol. 7:471-488. Hodgson, D. A., and A. E. Berry. 1998. Light regulation of carotenoid synthesis in Myxococcus xanthus, p. 185-211.

REGULATORY MECHANISMS In M. X. Caddick, S. Baumberg, D. A. Hodgson, and M. K. Phillips-Jones (ed.),Microbial Responses to Light and Time, Society for General Microbiology Symposium, vol. 56. Cambridge University Press, Cambridge, United Kingdom. Hodgson, D. A., and F. J. Murillo. 1993. Genetics of regulation and pathway of synthesis of carotenoids, p. 157-181. In M. Dworkin and D. Kaiser (ed.), Myxobacteria II. American Society for Microbiology, Washington, DC. Hoshino, T., R. Fujii, and T. Nakahara. 1993. Molecular cloning and sequence analysis of the crtB gene of Thermus thermophilus HB27, an extreme thermophile producing carotenoid pigments. Appl. Enuiron. Microbiol. 58:93-98. Hughes, K. T., and K. Mathee. 1998. The anti-sigma factors. Annu. Rev. Microbiol. 52:231-286. Iniesta, A. A. 2003. Expresidn de genes carotenogdnicos de la bacteria Myxococcus xanthus en Escherichia coli. Ph.D. thesis. University of Murcia, Murcia, Spain. Jarrett, J. T., C. W. Goulding, K. Fluhr, S. Huang, and R. G. Matthews. 1997. Purification and assay of cobalamin-dependent methionine synthase from Escherichia coli. Methods Enzymol. 28 1:196-2 13. Kaiser, D. 2003. Coupling cell movement to multicellular development in myxobacteria. Nut. Rev. Microbiol. 1:45-54. Kaiser, D. 2004. Signaling in myxobacteria. Annu. Rev. Microbiol. 58:75-98. Kato, F., T. Hino, A. Nakaji, M. Tanaka, and Y. Koyama. 1995. Carotenoid synthesis in Streptomyces setonii ISP539.5 is induced by the gene crtS, whose product is similar to a sigma factor. Mol. Gen. Genet. 247:387-390. Kleinig, H. 1972. Membranes from Myxococcus fuluus (Myxobacterales) containing carotenoid glucosides. Isolation and composition. Biochim. Biophys. Acta 274:489-498. Krubasik P., and G. Sandmann. 2000. Molecular evolution of lycopene cyclases involved in the formation of carotenoids with ionone end groups. Biochem. Soc. Trans. 28906-810. Lee, H. S., Y. Ohnishi, and S. Horinouchi. 2001. A sigma B-like factor responsible for carotenoid biosynthesis in Streptomyces griseus. J. Mol. Microbiol. Biotechnol. 3:95-101. Lonetto, M. A., K. L. Brown, K. E. Rudd, and M. J. Buttner. 1994. Analysis of the Streptomyces coelicolor sigE gene reveals the existence of a subfamily of eubacterial RNA polymerase u factors involved in the regulation of extracytoplasmic functions. Proc. Natl. Acad. Sci. USA 91:75737577. Lopez-Rubio, J. J., M. Elias-Arnanz, S. Padmanabhan, and F. J. Murillo. 2002. A repressor-antirepressor pair links two loci controlling light-induced carotenogenesis in Myxococcus xanthus. J. Biol. Chem. 2777262-7270. Lopez-Rubio, J. J., S. Padmanabhan, J. M. Lazaro, M. Salas, F. J. Murillo, and M. Elias-Arnanz. 2004. Operator design and mechanism for CarA repressor-mediated down-regulation of the photoinducible carB operon in Myxococcus xanthus. J. Biol. Chem. 279:28945-28953. Martinez-Argudo, I., R. M. Ruiz-Vazquez, and F. J. Murillo. 1998. The structure of an ECF-u-dependent, light-inducible promoter from the bacterium Myxococcus xanthus. Mol. Microbiol. 302383-893. Martinez-Laborda, A., J. M. Balsalobre, M. Fontes, and F. J. Murillo. 1990. Accumulation of carotenoids in structural

12. CAROTENOGENESIS IN M.

XANTHUS

and regulatory mutants of the bacterium Myxococcus xanthus. Mol. Gen. Genet. 223:205-210. Martinez-Laborda, A., M. Elias, R. Ruiz-Vazquez, and F. J. Murillo. 1986. Insertion of Tn5 linked mutations affecting carotenoid synthesis in Myxococcus xanthus. Mol. Gen. Genet. 205:107-114. Martinez-Laborda, A., and F. J. Murillo. 1989. Genic and allelic interactions in the carotenogenic response of Myxococcus xanthus to blue light. Genetics 122:48 1-490. Masuda, S., and C. E. Bauer. 2002. AppA is a blue light photoreceptor that antirepresses photosynthesis gene expression in Rhodobacter sphaeroides. Cell 110:613-623. McGowan, S. J., H. C. Gorham, and D. A. Hodgson. 1993. Light-induced carotenogenesis in Myxococcus xanthus: DNA sequence analysis of the carR region. Mol. Microbiol. 10:713-735. Meiser, P., H. B. Bode, and R. Muller. 2006. The unique DKxanthene secondary metabolite family from the myxobacterium Myxococcus xanthus is required for developmental sporulation. Proc. Natl. Acad. Sci. USA 103:19128-19133. Moraleda-Muiioz, A., J. PCrez, M. Fontes, F. J. Murillo, and J. Muiioz-Dorado. 2005. Copper induction of carotenoid synthesis in the bacterium Myxococcus xanthus. Mol. Microbiol. 56:1159-1168. Moreno, A. J., M. Fontes, and E J. Murillo. 2001. ihfA gene of the bacterium Myxococcus xanthus and its role in activation of carotenoid genes by blue light. J. Bacteriol. 183557569. Moskvin, 0.V., L. Gomelsky, and M. Gomelsky. 2005. Transcriptome analysis of the Rhodobacter sphaeroides PpsR regulon: PpsR as a master regulator of photosystem development. J. Bacteriol. 187:2148-2156. Nash, H. A. 1996. The HU and IHF proteins: accessory factors for complex DNA assemblies, p. 149-179. In E. C. C. Lin and A. S. Lynch (ed.), Regulation of Gene Expression in Escherichia coli. R. G. Landes Company, Austin, TX. Navarro-AvilCs, G., M. A. JimCnez, M. C. PCrez-Marin, C. Gonzalez, M. Rico, F. J. Murillo, M. Elias-Amanz, and S. Padmanabhan. 2007. Structural basis for operator and antirepressor recognition by Myxococcus xanthus CarA repressor. Mol. Microbiol. 63:980-994. Nicolas, F. J., M. L. Cayuela, I. M. Martinez-Argudo, R. M. Ruiz-Vazquez, and F. J. Murillo. 1996. High mobility group I(Y)-like DNA-binding domains on a bacterial transcription factor. Proc. Natl. Acad. Sci. USA 93:6881-6885. Nicolas, F. J., R. M. Ruiz-Vazquez, and F. J. Murillo. 1994. A genetic link between light response and multicellular development in the bacterium Myxococcus xanthus. Genes Dev. 8~2375-2387. Padmanabhan, S., M. Elias-Arnanz, E. Carpio, P. Aparicio, and F. J. Murillo. 2001. Domain architecture of a high mobility group A-type bacterial transcriptional factor. J. Biol. Chem. 276:4156641575.

225 Peiialver-Mellado, M., F. Garcia-Heras, S. Padmanabhan, D. Garcia-Moreno, F. J. Murillo, and M. Elias-Arnanz. 2006. Recruitment of a novel zinc-bound transcriptional factor by a bacterial HMGA-type protein is required for regulating multiple processes in Myxococcus xanthus. Mol. Microbiol. 61:9 10-926. PCrez-Marin, M. C., J. J. Lopez-Rubio, F. J. Murillo, M. EliasArnanz, and S. Padmanabhan. 2004. The N-terminus of Myxococcus xanthus CarA repressor is an autonomously folding domain that mediates physical and functional interactions with both operator DNA and antirepressor protein. J. Biol. Chem. 279:33093-33103. Reeves, R. 2001. Molecular biology of HMGA proteins: hubs of nuclear function. Gene 27763-81. Reeves, R. 2003. HMGA proteins: flexibility finds a nuclear niche. Biochem. Cell Biol. 81:185-195. Reichenbach, H., and H. Kleinig. 1984. Pigments of myxobacteria, p. 128-137. In E. Rosenberg (ed.), Myxobacteria: Development and Cell Interactions. Springer-Verlag, New York, NY. Robson, P. R. H. 1992. Towards a Mechanism of Carotenogenesis in Myxococcus xanthus. Ph.D. thesis. University of Warwick, Coventry, England. Ruiz-Vazquez, R. M., M. Fontes, and F. J. Murillo. 1993. Clustering and co-ordinated activation of carotenoid genes in Myxococcus xanthus by blue light. Mol. Microbiol. 10:25-34. Schumann, G., H. Nurnberger, G. Sandmann, and H. Krugel. 1996. Activation and analysis of cryptic crt genes for carotenoid biosynthesis from Streptomyces griseus. Mol. Gen. Genet. 252:65 8-666. Takano, H., D. Asker, T. Beppu, and K. Ueda. 2006a. Genetic control for light-induced carotenoid production in nonphototrophic bacteria. J. Ind. Microbiol. Biotechnol. 33: 88-93. Takano, H., T. Beppu, and K. Ueda. 2006b. The CarMLitRfamily transcriptional regulator: its possible role as a photosensor and wide distribution in non-phototrophic bacteria. Biosci. Biotechnol. Biochem. 70:2320-2324. Takano, H., S. Obitsu, T. Beppu, and K. Ueda. 2005. Lightinduced carotenogenesis in Streptomyces coelicolor A3(2): identification of an extracytoplasmic function sigma factor that directs photodependent transcription of the carotenoid biosynthesis gene cluster. J. Bacteriol. 187:1825-1832. Takayama, F., T. Egashira, and Y. Yamanaka. 2001. Singlet oxygen generation from phosphatidylcholine hydroperoxide in the presence of copper. Life Sci. 68:1807-1815. Whitworth, D. E., S. J. Bryan, A. E. Berry, S. J. McGowan, and D. A. Hodgson. 2004. Genetic dissection of the lightinducible carQRS promoter region of Myxococcus xanthus. J. Bacteriol. 186:7836-7846. Whitworth, D. E., and D. A. Hodgson. 2001. Light-induced carotenogenesis in Myxococcus xanthus: evidence that Cars acts as an anti-repressor of CarA. Mol. Microbiol. 425309419.

Structure and

Myxobacteria: Multicellularity and Differentiation Edited by David E. Whitworth 0 2008 ASM Press, Washington, D.C.

Zhaomin Yang, Xue-yan Duan, Mehdi Esmaeiliyan, Heidi B. Kaplan

Composition, Structure, and Function of the Myxococcus xanthus Cell Envelope

The cell envelope functions as the boundary and interaction surface between the bacterial cell and its environment. All cell-cell interactions and environmental responses are sensed and transduced through the cell envelope. A typical gram-negative bacterium cell envelope consists of the inner membrane, the periplasmic region containing the peptidoglycan (PG) and soluble proteins, the outer membrane, the lipopolysaccharide (LPS), and in some cases an extracellular matrix (ECM) of capsular exopolysaccharide (EPS) and proteins (Raetz and Whitfield, 2002). The myxobacteria are gram-negative soil bacteria that exhibit a variety of social behaviors during all stages of their life cycle. Growing cells exhibit at least two group behaviors: cooperative feeding and social motility. Cells that are starved at high density on a solid surface initiate the complex multicellular behavioral response of fruiting body development in which approximately 100,000 cells aggregate into haystack-shaped mounds and differentiate into environmentally resistant spherical myxospores. Coordination of these social behaviors requires the exchange of a number of extracellular signals that

13

stimulate signal transduction pathways. Since Myxococcus xunthus serves as the premier model bacterium for the study of social behaviors, the investigation of its cell envelope through which the signals are sensed and transduced has been an active research area. This chapter covers in detail the polysaccharide-containing components of the M. xanthus cell envelope including the PG, LPS, ECM, and EPS.

PEPTIDOGLYCAN The cell wall PG forms the network of glycans and amino peptides that determines cell shape, protects the cell from the internal turgor pressure applied by the cytoplasmic contents, and plays a key role in cell division. The PG subunits (muropeptides)are synthesized in the cytoplasm and transported to the outer surface of the cytoplasmic membrane, where they are polymerized by transglycosylases that link the glycan moieties and transpeptidases that cross-link the amino peptides (Atrich et al., 1999). Specifically, PG consists of glycan strands of alternating p-1,4-linked N-acetylglucosamine (GlcNAc) and

Zhaomin Yang, Department of Biology, Virginia Polytechnic and State University, Blacksburg, VA 24060. Xue-yan Duan, Department of Molecular and Cellular Biology, Baylor College of Medicine, Houston, TX 77030. Mehdi Esmaeiliyan, Department of Natural Sciences, University of Houston/Downtown, Houston, TX 77002. Heidi B. Kaplan, Department of Microbiology and Molecular Genetics, University of Texas Medical School, Houston, TX 77030.

229

230 N-acetylmuramic acid (MurNAc) disaccharides that are cross-linked by short peptides. The structure of the Escherichia coli PG is typical of gram-negative bacteria. The PG monomer consists of the two amino sugars GlcNAc and MurNAc, with a 1,6 anhydro-N-acetylmuramic acid end (Vollmer and Holtje, 2004). The pentapeptide [L-ala-D-glu-(y)-m-dap-D-ala-D-ala] is attached to the MurNAc with cross-links that are formed between the D-alanine (ala) at position 4 of one stem peptide and mdiaminopimelic acid (dap) at position 3 of a second stem peptide of a neighboring glycan strand (Vollmer and Holtje, 2004). There has been no comprehensive analysis of the M . xanthus PG structure in more than 3 decades. White et al. (1968) found that the murein components of M. xanthus PG were similar to those of other gram-negative bacteria. The overall composition of the PG was 1.0 glutamic acid (glu), 1.0 dap, 1.7 ala, 0.75 GlcNAc, and 0.75 MurNAc. This analysis of the M. xanthus murein components revealed two additional findings. First, the PG was associated with substantial amounts of glycine, serine, and glucose. Second, the vegetative cell wall PG was suggested to be discontinuous in that whole sacculi were not isolated and trypsin and sodium dodecyl sulfate were able to completely disassociate the PG (White et al., 1968). Since the shape of the bacterial cell is dictated by the PG, it would be expected that during differentiation the M . xanthus cells would remodel their PG as they change from rod-shaped vegetative cells to spherical spores. It is unclear if this transformation involves de novo synthesis and/or breakage and restructuring of the cross-links between the muropeptide polymers (O’Connor and Zusman, 1999). White et al. (1968)and Johnson and White (1972) compared the PG of vegetative cells to starvationindependent spores (often referred to as glycerol spores). Although the overall composition and proportion by weight (0.9%) of both cell types were similar, there appeared to be an increase in muropeptide cross-linking in the spherical spores. Interestingly, O’Connor and Zusman (1997)noted that M . xanthus P-lactamase activity was induced by the same compounds that induce starvation-independent spore formation. These compounds are cell-wall-damaging agents, including 0.5 M glycerol, lysozyme, p-lactam antibiotics, D-amino acids, glycine, dimethyl sulfoxide, D-cycloserine, phosphomycin, phenyl ethanol, and ethylene glycol (O’Connor and Zusman, 1997; Dworkin and Gibson, 1964). This connection prompted them to propose that these two processes share common pathway components. It is likely that muropeptides function as the intermediate signal, since Jacobs et al. (1994) found that

STRUCTUREAND METABOLISM P-lactamase activity in E. coli can be controlled by alternations in the cytoplasmic levels of muropeptides (Park, 1995). O’Connor and Zusman (1999) also showed that P-lactamase activity is part of the normal developmental program, as an increase in its activity can be detected as soon as the developmental program is initiated. Other evidence suggests that external PG components have a direct role in the developmental program of M . xanthus. Shimkets and Kaiser (1982a, 1982b)showed that exogenously added PG and its components, specifically a mixture of GlcNAc, MurNAc, dap, and ala at 2.5 mM each, can initiate rippling, which is the organized movement of cells during early development, and can rescue sporulation of a csgA mutant. The mechanism of this rescue is unknown; however, Janssen and Dworkin (1985) showed that various sugars including glucosamine and mannosamine could also rescue sporulation of a csgA mutant. Shimkets and Kaiser (1982a, 198213) suggested that the lysis of cells during development releases a necessary amount of PG and other components required for the developmental process. O’Connor and Zusman (1999) noticed that Plactamase induction seems to influence the course of development, as the addition of P-lactamase inducers expedites the onset of aggregation and sporulation in a developing population of cells. They proposed that the induction of P-lactamase is likely to play a role in aggregation and in the restructuring of PG that occurs during differentiation into spores. However, it should also be considered that P-lactamase induction might reflect an alteration in the balance of cell wall synthesis and degradation that is anticipated to accompany the entrance into development as the cells arrest their growth. This would be expected to lead to the accumulation of muropeptides that would induce p-lactamase and serve as an indicator of cell wall integrity. A number of genes would .be predicted to be directly or indirectly expressed in response to a muropeptide signal. One group would be the early developmental genes controlled by the Che3 chemosensory system due to the activity in this pathway of the CrdB regulator that is predicted to have a PG-interaction domain (Kirby and Zusman, 2003). Other potential responsive genes would be those expressed early in development, such as L2444.5, which is regulated by an extracytoplasmic function (ECF) sigma factor-anti-ECF sigma factor pathway that senses envelope stress (Rivera, 2002). It is anticipated that further investigations of the biosynthesis and assembly of M . xanthus PG and its role in developmental signaling will lead to important insights into the structure of the cell envelope and the control of the developmental program.

13. M.XANTHUS CELLENVELOPE

PERIPLASMIC PROTEINS In addition to the cell-wall peptidoglycan and the enzymes involved in its biogenesis, the periplasmic space of gramnegative bacteria contains many enzymes and proteins of structural and functional importance. Among them are lipoproteins, hydrolytic enzymes, binding proteins important for nutrient acquisition, folding factors, and chaperons, and proteins that perform signaling/sensory functions and energy metabolism. Readers are directed to recent reviews on these topics (Behrens, 2003; Bos and Tommassen, 2004; Dwyer and Hellinga, 2004; Lomovskaya and Totrov, 2005; Mogensen and Otzen, 2005; Schlieker et al., 2004). Although little is known about the M. xanthus periplasm specifically, there are a few developmentally relevant periplasmic proteins including the glucose inhibition division (GidA) protein (White et al., 2001) and the periplasmic heat shock proteins (HSPs) (Nelson and Killeen, 1986). M. xanthus GidA, a flavin adenine nucleotide binding protein, was localized to the periplasm by cell fractionation and immunoelectron microscopy (White et al., 2001). M. xanthus gidA mutants appear to be unstable but are proficient in development. Stable derivatives, designated gidA*, arise after passage of the original mutants on solid media. The gidA* derivatives are defective in development and produce small heatstable and protease-resistant extracellular molecules that inhibit the development of wild-type cells. Intriguingly, the disruption of aglU, which maps immediately downstream of, and is transcribed with, gidA, eliminates the presence of the gzdA* developmentally inhibitory molecules. In contrast to the aglU mutants that are defective in adventurous (A)-motility, the gidA mutants are not (White and Hartzell, 2000; White et al., 2001). In other bacteria, GidA has been shown to be important for virulence and may function as a global regulator of gene expression (Kinscherf and Willis, 2002; Sha et al., 2004). White et al. (2001) suggest that GidA may play an important role in sensing or mediating redox-related phenomena and possibly couple replication and growth with cell division and differentiation. Heat-shocked developing M. xanthus cells produce at least two periplasmic HSPs that are not made by vegetative cells (Nelson and Killeen, 1986). The identities and the functions of these periplasmic HSPs remain to be determined; however, heat shock treatment prior to starvation or glycerol addition accelerates myxospore formation (Killeen and Nelson, 1988). Only one of the two HSPs is produced by heat-shocked starvationindependent spores (Nelson and Killeen, 1986), supporting the differences between starvation-independent and

23 1 starvation-dependent spores (Downard and Zusman, 1985; Shimkets and Seale, 1975).

IS THERE A PERIPLASMIC MOTILITY APPARATUS? Are there M. xanthus proteins in the periplasm that are involved in gliding motility? Chain-like aggregates or strands have been observed and isolated from the M. xanthus periplasm (Freese et al., 1997; Lunsdorf and Reichenbach, 1989). The isolated strands appear to be composed of ring-like and centrally elongated elements (Fig. l a ) . It is proposed that these strands align and form ribbon-like structures (Fig. 1b) with a periodicity visible by electron microscopy (EM). It was further suggested that these periplasmic structures could be part of the gliding machinery that may utilize membrane potential to power adventurous (A) gliding motility of myxobacteria (Freese et al., 1997; Lunsdorf and Schairer, 2001). However, it is important to note that there is no direct evidence linking these structures to gliding and the identity of the structural components remains unknown.

OUTER MEMBRANE PROTEINS Among the outer membrane proteins of M. xanthus two lipoproteins, CglB and Tgl, are important for A and social (S) gliding motility, respectively (see chapter 6). The motility defects of both cglB and tgl mutants can be rescued by direct contact with cells that produce wildtype CglB and Tgl, respectively (Hodgkin and Kaiser, 1979a, 197910; Rodriguez and Spormann, 1999; Wall et al., 1998).The underlying mechanism for the stimulation of motility by physical contact, in the case of Tgl, is the transfer of this lipoprotein from Tgl+ to Tgl- cells (Nudleman et al., 2005, 2006). It remains to be determined whether CglB and Tgl are anchored to the outer or inner leaflets of the outer membrane. The mechanism of lipoprotein transfer also remains unknown (Bayan et al., 2006). MBHA is a developmentally regulated lectin that accumulates to its highest level during the aggregation phase of fruiting body formation (Cumsky and Zusman, 1979). Although the bulk of MBHA appears to be cell surface associated, as it can be easily washed off cells, osmotic shock treatment of M. xanthus cells reveals that 20% of the MBHA protein is localized within the periplasm (Nelson et al., 1981). At least two factors regulate MBHA accumulation. First, m b h A transcription, which is 0-54 dependent, increases during development (Romeo and Zusman, 1991). Second, the stability of the m b h A mRNA is increased during development (Romeo

STRUCTUREAND METABOLISM

232

and Zusman, 1992). It is suspected that MBHA plays a role in M. xanthus cell cohesion (Romeo and Zusman, 1987),which is known to be important for fruiting body development (Shimkets, 1986a). Consistent with a role in cell cohesion, MBHA has four highly conserved domains that are predicted to form a multivalent structure (Romeo et al., 1986). MBHA-deficient strains are delayed in development but are otherwise able to aggregate and sporulate (Romeo and Zusman, 1987). It was suggested that there might be two systems for cell cohesion during development, of which only one requires magnesium (Romeo and Zusman, 1987). MBHA is predicted to function in a magnesium-independent cell cohesion system, since the developmental defects of mbhA mutants are more noticeable in the absence of magnesium (Romeo and Zusman, 1987).

LPS In gram-negative bacteria, the lipid component of the outer layer of the cell envelope is composed of LPS, which forms a selective barrier between the environment and the periplasmic region of the cell. The M. xanthus LPS is similar in general structure to the LPS of other gramnegative bacteria (Fink and Zissler, 1989).Each LPS molecule is composed of three parts: a hydrophobic lipid A,

which comprises the lipid portion of the outer leaflet of the outer membrane; a covalently attached nonrepeating core oligosaccharide region; and a distal repeating polysaccharide termed 0 antigen (Raetz and Whitfield, 2002). This general structure of the M. xanthus LPS was revealed by studies of LPS biosynthesis mutants isolated in genetic screens using monoclonal antibodies (MAbs) generated against the cell surface of M. xanthus cells (Gill et al., 1985; Gill and Dworkin, 1986). One set of MAbs (2600, 1733, 1514, 1412, and 783) was determined to recognize LPS 0 antigen based on their reactivity to the ladder of bands of crude LPS preparations separated on polyacrylamide gels (Fink and Zissler, 1989). Another MAb, 2254, recognizes LPS core, as its reactivity is confined to the lowest band on these gels (Fink and Zissler, 1989). Five Tn.5 transposon mutants were identified, which are nonreactive to MAbs that recognize the LPS 0 antigen, suggesting that the insertion mutations altered the biosynthesis or assembly of the LPS 0 antigen (Fink et al., 1989).When analyzed in detail, these mutants were determined to have a number of phenotypes including a defect in S-motility, whereas A-motility remained unaffected (Bowden and Kaplan, 1998). Kaplan et al. (1991) identified the sasA locus using UV mutagenesis in a genetic suppressor screen designed to identify elements in a pathway linking extracellular A

Figure 1 Chain-like strand in the M. xantbus periplasm and the structural model. (a) Single strand fragment associated with a membrane fragment (MF).The ring elements (arrowheads) stand oblique to the plane of the carbon support. Inset: ring elements (circle) and central elongated elements (dots) are shown at higher magnification; arrow indicates the spoke-like central mass. Bar, 50 nm (inset, 30 nm). (b). Three-dimensional model of the location of strands within the cell wall of M. xanthus. These strands are proposed to be inserted between the flexible peptidoglycan sacculus and the outer membrane. RE, ring element, framed; EE, elongated element; CM, cytoplasmic membrane; OM, outer membrane; cema, central mass; pm, peripheral mass. Adapted from Freese et al., 1997.

13. 211. XANTHUS CELLENVELOPE signal to its responsive gene, 4521. One of the sasA alleles mapped to the wzm gene (formally rfbA).The wzm gene and its downstream gene wzt (formally rfaB) encode an ABC transporter, which is predicated to be required for the export of LPS 0 antigen into the periplasm (Guo et al. 1996). Recently, a similar genetic screen with the newly developed mini-Himarl transposon identified many more LPS biosynthesis genes, which encode a majority of the M. xanthus sugar biosynthesis enzymes (X.-Y. Duan, M. Esmaeiliyan, and H. B. Kaplan, unpublished data). Youderian and Hartzell (2006) conducted an extensive screen for S-motility mutants with the magellan-4 transposon and also identified many genes involved in LPS biosynthesis. Interestingly, all of the LPS mutants identified by our two laboratories mapped to three loci: two LPS O-antigen loci and one LPS core locus. All the LPS mutants that have been identified have developmental defects. They form defective fruiting bodies and sporulate at reduced efficiencies (Fink et al., 1989; Guo et al., 1996; Bowden and Kaplan, 1998; Youderian and Hartzell, 2006; Duan et al., unpublished). Further analysis of some LPS mutants showed that they are defective in S-motility,whereas they are wild type for A-motility (Bowden and Kaplan, 1998; Duan et al., unpublished). These data support the concept that different mechanisms control A- and S-motilityand indicate that S-motilityplays a major role in M. xanthus fruiting body formation. The carbohydrate composition of wild-type LPS consists of glucose, mannose, rhamnose, arabinose, xylose, galactosamine, glucosamine, KDO (2-keto-3-deoxyoctulosonic acid), and 3-O-methylpentose and 6-O-methylgalactosamine (Ashton, 1993). Interestingly, the 111. xanthus EPS is proposed to contain five of these monosaccharides: galactose, glucosamine, glucose, rhamnose, and xylose (Behmlander and Dworkin, 1994b). It is possible that the LPS preparations analyzed were contaminated with EPS, which would also partition to the aqueous phase with the LPS hot-phenol extraction method. Currently comprehensive analyses of both the LPS and EPS structures are being pursued (H. B. Kaplan and W. Shi, personal communication). Genomic and genetic analyses to identify all of the genes required for M. xanthus LPS biosynthesis and assembly are certain to be pursued in the future. In addition, a thorough biochemical analysis of the structures of mutant M. xanthus LPS molecules and the activities of the biosynthetic enzymes will be critical to elucidate the pathways involved in the production of the individual sugars and the assembly of the LPS. Furthermore, an analysis of the mechanisms underlying the increase in expression of early developmental genes, such as 4521, in the absence of LPS 0 antigen should reveal interesting

233 signaling pathways. It is likely that the LPS-dependent effects on 4522 developmental expression, which is controlled by the SasS/R/N histidine kinase three-component system (Yang and Kaplan, 1997; Xu and Kaplan, 1998; Guo et al., 2000), are responsive to envelope stress.

ECM STRUCTURE AND FUNCTION M . xanthus cells are covered by an ECM comprised of approximately equal amounts by weight of protein and polysaccharide (Kim et al., 1999; Merroun et al., 2003; Behmlander and Dworkin, 1994b), which was previously referred to as fibrils (Fig. 2). The word “fibril” has been used to describe the apparent filamentous fibers on the cell surface of M. xanthus cells observed by scanning EM (SEM) (Dworkin, 1999) (Fig. 2C and D). The fiber-like structures observed using SEM are most likely the result of dehydration of the polysaccharide during sample preparation. The polysaccharide component of the ECM is now referred to as EPS (Lancer0 et al., 2004; Lu et al., 2005; Xu et al., 2005; Yang and Li, 2005). M . xunthus ECM is required for cellular cohesion, Smotility, and fruiting body morphogenesis. Early genetic studies indicated that dsp mutants, which grow dispersed and are defectivein cellular cohesion, lack S-motilityand are unable to form fruiting bodies (Shimkets, 1986a, 198610). It was established later that dsp mutants are altered in their surface properties and lack ECM when examined by EM (Arnold and Shimkets, 198810; Behmlander and Dworkin, 1991). The correlation between the presence of ECM and cellular cohesion was also demonstrated using the diazo dye Congo red that can disrupt the formation of ECM and inhibited both cellular cohesion and fruiting body development (Arnold and Shimkets, 1988a, 198813).These observations provided the evidence that cohesive forces may be necessary to maintain the integrity of cell groups, which is important for S-motility and fruiting body formation. The ECM function in S-motility has become clearer in recent years. Yang et al. (2000) suggested that the ECM provides more than a cohesion force for S-motility (Yang et al., 2000). It was postulated that the ECM is involved in the activation of the S-motility motor in a cell-proximity-dependent manner (Yang et al., 2000). Li et al. (2003)suggested that EPS provides the anchor for the retraction of type IV pili (Tfp),which are now widely accepted as the motor for M. xanthus S-motility and bacterial twitching motility (see chapter 6 ) . In essence, EPS from neighboring cells may anchor Tfp at the distal end and this attachment or anchoring may in turn activate the retraction of Tfp at the base. It has been suggested that the signal for tethering or attachment at the distal end may be transmitted to the cell body mechanically

STRUCTUREAND METABOLISM

234 through structural restraints or shifts (Li et al., 2003; Black et al., 2006).Interestingly, EPS may not be the only material that can trigger or activate the S-motility motor. An increase in the viscosity of the medium by the addition of methylcellulose can stimulate the S-motility-dependent movement of individual cells (Sun et al., 2000). The EPS component of M. xanthus ECM is necessary for the structural integrity of the ECM (Behmlander and Dworkin, 199413). Whereas digestion of carbohydrate by periodic acid essentially destroys the observed ECM structure, EM observations indicate that protein removal by protease does not significantly alter the structure of isolated ECM. Furthermore, the ECM proteins are not required for M. xanthus S-motility as treatment with

proteases does not appreciably affect the activity of ECM in causing Tfp to disappear from cell surfaces (Li et al., 2003), and mutations in fibA, which encodes the only known ECM protein, result in no obvious defects in ECM structure and S-motility (Kearns et al., 2002). The ECM may be considered to be a layer of EPS associated with nonspecific outer membrane proteins.

REGULATION OF EPS PRODUCTION The production of EPS is a highly regulated process. Early work suggested that cellular cohesion and surface properties are likely regulated by cell density (Arnold and Shimkets, 1988b; Behmlander and Dworkin,

Figure 2 ECM visualized by different EM techniques. (A and B) Transmission EM micrographs prepared either by spray-freezing freeze substitution (A) or by staining with lanthanum and fixation with glutaraldehyde (B). (C and D) SEM images at different magnifications. Samples in panels C and D were fixed with glutaraldehyde and coated with platinum. Fig. 2A from Kim et al., 1999; Fig. 2B from Merroun et al., 2003; Fig. 2C and D from Behmlander and Dworkin, 1991.

13. 111. XANTHUS CELLENVELOPE 1991), and more recent work indicates a nutritional dependency of the two different motility systems (Hillesland and Velicer, 2005). Genetic evidence provides strong support for EPS regulation in M. xanthus. The dsp mutants were the first mutants known to display defects in cellular cohesion (Arnold and Shimkets, 1988b). Another group of mutants, the sticky (stk) mutants, show an increased ability to agglutinate and a strong tendency to adhere to surfaces (Dana and Shimkets, 1993). Both groups show aberrant ECM production with dsp mutants producing no detectable ECM and stk mutants overproducing ECM (Arnold and Shimkets, 1988b; Dana and Shimkets, 1993). In addition, mgl, tgl, and sgl mutants, which are all defective in S-motility, were found to have reduced levels of ECM by dye binding and/or EM (Dana and Shimkets, 1993). ECM biosynthesis and assembly are regulated by the dif locus, which was identified by mutants defective in development and encodes a set of proteins that are homologs of the well-studied chemotaxis proteins of the enteric bacteria and Bacillus subtilis (Black and Yang, 2004; Yang et al., 1998b) (see chapter 8). DifA is a homolog of the methyl-accepting chemotaxis proteins, DifC is a homolog of the coupling protein Chew, DifD is a homolog of the response regulator CheY, DifE is a homolog of the histidine kinase CheA, and DifG is a homolog of the protein phosphatase CheC. The Dif proteins both positively and negatively regulate EPS production. Mutations in difA, difC, and difE virtually eliminate EPS production, whereas those in difD and difG result in EPS overproduction (Bellenger et al., 2002; Black and Yang, 2004; Xu et al., 2005; Yang et al., 1998b; Yang et al., 2000). Genetic epistasis tests indicate that DifD does not function downstream of DifE in the regulation of EPS (Black et al., 2006). Based on these observations, a model is proposed for the regulation of EPS by a pathway consisting of Dif proteins (Fig. 3) (Black et al., 2006). In this model, DifA, DifC, and DifE are proposed to form a signaling complex anchored to the membrane by the MCP homolog DifA. This is analogous to the classical chemotaxis pathway where MCPs, Chew, and CheA form a membrane signaling complex (Webre et al., 2003). It is further proposed that stimulation of the DifA-DifC and DifE complex leads to increased DifE kinase activity. DifE would modulate EPS production by a phosphorylation-dependent mechanism through components yet to be identified. DifD is proposed as a phosphate sink and would be analogous in function to the CheY2 of Rhizobium rneliloti chemotaxis (Sourjik and Schmitt, 1996, 1998). DifG is proposed to accelerate the dephosphorylation of DifD

235 phosphate (DifD-P) (Black and Yang, 2004). Moreover, DifG must somehow inhibit the central pathway independently of DifD because mutations in d i p and difG show certain additive effects. The formation of a ternary complex by DifA, DifC, and DifE is supported by yeast two- and three-hybrid experiments (Yang and Li, 2005), although the proposed phosphorylation reactions between the Dif proteins have not been biochemically demonstrated. It was shown recently that signal perception by the Dif pathway involves Tfp (Black et al., 2006). First, Tfp was found to be required for EPS production in M. xanthus. Second, Dif proteins function downstream of Tfp as demonstrated by genetic epistasis tests. Finally, Tfp do not appear to function as either exogenous or endogenous signals for the Dif pathway. It can therefore be inferred that 111. xanthus Tfp function as a sensor or part of a sensory apparatus for the perception of signals for the Dif pathway in the regulation of EPS production. It was further proposed that S-motility involves a regulatory loop in which EPS triggers Tfp retraction

Figure 3 Model depicting the regulation of EPS production in M. xanthus by Tfp and the Dif pathway. Demonstrated interactions are indicated by solid lines, and proposed interactions are indicated by dashed lines. Arrows and bars indicate positive and negative regulation, respectively. See the text for details of the model. Adapted.from Black et al., 2006.

STRUCTUREAND METABOLISM

236 and Tfp provide proximity signals to the Dif pathway to modulate EPS production (Black et al., 2006). This

proposal is based on the following concepts: Tfp are the likely motors for S-motility, EPS appears to regulate Tfp retraction, and S-motility requires cell proximity for normal function. Besides the Dif proteins and Tfp, other proteins are likely to play regulatory roles in M. xanthus EPS production. Two of the best known are the DnaK homologs StkA (described above) and SglK. Whereas stkA mutants are more cohesive and overproduce EPS (Dana and Shimkets, 1993), sglK null mutants lack EPS (Weimer et al., 1998; Yang et al., 1998a) and show virtually no group movement on soft-agar plates. Surprisingly, an sglK insertion mutant at the 3’ end of the gene forms fractal patterns on soft-agar plates (MacNeil et al., 1994). The fibR gene, which is linked to the sglK locus, seems to encode a negative regulator of ECM, because a fibR mutant produces an increased amount of the ECM protein FibA (Weimer et al., 1998). The possible regulatory function of FibR is further supported by the fact that it is a homolog of the repressors of alginate production in Pseudomonas spp. It is likely that stkA, sglK, and fibR are all regulators of M . xanthus EPS biosynthesis based on the phenotypes of their corresponding mutants and that they are all homologs of regulatory proteins. There are a number of other potential or confirmed regulators of M . xanthus EPS biosynthesis. The NtrClike activator Nla24 that regulates both A- and S-motility (Caberoy et al., 2003; Lancero et al., 2004) is thought to regulate S-motility due to its control of EPS production (Lancero et al., 2004). The nla24 mutant lacks EPS production. Similarly, three additional NtrC-like activators, Nlal, Nla19, and Nla23, are potential regulators of M . xanthus EPS biosynthesis (Caberoy et al., 2003). Another regulator of M. xanthus EPS is the tyrosine kinase MasK (Thomasson et al., 2002). The masK815 mutant allele, which harbors two missense mutations, was isolated as a partial suppressor of the S-motility defect resulting from an mglA missense mutation. A masK81.S mglA8 double mutant overproduced both EPS and the ECM protein FibA. Distinct from other EPS regulators, masK appears to be an essential gene in M. xanthus. Other less well characterized proteins are involved in M. xanthus EPS biosynthesis in a regulatory or biosynthetic capacity. The protein pair RppA and MmrA are homologs of a sensory transducer and a multidrugresistant protein, respectively, that appear to be important for M. xanthus EPS or polysaccharide biosynthesis (Kimura et al., 2004). It is puzzling that although single mutations in either rppA or mmrA showed little effect on polysaccharide production, an rppA mmrA double

mutant showed a decreased amount of polysaccharide production. It is unclear how a transducer and a multidrug transporter could perform redundant or synergistic functions in either EPS regulation or biosynthesis. Other proteins with ambiguous functions in EPS production include RasA, EsgA, and MglA (Dana and Shimkets, 1993; Pham et al., 2005; Ramaswamy et al., 1997). Further investigation is necessary to fully understand the complexity of EPS regulation in M. xanthus.

ECM BIOGENESIS Preliminary biochemical studies indicated that M . xanthus EPS contains at least five monosaccharides: galactose, glucosamine, glucose, rhamnose, and xylose (Behmlander and Dworkin, 1994b). From the newly sequenced M . xanthus genome, pathways for producing all these sugars in either UDP-, GDP- or dTDP-activated forms can be constructed ( Z . Yang and H. B. Kaplan, unpublished data). The activated monosaccharides may be used for EPS production by synthetic enzymes, some of which are likely encoded by the eps locus (Lu et al., 2005). The deduced eps products include homologs of various polysaccharide synthesis proteins such as glycosyltransferases and polysaccharideexport proteins. Mutations in some of the esp genes and the two eas genes have been shown to result in defects in EPS production (Lu et al., 2005; Barbu et al., unpublished). The Nla24 protein (described above) is encoded by the eps locus (Lancero et al., 2004; Lu et al., 2005). The eps locus likely encodes additional regulators of EPS production because many of the open reading frames are homologs of regulatory proteins such as histidine kinases and response regulators (Lu et al., 2005). The exact functions of these potential biosynthetic enzymes and regulatory proteins remain to be elucidated. As described above, the only known ECM protein is FibA. It was identified by its reactivity with a monoclonal antibody (MAb2105) raised against cell surface antigens of M . xanthus (Behmlanderand Dworkin, 1994a; Gill et al., 1985).It was initially puzzling that MAb2105 recognized proteins with apparently different molecular weights or electrophoretic mobility. Behmlander and Dworkin (1994a) suggested that these proteins arose from the same protein product. Amino acid sequencing of the 31-kDa protein (Behmlander and Dworkin, 1994a) and the availability of M. xanthus genome sequence led to the discovery of the fibA gene (Kearns et al., 2002). The deduced polypeptide sequence suggests that FibA is a zinc metalloprotease. The immunoblot results suggest that FibA undergoes autoproteolytic processing that generates multiple polypeptides

23 7

13. M. XANTHUS CELLENVELOPE recognizable by MAb2105 (Kearns et al., 2002). The fibA mutants are defective in tactic responses to dilauroyl (12:O) phosphatidylethanolamine (PE) but have normal responses to dioleoyl (18:l) PE. The fruiting bodies formed by fibA mutants are elongated and irregularly shaped. Otherwise, fibA mutants are proficient in cellular cohesion and in both A- and S-motility, further suggesting that ECM proteins are not critical for the structure of ECM or its function in cohesion and S-motility. The mechanisms by which FibA is involved in PE taxis are not understood. FibA may be involved in sensory functions or responsible for the proteolytic processing of proteins that participate in the M. xanthus PE response.

PILI M. xanthus cells possess polar pili that can be visualized by negative staining and transmission EM (Dobson and McCurdy, 1979; Dobson et al., 1979; Kaiser, 1979). The pili are less than 10 nm in diameter and are usually present only at one of the two cell poles. Recent observations with the atomic force microscope confirm the dimensions (5 to 8 nm) and polar localization of M. xanthus pili (Pelling et al., 2005). Consistent with the polar localization, the sequence of M. xanthus pilA, which encodes the pilin subunit of the polymeric pilus fiber, indicates that M. xanthus pili belong to the class of Tfp which are found in gram-negative bacteria (Wu and Kaiser, 1995).Bacterial Tfp are important for many processes including surface movement, competency, and host colonization by pathogens. Readers are directed to recent reviews on Tfp (Craig et al., 2004; Mattick, 2002; Meibom et al., 2005; Pizarro-Cerda and Cossart, 2006; Wall and Kaiser, 1999). The atomic structures of pilins from a few organisms have been solved by X-ray crystallography or nuclear magnetic resonance (Craig et al., 2004; Ramboarina et al., 2005). Studies of M . xanthus pili focus on their involvement in S-motility. Dale Kaiser was the first to observe a tight association between pili and S-motility (Kaiser, 1979). This was confirmed by the observation that mechanical removal of M. xanthus Tfp was accompanied by the loss of S-motility (Rosenbluh and Eisenbach, 1992). Further studies provided additional support for the requirement of Tfp for S-motility (Nudleman et al., 2005,2006; Wall et al., 1998; Wall et al., 1999; Wu and Kaiser, 1995, 1996; Wu et al., 1997, 1998; Youderian and Hartzell, 2005). It is generally accepted that Tfp retraction provides the force for M. xanthus S-motility and the related surface motility known as bacterial twitching (Mattick, 2002; Nudleman and Kaiser, 2004). Tfp are proposed

to attach to their targets either on cells or other surfaces and pull the cells forward by retracting. The pulling force of Tfp retraction was measured by laser tweezers (Maier et al., 2002, 2004; Merz et al., 2000), and the retraction was observed directly using a novel combination of staining with amino-specific Cy3 fluorescent dye and total internal reflection microscopy (Skerker and Berg, 2001). There is also indirect evidence for pilus retraction in M. xanthus (Sun et al., 2000). Readers are directed to chapter 6 in this book for more information on the function of Tfp in M . xanthus S-motility.

References Arnold, J. W., and L. J. Shimkets. 1988a. Inhibition of cell-cell interactions in Myxococcus xanthus by Congo red. J. Bacteriol. 1705765-5770. Arnold, J. W., and L. J. Shimkets. 1988b. Cell surface properties correlated with cohesion in Myxococcus xanthus. J. Bacteriol. 1705771-5777. Ashton, A. 1993. Structural Studies of the Lipopolysaccharide from Myxococcus xanthus and Lipopolysaccharide Mutants. Ph.D. thesis. University of Minnesota, Minneapolis. Atrich, A., G. Bacher, G. Allmaier, M. P. Williamson, and S. J. Foster. 1999. Analysis of peptidoglycan structure from vegetative cells of Bacillus subtilis 168 and role of PBP 5 in peptidoglycan maturation. J. Bacteriol. 181:3956-3966. Bayan, N., I., Guilvout, and A. I?. Pugsley. 2006. Secretins take shape. Mol. Microbiol. 60:1-4. Behmlander, R. M., and M. Dworkin. 1991. Extracellular fibrils and contact-mediated cell interactions in Myxococcus xanthus. J. Bacteriol. 173:7810-7820. Behmlander, R. M., and M. Dworkin. 1994a. Integral proteins of the extracellular matrix fibrils of Myxococcus xanthus. J. Bacteriol. 176:6304-63 11. Behmlander, R. M., and M. Dworkin. 199413. Biochemical and structural analyses of the extracellular matrix fibrils of Myxococcus xanthus. J. Bacteriol. 176:6295-6303. Behrens, S. 2003. Periplasmic chaperones-preservers of subunit folding energy for organelle assembly. Cell 113556-557. Bellenger, K., X. Ma, W. Shi, and Z. Yang. 2002. A Chew homologue is required for Myxococcus xanthus fruiting body development, social gliding motility, and fibril biogenesis. J. Bacteriol. 1845654-5660. Black, W. P., and Z. Yang. 2004. Myxococcus xanthus chemotaxis homologs DifD and DifG negatively regulate fibril polysaccharide production. J. Bacteriol. 186:1001-100 8. Black, W. P., Q. Xu, and Z. Yang. 2006. Type IV pili function upstream of the Dif chemotaxis pathway in Myxococcus xanthus EPS regulation. Mol. Microbiol. 61:447-456. Bos, M. P., and J. Tommassen. 2004. Biogenesis of the Gramnegative bacterial outer membrane. Curr. Opin. Microbiol. 7~610-616. Bowden, M. G., and H. B. Kaplan. 1998. The Myxococcus xanthus lipopolysaccharide 0-antigen is required for social motility and multicellular development. Mol. Microbiol. 30:275-284.

238 Caberoy, N. B., R. D. Welch, J. S., Jakobsen, S. C., Slater, and A. G. Garza. 2003. Global mutational analysis of NtrC-like activators in Myxococcus xanthus: identifying activator mutants defective for motility and fruiting body development.]. Bacteriol. 185:6083-6094. Craig, L., M. E. Pique, and J. A. Tainer. 2004. Type IV pilus structure and bacterial pathogenicity. Nat. Rev. Microbiol. 2~363-378. Cumsky, M., and D. R. Zusman. 1979. Myxobacterial hemagglutinin: a development-specific lectin of Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 765505-5509. Dana, J. R., and L. J. Shimkets. 1993. Regulation of cohesiondependent cell interactions in Myxococcus xanthus. J. Bacteriol. 175:3636-3647. Dobson, W. J., and H. D. McCurdy. 1979. The function of fimbriae in Myxococcus xanthus. I. Purification and properties of M . xanthus fimbriae. Can. J. Microbiol. 25:1152-1160. Dobson, W. J., H. D. McCurdy, and T. H. MacRae. 1979. The function of fimbriae in Myxococcus xanthus. 11. The role of fimbriae in cell-cell interactions. Can. J. Microbiol. 25~1359-1372. Downard, J. S., and D. R. Zusman. 1985. Differential expression of protein S genes during Myxococcus xanthus development. J. Bacteriol. 161:1146-1155. Dworkin, M. 1999. Fibrils as extracellular appendages of bacteria: their role in contact-mediated cell-cell interactions in Myxococcus xanthus. Bioessays 21590-595. Dworkin, M., and S. M. Gibson. 1964. A system for studying microbial morphogenesis: rapid formation of microcysts in Myxococcus xanthus. Science 146:243-244. Dwyer, M. A., and G. J. Hellinga. 2004. Periplasmic binding proteins: a versatile superfamily for protein engineering. Curr. Opin. Struct. Biol. 14:495-504. Fink, J. M., and J. F. Zissler. 1989. Characterization of lipopolysaccharide from Myxococcus xanthus by use of monoclonal antibodies. /. Bacteriol. 171:2028-2032. Fink, J. M., M. Kalos, and J. F. Zissler. 1989. Isolation of cell surface antigen mutants of Myxococcus xanthus by use of monoclonal antibodies. J. Bacteriol. 171:2033-2041. Freese, A., H. Reichenbach, and H. Lunsdorf. 1997. Further characterization and in situ localization of chain-like aggregates of the gliding bacteria Myxococcus fulvus and Myxococcus xanthus. J. Bacteriol. 179:1246-1252. Gill, J., E. Stellwag, and M. Dworkin. 1985. Monoclonal antibodies against cell-surface antigens of developing cells of Myxococcus xanthus. Ann. Inst. Pasteur. Microbiol. 136A:ll-18. Gill, J. S., and M. Dworkin. 1986. Cell surface antigens during submerged development of Myxococcus xanthus examined with monoclonal antibodies. J. Bacteriol. 168505-5 11. Guo, D., Y. Wu, and H. B. Kaplan. 2000. Identification and characterization of genes required for early Myxococcus xanthus developmental gene expression. J. Bacteriol. 182~4564-4571. Guo, D., M. G. Bowden, R. Pershad, and H. B. Kaplan. 1996. The Myxococcus xanthus rfbABC operon encodes an ATPbinding cassette transporter homolog required for 0-antigen biosynthesis and multicellular development. J. Bacteriol. 178~1631-1639.

STRUCTUREAND METABOLISM Hillesland, K. L., and G. J. Velicer. 2005. Resource level affects relative performance of the two motility systems of Myxococcus xanthus. Microb. Ecol. 49558-566. Hodgkin, J., and D. Kaiser. 1979a. Genetics of gliding motility in Myxococcus xanthus: two gene systems control movement. Mol. Gen. Genet. 171:177-191. Hodgkin, J., and D. Kaiser. 1979b. Genetics of gliding motility in Myxococcus xanthus (Myxobacterales):genes controlling movement of single cells. Mol. Gen. Genet. 171:167-176. Jacobs, C., L. J. Huang, E. Bartowsky, S. Normark, and J. T. Park. 1994. Bacterial cell wall recycling provides cytosolic muropeptides as effectors for p-lactamase induction. EMBO J. 13:4684-4694. Janssen, G. R., and M. Dworkin. 1985. Cell-cell interactions in developmental lysis of Myxococcus xanthus. Dev. Biol. 112:194-202. Johnson, Y. R., and D. White. 1972. Myxospore formation in Myxococcus xanthus: chemical changes in the cell wall during cellular morphogenesis. J. Bacteriol. 1123349-855. Kaiser, D. 1979. Social gliding is correlated with the presence of pili in Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 7659.52-5956. Kaplan, H. B., A. Kuspa, and D. Kaiser. 1991. Suppressors that permit A-signal-independent developmental gene expression in M ~ X O C O C C xanthus. US J. Bacteriol. 173:1460-1470. Kearns, D. B., P. J. Bonner, D. R. Smith, and L. J. Shimkets. 2002. An extracellular matrix-associated zinc metalloprotease is required for dilauroyl phosphatidylethanolamine chemotactic excitation in Myxococcus xanthus. J. Bacteriol. 184:1678-1684. Killeen, K. P., and D. R. Nelson. 1988. Acceleration of starvation- and glycerol-induced myxospore formation by prior heat shock in Myxococcus xanthus. J. Bacteriol. 17052005207. Kim, S. H., S. Ramaswamy, and J. Downard. 1999. Regulated exopolysaccharide production in Myxococcus xanthus. J. Bacteriol. 181:1496-1507. Kimura, Y., S. Ishida, H. Matoba, and N. Okahisa. 2004. RppA, a transducer homologue, and MmrA, a multidrug transporter homologue, are involved in the biogenesis and/or assembly of polysaccharide in Myxococcus xanthus. Microbiology 150:631-639. Kinscherf, T. G., and D. K. Willis. 2002. Global regulation by gidA in Pseudomonas syringae. J. Bacteriol. 184:22812286. Kirby, J. R., and D. R. Zusman. 2003. Chemosensory regulation of developmental gene expression in Myxococcus xanthus. Proc. Natl. Acad. Sci. U S A 100:2008-2013. Lancero, H., N. B. Caberoy, S. Castaneda, Y. Li, A. Lu, D. Dutton, X. Y. Duan, H. B. Kaplan, W. Shi, and A. G. Garza. 2004. Characterization of a Myxococcus xanthus mutant that is defective for adventurous motility and social motility. Microbiology 150:4085-4093. Li, Y., H. Sun, X. Ma, A. Lu, R. Lux, D. Zusman, and W. Shi. 2003. Extracellular polysaccharides mediate pilus retraction during social motility of Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 1005443-5448. Lomovskaya, O., and M. Totrov. 2005. Vacuuming the periplasm. J. Bacteriol. 187:1879-1883.

13. M . XANTHUS CELLENVELOPE Lu, A., K. Cho, W. P. Black, X. Y. Duan, R. Lux, Z. Yang, H. B. Kaplan, D. R. Zusman, and W. Shi. 2005. Exopolysaccharide biosynthesis genes required for social motility in Myxococcus xanthus. Mol. Microbiol. 55:206-220. Lunsdorf, H., and H. Reichenbach. 1989. Ultrastructural details of the apparatus of gliding motility of Myxococcus fulvus (Myxobacterales).Microbiology 135:1633-1 64 1. Lunsdorf, H., and H. U. Schairer. 2001. Frozen motion of gliding bacteria outlines inherent features of the motility apparatus. Microbiology 147:939-947. MacNeil, S. D., A. Mouzeyan, and P. L. Hartzell. 1994. Genes required for both gliding motility and development in M y x o coccus xanthus. Mol. Microbiol. 14:785-795. Maier, B., L. Potter, M. So, C. D. Long, H. S. Seifert, and M. P. Sheetz. 2002. Single pilus motor forces exceed 100 pN. Proc. Natl. Acad. Sci. U S A 99:16012-16017. Maier, B., M. Koomey, and M. P. Sheetz. 2004. A forcedependent switch reverses type IV pilus retraction. Proc. Natl. Acad. Sci. U S A 101:10961-10966. Mattick, J. S. 2002. Type IV pili and twitching motility. Annu. Rev. Microbiol. 56:289-314. Meibom, K. L., M. Blokesch, N. A. Dolganov, C. Y. Wu, and G. K. Schoolnik. 2005. Chitin induces natural competence in Vibrio cholerae. Science 310:1824-1827. Merroun, M. L., K. Ben Chekroun, J. M. Arias, and M. T. Gonzalez-Munoz. 2003. Lanthanum fixation by Myxococcus xanthus: cellular location and extracellular polysaccharide observation. Chemosphere 52:113-120. Merz, A. J., M. So, and M. P. Sheetz. 2000. Pilus retraction powers bacterial twitching motility. Nature 407:98-102. Mogensen, J. E., and D. E. Otzen. 2005. Interactions between folding factors and bacterial outer membrane proteins. Mol. Microbiol. 57326-346. Nelson, D. R., M. G. Cumsky, andD. R. Zusman. 1981. Localization of myxobacterial hemagglutinin in the periplasmic space and on the cell surface of Myxococcus xanthus during developmental aggregation. J. Biol. Chem. 256:12589-12595. Nelson, D. R., and K. I? Killeen. 1986. Heat shock proteins of vegetative and fruiting Myxococcus xanthus cells. J. Bacterial. 168:1100-1 106. Nudleman, E., and D. Kaiser. 2004. Pulling together with type IV pili. J . Mol. Microbiol. Biotechnol. 752-62. Nudleman, E., D. Wall, and D. Kaiser. 2005. Cell-to-cell transfer of bacterial outer membrane lipoproteins. Science 309~125-127. Nudleman, E., D. Wall, and D. Kaiser. 2006. Polar assembly of the type IV pilus secretin in Myxococcus xanthus. Mol. Microbiol. 60:16-29. O’Connor, K. A., and D. R. Zusman. 1997. Starvationindependent sporulation in Myxococcus xanthus involves the pathway for p-lactamase induction and provides a mechanism for competitive cell survival. Mol. Microbiol. 24~839-850. O’Connor, K. A., and D. R. Zusman. 1999. Induction of plactamase influences the course of development in Myxococcus xanthus. J . Bacteriol. 181:6319-6331. Park, J. T. 1995. Why does Escherichia coli recycle its cell wall peptides? Mol. Microbiol. 17:421-426.

239 Pelling, A. E., Y. Li, W. Shi, and J. K. Gimzewski. 2005. Nanoscale visualization and characterization of Myxococcus xanthus cells with atomic force microscopy. Proc. Natl. Acad. Sci. U S A 102:6484-6489. Pham, V. D., C. W. Shebelut, B. Mukherjee, and M. Singe. 2005. RasA is required for Myxococcus xanthus development and social motility. J. Bacteriol. 187:6845-6848. Pizarro-Cerda, J., and P. Cossart. 2006. Bacterial adhesion and entry into host cells. Cell 124:715-727. Raetz, C. R., and C. Whitfield. 2002. Lipopolysaccharide endotoxins. Annu. Rev. Biochem. 71:635-700. Ramaswamy, S., M. Dworkin, and J. Downard. 1997. Identification and characterization of Myxococcus xanthus mutants deficient in calcofluor white binding.]. Bacteriol. 179:28632871. Ramboarina, S., P. J. Fernandes, S. Daniell, S. Islam, P. Simpson, G. Frankel, F. Booy, M. S. Donnenberg, and S. Matthews. 2005. Structure of the bundle-forming pilus from enteropathogenic Escherichia coli. J. Biol. Chem. 280:40252-40260. Rivera, J. J. 2002. An Extracytoplasmic Function Sigma Factor Operon Regulates Myxococcus xanthus Developmental Gene Expression. Ph.D. thesis. University of Texas, Houston. Rodriguez, A. M., and A. M. Spormann. 1999. Genetic and molecular analysis of cglB, a gene essential for single-cell gliding in Myxococcus xanthus.J. Bacteriol. 181:4381-4390. Romeo, J. M., B. Esmon, and D. R. Zusman. 1986. Nucleotide sequence of the myxobacterial hemagglutinin gene contains four homologous domains. Proc. Natl. Acad. Sci. U S A 83:6332-63 36. Romeo, J. M., and D. R. Zusman. 1987. Cloning of the gene for myxobacterial hemagglutinin and isolation and analysis of structural gene mutations. J. Bacteriol. 169:3801-3808. Romeo, J. M., and D. R. Zusman. 1991. Transcription of the myxobacterial hemagglutinin gene is mediated by a sigma 54-like promoter and a cis-acting upstream regulatory region of DNA. J. Bacteriol. 173:2969-2976. Romeo, J. M., and D. R. Zusman. 1992. Determinants of an unusually stable mRNA in the bacterium Myxococcus xanthus. Mol. Microbiol. 6:2975-2988. Rosenbluh, A., and M. Eisenbach. 1992. Effect of mechanical removal of pili on gliding motility of Myxococcus xanthus. J. Bacteriol. 17454064413. Schlieker, C., A. Mogk, and B. Bukau. 2004. A PDZ switch for a cellular stress response. Cell 117417-419. Sha, J., E. V. Kozlova, A. A. Fadl, J. I? Olano, C. W. Houston, J. W. Peterson, and A. K. Chopra. 2004. Molecular characterization of a glucose-inhibited division gene, gidA, that regulates cytotoxic enterotoxin of Aeromonas hydrophila. Infect. Imrnun. 72:1084-1095. Shimkets, L., and T. W. Seale. 1975. Fruiting-body formation and myxospore differentiation and germination in M x y o coccus xanthus viewed by scanning electron microscopy. J. Bacteriol. 121:711-720. Shimkets, L. J., and D. Kaiser. 1982a. Induction of coordinated cell movement in Myxococcus xanthus. J. Bacteriol. 152:451-461. Shimkets, L. J., and D. Kaiser. 1982b. Murein components rescue developmental sporulation of Myxococcus xanthus. J. Bacteriol. 152:462-470.

240 Shimkets, L. J. 1986a.Role of cell cohesion in Myxococcusxanthus fruiting body formation. J. Bacteriol. 1 6 6 9 4 2 4 4 8 . Shimkets, L. J. 1986b. Correlation of energy-dependent cell cohesion with social motility in Myxococcus xanthus. J. Bacteriol. 1665337-841. Skerker, J. M., and H. C. Berg. 2001. Direct observation of extension and retraction of type IV pili. Proc. Natl. Acad. Sci. USA 98:6901-6904. Sourjik, V., and R. Schmitt. 1996. Different roles of CheYl and CheY2 in the chemotaxis of Rhizobium meliloti. Mol. Microbiol. 22:427-436. Sourjik, V., and R. Schmitt. 1998. Phosphotransfer between CheA, CheY1, and CheY2 in the chemotaxis signal transduction chain of Rhizobium meliloti. Biochemistry 37: 232 7-23 35. Sun, H., D. R. Zusman, and W. Shi. 2000. Type IV pilus of Myxococcus xanthus is a motility apparatus controlled by the frz chemosensory system. Curr. Biol. 10:1143-1146. Thomasson, B., J. Link, A. G. Stassinopoulos, N. Burke, L. Plamann, and P. L. Hartzell. 2002. MglA, a small GTPase, interacts with a tyrosine kinase to control type IV pili-mediated motility and development of Myxococcus xanthus. Mol. Microbiol. 46: 1399-141 3. Vollmer, W., and J. V. Holtje. 2004. The architecture of the murein (peptidoglycan) in gram-negative bacteria: vertical scaffold or horizontal layer(s).J. Bacteriol. 1865978-5987. Wall, D., S. S. Wu, and D. Kaiser. 1998. Contact stimulation of Tgl and type IV pili in Myxococcus xanthus. J. Bacteriol. 180:759-76 1. Wall, D., and D. Kaiser. 1999. Type IV pili and cell motility. Mol. Microbiol. 32:l-10. Wall, D., P. E. Kolenbrander, and D. Kaiser. 1999. The Myxococcus xanthus pilQ (sglA) gene encodes a secretin homolog required for type IV pilus biogenesis, social motility, and development. J. Bacteriol. 181:24-33. Webre, D. J., P. M. Wolanin, and J. B. Stock. 2003. Bacterial chemotaxis. Cum Biol. 13:R47-R49. Weimer, R. M., C. Creighton, A. Stassinopoulos, P. Youderian, and P. L. Hartzell. 1998. A chaperone in the HSP70 family controls production of extracellular fibrils in Myxococcus xanthus. J. Bacteriol. 18053574368. White, D., M. Dworkin, and D. J. Tipper. 1968. Peptidoglycan of Myxococcus xanthus: structure and relation to morphogenesis. J. Bacteriol. 95:2186-2197. White, D. J., and P. L. Hartzell. 2000. AglU, a protein required for gliding motility and spore maturation of Myxococcus

STRUCTURE AND METABOLISM xanthus, is related to WD-repeat proteins. Mol. Microbiol. 36:662-678. White, D. J., R. Merod, B. Thomasson, and P. L Hartzell. 2001. GidA is an FAD-binding protein involved in development of Myxococcus xanthus. Mol. Microbiol. 42503517. Wu, S. S., and D. Kaiser. 1995 Genetic and functional evidence that Type IV pili are required for social gliding motility in Myxococcus xanthus. Mol. Microbiol. 18547-558. Wu, S. S., and D. Kaiser. 1996. Markerless deletions of pi1 genes in Myxococcus xanthus generated by counterselection with the Bacillus subtilis sacB gene. J. Bacteriol. 178: 58 17-5 82 1. Wu, S. S., J. Wu, and D. Kaiser. 1997. The Myxococcus xanthus pilT locus is required for social gliding motility although pili are still produced. Mol. Microbiol. 23:109-121. Wu, S. S., J. Wu, Y. L. Cheng, and D. Kaiser. 1998. The pilH gene encodes an ABC transporter homologue required for type IV pilus biogenesis and social gliding motility in Myxococcus xanthus. Mol. Microbiol. 29:1249-1261. Xu, D., C. Yang, and H. B. Kaplan. 1998. Myxococcus xanthus sasN encodes a regulator that prevents developmental gene expression during growth. J. Bacteriol. 180:6215-6223. Xu, Q., W. P. Black, S. M. Ward, and Z. Yang. 2005. Nitratedependent activation of the Dif signaling pathway of Myxococcus xanthus mediated by a NarX-DifA interspecies chimera. ]. Bacteriol. 187:6410-6418. Yang, C., and H. B. Kaplan. 1997. Myxococcus xanthus sass encodes a sensor histidine kinase required for early developmental gene expression. ]. Bacteriol. 179:7759-7767. Yang, Z., Y. Geng, and W. Shi. 1998a. A DnaK homolog in Myxococcus xanthus is involved in social motility and fruiting body formation.]. Bacteriol. 180:218-224. Yang, Z., Y. Geng, D. Xu, H. B. Kaplan, and W. Shi. 1998b. A new set of chemotaxis homologues is essential for Myxococcus xanthus social motility. Mol. Microbiol. 30:1123-1130. Yang, Z., X. Ma, L. Tong, H. B. Kaplan, L. J. Shimkets, and W. Shi. 2000. Myxococcus xanthus dif genes are required for biogenesis of cell surface fibrils essential for social gliding motility. J. Bacteriol. 18257934798. Yang, Z., and Z. Li. 2005. Demonstration of interactions among Myxococcus xanthus Dif chemotaxis-like proteins by the yeast two-hybrid system. Arch. Microbiol. 183:243-252. Youderian, P. A., and P. Hartzell. 2006. Transposon insertions of magellan-4 that impair social gliding motility in Myxococcus xanthus. Genetics 172:1397-1410.

Myxohacteria: Multicellularity and Differentiation Edited by David E. Whitworth 0 2008 ASM Press, Washington, D.C.

Patrick D. Curtis Lawrence J. Shimkets

Metabolic Pathways Relevant to Predation, Signaling, and Development

Genome analysis offers the ability to examine the central metabolism of organisms from a perspective that, while holistic in information content, lacks the accuracy of conclusions derived from sound biochemical and genetic analysis. Shadows of possibilities emerge that only silhouette necessary experimentation. Conversely, biochemical and genetic analyses are incomplete without the genomic knowledge. Here, previously determined biochemical and genetic data are analyzed through the scope of genomic analysis, with particular attention to catabolic and anabolic pathways involved in the basic biology of the myxobacteria. Genomic annotation was based upon the completed Myxococcus xanthus genome (Goldman et al., 2006). Metabolic pathways were examined by homology searching using the amino acid sequence for well-characterized enzymes. Metacyc (http://metacyc.org/) contains a compilation of most known pathways, enzymes, and genes and has a pathway analysis for 211. xanthus which was often used as a reference point (Caspi et al., 2006). A limitation of this analysis is its reliance on the annotation conducted by The Institute for Genomic Research (TIGR). To reduce problems with annotation errors, all pathways in the chapter were also annotated by hand

14

using BLASTp. Two considerations were used in assigning a function to a putative protein sequence, amino acid identity with known members of the EC class (>30%) and orthologous gene (COG) determination by TIGR. A strong hit satisfies both criteria. Gene and protein names adopted below are derived primarily from homologs found in Escherichia coli K-12 when applicable. This policy deviates from TIGR, which names the genes/ proteins after the closest named homolog in the database and does not specify the origin of the name. As with any homology-based search, the results should be considered “putative” until gene knockouts and enzyme assays confirm the identity and function of the homolog. The word putative is avoided for the sake of brevity but it should be understood that the proposed pathways represent little more than the best candidates. The first portion of this chapter, entitled “Catabolic Pathways,” deals with the catabolism of amino acids and lipids, as they are the principal carbon and energy sources derived from prey bacteria. The second portion of this chapter, entitled “Anabolic Pathways,” highlights the synthesis of lipids because of their unusual chemical structures in myxobacteria, and also the spore-specific components trehalose and ether lipids.

Patrick D. Curtis and Lawrence J. Shimkets, Department of Microbiology, University of Georgia, Athens, GA 30602.

241

STRUCTURE AND METABOLISM

242

CATABOLIC PATHWAYS Most myxobacteria, including M . xanthus, can catabolize prey microorganisms. M. xanthus utilizes amino acids and lipids as carbon and energy sources, incorporates purines and pyrimidines via salvage pathways, and fails to utilize sugars (Bretscher and Kaiser, 1978; Hemphill and Zahler, 1968a, 196810; Lau et al., 2002; Loebeck and Klein, 1956). The literature is extensive and is not reviewed here. Rather, genomic evidence for specific pathways involved in assimilation and catabolism is provided. Most of the research examining the sources of energy for the myxobacteria has focused on the assimilation of amino acids, not without reason (Bretscher and Kaiser, 1978; Dworkin, 1962). The average E. coli cell is composed of roughly 55% (dry weight) protein (Neidhardt and Umbarger, 1996), by far the largest component of the cell. Though lipids represent a much smaller fraction of the total cell mass (9% [Neidhardt and Umbarger, 1996]),their energetic content is considerable. The catabolism of serine to CO, yields approximately 11 ATP, while the catabolism of a 16:O fatty acid to CO, yields approximately 80 ATP. Therefore, both cellular fractions represent rich sources of energy.

Amino Acid Catabolism In most cases there is excellent agreement between the presence of a particular amino acid catabolic pathway and the ability of that amino acid to stimulate growth in defined and minimal media. The pathways are listed below according to amino acid, roughly in alphabetical order.

Alanine M . xanthus uses alanine dehydrogenase for catabolism of L-alanine (Fig. 1, reaction 1).L-Alanine can also be converted to D-alanine for use in peptidoglycan biosynthesis using alanine racemase (EC 5.1.1.1, Alr, MXAN7160). D-Alanine may be catabolized to pyruvate using the ironcontaining alcohol dehydrogenase MXAN5629. Arginine M. xanthus contains many pathways for catabolizing arginine, so it is surprising that arginine is not a major component in minimal and defined media (Bretscher and Kaiser, 1978). Collectively these pathways produce L-glutamate, putrescine (and hence other polyamines), L-proline, and succinate. The bulk of the arginine degradation in E. coli is carried out by the arginine succinyltransferase pathway, which is found in many Proteobacteria that use arginine as a sole carbon source. This pathway is not present

in M . xanthus, which may explain the lack of importance for arginine in minimal media. Several arginine catabolic pathways use arginase to hydrolyze L-arginine. Different organisms have variations of the arginase pathway depending on the fate of the catabolic products. The classical arginase pathway yields L-glutamate. There is weak homology for arginase, (26 to 31% amino acid identity), but beyond that the pathway is conserved in M. xanthus (Fig. 1, reaction 2). There is a branch in this pathway as ornithine cyclodeaminase (EC 4.3.1.12, MXAN7463) converts L-ornithine to L-proline. The arginine decarboxylase pathway produces putrescine as an intermediate for synthesis of other polyamines and succinate as an end product (Fig. 1,reaction 3).E. coli has two forms of arginine decarboxylase. Constitutively expressed SpeA is used for the biosynthesis of putrescine, while inducibly expressed AdiA is used for arginine catabolism. Interestingly, there is no homolog for AdiA but there is a strong homolog of SpeA in M . xanthus. All the other elements of the pathway to degrade arginine to succinate exist, suggesting that SpeA plays a dual role in the biosynthesis of putrescine and catabolism of arginine. M . xanthus contains two homologs of E. coli transaminase YgiG (MXAN3014 and MXAN7377). E. coli uses at least three different enzymes to catalyze the last reaction, Sad, AldA, and GabD. There are many homologs of each in M . xanthus, and though MXAN2844 is annotated as succinate semialdehyde dehydrogenase, there are other possibilities.

AsparagineIAspartate E. coli K-12 has three different L-asparaginases to deaminate asparagine, two of which are present in M . xanthus. E. coli aspartase (EC 4.3.1.1) removes the nitrogen from aspartate to produce fumarate, but there is no evidence for this enzyme in M. xanthus. Instead 111.xanthus appears to use a version of this pathway found in some gram-positive bacteria and mammals to produce phosphoenolpyruvate (Fig. 1, reaction 4).

Cysteine Two pathways are known for the catabolism of L-cysteine to pyruvate; at least one may exist in M . xanthus. In E. coli two desulfhydrases convert L-cysteine to pyruvate in a single step. Tryptophanase (TnaA, also used in tryptophan catabolism) does not appear to have a homolog in M . xanthus. MetC, P-cystathionase (also used in methionine synthesis), has four homologs in M . xanthus: MXAN2035, MXAN0969, MXAN0970, and MXAN1955 (Fig. 1, reaction 5). Of these, MXAN2035

14. PREDATION,SIGNALING,AND DEVELOPMENT (1)

+

L-alanine + NAD++ H,O

243

pyruvate + NADH + ammonia

EC 1.4.1.1 alanine dehydrogenase MXAN4146

(2)

L-arginine

+

L-ornithine EC 3.5.3.1 arginase MXAN4431

L-glutamatey-semialdehyde

EC 2.6.1.13 ornithine amino transferase MXAN7377

+

L-glutamate y-semialdehyde

pyrroline 5-carboxylate

Spontaneous

(3)

L-arginine

+

agmatine

EC 4.1.1.19 arginine decarboxylase MXAN2742

+

+

L-glutamate

EC 1.5.1.12 1-pyrroline-5-carhoxylatedehydrogenase MXAN5891

putrescine

+

4-amino-butyraldehyde

EC 3.5.3.11 EC 2.6.1.29 agmatinase putrescine transaminase MXAN4431 MXAN3014 or MXAN7377

4-amino-butyraldehyde

+

4-aminobutyrate

9 succinate semialdehyde

EC 1.2.1.19 EC 2.6.1.19 y-aminohutyraldehydedehydrogenase 4-aminobutyrate transaminase MXAN0921 MXAN3014

succinate semialdehyde

+

succinate

EC 1.2.1.16 succinate semialdehyde dehydrogenase MXAN2844

(4)

L-asparagine

+

L-aspartate + a-ketoglutarate

EC 3.5.1.1 L-asparaginase I MXAN5198 L-asparaginase I1 MXAN1160

Oxaloacetate + GTP

+

+

oxaloacetate + L-glutamate

EC 2.6.1.1 aspartate aminotransferase MXAN3386

phosphoenolpyruvate+ GDP + C 0 2

EC 4.1.1.32 phosphoenolpyrnvatecarboxykinase MXAN1264

(5)

L-cysteine + H,O

3 pyruvate + ammonia + H2S

EC 4.4.1.1 L-cysteine desulihydrase MXAN2035

Figure 1 Amino acid catabolic reactions 1 through 5 (alanine to cysteine). See the text.

has the most homology with the bacterial and yeast genes. The oxidation of cysteine to 3-sulfinoalanine and eventually to pyruvate is the major route of cysteine catabolism in mammals. The first step in this pathway involves cysteine dioxygenase (EC 1.13.11.20). There is weak evidence for a homolog in M . xantbus (MXAN4718). The remaining enzyme in the pathway, aspartate amino transferase (MXAN3386), is likely to be present (see catabolism of asparagine/aspartate).

Glutamine/Glutamate Two enzymes convert L-glutamine to L-glutamate: glutaminase and glutamate synthase. The biochemical properties of the two major E. coli glutaminases (EC 3.5.1.2) have been studied in detail, but the genes encoding them have not been identified. Two putative glutaminase genes have been described in E. coli, ybaS and yneH; however, neither has a homolog in M. xantbus. Nor is there a homolog of human glutaminase C. A number of amidotransferases, such as anthranilate synthetase, have

STRUCTUREAND METABOLISM

244 glutaminase activity and may contribute to glutamine catabolism. The biochemistry and genetics of glutamate synthase are known from a variety of bacteria. Glutamate synthase (EC 1.4.1.13) catalyzes the transamination of aketoglutarate to produce two glutamates. The Klebsiella aerogenes enzyme consists of a single 175-kDa protein, whereas the E. coli enzyme consists of 53- and 135-kDa subunits. M. xanthus has adjacent genes encoding two subunits like E. coli, MXAN3917 and MXAN3918. One of the major catabolic pathways for glutamate involves deamination to a-ketoglutarate by glutamate dehydrogenase. This is a key reaction in the catabolism of arginine, glutamine, histidine, and proline, all of which produce glutamate as an intermediate. Glutamate dehydrogenase catalyzes a reversible reaction that can be either anabolic or catabolic depending on the conditions and the organism. M . xanthus glutamate dehydrogenase (GdhA) is MXAN5873. Mammalian glutamate dehydrogenase uses both NAD’ and NADP+ as cofactors (EC 1.4.1.3). Most prokaryotic enzymes can use either NAD+ (EC 1.4.1.2) or NADP’ (EC 1.4.1.4). The TIGR annotation gives the M. xanthus enzyme as EC 1.4.1.3, suggesting that this enzyme may use both cofactors. The primary pathway for the use of L-glutamate as a carbon source in E. coli is transamination of oxaloacetate to form L-aspartate and a-ketoglutarate by AspC (Fig. 2, reaction 6). Aspartate aminotransferase (EC 2.6.1.1) resembling that of a Bacillus species is found in the M. xanthus genome, MXAN3386, and is dramatically different from the E. coli AspC. AspA (EC 4.3.1.1), which produces fumarate from L-aspartate by deamination, appears to be absent, suggesting that the aspartate is directed toward the synthesis of the aspartate family of amino acids or catabolized to oxaloacetate and phosphoenolpyruvate (see “Asparagine/Aspartate” above). Glutamate can also enter the tricarboxylic acid TCA cycle as succinate using the three-step glutamate decarboxylase pathway (Fig. 2, reaction 7) in which the final two steps are identical to those for putrescine degradation (see “Arginine” above). There is also evidence for a second bacterial pathway for catabolizing glutamate to succinate (Fig. 2, reaction 8).

Glycine Glycine plays a key role in C1 anabolism through the generation of the C1 donor Nlo-formyl-tetrahydrofolate. All elements of the pathway are present in M . xanthus with the exception of the anaerobic formate dehydrogenase (Fig. 2, reaction 9). These results suggest that the pathway is strictly anabolic unless another route exists for oxidation of formate. Four polypeptides are involved

in the glycine cleavage pathway, GcvHPT and LpdA, with the first three genes forming an operon in both E. coli and M . xanthus.

Histidine M . xanthus may catabolize L-histidine to L-glutamate (Fig. 3, reaction 10).The first three enzymes in the pathway have strong M. xanthus homologs, but here the trail runs dry. Bacillus subtilis uses EC 3.5.3.8 for the last step, but evidence for this enzyme in M. xanthus is lacking. Pseudomonas uses two enzymes for the last step, EC 3.5.3.13 and EC 3.5.1.68, to remove ammonia and formate sequentially. The first of these has a homolog in the M. xanthus genome (formiminoglutamate deiminase, MXANlOlO), but a homolog of the formate hydrolase is not apparent.

IsoleucinelLeucineNaline The branched-chain amino acids are unique in that they are essential for growth in all M. xanthus isolates that have been examined and are also essential for secondary metabolite production in myxobacteria and other organisms (Bretscher and Kaiser, 1978). Catabolism of these amino acids appears to be directed primarily at their use as primers for the synthesis of branchedchain fatty acids (see “Lipid Biogenesis” below). The key enzyme in this process is the branched-chain keto acid dehydrogenase (BCKAD) complex, also known as Esg in M. xanthus, which catalyzes the oxidative decarboxylation of the branched-chain a-keto acids derived from leucine, isoleucine, and valine (Fig. 3, reaction 11). In mammals the complex consists of 12 branched-chain a-ketoacid dehydrogenase ( E l )subunits and 6 dihydrolipoyl dehydrogenase (E3) subunits noncovalently associated with a core of 24 dihydrolipoyl transacylase (E2) components. A BCKAD kinase inactivates the complex by phosphorylation of alpha subunits of the heterotetrameric (a2p2) E l component; BCKAD phosphatase removes phosphates to activate the complex. The complex is present in M. xanthus, though there is no evidence for regulation by covalent modification.

Lysine At least nine pathways are known for the catabolism of lysine that vary in the initial products. The genes are known for only three of these pathways. In E. coli, lysine is decarboxylated to cadaverine by IdcC or CadA, neither of which is present in M . xanthus. In plants and animals, lysine is oxidized to saccharopine by using a unique dehydrogenase that is not present in M. xanthus. In fungi, lysine is degraded to glutarate following acetylation of the six-amino group with a unique lysine Wacetyltransferase,

14. PREDATION,SIGNALING,AND DEVELOPMENT (6)

L-glutamine + a-ketoglutarate

+

L-glutamate

+ oxaloacetate

EC 1.4.1.13 glutamate synthase MXAN3917, large subunit MXAN3918. small sununit

(7)

L-glutamate

(8)

L-glutamate

+

succinyl-CoA 3

L-aspartate + a-ketoglutarate

+

succinate semialdehyde

EC 2.6.1.19 4-aminobutyrateaminotransferase MXAN3014

a-ketoglutarate 3

EC 2.6.1.1 aspartate aminotransferase MXAN3386

3

EC 2.6.1.1 aspartate aminotransferase MXAN3386

3 4-aminobutyrate

EC 4.1.1.15 glutamate decarboxylase MXAN6783

245

+

succinate

EC 1.2.1.16 succinate semialdehyde dehydrogenase MXAN2844

S-succinyl-dihyrolipoamide 3 succinyl-Cod

EC 1.2.42 2-oxoglutarate dehydrogenase MXAN6035, E l component

EC 2.3.1.61 2-oxoglutarate dehydrogenase MXAN6036, E2 component

succinate

EC 6.2.1.5 succinyl-CoAsynthase MXAN3542, a subunit MXAN3541, p subunit

(9)

Glycine + tetrahydrofolate 3 5,lO-methylene-tetrahydrofolate EC 1.4.4.2 and EC 2.1.2.10 glycine dehydrogenase,MXAN3042 tetrahydrofolateaminomethyltransferase MXAN3041, MXAN3040, MXAN6341

5,lO-methylene-tetrahydrofolate 3

5,lO-methenyl-tetrahydrofolate

EC 1.5.1.15 methylenetetrahydrofoiate dehydrogeanse MXAN1095

5,lO-methenyl-tetrahydrofolate

+

N1o-formyl-tetrahydrofolate

EC35.4.9 methenyl tetrahydrofolatecyclobydrolase MXAN2226

N1o-formyl-tetrahydrofolate+ phosphate + ADP

+

formate + tetrahydrofolate + ATP

EC 6.3.4.3 formate tetrahydroformateligase MXAN0175

Figure 2 Amino acid catabolic reactions 6 through 9 (glutamine to glycine). See the text.

which is also missing in M. xanthus. Therefore, it is not clear whether lysine is catabolized.

Methionine The methionine catabolic pathway is similar to that observed in mammals and involves the synthesis of Sadenosyl-L-methionine for transmethylation reactions and then hydrolysis of adenosine to produce homocysteine (Fig. 3 , reaction 12).Homocysteine forms a branch point in the pathway and may either be recycled to Lmethionine, using the vitamin B,,-dependent methionine synthase (not shown), or degraded to succinate.

Phenylalanine/Tyrosine Only a single aerobic phenylalanine/tyrosine catabolic pathway is known. In this pathway phenylalanine is catabolized to tyrosine by phenylalanine 4-hydroxylase (EC 1.14.16.1, MXAN5127) and eventually to succinate. While homologs for genes involved in some later steps are found in M . xanthus, no homolog for the second step in the catabolic pathway (EC 2.6.1.5, tyrosine aminotransferase) is found. In M. xanthus, 14C-labeled phenylalanine is converted to tyrosine and the excess tyrosine is secreted into the growth medium (Hemphill and Zahler, 1968a), suggesting that this pathway is,

STRUCTURE AND METABOLISM

246 (10)

L-histidine -9

-9

urocanate

EC 4.3.1.3 histidine ammonia lyase MXAN3465

4-imidazolone-5-propionate

EC 4.2.1.49 urocanase MXAN4343

4-imidazolone-5-propionate -9

N-formimine-L-glutamate -9

EC 3.5.2.7 imidazolone-5-propionatase MXAN4345

(1 1)

L-isoleucine L-leucine L-valine

2-keto-3-methyl-valerate 2-keto-4-methyl-pentanoate 2-keto isovalerate

+

S-adenosyl-L-methionine -9

L-methionine -9

EC 2.5.1.6 methionine adenosyltransferase MXAN6517

S-adenosyl-homocysteine -9

homocysteine

cystathionine -9

EC 43.1.22 cystathionine p-synthase MXAN2041

propionyl CoA -9

S-adenosyl-homocysteine

EC 2.1.1.73 DNA methyltransferase MXAN3598

EC 3.3.1.1 S-adenosylhomocysteinehydrolase MXAN6516

Homocysteine -9

2-methylbutyr yl-CoA Isovaleryl-CoA Isobutyryl-CoA

-9

BCKAD Ela, MXAN4564 BCKAD Elp, MXAN4565 BCKAD E2, MXAN4217 BCKAD E3, MXAN4219

EC 2.6.2.42 branched chain amino acid aminotransferase MXAN2987

(12)

L-glutamate

??

-9

L-methionine

EC 2.1.1.13 methionhe synthase MXAN1971

2-oxobutanoate -9

EC 4.4.1.1 cystathionine y-lyase MXAN3917

propionyl CoA

no EC number or gene

(S)-methyl-malonyl-CoA -9 (R)-methyl-malonyl-CoA

EC 6.4.1.3 propionyl-CoA carboxylase MXAN1111, a subunit MXAN1113, p subunit

(R)-methyI-malonyl-CoA -9

EC 5.1.99.1 no gene associated with activity

succinyl-CoA

EC 5.4.99.2 methylmalonyl-CoAmntase MXAN2263, a subunit MXAN2264, p subunit

Figure 3 Amino acid catabolic reactions 10 through 12 (histidine to methionine). See the text.

indeed, blocked at the first step of tyrosine catabolism (Fig. 4, reaction 13). Proline There is a single pathway for degradation of proline that involves the sequential action of proline dehydrogenase and 1-pyrroline-5-carboxylate dehydrogenase to produce glutamate (Fig. 4,reaction 14). In bacteria the two enzyme domains are usually encoded by a single gene, whereas in eukaryotes they are separate genes. M. xanthus appears to have separate and unlinked genes.

Serine L-Serine is deaminated to pyruvate and ammonia in E. coli by three homologous serine deaminases, but only one is found in M. xanthus (Fig. 4, reaction 15). Threonine L-Threonine can be converted to many metabolites by pathways whose biochemistry has far surpassed the genetics. All the known pathways begin with either deamination to 2-oxobutanoate or oxidation to 2amino-3-oxobutanoate. In M. xanthus, the former reaction appears to be present while the latter reaction is

14. PREDATION,SIGNALING,AND DEVELOPMENT (13)

+

L-phenylalanine

247

L-tyrosine

EC 1.14.16.1 phenylalanine 4-hydroxylase MXAN5127

(14)

L-proliie

3

l-pyrroline 5-carboxylate

EC 15.99.8 proline dehydrogenase MXAN7405

(15)

L-serine

+

+

L-glutamate

EC 1.5.1.12 l-pyrroline-5-carboxylate dehydrogenase MXAN5891

pyruvate + ammonia

EC 4.3.1.17 serine deaminase MXAN6186

(16)

L-threonine

+

2-oxobutanoate + cysteine + ammonia 3

EC 4.3.1.19 threonine dehydratase MXAN5874 threonine deaminase MXAN6186

cystathionine

EC 4.4.1.1 cystathionine y-lyase MXAN2035

Figure 4 Amino acid catabolic reactions 13 through 16 (phenylalanine to threonine). See the text.

doubtful (Fig. 4, reaction 16). Threonine is then catabolized to cystathionine, an intermediate in methionine catabolism that can be used to generate succinate, methionine, or pyruvate (see “Methionine” above).

Tryptophan There are seven known pathways for the catabolism of tryptophan. The biochemistry has been examined more extensively than the genetics due to the fact that various products of tryptophan are plant hormones, dyes, or putrid by-products during cheese production. The simplest pathway is the conversion of tryptophan to pyruvate and indole in a single step by tryptophanase. This pathway is intriguing, as indole induces spore formation in Stigmatella and may be a cell-cell signal (Stamm et al., 2005). Neither a homolog of this enzyme nor any enzymes in the other pathways were detected in 111. xanthus. Additionally, no tryptophanase homolog could be found in the Stzgmatella aurantiaca genome. Some parts of a eukaryotic catabolic pathway are found, but several enzymes in the middle of the pathway are missing. In summary, genome evidence suggests that M. xanthus is missing pathways for catabolism of seven amino acids: leucine, isoleucine, valine, phenylalanine, tyrosine, tryptophan, and lysine. The absence of catabolic pathways for leucine, isoleucine, and valine is not surprising given that M. xanthus is auxotrophic for these amino acids (Bretscher and Kaiser, 1978) and also uses them in fatty acid biosynthesis (see “Fatty Acid Primer Synthesis”

below). Catabolism of these amino acids would reduce branched-chain fatty acid synthesis, which appears to be necessary for development (Toal et al., 1995). The absence of phenylalanine/tyrosine, tryptophan, and lysine catabolic pathways is a mystery. It is interesting that of the mixture of amino acids that comprise the A signal (an early developmental signal), leucine, isoleucine, phenylalanine, tyrosine, and tryptophan account for 63% of the A signal activity (Kuspa et al., 1992).

Purine and Pyrimidine Salvage A variety of labeling and analog toxicity studies indicate that purines and pyrimidines are efficiently salvaged in M. xanthus, and these studies are not reviewed here (Hemphill and Zahler, 1968a; Tsai and Westby, 1978). M. xanthus has a suite of purine and pyrimidine salvage pathways comparable to what is found in E. coli with several options for converting nucleobases and nucleosides into nucleotides. The preferred method of nucleoside and deoxynucleoside salvage in E. coli is to first remove the (deoxy)ribose moiety from the (deoxy)nucleoside and then use a phosphoribosyltransferase to create the nucleotide monophosphate. E. coli can also transport the (deoxy)nucleosides and then phosphorylate them. Adenosine and deoxyadenosine are salvaged through deamination to inosine/deoxyinosine (EC 3.5.4.4, Add, MXAN1519) and hydrolysis to hypoxanthine (purine nucleoside phosphorylase, DeoD, MXAN2306). Guanosine and deoxyguanosine are hydrolyzed to guanine

248 (purine nucleoside phosphorylase, DeoD, MXAN2306). Phosphoribosyltransferases were found for adenine (EC 2.4.2.7, Apt, MXAN5352) and uracil (EC 2.4.2.9, Upp, MXANO124). E. coli has two additional phosphoribosyltransferases, both of which function on hypoxanthine and guanine. The M. xanthus homolog appears to be more closely related to one with a bias for hypoxanthine (EC 2.4.2.-, Hyp, MXAN5070), but it should be assumed from labeling studies that M. xanthus can utilize guanine as well. Uridine and cytidine are phosphorylated to UMP/ CMP by uridine kinase (EC2.7.1.48, Udk, MXAN4159). Thymidine and thymine nucleotides are deoxy compounds with no ribonucleotide counterparts; thymidine is phosphorylated to TMP by thymidine kinase (EC 2.7.1.21, Tdk, MXAN5072). Deoxycytidine is deaminated to deoxyuridine and either degraded to uracil and/or phosphorylated to deoxyUMP by thymidine kinase (EC 2.7.1.21, Tdk, MXAN5072) and then converted to TMP by thymidylate synthase (EC 2.1.1.45, ThyA, MXAN.5942). It should be noted from the genomic studies here and from previous labeling studies (Hemphill and Zahler, 1968a) that the main fate of exogenous purinedpyrimidines is incorporation into nucleic acids and not degradation for energy generation.

Lipid Catabolism Lipid oxidation has been demonstrated by 14C-labeling experiments in Myxococcus virescens (Loebeck and Klein, 1956) and methyloleate feeding in M. xanthus (Lau et al., 2002). The principal, but not sole, source of lipids in prey bacteria is the phospholipids, which may be hydrolyzed by four classes of lipases in M. xanthus. Phospholipase D (MXAN6753) removes the head group, leaving phosphatidic acid at the sn-3 position on the glycerol backbone. Fatty acids are located at the sn-1 and sn-2 positions in phospholipids. Three classes of lipases generate glycerol and fatty acids (MoraledaMuiioz and Shimkets, 2007). Both are robust carbon sources, although glycerol utilization by 111. xanthus has not been reported. Interestingly, most lipase homologs identified here are not predicted to have sn positionspecific activity, perhaps indicating their importance for catabolism over lipid signaling, Lipases Patatins were originally identified as the major potato storage protein, but biochemical characterization revealed fatty esterase activity, usually on compounds containing a single fatty acyl chain. MXAN3852 contains the conserved active site, the oxyanion hole,

STRUCTURE AND METABOLISM and other features that suggest it may be catalytically functional. Three other patatin homologs are present in M. xanthus but may be missing motifs necessary for catalytic function. MXAN3852 is expressed only during starvation, and deletion results in no phenotype with regards to aggregation and sporulation on TPM agar (Tris-phosphate-magnesium starvation agar; MoraledaMuiioz and Shimkets, 2007). On clone fruiting (CF) agar there is a 24-h delay to both aggregation and sporulation, and the fruiting bodies are unusually large and amorphous. The a / p hydrolases contain an eight-stranded p sheet (stabilized by intervening a helices) formed of two antiparallel p strands followed by six parallel p strands (Ollis et al., 1992). Many a/p hydrolases are proteases, but at least two putative a / p hydrolases in M . xanthus are likely lipases. MXAN5522 (a triacyl glycerol lipase acting on all three sn positions) is located directly upstream of a lipase chaperone homolog (MXAN5523). Lipases often have chaperones to prevent catalytic activity until they are secreted. This lipase is expressed during both vegetative growth and development (MoraledaMufioz and Shimkets, 2007). Interestingly, it shows a sharp spike of expression 24 h into development. Deletion of MXAN5522 results in threefold-increased spore yield on both TPM and CF. While there is no defect in aggregation on TPM, this strain aggregates 24 h faster than the wild type on CF. A possible explanation for the increased rate of sporulation in this strain may be that it is unable to utilize storage lipids for energy, thereby starving faster than normal. MXAN4638 is a putative lysophospholipase, which removes the fatty acid from lysophospholipids. Interestingly, disruption of the gene upstream and operonic with MXAN4638 generates an A-motility defect (Youderian et al., 2003). M. xanthus contains two GDSL lipase homologs (MXAN.5500 and MXAN4569), which are general bacterial lipases that liberate fatty acids from either sn position in phospholipids. The C-terminal domain is a @-barrelporin which forms a pore in the outer membrane through which the N-terminal catalytic domain is secreted and ultimately anchors the lipase facing away from the membrane. Both M. xanthus homologs lack the C-terminal anchoring domain. There is one report of a GDSL lipase anchored to the membrane by an Nterminal acylation (Klingsbichel, 1996), and though MXAN4569 appears to have an acylation signal, MXANSSOO does not. The lack of an anchoring domain may be an indication that MXAN5.500 diffuses away from the cell. Deletion of MXAN4569 has no effect on aggregation or sporulation on TPM (Moraleda-Muiioz and Shimkets, 2007). On CF, aggregation shows a mild

14. PREDATION,SIGNALING,AND DEVELOPMENT fruiting body morphology defect and sporulation proceeds approximately 24 h faster than the wild type. The lipases from MXAN3852, MXAN5522, and MXAN4569 were all tested in vitro against various tagged lipid derivatives to elucidate fatty acyl chain specificity. All lipases showed preferences for short acyl chains and had the highest activity for two carbon chain lengths.

p Oxidation Fatty acids are usually degraded by P oxidation, where two carbon acetate units are sequentially removed from the carboxyl end of the fatty acid, also known as the A terminus, as opposed to the methyl end or w terminus (for a review, see Clark and Cronan, 1996). This process resembles fatty acid elongation in reverse. FadL (MXAN7040) translocates the free fatty acid across the outer membrane, which is then translocated across the inner membrane and esterified to a coenzyme A (CoA) moiety by FadD (Weimar et al., 2002). M. xanthus contains as many as 10 FadD homologs (EC 6.2.1.3; the most likely homologs are MXAN1573, MXAN7148, MXAN0216, and MXAN0225). Next, a trans double bond is introduced two carbons from the A terminus (A2)by FadE. While the M. xanthus FadE (EC 1.3.99.3) homologs are not clear, MXAN3795 and MXAN3797 are conspicuous possibilities as they appear to be in an operon with MXAN3791, an AtoB homolog. Also interesting is MXAN6989, which encodes a peptide with strong homology to the C terminus of FadE and appears to be in an operon with FadA and FadB homologs. Next, water is substituted into the double bond to create a P-hydroxy fatty acyl-CoA which is oxidized to a p-keto group. Both these reactions are performed by the multifunctional fatty acid oxidation complex P subunit FadB. Additionally, FadB epimerizes D-P-hydroxy fatty acyl CoA to L-P-hydroxy fatty acyl-CoA for further processing. M. xanthus contains two FadB homologs (EC 1.1.1.35, MXAN5371 and MXAN6987). Upon generation of the p-keto fatty acyl-CoA, the fatty acid oxidation complex 01 subunit (FadA) cleaves the chain at the P-keto group by adding a CoA moiety, creating acetylCoA and the fatty acyl-CoA truncated by two carbons. M. xanthus contains two FadA homologs (EC 2.3.1.16), MXAN5372 and MXAN6988, that are located next to FadB homologs. This P oxidation cycle continues until there are only four carbons left on the fatty acid chain when the chain is hydrolyzed to two acetyl-CoA units by AtoB (EC 2.3.1.9, MXAN3791 and MXAN5135). For fatty acids with unsaturations at odd numbers of carbons from the carboxyl terminus, FadB isomerizes the

249

cis-A3double bond to trans-A2.Fatty acids with unsaturations at even numbers of carbons from the carboxyl terminus are reduced by 2,4-dienoyl-CoA reductase (EC 1.3.1.34, FadH, MXAN3389). This enzyme reduces the second double bond of the trans-A2,&-A4 intermediate to trans-A2 fatty acyl-CoA, which is further oxidized by the FadE and the unsaturated fatty acid degradation pathway (Hubbard et al., 2003). Like E. coli, M. xanthus contains two sets of both major P oxidation pathway enzymes, FadA and FadB. In E. coli, one set functions under aerobic conditions while the other set works under anaerobic conditions (Campbell et al., 2003). While M . xanthus is a strict aerobe, it is possible that the cells may encounter periods where energy generation is needed under oxygen-limiting conditions, such as the interior of a fruiting body, where free fatty acids may be a plentiful energy source. Expression studies would help determine if and when each set of P oxidation enzymes is expressed. a Oxidation 01 oxidation removes a single carbon from the fatty acid (Caspi et al., 2006) and has been demonstrated in S. aurantiaca (Dickschat et al., 2005). Molecular oxygen is added to a free fatty acid to create P-hydroperoxyfatty acid by the fatty acid 01 dioxygenase (EC 1.11.13 .-, MXAN5217). This molecule spontaneously degrades, though it can be facilitated by the dioxygenase (Hamberg et al., 2005), to either a @-hydroxy-fattyacid, or a fatty aldehyde with a loss of CO, and H,O. The fatty aldehyde dehydrogenase (EC 1.2.1.3, MXAN6986) oxidizes the aldehyde to the acid and in the process reduces NAD to NADH. a oxidation only generates one NADH as opposed to the several generated as a result of P oxidation. Despite the two homologs, 01 oxidation was not observed in radiolabeling studies with M. xanthus (Bode et al., 2005). In peas, a oxidation is observed only during seed germination, indicating a situational role for this pathway (Saffert et al., 2000). By analogy, perhaps 01 oxidation is important during M . xanthus spore germination. w Oxidation

During w oxidation the last carbon in a chain is converted to a carboxyl group, creating fatty acids from alkanes and dicarboxylates from fatty acids. M. xanthus does not appear to have a homolog of the w fatty acid oxidase.

Carbohydrate Utilization M . xanthus and most myxobacteria (with the exception of Sorangiurn and Byssophaga species) cannot grow

250 on carbohydrates (Bretscher and Kaiser, 1978; Watson and Dworkin, 1968). Prey bacteria are not particularly enriched in carbohydrates (E. coli contains 2.5% glycogen [Neidhardt and Umbarger, 1996]),yet carbohydrates are energy rich and are available in the rhizosphere. The ability of M . xanthus to utilize carbohydrates is examined in two parts: carbohydrate assimilation by the phosphotransferase system and carbohydrate metabolism by glycolysis. Phosphotransferase System While sugars can be internalized through each of the major types of transport systems, the phosphotransferase system (PTS) is specific to carbohydrates and should be an indicator of carbohydrate utilization. The PTS pathway begins with the transfer of a phosphoryl group from phosphoenolpyruvate (PEP) to Enzyme I (EI).This phosphoryl group is then transferred to HPr, which then transfers the phosphoryl group, sometimes through another carrier protein, to the permease composed of IIA, IIB, IIC, and sometimes IID subunits, which phosphorylate the transported carbohydrate. M . xanthus has homologs of all the PTS components neatly located in an operon (EI, MXAN6530; HPr, MXAN6.531; and IID-A, MXAN6532-5). In E. coli, EI and HPr are common components of all PTS and substrate specificity is conferred by the sugar-specificI1 proteins. Unlike E. coli, which has over a dozen sets of I1 proteins, M. xanthus has only one, but it is homologous with the E. coli system that has the broadest substrate range. The M. xanthus system is a Class 4 PTS (for a review see Postma et al., 1993), due to the presence of a IID homolog, most closely resembling the mannose PTS in E. coli. The M . xanthus system has separate IIA and IIB subunits like Klebsiella pneumoniae and B. subtilis (Postma et al., 1993).The M . xanthus IIA and IIB homologs have the conserved His10 and His175 residues involved in the phosphorelay (Erni et al., 1987). Therefore, at this level of analysis M . xanthus has all the necessary components for a functional PTS system. The most homologous system is the E. coli mannose PTS system, which transports eight different sugars: mannose, N-acetylglucosamine, glucosamine, fructose, 2-deoxyglucose,glucose, trehalose,and methyl ct-glucoside (Postma et al., 1996). Many of these sugars have been examined as growth substrates in defined and minimal media for M . xanthus without success (Bretscher and Kaiser, 1978). N-acetylglucosamine has been used in peptidoglycan labeling studies with poor incorporation (L. Shimkets, unpublished data). As PTS is dependent on the availability of PEP, perhaps the cellular level of PEP is in short supply under the conditions examined in transport

STRUCTUREAND METABOLISM studies. Another possibility is that the PTS system may be expressed only under specific conditions. As the system is predicted to transport trehalose, a possible link with germination exists as the spore-specific sugar trehalose is first secreted and then disappears (see “Trehalose Biosynthesis” below). Glycolysis Monosaccharides are used for exopolysaccharide, peptidoglycan, and lipopolysaccharide biosynthesis. Additionally, M . xanthus produces glycogen (a homopolysaccharide of glucose monomers joined with 0l-1~4linkages and ct-1,6 branches) during early to middle stationary phase (Nariya and Inouye, 2003). Glycogen is a common carbon and energy storage polymer in many bacteria. Monosaccharides are produced by gluconeogenesis, which uses many of the glycolysis enzymes in reverse. Glycolysis has three kinase reactions not involved in gluconeogenesis: those performed by glucokinase, phosphofructokinase, and pyruvate kinase. For the first step in glycolysis, glucose must be phosphorylated to glucosel-phosphate. Extracellular glucose can be phosphorylated by the PTS system (see above), though activity has not been demonstrated in M . xanthus. While intracellular glucose is phosphorylated by glucokinase, there are no recognizable glucokinase homologs in the genome, and glucokinase activity was not observed in M . xanthus (Watson and Dworkin, 1968). Glycogen is consumed in E. coli by removing glucose monomers from the nonreducing end of the chain as glucose-l-phosphate by glycogen phosphorylase GlgP (Alonso-Casajus et al., 2006). N o glycogen phosphorylase homolog is found in the genome. The second unique kinase step, that performed by phosphofructokinase, has been carefully studied (Nariya and Inouye, 2002). M . xanthus has a phosphofructokinase (EC 2.7.1.11, Pfk, MXAN6373). The activity of Pfk is modulated by phosphorylation of Thr-226 by the protein serinekhreonine kinase Pkn4 (Nariya and Inouye, 2002). Additionally, the kinase activity of Pkn4 is controlled by a regulatory protein MkapB and may be involved in a larger signaling network (Nariya and Inouye, 2005). Deletion of phosphofructokinase leads to elevated glycogen levels and decreased spore production (Nariya and Inouye, 2003). Spore production can be restored by supplying the mutant with exogenous pyruvate, which is then presumably catabolized by the tricarboxylic acid cycle. The third unique kinase step is missing. Although M . xanthus contains two pyruvate kinase homologs (EC 2.7.1.40, Pyk, MXAN3514 and MXAN6299), enzymatic activity was not found in cell or spore extracts

14. PREDATION,SIGNALING,AND DEVELOPMENT (Watson and Dworkin, 1968). Pyruvate kinase is necessary for development in S. aurantiaca and binds indole, a known sporulation inducer in that organism (Stamm et al., 2005). Glycogen produced during stationary phase is consumed prior to sporulation, consumption of glycogen is necessary for efficient sporulation, and consumption of glycogen is linked to the regulated activity of phosphofructokinase (Nariya and Inouye, 2003). These results indicate that glycolysis occurs during development. However, the source of phosphorylated glucose for glycolysis is unclear as no glycogen phosphorylase homolog is found in the genome, and the later and critical glycolysis reaction perfomed by pyruvate kinase has not been detected. Additionally, during development monosaccharides are needed for trehalose (see “Trehalose Biosynthesis” below) and exopolysaccharide biosynthesis, suggesting competing destinations for glucose in the cell. One hypothesis is that the glycolytic pathway functions incompletely during development. The reaction following phosphofructokinase cleaves fructose-1, 6-bisphosphate into glyceraldehyde-3-phosphate and dihydroxyacetone phosphate. Dihydroxyacetone phosphate may be used in ether lipid biosynthesis (see “Ether Lipid Biosynthesis” below), compounds found only in M. xanthus during development. This unconventional use of glycolysis may explain the unusual method of enzymatic regulation of phosphofructokinase. Further testing, specifically during development, is required to determine if M. xanthus uses the full glycolysis pathway.

ANABOLIC PATHWAYS Lacking space in this chapter to focus on all aspects of anabolism, we chose to focus on unusual cell envelope molecules and spore-specific products. Over the years the structural components of the myxobacterial cell envelope have been examined. Lipids are by far the most structurally divergent group of macromolecules relative to other Proteobacteria and may perform roles in cell signaling (Curtis et al., 2006; Downard and Toal, 1995; Kearns et al., 2001). In the following sections we discuss phospholipids and ceramides. Steroids and polyketides are discussed in chapter 15. Additionally the synthesis and fate of the spore-specific carbohydrate trehalose are examined.

Lipid Biogenesis While phospholipids perform a basic structural role, the fatty acid diversity of myxobacteria is extraordinary and far exceeds that necessary for structural purposes. While most Proteobacteria have 3 to 5 different fatty acids (Cronan and Rock, 1996), vegetative M. xanthus

251 cells contain at least 18 different fatty acids (Curtis et al., 2006; Kearns et al., 2001). Within M. xanthus phospholipids, fatty acids can have straight chains or branched chains, and either type can be saturated or unsaturated. The principal phospholipid in 211. xanthus, phosphatidylethanolamine (PE), deviates markedly from PE species of other Proteobacteria in that it contains unsaturated fatty acids at the sn-1 position, which appear to have a role in cell signaling (Curtis et al., 2006). In this section, phospholipid (and ceramide) biosynthesis is examined with attention to sources of structural diversity.

Fatty Acid Primer Synthesis The biosynthesis of fatty acids begins with the generation of a primer that is extended in a series of repetitive cycles to form the fatty acid. For straight-chain fatty acids with even numbers of carbons the primer is acetyl-CoA. The branched-chain fatty acids are the most abundant fatty acids in M. xanthus. The primers for branched-chain fatty acids are derived from the branched-chain amino acids (see “Isoleucine/Leucine/Valine” in “Amino Acid Catabolism” above for detail). Fatty acids derived from leucine constitute the iso-odd fatty acid family, those derived from isoleucine constitute the anteiso-odd family, and those derived from valine constitute the iso-even family. In the first step of branched-chain primer synthesis, the amine group is removed, producing the aketoacid branched-chain derivative (EC 2.6.1.42, IlvE, MXAN2987). The a-ketoacids are decarboxylated and attached to CoA by the BCKAD complex to form the primer. Myxobacterial BCKAD enzymes may have more pronounced substrate specificities than usual. Deuterium-labeled leucine was incorporated into fatty acids of both M. xanthus and S. aurantiaca, whereas little labeled valine was incorporated (Bode et al., 2005); concordantly, the most abundant branched-chain fatty acids in these two organisms are the iso-odd fatty acids (Dickschat et al., 2005; Kearns et al., 2001; Ware and Dworkin, 1973). In Streptomyces, incorporation of valine into iso-even fatty acids was equal to that of leucine into isoodd fatty acids (Cropp et al., 2000). M . xanthus transposon insertions in the genes encoding Elci and E1P subunits (esg locus) block development within the first 5 h (Downard et al., 1993).These mutants show decreased amounts of branched-chain fatty acids and increased levels of the fatty acid 16:105c (Curtis et al., 2006; Kearns et al., 2001). It is possible that branched-chain fatty acids are necessary for the generation of a chemical signal required for development or for extracellular matrix production (Kim et al., 1999),which is essential for fruiting body formation (Behmlander and Dworkin, 1991; Lu et al., 2005; Yang et al., 2000).

STRUCTURE AND METABOLISM

252

52

ACP-S-C-CHa CH -(CH2),-CH3

FabUA H20

12

,

14

I

16

Mutant 2

Wild-type

*

18

Time [min]

Figurer 8 Identification of the chivosazol biosynthetic gene cluster by inactivation of PKS genes using plasmid integration by homologous recombination. (a) Inactivation plasmid containing the selection marker HygRand a homologous region obtained by PCR using degenerate PKS primers. (b)Biosynthetic gene cluster; localization of mutations is marked. ( c )HPLC chromatogram of culture extracts of S. cellulosum So ce56 wild type (wt) and chivosazolnegative mutants (Mutant1 and Mutant2). (d) Bioassay for chivosazol production using Hansenula anomala.

Recent Improvements in the Molecular Biology of Sorungium The development of genetic manipulation systems for Sorangium (transposon mutagenesis and insertion inactivation) was further improved by an electroporation protocol for foreign DNA transfer, which was successfully applied for So ce12 (Kopp et al., 2005). In contrast to other myxobacteria (for example, M. xanthus), a longer incubation time for the phenotypical expression of the selection marker is needed. About 20 h (more than two doubling times) was used to enable the transposition into the S . cellulosum So ce12 genome and the expression of the hygromycin resistance. Electroporation is a more convenient and efficient method than conjugation, and the development of this technique for Sorangium

strains is a significant improvement in the development of the genetic manipulation system. Moreover, protocols for quantitative reverse transcriptase PCR and proteome analysis in Sorangium have been established (Kegler et al., 2006; Y. Elnakady, A. Alici, R. Miiller, and K. Niehaus, unpublished data). Interestingly, it has been shown for the first time by realtime PCR analysis that the presumed large secondary metabolic transcripts (exemplified by the chivosazol and etnangien transcripts from So ce56, which are assumed to be approximately 90 kbp in size) seem to be unusually stable, with a half-life time of 30 min in contrast to control transcripts (e.g., phosphopantetheinyl transferase from So ce56), which were shown to be degraded completely after 10 to 15 min (Kegler et al., 2006).

19. SORANGIUMCELLULOSUM Identified Secondary Metabolite Biosynthetic Gene Clusters from Sorungium Strains Recently investigations concerning the molecular basis of the biosynthesis of secondary metabolites in strains of the genus Sorungium have become more extensive (Schupp et al., 1995; Julien et al., 2000; Tang et al., 2000; Ligon et al., 2002; Pradella et al., 2002; Gerth et al., 2003; Kopp et al., 2004, 2005; Perlova et al., 2006). Chivosazol, disorazol, soraphen, and epothilone genes represent examples for published biosynthetic gene clusters. Disorazol and chivosazol genes were analyzed recently (Carvalho et al., 2005; Kopp et al., 2005; Perlova et al., 2006), and the genetic basis of other secondary metabolites biosyntheses is currently being investigated (leupyrrin, sorangicin, etnangien, and spirangiene; 0. Perlova, M. Kopp, B. Frank, and R. Muller, unpublished data). The disorazol and chivosazol biosynthetic gene clusters belong to a group of lately described truns-acyltransferase (trans-AT)type I PKS (Cheng et al., 2003; Piel et al., 2004) with a discrete AT domain located on a separate protein (Kopp et al., 2005; Perlova et al., 2006). In addition, both gene clusters are characterized by highly unusual properties such as “split-modules” located on separate proteins and hybrid proteins containing NRPS and PKS modules on the same polypeptide (Wenzel and Miiller, 2005). Both biosynthetic gene clusters showed additional exceptions from the textbook rules, which are based on work with better-investigated secondary metabolite biosynthetic systems (Carvalho et al., 2005; Kopp et al., 2005; Perlova et al., 2006). More extensive biochemical studies are required to explain the details of the biosynthesis in each case.

Heterologous Expression of the Sorungium Biosynthetic Genes Although S. cellulosum strains possess great biotechnological importance, they are slow-growing bacteria and it is therefore desirable to produce the natural products of these microorganisms (e.g., epothilone as a potential anticancer agent, see above) in an alternative host with a better fermentation potential and thus enable the overproduction of the substance and/or generate altered products. Different heterologous expression systems are currently being developed (seeJulien and Shah, 2002; Zirkle et al., 2004a; and Wenzel et al., 2005). But the heterologous expression remains a challenging task, and only a relatively small number of successfully expressed genes in foreign strains are known to date. Several examples in the literature describe not only the expression of the biosynthetic gene clusters in phylogenetically related strains (e.g., expression of bacitracin from Bacillus licheniformis

343 in B. subtilis [Eppelmann et al., 20011 and expression of griseorhodin A from an environmental Streptomyces isolate in S. Zividuns [Li and Piel, 20021) but also the successful expression in less related organisms. Some efforts have been made for the heterologous expression of the biosynthetic gene clusters from Sorungium. The biosynthetic genes for the production of epothilone, soraphen, and flaviolin from Sorungium strains were expressed in different heterologous hosts (Tang et al., 2000; Julien and Shah, 2002; Zirkle et al., 2004a; Gross et al., 2006a). In 2000, genes for epothilone biosynthesis were successfully expressed in the nonrelated streptomycete S. coelicolor (Tang et al., 2000). In this experiment, the expression of the biosynthetic genes cloned into the plasmids was driven by the act1 promoter. The transformants produced epothilones A and B and, after the deletion of the epoK gene, the epothilones C and D. The initial yield of epothilones in S. coelicolor was 50 to 100 pg/liter, which is probably due to the bacteriostatic effect of this compound in the host strain. Next, the epothilone genes were introduced stepwise from different cosmids into the chromosome of M. xunthus, assuming that this microorganism possesses all required characteristics for the heterologous biosynthesis of epothilone (Julien and Shah, 2002), as this strain is a secondary metabolite producer itself (Wenzel et al., 2006; Simunovic et al., 2006). Despite some advantages which the related myxobacterial strain offers for the biosynthesis, the yields of heterologously produced epothilones are still low in comparison to yield improvements that have been achieved by classical strain mutagenesis of S. cellulosum. One of the disadvantages in using M. xunthus is that the genes had to be integrated stepwise into the chromosome from different cosmids because of a lack of plasmids in myxobacteria. However, the fermentation process in M. xunthus could be optimized by incorporating an adsorber resin, the identification of a suitable carbon source, the supplement of trace metals, and the implementation of a fed-batch culture. The yield of the produced epothilones increased from 0.16 to 23 mg/liter (Lau et al., 2002), which is comparable to the natural producer but still far from the titer of the optimized S . cellulosum strain C18-1. Recently, the heterologous production of epothilone in E. coli has also been achieved by the researchers from Kosan Biosciences (Mutka et al., 2006). However, the synthetic epo genes had to be designed for this purpose and introduced into the engineered E. coli strain, which had to undergo several modifications to allow the expression of each single protein of the gene cluster. Moreover, the expression of the largest protein-EpoD (765 kDa)-required separation into two polypeptides. Successful epothilone

STIGMATELLAAND SORANGIUM

344 production was possible only in combination with a lower temperature, the coexpression of chaperones, and alternative promoters. Another example which sheds light onto the molecular level of the biosynthetic machinery is the heterologous production of the antifungal polyketide soraphen A of S. cellulosum So ce26 by introducing the biosynthetic genes into S. lividuns (Zirkle et al., 2004a). These genes were cloned on two integrative plasmids and on an autonomously replicating expression plasmid. In order to facilitate soraphen biosynthesis, feedings with different precursors of soraphen were required. Although the yields of heterologously produced soraphen are still low (less than 0.3 mg/liter) and the host strain needs further improvement, this system allows us to draw conclusions about the genetics of the biosynthesis and the formation of the soraphen “glycolate” polyketide extender unit of unclear biosynthetic origin (Zirkle et al., 2004a). Recent studies have shown that also pseudomonadswhich are known as producers of secondary metabolites, have a codon usage and a GC content similar to that of myxobacteria, possess an intrinsic broad-substrate Ppant transferase (Gross et al., 2005), and grow very fast in culture-represent suitable hosts for the heterologous expression of myxobacterial gene clusters (Wenzel et al., 2005). This was demonstrated recently for the expression of a type I11 PIC5 protein from s. cellulosum So ce56 (Gross et al., 2006a). As has been described above (see the section on plantlike genes above), s. cellulosum So ce56 possesses two uncommon plant-like genes which seem to encode gene products similar to chalcone synthases. Extensive analysis of the secondary metabolites from this strain did not show any products which could be correlated to these genes, as inactivation mutants did not show phenotypical properties different from those of the wild type. Under the tested laboratory conditions both genes seem to be silent. The heterologous expression of one of these genes (rppA-like gene similar to other bacterial type I11 PKS, see above), however, helped to find a novel myxobacterial metabolite, the formation of which is catalyzed by the gene product (Gross et al., 2006a). HPLC analysis as well as mass spectrometry and nuclear magnetic resonancebased structure elucidation led to the conclusion that the biosynthetic product is flaviolin, which is also responsible for the pigmentation of the liquid culture of the heterologous host P. putida (Color Plate 8). Although it has been shown that P. putidu is a suitable heterologous host for some myxobacterial biosynthetic genes (Wenzel et al., 2005; Gross et al., 2005, 2006a), the system has to be further engineered for the expression of other gene

clusters. This was shown recently for the expression of the myxobacterial secondary metabolite myxothiazol, which required the introduction of the genes for the formation of methylmalonyl-CoA as biosynthesis precursor in P. putida (Gross et al., 2006b). Alternatively, the newly described group of fastgrowing thermophilic myxobacteria, among them also Sorungium strains, might replace the slow-growing isolates in favor of a cost-saving production of myxobacterial secondary metabolites; this group of myxobacteria can probably also be used as heterologous hosts for the expression of gene clusters from slow-growing strains (Gerth and Miiller, 2005). As discussed in this chapter, the fascinating microorganisms of the genus Sorangium attract more and more attention, because they undergo a complex life cycle, possess the largest bacterial genomes known to date, and show a high potential as producers of biotechnologically important natural products. Therefore, many new techniques are being developed to make work with these microorganisms more effective. When applied to Sorungium, these new methods provide the opportunity to obtain results which were not imaginable some years ago and to access and manipulate the enormous diversity of the natural products derived from these myxobacteria. Dedicated to Prof. J. E. Peterson for his pioneering work on Sorangium. K.G. and O.P. contributed equally to this chapter.

References Bentley, S. D., I1.4 pg/pl) . . . . . . . . 5 Pl Water (RNase-free) to . . . . . . . . . . . . . . . .15.5 pl Incubate RNA and oligo pd(N), at 70°C for 10 min. Chill on ice 10 min.

cDNA synthesis Mix according to the schemes: 50X aa-dUTP mix (2:3 ratio) dATP (100 mM) ..................... dGTP (100 mM) ..................... dCTP (100 mM) ..................... dTTP (100 mM) ..................... aa-dUTP (100 mM) . . . . . . . . . . . . . . . . . . . Total

..............................

5 PJ.1 5 PI 5 PI 2 PJ.1 3 Pl 20p1

1. Add and mix 10 pl of 1N NaOH and 10 pl of 0.5 M EDTA. Note: Use fresh NaOH every time. Aliquots can be stored frozen. 2. Incubate 15 min at 65°C. Neutralize with 10 pl of 1MHCl. Note: Neutralization with 1M HCl is not usually necessary if using a Zymo column for cleanup. Using 1ml of binding buffer for the Zymo binding step should obviate the need for any neutralization step. Cleanup To continue with the amino-ally1 dye coupling procedure, all contaminants must be removed from the reaction to prevent the monofunctional NHS-ester Cy-dyes from coupling to free amine groups from the buffer or unincorporated amino-ally1groups. Note: Cleanup can be carried out with commercial spin columns and the associated buffers. Since one does not know the exact composition of these buffers, and since you want to avoid any contamination with free amines, we recommend to prepare your own washing and elution buffers. This provides for higher labeling efficiency. Phosphate buffers can be stored at 4°C for 2 weeks. Use autoclaved deionized (DI) water only (see “Materials” in “Protocol for Hybridization of AminoSilane-Coated Microarray (TIGR)” below). 1M phosphate buffer stock solution, pH 8.5

-

........................ .16.5 g ........................ .0.68 g Autoclaved DI water . . . . . . . . . . . . . . . . 100ml l00ml Total ............................ K,HPO, KH2P0,

Cocktail, no. of pl per aliquot plper 4 1 0 buffer ~ (Stratagene) . . . . . . . . . . . . 3 (13.5) 50 X aa-dUTP mix . . . . . . . . . . . . . . . . 0.6 (2.7)

DTT(0.1M) .................... Stratascript RT (50 U/pl) . . . . . . . . . . .

3(13.5) 3 (13.5) 0.5 (2.3)

RNase inhibitor (Ambion) (20 U/pl) . . Water (RNase-free) . . . . . . . . . . . . . . . 4.5 (20.2)

Total

.........................

.14.6 (65.7)

Phosphate wash buffer, pH 8.0 5 mH KPO,, 80% EtOH) Ethanol ( 9 6 % ) . . . . . . . . . . . . . . . . . . . .42.13 ml Autoclaved DI water . . . . . . . . . . . . . . . 7.63 ml 1M phosphate buffer (add dropwise). . . 0.25ml Total

...........................

50ml

28. 211. XANTHUS: EXPRESSION ANALYSIS

485

Phosphate elution buffer, pH 8.5 (4 mM KPO,)

Cleanup

Autoclaved DI water . . . . . . . . . . . . . . . .49.8 ml 1 M phosphate buffer. . . . . . . . . . . . . . . . 0.2 ml

To remove unincorporated Cy-dyes, proceed with Zymo Spin Column. Alternatively, the cleanup can be done with a QiaQuick purification kit (Qiagen). The eluate can be concentrated in speed vacuum dryer.

Total

............................

50ml

Note: Single-stranded DNA has a lower affinity to the column matrix. To reduce loss of cDNA, add at least 7 volumes of binding buffer to the sample.

1. Load 500 pl of sample into a Zymo Spin Column and place column in a 2-ml collection tube. Centrifuge at full speed for 10 s and discard flowthrough. Repeat loading with remaining sample. Centrifuge again. 2. Add 200 pl of wash buffer to column and centrifuge at full speed for 30 s. Repeat wash and centrifuge again at full speed for 1min. 3. Add 30 p1 of elution buffer to column matrix, incubate for 1min, and elute by centrifuging at full speed for 1 min into a fresh 1.5-ml tube. Note: You can store the cDNA here at -20°C. Before going on, you should measure the amount of cDNA generated by the reaction. Hybridization of too little cDNA will result in weak signals and high background levels.

Coupling This reaction is performed in bicarbonate buffer to ensure the appropriate pH for specificity and efficiency of coupling. The sodium bicarbonate can be stored frozen, but use a fresh tube every time.

1. Resuspend monofunctional NHS-ester Cy3 or Cy5 dye: If using a fresh tube of Cy3/Cy5, contents of the dye tube may be resuspended in 10 pl of DMSO and coupled immediately by transferring 2 pl of dye solution into cDNA + sodium bicarconate buffer. Note: Alternatively, resuspend the entire tube in 10 pl of DMSO, transfer 2-pl aliquots into screw-cap tubes, and immediately dry in speed vacuum dryer without heat. Freeze instantly. 2. Concentrate cDNA in speed vacuum dryer until it is almost dry (-1 1.1). Add freshly prepared 0.1 M sodium bicarbonate, p H 9.3, to a final volume of 8 pl and mix. Add 2 p1 of DMSO-dissolved dye. Incubate for 2 h in the dark or use a shaker wrapped in aluminum foil to shelter from light. Note: Adding more than 2 pl of dye is possible but usually not necessary to achieve complete labeling.

1. Add 500 pl of DNA Binding Buffer to each tube. Load sample into a Zymo Spin Column and place column in a 2-ml collection tube. Centrifuge at full speed for 10 s and discard flowthrough. 2. Add 200 p1 of Wash Buffer (from kit) to the column and centrifuge at full speed for 30 s. Repeat wash with another 200 p1 of Wash Buffer and centrifuge for 1min. Repeat washing one more time. Note: The matrix of the spin columns should now be of distinct color: blue for Cy5- and red for Cy3labeled samples. 3 . Add 30 pi of H,O to column matrix and elute by centrifuging for 10 s into a fresh 1.5-ml tube. 4. Measure and record labeling efficiency with spectrophotometer. Notes: Measure absorbance at 550 nm for Cy3 and 650 nm for Cy.5. You can also take readings at 260 nm, but have in mind that you measure modified cDNA. In our lab, we use a NanoDroplOOO spectrophotometer that allows for simultaneous measurement of all three wavelengths in a l - k l sample volume. If labeling did not work properly, go back to fix the problem or else hybridization results may not be satisfying and expensive arrays will be wasted. 5. Concentrate labeled cDNA until it is almost dry (-1 1.1) in speed vacuum dryer. Store labeled cDNA in the dark at room temperature (or for longer storage at -20°C).

Protocol for Hybridization of Amino-SilaneCoated Microarrays (TIGR) Materials 20X SSC Buffer (Invitrogen; catalog no. 15557-044) ( 1 X SSC is 0.15 M NaCl plus 0.015 M sodium citrate) 10% (wthol) SDS Solution (Roth; catalog no. 2326.2) Bovine serum albumin (BSA) (Albumin Fraction V, Roth; catalog no. 8076.2) Formamide (Roth; catalog no. 6749.1; store desiccated) 2-Propanol (isopropyl alcohol) (Roth; catalog no. 6752.2) Lifterslip cover glasses 25 X 60 (IMPLEN; catalog no. 2260 or Erie Scientific Company; 25x601-2-4789)

MYXO BACTERIAL METHOD s

486 Hybridization chamber(s) (Corning; catalog no. 2551) Millipore or Corning Brand Mini-Miser Filter-Top Tube 0.22 p,m (Fisher Scientific; catalog no. 09761-34) Sheared salmon sperm DNA (10 mg/ml) (Ambion; catalog no. 9680; store at -20°C) Surfactant-free cellulose acetate syringe filters (0.45 p,M) (Millex or W R ; catalog no. 28196-114) Glass staining dishes (Fisher catalog no. 08-812) Metal slide racks with holders Autoclaved DI water Microarray printed slides (TIGR) Cy3/CyS-labeled cDNA probes (see above; store at - 80°C) l-mi tuberculin syringe with slip tip (VWR; catalog no. BD309602) Aluminum foil

Notes: Materials provided by other suppliers should work as well but have not yet been tested in our lab. Cy dyes, particularly Cy5, are sensitive to light and oxygen radicals. At ALL times they must be kept in darkened containers, as light exposure will cause photo bleaching of the dye. W and carbon-filtered MQ water (ELGA or NANOpure Infinity W)may be used as an alternative to autoclaved DI water. However, water must be completely filtered to eliminate any residual free radicals. Free radicals compete with Cy dyes, drastically inhibiting efficient coupling and hybridization. We recommend using only autoclaved DI- or ELGA-water. Prehybridized slides must be used immediately following prehybridization to ensure optimal hybridization efficiency. Prehybridized slides cannot be stored for later use. Therefore, prepare labeled cDNA first (see above), and then proceed with prehybridization.

2 0 SSC ~ ........................ 1 0 % S D S . . ...................... DI water ........................ Total

...........................

42°C water bath Rotary shaker Centrifuge with a flat four-plate adaptor 95°C heat block Microcentrifuge tube centrifuge (Optional) Clean compressed air You will need to have or prepare the following buffers and reagents before beginning this procedure: Low-stringency wash solution (2X SSC, 0.1% SDS)

10ml 890 ml lliter

Heat to 42°C. Note: The low-stringency wash solution has a high amount of salt, which may precipitate out if left on a bench too long. Higher background may be observed if the solution is not properly mixed. To prevent the precipitation, it is recommended that the solution be kept at 42°C at all times. Medium-stringency wash solution (0.1X SSC, 0.1 % SDS) 2 0 SSC ~ ........................ 1 0 % S D S . . ...................... DI water ........................ Total

...........................

5 ml 10ml 985 ml lliter

High-stringency wash solution (0.1x SSC)

........................ ........................ ...........................

2OXSSC DI water. Total

5ml 995 ml lliter

Procedure Prehybridization (i)Prepare prehybridization solution Prehybridization solution (5x SSC, 0.1% SDS, 1% BSA) 20X SSC ........................ 10%SDS .......................

............................ ........................ Total ...........................

BSA

DI water Equipment Required You will need the following instruments for this procedure:

100 ml

125 ml 5ml

5g

330 ml 500ml

Dissolve BSA by stirring on a magnetic stirrer for at least 0.5 h. Filter the prehybridization buffer with a 0.22-pm Mini-Miser (CA) filter. Transfer the solution to a clean glass dish and preheat at 42°C for about 30 min. (ii) Prehybridize the array slide Note: It is extremely important that slides be perfectly clean, or else you will have background problems. Wear gloves. Try not to scratch or touch the printed area. Do not let the slides dry out at any time.

28. M.XANTHUS: EXPRESSION ANALYSIS 1. Place the printed slide(s) carefully and not too tightly in a slide holder and transfer it in the dish with preheated prehybridization buffer (no more than 10 slides per dish-less, if possible, to reduce scratching). 2. Incubate at 42°C for at least 1h. (Recommended: 2 h.) Occasionally sway the rack gently. (iii)Washing prehybridized slides 1. Fill a glass staining dish with autoclaved DI water for every five slides you will wash. Place a clean slide holder in every dish. 2. Using forceps, carefully grip the slides by the label and then remove the slides from the prehybridization solution into the slide holder placed in the waterfilled dish. Note: Moving the whole slide holder would carry over too much SDS and BSA. No more than five slides should be used per holder. Do not place slides in the first or last slot of the holder and do not put slides together so that they touch. Do not let the slides dry. Slides must be completely submerged in water. 3. Attach the metal handle to the slide holder and shake the rack in the water for about 10 s. 4. Place the entire staining dish apparatus on top of a rotor shaker and let shake for approximately 2 min. 5. Change the water inside the staining dish every 2 min or take the slide holder into a new dish with fresh water and let shake. 6. Continue to wash until you have replaced the wash water at least five times. 7. Empty the staining dish and fill with isopropyl alcohol. (Or transfer rack into another dish with isopropanol.) Wash the slides in isopropanol for 2 min on the rotary shaker. 8. Leave the slides in the alcohol and take them immediately to the centrifuge. (iv) Drying slides Note: DO NOT let the slides dry before putting them in the centrifuge. Allowing the slides to slowly air dry will cause a rise in background signal.

1. Take the glass slide holder with the slides out of the isopropyl alcohol and remove the metal handle from the holder. Put the slides into a centrifuge with a flat plate-holder adaptor lined with paper towels. Place another slider holder with the same number of slides as a counterbalance. Make sure the centrifuge is clean inside. Make sure the centrifuge is at room temperature before centrifuging. 2. Centrifuge the slides at -1,000 rpm for 10 min at room temperature.

487 3. Hold slides up to the light to check for any degree of streaking or spots. If any of these appear, the slides must be rewashed and respun. Hybridization (i)Making the hybridization buffer

1. Prepare the following 1X hybridization buffer fresh for every 5 slides: 1X hybridization buffer (50% formamide, 5 X SSC, 0.1% SDS, 0.6 pg salmon sperm DNA) Formamide

.......................

20xSSC.. ........................ 1 0 % SDS.. .......................

......................... ............................

DI water. Total

500 pl 250pl lop1 240 pl 1ml

2. Using a 1-ml syringe, draw up the hybridization solution and filter it through a 0.45-pin-pore-size filter. 3. Add 60 pl of salmon sperm DNA. Set the hybridization solution aside at room temperature for later use. (ii) Preparing the probe for hybridization

1. Add 22 pl of 1X hybridization buffer to the labeled probes prepared previously (Cy3Ky.5 probes: see above). Do not throw away excess hybridization solution! 2. Resuspend the probe by vortexing at very low speed, pipetting, or finger flicking for -1 min. 3. Mix the corresponding Cy3Ky.5 samples. Final volume will be -45 pl. 4. Heat the probe mixture at 95°C for 3 min. (iii)Apply labeled probe mixture t o array

1. Prepare clean coverslips for every slide: dust with compressed air or clean with 70% ethanol and a dustfree wipe. Note: Ensure that the coverslip is at least 60 mm long, as anything smaller will not cover the entire print area. If using Lifterslips, ensure that the raised side of the coverslip faces down. The lifter slip creates a space to contain the hybridization solution and ensures uniform hybridization. The side containing the lifters can be determined by the reflection of light at the edges of the slip; the lifters do not reflect light, whereas glass does. Record the barcode and the corresponding sample that you are going to hybridize for each array. 2. Place a prehybridized microarray slide (array side up) in the bottom half of a hybridization chamber. 3. Pipette the entire labeled probe mixture (-45 ~ 1 ) onto the array (printed) area of the slide. Try to keep bubbles to a minimum.

MYXO BACTERIAL METHOD s

488 4. Place the coverslip over the printed area of the slide. A pipette tip may help to place the coverslip slowly and carefully. Don’t let the probe mixture squeeze out from under the coverslip! Notes: Alternatively, first place slip carefully on slide so that the array is covered and the lifters are facing down. Then, apply the boiled probe to the array: place the pipette tip on the surface of the slide, not touching the Lifterslip, and expel a small droplet. Move the tip and the droplet slowly to one corner of the Lifterslip. Capillary action should draw the liquid under the slip. Slowly dispense more of the probe into the same corner of the Lifterslip. Going slowly and having a clean array and Lifterslip will prevent the formation of bubbles. Once the probe has fully covered the surface of the array, dispense the remaining probe at the remaining three corners. The position of the Lifterslip on the slide can be adjusted with a pipette tip. Work any large bubbles toward the edge by gently pressing the coverslip surface. Remember that the coverslips are glass and will break with rough handling.

(iv) Slide incubation 1. To the small wells at each end of the chamber, add 20 p,1 of unused hybridization solution. This step is to prevent the prehybridization solution from drying under the coverslip. 2. Place the top of the chamber on and seal it. Note: Do not tilt or flip the slide chamber once the chamber is sealed with the slide inside. 3 . Incubate in a 42°C water bath for 16 to 20 h in the dark. (v) Posthybridization washes Note: At this point all the staining dishes used should be covered in aluminum foil to prevent light exposure. Do not let the slides dry out at any time! The probe will streak the surface, causing background problems, if the slides dry. 1. Preheat the low-stringency buffer to 42°C. 2. Prepare two clean glass dishes filled with warmed low-stringency wash buffer and place an empty slide holder in one of these dishes. 3 . After the incubation in “Slide incubation” above, remove the hybridization chamber from the water bath. Remove the slides from the chamber, taking care not to disturb the coverslip. 4. To remove the coverslip, grab the slide label with forceps (or gloved fingertips if the forceps prove difficult) and submerge it in the dish without the slide holder. Keep the array level when submerging in low-stringency wash solution. Once submerged, tilt the slide. With time the coverslip will slide free of the slide surface. It may be

necessary to lightly swish slides under solution to dislodge the slip and gently remove it. Take care not to scratch the printed area with the loose coverslip! Wash the slide for 10 s and place it in the slide holder in the second dish to prevent from drying. Proceed with the next slide until five slides are in the rack. When all slides are in the rack, plunge the rack up and down 10 to 20 times. Note: Change the low-stringency buffer in the first dish for every five slides. Do not place more than five slides in the same rack. Instead, use a new dish (or fresh buffer) for every five slides. The slides should be spread out so they do not touch each other or the sides of the holder (as above). 5. Cover the dish with aluminum foil and put it on a rotary shaker. Let the arrays gently shake for 5 min. 6. After the 5 min, transfer the rack to a dish with new low-stringency buffer (preheated to 42°C) and agitate on a rotary shaker for 5 min. Repeat. 7. Transfer the slides to a slide dish with mediumstringency buffer at room temperature and agitate 5 min. Repeat. 8. After the 5 min, transfer the slides to a slide dish with high-stringency buffer at room temperature and agitate 5 min. Repeat. Note: You can continue with a quick wash in DI water or proceed directly. 9. Without removing the slides from the buffer, carry them to the centrifuge and spin them down as in “Drying slides” above.

Note: Keep the slides in the dark until they can be scanned. Scan array within hours of washing, as the Cy dyes are unstable and will degrade. When scanning, note that the top of the print area begins on the part of the slide farthest from the barcode. A properly scanned slide, therefore, should have a legible barcode at the bottom of the scanned area. Analysis In our lab we use a GenePix 4000B microarray scanner with GenePix Pro 6.0 image analysis software. Normalization and data analysis are carried out with Acuity 4.0X software package. For statistical analysis we use the SAM Software from Stanford University (http://www-stat. stanford.edu/-tibs/SAM/index.html). However, it should be noted that there are other systems available too. References Bowtell, D., and J. Sambrook (ed.). 2002. D N A Microarrays: A Molecular Cloning Manual. Cold Spring Harbor Press, Cold Spring Harbor, NY. Diodati, M. E., F. Ossa, N. B. Caberoy, I. R.Jose, W. Hiraiwa, M. M. Igo, M. Singer, and A. G. Garza. 2006. Nla18, a key regulatory protein required for normal growth and development of Myxococcus xantbus. ]. Bacteriol. 188:1733-1743.

28. M.XANTHUS: EXPRESSION ANALYSIS Jakobsen, J. S., L. Jelsbak, L. Jelsbak, R. D. Welch, C. Cummings, B. Goldman, E. Stark, S. Slater, and D. Kaiser. 2004. oS4enhancer binding proteins and Myxococcus xanthus fruiting body development. ].Bacterial. 186:43614368. Kroos, L., A. Kuspa, and D. Kaiser. 1986. A global analysis of developmentally regulated genes in Myxococcus xanthus. Dev.Biol. 117:252-266.

489 Overgaard, M., S. Wegener-Feldbriigge, and L. SsgaardAndersen. 2006. The orphan response regulator DigR is required for synthesis of extracellular matrix fibrils in Myxococcus xanthus. 1. Bacteriol. 188:4384-4394. Pham, V. D., C. W. Shebelut, I. R. Jose, D. A. Hodgson, D. E. Whitworth, and M. Singer. 2006. The response regulator PhoP4 is required for late developmental events in Myxococcus xanthus. Microbiology 152:1609-1620.

Myxobacteria: Multicellularity and Differentiation Edited by David E. Whitworth 0 2008 ASM Press, Washington, D.C.

IGmberly A. Murphy Anthony G. Garza

Genetic Tools for Studying Myxococcus xanthus Biology

Myxococcus xanthus has a remarkable repertoire of complex multicellular behaviors, including social gliding motility, predation, rippling, and fruiting body formation. There are a few genetic tools such as an autonomously replicating plasmid and a defined library of M . xanthus mutant strains that have yet to be developed to study these complex behaviors. However, because of the efforts of several prominent lab groups over the last 30 to 40 years many, if not most, of the tools available to study model organisms such as Escherichia coli or Bacillus subtilis are now available to study M . xanthus biology. This large arsenal of genetic tools has allowed research on 111. xanthus to expand beyond single gene studies; we can now examine the relationships between multiple genes in the same genetic pathway and the interactions between different genetic pathways. This chapter contains a description of M. xanthus genetic tools and their practical applications.

GENERALIZED TRANSDUCTION Generalized transducing particles form when by mistake a bacteriophage head or capsid assembles around a fragment of a donor bacterium's chromosomal DNA or

29

around a plasmid instead of the phage genome. When these particles infect a recipient host, the donor bacterium's DNA is inserted into the recipient host and is free to undergo homologous recombination with the host cell chromosome.

Mx4 and Mx8 Transducing Phages In M. xanthus, generalized transducing phages are typically used for genetic mapping and for strain constructions. Generalized transduction between 111. xanthus strains is accomplished with lytic phage Mx4 (Campos et al., 1978; Geisselsoder et al., 1978) or the temperate phage Mx8 (Martin et al., 1978). Mx4 and Mx8 can package about 50 kb of DNA (Kaiser, 1984), which corresponds to about 0.5 % of the M . xanthus genome (Goldman et al., 2006). A temperature-sensitive derivative of Mx4 phage (Mx4 ts-27htf hrm-1)that transduces at high frequency in M . xanthus is commonly used (Geisselsoder et al., 1978). Because Mx4 ts-27htf hrm-l phage lysates prepared at 25°C do not replicate when introduced into an M . xanthus strain grown at 32"C, this phage is well suited for use in generalized transductions. In the case of phage Mx8, the clear plaque derivative Mx8 clp2 is often used for generalized transduction (Martin et al., 1978).

Kimberly A. Murphy and Anthony G. Garza, Department of Biology, Syracuse University, 130 College Place, Syracuse, NY 13244-1220.

49 1

-~

MYXOBACTERIAL METHODS

492

Genetic Mapping Generalized transducing phages such as Mx4 and Mx8 have been used to identify the sites of chromosomal point mutations by genetically linking them to antibiotic resistance markers. When two genetic markers are sufficiently close on the chromosome, a single transducing phage can transfer both genetic markers from the donor cell to the recipient cell. Genes that are transferred together to the recipient cell are referred to as coinherited genes. Typically, the phenotype associated with one genetic marker is selected (e.g., kanamycin resistance), and transductants that inherit this marker will be screened for the phenotype associated with inheritance of the second genetic marker (e.g., a defect in gliding motility). Cotransduction frequency is the number of transductants that coinherit both genetic markers divided by the total number of transductants. By applying the Wu equation (Wu, 1966), cotransduction frequency can be used to approximate the physical distance between two chromosomal markers. cotransduction frequency

=

[l- (distance between two markedlength of transduced DNA)I3

Thus, the shorter the distance between two chromosomal markers, the more frequently they will be coinherited by. generalized transduction. However, a limitation of genetic mapping by cotransduction frequency is that the two markers must be in close enough physical proximity to be packaged in the same phage particle. In addition, this approach for genetic mapping requires that a selection exists for at least one of the markers. Typically, one marker imparts antibiotic resistance to the bacterial cells that carry it. To facilitate genetic mapping in M . xanthus, Kuner and Kaiser (1981) developed a procedure to link chromosomal mutations to the selectable marker Tn5, a transposable genetic element that confers resistance to kanamycin (Fig. 1).To do this, a library of independent Tn5 insertions in the M. xunthus chromosome is generated by shuttling Tn5 from E. coli to M. xunthus by using phage P1. Mx4 or Mx8 phage are grown on a pool of strains from the M. xanthus insertion library, and the phage stocks are used to transduce the Tn5 insertions from the pooled cells into the appropriate mutant strain. Kanamycin-resistant transductants are then screened for

Motility- mutant

Motility- mutanl

m \

P1::TnS

Wild-type M xanthus strain

Library of independent Tn5 insertions in M xanthus chromosome

Grow Mx4 or Mx8 on strains from the library

I

Transduce motility- strain using Mx phage

I0/Mot+

Select for Kad, screen for gliding motility

\

or Mx8 on Mot' strain

Transduce motility- strain using Mx phage

1 0I

Mot'

Select for Kan', screen for gliding motility

Figure 1 Procedure for linking chromosomal mutations to Tn5 . The diagram shows a Tn5 insertion being genetically linked to a mutation (*) that causes a gliding motility defect (Mot-), but it is applicable to any chromosomal mutation that produces a distinguishable phenotype. Open rectangles represent M . xanthus cells, and the ovals inside the rectangles represent the M. xanthus chromosome. The small, shaded rectangles denote Tn5 chromosomal insertions that impart kanamycin resistance to recipient cells. Hexagons represent the indicated phages.

29. GENETIC TOOLS FOR STUDYINGM.XANTHUS coinheritance of the wild-type allele of the mutation of interest. To determine the genetic linkage between the mutation of interest and Tn.5, an Mx4 or Mx8 phage stock is generated from kanamycin-resistant transductants that carry the wild-type allele of the mutation, the phage stock is used to transduce the original mutant strain, and kanamycin-resistant transductants are screened for coinheritance of the wild-type allele (Fig. 1). Using a Tn5 library such as the one generated by Kuner and Kaiser (1981), it is possible, for example, to determine whether point mutations that produce similar phenotypes are near a particular Tn.5 marker; close physical proximity to the same Tn5 insertion implies that the mutations are in the same genetic locus in the chromosome. Furthermore, it is possible to identify the site of a mutation because close proximity to Tn5 and its kanamycin resistance cassette facilitates cloning, which in turn allows the region of DNA carrying the mutation to be subjected to DNA sequence analysis. By comparing this DNA sequence to that of the corresponding wildtype region in the M. xanthus genome sequence, the site of the point mutation can be identified. Recently, magellan-4, a transposon that confers kanamycin resistance, was used to create M. xanthus insertion libraries (Youderian et al., 2003; Youderian and Hartzell, 2006; Chavira et al., 2006). The magellan-4 transposon recognizes a TA dinucleotide (Plasterk et al., 1999) and therefore has the capacity to integrate into a large number of sites on the chromosome. It also has features that make cloning chromosomal DNA flanking the insertion site relatively easy (see below). Hence, the magellan-4 transposon library should greatly facilitate the genetic mapping and identification of chromosomal point mutations in M. xanthus.

Strain Construction Generalized transduction has been used extensively to place mutant alleles in different genetic backgrounds, which has provided functional information about the genes carrying the mutations. For example, M . xanthus gliding motility is controlled by the A-motility and S-motility systems (Hodgkin and Kaiser, 1979a, 1979b), and it is common practice to use a transducing phage to transfer an uncharacterized motility mutation (e.g., a Tn.5 insertion mutation) from a donor strain into A-motility and S-motility mutant recipient strains. Since gliding motility is completely abolished only when both motility systems are inactivated, one can determine whether the insertion mutation in the donor strain affects the A-motility or the S-motility system by screening the kanamycin-resistant transductants, which are double mutants, for a complete loss

493 of gliding motility. Similarly, to characterize mutations that affect fruiting body development it has been common practice to transduce developmentally regulated TnSlacZ reporter gene fusions into the strains carrying the mutations. By comparing the lac2 expression profiles of the mutant strains to those of wild-type cells, the approximate time when the developmental program goes awry in the mutants is revealed. This has been a particularly useful strategy for dissecting the genetic pathways that control the process of fruiting body development.

TRANSFER OF PLASMID DNA TO M . XANTHUS Methods for transferring plasmid DNA from E. coli to M.xanthus using phage P1 have been developed (O’Connor and Zusman, 1983; Gill et al., 1988). These methods take advantage of the fact that P l y which is an E. coli phage, is unable to multiply in M. xanthus (Kuner and Kaiseq 1981),but in some instances it is able to package plasmid DNA into phage particles and introduce them into M. xanthus cells. Another method of transferringplasmid DNA to M. xanthus cells is electroporation, which is by far the most common method used today (Kashefi and Hartzell, 1995). Because there are no plasmids available that replicate autonomously in M. xanthus, plasmids are propagated in, and isolated from, E. coli before they are introduced into M. xanthus cells. Hence, they must contain E. coli origins of replication. These E. colibased plasmids are modified in a variety of ways for use in M. xanthus research. However, the one thing that all of the plasmids have in common is that once they are inside M. xanthus they must integrate into the chromosome in order to be passed onto progeny cells. To promote sitespecific integration in the M. xanthus chromosome, some plasmids used in M. xanthus research carry an Mx phage attachment (att)site and a phage integrase (int)gene. For complementation or expression studies, which are the type of analyses typically done with these Mx att plasmids, the gene or genes of interest and their corresponding promoter elements are ligated into the plasmids. The plasmids are introduced into M. xanthus cells, strains that carry the plasmid integrated into phage attachment sites in the chromosome are identified, and the appropriate assays are performed. Other plasmids contain M. xanthus DNA that allows them to integrate into the chromosome via homologous recombination. For example, an internal fragment of an M. xanthus gene can be placed into an E. coli plasmid in order to inactivate the chromosomal copy of the gene, a procedure that is described in more detail below.

MYXO BACTERIAL METHOD s

494

PLASMID INTEGRATION AT PHAGE ATTACHMENT SITES In M . xanthus, phage attachment sites are often used for genetic complementation experiments; a wild-type copy of a gene and its native promoter are introduced into the phage attachment site of a strain carrying a mutation in that gene. Phage attachment sites have also been used for mutational analysis of M. xanthus genes and promoters. In M. xanthus, site-specific integration of Mx8 occurs between attP, the preferred site on the Mx8 genome, and attB, the preferred site on the M. xanthus chromosome (Stellwag et al., 1985), which is shown in Fig. 2. Within the Mx8 genome, a minimal 2.2-kb fragment has been defined that contains the attP site and the integrase gene int required for integration at attB (Magrini et al., 1999). Several plasmids that contain the attP site and the int gene have been constructed. When M. xanthus genes are introduced into these plasmids, they seem to have a high degree of preference for integration into the chromosomal attB site. For example, Tojo et al. (1993)

generated a plasmid carrying the Mx8 attP site and int gene, as well as a 1.3-kb fragment of the M. xanthus ZonD gene. When this plasmid was transferred to M. xanthus cells, the frequency of site-specific recombination at the chromosomal attB site was about 300-fold higher than the frequency of homologous recombination at the native lonD locus in the chromosome (Tojo and Komano, 2003). In some cases, expression of M. xanthus genes and their native promoters at the Mx8 attB site is relatively low compared to expression at the native locus on the chromosome (Li and Shimkets, 1988; Fisseha et al., 1996; Whitworth et al., 2004). Hence, using the attB site for complementation analysis or other analyses in which gene expression levels are important may be problematic. The advantage of using the Mx8 attB site or another phage attachment site in the chromosome for complementation or expression studies, for example, is that only a single copy of the gene and promoter of interest is provided to M. xanthus cells, which alleviates potential concerns about multicopy effects. Recently, Julien (2003) generated plasmid vectors for integration into the Mx9 phage attachment site in the M . xanthus chromosome. The integration gene int, phage attachment site, and site of integration in the M. xanthus chromosome were identified and characterized. Mx9 integrates into the M. xanthus genome by site-specific recombination at one of two sites, attB1 or attB2, and requires a single protein, the product of the int gene. Julien (2003) found that integration of plasmids containing piZA or mgZ promoter fusions to lac2 at either Mx9 attB site is expressed at higher levels than the same fusions integrated at the Mx8 attB site, suggesting that the Mx9 attB2 and attB2 sites may be better suited for complementation and expression studies in some cases.

RANDOM MUTAGENESIS

attL

attR

Figure 2 Site-specific recombination at a chromosomal Mx8 phage attachment site. A plasmid carrying a gene conferring kanamycin resistanc (Kan'), the Mx8 phage attachment site attP (gray box) located within the coding sequence for the integrase gene int (stippled box), and the locus of interest (cross-hatched box) is shown. The plasmid integrates into the M. xanthus chromosomal Mx8 phage attachment site via sitespecific recombination between the attP and attB sites. The integration event produces the indicated DNA arrangement in the chromosome.

In M. xanthus, random mutations have been generated using conventional chemical mutagens such as nitrosoguanidine and ethyl methanesulfonate, UV irradiation, and insertion mutagenesis. The advantage of using chemical mutagens and W irradiation is that they produce single-base-pair changes. Therefore, conditional mutations or mutations in essential genes that only partially disrupt the function of their corresponding protein products can be generated. Although these types of mutations are difficult to generate using transposon or random plasmid insertions, insertions are particularly useful for generating a large pool of mutations that inactivate chromosomal genes and for tagging mutated genes so that they can be identified. Below is a brief description of transposable elements that are commonly used to

29. GENETIC TOOLSFOR STUDYING M.XANTHUS make insertions in M. xanthus, as well as a method for making random plasmid insertions.

T n 5 and Derivatives In M. xanthus, Tn5, a transposable genetic element that confers resistance to kanamycin, can be used to generate a library of insertions in the M. xanthus chromosome or to link a particular locus to an antibiotic resistance marker for genetic mapping or for cloning. Tn5 is introduced into M. xanthus by the E. coli transducing phage P1::TnS (Kuner and Kaiser, 1981). In M. xanthus this virus serves as a suicide vector; P1::TnS is unable to replicate inside M. xanthus cells, but kanamycin-resistant transductants are formed when Tn5 transposes into the M . xanthus chromosome. Once Tn5 integrates into the chromosome of M. xanthus cells, the frequency of transposition into new sites on the chromosome appears to be about per cell (Avery and Kaiser, 1983; Sodergren and Kaiser, 1983). Hence, P1::TnS generates mutants with relatively stable chromosomal insertions. This property and the fact that P1::TnS tends to generate mutants that carry only a single Tn5 insertion help prevent complications when trying to identify the mutated gene that is responsible for the desired phenotype.

495 Tn5-Tp (Sasakawa and Yoshikawa, 1987) and Tn.5132 (Rothstein et al., 1981; Avery and Kaiser, 1983) are derivatives of Tn5 that retain most of the right and left IS Tn5O elements, but in the central region the kanamycin resistance module has been replaced with genes conferring resistance to trimethoprim and tetracycline, respectively (Fig. 3A). For the most part, these elements have been used to replace the kanamycin resistance cassettes in existing Tn5 insertions with different selectable markers. To do this, the strain carrying the Tn5 insertion is infected with P1::TnS-Tp or Pl::Tn5-132 and transductants resistant to either trimethoprim or tetracycline are identified. If a double homologous recombination event occurs between the IS50 sequences in Tn5-Tp or Tn5-132 and the IS50 sequences in the resident Tn5 element in the chromosome, then the kanamycin resistance cassette of the resident Tn5 element will be replaced by tetracycline or trimethoprim modules. The resulting strain has a single transposon insertion at the same chromosomal location as the original Tn5, but the new Tn5 element will confer resistance to tetracycline or trimethoprim instead of kanamycin. With a tetracycline or trimethoprim marker tagging the insertion of interest, it is then possible to introduce a new insertion that

A. IS50,

IS50,

B. I

I

IS50,

IS50,

I

IS50,

IS50,

I

IS50,I

C. I

IS50,l

I

phoA Figure 3 Structures of Tn5 and its derivatives. (A) Tn5 containing the genes that confer resistance to kanamycin (Kan'), tetracycline (Tet'), or trimethoprim (Tmp') flanked by IS50 elements. (B) A derivative of Tn5 used to create lac2 transcriptional fusions. In this Tn5 derivative, a segment of the ISSO, element is replaced with a promoterless lucZ reporter gene. (C) A derivative of Tn5 used to create PhoA translational fusions. A segment of the ISSO, element is replaced with a fragment of the phoA gene. The phoA fragment lacks sequences corresponding to the PhoA translational start site and sequences needed for export of the PhoA protein across the cytoplasmic membrane.

496 confers kanamycin resistance into this strain. Having the flexibility to place two different insertions in the same strain would, for example, provide an opportunity to do genetic epistasis experiments or to genetically map the locations of the insertions relative to one another.

magellan-4 In M . xanthus studies, plasmid pMycoMar (Rubin et al., 1999) has been used as a donor for the transposition of magellan-4, a minitransposon derived from the mariner element Himarl. magellan-4 contains the npt gene from Tn5, which confers resistance to kanamycin, flanked by the ends of Himarl. Adjacent to the magellan-4 element in plasmid pMycoMar is the Himarl transposase gene, which is under control of the mycobacerial T6 promoter (Barsom and Hatfull, 1996). This mycobacterial promoter seems to function well in M . xanthus; transposition of magellan-4 in M. xanthus cells is efficient after plasmid pMycoMar has been introduced by electroporation (Youderian et al., 2003). Plasmid pMycoMar cannot replicate autonomously in M . xanthus. However, expression of the plasmid-borne transposase promotes transposition of the magellan-4 element into the chromosome. Presumably, transposition occurs by a “cut and paste” mechanism (Lampe et al., 1996) that does not require any M . xanthus proteins. To identify the chromosomal site in which magellan-4 has inserted, magellan-4 and flanking M . xanthus DNA can be circularized by digesting and ligating the chromosomal DNA of an insertion mutant. The circularized DNA is transferred and propagated in E. coli (magellan-4 contains an origin for replication in E. coli) and then isolated and subjected to DNA sequence analysis. In recent studies, magellan-4 was used to identify new genes for A-motility and S-motility. When placed on a solid surface, the A-motility system controls individual cell movements, while S-motility controls the movements of multicellular groups (Hodgkin and Kaiser, 1979a, 1979b).Using magellan-4, Youderian et al. (2003)implicated approximately 30 new genes in A-motility. Similarly, Youderian and Hartzell (2006) implicated about 30 new genes in S-motility. In both studies, the desired motility mutants were found with a frequency of about 1%, which is fivefold higher than the frequency obtained through similar approaches with transposon Tn.5 (Youderian et al., 2003; Youderian and Hartzell, 2006). Therefore, magellan-4 seems to have a much broader spectrum of target sites in M. xanthus than does Tn.5.

Plasmid Insertion Libraries Although M. xanthus insertion libraries have typically been created using site-specific transposable elements, they can also be created by homologous recombination

MYXO BACTERIAL METHOD s of random cloned fragments of M . xanthus DNA. For example, Cho and Zusman (1999a) generated a plasmid library containing approximately 500-bp random M . xanthus DNA fragments. When electroporated into wild-type M. xanthus cells, these Kan‘ plasmids were unable to replicate autonomously, but they did integrate into the chromosome via homologous recombination of the cloned M . xanthus DNA. Since the SOO-bp fragments of cloned DNA were smaller than most M. xanthus genes, many of the Kan‘ transformants contained inactivated copies of chromosomal genes. Thus, Cho and Zusman (1999a)were able to generate a random pool of M. xanthus insertion mutants. Phenotypic characterization of these insertion mutants led to the discovery of the espAB locus, which seems to control the timing of M. xanthus sporulation, and the asgD locus, a locus involved in production of an M. xanthus cell density signal called Asignal and nutrient sensing (Cho and Zusman, 1999a, 1999b). In a recent study, Lu et al. (2005)screened the random plasmid insertion library generated by Cho and Zusman (1999a)for M . xanthus mutants with defects in exopolysaccharide (EPS) production, which is required for S-motility, and identified two genetic regions essential for EPS biogenesis: the EPS synthesis (eps)region and the EPS-associated (eas)region.

DIRECTED MUTAGENESIS Targeted Plasmid Insertions In recent years, targeted plasmid insertions have been used to inactivate a iarietyof M . xanthus genes. To make a targeted plasmid insertion, a partial fragment of the target gene is generated using PCR, the PCR fragment is cloned into a suicide plasmid that carries an appropriate antibiotic resistance gene, and the plasmid is electroporated into wild-type M . xanthus cells. Although the plasmid is incapable of autonomous replication in M . xanthus, it can integrate into the chromosomal copy of the target gene by homologous recombination of the cloned PCR fragment. A single crossover yields kanamycin-resistant electroporants with two incomplete copies of the gene separated by vector DNA (Fig. 4).Therefore, the plasmid insertion is likely to inactivate the gene and its corresponding protein product. Homologous recombination between plasmid-borne gene fragments and genes that reside in the M. xanthus chromosome, which is essential for inactivation of chromosomal genes and hence for this technique to work, appears to be constrained by the size of the plasmidborne fragment. For example, in a recent study we used this technique to inactivate 28 ntrC-like activator (nla)

29. GENETIC TOOLSFOR STUDYING211.

497

XANTHUS

rn plasmid

I 5' end target gene

3' end target gene

Figure 4 Strategy for making targeted insertions. An internal fragment of the gene of interest is generated using PCR, the PCR fragment is cloned into a plasmid vector that confers resistance to an antibiotic such as kanamycin, and plasmid DNA is electroporated into wildtype M. xanthus cells. A single homologous crossover produces a tandem duplication of the internal fragment and incorporation of the vector into the chromosomal copy of the gene. The likely result of the crossover is an inactivated copy of the gene.

genes (Caberoy et al., 2003). We were frequently unable to generate insertions when the plasmid-borne n h gene fragments had about 350 bp of homology with their chromosomal nlu gene targets. When we increased the region of homology to about 500 bp, most of plasmid-borne nla fragments that we tested yielded insertions, and when we increased the region of homology to about 600 bp, all of the plasmid-borne nlu fragments that we tested produced insertions. In some studies, we have been able to generate insertions using plasmid-borne fragments as small as 250 bp. However, we were able to generate the insertion mutants only when we used a relatively high concentration of plasmid DNA for the electroporation and when we allowed electroporated cells to recover from electric shock in rich medium for 12 to 24 h instead of 4 to 8 h, which is standard procedure. With the availability of the M. xanthus genome sequence (Goldman et al., 2006) (GenBank accession number CP000113), it is now possible to inactivate whole gene families using targeted plasmid insertions. For example, the M. xanthus genome sequence revealed 53 genes encoding NtrC-like proteins, transcriptional activators that participate in a variety of adaptive responses in bacteria. Using information from the

M. xunthus genome sequence and the targeted plasmid insertion strategy, almost all of the 41 uncharacterized ntrC-like genes have been inactivated and characterized in the past 5 years (Caberoy et al., 2003; Jakobsen et al., 2004; Jelsbak et al., 2005). These studies led to the discovery of 10 NtrC-like proteins that are required for fruiting body development to progress normally.

In-Frame Deletions Insertions have the potential to block transcription of genes located downstream of the insertion sites via polar effects. Therefore, it is often difficult to determine whether the phenotype caused by an insertion is due to inactivation of the gene that carries the insertion or due to the effect the insertion has on downstream transcription. This problem is eliminated by constructing a strain that carries an in-frame deletion of the gene of interest. An M. xunthus mutant containing an in-frame deletion in the target gene can be constructed by gene replacement using plasmids that provide a gene for positive screening such as the kanamycin resistance gene, and a gene for negative screening such as sucB (Ried and Collmer, 1987; Wu and Kaiser, 1996) or gulK (Ueki et al., 1996). Although both sacB and gulK work for negative

MYXOBACTERIAL METHODS

498 screening, the galK gene has been used more frequently for construction of in-frame deletions in M. xanthus. To generate the appropriate plasmid-borne in-frame deletion, a reverse PCR on a plasmid containing genes for both positive (kanamycin resistance) and negative (sacB or galK) selection, as well as the target gene and flanking M. xanthus DNA, can be performed. Typically, the primers used for the PCR correspond to the 5’ end and 3’ end of the target gene and they have restriction sites engineered into their 5’ tails. The restriction sites engineered into the primers are unique sites not found in the plasmid template, allowing a plasmid containing an in-frame deletion of the target gene to be generated by digesting the PCR product with the appropriate restriction enzyme and then ligating the digested DNA. The inverse PCRs are usually performed with mixtures containing Taq and high-fidelity DNA polymerases, the plasmid template and primers carrying the appropriate restriction sites. An alternative strategy to reverse PCR is now possible with the availability of the M. xanthus genome sequence (Goldman et al., 2006) (GenBank accession number CP000113). In this approach standard protocols are used to generate two PCR products, one corresponding to DNA upstream of the gene to be deleted and the other corresponding to DNA downstream of the gene to be deleted. Because the primers are designed to produce upstream and downstream PCR products with compatible restriction sites a t their 3‘ and 5‘ ends, respectively, the PCR products can be digested with the appropriate restriction enzyme and then ligated together to create a DNA fragment carrying a deletion of the target gene. The primers can also be designed to yield an upstream PCR product with a restriction site at its 5’ end and a downstream PCR product with a restriction site at its 3’ end, which allows the deletion-carrying DNA fragment to be cloned into a positive-negative selection vector. After introducing the in-frame deletion plasmid into M. xanthus cells by electroporation, strains with the plasmid integrated in the chromosomal copy of the target gene locus by homologous recombination are identified by selecting for resistance to kanamycin. Strains carrying the integrated plasmid are sensitive to galactose because of the plasmid-borne galK gene or to sucrose because of the plasmid-borne sacB gene, depending on the positivenegative selection vector that is used. Strains in which a plasmid excision event has occurred are identified by their resistance to galactose or sucrose and their sensitivity to kanamycin. In some cases, a plasmid excision event gives rise to a strain that has an in-frame deletion of the target gene.

GENE FUSIONS Gene fusions have been used to assay the transcriptional/ translational regulation of genes/proteins and to examine protein localization in bacterial cells (Silhavy and Beckwith, 1985). Fusions belong to one of two classes: transcriptional fusions or translational fusions. In the case of a transcriptional fusion, the reporter gene lacks a promoter, but it possesses a functional ribosome-binding site. Hence, the reporter gene is transcribed with the endogenous gene, but the reporter gene product is not fused to the product of the endogenous gene. In the case of a translational fusion, the reporter gene lacks a promoter and a functional ribosome-binding site. Therefore, a fusion is created between the products of the reporter gene and of the endogenous gene.

Transcriptional Fusions: TnSZuc In M. xanthus, TnSlac (Kroos and Kaiser, 1984), a derivative of Tn5 that contains a promoterless lacZ reporter gene, has been used to generate transcriptional fusions (Fig. 3B). TnSlac is introduced into M. xanthus cells using phage P1::TnSlac (Kroos and Kaiser, 1984). Once inside an M . xanthus cell, TnSlac can transpose into a variety of sites in the chromosome. If the promoterless lacZ gene in TnSlac is in the same orientation as the gene in which it has inserted, then it will place lacZ under transcriptional control of the promoter for that gene. Using TnSlac, Kroos et al. (1986) generated a collection of developmentally regulated lac2 reporter fusions by screening for increases in P-galactosidase production during fruiting body formation. This collection of lacZ reporters has been used extensively to characterize M. xanthus developmental mutants. By introducing a representative panel of these lacZ fusions into a developmental mutant and monitoring lac2 expression, the approximate time that the developmental program goes awry in the mutant can be determined. Kroos et al. (1990) also used this collection of TnSlac fusions, which create insertion mutations, to identify M. xanthus mutants with severe developmental defects.

Translational Fusions: TnphoA TnphoA (Fig. 3C) is a version of the Tn.5 transposon with the phoA gene from E. coli inserted at its left end (Manoil and Beckwith, 1985). The product of the phoA gene is the enzyme alkaline phosphatase, which must be exported to the periplasm or beyond to be active (Michaelis et al., 1983; Hoffman and Wright, 1985). All sequences needed for transcription, translation, and export have been deleted from the phoA gene in TnphoA. When TnphoA transposes into the chromosomal copy of a gene in the appropriate way, a fusion between PhoA

29. GENETIC TOOLSFOR STUDYING211.

499

XANTHUS

and the product of the gene is created. If the protein to which PhoA is fused contains sequences that allow PhoA to be exported, then alkaline phosphatase activity will be detected. Hence, TnphoA is an excellent probe to identify endogenous exported proteins. TnphoA was used by Kalos and Zissler (1990) to detect genes whose products are involved in cell interactions. Three of these TnphoA mutants carried insertions in the cgl locus, a locus that is important for the normal function of the A-motility system that controls individual cell gliding (Hodgkin and Kaiser, 1977, 1979a). Subsequent work has confirmed that one of the products of the cgl locus, CglB, is localized to the outer membrane (Rodriguez and Spormann, 1999; Simunovic et al., 2003). Although Kalos and Zissler (1990) used TnphoA fusions successfully in their study, TnphoA usage has gained little traction in subsequent years, which perhaps is due to the fact that M. xanthus strains have multiple Pho systems and hence a high background of alkaline phosphate activity.

INDUCIBLE PROMOTER SYSTEMS As the name implies, inducible promoter systems allow gene transcription to be induced under the desired circumstances. They are particularly useful tools when trying to evaluate the effects of expressing relatively low levels or relatively high levels of a particular gene under a particular condition. At this time, the availability of inducible promoter systems to study M. xanthus is limited. However, given that data from DNA microarraybased promotedgene expression studies and molecular studies of M. xanthus promoter elements are beginning to accumulate, the number of inducible promoter systems is likely to increase in the near future. One inducible promoter system that has been used in several studies is the M. xanthus light-inducible carQRS system. When growing cultures of M. xanthus cells are kept in the dark, the carQRS promoter is almost completely inactive (Letouvet-Pawlek et al., 1990; Hodgson, 1993). In contrast, exposing cells to blue light causes the carQRS promoter to become highly active. Hence, under conditions of vegetative growth carQRS is an ideal inducible promoter system. However, the blue light needed to induce expression from the carQRS promoter inhibits the process of fruiting body development (Li et al., 1992). Therefore, for the most part use of this promoter system has been restricted to studies involving vegetatively growing cells. Furthermore, it has been reported that light induction of the carQRS promoter at the Mx8 phage attachment site attB in the chromosome increases activity only about 6-fold, while light induction

of this promoter at the native carQRS locus on the chromosome increases activity about 100-fold (Whitworth et al., 2004). Thus, induction of the carQRS promoter at the Mx8 phage attachment site in the chromosome appears to be attenuated. Another inducible promoter system that has been used in M. xanthus is the tac promoter from E. coli (Letouvet-Pawlek et al., 1990), a hybrid consisting of parts of the trp promoter and lac promoter/operator. In E. coli, transcription from the tac promoter is repressed when the product of the lac1 gene binds to the lac operator and derepressed when cells are provided with isopropyl-p-D-thiogalactopyranoside(IPTG), which binds to and inactivates LacI. When the trp promoter and the lac1 gene were introduced into M. xanthus and the cells were grown vegetatively, there was a low background level of activity from the tac promoter. With the addition of IPTG the activity of the tac promoter in growing cultures increased about eightfold. The activity of the tac promoter was not tested under the starvation conditions that induce fruiting body development. However, in a recent study Jelsbak and Kaiser (ZOOS) generated a system in which a native M. xanthus promoter element can be induced during fruiting body development by adding IPTG. In wild-type cells, the promoter for pilA, a gene that codes for the external structural subunit of type IV pili, is active during vegetative growth and during fruiting body development. For this study, the leader sequence of chromosomal copy of pilA was replaced with the lac operator and a ribosomal binding site from E. coli. In addition, the lac1 gene was placed under control of the pilA promoter and then introduced into the Mx8 phage attB site in the chromosome. When no IPTG was provided to cells carrying this promoter system, expression of pilA increased about twofold during development, indicating that LacI was not able to fully repress pilA expression. When IPTG was provided, pilA expression in cells carrying these promoter constructs increased about sixfold during development. Thus, while this system is not optimal because of the background expression in the absence of IPTG, it does provide a way to generate relatively high levels of gene expression in developing cells.

References Avery, L., and D. Kaiser. 1983. In situ transposon replacement and isolation of a spontaneous tandem duplication. Mol. Gen. Genet. 191:99-109. Barsom, E. K., and G. F. Hatfull. 1996. Characterization of Mycobacterium smegmatis gene that confers resistance to phages L5 and D29 when overexpressed. Mol. Microbial. 2 1:159-1 70.

500 Caberoy, N. B., R. D. Welch, J. S. Jakobsen, S. C. Slater, and A. G. Garza. 2003. Global mutational analysis of NtrC-like activators in Myxococcus xanthus: identifying activator mutants defective for motility and fruiting body development. ]. Bacteriol. 185:6083-6094. Campos, J. M., J. Geisselsoder, and D. R. Zusman. 1978. Isolation of bacteriophage Mx4, a generalized transducing phage for Myxococcus xanthus. J. Mol. Biol. 119:167-178. Chavira, M., N. Cao, K. Le, T. Riar, N. Moradshahi, M. McBride, R. Lux, and W. Shi. 20 October 2006. PD-Allose inhibits fruiting body formation and sporulation in Myxococcus xanthus. J. Bacteriol. doi:l0.1128/JB.00792-06. Cho, K., and D. R. Zusman. 1999a. Sporulation timing in Myxococcus xanthus is controlled by the espAB locus. Mol. Microbiol. 34:7 14-725. Cho, K., and D. R. Zusman. 199913. AsgD, a new two-component regulator required for A-signalling and nutrient sensing during early development of Myxococcus xanthus. Mol. Microbiol. 34:268-281. Fisseha, M., M. Gloudemans, R. E. Gill, and L. Kroos. 1996. Characterization of the regulatory region of a cell interactiondependent gene in Myxococcus xanthus. ]. Bacteriol. 178: 2539-2550. Geisselsoder,J., J. M. Campos, and D. R. Zusman. 1978. Physical characterization of bacteriophage Mx4, a generalized transducing phage for Myxococcus xanthus. J. Mol. Biol. 119~179-189. Gill, R. E., M. G. Cull, and S. Fly. 1988. Genetic identification and cloning of a gene required for developmental cell interactions in Myxococcus xanthus. J. Bacteriol. 1705279-5288. Goldman, B. S., W. C. Nierman, D. Kaiser, S. C. Slater,A. S. Durkin, J. Eisen, C. M. Ronning, W. B. Barbazuk, M. Blanchard, C. Field, C. Halling, G. Hinkle, 0. Iartchuk, H. S. Kim, C. Mackenzie, R. Madupu, N. Miller, A. Shvartsbeyn, S. A. Sullivan, M. Vaudin, R. Wiegand, and H. B.Kaplan. 2006. Evolution of sensory complexity recorded in a myxobacterial genome. Proc. Natl. Acad. Sci. USA 103:15200-15205. Hodgkin, J., and D. Kaiser. 1977. Cell-to-cell stimulation of movement in nonmotile mutants of Myxococcus. Proc. Natl. Acad. Sci. USA 74:2938-2942. Hodgkin, J., and D. Kaiser. 1979a. Genetics of gliding motility in Myxococcus xanthus (Myxobacterales): genes controlling movement of single cells. Mol. Gen. Genet. 171:167-176. Hodgkin, J., and D. Kaiser. 1979b. Genetics of gliding motility in Myxococcus xanthus: two gene systems control movement. Mol. Gen. Genet. 171:177-191. Hodgson, D. A. 1993. Light-induced carotenogenesis in Myxococcus xanthus: genetic analysis of the carR region. Mol. Microbiol. 7:471-48 8. Hoffman, C., and A. Wright. 1985. Fusions of secreted proteins to alkaline phosphatase: an approach for studying protein secretion. Proc. Natl. Acad. Sci. USA 825107-5111. Jakobsen, J. S., L. Jelsbak, L. Jelsbak, R. D. Welch, C. Cummings, B. Goldman, E. Stark, S. Slater, and D. Kaiser. 2004. d4enhancer binding proteins and Myxococcus xanthus fruiting body development. J. Bacteriol. 186:4361-4368. Jelsbak, L., and D. Kaiser. 2005. Regulating pilin expression reveals a threshold for S motility in Myxococcus xanthus. J. Bacteriol. 1872105-2112.

MYXO BACTERIAL METHOD s Jelsbak, L., M. Givskov, and D. Kaiser. 2005. Enhancer-binding proteins with a forkhead-associated domain and the d4 regulon in Myxococcus xanthus fruiting body development. Proc. Natl. Acad. Sci. USA 102:3010-3015. Julien, B. 2003. Characterization of the integrase gene and attachment site for the Myxococcus xanthus bacteriophage Mx9. J. Bacteriol. 185:6325-6330. Kaiser, D. 1984. Genetics of Myxobacteria, p. 163-184. In E. Rosenberg (ed.),Myxobacteria: Development and Cell Interactions. Springer-Verlag, New York, NY. Kalos, M., and J. Zissler. 1990. Transposon tagging of genes for cell-to-cell signaling in Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 878316-8320. Kashefi, K., and P. L. Hartzell. 1995. Genetic suppression and phenotypic masking of a Myxococcus xanthus fat;-defect. Mol. Microbiol. 15:483-494. Kroos, L., and D. Kaiser. 1984. Construction of TnSlac, a transposon that fuses lacZ expression to endogenous promoters, and its introduction into Myxococcus xanthus. Proc. Natl. Acad. Sci. USA 815816-5820. Kroos, L., A. Kuspa, and D. Kaiser. 1986. A global analysis of developmentally regulated genes in Myxococcus xanthus. Dev. Biol. 117:252-266. Kroos, L., A. Kuspa, and D. Kaiser. 1990. Defects in fruiting body development caused by TnSlac insertions in Myxococcus xanthus. J. Bacteriol. 172:484-487. Kuner, J., and D. Kaiser. 1981. Introduction of transposon TnS into Myxococcus for analysis of developmental and other nonselectable mutants. Proc. Natl. Acad. Sci. USA 78:425-429. Lampe, D., M. E. Churchill, and H. M. Robertson. 1996. A purified mariner transposase is sufficient to mediate transposition in vitro. E M B O J. 155470-5479. Letouvet-Pawlek, B., C. Monnier, S. Barray, D. A. Hodgson, and J. F. Guespin-Michel. 1990. Comparison of P-galactosidase production by two inducible promoters in Myxococcus xanthus. Res. Microbiol. 141:425-435. Li, S., B.-U. Lee, and L. J. Shimkets. 1992. csgA expression entrains Myxococcus xanthus development. Genes Dev. 6:401-410. Li, S. F., and L. J. Shimkets. 1988. Site-specific integration and expression of a developmental promoter in Myxococcus xanthus. J. Bacteriol. 1705552-5556. Lu, A., K. Cho, W. P. Black, X. Y. Duan, R. Lux, Z. Yang, H. B. Kaplan, D. R. Zusman, and W. Shi. 2005. Exopolysaccharide biosynthesis genes required for social motility in Myxococcus xanthus. Mol. Microbiol. 55:206-220. Magrini, V., C. Creighton, and P. Youderian. 1999. Site-specific recombination of temperate Myxococcus xanthus phage Mx8: genetic elements required for integration. J. Bacteriol. 181:4050-4061. Manoil, C., and J. Beckwith. 1985. Tn phoA: a transposon probe for protein export sequences. Proc. Natl. Acad. Sci. USA 82: 8 129-8 133. Martin, S., E. Sodergren, T. Masuda., and D. Kaiser. 1978. Systematic isolation of transducing phages for Myxococcus xanthus. Virology 88:44-53. Michaelis, S., H. Inouye, D. Oliver, and J. Beckwith. 1983. Mutations that alter the signal sequence of alkaline phosphatase in Escherichia coli. J. Bacteriol. 154:366-374.

TOOLSFOR STUDYINGM. 29. GENETIC

XANTHUS

O’Connor, I