The Enzymes, Vol XIV: Nucleic Acids, Part A, 3rd Edition

The Enzymes VOLUME M V

NUCLEIC ACIDS Part A Third Edition

CONTRIBUTORS DAVID G. BEAR THOMAS BONURA BRUCE K. DUNCAN BRIAN ENDLICH ERROL C. FRIEDBERG MALCOLM L. GEFTER MARTIN GELLERT STANLEY HATTMAN RONALD D. KLEIN ARTHUR KORNBERG STEPHEN C. KOWALCZYKOWSKI I. ROBERT LEHMAN STUART LINN ZVI LIVNEH JACK D. LOVE KEVIN McENTEE CHARLES McHENRY STANFORD MOORE

HOWARD A. NASH BARBARA H. PHEIFFER ERIC H. RADANY ROBERT L. RATLIFF CHARLES C. RICHARDSON C. K. SINGLETON JOSEPH SPERLING BETSY M. SUTHERLAND KAREN M. TELANDER MUS KAVITCH INDER M. VERMA PETER H. VON HIPPEL JAMES C . WANG GEORGE M. WEINSTOCK BERNARD WEISS ARTHUR WEISSBACH ROBERT D. WELLS STEVEN B. ZIMMERMAN

ADVISORY BOARD MARTIN GELLERT I. ROBERT LEHMAN CHARLES C. RICHARDSON

THE ENZYMES Edited by PAUL D. BOYER Department of Chemist/ and Moleculur Biology Institute University of California Los Angeles, Culifornia

Volume XIV NUCLEIC ACIDS Part A

THIRD EDITION

198 1

ACADEMIC PRESS A Subsidiuty of Hurcourt Bruce Jovanovich, Publishers

New York London Toronto Sydney San Francisco

COPYRIGHT @ 1981, BY ACADEMIC PRESS, INC. ALL RIGHTS RESERVED. NO PART OF THIS PUBLICATION MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM OR BY ANY MEANS, ELECTRONIC OR MECHANICAL, INCLUDING PHOTOCOPY, RECORDING, OR ANY INFORMATION STORAGE AND RETRIEVAL SYSTEM, WITHOUT PERMISSION IN WRITING FROM THE PUBLISHER.

ACADEMIC PRESS, INC.

111 Fifth Avenue, New York, New

York 10003

United Kingdom Edition published by ACADEMIC PRESS, INC. (LONDON) LTD. 24/28 Oval Road, London NW1

IDX

Library of Coqress Cataloging i n Publication Data Main entry under t i t l e : The Enzymes.

Includes b i b l i q r a p h i c a l references. CONTENTS: v. 1. Structure and control. --v. 2. Kinetics and mechanism.--v. 3. Hydrolysis: 14. Mxleic acid, pt. A. peptide bonds.--[etc.I--v. I. Enzymes. I. Bo er, Paul 0. ed. [DNLM:

PuU5 ~ h e l 574.19'25 ISBN 0-12-122714-6 ( V . 14)

1. Enzymes.

QP601. €523

75-117107 AACfil

PRINTED IN THE UNITED STATES OF AMERICA 81828384

9 8 7 6 5 4 3 2 1

Contents List of Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xiii xvii

.

Section 1 DNA Polymerases and Related Enzymes

1

.

DNA Polymerases-A

Perspective

ARTHURKORNBERG

.

I Introduction . . . . . . . . . . . . . . . . . I1 . Variable Properties of Polymerases . . . . . . 111. Problems and Prospects in Polymerase Research IV. Conclusion . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

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

4 5 9 12

2 . DNA Polymerase I of Escherichia coli

I . ROBERTLEHMAN I . Purification . . . . . I1 . Physical Properties . I11. Reaction Catalyzed . IV. Biological Role . . . V. Research Applications

3

.

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

16 16

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

17 29 35

DNA Polymerase 111 Holoenzyme

CHARLES MCHENRY A N D ARTHURKORNBERG 1. Introduction . . . . . . . . . . I1 . DNA Polymerase I11 . . . . . . I11. DNA Polymerase I11 Holoenzyme 1V.Summary . . . . . . . . . . . . Note Added in Proof . . . . . .

V

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

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

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

39 40 43 49 49

vi 4

CONTENTS

.

1-Phage DNA Polymerase

I . ROBERTLEHMAN I. Introduction . . . . I1. T4 DNA Polymerase I11. T5 DNA Polymerase IV. T7 DNA Polymerase .

5

.

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

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

51 52

60 62

Cellular and Viral-Induced Eukaryotic Polymerases

ARTHURWEISSBACH I . Introduction and Perspective . . . . . . . . . . . . . . . . . . . . . . I1. DNAPolymerasea . . . . . . . . . . . . . . . . . . . . . . . . . . I11. DNA Polymerase p . . . . . . . . . . . . . . . . . . . . . . . . . . IV. DNA Polymerase . . . . . . . . . . . . . . . . . . . . . . . . . . V. Herpes Simplex Virus-Induced DNA Polymerase . . . . . . . . . . . . . VI . Vaccinia Virus-Induced DNA Polymerase . . . . . . . . . . . . . . . . VII . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.

6

67 69 73 76 80 83 86

Reverse Transcriptow

INDERM . VERMA I . Introduction . . . . . . . . . . . . . . . . . . I1. Purification and Properties . . . . . . . . . . . . 111 Biosynthesis of the Reverse Transcriptase . . . . IV. Mechanism of Reverse Transcription . . . . . . V. Applications to Molecular Biology . . . . . . .

.

7

.

. . . .

. . . .

. . . .

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

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

87 88 94 95 99

Terminal Deoxynucleotidyltransferase

ROBERTL . RATLIFF I . Introduction . . . . . . I1 Purification and Properties I11. The Reactions Catalyzed . IV. Practical Applications . . V Biological Role . . . . .

.

.

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

. . . . .

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

105 107 109

114 118

.

Section II DNA Nucleases and Related Enzymes

8

.

Deoxyribonucleases: Survey and Perspectives

STUARTLINN I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. Occurrence. Purification. and Molecular Properties . . . . . . . . . . .

122

.

124

vii

CONTENTS 111. I V. V. VI . VII.

9

.

Specificity . . . . . . . . . . . . . . . . . . . . . . . . . Assays: Designing Proper Substrates and Detection Procedures Biological Role . . . . . . . . . . . . . . . . . . . . . . . Control of Activities . . . . . . . . . . . . . . . . . . . . . Research Applications . . . . . . . . . . . . . . . . . . . .

. . . . . 124 . . . . . . 129 . . . . . 130 . . . . . 133 . . . . . 134

Type I Restriction Enzymes

BRIANENDLICH A N D STUART LINN I . Introduction . . . . . . . . . . . . . . . . . . . I1. Biological Role . . . . . . . . . . . . . . . . . . I11. Genetics . . . . . . . . . . . . . . . . . . . . . IV. Purification and Properties of Type I Enzymes . . . V. Reactions Catalyzed . . . . . . . . . . . . . . . . VI.Assays . . . . . . . . . . . . . . . . . . . . . . VII . On the Mechanisms of Cleavage-A Model Scheme VIII . Conclusions . . . . . . . . . . . . . . . . . . .

.

10

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

138 142 143 144 148 150

. . . . . . . . . . . 151 . . . . . . . . . .

156

Type II Restriction Enzymes

ROBERTD . WELLS. RONALDD . KLEIN.

AND

C . K . SINGLETON

I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1 . General Properties . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Catalytic Properties . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Ionic Strength and Solvent Effects on Enzyme Specificity . . . . . . . . . V. Synthetic Oligonucleotides as Substrates . . . . . . . . . . . . . . . . . VI . Substituted DNAs . . . . . . . . . . . . . . . . . . . . . . . . . . . VII . Inhibitor Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . VIII . Influence of Drugs and Other Ligands on Cleavage Specificities . . . . . . IX. Cleavage of Single-Stranded DNA Substrates by Certain Restriction Endonucleases . . . . . . . . . . . . . . . . . . . . . . . X . Cleavage of DNA-RNA Hybrids . . . . . . . . . . . . . . . . . . . . XI . Insolubilized Restriction Enzymes . . . . . . . . . . . . . . . . . . . XI1. Crystallization of Restriction Endonucleases . . . . . . . . . . . . . . . XI11 . Genes for Restriction Endonucleases . . . . . . . . . . . . . . . . . . XIV. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

158 159 167

172 176

177 181 183 185

187 188 188 189 191

1 1. Endonucleases Specific for Single-Stranded Polynucleotides

I . ROBERTLEHMAN I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . I1 . Neurosporu c'rassa Endonuclease . . . . . . . . . . . . . . . . 111. Nuclease S1 . . . . . . . . . . . . . . . . . . . . . . . . . IV. Mung Bean Endonuclease . . . . . . . . . . . . . . . . . . . V. Research Applications of Single-Strand-Specific Endonucleases . .

. . . . . . . .

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

193 194 198 199 201

viii 12.

CONTENTS

Exodeoxyribonucleases of Escherichiu coli

BERNARD WEBS

. . . . .

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

. .

203 206

I. Introduction . . . . , . . . . , . . . . . . . . . . . . . . . . . . . 11. Biological Role . , , , . . . . , . . . . . . . . . . . . . . . . . . . 111. Purification and Biophysical Properties , , . , . . , . , . . . , . . , . IV. The Reactions Catalyzed . . . . . . , . . . . . . . . . . . . . . . . . V. Models for the Mechanism of Action of the ATP-Dependent Double-Stranded DNA Exonuclease . , , . , . . , . . , . , . . . . . . . . . , . . . . VI. Conclusion . . . . , , . . . . , . . . . . . . . . . . . . . . . . . .

234 235 238 240

I. General Properties . 11. Specific Exonucleases

13.

, ,

. . . . . , . . . . .

.

,

recBC-like Enzymes: Exonucleuse V Deoxyribonucleases

KARENM. TELANDER MUSKAVITCH A N D STUART LINN

14.

Enzymes That Incise Damaged DNA

ERROLc. FRIEDBERG, THOMASBONURA,ERICH. JACK D. LOVE

RADANY, A N D

I. Introduction . . , . . . . . . . . . . , . . . . . . . . . . . . . . . 11. Enzymes That Attack Phosphodiester Bonds in DNA Following Hydrolysis of N-Glycosylic Bonds (AP Endonucleases) . . . . . , . , . , . . , . . 111. Enzymes that Attack Phosphodiester Bonds in Damaged DNA with Intact N-Glycosylic Bonds . . . . . , . . . . , . . . . . . . . . . . . Note Added in Proof . . . . . . , , . , , . . . . . . . . . . . . . .

15.

247 250

251 252 274 279

Pancreatic DNase

STANFORD MOORE I. Introduction . , . . , , . . . . . , . . . 11. Purification. . . , . . . . . . . . . . . . 111. Chemical Structure . . . . . . . . . . . . IV. Catalytic Properties , . . . . . , . . . . . V. Actin as an Inhibitor of DNase I , , . . . . VI. Research Applications . . . . , . , , . . .

. . . . . .

. . . . . . . . . . .

.

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

.

. . . . . .

281 282 285 288 293 295

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

299 301

. . . . .

. . . . . .

. . . . .

. . . . .

. .

. . . . . .

Section 111. DNA Modification 16.

Bacteriophage T4 Polynucleotide Kinare

CHARLES C. RICHARDSON I. Introduction

. . . . . .

. . . .

11. Isolation and Physical Properties

.

. .

. . . .

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

ix

CONTENTS

I11 . Catalytic Properties . . . . . . . . . . . . . . . . . . . . . . . . . . 305 I V. Role of Polynucleotide Kinase in Vivo . . . . . . . . . . . . . . . . . . 312 V. Research Applications . . . . . . . . . . . . . . . . . . . . . . . . . 313

.

17

Eukaryotic DNA Kinaser

STEVENB . ZIMMERMAN A N D BARBARA H . PHEIFFER I . Introduction and Perspectives . . . . . . . . . . . . . . . . . . . . . I1. Purification and Properties . . . . . . . . . . . . . . . . . . . . . . . 111. The Catalytic Reaction . . . . . . . . . . . . . . . . . . . . . . . . 1V. Comparison of the DNA Kinases with RNA Kinase and Polynucleotide Kinase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Biological Role . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Research Applications . . . . . . . . . . . . . . . . . . . . . . . . . Note Added in Proof . . . . . . . . . . . . . . . . . . . . . . . . .

315 316 318 326 327 329 329

18. Type I DNA Topoisomerarer

JAMES C . WANG I. I1 . 111. IV.

Introduction . . . . . . . . . . . . . . . . . Purification and Properties . . . . . . . . . . . The Reactions Catalyzed by the Enzymes . . . Biological Roles . . . . . . . . . . . . . . . . V . Research Application . . . . . . . . . . . . .

.

19

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . . . . 332 . . . . . . . 334 . . . . . . . . 337 . . . . . . . 341 . . . . . . . 343

DNA Gyrase and Other Type II Topoisomerases

MARTINGELLERT I . Introduction and Perspectives . . . . . . . . . . . . . . . . . . . . . I1 . Definitions and General Methods . . . . . . . . . . . . . . . . . . . . I11 . DNAGyrase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Other Type I1 Topoisomerases . . . . . . . . . . . . . . . . . . . . . V. Biological Role . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Research Applications . . . . . . . . . . . . . . . . . . . . . . . . .

20

.

345 347 348 359 361 366

DNA Unwinding Enzymes

MALCOLM L . GEFTER I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . I1 . Purification and Properties of the rep Protein . . . . . . . . . . I11. Isolation and Characterization of Helicase I11 . . . . . . . . . IV. Mechanism of Action of rep Protein and Helicase 111 . . . . . . V. The Biological Role of Enzymes That Catalyze Unwinding of DNA

. . . .

. . . .

. . 367 . . . 368 . . . 370 . . . 371

. . . . . 372

CONTENTS

X

.

21

Single-Stranded DNA Binding Proteins

STEPHENC . KOWALCZYKOWSKI. DAVID G . BEAR.A N D PETERH . VON HIPPEL I . Introduction and Overview . . . . . . . . . . . . . . . . . . . . . . . I1. Theoretical and Experimental Considerations . . . . . . . . . . . . . . 111. Protein Isolation and Purification: Procedures and Strategies . . . . . . . IV. Structure. Properties. and Nucleic Acid Binding Interactions of Several Single-Stranded DNA Binding Proteins . . . . . . . . . . . . . . . . V. DNA Binding Proteins as Research Tools . . . . . . . . . . . . . . . VI . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

22

.

374

. 375 . 383

. 388 . 441 442

The recA Enzyme of Escherichia coli and Recombination Assays

KEVINMCENTEEA N D GEORGEM. WEINSTOCK I. Introduction . . . . . . I1. Purification . . . . . . . I11. Physical Properties . . . IV. Reactions Catalyzed . . . V. Assays for Recombination VI . Biological Role . . . . . VII . Research Applications . .

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

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

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

445 447 448 453 464 465 470

23. Site-Specific Recombination Protein of Phage lambda

HOWARDA . NASH I . Introduction and Perspectives I1. Purification and Properties . . 111. Reactions Involving Int . . . IV. Biological Role . . . . . . . V. Research Applications . . . .

24

.

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

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

471 473 474 479 479

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

482 485 493 510 512

Photonractivating Enzymes

BETSYM . SUTHERLAND I . Introduction . . . . . . I1 . Purification and Properties 111. The Reaction . . . . . . IV. Biological Role . . . . . V . Research Applications . .

. . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

25 . DNA Methylation

STANLEYHATTMAN I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 517 I1 Nature of Methylated Bases and Methods of Analysis . . . . . . . . . . . 518

.

xi

CONTENTS

I11 . DNA Methylases . . . . . . . IV. Distribution of Methylated Bases V. Other DNA Modifications . . . VI . Biological Roles . . . . . . . . VII . Concluding Remarks . . . . .

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

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

521 531 536 537 547

26. DNA Base-Insertion Enzymes (Insertases)

ZVI LIVNEH A N D JOSEPH SPERLING I . Introduction . . . . . 1I.Assay . . . . . . . . 111. Purification . . . . . . IV. Properties . . . . . . V. Mechanism of Insertion VI . Biological Role . . . .

27

.

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

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

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

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

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

549 551 554 555 559 560

DNA Glycorylases

BRUCE K . DUNCAN 1. I1 . 111. IV.

Introduction . . . . Physical Properties . Enzymatic Properties Physiological Role . . V. Research Applications Note Added in Proof

. . . . . .

. . . . . .

. . . . . .

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

565 568 569 578 584 586

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

587

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

622

Subjecr Inc1e.r

Contents of Orher Volumes

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

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

6 45

This Page Intentionally Left Blank

List of Contributors Numbers in parentheses indicate the pages on which the authors' contributions begin.

DAVID G. BEAR (373), Institute of Molecular Biology and Department of Chemistry, University of Oregon, Eugene, Oregon 97403 THOMAS BONURA (25 l), Laboratory of Experimental Oncology, Department of Pathology, Stanford University, Stanford, California 94305 BRUCE K. DUNCAN (565), Institute for Cancer Research, Fox Chase Cancer Center, Philadelphia, Pennsylvania 191 1 1 BRIAN ENDLICH (137), Department of Biochemistry, University of California, Berkeley, California 94720 ERROL C. FRIEDBERG (251), Laboratory of Experimental Oncology, Department of Pathology, Stanford University, Stanford, California 94305 MALCOLM L. GEFTER (367), Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02 139 MARTIN GELLERT ( 3 4 3 , Laboratory of Molecular Biology, National Institute of Arthritis, Metabolism and Digestive Diseases, National Institutes of Health, Bethesda, Maryland 20205 STANLEY HATTMAN (5 17), Department of Biology, University of Rochester, Rochester, New York 14627 RONALD D. KLEIN (157), Department of Biochemistry, College of Agricultural and Life Sciences, University of Wisconsin, Madison, Wisconsin 53706 ARTHUR KORNBERG (3, 39), Department of Biochemistry, Stanford University Medical School, Stanford, California 94305 STEPHEN C. KOWALCZYKOWSKI' (373), Institute of Molecular Biology and Department of Chemistry, University of Oregon, Eugene, Oregon 97403 Present address: Department of Molecular Biology, Northwestern University Medical and Dental Schools, Chicago, Illinois 60611.

...

Xlll

xiv

LIST OF CONTRIBUTORS

I. ROBERT LEHMAN (15, 51, 193), Department of Biochemistry, Stanford University Medical School, Stanford, California 94305 STUART LINN (121, 137, 233), Department of Biochemistry, University of California, Berkeley, California 94720 ZVI LIVNEH (549),2Department of Organic Chemistry, The Weizmann Institute of Science, Rehovot, Israel JACK D. LOVE ( Z l ) , Laboratory of Experimental Oncology, Department of Pathology, Stanford University, Stanford, California 94305 KEVIN McENTEE (445),3 Department of Biochemistry, Stanford University School of Medicine, Stanford, California 94305 CHARLES McHENRY (39), Department of Biochemistry and Molecular Biology, University of Texas Medical School, Houston, Texas 77025 STANFORD MOORE (281), The Rockefeller University, New York, New York 10021 HOWARD A. NASH (4711, Laboratory of Neurochemistry, National Institute of Mental Health, National Institutes of Health, Bethesda, Maryland 20205 BARBARA H. PHEIFFER (315), Laboratory of Molecular Biology, National Institute of Arthritis, Metabolism and Digestive Diseases, National Institutes of Health, Bethesda, Maryland 20205 ERIC H. RADANY (25 l), Laboratory of Experimental Oncology, Department of Pathology, Stanford University, Stanford, California 94305 ROBERT L. RATLIFF (105), Genetics Group, Life Sciences Division, Los Alamos Scientific Laboratory, University of California, Los Alamos, New Mexico 87545 CHARLES C. RICHARDSON (299), Department of Biological Chemistry, Harvard Medical School, Boston, Massachusetts 021 15 C. K. SINGLETON (1571, Department of Biochemistry, College of Agricultural and Life Sciences, University of Wisconsin, Madison, Wisconsin 53706

* Present address: Department of Biochemistry, Stanford University, Stanford, California 94305. Present address: Department of Biological Chemistry, UCLA Medical School, Los Angeles, California 90024.

LIST OF CONTRIBUTORS

xv

JOSEPH SPERLING (549),4 Department of Organic Chemistry, The Weizmann Institute of Science, Rehovot, Israel BETSY M. SUTHERLAND (481), Biology Department, Brookhaven National Laboratory, Upton, New York 11973 KAREN M. TELANDER MUSKAVITCH (233), Department of Biochemistry, University of California, Berkeley, California 94720 INDER M. VERMA (87), The Salk Institute, San Diego, California 92138 PETER H. VON HIPPEL (373), Institute of Molecular Biology, Department of Chemistry, University of Oregon, Eugene, Oregon 97403 JAMES C. WANG (331), Department of Biochemistry and Molecular Biology, Harvard University, Boston, Massachusetts 021 15 GEORGE M. WEINSTOCK (445),6 Department of Biochemistry, Stanford University School of Medicine, Stanford, California 94305 BERNARD WEISS (203), Department of Microbiology, The Johns Hopkins University School of Medicine, Baltimore, Maryland 21205 ARTHUR WEISSBACH (67), Department of Cell Biology, Roche Institute of Molecular Biology, Nutley, New Jersey 071 10 ROBERT D. WELLS (157), Department of Biochemistry, College of Agricultural and Life Sciences, University of Wisconsin, Madison, Wisconsin 53706 STEVEN B. ZIMMERMAN (3 15), Laboratory of Molecular Biology, National Institute of Arthritis, Metabolism and Digestive Diseases, National Institutes of Health, Bethesda, Maryland 20205

Present address: Department of Biochemistry, Stanford University, Stanford, California 94305. Present address: The Biological Laboratories, Harvard University, Cambridge, Massachusetts 02138. ' Present address: Frederick Cancer Research Center, Frederick, Maryland 21701.

This Page Intentionally Left Blank

Preface This volume marks a change in organization, but not in the basic objectives, of “The Enzymes.” Until this volume, the Third Edition of “The Enzymes” has been organized as an updated companion to the First and Second Editions ; namely, a multivolume treatise that covers the present knowledge of groups of enzymes that catalyze similar reactions. But as the Third Edition progressed, as the field grew ever more extensive, and as molecular explanation of biological function continued to emerge, a different format seemed more appropriate. We felt that users of this treatise would be better served if enzymes were grouped by their biological function instead of by the nature of the reaction catalyzed. The objective for each enzyme type presented remains the same: Outstanding authorities in the field present important information about the molecular nature of enzymes and of the reactions they catalyze, together with a perspective of their biological role. The information contained thus far in this edition serves as a base for later volumes that will follow in an open-ended series. The first topic to be covered under the new format is the enzymology of nucleic acids. Volume XIV is one of two on this topic. The volumes have been planned with the help of a fine Advisory Board composed of I. Robert Lehman, Martin Gellert, and Charles C. Richardson. Volumes XIV and X V will give authoritative coverage of the enzymes that make, modify, cleave, recombine, repair, and degrade the nucleic acids. This enzymology makes possible the remarkable isolation and restructuring of DNA and RNA and allows penetrating experimentation about their biological function. Such achievements are at the core of a biological revolution. Volume XIV covers all enzymes with DNA as the primary substrate or product except DNA ligases, which will be treated with the RNA ligases xvii

xviii

PREFACE

in Volume XV. As with previous volumes, the quality of the present volume is gratifying and was achieved because nearly all of the first-choice contributors agreed to participate. Also, in keeping with a tradition already established for “The Enzymes,” the volume is timely. All manuscripts were received within a period of several months. Volume XIV was ready for distribution in less than a year from the time the first manuscript was received. The splendid cooperation of the authors and the publisher made this possible. Preparation of the volume was greatly facilitated by the capable participation of Lyda Boyer as Assistant Editor. Also, I take pleasure in acknowledging the warm and professional interest and cooperation of the staff of Academic Press. Paul D. Boyer

Section I

DNA Pulymerases and Related Enzymes

This Page Intentionally Left Blank

DNA Polymerases . A Perspective ARTHUR KORNBERG

I . Introduction . . . . . . . . . . . . . . . . . . . A . Invariant Rules of Polymerase Action . . . . . . I1. Variable Properties of Polymerases . . . . . . . . . A . Organization and Size . . . . . . . . . . . . . B . Template Preference . . . . . . . . . . . . . . C . Primer Preference . . . . . . . . . . . . . . . D . Tolerance for Substitution of a Nucleotide Analog . E . Fidelity . . . . . . . . . . . . . . . . . . . . F. Processivity . . . . . . . . . . . . . . . . . . G . Catalytic Efficiency . . . . . . . . . . . . . . . H . Nick Translation and Strand Displacement . . . . 1. Multiplicity and Abundance . . . . . . . . . . . J . Physiological Functions . . . . . . . . . . . . . K . Optima! Conditions . . . . . . . . . . . . . . . 111. Problems and Prospects in Polymerase Research . . . A . Isolation of a Scarce Enzyme . . . . . . . . . . B . Physical and Functional Properties . . . . . . . . C . The Template-Primer for Assay . . . . . . . . . D . Linkage to Related Replication Proteins . . . . . E . Biosynthesis and Regulation . . . . . . . . . . . IV. Conclusion . . . . . . . . . . . . . . . . . . . .

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

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

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . . .

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

. . . . . . .

. . . . .

. . . . .

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

4 4 5 5

5 6 6 6 7

7 7 8 8 8

9 9 10 11 11

12 12

3 THE ENZYMES. Vol . XiV

.

Copyright 0 1981 by Academic Press InC . All rights of reproduction in any form reserved ISBN 0-12-122714-6

4

ARTHUR KORNBERG

1. Introduction

DNA polymerases are found in nature wherever DNA is made (1-3). Assembly of a DNA chain by these deoxyribonucleotidyltransferases invariably follows certain rules, whether synthesis is extensive as in replication of a chromosome or brief as in the repair of a gap in the chain. A. INVARIANT RULESO F POLYMERASE ACTION(4) 1. The Primer

A nucleotide is added to the 3’-hydroxyl moiety of the terminal nucleotide of a preexisting chain, the primer. DNA polymerases, unlike RNA polymerases ( 4 7 , 5 ) cannot start a chain.

2. The Template The nucleotide added is selected in response existing DNA chain, the template, to which the base-paired double helix, the template-primer . tidyltransferase (6 ), which assembles a random direction, is not a polymerase by definition.

to directions by a preprimer is annealed in a Terminal deoxynucleochain without template

3 . The Nucleotide Substrate

The nucleotide added to the template-primer is a 2’-deoxyribonucleoside-5’-triphosphate chelated with a divalent metal, generally Mg’+. 4. Base Pairing

Selection of the nucleotide, dictated by the template chain, follows the Watson and Crick rules of pairing adenine with thymine, guanine with cytosine. Generally there is no recognition of a specific sequence of bases in the template. An animal cell polymerase copies bacterial DNA and the reverse is equally true. 1. T. Kornberg and A. Kornberg, “The Enzymes,” Vol. X, p. 119, 1974. 2. L. A. Loeb, “The Enzymes,” Vol. X, p. 174, 1974. 3. A. Kornberg, “DNA Replication.” Freeman, San Francisco, 1980. 4. Polymerase in this chapter will refer only to DNA polymerase although the term applies as well to RNA polymerases, the ribonucleotidyltransferases that assemble RNA chains by DNA template direction. 4a. P. Chambon, ”The Enzymes,” Vol. X, p. 261, 1974. 5 . M. J. Chamberlin, “The Enzymes,” Vol. X, p. 333, 1974. 6. F. J. Bollum, “The Enzymes,” Vol. X, p. 145, 1974.

I.

DNA POLYMERASESA PERSPECTIVE

5

5. Mechanism Nucleophilic attack by the 3'-hydroxyl group of the primer terminus (7), on the innermost (or a ) phosphorus of the nucleotide selected for addition, produces the 3'-5' phosphodiester bond of the DNA backbone and eliminates inorganic pyrophosphate.

6 . Polarity of Chain Growth Because the mechanism entails a nucleotidyl transfer to a 3'-hydroxyl primer terminus, DNA chain growth is in the 5' + 3' direction. The polarity of the newly synthesized chain is opposite to that of the template; the growing chain and template are antiparallel. 7. Reversal of the Reaction Pyrophosphorolysis of a DNA chain, and exchange between inorganic pyrophosphate and the p ,y group of a deoxynucleoside triphosphate during a synthetic step, are reversals of polymerization; as such, these reactions require a primer terminus base-paired with a template. An exchange rate more rapid than pyrophosphorolysis may be due to attack b y inorganic pyrophosphate at an enzyme site more favorable than that occupied by the primer terminus in pyrophosphorolysis. II. Variable Properties of Polymerases

DNA polymerases, isolated from a single cell or from different cells and organisms, vary widely in certain properties. A.

ORGANIZATION AND SIZE

They vary from a single polypeptide of 40,000 daltons (mammalian /3 polymerase) (8) to a seven-subunit complex of about 500,000 daltons (E. coli DNA polymerase I11 holoenzyme) (9). Inasmuch as multisubunit forms dissociate easily, a polymerase isolated as a single polypeptide may be part of a larger assembly in the cell.

PREFERENCE B. TEMPLATE The enormous variations in size, structure and composition of the template provide the basis for the clearest display of the variable proper7. P. M. J. Burgers and F. Eckstein, JBC 254, 6889 (1979). 8. D. Korn, P. A. Fisher, J. Battey, and T. S.-F. Wang, CSHSQB 43, 613 (1978). 9. C. McHenry and A. Kornberg, JBC 252, 6478 (1977).

6

ARTHUR KORNBERG

ties of polymerases. Extending a chain end at a nick, at a short gap or a long gap, displacing a strand, and even switching templates from one strand to the other, are all template-directed capacities that distinguish one polymerase from another.

C. PRIMERPREFERENCE

RNA priming is the general mechanism for starting DNA chains but probably not the universal one (10); a specific deoxynucleotidyl protein appears to provide the Y-start and the primer terminus of each chain of the adenoviral DNA duplex (11). Polymerases encoded by phages T7 and T4 extend tetra- and pentanucleotide RNA primers of nearly fixed composition; mammalian RNA primers for polyoma, SV40, and chromosomal replication are decanucleotides of relatively random composition (12). Escherichia coli primase, responsible for priming phage 6x174 and chromosomal DNA synthesis, incorporates deoxy- as well as ribonucleotides into chains only afew residues long that can be extended by DNA polymerase I11 holoenzyme (13). It seems likely that polymerases differ in their capacity and efficiency to use primers of varying size and composition, and that they may also interact directly with primases responsible for primer synthesis.

D. TOLERANCE FOR SUBSTITUTION OF A NUCLEOTIDE ANALOG An alteration in the base, sugar, or phosphate may be accepted in varying degrees. The same is true for comparable alterations in the primer and template and for substitution of Mn2+for Mg”. E. FIDELITY Insertion of the correct nucleotide into the chain is determined not only by the strictness of selection in the addition step, but also by availability of a subsequent proofreading step to remove a mismatched nucleotide from the primer terminus. The latter step can be accomplished by a 3’ + 5‘ exonuclease domain of the polymerase polypeptide, or perhaps by an 10. A. Kornberg, I n “RNA Polymerases” (R. Losick and M. Chamberlin, eds.), p. 331, Cold Spring Harbor Laboratory, Cold Spring Harbor, 1976. 11. M. D. Challberg, S. V. Desiderio, and T. J. Kelly, PNAS 77, 5105 (1980). 12. R. Eliasson and P.Reichard, JBC 253, 7469 (1978). 13. L. Rowen and A. Kornberg, JBC 253, 770 (1978).

1.

DNA POLY MERASES-A

PERSPECTIVE

7

associated exonuclease. Thus, correct base pairing may be required at the primer terminus as well as in its covalent extension when the incoming nucleotide is matched to the template.

F. PROCESSIVITY When the polymerase remains associated with the template-primer after the covalent addition of a nucleotide it is regarded as processive. (The opposite term is distributive .) The working polymerase molecule, if distributive, may equilibrate with free polymerase molecules after every addition event or, if processive, may not so do until chains, many thousands long, have been completed. The degree of processivity is determined by the state and nature of the polymerase, the template-primer, the nucleotide, the metal, and the reaction conditions (14).

G. CATALYTIC EFFICIENCY The rate of chain growth depends on the rate of the many events that comprise a nucleotide addition step: complex formation with primer terminus and template, binding of the correct deoxynucleoside triphosphate and rejection of incorrect ones, formation of the diester bond, elimination of inorganic pyrophosphate, and movement along the template-primer to the new primer terminus or dissociation from it. Turnover numbers, measured in vitro, vary from several hundred to many thousand residues polymerized per enzyme molecule per minute.

H. NICKTRANSLATION AND STRAND DISPLACEMENT A polymerase, such as E. coli polymerase I, has the capacity to start synthesis on duplex DNA at a nick in one of the strands and coordinately degrade the 5' end of the chain with its 5' -+ 3' exonuclease activity ( 3 ) . This concurrent polymerization and hydrolysis moves (linearly translates) the nick along the duplex without change in mass of the DNA. The nicktranslation capacity of E. coli polymerase I may be a manifestation of its repair function in removing an RNA primer from the 5' end of a chain, or in excising other uncommon nucleotides, such as a mismatched sequence in that location. Should the 5' chain end at the nick escape cleavage, E. coli polymerase I can displace that chain for even great distances simply by progressive polymerization. Most polymerases need the assistance of helicases and the expenditure of ATP energy to destabilize duplex DNA 14.

p. J. Fay, K.

0. Johanson, C. McHenry, and R. Bambara, J5C 256, 976 (1981).

8

ARTHUR KORNBERG

and achieve strand displacement; the presence of several distinctive helicases inE. coli (15) may be responsible for strand displacement associated with specialized functions, and may also provide for specific interactions with polymerases. I.

MULTIPLICITYA N D ABUNDANCE

Eukaryotic cells have one or a variety of polymerases in the nucleus; polymerases are also in cytoplasmic locations, as in mitochondria, chloroplasts, and certain viral factories. Even a prokaryotic cell, such as E. coli, has three distinctive polymerases. The molecular abundance of polymerases may be low, as for those associated with replicative forks, or many times higher for those designed for gap-filling and repair functions. Polymerases encoded by viruses are relatively numerous because they are produced to sustain rapid viral multiplication unregulated by strict copynumber controls, as is the host chromosome.

J. PHYSIOLOGICAL FUNCTIONS Certain polymerases seem designed for the extensive, highly processive, rapid chain growth of chromosome replication. In this role, the polymerase associated with other proteins that contribute to the progress of the replication fork may perhaps be part of a larger entity. Should the existence of such an organized complex become established, it might be called a replisome . It seems likely that the properties of a polymerase that is an integral part of such a larger complex will be markedly different from those observed in the purified enzyme. Other polymerases are designed for gap filling and serve in the excision and repair of mismatched regions, completion of chains undergoing recombination, and the removal and replacement of sections of RNA that served as primers to initiate DNA synthesis. While specialized in function, polymerases are also adaptable and fill other roles demanded by exigencies of mutation, cellular stresses, and invasion by viruses and plasmids. Thus a polymerase may provide an auxiliary or alternative means of synthesis to assist or partially replace a polymerase deficient in numbers or in function.

K. OPTIMALCONDITIONS In view of the multiplicity of polymerase forms, locations, and functions, there is (as might be expected) great variation in the physical and chemical IS. B. Kuhn, M. Abdel-Monern, and H. Hoffrnann-Berling, CSHSQB 43, 63 (1978).

I.

DNA POLYMERASESA PERSPECTIVE

9

components that define an optimal condition for DNA synthesis, and the range about this optimum that can be tolerated. Sharp distinctions are based on sensitivity to ionic strength, sulfhydryl-blocking agents, temperature, and pH. Especially noteworthy are the influences of DNA-binding agents such as polyamines, histones, specialized DNA-binding proteins, and DNA helicases.

111.

Problems and Prospects in Polymerase Research

A. ISOLATION OF A SCARCE ENZYME Lack of an adequate quantity of a homogeneous enzyme has been, and remains, the greatest obstacle to progress in understanding the nature and action of a polymerase. Chemical and physical studies with impure preparations have been of minimal value, and biochemical studies of such preparations have generally been misleading. The two major difficulties in obtaining a homogeneous preparation of a polymerase are its naturally low abundance and, especially in the case of multisubunit enzymes, their instability. DNA polymerase 111 holoenzyme ofE. coli, for example, is present in only ten to twenty copies per cell and activity is lost upon gentle manipulations (9). Yields of about 3 mg of a preparation, only about 60% pure, are obtained from 3 kg of cell paste after a laborious procedure consuming several weeks. The difficulties in isolating mammalian polymerases from cell cultures and tissues are even more formidable. Whereas skillful and patient application can circumvent enzyme instability, the pursuit of a trace-quantity enzyme can be a forbidding enterprise. No wonder that so little enzymology is done by so few people on so important a subject. To overcome the difficulties of isolating polymerases, two approaches can be taken. One is to locate a relatively abundant source; the other is to amplify the polymerase gene in order to overproduce the enzyme. The natural abundance of a virus-encoded polymerase is high because many copies of the viral chromosome are produced to facilitate rapid and extensive viral multiplication. Thus the phage T4- and T7-encoded polymerases have been isolated in pure form in relatively large amounts and studied extensively (16, 17). An attractive eukaryotic system is the Drosophifu embryo which doubles its DNA content every 10 minutes; 16. C. C. Liu, R. L. Burke, U. Hibner, J. Barry,and B. Alberts, CSHSQB 43,469 (1978). 17. C. C. Richardson, L. J. Romano, R . Kolodner, J. E. LeClerc, F. Tamanoi, M. J. Engler, F. B . Dean, and D. S . Richardson, CSHSQB 43, 427 (1978).

10

ARTHUR KORNBERG

from 1 kg of embryos, 1 mg of a nearly homogeneous a polymerase can be obtained (18). As a general method for obtaining an abundant source, the approaches offered by genetic engineering techniques are compelling. By identifying a chromosomal segment containing the polymerase gene or the messenger RNA transcribed from it, recombinant DNAs with the gene in viral or plasmid vectors can create major cellular factories for producing the enzyme. In this way, the abundance of polymerase I of E. coli (19) and the y subunit of polymerase I11 holoenzyme (20) have been raised a hundredfold and the yields of pure protein nearly a thousandfold. Similar avenues are now open for overproduction of eukaryotic as well as prokaryotic polymerases. PRGPERTIES B. PHYSICAL AND FUNCTIONAL

For even the most intensively investigated polymerases virtually nothing is known about amino acid sequence or three-dimensional structure. Based on a facile proteolytic cleavage of E. coli polymerase I into two functionally distinct fragments, the presence of an exposed hingelike region in this multifunctional enzyme can be inferred (I). Also a limited amount of information about the active sites has been obtained with chemical probes. However, the assembly and arrangement of subunits in the complex polymerases are unexplored, and the major harvest of knowledge from X-ray ditFraction analysis of crystal structure has yet to be made. With regard to functional properties, enough is known to formulate the invariant rules of polynucleotide assembly that have been cited. Yet the several component stages that comprise the template-directed, covalent addition of a nucleotide have not been dissected and analyzed. The number and nature of substrate binding sites are known in only a few instances. Allosteric effects have not been discovered or established. Nor has the role of zinc, possibly a universal component of polymerases, been explained. How each of the polypeptides of the multisubunit polymerases contributes to the catalytic efficiency, specificity, fidelity, and processivity is still a mystery. With the imminent availability of adequate quantities of homogeneous polymerases and sensitive methods for tracing them, there can be no excuse for neglecting the intensive and widespread investigation of their physical and functional properties. 18. G. Villani, B. Sauer, and I. R. Lehman, JBC 255, 9479 (1980). 19. W. S. Kelley, K. Chalmers, and N . E. Murray, PNAS 74, 5632 (1977). 20. U. Hiibscher and A. Kornberg, JBC 255, 11698 (1980).

I.

DNA POLYMERASES-A

c.

PERSPECTIVE

11

THETEMPLATE-PRIMER FOR ASSAY

That homogeneous polymerases are essential for serious studies has already been emphasized. “Not wasting clean thinking on dirty substances” is, after all, an elementary principle of chemistry. Yet the application of this dictum to the template-primer in the assay of a polymerase has not been sufficiently appreciated. The standard DNA preparations derived from natural sources and most synthetic homopolymers in general use have been a frequent source of confusion. Obtaining a clean, intact E. coli chromosome, namely the circular, duplex DNA molecule of four million base pairs, is still impossible. What can be obtained is a collection of random fragments with nicks, gaps, and variable ends. A comparably battered substrate is calf thymus DNA, often “activated” for assay use by further degradation with pancreatic DNase. Such templates provide targets for many adventitious actions by a polymerase and serve unpredictably as substrates for a large variety of nucleases, recombinases, and other DNA-directed enzymes. Inhomogeneities in size and structure of synthetic DNA polymers bedevil their use, as with most natural DNAs. Sources of intact natural DNA can be found in the form of viral chromosomes and plasmids and have proved to be attractive substrates for the template-primer role for polymerases. Although such molecules are generally inert for polymerases directly, they can serve in the discovery of the natural priming systems that make them susceptible to polymerase action. Improved techniques for isolating the DNAs of even large viruses and plasmids will soon make their use feasible too. Advances in the organic synthesis of relatively long DNA sequences are likely to provide the variety of templates and primers of defined primary and secondary structure needed for refined studies. D.

LINKAGE TO RELATED REPLICATION PROTEINS

It seems plausible that a DNA polymerase would interact in a precise way and for a sustained period with proteins that precede, facilitate, and follow its actions. Several examples can be suggested: (i) Polymerases can use very short RNA primers because they are held in place by the primases that make them (13, 21). (ii) Single-strand binding proteins that destabilize helical structures in some instances stimulate replication only by particular polymerases (16). (iii) Progress at the replicating fork by continuous synthesis by phage T4 polymerase is enhanced by the gene 41 and 61 proteins responsible for the priming actions in the discontinuous 21. G. Hillenbrand, G. Morelli, E. Lanka, and E. Scherzinger, CSHSQB 43,449 (1978).

12

ARTHUR KORNBERG

phase of synthesis (16); similarly in renlication of the duplex form of phage 4x174, a coupling may exist between DNA polymerase III holoenzyme and the complex of priming proteins, called the primosome (22, 23). (iv) Associated nucleases may contribute proofreading or lesion-excision functions (3, 24). (v) Machinery for biosynthesis of the deoxynucleoside triphosphate substrates may be part of the fabric that contains the polymerase (25). While still imaginary, a large replisome with functional responsibilities that go beyond replication to chromosome orientation and segregation, is an attractive investigative goal.

E. BIOSYNTHESISAND REGULATION A major current question in biology and medicine is what determines the initiation of a cycle of chromosome replication and thus the factors that govern the patterns of DNA synthesis in resting or dividing cells. Despite all that is known about the cell cycle from studies of its biology, it seems unlikely that a genuine understanding of the control of replication will emerge without a detailed knowledge of its biochemistry. More information about the identity of the polypeptides responsible for DNA synthesis and the genetic loci that encode them will make it possible to determine their levels as a function of the cell cycle, their assembly into larger complexes, and the mechanisms for regulation of their biosynthesis. In such studies, plasmid-bearing strains that overproduce the polypeptides, and mutants that produce defective polypeptides, will be of special value.

IV.

Conclusion

Fashion dictates the ebb and flow of scientific activity as much as it does other human social activities. Current attention to manipulating DNA, examining the immune system, and explaining physiologic and pathologic events has taken its toll in solving other problems, including some basic ones. Enzymology is one of the victims, particularly the study of complex and scarce enzymes such as the DNA polymerases. Fortunately there are now opportunities and forces that may direct more effort toward these neglected goals. An adequate start has been made in iden22. 23. 24. 25.

N . Arai and A. Kornberg, JBC 256, 5294 (1981). R. L. Low, K. Arai, and A. Kornberg, PNAS 78, 1436 (1981). D. W. Mosbaugh and R. R. Meyer, JBC 255, 10239 (1980). J . B. Flanegan and G. R. Greenbcrg, JBC 252, 3019 (1977).

I . DNA POLYMERASES-A PERSPECTIVE

13

tifying the subunit structure of the polymerases and the genes that encode them. Remarkable techniques are provided by genetic engineering for amplifying the levels of these scarce enzymes: Novel methods are available for isolating enzymes, such as the use of monoclonal antibodies resulting from cell fusions; isotopic and staining methods, a hundredfold more sensitive, can be used for tracing the enzymes; and improved techniques are in hand for obtaining well-defined DNAs as template-primers for assay. Intensive studies of polymerase enzymology will be repaid by insights into the protein-nucleic acid interactions fundamental to gene expression and by a better understanding of the factors that control replication, recombination, and repair of DNA.

This Page Intentionally Left Blank

DNA Polymerase I of Escherichia coli I. ROBERT LEHMAN

. . . . .

I. Purification . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . , , . . . . . A. Multiple Functions of DNA Polymerase I . . . . . B. General Features of the Polymerization Reaction . C. Processivity of Polymerization . . . . , . . . . . D. 3’ + 5‘ Exonuclease , . . . . . . . , . . . . . E. 5’ + 3’ Exonuclease . . . . . . . , , . . . , . IV. Biological Role . . . . . , . . . . , , . . , . . . A. DNARepair . . . . , . . . . . , . . . . . . . B. DNA Replication . . . . . . . . . . . . . . . . V. Research Applications . . . . . . . . . . . . . . . A. Preparation of Highly Radioactive DNA Probes . . B. Molecular Cloning . . . . . . . . . . . . . . . C. DNA Sequencing . , . . . , . . , , . . . , . .

11. Physical Properties 111. Reaction Catalyzed

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

. .

.

. . .

. . . . . . .

. .

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

. . . . . . . . . .

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

.

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

16 16 17 17 18 19 27 28 29 29 30 35 35 35 36

DNA polymerase I (pol I) of E. coli catalyzes the polymerization of nucleotides at the direction of a nucleic acid template. It was the first such erizyme to be discovered ( I ) . Since Pol I was the subject of a detailed review in 1974, in Volume X of this series (2), the major focus of this chapter is on knowledge of the enzyme that has developed during the ensuing 6 1. A. Kornberg, I. R. Lehman, M. Bessman, and E. S. Simrns, BBA 21, 197 (1956). 2. T.Kornberg, and A. Kornberg, “The Enzymes,” 3rd ed., Vol 10,p. 119, 1974. 15 THE ENZYMES, Vol. XIV Copyright 0 1981 by Academic Press. Inc. All rights of reproduction in any form reserved ISBN 0-12-122714-6

16

I . ROBERT LEHMAN

years; in particular, information regarding the processivity of polymerization, and the function of pol I in vivo as deduced from the analysis of mutant enzymes. 1.

Purification

Purification of pol I, previously a difficult task because of the relatively low concentration of the enzyme in wild-type E. coli (400 molecules per bacterium), has been greatly simplified by the molecular cloning of the polA gene in a lambda transducing bacteriophage (3). Extracts prepared from cells infected with this phage are up to one hundredfold enriched in pol I activity, and homogeneous enzyme can be obtained after approximately one hundredfold purification (4). This is to be compared with the five thousandfold purification previously required to achieve homogeneous enzyme (S). Pol I purified by this procedure is indistinguishable from that isolated from wild-type E. coli in its molecular weight and amino acid composition. II. Physical Properties

The amino acid sequence and three-dimensional structure of pol I have not yet been determined. However, considerable information is available regarding its structure and physicochemical properties (6 1, This information may be briefly summarized as follows. DNA polymerase I consists of a single polypeptide with a molecular weight 109,000 and an sz0,,+, value of 5.5-5.6 (5). It appears to be folded into two domains joined by a protease-sensitive peptide linker (7). The protein is approximately spherical, with a diameter of 65 A as determined from its hydrodynamic properties (8) and from electron microscopy (9). As judged by ORD measurements it contains a significant amount of alpha-helical structure (1%45%) (5). Pol I contains a single free sulfhydryl W. S. Kelley, K. Chalmers, and N. E. Murray, PNAS 74, 5632 (1977). W. S. Kelley and K. H. Stump, JBC 254, 3206 (1979). T. M. Jovin, P.T. Englund, and L. Bertsch, JBC 244, 6996 (1969). A. Kornberg, “DNA Replication,” Freeman, San Francisco, 1980. D. Brutlag, M. R. Atkinson, R. Setlow, and A. Kornberg, BBRC 37, 982 (1%9); H. Klenow and I. Henningsen, PNAS 65, 168 (1970). 8. P. T. Englund, M. P. Deutscher, T. Jovin, R. B. Kelly, N. R. Cozzarelli, and A. Kornberg, CSHSQB 33, 1 (1968). 9. J. Gnffith, J. A. Huberman, and A. Kornberg, J M B 55, 209 (1971). 3. 4. 5. 6. 7.

17

2. DNA POLYMERASE I OF Escherichia coli

group that can form a dissociable complex with HgZ+ (10) and can be carboxymethylated without loss of activity (5). As determined by atomic absorption spectroscopy, pol I contains one Zn2+atom per molecule (/I). As yet there is no direct evidence for the participation of this metal in the catalytic cycle of the enzyme. Indeed, the finding that the Zn2+chelator, o-phenanthroline is a potent inhibitor of pol I can no longer be taken as evidence for the involvement of Zn2+in pol I action. It has been found that the inhibition by phenanthroline is due to an inhibitory phenanthrolineCu(1) chelate, which forms in the presence of the trace contaminants of Cu2+and thiols present in pol I assay mixtures (12). An analysis of pol I by a variety of techniques including equilibrium dialysis and spectroscopic examination (NMR and ESR), as well as kinetic measurements, has yielded a picture of the active center of the enzyme in which there are three closely juxtaposed sites: (1) A DNA template-binding site, (2) a nucleoside-monophosphate-binding site, which presumably represents the site at which the primer and growing chain are bound, and (3) deoxynucleoside triphosphate site, which accommodates all four triphosphates (6 ). 111.

Reaction Catalyzed

A. MULTIPLEFUNCTIONS OF DNA POLYMERASE I (6) DNA polymerase I catalyzes the addition of mononucleotide units from deoxynucleoside 5'-triphosphates to the 3'-hydroxyl terminus of a primer chain. For this reaction, a template is absolutely required and directs the enzyme in its selection of the specific triphosphate according to the Watson-Crick base-pairing rules. This reaction may be written: (dNMP).

+ dNTP

(dNMPL,,

+ PP,

The turnover number for the enzyme in this reaction is 667 nucleotides polymerized per molecule of enzyme per minute at 37". In addition to the polymerization of nucleotides, pol I catalyzes the pyrophosphorolysis of DNA, a reaction which in fact represents the reversal of polymerization. (dNMP).,,

+ PP,

(dNMPX, + dNTP

10. T. M. Jovin, P. T. Englund, and A. Kornberg, JBC 244, 3009 (1969). 11. J. P. Slater, A. S. Mildvan, and L. A. Loeb, BBRC 44, 37 (1971). 12. D. S. Sigman, D. R. Graham, V. D'Aurora, and A. Stem, JBC 254, 12269 (1979).

18

I. ROBERT LEHMAN

It also catalyzes pyrophosphate exchange according to the following reaction: (dNMP),

+ dNTP + PP,* %(dNMP). + dNMP-PP* + PP,

This reaction represents a repetitive sequence of nucleotide addition and pyrophosphorolysis of the newly synthesized phosphodiester bond. In addition to nucleotide polymerization and phosphorolysis, pol I catalyzes the hydrolysis of phosphodiester bonds. There are two such hydrolytic activities; one is a 3‘ + 5‘ exonuclease and the other a 5‘ + 3‘ exonuclease. The two exonuclease activities are associated with the two different domains of the pol I molecule. Thus, exposure to subtilisin cleaves the enzyme into two active fragments, a “large” fragment (MW 76,0001, which contains polymerase and 3‘ + 5’ exonuclease activities, and a “small” fragment (MW 36,000), which contains the 5‘ -+3‘ exonuclease. The small fragment corresponds to the amino terminal portion of the molecule as judged by the identity of the N-terminal amino acid sequences of the intact enzyme and small fragment (13), and the finding that an amber peptide isolated from the polA1 mutant retains the 5‘ 3 3’, but not the polymerase activity of the enzyme (14).

B. GENERAL FEATURES OF THE POLYMERIZATION REACTION(6) DNA polymerase I can catalyze three different modes of polymerization: Gap filling, nick translation, and strand displacement. It is unique among DNA polymerases in its capacity to promote the two latter reactions (Fig. 1). 1. In gap filling, pol I acts to fill in the single-stranded regions of a gapped DNA duplex by nucleotide addition to the 3’-hydroxyl group at the gap. 2. Nick translation consists of the coordinated polymerization of nucleotides at the 3’-hydroxyl terminus and the exonucleolytic removal of nucleotides at the 5’ terminus of a nick in duplex DNA. The result of the concerted polymerization and 5‘ + 3‘ exonuclease action is the propagation or translation of the nick along the DNA duplex. 3. Strand displacement consists of polymerization by pol I uncoupled from 5’ -+ 3’ exonuclease action, so that nucleotide addition at the 3‘hydroxyl terminus at a nick displaces rather than hydrolyzes the 5’terminated strand ahead of it. 13. H. Jacobsen, H. Klenow, and K. Overgaard-Hansen, EJB 45, 623 (1974). 14. I. R. Lehman and J. R. Chien, JBC 248, 7717 (1973).

2. DNA POLYMERASE I OF Escherichia coli NICKEDDNA

U

w

3’

NICK TRANSLATION

19

1I I II I I I I I I I Ill

,;. Yr.LLull \ \ /-

STRAND DISPLACEMENT

GAPPED DNA

GAP FILLING

m

3’

f

FIG.1. DNA polymerase I action at a nick or gap in duplex DNA. Nicked DNA is subject to nick translation or strand displacement; gapped DNA can be filled in to generate a nick, which can then support nick translation or strand displacement. In each of these three modes of pol I action, the basic reaction is the same: Addition of a deoxynucleotidyl unit onto a 3’-hydroxyl terminus with the resultant pairing of the added nucleotides to the complementary nucleotide on the template strand. The extraordinary fidelity of DNA replication by pol I is a consequence of the high degree of selectivity during polymerization and the associated 3‘ -+ 5’ exonuclease (see Sections II1,C and D below). A kinetic analysis of polymerization with a defined series of primer-templates indicates that the presence of the template may facilitate rejection of the noncomplementary nucleotide (15). The template may also increase the a n ity of pol I for the incoming complementary nucleotide. Both of these features of polymerase action could contribute to the fidelity of polymerization. Further to this point, an NMR analysis of the conformation of deoxynucleoside triphosphate-Mn(I1) complexes bound to pol I in the absence of template has shown that there is puckering of the complex such that the base-deoxyribose conformation is close to that found in double helical DNA (16). Formation of such a structure should permit insertion of the nucleotide into the helix in proper alignment for WatsonCrick base pairing and would contribute to the fidelity of replication. The role of the 3’ + 5’ exonuclease component of pol I in ensuring the fidelity of replication is discussed in Section II1,D below.

c.

PROCESSIVITY OF POLYMERIZATION

An important question regarding the mechanism of pol I is whether the polymerization of nucleotides proceeds in a processive or nonprocessive 15. E. C. Travaglini, A. S. Mildvan, and L. A. Loeb, JBC 250, 8647 (1975). 16. D. L. Sloan, L. A. Loeb, A. S. Mildvan, and R. J . Feldman,JBC 250, 8913 (1975).

20

I. ROBERT LEHMAN

(distributive) manner. A processive reaction is one in which many nucleotides are incorporated into the growing chain before the enzyme dissociates. A nonprocessive mechanism is one in which dissociation of the polymerase from the primer-template occurs following addition of each nucleotide. To state this point somewhat differently, a nonprocessive mechanism is one in which dissociation of the enzyme from the primertemplate after each nucleotide addition is an obligatory feature of the catalytic cycle. Three approaches have been taken to determine the processivity of pol I-catalyzed deoxynucleotide polymerization: (1) Template challenge experiments, (2) determination of nucleotide incorporation ratios, and (3) measurement of the decrease in polymerization rate upon removal of one or more of the four deoxynucleoside triphosphates. 1.

Template Challenge

Template challenge experiments measure the rate at which the enzyme can redistribute itself between competing templates. In one such study, an excess of poly(dC) oligo(dG) was added to a reaction mixture in which pol I was replicating poly(dA) oligo(dT) (17). Incorporation of dTTP was promptly inhibited and, correspondingly, incorporation of dGTP began. The conclusion from these studies was that nucleotide polymerization by pol I is distributive. However, the time scale of the experiment was such that 100 nucleotides or more may have been polymerized before template switching occurred. In another experiment in which calf thymus DNA was added as the competing template and the temperature was lowered to 4" to slow polymerization, it was found that within the time required for turnover of one nucleotide the calf thymus DNA could compete effectively with the d(A-T) copolymer that served as the initial template (18). Again, the conclusion reached was that pol I is distributive. In this instance an important limitation to the interpretation is the lack of information regarding the affinity of the enzyme for the different polynucleotides at 4". 2 . Nucleotide Incorporation Ratios This method for the assessment of processivity takes advantage of the nucleotide sequence of the cohesive ends of bacteriophage lambda DNA (Fig. 2). With this DNA a known sequence of nucleotides is incorporated into the cohesive ends, and furthermore the composition of the nu17. L. M. s. Chang,JMB 93, 219 (1975). 18. W. R. McClure and T. M. Jovin, JBC 250, 4073 (1975).

21

2. DNA POLYMERASE I OF Escherichiu coli left-hand end 3 5

right-hand end

C C C G C C G C T G G A (G)

GGGCGGCGACCT

3

FIG.2. Terminal sequences of bacteriophage lambda DNA. The letters represent the nucleotides incorporated when the cohesive ends are completely filled in. The bracket indicates the nucleotides incorporated if only dGTP and dCTP are used for synthesis. The letter in parentheses represents the 3'-terminal dG of the left-hand end of native lambda DNA. This residue is exchangeable with dGTP during synthesis in the absence of dATP. The cohesive ends are represented by solid lines. The long double-stranded internal region of the molecule is not shown. cleotides incorporated changes in a predictable way as polymerization proceeds ( 1 9 ) . When dGTP and dCTP are added only the nucleotides enclosed by the bracket are incorporated into the right-hand end, and there is no synthesis at the left-hand cohesive end. If nucleotide polymerization occurs by a processive mechanism, at a low molar concentration of enzyme relative to DNA, the ratio of dG to dC incorporated remains constant and does not change with the extent of reaction. On the other hand, if the mechanism is nonprocessive, early in the reaction a large proportion of the lambda DNA molecules incorporate only 1 or 2 dG residues into their right-hand cohesive end, and the ratio of dG to dC is high. As the reaction proceeds to completion the dG : dC ratio drops to a value of 3, the ratio of the number of dG and dC residues incorporated into the right-hand cohesive end. When this type of analysis was applied to pol I at 6",the ratio of dG to dC incorporated remained close to three throughout a 10-hour period of incubation (Fig. 3). Thus, under these conditions pol I is processive for at least the 12 nucleotide length of the right-hand cohesive end of lambda DNA. In a control experiment in which there was a 60-fold excess of enzyme over 3'-hydroxyl termini, the dG : dC ratio decreased from 17 at 30 seconds to a value of 3 at 5 minutes. An interesting feature of this experiment was the finding that even in the presence of a large molar excess of enzyme over DNA termini, the time required to complete synthesis at the direction of any one cohesive end was considerably shorter than that required to complete polymerization at all ends (20). Thus, regardless of whether pol I was present in excess or was limiting, about 30 to 40% of the enzyme mole19. D. Uyemura, R. Bambara, and I. R. Lehman JBC 250, 8577 (1976). 20. R. A. Bambara, D. Uyemura, and I. R. Lehman,JBC 250, 4090 (1976).

22

I. ROBERT LEHMAN

0

0

1

2

4

10

TIME (HOURS)

FIG.3. Nucleotide incorporation into the cohesive ends of lambda DNA by DNA polymerase I at 6" in the presence of dGTP, dCTP, and limiting enzyme (19). cules began polymerization rapidly, but synthesis on the remaining molecules was significantly delayed (Fig. 4). This delay was not a consequence of a low rate of association between enzyme and primer-termini at the cohesive ends, but rather of the slow onset of polymerization at all ends after association had occurred. A plausible explanation for this effect is that two forms of the enzyme or enzyme-DNA complex exist, only one of which is active. Inasmuch as 30 to 40% of the enzyme molecules begin polymerization immediately, the normal equilibrium between the two forms would give a ratio of inactive to active molecules of two. By this argument the rate of conversion from inactive to active enzyme (or enzyme-DNA complex) is the rate-determining step for the initiation of polymerization. Indeed one consequence of the defect in pol I from strains bearing thepolAI2 mutation (see Section IV,B,l) is apparently to alter the equilibrium such that only a relatively small fraction of active molecules are present at any time. However, the rate of conversion from the inactive to the active form is not significantly altered by the mutation.

23

2. DNA POLYMERASE I OF Eschcrichici coli I

E

0

-B 3000

.

EXCESS ENZYME

0 I-

2

g a

2000

E

a

zn

+a-

LIMITING ENZYME

B 1000

-I

M

-0 n

20

60

120

180

TIME (MINUTES)

FIG. 4. Comparison of rates of nucleotide incorporation into the cohesive ends of lambda DNA at excess and limiting concentrations of DNA polymerase I(20). 3 . Polymerization in the Absence of One or More Deoxynucleoside Triphosphates

Determination of the rate of polymerization in the presence of one, two, or three deoxynucleoside triphosphates compared to the rate in the presence of all four provides a quantitative measure of processivity over a large range of values (one to several hundred). Under these conditions the rate of polymerization is determined in part by processivity; the rate with a full (as compared to a partial) complement of deoxynucleoside triphosphates can be used to distinguish processivity from other factors that affect the rate of polymerization (2 1 ). With a large excess of primer-termini over enzyme molecules, nucleotide polymerization may be considered to be a two-phase cyclic process. In the first phase free enzyme diffuses to and binds the primertemplate. In the second phase the enzyme catalyzes an ordered succession of dNTP binding, nucleotide condensation, and translocation until the enzyme dissociates from the primer-template. The entire cycle is then repeated. The average increment of time during one complete cycle, from the binding and reaction of the DNA polymerase with a 3’-primer terminus, through the dissociation and diffusion steps, to the rebinding of the enzyme to another reactive 3’ terminus, is defined as the “cycling time.” 21. R. A. Bambara, D. Uyemura, and T. Choi,JEC 253, 413 (1978).

24

I. ROBERT LEHMAN

The polymerization rate whenx dNTPs are present may be expressed as follows:

P,

=

ENJT,

(1)

where P, is the rate of polymerization, E the number of active enzyme molecules, N , the average number of nucleotides incorporated per polymerization cycle, and T, the cycling time. This is a general expression that applies both to synthesis with a limited complement of dNTPs (x = 1, 2, or 3) and to synthesis with a complete complement of dNTPs (x = 4). For DNA containing all four bases, N4 is the average number of nucleotides polymerized with a complete complement of dNTPs, and is therefore the processivity of polymerization. The aim of the calculations that follow is to express N4 in terms of parameters that can be measured experimentally. The ratio of polymerization rates with a limited versus a complete complement of dNTPs (PXz4) may be expressed as

or P,:4 = N,T4/N4T,

(x = 1, 2, or 3)

(3)

Values of N , may ultimately be expressed in terms of N4 and two statistical factors. The TJT, ratio must be determined for individual experiments with the use of an inhibitor that binds the DNA polymerase for a short period of time during the first phase of the polymerization cycle. The presence of the inhibitor increases the cycling time, T,, by an increment, Ti such that

Tx,i = Tf + Tx

(4)

where T,,{ is the cycling time in the presence of the inhibitor. The inhibitor should be chosen so that the value of Tf is not affected by the value of x. This requirement can be verified experimentally because processivity dewould not be equivalent unless Ti were indetermined from P I , and pendent of x . Using this definition of Ti, T,/T4 can be expressed in terms of rates of reaction. First, it follows from Eq. (1) that

T, = EN,/P,

(5)

With this form of the equation the cycling times for inhibited and uninhibited reactions can be related directly to measurable rates:

25

2. DNA POLYMERASE I OF Eschcrichia Cali

where Px,i is the rate of reaction when the inhibitor is present. Substituting for Tx,i from Eq. (4)

Ti can be evaluated as follows:

From Eq. (7), when four dNTPs are used in the reaction -T4 +

Ti - &

T4

P4.f

Substituting for Ti in Eq. (9) by using the value of Ti in Eq. ( 8 )

and solving for Tx/T4

Returning to Eq. (3), the measured reaction rates Px, P4, Px,i, and P4,t can be related to the processivity of polymerization:

This expression is rearranged as follows:

and

Equation (13) may also be rewritten

26

I. ROBERT LEHMAN

where P&, = P,:,T,/T,. Since only a single triphosphate is labeled in all experiments, a correction is made to allow use of the ratio of rates of radioactive nucleotide incorporation (R::,). The factor L relates RI ,: to P&, as follows:

R1.4

=

LP;:4

(16)

This expression is incorporated into Eq. (15) to yield N4 = LN,/R::,

(17)

The theoretical relationship between N4 and R::, for calf thymus DNA is shown in Fig. 5. The value of R::, is given by R::, = Rx:4(Tx/T4)

(18)

where R,:, is the ratio of label incorporated in limited (x = I , 2 , or 3) as compared to complete ( x = 4) reactions and T,/T, is calculated from Eq. (11).

An important limitation to the analysis is the possibility that the average length of the template is shorter than the extent of processivity. If so, the observed rate of polymerization with a full complement of dNTPs will be lowered and the ratio of polymerization rates will be increased, leading to an artifactually low processivity. The average length of template available to a DNA polymerase at each 3' terminus can be determined when there is an excess of enzyme molecules over primer-termini, and extents rather than rates of polymerization are measured. An equation similar to that used in the calculation of processivity can be used. In this case A,:, = LA,/A,

( x = I , 2 , or 3)

(19)

where Ax:,is the ratio of extents of nucleotide incorporated with a limited and a complete complement of dNTPs, and A, and A, are the average maximum number of nucleotides that can be added to a 3' terminus with a limited and complete complement of dNTPs, respectively. The relationship between A,:,, A,, and A, is the same as that between R,:,, N,, and N4 (see above). IfA, = N4 it may be assumed that template length limits the measured processivity of DNA polymerization. If A, > N4 the processivity may be assumed to be a consequence of the intrinsic tendency of the DNA polymerase to dissociate under the conditions of measurement. An analysis of the processivity of pol I by this method has shown that the enzyme can be highly processive, with values as high as 200, depending upon the primer-template and the conditions of polymerization. However, the processivity decreases at high ionic strength and at low temperatures. With nicked DNA templates at 37" and ionic strength 0.085,

2. DNA POLYMERASE I OF Escherichia coli

27

200 100

> 20

f

10

E

5

0

2 1

0.01

0.1

1.o

RATIO OF INCORPORATION RATES (R',,q)

FIG.5 . Theoretical relationship between RiZ4and processivity ( N ) , with activated calf thymus DNA as a primer-template. Curves are shown for five combinations of deoxynucleoside triphosphates in the limited reaction. The abscissa is the ratio of label incorporated @?::4) and the ordinate is the processivity ( N ) . The curves represent limited reactions containing (1) PHIdTTP: (2) PHIdTTP + dCTP or [3H]dTTP+ dGTP; (3) PHIdlTP + dCTP; (4) rH]dTTP + dCTP + dGTP; and ( 5 ) l3H]d=P + dATP + dCTP or PHJdlTP + dATP + dGTP ( 2 1 ) . processivity is 15 t o 20. With gapped DNA under the same conditions, the processivity increases to 40 to 50. With poly d(A-T), the processivity is 188. However, by increasing the ionic strength and decreasing the temperature, this value is reduced to 3. Of interest in this regard is the finding that the enzyme isolated from the polA5 mutant gives processivity values that are only one-fifth of wild-type pol I. To summarize then, pol I binds rapidly t o its primer-template. This is followed by a significant delay in the onset of polymerization because of a slow shift in the equilibrium between the inactive and active forms of the enzyme. Polymerization is processive, the extent of processivity being determined by a variety of factors (temperature, ionic strength) that appear t o influence the structure of the primer-template and its interaction with the enzyme.

D.

3'

+

5' EXONUCLEASE

The 3' + 5' exonuclease associated with pol I acts specifically on single-stranded DNA and hydrolyzes it to nucleoside 5'-monophosphates

28

1. ROBERT LEHMAN

(22). Of particular importance is its ability to remove a non-base-paired terminus. Thus, given a primer-template with one or more mismatched nucleotides at the 3' terminus, pol I removes the mismatched nucleotides by virtue of its 3' +. 5' exonuclease component before initiating polymerization. The removal of mismatched nucleotides by the 3' + 5' exonuclease thus represents a proofreading mechanism that contributes to the high fidelity of DNA replication (23). Following infection with phage T4, a DNA polymerase is induced with a similar 3' + 5' exonuclease, and defects in the balance of polymerase and exonuclease in various mutant phages have been correlated with mutator and antimutator effects observed in vivo (24,251. Thus far, no such effects have been observed for pol I isolated from E. coli polA mutant strains (see below). As described in the preceding section, the selective orientation of the incoming nucleotides by the template and the induced puckering of the deoxyribose-base bond may also be important in maintaining the fidelity of replication at the level of proper base selection. In fact, a "mutator" pol I isolated from Salmonella typhimurium appears to be defective in base selection rather than its 3' + 5' exonuclease (26).

E. 5'

+

3' EXONUCLEASE

The 5' + 3' exonuclease function of pol I degrades duplex DNA in the same direction as polymerization, starting at the 5' terminus of a nick in the duplex (27). In contrast to the 3' 5' exonuclease of pol I, it is inactive on single-stranded DNA; furthermore, it produces 5' phosphoryl-terminated oligonucleotides in addition to mononucleotides (28).The 5' -+3' exonuclease can excise damaged nucleotides from DNA following incision of the DNA duplex at or near the site of damage (29), and it can degrade the RNA component of a DNA RNA hybrid (30). As

+.

22. I. R. Lehman and C. C. Richardson, JBC 239, 233 (1964). 23. D. Brutlag and A. Kornberg, JBC 247, 241 (1972). 24. N . Muzyczka, R. L. Poland, and M . J. Bessman, JBC 247, 7116 (1972). 25. F. D. Gillin and N . G . Nossal, JBC 251, 5219, 5225 (1976). 26. M. Engler, and M. J. Bessman, C S H S Q B 43, 929 (1978). 27. R. P. Klett, A. Cerarni, and E. Reich, PNAS 60, 943 (1968). 28. N . R. Cozzarelli, R. B. Kelly, and A. Kornberg, JMB 45, 513 (1969). 29. R. B. Kelly, M. R. Atkinson, J. A . Huberman, and A. Komberg, Nature (London) 224, 495 (1969). 30. D. Baltimore, and D. F. Smoler,JBC 36, 185 (1972); J. P. Lies, I. Berkower, and J. Hunvitz, PNAS 70,466 (1973);0. Westergaard, D. Brutlag, and A. Kornberg,JBC 248, 1361 (1973).

2. DNA POLYMERASE I OF Escherichiu coli

29

indicated previously, 5‘ -+ 3’ exonuclease action is coupled with nucleotide polymerization at a nick in DNA, resulting in propagation or translation of the nick along the duplex (31). The process of nick translation ensures that long single-stranded stretches do not accumulate within a DNA duplex that is undergoing repair or replication. Although it was believed that the 5’ += 3’ exonuclease requires a 5’ terminus to initiate hydrolysis, it is now clear that under certain circumstances pol I can catalyze an endonucleolytic scission even in the absence of a terminus. This has been observed with negatively supertwisted circular DNA into which a single-stranded DNA or RNA segment has been assimilated to form a D loop. When such a structure is used by pol I as a primer-template the primer is elongated, resulting in an enlargement of the loop. In addition, however, endonucleolytic cleavage of the circular DNA occurs (32). This may be a consequence of 5’ 4 3’ exonuclease action at the strand not paired to the primer. The replication fork generated by polymerase action leads to a distortion in the duplex, which resembles that introduced by, for example, a thymine dimer and, hence, may become a site for the endonucleolytic action of the 5‘ + 3’ exonuclease.

IV.

Biological Role

Strains of E. coli with mutations in the polA gene are abnormally sensitive to ultraviolet irradiation and to radiomimetic agents (.33),and they are defective in chromosomal DNA replication (34). DNA polymerase I is therefore required for both the repair and the replication of DNA in vivo . Pol I is also essential for the replication of certain plasmids, for example the colicinogenic factor, ColE1 (35). A.

DNA REPAIR

The capacity of pol I to promote nick translation is clearly a key feature in its repair function. Thus, following incision at or near the site of damage (for example, a thymine dimer introduced following UV irradiation), pol I 31. R. B. Kelly, N . R. Cozzarelli, M. P. Deutscher, I. R. Lehman, and A. Kornberg, JEC 245, 39 (1970). 32. L. F. Liu and J. C. Wang, In “DNA Synthesis and Its Regulation” (M. Goulian and P. Hanawalt, eds.), p. 38. Benjamin, Menlo Park, 1975. 33. P. L. De Lucia, and J. Cairns, Nature (London) 224, 1164 (1%9). 34. P. L. Kuempel, and G . W. Veomett, BBRC 41, 973 (1970); R. Okazaki, M. Arisawa, and A. Sugino, PNAS 68, 2954 (1971). 35. D. T. Kingsbury, and D. R. Helinski. BBRC 41, 1538 (1970).

30

I. ROBERT LEHMAN

can catalyze the concerted 5' + 3' exonucleolytic removal of a DNA segment that includes the dimer, and the 5' + 3' polymerization required to restore the DNA duplex (31). Inasmuch as polA mutants are only moderately sensitive to UV irradiation, it is clear that alternative excision-repair mechanisms must exist that can substitute for pol I when the latter is defective (36). Indeed, pol I1 and/or pol I11 (which has an intrinsic 5' + 3' exonuclease) have been implicated in what has been termed "long-patch" repair, as opposed to the "short-patch" repair involving little degradation and the limited resynthesis characterization of wild-type strains with a fully functional pol I (36). B. DNA REPLICATION The joining of nascent DNA fragments is retarded in allpolA mutants that have been examined, indicating that pol I is required for the discontinuous replication of the E. coli chromosome (37). The severity of the defect differs considerably from one mutant strain to another; however, it is most pronounced at restrictive temperatures in conditionally lethal, temperature-sensitive mutants (37, 38). As in excision repair, the unique capacity of pol I to catalyze nick translation is the crucial feature of its function in discontinuous DNA replication. Thus, pol I can promote the coordinated 5' + 3' exonucleolytic removal of DNA primers at the 5' termini of Okazaki fragments, and the filling in of the gaps thus created (31). This coupled reaction permits the ligation of discontinuously synthesized DNA fragments to the growing DNA chain. polA mutants have been identified whose pol I is defective in the polymerase, the 5' + 3' exonuclease, and the capacity to coordinate the two. In each instance, there is an abnormal persistence of nascent DNA (Okazaki) fragments. The latter two mutant enzymes have been purified to homogeneity and are considered below. 1. Nonlethal PolA I2 Mutant

Strains bearing thepolA12 mutation show a temperature sensitive repair defect; that is, they are defective in the repair of DNA damage at 43", but not at 30" (39). They are, however, viable at both temperatures. The polAI2 polymerase is extremely thermosensitive. It is also rapidly dena36. P. Hanawalt, A. Burrell, P. Cooper, and W. Masker, In "DNA Synthesis and Its Regulation" (M. Goulian and P. Hanawalt, eds.), p. 774. Benjamin, Menlo Park, 1975. 37. I. R. Lehman, and D. Uyemura, Science 193, %3 (1976). 38. D. Uyemura, D. C. Eichler, and I. R. Lehman, JBC 251, 4078 (1976). 39. M. Monk, and J. Kinross, J . Bacteriol. 109, 971 (1972).

2. DNA POLYMERASE I OF Escherichirr coli

31

tured by even brief exposure to ionic strengths less than 0.1. Thus, purification of the mutant enzyme requires procedures that avoid or minimize exposure to solutions of low ionic strength (40). The temperature sensitivity and instability at low ionic strength of the polAI2 DNA polymerase appear to result from a significant alteration in the tertiary structure of the enzyme. The mutant protein has a significantly lower mobility than the wild-type enzyme in discontinuous polyacrylamide gel electrophoresis of the two native proteins. Furthermore, the polAI2 enzyme sediments at a lower rate than the wild type enzyme in a sucrose velocity gradient. The decrease in electrophoretic mobility taken together with the lower sedimentation coefficient suggests that thepolAI2 mutation has produced a misfolding of the mutant protein so that it is less compact than the wild-type enzyme. This is possibly the cause of its thermal instability and rapid inactivation in low salt solutions. A striking feature of the polAI2 enzyme, even at permissive temperatures, is its decreased ability to catalyze nick translation (40). This point is illustrated in Fig. 6. Given nicked PM2 DNA, wild-type pol I catalyzes the incorporation of nucleotides at the 3'-hydroxyl end and the release of nucleotides from the 5' end of the nick in equimolar amounts at both 30" and 43". With gapped PM2 DNA, the gaps are rapidly filled in, regenerating the nicks, which then become sites for nick translation (Fig. 7). The incorporation and release of nucleotides at a nick catalyzed by the polA I2 enzyme are also equivalent. However, the rate at which nick translation proceeds is tenfold lower than that seen with the wild-type enzyme at 30", and is even further reduced at 43" (Fig. 6). With DNA that contains gaps, the mutant enzyme rapidly fills in the gaps at nearly the same rate and to the same extent as the wild-type enzyme, then catalyzes very little further synthesis or release of nucleotides (Fig. 7). Since measurements of the polymerase and 5' + 3' exonuclease activities associated with thepolAI2 enzyme show them to be nearly normal at 30" (40), the abnormally low rate of nick translation suggests that there is a substantial defect in the coordination of polymerization and 5' -+ 3' exonuclease action. Since such coordination presumably demands a rigid spatial arrangement of the two active sites, this novel defect may be a consequence of the structural perturbation caused by the polAI2 mutation.

2. Conditionally Lethul polAexI Mutant The isolation of the temperature-sensitive, conditionally lethal mutant, E. coli polAexI, established that DNA polymerase I is essential for the 40.

D. Uyemura, and I . R. Lehman, JBC 251, 4085 (1976).

32

I. ROBERT LEHMAN 800

600

400

200

0 0

5

10

200

5

10

20

TIME (MINUTES 1

FIG.6. Action of wild-type and polA I2 DNA polymerases on nicked PM2 DNA (37). viability ofE. coli (41). Except for its conditional lethality,pofAex/ resembles otherpolA mutants in its retarded sealing of nascent DNA fragments, and in its sensitivity to methylmethane sulfonate and to ultraviolet irradiation. However, the rate of joining of nascent DNA fragments in the polAexl mutant is significantly more retarded than in nonlethal polA strains (38). In contrast to the polA12 enzyme, the pofAex1 polymerase is not particularly labile in low salt solutions. The pofAex1 protein comigrates with wild-type pol I in polyacrylamide gels containing sodium dodecyl sulfate as well as in native discontinuous, polyacrylamide gels; it also has the same sedimentation coefficient. Thus, there is no obvious structural alteration comparable to that observed with the polA12 polymerase (38). The polymerase and 3' 5' exonuclease activities of thepolAex1 enzyme do not differ significantly from those of the wild-type DNA polymerase I at either 30" or 43". In contrast, the 5' + 3' exonuclease activity is substantially reduced at both temperatures (Table I). Furthermore, the 5' + 3' exonuclease activity at 43" is significantly lower than at 30". With nicked PM2 DNA (Fig. 8) as template primer, thepolAex1 enzyme catalyzes the incorporation of nucleotides at a rate far in excess of the rate of hydrolysis of nucleotides from preexisting DNA. This contrasts sharply with the action of the wild-type polymerase, which maintains an almost perfect correspondence between nucleotide release and incorporation.

-

41. E. B. Konrad and I. R. Lehman, PNAS 71, 2048 (1974).

2. DNA POLYMERASE I OF Escherichiu coli

33

z a E K K

8

z

FIG.7. Action of wild-type andpolAI2 DNA polymerases on gapped PM2 DNA (37). The low rate of nucleotide release by the polAexl enzyme is presumably due to some nick translation that occurs at 30". The increment of nucleotide incorporation beyond the amount released must therefore be due to strand displacement; that is, polymerization in the 5' + 3' direction accompanied by unwinding of the strand preceding the enzyme molecule. The discrepancy between the rates of nucleotide incorporation and release is even greater at 43" than that at 30°, probably as a result of an increase in the rate of polymerization coupled with the decrease in 5' + 3' exonuclease activity at the elevated temperature. The polymerase activity TABLE I

DEFECTIVE5'

+

3' EXONUCLEASE ACTIVITYOF DNA POLYMERASE I ISOLATED FROM E. coli polAexl Enzyme activity (Nmol/mg protein)

W)

Polymerase

5' -+ 3' exonuclease

3' -+ 5' exonuclease

30 43 30 43

14.8 41.9 23.8 66.0

6.1 11.8 0.33 (0.19p

1.6 3.4 2.5 3.1

Temperature Enzyme polA+ polAexl

A portion (25-50%) of the apparent 5' -D 3' exonuclease activity at 43" may be attributable to 3' -D 5' exonuclease action. This estimate is based on the extent of hydrolysis observed on incubation of the nicked PM2 DNA with T4 DNA polymerase, which has 3' -P 5' but no 5' -+ 3' exonuclease activity (37).

34

1. ROBERT LEHMAN 600

I

400 300 -

I

1

30'

-

5

1

- - 430

500

0

1

-

10

20

-

0

p

I

?-!--

0

5

10

20

TIME (MINUTES)

FIG.

8. Action of wild-type and polAexl polymerases on nicked

PM2 DNA (37).

of the mutant is lower than that of the wild-type enzyme in this experiment (compare Fig. 8 and Table I). This may be a consequence of the 5' + 3' exonuclease defect of the mutant. When 5' + 3' exonuclease is reduced, the 5' terminated strand must be displaced ahead of enzyme molecule for polymerization to proceed (Fig. l), and this constraint might be expected to lower the polymerization rate. It therefore appears that, at 30°, the mutant enzyme can catalyze nick translation (at a low rate) and polymerization accompanied by strand displacement. At 43", nick translation is abolished and polymerization proceeds only with strand displacement. 3. Other PolA Mutants

Pol I has been purified to homogeneity from three otherpolA mutant strains, pofA'107 (42), polAS (43), and p0lA6 (44). In none of these instances is the mutation lethal. ThepolA'l07 enzyme is defective in the 5' + 3' exonuclease but not polymerase. Thus, it is similar to the polAexl enzyme. A direct comparison of the extent of the 5' + 3' exonuclease defect with that in thepolAexl mutant has not been made. ThepolA6 mutant enzyme shows an altered pH optimum for polymerization and a reduced binding affinity for DNA. The mutational defect 42. H. L. Heijneker, D. J. Ellers, R. H. Tjeerde, B. W. Giickman, B. van Dorp, and P. H. Pouwels, Mol. Gen. Genet. 124, 83 (1973). 43. S. W. Matson, F. N. Capaldo-Kimball, and R. A. Bambara, JBC 253, 7851 (1978). 44. W. S. Kelly and N. D. F. Grindley, Nucleic Acids Res. 3, 2971 (1976).

2. DNA POLYMERASE I OF Escherichitr cdi

35

appears to lie within the carboxyl-terminal large fragment of the enzyme because the fragment has the same pH optimum for polymerization as the intact enzyme, which is considerably more alkaline than that of the wildtype pol I. The large fragment also shows the reduced binding affinity for DNA that characterizes the intact mutant enzyme. The polA5 mutation which results in a decreased processivity of polymerization was mentioned in Section 111,C,3.

V. Research Applications

pol I has long been useful in the synthesis of defined homo- and copolymers [e.g., d(A-T), d(G-C), poly(dA)(dT)] that have served as well-defined structures for the physicochemical analysis of nucleic acids. More recently pol I has become a key reagent in the preparation of radioactive DNA probes, in molecular cloning, and in DNA sequence analysis. A. PREPARATION OF HIGHLYRADIOACTIVE DNA PROBES The capacity of pol I to catalyze nick translation has been used as an effective method for the preparation of highly labeled DNAs. Such labeled DNAs, and restriction endonuclease fragments derived from them, are then used as probes for detecting homologous sequences by measurements of reassociation kinetics or by in situ hybridization techniques. Indeed, labeled nick-translated SV40 DNA has been used to detect and quantitate DNA sequences present at the level of one SV40 DNA copy per haploid mouse genome, using only microgram quantities of cellular DNA (45).

B. MOLECULARCLONING The ability of pol I to fill in small gapped regions in duplex DNA molecules completely and efficiently has been used in the construction of recombinant DNA molecules in v i m . Using a closed circular duplex DNA molecule as a vector, the procedure involves cleaving the molecule to convert it to a linear structure, adding single-stranded homopolymeric stretches to the 3' termini with terminal transferase, adding the complementary homopolymeric sequence to the 3' termini of the DNA segment to be inserted, and annealing the segment to the linear form of the vector. At this point, gaps at the annealed juncture remain and are filled in 45. P. W. J. Rigby, M. Dieckmann, C. Rhodes, and P. Berg, J M B 113, 237 (1977).

36

I. ROBERT LEHMAN

with pol I so that covalent joining can be accomplished by DNA ligase (46 1.

C. DNA SEQUENCING Pol I is a key reagent in three methods for the sequence analysis of DNA. Each depends upon the ability of pol I to copy a particular sequence of single-stranded DNA starting from the terminus of the annealed primer. 1. Partial Ribonucleotide Substitution (47)

Pol I has the unique ability to incorporate a ribonucleotide in place of the corresponding deoxynucleotide when Mg2+ is replaced by MI?+ (48). The incorporated ribonucleotide then becomes a site of base specific cleavage of the chain by alkali, to yield fragments terminated by that particular ribonucleotide.

2. The Dideoxy dNTP Method (49) Pol I can incorporate the 2’-, 3’-dideoxy analogs of the deoxynucleoside triphosphates into a suitably annealed DNA primer, and in doing so block further chain growth (50). Using a known ratio of analog to each of the natural deoxynucleoside triphosphates results in an accumulation of chains at each point in the sequence at which the nucleotide occurs. As in the ribonucleotide insertion method, a family of fragments terminated with a specific nucleotide is obtained for further sequence analysis. 3 . The Plus-Minus Method (51) This method uses both pol I and the TCinduced DNA polymerase under conditions of limiting deoxynucleoside triphosphates. In the minus reaction, one of the four deoxynucleoside triphosphates is omitted from each of four separate reaction mixtures. Synthesis by pol I proceeds until the point in the sequence at which the missing deoxynucleoside triphosphate is required. For example, in the absence of dATP each chain is terminated at the 3‘ end before an A residue. In the plus reaction, synthe46. I. R. Lehman, Science 186, 790 (1974). 47. W. M. Barnes, J M B 119, 83 (1978). 48. P. Berg, H. Fancher, and M. Chamberlin,ln “Informational Macromolecules” (H. J. Vogel, B. Bryson, and J. 0. Lampen, eds.), p. 467. Academic Press, New York, 1963. 49. F. Sanger, S. Nicklen, and A. R. Coulson, PNAS 74, 5463 (1977). 50. M. R. Atkinson, M. P. Deutscher, A. Kornberg, A. F. Russell, and J. G . Moffatt, Biochemistry 8, 4897 (1969). 51. F. Sanger and A. R. Coulson, J M B 94, 441 (1975).

2. DNA POLYMERASE 1 OF Esrhrrichici coli

37

sis by T4 DNA polymerase, in the presence of a single triphosphate (dATP in this instance) results in all chains terminating at the 3' end with an A residue. A similar reaction is run with each of the other three deoxynucleoside triphosphates. After removal of the primer by cleavage with a restriction endonuclease, the nucleotide sequence of the chain that has been replicated can be deduced from the position of the bands in the eight reaction mixtures in an autoradiograph.

ACKNOWLEDGMENTS This work was supported in part by grants from the National Institutes of Health (GM 061%) and the National Science Foundation (PCM 74-00865).

This Page Intentionally Left Blank

DNA Polymerase 111 Holoenzyme CHARLES McHENRY

ARTHUR KORNBERG

I. Introduction . . . . . . . . . . . . . . . 11. DNA Polymerase I11 . . . . . . . . . . . . A. Detection and Isolation . . . . . . . . . B. Purification and Identification of Subunits . C. Nomenclature . . . . . . . . . . . . . D. Nuclease Activity . . . . . . . . . . . 111. DNA Polymerase 111 Holoenzyme . . . . . . A. Detection and Isolation . . . . . . . . . B. Purification and Subunit Structure . . . . C. Structural Genes for Holoenzyme Subunits D. Mechanistic Studies . . . . . . . . . . E. Physiological Role . . . . . . . . . . . 1V.Summary.. . . . . . . . . . . . . . . . Note Added in Proof , . , . . . . . . . . .

1.

. .

. . .

. . . .

. . . .

.

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

. . . . . . . .

. . .

. . .

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

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

. . . . . . . . .

. .

. . . .

.

.

.

. . . . .

. . . . . . .

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

39 40 40 41 41 42 43 43 43 46 47 48 49 49

Introduction

In this chapter we consider the structural, functional, genetic, and mechanistic aspects of the various forms of DNA polymcrase 111 of Escherichia coli. These forms are subassemblies of the subunits of DNA polymerase I11 holoenzyme. The holoenzyme has been implicated by both biochemical and genetic criteria as the polymerase responsible for replication of most of the E. coli chromosome. The nature of template-directed 39 THE ENZYMES, Vol. XIV Copyright 0 1981 by Academic Press. Inc. AU rights of reproduction in any form reserved ISBN 0-12-122714-6

40

CHARLES McHENRYANDARTHUR KORNBERG

DNA synthesis by DNA polymerases is reviewed in this volume Chapters 1-7, and elsewhere (I). DNA polymerases have generally been isolated using as an assay the filling of gaps created in duplex DNA by nuclease action. For years DNA polymerase I (pol I) was the only DNA polymerase recognized in E. coli, and was thought to be the polymerase solely responsible for synthesis of theE. coli chromosome. Subsequent studies (2) that led to the isolation of mutations in pol I (PolA), although erroneous in their original conclusion that pol I was dispensable for replication, sparked efforts to search for new DNA polymerases. The two enzymes found were designated DNA polymerase I1 (pol 11) (3-5) and DNA polymerase I11 (pol 111) (6). Beyond identification ofpolB as the structural gene for pol 11, little is known of the physiologic functions of this polymerase. Identification of pol I11 as the product of the dnaE (PofC)gene established it as an enzyme essential for DNA replication (7). Paradoxically, pol 111 demonstrated no unique properties in vitro that might distinguish it as a replicative enzyme. Later on, a more complex physiological form of pol 111, the DNA polymerase 111 holoenzyme, was isolated by using as an assay the in vitro replication of natural chromosomes rather than the repair of gaps in nuclease-damaged DNA. The natural chromosomes used were those of the single-stranded DNA phages M13, G4, and 4x174. The use of these phages to probe E. coli replicative functions is reviewed elsewhere (I, 8). As described in Section 111, the holoenzyme contains a pol I11 core and several auxiliary subunits that permit it to function as a natural replicative enzyme. II. DNA Polymerase 111

A. DETECTION AND ISOLATION Even with the availability of apolA mutant, the discovery of pol 111 was delayed by its lability and inhibition by even moderate levels of salt. Pol 1. A. Kornberg, “DNA Replication.” Freeman, San Francisco, 1980. 2. P. DeLucia and J. Cairns, Nature (London)224, 1164 (1969). 3. T. Kornberg and M. Gefter, BBHC 40, 1348 (1970). 4. R. Moses and C. C. Richardson, BERC 41, 1557, 1565 (1970). 5. R. Knippers, Nature (London) 228, 1050 (1970). 6. T. Kornberg and M. Gefter, PNAS 68, 761 (1971). 7. M. Gefter, Y. Hirota, T. Kornberg, J. Wechsler, and C. Barnoux, PNAS 68, 3150 (1971). 8. R. Schekman, A. Weiner, and A. Kornberg, Science 186, 987. (1974).

3.

DNA POLYMERASE 111 HOLOENZYME

41

I11 was initially detected as a peak of polymerase activity that was eluted at salt concentrations lower than those needed to elute pol I1 from phosphocellulose columns (6). After conditions were found that stabilized pol I11 and optimized its activity, the levels of pol I11 activity in extracts of polA cells approached those ofpol I inpolA+ cells. Maximal activity for pol I11 is attained at pH 7.0 in morpholinopropyl sulfonic acid buffer in the presence of minimal salt concentrations. Pol I11 is inhibited 50% by 50 mM KC1; addition of 150 mM KC1 results in complete inhibition. Ethanol (10%) stimulates pol I11 twofold and partially protects the enzyme from the inhibitory effects of salt (9). General comparisons of the properties of pol I11 with pol I and pol I1 have been reviewed elsewhere ( I ) .

B. PURIFICATION AND IDENTIFICATION OF SUBUNITS Purification attempts by two laboratories indicated that pol I11 contained at least a subunit of 140,000 daltons (10, / / ) . In these purification procedures, pol I11 was monitored by either the standard gap-filling assay (10) or by a complementation assay in which pol I11 stimulates replication by extracts of a dnaE mutant (11). Subsequently, pol I11 was purified 28,000-fold to 97% homogeneity (12). The resulting enzyme is composed of three subunits: a , E , and 8 of 140,000, 25,000, and 10,000 daltons, respectively. The three subunits chromatograph together on phosphocellulose and hydroxylapatite, and remain associated during gel filtration and electrophoresis on native acrylamide gels.

C. NOMENCLATURE DNA polymerase I11 (pol 111) is the term used for the simple gap-filling form of the enzyme originally isolated (6, 9) (Table I). The E and 8 subunits are not resolved from a using the chromatographic procedures described in any purification procedure (12). Thus, pol I11 is the simplest isolated enzyme that contains the dnaE gene product, and appears to be the catalytic core of the holoenzyme. The 140,000 dalton protein (a), which is the dnaE gene product (see Section III,C), will be referred to simply as the a subunit of the DNA polymerase I11 holoenzyme, whether or not it is found 9. T. Kornberg and M. Gefier, JBC 247, 5369 (1972). 10. D. Livingston, D. Hinkle, and C. Richardson, JBC 250, 461 (1975). 1 1 . B. Otto, F. Bonhoeffer, and H. Schaller, EJB 34, 440 (1973). 12. C. McHenry, and W. Crow, JBC 254, 1748 (1979).

42

CHARLESMcHENRYANDARTHURKORNBERG TABLE I

COMPONENTS OF DNA POLYMERASE 111 HOLOENZYME Subunits

Mass (daltons x

a

140

E

25 10 83 52 32 37

9 7

Y 6

P

Alternative designation dnaE protein, polC protein

-

dnaZ protein Factor 111, dnaX protein

Factor I, cop01 III*, dnaN protein

to exhibit any independent enzymatic activity. DNA polymerase I11 holoenzyme (or holoenzyme) is the term used for the complex of proteins that includes the core DNA polymerase 111; it is the intact soluble replicative polymerase responsible for most of the replication of the E. cofi chromosome. Intermediate forms of pol 111 that are subassemblies of pol I11 and holoenzyme auxiliary subunits will be referred to as pol I11 with a superscript (i.e., pol 111', pol III*). Individual subunits, or complexes of them, are referred to as such (i.e., p, y, 6, y . 6).

D. NUCLEASE ACTIVITY Pol I11 has both 3' + 5' and 5' 3' exonuclease activities (13). The 3' 5' activity is specific for single-stranded DNA and removes mismatched nucleotides from the 3' end of an otherwise duplex structure before continuing polymerization from the duplex 3'-OH terminus. Thus, like pol I, pol I11 has the capacity to proofread its product. The 3' -+ 5' exonuclease yields 5'-mononucleotides, but fails to degrade the 5'terminal dinucleotide. Unlike pol I, the 5' + 3' exonuclease of pol I11 is inactive on duplex DNA. It can degrade single-stranded DNA and proceed into duplex regions after degradation has begun. This indicates that pol 111 catalyzes 5' + 3' degradation processively (13). The role of this 5' .--, 3' activity in replication is not clear. It is incapable of removing duplex structures encountered during chain elongation. Possibly the 5' .--, 3' exonuclease is used for other functions, such as repair. The effect of the other holoenzyme subunits on the pol I11 nuclease activities is not known. -+

+

13. D. Livingston and C.

C. Richardson, JBC

250, 470 (1975).

3.

DNA POLYMERASE 111 HOLOENZYME

111.

DNA Polymerase 111 Hoioenzyme

A.

DETECTION AND ISOLATION

43

DNA polymerase 111 holoenzyme was discovered in the investigation of the conversion of single-stranded DNA phages to the duplex replicative form (14). The general reaction scheme for these studies [for a review see Refs. (I, S ) ] are summarized in Fig. 1. The phages M13, G4, and (6x174 differ in the mechanisms of primer formation. M13 DNA, coated with single-strand binding (SSB) protein, depends on the rifampicin-sensitive RNA polymerase; G4 uses the dnaC primase to synthesize its primer; 4x174 can use the dnaC primase only after the prepriming action of six other proteins (n, n’, n”, i, dnaB, and dnaC). Despite the diversity of priming systems, all primers are elongated by the action of the DNA polymerase I11 holoenzyme. DNA polymerase 111 alone will not substitute for holoenzyme, but holoenzyme, like pol 111, has been shown to depend on a product of thednaE (polC)gene (1.5). Early studies indicated that holoenzyme could be resolved by phosphocellulose chromatography into two components, pol III* and copol III* (14, 1.5). Pol IIP was distinguished from pol 111 by its ability to use a long, primed, single-stranded template in the presence of copol 111” and by its greater size and lability.

B. PURIFICATION AND SUBUNIT STRUCTURE The holoenzyme has been purified 7400-fold from HMS-83, apolA-poll3 strain (16). Using high-resolution, denaturing electrophoresis techniques ( I Z ) , 13 bands could be detected in this preparation, sedimenting as an 11 S entity in a glycerol gradient. Before a polypeptide can be regarded as a holoenzyme subunit more evidence than its mere presence in highly purified preparations is needed. Preliminary assignments of subunits have been based upon these criteria: (i) Purification of a subunit to homogeneity and a demonstration that it is essential for reconstituting holoenzyme activity on a natural chromosome. (ii) Physical association of a polypeptide with a known polymerase component in both the holoenzyme and a simpler subassembly of subunits; with this latter criterion, it is also necessary to show that the physical properties of the putative subunit change as 14. W. Wickner and A. Kornberg, JBC 249, 6244 (1974). 15. W. Wickner, R. Schekman, K. Geider, and A. Kornberg, PNAS 30, 1764 (1973). 16. C. McHenry and A. Kornberg, JBC 252, 6478 (1977).

44

CHARLES McHENRY AND ARTHUR KORNBERG

0-

5

SSB

ss

PRIMING

LOENZYME

DNA LIGASE

RF

FIG,1. Requirements for conversion of a single-stranded phage DNA (SS) to the duplex replicative form (RF). a result of the associations, and that the subunit comigrates with the other holoenzyme components on sizing columns and glycerol gradients. (iii) Demonstration that association of a proposed subunit to pol I11 makes the new complex more holoenzyme-like, Definitive evidence that a polypeptide is a subunit should include both the first criterion and genetic proof that the polypeptide is required for replication of the E. coli chromosome in vivo. Four polypeptides (a,/3 ,y , and 6) satisfy these rigorous criteria. Three additional ones ( E , 0, and T ) have been judged to be holoenzyme subunits by the other criteria. The composition of holoenzyme and its subunit subassemblies, based on present information, is as given in Table I. Resolution of holoenzyme by phosphocellulose chromatography generates pol III* and the /3 subunit (37,000 daltons) (16, 17). Alone, each is inactive, but together they reconstitute holoenzyme activity on G4, 4x174, and M13 templates. /3 appears to exist as a dimer upon dissociation from other holoenzyme components (17). Pol 111" can be further resolved to yield a complex of two subunits ( y . 6) by treatment with o-phenanthroline and mild heat (16). This chelator presumably removes an essential metal from pol 111. Although not yet found in pol 111, zinc is an essential component required for activity in a variety of DNA and RNA polymerases. Chelation by o-phenanthroline is likely responsible for rendering pol I11 sufficiently labile to permit its selective heat inactivation and dissociation from y . 6, thus permitting y . 6 to be purified as a distinct molecular species. The y . 6 complex has no known independent enzymatic activity, but together with pol I11 and /3 reconstitutes holoenzyme activity on phage templates (16). The y and 6 subunits have recently been resolved from each other. y was purified to 65% homogeneity from an overproducing strain that carried a plasmid with the dnaZ gene, the structural gene for y (see Section II1,C) (18). Free y was resolved from pol I11 by Blue Dextran Sepharose chromatography. The 17. K . Johanson and C. McHenry JBC 255, 10984 (1980). 18. U. Hiibscher and A. Kornberg, JBC, 255, 11698 (1980).

3. DNA POLYMERASE 111 HOLOENZYME

45

purified subunit exists as a dimer in its native state (18). 6 was purified to 50% homogeneity using, as an assay, its ability to reconstitute holoenzyme activity on a 4x174 template in the presence of pol 111, p, and y

(W.

Evidence that E and 8 subunits are holoenzyme subunits was presented in Section II,B. The polypeptide T (83,000 daltons), a seventh component of holoenzyme preparations, has also been implicated as a subunit (20). Purified as a complex of pol 111, it is termed pol 111’. Addition of T to pol I11 makes pol 111 heavier and more basic, as indicated by a higher affinity for phosphocellulose. When holoenzyme was immunoprecipitated with j3 antibody, T and a were coprecipitated in a 1 : 1 ratio (20). The T polypeptide appears to be a monomer when dissociated from holoenzyme and exhibits an ATPase activity that is dependent on single-stranded DNA (21). The functional properties of pol I11 are also altered by the addition of T. Pol 111‘, resembles holoenzyme, and differs from pol 111 by a slight capacity to use a randomly primed, long, single-stranded template in the presence of spermidine (20). Pol 111’ and holoenzyme are both stimulated by spermidine; by contrast, the low levels of pol I11 activity with these single-stranded templates are further decreased by spermidine. Pol 111’ is far less efficient than holoenzyme in this assay and cannot substitute for holoenzyme in the G4 system. Thus, the addition of T to pol I11 to form pol 111’ makes pol I11 more holoenzyme-like. The T polypeptide is not required for reconstitution of holoenzyme-like activity on single-stranded phage DNA templates (20). Perhaps some of the holoenzyme subunits, required for replication of the complex E. coli chromosome, are not needed for replication of the small phages. These simpler templates may not require all the components of the intact holoenzyme complex after it has been resolved into its constituent parts. For example, pol I11 alone is able to fill in gaps in nuclease-activated duplex DNA. Reports from the Hurwitz laboratory describe the isolation of three factors that, when added to pol 111, replicate single-stranded phage templates (22,231. Elongation Factors I and I11 anddnaZ protein probably correspond to p , 6 and y , respectively (Table I). 19. U. Hiibscher and A. Kornberg, PNAS 76, 6284 (1979). 20. C. McHenry, I n “Mechanistic Studies of DNA Replication and Genetic Recombination” (B.Alberts, ed.) ICN-UCLA Symp. Mol. Cellular Biol., Vol. 19. Academic Press, New York, 1980. 21. R. Meyer, J. Shlomai, and A. Komberg, personal communication. 22. J . Hurwitz and S. Wickner, PNAS 71, 6 (1974). 23. S. Wickner and J. Hurwitz, PNAS 73, 1053 (1976).

46

c.

CHARLESMcHENRYANDARTHURKORNBERG STRUCTURAL GENES FOR

HOLOENZYME SUBUNITS

The a, p , y , and 6 holoenzyme subunits have been identified as the products of replication (dna) genes (Table 11). Conditional lethal replication mutations in these genes have been correlated with defects in the corresponding subunit. Pol I11 and holoenzyme are known to contain the product of the dnaE (PolC) gene. Both pol I11 and holoenzyme are defective when assayed in extracts of temperature-sensitivednaE cells (7, 15). Some mutator mutations have also been mapped in the dnaE locus (24-26). Inasmuch as pol I11 contains three polypeptides (IZ),it was not known which subunit was the dnaE gene product. A 4500 base-pair segment of E. coli DNA that contains the dnaE gene has recently been cloned into the ret gene of pBR322. This plasmid complementsdnaE mutants and directs the synthesis of the Q polypeptide (27). Inserts of DNA into the Hind111 site on the cloned segment abolishes the ability of the plasmid to rescue dnaE mutants and blocks the expression of Q (27) in “maxicells” (28). Therefore, it has been concluded that dnaE is the structural gene for a . A specialized dnaE transducing phage has been constructed that can transduce several different dnaE mutants (29). Lysogens of this phage restore normal pol I11 activity to extracts. Yet, pol 111 is not amplified upon induction of these lysogens (29). Similarly, a colEldnaE hybrid plasmid from the Clarke and Carbon collection rescues dnaE mutants, but does not lead to an overproduction of pol I11 (30). These results may be due to an autoregulatory mechanism, with a! being rapidly proteolyzed if not bound to other holoenzyme components, or to the need for the E and 0 polypeptides for the expression of a activity. The structural gene for y is dnaZ. Early studies indicated that y . 6 contained a product of the dnnZ gene (16). Recent work in which y and 6 were purified independently has demonstrated that y specifically complements extracts from dnaZ cells (/8,3/). Similarly, 6 and p complements extracts from dnaX ( / 9 , 3 2 )and dnaN (33,341 cells, respectively, and are judged to 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34.

C. Sevastopoulos and D. Glaser, PNAS 74, 3947 (1977). E. B. Konrad, J . Bacferiol. 133, 1197 (1978). R. Hall and W. Brammer, Mol. Gen. Genet. 121, 271 (1973). C. McHenry and M. Welch, manuscript in preparation. A. Sancar, A. Hack, and W. Rupp, J . Bacferiol. 137, 692 (1979). H.Shizuya, D. Livingston, and C. C. Richardson, PNAS 71, 2614 (1974). S. Wickner, R. Wickner, and C. Raetz, BBRC 70, 389 (1976). C. Truitt and J. Walker, BBRC 61, 1036 (1974). J. Henson, H. Chu, C. Irwin, and J. Walker, Generics 92, 1041 (1979). P. Burgers, A. Kornberg, and Y. Sakakibara, PNAS, in press. Y. Sakakibara and T. Mizukami, Molec. Gen. Genet. 178, 541 (1980).

3. DNA POLYMERASE 111 HOLOENZYME

47

TABLE I1

CHARACTERISTICS OF DNA POLYMERASE 111 HOLOENZYME MUTANTS MutanP Characteristics

dnaE

dnaZ ~

Subunit Fast stop in DNA synthesis Growth of phages 4x174 h

T7 Role in repair

~

dnaX

dnaN

~~~~

a

Y

+

+

s

P

-

-

N.D. N.D. N.D. N.D.

-

+ -

+

+

+

N.D.

+ -

+ N.D.

N.D., not determined.

be the products of these genes. Thus, these four subunits, which were first identified by biochemical in v i m assays to be holoenzyme subunits, have now been shown by genetic criteria to be required for chromosomal replication.

D. MECHANISTIC STUDIES The specific roles of all the holoenzyme subunits in the replication process are not yet clear. However, preliminary work suggests a possible cycling of certain of the subunits during the replication process (35). Holoenzyme, in the presence of ATP, can form an isolatable initiation complex upon incubation with a primed single-stranded DNA template. Upon formation of this initiation complex, elongation proceeds in the presence of an antibody directed against copol III* (p) preparations. Furthermore, it has been suggested that when isolated separately, elongation Factor I11 (8) and dnaZ protein (y) can transfer elongation Factor I Cp) to the DNA template in an ATP- or dATP-requiring process (36), but the requirement for Factor I @) in further elongation remains undetermined. In later experiments, purified antibody prepared against homogeneous j3 has been shown to inhibit only reactions for which holoenzyme is required (17). Furthermore, as in the early experiments with copol III", the p antibody blocks the formation of an isolatable initiation complex, but not the subsequent elongation (17). Thus, j3 antibody can be used as a tool to 35. W. Wickner and A. Kornberg, PNAS 70, 3679 (1973). 36. S. Wickner, PNAS 73, 3511 (1976).

CHARLES McHENRYANDARTHUR KORNBERG

48

TABLE 111 PROCESSIVITY OF POL Processivity and effects Processivity, residues Spermidine effect On incorporation On processivity SSB" effect On incorporation On processivity

111 AND HOLOENZYME

Pol 111

Holoenzyme

10 to 15

>5000

Inhibition Two- to fourfold decrease

Stimulation Twofold increase

Strong inhibition N.D.b

Stimulation Fivefold increase

Single-strand binding protein N.D.. not determined

block reinitiation so that the processivity of holoenzyme can be determined. Holoenzyme can replicate most of the G4 genome (>SO00 nucleotides; without dissociating from it (37). Other experiments have indicated the processivity of pol I11 to be only 10-15 nucleotides (Table 111). Spermidine and single-strand binding protein, both of which stimulate holoenzyme and inhibit pol 111, increase the processivity of holoenzyme on single-stranded DNA by three- to fivefold (Table 111) (37).

E. PHYSIOLOGICAL ROLE The fact that four subunits of the holoenzyme are encoded by genes in which conditional lethal replication mutations have been found testifies to their participation in DNA replication. All of these mutants promptly stop DNA synthesis at the nonpermissive temperature (Table 11). Where it has been checked, cells with these mutations do not support the growth of phages that require E. coli polymerase functions (e.g., 4x174 and A), but do support the growth of phage T7, which codes for its own polymerase (Table 11). Combining genetic and biochemical evidence, it appears likely that the holoenzyme is responsible for most of the replicative DNA synthesis in E. coli. It is considered likely that the 10 to 20 molecules of the holoenzyme in an E. coli cell are sufficient to sustain the in vivo chain growth of about 1,000 nucleotides per second (I). However, on phage templates in vitro, holoenzyme can neither remove primers nor fill in the 37. P. Fay, K. Johanson, C. McHenry, and R. Bambara, JBC 256, 976 (1981).

3. DNA POLYMERASE I11 HOLOENZYME

49

short gap preceding the 5' terminus of these primers (see Fig. 1) (38).This is analogous to the synthesis of the lagging strand of the E. coli chromosome in which the nascent Okazaki fragments must be connected. Pol I performs this function in vitro and probably does so in vivo as well ( I ) . The dnaE gene product has also been judged essential for one pathway of excision-repair of UV irradiation damage of DNA (39), but the role of other holoenzyme mutations in this pathway has not yet been examined. IV.

Summary

DNA polymerase I11 holoenzyme is a complex, multisubunit enzyme responsible for most of the replicative synthesis in E. coli. It contains a core (pol 111) that can repair short gaps created by nuclease in duplex DNA. Pol I11 contains three subunits: a (dnaE protein), E and 8 . For efficient replication of the long, single-stranded templates that resemble structures encountered at natural replication forks, pol I11 requires the auxiliary subunits p , y , and 6 , encoded by dnaN, dnaZ, and dnaX genes, respectively. Processivity in elongation by the holoenzyme exceeds 5000 residues, compared to a value of only 10 for pol 111. The /3 subunit is required for initiation of replication but antibody directed against it does not block elongation. The intact holoenzyme complex contains at least one other component, a DNA-dependent ATPase, named T . The T polypeptide is not required for reconstitution of holoenzyme activity on singlestranded templates, but alters some properties of pol I11 to make it resemble the holoenzyme. Major uncertainties about the organization and functions of the subunits of the holoenzyme account for our limited understanding of the mechanism of its action. Note Added in Roof

New information is available about the following subunits: /3, 7,and 6 . The /3 subunit is part of the holoenzyme elongation complex in which all /3 antigenic determinants are buried after the initiation complex starts to function (40). The 7 subunit when added to pol I11 (presumably forming pol 111') increases its processivity from about 10 to 50 ( 4 / ) . Spermidine (4 mM 38. 39. 40. 41.

H. Tabak, J . Griffith, K. Geider, H. Schaller, and A. Kornberg,JEC 249,3049 (1974). D. Youngs and K . Smith, Nature New Biol. 244, 240 (1973). K. Johnson and C. McHenry, manuscript in preparation. P. Fay, K. Johanson, C. McHenry, and R . Bambara, manuscript in preparation.

50

CHARLES McHENRY AND ARTHUR KORNBERG

stimulates pol 111’ and increases its processivity to 100, whereas pol I11 is inhibited under these conditions. Pol IIIwhas a processivity of 50 which is increased to 150 by SSB, a protein which inhibits pol I11 and pol 111’ (41). Thus, there is a gradient of processivity among the forms of pol I11 reflecting their structural complexity. The T subunit, a DNA-dependent NTPase, hydrolyzes both AT and dATP to the corresponding diphosphates ; GTP and dGTP are hydrolyzed at 10-20% the rate of ATP. Pyrimidine deoxynucleoside triphosphates are hydrolyzed very slowly (42). Oligo(dA),, is 80% effective as G4 DNA in stimulating this reaction (42). The 6 subunits is a provisional designation for an activity required to reconstitute holoenzyme activity in the replication of single-stranded G4 DNA (43). This activity was previously supplied as an impurity in dnuG primase preparations. Thus, 4 may be the eighth holoenzyme subunit.

42. R. Meyer, D. Rein, and C. McHenry, manuscript in preparation 43. P. Burgers and A. Kornberg, unpublished results.

T-Phage DNA Polym erases I. ROBERT LEHMAN

I. Introduction . . . . . . . . . 11. T4 DNA Polymerase . . . . . . A. Purification and Properties . . B. Reactions Catalyzed . . . . . C. Role of T4 DNA Polymerase in 111. TS DNA Polymerase . . . . . . A. Purification and Properties . . B . Reactions Catalyzed . . . . . C. Role of T5 DNA Polymerase in IV. T7 DNA Polymerase . . . . . . A. Purification and Properties . . B. Reactions Catalyzed . . . . . C. Role of T7 DNA Polymerase in

I.

. . . . . . . . . . . . . . . . . . Vivo . . . . . . . . . . . . . . . . . Vivo . . . . . . . . . . . . . . . . . Vivo . . . .

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

. . . . . .

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

.

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

51 52 52 54 57 60 60 60 61 62 62 63 64

Introduction

Infection of Esclzerichia coli with the T series of bacteriophages (T1 through T7) results in the prompt termination of bacterial DNA replication and, after a brief interval, the onset of bacteriophage DNA replication. Each of these phages with the exception of T1 and T3 is known to induce a new DNA polymerase that is required for synthesis of the bacteriophagespecific DNA. This chapter is concerned with the T4-, T5-, and T7induced DNA polymerases, the enzymes that have been examined in 51 THE ENZYMES. Vol. XIV Copyright 0 19RI by Academic Press, Inc. All rights of reproduction in any form reserved

ISBN 0-12-122114-6

52

I. ROBERT LEHMAN

greatest detail and about which most is known. T4- and T7-induced polymerases very likely exist as part of multienzyme DNA replication complexes in vivo, and indeed act in association with accessory proteins (single-stranded DNA binding protein, primase, etc.) in duplex DNA replication in vitro. A similar situation may well hold for T5-induced DNA polymerase. Because of their close association with other proteins, the properties of the T-phage polymerases within such replication complexes may differ significantly from those of the isolated enzymes.

II. 14 DNA Polymerase

T4 DNA polymerase was the first of the phage-induced polymerases to be isolated in homogeneous form (I 1. It was also the first DNA polymerase to be identified with a specific genetic locus (T4 gene 43) (2, 3). A. PURIFICATION AND PROPERTIES T4 DNA polymerase was first purified from extracts of E. coli B infected with T4 phage bearing an umber (nonsense) mutation in gene 44 (I, 4). In such extracts the specific activity is five- to tenfold higher than in extracts from wild-type T4-infected cells. Nearly homogeneous polymerase is obtained after chromatography on phosphocellulose, DEAE-cellulose, and hydroxyalpatite ; 650 g of infected cells yield approximately 1 mg of pure enzyme. Alberts and colleagues (5) have developed an alternative purification procedure in which extracts enriched in polymerase as well as other replication proteins (the products of genes 44, 45, and 62) are prepared from cells infected with T4 bearing the regA mutation, along with mutations in genes 42 and 30. Using DNA-cellulose, hydroxylapatite, norleucineSepharose, and DEAE-cellulose chromatography, approximately 9 mg of homogeneous polymerase are obtained from 300 g of infected cells. 1. M. Goulian, 2. J. Lucas, and A. Kornberg, JBC 243, 627 (1968). 2. A. de Waard, A. V. Paul, and I. R. Lehman, P N A S 54, 1241 (1%5). 3. H. R. Warner and J. E. Barnes Virology 28, 100 (1966). 4. R. H. Epstein, A. Bolle, C. M. Steinberg, E. Kellenberger, E. B. De La Tour, R. Chevalley, R. S. Edgar, M. Susman, D. Denhardt, and A. Lielausis, C S H S Q B 27, 375 ( 1963).

5. C. F. Morris, H. Hama-Inaba, D. Mace, N. K. Senha, and B. Alberts, JBC 254, 6787 (1979).

53

4. T-PHAGE DNA POLYMERASES TABLE I

COMPARISON OF AMINOACIDCOMPOSITION OF T4 DNA POLYMERASE AND E . coli DNA POLYMERASE I" Amino acid Lysine Histidine Arginine Half-cystine Aspartic acid, asparagine Threonine Serine Glutamic acid, glutamine Proline Glycine Alanine Valine Methionine Isoleucine Leucine Tyrosine Phenylalanine Tryptophan

T4 DNA Polymerase Moles/ 1 14,000 g 85 16 46 15 113 29 65 113 40 61 55 55 35 86 59 41 12 12

E . coli DNA Polymerase I Moles/109,000 g 61 19 48 3 88 53

40 126 53 63 102 61 24 55 112 33 25 9

From Goulian er a / . ( I ) and Jovin et a / . (6).

Like DNA polymerase I (pol I), TCinduced DNA polymerase consists of a single polypeptide. However, its molecular weight, 114,000, is somewhat higher than that of the E. coli enzyme (MW 109,000).Its amino acid composition (Table I) also shows several significant differences from pol I. In particular, the T4 enzyme has 15 half-cystine residues compared to the E coli enzyme, which has only 3. This point is of interest in view of the great sensitivity of the phage-induced enzyme, and the relative insensitivity of pol I to sulfhydryl inactivation by, for example, N-ethylmaleimide (6). T4 polymerase displays a relatively broad pH optimum ranging from pH 8 to 9; at pHs 7.5 and 9.7, approximately 50% of optimal activity is observed. Maximal polymerase activity requires 6 mM M$+; Mn2+at an optimal concentration of 0.1 mM gives a rate approximately one-fourth of that with 6 mM M$+. There is no detectable activity in the absence of a divalent cation (Table 11). 6. T.

M. Jovin, P. T. Englund, and L. Bertsch, JBC 244, 2996 (1969).

54

I. ROBERT LEHMAN TABLE I1

COMPARISON OF PHAGE T4,T5 AND T7-INDUCED

Functions Polymerization: 5‘ + 3’ Exonuclease 3‘ 5’ single strands Exonuclease 3’ .+ 5‘ duplex DNA Template-primer Primed single strands Nicked duplex Activity pH optimum Inhibition by NEM Effect of 200 mM NaCl compared to 50 mM set at 100 Structure Homogeneity Molecular weight Number of subunits Genetic locus

DNA

POLYMERASES“

T4

T5

T7

+

+ + +

+ +

+

+

+

-

+

4

-

+

8-9

8-9

7.6-7.8

5

400

< 10

+

Yes 114,OOO 1 gene 43

+

-

+

Yes

Yes

%,OW

%,ooo

1

2 gene 5 , trxA

gene rs53

Adapted from A. Kornberg, “DNA Replication” Freeman, San Fracisco, 1980.

B. REACTIONS CATALYZED 1. Polymerization

Like all other DNA polymerases, T4 polymerase catalyzes the polymerization of mononucleotide units from deoxynucleoside 5‘triphosphates onto the 3’-hydroxyl terminus of a primer at the direction of a single-stranded DNA template. Thus, fully duplex DNA cannot serve as a template-primer. However, DNA made partially single-stranded by digestion from the 3’ termini with E. coli exonuclease I11 (7) becomes an effective template-primer for the T4 polymerase, and one which under optimal conditions for the enzyme permits a rate of polymerization that approaches the in vivo value (Fig. 1). Single-stranded DNA can also serve as a template-primer for T4 DNA polymerase. The product is a duplex structure in which the newly synthesized strand is covalently linked from its 5’ terminus to the 3‘ end of the template ( I ) . Formation of such a structure by T4 polymerase is plausibly explained by a model in which the single-stranded DNA loops back upon 7. C. C. Richardson and A. Kornberg, JBC 244, 29% (1%4).

55

4. T-PHAGE DNA POLYMERASES SINGLE STRANDED TEMPLATE-PRIMER

EXONUCLEASE 111 TREATED DUPLEX D N A

’1

3 - 5‘ EXONUCLEASE

5u1uwLLIIlLLUvL

5 -

3’

3 T4 POLYMERASE

FIG. 1. Action of single-stranded DNA and exonuclease 111-treated duplex DNA as template-primers for T4 DNA polymerase [adapted from Ref ( I ) . See text for details]. itself, the loop being stabilized by the annealing between regions along the same strand. Any unpaired nucleotides at the 3’ terminus are hydrolyzed by the 3‘ + 5’ exonuclease of the polymerase (see Section II,B,2) until a base-paired terminus is reached (Fig. 1). The template-primer thus formed is analogous to the exonuclease 111-digested duplex and can be replicated up to the 5‘ terminus of the template. Unlike pol I, the T4 polymerase is unable to use a DNA duplex that contains a phosphodiester break (a nick) as template-primer, presumably because of its inability to displace the 5‘-terminated strand at the nick. However, addition of the T4 gene 32 protein (the phage-induced singlestranded DNA binding protein) (4, 8) facilitates strand displacement and allows the T4 polymerase to replicate the nicked duplex. This effect is, however, observed only at low ionic strengths (9). When supplemented with accessory proteins, which include the products of genes 44, 62, and 45, in addition to gene 32, T4 DNA polymerase can initiate replication at a nick at physiological ionic strengths. DNA synthesis begins by covalent addition onto the 3‘-hydroxyl end and continues with strand displacement. 8. B. Alberts and L. Frey, Nature (London) 227, 1313 (1970). 9. N. G. Nossal, JBC 249, 5668 (1974).

56

I . ROBERT LEHMAN

Replication under these conditions is highly processive and proceeds at or near the in vivo rate of 800 nucleotides polymerized per second (10). When supplemented with the products of genes 41 and 61, these proteins can catalyze the initiation of short RNA primers on single-stranded circular DNA templates. The primers can be extended to form long DNA duplexes (10).

2. 3' + 5' Exonuclease T4 DNA polymerase lacks 5' 3' exonuclease activity but contains an extremely active 3' + 5' exonuclease that shows a strong specificity for single-stranded DNA; its turnover number is some 250-fold greater than 5' exonuclease associated with pol I and some 3-fold greater than the 3' the turnover number for polymerase action ( I ) . The products of hydrolysis are deoxynucleoside 5'-monophosphate and a single dinucleotide, derived from the 5' terminus of the polynucleotide. Thus, like E. coli exonuclease I, the 3' + 5' exonuclease of T4 polymerase is incapable of cleaving a dinucleotide (I/). The polymerase and 3' + 5' exonuclease activities associated with the 114,000 dalton DNA polymerase protein cannot be separated by, for example, proteolytic cleavage. However, nonSense mutants in gene 43 have been isolated that cause premature polypeptide chain termination, yielding a protein with a molecular weight approximately 80% that of the native protein (12). The mutant protein retains its exonuclease but not its polymerase activity. 5' exonuclease is The rate of hydrolysis of polynucleotides by the 3' dependent on chain length even at saturating levels of polynucleotide termini. This results from the ability of T4 polymerase to bind to internal nucleotides as well as to the 3'-hydroxyl terminus of a polynucleotide. Binding of the enzyme to internal sites does not result in hydrolysis and as a consequence the internal nucleotides act as inhibitors. Indeed, an increase in chain length of tenfold (from 300 to 3000 residues) results in a decrease in the rate of hydrolysis of approximately 100-fold ( / I ) .

-

-

3. Interaction of Polymerase and 3' -+ 5' Exonuclease Activities In the presence of the complementary deoxynucleoside triphosphates, a template, and a primer, T4 DNA polymerase may act either as a polymerase or exonuclease, depending upon the mode of association of the template and primer. Polymerization occurs when all three of the 10. C. C. Liu, R. L. Burk, U. Hibner, .I. Barry, and B. Alberts, CSHSQE 43,469 (1978). 11. W. M. Huang and I. R. Lehman, JBC 247, 3139 (1972). 12. N. G. Nossal and M. S. Hershfield, JBC 246, 5414 (1971).

4. T-PHAGE DNA POLY MERASES

57

following are available to the enzyme: (i) a polynucleotide template, (ii) a 3'-hydroxyLterminated primer at least one residue shorter than the template, and (iii) the appropriate deoxynucleoside triphosphate (or triphosphates) complementary to the template. In the presence of all of these components the enzyme acts as a polymerase; in the absence of any one, it functions as an exonuclease. Consider a structure in which the 3'-hydroxyl terminus of one of the two strands of a DNA duplex protrudes beyond the 5' terminus of the opposing strand (see Fig. 1). Hydrolysis of the protruding nucleotides by 3' + 5' exonuclease action will proceed until the 3' end occupies a site at least one residue interior to the 5' terminus. At this point all three of the components required for polymerase action are present and the missing nucleotide can be restored. Because fraying at the ends of the duplex creates a transient single strand, the enzyme can once again act as an exonuclease removing a residue from the 3' terminus. The cycle can be repeated again with the enzyme alternating as polymerase or exonuclease in a process that consumes deoxynucleoside triphosphates without net synthesis of DNA.

C. ROLEOF T4 DNA POLYMERASE in Vivo A functional T4 DNA polymerase is essential for the initiation and maintenance of viral DNA replication. Indeed, T4-induced polymerase was the first such enzyme to be directly linked with chromosomal replication ( 2 , 3 ) .It is now clear that T4 polymerase functions in T4 DNA replication as part of a replication complex that, in addition to the polymerase (the gene 43 product), contains six accessory proteins that permit the initiation of new DNA strands by RNA primers, and the replication of duplex DNA at rates near those observed in vivo (10). As previously noted, a chain-terminating (umber) mutant in T4 gene 43 induces the synthesis of a shortened polypeptide that lacks polymerase but retains exonuclease activity. By examining the immunological cross reactivity and molecular weights of the peptides generated following infection of E. coli with a series of amber mutants, a map of the T4 polymerase gene has been constructed (13). The physical map (Fig. 2) is in good agreement with the genetic map determined on the basis of recombination frequencies. From an analysis of the size and cross reactivity of the amber peptides it is clear that the T4 polymerase gene is translated in a counterclockwise direction along the circular genetic map of phage T4 (4). The results of temperature shift experiments with temperature-sensitive mutants in gene 43 have led to the suggestion that the production of a 13. W. M. Huang and I. R. Lehman, JBC 247, 7663 (1972).

58

I. ROBERT LEHMAN GENE 43 GENETIC El92 PHYSICAL MAP

I. I

1

:.'.I

I 100

I

.:I 80

' .

.

')'It

I

I

I

I

40

60

MW

.

.I

'

1

J

-

I 20

I

1

0

10-3

FIG. 2. Comparison between genetic and physical maps of T4 gene 43. The physical map was constructed on the basis of the gene 43 amber peptides as determined by their migration in poiyacrylamide gels in the presence of sodium dodecyl sulfate; the genetic map is that of E. F. Allen, I. Albrecht, and J . W. Drake, Genetics 65, 187 (1970). [From Huang and Lehman ( I - ? ) ] .

functional T4 polymerase, including its associated 3' + 5' exonuclease, involves the energy-dependent conversion of an inactive precursor to active enzymes (14). Thus, when E. coli is infected at 43" with the temperature-sensitive rsL.53 allele of the T4 DNA polymerase gene, T4 DNA polymerase and 3' + 5' exonuclease activities are undetectable in extracts assayed at 30". However, on shifting the culture from 43" to 30" both polymerase and exonuclease activity appear immediately and increase rapidly, even in the presence of sufficiently high concentrations of chloramphenicol, to completely block protein synthesis. Thus, during infection at 43" an inactive polymerase is synthesized that is converted to active enzyme on shifting to 30", a process that does not require protein synthesis. The appearance of active enzyme upon shifting to 30" does, however, require oxidative metabolism since it does not occur in the presence of 2,4-dinitrophenol, an inhibitor of oxidative phosphorylation (15) (Fig. 3). The nature of the energy-dependent conversion of inactive to active T4 tsL.56 DNA polymerase remains an intriguing but unsolved problem. Several, but not all, temperature-sensitive gene 43 mutants cause a large (in one instance up to 2000-fold) increase in reversion frequency of T4 rZZ mutants involving both transitions and transversions; that is, gene 43 acts as a mutator gene (16). Other temperature-sensitive gene 43 mutants produce a decrease in reversion frequency of rZl mutations, especially AT + GC transitions. These mutants, therefore, have the pro14. M. N. Swartz, H. Nakamura, and I. R. Lehman, Virology 47, 338 (1972). 15. J. Thorner, W. M. Huang, and I. R. Lehman, Virology 68, 338 (1975). 16. J. F. Speyer, J. D. Karam, and A. B. Leony, CSHSQB 31, 693 (1966).

4. T-PHAGE DNA POLYMERASES

59

I--

-430-

DNP

A-A-

01 0

' '

10

20

30

40

I

50

I

1

60

70

MINUTES AFTER INFECTION

FIG.3. Time course of appearance of T4tsLS6 DNA polymerase after infection at 43", followed by a shift to 30" in the presence of chloramphenicol (CAM) and in the presence and absence of 2,4 dinitrophenol (DNP). [From Thorner et a / . (IS)].

perties of antimutators. In vitro measurement of exonuclease-polymerase ratios of the purified mutant enzymes have shown the polymerase isolated from cells infected with a mutator mutant (rsL.76) to have a low exonuclease-polymerase ratio relative to the wild-type enzyme, and the polymerase from the antimutator (rsL144) mutant to have a correspondingly high exonuclease-polymerase ratio (17, 18). Inasmuch as the 3' + 5' exocuclease activity of pol I serves a proofreading function by excising mismatched nucleotides during DNA replication in vitro (19), mutation rates seen with these phages reflect, at least in part, the relative polymerase and 3' + 5' exonuclease activities of T4 DNA polymerase during DNA replication in vivo. On the other hand, the DNA polymerase induced by another mutator mutant, T4 rsLB8, shows no defect in the proofreading activity of its associated 3' + 5' exonuclease (20). In this instance, then, the mutator phenotype appears to result from a decrease in the accuracy of nucleotide selection during polymerization. As previously noted, T4 DNA polymerase functions in DNA replication in vivo as part of 17. N. Muzyczka, R. L. Poland, and M. J. Bessman, JBC 247, 7116 (1972). 18. M. J. Bessman, N. Muzyczka, M. F. Goodman, and R. L. Schnaar, J M B 88, 409 (1974). 19. D. Brutlag and A. Kornberg, JBC 247, 241 (1972). 20. F. D. Gillin and N . G. Nossal, JBC 251, 5219, 5225, (1976).

60

1. ROBERT LEHMAN

a multienzyme complex. Its association with accessory proteins within such a complex has in fact been shown to influence the fidelity with which it catalyzes nucleotide polymerization (10).

111.

T5 DNA Polymerase

A. PURIFICATION AND PROPERTIES T5-induced DNA polymerase has been purified to apparent homogeneity by DEAE-cellulose and phosphocellulose chromatography followed by either filtration through Sephadex G-100 or hydroxylapatite chromatography (21, 22). As judged by equilibrium sedimentation and polyacrylamide gel electrophoresis, the enzyme consists of a single polypeptide of molecular weight 96,000 (Z/, 22). Like T4-induced polymerase, the T5 enzyme requires a thiol (2-mercaptoethanol or dithiothreitol) for maximal activity, and is completely inhibited by N-ethylmaleimide. It is however strongly stimulated by 0.2 M monovalent cation, in particular, NI&+ and Na+. At this ionic strength T4 DNA polymerase retains only,about 5% of its activity. T5 polymerase has a pH optimum of 8.5; it is completely dependent upon added Mg2+(Table 11).

B. REACTIONSCATALYZED 1. Polymerization

T5 DNA polymerase can effectively use single-stranded DNA as a template, with either the 3’-hydroxyl end looped back or an oligonucleotide annealed to the chain to serve as a primer in a manner analogous to T4 polymerase (21, 22). Unlike the T4 enzyme, T5 DNA polymerase is able to initiate replication at a nick in duplex DNA even in the absence of accessory proteins (23). Replication at a nick is accompanied by displacement of the 5’-terminated strand, and is strongly influenced by temperature. At 25”, replication of denatured T7 DNA is two- to threefold greater than nicked duplex T7 DNA; at 37” it is nearly tenfold greater. The basis for the difference in behavior of the two types of template-primer is not known. In studies using the homopolymer pair poIy(dA)300.d(T),o as a 21. C. W. M. Orr, S. T. Herriott, and M. J. Bessman, JBC 240, 4652 (1965). 22. R. K . Fujimura and B. C. Roop, JBC 251, 2168 (1976). 23. S . K . Das and R. K. Fukimura, JBC 254, 1227 (1979).

4. T-PHAGE DNA POLY MERASES

61

template-primer, the T5 DNA polymerase is highly processive, with 150160 nucleotides added with each association and dissociation of the enzyme from the template-primer (24). A similar analysis of T4 DNA polymerase gave a processivity value of 12. Thus, replication by T5 DNA polymerase is intrinsically more processive than T4 polymerase. However, as noted previously, association of T4 polymerase with accessory replication proteins can substantially increase its processivity.

2. 3' -+ 5 ' Exonuclease T5 DNA polymerse has a 3' + 5' but not 5' + 3' exonuclease activity. However, it is noteworthy that bacteriophage T5 induces a 5' + 3' exonuclease separate from the polymerase that is essential for T5 DNA replication (25). The 3' + 5' exonuclease associated with T5 polymerase attacks both native and denatured DNA; however, the rate of degradation of native DNA is only one-fifth that of denatured DNA (26). The products of hydrolysis are deoxynucleoside 5'-monophosphates and a small proportion of dinucleotides. Possibly, the latter derive from the 5' termini of the polynucleotide. The optimal conditions for 3' + 5' exonuclease activity (e.g., pH, Mg 2+, thiol reagent, and ionic strength) are the same as those for polymerase action. Like T4 DNApolymerase,TS polymerase can promote the alternate insertion and hydrolysis of nucleotides at the 3' termini of a template-primer, leading to the conversion of deoxynucleoside triphosphates to their corresponding monophosphates without net synthesis of DNA. Although no accessory proteins have been identified, it is almost certain that like T4, replication of T5 DNA will require a multienzyme complex. Indeed, amber mutants of T5 have been isolated that represent at least six different genes that are essential for DNA replication. Several other mutations are known that affect the rate and extent of T5 DNA replication (27). Within such a complex, important features of the polymerase and exonuclease (template preference, processivity, and the balance between nuclease and polymerase) may be significantly altered. C. ROLEO F T5 DNA P O L Y M E R A S Vivo E~ Temperature-sensitive mutants in phage T5 that map in a specific genetic locus (fs.73)fail to synthesize T5 DNA at restrictive temperatures in 24. 25. 26. 27.

S. K. Das and R. K. Fujimura, JBC 252, 8700, 8708 (1977). G. D. Frenkel and C. C. Richardson, JBC 246, 4839, 4848 (1971). S. K. Das and R. K. Fujimura, J. Virol. 20, 70 (1976). H. E . Hendrickson and D. J. McCorquodale, J . Virol. 9, 981 (1972).

62

I. ROBERT LEHMAN

as assayed in vitro. Thus, T5 polymerase is essential for replication of the T5 chromosome (2, 27). The mutant polymerase has been purified to near homogeneity. Examination of its polymerase activity verified the thermolability first observed in partially purified preparation of the enzyme (28). The mutant enzyme is one-fifth as active at 43" as it is at 30", compared with the wild-type polymerase, which is fivefold more active at the higher temperature. In contrast, the 3' + 5' exonuclease activities of the mutant and wild-type enzymes are approximately equivalent at 43". The mutational defect therefore appears to be specifically in the polymerase activity of the T5induced enzyme.

vivo, and induce an abnormally thermolabile T5 DNA polymerase

IV. 17 DNA Polymerase

A. PURIFICATION AND PROPERTIES The DNA polymerase induced upon infection of E. coli with phage T7 has been isolated in nearly homogeneous form by chromatographic procedures similar to those employed for the isolation of the T4- and T5-induced enzymes (e.g., phosphocellulose, DEAE-Sephadex, and hydroxylapatite) (29, 30). The enzyme consists of two subunits: One, encoded by T7 gene 5 , has a molecular weight of 84,000; the other, which has a molecular weight of 12,000, is the bacterial protein, thioredoxin ( 3 / ) . Both subunits are essential for polymerase activity (32,33).Like the other T-phage DNA polymerases, the T7 enzyme has an absolute requirement for Mg 2f and is stimulated by 2-mercaptoethanol. Optimal activity is observed at pH 7.6-7.8 in phosphate buffer (Table 11). The two subunits of T7 polymerase can be separated by dialysis of the enzyme against 6 M guanidine-HC1followed by gel filtration (30).Alternatively, the gene 5 protein, free of the thioredoxin subunit, can be isolated from extracts of an E. coli thioredoxin (trxAf mutant infected with wildtype T7 (34). The larger, phage-specified, subunit retains the singlestranded-DNA-specific 3' + 5' exonuclease activity of the native en28. 29. 30. 31. 32. 33. 34.

R. K. Fujirnura and B. C. Roop, Biochemistry 15, 4403 (1976). P. Grippo and C. C. Richardson, JBC 246, 6867 (1971). S . Adler and P. Modrich, JBC 254, 11605 (1979). T. C. Laurent, E. C. Moore, and P. Reichard, JBC 239, 3436 (1964). P. Modrich and C. C. Richardson, JBC 250, 5515 (1975). D. F. Mark and C. C. Richardson, PNAS 73, 780 (1976). K. Hori, D. F. Mark, and C. C. Richardson, JBC 354, 11591 (1979).

4. T-PHAGE DNA POLYMERASES

63

zyme, but lacks polymerase and double-stranded DNA-specific 3' -+ 5' exonuclease (see Section IV, B,2). No polymerase-associated activities are detectable in the small, thioredoxin, subunit. Reconstitution of the native enzyme can be achieved by incubating the two subunits at 0" for an appropriate period, at molar ratios of thoredoxidgene 5 protein ranging from 12 to 150 depending upon the conditions and the concentration of subunits. The reconstituted enzyme is essentially indistinguishable from the native T7 polymerase in its polymerase and exonuclease activities (30, 34).

B. REACTIONSCATALYZED 1. Polymerization

Either single-stranded DNA or duplex DNA that has been made partially single-stranded by exonuclease action can serve as an effective template-primer for the T7 DNA polymerase; fully duplex DNA is inert. However, when complemented with the T7 gene 4 protein, T7 polymerase can promote the replication of duplex DNA at a nick. Under these conditions, T7 DNA polymerase catalyzes the polymerization of deoxynucleotides, while the gene 4 protein facilitates the unwinding of the duplex coupled to the hydrolysis of ATP or other nucleoside triphosphate (35,36). Moreover, in the presence of ribonucleoside triphosphates and a single-stranded DNA binding protein, the gene 4 protein can catalyze the synthesis of tetranucleotides on the displaced single strand resulting from polymerase action. The extension of the tetranucleotide primers by T7 polymerase then permits synthesis along the displaced strand. Thus, T7 polymerase acting together with the gene 4 protein and a single-stranded DNA binding protein is capable of generating a replication fork in which leading strand synthesis gives rise to a displaced single strand, which then permits lagging-strand synthesis by the generation of oligoribonucleotide primers, and their subsequent extension by deoxynucleotide polymerization (3, 36). 2. 3'

+

5 ' Exonuckases

T7 DNA polymerase has two separate 3' 4 5' exonuclease activities: One is active on single-stranded DNA and the other is active on duplex DNA (29, 30, 34). As previously noted, the active site for the single35. C. C. Richardson. L.. J. Romano, R. Kolodner, J . E. LeClerq, F. Tamanoi, M. J . Engles, F. B. Dean, and D. S. Richardson, CSHSQB 43, 427 (1978). 36. G . Hillenbrand, G . Morelli, E . Lanka, and E. Scherzinger, CSHSQB 43, 449 (1978).

64

I . ROBERT LEHMAN

stranded DNA-specific exonuclease resides entirely within the gene-5coded subunit (30,34); 3’ --* 5’ hydrolysis of duplex DNA like nucleotide polymerization requires the interaction of the gene 5 protein with thioredoxin. The specific activity of the double-stranded DNA-specific exonuclease is approximately twice that of the single-stranded DNAspecific exonuclease. The products of hydrolysis of both single-stranded and duplex T7 DNA are >98% deoxynucleoside 5’-monophosphates (SO). As judged by chromatographic analysis and sensitivity to 5’-nucleotidase, 5’-terminal dinucleotides are not formed. In contrast to the polymerase activity that is unable to initiate polymerization at a nick in duplex DNA in the absence of gene 4 protein, the double-stranded DNA-specific exonuclease can initiate hydrolysis at such nicks, as well as at the 3’-hydroxyl termini at the ends of duplex DNA molecules (30, 34). Under conditions of DNA synthesis, i.e., in the presence of the four deoxynucleoside triphosphates, the double-stranded DNA-specific 3‘ + 5’ exonuclease is inhibited, presumably reflecting the inaccessibility of the 3’ terminal nucleotides of the primer to exonuclease action as a consequence of nucleotide polymerization at the primer terminus (30,34). Similar effects have been noted with the T4 and T5 DNA polymerases. Surprisingly, the addition of one or more deoxynucleoside triphosphates to the gene 5 protein results in inhibition of the single-stranded DNAspecific exonuclease, despite the virtual absence of polymerase activity (30, 34). Similar inhibition has been noted upon addition of ribonucleoside triphosphates. Thus, the inhibition of exonuclease action may not simply reflect competition between synthesis and hydrolysis at the primer terminus. This finding further suggests that the phage-encoded gene 5 protein not only has a site for the 3’-hydroxyl terminus of a single-stranded polynucleotide, but a site for nucleoside triphosphates as well. Although there is no 5‘ 4 3’ exonuclease activity associated with T7 polymerase, gene 6, which is directly adjacent to gene 5 on the T7 chromosome, induces such an exonucleolytic activity (37-39). Like the analogous exonuclease induced by T5 infection, the 5‘ --* 3’ exonuclease specified by the gene 6 protein is essential for T7 DNA replication (39).

C.

ROLE OF

T7 DNA POLYMERASE in Vivo

T7 mutants defective in gene 5 synthesize an altered DNA polymerase and are unable to replicate T7 DNA in vivo (29, 3 2 , 3 9 ) .Similarly, whenE. 37. C. Ken and P. D. Sadowski, JBC 247, 305 (1972). 38. K. Shinozaki and T. Okazaki, Nucleic Acids Res. 5, 4245 (1978). 39. F. W. Studier, Science 176, 367 (1972).

4.

T-PHAGE DNA POLYMERASES

65

coli cells that carry a mutation in the structural gene for thioredoxin are infected with phage T7, neither T7 DNA replication nor active T7 DNA polymerase can be detected (32, 40, 41). Clearly, then, T7 DNA polymerase, a dimer composed of the phage-coded gene 5 protein and the host-specified thioredoxin, plays an essential role in T7 DNA replication. As noted previously, the homogeneous T7 DNA polymerase, while unable to polymerize nucleotides on a duplex DNA template, is able to do so when supplemented with a single-stranded DNA binding protein and the T7 gene 4 protein, an enzyme endowed with helix unwinding and ribooligonucleotide synthetic capabilities.

40. M. J. Chamberlin, J . Virol. 14, 509 (1974). W.Chase, and C. C. Richardson, Mol. G e n . Gene?. 155, 145 (1977).

41. D. F. Mark, I.

This Page Intentionally Left Blank

Cellular and Viral-Induced Eukaryotic Polymeruses A. WEISSBACH

I. Introduction and Perspective . . . . . . . . . 11. DNA Polymerase a . . . . . . . . . . . , . A. Purification and Properties. . . . . . . . . B. Biological Role . . . . . . . . . . . . . . 111. DNA Polymerase p . . . . . . . . . . . . . A. Purification and Properties . . . . . . . . . B. Biological Role . . . . . . . . . . . . . . IV. DNA Polymerase y . . . . . . . . . . . . . A. Purification and Properties . . . . . . . . . B. Biological Role . . . . . . . . . . . . . . V. Herpes Simplex Virus-Induced DNA Polymerase A. Purification and Properties . . . . . . . . . B. Biological Role . . . . . . . . . . . . . . VI. Vaccinia Virus-Induced DNA Polymerase . . . A. Purification and Properties . . . . . . . . . B. Biological Role . . . . . . . . . . . . . . VII. Conclusion . . . . . . . . . . . . . . . . .

1.

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

. . .

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

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

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

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

. .

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

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

67 69 69 73 73 74 76 76 77 79 80 80 83 83 84 85

86

Introduction and Perspective

A nomenclature for the known eukaryotic DNA polymerases was proposed in 1975 (/). This classification, which recognized three major 1. A. Weissbach, D. Baltimore, F. J. Bollum, R. C. Gallo, and D. Korn, Scicizcc 190, 401 (1975).

67 THE ENZYMES, Vol. XIV Copyright @ 1981 by Academic Press. Inc. All rights of reproduction in any form reserved

68

A. WEISSBACH

classes of cellular DNA polymerases--cY, /3, and y-has proved to be applicable to a wide range of species in the Animal Kingdom (2). In addition, the nomenclature scheme recognized the existence of a mitochondria1DNA polymerase and new viral-induced DNA polymerases that are produced in infected animal cells. This chapter limits itself to consideration of enzymes that have been well characterized, i.e., DNA polymerases a , P , and y from mammalian cells and the herpes simplex virus and vaccinia virus-induced DNA polymerases. A number of other reported DNA polymerase activities ( 2 4 ) , less well identified, are necessarily omitted. The three cellular DNA polymerases were named in the order of their discovery: DNA polymerase a was first isolated from calf thymus and characterized by Bollum (5, 6), and his work was an important starting point in the study of eukaryotic DNA polymerases. The enzyme has been identified in many species as a predominant activity, and has been purified from rat, murine, calf thymus, and human cells. In 1971 Weissbach et al. (7) reported a new, low molecular weight DNA polymerase in HeLa cell nuclei at the same time that Baril et al. (8)reported an enzyme with similar properties in rat liver nuclei. This enzyme has been designated as ppolymerase. Further identification and characterization of P-polymerase in calf thymus, rat liver, KB cells, and chick embryos followed shortly thereafter (9-11). In general, P-polymerase represents about 5% of the total DNA polymerase in growing, cultured cells, but is a major component of low DNA polymerase activity in quiescent cells. DNA polymerase y , first reported in 1972 by Fridlender et ul. (12), is a widely distributed enzyme (13) that shows a marked ability to copy ribohomopolymers, but a total inability to use natural RNA as a template. It thus has no relationship to the retrovirus reverse transcriptases. At first thought to be a minor activity in the cell, the DNA polymerase y level in 2. A. Weissbach, Annu. Rev. Biochem. 46, 25 (1977). 3. J . A. Kantor, X. H. Lee, J. G . Chirikjian, and W. G . Feller, Science 204, 511 (1979). 4. B. A. Brennessel, D. P. Buhrer, and A. Gottlieb,Anal. Biochem. 87, 411 (1978). 5. F. J. Bollurn, JBC 235, 2399 (1960). 6. M. Yoneda and F. J. Bollurn, JEC 240, 3385 (1965). 7. A. Weissbach, A. Schlabach, B. Fridlender, and A. Bolden,Notlrre New B i d . 231, 167 ( 197 1).

8. E. F. Baril, 0. E. Brown, M. D. Jenkins, and J. Laszlo, Binrhemistry 10, 1981 (1971). 9. L. M. S . Chang and F. J. Bollurn. JBC 246, 5835 (1971). 10. M. E. Haines, A . M. Holrnes, and I. R. Johnston, FEES (Fed.Eur. Biochem. SOC.) Lett. 17, 63 (1971). 11. H. Berger, Jr., R. C. Huang, and J. L. Irvin, JBC 246, 7275 (1971). 12. B. Fridlender, M. Fry, A. Bolden, and A . Weissbach, PNAS 69, 452 (1972). 13. A. Weissbach, Cell S, 101 (1975).

5. CELLULAR AND VIRAL-INDUCED DNA POLY MERASES

69

growing cultured cells is, in fact, equal to that of thep-polymerase, and in at least one case the y-polymerase is the major polymerase in the organism. Table I summarizes some relevant characteristics of the three cellular DNA polymerases obtained from various sources and also lists two wellcharacterized viral-induced DNA polymerases. It should be emphasized here that the properties, size, and behavior of the a-, P - , and y-polymerases can differ from those shown in Table I, depending on the source of the enzyme. The specific patterns shown by inhibitors, however, seem to be invariant. Each of these enzymes is considered in detail in the following sections.

II. DNA Polymerase a

DNA polymerase a has been extensively purified from calf thymus (14, 1 5 ) , human cells (16), murine cells (171, and others (2). Because of the extensive heterogeneity of the enzyme in various species, isolation of a-polymerase in a pure form has been difficult. Nevertheless, a nearhomogeneous preparation of a-polymerase has been obtained from human cells by Fisher and Korn (161, and from mouse myeloma cells by Chen et al. (17). A. PURIFICATION AND PROPERTIES Table I1 summarizes the purification of a-polymerase from cultured human KB cells as described by Fisher and Korn (16). In this procedure all buffers contained 1 rnM P-mercaptoethanol and 1 m M EDTA and, after fraction V, 20% glycerol. A protease inhibitor, p-toluenesulfonyl fluoride, is present when the cells are broken. The second DEAE-cellulose step, which offers little purification per se, is apparently important for subsequent steps. The purified protein from human cells has a specific activity of 206,000 unitdmg, a unit being defined as the incorporation of 1 nrnol of dTMP in DNMhour at 37”. The enzyme, as isolated, exists either as a monomer of 140,000 daltons or as dimers of 265,000-280,000 daltons. It has an isoelectric point of 5.0-5.2 and can be resolved in denaturing polyacrylamide gel 14. 15. 16. 17.

F. J. Bollum, Progr. Nucleic Acid Res. Mol. Biol. 15, 109 (1975). K. McKune and A. M. Holmes, Nucleic Acids Res. 6, 3341 (1979). P. A. Fisher and D. Korn, JBC 252, 6528 (1977). Y. C. Chen, E. W. Bohn, S. R. Planck, and S. H. Wilson, JBC 254, 11678 (1979).

TABLE I

EUKARYOTIC DNA POLYMERASES DNA polymerases

Source Human cells

Major cellular location Nucleus, cytoplasm

Molecular weight (X l t 3 )

Subunits (kilodaltons)

156

76, 66

(KB) Y

Novikoff hepatoma Chick embryo

Herpes simplexinduced Vacciniainduced

Infected HeLa cells Infected HeLa cells

Nucleus

3 I"

Nucleus, mitochondria

180

47

Nucleus

144

74,29

Cytoplasm

115 ~

a

~

~~~

~~~~~~~~

DNA polymerase /3 from other cells has a reported molecular weight of 40,OOO-45,OOO.

Inhibitors Aphidicolin, N-ethylmaleimide, Ara-ATP Dideoxynucleoside triphosphates, iodoacetate Dideoxynucleoside triphosphates, N-ethylmakimide Phosphonoacetate, Ara-ATP, N-ethylmaleimide Phosphonoacetate, N-ethylmaleimide

TABLE I1 PURIFICATION OF

Step Crude extract pH 5.5 precipitation

Ultracentrifugation First DEAE-cellulose

2

Second DEAE-cellulose Phosphocellulose Hydroxylapatite DNA-ceUulose Gel electrophoresis'

Fraction I I1 (resolubilized precipitate) 11' (supernatant) 111 (supernatant) 111' (pellet) IV (adsorbed) IV' (flow-through) V VI VII VIII IX

DNA POLYMERASE a FROM KB CELLS" Velum$ (ml)

Proteid (mg)

Activit9 (units)

9.9 2.2

43 19

880 890

2.8

9

2.9

9 x 10-1

1.1

4 x 10-1 4 x 10-2 1 x 10-2 I x 1 r 3 1.6 x 10-4

3.8 x lo-' 6.6 x lo-* 6.5 x lo-*

60 790 50 490 100 280 240 130

33

Specific activity (unitdmg)

Yield (%)

20 47

(100) 108

88

97

550

67

700 6,000 13,000 33,000 206,000

32 28

I5 4

" As described by Fisher and Korn (16). Reaction mixes contained in 2 5 0 ~ 1 10 , mM Tris, pH 9.2,20 mM mercaptoethanol, bovine serum albumin 200 pglml, 10 m M MgCb, activated salmon sperm DNA, 800 pglrnl, dATP, dCTP, dGTP, and dTTP, 50 g C L M each, and PHIdlTP at a final specific activity of 0.04 Ci/mmol, and enzyme. A unit is the amount of enzyme that catalyzes the incorporation of 1 nmol of labeled dTMP into an acid-insoluble product in 1 hour at 37". Quantities are expressed per gram wet weight of KB Cells. Aliquots, 400 p l , of fraction VIII were used for nondenatunng gel electrophoresis. The protein value was derived by densitometry of a stained gel. Recovery of DNA polymerase activity by elution of slices of parallel unstained gels varied between 50 and 95%. The specific activity value is based on units of loaded activity.

72

A. WEISSBACH

electrophoresis into two subunits of 76,000 and 66,000 daltons. The purified enzyme has a half-life at 0" of 14 months if stored in a concentrated form in the presence of sucrose and potassium phosphate. Optimal reaction conditions include a pH between 7.5-8.5, and Mg'+ at 4-8 mM. Salt concentrations above 50 mM are inhibitory, with about 50% of the activity lost at 100 mM KCl. a-Polymerase is markedly inhibited by Caz+ and Li+. a-Polymerase is most reactive with duplex DNA templates containing gapped regions with available 3'-OH termini (activated DNA). A surprising property of the purified enzyme is its inability to catalyze the synthesis of long DNA chains. It is only slightly processive, synthesizing an 11 2 5 nucleotide length before coming off the template (18, 19). DNA polymerase a does not act at nicks or in short gaps below 20 nucleotides in length, and does not utilize a blunt-ended DNA template. The enzyme binds to single-stranded DNA that contains 3'-OH ends and can catalyze synthesis of hairpin molecules from such templates (20, 21). Synthetic are copied at 20% the rate of DNA templates such as (dA), . (dT)12--IB activated DNA, whereas the corresponding synthetic RNA template, (A), . dTlz, copies at only 3% the rate of activated DNA. However, murine DNA polymerase a copies (dT), . rA2, faster than any other template (22).Spermidine has been found to increase the apparent K , for M. Purified calf DNA (20). The K , for dNTPs is in the range of 1-4 x thymus DNA polymerase a catalyzes both pyrophosphorolysis and pyrophosphate exchange (14). As isolated, purified human a-polymerase has no detectable nuclease activity. There have been reports that bone marrow and calf thymus contain an a-polymerase-like enzyme that contains a 3' + 5' exonuclease, and has been called DNA polymerase 8 (23). Whether this represents a new enzyme or an association of a cellular 3' + 5' exonuclease (24) with the a-polymerase is still unclear. Chen et al. (17) obtained two nearhomogeneous preparations of a-polymerase from mouse myeloma. These 18. P. A. Fisher, T. S-F. Wang, and D. Korn,JBC 254, 6128 (1979). 19. K. McKune and A. M. Holmes, BBRC 90, 864 (1979). 20. P. A. Fisher and D . Korn, JBC 254, 11033 (1979). 21. P. A. Fisher and D. Korn,JBC 254, 11040 (1979). 22. S. H. Wilson, A. Matsukage, E. W. Bohn, Y. C. Chen, and M. Sivarajan, Nucleic Acids RPS.4, 3981 (1977). 23. M. Y. W. Tsang-Lee, C. K. Tan, A. G . So, and K. M. Downey,Biochemistry 19,20% (1980). 24. G. Villani, S. Spadari. S. Boiteux, M. Defais, P. Caillet-Fauquet, and M. Radman, Biochimie 60, 1145 (1978).

5 . CELLULAR AND VIRAL-INDUCED DNA POLYMERASES

73

large molecular weight enzymes (MW = 190,000) contain subunits of 47,000 and 54,000 daltons. One of the a-polymerase species contains both 3' + 5' and 5' + 3' exonucleases associated with it. a-Polymerase has also been isolated in highly purified form from calf thymus (15) and regenerating rat liver; in both cases a number of subunits ranging from 50,000 daltons to 70,000 daltons seem to be associated with a catalytic polypeptide whose molecular weight is about 150,000. Thus, a common denominator in many of these studies is the heterogeneity of DNA polymerase a, a feature which may have important implications in the control and function of this enzyme. B.

BIOLOGICAL ROLE

It is generally agreed that DNA polymerase a has a key role in the replication of nuclear DNA and in the synthesis of the DNA of the viruses SV40, polyoma, and adenovirus (14, 25). Understanding the role of a-polymerase has been aided by the availability of specific inhibitors such as aphidicolin, or the arabinose-containing nucleotides such as ara-ATP. The use of these inhibitors supports the concept that DNA polymerase a is the major replicative polymerase in mammalian cells (26-30). It represents 90-95% of the total DNA polymerase activity of cultured growing mammalian cells and drops to low levels in cells that have ceased nuclear DNA synthesis. 111.

DNA Polymerase /3

DNA polymerase /3 is the smallest of the known eukaryotic DNA polymerases and shows remarkable chemical stability under various conditions. As a result, and although it represents only about 5% of the total DNA polymerase in growing, cultured mammalian cells, it was the first eukaryotic DNA polymerase to be isolated in a homogeneous state. This has been accomplished from calf thymus (3I ), human KB cells (32), mouse 25. 26. 27. 28. 29. 30. 31. 32.

A. Weissbach, ABB 198, 386 (1979). H. J. Edenberg, S. Anderson, and M. L. DeParnphilis, JBC 253, 3273 (1978). M. A. Waqar, M. J . Evan, and J. A. Huberman, Nucleic Acids Res. 5, 1933 (1978). M. Ohashi, T. Taguchi, and S . Ikegami, BBRC 82, 1084 (1978). E. Wist and H. Prydz, Nucleic Acids Res. 6, 1583 (1979). E. Wist, BBA 562, 62 (1979). L. M. S. Chang, JBC 248, 3789 (1973). T. S-F. Wang, W. D. Sedwick, and D. Korn, JEC 250, 7040 (1975).

74

A. WEISSBACH

myeloma (33),Novikoff hepatoma (34) and chick embryos (35). The purification of DNA polymerase /3 from the latter is summarized in Section II1,A. A. PURIFICATION AND PROPERTIES The procedure used by Stalker et al. (34) is shown in Table 111, and yields a homogeneous enzyme after a 200,000-fold purification with a remarkable apparent yield of 46%. The starting material, Novikoff hepatoma, is an ascites tumor with a generation time of 12 hours when maintained in rats, so relatively large quantities of cells can be obtained conveniently. The purification relies on the sequential use of three chromatographic separations on DEAE-Sephadex, phosphocellulose, and hydroxylapatite. The final step in the purification procedure uses singlestranded DNA cellulose as an affinity column, which in a 25-fold enrichment step provides the pure enzyme. The enzyme is stable at 4" during the isolation procedure, and stabilization of the enzyme during purification is facilitated by the use of 10% glycerol in the elution buffers and, at the final step, by having bovine serum albumin (1 mg/ml) present. With this method, 200 pg of purified DNA polymerase P are obtained per kilogram of cells. The enzyme in whole cells is stable at - 20" for months, and the purified enzyme has been stored for 1 year at - 196" without loss of activity. The enzyme, as isolated, has a molecular weight of 31,000 although the calf thymus (31), KB cell (32), and chick (35) and mouse P-polymerases (33) have been reported to have molecular weights of 44,000, 43,000 and 40,000 daltons, respectively. Purified DNA polymerase /3 has no detectable nuclease activity. It shows an alkaline isoelectric point (8.51, a pH optimum of 8.4-9.2 and a K, for deoxynucleoside triphosphates of 7-8 p M . For maximal synthesis Mg2+ at 5-10 mM is required; MI?+ (1 mM) can also be used. The enzyme is stimulated twofold by 50 mM NaCl or by 100-200 mM KC1. The latter salt levels severely inhibit a-polymerase. Phosphate and pyrophosphate are inhibitory to /3-polymerase and should be avoided in reaction mixes. Neither the mouse nor human enzyme can catalyze pyrophosphate exchange, pyrophosphorolysis, or dNTP turnover (33, 36). 33. K. Tanabe, E. W. Bohn, and S. H. Wilson, Biochemistry 18, 3401 (1979). 34. D. M. Stalker, D. W. Mosbaugh, and R. R. Meyer, Biochemistry IS, 3114 (1976). 35. M. Yamaguchi, K. Tanabe, Y. N . Taguchi, M . Nishizawa, T. Takahashi, and A. Matsukage, JBC 255, 9942 (1980). 36. T. S-F. Wang, W. D. Sedwick, and D. Korn, JBC 249, 841 (1974).

75

5 . CELLULAR AND VIRAL-INDUCED DNA POLYMERASES

TABLE I11

PURIFICATION OF NOVIKOFF HEPATOMA DNA POLYMERASE P" Protein (mg)

Total unitsh

Specific (unitdmg)

Purification (-fold)

Yield

Fraction 1. Cell extract 11. Ammonium sulfate 111. DEAE-Sephadex IV. Phosphocellulose V. Hydroxylapatite VI. DNA-Cellulose

14,500 4,610 8,866 81.2 1.34 0.031

3,880 3,780 9,820 4,390 3,150 1,800

0.268 0.820 11.3

1.0 3.06 42.2 202 8,770 217,000

100 97.4 253

54.1

2,350 58,100'

(%)

113

81.2 46.4

' I From Stalker et al. (35).The reaction mixtures contained the following components in a final volume of 125 111: 25 mM Tris-HCI, pH 8.4; 5 mM 2mercaptoethanol; 7 mM magnesium acetate; 0.5 mM EDTA; 0.015 m M each of dATP, dCTP, dGTP, and PHIdTTP (specific activity 975 rnCi/mmol); 50 m M NaCI; 15% ( w h ) glycerol; 250 Fg/ml activated DNA; and 0.01-0.3 units of DNA polymerase fraction. Incubations were carried out for 1 hour at 37" and acid-insoluble radioactivity was determined. When incorporation was not linear for 1 hour, the data were extrapolated from a 30-minute incubation. * A unit is defined as the incorporation of 1 nmol of total nucleotide into DNA per hour at 37". ' With several different preparations, the specific activity varied from 32,000 to 62,000 unitslmg.

A reported characteristic of DNA polymerase /3 is its relative insensitivity to urea, acetone, and alcohol (14). The enzyme is stabilized by glycols and stimulated by spermidine (up to 10 mM) (34). Another general property of the P-polymerases is their relative resistance to N-ethylmaleimide (NEM), which is a powerful inhibitor of DNA polymerases a and y . At 4 mM, NEM shows a 28% inhibition of DNA polymerase @ from Novikoff hepatoma, a value that is slightly higher than previously reported for the human enzyme (37). The NEM partial inhibition is not unexpected since p-hydroxymercuribenzoate inhibits @-polymeraseat concentrations above 50 p M (14, 34). An important characteristic of P-polymerase is its ability to copy a synthetic ribohomopolymer such as (A), . dTlz as well as the corresponding deoxyribohomopolymer (dA), . dTlz or activated DNA (34).This is in contrast to a-polymerase, which utilizes the deoxyribohomopolymer (dA), . dT,,_,, eight times better than (A), . dTlz, which is, in fact, copied at only 3% the rate of activated DNA (18). Rat DNA polymerase 0 has been reported to have a uniquely high requirement for primers when 37. K. W. Knopf, M. Yamada, and A. Weissbach, Biochemistry 15, 4540 (1976).

76

A. WEISSBACH

copying poly(A) templates, and can thus be distinguished from y-polymerase or oncornavirus reverse transcriptase (38). Steady-state kinetic measurements suggest an ordered BiBi mechanism for polymerization and a scheme depicting two DNA binding sites on the enzyme has been advanced (33).Although the specific activity of the purified Novikoff hepatoma DNA polymerase p prepared by Stalkeret al. (S4)is 58,000 with activated DNA as a template, Ono et al. (38)reported rat ascites hepatoma DNA polymerase preparations with a specific activity of lo6 units/mg on an (A), * dTlz-ls template.

B. BIOLOGICAL ROLE The level of DNA polymerase L,3 in quiescent or growing cells or during the cell cycle has been reported to be relatively constant, leading to the suggestion that it may be involved in DNA repair synthesis (39). Hiibscher et al. (40) have shown that &polymerase can participate in the repair of UV-damaged DNA in neuronal nuclei, an organelle in which DNA polymerase p is the only detectable polymerase activity. The further role of this enzyme in other types of DNA synthesis is unknown at the present time.

IV.

DNA Polymerase y

DNA polymerase y exists in at least two forms and is found in the nucleus, cytoplasm, and mitochondria (41-43). Like a-polymerase, DNA polymerase y readily undergoes reversible aggregations that, in v i m at least, are salt-dependent. It comprises about 5% of the total DNA polymerase activity in the growing, cultured mammalian cell, and therefore is about equal to the &polymerase level. In developing chick embryos, y-polymerase represents 45% of the total DNA polymerase activity and is present in larger amounts than either 0- or P-polymerases (44). It is clear that the so-called “mitochondrial” DNA polymerase is one species 38. K. Ono, A. Ohashi, K. Tanabe, A. Matsukage, M. Nishizawa, and T. Takahashi, Nucleic Acids Res. 7, 715 (1979). 39. G. Pedrali Noy, L. Dalpra’, M. A. Pedrini, G. Ciarrocchi, E. Gidotto, F.Nuzzo, and A. Falaschi, Nucleic Acids Res. 1, 1183 (1974). 40. U. Hiibscher, C. C. Kuenzle, and S. Spadari, PNAS 76, 2316 (1979). 41. S. Spadari and A. Weissbach, JBC 249, 5809 (1974). 42. G. Pedrali Noy and A. Weissbach, BBA 477, 70 (1977). 43. A. Bolden, G. Pedrali Noy, and A. Weissbach, JBC 252, 3351 (1977). 44. M. Yamaguchi, A. Matsukage, and T. Takahashi, JBC 255, 7002 (1980).

5 . CELLULAR AND VIRAL-INDUCED DNA POLY MERASES

77

of the y-polymerase class, and that the nuclear species of DNA polymerase y can be distinguished from it (43). Despite extensive efforts, the enzyme has not been prepared in pure form from mammalian tissues (45), perhaps due, in part, to its heterogeneity: but it has been purified to near-homogeneity from chick embryos (44). A.

PURIFICATION AND PROPERTIES

An outline of the purification of the y-polymerase from chick embryos as described by Yamaguchi et al. (44) is shown in Table IV. In this procedure, frozen 11-day-old embryos are minced and sonicated in a buffer containing 0.5 M KCl, 10% glycerol, and eventually, 0.5% Triton X-100. The purification scheme uses two phosphocellulose column chromatographic steps, a Sephadex G-200 gel filtration, and hydroxylapatite adsorption chromatography. Following the second phosphocellulose column, the enzymatic activity separates into two gel components, a 180,000- and a 280,000-dalton species, during gel filtration on a Sephadex G-200 column, and each species is further purified separately. The final separation step, which gives a 1000-foldenrichment, involves affinity chromatography on a double-stranded DNA cellulose column, and can be compared to the single-stranded DNA cellulose columns used in the purification of a- and P-polymerases. Attempts to purify DNA polymerase y by d n i t y chromatography on poly(rA)-Sepharose columns leads to inactivation of the enzyme. The purified enzyme can be stored at -80" but loses 50% of its activity in one freeze-thaw cycle. The purified enzyme sediments at 7.5 S and the molecular weight is estimated to be 180,000. SDS polyacrylamide gel electrophoresis shows a prominent polypeptide at 47,000 daltons, so the native enzyme appears to be a tetramer of this subunit. Based on this, the specific activity of the purified enzyme is calculated to be 660,000 unitslmg on a poly(A) template. With (A), dTlz-ls as a primer-template, the K , value for dTTP is about 1 p M, the optimal pH 8.5-9.0, and the optimal KCI concentration is 220 p M. In the presence of increasing levels of potassium phosphate, the optimal KCI concentration drops proportionately (45); and Mn2+ at 0.5-0.6 mM is fivefold more effective than the optimum Mg2+concentration of 12 mM. The structure of native DNA polymerase y in mammalian cells will probably differ somewhat from the avian enzyme. Rat liver DNA polymerase y can be obtained as a 4 S species (60,000 daltons) (43), whereas the smallest species of the chick native enzyme sediments at 7.5 45. K-W. Knopf, M. Yamada, and

A. Weissbach, Biochemistry

15, 4540 (1976).

TABLE IV

PURIFICATION O F DNA POLYMERASE y FROM CHICK

Step Crude extract First phosphocellulose and ammonium sulfate fractionation Second phosphocellulose Sephadex G-200 Hydroxylapatite Double-stranded DNA cellulose

Fraction I

I1

EMBRYOS~

Protein

Activityb units

(mg)

(%)

0,38 3s

80 4.8 4.1 x 10-' 1.0 x lo-'

I11 IV- 1 IV-2 v-1 v-2 VI- 1

9.0 x

4.1(14) 5.0(17) 3.6(12)

(VI-1-dT VI-2

(1.3 x 10-7 1.3 x lCV

(4 4.5( 15)

9.0 x 1C2 8.4 x 10-3 8.1 x 10-3

specific activity (unitdmg)

2x77) 8.4(28) 9.0(30)

56 84 100 490 620 400.000

Purification (-fold) 1 9.2 150

220 260 1,300 1,600 1,100,000 (1,500,000) 920,000

From Yamaguchi et al. (44). The assay mixture contained in 25pl,50 mM Tns, pH 8.5, 1 mM dithiothreitol, 0.5 mM MnC&.,80 pglml poly (rA) 16 pg/mI dT,,-,,, 0.1 mM [3H]dTTP(60cpdpmol), 15% glycerol, 400 pg/ml bovine serum albumin, 100-120 mM KCI, 20-40 mM potassium phosphate (pH 8.5) and enzyme. A unit is the amount of enzyme that catalyzes the polymerization of 1 nmol of dTMP in 60 minutes.

* Quantities are expressed per gram wet weight of

11-day-old chick embryos.

' Fractions in the peak of DNA polymerase activity (see Fig. 5A).

S. CELLULAR A N D VIRAL-INDUCED DNA POLY MERASES

79

S (150,000-180,000 daltons). By contrast, HeLa cell DNA polymerase y can be separated into two species on phosphocellulose chromatography, both with the same or similar apparent molecular weight of 110,000 (do), and human lymphoblast DNA polymerase y has a reported molecular weight of 120,000 (46). The interspecies difference is further illustrated by the report that sea urchin DNA polymerase y has a sedimentation value of 3.3 s (47). A salient feature of y-polymerases is their ability to copy ribohomopolymers at a rate greater than activated DNA. Under proper conditions the HeLa cell DNA polymerase y will utilize (A), * dT12-18five to ten times more efficiently than activated DNA (44). This is in contrast to the template characteristics of a-polymerase, which utilizes this synthetic template at 3% the rate at which it uses activated DNA. In addition, y-polymerase is active at potassium phosphate concentrations (50 mM) that are inhibitory to DNA polymerase /3(14,45), an enzyme that is known to copy (A), . dTlz-18 at about the same efficiency it copies activated

DNA. B. BIOLOGICAL ROLE Of the three cellular DNA polymerases, only DNA polymerase y is capable of synthesizing continuous long DNA chains in a processive manner (48). DNA polymerase a , by comparison, is highly discontinuous, polymerizing 10-15 nucleotides at a time and then leaving the template (18,49).,&Polymerase also shows discontinuous synthesis when copying a poly(A) template (49). The ability of the y-polymerase to carry out a processive and continuous synthesis of a DNA chain may explain its known physiological roles. One of the forms of the enzyme is responsible for mitochondrial DNA synthesis (43, 50, 51) and another, the nuclear y-polymerase, is involved in the replication of adenovirus DNA (52, 53). The synthesis of adenovirus DNA and mitochondria1 DNA share a strand-displacement step in their replication process; this has led to the 46. M. Robert-Guroff, A. W. Schrecker, B. J . Brinkman, and R. C. Gallo, Biochemistry 16, 2866 (1977).

47. A. Habara, H. Nagano, and Y. Mano, BBA 561, 17 (1979). 48. M. Yamaguchi, A. Matsukage, and T. Takahashi, Nature (London) 285, 45 (1980). 49. A. Matsukage, M. Nihizawa, T. Takahashi, and T. Hozumi, J . Eiochern. (Tokyo),in press (1980). 50. U. Hubscher, C. C. Kuenzle, and S . Spadari, PNAS 76, 2316 (1979). 51. W. timmemann, S-M. Chen, A. Bolden, and A. Weissbach,JBC 255, 11847 (1980). 52. P. C. Van der Vliet and M. M. Kwant, Nature (London) 276, 532 (1978). 53. H. Krokan, P. SchaEer, and M. L. DePamphilis, Biochemistry, 18, 4431 (1979).

80

A. WEISSBACH

suggestion that y-polymerase has a unique role in strand-displacement syntheses (52, 25). Since both mitochondria1 DNA and adenovirus DNA are synthesized in a continuous mode without the apparent formation of short intermediates, such as Okazaki fragments, the processive character demonstrated by y-polymerase in vitro is also reflected in vivo . However, it is apparent that the basic physiological role of DNA polymerase y in the nucleus of the cell remains unknown. V.

Herpes Simplex Virus-Induced

DNA

Polymerase

The recognition in 1963 that herpes simplex virus (HSV) induced a new DNA polymerase in infected cells (54, 5 5 ) followed shortly after the discovery of DNA polymerase a , and predates the identification of DNA polymerases p and y . The HSV-induced DNA polymerase is therefore one of the earliest eukaryotic DNA polymerases studied. The altered properties of the enzyme were recognized by Keir et al. ( 5 3 , and the enzyme was partially purified and characterized by Weissbach er a!. (56). Highly purified, near-homogeneous preparations of the HSV polymerase have been prepared from HSV-1-infected HEp-2 cells (57) and from African green monkey cells (58). Purification of the viral-induced enzyme is facilitated by the large amounts of virus that are produced in the infected cell (59). Thus, the amount of HSV-1 DNA polymerase in HSV-1-infected HeLa cells can rise to four times the combined level of all the host cell DNA polymerases. A. PURIFICATION AND PROPERTIES As described by Knopf (58), African green monkey cells (RC-37; Italdiagnostic Products) grown in monolayers were infected at 5 pfdcell with HSV-1 (Angelotti) that had previously been passed through RC-37 five times. Six hours after infection the cells were collected, disrupted by sonication in 0.25 M potassium phosphate, pH 7.5, containing 0.5% Triton X-100. All the subsequent purification steps shown in Table V were per54. H. M. Keir, J. Hay, J. M. Momson, and J. Subak-Shape, Nature (London)210, 369 (1966). 55. H. M. Keir, J. Subak-Shape, W. I. H . Shedden, D. H. Watson, and P. Wildy, Virology 30, 154 (1966). 56. A. Weissbach, S-C. L. Hong, J. Aucker, and R. Muller, JBC 248, 6270 (1973). 57. K . L. Powell and D. J . M. hrifoy, J . Vlrol. 24, 616 (1977). 58. K. Knopf, EJB 98, 231 (1979). 59. M. Yamada, G. Brun, and A. Weissbach, J . Virol. 26, 281 (1978).

TABLE V PURIFICATION OF

HSV-1-DNA POLYMERASE FROM INFECTED RC-37 CELLP ~~

Purification

Volume (ml)

Total protein (mg)

Total activity (units)

Specific activity (unitdmg protein)

Purification

(%)

Cell extract dialysate DEAE-cellulose Phosphocellulose DNA-cellulose DNA-cellulose peak (fraction 37)

370 575 247 30 0.9

1191.4 217.8 28.5 1.38 0.033

125,280 150,480 95,168 3 1,570 1,575

105.2 690.9 3339.2 22876.8 47727.3

1 6.6 31.7 217.5 453.7

100 120 76 25

~~

Total recovery

~

From Knoff (58). Reaction mixtures contained in lOOpl50 mM Tns-HCI (pH KO), 7.5 mM MgCl,, 100 m M ammonium sulfate, 5 0 p g bovine serum albumin, 0.5 mM dithiothreitol, 0.1 mM each of dATP, dCTP, dGTP, and PHld'lTP (0.4 Ci/mmol), and 25 pg of activated salmon sperm DNA prepared as described by Pedrali Noy and Weissbach (42).A unit is the amount of enzyme that catalyzes the polymerization of 1 nmol of nucleotide in 60 minutes under standard assay conditions. a

82

A. WEISSBACH

formed with buffers containing 0.5 mM dithiothreitol and 1 mM phenylmethylsulfonyl fluoride. The purification is relatively simple and involves three chromatographic separations on DEAE-cellulose, phosphocellulose, and double-stranded DNA cellulose, which yield a highly purified preparation after only a 450-fold purification. Using similar steps with DEAE-cellulose, phosphocellulose, and single-stranded DNA-cellulose separations, Powell and Purifoy (2 years prior to Knopf's report) purified the HSV-induced polymerase from HEp-2 cells almost 1700-fold with almost a 50% recovery (57). The purified enzyme, stored in 50 mM TrisHC1, 1 mM EDTA in 50% glycerol is stable at - 20 or -70". As isolated by Knopf, the purified enzyme shows a major polypeptide of 144,000 daltons on SDS polyacrylamide gel electrophoresis, which is in agreement with the 150,000-dalton species found by Powell and Purifoy (57). The enzyme isolated from RC-37 cells also shows the presence of two other polypeptides of 74,000 and 29,000 daltons, which were not observed by Powell and Purifoy and which may represent impurities. A prominent feature of HSV-DNA polymerase, and one that facilitates its identification, is its activity at high salt concentrations. The presence of 150 mM KCI or 100 mM (N&)2S04 leads to a two- to threefold enhancement of the enzymatic activity, whereas the cellular a-polymerase is inhibited nearly 90% at these salt concentrations. The HSV-1-induced DNA polymerase has a pH optimum of 8-8.5, and a M$+ optimum of 3 mM (in the presence of activated DNA template). Dithiothreitol ( 5 mM) also stimulates the enzyme threefold. The enzyme is inhibited by Zn2+, N-ethylmaleimide, and the pyrophosphate analogs, phosphonoacetic acid, or phosphoformate (60-62). Inorganic pyrophosphate does not inhibit the enzyme, which is able to catalyze pyrophosphate exchange into dNTPs. Aphidicolin, a powerful inhibitor of DNA polymerase a , also inhibits the HSV-induced DNA polymerase as well as the vaccinia-induced DNA polymerase described in Section VI (63). It has been observed that any inhibitor of DNA polymerase a also inhibits the HSV-induced DNA polymerase and the vaccinia-induced DNA polymerase, and vice versa (21). Since there is no known relationship or structural similarity between these viral-induced enzymes and DNA polymerase a , it will be of consid60. A. Bolden, J. Aucker, and A. Weissbach, J . Virol. 16, 1584 (1975). 61. S . Leinbach, J. M. Reno, L. Lee, A . F. Isbell, and J. A. Baezi, Biochtrnistry 15, 426 (1976). 62. B. Eriksson, A. L&rsmn,E. He!gstrand, N. G. Johansson, and B. Oberg, BBA607,53 (1980). 63. G . Pedrali Noy and S . Spadari, J . Virology 36, 457 (1980).

5 . CELLULAR AND VIRAL-INDUCED DNA POLYMERASES

83

erable interest to elucidate the active sites of these enzymes and compare them. The HSV-DNA polymerase contains a 3' + 5' exonuclease activity that copurifies with the enzyme and is apparently an intrinsic activity. This is in contrast to the purified host-cell DNA polymerases, which are devoid of nuclease activity in their most purified form, although preparations of DNA polymerase (Y with nuclease activity have been reported (22, 2 3 ) . Whether the exonuclease serves as a "proof-reading'' activity, as has been postulated for E. coli DNA polymerase I (64) and T4-DNA polymerase ( 6 3 , remains to be determined.

B. BIOLOGICAL ROLE Herpes virus contains a relatively large genome of about 10' daltons. A genome of this size would be expected to code for 100-150 proteins, and it would not be surprising if one of these proteins might be a new DNA polymerase. Genetic evidence for this exists since certain viral DNA negative mutations are located at the chromosomal site that determines the DNA polymerase expression (66, 67). It thus appears self-evident that HSV-induced DNA polymerase is required for synthesis of the viral DNA. In addition, almost all other members of the herpes group seem to induce a new DNA polymerase in host cells after infection (25).

VI.

Vaccinia Virus-Induced DNA Polymerase

The pox viruses, of which vaccinia virus is a member, are among the largest viruses and contain DNA genomes of 1.2-2 x 10' daltons. Jungwirth and Joklik (68) and Magee and Miller (69) suggested in the 1960's that vaccinia virus could induce a new DNA polymerase in infected cells. This viral-induced DNA polymerase was partially purified by Berns er al. (70), and later clearly separated from the host DNA polymerases by 64. M. P. Deutscher and A. Kornberg, JBC 244, 3019 (1969). 65. M. S. Hershfield and N . G. Nossal, JBC 247, 3393 (1972). 66. P. Chartrand, C. S . Crumpacker, P. S . Schaffer, and N. M. Wilkie, Virology 103,311 (1980). 67. L . E. Schnipper and C. S. Crumpacker, PNAS 77, 2270 (1980). 68. C. Jungwirth and W. K. Joklik, Virology 27, 80 (1965). 69. W. E. Magee and 0. V. Miller, Virology 31, 64 (1967). 70. K. I. Berns, C. Silverman, and A. Weissbach, J . Virol. 4, I5 (1969).

84

A. WEISSBACH TABLE VI

PURIFICATION OF

VACCINIAVIRUSDNA POLYMERASE"

Fraction

Activity (units x lo-'])

Protein (mg)

Specific activity (unitdmg)

55 28 6.6 3.9 2.8 1.7

1,495 268 10.9 1.3 0.52 0.089

36 104 610 2,800 5,400 19,000

I. Extract' 11. DEAE-cellulose'

111. IV. V. VI.

DNA-agarose Phosphocellulose Hydroxylapatite Glycerol gradient

From Challberg and Englund (72). A unit is the amount of enzyme that catalyzes the incorporation of 1 nmol of total nucleotide into an acid insoluble form in 30 minutes at 37". ' Activity in Fractions I and I1 includes both vaccinia and host polymerases.

Citarellaet al. (71). It has been purified to near homogeneity from infected HeLa cells by Challberg and Englund (72). A. PURIFICATION AND PROPERTIES Because of the relatively large amount of viral-induced DNA polymerase formed in the infected cell ( 7 / ) , Challberg and Englund (72) were able to isolate 100 p g of purified enzyme from 27 g of vacciniainfected HeLa cells. Vaccinia-infected HeLa cells, obtained 53 hours after infection, and stored at -2W, were broken by Dounce homogenization in 10 volumes of 10 mM NaC1, 2 mM Tris, pH 7.6, 0.1 mM benzamidine. The lysate was clarified by centrifugation at 15,OOOg, and the supernatant fluid containing 13% glycerol and 4 m M diisopropyl fluorophosphate (DFP) was incubated 1 hour at 0" and applied to a DEAE-cellulose column. The outline of the further purification procedure is shown in Table VI, and is unique in that the DNA affinity column step is performed before the phosphocellulose and hydroxylapatite steps. Elution of the enzyme activity in each chromatographic separation utilizes buffers containing 10% glycerol and yields about a 50% recovery of enzymatic activity in each step. The final step of the preparation yields an enzymatic activity that is at least 500-fold purified from the crude cytoplasmic fraction and is 95% homogeneous. 71. R. V. Citarella, R. Muller, A. Schlabach, and A. Weissbach, J. Virol. 10, 721 (1972). 72. M. D. Challberg and P. T. Englund, JBC 254, 7812 (1979).

5. CELLULAR AND VIRAL-INDUCED DNA POLYMERASES

85

The vaccinia DNA polymerase activity in the infected cells is stable for one month at -20” and is stable for 24 hours at 0” in the cytoplasmic extract (fraction I). The most purified preparations (fractions IV and VI) are stable for months at -20”. In the absence of protease inhibitors, such as DFP and benzamidine, proteolysis of the enzyme during purification occurs even at the phosphocellulose step. Native vaccinia-DNA polymerase is a single polypeptide with a molecular weight of 110,000-1 15,000. It is maximally active in the presence of 5 mM MgC1, and shows a pH optimum in 50 mM potassium phosphate, at 8-9. Its activity in Tris-HC1 at the same pH is 10% that shown in potassium phosphate buffers. The enzyme requires the presence of SH groups and is inhibited by 10 mM N-ethylamaleimide or 30 pM p chloromercuribenzoate. In contrast to the herpes simplex-induced DNA polymerase, the vaccinia-DNA polymerase is inhibited by salt (50% at 200 mM NaCl). The vaccinia DNA polymerase shows maximal activity in an activated DNA template, but will neither nick-translate nor strand-displace a nicked 4x174 DNA template. The enzyme seems sensitive to the secondary structure of the template since in copying 4x174 templates it pauses at regions that contain potential hairpin structures (73). As previously reported (71), the purified polymerase contains a strong exonuclease activity that is apparently part of the DNA polymerase polypeptide since both the polymerase and nuclease activity show the same kinetics of heat inactivation at 45”. The intrinsic nuclease activity is a 3’ + 5’ exonuclease that produces 5’-mononucleotides. Although the pH optimum of the exonuclease, 8-9, is similar to the pH optimum of the polymerase activity, the nuclease activity is twice as active in Mn2’ (50 pM MnC1,) as in the optimum MgCl, concentration (10 mM). In addition, the exonuclease is twice as active in Tris-HC1 as in potassium phosphate and is inhibited 50% by 50 mM NaCl. The polymerase-associated exonuclease hydrolyzes single-stranded DNA somewhat faster than the equivalent duplex DNA. This preference for single-stranded DNA increases as the size of the DNA piece becomes smaller.

B. BIOLOGICAL ROLE The vaccinia-induced DNA polymerase is assumed to be required for the synthesis of the viral DNA, although this remains unproved. Since the genetic loci for this enzyme on the vaccinia chromosome has not been 73. M. D. Challberg and P. T. Englund, J B t 254, 7XLU (IYIY). 74. A. Kornberg, “DNA Replication.” Freeman, San Francisco, 1980.

86

A. WEISSBACH

determined, genetic analysis of the components of DNA replication, as was done for the herpes virus, remains to be investigated. VII.

Conclusion

The mechanism(s) of DNA replication in the cell’s nucleus remain unknown. Further understanding of the physiological role of each of the cellular DNA polymerases will parallel the unraveling of the complex events that accompany and control the synthesis of nuclear DNA. The smaller viral chromosomes, which should be more vulnerable to genetic manipulation and analysis, would seem to offer a promising avenue of research, in parallel perhaps, to the extraordinary detail emerging from the studies of E. coli and its phages (74). The present lack of knowledge portends that our perception of the types and nature of eukaryotic DNA polymerases, as well as their roles, may change within the next few years.

Reverse Transcriptase INDER M. VERMA

I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Purification and Properties . . . . . . . . . . . . . . . . . . . .

A. Purification . . . . . . . . . . . B. Enzymatic Activities of Virions , . C. Physical Properties . . . . . . . 111. Biosynthesis of Reverse Transcriptase IV. Mechanism of Reverse Transcription . V. Applications to Molecular Biology . . A. Synthesis of Complementary DNA B. Synthesis of Double-Stranded DNA C. End Labeling . . . . . . . . . . D. Other Utilities . . . . . . . . .

1.

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

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

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

87 88 88 89 92 94 95 99 99 100 101 103

Introduction

Reverse transcriptase can faithfully transcribe RNA into complementary DNA (cDNA). It was first found in the purified virions of murine leukemia virus (MLV) by Baltimore (f)and Rous sarcoma virus (RSV) by Temin and Mizutani (2). The viral RNA acted as the template to direct the incorporation of deoxyribonucleoside triphosphates. The enzyme plays a central role during the life cycle of a retrovirus (3).Temperature-sensitive mutants with a lesion in the reverse transcriptase are unable to establish 1. D. Baltimore, Nature (London)226, 1209-1211 (1970). 2. H. Temin and S. Mizutani, Narure (London)226, 1211-1213 (1970) 3. I . M . Verma, RBA 473, 1-37 (1977). 87 THE ENZYMES, Vol. XIV Copynght 0 1981 by Academic Press. Inc. All rights of reproduciion in a n y form reserved ISBN 0-12-122714-6

88

INDER M. VERMA

infections (4 1. Reverse transcriptase manifests at least three enzymatic activities: (i) Conversion of RNA into DNA (RNA-dependent DNA polymerase); (ii) conversion of single-stranded DNA to double-stranded DNA (DNA-dependent DNA polymerase); and (iii) selective removal of RNA moiety from an RNA-DNA hybrid (RNase H) (3). It has been suggested that reverse transcriptase may also contain DNA endonuclease (5) and swivelase (“unwinding-like”) activities (6). In addition, reverse transcriptase also contains a “tRNA-primer binding site” (7). Due to its multiple activities, the enzyme reverse transcriptase has been alternatively referred to in the literature as RNA-dependent DNA polymerase, RNA-directed DNA polymerase and DNA polymerase of retrovirus (EC 2.7.7.7). Several review articles have been written on reverse transcriptase over the last few years (for general introduction, Temin and Baltimore (Sf, Temin (91, and Stephenson (10); for detailed review on purification, properties, etc., Green and Gerard ( I / ) , Sarngadharen er af. (12), Verma (3), and Gerard and Grandgenett (13). This chapter is limited to a brief review of the purification of the enzyme and its enzymatic activities, mechanism of reverse transcription, and research applications to molecular biology.

11.

A.

Purification and Propeflier

PURIFICATION

Reverse transcriptase resides in the core of the virion (14, 15) and can be easily solubilized by nonionic detergents like Nonidet P40 (16-19) or 4. M. Linial and W. S. Mason, Virology 53, 258-273 (1973). 5. D. P. Grandgenett, A. C. Vora, and R. D. Schiff, Virology 89, 119-132 (1978). 6. M. S . Collett, J. P. Leis, and A. J. Faras, J . Virol. 26, 498-509 (1978). 7. A. Panet, W.H. Haseltine, D. Baltimore, G. Peters, F. Harada and J. E. Dahlberg, PNAS 72, 2535-2539 (1975). 8. H. M. Temin and D. Baltimore, Advan. Virus Res. 17, 129-186 (1972). 9. H. M. Temin, Annu. Rev. Genet. 8, 155-177 (1974). 10. J. R. Stephenson (ed.) “Molecular Biology of RNA Tumor Viruses.” Academic Press, New York, 1980. 11. M. Green and G. F. Gerard, Progr. Nucleic Acid Res. Mol. Biol. 14, 187 (1974). 12. M. G. Sarngadharan, H. S . Allaudeen, and R. C. Gallo, Merhods Cancer Res. 12, 3-47 (1976). 13. G. F. Gerard and D. P. Grandgenett, in ”Molecular Biology of RNA Tumor Viruses (J. R. Stephenson, ed.), pp. 345” Academic Press, New York, 1980. 14. J. Tooze (ed.). “Molecular Biology ofTumor Viruses.” Cold Spring Harbor Laboratory, Cold Spring Harbor, New York, 1973.

6. REVERSE TRANSCRIPTASE

89

Triton X- 100 (20-22). The detergent-lysed virions manifest all three major activities associated with purified reverse transcriptase. The solubilized lysate is generally fractionated by one or a combination of ion-exchange chromatography, affinity chromatography, velocity sedimentation, and gel filtration. The details of these procedures have been described elsewhere (3, 13). Most of the studies have been carried out on reverse transcriptase obtained from avian retroviruses [avian myeloblastosis virus ( A M V ) ] , Rous sarcoma virus, and murine leukemia viruses. In our experience, the following protocol yields homogeneous preparations of reverse transcriptase relatively fast and in fairly high yields. Purified virions are lysed with nonionic detergent followed by centrifugation at 15,000g for 30 min. The supernatant is adsorbed onto a poly(C)agarose (23) or heparin-Sepharose column (24) and the enzyme eluted with a salt gradient. The enzymatic activity is monitored by utilizing poly(C) . oligo(dG)12-18 as template-primer and radiolabeled dGTP as substrate. The peak fractions of enzymatic activity are pooled and chromatographed on a phosphocellulose column. Material eluting from the phosphocellulose column is highly pure; however, it does contain traces of DNase activity. If the enzyme is to be used for synthesis of cDNA, the peak fraction of enzymatic activity obtained from the phosphocellulose column is further purified by velocity centrifugation on a glycerol gradient (17). The yield of the enzymatic activity from the starting material is about 50-60%. The enzyme should be stored at either -70” in 15% glycerol or -20” in 50% glycerol. B. ENZYMATIC ACTIVITIES OF VIRIONS

Purified reverse transcriptase exhibits both synthetic (DNA polymerase) and degradative (RNase H) activities. These and related enzyme activities found in purified virions are as listed below. 15. R. C. Nowinski, N . H. Sarkar, and E. Fleissner, Merhods Cancer Res. 8,237 (1973). 16. D. L. Kacian, K. F. Watson, A . Burny, and S. Spiegelman, EEA 246,365-383 (1971). 17. I. M . Verma and D. Baltimore, “Methods in Enzymology” Vol. 29, pp. 125-131, 1973. 18. D. P. Grandgenett, G. F. Gerrard, and M. Green, PNAS 70, 230-234 (1973). 19. A. J. Faras, J. M. Taylor, J. P. McDonnell, W. E. Levinson, and J . M. Bishop, Eiochetnisfry 11, 2334-2342 (1972). 20. J. Hurwitz and J. P. Leis, J . Virol. 9, 116-129 (1972). 21. L. H. Wang and P. H. Duesberg,J. Virol. 12, 1512-1521 (1973). 22. J. W. AbreU and R. C. Gallo,J. Virol. 12, 431 (1973). 23. S. L. Marcus, M. J. Modak, and L. F. Cavalieri, J . Virol. 14, 853 (1974). 24. M. Golomb and D. P. Grandgenett, JBC 254, 1606-1613 (1979).

90

INDER M. VERMA

1. DNA Polymerase

Reverse transcriptase can utilize both polyribonucleotides and polydeoxyribonucleotides as templates to direct the synthesis of complementary polydeoxyribonucleotides (25). The two activities are virtually inseparable and apppear to contain a common active site. The enzyme can efficiently transcribe both the homopolymers and heteropolymers. Like other known DNA polymerases, reverse transcriptase also requires a preformed primer to initiate DNA synthesis (26). Although both ribo- and deoxyribo-oligomers can serve as primers, the deoxyribo-oligomers are considerably more efficient primers (3). The primer for initiation of viral DNA synthesis is transfer RNA (27-29), which presumably provides the 3’-OH end to form a phosphodiester bond with the substrate. The direction of synthesis is from 5‘ to 3’ (2.5).The reverse transcriptase appears to be a zinc metalloenzyme (30, 31) and its divalent ion requirements have been tabulated elsewhere (3). Unlike the bacterial DNA polymerases, but like the eukaryotic DNA polyrnerases, reverse transcriptase lacks the 3‘-5’ exonuclease activity, which has been postulated to guarantee a high degree of fidelity during transcription (32). Not surprisingly, reverse transcriptase has been reported to catalyze the incorporation of an exceptionally large number of incorrectly paired bases (33) when homopolymeric templates are used. However, it is not known if the relatively high degree of mistakes is also introduced if heteropolymeric templates are used. 2 . Ribonuclease H The RNase H activity associated with purified reverse transcriptase specifically degrades the RNA moiety of an RNA-DNA hybrid (34). The degradation is not dependent on the concurrent synthesis of complemen25. D. Smoler, I. Molineux, and D. Baltimore, JBC 246, 7697 (1971). 26. I. M. Verma, N. L. Meuth, E. Bromfeld, K. F. Manly, and D. Baltimore, Narure New Biol. 233, 131-134 (1971). 27. R. C . Sawyer, F. Harada, and J. E. Dahlberg, J. Virol. 13, 1302-131 1 (1974). 28. A. J. Faras, J. E. Dahlberg, R. C. Sawyer, F. Harada, J. M. Taylor, W. E. Levinson, J. M. Bishop, and H. M. Goodman,./. Virol. 13, 1134 (1974). 29. G. Peters, F. Harada, J. E. Dahlberg, A. Panet, W. Haseltine, and D. Baltim0re.J. Virol. 21, 1031-1041 (1977). 30. B. J. Poiesz, N. Battula, and L. A. Loeb, BBRC 56, 959 (1974). 31. D. S . Auld, M. Kawaguchi, D. M. Livingston, and B. L. Vallee BBRC 62, 296 (1975). 32. A. Kornberg, “DNA Synthesis.” Freeman, San Francisco, 1974. 33. N. Battula and L. A. Loeb, JBC 251, 982 (1976). 34. K. Molling, D. P. Bolognesi, W. B. Auer, W. Biisen, H. W. Plassmann, and P. Hausen, Nature New Biol. 234, 240-243 (1971).

6. REVERSE TRANSCRIPTASE

91

tary DNA (cDNA) since preformed hybrids are susceptible to RNase H activity (35),which acts as an exoribonuclease and requires free ends (36). In contrast, the cellular RNase H acts as an endoribonuclease (37). Ribonuclease H associated with purified reverse transcriptase from either AMV or MLV cleaves at the 3’ end of the 3’-5‘phosphodiester bond to yield products containing 5’-phosphate and 3’-OH ends (35, 36, 38). The products generated during RNase H activity are 4 to 20 nucleotides long. No mononucleotides are detected in the reaction product. RNase H associated with AMV reverse transcriptase is a processive exonuclease (38, 39), while the RNase H activity associated with the purified MLV reverse transcriptase has been variously reported to act in a random (38) or processive manner (40). Although RNase H and DNA polymerase activities reside on the same polypeptide ( / 8 ) , they appear to have different functional sites. For instance, (a) DNA polymerase activity is more heat-labile than the corresponding RNase H activity (36, 4/); (b) RNase H activity is selectively inhibited by NaF (42); (c) digestion of reverse transcriptase with chymotrypsin leads to inactivation of DNA polymerase activity 8 to 10 times faster than the corresponding RNase H activity (43);(d) DNA polymerase activity purified from AMV is inactivated eight times faster than the corresponding RNase H activity by N-ethylmaleimide (44);and (e) differential sensitivity to low pH (441, sodium pyrophosphate (45), pyridoxal phosphate (46), etc. 3. DNA Endonuclease A DNA endonuclease has been reported in the purified virions and purified AMV reverse transcriptase. The virion-associated DNA endonuclease has a molecular weight of 32,000 and is referred to a ~ 3 ( 5 )2. It ~ has been purified from both AMV and ASV core structures and appears to 35. D. Baltimore and D. Smoler, JBC 247, 7282-7287 (1972). 36. J. Leis, I. Berkower, and J . Hurwitz, In “DNA Synthesis in Vitro” (R.D. Wells and R. B. Inman, eds.), pp. 287-308. University Park Press, Baltimore, Maryland, 1973. 37. W. Keller and R . Crouch. P N A S 69, 3360-3364 (1972). 38. 1. M. Verma,J. Virol. 15, 843-854 (1975). 39. D. P. Grandgenett and M. Green, JBC 249, 5148-5152 (1974). 40. K. Molling, J . Virol. 18, 418 (1976). 41. 1. M. Verma, W. S. Mason, S. D. Drost, and D. Baltimore, Nature (London) 251, 27-31 (1974). 42. L. Brewer and R . D. Wells, J . Virol. 14, 1494-1502 (1974). 43. M. T. Lai and I. M. Verma, J . Virol. 25, 652-663 (1978). 44. M . Gorecki and A. Panet, Biochemisfry 17, 2438-2442 (1978). 45. J. C. Meyers and S. Spiegelman, P N A S 75, 5329-5333 (1978). 46. M. J. Modak, BERC 71, 180-187 (1976).

~

92

INDER M. VERMA

be structurally related to the large, but not the small, subunit of AMV reverse transcriptase (47). Both the virion-associated ~32”’’endonuclease and the reverse transcriptase-associated DNA endonuclease can nick supercoiled DNA in the presence of Mn2+without any site specificity (47). However, ~ 3 2 ” endonuclease ” in the presence of Mg2+,nicks supercoiled E. coli ColEl plasmid DNA preferentially near the EcoRI cleavage site (13, 48). The purified reverse transcriptase can be treated with chymotrypsin to generate a 32,000 dalton fragment that exhibits DNA endonuclease activity (49). The DNA endonuclease activity appears to be more heat-stable than either the DNA polymerase or RNase H activities. No similar DNA endonuclease activity has been found to be associated with mammalian retrovirus purified reverse transcriptase, but an endonuclease activity has been demonstrated in MLV virions (13). Studies with temperature-sensitive mutants with lesions in reverse transcriptase may provide direct evidence of the viral origin of the DNA endonuclease activity. 4. DNA Swivelase

DNA swivelase removes superhelical turns from form I DNA, converting it to a series of relaxed covalently closed structures. These relaxed structures with various topological winding numbers can be separated by agarose gel electrophoresis. In the purified virions of ASV, DNA swivelase activity has been observed after treatment with nonionic detergent. Similarly, Collett el al. (6) have reported the presence of “unwinding-like’’ activity associated with purified AMV reverse transcriptase. However, it has not been shown unequivocally that the DNA swivelase activity is viral-coded protein. Several other enzymatic activities have been variously reported to be associated with purified retroviruses (8);however, this chapter is confined to viral-coded proteins only.

C. PHYSICAL PROPERTIES The physical structure of purified reverse transcriptase from avian, feline, and murine retroviruses has been studied extensively. Most of the structural work has been done on the purified reverse transcriptase obtained from AMV, ASV, M-MLV, and R-MLV (3, 13).Table I summarizes the physical and biochemical properties of purified AMV and M-MLV 47. R . D. S c h E and D. P. Grandgenett, . I Virol. . 28, 279-291 (1978). 48. R. D. Schiff and D. P. Grandgenett, .I.Virol. 36, 889-893 (1980). 49. D. P. Grandgenett, M. Golomb, and A . C. Vora, J . Virol 33, 264-271 (1980).

93

6. REVERSE TRANSCRIPTASE TABLE I

COMPARISON OF PROPERTIES OF M-MULV REVERSETRANSCRIFTASE, ISOLATED a A N D ap FORMS OF AMV REVERSETRANSCRIFTASE Properties

MuLV

a

ffp

Molecular weight Number of subunits rIl2 of DNA polymerase (min, at 45") f l , * of DNA polymerase with template (min at 45") Binding affinity of primer tRNA (liter/mole) Mode of action of RNAase H

84,000 1 6.5 6.5

65,000 1 7.5 7.5

170,000 2 7.0 15.5

<3 x 107

0.6 x 107

3-4 x 10-7

Random or processive

Random

Processive

reverse transcriptase. Briefly, avian reverse transcriptase has an average molecular weight of 170,000 and consists of two subunits of unequal size (16, 17). The large subunit p and the small subunit a are structurally related ( 5 0 , 5 / ) .The small subunit (Y has been isolated and manifests all the enzymatic activities associated with the holoenzyme cyp (18, 52). The enzymatically active form of the p subunit has also been demonstrated (53). In contrast, the purified murine retroviruses contain a single polypeptide of an average molecular weight of 80,000(38, 40). It manifests both the synthetic and the degradative activities. The murine enzyme appears to be very sensitive to proteolysis (40). It is possible that during or prior to purification the murine reverse transcriptase is rapidly degraded. Large precursors of reverse transcriptase ranging in size from 100,000 to 130,000 daltons can be identified in the MLV-infected cells (54, 55). Most likely, the viral reverse transcriptase is processed intracellularly and the 80,000 dalton form of the enzyme is packaged in the virions. The molecular size and number of subunits present in reverse transcriptase obtained from several other retorviruses has been reported. The simian-sarcoma-virus (SSV)-associated reverse transcriptase has a single subunit with a molecular weight of 70,000 (56), and that of gibbon ape 50. 51. 52. 53. 54. 55. 56.

W.Gibson and I . M . Verma, P N A S 71, 4991-4994 (1974). H. M. Rho, D. P. Grandgenett, and M. Green, JBC 250, 5278-5280 (1975). K . Molling, C S H S Q B 39, 969-973 (1975). A. Hizi and W. K . Joklik, JBC 252, 2281-2289 (1977). M . Rokutanda, Y. Maeda, and W. Watanabe, BBRC 80, 729-734 (1978). 0. N . Witte and D. Baltimore, J . Virof. 26, 750-761 (1978). J . W. Abrell, M. S. Reitz, and R. C. Gallo,J. Virol. 16, 1566-1574 (1975).

94

INDER M. VERMA

leukemia virus (GaLV) has a molecular weight of 68,000 (57).The purified reverse transcriptase from hamster leukemia virus (HaLV) is unusual among mammalian retroviruses reverse transcriptases in that it appears to have two subunits with apparent molecular weights of 68,000 and 53,000 that sediment as a complex with an S value corresponding to a molecular weight of 120,000 (58).The exact size of the purified reverse transcriptase obtained from mouse mammary tumor virus (M-MTV), a type B retrovirus, is disputed, but it appears likely to be a single polypeptide with a molecular weight of 100,000 (59). Another group of avian retroviruses, reticuloendotheliosis virus (REV), unrelated to the avian leukemia or sarcoma virus, contains a reverse transcriptase with a molecular weight of 70,000 (60) to 84,000 (61 ). Finally, purified reverse transcriptase from Mason-Nzer monkey virus (MPMV), squirrel monkey retrovirus and Po-1-LV (Langur monkey), and examples of type D retroviruses have an average molecular weight of 80,000 to 100,000 and contain a single polypeptide chain (62-64). 111.

Biosynthesis of the Reverse Transcriptase

The synthesis of reverse transcriptase has been extensively reviewed elsewhere (13, 65). Briefly, the initial translational product of the pol gene is Pr180 product, which can be immunoprecipitated by anti-gag or anti-polymerase antisera (65). Furthermore, immature virus particles newly released from cells contain Pr180 QnQ--pol rather than the mature 80,000 dalton MuLV reverse transcriptase (55). These immature virus particles manifest no polymerase activity, but upon incubation the Pr180QQQ--Po~ product is cleaved and enzymatic activity is observed (55). Cleavage of R180 gaQ--pOl to the mature 80,000 dalton reverse transcriptase occurs through two pd-containing intermediates of 135,000-145,000 dal57. P. S. Sarin and R. C. Gallo, BBA 454, 212-221 (1976). 58. I. M. Verma, N. L. Meuth, H. Fan, and D. Baltimore,J. Virol. 13, 1075-1082 (1974). 59. A. S. Dion, C. J. Williams, and D. H. Moore, J . Virol. 22, 187-193 (1977). 60. S. Mizutani and H. M. Temin, J . Virol. 16, 797-806 (1975). 61. K. Molling, G. Gelderblom, G. Pauli, R. R. Friis, and H. Bauer, Virology 65,546-557 ( 1975). 62. S . L. Marcus, N. H. Sarkar, and M. J. Modak, BBA 519, 317-330 (1978). 63. D. Colcher, R. L. Haberling, S. S . Kalter, and J . Schlom, J . Virol. 23, 294-301 (1977). 64. G. J. Todaro, R. E. Benueniste, C. J. Shem, J. Schlom, G. Schidlovsky, and J. R. Stephenson, Virology 84, 189-194 (1978). 65. R. B. Arlinghaus, G. A. Jmjoom, J. Kopchick, and R. B. Naso,fn "Cell Differentiation and Neoplasia" (G. F. Saunders, ed.), pp. 271-925. Raven, New York, 1978.

6. REVERSE TRANSCRIPTASE

95

tons and 145,000-150,000 daltons (65). These cleavages result in the removal of the gug sequences from the Pr180g"g-"0' product, but little is known about the exact mechanism of the cleavage.

IV.

Mechanism of Reverse Transcription

The mechanism of reverse transcription is quite complex and little is known about the precise details of the intermediary structures. However, enough experimental evidence has emerged to give broad outlines of the mechanism by which viral genomic RNA is transcribed into doublestranded DNA. Figure 1A shows the structure of viral RNA and of the integrated viral DNA; Fig. 1B displays the various steps involved in the process of reverse transcription of single-stranded viral genomic RNA into double-stranded viral DNA. In any model of reverse transcription, the following salient features should be reconciled: (1) The integrated, and some unintegrated, forms of viral DNA are 500 to 600 nucleotides larger than the genomic RNA (66-73). (2) The termini of the double-stranded viral DNA are redundant, and are referred to as long-terminal repeat (LTR) (66 ). The seven steps shown in Fig. 1B take these crucial features into account. Moloney MLV (M-MLV) DNA is used in the figure, but the general outline should be the same for other retroviral DNAs. This model is very similar to that described by Gilboa et al. (74), and depends on that model for experimental verification. Srep 1. Viral DNA synthesis initiates at the 3'-OH end of the tRNA primer located near the 5' end of the genomic RNA (75). In the case of 66. T. W. Hsu, J . L. Sabran, G. E. Mark, R. V. Guntaka, and J . M. Taylor, J . Virol. 28, 810-818 (1978).

67. J. L. Sabran, T. W. Hsu, C. Yeater, A. Kqji, W. S. Mason, and J. M. Taylor,J. Virol. 29, 170-178 (1979). 68. P. R. Shank, S. H. Hughes, H . 4 . Kung, J. E. Majors, N . Quintrell, R. V. Guntaka, J. M. Bishop, and H. E. Varmus, Cell 15, 1383-1395 (1978). 69. S. H. Hughes, P. R. Shank, D. H. Spector, H.-J. Kung, J. M. Bishop, H. E. Varmus, P. K . Vogt, and M. L. Brietman, Cell 15, 1397-1410 (1978). 70. E. Gilboa, S. Goff, A. Shields, F. Yoshimur, S. Mitra, and D. Baltimore, Cell 16, 863-874 (1979). 71. E. W. Benz Jr. and D. Dina, P N A S 76, 3294-3298 (1979). 72. G . F. Vande Woude, M. Oskarsson, L. W. Enquist, S. Nomura, S. Sullivan, and P. J. Fischinger, P N A S 76, 4464-4468 (1979). 73. R. A. Bosselman and I. M. Verma, J . Virol. 33, 487-493 (1980). 74. E. Gilboa, S. W. Mitra, S. Goff, and D. Baltimore, Cell 18, 93-100 (1979). 75. J. M. Taylor and R. Illmensee, J . Virol. 16, 553 (1975).

96

INDER M. VERMA t RNAPro

POI

env

viral RNA

u3

7 Reverse Transcription 7 PO I

e nv WcI

V V “3 u5

(A)

u3 u5

INTEGRATED viral DNA

FIG. 1. Proposed mechanism of reverse transcription. (A) Diagrammatic sketch of viral genomic RNA and integrated viral DNA. ( 0 )U, (also referred to as strong-stop DNA); (B) U,; (0)genomic terminal redundancy. U5 + Us sequences constitute a long-terminal repeat (LTR). C is the primer-binding site. (B) Model for reverse transcription. The genomic terminal redundancy is referred to as T strong-stop DNA as Us;and sequences from 3‘ end of the genomic RNA as Us. The transcript of 19 nucleotides of primer tRNA is referred to as C’; whereas the reverse transcript of primer binding site on genomic RNA is C. The complementary strands are referred to as T’, U; and U; .

M-MLV, 18 nucleotides at the 3’ end of the primer tRNApro are hydrogen-bonded to the genomic RNA (referred to as C in the figure) (76). The 3’-A-OH of the primer tRNA is not hydrogen-bonded and is located 146 nucleotides from the 5’-cap nucleotide of the genomic RNA (77, 78). It forms a phosphodiester bond with the first deoxynucleotide triphosphate (dATP in the case of M-MLV) and the synthesis of the complementary DNA proceeds until it reaches the 5’ nucleotide of the genomic RNA. Step 2. The 145 nucleotide long cDNA [“strong stop” DNA (79), referred to as U5 in the figure] covalently linked to the primer tRNAPro dissociates and hybridizes to terminally redundant sequences (region T in the figure) at the 3‘ end of the genomic RNA. The synthesis of DNA then proceeds from the 3‘ end of the RNA toward its 5‘ end. Steps 3 and 4 . A 600-bp-long DNA fragment of opposite polarity (+) 76. F. G. Harada, G. Peters, and J. E. Dahlberg, JBC 254, 10979-10985 (1979). 77. C. Van Beveren, J. G. Goddard, A. Berns, and I. M. Verma, PNAS 77, 3307-3311 (1980). 78. R. Dhar, W. L. McClements, L. W. Enquist, and G. F. Vande Woude, PNAS 77, 3937-3941 (1980). 79. W. A. Haseltine and D. Baltimore, In “Animal Virology” (D. Baltimore, A. S. Huang, and C. F. Fox, eds.), ICN-UCLA Symp. Mol. Cellular Biol., Vol. 4, pp. 175-213. Academic Press, New York, 1976.

98

INDER M. VERMA

strand, referred to as U3 in the figure, can be observed after the synthesis of 0.5-1.0 kb of the cDNA transcript [(-) strand]. The (+) strand DNA spans the U3,U, and T regions. In addition, it transcribes the C region of the tRNAPr". Several laboratories have observed the presence of an approximately 600 bp (+) strand during the reverse transcription in vivo and in vitro (80-82). Gilboaer al. (74) and Taylor and Hsu (83)have shown that the C region of the primer is reverse transcribed. The (-1 strand transcripts continue to elongate and transcribe the primer tRNA binding site (C region of the genomic RNA). Steps 5-7. The C' transcribed from tRNA primer (present on the (+) strand can base pair with the C region transcribed from genomic RNA (present on the (-1 strand) (73). The (-) strand then continues to transcribe the (+) strand and the (+) strand elongates by using the (-) strand as the template, leading to the formation of a double-stranded DNA molecule shown at the end of Step 7. The double-stranded DNA has long terminal repeats and is 517 (in the case of M-MLV) nucleotides longer than the genomic RNA (77, 84). If the reverse transcription is carried out in vitro in the presence of the drug actinomycin D, which inhibits the synthesis of the (+) strand of DNA, the double-stranded DNA synthesized appears to have no longterminal repeat (81, 85). It is not clear if the DNA synthesized in the presence of actinomycin D transcribes the tRNA primer-binding (C) region. In the cells infected with retroviruses, molecules containing linear and circular forms of double-stranded DNA can be identified. The circular forms of DNA that either contain or lack LTR can also be demonstrated (86). Several basic assumptions outlined in Fig. 1B remain elusive. For instance, how does the "strong stop DNA' dissociate and continue synthesis at the 3' end of the genomic RNA? What is the signal for the synthesis of the (+) strand of DNA? An unusual structure containing long tracts of purines and pyrimidines has been determined near the 5' end of the U, region. Finally, it is not known what form of viral DNA synthesized in vivo is integrated in the host-cell DNA. 80. H. E. Varmus, S . Heasley, H.-J. Kung, H. Opperman, V. C. Smith, J. M. Bishop, and P. R. Shank, J M B 120, 55-82 (1978). 81. I. M. Verma, J . Virol. 26, 615-629 (1978). 82. S. W. Mitra, S. Goff, E. Gilboa, and D. Baltimore, PNAS 76, 4355-4359 (1979). 83. J. M. Taylor and T. W. Hsu, J . Virol. 33, 531-534 (1980). 84. J. G. Sutclitfe, T. M . Shinnick, I . M . Verma, and R . A. Lerner, P N A S 77, 3302-3306 (1980). 85. E. Rothenberg and D. Baltimore, J . Virol. 21, 168-178 (1977). 86. F. Yoshimura and R. A. Weinberg, CeIl 16, 323-332 (1979).

6. REVERSE TRANSCRIPTASE

V.

99

Applications to Molecular Biology

The current revolution in molecular biology owes much to the discovery of reverse transcriptase. Prior to 1970 it would perhaps have been heretical to propose the existence of an enzyme that utilizes RNA as template to make DNA. It took Khorana and colleagues over a decade to chemically synthesize the 77 nucleotide-long phenylalanine tRNA, whereas it takes less than 10 minutes to synthesize a globin gene (-450 bp) by using reverse transcriptase and purified globin mRNA. Reverse transcriptase is now commonly used to synthesize DNA complementary to a variety of RNA templates. Some of the applications of reverse transcriptase in molecular biology are briefly reviewed in the following sections. A.

SYNTHESIS OF COMPLEMENTARY DNA

In 1972 three research groups independently reported the synthesis of DNA complementary to 10 S rabbit globin mRNA by using reverse transcriptase (87-89). They took advantage of the fact that mRNA has a stretch of poly(A) at its 3' end and hence an oligomer of (dT) can be used as a primer to initiate the synthesis of complementary DNA. Since then, a number of mRNAs [for instance, ovalbumin (YO), silk-fibroin ( Y I ) , insulin (921, and hormones (9.?)] have been used as templates to synthesize cDNA. In most cases, the mRNA contains a poly(A) stretch, and oligo(dT)12-le primer can be used. If the RNA of interest does not contain poly(A) sequences, as in the case of several negative-stranded viruses like influenza and VSV, poly(A) stretches can be added to the 3'-OH end of the RNA by poly(A) polymerase (W, Y5). In certain cases, like ribosomal RNA, a stretch of oligo(dG) can act as a primer (Verma and Weinberg, 87. I . M. Verma, G . F. Temple, H. Fan, and D. Baltimore, N a f u r e N e w B i d . 235, 163 (1972). 88. J . Ross, H. Aviv, E. Scolnick, and P. Leder, P N A S 69, 264-268 (1972). 89. D. L. Kacian, S. Spiegelman, A. Bank, M . Terada, S. Metafora, L. Dow, andP. A. Marks, Nature N e w Biol. 235, 167-169 (1972). 90. G . N . Buell, M. P. Wickens, P. Farhang, and R. T. Schimke, JBC 253, 2471 (1978). 91. P. M. Lizardi and D. D. Brown, CSHSQB 38, 701 (1973). 92. L. Villakomaroff, A. Efstratiadis, S. Browne, P. Lomedico, R. Tizard, S. P. Naber, W. L. Chick, and W. Gilbert, P N A S 75, 3727 (1978). 93. A. Ulrich, J . Shine, J. Chirgwin, R. Pictet, E. Tischer, W. J. Rutter, and H. M. Goodman, Science 196, 1313 (1977). 94. A. Hell, B. D. Young, and G . D. Birnie, BBA 442, 37-49 (1976). 95. J . S. Emtage, G . H. Gatlin, and N . H. Carey, Nucleic Acids Res. 6, 1221-1240 (1979).

100

INDER M. VERMA

unpublished results). The cDNA transcripts can be useful in many ways: (i) cDNA transcripts can be converted to double-stranded DNA and molecularly cloned in either bacteriophages or plasmids. (ii) cDNA transcripts can be used as probes to screen recombinant clones to identify the chromosomal DNA fragment that contains the entire gene. (iii) cDNA transcripts can be used as primers to make large transcripts. In many cases, it is not feasible to either identify or isolate relatively large quantities of purified mRNA for a given protein. However, if any amino acid sequence data are available, a primer corresponding to 3 or 4 amino acids can be annealed to the mRNA and extended with reverse transcriptase. The extended DNA fragment can then be used as a probe to screen for recombinants, or cloned in plasmids itself. (iv) The cDNA transcripts can be immobilized on either a Sepharose column or nitrocellulose paper and used to purify mRNA species. They can also be used to arrest translation of specific mRNA species in an irt vitro cell-free protein synthesizing system. (v) The cDNA transcripts can be used to perform S1 nuclease analysis (Fan and Verma, unpublished results). (vi) The cDNA transcripts can be labeled either at the 5’ end with polynucleotide kinase or at the 3’ prime end with terminal transferase and then subjected to nucleotide sequence analysis (77, 96-08),

B. SYNTHESIS OF DOUBLE-STRANDED DNA Reverse transcriptase can utilize single-stranded DNA or RNA-DNA hybrid as template to synthesize double-stranded DNA. If an RNA is used as a template in a reaction that contains reverse transcriptase, doublestranded DNA transcripts can be obtained (81, 99). The most likely mechanism of synthesis of the second strand of DNA is that the cDNA transcript forms a “hair-pin” by folding back on itself and the 3’-OH end then serves as a primer. This mechanism was first postulated when it was found that in using globin mRNA as a template (-450 nucleotides) DNA transcripts of an average size of 800-900 nucleotides could be identified on alkaline sucrose gradients. The “hair-pin’’ loop formation is now used as a standard methodology to synthesize double-stranded DNA for molecu96. W. A. Haseltine, A. Maxam, and W. Gilbert, PNAS 74, 989-983 (1977). 97. J. Shine, A. P. Czernilofsky, R. Friedrich, J . M. Bishop, and H . M. Goodman, PNAS 74, 14773-1477 (1977). 98. E. Stoll, M. A. Billeter, A. Palmenberg, and C. Weissman, Cell 12, 57-72 (1977). 99. I. M. Verma, G . F. Temple, H. Fan, and D. Baltimore, in “Symposia on Control of Transcription” (B. B. Biswas et d.,eds.), pp. 355-372. Calcutta, India, 1973.

6. REVERSE TRANSCRIPTASE

10 I

lar cloning purposes (100, 101). The double-stranded DNA formed by the hair-pin’’ mechanism is cleaved with single-strand specific nuclease S1 and then inserted in appropriate vector either by “tailing method” (102) or via “synthetic linkers” (103). Alternatively, the single-stranded cDNA transcript can be converted to double-stranded form by E. cofi DNA polymerase I (100). However, both ways of formation of doublestranded DNA have a basic inadequacy in that nucleotides involved in the formation of the “hair-pin’’ are removed by Sl digestion. Thus, the resulting double-stranded DNA is lacking sequences that represent the 5’ end of the RNA template. The 5‘ end of the RNA very often contains the ribosome-binding site and initiation codon AUG. Two ways can be used to overcome this problem: (i) Addition of a stretch of, for instance, oligo(dA) at the 3‘ end of the cDNA transcript by using terminal transferase, and isolate the (dA),-tailed cDNA transcript by binding to an oligo(dT1cellulose column. The (dA),-tailed cDNA transcript can then be converted to its double-stranded form by using oligo(dT)lz-18as primer (81) (it is somewhat surprising that this approach has not been extensively employed by other investigators; perhaps, the commercially available terminal transferase is not free of DNases). (ii) The cDNA transcript can be converted to a double-stranded form by using DNase-digested calfthymus DNA as primers (81, 104). The double-stranded DNA can then be fractionated on sucrose gradients or agarose gels to isolate various size molecules. This method is very efficient and can convert the entire cDNA transcripts to double-stranded form. The last two methods have the added advantage that they do not require S1 nuclease digestion prior to molecular cloning. “

C. ENDLABELING The reverse transcriptase, unlike the bacterial DNA polymerases, lacks the 3‘-5‘ and 5‘-3’ exonucleases and can thus be efficiently used for “end labeling” or “gap filling.” Two examples are given: (i) “Gap jlling.” A DNA fragment can be cleaved with either restriction endonucleases or exonuclease 111 digestion (Fig. 2A and B). The “gap filled” DNA can then 100. A. Efstratiadis and L. Villa-Komaroff, in “Genetic Engineering, Principles and Methods (A. Hollander and J . K . Setlow, eds.), Vol. 1. Plenum, New York, 1979. 101. I. R . Rose, J . Virol. 32, 404-41 I (1979). 102. A. C. Y. Chang, J. H. Nonberg, R. J . Kaufman, H. A. Erlich, R. T. Schimke, and S. N. Cohen, Nature (London) 275, 617-621 (1978). 103. K . Itakura and A. D. Riggs, Science 209, 1401 (1980). 104. J. M. Taylor, R. Illmensee, and J. Summer, BBA 442, 324-330 (1976).

102

INDER M. VERMA

5kAGCT - T

- ‘5

AOH~’ TTCGA,?‘

-~C0;3

3’

5‘

-‘3

Exo m treoted DNA

Hind IU cleaved frogment + 4 dNTP‘s Reverse Tronscriptose

+ 4 dNTP’s

+ Reverse Tronscriptase

+

5p ‘ AGCTT-

t

****

- ‘ 5

AAGCTOH~‘

t

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

........

3‘

3‘ 5’

(B)

FIG.2. “End labeling” by reverse transcription, (A) AHindIII cleaved DNA fragment is used as a template to fill the 5‘ overhangs. (B) An exonuclease 111-digested DNA fragment is used to fill in the single-stranded regions. be used for blunt-end ligation, addition of “synthetic linkers” or for “end labeling.” (ii) Selective “end labeling.” For instance, if a DNA fragment is cleaved by restriction endonucleases XbaI and BglII, then both ends can be selectively labeled as shown in Fig. 3. This is a very useful method for labeling and sequencing as it does not require strand separation or further cleavage of the labeled DNA fragment. This kind of labeling can also be used to make selectively labeled probes. However, “end labeling” by reverse transcriptase or DNA polymerase I cannot be used when restriction endonuclease-cleaved products have either blunt ends like Sma I or PvuII, or 3‘ overhangs like PstI, KpnI, or SacI.

Xba

I

Bgl II

I

I

TCTAGA-AG AGATCT-

b ‘CTAGA-

’

3~~~-

ATCT TCTAGA

I

t

+XbaI

+ BgI It ~ 0 ~ TCTAGp5‘

+ *dCTP “pCTAGA-C “t

~

’

-

+*dGTP ~ 0 “ ~ ’

TCTAG psi

5’pCTAGA 3‘~~~------

AG*

TCTAGp5’

FIG. 3. Selective “end labeling.” A DNA fragment cleaved with restriction endonuclease XbaI and BglII is used to demonstrate selective “end labeling.” In the presence of labeled dGTP as the only substrate, the XbaI cleaved end will be labeled, whereas if the labeled substrate is dGTP, only the &/I1 site will be labeled (*) Indicates labeled substrate.

6. REVERSE TRANSCRIPTASE

103

D. OTHERUTILITIES The RNase H activity associated with reverse transcriptase can be used to cleave RNA moiety from an RNA-DNA hybrid. Unlike E. coli or calf thymus RNase H, reverse transcriptase associated RNase H is an exonuclease and thus can be used to remove the ends of RNA from an RNA-DNA hybrid. Reverse transcriptase can also be used for “nick-translations” to yield high specific activity DNA. Finally, reverse transcriptase can be employed to induce site selective mutagenesis (10.5). Since it lacks the 3‘-5’ and 5’-3‘ exonuclease activities, reverse transcriptase is more error prone. At high substrate concentrations, and in the presence of different divalent cations like Mn2+instead of Mg2+,reverse transcriptase can introduce mismatched deoxynucleotide triphosphate. Exploitation of the use of reverse transcriptases for site-directed mutagenesis is potentially very promising.

ACKNOWLEDGMENTS I thank J. Rose for critically reading portions of the manuscript, Robert Bosselman for help in formulating the reverse transcription model (Fig. 1B) and Maureen Brennan for preparing the manuscript. The work was supported by research grants from the National Cancer Institute.

105. T. Kunkel, R . R . Meyer, and L. A . Loeb, PNAS 76, 6331-6335 (1979).

This Page Intentionally Left Blank

Termina1 Deoxynucleotidyltransferase ROBERT L. RATLIFF

I. Introduction . . . . . . . . . . . 11. Purification and Properties. . . . . 111. The Reactions Catalyzed . . . . . A. Deoxynucleoside Triphosphates B. Initiators . . . . . . . . . . . C. Metal Ions . . . . . . . . . . D. Buffers. . . . . . . . . . . . E. Kinetics . . . . . . . . . . . F. Statistics of Polymerization . . G. The Mechanism of the Reaction H. Assay Methods . . . . . . . . IV. Practical Applications . . . . . . . V. Biological Role . . . . . . . . . .

1.

. . ,

. . .

. . .

. .

.

. .

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

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

. . . . . .

. . .

.

. . .

.

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

105 107 109 109 110 111 111 111 112 113

114 114 118

introduction

During early attempts to isolate DNA polymerase from mammalian cells it was noted that a major activity recoverable from calf thymus glands was not due to a polymerase that was able to copy a template, but to a polymerase that formed DNA polymers whose composition was determined by the type and composition of deoxynucleoside triphosphates 105 THE ENZYMES, Vol. XIV Copyright 0 1981 by Academic Press, Inc. All rights of reproduction in any form reserved ISBN 0-12-122714-6

106

ROBERT L. RATLIFF

provided in the reaction mixture (1 -10). This enzyme was named terminal deoxynucleotidyltrunsferase (terminal transferase) to distinguish it from the replicative deoxynucleotidyltransferase (DNA polymerase). [For a previous review of terminal transferase see Bollum (I I ).I Among normal tissues terminal transferase is generally found only in mammalian bone marrow (12-15) and thymus (16-22). The presence of the enzyme in the bone marrow of congenitally athymic mice (23, 2 4 ) at levels equal to that found in bone marrow of normal mice suggests that lymphoid progenitor cells containing the enzyme can mature in bone marrow cells independent of thymic influence. Very high levels of terminal transferase activity have been detected in lymphoblastic leukemia (19, 25-32), in leukemic cells from patients with chronic myelogenous leukemia in blastic 1. J. S . Krakow, C. Coutsogeorgopoulos, and E. S. Cannellakis, BBRC 5, 477 (1961). 2. J. S . Krakow, H. 0. Kamen, and E. S . Cannellakis, BBA 53, 52 (1961). 3. J. S. Krakow, C. Coutsogeorgopoulos, and E. S . Cannellakis, BBA 55, 639 (1962). 4. F. J. Bollum, JBC 235, PC 18 (1960). 5. F. J . Bollum, JBC 237, 1945 (1962). 6. H. M. Keir, J. B. Sheperd, and J. Hay, BJ 89, 9P (1963). 7. H. M. Keir and M. J. Smith, BBA 68, 589 (1963). 8. F. J. Bollurn, E. Groeniger, and M. Yoneda, PNAS 51, 853 (1964). 9. M. Yoneda and F. J. Bollum, JBC 240, 3385 (1%5). 10. M. E. Gottesman and E. S. Cannellakis, JBC 241, 4339 (1966). 11. F. J. Bollum, “The Enzymes,” 3rd ed., Vol. 10, p. 145, 1974. 12. F. J. Bollum, PNAS 72, 4119 (1975). 13. N. H. Pazmino, R. N. McEwan, and J. N. Ihle, J . Immunol. 119, 494 (1977). 14. A. E. Silverstone, H. Cantor, G. Goldstein, and D. Baltimore,J. Exp. Med. 144, 543 ( 1976). 15. N. H. Pazmino, J. N. Ihle, and A. L. Goldstein, J . Exp. Med. 147, 708 (1978). 16. L. M. S . Chang, BBRC 44, 124 (1971). 17. M. S . Coleman, J. J. Hutton, and F. J. Bollum, BBRC 58, 1104 (1974). 18. P. C. Kung, A. E . Silverstone, R. P. McCaffrey, and D. Baltimore, J. Exp. Med. 141, 855 (1975). 19. R. P. McCafFrey, T. A. Harrison, R. Parkman, and D. Baltimore, New Engl. J. Med. 292,775 (1975). 20. R. Barton, J. Goldschneider, and F. J. Bollum, J . Immunol. 116, 462 (1976). 21. N. H. Pazmino and J. M. Ihle, J. Immunol. 117, 620 (1976). 22. L. M. S. Chang and F. J. Bollum, JBC 246, 909 (1971). 23. D. Baltimore, A. E. Silverstone, P. C. Kung, T. A. Harrison, and R. P. McCaffrey, CSHSQB 41, 63 (1977). 24. J. J. Hutton and F. J. Bollum, Nucleic Acids Res. 4, 457 (1977). 25. M. S. Coleman, J. J. Hutton, P. DeSimone, and F. J. Bollum, PNAS 71,4404 (1974). 26. R. McCaBey, D. F. Smoler, and D. Baltimore, PNAS 70, 521 (1973). 27. B. I. S . Srivastava and J. Minowada, BBRC 51, 529 (1973). 28. J. R. Bhattacharyya, BBRC 62, 367 (1975). 29. M. S. Coleman, M. F. Greenwood, J. J. Hutton, F. J. Bollum, B. Lampkin, and P. Holland, Cancer Res. 36, 120 (1976).

7. TERMlNAL DEOXYNUCLEOTIDYLTRANSFERASE

107

crisis (33-37), and in lymphoid cells transformed by Abelson murine leukemic virus (38). Thus the enzyme represents a useful marker of neoplastic cells in certain human diseases (19, 29, 34). Whereas no definitive role for terminal transferase has been established, speculation that the enzyme participates in immunological programming has been proposed (39, 40). Terminal transferase has proven to be a valuable tool for the synthesis of model polydeoxynucleotides, and more recently for the construction of recombinant DNAs.

11.

Purification and Properties

Terminal transferase from calf thymus was first purified to homogeneity by Chang and Bollum (22). The enzyme has a molecular weight of 32,360, measured V of 0.65 cm3 g-l, and can be dissociated into two polypeptide chains with molecular weights of 26,500 and 8,000. The amino acid composition has been determined for the homogeneous enzyme and each of its subunits; a significant excess of acidic (66) over basic (47) amino acids was found. This finding is surprising since the enzyme has an isoelectric pH of 8.6 and behaves as a basic protein on electrophoresis and chromatography (221. The terminal transferase from calf thymus has been studied most extensively. The pH optimum is approximately 7.2 and maximum activity is demonstrated in cacodylate buffer. The enzyme has a turnover number of about 50 moles of Mg2+:dATP min-’ mole of enzyme-I, and about 100 30. P. C. Kung, J . C. Long, R. P. McCaffrey, R. L. Ratliff, T. A. Harrison, and D. Baltimore, Amer. J. Med. 64, 788 (1978). 31. M. T. Shaw, J. M. Swyer, H. S. Allaudeen, and H. A. Weitzman, Blood 51, 181 (1978). 32. P. S. Sarin and R. C. Gallo, BBRC 65, 673 (1975). 33. P. S. Sarin, and R. C. Gallo, JBC 249, 8051 (1974). 34. P. S. Sarin, P. N. Anderson, and R. C. Gallo, Blood 47, 11 (1976). 35. J . J . Hutton and M. S. Coleman, Brir. J . Haemulo/. 34, 447 (1976). 36. S. M. Marks, D. Baltimore, and R. C. McCaEery, New Engl. J . Med. 298,812 (1978). 37. J. A. Donlon, E. S. JaEe, and R. C. Broylan, New Engl. J . Med. 297, 461 (1977). 38. A. E. Silverstone, N . Rosenberg, D. Baltimore, V. L. Sato, M. P. Scheid, and E. A. Boyse, in “Differentiation of Normal and Neoplastic Hematopoitic Cells” (B. Clarkson, J. Till, and P. Marks, eds.), p. 433. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York, 1978. 39. D. Baltimore, Narure (London) 248, 409 (1976). 40. F. J. Bollum, “Karl August Forster Lectures,” Vol. 14, p. 1. Franz Steiner Verlag, Wiesbaden, 1975.

108

ROBERT L. RATLIFF

moles of Co2+: dCTP min-' mole of enzyme-'. The enzyme is stable at low temperatures and at pH 4.5, but is not stable at temperatures above 40" or to protein denaturants such as urea, sodium dodecyl sulfate (SDS) and organic solvents (22). In contrast to Bollum's enzyme, there have been several reports of high molecular weight terminal transferases. A new terminal transferase has been isolated from calf thymus tissue by Johnson and Morgan (41) that is a single polypeptide with a molecular weight of 79,000. Unlike earlier reports of the isolation of a low molecular weight enzyme from whole blood cells of patients with acute lymphocytic leukemia (33, 42), Diebel and Coleman (43) purified terminal transferase that has a molecular weight of 63,000 and consists of a single polypeptide. Using an immunoprecipitation assay, Silverstone et rrl. (44) reported that murine terminal transferase is synthesized as a single polypeptide with a molecular weight of 60,000. Bollum and Brown (45) showed that immunoprecipitation of human lymphoblastoid tumor cells with a rabbit anti-calf terminal transferase precipitates a major protein band with a molecular weight of 58,000. Johnson and Morgan (41) observed that recovery of the high or low molecular weight terminal transferase depends on the purification procedure used. The high molecular weight terminal transferase is isolated if at no stage of the purification at low pH (i.e., (7.0) is used. They also reported that the DNA polymerase, isolated during the purification of terminal transferase, upon storage develops a high level of terminal transferase activity with a concurrent lower level of DNA polymerase activity. This suggests that the terminal transferase may actually be a proteolytic degradation product of template-requiring DNA polymerase. McCaErey et al. (19) reported that normal thymocytes and leukemic cells contain two forms of terminal transferase that can be separated by phosphocellulose chromatography. The ability to see two discrete peaks of terminal transferase from extracts of cells requires the addition of a protease inhibitor. Since the amino acid sequence of the high and low molecular weight terminal transferase has not been determined, no homologous sequence of amino acids are available for comparison to determine if the low molecular weight terminal transferase is a degraded product of the high molecular weight enzyme. 41. 42. 43. 44. 45.

D. Johnson and A. R. Morgan, BBRC 72, 840 (1976). F. A. Siddiqui and B. I. S. Srivastava, BEA 517, 150 (1978). M. R. Diebel and M. S. Coleman, JBC 254, 8634 (1979). A. Silverstone, L. Sun, 0. N. Witte, and D. Baltimore, JBC 255, 791 (1980). F. J. Bollum and M. Brown, Nature (London) 278, 191 (1979).

7. TERMINAL DEOXYNUCLEOTIDYLTRANSFERASE

111.

109

The Reactions Catalyzed

The reaction catalyzed by terminal transferase may be written ndNTP

M A' + d(pX), E - d(pX),d(pN). + nPP,

Initially an oligodeoxynucleotide initiator, d(pX),, is bound by the enzyme (E); this is followed by repetitive grafting of mononucleotide units from a deoxynucleoside triphosphate to the terminal 3'-hydroxyl of the initiator and the release of inorganic pyrophosphate. The reaction is irreversible, since no pyrophosphate exchange is demonstrated (46). A.

DEOXYNUCLEOSIDE TRIPHOSPHATES

Any of the common deoxynucleoside triphosphates can be polymerized by terminal transferase although the K , and V,,, values differ for each (46). Deoxynucleoside triphosphates with substituted amino groups are substrates for the termainl transferase, but the rate of polymer synthesis is much lower and the ultimate polymer length less than with the nonsubstituted bases. An apparent exception is the N-acetyldeoxyguanylate polymer, where the average size is larger than with deoxyguanylate, presumably because N-acetylation prevents aggregation of the dG residues and termination of synthesis (47-50). It has also been reported that deoxynucleoside triphosphates substituted at positions other than the amino-N, such as 5-methyl-dCTP (51) and 60-methyl-dGTP ( X ) , can function as substrates. Although limited ribonucleotide addition as well as some copolymerization of ribo- and deoxynucleotides can occur (53-58), 46. K. Kato, J. M. Gonclaves, G . E. Houts, and F. J. Bollum, JBC 242, 2780 (1967). 47. R. Fliigel and F. J. Bollurn, BBA 308, 35 (1973). 48. C. F. Lefler and F. J. Bollurn, JBC 244, 594 (1969). 49. E. Hansbury. V. N . Ken, V. E. Mitchell, R. L. Ratla, D . A. Smith, D. L. Williams, and F. N. Hayes, BBA 199, 322 (1970). 50. F. N. Hayes, E. Hansbury, V. E. Mitchell, R. L. Ratla, and D. L. Williams, ELIB 6, 485 (1968). 51. B . Zrnudzka, F. J. BoUum, and D. Shugar, J B M 46, 169 (1969). 52. P. J. Abbott, J. R. Mehta, and D. B. Ludlum, Biochemistry 19, 643 (1980). 53. H. Kossel and R. Roychoudhury, EJB 22, 271 (1971). 54. R. Roychoudhury and H. Kossel, EJB 22, 310 (1971). 55. R. Roychoudhury, R. Fischer, and H. Kossel, BBRC 45, 430 (1971). 56. G. Feix, FEBS (Fed. Eur. Biochem. Soc.)Left.18, 280 (1971). 57. G. Feix, BBRC 46, 2141 (1972). 58. R. Roychoudhury, JBC 247, 3910 (1972).

110

ROBERT L. RATLIFF

arabinonucleotides are not polymerized (49). Dideoxythymidine triphosphate (59) and cordycepin triphosphate produce chain-terminating addition. Terminal transferase could be used to synthesize oligodeoxynucleotides with specified sequences if substrates with easily removable 3’blocking groups were available. Unfortunately such substrates are not currently known, the 3’-O-acetyldeoxyadenosinetriphosphate was found not to function as a substrate in the terminal transferase reaction (48). B. INITIATORS At 35” the enzyme has an absolute requirement for an oligodeoxynucleotide containing at least three phosphate groups, d(pXpXpX), and a free 3’-hydroxyl (461, although d(pXpX) has been shown to be incorporated into a polymer if the reaction is done at 15 instead of 35” (60). Studies with radioactively labeled initiators have shown that the oligodeoxynucleotide is incorporated into the product and is present at the 5‘ terminus (60). The kinetic analysis of polymerization of dATP using a variety of oligodeoxynucleotides of different base composition and length as initiators have been reported (46, 60). The data indicate that as the initiator chain length is increased to 5-7 nucleotides the polymerization rate also increases. Longer chains interact well with the enzyme, but synthesis rates decrease. Conditions that enhance initiator incorporation are lower temperature, high buffer concentration, and high enzyme specific activity (60). For polymerization of dATP the order of preference for initiator is dA 3 dT > dC (46, 60). Terminal transferase can initiate polymerization at both 3’-hydroxyl groups of an oligodeoxynucleotide with an internal pyrophosphate linkage (61). Substitution on the S’-phosphate of an oligodeoxynucleotide does not seem to prevent its functioning as an initiator (62). Although oligoribonucleotides are not useful initiators in the terminal transferase reaction, an oligoribonucleotide terminated with one or two deoxynucleotides at the 3’-hydroxyl end is an efficient primer for the polymerization of deoxynucleotides (57). From the evidence presented it has been postulated that the competency of the initiator to be bound by the terminal transferase is an important factor in determining the extent of initiator incorporation into polymer. More than one configuration of the initiator-enzyme complex 59. N. R. Cozzarelli, R. B. Kelly, and A. Kornberg, J M B 45, 513 (1969). 60. F. N. Hayes, V. E. Mitchell, R. L. Ratliff, A. W. Schwartz, and D. L. Williams, Biochemistry 5, 3625 (1966). 61. A. W. Schwartz and F. N. Hayes, BBA 138, 604 (1967). 62. D. G. Ott, V. N. Kerr, E. Hansbury, F. N. Hayes, Anal. Biuchem. 21, 469 (1967).

7. TERMINAL DEOXYNUCLEOTIDYLTRANSFERASE

111

may be possible, but only one is competent to initiate. The enzyme conformation for proper binding at the 3’-hydroxyl is specific for the deoxyribose configuration, and for initiator molecules to interact with the enzyme the 3’-hydroxyl has to be freely available and not complexed with another polydeoxynucleotide chain. C.

METALIONS

Several divalent metals can substitute for magnesium in activating the enzymatic polymerization of deoxynucleotides. In certain instances more favorable polymerization rates are obtained with metals other than magnesium. For example, for the extension of chains with dCTP or dTTP low concentrations of cobalt are best (46). Human terminal transferase from leukemic cells show maximum activity when manganese is used as the divalent cation (19, 32, 33, 63, 64). This result suggests that different conformational changes in the substrate and/or in the enzyme are induced by different metals.

D. BUFFERS In addition to divalent metal ions, rates of polymerization of deoxynucleoside triphosphate and initiator utilization are affected by specific buffer ions and ionic strength. In 40 mM potassium phosphate buffer, dATP polymerization begins immediately. Increasing the phosphate buffer concentration above 40 mM decreases polymerization, whereas increasing cacodylate to 200 mM doubles the polymerization rate measured at 40 mM cacodylate. Addition of 250 m M NaCl to 40 mM cacodylate decreases the initial rate, but complete polymerization of the deoxynucleoside triphosphate occurs (46). Other buffers such as Tris-HC1 have been used for the assay of terminal transferase, particularly in leukemic cell extracts; but a much higher enzyme activity can be observed using the cacodylate buffer system (43, 64).

E. KINETICS The V,,, and K , have been reported for the polymerization of the four deoxynucleoside triphosphates and the kinetic constants for dATP polymerization using a series of oligodeoxynucleotide initiators differing in base composition and chain length (46). The highest polymerization rate 63. M. S. Coleman, ABB 182, 525 (1977). 64. P. C. Kung, P. D. Gottlieb, and D. Baltimore, JBC 251, 2399 (1976).

112

ROBERT L. RATLIFF

for purine deoxynucleoside triphosphates is obtained when cacodylate buffer and Mg2+are used, whereas the use of Co2+instead of Mg2+ increases the polymerization when the bases are pyrimidines. The kinetic constant derived from studies with the various initiators shows that the affinity of the terminal transferase for initiator is in the order of I > G > A > T > C (46). A maximum in V,,, for dATP polymerization with oligodeoxynucleotides of different chain length is observed when the length of the initiator is five to seven nucleotides. Larger polydeoxynucleotides can be used as initiators but the measured rate of polymerization of deoxynucleoside triphosphates is less favorab ,e probably because at a given total nucleotide concentration, as initiator chain length is increased the relative concentration of 3-hydroxyl termini is decreased. The copolymerization of two or more deoxynucleoside triphosphates shows that generally the substrate composition grossly influences the course of incorporation of each. Copolymerization of the two purine deoxynucleotides is exceptional in that the final product has a composition that agrees with the stoichiometric ratio of the triphosphates in the reaction mixture. The incorporation rate of a purine deoxynucleoside triphosphate initially exceeds that of the pyrimidine when both are present at equal concentration. Moreover the ultimate chain length of the product is short and pyrimidine deoxynucleotide incorporation is suppressed. In all cases of copolymerization of deoxyguanosine triphosphate with one or more of the other deoxynucleoside triphosphates the initial rate of incorporation of deoxyguanosine residues exceeds that of the other deoxynucleotides (65). The likely explanation seems to be the superiority of dGTP as a substrate for terminal transferase. The nucleotide sequences of copolymers synthesized by terminal transferase have been described as random (46, 66), but nearest-neighbor analysis of some of the copolymers gives values differing appreciably from those predicted for a random copolymer (67). Although the mechanism of heteropolymer synthesis catalyzed by terminal transferase leads to no preferred sequences, differences in length and frequency of repeating pyrimidine nucleotide segments are observed (65). F. STATISTICS OF POLYMERIZATION The stoichiometry of the polymerization has been studied by Kato et al. (46).For each mole of deoxynucleoside triphosphate incorporated into the 65. D. E. Hoard, R. L. Ratla, D. L. Williams, and F. N. Hayes, JBC 244, 5368 (1969). 66. R. L. Ratliff, D. E. Hoard, D. G . Ott, and F. N. Hayes,Biochemistry 6, 851 (1967). 67. R. L. Ratliff, A. W. Schwartz, V. N. Kerr, D. L. Williams, D. G. Ott, and F. N. Hayes, Biochemistry 7, 412 (1968).

7. TERMINAL DEOXYNUCLEOTIDYLTRANSFERASE

113

polymer one mole of pyrophosphate is formed. The equilibrium constant for the forward reaction has been estimated to be K,,, = 99 for dATP polymerization on an oligonucleotide initiator. If the terminal transferase reaction is a condensation in which a bifunctional monomer polymerizes on itself forming linear chains, it should give rise to a Poisson distribution of product lengths. If rn is the mean number of added nucleotide units that equals the average mole ratio of deoxynucleoside triphosphate reacting with initiator molecules, andx is the number of units added to an individual chain, the mole fraction of polymer molecules with x units added should fit the Poisson expression crnmx/x (68). Hayes er al. (69) demonstrated that a Poisson distribution is found when dATP is polymerized onto d(pT)B. However in some cases, such as dGTP polymerization (491, self-limiting effects such as product aggregation can give a much sharper distribution of product lengths. The product distribution can also be affected if incomplete utilization of initiator occurs. Chang and Bollum found that in the homopolymerization of d(pA), and dATP the product distribution varied considerably from the Poisson. With a longer initiator, such as d(pA),, a Poisson distribution is observed in the final product (70). G. THE MECHANISM OF THE REACTION Efforts to evaluate quantitatively the kinetics of the terminal transferase-catalyzed reaction are limited by uncertainties regarding the nature of the interactions between the enzyme and the required divalent metal cation. Experiments by Chang and Bollum (71) on the inhibition of terminal transferase by metal ligands and length of initiator required for synthesis suggest that at least two sites are involved in binding initiator molecules to terminal transferase. In their model the enzyme-metal ion complex binds to the 3’-hydroxyl and the phosphoryl group of the third nucleotide from the 3’ terminus. After binding and the addition of another nucleotide, the enzyme-bound metal dissociates from the initiator and again binds to the 3’-hydroxyl and phosphoryl group of the third nucleotide. This mechanism is supported by the fact that the polymerization of deoxynucleoside triphosphates catalyzed by the terminal transferase is strongly inhibited by metal chelators. An analysis of the kinetics of 68. P. J. Flory, “Principles of Polymer Chemistry.” Cornell Univ. Press, Ithaca, New York, 1953. 69. F. N. Hayes, V. E. Mitchell, R. L. Ratliff, and D. L. Williams,Biochemistry 6, 2488 (1967). 70. L. M. S. Chang and F. J. Bollum, Biochemistry 10, 536 (1971). 71. L. M. S. Chang and F. J. Bollum, PNAS 65, 1041 (1970).

114

ROBERT L. RATLIFF

o-phenanthroline inhibition demonstrates that the ligand is noncompetitive with the deoxynucleoside triphosphate substrate and is strictly competitive with oligonucleotide initiator (71).This model does not explain the improved pyrimidine triphosphate polymerization in the presence of cobalt, nor does it account for uninitiated synthesis (46). The improved addition of deoxynucleotides to the recessed ends of restriction endonuclease-treated SV40 DNA in the presence of cobalt could mean that the terminal transferase-cobalt complex binds more tightly to the 3’-hydroxyl, or that it makes the 3’-hydroxyl more accessible to the enzyme for the transfer of deoxynucleotidyl residues (72).

H. ASSAYMETHODS Purified terminal transferase can be assayed using a labeled deoxynucleoside triphosphate and an oligodeoxynucleotide initiator with a chain length of three to six nucleotides (9, 22). In crude extracts of cells that contain terminal transferase the low level of enzyme activity produces only several nucleotide additions, and this product may not be acidinsoluble. The use of an acid-insoluble oligodeoxynucleotide or polymer as initiator for assaying low levels of terminal transferase activity in cell extracts insures that each addition is made onto an acid-insoluble product (19, 33, 63). Antibodies to calf thymus terminal transferase have been developed in rabbits and mice, and have been found to cross-react with human and murine terminal transferase. These antibodies have been used to develop immunofluorescence and immunoprecipitation techniques for the detection of terminal transferase in whole cells and cell extracts (23, 30, 44, 45, 64, 73, 74).

IV.

Practical Applications

Terminal transferase has been used to synthesize a variety of model poiydeoxynucleotides, and to modify the ends of plasmids and genes preparatory to their union into recombinant molecules that can be replicated by a bacterium. The remainder of this section summarizes some of these applications. 72. R. Roychoudhury, E. Jay, and R. Wu, Nucleic Acids Res. 3, 863 (1976). 73. I. Goldschneider, K. E . Gregoire, R. W. Barton, and F. J. Bollurn, PNAS 74, 734 (1977). 74. K. E. Gregoire, I. Goldschneider, R. W. Barton, and F. J. Bollum, PNAS 74, 3993 (1977).

7. TERMINAL DEOXYNUCLEOTIDYLTRANSFERASE

I15

An oligomer of deoxythymidylate in 5'-ester linkage to cellulose particles was used as a primer for synthesis of block copolymer linked to cellulose. This type of copolymer has been used to study the mechanism of DNA polymerase and to assay for DNA ligase (75-77). In studies of the role of DNA structure in genetic regulations, Wells and his co-workers ( 7 8 4 4 ) synthesized a number of duplex block polymers, such as d(GoAlo)* d(T,,G,,); the synthesis utilized a series of enzymes including pancreatic DNase, terminal transferase, and T4 DNA polymerase. Thermal denaturation studies on these DNAs in the presence of actinomycin or netropsin, which bind respectively to GC or to AT base pairs, have confirmed that the properties of one region of a DNA can be influenced (telestabilized) by a remote region. The high-resolution proton NMR spectra of the synthetic DNA block polymers yielded results that have implication regarding the function of the AT- and GC-rich blocks found in DNA regulatory regions (83). A three-section block copolymer of deoxythymidylate, deoxyguanylate, and deoxyadenylate was synthesized by first incubating dGTP with a primer, d(pT6), and isolating the product, d(pT6Gl,). This polymer was then used as initiator for polymerization of dATP to yield d(pT6G10A213) (85). Sectional block polydeoxynucleotides, such as d(C44A169)and d(T66C,5),have been prepared for studying the mechanism of replication by the DNA polymerase from calf thymus (86) .

Polymers with distinctively labeled termini are readily prepared with terminal transferase. Radioactive oligodeoxynucleotide initiators have been extented with cold (or differently labeled) monomers to form specific substrates to test for 5' -+ 3' or 3' + 5' degradation. Thomas and Olivera (87) have studied a variety of deoxynucleases with regard to the pro75. T. M. Jovin, and A. Komberg, JBC 243, 250 (1968). 76. N . R. Cozzarelli, N. E. Melechen, T. M. Jovin, and A. Kornberg, BBRC 28, 578 (1967). 77. A. Panet and H. G. Khorana, JBC 249, 5213 (1974). 78. J. F. Burd and R. D. Wells, JBC 249, 7094 (1974). 79. J. F. Burd. J. E. Larson, and R. D. Wells, JBC 250, 6002 (1975). 80. R. M. Wartell, J. E. Larson, and R. D. Wells, JBC 250, 2698 (1975). 81. J. F. Burd, R. M. Wartell, J. B. Dodgson, and R. D. Wells,JBC 250, 5109 (1975). 82. R. M. Wartell and J . F. Burd, Biopolyrners 15, 1461 (1976). 83. T. A. Early, D. R. Kearns, J. F. Burd, J. E. Larson, andR. D. Wells,Biochemisrry 16, 541 (1977). 84. J . B. Dodgson and R. D. Wells, Biochemistry 16, 2367 (1977). 85. R . L. Ratliff and F. N . Hayes, BBA 134, 203 (1966). 86. F. N. Hayes, E. Hansbury, V. E. Mitchell, R. L. Ratliff, D. A. Smith, and D. L. Williams, JBC 246, 3631 (1971). 87. K. R. Thomas and B. M. Olivera, JBC 253, 424 (1978).

116

ROBERT L. RATLIFF

cessivity of their hydrolysis mechanism using double-labeled poly(dT) as substrate. The assay using double-labeled poly(dT) distinguishes between a nonprocessive enzyme, which dissociates after every hydrolytic event, and a processive enzyme, which does not dissociate from a DNA chain until complete hydrolysis has occurred. Only Escherichiu coli exonuclease I (3’ + 5 ’ ) and A-exonuclease (5’ + 3’) are processive enzymes. Rougeon et al. (88) used terminal transferase to add poly(dT) to the 3‘ end of cDNA, and oligo(dA) primer was used to prime the second-strand synthesis with DNA polymerase I. Rabbitts (89) used a different approach for copying mRNA. Terminal transferase was used to add poly(dT) tails to a linear plasmid, and the tailed plasmid was used to prime the copying of globin mRNA into cDNA by reverse transcriptase. Terminal transferase has also been used for the 3‘ end labeling of DNA for sequence analysis (72, 90,9 1 ) . The terminal transferase procedure for joining DNA segments and forming a closed loop of the product normally involves a series of steps. Restriction endonucleases are first employed to ( 1) cut a segment of duplex DNA that contains the promotor sequence and possibly one or more genetic markers from a bacterial plasmid, and (2) isolate a segment of duplex DNA that contains a desired piece of genetic information from another source; e.g., a cDNA copy of messenger RNA made with reverse transcriptase. The potential of the 3’ ends of the strands to function as initiators for the terminal transferase is enhanced by subjecting both duplex DNAs to limited digestion by A-exonuclease, which removes short sequences of nucleotides from the 5’-phosphoryl termini of the complementary strands. Terminal transferase is then employed to extend each strand at its 3‘ end with a short homooligonucleotide sequence; oligo(dA) and oligo(dT) sequences are added to plasmid DNA and cDNA respectively, so that when mixed the strands of different origin will be joined by hydrogen bonds formed between the oligonucleotide segments. The junctions are covalently closed by incubation with the enzymes, substrates, and cofactors needed for closure (92, 93). Different reaction conditions are required for addition of polypurine and polypyrimidine nucleotide segments to restriction endonuclease-treated closed circular DNA or 88. F. Rougeon, P. Kourilsky, and B. Mach, Nucleic Acids Res. 12, 2365 (1975). 89. T. H. Rabbitts, Nurure (London) 260, 221 (1976). 90. A. M. Maxam and W. Gilbert, PNAS 74, 560 (1977). 91. J. C. Chang, G.F. Temple, R. Poon, K. H. Neumann, and Y. W. Kan, PNAS 74,5145 (1977). 92. D. A. Jackson, R. H. Symons, and P. Berg, PNAS 69, 2904 (1972). 93. P. E. Lobban and A. 0. Kaiser, J M B 78, 453 (1973).

7 . TERMINAL DEOXYNUCLEOTIDYLTRANSFERASE

117

double-stranded cDNA. Polypurine nucleotide addition occurs better in the presence of Mg2+at low ionic strength, whereas polypyrimidine nucleotide addition occurs more rapidly in the presence of Co2+at higher ionic strength (92-98). Since the amounts of terminal transferase are in considerable molar excess over primer termini, the difference in optimal conditions could represent a lower affinity of the enzyme for poly(dA) in the presence of Mg2+ than for poly(dT) in the presence of Co2+. Roychoudury et al. (72) found that if cobalt ion replaced magnesium ion, the addition of homopolymer extensions to the 3’-hydroxyl termini using terminal transferase was possible without prior treatment of the DNA with A-exonuclease. Although different metal ions and buffers affect the ability of terminal transferase to add polypurine or polypyrimidine nucleotide segments to the 3‘ ends of DNA molecules, the limiting component in the reaction mixture is the concentration of 3’-hydroxyl termini. We have observed that if the reaction mixture volume is reduced to 50 pl when the amount of 3‘ ends is 1 pmole or less, the rate of addition as well as the total length of the homopolymer extension added to doublestranded cDNA and linear plasmid DNA is increased (J. K. Griffith and R. A. Walters, unpublished). The lengths of the homopolymeric extensions added to each of the 3’-hydroxyl end of a linear plasmid using terminal transferase and C d + are apparently equal. This has been confirmed by the addition of [ 3H]-poly(dT)tails to Barn HI-treated pBR322, and subsequent cleavage of the tailed plasmid with EcoRI. Analysis of the two fragments by gel electrophoresis revealed that 50% of the tritium label was present with each (J. K. Griffith and R. A. Walters, unpublished). An important feature of the poly(dA . dT) “connector” method is that a molecule of a linear vector with a poly(dT) extension can only be annealed with a tailed fragment of cDNA-d(A), while other intermolecular or intramolecular annealing is prevented. The yield of hybrid product is greatly increased under these conditions. This aspect is particularly important when attempting to select a specific molecule from a pool of several hundred different hybrids. An additional advantage of the method is that enzymatic ligation of the annealed-hybrid circular DNA is unnecessary; unsealed DNA preparations are infectious and are converted to covalently closed circular DNA molecules in vivo (97). 94. D. Brutlag, K. Fry, T. Nelson, and P. Hung, Cell 10, 509 (1977). 95. R. Higuchi, G. V. Paddock, R. Wall, and W. Salser, PNAS 73, 3146 (1976). 96. T. Maniatis, S. G . Kee, A. Efshatiadis, and F. C. Kafatos, Cell 8, 163 (1976). 97. L. Clark and J. Carbon, PNAS 72, 4361 (1975). 98. P. C. Wensink, D. J. Finnegan, J. E. Donnelson, and D. S. Hogness, Cell 3, 315 ( 1974).

118 V.

ROBERT L. RATLIFF

Biological Role

Terminal transferase attracted little attention from biologists until its unique tissue distribution was appreciated. Under normal conditions it is present only in thymus and bone marrow (12-2 1); circulating lymphoid cells, lymph node cells, and normal spleen cells contain no detectable enzyme activity (17, 19, 26). Terminal transferase is found at highest concentration in cortical thymocytes, and almost all of the terminal transferase-containing cells in the thymus and bone marrow are eliminated in mice and rats by cortisone treatment (17, 18, 20, 23,44). Regardless of which ontogenetic pathway cortical thymocytes may follow, it is clear that terminal transferase synthesis ceases before or immediately upon the attainment of immunologic competence by T cells, no measurable concentration of the enzyme is found in peripheral T cells. While no definitive role for terminal transferase has been established, speculation that in the early stages of lymphocyte differentiation the enzyme participates as a somatic mutator in the generation of immunological diversity has been advanced (39, 40). The model proposes that an endonuclease makes single-strand scissions in the variable-region gene in immunocytes that are in the process of commitment to synthesis of a specific immunoglobulin. The nicks are then elongated to different length gaps by an exonuclease, and the terminal transferase inserts a single, random nucleotide that does not form a base pair with the nucleotide on the opposite strand, and is therefore a mutated site. The gap is subsequently filled by a DNA polymerase and sealed by a ligase (39). It has been demonstrated that d(G * T) weak base pairing can occur (99). Therefore, if deoxyguanosine is substituted for deoxyadenosine, or thymidine for deoxycytidine, in the variable-region gene of an immunoglobulin chain by terminal transferase the mismatched base pairing between d(G * T) cannot be corrected by repair nuclease. Thus the hypothesis is that terminal transferase acts as a somatic mutator, diversifying the amino acid sequence in the variable region of immunoglobulin molecules by changing one of the nucleotides (39).

ACKNOWLEDGMENTS The author is indebted to Drs. A, G. Saponara, D. F. Hoard, R. A. Walters, and I. K. Griffith for their help and criticism in preparing this manuscript. This work was performed under the auspices of the U.S.Department of Energy,

99. D. M. Gray, and R. L. Ratliff, Biopolyrners 16, 1331 (1977).

Section II

DNA Nucleases and Related Enzymes

This Page Intentionally Left Blank

Deoxyribonucleases: Survey and Perspectives STUART LINN

I . Introduction . . . . . . . . . . . . . . . . . A . Historical . . . . . . . . . . . . . . . . . B. Classification of Reactions Catalyzed . . . . . I1. Occurrence. Purification. and Molecular Properties . . . . . . . . III. Specificity . . . . . . . . . . . . . . . . . . . . . . . . . . A . Duplex Versus Single-Stranded DNA . . . . . . . . . . . . B. Sugar Specificity . . . . . . . . . . . . . . . . . . . . . . C. Base and Sequence Specificity . . . . . . . . . . . . . . . . D . Specificity for Modified Nucleotides . . . . . . . . . . . . . E . Other DNA-Dependent Enzyme Activities . . . . . . . . . . IV. Assays: Designing Proper Substrates and Detection Procedures . . . V. Biological Role . . . . . . . . . . . . . . . . . . . . . . . . A . Function Determination . . . . . . . . . . . . . . . . . . . B . Possible Functions . . . . . . . . . . . . . . . . . . . . . VI . Control of Activities . . . . . . . . . . . . . . . . . . . . . . VII . Research Applications . . . . . . . . . . . . . . . . . . . . . A Analytical Applications . . . . . . . . . . . . . . . . . . . B . Preparative Applications . . . . . . . . . . . . . . . . . .

.

122 122 122 124 124 124 125 125 125 127 129 130 130 131 133 134 134 135

121 T H E ENZYMES. Vol . XIV Copyright 0 1981 by Academic Press. Inc . All rights of reproduction in any form reserved ISBN 0-12-122714-6

122 1.

STUART LINN

Introduction

A. HISTORICAL DNase activity was probably first reported in 1903 by Araki (I) who showed that extracts from animal tissues (thymus, intestine, pancreas, liver, or spleen) could liquify a gel of calf thymus DNA. Shortly therafter, Plenge (2) demonstrated similar activities in extracts from microorganisms, including a large cross section of bacterial forms and yeast. In 1913, de la Blanchardiere (3) reported a semiquantitative viscometric assay for liquefication of a thymus DNA gel, and used it to show that such liquefication was not accompanied by destruction of the DNA bases. From then until 1950 most progress was made through studies of bovine pancreatic DNase [see the review by Laskowski, Ref. (4)],which was the first deoxyribonuclease that was purified free of ribonuclease activity. It was also the first deoxyribonuclease to be crystallized, and indeed, upon crystallization Kunitz, in 1950, introduced the term deoxyribonuclease (5). Even by this time, mainly through simultaneous studies of spleen acid DNase [see the review by Bernardi, Ref. (611 it was clear that various types of DNase activities existed in nature. B. CLASSIFICATION OF REACTIONS CATALYZED By definition, deoxyribonucleases catalyze hydrolysis of the phosphodiester bond of polydeoxyribonucleotides. However, there are several subclasses noted among these enzymes. 1. Exonucleases Versus Endonucleases

Exonucfeases are best defined as enzymes that require a DNA terminus to act; i.e., they cannot degrade covalently closed DNA circles. In addition, exonucleases degrade DNA from the termini toward the center. Conversely, endonucleases are defined as enzymes that are capable of hydrolyzing covalently closed, circular substrates. An older operational definition states that exonucleases produce mononucleotides, whereas endonucleases produce oligonucleotides; however, this definition is inap1. 2. 3. 4. 5. 6.

T. Araki, Z . Physiol. Chem. 38, 84 (1903). H. Plenge, 2. Physiol. Chem. 39, 190 (1903). P. de la Blanchardiere, Z . Physiol. Chem. 87, 291 (1913). M. Laskowski, Sr. “The Enzymes,” 3rd ed., Vol. 4, p. 289, 1971. M. Kunitz, J . Gen. Physiol. 33, 349 (1950). G. Bernardi, “The Enzymes,” 3rd ed., Vol. 4, p. 271, 1971.

8. DEOXYRIBONUCLEASES

123

propriate because enzymes are known [e.g., the recBC enzyme, Ref. (711 that do not act upon circular DNA, but produce oligonucleotides that proceed from the terminus. Though most DNases are either exo- or endonucleases, some enzymes appear to exhibit both types of activity. 2 . Polarity of Exonucleases Exonucleases can hydrolyze DNA specifically in the 5’ + 3‘ or the 3‘ + 5’ direction. Most exonucleases have specificity for one direction or the other; if such an enzyme is specific for duplex DNA, only one-half of the DNA is hydrolyzed (on the average). However some enzymes, notably the recBC-like enzymes (7), have both specificities. 3. Processive Versus Dispersive Action Exonucleases may act in a processive manner, hydrolyzing one DNA substrate molecule to completion before commencing hydrolysis of another molecule. Conversely, if the enzyme disassociates from partially degraded substrate molecules, then reassociates more or less at random with any available substrate molecule, it is said to be acting dispersively. These alternatives are usually distinguished operationally by analysis of reaction intermediates under conditions of substrate excess. Processive enzyme reactions exhibit some intact and some very degraded molecules, whereas dispersive reactions exhibit a population of molecules, all of which are partially hydrolyzed. Endonucleases may also tend to have a processive or dispersive nature, but careful examination of this point remains to be done. 4. Nicks Versus Breaks by Endonucleases

When acting upon double-stranded DNA, an endonuclease may cleave both strands of the duplex at the same locus to produce a break, or it may hydrolyze only one strand to produce a nick. With circular substrates, the break results in a linear product, whereas the nick results in an open circular “Form 11” product. Some enzymes (e.g., E. coli endonuclease I) produce mainly breaks; some (e.g., pancreatic DNase) produce nicks as isolatable precursors to breaks; and some appear to produce both nicks and breaks. The selection of nicks versus breaks is often dependent upon reaction conditions.

5 . Nature of Phosphomonoester Termini Some DNases form 3’-phosphomonoester termini, whereas others form 5’-phosphomonoester termini. No enzyme is known that appears to form both. 7. K. M. T. Muskavitch and S. Linn, this volume, Chap. 13.

124

STUART LINN

II. Occurrence, Purification and Molecular Properties

The DNases are ubiquitous in nature, apparently occurring in all bacterial, plant, and animal cells. In addition, they are found associated with the more complex DNA-containing animal viruses, and are coded for or induced during infection by many simpler viruses. In E. coli K12 and B, where DNases have been most intensively studied, at least 14 DNases have been extensively characterized and others undoubtedly exist. DNases are isolated and purified using classical protein fractionation procedures. The various enzymes studied to date are too numerous to list here, but a compilation does exist (8). In a few instances apparently homogeneous and even crystalline preparations are obtainable, but in many instances, because of the small amount of enzyme per cell, one must utilize preparations that contain contaminating activities as well as extraneous inert material. This difficulty might eventually be circumvented in some cases by increasing the content per cell through DNA cloning techniques, although cells probably will not tolerate substantially increased levels of some DNases. Purified DNases vary greatly in size and complexity, ranging from single polypeptides of low molecular weight (- 15,000) to complex, multimeric structures of molecular weights greater than 500,000. In addition, eukaryotic enzymes often appear to exist in sets of isozymes with similar properties, or as proteolytic derivatives of one another, which appear to vary greatly in their catalytic abilities.

111.

Specificity

A.

DUPLEXVERSUSSINGLE-STRANDED DNA

Many DNases act without regard to secondary structure, but others when purified are quite specific for either single-stranded or ordered substrates. For example,E. coli exonuclease I is specific for a single-stranded DNA, whereas E. coli exonuclease 111 is specific for duplex DNA (9). Conversely, N. crassa endonuclease (10) and S1 endonuclease ( / I ) are specific for unordered DNA structures, whereas many restriction en8. P. Kowalski and M. Laskowski, Sr. I n “Handbook of Biochemistry and Molecular Biology, Vol. 11: Nucleic Acids” ( G . D. Fasman, ed.), 3rd ed., p. 491. CRC Press, Cleveland, 1976. 9. I. R. Lehman, “The Enzymes,” 3rd ed., Vol. 4, p. 251, 1971. 10. S. Linn and I. R. Lehman, JBC 240, 1294 (1965). 11. R. C. Weigand, G . N. Godson, and C. M. Radding, JBC 250, 8848 (1975).

8. DEOXYRIBONUCLEASES

125

donucleases are specific for duplex DNA (12). Using the single-strandspecific endonucleases, the covalently closed Form I circles can often be converted to Form I1 molecules, probably through the introduction of a single nick at a looped out region in the supertwisted substrate molecule (1.3). Exonucleases specific for single-stranded DNA might have a role in repair processes, trimming off non-base-paired regions that were previously recognized and incised by repair-specific endonucleases (14).

B. SUGARSPECIFICITY Although many DNases do not act on polyribonucleotides, many others are able to degrade RNA and/or RNA-DNA hybrids. In most instances, it is not known whether the activities on the RNA and DNA substrates are both of biological signtficance.

c.

BASEAND SEQUENCE SPECIFICITY

One feature that distinguishes RNases from DNases as a class is base or sequence specificity. Many RNases are base-specific, acting at a particular nucleotide, or at either purine or pyrimidine nucleotides. No DNase is known, however, that will cleave specifically next to all A, G, T, or C residues in a substrate molecule. Conversely, as opposed to the RNases, many sequence-specific DNases are found, at least among the bacteria. These enzymes have been generally termed restriction endonucleases, although in many instances their restriction function has not been verified (15).

D.

SPECIFICITY FOR

MODIFIEDNUCLEOTIDES

1. Repair Endonucleases

Some endonucleases specifically recognize abnormal or damaged regions of DNA as part of a repair process. The most abundant activity of this type is probably found in the AP endonucleases, a group of enzymes that recognize the baseless, apurinic or apyrimidinic sites (16). Such sites arise spontaneously, after chemically induced base damage, or after action 12. P. Modrich, Quanr. Rev. Biophys. 12, 315 (1979). 13. K. Bartok and D. T. Denhardt, JBC 251, 530 (1976). 14. E. Friedberg, T. Bonura, E. H. Radany, and J. Love this volume, Chap. 14. IS. 9. Endlich and S. Linn, this volume, Chap. 9. 16. S. Linn, I n “DNA Repair Mechanisms” (P. E. Hanawalt and E. C. Friedberg, eds.), p. 175. Academic Press, New York, 1978.

126

STUART LINN

of specific DNA-glycosylases (enzymes that remove abnormal bases from DNA through cleavage of the N-glycosylic bond between the base and the deoxyribose). While apparently ubiquitous in nature, AP endonucleases generally fall into one of two classes depending upon whether they cleave on the 5' or 3' side of the baseless site (see Scheme I). When acting together the two classes are able specifically to remove deoxyribose 5-phosphate from DNA (17). AP endonucleases are generally specific for duplex DNA. y1

7"" 7

51...p$p$p$p... /

3' \

Class I

\

Class II AP endonuclease CHO B,

Class II A P endonuclease

A possible second group of repair endonucleases, once thought to be the predominant group, are those that specifically recognize a particular type of damaged nucleotide, and cleave a nearby phosphodiester bond without the utilization of glycosylase activity as an intermediate step. The E. coli uvrABC gene products, for example, recognize pyrimidine dimers; however, these activities recognize certain other lesions as well (18). Members of this group of repair enzymes have not generally been very wellcharacterized, partly because of the confusion that existed between the true endonuclease activity and other activities that resulted from combinations ofN-glycosylase and AP endonuclease [see Ref. ( M ) ] .Hence, it is now uncertain whether endonucleases that cleave phosphodiester bonds next to particular damaged bases do, in fact, exist. A third group of repair endonucleases appear to recognize damage by virtue of the abnormalities that the damage causes in the DNA duplex 17. D. W.Mosbaugh and S . Linn, JBC 255, 11743 (1980). 18. E. Seeberg, PNAS 75, 2569 (1978).

127

8. DEOXYRIBONUCLEASES

structure. This type of endonuclease is thus less specific than the other repair endonucleases because various types of damage are recognized by the same enzyme. Examples of this type of endonuclease are E. coli endonuclease V (19) and rat liver repair endonuclease (20).

2. Endonucleases Recognizing Methylated Bases The true restriction endonucleases recognize specific DNA sequences and act only if both strands of the sequence have not been methylated by a corresponding modification methylase that recognizes and methylates the same sequence (21 ). For example, endonuclease EcoR1 recognizes and cleaves the sequence, G-A-A-T-T-C, as follows: 5 ’ - . . .-G-A-A-T-T-C- . . .

5 ’ - . . .-G

. . .-C-T-T-A-A-G- . . .-5‘

. . .-C-T-T-A-A

+

A-A-T-T-C-. . . G-.. .-5’

The sequence is protected, however, when 6-methyladenine residues (A*) are introduced on one or both strands of the sequence by the modification methylase: 5 ‘ - . . .-G-A-A-T-T-C-. . . . . .-C-T-T-A-A-G- . . .-5‘

5 ’ - . . .-G-A-A*-T-T-C- . . . . . .-C-T-T-A*-A-G- . . .-5‘

An interesting variation is a class of enzymes from Diplococcus pneumoniae, Dpn I, that cleaves DNA only if it is methylated: 5 ’ - . . .-G-A*-T-C-. . . . . .-C-T-A*-G-.. .-5’

5 ’ - . . .-G-A* . . .-C-T

+

T-C-. . . A*-G-. . .-5’

The function of this enzyme remains mysterious, particularly since the same organism contains a second enzyme, DpnII, that acts only on unmethylated 5’-. . .-G-A-T-C-. . . residues (22).

E. OTHERDNA-DEPENDENT ENZYME ACTIVITIES In addition to degrading DNA by phosphodiester bond cleavage, many DNases have the ability to carry out other DNA-dependent reactions. The catalytic sites for these reactions are not necessarily associated with a separate subunit of a multisubunit complex, but often appear to be a more intimate part of the nuclease protein. Although the number of such enzymes is too great to catalog here, several prominent examples are noted. 19. 1980. 20. 21. 22.

B. Demple, F. T. Gates, and S. Linn, “Methods in Enzymology,” Vol. 65, p. 224, J. L. Van Lancker and T. Tomura, BBA 353, 99 (1974). H. 0. Smith, Science 205, 455 (1979). S. Lacks and B. Greenberg, J M B 114, 153 (1977).

128

STUART LINN

I. DNA Polymerase Perhaps the most notable example is the association of bacterial DNA polymerases with exonuclease activities (23). In the case of E. coli DNA polymerase I, all of the catalytic activities are associated with a single, 109,000 dalton subunit. The 3’ + 5’ exonuclease is evidently involved in editing out replication errors, whereas 5 ‘ 3’ exonuclease, which also acts as an RNase H, is probably involved with DNA excision repair and/or removal of RNA primers.

-

2 . ATPase

Many DNases contain DNA-dependent ATPase activity. In the case of the exonuclease V-type enzymes, the activity seems to be required to unwind DNA duplexes, as well as to help the enzyme track along the DNA while degrading it to oligonucleotides (7). The enzyme is believed to track along the inside of a duplex during the unwinding process. The type I restriction endonucleases, on the other hand, show an ATPase activity that seems to be involved in helping the endonuclease move in a unique direction from the recognition site to the cleavage site, while remaining bound to the recognition site at all times (15). In this case the enzyme moves on the outside of the double helix. The E. coli uvrABC endonuclease is also associated with ATPase activity, although its function is stlll not clear. Finally, DNA gyrases appear to exhibit both ATPase and endonuclease activity; the ATPase drives the directionality of the change of DNA supercoiling via double-strand breakage and resealing (24). 3 . N-Glycosylase Some, but not all, repair endonucleases appear to be combinations of a specific DNA glycosylase and an AP endonuclease. Thus, in addition to AP endonuclease activity, E. coli endonuclease 111 shows DNA glycosylase activity for 5,6-hydrated thymines, whereas T4 UV endonuclease cleaves the glycosylic bond at the 5’ side of a pyrimidine dimer pair; both enzymes can act as an AP endonuclease at the sites generated by their respective endogenous glycosylase, or they can act at AP sites generated some other way (25). The M . luteus UV endonuclease also shows both N-glycosylase and AP endonuclease activity (26). 23. T.Kornberg and A. Kornberg, “The Enzymes,” 3rd ed., Vol. 10, p. 119, 1974. 24. N. Cozzarelli, Science 207, 953 (1980). 25. B. Demple and S. Linn, Nature (London) 287, 203 (1980). 26. W. A. Haseltine, L.K . Gordon, C. P. Lindon, R. H. Grafstrom, N. L.Shaper, and L. Grossman, Nature (London) 285, 634 (1980).

8. DEOXYRIBONUCLEASES

129

4. DNA Phosphatase

Escherichia coli exonuclease I11 is remarkable. It not only shows 3’ + 5‘ exonuclease activity specific for duplex DNA, an RNase H activity, and an AP endonuclease activity (“endonuclease VI”), but also a specific DNA 3’-phosphatase activity (27). The latter specifically removes 3’phosphomonoester groups from poly- or oligodeoxyribonucleotides, rendering them active either as substrates for exonucleases that require 3’hydroxyl termini, or as DNA polymerase primers. 5 . Multiple Forms of Nucleases from a Single Precursor

Although we know relatively little about protein processing in eukaryotes, a remarkable situation is coming to light in fungi, most notably in N . crassa. A single inactive precursor protein can evidently be acted upon by proteases in several alternative manners so as to affect the sugar (RNA : DNA) specificity, the specificity for duplex versus single-stranded substrate, the exo- versus endonucleolytic mode of action, the sensitivity to ATP inhibition, and the cellular location (28). Thus, through protease action (and regulation) a remarkable number of apparently quite different enzymes with quite different cellular locations and (presumably) functions are generated from a single protein. Indeed, certain mutants defective in UV repair, mutagen sensitivity, or genetic recombination may arise via altered protease activities.

IV.

Assays: Designing Proper Substrates and Detection Procedures

A major problem that has always been associated with the study of DNases is the assay-namely, obtaining the specific substrate in substantial amounts, identifying the proper product, and quantitating that product with sufficient specificity and sensitivity. Certainly the nonspecific very active DNases can be easily detected by the degradation of a large polynucleotide to acid-soluble products, which are monitored by radioactivity or UV absorption. (The older assays utilizing viscometry or hyperchromicity have been largely abandoned.) However, after these techniques became widely applied, it was clear that more sophisticated substrates and assay systems were required to identify, purify, and characterize new types of enzymes. Even in many cases where an enzyme had 27. B. Weiss, this volume, Chap. 12. 28. T. Y.-K. Chow and M. J. Fraser, Can. J . Biochem. 57, 889 (1979).

130

STUART LINN

been identified and purified by the simpler procedures [e.g., the exonuclease V-type enzymes, Ref. (7)], it was clear that without a specific substrate in hand (or at least knowledge of its specific in vivo product) and a proper detection system, we would not be able to characterize the correct, biologically relevant reaction. A major breakthrough came with the application of circular DNAs to DNase studies. Not only could endonucleases be monitored without interference by exonucleases, but as little as one phosphodiester bond cleavage per molecule could be detected by utilizing the infectivity of the DNA, sedimentation, gel electrophoresis or filtration through nitrocellulose filters. In this way several endonucleases involved in viral DNA replication have been discovered, isolated, and characterized. More importantly, the restriction enzymes were discovered and characterized. The restriction enzymes are in turn being used to construct substrates for studying other DNases. A major problem remains with studying DNases involved in repair and recombination. In the case of repair, one is limited by the unavailability of large amounts of substrates that contain one particular type of wellcharacterized DNA damage. Thus the only truly well-characterized repair DNases are those that recognize AP sites (specifically formed by aciddepurination or uracil-glycosylase), pyrimidine dimers (formed specifically at low doses of UV radiation), or perhaps hydrated thymine residues (formed with osmium tetroxide). Other DNA damaging agents (e.g., X rays) form too many types of potential substrates, whereas the use of simple single-strand-specific assays (e.g., trimming termini) leaves the function of the enzyme in doubt. In the case of recombination, the substrate often is not known, or if speculated upon cannot be prepared in substrate amounts. Finally, if substrate is obtained, the product is rare and must be quantitated by techniques, such as electron microscopy, that become too cumbersome to use routinely. In summary, assay procedures have been, and will continue to be, the major limiting factor in studying new DNases. Efforts put into designing and improving assay procedures will continue to be worthwhile. V.

Biological Role

A. FUNCTION DETERMINATION Correlating a DNase with a biological function is no easy task and often requires several indirect methods.

8. DEOXYRIBONUCLEASES

13 1

1. Genetic Techniques

In bacteria, the most noteworthy technique is genetic: The correlation of an alteration of enzyme activity with a mutation of a known gene, a mutation that leads to a defined phenotype. Thus, exonuclease V-type enzymes are implicated in genetic recombination because strains carrying mutations in recB or recC genes lack the enzyme activity and are rec- (7). However, such correlations obviously are not absolutely definitive; other possible interpretations always exist. In addition, the mutant approach is not generally applicable to higher eukaryotes.

2 . Localization The location of an enzyme in the cell also provides a clue. Extracellular enzymes are probably involved in a degradative function designed to eliminate foreign genetic information, or simply to scavenge nutrients. Likewise, periplasmic enzymes are unlikely to be involved in host chromosome replication. 3 . Substrate Specijcity and Reaction Products

The substrate specificity or reaction product is often indicative of function. Specificity for DNA containing particular types of damage clearly indicates repair; conversely, degradation of duplex DNA to acid-soluble products makes a role in replication unlikely, but specific nicking at a replication origin is clear evidence for an involvement in replication. However, unless one knows the mechanism involved for a particular function and can correlate an observed in vitro reaction with a similar one in vivo, these criteria may be logically weak. B.

POSSIBLEFUNCTIONS

1. Salvage and Scavenger Functions

Nutritionally derived DNA is often degraded for reutilization of bases and nucleosides, or for a nitrogen source. For this purpose DNases occur in digestive systems of multicellular organisms and in the extracellular environment of many fungi and bacteria. Moreover, certain viruses bring about increased DNase activity in order to degrade host DNA for reutilization of monomer units in viral genomes. 2 . DNA Replication

DNases most likely serve several functions during DNA replication. They create origins of replication, at least in viral DNA replication where several such cases are well-documented. DNase activity is also involved

132

STUART LINN

in the swivel mechanism-the relief of positive supertwisting induced by replication. DNases act to process large multimeric viral DNA intermediates into unit length genomes for viral packaging. In addition, exonucleases are probably involved in the maturation of certain viral DNAs containing redundant termini, though the exact process is not understood. Finally, 3' --* 5' exonuclease activity is utilized, at least by bacterial polymerases, to assure fidelity during the replicative process. 3 . Genetic Recombination

Endonucleases are obviously involved in breaking DNA for recombination, as well as for gene translocations, inversions, and deletions. In addition, genetic data with microorganisms, as well as direct molecular observations, implicate exonucleases and possibly endonucleases that are specific for single-stranded DNA in both the generation and the resolution of recombination intermediates (29). 4. Conjugation and Transformution

Besides having a role in the process of DNA recombination, DNase activity is apparently involved in conjugation and transformation during translocation stages of these processes, at least in some bacteria, The cases of transformation of B. subtilis (30) and D.pneumoniae (31) are the best documented. In these instances, translocation of DNA into the cell evidently is driven, in large part, by degradation of one of the strands of the duplex, allowing entry of one undegraded single strand. 5 . DNA Repair Mistakes made during the replication process as well as damage induced by radiation, environmental chemicals, and spontaneous breakdown must all be repaired. One very exact process involves incision of the phosphodiester bond(s) by specific repair endonucleases and usually subsequent excision by exonucleases. Damage might also be dealt with by nucleases in a recombinational process, an alternative of particular importance when both strands of the duplex acquire damage in the same location, as during crosslinking. The subject of DNA repair has been extensively reviewed by Hanawalt et d . (32), and by Friedberg et al. in this volume (14). 29. 30. 31. 32.

C. M. Radding, Annu. Rev. Biochem. 47, 449 (1978). W. F. Bodmer and A. T. Ganesan, Genetics 50, 717 (1964). S. Lacks and B. Greenberg, J M B 101, 255 (1976). P. C. Hnnwalt, P. K. Cooper, A. K. Ganesan, and C. A. Smith, Annu. Rev. 979).

8. DEOXYRIBONUCLEASES

133

6 . Protection Against Foreign DNA

To help protect itself against viral and other foreign DNAs, the cell is armed with a variety of DNases. The most notable class are the bacterial restriction endonucleases that recognize and cleave DNA by virtue of its foreign methylation pattern (15). Bacteria also have exonucleases that inactivate DNA, presumably by recognizing unprotected termini on a foreign DNA molecule. Some bacterial viruses induce, or contain in the virus structure, inhibitors to exonuclease V-type enzymes in order to overcome such a protective host mechanism (7). In E. coli both recBC DNase and exonuclease I degrade to completion partially degraded, restricted DNA, so that no genetic activity remains. Some viruses also cause DNases to be made in order to degrade partially host DNA, not for ultimate reutilization of monomer units but to inactivate it. Likewise, the primary purpose of lysosomal and serum DNases might be of a protective, rather than of a scavenging nature. VI.

Control of Activities

Whereas DNases are required for many cellular functions, their activities must be carefully controlled in order to preserve the cell’s DNA. In many cases the manner of control is not known, but several mechanisms have been observed in others. Localization is clearly important. Enzymes in the periplasmic space, for example, do not normally encounter nuclear DNA. It is also likely that many transported DNases are activated, possibly by proteolytic action, only when they reach their final destination as is suggested to be the case with N. crussu. Another manner of regulation might exploit a requirement for particular cofactors-divalent cations or ATP, for example. Alteratively , exogenous factors might play an inhibitory role. Protein inhibitors are known for many DNases, the inhibition of pancreatic DNase by actin being a prime example. The inhibition of E. cofi endonuclease I by tRNA is interesting: The enzyme degrades duplex DNA extensively by double-strand breakage in the absence of tRNA, but puts an average of one nick into supercoiled circles (33)in the presence of tRNA. It is not known whether either or both of these activities are of significance in the cell. Many viruses also induce the synthesis of inhibitors of DNases, presumably to counter the hosts’ restriction enzymes. A final example of inhibitors might be the chromatin structure of eukaryotes; undoubtedly the presence of nonDNA chromosomal material affects DNase sensitivity. 33. W. Goebel and D. R. Helinski, Biochemistry 9, 4793 (1970).

134

STUART LINN

Finally, DNases can be controlled by the availability of their specific substrates. Repair enzymes act only upon appropriately modified DNAs, whereas replication and recombination enzymes may act in vivo only upon certain very specific DNA structures. In summary, the control of DNase activity is like that of other enzymes. Regulation occurs at the transcription, translation, and protein modification levels. In addition, the availability of substrate to the enzyme coupled with the extreme substrate specificity of many of the DNases are important factors.

VII.

Research Applications

A. ANALYTICAL APPLICATIONS DNases have been the dominant element in determining the structure of particular DNA species. Primary DNA sequence determinations, nearest neighbor analyses, and base composition determinations are generally dependent upon DNases. Restriction endonucleases are certainly the key factor in the recent burst of DNA sequence determinations. Secondary structure also can be monitored by DNases, utilizing enzymes specific for duplex or single-stranded DNAs as testing agents. Likewise, circularity is easily determined by monitoring sensitivity to exonucleases ; E. coli exonuclease I is useful for single-stranded circles (34), whereas the recBC enzyme is useful for duplex circles (35).Even the presence (or absence) of nicks in duplex circles can be monitored, in this case by observing the sensitivity of the polymer to E. coli exonuclease I11 (36). Finally, models for the tertiary structure of DNA in chromatin are in large part due to studies of digestion products formed by DNase I and micrococcal nuclease. The location of DNA terminal phosphates is also determined by nuclease sensitivity. Classically, venom phosphodiesterase and spleen exonuclease are utilized; the former requires a 3’-hydroxyl terminus to degrade in a 3’ --* 5’ direction; the latter a 5‘-hydroxyl terminus to degrade in a 5’ + 3‘ direction. The additional utilization of a phosphomonoesterase should give unequivocal results in all possible cases. 34. P. J . Goldrnark and S. Linn, PNAS 67, 434 (1970). 35. D. Lackey and S. Linn, “Methods in Enzymology,” Vol. 65, p. 26, 1980. 36. P. Modrich and I. R. Lehman, JBC 245, 3626 (1970).

8. DEOXYRIBONUCLEASES

135

B. PREPARATIVE APPLICATIONS DNases have been used historically to prepare oligonucleotides or mononucleotides, as well as to separate duplex from single-stranded DNA by the selective degradation of one with appropriately specific DNases. The most spectacular use of DNases, however, has been in the utilization of restriction endonucleases for molecular genetics (37). Not only are these enzymes used to obtain specific DNA fragments, but in some cases, by virtue of the staggered termini formed by some of the endonucleases, the fragments anneal to one another spontaneously. The applications of DNases to molecular genetics are increasing daily; undoubtedly DNases will continue to play a primary role in this most exciting field of biology.

37. M. Zabeau and R. J. Roberts, In “Molecular Genetics” (J. H. Taylor, ed.), Vol. 111, p. 1. Academic Press, New York, 1979.

This Page Intentionally Left Blank

Type I Restriction Enzymes BRIAN ENDLICH

STUART LINN

I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . A . The Phenomenon of Host-Controlled Restriction . . . . . . . B The Three Types of Restriction Enzymes . . . . . . . . . . C . Survey Among Bacteria . . . . . . . . . . . . . . . . . . I1 Biological Role . . . . . . . . . . . . . . . . . . . . . . . . 111. Genetics . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Purification and Properties of Type 1 Enzymes . . . . . . . . . . A Purification . . . . . . . . . . . . . . . . . . . . . . . . B. Physical Properties . . . . . . . . . . . . . . . . . . . . . C. Recognition Sequences . . . . . . . . . . . . . . . . . . . V. Reactions Catalyzed . . . . . . . . . . . . . . . . . . . . . . A.DNase . . . . . . . . . . . . . . . . . . . . . . . . . . B.ATPase . . . . . . . . . . . . . . . . . . . . . . . . . . C . Methylase . . . . . . . . . . . . . . . . . . . . . . . . . VLAssays . . . . . . . . . . . . . . . . . . . . . . . . . . . A . Endonuclease . . . . . . . . . . . . . . . . . . . . . . . B.ATPase . . . . . . . . . . . . . . . . . . . . . . . . . . C. Methylase . . . . . . . . . . . . . . . . . . . . . . . . VII . On the Mechanism of Cleavage-A Model Scheme . . . . . . . . VIII . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . .

.

.

.

138 138 139 140 142 143 144 144 145 147 148 148 149 149 150 150 151 151 151 156

137 THE ENZYMES. Vol . XIV Copyright @ 1981 by Academic Press. Inc . All rights of reproduction in any form reserved ISBN 0-12-122714-6

138

BRIAN ENDLICH AND STUART LINN

I. Introduction

A. THE PHENOMENON OF HOST-CONTROLLED RESTRICTION Many bacteria contain enzymatic systems that act to restrict the expression of foreign DNA introduced through phage infection, conjugation, or transformation. This phenomenon of host-controlled specijcity was first described in the early 1950s by Luria and Human (I), who studied T-even phages, and by Bertani and Weigle ( 2 ) , who investigated the restriction of the host ranges of A and P2 phages. It was observed that a particular phage could have widely different efficiencies of infection on several closely related bacterial strains, but when phages that had initially plated with low efficiency were replated on the same bacterial strain, the efficiency of infection increased dramatically. This modification of host range was not a hereditary genetic adaptation, however, since it could be lost by subsequent propagation of the phage in another bacterial strain. It was subsequently shown that the host-specificity system described by Luria and Human was unique to the T-even phages, whereas the system described by Bertani and Weigle was more widespread. The type I restriction-modification systems described in this review are of the latter, more general, class. Modification of the T-even phages involves the glycosylation of hydroxymethylcytosine residues, and has been reviewed elsewhere (3). Further investigation of A phage host specificity by Arber and coworkers (4-7) led to the hypothesis of a molecular mechanism in which special DNA sequences are acted upon by appropriate restriction and modification enzymes. Foreign DNA that happens to contain such specific sequences is cleaved by a restriction endonuclease upon entering the cell. When these same sequences exist in the bacterium’s own DNA, however, they are protected by a modification methylase that imparts methyl groups to bases within the sequences, rendering them resistant to endonuclease act ion. This hypothesis was substantiated in the late 1960s when restriction endonucleases from theE. coli strains K (8)and B (9) were discovered and 1. 2. 3. 4. 5. 6. 7. 8. 9.

S. Luna and M. Human, J . Bacreriol. 64, 557 (1952). G. Bertani and J. Weigle, J . Bacreriol. 65, 113 (1953). H. Revel and S. Luria, Annu. Rev. Genet. 4, 177 (1970). W. Arber and D. Dussoix, J M B 5, 18 (1962). D. Dussoix and W. Arber, J M B 5, 37 (1962). W. Arber, S. Hattman, and D. Dussoix, Virology 21, 30 (1963). W. Wood, J M B 16, 118 (1966). M. Meselson and R. Yuan, Nuture (London) 217, 1110 (1968). S. Linn and W. Arber, PNAS 59, 1300 (1968).

9. TYPE I RESTRICTION ENZYMES

139

isolated in highly purified form. These nucleases were genetically identified as the restriction enzymes and observed to be complex in terms of cofactor requirements, subunit composition, and interactions with the DNA substrate. Other restriction enzymes were soon isolated from other bacteria such as Huemophilus injuenzae (10, 1 I ) . Many of these endonucleases have distinctly different properties from those of theE. coli B and K nucleases, and are distinguished as two additional types of restriction endonucleases. Whether all of these latter enzymes function in sifu in a restriction capacity is not clear. B.

THETHREETYPESOF RESTRICTIONENZYMES

The type-I systems are exemplified by E. coli B and K. The restriction endonucleases are very large complex molecules consisting of three types of subunits with molecular weights of 400,000 or greater (12, 13). Magnesium, S-adenosylmethionine and ATP are required as cofactors, and ATP is degraded during the reaction (8, 9). The modification methylases require only S-adenosylmethionine, but are stimulated by Mg'+ and ATP (8, 14). The EcoB ( M a ) enzyme can be resolved from the nuclease by purification and contains two of the three subunits of the endonuclease. Preparations of Eco K contain both nuclease and methylase activities that have remained inseparable by purification and sedimentation. A highly distinguishing feature of type I systems is that while the modification methylase acts directly at the recognition sequences, the site of restriction cleavage is random, occurring between 1 and 5 kb from the recognition site (15, 16). By comparison, the type I1 enzymes are less complex, with molecular weights of less than 80,000 (17, 18). Only Mg2+is required for nuclease activity, and only S-adenosylmethionine is necessary for modification activity. Since the discovery of Hind11 (10) and EcoRI (191, and the finding 10. H. Smith and K. Wilcox, J M B 51, 379 (1970). 11. T. Kelly and H. Smith, J M B 51, 393 (1970). 12. H. Boyer, Annu. Rev. Microbid. 25, 153 (1971). 13. W. Arber and S. Linn, Annu. Rev. Biochem. 38, 467 (1969). 14. U. Kuhnlein, S. Linn, and W. Arber, PNAS 63, 556 (1969). 14a. The nomenclature for restriction-modification systems used in this review is that proposed by H . Smith, and D. Nathans, J M B 81, 419 (1973). 15. K . Horiuchi and N. Zinder, PNAS 69, 3220 (1972). 16. S. Adler and D. Nathans, BBA 229, 177 (1973). 17. R. Roberts, CRC Crir. Rev. 4, 123 (1976). 18. H. 0. Smith, Science 205, 455 (1979). 19. M. Betlach, J . Hershfield, L. Chow, W. Brown, H . Goodman, and H. Boyer, FP 35, 2037 (1976).

140

BRIAN ENDLICH AND STUART LINN

that this class of endonucleases cleaves DNA within the recognition sequence, new type I1 restriction endonucleases have been highly sought after for use as reagents for site specific DNA cleavage. More than one hundred type I1 restriction enzymes have been isolated from widely diverse species of bacteria. For a review of type I1 restriction endonuclease mechanisms see Modrich (201, and Chapter 10 in this volume by Robert Wells. The endonucleases of Huemophilus influenzae serotype f (21) and of prophages Pl(22) and P15 (23,24)were originally grouped with the type I systems due to their complex properties, but they are now considered to be a third distinct class (21). These enzymes are actually of intermediate complexity, being comprised of two subunits with molecular weights in the range of 250,000. As with the type I enzymes, these endonucleases require Mg2+,S-adenosylmethionine, and ATP. However, the type I11 endonucleases, for which ATP is strictly required, do not exhibit any measurable ATP hydrolysis. The methylases require only S-adenosylmethionine as the methyl donor and occur in forms that contain, or are free of, nuclease activity. It has been suggested that the nuclease and methylase activities are components of the same molecular complex, which dissociates into distinct activities under appropriate conditions. The cleavage site of type I11 endonucleases is roughly ten to twenty nucleotides from the recognition sequence and has a degeneracy of one or two bases (21-24). C. SURVEY AMONGBACTERIA

Due to the intense search for restriction endonucleases to use in recombinant DNA research, we know that almost every major group in gramnegative and gram-positive bacteria has at least one genus from which restriction endonucleases can be isolated. In many cases it has not yet been demonstrated that these enzymes have corresponding methylases, or that they are involved with host-controlled restriction and modification in vivo . This is partly because the main interest in restriction endonucleases has been their utility for DNA-cleaving reagents, and partly because little is known about many of the organisms from which they are derived. 20. P. Modrich, Q.R e v . Biophys. 12, 315 (1979). 21. L. Kauc and J. F’ickarowicz, EJB 92, 417 (1978). 22. B. Bachi, J. Reiser, and V. Pirrotta,JMB 128, 143 (1979). 23. J. Reiser and R. Yuan, JBC 252, 451 (1977). 24. S. Hadi, B. Bachi, J. Shepard, R. Yuan, K. Ineichen, and T. Bickle, J M B 134, 655 (1979).

9. TYPE I RESTRICTION ENZYMES

141

However, several species of bacteria other than E. coli have been shown to exhibit host-controlled restriction and modification. Among the most highly characterized of these are Salmonella typhimurium, Haemophilus inBuenzae, and Bacillus subtilis . The genus Salmonella contains several species in which biological restriction has been demonstrated. The S . typhimurium SA and SB systems (2S,26) have been extensively characterized (27) and have been shown to be allelic to the E. coli B and K systems (28). It is, therefore, reasonable to assume that they belong to the type I class of endonucleases, even though the S. typhimurium enzymes have not yet been purified. Salmonella typhimurium also possesses a specificity system, LT, which is not allelic with SA or SB. It is not known to which class of restriction-modification systems LT belongs (25). The genus Haemophilus contains numerous strains from which restriction endonucleases have been isolated. Examples of each class of endonuclease (10, 21, 29) have been identified, and several examples have been shown to be biologically active in host-range restriction. Bacillus subrilis has also been shown to possess several host specificity systems ( 3 0 , 3 / ) .These endonucleases are believed to be of the type I1 class; one, BsuR, has been purified and characterized (32, 33). The genetics of this system have also been extensively studied (34). Several other species of bacteria, including Corynebacterium diphtheriae (39, Pseudomonas aeruginosa(36), Staphylococcus aureus (37), Streptococcus faecalis (38), and Rhizobium leguminosarium (39) have been found to have biologically active restriction-modification systems, although these systems have not been extensively characterized. Although eukaryotes 25. C. Colson, A. Colson, and A. van Pel, J . Gen. Microbiol. 58, 57 (1969). 26. C. Colson, A. Colson, and A. van Pel, J . Gen. M i c r o b i d . 60, 265 (1970). 27. L. Bullas and C. Colson, M o l . Gen. Genet. 139, 177 (1975). 28. L. Bullas, C. Colson, and B. Neufeld, J . Bacreriol. 141, 275 (1980). 29. R. Gromkova and S. Goodgal, . I Bacteriol. . 127, 848 (1967). 30. T. Trautner, B. Pawlek, S. Bron, and C. Anagnostopoulos, M o l . Gen. Genet. 131, 181 (1974). 31. S. Ikawa, T. Shibata, T. Ando, and H. Saito, M o l . Gen. Genet. 177, 359 (1980). 32. S. Bron, K. Murray, and T. Trautner, M o l . Gen. Genet. 143, 13 (1975). 33. S. Bron and K . Murray, M o l . Gen. Genet. 143, 15 (1975). 34. T. Shibata and T. Ando, M o l . Gen. Genet. 170, I17 (1979). 35. T. Lampidis and L. Barksdale, J . Bacteriol. 105, 77 (1971). 36. B. Holloway, Bacteriol. Rev. 33, 419 (1964). 37. D. Ralston and B. Baer, J . Gen. M i c r o b i d . 36, 25 (1964). 38. E. Collins, Virology 2, 261 (1956). 39. E. Schwinghamer,Aust. J . B i d . Sci. 18, 333 (1%5).

142

BRIAN ENDLICH AND STUART LINN

have endonucleases, none have been reported to have the sequencespecific characteristics of restriction enzymes. II. Biological Role

The most obvious role for the type I host specificity systems is to restrict the exchange of genetic material between bacteria of different strains or species. The first examples of host range restriction were the limitation of phage infection and of transducing ability of A and P2 phages (2). However, conjugation is also severely impeded in crosses between E. coli Hfiand F - strains (40),and DNA uptake by spheroplasts is controlled by host specificity (41 ). Thus the type I restriction-modification systems limit the exchange of genetic material and serve to isolate the genomes of bacteria possessing them. In this way speciation may be promoted between bacteria that occupy ecological niches that do not contain well-defined barriers to interspecies and intraspecies gene exchange (12, 13). Such barriers to gene exchange that promote species divergence in eukaryotes are often geographic or physiological. In addition, the ecological niches of eukaryotes are generally more highly defined than those of bacteria. Based on the characteristics of the type I1 and I11 host specificity systems, which are often carried on plasmids and lysogenic phages, respectively, Roberts (17) and Chang and Cohen (42) have proposed a role for these systems in the promotion of site-specific recombination. In these cases, restriction enzymes would have the opposite function of the type I enzymes: They would promote the shuffling of genetic information by expediting recombination of genes carried by plasmids and prophages. However, some type I1 systems [e.g., BsuR (30, 31) and EcoRI (19)l can act in a true restrictive capacity as well. Although the only biological function linked to type I restriction enzymes is host specificity, these enzymes seem to be unnecessarily complex, both in their subunit complement and in their mechanism of action. It might thus be speculated that the type I restriction endonucleases are also involved in yet another in vivo function, such as nonspecific recombination; in fact, they might fortuitously act to restrict invasive DNA. Mutants in host-controlled specificity that have been isolated to date might even retain this other function. It is particularly noteworthy that among 40. W. Arber, Pathol. Microbiol. 25, 668 (1962). 41. R. Benzinger, PNAS 59, 1294 (1968). 42. S. Chang, and S . Cohen, P N A S 74, 4811 (1977).

TYPE I RESTRICTION ENZYMES

9.

143

the mutants characterized, no deletions have been demonstrated, and generally the one subunit that is altered has not been shown to be absent. 111.

Genetics

The elucidation of the enzymatic aspects of restriction and modification was facilitated by the extensive genetic analyses that preceded attempts to purify the enzymes. Mapping studies by PI phage transduction in E. coli B and K showed that the restriction-modification (hsd) loci lie t o the left of serB at about 98 min (43-45). Transduction analyses by Boyer and coworkers (44) demonstrated that the E. coli B and K systems are allelic. Fine-structure mapping determined the order hsdM-hsdS-hsdR-serB (46) (see below). These results are in close agreement with the Salmonella typhimurium SA and SB systems, which map at 98 min nearpyrB and have been shown to be allelic with the E. coli B and K systems (27). The LT system ofS. typhimurium has been shown to be an independent system nonallelic to SA and SB, and maps at 9 min, nearproC (25, 47). Two phenotypes of restriction-modification deficiency have been isolated in E. coli, those deficient only in restriction, “ f m + , ” and those deficient in both restriction and modification, ‘3-m-.” No r+m- mutants have been isolated, presumably because these strains would degrade their own DNA and not be viable. Complementation analysis involving crosses between r-m+ and r-m- phenotypes restored the r+m+ phenotype (48). This demonstrated that at least two genes, designated hsdR and hsdS, respectively, were required for restriction activity and suggested that some aspect of both restriction and modification was determined by the same gene. When an rk-mK+mutant gene (hsdR ti-) of E. coli K was introduced as a merodiploid with an rB-mB- recipient bacterium, the resulting phenotype was rti+mk+.This lead to the hypothesis that an hsdS gene, common to both restriction and modification, determined the specificity of these activities (48,49).By this criterion the restriction endonuclease and modification methylase contain a common subunit enabling both enzymes to recognize and interact with the same DNA sequence. The derivation of two-step r-m- mutants from r- mf bacteria allowed 43. W. Arber and D. Wauter-Willerns, Mol. Gen. Genet. 108, 203 (1970). 44. H. Boyer, J . Bacteriol. 88, 1652 (1%4). 44a. W. Arber, Progr. Nucleic. Acid. R t ~ s .Mol. B i d . 14, 1 (1974). 45. B. Bachrnann, K . Low,and A. Taylor, Microbiol. Rev. 44, 1 (1980). 46. J. Bulkacz, Ph.D. Thesis. University of California, San Francisco, 1972. 47. K. Sanderson and P. Hartman, Microbiol. Rev. 42, 471 (1978). 48. H. Boyer and D. Roulland-Dussoix, J M B 41, 459 (1969). 49. S. Glover and C. Colson, Genet. Res. 13, 227 (1969).

144

BRIAN ENDLICH AND STUART LINN

the identification of a third gene involved in restriction and modification, the hsdM locus (50). Hubacek and Glover (5I)demonstrated that two classes of r-m- mutants could be defined by the fact that a cross between one (hsdS-) and the other (hsdM-) would result in restoration of the r+m+ phenotype. They postulated that the hsdM and hsdS loci suffice to determine modification activity, but that all three genes are necessary for restrict ion. This view of host specificity has been upheld by in vitro complementation analysis of extracts prepared from bacteria with various mutant restriction-modification phenotypes. Purified B modification methylase can be added to extracts of hsdM or hsdS bacteria to restore restriction activity (52). This observation confirmed the hypothesis that the methylase and endonuclease contain common subunits. Yuan and coworkers (53,54) partially purified complementation factors from strains of E. coli K that contained lesions in each of the hsdM, hsdS and hsdR genes. Characterization of the enzymatic properties of each of these factors verified the assignments of the gene functions postulated earlier through genetic analyses. The finding that one gene product determines the specificity of both endonuclease and methylase action suggests that altered hsdS genes might produce novel host specificities. Such a case was observed by Bullas et al. (28, 55) during the isolation of recombinants between the allelic S . typhimurium SA and S . potsdam SP host-specificity loci. Among the isolates was a strain exhibiting a unique host specificity, which was designated SQ and shown to be the result of a recombination event within the hsdS locus.

IV.

Purification and Properties of Type I Enzymes

A. PURIFICATION Almost all that is known about the enzymatic properties of type I restriction endonucleases has been derived from studies of the purified EcoK and EcoB enzymes. The EcoK enzyme was purified approximately 50. 51. 52. 53. 54. 55.

S . Glover, Genet. Res. 15, 237 (1970). T. Hubacek and S . Glover, J M B 50, 111 (1970). S. Linn, J. Lautenberger, B . Eskin, and D. Lackey, FP 33, 1128 (1974). S . Hadi and R. Yuan, JBC 249, 4580 (1974). R. Buhler and R. Yuan, JBC 253, 6756 (1978). L. Bullas, C. Colson, and A. van Pel, J. Gen. Microbial. 95, 166 (1976).

9. TYPE I RESTRICTION ENZYMES

145

5000-fold by high-speed sedimentation, ammonium sulfate fractionation, chromatography on DEAE-cellulose then on phosphocellulose, and glycerol gradient sedimentation (8). EDTA and 2-mercaptoethanol are present in all buffers during purification and the final product contains both the methylase and endonuclease activities (56). Modifications of this procedure have been described (53,57). The purification scheme for theEcoB (58,59) is somewhat different, and yields an approximately 1000-fold purification. It uses magnesium chloride, streptomycin sulfate, and ammonium sulfate precipitations; chromatography on phosphocellulose then DEAE-cellulose; and finally, sucrose gradient sedimentation. All steps are carried out in the presence of EDTA and dithiothreitol. This procedure yields EcoB endonuclease free of modification activity. The modification methylase is purified by essentially the same procedure used to prepare the endonuclease, except that the methylase is resolved from the nuclease during DEAE-cellulose chromatography (14) and further purified on DNA-cellulose (60). It is interesting that restriction and modification activities copunfy in Eco K preparations, whereas they can be resolved during Eco B purification. The differences between purified EcoB and EcoK preparations may reflect the differences in the purification procedures.

PROPERTIES B. PHYSICAL The EcoK molecule is composed of three types of subunits, as predicted by genetic analysis. Subunit (Y has a molecular weight of 135,000, and the two smaller subunits, p and y , have molecular weights of 62,000 and 52,000, respectively (61). The enzyme sediments in sucrose gradients with an szO,,,, of 12, corresponding to a molecular weight of roughly 400,OOO. A subunit composition of (r2p2y1has been proposed to account for the physical properties of the enzyme. The properties ofEcoB resemble those ofEcoK, but with some interesting differences. EcoB is also composed of three types of subunits:a, p and y , with molecular weights 135,000, 60,000 and 55,000, respectively (59). However, Eco B purifies as several active oligomeric species with szo,w values from 11 to 18 S, corresonding to molecular weights from 450,000 to 750,000. The predominant form has a proposed subunit composition of 56. 57. 58. 59. 60. 61.

A. Haberman, J. Heywood, and M. Meselson, PNAS 69, 3138 (1972). R. Yuan, D. Hamilton, and J. Burckhardt, Cell 20, 237 (1980). S. Linn and W. Arber, PNAS 59, 1300 (1%8). B. Eskin and S. Linn,JBC 247, 6183 (1972). J. Lautenberger and S . Linn, JBC 247, 6176 (1972). M. Meselson, R. Yuan,and J . Heywood, Annu. Rev. Biochern. 41, 447 (1972).

146

BRIAN ENDLICH AND STUART LINN TABLE I

PHYSICAL PROPERTIES OF TYPE1 RESTRICTION ENDONUCLEASES ~

Proposed structure

~

~~

Approximate MW

Enzyme

Subunits ( M )

sz0.,+

EcoK Endonuclease (methy lase)

a 135,000

12

%PZY

EcoB Endonuclease

~1

11-18

a&y2"

6.6 11.3 8.0

P nI

105,000

PSYl

240,000

Pzr1

175,000

1

450,000

P 62,000 52,000 135,000 P 60,000 Y 55,000 P 60,000 Y

EcoB Methylase

Y

(I

55.000

450,000-750,000

Predominant form.

cyzp4yz(59). Two active forms, which are enzymatically indistinguishable,

have also been isolated by native gel electrophoresis (52). The modification methylase of E. coli B, purified free of endonuclease activity, contains subunits that are electrophoretically identical to the p and y subunits of the endonuclease (60). The freshly purified enzyme has of 6.6, a molecular weight of 105,000, and a structure of plyI. an szo,w Upon storage, this enzyme disproportionates to an alternate form with an szo,wof 11.3, and a subunit composition of p3y1. Under appropriate conditions other forms of the methylase, including p Z y l ,may be obtained. All forms are enzymatically active, but whether all forms exist in vivo is not known (52). A comparison of the physical properties of EcoK and EcoB enzymes is shown in Table I. An attempt was made to match the enzyme subunits to the respective host-specificity genes; however, this investigation was not entirely successful. Linn and Arber (58) showed that extracts from bacteria carrying mutations in the restriction genes that did not contain active enzyme could be mixed to generate in vitro complementation. It was demonstrated that purified modification methylase could supply p and y subunits to generate restriction in hsdS- and h s d W extracts (52). From this it was concluded that the /3 and y subunits of the modification enzyme must be catalytically active in the restriction reaction. SincehsdR is the gene peculiar to restriction, although the cy subunit is not found in the methylase enzyme it is evidently controlled by thehsdR gene. Linn and co-workers (52)purified a factor from hsdS- bacteria that generates DNase and ATPase activities when added to purified modification enzyme. Sedimentation and electrophoretic analysis demonstrate a molecular weight of approximately

9. TYPE 1 RESTRICTION ENZYMES

147

130,000 for this factor, which Linnet 01. (52) concluded was a monomeric form of subunit a. To determine the correlation between hsdM and hsdS with /3 and y , preparations of p were isolated from native polyacrylamide gels. The eluted subunit was too unstable, however, to reproducibly complement either hsdM- or hsdS- extracts. Consequently the ,L3 and y subunits remain unassigned as gene products among hsdM and hsdS . C. RECOGNITION SEQUENCES The recognition sequence for EcoB is T-G-A(N&T-G-C-T (62-6S), where N8 seems to be any sequence of eight nucleotides. Methylation probably occurs as follows: T-G-A*-(Ns)-T-G-C-T A-C-T-(Ni)-A*-C-G-A

where A” is 6-methyladenine. The recognition sequence for Eco K (66 ) is A-A-C-(Ne)-G-T-G-C T-T-G-(N;)-C-A-C-G

The site modification of the EcaK sites is not known, although 6-methyladenine is the product, and both strands are known to be modified in vivo . Although both sequences are hyphenated, the interruptions are of different sizes, a surprising result in view of the allelism of the two systems. Nonetheless, when aligned, the sequences share four of seven specific base pairs, including at least one of the methylation positions: Eco B

Eco K

s’-T-G-A“-N-N -N -N -N -N -N -N -T-G-C-T A-c-T-N’-N~-N~-N‘-N’-N~-N~-N~-A-c-G-A-~~

I

s’-N A-A-C-N -N -N -N-N -N -G- T !B-N N’-T-T-G-N‘-N’-N’-N’-N’-N’-CA*-C-G-N’-5’

(The lower K strand is presumed to be methylated at the only adenine residue.) It is also possible, but not proved, that outside (nonspecific) base pairs are required for an active EcoK sequence. 62. R . Sommer and H. Schaller, Mol. G m . Gcwrr. 168, 331 (1979). 63. J . Lautenberger, N . Kan, D. Lackey, S . Linn, M . Edgell, C. A. Hutchinson 111, P N A S 75, 2271 (1978). 64. J. Ravetch, K . Horiuchi, and N . Zinder, PNAS 75, 2266 (1978). 65. J . Lautenberger, M . Edgell, C . A. Hutchinson 111, and G . Godson. J M B 131, 871 (1979). 66. N . Kan, J . Lautenberger, M. Edgell, and C. A. Hutchinson III,JMB 130, 191 (1979).

148

V.

BRIAN ENDLICH AND STUART LINN

Reactions Catalyzed

The type I restriction endonucleases effect two types of reaction, a DNase and an ATPase. In addition, EcoK acts as a modification methylase. A. DNASE The nuclease reaction has a pH optimum of approximately 8 (591, and requires a divalent cation, ATP, and S-adenosylmethionine. Duplex DNA that contains unmodified specificity sites is the only substrate. The nuclease does not appear to turn over, as no more than a single-strand scission is observed per enzyme molecule in the presence of excess substrate (8, 16, 59, 67). Consequently, two enzyme molecules are required to produce a double-strand DNA cleavage. The cleavages made by the endonuclease are one kb to several kb from the recognition sequence. In the case of EcoK, cleavage seems to occur on either side of the site. In the case ofEcoB, cleavage occurs only to the left of the sequence T-G-A(N,)-T-G-C-T. Molecules shorter than roughly 1000 base pairs are not cleaved by the enzyme, even if they contain a recognition site. Further details of the DNase reaction are given in Section VII. The role of S-adenosylmethionine in the DNase reaction (8, 52) is not totally understood. Neither an adenine, a methyl, nor a homoserine group seems to be transferred to a macromolecule during the reaction. Since the S-adenosylmethionine does not appear to be degraded during the restriction reaction but is needed for DNA binding, it is possibly an allosteric effector (68). Sucrose gradient sedimentation analysis has demonstrated that S-adenosylmethionine remains tightly bound to the enzyme-DNA complex, although not covalently, since S-adenosylmethionine does not band with either DNA or protein in CsCl (-52). The presence of S-adenosylmethionine is required for initiation of nuclease and ATPase activities, but is not necessary for their maintenance. The analogues, 5'methylthioadenosine, S-adenosylhomocysteine, and S-adenosylethionine cannot replace S-adenosylmethionine as a cofactor for the restriction reaction. S-adenosylethionine is a potent inhibitor of the reaction if present initially, but has no effect if added during the ATPase phase of the reaction, approximately 5 minutes after the DNase initiates. The apparent K , of S-adenosylmethionine is 0.40 pm for the nuclease reaction. 67. D. Roulland-Dussoix and H. Boyer, BBA 195, 219 (1969). 68. R. Yuan, T. Bickle, W. Ebbers, and C. Brack, Nature (London) 256, 556 (1975).

9. TYPE I RESTRICTION ENZYMES

149

ATP is also required for the DNase reaction. It can be replaced by GTP and dATP with about 30% efficiency, but not by ADP or the p,ymethylene -imido or -thio analogs of ATP. The endonuclease (and ATPase) reactions are mildly inhibited by ADPOPNP, but even high concentrations of ADP are not inhibitory (69).

B. ATPASE Associated with the endonuclease is an activity that hydrolyzes ATP to ADP and inorganic phosphate (69). The ATPase activity appears to be intrinsic to the endonuclease molecule, since the two activities have identical cofactor requirements and are inseparable by purification, sedimentation, and electrophoresis. In contrast to the nuclease, the ATPase is able to turn over, hydrolyzing approximately 10‘ ATP molecules per minute at 37” (70). Although the function of the ATPase activity is not known, it probably has a role in the complex mechanism whereby the nuclease introduces a single-strand break at a random site several kb from the recognition sequence. Unmodified DNA that contains the recognition site close to the end of the DNA molecule, while not restricted, supports limited ATP hydrolysis (7/ ):A peculiar feature of the restriction reaction is that ATP is hydrolyzed in vitro for up to several hours after nuclease activity has terminated.

C. METHYLASE The EcoB modification enzyme has a pH optimum of 6, requires S-adenosylmethionine and is stimulated by Mg2+and ATP, which it does not degrade (14, 60). S-Adenosylmethionine is the only compound that can act as a methyl donor, whereas S-adenosylhomocysteine, S-adenosylethionine, and 5’-methylthioadenosine inhibit methylation as they do the nuclease activity (52). DNA is methylated only if it is duplex and contains unmodified or half-modified recognition sites, and 6-methylaminopurine is the methylation product, with one methyl group being incorporated into each strand of the recognition sequence (72, 73). To escape restriction, both DNA parental strands must, of course, be methylated before each round of replication. Hence, half-methylated 69. B. Eskin and S. Linn, JBC 247, 6192 (1972). 70. H. Boyer, E. Schibiensky, H. Slocum, and D. Roulland-Dussoix, Virology 46, 703 (1971). 71. J. Rosamond, B. Endlich, and S . Linn,JMB 129, 619 (1979). 72. U. Kuhnlein and W. Arber, J M B 63, 9 (1972). 73. J. Smith, W. Arber, and U. Kuhnlein, J M B 63, 1 (1972).

150

BRIAN ENDLICH AND STUART LINN

sites are the normal substrate in this process, and, not surprisingly, halfmethylated recognition sequences are the most active substrate for this enzyme (74). In fact, the enzyme probably acts poorly on foreign, totally unmodified DNA in vivo, Like the endonuclease, the turnover of methylase activity is very slow (-51, 52, 56). Under optimal in vitro conditions turnover rate was measured as 7 hours with totally unmodified sites, and roughly 1% of that value with half-methylated sites (74).

The following assays were applied to the EcoB enzymes, but are clearly applicable to all type I restriction or modification enzymes. A. ENDONUCLEASE 1. Transfection This assay measures the decrease in infectivity of fd Replicative Form (RF) DNA on nonrestricting E. coli spheroplasts following treatment with the endonuclease (14).

2. Sucrose Gradient Sedimentation The extent of digestion of a DNA substrate is analyzed by velocity sedimentation in sucrose gradients (8, 52, 59). 3. Agarose Gel Electrophoresis A circular DNA substrate with a unique restriction site, such as fdlOl RF DNA in the case ofEcoB, is digested with endonuclease. The reaction products are run on 1% agarose gels, with which the conversion of Form I (supercoiled circles) to Form I1 (nicked circles) and Form I11 (linear DNA) can be visualized and quantitated (71). 4. Other Methods

The rendering of DNA supercoiled circles to a form susceptible to the recBC DNase (59, 75,76), and the use of nitrocellulose filters to distinguish cleavage products from a closed circular substrate have been described (16, 77). 74. 75. 76. 77.

G. Vovis, K. Horiuchi, and N . Zinder, PNAS 71, 3810 (1974). A. Karu, V. MacKay, P. Goldmark, and S. Linn, JBC 248, 4874 (1973). D. Lackey and S. Linn, "Methods in Enzymology," Vol. 65, p. 4, 1980. J. Reiser, C. Bentley, and R. Yuan, Anal. Biochem. 75, 555 (1976).

15 1

9. TYPE I RESTRICTION ENZYMES

B. ATPASE Since endonuclease has an intrinsic ATPase activity that is S adenosylmethionine-dependent,the enzyme can be assayed by including [yY2P]ATPin the reaction. The fraction of ATP that is degraded is determined either by measuring the 32Pthat does not absorb to charcoal or by chromatography. This is convenient for assaying activity during the early stages of purification when contaminating nucleases are present (59). Contaminating ATPase activities can be controlled by the omission of S -adenosylmethionine. C. METHYLASE 1.

Transfection

The increase in infectivity of fd RF DNA upon restricting E. coli lysozyme spheroplasts due to modification methylation is measured (14). 2. Incorporation Assay [3H-methyl]S-adenos ylmethionine is included in the reaction mixture, and the incorporation of the 3H-methyl label into acid-precipitable DNA is monitored (60).

VII.

On the Mechanisms of Cleavage-A

Model Scheme

Among the peculiar features of the type I restriction reactions is that cleavage is at random sites from roughly 1 kb to several kb from the recognition sequence. The enzyme introduces single-strand nicks and does not turnover as a nuclease, so that two enzyme molecules are required to complete a double-strand cleavage. The time course of the nuclease reaction is relatively short in vitro, terminating within minutes of initiation, whereas ATP hydrolysis may continue for several hours. The scheme described here (Fig. 1) is derived by interpreting many observations made with the EcoK and EcoB enzymes. The initial step in the restriction process is recognition of, and binding to, the unmodified DNA specificity site. The endonuclease first binds S-adensylmethionine, thus enabling it to form a recognition complex with DNA that contains specificity sites (68). In the case ofEcoK, these complexes can be trapped on nitrocellulose filters, but filter binding by EcoB has not been demonstrated. In both cases, however, the enzyme so bound appears to have a different configuration, as visualized by electron microscopy, which suggests an allosteric conversion or the loss of an enzyme

152

BRIAN ENDLICH AND STUART LINN

1

ATPase

I

ATRi!r

ADP + P,

Oligonucleotides

FIG. 1. Schematic representation of a hypothetical model by which the type-I restriction enzymes, EcoB andEcoK, degrade DNA. The EcoB enzyme and sequence were arbitrarily chosen for the figure. For this reason, the model is drawn showing the unique polarity ofBroB translocation: distal to the trimer portion of the sequence. No such unique polarity has been reported for EcoK. The supercoiled loops have been observed for EcoK, but not for EcoB (see text). At what stage of the reaction they exist is not really clear, so the model is arbitrarily drawn in this regard. Finally, S-adenosylmethionine (AdoMet) stays bound to enzyme throughout the process, but is not so shown. subunit (71, 78). ATP is not required for the formation of recognition complexes, and, in the absence of ATP, EcoK can act to methylate the site. The function of S-adensylmethionine in recognition is not known, but it may be an allosteric effector whose absence, as in times of methionine deprivation, prevents restriction, and thus degradation, of the host DNA. An implication of their hyphenated recognition sequences is that the EcoK and EcoB restriction endonucleases may have two sites for binding to the DNA duplex (57, 71). There are several features of the restriction reaction that are more easily explained if two binding sites are involved. A reaction intermediate observed by electron microscopy in EcoB (71, 79) and EcoK (57) reactions is a DNA loop, which is generated by the enzyme remaining bound to the recognition sequence and translocating along the DNA until it reaches the point of cleavage. It is possible that the enzyme 78. C. Brack, H. Eberle, T. Bickle, and R. Yuan, J M B 108, 583 (1976). 79. J. Rosamond, B. Endlich, K. Telander, and S. Linn, CSHSQB 43, 1049 (1979).

9. TYPE I RESTRICTION ENZYMES

153

remains anchored at the recognition sequence with one enzyme binding site while using the other for the process of DNA translocation (see Fig. 1). Upon reaching the site of cleavage a nick is made and, at least for EcoB, about 70 nucleotides are released as oligonucleotides (80). EcoB acts in an apparently random fashion with respect to which strand of the DNA duplex is initially cleaved (Endlich and Linn, unpublished). What triggers the nick, or how the enzyme knows it must travel at least 1 kb, is unknown. After making the nick the enzyme appears to stay bound to DNA, hydrolyzing ATP (71, 79). A two-binding-site model might also explain how the second enzyme molecule that converts a single-strand nick to a double-strand break recognizes where to insert its cleavage. Since the first endonuclease molecule is anchored at the recognition site and also bound at the site of singlestrand cleavage, these sites would be held in close proximity, allowing the second nuclease molecule to complete the cleavage at the proper site. It has been shown that no more than one double-strand cleavage is introduced for each recognition site, even with a large excess of enzyme. The looped intermediate, in which the recognition site is still occupied, could explain why the site cannot be used by another endonuclease molecule to translocate DNA and introduce a single-strand cleavage within the area spanned by the first molecule. On the other hand, the loops do not explain why sites of fully restricted, phenol-extracted product DNA molecules are not acted upon again. Some of these molecules may be too short, but even when full-length linear product of fdlOl RF DNA (on EcoB site) is so used, it is not cleaved again (B. Endlich and S. Linn, unpublished). (It was cleaved randomly about the circle at first.) Yet fd wild-type RF DNA is easily cleaved twice by virtue of its two recognition sites (59). No moiety of ATP or S adenosylmethionine is transferred to these inactive recognition sites, and no known enzyme subunit can be found there [(59), and Endlich and Linn, unpublished]. Indeed, these sites can be rnethylated by the modification enzyme (60). Perhaps a small, undetected part of the enzyme remains to block the restriction, but not the modification. This uncertainty remains one of the true mysteries of the restriction enzymes’ reactions. The loops of restriction intermediates of EcoB as observed by Rosamonder af. (71, 79) were of a relaxed topology when visualized by the polylysine technique of Williams (81). In contrast, Yuan et al. (57), utilizing a pentylamine (78) spreading technique, have visualized a restriction intermediate in which the loops contain supercoils. Several models have 80. M. Kimball and S. Linn, EERC 68, 585 (1976). 81. R. Williams, PNAS 74, 2311 (1977).

154

BRIAN ENDLICH AND STUART LINN

been proposed to account for the supercoiled restriction intermediates. Supercoiling would be expected if translocation of DNA involved tracking within one of the grooves of the helix such that, in effect, the DNA rotated while the enzyme remained anchored at the recognition site. An alternate model for DNA translocation, in which the movement of DNA is accomplished by wrapping DNA around the enzyme by full turns, would also result in the formation of a supercoiled loop. Mechanisms such as these, which have been proposed for other DNA translocating enzymes such as DNA gyrase (82-84), may thus be applicable to the translocation mechanism of the type I restriction enzymes. To test for similarities between the gyrase and restriction mechanisms, Yuan and co-workers (57) added the potent DNA gyrase inhibitors novobiocin and nalidixic acid to Eco K reactions. No effect on either the formation of filter-binding complexes or DNA cleavage was observed. Thus, while the mechanism of formation of a supercoiled restriction intermediate is unknown, the presence of the structures might give rise to physical constraints that limit the region of cleavage to between 1 and 5 kb from the recognition site. There are several explanations for the contrasting observations of Rosamond et al. with EcoB (71, 79) and Yuan et al. with EcoK (57). Rosamond et al., used the 6.4 kb fd RF DNA as substrate. Both linear and circular substrates were used, but only linear product was examined. Conversely, Yuan et al. observed twisted loops with circular, not linear, pBR322 DNA (approximately 4.4 kb). Linear DNA had supercoiled loops only if it were the longer A DNA. Although it is hard to visualize a mechanism in which the formation of a supercoiled loop, as a consequence of DNA rotation or winding at fixed DNA site, would be dependent on the length or topology of the DNA substrate, this remains a possibility. Alternatively, the difference in results might be a consequence of the respective spreading techniques used to prepare the samples for electron microscopy. In an effort to determine if the asymmetry of the recognition sequence affected the direction of DNA translocation, Rosamond et af. (7/, 79) utilized fd replicative DNAs that contained one or two EcoB sites in known positions and orientations. The DNAs were converted to a linear form with HincII, which has one unique cleavage site for this substrate. Agarose gel analysis of the EcoB cleavage products of the linear fd RF 82. L. Lin and J. Wang, PNAS 75, 2098 (1978). 83. M. Gellert, K. Mizuuchi, M. O D e a , H. Ohmori, and J. Tomizawa, CSHSQB 43, 35 (1978). 84. C. Peebles, N . Higgins, K. Kruezer, A. Morrison, P. Brown, A. Sugino, and N. Cozzarelli, C S H S Q B 43, 41 (1978).

9. TYPE I RESTRICTION ENZYMES

155

DNA (in addition to visualization by electron microscopy of restriction intermediates of the linear DNA substrate) demonstrated that EcoB translocates DNA only from the left side of the sequence 5'-T-G-A-(N&T-GC-T-3'. In contrast to these results, the EcoK enzyme appears to translocate DNA from either site of its recognition sequence (57). In light of the fact that many of the characteristics of these endonucleases are nearly identical, this result is puzzling. An explanation proposed by Yuan et al. (57) to account for the bidirectionality of EcoK is that it may contain a subunit arrangement that acts as a swivel, allowing EcoK to translocate DNA from either direction. Possibly the subunit arrangement of EcoB constrains its movement to one direction only. The termini of DNA restricted by type I enzymes are susceptible to exonucleases that degrade DNA containing 3'-hydroxyl and 5 ' phosphomonester groups (59). However, the 5' termini are resistant to phosphorylation by polynucleotide kinase, even following denaturation and phosphomonesterase treatment (59). It has been conclusively demonstrated that no moiety of S-adensylmethionine or ATP is incorporated into the restricted DNA [(59) and Endlich et al., unpublished]. It is possible that the 5' terminus is blocked by a small peptide in a covalent linkage, but this has not been demonstrated conclusively. Neither the 3' terminus of the restricted duplex nor the termini of the acid-soluble oligonucleotides released during DNA cleavage have a n y apparent specificity (Endrich and Linn, unpublished). The role of ATP in the restriction reaction remains a mystery. The function of the ATP hydrolysis that occurs during the nuclease phase of restriction may be that energy must be expended during the translocation of DNA, which forms the loop-containing intermediate. It has been demonstrated that a noncleavable DNA substrate that contains an sB+ site very near the terminus will support ATP hydrolysis and form looped restriction-type intermediates (71). The ATP hydrolysis that continues after restriction has terminated may be the consequence of an alteration in enzyme structure such that the molecule is inactive as a nuclease, hut active as an ATPase (69), or it may be that the ATP is being used to hold together the terminally looped product of the restriction reaction (see Fig. 1). It is obvious that if ATPase activity arises after restriction in vivo, the cell must have some means of controlling it. The subsequent degradation of the restricted DNA by the E. coli recBC DNase ( 7 4 , 8 5 , 8 h )and exonuclease I might serve to uncouple this reaction. 85. P. Goldmark and S. Linn, PNAS 67, 434 (1970). 86. P. Goldmark and S. Linn, JBC 247. 1849 (1972).

BRIAN ENDLICH AND STUART LlNN

156 VI I!.

Conclusions

Though many questions have been answered about type I enzymes, the answers have posed even more questions. We do not know the nature of the ATPase prior to or during the nuclease reaction; nor do we know the significance of it after cleavage has occurred. Why do we have a stable enzyme-S-adensylmethionine-restrictedDNA complex (with a terminal loop) that hydrolyzes ATP? What tells the enzyme where to cleave? Another class of questions centers upon the DNA product. Why are the 5' termini resistant to polynucleotide kinase? Why are recognition sites still substrate for modification, but no longer substrate for restriction? Why are the cleavage sites within a defined, but relatively great, distance from the recognition site? Finally, the differences between EcoB and Eco K that were summarized in this review should not cloud the overall close similarity of the very complex reactions that they catalyze, Perhaps the most significant question is why such complex reactions are used, only to make foreign DNA a substrate for ultimate disposal by exonucleases. Why is the simpler type I1 mechanism not used? Is this elaborate sequence of events necessary to prevent efficient rescue of the DNA [some rescue of restricted DNA markers does occur (511, or do these unique enzymes serve yet another unique function?

Type 11 Restriction Enzymes ROBERT D. WELLS C. K. SINGLETON

RONALD D. KLEIN

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Catalytic Properties . . . . . . . . . . . . . . . . . . . . . A. Kinetic Parameters . . . . . . . . . . . . . . . . . . . B. Reaction Pathway . . . . . . . . . . . . . . . . . . . . IV. Ionic Strength and Solvent Effects on Enzyme Specificity . . . A. Effect on EcoRI . . . . . . . . . . . . . . . . . . . . B. Other Enzymes . . . . . . . . . . . . . . . . . . . . . V. Synthetic Oligonucleotides a s Substrates . . . . . . . . . . . VI. Substituted DNAs . . . . . . . . . . . . . . . . . . . . . . A. Studies with Base Analogs . . . . . . . . . . . . . . . . B. Methylated DNAs . . . . . . . . . . . . . . . . . . . . VII. Inhibitor Studies . . . . . . . . . . . . . . . . . . . . . . A. Reagents That Modify Proteins . . . . . . . . . . . . . . B. Polynucleotides as Inhibitors . . . . . . . . . . . . . . . VIII. Influence of Drugs and Other Ligands on Cleavage Specificities . IX. Cleavage of Single-Stranded DNA Substrates by Certain Restriction Endonucleases . . . . . . . . . . . . . . . . . . X. Cleavage of DNA-RNA Hybrids . . . . . . . . . . . . . . . XI. Insolubilized Restriction Enzymes . . . . . . . . . . . . . . XII. Crystallization of Restriction Endonucleases . . . . . . . . . . XIII. Genes for Restriction Endonucleases . . . . . . . . . . . . . XIV. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . 11. General Properties

.

. . .

. . . . . .

. . . .

. . .

. . . . .

158 159 167 167 169 172 172 174 176 177 177 180 181 181 182 183

185 187 188 188 189 191

157 THE ENZYMES, Vol. XIV Copyright 0 1981 by Academic Press, Inc. All rights of reproduction in any form reserved ISBN 0-12-122714-6

158 1.

R. D. WELLS, R. D. KLEIN, A N D C . K. SINGLETON

Introduction

The current status of type I1 restriction endonuclease enzymology is quite unusual due to the history of this segment of molecular biology. More than 210 restriction endonucleases have been purified, at least partially, and their recognition sites identified. However, the genetic and biochemical characterization of these enzymes is meager. Indeed, they have usually been isolated as site-specific DNA endonucleases without genetically determining if they are involved in DNA restrictionmodification. Moreover, the vast majority of the restriction endonucleases are relatively crude preparations, and little or nothing is known about the properties of the enzymes. Never before in the history of biochemistry has there been such a large number of enzymes that have been so useful and widely available, but so poorly understood in biochemical terms. Some of the reasons for this situation are as follows: (i) Type I1 restriction endonucleases have been widely used for gene cloning experiments as well as for mapping studies. Due to the nature of these techniques, the majority of investigators have been more concerned with using the enzymes as tools than with characterizing them per se. (ii) The generous and expedient dissemination of information and bacterial strains has been a factor. Richard J. Roberts (Cold Spring Harbor Laboratory) has performed an important service in this regard. (iii) Several industrial firms that purify and sell the enzymes have also provided a valuable service. In general, the commercial preparations have been relatively reliable compared to the preparations of the somewhat more unstable and sophisticated enzymes described in other chapters in this volume. This is due in part to the relative ease of isolation and the stability of these restriction endonucleases. (iv) The development of DNA sequencing techniques has provided a rapid and reliable method for determining the specificity, and for indirectly determining the purity of restriction enzymes. Thus, the analytical methods for assessing the enzyme specificities are extremely precise, thereby allowing the utilization of less than completely pure endonuclease preparations. Since the endonucleases recognize precise sites on the DNA duplex, it is likely that these proteins will be further studied as model systems for understanding DNA-protein interactions. Much needs to be done to characterize the enzymes and their reactions and properties. The notion is gradually emerging that each restriction enzyme has its own characteristic properties. This is as expected by the enzymologist since each enzyme is a different protein. However, this concept has often been overlooked due to the considerations enumerated above. Likewise, we are beginning to appreciate subtle differences between various DNA

10. TYPE [I RESTRICTION ENZYMES

159

recognition sites for a given enzyme, and the role of the properties of neighboring sequences (I 1. To the best of our knowledge, no restriction endonucleases have been isolated from eukaryotic sources; this may be an important clue to our eventual understanding of their biological function. Several excellent reviews have focused on particular aspects of restriction enzymes ( 2 - 6 ) . The purpose of this chapter is to assess critically the status of the enzymology of this group of enzymes.

II.

General Properties

Table I lists properties of commonly used restriction endonucleases. This table was designed to be of maximum utility for the practicing molecular biologist. The majority of these enzymes are commercially available. The organization of the table is based on each enzyme’s recognition site. The recognition sites are centered around a “core” di- or tetranucleotide that permits identification of both isoschizomers and specific sequences of interest. Only one strand of the duplex canonical site is given; this strand reads in a 5‘ + 3’ direction. The complementary strand can be inferred. The phosphodiester bond that is cleaved is indicated by an arrow (v); the point of cleavage within the complementary strand is located by a twofold symmetrical relationship to the arrow. Where two bases appear in parentheses, either base may be present at that position. The name of the enzyme, in accordance with Smith and Nathans (See 69) and as cataloged by Roberts (6), appears next. The name of the microorganism from which an enzyme was isolated, the source of the microorganism, and additional DNA cleavage data can be obtained in Roberts (6) and the references therein. Several factors are responsible for the general lack of physical data and precise determinations of the conditions necessary for optimal activity of restriction endonucleases: (a) Purification to homogeneity is not critical in 1 . R. D. Wells, T. C. Goodman, W. Hillen, G . T. Horn, R. D. Klein, J. E. Larson, U. R. Muller, S. K. Neuendorf, N . Panayotatos, and S . M. Stirdivant, Progr. Nuclric Acids Res. Mol. B i d . 24, 167 (1980). 2. R. J . Roberts CRC Crit. Rev. Biocliem. 4, 123 (1976). 3. P. Modrich, Quart. Rev. Biophys. 12, 315 (1979). 4. L. Grossman and K. Moldave, “Methods in Enzymology,” Vol. 65, 1980. 5. J . G. Chirikjian, “Gene Amplification and Analysis, Vol. I: Restriction Endonucleases.” Elsevier-North Holland, Amsterdam, 1980. 6. R. J. Roberts, Gene 8, 329 (1980).

TABLE I

GENERALPROPERTIES OF COMMONLY USED RESTRICTION ENDONUCLEASES OFTIMUM REACTION

CONDITIONS

Ionic strength Recognition site' AT GATC G&C

-8

Number of Enzyme

MW

subunits" ( x 10-3)

Dpn I1

1

70

Dpn I

1

20

Temp. ("C)

Mg2+ (mM)

37b

26

PH

Buffer'

Salt

( m M Tris-HCI)

( m M NaCI)

7.6b

6.7

37b

2b

7.6

6.7

~GATC

Fnu EI

37b

6b

7.9b

6.0

~ A T C

Sou 3AI

30b

15b

7.9

6.0

6b

7.9b

~ A T C

ZGATCT TIGATCA

Mbo I BglII

370

6.0

10

9.5

20 mM glycine-

Be1 I

12b

7.46

12

&ATCC

Barn HI

lO(l3)#

8.5(7.5)

2O(m

GIGATCC

Bst I

0.2

7.5-8.0

10

CGAT&

Xor I1

6(12-24)

7.4

G'UTTC

Eco RI

5

7.1-7.5

I

27

30-40

NaOH

6.0 100

AN G'hTC GANTC

Hinfl Hha I1

2

24

37b 37b

1O b 7b

7.9 9.0b

6.6b 40 m M glycine-

NaOHb

References

TA 346

6b

7.9b

10

45

5

7.7-8.1

10

Xba I

37b

6b

7.9b

Ava I1

376

1O b

GGCC

HaeIII

70

GGCC

Sfa

GGCC GGCC GGCC

BsrtI Ngo I1

CGCG

Fnu DII

376

6b

7.9b

CGCG

Tha I

60

1O b

7.46

TGGCCA

Bal I

37b

6b

7.99

CGGCCG

Xma III

25

8

8.2

Ah1

37b

6b

7.9

Hind 111

370

5- 10

8- 9

CTGCAG CTGCAG CTGCAG

Pst I Xma I1 Sal PI

37b 376 37b

6.6b 6b 10

7.4b 7.9 7.9

G(+)GC(,A)C

Hgi A1

30-45

lob

7.5-8.5

GTAC

Rsa I

GTTAAC

Hpa I

TCTAGA

29, 30.5

d

6.0

GT W+,CC

7.5-9

20

5

7.5

50

6.6

7.56

GC

-z

AGCT AAGCTT

BSP

376

68 11 3s

376 55 37b

10 1 20

7.4 8.5 8-8.5

6.6 10 l00b 25 6.0

100-200 50 50- 150

10

6.0 10 6.0 10

30-60

TGCA 6.6 6.0 6.0 10

50 50 100- 150 (Continued)

TABLE I (Continued) Ionic strength Recognition site‘

Enzyme

Number of subunitsa

MW (X

10-7

Me2+

Temp. (“C)

(mM)

PH

BufferC (mM Tris-HCl)

salt

(mM NaCl)

CG 20 (10) m M sodium phosphate

37

5

6.8(7.4)b

Nci I

37b

6b

7.5

CCCGGG

Sma I

2S-37b

lob

7 9

10

CCCGGG

Xma I

37b

6b

7.9”

6.0

GCGC

Hha I

37b

6.6b

7.4b

6.6

506

Hae I1

376

6.6b

7.4

6.0

SOb

TCGA

Taq I

37bJ

CTCGAG

Xho I

CTCGAG

Sla I

CPyCGPuG

Ava I

37b

GTCGAC

Sal I

37

CCGG

Hpa I1

CC(E)GG

1(2)

40

6

50b

e

2 PuGCGCPy

GTPyPuAC

Hind11

-70

lob

7.40

10

376

6’

7.9b

6.0

37b

1Ob

7.96

10

5-25

7.5-8.8

20

6(5)

7.9(7.5)

50

100

5- 10

7.5-8.0

10

0

37b

l00b

N GGNCC

Sau96I

30b

Mb

7.4b

6.0

GCNGC

Fnu4Hl

37b

6b

7.90

6.0

CTNAG

Dde I

370

5

7.5

10

606

100

References

-8

GGTNACC

Est PI

55

TGACNNNGTC

Trh I I I I

60-70

GCCNNNNNGGC

Egl I

1

31

30-40(30)

CAAPuCA(N)II GTTPyGT(N),

TthlllII

1

95

65-70

GAAGA(N)B CTTCT(N),

Mbo I1

GGTGA(N), CCAC'UN), GACGC(N), CTGCG(N),o

10

7.56

90

6- 10

7.46

40(20)

9.5

20 mM glycineNaOH

2b

7.5

I0

376

6b

7.96

Hph I

376

106

7.46

Hga I

Vb

6b

7.9b

8.0

100 50-100

o(150)

(61)

120-150

(62) (63, 1 0 )

6.0 10

6.0

(59 )

(60)

6 mM KCIb

(64,6 5 ) (66-68)

Number in parentheses indicates subunits in active complex required for cleavage. General usage; not specified as optimal. In all cases optimal buffer concentration was not specified. Active at 200 m M NaCI. Inhibited by salt concentrations greater than 60 m M ; 10% activity at 100 mM NaCI. Stable to 70" and has been used at 50" with these conditions (unpublished, this lab). For several enzymes more than one set of optional conditions are reported; numbers in parentheses indicate one set of conditions, those without, another set. Hausler, B.,personal communication. ' Key to symbols: *, Indicates methylated base required for cleavage; Pu, purine; Py, pyrimidine; N, specific base not required. a

'

164

R. D. WELLS, R. D. KLEIN, AND C. K. SINGLETON

many instances, (b) the reaction requirements for cleavage by these enzymes are simple in comparison to many other types of enzymes, and (c) virtually all restriction endonucleases are stable for extended periods of time at 37”. As a result, it is unnecessary for some types of studies Gee., cloning or mapping) to rigorously characterize the properties and reaction conditions of these enzymes. Moreover, in many cases the reaction conditions listed as “general usage” in Table I are those utilized during enzyme purification and may not represent conditions that result in optimal activity of the purified enzyme. Table I also lists the molecular weight (MW) of the subunit(s) when this information is available. These data were usually obtained by SDSpolyacrylamide gel electrophoresis. In most cases the number of subunits present in the “active complex” was determined by gel filtration studies on the native enzyme complex. The reaction conditions that have been reported as giving the maximum DNA cleavage rate for a specific enzyme are also listed. These conditions are specified by the authors as optimal and are based on data presented in the references; studies where optimal conditions have not been conducted, are also indicated in the table. For several enzymes more than one set of optimal conditions are reported in the literature. All of the enzymes listed are active at 37” and are routinely used at this temperature even though it may not give a maximum cleavage rate. Optimal temperatures are cited as a single value or range of values in cases where they were determined. All type I1 restriction endonucleases require Mg2+as a cofactor, usually at a concentration near 5 mM. Studies on magnesium dependence of several endonucleases have demonstrated that for each enzyme a characteristic magnesium ion concentration is optimal. For some enzymes, this optimum concentration is related to the overall ionic strength. For example, with XorII and BglI, a higher Mg2+ concentration necessitates a lower concentration of NaCl. Optimal Mg‘+ concentrations are listed in Table I. Nearly all of the enzymes listed are active in the pH range 7.2-7.6. Optimal conditions, based on pH-dependent DNA cleavage studies, are indicated in the manner described above. The dependence of DNA cleavage on buffer concentration has not been studied for any of the enzymes listed. In most cases, Tris-HC1 is used for pH values below 8 . 5 , whereas glycine-NaOH is used to maintain higher pH values. The buffer concentrations listed in Table I are not specified as optimal and are those reported by the authors whose work is cited. There is a large variance in the salt tolerance of restriction endonucleases. This is most evident in the case ofHpaI andHpaII, both of which are isolated from the same organism, H . paruinfluenzae. The activity of

10. TYPE I1 RESTRICTION ENZYMES

165

HpaI is not affected by salt concentrations as high as 200 mM, whereas HpaII is inhibited by salt concentrations greater than 60 mM. Optimal salt concentrations are given in Table I; unusual salt effects are also indicated. Salt-dependent DNA cleavage studies for most enzymes have not been undertaken, and it is likely that many enzymes are currently being used under less than optimal reaction conditions. The references listed are the source of the physical and biochemical data presented in Table I (7-68). Studies concerning the isolation and 7. S. A. Lacks, “Methods in Enzymology,” Vol. 65, p. 138, 1980. 8. A. C. P. Lui, Nucleic Acids Res. 6 , 1429 (1979). 9. J. S. Sussenbach, C. H. Morfoort, R. Schiphof, and E . E. Stobberingh, Nucleic Acids Res. 3, 3193 (1976). 10. R. E. Gelinas, P. A. Myers, and R. J. Roberts,JMB 114, 169 (1977). 11. T. A. Bickle, V. Pirrotta, and R. Imber, “Methods in Enzymology,” Vol. 65, p. 132, 1980. 12. A. H. A. Bingham, T. Atkinson, D. Sciaky, and R. J. Roberts, Nucleic Acids Res. 5, 3457 (1978). 13. G. A. Wilson and F. E. Young, “Methods in Enzymology,” Vol. 65, p. 147, 1980. 14. B. Hinch and M. Kula, Nucleic Acids Res. 8, 623 (1980). 15. L. A. Smith and J. G. Chirikjian, JBC 254, 1003 (1979). 16. J . F. Catterall and N. E. Welker, 1.Bocreriol. 129, 1110 (1977). 17. R. Y.H. Wang, J. G. Shedlarski, M. B. Farber, D. Kuebbing, and M. Ehrlich, BBA 606, 371 (1980). 18. R. A. Rubin and P. Modrich, “Methods in Enzymology,” Vol. 54, p. 96, 1980. 19. K. N. Subramanian, B. S. Zain, R. J . Roberts, and S. Weissman, JMB 110, 297 (1977). 20. M. 9. Mann, R. N. Rao, and H. 0. Smith, Gene 3, 97 (1978). 21. S. P. Lynn, L. K. Cohen, S. Kaplan, J. F. Gardner, J . Bacferiol. 142, 380 (1980). 22. J. L. Hines, T. R. Chauncey, and K. L. Agarwal, “Methods in Enzymology,” Vol. 65, p. 153, 1980. 23. J. L. Hines and K. L. Agarwal, FP 38, 294 (1979). 24. B. S. Zain and R. J. Roberts, JMB 115, 249 (1977). 25. J. G. Sutcliffe and G. M. Church, Nucleic Acids Res. 5, 2313 (1978). 26. K. Murray, S. G. Hughes, J. S. Brown, and S. Bruce, BJ 159, 317 (1976). 27. R. W. Blakesley, J. B. Dodgson, I. F. Nes, and R. D. Wells, JBC 252, 7300 (1977). 28. R. WU, C. T. King, and E. Jay, Gene 4, 329 (1978). 29. S. Bron, K. Murray, and T. A . Troutner, Mol. Gen. Genet. 143, 13 (1975); S. Bron and K . Murray, ibid., p. 25. 29a. S. Bron, and W. Horz, “Methods in Enzymology,” Vol. 65, p. 112, 1980. 30. D. J. Clanton, S. W. Riggsby, and R. V. Miller, J . Bacreriof. 132, 1299 (1979). 31. P. Venetianer, “Methods in Enzymology,” Vol. 65, p. 109, 1980. 32. A. Kiss, 9 . Sain, E.Csordas-Toth, and P. Venetianer, Gene 1, 323 (1977). 33. C. Koncz, A. Kiss, and P. Venetianer, EJE 89, 523 (1978). 34. D. J. McConnell, D. Searcy, and G. Sutcliffe, Nucleic Acids Res. 5, 1979 (1978). 35. R. E. Gelinas, P. A. Myers, G. A. Weiss, R. J. Roberts, and K. E. Murray,JMB 114, 433 (1977). 36. W. Johannssen, H. Schiitte, F. Mayer, and H. Mayer, JMB 134, 707 (1979). 37. L. M. Kunkel, M. Silberklang, and 9. J. McCarthy, J M B 132, 133 (1979).

166

R. D. WELLS, R. D. KLEIN, AND C. K. SINGLETON

purification of these enzymes or determination of their recognition site may be found in several exceIlent reviews (4-6). Sulfhydryl reagents and bovine serum albumin are commonly added to restriction endonuclease reaction mixtures. These factors are not described in Table I. A 1.0-2.0 mM concentration of 2-mercaptoethanol or dithiothreitol is commonly maintained in the buffers during the isolation of restriction enzymes and in the reaction solutions. Only a few studies have reported the necessity of these agents. The absence or presence of 2-mercaptoethanol has no effect on the activity of HpaI and HpaII (22). However, it has recently been shown (Nath and Azzolina, personal com38. R. J. Roberts, P.A. Myers, A. Morrison, and K. Murray, J M B 102, 157 (1976). 39. H. 0. Smith and G. M. Marley, “Methods in Enzymology,” Vol. 65, p. 104, 1980. 40. N. L. Brown and M. Smith, FEBS (Fed. Eur. Biochern. Soc.) Lett. 65, 284 (1976). 41. D. I. Smith, F. R. Blattner, and J. Davies, Nucleic Acids Res. 3, 343 (1976). 42. S. A. Endow and R. J. Roberts, J M B 112, 521 (1977). 43. K. Charter, Nucleic Acids Res. 21, 1989 (1977). 44. N. L. Brown, M. McClelland, and P. R. Whitehead, Gene 9, 49 (1980). 45. J. L. Hines, T. R. Chauncey, and K. L. Agarwal, “Methods in Enzymology,” Vol. 65, p. 153, 1980. 46. R. Watson, M. Zuker, S. M. Martin, and L. P. Visentin, FEBS (Fed. Eur. Biochem. Soc.) Lert. 118, 47 (1980). 47. M. A. Marchionni and D. J. Roufa, JBC 253, 9075 (1978). 48. R. J. Roberts, P. A. Myers, A. Morrison, and K.Murray, J M B 103, 199 (1976). 49. C. D. Tu, R. Roychoudhury, and R. Wu, BBRC 72, 355 (1976). 50. R. J. Roberts, J. B. Breitmeyer, N. F. Tabachnik, and P. A. Myers, J M B 91, 121 (1975). 51. S. Sato, C. A. Hutchinson 111, and J. I. Harris, PNAS 74, 542 (1977). 52. T. R. Gingeras, P. A. Myers, J. A. Olson, F. A. Handberg, and R. J. Roberts, J M B 118, 113 (1978). 53. H. Takahashi, M. Shimizu, H. Saito, Y. Ikeda, and H. Sugisaki, Gene 5, 9 (1979). 54. J. R. Arrand, P. A. Myers, and R. J. Roberts, JMB 118, 127 (1978). 55. S. E. Halford, N. P. Johnson, and J. Grinsted, BJ 179, 353 (1979). 56. J. S. Sussenbach, P. H. Steenbergh, J. A. Rost, W. J. van Leeuwan, and J. D. A. van Embden, Nucleic Acids Res. 5, 1153 (1978). 57. D. W.Leung, A. C. P. Lui, H. Merilees, B. C. McBridge, and M. Smith, Nucleic Acids Res. 6 , 17 (1979). 58. R. A. Makulay and R. B. Meagher, Nucleic Acids Res. 8, 3125 (1980). 59. T. Pugatsch, Nucleic Acids Res. 6, 1429 (1979). 60. T. Shinomiya and S . Sato, Nucleic Acids Res. 8, 43 (1980). 61. Y. H. Lee and J. G. Chirikjian, JBC 254, 6838 (1979). 62. T. Shinomiya, M. Kobayashi, and S. Sato, Nucleic Acids Res. 8, 3275 (1980). 63. N. L. Brown, C. A. Hutchinson 111, and M. Smith, JMB 140, 143 (1980). 64. D. Kleid. “Methods in Enzymology,” Vol. 65, p. 163, 1980. 65. P. A. Sharp, B. Sugden, and J. Sambrook, Biochemistry 12, 3055 (1973). 66. N. L. Brown and M. Smith, PNAS 74, 3213 (1977). 67. H. Sugisaki, Gene 3, 17 (1978). 68. G. N. Godson and R. J. Roberts, Virology 73, 561 (1976).

10. TYPE I1 RESTRICTION ENZYMES

167

munication) that sulfhydryl groups are important for the activity of BumHI, PvuI, SmaI, PstI, HindIII, and A w I , but not for that ofEcoR1, SalI, BglII, HpaI, and SstII. Maintaining sulfhydryl groups in a reduced state may be necessary for the expression of optimal activity of other restriction endonucleases. Many authors and commercial producers of restriction enzymes recommend the addition of bovine serum albumin to a final concentration of 50 to 100 pglml. This stabilizes the enzyme during both storage and DNA cleavage reactions. No studies are available that relate bovine serum albumin concentration to optimal enzyme activity. The variance in the purity of these enzymes probably makes the inclusion of nuclease-free bovine serum albumin in the reaction solution advisable.

111.

Catalytic Properties

A.

KINETICPARAMETERS

Kinetic analyses of several restriction endonucleases have been carried out by various workers. The most comprehensive studies have involved EcoRI, and have been reviewed recently (3, 69a). As will become evident from the discussion of studies on EcoRI and other enzymes, the kinetic parameters and reaction mechanisms differ for the various restriction endonucleases, and are dependent upon reaction conditions and substrates employed. Values for the Michaelis constant for several restriction endonucleases are given in Table I1 [See (70-77)]. The K , values obtained for EcoRI using various substrates under similar conditions range from 3 to 10 nM. Two values that lie outside this range were obtained at a lower reaction 69. H. 0. Smith and D. Nathans, J M B 81, 419 (1973). 69a. W. E. Jack, R. A. Rubin, A. Newrnan, and P. Modrich, I n “Gene Amplification and Analysis, Vol. 1: Restriction Endonucleases” (J. G. Chirikjian, ed.), p. 166. Elsevier-North Holland, Amsterdam 1981. 70. P. Modrich and D. Zabel, JBC 251, 5866 (1976). 71. J . L. Woodhead, N . Bhave, and A . D. B . Malcolm, EJE 115, 293(81). 72. J. L. Woodhead and A. D. B. Malcolm, Nucleic Acids Res. 8, 389 (1980). 73. J. Langowski, A, Pingoud, M. Goppelt, and G. Maass, Nucleic Acids Res. 8, 4727 ( 1980). 74. K . L. Berkner and W. R. Folk, JBC 252, 3185 (1977). 75. P. J . Greene, M. S. Poonian, A. L. Nussbaurn, L. Tobias, D. E . Garfin, H. W. Boyer, and H. M . Goodman, J M B 99, 237 (1975). 76. B. Hinsch, H. Mayer, and M.-R. Kula, Nucleic Acids Res. S, 2547 (1980). 77. B. Hinsch and M.-R. Kula, Nucleic Acids Res. 5, 623 (1980).

TABLE II

KINETICPARAMETERS OF SOME RESTRICTION ENDONUCLEASES Turnover number Enzyme ~~

~

Eco RI EcoRI EcoRI EcoRI EcoRI EcoRI EcoRI EcoRI Barn HI EamHI BamHI Barn H I Hpa I Hpa I

K,(nM)

(rnin-')

~~~

Divalent cation

Temperature ("C)

DNA substrate

37 37 37 37 37 37 12 12 37 37 37 37 25 25

ColEl ColEl ColE 1 ColEl pBR322

~

8 3 3 0.4 5 10

30 7 x 103 0.3 0.36 0.9 170 X 10s 36x 103

4"

8

1 1.8 1.3" 1.5" 4

2.2

-

1.6 0.18 0.03

" Based on the enzyme in a dimer conformation. Linearized plasmid. All other substrates were covalently closed, supercoiled circular DNAs. B represents 5'-bromodeoxyuridine. P. Dwyer-Hallquist and K. Agarwal (personal communication).

A

SV40 d(pT-G-A-A-T-T-C-A) PJCN pJC80 NTP14 pJC80b d(G-G-T-T-A- A-C-C)

d(G-G-T-B-A-A-C-C)'

Reference

169

10. TYPE I1 RESTRICTION ENZYMES

temperature (30 nM), and by using Co'+ instead of Mg2+as divalent cation (0.4 nM). Under the more typical conditions (3T, Mg2+), the K, for

Bum HI is approximately an order of magnitude lower than that for EcoRI (77). However, when kinetic data were obtained for BamHI using a linear plasmid as substrate, unlike the studies above that employed covalently closed plasmid or phage DNAs, a threefold higher value for the K, was obtained (76). This difference in the K, obtained for a covalently closed circular substrate and linear substrate suggests that for BamHI, the substrate conformation is an important factor in the interaction between the enzyme and the DNA molecule. Different reactivities between linear and supercoiled substrates have also been observed in EcoRI reactions (55). Turnover numbers for EcoRI and Bum HI are also given in Table 11. The turnover number for each enzyme is somewhat low, ranging from 1 to 8 min-' for EcoRI and about 2 min-' forBumH1. Modrich has argued that, in the case of EcoRI, the slow turnover is an inherent property of this enzyme and does not reflect a large fraction of the isolated enzyme being inactive (3). Studies discussed below have determined that Bum HI and EcoRI have nonspecific binding affinities for most DNAs. Thus, it would be expected that the kinetic parameters shown in Table I1 would vary somewhat depending on the size of the substrate and number of recognition sites within the substrate. As pointed out by Langowski et ul. (731, the K m and V,,, obtained when using high molecular weight substrates are apparent values that differ from the intrinsic values. However, these authors demonstrate that the overall shape of the rate versus substrate concentration plot remains the same as for a standard Michaelis-Menten mechanism; i.e., the intrinsic K, and V,,, are decreased by the same factor (73).

B. REACTION PATHWAY A scheme outlining the EcoRI reaction pathway as deduced by Rubin and Modrich (78) is illustrated in Fig. 1. Studies on other restriction endonucleases also indicate a similar reaction scheme. Differences seem to be centered mainly on the relative values of k3 and k+, i.e., whether single-strand scission only or double-strand scission occurs before the enzyme -substrate complex dissociates. As discussed below, differences in mechanism at this juncture of the scheme are not only seen between various restriction endonucleases, but are also evidenced for a single enzyme under various conditions. Modrich and co-workers concluded from several lines of evidence that 78.

R. A. Rubin and P. Modrich, Nucleic Acids

Res. 5, 2991 (1978).

170

R. D. WELLS, R. D. KLEIN, AND C. K. SINGLETON

E+I

kl k-i

kz

k8

E . 1 +E.11

kfJl

k4

* E *I11--* E + 111

Lk-.

E+II

FIG.1.

Reaction pathway'ofEcoRI as outlined by Rubin and Modrich (78). I, duplex EcoRI recognition site; 11, the recognition site with one strand nicked; 111, product after double strand cleavage. The importance of and values for certain of the rate constants are discussed in the text.

the rate-limiting step in the proposed mechanism for EcoP.1 is controlled by k4, the release of end product from the enzyme (690). Thus the low turnover number is reflected in a relatively slow dissociation of the cleaved DNA molecule from EcoRI. Values for k, of greater than 40 min-' and k3 of 14 min-' were obtained at 30" (690). In Fig. 1, ifk, is much larger thank+ the result would be that both strand scissions would occur during a single binding step of the enzyme to the substrate. However, if k - 5 is much larger than k3, then an obligatory two-step mechanism would be expected in which dissociation of the enzyme from the substrate would occur after cleavage of a phosphodiester bond in only one of the two strands. Enzyme and singly-cleaved substrate would have to interact once again to effect double-strand cleavage. These two mechanisms are easily distinguished when the substrate employed is a covalently closed circular DNA possessing a single recognition site for the restriction endonuclease under examination. Form I (covalently closed circles), Form I1 (nicked circles), and Form I11 DNA (linear) are well-separated from each other by agarose gel electrophoresis. These types of assays have been carried out tadetermine the relative values 0fk3 and k --5 for several restriction endonucleases. Such studies on EcoRI have resulted in interesting findings. Form I1 DNA, the intermediate in Fig. 1, is not observed when EcoRI interacts with either ColEl or G4 RFI DNA at 37" under steady-state conditions (78, 70). Thus, under these conditions, both strands of the DNA within the recognition site are cleaved prior to dissociation of the Eco RI-DNA complex. Yet, for ColEl DNA, the results are highly dependent on temperature. At temperatures below 15" free Form I1 DNA was detected, and at o", this intermediate accumulated as the predominant form (70). Conflicting results have been obtained when SV40 DNA was used as substrate and analysed in the above manner. Ruben et al. (79) found a large accumulation of Form I1 DNA (- 70% at maximum) during SV40 79. G . Ruben, P. Spielman, C.-P. D. Tu, E. Jay, B. Siegel, and R . Wu, Nucieic Acids Res. 4, 1803 (1977).

17 1

10. TYPE I1 RESTRICTION ENZYMES

DNA cleavage withEcaR1. Their data led them to conclude that dissociation occurs after single-strand scission and before the second strand is cleaved, and thus a two-step mechanism was proposed (79). Rubin and Modrich, on the other hand, found that 75% of the SV40 DNA molecules underwent double-strand scission during a single binding event with EcoRI, indicating a k3/k-, ratio of about 3. These workers concluded that intermediate dissociation was not obligatory (78). Halford and co-workers camed out similar mechanistic studies of EcoRI employing pMB9 and A DNA as substrates (55). Unlike the studies previously discussed in which enzyme concentration was much lower than substrate concentration, Halford et al. used equivalent enzyme and substrate concentrations (within an order of magnitude); these concentrations were below the K , for EcoRI. These authors found an obligatory two-step mechanism (kV5> k 3 , Fig. 1) for both pMB9 and X DNA (55). The mechanism was independent of whether the substrate was linear or a covalently closed circle. The kinetic study of Halford et al. (55) evaluated the rate constants k , and k 2 in the scheme h

i!

Form I --&Form I1 -LForm 111

Although the authors found a two-step mechanism under various ionic and temperature conditions, the absolute values of k and k 2 were highly dependent on the conditions and ranged from 0.045 to 1.05 min-' for k l , and 0.020 to 0.42 for k 2 . Interestingly however, the ratio of k , to k 2 under the various conditions employed was approximately two to one in all cases. The authors suggest that this 2: 1 ratio is statistical in nature and due to the double strandedness of the DNA substrate. In generating Form I1 DNA from Form I DNA, cleavage can occur in one of two strands. However, in converting Form I1 into Form 111 DNA, cleavage must occur in the complementary strand (5.5). Thus, as suggested by Halford et al., binding and cleavage in both steps of the two-step reaction occur at equal rates after the statistical factor is accounted for. From the above discussions concerning a one-step versus a two-step mechanism for EcoRI endonuclease, it is apparent that the actual mechanism observed, and thus the relative values of k 3 and k P 5(Fig. l), is highly dependent on the reaction conditions, the reaction temperature, the nature and concentration of the substrate, and the substrate to enzyme ratio. Studies similar to these have also been carried out for other restriction endonucleases. A two-step mechanism has been demonstrated for Hpa I1 (79, 80), andHindIII (55). However, Sail (55) and MnoI (80)show double-

,

80. B . R. Baumstark, R. J. Roberts, and U. L. RajBhandary, JBC 254, 8943 (1979).

172

R. D. WELLS, R. D. KLEIN, A N D C . K. SINGLETON

strand scission during one binding event. For BumHI, a two-step mechanism has been interpreted in one instance ( / 5 ) ,whereas a one-step mechanism has been proposed under a different set of conditions (55). Thus the Bum HI data again points to the importance of the effect of reaction conditions on the mechanism of restriction endonucleases.

IV.

bnic Strength and Solvent Effects on Enzyme Specificity

A. EFFECTONEcoRI The influence of a variety of reaction conditions on the rates and specificities of restriction endonuclease reactions has been modestly studied. Some of the variables that have been examined include ionic strength, divalent metal ion (magnesium, manganese, cobalt, zinc, etc.), pH, solvent effects (glycerol, dimethylformamide, dimethyl sulfoxide), and enzyme concentrations. One of the most interesting observations is the ability of some enzymes to recognize a subset of their canonical recognition sequences as a function of one or more of these factors. As with most other topics covered in this chapter, the most extensive work in this area has been performed with EcoRI. Polisky et al. (81) were the first to demonstrate that the substrate specificity for EcoRI could be reduced from the canonical hexanucleotide (G-A-A-T-T-C) to the tetranucleotide (N-A-A-T-T-N). The reduction in site specificity occurred when the ionic strength of the reaction medium was kept low, the magnesium chloride concentration was reduced from 5 to 2 mM, and the pH was raised to approximately 8.5. The enzymatic activity that was responsible for cleavage at the N-A-A-T-T-N site was termed EcoRI". It was also pointed out by these workers that the different subset sites (N-A-A-T-T-N) were cleaved at substantially different rates, presumably dependent on differences in the sequence surrounding the AATT core. Other studies (82-84) have elaborated on this basic observation. Addition of managanese chloride to the reaction medium, or replacement of magnesium chloride with manganese chloride, also relaxes the specificity. A more thorough examination (84) of the types of conditions that enhance the EcoRI* activity demonstrated that 10 mM Tris buffer (pH 8.8) plus 81. B. Polisky, P. Greene, D. E. Garfin, B. J. McCarthy, H. M. Goodman, and H. W. Boyer, PNAS 72, 3310 (1975). 82. M. Hsu and P. Berg, Biochemistry 17, 131 (1978). 83. C. M. Clarke and B. S. Hartley, BJ 177,49 (1979). 84. T. I. Tikhonenko, E. V. Karamov, B. A. Zavizion, and B. S. Naroditsky, Gene 4, 195 (1978).

10. TYPE I1 RESTRICTION ENZYMES

173

2 mM managanese chloride was optimum. In addition, the presence of 40 to 50% glycerol, or several organic solvents in concentrations of 1 to 6%, enhanced the relaxation of specificity. A more complete evaluation of the role of divalent metal ions on the EcoRI and EcoRI* activity has been performed (Malcolm et al., personal communication). Cobalt and zinc, which have an ionic radius to charge ratio similar to magnesium, were found to act as cofactors for EcoRI. Malcolmet af. found that a general decrease in sodium chloride concentration and/or an increase in pH caused a stimulation of the EcoRI" activity. The salt concentration or pH value at which EcoRI" activity appeared depended on the cation under study. The relative order of the ability of metal ions to increase EcoRI* activity under these conditions was manganese > magnesium > zinc > cobalt. A detailed evaluation of restriction sites that are recognized by EcoRI* has been recently reported (85). Using the conditions reported previously (81, 82), Woodbury et al. attempted to gain further insight into the nature and selectivity of EcoRI* activity by carefully evaluating the rates of cleavage on 6x174 replicative-form DNA. This genome contains no EcoRI sites but does contain a number of EcoRI* sites. From kinetic analysis of the cleavage reactions it was possible to generate a hierarchy of the double-stranded recognition sequences: G-A-A-T-T-C, the canonical sequence, was most reactive. G-G-A-T-T-Twas the next most reactive sequence (G-G-A-T-T-A and G-G-A-T-T-G were not cleaved), followed by A-A-A-T-T-T and G-A-A-T-T-N (where N= A,T). N-A-A-T-T-N' (N and N' are unsepcified) were cleaved at a lesser rate, and sites having only the central A-T dimer were cleaved the slowest. Studies were also performed in the presence of managanese chloride at moderate salt concentrations and at near neutral pH values; the same heirarchy of cleavage sites was observed. Woodbury and co-workers utilized the cleavage specificity heirarchy to define possible contact sites between various functional groups of the bases within the recognition site and the EcoRI protein molecule. It should be noted that even under standard EcoRI digestion conditions (Table I) using only modest amounts of enzyme, someEcoR1" sites can be recognized as well as the canonical recognition site. Bishop demonstrated that under standard digestion conditions, G-A-A-T-T-A was recognized by EcoRI (86). However, instead of double-strand scission occurrhg at this site, only single-strand scission (specific nicking) occurred. The nick generated in this manner was found on the same strand in all molecules. 85. C. P. Woodbury, 0. Hagenbuchle, and P. H. von Hippel, JBC 255, 11534 (1980). 86. I. 0. Bishop, J M B 128, 545 (1979).

174

R . D. WELLS, R . D. KLEIN, ANDC. K. SINGLETON

B. OTHERENZYMES Heininger et al. (87) showed that the Bsu endonuclease, which cleaves in the middle of the tetranucleotide G-G-C-C, decreased its substrate specificity at high nuclease concentrations. Conditions that enhanced the recognition of the -G-C- sequence were high pH (approximately 8 . 3 , low ionic strength, and high glycerol content. Glycerol, (12%) was found to be particularly effective in generating additional cleavage products. In addition, sodium chloride enhanced the recognition of G-G-C-C by Bsu but inhibited the activity that recognizes the -G-C- sequence. The amount of Bsu that was required to show the first sign of degradation at the -G-Csequence depended strongly on the reaction medium. Considering all factors, the conditions that were most conducive to the formation of additional restriction fragments were 25 mM Tris buffer (pH 8.51, 10 mM magnesium chloride, and 25% glycerol. Under these conditions as little as a twofold excess of Bsu was sufficient to generate additional restriction fragments, whereas under the standard Bsu reaction conditions, reduced specificity was first observed with a 20- to 40-fold excess of the endonuclease. Single-stranded viral DNA was also cleaved by Bsu when these reaction conditions were used (see Section IX). Furthermore, Heininger et af. (87) found that HueIII, which also recognizes G-G-C-C, did not show a reduction in specificity (to -G-C-) under the high pH and glycerol conditions. This is additional evidence that the mechanism of recognition and cleavage may be different for different enzymes, even among isoschizomers (88) which cleave between the same base pairs. The activity of BurnHI can also be influenced by the presence of hydrophobic reagents such as glycerol, dimethyl sulfoxide, ethylene glycol, ethanol, and dioxane (89). The cleavage specificity ofBamHI in the presence of these reagents was distinct from that under normal reaction conditions. Cleavage did not occur at sites that are subsets of the BarnHI recognition sequence (G-G-A-T-C-C). Instead, cleavage occurred at sites completely unrelated to the canonical recognition site. Several lines of evidence indicated that this activity was not due to contaminating endonucleases, and thus, the activity was thought to be an intrinsic property of BarnHI (89). To the best of our knowledge, this is the first report of reagents that actually change the recognition site of a restriction endonuclease instead of simply relaxing the specificity to a subset of the canonical 87. K. Heininger, W. Horz, and H. G. Zachau, Gene 1, 191 (1977). 88. Isoschizomers are defined as enzymes which recognize and cleave within the same sequence. The precise site of cleavage is not necessarily the same. For example, SmaI (CCCkGG) and XmaI (CkCGGG) are considered to be isoschizomers. 89. J. George, R. W. Blakesley, and J . G . Chirikjian, JBC 255,6521 (1980).

10. TYPE I1 RESTRICTION ENZYMES

175

site ( 8 9 ~ )Heininger . et nl. may have observed a similar phenomena with Bsu (87). Another study reported on the alteration of specificity of several restriction endonucleases in the presence of organic solvents (90). The enzymes studied were EcoRI, XbaI, SalI, HhaI, PstI, BamHI, and SstI. However, the enzymes used in this investigation were not characterized as to their degree of purity. Hence, it is not possible to draw rigorous conclusions regarding the modification of site recognition by the enzymes in question; the observed cleavages could be due to contaminating activities. The specificity of BstI (G-&-G-A-T-C-C)is reduced to a BstI* site (NL-G-A-T-C-N) in the presence of large amounts of enzyme (>100 U/pg A DNA) (91). This reduced specificity was enhanced in the presence of glycerol as well as under other conditions that also promoteEcoR1" activity (91). It is not surprising that the kinetics and specificity of restriction endonucleases are altered by the presence of various reagents. The history of nucleic acid enzymology clearly indicates that a variety of other enzymes are influenced in a similar fashion. These include the E. coli and M. luteus DNA polymerases' capacity to incorporate ribonucleotides in place to of deoxyribonucleotides (92, 93), to carry out de novo reactions (W), utilize ribo- versus deoxyribo- templates and primers ( 9 3 , and to incorporate the wrong nucleotides (96). Furthermore, the type of divalent cation used in reactions of terminal transferase (97), E. coli RNA polymerase (98), and RNA ligase (991, influences the substrate specificity of each of these enzymes. Many other cases have also been documented. By selectively altering the specificity of restriction endonucleases it may be possible to increase their utility as reagents, as well as to provide new model systems for the study of DNA-protein interactions. The effect of incubation temperature has not been evaluated for many of the endonucleases (see Table I). However, it has been shown (27) that 89a. J . George and J. G. Chirikjian (personal communication) have recently found that BfiniHl actually cleaves subsites (G-G-A-N-C-C, G-G-N-T-C-C, G-A-A-T-C-C) of its canon-

ical recognition sequence in the presence of the hydrophobic reagents. 90. E. Malyguine, P. Vannier, and P. Yot, Gene 8, 163 (1980). 91. M. C. Catherine and B. S. Hartley, BJ 177, 49 (1979). 92. P. Berg, H. Fancher, and M. Chamberlin, "Symposium on Informational Macromolecules," p. 467. Academic Press, New York, 1963. 93. J. H. van de Sande, P. C. Loewen, and H. G. Khorana, JBC 247, 6140 (1972). 94. J. F. Burd and R. D. Wells, J M B 53, 435 (1970). 95. T. M. Tamblyn and R. D. Wells, Biochemisfry 14, 1412 (1975). %. N. Battula and L. A. Loeb, JBC 249, 4086 (1974). 97. K. Kato, J. M. Goncalves, G. E. Houts, and F. J. Bollum, JBC 242, 2780 (1967). 98. J. S . Krakow, G. Rhodes, and T. M. Jovin, In "RNA Polymerase" (R. Losick and M. Chamberlin, eds.), p. 127. Cold Spring Harbor Lab., Cold Spring Harbor, New York, 1976. 99. D. M. Hinton, J. A. Baez, and R. I. Gumport, Biochemisfry 17, 5091 (1978).

176

R . D. WELLS, R . D. KLEIN, ANDC. K. SINGLETON

Hue111 has remarkable heat stability, being stable at 70" for more than 1 hour with no decrease in activity. The rate of the Hue111 reaction increases approximately threefold from 37" to approximately 75". Even at this high temperature, the canonical -G-G-C-C- is faithfully recognized by Hue111 (27). Temperature may affect other enzymes differently, and thus may provide new insights concerning the mechanism of recognition and specificity. V.

Synthetic Oligonucleotides as Substrates

Studies by Greene et al. (75) demonstrated that the self-complementary octanucleotide d(pT-G-A-A-T-T-C-A)can serve as substrate for EcoRI endonuclease. Cleavage of the octamer required that it be base-paired. The optimum reaction temperature (15") was slightly below the T , of the duplex octamer (17-19"). Although the turnover number for EcoRI with this substrate was comparable to that with other substrates (Table 11), the value of the K , was about 200-fold higher. This discrepancy may reflect the fact that a significant proportion of the octanucleotide may have been single-stranded under the conditions employed. d(pT-G-A-A-T-T-U-A) was found not to be a substrate for EcoRI (75). Dwyer-Hallquist and Agarwal (personal communication) utilized four self-complementary octanucleotides that contain the recognition site for HpaI to investigate the interaction of this enzyme with the recognition site. The nonderivatized oligomer was d(G-G-T-T-A-A-C-C). The other three oligomers contained base analogs: 5-bromodeoxyuridine in place of the second thymidine, uridine in place of this residue, and inosine in place of the second guanosine. The kinetic data obtained using two of the oligomers is given in Table 11. The uridine- and inosine-containing oligomers did not serve as substrates for Hpa I. The bromodeoxyuridinecontaining oligomer did act as a substrate, albeit at a reduced efficiency as compared to the nonderivatized oligomer. Both oligomers had to be in the duplex form for cleavage to occur. The conclusions drawn by these workers were that some nonproductive binding of the bromodeoxyuridinecontaining oligomer to IJpaI was occurring, and that Hpa1 requires the presence of the 5-methyl group of the thymidine adjacent to its cleavage site for binding and cleavage. Synthetic duplex oligonucleotides ranging in length from 6 to 13 base pairs were used by Baumstark er al. in studies with HpaII and MnoI endonucleases (80). Both enzymes recognize the same four base-pair sequence (C-C-G-G). Both were shown to be inactive against single-strand substrates (Section IX). The data obtained in this study demonstrated that

177

10. TYPE I1 RESTRICTION ENZYMES

HpaII and MnoI, although possessing the same recognition sequence, differ in their interaction with, and cleavage of, the duplex substrates used (80). When the duplex 5' d(pG-A-A-C-C-G-G-A-G-A)

I

I

I

I

3' d(T-T -G-G-C-C

I

I

I

I

-T -C-Tp)

I

3' 5'

was employed as substrate, cleavage by HpaII occurred primarily in the upper strand. This is suggestive of a two-step mechanism for this enzyme as previously discussed. In contrast, MnoI cleaved both the upper and lower strands with equal efficiency. Finally, by using various combinations of the oligomers available to these workers, it was demonstrated that both HpaII and MnoI require at least one base preceding the 5' terminus of the recognition site for strand scission to occur. Goppelt et nl. (100) utilized several oligodeoxynucleotides to investigate specific and nonspecific interactions between EcoRI and DNA. The interaction between oligomers (some possessing the EcoRI recognition site and others not) and EcoRI was followed by circular dichroism. Each single-strand oligomer bound to EcoRI in the absence of Mg2+ ions with K ~ 107 = M - 1 . Complex formation with both specific (EcoRI sitepossessing) and nonspecific oligomers gave rise to a decrease of the ellipticity of the protein in the range of the absorption of the peptide bond. Such a change in ellipticity suggested a conformational change was occurring upon complexation with DNA (100). The dependence of complex formation on ionic strength was interpreted by Goppelt ef a!. (100) to correspond to an involvement of two ion-pair bonds between DNA and enzyme. The importance of electrostatic interactions in complex formation between EcoRI and DNA was also supported by nucleic acidcellulose affinity column studies carried out by these authors (100). Several studies on the use of oligonucleotides and dinucleotides as inhibitors of restriction endonuclease activity on high molecular weight substrates have been carried out (Section VII).

VI.

Substituted DNAs

A.

STUDIES WITH

BASE ANALOGS

Early work employing DNA substrates with modified bases to investigate restriction endonuclease-DNA interactions was carried out with 100. M.Goppelt, A. Pingoud, G . Maass, H. Mayer, H. Koster, and R. Frank, EJB 104, 101 (1980).

178

R. D. WELLS, R . D. KLEIN, AND C. K. SINGLETON

EcoRI. These studies have been reviewed (69a); the results are summarized here. The presence or absence of the 2-amino group of dG, a minor groove constituent, has no effect on binding and cleavage by EcoRI at its recognition site (G-A-A-T-T-C) (101).5-Hydroxymethylcytosine was also shown not to affect the restriction of DNA by EcoRI, suggesting that the 5-H of cytosine is unimportant in recognition and subsequent cleavage by this enzyme (102). These workers demonstrated, however, that glucosylation at this position rendered the DNA resistant toEcoRI cleavage, probably because of steric hindrance (102, 103). Berkner and Folk found that the V,,, and K , ofEcoRI were unaltered when thymidine was replaced with uridine (103). However, 5-hydroxymethyluridine substitution lowered the V,,, by 20-fold with no effect on K,, suggesting a stearic or electronic influence of the 5-hydroxymethyl moiety on EcoRI cleavage (103). Berkner and Folk investigated the effects of substituted pyrimidines in DNA on the activity of several restriction endonucleases (104). The studies involved endonucleases that contain A-T base pairs within their recognition site (HpaI, HindII, HindIII, EcoRI, BamHI, and HAEII, Table I) and those that contain only G-C base pairs within the recognition site (Hpa 11, and HhaI). DNA that contains glucosylated 5-hydroxymethylcytosine was resistant to cleavage by each of these enzymes, consistent with findings previously discussed (102, 103). These results suggest that a large group (glucosyl) situated within the major groove interferes with restriction by a number of enzymes (104). DNA that contained 5-hydroxymethyluridine (in place of thymidine) allowed discrimination between the two groups of enzymes that were studied. Endonucleases that contain A-T base pairs within their recognition site cleaved this substituted DNA at much diminished rates. However, no effect was observed on the rate of cleavage by HpaII and HhaI (104). Somewhat more interesting were the results Berkner and Folk obtained with DNA in which thymidine was replaced by uridine (104). No effect was seen on the cleavage rate of EcoRI and BamHI with uridinesubstituted DNA. HpaI, HindII, and HindIII, however, all exhibited a much reduced rate of cleavage with this substrate. All five of these enzymes contain A-T base pairs within their recognition sites, yet it appears that only HpaI, HindII, and HindIII utilize the 5-methyl group of 101. 102. 103. 104.

P. Modrich and R. A. Rubin, JEC 252, 7273 (1977). D. A . Kaplan and D. P. Nierlich, JBC 250, 2395 (1975). K. L. Berkner and W. R. Folk, JEC 252, 3185 (1977). K. L. Berkner and W. R. Folk, JBC 254, 2551 (1979).

10. TYPE I1 RESTRICTION ENZYMES

179

thymidine in the interaction with their sites of recognition, whereas EcoRI and BamHI do not. Several studies have been carried out on the interaction between restriction endonucleases and 5-bromodeoxyuridine-substitutedDNAs. Digestion of DNA that contained 5-bromodeoxyuridine (instead of thymidine) by EroRI, HaeII, HpaI, HpaII, BamHI, and Hind111 was slower than digestion of unsubstituted DNA (47, 104, 10-5). The most sensitive endonucleases with regard to initial velocity were Barn H I , HindII, and HpaI (102). The effects of 5-bromodeoxyuridine may be related to its alteration of the pyrimidine ring electronic properties. Such alterations include the possible influence of stacking, ionization, and tautomeric characteristics of the base (104). Results with HpaII (104) and SmaI (47), neither of which contains thymidine in its recognition site, indicate that the effect of 5-bromodeoxyuridine of altering the cleavage reaction lies outside the recognition site per s e . Marchionni and Roufa found that while the overall rate of cleavage for SmcrI on substituted and unsubstituted DNAs was the same, one of the three SrnaI sites of the DNA under investigation was highly resistant to SrnaI digestion with substituted DNA (47). Mention should be made of work by Petruska and Horn (105) who found that, unlike all other restriction endonucleases examined thus far, MboI activity was enhanced fivefold on 5-bromodeoxyuridine-substituted DNA. This unusual finding was interpreted in a manner that depended on the distance of the nearest thymidine residue from the actual cleavage site (105).

Finally, even though the presence of 5-bromodeoxyuridine in the DNA substrate reduces the rate of cleavage of most restriction endonucleases that have been examined (with the exception of MboI), the reduced rate forEroRI does not mask an earlier interesting finding. Thomas and Davis observed that EroRI exhibited differential rates of cleavage at the five EcoRI sites of A DNA (106). The difference in frequency of cleavage between the most preferred site and the least preferred site was about tenfold. Similar results were obtained by Forsblum et al. in noting a preference by EcoRI for the various recognition sites in adenovirus DNA (107). Marchianni and Roufa also observed the preference for certain sites in A DNA by EcoRI as seen by Thomas and Davis, even though the A 105. J. Petruska and D. Horn, BBRC 96, 1317 (1980). 106. M. Thomas and R . W. Davis, J M B 91, 315 (1975). 107. S . Forsblum, R. Rigler, M. Ehrenberg, U. Pettersson, and L. Philipson, Nucleic Acids Res. 3, 3255 (1976).

180

R. D. WELLS, R . D. KLEIN, AND C. K. SINGLETON

DNA contained 5-bromodeoxyuridine in place of thymidine and the overall rate of cleavage by EcoRI was about one-half that of unsubstituted A DNA (47). Thus, although cleavage was affected by the presence of 5-bromodeoxyuridine, preferential recognition of certain sites was not.

B. METHYLATEDDNAs Many investigations have focused on the interaction of restriction endonucleases with methylated DNAs. Such studies are important in determining the biological relationship between modification and restriction. However, findings on these interactions can also infer mechanistic details in a general way. Cleavage or lack of cleavage at a methylated site by a particular restriction endonuclease indicates whether or not the methylated residue is important in the interaction between the enzyme and its recognition site. Inferences of steric hindrance and/or electronic effects can be made from the results of studies on methylated DNA-restriction endonuclease interactions. The results of a number of such studies are summarized in Table I11 [see (108-1 / 8 ) ] , A fascinating new application of restriction endonucleases involves utilizing isoschizomers, one of which is inhibited by methylation of its recognition site while the other is not, to investigate the distribution and function of methylated bases in eukaryotic DNAs (119, 109, 120). For example, HpaII cannot cleave at its recognition site, C-C-G-G, when the internal C is methylated at the 5 position. Yet, MspI can cleave regardless of the presence or absence of 5-methylcytosine at this position (Table 111). Thus, comparison of the bands produced by MspI and those produced by Hpa I1 indicate the location of 5-methylcytosines within the sequence C-C-G-G. Other pairs of isoschizomers that can be employed in a similar 108. M. B. Mann and H. 0. Smith, Nucleic Acids Res. 4, 4211 (1977). 109. C. Waalwijk and R. A. Flavell, Nucleic Acids Res. 5, 3231 (1978). 110. T. W. Sneider, Nucleic Acids Res. 8, 3829 (1980). 111. P. H. Roy and H. 0. Smith, J M B 81, 427 (1973). 112. A. C. P. Lui, B. C. McBride, G. F. Vovis, and M. Smith, Nucleic Acids Res. 6, 1 (1979). 113. B. Dreiseikelmann, R. Eichenlaub, and W. Wackernagel, BBA 562, 418 (1979). 114. R. E. Streeck, Gene, in press (1980). 115. S. Lacks and B. Greenberg, JBC 250, 4060 (1975). 116. A. P. Bird and E. M. Southern, J M B 118, 27 (1978). 117. A. Dugaiczyk, J. Hedgepeth, H. W. Boyer, and H. M. Goodman, Biochemistry 13, 503 (1974). 118. P. H. Roy and H. 0. Smith, J M B 81, 445 (1973). 119. F. Gautier, H. Biinemann, and L. Grotjahn, EJB 80, 175 (1977). 120. A. Razin and A. D. R i g s , Science 210, 4470 (1980).

18 1

10. TYPE I1 RESTRICTION ENZYMES TABLE I11

RESTRICTIONENDONUCLEASES SHOWN METHYLATED DNA9 Enzyme

Site cleaved

HpaII Hap I1

(C)CGGb C(C)GG GGC(C) G(A)TC

Msp I HneIII Fnu EI MboI Dpn I Sau 3A1 BurnHI BglII Dpn 11 TuqI A I Hha I HaeII EcoRI Bsp I Hind11 Hind111 180

-

G(A)TC' G(A)TC GG(A)TCC AG( A)TCT

-

T(C)GA -

-

Site not cleaved

TO CLEAVE

Reference

C(C)GG C(C)GG (C)CGG GG(C)C G(A)TC GAT(C)"." G(A)TC

-

CPy(C)GPuG G(C)GC RG(C)GCY GA( A)TTC GG(CXCy= GTPyPu(A)C (A)AGCTT

The methylated residues, either 5-methylcytidine or bmethyladenosine, are enclosed in parentheses. Dashes designate not determined. * Half-methylation of the duplex recognition site. Methylation is required for cleavage. Nicking occurs in unmethylated strand. The position of the methylated C is uncertain.

manner to localize methylated residues at other sites can be inferred from Table 111. This approach is a very powerful technique, since information on distribution, tissue specificity, and function of methylated bases in eukaryotic DNAs can be obtained (120). VII.

A.

Inhibitor Studies

REAGENTSTHATMODIFY PROTEINS

Nath and Azzolina (personal communication) investigated the effect of sulfhydryl group inhibitors on the activity of several restriction endonucleases. BamHI was shown to be completely inactivated by 5 , S-dithio-

182

R. D. WELLS, R. D. KLEIN, A N D C . K. SINGLETON

bis(2-nitrobenzoic acid) (DTNB) and p-mercuribenzoate. The latter reagent also completely inactivated Pvu I. DTNB partially inhibited HindIII, PvuI, AvaI, SmaI, PstI, and SstII. p-Mercuribenzoate also demonstrated partial inhibition of AvaI, SmaI, and PstI. No inhibition with either reagent was found with SalI, BglII, and HpaI. Similarly, EcoRI (100) and BglI (61) were shown to be insensitive to sulfhydryl-inhibiting reagents. Thus, the importance of sulfhydryl groups for restriction endonuclease activity appears to be highly dependent upon the particular endonuclease. Methyl acetimidate, at concentrations of 3 to 33 mM was found by Woodhead and Malcolm to inactivate EcoRI endonuclease (72). This compound is specific for primary amino groups (1211, and the inhibition suggests that lysine residues may be important in the interaction of EcoRI with DNA (72). Lysine residues, as well as arginines, have also been implicated in the interaction of BglI with DNA. Modification of the BglI lysine residues by pyridoxal 5'-phosphate resulted in diminished binding and catalysis for this enzyme (61 1. 2,3-Butanedione modification of arginine residues of BglI inhibited the enzyme activity but did not affect the binding properties ofBglI (61). However, no study was performed of the influence of these modifications on the capacity of the modified enzyme to recognize noncanonical DNA sites, nor was there an attempt to determine which amino acid residues within the protein were modified. B. POLYNUCLEOTIDES AS INHIBITORS Woodhead and Malcolm (72) studied the ability of various DNAs to protect EcoRI from methyl acetimidate inhibition. Protection was shown to be independent of the presence or absence of Mg ions. Their investigation demonstrated that EcoRI will bind to DNA that does not contain the EcoRI recognition site. The kinetic analysis of the data led these authors to conclude that nonspecific binding ofEcoRI is 15,000 times weaker than binding to the recognition site (72). Although the finding that EcoRI endonuclease binds nonspecifically to DNA is in agreement with the studies on oligonucleotides (100; see Section V) and other studies discussed below, there is disagreement as to the relative strength of specific versus nonspecific binding. Studies on the nonspecific DNA binding properties of EcoRI have also been carried out by Langowski et al. (73). This work demonstrated that d(G-G-A-A-T-T-C-C), eukaryotic DNA, synthetic alternating copolymers (both duplex and single-stranded) and duplex homopolymers, and tRNA were competitive inhibitors of EcoRI cleavage of pBR322 DNA. Only 121. M. J . Hunter and M. L. Ludwig, JACS 84, 3491 (1962),

10. TYPE I1 RESTRICTION ENZYMES

183

poly(dG) . poly(dC) was found not to inhibit cleavage. The K, values for the various inhibitors ranged from to lo-" M (M nucleotides). This competitive inhibition was interpreted as representing nonspecific binding of EcoRI to DNA and RNA. By comparing results of inhibition between poly(dAT) and d(G-G-A-A-T-T-C-C), the authors concluded that nonspecific binding is about two orders of magnitude weaker than specific binding of DNA by EcoRI (73). The reason(s) for the discrepancy in the strength of nonspecific binding versus specific binding found in this work and in the studies by Woodhead and Malcolm (72) are not apparent at this time. Similar inhibition studies to those above have demonstrated that Bum HI, like EcoRI, possesses nonspecific DNA binding properties (76). Hinschet al. found that BamHI activity on pJC80 DNA was competitively inhibited by nonsubstrate polynucleotides with K, values from > to lop6 M ( M base pairs) (76). Inhibition was seen with polyribo- and polydeoxyribonucleotides. The amount of inhibition shown by a given DNA was dependent on base composition, base pairing, and helix conformation. The nonspecific DNA binding property demonstrated for EcoRI and Bum HI is a typical property for a protein that recognizes DNA at a given target site. Models of DNA-protein interaction for DNA binding proteins generally lead to the conclusion that these proteins possess an appreciable affinity for nontarget sequences leading to nonspecific binding that is important in the overall binding mechanism ( I , 122, 123). Finally, inhibition of Barn HI activity by dinucleotides has been investigated (76, 1244). The results of these two studies are conflicting, and there appears to be no good correlation between the amount of inhibition by a particular dinucleotide and its presence or absence from the BamHI recognition site. Thus, it may be that the enzyme binds the dinucleotides by a mechanism different from that with a true substrate.

VIII.

Influence of Drugs and Other Ligands on Cleavage Specificities

Several investigations have focused on the ability of region-specific DNA binding ligands to preferentially inhibit cleavage at certain restric122. P. H. von Hippel, In "Biological Regulation and Development, Volume I: Gene Expression" (R.F. Goldberger, ed.), p. 279. F'lenum, New York, 1979. 123. T. M. Jovin, Annu. Rev. Biuchem. 45, 889. 124. Y.H. Lee, D. Clanton, and J . G . Chirikjian, FP 38, 294 (1979).

184

R . D. WELLS, R. D. KLEIN, AND C. K. SINGLETON

tion sites. A number of ligands, such as carcinogens, dyes, and antibiotics, are known to bind specifically to DNA. Studies were undertaken in an effort to enhance the specificity of restriction enzyme cleavages, as well as to further understand the mode of interaction of these ligands with DNA. In 1976, Kania and Fanning (125) showed that the binding of 6,4’-diamidino-2-phenylindole(DAPI) to lambda and adenovirus DNAs caused a preferential inhibition ofEcoRI cleavage at some sites but not at others. Also, Nosikov el al. (126) found that distamycin A and actinomycin D preferentially inhibited some of the sites of the following enzymes: EcoRI, EcoRII, HindII, HindIII, HpaI, and HpaII. Distamycin A and DAPI preferentially complexes with A-T-rich regions, whereas actinomycin binds to G-C pairs. van de Sande and his associates (quoted in 1) have studied the capacity of a variety of drugs to inhibit the cleavage at some specific sites by DNA restriction enzymes. The drugs that demonstrated the largest effects were olivomycin and Hoechst 33258. At an appropriate drug/DNA ratio, the cleavage of one EcoRI site on lambda DNA was completely inhibited whereas cleavages at other sites were unaffected. Studies were also performed on HpaI cleavage of PM2 DNA and on HhaI reaction with MI3 replicative form DNA. Due to the relatively low concentration of drug necessary to cause these specific inhibitions, these workers suggested that these findings must be due to differences in the sequences surrounding the canonical recognition sites. A similar approach employed a broader range of drugs and antibiotics (127). The restriction enzymes studied were EcoR1, HindIII and BarnHI. An interesting observation from this work was that the restriction sites located in the center of lambda DNA were most sensitive to the inhibitory action of various intercalators, whereas these interior recognition sites were the least preferred when the intercalators were absent. Thus, the intercalators enhanced the discrimination shown by EcoRI against the interior sites. The binding of cis-dichlorodiammineplatinum(I1)to pSMI plasmid DNA causes a preferential inhibition of PstI cleavage at one of the four PstI sites. This inhibition is believed to be due to the presence of four G-C base pairs that neighbor the Pst1 site (128). 125. J. Kania and T. G. Fanning, EJB 67, 367 (1976). 126. V. V. Nosikov, E. A. Braga, A. V. Karlishev, A. L. Zhuze, and 0. L. Polyanovsky, Nucleic Acids Res. 3, 2293 (1976). 127. K. Nath, Absrr. / l r h Intern. Congr. Biochem. 01-5-HSI (1979). 128. G . L. Cohen, J. A. Ledner, W. R. Bauer, H. M. Ushay, C. Caravana, and S. J. Lippard, JACS 102, 2487 (1980).

10. TYPE I1 RESTRICTION ENZYMES

IX.

185

Cleavage of Single-Stranded DNA Substrates by Certain Restriction Endonucleases

In the early 1970s, it was believed that restriction enzymes only recognized sites on duplex DNA with twofold rotational symmetry. Later it was realized that HueIII specifically cleaved single-stranded DNAs from 6x174, M13, and fl viruses. This was an apparent violation of the “twofold rotational symmetry” notion. This observation was of substantial interest from both the standpoint of the enzymology of restriction endonucleases and of the study of properties of single-stranded viral genomes in solution. Further studies have demonstrated that HueIII plus several other enzymes apparently recognize folded-back duplex regions within the single-stranded viral genomes. The subject of single-stranded DNA cleavage has been recently reviewed (129). The enzymes that have been tested for their capacity to cleave singlestranded DNA substrates are listed in Table IV [See (130-/33)], with HueIII as probably the most thoroughly studied. A substantial body of data (27 and reviewed in 129) reveals that HueIII cleaves duplex DNA approximately 16 times faster than single-stranded viral DNA that contains potential short duplex regions. Although not yet proven, it is also possible that a b u m fide single-stranded G-G-C-C site may be cleaved, though at an even slower rate. One point of discrepancy in Table IV concerns the ability of HpaII to cleave single-stranded DNA. Two studies involving the interaction of chemically synthesized oligonucleotides and single-stranded viral DNAs with HpuII seem to validate the notion that single-stranded substrates are not cleaved, or are cleaved more slowly, by this enzyme (133, 134). Since single-stranded viral DNA has extensive duplex structure (reviewed in I ) , it is curious that such DNA is not a substrate for all restriction endonucleases that cleave replicative form DNA. Even though Hue111 and HhuI recognition sites are 100% G-C, high G-C content in the site is neither a necessary nor sufficient criterion for cleavage. MboI, MboII and HinI, whose sites are less than 50% G-C, cleave single129. s. K . Neuendorfand R. D. Wells, (1980). In “Gene Amplification and Analysis, Vol. I: Restriction Endonucleases” (J. G. Chirikjian, ed.), p. 101. Elsevier, North Holland, Amsterdam, 1981. 130. R. W. Blakesley and R. D. Wells, Nature (London)257, 421 (1975). 131. K . Hoviuchi and N . D. Zinder, PNAS 72, 2555 (1975). 132. 0. J . Yo0 and K . L. Agarwal,JBC 255, 10559 (1980). 133. B . R. Baumstark, R. J . Roberts, and U. L. RajBhandary, JBC 254,8943 (1979). 134. J . L. Hines and K. L. Agarwal, personal communication.

186

R . D. WELLS, R. D. KLEIN, A N D C . K. SINGLETON TABLE IV

RESTRICTION ENDONUCLEASES TESTED ON 4x174,M13 OR F1 VIRALDNAs Enzyme Enzymes that cleave HaeIII Hha I SfaI Hin fI MboI MboII Msp I Bsp RI Enzymes that do not cleave" Ah1 Bum HI EcoRI HaeII Hind11 Hind111 Hpa I HpaIIr Mno I Psr I SalI Sma I

Recognition site

Reference

GGCC GCGC GGCC GANTC GATC GAAGA CCGG GGCC

(/29-131, 6.0 (27, 65, 130) (130) (27, 65) (27) (65) (132) (33 )

AGCT GGATCC GAATTC PuGCGCPy GTPyPuAC AAGCTT GTTAAC CCGG CCGG CTGCAG GTCGAC CCCGGG

(130, 6 5 ) (27) (130, 7S)* (130)

(/.TI), 68) (130) (27)'' (130, 6 s . 131, 27, J33)b (133)b (27) (27) (2 7)

It has been shown that EcoRI (72, 73. 1 0 0 ) and BarnHI (76) bind to singlestranded DNA but d o not cleave. Binding to single-stranded DNA may be a general property of restriction endonucleases. * Tested on synthetic oligodeoxyribonucleotides. The discrepancy of results with HpaII is discussed in the text. K. L. Agarwal, personal communication.

stranded DNA. On the other hand, HpaII, MnoI, and SmaI, whose recognition sequences contain 100% G-C, do not cleave single-stranded DNA. In addition, sites for HaeII, which are a subset of those cleaved by HhaI, are not cleaved in 4x174 viral DNA (Table I V ) . Thus, subtleties in DNA structure or enzyme-substrate specificities, or both, determine if a particular enzyme recognition site in single-stranded DNA is cleaved. Another study (132) indicates that MspI can cleave single-stranded synthetic deoxyoligonucleotides at a slow rate with cleavage dependent on low temperatures and the surrounding nucleotide sequences. Similarly,

187

10. TYPE I1 RESTRICTION ENZYMES

MspI cleaves 4x174 viral DNA at a 27-fold slower rate than the cleavage of duplex 4x174 DNA. Koncz et al. (33) have shown that BspRI endonuclease specifically cleaves 4x174 viral DNA when a 100-fold excess of enzyme is present. BspRI is an isoschizomer of HaeIII, both recognizing G-G-C-C. These workers found that the canonical G-G-C-C sites were cleaved in the viral DNA like they were in the replicative form.

X.

Cleavage of DNA-RNA Hybrids

There has been substantial interest in the possibility of using DNA restriction endonucleases for the specific cleavage of DNA-RNA hybrids. Whereas several laboratories have attempted to evaluate this possibility, the presence of contaminating ribonucleases in the restriction endonuclease preparations gave rise to complications. If restriction endonucleases could specifically cleave both the DNA and RNA strands of DNA-RNA hybrids, this technique would provide a valuable tool for studying gene expression and for RNA sequence studies. Molloy and Symons (135) describe the cleavage of a DNA-RNA hybrid, prepared from cucumber mosaic virus RNA and its copy DNA, by the following restriction enzymes: EcoRI, HindII, SalI, MspI, HhaI, AluI, TaqI, andHaeII1. These investigators showed that the DNA strand of the hybrid was cleaved, at least in some cases, in the same place as for duplex DNA. This cleavage required 20- to 50-fold higher enzyme levels than those required for cleavage of duplex DNA. It was not certain if the RNA strand was cleaved. Alternatively, it is possible that the RNA was randomly degraded to small oligonucleotides since the appropriate determinations were not performed to rule out this possibility. As previously described, two of these enzymes (HhaI and HueIII) cleaved certain single-stranded viral DNAs. However, four of the enzymes (EcoRI, HindII, ScilI, Alrr I) that do cleave the DNA-RNA hybrids do not cleave viral single-stranded DNA. Thus, even though these authors have not studied the fate of the RNA strand, it is likely that DNA-RNA hybrids are, in fact, cleaved. This approach is likely to be used expediently in further studies. Brown (136) has reported that HaeIII will cleave a duplex DNA 5' 3'

. .

. .dG-dG-rC-rC. . . 3'

.

.dC-dC-dG-dG.

..

5'

135. P. L. Molloy and R. H . Symons, Nucleic Acids Res. 8, 2939 (1980). 136. N . L. Brown, FEBS (Fed. Eirr. Eiochern. Soc.) Lett. 93, 10 (1979).

188

R. D. WELLS, R. D. KLEIN, AND C. K. SlNGLETON

that contains a site partially substituted with two ribocytidine residues. However, no kinetic data were provided. XI.

lnsolubilized Restriction Enzymes

To the best of our knowledge, only one paper (137) has appeared that describes the covalent linking of restriction enzymes to an insoluble matrix. These workers used cyanogen bromide to couple Barn HI and EcoRI to Sepharose. Over 90% of the activities were found in an insoluble form, and the insoluble enzymes specifically cleaved lambda, adenovirus, and SV40 DNAs in the same manner as the soluble enzymes. In addition, EcoRI" activity was observed. The thermal stability of both enzymes was increased by at least lo", and the preparations could be lyophilized to dryness without loss of activity. In addition to BamHI and EcoRI, HpaI and TaqI have been carried through this procedure. The use of insolubilized restriction endonucleases has several advantages over standard techniques: the coupled resin can be employed in DNA cleavage reactions in a manner analogous to column chromatographic techniques, reactions can be terminated by rapidly pelleting the resin from solution by centrifugation, and finally, the recovered enzymes can be reused several times. It is quite likely that this technique will receive more attention in further studies. XII.

Crystallization of Restriction Endonucleases

Crystals of EcoRI have been obtained by two different groups. Both groups have obtained crystals of the endonuclease itself, and of the endonuclease complexed with an oligonucleotide that possesses the EcoRI recognition site (138, 139 and Rosenberg, personal communication). The unit cell of the EcoRI crystals contains four enzyme monomers per asymmetric unit (138, 139). At high enzyme concentrations in solution, the EcoRI subunit associates as a tetramer (70), in good agreement with the crystallographic data. The data for the Eco RI-oligomer complex crystals (G-A-A-T-T-C (139) and C-G-C-G-A-A-T-T-C-G-C-G(J. M. Rosenberg, personal communication)) suggests that the unit cell in both instances possesses four enzyme monomers plus two duplex DNA fragments in an 137. Y. H. Lee, R. W. Blakesley, L. A. Smith, and J. G . Chirikjian, Nucleic Acids Res. 5, 679 (1978). 138. J. M. Rosenberg, R. E. Dickerson, P. J . Greene, and H. W. Boyer, J M B 122, 241 (1978). 139. T.-S. Young, P. Modrich, E. Jay, and S.-H. !Grn,JMB, 145, 607 (1981).

10. TYPE I1 RESTRICTION ENZYMES

189

asymmetric unit. Thus, one DNA duplex is assumed to be associated with one enzyme dimer, the active form ofEcoR1 in solution (3). More detailed crystallographic data should prove to be highly enlightening with regard to understanding the interaction between EcoRI endonuclease and its recognition site. XIII.

Genes for Restriction Endonueleases

Only a small number of restriction and modification genes have been cloned to date. These include the genes for HhaII, EcoRI, EcoRII, PstI, and the methylase genes ofBspI, EcoRII, PstI andBsu. Attempts to clone several other restriction endonuclease genes have been futile. It is likely, however, that new approaches to cloning restriction enzyme genes will be made, since such clones will provide useful tools for studying the regulation and expression of these genes as well as being of practical value as a source of the enzymes. The restriction and modification genes of the HhaII system were the first to be cloned. Smith and his associates (20) cloned a three kilobasepair segment of DNA from Haemophilus haemolyticus using the plasmid pBR322 as a vector. Transformed cells carrying the appropriate recombinant plasmid expressed both theHha I1 restriction function as well as the modification function. E. coli cells harboring this plasmid were used for the partial purification of HhaII. Further studies from Smith's laboratory (140, and Hausler and Smith, unpublished) have generated another recombinant plasmid that contains a 2.2 kilobase-pair segment of Haemophilus haemolyricus DNA possessing both the restriction endonuclease and modification methylase genes. The DNA sequence of this insert has been compared to that of the EcoRI gene; no regions of extensive homology were observed (Hausler, Smith and Modrich, personal communication). Modrich and his associates and Boyer et al. have independently cloned 2.2 kilobase-pair segments of DNA that are required for the expression of the EcoRI restriction and modification phenotypes. These DNA segments were sequenced independently using different sequencing procedures. The DNA sequences obtained by these two laboratories are in complete agreement (14/-/43). Since the amino and carboxyl terminal sequences of 140. B. Hausler, FP 39, 946 (1980). 141. A. K. Newman, R. A. Rubin, S.-H. Kim, P. Modrich,JBC, 256, 2131 (1981). 142. R . A. Rubin, P. Modrich, and T. C. Vanaman, JBC, 256, 2140 (1981). 143. P. J. Greene, M. Gupta, H. W. Boyer, W. E. Brown, and J. M. Rosenberg,JBC, 256, 2143 (1981).

190

R . D. WELLS, R . D. KLEIN, AND C. K. SINGLETON

the restriction and modification enzymes have been partially determined (142), these investigators could unequivocally establish the reading frame of the DNA sequence. The DNA sequence indicated that the EcoRI endonuclease consists of 277 amino acid residues, whereas the methylase gene is 326 amino acid residues in length. The genes for the two proteins do not overlap and, in fact, are separated by a 29 nucleotide intercistronic region. Because both the restriction endonuclease and modification enzymes have the same recognition site (G-A-A-T-T-C), it was of interest to compare the amino acid sequences of the two proteins. Both groups concluded that there is a general lack of homology between the two polypeptides, suggesting different evolutionary origins for the two proteins. Moreover, both groups have concluded that, based on circular dichroism measurements and theoretical structural predictions, the two enzymes differ markedly in their secondary structure. Thus, the mechanism of interaction of these two enzymes with the same DNA recognition sequence remains an enigma. A few other restriction endonuclease genes have been cloned with virtually no characterization in terms of restriction mapping or DNA sequence analysis. The genes for EcoRII endonuclease and methylase have been cloned and exist together on a 5.8 megadalton DNA fragment (144). The modification methylase gene of Bacillus sphaericus R has been cloned inE. coli using a pBR322 vector. The gene is carried by a 2.5 kb restriction fragment (145). Though recombinant clones do not exhibit BspI restriction endonuclease activity, the level of methylase activity was found to be higher than in the parental strain (143). Also, the modification and restriction systems of Bsu 12451 and Bsu 124711 have been cloned into Bacillus subtilis 168 (146). The genes for PstI restriction and modification enzymes have also been cloned using the plasmid pBR322 as a vector. The smallest in a recombinant plasmid encoding both the modification and endonuclease genes was approximately 4,000 bp in length (147). When this plasmid was introduced intoE. coli minicells, two new proteins of 33,000 and 35,000 daltons were identified. The 35,000 dalton protein comigrated with the prominent component of a partially purified PstI endonuclease preparation.

144. 145. 146. 147,

V. G.Kosykh, Y. I. Buryanov, and A. A. Bayev, Mol. Gen. Genet. 178,717 (1980). E. Szomolanyi, A. Kiss, and P. Venetianer, Gene 10, 219 (1980). T. Shibata. S. Ikawa, Y.Komatsu, T. Ando, and H. Saito, J E 139, 308 (1979). R. Y. Walder, J. L. Hartley, J. E. Donelson, and J. A. Walder, PNAS 78, 1503 (1980).

10. TYPE 11 RESTRICTION ENZYMES

XIV.

191

Conclusions

Restriction endonucleases have been widely used for gene cloning and mapping, for investigations on gene expression in both prokaryotic and eukaryotic systems, and as models for the study of protein-DNA interactions. It is expected that this usage will be expanded in both academic and industrial directions. Most restriction endonucleases have been meagerly characterized in biochemical terms. Interestingly, however, the characterizations that have been carried out show that these enzymes, although grouped together as site-specific DNA cleavage enzymes, exhibit remarkable diversity. This diversity is evident in both the physical and kinetic properties of each enzyme, as well as whether a particular endonuclease can interact with and cleave methylated DNAs, base-analog containing DNAs, single-stranded DNA's, DNA-RNA hybrids, and noncanonical duplex DNA. Such diversity draws attention to the rich source of biochemical and enzymological knowledge that can be obtained by employing this class of enzymes as both biochemical reagents and subjects of investigation. ACKNOWLEDGMENTS This work was supported by the National Institutes of Health (CA 20279) and the National Science Foundation (PCM 15033). C. K. Singleton was supported, in part, by a fellowship from the American Cancer Society (PF-1904). We also wish to express our appreciation to our friends and colleagues who freely shared their unpublished results with us. Due to the extremely rapid pace of progress in this field, this chapter would be far less complete without their help. We thank R. Roberts for checking the accuracy of Table I.

This Page Intentionally Left Blank

Endonucleases Spec@ for Single-Strunded Poly nucleotides I. ROBERT LEHMAN

I. Introduction . . . . . . . . . . . . . . . . . . . . . . 11. Neurospora crassa Endonuclease . . . . . . . . . . . . A. Physical Properties . . . . . . , . . . . . . , . . . B. Reaction Catalyzed . . . . . . . . . . . . . . . . . C. Biological Role . . . . . . . . . . . . . . , . . . . 111. Nuclease S l . . . . . . . . . . . . . . . . . . . . . . IV. Mung Bean Endonuclease . . . . . . . . . . . . . . . . A. Physical Properties . . . . . . . . . . . . . . . . B. Reactions Catalyzed . . . . . . . . . . . . . . . . . V. Research Application of Single-Strand-Specifc Endonucleases I

1.

. . . . ,

. . . . .

. . . . . . . . . .

. . ,

. . . . . . .

.

. . .

. .

.

. . .

193 194 194 196 197 198 199 199 200 201

Introduction

Endonucleases with a high degree of specificity for polynucleotides that lack an ordered structure have been isolated from a variety of fungi, including Neurospora crassa ( I , 2 ) , Aspergillus oryzae (3, 4 ) , Penicillium cit1. 2. 3. 4.

S. Linn and 1. R. Lehman, JBC 240, 1287 (1965). S . Linn, and I. R. Lehman, JBC 240, 1294 (1965). T. Ando, BBA 114, 158 (1966). V. M.Vogt, EJB 33, 192 (1973). 193 THE ENZYMES,Vol. XIV Copyright 0 1981 by Academic Press, Inc. AU rights of reproduction in any form reserved ISBN 0-12-122714-6

194

I . ROBERT LEHMAN

rinum P (5, 6), and Ustilago maydis (7, 8 ) . Similar enzymes have been purified from mung bean sprouts (9, 1 0 ) and wheat seedlings ( I I ) . Because of their ability to discriminate between single- and double-stranded polynucleotides, these enzymes have become exceedingly useful as reagents for the analysis of nucleic acid structure. Two of these nucleases, from N. crassa and U . maydis, appear to be required for DNA repair and recombination in viva A summary of the properties of the six singlestranded polynucleotide specific endonucleases is given in Table I. The enzymes to be considered in detail are those from N. crassa, Aspergillus oryzae, and mung bean.

11,

Neurospora crassa Endonuclease

A. PHYSICAL PROPERTIES The N. crassa endonuclease was the first of the single-strand-specific endonucleases to be discovered ( I , 2). Though purified extensively from both conidia and stationary phase mycelia (approximately 10,000 fold), the enzyme has not yet been obtained in homogeneous form. As judged by sedimentation in sucrose density gradients, it has a sedimentation coefficient, s20,w, of 3.7 and a molecular weight (assuming a globular structure) of 55,000 (12). The important feature of the enzyme, and that which has made it a useful reagent, is its high degree of selectivity for single-stranded DNA or single-stranded regions within duplex DNA. Although the 10,000-fold purified enzyme preparation attacks native DNA at 2% the rate of denatured DNA, it is clear that hydrolysis of native DNA is due largely to the activity of a contaminating nuclease, which can be selectively inactivated by incubation at 55”, or by brief exposure to thiols. After such treatment, native DNA is attacked at less than 0.1% the rate found with denatured DNA. The partially purified enzyme is active under a variety of conditions including temperatures ranging from 25” to 65”, pHs ranging from 6 to 9 in 5. M. Fujimoto, A. Kuninaka, and H. Yoshino, Agr. Biol. Chem. 38, 777 (1974). 6. M. Fujimoto, A. Kuninaka, and H. Yoshino, Agr. Biol. Chern. 38, 785 (1974). 7. W. K . Holloman, and R. Holliday, JBC 248, 8107 (1973). 8. W. K. Holloman, JBC 248, 8114 (1973). 9. D. Kowalski, W. D. Kroeker, and M. Laskowski, Sr. Biochemistry, 15, 4457 (1976). 10. W. D. Kroeker, D. Kowalski, and M. Laskowski, Sr. Biochemistry. 15, 4463 (1976). 11. W. D. Kroeker and J. L. Fairley, JBC 250, 3773 (1975). 12. S. Linn, “Methods in Enzymology,” Vol. XII, Part A, p. 247, 1967.

TABLE I

PROPERTIES OF SINGLE-STRANDED POLYNUCLEOTIDE-SPECIFIC

Enzyme source

Degree of purity (%)

Molecular weight 55,000

Ratio DNasel RNase

Final products of hydrolysis

PH optimum

>90% Nucleoside

7.0-9.0

N . crassa (1. 2)"

Unknown

(1. maydis

Unknown

42,000

3

>90% Nucleoside 5'-phosphates

8.0

>90

32,000

5

>90% Nucleoside

4.0-4.3

1

A . oropzae (-3, 4 ) P. citrinum (5, 6 ) Mung bean (9. 1 0 ) Wheat seedlings

5.0

Zn" co2+ Mg'+ Mg2+ Ca2+ co2+ Zn" Zn" coz+ Zn2+

5.0

Zn2+

4.8-5.5

Zn2+required for stabilization

5'-phosphates

(7. 8 )

>90

Unknown

0.7

>90

39,000

1.2

>90

43,000

2

5'-phosphates >90% Nucleoside 5'-phosphates >90% Nucleoside 5'-phosphates >90% Nucleoside 5'-phosphates

(11)

Numbers in parentheses are reference numbers.

Divalent cations

ENDONUCLEASES

Inhibitors

Comments

2-Mercaptoethanol. Preferential attack at G potassium and dG residues phosphate, ATP 2- Me rc aptoethano I, Preferential attack at G potassium and dG residues phosphate, ATP

Potassium phosphate, NaF

Possesses 3'-nucleotidase activity; cold labile Possesses 3'-nucleotidase activity Possesses 3'-nucleotidase activity ; preferential attack at A and dA residues

196

I. ROBERT LEHMAN

the presence of a variety of divalent cations, and at ionic strengths ranging from 0.03 to 0.20. Although the enzyme shows some activity in the absence of added divalent cations (20% of maximal), it is strongly inhibited (95%) by EDTA at 0.1 mM. This inhibition cannot be overcome by an excess of Mg'+, but is specifically reversed by Coz+.The enzyme is also inhibited to some extent by potassium phosphate (80% inhibition at 33 mM) and ATP (50% inhibition at 0.5 r n M ) (13). The N. crussu endonuclease hydrolyzes RNA at approximately the same rate as DNA. The same enzyme is responsible for both activities. Thus, the two cochromatograph on hydroxylapatite, have the same pH and temperature optima, show similar rates of heat inactivation, respond in the same way to EDTA and to divalent cations, and have similar specificities for single-stranded polynucieotide substrates.

B. REACTION CATALYZED The N. crussa endonuclease hydrolyzes single-stranded DNA and RNA to a mixture of mono- and oiigonucleotides. As judged by the formation of acid-soluble products, the hydrolysis of single-stranded DNA proceeds in two phases: an initially rapid phase that yields a digest composed of a mixture of 5'-mononucleotides (approximately 30%) in which dGMP predominates, and oligonucleotides of various lengths ranging from dinucleotides to larger than pentanucleotides. At the end of the second phase, mononucleotides comprise approximately 45% of the digest and the remainder is mainly in the form of di- and trinucleotides. More prolonged incubation yields greater than 90% mononucleotides. The N. crussu nuclease can very effectively remove single-stranded regions in double-stranded DNA. This point is illustrated in Fig. 1. In the experiment shown, duplex bacteriophage T7 DNA was treated withE. coli exonuclease I11 to remove from 25 to 5000 nucleotides from the 3' termini of the duplex. Treatment of these DNAs with the N. c r a m endonuclease resulted in an amount of material made acid-soluble that was comparable to that removed initially by exonuclease 111. The value of the slope (0.86) is consistent with the extent of conversion to acid-soluble nucleotide expected under the conditions of the experiment. The N. crussu enzyme is thus able to remove single-stranded stretches in double-stranded DNA to within ten nucleotides of the hydrogen-bonded nucleotides; the value of ten represents the experimental error in the analysis. 13. M. J. Fraser, R. Tjeerde, and K. Matsumoto, Can. J . Biochem. 54, 971 (1976).

I I.

197

SINGLE-STRAND-SPECIFIC ENDONUCLEASES W -1

rn

14

I

I

I

1

I

I

I

3

slope = 0.89

0.1

-

-

-n 0

a

I

0

0

2

1

I

I

I

I

4 6 8 10 12 14 PERCENT MADE ACID SOLUBLE BY EXONUCLEASE 111

16

FIG. 1. Effect of pretreatment of ”P-labeled native T7 DNA with exonuclease 111 on its susceptibility to the N. crussa nuclease [from Linn and Lehman ( 2 ) ] . The abscissa gives the level of made acid-soluble after exonuclease 111 treatment; the ordinate gives the additional made acid-soluble by subsequent treatment with the N. crassu nuclease.

C. BIOLOGICAL ROLE Recent studies by Fraser and his colleagues have suggested that the N. endonuclease represents the end product of a series of proteolytic processing steps beginning with an inactive pronuclease (13, 14). Ronuclease can be purified from fresh mycelia. This form of the enzyme, which has a molecular weight of 88,000 based on sucrose density gradient sedimentation, can be activated by treatment with trypsin or by permitting the mycelial extracts to “age.” In the latter case activation is presumably a consequence of the action of endogenous proteases. The active enzyme, which has a molecular weight of 61,000, shows both endonuclease activity specific for single-stranded DNA and duplex DNA-specific exonuclease activity. Treatment of this form of the enzyme with trypsin or endogenous proteases results in its conversion to the single-strand-specific endonuclease that is devoid of exonuclease activity and has a molecular weight of

crassa

14. S. Kwoong and M. J. Fraser, Can. J . Biochern. 56, 370 (1978).

198

I. ROBERT LEHMAN

55,000 ( I , 2). It therefore appears that the single-strand-specific endonuclease and the double-strand-specific exonuclease activities are associated with a single polypeptide. This polypeptide is released into the culture medium from mycelia during the exponential phase of growth (15). The endo-exonuclease may function in recombination and DNA repair in vivo. In fact, in several respects it resembles therecBC nuclease ofE. coli, an enzyme that is required for normal recombination in this organism (16). Moreover, endo-exonuclease-deficientmutants of N. crussa (uvs-3 and nuh-4) have been isolated that are sensitive to ultraviolet light and to radiomimetic agents, and show an abnormally low frequency of mitotic recombination (17). In these two instances the level of inactive endoexonuclease precursor is higher than the wild type; thus the two mutations may result in the inability of the precursor to undergo conversion to the active form of the enzyme (18). A single-stranded DNA-specific endonuclease with properties identical to the Neurospora enzyme has been purified from the smut fungus Ustilago may& (7). It is also present at reduced levels in an ultravioletsensitive and recombination-defective mutant (8).

111.

Nucleare 51

Nuclease S1 was first identified by Ando (3) in preparations of “Takadiastase” from Aspergillus oryzae and purified some 1000-fold. It has subsequently been purified to near homogeneitv by Vogt (4). The purified protein has a sedimentation coefficient, S20,w of 3.3 and a molecular weight of 32,000. Like the N. crassa endonuclease, nuclease S1 appears to be a metalloenzyme. Dialysis of the enzyme against 1 mM EDTA results in its inactivation. Activity can be largely restored by the addition of either Co2+ or Zn*+. In contrast to the N . crassa enzyme, which has a broad pH optimum in the range of 7 to 9, nuclease S1 shows optimal activity at pH 4 . 6 4 . 3 . It is essentially inactive at pHs higher than 6.0. The enzyme is relatively insensitive to ionic strength; its activity at 0.4 M NaCl is 55% that at 0.1 M, the ionic strength at which it is optimally active. An exception is sodium phosphate, which at pH 4.6 inhibits the enzyme at concentrations as low 15. 16. 17. 18.

M. J. Fraser, Nucleic Acids Res. 6, 231 (1979). P. J. Goldrnark and S. Linn, JBC 247, 1849 (1972). E. Kafer and M. Fraser, M o t . Gen. Genet. 169, 117 (1979). T. Y.-K. Chow and M. J. Fraser, Can. 1. Eiochem. 57, 889 (1979).

11.

SINGLE-STRAND-SPECIFIC ENDONUCLEASES

199

as 10 mM. Although the rate of hydrolysis of denatured DNA is not greatly diminished at high ionic strengths, the extent of hydrolysis is reduced, presumably because of the tendency of DNA to renature under these conditions. Like the N. crassa endonuclease, nuclease Sl hydrolyzes both RNA and DNA; however, it is approximately fivefold more active on DNA than RNA. The products after extensive hydrolysis are nucleoside 5 ' monophosphates. The purified enzyme is highly specific for singlestranded DNA. Thus, under conditions where 96% of denatured DNA from bacteriophage lambda is rendered completely acid-soluble, less than one phosphodiester bond scission is introduced per ten molecules of native lambda DNA. On the other hand, both S1 and N. crassa endonucleases are able to recognize and cleave partially denatured regions in duplex DNA generated by the superhelicity of covalently closed duplex DNAs (19, 20) or by mismatches produced by mutational alterations in a DNA duplex (21). These enzymes can also cleave short single-stranded regions in duplex DNA (2, 22, 23) as well as the nonhydrogen-bonded loops in tRNA (24).

IV.

Mung Bean Endonucleate

A.

PHYSICAL PROPERTIES (3)

The mung bean nuclease has ,een obtainec in near homogeneous form. It has a molecular weight of 39,000 as determined by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate. Sucrose gradient sedimentation and Sephadex G-100 filtration yield values of 43,000 and 42,000, respectively. The enzyme contains one sulfhydryl group and three disulfide bonds. Approximately 70% of the enzyme molecules contain a peptide bond cleavage at a single site in the protein. Thus, reduction of the enzyme with 2-mercaptoethanol prior to gel electrophoresis in the presence of sodium dodecyl sulfate generates, in addition to the intact 39,000 dalton polypeptide two polypeptides of MW 25,000 and 15,000, 19. 20. 21. 22. 23. 24.

P. Beard, J . F. Morrow, and P. Berg, J . Virol. 12, 1303 (1973). A. C. Kato, K. Bartok, M. J . Fraser, and D. T. Denhart, BBA 308, 68 (1973) T. E. Shenk, C. Rhodes, P. W. J . Rigby, and P. Berg, PNAS 72, 989 (1975). J . E. Gerrnond, V. M. Vogt, and B. Hirt, EJB 43, 591 (1974). K. Shishido and T. Ando, BBA 390, 125 (1975). H. Tenehouse and M. J. Fraser, Can. J . Biochom. 51, 569 (1973).

200

I. ROBERT LEHMAN

which are presumably linked covalently by a disulfide bond. No difference in enzymatic activity can be detected between the intact and nicked forms of the enzyme. The mung bean endonuclease appears to be a glycoprotein containing 29% carbohydrate by weight. The mung bean enzyme is rapidly inactivated at pH 5, the pH at which it is optimally active. It can, however, be stabilized by the addition of 0.1 mM Zn2+and 1 mM cysteine or other thiols (glutathione or dithiothreitol). Dialysis of the enzyme at pH 5 results in complete loss in activity. However, activity can be restored by the addition of Zn2+and cysteine. In contrast, dialysis of the mung bean nuclease at pH 5 in the presence of 1 mM EDTA results in a loss of enzymatic activity that cannot be restored by the addition of Zn2+and cysteine; other divalent cations (Co2+,Mg2+, Mn2+,Caz+)are equally ineffective. Presumably a metal ion is removed by the EDTA dialysis that results in irreversible inactivation of the enzyme. B. REACTIONS CATALYZED (4) The mung bean endonuclease catalyzes the hydrolysis of singlestranded polyribo- and polydeoxynucleotides at approximately equivalent rates, to produce 5'-phosphoryl-terminatedmono- and oligonucleotides. It possesses an intrinsic 3'-nucleotidase which also acts on both mono- and oligonucleotides. Comparable endonuclease and 3'-nucleotidase activities are associated with a nuclease isolated from wheat seedlings (I I). Although showing a strong (up to 1000-fold) preference for polynucleotides lacking ordered structure, the mung bean endonuclease is less specific in this regard than the N. crassa or S1 nucleases. Thus, the mung bean endonuclease catalyzes as many as 50 double-strand cleavages in native T7 DNA at levels of enzyme that are required to convert denatured T7 DNA to >90% acid-soluble material. Under conditions that tend to destabilize the DNA duplex (i.e., lower ionic strength and increased temperature, 30" versus 22") native T7 DNA can be degraded completely. An analysis of the products generated under these conditions during the early phases of hydrolysis of native T7 DNA suggests that following an initially small number of endonucleolytic cleavages (possibly at very A-T rich regions), hydrolysis occurs preferentially from the ends of the duplex, generating a mixture of mono-, di-, and trinucleotides. Such a mode of hydrolysis is consistent with the degree of specificity of the mung bean nuclease for single-stranded polynucleotides. Thus, short single-stranded stretches that are formed transiently at the ends of the duplex, particularly under conditions of low ionic strength and elevated temperature, may be cleaved by the enzyme. Reiteration of this process should then lead to complete degradation of the DNA.

11.

SINGLE-STRAND-SPECIFIC ENDONUCLEASES

V.

Research Applications of Single-Strand-Specific Endonucleases

20 1

Because of their high degree of specificity for denatured regions in duplex DNA, all three of the enzymes described have served as useful probes for the identification and, if necessary, elimination of such regions. However, because of its very high degree of specificity for single-stranded as opposed to duplex DNA, its ease of preparation from readily available sources, and its relatively high specific activity the S1 endonuclease has become the enzyme of choice in the structural analysis of DNA and DNA-RNA hybrids. For example, the S1 enzyme has been widely used in DNA and DNA-RNA annealing experiments as a general reagent for the selective removal of nonhybridized, and hence single-stranded polynucleotide. The enzyme has also been utilized in mapping the location of small deletions in viral chromosomes. To cite a specific example, heteroduplex molecules formed from the complementary strands of a deletion mutant and wild-type SV40 DNA contain a single-stranded loop susceptible to the action of S1 nuclease at a position corresponding to the deletion. Incubation of the heteroduplex with S1 nuclease results in hydrolysis of the molecule at the loop, leaving fragments whose length corresponds to the position of the deletion within the SV40 molecule. In fact, by means of this technique S1 nuclease has been used to locate a deletion as short as 32 base pairs in the SV40 genome (21). S1 nuclease has been used widely for the analysis of spliced mRNAs generated as a consequence of intervening sequences in eukaryotic genes (25). The procedure consists of hybridization of unlabeled mRNAs to "P-labeled single-stranded DNA of high specific radioactivity. If the mRNAs are spliced, RNA-DNA hybrids flanked by single-stranded DNA, will result, together with loops of nonhybridized single-stranded DNA, at splice points in the mRNA. When these structures are treated with S 1 endonuclease, the single-stranded DNA is hydrolyzed, resulting in a fully duplex structure with discontinuities in the DNA at the splice points. These duplexes can be resolved and their size determined by electrophoresis through agarose gels.

25. A. J. Berk and P. A. Sharp, PNAS 75, 1274 (1978).

This Page Intentionally Left Blank

Exodeoxyribonucleases of Escherichia coli BERNARD WEISS

1. General Properties . 11. Specific Exonucleases A. Exonuclease I . . B. Exonuclease 111 .

. . . . C. Exonuclease IVA and IVB . D. Exonuclease VII . . . . . . E. Exonuclease VIII . . . . .

1.

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

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

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

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

203 206 206 211 225 226 229

General Properties

Deoxyribonucleases (DNases) catalyze the hydrolysis of phosphodiester bonds in polydeoxyribonucleotides. They are divided into two general classes: exonucleases and endonucleases. As classically defined ( I ), exonucleases hydrolyze linear chains stepwise from the 3' or 5' ends, releasing mononucleotides predominantly, whereas endonucleases attack chains more or less randomly at internal loci to release large fragments initially. Natural phenomena were not so easily categorized, however. DNases were soon discovered that attacked the ends of DNA molecules to release oligonucleotides as well as mononucleotides (Z), or to release 1. M. Laskowski, Ann. N . Y. Acad. Sci. 81, 776 (1959). 2. M. P. Deutscher and A. Kornberg, JBC 244, 3029 (1969).

203 T H E ENZYMES, Vol. XIV Copyright 0 1981 by Academic Press, Inc. All rights of reproduction in any form reserved ISBN 0-12-122714-6

204

BERNARD WElSS

dinucleotides almost exclusively ( 3 ) , or to release high molecular weight fragments initially (4). Accordingly, the terms are redefined as follows: An endonuclease can attack a covalently closed circular strand, but an exonuclease cannot because it requires a chain terminus. If an exonuclease preparation is contaminated with a small amount of an endonuclease, it is capable of rapidly degrading circular DNA molecules. Therefore, the following additional or alternative criteria are frequently used: (i) An exonuclease releases mononucleotides or small oligonucleotides early in the course of a digestion. (ii) An exonuclease usually displays directionality, degrading a strand 3' + 5' or 5' + 3'. (iii) The activity of an exonuclease is often affected by the concentration of free substrate termini in a reaction mixture; it may be enhanced, for example, by shearing the DNA substrate or by partially predigesting it with an endonuclease, or it may be blocked by terminal substituents such as 3'-phosphornonoesters or acetyl groups. It should be noted, however, that the terms exonuclease and endonuclease are merely descriptive and do not represent mutually exclusive categories. Some enzymes, such as exonucleases 111 and V of E. coli, possess both exo- and endonucleolytic activities, i.e., they can at times hydrolyze circular molecules and are capable of initiating hydrolysis from the ends of linear strands. Exonuclease activity is usually assayed by measuring the conversion of an acid-insoluble polynucleotide substrate into acid-soluble mono- or oligonucleotides. This digestion may be followed by measurement of radioactivity (if a radiolabeled substrate is used) or of Azso.The products may be readily identified by chromatography or by electrophoresis. If the substrate is a x2P-labeledDNA, the average chain length of the products may be determined from the proportion of radioactivity released into a Norit-nonadsorbable (i.e., base-free) form by E. coli alkaline phosphatase (a phosphomonoesterase). The 3'- and 5'-mononucleotide products may be distinguished by their susceptibility to 3'- or 5'-nucleotidase. Directionality of attack may be established through techniques introduced by Lehman (S), using terminally labeled substrates. Thus, the DNA may be uniquely labeled at its 3' end with the aid of radiolabeled nucleoside triphosphates and a DNA polymerase, or it may be labeled at its 5' end with the aid of "P-labeled ATP and polynucleotide kinase. The substrate is then incubated with the exonuclease to see if the terminal label is released early or late during the course of digestion. Exonucleases can also be characterized with respect to their tendency 3. D. M . Trilling and H. V. Aposhian, PNAS 60, 214 (1968). 4. E. A. Friedman and H. 0. Smith, JBC 247, 2846 (1972). 5. I. R. Lehrnan, JBC 235, 1479 (1960).

12. E. coli EXONUCLEASES

205

to act in either a processive or in a distributive (random) manner (6). The termprocessive (used in the sense of “going forward”) describes the behavior of a nuclease molecule that binds to a polynucleotide and continues to degrade it without becoming unbound, whereas a distributive enzyme catalyzes a single cleavage first on one chain and then on another at random. Some enzymes are intermediate in behavior. Three techniques have been used to determine processivity . The first two depend on a high molecular ratio of substrate to enzyme in the reaction mixtures. In the first (6), one starts with a homogeneous substrate preparation and then examines the length distribution of molecules in a partial digest. A distributive enzyme yields partially digested molecules with a Gaussian length distribution. A processive enzyme, however, yields a mixture that appears to consist entirely of completely digested and completely intact molecules; partially digested molecules are rare because they are equal in number to the enzyme molecules to which they are bound. By the second method, Thomas and Olivera (7) measured the relative rates of release of nucleotides from the 3‘ and 5’ ends of a substrate, e.g., from [3H]d(pT)~-[32P]d(pT)z used alone or annealed to poly(dA). Because a processive exonuclease tends to degrade a strand from one end to the other before proceeding to a second strand, it appears to release labeled material from both ends of the substrate simultaneously. In the third method, exemplified by the studies of Wu et al. (a), one observes the effect of the addition of a second substrate to an ongoing reaction; a processive enzyme remains bound to the first substrate and continues for a while to degrade it at the previous rate. The interpretation of experiments on processivity and directionality are often dependent on a knowledge of the molecular ratio betweeen substrate and enzyme. At a high substrate-to-enzyme ratio, a processive 3‘ + 5’ exonuclease appears to degrade DNA from 3’ and 5‘ ends simultaneously, as described above. If higher levels of enzyme are used, enough to saturate the ends of the chains, the enzyme appears to be distributive by the first two methods cited above. Because it is rare in most other areas of enzymology to use stoichiometric amounts of enzyme, many investigators do not concern themselves with these considerations. When dealing with exonucleases, however, it is quite possible to inadvertently employ stoichiometric amounts of enzyme in a reaction mixture for the following reasons: (i) Because the substrate can be uniformly, and hence highly, radioactively labeled, it is tempting to use small amounts of it; (ii) the 6. N . G. Nossal and M. F. Singer, JBC 243, 913 (1968). 7. K. R . Thomas and B. M. Olivera, JBC 253, 424 (1978). 8. R. Wu, G. Ruben, B. Siegel, E. Jay, P. Spielman, and C. D. Tu,Eiochemisrry 15, 734 ( 1976).

206

BERNARD WElSS

conversion of a significant percentage of a high molecular weight substrate into an acid-soluble form requires many catalytic events per molecule, necessitating much enzyme; (iii) a very long DNA strand may contain only one binding site for the enzyme, e.g., a 3’-hydroxyl terminus. The following are a few additional observations bearing on the properties of the exonucleases to be described in the sections to follow. (i) With few exceptions, the activity of an exonuclease is a function of the concentration of chain termini (rather than that of DNA-phosphorus), therefore, the K , should be reported in those terms. (ii) Specificity for singlestranded substrates has been customarily determined with denatured DNA, but single-stranded DNA undergoes transient intrastrand basepairing, making it susceptible to double-strand-specific enzymes; a homopolymer substrate is more appropriate as a standard of comparison. (iii) Most, but not all DNases, require bivalent cations (usually Mg2+)for optimum activity, but it is not generally appreciated that the optimum amount is not only a function of the nature of the enzyme but also of the substrate concentration; for that reason, optimum cation concentrations are not described here. II. Specific Exonucleares

Table I outlines some of the distinguishing characteristics of each of the exonucleases of E. coli. They all cleave phosphodiester bonds so as to produce 5’-rather than 3’-phosphomonoesters. In cases where directionality of an enzyme is known, it is always at least 3‘ + 5 ’ . Some of the enzymes have, in addition, a 5‘ + 3’ activity. The exonucleases of E. coli were covered in a review of bacterial DNases by Lehman in 1971 (9). Since then, exonucleases VII and VIII have been discovered, and there have been many new findings with respect to exonucleases 111 and v. The sections below describe only exonucleases I, 111, IV, VJI, and VIII. Separate chapters in this volume deal with exonuclease V (see Linn, Chapter 13 this volume) and with the DNA polymerases (see Section I, Chapters 1-3, this volume). A. EXONUCLEASE I 1. Introduction

Exonuclease I, discovered by Lehman (51, is a 3’ + 5‘ exonuclease that is specific for single-stranded DNA. It attacks a DNA strand processively 9. I. R. Lehman, “The Enzymes,” 3rd ed., Vol. 4, p. 251, 1971.

207

12. E. c d i EXONUCLEASES TABLE I

EXONUCLEASES O F E .coli Enzyme Exonuclease I

Gene sbcB

(Exonuclease 11) Exonuclease 111

Exonucleases IVA and IVB Exonuclease V (recBC DNase)

Characteristics” Single-strand specific, processive; leaves 5’-terminal dinucleotide intact 5’ exonuclease activity of Former name for 3‘ DNA polymerase I (see below) Double-strand-specific; associated activities: DNA-3’-phosphatase, AP endonuclease, and RNase H Single-strand-specific; degrades oligonucleotides completely; not well characterized* ATP-dependent; both 3’ + 5’ and 5’ --+ 3’; double-strand specific, processive; inactive at a nick; releases large oligonucleotides initially; also a single-strand specific endonuclease; a recombination enzyme Former name for 5’ + 3‘ exonuclease activity of DNA polymerase I (see below) Single-strand specific, processive; EDTAresistant; both 3’ -+ 5’ and 5‘ + 3’; oligonucleotide products Double-strand specific; not well characterizedb; product of the cryptic Rac prophage of E. coli K12; detected only in sbcA mutants 3‘ 5‘ activity that is single-strand specific; 5’ 3’ activity: double-strand specific; monoand oligonucleotide products; RNase H activity; repair functions 3’ 5’ only; single-strand specific 3’ + 5’ activity: single-strand specific; does not 3’ activity: singleattack dinucleotides; 5’ strand specific but can attack a duplex after initiating hydrolysis at a single-stranded end --+

xthA

recBC

(Exonuclease VI) Exonuclease VII

xse

Exonuclease VIII

recE

DNA polymerase I

polA

DNA polymerase I1 DNA polymerase Ill

polB p0lC

-+

-

-+

--+

‘I

Unless otherwise noted, directionality is 3‘ + 5’ and products are 5’-mononucleotides.

* Directionality is unknown.

(7), releasing 5‘-mononucleotides,and degrading it down to its 5’-terminal dinucleotide, which it leaves intact ( 1 0 ) (Fig. 1).

2. P~r$cationand Properties Purification methods have relied on the tendency of the enzyme to bind to DNA. In earlier procedures, the enzyme was coprecipitated with DNA 10. 1. R. Lehman and A. L. Nussbaum, JBC 239, 2668 (1964).

208

BERNARD WEISS

FIG. 1. The exonuclease I reaction: processive 3’ 5’ hydrolysis of 5’mononucleotides from single-stranded DNA, sparing the 5‘-terminal dinucleotide. n = number of nucleotides in the original chain. --f

when crude extracts of E. coli B were treated with protamine (5) or with streptomycin ( l o ) , and was then further purified by chromatography. In one method, it was obtained as a by-product of the purification of DNA polymerase I (10). Subsequently, the enzyme has been purified about 1200-fold to apparent homogeneity with the aid of affinity chromatography on denatured DNA-cellulose (1 1 ). The gene for exonuclease I was incorporated into multicopy bacterial plasmids (ColE 1 derivatives) by molecular cloning technology, yielding a strain that overproduced the enzyme 15-fold (12); further amplification may be possible with other cloning vectors. Such strains are preferred sources for the enzyme. Estimates of 70,000 and of 72,000 for the molecular weight of the exonuclease I protomer are based on electrophoresis in SDSpolyacrylamide gels (11, 13). An earlier value of 43,000 obtained by Sephadex gel filtration (see 15) was erroneously low, probably because of retardation of the protein under conditions of low ionic strength (13). The Lehman and Nussbaum method of purification, from E. coli B ( l o ) ,yields predominantly a monomeric protein; that is, the molecular weight of the enzyme as determined under nondenaturing conditions, by gel filtration and velocity sedimentation, agrees with that obtained by SDS-gel electrophoresis for the enzyme that has been denatured and reduced. Another method of purification, from E. coli K12 (/I), yields predominantly a different form of the enzyme, probably a dimer. Although its electrophoretic mobility in SDS is similar to that of the Lehman and Nussbaum preparation, it migrates more slowly than the latter under nondenaturing conditions. Both forms of the enzyme are active ( / I ) . 3 . The Reaction Catalyzed The enzyme is usually assayed by measuring the degradation of radiolabeled, heat-denatured DNA to acid-soluble products (10). Because the glucosylated DNA of bacteriophage T4 is resistant to most other 1 1 . R. K . Ray, R. Reuben, I. Molineux, and M. Gefter, JBC 249, 5379 (1974). 12. D. Vapnek, N. K. Alton, C. L. Bassett, and S. R. Kushner, PNAS 73, 3492 (1976). 13. V. MacKay and S. Linn, BBA 349, 131 (1974).

12. E . coli EXONUCLEASES

209

enzymes (14),its use as a substrate enhances the specificity of the assay (15).The enzyme requires a pH of 9.5 and the presence of Mg2+ for optimum activity. Studies with mutants indicate that with this assay at least 90% of DNase activity measured in crude extracts ofE. coli is due to exonuclease I(15). Exonuclease I has been characterized extensively (5, 10). It has no measurable activity on double-stranded DNA or RNA. An apparent exception is its ability to degrade almost completely some duplexes composed of strands of homopolymers or of an alternating copolymer, such as poly(dA) . poly(dT) or poly(dA-dT). The probable explanation is that the spontaneous breakage and reformation of hydrogen bonds in such molecules causes the creeping or slippage of one strand over another so that a single-stranded 3' end is always presented to the enzyme during digestion. The enzyme, however, is unable to digest poly(dG) * poly(dC), which has a more highly ordered structure. If a duplex substrate contains an unpaired 3' end, the enzyme digests the single-stranded region to within 6 to 8 nucleotides of the base-paired region (16). Exonuclease I attacks a DNA chain only at a free 3'-hydroxyl terminus. It is blocked by 3'-phosphoryl or 3'-acetyl end groups, and its action on strands with 3'-hydroxyl termini is inhibited in the presence of strands with 3'-phosphoryl termini. Oligonucleotides with chain length as low as 3 to 6 are also attacked, but with an estimated I(, about lo6higher than that for denatured DNA. Dinucleotides are resistant to cleavage. 4. Biological Role Exonuclease I is specified by the sbcB locus at 43.6 min (near his) on the genetic map of E. coli (17). The locus is near a major integration site for bacteriophage P2. Consequently, sbcB is among the genes lost by P2 eduction, a deletion of chromosomal genes accompanying the spontaneous curing of P2 lysogens. The fact that the eductants have no growth defect in nutrient media indicates that exonuclease I is dispensible. The proximity of the gene to a prophage attachment site, as well as its superfluous nature, raises the evolutionary possibility of a phage origin for the gene. Mutants lacking exonuclease I have no demonstrable defects in replication, recombination, or repair (15,18). On the contrary, the normal pres14. C. C. Richardson, JBC 241, 2084 (1966). 15. D. M. Yajko, M. C. Valentine, and B. Weiss, JMB 85, 323 (1974). 16. D. Brutlag and A. Kornberg, JBC 247, 241 (1972). 17. B. J . Bachmann and K . B. Low, Microbid. Rev. 44, 1 (1980). '18. S . R . Kushner, H. Nagaishi, A . Templin, and A . J. Clark, P N A S 68, 824(1971); S. R. Kushner, H. Nagaishi, and A. J. Clark, PNAS 69, 1366 (1972).

2 10

BERNARD WEISS

ence of the enzyme in wild-type E. coli has a deleterious effect on a recombination pathway. As reviewed by Clark (19), E. coli has several pathways for general (homologous) genetic recombination. One requires exonuclease V (the recBC gene product) and another requires the recF gene. recBC mutants are recombination-deficient, but this deficiency can be reversed by an sbcB mutation, provided the recF gene is functional (“sbc” is an acronym for “suppressor of recBC”). Presumably the sbcB gene product destroys a DNA intermediate in the recF pathway of recombination, so that the recF pathway does not normally operate in wild-type E. coli. There is a set of exonuclease I-deficient mutants, previously designated xonA (15), that do not suppress the deficiency of recBC mutants in assays for genetic recombination. They do, however, suppress to a variable extent their sensitivity to mitomycin C and to ultraviolet irradiation. Some xonA mutants produce a temperature-sensitive exonuclease I and are, therefore, known to be altered in the structural gene for that enzyme. The xonA and sbcB mutations belong to the same complementation group and are about equally contransducible with the his operon ofE. coli. While it is still possible that the sbcB mutations might be polar mutations affecting separate genes for exonuclease I and for the suppression of recombination deficiency, it is more likely that the xonA mutants are merely leaky sbcB mutants. For that reason, sbcB is the current designation for the exonuclease I gene (17). +

5 . Research Applications Exonuclease I has been used in a sensitive assay for single-strandspecific endonuclease activity. The assay is based on the principle that a single endonucleolytic cleavage in a circular DNA strand renders it almost completely degradable by exonuclease I (20). It can also be used to remove traces of single-stranded DNA in preparations of double-stranded DNA, but for this purpose it is more practical to use the readily obtainable endonucleases from Aspergillus oryzae ( S 1 nuclease) or from Pseudomonas BAL-31. Exonuclease I has enabled the measurement of the average length of single-stranded 3‘ tails in duplex molecules ( 2 / ) ,and its unique ability to degrade the glucosylated DNAs of the T-even phages (after DNA denaturation) has facilitated their chemical analysis (22). 19. 20. 21. 22.

A. J. Clark,Annu. Rev. Genet. 7, 67 (1973). P. Sadowski and J. Hurwitz, JBC 244, 6192 (1969). V. MacKay and S . Linn, JEC 249, 4286 (1974). I. R. Lehrnan and E. A. Pratt, JEC 235, 3254 (1960).

12. E . coli EXONUCLEASES

21 1

Because of its inability to attack a 5'-terminal dinucleotide, exonuclease I has remained an invaluable tool for studying the 5' ends of DNA. Limit digests produced by exonuclease I have been used to isolate and to identify the 5'-terminal dinucleotides in native DNAs (23) as well as those at the new termini created by site-specific endonucleases (24). In a similar manner, the DNA-5'-adenylate intermediate in the DNA ligase reaction was characterized after isolation of terminal deoxydinucleotides covalently bound to AMP (25). The digestion of high molecular weight DNA by exonuclease I is seldom more than 90% complete (10). For digestion up to the terminal dinucleotide, therefore, the enzyme is usually used in conjunction with an endonuclease, such as pancreatic DNase I. The endonuclease bypasses any .factitious 3'-phosphoryl termini and intrastrand duplex regions, and it increases the concentration of substrate termini.

B. EXONUCLEASE I11 1. Introduction

a. Description. Exonuclease I11 is a monomeric protein (MW = 28,000) with four catalytic activities (Fig. Z), namely (it a 3' 5' exonuclease specific for bihelical DNA (exodeoxyribonuclease activity) and (ii) for the RNA strand in an RNA-DNA hybrid duplex (RNase H activity); (iii) it hydrolyzes 3'-terminal phosphomonoesters fiom DNA (DNA-3'phosphatase activity); and (iv) it cleaves DNA endonucleolytically at an apurinic or apyrimidinic site (AP endonuclease activity) creating base-free deoxyribose 5-phosphate end groups. 6 . History. In 1964 Richardson and his colleagues (26) described a DNA phosphatase-exonuclease that was present in partially purified preparations of DNA polymerase I. It was discovered as a protein that enhanced the ability of calf thymus DNA to serve as a primer for the polymerase; it did so presumably by removing inhibiting 3'-phosphoryl end groups from the DNA. The phosphatase and exonuclease activities were copurified and shown to be inactivated by heat at the same rate, suggesting that they were properties of the same enzyme. In addition, the enzyme was shown to degrade a mixed copolymer of ribo- and 23. B. Weiss and C. C. Richardson, J M B 23, 405 (1967). 24. T.J. Kelly and H. 0. Srnith,JMB 51, 393 (1970). 25. I. R. Lehrnan, Srience 186, 790 (1974). 26. C. C. Richardson and A. Kornberg, JBC 239, 242 (1964); C. C. Richardson, I. R. Lehrnan, and A. Kornberg, ibid., p. 251.

212

BERNARD WEISS (a) exodeoxyri bonuclease :

; - I

;

-

_-

_"Ill - *

Ill

I

-.

'

*-

(b) RNase H

(c) DNA-3'-phosphatase

---m ---m (dl A P endonuclease

--

-m - - ---- m--

---P

sQ3--- 5: P

P

P

---P

PJ o $ p J - - -

FIG.2. The reactions catalyzed by exonuclease 111. deoxyribonucleotides, a finding from which it was later assumed that it had RNase H activity (27). The RNase H activity was directly demonstrated in 1972 by Keller and Crouch (28). The laboratories of W.Verly and D. Goldthwait were largely responsible for the initial discovery that E. coli extracts were capable of enzymatic cleavage at apurinic or apyrimidinic sites in DNA. Such sites are places where a base has been removed from DNA by hydrolysis of the glycosylic bond that joins it to a deoxyribose residue, leaving the sugarphosphate backbone intact. Depurination, which can occur spontaneously, is accelerated at elevated temperatures and at low pH. In these early experiments, the preparation of a partially depurinated substrate was facilitated by prior treatment of the DNA with methyl methanesulfonate, an alkylating agent; its major adducts are methylated bases (e.g., 7-methylguanine and 3-methyladenine) with unstable glycosylic bonds. 27. 1. Berkower, J . M. Leis, and J. Hurwitz,JBC 248, 5914 (1973). 28. W. Keller and R. Crouch, P N A S 69, 3360 (1972).

12. E. coli EXONUCLEASES

213

In 1969 Friedberg and Goldthwait (29) described “endonuclease 11,” an activity specific for alkylated DNA, but they did not know which of the many alkylation-produced lesions were specifically recognized by the enzyme. Verly and Paquette (30) reported an “endonuclease for apurinic sites” in extracts ofE. coli. When Hadi and Goldthwait ( 3 / ) found that endonuclease I1 preparations cleaved DNA that had been partially depurinated by acid, Paquette et al. (32) suggested that the two enzymes might be the same. They found that their AP endonuclease could be purified by the method used for endonuclease I1 and that it catalyzed cleavage only at the apurinic (alkali-labile) sites in an alkylated substrate, i.e., the DNA could be cleaved by the enzyme only to the extent that it could be cleaved by NaOH. That this endonucleolytic activity was catalyzed by exonuclease I11 was unsuspected until later work with bacterial mutants. Milcarek and Weiss (33) isolated a set of mutants deficient in exonuclease 111, as measured by the exonuclease and DNA-3’-phosphatase assays. Yajko and Weiss isolated another set deficient in “endonuclease 11” activity (W), as measured by an assay for the breakdown of alkylated DNA. When Marcus Rhoades discovered by chance that an “endonuclease 11” mutant was deficient in exonuclease I11 activity, all the mutants were analyzed and found to be deficient in both activities (34, 35). If one activity was unusually thermolabile in a mutant, so was the other, indicating that they were specified by a common structural gene. The mutations were also corevertible and cotransducible, and the multiple activities were specified by cloned DNA fragment about 3 kilobaae pairs in length (36). The endonuclease, exonuclease, and phosphatase activities were also copurified with a homogeneous monomeric protein (37). c. Nomenclature. From about 1973 to 1978 there was much apparent controversy and confusion in the literature. It stemmed more from conflicting terminology than from conflicting data. The term “endonuclease 11” was initially defined as an enzymatic activity purified by a standard 29. E. C. Friedberg and D. A. Goldthwait, PNAS 62, 934 (1969). 30. W. G . Verly and Y. Paquette, Can. J . Biochem. -50, 217 (1972). 31. S-M.Hadi and D. A. Goldthwait, Biochemisrry 10, 4986 (1971). 32. Y. Paquette, P. Crine, and W. G . Verly, Can. J . Biochem. 50, 1 1 9 9 (1972). 33. C. Milcarek and B. Weiss, J M B 68, 303 (1972). 34. D. M. Yajko and B. Weiss, PNAS 72, 688 (1975). 35. B. Weiss, S . G . Rogers, and A. F. Taylor, In. “DNA Repair Mechanisms” (P. C. Hanawalt, E. C. Friedberg, and C . F. Fox, eds.), p. 191. Academic Press, New York, 1978. 36. S . G . Rogers and B. Weiss, Gene, 11, 187 (1980). 37. B. Weiss, JBC 251, 1896 (1976).

214

BERNARD WEISS

scheme and detected by a standard assay (38); the equation for its reaction could not be written. The assay employed alkylated DNA and probably measured cleavage at apurinic sites (32) due mainly to exonuclease I11 (35). Later, the purification scheme (39) and assay (40) were changed, and “endonuclease 11” was used to refer to an enzyme that cleaved DNA at alkylated residues rather than at depurinated sites, and was clearly separable from exonuclease I11 (40, 41). Hence, there was confusion about the identity of exonuclease 111, endonuclease 11, and Verly’s enzyme for apurinic sites. When the existence of endonuclease I1 (as newly defined) could not be confirmed by its discoverers (42), the term was dropped. Meanwhile, Verly and Rassart (43) had purified to homogeneity the main AP endonuclease of E. coli. Its physical and enzymatic properties were not distinguishable from those of exonuclease I11 (44), which is responsible for about 85% of the AP endonuclease activity in crude extracts (35). Nevertheless, Gossard and Verly (44) referred to it as “endonuclease VI,” mainly to separate it from the controversy that surrounded “endonuclease 11” rather than to imply that it was distinct from exonuclease 111. For simplicity, we should adhere to the principle of one name per enzyme, and it is therefore urged that the term “endonuclease VI” be dropped in favor of the more familiar designation “exonuclease 111.” “Endonuclease VI activity” should therefore be referred to as “the AP endonuclease activity of exonuclease 111.” We will thus follow the convention by which the terms “exonuclease 11” and “exonuclease VI” were abandoned. d. Similar Enzymes. Although Hemophilus injuenzae is distantly related to E. coli, it possesses an enzyme that is similar to E. coli exonuclease 111 with respect to its physical properties and its multiple catalytic activities (45, 46). Streptococcus pneumoniue also contains a phosphatase-exonuclease similar to exonuclease 111, but AP endonuclease and RNase H activities have not been reported for it (47). 38. E. C. Friedberg, S-M.Hadi, and D. A. Goldthwait, JEC 244, 5879 (1969). 39. S-M. Hadi, D. Kirtikar, and D. A. Goldthwait, Biochemistry 12, 2747 (1973). 40. D. M. Kirtikar, G . R. Cathcart, and D. A. Goldthwait, PNAS 73, 4324 (1976). 41. D. M. Kirtikar, G. R. Cathcart, J. G. White, I. Ukstins, and D. A. Goldthwait, Biochemistry 16, 5625 (1977). 42. D. M. Kirtikar, G . R. Cathcart, J. G. White, I. Ukstins, and D. A. Goldthwait, Biochemistry 17, 4578 (1978). 43. W. G. Verly and E. Rassart, JEC 250, 8214 (1975). 44. F. Gossard and W. G . Verly, EJB 82, 321 (1978). 45. J. K. Gunther and S. H. Goodgal, JBC 245, 5341 (1970). 46. J. Clements, S. G. Rogers, and B. Weiss, JEC 253, 2990 (1978). 47. S. Lacks and H. Greenberg, JEC 242, 3108 (1967).

215

12. E. coli EXONUCLEASES TABLE I1

MOLECULARPROPERTIES OF EXONUCLEASE 111" Property Stokes radius, a Sedimentation coefficient, s20,w Frictional ratio,$!!,,, Molecular weight, native enzyme'' Molecular weight, reduced and denatured enzyme'

Value 2.29 nm 2.92 S 1.15

27,400 28,500

From Weiss (37). From sedimentation and diffusion coefficients with an assumed value for the partial specific volume. ' Determined by SDS gel electrophoresis. "

2. Purijicution and Properties Exonuclease 111 has been purified by ordinary methods of solvent or salt fractionation followed by ion exchange chromatography ( 2 6 , 3 7 , 4 3 , 4 8 ) In . one scheme (481, it was obtained as a by-product in the large-scale preparation of DNA polymerase I, from which it was separated by gel filtration chromatography. Two methods yielded preparations that were practically homogeneous by SDS gel electrophoresis (37, 43). Starting with wild-type cells, a 1600-fold purification was required. Currently, the enzyme is most easily purified from a strain that bears the gene for exonuclease 111 on a multicopy plasmid that is thermoinducible for DNA replication (36). Such cells can be induced to overproduce the enzyme 50- to 100-fold and to yield about 1 to 2 mg of pure (>98% homogeneous) enzyme per gram of cell paste after only a 15- to 30-fold purification (49). The physical parameters of exonuclease I11 (37) are summarized in Table 11. According to this data, the enzyme is a globular monomeric protein of MW = 28,000. Verly and Rassart (43) obtained values of 32,000 to 33,000 for the molecular weight of their endonuclease for apurinic sites, based on SDS gel electrophoresis and gel filtration chromatography; these differences are probably within experimental error. 3 . The Reactions Cutcilyzed u . Generul Properties. Rates for each of the four enzymatic reactions have been determined at equivalent unsaturating concentrations of each 48. T. M. Jovin, P. T. Englund, and L. L. Bertsch, JBC 244, 2996 (1969). 49. S. G. Rogers and B . Weiss, "Methods in Enzymology," Vol. 65, p. 201, 1980.

2 16

BERNARD WEISS

substrate and under optimum conditions for each assay (SO). These relative rates of phosphodiester bond cleavage are as follows: exodeoxyribonuclease, 100; AP endonuclease, 37; DNA-3'-phosphatase, 22; and RNase H [on poly(rA) . poly(dT)], 12. The pH optimum for the endo- and exonucleolytic activities is between 7.6 and 8.5 in Tris . HCl buffer (26,37); a lower pH optimum (6.8-7.4) has been reported for the phosphatase activity in phosphate buffer (26). The enzyme is partially active in the absence of added divalent cations, but this activity can be inhibited by EDTA, suggesting that the purified enzyme contains bound metal ions. Mg2+or Mn2+are required for optimum activity; they are about equally effective and can reverse the inhibition caused by EDTA. Ca2+can substitute for Mg2+in the AP endonuclease reaction (SO) but not in the exonuclease or phosphatase reactions. The exonuclease activity normally seen in the absence of added divalent cations is inhibited by the Ca'+. If Ca'+ is used in the AP endonuclease reaction, therefore, one can minimize the exonucleolytic degradation that would normally occur after the endonucleolytic cleavage (50). A similar effect has been reported for citrate in the presence of Mg2+(51). Zn2+ inhibits exonuclease I11 even in the presence of Mg'+ (>90% at lop4M ) (26 ) . Exonuclease I11 is inhibited by the sulfhydryl reagent p-chloromercuribenzoate (50-90% at M). Although dispensible during the standard assay, 2-mercaptoethanol stabilized the enzyme during prolonged incubations (26). b. Exodeoxyribonuclease Reaction. Exonuclease I11 has a 3' + 5' exodeoxyribonuclease activity specific for double-stranded DNA. It releases 5'-mononucleotides from the 3' end of DNA molecules so that partially digested duplexes have projecting 5' tails. The standard assay for the enzyme measures the exonucleolytic release of acid-soluble radioactive material (i.e., nucleotides) from radiolabeled linear DNA duplexes. The enzyme is highly specific for bihelical substrates; it has no detectable activity on single-stranded homopolymers. Because of intrastrand hydrogen-bonding, single-stranded DNA is also a substrate for the enzyme, but it is degraded at only + t o 3 the rate of duplex DNA (33). The glucosyl residues of bacteriophage T4 DNA render it largely resistant to the exonucleolytic activity of exonuclease I11 ( I # ) , whereas it remains a good source of substrate for the endonuclease and phosphatase reactions (33, 34). 50. S. G . Rogers and B. Weiss, unpublished data. 51. S. Ljungquist, B. Nyberg, and T. Lindahl,FEBS (Fed. Eur. Biochem. Soc.)Lerr. 57, 169 (1975).

12. E . coli EXONUCLEASES

2 17

The action of exonuclease III at a 3' terminus is not blocked by a phosphomonoester (which it can hydrolyze through its phosphatase activity), by one or two paired or mismatched ribonucleotides (52),or by up to 3 mispaired deoxyribonucleotides (16). The enzyme can initiate hydrolysis at a single-strand break, but not at the low temperatures and moderate salt concentrations that favor base-stacking across the nick (53). Under standard assay conditions (37), one molecule of the enzyme cleaves 150 phosphodiester bonds per minute. As the digestion proceeds, the duplex portion of the substrate becomes progressively shorter until there is an insufficient number of base pairs to hold the strands together. After 40 to 50% digestion of a duplex DNA preparation, the rate of hydrolysis slows abruptly because the remaining DNA is mostly singlestranded (26). When the substrates are poly(dA) . poly(dT) or poly(dAdT), the kinetics of degradation are different from those seen with naturally occurring DNAs. These synthetic polymers can maintain duplex structures even after 50% hydrolysis because their strands can slip or creep over one another through the transient breakage and reformation of hydrogen bonds. The exonuclease reaction therefore continues almost unabated until almost 100% of the substrate is digested. At 25" a strand can be degraded up to its 5'-terminal dinucleotide; at higher temperatures the limit products are longer 5'-terminal oligonucleotides (-52),probably because of the reduced stability of oligonucleotide duplexes. The mode and extent of degradation are influenced greatly by the incubation temperature. At 5" it is limited; at 23" it is processive; and at 37" it is largely distributive. Thus, at 5", in the presence of 270 mM NaCI, an enzyme molecule removes only 6 nucleotides from a 3' terminus and remains bound to the DNA in a complex that can be isolated by gel filtration chromatography: the addition of more enzyme leads to the removal of 6 more nucleotides by each newly bound enzyme molecule (53). The results suggest that enzyme molecules are bound in fixed positions and have a limited reach, or that their translocation is restricted under these conditions. Donelson and Wu (-5.3) found that the release of mononucleotides was not influenced by the length of the single-stranded 5' tail, and concluded therefore that the binding was not to the very end of that strand. When the temperature of their reaction mixtures or of their isolated enzyme-DNA complexes was shifted to 223", there was unlimited digestion of the polynucleotide substrates. A striking feature of exonuclease 111 is that at 23-28", in the presence of enough enzyme to saturate the DNA termini, the digestion of a homogeR. Roychoudhury and R. Wu, JBC 252, 4786 (1977). 53. J . E. Donelson and R. Wu,JBC 247, 2661 (1972).

52.

2 18

BERNARD WEISS

neous preparation of linear duplexes proceeds synchronously (? 5%) for the first 250 nucleotides (8). At 23" the enzyme is processive for at least 100 nucleotides as determined by an experiment in which the ongoing degradation of radiolabeled DNA by a limiting amount of enzyme was not impeded by the subsequent addition of a large excess of unlabeled substrate (8). At 37", however, the enzyme appears to work in a distributive or nonprocessive fashion; Thomas and Olivera ( 7 ) , using substrates with uniquely labeled 3' and 5' termini, found that exonuclease I11 removed less than 50 nucleotides before dissociating from a chain terminus. Exonuclease I11 may act in a halting or skipping fashion on DNA that is tightly bound to protein. Riley and Weintraub (54) found that when exonuclease I11 digests the 140-base pair DNA of the nucleosome core, it does so in a 3' -+5' direction on each strand, removing 10 nucleotides at a time. It was not determined if they were released as decanucleotides or as mononucleotides. The reaction required larger amounts of enzyme than for the digestion of free DNA, and it appeared to be nonprocessive, or else processive for 510 bases at a time. c. RNuse H Activity. Exonuclease I11 catalyzes the selective degradation of the RNA strand in an RNA-DNA hybrid duplex; the RNA strand is degraded at 10,000 times the rate of the DNA strand (35). 5'ribomononucleotides are released in what is presumably a 3' + 5' exonucleolytic reaction. This activity is distinct from that of the enzyme called RNase H of E. coli, which is an endonuclease, and from the RNase H activity associated with the 5' + 3' exonucleolytic activity of DNA polymerase I. RNase H activity is detected by the release of acid-soluble material (mononucleotides) from a radiolabeled ribohornopolymer in the presence of the complementary unlabeled deoxypolynucleotide, e.g., . poly(dT). There is no measurable activity of exonufrom p~ly([~H]rA) clease I11 on single- or double-stranded RNA molecules. At the time of its discovery, exonuclease I11 was found to be capable of degrading a mixed copolymer containing about one ribonucleotide for every eight deoxyribonucleotides (26). Without direct demonstration, exonuclease I11 was assumed to have RNase H activity (27) and was therefore used by Keller and Crouch (28) as an experimental control in studies of other enzymes with RNase H activity. These latter studies demonstrated that exonuclease I11 could attack poly(rA) poly(dT), break the phosphodiester bonds between ribonucleotides, and release 5 ' mononucleotides exclusively; however, the authors employed excessive amounts of an enzyme preparation of unstated purity and found that the DNA strand was also completely degraded. Roychoudhury and Wu (52) 54. D. Riley and H. Weintraub, Cell 13, 281 (1978).

12. E. coli EXONUCLEASES

2 19

found that the enzyme catalyzes the consecutive cleavage of two ribonucleotides from the 3’ end of a DNA duplex, thus suggesting a 3’ + 5 ‘ exonucleolytic mechanism. Finally, it was demonstrated that the RNase H activity copurifies with the exonuclease, and that the highly purified enzyme preferentially degrades the RNA strand in a hybrid duplex (35). This unusual specificity gives us some insight into how exonuclease 111 and perhaps some other double-strand-specific enzymes recognize their substrates. The simplest way for an enzyme to exercise specificity for a DNA-containing duplex is to recognize a deoxyribose (or even just a 2‘-hydrogen) on one strand and to attack the other strand. Thus, presented with an RNA-DNA hybrid duplex, exonuclease 111 binds to the DNA strand and degrades the RNA strand, but not vice versa (55). d. DNA--?’-phosphatuse Activity (26). 3’-terminal phosphomonoesters are hydrolyzed from DNA by exonuclease 111 to yield 3’-hydroxyl end groups and inorganic phosphate. The DNA is then subject to exonucleolytic degradation by the enzyme. The rate of hydrolysis of 3’phosphomonoesters from duplex DNA is about twice that for heatdenatured DNA. The enzyme has no measurable activity on 3‘phosphoryl-terminated RNA or on d(T-T-Tp), but it attacks a mixed polymer terminated at its 3’ end by ribonucleoside 3’-phosphate residues. Originally the activity was measured with a substrate consisting of uniformly labeled [32P]DNA into which 3’-phosphoryl termini were introduced through partial endonucleolytic digestion with micrococcal nuclease. The assay measured the release of radioactive material in an acidsoluble, Norit-nonadsorbable (base-free) form, i.e., as 34Pi. With crude extracts, this assay is subject to about 15% interference by other enzymes, presumably by other exonucleases coupled with 5’-nucleotidases. An alternative assay, employing a substrate uniquely labeled at its 3‘ termini, is >99% specific for exonuclease I11 in crude extracts (33). e . A P Endonucfease Reaction. Exonuclease I11 cleaves DNA molecules 5‘ to an apurinic or apyrimidinic site. At the newly created termini are nucleoside 3’-hydroxyl end groups and base-free deoxyribose 5-phosphate residues (35). Apurinic sites have been shown to be substrates for the enzyme even after their reaction with NaBH4 @), a treatment that reduces the free aldehyde groups on the base-free sugars. The enzyme has no measurable endonucleolytic activity on free DNAs lacking AP sites. It does, however, make endonucleolytic cleavages in chromatin 55. Curiously, this mechanism was presented initially only for the purpose of simplifying a figure (Ref. 37 and Fig. 3, Section II,B,3,f). It was later taken seriously, and its chief prediction, namely, the sparing of the DNA strand in the hybrid duplex, was tested and verified.

220

BERNARD WElSS

DNA at a point where the 5' end of a strand projects beyond the nucleosome core (54, 5 6 ) . The reaction is demonstrated most specifically by the cleavage of partially depurinated supercoiled DNA. 4x174 RFI VHIDNA (covalently closed circular duplexes) is treated with acid to introduce about 2 apunnic sites per molecule. The assay substrate is rapidly renaturable, but nicking at AP sites by exonuclease I11 permits the strands to remain separated after a mild denaturing treatment and so to be adsorbed to nictrocellulose filters (46). Other assays have measured the degradation of partially depurinated linear duplexes, either by the release of large fragments from DNA entrapped in a polyacrylamide gel (38)or by the release of acidsoluble material (4.7); they are less specific because they measure exonucleolytic degradation as well. A P sites have been introduced into these substrates by treatment of the DNA with alkylating agents or acid (43) or by the treatment of uracil-containing DNA with uracil-DNA glycosylase (46). Concomitant exonucleolytic degradation has been reduced by using glycosylated T4 phage DNA, citrate, or Ca'+ in the reaction mixture (see a , this section). f. Reaction Mechanisms. The relatively low molecular weight of exonuclease I11 suggests that a single active site may be responsible for all of its catalytic functions. According to the common-site model (Fig. 3), the enzyme contains three regions. One region recognizes adeoxyribose on the strand opposite the one that is cleaved, thus conferring a specificity for DNA-DNA or DNA-RNA duplexes. A second region recognizes and cleaves phosphoester bonds. A third region recognizes a topological feature common to all of the substrates, namely, a space created by a missing or displaced base. In the endonuclease substrate, the space is at an AP site. In the phosphatase substrate, the space is created by the missing terminal nucleoside. In the exonuclease substrates, an interstrand space is created by the transient unwinding that occurs commonly at ends of duplexes because the terminal bases are less constrained by stacking forces than is the rest of the DNA. Although alternatives are not precluded, the following observations are consistent with the common-site model: (i) Unlike other enzymes that work at 3' ends (such as DNA polymerases or exonuclease I), the enzyme does not require a free 3'-hydroxyl group, but attacks 3'-phosphorylterminated DNA; it therefore probably recognizes some other topological feature. (ii) Similarly, the enzyme does not recognize the chemical group 56. Perhaps, during the binding of DNA to histones, some nucleoside residues become unpaired and are rotated away from the central axis of the double helix to leave spaces that resemble AP sites (see Section II,B,3,D.

22 1

12. E . coli EXONUCLEASES

( A ) Enzyme:

(C)3' - Phosphatase

3 recognition sites

(B)Endonuclease

(D) Exonuclease

FIG. 3. The common-site model for exonuclease 111. The enzyme (shaded area) is shown attacking duplexes, cleaving phosphoester bonds at the arrows. The enzyme (A) is pictured as having three regions. One region recognizes and cleaves phosphoester bonds; a second region recognizes the duplex structure by recognizing a deoxyribose on the strand opposite the one that is cleaved. The thud region recognizes a space created by (B) a missing base, (C) a missing 3'-terminal nucleoside, or (D) terminal unwinding. From Weiss (-37), with permission. unique to an AP site, i.e., a free aldehyde; it cleaves at an AP site even if the aldehyde is reduced (44). (iii) The enzyme does not care whether a purine or pyrimidine is missing at the AP site (351,so it probably recognizes the space itself rather than some feature of the unpaired base. (iv) Although the exonuclease is specific for duplex substrates, it removes from one to three mispaired nucleotides from the 3' terminus (16). (v) The enzyme does not attack DNA at single-strand breaks at 5" in 70 mM NaCl ( 5 3 , conditions that favor the stacking of bases across the nick. (vi) In all the exonuclease I11 mutants that have been studied (12 of independent origin), the exonuclease, phosphatase, and endonuclease activities are similarly affected (33-35, 57). (vii) All the reactions produce new 3'57. B.

(1976).

J. White, S. J. Hochhauser, N. M.Cintron, and B. Weiss,J.

Bacteriol. 126, 1082

222

BERNARD WEISS

hydroxyl termini and result from cleavages 5' to the recognized moiety whether it be a terminal nucleotide, a terminal phosphate, or a base-free sugar (35). The model, therefore, correctly predicted that the endonucleolytic cleavage would be 5' to the AP site. An alternative model that had been first considered stated that an enzyme with two catalytic sites had evolved to mediate sequential steps in an excision-repair pathway. According to that model, however, endonucleolytic cleavage should have been 3' to the AP site so that subsequent 3' + 5' exonuclease activity could remove the lesion (i.e., the base-free sugar). In its most general form, the common-site model implies the cleavage of unpaired nucleotides that lie adjacent to paired regions. The following molecules, however, are not cleaved endonucleolytically although they contain such regions: (i) circular single-stranded 4x174 DNA, which has both relatively stable and transient bihelical regions (58); (ii) heteroduplexes of mutant A phage DNAs that contain unpaired regions ( 5 0 ) ; (iii) covalently closed circles of the alternating copolymer poly(dA-dT) (59); and (iv) duplexes with long, single-stranded 3' ends (16). From the terminal unwinding hypothesis, we might also expect that di- and trinucleotides would be released, in addition to mononucleotides, during exonucleolytic digestion, but such products have not been found (26). g. Biological Role. Exonuclease I11 is an abundant and very active enzyme in crude extracts of E. coli; there are about 3500 molecules per cell. Its structural gene, xthA, is located at 38 min on the chromosomal map (17,57) and is expressed constitutively. Exonuclease I11 production is not affected by exposure of the cells to mitomycin C (an alkylating agent) or by mutations in the gene for dUTPase (58), both of which result in the production of AP sites. From the study of xth mutants, we know that under the standard assay conditions exonuclease I11 accounts for about 85-90'36 of the AP endonuclease activity in crude extracts ofE. coli (34,35, 60). Within the cell, its exonuclease activity may be second only to that of exonuclease V, an ATP-requiring enzyme, although the relative activity of the latter under physiological conditions is uncertain (see Linn, Chapter 13, this volume). Most of the residual AP endonuclease activity in xth mutants appears to be due to endonuclease IV (60, 61). In crude extracts of Hemophilus influenzae, exonuclease I11 is also the major AP endonuclease and ATP-independent exonuclease, and this AP endonuclease activity far exceeds the restriction endonuclease activity (46). 58. A. F. Taylor and B. Weiss, unpublished data. 59. P. Modrich and I. R. Lehman, JBC 245, 3626 (1970). 60. S. Ljungquist, T. Lindahl, and P. Howard-Flanders,J . Bacteriol. 126, 646 (1976). 61. S. Ijungquist, JBC 252, 2808 (1977).

12. E. coli EXONUCLEASES

223

Tight xth mutants have the following abnormalities: (i) a slightly increased sensitivity to alkylating agents (34, 41, 62); (ii) an increased level of genetic recombination, or hyper-Rec phenotype (63);and (iii) an inability to tolerate a dut (dUTPase) mutation (35). These defects can be explained by their loss of AP endonuclease activity. The mutants are normal with respect to their growth rate, their sensitivity to ultraviolet irradiation, and their mutation frequency (331. The hypothesis that the exonuclease activity of exonuclease I11 was important in DNA repair was attractive because its properties were ideal for this purpose. The enzyme degrades only one strand in a given region, leaving an intact template opposite a 3‘-hydroxyl primer terminus. Because of its ability to remove mismatched nucleotides, it was implicated in the “proofreading” of newly synthesized DNA (16). There is no evidence for this assumption; xrh mutants do not have high mutation rates. The enzyme is clearly nonessential; a deletion mutant is viable (57). It is nevertheless possible that the exonuclease activity of the enzyme may have an important function for which other exonucleases can substitute; however, xth mutations have at least been transduced into a polAl mutant (deficient in the 3’ + 5‘ exonuclease activity of DNA polymerase I) and into an sbcB-recBC mutant (deficient in both exonucleases I and V) without noticeable effect (57). The effects of enzyme overproduction are also unremarkable. A strain bearing a multicopy xth plasmid grew normally despite a constitutive 30-fold overproduction of the enzyme (36). Moreover, a thermoinducible plasmid yielding a 120-fold overproduction could be isolated mostly intact after induction (36), suggesting that even at these levels of intracellular exonuclease activity, there is no significant degradation of the plasmid DNA. These findings suggest that there may be few, if any, free 3‘ ends in intracellular DNA; such termini may normally exist only within replication or topoisomerase complexes. The DNA-3’-phosphatase function of the enzyme was thought to be of possible importance (26) because 3’-phosphoryl-terminatedDNA is not a substrate for the DNA polymerases of E. coli, and because exonuclease I11 is responsible for >99% of this phosphatase activity in cells under normal growth conditions (33, 64). In E. coli, however, no enzymes are currently known to produce 3’-phosphoryl end groups in DNA. Similarly, 62. The apparent marked sensitivity of one mutant (Ref. 3 4 ) was due in part to a second uncharacterized mutation that was not cotransducible withxthA (B. Weiss, unpublished). A deletion mutant was not very sensitive to methyl methane sulfonate (Ref. 41). 63. J. Zieg, V. F. Maples, and S. R. Kushner, J . Barterid. 134, 958 (1978). 64. Alkaline phosphatase, which can also catalyze this reaction, is normally repressed except during growth in low-phosphate media.

224

BERNARD WElSS

it is hard to think of a use for the RNase H function. It would not etlectively degrade an RNA Drimer in DNA synthesis because it attacks the wrong end of the strand, and it should not selectively remove misincorporated ribonucleotides because its DNase activity is greater. Exonuclease I11 does, however, appear to be important in the repair of DNA that contains uracil residues or AP sites (see Friedberg, et nl., Chapter 13, this volume). AP sites can occur not only from spontaneous depurination but also from the action of glycosylases that specifically hydrolyze unusual bases, such as uracil, from DNA. AP sites are repaired through the sequential action of AP endonucleases, DNA polymerase I (via its 5‘ -+ 3’ exonuclease and its polymerase activities), and DNA ligase. Thus, a nonlethal dur (dUTPase) mutation, which enhances the misincorporation of uracil into DNA, becomes conditionally lethal in the presence of nonlethal mutations affecting exonuclease 111(35),DNA polymerase I (65),or DNA ligase ( 6 5 ) . The dut-xth double mutant is inviable, probably because of the accumulation of lethal AP sites. An ung (uracil DNA glycosylase) mutation restores its viability, presumably by blocking the removal of uracil and hence the formation of AP sites (35).While these results indicate that exonuclease 111 is essential for the efficient repair of AP sites in DNA, and while they imply that it is the AP endonuclease activity of the enzyme that is essential, they do not rule out the possibility that it is the exonuclease activity that is important (66). The basis for the hyper-Rec phenotype ofxth mutants is unknown. Enhanced recombination is not seen in assays that measure only the frequency of incorporation of foreign DNA during transduction, conjugation, or transformation. It has been seen only in an assay for recombination between two distant but homologous chromosomal regions (63). The hyper-Rec phenotype might be due to decreased degradation of recombination intermediates (as in the case of exonuclease I mutations) or to the production of recombinogenic lesions (perhaps AP sites) (67). 4. Research Applications

DNA that has been partially digested with exonuclease I11 is an ideal primer for most DNA polymerases and has been widely used for this 65. B-K. Tye and I. R. Lehman, J M B 117, 293 (1977). 66. One theory, for example, is that the exonuclease exerts an “anti-ligase” function; it would create a gap to prevent premature sealing of the incision site by DNA ligase, thereby giving DNA polymerase I more time to work. There is, however, no evidence that DNA ligase would catalyze a reaction involving a base-free sugar. 67. By analogy, dut mutants, which generate AP sites transiently, are also hyper-Rec. See S. J. Hochhauser and B. Weiss, J . Bacteriol. 134, 157 119781.

12. E . coli EXONUCLEASES

225

purpose. The enzyme can be used to create gaps in DNA at specific sites, such as restriction enzyme cleavage sites. These gaps can be repaired with triphosphates that contain base analogs or they can be treated with a mutagen, such as bisulfite (681, that is specific for single-stranded regions. The result is localized mutagenesis of a specific region of a genome. If the gaps are repaired by DNA polymerase and labeled nucleoside triphosphates, the region can be subjected to sequence analysis. Wu and his colleagues described methods for controlled synchronous exonucleolytic degradation that can be used to determine the nucleotide sequence for a limited distance at the 3' ends of DNA (8, 53). Exonuclease I11 has also been used in conjunction with a single-strand-specific endonuclease (S 1 nuclease) to generate small deletions at a restriction site in a cloned gene (69).

The AP endonuclease activity of the enzyme can be used to detect AP sites in DNA molecules and to distinguish them from sites of ribonucleotide misincorporation, which are also alkali-labile. In conjunction with other DNases, exonuclease I11 can also be used to identify, and to isolate, specific protein binding sites in a DNA molecule; the bound protein partially protects such sites from nucleolytic degradation (54).

c.

EXONUCLEASES IVA AND IVB

1. Introduction

Exonucleases IVA and IVB catalyze the degradation of oligonucleotides to nucleoside 5'-monophosphates. Our knowledge of them is limited to the information in one publication, by Jorgensen and Koerner (70), which was based on work with relatively impure fractions.

2 . Purijication and Properties Extracts of E. coli B were subjected to ammonium sulfate fractionation followed by DEAE-cellulose chromatography, and two separately eluting peaks of DNase activity were found. Although their catalytic properties were indistinguishable, they were still eluted at different salt concentrations when rechromatographed on DEAE-cellulose. By this method, 68. D. Shortle and D. Nathans, PNAS 75, 2170 (1978). 69. S . Sakonju, D. F. Bogenhagen, and D. D. Brown, Cell 19, 13 (1980). 70. S . E. Jorgensen and J. F. Koerner, JBC 241, 3090 (1966).

226

BERNARD WEISS

exonuclease IVA was purified 120-fold and exonuclease IVB was purified about 30-fold. 3. Catalytic Properties The exonucleases have a broad pH optimum, between 8.0 and 9.5. They require Mg2+,and are distinguished from exonuclease I by a greater heat stability and an ability to degrade dinucleotides. They are assumed to be exonucleases because early in the course of digestion, they release mononucleotides exclusively. The preparation of exonuclease IVB had an RNase activity (50% of DNase) that may have been due to contamination from a closely eluting RNase peak. The DNase activities were assayed by measuring the release of acidsoluble, ultraviolet-absorbing material from an oligonucleotide mixture. The substrate was prepared by digesting salmon sperm DNA with pancreatic DNase (an endonuclease) to the point of 20-25% acid solubility. Because this susbstrate preparation was degraded at about 20 times the rate of either native or heat-denatured salmon sperm DNA, it was concluded that exonucleases IVA and IVB were specific for oligonucleotides. This conclusion is unwarranted because the substrates were compared at widely different concentrations. The concentration of nucleotide phosphorus was the same in each reaction, and therefore, the short-chain oligonucleotides had, of course, a much higher concentration of 3' termini than did the native or denatured DNAs (71 1. Because of uncertainty about enzyme purity and about substrate specificity, therefore, we do not know if exonucleases IVA and IVB are unique, if they are other known enzymes, or if they are enzyme mixtures. D. EXONUCLEASE VII 1. Introduction Exonuclease VII (72) attacks single-stranded DNA processively from both the 3' and 5' ends, releasing large oligonucleotides that are then further degraded to smaller ones (Fig. 4). The enzyme has been thoroughly reviewed by Chase and Vales (73) and therefore is described briefly here. 71. It was perhaps for this reason that Jorgensen and Koerner noted, in a control experiment, that their preparation of authentic exonuclease I was more active on the oligonucleotide preparation than on denatured DNA. 72. J. W. Chase and C. C. Richardson, JBC 249, 4545, 4553 (1974).

227

12. E. coli EXONUCLEASES

FIG.4. The exonuclease VII reaction. Processive attack at both ends of the singlestranded DNA releases large oligonucleotides that are further degraded by enzyme molecules released from other chains.

2 . Purification and Properties The purification procedure (72), which has recently been improved (73), uses as an enzyme source a mutant that is deficient in both exonuclease I and DNA polymerase I. Streptomycin sulfate is used to coprecipitate the enzyme with cellular DNA, following which the enzyme is further purified by acetone fractionation and column chromatography. Tne native enzyme has a molecular weight of 88,000 as determined from gel filtration and velocity sedimentation data. An unusually high frictional coefficient of 3.07 indicates an asymmetric shape. If the molecule is rodshaped, it will cover a 100-nucleotide length of DNA, which is the approximate size of its initial degradation products. 3 . The Reactions Catalyzed

Exonuclease VII is highly specific for either single-stranded DNA or for the single-stranded ends of duplexes. Like exonuclease I (see Section II,A) it can digest poly(dA-dT), but at a slower rate than single-stranded DNA. The degradation, which is both 3’ 5’ and 5‘ + 3‘, is not affected by the presence or absence of phosphornonoesters at either terminus. Degradation is processive as evidenced by the lack of a significant fraction of partially digested DNA molecules in a partial digest. The initial products appear to be oligonucleotides of chain length 2100 nucleotides. These are further digested so that the limit products are 2 to 25 or more ---$

73. J. W. Chase and L. D. Vales, in “Gene Amplification and Analysis, Vol. 11: Analysis of Nucleic Acid Structure by Enzymatic Methods” (J. G . Chirikjian and T. S. Papas, eds.), in press. Elsevier-North Holland, Amsterdam, 1981.

228

BERNARD WEISS

nucleotides in length. No mononucleotides are produced. Of special interest is the enzyme's ability to excise thymine dimers from a duplex substrate that contains nicks near the dimers, introduced for example by an endonuclease of Micrococcus luteus. This reaction may be analogous to its ability to remove a small displaced or unpaired region from either end of a synthetic duplex, such as d(C)s-d(T)m * poly(dA) or d(Ts-d-(C)a. poMdA). The standard assay for the enzyme measures the conversion of sheared radiolabeled, denatured DNA to an acid-soluble form in the presence of EDTA. Exonuclease VII is unique among the exonucleases ofE. coli in that it is not inhibited by EDTA. The enzyme has a sharp pH optimum at 7.9 and has a 5- to 10-fold higher activity in phosphate as compared to Tris buffer.

4. Biological Role The structural gene for exonuclease VII, designated xseA, has been located with the aid of mutants that produce a thermolabile enzyme. It is at 53 min on the genetic map ofE. coli (17, 7 4 ) . Deletion mutants are fully viable, indicating that the gene is nonessential (75).xseA mutants have a slightly increased sensitivity to ultraviolet irradiation, are sensitive to nalidixic acid, and have a hyper-Rec phenotype (increased recombination frequency) (74). The defects of the exonuclease VII mutants may be difficult to observe because of redundant roles played by other enzymes. Therefore, an xseA mutation has been combined with apolAex mutation (defective in the 5' 3'-exonuclease of DNA polymerase I) and with recBC (exonuclease V) mutations. An xseA-polAex double mutant is more hyper-Rec and more temperature-sensitive than either alone, and although the .rseA mutation does not affect the rate of excision of thymine dimers from a wild-type strain, it markedly reduces the rate of such excision in a recBC-polAex multiple mutant (76-78). --f

5. Reseurch Applications (79) Exonuclease VII has been used in conjunction with other singlestrand-specific enzymes to determine the extent and location of single74. J. W. Chase and C. C. Richardson, J . Bacteriol. 129, 934 (1977). 75. L. D. Vales, J. W. Chase, and J. B. Murphy,J. Bocferiol. 139, 320 (1979). 76. J. W. Chase and W. Masker, J . Bncreriol. 130, 667 (1977). 77. W. E. Masker and J. W. Chase, in, "DNA Repair Mechanisms" (P. C. Hanawalt, E. C. Friedberg, and C. F. Fox, eds.), p, 261. Academic Press, N e w York, 1978. 78. J. W. Chase, W. E. Masker, and J. B. Murphy, J . Bacteriol. 137, 234 (1979). 79. For a more detailed discussion of the techniques mentioned in this section, see Ref. 73.

12. E. roli EXONUCLEASES

229

stranded regions in DNA molecules. Endonuclease S1 of Aspergillus oryzae digests away all single-stranded regions, whether they be terminally or internally located. Exonuclease I removes only a terminal 3' tail. Exonuclease VII removes both 3' and 5' tails. These enzymatic reagents have been used in various combinations to determine the structure of exonuclease V digestion intermediates that have long, single-stranded tails (80), to measure the length of inverted repeats and intervening sequences in a DNA after intrastrand annealing ( 8 / ) , to analyse DNA-mRNA hybrids for regions in the DNA that are not represented in the mRNA due to splicing (82), and to isolate material between inverted repeats, such as DNA cloned by the poly(dA) poly(dT) tailing technique (83). E.

EXONUCLEASE VIII

1. Ziitroduction

Exonuclease VIII (84, 85') is relatively specific for double-stranded DNA. Its reaction products and mode of attack have not been described. The enzyme is specified by the recE gene on an integrated cryptic prophage (the Rac prophage) carried by some lines ofE. coli K12. Enzyme production is normally repressed in wild-type cells; it is expressed at higher levels in sbcA mutants. The discovery of exonuclease VIII and of its evolutionary origin arose from three separate lines of investigation that yielded the following findings. (i) A recBC (exonuclease V) mutant is recombination-deficient, but if it is also rac -, it becomes recombination-proficient during mating at the time it receives a functional rac (recombination activation) locus from an Hfr donor cell (86). This phenomenon resembles zygotic induction, i.e., the derepression of prophage genes upon conjugational transfer into a repressor-free (nonlysogenic) recipient. (ii) recBC mutants can also become recombination proficient through ansbcA mutation (87) located in or near the rac (integrated Rac prophage) locus (88), and such mutants possess a new DNase activity that, unlike exonuclease V, does not require 80. V. MacKay and S. Linn, JBC 249, 4286 (1974). 81. P. Nisen, R. Medford, J. Mansour, M. Purucker, A. Skalka, and L. Shapiro, PNAS 76, 6240 (1979).

82. 83. 84. 85. 86. 87. 88.

A. J. Berk and P. A. Sharp, PNAS 75, 1274 (1978). S. P. Goff, and P. Berg, PNAS 75, 1763 (1978). S. R. Kushner, H . Nagaishi, and A. J. Clark, PNAS 71, 3593 (1974). J. R . Gillen, A. E. Karu, H. Nagaishi, and A. J. Clark,JMB 133, 27 (1977). K. B . Low, Mol. Gen. Genet. 122, 119 (1973). A. Templin, S. R. Kushner, and A. J. Clark, Genetics 72, 205 (1972). R . G. Lloyd and S. D. Barbour, Mol. Gen. Genet. 134, 249 (1974).

230

BERNARD WEISS

ATP (89).(iii) ex0 (previously called red) mutants of bacteriophage A lack an exonuclease required for phage-mediated recombination. After growth on rac+ cells, however, derivatives can be isolated that are recombination-proficient and have regained a DNase activity. These ArecE (or Arev, or A-reverse) phages have acquired by recombination a new region of DNA, not homologous with'h, but homologous with therac region of the chromosome (90-92). Exonuclease VIII is believed, therefore, to be a relative of the exonuclease of bacteriophage A, specified by A-like cryptic prophage. 2. Purijication and Properties The enzyme has been purified by conventional methods from an sbcA mutant of E. coli K12 (84) and from cells infected with ArecE (85). The phage-infected cells yielded about 12 times more enzyme than the sbcA mutants. The enzymes prepared from both sources were physically similar and were both precipitated by antiserum to the ArecE exonuclease. The enzyme appears to consist of a single polypeptide with a molecular weight of 1.4 x lo5, as determined by sedimentation analysis in a glycerol gradient and by sodium dodecyl sulfate gel electrophoresis (85, 93). Surprisingly, this value is quite different from that for A exonuclease (MW = S2,000),and antibody to the latter failed to inhibit exonuclease VIII significantly. Therefore, there must be considerable evolutionary divergence between these enzymes. It has been suggested that exonuclease VIII might be a gene fusion product, thus accounting for its higher molecular weight (85).

3 . Catalytic Properties The enzyme is assayed by the degradation of radiolabeled bacterial DNA to an acid-soluble form (84). The purified enzyme has a marked specificity for double-stranded versus single-stranded DNA (40-fold) (84, 85). Unlike exonuclease V (the recBC enzyme), which also functions in recombination, exonuclease VIII is ATP-independent and cannot degrade the glycosylated DNA of bacteriophage T4 (89). It requires Mg2' and has a pH optimum between 8.0 and 9.0. Questions of identity and origin have dominated investigations of this enzyme, leaving even the most basic 89. S. D. Barbour, H . Nagaishi, A . Templin, and A. J. Clark, PNAS 67, 128 (1970). 90. M. M. Gottesman, M. E. Gottesman, S. Gottesman, and M. Gellert, J M B 88, 471 (1974).

91. K. Kaiser and N. E Murray, Mol. Gen. Genet. 175, 159 (1979). 92. R. Evap, N. R. Seeley, and P. L. Kuempel, Mol. Gen. Gener. 175, 245 (1979). 93. A value of 1.22 x lo5 was separately obtained by the latter method [Ref. (SS)].

12. E . coli EXONUCLEASES

23 1

questions of mechanism unexplored. Thus, we do not know the nature of the reaction products or direction of attack. The only evidence that it is an exonuclease stems from its inability to degrade circular single-stranded DNA molecules (84); circular duplexes were not tried despite the enzyme's specificity for double-stranded DNA. It is assumed that the enzyme will prove to be similar to A exonuclease (W), i.e., a 5' -+ 3' processive exonuclease that releases 5'-mononucleotides and is unable to initiate degradation at a single-strand break; because of the physical differences between these enzymes, however, this assumption may not be warranted. 4. Biological Role

As discussed above, the enzyme can substitute for A exonuclease in A-mediated general recombination. It is also part of the recE pathway, an alternative pathway for homologous recombination in uninfected E. coli (19). TherecE pathway is manifest in mutants for the major recombination pathway (recBC mutants) that regain recombination proficiency concomitant with enhanced expression of the structural gene for exonuclease VIII (recE). This apparent derepression can occur through mutation in a control locus (sbcA) or through zygotic induction during conjugation. Neither loss of the prophage nor mutation in the sbcA locus significantly affects the frequency of genetic recombination in recB +C+cells. The theoretical role of exonucleases in recombination has been reviewed by Radding (95). By restriction endonuclease analysis, Kaiser and Murray (91) have studied the structure of several XrecE phages and of the rac region of the chromosome. Their results indicate that the Rac prophage undergoes excision from the bacterial chromosome and recombines with phage A to form ArecE. Two regions of the Rac prophage, amounting to no more than a few percent of its length, share homology with X DNA. The two phages are, therefore, distantly related.

94. J . W. Little, I. R . Lehman, and A . D. Kaiser, JBC 242, 672 (1967); J. W. Little, h i d . , p. 679; D. M . Carter and C. M. Radding, JBC 246, 2502 (1971); Y. Masamune, R . A. Fleischman, and C . C. Richardson, ibid.. p. 2680. 95. C. M. Radding, Annu. Rev. Biochem. 47, 847 (1978).

This Page Intentionally Left Blank

recBC-like Enzymes: Exonuclease V Deoxyribonucleases KAREN M. TELANDER MUSKAVITCH STUART LINN

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Biological Role . . . . . . . . . . . . . . . . . . . . . . . . .

111.

IV.

V. VI.

A. Recombination and Repair . . . . . . . . . . . . . . . . . B. Restriction . . . . . . . . . . . . . . . . . . . . . . . . . C. Viability . . . . . . . . . . . . . . . . . . . . . . . . . . D. Replication . . . . . . . . . . . . . . . . . . . . . . . . . Purification and Biophysical Properties . . . . . . . . . . . . . The Reactions Catalyzed . . . . . . . . . . . . . . . . . . . . A. ATP-Dependent Double-Stranded DNA Exonuclease . . . . . B. Single-Stranded DNA Exonuclease . . . . . . . . . . . . . C. Single-Stranded DNA Endonuclease . . . . . . . . . . . . . D. DNA-Dependent ATPase . . . . . . . . . . . . . . . . . . E. ATP-Dependent Double-Stranded DNA Unwinding . . . . . . Models for the Mechanism of Action of the ATP-Dependent DoubleStranded DNA Exonuclease . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . .

.

.

. .

. . . .

234 235 235 237 237 237 238 240 240 244 245 245 246 247 250

233 THE ENZYMES,Vol. XIV Copyright 0 1981 by Academic Press. Inc. All rights of reproduction in any form reserved ISBN 0-12-122714-6

234 1.

KAREN. TELANDER MUSKAVITCH AND STUART LINN

Introduction

The first report of an exonuclease V (exoV) activity, an ATP-dependent double-stranded DNase, was made by Tsuda and Strauss in 1964 ( I ) .They utilized the bacterium Micrococcus lysodiekticus, which is now known as M. luteus. Since then similar activity has been observed in nine additional bacterial species: Bacillus laterosporus (2 1, Mycobacterium smegamatis ( 3 ) , Escherichia coli (4-7), Diplococcus pneumoniae ( 8 ) , Haemophilus influenzae (9), Bacillus subtilis (10, I / ) , Bacillus cereus (12), Pseudomonas aeruginosa ( 13), and Alcaligenes faecalis (14). This widespread occurrence suggests that exoV enzymes are ubiquitous in bacteria. In eukaryotes, the only purified ATP-dependent double-stranded DNase was discovered by Gafurov et al. (15, 16) in sea urchin embryos (Strongylocentrotus intermedius). However, work on yeast mitochondria1 mutagenesis suggests the presence of an ATP-dependent DNase (17), and Neurospora crassa has an enzyme with a combination of nuclease activities resembling bacterial exoV (18). Thus, exoV-like enzymes may be present in all living organisms, prokaryotic or eukaryotic. The bacterial enzymes constitute a distinct class with a number of characteristics in common: (1) An ATP-dependent double-stranded DNA exonuclease that degrades DNA to short oligonucleotides, (2) a DNAdependent ATPase that hydrolyzes ATP to ADP and Pi, (3) a requirement for a divalent cation and an alkaline pH optimum, (4) single-stranded DNA exonuclease activity, and ( 5 ) a large multisubunit structure. Further Y. Tsuda and B. S. Strauss, Biochemistry 3, 1678 (1964). M. Anai, Seikagaku 39, 167 (1967). F. Winder and M. P. Coughlan, EBA 134, 215 (1967). G. Buttin and M. Wright, CSHSQE 33, 259 (I%@. M. Wright and G. Buttin, Bull. SOC. Chim. Bioi. 51, 1373 (1%9). M. Oishi, PNAS 64, 1292 (1969). S. Barbour and A. J. Clark, PNAS 65, 955 (1970). G. Vovis and G. Buttin, BEA 224, 29 (1970). E. A. Friedman and H. 0. Smith, JEC 247, 2846 (1972). 10. A. V. Chestukhin, M. F. Shemyakin, N. A. Kalinina, and A. A. Prozorov, FEBS (Fed. Eur. Biochem. Soc.) Lett. 24, 121 (1972). 11. S. Ohi and N. Sueoka, JEC 248, 7336 (1973). 12. G. Banfalvi, F. Antoni, and S. Csuzi, Stud. Eiophys. Berlin 50, 89 (1975). 13. R. V. Miller and A. J. Clark, J. Eacteriol. 127, 794 (1976). 14. J. D. C . Rosamond and M. R. Lunt, EJ 163, 485 (1977). 15. N. N. Gafurov and V. A. Rasskazov, Dokl. Akad. Nauk SSSR 219, 1495 (1974). 16. Y. M. Gafurov, L. L. Terentev, and V. A. Rasskazov, Biochemistry ( U S S R )44, 996 (1979). 17. R. N. Bastos and H. R. Mahler, JEC 249, 6617 (1974). 18. T. Y. K. Chow and M. J. Fraser, Can. J . Biochern. 57, 889 (1979). 1. 2. 3. 4. 5. 6. 7. 8. 9.

13. recBC-LIKE ENZYMES

235

work on some exoV enzymes has revealed associated single-stranded DNA endonuclease and double-stranded DNA unwinding activities. These similarities led Wilcox and Smith to propose a common nomenclature for these enzymes (191, in which the bacterial source abbreviations are those used for restriction endonucleases, and the enzyme class is noted as exoV. Thus Eco exoV, Hind exoV, etc., are used here. The exoV enzymes belong to a larger group of ATP-dependent DNases and DNA-dependent ATPases, which also includes such enzymes as type-I restriction endonucleases, helicases, and gyrases. This larger group has been reviewed by Whitehead (20). The approach of this review is a comparative one. Emphasis is placed on enzymes about which we have the most information: Eco exoV, Hind exoV, Bsu exoV, and Mlu exoV.

It.

Biological Role

A.

RECOMBINATION AND REPAIR

The observation of decreased or undetectable exoV activity in recombination and repair mutants of four bacteria has sparked a great deal of interest in these enzymes. Mutants in the recB or recC genes of E. coli form fewer recombinants in Hfr crosses than do wild-type strains (21-241, are more sensitive to X-rays, ultraviolet light, and mitomycin C (21, 2-3, 24), and lack exoV activity (5-7). Similarly, an H. irtfuenzae mutant selected for sensitivity to methylmethane sulfonate is deficient in recombinant formation following transformation and lacks Hind exoV activity (251, and a D . pneumoniae mutant isolated by sensitivity to y-rays produces transformants at 15% of the wild-type frequency, is UV sensitive, and lacks exoV activity (26). ExoV activity is also missing in therecE5 mutant of B. subtilis (27), but is only decreased five- to tenfold in another recstrain that is a poor recipient for transformation and transduction (10). 19. K. W. Wilcox and H . 0. Smith,JBC 251, 6122 (1976). 20. E. P. Whitehead, i n . "Macrmolecules in the Functioning Cell" (F. Salvatore, G. Marino, and P. Volpe, eds.), p. 159. Plenum, New York, 1979. 21. P. Howard-Flanders and R . P. Boyce, Radicrt. Res., Sicppl. 6 , 156 (1966). 22. B. Low, PNAS 60, 160 (1968). 23. N . S. Willetts and D. W. Mount, J . Bacreriol. 100, 923 (1969). 24. A. J. Clark, A m i i . Rev. Genet. 7 , 67 (1973). 25. M . L. Greth and M. R. Chevallier, BBRC 54, 1 (1973). 26. G. Vovis and G . Buttin, BBA 224, 42 (1970). 27. J. Doly, E. Sasarman, and C. Anagnostopoulos, Murcit. Res. 22, 15 (1974).

236

KAREN TELANDER MUSKAVITCH AND STUART LlNN

Recombination and repair lesions in bacteria that lack exoV are not absolute, so the deficiency must be carefully quantitated with well-defined strains. For E. coli recB and recC mutants, the frequency of recombinants following Hfr crosses is decreased no more than 100-fold below that of the wild type (23), and in A phage-prophage crosses it is decreased no more than twofold (28).Postreplication repair is almost normal in recB mutants (29), and recB and/or recC mutants are only slightly more sensitive than wild-type E . coli to ultraviolet light (23). Finally, work by Stahl and Stahl on A phage chi mutants indicates that these genes are essential to A phage-phage recombination, at least under special conditions (30). A thorough investigation of the H . injuenzae mutants has also been made. The results demonstrate the importance of relocating an allele from its original highly mutated background and the difficulty of quantitating transformation frequencies. When the mutant allele, add-18, originally isolated by Greth and Chevallier (2.51, is moved to an unmutagenized strain, the ability of the strain to form transformants, quantitated as transformants per competent cell, is decreased only twofold relative to the wild type. However, if quantitated as transformants per colony-forming unit, there is a sevenfold decrease, and if quantitated as transformants per microgram of DNA bound, there is a 17-fold reduction (31). Other mutants 1ackingHind exoV activity give similar results, and since the frequency of transformants per competent cell is almost the same as for wild-type cells, one might conclude that exoV is not involved in recombination. However, a small effect appears reproducible. These mutants are also slightly more sensitive to y-rays, UV light, methylmethane sulfonate, and mitomycin C, and these phenotypes have been used to group the mutant alleles into three complementation groups ( 3 2 ) . While the three complementation groups for loss of Hind exoV activity only suggest that this enzyme may have three subunits (32), in E. coli and B . subtilis direct identification of the exoV enzyme as the product of at least one rec gene has been possible. Tomizawa and Ogawa (33) have shown that a temperature-sensitive recB mutation produces a temperature-sensitive Eco exoV enzyme; however, a temperaturesensitive recC mutation has no effect on the enzyme. Further work by 28. P. Lin, E. Bardwell, and P. Howard-Flanders,PNAS 74, 291 (1977). 29. K. C. Smith and D. H. C. Meun, J M B 51, 459 (1970). 30. F. W. Stahl and M. M . Stahl,Generics 86, 715 (1977). 3 1. K . W. Wilcox and H . 0. Smith, J . Bacreriol. 122, 443 ( 197s). 32. J . Kooistra, G . D. Small, J . K . Setlow, and R. Shapanaka, J . Bacterial. 126, 31 (1976). 33. J . 1. Tomizawa and H. Ogawa, Nfiritre New Eiol. 239, 14 (1972).

13. wcBC-LIKE ENZYMES

237

Kushner revealed that of the DNase activities only the double-stranded DNA exonuclease activity of Eco exoV is made temperature sensitive by the temperature-sensitive recB allele (34). In B. subtifis, Doly and Anagnostopoulos (35) discovered that although a recES mutant has no exoV activity, the inactive enzyme can be purified and has an altered subunit. It should be noted that the five subunits of Bsu exoV identified by these workers may be the products of proteolysis (36).

B. RESTRICTION Studies of mutants in the E. coli recB and recC genes indicate that exoV is responsible for completing the degradation of foreign DNA after it has been acted upon by a host-restriction endonuclease (37).

C. VIABILITY It has been observed, both in E. coli and H . injuenzae, that mutants lacking exoV activity have a slower growth rate and reduced viability when compared to the wild type (32, 38). Typically, less than one-half of the cells present in a liquid culture are able to form colonies on agar. Studies of E. coli mutants indicate that reduced viability occurs because some of the cells cannot divide and others can divide only a limited number of times. The reason why mutants sometimes fail to divide is unknown (39). Attempts to clone the E. coli recB and recC genes have been unsuccessful, suggesting that the level of exoV is an important factor in determining the viability of the bacterium, perhaps in an area different from repair and recombination.

D. REPLICATION Involvement of Eco exoV in replication could explain the decreased viability of mutants. Although data on this point are only suggestive, studies of spheroplast lysates and toluene-treated E. coli indicate that 34. 35. 36. 37. 38. 39.

S. R . Kushner, J . Bacteriol. 120, 1219 (1974). J. Doly and C. Anagnostopoulos, EJB 71, 309 (1976). M. F. Shemyakin, A . A. Grepachevsky, and A. V. Chestukhin, EJB 98, 417 (1979). V. F. Simmon and S. Lederberg, 1.Bacteriol. 112, 161 (1972). F. Capaldo-Kimball and S. D. Barbour, 1.Bacteriol. 106, 204 (1971). F. N . Capaldo and S. D. Barbour, J . Bacleriol. 115, 928 (1973).

238

KAREN TELANDER MUSKAVITCH A N D STUART LINN

exoV is essential to ATP-dependent DNA synthesis observed under these conditions (40,41). In addition, Hendler and co-workers (42) have isolated a complex of E. coli exoV and DNA polymerase I that carries out ATPdependent DNA synthesis, and tentatively report the isolation of similar complexes with DNA polymerases I1 and 111. However, since recB and recC mutants do replicate, the Eco exoV activity involved in replication must still exist in these mutants, or is replaceable. 111.

Purification and Biophysical Properties

Some properties of various exoV enzymes, including the sea urchin (S. intermedius) enzyme, and their relative purity are summarized in Table I. Homogeneous enzymes have been obtained from E. coli, H. influenzae, and B. subtilis. The subunit structure of the Eco and Bsu exoV enzymes varies with the purification technique. In the less purified Eco exoV preparation of Oishi and co-workers (see 44,45), an additional smaller subunit is detected when compared with the preparation of Goldmark and Linn (43). The larger subunit fraction identified by Lieberman and Oishi may contain both of the large subunits reported by Goldmark and Linn, because the only characterization done by the former group was glycerol gradient sedimentation. The smaller subunit may be purified away from an enzyme core in the Goldmark and Linn preparation, or may be fortuitously present in the less purified Nobrega et a/. preparation (44). Two observations suggest that the small subunit may be a true subunit ofEco exoV: (1) It is required for recovery of activity after high salt treatment and separation of the two subunit fractions (45), and (2) Hind exoV has three subunits, two large and one small (46). The difference between the two B. subtilis preparations would seem to be due to proteolysis. Shemyakin et al. (36) add phenylmethylsulfonyl fluoride to the lysis buffer and report purification of a stable enzyme preparation with about an 8% yield. In contrast, Doly and Anagnostopoulos (35)report that the Bsu exoV is unstable at later stages of their purification and obtain a 1.7% yield. Ohi and Sueoka (11) also report 40. G. W. Brazill, R. Hall, and J . D. Gross, Nature (London)New Biol. 233, 281 (1971). 41. D. J. Harper, P. L. Chen, and P. L. Carl, BBA 474, 363 (1977). 42. R. W. Hendfer, M. Pereira, and R. Scharff, PNAS 72, 2099 (19751. 43. P. J. Goldmark, and S. Linn, JBC 247, 1849 (1972). 44. F. G . Nobrega, F. H. Rola, M. Pasetto-Nobrega, and M. Oishi, PNAS 69, 15 (1972). 45. R . P. Lieberman and M. Oishi, PNAS 71, 4816 (1974). 46. K. W. Wilcox, M. Orlosky, E. A. Friedman, and H . 0. Smith, FP 34, 515 (1975).

239

13. rerBC-LIKE ENZYMES TABLE I

PURIFICATION AND PROPERTIES OF

Source of enzyme

E. coli (43)

Purification (-fold)

Specific activity" (U/rng)

17,000

57,000

s

~

EXOV ENZYMES ~Molecular , ~

(S)

weight

12

270,000

E. coli (44, 4 5 )

3,500

21,600

12.4

350,000

H. influenme (9.46)

2,000

28,000

12

290,000

B. subtifis (35)

350

237

B . subtilis (.16)

5 ,000

$4,000

2,300 I50 100 743 4,200 97,000 820 200

67,000 54,500 8,700 28,000 2,450 91,OOO 1,900 4,700

M . luteus (47) B. laterosporus (48) M. smegmatis (49) D. pneumotiiae (8) B. cereus (50) P. aeruginosa (13) A. fueculis (14) S . inrermedius ( 1 6 )

-

270,000

300,000

-

300,000

-

450,000

Subunit molecular weight

140,OOO 128,000 170,000 60,000 115,000 107,000 68,000 81,000 70,000 62,000 52,500 42,500 155,OOO 140,000

-

-

" All values have been adjusted so that 1 unit is the amount of enzyme needed to render acid-soluble 1 nmol of duplex DNA-nucleotides in 30 rnin.

instability and a poor yield during purification in the absence of a protease inhibitory. Several generalizations can be made from Table I about bacterial exoV enzymes. They are around 300,000 daltons and are composed of two nonidentical, large subunits. In some cases a third smaller subunit may be part of the enzyme. The high level of purification required to obtain homogeneous enzyme indicates that relatively few enzyme molecules are present in the cell. 47. M . Anai, T. Hirahashi, and Y. Takagi, JBC 245, 767 (1970). 48. M. Anai, M. Mihara, M. Yamanaka, T. Shibata, and Y. Takagi,J. Biochem. (Tokyo) 78, 105 (1975). 49. F. G. Winder and M. F. Lavin, BBA 247, 542 ( 197 I ). 50. G . Banfalvi, S. Csuzi, A. Ohlbaum, and F. Antoni, ABB 14, 53 (1979).

240 IV.

KAREN TELANDER MUSKAVITCH AND STUART LINN

The Reactions Catalyzed

A. ATP-DEPENDENT DOUBLE-STRANDED DNA EXONUCLEASE ATP-dependent double-stranded DNA exonuclease activity defines an exonuclease V enzyme (51). In-depth characterization of this activity has been done for Eco exoV, Hind exoV, Bsu exoV, and Mlu exoV. A terminus is required on the double-stranded DNA substrate. The enzymes are active on linear duplex DNA from many sources (If, 3 6 , 4 3 , 47, 5 / , 52), and phosphomonoester groups are not required for Eco exoV activity (53). Covalently closed circular double-stranded DNAs are not substrates (36, 43, 51, S2, 54), nor are nicked duplex circles regardless of whether 3’- or 5’-phosphomonoesters are present at the nick (52-54). If a single-stranded gap of greater than 5 nucleotides is introduced into a duplex circular DNA, Eco exoV is able to degrade the DNA (53, 5 3 , having presumably opened the DNA with its single-stranded DNA endonuclease activity. It has been reported that Hind exoV is not active on gapped circular DNA, but these experiments were done under conditions where the single-stranded DNA endonuclease is not active (52). The degradation of linear, but not circular, duplex DNA by Eco exoV is an exploitable property, although its utilization is limited by the difficulty of purifying the enzyme. The enzyme can be used to assay restriction endonucleases (56) and to eliminate cellular DNA in plasmid preparations. Both Eco exoV andHind exoV are inhibited by cross-links within DNA in a manner that suggests the enzyme degrades the DNA from a terminus until it encounters a cross-link, and then stalls (57, 58). Eco exoV is inactive on duplex RNA but has limited activity on RNA-DNA hybrids, presumably degrading DNA termini up to a hybrid region (53).The treatment of circular or linear duplex DNAs with UV light or X-rays does not change their respective failure or success as substrates forEco exoV (53). ATP is a required cofactor for the double-stranded DNase of all exoV enzymes. Nucleoside triphosphate specificities for all the bacterial en51. 52. 53. 54. 55. 56. 57. 58.

M. Wright, G . Buttin, and J. Hurwitz,JBC 246, 6543 (1971). E. A. Friedman and H. 0. Smith, JBC 247, 2859 (1972). A. E. Karu, V. MacKay, P. J. Goldmark, and S . Linn, JBC 248, 4874 (1973). Y. Takagi, K. Matsubara, and M. Anai, BBA 269, 347 (1972). A. Prell and W. Wackernagel, EJB 105, 109’( 1980). D. Lackey and S. Linn, “Methods in Enzymology,” Vol. 65, p. 26, 1980. A. E. Karu and S. Linn, PNAS 69, 2855 (1972). M. Orlosky and H. 0. Smith,JBC 251,6117 (1976).

13. recBC- LIKE ENZYMES

24 1

zymes have been investigated: Eco exoV (43, 51, 591, Hind exoV (9),Bsu exoV ( l o ) ,Mlu exoV (60), Bla exoV (48), Msm exoV (49), Dpn exoV (8), Bce exoV ( 5 0 , Pae exoV (13), and Afa exoV (14). ATP is generally the most effective, and dATP nearly as effective. The efficacy of other riboand deoxytriphosphates varies with the enzyme. During DNA degradation the ATP is hydrolyzed to ADP and Pi and, as might be expected, all pure di- and monophosphates, as well as P-y ATP analogues, are ineffective. Interestingly, many ATP residues are hydrolyzed for each phosphodiester bond cleaved in the DNA. The optimum ATP concentration is usually between 10-s and M . In most cases, higher concentrations of ATP inhibit the double-stranded DNA exonuclease, but not the other activities of the enzymes. A divalent cation is required by all the exoV enzymes. Mg2+is generally best with an optimal concentration of around 10 mM. Mn*+is Less effective and has a lower optimal concentration. In some cases other divalent cations can stimulate the ATP-dependent double-stranded DNA exonuclease ( 3 , 4 7 ) but usually, when they are added in the presence of MgC12, they are inhibitory (3, 8, 9, 47, 50, 6 1 ) . As more fully discussed below, changes in the species or concentration of the divalent cation can have dramatic effects on the activities of the exoV enzymes (36, 61, 62). The pH optimum for all of the enzymes is around 9, and some have interesting ionic strength dependences. NaCl can inhibit Eco exoV if present at greater than 40 mM ( 5 / ) ,and there is evidence that K+, but not Na+, enhances the ability to degrade more than one DNA duplex (63). Hind exoV is 60% more active in 0.5 M Tris-HC1 than in 0.1 M , but is 50% inhibited by 0.125 M NaCl(9). In the case of Msm exoV, the identity of the salt is not important, but an ionic strength of approximately 0.3 M is required for maximum activity (49). Various factors that can inhibit the ATP-dependent double-stranded DNA exonuclease have been studied. In B. subtifis, an inhibitor was detected in crude extracts and later identified as small DNA fragments (48). Proteins that bind DNA can also decrease the production of acid-soluble oligonucleotides. Eschierichia cofi single-stranded DNA-binding protein inhibits Eco exoV (64), and low molecular weight, basic, relatively thermostable proteins from B. subtifis affect the Bsu exoV (65). Many E. coli 59. D. C. Eichler and I. R. Lehman, JBC 252, 499 (1977). 60. M. Anai, T. Hirahashi, M. Yamanaka, and Y. Takagi, JBC 245,775 (1970). 61. J. Rosamond, K. M. Telander, and S. Linn, JBC 254, 8646 (1979). 62. K . W. Wilcox and H. 0. Smith, JBC 251, 6127 (1976). 63. U. Hermanns and W. Wackernagel, EJB 76, 425 (1977). 64. V. MacKay and S. Linn, JBC 251, 3716 (1976). 65. A . V. Chestukhin, V. I. Fedchenkov, and M. F. Shemyakin, M d . B i d . 13,499 ( 1979).

242

KAREN TELANDER MUSKAVITCH AND STUART LINN

bacteriophages are also known to produce proteins that inhibit Eco exoV (6669). There has been only limited work on the effects of chemical inhibitors, the most interesting being the inhibition by pyridoxal phosphate of Eco exoV (70). The product of the ATP-dependent exonuclease activity is a mixture of acid-soluble 5'-phosphomonoester oligonucleotides. The average length of the oligonucleotides varies between 3 and 6 nucleotides depending upon the enzyme (9, 13, 36, 43, 47, 49, 71). When the time course of product formation is investigated using either sedimentation velocity analysis or viscometry , it is observed that the acid-soluble oligonucleotides are present from the very beginning of the reaction. When present in excess, undegraded substrate DNA molecules persist until the end of the reaction (8, 9, 43, 47, 49, 53, 71, 72). These data indicate that the ATP-dependent double-stranded DNase is exonucleolytic and processive, completely degrading one DNA molecule to oligonucleotides before beginning to degrade another. Studies with cross-linked duplex linear DNAs support this mechanism (13, 57, 58). The sedimentation velocity analyses indicate that intermediates in the double-stranded DNA exonuclease reaction do not tend to accumulate. However, they may be present in Dpn exoV and Bla exoV reactions (8, 71), and can be observed at short times in Hind exoV reactions (62, 73). Changes in the concentrations of ATP, Mgz+and salt can greatly increase the accumulation of intermediates in the processive exoV degradation. For Mfu exoV, low concentrations of ATP, which limit its availability, result in the production of sedimentable DNA fragments (54), but no further work has been done on the fragments. To increase the intermediate yield for the E. coli, H. influenzae, and 3. subtilis enzymes, the concentration of ATP is increased to 1-5 mM (36, 53, 62) which, as noted previously, inhibits the production of acid-soluble oligonucleotides. Additional modifications of the reaction conditions can further favor the production of intermediates: for Eco exoV the ionic strength is increased slightly (53); for Hind exoV both the concentration of Mg2+and the ionic strength are 66. Y. Sakaki, J . Virol. 14, 1611 (1974). 67. A. E. Karu, Y. Sakaki, H. Echols, and S. Linn, JBC 250, 7377 (1975). 68. R. Pacumbaba and M. S. Center, J . Virol. 16, 1200 (1975). 69. F. VanVliet, M . Couturier, J. DeLafonteyne, and E. Jedlicki, Mol. Gen. Genef. 164, 109 (1978). 70. M. Anai, T. Fujiyoshi, J. Nakayama, and Y. Takagi, JBC 254, 10853 (1979). 71. M . Anai, M. Yamanaka, T. Shibata, T. Mihara, T. Nishimoto, and Y . Tagaki, J. Biochem. (Tokyo) 78, 115 (1975). 72. P. J. Goldmark and S. Linn,PNAS 67, 434 (1970). 73. E. A. Friedman and H. 0. Smith, N ~ f u r eNew Biol. 241, 54 (1973).

243

13. recBC- LIKE ENZYMES TABLE I1

CHARACTERISTICS OF REACTION INTERMEDIATES“ Characteristics Tailed fragments Length of duplex region (base pairs) Length of tails (nucleotides) Range Mean Termini of tails Single-stranded fragments Length (nucleotides)

Eco exoV (75)

Hind exoV (74, 6 3 )

300-27,OOO

1000-36,000b and -2000b

800-5,OOO

51.5,00Ob

3,200 5’ and 3‘

4,200 3’

100-500

5 21 ,OOOb

T7 phage DNA, the substrate in these experiments, is 39,000 base pairs long. Estimated from data reported in micrometers assuming 2800 base pairs or 2900 nucleotides per micrometer for duplex and single-stranded DNA, respectively.

decreased (62, 73); forBsu exoV the pH is decreased (36). The amounts of Eco exoV and Hind exoV are also increased to give a more synchronous reaction. Biochemical (53, 62, 73, 74) and electron micrographic (62, 7 4 ) characterization of the Eco exoV and Hind exoV intermediates reveal that they are of two classes: (1) Duplex DNA fragments with terminal singlestranded tails, and (2) large single-stranded fragments. Table I1 summarizes the details of their properties. While similarities between the intermediates produced by Eco exoV and Hind exoV are evident, several important differences exist. Hind exoV produces only 3’ tails, whereas Eco exoV produces about equal numbers of 3‘ and 5’ tails. Only rarely do intermediates of the Hind exoV reaction have two tails, but it is not uncommon to observe a single-stranded tail on each end of a duplex Eco exoV intermediate. This latter difference may reflect the fact that one enzyme per terminus was added in the Hind exoV study, whereas three to four per terminus were added in the Eco exoV study. Hind exoV produces a distinct class of duplex intermediates about 2000 base pairs long, possibly with single-stranded tails, which accumulate during the course of the reaction. No similar, distinct class is observed for Eco exoV. Further studies of Eco exoV intermediates have been carried out using a very gentle technique to spread the molecules for electron microscopy (75). The presence of terminal single-stranded loops associated with one 74. V. MacKay and S. Linn, JBC 249, 4286 (1974). 75. K . M. T. Muskavitch and S . Linn, in, “Mechanistic Studies of DNA Replication and Genetic Recombination” (B. Alberts and F. C. Fox, eds.), ICN-UCLA Symp. Mol. Cell. Biol., Vol. 19, p. 901, 1980.

244

KAREN TELANDER MUSKAVITCH AND STUART LINN

or two single-stranded tails is revealed, as is the rare occurrence of pairs of single-stranded loops located within otherwise duplex regions. The relevance of all these data to the mechanism of the double-stranded DNase is discussed in Section V. Unfortunately, extensive characterization of the intermediates produced by Bsrr exoV has not yet been done, but sedimentation velocity analysis (36) of the intermediates suggests that they may be similar to the intermediates produced by Eco exoV and Hind exo V. Along with partially degraded DNA molecules, complexes between enzyme and DNA may be regarded as reaction intermediates. Whereas no one has succeeded in detecting Eco exoV-DNA complexes, complexes have been detected with the enzymes from H. itzjkenzue (19), M . linteus (761, and M . sntegrnatk (77). The optimum ATP and Mgz+concentrations for complex stability vary. When testing with SV40 DNA, Hind exoV binds only to the linear duplex form of DNA and only at termini. Mlu exoV binds best to blunt duplex DNA such as T7 DNA, and more poorly to molecules with single-stranded tails, such as hDNA or T7 DNA following limited exonuclease 111 digestion. The binding of Bsu exoV to duplex DNA is unique; it appears to bind all over the DNA and cause a nonspecific pairing of duplexes, the extent of which depends upon the concentration of enzyme (78). ATP is not required.

B. SINGLE-STRANDED DNA EXONUCLEASE With the exception of Pric exoV, all of the known bacterial exoV enzymes have associated single-stranded DNA exonuclease (8, 9, I I , 13, 14, 35, 36, 42,47-50, 72). ATP dependence varies from none for one preparation ofBsrr exoV (11) to total forEco exoV (721, Hind exoV (9), Mll4 exoV (47),and Afii exoV (14). The rate of degradation of single-stranded DNA is always less than that for duplex DNA, but the degree varies between a twofold decrease for Eco exoV (43) and a 100-fold decrease for Blri exoV (48). [This difference probably explains why molecules with singlestranded tails are degraded more slowly than duplex molecules with blunt ends (55, 62).] Reaction conditions for maximum activity are usually the same as for the duplex exonuclease. The single-stranded DNA exonuclease activity ofEco exoV is not inhibited by high concentrations of ATP, but Afrr exoV has the same ATP dependence for both of its exonuclease 76. B . Van Dorp, M . T. E . Ceulen, H. L. Heijneker, and P. H. Pouwels, BBA 299, 65 (1973). 77. F. G . Winder and P. A. Sastry,FEBS (Fed. Eitr. Biochetn. Soc.) L e u . 17, 27 (1971). 78. S. Ohi, D. Bastia, and N . Sueoka, Notidre (Lotidoti) 248, 586 (1974).

13. rrcBC- LIKE ENZYMES

245

activities. Studies of Eco exoV indicate that the single-stranded DNA exonuclease is processive and produces acid-soluble oligonucleotides (72).

C. SINGLE-STRANDED DNA ENDONUCLEASE Only three species of exoV have been tested for the presence of singlestranded DNA endonuclease activity. It is present in Eco exoV (72) and Hind exoV (52),but absent in Prie exoV under a variety of reaction conditions (13). It is important that the enzyme be tested under more than one set of reaction conditions because this activity is not maximally active under conditions optimal for the ATP-dependent double-stranded DNA exonuclease. The Hind exoV single-stranded endonuclease is active only at Tris-HC1 concentrations less than 0.25 M , and so was originally overlooked (32).Eco exoV single-stranded DNA exonuclease has a pH optimum of 7 and is stimulated sevenfold by ATP (72). The optimum ATP concentration is about 2 mM and inhibition is not observed at high ATP concentrations (4.3). It can be inhibited by E. c d i single-stranded DNA binding protein (SSB) ((54) and CaC1,(61 1. With fd single-stranded circular DNA in the absence of ATP, the endonuclease acts randomly (72), but recent work by Prell and Wackernagle (55) suggests that this is not the case for single-stranded gaps located within duplex DNA. D. DNA-DEPENDENT ATPASE All exoV enzymes have a DNA-dependent ATPase activity that hydrolyzes ATP to produce ADP and inorganic phosphate (8, 11, 13, 14, 43, 48-51, 60, 79). The optimal conditions for the ATPase are generally those for the exonuclease activities. A DNA cofactor is required: Any DNA that can be hydrolyzed by the enzyme suffices, but single-stranded DNA is less effective than double-stranded DNA (11, 43, 50). An exchange reaction between ATP and ADP that does not require DNA has been observed for Mlu exoV (80). Normally, the hydrolysis of ATP is coupled to the degradation of DNA in such a way that the number of ATPs hydrolyzed per phosphodiester bond cleaved can be determined. This number is greater than one and varies between approximately 3 and 40 for different enzymes (8, 11, 13, 14, 43, 48-50, hU, 79). It can also vary with reaction conditions (14, 4 3 ) . Several examples of the uncoupling of ATPase from DNase are known. Eco 79. H . 0. Smith and E. A. Friedman, JBC 247, 2854 (1972). 80. M . Anai and Y. Takagi, JBC 246, 6389 (1971).

246

KAREN TELANDER MUSKAVITCH AND STUART LINN

exoV, Hind exoV, and Pue exoV form a complex with cross-linked, linear DNA that (although unable to degrade the DNA) continues to hydrolyze ATP (13, 57, 58). A similar phenomenon is observed for Eco exoV in the presence of an RNA-DNA hybrid (53).SSB inhibits the DNase activity of Eco exoV but not the ATPase such that more ATP molecules are hydrolyzed for each phosphodiester bond cleaved (64). High levels of ATP inhibit the double-stranded DNA exonuclease of Eco exoV and Hind exoV but do not so affect the ATPase ( 4 3 , 5 3 , 5 9 , 6 2 ,79). The effect can be enhanced for Hind exoV if the concentration of Mg2+ is decreased (62). For Eco exoV the most dramatic uncoupling is seen in the presence of Caz+ (81). The DNase activities can all be completely inhibited without significantly affecting the ATPase (61). It should be noted that not everything that inhibits the double-stranded DNA exonuclease activity is also an uncoupler. Pyridoxal phosphate (70) and a variety of divalent metallic salts (61) inhibit both DNase and ATPase. E. ATP-DEPENDENT DOUBLE-STRANDED DNA UNWINDING The ability of exoV to catalyze unwinding is suggested by the nature of the double-stranded DNA exonuclease reaction intermediates and by the lack of any obvious strict coupling between ATPase and DNase. Intermediates with single-stranded regions are formed by Hind exoV if the concentration of Mg2+is lowered (62). In a similar way, unwinding is favored over DNase by Eco exoV in the presence of SSB (64)or high levels of ATP (53,59, 74). In the presence of Ca2+,only ATP-dependent unwinding and associated ATP hydrolysis are catalyzed by Eco exoV (61,811. Under nearly physiological conditions, which include 1 mM MgC12, 1 mM CaCl,, and 5 mM ATP, linear duplex T7 DNA appears to be unchanged by Eco exoV, although ATP is hydrolyzed. If SSB is also present in such a reaction “forks” (two single-stranded tails at the same terminus) are formed at the ends of duplex DNA and grow larger with time until only single-stranded DNA remains. This single-stranded DNA is about one-third unit length because of a very low level of nicking activity that is still retained by Eco exoV. The exact amount of nicking vanes with the reaction conditions (61). If SSB is added after the reaction has begun but just shortly before it is terminated, small denatured regions, or “bubbles,” are observed within the duplex. Thus, with Ca2+present, Eco exoV apparently moves along in the interior of the DNA double helix, transiently denaturing the duplex in its vicinity as it travels down the molecule. When present, SSB prevent 81. J. Rosamond, B. Endlich, K. M. Telander, and S. Linn, CSHSQB 43, 1049 (1979).

13. rPcBC-LIKE ENZYMES

247

reannealing of the strands unwound by the enzyme, and single-stranded DNA accumulates. It has been suggested that Hind exoV moves within a DNA double helix in a similar manner ( 6 2 ) . Recent electron microscopic studies by Taylor and Smith indicate that the unwinding by Eco exoV in the presence of Ca2+proceeds by way of single-stranded loops associated with two single-stranded tails (8.2,83 ) in a manner reminiscent of the double-stranded DNA exonuclease (75). However, under the unwinding conditions internal paired single-stranded loops (twin loops) are observed more frequently (see below).

V.

Models for the Mechanism of Action of the ATP-Dependent Double-Stranded DNA Exonuclease

Only the H. itiflirenzue and E. coli enzymes have been studied in sufficient depth to formulate detailed models for their action on doublestranded DNA. Wilcox and Smith have proposed the model shown in Fig. 1 for Hind exoV ( 6 2 ) . Following binding to a terminus,Hind exoV begins to travel along the interior of the double helix if ATP is present. At approximately 6000 base pair intervals, the enzyme begins to hydrolyze one strand, forming single-stranded fragments. After continuing in this mode for approximately 4000 base pairs, the enzyme nicks the previously undegraded strand, releasing a duplex of roughly 2000 base pairs with a 3’ single-stranded tail about 4000 nucleotides long. The enzyme remains bound to the original DNA molecule and reverts to its “tunneling” mode, again transiently denaturing the double helix as it moves through it until, after about 2000 base pairs, it begins again to degrade the same strand. Hiri d exoV continues this processive fragmentation until it completes the formation of intermediates from one molecule. It can then begin either to degrade another intact duplex, or to degrade an intermediate to acidsoluble oligonucleotides. The former is the preferred course. Figure 2 depicts a mechanism for Eco exoV recently proposed by Muskavitch and Linn (75). It is based upon an earlier model (74) but is more consistant with further electron microscopic observations made by them (75) and by Taylor and Smith (8.2,83). It is proposed to apply to reactions with or without Ca2+.As shown, the enzyme binds to a duplex terminus (Fig. 2A) and begins to unwind the DNA double helix, forming one strand 82. A. Taylor and G. R . Smith,iri. “Mechanistic Studies of DNA Replication and Genetic Recombination” 1B.Alberts and F. C. Fox, eds.), ICN-UCLA Symp. Mol. Cell. Biol., Vol. 19, p. 909, 1980. 83. A. Taylor, and G . R . Smith, CeII, 22, 447 (1980).

248

KAREN TELANDER MUSKAVITCH AND STUART LINN

M9'* (1) Bindina

1 ATP

-.

( 2 ) Processive fragmentation

-\

2e .

\=-

--

' I

3-

---~ ,--

L----%

,--.-

? %

-'

--A*.-\-

-

-,

---

1

e-

--'

(3) Recycling

-, or

-t

(4) Dcgrodotion

oligonuc loot ides

>

-r

7-

.c

,'

- 1 ,

,'

A

FIG.1. Schematic illustration of the degradation of duplex DNA by Hind exoV according to Wilcox and Smith (62), reprinted with permission.

,

;;-.

\

C'

B"

b RE

FIG.2. Model for the degradation of duplex DNA by Eco exoV (75). The structures in the circle represent true reaction intermediates; those on the periphery are formed by disruption of the intermediates during observation procedures (e.g., electron microscopy).

13. recBC-LIKE ENZYMES

249

into a loop and displacing the other as a tail (Fig. 2B). Both 3’- and 5’-terminated tails are formed with equal frequency. If SSB is present when such a loop-plus-tail structure is disrupted (during electron microscopic sample preparation, for example), renaturation (Fig. 2B‘) is prevented and a fork is formed (Fig. 2B”). When the loop is several thousand nucleotides in length it is hydrolyzed, beginning at its terminus to form single-stranded fragments several hundred nucleotides long. During this hydrolysis, a loop and two tails are observed at a terminus (Fig. 2C), or, if disrupted, either a fork (Fig. 2C”) or a single-stranded tail (Fig. 2C’), depending upon whether SSB is present. If the strand composing the loop is not cleaved but is partially threaded past the enzyme, it can reanneal with the complementary region of the tail. The DNA still held by exoV in the loop, and that complementary to it in the tail, then form a pair of single-stranded loops (twin-loops), previously called “rabbit ears” (Fig. 2,RE). This process may occur at times when loop DNA is being threaded through the enzyme before it is cut off as a single-stranded fragment. SSB in the reaction does not prevent the formation of twin loops. Complete hydrolysis of the loop to single-stranded fragments produces a molecule with a single-stranded tail and the enzyme bound at the single strand : double-strand junction (Fig. 2D). Since the enzyme is observed to be essentially processive, hydrolysis of the tail must occur in a manner that allows the enzyme to remain associated with the same duplex DNA molecule. This could be accomplished by hydrolysis of the tail from its junction with the duplex toward its terminus as depicted in Fig. 2E. The enzyme, having completed hydrolysis of the tail, is able to reinitiate the reaction cycle (Fig. 2A). Repetition continues until the duplex molecule is completely reduced to single-stranded fragments. The released enzyme can then begin to unwind and hydrolyze another DNA molecule, or to degrade the single-stranded fragments to acid-soluble oligonucleotides. It is obvious that the proposed mechanisms for Eco exoV and Hind exoV are very similar. The one essential difference is that, unlike Eco exoV, Hind exoV appears to proceed via a two-stage process, first forming a distinct class of intermediates and then degrading them. Both act processively from the end of a linear molecule and simultaneously unwind the double helix as they move within its interior (rather than along its exterior). As previously suggested (84), ATP hydrolysis is believed to provide the energy for this combination of translocation and denaturation. The type-I restriction endonucleases represent, in contrast, a class of enzymes that use ATP hydrolysis to power translocation along the ex84. F. G. Winder, Natirre New B i d . 236, 75 (19721.

250

KAREN TELANDER MUSKAVITCH A N D STUART LINN

terior of the DNA double helix (85). The details of the manner in which Eco exoV and Hind exoV unwind the DNA may be slightly different, since Hind exoV produces only 3' tails, where Eco exoV produces both 3' and 5' tails. However, it is reported that Hind exoV also forms both loopplus-tail structures and twin-loops (8.3). The preeminence of unwinding, even in reactions that degrade DNA to oligonucleotides, and particularly under conditions believed to approximate those within the bacterium, suggests that unwinding is an important function of the enzyme in vivo. Further credence is lent to this view by work with temperature-sensitive mutants of E. coli . At the nonpermissive temperature, the mutants display the decreased viability characteristic of recBC mutants (86). Among the nuclease activities, only the doublestranded DNA exonuclease of the isolated enzyme is found to be temperature sensitive (34).It is possible that a deficiency in unwinding is the lesion in these mutants causing both decreased viability and.loss of the doublestranded DNA exonuclease. Unwinding, which creates single-stranded DNA, may also explain the role of Eco exoV in recombination. Single-stranded DNA is believed to play an important role in recA-mediated recombination (87). Since other cellular enzymes also are able to unwind DNA (88, 8 9 ) , loss of exoV activity need not result in a total inability to carry out recombination. VI.

Conclusion

ExoV enzymes are found throughout the bacterial kingdom. All have ATP-dependent double-stranded DNA exonuclease and DNA-dependent ATPase activities. Also associated with the enzyme may be ATPdependent single-stranded DNA exonuclease, ATP-stimulated singlestranded DNA endonuclease, and ATP-dependent DNA unwinding. Investigation of the mechanism of action of the enzyme indicates that DNA unwinding is fundamental to several of the enzyme's activities and may well be its most important function in the cell. The complete degradation of DNA to acid-soluble oligonucleotides may only serve a function in restriction. 85. B. P. Endlich and S. Linn. (1980), this volume, Chap. 9. 86. S. R. Kushner, J . Bacterial. 120, 1213 (1974). 87. T. Shibata, C. DasGupta, R. P. Cunningham, and C . M. Radding, PNAS 76, 1638 ( 1979). 88. M. Abdel-Monem and H. Hoffman-Berling, H E 65, 431 ( 1976). 89. M. Abdel-Monem, M . C . Chanal, and H. Hoffman-Berling, H E 79, 33 (1977).

Enzymes That Incise Damaged DNA ERROL C. FRIEDBERG THOMAS BONURA ERIC H. RADANY JACK D. LOVE

.

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . .

251

11. Enzymes That Attack Phosphodiester Bonds in DNA Following

Hydrolysis of N-Glycosylic Bonds (AP Endonucleases) . . . . . . A. DNA Glycosylase-Associated AP Endonuclease Activities . . . . B. A P Endonucleases Associated with Other Catalytic Activities . . C. AP Endonucleases Without Associated Catalytic Activities. . . . D. Concluding Comments on AP Endonucleases . . . . . . . . . . 111. Enzymes That Attack Phosphodiester Bonds in Damaged DNA with Intact N-Glycosylic Bonds . . . . . . . . . . . . . . . . . . . A. UV Endonuclease of Escherichia coli . . . . . . . . . . . . . B. UV Endonuclease Activity in Other Organisms . . . . . . . . . C. Endonuclease V of Escherichiu coli . . . . . . . . . . . . . . D. Endonuclease Activity Directed at Photoalkylated Purines in DNA Note Added in Proof . . . . . . . . . . . . . . . . . . . .

1.

252 252 263 267 273 274 274 277 277 278 279

Introduction

The excision of damaged or inappropriate nucleotides from DNA can occur by a number of different biochemical pathways, depending on both the nature of the specific base damage in question and on the particular organism under investigation (I). Evidence to date indicates that an integ1. P. C. Hanawalt, P. K. Cooper, A. K. Ganesan, and C. A. Smith,Annu. Rev. Biochem. 48, 783 ( 1979).

25 1 THE ENZYMES,Val. XIV Copyright 01981 by Academic Press, Inc. All rights of reproduction in any form reserved ISBN 0-12-122714-6

252

E. FRIEDBERG, T. BONURA, E. RADANY, J. LOVE

ral component of most (if not all) pathways of excision repair of DNA is the enzyme-catalyzed hydrolysis of phosphodiester bonds by specific enzymes that we collectively designate as DNA-incising uctivities . Such enzymes fall into two major classes: ( I ) Those that attack phosphodiester bonds in DNA subsequent to the hydrolysis of the associated glycosylic bond that links a nitrogenous base to the deoxyribose-phosphate backbone: and (2) those that directly attack phosphodiester bonds in damaged DNA. The former class of enzymes is generally designated as AP (apurinicl apyrimidinic) endonucleases (2 ) because their endonuclease activity is confined to sites of base loss in DNA. Such substrate sites arise by the spontaneous hydrolysis of N-glycosylic bonds in DNA (the rate of which is markedly accelerated by certain chemical modifications of bases), or by enzyme-catalyzed hydrolysis of these bonds by DNA glycosylases (31. Most known AP endonucleases appear to be free of known DNA glycosylase activities. However, other AP endonucleases have not yet been physically separated from certain specific DNA glycosylases . Until evidence to the contrary is available, we will consider DNA glycosylaseassociated AP endonucleases as a distinct category of DNA-incising activities. In this chapter attention is focused principally on DNA-incising activities from Escherichicr coli, since this organism has been the subject of extensive genetic and biochemical investigations on the excision repair of DNA ( I ) . However, where appropriate, enzyme activities from other biological sources are considered. II. Enzymes That Attack Phorphodiester Bonds in DNA Following Hydrolysis of N-Glycosylic Bonds (AP Endonucleases)

A. DNA GLYCOSYLASE-ASSOCIATED AP ENDONUCLEASE ACTIVITIES 1. UV DNA-lticising Activity .from Bncteriophage T4-l~fected Esclierichicr coli

The earliest report of a gene product coded by phage T4 involved in the excision of pyrimidine dimers from UV-irradiated DNA, was by Takagi et 2. S . Linn, ii7 "DNA Repair Mechanisms" (P. C. Hanawalt, E . C. Friedberg, and C. F. Fox, eds.), p . 175. Academic Press, New York, 1978. 3. T. Lindahl, Progr. Nircleic Acids Res. Mol. B i d . 22, 135 (1979).

14. ENZYMES THAT INCISE DAMAGED DNA

253

in 1968 (4). These investigators reported that incubation of UVirradiated DNA with extracts of TCinfected E. coli resulted in the loss of thymine-containing pyrimidine dimers from the acid-insoluble fraction of the DNA. Subsequently it was demonstrated that extracts of TCinfected E. coli contain an activity that catalyzes the preferential nicking of UVirradiated duplex DNA (5-7). This activity was shown to be dependent on a functional denV gene (8) formerly called the v gene (9) of phage T4. The isolation of denV gene mutants that are temperature-sensitive with respect to both UV sensitivity and UV DNA-incising activity provided direct evidence that the denV locus is the structural gene for an enzyme activity required for the incision of UV-irradiated DNA (10). This activity has been variously referred to in the literature as the T4 UV endonuclease (5, 6 , I / ) ,or endonuclease V (12, 13) of phage T4. For reasons that will become evident in the course of later discussion, we suggest that it is more appropriate to refer to this activity as the T4 UV DNA-incising activity. The T4 U V DNA-incising activity has been extensively purified in a number of laboratories, (6, 7, 12, 14, 1 3 , generally using assays that directly measure hydrolysis of phosphodiester bonds in UV-irradiated DNA. An alternative assay described by Seawell et ul. (16) measures the direct binding of protein to UV-irradiated DNA. The interested reader is referred to the references previously cited for detailed descriptions of the purification protocols. It is important to note that in all published purification schemes it has been assumed that the T4 UV DNA-incising activity was an endonuclease that directly catalyzed the hydrolysis of phosphodiester bonds 5’ to pyrimidine dimer sites in DNA. However studies, described later in this chapter, suggest that the T4 UV DNA-incising id.

4. Y. Takagi, M. Sekiguchi, S. Okubo, H. Nakayama, K. Shimada, and S. Yasuda, T. Nishimoto, and H. Yoshihara, CSSSQB 33, 219 (1968). 5. E. C. Friedberg and J . J . King, BBRC 37, 646 (1969). 6. S. Yasuda and M. Sekiguchi, PNAS 67, 1839 (1970). 7. E. C. Friedberg and J . J. King, J . Bacteriol. 106, 500 (1971). 8. W. B. Wood and H. Revel, Bacteriol. Rev. 40, 847 (1976). 9. W. Harm, Virology 19, 66 (1%3). 10. K. Sato and M. Sekiguchi, J M B 102, I5 (1976). 11. K. Minton, M . Durphy, R . Taylor, and E. C. Friedberg, JBC 250, 2823 (1975). 12. S . Yasuda and M . Sekiguchi, BBA 442, 197 (1976). 13. Y. Nishida, S. Yasuda, and M . Sekiguchi, BBA 442, 208 (1976). 14. E. C. Friedberg, A. K. Ganesan, and P. C. Seawetl, “Methods in Enzymology,” Vol. 65, p. 191, 1980. 15. P. C. Seawell, E. C. Friedberg, A. K. Ganesan, and P. C. Hanawalt, in, “DNA Repair: A Laboratory Manual of Research Procedures” (E. C. Friedberg and P. C. Hanawalt, eds.), Dekker, New York, 1981. 16. P. C. Seawell, T. J. Simon, and A. K. Ganesan, Biochemisrry 19, 1685 (1980).

254

E. FRIEDBERG, T. BONURA, E. RADANY, J. LOVE

activity consists of a pyrimidine-dimer-specific DNA glycosylase plus an associated AP endonuclease. T4 UV DNA-incising activity has a calculated MW -18,000 as measured by gel filtration (1l ) .The activity has no requirement for any known cofactor and is fully active in the presence of 10 mM EDTA (6, 7, 11). In the absence of EDTA, activity is slightly stimulated in the presence of Mg2+or Mn2+;other divalent cations and monovalent ions have little effect on its activity (6, 7, I / ) . Yasuda and Sekiguchi (6) reported that the T4 UV DNA-incising activity is sensitive to inhibition in 0.4 mM p-chloromercuribenzoate. However, Friedberg el a/.(14) observed only -20% inhibition by 0.1 mM p-chloromercuriphenylsulfonic acid, and Friedberg and King (7) observed no loss of activity in the presence of 0.1 mM p-chloromercuribenzoate. Enzyme activity has a broad pH optimum between 7-9 with a maximum at pH 7.2 (6, 7, if). The activity is induced very early after infection of E. coli, reaching maximum levels about 10 minutes after infection (7, 12). The T4 UV DNA-incising activity attacks cyclobutyl dipyrimidines (pyrimidine dimers) in both duplex and single-stranded DNA (6, 7, 11, 12). Both linear and circular DNA that contain pyrimidine dimers are recognized as substrate. Dimers that contain glucosylated hydroxymethylcytosine (present in the DNA of the T-even phages) are also recognized as substrate. Single-strand incisions in duplex UV-irradiated DNA are catalyzed 5‘ to pyrimidine dimers (11, 12), leaving 3’-OH and 5’-P termini (11). Within the limits of accurate measurement, saturating amounts of enzyme activity catalyze the formation of 1 nick per pyrimidine dimer (17). When heteroduplex DNA that contains pyrimidine dimers in only one strand is treated with T4 UV DNA-incising activity, all detectable nicks occur exclusively in the dimer-containing DNA strands (18). In addition, enzymatically photoreactivated UV-irradiated DNA is no longer a substrate for the enzyme (19). DNA treated with nitrogen mustard, methyl methanesulfonate, 4-nitroquinoline-l-oxide,mitomycin C, N methyl-N’-nitro-N’-nitrosoguanidine, or ionizing radiation is not a substrate for the enzyme (13, 2 0 , 2 / ). These studies indicate a strict specificity of the enzyme activity for cyclobutyl dipyrimidines in DNA, although the possibility that the enzyme recognizes this photoproduct in RNA has not been excluded. 17. E. C. Friedberg, i n , “Molecular Mechanisms for Repair of DNA’ (P. C. Hanawalt and R. B. Setlow, eds.), p. 125. Plenum, New York, 1975. 18. T. J. Simon, C. A. Smith, and E. C. Friedberg, JBC 250, 8748 (1975). 19. A. K. Ganesan, JME 87, 103 (1974). 20. E. C. Friedberg, Mutation Res. 15, 113 (1972). 21. M. It0 and M. Sekiguchi, Jap. J . Genet. 51, 129 (1976).

14. ENZYMES THAT INCISE DAMAGED DNA

255

Further investigations have demonstrated that this stringent specificity is a property of a DNA glycosylase activity and not of an endonuclease. The pyrimidine dimer-DNA glycosylase function of a UV DNA-incising activity was first demonstrated by Grossmanef 01. (22) and by Haseltineet a / . ( 2 3 ) using a preparation of enzyme from M. lufeus (see Section II,A,2). Direct evidence for such activity in T4 enzyme preparations has been provided by studies from this laboratory ( 2 4 , 2 5 ) .DNA labeled in thymine with tritium was UV-irradiated under photosensitizing conditions t o produce 17% of thymine in thymine-containing pyrimidine dimers. The DNA was incubated with saturating amounts of T4 UV DNA-incising activity dnd then reirradiated at 254 nm to photoreverse pyrimidine dimers (Fig. 1). This protocol was found to promote the release of free thymine from DNA. The generation of free thymine was shown to be strictly dependent on the presence of thymine-containing pyrimidine dimers in the substrate DNA, incubation with T4 denV gene product, and photoreversal (24, 35). These results are consistent with the notion that the thymine detected is derived from pyrimidine dimers in which N-glycosylic bonds were hydrolyzed by catalytic action of the T4 enzyme preparation. Experiments have also demonstrated that the amount of radioactivity measured in free thymine is one-half the amount lost from thyminecontaining pyrimidine dimers by photoreversal at several reirradiation fluences (Fig. 2). The amount of free thymine liberated from enzymetreated DNA irradiated with 12,000 JimZof photoreversing fluence corresponds to 60% of the thymine-containing pyrimidine dimers, or 5% of the total radioactivity. Two important conclusions can be drawn from this data. First, since no other photoproducts are known to be present at such high concentrations in UV-irradiated DNA, it is highly improbable that free thymine is derived from any source other than thymine-containing pyrimidine dimers, especially in light of the kinetic correlation between free thymine release and pyrimidine dimer photoreversal. Second, in the experiment previously quoted, in which enzyme-treated DNA was exposed to 12,000 J/m2 of photoreversing light, at least 60% of thyminecontaining pyrimidine dimers in DNA must have had an N-glycosyl bond hydrolyzed by T4 enzyme treatment, so pyrimidine dimer-DNA glycosylase activity cannot be a minor component of T4 UV DNA-incising 22. L. Grossman, S . Riazuddin, W. A. Haseltine, and C. Lindan, CSHSQE 43, 947 (1979). 23. W. A. Haseltine, L. K. Gordon, C. P. Lindan, R. H. Grafstrom, N. L. Shaper, and L. Grossman, Nufure (London) 285, 634 (1980). 24. E. H. Radany and E. C. Friedberg, Natrtre (London) 286, 181 (1980). 25. E. H . Radany, J. D. Love, and E. Friedberg,in, “Chromosome Damage and Repair” (E. Seeberg and K. Kleppe, eds.). Plenum, New York, in press.

256

E. FRIEDBERG, T. BONURA, E. RADANY, J. LOVE

PYRIMIDINE DlMER DNA GLYCOSYLASE

3' A P ENDONUCLEASE

MONOMER IZATION OF THYMINE DlMER

FIG. 1. Model for two-step incision of UV-irradiated DNA. The figure shows a diagrammatic representation of part of a single strand of DNA that contains a thymine-thymine dimer (a). The dimer is shown to be attacked successively by a pyrimidine dimer-DNA glycosylase (which specifically hydrolyzes the 5' glycosyl bond of the dimer) (b), and by an AP endonuclease (c). The latter is shown to cleave the phosphodiester backbone 3' to the apyrimidinic site. [Also shown (d), is the predicted release of free thymine following monomerization of the dimer after it has been attacked by the DNA glycosylase activity.] From Grossman et a / . ( 2 2 ) and Haseltine et a / . (23). activity. This would be expected, since genetic studies have shown that both pyrimidine dimer-DNA glycosylase activity (3)and UV DNAincising activity (5) are associated with the denV gene of phage T4. Finally, the 1 : 2 stoichiometry between radioactivity in free thymine following photoreversal, and that in thymine-containing pyrimidine dimers lost following photoreversal, indicates that only one of the N-glycosylic bonds

14.

257

ENZYMES THAT INCISE DAMAGED DNA

c

w

I

Y

-

. .

PP

‘ h b m --a v)

0

1

2

3 4 5 6 7 PHOTOREVERSAL OF FLUENCE ( kJ/m2)

8

9

FIG.2. Quantitative relationship between the amount of radioactivity measured in free thymine and that associated with the reversal of thymine-containing pyrimidine dimers. The latter parameter is calculated from the measured loss of thymine-containing dimers in ENA. (From Radany ef a / . (15); reproduced with permission of the publishers.) in the dimer is hydrolyzed by the enzyme. The previously documented evidence that nicking of UV-irradiated DNA is catalyzed 5’ with respect to pyrimidine dimers ( I ] , 12) indicates that the 5’-glycosyl bond is uniquely hydrolyzed by the enzyme preparation. Further evidence for such a proposed mechanism of action of the T4 UV DNA-incising activity stems from studies by Seawell et al. (16) who have shown that when UV-irradiated DNA is incubated with T4 enzyme preparations at low temperature, alkaline-labile (presumably apyrimidinic) sites are demonstrable. In addition, the hydrolysis of phosphodiester bonds in such a substrate is effected under near neutral pH by subsequent incubation of the DNA with exonuclease 111 of E. coli. (Exonuclease 111 contains the quantitatively major AP endonuclease activity of E. coli; see Section II,B,l). Warner et d.(26) have also provided evidence for a T4 pyrimidine 26. H. R . Warner, B. F. Demple, W. A. Deutsch, C. M. Kane, and S. Linn, PNAS, in press.

258

E. FRIEDBERG, T. BONURA, E. RADANY, J. LOVE

dimer-DNA glycosylase. They utilized the observation that DNA polymerase I ofE. coli catalyzes DNA strand elongation at a nick containing a 3' terminal nucleotide significantly more rapidly than at a nick containing a 3' terminal apurinic or apyrimidinic site. When testing the primer-template activity of duplex UV-irradiated DNA incubated with T4 UV DNA-incising activity, they observed it to be low. However, the addition to the T4 enzyme-treated DNA of an AP endonuclease that catalyzes incision of DNA on the 5' side of a site of base loss (5' AP endonuclease) restored normal priming activity for DNA polymerase I. The authors reasoned that the original incubation of UV-irradiated DNA with T4 enzyme must have generated nicks with 3' terminal AP sites, which were then excised by the subsequent addition of 5' AP endonuclease (Fig. 3). The formation of AP sites in UV-irradiated DNA by the T4 enzyme implied the catalytic hydrolysis of one or more DNA glycosyl bonds, presumably at pyrimidine dimers. Warner et al. (26) also demonstrated the photoreversal-dependent release of free thymine from UVirradiated DNA incubated with purified T4 enzyme preparations. The release of free thymine following the monomerization of thyminecontaining pyrimidine dimers in DNA provides a convenient assay for the purification of the TCpyrimidine dimer-DNA glycosylase activity. Such studies are in progress. The incision of UV-irradiated DNA by T4 enzyme preparations demonstrates that the preparations purified thus far contain AP endonuclease activity in addition to the pyrimidine dimer-DNA glycosylase. Attempts to physically separate T4 pyrimidine dimer-DNA glycosylase and AP endonuclease activities have not been successful.

2 . UV DNA-Incising Activity from Micrococcus luteus The discovery of enzyme activity from extracts of M. luteus (formerly called M. lysodeikticus) that preferentially attacks UV-irradiated DNA dates back to a report by Strauss (27) in 1962. Subsequent evidence for such an activity has been reported by a number of laboratories (4,28-35). A UV DNA-incising activity assumed to be a direct-acting endonuclease 27. B. S. Strauss, PNAS 48, 1670 (1%2). 28. A. Rorsch, C. van der Kamp, and J. Adema, BBA 80, 346 (1964). 29. R. L. Elder and R. F. Beers, J . Bacrerriol. 90, 681 (1965). 30. B. Strauss, T. Searashi, and M. Robbins, PNAS 56, 932 (1966). 31. E. Moriguchi and K . Suzuki, BBRC 24, 195 (1%6). 32. H. Nakayarna, S. Okubo, M . Sekiguchi, and Y. Takagi, BBRC 27, 217 (1%7). 33. K. Shimada, H. Nakayama, S. Okubo, M. Sekiguchi, and Y. Takagi, BBRC 27, 539 (1967). 34. S. Okubo, H. Nakayama, M. Sekiguchi, and Y. Takagi, BBRC 27, 224 (1%7). 35. W. L. Carrier and R . B . Setlow, EBA 129, 318 (1966).

14. ENZYMES THAT INCISE DAMAGED DNA

259

5’ A P ENDONUCLEASE

FIG. 3. The figure demonstrates the presence of a putative 3’-acting AP endonuclease associated with the phage T4 pyrimidine dimer-DNA glycosylase. The former activity produces a 3’ terminal deoxyribose-phosphate residue (a), which is a poor primer for DNA polymerase I of E. coli. However, following fiwther incubation with a S’-AP endonuclease, the deoxyribosephosphate residue is excised (b), leaving a normal 3’ terminus. The template DNA strand is omitted for clarity. specific for pyrimidine dimers was first purified by Kaplan et a/. (36) and by Carrier and Setlow (37). In 1971 Nakayama et a/. (38)reported that two species of U V DNA-incising activity could be distinguished by isoelectric focusing and by column chromatography, and in 1977 Riazzudin and Grossman (39) reported the extensive purification of two species of U V DNA-incising activity from M. luteus. which they designated as “correndonucleases” I and 11. A mixture of the two “correndonucleases” from M. luteus was used by Grossman ef id. (22) and Haseltine et a / . (23) to explore the distribution of pyrimidine dimers in a sequenced segment of the operator-promotor region of the E. coli lac operon. The rationale of these experiments was that enzyme-catalyzed nicking of the sequenced DhA immediately 5’ to all pyrimidine dimers by a direct-acting endonuclease (as anticipated by the investigators), or by a combined 5’-DNA glycosylase plus 5’-AP endonuclease (as shown in Fig. 4) should, following denaturation of the duplex DNA, yield fragments of precisely predictable length from the “P-labeled 36. 37. 38. 39.

J. C. Kaplan, S. R. Kushner, and L. Grossman, PNAS 63, 144 (1969). W. L. Carrier and R. B. Setlow, J. Bucteriol. 102, 178 (1970). H. Nakayama, S. Okubo, and Y. Takagi, BBA 228, 67 (1971). S. Riazzudin and L. Grossman, JBC 252,6280 (1977).

260

E. FRIEDBERG, T. BONURA, E. RADANY, J. LOVE

1 FIG. 4. This figure schematically indicates the method by which Haseltine et c d (23) detected a pyrimidine dimer-DNA glycosylase activity in enzyme preparations from M . lrrterrs . The bars under the polynucleotides indicate relative lengths of 5’ radiolabeled fragments (*) following gel electrophoresis of the denatured products of incubation of UVirradiated DNA with M. hrterw “correndonucleases.” Note that the concerted action of a S’-dimer-DNA glycosylase and a 3’-AP endonuclease generates fragments, the size of which is reduced at their 3’ ends by I nucleotide after treatment with alkali due to the presence of an alkaline-labile 3’ terminal apyrimidinic site. Enzyme-catalyzed hydrolysis of phosphodiester bonds by a 5’-endonuclease, or direct hydrolysis of phosphodiester bonds acting immediately S to the dimers, would not result in the alkaline-dependent size reduction of DNA fragments observed.

-

5’ ends (Fig. 4). Surprisingly, these investigators observed by agarose gel electrophoresis that the labeled DNA fragment lengths were -1 nucleotide longer than expected, suggesting that phosphodiester bond hydrolysis catalyzed by the UV DNA-incising activity was between the dimerized pyrimidines rather than on their 5‘ side. The predicted size distribution of denatured DNA fragments resulting from incubation of the UV-irradiated DNA with enzyme was obtained if prior to gel electrophoresis the DNA was subjected to treatment in strong alkali. This result suggested that alkaline-labile sites (presumably sites of depyrimidi-

14. ENZYMES THAT INCISE DAMAGED DNA

26 1

nation) were present 5‘ to the sites of phosphodiester bond hydrolysis (Fig. 4). Based on these findings, these investigators proposed the twostep model for the incision of DNA that contains pyrimidine dimers, shown in Fig. 1. The previously reported isolation of two “correndonuclease” activities may reflect the purification of two AP endonucleases from M. luteus, each of which was contaminated with a single pyrimidine dimer-DNA glycosylase. In 1981 Pierre and Laval (40,4/) reported the purification and characterization of two chromatographically distinct AP endonucleases from M. lureus (AP endonucleases A and B), one with a PI of 4.8 and the other with a PI of 8.8 (see Section 11,C,2). These values correspond very closely with the PIS of 4.7 and 8.7 originally reported for “correndonucleases” I and 11. However, Pierre and Laval (40,411 reported that both AP endonucleases catalyze phosphodiester bond breakage 5‘ to sites of base loss. This observation is inconsistent with the results of the gel electrophoresis of DNA reported by Grossman el al. (22) and Haseltine et al. (23 ), which showed alkaline-labile sites exclusively at the 3’ end of DNA fragments, indicating the action of 3‘-AP endonuclease. Hopefully further studies on the AP endonucleases of M. luteus will resolve this apparent contradiction. Haseltine et (11. (23) purified the M. luteus pyrimidine dimer-DNA glycosylase further and achieved a substantial reduction in AP endonuclease activity. Although the most purified fraction still contains detectable AP endonuclease, it is uncertain whether or not the M. luteus pyrimidine dimer-DNA glycosylase contains a physically associated AP endonuclease activity. In this respect it is of interest to note that Tomilinet al. (42) isolated only a single “UV endonuclease” activity fromM. luteus that also contained AP endonuclease activity. The kinetics of heat inactivation of both the “UV endonuclease” and AP endonuclease activities were identical. Furthermore, the two catalytic activities were inactivated at the same concentrations of cyanide and proflavin. Tomilin et al. (42) also reported that the UV DNA-incising activity of their M. luteus fraction was competed for by the addition of duplex DNA that contained apurinic sites to enzyme reactions with UV-irradiated DNA. While these experiments do not provide definitive evidence that the AP endonuclease and UV DNAincising activity are resident in the same molecule, they indicate that the 40. J. Pierre, and J. Laval, Biochemistry, in press. 41. J. Pierre, and J. Laval, Biochemistry, in press. 42. N. V. Tomilin, E. B. Raveltchuk, and T. V. Mosevitskaya, Eur. J . Biochem. 69, 26s (1976).

262

E. FRIEDBERG, T. BONURA, E. RADANY, J. LOVE

AP endonuclease activity in question is required for UV DNA-incising activity. The precise nature of the physical relationship between pyrimidine dimer-DNA glycosylase and AP endonuclease activities in the T4 and M. Iuteus enzyme preparations is obviously an important area for further detailed investigation. (See Note Added in Proof, p. 279.) 3 . Endonuclease III of Escherichia c d i

Endonuclease 111 of E. coli is another enzyme activity in which AP endonuclease and DNA glycosylase functions have not been physically separated. This activity was originally described by Radman (43) who was searching for endonucleases involved in the correction of mismatched base pairs in DNA and used heavily UV-irradiated DNA as a substrate. These studies led to the isolation of a DNA incising activity that attacks UV-irradiated, but not unirradiated, DNA. However, in contrast to UV DNA-incising activities specific for pyrimidine dimers in DNA, extensive photoreactivation of UV-irradiated DNA did not eliminate substrate sites value for endonuclease 111. The enzyme activity was shown to have a szo,w of 2.6, with no cofactor requirement. The pH optimum was -7.0 and activity was inhibited in the presence of tRNA or 1.0 M NaCl. Independent investigations by Gates and Linn (44)led to the isolation of a DNA incising activity with properties indistinguishable from those of endonuclease 111. These investigators purified the enzyme approximately 1600-fold, but not to physical homogeneity. The purified enzyme was clearly shown to recognize a photoproduct in DNA other than thyminecontaining pyrimidine dimers. Thus, PM2 DNA irradiated under standard conditions contained -85 thymine-containing pyrimidine dimers and one enzyme-sensitive site per molecule. However, when dimers were introduced into DNA by irradiation at 309 nm in the presence of a photosensitizer (a protocol that results in far fewer nondimer photoproducts in DNA), the resulting substrate contained -95 thymine-containing pyrimidine dimers per PM2 DNA molecule, but was not nicked by preparations of endonuclease 111. Gates and Linn examined the activity of endonuclease I11 on a variety of DNA substrates and demonstrated nicking on heat- and acid-treated DNA and X-irradiated DNA. Both substrates share in common the presence of sites of base loss. Whereas the enzyme was shown to attack duplex DNA that contained only apurinic sites, alkali-stable lesions in UV-irradiated and Os0,-treated DNA were also attacked by the enzyme preparation. A reasonable explanation for this observation stems from 43. M. Radman, JBC 251, 1438 (1976). 44. F. T. Gates, Ill. and S. Linn, JBC 252, 2802 (1977).

14. ENZYMES THAT INCISE DAMAGED DNA

263

studies by Linn and his colleagues (45, 46) that demonstrate DNA glycosylase activity in purified preparations of endonuclease I11 that excises, 5,6-dihydrothymine and 5,6-dihydroxydihydrothyminefrom DNA as the free base. These products had previously been shown by Cerutti (47) to be present in UV-irradiated or X-irradiated DNA, as well as in DNA treated with OsO,. Attempts to physically separate the DNA glycosylase and AP endonuclease activities have not as yet been successful. Endonuclease I11 is believed to catalyze incision of duplex DNA 3‘ to sites of base loss. This conclusion is based on experiments similar to those previously described (see Section II,A,l and Fig. 3), i.e., apurinic DNA treated with endonuclease 111 is a poor primer-template for E. coli DNA polymerase I. The DNA becomes a much better substrate for polymerization when either endonuclease IV or endonuclease VI (both of which are believed to be 5’-acting AP endonucleases, see Sections II,B, 1 and II,C, 1) is added to the reaction prior to the polymerization reaction. The interpretation offered is that endonuclease I11 leaves a 3’-depurinated terminus that is not favored by DNA polymerase I as a primer for DNA synthesis. The excision of the 3’-deoxyribose-phosphateresidue by the 5’-AP endonuclease action of endonuclease IV or AP endonuclease function of exonuclease I11 leaves a normal 3‘ terminus that is more effectively utilized by E. coli DNA polymerase I.

B. AP ENDONUCLEASES ASSOCIATED WITH OTHER CATALYTIC ACTIVITIES I . AP Endonuclease Activity of Eronuclease III from Escherichin coli The quantitatively major AP endonuclease activity of E. coli has been referred to in the literature as endonuclease 11, endonuclease VI, or as the AP endonuclease function of exonuclease 111 (see Chapter 12 by B. Weiss, this volume). Endonuclease I1 was discovered by Friedberg and Goldthwait (48-50) and defined as an activity that catalyzes the incision of 45. B. Demple and S. Linn,Nature (London) 287, 203 (1980). 46. S. Linn, B . Demple, D. W. Mosbaugh, H. R. Warner, and W. A. Deutsch, in “Chromosome Damage and Repair” (E. Seeberg and K. Kleppe, eds.), Plenum Press, New York, in press. 47. P. Cerutti, Nuturwissenschuften 61, 51 (1974). 48. E. C. Friedberg and D. A. Goldthwait, CSHSQB 33, 271 (1968). 49. E. C. Friedberg and D. A. Goldthwait, PNAS 62, 934 (1969). 50. E. C. Friedberg, S-M. Hadi, and D. A. Goldthwait, JBC 244, 5879 (1%9).

264

E. FRIEDBERG, T. BONURA, E . RADANY, J. LOVE

DNA alkylated with methyl methanesulfonate. Subsequently, Hadi and Goldthwait (51) demonstrated that the activity also attacked depurinated DNA. Independent studies by Verly and his colleagues (52,53)resulted in the isolation and purification of an AP endonuclease originally thought to be distinct from endonuclease I1 sinceall substrate sites in alkylated DNA attacked by this AP endonuclease were shown to be alkaline-labile, i.e., sites of depurination arising from the spontaneous hydrolysis of alkylated purines. In contrast, endonuclease I1 was shown to attack both alkali-labile and alkali-stable sites in alkylated DNA. The discovery of a specific 3-methyladenine-DNA glycosylase in extracts of E. coli (54) provides an explanation for this confusion. Endonuclease I1 of E. coli as originally defined was presumably a mixture of 3-methyladenine-DNA glycosylase and AP endonuclease activities. However, prior to this and other retrospective clarifications about the identity of endonuclease I1 and the major AP endonuclease of E. coli, Verly designated the latter enzyme as endonuclease VI (55 1. We recommend that the terms endonuclease I1 and endonuclease VI be dropped, because the AP endonuclease activity appears to be clearly associated with the previously described enzyme, exonuclease 111. Milcarek and Weiss (56) and Yajko and Weiss (57) isolated a series of mutants of E. coli defective in both the 3’-exonuclease and the associated 3’-phosphatase functions of exonuclease 111. Of the mutants characterized genetically, all mapped to the xthA locus, the structural gene coding for exonuclease 111, which is situated between the pfkB and pncA loci of E. coli at approximately 38.5 min on the E. coli genetic map. Studies with preparations of exonuclease 111 purified to 98% homogeneity demonstrated the presence of AP endonuclease activity that could not be separated from exonuclease and phosphatase activities by electrophoresis, sedimentation, or gel filtration (58). All xthA mutants were defective in exonuclease I11 activity and were also shown to be defective in AP endonuclease activity (58). Exonuclease I11 as purified by Weiss in terms of its exonucleasephosphatase activity is a monomeric protein of MW -28,000. The enzyme

-

51. S-M. Hadi and D. A . Goldthwait, Biochemistry 10, 4986 (1971). 52. W. G. Verly and Y. Paquette, Con. J . Biochem. 50, 217 (1972). 53. Y. Paquette, P. Crine, and W. G . Verly, Can. J . Biochem. 50, 1199 (1972). 54. S . Riazzudin and T. Lindahl, Biochemistry 17, 21 10 (1978). 55. W. G. Verly,in, “DNA Repair Mechanisms” (P. C. Hanawalt, E. C. Friedberg, and C. F. Fox, eds.), p. 187. Academic Press, New York, 1978. 56. C. Milcarek and B . Weiss, J M B 68, 303 (1972). 57. D. M. Yajko and B . Weiss, PNAS 72, 688 (1975). 58. B. Weiss, JEC 251, 1896 (1976).

14.

ENZYMES THAT INCISE DAMAGED DNA

265

has also been purified about 8000-fold by Verly and Rassart (59) as a monomeric protein of similar molecular weight. The AP endonuclease activity is optimal at pH 8.5 and is maximally stimulated by Mg2+ at 20 mM at pH 8.0 (60). In the absence of M&+, activity is present, but is completely inhibited in the presence of EDTA at concentrations > 1 mM. If the enzyme is dialyzed against 10 mM EDTA to remove divalent cations, and then against buffer without EDTA, no enzyme activity is detected (60). Activity can be restored by the addition of Mg2+,but not by Ca2+,Mn'+, Zn2+,or Cu2+. These observations suggest that the enzyme may be a metallo-protein containing bound Mg'+. A number of different experimental approaches have yielded results consistent with the conclusion that the enzyme catalyzes the hydrolysis of phosphodiester bonds 5' to apurinic sites in duplex DNA, leaving 3'-OH and 5'-Ptermini (60). The AP endonuclease activity of exonuclease I11 of E. coli does not degrade native DNA. It is generally assumed that the enzyme attacks sites of pyrimidine or purine loss in duplex DNA with equal facility, however quantitative comparisons of this parameter have not been documented. The enzyme is at least as active on DNA that contains AP sites reduced with sodium borohydride as on the unreduced substrate, indicating that the aldehyde function of C, in deoxyribose is not required for enzyme action. Gossard and Verly (60) suggested that the mechanism of action of the enzyme is probably distinct from the @elimination reaction catalyzed by alkali at sites of base loss in DNA. This conclusion is supported by the observation that alkaline-catalyzed /3-elimination most frequently results in hydrolysis of phosphodiester bonds on the 3' side of sites of base loss (611, and has an apparent requirement for the aldehyde function at C1 of the deoxyribose residue, since reduction of C, to the alcohol prevents the reaction. Hydrolysis of apurinic DNA 5' to sites of base loss would necessitate excision of the deoxyribose-phosphate moiety in the 5' --* 3' direction during the excision repair in vivo of base damage to DNA. Such a biochemical pathway provides no obvious function for the 3' exonuclease function of exonuclease 111. Gossard and Verly (60) suggested that the exonuclease may remove one or more nucleotides from the 3' end of sites of phosphodiester bond hydrolysis, thus precluding ligation of enzymecatalyzed phosphodiester bond breaks. However such a function for exonuclease 111 has not been directly demonstrated, nor is there evidence 59. W. G. Verly and E. Rassart, JBC 250, 8214 (1975). 60. F. Gossard and W. G . Verly, Eitr. J . Biochem. 82, 321 (1978). 61. C. R. Bayley, K . W. Brarnrner, and A. D. Jones,JCS., 1903 (1961).

266

E. FRIEDBERG, T. BONURA, E. RADANY, J. LOVE

that a 5’-Pterminus associated with an apurinic site can be enzymatically ligated to an adjacent 3’-OH end. An intriguing model reconciling the multiple catalytic activities of exonuclease I11 has been suggested by Weiss (58, and Chapter 12, this volume). He proposes that the enzyme recognizes “spaces” in DNA duplexes that provide substrate sites for each of its 3 catalytic domains: The exonuclease recognizes a “space” created by the unwinding of a terminal base pair due to the lowered base-stacking intcractions operating at the ends of DNA duplexes; the AP endonuclease recognizes the “space” created by loss of a base; and the 3’-phosphatase recognizes the “space” corresponding to a “missing” 3’-terminal nucleoside residue. The exact biological role of the AP endonuclease activity of exonuclease I11 is not clear. Surprisingly, mutants of E. coli (including deletion mutants) defective in exonuclease I11 (xth-) are not particularly sensitive to treatment with alkylating agents such as methyl methanesulfonate (3). This may reflect the ability of other AP endonucleases to assume the essential function(s) of this enzyme during the repair of sites of base loss. Alternatively, repair of such sites may occur by the action of DNA purine insertase activity. In contrast,E. coli mutants defective in bothxth and duf genes are only conditionally viable (62). One possible explanation for this difference is that in the absence of AP endonuclease activity, the apyrimidinic sites created by incorporation of dUMP into DNA, and the subsequent excision of uracil by ura-DNA glycosylase, are lethal lesions. This apparent contradiction suggests that the function(s) that operates in the repair of apurinic sites in the absence of AP endonuclease activity is not able to deal effectively with apyrimidinic sites. Such a conclusion is consistent with the apparent failure to discover DNA pyrimidine insertase activity in any biological source, but is equally consistent with the notion that some AP endonucleases do not attack apurinic and apyrimidinic sites in DNA with equal efficiency. 2. Exonuclease III-Like AP Endonucleuse Activity from Other Organisms Haemophilus injuenzae is another biological source from which an AP endonuclease with associated 3’-exonuclease and 3’-phosphatase activities has been isolated. Clements et al. (63) partially purified such an enzyme with a MW -30,000. Divalent cation is required for maximal 62. B. Weiss, S. G. Rogers, and A. F. Taylor, in “DNA Repair Mechanisms” (P. C. Hanawalt, E. C. Friedberg, and C. F. Fox, eds.), p. 191. Academic Press, New York, 1978. 63. J . E. Clements, S. G . Rogers, and B . Weiss, JBC 253, 2990 (1978).

14.

ENZYMES THAT INCISE DAMAGED DNA

267

activity, Mg2+and Mn2+being equally effective at optimal concentrations of 5 mM. Like the AP endonuclease activity ofE. coli, activity is detected in the absence of divalent cation, and the enzyme is totally inactivated in the presence of 1 mM EDTA. AP endonuclease and exonuclease activities cannot be resolved by electrophoresis, sedimentation, or gel filtration. In addition, a mutant of H . injuenzue (corn-lU) known to be defective in exonuclease activity, is also defective in AP endonuclease activity. Like the AP endonuclease activity of exonuclease I11 of E. coli, the AP endonuclease of H. itzfluenzue catalyzes the hydrolysis of phosphodiester bonds 5' to sites of base loss in DNA, creating 3'-OH and 5'-P termini. C. AP ENDONUCLEASES WITHOUTASSOCIATED CATALYTIC ACTIVITIES

I . Etidonuclemse IV qf Escherichicr coli The AP endonuclease activity of exonuclease I11 is quantitatively the major AP endonuclease detectable in crude extracts of wild-type E. coli. In extracts ofxth- mutants defective in exonuclease 111, about 10% of the normal level of AP endonuclease active on duplex AP DNA is detectable, essentially all of which can be accounted for by an activity designated as endonuclease IV (641. Endonuclease IV has been purified about 3000-fold by Ljungquist (641, but this preparation is not physically homogeneous. The enzyme has a relative S value of 3.4 with a Stokes radius of 25 A andf/fo = I . 18. These results are consistent with a globular protein of MW -30,000. The activity has a broad pH optimum between 8.0-8.5 and has no requirement for known cofactors. Activity is sensitive to inhibition by sulfhydryl blocking agents. Normal levels of activity are expressed in the presence of EDTA and there is no detectable stimulation or inhibition of the enzyme activity when EDTA in the standard reaction is replaced by MgCl, (0.5-10 mM) or CaCl, (1.0 mM). Higher concentrations of CaCI, (5.0 mM) are inhibitory. Endonuclease IV activity is unusually resistant to NaCl and retains 50% maximal activity in 0.56 M salt. The enzyme is also quite heat resistant: heating to 60" for 5 minutes in the presence of 0.2 M NaCl results in no detectable loss of activity. Activity is not affected by tRNA at concentrations between 20-500 pglml. Endonuclease IV does not attack covalently closed circular PM2 DNA but does catalyze the stoichiometric nicking of such DNA when it con64. S. Ljungquist, JBC 252, 2808 (1977).

268

E. FRIEDBERG, T. BONURA, E. RADANY, J. LOVE

tains apurinic sites. The enzyme also degrades heavily UV-irradiated or X-irradiated DNA, however all substrate sites in such DNA are satisfactorily accounted for as sites of base loss. The polarity of DNA incision relative to sites of base loss, and the chemistry of the termini created, have not been reported. However, Linn and his colleagues (26, 46) have observed that the addition of endonuclease IV to reactions of apurinic DNA with endonuclease I11 enhances the priming activity of the product for DNA synthesis by E. coli DNA polymerase I (see Section II,A,3). This result suggests that like the AP endonuclease activity of exonuclease 111, endonuclease IV catalyzes hydrolysis of phosphodiester bonds 5’ to sites of base loss in DNA. 2 . AP Eiidotiuclecises of Micrococcus luteus Two AP endonucleases have been extensively purified and characterized from M . luteus and are referred to as AP endonucleases A and B (4U, 4 1 ) . As judged by polyacrylamide gel electrophoresis, each is a monomeric protein of MW -35,000 and neither protein has detectable nonspecific endonuclease, exonuclease, phosphatase, 3-methyladenine-DNA glycosylase, or uracil-DNA glycosylase activity. The two AP endonuclease activities are distinguishable by a different affinity for phosphocellulose and by markedly different isoelectric points (4.8 and 8.8 for enzymes A and B, respectively). In addition, AP endonuclease A has a half-life of 4 minutes at 45”, whereas AP endonuclease B has a half-life of 10 minutes at the same temperature. Activity A has a narrow pH optimum around 7.5, whereas activity B has a broad pH optimum between 6.5-8.0. Neither requires divalent cations for activity, although both are stimulated by Mg’+, are inhibited by Ca2+ or Mn”, and are inhibited 90% in the presence of 0.4 mM EDTA. A number of distinct experimental approaches indicate that both activities catalyze the hydrolysis of phosphodiester bonds 5’ to sites of base loss in duplex DNA (however see discussion in Section 11,A,2), leaving 3’-OH and 5’-P termini. Whether or not AP endonucleases A and B represent independently coded gene products is an important question that remains to be resolved. Conceivably one is a posttranslationally modified form of the other, or arises as an artifact of proteolysis in cell-free preparations of M. luteus. It would be of distinct interest to examine the antigenic relatedness of these two enzymes. AP endonuclease activities with properties similar to those of endonuclease IV ofE. coli have been purified and characterized to varying degrees from a number of other biological sources, including calf thymus, calf liver, plants, mouse epidermal cells, S . cerevisiae, B. subtilis, B.

14. ENZYMES THAT INCISE DAMAGED DNA

269

stenrothermophilits, and human placenta. A summary of the essential features of these enzymes (65-77) is presented in Table I.

3. Etidotzucleiise VII of Escherichicr coli

The AP endonucleases thus far discussed have been characterized in terms of their activity on duplex DNA that contains sites of base loss. Studies in this laboratory have led to the identification of an AP endonuclease that catalyzes the degradation of single-stranded polydeoxyribonucleotides or DNA that contains depyrimidinated sites, but not of intact single-stranded polymers or DNA. This activity is present in extracts of both wild type and xth- strains of E. coli and has been designated as endonuclease VII of E. coli (78-79~). The standard assay of endonuclease VII activity measures the degradation to acid-soluble products of poly(dU) . [3H](dT)200that contain depyrimidinated sites produced by the excision of uracil from the polymer by preincubation with purified ura-DNA glycosylase. The detection of endonuclease VII in crude extracts of wild-type E. coli by this assay is complicated by the presence of two contaminating activities. One is due to low molecular weight basic proteins and polyamines that are known to promote p-elimination reactions at sites of base loss (3) and result in the selective degradation of the depyrimidinated polymer. Unlike endonuclease VII, however, these substances do not bind to DEAE-cellulose and are readily separated from the enzyme by column chromatography. The 65. L. Thibodeau and W. G. Verly, FEBS (Fed. Eur. Biochem. Soc.) Lett. 69, 1 (1976). 66. L. Thibodeau and W. G. Verly, JBC 252, 3304 (1977). 67. V. Bibor and W. G . Verly, JBC 253, 850 (1977). 68. T. Inoue and T. Kada,JBC 253, 8559 (1978). 69. P. R. Armel and S. S. Wallace, Nucleic Acids Res. 9, 3347 (1978). 70. A. B . Futcher and A. R. Morgan, Can. J. Biocllem. 57, 932 (1979). 71. S . Ljungquist and T. Lindahl, JBC 249, 1530 (1974). 72. S. Ljungquist, B. Nyberg, and T. Lindahl,FEBS (Fed. Eur. Biochem. S o c . ) L e t t . 57, 169 ( 1975). 73. J. P. Kuebler and D. A. Goldthwait, Biochemistry 16, 1370 (1977). 74. G. Ludwig and H. W. Thielmann, Nucleic Acids Res. 6, 2901 (1979). 75. T. P. Brent, Niicleic Acids Res. 4, 2445 (1977). 76. U . Kuhnlein, E. E. Penhoet, and S. Linn, P N A S 73, 1169 (1976). 77. W. S. Linsley, E. E. Penhoet, and S. Linn,JBC 252, 1235 (1977). 78. E. C. Friedberg, T. Bonura, R. Cone, R. Simmons, and C. Anderson, in "DNA Repair Mechanisms" (P. C. Hanawalt, E. C. Friedberg, and C. F. Fox, eds.), p. 163. Academic Press, New York, 1978. 79. E. C . Friedberg, C. T. M. Anderson, T. Bonura, R. Cone, E. H. Radany, and R. J. Reynolds, Progr. Nucleic Acid Res. M o l . Biol., 26, 197 (1981). 79a. T. Bonura, R. Schultz and E. C. Friedberg, in preparation.

TABLE I

AP ENDONUCLEASES FROM A VARIETY OF BIOLOGICAL SOURCES

Source

Molecular weight

Cofactor requirement

Plmsoeolirs rnti(tifl0ru.r

40,000

None

R . stearothertno~kilrrs

28.000

None

56,000

None

Stimulation or inhibition of activity Mg2+or Mn2+stimulate about Cfold; inhibited in presence of NaCl Stimulated by monovalent cations: inhibited by divalent cations Stimulated by 50 mM NaCl, and by Mg2+,Mn2+or Ca2+:inhibited in presence of 500 mM NaCl

General comments

Reference

Temperature optimum at 60" -

S. cerei7isine

(Endonuclease D) (Endonuclease E)

MgZ+ None None

S . rereifsine

Calf thymus Calf liver Mouse epidermis

35,000 28 ,000 3 1 .OW

Human lymphocytes Human skin fibroblasts

Human placenta

Mgz+ Mg2+ None MgZ+

27,000-31,000

None

Inhibited in presence of 400 mM NaCl Inhibited by 1 mM EDTA Inhibited by Ca2+,EDTA, or tRNA Stimulated by Mg2+,or KCI: inhibited by high ionic strength Inhibited by EDTA

Stimulated by Mg2+, Mn2+,Coz+, Zn2+ or 40 mM KCl: inhibited by CaZ+or EDTA

Two activities identified in extracts Temperature optimum at 40" Optimal pH at 9.5 -

Activity in XP group A and D cells has 5 to 10-fold higher K , than normal Six chromatographically separable forms identified

(76)

(77)

14.

ENZYMES THAT INCISE DAMAGED DNA

27 1

second contaminating activity is due to exonuclease VII in extracts of E. coli. This enzyme catalyzes the exonucleolytic degradation of both the intact and the depyrimidinated polymer in the presence of EDTA (80). Exonuclease VII also binds to DEAE-cellulose, but is totally separable from endonuclease VII by successive chromatography on DEAEcellulose and phosphocellulose. Endonuclease VII has been purified free of nonspecific endonuclease and has a relative S value of 4.2 measured by sedimentation in 15-40% glycerol gradients in the presence or absence of 0.25 M KCI.A molecular weight of -55,000 has been calculated from gel filtration of the enzyme, relative to appropriate standard markers. Activity is unaffected by the presence of 5 mM EDTA in the standard reaction. In the absence of EDTA, activity is slightly stimulated by 5 mM MgC1, and by 1-2 mM CaCI,. Endonuclease VII activity is inhibited 50% in the presence of 0.15 M NaCl or in the presence of 0.58 nM (as nucleotide) tRNA. Endonuclease VII can be differentiated from previously described AP endonucleases of E. coli by a number of criteria (Table 11). Note that the only other enzymes active in the presence of EDTA are endonucleases I11 and IV. Both have been directly examined for activity against depyrimidiand found to be inactive under conditions that nated poly(dU) . [3H](dT)200 promote readily detectable degradation of depurinated duplex PM2 DNA. The substrate specificity of endonuclease VII is uncertain. In addition to the single-stranded depyrimidinated heteropolymer mentioned above, the enzyme degrades depyrimidinated poly(dU) and single-strandeddepyrimidinated PBS2 DNA, as well as single-stranded circular M 13 DNA that contains depyrimidinated sites. (The latter substrate is produced by growing phage M 13 in a strain of E. coli defective in dUTPase and in uracil-DNA glycosylase activities. The resulting phage DNA contains dUMP residues that can be converted to depyrimidinated sites by preincubation with uracil-DNA glycosylase .) Duplex form I PM2 DNA that contains apurinic sites (created by beat and acid treatment of the DNA) is not attacked by the enzyme, nor is duplex PBS2 DNA containing depyrimidinated sites a substrate for endonuclease VII. Depyrimidinated sites in single-stranded DNA that have been reduced with sodium borohydride are also not attacked by the enzyme. The reduction of the aldehyde function of the deoxyribose moiety to the alcohol stabilizes the associated phosphodiester bond against p elimination. Thus it is possible that endonuclease VII catalyzes phosphodiester bond hydrolysis by a p elimination mechanism. Single-stranded DNA substrates containing apurinic sites have not yet been examined with endonuclease VII. 80. J . W. Chase and C. C. Richardson, JBC 249, 4545 ( 1974).

TABLE I1

APURINIC/hYRMIDINIC (AP) ENDONUCLEASES OF Escherichio coli Enzyme Endonuclease 111

Endonuclease IV

Endonuclease V'

AP Endonuclease of Exonuclease 111 Endonuclease VI1

Principal properties No requirement for divalent cation: inhibited in presence of tRNA; 2.7 S: MW -27,000; pH optimum -7.0; requires duplex DNA with AP sites No requirement for divalent cation; not stimulated by Mg2+ or Ca2+:no inhibition in presence of tRNA: 3.4 S: MW -33,000; pH optimum -8.0-8.5; inhibited in presence of PCMB; requires duplex DNA with AF' sites Requires Mg2+for activity: inhibited in presence of tRNA: 2.3 S: MW -20,000: pH optimum 9.25; requires duplex DNA with AP sites Requires Mg2+for optimal activity; inhibited in presence of EDTA; 2.9 S ; MW -32,000 pH optimum -8.5 ; requires duplex DNA with AP sites No requirement for divalent cation; stimulated by Mg2+or Caz+;inhibited in presence of tRNA; 4 . 3 S ; MW -45,000: pH optimum -7.0; insensitive to PCMB; attacks single-stranded DNA and polydeoxypyrimidines with depyrimidinated sites; duplex DNA with apurinic sites is not a substrate

Associated catalytic activities

Reference

DNA glycosylase activity that recognizes 5,bsaturated thymine photoproducts None detected

(43,4446)

Acts on a variety of damaged DNAs as well as certain native DNAs 3', 5'-exonuclease (exonuclease HI), 3'-phosphatase, RNase H Not determined

(48-53 ) ( 5 5 4 U ,6 2 ) (78-790)

" Endonuclease V is not considered in the text as an AP endonuclease, but is included in the table for the sake of completeness since it has been reported to attack DNA heated at pH 5.2. From Gates and Linn (99).

14. ENZYMES THAT INCISE DAMAGED DNA

273

D. CONCLUDING COMMENTS ON AP ENDONUCLEASES AP endonucleases, discovered in the late 1960s, were originally thought to constitute a specific class of endonucleases required for the repair of sites of base loss that arose spontaneously in either native DNA, or in DNA treated with chemical agents (such as certain alkylating agents) that promote increased lability of N-glycosylic bonds. The discovery of DNA glycosylases as a general class of enzymes involved in the excision of damaged or inappropriate bases (8/) provided a more obvious role for AP endonucleases in DNA repair. The findings reviewed above, that some AP endonuclease activities may be physically associated with specific DNA glycosylase activities, suggests that this class of enzymes may be fundamental to understanding the enzymology of the incision of a major fraction (if not all) of DNA damage in living cells. Thus for example, it is possibly significant that endonuclease I11 (which apparently has an associated DNA glycosylase activity specific for monoadduct thymine photoproducts with saturated 5,6 bonds) and the AP endonucleases associated with the pyrimidine dimer-DNA glycosylases of phage T4 and M. luteus all catalyze the hydrolysis of phosphodiester bonds 3’ to sites of base loss. On the other hand endonuclease IV and the AP endonuclease activity of exonuclease 111, both of which are thought to be 5’-acting AP endonucleases, are apparently not associated with DNA glycosylase activities. Conceivably 3’-acting AP endonucleases function exclusively in concert with specific DNA glycosylases, while the 5’-AP endonucleases are more general in their action. Clearly it is necessary to establish more information on the polarity of DNA incision by all known AP endonucleases and to survey them thoroughly for associated DNA glycosylase activity. It also remains to be satisfactorily demonstrated whether or not all AP endonucleases attack sites of purine and pyrimidine loss with equal facility. This potential basis for substrate specificity may provide further understanding of the biological relevance of the multiplicity of AP endonucleases detected in different organisms. In this regard it should be noted that with the exception of the AP endonuclease activity of exonuclease 111, no mutants of E. co/i have been isolated that are defective in AP endonucleases. Such strains would be of enormous value in understanding the biological role of the various AP endonucleases in this organism. Finally, in view of the apparent central role that AP endonucleases play in the excision repair of DNA, the importance of making the distinction between a direct-acting endonuclease and an AP endonuclease that acts subsequent to a physically associated or distinct DNA glycosylase, is 81. T. Lindahl, Nuture (London) 259, 64 (1976).

274

E. FRIEDBERG, T. BONURA, E. RADANY, J. LOVE

obvious. All newly discovered endonuclease activities that attack modified or damaged DNA should be very carefully investigated from this point of view.

111.

Enzymes That Attack Phosphodiester Bonds in Damaged DNA with Intact N-Glycosylic Bonds

OF Escherichiu coli A. UV ENDONUCLEASE

The phenomenon of pyrimidine dimer excision during post-UV incubation of irradiated E. coli was reported as early as 1964 (82, 83). The isolation of a series of UV-sensitive mutants defective in pyrimidine dimer excision established the involvement of at least 3 genes (uvrA, -B, and -C) in the incision of UV-irradiated DNA (I). However the isolation of the products of these genes and the direct demonstration of UV DNA-incising activity in cell-free preparations was achieved much later. Braun and Grossman (84)reported the partial purification of an endonuclease activity that selectively attacked UV-irradiated and not unirradiated DNA. This activity also has the property of binding to UVirradiated DNA that contains pyrimidine dimers . When irradiated DNA was subjected to monomerization of pyrimidine dimers by incubation with yeast photoreactivating enzyme the binding activity was lost. The activity was not detected in extracts of uvrA or uvrB mutants of E. coli, but was present at normal levels in extracts of uvrC mutants. The molecular weight of this activity was estimated at - 14,000. Subsequent studies by Seeberg and his colleagues yielded different results. These investigators prepared lysates of E. coli using lysozyme lysis of sucrose-permeabilized cells (8.5).In the absence of added ATP, extracts of wild-type and of uvrA, -B, and -C mutants showed similar nicking of either unirradiated or UV-irradiated supercoiled ColE 1 DNA. However, in the presence of 1.5 mM ATP, a fivefold increase in activity that nicked UV-irradiated DNA was observed only with extracts of wild-type strains. The addition of an extract of any one of the three uvr mutants to that prepared from either of the other two resulted in the restoration of ATPdependent nicking of UV-irradiated DNA. This observation provided the 82. 83. 84. 85.

R. B. Setlow and W. L. Carrier, PNAS 51, 226 (1964). R. P. Boyce and P. Howard-Flanders, PNAS 51, 293 (1964). A. Braun and L. Grossman, PNAS 71, 1838 (1974). E. Seeberg, J. Nissen-Meyer, and P. Strike, Nature (London) 263. 524 (1976).

14. ENZYMES THAT INCISE DAMAGED DNA

275

basis for a complementation assay that has been extremely useful for the purification of the uvrA, -B, and -C gene products (86,87). The uvrB and -C gene products cochromatograph on DEAE-cellulose. The uvrA gene product does not bind to DEAE-cellulose but does bind to phosphocellulose, eluting at about 250 m M KCI. Neither the isolated uvrA gene product, nor the combined uvrB and -C gene products alone show appreciable U V endonuclease activity under standard assay conditions. However, when these fractions are combined, an ATP-dependent UV DNA-incising activity is demonstrated. This activity recognizes pyrimidine dimer sites in DNA, since substrate sites are lost when UVirradiated duplex DNA is pretreated with photoreactivating enzyme. Gel filtration measurements yielded a MW 100,000 for the uvrA gene product and a value of -70,000 for the uvrB gene product (86). The uvrA, -B, and -C genes have been cloned into bacterial plasmid vectors in a number of laboratories (87-91). Sancar et al. (88-90) developed a so-called “maxicell” procedure for the specific radiolabeling of proteins coded by infecting plasmids. Applying this procedure to cells infected with plasmids carrying either the uvrA or uvrB genes, they demonstrated monomeric gene products of MW = 114,000 and 84,000, respectively, by pol yacrylamide gel electrophoresis. Van Sluis et al. (personal communication) measured a MW = 27,000 for the product of the cloned uvrC gene expressed in minicells. The uvrA protein has been purified to apparent homogeneity and has been reported to bind to unirradiated single-stranded DNA, UV-irradiated single-stranded DNA, and UV-irradiated duplex DNA (87, 92, W).It is likely that in the latter substrate the protein binds to locally denatured regions associated with pyrimidine dimers. DNA binding activity requires the presence of Mg2+and is stimulated by either ATP or GTP. In contrast, the ATP dependence for UV DNA-incising activity by the uvrABC gene product complex cannot be replaced by GTP, ADP, or AMP (87, 92, 94).

-

86. E. Seeberg, PNAS 75, 2569 (1978). 87. E. Seeberg, in “DNA Repair Mechanisms” (P. C. Hanawalt, E. C. Friedberg, and C. F. Fox, eds.), p. 225. Academic Press, New York, 1978. 88. A. Sancar, R . P. Wharton, S. Seltzer, B . M. Kacinski, N . D. Clarke, and W. D. Rupp, J . Ma/. B i d . 148, 45 (1981). 89. A. Sancar, N . D. Clarke, J . Griswotd, W. J . Kennedy, and W. D. Rupp,J. Mol. B i d . 148, 63 (1981). 90. A. Sancar and W. D. Rupp, BBRC 90, 123 (1979). 91. H. Pannekoek, I. A. Noordermeer, C. A. Van Sluis, and P. van de Putte,J. Bncferiol. 133, 884 (1978). 92. E. Seeberg and A-L. Steinum, in “Chromosome Damage and Repair” (E. Seeberg and K. Kleppe, eds.). Plenum, New York, in press.

276

E. FRIEDBERG, T. BONURA, E. RADANY, J . LOVE

Rupp and his colleagues (9-?)reported that the rrvrA protein binds to both unirradiated single- and double-stranded DNA and that UV-irradiation increases the binding affinity of the protein for both substrates, particularly for duplex DNA. These investigators also reported that 10 mM ATP increases binding affinity for unirradiated single-stranded, but not double-stranded, DNA. The MWAprotein also catalyzes the hydrolysis of ATP in the absence of DNA (92, 94). The K , value for this reaction corresponds to that measured for the ATP requirement for endonuclease activity by the complex, suggesting that the ATPase activity may reflect the ATP requirement for nicking. The purified uvrB protein does not have any demonstrable DNA binding or ATPase activity. Very little is known about the properties of the MVVCprotein or of the composition of the IIWABCcomplex in wild-type cells. The irvrABC gene product complex is discussed in Section 111 of this chapter because there is no direct evidence that this complex functions by a DNA glycosylase/AP endonuclease mechanism, as has been demonstrated for the M. luteirs and T4 UV DNA-incising activities. However, the evidence indicating that the complex functions as a direct-acting endonuclease is still limited and indirect. The standard assay for activity on UV-irradiated DNA measures the hydrolysis of phosphodiester bonds without the use of conditions that promote p-elimination of apyrirnidinic sites in the DNA. However, this observation is not inconsistent with a two-step mechanism of DNA incision if the complex contains both PD-DNA glycosylase and AP endonuclease activities. The uvrABC protein complex has also been shown to attack DNA that contains substrate sites other than pyrimidine dimers, e.g., in DNA treated with 8-methoxypsoralen plus 360 nm light, both sites of psoralen monoadduct and of psoralen interstrand crosslink formation are attacked by the enzyme activity (94).This supports the results of in vivo studies that have shown uvrABC mutants of E. coli to be abnormally sensitive to treatment with psoralen plus long wavelength UV radiation (9.5). Such mutants are additionally sensitive to Cnitroquinoline- I-oxide, nitrogen mustard, and mitomycin C (see Ref. 1 ) suggesting that the rrvrABC complex recognizes conformational distortions in the DNA duplex rather than specific forms of base damage to DNA. Such a broad substrate specificity is more consistent with the expression of a direct-acting endonuclease than with a DNA 93. w. D. Rupp, A. Sancar, B. Kacinski, R . Wharton, S. Seltzer, J . Griswold, and N . Clarke, in “Chromosome Damage and Repair” (E. Seeberg and K. Kleppe, eds.), Plenum, New York, in press. 94. E. Seeberg, Prog. Ncic. Acids Res. Mol. Biol., in press. 95. R . S . Cole, D. Leortan, and R. R. Sinden, J M R 103, 39 (1976).

14. ENZYMES THAT INCISE DAMAGED DNA

277

glycosylase activity. However, the molecular mechanism of incision of DNA that contains the various forms of damage previously mentioned need not be the same in each case. Thus it remains to be definitely established that the incision of DNA that contains pyrimidine dimers, for example, does not uniquely occur by a DNA glycosylase/AP endonuclease mechanism.

B.

uv ENDONUCLEASE ACTIVITY IN OTHERORGANISMS

There have been numerous reports in the literature of eukaryote DNAincising activity that preferentially attacks UV-irradiated relative to unirradiated DNA, however few of these activities have been purified or extensively characterized (see Ref. %). Van Lancker and Tomura (97) reported the purification to apparent homogeneity of an enzyme from rat liver with an estimated MW 15,000. Sedimentation velocity analyses in neutral and alkaline sucrose density gradients indicate that the purified enzyme catalyzes the formation of single-strand breaks in duplex UVirradiated or acetylamino fluorene-treated DNA. The enzyme is active without added MgC12, but is optimally stimulated at 5-10 mM MgC12. Waldstein et d.(98) reported a %fold purification of a UV DNA-incising activity from calf thymus that apparently recognizes pyrimidine dimers specifically. Maximum activity of the enzyme is observed between pH 7.5-8.0, and 5-15 mM MgC1, is optimal for activity. No activity is observed in the presence of 10 mM EDTA. The activity is extraordinarily labile, which has precluded significant purification thus far. However, preliminary gel filtration studies suggest that the activity is associated with a protein of high molecular weight.

-

C. ENDONUCLEASE V OF Escherichici coli

Endonuclease V is a protein that has been purified -6000-fold by Gates and Linn (99) on the basis of its ability to cleave single-stranded fd DNA. The enzyme also acts on either Form I or Form I1 duplex DNA at a rate about 10% of that on single-stranded DNA. However, duplex DNA exposed to UV radiation, heating at pH 5.2, partial denaturation and renaturation, or treatment with OsO, is degraded several times faster than is 96. E. C. Friedberg, K . H . Cook, J. Duncan, and K . Mortelman,Photochem. Phorobiol. Rev. 2, 263 ( 1977).

97. J. L. Van Lancker and T. Tomura, BBA 353, 101 (1974). 98. E. A. Waldstein, S. Peller, and R. B . Setlow, PNAS 76, 3746 (1979). 99. F. T. Gates, Ill, and S. Linn, JEC 252, 1647 (1977).

278

E. FRIEDBERG, T. BONURA, E. RADANY, J. LOVE

native DNA. A substrate particularly favored by the activity is duplex PBS2 DNA (which contains uracil instead of thymine). Mg2+is required for activity on fd, PM2, UV-irradiated PM2, or PBS2 DNA. With PBS2 or fd DNA as substrate, the enzyme has a sharp pH optimum at 9.25. When PBS2 DNA labeled with both [32P]-and [3H]uridine was incubated with endonuclease V, the acid-soluble products of degradation were shown to be oligonucleotides. No free uracil or dUMP were detected. The substrate specificity and mechanism of action of endonuclease V are unclear. Gates and Linn (99) suggested that the enzyme might recognize regions of DNA with abnormal secondary structure. Based on this premise these authors argued that at the pH optimum of 9.2-9.5, transient denatured regions may form in duplex DNA by titration of uracil or thymine residues, with consequent disruption of A-U or A-T base pairs, respectively. Surprisingly however, the activity of endonuclease V on heat denatured PBS2 DNA is less than that on the duplex form, and the pH optimum with fd DNA (which is naturally single-stranded) is also about 9.5. AT D. ENDONUCLEASE ACTIVITY DIRECTED PHOTOALKYLATED PURINESIN DNA

Livnehet al. (100) have reported the identification of an enzyme activity that recognizes a specific substrate produced by photoalkylation of DNA. Reaction of Form I PM2 DNA with isopropylalcohol in the presence of a free radical photoinitiator plus UV light at 300 nm, leads to the specific substitution of purine moieties in the DNA, yielding 842hydroxy-2-propy1)adenineand 8-(2-hydroxy-2-propyl)guanine.These investigators identified an activity in extracts of M. luteus that degrades this DNA. The activity is not completed by the simultaneous presence of single-stranded undamaged DNA or single-stranded UV- or X-irradiated DNA. However, the inclusion of duplex UV- or X-irradiated DNA results in inhibition of activity. Divalent cations are not required for the activity, which is fully active in 1 mM EDTA. ATP (0.1 mM) results in about 40% inhibition, and 0.1 mM caffeine about 30% inhibition of activity. No release of free modified purines is detected during incubation of photoalkylated DNA with M. luteus extract. In addition, when substrate is present at saturating concentrations, the addition of a tenfold excess of depurinated duplex DNA does not inhibit enzyme activity. Both of these observations argue against a DNA glycosylase/AP endonuclease mechanism of incision of the DNA. The authors thus conclude that this endonu100. Z. Livneh, D. Elad, and J. Sperling, PNAS 76, 5500 (1979).

14. ENZYMES THAT INCISE DAMAGED DNA

279

clease directly attacks phosphodiester bonds in photoalkylated DNA. In speculating about the nature of the substrate sites recognized by the endonuclease, the authors point out that 8-substitution of purines by a group as small as bromine results in destabilization of the normal anti conformation of the affected nucleotide. Physical studies on such substitutions have shown that a stable configuration requires rotation around the N-glycosylic bond to assume the syn conformation. It is possible that the syn conformation assumed by any 8-substituted purines in DNA provides the molecular basis for the specific substrate recognized by this class of endonucleases. If so, one would assume that the purified enzyme might also attack DNA that contains N-acetoxy-2-acetylaminofluorene adducts, since this compound has also been shown to bind at the %position of guanine in DNA and to promote the conformation change previously discussed (101). ACKNOWLEDGMENTS Studies in the senior author’s laboratory were supported by research grants from the U.S. Public Health Service (USPHS) (CA-12428), the American Cancer Society (NP 174), and The Foundation-March of Dimes (1-6721, as well as by contract EY-76-S-03-0326 with the U S Department of Energy. E. H. Radany is a predoctoral fellow supported by USPHS training grant GM-07364, and J. D. Love is a USPHS postdoctoral fellow supported by grant CA-06441. We thank numerous colleagues for providing preprints of their work cited in this paper.

Note Added in Proof

Very recent investigations in this laboratory have provided genetic and biochemical evidence for the physical association of pyrimidine dimer-DNA glycosylase and AP endonuclease activities on the polypeptide encoded by the d m V gene of phage T4. [S. L. H. McMillan, E. H. Radany, and E. C. Friedberg, Fed. Proc. 40, 1763 (1981); S. McMillan, H. J . Edenberg, E . H. Radany, R. C. Friedberg, and E. C. Friedberg, J. Virol., in press.]

101. D. Grunberger and I. B. Weinstein,in “Biology of Radiation Carcinogenesis” (J. M. Yuhas, R. W. Tennant and J. Regan, eds.), p. 175. Raseen, New York, 1976.

This Page Intentionally Left Blank

Pancreatic DNase STANFORD MOORE

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ill. Chemical Structure . . . . . . . . . . . . . A . Sequence, . . . . . . . . . . . . . . . B . Essentiality of Specific Residues . . . . IV. Catalytic Properties . . . . . . . . . . . . . A . Roles of Divalent Metal Ions . . . . . . B . Substrate Specificity . . . . . . . . . . . V. Actin as an Inhibitor of DNase I . . . . . . VI. Research Applications . . . . . . . . . . . 11. Purification

1.

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

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

. . . . . . . . . .

281 282 285 285 286 288 288 290 293 295

Introduction

Bovine pancreatic deoxyribonuclease is the most thoroughly studied of the enzymes of the DNase I class (EC 3.1.21. l), defined as enzymes that cleave the substrate endonucleolytically to yield primarily S-phosphodiand oligonucleotide end products. In this chapter, the term DNase will refer to the bovine enzyme, unless otherwise specified. Bovine pancreatic DNase I can be resolved into four components of similar catalytic activity, DNases A, B, C, and D. For most purposes the mixture of the four is a suitable catalyst, subject to the degree of freedom from contaminating proteases or ribonucleases : when one of the subfractions has been studied, the letter designation will be added. The information on DNase to 1970 has been summarized by Laskowski 28 1 THE ENZYMES, Vol. XIV Copyright 0 1981 by Academic Press, Inc. All rights of reproduction in any form reserved ISBN 0-12-122714-6

282

STANFORD MOORE

(I) in Volume IV of this series. This chapter will cover primarily the information that has been gathered in the subsequent decade. Comparative studies on pancreatic DNases from different species have shown that the ovine and human enzymes are closely similar to the bovine catalyst, as cited in the following section on purification. The characterization of DNases from other tissues has included the finding that the DNase I secreted by the bovine parotid gland ( 2 , 3 )is very similar but not identical, chemically (4), to the bovine pancreatic enzyme. The results provide one of the several examples of enzymes built to the same basic design via different genes in different tissues of the same species. Moving further afield in terms of biological source, the DNase in germinating barley that Brawerman and Chargaff (5) showed had some enzymatic properties in common with bovine pancreatic DNase, has been reinvestigated by ‘iao (6). By applying current techniques, he has succeeded in purifying and characterizing the enzyme from malt diastase and finds it to be remarkably homologous in all of its chemical and enzymatic properties with the mammalian enzyme. Thus, the detailed picture of the chemical structure and catalytic properties of bovine pancreatic deoxyribonuclease summarized in the following pages applies to a broader spectrum of enzymes than was envisaged at the outset of the research.

II. Purification

The starting material for the chromatographic purification of pancreatic DNase has usually been the amorphous enzyme prepared from bovine pancreas by ammonium sulfate fractionation according to the procedure of Kunitz (7). In an extension of the chromatographic experiments of Price et al. (8) on sulfoethyl-Sephadex, Salnikow et al. (9) observed four active components by chromatography on phosphocellulose (Fig. 1). Pancreatic juice submitted to initial chromatography on DEAE-cellulose gave a simi1. M . Laskowski, Sr., “The Enzymes,” 3rd. Ed., Vol. 4, p. 289, 1971. 2. W. D. Bail and W. J. Rutter, J . Exp. Z o d . 178, I (1971). 3. W. D. Ball, BBA 341, 305 (1974). 4. R. L. Lundblad, S . Hoffman, C. M. Noyes, and H. S. Kingdon,J. Dent. Res. 56, 320 (1977). 5. G . Brawerman and E. Chargaff, JBC 210, 445 (1954). 6. T.-H. Liao, Phytochernistry 16, 1469 (1977). 7. M. Kunitz, J . Gen. Physiof. 33, 349 (1950). 8 . P. A. Price, T.-Y. Liu, W. H. Stein, and S. Moore, JBC 244, 917 (1%9). 9. J. Salnikow, S. Moore, and W. H. Stein, JBC 245, 5685 (1970).

15.

283

PANCREATIC DNase

1

1

’M f”

-

> .-

u

a 0

300

Effluent

600

900 ...

(ml)

FIG. 1. Chromatography of bovine pancreatic DNase (Worthington DP grade) on phosphocellulose. Column, 2 x 75 cm: temperature, 25”; flow rate, 35 ml per hour: 4-mI fractions were collected: column equilibrated with 0.25 M sodium acetate at pH 4.7: initial eluent, 150 ml of 0.38 M sodium acetate buffer, pH 4.7; linear gradient with 400 ml each of the initial and the limit buffer, 0.7 M sodium acetate, pH 4.7: (---) absorbance at 280 nm: (0-0) enzymatic activity. From Salnikow ei nl. (Y), reproduced with permission. lar pattern of DNases on phosphocellulose. The major enzyme, DNase A, was the protein taken for detailed sequence analysis; it is a glycoprotein with a neutral carbohydrate side chain. The protein moiety of DNase B is indistinguishable from that of DNase A; the carbohydrate side chain contains sialic acid. From amino acid analyses, DNase C was characterized as being the same as A except for a proline residue substituted for a histidine. This conclusion was confirmed by peptide maps (10) interpreted in the light of the sequence studies. Liao ( 1 1 ) has shown that DNase D has the same sequence as C but contains sialic acid in the carbohydrate portion. In order to obtain the small amount of DNase D in a stable form ( / I ) , the diisopropyl fluorophosphate-treated protein fraction was rechromatographed on DEAE-cellulose with use of a CaC1, gradient, thus combining the experience of Hugli (f2)on the chromatographic removal of traces of chymotrypsin and chymotrypsinogen B and of Price et nl. (8) on the stabilization of DNase against proteolysis by Ca2+. Liao ( I / ) has summarized the differences in the four enzymes (Table I). In Fig. 1 it can be noted that the activity-to-protein ratio may be slightly lower for the sialylated DNases B and D than for A and C; this observation is borne out by the specific activities in Table I. When DNase I is used as a specific biochemical reagent in the preparation of nuclear ribonucleic acids or nuclear proteins, the purification prob10. J. Salnikow and D. Murphy, JBC 248, 1499 (1973). 11. T.-H. Liao, JBC 249, 2354 (1974). 12. T. E. Hugli, JBC 248, 1712 (1973).

284

STANFORD MOORE TABLE 1

ANALYSESOF DNASESA, B ,

c, A N D D“

Analysis for

DNase A”

DNase B”

DNase C”

DNase Dh

Mannose Galactose N - Acet ylglucosamine Sialic acid Proline Histidine Specific activity (unitsimg)

5.8 0.0 I .9 0.0 9.0 6.2 1158

4.5 I .O 3.4 1 .O 9.2 5.7 92 1

4.7 0.0 1.9 0.0 10.1 4.9 I045

4.3 I .O 3.3 0.8 9.9 5.1 837

“ From Liao ( / I ) ,

reproduced with permission.

’ Constituents expressed as residues per molecule

lem is a special one. The separation of DNases A , B, C, and D from one another is not necessary, but freedom from traces of ribonucleases or proteases is essential. Wang and Moore (13) have described a procedure that effects complete removal of trypsin, chymotrypsin, and chymotrypsinogen by a combination of affinity chromatography and salting-out adsorption on lima bean protease inhibitor coupled to Sepharose, an extension of the method of Otsuka and Price (14). DNase is extremely sensitive to inactivation by proteases in the absence of Ca2+; the protease-free preparation retains full stability in the absence of Ca’+ for more than 10 days at pH 8 and 37”. Removal of the last traces of RNase has been accomplished (7) by affinitychromatography on a long (72 cm) column of 5-(4-aminophenylphosphoryl)uridine2’(3’)-phosphate-Sepharose (15, 16). The fully active product, obtained in quantitative yield, has less than I part of RNase per 10 million parts of DNase. Wadano et (11. (171, in the course of isolating DNases from ovine pancreas, have found chromatography on conconavalin A-agarose to be an effectivestep in the purification of DNases with neutral carbohydrate side chains: their method of isolation of DNase, which includes the use of phenylmethanesulfonyl chloride (18) as a protease inhibitor and chromatography on CM-cellulose with Ca2+-containingbuffers, offers the possibility of preparing bovine DNases A and C in higher yield than that obtained in the initial steps of the Kunitz (7) procedure. They also found 13. 14. 15. 16. 17. 18.

D. Wang and S. Moore, JBC 253, 7216 (1978). A. S. Otsuka and P. A. Price, A I I ~Biochem. . 62, 180 (1974). M . Wilchek and M . Gorecki, “Methods in Enzymology,” Vol. 34, p. 492, 1974. 0 . Brison and P. Chambon, Anti/. Biockern. 75, 402 (1976). A. Wadano, P. A. Hobus, and T.-H. Liao, Biocliemisrry 18, 4124 (1979). D. E. Fahrney and A . M. Gold, JACS 85, 997 (1963).

15. PANCREATIC DNase

285

that adsorption on Con A-agarose provided a means of obtaining a protease-free preparation of bovine pancreatic DNase. Funakoshi et a/. (19) have isolated from human duodenaljuice a DNase I that has properties very similar to those of the bovine pancreatic enzyme. Love and Hewitt (2U) have purified the human pancreatic enzyme with similar results. In their experiments they introduced a fluorometric DNase assay based upon the use of a circular DNA substrate and the binding of the denatured split products to ethidium bromide. In the other papers covered in this chapter, authors define their own modifications of the assays ( 1) based upon hyperchromicity, proton release, or acid-soluble nucleotides determined by absorbance or radioactivity.

111.

Chemical Structure

A.

SEQUENCE

Through study of the peptides yielded by tryptic or chymotryptic hydrolysis and by cyanogen bromide cleavage, Salnikowet al. (21)and Liaoet LII. (22) derived a sequence for the amino acid residues in reduced and carboxymethylated DNase A. The disulfide bridges were characterized by drawing upon the observation of Price et ril. (23) that in the presence of Ca2+one of the two S-S bonds could be selectively reduced by mercaptoethanol ; cyanogen bromide cleavage of the alkylated derivative provided data on the pairings. The result of the sequence study is given in Fig. 2. The sequence is a working hypothesis based upon all of the data available at this time. The molecule corresponds to a protein of 257 residues with carbohydrate attached through an aspartamido-hexose linkage at one position (Asn-18 in an Asn-X-Thr sequence). The amide -NH3 value is 21, which agrees with the determination by Lindberg (24) recalculated for the present molecular size. The molecular weight of DNase A, calculated from the amino acid and carbohydrate composition, is 30,072. The numbers of individual residues are Lys-9, His-6, Arg-11, Asp-20, Asn-12, Thr-15, Ser30, Glu-10, Gln-9, Pro-9, Gly-9, Ala-22, +Cys-4, Val-24, Met-4, Ile-11, Leu19. A. Funakoshi, Y. Tsubota, H. Wakasugi, H. Ibayashi, and Y. Takagi, J . Biochem. (Tokyo) 82, 1771 (1977). 20. J. D. Love and R. R. Hewitt, JBC 254, 12588 (1979). 21. J. Salnikow, T.-H. Liao, S . Moore, and W. H. Stein,JBC 248, 1480 (1973). 22. T.-H. Liao, J. Salnikow, S. Moore, and W. H. Stein, JBC 248, 1489 (1973). 23. P. A. Price, W. H. Stein, and S . Moore, JBC 244, 929 (1969). 24. U . Lindberg, Biochemistry 6, 335 (1967).

286

STANFORD MOORE

Leu-Lys-Ile-Ala-Ah-Phe-Asn-I

10 Corb. le-Arg-Thr-Phe-Gly- Glu-Thr-Lys-Met-Ser-Asn-

20

30

Alo -Thr-Leu- Alo-Ser-Tyr -1le-Vol- Arg-Arg -Tyr-Asp-Ile-Val-Leu-Ile-Glu-Gln-Val40

Arg-Asp-Ser-His-Leu-Val-

50 Ala- Val-Gly-Lys-Leu-Leu-Asp-Tyr

- Leu-Asn-Gln- Asp-Asp70

60

Pro - As n -T hr - Ty r - H i s -Ty r - Vo I- Vo I- Ser - GI u - Pro - Leu- GIy - Arg - Asn- Ser -Ty r - Lys -G Iu 80

YO

- Gln-Tyr -

Arg-Tyr-Leu-Phe-Leu-Phe-Arg-Pro-Asn-Lys-Val-Ser-Val-Leu-Asp-Thr-Tyr I00

110

Asp-Asp-Gly-Cys- Glu- Ser-Cys-Gly -Am-Asp-Ser-Phe-Ser- Arg-Glu-Pro- Alo-vai-vai -

u

I30

Lys-Phe-Ser-Ser-His-Ser-Thr-Lys-Val-Lys-Glu-Phe-Ala-Ile-

Val-Alo-Leu-His-Ser-

140

Alo-Pro-Ser-Asp-Ala-Val-

I50

Ala-Glu-lle-Asn-Ser-Leu-Tyr

-Asp-Val-Tyr -Leu-Asp-Val

160

-

I70

Gln-Gln- Lys -Trp-His-Leu-Asn-Asp-Vol-Met-Leu-Met-Gly-Asp-Phe-Asn-Alo-Asp-CysI80

Ser-Tyr- Vol-Thr-Ser-Ser-Gln-Trp-

Ser- Ser- Ile- Arq-Leu- Arg-Thr- Ser-Ser -Thr-Phe200

190

Gln-Trp-Leu- I l e - Pro-Asp-Ser-Alo-Asp-Thr-Thr-Alo-Thr-Ser-Thr-Asn-Cys-Aia 210

-Tyr-

220

Asp-Arg-Ile- Val-Val- Alo-Gly-Ser-Leu-Leu-Gln-Ser-Ser-Vol230 A lo-Pro-Phe-Asp-Phe-GIn250

Ser-Asp-His-Tyr-Pro-Val-

Vol- Gly-Pro-Ser- Ala-

240

Ala-Ah-Tyr-Gly- Leu-Ser- Asn-Glu-Mel-Ala-Leu-Alo-Ile

-

2 57

Glu-Vol-Thr-Leu-Thr

FIG. 2. Sequence of bovine pancreatic deoxyribonuclease A. From Sainikow et rrf. (21) and Liao et a / . ( 2 )reproduced . with permission. 23, Tyr-15, Phe-11, Trp-3, Man-5.8, and GlcNAc-1.9. The carbohydrate side chain (25) is probably a mixture of oligosaccharides of slightly different chain lengths. The pl is about 5 (7).

B. ESSENTIALITY OF SPECIFIC RESIDUES

The enzyme is inactivated by iodoacetate at pH 7.2 in the presence of Mn2+ or Cuz+ with the formation of one residue of 3carboxymethylhistidine (26). The sequence of the tryptic peptide that con25. B. J. Catley, s. Moore, and W. H. Stein, JBC 244, 933 (1969). 26. P. A. Price, S. Moore, and W. H. Stein, JBC 244, 924 (1969).

15. PANCREATIC DNase

287

tained the modified amino acid permits the assignment of the crucial histidine residue to position 13 1 (22). The conclusion from studies with small substrates (I) that there may be a positive charge near the active center histidine residue fits with the fact that this imidazole ring is alkylated by iodoacetate but not by iodoacetamide (16). The histidine residue in DNase A that is substituted by proline in DNase C is His- 118 (10): this residue is thus not essential to the activity of the enzyme. Loss of activity by nitration has been found to parallel the formation of a single 3-nitrotyrosine residue per enzyme molecule (27). By reference to Fig. 1, the composition and sequence of the Tyr-(3-N02)-containing peptide isolated from an enzymatic digest permits assignment of the tyrosine residue to position 62 (2). Ca2+ was not able to stabilize the nitrated enzyme toward chymotryptic digestion, thermal denaturation, or mercaptoethanol reduction of the essential disulfide linkage. Tyr-62 may contribute to the formation of a Cazf binding site on the molecule that is not coincident with the region of His- 131, since the latter remains sensitive to specific carboxymethylation in the nitrated enzyme (27). The disulfide bond that can be reduced without loss of activity (13)is the one involved in the small loop between residues 98 and 101 (22). The disulfide bond between residues 170 and 206 is essential for maintenance of the active molecule. By controlled proteolysis with chymotrypsin, Hugli (28) was able to split the bond between Trp and Ser at positions 178 and 179; the product retained nearly the full activity of the native enzyme. When the cleaved molecule was further digested with carboxypeptidase-Y, the residues Thr-Ser-Ser-Gln-Trp (residues 1 7 4 178) were removed and the molecule still retained 80% of its activity. Thus, five residues h a t e d in the central portion of the peptide chain are functionally expendable. The COOH-terminal residues of DNase, which are normally unavailable to carboxypeptidases, become susceptible to removal when the enzyme is denatured in 0.005% sodium dodecyl sulfate (29). Study of the effect of carboxypeptidase action upon the enzyme required development of a procedure for restoring activity to DNase that has been denatured by the detergent. The inactivation of DNase could be completely reversed by diluting the enzyme solution tenfold into 6 M guanidinium chloride before a 100-fold dilution for assay. A loss of regenerable activity could be correlated with the removal of 1 or 2 amino acid residues (-Leu-Thr) from the 27. T. E. Hugli and W. H. Stein,JBC 246, 7191 (1971). 28. T. E. Hugli, JBC 248, 1712 (1973). 29. T.-H. Liao,JBC 250, 3831 (1975).

288

STANFORD MOORE

COOH-terminal sequence. DNase thus resembles RNase (30,31) in being one of the several enzymes in which the residues at the COOH terminus have a determining effect upon the folding of the chain into the active conformation. By following the kinetics of the reaction of N-bromosuccinimide with DNase by amino acid analysis for tryptophan, Sartin er ul. (32)have been able to show that modification at Trp-155 is the change most crucial to inactivation by that reagent. Methanesulfonyl chloride at pH 5 inactivated DNase by modification of the hydroxyl group of an as yet unidentified serine residue (3.3). DNase is inactivated by 2-nitro-5-thiocyanobenzoicacid (34)by a reaction that has been shown to involve cleavage of the peptide chain at the hydroxyamino acids at positions 14, 40, 72, and 135 (35). In more general derivatization experiments, guanidination of the nine E-NH, groups or picolinimidylation of the a- and €-amino groups yields active derivatives (36). The NH, groups are thus not essential per se, but when positive charges on the enzyme are removed by carbamylation with cyanate, activity is progressively lost. Modification of the carboxyl groups by condensation with glycine ethyl ester in the presence of a carbodiimide causes major inactivation in the absence of Ca2+(-?3).The presence of the bivalent cation slows the rate of the inactivation.

IV.

Catalytic Properties

A. ROLESOF DIVALENT METALIONS The conformation and the activity of the molecule are markedly dependent upon the presence of metal ions. The resistance to proteolysis conveyed by Ca2+ (8) has been important in the purification of the protein. The role of Caz+ in the refolding of the reduced enzyme to the active conformation is crucial (23). The apparent molecular weight of DNase by gel filtration in Tris buffer increases with pH in the range from pH 7.5 to 9 30. 31. 32. 33. 34. 35. 36.

C. B. Anfinsen, J5C 221, 405 (1956). M . C. Lin,JBC 245. 6726 (1970). J . L. Sartin, T. E. Hugh, and T.-H. Liao, JBC 255, 8633 (1980). T. L. Poulos and P. A. Price, JBC 249, 1453 (1974). T.-H. Liao and L. J . McKenzie, JBC 254, 9598 (1979). T.-H. Liao and A. Wadano, JBC 254, 9602 (1979). B . V. Plapp, S. Moore, and W. H. Stein, JEC 246, 939 (1971).

15.

289

PANCREATIC DNase

TABLE I1

CA" REQUIREMENTFOR DNASEACTIVITYAT P H 8 '* H yperchromicity assay

Metal ions in assay

midmg)

Maxi mu m specific activity (9%)

2.5 mM MgCI, 2.5 mM MgCI2, Dowex-purified 2.5 mM MgCI, + EGTA lo-' M CaC12b 2.5 mM MgCI, + IO-'M CaC12b

10-18 3-6 0.1 8 710'

1.4-2.5 0.4-0.9 0 1.1 100

Specific activity (&so

I

pH-stat assay Specific activity (NaOHI min/DNase) 10

3 0 5 650'

Maximum specific activity (%)

1.5 0.5

0 0.8 100

" From Price (42), reproduced with permission. Each value represents the average or range of values for three or more assays of metal ion-free DNase A. Hyperchromicity assays were in 5 mM ionic strength Tris-chloride buffer at pH 8 and 25". pH-stat assays were in buffer-free solution at 25". DNA concentration is 0.04 mg/ml. ' Addition of lo-" M EGTA has no effect on these activities.

(37,38), a hydrodynamic change that is reversed by the presence of Caz+.

By gel filtration of DNase A at pH 7.5 with 45Ca2+,Price (39) found 2 Ca2+ bound with an average Kd of 1.4 x and 3 bound with a K d of 2 x With Mg2+the Kd for two sites was 2.3 x One of the two strong Ca2+ binding sites is not subject to competition from Mg2+.Fifty percent of the maximum transition in the CD spectrum occurs near M CaCl, and half-maximum protection against action by trypsin is achieved near 1.3 x M CaCl, (40).The uv spectra show a conformational change indicative of increased interiorization of tryptophan and tyrosine residues in the presence of Ca'+ (41). Price (42) has made the key observation (Table 11) that with reagents that have been purified to reduce the Ca2+contamination to a minimum, DNase in the presence of Mg2+is about 99.5% inactive. In the presence of B . Librraga, C . Bustamante, A. Gil, and E. Melgar, BBA 579, 298 (1979). B . Lizarraga, D. Sanchez-Romero, A. Gil, and E. Melgar,JBC 253, 3191 (1978) P. A. Price, JBC 247, 2895 (1972). T. L. Poulos and P. A. Price, JBC 247, 2900 (1972). 41. R . Tbllis and P. A. Price, JEC 249, 5033 ( 1974). 42. P. A. Price, JBC 250, 1981 (1975).

37. 38. 39. 40.

290

STANFORD MOORE

a very low concentration (0.01 mM) of EGTA, a chelating agent that binds Ca*+about lo6 times more strongly than it binds Mg2+,DNase activity in the presence of 2.5 mM Mg2+becomes undetectable. A Ca2+concentration of 0. I mM yields maximum activity; concentrations as high as 1 mM are inhibitory. Earlier experiments on the low activity of DNase when only MgCl, is added are attributed to the effect of traces of Ca2+in the reagents. Bivalent metal ions serve two essential roles; Ca2+ must be bound to the enzyme and ions such as Mg2+to the substrate. The concento 3 x tration of Ca2+in bovine pancreatic juice is 4 x M ( 4 3 ,a range that can contribute to keeping DNase functional in its physiological environment. The effects of various bivalent metal ions are summarized in Table I11 (42). Sr2+and Ba2+can substitute for Ca2+,but are leass effective. Mn2+ and Co2+ can substitute for Mg2+, the latter with only about 10% efficiency. The results permit earlier studies on metal ion effects [see, e.g., Refs. (44,431 to be interpreted in more operational terms. The degree to which the enzyme functions optimally depends upon two factors, the Ca2+concentration and the Mg2+ or Mn2+ concentration, with the former being subject to variation from trace amounts of Ca2+as a contaminant in the latter. Double-strand scission [see ( I ) ] can be expected when the affinity of the enzyme for substrate is at a maximum in the presence of both CaLf and Mg2+;single-strand scission and changes in specificity are likely to be associated with suboptimal concentrations of Ca2+ present as contaminants when Ca2+is not deliberately added (42). Douvas and Price (46) have shown that 1 Mg2+ per 2 DNA-phosphorus is optimum. Na+-DNA is inhibitory and maximum rates are obtained with Mg2+-DNAas substrate rather than with Na+-DNA plus MgC1,. Preincubation of DNase with Ca2+ before addition to the substrate increases the initial rate of hydrolysis twofold over that obtained with Ca2+-free DNase with double-stranded calf thymus DNA as substrate. B. SUBSTRATE SPECIFICITY

Junowicz and Spencer (45) have conducted an extensive enzymatic and chromatographic study of the terminal purines and pyrimidines in the oligonucleotides liberated by the digestion of calf thymus DNase A with a variety of bivalent ion mixtures for different times; the experiments were 43. 44. 45. 46.

A. Frouin and P. Gerard, C.R. SOC.Bid. 72, 98 (1912). E. Junowicz and J. H. Spencer, BBA 312, 72 (1973). E. Junowicz and J . H. Spencer, BBA 312, 85 (1973). A. Douvas and P. A. Price, BBA 395, 201 (1975).

IS.

29 1

PANCREATIC DNase TABLE 111

ACTIVITYO F DNASEWITH DIFFERENT METAL IoNs"*~ Activities under following conditions of assay*

Metal ion

2.5 mM MgCI,, lo-' M metal ion

2.5 m M Metal ion, M Caz+

9 0 700 440 220 50 30

700 700 8 0.9 4.2 750 81 0 0

MgCI, MgCI, + 10-5M EGTA CaCI, SrCI, BaCI, MnCI,' COCI," CdCI,, SnCI,, FeCI,, NiCI,, CuCI, ZnCI,, EuCI,, SmCI,, NdCI,

II 0

From Price (42), reproduced with permission. Hyperchromicity assays in 5 rnM Tris, pH 8, 25". with 0.04 mdml of DNA. Activity values are AA,,, per midmg of DNase. The possibility exists that traces of CaZ+in MnCI, and CoCI, account for these values.

conducted before the effects of traces of Ca2+in most bivalent metal salts (42) were fully appreciated. Under near optimal conditions (e.g., with Mg2+plus C a 2 + ) , the enzyme gives the molar yields of the four nucleosides in the end positions listed in Table IV; in these experiments the enzyme is impressive for its versatility rather than its selectivity. The results did not vary greatly with the time of hydrolysis. However, with several bivalent TABLE 1V

DEOXYRIBONUCLEOTIDES AT T H E 5' AND 3' ENDSOF T H E DNAovb OLIGONUCLEOTIDES RELEASEDFROM C A L F THYMUS Mole% nucleosides ~

~~

5' end

3' end

Digestion time (rnin)

dT

dC

dG

dA

dT

dC

dG

dA

1.5 30.0

29.0 34.8

18.2 22.2

23.8 17.4

29.0 25.5

29.0 27.7

10.5 8.5

28.4 30.1

32.1 33.7

" From Junowicz and Spencer (4.5), reproduced with permission. Activation by 33 m M Mg2+ and 0.1 mM Ca'+.

292

STANFORD MOORE

metal additions that gave less than maximum specific activity, the molar ratios of the terminal nucleosides varied markedly and the proportions changed with the time of digestion. At the 5’ end, the ratio of dT or dC to dG or dA varied from 1 : 50 to 1 : 1; at the 3’ end, the ratio of dA or dC to dG or dT varied from 1 : 5-25 to 1 : 1. In these digests and with the singlestranded DNA from E. coli K12, consistent yields were obtained of long oligonucleotides lacking dA at the 3’ end. Simon et al. (47) have extended to DNase their use of crab d(A-T) polymer as a substrate for nucleases. This unique polymer, which is composed predominently of alternating A and T but contains about 3% of G and C residues integrated into its structure, was submitted to controlled hydrolysis by DNase in the presence of varying concentrations of MgCI, to obtain a hexanucleotide fraction enriched in C and G. Pruch and Laskowski (48) have subsequently undertaken to determine whether the I-3% of ribonucleotides, still detectable in preparations of crab d(A-T), that have been exhaustively treated with RNases, are built-in components. Digestion with DNase A to dinucleotides led to chromatographic evidence for the mixed dinucleotides dC-rG, dT-rA, and dT-rG. The authors conclude that crab d(A-T) polymer from C . borealis contains covalently bound ribonucleotides and that, as a corollary premise, sugar specificity of DNase may be limited to the nucleotide following the point of cleavage. The result will stimulate experimentation; the finding fits the function of DNase as an enzyme with which dinucleotides are key end products. Pancreatic DNase has been an enzyme for which no small synthetic substrate is hydrolyzed rapidly. p-Nitrophenyl esters have proved to be convenient synthetic substrates in spectrophotometric assays for use in studies on the kinetics and mechanisms of action of a number of phosphodiesterases. With DNase I, Liao (49) has examined deoxythymidine, 3’, 5’-di-p-nitrophenyl phosphate, a substrate that Razzell and Khorana (50) studied with snake venom phosphodiesterase, and that Cuatrecasas et cil. (51) found to be highly susceptible to the action of staphylococcal nuclease. The compound is rapidly hydrolyzed at a single bond by DNase at pH 7 . 2 with the liberation of p-nitrophenol measurable at 400 nm. The binding is not strong: at 10 mM substrate, the enzyme is not saturated, but the initial rate of hydrolysis (in the first 1 to 5 min) varies linearly with 47. M . Simon, H.-C. Chang, and M . Laskowski, Sr. BBA 232, 462 (1971). 48. J. M. Pruch and M. Laskowski, Sr. JBC 255, 9409 (1980). 49. T.-H. Liao, JBC 250, 3721 (1975). 50. W. E. Razzell and H. G . Khorana, JBC 234, 2105 (1959). 51. P. Cuatrecasas, ‘M. Wilchek, and C. B . Anfinsen, Biochemistry 8, 2277 (1969).

15.

PANCREATIC DNase

293

enzyme concentration in the range from 1 to 6 pg of DNase in 110 pl. Bivalent metals are essential; the maximum facilitation was obtained with 10 mM MnCI, and 1 mM CaCl,. The enzymatic activities toward DNA and N02Ph-pdTp-NOzPhwere lost in parallel upon carboxymethylation of His-131 (26). Since the photometric response is not specific for DNase, the assay has use only with the purified enzyme. Examination of the products of hydrolysis by paper electrophoresis gave the unexpected result that DNase liberates p-nitrophenol from the 3’-ester group. Snake venom phosphodiesterase, which like DNase yields 5’-nucleotides from DNA, liberates p-nitrophenol from the 5’-ester group of this substrate (SO).Staphylococcal nuclease (51 ) gives predominantly p-nitrophenyl phosphate from the 5’ position. Thus, these three diesterases hydrolyze N0,Ph-pdTp-N0,Ph in three different ways. DNase A has been cross-linked to RNase by Wang (52) to prepare a bifunctional enzyme. The coupling was via the thiolation of each protein with N-acetyl-DL-homocysteinethiolactone to yield a disulfide bridge between the two enzymes. The product, which contained one molecule each of DNase and RNase, hydrolyzed thymus DNA and yeast RNA with 75 and 40%, respectively, of the efficiencies of the parent catalysts. The RNA strand of the hybrid substrate phage f l DNA * [3H]RNA was hydrolyzed rapidly by the Mn2+-activatedhybrid enzyme; the RNA strand in the complementary combination was not hydrolyzed significantly by RNase alone. Although an -S-Slinkage may not be the most practical linkage, the concept of the conjugation offers the possibility of delivering in vivo two enzymes that differ in size, charge, and biological function to the same site at the same time.

V.

Actin as a n Inhibitor of DNase I

The presence in most mammalian cells of a protein that will inhibit pancreatic DNase has been known for over thirty years (I). Lazarides and Lindberg (S3)found in 1974 that the protein is cytoplasmic actin. Purified actin from chicken skeletal muscle inhibits DNase; the DNase inhibitor isolated from various tissues and cells is found to be closely similar to actin in its physical and chemical properties. Antibodies to the purified DNase inhibitor show reactivity toward actin-containing fibers in human skin fibroblasts. The inhibitory protein is thus a major cellular component that usually constitutes 5-10% of the soluble protein. The attribution of a 52. D. Wang, Biochemistry 18, 4449 (1979). 53. E. Lazarides and U . Lindberg, PNAS 71, 4742 (1974).

294

STANFORD MOORE

physiological role to the DNase-inhibitory action of actin is limited by the paucity of information on the existence of DNase I-type enzymes in nonpancreatic cells. The situation is very different from that which pertains to the highly specific cytoplasmic RNase inhibitor (54), K i= 3 X lo-’’’ M, which is present in trace amounts (about 0.01-0.02% of the soluble protein) along with a one-fifth to one-tenth molar quantity of a pancreatictype RNase, the activity of which is thus modulated. In line with Laskowski’s (I) suggestion, inhibition by actin can be a key criterion for characterization of intracellular nucleases of the DNase I type. Actin can be removed from cellular extracts by adsorption on a column of DNase coupled to agarose (55); so far, it has not been possible to elute the actin from the affinity column in a way that will preserve the inhibitory activity. Elution with 3 M guanidinium chloride, 1 M in sodium acetate (pH 6 3 , and 30% in glycerol yields the actin-like protein, but with more than 90% of the inhibitory activity lost. A pH 2.8 buffer did not release the actin from the adsorbent. Hitchcock et al. (56) and Mannherz et al. (57) have shown that DNase I causes depolymerization of filamentous actin to form a stable complex of 1 mole of DNase I with 1 mole of globular actin (MW -49,000). Wang and Goldberg (58) have utilized the affinity of DNase for actin-containing fibers to visualize microfilament bundles in nonmuscle cells: DNase I was added to formaldehyde-fixed and acetone-extracted chick or human fibroblasts followed by antibody to DNase for indirect immunofluorescence microscopy, or rhodamine-conjugated DNase was used for direct fluorescent microscopy. A selective assay for monomeric and filamentous actin in cell extracts has been developed by Blikstad et al. (59); the inhibition of DNase activity by G-actin is measured a few seconds after the addition of the enzyme and also after depolymerization of F-actin by 0.75 M guanidinium chloride, 0.5 M in sodium acetate, 0.5 mM in CaCl,, 0.5 mM in ATP, and 10 mM in Tris-HC1 (pH 7.5). Under this condition for 5 min at O”, the F-actin is converted to G-actin without loss of the DNase-inhibitory activity in the monomer. With fluorescently labeled DNase, Mannherz et al. (60) estimated the 54. P. Blackburn and S. Moore, “The Enzymes,” 3rd Ed., Vol. 15, in press. 55. U. Lindberg and S. Eriksson, EJB 18, 474 (1971). 56. S. E. Hitchock, L. Carlsson, and U . Lindberg, Cell 7, 531 (1976). 57. H. G. Mannherz, J. Barrington Leigh, R. Leberman, and H. Pfrang,FEBS (Fed.Eur. Biochem. S o c . ) Lett. 60, 34 (1975). 58. E. Wang and A. R. Goldberg, J . Histochem. Cytochem. 26,745 (1978). 59. I. Blikstad, F. Markey, L. Carlsson, T. Persson, and U. Lindberg, Cell 15,935 (1978). 60. H. G. Mannherz, R . S. Goody, M. Konrad, and E. Nowak, EJB 104, 367 (1980).

15. PANCREATIC DNase

295

binding constant for DNase I and monomeric rabbit skeletal muscle actin to be 5 x 10*M-' and the inhibition to be competitive. With fluorescently labeled actin, Ikkai et al. (61) obtained a K b value of 1 x lo6 M-I. A binding constant of 1.2 x 104 M-' was obtained with filamentous actin (60). Crystals of actin . DNase I complexes have been obtained for cystallographic studies (62, 6 3 ) . When Rohr and Mannherz (64)examined rat pancreatic juice by SDS gel electrophoresis for bands coinciding with actin and DNase actin complex, both were observed. When the juice was treated with 0.25 N H2S04, as in the first steps of the Kunitz (7) procedure for the isolation of pancreatic nucleases, the DNase activity doubled; the inhibitory action of actin is destroyed in this step but DNase is stable. The complex could also be dissociated under physiological conditions by rat or human bile; an activating component has been found to be 5'-nucleotidase (65). The enzyme from snake venom can produce the same results; the process is slow (14 hours at 23") with 10 mM DNase * actin complex and 2 mM nucleotidase. Since 5'-nucleotidase is a constituent of plasma membranes, they were tested (66) and found to effect a slow liberation of DNase from the complex; Grazi and Magri (67) suggest that phosphorylation of actin may have a role in the process. It has been proposed (60) that the interaction of DNase and actin may be a physiological process in the extracellular space of the gastrointestinal tract. VI.

Research Applications

DNase I has received wide use as a probe in the study of the structure of chromatin. Active genes are expected to be more readily accessible to digestion by DNase than the transcriptionally inert segments. Experiments in several laboratories (68-71 ) have demonstrated that limited 61. T. Ikkai, K . Mihashi, andT. Kouyarna,FEBS (Fed. Eur.Eiochem. S o c . ) L e t t . 109,216 (1980). 62. H. G. Mannherz, W. Kabsch, and R . Leberman, FEBS (Fed. E w . Biochem. S o c . ) Lett. 73, 141 (1977). 63. H . Sugino, N. Sakabe, K. Sakabe, S. Hatano, F. Oosawa, T. Mikawa, and S. Ebashi, J . Biochem. (Tokyo) 86, 257 (1979). 64. G. Rohr and H. G. Mannherz, EJB 89, 151 (1978). 65. H. G. Mannherz and G . Rohr, FEBS (Fed. E r r . Eiochem. Soc.) Lett. 95, 284 (1978). 66. G . Rohr and H . G . Mannherz, FEBS (Fed. Eur. Eiochem. Soc.) Lett. 99, 351 (1979). 67. E. Grazi and E. Magri, FEES (Fed. Eur. Eiochem. S o c . ) Lett. 104, 284 (1979). 68. H. Weintraub and M. Groudine, Stience 193, 848 (1976). 69. A. Garel and R. Axel, PNAS 73, 3966 (1976). 70. B . Sollner-Webb and G. Felsenfeld, Cell 10, 537 (1977). 71. S. Weisbrod and H. Weintraub, PNAS 76, 630 (1979).

296

STANFORD MOORE

digestion by DNase can provide information on the structure of the nucleosome. High resolution electrophoretic separation of solubilized oligonucleotides has provided evidence (72-74) for a repetitive internal structure of the chromatin subunit. The periodicity in the fragment lengths can be correlated with helical and super helical orientation of DNA in the nucleosome core (75). In a study of the role of postsynthetic modification of histones in gene activation, Vidali et NI. (76) have used DNase I digestion to show that DNA is more readily liberated from chromatin that contains an increased quantity of acetylated histones; Kastern et al. (77) have used DNase coupled to Sepharose in the preparation of DNA-dependent RNA polymerase 11; the use of an immobilized enzyme makes it possible to release RNA polymerase from actively transcribing genes without having the product contaminated by DNase. The interaction of DNase and actin has formed the basis of a procedure for visualizing cellular microfilaments by fluorescent microscopy (58) and for a differential assay for monomeric and filamentous actin in cell extracts (59).

Thoroughly RNase-free DNase (13) has proved useful (78, 79) to digest chromatin in the preparation of nuclear RNA by the procedure of Penman (80).The protease-free nature of the same DNase preparation is of potential value in the isolation of nuclear proteins. On the basis that the process of replication in virus-infected cells is preceded by liberation of the viral nucleic acid from its protective protein coating, Trukhachev and Salganik (81) studied the inhibition of viral synthesis by DNase in cell cultures. In a clinical study, favorable results (82) have been reported from the local application of DNase to patients suffering from herpes infections of the eye. M . NOH, Nitcleic Acids Res. 1, 1573 (1974). L. C. Lutter, J M B 117, 53 (1977). L. C. Lutter, Nircleic Acids Res. 6, 41 (1979). A. Prunell, R. D. Kornberg, L. Lutter, A. Klug, M. Levitt, and F. H. C. Crick, Science 204, 855 (1979). 76. G. Vidali, L. C. Boffa. E. M . Bradbury, and V. G . Allfrey, /“AS 75, 2239 (1978). 77. W. H. Kastern, J. D. Eldridge, and K. P. Mullinix, JBC 254, 7368 (1979). 78. I. Tamm and T. Kikuchi, P N A S 76, 5750 (1979). 79. I. Tamm, T. Kikuchi, J. E. Darnell, Jr., and M. Salditt-Georgieff, Biockemisrry 19, 2743 (1980). 80. S. Penman, in “Fundamental Techniques in Virology” ( K. Habel and N. P. Salzman, eds.), p. 35. Academic Press, New York, 1969. 81. A . A. Trukhachev and R. 1. Salganik, Virology 33, 552 (1967). 82. A . A. Colain, R . 1. Salganik, I . E . Mikhailovskaya, and I . M. Gorban,Ann. h i / . 203, 371 (1970). 72. 73. 74. 75.

Section Ill

DNA Modification

This Page Intentionally Left Blank

Bacteriophage T4 Polynucleotide Kinase CHARLES C. RICHARDSON

I . Introduction

,

.

. .

.

. . .

. .

.

.

. .

. .

. .

. .

.

.

. .

. .

11. Isolation and Physical Properties . . . . . . . . . . . . . . . .

A . Assay . . . . . . . . . . . . . . . . . . B . Purification , . . , . . . , . . . . . , . . C . Physical Properties . . . . . . . . . . . . 111. Catalytic Properties . . . . . . . . . . . . . . A . Reactions Catalyzed by Polynucleotide Kinase B . Phosphorylation of 5’-Hydroxyl Termini . . . C. Reversal of the Kinase Reaction . . . . . . D. 3‘-Phosphatase Activity . . . . . . . . . . 1V. Role of Polynucleotide Kinase in Vitw . . , . . V. Research Applications . . . . . . . . . . . .

1.

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

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

.

.

. . .

.

. .

. . .

299 30 1 30 1 301 303 305 305 305 309 310 312 3 13

Introduction

Polynucleotide kinase is an enzyme that catalyzes the transfer of the y-phosphate of a nucleoside 5’-triphosphate to the 5’-hydroxyl terminus of a deoxyribonucleic acid or ribonucleic acid molecule. The reaction products are the nucleoside 5’-diphosphate and a 5’-phosphoryl-terminated polynucleotide. The discovery of polynucleotide kinase was made during an attempt to identify enzymes that catalyze the formation of a 5’-terminal triphosphate 299 THE ENZYMES, Vol. XIV Copyright 0 1981 by Academic Press, Inc. All rights of reproduction in any form reserved ISBN 0-12-122714-6

300

CHARLES C. RICHARDSON

group on either an RNA or a DNA molecule. The impetus for such a search was the hypothesis that the formation of a phosphodiester bond to join polynucleotide strands covalently, as is the case in recombination and repair, would require activation of the 5’ terminus of one of the polynucleotides. A nucleophilic attack of the 3’-hydroxyl group of another polynucleotide on the a-phosphate of the triphosphate could result in phosphodiester bond formation and the release of pyrophosphate, a reaction known to occur in the polymerization of nucleotides by RNA and DNA polymerases. Although no such activities were identified that could catalyze this addition of phosphoryl groups to 5’-phosphoryl-terminated polynucleotides, these studies did lead to the discovery of a different enzyme activity-polynucleotide kinase-that could phosphorylate 5 ’ hydroxyl termini. Subsequent studies showed that covalent joining of polynucleotides involved a different enzyme, DNA ligase, via the activation of the 5’ terminus of the polynucleotide by adenylation to form a pyrophosphate bond. The rationale to search for enzymes that catalyze the synthesis of 5’-triphosphate termini was not unreasonable: just such an enzyme has been isolated from vaccinia virus cores (I). Polynucleotide kinase was first identified in Escherichici coli cells infected with bacteriophage T2 or T4 (2, 3 ) . Although no polynucleotide kinase activity has been found in uninfected bacteria, a similar activity has been identified (4, 5 ) and partially purified (6, 7) from rat liver nuclei. Of these three polynucleotide kinases, the bacteriophage T2- and T4induced enzymes have been most extensively purified. While the T2 and T4 enzymes are very similar, the T4 polynucleotide kinase has been more extensively characterized. Therefore, this chapter focuses on the bacteriophage T4 polynucleotide kinase, with reference to studies on the other enzymes only when they supplement or differ from those obtained with the T4 enzyme. Whereas studies have provided some information on the role of polynucleotide kinase in T4 phage-infected cells, the major importance of the enzyme is as a reagent in nucleic acid studies. In this chapter major emphasis is placed on (1) the purification of T4 polynucleotide kinase, (2) the properties and substrate specificities of the reactions catalyzed by the enzyme, and (31 the research applications of the enzyme. 1. E. Spencer, D. Loring, J. Hurwitz, and G . Monroy, PNAS 75, 4793 (1978). 2. C. C. Richardson, P N A S 54, 158 (1965). 3. A. Novogrodsky and J . Hurwitz, JBC 241, 2923 (1966). 4. A. Novogrodsky, M. Tal, A. Traub, and J. Hurwitz,JBC 241, 2933 (1966). 5 . M. Ichimura and K . Tsukada, J. Biochem. 69, 823 (1971). 6. H. Teraoka, K. Mizuta, F. Satl, M . Shimoyachi, and K . Tsukada, EJB 58, 297 (1975). 7. C. J. Levin and S. B . Zimmerrnan,JEC 251, 1767 (1976).

16. BACTERIOPHAGE T4 POLYNUCLEOTIDE KINASE

II.

A.

30 1

Isolation and Physical Properties

ASSAY

A convenient and quantitative assay for pol ynucleotide kinase measures the conversion of acid-soluble radioactivity in [y-32P]ATPto an acidinsoluble product (2, 3 , 8 ) .The most commonly used acceptor DNA substrate is prepared by partial digestion of duplex DNA with micrococcal nuclease (2. 3, 8 ) , a nuclease that hydrolyzes phosphodiester bonds to produce oligonucleotides terminated by 5’-hydroxyl and 3’-phosphoryl groups. Alternatively, 5’-hydroxyl termini can be introduced into DNA by first sonically irradiating or partially hydrolyzing DNA with pancreatic DNase followed by dephosphorylating the 5’ termini with E. coli alkaline phosphatase (9, 10). These latter substrates have been used to follow the purification of the T4 (11) and rat liver (6) enzymes, respectively.

B.

PURIFICATION

The purification procedure for polynucleotide kinase from bacteriophage TCinfected cells that was described initially consisted of a series of six steps ( 2 ) . In order to obtain a physically homogeneous enzyme and to remove traces of contaminating nuclease activity a seventh step, chromatography on hydroxylapatite, was added to the purification procedure (12). For convenience the published purification procedure (8) is summarized in Table I. An alternative procedure for the purification of the T2 polynucleotide kinase has been described ( 3 ) . The source of the enzyme described in Table I is from E. coli cells infected with T4 r+ phages. A 50 to 100% increase in polynucleotide kinase activity can be obtained by infecting E.coli B su- (nonpermissive host) with one of several T4 mutant phages defective in lysis, such as T4 amN8-7 (13) or T4 N55SP62 (14). In addition, by choosing the appropriate T4 mutant for infection, it is possible to eliminate certain potential enzyme contaminants from the very start of the purification. For example, extracts prepared from E. coli B infected with T4 amXF1 (genes 41-45) are free of 8. C. C. Richardson, Proced. Nucleic Acid Res. 2, 815 (1972). 9. C. C. Richardson, J M B 15, 49 (1966). 10. B. Weiss, T. R . Live, and C. C. Richardson, JBC 243, 4530 (1%8). 1 I . K. Sirotkin, W. Cooley, J. Runnels, and L. R. Snyder, J M B 123, 221 (1978). 12. A. Jacquemin-Sablon and C. C. Richardson, JMB 47, 477 (1970). 13. A. Panet, J . H. van de Sande, P. C. Loewen, H. G. Khorana, A. J. Raae, J. R. Lillehaug, and K. Kleppe, Biochemistry 12, 5045 (1973). 14. K. L . Berkner and W. R. Folk,JEC 252, 3176 (1977).

302

CHARLES C. RICHARDSON TABLE I

mRIFICATI0N OF POLYNUCLEOTIDE KINASEFROM OF E. coli INFECTED WITH PHAGE T4" Total units Fraction I 11

I11 IV V VI VII

Step Extract Streptomycin Auto1ysis Ammonium sulfate DEAE-cellulose Phosphocellulose H ydroxylapatite

(X

10-3)

95

90 1 I5 105 68 40 27

25 GRAMS

Specific activity (unitdmg) 40 60

I50 420 8,500 59,000

-b

From Richardson ( 8 ) . Protein concentration insufficient to obtain an accurate specific activity.

T4 DNA polymerase, since gene 43 is the structural gene for T4 DNA polymerase (15, 16). Contamination of the purified kinase by DNA ligase can be reduced by the use of T4 amH39X (gene 30)-infected cells, since gene 30 is the structural gene for the T4 DNA ligase (17). The same procedure for growth and infection as that used for wild-type T4 is followed when such mutant phages are used, except that the cells are harvested at a later time (8). Similarly, attention to the E. coli host used to prepare the infected cells can also circumvent troublesome contaminating activities. For instance, ribonuclease activity can be greatly reduced at the outset by using ribonuclease mutants of E. coli (18). The structural gene for polynucleotide kinase, the pseT gene of phage T4, also codes for the T4 3'-phosphatase (11, 1 9 ) . Although the 3'phosphatase activity does not normally interfere with the use of the polynucleotide kinase, there are occasions when it is desirable to prepare oligonucleotides having both 3'- and 5'- terminal phosphates. Escherichia coli cells infected with the T4 mutant, p s e T f , synthesize an altered polypeptide that has relatively normal kinase activity but greatly reduced levels of phosphatase activity (I 1, 20). Thus, polynucleotide kinase prep15. 16. 17. 18. 19. 20.

A. De Waard, A. V. Paul, and I. R . Lehman, PNAS 54, 1241 (1965). H. R . Warner and J. E. Barnes, Virology 28, 100 (1966). G. C. Fareed, and C. C. Richardson, PNAS 58, 665 (1967). M. Takanami, JMB 23, 135 (1967). V. Cameron, and 0. C. Uhlenbeck, Biochemistry 16, 5120 (1977). V. Cameron, D. Soltis, and 0. C. Uhlenbeck, Nucleic Acids Res. 5, 825 (1978).

16. BACTERIOPHAGE T4 POLYNUCLEOTIDE KINASE

303

arations essentially free of 3'-phosphatase activity can be obtained by purifying the kinase from extracts of T4 pseTI-infected E. coli (20). The pseTl polynucleotide kinase behaves analogous to wild-type kinase during purification. Minor modifications have been made in the purification procedure summarized in Table I by several laboratories. For example, low concentrations of ATP have been added to the buffers during purification in order to stabilize the enzyme (13,2I). Frequently it is advantageous to purify the T4 DNA polymerase and the T4 polynucleotide ligase from the same extract of infected cells. These enzymes can be purified by published procedures for the polymerase (22) and ligase (23) using the appropriate side fractions obtained from the kinase purification procedure summarized in Table I. Such simultaneous purification of the three enzymes has been described (13). In view of the extensive use of polynucleotide kinase in nucleic acid studies, considerable time, effort, and cost would be saved if the enzyme were amplified in E. coli by cloning the kinase gene, pseT. It seems likely that the pseT gene will be cloned because the difficulties encountered in generating restriction enzyme fragments of the glucosylated T4 DNA can be circumvented (24). Furthermore, restriction fragments of phage T4 DNA that contain at least portions of the kinase gene have already been cloned in E. coli (25).

C. PHYSICALPROPERTIES 1. Physical Homogeneity

T4 polynucleotide kinase has been purified to physical homogeneity as judged by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate at pH 8.3 after denaturation and reduction; a single band is also observed at pH 6.8 in the presence of 2-mercaptoethanol or during electrophoresis through urea gels (26). Sedimentation studies of the purified enzyme using the analytical ultracentrifuge also indicate that the 21. R. Wu and A. D. Kaiser, P N A S 57, 170 (1967). 22. B. Weiss, A. Jacquemin-Sablon, T. R. Live, G. C. Fareed, and C. C. Richardson, JBC 243, 4543 (I%@. 23. M. Goulian, Z. J. Lucas, and A . Kornberg, JBC 243, 627 (1968). 24. G. G. Wilson, V. I. Tanyashin, and N . E. Murray, Mol. Gen. Genef. 156, 203 (1977). 25. A. J. Mileham, H. R . Revel, and N. E. Murray, M o l . Gen. Genet. 179, 227 (1980). 26. J. R. Lillehaug, EJB 73, 499 (1977).

304

CHARLES C. RICHARDSON

enzyme is physically homogeneous. Only a single N-terminal amino acid, phenylalanine, is present in the purified enzyme (26). 2. Molecular Weight

The molecular weight of native polynucleotide kinase, as estimated by filtration through a Sephadex (3-200 column calibrated with proteins of known molecular weight, is 140,000 (13). The apparent molecular weight of the denatured and reduced form of the kinase, as determined by comparison with the mobilities of proteins of known molecular weight on polyacrylamide gels, is 33,000 (13, 26). Since phenylalanine is the exclusive N-terminal amino acid, it is likely that T4 polynucleotide kinase contains four identical subunits. In support of this structure is the finding of minor bands of 60,000 and 120,000daltons during electrophoresis of the denatured and reduced protein on polyacrylamide gels containing dodecyl sulfate at pH 6.8 (26). The szo,wof the purified enzyme has been determined under a variety of conditions (26). In 0.1 M potassium phosphate buffer, pH 7.8, containing 1 mM 2-mercaptoethanol, the enzyme preparation contains approximately 30% of a 2.9 S and 70% of a 6.5 S species, presumably the monomer and tetramer, respectively. The molecular weights of these two species of kinase have been determined by sedimentation equilibrium ultracentrifugation and were found to be 33,200 and 147,300, respectively. The activity of T4 polynucleotide kinase is stimulated by NaCl and KCl and by polyamines such as spermine (27). Potassium chloride is necessary to maintain an oligometric structure, and incubation with spermine converts the protein to a pure 6.2 S species (tetramer) (26). The polynucleotide kinase isolated from rat liver nuclei has an apparent molecular weight of 80,000 as estimated by gel filtration (6, 7). The sedimentation coefficient is 4.4 S (6). 3. Amino Acid und Spectrophotometric Analyses The amino acid composition of T4 polynucleotide kinase has been determined (26); the N-terminal amino acid is phenylalanine. The protein has two sulfhydryl groups, one exposed to the environment and the other more buried, as determined by reactivity to 5 , 5’-dithiobis(2-nitrobenzoic acid). Analysis of the ultraviolet absorption spectrum (26) shows the maximum absorption at 276 nm. The 280-260 nm absorbance ratio is 1.28. The absorption coefficient at 276 nm is 11.0 cm-’. The CD spectrum of the protein indicates that 45 to 55% of the polypeptide chain is in an a-helical conformation (26 ) . 27. J. R. Lillehaug and K. Kleppe, Biochemistry 14, 1225 (1975).

16. BACTERIOPHAGE T4 POLYNUCLEOTIDE KINASE

111.

Catalytic Properties

A.

REACTIONSCATALYZED BY POLYNUCLEOTIDE KINASE

305

Polynucleotide kinase of bacteriophage T4 catalyzes the transfer of the y-phosphate of a nucleoside 5‘-triphosphate to the 5’-hydroxyl group of a nucleoside 3’-phosphate, oligonucleotide, or polynucleotide (Fig. 1). The products of the reaction are a nucleoside 5’-diphosphate and the 5’phosphoryl nucleotide or polynucleotide (2-4, 8, 28). The reaction shown in Fig. 1 is reversible (14,29,30). In the presence of a nucleoside 5’-diphosphate, a 5’-phosphoryl polynucleotide is dephosphorylated to yield a 5’-hydroxyLterminated polynucleotide and the nucleoside 5’-triphosphate. The fact that this reaction is reversible makes possible the exchange of radioactivity between the y-phosphate of a nucleoside 5‘-triphosphate and the 5’-phosphoryl group of a polynucleotide. In addition to the phosphorylation activity of T4 polynucleotide kinase, the protein is also a 3’-phosphatase (11, 19, 3 1 ) . The enzyme’ catalyzes the hydrolysis of 3’-phosphoryl groups of deoxynucleoside 3‘monophosphates, deoxynucleoside 3’ ,5’-diphosphates, and of 3’-phosphoryl polynucleotides to yield inorganic orthophosphate and a 3‘hydroxyl group (Fig. 2).

B. PHOSPHORYLATION OF 5’-HYDROXYL TERMINI 1. Acceptor Substrutes

A variety of nucleic acid compounds can be phosphorylated in the pol ynucleotide kinase reaction provided they have a nucleotide bearing a free 5’-hydroxyl group with a phosphoryl group esterified at the 3‘ position (Table I I , 2 4 , 8, 28). Thus, the substrates include DNA and RNA, oligonucleotides, and nucleoside 3‘-monophosphates. All terminal 5’nucleotides found in DNA or RNA can be phosphorylated. The enzyme also catalyzes the phosphorylation of nucleotides whose bases contain chemically protected groups (32), such as those used in the chemical synthesis of polynucleotides . The apparent Michaelis constants and V,,, values for oligonucleotides and 3‘-mononucleotides vary depending on the 28. 29. 30. 3I . 32.

B. Weiss and C. C. Richardson, C S H S Q B 31, 471 (1966). J . H. Van de Sande, K. Kleppe, and H. G . Khorana, Biochemistry 12, 5050 (1973). ti. Chaconas, J. H. van de Sande, and R. B. Church, BBRC 66,962 (19753. A. Becker and J. Hurwitz, JBC 242, 936 (1%7). J. H. van de Sande and M. Bilsker, Biochemistry 12, 5056 (1973).

306

CHARLES C. RICHARDSON

0

k

R

FIG. 1. The transfer of phosphate from a nucleoside 5’-triphosphate to the 5’-hydroxyl group of an acceptor molecule by polynucleotide kinase. B = adenine, guanine, cytosine, thymine, or uracil; R = H, a nucleoside, a nucleotide, or a polynucleotide: R’ = H or OH. identity of the 5’4erminal base residue and the length of the oligonucleotide (33). The apparent Michaelis constant for 3’-mononucleotides is between 22.2 and 143 p M ;the apparent Michaelis constant for micrococcal nuclease-treated DNA, the substrate normally used to follow the purification, is 7.6 pM (33). Similar specificities have been found with the T2 polynucleotide kinase (3, 4). However, the kinase isolated from rat liver nuclei cannot phosphorylate RNA molecules, nor can it phosphorylate oligodeoxynucleotides of chain length less than 10 to 12 residues (7). Polynucleotide kinase catalyzes the phosphorylation of high molecular weight single-stranded and double-stranded DNA molecules in which the 5’-hydroxyl groups are in a variety of configurations. For example, it is possible to phosphorylate all of the 5’-hydroxyl groups (produced by alkaline phosphatase) in a preparation of bacteriophage T7 DNA (9, 3 4 ) , a duplex molecule consisting of 40,000 base pairs with blunt ends. However, the quantitative phosphorylation of 5’-hydroxyl groups located at single-strand interruptions in duplex DNA (nicks) is extremely difficult (10); even with prolonged incubation and large amounts of enzyme, it is not possible to phosphorylate more than 70% of such internally located groups. In order to obtain a more quantitative analysis of the phosphorylation of these large substrates, several synthetic double-stranded DNAs containing defined 5’-hydroxyl end groups have been used as substrates in the kinase reaction (35). The rate of phosphorylation of 5’-hydroxyl groups located at gaps in duplex DNA is approximately 10-fold slower than for 5‘-hydroxyl groups on the corresponding single-stranded DNA. With high concentrations of ATP complete phosphorylation of 5’-hydroxyl groups at 33. J. R. Lillehaug and K. Kleppe, Biochemisrry 14, 1221 (1975). 34. B. Weiss and C. C. Richardson, JMB 23, 405 (1967). 35. J. R. Lillehaug, R. K. Kleppe, and K. Kleppe, Biochemistry 15, 1858 (1976).

307

16. BACTERIOPHAGE T4 POLYNUCLEOTIDE KiNASE

+

0

It

HO-P-O-

I -0

-0

FIG. 2. The hydrolysis of a 3'-phosphoryl group by the 3'-phosphatase activity of polynucleotide kinase. B = adenine, guanine, cytosine, or thymine; R = H, PO,2-, a nucleotide, or a polynucleotide.

gaps can be achieved, but not at nicks. The 5'-hydroxyl groups on protruding single-stranded ends of duplex molecules are more readily phosphorylated than are 5'-hydroxyl groups on blunt ends or on molecules with protruding 3'-hydroxyl group ends (recessed 5'-hydroxyl groups). However, with sufficiently high concentrations of ATP the 5'-hydroxyl termini in all of these latter substrates can be completely phosphorylated. With the exception of DNA substrates that contain protruding 5'-hydroxyl TABLE I1 SUBSTRATES FOR POLYNUCLEOTIDE

KINASE

Substrate DNA 5'-OH-terminated T7 DNA (micrococcal nuclease treated) 5'-OH-terminated E. coli DNA (micrococcal nuclease treated) RNA E . coli tRNA (phosphatase treated) Oligonucleotides d(TpTpTpTp) d(ApApApAp) d(GpGpG p) d(CpCpCp) uPU APA

CPC GpCl

3'-AMP 3'-dAMP

Rate" (nmoVmidmg)

2200 2500 1700 2100 2300 2400 2000 2000 2600 2700 2600 2200 2900

" The rate of reaction was measured in the standard reaction mixture using Fraction VI of T4 pol ynucleotide kinase.

308

CHARLES C. RICHARDSON

ends, the phosphorylation reaction is inhibited by the addition of excess KCI (-35).

5' -Triphosphu tes A variety of nucleoside 5'-triphosphates can function as the phosphate donor. Although ATP is routinely used in the standard assays, CTP, UTP, GTP, dATP, and dTTP have all been found to function equally well. The apparent Michaelis constant for ATP is between 1.4 x M and 1.3 x W5 M depending on the DNA substrate used (4, 30). The apparent Michaelis constants for ATP, UTP, GTP, and CTP in the T2 polynucleotide kinase reaction, using micrococcal nuclease-treated DNA, are 1.4, 1.5, 3.3, and 2.5 x M , respectively ( 4 ) . 2.

Nil cleoside

3. Effect of p H mid Divaleiit Cutioiu The optimal pH range for the enzyme is 7.4 to 8.0 in Tris-HC1 buffer with maximal activity obtained at pH 7.6 (2). In the absence of added MgCl, there is no detectable activity. The optimal Mg2+concentration at pH 7.6 is 1 x M ( 2 ) . Mn2+can partially fulfill the metal requirement; at the optimal concentration of 3.3 x M , 50% of the maximal activity obtained with Mg2+is observed.

4. Sidjliydryl requirement^ Maximal activity is obtained with 5 mM dithiothreitol in the reaction mixture ( 8 ) . Protection is also achieved with 10 mM 2-mercaptoethanol and 10 mM glutathione ( J ) , but only 80 and 70%, respectively, of that observed with dithiothreitol. In the absence of a sulfhydryl compound, there is 2% of the optimal activity.

5 . liihibitov and Stirnulatory Fuctors Phosphate and pyrophosphate anions are inhibitory to T2 and T4 polynucleotide kinase (2, 4, 27). At pH 7.6 in either 70 mM sodium or potassium phosphate buffer, 5% of the value observed in Tris-HC1 buffer is obtained. Furthermore, the addition of inorganic phosphate or pyrophosphate to the standard assay mixture containing Tris-HC1 buffer results in inhibition; 20 mM inorganic phosphate or 5 mM pyrophosphate give approximately 50% inhibition (27). Since phosphate ions are relatively more inhibitory to E. coli alkaline phosphatase, at appropriate concentration of phosphate ion, polynucleotide kinase can phosphorylate polynucleotides in the presence of E . coli alkaline phosphatase (22; see also 4 8 ) . This is a convenient procedure for end-group labeling in which prior dephosphorylations by phosphatase is required. A variety of salts, such as NaCl, stimulate the reaction when single-

16. BACTERIOPHAGE T4 POLYNUCLEOTIDE KINASE

309

stranded DNA substrates are used (27). However, as previously discussed, these salts inhibit the phosphorylation of 5’-hydroxyl groups in certain duplex substrates. Polyamines such as spermine also stimulate the reaction: spermine at 1.7 mM stimulates the rate of phosphorylation approximately 30-fold (27).

c.

REVERSALOF THE KlNASE REACTION

1. Dephosphorylation of Polynucleotides

In the presence of ADP, polynucleotide kinase catalyzes the dephosphorylation of a 5’-phosphoryl group of a polynucleotide to yield a 5 ’ hydroxyl-terminated polynucleotide and ATP in a reversal of the phosphorylation reaction (29). The optimal pH for dephosphorylation is 6.2 in imidazole-HC1 buffer (14) as compared to the optimal pH for phosphorylation of 7.6 in Tris-HCI buffer. A direct comparison of the kinetic parameters of phosphorylation and dephosphorylation using 5’-protruding ends of fragments produced by EcoRI endonuclease show a striking difference in the two reactions (14). The apparent Michaelis constants for ATP and ADP in the phosphorylation and dephosphorylation reactions, respectively, are 4 ph4 and 200 p M : the corresponding V,,, are 3 1.3 and 4.3 pmol/min/pg of enzyme. The apparent Michaelis constant for ADP using a single-stranded oligonucleotide substrate is 0.22 p M (29). Even with large amounts of enzyme, however, only partial dephosphorylation is obtained (14, 29). 2. Erchcinge Retictioil Due to the reversibility of the reaction, in the presence of ADP polynucleotide kinase catalyzes an exchange of 32Pbetween the y-phosphate of ATP and the 5’-phosphoryl group of a polynucleotide (29). This exchange reaction provides a convenient method for labeling the 5’ termini of polynucleotides radioactively without prior removal of existing 5‘phosphoryl groups ( M , 29, 30). Like the forward reaction, the exchange reaction requires Mg2+ and 2-mercaptoethanol for optimal activity (29).The exchange reaction is also inhibited by phosphate ions: 50 mM potassium phosphate results in a 60% inhibition (29). The optimal pH is 6.2 in imidazole-HC1 buffer (29). The optimal rate of exchange of 3aPiinto the termini of EcoRI DNA fragments at pH 6.6 occurs at an ATP concentration of 10 p M , and an ADP concentration of 300 F M (14). The 5‘-phosphoryl end groups in a variety of polynucleotides can be labeled radioactively using the exchange reaction. Although the labeling

3 10

CHARLES C. RICHARDSON

of 5'-phosphoryl termini by the exchange reaction is usually incomplete, the extent of exchange is proportional to the number of such groups for a given amount of polynucleotide kinase (14). Furthermore, there is no significant effect of molecular weight of the acceptor nucleic acid from 4 X lo5 to 14 x lo6 on the rate and extent of labeling (14). Single-stranded oligonucleotides are most easily labeled in the exchange reaction: a 5'phosphorylated nonacosanucleotide is labeled to an extent of 96% (29). The 5'-phosphoryl recessive termini are more difficult to label, although with sufficient amount of enzyme 70% of the termini are labeled (14). The inaccessibility of 5'-hydroxyl groups at nicks to phosphorylation holds true for 5'-phosphoryl groups in the exchange reaction. The 5'-phosphoryl groups located at nicks in duplex DNA are labeled in the exchange reaction 30-fold less efficiently than 5'-phosphoryl protruding ends (14). The exchange reaction presents difficulties if one wishes to label 5'hydroxyl termini specifically. This problem can be circumvented by carrying out the phosphorylation reaction at @', a temperature at which the rate of the exchange reaction is greatly reduced (36). At 0" the exchange rate is &th of the rate at 37", whereas the rate of phosphorylation of 5'-hydroxyl termini at 0" is Ath the rate at 37". In the absence of added ADP an exchange of 32P between 5'-32Pphosphoryl groups of a polynucleotide and ATP is observed (29). The radioactive products formed are inorganic phosphate, ATP, and adenosine 5'-tetraphosphate. Although it has been proposed that the formation of Pi and adenosine 5'-tetraphosphate reflect a phosphorylated kinase intermediate (291, no such intermediate has been isolated. D. 3'-PHOSPHATASE ACTIVITY Polynucleotide kinase, in addition to catalyzing the phosphorylation of 5'-hydroxyl groups in nucleic acids, also catalyzes the hydrolysis of 3'phosphoryl groups of deoxynucleotides and other nucleic acids (Fig. 2). Initially, a 3'-phosphatase was identified and purified from E. coli cells infected with bacteriophage T4 (31 ). The purified enzyme was found to catalyze the removal of 3'-phosphoryl groups from deoxyribonucleotides, oligodeoxyribonucleotides,and DNA. The enzyme had no effect on 3'ribomononucleotides, 3'-phosphoryl groups of RNA, or any 5'-phosphate esters (31). The optimal pH of the 3'-phosphatase was between 5.8 and 6.2 in Tris-maleate buffer. In the absence of added MgClz, only 2% of maxi36. ( 1975).

R. Okazaki, S . Hirose, T. Okazaki, T. Ogawa, and Y. Kurosawa, BBRC

62, 1018

16. BACTERIOPHAGE T4 POLYNUCLEOTIDE KINASE

31 1

ma1 activity was observed: the optimal magnesium concentration was 8 mM ; and 2-mercaptoethanol , glutathione, or dithiothreitol were required for sustained activity (31). Subsequently, purified preparations of polynucleotide kinase were found to contain a 3’-phosphatase activity (19). The kinase and phosphatase activities purified together through all stages of purification, and the two activities migrated together as the major protein band during electrophoresis in polyacrylamide. These and other studies, including heat inactivation, strongly suggest that the 3’-phosphatase and kinase activities reside in the same protein molecule (19). Is the 3’-phosphatase of polynucleotide kinase identical to the previously described T4 3’-phosphatase? The two 3’-phosphatase activities have similar requirements. Both require magnesium or cobalt ion, and both have an optimal pH around 5.9 (19,31).Their substrate requirements are similar in that both activities can hydrolyze deoxynucleoside 3’monophosphates, but not 3’-ribonucleoside 3'-monophosphates . The single difference in their substrate specificity is the finding that the 3’phosphatase of polynucleotide kinase can use 3’-phosphoryl oligoribonucleotides as a substrate (19), wherea.; the previously described T4 3‘phosphatase was reported to be unable to do so (31). Additional evidence, however, for the identity of the two 3‘phosphatases comes from genetic data. A mutant of phage T4, T4 pseT, fails to induce the T4 3’-phosphatase (37) as measured by the standard 3’-phosphatase assay (31). This activity represents essentially all of the 3’-phosphatase activity found in extracts of bacteriophage TCinfected cells. Analysis of the mutant reveals that it also fails to induce polynucleotide kinase (11). A T4 mutant lacking polynucleotide kinase (38) is similarly defective in inducing the 3’-phosphatase activity (11). Specific mutations in the phage pseT gene lead to an altered polynucleotide kinase that lacks the 3’-phosphatase activity (11, 20). Whereas, mutations in thepseT gene may affect both the kinase and the phosphatase activities, polynucleotide kinase, free of the 3’-phosphatase activity, can be prepared by purifying the enzyme from E. coli cells infected with T4pseTl (20). This phage mutant induces normal amounts of kinase activity but greatly reduced levels of 3’-phosphatase (11, 20, 37). The ratio of the two activities in a wild-type enzyme can be changed significantly by inactivation of one, but not the other, activity by heating in the presence of specific substrates or products of the two reactions (19). 37. R . E. Depew and N . R. Cozzarelli, J . Viral. 13, 888 (1974). 38. V. L. Chan and K. Ebisuzaki, Mol. G e n . Genet. 109, 162 (1970).

3 12 IV.

CHARLES C. RICHARDSON

Role of Polynucleotide Kinase in Vivo

The early appearance (2, 3 ) and specificity of polynucleotide kinase, 3’-phosphatase suggest that it may be involved in the metabolism of T4 DNA. One can envision a number of roles in vivo for the enzyme on the basis of the reactions catalyzed in virro. The two activities of the enzyme would assure the presence of 3’-hydroxyl and 5’-phosphoryl groups on nucleic acid molecules; such groups are a prerequisite for most enzymatic reactions, such as ligation, involving the termini of polynucleotides. One could postulate that the two activities reside on the same polypeptide in order to carry out the simultaneous removal of a 3’-phosphoryl group and the phosphorylation of the 5’-hydroxyl group of a given molecule. For example, the two activities are capable of converting deoxyribonucleoside 3’-monophosphates to deoxyribonucleoside 5’-monophosphates, the precursors for DNA synthesis. However, it is not known whether such 3’phosphoryl-terminated mono- and polynucleotides exist in vivo . The most direct and definitive approach to determine the role of the enzyme is to examine the biochemical consequences of mutations that directly affect the enzyme. The first mutants of phage T4 found to lack polynucleotide kinase were found by direct assay of polynucleotide kinase in a variety of T4 strains that harbored other mutations (28). However, after the mutation affecting the kinase was separated from other mutations, the mutants had no observable phenotype, and hence the mutations were not mapped, and the mutants were not subjected to further study. The first detailed analysis of the polynucleotide kinase gene came, unknown, from the isolation and characterization of T4 mutants that lacked 3’phosphatase activity. The subsequent finding that 3’-phosphatase and polynucleotide kinase activities purify together (19) led to the demonstration that both activities are coded by the same structural gene (20, 37). Mutants of T4 that lack 3’-phosphatase activity were initially isolated by direct assay of 3‘-phosphatase activity in extracts of E. coli cells infected with mutagenized T4 phage (37). One such mutant, T4pseTf, grew normally on wild-type E. coli and exhibited no observable phenotype on a variety of E. coli mutants in which one could postulate a requirement for the phosphatase. However, T4pseTf failed to produce viable phage (less than 2% of a wild-type infection) on a TCsensitive Hospital strain, E. coli CT196 (37). Using this phenotype, a secondpseT mutant was isolated as well as an extracistronic suppressor T4 mutation (sfp). ThepseT gene lies between genes 63 and 31 (37).A fine structure map of the gene has been described ( I f ) , and a restriction enzyme map of this region has been obtained (25). The srp gene lies in the nonessential region between rZZB and ac (37).

16.

BACTERIOPHAGE T4 POLYNUCLEOTIDE KINASE

313

What is the biochemical manifestation of enzyme deficiency in T4pseTinfected CT196 cells that accounts for the lack of phage growth? First, the growth defects are accompanied by a 50% reduction in the rate of T4 DNA synthesis and a decrease in the length of the DNA product as measured on an alkaline sucrose gradient (37). Relatively little DNA is packaged to yield mature phage particles. Furthermore, there is a dramatic reduction in the rate of late T4 gene expression (true-late gene expression) in T4 pseT-infected E. coli CTr5X (a hybrid between K12 strains and CT196) (11 ). Hybridization experiments suggest that the defect is at the transcriptional level ( I / ) and hence the kinase, 3‘-phosphatase is probably required for normal T4 true-late transcription. The relationships, however, between these multiple defects is not known. No information is available concerning the mechanism by which the lack of kinase or 3’phosphatase activity leads to these abnormalities.

V.

Research Applications

Polynucleotide kinase is an indispensable enzyme in studies on nucleic acid structure and metabolism. Its primary value derives from its ability to phosphorylate specifically 5’-hydroxyl termini in RNA and DNA molecules. Although it is not feasible to enumerate all of the many applications of the enzyme, it is possible to cite a few that illustrate its past and present usefulness as a reagent in the study of nucleic acids. Initially the ability to label specifically 5’-termini in polynucleotides radioactively provided the only method to determine the number and identity of such end groups in polynucleotides of high molecular weight or in molecules that could not be obtained in sufficient quantities for analysis by other methods (8, 28). Similarly, this technique was useful in characterizing phosphodiester bond cleavages introduced into DNA by shearing or nuclease action (8, 28). In studies of this type it is usually necessary to distinguish between, and to determine, the number of 5’-hydroxyl and 5’-phosphoryl end groups. For this reason E. coli alkaline phosphatase is frequently used in conjunction with polynucleotide kinase. An analysis of the extent of phosphorylation of denatured nucleic acid before and after phosphatase treatment makes possible the determination of the relative number of 5’-hydroxyl and 5’-phosphoryl end groups in a preparation. Furthermore, under appropriate conditions these two enzymes can be used to determine the number and identity of external and internal (located at nicks) 5‘-end groups in DNA (10). Polynucleotide kinase is often used to prepare radioactively labeled

3 14

CHARLES C. RICHARDSON

molecules that can be used as substrates to identify or to characterize other enzymes. For example, assays for DNA and RNA ligase make use of polynucleotide substrates containing 5’-32P-phosphorylend groups (394 1 ) . Likewise, 5’-32P-labeledRNA and DNA have been used to characterize the 5’ + 3‘ hydrolytic activity ofE. coli DNA polymerase 1(42-44) and a TCinduced 5’-phosphatase (31). The enzyme has been an important reagent in the chemical synthesis of double-stranded DNAs of specific nucleotide sequence. For example, the synthesis of the structural genes for an alanine and tyrosine tRNA make extensive use of end group labeling, both to monitor and to achieve the joining of fragments (45, 46). Another use of polynucleotide kinase and end group labeling is to prepare radioactive polynucleotide molecules that cannot be labeled unifonnlyin vivo or obtained in sufficient amounts to make possible biochemical studies by methods other than those involving radioactive labeling. Shortly after its discovery the kinase was used to label digests of DNA and RNA radioactively in order to obtain fingerprints (47). Similarly, the physical mapping of restriction enzyme fragments is frequently carried out using polynucleotide kinase to label the 5’ termini of the fragments (48). Currently the most extensive use of polynucleotide kinase is in the sequencing of RNA and DNA molecules (49-51 ). Terminally labeled polynucleotides can be sequenced rapidly using partial, base-specific cleavages by chemical or enzymatic procedures such that the length of the radioactively labeled fragments identify the positions of the specific bases at which each cleavage occurs.

39. C. C. Richardson, Antiu. Rev. Biochem. 38, 795 (1%9). 40. R. Silber, V. G. Malathi, and J. Hurwitz, PNAS 69, 3009 (19721. 41. A. Sugino, T. J. Snopek, and N. R. Cozzarelli, JBC 252, 1732 (1977). 42. R. P. Klett, A. Cerami, and E. Reich, PNAS 60, 943 (1968). 43. M. P. Deutscher and A. Kornberg, JBC 244, 3029 (1969). 44. Y. Masamune, R. A. Fleischman, and C. C. Richardson,.UK 246, 2680 (1971). 45. H.G. Khorana, K. L. Agarwal, H. Biichi, M. H. Caruthers, N. K. Gupta, K. Kleppe, A. Kumar, E. Ohtsuka, U. L. Raj Bhandary, J. H. van de Sande, V. Sgaramella, T. Terao, H. Weber, and T. Yamada, JMB 72, 209 (1972). 46. E . L. Brown, R. Belagaje, M. J. Ryan, and H. G. Khorana, “Methods in Enzymology” Vol. 68, p. 109, 1979. 47. M. Szekely, Proced. Nucleic Acid Res. 2, 780 (1971). 48. G. Chaconas and J. H. van de Sande, “Methods in Enzymology” Vol. 65, p. 75, 1980. 49. H. Donis-Keller, A. M. Maxam, and W. Gilbert, Nucleic Acids Res. 4, 2527 (1977). 50. A. Simoncsits, G. G. Brownlee, R. S. Brown, J. R. Rubin, and H. Guilley, Nature (London) 269, 833 (1977). 51. A. M. Maxam and W. Gilbert, “Methods in Enzymology” Vol. 65, p. 499, 1980.

Eukaryotic DNA Kinases STEVEN B . ZIMMERMAN BARBARA H . PHEIFFER

I . Introduction and Perspectives . . . . . . . . . . . . . . . . . . I1 . Purification and Properties . . . . . . . . . . . . . . . . . . . .

A . Purification . . . . . . . . . . . . . . . . . . . . . . . . . B . Physical Properties . . . . . . . . . . . . . . . . . . . . . 111. The Catalytic Reaction . . . . . . . . . . . . . . . . . . . . A . Description of the Reaction . . . . . . . . . . . . . . . . . . B . Assay Procedures . . . . . . . . . . . . . . . . . . . . . . C. Stoichiometry and Identification of Products . . . . . . . . . . D. Requirements for Activity . . . . . . . . . . . . . . . . . . E . Reversal of the Reaction and Labeling by Exchange . . . . . . F. Kinetics and Mechanism . . . . . . . . . . . . . . . . . . . G . Inhibitors . . . . . . . . . . . . . . . . . . . . . . . . . IV. Comparison of the DNA Kinases with RNA Kinase and Polynucleotide Kinase . . . . . . . . . . . . . . . . . . . . . . . . . . . . V . Biological Role . . . . . . . . . . . . . . . . . . . . . . . . . VI . Research Applications . . . . . . . . . . . . . . . . . . . . . Note Added in Proof . . . . . . . . . . . . . . . . . . . . . .

1

.

315 316 316 318 318 318 318 319 320 322 324 324 326 327 329 329

Introduction and Perspectives

DNA kinase activity from a eukaryotic source was first demonstrated by Novogrodsky ef (11 . ( 1 ) . Extracts of rat liver nuclei were shown t o transfer phosphate groups from ATP to 5'-hydroxyl termini in DNA . Since then. enzymes with this activity have been partially purified from 1 . A . Novogrodsky. M . Tal. A . Traub. and J . Hurwitz. JBC 241. 2933 (1966)

315

.

THE ENZYMES Vol XIV Copynght @ 1981 by Academic Press. lnc All rights of reproduction in any form reserved ISBN 0-12-122714-6

3 16

S . B . ZIMMERMAN AND B . H. PHEIFFER

rat liver (2, 3 ) and calf thymus ( 4 ) . The DNA kinase from rat liver has proved to be highly specific for DNA ( 3 ) .A preparation of an enzyme from calf thymus with otherwise quite similar properties shows low activity on RNA chains in addition to its activity on DNA (4).In addition, a relatively specific RNA kinase has been partially purified from HeLa cell nuclei (5 ). The restricted acceptor specificity of these eukaryotic enzymes may be contrasted to the broad specificity of the polynucleotide kinase from T2-, T4-, or T6-infected Escherichicr coli, which has comparable activities on both DNA and RNA as well as on oligonucleotides and even on 3’rnononucleotides (6, 7). In this chapter, we will describe the properties of the eukaryotic DNA kinases and contrast them with those of the RNA kinase and of the polynucleotide kinase. The polynucleotide kinase is reviewed by Richardson in this volume (7); both DNA kinase and polynucleotide kinase have been reviewed by Kleppe and Lillehaug (6).

II. Purification and Properties

A. PURIFICATION The most thoroughly characterized DNA kinase is the activity from nuclei of rat liver (1-3,8). This enzyme has been partially purified by two independent procedures. Both preparations start by isolating nuclei. The enzyme is extracted from the nuclei with 0.15-0.2 M NaCl. Subsequent steps in the procedure of Teraoka et ul. ( 2 ) include removal of inactive materials by precipitation at pH 5, gradient elution from a phosphocellulose column, and finally gel filtration on Sephadex G-150. In the procedure of Levin and Zimmerman (-?I the nuclear extract is subjected to gradient elution from a phosphocellulose column and stepwise elution from a sulfopropyl Sephadex column. The specific activities of the final fractions from either procedure are similar and correspond to approximately a 1000-fold purification relative to a crude extract of rat liver. Two useful aspects of the procedure of Levin and Zimmerman (-3) may be 2. 3. 4. 5. 6. 7. 8.

H. Teraoka, K. Mizuta, F. Sato. M . Shimoyachi, and K. Tsukada, EJB 58,297 (1975). C. J. Levin and S. B. Zimmerman, JBC 251, 1767 (1976). G . E. Austin, D. Sirakoff, B . Roop, and G . H. Moyer, BBA 522, 412 (1978). S. Shuman and J. Hurwitz, JBC 254, 10396 (1979). K. Kleppe and J. R . Lillehaug, Adiwn. En:ymo/. 48, 245 ( 1979). C. C. Richardson, this volume, chap. 16. M . Ichimura and K. TsukadaJ. Biochern. (Tokyo)69, 823 (1971).

17. EUKARYOTIC DNA KINASES

3 17

mentioned. First, the phosphocellulose fraction has proved to be very stable, losing little or no activity for periods of 4-6 months at 4" (9); this fraction has been used extensively to characterize the activity. Second, a highly purified DNA ligase is also obtained in separate fractions of the phosphocellulose chromatography (10 ). Although preparations from both procedures are heterogeneous based upon their gel electrophoresis patterns ( I I ) , they are relatively free from interfering activities, as implied by their use for labeling the 5'-hydroxyl termini at single-strand interruptions (nicks) within duplex DNA (8, 10). More direct assays have indicated a general lack of contamination with nuclease, phosphodiesterase, or DNA ligase activities (2, 3 ) ; no phosphatase activity was detected on p-nitrophenyl phosphate (2) or 5 ' phosphate groups in DNA ( 3 ) . A very low level of nuclease activity on denatured DNA can be demonstrated with concentrated samples of the phosphocellulose fraction (/I); this activity, which is apparently the enzyme described by Cordis et d.(/.?), can be abolished with little loss of kinase activity by substituting Ca*+(0.01 M ) for Mg2+in the kinase incubation mixture (/I). A DNA kinase activity has been partially purified from homogenates of calf thymus ( 4 ) . The procedure involved protamine sulfate precipitation of inactive materials, ammonium sulfate fractionation, gradient elution from columns of phosphocellulose, hydroxyapatite and sulphopropyl Sephadex , and finally centrifugation through a glycerol gradient. The final fraction was about 1600-fold purified relative to the crude extract. The purified fractions were relatively unstable. DNA ligase, nuclease, and phosphatase (on 5'-phosphate groups in DNA) were not detected in these fractions. DNA kinase has also been partially purified from extracts of Chinese hamster lung cells grown in tissue culture (II). Extracts of washed cells in 0.2-0.4 M NaCl were made by several cycles of freezing and thawing. Diluted extracts were subjected to the same phosphocellulose chromatography that was used for the rat liver enzyme (3).A single peak of activity with the characteristic inhibition by inorganic sulfate (see Section I11 ,G,3) appeared at the same place in the gradient as did the enzymes from liver (2, 3 ) or calf thymus (4). 9. The sulfopropyl Sephadex fraction of this procedure has a half-life of a few weeks U). Both the phosphocellulose and Sephadex fractions of Teraoka et a / . are stated to be stable for at least a week at 0-4" (2). 10. S. B . Zimmerman and C. J. Levin,JBC 250, 149 (1975). 1 1 . S . B . Zimmerman, C. J . Levin, and B . H . Pheiffer, unpublished results. 12. G . A . Cordis, P. J. Goldblatt, and M. P. Deutscher, Biochemisrry 14, 2596 (1975).

318

S. B. ZIMMERMAN AND B. H. PHEIFFER

B. PHYSICAL PROPERTIES The molecular weight of the DNA kinase from rat liver has been estimated at 8 x lo4 based upon its gel filtration properties (2, 3 ) . A sedimentation coefficient ( s ~ , , ,=~ 4.4) was determined by Teraoka et al. (2). The enzyme from calf thymus is similar in size. A molecular weight of 7 x lo4 was estimated from a sedimentation coefficient = 4.3) and a Stokes radius from gel filtration of 3.9 nm, using an assumed value for the partial specific volume (4). 111.

The Catalytic Reaction

A. DESCRIPTION OF THE REACTION DNA kinase catalyzes the reversible transfer of a phosphate group between a nucleoside triphosphate and the 5’-hydroxyl moiety at the terminus of a DNA chain ( 2 - 4 ) NTP

+ 5’-hydroxyl terminus in DNA e NDP + 5’-phosphate terminus in DNA

(1)

Although the reaction is customarily assayed in the forward direction with ATP as the phosphate donor, studies of the specificity of the reverse eaction (13) indicate that the enzyme can use a number of other nucleotides. DNA is indicated as the phosphate acceptor in Eq. (1); the kinase from calf thymus may also have limited activity on 5’-hydroxyl termini of RNA chains (4).

B. ASSAYPROCEDURES The routine assay for DNA kinase measures the rate of transfer of the radioactive phosphate group of [y-32P]ATPinto an acid-insoluble form in the presence of a DNA acceptor containing 5’-hydroxyl termini ( 1 - 4 , 8). In preparation for its use as an acceptor, the DNA is either partially digested with pancreatic DNase to form 5’-phosphate termini followed by treatment with a phosphatase to yield 5’-hydroxyl termini, or is partially digested with micrococcal nuclease, which directly yields 5’-hydroxyl groups. Estimates of DNA kinase activity in crude extracts of cells or nuclei should be evaluated cautiously. Other enzymes can transfer the terminal phosphate of ATP to acid-precipitable acceptors that may be present in 13. B. H. Pheiffer and S. B. Zimmerman, Biochemistry 18, 2960 (1sv79).

17. EUKARYOTIC DNA KINASES

3 19

crude extracts. The characteristic inhibition of DNA kinase by relatively low concentrations of inorganic sulfate (see Section III,G,3) may prove useful in such situations. Other assays have been used for special purposes. For example, the rate of phosphorylation of relatively low molecular weight acceptors that are not acid-precipitable may be followed by adsorbing them to charcoal ( I , 3 ) . Also, the reverse reaction may be assayed by the rate of the nucleoside diphosphate-dependent release of radioactivity from [5'-3'P]phosphate termini in DNA into an acid-soluble form (13).

c.

STOICHIOMETRY AND IDENTIFICATION OF

PRODUCTS 1.

The Forward Reuctioti

The labeled product of phosphorylation of DNA with [y-32P]ATPby the rat liver DNA kinase was characterized as a 5'-phosphate terminus of a DNA chain by a number of criteria. As expected for such a product, the label was rendered acid-soluble by treatment with E. coli alkaline phosphatase or pancreatic DNase (3, 8) but not by treatment with pronase, pancreatic RNase, or alkali ( 3 ) .Combined treatment with a phosphodiesterase plus 5'-nucleotidase yielded 32Pi, implying the formation of a labeled 5'-phosphate group (2). The product of the kinase and [y-32P]ATP on dephosphorylated nicked DNA was characterized in some detail. When treated with pancreatic DNase and venom phosphodiesterase, the radioactivity was quantitatively recovered in the four isolated 5'deoxynucleoside monophosphates, clearly indicating the formation of 5 ' phosphate groups by the kinase (10). Further, the labeled product was sealed into phosphodiester linkage by DNA ligase from either rat liver (10, 14) or E. coli (11); these enzymes require the presence of a 5'-phosphate group for their action. The other product of the forward reaction was identified as ADP by its cochromatography with an ADP standard on a DEAE-cellulose column. The amount of ADP formed approximately matched the amount of DNA phosphorylated, consistent with the reaction as written in Eq. (1) ( 2 ) . The product formed by phosphorylation of DNA with the calf thymus enzyme was also identified as a phosphate terminus (4), based upon the sensitivity of the product to degradation by pancreatic DNase, micrococcal nuclease, or E. coli alkaline phosphatase. The radioactivity was released in a form not adsorbed by charcoal after combined treatment with 14. K . Tsukada and M . Ichimura, BBRC 42, 1156 (1971).

320

S. B. ZIMMERMAN AND B. H . PHEIFFER

venom phosphodiesterase and 5’-nucleotidase, as expected for a 5 ’ nucleotide product. The acid-precipitable product of the calf thymus enzyme on RNA was identified by its solubilization by pancreatic RNase or NaOH, but not by pancreatic DNase or pronase. 2 . The Reverse Recictioti The reverse reaction catalyzed by rat liver DNA kinase releases radioactivity from DNA bearing [5’-32P]phosphatetermini in the presence of a suitable nucleoside diphosphate ( I S ) . The acid-soluble product of the reverse reaction with ADP and P2P]DNA was identified as PPIATP by its cochromatography with authentic ATP on a DEAE-cellulose column. The 32P released from the DNA was quantitatively accounted for by the amount of [32P]ATPformed. No significant radioactivity was associated with Pi, AMP, ADP, or adenosine tetraphosphate. FOR ACTIVITY D. REQUIREMENTS

1. p H

The rates of the forward (2, 3 ) and reverse reactions (I.?) of the DNA kinase from rat liver have a similar sharp optimum at about pH 5.5, with diminished but significant activity up to at least pH 8 (Fig. 1). The rate of the forward reaction of the calf thymus enzyme has a similar pH dependence ( 4 ) . 2 . Divu I en t Cations The DNA kinase from both rat liver and calf thymus requires a divalent cation for significant activity. For the rat liver enzyme ( 2 , 3 ) ,a number of metals support kinase activity (Mg2+,Mn’+, Co2+,Zn2+,Ni2+,and Ca2+), whereas Cu2+is inhibitory. With the calf thymus enzyme (4),Mg2+, Mn2+, and Zn2+were shown to be active. 3 . Specijicity .for Nucleoside Triphosphrites ATP is the only nucleoside triphosphate that has been shown to be active in the forward reaction of the DNA kinase ( K , = 2 4 y M ) (/-4). However, the wide range of nucleoside diphosphates that are active in the reverse reaction ( I 3 1 suggests that all of the common ribo- and deoxyribonucleoside triphosphates can participate in the forward reaction. The forward reaction with labeled ATP is inhibited by the presence of these unlabeled ribo- and deoxyribonucleoside triphosphates in a manner consistent with such a broad specificity for the phosphate donor ( 3 , 4 ) .

32 1

17. EUKARYOTIC DNA KINASES

PH

FIG. 1. Effect of pH on rat liver DNA kinase. Open and closed circles indicate Trismaleate and sodium succinate buffers, respectively. From Levin and Zimmerman ( 2 ) , reproduced with permission.

4. Specificity .for the Phosphate Acceptor

The first descriptions of DNA kinase activity in crude extracts did not define its ability to phosphorylate other than DNA acceptors (I, 8). Subsequent detailed studies indicated that the DNA kinase purified from rat liver is specific for DNA (and long oligodeoxynucleotides) and has little or no activity on RNA (3, 15). In these studies, all of the potential substrates tested were also assayed with the polynucleotide kinase from T4-infected cells (7) to ensure that negative results with the rat liver enzyme were a function of the specificity of that enzyme, and not due to a defect in the substrate. a. DNA versus R N A . The routine substrate for the DNA kinase from rat liver is DNA that has been enriched in 5’-hydroxyl termini (see Section 15. Teraoka et r d . ( 2 ) cite preliminary results with the purified enzyme from rat liver which indicated to them that RNA can act as a phosphate acceptor. In the absence of their experimental results and in view of the experimental evidence cited here in opposition to their conclusion, it is our opinion that the rat liver DNA kinase is indeed highly specific for DNA relative to RNA.

322

S . B. ZIMMERMAN A N D B. H. PHEIFFER

111,B). In contrast to its activity on such DNA preparations, the DNA kinase was inactive on similarly treated samples of RNA or of poly(rA) (Table I) (3, 15). The lack of activity on these materials was not due to contaminating nucleases that destroyed either the phosphate acceptors or the phosphorylated products. Such possibilities were ruled out by the results of sequential incubations of the rat liver and T4 kinases. The lack of activity of the rat liver enzyme on RNA was also maintained in the presence of divalent cations other than Mg2+, as well as under the T4 kinase assay conditions (3). The calf thymus preparation differed from that of rat liver in that the thymus preparation phosphorylated RNA containing 5'-hydroxyl termini at about 10% of the rate with which it acted upon DNA (4). This apparent difference in acceptor specificity is further discussed in Section IV. The rat liver kinase phosphorylates 5'-hydroxyl termini in DNA that are joined to any of the four usual bases (see Section III,C, 1). Further, any of these four bases can also be present on the 3'-hydroxyl side of a nick that is being phosphorylated (10). Both native and denatured DNA were generally phosphorylated at similar rates (Table I) and to similar extents, except that the extent of phosphorylation of micrococcal nuclease-treated DNA increased greatly upon denaturation, suggesting an inhibition by 3'-phosphate groups adjacent to a site of phosphorylation (3). b . Nucleotides and Oligonucleotides as Phosphate Acceptors. Several deoxydinucleoside monophosphates and deoxynucleoside 3'-mOnOphosphates (as well as ribonucleoside 3'-monophosphates) were not substrates for the rat liver enzyme (Table I), although they were all readily phosphorylated by the T4 kinase ( 3 ) . The dependence of the kinase activity on the chain length of oligodeoxynucleotides was tested with a series of partial DNase I digests of calf thymus DNA (3). After dephosphorylation, the digests were incubated with DNA kinase and [y-32P]ATP.Digests with average chain length of -6-9 residues were relatively inactive, whereas digests with average chain length of 13 or more residues were acted upon at rates similar to the rate on denatured DNA. These results suggested that a chain length of more than 10-12 residues is required for rapid phosphorylation. The size distribution of labeled oligodeoxynucleotides in a partial digest was consistent with this conclusion. E.

REVERSALOF THE REACTIONAND LABELING BY

EXCHANGE Reversal of the kinase reaction could be readily demonstrated (131, although the rate was several orders of magnitude slower than the rate of

323

17. EUKARYOTIC DNA KINASES

TABLE I

PHOSPHATE ACCEPTORSPECIFICITY OF RAT LIVERDNA KINASE~ Relative rate of kinase activity Experiment

Substrate

(%)

A

DNA, micrococcal nuclease-treated DNA, micrococcal nuclease-treated, heat 5 min at loo", quench poly(rA), micrococcal nuclease-treatedb RNA, micrococcal nuclease-treated" RNA, micrococcal nuclease-treated, heat 5 min at loo", quench DNA, micrococcal nuclease-treated Ado 3'-P, Guo 3'-P, Cyt 3'-P, or Urd 3'-P dCyt 3'-P or dThd 3'-P dT-dC or dC-dT

100

B

83 <2 <2

Adapted from Levin and Zimmerman (3).

'' The lack of activity of the kinase on these substrates was observed on samples treated with levels of micrococcal nuclease producing from 6 to 50% acid-soluble materials. With the T4 kinase, these samples gave rates of activity within twofold of the rate of that enzyme upon micrococcal nuclease-treated DNA.

the forward reaction. The reverse reaction required the presence of a divalent cation and any one of a variety of nucleoside diphosphates; ADP, GDP, CDP, UDP, dADP, dGDP, dCDP, and dTDP all supported the reverse reaction, whereas ATP, AMP, PPi, and Pi were not active. (Identification of products is summarized in Section III,C,2.) The pH dependence of the reverse reaction showed a sharp optimum at -pH 5.5, which was indistinguishable from that of the forward reaction. There was also no obvious differential effect of incubation temperature on the forward and reverse reactions, both rates increasing about fourfold between 20"and 37" and more than eightfold between 0 and 20". The reverse reaction with heat-denatured DNA occurred at a rate 5 times less than that with native DNA. It is pJssible to introduce radioactivity at nicks without prior dephosphorylation by using the DNA kinase in an exchange reaction (13) similar to that previously demonstrated for polynucleotide kinase (7). DNA bearing unlabeled 5'-phosphate termini was incubated with a mixture of ADP and [y-32P]ATPin the presence of the enzyme. As expected for an exchange reaction, the rate of incorporation of label into DNA is increased by the presence of the ADP, as opposed to labeling of dephosphorylated nicks by the forward reaction where ADP acts as an inhibitor.

324

S . B. ZIMMERMAN AND B. H. PHEIFFER

F. KINETICSAND MECHANISM The reaction catalyzed by the rat liver DNA kinase apparently proceeds by a sequential mechanism (i.e., without involvement of a covalently linked enzyme-substrate complex) (13). Double-reciprocal plots of the initial velocity as a function of ATP concentration at different DNA levels or of DNA concentration at several ATP levels both gave series of nonparallel lines, indicative of a sequential mechanism. Also, attempts to demonstrate an enzyme-dependent exchange reaction between nucleoside di- and triphosphates that would be expected to occur with a covalent enzyme-substrate complex were unsuccessful. Either DNA alone or ATP alone protected the kinase from thermal inactivation under assay conditions ( 1 3 ) . Hence, the enzyme can apparently interact with either substrate in the absence of the other, suggesting a random order of interaction with the substrates. There is no information on the order of release of products. G. INHIBITORS 1. Ionic Strength

The kinases from rat liver (3)and calf thymus ( 4 )are subject to a similar nonspecific inhibition by monovalent salts, with 50% inhibition at about 0.2 M salt. The rat liver enzyme shows a small stimulation at lower ionic strengths (3 ).

2. Sirlfi ydryl Conipouiitis The enzymes from both liver (2, 3 ) and thymus (4) appear to have essential sulfhydryl groups. A mercaptan such as dithiothreitol or P-mercaptoethanol was generally added to enzyme preparations and their assay mixtures. Although omission of the mercaptan from these media usually had only slight effects, dialysis of enzyme in its absence caused significant losses of activity. The rat liver kinase was inhibited by p-chloromercuribenzoate (2, 3 ) ; both the rat liver and thymus enzymes were inhibited by iodoacetate or AgN03 (3, 4). 3. Inorganic Sulfnte and Other Inorganic Anioris The forward reaction of the rat liver kinase is inhibited by remarkably low levels of inorganic sulfate (-50% inhibition by 0.3 m M N%S04under routine assay conditions) and related compounds, such as N+Se04 or Na2W04( 2 , 3 )(Fig. 2). The inhibition by Na,S04 is competitive with ATP ( K i = 0.2 mM) and noncompetitive with DNA (13). Inhibition of the

SALT (rnM)

FIG.2. Effect of various anions on rat liver DNA kinase. From Levin and Zrnmerman (3). reproduced with permission.

326

S. B. ZIMMERMAN AND B . H. PHEIFFER

reverse reaction requires much higher levels of sulfate (50% inhibition at 4 mM N%,S04)(13).The kinase activity is inhibited by relatively low levels of PPi (50% inhibition at 0.2 mM), but not by NaF (> 10% inhibition at 10 mM) (2, 3 ) . The calf thymus enzyme appears to be less sensitive to inorganic sulfate [60% inhibition at 0.01 M (NH4)2S04]and is also inhibited by PPi (4). 4. Organic Sufur. Compounds

The rat liver DNA kinase was inhibited by low levels of certain sulfurcontaining polymers (3, 13). For example, dextran sulfate or heparin gave 50% inhibition at -20 or 60 ng/ml, respectively, under assay conditions, whereas samples of agar or chondroitin sulfates A or C (or dextran) did not inhibit significantly even at 15 pg/ml. Dextran sulfate inhibition was competitive with respect to both DNA and ATP, and inhibited the forward and reverse reactions to similar extents. The calf thymus enzyme was inhibited 75% by a sample of heparin at 1 mg/ml and was not inhibited by agar at three times that level (4). 5. Miscellaneous Compounds

A number of compounds that either influence the rate of the microbial polynucleotide kinase or that are of intrinsic interest were tested with the rat liver kinase (spermine, spermidine, adenosine 3', 5'-monophosphate, adenosine 5'-phosphosulfate); all were without strong effect (3, 13). Actinomycin D did not inhibit the activity of the enzyme in crude extracts of liver nuclei (1, 8).

IV.

Comparison of the DNA Kinases with RNA Kinase and Polynucleotide Kinase

This review has been primarily concerned with the isolation and properties of the DNA kinases from rat liver and calf thymus. These enzymes are notable for the much greater rate at which they phosphorylate DNA termini than RNA termini, although the calf thymus preparations show some activity on RNA termini. In this connection, the report by Shuman and Hurwitz (5) of a eukaryotic RNA kinase is particularly interesting. They found that extracts of nuclei from HeLa cells contained two physically separable kinase activities, namely, a DNA kinase activity that is probably similar to those previously described, and in addition an RNA kinase with distinctly different properties. The RNA kinase has been partially purified and characterized in some detail. Although it is relatively

17. EUKARYOTIC DNA KINASES

327

specific for RNA termini, it also shows a low rate of phosphorylation on DNA termini. The demonstration of the RNA kinase emphasizes a need for caution in evaluating the low levels of activity on RNA that were shown by partially purified preparations of DNA kinase from calf thymus. As suggested by Austin et ril. ( 4 ) ,it is possible that the activities they observed on DNA termini and RNA termini were actually a result of the presence of two separate enzymes. In addition to these eukaryotic enzymes that phosphorylate 5’-hydroxyl termini of DNA or RNA, there is also the polynucleotide kinase from T2-, T4-, or T6-infected E . coli, which phosphorylates both species of termini, as well as other acceptors (6, 7). A comparison of some distinctive properties of all of these kinases is made in Table 11.

V.

Biological Role

The biological function of the DNA kinase is not known. It seems reasonable to ascribe to the enzyme a role in the repair of damage to DNA (2-41, particularly in view of its activity at single-strand breaks in duplex DNA. DNA kinase can convert 5’-hydroxyl termini to 5’-phosphate groups at such nicked locations, thus providing the proper substrate for subsequent joining to apposed 3’-hydroxyl groups by the DNA ligase (19, 20). The distribution of the DNA kinase is consistent with a role in DNA repair processes. The activity is apparently confined to the cell nucleus in rat liver (2, 3 ) . The enzyme is found in cells of several species: rat ( I ,3 ) , hamster ( I I ) , and probably also in human (5, 11, 11, 21) and frog (13). There is no evidence as yet for a control function of the DNA kinase. Its activity was not changed after partial hepatectomy in the rat (2),nor was it significantly different in extracts of Chinese hamster lung cells from the exponential or stationary phases of growth ( I I ). The apparent polynu16. C . C. Richardson, PNAS 54, 158 (1965). 17. R. Wu, BBRC 43, 927 (1971). 18. J . R. Lillehaug and K. Kleppe, B i o c h e r n i s ~ y14, 1225 (1975). 19. I . R. Lehman, Scietice 186, 790 (1974). 20. S. Soderhall and T. Lindahl, FEES (Fed. O r r . Riocliem. S o c . ) Lett. 67, 1 (1976). 21. G . C. F. Pedrali Noy, L. Dalpra’, A. M. Pedrini, G . Ciarrocchi, E. Giulotto, F. Nuzzo, and A. Falaschi, Nircleic Acids Res. I , 1183 (1974). 22. A. M. Pedrini, L. Dalpra’, G. Ciarrocchi, G. C. F. Pedrali Noy, S. Spadari, F. NUZZO, and A. Falaschi, Nrrcleic Acids Res. 1, 193 (1974). 23. H. Saiga and T. Higashinakagawa, Nircleic Acids Res. 6, 1929 (1979).

TABLE I1

KINASES,RNA KINASEAND T4 POLYNUCLEOTIDE KINASE

COMPARISONS OF DNA

DNA kinase Rat liver Characteristic pH optimum Apparent K , for ATP ( p m ) Phosphate acceptors Divalent cations

Teraoka et al. ( 2 )

Levin and Zimmerman ( 3)

5.5

5.5

2

2

DNA

Mg2+, Mn2+, Cap+

0.5

5.5 4

RNA kinase ( 5 )

DNA, RNA, oligonucleotides, nucleoside 3'-phosphates (6. 7) Mg2+, Mn'+ , Znp+, Co2+ , Nil+ , Ca2+ (Cu2+inhibitory) ( / )

1

(Cu2+ inhibitory) 0.3

8 x 10'

8 x 104

T4 Polynucleotide kinase

7.9 to 8.9 500

DNA, oligodeoxynucleotides with n > 10-12 Mg2+, Mn2+, Zn2'. COW Ni2+ Ca2+ 7

[SOZ-] for 50% inhibition (mM) Molecular weight

Calf thymus Austin et a / . (4)

b

7 x 104

-

14 x lo4 ( 6 )

10 mM (NH,),SO, produced 60% inhibition ( 4 ) .

* Inorganic sulfate is inhibitory to the T4 polynucleotide kinase (K, = 600 &ml) actually stimulate activity in assays conducted at low ionic strength (18).

(/7), although sulfate at high concentrations has been reported to

17. EUKARYOTIC DNA KINASES

329

cleotide kinase activity did not undergo large changes in lymphocytes as a function of time after phytohemaglutinin stimulation (2f,24), nor was it greatly different in tissue culture cells from several patients with xeroderma pigmentosum as compared to control cells (2,24). It is not known whether the DNA kinase has an intrinsic 3’-phosphatase activity in addition to its phosphorylating capacity. If the eukaryotic enzyme turns out to be similar to polynucleotide kinase (6, 7) in having both activities, this will certainly strengthen the case for the involvement of the DNA kinase in DNA repair. Finally, it should be noted that preparations of both the DNA kinase from calf thymus (4) and the RNA kinase from HeLa cells (5) have appreciable activity on both RNA and DNA termini, albeit with quite different relative rates, pH optima, etc. If the relatively low activity on RNA of the calf thymus DNA kinase preparations turns out to be due to the DNA kinase per se, the latter enzyme would be a second agent (along with the RNA kinase) capable of phosphorylating RNA termini. As noted by Shuman and Hurwitz (5) and by Winicov (3), such phosphorylation may well be involved in the sequence of reactions for “capping” and “splicing” during RNA processing. In a related way, it may be noted that the RNA kinase activity on DNA termini provides a second source of phosphorylation of DNA termini. VI.

Research Applications

The DNA kinase of rat liver has been used for some years to introduce labeled 5’-phosphate groups at single-strand breaks in duplex DNA (14). Such labeled DNA may be conveniently used to assay DNA ligase activity (14, 2 ) . Also, we note that although such application has not yet been made, the specificity of the various eukaryotic kinases discussed in this review suggests their possible use to quantitate RNA and DNA termini in heterogeneous samples. Note Added in Proof

Since this paper was submitted, a further purification of the bovine DNA kinase has been described [S. Tamura, H. Teraoka, and K. Tsukada, EJB 1 15, 449 (1981)l. In contrast to earlier results ( 4 ) , the more highly purified enzyme is inactive on RNA; hence both the bovine and rat DNA kinases appear to be specific for DNA. 24. The kinase assays in references (21)and (22)were done under conditions optimal for the T4 polynucleotide kinase and may not be a measure of the DNA kinase in these cells. 25. 1. Winicov, Biorhemistry 16, 4233 (1977).

This Page Intentionally Left Blank

Type I DNA Topoisomerases JAMES C . WANG

I . Introduction . . . . . . . . . . . . . . . . . I1 . Purification and Properties . . . . . . . . . . . A . Bacterial Enzymes . . . . . . . . . . . . B . Eukaryotic Enzymes . . . . . . . . . . . C . Viral Enzymes . . . . . . . . . . . . . . I11 . The Reactions Catalyzed by the Enzymes . . . . A . Interconversions between Topological Isomers B . Mechanistical Considerations . . . . . . . . IV. Biological Roles . . . . . . . . . . . . . . . A . Bacterial Enzymes . . . . . . . . . . . . B . Viral Enzymes . . . . . . . . . . . . . . C . Other Enzymes . . . . . . . . . . . . . . V . Research Applications . . . . . . . . . . . . .

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

. . . . . . .

. . . . . . .

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

. . . .

. . . .

. . . .

. . . .

. . . . . . . .

. . . .

. . . . .

332 334 334 334 335 331 337 331 341 341 342 342 343

331 THE ENZYMES. Vol . XIV Copyright 0 1981 by Academic Press Inc . All rights of reproduction in any form W e N e d ISBN 0-12-122714-6

.

332 1.

JAMES C . WANG Introduction

DNA topoisomerases are enzymes that catalyze the breakage and rejoining of DNA strands. The breakage of a DNA backbone bond by a topoisomerase is followed efficiently by the rejoining of the same bond. Therefore the disjoined state is a transient one, and each cycle of breakage and rejoining of a bond is almost invariably detected by the conversion of one DNA topological isomer to another, and hence the name topoisomerase (I ). There are several classes of activities that can promote the breakage and rejoining of DNA backbone bonds. Some differ from the archetype DNA topoisomerases mainly in that the disjoined state may persist for a long period of time. The well-known examples of such activities are the gene A product of phage 4x174 (2-5) and the gene I1 product of fd ( 6 ) . The 4x174 gene A protein, for example, cleaves the viral strand of a negatively supercoiled double-stranded 4X 174 DNA at a unique position, generating a 3’-hydroxyl group, and the protein itself becomes linked covalently to the 5‘-phosphoryl end. This covalent complex can be isolated readily. In the presence of other proteins and cofactors that are required for the net synthesis of 4x174 viral strands, extension of the gene A protein nicked-viral strand proceeds by the addition of nucleotides to the 3’ terminus. Eventually, after one round of DNA replication, displacement of the covalently linked gene A protein occurs with the concomitant rejoining of the 5’-phosphoryl group to a 3’-hydroxyl group to give a complete copy of the single-stranded circular genome. In other cases, although the backbone bonds that are broken by such enzymes are of the same type as the backbone bonds that are reformed, the pair of groups that are brought together in the rejoining step may not be the same pair of groups disjoined by the enzyme. The h inr enzyme, for example, can perform the transient cleavage of two viral DNA strands and two host DNA strands when combined with host proteins. Rejoining of these two pairs of broken duplexes occurs, however, after pairwise switches, forming two double-stranded junctions between the host and I . J. C. Wang and L. F.Liu, in “Molecular Genetics, Part 111” (J. H. Taylor, ed.), p. 65. Academic Press, New York, 1979. 2. J.-F. Ikeda, A. Yudelevich, and J. Hurwitz, PNAS 73, 2669 (1976). 3. J. F. Scott, S. Eisenberg, L. L. Bertsch, and A. Korngerg, PNAS 74, 193 (1977). 4. J.-E. Ikeda, A. Yudelevich, N . Shimamoto, and J . Hurwitz, JBC 254, 9416-9428 (1979). 5 . S . Eisenberg, J. Grfith, and A. Kornberg, P N A S 74, 3198 (1977). 6. T. F. Meyer and K . Geider, JBC 254, 12642 (1979).

18. TYPE I DNA TOPOISOMERASES

333

viral DNAs. This strand breakage, transfer, and rejoining leads to the linear insertion of the double-stranded circular viral genome into the host genome (for reviews, see Refs. 7 and 8). The distinctions between the archetype DNA topoisomerases and enzymes such as 4x174 gene A protein, and A int protein are not always apparent. Under certain conditions 4x174 gene A protein, fd gene I1 protein, and A irzt protein all exhibit classical DNA topoisomerase activities. Relaxation of a negatively supercoiled DNA by these enzymes has been demonstrated (4, 6, 9). There is also the possibility of reciprocating analogies, i.e., the archetype topoisomerases might under certain conditions show activities similar to those of the viral enzymes described here (see Section IV,A). In this chapter, discussion of proteins of the 4x174 gene A product and the A int types is of a limited nature; the emphasis is placed on the more classical topoisomerases. The known DNA topoisomerases can be divided into two categories, type I and type I1 (10). The type I enzymes appear to catalyze the breakage and rejoining of DNA strands one at a time. As a general rule they require no energy cofactors such as ATP or NAD. A number of enzymes that were discovered in the early 1970s, including Esclzerichia coli DNA topoisomerase I [o protein (I1)] and the rat liver DNA topoisomerase I [the nicking-closing enzyme (1211, belong to this category. The type I1 enzymes appear to catalyze the breakage and rejoining of DNA strands in a coordinated way such that in the disjoined state the DNA can be modeled as containing a double-stranded break. Several enzymes that were discovered in the late seventies, including E. coli DNA gyrase (13 1, the ATP-dependent topoisomerases from phage T4 ( 1 4 , and several eukaryotic organisms (10, 15-17), belong to this category. These enzymes are usually ATP-dependent, at least in some of the topoisomeri7. H . Nash, Curr. Topics Microbiol. Immirnof. 78, 171 (1977). 8. R . A. Weisberg, S. Gottesman, and M. E. Gottesman, in “Comprehensive Virology” (H. Frankel-Conrat and R . R. Wagner, eds.), Vol. 8 , p. 197. Plenum, New York, 1977. 9. Y. Kikuchi and H. Nash, PNAS 76, 3760 (1979). 10. L. F. Liu, C.-C. Liu, and B. M. Alberts, Ceff 19, 697 (198G). 1 1 . J. C. Wang,JMB 55, 523 (1971). 12. J . J . Champoux and R . Dulbecco, PNAS 69, 143 (1972). 13. M. Gellert, K . Mizuuchi, M. H. O’Dea, and H. A. Nash, PNAS 73, 3872 (1976). 14. L . F. Liu, C.-C. Liu, and B . M . Alberts, Nrrture (London) 281, 456 (1979). 15. T.-S. Hsieh and D. Brutlag, Cefl 21, 115 (1980). 16. M. I . Baldi, P. Benedetti, E . Mottoccia, and G . P. Tocchini-Valentini, CefI 20, 461 (1980). 17. L . F. Liu, in “Mechanistic Studies of DNA Replication and Genetic Replication,” ( B . M . Alberts and C. F. Fox, eds.), ICN-UCLA Symp. Mol. Cellular Biol., Vol. 19. Academic Press, New York, 1980.

334

JAMES C. WANG

zations they catalyze. Only type I enzymes are reviewed in this chapter; type I1 activities are reviewed in Chapter 19 of this volume (18). II.

Purification and Properties

A. BACTERIALENZYMES Following the initial discovery and purification of the E. coli DNA topoisomerase I (11, 19), similar enzymes were purified from Micrococcus luteus (20, 2 I ) , Bacillus magaterium (Z),Salmonella typhimurium ( I 1, and Agrobacteriurn tunzefaciens, a bacterium that induces crown gall tumors in plants (23). The purification of the bacterial enzymes follows in general the widely practiced steps of cell lysis, nucleic acid removal, ammonium sulfate precipitation, and chromatographic separations. The relaxation of a negatively supercoiled DNA is commonly used to monitor the enzymatic activity during purification. A discussion of the various assays that can be used to monitor type I and type I1 topoisomerases can be found in (24). Several of the bacterial enzymes have been purified to homogeneity or near homogeneity (f9-22). All of these are found to be single-subunit proteins with a molecular weight of 100,000-120,000. Mg2+ ions are required for activity. The enzymes are in general insensitive to the sulfhydry1 reagent N-ethylmaleimide, with the possible exception of the B. magaterium enzyme (22). There have been no systematic studies on the sequence homologies between these enzymes. The enzymes from the gram-negative bacterium E. coli and the gram-positive bacterium M . luteus are sufficiently different to be devoid of cross-reactivity toward rabbit antibodies directed against them (20). B. EUKARYOTIC ENZYMES

The first eukaryotic activity studied was from mouse embryo nuclei (12). Similar activities have since been found in mouse, monkey and 18. M. Gellert, This volume, Chapter 19. 19. R. E. Depew, L. F. Liu, and J. C. Wang, JBC 253, 511 (1978). 20. V. T. Kung and J. C. Wang, JBC 252, 5398 (1977). 21. R . Hecht and H. W. Thielman, Nucleic Acids Res. 4, 4235 (1977). 22. M. G. Burrington and A. R. Morgan, C m . J . Biochem. 54, 301 (1976). 23. J. M. LeBon, C. I. Kado, L. J. Rosenthal, and 3. G . Chirikjian, PNAS 75, 4097 (1978). 24. J. C. Wang and K. Kirkegaard, in “Gene Amplification and Analysis, Vol. 11: Analysis of Nuclei Acid Structure by Enzymatic Methods” (J. G. Chirikjian and T. S. Papas, eds.). Elsevier, North Holland, 1981.

18. TYPE I DNA TOPOISOMERASES

335

human tissue culture cells (25-27), rat and bovine liver (28,29),yeast OO), calf thymus (-?f1, duck (32) and chicken U3) erythrocytes, and the eggs of drosophila (15, 341, frog (35,361, and sea urchin (37). The purification of the eukaryotic type I topoisomerases involves typically the isolation of cell nuclei as the initial step. Subsequent fractionations by polyethylene glycol extraction and chromatography are commonly carried out. The earlier preparations yielded typically proteins with molecular weights of about 65,000. The rat liver enzyme, for example, was purified to apparent homogeneity as a 66,000 dalton polypeptide (28). Similarly, the enzymes from mouse L cells and human KB cells have been fractionated to a high degree of purity with molecular weights of 68,000 (25) and 60,000 (27), respectively. These preparations showed no dependence on divalent ions. The use of proteinase inhibitors at higher concentrations and more rapid fractionation procedures has yielded enzymes with molecular weights around 100,000 (17,38). Controlled proteolytic cleavage of these fractions shows that the 67,000 dalton species gives fragments that are a subset of the proteolytic fragments of the 100,000 dalton species. Thus it appears that the 67,000 dalton protein is itself a proteolytic fragment of the 100,000 dalton enzyme (17). Type I topoisomerases with molecular weights around 100,000 have been purified from HeLa (17), wheat germ (39), and Drosophilci embryos (38). The dependence of the 100,000 dalton enzyme on divalent ions is yet to be established. The HeLa enzyme appears to be greatly stimulated by Mg2+(17).

C. VIRALENZYMES DNA topoisomerase activity is clearly detectable in purified gene A protein of phage 4x174 and gene I1 protein of phage fd when DNAs that 25. H. P. Vosberg and J. Vinograd, BBRC 68, 456 (1976). 26. R. J. DeLeys and D. A. Jackson, Nucleic Acids Res. 3, 641 (1976). 27. W. Keller, PNAS 72, 2550 (1975). 28. J. J. Champoux and B . L. McConaughy, Biochemistry 15, 4638 (1976). 29. D. Kowalski, A t i d . Biochem. 107, 31 I (1980). 30. J. M . Durnford and J . J. Champoux, JBC 253, 1086 (1978). 31. D. E. Pulleyblank and A. R . Morgan, Can. J . Eiochem. 14, 5205 (1976). 32. R . D. Camerini-Otero and G. Felsenfeld, N i d e r c Acids Rcs. 4, 1159 (1977). 33. M . Bina-Stein, T. Vogel, D. S. Singer, and M . F. Singer, JBC 251, 7363 (1976). 34. W. A. Baase and J. C. Wang, Biochemistry 13, 4299 (1974). 35. E. Mattochia, D. G . Attardi, and Tochini-Valentini PNAS 73, 4551 (1976). 36. R . A. Laskey, A. D. Mills, and N. R . Morris, Ce// 10, 237 (1977). 37. D. L. Poccia, D. LeVine, and J. C. Wang, Dcv. B i d . 64, 273 (1978). 38. J . C. Wang, R . I. Gumport, K . J . Javaherian, K. Kirkegaard, L. Klevan, M . L. Kotewicz, and Y.-C.Tse, in "Mechanistic Studies of DNA Replication and Genetic Repli-

336

JAMES C. WANG

contain unique sequences recognized by these enzymes are employed as the substrates (4, 6 ) . I n viiw these enzymes are involved in the initiation of the viral DNA replication in a manner different from the archetype topoisomerases, as described for the #XI74 gene A protein in Section I. The nucleotide sequence of genes coding for these enzymes has been determined (40-42), and therefore the protein sequences of the enzymes are readily deduced. Their molecular weights calculated from electrophoretic mobilities in SDS-polyacrylamide gels and from gel filtration [56,000 for the 4x174 gene A protein (43) and 45,000 for the fd gene I1 protein (#)I are in excellent agreement with those calculated from the sequence data. As mentioned in Section I, the phage A int gene product also shows DNA topoisomerase activity under certain assay conditions. In contrast to the integration reaction, which is sequence specific, the topoisomerase activity shows little sequence specificity (9). The int gene sequence has also been determined ( 4 3 , from which it is deduced that the enzyme contains a single subunit of a peptide weight 40,000. Of the 356 amino acids making up the polypeptide chain, 69 are basic and 46 are acidic. The molecular form of the int enzyme in solution is unknown. At least in its catalysis of the sequence-specific integration reaction in conjunction with a host protein, a multimeric species appears to be the active form (46,47). A topoisomerase activity from vaccinia virus core has been partially purified (48).The most highly purified fraction of this activity contains two polypeptides of molecular weights 24,000 and 35,000, and the activity appears to be associated with fractions that are enriched in the 35,000molecular-weight species. The classification of the vaccinia enzyme as a type I topoisomerase is a tentative one, since no test has yet been reported in this regard. cation” (B. M. Alberts and C. F. Fox, eds.), ICN-UCLA Symp. Mol. Cellular Biol., Vol. 19 Academic Press, New York, 1980. 39. W. S. Dynan, J. J. Jendrisak, D. A. Hager, and R. R. Burgess, JRC 256,5860 (1981). 40. F. Sanger, G. M. Air, B. G. Barrell, N. L. Brown, A. R. Coulson, J. C. Fiddes, C. A. Hutchinson 111, P. M. Slocombe, and M. Smith, Nnrure (London) 256, 687 (1977). 41. F. Sanger, A. R. Coulson,T. Friedmann,G. M. Air, B. G. Barrell, N. L. Brown, J. C. Fiddes, C. A. Hutchison, Ill, P. M. Slocombe, and M. Smith, J M B 125, 225 (1978). 42. E. Beck, R. Sommer, E . A. Auerswald, Ch. Kurz, B. Zink, G. Osterburg, and H. Shaller, Niicleic Acids. Res. 5, 4495 (1978). 43. T. J. Henry and R. Knippers, PNAS 71, 1549 (1974). 44. T. F. Meyer and K. Geider, JBC 254, 12636 (1979). 45. R. H. Hoess, C. Foeller, K. Bidwell, and A. Landy, PNAS 77, 2482 (1980). 46. W. Ross, A. Landy, Y. Kikuchi, and H. Nash, Cell 18, 297 (1979). 47. R. W. Davies, P. H. Schreier, M. L. Kotewicz, and H. Echols, Nucleic Acids Res. 7, 2255 (1979). 48. R. W. Bauer, E. C. Ressner, J. Kates, and J. V. Patzke, PNAS 74, 1841 (1977).

18.

111.

TYPE I DNA TOPOISOMERASES

337

The Reactions Catalyzed by the Enzymes

A. INTERCONVERSIONS BETWEEN TOPOLOGICAL ISOMERS Four DNA toposiomerization reactions catalyzed by the type I enzymes have been reported: The relaxation of negatively or positively supercoiled DNA (//, / 2 ) , the intertwining of single-stranded rings of complementary sequences (49, 50), the interconversion between singlestranded rings with and without topological knots ( 5 / ) ,and the catenation and decatenation of a pair of double-stranded rings with at least one preexisting single-chain scission in one of the rings (52). The intertwining of single-stranded rings of complementary sequences can be more appropriately viewed as a special case of the relaxation of a negatively supercoiled DNA, since two unlinked complementary rings can be considered as a duplex ring in a very underwound or highly negatively supercoiled state. B . MECHANISTICALCONSIDERATIONS 1. Single-Stranded Breaks and Intermediate

It is apparent from an examination of the reactions described in Section III,A that a type I enzyme must be capable of introducing transient single-stranded breaks into a DNA molecule. One of the earliest postulations on the mechanism of a topoisomerase-catalyzed reaction is that the transient breakage of a DNA backbone bond is accompanied by the simultaneous formation of a covalent protein-DNA bond (11). This postulate was based on the lack of an energy cofactor requirement for the E. coli type I enzyme, and the knowledge from studies with DNA ligases that if hydrolysis of a DNA phophodiester bond had occurred there would have been an energy cofactor requirement. For bacterial type I topoisomerases, it has been shown that addition of a protein denaturant upon incubation of an enzyme with single-stranded DNA, or with a negatively supercoiled DNA in the absence of Mg2+ so the DNA is not relaxed, leads to the cleavage of a DNA strand to give a 3’-hydroxyl group. The enzyme is found linked covalently to the 5’-phosphoryl group at the other side of the breakage point (19). Enzymatic and chemical hydrolysis of the covalent 49. J. J . Champoux, P N A S 74, 5328 (1977). 50. K . Kirkegaard and J. C. Wang, Nitcleic Acids. Res. 5, 3811 (1978). 51. L. F. Liu, R . E . Depew, and J . C. Wang, JMB 106, 439 (1976). S2. Y.-C. Tse and J. C. Wang, Cell 22, 269 ( 1980).

338

JAMES C. WANG

protein-DNA complex identifies a DNA phosphoryl group linked to the 04-position of a protein tyrosyl residue (53), thus the covalent proteinDNA bond appears to be a phosphotyrosine diester linkage. For type I enzymes from higher eukaryotes, similar treatments also lead to the formation of covalent protein-DNA complex (54,55). The eukaryotic enzymes differ from the bacterial enzymes in that the protein is found linked to a 3’-phosphoryl group. For the bacterial enzymes of E. coli DNA topoisomerase I type, the existence of the covalent enzyme-DNA intermediate has not been demonstrated directly without denaturation of the protein. The formation of a stable covalent complex between 4x174 gene A enzyme and DNA is well-documented (25).No such complex has been found, however, for the gene I1 enzyme of fd (6), in spite of the genetic and other biochemical evidence that these enzymes act in a rather similar fashion. Interestingly, purified gene I1 protein converts negatively supercoiled fd DNA to two forms, the covalently closed relaxed topoisomers and a nicked species without a covalently attached protein. The nicked species cannot be closed by the gene I1 enzyme (6). It thus appears that the gene I1 enzyme forms a rather labile covalent complex with DNA such that the covalent linkage can be displaced by a water hydroxyl group as well as by a DNA 3’-OH group. For the enzymes from rat liver (56) and calf thymus (55), the covalent enzyme-DNA complex has been identified without denaturation of the enzyme. Incubation of a single-stranded DNA with these enzymes, followed by zone sedimentation through a sucrose gradient, yields shorter DNA fragments with protein moieties linked to the 3’ ends. 2. Strand Crossing Over a . Topological Considerations. To convert a trefoil knot to a simple ring, or vice versa, one part of the continuous ring must cross over or pass through a transient break in another part of the ring. Similarly, the linking of two rings into a catenane must involve the transient breakage of one ring and the passage of the other through this break. The interconversion between knotted or intertwined species that are more complicated in topology requires more than one crossing-over event. For the relaxation of a supercoiled DNA, changing the linking number of the DNA by an amount n requires a net of n single-stranded passage events. This can be readily 53. Y.-C. Tse, K. Kirkegaard, and J . C . Wang, JBC 255, 5560 (1980). 54. J . J . Champoux, PNAS 74, 3800 (1977). 55. B. Prell and H. P. Vosberg, EJB 108, 389 (1980). 56. M. D. Bean and J . J. Champoux, Nucleic Acids Res. 8, 6129 (1980).

18. TYPE I DNA TOPOISOMERASES

339

deduced by using an algorithm relating strand crossovers to the linking number (57, 58). Conversely, one single-strand crossing-over event changes the linking number by 1. The relaxation of a supercoiled DNA can also be achieved by the repetitive passage of a double-stranded segment through a double-stranded transient break. In this case, the linking number changes in units of 2 for a simple ring (59), or a knot if the complexity of the knot is unaltered, provided that relative rotation of the segments straddling the transient break around their helical axis is forbidden. In agreement with the notion that type I topoisomerases introduce transient single-chain breaks, treatment of a DNA of a fixed linking number with such an enzyme generates a Boltzmann distribution of topoisomers that differ from the linking number of the original DNA by both odd and even linking numbers (60, 61). With type I1 enzymes, the linking numbers of the products differ from that of the reactant by even numbers only (for a review, see 18, 62). These topological considerations leave little doubt that type I enzymecatalyzed topoisomerizations involve the crossing of one single strand over another with the introduction of a transient single-stranded break. However, the mechanisms by which this occurs is not clear. b. Stepwise Versus Single-Hit Mechanism. Although changing the linking number by n requires a net of n single-strand crossing-over events, mechanistically there is no a priori limit on how many crossing-over events can occur between the generation of a transient break and the resealing of the break. The limiting case in one extreme is the single-hit mechanism, i.e., a supercoiled DNA becomes relaxed by a single breakage-rejoining cycle. At least under conditions such that the reaction is sufficiently slow, stepwise rather than single-hit relaxation of the supercoiled DNA is observed (11). c . State of the DNA Termini. As discussed in Section III,B,l, there is substantial evidence that the transient breakage of a DNA phosphodiester bond is accompanied by the covalent linkage of the DNA to the protein. For the bacterial enzymes, the 5' end of the interrupted DNA chain is the one that becomes linked to the protein; for the enzymes from higher eukaryotes, it is the 3' end. Especially when a single-stranded DNA is the 57. D. Glaubiger and J . E. Hearst, Biopolymers 5, 691 (1%7). 58. F. H. C. Crick, PNAS 73, 2639 (1976). 59. F. B. Fuller, PNAS 75, 3557 (1978). 60. D. E. Pulleyblank, M . Shure, D. Tang, J. Vinograd, and H. P. Vosberg,PNAS 72, 4280 (1975). 61. P. 0. Brown and N. R. Cozzarelli, Science 206, 1081 (1979). 62. N. R. Cozzarelli, Science 207, 953 (1980).

340

JAMES C. WANG

substrate, as in the knotting-unknotting reaction with single-stranded rings, a question of interest is the state of the other end of the interrupted chain that is not covalently linked to the enzyme. For enzymes of the E. coli DNA topoisomerase I type, it appears that the 3'-OH end of the interrupted single-stranded DNA is not free. No intermolecular rejoining, or intramolecular cross-joining (the linking of the 3' end of one break to the 5' end of another), has been detected (51). The possibility that the transient break occurs in a double-stranded region to keep the ends in close proximity, was suggested (51). With doublestranded DNA as the substrate, attempts to detect the presence of a 3'-OH during the course of DNA topoisomerization, by using exonuclease I11 or E. coli DNA polymerase I plus deoxynucleoside triphosphates, were unsuccessful (53).This is to be contrasted with the type I enzymes from higher eukaryotes, which appear to leave free 5'-OH termini with a singlestranded DNA substrate, as evidenced by the zone sedimentation experiment described in Section III,B,l, and by the accessibility of these ends to polynucleotide kinase (63). With purified 4x174 gene A protein, the cleavage of the negatively supercoiled 4X 174 DNA by this enzyme gives a 3'-OH group that is sensitive to nucleolytic attack by exonuclease I11 (4). Studies on DNA gyrase provide strong evidence that when a doublestranded DNA is broken by the enzyme, both termini resulting from the breakage are associated with the enzyme by covalent and noncovalent interactions (18, 62, 6 4 ) . This raises the interesting question whether for type I topoisomerases both termini resulting from the breakage of a single strand are also bridged by the enzyme: one end by covalent and perhaps also noncovalent interaction and the other end by noncovalent interaction (65). The evidence available is insufficient to draw a conclusion on this point, but it appears that there is a significant spread in the degree of freedom of the noncovalently linked ends generated by different type I topoisomerases . d . The Catenation of Double-Stranded Rings. The requirement of a preexisting single-chain scission for the type I enzyme-catalyzed catenation of a pair of double-stranded DNA rings (52) is well-documented experimentally but a mechanistical interpretation of this requirement is lack63. J. J. Champoux, in, "Mechanistic Studies of DNA Replication and Genetic Replication" (B. M. Alberts and C. F. Fox, eds.), ICN-UCLA Symp. Mol. Cellular Biol., Vol. 19. Academic Press, New York, 1980. 64. K. Kirkegaard and J. C. Wang, Cell 23, 721 (1981). 65. A. Morrison, P. 0. Brown, K. N . Kreuzer, R . Otter, S . P. Gerrard, and N. R . Cozzarelli, in "Mechanistic Studies of DNA Replication and Genetic Replication'' (B. M. Alberts and C. F. Fox, eds.), ICN-UCLA Symp. Mol. Cellular Biol., Vol. 19. Academic Press, New York, 1980.

18. TYPE I DNA TOPOISOMERASES

34 1

ing. It is unknown whether the transient break leading to catenation occurs directly opposite to the preexisting scission, and therefore generates the equivalent of a double-stranded break, or the transient break can be anywhere but only a segment of the single-strand directly opposite to the preexisting nick can pass through a transient break generated by the enzyme. IV.

Biological Roles

A. BACTERIALENZYMES Mutants defective in E. coli DNA topoisomerase I have been identified (66, 67). The structural gene, top, encoding the enzyme maps at 28 min on

the E. coli chromosome, flanked by CysB on one side and more distantly by Trp on the other. Strains deleted in top are viable. The growth rates of these strains are somewhat slower than that of the wild-type, and on agar plates they often give smaller colonies. The strains can support the growth of phages T7, T4, and A, and can be transformed with covalently closed plasmid DNAs. Thus the topological problem associated with the continuous reduction in the linking number between the parental strands of a covalently closed circular DNA during replication can be alleviated by enzymes other than topoisomerase I. The role of DNA gyrase in replication is now wellestablished (18, 62), and the possible existence of other topoisomerases in bacteria is yet to be explored. As one might anticipate from the in vitro function of the type I enzymes, plasmid DNAs isolated from top mutants have significantly higher negative superhelicity. This suggests that the superhelicity of DNA in bacteria is determined by a dynamic balance between gyrase, which supercoils the DNA, and topoisomerase I, which relaxes the DNA. Thus inhibition of gyrase leads to a decrease in negative superhelicity, and deficiency in the top gene product leads to an increase in negative superhelicity. The regulation of these enzymes with diametrically opposing activities, and the possibility of regulating certain physiological aspects of a large cluster of genes or the whole genome by controlling the superhelicity of a loop, or the whole genome, are interesting problems that require further studies. 66. R. Stemglans, S . Dinardo, J. C . Wang, Y. Nishimura, and Y. Hirota, in “Mechanistic Studies of DNA Replication and Genetic Replication” (B. M. Alberts and C. F. Fox, eds.), ICN-UCLA Symp. Mol. Cellular Biol., Vol. 19. Academic Press, New York, 1980. 67. R. Sternglanz, S . DiNardo, K. A. Voelkel, Y. Nishimura, Y. Hirota, K. Becherer, L. Zumstein, and J. C. Wang, PNAS 78, 2747 (1981).

342

JAMES C. WANG

Probably as a result of the increase in negative superhelicity in top mutants, the levels of transcription of several genes are found to be elevated. This is reminiscent of the phenotype of supX mutants that have been studied in detail in Salmonella typhimurium (68, 69). supX was originally identified as a mutation that suppresses phenotypically or promoter mutation in an operon involved in leucine biosynthesis, and it has been mapped near CysB in Salmonella typhimurium. It is most likely that supX is identical to top. The other known phenotypes of top mutants include an increase in sensitivity to inactivation by ultraviolet light or chemical treatment with methyl methanesulfonate. The plating efficiency with phage mu is also much lowered. Because of the pleiotropic effects of top mutations, however, the molecular interpretations of these phenotypical properties are uncertain.

B. VIRALENZYMES As discussed in Section I, the coliphage 4x174 gene A product, fd gene I1 product, and A int gene product differ from the archetype topoisomerases in several respects, but all three can catalyze the relaxation of negatively supercoiled DNAs under certain conditions. Similar activities coded by other bacteriophages are presumably common. The biological roles of these coliphage enzymes, the gene A and gene I1 proteins in replication, and the int protein in the integration and excision of the viral genome are well known, as previously described.

C. OTHERENZYMES

No mutants or specific inhibitors of type I topoisomerases other than those discussed in the sections above are known. Therefore the possible biological roles of these enzymes are usually speculations based on their in vifro activities, or are inferred from their association with cellular constituents or their levels during cellular processes. These aspects have been summarized in earlier reviews (1, 70, 7 1 ) . 68. 69. 70. 71.

F . Dubnau and P. Margolin, Mol. Gen. Genet. 117, 91 (1972). E. Dubnau. A. B. Lenny, and P. Margolin, Mol. Gen. Genet. 126, 191 (1973). J. J. Champoux,Annu. Rev. Biochem. 47,449 (1978). W. R. Bauer, Annu. Rev. Biophys. Bioeng. 7 , 287 (1978).

18. TYPE I DNA TOPOISOMERASES

V.

343

Research Applications

Topoisomerases are useful in preparing covalently closed DNAs of different degrees of superhelicity. The complete relaxation of a covalently closed DNA by a topoisomerase is formally equivalent to the combined actions of a nuclease that introduces single-chain scissions in the DNA and an excess of ligase that reseals all the breaks. The use of a eukaryotic activity is preferable because the bacterial topoisomerases exhibit a strong dependence on negative superhelicity and the rate slows down as the negative superhelicity of the DNA substrate decreases (11, 20). To ensure the complete relaxation of a DNA by a topoisomerase, two DNA samples with average linking numbers above and below the expected average linking number of the completely relaxed DNA can be used to see if they give identical products. To prepare negatively supercoiled DNA samples of different degrees of superhelicity, relaxation of the DNA can be carried out in the presence of different amounts of ethidium. In a medium containing 0.2 M monovalent cations typically employed for eukaryotic topoisomerases, binding of ethidium to DNA is reasonably strong and is nearly quantitative at a DNA concentration of 50 pglml or higher (72). Since each bound ethidium unwinds the DNA by 26" (73), 14 bound ethidium molecules are expected to reduce the linking number by 1 when the DNA is relaxed. Termination of the reaction can be achieved by the rapid mixing of the incubation mixture with phenol, which also extracts ethidium from the DNA in the aqueous phase. Positively supercoiled DNA samples can be prepared by relaxing the DNA with a eukaryotic type I topoisomerase in the presence of stoichiometric amounts of gyrase (74). ATP is omitted in this reaction to avoid the ATP-dependent negative supercoiling of DNA by gyrase. The difference in linking numbers of two identical DNA samples relaxed in the presence and absence of a bound agent has been used to provide information on the structural change of the DNA caused by the binding of the agent. Examples of such studies are the unwinding of the DNA helix by RNA polymerase (75, 76). The coiling of DNA around histones in a nucleosome (77, 7 8 ) ,and around DNA gyrase (74). 72. J.-B. LePecq and C. Paoletti, J M B 27, 87 (1967). 73. J . C. Wang, J M B 89, 783 (1974). 74. L . F. Liu and J . C. Wang, PNAS 75, 2098 (1978). 75. J.-M. Saucier and J . C. Wang, Nrrrure New Biol. 239, 167 (1972). 76. J. C. Wang, J . H. Jacobsen, and J.-M. Saucier, Nuclerc Acids Res. 4, 1225 (1977). 77. J . E . Germond, B. Hut, P. Oudet, M. Gross-Bellard, and P. Chambon, P N A S 72, 1843 (1975). 78. W. Keller, PNAS 72, 4876 (1975).

344

JAMES C. WANG

Another area where topoisomerases are used in DNA research is for the preparation of novel topological forms of DNA, which are in turn used as substrates for various reactions. Combinations of the different modes of topoisomerization reactions described in Section II1,A can yield a plethora of species, some of which have been discussed in a recent review (24) . ACKNOWLEDGMENT Much of the work on DNA topoisomerases from this laboratory has been supported by grants GM 24544 and its predecessor GM 14621 from the U.S. Public Health Service.

DNA Gyrase and Other Type XI Topoisomerases MARTIN GELLERT

I. Introduction and Perspectives . . . . . . . . . . . . . . . . . . 11. Definitions and General Methods . . . . . . . . . . . . . . . . .

111. DNAGyrase. . . . . . . . . . . . . . . . . . . . A. Purification and Properties . . . . . . . . . . . B . Reactions of DNA Gyrase . . . . . . . . . . . C. Mechanistic Models of the Action of DNA Gyrase D. A Second Gyrase-Related Topoisomerase . . . . IV. Other Type I1 Topoisomerases . . . . . . . . . . . V. Biological Role . . . . . . . . . . . . . . . . . . . A. DNA Replication . . . . . . . . . . . . . . . . B. Transcription . . . . . . . . . . . . . . . . . . C. DNA Recombination and Repair . . . . . . . . VI. Research Applications . . . . . . . . . . . . . . .

. . . . . .

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

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

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . .

. . . . .

. . . . . .

345 347 348 348 349 356 359 359 361 362 364 365 366

I. Introduction and Perspectives

DNA topoisomerases are enzymes that catalyze changes in the topological structure of DNA molecules. In the conceptually simplest process, the substrate is a closed circular duplex DNA and the reaction alters the number of times the two strands are wound around each other. This reaction interconverts relaxed and superhelical, or supercoiled, forms of DNA. Interconversion between circular duplex DNA and more complex knotted and catenated forms is also catalyzed by topoisomerases. 345 THE ENZYMES,Vol. XIV Copyright Q 1981 by Academic Press, Inc. All rights of reproduction in any form reserved ISBN 0-12-122714-6

346

MARTIN GELLERT

The basic chemical reaction carried out by topoisomerases is a cycle of breakage and rejoining of DNA chains, in the course of which the phosphodiester bond energy is conserved by the formation of a covalent enzyme-DNA intermediate. Within this general scheme, two classes of topoisomerases have been distinguished. The first class that was found (type I) is typified by DNA topoisomerase I of E. coli, also known as w protein, and by the so-called nicking-closing enzyme from rat liver [for reviews see (1-3)]. These enzymes operate by making transient singlestrand breaks in DNA. They are considered in Chapter 18 of this volume. This chapter concerns itself with type I1 topoisomerases, whose basic mode of operation involves the generation of transient double-strand breaks in DNA. Among type I1 enzymes, two subclasses must be distinguished. DNA gyrases primarily catalyze the conversion of relaxed DNA to a superhelical form, thus transforming the DNA structure to one of higher free energy at the expense of ATP hydrolysis. Other type I1 topoisomerases stimulate the reverse reaction, converting supercoiled DNA to more relaxed forms. Although this is the thermodynamically favored direction, these enzymes are also ATP-dependent (with some minor exceptions mentioned in Section IV). In contrast, no type I topoisomerases have been found to require an energy-supplying cofactor. There are thus two distinctions between type I and type I1 enzymes: single versus double-strand breakage and (with exceptions) the energy requirement of type I1 enzymes. DNA gyrase has been isolated only from prokaryotic sources; it is not clear whether such an activity exists in eukaryotic cells. The best-studied enzymes are those from Escherichia coli and Micrococcus luteus. These are discussed in detail in this chapter, and are covered in two recent reviews ( 3 , 4 ) .Similar activities from Bacillus subtilis and Pseudomonas aeruginosa have been identified but less fully characterized. Other type II topoisomerases have been isolated from both prokaryotic and eukaryotic sources. The enzymes purified from E. coli infected with bacteriophage T4 and from Drosophila embryos are discussed below. Similar activities have been detected in Xenopus laevis germinal vesicles (9,in HeLa cells and calf thymus ( 6 ) ,and in rat liver mitochondria (7). 1 . J. J. Champoux, Annu. Rev. Biocliem. 47, 449 (1978). 2. J. C. Wang and L. F. Liu,in “Molecular Genetics” (J. H . Taylor, ed.), Part 111, p. 65. Academic Press, New York, 1979. 3. M. Gellert, Annu. Rev. Biochem., in press (1981). 4. N. R. Cozzarelli, Science 207, 953 (1980). 5. M. I . Baldi, P. Benedetti, E. Mattoccia, and G . P. Tocchini-Valentini, Cell 20, 461 (1980). 6. L. F. Liu, in “Mechanistic Studies of DNA Replication and Genetic Recombination”

19. TYPE I1 TOPOISOMERASES

347

Supercoiling turns out to have important consequences for the biological activity of DNA. Thus DNA gyrase is an essential enzyme in prokaryotes, being required for DNA replication and having a profound influence on transcriptional and recombinational processes. T4 DNA topoisomerase is also involved in DNA replication. A brief discussion of the biological role of these enzymes is included in this chapter.

II.

Definitions and General Methods

The topology of DNA supercoiling has been extensively discussed (8-11 ) and only a brief outline is given here. We use the notation of Liu and Wang (12). Supercoiling of circular duplex DNA is measured in terms of the linking number a, which specifies the number of times the two strands

are intertwined. The linking number of a molecule is necessarily an integer. A convenient way to arrive at a is to consider a projection of the molecule in a plane, and to Count the excess of right-handed over lefthanded crossings of one strand over the other. As DNA is a right-handed helix, the linking number is normally positive. DNA in solution has a conformation close to the B form, with a pitch that has been measured as 10.4 base pairs per helical turn (13). Closed circular DNA with this pitch is under no torsional strain and is termed relaxed. The linking number (YO of a population of relaxed DNA molecules will be distributed over a narrow range of integer values (14, 15) centered around N/10.4, where N is the number of base pairs in the DNA.DNA with a mean linking number a less than is termed negatively supercoiled, or underwound; DNA with LY larger than (YO is positively supercoiled, or overwound. A convenient measure of supercoiling that is independent of molecular length is given by the specific linking difference [(a-a")/a"]. [A closely related quantity was orig(YO

(B. M . Alberts and C . F. Fox, eds.), ICN-UCLA Symp. Mol. Cellular Biol., Vol. 19.

Academic Press, New York, 1980. 7. F. J. Castora, G . G. Brown, and M . F. Simpson,in "The Organization and Expression of the Mitochondria1 Genome" (A. M. Kroon and C. Saccone, eds.), in press. Elsevier, Amsterdam, 1980. 8. F. B . Fuller, PNAS 68, 815 (1971). 9. F. B. Fuller, PNAS 75, 3557 (1978). 10. F. H . C. Crick, PNAS 73, 2639 (1976). 1 I . W. R . Bauer, Annu. Re,,. Biophys. Bioeng. 7, 287 (1978). 12. L. F. Liu and J. C. Wang, JBC 254, 11082 (1979). 13. J. C. Wang, PNAS 76, 200 (1979). 14. R. E. Depew and J. C. Wang, PNAS 72, 4275 (1975). 15. D. E. Pulleyblank, M. Shure, D. Tang, J. Vinograd, and H.-P. Vosberg, PNAS 72, 4280 (1975).

348

MARTIN GELLERT

inally called the superhelix density (1611. The excess free energy per unit length of a supercoiled DNA is proportional to (Aru/ao)z[reviewed in (11)J. The free energy can be quite considerable. For natural DNA species, whose specific linking difference usually falls in the range from -0.04 to -0.09, a change of linking number by one unit toward the relaxed value is favored by - 6 to - 14 kcal/mole. The binding of a protein that relieves the superhelical stress by unwinding the DNA is similarly favored. DNA topoisomerization reactions are conveniently assayed by electrophoresis in agarose gel (17). This method is simple to use and can handle many samples at one time. For DNA with molecular weight less than lo7,gel electrophoresis also has high resolving power, allowing resolution of topological isomers differing by 1 in linking number in a suitable range of ACU/CY". In this way detailed information about the distribution of topological isomers can be collected. Gel electrophoresis can also be used to identify knotted and catenated molecules. For rapid assay of topoisomerization reactions, measurement of the fluorescence of a bound dye has also been used (18, 19). Centrifugation methods that are sensitive to the topological state of the DNA (16, 20, 2 1 ) are generally more useful for analytical purposes than for enzyme assays. 111.

DNA Gyrare

A. PURIFICATION AND PROPERTIES We consider DNA gyrase first, and other type I1 topoisomerases subsequently. DNA gyrase from E. coli and M . lureus is made up of two protein subunits. The E. coli enzyme has been purified to near homogeneity either as the active complex (22) containing equimolar amounts of the two proteins, or as the two separate protein components, the gyrase A and B proteins, from which full activity is reconstituted by mixing (L?,24). Cells 16. W. Bauer and J. Vinograd, J M B 33, 141 (1968). 17. W. Keller, PNAS 72, 4876 (1975). 18. A. R . Morgan, J. S. Lee, D. E. Pulleyblank, N . L. Murray, and D. H . Evans, Nitcleic Acids Res. 7, 547 (1979). 19. A. R . Morgan, D. H. Evans, J. S . Lee, and D. E. Pulleyblank, Nircleic Acids Res. 7, 57 1 ( 1979). 20. L. V. Crawford and M. J. Waring, J M B 25, 23 (1967). 21. R . Radloff, W. Bauer, and J. Vinograd, PNAS 57, 1514 (1%7). 22. K . Mizuuchi, D. H . O'Dea, and M. GeUert, PNAS 75, 5960 (1978). 23. A. Sugino, C. L. Peebles, K . N . Kreuzer, and N . R. Cozzarelli, PNAS 74, 4767 (1977). 24. N . P. Higgins, C. L. Peebles, A. Sugino, and N . R . Cozzarelli,PNAS 75, 1773 (1978).

19. TYPE I1 TOPOISOMERASES

349

normally contain several times more gyrase A protein than B protein (25). Both proteins have also been purified separately from E. coli strains in which the corresponding structural genes have been cloned on plasmids that permit a significant overproduction of the gyrase subunits, particularly of the gyrase A protein ( 2 5 1 ) .The increased amount of the proteins in the cells allows the use of simplified purification procedures and leads to higher yields of the homogeneous proteins. In the author’s laboratory, purification from 40 g of packed cells of the two plasmid-carrying strains has yielded 100 rng of gyrase A protein or 5 mg of B protein. Chromatography on valine-Sepharose, with reverse salt gradient elution, has been found useful as a final step to remove minor contaminants. The E. coli gyrase A and B proteins have denatured molecular weights of 100,000-105,000 (22, 23) and 90,000-95,000 (22, 2 4 ) , respectively. The native gyrase A protein exists in solution as a dimer (23). Purification of DNA gyrase activity from M . luteus has also led to the isolation of two separate subunits, originally named a and p , whose molecular weights in denatured form are 115,000 and 97,000, respectively (26). It has been shown that a can complement E. coli gyrase B protein and /?can complement gyrase A protein to give the activities of DNA gyrase, thus identifying a and p as the functional equivalents of the E. coli gyrase A and B proteins, respectively (2.5). The M. luteus gyrase subunits have therefore been renamed A and B to be consistent with the naming of theE. coli proteins (27). Most properties of the E. coli andM. luteus enzymes are very similar, and the two enzymes are thus discussed together in the following sections. DNA gyrase from M . luteus has been shown to bind to DNA as an A2B, complex; by cross-linking experiments, at least some of the free enzyme exists in the same form (28). All the known activities of DNA gyrase require the presence of both subunits. A report of DNA relaxation by the E. coli gyrase A protein alone (23) was later shown to be incorrect (24). B.

REACTIONSOF DNA GYRASE

DNA gyrase carries out the following reactions: 1. In the presence of ATP, it catalyzes the negative supercoiling (decrease of the linking number) of closed circular duplex DNA (29). 25. P. 0 . Brown, C. L. Peebles, and N . R. Cozzarelli, PNAS 76, 61 10 (1979). 25a. K. Mizuuchi, M . Mizuuchi, M. H . O’Dea, and M. Gellert, in preparation. 26. L. F. Liu and J . C. Wang, PNAS 75, 2098 (1978). 27. Y.-C. Tse, K. Kirkegaard, and J . C. Wang, JBC 255, 5560 (1980). 28. L. Klevan and J. C. Wang, Biochemistry 19, 5229 (1980). 29. M . Gellert, K. Mizuuchi, M. H . O’Dea, and H. A. Nash, PNAS 73, 3872 (1976).

350

MARTIN GELLERT

2. In the absence of ATP, it stimulates the relaxation of superhelical DNA. 3. In the presence of the inhibitor oxolinic acid, double-strand breakage is produced at specific sites in DNA. 4. It carries out a DNA-dependent hydrolysis of ATP. 5 . The enzyme promotes the formation and resolution of catenated and knotted duplex DNA structures. These reactions are discussed in more detail in the following sections. The study of DNA gyrase has been greatly helped by the use of two families of specific inhibitors of the supercoiling reaction; one includes coumermycin Al and novobiocin; the other, nalidixic acid and oxolinic acid (23, 30, 31). These antibiotics are known to be inhibitors of DNA replication (32-34), suggesting immediately that DNA gyrase is involved in that process. Mutants of E, coli are known that are resistant to one or the other group of drugs; DNA gyrase isolated from a resistant strain is also drug resistant (23, 30, 31). The map locations of the genetic loci responsible for resistance have been determined (32, 34), and it has been shown that drug resistance of the cells and of DNA gyrase are closely linked genetically. Thus DNA gyrase appears to be the principal target of these antibiotics, with one subunit of the enzyme being associated with each genetic locus. The genetic loci have been renamed to reflect this identification (35).The locus for resistance to nalidixic and oxolinic acid at 48 minutes on the standard E, coli map, formerly nu&, is now known as gyrA. The locus for resistance to coumermycin Al and novobiocin [and clorobiocin (36)] at 82 min on the map, formerly cou, is now called gyrB. The gyrase A and B proteins are the products of the corresponding genes. The following sections summarize what is known about the reactions of DNA gyrase. 1. DNA Supercoiling

The supercoiling reaction catalyzed by DNA gyrase requires ATP and a divalent metal ion, usually Mg2+;Ca2+and Mn2+are less effective. As with many ATPases, K+ is stimulatory and Na+ is inhibitory. The pH optimum 30. M. Gellert, M. H. O’Dea,T. Itoh, and J. Tomizawa, PNAS 73, 4474 (1976). 31. M. Gellert, K. Mizuuchi, M. H. O’Dea, T. Itoh, and J. Tomizawa, PNAS 74, 4772 ( 1977). 32. M. J. Ryan, Biochemistry 15, 3769 (1976). 33. W.A. Goss, W. H. Deitz, and T. M. Cook,J . Bacteriol. 89, 1068 (1%5). 34. G. J. Bourguignon, M. Levitt, and R. Stemglanz,Antirnicrob. A g . Chemother. 4,479 (1973). 35. F. G . Hansen and K. von Meyenburg, Mol. Gen. Genef. 175, 135 (1979). 36. N. F. Fairweather, E. Orr, and I. B . Holland, J . Bacteriol. 142, 153 (1980).

19. TYPE I1 TOPOISOMERASES

35 1

is between 7 and 8. Under some conditions the reaction is stimulated severalfold by spermidine (29). The substrate DNA must be a closed circular duplex molecule: DNA gyrase cannot join DNA ends except as part of its coupled breakage-rejoining reaction. All species of closed circular duplex DNA tested have been effective substrates, though in principle one could anticipate differences based on the DNA's content of strong binding sites for the enzyme (see Section 111,B,5). The supercoiling reaction always reduces the linking number of the DNA, driving it toward negative supercoiling (29).This is the direction of supercoiling found in DNA isolated from cells. There is a limit to the extent of supercoiling that can be reached in the enzymatic reaction. With high concentrations of E. coli DNA gyrase, the specific linking difference reaches a value of about -0.10 (29). The smaller extent of supercoiling normally found in DNA species isolated from E. coli (specific linking difference of about -0.06) may well reflect the relaxing activity of competing topoisomerases in vivo. The supercoiling activity of DNA gyrase is catalytic; the linking number of all DNA molecules in the solution is changed by many units, even with a considerable molar excess of DNA over enzyme Q2, 24). The nucleotide requirement is quite strictly specific for ATP, with a K , of 0.3 mM (37). Deoxy ATP can substitute for ATP, but with a much higher K,; other nucleoside triphosphates are inactive (29). Supercoiling is inhibited by novobiocin and coumermycin (301,which inhibit competitively with ATP. Inhibition constants of 4 x loT9M for coumermycin and lo-* M for novobiocin have been reported (37). Inhibition is also obtained with oxolinic acid (23,31),the half-inhibitory concentration being 4 x M. Nalidixic acid, though an effective antibiotic, is a considerably weaker inhibitory than oxolinic acid and has not been widely used for in vifro experiments. The enzyme is also inhibited by N-ethylmaleimide. The ATP analogue (p, y-imido)ATP inhibits the ATP-driven supercoiling reaction. However, (p,y-imido)ATP itself is capable of supporting a limited supercoiling reaction. When a large amount of DNA gyrase is incubated with DNA and (p,y-imido)ATP, the DNA becomes supercoiled to an extent stoichiometric with the amount of enzyme added (37). The linking number is reduced by about 1.4 for every enzyme tetramer added; this value must be a lower limit because of the possibility that some of the enzyme is inactive. As the analogue is not noticeably hydrolyzed, its binding alone evidently triggers one cycle of supercoiling. Hydrolysis of ATP is apparently required to return the enzyme to the starting state for 37. A. Sugino, N . P. Higgins, P. 0. Brown, C . L. Peebles, and N. R. Cozzarelli,PNAS 75, 4838 (1978).

352

MARTIN GELLERT

another round of supercoiling (37). The decrease of the linking number is thus coupled to binding of the triphosphate rather than its hydrolysis. 2. DNA Relaxation

In the absence of ATP, DNA gyrase causes the relaxation of negatively supercoiled DNA (23, 31). This activity is much weaker than the supercoiling activity, with 20-40 times as much enzyme required to give a comparable rate (24, 38). Optimal reaction conditions are also somewhat different: the relaxation reaction prefers a higher Mg2+concentration. Relaxation is inhibited by oxolinic acid at concentrations similar to those needed to block supercoiling, but is not inhibited by coumermycin or novobiocin. Because relaxation requires only the breakage-rejoining reaction without the energy coupling aspect required for supercoiling, one can infer that oxolinic acid interferes with the breakage-rejoining step and that the gyrase A protein, the target of oxolinic acid, is involved primarily in that step. In the absence of nucleotide, DNA gyrase is unable to relax positively or to reequilibrate the linking number of a relaxed supercoiled DNA (3) DNA (39). The torsional stress of negative supercoiling must therefore be used in some unknown way to drive relaxation. Interestingly, DNA gyrase becomes capable of relaxing positively supercoiled DNA when (ply-imido)ATP is added (39). This reaction is in addition to the single cycle of supercoiling induced by the analogue. Apparently the binding of (/3,y-imido)ATPalters the functional state of the enzyme, and possibly its conformation. 3.

Oxolinic Acid-Dependent Breakage of DNA

If E. coli DNA gyrase is incubated with DNA and oxolinic acid, double-strand breaks appear in the DNA on subsequent addition of sodium dodecyl sulfate (23, 31). The DNA is broken into defined fragments, but not all sites of breakage are used with equal efficiency. Micrococcus luteus DNA gyrase behaves similarly, but breakage is less efficient and occurs more readily in this case if the complex is treated with alkali instead of detergent (27). It has been shown subsequently that the broken ends always have the same chemical structure: the two strands are broken four base pairs apart, with the 5' ends protruding and blocked by protein (39-41). It is the A protein that is attached, and the linkage is a phosphotyrosine bond, as is found with type I topoisomerases (27). 38. M. Gellert, L. M. Fisher, and M . H. O'Dea, PNAS 76, 6289 (1979). 39. M. Gellert, L. M. Fisher, H. Ohmori, M. H. O'Dea, and K. Mizuuchi, CSHSQB 45, in press (1980).

19. TYPE I1 TOPOISOMERASES

353

Strongly preferred cleavage sites are quite rare in DNA, occurring on the average less often than once in a thousand base pairs. This argues for a high degree of base sequence specificity in determining the sites. Until now, however, inspection of nucleotide sequences around several sites has not revealed any extensive homology. One study of sites cleaved by E. coli DNA gyrase in SV40, ColEI, and QX174 RFI DNA indicated that cleavage of one strand commonly occurred within a TpG sequence, and it was suggested that this dinucleotide was a general feature of gyrase cleavage sites (40). Even this limited homology was not found in an investigation of sites in the plasmids pBR322 and pNTl (a ColE1 derivative); breakage within a TpG dinucleotide occurred at only one out of five sites (39). Similarly, an analysis of sites cleaved by M. luteus DNA gyrase revealed one TpG cleavage among three sites (41). Because DNA gyrase binds to a region of DNA about 140 base pairs long (see Section 11I,B,5), the sequence that is recognized could be dispersed anywhere within that length. Further study should clarify the recognition specificity. The general pattern of fragmentation of a given DNA by DNA gyrase from E. coli, M. luteus, and B. subtilis is very similar, so the recognition sequences for all three enzymes may be common (25,42). The similarity has not yet been tested at the level of nucleotide sequence. The efficiency of DNA cleavage is increased somewhat by addition of ATP or (p,yimido)ATP, and the pattern of preference among cleavage sites is changed (37, 43). It was once proposed that ATP or its analogue caused the enzyme to move and cleave uniformly 400 base pairs away from its previous site (37). A reinvestigation has shown that no movement occurs; gyrase remains bound at its original locations, but the nucleotides enhance the cleavage at some of these sites (44). 4. Hydrolysis of ATP

DNA gyrase hydrolyzes ATP to ADP and phosphate. The ATPase activity is strongly stimulated by duplex DNA but only slightly by singlestrand DNA (22, 37, 45). Because some synthetic duplex DNA homopolymers and alternating copolymers are good cofactors, recognition of specific sites in DNA must not be important for this reaction (45). On the other hand, the tertiary structure of DNA clearly has an effect on 40. A . Morrison and N . R . Cozzarelli, CeIl 17, 175 (1979). 41. K . Kirkegaard and J . C. Wang, Cell 23, 721 (1981). 42. A. Sugino and K . F. Bott, J . Bncferiol. 141, 1331 (1980). 43. C . L. Peebles, N . P. Higgins, K. N . Kreuzer, A. Morrison, P. 0. Brown, A. Sugino, and N. R. Cozzarelli, CSHSQB 43, 41 (1978). 44. A. Morrison, N. P. Higgins, and N. R. Cozzarelli, JBC 255, 2211 (1980). 45. A. Sugino and N. R. Cozzarelli, JBC 255,6299 (1980).

354

MARTIN GELLERT

the ATPase activity. Linear, nicked circular, and relaxed closed circular duplex DNA stimulate the ATPase strongly, but highly negatively supercoiled DNA is less effective (22, 45). This result provides evidence for coupling between ATP hydrolysis and the reaction leading to DNA supercoiling. A degree of coupling is also suggested by the approximate stoichiometry between ATP hydrolyzed and superhelical turns introduced in the early stages of reaction with a relaxed DNA (45). Hydrolysis of ATP is inhibited by novobiocin and coumermycin, but not appreciably by oxolinic acid (22, 37). Novobiocin and coumermycin compete kinetically with ATP (37) even though they have no obvious structural similarity to it. Two additional experiments indicate that these drugs interfere with the binding of ATP to the enzyme. First, they prevent the alteration in the oxolinic acid-induced cleavage of DNA that is caused by ATP or (P,y-imido)ATP 07). Because this alteration does not require hydrolysis of the triphosphate, it is plausible that the drugs block its binding. Second, novobiocin prevents the covalent binding of the ATP analogue oATP (46) to the gyrase B protein (22). In a general sense, coumermycin and novobiocin thus interfere with the energy-coupling component of the supercoiling reaction, while oxolinic acid blocks the breakage-rejoining step.

5 . Binding of DNA Gyrase to DNA The complex between DNA and DNA gyrase has several features that are important to an understanding of the enzyme’s action. Binding of M. luteus DNA gyrase has been shown to protect stretches of DNA 140 base pairs long from digestion by staphylococcal nuclease (47); the E. coli enzyme protects regions of similar size against nuclease (39). The great length of the protected region immediately suggests that the DNA may be wrapped around the enzyme. More evidence for wrapping comes from studies of digestion of the gyrase-DNA complex with pancreatic DNase; this generates a series of DNA fragments whose lengths differ by 10 2 1 bases (47), as has been found in nucleosomes where the DNA is known to be bound around a protein core (48). Furthermore, if DNA gyrase is bound to a nicked circular DNA, which is then sealed with DNA ligase and deproteinized, the linking number of the DNA is increased from the value characteristic of relaxed DNA by about one unit for each gyrase tetramer bound (26). In other words, the 46. S. B . Easterbrook-Smith,J. C. Wallace, and D. B . Keech, EJB 62, 125 (1976). 47. L. F. Liu and J. C. Wang, Cell 15, 979 (1978). 48. J. D. McGhee and G . Felsenfeld,Annu. Rev. Biochem. 49, I l l 5 (1980).

19. TYPE I1 TOPOISOMERASES

355

DNA becomes positively supercoiled. The wrapping of the DNA must thus be asymmetric. More detailed studies have shown that DNA gyrase from either E. coli or M . luteus binds specifically to DNA fragments that are known to contain a preferred site for oxolinic acid-dependent cleavage (39, 41, 44). By the use of “footprinting” methods (49) it has been shown that the sequence protected by DNA gyrase, both in the absence and presence of oxolinic acid, is roughly centered on the cleavage site, with strongest protection found near that site (39, 41). Flanking regions, though still protected, show points of enhanced DNase I sensitivity at 10-1 1 base pair intervals. The DNase cutting points on complementary strands are staggered by two base pairs on the average, which is again reminiscent of the action of DNase I’on nucleosomal DNA (50). In one case, M. luteus gyrase was found to protect a region in which cleavage did not occur (41); thus some caution is necessary in identifying binding sites with cleavage sites. 6 . Formution and Resolution of Knotted and Catenated DNA

The spectrum of reactions performed by type I1 topoisomerases was recently expanded by the discovery that the ATP-dependent T4 DNA topoisomerase, discussed in more detail in Section IV, can introduce or remove knots in covalently closed duplex DNA (51, 52). Such reactions can only take place if there is a coupled breakage and rejoining of both strands of DNA. DNA gyrase has also been shown to remove knots from DNA (52, 53), form catenanes (54), and separate catenated rings (53, 54). While unknotting and separation of catenanes occur under normal gyrase reaction conditions, formation of catenanes requires high spermidine concentrations (54). Spermidine presumably aggregates the DNA and increases the local concentration of DNA chain segments. All these reactions require ATP (53, 54) and are inhibited by both novobiocin and oxolinic acid (29, 54); their requirements are thus more similar to the supercoiling than the relaxation reaction of DNA gyrase. Unlinking of catenanes and unknotting are slower reactions than supercoiling of the same amount of DNA. A possible explanation is that these more complex topological transitions require the juxtaposition of two distant 49. D. J. Galas and A. Schmitz, Nucleic Acids R e s . 5 , 3157 (1978). SO. B . Sollner-Webb and G. Felsenfeld, Cell 10, 537 (1977). 51. L. F. Liu. C.-C. Liu, and B . M. Alberts, Nature (Lo,7don) 281, 456 (1979). 9 . L. F. Liu, C.-C. Liu, and B . M . Alberts, Cell 19, 697 (1980). 53. K. Mizuuchi, L . M. Fisher, M. H. O’Dea, and M . Gellert, PNAS 77, 1847 (1980). 54. K. N . Kreuzer and N . R . Cozzarelli, Cell 20, 245 (1980).

356

MARTIN GELLERT

segments of DNA while supercoiling may involve DNA motions within a single small region. This point is discussed below in connection with models of DNA gyrase reactions (Section 111,C). 7. Trcinsient Double-Strund Breakage of DNA in the Supercoilitzg Reuction

The reactions described in the last section interconvert closed circular duplex DNA with knotted or catenated forms that are also covalently continuous. The pathway of these reactions must involve breakage and rejoining of both DNA strands; however, demonstration of a similar pathway for the supercoiling reaction has required a more indirect test. Fuller (9) pointed out that passage of one two-stranded segment of DNA through another, without rotation of the strands, would change the linking number of the DNA by two units. B y contrast, breakage of one strand, rotation around the unbroken strand by 360°, and rejoining would alter the linking number by one. For an experimental test, circular DNA with a single linking number must be isolated. Since DNA molecules are separated into distinct bands in gel electrophoresis according to their linking number, a single topological isomer can be recovered from such a gel. It is found that DNA gyrase does indeed change the linking number of a single topological isomer in steps of two, either during supercoiling or during relaxation (53, 55). Although more complicated mechanisms (e.g., singlestrand breakage with obligatory rotation by two turns before resealing) could also explain this result, the most economical interpretation that assembles the quantized linking number change with the knot and catenane interconversions is that all the topological changes induced by DNA gyrase involve double-strand breaks in DNA.

c.

MECHANISTIC MODELSOF THE ACTIONOF DNA GYRASE

The various experiments described above can be assembled into a general scheme that accounts for the reactions of DNA gyrase. Several models with varying levels of detail have been elaborated, differing in the mechanics of strand transport and in the means chosen to give the supercoiling reaction the correct polarity (52, 53, 55-57). One model (53) is 55. P. 0. Brown and N. R. Cozzarelli, Science 206, 1081 (19791. 56. P. Forterre, J . Theoret. B i d . 82, 255 (1980). 57. .I. C. Wang, R. I. Gumport, K. Javaherian, K. Kirkegaard, L. Klevan, M. L. Kotewicz, and Y.-C. Tse, in "Mechanistic Studies of DNA Replication and Genetic Recombination" ( B . M . Alberts and C. F. Fox, eds.), ICN-UCLA Symp. Mol. Cellular Biol., Vol. 19. Academic Press, New York, 1980.

19. TYPE I1 TOPOISOMERASES

357

discussed here, and the different features of some other versions are briefly mentioned. The model sketched in Fig. 1 begins with binding of the enzyme at the specific sites where oxolinic acid-stimulated cleavage is found; these sites are suggested also to be the locations of double-strand breakage in the supercoiling reaction. It is supposed that the DNA segment to be translocated through the break is contained within or very near the same protected region as this site, and becomes wrapped on the enzyme (Fig. 1B) with the positive sense of supercoiling found in the experiments of Liu and Wang (26). This wrapping provides the directionality of the reaction. Binding of ATP then leads to a coupled opening of the double-strand break and transport of the wrapped DNA segment through it, presumably accompanied by a conformation change in the enzyme; afterwards, the break is resealed. This process decreases the linking number of the DNA by two units, leading to the observed negative supercoiling (Fig. 1C). Because the nonhydrolyzed ATP analogue (P,y-imido)ATP produces one cycle of

t

FIG. 1. A model for DNA-gyrase-induced DNA supercoiling by means of transient double-strand breaks. The enzyme binds preferentially to certain sites on DNA and induces a positive superhelical wrapping of a local DNA region (B). ATP binding then leads to transport of the upper double helix through the lower, via a transient double-strand break ( C ) , with an accompanying conformational change in the enzyme. This reaction decreases the linking number of the DNA by two. Subsequent hydrolysis of ATP and release of the transported DNA segment prepares the system for another cycle of supercoiling (D-G). During relaxation of negatively supercoiled DNA (bottom series of drawings), the superhelical coiling causes a loop of DNA to fold over the enzyme with the opposite handedness to that used in the supercoiling reaction (H). Transport through a transient double-strand break causes an increase of linking number (relaxation) by two units (I-J). [Reprinted from (53)].

358

MARTIN GELLERT

supercoiling but then blocks further reaction @7), release of the translocated segment and return of the enzyme to its starting conformation may be coupled to hydrolysis of ATP (Fig. ID). Another cycle of reaction (Fig. lE, F, G) would then result in a further reduction of DNA linking number by two units. To retain the supercoiling during the reaction, the broken DNA ends must not be allowed to rotate relative to each other. The ends are presumably fixed by the enzyme, with the 5‘ ends attached by reversible covalent linkage to the gyrase A protein, this being the bonding found after oxolinic acid-promoted cleavage. Because DNA gyrase binds to DNA as a tetrameric A,B2 complex, one can envisage the possibility that the enzyme-DNA complex opens on one side to admit the transported DNA segment and then, in a coupled process, closes on that side and opens on the other to expel the segment. With such a two-step reaction, a rotation that would frustrate the supercoiling reaction could more easily be avoided. In this model, relaxation of negatively supercoiled DNA requires the DNA to loop over the enzyme in a right-handed sense, opposite to that used in the supercoiling reaction (Fig. 1H-J). The higher efficiency of supercoiling compared to relaxation is thought to be due partly to the use of positive superhelical wrapping and partly to the ability of ATP to speed up the cycling of the enzyme’s gating mechanism. Reactions involving knots and catenanes would also be expected to be less efficient because they require the transport of a more distant DNA segment, whose binding has to compete with the preferred local wrapping of a DNA loop. This model and those mentioned below still leave unspecified the complex maneuvers needed to translocate a DNA chain. Understanding the mechanics of this process will probably require the isolation of reaction intermediates. In a model termed “sign inversion” ( 5 3 , the enzyme is envisaged as binding two segments of DNA so they cross to generate a node of positive superhelical sense (sign); the sign of the node is thought to be determined either by local wrapping or by binding of a more remote DNA segment with a polarity determined by a sequence-specified orientation. Binding of ATP again causes passage of one duplex through a break in the other, with a reversal of the local sense of supercoiling and a net decrease of linking number by two units. Another model (57) proposes that the positive superhelical wrapping of DNA around the enzyme does not change during the reaction cycle, but that a second DNA loop is translocated from outside to inside this wrapped complex by the ATP-driven breakage and rejoining reaction, again with a linking number change of two units. ATP hydrolysis then leads to

19. TYPE I1 TOPOISOMERASES

359

release of the inside DNA segment and repetition of the cycle. In this model, the orientation of the strand crossings that lead to net decreases of linking number is not specified in detail, but could presumably be determined by a requirement for binding within a small DNA loop of preferred sense. D.

A SECOND GYRASE-RELATED TOPOISOMERASE

Extracts of E. coli contain a second protein, in addition to the gyrase B protein, that complements the gyrase A protein to produce a DNArelaxing activity. The protein, of molecular weight 50,000, has been purified to homogeneity (25, 38). Partial proteolytic digestion of this protein and the gyrase B protein generates many pairs of fragments with identical gel mobility, strongly suggesting that the 50,000 molecular weight species is a fragment of the gyrase B protein (38). The name topoisomerase 11’ has been suggested for the B fragment-A protein complex (25). The complex relaxes negatively superhelical DNA (25, 38) and, unlike DNA gyrase, can also relax positively superhelical DNA in the absence of ATP analogues (25).This activity alters linking numbers in steps of two (M. Gellert, unpublished results) and thus shares the basic mechanism of double-strand breakage and rejoining with DNA gyrase and other type I1 topoisomerases. DNA relaxation is inhibited by oxolinic acid but not by novobiocin. Oxolinic acid-dependent cleavage of DNA is seen, and produces the same pattern of fragments as with DNA gyrase, but in this case the pattern is not changed in the presence of ATP. Furthermore, the complex lacks the ATP-dependent supercoiling and DNA-dependent ATPase activities of DNA gyrase. The B fragment has apparently lost a portion of the B protein required for energy coupling. It is not known by what mechanism the fragment is produced; nor has it been shown to be a product of thegyrB gene. If the severalfold excess of B fragment over B protein found in extracts accurately reflects the intracellular situation, the B fragment may contribute appreciably to the total topoisomerase activity in the cells.

IV.

Other Type II Topoisomerases

Type I1 topoisomerase activities other than DNA gyrase have usually been detected first as ATP-dependent DNA-relaxing activities; all such enzymes tested have later been shown to work by a type I1 mechanism. The first enzyme of this class was purified from E. coli infected with

360

MARTIN GELLERT

phage T4 (51.58). The activity is absent from cells infected with T4 phages mutant in genes 39 or 52 (or possibly gene 60, though here the genetic situation is more complicated). Mutants in these genes have a “DNAdelay” phenotype; T4 DNA replication is appreciably reduced at early times after infection (59). The purified T4 DNA topoisomerase has protein subunits of molecular weight 63,000 and 52,000 identified as the products of genes 39 and 52, respectively (51,581. Some preparations also contain a 16,000 molecular weight protein, which may be the product of gene 60 (51). The enzyme requires ATP hydrolysis to relax DNA catalytically, though a limited DNA relaxation is found with substrate levels of enzyme and the ATP analogue ATP-yS [riboadenosine 5 ’ 4 4 3-thiotriphosphate)] (51). It has been suggested that this reaction may be analogous to the single cycle of DNA supercoiling produced by DNA gyrase in the presence of (p,y-imido)ATP. In the ATP-stimulated reaction the linking number of a DNA molecule is altered by one to two units for each molecule of ATP hydrolyzed (51). The enzyme can relax both positively and negatively supercoiled DNA. No inhibition by novobiocin, and little or no inhibition by oxolinic acid, has been found. Hydrolysis of ATP yields ADP and phosphate as products, and is strongly stimulated by DNA; single-stranded DNA is less effective than duplex DNA (51). T4 DNA topoisomerase also promotes more complex topological reactions. When a high concentration of the enzyme is incubated with circular duplex DNA in the absence of ATP, knotted molecules of varying degrees of complexity are formed (52). These knotted circles are efficiently restored to simple circular form by much lower (catalytic) amounts of enzyme in the presence of ATP. Both reactions occur with covalently closed DNA, and so an intermediate with a double-strand break must be involved. The relaxation of DNA by T4 DNA topoisomerase changes the linking number by multiples of two, showing that this reaction also involves transient double-strand breaks (52). An ATP-dependent topoisomerase similar in many ways to the T4 enzyme has been isolated from Drosophila embryos (60). This enzyme relaxes either negatively or positively supercoiled DNA, using a doublestrand breakage mechanism. It can separate catenated rings and resolve knotted DNA molecules; in the presence of a second protein component, it can also form large catenated networks from circular DNA. The second 58. G. L. Stetler, G . J. King, and W. M. Huang, PNAS 76, 3737 (1979). 59. R. H. Epstein, A. Bolle, C. H. Steinberg, E . Kellenberger, E. Boy de la Tour, R. Chevalley, R . S. Edgar, M . Susman, G. Denhardt, and A. Lielausis, CSHSQB 28, 375 (1963). 60. T.-S. Hsieh and D. Brutlag, Cell 21, 1 I5 (1980).

19. TYPE I1 TOPOISOMERASES

36 I

protein has not been extensively purified, but can be replaced by histone H1 (60).Histone H1 is known to condense DNA (61), and so may function by raising the local strand concentration to the point where catenation is favored. Novobiocin inhibits DNA relaxation by the Drosophilu enzyme, but only at a much higher concentration (above M ) than is needed to inhibit DNA gyrase (60). Novobiocin appears to compete with ATP, and so may operate by the same mechanism on this enzyme as on DNA gyrase. Nalidixic acid is not inhibitory. The role of ATP in the activity of these type I1 topoisomerases has not been entirely clarified. Unlike DNA gyrase, they do not require an energy source to drive the reaction because DNA relaxation is thermodynamically favored. One could envisage that ATP works by stimulating the action of a gating mechanism like the one previously described for DNA gyrase. It has been proposed that the other type I1 enzymes may work very similarly to DNA gyrase, except that they are unable to wrap DNA with a positive twist, and thus are unable to supercoil DNA (60). Without this orienting factor, DNA segments should more frequently pass through a transient break with a sense determined by the DNAs supercoiling, and so the DNA will be relaxed. It is worth emphasizing that there is no necessary correlation between a type I1 mechanism and an ATP requirement. For example, DNA relaxation by DNA gyrase and the related topoisomerase 11’ does not require ATP. V.

Biological Role

The most extensive information about the physiological function of topoisomerases concerns DNA gyrase. It is now clear that DNA gyrase is responsible for DNA supercoiling in E. coli, and that DNA in these cells is maintained under torsional stress. By inference, the same conclusion pertains to other prokaryotic cells. The existence of torsional strain in DNA has far-reaching consequences for its reactivity. The evidence about the intracellular state of DNA will be summarized first. The first indication that DNA in E. coli is actively supercoiled was provided by superinfection of A-lysogenic cells with phage A. Under these conditions of repression, the infecting DNA is circularized and normally is isolated in supercoiled form (62); no further development of the phage 61. M . W. Hsiang and R . D. Cole, PNAS 74, 4852 (1977). 62. E. T. Young, 11, and R . L. Sinsheimer, J M B 10, 562 (1964).

362

MARTIN GELLERT

occurs. When the cells are superinfected with X in the presence of coumermycin or oxolinic acid, however, the closed circular A DNA is reisolated with very little supercoiling (3U, 31). It was estimated that at most 15% of normal supercoiling is introduced by mechanisms independent of DNA gyrase (30). The chromosomal DNA of E. coli cells treated with coumermycin also loses a large proportion of its supercoiling; there is a 3- to 4-fold decrease within 10-20 minutes after addition of the drug (63). N o decrease is found in a gyrB mutant strain that contains a coumermycin-resistant DNA gyrase. Direct evidence that E. coli DNA is torsionally strained in the cellular environment has come from measurement of the rate of covalent photobinding of trimethylpsoralen to chromosomal DNA (64).This rate is increased when DNA is under the torsional stress of negative supercoiling. Chromosomal DNA in E. culi cells is found to react like purified superhelical DNA with a specific linking difference of -0.05 (64),a value close to that found for the isolated chromosome (65, 66). It should be emphasized that the state of DNA in eukaryotic cells is quite different. In all likelihood, intracellular eukaryotic DNA is not torsionally stressed. Although circular DNA isolated from eukaryotes is also negatively supercoiled, a combination of histone binding and relaxation by a topoisomerase seems to be sufficient to account for the degree of supercoiling found (67, 68). Binding of trimeth ylpsoralen also indicates that DNA in eukaryotic cells is free of torsional strain (64). A . DNA REPLICATION

Topoisomerases could have several functions in the replication of closed circular duplex DNA. 1 . Supercoiling induced by DNA gyrases could assist the binding of proteins required to start replication. 2. During chain elongation, a topoisomerase capable of relaxing positive superhelical turns could work as a swivel (69) to remove the positive supercoiling that would otherwise accumulate in the unreplicated part of the molecule. A DNA gyrase could satisfy this requirement and, in addi63. K . Drlica and M. Snyder, JMB 120, 145 (1978). 64. R. R. Sinden, J. 0. Carlson, and D. E. Pettijohn, Cell 21, 773 (1980). 65. A. Worcel and E. Burgi, JMB 71, 127 (1972). 66. D. E. Pettijohn and R. Hecht, CSHSQB 38, 31 (1973). 67. J. E. Germond, B. H i t , P. Oudet, M. Gross-Bellard, and P. Chambon,PNAS 72, 1843 (1975). 68. G. Felsenfeld, Nature (London) 271, 115 (1978). 69. J. Cairns, JMB 6, 208 (1963).

19. TYPE 11 TOPOISOMERASES

363

tion, could keep the DNA under negative superhelical strain and thus facilitate unwinding at the replication fork (29). 3. At the end of a round of replication, a topoisomerase could unlink catenated daughter DNA molecules. There is convincing evidence that DNA gyrase activity is essential for replication of circular duplex DNA in E. coli. Novobiocin and coumermycin block chromosomal replication in whole and toluenized cells and block replication of a variety of plasmid and viral DNA species in cell-free systems [see ( 3 )for a recent review]. In a few cases, the sensitive step has been identified. Thus, replication of phage 4x174 RFI DNA begins with a single-strand break made at a specific site by the viral cisA protein; this reaction requires a supercoiled DNA substrate. Oxolinic and nalidixic acid also block replication of E. coli DNA, but here the inhibition may not be due simply to interference with supercoiling [ reviewed in (-?)I. It takes a 100-fold higher concentration of oxolinic acid to block intracellular supercoiling of A DNA than to inhibit chromosomal DNA replication 0 1 ) . On the other hand, exposure of cells to low concentrations of oxolinic acid leads to formation of a gyrase-DNA complex that, when isolated and exposed to sodium dodecyl sulfate, produces breaks in the DNA (70). This reaction is similar to the cleavage of DNA by purified gyrase and oxolinic acid. It has been suggested that the oxolinic acid-gyrase-DNA complex may itself be inhibitory to DNA replication (71).

Temperature-sensitive mutations in the gyrA (72) and gyrB (7-?)loci are likewise known to block cellular DNA replication. T4 DNA topoisomerase is also involved in DNA replication. As previously mentioned, mutants defective in the enzyme make little DNA soon after infection. Because the mutants eventually produce a burst of phage, it has been proposed that the missing function is partly replaced by a host function (74).Growth of these phage mutants is much more sensitive to coumermycin and novobiocin than growth of wild-type T4, and so the complementing function may plausibly be DNA gyrase (7.5). Liu et 01. (51) have extended this line of argument by suggesting that T4 DNA topoisomerase, though so far known only to relax DNA, may have a gyrase-like supercoiling activity when bound at specific DNA sites form70. M . Snyder and K. Drlica, J M B 131, 287 (1979). 7 I . K. Drlica, E. Engle, and S. H. Manes, PNAS 77, 6879 ( 1980). 72. K . N . Kreuzer and N . R. Cozzarelli, J . Bircreriol. 140, 424 ( 1979). 73. E. Orr, N . F. Fairweather, I. B . Holland, and R. H . hitchard, Mol. Gen. Genet. 177, 103 (1979). 74. S. Mufti, and H . Bernstein, J . Virol. 14, 860 (1974). 75. D. McCarthy, J M B 127, 265 (1979).

364

MARTIN GELLERT

ing a loop around the T4 replication origin, and so may help to melt out the DNA sequence at the initiation point. No system has been identified in which inhibition of topoisomerase activity leads to abnormal accumulation of catenated DNA. Catenanes are seen as transient products of the replication of SV40 DNA (76) and of bacterial plasmids (77, 78). It is quite possible that they are later unlinked by the action of a topoisomerase; this aspect of DNA replication deserves more study. B . TRANSCRIPTION

It has been known for some time that transcription of negatively supercoiled DNA with E. coli RNA polymerase is generally enhanced over that of relaxed, nicked, or linear DNA (79-84). Because binding of RNA polymerase unwinds DNA by almost one turn (85), its binding to negatively supercoiled DNA is accompanied by a favorable free energy change possibly as large as -9 kcal/mole (for DNA with a specific linking difference of -0.06), corresponding to an additional factor of more than lo6 in equilibrium constant. However, more recent work has shown that the effects of supercoiling on RNA synthesis are not so simple and uniform. Transcription from various promoter sites has quite different sensitivities to inhibition of DNA gyrase. For instance, it has been found that the expression in E. coli cells of the maltose and lactose operons is reduced 5- to 10-fold by nalidixic acid, while that of the threonine, tryptophan, and tryptophanase loci is almost unchanged, and expression from the lactose repressor (krc 19) a n d h c UV5 mutant promoters is even increased (86). Novobiocin had similar but smaller effects. It was suggested that catabolite-repressible (CAP-protein dependent) promoters are particularly sensitive to the inhib76. R . Jaenisch and A . J . Levine, J M B 73, 199 (1973). 77. Y. M. Kupersztoch and D. R . Helinski, BBRC 54, 1451 (1973). 78. R. P. Novick, K . Smith, R. J . Sheehy, and E. Murphy, BBRC 54, 1460 (1973). 79. Y. Hayashi and M. Hayashi, Bioclicmistry 10, 4212 (1971). 80. P. Botchan, J. C. Wang, and H. Echols, PNAS 70, 3077 (1973). 81. J. C. Wang,JMB 87, 797 (1974). 82. P. Botchan,JMB 105, 161 (1976). 83. P. H. Seeburg, C. Nusslein, and H. Schaller, E J B 74, 107 (1977). i (D.Schlessinger, ed.), p. 163. Am. 84. A. Levine and W. D . Rupp, i ~ "Microbiology" SOC.Microbiol., Washington, D.C., 1978. 85. J.-M. Saucier and J. C. Wang, Naritre New B i d . 239. 167 (1972). 86. B . Sanzey, J . BucfcrioI. 138, 40 (1979).

19. TYPE I1 TOPOISOMERASES

365

itors. Differential sensitivity has been found in a number of other systems as well (8744)). With the E. coli phage N4, a particularly close connection is seen between transcription and DNA conformation. Purified N4 RNA polymerase transcribes only single-stranded DNA. Intracellular transcription of N4 is abolished by gyrase inhibitors or a gyrA temperaturesensitive mutation (91). Apparently the unwinding of DNA that is facilitated by negative supercoiling is essential to allow N4 RNA polymerase to function. The influence of supercoiling on transcription is seen most directly in a cell-free coupled transcription-translation system. For example, with relaxed ColE1 DNA as template, expression of the colicin gene is strongly depressed by gyrase inhibitors, while other plasmid-coded proteins are less affected. There is no inhibition if the DNA is added in supercoiled form; thus the need for gyrase function can be bypassed if the DNA is already supercoiled. With rifampicin added to block later initiation events, the growth of previously initiated RNA chains is unaffected by inhibition of DNA gyrase. Thus supercoiling affects only chain initiation (92).

c.

DNA RECOMBINATION A N D REPAIR

DNA gyrase was first detected as a factor needed to activate DNA for the site-specific phage A integrative recombination reaction in vifro (29). The requirement for gyrase activity can be bypassed by using supercoiled DNA (93, 94). Although reaction conditions can be changed to avoid an absolute requirement for supercoiling, recombination is always much faster with a supercoiled substrate (95). Intracellular X integrative recombination is inhibited by coumermycin, and so also seems to require supercoiling (%). Since only the molecule containing the phage attachment site litfP, and not the one with the bacterial attachment site atfB, needs t o be 87. A. Puga and 1. Tessman, JME 75, 99 (1973). 88. C. L. Smith, M . Kubo, and F. Imamoto, Nurure (London) 275, 420 (1978). 89. M. Kubo, Y. Kano, €3. Nakamura, A. Nagata, and F. Imamoto,Gene 7, 153 (1979). 90. H. Shuman and M . Schwartz, EERC 64, 204 (1975). 91. S . C. Falco, R. Zivin, and L. B. Rothman-Denes, P N A S 75, 3220 (1978). 92. H.-L. Yang, K. Heller, M . Gellert, and G . Zubay, PNAS 76, 3304 (1979). 93. H. A. Nash, K . Mizuuchi, R. A. Weisberg, and M . Gellert, in “DNA Insertion Elements, Plasmids, and Episomes” (A. 1. Bukhari, J. A . Shapiro, S. L. Adhya, eds.), p. 363. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York, 1977. 94. K. Mizuuchi, M. Gellert, and H . A. Nash, JME 121, 375 (1978). 95. T. J. Pollock and K. Abremski,JME 131, 651 (1979). 96. Y. Kikuchi and H. A . Nash, C S H S Q B 43, 1099 (1978).

366

MARTIN GELLERT

supercoiled, supercoiling presumably affects binding of a specific protein (97).

Supercoiling also stimulates repair of DNA damage in UV-irradiated E. coli, as well as some types of general recombination (98,99); both processes are inhibited by coumermycin. As supercoiling is known to aid the uptake of single strands into duplex DNA in vitro [reviewed in ( I O U ) ] , its effect on recombination in vivo may involve this step. A more direct participation of DNA gyrase in one type of recombination has also been found. Illegitimate recombination (i.e. between nonhomologous sequences) in a cell-free E. coli system is stimulated by irzhibition of DNA gyrase by oxolinic acid (201). The stimulation is abolished by coumermycin, and is not found in extracts of oxolinic acidresistant mutant cells. The authors propose that the oxolinic acid stimulation may involve cleavage by the gyrase-oxolinic acid combination, DNA strand crossing over by an exchange of gyrase subunits, and resealing of DNA. VI.

Research Applications

DNA gyrase can of course be used to prepare negatively supercoiled DNA. Although there are alternate methods of preparation, such as relaxation of DNA by a topoisomerase in the presence of an intercalating dye, the use of DNA gyrase is preferred for large and delicate DNA structures such as nucleoids (102) to avoid the manipulations needed to remove the dye. Binding of large amounts of DNA gyrase to nicked DNA, followed by sealing of the DNA, can also be used to prepare positively supercoiled DNA (26). As most DNA-binding proteins induce either negative supercoiling or none at all, DNA gyrase binding is now the best method for preparing DNA with a high degree of positive supercoiling in the absence of intercalating compounds (25, 39). All type I1 topoisomerases can be used to resolve knotted and catenated structures without permanently breaking the DNA. An example is the separation of individual DNA circles from the very complex catenated networks found in the kinetoplasts of trypanosomes (103). 97. K. Mizuuchi and M. Mizuuchi, C S H S Q B 43, 1111 (1978). 98. J. B. Hays and S. Boehmer, PNAS 75, 4125 (1978). 99. .I. L. Raina and A . W.Ravin, Mol. Gen. Genet. 176, 171 (1979). 100. C. M. Radding, Anriir. Rev. Biochem. 47, 847 (1978). 101. H. Ikeda, K. Moriya, and T. Matsumoto, CSHSQB 45, in press (1980). 102. A. Akrigg and P. R. Cook,Nucleic Acids Res. 8, 845 (1980). 103. J. C. Marini, K. G. Miller, and P. T. Englund, JBC 255, 4976 (1980).

DNA Unwinding Enzymes MALCOLM L. GEFTER

I. Introduction . , . . . , . , . , . , . , . , . . , . . Purification and Properties of the rep Protein . . . , . . Isolation and Characterization of Helicase 111 , . , . . . Mechanism of Action ofrep Protein and Helicase I11 . . . The Biological Role of Enzymes That Catalyze Unwinding ofDNA , . , , , . . , . . . . . , . . . , , . , . .

11. 111. IV. V.

1.

. .

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

367 368 370 371

,

,

.

372

.

.

,

.

Introduction

Proteins that act to promote separation of complimentary strands of DNA have been isolated from a variety of sources, including E. coli (I -8) and E. coli infected with bacteriophages T7 and T4 (9-12). Typically, P. J . Goldmark and S. Linn, JBC 247, 1849 (1972). M. Wright, S. Wickner, and J. Hurwitz, PNAS 70, 3120 (1973). S. Wickner and J. Hurwitz, PNAS 72, 4 (1975). E. Richet and A. Kohiyanna, JBC 251, 808 (1976). M. Abdel-Monem, H. Diirwald, and H. Hoffmann-Berling, EJB 65. 441 (1976). M. Abdel-Monem. M.-C. Chanal, and H. Hoffmann-Berling, EJB 79, 33 (1977). J. F. Scott and A. Kornberg, JBC 253, 3292 (1978). 8. G. M . Weinstock, K. McEntee, and I. R. Lehman, PNAS 76, 126 (1979). 9. E. Scherzinger, E . Landa, G . Morelli, D. Seitfert, and A. Yuki, EJB 72, 543 (1977). 10. R . Kolodner and C. C. Richardson, PNAS 74, 1525 (1977). 1 1 . C. C. Liu, R. L. Burke, U. Hibner, J. Barry, and B. Alberts,CSHSQB 43,469 (1979). 12. H. Krell, H. Durwald, and H. Hoffmann-Berling, EJB 93, 387 (1979).

1. 2. 3. 4. 5. 6. 7.

367 THE ENZYMES, Vol. XIV Copyright 0 1981 by Academic Press. tnc. All rights of reproduction in any form reserved ISBN 0-12-122714-6

368

MALCOM L. GEFTER

these proteins consume ATP during the process of strand separation and have been characterized as part of a larger class of enzymes that demonstrate DNA-dependent ATPase activity. The first two of these unwinding proteins to be described, helicase I and I1 (13, 141, apparently act by complexing in stoichiometric amounts with the product of the reaction, i.e., single-stranded DNA. The rep protein (15, 161, the recBC enzyme (17), the T7-encoded gene 4 protein ( l o ) , the TCencoded 44/61 and 45 complex, and the E. coli helicase 111 protein act to promote strand separation catalytically, i.e., one enzyme molecule can catalyze the melting of thousands of DNA base pairs. This chapter is concerned with the latter class of enzymes. Specifically it focuses on the rep and helicase 111 proteins of E. coli, two enzymes studied in the author's laboratory. As will become apparent, these two enzymes are quite similar in the mechanism of their action and may provide a suitable model to explain catalytic DNA strand separation in general. Sorting out the biochemistry of DNA replication has required the development of the genetics of the system as well. Historically these disciplines have not developed synchronously. The discovery of the requirement for therep gene function in the replication of bacteriophage +X 174 by Denhardt et 01. (18) preceded the elucidation of the biochemistry of the gene product by more than 10 years. The detailed biochemistry of the replication of small DNA bactriophages carried out primarily in the laboratories of A. Kornberg, J. Hurwitz, and s. Wickner over the past several years was essential for our understanding of the rep protein function. Similarly, the pioneering work of B. Alberts and our laboratory on DNA binding proteins (helix-destabilizing proteins) was essential to the development of an understanding of the process of catalytic unwinding of DNA. II. Purification and Properties of the rep Protein

The isolation and purification of the rep protein was based upon its requirement for the replication in vitro of the replicative form of bac13. M. Abdel-Monem, H. F. Lauppe, J. Kartenbeck, H. Durwald, and H. HoffmannBerling, JMB 110, 667 (1977). 14. M . Abdel-Monem, H. Durwald, and H. Hoffmann-Berling, EJB 79, 39 (1977). 15. A. Kornberg, J. F. Scott, and L. L. Bertsch, JBC 253, 3298 (1978). 16. G. T. Yarranton and M. L. Gefter, PNAS 76, 1658 (1979). 17. V. Mackay and S. Linn,JBC 251, 3716 (1976). 18. D. T. Denhardt, M. Iwaya, and L. L . Larison, Virology 49, 486 (1972).

20. DNA UNWINDING ENZYMES

369

teriophage 4 X 174 (19, 20). The components of that system include DNA polymerase 111 and its ancillary factors, the product of cistron A of 4 x 1 7 4 , E. coli DNA binding protein, and ATP. It was shown initially that the role of the rep protein in that overall reaction was to separate the complimentary strands of the 4 x 1 7 4 replicative form DNA (21). This process could be uncoupled from DNA synthesis because the &A protein cut 4 x 1 7 4 DNA served as a substrate for unwinding provided that ATP, DNA binding protein, and rep protein were present. In fact, any DNA could be unwound by the action of rep protein, DNA binding protein, and ATP provided that an initiator region was present on the otherwise duplex DNA. This initiator region could be a “gap” in duplex DNA or a break, which could be introduced into duplex DNA by the cisA protein (22). ATP is consumed during the process of strand separation to the extent that 2 moles of ATP are converted to ADP and Pi for each mole of base-pair unwound (22, 23). The purified enzyme also displays an ATPase activity dependent upon single-stranded DNA as a cofactor . (The relationship of these two activities is described in Section 111). The purification of the rep protein was achieved using conventional procedures (23). A significant advantage in the purification was realized by using a rep protein overproducing strain of E. coli as starting material. The protein was purified by ammonium sulfate precipitations and column chromatographic steps on supports of DEAE-cellulose, Bio-Rex 70, and DNA cellulose. The final preparation is nearly homogeneous and is composed of a single polypeptide chain of molecular weight 65,000. The turnover number of the enzyme as measured by ATP consumption during strand separation is 6,800 moleculeslmin (23 ). As a DNA-dependent ATPase, the enzyme requires single-stranded DNA (limit of dT,), ATP, and Mg2+.The reaction is inhibited by thiol reagents and, most interestingly, by DNA binding protein (23, 24; See also 26). It is curious to note that for catalyzing the unwinding of DNA, DNA binding protein is required; however, it is an inhibitor of the enzyme when the single-stranded DNA-dependent ATPase activity is measured. This latter result is due to the fact that both proteins compete for binding to single-stranded DNA substrate (25). At the site of a replication fork (i.e., the junction between single- and double-stranded DNA) the rep protein is 19. 20. 21. 22. 23. 24. 25.

C. Sumida-Yasumoto, A. Yudelevich, and J . Hurwitz, PNAS 73, 1887 (1976). S. Eisenberg, J . F. Scott, and A. Kornberg, PNAS 73, 3151 (1976). J. F. Scott, S. Eisenberg, L. L. Bertsch, and A. Kornberg, PNAS 74, 193 (1977). J . Yarranton and M. L.Gefter, PNAS 76, 1658 (1979). J . F. Scott and A. Kornberg, JBC 253, 3292 (1978). A. Kornberg, J. F. Scott, and L. L. Bertsch, JBC 252, 3298 (1978). J. Yarranton, R . H. Das, and M. L. Gefter, JBC 254, 12002 (1979).

370

MALCOM L. GEFTER

not inhibited by DNA binding protein and, in fact, requires DNA binding protein for its continued action, measured by its ATPase activity or its unwinding activity ( 2 4 , 2 6 ) .As described in more detail in Section IV, the latter result is presumable due to the requirement of the DNA binding protein to maintain the strands of DNA in single-stranded form as they are generated in a step-wise fashion at the fork by the action of therep protein (26). During the course of analyzing the action of the rep protein on duplex DNA, it was demonstrated that in order for the protein to catalyze unwinding, a “leader” or single-stranded protruding region was required. In keeping with theoretical considerations of a step-wise action (17) it was demonstrated that the rep protein migrates along this leader segment unidirectionally and then invades the duplex DNA. The polarity of movement along the leader is 3’ + 5’ only (11). This is the polarity expected for the rep protein considering the requirement for generating a “rolling circle” structure during qhX RFI replication. 111.

Isolation and Characterization of Helicase 111

In considering the facts that the rep mutation was not lethal to E. coli and that catalytic strand separation appeared to be an essential function for DNA replication, a search was made for arep-like activity in cells that harbored the rep mutation. The rationale for purification was that the protein should be a single-stranded DNA-dependent ATPase (a far more convenient assay than strand separation) inhibitable by DNA binding protein. Accordingly, a protein was isolated by a series of routine purification procedures that satisfied these criteria. Purification to greater than 90% purity was achieved by ammonium sulfate precipitation and chromatography on phosphocellulose, DNA-cellulose, and DEAE-cellulose (26). The protein is composed of two polypeptide chains each of molecular weight 20,000 (presumably they are identical), its specific activity is comparable to the rep protein. The DNA-dependent ATPase activity requires single-stranded DNA (minimum of dT,,,), and Mg2+.The product of the reaction is inhibited by thiol reagents and the nonhydrolyzable ATP analogue imido ATP (AMP-PNP). Just as with rep, the ATPase is abolished in the presence of DNA binding protein (27). The enzyme also catalyzes strand separation of duplex DNA (provided the DNA contains a leader of single-stranded DNA), which requires ATP 26. M. Duguet, G. Yarranton, and M. Gefter, CSHSQB 43, 335 (1979). 27. J. Yarranton, R. H.Das, and M. L. Gefter, JBC 254, 1197 (1979).

20. DNA UNWINDING ENZYMES

37 1

and DNA binding protein. In contrast to rep, however, helicase I11 (27u) migrates unidirectionally on the leader segment in 5' --* 3' direction (25). An additional activity observed for helicase 111 is its ability to promote a DNA-independent ADP-ATP exchange reaction. As described below, this latter activity suggests the existence of an enzyme-phosphate complex that has implications for the proposed mechanism of DNA unwinding (29).

IV.

Mechanism of Action of rep Protein and Helicase 111

The suggested mechanism of action of rep protein and helicase I11 is based upon several observations, which may be summarized as follows: The DNA binding protein binds to the phosphodiester backbone of DNA and competes for rep and helicase I11 binding (25). This suggests that the unwinding enzymes may also bind to the phosphate backbone. The DNA binding protein binds only to single-stranded DNA, not to doublestranded DNA. Thus the conformation of the backbone can be used to discriminate between these two forms of DNA. Alternatively, a protein binding to the backbone of double-stranded DNA and then altering its conformation such that the phosphates to which it is bound are no longer compatible with a stable DNA duplex could lead to melting of the duplex. The energy for that reaction may be derived from the hydrolysis of ATP. It has been demonstrated that the properties of helicase 111 are in keeping with such a mechanism (29). The free enzyme binds to both single- and double-stranded DNA. On binding of ATP (which is converted into bound phosphate and free ADP) the enzyme looses its ability to bind to doublestranded DNA but retains its ability to bind to single-stranded DNA. The enzyme in the presence of single-stranded DNA is an ATPase and thus releases the bound phosphate and regains its ability to bind doublestranded DNA. Thus the enzyme sitting at a replication fork (i.e., the junction of single- and double-stranded DNA) can alternatively, at the expense of ATP hydrolysis, bind to duplex DNA, destabilize it, convert it to single-stranded DNA, and then repeat the cycle. Being endowed with the property of unidirectional movement along DNA, as is the case for 27a. The enzyme was named helicase I11 because helicases I and II were previously described and they too act to unwind DNA undirectionally (5' + 3' on the leader strand). In studies carried out independently, the rep protein was also given the name helicase 111 (28). In considering the polarity of movement, we propose that the rep protein not be renamed helicase 111. 28. S. Takashi, C . Hours, A. Chu, and D. T. Denhardt, Can. J . Chem. 57, 855 (1979). 29. R . H. Das, G . T. Yarranton, and M. L. Gefter, JBC 17, 8069 (1980).

372

MALCOM L. GEFTER

these enzymes, duplex DNA can be unwound in a step-wise fashion. The requirement for DNA unwinding and unidirectional movement may explain why this process consumes a great deal of energy, i.e., 2 molecules of ATP per base-pair unwound. A more detailed description of the mechanism has been published (39).

V.

The Biological Role of Enzymes That Catalyze Unwinding of DNA

To date there are no reports that demonstrate absolutely the role in normal cells for helicases orrep protein. One report suggests that helicase I11 may be essential for observing DNA synthesis in vitro (39). This work obviously must be extended. The analysis of the action of these proteins in vitro suggests a possibility as to why no mutants that dramatically affect DNA replication have yet appeared that are defective in rep or helicases. These proteins can catalyze overlapping or parallel functions. Given the polarities of movement [i.e.,rep migrating in a 3' + 5' direction and helicase (30) migrating in a 5' + 3' direction] suggests that they may sit on opposite strands of DNA at the replication fork and coordinately catalyze unidirectional unwinding. The properties of the rep mutants are compatible with this hypothesis. Chromosome replication is only mildly affected by the loss ofrep function, but 6 x 1 7 4 rolling circle (unidirectional) replication is abolished. It can be suggested that mutations that affect DNA unwinding can be sought in a rep-defective background, i.e., one that makes replication dependent upon helicase action. It is anticipated that catalytic unwinding of parental DNA will be shown to be essential for the replication of DNA.

30. It is not possible to assign different functions for helicases 1, 11, and 111; they are treated as one here.

Single-Stranded DNA Bin ding Proteins STEPHEN C . KOWALCZYKOWSKI PETER H . VON HIPPEL

DAVID G . BEAR

1. Introduction and Overview . . . . . . . . . . . . . . . . . II . Theoretical and Experimental Considerations . . . . . . . . . A . The DNA Substrate . . . . . . . . . . . . . . . . . . . B . Binding Parameters and Interactions . . . . . . . . . . . . . C . Methods for Monitoring Binding . . . . . . . . . . . . . . . 111 . Protein Isolation and Purification: Procedures and Strategies . . . . A . Production of Cellular Extracts . . . . . . . . . . . . . . . . B . DNA-Affinity Chromatography . . . . . . . . . . . . . . . . C . Additional Purification Procedures . . . . . . . . . . . . . . D. Assays and Criteria of Purity . . . . . . . . . . . . . . . . . IV . Structure, Properties and Nucleic Acid Binding Interactions of Several Single-Stranded DNA Binding Proteins . . . . . . . . . . A . Bacteriophage T4-Coded Gene 32 Protein . . . . . . . . . . . B . Filamentous Phage Gene 5 Protein . . . . . . . . . . . . . . C . Esclwriclrin coli Single-Stranded DNA Binding Protein . . . . . D . Bacteriophage T7-Coded Single-Stranded DNA Binding Protein . E . Eukaryotic Single-Stranded DNA Binding Proteins . . . . . . . V . DNA Binding Proteins as Research Tools . . . . . . . . . . . . . A . Electron Microscopy . . . . . . . . . . . . . . . . . . . . B . Biochemical Assays . . . . . . . . . . . . . . . . . . . . VI . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . .

374 375 375 377 381 383 384 386 387 388 388 389 412 423 43 1 432 441 441 441 442

373 THE ENZYMES. Val . XIV Copyright 0 1981 by Academic Press. Inc . All rights of reproduction in any form reserved ISBN 0-12-122714-6

374 1.

S. KOWALCZYKOWSKI, D. BEAR, A N D P. VON HIPPEL

Introduction and Overview

In its broadest context, the title of this chapter could easily encompass the subject matter of many of the chapters of this and the companion volume. As we define it the topic is more circumscribed, and our discussion is restricted to proteins that ( 1) bind preferentially and relatively nonspecifically to single-stranded DNA, and (2) have no other (enzymatic) activity. It is becoming apparent that proteins that fit the above definition are essential to many physiological functions, including replication, recombination, and repair, in a host of organisms ranging from bacteriophage to higher eukaryotes. Furthermore, despite the apparent simplicity of their central function, it turns out that many of these proteins have subtle and sophisticated features that not only permit them to participate in these processes, but also to play important roles in controlling and directing them. In this sense current single-stranded DNA binding proteins may represent systems that have evolved substantially beyond primitive precursors, which may only have been capable of direct and uncontrolled nucleic acid binding. Therefore in this chapter we focus not only on the DNA binding properties of the proteins, but also (to the extent information permits) on molecular aspects of their involvement in entire systems of DNA replication, recombination, and repair. To this end we attempt to bring out the structural, thermodynamic, and functional principles that unify the singlestranded DNA binding proteins and serve to define them as a class. After a brief summary of the nature and measurement of DNA-protein interactions (Section 11), and a presentation of general purification strategies for single-stranded DNA binding proteins (Section III), we discuss in some detail the properties of representative members of this class (Section IV). These “case histories” deal with proteins isolated from both prokaryotic and eukaryotic cells, and although significant differences in some properties exist, we hope the reader will be more impressed by the many apparent underlying similarities of the proteins in terms of their functions, and of the structures that have evolved in support of the functions. In Section V we review briefly certain ways in which the singlestranded DNA binding proteins have been exploited as tools in molecular biological research, particularly in the electron microscopy of biological macromolecules, and in certain biochemical assays. In conclusion (Section VI) we attempt to draw these threads together to present a series of generalizations to help involved workers develop an overview of the field, and also to set up certain experimental criteria that

21. DNA BINDING PROTEINS

375

might guide and facilitate the characterization and functional interpretation of the properties of proteins that will be examined in the future. We also hope that a better understanding of the structure and properties of the single-stranded DNA binding proteins will help to provide further insight into the mechanistic details of the physiological systems of which these proteins form an integral part. The nomenclature of the single-stranded DNA binding proteins is still in a relatively unsatisfactory state. This reflects, in part, the fact that the role(s) of these proteins in the various integrated physiological systems in which they are involved is still far from completely understood. Most names used in the past reflect the (at least potential) ability of these proteins to shift nucleic acid helix-coil conformational equilibria by binding preferentially to the single-stranded “coil” form of DNA or RNA. For this reason these ligands have been called melting proteins, unwinding proteins, helix-destabilizing proteins (HDPs), and single-strand binding (SSB) proteins. An effort initiated by Bruce Alberts ( I ) to arrive at a consensus on the use of the term “helix-destabilizing protein” as a generic name for this class has been only partially successful. In the absence of a final consensus, and to avoid further confusion, we generally use (or, at least, indicate) the names employed in the original articles in presenting detailed descriptions of individual protein systems in Section IV. In general discussions we use the terms single-stranded (DNA) binding protein (SSBP) and helix-destabilizing protein (HDP) interchangeably.

II. Theoretical and Experimental Considerations (2)

A.

THEDNA SUBSTRATE

Single-stranded nucleic acid sequences comprise the primary binding substrates for the proteins discussed here, and thus we consider briefly the main structural features of these target lattices. First, however, we must recall that almost all of the nucleic acid components of the cell are double-stranded in nature, either as double-helical DNA or as base-paired 1. B. M. Alberts and R . Sternglanz, Nriiure ( L o ~ h 7269, ) 655 (1977). 2. We note that such binding interactions (with either single- or double-stranded nucleic acids) comprise a central component of the general interaction of virtually all genome regulatory proteins (including repressors, polymerases, nucleases, gyrases, helicases, etc.) with nucleic acid lattices. Thus it is important to note that the theoretical and experimental considerations outlined in this section in terms of HDPs are general, and form, in exactly the same terms, a part of the description of all nucleic acid binding proteins.

376

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

regions of secondary structure in the various cellular RNAs ( 3 ) . These double-stranded structures enter our considerations in two ways: (1) They can compete directly (depending on relative binding affinities) for DNA binding proteins; and (2) they can serve (via local fluctuation-driven double-helix e coil transconformation reactions) as additional sources of single-stranded binding sites. The first problem is generally obviated by the fact that these proteins bind sufficiently weakly to base-paired sequences to prevent double-stranded nucleic acids from serving as effective binding competitors at physiological levels of protein and nucleic acid, and at physiological salt concentrations. The second problem is central, and has both equilibrium and kinetic aspects. All duplex nucleic acid sequences are, of course, in potential equilibrium with their single-stranded constituents. Thus, from an equilibrium perspective, these latter forms will bind protein if the (favorable) binding free energy made available on forming such protein-single-stranded nucleic acid complexes exceeds the conformational free energy lost on disrupting the otherwise stable duplex sequences. The point at which complex formation is favored for each sequence depends on the length of the particular double-stranded segment, its structure (i.e., does it contain single-stranded loops or mispaired bases?), its base composition, the solvent environment (increasing salt concentrations generally stabilize the nucleic acid duplex and destabilize protein-nucleic acid complexes), the binding constant, and the concentration of free protein. (For a further discussion of these aspects, see Refs. 4, 5 . ) In addition to considering the equilibrium situation, one must also ask whether conformational equilibrium is actually reached in a finite time under various experimental conditions. For example, results with gene 32 protein (4, 6, 7) suggest that this protein is kinetically blocked from invading double-stranded DNA (and probably RNA) sequences of significant length, even under conditions where single-stranded nucleic acid binding is favored at equilibrium. Thus, for at least some DNA binding proteins, "opening" fluctuations of the DNA duplex may not be of sufficient size or 3. RNA lattices are included in this discussion because they can serve as binding sites for autogenous regulatory interactions (at least for phage T4-coded gene 32 protein) and because they comprise an appreciable fraction of the nucleic acid composition of the cell, and thus may compete with the primary single-stranded DNA target sequences for free singlestranded binding protein. 4. D. E. Jensen, R. C. Kelly, and P. H. von Hippel,JBC 251, 7215 (1976). 5 . J . W. Newport, N. Lonberg, S. C. Kowalczykowski, and P. H. von Hippe1,JMB 145, 105 (1981).

6. B. M. Alberts and L. Frey, Nurure (London) 227, 1313 (1970). 7. J. W. Newport, Ph.D. Thesis, University of Oregon, Eugene, Oregon, 1980.

21. DNA BINDING PROTEINS

377

frequency to nucleate the transition to the single-stranded-proteincomplexed form. As discussed in Section IV,A, such kinetic blocks may serve as important physiological control elements. Other single-stranded DNA binding proteins can take advantage of thermally driven conformational fluctuations of the DNA double-helix to melt such structures to equilibrium. However we note that in at least some replication complexes (e.g., T4, E. coli) other protein components are involved in opening the replication fork and making binding sites available for the subsequent (and passive?) binding of the S S B P . The actual interaction of a DNA binding protein with a single-stranded nucleic acid lattice clearly involves the participation of many functional groups. A substantial component of the binding free energy is generally electrostatic in nature, involving interaction of DNA phosphates with appropriately placed, positively charged (Arg, Lys, His) amino acid residues of the protein. In addition, hydrogen bonding, dipolar, and hydrophobic (solvent-driven) interactions between functional groups of the protein and those located on the various components of the nucleic acid may provide some additional stability to the binding complex, as well as varying degrees of binding specificity. (For a recent discussion of these aspects see Ref. 8.1 B.

BINDINGPARAMETERS A N D INTERACTIONS

I . Tl?ertnod.yttamicso f Bindirig A description of the nonsequence-specific binding of proteins to nucleic acid lattices involves thermodynamic considerations beyond those needed to characterize the simple binding of ligands to independent binding sites. On binding to a nucleic acid lattice, a DNA binding protein generally covers (i.e., makes unavailable to another incoming protein) more than one nucleotide residue. If binding is nonspecific each nucleotide residue of the lattice can then be considered, in principle, to comprise the beginning of a potential protein-binding site n residues in length, where n (in units of nucleotide residues) represents the protein site size. When binding to an otherwise “naked” lattice the protein not only occludes the lattice site to which it actually binds, but it also partially covers 3n minus two other potential binding sites. As more protein is bound, however, the number of potential binding sites occluded per binding event decreases, and the number of binding sites remaining unoccupied at any particular protein 8. P. H . von Hippel, i f i “Biological Regulation and Development” (R.F. Goldberger, ed.), Vol. I . p. 279. Plenum, New York, 1979.

378

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

binding density is not a linear function of the number of ligands bound. As a consequence of this “overlap” type of binding, curved Scatchard plots are obtained even for the noncooperative binding of homogeneous protein ligands (9, lo). General approaches for extracting intrinsic ligand-lattice binding constants (K), as well as site size parameters ( n ) , from binding data on such systems have been developed elsewhere (9-11). We note that because of this overlap effect, complete equilibrium saturation of a nucleic acid lattice by a noncooperative binding protein cannot be attained. As the binding density of protein on the lattice increases, most of the remaining vacant binding loci are less than n residues in length, and thus cannot be used. To create more usable binding sites some of the smaller vacancies must be rearranged into fewer larger ones. The unfavorable entropy involved in such rearrangements eventually exceeds the potential free energy to be gained by binding another protein molecule, and the binding stops short of lattice saturation. This problem can be overcome and binding saturation attained if binding is cooperative, i.e., if binding of a ligand adjacent to one previously bound is more favorable than isolated binding. Many physiologically active DNA binding proteins show such binding cooperativity, which is characterized thermodynamically by the unitless parameter, w . This parameter corresponds to the equilibrium constant for moving a protein already bound to the lattice from an isolated to a contiguous binding position. As an indication of the magnitudes involved, w = 2 x lo3 for the cooperative binding of T4 gene 32 protein to a single-stranded DNA or RNA lattice. This value of w represents a favorable increment in the net binding free energy (for contiguous over isolated protein binding) of about -4 kcal per mole of protein monomer. These three parameters ( n , K and w) suffice for a complete thermodynamic description of protein-nucleic acid binding interactions of this type. The definitions of these parameters are summarized and illustrated in Fig. 1.

2. Molecular Characterization of the Binding Interaction In addition to the “bare-bones’’ thermodynamic description of a protein-nucleic acid interaction provided by n , K , and w , it is often possible to learn more about the protein binding domain, and about the com9. J. D. McChee and P. H. von Hippel, J M B 86, 469 (1974). 10. J. A. Schellman, fsr. J . Chem. 12, 219 (1974). 1 1 . S . C. Kowalczykowski, L. S. Paul, J. W. Newport, N. Lonberg, and P. H. von Hippel, in preparation.

21. DNA BINDING PROTEINS -n-

-n

-

379

it -&+-AIsolot ed Binding

Contiguous Binding

FIG. 1. Definitions of the thermodynamic parameters describing the interaction of a binding protein with a nucleic acid lattice. Each arrowhead represents a lattice site (i.e., a nucleotide residue) and the illustrated protein covers three such sites in = 3). I< (in M - * )is the intrinsic association constant for protein binding to the lattice at an isolated site, and w (dimensionless) represents the cooperativity of binding (w is defined as the equilibrium constant for moving a protein from an isolated to a contiguous binding site). Thus K w is the net binding constant per contiguously-bound protein molecule. If contiguous binding is fav0red.w > 1; ifcontiguous bindingisdisfavored,w < 1; and ifthe bindingisnoncooperative, w = I. plementary nucleic acid surface, by a variety of other approaches. Thus the number of nucleotide residues actually interacting with the protein On),as opposed to merely being covered by it ( n ) , can often be determiried by measuring the apparent binding constants of a series of oligonucleotides (of lengthl) to the binding site; a particularly simple system of this type is portrayed in Fig. 2. When the length of the test oligonucleotide ( I ) exceeds the number of interacting residues (m), further increase in lattice length should merely increase the apparent oligonucleotide binding constant by a statistical factor, I - t n + 1 (12). In addition, at I = n i , the

FIG. 2. Definitions of molecular binding (interaction) parameters. Here the arrowheads represent the nucleoside (sugar-base) residue, the negative charges in the backbone represent backbone phosphates, the positive charges in the protein represent basic amino acid residues, and the positive changes in the solution represent monovalent counterions. The illustrated protein here cot'ers (occludes) six nucleotide residues (n = 6), but interacts with only three sugar-base units (rn = 3) and forms two charge-charge interactions ( i n ' = 2) with the nucleotide backbone.

3 80

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

apparent protein-oligonucleotide binding constant should equal the intrinsic K obtained with long DNA or polynucleotide lattices. Study of the base and sugar dependence of oligonucleotide binding constants may also provide information about binding specificities (12-14). This approach depends on several crucial premises, including the assumptions (1) that there is no change in protein conformation or binding site geometry in going from the oligonucleotide binding form of the protein to the (often cooperatively bound) polynucleotide binding form, and (2) that the oligonucleotide is free to bind statistically (i.e,, to shuffle) in the protein binding site. We note that neither of these assumptions appear to be valid for the T4 gene 32 protein-nucleic acid binding interaction (14) (see Section IV,A). Record et d . (15) have shown that monitoring the dependence of K (or K w for cooperatively binding proteins) on salt concentration can be developed into a molecular probe of considerable power and generality. These workers point out that (in simple cases) one can treat a protein-nucleic acid interaction as a three component system involving the protein ligand, the nucleic acid lattice, ond the counterions that are bound tightly to this lattice. On binding the (in this sense) plycationic protein ligand, some of these counterions are displaced from the lattice: this results in a dependence of the observed binding constant on salt concentration, which can be interpreted to determine nz ’ , the number of charge-charge interactions involved (per protein monomer) in the formation of the protein-nucleic acid complex (see Fig. 2). This viewpoint also shows clearly that the major source of binding free energy for such electrostatic interactions comes from the entropy of mixing (or dilution) of the ions displaced from the polynucleotide lattice. Such determinations of m ’ have been made for a number of systems ( 1 4 4 7 ) . In some situations anion as well as cation displacement may be involved, and the situation can be more complicated (14). 3. Kirietics of Birzdirzg and Binding Pathways

All of the previous considerations are essentially equilibrium in nature. In “real life” binding, equilibrium is often not attained at every step, 12. R . C. Kelly, D. E. Jensen, and P. H . von Hippel, JBC 251, 7240 (1976). 13. D. M. Draper, and P. H . von Hippel, JMB 122. 321 (1978). 14. S. C. Kowalczykowski, N. Lonberg, J. W. Newport, and P. H. von Hippe1,JMB 145, 75 (1981). 15. M . T. Record, Jr., T. M. Lohman, and P. L. deHaseth, JMB 107, 145 (1976). 16. P. L. deHaseth, T. M. Lohman, and M. T. Record, Biockemirrry 16, 4883 (1977). 17. A. Revzin and P. H. von Hippel, Biochernisrry 16, 4769 (1977).

21. DNA BINDING PROTEINS

38 1

especially in complex and interlocking multicomponent systems. Thus information about association and dissociation rates and pathways is often required as well. Complete information of this type is not at hand for any DNA binding protein, though approaches have been made and some kinetic questions relevant to physiology have been raised (see Section IV,A,2) (18-22). c . METHODSFOR MONITORING BINDING 1. Meusitremetit o j Bitiding Purcimeters

The binding parameters previously described ( i t , fi, o,m , t w ‘ and the related kinetic constants) can be measured in a given system by a variety of approaches. Generally titrations are involved, in which the progress of the reaction is measured by monitoring either spectroscopic or hydrodynamic changes due to complex formation. Binding site size ( n ) can be (and usually is) determined independently by titrating protein with nucleic acid (or vice versa) to saturation under tight-binding (typically low salt) conditions ( 4 ) .Binding constants ( K )and cooperativity parameters ( w ) can be determined by titrating under conditions where binding is less tight, and appreciable (measurable) concentrations of free protein ligands and free nucleic acid binding sites are present in equilibrium with the complexes (14, I h , 17, 3 - 2 5 ) . In addition, all these parameters can also be obtained by analyzing the shapes of binding protein-perturbed nucleic acid melting profiles (4, 2 6 ) . u. Nucleic Acid Signals. The interaction of DNA binding proteins with single-stranded nucleic acid sequences often results in appreciable deformation of the sugar-phosphate backbone, with concomitant unstacking of 18. B. F. Peterman, and C. W. Wu, BiocliemiFtry 17, 3889 (1978). 19. I. R. Epstein, Biopolvmers 18, 2037 (1979). 20. S. C. Kowalczykowski, N. Lonberg, J. W. Newport, L. S. Paul, and P. H. von Hippel BJ 32, 403 (1980). 21. T. M. Lohman, BJ 32, 458 (1980). 22. B. M. Alberts, J. Barry, P. Bedinger, R. L. Burke, U. Hibner, C. C. Liu, and R. Sheridan, in “Mechanistic Studies of DNA Replication and Genetic Recombination” (B. Alberts and C. F. Fox, eds.), ICN-UCLA Symp. Mol. Cellular Biol., Vol. 19. Academic Press, New York, 1980. 23. D. E. Jensen and P. H . von Hippel, JBC 251, 7198 (1976). 24. D. E. Draper and P. H. von Hippel, Bioclienzisrry 18, 753 (1979). 25. T. L. Lohman, C. G. Wensley, J. Cina, R. R. Burgess, and M. T. Record, Jr. Bioclirtnisrry 19, 3516 (1980). 26. J. M. McGhee, Biopolymers 15, 1345 (1976).

382

S. KOWALCZY KOWSKI, D. BEAR, AND P. VON HIPPEL

adjacent nucleotide bases. As a result these processes can frequently be followed by monitoring changes in the spectroscopic properties of the nucleic acid; circular dichroism [backbone deformation; see, e.g., Refs. (4, 27-29)] or UV hyperchroism [base unstacking; see, e.g., Refs. (4, 611 are generally used. In addition (covalently) modified nucleic acids are often useful. For example, the chemical modification of pol yriboadenylic acid to polyriboethenoadenylic acid makes this moiety fluorescent (30), and the fluorescence is greatly enhanced by the base-unstacking brought about, for example, by T4 gene 32 protein (14,31). Nitroxide spin labels have also been attached to polynucleotides, and changes in the resulting ESR spectra on protein binding have been followed (32). b. Protein Signals. Many single-stranded DNA binding proteins show appreciable quenching of intrinsic protein fluorescence on interacting with nucleic acid lattices. These changes (generally in tryptophan, but sometimes in tyrosine fluorescence) can also be monitored to follow proteinnucleic acid binding reactions [see, e.g., Refs. (12, 33-36)]. c. Other Approaches. Changes in sedimentation, electrophoretic, and gel exclusion chromatographic behavior are also used to follow proteinnucleic acid interactions [see, e.g., Refs. (4, 23-25, 3 7 ) ] . DNA cellulose chromatography (38), utilizing the nucleic acid lattice as the stationary phase and the protein ligand as the mobile phase, has also been applied to the measurement of DNA-protein binding constants (14, 16,391. Quantitative photoaffinity cross-linking studies, in which the competition for a DNA binding protein between a photoaffinity-labeled nucleic acid component and a nonlabeled polynucleotide is monitored, also can be made to yield measurements of binding parameters (40). 27. R. A. Anderson and J. E. Coleman. Biochemistry 14, 5485 (1975). 28. A. P. Butler, A. Revzin, and P. H. von Hippel, Biochemistry 16, 4757, (1977). 29. L. A. Day, Biochemistry 12, 5329 (1973). 30. J. A . Secrist, R. J. Bario, N . J. Leonard, and G . Weber, Biochemistry 11,3499 (1972). 31. J. J. Toulme and C. Helene, BBA 606, 95 (1980). 32. A. M. Bobst and Y.-C. Pan, BBRC 67, 562 (1975). 33. R. C. Kelly and P. H. von Hippel, JBC 251, 7229 (1976). 34. C. Helene, F. Toulme, M. Charlier, and M. Yaniv, BBRC 71, 91 (1976). 35. H. T. Pretorius, M. Klein, and L . A . Day, JBC 250, 9262 (1975). 36. I. J. Molineux, A. Pauli, and M. L. Gefter, Nucleic Acids Res. 2, 1821 (1975). 37. K. R. Yamarnoto and B. M. Alberts, JBC 249,7076 (1974). 38. B . M. Alberts, F. J. Amodio, M. Jenkins, E. D. Gutmann, and R. L. Ferris, CSHSQB 33, 289 (1%8). 39. P. L. deHaseth, T. M. Lohman, R. R. Burgess, M. T. Record, Jr. Biochemistry 17, 1612 (1978). 40. S . C. Kowalczykowski and L. S. Paul, unpublished results.

383

21.

DNA BINDING PROTEINS

111.

Protein Isolation and Purification: Proeedures and Strategies

Techniques used to isolate and purify single-stranded DNA binding proteins generally share many common features. In particular, one or more DNA-cellulose columns, first developed by Alberts and co-workers (-18)as a form of affinity chromatography for DNA binding proteins, occupy a central position in nearly every purification scheme. Here we outline some general procedures and strategies that have been used; these procedures are summarized (in outline) in Table I where we present a sample scheme (generally the most recent or most widely used variant if several procedures are in common use) for the isolation of four of the best-characterized prokaryote single-stranded DNA binding proteins. Our purpose in this section is to pinpoint certain general approaches that have been widely employed to purify SSBPs and should probably be considered in developing procedures for the isolation of new members of this protein class. TABLE I PURIFICATION OF PROKARYOTIC SINGLE-STRANDED

DNA BINDINGPROTEINS Protein T4 Gene 32 protein

fd Gene 5 protein

E. coli SSB protein

"7 DNA binding protein

a

Procedure Lysis (sonication and blending), DNase, LSC", D, ssDNAcellulose, norleucineSepharose, pho sphocellulose or phenyl-Sepharose Lysis (sonication), DNAse, LSC', HSC', ssDNA-cellulose, DEAE-cellulose Lysis (sonication), LSC,' PEG' ppt. of DNA, LSC', Dr, ssDNA-cellulose, DEAESephadex Lysis (Sonication), LSC', PEGe ppt. of DNA, LSC', D', ssDNA-cellulose, M&+ ppt., DEAE-cellulose

Yield (mg)

Referencesb

- I6

(56 )

-60

(41)

- 100-250d

-3 -30"

(42)

-3

(161)

Yield from 100 gm (net weight) of E. coli or phage-infected E. coli.

* Only the most recent, or most detailed, reference is listed.

Abbreviations: LSC, low-speed centrifugation to remove cellular debris; HSC, highspeed centrifugation to remove ribosomes; PEG, polyethyleneglycol; D, dialysis into low ionic strength buffer + EDTA. Yield from overproducing strain of E. coli or phage.

384

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

A. PRODUCTION OF CELLULAR EXTRACTS Because most of the proteins discussed in detail in this chapter are derived from uninfected or phage-infected bacterial cells, we confine the majority of our remarks in this section to the production of cellular extracts from prokaryotic cells. Specific references should be consulted for procedures involved in the growth and lysis of eukaryotic cells. 1. SrrL1itz.r

The choice of a strain for the isolation of bacterial and phage-coded, single-stranded DNA binding proteins depends on the nature of the experiments to be undertaken. If the amount of protein required is small, isolation from wild-type strains is often adequate. For example, -8 mg of T4 gene 32 protein can be isolated from 50 g of E. coli cells infected with wild-type T4 phage (6), and -30 mg of fd gene 5 protein can be obtained from 50 g of Ff-infected E. coli (41 1. However, in some cases the yield of protein from wild-type strains is rather poor, or at least insufficient for large-scale physical studies: for example, only -0.5 to 1.5 mg of SSB protein are obtained per 50 g of wild-type E. coli (42). Thus it is often useful to isolate strains of bacteria or phage that will overproduce the protein desired. Sometimes this can be done in a fairly straightforward fashion; thus strains of E. coli that overproduce SSB protein were obtained by simply inserting plasmids or A phage carrying the ssb gene into E. coli ( 4 3 . Overproducers of E. coli lac repressor were found by screening for “up” promoter mutants in the luc i-gene (repressor) promoter (4.0. Sometimes, however, straightforward approaches may not work, either because the overproduced protein is lethal to the cell or because the wild-type free protein level is controlled by autoregulatory feedback mechanisms. In the latter case an appropriate overproducer may be designed by taking advantage of the operation of the regulatory system. For example, the free concentration of T4 gene 32 protein is autoregulated at the translational level (44,451. The system works on the basis of a cascade of binding specificities, as follows (see Section IV,A). First, sufficient protein is produced to saturate all the single-stranded DNA sequences in the cell. Then, after the free concentration of gene 32 protein has risen to a 41. B. M. Alberts, L. Frey, and H. Delius, JMB 68, 139-152 (1972). 42. J. W. Chase, R. F. Whittier, J. Auerbach, A. Sancar, and W. D. Rupp, Nltcleic Acids Res. 8, 3215-3227 (1980). 43. B. Muller-Hill, L. Crapo, and W. Gilbert, PNAS 59, 1259 (1968). 44. H. M. Krish, A. Bolle, and R . H. Epstein, J M B 88, 89 (1974). 45. L. Gold, P. 2. O’Farrell, and M. Russel, J B C 251, 7251 (1976).

21.

DNA BlNDlNG PROTEINS

385

critical level, the protein binds specifically and reversibly to a critical control sequence on gene 32 mRNA and prevents further synthesis (5,46). The key to overproducing this protein, then, was to find a way to keep the .free protein level below the shut-off level while the total protein concentration was increased. This was achieved for gene 32 protein by infecting with a T4 phage that had mutations in several nucleic acid processing enzymes. The net effect of these mutations is to increase greatly the steady-state concentration of single-stranded sequences in the newly synthesized phage DNA. These regions then bind greatly increased quantities of gene 32 protein, and thus permit overproduction by delaying shut-off synthesis (45). In general, before time and effort is put into cloning a gene for the purpose of overproducing a specific protein, it is important to establish whether the protein may be toxic in excess or whether its synthesis is autoregulated, in order that an appropriate overproducing strategy can be devised. 2 . Cell Lysis iitid Processing of the Proteitt Extract Procedures commonly used for bacterial cell lysis, such as grinding with glass beads or alumina, pressure disruption, sonication, and lysozymedetergent treatment, have been used in isolating single-stranded DNA binding proteins. Pressure disruption or lysozyme-detergent treatment are generally the methods of choice for large-scale preparations. A problem that is sometimes encountered with bacteria (particularly B. subtilis), and is very common with cells of higher organisms, is.that of intracellular proteases. Thus, after the cell is broken, normally compartmentalized or membrane-bound proteases may start to attack the desired protein product. This not only reduces yields considerably, but can also generate proteolytic fragments that are difficult to separate from the intact protein. Both general covalently binding protease inhibitors, such as phenylmethylsulfonyl fluoride (PMSF) and diisopropyl fluorophosphate (DFP), and specific complex-forming moieties, such as soybean trypsin or chymotrypsin inhibitors, have been employed. An excellent review of this subject is available (47). Other types of covalent modification of protein during isolation are a potential problem. For example, in several eukaryotic proteins the binding activity depends on the level of protein phosphorylation (see Section IV,E), and these proteins may be inactivated by phosphatases acting in the cell extract during purification. 46. G. Lernaire, L. Gold, and M. Yarus, JMB 126, 73 (1978). 47. J. R. F’ringle, Merlrods Cell B i d . 12, 149-184 (1975).

386

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

After the cells have been lysed, free DNA may make the lysate enormously viscous. This viscosity must be reduced to permit effective removal of cell debris by centrifugation prior to fractionation of the extract. Sonication and treatment with DNase I are the two techniques most frequently used to degrade the free DNA. Preparations of DNase I should be treated with PMSF prior to use to remove contaminating proteases (48). The concentration of salt present in lysing and extraction buffers during protein purification is also often crucial. High salt concentrations are generally used in early steps to liberate the desired protein from single- or double-stranded DNA fragments. The (protein-free) DNA is then removed by procedures such as PEG-dextran two-phase extraction, or precipitation with streptomycin or polyethyleneimine. After initial centrifugation to remove cell debris, various approaches can be employed to achieve a gross fractionation of the cell extract. These techniques may include further centrifugation (to remove ribosomes), heat treatment, ammonium sulfate precipitation, and dextran sulfate or PEG extraction. The final extract is generally dialyzed against low salt buffer containing EDTA (to inhibit nucleases) in preparation for DNAaffinity chromatography. B. DNA-AFFINITYCHROMATOGRAPHY At the heart of almost every scheme for purifying a DNA binding protein lies one or more DNA-cellulose or DNA-agarose column chromatography steps. The columns carry either single- or double-stranded DNA, and the protein extract to be resolved is generally loaded on the columns at low salt and then eluted with a continuous or a step salt gradient. This approach offers a powerful means to isolate and separate the DNA binding proteins of the cell. On the other hand, the simple criterion that a particular protein binds to a DNA affinity column at low salt concentrations has often been used to identify a protein isolated from an otherwise uncharacterized extract as a binding protein. This may be a mistake; at low salt concentrations, in particular, a DNA column can function as a nonspecific cation-exchanger and many proteins stick only because of nonspecific charge-charge interactions. In functional terms proteins isolated this way may have absolutely nothing to do with DNA metabolism or genome function. It is therefore essential to prove by genetic complementation, or by some other biochemical assay, that a particular DNA binding protein does, in fact, play a role in genome manipulation in vivo. In contrast it is also possible for proteins that do have an actual DNA-binding function in vivo to be 48. P.

A. Price, T. Y. Liu, W. H. Stein, and S. Moore,JBC

244, 917 (1969).

21. DNA BINDING PROTEINS

387

lost, damaged, or modified during cell fractionation procedures, and thus fail to bind to DNA-affinity columns. A general procedure for the identification and selective purification of biologically significant DNA binding proteins has been proposed (49). In this procedure the cell extract is first passed through a native DNAcellulose column under defined conditions, and is then loaded onto a column containing denatured DNA. After a washing step, a solution of dextran sulfate is applied to the column. This step should remove proteins that stick to the column only because of weak nonspecific ionic interactions. The remaining proteins are then eluted with high salt. While this protocol may not fractionate all single-stranded DNA binding proteins, any proteins that survive the double-stranded DNA cellulose column and dextran sulfate cuts are reasonable candidates for consideration as biologically relevant DNA binding proteins. The support matrix, and the method used to couple DNA to it, are extremely important considerations in DNA affinity chromatography. Initially, noncovalently linked DNA-cellulose columns (50) and DNAagarose columns (51) were prepared by mixing DNA with such support materials and then drying to attach the DNA. Under certain circumstances such preparations are quite suitable for quantitative protein fractionation. However, such columns often shed DNA, and thus contaminate protein fractions with large quantities of nucleic acid that may be difficult to remove. In addition, the useful lifetime of affinity columns prepared in this manner is often short. These problems have been circumvented by covalently coupling the DNA to the cellulose matrix by ultraviolet photocrosslinking (52) and to agarose by cyanogen bromide activation (53).The relative merits of agarose and cellulose must be decided in each case. Nonspecific adsorption of the protein to the matrix can effect the salt concentration required to elute a DNA binding protein from the column; thus elution patterns for DNA-agarose and DNA-cellulose are not always the same. The use of chelating agents to inhibit divalent-cation- lependent nucleases increases the lifetime of the DNA affinity columns significantly.

C. ADDITIONAL PURIFICATION PROCEDURES Although DNA affinity chromatography plays the most significant role in the purification of DNA binding proteins, most proteins are still some49. G . Herrick and B. M. Alberts, JHC 251, 2124 (1976). 50. B. M. Alberts and G . Herrick, “Methods in Enzymology,” Vol. 21D, p. 198, 1971. 51. H. Schaller, C. Nusslein, F. J. Bonhoeffer, C. Kurz, and I. Nietzschmann, EYB 26, 52. R. M. Litrnan, JBC 243, 6222-6233 (1968). 53. D. J . Arndt-Jovin, T. M . Jovin, W. Bahr, A. Fischauf, and M . Marquardt, W B 54, 41 1 (1975).

388

S . KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

what contaminated after this step. Further purification by ion exchange, gel filtration, and hydrophobic or affinity chromatography (see Table I for examples) is usually required. Affinity chromatography on columns containing conjugated dyes (54) or nucleotides (55) has become increasingly popular. Hydrophobic chromatography has also been employed to remove nuclease contaminants from T4 gene 32 protein (56).

D. ASSAYSA N D CRITERIA OF PURITY Since DNA binding proteins are not enzymatically active, they are difficult to assay in crude extracts. Occasionally an empirical criterion, such as the retention ofE. coli SSB protein on nitrocellulose filters at very high salt concentrations, can be used (57). And sometimes activation of DNA polymerases or nucleases by specific DNA binding proteins can serve as the basis for a biochemical (complementation) assay. However such approaches tend to suffer from considerable variability in crude cell extracts. Generally the purity of DNA binding proteins is established by demonstration of a homogeneous band (or set of bands) on an SDSpolyacrylamide gel, and by the absence of contaminating nucleic acids and enzymatic activities. Many of the tryptophan-containing proteins described in Section IV show 280 : 260 nm absorbance ratios as high as 1.5 to 2.0; lower 280 : 260 ratios generally reflect nucleic acid contamination. In addition to testing for chemical purity as above, it is also important to demonstrate that the purified protein retains biological activity, if such a property has been established. Thus the operation of the pure protein as a specific activator of an in vitro DNA replication, recombination, or repair system can sometimes be monitored. Such assays may reveal the need for a specific cofactor, a special state of aggregation of the protein, or a covalent modification of the protein that is required for function. IV.

Structure, Properties, and Nucleic Acid Binding Interactions of Several Single-Stranded DNA Binding Proteins

In this section we describe several single-stranded DNA binding proteins and their interactions with various types of nucleic acid “substrates,” as well as with other proteins of the relevant DNA replication 54. R . R. Meyer, J. Glassberg, J. Y. Scott, and A. Komberg,JBC 255,2897-2901 (1980). 55. E. Calva and R . R. Burgess,JBC 255, 11017 (1980). 56. M. Bittner, R . L. Burke, and B. M. Alberts, JBC 254, 9565 (1979). 57. R. F. Whittier and J. W. Chase, A n d . Bioclwm. 106, 99 (1980).

21.

DNA BINDING PROTEINS

389

complexes. (Very little information is available at the molecular level about the participation of any of these proteins in integrated recombination or repair systems.) Our knowledge of each of these systems is incomplete, but to a certain extent the available information is complementary, and the reader is urged to take an integrated view. To aid in this we present an extensive summary of the known properties and interactions of several of the major single-stranded DNA binding proteins isolated from prokaryotes in Table 11; a similar (though much more limited) compilation is presented for certain eukaryotic SSBPs in Table 111. The following considerations suggest how one might profitably consider the results presented here in a broader context. T4 gene 32 protein is by far the most carefully studied protein from a physiochemical point of view. Many features of other binding proteins are, at least qualitatively and semiquantitatively, similar to those of gene 32 protein. Thus the approach to (and results from) the measurement of thermodynamic and molecular parameters (e.g., t z , K , o,111, ni’) for this system certainly provides methodology (and probably also reflects orders of magnitude) that may well apply to similar systems. The only SSBP for which a detailed molecular structure is in hand is fd gene 5 protein. Careful analysis of this structure certainly suggests ways that other SSBPs might be built, and ways they could interact with their nucleic acid substrates. The interactions of the E. coli SSB protein with other components of reconstituted replication and recombination systems have been extensively studied: consideration of the enormous “skein” of interactions that involve this protein in so many facets of genome regulation and expression in E. coli certainly provide insight into what one might expect from similar proteins in less well-studied systems. At the same time these are all individual proteins, and will clearly be found to have many features that are less general. Thus, fd gene 5 protein serves a special role in the life cycle of this unusual phage, and its molecular structure and interactions with DNA may well reflect some elements of this. Similarly, while gene 32 protein binds to DNA lattices as a monomer, gene 5 protein probably binds as a dimer and E. coli SSB protein binds as a tetramer. These binding differences require somewhat different interpretations of thermodynamic parameters such as site size ( n ) and cooperativity (w).

A. BACTERIOPHAGE TCCODEDGENE32 PROTEIN The product of bacteriophage TCcoded gene 32 was the first DNA binding protein to be isolated utilizing the DNA-cellulose chromatography method (38). Identification of the protein was achieved by infecting with

TABLE 11

PROPERTIES OF PROKARYOTIC SINGLE-STRANDED DNA BINDINGPROTEINS

Property Physical Molecular weight (monomer) (kdal) Oligomeric state Sedimentation coefficient Number of amino acid residues Extinction coefficient Isoelectric point Binding Polynucleotide site size ( 1 1 ) Relative affinity Cooperativity Ability to melt duplex DNA Ability to reanneal denatured DNA Ability to melt poly(dA-dT) [NaCl] for elution from ssDNA-cellulose" Biological Copies per cell Stimulation of DNA polymerase Stimulation of DNA repair Stimulation of DNA recombination

fd Gene 5 protein

"7 DNA binding protein

9.7 Monomer e dime?

31 (25) Monomer N.D. N.D. N.D.

5.5

30,000 M-' cm-' (280 nm) 6.0

1 . 3 5 a 1.95 87 7100 M-' cm-1 (276 nm) 8.0

7 nuc/monomer DNA > RNA Yes No (GP32*1, yes) Yes Yes >0.6 M, <2.0 M

8 nuc/monomer DNA > RNA Yes Yes Yes' Yes 1.0-2.0 M

4 nuc/monomer DNA > RNA Yes YeS No Yes 0.6 M

N.D. N.D. N.D. N.D. N.D. Yes 0.4-0.6 M

10,000 (monomer) T4 pol Yes Yes

300-800 (tetramer) pol II, pol 111, T7 pol Yes Yes

75,000 (dimer) pol IId N.D.e N.D.

N.D. T7 pol N.D. N.D.

T4 Gene 32 protein

34 Monome? & dimer s?infinite N.D. 301 37,000 M-' cm-I (280 nm)

E. coli SSB protein

19.5 Tetramer 4.7

s

- 190

Because of cooperativity this is a function of protein concentration.

* Probable nucleic acid binding form.

Requires presence of polyamines. Stimulates amount of DNA synthesis, not rate, only at low protein concentration. N.D., not determined.

Acidic

TABLE 111

PROPERTIES OF Property

w

'0

Physical Molecular weight (monomer) (kdal) Isoelectric point Oligomeric state Sedimentation coefficient Binding Polynucleotide site size (n) Relative affinity (DNA vs. RNA) Cooperativity Ability to melt duplex DNA Ability to melt poly(dA-dT) [NaCI] for elution from DNA-cellulose Biological Copies per cell Stimulation of DNA polymerase

EUKARYOTIC SINGLE-STRANDED DNA BINDINGPROTEINS Calf thymus HDP-I

Mouse myeloma

Ustilago maydis

24 7.8 Monomer N.D.'

27 6.6 Monomer N.D.

20 N.D. Monomer 2.6 S

35 N.D. Monomer 3.3 s

72 7.5" Monomer 3.3 s

7 nuc DNA > RNA No Yes Yes

5-7 nuc DNA > RNA No N.D. Yes 0.4 M

3-7 nuc N.D.

N.D. DNA > RNA

7 nuc DNA > RNA

b

b

b

N.D. Yes 1.6 M

Yes N.D. N.D.

N.D. No 1.0 M

1,000,OOO Yes

300,000 Yes

N.D. N.D.

70,000,000 Yes

0.4 M

800,000

" Nonphosphorylated species. Possibly cooperative, but not fully established. N.D. = not determined.

Yes

Lily

Adenovirus

392

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

phage that containedamber mutations in gene 32 and by demonstrating the loss of the corresponding polypeptide from the resulting polyacrylamide gel patterns; controls using phage defective in other genes ( 3 0 , 4 1 4 5 ) had no effect on the band identified as the product of gene 32. Both because gene 32 protein had been shown by genetic analysis to be essential to phage DNA replication (57a) and recombination (58, 59), and because it is produced in large (stoichiometric) quantities (60, 6 1 ) [- 10,000 copieslcell (631, Alberts et al. (6, 38) undertook biochemical studies of this protein. These early studies showed that (1) the protein binds to single-stranded DNA in preference to double-stranded DNA; (2) binding appeared to be cooperative in protein concentration; (3) gene 32 protein can denature duplex poly(dA-dT), but not native duplex DNA; (4) the protein accelerates the renaturation of denatured DNA; and ( 5 ) the protein appears to stimulate specifically the activity of T4 DNA polymerase. Each of these properties of gene 32 protein has since been studied in more detail, and the results are summarized below. 1. Physicul Properties of the Protein Gene 32 protein has a monomeric molecular weight of approximately 35,000, as determined by SDS gel electrophoresis and by a combination of sedimentation and gel filtration studies; it behaves hydrodynamically as if it were a prolate ellipsoid with an axial ratio of 4 : 1 (=120 A in length) (6, 38). A determination of the complete amino acid sequence of the protein yielded a more exact molecular weight of 33,466 (63). u. Aggwgatim State. Although gene 32 protein exists predominantly as a monomer in dilute solution (c0.025 mg/ml), the protein has been shown to undergo extensive self-association at higher concentrations (64, 65). The extent of this aggregation is dependent on many variables, including ionic strength, pH, temperature, and glycerol concentration. The aggregation process seems to consist of at least two different types of 57a. R . H. Epstein, A. Bolle, C. M. Steinberg, E. Kellenberger, E. Boy de la Tour, R. Chevalley, R. S . Edgar, M . Susman, G. H. Denhardt, and A. Lielausis, CSHSQB 28, 375 (1%3).

58. J. Tomizawa, N . Anraku, and Y. Iwama, JMB 21, 127 (1966). 59. A. W. Kozinski and Z. Z. Felgenhauer,J. Virol. 1, 1193 (1967). 60. D. P. Snustad, Virology 35, 550 (1968). 61. N. K. Sinha and D. P. SnustadJMB 62, 167 (1971). 62. B. M. Alberts, Ff 29, 1154 (1970). 63. K. R . Williams, M. B . LoPresti, M. Setoguchi, and W. H. Konigsberg, PNAS 77, 4614 (1980). 64. R. B. Carroll, K. E. Neet, and D. A. Goldthwait, PNAS 69, 2741 (1972). 65. R . B. Carroll, K. E. Neet, and D. A. Goldthwait,JMB 91, 275 (1975).

21. DNA BINDING PROTEINS

393

association. One leads to a unique dimeric species; the other, via an indefinite association process, leads to large aggregates. The indefinite aggregation is particularly susceptible to various agents, being virtually completely inhibited by high concentrations of salt (>1.0 M KCl), high temperature, high pH (>pH lo), and the addition of saturating concentrations of oligonucleotides [e.g., d(pT),,]. Under each of these conditions the limit aggregate is a dimeric species, even at protein concentrations as high as 1 mg/ml. Although these protein-protein association reactions might be considered to reflect the interactions responsible for the cooperative binding of gene 32 protein to nucleic acid lattices, quantitative study of the nucleic acid binding reaction shows that this interpretation is not consistent with the facts. For example, the unitary free energy of the aggregation process is about - 9 kcal/mol, whereas the free energy of the monomer-monomer interaction involved in cooperative binding is only about - 4 kcaVmol under the same conditions ( I ? ) . In addition, while the indefinite selfassociation shows a very large dependence on salt concentration, the cooperativity parameter for nucleic acid binding is not dependent on this variable ( 5 ) . Thus the nonelectrostatic part of the protein-protein selfassociation free energy may contribute to the cooperativity of nucleic acid binding, but it is clear that the interactions involved in self-association are not, in toto, the same as those involved in cooperative binding of this protein on a nucleic acid lattice. 6. Protein Domain Structure. The observation that gene 32 protein can bind to (polyanionic) DNA-cellulose, despite the fact that it carries a net negative charge at pH 8 and is eluted from DNA-cellulose at high concentrations of salt, suggested very early that gene 32 protein must contain a positively charged binding site ( 6 , 3 8 ) .The idea that gene 32 protein might be composed of several functionally distinct domains was given a firmer basis when it was discovered that the native protein can be converted into three discrete products by limited tryptic digestion. These products have been called G32P*I, G32P*II and G32P*III (27, 66,67). The G32P*I fragment results from the removal of -50 amino acid residues from the carboxy terminus of the native protein, the G32P"II product is obtained by removing -20 amino acid residues from the amino terminus, and the G32P*III fragment is produced by the removal of both the amino and carboxy termini. The G32P*III protein core is quite resistant to further proteol ytic degradation. 66. J. Hosoda, B. Takas, and C. Brock, FEBS (Fed. Eur. Biochem. Soc.) Lerr. 47, 338 (1974). 67. H.Moise and J. Hosoda, Nufure (London) 259, 455 (1976).

394

S. KOWALCZYKOWSKI, D. BEAR, A N D P. VON HIPPEL

These proteolytic products differ in their elution properties from DNA-cellulose columns. Thus the native protein and G32P"I (at fairly high protein concentrations approaching DNA site saturation within the protein band) require -2.0 M NaCl for elution from single-stranded DNA-cellulose, whereas G32P"II and G32P*III can be eluted at -0.4 M NaCl concentrations. This latter salt concentration is also sufficient to elute noncooperatively bound native gene 32 protein (38), in keeping with the conclusion (see Section IV,A,2,c) that G32P*II and G32P*III retain most of their intrinsic affinity for single-stranded DNA, but no longer can bind cooperatively. In contrast G32P*I (and, to some extent, G32P"III) bind more tightly to double-stranded DNA-cellulose than the native protein or G32P"II (67), presumably reflecting the fact that the *I (and the *III) species can denature the duplex DNA and bind to the resulting single-stranded sequences (66, 68) (see also Section IV,A,2,c). Additional evidence for the differential involvement of the amino and the carboxy-terminal domains of gene 32 protein in nucleic acid binding was provided when it was shown that the rate at which these regions are removed by proteolytic cleavage depends on whether or not the protein is bound to polynucleotides (69, 70). The cooperative binding of gene 32 protein to polynucleotides was found to enhance the rate of cleavage at the carboxy terminus, but to protect the amino terminus against proteolytic attack. These effects were not observed in the presence of short oligonucleotides [
21. DNA BINDING PROTEINS

3 95

the *II\,a,,hand ii:IIsta,,l,proteins do not self-associate appreciably. Thus it appears that the residues controlling both cooperative DNA binding (74) and protein self-association must be located within 9 amino acids of the N terminus, and that the part of the polypeptide chain responsible for control of DNA melting ability (and interaction with T4 DNA polymerase: see below) is included within the first 25 C-terminal residues. The amino acid sequence data for gene 32 protein can provide additional insight into the domain structure of this protein. Using an empirical secondary structure prediction scheme (7.51, the secondary structure for this molecule has been determined (715).The results suggest that the molecule can be divided into three regions on this basis: The amino terminal (residues 1-35) and the carboxy terminal (residues 187-301) sequences appear to be primarily a-helical, and the middle (core) sequence is predicted to be primarily P-sheet. Whether these predictions reflect reality will eventually be established crystallographically; meanwhile, we find that these numbers are in reasonable accord with circular dichroic results for the whole protein. Thus CD data (27, 77) suggest that gene 32 protein contains -20% a-helix, 20-2596 P-sheet, and 55-60% disordered regions, whereas the secondary structure prediction approach gives totals of 36, 18, and 46% in each of these categories. c. Ccilorimstry. Calorimetric studies on the denaturation of gene 32 protein and its proteolytic products have shown that removal of either the N- or the C-terminal peptide decreases the thermal stability of the protein (73, 78). Under comparable conditions these proteins denature at about 56" (gene 32 protein), 54" (G32P*II), 51" (G32P*I), and 46" (G32P"III). Each protein is further stabilized by single-stranded polynucleotide binding. The thermal denaturation of protein-nucleic acid complexes of gene 32 protein and G32P*I takes place over a much narrower temperature range than does the denaturation of either the free proteins or complexes of either G 3 2 P I I or G32PIII with nucleic acid lattices. This suggests that native gene 32 protein and G32P"I bind cooperatively, whereas G 3 2 P I I and G32P"III do not.

2. Proteiri-Nircleic Acid Iriterrictioris As previously pointed out, several physicochemical features of the binding of gene 32 protein to nucleic acid substrates were recognized in 74. N . Lonberg, S. C . Kowalczykowski, L. S. Paul, and P. H. von Hippe1,JMB 145, 123 ( 198 1).

75. 76. 77. 78.

P. Chou and G. Fasman, Advun. En:.vmo/. 47, 45 (1978). K . R. Williams, M. LoPresti, and M. Setoguchi, JBC 256, 1754 (1981). J . Greve, M. F. Maestre, H . Moise, and J. Hosoda, Biochemistry 17, 887 (1978). K. R . Williams, L. Sillerud, D. Schafer, and W. H. Konigsberg,JBC 254,6426 (1979).

396

S. KOWALCZYKOWSKJ, D. BEAR, AND P. VON HJPPEL

the initial studies of protein retention on DNA-cellulose columns (38).The fact that the protein elutes from double-stranded DNA-cellulose at lower salt concentrations than from single-stranded DNA-cellulose also suggested that the net affinity of the protein for single-stranded DNA should exceed that for double-stranded DNA, and that binding involves ionic interactions. Also, the observation that the amount of gene 32 protein retained on the single-stranded DNA cellulose column depends on the initial concentration of the protein suggested that the binding must be cooperative in nature: no effect of protein concentration on binding affinity was observed with double-standed DNA. t i . Interactions with Single-Strtinded DNA. These early observations formed the basis for subsequent, more quantitative, investigations. Alberts and Frey (6 ) showed, using sedimentation techniques, that saturation of single-stranded (fd phage) DNA occurred at a ratio of -10 nucleotide residues per protein monomer. In addition, these workers estimated that the affinity of gene 32 protein for a site adjacent to a bound protein molecule should be at least 80-fold greater than its affinity for free DNA. Electron microscopic visualization of gene 32 protein-nucleic acid complexes supported the idea that the protein binds cooperatively to single-stranded nucleic acids. With subsaturating amounts of gene 32 protein, binding was seen to occur in clusters; some fd DNA molecules were coated entirely while others showed no bound protein. The protein coats single-stranded DNA to form a flexible, rodlike complex with a diameter of 60 A and an internucleotide spacing of 4.6 A (79). In addition, these studies also showed that although gene 32 protein does not melt native double-stranded DNA, it can invade A-T rich regions of duplex DNA under irreversible binding conditions (e.g., here with glutaraldehyde as a covalent binding agent). i . Polynitcleotides. Quantitative studies of the interaction of gene 32 protein with nucleic acids have utilized changes in the optical properties of either the protein or the nucleic acid that result from complex formation (Section 11,B). Because binding of gene 32 protein results in an unstacking of the bases within the nucleic acid, changes in circular dichroism (4, 27, 77), UV absorbance ( 4 , 5 , 141, and fluorescence of nucleic acid analogues (5, 14) can be used to monitor binding. In general, the binding of gene 32 protein to polynucleotides results in changes similar to those observed upon thermal denaturation of polynucleotides; i.e., a decrease in the CD signal and an increase in the U V absorbances of the solution. In addition the binding of protein is monitored by following the quenching of the 79. H. Delius, N. J. Mantell, and B. M. Alberts, J M B 67, 341 (1972).

21.

397

DNA BlNDlNG PROTEINS

intrinsic fluorescence of the protein. Upon binding to nucleic acids this fluorescence is quenched as much as 60%; the exact value depends on the length and base composition of the nucleic acid lattice used (5, 12, 14,33, 74).

Under conditions of tight binding, the binding site size (n)can be determined (see Section II,B,2). Values of n ranging from 5 to 11 nucleotide residuedprotein monomer (12, 27) have been obtained, with most values clustering at tz = 7 ? 1 (4, 5, 14, 3 4 ) . Under such stoichiometric binding conditions [in general (NaCI) 5 0.1 MI, it is only possible to determine a lower iimit for the net binding constant; i.e., K w e 10*M-' (4, l2,27). However, by performing titrations in increasing concentrations of NaCI, it is possible to weaken the affinity of the protein so that nonstoichiometric titrations are observed (5, 14). As shown in Fig. 3, at 0.01 A4 NaCl the binding is very tight; however, at 0.35

5

15

10 [Gene 32 Protein]

(p MI

FIG.3. Titration curves for the binding of T4 gene 32 protein to a poly(rA) lattice at 25" as a function of NaCl concentrations. The solid curves (except for that labeled "0.01 M NaCI," for which binding was stoichiometric) are calculated using the following best-fit parameters: n = 7 nucleotide residues, w = 2 x 103 and K determined from the measured value of A'W. The points represent the experimental data [taken from Ref. (14)l.

398

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

M NaCl, there is a slight (sigmoid) lag before binding and this lag becomes very pronounced at still higher NaCl concentrations. These data display classical cooperative (sigmoid) binding profiles, and quantitative analysis shows that each of these curves can be fit with.a cooperativity parameter ( w ) of -2 x lo3. Values of the net affinity constant, Kw, can be determined from each set of data by utilizing the fact that at the midpoint of each titration, Kw = l/[protein],,,, (9). The values of Kw for the four nonstoichiometric binding curves in Fig. 3 are 1.7 x I 0 6 M - l , 8 . 2 x lo5 M - l , 2.5 x lo5 M - l , and 1.6 x 105 M - l , reading from left to right. This represents a significant dependence of the net binding affinity of gene 32 protein for polynucleotides on salt concentration, with K w decreasing -7 orders of magnitude per 10-fold increase in salt concentration (14). These direct titration procedures, as well as a salt-induced dissociation procedure, were used to determine values of K w for a variety of polynucleotides as a function of salt concentration (5, 14). The resultant plots of log Kw versus log [NaCl] demonstrate the steep dependence of the net protein affinity on [NaCI] (dlog Kwldlog [NaCl] = -7), as well as the fact that there is a distinct specificity of binding in terms of polynucleotide base and sugar composition. Qualitative experiments involving the competition of two nucleic acid lattices for limiting amounts of gene 32 protein have also established the existence of such a nucleic acid binding specificity by using either changes in electron spin resonance to monitor competition with nitroxide-labeled polynucleotides (32),or by using differences in hyperchromism in polynucleotides upon binding ( 5 ) . The hierarchy of affinities that has been established by direct measurements of K w is (in order of increasing affinity at 0.2 M NaC1): poly(rC) < poly(rU) < poly(rA) < poly(dA) < poly(dC) < poly(dU) < poly(r1) < poly(d1) < poly(dT) [see Table I1 of Ref. ( 5 ) for a quantitative comparison of K w values]. In addition, quantitative analysis of gene 32 protein binding isotherms has yielded separate values of K and w for a number of polynucleotides at different salt concentrations (5, 14). The results of these analyses show that the cooperativity parameter is independent of salt concentration and essentially independent of polynucleotide type ( w values range from 103 to 1@),and that virtually all of the salt dependence of Kw resides in the intrinsic binding constant, K. The results of these equilibrium studies on the interaction of gene 32 protein with single-stranded polynucleotides can be summarized as follows: (1) Binding is highly salt-dependent (dog Kwldlog [NaCl] = -7); (2) binding is highly cooperative (w E lo3 - 104); (3) w is independent of salt concentration and relatively independent of polynucleotide type; (4) K is highly salt-dependent (dlog Kwldlog [NaCl] = -7); ( 5 ) there is distinct specificity of binding with respect to polynucleotide types, and this is

399

21. DNA BINDING PROTEINS

mainly reflected in differences in K ; ( 6 ) the affinity for a deoxypolynucleotide is always greater than that for the homologous ribopolynucleotide (differences in K w range between 10' to lo4); and ( 7 ) values of K w for randomly copolymerized pol ynucleotides and natural DNA can be calculated from compositionally weighted averages of K w for the homopolynucleotides representing each of the constituent bases ( 5 ) . The approach of Record et ul. (1.5) (see Section II,B,3) was used to analyze the salt (both cation and anion) concentration dependence of the binding of gene 32 protein to polynucleotide lattices (14). The results indicate that over one-half of the dependence of K w on NaCl concentration results from the release of - 5 chloride ions from the protein on pol ynucleotide binding, whereas the remainder of the salt dependence is due to the formation of 3 charge-charge interactions between basic amino acid residues on the protein and negatively charged phosphates of the polynucleotide backbone. For the other anions studied, the classical Hofmeister series of ion binding affinity was followed (80);e.g., only -2 acetate ions, and -0-1 fluoride ions were released per protein monomer bound. ii. Oligonucleotides. The interactions of short (I 5 8 nucleotide residues) oligonucleotides with gene 32 protein differ quantitatively and qualita14, 3 ).The followtively from those observed with polynucleotides (I, ing generalizations emerge from these studies: (i) The oligonucleotide binding interaction shows very little salt dependence (dog KO,,,,/ dlog [NaCl] = -0.3); (ii) this interaction shows little dependence on oligonucleotide base composition (factors of less than 2 to 3 in KO,,,); (iii) the dependence of the interaction on sugar type is small ( Koligo twofold greater for oligodeoxyribonucleotidesthan for the homologous oligoribonucleotides); and (iv) there is little dependence of the observed value of on lattice length (i.e., the expected statistical factor is not observed; see Section II,B,3). These results may be interpreted as follows: The low salt dependence suggests that, at most, one (and possibly no) ionic interactions are involved in the formation of the protein-oligonucleotide complex. The absence of the statistical factor in KOiigosuggests that the oligonucleotide is not able to move freely within the binding site, and thus that the protein may bind to an end of the oligonucleotide lattice. This also suggests that a dinucleotide monophosphate constitutes the minimum binding unit for Finally, the dinucleotides pApA and ApAp oligonucleotide binding (I_)). bind to gene 32 protein with identical affinities but cause very different levels of fluorescence quenching ( 2 versus 17%). This shows that binding

-

-

80. P. H . von Hippel and T. S c h l e i c h , A c c . Clwrn. Re.?. 2, 259 (1969).

21. DNA BINDING PROTEINS

40 1

of oligonucleotides to the protein is polar; i.e., that an oligonucleotide can interact with the binding site in only one orientation. iii. Birzding models. The facts discussed in the previous sections concerning the interaction of gene 32 protein with polynucleotides and oligonucleotides, as well as some of the results of binding studies with the proteolytic products G32P*I and G32P*III, can be incorporated into a schematic model of gene 32 protein binding to nucleic acids, which is presented in Fig. 4. Two different types of binding conformations are proposed; these are termed the oligonucleotide (a), and the polynucleotide (b and c) binding modes. In Fig. 4a (the oligonucleotide binding mode) the protein is shown to interact nonelectrostatically with two residues at the end of an oligonucleotide. This is consistent with the fact that a dinucleotide binds better than a mononucleotide and that further increases in length of the oligonucleotide has no effect on Koliso.The lack of statistical effect on binding for larger oligonucleotides is accounted for in the model by the presence of a steric constraint (the “arm” or “flap” in the drawing). No electrostatic interactions are shown for oligonucleotide binding. In Fig. 4 (the polynucleotide binding mode), an additional binding subsite becomes available due to a displacement of the flap. We assume this displacement occurs because polynucleotides provide very few lattice ends relative to the number of potential internal binding sites. Displacement of the flap is a free energy-requiring process; thus almost all of the values of K for polynucleotides are lower than the values of Koligo.However the displacement of the flap uncovers -3 positively charged amino acid residues on the protein that can form charge-charge interactions with the phosphate residues of the polynucleotide. Simultaneously, an anion binding site (or sites) is destroyed, resulting in the release of the anions upon formation of the complex. In addition, Fig. 4 shows that gene 32 protein covers -7 nucleotide residues when bound, and indicates that FIG. 4. Schematic models of three modes of gene 32 protein binding to nucleic acid lattices. (a) Binding in the oligonucleotide binding mode: note the presence of the block to statistical “shuffling” of the oligonucleotide in the binding site, as well as the presence of the anion binding site. (b) Isolated binding in the polynucleotide mode: note that the “shuffling block” has been moved away, exposing the positively changed binding subsite, and the anion binding site has been disrupted. Also the nucleic acid lattice between the two binding subsites is somewhat stretched, and the nonelectrostatic (XpX) binding subsite is somewhat altered, indicating that gene 32 protein binding in this mode (conformation) shows somewhat enhanced base compositional specificity. (c) Continguous binding in the polynucleotide mode; the gene 32 protein binding conformation is unchanged from that of (b), except for protein-protein interactions and cooperative extension of the nucleic acid lattice between and through contiguously bound protein monomers [taken from Ref. (W1.

402

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

binding results in increasing the internucleotide spacing of the DNA lattice to - 5 A. In Fig. 4c (the cooperative polynucleotide binding mode) contiguous proteins are shown to interact with one another: these interactions are nonelectrostatic since the magnitude of w is salt-independent and reflects primarily protein-protein interactions [see Ref. (7411. These models are also consistent with proteolytic digestion studies of gene 32 protein complexed with nucleic acids (69, 70). The enhancement of the rate of proteolysis of the carboxy terminus of DNA-bound gene 32 protein relative to that of the free protein, suggests that the flap in Fig. 4a may comprise the carboxy terminus of the polypeptide chain. In contrast, cooperative polynucleotide binding protects the amino terminus, just as the “bump” on the left of each protein monomer is shown to be protected in Fig. 4c. Finally, the binding of oligonucleotides has no effect on the proteolytic digestion patterns obtained (relative to those obtained with the free protein), which is consistent with the model shown in Fig. 4a. This schematic model of gene 32 protein binding also predicts that G32P”I should display altered oligonucleotide binding properties, that G32P*II should not bind polynucleotides cooperatively, and that G32P”III should share both of the above properties. These predictions are consistent with experimental observations (see Sedion c below). b. Interaction with Double-Stranded DNA. Most studies on the interaction of gene 32 protein with duplex DNA have focused on the effect of protein on the thermal denaturation profile of the nucleic acid (4, 6, 77). It has been shown that gene 32 protein will denature poly(dA-dT), but is incapable of denaturing native T4 or T7 DNA (4. 6). The binding of gene 32 protein to duplex DNA is quite weak, with values of K ranging from 4 x lo4 M-’ at 0.02 M NaCl to 8.0 x lo3 M-’ at 0.05 M NaCl ( 4 ) . In binding to duplex DNA, gene 32 protein covers approximately 5 base pairs and forms 1 to 2 electrostatic interactions with phosphate groups. There is no evidence for cooperative binding to duplex DNA. These binding parameters for gene 32 protein to duplex DNA can be utilized, together with theoretical approaches to ligand-perturbed double-helix coil transitions, to determine values of K and w for the gene 32 protein interaction with single-stranded DNA (26). The results of such calculations, based on the thermal denaturation measurements of 103 poly(dA-dT) at 0.01 M NaCl, are that IZ = 7.5, K = lo7M - l , and w (4); these parameters are in excellent agreement with those extracted from direct measurements of the affinity of gene 32 protein for single-stranded nucleic acids. Using these values, it is possible to calculate the expected thermal denaturation temperature (T,) of natural duplex DNA in the presence of

21.

DNA BINDING PROTEINS

403

gene 32 protein. Although it can be calculated that gene 32 protein should lower the T,, of the T7 DNA by -60°, no destabilization was observed under any conditions. No melting of the duplex DNA by gene 32 protein was observed for either whole or sonicated T7 DNA, even after 8 hours at a temperature 20" below the unperturbed T,. Similar results were obtained for other natural DNAs, including those extracted from calf thymus (58% A-T), Clostridium perfringens (69% A-T), and Micrococcus lysodeikticus (28% A-T). These results strongly suggest that the melting of native duplex DNA by gene 32 protein is kinetically blocked. c . Intermtion of the Proteolytic Digestion Products of Gene 32 Protein with Nircleic Acids. The digestion products of gene 32 protein formed by limited proteolysis have been shown to possess different DNA binding properties, as assessed by their affinities for DNA-cellulose (67) (also see Section b above]. To more fully understand the molecular aspects of the interactions of gene 32 protein with nucleic acids, and to investigate the functional role of the N- and C-terminal domains in these interactions, the binding of these products to nucleic acids has been studied (67, 68, 7 4 , 8 1 ) . i. G32P"I. The oligo- and polynucleotide binding properties of G32P"I are very similar to those of the native protein (74). With respect to oligonucleotide binding, all interactions are essentially the same for the two proteins except that G32P"I shows a greater electrostatic component of the binding free energy and a greater salt dependence of binding for 6-mers and 8-mers, with the nonelectrostatic component remaining the same for both species. This result is most easily interpreted in terms of the model shown in Fig. 4a, by postulating that this proteolytic cleavage removes at least part of the flap on the lower right-hand side of the model, thus making the charge-containing binding protein subsite available to the longer oligonucleotides, at least in part. The polynucleotide binding properties of the native protein and G32P*I are also virtually identical; the only significant difference is that the value of K is 2- to 3-fold greater for G32P"I for all the polynucleotides. This single (and small) thermodynamic difference between the two proteins is particularly noteworthy when we recall that the gene 32 protein cannot melt native double-stranded DNA, whereas G32P"I can ( 66, 68) .It seems unlikely that this small difference in binding affinity can explain the melting differences, suggesting that melting by the two proteins must involve very different kinetic pathways (4). This conclusion, and the thermodynamic differences have been confirmed by comparing the differences between the melting temperature depressions induced by gene 32 protein and G32P"I on poly(dA-dT). The 81. E. K . Spicer, K . R. Williams, and W. H. Konigsberg,JBC 254, 6433 (1979).

404

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

observed change in TI,,(-- 12”) (70) corresponds to about a twofold difference in net binding affinity. Based on these results (since G32P*I denatures native T4 DNA at -70” below the unperturbed TI,,in 0.01 M NaCI) (37),we would expect that the TI,,of this DNA should be lowered -60” in the presence of native gene 32 protein if this melting goes to equilibrium. A T , , depression of exactly this magnitude has, in fact, been calculated, but was not observed ( 4 ) . This confirms that gene 32 protein is indeed kinetically blocked from melting native DNA, and that this block is effectively removed when G 3 2 P I serves as the melting protein. ii. G32P*IZI. This proteolytic product of gene 32 protein differs appreciably in its nucleic acid binding properties from both the native protein and G32P*I, though it retains some features of these precursors. The most striking change is that binding cooperativity is abolished: G32P*III binds to polynucleotides with a measured value of w = 1 (74). G32P”III has a smaller site size (n = 5-6) than either the native protein or GP32*1; in addition, the magnitude of the salt dependence of binding of this product to both polynucleotides and oligonucleotides is somewhat changed, though the overall relative specificity of polynucleotide binding displayed by the native protein is approximately retained. Despite the fact that G32P”III binds to polynucleotides noncooperatively ,binding induces the same changes observed with the cooperatively bound proteins in the optical properties of the polynucleotides (74). This suggests that lattice distortion [lattice-mediated cooperativity ; see Ref. (82)]is not centrally involved in cooperative binding, and that proteinprotein interactions are predominantly responsible for the binding cooperativity of gene 32 protein (and GP32”I). This cooperativity appears to be localized at the amino terminus of the protein, and the results of the Stclphalococcus N U ~ ~ Nprotease S digestion experiments suggest that essential residues for this property fall within 9 residues of the N terminus (71). The “ball” of the “ball-and-socket’’ interaction between adjacent protein molecules in Fig. 4c provides a simple functional representation of the residues that are cleaved off in forming G32P*III (and G32P*II). iii. G-VP*II. No direct quantitative studies of the interaction of G32P*II with nucleic acids have as yet been carried out. However, based on its DNA-cellulose elution behavior and on the studies with G32P”I and *HI, it is possible to establish some of the properties of this proteolytic product by inference. Since the N terminus is required for cooperative interactions, G32P*II should bind to polynucleotides noncooperatively . This conclusion is con82. P. H. von Hippel. D. E. Jensen, R. C. Kelly, and J. D. McGhee, in “Nucleic AcidProtein Recognition” (H. J. Voegl, ed.), p. 65. Academic Press, New York, 1977.

21. DNA BINDING PROTEINS

405

sistent with both its single-stranded DNA-cellulose binding behavior (67) and its calorimetric properties (73) (see above). In addition, since the N terminus may also be responsible, at least in part, for protein selfassociation, the G32PII product should not form indefinite aggregates; this is also confirmed by preliminary studies (71, 73). Furthermore, since G32P*II retains its C-terminal peptide, we may also expect that this product (like the native protein) will be kinetically blocked from melting native double-stranded DNA. This is supported by the fact that G32P"II (like native gene 32 protein but unlike G32P"I and *IIJ), does not bind to double-stranded DNA-cellulose (67). d. Kinetics of the Binding of Gene 32 Protein to Nucleic Acids. As discussed in the preceding section, the kinetics (as well as the thermodynamics) of the interaction of gene 32 protein with its various nucleic acid substrates must be elucidated in order to develop a complete understanding of the physicochemical and biological properties of this protein. Such studies are quite incomplete, and have focused mainly on doublestranded DNA, poly(dA-dT) denaturation and renaturation rates (6), and the kinetics of the binding of gene 32 protein to single-stranded DNA (18, 3, ' 1 , 83). i. Denaturation of poly(dA-dT). Since gene 32 protein is kinetically blocked from denaturing double-stranded DNA (6, 12), denaturation rate studies with the native protein have been possible only with poly(dA-dT). The melting of this model DNA duplex has been monitored spectrophotometrically, using the increase in OD,,, to follow the reaction. The rate of melting is slow, and as expected depends strongly on salt concentration (6): e.g., a half-time for denaturation (at 25") of -20 min was observed in 10 m M MgS04. This f,,, increased to -300 min in -40 mM MgSO,. More detailed kinetic studies on this system, and of the kinetics of the denaturation of native DNA by G32P*I, are in progress in this laboratory (84 1. ii. Renaturation of double-stranded DNA. Since DNA denaturation is reversible the perturbation of the rate of renaturation of DNA by gene 32 has also been studied. The DNAs used in these studies were first denatured by alkali, then (after neutralization of the solution) gene 32 protein in various concentrations was added, and finally MgSO, was added to induce renaturation. The reactions were monitored by following decreases in ODz6,. Studies of this sort have shown that renaturation rates can be accelerated over 1000-fold by gene 32 protein. As observed in the denaturation rate studies, these effects are salt-dependent, with the stimulatory 83. P. Suau, J. J. Toulme, and C. Helene, Nucleic Acids Res. 8, 1357 (1980). 84. N. Lonberg and S . C. Kowalczykowski, unpublished observations.

406

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

effect of gene 32 protein on the rate decreasing as the salt concentration is increased. Studies in which the concentration of DNA has been varied at less than saturating concentrations of gene 32 protein have shown that the process that is being accelerated is primarily the bimolecular (nucleation) step in the reassociation of the complementary DNA strands (6); the rate of the subsequent “zippering” process is not appreciably affected. It has been concluded that gene 32 protein accelerates renaturation by holding the single-stranded DNA moieties in a favorable, unfolded conformation, which increases the probability of pairing by complementary sequences during strand collisions. It has also been suggested, on the basis of these studies, that bound gene 32 protein might be displaced as renaturation proceeds (6). iii. Kinetics of the association and dissociation reactions of gene 32 protein with single-strunded nucleic acids. The kinetics of the interaction of gene 32 protein with various single-stranded nucleic acids has been investigated by using stopped-flow techniques; the course of the reaction is followed by monitoring the changes in optical properties of either the protein or polynucleotide upon complex formation (18, 32,21, 85). Dissociation kinetics can be studied by subjecting the preformed protein-nucleic acid complex to an ionic strength jump, resulting in dissociation of the complex (18). The rate of dissociation was found to depend strongly on the extent of protein saturation of the nucleic acid lattice (18, 20, 21), as well as on the composition of the nucleic acid. In fact, the same order of polynucleotide affinities observed in the equilibrium experiments is reflected in the dissociation rates as well (20, 2 1 ) . It has been suggested that the dissociation of gene 32 protein occurs primarily from the ends of cooperatively bound clusters of protein, and that this type of mechanism facilitates the renaturation of DNA strands during various biological processes (18). In addition, similar studies have suggested that the bound protein molecules do not behave independently during dissociation, but may be somewhat mobile on the DNA lattice (20). Studies of the kinetics of the association of gene 32 protein have shown that this is a multistep process (85). The steps include (at least) the preequilibrium formation of noncooperatively bound protein, the growth of cooperatively bound clusters of gene 32 protein, and finally the redistribution of the clusters to form a final equilibrium state. The measured bimolecular association rate constant is 3 x lo6M-’(nucleotide) sec-’ [or 2 x lo7 M-I (protein) sec-’1. In addition, the data suggest that the

85. T. M. Lohman and S. C. Kowalczykowski, J M B , in press (1981).

21.

DNA BINDING PROTEINS

407

cooperative growth step may occur by a process that involves translocation of bound proteins along the DNA lattice (86). e. Moleculiir Detuils of the Birding Interaction. Several approaches have been made to attempt to elucidate additional molecular aspects of the nucleic acid binding reaction of gene 32 protein. Equilibrium binding studies (see Section a above) have suggested that two nucleotide residues and two or three nucleotide phosphates interact directly with the protein in the polynucleotide binding mode. We ask here which protein residues are involved in the interaction? This question can be quite definite since the amino acid sequence of the protein is available (62). The involvement of aromatic amino acids in protein-nucleic acid interactions has been much investigated because of the potential of these moieties to stack on, or intercalate between, nucleotide bases [e.g., see (92 )] . Furthermore tyrosine residues have been strongly implicated in the binding of phage fd gene 5 protein to DNA (see Section IV,B,2). Thus chemical modification studies of the nucleic acid binding site of gene 32 protein have also focused on aromatic residues. Tetranitromethane has been used as a chemical probe to determine the accessibility of tryosine residues in gene 32 protein. It was found that 4 or 5 out of 8 tyrosine residues on the native protein can be modified with this reagent. Nitration of these residues completely abolishes the DNA binding activity of the protein: furthermore, no tyrosine residues are modified when the protein is bound to single-stranded DNA (27). These results strongly implicate tyrosine residues in the nucleic acid binding interaction of gene 32 protein. The position within the protein primary structure of the tyrosine residues nitrated in the above study is not known. However the amino acid sequence of gene 32 protein shows a very suggestive distribution of tyrosine residues. Five of the eight tyrosine residues are located within the proteolytically resistant protein core region, and are distributed at 7 to 9 residue intervals along the sequence (63). Whether these residues actually comprise part of the DNA binding site must await further characterization of the protein. 86. The processes by means of which E . coli I N C repressor can translocate on DNA have been extensively studied, both theoretically and experimentally (87-91 ). These results may provide insight into analogous kinetic mechanisms which may be involved in single-stranded DNA binding protein-nucleic acid interactions. 87. P. H. Richter and M. Eigen, Biophys. Clicvn. 2, 255 (1974). 88. 0. G . Berg and C. Blomberg, Biopkys. Cliern. 4, 367 (1976). 89. M. D. Barkley, P. A. Lewis, and G. E. Sullivan, 5J 32, 452 (1980). 90. 0. G . Berg, R. B. Winter, and P. H. von Hippel, Biochemistry, in press (1981). 91. R. B . Winter, 0. G . Berg, and P. H. von Hippel, Bioc,hrmisrry. in press (1981). 92. C. Helene, in “Excited States in Organic Chemistry and Biochemistry” (B. Pullman and N . Goldblum, eds.), p. 65. D. Reidel, Holland, 1977.

408

S . KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

The involvement of aromatic amino acids in the nucleic acid binding site of gene 32 protein is also currently being examined by l9F-NMR techniques using fluorine-substituted analogues of these amino acid residues. Preliminary results show that all five tryptophan resonances can be resolved, and that none are shifted when the protein binds to various oligonucleotides (an upfield shift would be expected if extensive stacking interaction between tryptophan residues and nucleotide bases accompanied binding; see Section IV,B,2,e for a comparable study on fd gene 5 protein). Shifts of some of the fluorotyrosine resonances are observed on single-stranded DNA binding, suggesting that some tyrosine residues may be involved in polynucleotide binding (93). 3. Biologicul Roles Replication. Numerous genetic studies have shown that gene 32 (1. protein is absolutely essential to the DNA replication of the T4 phage (-38, 57tr, 59, 94). Gene 32 protein is required continuously throughout the replication process, and if a phage that contains a temperature-sensitive mutation in gene 32 protein is switched to a nonpermissive temperature, DNA replication ceases immediately (38, 94). Some of the possible roles of gene 32 protein in DNA replication can also be demonstrated in in vitro experiments. Gene 32 protein has been shown to bring about a 5- to 10-fold increase in the rate of DNA polymerization by T4 DNA polymerase on a primed single-stranded homopolynucleotide template (95). Also sedimentation analyses have demonstrated that gene 32 protein binds weakly but specifically to T4 polymerase in solution (95). Neither of these effects is observed in heterologous systems, i.e., with the substitution of E. coli polymerase I for T4 polymerase, or the substitution of E. coli binding protein (SSB) for gene 32 proteins ( % -98). A recent kinetic investigation has confirmed and extended these results by demonstrating that the T4 DNA polymerase engaged in synthesizing a complementary strand to single-stranded fd phage DNA will pause at sites on the template that are capable of forming stable (about - 15 kcal/mol) 93. S . C. Kowalczykowski, R. A. Anderson, V. Ochs, F. W. Dahlquist, and P. H. von Hippel, unpublished observations. 94. S. Riva, A. Cascino, and E. P. Geiduschek, JMB 54, 85 (1970). 95. J. A. Huberman, A. Komberg, and B. M. Alberts, JMB 62, 39 (1971). %. N. Segal, H. Delius, T. Komberg, M. L. Gefter, and B. M. Alberts, P N A S 69, 3537 (1972).

97. C. C. Liu, R. L. Burke, U . Hiber, J. Barry, and B. M. Alberts, CSHSQB 43, 469 (1979). 98. R. L. Burke, B. M. Alberts, and J. Hosada, JBC 255, 11484 (1980).

21. DNA BINDING PROTEINS

409

secondary structures, and that the addition of gene 32 protein causes a 7.5-fold stimulation of the rate of DNA polymerization in this system, presumably as a consequence of the destabilization of the hairpin structures (99). If nicked double-stranded DNA is used as a template for DNA polymerase the situation becomes more complex, and the addition of gene 32 protein has only a limited effect (100, 101); less than 1% of the maximum in vitro rate is obtained (101). In addition, the product formed is largely A-T rich, and rapidly renaturable, indicating that A-T rich regions in the DNA are being copied preferentially (100). However, reasonable rates of DNA polymerization can be achieved if additional T4-coded proteins are added to the reaction mixture [for reviews see (22, 97, 10210411. The T4 accessory proteins (gene products 44, 45, and 62), together with gene 32 protein, are essential for efficient replication of doublestranded DNA on a nicked duplex template. The exact role of gene 32 protein in this process is not clear, but the free concentration of gene 32 protein in such in vitro reconstituted replication systems is very important. If gene 32 protein is omitted, no synthesis is observed; furthermore, the rate of replication fork movement increases almost linearly with increasing concentration of gene 32 protein, up to a rate of -200 nucleotides/sec at 200 pglml(22). Clearly, gene 32 protein plays an important role in helix destabilization in this system. However, there must be additional sources of free energy for helix unwinding since the rate of this five-protein system is well below that observed in vivo [-750 nucleotides/sec (105)],and also below that observed when additional T4 proteins are included in the mixture to reconstitute the seven-protein system [ -500 nucleotides/sec (22, l o / ) ] . In addition to the stimulatory effects on DNA polymerization rates, which are due to the ability of gene 32 protein to destabilize weak hairpins in template DNA, this protein may also stimulate DNA synthesis by increasing the processivity of the T4 polymerase. An increase in pro99. C. C. Huang and J. E. Hearst, A n d . Biochem. 103, 127 (1980). 100. N. G . Nossal, .,’BC 249, 5668 (1974). 101. N. K . Sinha, C. F. Morris, and B. M. Alberts, JBC 255, 4290 (1980). 102. B. M. Alberts, C. F. Morris, D. Mace, N. Sinka, M. Bittner, and L. Moran, in “DNA Synthesis and Its Regulation” (M. Gouliun and P. Hanawalt, eds.), Vol. 111, p. 241. Benjamin, Menlo Park, California, 1975. 103. B. M. Alberts, J. Barry, M. Bittner, M. Davies, H. Hama-Inaba, C. C. Liu, D. Mace, L. Moran, C. F. Morris, J. Piperno, and N. K. Sinka, in “Nucleic Acid-Protein Regulation” (H. J . Vogel, ed.), p. 31. Academic Press, New York, 1977. 104. N . G . Nossal and B. M. Peterlin, JBC 254, 6032 (1979). 105. D. McCarthy, C. Minner, H. Bernstein, and C. Bernstein,JMB 106, %3 (1976).

4 10

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

cessivity allows the polymerase to remain attached to the DNA primertemplate for a longer time per polymerase binding event, thereby increasing the macroscopic rate of DNA synthesis without a concomitant change in the microscopic polymerization rate (22, lUl-lU-?). Recent quantitative studies on the processivity of the T4 replication systems have demonstrated that gene 32 protein is required for processive synthesis in vitro; if this protein is omitted, replication becomes much less processive or completely dispersive (7, 1U6). It has been shown, by investigating the effects of replacing gene 32 protein by G32P”I in in vitro replication systems, that the carboxyterminal domain of this protein is essential in at least two aspects of the replication process. G32P*I does not interact properly with T4 polymerase and also inhibits RNA primer formation (72, 98). T4 DNA polymerase does not cosediment with G32PXI as it does with gene 32 protein, and G32P*I inhibits DNA synthesis by T4 polymerase when single-stranded templates are used. In addition, G32P”I does not interact with the priming protein (gene product 61), leading to the disruption of both primer synthesis and primer utilization (72, 98). b. Recombination and Repair. Numerous investigations have demonstrated that gene 32 protein is essential for recombination in the T4 genome (58, 59, 107-109). Recombination in T4 does not occur (591, or is greatly reduced, if the phages are grown under semipermissive conditions (107). In addition the formation of branched DNA molecules (intermediates in recombination) is reduced 10-fold in phages that carry a defective gene 32 protein gene (108). It has been suggested that the helix-destabilizing capacity of gene 32 protein and, perhaps more importantly, the DNA renaturation activity of this protein, may play a role in genetic recombination (6). The notion that this protein might catalyze the formation of heteroduplex molecules of DNA has been supported by in vitro experiments showing that the formation of joint DNA molecules infective in a spheroplast assay is stimulated -5-fold by added gene 32 protein (109). Gene 32 protein has also been implicated in DNA repair. Phages that carry a temperature-sensitive gene 32 are defective in the repair of UVinduced lesions at nonpermissive temperatures; this may be due to the

-

106. J. W. Newport, S . C. Kowalczykowski, N . Lonberg, L. S. Paul, and P. H. von Hippel, it? “Mechanistic Studies on DNA Replication and Genetic Recombination” (B. M . Alberts and C. F. Fox, eds.), ICN-UCLA Symp. Mol. Cellular Biol., Vol. 19. Academic Press, New York, 1980. 107. H. Berger, A. J. Warren, and K. E. Fry, J . Virnl. 3, 171 (1969). 108. T. R. Broker and I. R . Lehman, JMB 60, 131 (1971). 109. W. Wackernagal and C. M . Radding, P N A S 71, 431 (1974).

?I.

DNA BINDING PROTEINS

41 1

inability of the mutant gene 32 protein to bind to (and protect against nucleases) the single-stranded, gapped sections of DNA produced in the excision-repair process ( 1 10). This supposition has been strengthened by the observation that T4 phages that are temperature-sensitive in gene 32 suffer rapid nucleolytic degradation of their DNA when shifted to nonpermissive temperatures (1 I 1). c . Aritogenoins Regulation of Gene 32 Protein Synthesis. Gene 32 protein regulates its own expression at the translational level ( 4 4 4 6 , 112-114). The observations that have contributed to this conclusion include ( 1) nonsense mutations in gene 32 protein overproduce the nonsense fragment peptides, (2) overproduction in mixed infections is recessive, (3) the amount of overproduced gene 32 protein is related to the amount of single-stranded DNA present (44, 451, and (4) gene 32 protein mRNA is very stable. These primarily genetic conclusions have been confirmed by iii vitro experiments showing that purified gene 32 protein can specifically inhibit translation of its own mRNA, and this inhibition occurs only at concentrations of gene 32 protein sufficient to complex all the singlestranded DNA present [for a review, see Ref. (11311. These data are consistent with a model in which gene 32 is synthesized as needed to saturate regions of single-stranded DNA that are produced in the course of replication, recombination, and repair. When sufficient protein has been produced the excess gene 32 protein binds specifically to gene 32 protein mRNA, and reversibly shuts down further translation of this message. The molecular basis of this recognition and specific shutdown have been speculated upon elsewhere (5, 46, /Oh). However we note, at a minimum, that the protein must (i) demonstrate a strong effective preference for single-stranded DNA over RNA sequences, and (ii) show a significant preferential affinity for its own mRNA relative to other T4 mRNAs. Physical chemical studies have shown that the first requirement is met as a consequence of the high degree of cooperativity of gene 32 protein binding (5). The exact nature of the binding site on gene 32 mRNA may soon be elucidated, since the relevant piece of DNA has been cloned and partially sequenced ( I 14). 110. J. R . Wu and Y-C. Yeh,J. Vi,o/. 12, 758 (1973). I 1 I . M. J . Curtis and B . M. Alberts, J M B 102, 793 (1976). 112. M. Russel, L. Gold, H . Morrissett, and P. Z. O’Farrell,JBC 251, 7263 (1976). 113. L. Gold, G. Lemaire, C. Martin, H . Morrissett, P. O’Conner, P. O’Farrell, M. Russel, and R . Shapiro, in “Nucleic Acid-Protein Interactions” (H. J. Vogel, ed.), p. 91. Academic Press, New York, 1977. 114. H. M. Krish, R. M. Duvoisin, B. Allet, and R . H . Epstein, in “Mechanistic Studies on DNA Replication and Genetic Recombination” ( B . M. Alberts and C . F. FOX, ed.), ICN-UCLA Symp. Mol. Cellular Biol., Vol. 19. Academic Press, New York, 1980.

412

S. KOWALCZY KOWSKI, D. BEAR, AND P. VON HIPPEL

4. Getietic Approaches Elegant genetic studies involving a large number of temperaturesensitive and amber mutations in gene 32 protein have demonstrated that this protein participates in a variety of phage functions, and researchers have mapped the regions of the protein that are involved in these interactions (115, 116). The results show that gene 32 protein interacts with T4 DNA polymerase, ligase, and gene 4 6 4 7 nuclease. In addition these studies have shown that the N-terminal domain of the protein is involved in DNA binding and in interactions with proteins that initiate DNA replication and recombination, whereas the C-terminal region may be required to modulate the activity of the nucleases that act during recombination, and to protect the DNA from excessive degradation. Several E. coli gene products (dtzoc, d n a G ) may be able to substitute for gene 32 protein in the first round of DNA replication, but the specific interaction of gene 32 protein and T4 DNA polymerase is essential to subsequent DNA replication and recombination. Thus gene 32 protein must play a central role in coordinating and controlling the activities of many of the enzymes involved in T 4 regulation and expression.

B . FILAMENTOUS PHAGEGENE 5 PROTEIN The DNA binding protein of the filamentous phages M13, fd, 23-2 and f 1 is the product of gene 5 of these phage genomes (41, I17j. The protein is essential to the life cycle of the phages, and has been shown to be essential in controlling the switch to the production of single-stranded viral DNA from the double-stranded replicative form during replication ( I 18). As observed for T4 gene 32 protein, fd gene 5 protein is produced in large quantities (-75,000 copieslcellj and binds preferentially and cooperatively to single-stranded DNA, thereby lowering the T, of both poly(dA-dT) and native DNA (41, 117). Gene 5 protein is found closely associated with single-stranded viral DNA in a molecular ratio of 1600 gene 5 protein monomers to one viral DNA molecule. It is clear that gene 5 protein is not a structural protein of the fd viral coat (119). Although gene 5 protein has DNA binding properties similar to those of the other prokaryotic HDPs, its role in phage replication is unique, and is discussed in the following sections.

-

115. 116. 117. 118. 119.

A. M. Breschkin and G. Mosig, JMB 112, 279 (1977). A . M.Breschkin and G . Mosig, J M B 112, 295 (1977). J. L. Oey and R . Knippers, J M B 68, 125 (1972). J. S. Salstrom and D. Pratt, J M B 61, 489 (1971). T. J. Henry and D. Pratt, /"AS 62, 800 (1969).

21. DNA BINDING PROTEINS

413

1. Pliysiccil Properties o j the Proteiti The product of fd gene 5 is a small protein, with a molecular weight of sequences of both the M 13 and the fd gene 5 protein have been determined and the exact value of the molecular weight (from the amino acid composition) is 9688 ( 1 3 , 121). The sequences of the M13 and fd proteins are identical. N. Aggregcrtioti State. Gene 5 protein exists in solution mainly in a monomer-dimer equilibrium state (3S,117, 122). Early sedimentation studies showed that gene 5 protein sediments either as a monomer (1.3 s)or as a dimer (1.9 S), depending on the ionic composition of the solution (117). Increasing concentration of salt induces dissociation of the dimer, so only monomer is present at salt (NaCI or NaCIO,) concentrations exceeding -0.7 M (35).Estimates of the monomer-dimer association constant range from lo6 M-' in 0.15 M NaCl to lo3 M-' in 0.68 M NaCl. In contrast, a subsequent sedimentation equilibrium study showed that the dimeric species appears to be stable to dilution to concentrations as low as 0.075 mg/ml. Under the conditions used the dimer was also found to be stable to extremes of salt (0-0.5 M KCl), pH (5-1 l ) , and temperature (5' and 20") (122). Only at concentrations greater than - 1 mg/ml did some formation of tetramer ( 7 4 % ) become apparent. The protein remains dimeric in saturating amounts of d(pT),, but in the presence of d(pTIBgene 5 becomes tetrameric. This may be due to the binding of two protein dimers of this oligonucleotide, and as such may reflect the cooperative bonding to DNA seen with longer single-stranded lattices. h. Protein Structure. The successful crystallization and subsequent determination of the crystal structure of gene 5 protein to a resolution of 2.3 8, (123-13) makes possible an examination of molecular details at a level that is not accessible for any other DNA binding protein. The monomer of gene 5 protein has molecular dimensions of 45 A x 25 8, x 30 A. Its secondary structure consists entirely of antiparallel P-sheets, and contains no a-helical sequences whatsoever. This result is in agreement with sec-

- 10,000 (41, 117, 119). The complete amino acid

120.Y. Nakashima, A. K . Dunker, D. A. Marvin, and W. Konigsberg, FEBS ( F e d . Eur. Biockrm. S o c . ) Lett. 40, 290 (1974).

121. T. Cuypers, F. J. van der Ouderaa, and W. W. de Jong, BBRC 59, 557 (1974). 122. S . J . Cavalier, K. E. Neet, and D. A. Goldthwait, JMB 102, 697 (1976). 123. A. McPherson, I. J. Molineux, and A. Rich, JMB 106, 1077 (1976). 124. A. McPherson, F. Jurnak, A. Wang, F. Kolpak, 1. J. Molineux, and A. Rich, C S H S Q B 43, 21 (1979). 125. A. McPherson, F. A. Jurnak, A. H. J. Wang, I . Molineux, and A. Rich, J M B 134, 379 (1979). 126. A. McPherson, F. Jurnak, A. Wang, F. Kolpak, A. Rich, I. J. Molineux, and P. Fitzgerald, BJ 32, 155 (1980).

4 14

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

ondary structure estimates from circular dichroism (29), and with secondary structure predictions based on amino acid sequence (127). The secondary structure consists basically of three elements: A threestranded antiparallel P-sheet (residues 12-49), a two-stranded antiparallel P-ribbon (residues 5O-70), and a second two-stranded antiparallel ribbon (residues 7 1-82). The first P-ribbon is involved in the protein-protein interactions that maintain the dimer in solution, whereas the second P-ribbon is believed to participate in the nearest-neighbor interactions responsible for cooperative binding (126 ). Although the structure of the protein-nucleic acid complex has not been determined, studies on the unliganded gene 5 protein show that a 30 A-long groove exists in the molecule. It is believed that this groove comprises the DNA binding site, based on its size and shape as well as on the presence of amino acid residues in this region that have been identified in solution studies as being involved in DNA binding (see Section IV,B,2,d). A three-dimensional representation of gene 5 protein is shown in Fig. 5. 2. Protein-Nuclei Acid Interactions N. DNA Binding. The DNA binding properties of gene 5 protein were first analyzed using a sedimentation velocity technique (41, 117); these studies showed that gene 5 protein can saturate single-stranded DNA at a ratio of 1 protein monomer per 4 DNA nucleotide residues. It was also observed that a 10-fold increase in protein concentration dramatically increased the affinity of the protein for DNA, suggesting that binding is cooperative. From these data it was concluded that the affinity of the protein for a contiguous binding site must be at least 60-fold greater than that for an isolated site (41). In addition, under the conditions employed, gene 5 protein exhibited no affinity for double-stranded T4 DNA, for double-stranded (replicating form) fd DNA, or for ribosomal RNA. Filter binding has also been used to monitor the formation of the gene 5 protein-nucleic acid complex (117, 122, 128). A semiquantitative analysis of the filter binding data demonstrated that the binding of the protein to DNA is neither totally noncooperative nor infinitely cooperative (128). These results also suggested that the affinity of the protein for a contiguous binding site is -100Gfold greater than for an isolated site; this estimate of the cooperativity parameter is of the same magnitude as that measured for the T4 gene 32 protein.

-

127. R. A. Anderson, Y. Nakashirna, and J. E. Coleman, Biocheinistry 14, 907-917 (1475). 128. A. K. Dunker, FEES (Fed. Eur. Bioclrern. Soc.) Lett. 52, 323 (1975).

4 15

21. DNA BINDING PROTEINS

The optical properties of both the protein and the nucleic acid are altered upon complex formation. The intensity of a CD band that has been attributed to the interaction of tyrosine residues is reduced on nucleic acid binding: this change saturates at -4 nucleotide residues per protein monomer (29). The observed effects on the protein CD spectrum suggest that no major alteration in protein secondary structure accompanies DNA binding; however, as indicated above, the spectral changes that were seen strongly implicate alterations on binding in the environment of at least some tyrosine residues (29). Similar conclusions have been derived from an examination of the effect of DNA binding on the tyrosine fluorescence of the protein: a quenching of up to 70% is observed at a binding stoichiometry of 4 nucleotide residues per protein monomer (35). The optical properties of the nucleic acid showed changes in the CD and UV absorbance spectra of protein binding that are consistent with base unstacking (29, 127);again these effects saturate at a nucleotide residue to protein monomer ratio of -4 : I . The association constant under these low salt conditions has been estimated to be greater than lo8M-' for gene 5 protein binding to single-stranded DNA or RNA (127).As seen with gene 32 protein, this affinity is greatly reduced by increasing the salt concentration; the gene 5 protein-fd DNA complex dissociates completely at 0.5 A4 NaCl or 0.1 M MgClz. Binding to short oligonucleotides shows both base specificity and oligonucleotide length dependence ( 127, 129), with binding affinity increasing markedly with oligonucleotide lattice length to at least the 8-mer level. b. Melting of Double-Stranded D N A . Since gene 5 protein binds tightly to single-stranded DNA, but shows little or no affinity for double-stranded DNA under the same conditions, it should destabilize the duplex form of DNA at equilibrium. Melting profiles of DNA in the presence of excess gene 5 protein confirm this expectation; the T, values of poly(dA-dT), C . pe[fingeizs DNA, and T4 DNA are all lowered by -40" relative to those of the free DNA in 0.03M KC1 (41).Thus, unlike T4 gene 32 protein, fd gene 5 protein does not seem to be kinetically blocked with respect to melting duplex DNA. This DNA denaturation is relatively nonspecific with respect to base composition, but shows a slight preference for A-T rich DNA, which appears to be destabilized somewhat more than G-C rich DNA under the same conditions. The rate of DNA denaturation was found to be slow unless a small amount of M g + is present ( < l o mM), but higher concentrations of Mg2+ inhibit the reaction. Thus Mg2+ may play a special role in either the structure of the protein or of the protein-

-

129. J. E. Coleman, R . A . Anderson, R . Ratcliffe, and I. M. Armitage, 6iochetnisrr.v 15, 5419 (1976).

416

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

(A)

Z

\

*X '

THREE STRAND ANT I- PAR ALL E L GENE 5 PRODUCT

l.'

FIG.5 . (A) A schematic diagram that shows the three components of &structure that comprise the major part of the gene 5 protein molecule. The amino acid residues forming the three-stranded sheet are indicated; these amino acids are primarily involved in the interaction with single-stranded DNA. (B) This schematic three-dimensional drawing, showing the course of the gene 5 polypeptide backbone, is taken directly from the electron density map. Domain (I) indicates the three-stranded p-sheet, which forms the major part of the DNAbinding interface. Region (11) shows the two strands ofp-ribbon, which appear to be primarily responsible for maintaining the molecule as a dimer in solution by interaction with a symmetry-related p-ribbon. The second p-ribbon (111) is established diagonally across the p-sheet (I). This component may be the primary participant in the lateral interactions from which the cooperativity of the DNA binding arises. [After McPherson er ul., ( / 2 4 ) ] . DNA complex. Finally, in contrast to T4 gene 32 protein, no conditions could be found under which the gene 5 protein catalyzes the renaturation of denatured DNA (41); in this regard, gene 5 protein resembles the E. coli HDP. c . Electron Microscopy. The gene 5 protein-fd DNA complex appears in the electron microscope as a rigid rod interrupted by occasional branched structures (41). These images contrast strikingly with those observed for complexes with DNA of T4 gene 32 protein, or of the E. coli or calf thymus HDPs, which all form expanded open circles with singlestranded fd DNA. The internucleotide spacing in the gene 5 protein-

21.

417

DNA BINDING PROTEINS

(B)

FIG.5B.

fdDNA complex has been estimated to be at least 3.8 A; in contrast the spacing for the T4 gene 32 protein-DNA complex is 5.3 %i per nucleotide residue, and -2.1 A per residue for the E. coli SSB protein-DNA complex (27).The linear rodlike structure of the gene 5 protein-DNA complex seems to reflect the lateral association of two regions of protein-covered DNA, stabilized by the formation of back-to-back dimers. This complex, formed in vitro with fd DNA, is 100 8, wide and 7300 A long, and is wound into an overall helix with a longitudinal repeat distance of 65-70 A. Each turn of the helix contains at least six gene 5 protein dimers; we note that the asymmetric unit of gene 5 protein cocrystallized with oligonucleotides also contains 6 protein dimers (126). Studies of complexes formed in vivo, however, show some differences from those formed in vifro. The in vivo complex consists of fibers -40 8, in

-

418

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

-

width, which are supercoiled to an overall width of 160 8, with a supercoil repeat of 160 A; the length of these complexes is 1.1 p m (130). It has been observed that the binding stoichiometry of the complexes isolated in vivo differs from that of those formed in vitro ( n = 4.7 and 4.0 nucleotide residues, respectively); this difference is not thought to be due to selective losses of protein in the isolation process (35).Also the in vivo complex appears to be more stable to salt dissociation. However, the addition of excess protein to the irz vivo complex seems to result in the formation of the irz vifro complex, as defined by sedimentation properties or stoichiometry. Whether these differences reflect isolation artifacts, are due to the unique ionic environment of the cell (e.g., critical concentrations of Mgz+, spermidine, etc.), or represent different equilibrium or kinetic forms generated because of limited availability of protein during complex formation in vivo, remains to be seen. d. Chemicul Modificution. The spectral results that appeared to implicate tyrosine residues in gene 5 protein-nucleic acid complex formation have been followed up by chemical modification studies (127). Reaction of gene 5 protein with tetranitromethane results in the nitration of three of the five tyrosine residues, and in a greatly reduced affinity of the modified protein for DNA (127). However all of the tyrosines are protected from the reagent if the protein is complexed with DNA prior to modification. The three tyrosine residues that are modified were identified as residues 26,41, and 56. The presence of these residues on the surface of the protein has been confirmed by solvent perturbation (35) and by NMR studies (see next section). Lysine residues also play a role in stabilizing the protein-nucleic acid complex, since reaction of the six lysine residues in gene 5 protein with N-acetylimidazole also abolishes the ability of the protein to bind DNA. Here, however, the residues are not protected against reaction by DNA binding (127). Gene 5 protein also carries a single cysteine residue that is resistant to modification by DTNB unless the protein has been previously denatured (35,127). However, this sulfhydryl is accessible to Hg2+,and modification by this agent prevents binding of the protein to DNA. Conversely, complexation with DNA protects the sulfhydryl from modification (127). Approximately 30% of the protein in an in vivo gene 5 protein-DNA complex can be covalently crosslinked to DNA by irradiation with UV light (131, 132). The covalent complex that results from irradiation of the

-

-

130. D. Pratt, P. Laws, and J. Grifiith, JMB 82, 425 (1974). 131. E. Anderson, Y. Nakashima, and W. H. Konigsberg, Nucleic Acids Res. 2, 361 (1975). 132. L. Lica and D. S. Ray,JMB 115, 45 (1977).

21. DNA BINDING PROTEINS

4 19

in vivo complex has been isolated, and the site of crosslinking has been

shown to be between residues 70 and 77; it has been suggested that Ser-75, Gly-74, Phe-73, and Gly-71 might be the actual amino acid residues involved (133). In contrast, irradiation of the in vitro complex results in crosslinking at residue Cys-33, with an efficiency of crosslinking as high as 21% for poly(dT) (131). Whether this result represents another reflection of structural differences between in vivo and in virro complexes is unclear. e. N M R Srudies. The role of aromatic residues in D N A binding has been extensively studied using NMR techniques, and on the basis of these results it has been suggested that intercalation of two tyrosines and one phenylalanine are involved in the recognition of the nucleic acid ( 1 7 , 129, 134437). l9F-NMR has been used to probe the role of these protein residues in binding, 31P-NMR has been employed to study the effects of complex formation on polynucleotide backbone structure, and 'H-NMR has been utilized to investigate the structural changes in both the protein and the nucleic acid accompanying complex formation. The fluorotyrosine derivative of gene 5 protein can be prepared by infection (with the wild-type bacteriophage) of E. coli tyrosine auxotrophs that have been grown in a medium containing rn-fluorotyrosine ( 1 7 , 129). This fluorotyrosine-substituted gene 5 protein has binding properties similar to those of native protein. The IYF-NMR of this protein shows 5 resonances that correspond to the 5 tyrosine residues within the primary structure (127, 129). Three of the five resonances are located at spectral positions corresponding to free m-fluorotyrosine; the remaining two resonances are shifted downfield. The three upfield resonances have been assigned as sutfwe residues (i.e., those accessible to nitration). This assignment is based on their spectral location, on their perturbation upon nucleic acid binding, and on the fact that they show a greater degree of rotational mobility than the other resonances as determined by the effect of proton decoupling on the intensity of the fluorine spectrum (nuclear Overhauser effect)( 13.5). Based on similar reasoning, the two downfield resonances have been assigned as buried tyrosine residues. Upon formation of a complex with either d(pT), or d(pAIB,the 19Fresonances of fluorotyrosine gene 5 protein show two general characteris133. P. R . Paradiso, Y. Nakashima, and W. H . Konigsberg, JBC 254, 4739 (1979). 134. G. J. Garssen, C. W. Hilbers, J. G. G . Schoenmakers, and J . van Boom,EJB 81,453 ( 1977). 135. J . E . Coleman and I . M. Armitage, Biochemi.siry 17, 5038 (1978). 136. G. J. Garssen, R . Kaptein, J. G . G. Schoenmakers, and C. W. Hilbers, P N A S 75, 5281 (1978). 137. G. J . Garssen, G . I . Tesser, J . G . G. Schoenmakers, and C. W. Hilbers, BBA 607, 361 (1980).

420

S . KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

tics: The downfield buried peaks shift slightly downfield, whereas the upfield surface resonances shift upfield (135). Two of the upfield peaks show chemical shifts upon complex formation, the magnitudes of the shifts depending on the base composition of the nucleic acid; a shift of -0.4 ppm is observed with d(pT), and a shift of -0.7 ppm with d(pA),. This upfield shift is of the proper magnitude and direction to correspond to the nucleotide-base-induced ring current shifts that would be expected if intercalation of tyrosine residues between bases of the nucleic acid were taking place. It has therefore been suggested that two of the three tyrosine surface residues intercalate between nucleic acid bases upon complex formation (129, 135). Studies involving proton NMR of gene 5 protein have confirmed the results based on lgF-NMR, and have provided a more detailed picture of the intercalation mechanism (129, 134437). These studies have shown that 30 to 40% of the aromatic protons of gene 5 protein show upfield shifts upon complex formation with oligonucleotides (129). The upfield shifts are on the order of 0.3 ppm for d(pT),, and -0.8 ppm for d(pA),, which are similar to the values observed in the l9F studies. Concommitantly with these changes, the C-6 proton resonances of the thymidine residues also shift upfield by 0.1 to 0.2 ppm. This observation is also consistent with intercalation, and could represent the expected upfield shift in the nucleotide base protons due to ring current effects from the aromatic residues of the protein. Closer examination of the NMR proton spectra of gene 5 protein has shown that the 3,5-ring protons of the tyrosines are more affected by complex formation than the 2,6 protons (134-1-37), suggesting that only the leading edge of the tyrosine residues is inserted between the bases. In addition to the effect on tyrosine protons that accompanies binding, it is clear that phenylalanine protons are also shifted upfield, and that these shifts are dependent on the base composition of the oligonucleotides of the complex. Thus, in addition to the two tyrosine residues, at least one phenylalanine is also involved in the intercalation process (129, 137). Chemically induced dynamic nuclear polarization NMR has also been used to support the conclusions cited above on the basis of conventional NMR and chemical modification data (136). This technique has shown that tetranucleotide binding shields three tyrosine surface residues from reaction with a flavin dye. On the other hand, the dinucleotide r(A)* seems to protect only one residue when bound. Thus the decreased affinity of the shorter oligonucleotides (see above) may reflect the fact that the dinucleotides can interact with only one aromatic residue rather than with three.

21.

DNA BINDING PROTEINS

42 1

Proton NMR has also been used to investigate the effects of nucleic acid binding on nonaromatic amino acid residues (129). The methyl protons of the €-carbon of lysine, for example, show little chemical shift or line broadening upon complex formation, even though it is clear from chemical modification data that these groups are important in DNA binding. It has been concluded, since these groups show a large degree of rotational freedom on the NMR time scale, that they do not form rigid salt bridges with the phosphates of the nucleic acid, but are involved in a more delocalized binding interaction. The resonances due to the arginine 6-CH2 groups also show chemical shifts and/or broadening upon oligonucleotide binding (129). These changes may be due to a direct interaction of the amino acid residues with the phosphate backbone, but since the changes in the resonances differ if a tetranucleotide is used instead of an octanucleotide, these effects could also reflect a more general change in protein structure. This latter interpretation is consistent with the distinct changes that are observed in the aliphatic region of the spectrum and are dependent on the size of the oligonucleotide. The structure of the nucleic acid backbone in the gene 5 protein complex has been studied using 31P-NMR (129, 134). The diester phosphate resonances are shifted very little, indicating that gene 5 protein binding probably does not alter the conformation of the sugar-phosphate backbone appreciably. However, the fact that the resonances are broadened suggests that the phosphates are rigidly held in the complex. In addition, the apparent pK of the 5’-terminal phosphate is shifted in the complex, suggesting that the group is in close proximity to a positively charged amino acid residue of the protein (134). f. Correlatioti of the Solution Results with the Protein Structure. The 30 8, cleft in the uncomplexed gene 5 protein structure, which has been proposed as the site of DNA binding, consists primarily of residues 12-49, 50-56, and 66-69 (126). The aromatic amino acids that are present in this region are the surface tyrosine residues 26, 41, and 56, as well as the buried Tyr-34, in addition, phenylalanine residues 13 and 68 are nearby. Of these residues only Tyr-56 and Phe-68 appear to point into the binding groove, but the other residues can be brought into the cleft by rotation about their respective P-carbon atoms. All of these residues except Tyr-26 are clustered at one end of the binding cleft, and the side chains of Tyr-41, Tyr-34, and Phe-13 form a triple-stacked structure in which the residues are arranged in a fanned-out array. These tyrosines are probably located in the upfield region of the NMR spectrum, and may also be responsible for the protein circular

422

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

dichroism band at 228 nm, which has been attributed to Tyr-Tyr interactions. Binding to DNA results in a reduction of this signal by -30%, perhaps reflecting an unstacking of the residues. Cys-33 is also located on the inside of the DNA binding groove. This position is consistent with the fact that the bulky reagent, DTNB, is not able to penetrate to this residue, while Hg2+can make contact. The fact that DNA binding prevents mercuration of Cys-33, and that this residue may be the site of UV-induced crosslinking to the nucleic acids (133), provides additional support for the identification of this region as the DNA binding site. The distribution of the charged and uncharged residues within the binding groove is also noteworthy. Thus, while aromatic residues are located predominantly along the outside edge of the cleft, positively charged residues (particularly Arg-21, Arg-80 and Arg-82 and Lys-24 and Lys-46) are located within the groove. This suggests that the phosphate backbone of the nucleic acid is aligned and drawn into the binding region by these charged residues, and that subsequent stabilization of the complex may occur through interaction with the aromatic amino acids that rotate into position to react with the polynucleotide bases. Final confirmation of this scheme must await direct high resolution X-ray data on crystals of gene 5 protein-nucleic acid complexes. The limited data presently available from such complexes indicate that the asymmetric unit consists of six gene 5 protein dimers. By slightly distorting the observed cylinderical arrangement into a helix, McPherson et ul. (126) have proposed a structure for the gene 5 protein-DNA complex in which DNA strands bound to the gene 5 protein dimers are -25 8, apart; the structure has a linear repeat of 80-90 A and a diameter of -100 A. This proposal is in reasonable accord with some of the electron microscopic studies of gene 5 protein-DNA complexes (41).

3. Biological Roles u. Replication. The main function of gene 5 protein in the life cycle of the filamentous phages is to control the switch from RF DNA replication to the replication of the single-stranded viral genome (118, 138-140). It has been clearly demonstrated, using phages that carry temperature-sensitive mutations in gene 5 , that this protein is essential to the maintenance of the 138. B . J. Mazur and P. Model, J M B 78, 285 (1973). 139. K. Geider, and A. Kornberg, JBC 249, 3999 (1974). 140. A. Kornberg, “DNA Replication.” Freeman, San Francisco, 1980.

21. DNA BINDING PROTEINS

423

production of single-stranded viral DNA (118 1; at nonpermissive temperatures with such mutants, single-stranded viral DNA is lost by continued synthesis of the double-stranded replicative form. The free concentration of gene 5 protein in the cell may determine whether replicative form or single-stranded DNA is synthesized. This has been shown in phage in which DNA replication is blocked by a temperature-sensitive mutation in gene 2 (138).At nonpermissive temperatures gene 5 protein synthesis continues at a constant rate; on shifting to permissive temperatures only viral single-stranded DNA is accumulated, and no RF form can be detected. 111 v i m experiments have also confirmed the iiz viva observation that gene 5 protein inhibits the replication of viral single-stranded DNA (1.79). An iiz vitvo E . coli replication system has been established that can convert viral DNA into RF DNA using purified DNA polymerase I and 11, RNA polymerase, E. coli HDP, and DNA ligase. It was observed that purified M13 gene 5 protein could not substitute for the E. coli HDP: in fact the addition of gene 5 protein virtually abolished the reaction when present in excess of one protein per 4 nucleotide residues. Only at very low levels did gene 5 protein show a positive effect; under these conditions the protein appeared to stimulate the synthesis of RNAprimed MI3 DNA. It is clear that although the fd gene 5 product is a single-stranded DNA binding protein, its physiological role is very different from that of T4, T7, and E. coli (see the following sections) HDPs. The gene 5 protein plays a special role in the life cycle of the filamentous phages; unlike the HDPs previously listed, it acts as a specific switch in directing the DNA replication of this phage. In addition, it may be responsible for “prepackaging” the newly synthesized viral DNA and protecting it from nucleases until it reaches the cell membrane, at which point the gene 5 protein is displaced by the fd gene 8 coat protein.

C. Escherichim coli SINGLE-STRANDED DNA BINDING PROTEIN In an attempt to determine whether a protein analogous to the T4-coded gene 32 protein exists in uninfected E . coli cells, Sigal et ml. (96) chromotographed crude protein extracts of E. coli on denatured DNA cellulose columns. A single-stranded DNA binding protein with a monomeric molecular weight of about 20,000 was discovered and purified to homogeneity (see Section 111). Subsequently this protein has been

424

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

shown to be necessary for DNA replication (96. 141-143), and has been implicated in recombination and repair (144-148). About 300-800 copies are present per bacterial cell ( 9 6 , 142). This protein has unfortunately been called by several different names, including “DNA unwinding protein” [Chapter 18, this volume, and Ref. ( 9 6 ) ] ,“DNA binding protein” ( 1 4 9 ) , “ E . coli helix destabilizing protein I” (I), and “ E . coli single-strand binding protein (SSB)” (148). We call this protein the E. coli SSB protein in accord with the nomenclature that has been used by the authors of the more recent papers, and in consideration of the fact that the gene for this protein has now been formally named ssb (150). 1. Physical Properties of the Protein a. Size and Structural Features. The E. coli SSB protein in its native form exists as a tetramer of four identical subunits (151 ). Each monomeric subunit has a molecular weight of about 20,000, with exact estimates (based on sodium dodecyl sulfate gel electrophoresis) ranging from 18,500 (142) to 22,000 (27. 191). From amino acid composition the molecular weight is estimated at 19,500 (142). The sedimentation coefficient of the native protein is -4.8 S (96, 142, 152) corresponding to a tetramer of total molecular weight 76,000. The C.coli SSB protein (27, 73, 142) contains a rather large number of acidic residues. The isoelectric point was determined to be 6.0 2 0.1 (142). Based on circular dichroism spectra, it has been estimated that E . coli SSB protein is about 20% a-helix and 20% /3-sheet (27). 6 . Spectroscopic Properties. The molar extinction coefficient (at 280 nm) for E . coli SSB, based on a molecular weight of 20,000 is 3.0 x lo4 M - ’ cm-’ (153). The protein displays a fluorescence spectrum characteristic of many tryptophan-containing proteins; the wavelength of maximum excitation is 285 nm and the wavelength of maximum emission is 345 nm.

-

141. T. A. Kunkel, R . R. Meyer, and L. A. Loeb, /“AS 76, 6331 (1979). 142. J . H. Weiner, L. L. Bertsch, and A . Kornberg, JBC 250, 1972 (1975). 143. J. F. Scott, S . Eisenberg, L. L. Bertsch, and A . Kornberg, PNAS 73, 1594 (1977). 144. J. Tomizawa, N . Anraku, and T. Iwama, JMB 21, 247 (1966). 145. M. W. Baldy, Virology 40, 212 (1970). 146. B. F. Johnson, M o l . Gen. Genet. 157, 91 (1977). 147. J . Glassberg, R. R. Meyer, and A . Kornberg, J . Bocteriol. 140, 14 (1979). 148. R . R. Meyer, J. Glassberg, and A. Kornberg,PNAS 76, 1702 (1979). 149. I. J. Molineux and M. L. Gefter, PNAS 71, 3858 (1974). 150. B. J. Bachmann and K. B. Low, Microbial. Rei,. 44, 1 (1980). 151. I. J. Molineux, S. Friedman, and M. L. Gefter, JBC 249, 6090 (1974). 152. W. T. Ruyechan and J. G. Wetmur, Biochemistry 15, 5057 (1976). 153. W. T. Ruyechan and J. G. Wetmur, Biochemistry 14, 5529-5534 (1975).

21.

DNA BINDING PROTEINS

425

The binding of polynucleotides toE. coli results in up to 70% quenching of intrinsic protein fluorescence, unaccompanied by any shift in the wavelength of maximum emission (36). c. Aggregation State and Heat Stabilio. The SSB protein exists as a tetramer within the protein concentration range of 75-750 pglml (151) at ionic strengths as high as 1.0 M (152). TheE. coli SSB protein also appears to be remarkably heat stable. It has been reported that the protein loses none of its known activities as a consequence of heating to 100" for 2 minutes (142).This property has been exploited to purify the protein away from other proteins that precipitate upon heating. Even a temperaturesensitive mutant of the SSB protein is heat resistant and will function normally when cooled from 100" to the permissive temperature (148). However, preliminary differential scanning calorimetry studies have indicated that the wild-type protein undergoes an irreversible transition at about 70" when left at high temperature for an extended period of time (73). It is therefore important to exercise care in the use of heating in preparative protocols for the SSB protein. 2. Protein-Nideic Acid Interactions

The binding ofE. coli SSB protein shares many of the characteristics of the T4 gene 32 protein interaction with DNA; most importantly, like gene 32 protein, and SSB protein binds preferentially and cooperatively to single-stranded nucleic acids (96). In contrast to gene 32 protein, however, the SSB protein can, under some circumstances, denature doublestranded DNA. In addition, unlike gene 32 protein, which binds to the DNA as a monomer (and fd gene 5 protein, which seems to bind as a dimer), the SSB protein probably binds as a tetramer. a . Binding to Single-Stranded Nucleic Acids. It has been demonstrated that SSB protein displays a large preference for single-stranded DNA, and shows no detectable binding to native duplex DNA or bacteriophage R17 RNA (36, 96, 142). Chemical modification studies (154) have strongly implicated lysine residues in the binding of SSB protein to single-stranded DNA, and indeed, in common with the other proteins discussed in this review, a significant salt dependent of the interaction is seen. SSB protein elutes from denatured DNA-cellulose columns at between 1 and 2 M NaCl (96).A number of groups have reported that the affinity of the protein for both poly- and oligonucleotides decreases significantly as the ionic strength increases above 0 . 1 - 0 . 2 M in NaCl(36, 142,151453). Chemical modification studies 154. P. K. Bandyopadhyay and C. W. Wu, Biochemisiry 17, 4078 (1978).

426

S . KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

with reagents specific for cysteine, arginine, and tyrosine seem to rule out the significant participation of these residues in the interaction of the protein with polynucleotides (154). In addition to (monovalent salt concentration, both pH and Mg" concentrations also affect the SSB protein-nucleic acid interaction. Binding is strongest between pH 7.5 and 8.5, and the strength of the interaction appears to fall off rapidly below pH 6.5 and above pH 9.0 (152). Mg2+ concentrations above 10 mM decrease significantly the affinity of SSB protein for single-stranded oligo- and polynucleotides; thus most in vitro replication assays are carried out in the 1-10 mM Mgz+ concentration range, One should be careful to distinguish this effect of Mg2+from the Mg"-induced stabilization of secondary structure in single-stranded DNA and RNA. In the latter case a decrease in the binding of the protein to nucleic acid occurs as a consequence of the stabilization of competing hairpin (duplex) structures, and this decreases the concentration of single-stranded nucleic acid available for binding. The effect ofE. coli SSB protein on base stacking interactions in singlestranded DNA has been examined by circular dichroism (27), sedimentation (961, and electron microscopic techniques (96). Circular dichroism spectra indicate that the protein unstacks the bases in a manner similar to that due to thermal denaturation, and density gradient centrifugation analyses suggest that the DNA is also held in a more open conformation, thus increasing the length of the polynucleotide backbone. However, electron microscopy studies of glutaraldehyde-fixed complexes are consistent with a 35% decrease in internucleotide spacing. This is to be contrasted to a 50% increase in length for DNA complexed to gene 32 protein (79); EM artifacts of fixation or spreading may account for this discrepancy. b. Binding Specificity. The interaction of E. coli SSB protein with synthetic polynucleotides has been studied to look for base and sugar composition-based differences in binding affinity. Gel filtration studies (143) lead to the following order of SSB protein-polynucleotide affinities: poly(dT) > ssDNA (4x174) > poly(rU) > poly(dA) > dsDNA (T4) > poly(rA). By monitoring quenching of protein fluorescence, a somewhat different series was obtained: poly(dT), poly(dA) > ssDNA (fd) > poly(rU) = poly(dC) > poly(rA) > tRNAp'" > poly(rC) + dsDNA (P22) (36). Obviously genera1 preferences can be seen for single-stranded over double-stranded DNA, and for deoxyribose-containing over ribosecontaining polynucleotides. Also poly(dT) seems to be bound more tightly than all other polynucleotides; these general features are also seen with gene 32 protein. Little, if any, nucleotide composition or sequence dependence is observed for the binding of hexanucleotides to SSB protein.

427

21. DNA BINDING PROTEINS

c. Dependence oj' Binding Aj'finity on Oligonrrcleotide Lattice Length. Equilibrium dialysis experiments (152) have suggested that the apparent binding affinity of SSB protein for various multimers of d(pCpT) depends on the length of the overall oligonucleotide lattice. The apparent binding constant to such oligomers increased -10-fold in going from d(pCpT)2to d(pCpT),, and only -2-fold further in going from d(pCpT)3to d(pCpT)6+. In partial contrast to these results, intrinsic protein fluorescencequenching studies have shown that d(pT),6 binds SSB protein -200-fold more strongly than dT-containing oligomers 8 residues in length or less (36). The longer oligomers quench protein fluorescence much more, effectively than the shorter lattices. These results are also in general accord with those observed with gene 32 protein. d. Polynucleotide Binding Site Size and the Oligomeric State of SSB Protein. A variety of techniques have been used to determine the site size ( n ) of SSB protein for binding to polynucleotides; most approaches have yielded values ofti of -8 nucleotide residues per protein monomer (36, 96, 143). The molecular significance of this parameter has been hard to define because neither the state of aggregation of the protein in its binding form, nor the number of binding sites utilized per bound protein, have been established unequivocally. EM studies have suggested that the protein binds to single-stranded DNA as a tetramer, or alternatively, that it aggregates to tetramers on binding (96). It was also demonstrated by sedimentation experiments with oligo d(pCpT),-, that no change in the state of aggregation occurs upon binding (152). This result has been confirmed by measurements of the rotational correlation time of the free protein and the complex by time-dependent emission anisotropy ; no significant change in the protein structure upon binding to oligo(dT),, was revealed (154).In light of these results, a site size of about 32 nucleotides per tetramer may be a more physically significant way to view the binding of SSB protein to single-stranded DNA. More detailed work on the topography of this complex is obviously required. e. Binding Cooperativity. The cooperative nature of the binding of E. coli SSB protein to single-stranded DNA and to synthetic polynucleotides has been demonstrated by several different techniques, including electron microscopy (96, /S3), gel filtration (142), density gradient centrifugation ( 9 6 ) , and fluorescence quenching ( 3 6 ) . By comparing the binding constants for the interaction of SSB protein with pT8 and pT16, it was e d mated that the value of the cooperativity parameter is at least 50 on the basis of fluorescence measurements ( 3 6 ) , whereas on the basis of electron microscopic studies, w was estimated to be 10' (153). The estimate based on fluorescence is only a lower limit; however, the value obtained by electron microscopy may be high due to artifacts in the technique. Thus

-

428

S . KOWALCZYKOWSKI, D . BEAR, A N D P. VON HIPPEL

the use of glutaraldehyde in the fixation of the complexes prior to spreading could have artificially increased the size of the protein clusters beyond equilibrium expectations. Such artifacts have been seen in EM estimates of cooperativity for gene 32 protein (79). An accurate measure of w for SSB protein, obtained under physiological conditions, is not available. In addition, preliminary evidence (1.33) suggests that, unlike the situation with gene 32 protein, w for SSB protein may be salt concentration dependent. f. Dennturution and Renaturution of Duplex D N A . As noted above, E. coli SSB protein at low ionic strength can slowly denature duplex DNA (96). Surprisingly, this effect has not been heavily investigated, in contrast to the denaturation induced by the gene 32 protein *I fragment. Upon addition of E. coli SSB protein, it was found that denaturation of duplex DNA goes to completion in 12 to 30 min at 37” for duplex DNA; the DNA has an unperturbed T,,,Of 55” under the same solvent conditions (96). The effect is stoichiometric rather than catalytic; however, the quantitative aspects of the stoichiometry have not been characterized. In electron microscopic studies involving glutaraldehyde-fixed protein-DNA complexes it was found that the E. coli SSB protein attacks A-T rich regions, and tends to expand denatured loops, rather than to initiate new ones. The addition of Mg2+completely reverses the denaturation effect. Several groups have reported little or no affinity of E. coli SSB protein for double-stranded DNA. Also the addition of 5 mM spermidine to denatured DNA strongly decreases the affinity of the protein, presumably by stabilizing the secondary structure of the nucleic acid. I t has also been demonstrated that E. coli SSB protein does not retain duplex DNA on nitrocellulose filters (142). The binding affinity of SSB protein to duplex DNA has been estimated to be at least 3 orders of magnitude weaker than that to denatured DNA (36). These data suggest that, like the carboxy-terminal cleaved gene 32 protein (G32P*I), E . coli SSB protein binds to and traps sequences of DNA transiently opened by thermal fluctuations. Since A-T rich sequences are less stable with respect to thermal melting than those rich in G-C base pairs, the former are presumed to be prime candidates for such nucleating interactions. The protein, due to its significant cooperativity, would then tend to bind at contiguous sites and expand the denatured regions. The biological significance of such SSB protein-driven melting processes is questionable, due to the sensitivity of melting to the concentration of mono- and polyvalent ions. We return to this question in the following sections. It has been reported that under certain conditions E. coli SSB protein

21.

D N A BINDING PROTElNS

429

can catalyze the renufuration of denatured DNA (155). In the presence of NaCl(0.2 M or less), and in the presence or absence of MgC12,E. coli SSB protein does not significantly affect the rate of renaturation of h DNA at pH 7.0. However, in the presence of 2 mM spermidine or spermine the renaturation rate is enhanced by a factor of about 5000. This rateenhancement effect requires saturating amounts of SSB protein, and its efficiency increases with the length of the DNA. Because of the complex nature of the dependence on counterions and pH, the simple removal of intrastrand hairpin loops by SSB protein binding does not appear to fully explain this effect. 3. Biological Roles

a. Replication. For several years after the initial discovery of E. coli SSB protein, the effects of this protein on the in virro replication of viral DNA by various viral and host-coded polymerases has been heavily investigated. However, it was not until a temperature-sensitive mutant was isolated (148) and found to be defective in DNA replication that one could be sure of the importance of E. coli SSB protein in phage and cellular metabolism in vivo . The mutation (designated ssb- 1) has been localized at 90 to 91 minutes on the E. coli linkage map. In virro studies have demonstrated that the E. coli SSB protein specifically stimulates polymerase 11-directed DNA synthesis on various phage templates (96, 150). This stimulation appears to be dependent on the ratio of binding protein to DNA, and independent of polymerase concentration, suggesting that the primary effect is at the DNA level (151). The addition ofE. coli SSB protein to in vitro replication systems has a multitude of effects, including (i) stimulation of the initiation of RNA primer-directed DNA synthesis by polymerase I11 holoenzyme on singlestranded phage DNA templates (139, 142), (ii) stimulation of the elongation rate of DNA synthesis by pol I11 (142), (iii) increase in the fidelity of DNA synthesis by pol I11 ( / 4 / ) ,and (iv) stimulation of the initiation and elongation of DNA synthesis by pol II on gapped and single-stranded templates (96,150, 156). The protein is required in an in virro polymerase I11 replication mix in order to convert bacteriophage G4 single-stranded DNA into its replicative form (142); and also to convert the replicative form back to single-stranded circular DNA (143). All these activities may be explained by the ability of SSB protein to bind to single-stranded DNA, trapping the DNA in the open form and 155. C. Christiansen and R . L. Baldwin, JMB 115, 441 (1977). 156. 1. J. Molineux and M. L. Gefter,JMB 98, 811 (1975).

430

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

melting out double-stranded regions by contiguous cooperative binding. This makes the bases more accessible as templates for polymerases, and thus may increase the fidelity of DNA synthesis as well as the elongation rate. However, there is ample evidence to suggest that E. coii SSB protein also interacts specifically with polymerases I1 and 111. The SSB protein also inhibits the exonuclease activity of pol I and the T4 DNA polymerase, and the activity of the single-stranded DNases from Aspergillus and Neurospera crassa . In contrast it does not inhibit, and even stimulates to a small extent, the exonuclease activity of T7 DNA polymerase, pol 11, and exonuclease I(156). Both pol I1 and exonuclease I form complexes with SSB protein. While no complex appears to be formed between SSB protein and T7 polymerase, it has been shown that SSB protein can substitute for T7 DNA binding protein (157) in a T7 replication system. Other DNA binding proteins, such as fd gene 5 , T7 DNA binding protein, and T4 gene 32 protein, cannot substitute for SSB protein in stimulating pol I1 (157). Thus the interaction of E. coli SSB protein with pol I1 is probably physiologically important, whereas the role of the DNA binding protein in the T7 replication system may be to interact primarily with the DNA. Demonstration of meaningful specific interactions with pol I11 must await a complete characterization of the holoenzyme [see Chapter 3, this volume and Ref. (140) for a review of this aspect]. Antibody titration studies have shown that there about 300 copies ofE. coli SSB protein tetramer per log phase bacterial cell (142). Based on the binding information previously presented, and the fact that there are approximately six replication forks per cell (96),we can calculate that -1600 nucleotides of DNA are covered per replication fork. Both pfasmids and transducing phages carrying the ssb gene have been used to overproduce SSB protein (42). Thus it is unlikely that the synthesis of this protein is autogenously regulated in vivo . b. Recombination and Repair. Further characterization of the ssb- 1 mutant (148)has revealed that the mutant strain is about one-fifth as active in recombination as the wild type, and extremely sensitive to UV damage. The lexC gene proposed as a regulator of UV and X-ray inducible repair (146, 158) has been tentatively found to be allelic withssb. Both genes map at the same locus, and SSB protein from a temperature-sensitive lexC mutant is temperature-sensitive when tested as an accessory protein in bacteriophage G4 replication. While an exact role for the E. coli SSB protein in recombination and repair has yet to be established, it has been demonstrated that SSB protein catalyzes recA -mediated single-stranded DNA 157. R. C. Reuben and M. L. Gefter, JBC 249, 3843 (1974). 158. J. Greenberg, L. J. Berends, J. Donch, M. H. L. Green, Genet. Res. (Cambridge) 23, 175 (1974).

21.

DNA BINDING PROTEINS

43 1

assimilation into homologous double-stranded DNA in vitro (159, 160). The concentration of recA protein required for such D-loop formation is also reduced in the presence of SSB protein. D.

BACTERIOPHAGE T7-CODED SINGLE-STRANDED DNA BINDING PROTEIN

1. Protein Properties and Interactions with DNA In searching for an analogue of T4-coded gene 32 protein that might serve similarly in phage T7 replication, two groups independently discovered a T7-coded, single-stranded DNA binding protein that stimulates replication by T7 DNA polymerase on either single-stranded or gappedduplex DNA templates (16l, 162). Although the original estimates of molecular weight of the two proteins differ considerably [31,000 (161) versus 25,000, (162)], the proteins appear to be the same on the basis of their other properties. Like gene 32 protein, the T7 HDP does not bind to duplex DNA and carries a net negative charge at neutral pH. NaCl concentrations in excess of 0.25 M are required to elute this protein from DEAE-Sephadex; under the same conditions E. coli SSB protein (PI = 6.0) is eluted at -0.15 M NaCI. The protein appears to be monomeric in 2 M NaCI, as judged by its behavior on gel filtration columns. The T7 HDP denatures duplex poly(dA-dT) (162). We estimate, on the basis of incomplete data (1621, that the T7 protein lowers the T, of poly(dA-dT) by about 40" in 0.04 M NaCI; this AT, is very comparable to that induced by T4 gene 32 protein under the same conditions. The T7 protein also stimulates the exonuclease activity of T7 DNA polymerase (the T7 gene 5 protein-E. coli thioredoxin complex) on a duplex DNA template (163). These rather scanty data suggest that the nucleic acid binding properties of the T7 protein may be quite similar to those of the other prokaryotic single-stranded DNA binding proteins described in the preceding sections. 2. Biological Roles

The biological properties of the T7 single-stranded binding protein also resemble those of the analogous T4 and E. coli proteins. The T7 protein 159. K. McEntee, G. M. Weinstock, and I. R. Lehman, PNAS 77, 857 (1980). 160. T. Shibata, C. DasGupta, R. P. Cunningham, and C. M. Radding, PNAS 77, 2606 (1980). 161. R. C. Reuben and M. L. Gefter,PNAS 70, 1846 (1973). 162. E. Scherzinger, F. Litfin, and E. Jost, Molec. Geti. Getter. 123, 247 (1973). 163. K . Hori, D. F. Mark, and C. C. Richardson,JBC 254, 11598 (1979).

432

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

stimulates the polymerization activity of T7 DNA polymerase on singlestranded templates 10- to 15-fold at low temperatures; under these conditions the T7 DNA polymerase is otherwise quite ineffective (161). R e sumably this stimulation reflects the denaturation of DNA hairpins, which are stable under these conditions, and which, if present, inhibit the polymerase. The T7 protein does not, however, stimulates T4 DNA polymerase or E. coli DNA polymerases I, 11, or I11 (161). In spite of this demonstration of functional specificity, no physical interaction between the T7 DNA binding protein and T7 polymerase has been observed (162). In contrast, the T4 gene 32 protein and the E. cofi SSB protein do bind preferentially to their homologous polymerases in free solution. In vitro, E. coli SSB protein can substitute for the T7 protein with regard to its stimulatory effect on the activity of T7 DNA polymerase (162). Thus, despite the fact that the T7 SSB protein seems to be specific (i.e., it can only stimulate its homologous DNA polymerase), the T7 DNA polymerase can be stimulated by both E. coli and T7 SSB proteins. This may also explain why mutations in T7 SSB protein are not lethal to T7 growth; presumably the host SSB protein substitutes for the phage protein, thus “rescuing” the mutant phage.

E. EUKARYOTIC SINGLE-STRANDED DNA BINDING PROTEINS A large number of proteins that show affinity for either single-stranded or double-stranded DNA have been isolated from a variety of eukaryotic organisms [for reviews, see Refs. (/64-166)]. In general, these proteins have been isolated from cell extracts on the basis of binding to DNAcellulose columns: in the absence of genetic information and mutants the physiological role of many of these proteins has been difficult to assess. The proteins that are included in the following sections have been selected for discussion because they show (i) binding to DNA, (ii) a preferential affinity for single-stranded over double-stranded DNA (thus they are, at least potentially, HDPs), and (iii) a presumed biological activity (usually manifested as a stimulatory effect in an in vitro replication system with the homolgous DNA polymerase). As a consequence many interesting proteins have been omitted, but we hope that those discussed will prove to be representative of this potentially important class. Some of the more important properties of these proteins are summarized in Table 111. 164. J. E. Coleman and J. L. Oakly, Cril. Rev. Biochem. 7, 247 (1980). 165. J. J. Champoux, Annu. Re\’. Biochem. 47, 449 (1978). 166. A. Falaschi. F. Cobianchi, and S. Riva, Trends Biochem. Sci. 5, 154 (1980).

21.

DNA BINDING PROTEINS

433

1. The Lilium (Lily)DNA Binding Protein

A unique DNA binding protein has been isolated from the meiotic cells of lily plants. In contrast to the other prokaryotic and eukaryotic DNA binding proteins discussed here, this protein is primarily involved in meiotic recombination, rather than in DNA replication (167-169). This protein, referred to as R-protein, is synthesized during meiotic prophase and is localized within the nucleus. The fact that it is only present (and active) in germ cells during the portion of the cell cycle corresponding to meiosis strongly suggests that it is somehow involved in the processes of chromosome pairing and recombination (168). The isolated protein has a molecular weight of -35,000, and forms a stable complex with single-stranded DNA even at 2.0 M NaCl. Minimal concentrations (5 mM) of Mg2+or Caz+are absolutely required for binding (168). Binding is specific for single-stranded DNA, with no detectable binding to RNA. Furthermore, like T4 gene 32 protein and E. coli SSB protein, this protein catalyzes the renaturation of denatured DNA. As previously noted, the ability to catalyze the renaturation of denatured DNA might be quite important in recombination, and is thus consistent with the proposed role of this protein in meiosis. The lily protein can be phosphorylated by a specific CAMP-dependent protein kinase, and the level to which it is phosphorylated determines its in vifro properties (169). The native (phosphorylated) protein exhibits a definite preference for binding to single-stranded DNA, and although the dephosphorylated protein has an increased affinity to single-stranded DNA, it shows an even greater increased affinity for double-stranded DNA. The properties of the dephosphorylated protein revert to those of the native protein by treatment with the protein kinase, which adds two phosphate groups per protein monomer. The native protein stimulates both duplex DNA denaturation and denatured DNA renaturation, but the dephosphorylated protein shows neither of these activities. A very similar protein has been isolated from rat spermatocytes (170). The DNA binding properties and renaturation activity of the rat protein are also modulated by kinase-driven phosphorylation-dephosphorylation activities. In addition a DNA binding protein of mouse acites cells has been shown to vary in its stimulatory effect on DNA replication, depending on its level of phosphorylation. These results suggest that control of the level of phosphorylation of DNA binding proteins may serve as a 167. 168. 169. 170.

Y. Hotta and H. Stern, Dev. B i d . 26, 87 (1971). Y. Hotta and H . Stern, Nnrltre (London)N ~ Bwi d . 234, 83 (1971). Y. Hotta and H. Stern, EJB 95, 31 (1979). J. Mather and Y. Hotta, Exp. Cell Res. 109, 181 (1977).

434

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

general mechanism to modulate the activity of these proteins during the cell cycle in eukaryotes [for a review of eukaryotic protein phosphorylation, see Ref. ( / 7 / ) 1 . 2. Cay Thymus DNA Binding Proteins Three DNA binding proteins have been isolated from calf thymus by Herrick and Alberts (49, 172, 173) using a general purification protocol designed by these workers for the isolation of eukaryotic HDPs (49). The proteins isolated are UP1 [UP for unwinding protein; by current nomenclature this protein would be calf thymus (CT) HDP-I]; a “high salt-eluting fraction” (CT HDP-111, and a “low salt-eluting fraction”. CT HDP-I has a molecular weight of 24,000 and is present in the thymus at -800,000 copies per cell (49). It exists as a monomer in solution, and has an isoelectric point near neutrality. Isoelectric focusing shows that it is composed of four or five subspecies, probably reflecting intrinsic heterogeneity or limited protease action during isolation. These fractions show different affinities for single-stranded DNA; the most acidic fraction binds most weakly. This protein has a marked preference for single-stranded over double-stranded DNA, as demonstrated by its ability to (reversibly) depress the Tm of poly(dA-dT), poly(rA-rU), and C. perfringens duplex DNA (172). This helix destabilization effect is greatest for poly(dA-dT), less for poly(rA-rU), and least for the C . perfringens DNA, suggesting that the protein may have some base- and sugar-binding specificity. From these studies it has been estimated that the affinity for single-stranded DNA is 1500-fold greater than for double-stranded DNA. The acidic subfraction of CT HDP-I shows a much smaller ATm, as might be expected from the smaller affinity of this protein for single-stranded DNA cellulose. Sedimentation studies (173) and optical studies (174) have shown that a stoichiometric complex is formed at a ratio of 7 DNA nucleotide residues per protein (CT HDP-I) monomer. The formation of this complex is very dependent on salt concentration, and the data suggest that up to 6 ionic interactions per protein monomer may be involved (174). The Tm depression of duplex DNA induced by this protein is salt-dependent, decreasing with increased NaCl concentration (172). Although the protein shows a high affinity for single-stranded DNA, unlike the prokaryotic HDPs it does not bind cooperatively (17.3). Electron microscopy shows an extended DNA complex in which the 171. C. S. Rubin and 0. M. Rosen, Annu. Rev. Biochem. 44, 831 (1975). 172. G . Herrick and B . M. Alberts, JBC 251, 2133 (1976). 173. G. Herrick, H. Delius, and B. M. Alberts, JBC 251, 2142 (1976). 174. R. L. Karpel and A. C. Burchard, Binchemisrry 19, 4674 (1980).

21.

DNA BINDING PROTEINS

435

contour length increases by -17% (17.3); the comparable increase in length for DNA complexed with T4 gene 32 protein is 46%. Also, as expected, CT HDP-I (and CT HDP-11) produce denaturation “bubbles” in supercoiled SV40 DNA. In addition changes in the polynucleotide circular dichroism and UV absorbance spectra indicate considerable unstacking of the bases (174). The accessibility of nucleotide bases in this protein-nucleic acid complex was probed by chemical modification and hydrogen exchange techniques. The results suggest that the bases are as exposed in the complex as in free DNA, but that they are also unstacked in the complex (17.5). The CT HDP-I1 preparation contains several protein fractions; the most prominent has a molecular weight of -33,000 and an isoelectric point (PI) of 5.2-5.6. This protein is very similar to CT HDP-I in that (i) it binds noncooperatively to a single-stranded DNA at a nucleotide residue to protein stoichiometry of -10: 1, (ii) it strongly depresses the T, of poly(dA-dT) and poly(rA-rU), and (iii) it forms an extended complex with fd DNA (49, 172). In contrast, the low-salt fraction (molecular weight -33,000) does not denature native DNA and does not form a stable complex visualizable in the EM. In addition, this fraction does not stimulate calf thymus DNA polymerase activity. While the physiological role of these proteins has not been defined, both CT HDP-I and -11 stimulate the activity of the calf thymus DNA polymerase a , but not that of polymerase /3 (173). T4 gene 32 protein does not stimulate polymerase a activity; thus stimulation by the calf thymus proteins may be specific. The amount by which this polymerase is stimulated by CT HDP-I depends on the type of DNA template used, and ranges from a 10-fold stimulation on exonuclease-treated DNA to less than a twofold effect on heat-denatured DNA. As found with T4 gene 32 protein, excess CT HDP-I inhihits DNA synthesis. At optimal concentrations both CT HDP-I and -11 stimulate calf thymus polymerase a activity on oligo(dG) primed-poly(dC) templates more than 5-fold. CT HDP-I can also catalyze the renaturation of tRNA and 5 S RNA to their active (native) forms (176). This renaturation occurs despite the fact that CT HDP-I will not renature denatured DNA; this difference may reflect the fact that the tRNA renaturation is an intramolecular process 175. T. Kohwi-Shigematsu, T. Enomoto, M. Yamada, M. Nakanishi, and M. Tsuboi, PNAS 75, 4689 (1978). 176. R . L. Karpel, N . S. Miller, and J. R. Fresco, iff “Molecular Mechanisms in the Control of Gene Expression” (D. P. Nierlich, W. J . Rutter, and C. F. Fox, eds.), p. 411. Academic Press, New York, 1976.

436

S . KOWALCZY KOWSKI, D. BEAR, AND P. VON HIPPEL

and DNA renaturation is intermolecular (172). In addition, CT HDP-I seems to be located primarily in the cytoplasm of the cell, suggesting that it may be involved in RNA manipulation (176). By using the complementary approach of isolating a protein that stimulates the activity of polymerase a, and then comparing it to the proteins that have been previously described, a protein very similar to the most basic component of CT HDP-I has been purified (177). Its binding is specific for single-stranded DNA, and it stimulates the activity of DNA polymerase LY up to 8-fold depending on the template, whereas polymerase /3 is stimulated less than 2-fold.

3. Ustilago mrrydis D N A Binding Protein An HDP has been isolated from mitotic cells of the basidiomycete fungus, U . maydis. This protein may be involved in DNA replication in this organism in that it stimulates U . rnciydis DNA polymerase (178, 179). The protein has a molecular weight of -20,000, and exists in solution as a monomer. Filter binding experiments show that it binds tightly to singlestranded DNA, but not to double-stranded DNA. The T,,,of poly(dA-dT) is lowered by -50" in the presence of saturating amounts of this protein, and Mg'+ was found to increase the rate of renaturation. As with T4 gene 32 protein and the lily R-protein, the I / . mcrydis HDP also catalyzes the renaturation of denatured DNA (178). The protein stimulates by twofold the activity on denatured DNA of the only DNA polymerase that has been isolated from U . maydis. As with many of the other HDPs, excess binding protein inhibits the activity of the polymerase. Since the U . muydis HDP does not stimulate the activity of polymerases from E. coli. M. lureus, T4, or T7, the stimulation of the U . maydis polymerase may be specific. However, no specific interaction of polymerase with binding protein could be detected. In addition to the stimulation of polymerase activity, the U . mciydis HDP in stoichiometric excess also inhibits the nucleolytic digestion of DNA. The stimulatory effect of this protein on the U . maydis polymerase was found to be due to an increase in the rate rather than in the extent of DNA replication. This stimulation was found to arise from several effects that were produced by the presence of binding protein (179). When denatured DNA was used as a template for U . rnaydis polymerase, the apparent K , for nucleoside triphosphates was increased 3- to 4-fold in the presence of binding protein. In addition, the apparent affinity of the polymerase for

-

177. F. Cobianchi, S . Riva, G. Mastromel, S. Spadari, G. Pedrali-Noy, and A. Falaschi, CSHSQB 43, 639 (1979). 178. G . R. Banks and A. Spanos, J M B 93, 63 (1975). 179. G. Yarranton, P. D. Moore, and A. Spanos, M d . Get?. Genet. 145, 215 (1976).

21.

DNA BINDING PROTEINS

437

DNA was increased fourfold by the U . may& HDP. And finally, the V,,, of the polymerization reaction increased -50%. All of these effects have been interpreted to indicate that the U . maydis binding protein stimulates replication both by removing secondary structure in the DNA template, which may impede the polymerase, and by providing a specific proteinnucleic acid complex with which the DNA polymerase can interact. 4. Mouse DNA Binding Proteins Several DNA binding proteins have been isolated from mouse tissue, including proteins isolated from 3T6 cells (180), ascites cells (181), and myeloma cells (182). Each of these proteins is a helix-destabilizing protein, and the last two stimulate mouse polymerase a. The role of the protein isolated from 3T6 cells is unknown; it is found primarily in the cytoplasm, and more is found in growing than in resting cells. The HDP found in mouse ascites cells has a molecular weight of -35,000, and exists as a monomer in solution (181). It shows a high affinity for single-stranded DNA or RNA, and only a slight affinity for doublestranded DNA. Sedimentation experiments suggest that the proteinnucleic acid complex is asymmetric and highly extended, and that saturation of the DNA occurs at -6-10 nucleotide residues per protein monomer. The properties of this protein suggest that it is analoJous to the high-salt fraction protein isolated from calf thymus. The mouse protein also stimulates the activity of mouse DNA polymerase a fourfold, but only on a heat-denatured DNA template. There is no stimulation of activity on pancreatic DNAse-treated (activated) DNA templates, but we note that the activity of the DNA polymerase is already sevenfold greater on the activated than on the denatured DNA template. This helix-destabilizing protein may specifically stimulate polymerase a , since assays of the effect of this protein on mouse polymerase p, E. coli DNA polymerase, and T7 DNA polymerase all show only very slight stimulatory effects. The effect on polymerase a has been surmised to be at the elongation step. It has also been showed that DNA binding protein increases the affinity of polymerase a for DNA cellulose (183) similar to the effect shown with the U . maydis protein. Although no direct association between polymerase and binding protein has been seen, these results suggest that the binding protein may stabilize the polymerase a-DNA complex.

-

180. 181. 182. 183. (1978).

R. L. Tasi and H. Green, J M B 73, 307 (1973). B . Otto, M. Baynes, and R. Knippers, EJB 73, 17 (1977). S. R. Planck and S. H. Wilson,JBC 255, 11547 (1980). A . Richter, R . Knippers, and B. Otto, FEBS (Fed. E m . Biochem. SOC.Lett.) 91,293

438

S . KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

As with the protein isolated from lily, the mouse DNA binding protein can be phosphorylated (1 phosphate/protein monomer) by a chromatinassociated kinase, resulting in an alteration of its binding properties. Although phosphorylation has no effect on single-stranded DNA affinity, the phosphorylated protein shows a reduced affinity for double-stranded DNA and a greatly reduced ability to stimulate DNA polymerase a. Thus this system seems to represent another example of HDP activity controlled by the level of covalent phosphorylation. Another protein that has been isolated from mouse cells (myeloma), and is clearly not the same as those obtained from 3T6 or ascites cells, is mouse HDP-I (182). This protein is heterogeneous in molecular weight, ranging from 24,000 to 33,000, with a predominant species at -27,000. This apparent heterogeneity is not due to different levels of phosphorylation, acetylation, or glycosylation; rather peptide mapping and tryptic digestion studies implicate protease activity. This HDP is localized predominantly in the cell nucleus, and is not associated with the chromatin; however -25% is also found in the cytoplasm. Note that the amino acid compositions and the molecular weights of the mouse HDP-I and CT HDP-I are very similar. Sedimentation experiments with mouse HDP-I and denatured DNA indicate that the protein binds noncooperatively to single-stranded DNA, with a site size of 5 to 7 nucleotide residues per protein monomer and a binding constant of -4 x lo5M - I . In addition, saturating amounts of this protein depress the T,,, of poly(dA-dT) by -25” in 0.01 M salt. The binding of DNA to this protein induces a fluorescence increase of -35%, in contrast to fluorescence quenching typically seen with prokaryotic DNA binding proteins (182). Studies of the products of limited tryptic hydrolysis of the mouse HDP-I have shown that different products are obtained, depending on whether or not the protein is bound to single-stranded nucleic acids in the digestion process. This effect depends on the type of nucleic acid present, with poly(dT) and denatured DNA protecting the protein most effectively. Poly(dI), poly(rA), and poly(dC) alter the digestion pattern to a lesser extent, and poly(dA) and poly(dA-dT) are quite ineffective. Thus mouse HDP-I binding may display some base composition-dependent binding specificity. Short oligonucleotides are much less effective in protecting the -19,000 molecular weight digestion product, and this spectrum of protection effectiveness has been used to measure the affinity of oligonucleotides for HDP-I. In addition the DNA binding properties of two proteolytic products (molecular weights of about 19,000 and 22,000) that lack the amino terminus are identical to that of native HDP-I, suggesting that -65

21.

DNA BINDING PROTEJNS

439

residues at that end of the polypeptide are not essential for DNA binding (182). The mouse HDP-I seems to serve as an accessory protein to mouse DNA polymerase a (184). This conclusion is based on the fact that HDP-I increases the processivity of DNA polymerase a in a manner very similar to that observed for T4 gene 32 protein. In the absence of HDP-I, polymerase a can processively extend an RNA-primer, yielding a Poisson distribution of products with a maximum (nucleotides added) length of 8-9 residues. When HDP-I is added to this system, the peak of the distribution of added nucleotides is increased to 21 nucleotide residues, suggesting that eukaryotic binding proteins have similar effects on eukaryotic in vitro replication as do the procaryotic proteins (106, 185).

5 . Adenovirus DNA Binding Protein Early in infection of human cells by oncogenic (DNA) adenoviruses (types 2 and 51, a 72,000 molecular weight viral-coded DNA binding protein (Ad DNA binding protein) is synthesized in very large numbers (--lo7 copies per cell) (186, 187). The gene coding for this DNA binding protein has been mapped on the adenovirus genome (188), and a temperaturesensitive mutant (HSts 125) has been isolated (189, 190). Adenovirus carrying this mutation is defective in the initiation of viral DNA replication (19/), and the DNA binding protein isolated from cells infected with the mutant strain is defective, both in binding to single-stranded DNA (190) and in complementation assays for in vitro replication (191a). The protein is phosphorylated in wivo (187, 192-194). It also appears to be a major S. D . Detera, S. P. Becerra, J . Swack, and S. H. Wilson, JBC, in press (1981). C. C. Huang, J. E. Hearst, and B. M . Alberts, JBC 2.56, 4087 (1981). P. C. Van der Vliet and A. J. Levine, Ntrtrrre Nen, B i d . 246, 170 (1973). T. Linne, H . Jornvale, and L. Pullipson, EJB 76, 481 (1977). J. B. Lewis, J . F. Atkins, P. R . Barn, R . Solem, R. F. Gesteland, and C. W. Anderson, Cell 7, 141 (1976). 189. M. J. Ensinger and H. S. Ginsberg, J . Virol. 10, 328 (1972). 190. P. C. Van der Vliet, A. J. Levine, M. J. Ensinger, and H . S. Ginsberg,J. Virol. 15, 184. 185. 186. 187. 188.

348 (1975). 191. P. C. Van der Vliet and J . S. Sussenbach, Virology 67, 415 (1975). 191a. M. S. Horiwitz, L . M. Kaplan, M. Abboud, J. Maritato, L. T. Chow, and T. R . Broker, C S H S Q B 43, 769 (1978). 192. N. Axelrod, V i r o l o g y 87, 3 6 3 8 3 (1978). 193. Y.-H. Jeng, W. S. M . Wold, K . Sugawara, Z. Gilead, and M. Green,J. Virol. 22,402 (1977). 194. A. D. Levinson, E. H. Postel, and A. J. Levine, Virology 79, 144 (1977).

440

S. KOWALCZYKOWSKI, D. BEAR, A N D P. VON HIPPEL

component of adenovirus replication complexes isolated from infected cells (19.5497). Whereas this DNA binding protein has been shown to be important in initiation, recent evidence suggests that it may also function during the elongation step of adenovirus DNA replication (198). The Ad DNA binding protein appears to be fibrous (199, 200), and to exist as a monomer in solution. It can be cleaved into two subfragments of molecular weights about 26,000 and 44,000 (201, 202). The isolated 26,000 fragment is derived from the amino-terminal domain of the protein, contains most of the sites of protein phosphorylation, and does not bind single-stranded DNA (201). The 44,000 fragment is derived from the carboxy terminus of the protein, binds to single-stranded DNA as an isolated fragment, and can function in in v i m DNA replication (201, 202). The defined mutation site (HSts125) is located in the larger fragment. The DNA binding properties of this protein have only been roughly characterized. Although the protein was initially thought to be a singlestranded specific DNA binding protein (site size -7 nucleotide residues per protein monomer) on the basis of DNA-cellulose (186) and nitrocellulose filter binding assays (200), it has since been shown that the protein does not melt poly(dA-dT) and may in fact stabilize this duplex polynucleotide (203). In addition, it has been demonstrated that Ad DNA binding protein can bind to double-stranded adenovirus DNA, although it shows a much higher affinity for the termini of duplex DNA molecules (203). As previously indicated, the Ad DNA binding protein has been shown to be phosphorylated in vivo at several sites (187, 192-194). Although the role of phosphorylation in the function of Ad DNA binding protein has been elusive, it appears that the more extensively phosphorylated species have a lower affinity for single-stranded DNA-cellulose (2U1). Further studies suggest that newly phosphorylated DNA binding protein associates preferentially with replicating viral DNA; after a period of time the protein is also found associated with the mature duplex DNA (195). A better understanding of the relationship between phosphorylation, DNA 195. C. H. Shaw, W. C. Russell, and M. K. Rekosh, Virology 92, 436 (1979). I%. C. Kedinger, 0. Brison, F. Perrin, and J. Wilhelm, J.Virol. 26, 364 (1978). 197. M . h e n s , T. Yamashita, R. Padmanbhian, T. Tsuruo, and M. Green,JBC 252,7947 (1977). 198. p. C. Van der W e t , J. Zandberg, and H. S. Jansz, Virology 80, 98 (1979). 199. K. Sugawara, Z. Gilead, and M. Green, J . Virol. 21, 338 (1977). 200. P. C. Van der Vliet, W. Keegstra, and H . S. Jansz, EJB 86, 389 (1978). 201. H. Klein, W. Maltman, and A. J. Levine, JBC 254, 11051 (1979). 202. H . Ariga, H. Klein, A. J. Levine, and M. S. Horwitz, Virology 101, 307-310(1980). 203. D. M. Fowlkes, S. T. Lord, T. Linne, U. Pettersson, and L. Philipson, JMB 132, 163 (1979).

21. DNA BINDING PROTEINS

44 1

binding properties, and the role of this protein in DNA replication awaits further experimentation.

V.

DNA Binding Proteins as Research Tools

A.

ELECTRON MICROSCOPY

Single-stranded DNA binding proteins are, in general, too small to discern as individual molecules in the electron microscope. However complexes formed by these proteins with DNA and RNA cause distinct morphological changes in these entities. This property of the binding proteins has been exploited as a tool for the electron microscopic visualization of single-stranded DNA, which under normal conditions is difficult to see as a “naked” species, and may be severely deformed during spreading and grid preparation. In this way the T4 gene 32 protein has been shown to be useful in the mapping of single-stranded regions of DNA-DNA and DNA-RNA hybrids (204, 205). This technique has been applied to the mapping of ribosomal RNA and tRNA genes on 480 phage genomes (2051, the mapping of histone and ribosomal genes in Drosophilu (206, 207), and the mapping of terminal sequences in the adenovirus genome (208). The E. coli SSB protein has been used to stabilize and visualize singlestranded DNA sequences generated byrecBC enzyme [see Chapter 13, this volume, and Refs. (209, 210)l and to visualize single-stranded ends of reconstituted histone-DNA complexes (21 1 ) .

B. BIOCHEMICAL ASSAYS Single-stranded DNA binding proteins have been used as traps for sequences of single-stranded DNA that are transiently formed during enzymatic assays. This approach has been particularly useful in studying the (ATP-dependent) unwinding reaction of the class of enzymes known as 204. 205. 206. (1976). 207. 208. 209. 210. 211.

C. Brack, T. A. Bickle, and R. Yuan,JMB 96, 693 (1975). M. Wu and N. Davidson, PNAS 72, 4506 (1975). M. Wu, D. S . Holmes, N. Davidson, R. H. Cohn, and L. H . Kedes, Cell 9, 163

M. Pellegrini, J. Manning, and N. Davidson, Cell 10, 213 (1977). M. Wu, R. J. Roberts, and N. Davidson, J . Virol. 21, 766 (1977). J . Rosamond, K. M. Telander, and S. Linn, JBC 254, 8646 (1979). A. Taylor and G . R. Smith, Cell 22, 447 (1980). K. Dunn and J. D. Griffith, Nucleic Acids Res. 8, 555 (1980).

442

S. KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

DNA helicases [see Chapter 20 of this volume and Refs. (2/2-215)]. Specifically, eitherE. coli HDP or T4 gene 32 protein have often been used to sequester single-stranded DNA regions formed by the unwinding of duplex DNA by helicases. After stopping the enzymatic reaction, the HDP-nucleic acid complex is dissociated and the free single-stranded DNA is digested by S1 nuclease. The extent of unwinding can then be determined by measuring (typically, radiochemically) the amount of S1 nuclease-resistant duplex DNA that remains. The unwinding activity ofE. coli helicase I11 and rep protein have been measured in this manner, and it was shown that HDP is required to trap the unwound DNA (215). In contrast E. coli helicases I and 11, and the T4 helicase (at high protein concentration) can unwind duplex DNA in the absence of HDPs (213).

VI.

Conclusions

Although the specific details of the interactions of the single-stranded DNA binding proteins with nucleic acids, and their exact biological functions, differ somewhat for each protein considered in Section IV, some overriding generalizations emerge. Thus all these proteins seem to operate stoichiometrically (as opposed to catalytically), in that they are present at intracellular levels sufficient to effectively saturate the single-stranded DNA intermediates produced during replication, recombination, and repair. To avoid dissipation over the great excess of double-stranded DNA present in the cell, most of the proteins show appreciably more affinity for single-stranded than for double-stranded DNA. Furthermore this net difference in affinity is (at least for the prokaryotic proteins) amplified by the fact that binding to single-stranded nucleic acids is cooperative. This binding cooperativity is essential in permitting complete coverage of singlestranded sequences, and also in effectively destabilizing the small duplex hairpins formed by intrastrand base pairing in single-stranded DNA. [When present, such hairpins can slow down or stop the utilization of the involved sequence as a template for DNA polymerase [see Refs. (99, 1831. In addition, uncomplexed single-stranded sequences are very susceptible to attack by intracellular endonucleases. It seems most likely that removing hairpins from transient single-stranded DNA sequences, and 212. M. Abdel-Monem, H. F. Lauppe, J. Kartenbeck, H. Durwald, and H . HoffmanBerling,JMB 110, 667 (1977). 213. B. Kuhn, M. Abdel-Monem, H. Krell, and H. Hoffman-Berling, JBC 254, 11343 ( 1979). 214. M. Duguet, G. T. Yarranton, and M. L. Gefter, C S H S Q B 43, 335 (1978). 215. G . T. Yananton, R. H. Das, and M. L. Gefter, JBC 254, 12002 (1979).

21. DNA BINDING PROTEINS

443

protecting these sequences against nucleases, comprises the central task of the single-stranded DNA binding proteins in vivo. Of course these proteins must function within multicomponent complexes. As a consequence most appear to have developed some degree of interactional specificity with other proteins of the homologous complex: this is manifested particularly (in in virro assays) by specific stimulation of certain homologous polymerases. These effects might proceed by a variety of mechanisms. For example, interaction with the homologous polymerase could prevent the destabilization of the primer-template complex by single-stranded DNA binding protein. Alternatively, the binding protein could put the single-stranded template into an optimal geometry for utilization as a companion template by the homologous polymerase. Another possibility is that only the homologous binding protein can be effectively removed from the single-stranded DNA by components of the homologous replication (or recombination) complex. Further experiments with complete in vitro systems are required to choose between these possibilities or others. In order to effectively discharge its primary function(s), the binding of binding protein to single-stranded DNA lattices must be relatively nonspecific with respect to nucleotide sequence or composition. At the same time, for effective functioning of replication (and probably recombination and repair) complexes, the presence of either too much or too little single-stranded binding protein could be inhibitory, or even lethal. Thus some mechanism for controlling the intracellular concentration of SSB proteins is probably generally required. For T4 gene 32 protein this regulation is autogenous at the translational level, and involves differences in affinity for nucleic acid sequences based both on sugar and on nucleotide residue type. These affinity differences are relatively small at the level of the binding of the individual protein molecule, as required to avoid problems of incomplete saturation of single-stranded DNA. Yet these differences are also large enough to be amplified, by binding cooperativity , into control systems of considerable overall specificity (see Ref. 5 ) . The molecular details of the interactions of single-stranded DNA binding proteins with nucleic acid lattices, resulting in strong overall binding modulated by some binding specificity, are just beginning to emerge. Electrostatic interactions are generally involved; at the same time more specific binding interactions, based on hydrogen bonding and possible staclung interactions of bases with aromatic acid residues in the binding site, may also participate. The principles outlined above are probably involved, in various combinations, in the in v i i v functioning of most single-stranded binding proteins.

444

S . KOWALCZYKOWSKI, D. BEAR, AND P. VON HIPPEL

To the extent that present results permit us to judge, the eukaryotic SSB proteins have many features in common with the better-studied prokaryotic proteins. However some significant differences are seen. For example, some or most of the eukaryotic proteins that have been examined (see Table 111) may bind noncooperatively to single-stranded nucleic acid lattices. The significance of this is not clear; perhaps these eukaryotic proteins operate in vivo in collaboration with factors or proteins yet to be discovered to achieve the binding saturation brought about by binding cooperativity in the prokaryotic systems. We note also that both the biological activities and the DNA binding properties of several of the eukaryotic binding proteins seem to be modulated in vivo by enzymatically catalyzed covalent phosphorylation and dephosphorylation reactions. Such processes have not been observed with prokaryotic singlestranded binding proteins, and when fully understood may turn out to be involved in controlling the effective binding levels and specificities (and binding cooperativity?) of the eukaryotic SSB proteins. Overall, patterns of single-stranded DNA binding protein properties are starting to emerge (see Tables I1 and 111). However the total range of function in which these proteins participate will not be clear until we understand, in molecular detail, the entire physiological systems of which these proteins form an integral part. ACKNOWLEDGMENTS We are pleased to acknowledge many conversations with colleagues, both at the University of Oregon and elsewhere, that have helped to focus our thinking and improve our understanding of the subject matter of this review. We are also grateful t o many colleagues, including Drs. Bruce Alberts, Rae Lynn Burke, Joseph Coleman, Malcolm Gefter, Larry Gold, Jack G r f i t h , Junko Hosoda, Richard Karpel, William Konigsberg, Timothy Lohman, Alexander McPherson, Thomas Record, Kenneth Williams and Samuel Wilson, who provided unpublished data or preprints of their relevant papers prior t o publication. We wish to thank Ms. Nancy Caretto, who patiently typed and retyped the many drafts of this review. The research described here that has been carried out in our laboratory has been supported, in part, by U.S. Public Health Services Research Grant GM-15792, American Cancer Society Postdoctoral Fellowship PF-1301 (to S. Kowalczykowski) and USPHS Postdoctoral Fellowship GM-06676 (to D. Bear).

The recA Enzyme of Escherichia coli and Recombination Assays KEVIN McENTEE GEORGE M . WEINSTOCK

I . Introduction . . . . . . . . . . . . . . . . . I1 . Purification . . . . . . . . . . . . . . . . . . 111. Physical Properties . . . . . . . . . . . . . . IV. Reactions Catalyzed . . . . . . . . . . . . . . A . Single-Stranded DNA-Dependent NTPase . . B . Duplex DNA-Dependent ATPase . . . . . . C . Reassociation of Single-Stranded DNA . . . D. Strand Assimilation or Strand Uptake . . . E . Protease Activity . . . . . . . . . . . . . . V. Assays for Recombination . . . . . . . . . . VI . Biological Role . . . . . . . . . . . . . . . . A . Strand Annealing and Assimilation in Vivo . B . Role of Protease Activity in Vivo . . . . . C. Addition Role of recA Protein in Vivo . . . VII . Research Applications . . . . . . . . . . . . .

I

.

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . . . . . . .

. . . . . . . . .

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

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

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

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

445 447 448 453 453 455 456 457 463 464 465 468 469 469 470

INTRODUCTION

The recA protein of Escherichia coli plays a fundamental role in DNA metabolism . Genetic studies have demonstrated that the recA product is 445 THE ENZYMES. Vol . XIV Copyright 0 1981 by Academic Pms. Inc. All rights of reproduction in any form reserved ISBN 0-12-122714-6

446

KEVIN McENTEE AND GEORGE M. WEINSTOCK

essential for homologous genetic recombination (/-.?), DNA repair (4, 5 ) , and the expression of SOS functions (functions that alter cellular metabolism following DNA damage) (6, 7) such as prophage induction (8, 9 ) , mutagenesis ( l o ) , and cell division inhibition (1I ) , which occur after cellular DNA is damaged. The protein was originally identified using specialized transducing phages that carried the recA genetic region (IZ), and was subsequently shown to be identical to protein X, a polypeptide whose synthesis is induced following DNA damage or replication blockage (1.717). Since its discovery in 1976, the recA protein has become the subject of considerable biochemical interest with the result that the characterization of its properties has progressed rapidly. To a considerable extent the phenotypic complexity of recA- mutations is reflected in the novel biochemical properties of the recA protein. Several activities are known for this multifunctional enzyme including (a) DNA-dependent hydrolysis of nucleoside triphosphates (18, 19); (b) ATP-dependent pairing of DNA molecules between either complementary single strands (annealing) (19) or single strands and homologous duplex DNAs (strand assimilation or strand uptake) (20, 2 1 ) ; and (c) ATP- and polynucleotide-dependent proteolysis of specific regulatory proteins (22, 2 3 ) . Remarkably, these activities reside in a single polypeptide with a molecular weight under 40,000. Taken together these activities account for the range of processes dependent upon a functional recA gene in viva. The DNA pairing activities ofrecA protein reflect the function of this enzyme in homology-dependent recombination and postreplication repair of damaged DNA (2,24,25).The protease activity of recA protein functions in viva to coordinately regulate several critical genes that are expressed in response to DNA damage (6, 7, 14, 26). 1. A. J. Clark, Annu R e v . Gen. 7, 67 (1973). 2. C. M. Radding, Annu. Rev. Biochem. 47, 847 (1978). 3. A. J. Clark and A. D. Margulies, PNAS 53, 451 (1965). 4. P. Howard-Flanders and L. Theriot, Genetics 53, 1137 (1966). 5. P. C. Hanawalt, P. K. Cooper, A. K. Ganesan, and G. A. Smith,Annu Rev. Biochem. 48, 783 (1979). 6 . M. Radman, in “Molecular Mechanisms for Repair of DNA’ (P. C. Hanawalt and R. B. Setlow, eds.), Part A, p. 355. Plenum, New York, 1975. 7. E. M. Witkin, Bacteriol. Rev. 40, 869 (1976). 8. K . Brooks and A. J. Clark, J . Virol. I , 283 (1967). 9. I. Hertman and S. W. Luria, JMB 23, 117 (1967). 10. A. Miura and J . 4 . Tomizawa, Mol. Gen. Genet. 103, 1 (1968). 1 1 . M. Inouye, J . Bacteriol. 106, 539 (1971). 12. K. McEntee, J. E. Hesse, and W. Epstein, PNAS 73, 3979 (1976). 13. M. Inouye and A. B. Pardee, JBC 245,5813 (1970). 14. K. McEntee, PNAS 74, 5275 (1977). 15. L. J. Gudas and D. W. Mount, PNAS 74, 5280 (1977).

22. recA ENZYME O F E . coli

447

This chapter focuses on our understanding of the biochemica1 and enzymatic properties of the recA enzyme, and describes the assays used to analyze these reactions in detail. Furthermore, where possible we attempt to relate the biochemical properties of t h e recA protein to its multiple biological roles.

II.

Purification

In exponentially growing Escherichitr coli. recA protein is present at a level of approximately 2000-5000 molecules per cell (27). This level increases to more than 150,000 molecules following DNA damaging treatments (mitomycin C, nalidixic acid, UV irradiation). Induction by DNA damage can be further amplified by employing bacterial strains that contain multicopy plasrnids containing the cloned recA gene (14, 28). Under appropriate conditions such plasmid-containing cells overproduce recA protein to levels approaching 10% of the total cellular protein (14, 27, 28). Alternatively, large quantities of recA protein are produced during vegetative growth of hprecA specialized transducing phages (29). In either case extracts prepared from plasmid-containing or phage-infected cells provide a rich source of recA protein. At least four purificiation procedures are currently available that permit simple and rapid preparation of several hundred milligrams of nearly homogeneous recA protein from a few hundred grams of cell paste (18, 19, 23, 30).Most of the purification schemes 16. P. T. Emmerson and S. C. West, M o l . Gm. Genet. 155, 77 (1977). 17. J . W. Little and D. G. Kleid, JBC 252, 6251 (1977). 18. T. Ogawa, H. Wabiko, T. Tsurirnoto, T. Horii, H. Masukata, and H. Ogawa, C S H S Q B 44, 909 (1979). 19. G. M. Weinstock. K. McEntee, and I. R. Lehman, P N A S 76, 126 (1979). 20. T. Shibata, C. Das Gupta, R. P. Cunningham, and C. M . Radding, P N A S 76, 1638 (1979). 21. K. McEntee, G . M. Weinstock, and I. R. Lehman, P N A S 76, 2615 (1979). 22. J. W. Roberts, C. W. Roberts, and N. L. Craig, P N A S 75, 4714 (1978). 23. J. W. Roberts, C . W. Roberts, N. L . Craig, and E. M. Phizicky, CSHSQB 44, 917 ( 1979). 24. W. D. Rupp and P. Howard-Flanders, J M B 31, 91 (1968) . 25. K . C. Smith and D. H. C. Meun,JMB 51, 459 (1970). 26. N. L. Craig and J. W. Roberts, Nrrture (London) 283, 26 (1980). 27. C. Paoletti, in preparation. 28. K. McEntee, in "DNA Repair Mechanisms" (P. C. Hanawalt, E. C. Friedberg, and C. F. Fox, eds.), ICN-UCLA Symp. Mol. Cell. Biol., Vol. 19. p. 349, Academic Press, New York, 1978. 29. J. W. Little, S. H. Edmiston, L. Z. Pacelli, and D. W. Mount, P N A S 77, 3225 (1980). 30. M. M. Cox, K. McEntee, and I. R. Lehman, JBC 256, 4676 (1981).

448

KEVIN McENTEE AND GEORGE M . WEINSTOCK

involve conventional column chromatographic elution procedures from either phosphocellulose or single-stranded DNA-cellulose. A later purification procedure (30) is based upon the observation that certain nucleoside triphosphates or diphosphates reduce the affinity of recA protein for single-stranded DNA. At relatively low ionic strength (below 100 mM KCl), recA protein binds to single-stranded DNA-cellulose. Treatment of these recA protein-DNA complexes with ATP rapidly releases recA protein from the DNA. This dissociation is achieved with concentrations of ATP below 1 mM. Furthermore, the fact that ADP and dTTP efficiently promote this release argues that hydrolysis of the nucleotide is not necessary for dissociation (31).recA protein prepared by ATP elution is more than 99% pure Y O ) and contains none of the activities that are often found to contaminate material prepared by other methods, including polynucleotide phosphorylase, endonuclease, and DNA independent ATPase (23, 30,32). The recA protein obtained by ATP elution catalyzes all of the activities mentioned earlier including proteolysis of bacteriophage A repressor (30). Several mutant forms of recA protein have been purified including the tij7 and lexB.10 mutant enzymes (19, 33, -?4),and a cold labile recA protein (recA629 protein) from a conditional (cold-sensitive) recombinationdeficient recA- strain. In v i m the purified recA629 protein is impaired in its ability to hydrolyze ATP and to hybridize DNA molecules at the nonpermissive temperature (19). Several differences have been noted in the behavior of the mutant gene products during purification suggesting alterations in the properties of these polypeptides. Because recA protein is generally prepared from induced cells (i.e., cells treated with a DNAdamaging agent, or inhibitor of DNA replication) the suggestion has been made that this form of recA protein might differ from the uninduced form due to covalent modification or structural alteration of the polypeptide. There is no direct evidence that supports this idea.

111.

Physical Properties

The complete nucleotide sequence of the recA+ gene and its control region have been determined, and the predicted primary sequence of the protein has been confirmed by sequencing portions of the amino and car31. 32. 33. 34.

K. McEntee, G . M. Weinstock, and 1. R. Lehman,JBC, in press (1981). G . M. Weinstock, K . McEntee, and I. R. Lehman, J B C , in press (1981). M. Castellazzi, J. George, and G . Buttin, M o l . Gen. Gener. 119, 139 (1972) P. Morand, M. Blanco, and R. Devoret, J . Bacterial. 131, 572 (1977).

449

22. recA ENZYME OF E. coli

boxyl termini of the polypeptide (3.5,3 6 ) . The recA protein is composed of 352 amino acids with a total monomer molecular weight of 37,842. The protein is relatively low in tyrosine (seven) and tryptophan (two) (25, 36). Although analysis of the protein sequence does not indicate which regions of the polypeptide are involved in binding to DNA, the carboxyl-terminal half of the protein contains more than 70% of the basic amino acid residues, which may participate in DNA interactions (36). The aminoterminal half of the recA protein chain contains sequences that are homologous to peptide sequences in and around the active sites of serine proteases (-?6). These primary structural features are suggestive of the indicated role, but their real significance must await more direct demonstrations of their functionality. An important first step toward this goal is the preparation of crystals of recA protein that are suitable for X-ray diffraction analysis (-?7). Physicochemical studies indicate that the recA protein is highly aggregated in solution under conditions where it is enzymatically active (18,38). At neutral pH, recA protein sediments in sucrose gradients as a heterogeneous population of multimeric species (tetrameric or greater) (38),even at relatively high ionic strenth (18, 38). By lowering the pH to approximately 6.2, the aggregates dissociate and the protein sediments uniformly (Fig. 1). The protein retains full enzymatic activity at this pH and sediments like an asymmetric dimer in the presence of MgClz (38). A dramatic increase in the sedimentation velocity of recA protein occurs upon addition of ATP (or its nonhydrolyzable analog ATP[yS]) (Fig. 1) (18, 38). Although under these conditions heterogeneous sedimentation behavior is once again observed for the protein, electron microscopic examination of the fast sedimenting material reveals long, highly ordered filaments of recA protein. In the absence of ATP at the lower pH, only short protein filaments are detected. Moreover, at pH 7 . 5 , where the recA protein sediments anomalously, few if any ordered filamentous structures are discernible, but irregular protein aggregates are evident. Thus the effect of low pH is to dissociate these aggregates and, in the presence of ATP, permit polymerization of subunits into highly ordered filaments (38). Filamentation is not efficiently induced by UTP (Fig. 11, a nucleoside triphosphate that is rapidly hydrolyzed by recA protein but does not substitute efficiently for ATP in DNA annealing or assimilation reactions (38). +

35. T. Horii, T. Ogawa, and H. Ogawa, PNAS 77, 313 (1980). 36. A. Sancar, C. Stachelek, W. Konigsberg, and W. D. Rupp, PNAS 77, 2611 (1980). 37. D. B. McKay, T. A. Steitz, I. T. Weber, S. C. West, and P. Howard-Flanders, JBC 255, 6662 (1980). 38. G. M. Weinstock, K . McEntee, and I. R. Lehman. JBC, in press (1981).

450

KEVIN McENTEE A N D GEORGE M. WEINSTOCK I

I

I

8

12

1

I

I

I

I

I

I

20

24

28

32

36

40

10

2 0

0

4

16

FRACTION

FIG.1. Effects of ATP and UTP on the sedimentation properties of recA protein. All incubations and sucrose gradient (10% to 30%) sedimentations were performed in maleate buffer pH 6.2 (38). A detailed structural analysis of the subunit arrangement in these filaments, which are composed of several hundred to several thousand recA protein monomers, has not been performed. Presumably the filamentous form of recA protein is necessary for binding and unwinding duplex DNA (see Section IV,B). Moreover, the alignment of DNA molecules during strand assimilation could be accomplished using protein filaments to move one DNA molecule relative to another. Based upon a steady state kinetic analysis of ATP hydrolysis catalyzed by recA protein (39) and photoaffinity labeling experiments with [32P]azidoATP(40), there appears to be a common binding site(s) for nucleoside triphosphates and the phosphothiolate analogs, adenosine 5’-0-3-thiotriphosphate, ATP[yS], and uridine 5’-0-3-thiotriphosphate, UTP[yS]. Using radioactively labeled ATP[yS] it has been possible to demonstrate that there is one high-affinity binding site for this analog per monomer of recA protein (Fig. 2) ( 4 / ) , Competition experiments with single-stranded and double-stranded DNA indicate that 5- 10 nucleotides of DNA bind per recA protein monomer, and that single-stranded and 39. G . M. Weinstock, K. McEntee, and I. R. Lehman, J B C , in press (1981). 40. K. McEntee, G . M. Weinstock, and I. R. Lehman, Prog. Nircleic Acid Res. Mol. Biol. in press (1981). 41. G . M. Weinstock, K. McEntee, and I. R . Lehman, J B C , in press (1981).

45 1

22. recA ENZYME OF E . coli

' b -

.9

.8

-

.7

-

.6

-

% ?!

0

E

1

n

z

3

-8 v)

x

I

z

n.

.4 -

.5

0

2

4 6 8 10 12 NUCLEOTIDES/ recA MONOMER

14

16

FIG. 2. Titration of single- and double-stranded DNA for the formation of recA protein-ATP[yS] complexes. The binding curves for single-stranded DNA (pH 8.0) and double-stranded DNA (pH 6.2) are identical and indicate that at saturation, approximately 1 mole of ATP[yS] is tightly bound to each mole of recA protein (41). duplex DNA binding sites on the protein are identical or overlapping (.?/). Interestingly, oligonucleotides that contain fewer than 12-18 residues bind poorly to recA+ protein (-?I, 3 2 ) . These results argue that the recA polypeptide contains a binding site for nucleoside triphosphates, one or more binding sites for duplex or single-stranded DNA, and regions that participate in protein-protein interactions. That these sites or domains are not independent but interact with one another is apparent from the biochemical and enzymatic properties of this protein. For example, tight binding of polynucleotides accompanies binding of the nonhydrolyzable analog ATP[ y S ] . Conversely, tight binding of the ATP[ y S ] requires the presence of a polynucleotide effector. In the absence of the other cofac-

452

KEVIN McENTEE AND GEORGE M. WEINSTOCK

tor, either polynucleotide or ATP[yS] dissociates readily from the protein ( 2 / , 4 0 , 4 / )Thus . binding of ATP[yS] influences the affinity ofrecA protein for polynucleotide and vice versa. The effect of the ATP[yS] analog, unlike ATP, is to lock the recA protein into a complex that contains both ATP[yS] and polynucleotide and dissociates extremely slowly (31, 40,41). In contrast, ATP stimulates the dissociation of recA protein from singlestranded DNA (Fig. 3) ( 3 0 , 3 / ) ,a result that has been useful in purification of this protein (30). However, this latter effect is not specific for ATP and can be achieved with dTTP or ADP. Although neither of these nucleotides is hydrolyzed, both of the effectors bind to recA protein and inhibit its ATPase and DNA binding activities (31, 3 2 ) . Furthermore, ATP and ATP[yS] influence recA protein-protein interactions by inducing filamentation of the protein without hydrolysis.

100

LI

\

90

80

-E

70

0

z

s3

60

u)

w

50

-I

s

40

30 20

10 0

0

2

4

6 TIME ( m i d

8

10

12

FIG.3. Effects of NTPs, NDPs, and their analogs on the dissociation of recA proteinsingle-strandedDNA complexes (31).

453

22. r w A ENZYME OF E. coli

IV. Reactions Catalyzed

Several reactions catalyzed by recA protein have been studied in considerable detail. These reactions differ remarkably in their requirements for recA protein, nucleoside triphosphate, polynucleotide, salt, pH, and divalent cations. A brief description of these reactions is presented in Section A to E below. A.

SINGLE-STRANDED DNA-DEPENDENT NTPASE

In the presence of single-stranded (ss) DNA or deoxyhomopolymers, recA protein catalyzes the hydrolysis of rATP, dATP, rUTP, dUTP, rCTP, and dCTP to the corresponding dinucleotide and inorganic phosphate. The relative rates of hydrolysis are as follows: dATP, rATP 2 dUTP, rUTP > dCTP, rCTP (Table I). Hydrolysis is stimulated by DNA and deoxyribohomopolymers that exceed 12-18 residues in length, but not by shorter polymers. The rate of ATP hydrolysis is linear with enzyme concentration above 0.2 p M , independent of pH between pH 5.5 and 9.0, inhibited 50% by 200 mM KCI, and shows optimal activity above 5 mM MgC12. Mn2+ion substitutes poorly for Mg2+in the ATP hydrolysis reaction, and Cay+and Zn2+ are completely inactive. At 37", the turnover number is approximately 10 moles of ADP formed/mole recA proteidmin (-32). At least two groups have reported that the KkTPfor the enzyme in the TABLE I

SPECIFICITY OF NUCLEOSIDE TRIPHOSPHATE HYDROLYSIS BY recA PROTEIN^ CATALYZED Turnover number" Nucleoside triphosphate

ssDNA (pH 7.5)

dsDNA(pH 6.2)

rATP dATP rUTP dUTP rCTP dCTP rGTP dGTP dTTP

7.8 9.44 3.74 6.16 2.11 3.12 0.16 0.31 0.3 1

2.0 3.04 1.24 1.72 0.80 1.24 0.14 0.50 0.14

,I Reaction conditions are described in Ref. (32). Moles NDP formedhole recA proteidmin

454

KEVIN McENTEE AND GEORGE M . WEINSTOCK

ssDNA-dependent reaction is relatively high, between 0.5 and 1.O mM (23, 26, 4 2 ) . However, careful measurements of the initial velocity of the reaction indicate that the KATPis actually very low, approximately 20 p M (32). The higher values obtained for KATP may be due to the inhibitory effects of ADP on the reaction since in one case the rate determinations were made when approximately 50% of the ATP was hydrolyzed (42). Unlike ATP hydrolysis catalyzed by recA protein, UTP (dUTP) hydrolysis is dependent upon pH with an optimum near pH 6.2. At this pH the rate of hydrolysis is approximately twice that observed at pH 7.5. Other differences such as salt sensitivity, substrate K , , and enzyme concentration dependence distinguish this UTPase activity from the ATPase activity ofrecA protein (43). Steady-state kinetic evidence, nevertheless, indicates that ATP competitively inhibits UTP hydrolysis and, conversely, UTP competitively inhibits ATP hydrolysis. These results, as well as the results of photoaffinity labeling experiments, indicate that ATP and UTP share a common catalytic site on the recA protein (32, 40, 4 3 ) . Single-stranded DNAs and most but not all deoxyribohomopolymers are effectors for both ATPase and UTPase activities. Poly(dG) is incapable of stimulating recA protein-catalyzed ATP or UTP hydrolysis due to the inability of this polynucleotide to bind to recA protein (3I , 32). On the other hand, several ribohomopolymers bind efficiently to recA protein but fail to stimulate its ATPase or UTPase activities. Short oligonucleotides such as dT12 do not stimulate recA protein-catalyzed ATP or UTP hydrolysis (Table 11). The mutant tifl product is altered in its interaction with cofactors for ATP hydrolysis. Certain ribohomopolymers stimulate hydrolysis by the mutant enzyme (K. McEntee, unpublished), but short oligonucleotides (dTI2) do not. The active form of the recA enzyme for ATP hydrolysis is not known, although it is likely to be an oligomer, perhaps a tetramer (18, 39). Hill coefficients determined for the hydrolysis reactions indicate that more than one ATP molecule must be bound to the active form of the enzyme for maximal activity (39). Interestingly, at pH 6.2 the Hill coefficient is lower than at pH 7.5, suggesting that the active form of the enzyme depends upon pH (39).Since the enzyme structure is remarkably different at these pH values, it is tempting to speculate that the changes in quaternary structure are responsible for the reduction in Hill coefficient. A qualitatively similar result has been obtained for UTP hydrolysis catalyzed by recA protein (43). 42. T. Shibata, R. P. Cunningham, C. Das Gupta, and C. M. Radding, PNAS 76, 5100 (1 979).

43. G . M. Weinstock, K. McEntee, and I. R. Lehman, JBC, in press (1981).

455

22. r r r A ENZYME OF E. c d i TABLE I1

HYDROLYSIS OF NUCLEOSIDE TRIPHOSPHATES BY recA PROTEIN"

Turnover numbeln Polynucleotide

ATP

UTP

6.4 6.9
8.8 7.6
N.D.' 13.2
'' Reaction conditions are described in Ref. (32) "

Moles NDP formedmole rerA proteidmin N.D. = not determined

B. DUPLEXDNA-DEPENDENT ATPASE In the presence of duplex DNA, recA protein catalyzes ATP hydrolysis at rates approaching those of the ssDNA-dependent reaction (2 I, 32 1. Unlike the latter reaction, however, the duplex DNA-dependent activity is strikingly pH-sensitive with an optimum at or near pH 6.2 (32). It is noteworthy that the optimal conditions for binding duplex DNA and hydrolyzing ATP in its presence are the conditions that promote ATPdependent oligomerization of recA protein (-?I,S2, 38). Duplex DNAdependent ATP hydrolysis depends nonlinearly on the enzyme concentration, presumably because of a cooperative binding of recA protein to duplex DNA (31, 3 ) .Hydrolysis of ATP is accompanied by release of recA protein from the double-stranded (ds)DNA, a result that led to the suggestion that ATP hydrolysis is needed to induce a conformational change that releases recA protein from the duplex effector (21, 3 1 , 40). ATP hydrolysis in the presence of duplex DNA is considerably more salt-sensitive than the ssDNA-dependent ATPase reaction because recA protein binding to duplex DNA is more sensitive to increased ionic strength (31 ). Although the ssDNA-dependent and dsDNA-dependent ATPase activities of vecA protein have several different characteristics, they display the same relative nucleoside triphosphate specificity (Table I) and are inhibited competitively by the ATP analog, ATP[ySl (32). At

456

KEVIN McENTEE AND GEORGE M. WEINSTOCK

higher pH values, the kinetics of the dsDNA-dependent ATPase reaction become nonlinear; hydrolysis proceeds after a lag of several minutes at pH 7.5. The requirement for low pH in this reaction is not due to pH effects on the DNA but is very likely due to significant changes in enzyme conformation that were described earlier (32).

C. REASSOCIATION OF SINGLE-STRANDED DNA recA protein catalyzes homologous pairing of single-stranded DNA chains (strand annealing) coupled to hydrolysis of ATP. The kinetics of recA protein-catalyzed strand reassociation are first-order with respect to DNA concentration (Fig. 4B) (K. McEntee, unpublished), a result that is in striking contrast to spontaneous renaturation or renaturation promoted by DNA binding proteins, both of which are kinetically second-order reactions (44 -46). These results indicate that the rate-limiting step in recA protein-catalyzed reassociation occurs after the formation of a recA protein-DNA complex. In addition, the rate of reassociation is linearly dependent upon enzyme concentration when the DNA substrate is in excess (between 20-100 nucleotides per recA protein monomer) (K. McEntee, unpublished). Challenge experiments with heterologous singlestranded DNA demonstrate that addition of noncomplementary DNA inhibits renaturation within two minutes. This result, together with the ability of ATP to enhance dissociation of recA protein from DNA, suggests that strand reassociation is either nonprocessive or modestly processive under the experimental conditions used (3I ). Although recA protein catalyzes the hydrolysis of UTP as efficiently as ATP in the presence of single-stranded DNA, the rate of DNA reassociation is 3-5 times greater in the presence of ATP compared to UTP (Table 111) [Ref. (19), and K. McEntee, unpublished]. This difference between the hydrolytic reactions and DNA reassociation may indicate that the oligomers of recA protein induced by ATP are important in the renaturation reaction but are not obligatory for nucleoside triphosphate hydrolysis. The role of recA protein in catalyzing strand reassociation is not simply to remove secondary structure from the single-stranded DNA chains (47). Under certain conditions (in the presence of the nonhydrolyzable ATP[yS] 44. J . G. Wetrnur and N. Davidson, J M B 31, 349 (1968). 45. B. M. Alberts and L. Frey, Nature (London) 227, 1313 (1970). 46. C. Christiansen and R. L. Baldwin, JMB 115, 441 (1977). 47. T. Shibata, C. Das Gupta, R. P. Cunningham, and C. M. Radding, PNAS 77, 2606 ( 1980).

457

22. recA ENZYME OF E . cnli

TABLE 111

REQUIREMENTS FOR recA PROTEIN-CATALYZED SINGLE-STRANDED DNA REASSOCIATION Reaction mixture

DNA in duplex (pmol)h

Complete" -recA protein - ATP -ATP + U P -ATP + GTP -ATP + dTTP +ADP (1.7 mM) - MgCI, - MgCI2 + MnCI, (10 mM)

1384

<5 104 422 41 34 167 45 66

The conditions for the reaction are essentially those described in Ref. (19) DNA resistant to S 1 nuclease digestion (19) ('

analog) recA protein can disrupt secondary structure; however, these conditions do not favor DNA renaturation (19,40). An alternative model is that recA protein binds rapidly to single-stranded DNA and the enzymeDNA complexes aggregate due to recA protein-protein interactions. According to this model, the rate of renaturation is accelerated due to the high local DNA concentration within tl-: complexes. Although features of this model are consistent with the properties of several basic proteins that tend to aggregate in solution and promote DNA renaturation (481, its application to the recA protein-catalyzed reaction must be modified to account for the role of ATP in this process. recA protein fails to promote a significant rate of renaturation in the absence of ATP (Table 111), despite the fact the protein binds single-stranded DNA and is highly aggregated under these conditions (31,38). Moreover, ATP stimulates dissociation of recA protein from DNA, thereby hindering formation of protein-DNA aggregates. Thus, recA protein-catalyzed reassociation cannot be explained by the simple aggregation model. D.

STRAND ASSIMILATION OR STRAND UPTAKE

recA protein promotes pairing between single-stranded and homologous duplex DNAs. The products of this reaction, which is coupled to ATP 48. M. M. Cox and I. R. Lehman, Ntrcleir Acids Res. 9, 389 (1981).

458

KEVIN McENTEE AND GEORGE M. WEINSTOCK

hydrolysis, are displacement loops (D loops) containing exogenous single strands base-paired to complementary regions in the duplex. The additional complexity of this reaction compared to strand reassociation involves the unwinding of a region of the duplex DNA in order to render sequences available for pairing with the invading single-stranded DNA. Originally strand uptake was observed using superhelical duplex DNA substrates (20). Later it was demonstrated that a variety of relaxed DNAs, including circles containing nicks or gaps as well as linear duplexes, are substrates for the strand assimilation reaction ( 2 1 , 4 9 4 1 ) .Joint molecules are also formed between closed circular double-stranded DNA and gapped DNA, indicating that recA protein can promote the intermolecular transfer of a DNA strand from one duplex molecule to another (SO, S/). Presumably this reaction requires unwinding of both the “donor” and “recipient” duplexes prior to, or concommitant with, annealing of the invading strand. Although they appear varied, the strand assimilation or transfer reactions promoted by recA protein are limited by the topological constraints of the DNA molecules themselves (51). Thus, recA protein cannot promote strand transfer between a closed circular duplex DNA and a circular single-stranded molecule. These results are consistent with the inability of recA protein to transiently interrupt a phosphodiester bond in either a single- or double-stranded DNA substrate. Unlike strand reassociation that is efficiently catalyzed by recA protein when single-stranded DNA is in excess, strand uptake reactions require a high concentration of recA protein that is determined by the amount of single-stranded DNA participating in the reaction. Thus, when recA protein is in excess over single-stranded DNA, the rate of strand assimilation increases linearly with single-stranded DNA concentration. Optimal strand uptake occurs at a ratio of 3-5 nucleotides of single-stranded DNA per monomer of recA protein ( 2 0 , 2 / ) .This value is close to that observed for saturating the single strands with recA protein (411). Increasing singlestranded DNA beyond this ratio results in inhibition of strand assimilation (Fig. 4A) (52). This inhibition by excess single-stranded DNA is reversed by addition of single-stranded DNA binding (SSB) protein which acts by masking the single-stranded DNA (Fig. 4). Under these conditions, the rate of strand uptake increases linearly with the single-stranded DNA and recA protein concentrations (47, 52). 49. E. Cussuto, S . C. West, J. Mursalim, S. Conlon, and P. Howard-Flanders, P N A S , in press (1980). 50. R. P. Cunningham, C. Das Gupta, T. Shibata, and C. M. Radding, Cell 20,223 (1980). 51. C. Das Gupta, T. Shibata, R. P. Cunningham, and C. M. Radding, Cell 22,437 (1980). 52. K . McEntee, G. M. Weinstock, and I . R. Lehman, PNAS 77, 857 (1980).

459

22. recA ENZYME OF E . coli

-

600 -

+SSB

-

A

0

4

I

I

1

I

I

8

12

16

20

24

SS DNA (pM)

FIG. 4. The rate-dependence of single-stranded DNA concentration for (A) strand assimilation and (B) strand reassociation. In (A) the maximal rate occurs at 4 nucleotides of ssDNAlrrcA monomer. SSB protein stimulates strand assimilation in the presence of excess single-stranded DNA (ssDNA) (A), whereas SSB protein inhibits reassociation of ssDNA (B). [Ref. (-52) and K . McEntee, unpublished results.) It has been possible to dissect the strand assimilation reaction into several partial reactions (steps) and to study the requirements, substrates, and products of each step. As a first step, recA protein preferentially binds to single-stranded DNA (21, -?/I. In a subsequent step, this proteinsingle-stranded DNA complex interacts with duplex DNA in the presence of ATP or ATP[yS]. When the nonhydrolyzable analog ATP[ySl is used instead of ATP, stable complexes are formed that contain recA protein, ATP[yS], duplex DNA, and single-stranded DNA (42, 52). No such com-

460

KEVIN McENTEE AND GEORGE M . WEINSTOCK 70

I

f

f

I

I

f

I

1

f

1

60

50

-ag s Z

z

40

L

K

a Z

n

30

N N 0

20

10 A I

-0

400

800

1200

1600

SS P22 DNA or SS P22 DNA SSB

2000

(prnole)

FIG.5 . SSB protein stimulation of recA protein-DNA complexes in the presence of ATP. TherecA protein-duplex P22 DNA complexes were measured by a nitrocellulose filter binding assay (52). Assimilation was measured in the same experiment by treating the reaction with sodium dodecyl sulfate (1%) before filtering (.?/, 2). plexes are detected with ATP. Homology between the single-stranded and duplex DNAs is not necessary for complex formation and, because intact circular DNA molecules can participate in the reaction, free ends or strand interruptions are not required in either DNA (40, 42). These recA protein-DNA complexes survive incubation at high ionic strength (1 M NaCl) for more than 60 minutes at 37” but are disrupted by detergent treatment. No covalent attachment of recA protein to DNA or ATP[yS] has been detected in these complexes (40). The duplex DNA appears to be partially unwound, perhaps at regions corresponding to where recA protein is tightly bound. The combined binding and unwinding of duplex molecules by recA protein does not require hydrolysis of a nucleo-

22. recA ENZYME OF E . coli

46 I

side triphosphate, but requires binding of an appropriate cofactor (40, 42, 53). The recA protein-DNA complexes can be formed with ATP when single-stranded DNA binding (SSB) protein is included in the reaction (Fig. 5 ) (52). The role of SSB protein may be to decrease the rate of ATP hydrolysis as well as to retard dissociation of recA protein from singlestranded DNA in the presence of ATP. SSB protein also prevents excess single-stranded DNA from inhibiting complex formation in the presence of ATP or ATP[yS] (Fig. 5 ) (52). Complexes formed with heterologous or homologous DNAs in the presence of ATP[yS] are held together by protein-DNA interactions, as demonstrated by the fact that their exposure to 1% sodium dodecyl sulfate quickly releases the DNA. recA protein also promotes complexes between duplex DNA and single-stranded ribo- and deoxyribohomopolymers that bind to the enzyme (K. McEntee, unpublished). Unlike complex formation, which does not require ATP hydrolysis, strand assimilation is blocked by the ATP[yS] analog (21,421. Presumably, ATP hydrolysis is required for aligning the homologous sequences of the duplex and single-stranded DNAs. The mechanism of this reaction is not completely understood but this alignment process is rate-limiting for the overall uptake reaction. Any mechanism must explain the observation that a limited amount of recA protein-dependent strand assimilation occurs in the presence of ATP[yS] when SSB protein is included in the reaction (52). The assimilation of single strands into duplex DNA promoted by recA protein may result in a hybrid joint containing anywhere from a few base-pairs to several hundred base-pairs formed between duplex and homologous single-stranded DNAs (21, 42, 50, 51). It has been found that complete strand displacement or exchange can occur between linear 6x174 duplex DNA and single-stranded viral circles (M. Cox and I. R. Lehman, personal communication). In addition to recA protein and ATP, the reaction is greatly stimulated by SSB protein. The products of this reaction, 6x174 RFII and linear single strands, are readily separated from the substrate DNAs. The reaction appears to be a coupled unwinding process, which indicates that the combined action of recA protein and SSB protein can initiate strand transfer and enzymatically drive branch migration at a crossover junction for several thousand base-pairs (Fig. 6).

53. R. P. Cunningham, T. Shibata, C. Das Gupta, and C. M. Radding, Nature (London) 281, 191 (1979).

a

2

0

4

I-

3 U

0

.-

+

a

a

a

a

m

m

v1 v1 TJ

m

TJ

22. recA ENZYME OF E. coli

463

E. PROTEASE ACTIVITY In addition to catalyzing DNA reassociation and assimilation reactions, recA protein is a highly specific protease (22, 26). Three substrates for this proteolytic activity have been identified: The repressors of bacteriophages A (25, 54) and P22 (23,55),and thelexA protein (29, 56), which is believed to repress the recA gene in vivo as well as other SOS functions (14,26,29). The protease cleaves each of these proteins into two fragments. In the case of A repressor, the site of this cleavage is located within a proposed “hinge” segment connecting DNA-binding and protein-interacting regions of the repressor. The sites for cleavage of phage P22 repressor and /exA protein have not been identified. Proteolysis by recA protein requires nucleoside triphosphate and polynucleotide cofactors. Unlike strand annealing and assimilation, recA -dependent cleavage of A repressor occurs more rapidly in the presence of ATP[yS] than in the presence of ATP (26). This result argues that net ATP hydrolysis is not likely to be required for repressor cleavage. Based upon the DNA binding properties of recA protein, one can reasonably conclude that a recA protein-single-stranded DNA complex is the active form of the enzyme required for proteolysis. Whether recA protein dissociates from the polynucleotide during or following proteolysis of repressors is not known. The absolute rate of repressor cleavage is slow compared to the hydrolytic, annealing, and assimilating activities of recA protein, and appears to be determined by the rate of dissociation of repressor dimers into monomers, this latter form of repressor being the substrate for proteolysis by recA protein (55). Like strand assimilation, proteolysis of A repressor is inhibited by excess polynucleotide (greater than 3-5 nucleotideslrecA protein monomer). This effect is on therecA protein rather than on the repressor ( 5 5 ) . Binding proteins (SSB protein and phage T4 gene 32 protein) eliminate this inhibition, presumably by a mechanism similar to their effect on strand assimilation (Weinstock, G. M., and McEntee, K., in preparation). Repressor cleavage by recA protein is inhibited by ADP, which suggests that ATP hydrolysis could actually inhibit proteolysis. By replacing Mg2+ with Mn2+in the reaction, the rate of repressor cleavage increases markedly in the presence of ATP, whereas the rate of ssDNA-dependent hydrolysis of ATP is reduced to less than 10% (Weinstock, G. M., and McEntee, K., in preparation). This result is taken as additional evidence 54. J. W. Roberts, C. W. Roberts, and D. W. Mount, P N A S 74, 2283 (1977). 55. E. M. Phizicky and J. W. Roberts,JMB 139, 319 (1980). 56. R. Brent and M. Ptashne, P N A S 77, 1932 (1980).

464

KEVIN McENTEE AND GEORGE M. WEINSTOCK

that ATP hydrolysis is not required for repressor cleavage by recA protein. The tif/ mutant form of the recA protein is altered in its cofactor requirements for proteolysis of A repressor. Although proteolysis by this mutant enzyme still requires a single-stranded nucleic acid cofactor, short oligonucleotides (12- 16 residues) support tijl protein-directed cleavage of A repressor, whereas these same oligonucleotides are ineffective in stimulating the recA+ form of the enzyme. Additionally, poly(rU) serves as a cofactor for the tiJl form of recA protein in both protease and ATPase activities but does not stimulate the recA+ enzyme (K. McEntee, unpublished results). The fact that short single-stranded oligonucleotides activate the tifl variant enzyme suggests an explanation for the properties of this mutant in vivo (see Section V1,B).

V.

Assays for Recombination

Several assays have been developed for detection and measurement of joint molecules formed by recA protein action. The products of these reactions are not mature recombinant molecules per se but are presumed to be early intermediates in homology-dependent recombination. In addition to electron microscopy, filter binding and agarose gel assays have been used to quantitate and characterize the reaction products. The filter binding assay relies on the ability of nitrocellulose filters to retain DNA structures with single-stranded regions in the presence of high salt. Strand assimilation promoted by recA protein hybridizes a single-stranded DNA molecule to a homologous region within a labeled duplex DNA. These joint molecules contain an unpaired, locally displaced strand that can be retained on nitrocellulose filters (Fig. 7). It is important to note that a small amount of exonuclease contamination can also result in the retention of duplex material on nitrocellulose filters, and several controls must be performed in order to rule out this possibility. Controls include measurements of the amount of acid-soluble radioactivity produced during the reaction, demonstration that only homologous single-stranded DNA leads to retention on the filter, and that the joint molecules can be destroyed by incubating at elevated temperatures to encourage branch migration of the assimilated strand. Although superhelical DNA is not required for strand uptake it is a convenient substrate to use because assimilation of single-stranded DNA removes negative supertwists from the duplex molecules. The removal of the topological turns alters the electrophoretic migration properties of the molecules in an agarose gel. Approximately 10 base-pairs formed

465

22. recA ENZYME OF E . coli 90

I

1

I

I

I

1

10

20

30

40

50

60

80

-ae

I

70

0

60

-4 50

U 40 0 I

U

K

E

30 20 10

0

TIME ( m i d

FIG. 7 . Kinetics of recA protein-catalyzedstrand assimilationin the presence of DNA binding proteins (SSB and T4 gene 32 protein (P32) and nonhomologous DNA (M13SS, bX174SS). Measurements were made using a nitrocellulose filter binding assay for retaining joint molecules (21, 52).

during strand assimilation remove one negative superhelical turn. Thus removal of all superhelical turns produces a joint molecule that is retarded relative to the substrate DNA. The assimilated molecules may contain single-stranded tails of varying lengths, which further alter the migration of these molecules in agarose gels (Figs. 8 and 9).

IV.

Biological Role

Mutants of Escherichia coli that are altered in recA gene function show striking pleiotrophic effects. In addition to their inability to perform homologous genetic recombination, recA - mutants are unusually sensitive to killing by ultraviolet and ionizing radiation, as well as to several mono-

466

KEVIN McENTEE AND GEORGE M . WEINSTOCK

FIG.8. Agarose gel assay for recA protein-catalyzed strand assimilation. Supercoiled 4x174 RFI, homologous linear single strands, ATP (or ATP[yS]) and increasing amounts of recA protein were incubated at 37" and the reaction mix was electrophoresed in an agarose gel (0.7%). The positions of the RFI and assimilated molecules are indicated. Complete assimilation occurs at a ratio of 1 recA monomer/4 nucleotides of single-stranded DNA. Over the same range of recA protein concentrations, little or no assimilation is observed in the presence of ATP[yS] (.?I, 51 ). and bifunctional alkylating agents (I, 3, #).Unlike other recombination deficient strains (recBC- or recBC-recF-), recA - strains cannot be mutagenized by ultraviolet irradiation, and considerable DNA degradation occurs following such treatment (7, 10). Moreover, A prophage is not inducible by DNA damaging agents inrecA - cells, and its repressor protein is not cleaved, as in the case of recA+ lysogens (8, 9, 57). Several mutations located within the recA gene have demonstrated that the recombination functions ofrecA protein can be genetically separated from the inducible functions of this protein. The lexB30 mutation blocks prophage A induction and expression of other SOS functions (such as mutagenesis and W-reactivation), but has little or no effect upon homologous recombina57. J. W. Roberts and C. W . Roberts, PNAS 12, 147 (1975).

FIG. 9. The products of r e c A protein-catalyzed strand assimilation. Joint molecules (D loops) formed between 6x174 RFI and homologous single strands in the presence ofrecA protein and ATP ( 1 recA monomeri3.5 mucleotides of 6 X viral single-stranded DNA). The arrows indicate the position of the displacement loops. Electron microscopy was performed as described (19).

468

KEVIN McENTEE A N D GEORGE M. WEINSTOCK

tion (34,58).The tifl mutation shows normal levels of homologous recombination, as well as expression of SOS function. However, in addition tifl mutant strains express SOS functions at 42” in the absence of DNA damage (7). Revertants of the tifl mutations show a wide variety of phenotypes including complete loss of recA gene function (33). Considerable genetic and biochemical evidence suggests that the DNA annealing and assimilation activities of recA protein are directly involved in homologous recombination and DNA repair processes, and its proteolytic activity is important for its role in coordinately regulating a set of genes that are expressed following DNA damage. A N D ASSIMILATION in vivo A. STRAND ANNEALING

In cells lacking recA gene function, no heteroduplex molecules or other recombination intermediates can be detected (59451, indicating that the block to this process is at an early step. Furthermore, 4x174joint molecules, formed nonenzymatically in vitro, can produce recombinant phage in recA- mutant cells (66). A conditional recA- mutation, recA629, results in a cold-sensitive RecA- phenotype; cells are RecA+ at temperatures above 37” and RecA - at temperatures below 30”. The recA protein purified from such a mutant has properties consistent with this phenotype. The recA629 protein catalyzes annealing and strand pairing at 37” but not at 28”. The wild-type recA protein catalyzes these reactions at 28” and 37” (19). The phenotype of the r ~ c A 6 2 9mutation also extends to sensitivity to ultraviolet light and radiomimetic agents. This result indicates that a considerable amount of DNA repair, presumably postreplication repair, is the result of annealing and strand transfer between daughter strands (24).The filling of daughter strand gaps that are created during incomplete replication of a damaged chromosome is analogous to the assimilation reaction, and is thought to involve hybridization of a single-stranded region to its homolog. +

58. B. W. Glickman, N . Guijt, and P. Morand, Mu/. Gen. Genet. 157, 83 (1977). 59. B. M. Wilkins, J . Bacteriul. 98, 599 (1969). 60. R. M. Benbow, A. J. Zuccarelli, and R. L. Sinsheimer, P N A S 72, 235 (1975). 61. H. Potter and D. Dressler, PNAS 73, 3000 (1976). 62. E. A. Birge and K . B. Low, J M B 83, 447 (1974). 63. H. E. N . Bergmans, W. P. M. Hoekstra, and E. M. Zuidweg, Molec. Gen. Genet. 137, l(1975). 64. J. Doniger, R. C. Warner, and I. Tessman, Nature (London) New B i d . 242,9 (1973). 65. J. R. Bedbrook and F. M. Ausubel, Cell 9, 707 (1976). 66. W. K . Holloman and C. M. Radding, P N A S 73, 3910 (1976).

22. rrcA ENZYME OF E . coli

469

B. ROLEOF THE PROTEASE ACTIVITYin vivo The protease activity of recA protein is necessary for induction of SOS functions. Cleavage of phage A repressor as well as the lexA protein, which behaves genetically like a repressor of several genes, occurs following DNA damage or expression of the tij'l mutation at 42". Noninducible mutations in these repressor genes (hind- and lexA-) prevent cleavage of these proteins in recA+ cells (57). In vitro neither the hind- repressor nor the lexA- product is cleaved by recA protein from the ti3 mutant (29, 54). Since the lexA protein may repress several genes (including the recA gene, the uvrA excision-repair endonuclease subunit and genes involved in septation inhibition, sj) the endopeptidase activity provides a novel means of coordinate gene regulation. Single-stranded DNA is probably an important controlling element in vivo for recA protease activity. In undamaged cells high levels of recA+ protein do not cause prophage induction (/4, 28), a result that indicates that another factor, perhaps single-stranded DNA, is limiting. Treatments that create single-stranded DNA regions stimulate prophage induction and other SOS functions. The tifl mutation leads to prophage induction and expression of SOS functions in the absence of DNA damage. fn vitro the purified ti$ protein cleaves A repressor in a reaction dependent upon single-stranded polynucleotide. However, unlike the recA+ protein, the t i j enzyme can utilize short single-stranded oligonucleotides such as dTlz, which do not bind to recA+ protein (K. McEntee, unpublished). Thus the@ proteinin vivo may be activated by short single-stranded regions near the replication fork, or other limited regions of single-stranded DNA that occur during gap filling. Such short single-stranded regions would be accessible to the ti$ protein but not to the vecA+ protein. It is not clear what form of single-stranded DNA serves as an effector in vivo. DNA degradation is correlated with induction, and evidence for induction by a specific oligonucleotide degradation product of DNA has been reported (67). In the absence of DNA degradation, single-strand gaps produced by incomplete replication have also been implicated (68, 69). C.

ADDITION ROLEOF recA PROTEIN in Vivo

Several other cellular processes are affected by mutations at the recA locus but it is not yet clear which enzymatic activities of the protein 67. M. Oishi, C. L. Smith, and B. Friefeld, C S H S Q B 44, 897 (1979). 68. J. W. Little and P. C. Hanawalt, Mol. Cen. Gene?. 150, 237 (1977). 69. R. C. Bockrath and P. C. Hanawalt, J . Bacreriol. 143, 1025 (1980).

470

KEVIN McENTEE AND GEORGE M. WEINSTOCK

are involved. The extent of DNA degradation during repair is influenced by the level of recA protein (70) suggesting a stoichiometric requirement for recA protein in controlling DNA degradation. This may simply reflect a protection of single-stranded DNA regions from nucleases (principally, exonuclease V) by binding of recA protein (70). Stable DNA replication (71), a damage-inducible process, has a continuous requirement for recA protein (72), implying a direct participation of recA protein rather than a regulatory role. In addition to these effects, recA - mutations decrease viability in the absence of DNA damage (73), are inviable in combination with mutations in the polA locus (74), and impair growth of temperate phage P1 (75). These results indicate that the recA protein functions in numerous cellular processes. VII. Research Applications

The strand assimilation activity of recA protein has proved useful for development of an in vitro site-directed mutagenesis technique (76 1. Briefly, the procedure utilizes recA protein to assimilate specific singlestrand restriction fragments into homologous DNA duplexes. The displaced single-strand loop created in the reaction is cleaved with nuclease S1, producing molecules with nicks localized to a region defined by the assimilated single-stranded segment. Nicks are converted to short gaps by exonuclease treatment (5’-3’exonuclease of M . futeus DNA polymerase I) and the resultant gapped molecules treated with sodium bisulfite to preferentially deaminate cytosine in the single-stranded region. Molecules are then repaired by DNA polymerase action and transformed into cells. This technique permits the generation of mutations in any predetermined region of a DNA molecule. ACKNOWLEDGMENTS The authors’ work reported here was supported in part by grants from the National Institutes of Health (GM 061%) and the National Science Foundation (PCM74-00865) to Dr. I. R. Lehman. In addition, K. M. with a Senior Fellow of the American Cancer Society (California Chapter) and G. M. W. was supported by a Bank of America-A. P. Giannini Fellowship. 70. 71. 72. 73. 74. 75. 76.

G. Satta, L. J. Gudas, and A. B. Pardee, Mol. Gen. Genet. 168, 69 (1979). T. Kogoma and K. G. Lark,JMB 52, 143 (1970); ibid. 94, 253 (1975). K . G. Lark and C. A. Lark, CSHSQB 43, 537 (1979). J. E. Miller and S. D. Barbour, J. Bacreriol. 130, 160 (1977). M. Monk and J . Kinross, J. Bacreriol. 109, 971 (1972). S. Zabrovitz, N. Segev, and G. Cohen, Virology 80, 233 (1977). D. Shortle, D. Koshland, G. M. Weinstock, and D. Botstein, PNAS 77, 5375 (1980).

Site-Specijic Recombination Protein of Phage Lambda HOWARD A. NASH

I. Introduction and Perspectives . . . . . . . . . . . . . . . . . .

471

11. Purification and Properties . . . . . . . . . . . . . . . . . . . . 111. Reactions Involving Int . . . . . . . . . . . . . . . . . . . . . A. Specific Binding to DNA . . . . . . . . . . . . . . . . . . .

473 474

B. Topoisomerase Activity . C. Integrative Recombination D. Excisive Recombination . IV. Biological Role . . . . . . . V. Research Applications . . .

1.

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

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

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

474 476 477

478 479 479

Introduction and Perspectives

Recombination between the genomes of E. coli and bacteriophage A is a key step in the lysogenic pathway of this virus-host system. As first hypothesized by Campbell (I), recombination between the phage and bacterial circular chromosomes is reciprocal and creates a single circular 1. A. M. Campbell, A d w ) ~Genet. . 1 I , 101 (1962).

47 1 THE ENZYMES, Val. XIV Copyright @ 1981 by Academic Press. Inc. All rights of reproduction in any form reserved ISBN 0-12-122714-6

47 2

HOWARD A . NASH

chromosome in which the viral DNA is inserted into the DNA of its host [reviewed in (21, (311. The recombination that accomplishes the integration of the viral genome is limited to specialized sites on both the viral and host chromosomes. The viral site, attP, and the host site, atfB, are functionally distinct. This is reflected in the difference in their size-uttP spans a region of about 250 base pairs whereas attB involves no more than 30 pairs (4-6). However, the two sites are identical over a 15 base-pair region that is called the attachment site core (7). Crossover between urtP and atfB occurs within this core, yielding a pair of hybrid sites, atfL and attR, that flank the integrated viral or prophage DNA. It was early suggested that viral functions might play a role in site-specific recombination (8). Subsequently, a phage gene, inr, was identified by mutations that depressed A lysogeny without affecting either the ability of the virus to infect cells or to repress the lytic viral functions (2, 3). The int gene has been the subject of intensive genetic and physiological studies; the polypeptide it encodes, Int, is the subject of this chapter. Many other temperate bacteriophages exhibit site-specific integration. Some appear to use the same bacterial site as does A; others integrate at specific sites that are remote from uttB [ reviewed in ( 9 ) ] .Although some of the most important of the early work on int gene function involved these phages, no biochemical studies on their site-specific recombination or inr gene product have been reported. They remain a potential source of useful comparative information. Integrative recombination between attB and attP, and excisive recombination between attL and utrR proceed efficiently in vitro (10, I I ). This has provided a functional assay for the purification of the proteins involved in site-specific recombination. As a result, Int protein has been shown to be a direct participant in these reactions. However, it should be pointed out that we do not know if Int has the capacity to function as a catalyst in recombination since it is usually supplied in stoichiometric or greater amounts relative to DNA substrate: thus it is not certain whether Int 2. R. A . Weisberg, S. Gottesman, and M. E . Gottesman, in “Comprehensive Virology” (H. Fraenkel-Conrat and R. R. Wagner, eds.), Vol. 8, p. 197. Plneum, New York, 1977. 3. H . A. Nash, Curr Topics Microbiol Irnmunol 78, 171 (1977). 4. P.-L. Hsu, W. Ross, and A . Landy, Nature (London) 285, 85 (1980). 5. M. Mizuuchi and K. Mizuuchi, P N A S 77, 3320 (1980). 6. K . Mizuuchi, R. Weisberg, L. Enquist, M. Mizuuchi, M. Buraczynska, C. Foeller, P.-L.Hsu, W. Ross, and A. Landy, C S H S Q B 45, 429 (1981). 7. A. Landy and W. Ross, Science 197, 1147 (1977). 8. E. R. Signer and J. R. Beckwith, J M B 22, 33 (1966). 9. A. M. Campbell, “Episomes.” Harper, New York, 1969. 10. H. A. Nash, P N A S 72, 1072 (1975). 1 1 . S . Gottesman and M. Gottesman, P N A S 72, 2188 (1975).

23. RECOMBINATION PROTEIN OF PHAGE A

473

fulfills the strict definition of an enzyme. In addition to promoting recombination, purified Int has been shown to bind specifically to DNA sequences within attP and attB and to possess topoisomerase activity. These activities are presumably related to the primary function of Int in recombination, and therefore provide insight into the mechanism of action of this protein. II.

Purification and Properties

Most purifications of Int have begun with cells in which the in? gene is expressed from a mutant promoter (12). This promoter, pintc,expresses the gene efficiently even in the absence of the phage-encoded activators that are normally required for the production of Int [reviewed in (13)]. Recombinant plasmids bearing the in? gene and the constitutive promoter (14) appear to be the most convenient source of Int (15, 16). In the author's laboratory 80 liters of midlog cells bearing this plasmid typically yield 2-5 mg of highly purified Int (17). The purification of Int is characterized by high affinity of the protein for phosphocellulose and DNA-cellulose (15, 18, 19). The most highly purified material displays a single polypeptide of MW -40,000 on gel electrophoresis under denaturing conditions (15, 19, 20). Int from cells infected with wild-type (pint+) phage has the same subunit molecular weight and purification behavior as Int expressed from Pintc (19). The molecular weight of the purified protein agrees with that determined for the int gene product in studies of phage A proteins labeled after infec12. K. Shimada and A. Campbell, PNAS 71, 237 (1974). 13. 1. Herskowitz and D. Hagen, Annu. Rev. Genet. 14, 399 (1980). 14. A. Honigman, S.-L. Hu, and W. Szybalski, Virology 92, 542 (1979). 15. Y. Kikuchi and H. Nash, CSHSQB 43, 1099 (1979). 16. R. W. Davies, P. H. Schreier, M. L. Kotewicz, and H. Echols, Nucleic Acids Res. 7 , 2255 (1979). 17. The original description (18) of the preparation of the calcium phosphate gel for the purification of Int failed to note that the calcium phosphate precipitate was adjusted to pH 7.4 with 1 M acetic acid and rinsed exhaustively with water before the final resuspension. K. Abremski (personal communication) has found that chromatography of Int on hydroxylapatite is a satisfactory alternative to the calcium phosphate cellulose step of the procedure of ref. (IS). 18. Y. Kikuchi and H. A. Nash,JBC 253, 7149 (1978). 19. M. Kotewicz, E. Grzesiuk, W. Courschesne, R. Fischer, and H. Echols, JBC 255, 2433 (1980). 20. The purified protein is quite stable when stored at -70". The earliest preparations of purified Int (18)were unstable upon dilution unless DNA was present in the diluent but this has not been observed with Int prepared from plasmid-containing cells.

474

HOWARD A. NASH

tion of ultraviolet irradiated cells [reviewed in (21)]; in addition the DNA sequence of the int gene corresponds to a protein of molecular weight 40,330 (22,23). It thus appears that little processing of theint gene product occurs in vivo or during purification. However, the amino terminus of purified Int is serine (M. Waxdal, Y. Kikuchi and H. Nash, unpublished observation), indicating that the N-terminal formylmethionine is removed. In the high ionic strength solutions in which it is purified, Int behaves as a monomer. Under these conditions, Int has a sedimentation coefficient of 3.0 S (18,241. Furthermore, as determined by gel permeation chromatography on Sephadex G-150, the diffusion coefficient is 7.57 x lo-' cm sec-I, indicating a Stokes radius of 29w ( Y .Kikuchi and H. Nash, unpublished observation). Assuming a partial specific volume of 0.725 cmYg, a native molecular weight of 36,000 is calculated. These measurements also indicate that Int is a rather asymmetric protein having a frictional ratio of 1.28, the value expected for a prolate elipsoid whose major and minor axis are in the ratio of 5.5 : 1. The Int protein purified by Kotewicz and colleagues may be even more asymmetric (or partially unfolded) since a higher Stokes radius and a lower sedimentation coefficient are observed (19). Although Int behaves as a monomer at high ionic strength, we do not know if this behavior is changed at the low ionic strengths typical of recombination reactions.

111.

Reactions Involving Int

A. SPECIFIC BINDINGTO DNA The specific interaction of Int and DNA that contains attP was first discovered in filter binding studies using crude preparations of Int (25). Purified Int also forms complexes with attP that are retained on filters; such complexes are characterized by a lifetime greater than 20 minutes, resistance to challenge by heparin, and an apparent dissociation constant of 2 x lo-" M (18, 19). Specific complexes between Int and attP have also been visualized in the electron microscope; the size of the compIex indicates that 8-10 Int monomers are located at attP (26). The interaction between attP and Int has been resolved at the nucleotide sequence level. 21. 22. 23. 24. 25. 26.

N. Katzir, A. Oppenheim, M. Belfort, and A . B. Oppenheim, Virology 74,324 (1976). R. H. Hoess, C. Foeller, K. Bidwell, and A. Landy, PNAS 77, 2482 (1980). R. W. Davies, Nucleic Acids Res. 8, 1765 (1980). Y. Kikuchi and H. A. Nash, PNAS 76, 3760 (1979). D. Kamp, Ph.D. Thesis, University of Cologne, 1973. D. Hamilton, R. Yuan, and Y. Kikuchi, J M B . in press (1981).

23.

RECOMBINATION PROTEIN OF PHAGE A

475

Footprinting experiments, i.e., determination of the DNA sequences protected by Int from digestion by nucleases or other reagents, show that Int binds to 4 distinct sites that are distributed over the 250 base-pair atfP region (4, 27). These 4 sites are designated: PI (- 148 to - 129), P2 (- 116 to -98), core (- 17 to + 19), and P‘ (+ 50 to + 86): (The location of each site is given in parenthesis in coordinates measured from the center of the 15-base-pair core.) It is clear that the binding sites are different from one another. For example, the PI and P2 sites are approximately 17-base-pairs long while the core and P’ sites each cover about 35 base pairs. Moreover, after challenge by heparin, Int continues to protect the P‘ site but not the others (27). However, sequence homologies are found between the P l , P2, and P‘ sites. Specifically, the sequence 5’-G-T-C-A-C-T-A-T-3‘, or a closely related derivative of this sequence, is found in P1, in inverted orientation in P2, and as a directly repeated pair in P’ (4).This sequence is not found at the core site. Indeed, no significant homologies have been noted between the core and the other sites, suggesting that Int may have two separate binding specificities. Several of the binding sites contained within attP have been shown to interact with Int, even when isolated from their usual contexts. For example, restriction fragments of attP that contain only PI or P2 can each be specifically, albeit weakly, retained on membrane filters (16). In addition, a fragment that contains u f t sequences only to the right of position +66 is also retained on filters (16). This means not only that the P’ site can interact with Int in the absence of the other sites contained in attP but also that half of the P’ site suffices for this interaction. Nuclease protection studies also indicate that the P1 and P‘ sites can each interact with Int in the absence of the other sites ( 4 ) . It is tempting to speculate that each G-T-C-A-C-T-A-T sequence comprises an Int binding site. The larger size of the P’ site and its resistance to heparin might follow from the close positioning of two of these binding sequences in the P’ region. The core binding site ofnftP is of particular interest since it is at this site that the recombination crossover occurs. The binding of Int to sequences at this site ( - 17 to + 17) can be compared to the interaction of Int with attB. Although filter binding assays failed to detect specific complex formation between Int and a f t B (15, 18, / 9 ) , a region around the core of arrB (-15 to +4) is protected by Int from nuclease digestion (27). Thus, the core regions of uttB and attP are both protected, but to different extents. This implies that sequences outside, perhaps immediately flanking, the core are important in specifying interactions with Int. No detailed studies concerning equilibrium and kinetic aspects of the 27. W. Ross, A. Landy, Y. Kikuchi, and H. Nash, CeN 18, 297 (1979).

476

HOWARD A. NASH

interaction of Int with the various binding sites in attP and arrB have been reported. In addition there are no published analyses at the nucleotide sequence level of the interaction of the prophage sites with Int.

B. TOPOISOMERASE ACTIVITY Int can reversibly break and reseal the backbone of DNA. This activity is conveniently assayed by the relaxation of negatively (or positively) supercoiled DNA to closed circles lacking supertwists. The ability of Int thereby to convert one topological form of DNA to another identifies Int as a protein with topoisomerase activity. The evidence supporting the hypothesis that the topoisomerase activity is intrinsic to the Int protein includes the cosedimentation of recombination and relaxation activities, the parallel inactivation of these two activities by heat and specific antiserum, and the relative deficiency of both activities in crude extracts of cells expressing a mutant int gene (24). Relaxation of DNA by Int is accomplished in the absence of cofactors like ATP or NAD that are required for ligation of hydrolyzed phosphodiester bonds. This probably means that, as for other topoisomerases, the breakage step in relaxation by Int involves a covalent intermediate in which a phosphodiester bond links the DNA and protein. However, no such intermediate has been detected. Int appears to be a type I topoisomerase (28), i.e., an enzyme that nicks one strand of the double helix and subsequently reseals it, rather than a type I1 topoisomerase that breaks both strands of DNA simultaneously [for review see Chapters 18 and 19, this volume, and Ref. (2911. This is inferred because the change in linking number caused by Int is not restricted to multiples of 2 (28). Relaxation of DNA by Int is inefficient: on the average one molecule of purified Int removes one turn from DNA in 100 minutes (24). However, Int may function catalytically in relaxation; given enough time one molecule of Int can remove all 30 turns from more than one DNA molecule. Relaxation is not more efficient on circular DNA that contains an attachment site than on one that lacks such a site (24).This means that either Int topoisomerase is nonspecific or, if Int is a specific topoisomerase its specificity involves a szquence within the attachment site that is also present at moderate frequency in other contexts. One appealing possibility is that the active site for the topoisomerase coincides with the region of Int that specifies binding to the core of u t t P . This might define a domain of Int concerned with strand exchange. A second domain of Int, differing 28. H.A.Nash,K.Mizuuchi,L. W. Enquist,andR.A.Weisberg,CSHSQB45,417(1981). 29. M. Gellert, Annu. Rev. Binchern. 50, 879 (1981).

23.

RECOMBINATION PROTEIN OF PHAGE A

477

both in sequence specificity and sensitivity to inactivation by N-ethylmaleimide (24), would be concerned with the interaction of Int with the arms of attP.

C. INTEGRATIVE RECOMBINATION The definitive reaction that involves Int is recombination between atrP anduttB to produce the prophage sites attL andattR. Depending upon the arrangement of the attB and attP sites on the DNA substrate for this reaction, the product can take one of several forms. For example, if aftP and attB are each on a separate circle of DNA, the product is a single circle that encompasses the two substrate circles (30).If attP and attB are both located on the same circular substrate, two outcomes are possible. If the sites are oriented in a head to tail fashion, attL andattR are each found on a separate product circle; the two circles are usually linked to one another in a catenane (31). On the other hand, ifattB andattP are oriented head to head, recombination between them inverts a segment of the substrate circle, yielding a circle that is the same size as the substrate but bears atfL andatrR; this product circle is often knotted (32). Each of these cases, as well as others involving combinations of circular and linear DNAs, can be conveniently assayed by electrophoresis or reaction mixtures on agarose gels since the recombination-produced rearrangement usually changes the size, mobility, or restriction profile of the substrates. Other assays that examine the genetic (10) or other physical (18) consequences of recombination are also available, but these are judged to be relatively inconvenient. There are two unconditional requirements for integrative recombination in vitro-Int and IHF, a protein factor that can be obtained from uninfected, nonlysogenic E. coli (33). No other proteins are required. Under standard conditions of buffer and salt, integrative recombination also requires supercoiling of the substrate DNA (35),particularly attP (30). This requirement is bypassed in reaction mixtures carried out at lower ionic 30. K. Mizuuchi and M. Mizuuchi, CSHSQB 43, 1111 (1979). 31. K. Mizuuchi, M. GeIlert, R. A. Weisberg, and H. A. Nash,JMB 141, 485 (1980). 32. K. Mizuuchi, L. M. Fisher, M. H. O'Dea, and M. Gellert, PNAS 77, 1847 (1980). 33. IHF activity has been purified; it consists of two polypeptides (33a). One is the product of the E. coli himA gene (33b) and the other is probably coded for o r controlled by the E. coli hip gene (34). 33a. H. A. Nash and C. A. Robertson, JBC, in press (1981). 33b. H. I. Miller and H. A. Nash, Nature (London) 290, 523 (1981). 34. H. I. Miller, A. Kikuchi, H. A. Nash, R. A. Weisberg, and D. I. Friedman, CSHSQB 43, 1121 (1979). 35. K. Mizuuchi, M. Gellert, and H. A. Nash, J M B 121, 375 (1978).

478

HOWARD A. NASH

strengths (.36), but recombination of nonsupercoiled DNA is always at least 5- to 10-fold slower than recombination of supercoiled substrate. The final requirement is for a polycation, Mg2+or spermidine. This dependence is reduced in reactions that are carried out at elevated ionic strength ( 3 3 ~ It ) . is important to note that a high-energy cofactor is neither required nor stimulates integrative recombination. Under all conditions tested, the bulk of the products of recombination are continuous double helices containing attL and attR. No nicks or gaps have been found to be associated with recombination ( 3 / ) .Taken together with the simple cofactor requirements, this means that recombination is not accomplished by separate steps involving hydrolysis of the phosphodiester backbone by a nuclease and subsequent rejoining of the backbone in a novel configuration by a ligase. Instead, a topoisomeraselike mechanism that permits cleavage and rejoining without an external energy source must be used. A model that suggests how Int topoisomerase activity might be used in recombination has been presented (28). This mechanism also accounts for the fact that most of the superhelical turns in the substrate are retained in the product circles (3/1. Integrative recombination in vitro can be a vigorous reaction. Under optimal conditions, over 50% of the substrate is converted by purified Int and IHF to recombinant products in 3-5 minutes ( 3 3 ~ )Typically, . the amount of Int required to achieve these levels of recombination is quite large (30-50 monomers of Int per recombinant formed). The meaning of this ratio is uncertain not only because the proportion of active molecules in purified Int preparations is unknown but also because the amount of Int required to produce a given amount of recombinant varies with the amount of substrate DNA. Integrative recombination is maximal at pH values between 7.7 and 8.7, at KCI concentrations below 150 mM and at temperatures between 25"-35". D. EXCISIVE RECOMBINATION Recombination between the prophage attachment sites, utrL and attR also takes place efficiently in v i m (/I). As in the case of integrative recombination, excisive recombination requires a supertwisted substrate at elevated ionic strength (36, 37); it is not yet known whether this requirement pertains to only one of the two prophage attachment sites. Under one set of conditions excisive recombination displays the identical protein

36. T. J. Pollock and K. Abremski, JMB 131, 651 (1979). 37. K. Abremski and S. Gottesman, J M B 131, 637 (1979).

23. RECOMBINATION PROTEIN OF PHAGE A

479

requirements as integrative recombination, Int and IHF, suggesting a simple reversal of the integrative reaction (5). However, at a higher ionic strength, that may be more representative of physiological conditions, excisive recombination requires a third protein-probably the product of the xis gene (Abremski and Gottesman, manuscript in preparation). The purification of this protein is underway but little is known of its interaction with Int, IHF, or DNA.

IV.

Biological Role

Promoting the integration-excision cycle of A virus is undoubtedly the primary function of Int protein. Bacteriophage containing any one of several hundred different mutants in theinr gene show little or no alteration in physiology other than a defect in integration and/or excision (usually both recombinations are affected) (38).Under special circumstances, the ability of Int to promote recombination may have alternative consequences. For example, in E. coli deleted for attB, X can integrate at low frequency into one of many alternative attachment sites. Integration into any of the secondary sites that are distributed around the E. coli chromosome depends entirely on a functional inr gene (39). More subtle roles for Int protein pertain to the growth of A under conditions where general, homologous recombination is blocked. Here, the ability of Int to promote recombination between the attP sites on two phage chromosomes can lead to recombination of genetic markers that span the crossover locus or, more rarely, . addition, since Int-promoted recombination is are adjacent to it ( 4 0 , 4 / ) In reciprocal, this should produce dimeric A circles (42), that are preferred substrates for the packaging of phage DNA. Thus, under some restrictive conditions, Int can have an effect on phage yield.

V.

Research Applications

The ability of Int to promote site-specific recombination in vitro has been valuable in studying the characteristics of attachment sites. For example, the extent of DNA sequence required for attachment-site func38. 39. 40. 41. 42.

L. W. Enquist and R . A. Weisberg, J M E 1 1 I , 97 (1977). K . Shimada, R. A. Weisberg, and M. E. Gottesman, J M B 63, 483 (1972). L. W. Enquist, H. Nash, and R. A. Weisberg, PNAS 76, 1363 (1979). H. Echols and L. Green, Genetics 93, 297 (1979). D. K. Chattoraj, CeN 19, 143 (1980).

480

HOWARD A. NASH

tion was determined by measuring the efficiency of site-specific recombination in vitro with substrates derived by resection and subcloning of plasmids that contain a f f Por attB (4-6). As a second example, the structure of the recombination crossover locus has been determined at the molecular level by measuring the nucleotides transferred from one parent to the recombinant attachment sites during in vitro recombination (6). In these examples, as well as others (30, 3 / ) ,Int has been successfully used as a reagent to permit the study of the recombination mechanism itself. Site-specific recombination may prove to be a useful cloning tool. Since Int promotes reciprocal recombination, any circle of DNA that contains an attachment site can be incorporated by site-specific recombination in a single step into a second DNA that also bears an attachment site. In this way, a plasmid can be added to a phage genome, forming a phasmid (43)or two plasmids can be fused to make a composite (30). This union by sitespecific recombination can be carried out in vivo or in vitro. An extreme example of this kind of additive recombination has been suggested by M. Ptashne. He has pointed out that the injection of oocyte nucleii with Int, IHF, and a circular DNA containing aftP could result in the insertion of the injected DNA into the oocyte chromatin, presumably at an attB-like fertilization and implantation may then produce stable chimeras. ACKNOWLEDGMENT I thank Drs. N. L. Craig and M. Gellert for their criticism of this manuscript.

43. J. Karn, S. Brenner, L. Barnett, and G. Cesareni, €“AS 77, 5172 (1980).

Photoreactivating Enzymes BETSY M . SUTHERLAND

. I . Introduction I1. Purification and Properties . . . . . . . . . . . . . A . Streptomyces griseus Enzyme . . . . . . . . . B Bakers’ Yeast Enzyme . . . . . . . . . . . . . C . Escherichia coli Enzyme . . . . . . . . . . . . D . Anacystis nidulans Enzyme . . . . . . . . . . E . Mammalian Enzymes . . . . . . . . . . . . . 111. The Reaction . . . . . . . . . . . . . . . . . . A . Nature of the Reaction . . . . . . . . . . . . B . Reaction Requirements . . . . . . . . . . . . C . Kinetics of Photoreactivation . . . . . . . . . . D . Intermediates: The Enzyme-Substrate Complex . E . Mechanism . . . . . . . . . . . . . . . . . . F. Control of Enzyme Synthesis and Function . . . IV. Biological Role . . . . . . . . . . . . . . . . . . A . Molecular and Cellular Photoreaction . . . . . . B . Mutants in Photoreactivation . . . . . . . . . . V . Research Applications . . . . . . . . . . . . . . A . Photoreactivation in Vitro . . . . . . . . . . . B. Photoreactivation in Vivo . . . . . . . . . . . C. Practical Considerations . . . . . . . . . . . . .

.

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

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . . .

. . . . .

. . . . .

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

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

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

482 485 486 487 489 491

4w 493 493 496 501 502 504 508 510 510 510 512 512 513 514

48 1 THE ENZYMES. Vol . XIV Copyright 0 1981 by Academic Press. Inc . All rights of reproduction in any form reserved ISBN 0- 12-1227 14-6

482 1.

BETSY M. SUTHERLAND

Introduction

Photoreactivating enzymes (PRE) also called photolyases (EC 4.1.99.3) catalyze the light (300-600 nm)-dependent monomerization of cyclobutyl pyrimidine dimers, formed between adjacent pyrimidines on the same DNA strand, upon exposure to ultraviolent (UV) irradiation (220-320 nm). Although much is known about the substrate and product of these unusual enzymes, their identification required the development and synthesis of such fields as photochemistry, biochemistry, and microbiology. Photoreactivation was first known as a biological recovery phenomenon: Cells exposed to visible lightfollowing UV irradiation showed higher survival than those kept in the dark. Early investigators examined the photoreactivability of an enormous range of cellular damage in both prokaryotes and eukaryotes. The first real insight into the molecular basis of photoreactivation came with the pioneering work of Goodgal, Rupert, and Herriott ( I ) . Using the Huemophilus injuenzae-transforming DNA assay, they showed that the photoreactivation reaction was mediated by a photoreactivating enzyme. The enzyme followed Michaelis-Menten kinetics in its actions: First, the formation of an enzyme-substrate complex, with UV-irradiated DNA as substrate; and second, upon exposure to visible light, the repair of that DNA (2,3). Although the fragmentary knowledge of cellular photobiology and photochemistry did not allow identification of the lesion(s) serving as substrate in this reaction, the experiments with transforming DNA and the solid background pinpointing DNA as the cellular target for UV damage indicated that nucleic acid photochemistry was to play an important role. The essential idea of reversibility accompanied the beginnings of nucleic acid photochemistry. In 1949 Sinsheimer and Hastings found that UV irradiation of uracil in solution led to a disappearance of its characteristic absorption spectrum (4). Heat or acid treatment restored the absorbance. Although Errera had shown as early as 1952 that UV irradiation of DNA led to destruction of pyrimidines (3,the nature of the molecular alteration of the DNA was entirely unknown. The first insight into these UV-induced changes was provided by Beukers et af., who found that thymine irI . S. H. Goodgal, C. S. Rupert, and R . M. Herriott, in “The Chemical Basis of Heredity” (W. D. McElroy and B. Glass, eds.), p. 341. Johns Hopkins Univ. Press, Baltimore, 1957. 2. C . S . Rupert, J . Gen. Physiol. 45, 703 (1962). 3. C. S. Rupert, J . Gen. Physiol. 45, 725 (1962). 4. R. L. Sinsheimer and R. Hastings, Science 110, 525 (1949). 5. M. Errera, EBA 8, 30 (1952).

24. PHOTOREACTIVATING ENZYMES

483

radiated in ice also showed decreased absorbance ( 6 ) , and that further irradiation of the thawed solution led to reappearance of the characteristic absorption spectrum (7, 8). They suggested that this reversible alteration of DNA by UV might be related t o biological photoreactivation. The involvement of thymine alteration in UV photobiology became more likely after the demonstration that UV irradiation of DNA led t o the disappearance of cytosine and thymine from apurinic acid (9), the identification of thymine dimers in DNA (10, I / ) ,and their UV reversibility (12, 13). There were two critical questions concerning the relation of the thymine dimer-monomer conversion to biological photoreactivation: (1) Were dimers the substrate for the enzyme and monomers its product? (2) Did photoreactivation proceed via the UV-reversal mechanism? Evidence against the direct UV reversal mechanism arose quite soon. First, Wang (12) showed that irradiation of thymine dimers in solution with 370 nm light (which was highly efficient in driving the enzyme-mediated photoreactivation reaction) did not lead to the appearance of thymine monomer. In 1961 Setlow measured an action spectrum for reversal of thymine dimers in solution, showed that direct UV-reversal had an entirely different wavelength dependence than the enzymatic or biological photoreactivation reactions, and concluded that direct reversal was not the mechanism underlying biological photoreactivaton (13). Was the thymine dimer substrate for the photoreactivation reaction? Wulff and Rupert (14) provided the first evidence that this was the case; they showed that thymine dimers disappeared from UV-irradiated DNA treated with photoreactivating enzyme in the presence of light. However, this conclusion was challenged by several additional findings. Rupert found that neither thymine dimer, thymidine dimer, dinucleotide TpT in which the thymines were dimerized, UV-irradiated apurinic acid, nor a preparation of d A : dT acted as a competitive inhibitor of the yeast enzyme acting on UV-irradiated-transforming DNA (15). Moreover, UVirradiated dG : dC (which contained no thymine dimers) did compete. R. Beukers, J . Ylstra, and W. Berends, Rec. Trav. Chim. 77, 729 (1958). R. Beukers, J. Ijlstra, and W. Berends, Rec. Trci~’.Chim. 78, 883 (1959). R. Beukers, J. Ijlstra, and W. Berends, Rec. Tru18. Chim. 78, 879 (1959). R. Beukers, J. Ijlstra, and W. Berends, Rec. Truv. Chim. 78, 247 (1959). 10. R. Beukers and W. Berends, BBA 41, 550 (1960). 11. A . Wacker, H. Dellweg, and D. Weinblum, Nafurwissenschajien 47, 477 (1960). 12. S. Y . Wang, N u f w e (London) 188, 844 (1960). 13. R . Setlow, BBA 49, 237 (1961). 14. D. L. W u E and C. S. Rupert, BBRC 7, 237 (1962). 15. C. S. Rupert, Phofochem. Phorobiol. 3, 399 (1964).

6. 7. 8. 9.

484

BETSY M. SUTHERLAND

These apparent contradictions were resolved as follows: As Rupert (15) suggested and Setlow and Bollum (16) showed, the dimer-containing species smaller than 10 nucleotides in length were too small to be bound by the enzyme. J. K. Setlow, Boling, and Bollum showed that the homopolymer duplex dA : dT, in which pyrimidine dimer formation can be shown by spectroscopy or by chemical analysis, competes for the enzyme, while the alternating polymer dAdT: dAdT (in which no dimers are formed) does not (17). They also showed that polymers containing dimers of cytosine (dI : dC, dG : dC and dC) were photoreactivable, indicating that both thymine-containing and cytosine-containing dimers were substrates for the enzyme. Further, Setlow and Setlow showed that dimers were the only substrate for the enzyme (18). If dimers were the substrate for the enzyme, what was the product of enzyme action on the dimer? The idea of reversible damage to pyrimidines, which stemmed from the early nucleic acid photochemistry, provided a working hypothesis: In analogy with direct UV-reversal of thymine dimers to thymine in solution, the enzyme might mediate reversal of pyrimidine dimers to monomers in DNA. Indeed the experiments of Wulff and Rupert (14) seemed to support just this interpretation: They found that dimers disappeared from the DNA upon action of the enzyme and light. However, Wulff (19) pointed out that another interpretation of the data was possible: Since their procedure involved extraction of the DNA after enzyme treatment, they might have missed dimers solubilized into small DNA fragments by the enzyme. Setlow showed that this was not the case by repeating the Wulff and Rupert experiment, but measuring the total dimer content of the samples rather than that in the highmolecular-weight DNA fraction (20). This experiment showed that dimers had indeed disappeared from the reaction mixture, and were not merely transferred into a low-molecular-weight fraction. [Note that the idea of transfer of dimers into a low-molecular-weight fraction is a precursor to the later experiments of Setlow and Carrier (21) and Boyce and HowardFlanders (22), which showed that cells were able to excise dimers from their DNA without the light-dependence of photoreactivation.] These experiments did not show whether the dimers were actually reversed to monomers, or were converted to unknown products. Setlow, Carrier, and 16. 17. 18. 19. 20. 21. 22.

J . K. Setlow and F. J. Bollum, EEA 157, 233 (1968). J. K . Setlow, M. E. Boling, and F. J. Bollum, PNAS 53, 1430 (1965). J. K . Setlow and R . B. Setlow, Nature (London) 197, 560 (1963). D. L. Wulff, Thesis, California Institute of Technology, Pasadena, California, 1962. J. K. Setlow, Photochem. Photobioi. 3, 405 (1964). R. B. Setlow and W. L. Carrier, PNAS 51, 226 (1964). R. Boyce and P. Howard-Flanders, PNAS 51, 293 (1%4).

24. PHOTOREACTIVATING ENZYMES

485

Bollum provided direct evidence for the monomerization of dimers by the enzyme (23). They irradiated the synthetic polynucleotide dI :dC, then heated it to convert the C[lC dimers to U[]U dimers. (Note that monomer Cs deaminate at a much lower rate than do 5,6-saturated cytosines in the dimer.) After photoreactivation, hydrolysis and chromatography, they found that monomer U appeared, which resulted from the photoreactivation of U[]U dimers. Cook showed that photoreactivation of thyminecontaining dimers in DNA also resulted in production of thymine monomers (24). These experiments-which defined the role of the enzyme in photoreactivation, its substrate, and the product of the reaction-provided the basics for biochemical and biophysical study of the enzyme. These major characteristics are shared by all DNA photoreactivating enzymes characterized to date, although, as discussed in this chapter, the structures and mechanisms of dimer photolysis of the enzymes from different organisms may vary greatly. Biological photoreactivation and photoreactivating enzymes are almost ubiquitous, but two groups of organisms were found early t o be deficient in photoreactivation (25)-transformable bacteria and mammals. Soon revisions of this generalization were necessary. Photoreactivation was found in one group of mammals, the marsupials (26), and later among the placental mammals (27).Esclzerichiu cofi, which possesses photoreactivating enzyme, was found to be transformable under special conditions (28). Photoreactivating enzyme levels seem to be regulated within an organism, so PRE activity may vary from tissue to tissue within an organism, for different growth stages of a culture, according to the developmental stage of the organism, or for the same cells in different culture conditions. The effectors are unknown [except in the case of E. coli, in which adenine auxotrophs starved for adenine show greatly increased PRE activity (29)l. 11.

Purification and Properties

I n vitro photoreactivating activities from a wide variety of prokaryotic and eukaryotic sources have been studied and purified to different extents. 23. R. B. Setlow, W. L. Carrier, and F. J . Bollum, PNAS 53, 1 1 1 1 (1965). 24. J. S. Cook, Phorochem. Phorohiol. 6, 97 (1967). 25. J . S. Cook,in “Photophysiology” (A. C. Giese, ed.), Vol. V, p. 191. Academic Press, New York, 1970. 26. J. S. Cook and J. D. Regan, Ncitrcre (London) 223, 1066 (1969). 27. B. M. Sutherland, Nature (London) 248, 109 (1974). 28. S. P. Cosloy and M. Oishi, PNAS 70, 84 (1973). 29. H. Nishioka and W. Harm, Mutat. Rex. 16, 121 (1972).

486

BETSY M. SUTHERLAND

Frequently, extensive purification has been impeded by the lability of the enzyme, both in cellular extracts and purified preparations. Although many properties of the enzyme, especially those of action spectra, have been successfully studied in cell extracts and have proven useful for correlation with in vivo photoreactivation action spectra, I discuss here only enzymes that have been purified to apparent homogeneity or contain, at most, only traces of contaminating proteins. A.

Streptomyces mgriserts ENZYME

Eker and Fichtinger-Schepman purified a PRE from S . griseus to apparent homogeneity by sonication, centrifugation, and filtration; then chromatography on Spherosil type D, UV-irradiated DNA-cellulose, DEAE-cellulose, and single-stranded DNA-agarose (30). The resulting preparation shows a single major protein band after polyacrylamide gel electrophoresis, which coincides with the major band of enzyme activity. [It is interesting to note that some of the S . griseus PRE activity remained at the origin, a problem traced in E. coli PRE to its nature as a glycopro. protein has a molecular weight of 41,000, and shows a pH tein ( 3 / ) ]The optimum at 7.0, and ionic strength optimum of 0.04 (A. Eker, personal communication). The action spectrum for cellular photoreactivation extends from 350 to 500 nm, with a peak about 435 nm (32); thus the presence of a moiety associated with the Streptomyces enzyme that showed an absorption maximum at 445 nm (and a shoulder at 425 nm) indicated that the enzyme might possess an intrinsic chromophore responsible for absorption of photoreactivating light (33). Eker has presented several additional lines of evidence showing the involvement of the 445 nm-absorbing moiety in enzyme action. Although in solution at 10” the PRE dissociated from the 445-nm species with a half-life of 46 hours, addition of unirradiated DNA raised the half-life to 330 hours, and UV-irradiated DNA stabilized the enzyme entirely (33, 34). On the basis of absorption and fluorescence spectroscopy, Eker proposed that this chromophore is a 7,8-didemethyl-&hydroxy-5-deazaflavinderivative (35). Upon irradiation 30. A. Eker and A. Fichtinger-Schepman, EEA 378, 54 (1975). 31. R. M. Snapka and B. M. Sutherland, Biochemistry 19, 4201 (1980). 32. J. Jagger, R. S. Stafford, and J. M. Snow, Photochem. Photobiol. 10, 383 (1969). 33. A. P. M. Eker, “Spectral Properties of a DNA-Photoreactivating Enzyme and its Chromophoric Group from Sfreptomyces griseus, ” Abstr. Ifit. Congr. Photobiol., 7111, p. 179 (1976). 34. A. P. M. Eker, Photochem. Photobiol. 32, 593 (1980). 35. A. P..M. Eker, “Structure of the Chromophoric Cofactor of Photoreactivating Enzyme from Streptomyces griseus,” Abstr. I n t . Congr. Phofobiol.. Bth, p. 317. Strasbourg, July 20-25, 1980.

24. PHOTOREACTIVATING ENZYMES

487

with 420 nm light, such derivatives can sensitize the monomerization of thymine dimers in solution.

B. BAKERS’YEASTENZYME The high nuclease level in E. coli cells, which precluded their use with the transforming DNA assay, led Rupert to begin the first purification of a photoreactivating enzyme using yeast cells (36). The relatively low specific activity, and the low cellular PRE content soon made obvious the advantages of using commercially available bakers’ yeast as a source of the enzyme. This choice proved an immense boon in the study of photoreactivating enzyme because it provided large quantities of nuclease-free material for the development of purification procedures, as well as partially purified preparations for the extensive and elegant studies of Walter and Helga Harm and C. S. Rupert [for a partial summary, see Ref (37)]. However, problems appeared in the development of the purification procedures: Different stability of the enzyme, different behavior of the enzyme activities, and different spectral properties of the enzyme from one purification procedure to the next [(38,39) and Madden, personal communication]. These properties led Madden to propose that bakers’ yeast contains two photoreactivating enzymes with rather different chromatographic and spectral properties, as well as different molecular weights and subunit compositions (40).An enzyme preparation from a batch of bakers’ yeast may contain one, both, or neither of the PRes. (For this information, and that given in the following two sections, the author is indebted to personal communications from Drs. Madden and Werbin.) 1. Bakers’ Yeast Enzyme I (38)

The enzyme was purified from commercial bakers’ yeast cakes by toluene lysis of the cells, ammonium sulfate fractionation, chromatography on phosphocellulose, and ammonium sulfate concentration of the eluate, followed by chromatography on UV-irradiated DNA-cellulose. The resulting preparation showed a major band coincident with enzyme activity, as well as several minor protein bands. Gel filtration in Sephadex G-200 in 36. C. S. Rupert, J . Gen. Physiol. 43, 573 (1960). 37. C. S. Rupert, A d v . Radial. Eiol. 2, 81 (1966). 38. S. Minato and H. Werbin, Eiochernistr.v 10, 4503 (1971). 39. D. T. Boatwright, J. J. Madden, J. Denson, and H. Werbin, Biochemistry 14, 5418 (1975). 40. J. J. Madden, “Yeast DNA Photolyase: Purification, Structure and Utilization as a tool for counting Pyrimidine Dimers,” Abstr. Ann. Mfg, A m . SOC. Phorobiol., 71h, p. 124 (1979).

488

BETSY M . SUTHERLAND

buffer plus 0.4 M KCI indicated an apparent molecular weight of 53,000. In addition to the protein absorbance at 273 nm, the enzyme showed a weak absorption band from 350 to 420 nm, with a maximum at about 380 nm. For excitation at 380, the enzyme had an emission maximum between 485 and 490 nm; for emission monitored at 470 nm, the excitation maximum was approximately 385 nm. It seems likely that Iwatsuki, Joe and Werbin (41) also studied the yeast PRE I of Minato and Werbin (38).Although the former group reported that their enzyme activity eluted from phosphocellulose at 0.37-0.47 MKCI, whereas the latter authors found it between 0.25 and 0.3 M KCl; both reported similar elution properties on UV-irradiated DNA-cellulose, and similar molecular weights. Further, Iwatsuki et al. found that the enzyme migrated as a protein of 51,000 5 1000 on SDS polyacrylamide gels. Taken along with Minato and Werbin’s figure of 53,000 f 1000 from gel filtration of the native enzyme, this indicates that the enzyme is a single polypeptide. Iwatsukiet al. also found similar optical properties of their enzyme to that of Minato and Werbin. It had absorbance maxima at 275 and 380 nm, fluorescence emission maximum at 460 nm (for excitation at 380 nm) and a fluorescence excitation maximum at 380 nm (for emission at 460 nm). By denaturation of the enzyme by heat or 8 M urea, they were able to isolate an oxidized flavin adenine dinucleotide from the enzyme. Examination of the optical properties of the native enzyme led them to propose that 4a,5-reduced flavin is the chromophore for yeast enzyme I(41). Although the early purification work on the yeast enzyme yielded a preparation that showed sedimentation properties of a protein of molecular weight 30,000 (42), it is possible that this preparation was also yeast enzyme I; Madden has found that partially purified preparations of this enzyme upon sedimentation give an apparent molecular weight of 30,000, but upon further purification and gel electrophoresis under denaturating conditions yield the expected molecular weight of 53,000. 2. Bakers’ Yeast Enzyme II (39) The description of a second yeast PRE by Boatwright et al. (39) in 1975 introduced a new complexity. Their enzyme, called a photolyase in accord with a nomenclature recommendation of Minato and Werbin (431, has distinctly different properties than the first yeast PRE described (36). The molecular weight Boatwright et al., obtained for their enzyme preparation was 130,000, in contrast to those of Muhammed (42), 30,000; Cook and 41. N. Iwatsuki, C. 0. Joe, and H.Werbin, Biochemistry 19, 1172 (1980). 42. A . Muhammed, J . Biol. Chem. 241, 516 (1966). 43. S. Minato and H. Werbin. Photochem. Phorobiol. 13, 97 (1972).

24.

PHOTOREACTIVATING ENZYMES

489

Worthy (44), 69,000; and Minato and Werbin (38),53,000. However, their extensive characterization left little doubt about the existence of the 130,000 dalton species. [Madden (40) has since shown that the two yeast enzymes can be resolved by elution from UV-irradiated DNA-cellulose, with enzyme I eluting at 0.4-0.7 M KCl, and enzyme I1 at 0.2-0.4 M.] Boatwright et a!. (39) found that the enzyme activity migrated to a position in gradient gels corresponding to 130,000 molecular weight, and sedimentation velocity experiments under moderate salt conditions gave a similar value (136,000). However, sedimentation under higher salt conditions (1 M KCl) gave activity at a position corresponding to 68,200 daltons, which was bracketed by two absorbance peaks with approximate molecular weights of 54,000 and 82,500. The activity peak at 68,200 daltons corresponds well with the figure of 69,000 daltons obtained by Cook and Worthy (44),whose gradients and gel filtration columns also included 1 M KC1. Boatwright et al. also found that neither of the two absorbance peaks (54,000 or 82,500 daltons) had photoreactivating activity alone, but if mixed in the presence of UV-irradiated DNA substrate, they reconstituted to give active enzyme. They obtained additional evidence for two subunits in yeast enzyme I1 SDS gel electrophoresis, in which two protein bands of 85,000 and 60,000 could be resolved. The action spectrum for the enzyme shows a broad maximum with a peak at 366 nm (45). The absorption spectrum of yeast enzyme I1 contains a broad peak at about 265 nm but, unlike that of yeast enzyme I, does not contain discernable absorption in the wavelength range 300-380 nm (46). Madden and Werbin report, however, that it does fluoresce when excited at 380 nm; its fluorescence properties are generally similar to those of the yeast enzyme I (47). It is not known which subunit is associated with the fluorescent species, nor the stoichiometry of the fluorescent moieties relative to enzyme molecules. C. Escherichia coli ENZYME Rupert er al. provided the first partial purifications of the E. coli PRE (48), but found that the high nuclease level in these cells greatly impeded the measurement of enzyme activity by the transforming DNA assay (36). Using a nuclease-insensitive assay for photoreactivation, Sutherland et al. 44. 45. 46. 47. 48.

J. S. Cook and T. E. Worthy, Biochemistry 11, 388 (1972). J. J. Madden and H. Werbin, Biochemistry 13, 2149 (1974). H. Werbin and J . J. Madden, Photochem. Phorobiol. 25, 421 (1977). J. J . Madden and H. Werbin, BBA 383, 160 (1975). C. S. Rupert, S. H. Goodgal, and R. M. Herri0tt.J. Gen. Physiol. 41, 451 (1958).

490

BETSY M. SUTHERLAND

(49) developed a purification for PRE from E. coli B, a small-scale and a

large-scale purification for enzyme from a PRE overproducer strain (SO). The preparation resulting from the latter strain showed a single rather broad band of molecular weight about 40,000 after SDS gel electrophoresis, and a marked tendency toward aggregation. The enzyme was sensitive to thermal denaturation and protease digestion. The action spectrum showed a broad band with a maximum at about 366 nm, similar to that for photoreactivation in the cells. Although the purification methods of Sutherland e? al. provided apparently homogeneous enzyme, they depended on acetone precipitation (method 2 II), and ion-exchange chromatography (all three methods), which sometimes led to large losses in enzyme activity. Snapka and Sutherland (31 ) developed an improved purification procedure that routinely yields large quantities of high specific activity, apparently homogeneous enzyme. In this procedure, cells are lysed by blending with glass beads, and the resulting cell extract is fractionated by streptomycin and ammonium sulfate precipitation (fraction 111 of Snapka and Sutherland). The resulting partially purified preparation is stable for years in 40% glycerol at -20". For further purification the enzyme is passed through Bio-Gel P-4, focused in an isoelectric focusing column with pH 4-6 ampholytes, then chromatographed on Bio-Gel P- 100. The resulting enzyme (fraction V of Snapka and Sutherland) is homogeneous by the criteria of (1) the appearance of a single precipitin band upon immunodiffusion of fraction V enzyme versus a poly-specific antiserum that recognizes multiple species in fraction 111 enzyme, (2) the appearance of a single symmetric peak upon velocity sedimentation in neutral buffer, and (3) the appearance in SDS gels of the dansyl-labeled protein as the monomer band of 35,200 -+ 200 and its higher multiples. This enzyme contains neither unusual nor modified amino acids, and seems to lack tryptophan. It contains a variable percentage of carbohydrate, which does not dialyze away from the enzyme. In the high carbohydrate fraction of Snapka and Sutherland, the composition is five mannose/seven galactosehhree glucose/three N-acetylglucosamine, with an indication of an N-glycosidic linkage to asparagine. The E. coli enzyme is also associated with a small RNA that contains ribose [as analyzed by gas-liquid chromatography ( 3 / ) ]and the four RNA bases, uracil, adenine, guanine, and cytosine (51). Spectroscopic measurements give estimates of 10-15 bases/PRE monomer; the finding by 49. B. M. Sutherland, M . J. Chamberlin, and J . C. Sutherland, JBC 248, 4200 (1973). 50. B. M. Sutherland, D. Court,and M. J. Charnberlin, Virology 48, 87 (1972). 51. R. M. Snapka and C. 0. Fuselier, Photochem. Photobiol. 25, 415 (1977).

24. PHOTOREACTIVATING ENZYMES

49 1

Koka and Sutherland (52) of RNA species isolatable from the enzyme in the size range 10-15 bases long implies that each enzyme is associated with one RNA oligonucleotide. Removal of the RNA by charcoal adsorption or dialysis results in stoichiometric loss of enzyme activity; the enzyme activity is also sensitive to pancreatic RNase digestion. The isolated RNA is cleaved by T1 and pancreatic RNase. Snapka and Sutherland ( 3 / )determined the molecular weight of the apoprotein by SDS gel electrophoresis of dansyl-labeled enzyme to be 35,200 and by molecular seiving on agarose A-5M in 6 M guanidinium chloride of the reduced and carboxymethylated enzyme to be 35,000. By following the absorbance of the RNA, they were able to obtain an estimate of the molecular weight of the holoenzyme as 36,800, assuming that the enzyme is approximately spherical in solution. Support for the spherical shape of the enzyme comes from electron micrographs of uranyl acetate-shadowed fraction V enzyme; the enzyme appears as round globules approximately 40 A in diameter (Rothman and Hainfeld, personal communication). The absence of tryptophan and low content of other aromatic amino acids, plus the presence of the RNA, lead to an absorption spectrum in the 200-300 nm region dominated by the RNA (31).Since the enzyme aggregates in solution, absorption spectra of the intact enzyme show apparent optical density above 300 nm, the region of photoreactivating enzyme action in vivo and in vitvo. However, Sutherland found that correction for scattering by experimental or theoretical means indicates that there is no true absorption in this region in the intact E. coli holoenzyme (53).The origin of the photoreactivating absorption band for this enzyme is discussed in Section 111. D . Anacysris nidulans ENZYME

A photolyase from the blue-green alga Anacystis nidulans was first purified by Saito and Werbin (54) and then more extensively by Minato and Werbin (43).Although the latter authors did not include an evaluation of the purity of their preparation, Saito and Werbin indicate that their preparation contained several protein bands detectable after electrophoresis in polyacrylamide gels under nondenaturing conditions. They found a molecular weight by gel filtration on Sephadex G-200 of 93,000. 52. P. Koka and B . M. Sutherland, “ E . coli Photoreactivating Enzyme Requires Nucleic Acid for Activity,” Ahstr. I n f . Congr. Phofohiol., 8th. p. 316 (1980). 53. J. C. Sutherland, in “DNA Repair Mechanisms,” (P. C. Hanawalt, and E. C. Friedberg, eds.), p. 137. Academic Press, New York, 1978. 54. N . Saito and H. Werbin, Biochemistry 13, 2610 (1970).

492

BETSY M. SUTHERLAND

The action spectrum had a maximum at 436 nm, and the enzyme preparation showed absorption at 418 nm as well as at 275 nm. Minato and Werbin found that their more highly purified preparation also contained the 418 nm absorbing material, which showed a fluorescence excitation maximurn at 420 nm and a shoulder at about 330 nm (for emission at 470 nm) and an emission maximum at about 470 nm (for excitation at 420 nm). E.

MAMMALIAN ENZYMES

Although early searches for photoreactivating enzymes or cellular photoreactivation in mammalian cells had been negative [for a review see Ref. (2.5)], Cook and Regan (26)found both in cells of the mammalian order Marsupialia (Calurornys derbianus, the South American woolly opossum; Potorus tridactylis, the Tasmanian rat-kangaroo or potoroo; and Didelphis marsupialis, the common North American opossum). However, no evidence could be found for a photoreactivating enzyme in other mammalian orders until Sutherland (27) isolated and characterized an enzyme from human leukocytes. She showed that the photoreactivating activity requires UV-irradiated DNA as substrate, light for catalysis, and is inactivated by heat or trypsin treatment. The enzyme elutes from a Bio-Gel P-100 column as a macromolecule of 40,000 daltons, and has an isoionic pH of 5.4. The pH optimum for the reaction is 7.2 and, unlike many other photoreactivating enzymes, has an ionic strength optimum of 0.05. [A. Eker (personal communication) has found the ionic strength optimum of the S . griseus enzyme to be 0.04.1 The enzyme catalyzes the lightdependent disappearance of dimers from DNA, converting the dimers to monomers (55). The action spectrum for the human enzyme extends from 300 to at least 577 nm, with a maximum about 400 nm. The ability of this enzyme to utilize light of the 500-600 nm region was unique among photoreactivating enzymes known at that time: however, Chiang and Rupert (56) found that the action spectrum of the PRE activity in cells of the marsupial potoroo also extends to wavelengths longer than 500 nm. Several factors may have contributed to the difficulty of enzyme detection in the early studies. First, even in lower vertebrates the enzyme is not present in all tissues nor at all developmental stages (55, 57). Second, levels of the enzyme are regulated in cells in culture (58, 5 9 ) , even in E. 55. B. M. Sutherland, P. Runge, and J. C. Sutherland, Biochernisfry 13, 4710 (1974). 56. T. Chiang and C. S. Rupert, Photochem. Photobiol. 30,525 (1979). 57. I. S. Cook and J. R. McGrath, PNAS 58, 1359 (1967). 58. B. M. Sutherland and R. Oliver, BBA 442, 358 (1976). 59. K. Mortelmans, J. E. Cleaver, E. C. Friedberg, M. C. Paterson, B. P. Smith, and G. H. Thomas, Murat. Res. 44, 433 (1977).

24. PHOTOREACTIVATING ENZYMES

493

coli (29). Third, the lower ionic strength optimum (27) and the unusual extension of the action spectrum to the 500-600 nm range (55) [which in

the usual yellow photoreactivating “safelights” is present in sufficient intensities to catalyze photoreactivation effectively by the mammalian enzyme (60) ] may have precluded detection of enzyme in mammalian cell extracts .

111.

The Reaction

A.

NATURE OF THE REACTION

Photoreactivating enzymes catalyze the light-dependent monomerization of cyclobutyl pyrimidine dimers in DNA in a two-step reaction: the enzyme binds to dimer-containing DNA to form a stable enzymesubstrate complex; upon absorption of light in the range 300-600 nm, the dimer is monomerized and the enzyme released. 1. Binding

Rupert first demonstrated the formation of a complex of photoreactivating enzyme and UV-irradiated DNA substrate in his classic 1962 paper (3) entitled “Photoenzymatic Repair of Ultraviolet Damage in DNA. 11. Formation of an Enzyme-Substrate Complex.” In the absence of photoreactivating light the complex was stable enough to permit its sedimentation in a sucrose gradient. He also showed that the reaction could be described by Michaelis-Menten kinetics (2). This and the preceding companion paper were quite remarkable for their insight into the molecular mechanism of the biological photoreactivation of transformation; Rupert provided early evidence for a DNA repair enzyme although he was also working with a crude cell extract at a time when specific damages to DNA were neither identified nor quantitated. In these papers he also introduced into the study of photoreactivating enzymes the use of competition by UV-irradiated nontransforming DNA, providing the basis for the measurement of lesions (dimers) in any polydeoxyribonucleotide. Harm and Rupert (61 ) were able to study the binding reaction further by developing a flash photolysis method. They showed that exposure of the enzyme-substrate complex to a single high-intensity, millisecond flash photolyzed all the complexes; thus by administering the flash under varying conditions or times of incubation of PRE with substrate (and com60. J . C. Sutherland and B. M. Sutherland, Biophys. J . 15, 435 (1975). 61. H . Harm and C. Rupert, Mutar. Res. 6, 355 (1968).

494

BETSY M. SUTHERLAND

petitors, if relevant), they were able to study the binding reaction alone. Since their extensive studies were carried out with a partially purified preparation that may have contained either or both of the yeast enzymes, and since there are doubtless quantitative differences among PREs from the various species, I do not attempt to discuss their work in detail. However, their major findings are applicable to most PREs. Harm and Rupert ( 6 1 ) first showed that due to the low concentrations of enzyme and substrate in their assays the binding reaction required minutes for completion. Complexes formed faster at 37" than at 5"; complexes formed at 37" did not dissociate rapidly upon shifting to 2". They were able to follow the dissociation of the enzyme-substrate complexes by adding irradiated nontransforming DNA to serve as a "sink" for released enzymes. Surprisingly, they found that PRE complexes with the irradiated synthetic polynucleotides dA :dT and dG :dC were more stable than most of those with native DNA. Madden et al. (62) later developed a filter-binding assay that enabled them to measure the binding reaction without the requirement for subsequent photolysis. They used this procedure both to locate the enzyme and to characterize the kinetics of binding (see Section 111,C). 2. Phutulysis

After the enzyme is bound to the UV-irradiated DNA substrate, absorption of a photon by the complex leads to disappearance of the dimer from the DNA and release of the enzyme. That the disappearance of the dimer was actually monomerization was established by Setlow el al. (23) in an elegant experiment based on the susceptibility of 5,6-saturated cytosines to deamination upon heating. They used the synthetic polynucleotide dI :dC labeled with ['qC]cytosine and irradiated to produce C[]C dimers, which then deaminate upon heating to give U[]U dimers. (Note that undimerized cytosine deaminates much more slowly than the C[]C dimers.) After treatment of the polymer with yeast photoreactivating enzyme and photoreactivating light, they hydrolyzed and chromatographed the polymer. Assessment of the radioactive species showed the presence of cytosine (from monomers, never in dimers), U[]U dimers (those that were not photoreactivated by the enzyme), and monomer uracil, resulting from photomonomerization of the U[]U dimers. Later Cook showed that thymine-containing dimers in natural DNAs were also converted to thymine monomers by the photoreactivating enzyme (24). A priuri there is no reason to assume that a protein required for photoreactivation acts catalytically. Rupert (36) showed that the activity did 62. J. J . Madden, H. Werbin, and J. Denson, Phorochern. Phorobiol. 18, 441 (1973).

24. PHOTOREACTIVATING ENZYMES

495

correspond to that of an enzyme: After photoreactivation to completion of one UV-irradiated DNA sample, the enzyme was able to work equally well on a second sample. Later, using samples in which both enzyme and dimer concentrations were known, Farland (personal communication) used a large excess of dimers over PREs, allowed photoreactivation to completion by the E. coli enzyme, and showed that the enzyme could also photoreactivate dimers in a second DNA sample that contained an excess of dimers. Sutherland (unpublished data) has obtained similar results for the enzyme from human leukocytes. There is a possibility that, due to photolability of the S . griseus photolyase chromophore, (see Section I1,A) the holoenzyme (protein plus cofactor) does not recycle, but must replenish its cofactor after each photolysis event. This remains in the realm of speculation. 3 . Measurement of the Complete Reaction

Photoreactivation has been measured chiefly by three methods (see Ref. 6 3 ) . In the first, the transformation assay (M), one measures the lightdependent ability of an extract to restore transforming ability to UVirradiated transforming DNA. Because this requires expression of the transforming DNA’s genetic markers (generally drug resistance) and thus overnight growth of the transformed cells into colonies on plates, it is rather slow for use during enzyme purification. It is also sensitive to nucleases in the cellular extracts: it has the advantage of great sensitivity, requiring only a few pmole of DNA. Direct measurement of pyrimidine dimer content in the DNA before and after photoreactivation forms the basis of the second assay, in which the radioactive DNA is hydrolyzed and the resulting pyrimidine monomers and dimers separated by chromatography (65). The chromatographic assay requires much larger quantities of DNA, the exact amount depending on its specific radioactivity. It is rather slow even with thin-layer chromatography, but is insensitive to nucleases and offers the great advantage of identification of both substrate and products in the reaction. A nuclease-digestion assay was developed especially for use in enzyme purification (66); it is specific, reasonably rapid, insensitive to nucleases in cell extracts, and requires only a few pmol of radioactive DNA. It depends on the resistance to nuclease-digestion of the nucleoside-phosphate 63. B. M. Sutherland, in “Molecular Mechanisms in DNA Repair” (P. C. Hanawalt and R. B. Setlow, eds.), p. 103. Plenum, New York, 1975. 64. B. J. Barnhart and R. M. Herriott, BBA 76, 25 (1963). 65. W. L. Carrier and R. B. Setlow, “Methods in Enzymology,” Vol. XXI(D), p. 230, 1971. 66. B. M. Sutherland and M. J. Chamberlin, Anal. Biochern. 53, 168 (1973).

496

BETSY M. SUTHERLAND

bond of residues involved in the pyrimidine dimer (67). Thus digestion (with DNase I, crude snake venom and alkaline phosphatase) of UVirradiated DNA yields mononucleosides, inorganic phosphate and dimercontaining oligonucleotides. In the case of the original assay, the DNA was labeled with 32P,and the [32P]oligonucleotideswere separated from 32Piby adsorption to charcoal. In the extension of this assay to [3H]thymine-labeled DNA, the [ 3 H ] ~ l i g ~ n u c l e ~ tare i d eseparated ~ from the [3H]mononucleosides by adsorption to a DEAE-filter disk (68). In either case, photoreactivation appears as a loss of radioactive counts from the charcoal or DEAE-disk.

B. REACTION REQUIREMENTS In addition to the presence of an intact photoreactivating enzyme in a suitable buffer with a dimer-containing DNA substrate, the photoreactivation raction has a unique second requirement-light. Further, the reaction has a temporal requirement; the enzyme must be associated with the first substrate, the dimer in DNA, before the absorption of the photon can occur and thus drive the photocatalysis of the dimer monomerization. (Also see Section III,F,2 for other effects of light on the enzyme.) It should be stressed that there is no evidence for storage by the PRE of the energy of the photoreactivating photon for later use in a photoreactivating reaction. Further, some PREs do not absorb 300-600 nm light (69) in absence of UV-irradiated DNA. The appearance of the absorption only in the enzyme-substrate complex might thus protect the free enzyme against inactivation during sunlight exposures. 1. Solution Conditions

As indicated in Section 11, all PREs that have been characterized are composed of an apoprotein and a nonprotein component. In the cases of the yeast enzyme I and S. griseus enzyme, the nonprotein components seem to possess the properties expected for the chromophore responsible for photoreactivating light absorption. Even though the E. cofi and yeast I1 enzymes do not absorb in the spectral region of the photoreactivation reaction, the E. cofi enzyme contains the RNA cofactor required for enzyme activity (31). The presence in the yeast I1 enzyme of the 265 nm absorption peak might lead to speculation that this enzyme also possesses 67. R. B. Setlow, W. L. Carrier, and F. J. Bollurn, BBA 91, 446 (1964). 68. W. H. Farland and B. M. Sutherland, Anal. Biochem. 97, 378 (1979). 69. K . L. Wun, A. Gih, and J. C. Sutherland, Biochemistry 16,921 (1977)

24. PHOTOREACTIVATING ENZYMES

497

a nucleic acid cofactor, but there is no additional evidence supporting such speculation. The association of PRE aproproteins and cofactors is noncovalent, as the holoenzymes dissociate upon dialysis (36, 31) or upon incubation in solution (Eker, personal communication). The presence of UV-irradiated DNA substrate stabilizes the holoenzyme to inactivation by heavy metals or by thermal denaturation (3). The enzymes have neutral pH optima, approximately 7.2, and require neither added metals nor cofactors (other than the cofactor part of the holoenzyme) to function. They differ in ionic strength optima, with the S . griseus and human enzymes requiring rather lower ionic strength conditions (0.04 and 0.05), respectively) than others, which require about 0.2 [A. Eker, personal communication, and Ref. (27)]. Most are stabilized by dithiothreitol or /3-mercaptoethanol. The ability of many of these enzymes to function even in the presence of rather high EDTA concentrations (10 mM) allows their use (even as partially purified preparations containing Mg-dependent deoxyribonucleases) in reactions requiring photoreactivation without nicking (70) (see Section V). Photoreactivating enzymes are inactivated by repeated freezing and thawing; two practical solutions to this problem are the storage at -80" of the enzyme in quantities suitable for a single experiment (Madden, personal communication), or the storage at - 20" of the enzyme in a 40% glycerol solution, which prevents freezing of the solution (see Ref. 71).

2 . The Substrate: Pyrimidine Dimers in DNA The enzyme will monomerize cis-syn cyclobutyl dimers in DNA. Other dimer configurations are not usually found in DNA, although Ben Hur and Ben Ishai (72) were able to form the trans-syn thymine dimer in minor concentrations by UV irradiation of denatured DNA. This dimer was not photoreactivated by the yeast photoreactivating enzyme. In addition to pyrimidine-pyrimidine adducts joined by the cyclobutyl ring at the 5,6 positions of the pyrimidines, one can also produce a psoralen-pyrimidine monoadduct joined by a cyclobutyl ring. Song (personal communication) produced such a DNA with radioactive 5-methoxypsoralen-pyrimidine monoadducts; Farland and Sutherland (unpublished data), and Song and his collaborators tested the ability of the E. coli and human PREs to photoreactive such adducts. Although the 70. F. E. Ahmed and R. B . Setlow, Cancer Res. 39, 471 (1979). 71. B. M. Sutherland, in "DNA Repair: A Laboratory Manual of Research Procedure (E. C. Friedberg and P. C. Hanawalt, eds.) Part B, pp. 219-227. Dekker, New York, 1980. 72. E. Ben Hur and R. Ben Ishai, BBA 166, 9 (1968).

498

BETSY M . SUTHERLAND

enzyme preparations were active in removing dimers from UV-irradiated DNA, neither showed photoreactivating activity toward the psoralenpyrimidine adducts. It should be noted that although halogenated pyrimidines (e.g., bromodeoxyuridine) can be incorporated into DNA, they do not form pyrimidine dimers upon UV-irradiation, and thus their damage is not subject to photoreactivation (see Ref. 73). In addition to thymine-thymine dimers in DNA, the enzymes can act on dimers produced between cytosine and thymine, cytosine and cytosine, uracil and thymine, or uracil and uracil (the latter two being formed from the thermal deamination of cytosine-thymine and cytosine-cytosine dimers, respectively). Because of the polarity of DNA, it should be realized that a C[]T dimer differs stereochemically from a T[]C, and thus the enzyme has at least seven dimer substrates in DNA (17, 23): the homodimers T[]T, C[]C and U[lU, as well as the heterodimers T[]C, C[]T, WIT, and T[]U (presumably the heterodimers U[lC and C[lU might also occur in partially deaminated DNA). The Setlows have shown that the cis-syn cyclobutyl dimer in DNA is the only substrate for the photoreactivating enzyme (18);direct examinations of other dimers and adducts, as previously discussed, support this conclusion. Harm produced indirect evidence that an additional photoproduct might serve as a minor substrate for cellular photoreactivation (74), but no direct tests of the possible identity of such a substrate or the actions of the various photolyases have been carried out. In addition to presence of the cis-syn dimer, the enzyme requires that the dimer be a part of a deoxyribonucleic acid polymer at least 10 bases long (16). The enzyme will not act on isolated dimers in solution, nor on dimers in small oligonucleotides. These oligonucleotides do not compete with UV-irradiated DNA for photoreactivating enzyme, so either the enzyme does not bind them, or the binding to DNA is so much stronger that the overall rate of photoreactivation is unaffected. Although cis-syn dimers are formed in RNA, the enzyme will not monomerize dimers in RNA (151, nor does RNA compete with UV-irradiated DNA, even at a ratio of UV-irradiated RNA to DNA of 3000 : 1 (Hausrath and Sutherland, unpublished data). Although photoreactivation of a number of RNAs, including E. cofi RNA phages and plant viral RNAs, has been seen [see ref. (75) for a 73. J . K . Setlow, in “Comprehensive Biochemistry” (M.Florkin, and E. H. Stotz, eds.), Vol. 27, p. 157. Elsevier, Amsterdam, 1957. 74. W. Harm, in “DNA Repair Mechanisms” (P. C. Hanawalt and E. C. Friedberg, eds.), p. 147. Academic Press, New York, 1978. 75. J. Hurter, M. P. Gordon, J. P. Kirwan, and A. D. McLaren, Phorochem. Phorobiol. 19, 185 (1974).

24. PHOTOREACTIVATING ENZYMES

499

discussion], and biological photoreactivation in an insect egg has been reported to be due to dimer reversal in RNA (76-78), the agent responsible has not been identified. Hurter et nl. (7.5) have, however, studied the in v i m photoreactivation of tobacco mosaic virus RNA by tobacco leaf extracts. They found that the activity was destroyed by boiling, precipitated with ammonium sulfate, and excluded from Bio-Rad P-100, and thus concluded that the reaction was mediated by a protein. They also found that neither the yeast PRE nor the PRE from bean seedlings was active on the UV-irradiated tobacco mosaic virus RNA. Thus it seems likely the DNA and RNA photoreactivating activities are separate entities, and their nature, composition, and mechanism of action await further characterization.

3 . The Second Requirement: Light Photoreactivating enzyme function by absorbing light in the range 300600 nm, the exact maximum and range varying between the different enzymes. The short wavelength cut-off, usually about 300 nm, probably results from the increasing ability of the shorter wavelengths to form pyrimidine dimers by direct absorption by the DNA (See Ref. 79). The maximum long wavelength cutoff presumably depends on the absorption of the chromophore of the enzyme, whether it is intrinsic t o the holoenzyme or results from absorption induced upon complex formation. Individual enzymes are frequently distinguishable by the wavelength requirements for activity. Action spectra measurements are fraught with pitfalls; Jagger (80) gives an excellent description of the methodology. A “basic” spectrum shared by the yeast and E. coli enzymes, as well as those from silkworm and maize, shows a broad, structureless peak in the 366-385 region, and does not extend beyond 500 nm. The action spectrum reported for silkworm (8/) is almost independent of wavelength in the region 300-460 nm; Muraoka et al. (81) propose that it falls into a class distinct from other enzymes that are active in the wavelength range. The action spectrum for photoreactivation in S . gviseus has a maximum at 436 nm, but does not extend beyond 500 nm (32); a shoulder at 313 nm was shown not to result from enzymatic photoreactivation (82). A similar 76. K. Kalthoff, Photochern. Photobiol. 23, 93 (1976). 77. H. Jackle and K . Kalthoff, Photochem. Phofobiol. 27, 309 (1978). 78. K. Kalthoff, K . Urban, and H. Jackle, Photochern. Photobiol. 27, 317 (1978). 79. R. B . Setlow, Science 153, 379 (1966). 80. J. Jagger, “Introduction to Research in Ultraviolet Photobiology.” Prentice Hall, Englewood Cliffs, New Jersey, 1967. 81. N. Muraoka, A. Okuda, and M. Ikenaga, Phorochem. Photobiol. 32, 193 (1980). 82. M. Ikenaga and J . Jagger, Phorochem. Phorobiol. 13, 459 (1971).

500

BETSY M . SUTHERLAND

spectrum with a 436 nm peak that does not extend beyond 500 nm was reported for the A. nidulans photolyase by Saito and Werbin (54). A third type of spectrum shows a broad peak in the 366-400 nm range [about 366 nm for marsupial PRE (56) and 400 for the human enzyme (SS)], but with the distinctive feature of extension of the spectrum to wavelengths longer than 500 nm. Rupert and To (83) have shown that the action spectrum for a yeast enzyme preparation depends on the nature of the polymer that contains and the dimer. The enzyme showed both greatly reduced efficiency (14)) altered spectral distributions for the UV-irradiated synthetic polymers poly(dG) :poly(dC) and poly(dA) : poly(dT). The spectrum for the enzyme acting on dA: dT had a single maximum at 380 nm, while on dG: dC it showed a major peak at about 355 nm, with a second, slightly smaller one at 380 nm. Rupert and To showed that these differences do not arise from the different dimers (T[]T versus C[]C) by measuring the action spectra for the same enzyme preparation acting on natural DNAs in which pyrimidine dimers were produced by UV-irradiation (which produces T[]T, C[]C and T[]C dimers) or by acetophenone sensitization [which produces only T[]T dimers (84)].These spectra were almost indistinguishable, and were strikingly different from either spectrum for photoreactivation of the synthetic polymers. Thus it is necessary to specify the substrate as well as the species of origin of the enzyme. In the previous discussion, we assumed that the photolytic step requires only one photon. Evidence that this is the case is provided by the work of Setlow and Boling (851, who found that the rate of photoreactivation of the yeast enzyme was linearly dependent on the intensity of the photoreactivating light (rather than showing a dependence on the square of the intensity, as would be expected for a biphotonic process). Similar results were obtained for photoreactivation in vivo of E. coli bacteriophage (86). Harm has reported that photolysis by the yeast photolyase system is enhanced (for some enzyme-substrate complexes) when two flashes of light are given within a short period (87). It might also be asked if the enzyme can use alternate energy sources instead of light for the catalytic reaction. Although a number of highenergy compounds were tested (J. K. Setlow, personal communication) none was found effective in driving the reaction. 83. 84. 85. 86. 87.

C. S. Rupert and K . To, Phorochem. Photobiol. 24, 229 (1976). A. A. Lamola, Photochem. Phorobiol. 9, 291 (1966). J. K . Setlow and M . E. Boling, Photochem. Photobiol. 2, 471 (1963). R. Dulbecco, J . Bacreriol. 59, 329 (1950). W. Harm,Murar. Res. 60, 121 (1979).

50 1

24. PHOTOREACTIVATING ENZYMES

c.

KINETICSOF PHOTOREACTIVATION

Most of the studies of kinetics of photoreactivation have been carried out with partially purified yeast enzyme preparations, which (as discussed in Section II,B) may have contained a mixture of two photolyases. However, the similarity of the kinetic parameters obtained using these preparations to those obtained using purified preparations implies that either the same enzyme was present in both types of preparations, or that the kinetic properties of the mixture were not distinguishable from those of the purified enzyme. Some of the properties of the E. coli enzyme have been determined in cells or in partially purified preparations, and are discussed briefly. All these studies have centered around the determination of the kinetic parameters describing the binding and photolysis reactions. These may be formally represented as E

+ . . . Py[ ]Py . . .

LI

12

L

. , . Py[ ]Py . . . h y - E + . . . PyPy . . .

(1)

_ i

E

1. The Binding Reaction

The association and dissociation of the enzyme-substrate complex have been studied by continuous illumination (2), the single flash method (61), a repetitive flash technique (88), and a filter-binding assay (45, 62). The similarity of the results, even with the different techniques and possible heterogeneity of the enzymes, allow one to draw general conclusions about the kinetic properties of the yeast enzyme. Values for k , of lo7 (2), 2.6 x lo7(61), and 2.7 x lo7 liter mol-' sec-' at 22" (44) and for kz of lov3 sec-' at 22" (45) were obtained by the (2), 1.5 x (6/), and 1.4 x three methods. These values allow calculation of K,, ( = k , / k z ) of lo'', 1.7 and 1.93 x 10" literhole and ofK,, of lo-'', 5.9 x lo-", and 5.2 x lo-" mole/liter for the three methods, respectively. In addition, by measuring k, and k, at different temperatures, energies of activation of 9.3 and 5.5 kcal/mole were obtained by Harm and Rupert for the association and dissociation reactions, respectively. The value of 6.9 kcaYmole for the dissociation reaction of Madden and Werbin agrees reasonably well with that obtained by Harm and Rupert; the former authors attribute the difference between their value for the association reaction of 4.4 kcallmole and Harm and Rupert's value of 9.3 to technical difficulties in measuring klat

0". 88. H . Harm and C . S . Rupert, Muraf. Res. 10, 291 (1970).

502

BETSY M. SUTHERLAND

Harm (89) studied the kinetics of photoreactivation in E. coli cells by a repetitive flash method, and obtained values of k l of 1 . 1 x 10' liter/mole/ to 1.3 x lo-? sec-I, yielding sec, and a range of values of k2 of 1.9 x Keq in the range 8.5 x lo' to 5.8 x 10' liter/mole and a corresponding Kd range of 1.2 x lop8to 1.7 x mole/liter. He also obtained energies of activation for the association and dissociation reactions of 1 1 and 4.5 kcdmole, respectively. As Madden and Werbin (45) pointed out, it is of interest to compare these values for the photolyase-UV-irradiated DNA system with the corresponding parameters for the lac repressor system obtained by Riggs et al. (90). They reported k, of 7 x lo9 liter/mole/sec and kz of 6.2 x lou4 sec-' at 23", yielding K , , of 9 x mole/liter and K , , of 1 x lOI3 litedmole. The greater substrate range of PREs, encompassing all possible cis-syn dimers in all possible flanking sequences, versus the unique repressor-repressor binding site relationship may account for the greater stability of the latter system.

2. Photolysis and the Complete Reaction The rate constant k3 is the reciprocal of the lifetime of the enzymesubstrate complex in the presence of photoreactivating light. Depending on the light source used, values of k3 of 10+ to 10-I sec-' were obtained by Rupert (2 [also see Ref. (88)l. It is clear that this rate constant depends both on the properties of the enzyme-substrate complex and the quantity and wavelength of the incident photoreactivating light. Thus Harm (89) and Harm and Rupert (88) have used a photolytic constant k, = k3Z where Z is the intensity of the incident light and k, varies with the wavelength of the incident light. For 366 nm, H. Harm and Rupert obtained a value of 9.5 x mm2 erg-' for the yeast enzyme, and W.Harm (89) obtained a k, value of 1.37 x mm2 erg-' for photoreactivation in E. coli cells,

D. INTERMEDIATES: THE ENZYME-SUBSTRATE COMPLEX The stability, in the absence of light, of the complex of PRE and UVirradiated DNA allowed its isolation by centrifugation or gel filtration in Rupert's early experiments (3). Although the impurity of most of the enzyme preparations precluded most characterizations of the complex, information on the absorption properties of the complex could be obtained from the action spectra of the photolytic event. This results from the dependence of the photolysis on the wavelength and intensity of the inci89. W.Harm, Mufar. Res. 10, 277 (1970). 90. A. D. Riggs, S. Bourgeois, and M. Cohn,J M B 53, 401 (1970).

24. PHOTOREACTIVATING ENZYMES

503

dent light (“concentration” of the second substrate, the photon), on the probability that an incident photon will be absorbed ( E , the molar extinction coefficient of the complex), and the probability that an absorbed photon will be effective in driving the catalytic event (a, the quantum yield). For a known wavelength and light intensity, by measuring k, or k , (see Section III,C), one can determine the product over the wavelength range effective in photoreactivation. Assuming that Q, is constant over the photoreactivation action spectrum (a reasonable assumption, as long as one absorption band is responsible for the photolysis), determination of the product E @ as a function of wavelength should give values proportional to E , and thus a predicted absorption spectrum for the complex. [Note that Jagger (80) has discussed reasons why action and absorption spectra may not agree.] For enzymes like the S. griseus and yeast enzyme I, the absorption spectra of their intrinsic cofactors seem to correspond reasonably well to the absorption spectra expected on the basis of their action spectra. However, for the E. coli, Thermobia dornestica, and yeast enzyme 11, the absorption spectra measured for the purified enzymes indicate an absence of absorption in the spectral region responsible for photoreactivation. Characterization of the latter two enzymes has not yet progressed sufficiently to provide a clear cut solution to this problem; in the case of the E. coli enzyme, Wun et a/. (69) showed that a new absorption band appeared when the enzyme bound to UV-irradiated DNA. This band was of the correct spectral range (- 300-450 nm) and molar extinction coefficient (-7000 liters/mole/cm) to be that expected for the photoreactivating chromophore. The band did not appear when the enzyme was mixed with unirradiated DNA, or with irradiated DNA at sufficiently high ionic strength to preclude enzyme-substrate complex formation, and disappeared upon depletion of the E ’ S complexes upon photoreactivating illumination. The increase in absorbance in the 300-450 nm range was accompanied by a decrease in absorbance at about 260 nm. It is not clear whether this absorbance decrease reflects changes in the RNA cofactor, the DNA substrate, or both. It is interesting to speculate that binding of the PRE involves changes in the DNA conformation, either through association with the RNA (in which case, part of the hypochromicity at 260 nm might result from increased stacking or order of the RNA bases) or with the protein, or both. Some support for the idea of changes in the RNA come from the experiments of Cimino and Sutherland (unpublished) who found that the degree of double strandedness of the RNA (while associated with the enzyme) could be altered by changing the ionic strength or temperature of the solution or by protease treatment of the enzyme. A full under-

504

BETSY M. SUTHERLAND

standing of the groups involved in binding of the enzyme-substrate complex, of possible conformational changes upon enzyme-substrate complex formation “in preparation for” photolysis, and of groups involved in the photolytic event, await further characterization of these enzymes. E.

MECHANISM

Pyrimidine dimers can be monomerized by reactions other than enzymatic photoreactivation. In fact, one of the problems in understanding the action of PREs has been to decide which-if any-of the reaction paths suggested by the various systems is applicable to the photolyase. Three model systems for dimer breakage will be discussed first, followed by a review of the current evidence on the mode of action of the PRE. 1. Models For Pyrimidine Dimer Breakage

In addition to dimer monomerization by direct photon absorption, the photoreversal can proceed in a photosensitized reaction either using small molecules (91- 9 3 , tryptophan-containing peptides, or proteins (94). Each process has its own similarity to (and some have differences from) the enzymatic process. Each of the model systems is reviewed with special attention to its relevance to the enzymatic process. a. Direct Photon Absorprion. As discussed in Section I, the early studies of pyrimidine dimer formation were accompanied by the observation of their reversibility to pyrimidine monomers upon further irradiation in solution. The early suggestion that the reversal might be the mechanism of biological photoreactivation was disproved by the demonstration of Wang (12) that 366 nm radiation (which was quite efficient in producing biological photoreactivation) was ineffective in dimer photodestruction. Setlow (13) showed that the action spectrum for dimer monomerization approximated the absorption spectrum of the dimer (high at shorter wavelengths, e.g., 230 nm, and decreasing at longer wavelengths, eg., 280 nm). Wavelengths longer than 300 nm were ineffective in monomerizing the dimers; since action spectra for photoreactivation had short wavelength extents of about 300 nm, and maxima at even longer wavelengths, Setlow concluded that direct photon absorption could not be the mechanism of action of biological photoreactivation. This method of analysis points out one of the great strengths (and a 91. 92. 93. 94.

A. A. Lamola, JACS 88, 813 (1966). I. Rosenthal and D. Elad, BBRC 32, 599 (1968). C. Helene and M. Charlier, BBRC 43, 252 (1971). C. Helene and M. Charlier, Phofochern. Phorobiol. 25, 429 (1977).

24. PHOTOREACTIVATING ENZYMES

505

weakness) of using action spectra to identify photon absorbers in biological processes: If two action spectra are quite different in wavelength maximum andor range, one can rather safely conclude that the two processes are distinct. However, if they are similar, one has only circumstantial evidence that they correspond to the same process, and if at all possible additional lines of evidence should be sought. Although direct photon absorption can be excluded as a mechanism of enzymatic photoreaction, it shares one hallmark with the enzymatic process: The product of reversal by direct photon absorption and enzymatic action is the pyrimidine monomer, with no side reactions. The identity of substrates and products in the two reactions has allowed the use of direct photoreversal as one tool in evaluating the role of pyrimidine dimers in producing biological damage. Two critical factors must be considered in using the direct photoreversal: First, the reaction involves a photochemical equilibrium, as hv

PYPY

'hv PY[ IPY

(2)

with the wavelength of the incident light and the relative absorptions of monomer and dimer at that wavelength determining the proportions of the two species (see Ref. 79). At longer wavelengths (e.g., 289 nm), since the monomer absorbs reasonably well and the dimer poorly, the equilibrium can be driven to favor the dimer. At shorter wavelengths (e.g., 239 nm), the dimer absorbs more strongly (although of course the monomer also absorbs at this wavelength), and the reaction can be driven to the left. In DNA the reaction does not proceed to zero dimer content, however, as the 239 nm light also produces dimers. This equilibrium dictates that biological systems used for evaluation by the 239 nm-reversal must be able to show biological activity even with the residual (post 239 nm-irradiation) level of dimers (95). Second, the property of the photochemical equilibrium, in which the short wavelength irradiation reduces but does not eliminate dimers in the DNA, is markedly different from enzymatic photoreactivation, in which virtually all the dimers in the DNA can be reversed. This property, along with the differences in the action spectra, allows clear differentiation between reversal by direct photon absorption by the dimer and by PRE action. 6 . Sensitized Dimer Reversal. A wide variety of small molecules are effective in sensitizing pyrimidine dimer monomerization (91, 92, 96,97). It 95. R. B. Setlow and J. K. Setlow, Photochem. Photobiol. 4, 939 (1965). 96. H. D. Roth and A. A. LarnoIa,JACS 94, 1013 (1972). 97. I. Rosenthal, M. M. Rao, and J. Salomon, EEA 378, 165 (1975).

506

BETSY M . SUTHERLAND

is difficult to extrapolate from the properties of a sensitizer in solution acting on isolated dimers to the properties of that sensitizer within a protein matrix acting on dimers in DNA. However, in some cases sufficient photophysics or photochemistry of the system is known to allow at least reasonable guesses on their relevance to the mechanism of action of the enzyme. Wackerer al. (98) first showed that illumination of thymine dimers in the presence of uranyl acetate led to the production of monomer thymine plus other photolysis products. Lamola ( 9 / )found that 2-triphenylenesulfonate was able to photosensitize dimer breakage by 313 nm light, but with a quantum yield of 0.001 (compared to 1.0 for direct photolysis.) At that time, energy transfer from sensitizer to dimer was considered a possible mechanism for sensitized monomerization (and for enzyme action), but further consideration of the relative energy levels of donor and acceptor, as well as the low quantum yields, make this mechanism unlikely. Helene and his co-workers (93, 94) have extensively studied the photosensitized splitting of dimers by indoles, indole derivatives, and tyrosine. These moieties donate an electron to the dimer during the monomerization process. They showed that excitation of tryptophan or S-hydroxytroptophan (at wavelengths absorbed by these moieties) plus dimers in fluid or frozen aqueous solution led to monomerization of dimers. They also found that tryptophan in oligopeptides (e.g., Lys-Trp-Lys) or in DNA-binding proteins (99) was effective in splitting dimers in DNA. Upon mixing tryptophan- or tyrosine-containing peptides with nucleic acids, they observed a new absorption that extends to wavelengths greater than 300 nm (100). The great power of the Trp system is the ease with which dimer photolysis can be examined by the isolated tryptophan in solution or by tryptophan in a synthetic oligopeptide or naturally occurring protein. This great flexibility allows the best evaluation of the role of tryptophanmediated photolysis in PRE action. The appearance of new absorption bands upon complex formation between UV-irradiated polynucleotides and Tyr- or Trp-containing peptides would seem, at first glance, to be a likely origin of the new absorption band that appears upon UV irradiation of the E. coli enzyme-DNA complex. However, for the Tyr- and Trpcontaining peptides, the new absorbance is greater at the shorter wave98. A. Wacker, H. Dellweg, L. Trager, A. Kornhauser, E. Lodemann, G . Turck, R. Selzer, P. Chandra, and M. Ishimoto, Phorochem. Phorobiol. 3, 369 (1964). 99. C. Helene, F. Toulme, M. Charlier, and M. Yaniv, BBRC 71, 91 (1976). 100. F. Toulme and C. Helene,J. Chim. Phys., in press (1980).

24.

PHOTOREACTIVATING ENZYMES

507

*

lengths (e.g., 313 334 nm) and undetectable above 350 nm (where most PREs show maximum efficiency). Further, Sutherland and Griffin (/O/) showed that the tripeptide Lys-Trp-Lys could mediate destruction of dim e n in DNA by 313 nm light, but not by 334,365, or 405 nm, wavelengths effective in driving the photoenzymatic reaction. The absence of Trp in theE. coli enzyme (.I/), of course, also precludes tryptophan photolysis as its mechanism of action. In addition to its characteristic activity in breaking dimers, a hallmark of repair photoreactivating enzymes is the lack of side reactions, thus mediating biological repair. Chen er al. (102) tested the ability of indole derivatives to monomerize dimers in RNA and thus mediate biological photoreactivation of tobacco mosaic virus RNA. Although these compounds were able to monomerize dimers, they not only failed to mediate biological photoreactivation, but inactivated the RNA. The sum of the evidence-the strikingly different absorption spectra and wavelength dependence of action, the low efficiency, and the absence of Trp in at least one photoreactivating enzyme-indicate that it is unlikely that tryptophan-mediated dimer photolysis is a general mechanism for the enzyme action. Moreover, the propensity of these agents to photoinactivate nucleic acids renders quite unlikely that they would be effective in biological repair. Another general class of compounds effective in dimer photolysis is the good electron acceptors (for example, 2-anthraquinone sulfonate (96), K,Fe(CN),, and UO,SO,) (97). From CIDNIP (chemically induced dynamic nuclear polarization) studies of the cis-syn dimer of 1,3dimethylthymine irradiated at wavelengths greater than 340 nm in the presence of 2-anthroquinone sulfonate, Roth and Lamola (96) obtained evidence that transfer of the electron actually occurs during the photomonomerization process. The high quantum yields for dimer spIitting catalyzed by these electron acceptors (e.g., Q, = 1 for 2-anthraquinone sulfonate) makes it tempting to propose such a mechanism for action of the photoreactivating enzyme. Their efficiency in the wavelength region effective in photoreactivation makes this hypothesis even more attractive. No studies using electron acceptors in a protein environment (in analogy with the Lys-Trp-Lys, Gene 32 protein studies for electron donors) have been carried out, so the ability of such compounds to mediate biological photoreactivation remains unknown. 101. J. C. Sutherland and K. P. Griffin, Radirrr. Rcs. 83, 529 (1980). 102. I. Chen, C. W. Huang, L. Hinman, M. P. Gordon, and D. A. Derankau, J . Theor. B i d . 62, 53 (1976).

508

BETSY M. SUTHERLAND

F. CONTROL OF ENZYMESYNTHESIS AND FUNCTION Photoreactivating enzyme levels are regulated in prokaryotes and eukaryotes, and depend on a variety of factors (see Ref. 5 7 ) . In addition, cells of prokaryotes or eukaryotes in culture regulate their PRE levels in response to the medium (29, 58, 59). In one case, the enzyme seems to be induced by illumination (103), and evidence has been presented for activation by light of the enzyme in vitro (104). 1. Regulation of Enzyme Levels in Cells and Organisms

Cook and McGrath (57) showed that the activity of photoreactivating enzyme depended on the organ of origin of tissues (for example, highest in the leukocytes and absent in the erythrocytes of the toad Bufo marinus) and the stage of development of the organism (whole chick embryos showed activity while tissues from the adult chicken were deficient in the activity). Among prokaryotic and eukaryotic cells in culture, PRE activity levels vary with the growth phase of the culture. Boling and Setlow (105) showed that stationary-phase yeast cells contained between 4 and 35 times more PRE activity than logarithmically growing cells. Tyrrell et al. (106) also found higher photoreactivating enzyme levels in stationary phase than in log phase E. coli, and Sutherland el al. obtained similar data for cultured murine cells (55). In addition, both prokaryotes and eukaryotes regulate their PRE levels in response to the culture medium; E. Cali adenine auxotrophs starved for adenine showed striking increases in PRE levels (29). Cultured human cells grown in a rich Dulbecco’s modified Eagle’s medium were shown to contain measurable photoreactivating enzyme and to undergo cellular dimer photoreactivation, but cells grown in minimal essential medium contained no detectable levels of enzyme and were not capable of dimer photoreactivation (58, 59). The enzyme also depends on the genetic composition of the cell. Resnick and Setlow found that the PRE level was directly proportional to the ploidy (i.e. the number of copies of the gene) of the cells (107); Harm was able to obtain both PRE(-) E. coli (lacking photoreactivating enzyme) and PRE (+ + ) cells that produced about five times the normal enzyme level (108). Sutherland et al. (109) and Sutherland and Oliver (110) found that some primary human cell strains contained lowered PRE levels, and that 103. J. Diamond, J. A. Schiff, and A. Kelner, ABB 167, 603 (1975). 104. H. Harm and C. S. Rupert, Murar. Res. 34, 75 (1976). 105. M. E. Boling and J. K. Setlow, BBA 145, 502 (1967). 106. R. M. Tyrrell, S. H. Moss, and D. J. G. Davies, Murut. Res. 16, 345 (1972).

24.

PHOTOREACTIVATING ENZYMES

509

the lowered level showed simple Mendelian inheritance patterns (I I I). Although it would seem that the enzyme should be inducible by light, extensive evidence for such induction has been presented only in the case of Euglena (103).Diamond el a f . showed that the increase of PRE specific activity in cells exposed to light was prevented by the addition of cyclohexamide. They found that the wavelengths effective in photoreactivation were active in triggering induction of the enzyme. 2 . Effects of Light on Enzyme Activity

Harm and Rupert reported that illumination of yeast enzyme preparations, either as an ammonium sulfate-precipitated fraction (approximately fivefold purified relative to the cell extract) or as a 650-fold purified fraction (compare with the 70,000-fold purification cited by Minato and Werbin for purification to a major protein band on polyacrylamide gels), led to higher enzyme activity with UV-irradiated DNA in the presence of photoreactivating light (104).The preillumination effect did not obviate the requirement for photoreactivating light. The preillumination-induced increase in activity decayed slowly in the dark, but could be repeatedly regained by reillumination of the enzyme. The action spectrum for preillumination enhancement had two peaks, one at about 366 nm, and the other near 580 nm (87). Although the shorter wavelength peak coincided with the action spectrum for photoreactivation, the longer wavelength peak did not seem to correspond to major absorption bands for purified yeast I or yeast I1 photolyase. Thus, while the preillumination effect is clearly of practical importance to those working with the enzyme, whether it represents an effect on the binding, photolytic properties of the enzyme, or some indirect effect can not be clearly resolved. Tyrrell et a f . have reported that very large doses of 365 nm radiation (much greater than those used by Harm and Rupert) destroy the E. cofi enzyme in vivo and the yeast enzyme in vitro (112). Although the yeast I photolyase contains a cofactor that absorbs 365-nm light, neither yeast I1 nor E. cofi enzymes possess intrinsic absorption at this wavelength. Of course, in E. cofi cells a variety of other components might absorb this radiation leading indirectly to inactivation of the enzyme. 107. 108. 109. 110. 11 1. 112.

M. A. Resnick and J. K. Setlow, J . Bacteriol. 109, 1307 (1972). W. Harm,Mutat. Res. 8, 41 1 (1969). B. M. Sutherland, M. Rice, and E. K. Wagner, PNAS 72, 103 (1975). B. M. Sutherland and R. 0.Oliver, Nature (London)257, 132 (1975). B. M. Sutherland and R. Oliver, Phorochem. Photobiol. 24, 449 (1976). R. M. Tyrrell, R. B. Webb, and M. S . Brown, Photochem. Photobid. 18,249 (1973).

5 10 IV.

BETSY M. SUTHERLAND

Biologicol Role

Photoreactivation was f i s t known as a biological repair phenomenon: UV-irradiated cells exposed to longer wavelength light ufrer the UV had higher survival rates than those exposed to UV and kept in the dark. Not until about ten years after the discovery of biological photoreactivation were the foundations of its enzymology established.

A. MOLECULAR A N D CELLULAR PHOTOREACTION After the characterization of the enzyme, the same basic properties of photoreactivation by the enzyme were shown to operate in cellular photoreactivation: In siru disappearance of dimers; the lack of side reactions (leading to error-free repair); the use of light rather than cellular metabolic energy; and the absence of the incision, excision, and de noto synthesis characteristic of other repair processes. In addition, several properties of the photoreactivation process in vivo and in vitro provided powerful tools for testing biological repair phenomena to ascertain whether they were mediated by a photoreactivating enzyme [for a discussion, see Ref. (113)l. First, the reaction must require light, and light must be effective only when administered after the original UV insult. Second, the wavelengthdependence of the light requirement (action spectrum for cellular photoreactivation) must be similar to the action spectrum for the isolated enzyme. Third, dimers must disappear, not only from the high-molecularweight DNA, (as would also occur in excision repair) but from the total system (DNA, cell plus medium). Fourth, it is frequently possible to correlate PRE levels measured in extracts with cellular rates of photoreactivation (but note possible complications from competition of other proteins, such as repair enzymes, with the photolyase for the dimer sites).

B. MUTANTS IN PHOTOREACTIVATION 1. Escherichia coli

Walter Harm reported the first photoreactivation-deficient E. coli cells (phr-) along with several strains with higher photoreactivation capacities (phr++)(108).The latter strains proved difficult to work with but showed fascinating properties. They contained much as six times more photoreactivating enzyme than the parental strains, they were all auxotrophs, and 113.

B. M. Sutherland, Inr. Rev. Cyrol., Suppl.

8, 301 (1978).

24.

PHOTOREACTIVATING ENZYMES

511

from them prototrophic “revertants” arose at the level of 10-5 to The revertants to prototrophy were also revertants for PRE production, stationary phase titer, and UV sensitivity. Later work by Nishioka and Harm indicated that the auxotrophic requirement was due to a defect in adenine synthesis (29). Since adenylosuccinate supported growth of the phr++ cells but inosine monophosphate did not, they concluded that the cells were deficient in adenylosuccinate synthetase, for which the structural gene is purA. Moreover, they found that excess PRE production occurred only during adenine starvation, suggesting the regulation of PRE levels in E. coli by adenine or an adenine derivative. Thephr- mutation of Harm was mapped by Van der Putte to be neargal (114). Using deletion mutants, Sutherland et al. mapped a p h r gene between gal and A att, and found that upon induction of a A lysogen (containing the gal-A atr interval) levels of photoreactivating enzyme could be increased several thousandfold (50). Sancar and Rupert (1 15) and Bonura and Smith (1 16) mapped aphr gene about 1 minute away on theE. coli map from the other phr gene. Sutherland and Hausrath have suggested that these be termedphrA (at 17 minutes) andphrB (at 15.9 minutes), respectively (117). Although the presence of RNA and protein in theE. coli PRE makes it tempting to speculate that one of these genes might be the structural gene for the apoprotein and the other code for the RNA, no current evidence excludes possible control or processing functions of these loci.

2.

Yeast Resnick reported the production of a phr mutant of the yeast Saccharomyces cerevisiae, isolated after UV irradiation to about 20% survival (118). One phr mutant was obtained for about 20,000 colonies tested. Resnick treated the resulting mutant as resulting from a single gene change. The frequency of appearance of the mutant and the high survival rate would tend to argue against the presence of two separatephr genes in yeast, both of which would have to be mutated to produce the PRphenotype). It might be possible that the laboratory strain of S. cerevisicle was mutant for one of the photolyases, and thephr mutation of Resnick inactivated the second gene. It is also possible that bakers’ yeast contains more than one strain, each of which has its own enzyme. 114. (1965). 115. 116. 117. 118.

P. Van de Putte, C. S. van Sluis, J. Van Dillewijn, and A. Rorsch, Murar. Res. 2,97 A. Sancar and C. S . Rupert, Mutat. Res. 51, 139-143, (1978). D. Youngs and K . C. Smith, Mutat. Res. 51, 133-137, (1978). B. M. Sutherland and S. G. Hausrath, J . Bacreriol. 138, 333 (1979). M . A. Resnick, Phorochem. Photobiol. 9, 30 (1969).

512

BETSY M. SUTHERLAND

3. Man Cells from patients with xeroderma pigmentosum (xp), the inherited propensity to sunlight-induced skin cancer, have lowered levels of photoreactivating enzymes (109). The low enzyme levels are not the result of low growth rates, cell culture conditions, assay conditions, age of the cell donor at biopsy, the presence of inhibitors, or mycoplasma contamination. Sutherland and Oliver (110) found that the low level of PRE in xp cells followed a pattern of simple Mendelian inheritance in which pp represented individuals with very low PRE content (<20% of normal), Pp those with intermediate levels, and PP those with normal PRE levels. It is not known if the defect in PRE activity represents an alteration in the primary structure of the enzyme, or in a regulatory or processing element.

V.

Research Applications

The ability of PRE to reverse specifically in ,a light-dependent reaction, cis-syn-cyclobutyl pyrimidine dimers in DNA allows its use as a powerful analytical tool. By treating DNA in vitro, allowing cellular photoreactivation (in photoreactivation-competent cells) or inserting exogenous PRE (into photoreactivation-defective cells) and then allowing photoreactivation, one can specifically remove pyrimidine dimers from the DNA and evaluate their biophysical, biochemical, and biological consequences [See Ref, (113) for a discussion]. A.

in Vitro PHOTOREACTIVATION

DNA that contains pyrimidine dimers, produced either by direct irradiation or by sensitization (84), can be treated with photoreactivating enzyme and photoreactivating light, and its dimer-content evaluated in a number of ways. For example, Eker used the ability of DNA to serve as a template for RNA polymerase to evaluate its dimer content (30). Ahmed and Setlow used the specific cleavage of DNA at pyrimidine dimers by the M. luteus endonuclease to evaluate dimer content before and after photoreactivation (70). And of course radioactive DNA may simply be digested with nucleases or hydrolyzed with formic acid and the dimer content evaluated directly (65). These assays are of particular use in determining whether a given treatment (for example, a new light source) produces pyrimidine dimers. If a fraction of the lesions is photoreactivable, one has reasonable evidence that they were pyrimidine dimers. These methods have been used

24.

PHOTOREACTIVATING ENZYMES

5 13

to show that FS-40 and FS-20 sunlamps produce dimers in DNA, in cells, and in skin. It is also possible by photoreactivation to evaluate the fraction of biologically important damage that is produced by pyrimidine dimers: First photoreactivate all the dimers in the DNA (and ascertain by one of the above methods that all measurable dimers have been photoreactivated), and then evaluate the proportion of biological damage remaining. The H. influenzae-transforming DNA system has been of cardinal importance in such evaluation of the biological activity of DNA treated in vivo. Presumably other systems, such as DNA transfection, could be used successfully as well. A third important analytical use of PREs is the determination of the substrate range of other repair enzymes. Ahmed and Setlow (70) photoreactivated UV-irradiated DNA with the E. coli photoreactivating enzyme, and found that samples of DNA plus PRE exposed to light no longer contained substrate for the M. luteus UV-endonuclease, whereas equivalent samples kept in the dark contained substrate for this enzyme. Seeberg (personal communication) attempted a similar experiment and obtained the puzzling result that the treatment of UV-irradiated DNA with E. coli PRE, whether in the light or in the dark, resulted in a loss of sites for the uvrABC endonuclease of E. coli. The puzzle was solved when Seeberg treated the DNA with PRE plus light (and PRE in the absence of light), then extracted the DNA. The light-treated sample had lost endonuclease sites, whereas the sample kept in the dark had not. Evidently the uvrABC endonuclease could not displace PRE bound to dimer sites (but note that the M. luteus endonuclease is able to do so, Oliveria and Sutherland, unpublished data). This probably results from the relative size and/or binding constant of the various enzymes for dimer sites on the DNA.

B. PHOTOREACTIVATION in Vivo The cellular role of pyrimidine dimers can .be evaluated in a similar fashion to that discussed above for transforming DNA. In the case of photoreactivation-competent cells, one can simply apply photoreactivating light after the initial damage and measure the change in biological activity after photoreactivation. Of course it is essential to show that the light-mediated recovery results from true enzymatic photoreactivation by the criteria discussed above. The above technique has been limited to cells with reasonable cellular levels of photoreactivating enzymes. Several cell types, both eukaryotic (e.g., rodent cells and xeroderma pigmentosum cells) and prokaryotic

5 14

BETSY M. SUTHERLAND

(e.g., the naturally transformable bacteria, including Bacillus subrilis) contain little or no endogenous PRE. Permeabilization methods have been developed to permit insertion of photoreactivating enzyme into mammalian (119) and bacterial (Yasbin, unpublished data) cells. Both studies employed partially purified E. coli photoreactivating enzyme, and showed that the enzyme could monomerize dimers in DNA inside the recipient cell.

C. PRACTICAL CONSIDERATIONS Although photoreactivation is one of the easiest repair pathways to control, the reaction is sufficientlyunusual to warrant an outline of several important points. First, the enzymes are generally fragile and are inactivated by freezing and thawing or by storage in ice or in a refrigerator. They can be successfully stored at - 20" by the addition of sufficient glycerol (-67 g to 100 ml of solution) to prevent freezing. The relative concentrations of substrates and enzyme must be chosen carefully to fit the experiment. To measure enzyme concentrations, use saturating concentrations of dimers in DNA and saturating light intensity, but limiting enzyme concentrations. It should be stressed that the initial rate of photoreactivation, but not the final extent of photoreactivation, is a function of enzyme concentration. This is especially important in cellular experiments. For measurement of an action spectrum, which is derived from the probability that a photon of a given wavelength will be absorbed by the enzyme-substrate complex, use enzyme-substrate complexes in excess and limiting light fluxes [see Ref. (80) for a discussion]. After selection of substrate and enzyme concentrations, it is essential that the solution conditions be optimal for the enzyme. The various enzymes vary in ionic strength optima, but most prefer pHs of about 7.2 and show no requirement for added metal. Both the yeast and E. coli activities are inhibited by heavy metal ions and stabilized by the presence of sulfhydryl reagents. After formation of the enzyme-substrate complex, (which, at usual enzyme assay conditions requires minutes, not milliseconds), the second substrate, light, must be administered. This time sequence of complex formation before photoreactivating light is critical, as the enzyme can not store energy for later use in the photolytic process. The light must be of the correct wavelength range to mediate photolysis. The reader is referred to the reports of action spectra of the individual enzymes for selection of 119.

B. M. Sutherland and S . G . Hausrath, Nafure (London) 286, 510 (1980).

24. PHOTOREACTIVATING ENZYMES

5 15

the best spectral range. It should not be assumed that lamps adequate for one enzyme (e.g., 365 nm black lights for the E. coli enzyme) will drive the photolytic action of another enzyme. In addition, one must select a safe light for forming and handling enzyme-substrate complex when photolysis is not desired. Again the action spectrum of the enzyme determines the wavelengths ineffective in producing photolysis. For many PREs, yellow “bug lamps” obtainable in supermarkets are adequate safe lamps as they contain no light below 500 nm (60). For the marsupial and human enzymes, a red lamp emitting no wavelengths shorter than 600 nm can be used as safe lamps (60). For enzymes showing a preillumination effect (87, 104), only brief exposures of the enzyme (even before substrate addition) to a dim red lamp should be permitted. In the case of enzymes of unknown action spectrum, a dim red lamp will permit experimental operations without producing significant photoreactivation. For cellular experiments, additional factors must also be considered (see Ref. 113). First, is the enzyme present? Since biochemical measurements of PRE activity are much faster than biological determinations, it is advisable to determine if the enzyme is present before looking for its cellular action. Second, what are the other repair capacities of the cell? As discussed above, excision-repair enzymes compete in vitro with PREs for dimer lesions in DNA. That they may also do so in vivo is reflected by the much larger effect of photoreactivation in excision-deficient than in excision-proficient cells (120, IZI). The reader is referred to the elegant studies of Harm in studyingE. coli PRE action inE. coli cells. His methods permit almost as much flexibility as in v i m assays, while retaining the natural milieu of the enzyme and its substrate. ACKNOWLEDGMENTS I thank the National Cancer Institute of the National Institutes of Health (CA26492 and CA23096). and the Department of Energy for support.

120. W. Harm, C. S. Rupert, and H. Harm, in “Molecular and Cellular Repair Processes” (R.F. Beers, R. M. Herriott, and R. C. Tilghman, eds.), p. 53. Johns Hopkins Univ. Press, Baltimore, 1971. 121. E. E . Henderson, Cancer Res. 38, 3256 (1978).

This Page Intentionally Left Blank

DNA Methyla tion STANLEY HATTMAN

I . Introduction . . . . . . . . . . . . . . . . . . . . I1. Nature of Methylated Bases and Methods of Analysis A . Quantitation and Identification . . . . . . . . . B . Sequence Specificity . . . . . . . . . . . . . . 111. DNA Methylases . . . . . . . . . . . . . . . . . A.Assay . . . . . . . . . . . . . . . . . . . . . B . Prokaryote Enzymes . . . . . . . . . . . . . . C. Eukaryote Enzymes . . . . . . . . . . . . . . IV. Distribution of Methylated Bases . . . . . . . . . A . Survey Among Prokaryotes . . . . . . . . . . B . Survey Among Eukaryotes . . . . . . . . . . V . Other DNA Modifications . . . . . . . . . . . . . . VI . Biological Roles . . . . . . . . . . . . . . . . . . A . Prokaryotes . . . . . . . . . . . . . . . . . . B . Eukaryotes . . . . . . . . . . . . . . . . . . VII . Concluding Remarks . . . . . . . . . . . . . . . .

.

1

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

. . . . . . .

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

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

517 518 518 520 521 521 523 527 531 531 532 536 537 537 543 547

Introduction

Almost all biological macromolecules undergo some form of processing event subsequent to their biosynthesis . Polypeptides are specifically folded. cleaved. phosphorylated. acetylated. methylated. or covalently bound to other polypeptides . RNA molecules are specifically cleaved. lengthened. spliced. methylated. thiolated. or otherwise base.modified . These processes are obligatory events in the establishment of functional expression of these molecules . It has been known for almost three de517

.

THE ENZYMES Vol . XIV Copyright 0 1981 by Academic Press. Inc . AU rights of reproduction in any form reserved

ISBN 0-12-122714-6

5 18

STANLEY HATTMAN

cades that DNA is also subject to post-replication modification. The main topic of this chapter is to consider the phenomenon of DNA methylation, its nature, distribution, analysis, specificity, and biological function. In addition, I briefly review other DNA modifications known to occur among prokaryotes and eukaryotes. II. Nature of Methylated Bases and Methods of Analysis

In general, there are only two naturally occurring methylated bases found in DNA: N6-methyladenine (MeAde) and 5-methylcytosine (MeCyt). When present they usually comprise only a minor fraction of the total DNA bases, varying from as low as 0.17 mole % MeCyt (of the cytosines) in insects up to 8 mole % MeCyt in mammals; and in certain higher plants, the level may rise to as high as 50% (1-7).Although there are occasional reports that a given higher organism may lack DNA methylation, it appears that there are really no exceptions. AND IDENTIFICATION A. QUANTITATION

The nature and extent of DNA methylation has been measured by a variety of techniques. In the early studies relatively large amounts (mg) of DNA were acid-hydrolyzed to bases and then separated by paper or thinlayer chromatography in one or two dimensions; the concentration of each base could then be determined from spectrophotometric measurements. The identity of MeCyt or MeAde was established by spectral data and by cochromatography with authentic marker in several solvent systems. In later years, in vivo labeling with radioisotopic purine or pyrimidine precursors, Hg32P04,or [I4C-or 3H-methyl]methionine has been applied successfully. Methionine labeled in the methyl group is a precursor of S-adenosyl-L-methionine (AdoMet), which is the methyl donor in the enzymatic reaction (see Section I11 below). Following extraction, purified G . R. Wyatt, BJ 48, 584 (1951). D. B. Dunn and J. D. Smith, BJ 68, 627 (1958). J. DoskoEil and Z. Sormova, BBA 95, 513 (1965). H. S. Shapiro and E. Chargaff, BBA 39, 68 (1960). B. F. Vanyushin, A. N. Belozersky, N. A. Kokurina, and D. X. Kadirova, Nature (London) 218, 1066 (1%8). 6. B. F. Vanyushin, S. G . Tkacheva, and A. N. Belozersky, Nature (London)225, 948 1. 2. 3. 4. 5.

(1970). 7. R. L. P. Adams, E. L. McKay, L. M. Craig, and R. H. Burdon, BBA 563, 72 (1979).

25.

DNA METHYLATION

5 19

DNA can be enzymatically digested to oligonucleotides, mononucleotides or nucleosides; appropriate chemical degradation can produce free bases, free purines and apurinic acid, or free purines and pyrimidine tracts (Py, P,+l). Depending on the method of degradation any number of subsequent techniques may be applied. Bases, nucleosides or mononucleotides can be analyzed by thin-layer or paper chromatography, paper electrophoresis, ion-exchange chromatography, or high-performance liquid chromatography. Quantitative determination of the level of methylation depends on the ability to effect resolution of the normal and methylated forms (from each other and from other components), and to have sufficient radioactivity or mass (for spectrophotometric analysis). In general, the available methods have been adequate for studying the nature and level of MeCyt and MeAde in most organisms. A different problem arises when methyl-labeled methionine is used as a precursor. In addition to the need for removal of proteins (labeled in methionine) and RNA (labeled in MeCyt and MeAde), radioactivity may appear in the methyl group of thymine and purine rings (8, 9). Also, it is not straightforward to calculate the mole % of a methylated base, since the normal counterpart is not radioactive. The problem of spurious labeling has been circumvented in many cases by adding formate to the labeling medium to dilute the C-1 pool (8, 10). Nevertheless, it should not be taken for granted, as has been too frequently the case, that incorporation of methylmethionine label is specific for methylated bases; each system must be separately analyzed to determine how the radioactivity is distributed among the DNA bases. Depending on the organism, radioactive labeling may not be feasible, or sufficient quantities of DNA may not be readily obtainable to analyze very low levels of methylation (e.g., 0.01% of the normal base). Thus, to achieve greater sensitivity and resolution, more recent studies have employed high-performance liquid chromatography ( I 1-I6a 1, gas 8. E. Winocour, A. M. Kaye, and V. Stollar, Virology 27, 156 (1965). 9. A. Razin, J . Sedat, and R. L. Sinsheimer, JME 53, 251 (1970). 10. M. M. K. Nass,JMB 80, 155 (1973). 1 1 . P. H. Roy and H. 0. Smith, JMB 81, 445 (1973). 12. J. Singer, R. H. Stellwagen, J. Roberts-Ems, and A. D. Riggs, JEC 252,5509 (1977). 13. J. M. Pollock, Jr., M. Swihart, and J. H. Taylor, Nucleic Acids Res. 5, 4855 (1978). 14. J . Singer, J. Roberts-Ems, F. W. Luthardt, and A. D. Riggs, Nucleic Acids Res. 7, 2369 (1979). 15. J. N . Lapeyre and F. F. Becker, BERC 87, 698 (1979). 16. W. G . Burton, C. T. Grabowy, and R. Sager, PNAS 76, 1390 (1979). 16a. R. Baur, H. Wohlert, and H. Kroger, Hoppe-Seyler’s Z . Physiol. Chem. 360, 1263 (1979).

520

STANLEY HATTMAN

chromatography (17), or mass spectrometry (18-20). The latter method promises to be an extremely useful tool for both quantitative and qualitative analyses of all DNA modification. Finally, an interesting and innovative approach utilizing antibodies to MeCyt has been developed (21-25).

B. SEQUENCE SPECIFICITY The next higher level of methylation analysis is determination of the primary base-sequence in which it occurs. Among the strategies adopted to achieve this are: identification of 5' and 3' nearest neighbors ( I / , 26-30), analysis of pyrimidine (4, 31 -38) and purine Q9) tracts, direct sequencing of oligonucleotides or restriction nuclease fragments that contain methylated bases (26, 27, 29, 40-48), and protecting DNA against cleavage by 17. A. Razin and J. Sedat, Anal. Biochem. 77, 370 (1977). 18. J. Deutsch, A. Razin, and J. Sedat, Anal. Biochem. 72, 586 (1976). 19. F. Gautier, H. Biinemann, and L. Grotjahn, EJB 80, 175 (1977). 20. J. Singer, W. C. Schnute, Jr., J. E. Shively, C. W. Todd, and A. D. Riggs, Anal. Jiochem. 94, 297 (1979). 21. 0. J. Miller, W. Schnedl, J. Allen, and B. F. Erlanger, Nature (London) 251, 636 (1974). 22. W. Schnedl, V. G. Bev, R. Tantravahi, D. A. Miller, B. F. Erlanger, and 0. J. Miller, Chromosoma (Berlin) 52, 59 (1975). 23. B. W. Lubit, T. D. Pham, 0. J. Miller, and B. F. Erlanger, Cell 9, 503 (1976). 24. R. R. Schreck, V. G. Bev, B. F. Erlanger, and 0. J. Miller, Chromosoma (Berlin) 62, 337 (1977). 25. E. M. Eastman, R. M. Goodman, B. F. Erlanger, and 0. J. Miller, Chromosoma (Berlin) 79, 225 (1980). 26. M. B. Mann and H. 0. Smith, Nucleic Acids Res. 4, 4211 (1977). 27. S . Hattman, H. van Ormondt, and A. de Waard, JMB 119, 361 (1978). 28. S . Hattman, T. Keister, and A. Gottehrer, f M B 124, 701 (1978). 29. S. Hattman, J. E. Brooks, and M. Masurekar, JMB 126, 387 (1978). 30. U. Giinthert, K. Storm, and R. Bald, EJB 90, 581 (1978). 31. J. DoskoEil and F. Sorm, BBA 55, 953 (1962). 32. R. Salomon, A. M. Kaye, and M. Herzberg, JMB 43, 581 (1%9). 33. R. Salomon and A. M. Kaye, BBA 204, 340 (1970). 34. T. W. Sneider, JBC 246, 4774 (1971). 35. T. W. Sneider, JBC 247, 2872 (1972). 36. M. S. May and S . Hattman, J . Bacteriol. 122, 129 (1975). 37. K. Harbers, B. Harbers, and J. H. Spencer, BBRC 66, 738 (1975). 38. B. F. Vaynushin and A. P. Dobritsa, BBA 407, 61 (1975). 39. B. F. Vanyushin, Ya. I. Buryanov, and A. N. Belozersky, Nature New Biol. 230,25 ( 1971). 40. H. van Ormondt, 3. A. Lautenberger, S. Linn, and A. de Waard, FEES (Fed. Eur. Biochem. Soc.) Lett. 33, 177 (1973). 41. H. W.Boyer, L. T. Chow, A. Dugaiczyk, J. Hedgpeth, and H. M. Goodman, Nature New Biol. 244, 40 (1973).

25.

DNA METHYLATION

52 1

site-specific nucleases of known specificity (49-51). The last approach is limited, however, by the possibility that the recognition site may have two adenines or two cytosines, and, consequently, the position of the methylated base cannot be inferred; moreover, if the base specificity of a methylase is not established, presence of both Ade and Cyt in the site presents the same problem. Determining the base-sequence of a region containing a methylated base will not establish a methylation sequence. Because the length of the site is not known it is necessary to characterize several such regions and determine the shortest nucleotide sequence common to all sites. In this regard, it is interesting to note that the Maxam-Gilbert sequencing technique can reveal the position of MeCyt residues (52). Finally, a recent development in strategy has been to employ computer technology. Following in vitro methylation of a DNA of known base sequence, fragments are produced by restriction nucleases; those containing methylation are analyzed for the presence of a common sequence unique to methylated fragments. This method has been used successfully in two instances (62, 85a). 111.

DNA Methylases

A. ASSAY The DNA methylases of prokaryotes and eukaryotes transfer methyl groups from donor S-adenosyl-L-methionine (AdoMet) to acceptor DNA (53, 54). Depending on the enzyme, both double-stranded and singlestranded DNAs may be suitable substrates (discussed further in Sections III,B and C), but native DNA is likely to be the in vivo acceptor. For routine assays we have found that commercial preparations of Micrococ42. A. Dugaiczyk, M. Kimball, S. Linn, and H. M. Goodman, BBRC 61, 1133 (1974). 43. A. Dugaiczyk, J. Hedgpeth, H. W. Boyer, and H. M. Goodman, Biochemistry 13,503 ( 1974). 44. J . P. Brockes, P. R. Brown, and K. Murray, J M B 88, 437 (1974). 45. M. May and S. Hattman, J . Bncreriol. 123, 768 (1975). 46. B. Bachi, J. Reiser, and V. Pirrotta, JMB 128, 143 (1979). 47. G. E. Geier and P.Modrich, JBC 254, 1408 (1979). 48. J. R. Miller, E. M. Cartwright, G. G. Brownlee, N. V. Fedoroff, and D. D. Brown, Cell 13, 717 (1978). 49. S. Lacks and B. Greenberg, JMB 114, 153 (1977). 50. S.Schlagman, S. Hattman, M. S. May, and L. Berger,J. Bacferiol. 126,990 (1976). 51. M. C. Gomez-Eichelmann, J . Bncteriol. 140, 574 (1979). 52. H. Ohmori, J. Tomizawa, and A. H. Maxam, Nucleic Acids Res. 5, 1479 (1978). 53. M. Gold and J. Hurwitz, CSHSQB 28, 149 (1964). 54. B. Sheid, P.R. Srinivasan, and E. Borek, Biochemistry 7, 280 (1%8).

522

STANLEY HATTMAN

cus luteus DNA (formerly M. lysodeikticus) is an excellent substrate for both prokaryote and eukaryote DNA methylases. The basic strategy of the assay is to measure the transfer of radioisotopically labeled methyl groups from [ ' T I - or 13H]AdoMet to an acid-insoluble product. This is accomplished most easily by acidifying the reaction mixture and collecting the acid-precipitable fraction by filtration or centrifugation. The reaction can be summarized as follows: enzyme

Acceptor DNA

+ AdoMet (?Hor ''C-methyl-labeled) --+ [3H-methyIl- or [Y-rnethyl1DNA

+ S-adenosylhomocysteine

The methylases studied to date have not shown any cofactor requirements, although some may be stabilized by sulfhydryl compounds. Absence of a requirement for divalent cations is an advantage in that the reaction can be carried out in the presence of EDTA; thus, in enzyme fractions that contain DNase activity, it is frequently possible to assay for methylase activity and not be concerned with DNA degradation. The requirement for AdoMet cannot be met by other methyl donors, and product S-adenosylhomocysteine is a potent competitive inhibitor of the reaction (53-55). Some DNA methylases are stimulated by the addition of ATP and Mg2+,but these are unusual cases (56). A number of suggestions are in order concerning the assay procedure. There is frequently a high background of radioactivity, even in the absence of added protein fraction; this is a particular problem for enzymes with few recognition sites per acceptor DNA when AdoMet of very high specific radioactivity has to be used. We found that including cold, carrier AdoMet during acidification significantly reduces the nonspecific label retained by glass fiber filters. Acidification in cold 5% TCA or PCA is effective in precipitating the DNA; following collection onto glass fiber disks, samples can be washed with 1% TCA (the rate of filtration does not appear to be important). The disks are dried in air or under infrared illumination, and the radioactivity determined in a liquid scintillation counter. Inclusion of agents that solubilize DNA and remove it from the filter significantly increase the counting efficiency. This may be an important factor to consider when using rHIAdoMet because increasing amounts of protein collected on a filter will eventually act as a quenching agent. Finally, considerable caution is advised when assaying for DNAmethylase activity in crude protein fractions; this is particularly relevant to eukaryote extracts (57) where there are significant protein and RNA 55. H. Sano and R. Sager, EJB 105, 471 (1980). 56. S. Linn, this volume, Chapter 8. 57. F. Kalousek and N . R. Morns, JBC 243, 2440 (1968).

25. DNA METHYLATION

523

methylase activities. Thus, with crude fractions, the assay described above is not specific for DNA methylation, and it is necessary to deproteinize and alkali-digest reaction mixtures prior to acidification (57). This is an obligatory step, notwithstanding the fact that exogenous DNA stimulates the reaction. B.

PROKARYOTE ENZYMES

DNA methylase activities have been partially purified from a wide variety of gram-negative and gram-positive bacteria. It has been clear for some time that these enzymes show species and strain specificity: i.e., although a given enzyme is incapable in vitvo of methylating (homologous) DNA from the same strain, it is capable of methylating heterologous DNAs (53, 58-60). These early findings are now clearly understood in terms of there being a wide variety of sequence-specific DNA methylases existing among bacteria. Table I contains a summary of enzymes that have been characterized as to their sequence recognition (see Refs. 61-78). 58. M. Gold, J. Hurwitz, and M. Anders, PNAS 50, 164 (1%3). 59. M. Gold and J. Hurwitz, JBC 239, 3866 (1964). 60. D. Fujimoto, P. R. Srinivasan, and E. Borek, Biochemistry 4, 2849 (1965). 61. P. J. Greene, M. S. Poonian, A. L. Nussbaum, L. Tobias, D. E. Garfin, H. W. Boyer, and H. M. Goodman, J M B 99, 237 (1975). 62. S. M. Hadi, B. Bachi, J. C. W. Shepherd, R. Yuan, K. Ineichen, and T. A. Bickle, J M B 134, 655 (1979). 63. J. A. Lautenberger, N. C. Kan, D. Lackey, S. Linn, M. H. Edgell, and C. A. Hutchison 111, PNAS 75, 2271 (1978). 64. J. V. Ravetch, K. Horiuchi, and N. D. Zinder, PNAS 75, 2266 (1978). 65. N. C. Kan, J. A. Lautenberger, M. H. Edgell, and C. A. Hutchison II1,JMB 130, 191 (1979). 66. S. Hattman, J. Virol. 34, 277 (1980). 67. Ya I. Buryanov, I. G. Bogdarina, and L. M. Vagabova, Dokl. Akad. Nauk S S S R 230, 431 (Engl. transl.) (1976). 68. V. F. Nesterenko, Ya I. Buryanov, and A. A. Baev, Dokl. Akad. Nauk S S S R 2.50, 72 (Engl. transl.) (1980). 69. I. I. Nikolskaya, N. G. Lopatina, N. V. Anikeicheva, and S. S . Debov, Niicleic Acids Res. 7, 517 (1979). 70. H. 0. Smith, Science 205, 455 (1979). 71. J . Yo0 and K. L. Agarwal, JBC 255, 6445 (1980). 72. S. Hattman, Virology 42, 359 (1970). 73. H. R. Revel and S. Hattman, Virology 45, 484 (1971). 74. R. Hehlmann and S. Hattman, J M B 67, 351 (1972). 75. J. E. Brooks and S. Hattman,JMB 126, 381 (1978). 76. H. 0. Smith and D. Nathans, JMB 81, 419 (1973). 77. M. Gold, R. Hausmann, U. Maitra, and J. Hurwitz, PNAS 52, 292 (1964). 78. K . Oda and J. Marmur, Biochemistry 5 , 761 (1966).

524

STANLEY HATTMAN

TABLE I

PROKARYOTE DNA-METHYLASES ~~

~~~

Organism" DNA-adenine methylases Bacillus brevis

~~~~~~~

Enzymeb designation

Diplococcus pneumoniae Escherichia coli (RI) E. coli B and K12 E. coli (Pl) E . coli (P15) E. coli B

M BbvII M . BbvIII M . DpnII M . EcoRI M . Eco dam M EcoPl M . EcoPl5 M EcoB

E. cob K12

M , EcoK

E. coli (T2) M ' T2 damh E . coli (T2) M . T2 d a m + Haemophilus injluenzae Rd M . Hind1 M . Hind11 M . Hind111 M . HindIV M . HinffII DNA-cytosine methylases tlaciilus amyloliquefaciens M . Bum I M Bum11 Bacillus brevis M . BbvI Bacillus subtilis M . BsuR Escherichia coli (RII) M EcoRII E. coli (K12) M . Eco dcmI E. coli C and MRE600 M Eco dcmII E. coli SK M . EcodcmIII M . Eco dcm IV Haemophilus aegyptius M HaeIII H . haemolyticus M . HhaI H. parainjluenzae M . HpaII M HpaII' Moraxella species M . MspI

~

~

Methylatedc*d*'" sequence

Ref.

(28) X-G-A*-T-(C/Y) X-A-A*-G-Y (28) G-A*-T-C (49) (42, 43, 6 1 ) G-A-A*-T-T-C (29, 47, 4 9 ) G-A*-T-C A-G-A"-C-C7 (29.46) C-A?-G-C-A -G (62) 5' .T-G-A*(8N)T-G-C-T.3' (63, 6 4 ) 3' .A-C:T(8N)A*-C-G-A.S' 5'.G-C-A*-C(GN)G-T-T.3' (65, 6 6 ) 3' .C-G-T-G(GN)C-A*-A.5' G-A*-Py (27) G-A*-T (27, 7 5 ) X-C-A*-C-Y (11) G-T-Py-Pu-A*-C (11) A*-A-G-C-T-T (11) X-G-A:-T-Y (I/) C-G-A.-A7-T (85a )

G-G-A-T-C*-C G-C*-(A/T)-G-C G-C*-(A/T)-G-C G-G-C*-C C-C*-(A/T)-G-G C-C*-(A/T)-G-G Pu-C*-C-G-G C*-C-(A/T)-G-G G-G-(A/T)-C-C* G-G-C*-C G-C*-G-C C-C*-G-G C-C*-G-G C*-C-G-G

(28) (28) (28, 38) (30) (35, 45, 50) (36, 41, 45) (67, 6 8 ) (69) (88)

(26) (70) (26) (71) (97)

" For space economy, all strain designations are not cited even though the presence or absence of a particular enzyme is strain-specific. (RI) and (RII) refer t o the drug-resistance transfer R-factors specifying hspI and hspII restrictiodmodification systems; (Pl) and (P15) refer to the phages; M T2 dam and M T2 damh refer to the wild-type and mutant forms of the phage T2 DNA methylase (72-75) produced following infection of E. coli ( 7 7 ) . The designations are based upon the system of nomenclature proposed by Smith and Nathans ( 7 6 ) .Original literature designations have not always adhered to this system; I have taken the liberty to arbitrarily make assignments in conformity with their proposal. +

25. DNA METHYLATION

525

Details about the purification procedures can be obtained from the original references; in almost all cases, it was of considerable advantage that the enzyme was retained by phosphocellulose. I n view of the large range of known site-specific endonuclease targets, it is likely that new methylation specificities will eventually be characterized (assuming that cells contain modification methylases corresponding to these nucleases). The DNA methylases are quite diverse with respect to size and complexity. M .&OK, -B, -P1 and -P15 are large multimeric proteins composed of several different subunits (79-85). Depending upon the purification procedure, methylase activity may be isolated alone or in combination with restriction nuclease activity. M * HpaII activity is unusual in that it has been observed as two distinct monomeric polypeptide species: M . HpaII (38,500) and M . HpaII’ (41,500) (71). Preliminary studies sug-

79. A. Haberman, J. Heywood, and M. Meselson, PNAS 69, 3138 (1972). 80. J. A. Lautenberger and S. Linn, JBC 247, 6176 (1972). 81. B. Eskin and S. Linn,JBC 247, 6183 (1972). 82. G. F. Vovis, K. Horiuchi, and N. D. Zinder, PNAS 71, 3810 (1974). 83. G. F. Vovis and N. D. Zinder, JMB 95, 557 (1975). 84. J. Reiser and R. Yuan,JBC 252, 451 (1977). 84a. R. Yuan, Annu. Rev. Biochem, in press (1981). 85. J. Burckhardt, J. Weisemann, and R. Yuan, JBC 256, 4024 (1981). 85a. A. Piekarowicz, T. A. Bickle, J. C. W. Shepherd, and K. Ineichen, JMB 146, 167 ( 1981).

The methylated sequences (written in the 5’ + 3’ direction) have been deduced from a variety of approaches including pyrimidine tract analysis, nearest-neighbor analysis and two-dimensional fingerprinting of in vivo or in virro labeled DNA. In some instances, knowledge of the restriction-nuclease cleavage site aided in deducing the sequence (following determination of the 5’ and 3’ nearest neighbors). Where only partial sequence information is known, the 5’ and 3’ positions are indicated by unknown bases, X and Y, Pu and F‘y denote purine and pyrimidine, respectively. N denotes any of the 4 standard bases. The site of A methylationb) for M . HinfIII and M . EcoP15 are not yet known. Due to the complex nature of the M EcoK and M EcoB sites, the sequences of both strands are illustrated; 6N and 8N denote 6 and 8 bases, respectively. The positions of A* in the M . EroK site were only postulated ( 6 5 ) ; indirect support for C-A*-C methylations has been obtained (66). Note that most sites show an axis of rotational symmetry; M . EcoP1, M . EcoP15, M HinflII and M . T2 dam are notable exceptions. M . EcoPl may methylate some A-G-A-C-T sites (29). Salmonella typhi and S . fyphimurium appear to methylate the same sequences as M . Eco dam and M . Eco dcm I (S/). The T2 darn’ site is not entirely clear at the present. Although direct sequencing studies follcwing long-term methylation suggested G-A*-Py, both in vivo and short term in vitro methylation indicated a relatively poor ability to methylate G-A-C (75).

526

STANLEY HATTMAN

gested that the two molecular forms are related. M * EcoRI is also a monomeric protein (39,000) (86). The smallest enzyme appears to be the T2 dam methylase, which has a molecular weight of only 14,400 (87). The ability to methylate double-stranded versus single-stranded DNA has not been analyzed in all cases. It is known that several of the Bacillus and Haemophilus enzymes (e.g., M * BsuR, M .HaeIII, M . HpaII) methylate denatured DNA (71, 75). The size of duplex DNA required for methylation does not seem to be critical because degradation by sonication or partial nucleolytic cleavage does not inhibit methylation; in some cases this actually stimulates methylation. It is also possible to methylate restriction nuclease fragments of DNA, or even synthetic deoxynucleotide oligomers. For example, M HpaII methylates a 9-base-pair synthetic duplex oligomer (26); it can also methylate the single-stranded chains, but at a much reduced rate. M * EcoRI is capable of methylating an 8base-pair synthetic oligonucleotide duplex (6 I ). It is interesting that only specific residues within a recognition site are methylated; i.e., the methylases are capable of distinguishing between adjacent adenine or cytosine residues and specifically methylating only one of them (see Table I). In this regard, the restriction nucleases also distinguish between adjacent A and A*, and C and C*. Although a restriction nuclease and its corresponding modification methylase may recognize the same base sequence, it is evident that they “read” the site differently. For example, when G-A-A-T-T-C is modified to I-A-A-T-T-C or GA-A-U-U-C, M .EcoRI methylase activity is lost, but the R .Eco RI restriction nuclease still cleaves the site (88). Conversely, G-A-A-T-Tglucosyl hydroxymethyl C does not block M * Eco RI methylatioqbut inhibits cleavage by R EcoRI (89, 90). The sequence specificity may in some cases be modified by altering reaction conditions, e.g., M . EcoRI will methylate N-A-A-T-T-N at appropriate pH and salt concentration (91). Likewise the corresponding restriction nuclease, albeit a separate protein species, can be “fooled” into cleaving unmethylated A-A-T-T sites (92). Mutation in the structural gene may cause an alteration in methylase specificity; although the precise change is not fully understood, it is known that the T2 dam + T2 darn mutation leads to a change in methylation ability (72-75). +

+

86. R. A. Rubin and P. Modrich, JBC 252, 7265 (1977). 87. M. Masurekar and S. Hattman, unpublished observation (1978). 88. P. Modrich and R. A. Rubin, JBC 252, 7273 (1977). 89. D. A. Kaplan and D. P. Nierlich, JBC 250, 2395 (1975). 90. K. L. Berkner and W. R. Folk, JEC 252, 3185 (1977). 91. K. L. Berkner and W. R. Folk,Nucleic Acids Res. 5, 435 (1978). 92. B. Polisky, P. Greene, D. E. Garfin, B. J. McCarthy, H. M. Goodman, and H. W. Boyer, PNAS 72, 3310 (1975).

25. DNA METHYLATION

527

Methylation sequences are either asymmetric or show twofold rotational symmetry (palindromes) (for an interesting discussion of recognition mechanisms possible for such sites, see Ref. 70). When a palindrome is the recognition site, it is evident that both complementary DNA strands can be methylated. Nevertheless, methylation of only one strand (hemimethylation) is sufficient to block cleavage by the corresponding restriction nuclease, but does not interfere with complete methylation by the modification enzyme. M . Eco B and M . Eco K have a strong preference for hemi-methylated sites compared to completely unmethylated sites (82, 85), and methylation of these sites is stimulated by the presence of ATP and Mg2+.M . EcoK activity is associated with a complex protein that can also restrict DNA (79) in a reaction that requires AdoMet, ATP, and Mg'+. Detailed studies on the mechanism of restriction-modification modes of the EcoK complex show that the protein is activated by AdoMet regardless of whether it will cleave or modify a DNA (85). The decision to cleave or methylate occurs in a subsequent step. In contrast, M . EcoRI methylates both hemi-methylated and unmethylated sites at similar rates (86). A kinetic analysis of the reaction showed that the enzyme functions as a monomer, and after addition of a single methyl group to an unmodified site detaches from the template (86). Thus, methylation of each strand within the (palindrome) site occurs independently and at similar rates. The ability to recognize hemi-methylated sequences is essential inasmuch as DNA replication generates such sites. Although the M .EcoB and M . EcoK methylation sequences are not palindromes, both strands are methylated (Table I). It is interesting that M . EcoP1, M . EcoP15 and M . HinfIII sites are asymmetric and only 1 strand is methylated. One would expect DNA replication to generate unmodified sites that are targets for the restriction enzyme. Nevertheless, the cell DNA is not cleaved in vivo; either the modification enzyme methylates the DNA rapidly enough, or partially modified heteroduplex DNA is a poor substrate for restriction. In this regard, artificial heteroduplexes of phage A DNA are not cleaved by R .Eco P I in virw (29).

C. EUKARYOTE ENZYMES Studies of eukaryote methylases have lagged behind those of prokaryote enzymes. Nevertheless, it is already clear that there are general differences. For example, all eukaryote enzymes studied so far are able to methylate denatured DNA; in some cases, they may even prefer singlestranded DNA. All eukaryote enzymes are capable of methylating

528

STANLEY HATTMAN

homologous DNA in v i m ; this may be more of a reflection of the difference between prokaryote and eukaryote DNA than of the enzymes. That is, extensive association of histone and non-histone proteins with eukaryote DNA may interfere with saturation methylation in vivo . Nevertheless, in vivo levels of DNA methylation in mammals are considerably higher than in prokaryotes, and the distribution of MeCyt in pyrimidine tracts is exceedingly complex (32-35). Yet, there is considerable conservation in the pattern among vertebrates (93).It remains to be seen whether eukaryotes have a methylase with low specificity or several enzymes with high specificity. The (partial) sequences characterized so far do not appear to be palindromes (see Table 11). Thus, ability to methylate denatured DNA may be due to the fact that these enzymes do not have to make bilateral symmetric contacts in a duplex site (70); i.e., their recognition involves asymmetric contact with only one strand of DNA. Preference for native versus denatured DNA has proved to be a complicated situation, since origin of the DNA and salt concentration each significantly affects the relative rates of methylation. The first experiments aimed at analyzing the interaction between a DNA methylase and DNA were performed with a partially purified enzyme fraction from rat liver (101). The results of these studies indicated that the enzyme forms two types of complexes with DNA in the absence of AdoMet: (i) A weak complex (dissociates in 0 . 2 M NaCl at 0");and (ii) a tight complex (formed at higher temperatures and stable in 0.2 M NaCl). The formation of the tight complex appeared to be required for methylation to occur; it was not known whether formation of this tight complex involves recognition of specific binding sites. It was later found (102) that the enzyme can form a tight complex at 0" with denatured DNA, but not with native DNA. This was interpreted to mean that helical DNA may have to locally unwind in order for tight binding to occur. It was also noted that tight binding can occur with nonacceptor DNA, such as phage 93. M. J. Browne and R. H. Burdon, Nucleic Acids Res. 4, 1025 (1977). 94. D. Simon, F. Grunert, U. v. Acken, H. P. Doring, and H. Kroger, Nucleic Acids Res. 5, 2153 (1978). 95. M. J. Browne, J. F. Turnbull, E. L. McKay, R. L. P. Adams, R. H. Burdon, Nucleic Acids Res. 4, 1039 (1977). 96. R. L. P. Adams, E. L. McKay, L. M. Craig, and R. H. Burdon, BBA 561,345 (1979). 97. T. W. Sneider, Nucleic Acids Res. 8, 3829 (1980). 98. L. H. T. van der Ploeg, J. Groffen, and R. A. Flavell, Nucleic Acids Res. 8, 4563 (1980). 99. P. H. Roy and A. Weissbach, Nucleic Acids Res. 2, 1669 (1975). 100. S. Bromberg, K. Pratt, and S. Hattman, submitted for publication (1981). 101. D. Drahovsky and N . R. Morris,JMB 57, 475 (1971). 102. D. Drahovsky and N. R. Morris, J M B 61, 343 (1971).

529

25. DNA METHYLATION TABLE I1

EUKARYOTE DNA METHYLASES Source of enzyme

(I

Rat liver (normal and regenerating) Krebs I1 mouse ascites tumor cells Novikoff rat hepatoma cell lines HeLa cells Chlumydornonas reinhardi Tetrahymena thermophila

Molecular weight 115,000 184,000

Not analyzed 120,000 55,000-58,000

Not analyzed

Methylatedb sequence X-N-C*-N-Y X-N-C*-N-Y X-N-C*-N-Y X-(G/C)-C*-G-Y X-T-C*-(PU/C)-Y X-N-A*-T-Y

Ref. (94 ) (95, 96) (34,35) (99) (55) (100)

“ DNA methylases have been extracted from a variety of organisms: they are not listed due to a lack of data relevant to the table. * The sequence presented for the rat liver and cultured cells is derived from in vitro and in vivo studies (34, 35, 95, 97, 98). In fact, it is difficult to deduce a sequence because of the complex C* distribution in pyrimidine tracts. Furthermore, it is not clear yet whether there is only a single methylase species; thus, the methylation pattern could be a composite of two or more patterns. It should be noted that C*-C (33, 97, 98) and C’-T (.?3,35)have been found in vertebrate DNA. In the three cases listed above, A has not yet been demonstrated at the 3’ N position. The Chlamydomonas sequence differs from that proposed originally [namely, T-C*-Pu (55)]. The dinucleotide, C-C,* was observed as a significant fraction; therefore, it must be included in the sequence. The Tetmhymena sequence was deduced from in vivo methylation studies; in britro both T and C are 3‘ nearest neighbors, however.

T4 DNA. Thus, tight binding may be necessary, but not sufficient, for methylation t o ensue. It was also shown that the rat liver DNA methylase functioned processively; i.e., it continues to “walk” along the DNA transferring methyl groups at appropriate sites without detaching (101). A reinvestigation of the rat liver DNA methylase was carried out using a more purified preparation (94); these workers confirmed the temperature-dependent formation of a salt-resistant, tight complex. The ability of highly purified DNA methylase from mouse Krebs I1 ascites cells to form a tight complex with native DNA, however, was dependent upon the source of the DNA (96). Finally, a DNA-cytosine methylase from Chlumydomonas was analyzed for its interaction with substrates AdoMet and DNA (55). Kinetic data were consistent with the enzyme randomly binding to either DNA or AdoMet, and then binding the other to form a ternary complex. The sequence specificity of eukaryote DNA methylases has been studied by in vitro and in vivo labeling of DNA (see Section IV,B), and by analyzing the susceptibility of DNA to cleavage by site-specific endonucleases of known sequence specificity and methylation susceptibility. DoskoEil and Sorm (31) were the first to analyze the distribution of MeCyt in pyrimidine tracts. In wheat germ and mammalian DNAs a large fraction of

530

STANLEY HATTMAN

the MeCyt is in the sequence Pu-C*-Pu (up to 50%); the remaining MeCyt is in other oligopyrimidine tracts (Py,, C*). In this study, the identity of the Pu base was not established. Based on Sinsheimer's earlier work (103, 104), the authors concluded that Pu must be G. Subsequently, Sinsheimer's studies have been repeatedly (mislquoted as having shown that C* is mainly in the sequence C*-G. In actuality, the original observation was quite different. Following pancreatic DNase I digestion, the resulting dinucleotides were isolated and analyzed for base composition. For wheat germ DNA, 70% of the MeCyt was in the dinucleotide (C*, G); almost all the calf thymus DNA MeCyt was in this fraction. Contrary to what is generally quoted, the sequence of the dinucleotide was G-C* (104). It is not clear why no 3' nearest neighbor was observed in any of the dinucleotides, or why no Py was found as a 5' nearest neighbor; although DNase I exhibits some cleavage specificity (105, 106) our experience indicates it should be possible to recover all sequence isomers. An indirect demonstration of C*-G came from studies with the sea urchin (1071, where it was shown that 60% of the MeCyt was in Pu-C*-Pu and that greater than 90% of MeCyt was in the (C*,G) dinucleotide. Since the (C*,G) dinucleotide was not sequenced, it can only be inferred that it contained both G-C* and C*-G. A number of other studies have shown that G is not the only 3' nearest neighbor of MeCyt. Sequence analysis of pyrimidine tracts containing MeCyt have clearly demonstrated C*-C and C*-T, and C* may occupy internal positions in tripyrimidine tracts (34, 35, 97, 98). In fact, for rat DNA, the sequence C*-C-G-G occurs about 6 to 7 times more frequently than C-C*-G-G (97). On the other hand, direct sequence analysis of Xenopus luevis 5 S DNA fragments showed that in the G-C-rich region, all C-Gs are at least partially methylated and G is the only 3' nearest neighbor of C* (48). However, in other vertebrates, many C-G sites are unmethylated or variably methylated (97, 98; see later refs. 151-169). The X. luevis 5 S DNA study also yielded valuable information regarding the 5' nearest neighbors as well as the penultimate 3' and 5' bases (48).The sequences surrounding nearly two dozen methylated sites were established. From these data one can deduce the recognition site (Pu/C)-N-C*-G-N. It can be seen that this sequence is related to the other vertebrate methylase sites listed in Table 11. 103. R. L. Sinsheimer,JBC 208, 445 (1954). R. L. Sinsheimer,JBC 215, 219 (1955). S. D. Ehrlich, U. Bertazzoni, and G . Bernardi, EJB 40, 143 (1973). G. Bernardi, S. D. Ehrlich, and J. P. Thiery, Nurure New B i d . 246, 36 (1973). P. Grippo, M. Iaccarino, E. Parisi, and E. Scarano,J M B 36, 195 (1968).

104. 105. 106. 107.

2s. DNA METHYLATION

IV.

Distribution of Methylated Bases

A.

SURVEY AMONGPROKARYOTES

53 1

With a few exceptions all prokaryotes contain either one or both methylated bases. The gene(s) controlling DNA-methylase activity may be located on the cell chromosome, phages, or plasmids (see Table I). Previous reports that phage 4x174 specifies a DNA-cytosine methylase (108, 109) have not been substantiated (110, 111). Although it has been generally assumed that DNA methylase is synthesized constitutively, there are several cases where enzyme production is inducible (78, 112-115). In one instance, DNA-methylase induction may be correlated with induction of a defective prophage (112-114). An interesting case is that of B. subtifis, where enzyme activity is correlated with physiological state; namely, onset of transformation competence (78, 115). In general, all potential methylation sites appear to be methylated in vivo . This has been repeatedly established by showing that DNAmethylase activity from any given strain is unable to methylate its homologous DNA in vitro. However, DNA isolated from phage does not have all sites methylated (36, 116-119); this is likely t o result from competition between DNA methylation and maturation into phage particles. With a sufficient high level of enzyme activity it is possible to methylate all potential sites before phage DNA is packaged into virions (36, 118, 120 1. Finally, there is the extraordinary case of phage XP12,grown in Xunthomonus oryzue, in which all cytosine residues are replaced by MeCyt (121, 122). This is not due to modification at the polynucleotide level, how108. A. Razin, P N A S 70, 3773 (1973). 109. A. Razin, D. Goren, and J. Friedman, Nucleic Acids R e s . 2, 1967 (1975). 110. S. Hattman, C. Gribbin, and C. A. Hutchison 111, J . Virol. 32, 845 (1979). 111. A. Razin, S. Urieli, Y.Pollack, Y. Gruenbaum, and G. Glaser, Nucleic Acids R e s . 8, 1783 (1980). 112. A. Yudelevich and M. G o l d , J M B 40, 77 (1969). 113. G. Medoff and M. N. Swartz, J . Gen. Virol. 4, IS (1969). 114. U. Giinthert, B. Pawlek, J. Stutz, and T. A. Trautner, J . Virol. 20, 188 (1976). 115. A. T. Ganesan, J. Eacreriol. 139, 270 (1979). 116. S. Hattman, J. Virol. 10, 356 (1972). 117. S. Hattman, J M E 74, 749 (1973). 118. S. G. Hughes and S. Hattman, J M B 98, 645 (1975). 119. S . Hattman, J . Barreviol. 129, 1330 (1977). 120. S. Hattman, S. Schlagman, and L. Cousens, J . Bucteriol. 115, 1103 (1973). 121. T. T. Kuo,T. C. Huang, and M. H. Teng, JMB 34, 373 (1968). 122. M. Ehrlich, K. Ehrlich, and J. A. Mayo, BBA 395, 109 (1975).

532

STANLEY HATTMAN

ever. The direct incorporation of MedCMP into DNA accounts for its complete replacement of Cyt (123).

B. SURVEY AMONGEUKARYOTES Until several years ago the only methylated base clearly demonstrated in eukaryote DNA was MeCyt. It is now evident that many unicellular eukaryotes contain MeAde; e.g., certain green algae (16, 124, 125), protozoa (126-128), and dinoflagellates (129). The presence of a low level of MeAde (- 0.1% of the adenines) has also been reported for the mosquito Aedes afbopictus (7). It is interesting to note that several of the green algae contain both MeAde and MeCyt (124, /25), while the protozoa contain only MeAde (125-128, 130). With the exception of the mosquito, all higher eukaryotes contain only MeCyt, and the range of methylation varies from -0.17% for insects up to 50% for higher plants. It is a matter for speculation as to why DNA-adenine methylation has been discarded in the process of evolution to multicellular organisms. The phenomenon of DNA methylation has been studied from a variety of viewpoints. Early studies included analyses of MeCyt content in various tissues where it appeared that methylation levels were tissue-specific (131, 132). There is some disagreement as to whether the reported differences in overall MeCyt content are really significant or not; however tissue-specific differences in methylation are clearly evident when specific sites are analyzed (Section VI,B,3). At the cellular level, DNA from organelles was also a focus of attention with respect to methylation content. Again, there is no universal pattern; i.e., depending on the organism, mitochondria1 DNA is or is not methylated (10, 133-135). In plants and algae chloroplast DNA is generally not methylated (16, 125, 135-137) 123. M. Ehrlich, F. U . Lin, K. Ehrlich, S. L. Brown, and J. A. Mayo,J. Virol. 23, 517 (1977). 124. M. W. Pakhomova, G . N . Zaitseva, and A. N. Belozerskii, Dokl. Akad. Nauk SSSR 182, 712. (Engl. transl.) (1968). 125. S . Hattman, C. Kenney, L. Berger, and K. Pratt,J. Bacreriul. 135, 1156 (1978). 126. M . A. Gorovsky, S. Hattman, and G . L. Pleger, J . Cell B i d . 56, 697 (1973). 127. D. J. Cummings, A . Tait, and J. M . Goddard, BBA 374, 1 (1974). 128. P. M. M. Rae and B. B. Spear, PNAS 75, 4992 (1978). 129. P. M. M. Rae, Science 194, 1062 (1976). 129a. R. E. Steele and P. M. M. Rae, Nucleic Acids Res. 8, 4709 (1980). 130. K. Pratt and S. Hattman, M o l . CeIl. Bid 1, in press (1981). 131. J. W. Kappier,J. Cell. Physiul. 78, 33 (1971). 132. B. F. Vanyushin, A. L. Mazin, V. K . Vasilyev, and A. N . Belozersky, BBA 299,347 (1973). 133. I . B. Dawid, Science 184, 80 (1974).

25. DNA METHYLATION

53 3

(however, see Section VI,B,2). In the protozoan, Tetrahymena thermophiia, macronuclear DNA contains MeAde, but micronuclear DNA lacks MeAde (126, 128). It was reported that both DNAs are methylated in Paramecium, however (127). Since the micronucleus is the germinal nucleus in these organisms, one might expect to see a general pattern, so it is hoped that these are only experimental differences. It is important to emphasize that eukaryote DNA is not methylated to saturation in vivo. As mentioned earlier, eukaryote DNA can be further methylated in vitro by the homologous enzyme(s), although the degree of undermethylation has not yet been determined. It is likely that the extensive association of histones and other proteins with eukaryote DNA sterically interferes with DNA methylation. In this regard, recent investigations have also analyzed the distribution of methylated bases with respect to chromatin structure. Chromatin is known to be composed of basic repeating units (nucleosomes); the repeat length may vary from 150 to 240 base pairs (bp), depending on the organism or tissue. The nucleosome core contains 145 bp of DNA associated with an octamer of histones H2A, H2B, H3, and H4; the heterogeneity in nucleosome repeat length is generally due to variability in the length of internucleosome core DNA, i.e., linker DNA (see reference 138 for a review). Analyses of MeCyt distribution in chromatin of a variety of mammalian cells have led to three different conclusions: MeCyt is randomly distributed (1391, is preferentially in core DNA (140, /41), and is preferentially in linker DNA (142). There is no obvious explanation for these contradictions other than each study used different methods and different organisms. In our laboratory we observed that there is a severalfold enrichment of MeAde in Tetrahymena linker DNA (i.e., staphylococcus nuclease-sensitive DNA) (130). The distribution of MeCyt with respect to chromosome structure has also been investigated by in siru indirect immunofluorescence techniques using antibodies against MeCyt (2/-25). The results of these studies indi134. (1974). 135. 136. 137. 138. 139. 4, 3097 140. 141. 142.

B. F. Vanyushin and M. D. Kirnos, FEES (Fed. Eur. Biochem. SOC.)Lett. 39, 195

G. Brawerman, D. A . Hufnagel, and E. Chargaff, BBA 61, 340 (1962). G. Brawerman and J. M. Eisenstadt, BBA 91, 477 (1964). D. S. Ray and P. C. Hanawalt, J M B 9, 812 (1964). J. D. McGhee and G. Felsenfeld, Annu. Rev. Biochem. 49, 11I5 (1980). R. L. P. A d a m , E. L. McKay, J. T. Douglas, and R. H. Burdon, Nucleic Acids Res. (1977). A. Razin and H. Cedar, PNAS 74, 2725 (1977). A . Solage and H. Cedar, Biochemistry 17, 2934 (1978). N. N . Khodarev, I. I. Votrin, N . N . Sokolov, and A. G. Basnakyan, Biokhimiya 44, 1058 (Engl. transl.) (1979).

534

STANLEY HATTMAN

cate that methylation is heavy in some, but not all, classes of constitutive heterochromatin (particularly the centromeric regions). Since heterochromatic regions appear to be transcriptionally inactive, there may be an inverse correlation between gene activity and DNA methylation; this is discussed further in Section VI,B,3. At the molecular level, other studies have analyzed methylation distribution according to sequencerepetitiveness and “satellite” character. For example, mouse cell DNA can be fractionated by equilibrium density centrifugation into a main band (42% G-C) and a satellite band (35% G-C) (143). The satellite DNA appears to originate from the centromeric ends of nearly every chromosome (144), and it corresponds to the constitutive heterochromatin previously discussed. The satellite is both highly repetitive and highly methylated (32, 33, 37). i.e., the MeCyt content is two- to threefold higher than main band DNA (32, 37). Other examples of higher methylation in satellite DNA are known, e.g., the highly reiterated HS-P satellite in the kangaroo rat is methylated sevenfold higher than bulk DNA (145). This is not a constant feature of all satellite DNAs, however, because two other satellites in the same organisms are not more highly methylated than bulk DNA (145). There are numerous reports demonstrating that repetitive sequences are relatively more highly methylated (146-148); “fold-back” (inverted-repeat) sequences have been shown to be more highly methylated than bulk DNA in mouse mastocytoma cells (149, 150). Specific, amplified genes have been investigated for methylation. In the African toad, X. laevis, amplified ribosomal DNA (rDNA) genes that are actively transcribed in oocytes are not methylated; in contrast, nonamplified somatic rDNA genes contain MeCyt (151). The same relationship was observed in Physarum polycephalum rDNA with respect to methylation of C-C-G-G sequences (152). In the protozoan, Tetrahymena therrnophila, macronuclear rDNA is amplified and extrachromosomal in 143. S. Kit,JMB 3, 711 (1961). 144. M. L. Pardue and J. Gall, Science 168, 1356 (1970). 145. K. Fry, R. Poon, P. Whitcome, J. Idriss, W. Salser, J. Mazrimas, and F. Hatch, PNAS 70, 2642 (1973). 146. M. Brahic and M. J. Fraser, BBA 240, 23 (1971). 147. M. H. Schneiderman and D. Billen, BBA 308, 352 (1973). 148. J. Sawecka, L. Kornacka, and J. Malec, Experientia 35, 1166 (1979). 149. D. Drahovsky, T. L. J. Boehm, and W. Kreis, BBA 563, 28 (1979). 150. T. L. J. Boehm and D. Drahovsky, Z. Nuturforsch. 3 9 , 61 1 (1980). 151. I. B. Dawid, D. D. Brown, and R. H. Reeder,JMB 51, 341 (1970). 152. J. C. Redly, R. Braun, and C. A. Thomas, Jr. FEBS (Fed. Eur. Biochern. Soc.) Lett. 116, 181 (1980).

535

25. DNA METHYLATlON

vegetative cells; however, unlike higher eukaryote systems, the amplified rDNA is methylated to an extent similar to that in bulk DNA (153). The most recent development in the study of eukaryote DNA methylation has been the application of site-specific nucleases to generate profiles of methylation within specific gene regions (97, 98, 152, 1.54-165). This strategy has also been applied to the analysis of virus-specific genes in transformed and infected cells (166-170). For example, the methylation status of the sequence C-C-G-G can be probed by examining the cleavage fragments generated by R HpaII (will not cleave C-C*-G-G)and R . MspI (will not cleave C*-C-G-G); both enzymes cleave C-C-G-G, or sites methylated at the “wrong” C position. Thus, it has been possible to establish for certain regions that particular C-C-G-G sites are unmethylated, always methylated or variably methylated. By analyzing DNAs from appropriate tissues, it has been possible to compare genes in transcriptionally active versus inactive states. These studies suggest that certain active genes are undermethylated at specific sites; this issue is discussed further in Section VI,B,3. Site-specific nucleases were used in an elegant study of the sea urchin Echintis esculentus (164). Whereas DNA methylation was believed to be distributed thoughout, in the sea urchin the genome can be subdivided into two distinct classes: (i) Forty percent of the DNA consists of long tracts (15 to more than 50,000 bp in length) in which many sites are consecutively methylated; and (ii) 60% of the genome is not significantly methy1

153. P. M. M. Rae and R. E. Steele, Biosysfems 10, 37 (1978). 154. J. Singer, J. Roberts-Ems, and A. D. Riggs, Science 103, 1019 (1978). 155. A. P. Bird and E. M. Southern,JMB 118, 27 (1978). 156. A. P. Bird, JMB 118, 49 (1978). 157. M. J. Browne, A. C. B. Cato, and R. H. Burdon, FEBS (Fed. Eur. Biochem. Soc.) L e u . 91, 69 (1978). 158. C. Waalwijk and R. A. Flavell, Nucleic Acids Res. 5, 3231 (1978). 158a. C. Waalwijk and R. A. Flavell, Nucleic Acids Res. 5 , 4631 (1978). 159. J. D. McGhee and G. D. Ginder, Nafure (London) 280, 419 (1979). 160. H. Cedar, A. Solage, G. Glaser, and A. Razin, Nitcleic Acids Res. 6, 2125 (1979). 161. J. L. Mandel and P. Chambon, Nircleic Acids Res. 7, 2081 (1979). 162. M. T. Kuo, J. L. Mandel, and P. Chambon, Niicleic Acids Res. 7, 2105 (1979). 163. J. Kaput and T. W. Sneider, Nitcleic Acids Res. 7, 2303 (1979). 164. A. P. Bird, M. H. Taggart, and B. A. Smith, Cell 17, 889 (1979). 165. A. P. Bird and M. H. Taggart, Nrtcleic Acids Res. 8, 1485 (1980). 166. R. C. Desrosiers, C. Mulder, and B. Fleckenstein, P N A S 76, 3839 (1979). 167. L. Vardimon, R. Neumann, I. Kuhlmann, D. Sutter, and W. Doerfler, Nucleic Acids Res. 8, 2461 (1980). 168. D. Sutter and W. Doerfler, P N A S 77, 253 (1980). 169. J. C. C. Cohen, Cell 19, 653 (1980). 170. D. Eick, S. Stabel, and W. Doerfler,J. Virol. 36, 41 (1980).

536

STANLEY HATTMAN

lated. Moreover, the two classes do not appear to overlap with respect to the unique sequences included in each, and the sequence composition of each class is the same in sperm, embryo and adult (intestine) DNA. The functional significance of these (un)methylated compartments is not clear at this time; however, they are not confined to Echinus because a variety of other marine invertebrates exhibit a similar pattern with respect to methylation of C-C-G-G and G-C-G-C sequences (16.5 1. In contrast to bacterial viruses, animal virus DNA (isolated from virions) is unmethylated (171-174). Considering that cell DNA is modified, it is surprising that these viruses are unmethylated, even though they replicate in the host nucleus. This cannot be due to a lack of methylase recognition sites because viral DNA is methylated after integration into the host genome of transformed cells (166-170). It has been reported that intracelMar herpes simplex virus (HSV-I) DNA is transiently methylated during replication (175), suggesting that there may be a subsequent demethylation event. Since alternative explanations were not ruled out, we must await further studies.

V.

Other DNA Modifications

In addition to methylated adenine and cytosine residues, a variety of other DNA modifications are known (see Ref. 176 for a review) and occur mainly, but not exclusively, in bacterial viruses (Table 111). These modifications frequently originate at the level of nucleotide metabolism; as one would expect, replacement of the normal base is extensive in such instances. A remarkable example is that of 5-hydroxymethyluracil, which is incorporated in place of thymine into the DNA of phages 4W14 (177, 178) and SP8 (179). Then, at the level of the polynucleotide, hydroxymethyluracil may be modified to thymine or hypermodified to a-putrescinyl thymine. With the exception of phage Mu (180) and phage DDVI (176, 177) it is 171. 172. 173. 174. 175. 176. 177. 178. 179. 180.

A. M. Kaye and E. Winocour, JMB 24, 475 (1967). M. Low, J. Hay, and H. M. Keir, JMB 46, 205 (1969). U. Giinthert, M. Schweiger, M . Stupp, and W. Doerfler, P N A S 73, 3923 (1976). U. v. Acken, D. Simon, H.-P. Doring, and H. Kroger, Virology 99, 152 (1979). S. Sharma and N. Biswal, Virology 82, 265 (1977). R. A. J. Warren,Annu. Rev. Microbiol. 34, 137 (1980). A. M. B. Kropinski, R. J. Bose, and R. A. 1. Warren,Biochmistry 12, 151 (1973). R. A. Kellin and R. A. J. Warren, J . Virol. 12, 1427 (1973). R. G . Kallen, M. Simon, and J . Marmur, J M B 5, 248 (1962). S. Hattman, J . Virol. 32, 468 (1979).

25. DNA METHYLATION

537

likely that other postreplicative phage modifications occur without regard to sequence specificity. However, further studies are required (mutants would be useful) to answer this question. It is extraordinary that 7-MeGua could be a natural substituent of phage DDVI DNA (18/, 182). From studies on the stability of modified guanines and adenines produced by alkylating agents, it has been shown that 7-MeGua is not a stable constituent of DNA (183-185). Even in aqueous solution at neutral pH, 7-MeGua is released by spontaneous hydrolysis of the glycosidic bond. Thus, it would be of interest if identification of this base were independently verified by methods other than chromatographic behavior and absorption spectrum. (See also Refs. 186-194.) VI.

Biological Roles

A.

PROKARYOTES

1. Restriction-Modijication Systems

Although the phenomena of host-induced modification (f95, 196) and DNA methylation were discovered in the early 1950s, it was at least 10 years before the connection could be demonstrated. In brief, bacteria may contain a site-specific (restriction) endonuclease(s) that can degrade exo181. I. I. Nikolskaya, G. N. Trushinskaya, and T. I. Tikhonenko, Biokhimiya 37, 178 (Engl. transl.) ( I 972). 182. I. I. Nikolskaya, N. G. Lopatina, and S. S. Debov, BBA 435, 206 (1976). 183. P. D. Lawley and C. A. Wallick, Chem. Ind.. p. 633 (1957). 184. P. D. Lawley and P. Brookes, Biochem. J . 89, 127 (1963). 185. G. P. Margison and P. J. O’Connor, BBA 331, 349 (1973). 186. M. D. Kirnos, I. Y.Khudyakov, N. I. Alexandrushkina, and B. F. Vanyushin, Nature (London) 270, 369 (1977). 187. G. R. Wyatt and S. S. Cohen, Biochem. J . 55, 774 (1953). 188. I. R. Lehman and E. A. Pratt, JBC 235, 3254 (1960). 189. I. Takahashi and J. Marmur, Nature (London) 197, 794 (1963). 190. J. Marmur, C. Brandon, S. Neubort, M. Ehrlich, M. Mandel, and J. Konvicka, Nature New B i d . 239, 68 (1972). 191. C. Brandon, P. M. Gallop, J. Marmur, H. Hayashi, and K. Nakanishi, Nature (London) New Biol. 239, 70 (1972). 192. C. Brandon, Ph.D. Thesis, Yeshiva Univ., New York, 1974. 193. K. L. Maltman, J. Neuhard, H. A. Lewis, and R. A. J. Warren,J. Virol. 34, 354 (1980). 194. D. Swinton, S. Hattman, R. Benzinger, V. Buchanan-Wollaston, and J. E. Beringer, unpublished observations (1981). 195. S. E. t u n a and M. L. Human, J . Bacteriol. 64, 557 (1952). 196. G. Bertani, and J. Weigle, J . Bacteriol. 65, 113 (1953).

TABLE III

OTHERMODIFIEDBASES NATURALLY OCCURRING IN DNA" Base $ 00

Organisms

Comments

Ref.

Eukaryotes Pyrimidines 5-Hydroxymethyluraci1

Dinoflagellates (Pyrrophyra)

Depending on strain, 12 to 38% replacement of thymine, distributed nonrandomly in DNA

(129. 129a)

Prokaryotes Purines 7-Methylguanine

Shigella and E . coli

1.5% of guanines modified

(181, 182)

Complete replacement of adenine Sequence specific modification of 15% of adenines; added substituent contains a free-COOH group

( 186 ) (66, 180)

Complete replacement of cytosine due to modification at nucleotide level; also contains glucose residues added at level of polynucleotide. Complete replacement thymine

(182, 187, 188)

2-hinoadenine Undefined adenine derivative Pyrimidines 5-Hydroxymethylcytosine

phage DDVI Cyanophage S-2L E. coli phage Mu Shigella and E. coli

phages DDVI, T2, T4, T6 5-Hydrox ymethyluracil

B . subrilis phages SPO1, SP8, SP82G, $25, &, 2C

(179)

5-H ydroxymethyluracil a-htrescinylthymine Uracil

5-(4’,5’-Dihydroxylpentyl)uracil

Pseudomonas phage 6W14 Pseudomonas phage 6W14 B. subtilis phage PBSl and PBS2 B . subtilis phage SPl5

a-Glutamylthymine

B. subtilis phage SPlOc

Undefined cytosine derivative

Rhizobium phage RL38J1

78)

50% replacement of thymineb 50% replacement of thymineb

( 176-1

Complete replacement of thymine

(189)

40% replacement of thymine; modification occurs

(190-192)

at nucleotide level; base is “hypermodified” by addition of glucose residues as postreplication event 15-20% replacement of thymine; the a-carboxyl group is further modified by an unidentified substituent Complete replacement of cytosine

(176. 193)

( 176-1 7 8 )

(194)

“ The modifications tabulated are not always the result of a postreplicative event; cases where modification is known to occur at the level of nucleotides are indicated. In phages SPlOc and 6W14, 5-hydroxymethyluracil (as HM dUTF’) is incorporated in place of thymine during DNA replication (171). As a vI $ postreplicative event the HMU is either modified to thymine or hypermodified.

540

STANLEY HATTMAN

genous DNA transferred into the cell by conjugation, transformation, or infection (197, 198). The molecular fate of transferred DNA depends upon its origin-r host specificity, i.e., on the host in which it had last replicated. Host specificity refers to the presence or absence of methyl groups within specific nucleotide sequences; absence of methylation in these recognition sites results in cleavage of the DNA, either within the site or elsewhere [two exceptional cases are R .DpnI (48) and R .ApyI (/I/), which require methylation for cleavage activity]. Thus, many DNA methylases are part of restriction-modification systems, and they serve to protect the host DNA against action of the indigenous nuclease activity. 2. Mutation In E. coli the modification methylases account for only a fraction of the total DNA-adenine methylation activity. Mutants lacking the major DNA-adenine methylase activity (designated dam-) have been isolated (199) and shown to exhibit a variety of phenotypic properties different than the parental (darn+) wild type. For example, dam- mutants have higher spontaneous mutation frequencies, slightly higher sensitivity to UV-inactivation, increased levels of spontaneous A-prophage induction, increased recombination following conjugation, and increased mutagenesis by base analogs or alkylating agents (200-204); mutants that lack the major E. coli DNA-cytosine methylase do not exhibit such properties (203). To explain the elevated mutability ofdum- strains, it was suggested that these mutants have lost the selectivity of “methylation-instructed mismatch correction” (204). Evidence for the occurrence of base-mismatch repair had been obtained in a variety of systems, and it was proposed that DNA methylation might be the basis for ensuring discrimination between parental and nascent progeny strands (205). This was supported by the observation that mismatch repair of heteroduplex A DNA occurs preferentially on the unmethylated versus methylated strand (’05a). Thus, in the W. Arber and S. Linn, Annu. Rev. Biochem. 38, 467 (1969). H. W. Boyer, Annu. Rev. Microbid. 25, 153 (1971). M. G. Marinus and N. R. Morris, J . Bacterial. 114, 1143 (1973). M. G. Marinus and N. R. Morris, JMB 85, 309 (1974). M. G . Marinus and N. R. Morris, Mutat. Res. 28, 15 (1975). M. G. Marinus and N . R. Morris, Mol. Cen. Genet. 149, 273 (1976). A. Bale, M. D’Alarcao, and M. G. Marinus, Mutat. Res. 59, 157 (1979). B. Glickman, P. van den Elsen, and M. Radman, Mol. Gen. genet. 163, 307 (1979). R. Wagner, Jr. and M. Meselson, PNAS 73, 4135 (1976). 205a. M. Radman, R. E. Wagner, Jr., B. W. Glickman, and M. Meselson. Progr. Environ. Mutagen., p. 121 (1980). 197. 198. 199. 200. 201. 202. 203. 204. 205.

54 1

25. DNA METHYLATION

absence of DNA-adenine methylation, the parental strand and the nascent progeny strand cannot be distinguished, and mismatch repair can occur on either strand; it follows that when the parental strand is repaired, replication produces two daughter molecules, each with a single base change from the wild-type. [Studies in my laboratory indicate that a high MeAde content increases the frequency of spontaneous mutation in phage T2gt(205b), suggesting that MeAde can act as a base analogue mutagen.] This model is attractive because it not only suggests a novel mechanism for maintaining fidelity during DNA replication, but also proposes a new and important biological function for DNA-adenine methylation. Although the major DNA-cytosine methylase ofE. coli, specified by the dcm (or mec) gene (120,200) does not appear to have a biological function, it has the same sequence specificity as the RII modification enzyme (36, 41,45,49, 119). As indicated above, dcm- (ormec-) mutants do not exhibit the dam- phenotype. However, they do lose the ability to protect DNA against RII restriction (120) and they no longer exhibit the two specific 2-aminopurine mutational “hot-spots” in the lac I gene (206). The notion that enzymatic deamination of MeCyt residues (to produce Thy) might be a pathway for spontaneous transition mutations during cell differentiation was proposed sometime ago (207). In the case of the lac I gene sites, however, recent findings argue against 2-aminopurine-induced deamination of MeCyt as the molecular basis for the hot-spots (208). 3 . Expression of Phage Mu Mom + Modification Escherichia cofi phage Mu DNA is resistant to a variety of restriction systems in vivo (209, 210). This phenotype requires expression of the phage mom gene and the host dam gene (209-212). Fifteen percent of the adenine residues in Mu DNA are modified to a new form, A, (180); although the molecular nature is not yet defined, the added substituent appears to contain a free carboxyl group (180). The modification occurs in specific sequences (66, 212), and it requires expression of both the mom and dam + genes. These results suggest that A, modification is responsible for protecting Mu DNA against restriction. The role of the host dam+ +

+

+

205b. S. Hattman, A. Iannotti, and S. Schlagman, Manuscript in preparation (1981). 206. C. Coulondre, J. H. Miller, P. J . Farabaugh, and W. Gilbert, Nature (London) 274, 775 (1978). 207. P. Grippo, E. Parisi, C. Carestia, and E. Scarano, Biochemistry 9, 2605 (1970). 208. B . K . Duncan and .I. H. Miller, Nature (London) 287, 560 (1980). 209. A. Toussaint, Virology 70, 17 (1976). 210. A. Toussaint, J . Virol. 23, 825 (1977). 211. H. Khatoon and A. I. Bukhari,J. Bacteriol. 136, 423 (1978). 212. R. Kahmann and D. Kamp, personal communication (1979).

542

STANLEY HATTMAN

methylase is not understood; we do know that the mom+ protein does not alter the methylase sequence specificity. Transcription of the phage mom' gene may require site-specific M * Eco dam methylation. This is currently under investigation.

4. Methylations Looking for a Function Two sets of observations are worth mentioning here briefly, even though their significance with respect to biological function remains to be shown: (i) In B. subtilis, DNA-cytosine methylase activity changes as a function of physiological state (78) and reaches the maximum level when cells become competent for DNA transformation (I 14). Coincident with this increase in activity is the extensive methylation of resident and donor DNA, and termination of nucleolytic activity on resident DNA (114). It remains to be seen whether the methylase is involved in the transformation and recombination processes. (ii) The region of DNA presumed to contain the origin of DNA replication (ori C) has been cloned and sequenced (213,214). The E. coli and S . typhimurium regions both have the interesting feature of including a large number of G-A-T-C sequences. This sequence is the methylation site for the major DNA-adenine methylase in these cells (Table I). Thus, the MeAde content of this region (15% of the Ade residues modified to MeAde) should be an order of magnitude higher than the overall methylation level of total DNA. It is known that MeAde * T base pairs are more readily denatured than Ade * T base pairs (215). Thus, the ori C region may be more readily melted than bulk DNA, and this may be involved in its function as an origin of replication. However, this cannot be an obligatory function, otherwise dammutations would be lethal. There have been reports suggesting that DNA methylation is essential for replication of phage 4x174 DNA (109) and E. coli DNA (216). These conclusions appear to be invalid in the light of later studies (110, 111, 199). The fact that certain bacterial strains are devoid of MeAde and MeCyt (99, 217, 218) rules out a required role of methylation in replication or transcription. 213. K . Sugimoto, A. Oka, H. Sugisaki, M. Takanami, A. Nishimura, Y. Yasuda, and Y. Hirota, PNAS 76, 575 (1979). 214. M. Meder, E. Beck, F. G. Hansen, H. E. N. Bergmans, W. Messer, K . von Meyenburg, and H. Schaller, PNAS 76, 580 (1979). 215. J. D. Engel and P. H. von Hippel, JBC 258, 927 (1978). 216. C. Lark, J M B 31, 401 (1968). 217. J. T. Wachsman and V. Irwin, J . Bacreriol. 104, 814 (1970). 218. A. Schein, B. J. Berdahl, M. Low,and E. Borek, BBA 272, 481 (1972).

25. DNA METHYLATION

543

B. EUKARYOTES 1. Restriction-ModiJcution Systems

The restriction-modification systems in prokaryotes are obvious models for higher organisms. However, no one has been able to demonstrate virus restriction based on DNA host-specificity. This comes as no surprise in view of the absence of DNA-methylation in virion DNA. Attempts to isolate prokaryote-like restriction-nucleases have also been largely unsuccessful. However, there are candidates for restriction-enzyme status, e.g., the AdoMet-dependent DNase from nuclei of cultured hamster kidney (BHK) fibroblasts (219). This enzyme does not require ATP or Mg'+ for activity and is inhibited by the AdoMet analogue, S-adenosylhomocysteine. A variety of native and denatured DNAs are substrates for cleavage, including phage A DNA and BHK 21 DNA (from methionine-deprived cells); however, no discrete fragments are produced. When BHK 21 cells are allowed to grow in medium containing methionine, following a growth-deprivation period, the DNA becomes insensitive to cleavage. Moreover, when A DNA is methylated in v i m by a BHK 21 methylase, it is made resistant to cleavage by the BHK nuclease. These results bear striking similarity to prokaryote restriction and modification enzymes. An endonuclease stimulated by ATP or AdoMet has been found in African green monkey cells (2201, and finally, endonuclease activities have been isolated from Chlumydomonus (221) and Epstein-Barr virus-infected cells (222) that may be site-specific. Needless to say, more information is needed about the properties and functions of these enzymes. 2 . Muternul lnheritunce in Chlumydomonus The green alga, Chlamydomonus, contains MeCyt and MeAde in the nuclear DNA of vegetative cells (124, 12.5); there may be strain differences, insofar as some have been reported to lack MeAde (16). In contrast, chloroplast DNA is generally not methylated in vegetative cells (16, 12.5, 223, 224). Genetic studies have convincingly demonstrated that maternally inherited markers in Chlumydomonus are in chloroplast DNA (16, 219. (1979). 220. 221. 222. 223. 224.

A. C. B. Cat0 and R. H. Burdon, FEBS (Fed. Eur. Biochem. S O C . )Lett. 99, 33

F. L. Brown, P. R. Musich, and J . J. Maio, Nircleic Acids Res. 5 , 1093 (1978). W. G . Burton, R. J. Roberts, P. A . Meyers, and R. Sager, PNAS 74, 2687 (1977). W. Clough, PNAS 77, 6194 (1980). R. Sager and D. Lane, PNAS 69, 2401 (1972). H. D. Royer and R . Sager, PNAS 76, 5794 (1979).

544

STANLEY HATTMAN

225,226). In this system, genetic crosses lead to transmission of markers by the mt mating type (maternal parent) and not by the mt- mating type (paternal parent). Physical, chemical, and enzymatic studies by Sager and co-workers (16, 223, 224) indicate that m t f chloroplast DNA survives destruction because it is heavily methylated during gametogenesis and again in the young zygote. In contrast, mt- chloroplast DNA is only sparingly methylated during gametogenesis and is degraded shortly after gametic fusion. These studies nicely explain maternal inheritance on the basis of differential methylation and degradation of DNA. However, a number of questions arise from the experiments; e.g., Why was uracil observed in gamete and zygote chloroplast DNA, and how can zygote chloroplast DNA be simultaneously resistant to cleavage by R . MspI, R . HpaII, R BamHI and R EcoRI? The latter observation has very interesting implications [note that mt gamete chloroplast DNA (but not m t - gamete) is resistant to R . HpaII only; resistance to the three other enzymes is only partial in the gamete, but almost complete in the zygote {224)].Based on the known specificities of these nucleases, it appears that two distinct waves of methylation occur during gametogenesis and gamete fusion, respectively. In the first, C-C-G-G sites are all methylated to C-C*-G-G and become resistant to R HpaII; then, in the zygote, there follows methylation of C-C*-G-G to C*-C*-G-Gand all T-C to T-C*. Since R . MspI does not cleave C*-C-G-G (97), it is necessary to methylate the 5’ C of C-C-G-G and C-C*-G-G to protect these sites. Also, methylation of T-C is a reasonable possibility considering that a Chalmydomonas methylase is known to produce T-C*-(PdC) (55) (see Table 11). This methylation would in fact protect all the Bam HI sites, since R . Bum HI does not cleave G-G-A-T-C*-C (28). What is most unusual is that RI-sites (G-AA-T-T-C) are not cleaved. Although RI modification is mediated by adenine-methylation to produce G-A-A*-T-T-C, C methylation to GA-A-T-T-C* might also protect against R . EcoRI. If so, this is quite surprising in view of the fact that G-A-A-T-T-hydroxymethyl C is sensitive to R .EcoRI (89, 90). This possibility should be kept in mind when using R * EcoRI to cleave DNA that contains a high level of MeCyt because some RI sites may be resistant. Another conclusion that may be drawn from the above studies is that Chlamydomonas contains two or more DNA-cytosine methylases that are subject to regulatory controls. The enzyme described by Sano and Sager (55) methylates T-C*-(PdC); therefore, one or more activities are required to methylate the C-C-G-G sequences. Furthermore, G-G-C-C sites in

-

225. R. Sager, “Cytoplasmic Genes and Organelles.” Academic Press, New York, 1972. 226. D. M. Grant, N. W. Gillham, and J. E. Boynton, PNAS 77, 6067 (1980).

25. DNA METHYLATION

545

chloroplast DNA are known to be methylated (226a) in certain mtstreptomycin-dependent strains (as determined by resistance to R . HaeIII); thus, there may be another activity as well. It was also reported (226a) that these mt- strains contain chloroplast DNA with 50% of the cytosines modified to MeCyt, even during vegetative growth. Nevertheless, in crosses to normal mt+ strains, the usual pattern of uniparental inheritance is observed. In the light of this finding, the question remains open whether methylation has anything to do with the phenomenon of maternal inheritance. 3 . Gene Regulation and Differentiation

The first attempts at gaining insight as to whether DNA methylation is involved in gene regulation were to compare MeCyt contents in DNAs from cells or tissues in different stages of differentiation; e.g., after partial hepatectomy in rats (2271, after administration of hormones (228) or carcinogens (229-234), in normal versus neoplastic cells (173, 234-242), and during different stages of growth and development (12, 13,14, 107,243, 244). Notwithstanding these intensive efforts, no definitive conclusions could be drawn concerning the function of methylation. On the other hand, there are reports that claim that methylation is implicated in learn226a. D. Grant, D. C. Swinton, P. L. Bolen, K.-S., Chiang, N. C. Gillham, and J. E. Boynton, J . Cell Biol. 83, (Part 2 , Abstr. MC 1930) (1979). 227. N. R. M o m s and K. D. Pih, Cancer Res. 31, 433 (1971). 228. R. W. Turkington and R. L. Spielvogel, JBC 246, 3835 (1971). 229. R. K. Datta and B. Datta, Exp. Mol. Pathol. 10, 129 (1969). 230. J. A. Miller, Cancer Res. 30, 559 (1970). 231. B. Singer, Progr. Nucleic Acid Res. Mol. Biol. 15, 219 (1975). 232. P. Lawley, in. “Chemical Carcinogenesis” (C. E. Searle, ed.), p. 84. ACS Monograph 173, (1976). 233. P. N. Magee, R. Montesano, and R. Preussman, i n , “Chemical Carcinogenesis” (C. E. Searle, ed.), p. 491. ACS Monograph 173, (1976). 234. A. E. P e a , Advan. Cancer Res. 25, 195 (1977). 235. N. A. Fedorov, V. A. Kuzmichev, G. A. Kritskii, and Yu. E. Vinogradova, Biokhimiya 42, 791 (Engl. transl.) (1977). 236. R. Silber, E. Berman, B. Goldstein, H. Stein, G. Farnham, and J. R. Bertino, BBA 123, 638 (1966). 237. T. W . Sneider and V. R. Potter, JMB 42, 271 (1969). 238. L. S. Desai, U. C. Wulff, and G. E. Foley, Exp. Cell Res. 65, 260 (1971). 239. P. D. Lawley, A. R. Crathorn, S. A. Shah, and B. A. Smith, EJ 128, 133 (1972). 240. S. Shirakawa and G. F. Saunders, Proc. SOC.Exp. Biol. Med. 138, 369 (1971). 241. L. A. Culp and P. H. Black, BBA 247, 220 (1971). 242. E. D. Rubery and A. A. Newton, BBA 324, 24 (1973). 243. D. E. Comings, Exp. Cell Res. 74, 383 (1972). 244. H. E. Evans and T. E. Evans, JBC 245, 6436 (1970).

546

STANLEY HATTMAN

ing (245) and aging (246), and that hydrocortisone induces reversible, hypermethylation in rat liver DNA (247). In 1977 the first experimental data appeared suggesting that loss of DNA methylation influenced gene expression (248). Friend erytholeukemia cells were induced to produce hemoglobin when grown in the presence of appropriate levels of ethionine, dimethyl sulfoxide, or butyrate. Each of these agents interfered with in vivo DNA methylation. Since this report, many workers have examined the question of gene activity and DNA methylation. As mentioned above (Section IV,B), methylation-sensitive, site-specific nucleases have been utilized to construct DNA methylation profiles, both in specific gene regions and in bulk DNA. These investigations appear to show that transcriptionally active (or previously active) genes are frequently unmethylated at specific sites that are methylated in cells not transcribing those genes. This has been observed for chicken P-globin (158, 159), chicken ovalbumin (161, 162), in virally transformed cell lines (/67-/70), and for the human y8p-globin locus (98, 164.1).Yet, there are also examples in which no clear correlation can be made between site-specific methylation and gene activity (14, 48, 158u, 164,249). That most DNA methylation is involved in gene regulation is difficult to imagine when one considers the range of MeCyt contents observed in higher organisms (0.17 to 50% of the cytosines). Whether site-specific methylation is important in regulatory control processes is an exciting issue that remains to be resolved. 4. Mutation

Deamination of MeCyt in DNA creates a T * G mismatched base pair. In the absence of repair, replication will produce one normal and one altered DNA molecule, the latter undergoing a G MeCyt + A * T transition. This process has been postulated to be important during sea urchin embryogenesis (207, 250). Indirect support for MeCyt deamination has been presented by Bird (251). Because CpG is frequently methylated in vertebrates, but not insects, Bird reasoned that CpG would occur in vertebrate

-

245. B. F. Vanyushin, N. A. Tushmalova, and L. V. Guskova, Dokl. Akud. Nuuk SSSR 219, 569 (Engl.transl.) (1974).

246. G . G . Zinkovskaya, G . D. Berdyshev, and B. F. Vanyushin, Biokhirniyu 43, 1484

(Engl.transl.) (1978). 247. B. F. Vanyushin and E. B. Romanenko, Biokhirniyu 46, 65 (Engl. transl.) (1979). 248. J. K . Christman, P. Price, L. Perdrinan, and G . Acs, EJB 31, 53 (1977). 249. A. Pellicer, D. Robins, D. Wold, R. Sweet, J. Jackson, 1. Lowy, J. M. Roberts, G. K. Sim, S. Silverstein and R. Axel, Science 209, 1414 (1980). 250. E. Scarano, M. Iaccarino, P. Grippo, and E. Parisi, PNAS 57, 1394 (1967). 251. A. P. Bird, Nucleic Acids Res. 8, 1499 (1980).

25. DNA METHYLATION

547

genomes at a frequency much lower than predicted from base composition. Moreover, the TpG plus CpA frequency would be concomitantly higher. In contrast, organisms that don’t methylate CpG would contain the predicted levels of these doublets. A survey of vertebrates, invertebrates, and arthropods confirmed these predictions. Since deamination of MeCyt G doublets lowers the CpG frequency and increases TpG plus CpA, these observations are consistent with such a process operating during vertebrate genome evolution.

VII.

Concluding Remarks

The recent applications of cloning, sequencing and restriction nuclease technologies have resulted in exciting advances of our knowledge about eukaryote DNA methylation. In some ways this progress has also been disappointing because of the attendant proliferation of new questions, while old questions remain just that. In 1975, three articles appeared that proposed roles for DNA methylation in gene function, differentiation, and chromosome recognition (252-254). An essential feature of each of these models was the notion that presence or absence of methyl groups in specific sequences affects protein-DNA interaction at those sites. We know that specific methylation is, in fact, a crucial factor in determining the nature of the interaction of a restriction nuclease with its recognition site; moreover, the E. coli lac repressor loses affinity for the lac operator when a specific thymine residue is replaced by uracil or cytosine (255). Thus, it is plausible that gene regulation in eukaryotes could be mediated by site-specific methylation. An additional feature in two of the models was the postulate that methylation sites are symmetrical sequences that are in only one of two states; namely, fully methylated or fully unmethylated. It was suggested that the resident DNA methylase acts mainly at hemi-methylated sites (generated during DNA replication), but not at fully unmethylated sequences. Consistent with this notion were the studies of Bird (156) who showed that Xenopus rDNA had only fully methylated or unmethylated sites, and that only progeny DNA strands were methylated in S-phase. However, all eukaryote enzymes examined so far are capable of methylating both heterologous (prokaryote) DNA and homologous DNA in v i m . This 252. 253. 254. 255.

A. D. Riggs, Cytogenet. Cell Genet. 14, 9 (1975). R . Holliday and J. E. Pugh, Science 187, 226 (1975). R. Sager and R. Kitchin, Science 189, 426 (1975). E. F. Fisher, and M. H. Carruthers, Nucleic Acids Res. 7, 401 (1979).

548

STANLEY HAlTMAN

would argue that these enzymes are capable of recognizing fully unmethylated sites. Unless there is a difference between in vivo and in vitro methylation specificity (sequencing data does not support this) or accessibility, this aspect of the above models has to be reconsidered. Another issue that requires reflection concerns the basis of tissue- and site-specific differences in methylation. A given restriction nuclease probes only a fraction of the MeCyt-containing sites. In addition, the complex structure of eukaryote chromatin itself may influence the DNA methylation pattern (130, 140, 142). For example, there is mounting evidence that there are specific nucleosome phasing patterns (253-260) in certain chromosomal regions. A recent report suggests that there is a difference in pattern for an inactive vs. active gene (260). However, it is clear that some nucleosome structure alteration is associated with gene activity (261-265). Thus, it is reasonable to wonder whether site-specific hypomethylation is only a reflection of phasing and nucleosome alteration. That is, a phasing pattern for a particular gene region could lead to site-specific hypomethylation if nucleosome placement sterically influenced methylation of potential sites. Then modification of nucleosome structure resulting from gene activity could change methylation at certain sites without necessarily appreciably affecting phasing. Such a model might well account for the observed differences in methylation profile for active vs. inactive genes. The implication of this model is that site-specific hypomethylation is merely a consequence of gene activity rather than a regulator. Whatever the truth may be, this is an exciting time for students of DNA modification. As new information continues to be gained, there is an increasing sense that solutions to these difficult questions are imminent.

256. 257. 258. 259. 260. 261. 262. 263. 264. 265.

B. A. Ponder and C. V. Crawford, Cell 11, 35 (1977). P. R. Musich, J. J. Maio, and F. L. Brown, J M E 117, 657 (1977). B. Wittig and S . Wittig, Cell 18, 1173 (1979). J. M. Gottesfeld and L. S. Bloomer, Cell 21, 751 (1980). W. Pfeiffer and H. G . Zachau, Nitcleic Acids Res. 8, 4621 (1980). C. Louis, P. Schedl, B. Samal, and A. Worcel, Cell 22, 387 (1980). B. Samal, A. Worcel, C. Louis, and P. Schedl, Cell, 23, 401 (1981). A . Garel and R. Axel, PNAS 73, 3966 (1976). H. Weintraub and M. Groudine, Science 193, 848 (1976). S. J. Flint and H. M. Weintraub, Cell 12, 783 (1977).

DNA Base-Insertion Enzymes (Insevtases) ZVI LIVNEH JOSEPH SPERLING

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . II.Assay., . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Purification . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Acceptor DNA Substrates . . . . . . . . . . . . . . . . . B. Purine Donor Substrates . . . . . . . . . . . . . . . . . . C.Specificity . . . . . . . . . . . . . . . . . . . . . . . . . D. Effect of Small Molecules and Various Treatments . . . . . . E. Depurinated DNA and Purine Base-Binding Activities . . . . . V. Mechanism of Insertion . . . . . . . . . . . . . . . . . . . . . VI. Biological Role . . . . . . . . . . . . . . . . . . . . . . . . .

I.

. .

549 551 554 555 555 556 558

. .

558 559 559 560

Introduction

DNA base-insertion enzymes (insertases) are defined as a class of enzymes that catalyze the reinsertion of a missing base into the appropriate apurinic or apyrimidinic sites (AP sites) in DNA. So far, only enzymatic activities that represent purine insertases have been described; however, the available data do not preclude the existence of pyrimidine insertases. The existence of insertase activities, and their possible role in DNA repair, was first implied in a study by Hennings and Michael (I), who 1 . H. Hennings and D. Michael, Cancer Res. 36, 2321 (1976).

549 THE ENZYMES, Val. XIV Copyright @ 1981 by Academic Press. lnc. All rights of reproduction in any form reserved ISBN 0-12-122714-6

550

ZVI LIVNEH AND JOSEPH SPERLING

reported that mouse skin cells exposed to low levels of N methyl-N’-nitro-N-nitrosoguanidineincorporate label from deoxyguanosine but not from thymidine. These results led them to suggest an apparent distinction between the observed guanine-specific repair “in which a single base apparently replaces a base lost by depurination” and the various forms of excision repair. The recent search for DNA baseinsertion enzymes has been motivated by the notion that they may offer an alternative, single-step pathway for the repair of AP sites in DNA. Furthermore, by operating in concert with DNA-N-glycosylases (2) they should be expected to constitute a base-replacement mechanism for the repair of DNA that contains modified or damaged bases. The base-deoxyribose bonds in DNA are susceptible to spontaneous cleavages that release free bases and lead to the formation of apurinic or apyrimidinic sites in the DNA. Spontaneous base release under physiological conditions, proceeds slowly but with a significant rate. The rate constant for spontaneous depurination was estimated to be close to 3 x lo-” sec-I (3),and that for spontaneous depyrimidination was found to be about 20 times slower (4). This means that the DNA of a growing mammalian cell loses about 10,000 purine and several hundred pyrimidines during each generation. The rate of depurination or depyrimidination is highly increased for some base derivatives that are formed in the DNA upon its exposure to radiation or chemical agents. Thus, N-3 or N-7 alkyl purines are easily released from the DNA, leaving apurinic sites ( 5 ) ; and irradiation of DNA with ultraviolet light (6, 7) or with ionizing radiation (7) is known to produce apyrimidinic sites in the DNA, presumably through intermediate 5,6-dihydropyrimidine derivatives since some of the latter have a highly labile glycosylic bond (8). In addition to their formation by spontaneous hydrolysis of glycosylic bonds, AP sites are also generated enzymatically by the action of DNA-N-glycosylases. These enzymes recognize unnatural bases (e.g., uracil, hypoxanthine) or modified bases (e.g., 3-methyladenine, 5,6-dihydrothymine) in the DNA and release them by cleavage of the glycosylic bond (2, 9). Since AP sites in cellular DNA interfere with correct readout of the 2. B. Duncan, this volume, Chapter 27. 3. T. Lindahl and B. Nyberg, Biochemistry 11, 3610 (1972). 4. T. Lindahl and 0. Karlostrom, Biochemistry I t , 5151 (1973). 5. P. D. Lawley and W. Warren, Chem. B i d . Interact. 12, 211 (1976). 6. T. N . Menshonkova, N. A. Simukova, E. I. Budowsky, and L. B. Rubin, FEES (Fed. Eur. Biochem. S o c . ) L e t t . 112, 299 (1980). 7. S. Ljungquist, A. Andersson, and T. Lindahl, JBC 249, 1536 (1974). 8. B. Dunlap and P. Cerutti, FEES (Fed. Eur. Biochem. Soc.) Lett. 51, 188 (1975). 9. T. Lindahl, Prog. Nucleic Acids Res. Mol. Eiol. 22, 135 (1979).

26. DNA BASE-INSERTION ENZYMES

55 1

genetic information, the ability of cells to repair this type of damage in their DNA seems to be a vital need for their survival, and indeed most cells were found to have the ability to repair AP sites efficiently (9). The prevailing mechanism for the repair of this type of lesion has been shown to involve nucleotide excision repair. It is initiated by an endonuclease specific for AP sites (AP endonuclease), which incises the DNA at the AP site. Following this incision the terminal deoxyribose together with some of its neighboring nucleotides are excised by an exonuclease, the missing sequence is repolymerized by a DNA polymerase, and finally a ligase covalently seals the nick, leading to restoration of DNA integrity (9). This mechanism has been verified by the isolation of AP endonucleases from a variety of sources (/0-16), and by the demonstration of the total excision repair of AP sites in v i m using bacterial (17) and human enzymes (18) (see also Chapter 14 of this volume). This chapter summarizes experiments that led to the discovery of purine insertase enzymatic activities in cultured human fibroblasts (19) and in E. cofi (20). These activities insert missing purines directly and specifically into the appropriate apurinic sites, with no apparent need for rupture of the phosphodiester bond at the site or excision of undamaged nucleotides. The possible action of purine insertases in vivo offers an alternative pathway for the repair of apurinic sites in a single enzymatic step. II.

Assay

To assay for purine insertase activity one usually measures the incorporation of radioactivity from a monomeric-labeled purine derivative (the donor substrate) into depurinated DNA (the acceptor substrate). Incorporation into the DNA can be monitored in three ways (20): (1) Acid precipitation-typically the DNA is precipitated on glass filters with 5% trichloroacetic acid and the excess of donor substrate is washed off. (2) 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.

S. Ljungquist and T. Lindahl, JBC 249, 1530 (1974). W. G . Verly and E. Rassart,JBC 250, 8214 (1975). G . W. Teebor and N . J. Duker, Nature (London) 258, 544 (1975). W. S. Linsley, E. E. Penhoet, and S. Linn,JBC 252, 1235 (1977). J. P. Kuebler and D. A. Goldthwait, Biochemistry 16, 1370 (1977). S. Ljungquist, JBC 252, 2808 (1977). J. Laval, Nature (London) 269, 829 (1977). W. G . Verly, F. Gossard, and P. Crine, PNAS 71, 2273 (1974). K . Bose, P. Karran, and B. Strauss, PNAS 75, 794 (1978). W. A. Deutsch and S. Linn, PNAS 76, 141 (1979). 2. Livneh, D. Elad, and J . Sperling, PNAS 76, 1089 (1979).

552

ZVI LIVNEH AND JOSEPH SPERLING

Adsorbtion to an ion exchanger-the DNA is most conveniently bound to a square of DEAE-cellulose filter paper and is washed free of the donor substrate. (3) Gel electrophoresis-following incubation and phenol extraction the DNA is electrophoresed on an agarose gel; the gel is washed extensively to remove unbound donor substrate, and after staining the DNA band is cut out, solubilized, and counted (Fig. 1). The acid precipitation procedure is preferred because it is fast and has a low background. The acceptor substrate is usually prepared by heating DNA in a mild acidic buffer (3,19, 20). Double-stranded DNA from any source can be used as a substrate. The gel assay, however, requires DNA preparations of homogeneous size. Free purine bases, purine deoxyribonucleosides, or purine deoxyribonucleoside triphosphates (dNTP) can serve as donorsdepending on the source of partially purified purine insertases, or of the crude protein extracts that contain the insertase activity. Reactions with the free bases or the deoxyribonucleosides normally have a low background, caused mainly by nonspecific binding of the donor to the DNA. However, with dATP and dGTP as base donors for assaying crude extracts, a mixture of both 3H-base-labeleddNTP and [a-3*P]dNTPshould be used to provide an appropriate blank for background incorporation by polymerization (20). It should be noted that some bacterial crude extracts

FIG.

1 . Gel analysis of PM2 DNA after insertion incubation with extracts from E. coli

H502 and with IT-labeled dATP and dGTP as the purine donors. Upper panel: Ethidium bromide-stained gel; lower panel: Radioactivity distribution almng the gel. Gel a (O), PM 2 DNA (0.4 pg) with 10 apurinic sites per molecule; gel b ( O ) ,intact PM2 DNA (0.4 pg); gel c (A), as in gel a, without extract; gel d, (A)as in gel b, without extract. From Livneh et a / . 20, reproduced with permission.

-

553

26. DNA BASE-INSERTION ENZYMES - P -dRib-

P

-

T A 1 - P -dRib

NICK TRANSLATION

- P 1dRib'p

Poll

-P-

dATP.dGTP,dCTP dUTP'

r----q

I

P-

"

L-K;

?

-P-dRibPt-D,NA URACl L-GLVCOSYLASE c-P-[ dRib'i-P

-

II

L__-__,

-P-

'!

-

? - P-

dRib

LABELED AP SITE

BASE INSERTION REPAIR

/ \ - {i-ik-k -

AP ENDONUCLEASE EXCISION REPAIR p

I

A

I

-P-dRib-PTOTAL DIGESTION TO NUCLEOSIDES dAdo,OGuo,dCyd dThd*[=dRib'-Thy)

1

LABELED dThd

FIG. 2.

p

I ; ;

.___

I

--.d

!

A

I

- -

P TOTAL DIGESTION TO NUCLEOSIDES dAdo,dGuo.dCyd dThd -P-

dRib

1

UNLABELED NUCLEOSIDES

AP site-labeling method for the assay of possible pyrimidine insertase ac-

tivities.

exhibit a high level of polymerization, which does not allow measurement of insertion incorporation with the dNTPs as donors. For some other bacterial strains active extracts can be prepared only if cultures are grown in the presence of sublethal doses of alkylating agents (21). An alternative approach to the assay of insertases over a high background of polymerization is the use of DNA substrates with radioactively labeled AP sites (21). The preparation of such DNA substrate for a pyrimidine insertase and an outline of the assay procedure are summarized in Fig. 2. The DNA is nick translated in the presence of [''Cl- or [3H]deoxyuridine triphosphate labeled uniformly in both base and sugar. Treatment of the DNA with uracil-DNA-N-glycosylase (9) releases uracil residues leaving deoxyribose-labeled AP sites. Following insertion incubation the DNA is hydrolyzed to nucleosides, which are analyzed chromatographically. Since the donor is unlabeled TTP, excision and polymerization should result in unlabeled thymidine in the total hydrolysate. Only insertion of thymine into the labeled AP site should yield labeled thymidine upon hydrolysis. In an analogous way, guanine apurinic 21. Z. Livneh and J. Sperling, in "Chromosome Damage and Repair" (E. Seeberg, ed.), in press, 1981.

554

ZVI LIVNEH AND JOSEPH SPERLING

sites have been introduced into DNA by using uniformly labeled dITP in the nick-translation reaction and hypoxanthine-DNA-N-glycosylase for the removal of hypoxanthine residues. Insertion of guanine from unlabeled dGTP yields labeled deoxyguanosine after total enzymatic hydrolysis of the DNA. Insertion could not be assayed in crude extracts of cultured human fibroblasts or HeLa cells (19), most likely because of interference from AP endonucleases. As has been shown (22), DNA incised at AP sites cannot serve as an acceptor substrate for purine insertase. In such cases, where the insertase activity copurifies with AP endonuclease activities, the insertases are assayed as depurinated DNA-binding proteins on nitrocellulose filters (22). Glycerol gradient sedimentation separates the insertase activity from AP endonucleases and this permits a direct insertion assay (221. 111.

Purification

Purine insertase has not yet been purified to homogeneity, and only partial purification of the enzyme from cultured human fibroblasts has been achieved (22). The purification procedure is complicated by the fact that purine insertase copurifies with AP endonucleases and can not be assayed in the presence of these enzymes because incised AP sites cannot serve as acceptor substrates for the insertase. Since no specific inhibitor of A P endonuclease is known, the purine insertase was originally assayed indirectly by utilizing its depurinated-DNA binding activity. As a matter of fact, the fibroblast purine insertase was first discovered as a protein that bound to depurinated DNA and did not show any AP endonucleolytic activity (19). The purification scheme (22) consisted of sonic disruption of the fibroblast cells, followed by DEAE-cellulose chromatography and fractionation on a phosphocellulose column using a gradient of 0.1-0.5 M potassium phosphate as the eluent. A small fraction of the insertase and AP endonucleases coeluted with the flowthrough, and a broad peak of these activities coeluted between 0.24 and 0.27 M potassium phosphate. The purine insertase and the AP endonuclease activities were resolved by sedimentation through a gradient of 20-40% glycerol, and the partially purified insertase fraction was used for further characterization and studies of the activity. The protein has a sedimentation constant of 6.4 S, and a molecular weight of about 120,000 (19). This purification procedure (Table I) resulted in a 22. W.A. Deutsch and S. Linn, JBC 254, 12099 (1979).

555

26. DNA BASE-INSERTION ENZYMES TABLE I PURIFICATION SCHEME FOR PURINE INSERTASE FROM

HUMANFIBROBLASTS Total protein Fraction I I1 Ill IV

Crude extract DEAE-cellulose Phosphocellulose Glycerol gradient

(me)

Specific activity'' (unitsfmg)

18.5 13.5 0. I5 0.006

920 1,400 4,600 11,000

" Activity as a depurinated-DNA binding protein. Data from Ref. 22.

12-fold purification of the insertase, and the yield of recovered activity was less than 1%. This low recovery is probably due to the instability of the enzyme after the cells are opened. It is clear that optimal conditions that stabilize the enzyme in crude extracts and in a variety of fractionation steps should be looked for before further purification procedures are attempted. The E. coli insertase has not yet been purified. The fractionation of the activity on a phosphocellulose column can be monitored by the direct assay of the insertion activity, but the partially purified fractions exhibit extremely high lability and loose activity within a day or two.

IV.

Properties

A.

ACCEPTOR DNA SUBSTRATES

Purine insertase catalyzes the insertion of adenine and guanine into depurinated DNA. A summary of the properties of this reaction is presented in Table 11. The only requirements for acceptor DNA substrates seem to be a double-stranded structure and the presence of intact apurinic sites. The following depurinated DNAs acted equally well as acceptor substrates (19-22): Phages PM2, PBS2, T7 and 4x174 (RFI and RFII DNAs); E. coli DNA; poly(dA-dT) and poly(dG-dC) (tested for the fibroblast activity); and calf thymus DNA (tested for the bacterial activity). Depurinated single-stranded DNA from phage 4x174 was inactive, as were untreated double-stranded DNAs or depyrimidinated PBS2-DNA. An important characteristic of the human insertase is its inability to act at incised apurinic sites (Table 111). Furthermore, it seems as if nicked apurinic sites inhibit the enzyme from acting at other sites (22).

556

ZVI LIVNEH AND JOSEPH SPERLING TABLE I1 PROPERTIES OF PURINE INSERTASES

Properties ~

Human purine insertase (19, 22)

E. coli purine insertase (20, 21)

~

DNA substrates Depurinated linear dsDNA Depurinated open circular dsDNA Depurinated superhelical dsDNA Depurinated ssDNA Depurinated dsDNA nicked at AP sites Depyrimidinated dsDNA Untreated dsDNA Purine donor substrate Purine bases Purine deoxyribonucleosides Purine ribonucleosides dATP and dGTP Metal ion requirements Inhibition of activity by EDTA (1 mM) Caffeine (1 m M ) ATP (0.4 m M ) 2-Mercaptoethanol (1 m M ) Inactivation by Heat treatment Dialysis Freezing Mass (kilodaltons)

+

+ + n.d." n.d. -

+ n.d. -

+* Mg2'

+ +

+ +

n.d. -

f

-

+

+" +

120

n.d.

+c

+

-

" n.d., not determined. 'I

In strain E. coli HS02. loo" for 5 min 60" for 15 min.

B. PURINE DONORSUBSTRATES Both the free purine bases and the purine deoxyribonucleosides are substrates for the human purine insertase activity, while dATP and dGTP seem to be inactive (19, 22). Evidently, pyrimidines and their derivatives are not inserted into the apurinic sites in the DNA, and purine ribonucleosides or ribonucleotides fail to serve as substrates (Table IV). Purine base insertion with the human enzyme follows Michaelis-Menten kinetics with regard to the purine bases, with an apparent K , of 5 p M for both guanine and adenine. The V,,, observed for guanine was found to be 7 times higher than that for adenine. However, these data may be in error

557

26. DNA BASE-INSERTION ENZYMES TABLE 111

INSERTION OF [3H]GUANINE INTO DEPURINATED PM2 DNA NICKEDBY AP ENDONUCLEASE' Nicks introduced (fmol)

AP DNA sites (% nicked)

Guanine incorporation fmol (%)

92 184 460 920

8 16 44 88

218 (100) 199 (91) 143 (66) 16 (7) <16 (<7)

' I Reaction mixtures contained 9 nrnol of PM2 DNA containing 1044 fmol of AP DNA sites. Nicks were introduced at AP DNA sites with a human AP endonuclease preparation. Data from Ref. 2 2 .

due to lack of knowledge of the exact number of guanine versus adenine apurinic sites in the substrate DNA (22). The situation seems to be more complex in the case of the E. coli activity, since different strains exhibit different substrate specificities. In extracts prepared from E. coli H502 (endol-, thy-, uvrA, su-), dATP and dGTP served as the purine donors (20) and for the incorporation of the TABLE IV

INSERTION OF LABELED P U R I N E DERIVATIVES INTO AP DNA S I T E S BY T H E HUMAN PURINE INSERTASE'

Experiment

Purine derivative

Base incorporation (fmol)

1

Gua Ade Thy dGuo dAdo dThd Gua Ade dGTP dATP dCTP

638 92 <2 121 23 <4 1100 400 <20 <20 <20

2

" Reaction mixtures contained 8.8 nmol of E. coli DNA with 2040 frnol of apurinic sites for experiment 1 , and IS nmol of T7 DNA with 3480 fmol of apurinic sites for experiment 2. Data from Refs. 19 and 22, respectively.

558

ZVI LIVNEH AND JOSEPH SPERLING

base residues, but the insertion of neither the deoxyriboses nor the a-phosphates into the depurinated DNA has been demonstrated. This incorporation could be faithfully monitored due to a very low background of polymerization in the crude extracts of that particular strain. In two other strains that were studied-E. coli KMBL 1719 ( F - , bio-, endoA 101, thyASUi, uvrE50.2) and E. coli KMBL 1720 (same as KMBL 1719 but uvrE+)-a high background of polymerization prevented accurate measurement of the insertion of purines when using dATP and dGTP as substrates. Nevertheless, efficient incorporation of free adenine (but not guanine) into depurinated DNA was catalyzed by these extracts (Fig. 3 and Ref. 21). The reasons for heterogeneity in substrate utilization are unclear. It may be caused by variations in the internal pools of base derivatives in the different strains or it may reflect the presence of different agents in the various extracts, which modify differentially the ability of the purine insertase to utilize the purine donor substrates. C. SPECIFICITY Purine base insertion is a specific process. No insertion of pyrimidine derivatives could be observed either with bacterial or human enzymes. In addition, it was found that adenine derivatives cannot substitute for guanine derivatives and vice versa. Thus, the human insertase incorporated guanine but not adenine into depurinated poly(dG-dC), and adenine but not guanine into depurinated poly(dA-dT) (19). This property was demonstrated also in competition experiments performed with the E. cofi activity (20). It was shown that a 100-fold excess of unlabeled dGTP did not alter the incorporation of labeled dATP into depurinated DNA and vice versa. D. EFFECTOF SMALL MOLECULES AND VARIOUS TREATMENTS The human purine insertase requires K+ ions, and these cannot be substituted for by NH4+,Na+, Mgz+or Mn'+ ions. Only Li+ supports the activity to a small degree (22). In contrast, theE. coli activity requires Mg2+ ions and is totally inactive in their absence (20, 21). Both activities are inhibited by EDTA and by caffeine, and are insensitive to phenylmethanesulfonyl fluoride and to 2-mercaptoethanol. The bacterial enzyme is inhibited by ATP. Heat treatment (60" for 10 min) or dialysis cause total inactivation of both purine insertases, while freezing inactivates the human activity but not the E. coli activity (Table 11).

26.

559

DNA BASE-INSERTION ENZYMES

30

60

90

120

pg protein/ reactim

10 20 30 Adenine (pM)

FIG.3 . Incorporation of [2-3H]adenine into depurinated DNA by an extract from E. coli KMBL 1720. 0, Incorporation into 1.5 nmol depurinated PM2 DNA with 0.8 pmol apurinic sites; 0 , incorporation into intact PM2 DNA. Data from Ref. 2 1 .

E. DEPURINATED DNA ACTIVITIES

AND PURINE

BASE-BINDING

In contrast to the insertion reaction, the binding of human insertase to depurinated DNA does not require Kf ions (22). Addition of adenine or guanine weakens the binding of the insertase to the DNA, probably because of their insertion into the apurinic sites, which leads to a decrease in the concentration of available apurinic sites and to dissociation of the enzyme. The protein binds adenine and guanine with a remarkable difference in the affinity of the protein towards the two bases. Adenine saturates at less than 1 p M ; whereas guanine does not saturate, although the curve breaks between 3 and 5 p M . Although purine binding and purine insertion activities decay concomitantly, it is not clear yet whether they reside on different polypeptides or on the same protein. V.

Mechanism of insertion

The specificity of purine insertion, and the absolute requirement for double-stranded DNA substrates suggest that base pairing of the inserted purine with the pyrimidine base located on the intact strand opposite to the apurinic site is an intermediate step in the insertion process. Further data that will allow unambiguous characterization of other mechanistic features of the insertion reaction are not available. An intriguing question is that of the energy source for the insertion reaction. Although the free energy of formation of glycosylic bonds within a DNA molecule has not been measured directly, it is conceivable that such a reaction would require energy. However, at least the human en-

560

ZVI LIVNEH AND JOSEPH SPERLING

zyme does not seem to need an energy-rich cofactor. The re-formation of the base-pairing and base-stacking might be a source of free energy, although this is not likely to be as rich as triphosphate hydrolysis. The use of dATP and dGTP by the E. coli activity may explain the source of energy, but this explanation becomes doubtful with the discovery that free adenine can be inserted into depurinated DNA in other E. coli extracts, and that ATP (added as a possible energy source) totally inhibits this insertion. Despite the fact that there seems to be no cofactor requirement, it is still possible that the insertase preparations are contaminated with an unknown energy-providing factor. A pure enzyme preparation will help to clarify this point. Enzymes that catalyze reactions analogous to insertions are known to be active in salvage pathways of nucleotide biosynthesis. For example, hypoxanthine-guanine phosphoribosyltransferase converts a purine base into a purine nucleotide using 5-phosphoribosyl 1-pyrophosphate ( 2 3 , and nucleoside phosphorylases catalyze the equilibrium in the interconversion of bases and nucleosides with intermediate formation of ribose 1-phosphate, or deoxyribose 1-phosphate (24). These reactions, however, involve phosphorylation and hydrolysis of phosphate groups and, most probably, proceed by a mechanism different from the insertions. Another similar reaction is performed by the tRNA guanylylation enzyme, which replaces the Q base of certain tRNAs with guanine (25, 26). Like the purine insertases, this enzyme acts in the absence of any apparent added energy cofactor. However, one should bear in mind that the guanylylation is a replacement reaction that involves both cleavage and re-formation of a glycosylic bond, while the purine insertion involves only the formation of a glycosylic bond. Further, the guanylylation occurs at the anticodon region of the tRNA and does not seem to involve base pairing to direct the reaction. The insertion, however, requires an intact strand opposite the AP site. It seems likely, therefore, that the guanylylation enzyme is specific for the replacement of Q bases and, from the mechanistic point of view, bears little relevance to the insertion reactions. VI.

Biological Role

The existence and the properties of purine insertase activities strongly argue for their involvement in the repair of apurinic sites in vivo . Repair of 23. 24. 25. 26.

C. T. Caskey and G . D. Kruh, Celt 16, 1 (1979). E. W. Yamada, "Methods in Enzymology," Vol. 51, p. 423, 1978. N . Okada, F. Harada, and S. Nishimura, Nucleic Acids Res. 3, 2593 (1976). W. R. Farkas and R. D. Singh, JBC 248, 7781 (1973).

26. DNA BASE-INSERTION ENZYMES

56 1

apurinic sites by purine insertase offers a simple, direct and specific repair pathway, with no need for breakage of the phosphate-sugar backbone or for excision of undamaged nucleotides, as is needed in the multistep excision repair pathway. There is no direct evidence for a functional role of purine insertases in the in vivo repair of apurinic sites in the bacterial or the mammalian cells studied. However, two observations have been made in the eukaryotic systems that suggest the existence of an alternative pathway for the repair of apurinic sites, and are compatible with the notion that purine insertases provide this function. First, Hennings and Michael (I) found that mouse skin cells exposed to low levels of N-methyl-N’-nitro-N-nitrosoguanidine incorporate label from deoxyguanosine but not from thymidine. These findings led them to suggest the existence of a guanine-specific repair in which a single base replaces a base lost by depurination. Second, in their studies of the host cell reactivation of heat and acid depurinated or methyl methanesulfonate-treated transfecting SV40 viral DNA, Kudrnaet al. (27) found that the xeroderma pigmentosum D cell lines, that lack a major AP endonuclease, reactivate the depurinated DNA more effectively than normal fibroblasts. This observation also suggests the existence of an alternative pathway for the repair of apurinic sites that is independent of AP endonuclease. The role of purine insertases in the repair of apurinic sites in bacteriain vivo is unclear. Although several E. coli mutants deficient in the major AP endonuclease have been isolated [xth- mutants; see Refs. (28, 2 9 ) ] , they contain another AP endonuclease activity-endonuclease I V (15). Clearly, the isolation of mutants totally deprived of AP endonuclease activity may help in the evaluation of the role of purine insertases in vivo. Evidence that insertase activity may be inducible is provided by the observation (21) that E. coli strain AB3027 (xrh-, pol I-), which yields protein extracts with no measurable purine insertase activity, yields active extracts if cultures are grown in the presence of sublethal doses of alkylating agents, such as methyl methanesulfonate or N-methylN-nitro-N’-nitrosoguanidine (Table V). Other strains that behave similarly include E. coli AB1157 and E. coli NH5016 (xth-). It is possible that alkylating agents act indirectly by neutralizing factors that do not allow the assay of the purine insertase in crude extracts. A more attractive interpretation of these results is that the purine insertase is an induced function, or that its expression is amplified as a response to the exposure 27. R. D. Kudrna, J. Smith, S. Linn, and E. E. Penhoet, Mu?. Res. 62, 173 (1979). 28. D. M. Yajko and B. Weiss, PNAS 72, 688 (1975). 29. S. Ljungquist, T. Lindahl, and P. Howard-Flanders, J . Bacreriol. 114, 1151 (1976).

562

ZVI LIVNEH AND JOSEPH SPERLING

TABLE V INCORPORATION OF [3H]ADENINE INTO DEPURINATED DNA WITH EXTRACTS FROM E. coli AB3027 ( x t h -

Assay conditions Extracts prepared from bacteria grown in a normal medium MgCI2 (5 mM) Extracts prepared from bacteria grown in the presence of methyl methanesulfonate + MgCI2 (5 mM) -MCI2 + MgCI2 ( 5 mM), + ATP (0.4 mM)

*

Incorporation of [RH]Ade bmol (cpm)l

0.01 (200)

0.17 (3400) 0.01 (200) 0.02 (500)

" Reaction mixtures contained 1.2 nmol depurinated PM2 DNA with 0.6 pmol apurinic sites. Data from Ref. 2 1 .

to alkylating agents. In this context it should be interesting to look for any possible relationships between purine insertase activities and the adaptive response of E. coli, which is manifested as a higher survival and a lower mutability of cultures in response to exposure to alkylating agents, and is acquired upon pretreatment with low doses of alkylating agents (30, 31). The scheme for the repair of bacterial DNA damaged by UV or alkylating agents that emerges from later studies assigns a central role to DNA-N-glycosylases ( 2 , 3 2 , 3 3 ) .It seems that the specificity of base excision repair resides initially with damage-specific DNA-N-glycosylases, while subsequent incision steps require AP endonuclease specificity. DNA base insertion enzymes may provide an alternative route whereby AP sites are correctly filled with the missing bases in a single step. The possible collaborative action of DNA-N-glycosylases and DNA-insertases in vivo should thus constitute a specific base-replacement repair pathway. The presumed generality and specificity of base replacement repair mechanisms call for the existence of pyrimidine insertases along with the purine insertases described here. The former, however, have not been discovered. Whether this is due to the fact that depyrimidination events in vivo are more rare than depurinations remains to be tested. In conclusion, it should be pointed out that in addition to their function in the repair of AP sites, insertases may be needed in other processes 30. 31. 32. 33.

L. Samson and J. Canris, Nuture (London) 267, 281 (1977). P. Jeggo, M. Defais, L. Samson, and P. Schendel, Mol. Gen. Genet. 157, I (1977). L. Grossman,S. Riazuddin, W. Haseltine, and C. Lindan, CSHSQB 43, 947 (1978). B. Demple and S. Linn, Nuture (London) 287, 203 (1980).

26. DNA BASE-INSERTION ENZYMES

563

where formation of DNA sequence diversity is required, or in regulatory mechanisms, where replacement of naturally occurring modified bases by unmodified bases may trigger developmental or regulatory mechanisms. ACKNOWLEDGMENTS The work in the authors' laboratory was supported by grants from the United StatesIsrael Binational Science Foundation (BSF), Jerusalem, Israel, and from the Ber-Lehmsdorf Foundation of the Israel Cancer Association. We are grateful to Dr. G. R. Stark for suggestions concerning the manuscript.

This Page Intentionally Left Blank

DNA Glycosylases BRUCE K . DUNCAN

I. Introduction . . . . . . . . . . . . . . . . . . . .

.

. . . . .

11. Physical Properties . , . . . . . . , , . . . . . . . . . . . . . 111. Enzymatic Properties . . . . . . . . . . . . . . . . . . . . . .

A. Methods of Assay . . . . . . . . . . B. Enzyme Purification . . . . . . . . . C. Reaction Mechanism . . . . . . . . D. Substrate Specificities . . . . . . . . E. Glycosylase Inhibitors . . . . . . . . IV. Physiological Role . . . . . . . . . . . A. DNA Glycosylases as Repair Enzymes B. Base Excision Repair . . . . . . . . V. Research Applications . . . . . . . . . Note Added in Proof . . . . . . . . . .

1.

.

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

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

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

. . . .

. . . . . . . .

. . . .

565 568 569 569 572 573 575 577 578 578 582 584 586

Introduction

DNA glycosylases hydrolyze nucleoside base-glycosidic bonds (I ); the reaction results in the production of an apurinic/apyrimidinic (AP) site in DNA and a free nucleoside base. These enzymes participate in DNA repair processes by catalyzing the removal of unconventional or damaged bases from DNA. The structures of some of the bases released by the various glycosylases are shown in Fig. 1. Prior to the discovery of DNA glycosylases in 1974, most of the enzymes responsible for initiating DNA repair processes were widely be1 . T. Lindahl, P N A S 71, 3649 (1974). 565 THE ENZYMES,Vol. XIV Copyright 0 1981 by Academic Press, Inc. All rights of reproduction in any form reserved

ISBN 0-12-122714-6

566

BRUCE K. DUNCAN

Uracil

(I)

Hypoxonthine

(11)

3-methylodenine

(rn)

0

dRib

2,4-diamino-6hydroxy- 5 - Nmethyl formamidopyrimidine

5 , 6 - dihydro-5,6 dihydraxythymine

Pyrimidine dimer

b x1

(P)

(IE)

FIG. 1. Structure of nucleoside bases excised by DNA glycosylases. Substituted pyrimidines are derived from purines. The foramidopyrimidine (IV) is derived from 7-methylguanine. Pyrimidine dimers (VI) have only one of the two glycosidic bonds hydrolyzed. lieved to be nucleases (2). Subsequent work on several of these enzymes, now known to be glycosylases, revealed that DNA-repair nucleases were often impure preparations consisting of DNA glycosylase and AP endonuclease activities. DNA glycosylases are ubiquitous. One enzyme, uracil-DNA glycosylase, has been identified in every tissue that has been examined [Table I; see also (.5-/6)], with the exception of Drosophila (3). Human cells, and probably other eukaryotic cells, have both a nuclear and a mitochondria1 uracil glycosylase [ ( 4 ) , confirmed by M. Sirover, personal communication]. One organism, Escherichia coli, has been shown to have 2. T. Lindahl, Nature (London) 259, 64 (1976). 3. E. C. Friedberg, T. Bonura, R. Cone, R. Simmons, and C. Anderson, in “DNA Repair Mechanisms” (P. C. Hanawalt, E. C. Friedberg, and C. F. Fox, eds.), p. 163. Academic Press, New York, 1978. 4. C. T. M. Anderson and E. C. Friedberg, Nucleic Acids Res. 8, 875-888 (1980). 5. T. Lindahl, S. Ljungquist, W. Siegert, B. Nyberg, and B. Sperens,JBC 252,3286-3294 (1977).

567

27. DNA GLYCOSYLASES TABLE I

OCCURRENCE AND

VARIETY OF DNA

Enzyme Class I enzymes Uracil glycosylase' E. coli E . subtilis Human blast cells Calf thymocytes Hypoxanthine glycosylase E. coli Human fibroblasts Calf thymocytes 3-Methyladenine glycosylase E. cnli Human Formamidopyrimidine glycosylase E. coli Class I1 enzymes (with associated AP endonuclease) Thymine hydrate glycosylase" Pyrimidine dimer glycosylase" T4 endonuclease V M . luteus "

1 unit

=

GLYCOSYLASES

Activity in crude extracts (pU/mg protein)"

Structure of released product*

Reference

3500 11

0.0007 0.0002 0.12 0. I 0.2

3.3 NC 0.01

NC NC

1 pmol base release min-'. NC = no comparable data.

* The structures of the released nucleoside bases are shown in Fig.

1.

' Uracil glycosylase has been found in several bacterial species, corn, and several different mammalian tissues, but not in Drosophiln, embryos, or cell cultures. Same enzyme as endonuclease 111 in E. eel;. ' Missing in E. coli, this enzyme activity is induced by phage T4 infection (14, / 5 ) .

"

6. R. Cone, J. Duncan, L. Hamilton, and E. C. Friedberg,Biochemistr.~16, 3194 (1977). 7. S. J. Caradonna and Y. C. Cheng, JBC 255, 2293 (1980). 8. M.Talpaert-Borle, L. Clerici, and F. Campagnari, JBC 254, 6387 (1979). 9. P. Karran and T. Lindahl, JBC 253, 5877 (1978). 10. P. Karran and T. Lindahl, Biochemistry 19, 6005 (1980). 11. S. Riazuddin and T. Lindahl, Biochemistrv 17, 21 10 (1978). 12. Thomas P. Brent, Biochemistry 18, 911 (1979). 13. C. F. Chetsanga and T. Lindahl, Niicleic Acids Res. 6, 3673 (1979). 14. B. Demple and S. Linn, Nature (Lotidon) 287, 203 (1980). 15. E. H. Radany and E. C. Friedberg, Nrrtrire (Lordon)286, 182 (1980). 16. W. A. Haseltine, L. K. Gordon, C. P. Lindan, R. H. Grafstrom, N. L. Shaper, and L . Grossman, Nature (London) 285, 634 (1980).

568

BRUCE K.DUNCAN

all of the known DNA glycosylases although one of them, the thyminedimer glycosylase, is made only in T4 phage-infected cells (2, 9, 13-15). Six different DNA glycosylase activities are known and are divided into two classes (see Table I and Fig. 1). Class I enzymes are specific for uracil (I), hypoxanthine (111, 3-methyladenine (III), and formamidopyrimidine (IV) residues in DNA. They have been purified free of contaminating nuclease activities. Class I1 enzymes appear to have an associated AP endonuclease activity that is an integral part of the enzyme. DNA molecules that contain 5,6-dihydro-5,6-hydroxythymine(V) and pyrimidine dimers (VI) are degraded by class I1 enzymes. Probably there are other DNA glycosylases yet to be discovered; these enzymes will presumably be specific for other damaged nucleoside bases. Lindahl has provided an illuminating discussion of the chemical considerations that eventually led to the search for, and discovery of, formamidopyrimidine-DNA glycosylase (13). This chapter deals primarily with the class I glycosylases. Chapter 14 in this volume by E. C. Friedberg et a / . contains additional information on the class I1 enzymes. Lindahl has also published a comprehensive review of base excision repair (17).

II. Physical Properties

The DNA glycosylases are small, single-chain proteins with molecular weights ranging from 20,000 to 35,000 (Table 11). None of the enzymes that have been characterized have given any indication of a multisubunit structure. Although the molecular weights of the various E. coli glycosylases are quite similar, biochemical and genetic studies clearly show these enzymes, which act on different substrates, to be different proteins (9, 11, 13). The only enzyme whose physical properties have been characterized beyond molecular weight and substrate specificity is E. coli uracil glycosylase (5). Its ultraviolet spectrum is that of a typical protein. No metal ion analyses have been performed. An amino acid analysis reveals a single cysteine residue per 24,600 dalton enzyme molecule. The cysteine does not appear to be involved in catalysis, since exposure to 5 mM N-ethylmaleimide or iodoacetate does not inhibit enzyme activity. The turnover number is about 13 sec-' under the standard assay conditions. The homogeneous enzyme is about 11,000-fold purified, indicating there are about 300 to 400 molecules per cell. 17. T. Lindahl, Progr. Nircleic Acid Res. Mol. Biol. 22, 135 (1979).

27.

569

DNA GLYCOSYLASES TABLE I1

PROPERTIES OF DNA GLYCOSYLASES Glycosylase

Molecular weight

Class I enzymes Uracil E . coli B. subrilis

Lymphoblast H ypoxanthine E. coli Thymocyte 3-Methyladenine E. coli Lymphoblast Formamidopyrimidine Class I1 enzymes Thymine hydrate E. coli Pyrimidine dimer T4 endoV M. luteus

24,600 24,000 30,000

K mu

4 x lo-" 1 x 10-9 3 x 10-7

ND'

30,000 3 1,000

5-9 x 10-7

19,000 34,000 30,000

6 x lo-' ND 2 x 10-9

Preferred substrateb

Product inhibitors

ss > ds ss > ds ds

Uracil Uracil Uracil

ss < ds ss < ds

ND none

ds ds ds

3-Methyladenine 3-Methyladenine ND

24,000

ND

ds

ND

18,000 ND

ND ND

ss,ds ds

ND ND

" Units are M of modified base residue. ss, single strand or denatured DNA; ds, double strand or native DNA. ND, not determined.

111.

Enzymatic Properties

A.

METHODS OF ASSAY

DNA glycosylases were recognized as a new class of enzymes when crude extracts of E. coli and B. subtilis were found to release free uracil from uracil-containing DNA ( I ) . Product analysis showed that the other reaction product was an AP site in the DNA; no phosphodiester bonds were broken. Other glycosylases were subsequently discovered. Sensitive assay procedures have been devised for detecting DNA glycosylases in crude extracts. Based on what is known about DNA glycosylases, and if the appropriate substrate is available, it is reasonable to consider assaying for previously unrecognized glycosylases. 1. Substrate Preparation

The choice and source of the substrate is determined by the enzyme activity to be measured. A glycosylase substrate can be made by several

570

BRUCE K . DUNCAN

methods. The most generally useful method is to treat radioactive DNA (labeled in the appropriate nucleoside base) with various chemical (alkali, alkylating agents, Os04)or physical (heat, radiation) agents to produce the desired DNA damage. It is useful to isolate plasmid or phage DNA from nutritional mutants because the DNA is easy to isolate and any of the four nucleoside bases in the DNA can be labeled. Thymine auxotrophs (thyA mutants) can be used to label thymine or cytosine residues by adding labeled thymine or uracil to the growth medium (18). Similarly, guanine auxotrophs (gua mutants) can be used to preferentially label adenine or guanine. Adding labeled thymidine to the growth medium ofthy+ cells can also be used to make thymine-labeled DNA (19). Substrates for uracil, thymine hydrate, and pyrimidine dimer glycosylases have been made in this way (5, 14-16). Radioactive DNA may also be synthesized by polymerizing or copolymerizing an unusual nucleotide (or its precursor) into DNA with terminal transferase or DNA polymerase (I, 6, 9 , 2 0 ) . The suitability of this approach depends upon the ability of the polymerase to incorporate unusual nucleotides into DNA. Uracil and hypoxanthine glycosylase substrates have been successfully prepared (1, 6, 9). P~ly(dA-[~HldT), after OsO, treatment (14), could be a useful substrate for the thymine hydrate glycosylase. Alkylating agents of high specific radioactivity such as [3H]dimethylsulfate or [3H]methylmethanesulfonate have been used to label nonradioactive DNA in preparing substrates for 3-methyladenine and formamidopyrimidine glycosylases (1 1-13). Uracil-containing DNA, the substrate for uracil glycosylase, can be isolated from the B. subtilis phage PBS2 (21), from deoxyuridine triphosphatase, uracil-DNA glycosylasedeficient (dut ung) E. coli mutants, or various coliphage grown on the dur ung bacteria (22,231. PBS2 DNA, which contains uracil in its DNA, is the only known example of a naturally occurring DNA that is also a substrate for a DNA glycosylase. After making the DNA substrate, the activity of a given glycosylase can be determined by assays described in the following sections. 18. A. D. Frankel, B. K . Duncan, and P. E. Hartman, J . Bncteriol. 142, 335 (1980). 19. M. G. Wovcha and H. R. Warner, JBC 248, 1746 (1973). 20. J. E. Clements, S. G . Rogers, and B. Weiss, JBC 253, 2990 (1978). 21. T. Lindahl, “Methods in Enzymology,” Vol. 65, p. 284, 1980. 22. H. Warner. R. B. Thompson, T. J. Hozer, and B. K . Duncan, JBC 254, 7534-7539 (1979). 23. H. R. Warner, B. K . Duncan, C. Garrett, and J . Neuhard, J . Bacteriol., in press (1981).

27. DNA GLYCOSYLASES

57 1

2. Buse-Release Assuy Following digestion of the DNA substrate by the glycosylase preparation, the DNA is precipitated with alcohol or acid, and the radioactive nucleoside base remains in the supernatant fractisn. Because all known glycosylases are active in the absence of added divalent cations, a chelating agent such as EDTA is included in the initial incubation. Nuclease activities, which could also release nucleotides into the supernatant, are thereby inhibited. If nucleases interfere with the assay the reaction mix can be passed through a Dowex-1 column. All. nucleotides and DNA are bound to the Dowex and only the released base can pass through the column (24).The base-release assay is simple and offers the advantage that the reaction product can be properly identified by chromatographic analysis. It may be necessary to employ a two-step procedure using precipitation and paper chromatography to quantitate a single DNA glycosylase (13). The base-release assay is not suitable for dimer glycosylases: Only one of the pyrimidine-dimer glycosidic bonds is broken by the dimer glycosylase, so the dimer is not released from the DNA. However, the pyrimidine-dimer glycosylases have been assayed by irradiating the reaction mix with U V light, after the enzyme incubation but before precipitation of DNA, to allow release of thymine from the dimer by “photoreversal” (14, 15). 3 . Filter-Binding Assuy

This assay procedure requires the use of covalently closed circular DNA as substrate (12, 25). It measures the conversion of closed circular DNA to its open (nicked) circular form. The glycosylase removes a nucleoside base to generate an AP site in the substrate. The glycosylase product is then nicked to the open circle form by AP endonuclease, which is included in the reaction mix. After denaturing the DNA in alkaline conditions for a short time, the reaction mix is neutralized and filtered through a cellulose nitrate filter. DNA nicked by the AP endonuclease does not renature; it binds to the filter, whereas the unreacted, closed circular DNA passes through. Basically this assay is a modified endodeoxyribonuclease assay. It is more sensitive than the base-release assay because the generation of a single AP site results in the binding of several hundred molar equivalents 24. B . K . Duncan, P. A. Rockstroh, and H. R. Warner,J. Barrerid. 134, 1039 (1978). 25. F. T. Gates and S. Linn, JBC 252, 2802 (1977).

572

BRUCE K . DUNCAN

of nucleotide. Furthermore, the radioactive label need not be located in the damaged nucleoside base to detect a glycosylase activity. It is important to establish that a glycosylase and not a nuclease is being measured. This is done by demonstrating an AP endonuclease requirement for binding DNA to the filter. The presence of an AP site should always be demonstrated with an AP endonuclease. However, after a glycosylase activity has been established an alkaline hydrolysis step can sometimes be substituted for the nuclease treatment in the filter-binding assay; an AP site is very sensitive to alkaline hydrolysis. The measurement of class I1 DNA glycosylases by this assay is complicated by the intrinsic AP endonuclease activity. Excess AP endonuclease should still be added to the assay because the AP endonuclease activity of the glycosylase is not sufficient to assure complete hydrolysis of all of the AP sites made by the glycosylase (26).

B. ENZYME PURIFICATION Procedures for preparing highly purified (- 1000-fold) DNA glycosylases from different sources have been published (see Table I). A typical procedure uses a series of standard chromatography steps: Gel filtration, ion exchange (phosphocellulose), hydroxylapatite, and affinity (DNA-cellulose) chromatography. The major problem is to obtain the desired enzymes free of contaminating glycosylases and nucleases. For example, in crude extracts of E. coli, the specific activity of uracil-DNA glycosylase (3 mU/mg) is from lo3 to 10; higher than the 3-methyladenine (3 pUlmg), hypoxanthine (0.1 pU/mg), or formamidopyrimidine (0.01 pUlmg) glycosylases (Table I). Thus, while it is relatively easy to purify uracil glycosylase, it may be difficult to obtain other glycosylases free of uracil glycosylase because the latter is so abundant. Endonucleases can also be troubling contaminants in glycosylase preparations. Three AP endonucleases have been purified from E. coli: exonuclease I11 (also called end011 and endoVI) (20, 27), endonuclease I11 (14, 26), and endonuclease IV (28). Uracil glycosylase copurifies with exonuclease I11 during gel filtration and phosphocellulose chromatography (B. Duncan, unpublished results). Similar difficulties have occurred in the purification of the M. luteus pyrimidine dimer glycosylase (16). It is useful 26. H. R. Warner, B. Demple, W. Deutsch, C. Kane, and S. Linn, PNAS 77, 4602 (1980).

27. B. Weiss, S. G . Rogers, and A. F. Taylor, in “DNA Repair Mechanisms” (P. C. Hanawalt, E. C. Friedberg, and C. F. Fox, eds.), p. 191. Academic Press, New York, 1978. 28. S. Ljungquist, JBC 252, 2808 (1977).

27. DNA GLYCOSYLASES

573

to measure nuclease activities during a glycosylase purification to determine whether nuclease activities are contaminants or are an integral part of a class I1 glycosylase. The more definitive way to determine if a DNA glycosylase and an AP endonuclease activity reside in a single protein is to isolate enzyme-deficient mutants and see if both enzyme activities are lost (27). In purifying bacterial glycosylases, mutant strains that lack some of the above undesirable nucleases or glycosylases are useful as enzyme sources. An exonuclease 111-deletion mutant has been isolated that has only 10% of the AP endonuclease activity normally found in wild-type E. coli (27). Uracil glycosylase-deficient (ung) (24, 29) and 3-methyladenine glycosylase-deficient ( l a g ) mutants also exist; these have some residual glycosylase activity (I I , 24, SO). More recently ung insertion mutants have been isolated that have no (<0.1%) uracil glycosylase activity (B. Duncan, unpublished results). Another approach to preparing large amounts of bacterial glycosylases that are free of interfering enzyme activities is to amplify the desired enzyme activity by recombinant DNA techniques. Lambda ung specialized transducing phage and ColE ling plasmids from E. coli strains have been isolated. Either the plasmid, pLC 34-44 (.?I), or the phage vector can be used to increase the amount of uracil glycosylase activity in crude extracts (B. Duncan, unpublished results).

C. REACTIONMECHANISM The general DNA glycosylase-catalyzed reaction is shown in Fig. 2. An unusual or damaged nucleoside base (B’) is excised from a DNA polymer by the hydrolysis of the deoxyribose N-glycosidic bond to yield a free base and an AP site in the position formerly occupied by the base. Thus, in formal nomenclature, the enzyme is an N-glycohydrolase (EC 3.2.2). Most of the enzymes catalyzing glycosidic bond cleavages of nucleosides and nucleotides are reversible, Mg’+-dependent phosphorolysis reactions. Lindahl clearly demonstrated the hydrolytic nature of the glycosylases by showing that uracil glycosylase has no phosphate requirement, nor does it incorporate phosphate into DNA ( 5 ) . The equilibrium constant for the glycosylase reaction favors hydrolysis. Lindahl found no evidence for the reverse reaction, which would be a base “insertase” activity (5). Uracil 29. F. Makino and N. Munakata, J . Bacreriol. 131, 438 (1977). 30. P. Karran, T. Lindahl, I. Ofsteng, G. B . Evensen, and E. Seeberg, JMB 140, 101 (1980). 31. L. Clarke and J. Carbon, “Methods in Enzymology,” Vol. 68, p. 396, 1979.

574

BRUCE K. DUNCAN

I

- PO4-dRib - P04-CH2 B

B HY o $ H

I

I

. . -dRib -POq-dRib-POq

H

I

PO4- dRib \r

B

1

*

a

H20

1

B

..-

+

B

I

--dRib- PO4-dRib-PO4

FIG.2. The DNA glycosylase reaction. B’ nucleoside base: A, C, G or T.

=

OH

+

B‘

I

H

H

bases shown in Fig. 1. B = conventional

glycosylase cannot catalyze a trans-N-deoxyribosylase reaction, an activity that might also permit the reinsertion of a nucleoside base into DNA after the excision of uracil ( 5 ) . The criterion for identifying a DNA glycosylase is the identification of the reaction products: An excised nucleoside base and an AP site in the DNA. The DNA glycosylase reaction probably is mechanistically similar to other glycohydrolases and nucleosidases, such as lysozyme and P-galactosidase, whose mechanisms are considered in Volume VII of this series (32, 33). AMP nucleosidase and NAD glycohydrolase have also been carefully, but less thoroughly, characterized (34). All of these enzymes are believed to catalyze the formation of a “carboxonium ion” intermediate. Garrett tried to detect a 5,6-dihydropyrimidine intermediate that might be formed during the uracil glycosylase reaction because such an intermediate would be more easily hydrolyzed than its unsaturated counterpart. Double labeled [2-’4C,6-3H]uracil-DNA(PBS2) was used to 32. T. Imoto, L. N. Johnson, A. C. T. North, D. C. Philips, and J. A. Rupley, “The Enzymes,” 3rd ed., Vol. 7, p. 665, 1972. 33. K. Wallenfel and R. Weil, “The Enzymes,” 3rd ed., Vol. 7, p. 617, 1972. 34. E. G. Cordes and H. G. Bull, in “Transition States of Biochemical Processes” (R. D. Gandour and R. L. Schowen, eds.), p. 429. Plenum, New York, 1978.

27. DNA GLYCOSYLASES

575

test for an a secondary tritium isotope effect. Since the k , , / k Hratio was 1.00 ? 0.02 no conclusion concerning the proposed intermediate could be made (35). D. SUBSTRATE SPECIFICITIES Detailed studies of the substrate specificity of the DNA glycosylases are limited by the availability of suitable substrates for testing. The results obtained using the easily prepared nucleoside base modifications suggest that these enzymes are highly specific; however, subtle modifications of the natural substrates have not been tested. The most extensively studied enzymes are the uracil glycosylases. Several studies have shown that the enzyme has an absolute specificity for DNA polymers that contain uracil. The simplest and smallest substrate for the bacterial uracil glycosylases is (dU),; this compound is a poor substrate and it is not known how well it binds to the enzyme (6). Smaller oligodeoxyuridylates, dUMP, deoxyuridine, uridine, and RNA are not substrates. Other unconventional DNAs have been tested as substrate with uracil glycosylases: DNAs that contain 5-bromouracil, 5-fluorouracil, 5-hydroxymethyluracil, xanthine, hypoxanthine, pyrimidine dimers, 3methyladenine, 7-methylguanine, or other alkylated base residues and their breakdown products ( I , 5 , 7). Of these, only 5-fluorouracil-DNA is a substrate for E. coli-uracil glycosylase (36). Substrates that have not been tested with E. coli-uracil glycosylase include deoxyribouridylate residues in RNA and ribouridylate residues incorporated into DNA. Uracil glycosylase quantitatively releases small amounts of uracil from DNA. Complete digests of DNA with large amounts of uracil [PBSZ. or poly(dU,dC)] do not give 100% uracil release ( I , 5 ) . Chain termini, AP sites, or nearest neighbor uracil residues appear to decrease enzyme activity in an unknown fashion. To the extent to which they have been tested, the other glycosylases appear to have similarly restricted substrate specificities. 3-Methyladenine glycosylase excises 3-ethyladenine, but probably not hypoxanthine or 3Tethylguznine ( Q ). The pyrimidine dimer-specific enzyme hydrolyzes TIY, C!, TC and CC. Alternative substrates for the hypoxanthine, thymine hydrate, and formamidopyrimidine glycosylases have not been described. The DNA molecule presents multiple sites for glycosylase action. Enzymes that modify DNA interact with their macromolecular substrate in 35. C. E. Garrett, Ph.D. Thesis, University of California, San Francisco, 1979. 36. H . R. Warner and P. A. Rockstroh,J..Bocteriol. 141, 680 (1980).

576

BRUCE K. DUNCAN

two ways. Distributive enzymes dissociate from the DNA after they catalyze a reaction, whereas processive enzymes remain attached and move along the DNA to catalyze successive reactions without dissociating from the substrate. The processiveness of DNA glycosylases has not been examined. In addition, the relative affinity of the enzymes for substrate DNA and nonsubstrate DNA needs to be examined. Both the K , and turnover numbers can depend on the frequency of glycosylase-sensitive sites under certain conditions. The K , values of several glycosylases have been reported in the range to lo-" M substrate residues (Table 11). The concentration basis for determining the K , values is uncertain because the K , values have been determined with single DNA preparations whose mole percentage of substrate nucleotide varied from 100% to less than 0.001% (6, 13). In one case the K , value of calf thymus hypoxanthine glycosylase has been reported in the range 5 x lo-' to 9 x M dIMP for two substrates containing 50% [poly(dI) . poly(dC)] or 0.05% [poly(dA,dI) poly(dT)] hypoxanthine. In this case it seems that hypoxanthine glycosylase acts in a distributive manner and recognizes only a small portion of the DNA near the nucleoside base to be excised. The K , for poly(d1) was not reported. Escherichiu coli uracil glycosylase has a K , of about 4 x M dUMP for native and denatured PBS2 DNA (5). The B. subtilis enzyme has a K , of M dUMP using poly(dU) as a substrate; the addition of excess poly(dA) does not cause any inhibition (6). The 3-methyladenine and formamidopyrimidine glycosylases appear to have an absolute requirement for double-stranded DNA (1 1-13]. Other glycosylases show a distinct preference but not a requirement for either native or denatured DNA. Poly(d1) and poly(dU) are degraded at about 10% of the rates of their double-stranded counterparts, poly(d1) . poly(dC) and poly(dA) poly(dU), by hypoxanthine glycosylases and by mammalian uracil glycosylase, respectively (8, 10). The bacterial uracil glycosylases are the only enzymes known to hydrolyze single-stranded DNA preferentially. Denatured PBS2 DNA is hydrolyzed 2.5-fold faster than native DNA ( 5 ) . The effects of temperature and ionic strength cannot be ignored when synthetic homopolymers are used to probe the requirements of a DNA glycosylase for a helical substrate. The melting temperatures of PBS2 DNA, poly(dA-dU), poly(dA) * poly(dU), and poly(d1) * poly(dC) are 57", 35", 37", and 28" in 0.01 M NaCl(37). Thus all of these DNAs are partially single-stranded at 37". Without varying the reaction temperature or salt

-

37. G . D. Fasman, (ed.) "Handbook of Biochemistry and Molecular Biology," 3rd edition, Vol. I, p. 560, 575. Chemical Rubber, Cleveland, 1975.

27.

DNA GLYCOSYLASES

577

concentration it is difficult to evaluate the importance of DNA structure for substrate binding by a DNA glycosylase. Most DNA glycosylases are mildly inhibited by salt: calf thymus DNA glycosylase activity is stimulated fourfold by adding 0.1 M KCI t o the standard assay when poly(d1) * poly(dC) is used. Uracil and hypoxanthine form conventional A : U and I : C base pairs in PBS2 DNA and poly(d1) . poly(dC),7qespectively. Unpaired or unorthodox base pairs, G : U and I :T, have been prepared by deaminating cytosine residues in DNA (G : C -+ G : U) and by synthesizing poly(dA, 0.1% dI) . poly(dT). Lindahl has shown that uracil and hypoxanthine glycosylases are able to excise uracil and hypoxanthine from paired and mispaired structures in double-stranded DNA molecules (5, 10).

E. GLYCOSYLASE INHIBITORS Uracil is a noncompetitive inhibitor of bacterial and mammalian uracil glycosylases with apparent Ki values in the range lo-' to lop3M . Several uracil derivatives have been tested for their capacity to inhibit the different uracil glycosylases but so far no other inhibitors have been found. The list of derivatives tested includes deoxyuridine, thymine, orotic acid, 2-thioruacil, 5-bromouracil, 5-fluorouracil, 5-aminouracil, 6-azauracil, 3-deazauracil, 2-hydroxypyrimidine, 4-hydroxypyrimidine, and barbituric acid (5, 7). The failure to find any inhibitors seems to confirm the high substrate specificity of uracil gly'iosylase. Fluorouracil-DNA is a substrate for E . coli uracil glycosylase, but fluorouracil has not been tested as an E . coli glycosylase inhibitor, though it does not inhibit the mammalian enzyme. The product inhibition of other glycosylases has not been extensively studied. Based on the limited substrate specificity of the glycosylases it seems reasonable to assume that few product analogs will inhibit enzyme activity. Thus, only 3-methyl or 3-ethyladenine inhibit 3-methyladenine glycosylase; adenine, I-methyladenine, 7-methyladenine, 3-methylguanine, and @-methylguanine fail t o inhibit this enzyme. Interestingly, calf thymus hypoxanthine-DNA glycosylase is not inhibited by hypoxanthine. Two proteins have been identified that inhibit uracil-DNA glycosylase. Bacteriophages PBS2 and T5 each induce the synthesis of a specific protein that inhibits uracil glycosylase in vivo and in vitvo. The PBSZcoded inhibitor inhibits not only its homologous B. subtilis enzyme but also the glycosylases from E. coli and human tissue ( 4 ) . It is not known if either of these proteins can inhibit other DNA glycosylases or other enzymes (13).

578 IV.

BRUCE K. DUNCAN

Physiological Role

A. DNA GLYCOSYLASES AS REPAIRENZYMES Hydrolytic reactions are intrinsic to the nature of DNA and cannot be avoided in the cellular milieu. The deoxyribose moiety of DNA makes DNA more resistant to nonenzymatic hydrolysis than RNA, and the double helical structure of DNA provides some protection against hydrolytic reactions of the nucleoside bases in DNA. Nevertheless, cytosine, adenine, and guanine deaminate to uracil, hypoxanthine and xanthine, respectively, at significant rates. Other reactions that affect the basepairing properties of cellular DNA are depurinations, hydrolysis of purine imidazole rings, and purine or pyrimidine hydration. Environmental agents such as ultraviolet, gamma, and X irradiation, and various chemicals are also effective in producing DNA damage (38-40). A short catalog of some of the most common damaged nucleoside bases is shown in Table 111. A comparison of the substrate specificities of DNA glycosylases (Table I) with the most common types of damaged DNA (Table 111) shows that specific DNA glycosylases exist for most of the common damaged nucleoside bases. The primary physiological function of DNA glycosylases is probably DNA repair. The cell must be able to repair these chemical alterations in its DNA if it is to maintain the functional integrity of its genome and have an acceptably low mutation frequency. Experiments with bacterial mutants defective in either uracil glycosylase or 3-methyladenine glycosylase activities clearly indicate the DNA repair function of these enzymes. DNA glycosylases are essential for the rapid removal of damaged nucleoside bases from DNA. There are many different chemical modifications of DNA residues; some modifications are potentially lethal, some are mutagenic, and some have no noticeable effect. DNA repair seems to be directed against potentially lethal or mutagenic lesions. Failure to remove deaminated cytosine (uracil) or 3-methyladenine residues causes mutations (30, 4 / ; see later 44). Another important consequence of failing to excise damaged nucleoside bases is to modify protein-DNA interactions that are important in the control of gene expression and replication. 38. D. M. Brown, in “Basic Principles in Nucleic Acid Chemistry” (P. 0. P. Ts’o, ed.), Vol. 2, p. 1. Academic Press, New York, 1974. 39. B. Singer, JNCI 62, 1329 (1979). 40. N . K . Kochetkov and E. I. Budovskii, “Organic Chemistry of Nucleic Acids,” pp. 269-609. Plenum, New York, 1972. 41. B. K. Duncan and B. Weiss, in “DNA Repair Mechanisms” (P. C. Hanawalt, E. C . Friedberg, and C. F. Fox, eds.), p. 183. Academic Press, New York, 1978.

27.

579

DNA GLYCOSYLASES TABLE 111

IMPORTANT DAMAGED NUCLEOSIDEBASES

DNA damage Uracil-DNA

Hypoxanthine-DN A

Xanthine-DNA

3-Methyladenine-DNA @-Met hylguanine-DN A 7-Met h ylguanine-DNA 0'-Methylcytosine O'-Methylthymine 0 '-Methylthymine

Aminopyrimidine-DNA or formamidopyrimidine-DNA

Thymine hydrate-DNA or dihydrothymine-DNA

Source of DNA damage Spontaneous and mutagen-induced deamination of DNA cytosine residues, d U T P incorporation by DNA polymerase Spontaneous and mutagen-induced deamination of DNA adenine residues Spontaneous and mutagen-induced deamination of DNA guanine residues Nonenzymatic alkylation by mutagens Nonenzymatic alkylation by mutagens (39) Nonenzymatic alkylation by mutagens (39) Nonenzymatic alkylation by mutagens (3Y) Nonenzymatic alkylation by mutagens (39) Nonenzymatic alkylation by mutagens (3Y) Ionizing radiation, hydrolysis of purine imidazole rings, including 7-methylpurines and purine hydrates from mutagens (13, 4 0 ) Ultraviolet radiation, ionizing radiation, chemical mutagens

-

Major base pair properties A:U

I:C

x:c MeA : T MeG : C MeG : C MeC : A MeT: G MeT: A MeT: G MeT: A Unknown

Unknown

(14, 40)

Pyrimidine dimer-DNA AP-DNA

Ultraviolet radiation Spontaneous or enzymatic loss of nucleoside base

Unknown None

3-Methyladenine DNA glycosylase-deficient (rag) mutant bacteria (30) are hypersensitive to the lethal effects of the alkylating agent methyl methanesulfonate (MMS). These bacteria fail to reactivate MMS-treated bacteriophage, and are hypermutable when treated with MMS. Two different tag mutants have been studied; one has a temperature-sensitive methyladenine glycosylase activity that makes the cells MMS-sensitive only at high temperatures. All of these phenotypes are a consequence of

580

BRUCE K. DUNCAN

the observed inability of mutant cells to excise 3-methyladenine effectively from DNA. When tag mutants are treated with sublethal doses of MMS they are able to continue growing at normal growth rates, even though the chromosome may contain as many as 80 3-methyladenine (and 450 7-methylguanine) DNA residues per cell. Thus small amounts of 3-methyladenine are tolerated by E . coli, although higher levels of DNA damage cause cell death. Uracil-DNA glycosylase-deficient (ung f mutants have an increased spontaneous mutation rate ( 4 / )and have some uracil residues substituted in place of thymine in their chromosome (23).irng mutants permit uracilDNA phage to replicate, whereas ling+ cells restrict the growth of uracilDNA phage (22). These phenotypes are due to the inability ofung mutants to excise uracil residues from DNA. Lindahl and Nyborg predicted from studies of spontaneous cytosine deamination at pH 7.4 in physiological saline at different temperatures that mutagenic cytosine deamination should occur at a high rate (42).The rate constants for cytosine deamination in single- and double-stranded DNA are calculated to be about 2 x lo-“’ sec-I, and lo-” sec-l at 37” (17). This extrapolates to an expected mutation frequency in the range 2 x to 4 x lo-’ for a single G : C base pair in a log culture with a 30 min doubling time. The measured G :C + A : T mutation frequency for a single base pair (by reversion of a frpA allele, trpA446) in ung+ E . coli is and lo-’”. ung mutant cells were found less than expected: Between to have a G : C + A : T base substitution mutation specificity and revert the fupA446 allele at a frequency of more than These results are consistent with the idea that uracil-DNA glycosylase is involved in the repair of chromosomes damaged by cytosine deamination (41). Methylcytosine residues deaminate at about the same rate as cytosine, and the product (thymine) is not a uracil glycosylase substrate. In studies of the lac I gene, Coulondre et al. found that methylcytosine-DNA residues are mutagenic “hot spots” by genetic and DNA-sequence analysis (43).In finding that the hot spots disappeared into the background of other GC + AT transition mutations in ung mutants, Duncan and Miller demonstrated that cytosine deamination occurs at a high enough rate in vivo to constitute a significant mutagenic challenge to the genetic stability of the cell (44). The cell therefore must be able to repair deaminated cytosine residues in DNA if the mutation rate is to be kept at the observed low level. Uracil glycosylase excises the deaminated cytosine. 42. T. Lindahl and B. Nyberg, Biochemistry 13, 3405 (1974). 43. C. Coulondre, J. H. Miller, P. J. Farabaugh, and W. Gilbert, Ntifrtre (londurr) 274, 775 (1978). 44. B . K. Duncan and J. H. Miller, Nature (London) 287, 560 (1980).

27.

DNA GLYCOSYLASES

58 1

The DNA isolated from ung cells also contains occasional uracil residues base-paired with adenine, a consequence of dUTP incorporation. The enzyme deoxyuridine triphosphatase (dUTPase) largely prevents the incorporation of more than a few tenths of a mole percentage uracil into DNA by keeping the in vivo dUTP pool at a n undetectabIy low level. The small amount of uracil incorporated into DNA from dUTP is excised by uracil glycosylase (45,46). By combining a dUTPase- (dut) mutation with an ung mutation the mole percentage uracil increases to about 4% in viable E. coli and to as much as 10% in T4 phage ( 2 3 , 4 7 ) .Dut cells grow poorly, either because of a partial thymidine nutritional requirement [they require TdR at high temperature (B. Weiss, personal communication)], or owing to the high level of uracil-DNA repair, or both. The ung mutation suppresses most of the dut phenotypic properties and dut ung bacteria are healthier than dut cells. Many phages grown in dur ung bacteria incorporate uracil into their DNA (fl, P1, A , TQlc, and TSdut), but T4+ or T5+do not (22). These phages have a low plating efficiency on ung+ cells because the defective repair of many uracil residues in each DNA strand causes lethal damage to the DNA, probably double-strand chain breaks. This behavior of uracil-DNA phage is similar in some ways to the host modification-restriction systems discovered by Arber (48).In this case the DNA modification is uracil-DNA; the low relative plating efficiency of uracil-DNA phage shows that uracil glycosylase can function as a restriction enzyme. Bacteriophages often have mechanisms for inactivating restriction enzyme systems. The phages PBS2 and T5 both make uracilDNA glycosylase inhibitors (49, 50). It is clear that uracil glycosylase functions to exclude uracil from DNA. Uracil glycosylase cannot distinguish between dA : dU and dG :dU base pairs. If the function of uracil glycosylase is to suppress the potentially mutagenic effects of cytosine deamination it must also exclude uracil frQ>m normal Watson-Crick (A : U) base pairs. Therefore, one may argue thit uracil cannot be a normal constituent of DNA. The exclusion of uracil from A :U base pairs may play an important role in the control of gene expression. Fisher and Caruthers have shown that the methyl substituent of a specific A : T base pair in the lac 0 gene is essential for proper recognition by the repressor. A :U and G :C base pair 45. B. K. Tye, J. Chien, I. R. Lehman, B. K. Duncan, and H. R. Warner, P N A S 75, 233 (1978). 46. F. Makino and N. Munakata, J. Bacrariol. 134, 24 (1978). 47. H . R. Warner and B. K. Duncan, Nature (Lorzdon) 272, 32 (1978). 48. W. Arber, Progr. Nircleic Acid Res. Mol. B i d . 14, 1 (1974). 49. H. R. Warner, L. K. Johnson, and D. P. Snustad, J. Virol. 33, 535 (1980). 50. E. C. Friedberg, A. K. Ganesan, and K. Minton, J. Virol. 16, 315 (1975).

582

BRUCE K. DUNCAN

substitutions in place of the position 13 A :T base pair abolish repressoroperator recognition, whereas a G : MeC substitution preserves the specific repressor binding properties of the lac operator (51). Similar results have been obtained in vivo with dut ung bacteria; uninduced p-galactosidase is overproduced about 20-fold ( B . Duncan, unpublished result). Thus uracil-DNA (in A : U base pairs) may interfere with gene expression. Seeman et al. have described the likely protein-nucleoside base interaction sites of DNA; nucleoside modifications should affect some of these interactions (52). Other nucleoside base modifications such as hydrations or alkylations should also affect gene expression. Less is known about the specific function of other DNA glycosylases but it seems safe to assume that they also protect the cell against the effects of damage to DNA nucleoside bases. Hypoxanthine-DNA glycosylase, for example should repair mutagenic (A :T + G : C) adenine deaminations (10). The cell does not have to prevent dITP incorporation into DNA since dITP is not biosynthesized. The formamidopyrimidine glycosylase excises a degradation product of 7-methylguanine. It may also excise purine derivatives resulting from spontaneous or radiation-induced hydrolysis of the purine imidazole ring (13). The ability of the enzyme to excise damage from irradiated DNA has not been tested.

B. BASEEXCISION REPAIR AP sites in DNA are formed by DNA glycosylases or by nonenzymatic hydrolysis of glycosidic bonds. The repair of AP-DNA requires at least one nuclease, a polymerase, and a DNA ligase. The newest proposal of the E. coli base-excision repair pathway is shown in Fig. 3 (14). DNA glycosylases initiate repair by base excision. AP endonucleases incise AP-DNA at the AP site to generate nicks with 3' or 5' termini of base-free deoxyribose (AP) residues. Other nucleases or DNA polymerases remove the deoxyribose residues from the DNA. DNA polymerase incorporates new nucleotides to make nicked DNA, which can be ligated by DNA ligase. Substantial genetic evidence for the participation of DNA ligase, DNA polymerase, and exonuclease 111 in base-excision repair exists [Chapter 14, this volume and (27, 5311. However this repair pathway should only be regarded as provisional. Other enzymes, such as exonuclease VII (54), may function in addition to, or instead of, the enzymes 51. 52. 53. 54.

E. F. Fisher and M. H. Caruthers, Nircleic Acids Res. 7, 401 (1979). N . C. Seeman, J. M. Rosenberg, and A. Rich, PNAS 73, 804 (1976). B. K. Tye and I. R. Lehman,JMB 117, 293 (1977). J. W. Chase and C. C. Richardson, JBC 249, 4553 (1974).

3'T C G A G T A C A A G T A G C T C A T G T T C A

dUTP INCORPORATION or

-

CYTOSINE DEAMINATION

-3'----

i c

5' G A G T A C A A G :

A G U T C A U G T T C A

5'-

T C G A G T A C A A G T A G C T C A T G T T C A

NICKED-DNA

T C G A G T A C A A G T A G T C A G T T C A

AP-DNA

FIG.3. The base excision repair pathway of E. coli. Uracil-DNA repair is shown. DNA glycosylase initiates repair by base excision. AP endonucleases incise AP-DNA at the AP site to generate nicks with 3' or 5' termini of base-free deoxyribose (AP) residues. Other nucleases and/or DNA polymerases remove the deoxyribose residues from the DNA. DNA polymerase incoporates new nucleotides to make nicked DNA, which is ligated by DNA ligase. This model incorporates the results obtained by Warner e t a / . (55) into Lindahl's model ( 1 7 ) . Further modifications will be necessary as new information concerning the activity of exonuclease 111, endonuclease 111, endonuclease IV, and DNA polymerase on the nicked AP-DNA repair intermediates is obtained.

584

BRUCE K. DUNCAN

shown. A similar repair pathway probably functions in other organisms, since DNA glycosylases and AP endonucleases are found in many organisms. The addition of a second parallel pathway for AP-DNA repair to the overall base excision repair pathway originally proposed by Lindahl (2) (see Fig. 3) is necessitated by the recent discovery that the AP endonuclease cleavage product of endonuclease I11 differs from the endonuclease IV and exonuclease 111 products [Chapters 8 and 14 this volume and (14, 55)]. Endonuclease IV or exonuclease I11 cuts AP-DNA to yield a deoxyribose 5-phosphate (AP) residue on the 5' terminus of the nicked DNA chain. The 5' + 3' exonuclease activity of DNA polymerase I(56) or I11 (57) (or possibly an endonuclease I11 activity) can efficiently excise the AP residue during nick translation. In contrast, endonuclease III\:uts the AP-DNA to leave an AP residue on the 3' terminus of the nick that DNA polymerase I cannot excise. In this case the AP residue blocks DNA synthesis until exonuclease 111, or possibly endonuclease IV, excises the AP residue from the DNA. Of the two AP-DNA repair pathways shown, the exonuclease IIUendonuclease IV pathway would seem to be the simpler and more important pathway for class I glycosylases because exonuclease I11 is the major AP endonuclease activity in the cell (27). Class I1 glycosylases may channel repair through the endonuclease 111-dependent pathway because all of the known class I1 glycosylases have endonuclease 111-like AP endonuclease activities. The endonuclease I11 pathway still requires exonuclease I11 for the excision of the AP residue.

V.

Research Applications

All of the DNA glycosylases are potentially useful for detecting specific lesions in DNA by either the direct measurement of released bases or by the indirect but more sensitive method of introducing alkali-sensitive AP sites into DNA. Uracil glycosylase has been used to detect uracil residues formed by treating DNA with heat, NaHSO, or nitrous acid ( 5 , 18). The nitrous acid experiments allowed Frankel et (11. (18) to conclude that deaminated cytosine is not the major mutagenic lesion induced by nitrous acid treatment of pure DNA. The enzyme has also been used to detect uracil-DNA, to estimate the mole percentage of uracil in DNA (45,581 and 55. H. R. Warner, B. Demple, W. Deutsch, C. Kane, and S. Linn, PNAS 77,4602 (1980). 56. R. B. Kelly, K. R. Atkinson, J. A. Huberman, and A. Kornberg, Nurure (London) 224, 495 (1969). 57. D. M. Livingston and C. C. Richardson, JBC 250, 470 (1975). 58. J. B. Hays, B. K. Duncan, and S. Boehmer,J. Bacreriol. 145, 306 (1981).

27. DNA GLYCOSYLASES

585

to estimate the dUTP pool size of tissue culp-%cell? (59). AApyrimidine dimer glycosylase has been used to detect TT, CT, TC, and CC dimers in DNA sequences and thus demonstrate the substrate specificity of this enzyme (16 ). DNA glycosylases should be useful synthetic tools for synthesizing AP-DNA sequences. [ c I - ~ ~ P I ~ Uhas T P been copolymerized into poly(dAdT) and treated with uracil glycosylase to make an AP-DNA substrate for determining the site of phosphodiester bond cleavage of AP endonucleases (20). From I 4x174 or fl DNA prepared from ung bacteria should contain 2-4 uracil residues per molecule (45). Digestion of this DNA with uracil glycosylase should yield a useful AP endonuclease substrate because it can be prepared daily with a high degree of reproducibility. It is possible to prepare large amounts of synthetic, defined sequence oligodeoxynucleotides for crystal and solution structure studies. Uracil glycosylase can be used to prepare AP-DNA for structural studies. The selfcomplementary structure dG-dC-dG-dRib-dA-dA-dT-dT-dG-dC-dG-dC could be made from dG-dC-dG-dU-dA-dA-dT-dT-dG-dC-dG-dC. This structure cannot be made by other methods due to the chemical instability of the AP site. Finally, the uracil, hypoxanthine, and 3-methyladenine glycosylases are potentially useful in specialized applications of DNA sequencing (60, 61). Although DNA glycosylases have obvious uses in nucleic acid research, there are potential difficulties associated with their use. To prepare uracil-free DNA it is important to avoid exposing DNA to conditions that induce cytosine deamination. Heat denaturation of DNA in a Trisbuffered solution will deaminate enough cytosine residues to produce a shift in the alkaline sucrose gradient profile of uracil glycosylase-treated lambda DNA (B. Duncan, unpublished result). Most cytosine compounds are contaminated with their uracil analogs. It is important to purify these compounds before using them for synthesizing DNA if uracil-free DNA is desired; for example, commercial preparations of dCTP may contain as much as 10% dUTP. The simplest way to remove contaminating dUTP from nucleoside triphosphates is to add purified dUTPase (62, 63) to the triphosphates before using them to synthesize DNA. If uracil is incorporated into DNA, by design or by chance, the presence of uracil glycosylase as a contaminant of other enzyme preparations 59. 60. 61. 62. 63.

S . Nilsson, P. Reichard, and L. Skoog, JBC 255, 9552 (1980). A. M . Maxam and W. Gilbert, “Methods in Enzymology,” Vol. 65, p. 499, 1980. A. J. H. Smith, “Methods in Enzymology,” Vol. 65, p. 560, 1980. A. F. Taylor, P. G. Siliciano, and B . Weiss, Gene 9, 321 (1980). J. Shlomai and A. Kornberg JBC 253, 3305 (1978).

586

BRUCE K. DUNCAN

will cause DNA degradation. Terminal transferase preparations may contain some uracil glycosylase that would make the synthesis of poly(dU) difficult (21). Similar problems have occurred in DNA replication studies (63). Some of these undesirable effects can be minimized by inhibiting uracil glycosylase with uracil (64) or a uracil glycosylase-inhibitor protein. Enzyme deficient mutants can also be used to prepare extracts devoid of the glycosylase activity. ACKNOWLEDGMENTS Work in the author’s laboratory was supported by NIH grants GM 27813, CA 06927 and RR 05539 and by an appropriation from the Commonwealth of Pennsylvania. The author wishes to thank T. Lindahl for making available copies of manuscripts prior to publication. Special thanks are due Maryann Goetz for help in preparing this manuscript.

Note Added in Proof A urea-DNA glycosylase has been identified in E. coli [see L. Breimer and T. Lindahl, Nucleic Acids Research 8, 6199 (1980)l. A 7-methylguanine-DNA glycosylase has been discovered by three groups: J. Laval, J. Pierre, and F. Laval [ P N A S 78, 852 (1981); B. Singer and T. P. Brent (ibid, p. 856); and G. P. Margison and A. E. Pegg(ihid. p. 861). Strong biochemical and genetic evidence indicating that the T4denV gene product, a single polypeptide chain, is indeed a class I1 pyrimidine dimer-DNA glycosylase with AP endonuclease activity is presented by Y. Nakabeppu and M. Sekiguchi [ P N A S 78, 2742 (l98l)l.

64. K. Brynolf, R. Eliasson, and P. Reichard, Cell 13, 573 (1978).

Author Index Abbott, P. J., 109 Abdel-Monem, M., 8, 250, 367, 368, 442 Abboud, M., 439 Abrell, J. W., 89, 93 Abremski, K., 365, 478 Acs, G . , 546 Adams, R. L. P., 518, 528, 529(95, 96), 532(7), 533 Adema, J., 258 Adler, S . , 62, 63(30), 64(30), 139, 148(16), l50( 16) Agarwal, K. L., 161(22, 231, 162(22, 23,45), 165(45), 166(22), 168, 185, 186(132), 314, 523, 524(71), 526(71) Ahmed, F. E., 497, 512(70), 513 Air, G. M.,336 Akrigg, A., 366 Alberts, B., 9, 11(16), 12(16), 52, 55, 56, 60(10), 333, 355, 356(52), 360(51, 52). 363(51), 367, 370(11), 375, 376, 379(1), 381, 382(6), 383(41, 56). 384(6), 387, 388, 389(38), 392(38, 41). 393(6, 38), 394(38), 396(38), 402(6), 404(37), 405(6), 406(6), 408(38), 409(22, 97). 410(6, 22, 98, 101, 102, 103). 411,412(41), 413(41), 414(4l), 4 15(4I ), 4 l 6 ( 4 l ) , 422(41), 423(1, 96), 424(1, 96). 425(96), 426(79, 96), 427(96), 428(79, 96), 429(96), 430(96), 434(49), 435(49, 172, 173), 436(172), 439, 442(185), 456 Alexandrushkina, N . I., 537, 538(186) Allaudeen, H. S., 88, 106(31), 107 Allen, J., 520. 533(21) Allet, B., 411 Allfrey, V. G., 296 Alton, N. K., 208

Arnodio, F. J., 3 8 i , 382(38), 389(38), 392(38), 393(38), 394(38), 396(38), 408(38) Anagnostopoulas, C., 141, 142(30), 235,237, 238, 244(35) Anai, M., 234, 239, 240(47), 241(47, 48), 242(47, 541, 244(47, 48), 245(48, 60). 246(70) Anderson, C. T. M., 269, 272(78, 791, 566, 577(4) Anderson, C. W., 439 Anderson, E., 418, 419(131) Anderson, P. N., 107 Anderson, R. A., 382, 393(27), 395(27), 396(27), 397(27), 407(27), 408, 414, 415(127), 417(27), 418(127), 419(127, 129), 420(129), 421(129), 424(27), 426(27) Anderson, S., 73 Anderson, A,, 550 Ando,T., 141, 142(31), 190, 193, 195(3), 198, 1990) Anfinsen, C. B., 288, 292, 293(51) Anikeichera, N. V., 523, 524(69) Anraku, N., 392, 410(58), 424 Antoni, F., 234, 239, 241(50), 244(50), 245(50) Aposhian, H. V., 204 Arai, N., 12 Araki, T., 122 Arber, W . , 138, 139(9), 142(13), 143, 145(14), 146, 149(14), 151(14), 156(5), 540, 581 Arens, M., 440 Ariga, H., 440 Arisawa, M., 29

587

588 Arlinghaus, R. B., 94 Armel, P. R., 269, 270(69) Armitage, I. M.,415, 419(129), 420(129, 135), 421(129) Arndt-Jovin, D. J., 387 Arrand, J. R., 162(54), 165(54), 166 Atkins, J. F., 439 Atkinson, K. R., 584 Atkinson, M.R., 16, 28, 36 Atkinson, T., 160(12), 165 Attardi, D. G., 335 Aucker, J., 80, 82 Auer, W. B., 90 Auerbach, J., 383(42), 384, 392(42), 430(42) Auerswald, E. A., 336 Auld, D. S., 90 Austin, G. E., 316, 317(4), 318(4), 319(4), 320(4), 322(4), 324(4), 326(4), 327, 328 Ausubel, F. M., 468 Aviv, H., 99 Axel, R., 295, 546, 548 Axelrod, N., 439, 440(192) Baase, W. A., 335 Bachi, B., 140,520(46), 521(62), 523,524(46, 62) Bachmann, B., 143 Bachmann, B. J . , 209, 210(17), 222(17), 228(17), 424, 429(150) Baer, B., 141 Baev, A. A., 523, 524(68) Baez, J. A., 175 Baezil, J. A., 82 Bahr, W., 387 Bald, R., 520, 524(30) Baldi, M. I., 333, 346 Baldwin, R. L., 429, 456 Baldy, M. W., 424 Bale, A., 540 Ball, W. D., 282, 287(2) Baltimore, D., 28, 67, 87, 88(17), 89,90,91, 92(8), 93(17), 94(55), 95, 96, 98(74), 99, 100, 106(30), 107(19), 108(19), I 1 1(19), 114(19, 23, 30, 64),118(14, 18, 19, 23, 26, 39, 44) Bam, P. R., 439 Bambara, R. A,, 21, 22(19), 23(20), 34, 48 Bandyopadhyay, P. K., 425, 426(154), 427( 154) Banfalvi, G., 234, 239, 241(50), 244(50), 245(50)

AUTHOR INDEX Bank, A., 99 Banks, G. R., 436 Barbour, S., 234, 235(7), 237 Barbour, S. D., 229, 230,470 Bardwell, E., 236 B a d , E. F., 68 Bario, R. J., 382, 392(30) Barkley, M. D., 407 Barksdale, L., 141 Barnes, J. E., 52, 57(3), 302 Barnes, W. M.,36 Barnett, L., 480 Barnhart, B. J., 495 Barnoux, C., 40, 46(7) Barrell, B. G., 336 Barrington Leigh, J., 294 Barry, J., 9, 11(16), 12(16), 56, 60(10), 367, 370(11), 381, 408, 409(22, 97), 410(22, 103) Bartok, K., 125, 199 Barton, R. W., 106, 114, 118(20) Basnakyan, A. G., 533, 548(142) Bassett, C. L., 208 Bastia, D., 244 Bastos, R. N., 234 Battey, J., 5 Battula, N., 90,175 Bauer, H., 94 Bauer, R. W., 336, 343, 347, 348(11) Bauer, W. R., 184 Baumstark, B. R., 171, 176(80), 177(80), 185, 186(133) Baur. R., 519 Bayev, A. A., 190 Bayley, C. R., 265 Baynes, M., 437 Beard, P., 199 Becerra, S. P., 439 Becherer, K., 341 Beck, E., 336, 542 Becker, A., 305, 310(31), 311(31), 314(31) Becker, F. F., 519 Beckwith, J. R., 472 Bedbrook, J. R., 468 Bedinger, P., 381, 409(22), 410(22) Beers, R. F., 258 Belagaje, R., 314 Belfort, M.,474 Belozerskii, A. N., 532, 543(124) Belozersky, A. N., 518, 520, 532 Benbow, R. M., 468

589

AUTHOR INDEX

Benedetti, P., 333,346 Ben Hur, E., 497 Ben Ishai, R., 497 Bentley, C., 150 Benueniste, R. E., 94 Benz, E. W., Jr., 95 Benzinger, R.,142,537,539(194) Berdahl, B. J., 542 Berdyshev, G. D., 546 Berends, L. J., 430 Berends, W., 483 Berg, 0. G., 407 Berg, P.,35,36, 116,117(92), 172,173(82),

175,199,201(21), 229 Berger, H.,410 Berger, H., Jr., 68 Berger, L., 521,524(50), 532,543(125) Bergmans, H. E. N., 468,542 Beringer, J. E., 537,539(194) Berk, A. J., 201,229 Berkner, K.L., 167,168(74), 178,179(104),

301,309(14), 310(14), 526,544(90) Berkower, I., 28.91,212,218(27) Berman, E., 545 Bernardi, G., 122,530 Berns, A . , 96,98(77), lOO(77) Berns, K.I., 83 Bernstein, C., 409 Bernstein, H.,363,409 Bertain, G., 138,537 Bertazzoni, U.,530 Bertino, J. R . , 545 Bertsch, L., 16,17(5), 53 Bertsch, L. L., 215,332,368,369,370(24),

424, 425(143), 426(143), 427(143), 428(143), 429(143, 144), 430(143) Bessman, M. J., IS,28,59, 60 Betlach, M.,139. 142(19) Beukers, R . , 483 Bev, V. G., 520,533(22,24) Bhattacharyya, J. R . , 106 Bhave, N., 167,168(71) Bibor, V., 269,270(67) Bickle, T., 140,148,151(68) Bickle, T.A., 160(11), 165,441,521(62, Ma), 523,524(62, 85a). 525 Bidwell, K., 474 Billen, D., 534 Billeter, M.A . , 100 Bilsker, M., 305 Bina-stein, M., 335

Bingham, A. H. A . , 160(12), 165 Bird, A. P.,180, 181(116),530(155, 156,164,

16S),535,536(165), 546(164), 547 Birge, E. A., 468 Birnie, G. D., 99 Bishop, J. M., 88(19), 89,90,95,98,100 Bishop, J. O., 173 Biswal, N.,536 Bittner, M.,383(56), 388,409,410(102, 103) Black, P. H., 545 Blakesley, R . W.,161(27), 174, 175(27),

176(27), 185(27), 186(27, 130),188 Blanco, M., 448,468(34) Blattner, F. R., 161(41), 165(41), 166 Blikstad, I., 294,296(S9) Blomberg, C., 407 Bloomer, L. S., 548 Boatwright, D. T.,487,488(39) Bobst, A. M., 382 Bockrath, R. C., 469 Bodmer, W. F., 132 Boehm, T.L. J., 534 Boehmer, S., 366,584 Boffa, L. C., 296 Bogdarina, I. G., 523,S24(67) Bogerhagen, D.F., 225 Bohn, E. W., 69,72(17), 74,76(33), 83(22) Boiteux, S., 72 Bolden, A,, 68,76,77(43), 79(43), 82 Bolen, P.L., 545 Boling, M.E., 484,498(17), 500, 508 Bolle, A . , 52,55(4), 57(4), 360,384,392(44),

408(57a), 411(44) Bollum, F. J., 4,67,68,69,72(14), 73(14),

75(14), 79(14), 106,107(29,47),108(22), 109, llO(46, 48), 111(46), 112(46), 113, 114(9, 22, 45, 46, 71), 118(12, 17,20, 40), 175,484,485, 494(23), 496,498(16, 23) Bolognesi, D. P., 90 Bonhoeffer, F., 41 Bonhoeffer, F. J., 387 Bonura, T . , 125, 126(14), 132(14), 269, 272(78,79,79a), 51 I , 566 Borek, E., 521,522(S4), 523,542 Bose, K.,551 Bose, R. J., 536,539(177) Bosselman, R.A., 95,98(73) Botchan, P.,364 Botstein, D., 470 Bott, K.F.,353

590 Bourgeois, S., 502 Bourguignon, G. J., 350 Boyve, R., 484 Boyce, R. P., 274 Boy de la Tour, E., 360, 392, 408(57a) Boyer, H. W., 139, 142(12, 19), 143, 148, 149, 167, 168(75), 172, 173(81), 176(75), 180, 181(117), 186(75), 188, 189, 190(143), 520(43), 521, 523, 524(41, 43, 611, 526(61), 541(41) Boynton, J. E.. 544, 545 Boyse, E. A., 107 Brack, C., 148, 151(68), 441 Bradbury, E. M., 296 Braga, E. A., 184 Brahic, M., 534 Brammer, K. W., 265 Brammer, W., 46 Brandon, C., 537, 539(190, 191, 192) Braun, A., 274 Braun, R., 530(152), 534, 535(152) Brawerman, G., 282, 286(5), 532(135, 136), 533 Brazill, G. W., 238 Breimer, L., 586 Breitmeyer, J. B., 162(50), 165(50), 166 Brenner, S., 480 Brennessel, B. A., 68 Brent, R., 463 Brent, T. P., 269, 270(75) Brent, Thomas, P., 566(12), 567, 570(12), 571(12), 576(12) Breschkin, A. M., 412 Brewer, L., 91 Brietman, M. L.,95 Brinkman, B. J., 79 Brison, O., 284, 440 Brock, C., 393, 394(66), 403(66) Brockes, J. P., 520(44), 521 Broker, T. R., 410, 439 Brornberg, S., 528, .529(100) Bromfeld, E., 90 Bron, S., 141, 142(30), 161(29), 165 Brookes, P., 537 Brooks, J. E., 520,523,524(29), 525(29,75), 526(75), 527(29, 75) Brooks, K., 446, 466(8) Brown, D. D., 99, 225, 520(48), 521, 530(48, 151), 534, 540(48), 546(48) Brown, D. M., 578

AUTHOR INDEX Brown, E. L., 314 Brown, F. L., 543, 548 Brown, G. G., 346(7), 347 Brown, J. S., 161(26), 162(26), 165 Brown, M., 108, 114(45) Brown, M. S., 509 Brown, N. L., 161(40, 44), 162(40), 163(63, 66), 16340, 44,63, 66), 166, 187, 336 Brown, 0. E., 68 Brown, P., 154 Brown, P. O., 339, 340, 349, 351, 352(25, 371, 353(25, 37), 354(37), 356, 358(37, 55). 359(25), 366(25) Brown, P. R., 520(44), 521 Brown, R. S., 314 Brown, S. L., 532 Brown, W.,139, 142(19) Brown, W. E., 189, 190(143) Browne, M. J., 528, 529(95), 530(157), 535 Browne, S., 99 Brownlee, G. G . , 314, 520(48), 521, 530(48), 540(48), 546(48) Broylan, R. C., 107 Bruce, S., 161(26), 162(26), 165 Brun, G . , 80 Brutlag, D., 16, 28, 59, 117, 209, 217(16), 221(16), 222(16), 223(16), 333, 335(15), 360, 361(60) Brynolf, K., 586 Buchanan-Wollaston, V.,537, 539(194) Biichi, H., 314 Budovskii, E. I., 578, 579(40) Budowsky, E. I., 550 Buell, G. N., 99 Buhler, R., 144 Buhrer, D. P., 68 Bukhari, A. I., 541 Bulkacz, J., 143 Bull, H. G., 574 Bullas, L., 141, 143(27), 144 Biinemann, H . , 180, 520 Buraczynska, M., 472, 480(6) Burchard, A. C., 434, 435(174) Burckhardt, J., 145, 152(57), 153(57), 154(57), 153571, 525, 527(85) Burd, J. F., 115, 175 Burdon, R. H . , 5 18, 528, 529(95, 96), 530(157), 532(7), 533, 535, 543 Burgers, P., 46 Burgers, P. M. J., 5

AUTHOR INDEX Burgess, R. R., 381, 382(25), 388 Burgi, E., 362 Burk, R. L., 56, @(lo) Burke, R. L.,9, 11(16), 12(16),367, 370(11), 381, 383(56), 388, 394, 408, 409(22, 97), 410(22, 72, 98) Burny, A., 88(16), 89, 93(16) Burrell, A., 30 Bunington, M. G.. 334 Burton, W. G., 519, 532(16), 543(16), 544(16) Buryanov, Y. I., 190, 520, 523, 524(67, 68) Biisen, W., 90 Bustamante, C., 289 Buttin, G., 234, 235(5), 240, 241(8, 51), 242(8), 244(8), 245(8, 51), 448, 468(33) Caillet-Fauquet, P., 72 Cairns, J., 29, 40, 362 Calva, E., 388 Camerini-Otero, R. D., 335 Cameron, V., 302, 303(20), 305(19), 311(19, 20), 312(19, 20) Campagnari, F., 566(8), 567, 576(8) Campbell, A. M., 471, 472, 473, 477(9) Cannellakis, E. S., 106 Cantor, H., 106, 118(14) Capaldo, F. N., 237 Capaldo-Kimball, F. N., 34, 237 Caradonna, S. J., 566(7),567, 575(7), 577(7) Caravana, C., 184 Carbon, J., 117, 573 Carestia, C., 541, 546(207) Carey, N. H., 99 Carins, J., 562 Carl, P. L., 238 Carlson, J. O., 362 Carlsson, L., 294, 296(59) Carrier, W. L., 258, 259, 274, 484, 485, 494(23), 495, 496, 498(23), 512(65) Carroll, R. B., 392 Carmthers, M. H., 547, 548(255) Canithers, M. H., 314, 582 Carter, D. M., 231 Cartwright, E. M., 520(48), 521, 530(48), 540(48), 546(48) Cascino, A., 408 Caskey, C. T., 560 Castellazzi, M., 448,468(33) Castora, F. J., 346(7), 347

59 1 Cathcart, G. R., 214, 223(41) Catherine, M. C., 175 Catley, B. J., 286 Cato, A. C. B., 530(157), 535, 543 Catterall, J. F., 160(16), 165 Cavalier, S. J., 413, 414(122) Cavalieri, L. F., 89 Cedar, H., 530(160), 533, 535, 548(140) Center, M. S., 242 Cerami, A., 28, 314 Cerutti, P., 263, 550 Cesareni, G., 480 Ceulen, M. T. E., 244 Chaconas, G., 305, 308(30,48), 309(30), 314 Challberg, M. D., 6, 84, 85 Chalmers, K., 10, 16 Chamberlin, M., 36, 175 Chamberlin, M. J., 4, 65, 490, 495, 511(50) Chambon, P., 4, 284, 344, 362, 530(161, 162), 535, 546(161, 162) Champoux, J. J., 333, 334(12), 335, 337(12), 338, 340, 343, 346, 432 Chan, V . L., 311 Chanal, M. C., 250, 367 Chandra, P., 506 Chang, A. C. Y.,101 Chang, H.-C., 292 Chang, J. C., 116 Chang, L. M. S . , 20,68,73,74(31), 106, 107, 108(22), 113, 114(22, 71), 118(16) Chang, S., 142 Chargaff, E., 282, 286(5), 518, 520(4), 532(135), 533 Charlier, M., 382, 397(34), 504, 506 Charter, K., 161(43), 165(43), 166 Chartrand, P., 83 Chase, J. W., 65, 226, 227(72), 228(73), 280, 383(42), 384, 388, 392(42), 430(42), 582 Chattoraj, D. K., 479 Chaurcey, T. R., 161(22), 162(22, 4 3 , 165(45), 166(22) Chen, I., 507 Chen, P. L., 238 Chen, Y. C., 69, 72, 83(22) Chen, S. M., 79 Cheng, Y.C., 566(7), 567, 575(7), 577(7) Chestukhin, A. V . , 234, 235(10), 237, 238(36), 240(36), 241(10, 36), 242(36), 243(36), 244(36) Chetsanga, C. F., 566(13), 567, 568(13),

AUTHOR INDEX 570(13), 571(13), 576(13), 577(13), 579(13), 582(13) Chevalley, R., 52, 55(4), 57(4), 360, 392, 408(57a) Chevallier, M. R., 235, 236 Chiang, K.-S., 545 Chiang, T., 492, SOO(56) Chick, W. L., 99 Chien, J., 581, 584(45), 585(45) Chien, J. R., 18 Chirgwin, J., 99 Chirikjian, J. G., 68, 159, 160(15), 163(61), 165(61), 166(5), 172(15), 174, 182(61), 183, 188, 334 Choi, T., 23 Chou, P., 395 Chow, L., 139, 142(19) Chow, L. T., 439, 520, 524(41), 541141) Chow, T. Y.-K., 129, 198, 234 Christiansen, C., 429, 456 Christman, J. K.,546 Chu, A., 371 Chu, H., 46 Church, G . M., 161(25), 165 Church, R. B., 305, 308(30), 309(30) Ciarrocchi, G., 76, 327, 329(21, 22) Cina, J., 381, 382(25) Cintron, N. M., 221, 222(57), 223(57) Citarella, R. V., 84, 85(71) Clanton, D. J., 161(30), 165, 183 Clark, A. J., 209, 210, 229, 230(84, 8 5 ) , 231(19, 84). 234, 235(7), 241(13), 242(13), 244(13), 245(13), 246(13), 446, 466(1, 3, 8) Clark, L., 117 Clarke, C. M., 172 Clarke, L., 573 Clarke, N. D., 275, 276 Cleaver, J. E., 492, 508(59) Clements, J., 214, 220(46), 222(46) Clements, J. E., 266, 570, 572(20), 585(20) Clerici, L., 566(8), 567, 576(8) Clough, W., 543 Cobianchi, F., 432, 436 Cohen, G . , 470 Cohen, G. L., 184 Cohen, J. C. C., 530(169), 535, 536(169), 546( 169) Cohen, L. K.,161(21), 165

Cohen, S., 142 Cohen, S. N., 101 Cohen, S. S., 537, 538(187) Cohn, M., 502 Cohn, R. H., 441 Colain, A. A., 296 Colcher, D.,94 Cole, R. D., 361 Cole, R. S., 276 Coleman, J. E., 382, 393(27), 395(27), 396(27), 397(27), 407(27), 414, 415(127), 417(27), 418(127), 419(127, 129), 420(129, 135), 421(129), 424(27), 426(27), 432 Coleman, M. S., 106, 107(29), 108, I 1 1(43), 1 l4(63), 1 l8( 17) Collett, M. S., 88, 92 Collins, E., 141 Colson, A., 141, 143(25, 27) Colson, C., 141, 143(25), 144(28) Comings, D. E., 545 Cone, R., 269, 272(78, 79), 566(6), 567, 570(6), 575(6), 576(6) Conlon, S., 458 Cook, J. S., 485, 488(44), 489, 492(25), 494(24), 508(57) Cook, K.H., 277 Cook, P. R., 366 Cook, T. M., 350 Cooley, W., 301, 302(11), 305(11), 311(11), 312(11), 313(11) Cooper, P., 30 Cooper, P. K.,132, 234, 276(1), 446 Cordes, E. G., 574 Cordis, G. A., 317 Cosloy, S. P., 485 Coughlan, M. P., 234, 241(3) Coulondre, C., 541, 580 Coulson, A. R., 36, 336 Courschesne, W., 473, 474(19), 475(19) Court, D.,490, 511(50) Cousens, L., 531, 541(120) Coutsogeorgopoulos, C., 106 Couturier, M., 242 Cox, M. M., 447, 452(30), 457 Cozzarelli, N., 154 Cozzarelli, N. R., 16, 28, 29, 30(31), 110, 115, 128,311,312(37), 313(37), 314,339, 340(62), 341(62), 346, 348, 349(23, 24),

AUTHOR INDEX 350(23), 35 l(23, 24), 352(23, 24, 25, 37), 353(25, 37), 354(37, 4 9 , 355(44), 356, 358(37, 55). 359(25), 363, 366(25) Craig, L. M., 518, 528, 529(96), 532(7) Craig, N. L., 446(22, 23, 26), 447, 448(23), 454(23, 26), 463(22, 23, 26) Crapo, L., 384, 392(43) Crathorn, A. R., 545 Crawford, C. V., 548 Crawford, L. V., 348 Crick, F. H. C., 296, 339, 347 Crine, P., 213, 214(32), 264, 272(53,55), 551 Crouch, R., 91,212, 218 Crow, W., 41,43(12) Crumpacker, C. S., 83 Csordas-Toth, E., 161(32), 165 Csuzi, S., 234,239, 241(50), 244(50), 24350) Cuatrecasas, P., 292, 293(51) Culp, L. A., 545 Cummings, D. J., 532, 532(127) Cunningham, R. P., 250, 431, 446(20), 447, 454, 456, 458(20), 459(42), 460(42, 50, 51), 461(42, 47) Curtis, M. J., 411 Cussuto, E., 458 Cuypers, T., 413 Czernilofsky, A. P., 100 Dahlquist, F. W., 408 Dahlberg, J. E., 88, 90,96 D’Alarcao, M., 540 Dalpra, L., 76, 327, 329(21, 22) Darnell, J. E., Jr., 296 Das, R. H., 369, 370, 371(25), 372(29), 442 Das, S . K., 60,61 Das Gupta, C., 431, 446(20), 447, 454, 456, 458(20), 459(42), 460(42,50, 51), 461(42, 47) DasGupta, C., 250 Datta, B., 545 Datta, R. K., 545 D’Aurora, V., 17 Davidson, N., 441, 456 Davies, D. J. G., 508 Davies, J., 161(41), 165(41), 166 Davies, M., 409, 410(103) Davies, R. W., 336, 473, 474 475(16) Davis, R. W., 179 Dawid, I. B., 530(151), 532, 534

593 Day, L. A., 382, 413(35), 414(29), 415(29, 35),418(35) Deckstein, F., 5 Dean, F. B., 9, 63 Dean, M. D., 338 Debov, S. S., 523, 524(69), 537, 538(182) Defais, M., 72, 562 deHaseth, P. L., 380, 381(16), 382(16), 399(15) Deitz, W. H., 350 de Jong, W. W., 413 de la Blanchardiere, 122 De Laforteyne, J . , 242 De La Tour, E. B., 52, 5341, 57(4) De Leys, R. J., 335 Delius, H., 383(41), 384, 392(41), 396, 408, 412(41), 413(41), 414(41), 415(41), 4 I6(4l), 422(41), 423(96), 424(96), 425(96), 426(79, 96), 427(96), 428(79, 96), 429(96), 430(%), 434, 435(173) Dellweg, H., 483, 506 De Lucia, P., 40 De Lucia, P. L., 29 Demple, B., 126, 128, 257, 258(26), 263, 268(26, 46), 272(45, 46), 562, 566(14), 567, 568(14), 570(14), 571(14), 572(14), 579(14), 582(14), 583(55), 584(14) Denhardt, D., 52, 55(4), 57(4) Denhardt, D. T., 125, 199, 368, 371, 392, 408(57a) Denhardt, G., 360 Denson, J., 487, 488(39), 494, 501(62) De Pamphilis, M. L., 73, 79 Depew, R. E., 311, 312(37), 313(37), 334, 337(19), 340(51), 347 Derankau, D. A., 507 Desai, L. S., 545 Desiderio, S . V., 6 De Simone, P., 106 Desrosiers, R. C., 530(166), 535, 536(166) Detera, S. D., 439 Deutsch, J., 520 Deutsch, W., 572, 583(55), 584 Deutsch, W. A., 257, 258(26), 263, 268(26, 46), 272(46), 551, 554(19), 55319, 22), 556(19,22), 557(22), 558(19,22), 559(22) Deutscher, M. P., 16,29,30(31), 36,83,203, 314, 317 Devoret, R., 448, 468(34)

594 De Waard, A., 302 de Waard, A., 52, 57(2), 62(2), 520, 524(27) Dhar, R., 96 Diamond, J., 508, 509(103) Dickerson, R. E., 188 Dieckrnann, M., 35 Diebel, M. R., 108, 111(43) Dina, D., 95 Di Nardo, S., 341 Dion, A. S., 94 Dobritsa, A. P., 520, 524(38) Dodgson, J. B., 115, 165, 175(27), 176(27), 185(27), 186(27) Doeffler, W., 530(167, 168), 535, 536(167, 168, 170), 545(173), 546(167, 168, 170) Doly, J., 235, 237, 238, 244(35) Donch, J., 430 Donnelson, J. E., 117 Donelson, J. E., 217, 221(53), 22353) Doniger, J., 468 Donis-Keller, H., 314 Donlon, J. A., 107 Doring, H. P., 528, 529(94), 536 Doskocil, J., 518, 520, 529 Douglas, J. T., 533 Douvas, A., 290 Dow, L., 99 Downey, K. M., 72, 83(23) Drahovsky, D., 528, 529(101), 534 Draper, D. E., 381, 382(24) Draper, D. M., 380 Dreiseikelrnann, B., 180, 181(113) Dressler, D., 468 Drlica, K., 362, 363 Drost, S. D., 91 Dubnau, E., 342 Duesberg, P. H., 89 Dugaiczyk, A., 180, 181(117), 520(42, 43). 521, 524(41, 42, 43), 541(41) Duguet, M., 369(26), 370, 442 Duker, N. J., 551 Dulbecco, R., 333, 334(12), 337(12), 500 Duncan, B., 550, 562(2) Duncan, B. K., 541, 570, 571, 573(24), 578(44), 580(22, 23, 41), 581(22, 23), 584(18, 451, 585(45) Duncan, J., 277, 566(6), 567, 570(6), 575(6), 576(6) Dunher, A. K., 414 Dunlap, B . , 550

AUTHOR INDEX Dunn, D. B., 518 Dunn, K., 441 Durnford, J. M., 335 Durphy, M., 253, 254(11), 257(11) Diirwaid, H., 367, 368, 442 Dussoix, D., 138, 156(5) Duvoisin, R. M., 41 1 Dwyer-Hallquist, P., 168 Dynan, W. S., 335(39), 336 Early, T. A., 115 Easterbrook-Smith, S . B., 354 Eastman, E. M., 520, 533(25) Ebashi, S., 295 Ebbers, W., 148, 151(68) Eberle, H., 151 Ebisuzaki, K., 311 Echols, H., 242, 336, 364, 473, 474(19), 475(16, 19), 479 Edenberg, H. J., 73 Edgar, R. S . , 52, 55(4), 57(4), 360, 392, 408(57a) Edgell, M., 147 Edgell, M. H., 523, 524(63, 65), 525(65) Edmiston, S. H., 447, 463(29), 469(29) Efshatiadis, A., 117 Efstratiadis, A . , 99, 101 Ehrenberg, M., 179 Ehrlich, K., 531, 532 Ehrlich, M., 160(17), 165, 531, 532, 537, 539(190) Ehrlich, S. D., 530 Eichenlaub, R., 180, 181(113) Eichler, D. C . , 30, 32(38), 241, 246(59) Eick, D., 535, 536(170), 546(170) Eigen, M., 407 Eisenberg, S., 332, 369,424, 429(144) Eisenstadt, J. M., 532(136), 533 Eker, A. P. M., 486, 512(30) Elad, D., 278, 504, 505(92), 551, 552(20), 555(20), 556(20), 557(20), 558(20) Elder, R. L., 258 Eldridge, J. D., 296 Eliasson, R., 6, 586 Ellers, D. J., 34 Emrnerson, P. T., 446(16), 447 Emtage, J. S., 99 Endlich, B., 125, 128(15), 133(15), 149, 150(71), 151(71), 152(71), 153(71, 79), 154(71, 79), 155(71), 246, 250

AUTHOR INDEX Endow, S. A., 161(42), 163(42),165(42),166 Engel, J. D., 542 Engle, E., 363 Engler, M., 28 Engler, M. J., 9 Englss, M. J., 63 Englund, P. T., 16, 17(5),53,84,85,?15,366 Enomoto, T., 435 Enquist, L. W., 95, 96, 472, 476, 478(28), 479, 480(6) Ensinger, M. J., 439 Epstein, I. R., 381 Epstein, R. H., 52, 534). 57(4), 360, 384, 392(44), 408(57a), 41 l(44) Epstein, W., 446 Eriksson, B., 82 Eriksson, S., 294 Erlanger, B. F., 520, 533(21, 22, 23, 24, 25) Erlich, H. A . , 101 Errera, M., 482 Eskin, B., 144, 145, 146(52, 59), 147(52), 148(52, 59), 149(52), 150(52, 59), 151(59), 153(59), 155(59, 69), 525 Evans, D. H . , 348 Evans, H. E., 545 Evans, M. J., 73 Evans, R . , 230 Evans, T . E., 545 Evensen, G. B., 573, 578(30), 579(30) Fahrney, D. E., 284 Fairley, J. L., 194, 195(11),200(11) Fairweather, N. F., 350, 363 Falaschi, A . , 76, 327, 329(21, 22), 432, 436 Falco, S. C., 365 Fan, H . , 94, 99, 100 Fancher, H., 36, 175 Fanning, T . G., 184 Farabaugh, P. J., 541, 580 Faras, A . J., 88(19), 89, 90, 92(6) Farber, M. B., 160(17), 165 Fareed, G. C., 302, 303, 308(22) Farhang, P., 99 Farkas, W.R., 560 Farland, W. H., 496 Farnham, G., 545 Fasman, G., 395 Fasman, G. D., 576 Fay, P., 7, 48 Fedchenkov, V. I., 241

595 Fedoroff, N . V., 520(48), 521, 530(48), 540(48), 546(48) Fedorov, N. A . , 545 Feix, G., 109, llO(57) Feldman, R. J., 19 Felgenhauer, Z. Z., 392, 408(59), 410(59) Feller, W.G., 68 Felsenfeld, G., 295, 335, 354, 355, 362, 533 Fems, R. L., 382, 383(38), 389(38), 392(38), 393(38), 394(38), 396(38), 408(38) Fichtinger-Schepman, A., 486, 512(30) Fiddes, J. C., 336 Finnegan, D. J., 117 Fischauf, A . , 387 Fischer, R., 109, 473, 474(19), 475(19) Fischinger, P. J., 95 Fisher, E. F., 547, 548'(255), 582 Fisher, L. M., 352,353(39), 354(39), 355(39), 356(53), 357(53), 359(38), 366(39), 477 Fisher, P. A . , 5, 69, 71, 72, 75(18), 79(18), 82(21) Fitzgerald, P., 413, 414(126), 417(126), 421(126), 422(126) Flangan, J. B., 12 Flavell, R. A., 180, 181(109), 528, 529(98), 530(98, 158, 158a), 535(98), 546(98, 158, 158a) Fleckenstein, B., 530(166), 535, 536(166) Fleischman, R. A . , 231, 314 Fleissner, E., 88(15), 89 Flint, S. J., 548 Flory, P. J., 113 Fliigel, R., 107(47), 109 Foeller, C., 472, 474, 480(6) Foley, G. E., 545 Folk, W. R . , 167, 168(74), 178, 179(104), 301. 309(14), 310(14), 526, 544(90) Forsblum, S., 179 Forterre, P., 356 Fowlkes, D. M., 440 Frank, R., 177, 182(100), 186(100) Frankel, A . D., 570, 584(18) Fraser, M. J., 129, 196, 197(13), 198, 199, 234, 534 Frenkel, G. D., 61 Fresco, J. R . , 435, 436(176) Frey, L., 55, 376, 382(6), 383(41), 384(6), 392(6, 41), 393(6), 396, 402(6), 405(6), 406(6), 410(6), 412(41), 413(41), 414(41), 415(41), 416(41), 422(41), 456

596 Fridlender, B.. 68 Friedberg, E., 125, 126(14), 132(14) Friedberg, E. C., 213, 214, 220(38), 253, 254(7, I I ) , 255, 256(5, 241, 257(11, 2 3 , 263, 269, 272(48, 49, 50, 78, 79, 79a). 277, 492, 508(59), 566(6, 15), 567, 568(15), 570(6, 15), 571(15), 575(6), 576(6), 577(4), 581 Friedman, D. I., 477 Friedman, E. A., 204, 234, 238, 240, 241(9), 242(9), 243(73), 244(9), 245(52), 246(79) Friedman, J., 531, 542(109) Friedman, S., 424. 425(151), 429(151) Friedmann, T., 336 Friedrich, R., 100 Friefeld, B . , 469 Friis, R. R., 94 Frouin, A., 290 Fry, K., 117, 534 Fry, K. E., 410 Fry, M., 68 Fujimoto, D., 523 Fujimoto, M., 194, 1935, 6) Fujimura, R. K., 60, 61, 62 Fujiyoshi, T., 242, 246(70) Fuller, F. B., 339, 347, 356 Funakoshi, A. , 285 Fuselier, C. O., 490, 491(51) Futcher, A. B., 269, 270(70) Gafurov, N. N., 234 Galas, D. J., 355 Gall, J., 534 Gallo, R. C., 67, 79, 88, 89, 93, 94, 106(32), 107, 108(33), 11 l(32, 331, 114(33) Gallop, P. M., 537, 539(191) Ganesan, A . K., 132, 234, 253, 254(14), 257(16), 276(1), 446, 581 Ganesan, A. T., 132, 531 Gardner, J. F., 161(21), 165 Garel, A . , 295, 548 Garfin, D. E., 167, 168(75), 172, 173(81), 176(75), 186(75), 523, 524(61), 526(61) Garrett, C. E., 570, 575, 580(23), 581(23) Garssen, G. J . , 419, 420(134, 136, 137). 42l( 134) Gates, F. T . , 126, 571 Gates, F. T., 111, 262, 272(44, 99),277, 278 Gatlin, G. H . , 99 Gautier, F., 180, 520

AUTHOR INDEX Gefter, M., 40, 41(6), 46(7), 208, 382, 383(161), 408, 423(96), 424(96, 1411, 425(36, 96, 151), 426(36, 96), 427(36, 96), 428(36, 96). 429(96, 151), 430(96, 156). 431, 432(161), 442 Gefter, M. L., 368, 369(26), 370, 371(25), 372(29) Geider, K., 43, 46(15), 49, 332, 333(6), 336(6), 338(6), 422, 423(139), 429(139) Geiduschek, E. P., 408 Geier, G. E., 520(47), 521, 524(47) Gelderblom, G . , 94 Gelinas, R. E., 160(10, 351, 163(10), 165 Gellert, M., 154, 230, 333, 334, 339(18), 340(18), 341(18), 346, 348, 349(22), 350, 351(22, 29, 30, 31), 352(31), 353(22, 391, 354(22, 39), 355(39), 356(53), 357(53), 359(38), 362(30, 31), 363(3, 29, 311, 365(29), 366(39), 476, 477, 478(31), 480(3 I ) George, J., 174, 448, 468(33) Gerard, G. F., 88(18), 89(13), 92(13), 94(13) Gerard, P., 290 Germond, J. E., 199, 344, 362 Gerrard, S . P., 340 Gesteland, R. F., 439 Gibson, W., 93 Gih, A . , 496, 503(69) Gil, A., 289 Gilbert, W., 99, 100, 116, 314, 384, 392(43), 541, 580, 585 Gilboa, E., 95, 98 Gilead, Z., 439, 440(193) Gillen, J. R., 229, 230(85) Gillham, N. C., 545 Gillham, N. W., 544 Gillin, F. D., 28, 59 Ginder, G. D., 530(159), 535, 546(159) Gingeras, T . R., 162(52). 165(52), 166 Ginsberg, H. S., 439 Giulotto, E., 76, 327, 329(21) Glaser, D., 46 Glaser, G., 530(160), 531, 535, 540(111), 542(1 1I ) Glassberg, J., 388, 424, 425(149), 429(149), 430( 148) Glaubiger, D., 339 Glickman, B. W., 34, 468, 540 Glover, S., 143, 144, 150(51) Goddard, J. G., 96, 98(77), lOO(77)

AUTHOR INDEX Goddard, J. M., 532, 533(127) Godson, G., 147 Godson, G. N., 124, 163(68), 165(68), 166, 186(68) Goebel, W., 133 Goff, S., 95, 98(74) Goff, S. P., 229 Gold, A. M., 284 Gold, L., 384, 385(45), 392(45), 411(45, 46) Gold, M., 521, 522(53), 523(53), 531 Goldberg, A. R., 294, 296(58) Goldblatt, P. J., 317 Goldmark, P., 150, 155 Goldmark, P. J., 135, 198, 238, 240(43), 241(43), 242(43, 53), 243(53), 244(43, 72), 245(43, 72), 246(43, 53), 367 Goldschneider, I., 106, 114, 118(20) Goldstein, A. L., 106, 118(15) Goldstein, B., 545 Goldstein, G., 106, 118(14) Goldthwait, D. A., 213, 214, 220(38), 223(41), 263, 264, 269, 270(73), 272(48, 49, 50, 511, 392, 413, 414(122), 551 Golomb, M., 89, 92 Gomez-Eichelmann, M. C., 521, 525(51) Gonclaves, J. M., 109, 110(46), 111(46), 112(46), 114(46), 175 Goodgal, S., 141 Goodgal, S. H., 214, 482, 489 Goodman, H., 139, 142(19) Goodman, H. M., 90, 99, 100, 167, 168(75), 172, 173(81), 176(75), 180, 181(117), 186(75), 520(42, 43), 521, 523, 524(41, 42, 43, 61). 526(61), 541(41) Goodman, M. F., 59 Goodman, R. M., 520, 533(25) Goodman, T. C., 159, 183(1) Goody, R. S., 294, 295(60) Goppelt, M., 167, 168(73), 169(73), 177, 182(73, loo), 183(73), 186(73, 100) Gorban, I. M ., 296 Gordon, L. K., 128, 255, 256(23), 259(23), 260(23), 261(23), 566(16), 567, 570(16), 572(16), 585( 16) Gordon, M. P., 498, 499(75), 507 Gorecki, M., 91, 284 Goren, D., 531, 542(109) Gorovsky, M. A . , 532, 533(126) Goss, W. A., 350 Gossard, F., 214, 219(44), 265, 272(60), 551

597 Gottehrer, A., 520, 524(28), 544(28) Gottesfeld, J. M., 548 Gottesman, M. E., 106, 230, 472, 478(11), 479 Gottesman, M. M., 230 Gottesman, S., 230, 472, 478(11) Gottlieb, A., 68 Gottlieb, P. D., 111, 114(64) Goulian, M., 52, 53, 54( I), 55( I), 56( I ) , 303 Grabowy, C. T., 519, 532(16), 543(16), 544( 16) Grafstrom, R. H., 128,255,256(23), 259(23), 260(23), 261(23), 566(16), 567, 570(16), 572(16), 585(16) Graham, D. R., 17 Grandgenett, D. P., 88(18), 89(13), 91(5), 92(13), 93, 94(13) Grant, D. M., 544, 545 Gray, D. M., 118 Grazi, E., 295 Green, H., 437 Green, L., 479 Green, M., 88(18), 89, 91, 93, 439, 440(193) Green, M. H. L., 430 Greenberg, B., 127, 132, 180, 181(115), 521, 524(49), 541(49) Greenberg, G. R., 12 Greenberg, H., 214 Greenberg, J., 430 Greene, P. J., 167, 168(75), 172, 173(81), 176, 186(75), 188, 189, 190(143), 523, 524(61), 526(61) Greenwood, M. F., 106, 107(29) Gregoire, K. E., 114 Grepachevsky, A. A., 237, 238(36), 240(36), 241(36), 242(36), 243(36), 244(36) Greth, M. L., 235, 236 Greve, J., 382, 394, 403(68) Gribbin, C., 531, 542(110) Griffin, K.P., 507 Griffith, J. D., 16, 49, 332, 418, 441 Grindley, N. D. F., 34 Grinsted, J., 162(55), 165(55), 166, 169(55), 171(55), 172(55) Grippo, P., 62, 63(29), 64(29), 530, 541, 54%107), 546(207) Griswold, J., 275, 276 Groeniger, E., 106 Groffen, J., 528, 529(98), 530(98), 535(98), 546(98)

AUTHOR INDEX Gromkova, R., 141 Gross, J. D., 238 Gross-Bellard, M., 344, 362 Grossman, L., 562, 566(16), 567, 570(16), 572(16), 585(16) Grossman, L., 128, 159, 166(4), 255, 256(23), 259(23), 260(23), 261(23), 274 Grotjahn, L., 180, 520 Groudine, M., 295, 548 Gruenbaum, Y.,531, 540(111), 542(111) Grunberger, D., 279 Grunert, F., 528, 529(94) Grzesiuk, E., 473, 474(19), 47319) Gudas, L. J., 446, 470 Guijt, N., 468 Guilley, H., 314 Gumport, R. I., 175, 335, 356, 358(57) Guntaka, R. V., 95 Gunther, J. K., 214 Giinthert, U., 520, 524(30), 536, 542(114), 545(173) Gupta, M., 189, 190(143) Gupta, N. K., 314 Guskova, L. V., 546 Gutmann, E. D., 382, 383(38), 389(38), 392(38), 393(38), 394(38), 396(38), 408(38) Habara, A., 79 Haberling, R. L., 94 Haberman, A., 145, 150(56), 525, 527(79) Hack, A., 46 Hadi, S., 140, 144, 145(53) Hadi, S. M., 213, 214, 220(38), 263, 264, 272(51), 521(62), 523, 524(62) Hagen, D., 473 Hagenbiichle, O., 173 Haines, M. E., 68 Halford, S. E., 162(55), 165(55), 166, 169(55), 171(55), 172(55) Hall, R., 46, 238 Hama-Inaba, H., 52, 409, 410(103) Hamilton, D., 145, 152(57),153(57), 154(57), 155(57), 474 Hamilton, L., 566(6), 567, 570(6), 575(6), 576(6) Hanawalt, P., 30 Hanawalt, P. C., 132, 251, 253, 276(1), 446, 469, 532(137), 533 Handberg, F. A., 162(52), 165(52), 166

Hansen, F. G., 350, 542 Hansbury, E., 109, 110(49), 113(49), 115 Harada, F., 88, 90, 96, 560 Harbers, B., 520, 534(37) Harbers, K., 520, 534(37) Harm, H., 493, 494, 501(61), 508, 509(104), 515(104)

Harm, W., 253, 485, 493(29), 498, 500, 502(88), 508(29, 108), 509(87), 510(108), 51 l(29), 515(87) Harper, D. J., 238 Harris, J. I., 162(51), 165(51), 166 Hamson, T. A., 106(30), 107(19), 108(19), 111(19), 114(19,23, 30), 118(19, 23) Hartley, B. S., 172, 175 Hartley, J. L., 190 Hartman, P., 143 Hartman, P. E., 570, 584(18) Haseltine, W. A., 88, 90,96, 100, 128, 255, 256(22), 259(22), 260, 261(22), 562, 566(16), 567, 570(16), 572(16), 585(16) Hastings, R., 482 Hatano, S., 295 Hatch, F., 534 Hattman, S., 138, 520(45), 521, 523, 524(27, 28,29,36,45,50,66), 525(29,66,72,73, 74, 7 9 , 526(75), 527(29, 72-75), 528, 529(100), 531(36), 532, 533(126, 1301, 536, 537, 538(66, 180), 539(194), 541(36, 45, 66, 119, 120, 180), 542(110), 542(125), 544(28), 548(130) Hausen, P., 90 Hausler, B., 189 Hausmann, R., 523 Hausrath, S. G., 511, 514 Hay, J., 80, 106, 536 Hayashi, H., 537, 539(191) Hayashi, M., 364 Hayashi, Y.,364 Hayes, F. N., 107(66), 109, 110(49), 112, 113, 115 Hays, J. B., 366, 584 Hearst, J. E., 339, 409, 439, 442(99, 185) Heasley, S., 98 Hecht, R., 334, 362 Hedgepeth, J., 180, 181(117), 520(43), 521, 524(41, 42). 541(41) Hehlmann, R., 523, 525(74), 527(74) Heijneker, H. L., 34, 244 Heininger, K., 174, 175(87)

AUTHOR INDEX Helene, C., 382, 397(34), 405, 407, 504, 506 Helinski, D. R., 29, 133, 364 Helgstrand, E., 82 Hell. A., 99 Heller, K., 365 Henderson, E. E., 515 Hendler, R. W., 238 Hendrickson, H . E., 61 Hennings, H., 549, 561 Henningsen, I., 16 Henry, T. J., 336, 412, 413(119) Henson, J., 46 Her-manns, U., 241 Herrick, G., 387, 434(49), 435(49, 172, 173), 436( 172) Hemott, R. M., 482, 489, 495 Hemott, S. T., 60 Hershfield, J., 139, 142(19) Hershfield, M. S . , 56, 83 Herskowitz, I., 473 Hertman, I . , 446, 466(9) Herzberg, M., 520, 528(32), 534(32) Hesse, J. E., 446 Hewitt, R. R., 285 Heywood, J., 525, 527(79) Heywood, J., 145. 150(56) Hibner, U.,9, I I ( 161, 12(16), 56,60( lo), 367, 370(11), 381, 408, 409(22, 97), 410(22) Higashinakagawa, T., 327 Higgins, N., 154 Higgins, N. P., 348,349(24), 351(24), 352(24, 37), 353(37), 354(37), 355(44), 358(37) Higuchi, R., 117 Hilbers, C. W., 419, 420(134, 136, 137), 421( 134) Hillen, W., 159, 183(1) Hillenbrand, G., 11, 63 Hinch, B., 160(14), 165, 167, 168(76, 77), 169(76, 77), 183(76), 186(76) Hines, J. L., 161(22, 23), 162(22, 23, 44), 165(44), 166(22), 185 Hinkle, D., 41 Hinman, L., 507 Hinton, D. M., 175 Hirahashi. T., 239,240(47), 241(47), 242(47), 244(47), 245(60) Hirose, S., 310 Hirota, Y.,40, 46(7), 341, 542 Hirt, B., 344, 362 Hitchock, S. E., 294

599 Hizi, A., 93 Hoard, D. E., 107(66), 112 Hobus, P. A., 284 Hochhauser, S. J., 221,222(57), 223(57), 224 Hoekstra, W. P. M., 468 Hoess, R. H., 336, 474 Hoffman, S., 282 Hoffmann-Berling. H., 8, 250, 367, 368,442 Hogness, D. S., 117 Holland, I. B., 350, 363 Holland, P., 106, 107(29) Holliday, R., 194, 195(7), 198(7), 547, 548(253) Holloman, W. K., 194, 195(7,8), 198(7), 468 Holloway, B., 141 Holmes, A. M.,68, 69, 72, 73(15) Holmes, D. S., 441 Hong, S-C. L., 80 Honigman, A., 473 Hori, K., 62, 63(34), 64(34), 431 Horii, T., 446(18), 447, 449(18), 454(18) Horiuchi, K., 139, 147, 150, 155(74), 523, 524(64), 525, 527(82) Horwitz, M. S., 439, 440 Horn, D., 179 Horn, G. T., 159, 183(l) Horz, W., 174, 175(87) Hosoda, J., 382, 393, 394(66, 67), 402(70), 403(66, 67), 404(70, 71), 405(67, 71), 408, 410(72) Hotta, Y.,433 Hours, C., 371 Houts, G. E., 109, I10(46), 11 1(46), 112(46), 114(46), 175 Hoviuchi, K., 185, 186(131) Howard-Flanders, P., 222, 235, 236, 274, 446(24), 447, 449, 458, 466(4), 468(24), 484, 561 Hozer, T. J., 570, 580(22), 581(22) Hozumi, T., 79 Hsiang, M. W., 361 Hsieh, T.-S., 333, 335(15), 360, 361(60) Hsu, M.,172, 173(82) HSU,P.-L., 472, 473, 475(4), 480(4, 6) Hsu, T. W., 95, 98 Huang, C. C., 409, 439, 442(99, 185) Huang, C. W., 507 Huang, R. C., 68 Huang, T. C., 531 Huang, W. M.,56, 57, 58, 59(15), 360

600 Hubacek, T., 144, 150(51) Huberman, J. A., 16, 28, 73, 408, 584 Hiibscher, U., 10, 44, 45, 46(18, 19), 76, 79(40) Hufnagel, D. A,, 532(135), 533 Hughes, S. G., 161(26), 162(26), 165. 531 Hughes, S. H., 95 Hugli, T. E., 283, 287, 288 Human, M., 138, 142(2) Human, M. L., 537 Hung, P.,117 Hunter, M. J., 182 Hurter, J., 498, 499 Hurwitz, J., 28,45,89,91,210,212,218(27), 240,241(51), 245(51), 300,301(3), 305(3, 41, 306(3, 41, 308(4), 310(31), 311(31). 312(3), 314(31), 315, 316(1), 318(1), 319(1), 320(1), 321(1), 326(1), 327(1, 5 ) , 327(1), 328(1, 5 ) , 329(5), 332, 333(4), 336(4), 340(4), 367, 369, 521, 522(53), 523(53) Hutchinson, C. A., III., 147, 162(51), 163(63), 165(51. 63), 166, 336, 523, 524(63, 65), 525(65), 531, 542(110) Hutton, J . J., 106, 107(29), 118(17) laccanno, M., 530, 545(107), 546 lannotti, A., 541 Ibayashi, H., 285 Ichimura, M., 300, 316, 317(8), 318(8), 319(8), 321(8), 326(8), 329(14) Idriss, J., 534 Ihle, J. N., 106, 118(13, 15, 21) Ijlstra, J., 483 Ikawa, S., 141, 142(31), 190 Ikeda, J.-F., 332, 333(4), 336(4), 340(4) Ikeda, H., 366 Ikeda, Y.,162(53), 165(53), 166 Ikegami, S., 73 Ikenaga, M., 499 Ikkai, T., 295 Illmensee, R., 95, 101 Imamoto, F., 365 Imber, R., 160(11), 165 Imoto, T., 574 Ineichen, K., 140, 521(62,85a), 523, 524(62, 85a). 525 Inoue, T., 269, 270(68) Inouye, M., 446 Irvin, J. L., 68

AUTHOR INDEX Irwin, C., 46 Irwin, V., 542 Isbell, A. F., 82 Ishimoto, M., 506 Itakura, K., 101 110, M., 254 Itoh, T., 350, 351(30, 31), 352(31), 362(30, 31), 363(31) Iwama, Y.,392, 410(58), 424 Iwatsuki, N., 488 Iwaya, M., 368 Jack, W. E., 167, 170(69a), 178(69a) Jackie, H., 499 Jackson, D. A., 116, 117(92), 335 Jackson, J., 546 Jacobsen, H., 18 Jacobsen, J. H., 344 Jacquemin-Sablon, A., 301, 303, 308(22) Jaenisch, R., 364 Jaffe, E. S., 107 Jagger, J., 486, 499(32), 503, 514(80) Jamjoom, G. A., 94 Jansz, H. S . , 440 Javaherian, K., 356, 358(57) Javaherian, K. J., 335 Jay, E., 114, 116(72), 117(72), 161(28), 165, 170, 171(79), 188, 205, 218(8), 225(8) Jedlicki, E., 242 Jeggo, P.,562 Jeng, Y.-H., 439, 440(193) Jenkins, M., 382, 383(38), 389(38), 392(38), 393(38), 394(38), 396(38), 408(38) Jenkins, M. D., 68 Jensen, D. E., 376, 380, 381(4), 382(4, 12, 23), 393(12), 396(4), 397(4, 12), 399(12, 23), 402(4), 403(4), 404(4), 405(12) Joe, C. O., 488 Johannssen, W., 161(36), 165 Johanson, K., 7, 44, 47(17), 48 Johansson, N. G., 82 Johnson, B. F., 424, 430(147) Johnson, D., 108 Johnson, L. K., 581 Johnson, L. N., 574 Johnson, N. P., 162(55), 16505). 166, 169(55), 171(55), 172(55) Johnston, 1. R., 68 Joklik, W. K., 83, 93 Jones, A. D., 265

AUTHOR INDEX

Jorgensen, S. E., 225 Jornvale, H., 439, 440(187) Jost, E., 431, 432(162) Jovin, T. M., 16, 17(5), 20.53, 115, 175, 183, 215, 387 Jungwirth, C., 83 Junker, A. K., 413 Junowicz, E., 290, 291 Jurnak, F., 413, 414(126), 416(124), 417( 126), 42 1(126), 422( 126) Jurnak, F. A., 413 Kabsch, W., 295 Kacian, D. L., 88(16), 89, 93(16), 99 Kacinski, B. M., 275, 276 Kada, T., 269, 270(68) Kadirova, D. Y., 518 Kado, C. I., 334 Kiifer, E., 198 Kafatos, F. C., 117 Kahmann, R., 541 Kaiser, A. D., 231, 303 Kaiser, A. O., 116, 117(93) Kaiser, K., 230, 231 Kaji, A,, 95 Kalinina, N. A,, 234, 235(10), 241(10) Kallen, R. G., 536, 539(179) Kalousek, F., 522, 523(57) Kalter, S. S., 94 Kalthoff, K., 499 Kamen, H. O., 106 Kamp, D., 474, 541 Kan, N., 147 Kan, N. C., 523, 524(63, 65), 525(65) Kan, Y. W., 116 Kane, C., 572, 583(55), 584 Kane, C. M., 257, 258(26), 268(26) 184 , Kania, .I. Kano,Y., 365 Kantor, J. A., 68 Kaplan, D. A., 178, 179(102), 526, 544(89) Kaplan, J . C . , 259 Kaplan, L. M., 439 Kaplan, S., 161(21), 165 Kappler, J. W., 532 Kaptein, R., 419, 420(136) Kaput, J., 530(163), 535 Karam, J. D., 58 Karamov, E. V., 172 Karlishev, A. V., 184

Karlostrom, O., 550 Karn, J., 480 Karpel, R. L., 434, 4331741, 436(176) Karran, P., 551, 566(9, lo), 567, 568(9), 570(7), 573, 576(10), 577(10), 578(30), 579(30), 582( 10) Kartenbeck, J., 368, 442 Karu, A., 150 Karu, A. E., 229, 230(85), 240, 242(53, 57), 243(53), 246(53, 57) Kastern, W. H., 296 Kates, J., 336 Kato, A. C., 199 Kato, K., 109, 110(46), 1 I1(46), 112(46), 114(46), 175 Katzir, N., 474 Kauc, L., 140, 141(21) Kaufman, R. J., 101 Kawaguchi, M., 90 Kaye, A. M., 519, 520, 528(32, 33), 529(33), 534(32, 33), 536, 539(171) Kearns, D. R., 115 Kedes, L. H., 441 Kedinger, C., 440 Kee, S. G., 117 Keech, D. B.. 354 Keegstra, W., 440 Keir, H. M., 80, 106, 536 Keister, T., 520, 524(28), 544(28) Kellenberger, E., 52, 55(4), 57(4), 360, 392, 408(57a) Keller, W., 91, 212, 218, 335, 344, 348 Kelley, W.S., 10, 16, 34 Kellin, R. A., 536, 539(178) Kelly, R. B., 16, 28, 29, 30(31), 110, 584 Kelly, R. C., 376, 380, 381(4), 382(4, 121, 393(12), 396(4), 397(4, 12, 33), 399(12), 402(4), 403(4), 404(4), 405(12) Kelly, T., 139 Kelly, T. J., 211 Kelner, A., 508, 509(103) Kennedy, W. J., 275 Kenney, C.. 532, 543(125) Kerr, C., 64 Kerr, V. N., 109, I 10(49), 112, 113(49) Khatoon, H., 541 Khodarve, N. N., 533, 548(142) Khorana, H . G., 115, 175,292,293(50), 301, 303(13), 304(13), 305, 309(29), 310(29), 314

602 Khudyakov, I. Y., 537, 538(186) Kikuchi, A., 477 Kikuchi, Y., 333, 336(9), 365. 473, 474(18), 475(15, 181, 476(24), 477(18, 24) Kim, S.-H., 188, 189 Kimball, M., 153, 520(42), 521, 524(42) King, C.T., 161(28), 165 King, G. J., 360 King, J. J., 253, 254(7), 256(5) Kingdon, H. S., 282 Kingsbury, D. T., 29 Kinross, J., 30, 34(39), 470 Kirnos, M. D., 532(134), 533, 537, 538(186) Kirkegaard, K., 334, 335, 337, 338, 340(53), 344(24), 349, 352(27), 353, 355(41), 356, 358(57) Kirtikar, D., 214, 223(41) Kirwan, J. P., 498, 499(75) Kiss, A., 161(32,33), 165,186(33), 187(33),190 Kit, s., 534 Kitchin, R., 547, 548(254) Kleid, D., 163(64), 165(64), 166 Kleid, D. G., 446(17), 447 Klein, H., 440 Klein, M.,382, 413(35), 415(35), 418(35) Klein, R. D., 159, 183(1) Klenow, H., 16, 18 Kleppe, K., 301, 303(13), 304(13), 305, 306, 308(27), 308(35), 309(27, 29), 310(29), 314, 316, 327(6), 328(6, 18), 329(6) Kleppe, R. K., 306, 308(35) Klett, R. P., 28, 314 Klevan, L., 335, 349, 356, 358(57) Klug, A,, 296 Knippers, R., 40, 336, 412, 413(117), 414(117), 437 Knopf, K. W., 75, 77, 79(45), 80, 81 Kobayashi, M., 163(62), 165(62), 166 Kochetkov, N. K., 578, 579(40) Koerner, J. F., 225 Kogoma, T., 470 Kohiyanna, A., 367 Kohwi-Shigematsu, T., 435 Koka, P., 491 Kokurina, N. A., 518 Kolodner, R., 9, 63, 367, 368(10) Kolpak, F., 413, 414(126), 416(124), 417( 126), 42 1(126), 422( 126) Komatsu, Y.,190

AUTHOR INDEX Koncz, C., 161(33), 165, 186(33), 187 Konigsberg, W. H., 392, 394, 395(73), 402(69), 403, 405(73), 407(63), 413,418, 419(131), 422(133), 424(73), 425(73), 449 Konrad, E. B., 32, 46 Konrad, M., 294, 295(60) Konvicka, J., 537, 539(190) Kooistra, J., 236, 237(32) Kopchick, J., 94 Korn, D., 5 , 67, 69, 71, 72, 73, 74(32), 75(18), 79(18), 82(21) Kornacka, L., 534 Kornberg, A., 4, 5 , 6, 7(3), 9(9), lo(]), 11(13), 12(3), 15, 16, 17(6), 28, 29, 30(31), 36,40,41(1), 43(1, 8), 44(16), 45, 46(15, 16, 18, 19), 47, 48(1), 49(1), 52, 53(1), 54(1), 5311, 56(1), 59, 83, 85, 86(74), 90, 110, 115, 128, 203, 209, 211, 215(26), 216(26), 217(16, 26), 218(26), 219(26), 221(16), 222(16, 26), 223(16, 261, 303, 3 14, 367, 368, 369, 370(24), 388, 408, 422, 423(139), 424, 425(143, 149), 426(143), 427(143), 428(143), 429(139, 143, 144, 149), 430(96, 140, 143, 148). 584, 585, 586(63) Kornberg, R. D., 296 Kornberg, T., 4, 10(1), 15, 40, 41(6), 46(7), 128, 408, 423(96), 424(96), 425(96), 426(96), 427(96), 428(96), 429(96) Korngerg, A., 332 Kornhauser, A,, 506 Koshland, D., 470 Kossel, H.,109 Koster, H., 177, 182(100), 186(100) Kosykh, V. G., 190 Kotewicz, M. L., 335, 336, 356, 358(57), 473, 474(19), 475(16, 19) Kourilsky, P., I16 Kouyama, T., 295 Kowalczykowski, S. C., 376, 378, 380, 381(14), 382(14), 385(5), 393(5), 395, 396(5, 14). 397(5, 14), 398(5, 14). 399(5, 14), 401(14), 402(74), 403(74), 404(74), 405(20), 406(20), 408, 410, 411(5, 106), 439(106), 443(5) Kowalski, D., 194, 1939, lo), 335 Kowalski, P., 124 Kozinski, A. W., 392, 408(59), 410(59) Krakow, J. S . , 106, 175

AUTHOR INDEX Kreis, W., 534 Krell, H., 367, 442 Kreuzer, K. N., 340, 348, 349(23), 350(23), 351(23), 352(23). 353, 355, 363 Krish, H. M., 384, 392(44), 41 l(44) Kritskii, G. A., 545 Kroeker, W. D., 194, 195(9, 10, I I ) , 200(11) Kroger, H., 519, 528, 529(94), 536 Krokan, H., 79 Kropinski, A. M. B., 536, 539(177) Kruezer, K., 154 Kuebbing, D., 160(17), 165 Kubo, M., 365 Kubota, I . , 394, 410(72) Kudrna, R. D., 561 Kuebler, J. P., 269, 270(73), 551 Kuempel, P. L., 29, 230 Kuenzle, C. C., 76, 79(40) Kuhn, B., 8, 442 Kuhlmann, I., 530(167), 535, 546(167) Kuhnlein, U., 139, 145(14), 149(14), 151(14), 269, 270(76) Kukuchi, T., 296 Kula, M. R., 160(14), 165, 167, 168(76, 77), 169(76, 77), 183(76), 186(76) Kumar, A., 314 Kung, H.-J., 95, 98 Kung, P. C., 106(30), 107, 111, 114(23, 30, 64), 118(18, 23) Kung, V. T., 334, 343(20) Kuninaka, A., 194, 195(5, 6) Kunitz, M., 122, 282, 284(7), 286(5) Kunkel, L. M., 161(37), 165 Kunkel, T., 103 Kunkel, T. A , , 424, 429(142) Kuo,M. T., 530(162), 535, 546(162) Kuo, T. T., 531 Kupersztoch, Y. M., 364 Kurosawa, Y.,310 Kurz, Ch., 336 Kurz, C., 387 Kushner, S. R., 208, 209, 223, 224(63), 229, 230(84), 231(84), 237, 250(34), 259 Kuzmichev, V. A , , 545 Kwant, M. M., 79, 80(52) Kwoong, S., 197 Laanka, E., 63 Lackey, D., 134, 144, 146(52), 147(52),

148(52), 149(52), 150(52), 240, 523, 524(63) Lacks, S. A., 127, 132, 160(7), 165, 180, lHl(l15), 214, 521, 524(49), 541(49) Lai, M. T., 91 Lamola, A . A., 500, 504, 505(91), 506, 507(96), 512(84) Lampidis, T., 141 Lampkin, B., 106, 107(29) Landa, E., 367 Landy, A , , 336, 472, 474, 475(4), 480(4, 6) Lane, D., 543, 544(223) Langowski, J., 167, 168(73), 169, 182, 183(73), 186(73) Lanka, E., 11 Lapeyre, J. N., 519 Larison, L. L., 368 Lark, C., 542 Lark, C. A , , 470 Lark, K. G . , 470 Larson, J. E., 115, 159, 183(1) Larsson, A., 82 Laskey, R. A., 335 Laskowski, M., 194, 195(9, lo), 203 Laskowski, M., Sr., 122, 124, 282, 287(1), 292, 294 Laszlo, J., 68 Lauppe, H. F., 368,442 Laurent, T. G., 62 Lautenberger, J . , 144, 145, 146(52, 60), 147(52), 148(52), 149(52, 60), 150(52), 151(60), 153(60) Lautenberger, J. A., 520, 523, 524(63, 65), 525(65) Laval, F., 586 Laval, J., 261, 268(40, 41), 551 Lavin, M.F., 239, 241(49), 242(49), 244(49), 245(49) Lawley, P. D., 537, 545, 550 Laws, P., 418 Lazarides, E., 293 Leberman, R., 294, 295 Le Bon, J. M., 334 LeClerc, J. E., 9 LeClerq, J. E., 63 Leder, P., 99 Lederberg, S., 237 Ledner, 3. A., 184 Lee, J. S., 348

604 Lee, L., 82 Lee, X. H., 68 Lee, Y. H., 163(61), 165(61), 166, 182(61), 183, 188 Lefler, C. F., 109, llO(48) Lehman, I. R., 10, 15, 18,21,22(19), 23(20), 28, 29, 30(31), 31, 32(37, 38), 33(37), 36, 52,56,57(2), 58,59(15), 62(2), 124, 134, 193, 194(1, 2), 195(1, 2), 197(1), 199(2), 204, 206, 207, 208(5, 10). 209(5, 10). 210, 211(10), 215(26), 216(26), 217(26), 218(26). 219(26), 222(26), 223(26), 224, 231, 241, 246(59), 302, 327, 367, 410, 431, 446(19, 211, 447, 448(19), 449, 450(38), 451(31, 32, 411, 452(21, 30, 31, 32, 40, 411, 453(32),454(31, 32, 39, 40), 455(21, 31, 32, 38, 40). 456(19, 31, 32), 457(19, 31, 38, 40). 458(21, 40), 459(21, 52), 460(21, 521, 461(21, 31, 40, 52), 465(21, 521, 466(21, 52), 467(19), 468(19), 537, 538(188), 581, 582, 584(45), 585(45) Leinbach, S . , 82 Leis, J. M., 212, 218(27) Leis, J. P., 88, 89, 91, 92(6) Lemaire, G.,385, 411(46) Lenny, A. B., 58, 342 Leonard, N. J., 382, 392(30) Leortan, D., 276 LePecq, J.-B., 343 Lerner, R. A., 98 Leung, D. W., 162(57), 165(57), 166 Levin, C. J., 300,304(7), 306(7), 316, 317(31, 318(3), 319(3, 10, l l ) , 320(3), 321,322(3, lo), 323, 324(3), 325, 326(3), 327(3, 11). 328, 329(3) Levine, A., 364 Levine, A. J., 364, 439, 440(186, 194) LeVine, D., 335 Levinson, A. D., 439, 440(194) Levinson, W. E., 88(19), 89, 90 Levitt, M., 296, 350 Lewis, H. A., 537, 539(193) Lewis, J. B., 439 Lewis, P. A,, 407 Liao, T.-H., 282, 283, 284, 285, 286(21), 287(22), 288, 292 Lica, L., 418 Lieberman, R. P., 238

AUTHOR INDEX Lielausis, A., 52, 55(4), 57(4), 360, 372, 408 (574 Lies, J. P., 28 Lillehaug, J. R., 301, 303(13), 304(13, 261, 306, 308(27, 3 9 , 309(27), 316, 327(6), 328(6, 18), 329(6) Lin, F. U., 532 Lin, L., 154 Lin, M. C., 288 Lin, P., 236 Lindahl, T., 216, 222, 252, 264, 266(3), 269(3), 270(71, 72), 273, 327, 550, 551(9),552(3,9), 553(9),561,565,566(9, 10, 1 I , 13), 567(5), 568(2, 5 , 9, I t , 131, 569(1), 570(1, 5 , 9, 11, 13), 571(13), 573(5, I t ) , 574(5), 575(1, 5 , 111, 576(5, 10, 11, 13), 577(5, 10, 13), 578(30), 579(13, 30), 580, 582(10, 13), 583, 584(5), 586(21) Lindan, C. P., 255, 256(22, 23), 259(22, 23), 260(23), 261(22, 23), 562, 566(16), 567, 570(16), 572(16), 585(16) Lindberg, U., 285, 293, 294, 296(59) Lindon, C. P., 128 Linial, M.. 88 Linn, S., 123, 124, 125, 126, 128(7, 1 3 , 130(7), 131(7), 133(7, IS), 134, 138, 139(9), 142(13), 144, 145(14), 146(52,59, 60), 147, 148(52, 59). 149(14, 52, 601, 150(52, 59, 71), 151(14, 59, 60, 71), 152(71), 153(59,60,71, 79), 154(71, 79), 155(59,69,71), 193, 194(1,2), 195(1, 2), 197(1), 198, 199(2), 210, 229, 238, 240(43), 241(43), 242(43, 53, 57), 243(53), 244(43, 72), 245(43,61,64,72), 246(43, 53, 57, 61, 64, 74), 247(74, 751, 248(75),250,252,257,258(26), 262,263, 268(26,46), 269,270(76, 77), 272(44,45, 46,99), 277,278, 367,368, 370(17),441, 520(42), 521, 522, 523, 524(42, 63), 540, 551, 554(19), 555(19, 22), 556(19, 22), 557(22), 558(19, 22), 559(22), 561, 562, 566(14), 567, 568(14), 570(14), 571(14), 572(14), 579(14), 582(14), S83(55), 584(14) Linni5, T., 439, 440(187) Linsley, W. S . , 269, 270(77), 551 Lippard, S . J., 184 Litfin, F., 431, 432(162)

AUTHOR INDEX Litman, R. M., 387 Little, J. W., 231, 446(17), 447, 463(29), 469(29) Liu, C. C., 9, 11(16), 12(16), 56,60(10), 333, 355, 356(52), 360(51, 52), 363(51), 367, 370(11), 381, 408, 409(22, 97), 410(22, 103) Liu, L. F., 29, 332, 333, 334(1), 335(17), 337(19), 340(51), 343(1), 344, 346, 347, 349, 354(26), 355, 356(52), 357, 360(51, 521, 353(51), 366(26) Liu, T.-Y., 282, 288(8), 386 Live, T. R., 301, 303, 306(10), 308(22), 313(10) Livingston, D., 41, 42, 46 Livingston, D. M., 90,584 Livneh, Z., 278, 551, 552(20), 553, 555(20, 211, 556(20, 21), 557(20), 558(20, 21), 559(2 1), 561(2 1), 562(2I ) Lizardi, P. M., 99 Lizarraga, B., 289 Ljungquist, S., 216, 222, 267, 269, 270(71, 721, 272(64), 550, 551, 561(15), 566, 567(5), 568(5), 570(5), 572, 573(5), 574(5), 575(5), 576(5), 577(5), 584(5) Lloyd, R. G., 229 Lobban, P. E., 116, 117(93) Lodemann, E.. 506 Loeb, L. A., 4, 17, 19, 90, 103, 175, 424, 429( 142) Loewen, P. C., 175, 301, 303(13), 304(13) Lohman, T. M., 380, 381(16), 382(16, 25), 399(15), 40x2 I ), 406(2 I) Lomedico, P., 99 Lonberg, N., 376,378,380,381(14), 382(14), 385(5), 393(5), 395, 396(5, 14), 397(5, 14), 398(5, 14), 399(5, 14), 401(14), 402(74), 403(74), 404(74), 405(20), 406(20), 410, 411(5, 106), 439(106), 4430) Long, J. C., 106(30), 107, 114(30) Lopatina, N. G., 523, 524(69), 537, 538(182) LoPresti, M. B., 392, 395, 396(77), 407(63) Lord, S. T., 440 Loring, D., 300 Louis, C., 548 Love, J.. 125, 126(14), 132(14) . 255, 257(25), 285 Love, .ID., Low, B., 235

Low, K., 143 Low, K. B., 209, 210(17), 222(17), 228(17), 229, 424, 429(150), 468 Low, M., 536, 542 Low, R. L., 12 Lowy, I., 546 Lubit, B. W., 520, 533(23) Lucas. Z. J., 52, 53(1), 54(1), 55(1), 56(1), 303 Ludlum, D. B., 109 Ludwig, G., 269, 270(74) Ludwig, M. L., 182 Lui, A. C. P., 160(8), 161(8), 162(57), 165(57), 166, 180, 181(112) Lundblad, R. L., 282 Lunt, M. R., 234, 241(14), 244(14), 245(14) Luria, S., 138, 142(2) Luria, S. E., 537 Luria, S. W., 446, 466(9) Luthardt, F. W., 519, 545(14), 546(14) Lutter, L. C., 296 Lynn, S. P., 161(21), 165 McBridge, 9. C., 162(57), 165(57), 166, 180, I8 1 (1 12) McCaffrey, R. P., 106(26, 30), 107(19, 36), 108, 111(19), 114(19,23,30), 118(18, 19, 23, 26) McCarthy, 9. J., 161(37), 165, 172, 173(81), 526 McCarthy, D., 363,409 McClelland, M., 161(44), 165(44), 166 McClements, W. L., 96 McClure, W. R., 20 McConaughy, 9. L., 335 McConnell, D. J., 161(34), 165 McCorquodale, D. J., 61 McDonnell, J. P., 88(19), 89 McEntee, K., 367,431,446(19,21), 447(14), 448(19), 449, 450(38), 451(31, 32, 41), 452(21, 30, 31, 32, 40, 41), 453(32), 454(31, 32, 39, 40), 455(21, 31, 32, 38, 40), 456(19, 31, 32), 457(19, 31, 38, 40), 458(21, 40), 459(21, 52), 460(21, 52), 461(21, 31,40, 52), 463(14), 465(21, 521, 466(21,52),467(19),468(19),469(14,28) McEwan, R. N., 106, 118(13) McGhee, J . D., 354, 378, 398(9), 404, 530(159), 533, 535, 546(159)

606 McGhee, J. M., 381, 402(26) McGrath, J. R., 492, 508(57) McHenry, C., 5 , 7, 9(9), 41, 43(12), 44(16), 45, 46(16), 47(17), 48 McKay, D. B., 449 McKay, E. L., 518,528,529(95,96), 532(7), 533 MacKay, V., 150, 208, 210, 229, 240, 241, 242(53), 243(53), 245(64), 246(53, 64, 74), 247(74), 368, 370(17) McKenzie, L. J., 288 McKune, K., 69, 72, 73(15) McLaren, A. D., 498, 499(75) McPherson, A ., 413, 414(126), 416, 417(126), 421(126), 422 Maass, G., 167, 168(73), 169(73), 177, 182(73, IOO), 183(73), 186(73, 100) Mace, D., 52,409, 410(102, 103) Mach, B., 116 Madden, J. J., 487, 488(39), 489, 494, 501(44, 45, 62), 502 Maeda, Y., 93 Maestre, M. F., 382, 394, 403(68) Magee, P. N., 545 Magee, W. E., 83 Magri, E., 295 Mahler, H. R., 234 Maio, J. J., 543, 548 Maitra, U., 523 Majors, J. E., 95 Makino, F., 573, 581 Makulay, R. A., 162(58), 165(58), 166 Malathi, V. G., 314 Malcolm, A. D. B., 167, 168(71, 72), 182(72), 183 Malec, J., 534 Maltman, K. L., 537, 539(193) Maltman, W., 440 Malyguine, E., 175 Mandel, J. L., 530(161, 162), 535, 546(161, 162) Mandel, M., 537, 539(190) Manes, S. H., 363 Maniatis, T., 117 Manly, K. F., 90 Mann, M. B., 160(20), 165, 180, 181(108), 189(20), 520, 524(26), 526(26) Mannherz, H. G., 294, 295(60) Manning, J., 441

AUTHOR INDEX Mano, Y., 79 Mansour, J., 229 Mantell, N. J., 396, 426(79), 428(79) Maples, V. F., 223, 224(63) Marchionni, M. A., 162(47), 165