Advances in Protein Chemistry, Volume 36

ADVANCES IN PROTEIN CHEMISTRY Volume 36

CONTRIBUTORS TO THIS VOLUME Pierre Douzou L. Giri W. E. Hill David Kristofferson

M. F. Perutz Gregory A. Petsko Daniel L. Purich Kai Simons Graham Warren H. G. Wittmann B. Wittmann-Liebold

ADVANCES IN PROTEIN CHEMISTRY EDITED BY

C. B. ANFINSEN

JOHN T. EDSALL

Department of Biology The Johns Hopkins University Baltimore, Maiyland

Department of Biochemistry and Molecular Biology Harvard University Cambridge, Massachusetts

FREDERIC M. RICHARDS Department of Molecular Biophysics and Biochemistry Yale University New Haven. Connecticut

VOLUME 36

1984

ACADEMIC PRESS, INC. (Harcourt Brace Jovanovich, Publishers)

Orlando San Diego San Francisco New York London Toronto Montreal Sydney Tokyo S ~ Paulo O

COPYRlOHT @ 1984, BY ACADEMIC -9, INC. ALL R I o m RESBRVED. NO PART OF TIUS PUBLICATION MAY B e REPRODUCBDon TMNSMITTED IN ANY FORM OR BY ANY MEANS, ELECIXONIC OR MECHANICAL, INCLUDING PHOTOCOPY, RECORDINO, OR ANY INFORMATION STORAOE AND RETRIEVAL SYSTEM, WITHOUT PBRM188ION IN WRXTfNO FROM THE PUB-

ACADEMIC PRESS INC.

Orlando, Florida a2884

United Kingdom Edition published by

ACADEMIC PRESS, INC. ( L O N D O N ) 24128 Oval Road, London NWI 1 D X

PRINTED IN THE UNITED STATES OF AMERICA 84 85 86 81

9 8 1 65 4 3 2 1

LTD.

CONTENTS CONTRIBUTORS TO VOLUME 36

.

.

.

.

.

.

.

.

vii

Ribosomal Proteins: Their Structure and Spatial Arrangement in Prokaryotic Ribosomes

L. GIRI,W. E. HILL,H. G. WITTMANN, AND B. WIITMANN-LIEBOLD I. Introduction . . . . . . . . . . 1 11. Ribosomal Proteins . . . . . . . . . 2 111.. Ribosomal RNAs . . . . . . . . . 23 IV. Topography of the Ribosome. . . . . . . 28 V. Summary and Outlook . . . . . . . . 47 References. . . . . . . . . . . 48 Appendix: Primary Structure of Escherichiu coli Ribosomal Proteins . . . . . . . . . 56 Appendix References . . . . . . . . 77 Semliki Forest Virus: A Probe for Membrane Traffic in the Animal Cell

I. 11. 111. IV.

KAI SIMONS AND GRAHAM WARREN Introduction . . . . . . . . Structure . . . . . . . . . The Life Cycle of Semliki Forest Virus . . Perspectives. . . . . . . . . References . . . . . . . . .

. . . . . . . . . .

79 81 98 124 125

Microtubule Assembly: A Review of Progress, Principles, and Perspectives

DANIELL. PURICHAND DAVIDKRISTOFFERSON I. Introduction . . . . . . . . . 11. Biochemical Properties of Microtubule Proteins . 111. Nucleation or Initiation of Microtubule Assembly . V

.

. .

133 137 158

vi

CONTENTS

IV . Microtubule Elongation . . . . V . Protomer-Polymer Equilibria and Critical Concentration Behavior . . . . VI . Microtubule Length Redistribution . VII . Protomer Flux with Assembled Polymers (Microtubule “Treadmilling”) . . . VIII . Concluding Remarks . . . . References . . . . . . .

.

.

.

.

168

. . . . . . . .

182 190

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

194 206 207

Species Adaptation in a Protein Molecule

M . F. PERUTZ I . Introduction . . . . I1. Fish Hemoglobins . . . I11. Amphibia . . . . . IV . Reptiles . . . . . V . Birds . . . . . . VI . Mammals . . . . . VII . Species Adaptation in Enzymes VIII . Discussion . . . . . References . . . . .

. . . . . . 213 . . . . . . 217 . . . . . . . 225 . . . . . . 227

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

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

232 233 235 236 240

Proteins at Work: “Stop-Action” Pictures at Subzero Temperatures

I. I1. I11.

IV . V. VI .

PIERREDouzou AND GREGORY A . PETSKO Introduction . . . . . . . . . Trapping of Enzyme-Substrate Intermediates in Solution . . . . . . . . . . X-Ray Studies at Subzero Temperatures: Physical-Chemical Basis of Cryoprotection of Proteins in Solution and in the Crystalline State and Related Problems . . . . . . . Devices . . . . . . . . . . . Results . . . . . . . . . . . Prospects and Problems . . . . . . . References . . . . . . . . . .

.

.

.

.

.

.

.

280 321 328 353 358

. .

. .

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

.

. .

SUBJECT INDEX

.

.

247

.

.

.

.

246

.

AUTHORINDEX

CONTENTS OF PREVIOUS VOLUMES

.

.

.

363 383 387

CONTRIBUTORS TO VOLUME 36 Numbers in parentheses indicate the pages on which the authors' contributions begin.

PIERREDouzou (245), Institut de Biologie Physico-Chimique, U 128 INSERM, Paris 75005, France L. GIRI'(l),Department of Biochemistry, University of Massachusetts, Amherst, Massachusetts 01003 W. E. HILL( l ) , Department of Chemistry, University of Montana, Missoula, Montana 59812 DAVIDKRISTOFFERSON~ (1 33), Department of Chemistry, University of California, Santa Barbara, California 931 06

M. F. PERUTZ(213), MRC Laboratory of Molecular Biology, Cambridge CB2 2QH, England

GREGORY A. PETSKO (245), Department of Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139 DANIELL. PURICH(133), Department of Chemistry, University of California, Santa Barbara, California 93106 KAI SIMONS (79),European Molecular Biology Laboratory, 6900 Heidelberg, Federal Republic of Germany GRAHAM WARREN (79),European Molecular Biology Laboratory, 6900 Heidelberg, Federal Republic of Germany H. G. WITTMANN (l), Max-Planck-Institut fur Molekulare Genetik, D-1 000 Berlin-Dahlem 33, Federal Republic of Germany B. WITTMANN-LIEBOLD ( l ) , Max-Planck-Institut fur Molekulare Genetik, D-1000 Berlin-Dahlem 33, Federal Republic of Germany Present address: Pharmacia Fine Chemicals, 800 Centennial Avenue, Piscataway, New Jersey 08854. * Present address: Department of Biochemistry and Biophysics, University of California Medical Center, San Francisco, California 94 143. vii

This Page Intentionally Left Blank

ADVANCES IN PROTEIN CHEMISTRY Volume 36

This Page Intentionally Left Blank

RIBOSOMAL PROTEINS: THEIR STRUCTURE AND SPATIAL ARRANGEMENT I N PROKARYOTIC RIBOSOMES By L GIRI;~' W. E. HILL? and H. G. WllTMANNI Wlth an APPENDIX: Prlmary Structure of Escherlchla coll Ribosomal Proteins By B. WITTMANN-LIEBOLD* 'Department of Biochemistry, Unlvenity of Massachusetts, Amherst, Massachusetts, *Department of Chemlrtry, University of Montana, Missoula, Montana, and *Max-#anck-lnsfflut fur Molekuiare Genetlk, Berlln-Dahlem, Federal Republic of Germany

. . . . . . . . . . . . A. Preparation of Ribosomal Proteins B. Primary Structure . . . . C. Secondary Structure . . . . D. Tertiary Structure . . . . E.Shape. . . . . . . . 111. RibosomalRNAs. . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . , . . . . A. Primary and Secondary Structures . . . . . B. Spatial Arrangement of the RNAs in Situ . . . Topography of the Ribosome . . . . . . . A. Size and Shape of Ribosomal Subunits . . . . B. Crystals of Ribosomal Particles . . , . . . C. Immune Electron Microscopy (IEM) . . . . D. Neutron Scattering . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Singlet-Singlet Fluorescent Energy Transfer between Proteins . . F. Cross-Linking. . . . . . . . . . . . . . . . G. Protein-Binding Sites on RNA . . . . . . . - . . . H. Fragments of Ribosomal Particles . . . . . . . . . . I. Protein Complexes . . . . . . . . . . . . . . J. Chemical Reactivity . . . . . . . . . . . . . . K. Assembly of Ribosomal Subunits . . . . . . . . . . Summary and Outlook . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . Appendix: Primary Structure of Escherichia coli Ribosomal Proteins . . Appendix References. . . . . . . . . . . . . . .

I. Introduction

11. Ribosomal Proteins

IV.

V.

.

. .

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

1

2 2 4 9 12 15

23 23 27 28 28 32 32 34 37 38 40 43 44 44 45 47 48 56

77

I. INTRODUCTION Translation of the genetic message into a polypeptide chain is a universal process that takes place on ribosomes. These particles play an

'

Present address: Pharmacia Fine Chemicals, 800 Centennial Avenue, Piscataway, New Jersey 08854. 1 ADVANCES IN PROTEIN CHEMISTRY, Vol. 36

Copyright 0 1984 by Academic Reas. Inc. All righu of reproduction in any form reserved.

ISBN 0-12-094236-7

2

L. GIRl ET AL.

integral role in this process which is not yet clearly defined. Containing 50-80 different components, the complexity of the ribosome itself is indicative of an intricate mechanism with multifaceted functions. For instance, it is apparent that ribosomes must play a specific role in initiation, elongation, and termination of the synthesis of polypeptide chains. In addition, defined interactions must occur between the ribosome and tRNA, mRNA, and the nascent polypeptide chain itself. Ribosomes are found in all organisms and consist of two unequal subunits. Prokaryotic ribosomes (e.g., those from Escherichia coli) are the most widely studied and those with which this article will be primarily concerned. The 50 S subunit of the E. coli ribosome consists of one strand each of 5 S and 23 S rRNA, and 32 different proteins. The 30 S subunit contains 16 S rRNA and 21 different proteins. It is the purpose of this article to review the characteristics of the ribosomal proteins which have been studied to date. By necessity, the conformation of the ribosomal RNA and the architecture of the ribosomal subunits themselves must be discussed since the placement of the proteins on the subunits is essential in elucidating the function of the ribosome. 11. RIBOSOMAL PROTEINS

Since the early 1960s, when it was shown that ribosomes contain a large number of different proteins, a tremendous effort has been made to isolate and characterize each individual protein. The first problem was to prepare individual, homogeneous proteins. Determination of the primary, secondary, and tertiary structure, as well as of the overall physical characteristics, was then approached. A . Preparation of Ribosomal Proteins Three major methods have been employed for the isolation of all E. coli ribosomal proteins. 1. After dissociation of the 70 S ribosome into its two subunits followed by zonal centrifugation for the separation and isolation of the 30 S and 50 S subunits on a preparative scale, the ribosomal proteins were extracted by acetic acid and then separated by cellulose ion exchange chromatography and by gel filtration on Sephadex in the presence of 6 M urea. In this way all the 53 individual ribosomal proteins have been isolated (Wittmann, 1974). Proteins prepared in this manner have been used for physical studies (Brimacombe et al.,' 1978; Wittmann, 1982) as well as for immunological investigations (Stoffler el al., 1980; Lake,

RIBOSOMAL PROTEINS

3

1980) and protein-sequence analyses (Wittmann et al., 1980; WittmannLiebold, 1980b). 2. In order to avoid the use of acetic acid and urea which can lead to denaturation of proteins, a new gentler procedure has been developed (Dijk and Littlechild, 1979). The ribosomal subunits were first washed stepwise with salts (e.g., LiCl or NaCl) of increasing concentrations, which remove specific groups of proteins from the particles. The proteins in each of the groups were further separated by column chromatography on carboxymethyl-Sephadex, followed by gel filtration on Sephadex G-75 in buffers of high ionic strength but without urea. The proteins isolated by this method have been used for physical studies such as circular dichroism (CD), nuclear magnetic resonance (NMK), shape determination, and crystallization (Wittmann et al., 1980; Wittmann, 1982), which require conservation of the native structure as much as possible. 3. Isolation of 50 S ribosomal proteins that are still functionally active has also been accomplished in the following way (Wystup et al., 1979): first, prefractionation by washing 50 S subunits with salts of different concentrations; second, gel filtration on Sephadex; and third, column chromatography on carboxymethylcellulose in buffers containing 6 M urea, after which the proteins are extensively dialyzed against renaturing buffers. Proteins isolated in this way regain by renaturation their biological activity in reconstitution and functional tests. Therefore, they are being extensively used for total reconstitution of 50 S subunits followed by neutron-scattering studies (see below).

Molecular weights of each of the purified proteins were initially determined by sedimentation equilibrium and SDS-gel electrophoresis (Wittmann, 1974). More recently, as the primary sequences have been determined, exact molecular weights can be calculated. The molecular weights obtained from the physical methods were consistently higher than those found from the chemical means. This discrepancy is most pronounced for small and very basic proteins and is not surprising since some of the assumptions used for molecular-weight determinations by SDS-gel electrophoresis rapidly lose viability as the molecular weight of the protein descends below 10,000- 12,000. Furthermore, the physical methods give a molecular weight of the protein and associated salts, whereas sequence results provide a value for the polypeptide only. After isolation in a pure state, each of the ribosomal proteins was injected into rabbits and/or sheep, and antibodies were obtained. It was demonstrated by various immunological techniques that there is no significant immunological cross-reaction among any of the E. colz ribosomal

4

L. GIRI ET AL. TABLE 1. Number of Amino Acid Residues Protein

Residues ASP Thr Ser Glu Pro GlY Ala Val Met 1le

Leu TYr Phe His LYS

Arg =rP CYS Gln Asn

S1

S2

S3

S4

S5

S6

S7

S8

S9

S10

557

240

232

203

166

135

177

129

128

103

43 25 25 60 10 48 48 67 6 30 45 6 17 8 43 30 7 2 14 23

17 13 11 16 7 17 25 17 7 14 21 2 12 6 19 13 3 1 7 12

9 11 9 16 10 19 24 22 4 18 13 4 5 4 23 21 4 0 9 7

8 6 10 16 6 15 18 16 3 8 20 8 4 3 20 22 1 1 10 8

2 8 7 11 5 21 20 19 6 10 9 2 4 4 12 10 0 0 5 11

10 7

5 5 15 14 7 10 22 16 4 5 16 4 5 3 14 18 2 0 5 7

6 6 7 5 11 14 12 5 9 10 3 3 0 12 8

3 5 7 10 3 13 10 10 3 7 9 5 4 1 10 18

8 8 2 6 5 4 9 9 2 9 10 1 2 3 5 12

0

0

0

1 5 3

0 6 4

0 6 2

4

20 5 5 14 10 7 7 5 4 4 6 6 12 1 0 4 4

9

proteins, with the exception of proteins L7 and L12,which showed complete cross-reaction (St6ffler and Wittmann, 1971a,b). This is to be expected since these two proteins differ only by an acetyl group. The noncross-reactivity among antibodies raised against the individual ribosomal proteins was an important prerequisite for the immune electron microscopic studies described below.

B . Primary Structure Most of the E . colz ribosomal proteins are rather basic with high isoelectric points (Kaltschmidt, 1971) and a high content of basic amino acids (Tables I and 11). The complete primary structures of all E. colz ribosomal proteins have been determined by Wittmann-Liebold and coworkers (see Table I11 and Appendix). The elucidation of the primary structure of the ribosomal proteins was very much facilitated by several improvements in the amino acidsequencing technique, especially on the micro scale. These resulted in

5

RlBOSOMAL PROTEINS in E . coli 30 S Ribosomal Proteins

Protein

Residues ASP Thr Ser Glu Pro GlY Ala Val Met Ile Leu TYr Phe His LYS Arg

=rP CYS Gln Asn

S11 S12

S13 S14

S15 S16 S17 S18

S19 S20 S21

128

123

117

87

74

91

86

4 9 6 4 8

3 8 6 4 7 11 9

3 6 3 2 2 3 7 2 0 6 5 6 3 1 6 12 0 1 5 1

4 6 4 4 6 7 6

2 2 4 3 1 2 19

98

6 5 6 2 6 8 6 4 4 4 13 11 5 17 11 11 1 0 1 5 7 1 0 2 2 9 3 12 4 3 8 9 8 1 4 2 1 4 1 1 3 3 3 3 1 9 13 11 11 14 15 15 14 1 0 0 1 2 4 1 1 3 4 2 5 7 5 2 3

82

83

5 4 4 6 3 6 6 3 4 5 5 7 0 2 2 5 6 4 7 11 2 5 5 8 1 1 1 1 3 6 7 12 4 5 2 1 1 2 4 2 4 2 3 6 5 10 11 10 7 0 1 1 0 0 2 5 2 1 2 4 2

2 2 4 7 1 4 5 13 7 1 0 1 2

70

1 3 2 8 3 1 9 7 2 6 3 0 7 1 3 4 1 2 2 3 3 1 14 9 7 14 0 0 0 1 5 0 6 2

sequencing the proteins more quickly, more reliably, and with less material and labor. Much progress in this direction has been made by numerous technical improvements to the Beckman liquid-phase sequenator (Wittmann-Liebold, 1973, 1980a),including the construction of an automated conversion device for phenylthiohydantoin-amino acids (Wittmann-Liebold et al., 1976); improvements in the solid-phase sequencing technique (Wittmann-Liebold, 1980a, 198la); isolation of peptides with thin-layer and high-performance liquid chromatography (Hitz et al., 1977; Wittmann-Liebold et al., 1977a; Wittmann-Liebold, 1981b); and the introduction of a new reagent (Chang et al., 1978) for the Edman degradation reaction that results in a 10- to 20-fold higher sensitivity than previous techniques. A number of ribosomal proteins contain modified amino acids at the N terminus or at other positions of the protein chain (Table IV). The N termini of three proteins (S5, S18, and L7) are acetylated, thus they cannot be subjected successfully to manual or automatic Edman degradation because of their blocked N termini. Mutants have been isolated in

TABLE I1 Number of Amino A d Residues in E. coli 50 S Ribosomal Proteins Protein

Residues

ASP Thr SeT

Glu Q,

Pro GlY Ala Val Met Ile

Leu TYr Phe His LYS

k g =rP CYS Gln

Asn

L1

L2

L3

L4

L5

L6

L9

L10

L11

L7/12

L13

L14

L15

L16

L17

L18

233

272

209

201

178

176

148

165

141

120

142

123

144

136

127

117

14 13 8 11 7 20 33 29 6 11 17 3 4 2 23 11 0 0 9 12

10 14 10 11 17 34 21 26 5 11 15 6 4 9 25 29 2 2 6 15

9 16 9 12 6 26 19 28 4 8 11 1 8 3 19 12 2 0 8 8

12 12 12 12 5 13 23 21 5 8 19 2 6 3 18 14 2 0 10 4

13 9 7 11 6 13 16 12 5 14 13 6 10 1 16 15 1 1 4 4

10 12 4 8 8 18 20 22

6 6 5 14 3 11 24 20 1 7 12 1 6 2 11

6 9 6 15 5 9 33 15 5 5 15 3 6 1 12 13 0 1 3 3

6 9 8 6 9 12 20 13 5 9

6 3 6 16 2 8 28 16 3 6 8 0 2

7 12 0 8 6 12 14 12 4 8 9 6 3 7 14 11 1 0 4 4

5 5 6 6 5 12 7 14 4 12 8 1 3 1 11 12 0 2 3 6

2 9 7 9 4 26 15 9 2 10 12 1 4 1 13 14 0 0 3 3

3 9 2 7 7 13 14 11 6 6 10 2 6 1 16 14 2 0 5 2

4 6 7 9 4 7 17 6 4 6 11 2 5 3 8 18 0 1 2 7

5 5 5 7 2 8 22 10 1 5 8 3 3 3 9 13 0

1 8 12 5 3 3 16 10 1 1 7 7

7 0 0 4 8

7 2 4 0 15 4 0 1 7 4

0

13 1 0 0 1 1

0

6 2

Protein

Residues

ASP

Thr Ser Glu

Pro GlY Ala

Val Met -3

Ile Leu =Yr Phe His LYS Arg Trp CYS Gln

Asn

L19

L20

L21

L22

L23

L24

L25

L27

L28

L29

L30

L31

L32

L33

L34

114

117

103

110

99

103

94

84

77

63

58

62

56

54

46

3 4 8 9 3 8 7 15 1

4 1 4 2 0

3 5

9 6

3 5 4

5 3 5 6 3 10 9 13 0 8 4

5 2 2 7 4 5 11 8 3 7 5 3 4 3 11 6 0 0 5 3

3 3 5 5 1 13 9 6 0 4 3 0 5 3 11 8 0 1 1 3

2 5 5 4 2 5 5 8 1 2 6 1 3 3 7 11 1 1 1 4

1 2 3 7 0 2 6 4 2 0 12 0 1

3 1 5 5 5 2 1 4 1 2 4 3 5 3 4 4 1 0 2 4 3 3 2 2 0 2 4 2 6 1 2 3 7 0 0 0 0 2 1 1 1

0 3 3 0 1 3 5 3 2 0 3 0 2 1 5 11 0 0 2 2

6 2 4 2 11 13 1

6 22 9 0 10 6 6 5 1 14 16 1

0

0

6 4

7 3

7

3

7

9 1 1

4 2

7 14 2 7 3 2 5 4 10 8 1 0 6 1

2 4 13 11 3 9 8 1 1 4 13 10 0 0 3 3

1

0 1 6 10 14 2 4 8 1 2 2 14 7 0 0 3

3

0

5 1 16 6 0 0 3 6

1

6 6 0 0 6 4

1 6 2 4 2 5 5 4 2 6 5 0 1 2 5 6 0 0 1 1

5 4 4 2 2 5 2 5 2 3 2 1 3 3 7 4 0 4 1

3

8

L. GIRI ET AL. TABLE I11 Number of Amino Acid Residues and Mohcukar Weights of Ribosomal and Related Proteins from E . coli" Protein

Residues

MW

Protein

Residues

MW

s1 s2 s3 s4 s5 S6 S7K S7B S8 s9 s10 s11 s12 S13 S14 S15 S16 S17 S18 s19 S20-L26 s 21 L1 L2 L3 L4 L5 L6 L7 L9

557 240 232 203 166 135 177 153 129 128 103 128 123 117 98 87 82 83 74 91 86 70 233 272 209 20 1 178 176 120 148

61,160 26,6 13 25,852 23,137 17,515 15,704 19,732 17,131 13,996 14,569 11,736 13,728 13,606 12,968 11,191 10,001 9,191 9,573 8,896 10,299 9,553 8,369 24,599 29,730 22,258 22,087 20,171 18331 12,220 15,696

L10 L11 L12 L13 L14 L15 L16 L17 L18 L19 L20 L2 1 L22 L23 L24 L25 L26-S20 L27 L28 L29 L30 L31 L32 L33 L34 IF- 1 IF-3 EF-TU NS 1 NS2

165 141 120 142 123 144 136 127 117 114 117 103 110 99 103 94 86 84 77 63 58 62 56 54 46 71 181 393 90 90

17,737 14,874 12,178 16,019 13,541 14,981 15,296 14,364 12,770 13,002 13,366 11,565 12,227 11,013 11,185 10,694 9,553 8,993 8,875 7,274 6,411 6,971 6,315 6,255 5,381 8,119 20,695 43,225 9,226 9,535

According to Wittmann-Liebold (1980b) and Wittmann (1982).

which the acetyl groups cannot be incorporated into the appropriate proteins because the acetylating enzymes are mutationally impaired. Using these mutants it could be shown that each of the three proteins is acetylated by a different enzyme (Isono, 1980).The most heavily methylated protein is L11 which contains nine methyl groups attached to amino nitrogens: three at the N-terminal amino acid (trimethylalanine) and six at the two internal trimethyllysines at positions 3 and 39, respectively.

9

RIBOSOMAL PROTEINS

Protein

Position

s5

I

s1 I

I

s12

88

S18

1

L3

I50

L7

1

L7/ L12

81

Modified amino acid N-Acetylalanine N-Monomethylalanine Derivative of aspartic acid N-Acetylalanine N5-Monornethylglutamine N-Acetylserine N-Monomethyllysine

Protein

Position

LI 1

1

LI 1

3

LI 1

39

LIG

1

L16

81

L33

1

IF-3

1

Modified amino acid N-Trimethylalanine N-Trimethyllysine N-Trimethyllysine N-Monomethylmethionine Derivative of arginine N-Monomethylalanine N-Monomethylmethionine

According to Wittrnann et al. (1980).

In addition to the ribosomal proteins, the two initiation factors IF-1 (Pon et al., 1979) and IF-3 (Brauer and Wittmann-Liebold, 1978), the elongation factor EF-Tu (Arai et al., 1980), and the two proteins NS-1 and NS-2 (Mende et al., 1978), which bind to ribosomes and to DNA, have been completely sequenced (Table 111). Because the sequencing of DNA has become so straightforward, genes of several ribosomal proteins have been sequenced to allow an independent determination of the primary sequence of some of the proteins (Post et al., 1979; Olins and Nomura, 1981). These studies have confirmed the sequences previously elucidated by protein chemical techniques. In the case of S1 (the largest of all E. coli proteins), the combination of amino acid- and nucleotide-sequence determinations was used to provide the sequence (Schnier et al., 1982). C . Secondary Structure 1. Prediction of Secondary Structure

Using the primary sequence, the secondary structure of the proteins has been predicted by the application of four different statistical proce-

10

L. GIRI ET AL.

dures (Wittmann et al., 1980),These analyses give the location of secondary structural elements along the polypeptide chain. Many of the 30 S proteins are predicted to have a high a helix content, whereas the 50 S proteins are more diverse and show a larger contribution of /.3 sheet. Figure 1 illustrates the secondary structures of proteins S11 and L11 predicted by the four different methods.

2. Circular Dichroism The measurement of optical rotatory dispersion (ORD) and circular dichroism (CD) in protein chemistry has become very general practice for the characterization of proteins and their conformational changes resulting from denaturation, binding, aggregation, and chemical modification. Considerable information concerning especially the secondary structure of proteins in solution can be obtained from measurement of the optical activity. ORD is a measurement, as a function of wavelength, of a molecule’s ability to rotate the plane of linearly polarized light; CD is a measurement of the molecule’s unequal absorption of right- and lefthanded circularly polarized light. It is the conformation of the protein, that is, the asymmetric and periodic arrangement of peptide units in space, which gives rise to their characteristic ORD and CD spectra (Adler et al., 1973). In the last one and one-half decades many studies have been made on the ORD and CD of ribosomal proteins. Early studies (McPhie and Gratzer, 1966; Sarkar et al., 1967; Cotter and Gratzer, 1969)were made on a mixture of proteins, and the general conclusion was that both in the ribosome and in the isolated state (usually after acetic acid and urea extraction) the protein moiety contained approximately 25% a helix together with some P-pleated sheet and random-coil conformation. CD studies have recently been made on urea-isolated and renatured individual proteins from the small ribosomal subunit (Venyaminov and Gogia, 1982). In another study (Dijk et al., 1983a)’ many proteins obtained from both the small and large ribosomal subunits by a gentler salt extraction method were measured with the CD technique. It was found that, in general, the 30 S proteins are rich in a helix and contain a rather small amount of /3 sheet, whereas the 50 S proteins are more diverse, especially in their a helix content, and most are relatively rich in /.3 sheet (Dijk et al., 1983a). A comparison between the amount of a helix predicted from the primary structure (Wittmann et al., 1980) with that obtained from CD measurements (Dijk e l al., 1983a) gives a good agreement for a number of the proteins (e.g., S4, S8, L l l , L27). However, there are discrepancies

s 76 F 82 N77

R 76 PRE

FIG. 1. Secondary structure of E. coli ribosomal proteins Ll 1 and S11 as predicted from their amino acid sequences. The prediction was carried out using four different methods represented by four different lines (S74, F82, N77, and R76). The line "PKE" summarizes the secondary structure obtained when at least three out of the four predictions are in agreement. The symbols represent residues in helical (A), turn or bend (B), extended (C), and coil (D) conformational states, respectively. For details see Wittmann-Liebold el al. (1977b) and Dzionara rt af. (1977).

12

L. GIRI ET AL.

in other cases (e.g., with S20,S21, and L25),where a high a helix value is predicted and a low value is found experimentally.

D. Tertiary Structure Information concerning the tertiary structure of the proteins has been obtained from fluorometry, proton magnetic resonance spectroscopy, limited proteolysis, and X-ray analysis of protein crystals. 1. Fluoromtry

In fluorometric studies advantage has been taken of intrinsic fluorescent chromophores in proteins (aromatic side chains of tyrosine and tryptophan). Using this technique, the properties of proteins S4 and S7 have been studied (Gerard et al., 1975). It was found that emissions observed were due to tyrosine and tryptophan in protein S4 and to tryptophan only in protein S7.The single tryptophan in protein S4 was found to be buried, but it could be exposed by heating to 35°C and above. In protein S7 the tryptophan seems to be exposed or partially buried, depending on preparative conditions. The spectral characteristics of the tryptophan residue in S4 and of the partially buried one of S7 suggest that these residues are located in a-helical regions of the proteins. The observation that energy transfer from Tyr + Trp in protein S4 is very inefficient implies an extended conformation for this protein, suggesting that the N-terminal region of the protein, which contains six of the eight' tyrosines, is far from the tryptophan in position 167. In another fluorometric study (Brochon et al., 1976),it has been shown that protein S4 exists in two conformations at room temperature, but at 42°C under reconstitution conditions one conformation predominated. The presence of the two conformations indicates a nonhomogeneous preparation of the protein. In these studies, ribosomal proteins were used that had been extracted with acetic acid and then treated with 6 M guanidine hydrochloride before being dialyzed against reconstitution buffer. 2. Magnetic Resonance Studies Nuclear magnetic resonance (NMR) represents a spectroscopic technique capable of analyzing individual atoms within a molecule. The recording of proton NMR (PMR) spectra has been widely applied to the study of small molecules of biological importance. Although the large number of protons present and overlapping signals limit its application in the study of protein molecules, this technique is now being widely used in the tertiary structure determination of proteins. Much of the enhancement in the use of NMR has been due to circumventing its

RIBOSOMAL PROTEINS

13

limitations by improvements in instrumentation and experimental techniques. The NMR spectrum given by a globular protein with a well-defined tertiary structure differs from that of the same protein under denaturing conditions in two respects. First, the reduction in mobility of residues when the protein folds into a stable tertiary structure produces a broadening of resonances. Second, alterations in resonances caused by chemical shifts arise due to the stable placement of specific protons in unique chemical environments which leads to the appearance of resonances in new positions. The PMR spectrum of protein S1 suggests that the protein has considerable tertiary structure in physiological buffer and is more flexible than normal globular proteins of its molecular weight (Moore and Laughrea, 1979). No difference was observed when the protein was prepared in the presence or absence of urea at neutral pH. The spectra obtained in this study resemble those previously obtained with salt-extracted S 1 by Littlechild and Malcolm (1978). PMR studies have been performed on a number of other ribosomal proteins isolated by the acetic acidhrea method (Morrison et al., 1977a). The results of these studies have shown that acetic acidlurea-extracted proteins contain little tertiary structure. However, some structure was seen in protein S4 and especially in protein S16 as indicated by the appearance of ring-current shifted resonances in the apolar region of the spectrum (Morrison et al., 1977b). These are due to the interaction of apolar methyl groups with aromatic amino acids in the tertiary structure of the protein. The PMR spectra were recorded either in water or in dilute phosphate buffer at pH 7.0-conditions under which the proteins were soluble. A very extensive PMR study of most of the salt-extracted 30 S and 50 S ribosomal proteins has been reported by Littlechild et al. (1982). The results show that a majority of the proteins have a uniquely folded tertiary structure. This is especially true for proteins S15, S16, S17, and L30 which are very resistant to limited protease digestion. A good agreement between the amount of folding deduced from the PMR studies and the degree of proteolytic resistance also has been found for most of the other proteins. A comparison of the spectra of salt-extracted proteins with those obtained from proteins purified under denaturing conditions shows that the former proteins have a much higher degree of tertiary structure. When urea-denatured preparations of protein L11 are introduced into physiological buffers, two different conformations occur as shown by NMR studies (Kime et al., 1980). One form is distinctly folded while

14

L. GIRI ET AL.

the other form is intermediate between folded and unfolded conformations. The conformation obtained depends on details of the method used for returning samples to nondenaturing conditions. The NMR spectra of the renatured L11 preparations (Kime et al., 1980)were similar to those reported earlier for L11 prepared by milder methods (Morrison et al., 1977~).That proteins isolated in the presence of urea can be renatured and regain a folded conformation has also been shown with proteins S4 (Serdyuk et al., 1980), S15 (Gogia et al., 1979), Ll1 (Tumanova et al., 1981), and L25 (Kime et al., 1981). 3. Limited Proteolysis The existence of flexible regions in ribosomal proteins can be explored by studying the action of proteolytic enzymes under mild conditions. It has been found that many E. coli ribosomal proteins consist of two domains: one compactly folded and resistant to proteolysis, the other flexible and vulnerable to proteases (Littlechild et al., 1983). Some proteins (S15, S16, S17, and L30) are very resistant whereas others (S2, S6, S9, L2, L27, L29, and L33) are completely degraded without the appearance of discrete fragments. The remaining proteins yield fragments of various size under these conditions. 4 . Microcalorimetry One method to show the existence of tertiary structures in proteins is to study the disruption of the structure during the process of heat denaturation using scanning microcalorimetry (Privalov, 1974, 1979, 1982). This is in principle achieved by the direct measurement of the energy absorbed by proteins undergoing thermal order-disorder transitions. The heat capacity of the sample is recorded as a function of temperature as it is continuously heated from 20 to near IOO'C, the heat absorption peaks being a direct indication of the existence of a tertiary structure in proteins. The enthalpies of melting are obtained by integrating the area under the peaks and can be used to derive a number of other thermodynamic parameters. Calorimetric studies have been made on proteins S4, S7, S8, S15, S16, S18, L11, and L7 (Khechinashvili et al., 1978; Gudkov and Behlke, 1978). Most of these proteins displayed a cooperative tertiary structure in solution. Proteins S4, S7, S15, and S l 8 were extracted from the ribosome by a urea-LiC1 technique followed by renaturation, whereas proteins S8, S16, and L11 were prepared by the mild isolation method. A calorimetric study on protein S1 showed a noncooperative transition around 7O-8O0C, suggesting a flexible tertiary structure (L. Giri, unpublished).

RIBOSOMAL PROTEINS

15

5. Crystallamtion and X-Ray Analysk Since the X-ray structural analysis of crystallized proteins yields the most direct information on the tertiary structure, many attempts have been made in the last decade to crystallize individual ribosomal proteins. However, it was many years before any progress in this field was made. The N-and C-terminal fragments of the E. coli protein L7/L12 have been crystallized, and the crystals diffract to 4 and 2.6 A, respectively (Liljas et al., 1978). According to the X-ray analysis, the C-terminal fragment (positions 53- 120) has a compact, plum-shaped tertiary structure with three a helices and three p sheets (Leijonmarck et al., 1980). The only other E. coli ribosomal protein whose crystallization has so far been reported is L29 (Appelt et al., 1981). On the other hand, attempts to crystallize ribosomal proteins from the thermophilic Bacillus stearothermophilus have been more successful. Protein BL17, which according to its amino acid sequence (Kimura et al., 1980) corresponds to protein L9 from the E. coli ribosome (Kimura et al., 1982), was the first intact ribosomal protein to give crystals useful for X-ray structural analysis (Appelt et al., 1979). Several other B . stearothemnophzlus ribosomal proteins, namely BL6 and BL30 (Appelt et al., 1981,1983) from the large and BS5 (Appelt et al., 1983) from the small subunit have been crystallized, and the determination of their three-dimensional structure at a resolution of better than 3 A is now in progress. Furthermore, crystals of a B. stearothenophilus ribosomal protein complex, which corresponds to the complex (L7/L12)4 L10 from E. coli ribosome, have been obtained (Liljas and Newcomer, 1981).

E . Shape 1 . Methods

To determine the shape of ribosomal proteins in solution, ultracentrifugation, digital densimetry, viscosity, gel filtration, quasi-elastic light scattering, and small-angle X-ray or neutron scattering have all been used. With each technique it is possible to obtain a physical characteristic of the protein. Combining these techniques should allow the size and shape of the protein to be characterized quite well. However, the values determined in various laboratories for the same ribosomal proteins differ considerably. To help understand some of the reasons we will initially discuss each method briefly as it relates to proteins and then review the size and shape of the ribosomal proteins that have been so characterized. a. Analytical Ultracentrifugation. Analysis of proteins by analytical ul-

16

L. GlRl ET AL.

tracentrifugation can provide valuable information on the size and shape of the proteins. Sedimentation velocity experiments give sedimentation coefficients that are inversely related to the frictional coefficients, and they can be used to assess the amount of axial asymmetry of the protein. However, this measurement is quite difficult because many ribosomal proteins are relatively small and diffuse rapidly, thereby giving a broad concentration gradient. Diffusion coefficients can also be obtained from sedimentation velocity experiments, but the precision is quite low and subject to some question. Quasi-elastic light scattering is a much more useful technique to obtain diffusion coefficients (see below). The absolute molecular weight of the protein in solution (with associated salt) can be determined using sedimentation equilibrium. Since the sequences of all E. coli proteins are now known (Wittmann-Liebold, 1980b; Wittmann, 1982), these values may now provide some indication of the amount of salt associated with the proteins. One critical piece of information to be gleaned from sedimentation equilibrium experiments is that of sample monodispersity. Using a point-average molecular weight analysis, the molecular weight of the sample at each point within the cell may be calculated. From these values aggregates are easily discernible and the amounts can be readily estimated. This information is of great value in evaluating sample quality. b. Densimetry. The buoyancy term (1 - Op) or its equivalent is present in sedimentation-velocity and sedimentation-equilibrium equations and in the molecular-weight equations used with small-angle X-ray or neutron scattering. The accurate measurement of the density of the solution is critical in the determination of the partial specific volume 3. Whereas the apparent ii for proteins can be estimated with reasonable accuracy using the specific volumes of the constituent amino acids, one can determine fi directly to an accuracy of 2 parts in lo6 using digital densimetry (Kratky et ul., 1969). An accurate determination of the protein concentration is necessary for this measurement. c. Vkcometry. Viscosity measurements represent a simple hydrodynamic method for the study of macromolecules. The viscosity of a liquid in general is increased by the presence of a solute. The magnitude of increase depends on the concentration, hydrated volume, and structural properties of the solute particles. Viscosity measurements have provided useful shape information on several ribosomal proteins. However, this is not a preferred method because it requires a large quantity of sample and assumes the protein molecules to be ellipsoids of equivalent volume and molecular mass which realistically may not be the case. d. Gel Filtration. Hydrated volumes of particles can be measured

RIBOSOMAL PROTEINS

17

quickly by determining the partition coefficient using gel filtration. This method assumes that the partition coefficient is a reliable estimate of the protein volume. Although this method requires relatively large amounts of sample, it is a direct approach to the determination of the approximate shape of the macromolecule. e. Quasi-Elastic Light Scattering. This method is one of great power and versatility. By impinging a laser beam on the protein solution, timedependent fluctuations of the scattered intensity can be measured and used in the calculation of the diffusion coefficient. Since the diffusion coefficient is inversely proportional to the frictional coefficient, a measurement of the axial ratio and/or hydration is immediately available. This method is precise to 2-3% and is therefore the method of choice to provide axial ratios. Another significant advantage of quasi-elastic light scattering is the ability to discriminate sample quality. Aggregates of only a few percent can be discovered and by various computational methods excluded from the diffusion determination. f. Small-Angle X-Ray Scattering. This powerful technique has been used considerably to provide data on ribosomal proteins. The radii of gyration can be measured directly, and through curve fitting the shape of the particle can be estimated. Inasmuch as the scattered intensity is proportional to the square of the number of scattering centers, aggregates can skew the curves disproportionately. Unlike quasi-elastic lightscattering measurement, it is not easy to discern or remove the effects of aggregates. Therefore, it is highly dependent on the quality of the sample. Under appropriate conditions, it provides direct and valuable information about the size and shape of the studied protein. g. Small-Angle Neutron Scattering. Although this technique and smallangle X-ray scattering are fundamentally the same, neutron scattering offers the advantage of contrast variation. This method allows the buffer density to be increased by D20 so as to give a progressive decrease of the net scattering of the particle. Using this method, upon extrapolation it is possible to obtain the radius of gyration at infinite contrast, which gives a more precise value for the radius of gyration. However, due to lower flux densities, the neutron-scattered intensity is significantly lower than that of X-ray scattering.

2. Results on Ribosomal Proteins The physical characterization of ribosomal proteins has presented an exceptional challenge since they are difficult to purify, are relatively insoluble in aqueous solutions, and have a great propensity to aggregate. Polydispersity is an especially severe problem that has not been well addressed in most studies to date. Aggregation can cause substantial

18

L. GlRI E T AL.

errors, especially in scattering studies. While gel electrophoresis can indicate the presence of predominant species, small amounts of contaminants or aggregates may not be apparent. Careful sedimentation-equilibrium runs or quasi-elastic light-scattering studies will provide definite estimates of the amount of heterogeneity or polydispersity. The method of preparation of the proteins is of great importance. Not only does it affect the amount of polydispersity, but the conformation of the proteins themselves varies depending upon the conditions of preparation. Urea-treated proteins can be renatured if sufficient care is taken to prevent precipitous changes in ionic conditions. Proteins have also been prepared using LiCl which avoids the harsh effects of urea. In most of the physical studies made to date there is little evidence of sufficient care being taken to assure sample quality. Especially pronounced is the effect this has on scattering results; curvature of the Guinier plots (which provide the radius of gyration) due to polydispersity allows only an approximation of the radius of gyration at best. As a result of the problems with protein purity, preparation methods, and sample analysis, there are many discrepancies in the values obtained from various studies. Some of these will be emphasized as we review a few of the more noteworthy proteins specifically. a. Protein S4. Of all the proteins studied to date perhaps protein S4 is most central in the structural controversy as to whether proteins are spherical or elongated in shape. According to early studies using both hydrodynamic and scattering methods, protein S4 was concluded to be very elongated (Rohde et al., 1975; Paradies and Franz, 1976; h e r b e r g et al., 1977a).On the other hand, more recent neutron-scattering studies showed S4 to have a radius of gyration of 17.5-19.5 A suggesting an almost uniform sphere having 0.4 g HnO/g protein hydration (Serdyuk et al., 1980).Another study using neutron scattering indicated a value of 3 1 A for the radius of gyration of protein S4 in situ (Ramakrishnan et al., 1981). This value is similar to that (26 A) found by Gulik et al. (1978) from X-ray scattering. Preliminary results from quasi-elastic light scattering also suggest a similar value (Dodd and Hill, unpublished observation). Although additional studies on S4 under well-defined conditions are essential in order to specify its size and shape more precisely, this protein does not appear to be spherical. b. Protein S1. This protein is the largest of all ribosomal proteins, is very asymmetric, and it appears to have many functions (Subramanian, 1983).Protein S1 is relatively easy to prepare and in so doing it is unnecessary to use a high salt concentration or denaturants such as urea or guanidine hydrochloride. Also, S1 can be concentrated to above 5 mg/ ml without apparent aggregation. These features coupled with its rela-

RIBOSOMAL PROTEINS

19

tively large molecular weight (6 1,000) have made it readily amenable to physical studies. The four physical studies that have been made (Laughrea and Moore, 1977; Giri and Subramanian, 1977; Osterberg et al., 1978; Labischinski and Subramanian, 1979) all show this protein to be very extended, having an axial ratio of about 10: 1. The longest dimension is about 250 A and the shortest about 10 A. The long dimension of S1 is equivalent to the full length of the 30 S ribosomal subunit. Putative models for the structure of S1 suggest a “V” shape (Osterberg et al., 1978) or an elliptical cylinder (Labischinski and Subramanian, 1979). NMR and fluorescence-polarization studies suggest a flexible structure for this protein (Moore and Laughrea, 1979; Chu and Cantor, 1979). c. Other 30 S Proteins. A compilation of the physical data obtained on the 15 30 S ribosomal proteins that have been studied to date is given in Table V. At first perusal the data suggest considerable dissimilarity between the various proteins. However, upon closer analysis it may be possible to group the proteins into three groupings for the sake of discussion. 1. There are five proteins that are quite globular in conformation, namely S6, S8,S13, S15, and S16. Of these, S8 is apparently the most spherical, with S16 following closely behind. Proteins S6, S13, and S15 can be considered somewhat asymmetric, but hardly elongated. For proteins S6 and S13 only hydrodynamic data, which merely indicate a low axial ratio for the particle, are available. Proteins S8, S15, and S16 have all been studied using both small-angle X-ray (Osterberg et al., 1978) and small-angle neutron scattering (Serdyuk et al., 1978; Gogia et al., 1979). However, the results are quite disparate. The X-ray studies gave radii of gyration (Rg)of 21-27 A, whereas neutron scattering gave values of 12-13 A for the three proteins. This is a remarkable disparity. It should be noted that for a protein of molecular weight 14,000, the anhydrous sphere would have an R , of 12.3 A. Assuming 0.3 g HpO/g protein, the R , would be 13.6 A if the particle were completely spherical. These approximations would indicate that the neutron-scattering values are too low. On the other hand, some curvature in the Guinier region of the X-ray-scattering data suggests that aggregates may have been present which would affect the values. If one calculates theflfo values using X-ray data for protein S8 for instance, one obtains 5 : 1 instead of the 3 : 1 value obtained from the hydrodynamic studies. 2. Proteins S2, S3, S5, and S21 can be grouped into the moderately elongated category. Axial ratios of about 5 : 1 are common to this group

TABLE V H$roa)namic Propcrhcs of E coli Ribosomal Proteins Molecular weight

hl 0

Dmw

preparation"

From sedimentation equilibrium

From sequence

s2

A A D

63,300-66,500 68,000 29,200

61.159 26,613

3.2 3.2 2.36

s3

C

25,800

25,852

2.1

B

22,400 25,000 18.000 17,000 15,700 16,400 18,700 13,400 10,200

23,138 17,515 15,704 19,732 13,996 10,001

Method of Proteins

s1

s4 s5 S6

s7 s8 S15

C B C B B B

D C

-

x 107 (cm2/sec)

Sp0.w

x 1013

171*

Axialc

(ml/g)

ratio

-

10:lPE 10: 1 PE 5.4 PE 7.5 OE 5 : 1 PE 6:lOE

-

f

0.1

1.65 f 0.1 1.95 1.49 2 0.23 1.44 1.75 f 0.1 1.66 1.66 f 0.1 1.62 1.5

-

4.5 7.4

9.8 6.4

7.4 2 0.5

5.8

3.7

-

8.0

-

-

9.7 13.1

f

0.2

12.5

-

7.3

-

2.5-3.5 6.0

-

15:l PE 6:lPE

-

2.3 : 1 5:lPE

flfo

Referenced

1.55 1.55 1.45

b

1.46; 1.26

d

1.67 1.69 1.56 1.47 1.39 1.45; 1.25 1.57 1.17 1.26

e f e g e h e

a C

I

j

S18 s20 L1 L3 L6 L7U2 L9 Lll L24 L25 L29 L30

t?

B C B D D D C A D D D D

B D B

8,900 9,OOO 14,200 25,000 24,500 19,Ooo 24,000

-

17,300 16,600 12,000 12,000 6,900

8,896

-

9,553 24,599 22,258 18,832 24,398

-

6,000

17,014 14,874 11,184 10.6% 7,274 6,411

6,300

-

1.3 1.15 f 0.05 2.0 f 0.2 2.1 f 0.1 1.78 1.4 1.6 f 0.02 1.8 f 0.15 1.7 1.5 f 0.05 1.8 f 0.1 1.17 1.0 f 0.1 1 .o

10.5

10.0

8:1 PE

-

-

-

8.4 f 0.8 7.6 9.2 5.6

4.0 f 0.3 6.0 2 0.4 4.6-5.0 27.7 28 f 0.4 5.5 f 0.3 4.7 4.0 f 0.2 3.6 f 0.2

4.5:1 PE 5: 1 PE 3: 1 PE 3:l PE

3.2 f 0.1

2.5 : 1 PE

-

9.0 f 1.0 7.4 11.0 14.0 f 0.9 15.3 15.0 f 1 15.6

-

5:l PE 5.5:1 PE 4.2:1 PE 19:1

-

-

1.4 1.45 1.8 1.3 1.46 1.1 1.97-2.01 1.90 1.2 1.4 1.15 1.15 1.o 1.15 1 .o

h j e k 1 1

m n

k 0

1

k P

k

P

Method of preparation: A, NH&l wash and chromatography; B, acetic acid and urea followed by renaturation; C, acetic acid and urea; D, salt extraction and absence of urea. * [q]Denotes intrinsic viscosity. Abbreviations: PE, prolate ellipsoid; OE, oblate ellipsoid. References: a, Laughrea and Moore (1977);b, Gin and Subramanian (1977);c, Georgalis et al. (1981);d, Franz et al. (1979);e, Rohde et al. (1975);f, Paradies and Franz (1976);g, Georgalisand Gin (1978);h, Georgalis and Giri (1980);i, Gin et al. (1977); j , Giri and Franz (1978);k, Gin et al. (1979);I. Gin and Dijk (1979);m, Wong and Paradies (1974);n, Kar and Aune (1981);0, Gin et al. (1978);p. Gudkov et al. (1982).

22

L. GlRI ET AL.

of proteins. Although globular, they are nonetheless asymmetrical. The data for proteins S2, S5, and S2 1 are derived from hydrodynamic studies (Georgalis and Giri, 1978; Georgalis et al., 1981). Of these only S2 and S5 have been fully characterized, both showing a frictional coefficient of 1.45. Protein S5 has been analyzed in both T M K buffer (0.35 M KCl, 2 mM MgC12, 30 mM Tris-HC1, pH 7.4) and in low-salt buffer (0.04 M KC1, 1 mM Tris-HC1, pH 7.0), and has been shown to be somewhat more asymmetric in the low-salt buffer (Georgalis and Giri, 1978). Protein S3 has been studied by small-angle X-ray scattering and hydrodynamically (Gulik et al., 1978; Franz et al., 1979). From these studies protein S3 appears to have an axial ratio of 4 : 1 or 5 : 1. 3. The remaining five proteins that have been analyzed, namely S1, S4, S7, S18, and S20, fall into the category of extended structures. Protein S18 appears quite extended even though its mass is but 9000 d (Georgalis and Giri, 1980; Giri and Franz, 1978). Protein S20 has been shown to have an axial ratio greater than 8 : 1 by both sedimentation analysis (Rohde at al., 1975) and small-angle X-ray scattering (Gulik et al., 1978). Protein S7 has been analyzed using both X-ray (Gulik et al., 1978) and neutron scattering (Serdyuk et al., 1978), and as with some of the other proteins, there is a large discrepancy between the reported R , values. The value from neutron scattering is quite close to that calculated for an anhydrous sphere. This is markedly different from the almost 8 : 1 axial ratio deduced from the results of the other studies.

d. Protein L7/L12. One of the most intensively studied ribosomal proteins is a complex containing proteins L7 and L12 which are identical except for the presence of an acetyl group on the N-terminal end of L7. This complex also plays a unique structural role in that it forms the “L7/ L12 stalk,” a thin protuberance on the 50 S subunits. The L7/L12 complex apparently occurs as a tetramer of L7/L12 in conjunction with L10 (Pettersson et al., 1976; Dijk et al., 1977), and it is now believed that all four copies occur in a parallel fashion (Liljas, 1982) in the stalk although there are some data to the contrary (Dijk et al., 1983b). The physical studies from several groups have all produced similar results. The L7/L12 complex is very elongated, having an axial ratio of at least 10 : 1. This finding has been substantiated by scattering studies giving radii of gyration of about 40 (Wong and Paradies, 1974; Osterberg et al., 1976a; Serdyuk et al., 1980), and by hydrodynamic studies showing high viscosities and large axial ratios (Wong and Paradies, 1974; Kar and Aune, 1981). The C-terminal fragment of L7/L12 has been crystallized and studied by X-ray structural analysis (Leijonmarck

RIBOSOMAL PROTEINS

23

et al., 1980). The picture emanating from all of these studies is that of an elongated structure over 100 A in length and having some more densely packed regions at the C termini. e. Other 50 S Proteins. The other 50 S proteins studied by physical techniques are not as elongated as L7/L12 (Table V). Several of them (Ll, L6, L9, L24, L25, and L30) are quite globular in structure. Proteins L3, L11, L18, and L26 (=S20)appear to be more elongated than the ones in the previous group. According to gel-filtration experiments conducted by P. Wills and J. Dijk (personal communication), proteins L17, L25, L28, L29, and L30 are compact; L1, L4, L5, L6, L13, L16, L19, and L24 are moderately elongated; and L2, L3, L9, L11, L15, L23, L27, L32, and L33 are quite extended. A discrepancy between these results and those mentioned earlier is protein L9 which appears to be globular from hydrodynamic measurements (Giri et al., 1979), but the Stokes radius calculated from gel-filtration experiments was found to be quite large, suggesting an elongated particle. From the data described above, it is apparent that the shape of the proteins on both subunits can apparently range from very compact (e.g., proteins L6, L24, S8, S15) to very elongated (e.g., S1, L7/L12), with many structures intermediate. As discussed above, it appears from physical studies, especially with the NMR technique, that the tertiary structure of ribosomal proteins isolated in the presence of 6 M urea and then carefully renatured under appropriate conditions is very similar to those proteins prepared in the complete absence of urea. The question can be raised as to whether the structure of the proteins within the ribosomal particle is the same as in the isolated state. The only direct evidence we have that the structures of proteins are not changed upon incorporation into the subunit is provided by the neutron-scattering studies of Nierhaus et al. (1983b). They showed that individual proteins in solution had radii of gyration indistinguishable from those obtained from their counterparts on the ribosomal subunits in the same buffers and under identical preparation conditions. 111. RIBOSOMAL RNAs

A. Primary and Secondaly Structures

Bacterial ribosomes contain three RNA molecules: the 5 S, 16 S, and 23 S RNAs. The nucleotide sequence of the 5 S RNA with 120 nucleo-

"BERLIN"

"

C AL IFOR N IA "

"

STRASBOURG "

" F

b

C

u

RIBOSOMAL PROTEINS

25

tides was published in 1967 by Brownlee et al., but the elucidation of the complete primary structures of the 16 S RNA with 1542 nucleotides (Brosius et al., 1978; Carbon et al., 1978) and of the 23 S RNA with 2904 nucleotides (Brosius et al., 1980) was accomplished only after rapid and powerful sequencing techniques became available. The knowledge of the primary structure was the basis for the construction of models of the secondary structure of the RNA molecules. Different approaches have been used in several laboratories to get experimental support for developing secondary structure models: for example, chemical modification of the RNA, treatment with single- or double-strand-specificnucleases, intramolecular RNA cross-linking, isolation and sequence analysis of double-stranded RNA, and, last but not least, comparison of ribosomal RNA sequences from different organisms (reviewed by Brimacombe et al., 1983). In this way three models each for the secondary structures of the 16 S RNA (Woese et al., 1980; Glotz and Brimacombe, 1980; Stiegler et al., 1981) and for the 23 S RNA (Glotz et al., 1981; Branlant et al., 1981; Noller et al., 1981) have been proposed. As illustrated in Fig. 2 for the 16 S RNA, there is good agreement among the models of the various groups although different methods were used by the groups to construct the models. When the ribosomal RNAs from a wide range of organisms were compared it became obvious that the secondary RNA structure is more conserved than the primary structure. The different lengths of the RNA strands from different organisms are caused by insertions or deletions (“amputations”) of certain structural RNA elements such as stems and loops, or by shrinkage of the whole domains (Fig. 3). In many cases, a base-paired stem in the E. coli ribosomal RNA also remains a stem in the ribosomal RNA of other organisms (e.g., yeast), although the nucleotide sequences within the corresponding regions of the two species differ drastically. This is achieved by compensating base changes; for example, an A-U base pair in a given stem of the E. coli RNA is replaced by a G-C pair in the corresponding site of the yeast RNA. Examples are known where the majority of the base pairs within a given stem are such compensating pairs, apparently to conserve the structurally and/or functionally important stem character of the particular RNA region. FIG.2. Secondary structure of the E. coli 16 S RNA. Model “Berlin”according to Glotz and Brimacombe (1980); “California”according to Woese et al. (1980); “Strasbourg”according to Stiegler ei al. (1981). For details see Brimacombeet al. (1983). Here and in Figs. 3 and 13, section a of each structure includes the 5’ end of the molecule, which is marked; b includes the middle portion; and c includes the marked 3’ end. Arrows indicate connections between a, b, and c. Reproduced with permission from Brimacombe et al. (1983).

HUMAN MITOCHONDRION 12s rRNA m

b

E.COLI 16s rRNA

YEAST185 r RNA am.-

b

-

Ua

FIG.3. Secondary structure comparison between the RNA molecules from the small ribosomal subunits of human mitochondrion, E . coli, and yeast (Brimacombe, 1983). Relations between a, b, and c are as in Fig. 2.

27

RIBOSOMAL PROTEINS

B . Spatial Arrangement of the RNAs in Situ For the elucidation of the spatial arrangement of the 16 S and 23 S RNA strands within their cognate ribosomal subunits, several approaches have been used. Two of them, namely (immune) electron microscopy and RNA-RNA cross-links, will be described here. Several attempts have been made to elucidate the shape of the isolated 5 S RNA (Tesche et al., 1980; Sieber et al., 1980), the 16 S RNA (Vasiliev et al., 1978; Sieber et al., 1980; Edlind and Bassel, 1980; Boublik et al., 1982), and the 23 S RNA (Sieber et al., 1980; Edlind and Bassel, 1980; Vasiliev and Zalite, 1980; Boublik et al., 1982) by direct visualization in the electron microscope. In some studies (Vasiliev et al., 1978; Vasiliev and Zalite, 1980), both the 16 S and the 23 S RNAs have been seen in a specific and compact structure. According to these results, the isolated 16 S RNA has a V-like structure, and the isolated 23 S RNA has a size and shape which can be accommodated within the 50 S subunit. However, the conclusion that the shape and size of the in situ RNA and of the isolated RNA are very similar has been questioned by other electron microscopy results (Boublik et al., 1982), and it is also in disagreement with physical studies (Tam et al., 1981a,b; Robakis and Boublik, 1981). The locations of several characteristic regions of the ribosomal RNA strands within their subunits have been mapped by immune electron microscopy. In this way the positions of the following regions on the 16 S RNA were determined (Fig. 4): its 5’ end (Mochalova et al., 1982), the

-

BASES 925 1395 I PROTEINS 51. 59. 510,513,5191

_. 2090-2200 BASES I PROTEIN Ll I

-

165 RNA

N7-METHYL GUANOSINE

13’- EN01

305 SUBUNIT

50s SUBUNIT

FIG.4. Location of RNA regions in the small and large ribosomal subunits of E . coli. For references see text. Reproduced with permission from Wittmann (1983).

28

L. GlRI E T AL.

N7-methylguanosine at position 526 (Trempe et al., 1982), the N6-dimethyladenosines at positions 1517 and 1518 (Politz and Glitz, 1977; Stoffler and Stoffler-Meilicke, 198l), and its 3’ end (McKuskie-Olson and Glitz, 1979; Shatsky et al., 1979; Luhrmann et al., 1981). Similarly, the 3’ ends of the 5 S RNA (Shatsky et al., 1980a; Stoffler-Meilickeet al., 1981)and of the 23 S RNA (Shatsky et al., 1980b; Stoffler-Meilicke et al., 1981) were localized. The locations of the RNA regions mapped on the small and large ribosomal subunits are valuable fixed points when attempting to elucidate the spatial arrangement of the ribosomal RNAs in situ. As mentioned above, there exist reliable models for the secondary structure of both the 16 S and the 23 S RNAs. Information about the three-dimensional packing of the RNA strands within the subunits can be obtained by producing cross-links between RNA regions which are distant in the secondary structure model. If many such cross-links were produced and identified it should be possible to fold the RNA strand into three dimensions in the ribosomal model. T o this end, a number of intra-RNA cross-links induced by UV irradiation have been localized by sequence analysis (Zwieb and Brimacombe, 1980; Glotz et al., 1981; Stiege et al., 1982). Further cross-links within the 16 S RNA caused by intercalating psoralen derivatives were investigated by electron microscopy (Wollenzien et al., 1979; Thamrnana et al., 1979; Wollenzien and Cantor, 1982a,b) in order to identify the approximate positions along the 16 S RNA molecule of the cross-linked regions. The resolution of this approach has recently been improved by applying sequencing techniques (Turner et al., 1982) instead of electron microscopy. However, the results obtained from all cross-linking studies are still too sparse to allow a precise packing of the RNA strands within their cognate subunits. IV. TOPOGRAPHY OF THE RIBOSOME Determination of the spatial juxtaposition of the proteins and RNA within the ribosomal subunits is a major goal in ribosome research. Key functional characteristics can be clearly understood only if the ribosome topography is known. As our understanding of the structure of the individual components increases, there is an increasing need to piece together the structure to provide the functional understanding desired. A. Size and Shape of Ribosomal Subunits In order to place the ribosomal components in the appropriate topography, the size and shape of the ribosomal subunits must be defined.

RIBOSOMAL PROTEINS

29

This has been done mainly using two techniques, electron microscopy and small-angle scattering. For the 30 S ribosomal subunit, the electron micrographs show an asymmetric model in which there is a “head” region comprising about one-third and a “base” region comprising the other two-thirds of the particle. These regions are separated by a constriction or neck. There are various models of the 30 S subunits varying mainly in the degree of asymmetry and in some structural details such as the depth of the cleft and the number and shape of the protuberances. This is illustrated in Fig. 5. From small-angle X-ray scattering in solution, the 30 S subunit is estimated to be a reasonably uniform, 4 : 1 oblate ellipsoid having a long dimension of about 22 nm (Hill et al., 1969; Smith, 1971). This shape differs considerably from that observed in electron micrographs, and this discrepancy has not been fully resolved. Small-angle solution-scattering curves cannot predict a unique model, but it can be judged whether ribosomal models obtained by other methods are consistent with the scattering curves. Comparisons of the calculated scattering curves from electron microscopy 30 S models, with the experimental curves for these particles, have shown some agreement (Spirin et al., 1979; Tardieu and Vachette, 1982),but the fits are not as good as those obtained with ellipsoidal models (Hill and Fessenden, 1974).It should be kept in mind that electron micrographs are made on particles which have been dried and treated with an acidic staining solution. Since ribosomes are highly hydrated, adverse effects from the preparative methods used for electron microscopy are very likely. The 50 S subunit as seen by electron microscopists is an asymmetric, trinodal particle. Three protuberances-the so-called L7/L12 stalk, the nose, and the L1-shoulder-appear on a hemispherical body. The models proposed by various groups are shown in Fig. 5. The small-angle X-ray-scattering results with 50 S subunits show a best-fit model to be a 2 : 1 ellipsoid with a long dimension of about 25 nm (Hill et al., 1969; Smith, 1971; Tolbert, 1971). However, the curves do not compare well with this ellipsoid, suggesting that the 50 S structure is quite nonuniform. On the other hand, recent neutron-scattering studies are consistent with a trinodal structure for the 50 S subunit (Stuhrmann et al., 1977). According to these results, models similar to those seen in electron micrographs (Fig. 5) approximate those obtained from neutron scattering quite well. For both the 30 S and 50 S subunits, additional neutron-scattering studies have been made to determine the distribution of the protein and the RNA within the subunits. It has been shown in both cases that the

30

L. GIRI ET AL.

FIG. 5. Models of the small and large ribosomal subunits of E . coli. (a) Kastner el al. (1983);(b) Lake (1980);(c) Boublik (1984);(d) Vasiliev (1974) and Vasiliev et al. (1983); (e) Korn et ul. (1982); (f) Spiess (1978). Reproduced with permission from Wittrnann (1983).

RIBOSOMAL PROTEINS

31

protein complement is more to the outside and the RNA more internal, both being reasonably concentric (Stuhrmann et al., 1976,1978; Serdyuk et al., 1977, 1979; Moore, 1980). The 70 S ribosome itself, as depicted in Fig. 6, appears to be rather symmetric, with the 30 S subunit lying somewhat obliquely across the 50 S subunit with the head of the 30 S subunit near the L1 shoulder on the 50 S subunit. The platform region of the 30 S subunit faces the 50 S subunit providing an interesting shielded area where the ribosomal function apparently occurs. It appears from electron micrographs that the structure of the subunits is not greatly altered upon their association to form the 70 S ribosome.

FIG. 6. Models of the E. coli 70 S ribosome. (a) Kastner et al. (1983); (b) Lake (1980);(c) Boublik (1984); (d) Vasiliev et al. (1983). Reproduced with permission from Wittmann (1983).

32

L. GIRI ET AL.

B . Crystals of Ribosomal Particles Though solution scattering and electron microscopy can provide information on the shape and size of the ribosome and its subunits, diffraction techniques, such as X-ray analysis, are expected to yield an insight into the ribosomal structure at a much higher resolution. Two-dimensional periodic organization of eukaryotic and prokaryotic ribosomes occurs under special conditions in vivo (Byers, 1967; Kress et al., 1971; Taddei, 1972; OBrien et al., 1980) or in vitro (Barbieri, 1979; Clark et al., 1982). The two-dimensional sheets have been analyzed by image-reconstruction techniques (KUhlbrandt and Unwin, 1982; Clark et al., 1982). Three-dimensional crystals have been obtained with 50 S ribosomal subunits from B . stearothermophilus (Yonath et al., 1980, 1982a,b) and with 70 S ribosomes from E. colz (Wittmann et al., 1982) as shown in Fig. 7. Image-reconstruction studies have been performed on thin sections through the 50 S crystals (Leonard et al., 1983), and at least one crystal form seems to be suitable for X-ray-crystallographic studies (K. R. Leonard, B. Tesche, H. G. Wittmann, and A. Yonath, unpublished results). C . Immune Electron Microscopy (IEM) An IgG-antibody against an individual ribosomal protein binds specifically only to this protein in a ribosomal subunit. Since the antibody is divalent it can form a bridge between the identical proteins in two subunits, leading to a dimer that can be examined under the electron microscope. The location of the bound antibody on the subunit surface can be determined, defining the position of the antigenic determinant of a particular protein. The method relies on the fact that IgG-antibodies are able to react with specific proteins within the intact ribosomal subunits and that both subunits have discernible shapes with recognizable morphological landmarks. Since this technique was first used by Wabl (1973) for studies of ribosomes, it has been used extensively by StGffler et al. (1980) and Lake (1980) as a powerful tool by which to delineate the external topography of ribosomal subunits. Initially both groups made comprehensive preliminary studies to define the loci of many ribosomal proteins on both subunits (Tischendorf et al., 1974, 1975; Lake 1976, 1978). However, difficulties with antibody purity and specificity as well as differences in the models proposed for the ribosomal subunits caused a considerable discrepancy between the two groups. The results were reexamined, and

FIG. 7. (a) Crystals of E . cdi 70 S ribosomes. (b) and (c) Electron micrographs of sections through three-dimensional crystals shown in (a) in two orthogonaldirections (Wittmannet d.,1982). (d) and (e)Crystals and computed filtered image of a section through a crystal of B a d u s stearotlvnnophilus 50 S ribosomal subunits (Yonath et al., 1982a,b; Leonard el d.,1983). (d) and (e) are related to two different crystal forms. Reproduced with permission from Wittmann (1983).

34

L. GIRI ET AL.

there is now much more agreement for the location of the proteins, as illustrated in Figs. 8 and 9 for the 30 S and 50 S particles, respectively. IEM has also been used to map the locations of functional domains in both ribosomal subunits (Fig. 10).The current knowledge has been summarized by Wittmann (1983).

D. Neutron Scattering One of the most powerful techniques by which protein-protein neighborhoods within the ribosomal particles can be elucidated is neutron scattering. When using this method to determine the relative positions of proteins in the 30 S subunit, the particle is reconstituted with two specific proteins that are deuterated whereas all other ribosomal components are in the protonated form (Moore, 1980). The subunits containing the two deuterated proteins give additional contributions to the scattering curves which provide information on the lengths of the vectors between the two deuterated proteins. The length-distribution studies were recently extended to include an estimate of the radii of gyration of the protein in situ (Ramakrishnan et al., 1981). The constraint used was that the proteins must have a radius of gyration greater than that of an anhydrous sphere. From their results Ramakrishnan et al. (1981) conclude that only S1 and S4 show signs of an extended conformation in situ, whereas 12 proteins (S3, S5, S6, S7, S8, S9, S10, S11, S12, S14, S15, and S18) appear quite compact and globular. However, the experimental errors of these estimations are very large. Using 59 data pairs sufficient data exist to rather accurately position the 14 proteins relative to each other (Fig. 11). It is gratifying that there is very good agreement between these results and those from IEM. The method applied by Nierhaus et al. (1983a,b) for their neutronscattering studies on 50 S subunits is somewhat different from that used with 30 S subunits in that they used 5 S and 23 S rRNA and proteins which had been deuterated sufficiently (76 and 84%, respectively) to provide a density match for 100%D20 solvent. Specific protonated proteins are then reconstituted into the otherwise deuterated particle, either singly for shape determination or in pairs for distance measurements. Scattering from these particles in 100%D20 derives only from the proteins that are protonated. In this way, the shapes of proteins L1, L3, L4, and L23 have been studied in situ giving radii of gyration of 26, 22, 20, and 13 A, respectively, corresponding to axial ratios of 3 : 1 to 5 : 1. The distance between the centers of gravity were found to be about 120, 120, and 70 A for protein couples L2-L3, L2-L4, and L4-Ll6, respectively

RIBOSOMAL PROTEINS

35

FIG. 8. Mapping of proteins on the E. coli 30 S ribosomal subunit by immune electron microscopy.(a) Sttbffler-Meilicke and Stiiffler (1983); (b) Winkelmann d al. (1982).Reproduced with permission from Wittmann (1983).

36

L. GlRl ET AL.

FIG. 9. Mapping of proteins on the E. coli 50 S ribosomal subunit by immune electron microscopy. (a) Noah etal. (1983); (b) Lake and Strycharz (1981). Reproduced with permission from Wittmann (1983).

RIBOSOMAL PROTEINS I F 3 BINOING

37

CHLORAMPHENICOL

DECODING RfGlON

BINDING SITE

305 SUBUNIT 50s SUBUNIT FIG. 10. Functional domains on E . Cali ribosomal subunits as determined by immune electron microscopy. For details see Wittmann (1983). Reproduced with permission from Wittmann (1983).

(Nierhaus et al., 1983a).These data are still too sparse to allow a comparison with the results from immune electron microscopy.

E . Singlet-Singlet Fluorescent Energy Transfer between Proteins Cantor et al. (1974) introduced the possibility of using singlet-singlet energy transfer between a pair of fluorescently labeled proteins to obtain topographical information on the arrangements of ribosomal proteins. The principle of the method lies in covalently attaching fluorescent labels to proteins and then, after reassembly in vitro, to measure the efficiency of energy transfer between various protein pairs. This approach gives a measure of the relative distance between the centers of mass of the proteins concerned. The use of this approach has been further extended to study the binding site of antibiotics on the ribosome. The distance between the

0 0 0

within FIG. 11. Map of proteins wit1 iin the E . cob 30 S ribosomal subunit as determined by (Moore,:, 1980; Moore et al., 1984). Reproduced with permission neutron scattering studies (Moore from Wittmann (1983).

38

L. CIRl ET AL.

erythromycin binding site and proteins L7/L12 in the 50 S subunit has been measured (Langlois et al., 1976). Also the interaction of a fluorescent streptomycin derivative with the ribosome has been studied (Hall et al., 1977). It was further shown that the four copies of proteins L7/L12 are far away from the main body of the 50 S subunit, and the C-terminal regions of these proteins are very accessible to the solvent. The results indicate that the C terminus of L7/L12 must be more than 70 8, away from both the 3' end and the anticodon region of ribosome-bound tRNAs (Lee et al., 1981a,b). Additional fluorescence studies have been performed with proteins L6 (Steinhauser et al., 1982), L10 (Zantema et al., 1982a,b),L7/L12 (Zantema et al., 1982a,b),S1 (Odom et al., 1983),S4 (Epe et al., 1982), and S17 (Epe et al., 1982).

F. Cross-Linking By the use of'suitable bifunctional reagents, the local neighborhoods of associated macromolecules can be probed. Studies of this nature do not provide the location of a given site relative to the entire structure, but rather relative to its nearest neighbors. It should also be pointed out that most probes are specific for certain side chains and groups. Where these are lacking, there is a null test for proximity even if molecules are actually in close proximity. 1. Protein-Protein Cross-Linking Bifunctional reagents, such as bis-imido esters, have been widely used because they react under mild conditions specifically with the amino groups of proteins. Bifunctional imido esters that introduce a disulfide bond in the cross-link are especially useful since these bonds can be readily cleaved by reduction, allowing individual proteins to be regenerated. The reagent most commonly used for ribosomal proteins is 2iminothiolane (Traut et al., 1980). Ribosomal subunits are incubated with the cross-linking reagent after which the proteins are extracted, and those proteins cross-linked are identified. The identification is often difficult due to the relatively low concentration of cross-linked pairs and the limiting sensitivity of the method of assay. When using noncleavable reagents it is necessary to apply antisera to identify the cross-linked proteins (Lutter et al., 1972). With cleavable reagents, the cross-linked protein pair can be split and the individual proteins identified by a diagonal gel electrophoresis. This has led to the identification of many cross-linked components (Sommer and Traut, 1976).It has also been shown in some cases that the biological activity of the ribosomal subunits is not seriously reduced after crosslinking and that cross-linked pairs of protein could be reconstituted into

39

RIBOSOMAL PROTEINS

the 30 S subunit without impairing functional activity (Slobin, 1972; Lutter and Kurland, 1973). Many pairs of cross-linked proteins have been identified in the 30 S and 50 S subunits (Traut et al., 1980). A schematic two-dimensional arrangement of protein neighborhoods for both the 30 S and 50 S subunits is shown in Fig. 12. There is general agreement between many results from the cross-linking experiments and those from the IEM or neutron-scattering studies. For instance, cross-links between S3-S4, S3-S5, S3-SlO, S4-S5, S5-S8, S7-S9, and S8-Sl5 are in good agreement. On the other hand, cross-links such as S7-S8 and S 5 4 9 are not consistent with other topographical studies. Bifunctional reagents have also been used to identify the ribosomal proteins present on the subunit interface on both the 30 S and 50 S subunits (Lambert and Traut, 1981; Cover et al., 1981). Figure 12 summarizes the identified cross-linked pairs containing one protein from each subunit, suggesting that these proteins are located at the subunit interface. It is essential to note that cross-linking results are indicative that por-

a

b

30s CROSS-LINKS Scheme I

Schamr 2

CROSS-LINKS TO L 2

Scheme 3

FIG.12. Cross-linkingof proteins within the E . coli (a) 30 S and (b) 50 S subunits (Traut 1980). Asterisks denote proteins cross-linked to initiation factors. (c) Protein neighborhoods at the subunit interface (Lambert and Traut, 1981). Scheme 1 shows the crosslinks found in highest amount;Scheme 2 those among 50 S proteins L1 and L2 and several 30 S proteins; and Scheme 3 those between 5 S RNA binding proteins and 30 S proteins. Reproduced with permission from Wittmann (1983). et al.,

40

L. GlRI ET AL.

tions of the proteins are within moderate proximity of one another but that the centers of mass may be somewhat more distant. Since the amino acid sequences of all E. coli ribosomal proteins are now known (Wittmann-Liebold, 1980b; Wittmann, 1982), it is possible to identify those amino acid residues which are cross-linked in the protein pair. This has been accomplished for the pairs S5-S8 (Allen et al., 1979) and L7-Ll2 (Maassen et al., 1981). 2. Protein-RNA Cross-Linking Several methods have been used to cross-link proteins with RNA. The most direct is that of irradiation with ultraviolet light which generates a number of protein-RNA cross-links in the ribosome (Gorelic, 1975,a,b; Miiller and Brimacombe, 1975; Baca and Bodley, 1976; Turchinsky et al., 1978). Photoactivatable dyes, such as methylene blue (Zook and Fahnestock, 1978) and other more specialized reagents, have also been used. Most useful are bifunctional reagents containing different functional groups at either end, one reacting with RNA and the other with protein. These have the advantage that they are more specific and that each step of the reaction can be controlled. Several reagents of this nature have been developed (Fink et al., 1980; Rinke et al., 1980; Millon et al., 1980; Politz et al., 1981),each having specific advantages. Cleavable bifunctional reagents have also been applied (Baumert et al., 1978; Expert-Bezancon and Hayes, 1980), including 2-iminothiolane, the common protein-protein cross-linking reagent. This reagent is used by allowing the imido ester function to react first with protein and then using mild ultraviolet irradiation to cross-link to the RNA (Wower et al., 1981). The results showed that most, if not all, proteins can be cross-linked to rRNA, suggesting that there is much rRNA in close proximity to the proteins. Studies to determine the exact sites on the rRNA which crosslink to the proteins have led to the identification of such sites for a number of proteins within both subunits (Brimacombeet al., 1983). This is illustrated for the 16 S RNA in Fig. 13 and for the 23 S RNA in Fig. 14.

G . Protein-Binding Sites on RNA To determine the binding sites of proteins on the rRNA different methods have been used, most notably enzymatic digestion of the rRNA around the protein that protects the RNA binding site from nuclease attack. Details and results of these studies have been reviewed (Zimmermann, 1980; Wittmann, 1982; Brimacombe et al., 1978, 1983). The “binding sites” so obtained vary in length from about 50 nucleotides for protein S8 to about 500 nucleotides for protein S4. The 5’ region of the 16 S rRNA contains the sites for proteins S4, S12, and S20. The 3’

a

520

I-'

u . __-_-___-I

L J FIG.13. The secondary structure of the E . coli ribosomal 16 S RNA, showing protein-binding sites, RNA-protein cross-link sites, and intra-RNA cross-link sites. The relations between a, b, and c are represented as in Figs. 2 and 3. Reproduced with permission from Brimacornbe d al. (1983).

I-------\

f

L6

SSRN4, L5. Ll8. LZS

FIG.14. The secondary structure of the E . coli ribosomal 23 S RNA, showing protein-binding sites, RNA-protein crossiink sites, and intra-RNA cross-link sites. The relationships among a-f are as in Figs. 2 and 3. Reproduced with permission from Brimacombe et d.(1983).

RIBOSOMAL PROTEINS

43

region is bound by protein S7 with additional interaction by proteins S9, S10, S13, S19, S21, and possibly S1. The central portion contains the sites for S8 and S15 and also interacts with S6 and S18 (Fig. 13). Similar experiments on the 23 S rRNA molecule have shown binding regions for the 50 S proteins extending from about 50 nucleotides for Ll 1 to about 500 nucleotides for L24 (Fig. 14).Proteins L3 and L16 bind to the 3' end of the 23 S rRNA, whereas the 5' end interacts with proteins L4, L l l , and the (L7/L12)4 L10 complex. The central portion of the rRNA interacts with proteins L2 and L13 (Zimmermann, 1980; Brimacombe et al., 1983). Studies were done in order to precisely define the interactions taking place between specific nucleotides on the RNA and amino acids on the proteins (Maly et al., 1980). Although these analyses are extremely difficult, several such interactions have been recently characterized in this way (Wower et al., 1981). H . Fragments of Ribosomal Particles If the ribosomal subunits are treated with ribonucleases under suitably mild conditions, specific RNA-protein fragments can be isolated. Proteins found together in such fragments have been considered to be neighbors in the intact particles. This is a reasonable assumption since it is unlikely that a ribonucleoprotein complex consisting of widely separated groups of proteins joined by an unprotected RNA strand could survive the nuclease treatment. These fragments have been isolated by both gel electrophoresis and sucrose density gradient centrifugation, and their protein content has been determined by gel electrophoresis (Zimmermann, 1974; Brimacombe et al., 1978). Limited nuclease treatment of 30 S subunits yields two main fragments of unequal size; one containing proteins S4, S5, S6, S8, S15, S17, S20, and possibly S13 or S14, and the other S7, S9, S10, S19, and S13 or S14. Various smaller fragments have been isolated consisting of subsets of these two groups of proteins, the smallest one containing S8 and S15 and another one S7 and S19 (Morgan and Brimacombe, 1972, 1973; Roth and Nierhaus, 1973; Yuki and Brimacombe, 1975). A subdivision of 30 S proteins into three groups has been obtained by analyzing hydrolyzates of a protein-deficient reconstitution intermediate (RI) particle. In this case, the protein groups consisted of S4, S16, S17, and S20, with RNA from the 5' proximal region of the 16 S RNA; S6, S8, S15, and S18, with RNA from the central region; and S7, S9, S13, and S19, with RNA from the 3' proximal region (Zimmermann et al., 1974, 1975). The 50 S subunit is much more resistant to this mild nuclease treat-

44

L. CIRl ET AL.

ment than the 30 S subunit and remains intact in spite of some cuts in the 23 S RNA. However, upon loosening the 50 S subunit structure before treating with nuclease, several fragments can be obtained. These are compiled by Nierhaus (1982). It should be noted that since proteins are not covalently attached to the RNA, under low-salt conditions they may lose their specific sites of attachment to the RNA and be free to exchange positions (Newton et al., 1975). However, Spitnik-Elson and Elson (1979) have described a method to prepare specific ribonucleoprotein (RNP) fragments reproducibly. It is critical that the ionic environment remains unchanged during the preparative procedure. Using this method two RNP fragments have been isolated, and the analysis yielded additional evidence that there are long-range interactions between RNA regions far apart in the primary structure (Spitnik-Elson et al., 1982a,b).

I . Protein Complexes Among some groups of proteins there is a considerable propensity for association among themselves which may take place before the assembly process. For instance, proteins L7 and L12 are most often found together (Moller et al.! 1970, 1972) or as an (L7/L12)4-L10 complex (Pettersson et al., 1976; dsterberg et al., 1977b; Dijk et al., 1979; Tokimatsu et al., 1981). Although each of these complexes forms spontaneously in solution, they are also well defined structurally on the 50 S subunit. It is very likely that this is also true for proteins S3-S4, S3-S5-S10, S4-S5, S 5 4 8 , S5-Sl0, and S6-Sl8 which all have high association constants (Rohde and Aune, 1975; Rohde et al., 1975; Aune, 1977; Prakash and Aune, 1978; Tindall and Aune, 1981). These protein pairs agree remarkably well with nearest neighbor results emanating from neutronscattering, protein-protein cross-linking, and IEM studies. It is possible that the protein-protein interactions responsible for their association may also be functionally important in the assembly of the ribosomal subunit. J . Chemical Reactivity Another method of probing the topography of the ribosome is to use chemical reagents selective for specified groups on the proteins or the RNA. A number of such probes, such as trypsin (Spitnik-Elson and Breiman, 1971; Crichton and Wittmann, 1971; Chang and Flaks, 1971), fluorescent reagents (Spitnik-Elson et al., 1976), N-ethylmaleimide (Ginzburg and Zamir, 1976; Ghosh and Moore, 1979), kethoxal (Noller, 1980), RNase T 1 (Yuki and Brimacombe, 1975), and iodine (Michalski and Sells, 1975; Litman et al., 1976; Lam et al., 1979; Wower et al., 1983)

RIBOSOMAL PROTEINS

45

have been used with some degree of success. For instance, using iodination by lactoperoxidase, it was found that about 30 pairs of proteins could be identified which affect each other’s reactivity. These results imply a near-neighbor relationship between these proteins (Changchien and Craven, 1979). It was also noted that when iodination of proteins in isolated subunits and in 70 S ribosomes was compared, seven proteins were labeled to a greater extent and four to a lesser extent in the subunits than in the intact ribosome, indicating shielding and/or conformational changes upon subunit association (Michalski and Sells, 1975). Kethoxal reacts with guanines in single-stranded RNA regions, and this probe has been used in mapping exposed regions of the rRNA in the subunits. Upon association of the subunits, protection for some of the guanines was found whereas others became more reactive (Noller, 1980), implying that a conformational change has occurred upon association. K . Assembly of Ribosomal Subunits One of the preeminent accomplishments within the ribosomal field is the in vitro reconstitution of the 30 S subunit (Traub and Nomura, 1968). As the mechanism of the process was further refined, an assembly map was developed which defined the sequence of protein incorporation into the ribosomal intermediates. Initially about 10 proteins are bound to the RNA, and after heating the intermediate particle, the remainder of the proteins can bind (Nomura and Held, 1974). The assembly map is shown in Fig. 15. The results show a close correlation between protein proximity as deduced from immune electron microscopy, neutron scattering, and cross-linking on the one hand, and assembly interaction on the other hand. There are some near-neighbor proteins that are not related by assembly interactions, however. Nonetheless, the agreement is gratifying and suggests protein-protein interactions as being necessary for the assembly of the 30 S ribosomal subunit. The reconstitution of active E. coli 50 S subunits, in contrast to that of 50 S particles from B. stearothermophilus (Nomura and Erdmann, 1970), requires a two-step incubation procedure (Nierhaus and Dohme, 1974; Dohme and Nierhaus, 1976). The assembly process occurs in four steps from 23 S RNA to 50 S particles, leading to formation of 33 S, 4 1 S, and 48 S intermediates. The step from 33 S to 41 S consists of a compact folding of the 33 S intermediate, without addition to any protein component. This drastic conformational change has been demonstrated by biochemical and electron-microscopic studies (Sieber and Nierhaus, 1978; Sieber et al., 1980; Nierhaus, 1982). Kinetic analyses performed at

L. GIRI ET AL.

46

b FIG. 15. Assembly map of the E . coli (a) small (Nomura and Held, 1974) and (b) large (R6hl el aL, 1982) ribosomal subunits.

RIBOSOMAL PROTEINS

47

various temperatures revealed that the rate-limiting step in both incubation steps is a first-order reaction, the activation energies being 290 and 255 kJ/mol for the first and second incubation, respectively. There are at least two “assembly domains,” namely the L20 domain and the L15 domain, in the 50 S assembly map (Fig. 15).Proteins within the L20 domain are essential for the assembly but not for the function of the 50 S subunit whereas those in the L15 domain are functionally important proteins whose assembly occurs at a late state. As with the 30 S subunit, the assembly map of the 50 S subunit (Rohl and Nierhaus, 1982)not only reflects the assembly dependence but also the topographical relationship of the proteins within the ribosomal particle. This conclusion is supported by a good correspondence between the assembly map on the one hand, and results from cross-linking studies and from the sequential removal of proteins from the particle by LiCl on the other hand. There is also a correlation between the interdependence of proteins during the assembly process and the arrangement of their genes on the E. coli chromosome (Rohl et al., 1982).

V. SUMMARY AND OUTLOOK During the last 15 years of ribosomal protein study, enormous progress has been made. Each of the proteins from E. coli ribosomes has been isolated, sequenced, and immunologically and physically characterized. Ribosomal proteins from other sources (e.g., from some bacteria, yeast, and rat) have been isolated and studied as well. Several proteins have recently been crystallized, and from the X-ray studies it is expected that much important information on the threedimensional structure will be forthcoming. Many other proteins can probably be crystallized if suitable preparative procedures and crystallization conditions are found. Tremendous progress has also been made in deciphering the architecture of the ribosome. A battery of different methods has been used to provide the nearest neighbor distances of the ribosomal proteins in situ. Definitive measurements are now emanating from neutron-scattering experiments which also promise to give reasonably accurate radii of gyration of the proteins in situ. In turn, refined immune electron microscopy results supplement the neutron-scattering data and also position the proteins on the subunits themselves. This cannot be done by the other methods. Determination of the three-dimensional RNA structure within the ribosome is still in its infancy. Nonetheless, it is expected that by combin-

48

L. GIRI ET AL,

ing the data from protein-RNA and from RNA-RNA cross-linking studies, the structure of the RNA in situ can be unraveled. Of great interest is the fact that ribosomal subunits and ribosomes themselves have now been crystallized, and low-resolution structural maps have already been obtained. However, to grow suitable crystals and to resolve the ribosomal structure at a sufficiently high resolution remains a great challenge and task to biochemists and crystallographers.

REFERENCES Adler, A. J., Greenfield, N. J., and Fasman, G. D. (1973).In “Methods in Enzymology” (C. H. W. Hirs and S.N. Timasheff, eds.), Vol. 27, p. 675. Academic Press, New York. Allen, G., Capasso, R., and Gualerzi, C. (1979).J. Bwl. Chem. 254, 9800. Appelt, K., Dijk, J., and Epp, 0. (1979). FEBS Lett. 103, 66. Appelt, K., Dijk, J., Reinhardt, R., Sanhuesa, S.,White, S.,Wilson, K. S.,and Yonath, A. (198l).J. Biol. C h n . 256, 11787. Appelt, K.,White, S. W., and Wilson, K. S. (1983).J. Biol. Chem. 258, 13328. Arai, K.. Clark, B. F. C., Duffy, L.,Jones, M. D., Kaziro, Y.,Lauresen, R. A,, L‘ltalien, J., Miller, D. L., Nagarkatti, S., Nakumura, S., Nielsen, K. M., Petersen, T. E., Takahashi, K., and Wade, M. (1980). Proc. Natl. Acad. Sn‘. U.S.A. 77, 1326. Aune, K. C. (1977). Arch. Biochem. Biophys. 180, 172. Baca, 0. G., and Bodley, J. W. (1976). Biockm. Biophys. Res. Commun. 70, 1091. Barbieri, M. (1979).J. Supramol. Struct. 10, 349. BBumert, H. G., Skbld, S. E., and Kurland, C. G. (1978). Eur. J. Biochem. 89, 353. Boublik, M. (1984). In “Anatomy of Esckrichia coli” (N. Nanninga, ed.). Academic Press, New York, in press. Boublik, M., Robakis, N.,Hellmann, W., and Wall, J. S. (1982). Eur. J. Cell Biol. 27, 177. Branlant, C., Krol, A., Machatt, M. A., Pouyet, J., Ebel, J. P., Edwards, K., and Kbssel, H. (1981). N u c k A d Res. 9,4303. Brauer, D., and Wittmann-Liebold, B. (1978). FEBS Lett. 96,395. Brimacombe, R. (1982). Biochem. SOC.Spp. 47,49. Brimacornbe, R., Stbffler, G., and Wittmann, H. G. (1978). Annu. Rev. Biochem. 47, 217. Brimacombe, R., Maly, P., and Zwieb, C. (1983). Prog. Nuclcic AcidRes. Mol. Bwl. 28, 1-48. Brochon, J.-C., Wahl, P., Vachette, P., and Daune, M. P. (1976). Eur.1. Biochem. 65, 35. Brosius, J., Palmer, M. L., Kennedy, P. L., and Noller, H. F. (1978). Proc. Natl. Acad. Sci. U.S.A. 75,4801. Brosius, J., Dull, T. J., and Noller, H. F. (1980). Proc. Nafl. Acad. Sci. U.S.A. 77, 201. Brownlee, G. G., Sanger, F., and Barrel], B. (1967). Nature (London) 215, 735. Byers, B. (1967).J. Mol. Biol. 26, 155. Cantor, C. R., Huang, K., and Fairclough, R. (1974). In “Ribosomes” (M. Nomura, A. Tissihres, and P. Lengyel, eds.), p. 587. Cold Spring Harbor Laboratory, Cold Springer Harbor, New York. Carbon, P., Ehresmann, C., Ehresmann, B., and Ebel, J. P. (1978). FEBS Lett. 94, 152. Chang, F. N., and Flaks, J. G. (1971).J. Mol. Bwl. 61, 387. Chang, J. Y., Brauer, D., and Wittmann-Liebold, B. (1978). FEBS Lett. 93, 205. Changchien, L. M., and Craven, G. R. (1979).J. Mol. Biol. 113, 103. Chu, Y. G., and Cantor, C. R. (1979). N u c k Acidr Res. 6, 2363. Clark, M. W., Leonard, K.,and Lake, J. A. (1982). Science 216,999. Cotter, R. I., and Gratzer, W. B. (1969). Eur. J. Biochem. 8, 352.

RIBOSOMAL PROTEINS

49

Cover, J. A., Sommer, A., and Traut, R. R. (1981).J.Biol. Cha. 256,8222. Crichton, R. R.,and Wittmann, H. G. (1971).Mol. Gen. Genet. 114,95. Dijk, J., and Littlechild, J. (1979).In “Methods in Enzymology” (K. Moldave and L. Grossman, eds.), Vol. 59,p. 481.Academic Press, New York. Dijk, J., Littlechild, J., and Garrett, R. A. (1977).FEES Lett. 77, 295. Dijk, J., Garrett, R. A., and MUler, R. (1979).Nwleic Acids Res. 6, 2717. Dijk, J., Littlechild, J. A., Freund, A. M., Pouyet, J., Daune, M., and Provencher, S. W. (1983a).J. Biol. Chem. 258, 15021. Dijk, J., Georgalis, Y.,Labischinski, H., and Wills, P. (1983b).Biochemisty 22, 5426. Dohme, F., and Nierhaus, K. H. (1976).J.Mol. B i d . 107,585. Dzionara, M., Robinson, S. M. L.,and Wittmann-Liebold, B. (1977).H@e-Seyler’s Z. Physwl. Chem. 358, 1003. Edlind, T. D., and Bassel, A. R. (1980).J.Bactm’ol. 141,365. Epe, B.,Woolley, P.,Steinhiuser, K., and Littlechild,J. A. (1982).Eur.J. Baochem. 129,211. Expert-Benzancon, A., and Hayes, D. (1980).Eur. J. Biochem. 103,365. Fink, G., Fasold, H., Rommel, W., and Brimacombe, R. (1980).Anal. Biochem. 108, 394. Franz, A., Georgalis, Y.,and Giri, L. (1979).Biochim. Biophys. Acka 578, 365. Georgalis, Y.,and Giri, L. (1978).FEES Lett. 95,99. Georgalis, Y.,and Giri, L. (1980).Biochem. Int. 1, 105. Georgalis, Y.,Giri, L., and Littlechild, J. (1981).Biochemistry 20, 1061. Gerard, D., Lemieux, G., and Laustriat, G. (1975).Photochem. Photobiol. 22, 89. Ghosh, N.,and Moore, P. B. (1979).Eur. J . Biochem. 93, 147. Ginzburg, I., and Zamir, A. (1976).J.Mol. Bwl.100,387. Giri, L.,and Dijk, J. (1979).Arch. Biochem. BiOphys. 193, 122. Giri, L.,and Franz, A. (1978).FEBS Lett. 87,31. Giri, L., and Subramanian, A. R. (1977).FEES Lett. 81, 199. Giri, L., Littlechild, J., and Dijk, J. (1977).FEES Lett. 79,238. Giri, L., Dijk, J., Labischinski, H., and Bradaczek, H.(1978).Biochemistq 17, 745. Giri, L., Franz, A., and Dijk, J. (1979).Biochemitq 18, 2520. Glotz, C.,and Brimacombe, R. (1980).Nucleir Acidc Res. 8, 2377. Glotz, C., Zwieb, C., Brimacombe, R., Edwards, K., and Kisssel, H. (1981).N w h Acids Res. 9,3287. Gogia, Z. V., Venyaminov, S. Yu.,Bushuev, V. N., Serdyuk, I. N., Lim, V. I., and Spirin, A. S. (1979).FEES Lett. 105,63. I Gorelic, L. (1975a).Biochemkty 14,4627. Gorelic, L. (1975b).Biochim. Blophys. Acta 390, 209. Gudkov, A. T., and Behlke, J. (1978).Eur. J.Biochem. 90,309. Gudkov, A. T., Venyaminov, S.Y.,Behlke, J., and Bushuev, V. N. (1982).FEES. Lett. 141, 254. Gulik, A., Freund, A. M., and Vachette, P. (1978).J.Mol. Biol. 119,391. Hall, J., Davis, J. P., and Cantor, C. R. (1977).Arch. B i o c h a . Biophys. 179, 121. Hill, W. E., and Fessenden, R. J. (1974).J.Mol. Biol. 90,719. Hill, W. E., Thompson, J. D., and Anderegg, J. W. (1969).J.Mol. Biol. 44,89. Hitz, H., Schifer, D., and Wittmann-Liebold, B. (1977).Eur. J. Biochem. 75,497. Isono, K. (1980).In “Ribosomes: Structure, Function, and Genetics” (G. Chambliss, G. R. Craven, J. Davies, K. Davis, L. Kahan, and M. Nornura, eds.), p. 641.Univ. Park Press, Baltimore, Maryland. Kaltschmidt, E. (1971).Anal. Biochem. 43, 25. Kar, E. C., and Aune, K. C. (1981).Biochemitq 20,4638. Kastner, B., StaMler-Meilicke, M., and Staffler, G. (1981).Proc. Natl. Acod. Sci. U.S.A. 78, 6652.

50

L. GIRl ET AL.

Kastner, B., StbfRer-Meilicke,M., and Sttimer, G. (1983).Electron Microsc. Int. Congr. loth, 1982 3,105. Khechinashvili, N. N., Koteliansky, V. E., Gogia, Z. V., Littlechild, J., and Dijk,J. (1978). FEBS Lett. 95,270. Kime, J. M., Ratcliffe, R. G . , Moore, P. B., and Williams, R. J. P. (1980). Eur. J. Biochem. 110,493. Kime, J. M., Ratcliffe, R. G., Moore, P.B.,and Williams, J. P. (1981).Eur.J. Biochem. 116, 269. Kimura, M., Dijk, J., and Heiland, I. (1980). FEES Lett. 121, 323. Kimura, M., Mende, L., and Wittmann-Liebold, B. (1982). FEBS Lett. 149, 304. Korn, A. P., Spitnik-Elson, P., and Elson, D. (1982).J. Biol. Chem. 257, 7155. Kratky, O., Leopold, H., and Stabinger, H. (1969). 2.Angm. Phys. 27, 273. Kress, Y., Wittner, M., and Rosenbaum, R. M. (1971).J . Cell Biol. 49, 773. Kiihlbrandt, W., and Unwin, P. N. T. (1982).J. Mol. Eiol. 156,431. Labischinski, H., and Subramanian, A. R. (1979). Eur. J . Biochem. 95, 359. Lake, J. A. (1976).J. Mol. Biol. 105, 131. Lake, J. A. (1978). I n “Advanced Techniques in Biological Electron Microscopy 11” (J. Koehler, ed.), p. 173. Springer-Verlag, Berlin and New York. Lake, J. A. (1980).In “Ribosomes: Structure, Function, and Genetics” (G. Chambliss, G. R. Craven, J. Davies, K. Davis, L. Kahan, and M. Nomura, eds.), p. 207. Univ. Park Press, Baltimore, Maryland. Lake, J. A., and Strycharz, W. A. (198l).J. Mol. Biol. 153, 929. Lam, M. K. T., Changchien, L. M., and Craven, G. R. (1979).J. Mol. Biol. 128, 561. Lambert, J. M., and Traut, R. R. (1981).J. Mol. Biol. 149,451. Langlois, R., Lee, C. C., Cantor, C. R., Vince, R., and Pestka, S. (1976).J. Mol. Biol. 106, 297. Laughrea, M., and Moore, P. B. (1977).J. Mol. Biol. 112, 399. Lee, C. C., Cantor, C. R., and Wittmann-Liebold, B. (1981a).J. Biol. Chem. 256,41. Lee, C. C., Wills, B. D., Fairclough, R. H., and Cantor, C. R. (1981b).J. Biol. Chem. 256,49. Leijonmarck, M., Eriksson, S.,and Liljas, H. (1980). Nature (London) 286, 824. Leonard, K. R.,Arad, T., Tesche, B., Erdmann, V. A., Wittmann, H.G.,and Yonath, A. E. (1983). Electron Mirrarc. Int. Congr. loth, 1982 $9. Liljas, A. (1982). Prog. EiopnYr. Mol. Bwl.46, 161. Liljas, A., and Newcomer, M. E. (1981).J. Mol. Biol. 153, 393. Liljas, A., Erikson, S., Donner, D., and Kurland, C. G . (1978). FEBS Lett. 88, 300. Litman, D. J., Beckman, A., and Cantor, C. R. (1976). Arch. Biochem. Blophys. 174, 523. Littlechild, J., and Malcolm, A. L. (1978).Biochembhy 17, 3363. Littlechild, J., Malcolm, A., Paterakis, K., Ackermann, I., and Dijk, J. (1983).J. Biol. Chem. 458, 15027. Luer, C. A., and Wong, K.-P. (1980). Biochembtry 19, 176. Liihrmann, R., Bald, R., Stl)mer-Meilicke, M., and Sttimer, G. (1981).Proc. Natl. Acad. Sci. U.S.A. 78, 7276. Lutter, L. C., and Kurland, C. G. (1973). Nature (London) New Eiol. 243, 15. Lutter, L. C., Zeichhardt, H., Kurland, C. G., and Stamer, G. (1972).Mol. Gen. Genet. 119, 357. Maassen, J. A., Schop, E. N., and Mbller, W. (1981). Biochemistry 20, 1020. McKuskie-Olson, H., and Glitz, D.G. (1979). Proc. Natl. A c d . Sci. U.S.A. 76, 3769. McPhie, P., and Gratzer, W . B. (1966). B i o c h b t r y 5, 1310. Maly, P.. Rinke,J., Ulmer, E., Zwieb, C., and Brimacombe, R. (1980).Biochemishy 19,4 179. Mende, L., Timm, B., and Subramanian, A. R. (1978).FEBS Lett. 96,395.

RIBOSOMAL PROTEINS

51

Michalski, C. J., and Sells, B. H. (1975). Eur. J. Biochem. 52, 385. Millon, R.,Olomucki, M., LeGall,J. Y., Golinska, B., Ebel,J. P., and Ehresmann, B. (1980). Eur. J. B i o c h . 110, 485. Mochalova, L. V., Shatsky, I. N., Bogdanov, A. A., and Vasiliev, V. D. (1982).J.Mol. Biol. 159,637. MBller, K., and Brimacombe, R. (1975). Mol. Gen. Genet. 141, 343. MBller, W., Castleman, H., and Terhorst, C. P. (1970). FEBS Lett. 8, 192. MBller, W., Groene, A., Terhorst, C., and Amons, R. (1972). Eur. J. Eiochem. 25, 5. Moore, P. B. (1980). In “Ribosomes: Structure, Function, and Genetics” (G. Chambliss, G. R. Craven, J. Davies, K.Davis, L. Kahan, and M. Nomura, eds.), p. 1 11. Univ. Park Press, Baltimore, Maryland. Moore, P. B., and Laughrea, M. (1979). Nucleic Acids Res. 6, 2355. Moore, P. B., Engelman, D. M., Langer, J. A., Ramakrishnan, V. R., Schindler, D. G., Schoenborn, B. P., Sillers, I. Y., and Yabuki, S. (1984). In “Basic Life Science Series” (B. P. Schoenborn, ed.). Plenum, New York, in press. Morgan, J., and Brimacombe, R. (1972). Eur. J. Biochcm. 29,542. Morgan, J., and Brimacombe, R. (1973). Eur. J. Biochem. 37,472. Morrison, C. A., Bradbury, E. M., and Garrett, R. A. (1977a). FEBS Lett. 81,435. Morrison, C. A., Garrett, R. A., and Bradbury, E. M. (1977b). Eur. J. Biochem. 78, 153. Morrison, C. A., Bradbury, E. M., Littlechild, J., and Dijk, J. (1977~).FEBS Lett. 83, 348. Newton, J., Rinke, J., and Brimacombe, R. (1975). FEBS Lett. 51, 215. Nierhaus, K. H. (1982). Curr. Top. Mbobiol. Immunol. 97,82. Nierhaus, K. H., and Dohme, F. (1974). Proc. Natl. Acad. Sci. U.S.A. 71,4713. Nierhaus, K.H., Lietzke, R., Nowotny, V., Schulze, H., Wurmbach, P., Stuhrmann, H. B., and May, R. (1983a). Physica 140B,426. Nierhaus, K. H., Lietzke, R., May, R. P., Nowotny, V., Schulze, H., Simpson, K., Wurmbach, P., and Stuhrmann, H. B. (1983b). Proc. Natl. Acad. Sci. U.S.A. 80,4213. Noah, M., Stbffler-Meilicke, M., and Stofller, G. (1983). Electron Microsc. Int. Cungr. loth, 1982 3, 101. Noller, H. F. (1980). In “Ribosomes: Structure, Function, and Genetics” (G. Chambliss, G. Craven, J. Davies, K.Davis, L. Kahan, and M. Nomura, eds.), p. 3. Univ. Park Press, Baltimore, Maryland. Noller, H. F., Kop, J., Wheaton, V.,Brosius, J., Gutell, R. R., Kobylov, A. M., Dohme, F., Stahl, D.A., Gupta, R., and Woese, C . R. (1981). Nuclcic Acidr Res. 9,6167. Nomura, M., and Erdmann, V. A. (1970). Nature (London) 228,744. Nomura, M., and Held, W. A. (1974). In “Ribosomes” (M. Nomura, A. Tissieres, and P. Lengyel, eds.), p. 193. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. OBrien, L., Shelley, K., Towfighi, J., and McPherson, A. (1980). Proc. Natl. Acad. Sci. U.S.A. 77, 2260. Odom, 0.W., Robbin, D., Kramer, G., Hardesty, B., and Subramanian, A. R. (1983). Arch. Biochnn. Biophys., in press. Olins, P. O., and Nomura, M. (1981). Cell 26,205. h e r b e r g , R., Sjbberg, B., Liljas, A,, and Pettersson, I. (1976a). FEBS Lett. 66,48. h e r b e r g , R., Sjbberg, B., and Garrett, R. A. (1976b). FEBS Lett. 65, 73. bterberg, R., Sjbberg, B., Garrett, R., and Littlechild, J. (1977a). FEBS Lett. 73, 25. Ckterberg, R.. SjBberg, B., Pettersson, I., Liljas, A,, and Kurland, C. G. (1977b). FEBS Lett. 73,22. h e r b e r g , R., Sjbberg, B., and Littlechild, J. (1978). FEBS Lett. 93, 115. Paradies, H. H.,and Franz, A. (1976). Eur.1. B i o c h . 67, 23.

52

L. GIRI ET AL.

Pettersson, l., Hardy, S.J. S., and Liljas, A. (1976). FEBS Lett. 64, 135. Poliu, S. M.. and Glitz, D. C. (1977). Prof. Natl. Acad. Sci. U.S.A. 74, 1468. Poliu, S. M., Noller, J. F., and McWhirter, P. D. (1981). Biochemistry 20, 372. Pon, C., Wittmann-Liebold, B., and Gualerzi, C. (1979). FEES Lett. 101, 157. Post, L. E., Strycharz, G. D., Nomura, M., Lewis, H., and Dennis, P. P. (1979). Proc. Natl. Acad. Sci. U.S.A. 76, 1697. Prakash, V., and Aune, K. C. (1978). Arch. Biochem. Biophys. 187,399. Privalov, P. L. (1974). FEBS Lett. 40 (suppl.), 140. Privalov, P. L. (1979). Adu. Protein Chem. 33, 167. Privalov, P. L. (1982). Adu. Protein Chem. 35, 1. Ramakrishnan, V. R., Yabuki, S., Sillers, LY., Schindler, D. G., Engelman, D. M., and Moore, P. B. (1981).J. Mol. Biol. 153, 739. Rinke, J., Meinke, M., Brimacombe, R., Fink, G., Rommel, W.,and Fasold, H. (1980).J. Mol. Biol. 137, 301. Robakis, N., and Boublik, M. (1981). Biochem. Biophys. Res. Commun. 103, 1401. Rohde, M. F., and Aune, K. C. (1975). Biochemistry 14, 4344. Rohde, M. F., O’Brien, S., Cooper, S.,and Aune, K. C. (1975). Biochemistry 14, 1079. Rtihl, R., and Nierhaus, K. H. (1982). Prod. Nntl. Acad. Sd. U.S.A. 79, 729. Rtihl, R., Roth, H. E., and Nierhaus, K. H. (1982). HopPC-Seyler’sZ. Physiol. C h . 363, 143. Roth, H. E., and Nierhaus, K. H. (1973). FEBS Lett. 31,35. Sarkar, P. K., Yang, J. T., and Doty, P. (1967). Biopolpers 5, 1. Schnier, J., Kimura, M.,Foulaki, K., Subramanian, A. R., lsono, K., and WittniannLiebold, B. (1982). Proc. Natl. Acad. Sci. U.S.A. 79, 1088. Serdyuk, I. N., Grenader, A. K., and Koteliansky, V. E. (1977). Eur. J. Biochem. 79, 504. Serdyuk, I. N., Zaccai, G., and Spirin, A. S. (1978). FEBS Lett. 94, 349. Serdyuk, I. N.,Grenader, A. K., and Zaccai, G. (1979).] Mol. Biol. 135, 691. Serdyuk, I. N., Gogia, Z. V., Venyaminov, S. Yu.,Khechinashvili, N. N., Bushuev, V. N., and Spirin, A. S. (1980). J . Mol. Biol. 137, 93. Shatsky, I. N., Mochalova, L. V., Kojouharova, M. S.,Bogdanov, A. A., and Vasiliev, V. D. (1979).J . Mol. Biol. 133, 501. Shatsky, I. N., Estafieva, A. G., Bystrova, T. F., Bogdanov, A. A., and Vasiliev, V. D. (1980a). FEBS Lett. 121,97. Shatsky, 1. N., Estafieva, A. G., Bystrova, T. F., Bogdanov, A. A., and Vasiliev, V. D. (1980b). FEHS Lett. 122, 25 I . Sieber, G., and Nierhaus, K. N. (1978). Biochemistry 17, 3505. Sieber, G., Tesche, B., and Nierhaus, K. H. (1980). Eur. J . Biochem. 106, 515. Slobin, L. I. (1972). Proc. Natl. Acad. Sd. U.S.A. 69, 3769. Smith, W. S. (1971). Ph.D. Thesis, University of Wisconsin, Madison, Wisconsin. Sommer, A., and Traut, R. R. (1976).J. Mol. Bwl. 106,995. Spiess, E. (1978). FEBS hit. 91, 289. Spirin, A. S., Serdyuk, I. N., Shpungin, J. L.,and Vasiliev, V. D. (1979). Proc. Natl. Acad. Sci. U.S.A. 76,4867. Spitnik-Elson, P., and Breiman, A. (1971). Biochim. Biophys. Acla 254,457. Spitnik-Elson, P., and Elson, D. (1979). In “Methods in Enzymology” (K. Moldave and L. Grossman, eds.), Vol. 59, p. 461. Academic Press, New York. Spitnik-Elsod, P., Schechter, N., Abramowitz, R., and Elson, D. (1976). Biochemistry 15, 5246. Spitnik-Elson, P., Elson, D., Avital, S., and Abramowitz, R. (1982a). Nucleic Acids Res. 10, 1995.

RIBOSOMAL PROTEINS

53

Spitnik-Elson, P., Elson, D., Avital, S., and Abramowitz, R. (1982b).Nucleic Acidr Res. 10, 4483. Steinhiuser, K. G., Woolley, P., Epe, B., Dijk, J. (1982).Eur. J. Eiochm. 127,587. Stiege, W., Zwieb, C., and Brimacombe, R. (1982).Nuclcic Acids Res. 10,721 1. Stiegler, P., Carbon, P., Zuker, M., Ebel, J. P., and Ehresmann, C. (1981).Nvcleic Acids Res. 9,2153. StBfRer, G., and StBffler-Meilicke, M. (1981).Int. CellBiol., Int. Con&. CellBiol., 2nd, p. 93. Sttimer, G., and Wittmann, H. G. (1971a).J.Mol. Eiol. 62,407. StBfRer, G., and Wittmann, H. G. (1971b).Proc. Natl. Acad. Sci. U.S.A. 68, 2283. StBfRer, G., Bald, R., Kastner, B., Luhrmann, R., StBffler-Meilicke, M., Tischendorf, G., and Tesche, B. (1980).In “Ribosomes: Structure, Function, and Genetics” (G. Chambliss, G. R. Craven, J. Davies, K. Davis, L. Kahan, and M. Nomura, eds.), p. 171.Univ. Park Press, Baltimore, Maryland. StOffler-Meilicke.M., and StBfRer, G. (1985).Electron MiCrosc. Int. C q p . loth, 1982 3,99. StBffler-Meilicke,M., Kastner, B., Noah, M., Tischendorf, G., and StBfHer, G. (1980).Eur. J . Cell Bwl. 33, 134. StllfRer-Meilicke, M., StBffler, G., Odom, 0. W., Zinn, A., Kramer, G., and Hardesty, B. (1981).Proc. N a L Acad. Sci. U.S.A. 78, 5538. Stuhrmann, H., Haas, J., Ibel, K., De Wolf, B., Koch, M. H. J., Parfait, R., and Crichton, R. R. (1976).Proc. Natl. Acad. Sci. U.S.A. 73, 2379. Stuhrmann, H. B., Koch, M. H. J., Parfait, R., Haas, J., Ibel, K., and Crichton, R. R. (1977). Proc. Natl. Acad. Sci. U.S.A. 74,2316. Stuhrmann, H. B., Koch, M.H. J., Parfait, R., Haas, J., Ibel, K., andcrichton, R. R. (1978). J . Mol. Biol. 119,203. Subramanian, A. R. (1983).Prog. N w k c AcidRes. Mol. Biol. 28, 101. Taddei, C. (1972).Ex$. CellRes. 70, 285. Tam, M. F., Dodd, J. A., and Hill, W. E. (1981a).J.Eiol. Chm. 256, 6430. Tam, M. F., Dodd, J. A., and Hill, W. E. (1981b).FEES Lett. 130, 217. Tardieu, A,, and Vachette, P. (1982).EMEOJ. 1, 35. Tesche, B., Schmiady, H., Lorenz, S., and Erdmann, V. A. (1980).In “Electron Microscopy” (P. Brederoo and W. de Priester, eds.), p. 534. Eur. Congr. Electron Microscopy, Leiden. Thammana, P., Cantor, C. R., Wollenzien, P., and Hearst, J. E. (1979).J.Mol. Eiol. 135, 271. Tindall, S. H., and Aune, K. C. (1981).Eiochemistty 20,4861. Tischendorf, G . W., Zeichhardt, H., and Stoffler, G. (1974).Mol. Gen. Genet. 134, 209. Tischendorf, G. W., Zeichhardt, H.. and StBf€ler, G. (1975).Proc. Natl. A d . Sci. U.S.A. 72, 4820. Tokimatsu, T., Strycharz, W. A., and Dahlberg, A. E. (1981).J.Mol. E d . 152, 397. Tolbert, W. R. (1971).Ph.D. Thesis, University of Wisconsin, Madison, Wisconsin. Traub. F., and Nomura, M. (1968).Proc. Natl. Acad. Sci. U.S.A. 59, 577. Traut, R. R., Lambert, J. M., Boileau, G., and Kenny, J. W.(1980).In “Ribosomes: Structure, Function, and Genetics” (G. Chambliss, G. R. Craven, J. Davies, K. Davis, L. Kahan, and M. Nomura, eds.), p. 89.Univ. Park Press, Balitmore, Maryland. Trempe, M. R., Ohgi, K., and Glitz, D. G. (1982).J.Biol. Chm. 257, 9822. Tumanova, L. G., Gudkov. A. T., Bushuev, V. N., and Okon, M. S. (1981).FEBS Lett. 127, 241. Turchinksy, M. J., Bronde, N. E., Kussova, K. S., Abduraschidova, G. G., Muchamedganova, E. V., Shatsky, J. N., Bystrova, T. F., and Budowsky (1978).Eur. J. Biochem. 90,83.

54

L. GIRI ET AL.

Turner, S., Thompson, J. F., Hearst, J. E., and Noller, H. F. (1982). Nzrcleic Acids Res. 10, 2839. Vasiliev, V. D. (1974). A c h Biol. Med. Ger. 33, 779. Vasiliev, V. D., and Zalite, 0. M. (1980). FEES Lett. 121, 101. Vasiliev, V. D., Salinonova. 0. M., and Koteliansky, V. E. (1978). FEES Left. 95, 273. Vasiliev, V. D., Selivanova, 0. M., Baranov, V. I., and Spirin, A. S. (1983). FEES Leff.155, 167. Venyaminov, S. Y., and Gogia, Z. G. (1982). Eur. J. Biochem. 126,299. Wabl, M. R. (1973). Ph.D. Thesis, Free University, Berlin. Winkelmann, D. A., Kahan, L., and Lake, J. A. (1982). Proc. Natl. Acad. Sci. U.S.A. 79, 3111. Wittmann, H. G. (1974). In “Ribosomes” (M. Nomura, A. Tissieres, and P. Lengyel, eds.), p. 93. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Wittmann, H. G. (1982). Annu. Rev. Biochcm. 51, 155. Wittmann, H. G. (1983). Annu. Rev. Biochcm. 52,35-65. Wittmann, H. G., Littlechild, J. A., and Wittmann-Liebold, B. (1980). In “Ribosomes: Structure, Function, and Genetics” (G. Chambliss, G. R. Craven, J. Davies, K. Davis, L. Kahan, and M. Nomura, eds.), p. 51. Univ. Park Press, Baltimore, Maryland. Wittmann, H. G., Mlissig, J., Piefke, J., Gewitz, H. S., Rheinberger, H. J., and Yonath, A. (1982). FEES Lett. 146,217. Wittmann-Liebold, B. (1973). HopPC-Scykr’s Z . Physiol. Chem. 254, 1415. Wittmann-Liebold, B. (1980a). In “Polypeptide Hormones” (R. F. Beers and E. G . Bassett, eds.), p. 87. Raven, New York. Wittmann-Liebold, B. (1980b). In “Genetics and Evolution of RNA Polymerase, tRNA, and Ribosomes” (S. Osawa, H. Ozeki, H. Uchida, and T. Yura, eds.), p. 639. Univ. of Tokyo Press, Tokyo. Wittmann-Liebold, B. (1981a). In “Chemical Synthesis and Sequencing of Peptides and Proteins” (T. Y. Liu, A. N. Schechter, R. L. Heinrikson, and P. G. Condliffe, eds.), p. 75. Elsevier, Amsterdam. Wittmann-Liebold, B. (1981b). In “High Performance Chromatography in Protein and Peptide Chemistry” (F. Lottspeich, A. Henschen and K. P. Hupe, eds.), p. 223. de Gruyter, Berlin. Wittmann-Liebold, B., Graffunder, H., and Kohls, H. (1976). Anal. Biochem. 75, 621. Wittmann-Liebold, B., Brauer, D., and Dognin, J. M. (1977a). In “Solid Phase Methods in Protein Sequence Analysis” (A. Previero and M. Previero-Coletti, eds.), p. 219. Elsevier, Amsterdam. Wittmann-Liebold, B., Robinson, S. M. L., and Dzionara, M. (1977b). FEES Lett. 81,204. Woese, C. R., Magrum, L. J., Gupta, R., Siegel, R. B., Stahl, D. A,, Kop, J., Crawford, N., Brosius, J., Guttel, R., Hogan, J. J., and Noller, H. F. (1980). Nucleic Acids Res. 8, 2275. Wollenzien, P. L., and Cantor, C. R. (1982a).J. Mol. Biol. 159, 151. Wollenzien, P. L., and Cantor, C. R. (1982b). Proc. Natl. Acad. Sci. U.S.A. 79, 3940. Wollenzien, P. L., Hearst, J. E., Thammana, P., and Cantor, C. R. (1979).J. Mol. Biol. 135, 255. Wong, K.-P., and Paradies, H. H. (1974). Biochcm. Biophys. Res. Commun. 61, 178. Wower, I., Wower, J., Meinke, M., and Brimacombe, R. (1981). Nucleic Acids Res. 9,4285. Wower, J., Maly, P., Zobawa, M.,and Brimacombe, R. (1983). Biochemistry 22,2339. Wystup, G., Teraoka, H., Schulze, H., Hampl, H., and Nierhaus, K. H. (1979). Eur. J . Biochcm. 100, 101.

RIBOSOMAL PROTEINS

55

Yonath, A., Miissig, J., Tesche, B., Lorenz, S., Erdmann, V. A., and Wittmann, H. G. (1980). B i o c h . fnt. 1,428. Yonath, A., Miissig, J., and Wittmann, H. G. (1982a).J. Cell. Bwchem. 19, 145. Yonath, A., Khavitch, G., Tesche, B., Miissig, J,, Lorenz, S., Erdmann, V. A., and Wittmann, H. G. (1982b). Biochem. fnt. 5,629. Yuki, A., and Brimacombe, R. (1975). Eur. J. B i o c h . 56, 23. Zantema, A., Maassen, J. A., Krick, J., Moller. W. (1982a). BiOchiShy 21, 3077. Zantema, A., Maassen, J. A., MBller, W. (1982b). Biochemkhy 21, 3069. Zimmermann, R. A. (1974). In “Ribosomes” (M. Nomura, P. Lengyel, and A. Tissitres, eds.), p. 225. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Zimmermann, R. A. (1980). In “Ribosomes: Structure, Function, and Genetics” (G. Chambliss, G.R. Craven, J. Davies, K. Davis, L. Kahan, and M. Nomura, eds.), p. 135. Univ. Park Press, Baltimore, Maryland. Zimmermann, R. A., Muto, A., and Mackie, G. A. (1974).J. Mol. Biol. 86, 433. Zimmermann, R. A., Mackie, G. A., Muto, A., Garrett, R. A., Ungewickell, E., Ehresmann, C., Stiegler. P., Ebel, J. P., and Fellner, P. (1975). Nwleic Acids Res. 4, 279. Zook, D. E.,and Fahnestock, S. R. (1978). Biochim. Biophys. Acta 517,400. Zwieb, C., and Brimacombe, R. (1980). Nuclcic Act& Res. 8, 2397.

56

L. GIRI ET AL.

APPENDIX:PRIMARY STRUCTURE OF Escherichia coli RIBOSOMAL PROTEINS

In this appendix the primary structures of all E. coli ribosomal proteins are listed. First the proteins from the small (30 S) ribosomal subunit and then those of the large (50 S) subunit are given. The amino acids are designated by their one-letter code as follows: A, alanine; C, cysteine; D, aspartic acid; E, glutamic acid; F, phenylalanine; G, glycine; H, histidine; I, isoleucine; K, lysine; L, leucine; M, methionine; N, asparagine; P, proline; Q, glutamine; R, arginine; S, serine; T, threonine; V, valine; W, tryptophan; and Y,tyrosine.

RIBOSOMAL PROTEINS PROTEIN

1 H 26 C 51 E 76 E 161 ~ 126 C 151

EC S 1

57

557 R E S I D U E S

25 E T R P C S I V R 50 V V V A I D K D V V L V D I G L K S E S k I P k

T E S F A Q L F E E S L K E I

-c

I J

Q F k N A Q C E L E I Q V C D E V D V A L D A V 100 D C F C E T L L S R E K A k R H E A W I T L E k 125 Y E D A E T V T C V I N C k V K C C f T V E L D 150 I R A F L P C S L V D V R P V R D T L H L E C K 175

E L E F K V I K L D Q K R N N V V V S R R I V I E 176 S 26 1 N 226 K 25 1 E 276 E 361 E 326 D 35 1 A 376

200 E N S I E R D Q L L E N L Q E C M E V K C I V k 225 L T D Y C A F V D L C C V D C L L H 1 T I M A b J 2 58 R V K H P S E I V N V C D E I T V K V L t ' F D R 275 R T R V S L C L K Q L C E D P U V A I A K R Y P 300 C T I L T C R V T N L T D Y C C F V E I E E C V 325 C L V H V S E t l D U T N K N I H P S L V V N ~ C 3 56 V V E Y tl V L D I D E E R R R I S L C L Ir. D C K 375 N F W Q Q F A E T H N K C D R V E G K I K S I T 4 00

D f C I F I G L D C C I D C L V H L S D l S U N V 401 A 426 R 451 C 476 C 581 V 526 D 55 1 F

4 25 C E E A V R E Y K K C D E I A ~ V V L Q V D A E 4 50 E R I S L C V K Q L A E D P F N N O V A L N K K 4 75 A I V T G K V T A V D A K C A T V E L A D C V E 580 Y L R A S E I S R D R V E D R T L V L S V C D E 525 E A K F T C V D R K N R b I S L S V R A K D E A 550 E K D A I A T V N K Q E D ~ ~ N F S N N A ~ A E A 557 K A A K C E

58

L. GIRl ET AL.

PROTE I N I A 26 H 51 E 76 S 181 T

EC S 2

2 4 8 RESIDUES

25 T V S R R D H L K A C V H F C H Q T R Y U N P K 58 K P F I F C A R N K V H I I H L E K T V P H F N 75 R L A E L N K I A S R K C K I L F V C T K R A A 188 E A V K D A A L S C D Q F F V N H R W L C C R L 125 N U K T V R Q S I K R L K D L E T Q S Q D C T F

126 i 58 D K L T K K E A L H R T R E L E K L E N S L C G I 151 175

K D H C C L P D A L F V I D A D H E H I A I K E A 176 200 N N L C I P V F A I V D T N S D P D C V D F V I P 20 1 225 C N D D A I R A V T L Y L C A V A A T V R E C R S 226 248 Q D L A S Q f i E E S F V E A E

PROTEIH

EC S3

232 RESIDUES

1 23 C Q K V H P N C I R L C I V K P U N S T U F A N T 26 50

K E F A D H L D S D F K V R Q Y L T K E L A K A S 51 75 V S P I V I E R P A K S I R V T I H T A R P G I V i6 108 I C # K C E D V E K L R K V V A D I A C V P A Q I l@l 125 N I A E V R K P E L D A K L V A D S I T S Q L E R 126 158

R V R F R R A H K R A V B N A H R L C A K C I K V 151 E 176 T 201 I 226 Q

175 V S C R L C C A E I A R T E U Y R E C R V P L H 200 L R A D I D Y N T S E A H T T Y C V I C V l V U 225 F K C E X L C C H 6 A V E O P E K P A A ~ p K ~ 2 32 Q R K C R K

59

RIBOSOMAL PROTEINS PROTEIN

283 R E S I D U E S

EC S 4

25

1

D L F L K S C V R

' f i R 'i' L C P K L K L S R R E C T

26

58

~ I D T K ~ K I E Q ~ P G Q H G ~ R K P R L S D Y 51 75 C V Q L R E K Q K V R R I Y C V L E R Q F R N Y Y

76

108

K E I A R L K C N f C E N L k L L Q C R L D N V V

101 Y

125 R

~

G

F

C

A

T

R

A

E

A

R

Q

L

V

S

H

K

~

126

I

~

i se

V

N

C

R V V N I ~ S Y Q V D P N S Y V I R E K A K K E S

175

151

R V K A A L E L A E Q R E K P T U L E V D ~ C K H

1 i6

200

E G T F K R K P E R S D L S A D I N E H L I V E L 26 1 203 Y S K

EC s 5

PROTEIN

1

a

H I E

c

K Q A

166 R E S I D U E S

E L

a

E K L I

v

p1

N R

v s

P6

K T

v

25 K

58

C C R I F S F T A L T V V C D G N C R V C F C Y C 51 75 K A R E V P A ~ I Q K ~ ~ E K ~ R R N M I N

V

A

L

108

76

N N C T L Q H P V K C V H T C S R V F H Q P A S E 125 C T C I I A G C A H R A V L E V A C V H N V L A K 126 i 58 ~ Y G S T N P I N ~ V R ~ T I D ~ L E N ~

101

151 ~

166 ~

~

A

K

R

G

K

S

~

E

E

I

L

G

K

N

S

P

60

L. GIRI ET AL.

PROTE?N

EC S 6

135 RESlDUES

1 25 H R H Y E I V F B V H P D Q S E Q V P G H I E R Y 96 50 T A A I T C A E C K I H R L E D Y C R R Q L A Y P 51 75

I N K L H K A H Y V L H N V E A P Q E V I D E L E 76 160 T T F R F N D A V I R S H V H R T K H A V T E A S 101 125 P H Y K A K D E R R E R R D D F A H E T A D D A E 126 135 A C D S E E E E E E

PROTEIN

EC S 7

1 7 7 RESIDUES

1 25 P R R R V I C Q R K I L P D P K F G S E L L A K F 26 50 V N I L M V D C K K S T A E S I V Y S A L E T L A 51 75 Q R S C K S E L E A F E V A L E N V R P T V E V K 76 160 S R R Y C C S T Y Q V P V E V P V R R H A L A H R 101 125 U I V E A A R K R C D K S H A L R L A N E L S D A 126 158 A E N K C T A V K K R E D V H R ~ A E I N K A F A 151 175 H Y R U L S L R S F S H Q A C A S S K Q P A L C Y 176 177 L N

RIBOSOMAL PROTEINS EC S S

PROTEIN

61

129 R E S I D U E S

1 25 S H Q D P I A D I I L T R I R N C Q A A N g A k V T 26 50 H P S S K L K Q A I R N V L E E E C F I E D F K V

51

75

E C D T K P E L E L T L K Y F Q C K A V V E S I Q 76 180 R V S R P C L R I Y K R K D E L P K V M A C L C I

181 A V V S 126 C Y V A

PROTEIN

125 T

S

K

C

V

~

T

D

R

A

~

R

Q

A

G

L

C

C

E

I

129

EC S 9

128 R E S I D U E S

1 25 A E N Q Y Y C T C R R K S S A ~ R V F I K P C N C 26 50 K I V I N Q R S L E Q Y F C R E T A R I I V V R Q P 51 75 L E L V N H V E K L D L Y I T Q K C G C I S C Q A

76

108

G R I R H C I T R A L l l E Y D E S L R S E L R K A

lei

125

C F V T R D A R Q V E R K K V C L R K A R R P E F 126 128 S K R

I

62

L. GIRI ET AL.

PROTEIN

EC $ 1 0

1 0 3 RESIDUES

1

25

H Q N B R I R I R L K A F D H R L I D Q A T A E I 26

58 V E T A K R T C A Q V R C P I P L P T R K E R F T 51 75

V L I S P H V N K D A R D Q Y E I R T H L R L V D 76 188 I V E P T E K T V D R L H R L D L A A C V D Y Q I 101 183 S L C

PROTEIN

1 A 26 F 51 F 76 Y 101 A 126 R

1 2 8 RESIDUES

25 K A P I R A R K R V R K Q V S D C V A H I H A S 58 N N T I V T I T D R Q C N ~ L C W A T A C C S C 75 R C S R K S T P F R A Q V A A E R C k D A V K E 108 C I K N L E V H V K C P C P C R E S T I R A L N i 25 A C F R I T N I T D V T P I P H H G C ~ P P K ~ 128 R Q

PROTEIN 1 A 26 C 51 V 76 H 101 L

EC S 1 1

EC S 1 2

1 2 3 RESIDUES

25 T Q N Q L V R K P R A R K Y A K S N V P A L E k 5% P Q K R C V C T R V Y T T T P K K P N S A L R K 75 C R V R L T N C F E V T S Y I C C E C H N L Q E 188 S V I L I R C C R V K D L P G V R Y H T V R C A 123 D C S C V K D R K Q A R S K Y C Y K R P K A

RIBOSOMAL PROTEINS

PROTEIN

EC S13

63

117 R E S I D U E S

25

1

A R I A C I N I P D H K H A V I A L T S I Y C V G 26 K 51 Q 76 I 101 T

58 T R S K A I L A A A C I A E D V K I S E L S E C

75 I D T L R D E V A K F V V E C D L R R E I S H S 188 K R L ~ D L C C Y R C L R H R R C L P V R C Q R 117 K T N A R T R K C P R K P I K K

PROTEIN

1 A 26 L 51 P 76 F

EC S 1 4

98 R E S I D U E S

25 K P S H K I R E V K R V A L A D K Y F A K R A E 58 K A I I S D V N A S D E D R U N A V L K L O T L 7s R D S S P S R O R N R C R Q T C R P H G F L R K 98 G L S R I K V R E A A H R C Q I P C L K K S

PROTEIN

E C Sl5

87 R E S I D U E S

1 25 S L S T E A T A K I V S E F C R D A N D T G S T E 26 58 V Q V A L L T A Q I N H L Q C H F A E K K D H H S

51 75 R R C L L R M V S Q R R K L L D V L K R K D V A R 76 87 Y T Q L I E R L C L R R

64

L. GIRI ET AL.

PROTEIN

EC S 1 6

82 R E S I D U E S

2s 1 ~ V T I R L A R H C A K K R P F Y Q V V V A D S R 26 50 N A R N C R F I E R V C F F N P I A S E K E E C T 91 75 R L D L D R I A H Y V C Q C A T I S D R V ~ ~ L 16 82 K E V N K A A

PROTEIN

EC S 1 7

83 R E S I D U E S

25 1 T D K I R T L R C R V V S D K H E K S I V V A I E 26 58

R F V K H P I Y C K F I K R T T K L H V H D E N N 51 7s E C I C D V V C E I R E C R P L S K T E S W T L V 76 83 R V V E K A V L

PROTEIN

1

EC S 1 8

74 RESIDUES

25

A R Y F R R R K F C R F T A Q C V Q E I D Y K D I 26

50

A T L K N Y I T E S C K I V P S R I T C T R A K Y 51 74 Q R Q L A R A I K R A R Y L S L L P Y T D R H O

I

RIBOSOMAL PROTEINS PROTEIN

EC S 1 9

65

9 1 RESIDUES

25 1 P R S L K K C P F I D L H L L K K V E K A V E S G 26 50

D K K P L R T U S R S T I F P D R H I C L T I A V 51 75 H N C R Q H V P V F V T D E ~ V C H K L G E F A P 76 91 T R T Y R C H A A N K K A K K K

PROTEIN

1 A 26 H 51 N 76 A

EC S 2 8

86 R E S I D U E S

25 N I K S A K K R A I Q S E K A R K H N A S R R S 50 H R T F I K K V Y A A I E ~ C D K A k A Q K L F 75 E H Q P I V D R Q A A K C L I H K N K A ~ R H K 86 N L T A Q I N K L A

PROTEIN

EC S 2 1

7 0 RESIDUES

1 25 P V I K V R E N E P F D V A L R R F K R S C E K A 26 50 C V L A E V R R R E F Y E K P T T E R K R A K h S 51 70 ~ V K R H A K K L ~ ~ R E N A R R T R L Y

66

L. GlRI ET AL.

PROTEIN

EC L l

233 R E S I D U E S

1

25

A K L T K R H R V I R E K V D A T K Q Y D I N E A 26 I 31 A 76 V 181 D 126 L 151 A 176 K 20 1 T 226 A

50 A L L K E L A T A K F V E S V D V A V N L C I D 73 R K S D Q N V R C A T V L P H C T C R S V R V A 100 F T Q C A N A E A A K A A C A E L V C H E D L A 125 Q 1 . K K C E I! N F D V V I A S P D A M R V V C Q 150 C Q V L C P R C L H P N P K V C T V T P N V A E 175 V K N A K A C Q V R Y R N D K N C I I H T T I C 200 V D F D A D K L K E N L E A L L V A L K K A K P 225 Q A K C V Y I K K V S I S T T H C A C V A V D Q 2 33 C L S A S V N

PROTEIN

EC L2

272 R E S I D U E S

1 25 A V V K C K P T S P C R R H V V K V V N P E L H K 26 50 C K P F A P L L E K N S K S C C R N N N C R I T T 51 75

R H I C C C H K Q A Y R I V D F K R N K D C I P A 76 180 V V E R L E Y D P N R S A N I A L V L Y K D C E R 101 125

R Y I L A P K C L K A C D Q I Q S C V D A A I K P 126 C 151 C 176 R 28 1 L

226

P 25 1 T

150 N T L P H R N I P V C S T V H N V E M K P C K C 175 Q L A R S l C T Y V Q I V A R D G A Y V T L R L 200 S C E H R K V E A D C R A T L C E V C M A E H H 225 R V L C K A C A A R U R C V R P T V R C T A H N 2 50 V D H P C C C H E C R N F C K H P V T P U C V Q 272 K C K K T R S N K R T D K F I V R R R S K

67

RIBOSOMAL PROTEINS

PROTEIN

EC L 3

209 R E S I D U E S

25

1

H I C L V C K K V C H T R I F T E D C V S I P V T 26 58 V I E V E A N R V T Q V K D L A N D C Y R A I Q V

51

75

T T C A K K A N R V T K P E ~ C H F R K A G V E A

76

188

C R C L U E F R L R E C E E F T V C Q S I S V E L

lei

125

F A D V K K V D V T C T S K C K C F A C T ~ K R U

126

158

N F R T Q D A T H C N S L S H R V P C S I C Q N Q

151

175

T P C K V F K C K K H R C Q H C N E R V T V Q S L 280 176 D V V R V D A E R N L L L V K C A V P C R T C S D 28 1 289 L I V K P ~ ~ V K A

PROTEIN

EC L 4

281 R E S I D U E S

1

25

H E L V L K D R Q S A L T V S E T T F C R D F N E 26 58 ~ L V H Q V V V A Y A A G A R Q G T R A ~ K T

51

75

E V T C S C K K P Y R Q K C T C R ~ R S C S I K S 16 i aa P I Y R S C C V T F A A R P Q D H S Q K V N K K H

lei 125 Y R C R L K S I L S E L V R Q D R L I U V E K F S 126 V E A P K T K L L A Q K L K D H A L E D V L I

151

158

I

T

175

G E L D E N L F L A A R N L H K V D V R D R T C I 288 D P V S L I A F D K V V H T A D A V K B V E E H L

176

28 1 1

281

R

~

~

68

L. CIRI ET AL.

PROTEIN

EC L S

1 7 8 RESIDUES

1

25

A K L H D Y Y K D E V V K K L H T E F N Y N S V M 26

50

Q V P R V E K I T L N H C V C E A I A D K K L L D 51 75 N A ~ A D L A A I S C Q K P L I T K A R K S V A C ?6 i Be

F K I R Q C Y P I C C K V T L R C E R M W E F F E 101 125 R L I T I A V P R I R D F R C L S R K S F D C R C 126 158

N Y S t l C V R E Q I I F P E I D Y D K V D R V R G 151 175 L D I T I T T T A K S D E E C R R L L A A F D F P .1?6 178

F R K

PROTEIN 1

EC L 6

1 7 6 RESIDUES 25

S R V A K A P V V V P A C V D V K I N C Q V I T I 26 K 51 F 76 I 101 V 126 T 151 R 176 K

50 C K N C E L T R T L N D A V E V K H B D N T L T 7s C P R D C Y A D C U A Q A C T ~ R A L L N S N V 108 C V T E D F T K K L Q L V C V C Y R A A V K C N 125 I N L S L C F S H P V D H Q L P A C I T A E C P 150 Q T E I V L K C A D K Q V I C Q V A A D L R A Y 1i s R P E P Y K C K C V R Y A D E V V R T K E R K K 176

RIBOSOMAL PROTEINS

PROTEIN 1 H 26 L 51 R 76 E

101 D 126 H

148 RESIDUES

25 P V I L L D K V A N L C S L C B P V N V K A C Y 50 R N F L V P P C K L V P A T K K N I E F F E L R 75 A E L E A K L A E V L L L L N L R L E K I N L L 188 T V T I A S K L C D E C K L F C S I C T R D I L 125 A V T I A D C V E V L K S E V R L P N G V L R T 148 C E E V S F Q V H E V F L K V I V N V V A E

PROTEI N 1 A 26 V 51 H ?6 V 101

EC L 9

69

EC L l 0

165 R E S I D U E S

25 L N L O D K P A I V A E V S E V A K C A L S L V 58 A D S R C V T V D K H T E L R K L C R E A C V Y 75 R V V R N T L L R R A V E C T P F E C L K D L F 100 C P T L I A Y R S H E H P C A A L R L F K E F L 125

K A N L K F E V K A A A F E C E L I P A S P I D R

126 1 se L A T L P T Y E E A I A R L H I T H K E L S L G K 151 165 L V R T L A A V R D L K E A A

70

L. GIRl ET AL.

PROTEIN

EC L 1 1

141 RESIDUES

1 25 A K K V O A Y V K L O V A A C H A N P S P P V C P 26 58

A L C Q O C V N I H E F C K A F N I K T D S I E K 51

75

C L P I P V V I T V Y A D R S F T F V T K T P P A

i ee

76

A V L L K K A A C I K S C S C K P N K D K V C K I 101 125 S R ~ Q L Q E I A Q T K A A D H T C A D I E A H T 126 141 R S I E C T A R S H C L V V E D

PROTEIN

EC L 1 2

128 RESIDUES

1

25

S I T K D Q I I E A V A A H S V M D V V E L I S A 26 M 51 K 76 T 101 D

58 E E K F C V S A A A A V A V A I C P Y E A ~ E E 75 T E F D V I L K A A G A N K V A V I K A V R G a 100 C L C L K E A K D L V E S A P A A L K E G V S K 128 D A E A L K K A L E E A C A E V E V K

PROTEIN 1 H 26 G 51 C 76 H

EC L 1 3

142 RESIDUES

25 K T F T A K P E T V K R D U Y V V D L T G K T L 58 R L A T E L A R R L R C K H K A E Y T P H V D T 75 D Y I I V L N I D K V A V T C N K R T D K V Y Y 188 H T C H I C C I K B A T F E E H I A R R P E R V lei 125 I E I A V K C H L P K C P L C R A H F R K L K V Y 126 142 A C N E H N H A A Q Q P O V L D I

RIBOSOMAL PROTEINS

PROTEIN

EC L 1 4

71

123 R E S I D U E S

1 25 H I Q E Q T n L N V A D N S C A R R V M C I K V L 26 50

C C S H R R Y A C V C D I I K I T I K E A I P R C 51 75 K V K K C D Q L K A V V V R T K K C V R R P D C S 76 188 Q I R F D C N A C V L L N N N S E Q P I C T R I F 181 123 C P V T R E L R S E K F H K I I S L A P E V L

PROTEIN 1 H R L 26 C L C 51 E C C 76 E I R 101

EC L l 5

1 4 4 RESIDUES

25 N T L S P A E C S K K A C K R L C R C I C S 58 K T C C R C H K C Q K S R S C C C V R R C F 75 Q H P L Y R R L P K F C F T S R K A A I T A 100 L S D L A K V E C C V V D L N T L K A A N I 125 I C I Q I E F A K V I L A C E V T T P V T V R C L 144 126 R V T K C A R A A I E A A C C K I E E

PROTEIN EC L l 6 136 R E S I D U E S 1 25 H L Q P K R T K F R K M H K C R N R C L A Q C T D 26 58 V S F C S F C L K A V C R C R L T A R Q I E A A R

51

7s

R A H T R A Q K R Q C K I Y I R V F P D K P I T E 76 K P L A Y R N C K C K C N V E Y U V A L I Q P 101 V L Y E H D C V P E E L A R E A F K L A A A K 126 I K T T F V T K T V H

180 C K 125 L P

136

72

L. GlRI ET AL.

PROTEIN

EC L 1 7

127 R E S I D U E S

1 25 H R H R K S C R O L N R N S S H R O A H F R N H A ?6 50

C S L V R H E I I K T T L P K R K E L R R V V E P

51 L 76 V 101 C 126 A

75 I T L A K T D S V A N R R L A F A R T R D N E I 188 A K L F N E L C P R F A S R A C C Y T R I L K C 125 F R A C D N A P R A Y I E L V D R S E K A E A A 127 E

PROTEIN

EC L 1 8

117 R E S I D U E S

1 25 H D K K S A R I R R A T R A R R K L O E L C A T R 26 58 L V V H R T P R H I Y A Q V I f i P N C S E V L V A 51 75 A S T V E K A I A E O L K Y T C N K D A A A ~ V C 76 188

K A V A E R A L E K C I K D V S F D R S G F O Y H 161 C R V Q A L A D A A R E A C L O F

PROTEIN 1

EC L 1 9

117

114 R E S I D U E S 25

8 N I I K O L E O E O H K O D V P S F R P C D T V 26 E 51 N 76 W 101 E

58 V K V Y V V E C S K K R L O A f E C V U I ~ l ~ 75 R C L H S A F T V R K I S N C E C V E R V F O T 188 S P V V D S I S V K R R C A V R K A K L Y Y L R 114 R T C K A A R I K E R L N

RIBOSOMAL PROTEINS

PROTEI N

EC L 2 0

73

117 R E S I D U E S

1 25 ~ R V K R G V I A R ~ ~ R H K K I L K Q A K G Y Y G 26 58 l R S R V Y R V A F Q A V I K A C Q Y A Y R D R R

51

75

~ R K R ~ F R Q L U I A R I N A A A R Q N G I S Y 16 198

S K F I N C L K K A S V E I D R K I L A D I A V F

1 l?

181 D K V A F T A L V E K A K A A L A

PROTEIN

EC L 2 1

193 R E S I D U E S

1 25 H Y A V F Q S C C K Q H R V S E C Q T V R L E K L 26 58 D I A T C E T V E F A E V L ~ I A N C E E V K I C 51 75

V P F V D C C V I K A E V V A H C R C E K V K I V 76

199

K F R R R K H Y R K Q Q C H R Q Y F T D V K I T G 101

183

I S A

PROTEIN

EC L 2 2

1

110 R E S I D U E S

25

H E T I A K H R H A R S S A Q K V R L V A D L I R 26 58 C K K V S Q A L D I L T Y T N K K A A V L V K K V

51

75

L E S L I A N A E H N D C A D I D D L K V T K I F 76 108 V D E C P S H K R I ~ P R A K C R A D R I L K R T

181

S H I T V V V S D R

119

74

L. GIRI ET AL.

PROTEIN

1 M 26 K 51 F 76 R

EC L 2 3

99 R E S I D U E S

25 I R E E R L L K V L R A P H V S E K A S T A R E 58 S N T I V L K V A K D A T K A E I K A A V Q K L 75 E V E V E V V N T L V V K C K V K R H C Q R I C 99 R S D K K A Y V T L K E C Q N L D F V C C A E

PROTEIN

EC L 2 4

183 RESIDUES

1 25 A f l K I R R D D E V I V L T C K D K C K R C K V K 26 58

N V L S S C K V I V E C I H L V K K H P K P V P A 51 75 L N B P C C I V E K E A A I P V S N V A I F N A A 76 1 ee T C K A D R V C F R F E D C K K V R F F K S N S E

lei

i 83

T I K

PROTEIN

1 R 26 F 51 Q ?6 D

EC L 2 5

94 RESIDUES

25 F T I N A E V R K E Q C K C A S R R L R ~ ~ M K 58 P A I I Y C C K E I P L A I E L D H D K V ~ N H 75 A K A E F Y S E V L T I V V D C K E I K V K ~ Q 94 V Q R H P Y K P K L Q H I D F V R A

RlBOSOMAL PROTEINS

PROTEIN 1 A 26 C 51 C 76 R

EC L 2 7

75

84 R E S I D U E S

25 H K K f i C C S T R N C R D S E A K R L C V K R f 58 C E S V L A C S I I V R O R C T K F H A G f i N V 75 C C R D H T L F A K A D C K V K F E V K C P K N 84 K F I S I E A E

PROTEIN

EC L 2 8

77 RESIDUES

1 25 S R V C Q V T C K R P V T C N N R S H A L N A T K 21 50 R R F L P N L H S H R F U V E S E K R F V T L R V 51 7s

S A K C H R V I D K K C I D T V L A E L R ~ R G E 76 K Y

PROTEIN

77

EC L 2 9

63 RESIDUES

1 25 ~ K ~ ~ K E L R E K S V E E L N T E L L N L L R E Q 26 58 F N L R H Q A A S C Q L Q Q S H L L K Q V R R D V 51 63 A R V K T L L N E K A C A

PROTEIN

EC L 3 8

58 RESIDUES

1 25 A K T I K I T Q T R S f i I C R L P K H K A T L L C 26 58

L C L R R I C H T V E R E D I P A I R C H I N A V 51 S F H V K V E E

58

76

L. GIRI ET AL.

PROTEIN

EC L 3 1

62 R E S I D U E S

25

1

H K K D I H P K Y E E I T A S C S C C N V ~ K I R 26 58 S T V C H D L N L D Y C S K C H P F F T C K Q R D 51 62 V A T C C R V D R F N K

PROTEIN

EC L32

56 R E S I D U E S

1 25 A V Q Q N K P T R S K R C ~ R R S H D A L ~ A V ~ 26 58 S L S Y D K T S C ~ K H L R H H I T A D G Y Y R C 51 56 R K V I A K

PROTEIM

EC L 3 3

54 R E S I D U E S

1 2s A K C I R E K I K L V S S A C T C H F Y T T T K N 26 5e K R T K P E K L E L K K F D P V V R Q H V Y I K E 51 54 A K I K

PROTEIN

EC L34

46 R E S I D U E S

1 25 H K R T F O P S V L K R N R S H C F R A R ~ A T K 26 46 N C R Q V L A R R R f i K G R A R L T V S K

RIBOSOMAL PROTEINS

77

APPENDIXREFERENCES s1

s2 s3 s4 s5 S6 s7 S8 s9 s10 s11 s12 S13 S14 S15 S16 S17 S18 s19 s20 s 21 L1 L2 L3 L4 L5 L6 L7 L9 L10

Schnier, J.. Kimura, M., Foulaki, K., Subramanian, A. R.,Isono, K., and Wittmann-Liebold, B. (1982). Proc. Natl. Acad. Sci. U.S.A.79, 1008-101 1. Kimura, M., Foulaki, K., Subramanian, A. R., and Wittrnann-Liebold, B. (1982). Eur. J. Ewchem. 123,37-53. Wittmann-Liebold, B., and Bosserhoff, A. (1981).FEES Lett. 129, 10-16. Brauer, D., and Roeming, R. (1979).FEES Lett. 106, 352-357. Reinbolt, J., and Schiltz, E. (1973). FEES Lett. 36,250-252. Schiltz, E., and Reinbolt, J. (1975). Eur. J. Biochem. 56,467-481. Wittmann-Liebold, B., and Greuer, B. (1978). FEES Lett. 95, 91-98. Hitz, H., Schaefer, D., and Wittmann-Liebold, B. (1975).FEES Lett. 56,259-262; (1977). Eur. J. Ewchem. 75,497-512. Reinbolt, J., Tritsch, D., and Wittmann-Liebold, B. (1978). FEES Lett. 91, 297301; (1979). Eiochimie 61,501-522. Allen, G., and Wittmann-Liebold,B. (1978). Hoppe-Seyler’s 2.Physiol. Chem. 359, 1509-1525. Chen, R., and Wittmann-Liebold,B. (1975). FEES Lett. 52, 139-140. Chen, R. (1977).Hoppe-Seyler’s Z . Physwl. C h . 358, 1415-1430. Yaguchi, M., Roy, C., and Wittmann, H. G. (1980). FEESLett. 121, 113-116. Kamp, R., and Wittmann-Liebold, B. (1980). FEES Lett. 121, 117-122. Funatsu, G., Yaguchi, M., and Wittmann-Liebold, B. (1977).FEESLett. 73,12-17. Lindemann, H., and Wittmann-Liebold, B. (1976). FEES Lett. 71, 251-255; (1977). Hoppc-Seylc*’s Z . Physiol. C h m . 358,843-863. Yaguchi, M., Roy, C., Reithrneier, R. A. F., Wittmann-Liebold,B., and Wittrnann, H. G. (1983). FEES Lctt. 154921-30. Morinaga, T., Funatsu, G., Funatsu, M., and Wittmann, H. G. (1976). FEES Lett. 64,307-309. Vandekerckhove,J., Rombauts, W., and Wittrnann-Liebold,B. (1977).FEES Lett. 73, 18-21; (1977). Hoppe-Seyler’s 2.Physiol. Chem. 358, 989-1002. Yaguchi, M., and Wittmann, H. G. (1978). FEES Lett. 87, 37-40. Yaguchi, M. (1975). FEES Lett. 59, 217-220. Yaguchi, M., and Wittmann, H. G. (1978). FEBS Lett. 88,227-230. Wittrnann-Liebold, B., Marzinzig, E., and Lehmann, A. (1976). FEES Left. 68, 110-1 14. Vandekerckhove,J., Rombauts, W., Peeters, P., and Wittmann-Liebold,B. (1975). Hoppc-Seyler’s 2. Physiol. Chem. 356,1955-1976. Brauer, D., and Oechsner, 1. (1978). FEES Lett. 96, 317-321. Kimura, M., Mende, L., and Wittmann-Liebold, B. (1982). FEES Lett. 149, 304312. Muranova, T. A., Muranova, A. V., Markova, L. F., and Ovchinnikov, Y. (1978). FEBS Lett. 96, 301-305. Kimura, M., and Wittrnann-Liebold, B. (1980). FEES Lett. 121, 317-322. Chen, R., and Ehrke, G. (1976). FEES Lett. 69,240-245. Chen, R., Arfsten, U., and Chen-Schmeisser, U. (1977).Hoppe-Seyler’s Z . Physiol. Chem. 358,531-535. Terhorst, C., Mbller, W., Laursen, R. A., and Wittmann-Liebold, B. (1972). FEES Lett. 98, 325-328; (1973).Eur. J. Biochem. 34, 138-152. Kamp, R. M., and Wittrnann-Liebold, B. (1982).FEES Lett. 149, 313-319. Heiland, I., Brauer, D., and Wittmann-Liebold, B. (1976).Hoppe-Seyler’s Z . Physiol. Chem. 357, 1751-1770.

78

L11 L12 L13 L14 L15 L16 L17 L18 L19 L20 L2 1 L22 L23 L24 L25

L27 L28 L29 LSO L3 1 L32 L33 L34

L. GlRI E T AL. Dovgas, N. V., Vinokurov, L. M., Velmoga, I. S., Alakhov, Y. B., and Ovchinnikov, Y.A. (1976).FEBS Lett. 67, 58-61. Dognin, M. J., and Wittmann-Liebold, B. (1977). FEES Lett. 84, 342-346; (1980). EUT.J . Bwchem. 112, 131-151; (1980). Ho@~-S&'S Z . P h y i d . C h m . 361, 1697- 1705. Terhorst, C., M&ller,W., Laursen, R. A., Wittmann-Liebold, B. (1972).FEBS Lett. 28, 325-328; (1973). E m J. Biochem. 34, 138-152. Mende, L. (1978). FEBS Lett. 96, 313-316. Kimura, M., and Wittmann-Liebold, B. (1982). Bwchem. Int. 4,567-574. Giorginis, S., and Chen, R. (1977). FEBS Lett. 84, 347-350. Brosius, J., and Chen, R. (1976). FEBS Lett. 68, 105-109. Rombauts, W., Feytons, V., and Wittmann-Liebold, B. (1982). FEBS Lett. 149, 320-327. Brosius, J., Schiltz, E., and Chen, R. (1975).FEBS Lett. 56, 359-361. Brosius, J., and Arfsten, U. (1978). Biochemkty 17,508-516. Wittmann-Liebold,B., and Seib, C. (1979). FEBS Lett. 103, 61-65. Heiland, I., and Wittmann-Liebold, B. (1979).Biochemisty 18,4605-4612. Wittmann-Liebold,B., and Greuer, B. (1980). FEES Lett. 121, 105-1 12. Wittmann-Liebold, B., and Greuer, B. (1979). FEES Lett. 108,69-74. Wittmann-Liebold,B. (1979). FEBS Lett. 108,75-80. Dovgas, N. V., Markova, L. F., Mednikova, T. A., Vinokurov, L. M., Alakhov, Y.B., and Ovchinnikov, Y. A. (1975). FEBS Lett. 53, 351-354. Bitar, K.G., and Wittmann-Liebold,B. (1975).Hoppe-Seyler's Z. Physiol. Chem. 356, 1343-1352. Chen, R., Mende, L., and Arfsten, U. (1975). FEBS Lett. 59,96-99. Wittmann-Liebold,B., and Marzinzig, E. (1977). FEES Lett. 81, 214-217. Wittmann-Liebold,B., and Kamp, R. (1980). Biochem. Int. 5,436-445. Ritter, E., and Wittmann-Liebold, B.(1975). FEBS Lett. 60, 153-155. Brosius, J. (1978). Biochemistty 17,501-508. Wittmann-Liebold,B., Greuer, B., and Pannenbecker, R. (1975). Hoppe-Sqrler's 2. PhySw1. C h . 356, 1977-1979. Wittmann-Liebold,B., and Pannenbecker, R. (1976). FEBS Lett. 68, 115-1 18. Chen, R., and Ehrke, G. (1976). FEBS Lett. 63,215-217. Chen, R. (1976). Hoppe-Sqrler's 2.Physiol. Chem. 357, 873-886.

SEMLlKl FOREST VIRUS: A PROBE FOR MEMBRANE TRAFFIC IN THE ANIMAL CELL By KAI SIMONS and GRAHAM WARREN European Molecular Biology Laboratory, Heid.lb.rg. Federal Republk of Germany

. . . . . . . . . . . . . . . . . . . . . . . . . . . A. The Nucleocapsid. . . . . . . . . . . B. The Viral Envelope . . . . . . . . . . 111. The Life Cycle of Semliki Forest Virus. . . . . . A. Infection . . . . . . . . . . . . . B. Synthesis . . . . . . . . . . . . . C. Intracellular Transport of the Viral Glycoproteins. . D. Budding . . . . . . . . . . . . . IV. Perspectives . . . . . . . . . . . . . References. . . . . . . . . . . . . . I. Introduction

11. Structure

. . . . .

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

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

79 81 82 86 98 98 104 111 120 I24 125

I. INTRODUCTION During the past 25 years a considerable body of data has been accumulated, often to atomic resolution, on the structure and function of proteins. In contrast we know far less about the life cycle of these proteins-those processes which put a protein in the part of the cell in which it is to function and the cellular movements (if any) of this protein as it carries out its function. We know even less about those processes which eventually single out the protein for degradation. The first evidence that the routing of proteins to their correct destination in the cell is encoded in the primary structure of the protein came from work on secretory proteins (Milstein et al., 1972). Mainly through the work of Blobel and associates it was found that the amino-terminal extension of secretory proteins, termed the signal peptide, directs the ribosomal complex to the endoplasmic reticulum (ER),and, as a result, the polypeptide chain is transferred through the ER membrane during synthesis and segregated into the lumen (Blobel and Dobberstein, 1975a,b). Not only secretory proteins but also a number of other proteins are synthesized with amino-terminal signal peptides (see Warren, 1981). Those proteins which have their domiciles in such diverse organelles as the cell surface membrane, lysosomes, and secretory granules are synthesized on membrane-bound ribasomes and are initially found in the same compartment, the ER.From here the proteins are transported 79 ADVANCES IN PROTEIN CHEMISTRY, Vol. 34

Copyright 0 1984 by Academic Press, Inc. All rights of reproduction in any form reserved.

ISBN 0-12-034236-7

80

KAI SIMONS AND GRAHAM WARREN

to the Gold complex, the organelle which routes the proteins to their final destinations (see Tartakoff, 1980). The traffic among the cellular compartments is thought to be mediated by membrane vesicles which bud from one compartment and fuse with the next (Palade, 1975). Despite the continuous exchange of membrane components among them, the organelles maintain their characteristic protein and lipid compositions so that the traffic remains selective, thus avoiding intermixing of components. Membrane must also be recycled backward to compensate for loss of membrane in the forward movement. Membrane recycling at the cell surface has been especially studied and it is now thought that the cell membrane of all animal cells is being continuously and rapidly endocytosed (Silverstein et al., 1977). This membrane traffic recycles components from the cell surface to the interior of the cell and back to the cell surface again (Anderson and Kaplan, 1983). Some of the surface membrane components are channeled at least in part to the lysosome to be degraded. In some cells with considerable secretory activity, recycling of surface components back to the Golgi complex has also been detected (Farquhar and Palade, 1981). The membrane traffic between the ER and the cell surface involves a major sorting problem (Rothman, 1981). Little is known of how the animal cell has solved this problem in molecular terms. Such processes are exceedingly difficult to study in the cell in which a multitude of proteins is synthesized simultaneously with a sizable proportion of them initially routed into the ER. In vitro systems have been developed to study the first phase of assembly into the ER during protein synthesis (Blobel and Dobberstein, 1975a,b). Attempts to reconstruct other phases in the transport of proteins from the ER to other organelles have begun (Fries and Rothman, 1980) but are still in their infancy. Other simplifications are obviously needed to make possible studies of these processes at the molecular level. One experimental tool in this direction is provided by some enveloped animal viruses which mature at the cell surface of infected cells (KZiriainen and Renkonen, 1977; Lenard, 1978). Such viruses include influenza virus, Semliki Forest virus (SFV), Sindbis virus, and vesicular stomatitis virus (VSV). They are extremely simple in makeup and hence are very well characterized. They can be tagged with biochemical probes in many different ways. They infect many animal cells in culture, and after infection turn the cells into factories for the production of virus progeny. The protein-synthesizing machinery of the host cell is programmed by the viral RNA to make viral proteins exclusively and these include the viral surface glycoproteins.These are synthesized with signal peptides and inserted into the ER membrane (Katz et al., 1977; Garoff et

SEMLIKI FOREST VIRUS

81

al., 1978; Bonatti f t al., 1979), from which they are transported to the cell surface via the Golgi complex (Bergmann et al., 1981; Green et al., 1981). The net effect is the same as if the cell were to divert most of its protein-synthesizing capacity to the making of only one or two of its own plasma membrane glycoproteins. This amplification is the key to the use of viral glycoproteins as probes for membrane traffic from the ER. The endocytic route from the cell surface can also be studied with enveloped viruses, because this is the route they use to infect the cell (Helenius et al., 1980). The purpose of this article is to illustrate the use of enveloped viruses as tools in the study of membrane traffic in the animal cell. We will do this in the context of the life cycle of the virus in the host cell. The article will be concerned mainly with Semliki Forest virus (SFV) which is the virus we have worked with. SFV belongs to the alphaviruses, a genus of the togavirus family. Another well studied and closely related alphavirus is Sindbis virus. For more information on the biology of these viruses see the monograph on togaviruses (Schlesinger, 1980). There are also a number of reviews on the structure and assembly of alphaviruses which overlap but also cover aspects not treated here (Strauss and Strauss, 1977; Kaariainen and Soderlund, 1978; Garoff et al., 1982a).

11. STRUCTURE The alphavirus particle consists of RNA, protein, and lipid. The viral 42 S RNA molecule is single stranded and has a molecular weight of 4.14.3 x lo6 (see Kaariainen and Soderlund, 1978; Kennedy, 1980). Together with the capsid protein which has a molecular weight of 29,700 (Garoff et al., 1980a; Rice and Strauss, 1981) it forms the nucleocapsid which is encapsulated by the viral envelope, a lipid bilayer studded with spikes of viral glycoproteins. In SFV each viral spike glycoprotein is formed from three polypeptide chains (molecular weights in parentheses): El (50 X lo3), E2 (50 X lo3), and E3 (10 X lo3), each of which contains covalently bound carbohydrates (Garoff et al., 1974; Ziemiecki and Garoff, 1978). In Sindbis virus the viral spike is a two-chain structure containing only El and E2 (Schlesinger and Schlesinger, 1972; Rice and Strauss, 1982). The weight of the SFV particle was, until recently, thought to be about 60 x lo6 daltons (Laine et al., 1973; Kaiiriainen and Soderlund, 1978). This estimate was based on the chemical composition of the virus and on the molecular weight of the viral RNA. This value has now been shown to be significantly wrong by three independent methods.

82

KAI SIMONS AND GRAHAM WARREN

1. The sedimentation coefficient ( s ~ ~measured , ~ ) in the analytical ultracentrifuge is 274 & 2 S.The diffusion coefficient from light scattering measurements ( D Z ~ is, ~6.35 ) 2 0.1 X lo-* cm2 sec-l. These values, together with the partial specific volume of the virus (0.75ml g-l) give a weight for the viral particle of 42 X lo6 daltons (Jacrot et al., 1983). 2. From neutron scattering measurements using different concentrations of D20, an independent molecular weight estimate of 40.8 x lo6 has been derived (Jacrot et al., 1983). 3. Mass determination of unstained virus specimens in the scanning transmission electron microscope gives a value of 35 2 7 x lo6 daltons (Freeman and Leonard, 1981). These new estimates of the particle weight change the number of copies of the viral glycoprotein per viral particle. The polypeptides (capsid, E l , E2, and E3) are present in equimolar amounts (Garoff et al., 1974).Since 56.6% of the virus is protein (leaving out the carbohydrate content) the viral particle (using a molecular weight of 41-42 x lo6) should contain about 180 copies of each protein. Electron microscopic studies have suggested that the alphavirus particle has icosahedral symmetry (see below). The triangulation number is not certain, however (Murphy, 1980).Previous estimates for the molecular weight were compatible with 240 subunits per virus particle, and electron micrographs appear to show a T = 4 surface lattice (von Bonsdorff and Harrison, 1975). More information is now needed to determine the surface organization, since compositional data show fewer than 240 subunits. A. The Nucleocapsid The nucleocapsid can be isolated from purified viral particles using mild detergents to solubilize the viral envelope (see Kaariainen and Siiderlund, 1978). It can also be purified from extracts of infected cells. The weight of the SFV nucleocapsid, recently determined by neutron scattering analysis, is 9.46 X lo6 daltons of which the RNA accounts for 4.1 x lo6 daltons (Jacrot et al., 1983).This leaves 5.36 x lo6 daltons for the capsid proteins, which is exactly the value calculated for 180 protein subunits each having a molecular weight of 29.7 X 10’. These values are in good agreement with the RNA percentage (42%) determined by chemical analysis (Jacrot e l al., 1983).The diameter of the nucleocapsid in the virus particle has been determined by low-angle X-ray scattering (Harrison et al., 1971;S. C. Harrison and L. Kaariainen, unpublished

SEMLIKI FOREST VIRUS

83

data). The diameters of the Sindbis virus and SFV nucleocapsid are 400 and 380 8,, respectively. Recent neutron scattering values give a slightly larger value of 410 8, for the diameter of the isolated SFV nucleocapsid (Jacrot et al., 1983). The surface structure of the nucleocapsid appears to be icosahedral. Although the exact organization of the capsid proteins in the nucleocapsid is not yet clear (see Murphy, 1980),the composition is compatible with T = 3 icosahedral symmetry. The difficulty in determining the surface lattice of the capsid protein is probably due to the organization of the protein within the nucleocapsid. Neutron diffraction studies suggest that the proteins do not form a shell around the RNA (Jacrot et al., 1983). Instead, the proteins and the RNA appear to be rather uniformly distributed within the particle and this might make the protein subunits rather difficult to visualize in electron micrographs. The capsid protein is 267 amino acids in length in SFV, and 264 in Sindbis virus (Garoff et al., 1980a; Rice and Strauss, 1981; Boege et al., 1981). Their amino acid sequences have been established both by conventional methods and by DNA sequencing (Fig. 1). The capsid proteins from both viruses contain a striking cluster of lysine, arginine, and proline residues in the amino-terminal third of the polypeptide chain. A number of nucleocapsid proteins from other viruses have similar base sequences and this region is probably involved in the interaction of the protein with the RNA (e.g., Fiers et al., 1978; Shinnik et al., 1981; Kitamura et al., 1981). However, there is less sequence homology in this part of the molecule between the two viruses than in the carboxyl-terminal part of the proteins (residues 166-267) where 76% of the residues are identical and another 6% are conservative substitutions. This carboxylterminal region is probably important for other critical functions of the protein such as those protein-protein interactions that stabilize the nucleocapsid structure and those that are formed between the spike proteins and the capsid proteins during viral budding. Serine-219 in the SFV capsid protein is part of a sequence (Gly-Asp-Ser-Gly) characteristic of serine proteases (Boege et al., 1981). The same sequence is found in an analogous position in the Sindbis virus capsid protein. This might be part of the catalytic site of the putative autoprotease activity of the capsid protein (see Section III,B, 1). The amino-terminal methionine is acetylated in the Sindbis virus capsid protein (Bell and Strauss, 1981) and in SFV it is also blocked (Kalkkinen et al., 1980). Such blocks are fairly common in the structural proteins of viruses and might protect them against proteolytic degradation by cellular exopeptidases (Jornvall, 1975).

84

KAI SIMONS AND GRAHAM WARREN

vEcoR1 ATT GGT GCG TTA ATA CAC ~ G AATT STG ATT ATA GCG CAC TAT TAT AGC ACC ATG AAT TAC ATC CCT ACG CAA ACG -C

(C)

MET ASN TYR I L E PRO THR GLN THR

---

--- ARG GLV

PHE PHE A f N MET

TTT TAC GGC CGC CGG TGG CGC CCG CGC CCG GCG GCC CGT CCT TGG CCG TTG CAG GCC ACT CCG GTG GCT CCC GTC

(c)

PHE TYR GLY@TRP@PRO LE"

-_- --_

PRO PHE

Q

Q

ARG PRO ALA ALA RG PRO TRP PRO LEU GLN ALA THR PRO VAL AMPRO VAL

--- A A --- THR --- M T

TRP ARG

--- ARG --- --- ALA --- MgT PRO ALA ARG ARG ARG

GTC CCC GAC TTC CAG GCC CAG CAG ATG CAG CAA CTC ATC AGC GCC GTA AAT GCG CTG ACA ATG AGA CAG AAC GCA ASN ALA LEU THR MET@GLN

ASN ALA

SER

ALA THR

--- ---

VAL I $ E GLV

---

ATT GCT CCT GCT AGG CCT CCC A M CCA AAG AAG AAG AAG ACA ACC AAA CCA AAG CCG AAA ACG CAG CCC AAG AAG

ATC AAC GGA AAA ACG CAG CAG CM AAG AAG AAA GAC AAG CAR GCC GAC AAG AAG AAG AAG AAA CCC GGA AAA AGA

(c)

ILE ASN GLY@THR L y S L"S PRO

--- ---

mpRo---a

GLN GLN GLN LYS LYS YS ASP LYS GLN ALA ASP LYS LYS LYS LYS LYS PRO GLY LYS ~ R G

--- GLU Lys --- ---

---

ALA

--

GAA AGA ATG GGC ATG AAG A T 1 GAA AAT GAC TGT ATC TTC GAA GTC AAA CAC GAA GGA AAG GTC ACT GGG TAC GCC

ASP

TGC CTG GTG GGC GAC AAA GTC ATG AAA CCT GCC CAC GTG AAA GGA GTC ATC GAC AAC GCG GAC CTG GCA AAG CTA

GCT TTC AAG AAA I C G AGC AAG TAT GAC CTT GAG TGT GCC CAG ATA CCA GTT CAC ATG AGG TCG GAT GCC TCA AAG

FIG.1. Nucleotide sequence of the SFV 26 S RNA (top row), the corresponding amino acid sequence (middle row), and the amino acid sequence of the Sindbis virus structural proteins (bottom row). Nucleotides are numbered from the 5' end of the RNA molecule and all amino acids from the amino terminus of each protein. The amino- and the carboxyl-terminal ends of each protein are indicated by arrows, glycosylation sites by triangles, and membrane-spanning regions of the viral glycoproteins by underlines for Sindbis virus and overlines for SFV. Amino acids in boxes are negatively charged (Asp and Glu), and those circled are positively charged (Lys and Arg). Some restriction endonuclease cleavage sites are shown on the nucleotide sequence. The alignment of the amino arid

SEMLIKI FOREST VIRUS

*

TAC ACG CAT GAG AAG CCC GAG GGA CAC TAT PAC TGG CAC CAC GGG GCT GTT CAG TAC AGC GGA GGT AGG TTC ACT

ATA CCG ACA GGA GCG GGC AAA CCG GGA GAC AGT GGC CGG CCC ATC TTT GAC AAC AAG GGG AGG GTA GTC GCT ATC

___ ___ GTC CTG GGC GGG GCC AAC GAG GGC TCA CGC ACA GCA CTG TCG GTG GTC ACC TGG AAC AM GAT ATG GTG ACT AGA (c)

___ ___ ___ _-____

.

LEU SER VAL VAL THR TRP ASN --- __-THR --- _-_AM --- --- --- --- --- --- --- --SER

VAL LEU GLY GLY ALA A S N ~ G L YSER@THR Asp

*

GTG ACC CCC GAG GGG TCC GAA GAG TGG TCC GCC CCG CTG ATT ACT GCC ATG TGT GTC CTT GCC AAT GCT ACC TTC

m h I CCG TGC TTC CAG CCC CCG TGT GTA CCT TGC TGC TAT GAA AAC AAC GCA GAG GCC ACA CTA CGG ATG CTC GAG GAT

(D) PRO CYS PHE GLN PRO PRO CYS VAL PRO CYS CYS

_-- ___

ASP ARG

--- ---

THR

TYRHASN AMMALA LEU@NET ---

--- --- THR

ASN

LEU=]

THR

ARG GLU PRO SER ARG ALA

ASP ILE

--- --- G L U

AAC GTG GAT AGG CCA GGG TAC TAC GAC CTC CTT CAG GCA GCC TTG ACG TGC CGA AAC GGA ACA AGA CAC CGG CGC

(B) ASN V A L ~ P R GLY O TYR --- __-ASN H I S GLU A t A ---

TYRBLEU ALA --- ----- --- --LEU GLN

ASP THR

AIN

ALA LEU THR C Y S @ A ~ N GLY THR@HIS ILE

ARG

GLV SER SER GLV

--- SER

sHinfl CAC TCG TGT CAT AGC CCC GTA GCA ATT GAA GCG GTC AGG TCC GAA GCT ACC GAC GGG ATG CTG AAG AT1 CAG TTC

(R) HIS SER CYS HIS SER PRO VAL ALA I L E ~ A VM AL@SER~ALA GLU PRO

--- PHE ---

--- ---

LVS

--- ---

GLN

---

TRP ASP

---

THRBGLY NET LEU@ILE

GLN PHE

--- ASP --- ASN THR I L*E ARG . --- ---

THR

FIG. 1 (continued).

sequences of the two alphaviruses has been made to niaxiniize homology and therefore numerous small deletions (empty spaces) and insertions (aminoacids below each other) are present. A dashed line in the position of an amino acid in the Sindbis virus sequence indicates homology with the SFV sequence. A dot under an amino acid in the Sindbis virus sequence indicates a conserved change. From Garoff et al. (1982a), with permission.

86

KAI SIMONS AND GRAHAM WARREN (1275)

TCG GCA CAA ATT GGC ATA GAT AAG AGT GAC AAT CAT GAC TAC ACG AAG ATA AGG TAC GCA GAC GGG CAC GCC ATT

SER LEU LVS

GAG AAT GCC GTC CGG TCA TCT TTG AAG GTA GCC ACC TCC GGA GAC TGT TTC GTC CAT GGC ACA ATG GGA CAT TTC

(I331

PHE VAL HIS GLY THR NET GLY HIS PHE ARG ARG LEU SER TVR LVS

---

TVR

---

ATA CTG GCA AAG TGC CCA CCG GGT GAA TTC CTG CAG GTC TCG ATC CAG GAC ACC AGA AAC GCG GTC CGT GCC TGC (Q)

I L E LEU ALA@CYS

PRO PRO GLY --- --- --- _-- --_ -__---

Liu

Q

ASP THR ARG ASN ALA VAL ARG ALA CYS SER AtA T

QS$Ro---

SER

---

AGA ATA CAA TAT CAT CAT GAC CCT CAA CCG GTG GGT AGA GAR AM TTT ACA ATT AGA CCA CAC TAT GGA AM GAG

( ~ 2 ) AR ILE GLN TYR HIS HIS ASP PRO GLN PRO VAL GLY-PHE QLiU

ALA ARG LVS ILE

--- LVS PHE ---

---

THR LIE --- --- --- TLR ASP LEU

PRO HIS TYR G

L

Y

~

ATC CCT TGC ACC ACT TAT CAA CAG ACC ACA GCG GAG ACC GTG GAG GAA ATC GAC ATG CAT ATG CCG CCA GAT ACG (E2)

--- ___

I L E PRO CYS THR THR TYR GLN GLN THR THR A L A m T H R

_-- ---

VAL

---

ASP ARG LEU LYS GLU THR

---

CCG GAC AGG ACG TTG CTA TCA CAG CAA TCT GGC AAT GTA AAG ATC ACA GTC GGA GGA AAG AAG GTG AM TAC AAC

Q

( ~ 2 ) P R O ~ T H LEU R LEU SER GLI~ GLN SER GLY ASN VAL LYS ILE THR VAL GLY GLY LYS LYS VAL LYS TYR A ~ N ALA TVR THR S i R TVR

---

GLU GLU SER

--- ---

LVS

---

A A LVS PRO PRO SER ---QIbEQ---

GLU

TYR

TGC ACC TGT GGA ACC GGA AAC GTT GGC ACT ACT AAT TCG GAC ATG ACG ATC AAC ACG TGT CTA ATA GAG CAG TGC (Ez)

CYS THR CYS GLY THR GLY ASN VAL GLY THR THR ASN SER

___

LVS

-_- _-_ASP TVR LVS THR --- --- VAL

SER TgR

NET THR I L E ASN THR CYS LEU I L E m G L N CYS THR GLU

--- THR GLV ---THRALA --- LVS --- ---

FIG. 1 (continued, see legend on pp. 84-85).

B . The Viral Envelope 1 . Viral Glycoproteans The systematic study of how detergents solubilize the viral proteins laid the basis for our understanding of how the viral particle is built. The

87

SEMLIKI FOREST VIRUS AAA GGC AAA GTC CAT ATC CCA TTC CCG TTG GAC AAC ATC ACA TGC AGA GTT CCA ATG GCG CGC GAA CCA ACC GTC

ATC CAC GGC AAA AGA GAA GTG ACA CTG CAC CTT CAC CCA GAT CAT CCC ACG CTC TTT TCC TAC CGC ACA CTG GGT

( ~ 2 ) LIE HIS GLY LYS RG GLU VAL THR LEU HIS LEU HIS

-_- --- -_-W

I

L

E SER

PROMHIS LEU PHE PRO THR

SER TYR@THR

--- GLN --- ASP THR --- --- LEU --- --- LEU T i R

THR

LEU GLY

--- ARG --- ---

GAG GAC CCG CAG TAT CAC GAG GAA TGG GTG ACA GCG GCG GTG GAA CGG ACC ATA CCC GTA CCA GTG GAC GGG ATG

GAG TAC CAC TGG GGA AAC AAC GAC CCA GTG AGG CTT TGG TCT CAA CTC ACC ACT GAA GGG A M CCG CAC GGC TGG ( ~ 2 ) ~ T Y HSI R TAP GLY ASN A S N ~ P R OVAL@LEU

--- ---

ILE

--- --- ---

H I S GkU

--- --- ---

TRP SER GLN LEU THR T H R ~ G L Y @ P R O HIS GLY TRP

VAL TYR *LA

---

GLu

sgR

ALA PRO

---

Asp

_-____

___ ___

CCG CAT CAG ATC GTA CAG TAC TAC TAT GGG CTT TAC CCG GCC GCT ACA GTA TCC GCG GTC GTC GGG ATG AGC TTA

(D PRO H I S GLN I L E VAL GLN TYR TYR TYR GLY LEU TYR PRO ALA ALA THR VAL SER ALA VAL VAL GLY MET SER LEU

___ ___

GLU

--- --- ---

HIS

--- --- H I S ARG H I S - - - V i L

TVR

---

I t E LEU

--- --- A P SER ALA TgR

VtL

CTG GCG TTG ATA TCG ATC TTC GCG TCG TGC TAC ATG CTG GTT GCG GCC CGC AGT AAG TGC TTG ACC CCT TAT GCT

(R) LEU ALA LEU ILE SER LIE PHE ALA SER CYS TYR MET LEU VAL ALA ALA@SER@CYS ALA HE1 HgT

--- GLV V$L

THR VAL ALA VAL L E U CYS ALA CVS LVS

---

LEU THR PRO TYR AM

--- ARC GLU --- --- --- --- --- ---

TTA ACA CCA GGA GCT GCA GTT CCG TGG ACG CTG GGG ATA CTC TGC TGC GCC CCG CGG GCG CAC GCA GCT AGT GTG (u/80 LEU THR PRO GLY AMALA VAL PRO TRP THR LEU GLY ILE LEU CYS CYS ALA PRO@ALA

--- ALA --- ASN ---

V$L I t E

---

THR S I R

---

ALA LEU

--- --- ---

VOL ARG SER

E2-6K

HIS ALA ALA SER VAL

--- ASN --- GLU TIJR

PHE

GCA GAG ACT ATG GCC TAC TTG TGG GAC CAA A4C CAA GCG TTG TTC TGG TTG GAG TTT GCG GCC CCT GTT GCC TGC

(wALA ~ ,HI1

T H RET R ALA TYR LEU TRP

GLN ASN GLN ALA LEU PHE TRP LEU

_-_-____-SER --- __---_ SER A i N

SER

---

PRO PHE

--- ---

PHE ALA AMPRO VAL ALA CYS

VOL GLN LEU

___

FIG. I (continued, see legend on pp. 84-85).

results obtained also led to important insights into the mechanisms by which detergents solubilize biological membranes (Helenius and Simons, 1975). These results have been reviewed previously (Simons et al., 1977, 1978) and will not be covered here. a. Subunit Structure and Topology. The subunit structure of the mem-

88

KAI SIMONS AND GRAHAM WARREN

ATC CTC ATC ATC ACG TAT TGC CTC AGA AAC GTG CTG TGT TGC TGT AAG AGC CTT TCT TTT TTA GTG CTA CTG AGC (6K)

I L E LEU ILE I L E THR TYR CYS LEU

8

(2475)

ASN VAL LEU CYS CYS CYS YS SER LEU SER PHE LEU VAL LEU LEU SER

--- ALA GLV ALA

(53)

MET ARG CVS

(118)

CTC GGG GCA ACC GCC AGA GCT TAC GAA CAT TCG ACA GTA ATG CCG AAC GTG GTG GGG TTC CCG TAT AAG GCT CAC

(259)

--- PRO LEU A U ALA PtJE

I L E V&L

--- SER ---

S LEU PRO PHE LEU V&L

*

6K-El

Q-----a-----

LEU GLY ALA THR ALA RG ALA TYR GLU HIS SER THR VAL NET PRO ASN VAL VAL GLY PHE PRO TYR@ALA TVR LEU

(u)

LVS VOL

ALA

THR V&L

--- --- --- PRO GLN I L E --- ---

HSI

--- LEU

(IS)

(IS)

A T 1 GAA AGG CCA GGA TAT AGC CCC CTC ACT TTG CAG AT6 CAG GTT GTT GAA ACC AGC CTC GAA CCA ACC CTT AAT

(2625)

I L E ~ P R GLY O TYR SER PRO LEU THR LEU GLN NET GLN VAL VAL GLU THR SER LEU GLU PRO THR LEU ASN

---

(43) (43)

TTG GAA TAC ATA ACC TGT GAG TAC AAG ACG GTC GTC CCG TCG CCG TAC GTG AAG TGC TGC GGC GCC TCA GAG TGC

(27u))

VbL

(u)

---

--- --- A U --- --- ALA --- --- ASN --- GLU I k E THR --- M

-_---- _-- --- ---

L y s ICE

SLR THR

- --- CYS _--GLY --- ALA

TYR VALQCYS

L E U ~ T Y LIRE THR CYS GLN

i T u S i R GLU V & g - - -

SER SLEU E

R--~ C--Y S

(a) (68)

pnlntl TGC TTC TGC GAC TCA GAA AAC ACG CAA CTC AGC GAG GCG TAC GTC GAT CGA TCG GAC GTA TGC AGG CAT GAT CAC

(a)

4

CYS PHE C Y S ~ S E R ~ A THR S N GLN LEU S E R ~ A U I TYR V A L ~ S E ASP R VAL CYS ARC HSI

ASP HSI

__---_ _-- --- _-- -_-_-- S i R _--M i l --- --- --- --- --_ GLU LEU --- AL --- --- Q S E R U - - -

GCA TCT GCT TAC AAA GCC CAT ACA GCA TCG CTG AAG GCC AA4 GTG AGG GTT ATG TAC GGC AAC GTA AAC CAG ACT

(u)

(u)

ALA SER ALA TYR@ALA

---

GLN

---

ILE

S THR ALA SER LEU@ALA --- V&L HI----- --- ALA MET --- VOL

VAL NET TYR GLY ASN VAL

A%

I t E VOL

THR SER PHE

--- --- --- THR

GLN THR

(2850)

(u)

(u) (2%) (143)

(143)

GTG GAT GTT TAC GTG AAC GGA GAC CAT GCC GTC ACG ATA GGG GGT ACT CAG TTC ATA TTC GGG CCG CTG TCA TCG

(3ooo)

V A L ~ V ATYR L VAL ASN GLY ASP HIS ALA VAL THR LIE GLY GLY THR GLN PHE LIE PHE GLY PRO LEU SER SER

(168) (168)

--- --- -_---- --- --- U T H R PRO GLV --- SER LVS ASP LEU LVS VAL -:ALA - --- --- I k E --- ALA FIG. I

(continued, see legend on

pp. 84-85).

brane glycoproteins has been deduced from studies in which the virus is cross-linked with bifunctional, amino-reactive reagents, and the products solubilized with the aid of detergents (Simons et al., 1973a; Ziemiecki and Garoff, 1978; Rice and Straws, 1982). In SFV the com-

89

SEMLIKI FOREST VIRUS GCC TGG ACC CCG l l C GAC AAC AAG ATA GTC GTG TAC pJ\R GAC GAA GTG TTC AAT CAG GAC TTC CCG CCG TAC GGA

(BE)

TCT GGG CAA CCA GGG CGC TTC GGC GAC ATC CAA AGC AGA ACA GTG GAG AGT AAC GAC CTG TAC GCG AAC ACG GCA

(fl90)

GLY ___ -__@PHE --- --- --

A U

m,

Lys

LEU TYR ALA ASN THR ALA

LIE GLN

SER GLV GLN PRO GLY

---

&LA

---

ILE

--- SER --- ASP

CTG AAG CTG GCA CGC CCT TCA CCC GGC ATG GTC CAT GTA CCG TAC ACA CAG ACA CCT TCA GGG TTC

CTA AAG GAA AAA GGG ACA GCC CTA AAT ACG AAG GCT CCT TTT GGC TGC CAA ATC AA4 ACG AAC CCT GTC AGG GCC

CaOO)

L

E

___

U

~

ALA@PRO LEU L l s

G

A u L y S ASN

L THR Y ALA LEU ASN

A u SER

G U MET

--y

THRGJALA PRO PHE GLY CYS GLN ILE@HR

ASN PRO VAL@ALA

0

(ED

___ ___

(268)

ATG ARC TGC GCC GTG GGA AAC ATC CCT GTC TCC ATG AAT TTG CCT GAC AGC GCC TTT ACC CGC A T 1 GTC GAG GCG

(395)

LYS ASN ASN SER

(a) NET

---

ARG PRO

---

G t N GLU THR

---

--- ---

--- ---

CYS ALA VAL GLY ASN ILE PRO VAL SER NET ASN LEU PRO

ASI

---

V i L ASP

(El)

Q

CmS,

m

ItE A p

(El)

TAT TGG

a)

PRO GLY RET VAL HIS VAL PRO TYR THR GLN THR PRO SER GLY PHE LYS TYR TRP ___ SER _____ ___ --- --- --- --- ---------

a) LEU@LEU

(a)

pJ\R

(w1)

SER TYR

--- --- --- ---

IkE

---

I t E ASP

ItE

---

___

L ~ S

AM

___ ___

V A ~

SER AM PHE THR@ ALA

--- ---

ILE

-

LEU

LIE VAL TUR SER

CCG ACC ATC AT1 GAC CTG ACT TGC ACA GTG GCT ACC TGT ACG CAC TCC TCG GAT TTC GGC GGC GTC TTG ACA CTG

(YFjO)

ACG TAC AAG ACC AAC AAG AAC GGG GAC TGC TCT GTA CAC TCG CAC TCT AAC GTA GCT ACT CTA CAG GAG GCC ACA

W)

0 8

a

___ --- --- --- ---

THR TYR LYS THR ASN LYS ASN GLY ASP CYS SER VAL H I S SER H I S SER ASN VAL ALA THR LEU GLN GLU ALA THR GLN

---

GCA

Ap9.

L SER ASP A6.

GLU

---

---

SER THR

--- --- --- ---

---

GTG AAG ACA GCA GGT AAG GTG ACC TTA CAC TTC TCC ACG GCA AGC GCA TCA CCT TCT l l T GTG GTG TCG

---Q---Q

ALA YS VAL LY THR ALA GLY YS VAL THR LEU H I S PHE SER THR ALA SER ALA SER PRO SER PHE VAL VAL SER V g L R

GLU LYS

--- ---

VPL

--- --- --- --- --- ---

PRO GLN ALA ASN

---

kE

I

--- ---

(W)

(W)

(3Mo) (168) (168)

FIG. I (continued, see legend on pp. 84-85).

plex of E l , E2, and E3 is held together by weak interactions and can be solubilized intact using the mild nonionic detergent Triton X-100. However, if antibodies to either the E l or the E2 proteins are added, the polypeptide chains dissociate from each other. This separation can also be observed when deoxycholate is used to solubilize the spike glycoproteins (Helenius et al., 1976).

90

KAI SIMONS AND GRAHAM WARREN

CTA TGC AGT GCT AGG GCC ACC TGT TCA GCG TCG TGT GAG CCC CCG AAA GAC CAC ATA GTC CCA TAT GCG GCT AGC

(u)LEU CYS SER ALA@ALA

THR CYS SER ALA SER C Y S ~ P R OP R O ~ ~ H LIE S VAL PRO TYR ALA ALA SER

__---- GLY LYS LYS THR --- --- ASN --- GLU --- LYS --- --- ALA

--- --- --- ---

SER THR PRO H I S LYS

CAC AGT AAC GTA GTG TTT CCA GAC ATG TCG GGC ACC GCA CTA TCA TGG GTG CAG AAA A I ? % %

(ED HSI

SER ASN VAL VAL PHE PRO

MET S E R ~ L YTHR ALA LEU SER TRP VAL GLN IbE

ASN ASP GkN GLU PHE GLN ALA

GGT CTG GGG

--- LYS --- SER

TRP

--- ---

L I U PHE

GCC TTC GCA ATC GGC GCT ATC CTG GTG CTG GTT GTG GTC ACT TGC ATT GGG CTC CGC AGA TAA GTT AGG GTA GGC El

(u)ALA PHE ALA SER LEU I L E IEll

LIE GLY ALA ILE LEU VAL LEU VAL VAL VAL THR CYS LIE GLY

--- --- LEU MET I k E PHE ALA CYS

LEU^ --- --+I

SER MET MET LEU THR SER THR

AAT GGC AT1 GAT ATA GCA AGA AAA TTG AAA ACA GAA AAA GTT AGG GTA AGC RAT GGC ATA TAA CCA TAA CTG TAT

(XU)

pAvol AAC TTG TAA CAA AGC GCA ACA AGA CCT GCG CAA TTG GCC CCG TGG TCC GCC TCA CGG AAA CTC GGG GCA ACT CAT

(3975)

TTG CAA TTG GTT TTT AAT ATT TCC

FIG. 1 (continued, see legend on pp. 84-85).

If the virus is treated with proteolytic enzymes the fuzzy layer formed by the viral spikes is removed (Osterrieth, 1965;Compans, 1971;Gahmberg el al., 1972; Sefton and Gaffney, 1974; Utermann and Simons, 1974).Remnants of both El and E2 are left in the bilayer. These have a hydrophobic amino acid composition, and are soluble in lipid solvents such as chloroform-methanol. The amphiphilic nature of the spike protein is also evident from its capacity to bind Triton X-100(0.6 g/g protein) which binds to the hydrophobic part to form a water-soluble protein-detergent complex (Simons et al., 1973a). The ability of amphiphilic proteins to bind Triton can be used to separate them from hydrophilic proteins using an extraction procedure recently described

SEMLIKI FOREST VIRUS

91

by Bordier (1981). The virus membrane is solubilized with Triton X- 114, another detergent of the octylphenolpolyoxyethyIene series, and hydrophilic proteins are separated from the amphiphilic ones simply by raising the temperature to 30°C. At this temperature, Triton X-114 separates into a detergent phase containing the viral spike glycoproteins leaving the viral nucleocapsids in the aqueous phase (G. Warren, unpublished observations). The hydrophobic peptide segments of El and E2, which attach the spike protein to the lipid bilayer, can be localized on the polypeptide chains by a mapping procedure first used by Dintzis (1961)to show that the synthesis of polypeptide chains begins at the amino-terminal end. The hydrophobic stubs left in the viral membrane after protease treatment are found at the carboxyl-terminal ends of both the E 1 and the E2 polypeptides (Garoff and Soderlund, 1978). Further studies have shown that not only do the carboxyl-terminal regions of the El and the E2 proteins penetrate into the lipid bilayer, but the E2 chain also spans the membrane. When the virus is labeled from the outside and from both sides with f~rmyl[~~S]rnethionyl sulfone methyl phosphate, one additional basic peptide derived from the E2 chain can be labeled (Garoff and Simons, 1974;Simons et al., 1980).This is assumed to be derived from the internal domain of the E2 chain. This internal domain can be demonstrated more directly in vesicles derived from ER membrane after assembly of the viral glycoprotein in the infected cell. These vesicles are “inside out” when compared to the viral particle. Protease digestion of such vesicles removes about 25-30 amino acids from the carboxyl-terminal region of the E2 chain (Garoff and Soderlund, 1978).No comparable evidence has been obtained for the E l chain. Another approach using the cross-linker dimethyl suberimidate, which cross-links reactive groups that are about 1 1 A apart, shows that the spike glycoproteins in the SFV can be cross-linked to the underlying nucleocapsid probably by links between the internal domain of the E2 chain and the capsid protein (Garoff and Simons, 1974).However, it has not been possible to isolate glycoprotein-capsid oligomers, mainly because the basic capsid proteins prefer to cross-link with each other forming large polymers that do not penetrate into polyacrylamide gels (Garoff and Simons, 1974; Richardson and Vance, 1978a). But nucleocapsids cross-linked to glycoproteins can be isolated by density gradient centrifugation after detergent treatment. These contain up to 65% of the spike proteins in the original viral particle. With Sindbis virus, bifunctional amino-reactive reagents have not led to cross-linking of the spike proteins with the nucleocapsid (Rice and Strauss, 1982).However, treatment of the virus with formaldehyde results in such cross-links,

92

KAI SIMONS AND GRAHAM WARREN

suggesting that similar interactions also exist in Sindbis virus (Brown et aL, 1974). More evidence that the spike proteins are attached to the nucleocapsid can be obtained using mild detergents that solubilize the lipids from the viral particles but leave most of the spike proteins still attached to the nucleocapsid (Helenius and Kartenbeck, 1980). When SFV is treated with 22 mM octyl P-D-glucoside at neutral pH and at low ionic strength (10 mM NaCl), 80% or more of the spike proteins remain bound to the nucleocapsid. The bound spikes can still be seen after negative staining in the electron microscope. When either the pH is increased or the ionic strength is raised above 50 mM, the spike proteins dissociate from the nucleocapsid. This sensitivity to pH and to salt concentration suggests that the interaction of the spike proteins with the nucleocapsid depends on charged groups. If the pH is lowered to about 6.0 the SFV particle undergoes a dramatic decrease in diameter of about 70 A which is due to the contraction of the nucleocapsid (Siiderlund et al., 1972; von Bonsdorff, 1973). The viral membrane apparently adheres to the nucleocapsid during the contraction, and excess membrane is extruded in the form of blebs. Interestingly, few spike proteins are seen on these blebs suggesting that they contain only lipid and that the spike proteins remain bound to the nucleocapsid during shrinkage. These observations are all in keeping with the postulated interaction of the spike protein with the capsid protein (Garoff and Simons, 1974), and though they do not show exactly how these proteins interact, it is probable that each capsid protein binds one spike protein via the internal domain of the E2 chain. b. Primary Structure. The complete amino acid sequence (Fig. 1) of each of the viral glycoproteins has now been established (Garoff et al., 1980b; Rice and Strauss, 1981). They have been deduced from the sequence of the DNA complementary to the 26 S RNA messenger which codes for the structural proteins of the virus (see Section IILB). The coding region for the different proteins was localized from the aminoterminal and carboxyl-terminal amino acid sequences determined by conventional methods (Bell el al., 1978; Bonatti and Blobel, 1979; Kalkkinen, 1980; Boege et al., 1981; Kalkkinen et al., 1980; Welch et al., 1981; Garoff et al., 1982b). The results confirm and extend previous results showing that the genes for the structural proteins are arranged on the RNA in the order 5'-capsid-E3-E2-E1-3' (Clegg, 1975; Lachmi et al., 1975; Garoff and Siiderlund, 1978). The E3 chain is composed of 66 amino acid residues in SFV, and 64 in Sindbis virus. The E2 protein is 422 amino acids in length in SFV and

SEMLIKI FOREST VIRUS

93

423 in Sindbis. The El protein is slightly longer being 438 amino acids in SFV and 439 in Sindbis. The overall homology between the structural proteins of the two alphaviruses is striking; 47% of the residues are identical while another 12% represent conservative substitutions. The most gratifying aspect of these amino acid sequences is that they are fully consistent with the biochemical evidence on the organization of the El, E2, and E3 glycoproteins with respect to the lipid bilayer. Using the criteria proposed by Segrest and Feldmann (1974) to search for hydrophobic segments that could be embedded in lipid bilayers, three such segments can be found in the SFV protein sequences. Two of these are in the same hydrophobicity range and of the same length as the transmembrane segment of glycophorin (more than 25 residues in length and uninterrupted by charged amino acids) (Tomita and Marchesi, 1975). One is located in the carboxyl-terminal region of the E2 protein between glutamine-352 and alanine-391 in SFV and between proline-364 and cysteine-390 in Sindbis virus. The other is in the carboxyl-terminal region of the El protein between isoleucine-413 and leucine-436 in SFV and between threonine-405 and threonine-437 in Sindbis. In all of these segments there is a cluster of basic amino acids marking the carboxyl-terminal end of the hydrophobic peptide, and this is followed in SFV E2 by 3 1 more residues before the carboxyl terminus is reached, and in Sindbis E2 by 33 residues. This internal domain contains a lysine residue in position 440 in SFV (corresponding to an arginine in Sindbis) which is presumably the lysine which was labeled by f~rmyl[~~S]methionyl sulfone methyl phosphate or cross-linked to the capsid protein by dimethyl suberimidate (Garoff and Simons, 1974; Simons et al., 1980). This domain also contains a tyrosine at position 443, a likely cleavage site for chymotrypsin in experiments in which 25-30 amino acids were cleaved from the carboxyl-terminal region of E2 in microsomal vesicles (Wirth et al., 1977; Garoff and Soderlund, 1978). This internal domain in the E2 protein shows strong homology between SFV and Sindbis virus, and is probably involved in the interaction with the capsid protein in the viral particle. In the E l protein there are only two arginine residues on the carboxyl-terminal side of the hydrophobic segment. This explains why the approaches used to detect the internal domain of the E l protein failed. Although formal evidence is lacking it seems most likely that the hydrophobic segment of the El chain also spans the membrane and the two arginine residues are on the internal side of the bilayer. Further confirmation for the location of the membrane-spanning domains of the E 1 and of the E2 polypeptide chains has come from studies of Sindbis virus. After chymotrypsin digestion of the viral particle, the

94

KAI SIMONS AND GRAHAM WARREN

hydrophobic stubs left in the membrane have been isolated and sequenced. As expected, the amino-terminal amino acid sequence showed that the E l and the E2 proteins had been cleaved on the amino-terminal external side of the hydrophobic segments at phenylalanine-398 in E 1 and at tyrosine-359 in E2 (Rice et al., 1982). A number of putative transmembrane segments have been sequenced in several viral and cellular glycoproteins and a comparison of these sequences reveals certain common features (see Warren, 1981; Garoff et al., 1982a; Rice et al., 1982). Each segment has at least 20 residues and contains predominantly hydrophobic amino acids. Charged amino acids (Asp, Glu, Lys, Arg) are excluded as is Pro. These rules are probably most useful in showing which parts of a polypeptide could not span a lipid bilayer. However, they give no indication as to which residues of a putative spanning segment are actually within the lipid bilayer. This can be illustrated for the putative spanning sequence of E l for both SFV and Sindbis. By the above criteria the spanning segment would be eight residues longer in Sindbis than in SFV (Rice et al., 1982).Whether more of the Sindbis E 1 chain is actually located within the lipid bilayer than of the SFV E l will demand a more direct method of analysis. The third hydrophobic region found by the Segrest and Feldmann criteria is in the E 1 protein (Garoff el al., 1980b).This segment is located between valine-80 and cysteine-96 both in SFV and Sindbis virus. The segment is more highly conserved than the spanning segment of the E 1 and the E2 proteins. It does not conform to the criteria for spanning sequences because it is interrupted by a proline residue in both viruses. The function of this segment is not known but it may involve the fusion activity which appears to be a function of the El protein (see Section III,A,4). c. Carbohydrate Side Chains. The oligosaccharides bound to the E l , E2, and E3 proteins are of the N-glycosidic type with N-acetylglucosamine attached to the amide nitrogen of asparagine. Both high-mannose type and complex oligosaccharides are found in the SFV and Sindbis virus proteins (Sefton and Keegstra, 1974; Keegstra and Burke, 1977; Burke and Keegstra, 1979; Mattila et al., 1976; Mattila and Renkonen, 1978; Pesonen and Renkonen, 1976; Pesonen et al., 1979). They have a common Man-(1 -+ 6)-Man-(1 -+ 3)-Man-(1 -+ 4)-GlcNAc(1 + 4)-GlcNAc pentasaccharide core. The viral polypeptides are glycosylated by host cell enzymes, and carbohydrate side chains with similar structures are found in both cellular and viral glycoproteins (see Staneloni and Leloir, 1982).The ratio of high-mannose type to complex glycans in the SFV and Sindbis proteins varies with the host cell (Keegstra etal., 1975;Kaariainen and Pesonen, 1982).The glycosylationsites are determined by the structure of the protein. Not all of the asparagines can

SEMLIKI FOREST VIRUS

95

be glycosylated in part because potential sites must conform to the sequences Asn-X-Ser or Asn-X-Thr (Neuberger et al., 1972). The El protein has a single glycosylation site in SFV, which is glycosylated (Garoff et al., 1974; Mattila et al., 1976; Garoff et al., 1980b). When synthesized in BHK-21 cells it appears to contain a two-branched complex oligosaccharide chain, whereas the glycans of the El protein made in chick embryo fibroblasts are heterogeneous, consisting of multibranched and two-branched complex chains as well as of high-mannose chains (Mattila et al., 1976; Rasilo and Renkonen, 1979; Kaariainen and Pesonen, 1982).The cause of this heterogeneity is not known. In Sindbis virus the El protein has two potential sites, asparagine-139 and -245, both of which are glycosylated (Burke and Keegstra, 1976; Rice and Strauss, 1981). Both of the oligosaccharides are of the complex type when the virus is grown in BHK-2 1 cells, but in chick embryo fibroblasts only one is complex whereas the other is of the high-mannose type (Sefton and Keegstra, 1974). The Sindbis virus E2 protein has two potential sites, asparagine- 196 and -318 (Rice and Strauss, 1981). The former has a complex chain, whereas asparagine- 3 18 carries a high-mannose-type oligosaccharide (Sefton and Keegstra, 1974; Burke and Keegstra, 1976). The complex side chain is a two-branched structure of the type that is found also in other proteins (see Staneloni and Leloir, 1982). The E2 protein of SFV also has two potential sites, asparagine-200 and -264, both of which are glycosylated (Garoff el al., 1980b; Mattila et al., 1976; Rasilo and Renkonen, 1979). In BHK-21 cells, the E2 protein has one complex and one high-mannose side chain whereas in chick embryo fibroblasts both seem to be of the high-mannose type (Kaariainen and Pesonen, 1982). The E3 protein in Sindbis virus has one glycosylation site (asparagine-14) which is glycosylated (Welch and Sefton, 1979; Rice and Strauss, 1981). This protein, which in contrast to SFV is shed into the extracellular medium of infected cells, contains a single complex oligosaccharide. The E3 protein in SFV has two potential glycosylation sites, asparagine-13 and -60, and it carries only complex glycans. At least asparagine-13 appears to be glycosylated (Garoff et al., 1980b; Kalkkinen et al., 1980). 2. Lipids

The lipids in the viral envelope are taken from the host cell. Pfefferkorn and Hunter (1963) had already shown that the viral phospholipids are largely derived from cellular phospholipids synthesized before infection. Subsequent studies of the phospholipid, glycolipid, and cholesterol content of the alphaviruses have shown that the lipid composi-

96

KAI SIMONS A N D GRAHAM WARREN

tions are very similar if not identical to that of the host cell plasma membrane (Renkonen et al., 1971; Laine et al., 1972; Quigley et al., 1971; Hirschberg and Robbins, 1974). By growing the viruses in different host cells, large differences can be obtained in the viral lipid composition (Luukkonen et al., 1976). The small differences observed between the lipid compositions of the viral envelope and of the host plasma membrane can in general be attributed to the contamination of the plasma membrane preparations which cannot be purified to the same extent as those of the virus. Whether alphaviruses assert any selectivity on the set of lipids they take with them from the host cell plasma membrane is therefore difficult to ascertain. On the other hand, the lack of demonstrable specificity is consistent with what is generally known of proteinlipid interactions in biological membranes (Chapman et al., 1979; Seelig and Seelig, 1980). With present methods, specificity cannot usually be demonstrated. It is therefore reasonable to conclude that viral lipids are more or less passively incorporated into the viral particle during budding from the plasma membrane. The viral lipids are organized into a bilayer about 50 A in width (Harrison et al., 1971). The distribution of the different phospholipids between the two monolayers has been studied with SFV grown in BHK21 cells (van Meer et al., 1981).The phospholipids are localized by using phospholipid exchange proteins, by digestion with phospholipases, and by labeling with trinitrobenzenesulfonate. Phosphatidylcholine appears to be about equally distributed whereas phosphatidylethanolamine and sphingomyelin are enriched in the inner monolayer. Phosphatidylserine has not yet been localized. Altogether, 30% of the viral phospholipids can be assigned to the outer monolayer and 50% to the inner monolayer; 20% (phosphatidylserine and some minor phospholipids) have not yet been assigned. Apart from the phospholipids (48 mol% of the total lipid), SFV grown in BHK-21 cells contains 48 mol% cholesterol and 4 mol% glycolipids (Renkonen et al., 1971). Comparable experiments using VSV grown in BHK-2 1 cells have given similar distributions of lipids in the two halves of the bilayer (see Patzer et al., 1979). Table I gives a compilation of the molecular composition of SFV grown in BHK-2 1 cells, based on the revised weight for the viral particle of 41-42 x lo6 daltons (Jacrot et al., 1983). If one assumes that each hospholipid-cholesterol pair takes up a surface area of about 90- 100 (Israelachvili and Mitchell, 1975) and each glycolipid about 55 A* (Pascher and Sundell, 1977), then about 80% of the surface area in the bilayer is occupied by the lipids, leaving about 20% for the spanning proteins. This is somewhat more than would be expected if 180 spike proteins span the bilayer, each having two transmembrane a helical segments.

fii2

97

SEMLIKI FOREST VIRUS

TABLE I Molecular Composition of Semliki Forest Virus Based on a T = 3 SymmetT Model Component Nucleocapsid RNA Protein Envelope proteins El E2 E3 Lipids Phospholipids Cholesterol GIycolipids Virion

MW of component

Molecules per virion

Total MW x

4.1 x lo6 29.7 x 103

180

9.4 4.1 5.3

49 52 10

I80 180 180

8.8 9.4 1.8

10,000 10,000 650 -

7.8 3.9 0.8 41-42

x 103 x 103 x 109

775 385 1,200 -

I

The underlying assumptions in Table I are that the nucleocapsid is built according to T = 3 icosahedral symmetry, and that the symmetric arrangement of the spike proteins would be dictated by the direct interaction of one spike glycoprotein with one capsid protein. This interaction is assumed to be the basic structural design of the viral particle. There is no evidence that the nucleocapsid penetrates into the lipid bilayer and interacts with the lipids directly. The isolated nucleocapsid does not bind Triton X-100 and its primary structure shows no obvious hydrophobic regions (Helenius and Soderlund, 1973; Garoff et al., 1980a). To prove the basic design of the viral particle, the structure of the virus would have to be determined to high resolution. This has not yet been possible. Electron micrographs of thin sections from pellets produced by ultracentrifugation of SFV have shown that the regular arrays of particles seen represent three-dimensional crystals, the largest being up to 5 pm on one side (Wiley and von Bonsdorff, 1978). The probable space group was found to be F23. The diffraction pattern of the electron micrographs of the embedded and sectioned crystals of SFV revealed crystalline order to only 100 A resolution. R. Leberman (EMBL) was able to crystallize SFV by conventional methods. However, X-ray diffraction analyses of these crystals showed that they were difficult to handle and they did not diffract to high resolution. The length of the unit cell edge was 890 A indicating a nearest neighbor distance between viral

98

KAI SIMONS AND GRAHAM WARREN

particles of 630 (F. Winkler, unpublished observations).The particles seemed to be ,packed into a face-centered cubic lattice with specific neighbor contacts, but there was only short-range order. Thus particles separated by more than a few unit cells did not scatter coherently to better than about 40 A resolution. 111. THELIFECYCLEOF SEMLIKI FORESTVIRUS

The structure of an alphavirus particle is simpler than that of all known cellular organelles, but it is built according to the same principles. This is because the viral genome is small and the virus must use for its construction those cellular components normally engaged in the biogenesis of host cell membranes. This means that studies of viral replication can be exploited to study cellular funktions at the molecular level. Naturally viral infections also perturb cellular physiology, but there is usually enough time early in infection for studies to be carried out before cellular malfunction becomes a source of error. The life cycle of SFV is initiated by the delivery of the viral RNA into the cytoplasm of the host cell. The viral RNA is then transcribed into new 42 S RNA molecules and into 26 S messenger RNA molecules which are translated into viral structural proteins. Nucleocapsids are formed in the cytoplasm from the 42 S RNA molecules and capsid proteins. The viral glycoproteins are assembled in the ER membrane, then modified and transported via the Golgi complex to the cell surface. The newly made nucleocapsids bind to the cytoplasmic face of the plasma membrane via the viral glycoproteins, and function as a template for binding more spike glycoproteins. The plasma membrane becomes modified as it wraps around the nucleocapsid and is finally released into the extracellular medium. Only a few functions needed for virus replication are specified by the viral RNA; for the most part the virus exploits the normal function of the host cell. What is special about alphaviruses (and some other enveloped viruses) is that they have specifically adapted the mechanisms by which they enter and leave the host cell to existing routes of membrane traffic connecting the internal cellular compartments with the cell surface. A . Infection 1 . Binding to the Cell Surface The first phase in the entry of the virus into the cell is its binding to the cell surface. SFV can infect a wide variety of cultured cells of mammalian, avian, or invertebrate origin, suggesting that the virus must recog-

SEMLIKI FOREST VIRUS

99

nize surface structures common to many different cell types (Mussgay et al., 1975). For BHK-21 cells, SFV binds with an apparent binding constant of 3 X 10"M-' at pH 6.8; the apparent number of sites is 50 x lo3 (Fries and Helenius, 1979). Using the SFV spike proteins isolated in a water-soluble form as (El, E2, E3) octamers, binding to the H2 and HLA cell surface glycoproteins can be demonstrated (Helenius et al., 1978). However, more recent studies show that these common cell surface antigens cannot be the sole receptors for SFV. There are murine cell lines which do not express major H2-histocompatibility antigens but the cells can nevertheless be infected with SFV (Oldstone et al., 1980).The major problem confronting researchers studying virus-receptor interactions is the multivalency of the binding which leads to tight attachment of the virus to the cell surface even if the interaction between one individual spike protein and one cell surface receptor is of low affinity. The low affinity makes biochemical studies of cell surface receptors for SFV difficult, and which surface molecules can function as receptors for the virus remains unknown. The interaction does not seem to involve sialic acid, since this sugar is not found in mosquito cells (Stollar et al., 1976). Morphological studies show that SFV particles bound to BHK-21 cells are preferentially associated with the microvillar projections of the cell surface membranes (Helenius et al., 1980). Many of the virions which are not bound to microvilli (5% of all the cell surface viruses) are located in coated pits. The coated pits are invaginations of the plasma membrane, with a characteristic electron-dense coat composed of clathrin and other proteins on the cytoplasmic face (Pearse and Bretscher, 1981). Many of the coated pits are localized close to the base of microvilli.

2. Endocytosis Binding to the cell surface proceeds at O'C, but the cells are not infected (Helenius et al., 1980). When the cells are warmed to 37°C the virus is rapidly removed from the cell surface and infection ensues. In general there are two ways to envisage the entry of enveloped viruses into cells-either by penetration directly through the plasma membrane, or by endocytosis (engulfment by a plasma membrane-derived vesicle) (see Lonberg-Holm and Philipson, 1974). In both cases delivery of the nucleocapsid with the RNA would have to involve a fusion reaction between the viral envelope and either the cell surface membrane or the vesicle membrane. Paramyxoviruses are known to fuse their envelopes with the plasma membrane (see Hosaka and Shimizu, 1977). However, whether this process leads to productive infection has not yet been settled. Careful studies by electron microscopy have shown that SFV enters

100

KAI SIMONS AND GRAHAM WARREN

the cell by endocytosis (Helenius et al., 1980). No evidence for fusion at the cell surface has ever been obtained. The cell-bound viruses are rapidly (within seconds) internalized at 37°C into coated vesicles which form by invagination of the coated pits and usually carry only one viral particle. After 1 minute or longer, viruses can be observed accumulating into larger, irregularly shaped vacuoles. These prelysosomal vacuoles have been named endosomes (see Marsh et al., 1983) or receptosomes (see Pastan and Willingham, 1981). The endosomes are devoid of lysosomal enzymes, and are probably important in membrane recycling. From these vacuoles some viruses find their way into lysosomes, as demonstrated by lysosome-specific staining procedures (Helenius et al., 1980). The internalization process can be followed quantitatively by using radiolabeled virus and the susceptibility of bound virus to protease treatment; proteinase K at 0°C removes surface-bound viruses but not internalized viruses (Helenius et al., 1980). The uptake process is extremely efficient. The half-life of a virus particle on the cell surface is less than 20 minutes (Marsh and Helenius, 1980). The uptake is saturable at high virus concentrations. However, the saturation observed is due to saturation of binding and not of endocytosis. The highest average rate of uptake measured is around 2000 viral particles per minute per cell. Since, on average, 1.3 viral particles are internalized per coated vesicle, the measured uptake of 2000 viral particles per minute means that, on average, 1600 coated vesicles internalize viruses from the cell surface per minute. This uptake is not induced by the virus. The maximal rate of virus uptake corresponds to the ongoing endocytic rate of BHK-21 cells measured by [3H]sucrose uptake. Thus, SFV uptake must occur by a continuous cellular process. This process is clearly distinguished from phagocytosis which is blocked by cytochalasin B, does not involve small coated vesicles, shows different kinetics, and is induced by the particle to be phagocytosed (Silverstein et al., 1977). Instead, the SFV uptake into BHK-2 1 cells has exactly the same characteristics as receptor-mediated endocytosis (Goldstein et al., 1979).This interpretation is supported by a large number of similarities between SFV uptake and the endocytosis of physiological ligands such as low-density lipoproteins, asialoglycoproteins, epidermal growth factor, cup-macroglobulin, and lysosomal enzymes (Kaplan, 1981; Anderson and Kaplan, 1983). These similarities include temperature dependence; kinetics; the involvement of coated pits, coated vesicles, endosomes, and lysosomes; and the effect of inhibitors. The only efficient way to inhibit receptor-mediated endocytosis is to lower the temperature below 10°C. Inhibitors of oxidative phosphorylation also have a marked inhibitory effect on the rate of endocytosis, at least in BHK-21 cells. As shown for a number of physiological ligands,

SEMLIKI FOREST VIRUS

101

the number of receptors on the cell surface does not decrease by adding saturating amounts of SFV and letting the viruses endocytose. The cell surface receptors are probably taken into the cell with their ligands which then dissociate, and the receptors are then recycled back to the surface within a matter of minutes (Pearse and Bretscher, 1981). 3. The Endocytic Route Leads to Infection The first indication that the endocytic route leads to infection came from inhibitor studies. Five well-characterized lysosomal inhibitors (chloroquine, NH,Cl, amantadine, tributylamine, and methylamine) block infection by SFV (Helenius et al., 1980, 1982). These agents have no direct viricidal effect, and no effect on cell viability during the span of the experiments. Moreover, they do not block binding or endocytosis of the virus. They prevent release of the RNA from an intracellular vacuole into the cytoplasm. This release can be assayed by lysing cells and using ribonuclease as a probe for RNA location; when the nucleocapsid has penetrated into the cytoplasm the viral RNA becomes susceptible to ribonuclease attack. The common target of these lipophilic bases is the lysosome. Being weak bases they rapidly accumulate in this acidic organelle, and raise the lysosomal pH (De Duve et al., 1974; Poole and Ohkuma, 1981). Inhibition of SFV infection can be achieved using concentrations of the inhibitors which have been shown to increase the pH of the lysosomes from 4.8 to around 6. Recent results indicate that not only do the lysosomes have an acidic pH, but that the endosomes are acidic as well (Tycho and Maxfield, 1982). Thus, lysosomal inhibitors could also assert their inhibitory effect on SFV infection by raising the pH of the endosomes. In fact, it seems likely that this is the case (Marsh et al., 1983). Since lysosomal inhibitors like chloroquine and NH&l raise the lysosomal pH within seconds, they can be used to pinpoint the time at which infection can no longer be inhibited. Such studies show that addition of the inhibitor within 4 minutes after viral entry from the cell surface gives virtually complete inhibition of viral infection whereas after 6 minutes the agent is largely ineffective in preventing the penetration of the viral nucleocapsid into the cytoplasm. These times are more consistent with the effect being localized to endosomes since morphological studies show that viruses do not reach the lysosomes within 6 minutes after leaving the cell surface (Helenius et al., 1980; Marsh et al., 1983). More data have been obtained using other ligands. Dunn et al. (1980) have shown that in perfused livers the passage of asialoglycoprotein from endosomes to lysosomes can be prevented by simply lowering the temperature to 20°C. Endocytic uptake into the endosomes is slowed but not inhibited. At this temperature

102

KAI SIMONS AND GRAHAM WARREN

Marsh et al. (1983) have shown that SFV can still infect BHK-21 cells. Moreover, the release of nucleocapsid into the cytoplasm (as measured by the ribonuclease assay) proceeds, albeit 40% more slowly than at 37°C. Thin-section electron microscopy confirms that SFV particles are not transferred to lysosomes at this temperature. Although strongly suggestive, these experiments do not prove that the normal pathway of infection involves only the endocytic route. Infection could also occur by fusion of the virus with the plasma membrane. This was shown not to be the case by allowing uptake of SFV into BHK-21 cells for 10 minutes at 37°C in the presence of inhibitory concentrations of NH&l (Helenius et al., 1982).All of the viruses left on the cell surface were then removed by proteinase K digestion at OOC, and after removal of the inhibitor, the incubation continued at 3TC. The intracellular viruses were shown to infect the cells almost as efficiently as in control cells. 4. Penetration by Fusion

One would assume that the mechanism for delivery of the nucleocapsid through the membrane of the intracellular vacuole has to be provided by the virus. There are no known precedents in normal cell physiology for the passage of macromolecular assemblies like the viral nucleocapsids into the cytoplasm. The most likely mechanism would be fusion of the viral envelope with the vacuolar membrane and subsequent release of the nucleocapsid into the cytoplasm. But if penetration occurs by fusion why would this occur intracellularly and not at the cell surface? The clue comes from the low pH dependence of the infection. Low pH has been shown to induce an extremely efficient membranefusion activity (Helenius et al., 1980). The fusion activity of SFV is expressed first at pH values of 6 or lower. The most important tool to study this fusion process has been a quantitative assay based on the ribonuclease (RNase) sensitivity of the nucleocapsid RNA. Liposomes filled with RNase and mixed with SFV below pH 6 degrade the viral RNA introduced into the liposome interior after fusion of the viral envelope with the liposomal membrane. The viral glycoproteins are integrated into the lipid bilayer with the same orientation as in the viral particle; the spikes project from the external surface of the liposome. This assay has shown that besides low pH, the fusion reaction requires the viral spike glycoproteins and cholesterol (optimally one molecule per two phospholipid molecules) in the target membrane (White and Helenius, 1980). The fusion reaction takes place within seconds, does not require divalent cations, is not leaky (if the virus is not damaged, for instance, by freezing and thawing), and is more than 90% efficient.

SEMLIKI FOREST VIRUS

103

Although fusion between SFV and the plasma membrane normally does not occur, cell surface-bound viruses can be induced to fuse simply by decreasing the extracellular pH below 6 for a few seconds (White et al., 1980; Vaananen et al., 1981).As a result of its fusion activity SFV can hemolyze red blood cells at pH 5.8 (Vaananen and Kaariainen, 1979, 1980). However, the lysis occurs only with virus damaged by freezing and thawing. Cells can also be made to fuse with each other using SFV at low pH (White et al., 1981). The membrane fusion activity is probably a function of the El protein, since the hemolytic activity of Sindbis virus can be inhibited by monoclonal antibodies specific for E l (Chanas et al., 1982). Moreover, studies using cDNA molecules coding for the spike proteins have shown that if the spike protein is expressed at the cell surface, fusion between cells is induced at low pH. However, when the p62 protein is expressed alone, no fusion occurs (Kondor-Koch et al., 1983). Although the evidence is still incomplete, Sindbis virus seems to enter its host cells by the same mechanism as SFV (Talbot and Vance, 1980, 1982). Moreover, this mechanism is more general since influenza virus and VSV also enter canine kidney cells (MDCK) by endocytosis and by low-pH-mediated fusion ( M a t h et al., 1981, 1982). Influenza virus is especially interesting in this context. The major spike glycoprotein, the hemagglutinin, is responsible for the fusion process (Maeda and Ohnishi, 1980; Huang et al., 1981; White et al., 1981). The protein consists of two polypeptide chains HA1 and HA2 (see Simons and Garoff, 1980).The amino-terminal end of the HA2 polypeptide chain is hydrophobic (Skehel and Waterfield, 1975; Porter et al., 1979), and it is this part of the molecule which seems to be involved in the low-pHinduced fusion activity (Richardson et al., 1980). Bromelain treatment releases a water-soluble spike of the hemagglutinin molecule from the viral particle, leaving a short hydrophobic stub in the viral membrane (Brand and Skehel, 1972). The bromelain-released spike protein has been crystallized, and its three-dimensional structure has been determined to 2.8 8, resolution (Wilson et al., 1981). From this structure it is known that the amino-terminal end of the HA2 polypeptide is not available for direct interaction with a target lipid membrane. If, however, the water-soluble spike is subjected to a pH below 6, a drastic conformational change is induced which can be followed by circular dichroism (Skehel et al., 1982). As a result of the low-pH treatment, the spike protein becomes hydrophobic; it binds Triton X-100, it aggregates to form protein micelles, and it attaches to liposomes. Model building based on the spike protein structure suggests that the amino-terminal end of the HA2 protein can be exposed by a conformational change so

104

KAI SIMONS AND GRAHAM WARREN

as to participate in hydrophobic interactions. Through this interaction the viral membrane and the target membrane might be brought into the close proximity needed to initiate the fusion reaction between the two lipid bilayers. The viral fusion systems provide a unique opportunity to study membrane fusion at the molecular level. It would be surprising if the mechanism of spike protein-mediated fusion did not have features in common with the mechanisms involved in the fusion reactions occurring during membrane vesicle traffic in the cell. It may even be that the viral fusion proteins have evolved from cellular fusion proteins. The trigger for cellular fusion would of course be different and other proteins would be needed to specify which membranes are to fuse together. Analysis of SFV entry has thus shown that the virus binds to receptors on the cell surface and moves by lateral diffusion into coated pits to be internalized by coated vesicles. The endocytosed virus is delivered into endosomes, Here presumably, the viral envelope is activated by the low pH prevailing in this compartment to fuse with the vacuolar membrane. This results in the release of the viral nucleocapsid into the cytoplasm. During normal infection, the virus might not enter into lysosomes although SFV particles have been identified in this compartment using the large loads of virus needed to visualize the entry process by electron microscopy. Even if this were to happen normally, the viral nucleocapsid would escape destruction because of the rapidity of the fusion mechanism.

B. Synthesis After the nucleocapsid has been expelled into the cytoplasm, the RNA must be released from the capsid proteins to allow the viral 42 S RNA to function as a messenger. The uncoating of the nucleocapsid is probably induced by some mechanisms related to the penetration event. The incoming nucleocapsid must be changed in some way to make it different from newly made nucleocapsids, which, after assembly in the cytoplasm, form new virions. It might be that the low pH is the crucial factor. Low pH has a drastic effect on the SFV nucleocapsid causing it to shrink from a diameter of about 400 to 320 A (Soderlund et al., 1972). This conformational change may induce uncoating of the RNA, although clearly some other factor is also required since uncoating does not occur by low pH alone. The Sindbis virus nucleocapsid does not undergo a similar shrinkage, although the lowest pH tested was 6 and the pH in the endosomes and in the lysosomes is lower than this (Sdderlund et al., 1979). The incoming 42 S RNA serves as a messenger RNA for several non-

SEMLIKI FOREST VIRUS

105

structural proteins which are translated from one initiation site at the 5’ end of the molecule (Cancedda et al., 1975; Glanville et al., 1976a; Kaariainen and Soderlund, 1978). The 3’ end of the 42 S RNA contains the genes for the structural proteins but these are not translated. The nonstructural proteins are subunits of one or more RNA-dependent RNA polymerases which are needed for the replication of the viral RNA (Clegg et al., 1976; Lachmi and Kaariainen, 1976; Ranki and Kaariainen, 1979; Sawicki and Sawicki, 1980; Lehtovaara et al., 1980). Two major species of RNA molecules, new 42 S and 26 S RNA, are formed during viral infection (see Strauss and Strauss, 1977). The 26 S RNA molecule is a subgenomic copy of the 3’ end of the 42 S RNA and serves as a messenger RNA for the structural proteins (Kennedy, 1976; Wengler and Wengler, 1976; Schlesinger and Kaariainen, 1980). The viral RNA polymerase first transcribes the incoming positivestranded 42 S RNA into negative-stranded 42 S RNA, which in turn serves as a template for the synthesis of both new positive-stranded 42 S RNA molecules and 26 S RNA molecules. The transcription of the 26 S RNA is initiated internally on the negative-stranded 42 S RNA. RNA replication seems to take place on membranes in characteristic cytoplasmic vacuoles, called cytopathic vacuoles I (CPV I), which appear soon after infection and are not seen in uninfected cells (Grimley et al., 1968, 1972; Friedman et ab, 1972). The detailed mechanism of the RNA replication will not be dealt with here (for reviews see Strauss and Strauss, 1977; Kaariainen and Soderlund, 1978; Kennedy, 1980). The nucleotide sequence of the 26 S RNA has been determined for both SFV and Sindbis virus (Garoff et al., 1980a,b; Rice and Strauss, 1981; Riedel et al., 1982). In addition, the junction where the 5’ end of the 26 S RNA resides on the 42 S RNA has been sequenced and shows that all three reading frames are efficiently blocked before the first gene of the structural proteins is reached (Riedel el al., 1982). There is thus no overlap between viral genes for the nonstructural and the structural proteins. The 26 S RNA has one initiation site for protein synthesis located 50 nucleotides from the 5’ end (Clegg, 1975; Cancedda et al., 1975; Glanville et al., 1976a; Garoff et al., 1982a; Riedel et al., 1982). The four different structural proteins of the virus are generated by proteolytic cleavage occurring both during and after translation (Fig. 2). The sequences flanking the initiator AUG in the SFV 26 S RNA are accommodated within the structure CAXXAUG: that has been considered a possible consensus sequence for a eukaryotic initiation site for translation (Kozak, 1981). Downstream from the initiation codon by 7 bases is a sequence of 11 nucleotides (AUCCCLJACGCA), 9 of which (those underlined) are complementary to the purine-rich tract close to

106

KAI SIMONS AND GRAHAM WARREN C-TERMINUS OF C A P S I D PROTEIN

1

I

GLY-SER-GLU-GLU-TRP-SER-ALA-PRO-LEU-ILE

N-TERMINUS

C-TERMINUS

OF

OF

E3

OF

6K

OF

El

E2

11

I

PRO-ARG-ALA-HIS-ALA-ALA-SER-VAL-ALA-GLU

N-TERMINUS

C-TERMINUS

Ill

OF

6K

I ALA-THR-ALA-ARG-ALA-TYR-GLU-HIS-SER-THR

N-TERMINUS

C-TERMINUS

IV

-v

OF

E3 I

I

THR-ARG-HIS-ARG-ARG-SER-VAL-SER-GLN-HIS

N-TERMINUS OF E2 FIG.2. Proteolytic cleavages involved in the formation of the SFV structural proteins. Cleavages 1-111 take place during translation of the 26 S RNA, and cleavages IV-V during intracellular transport.

the 3’ end of the ribosomal 18 S RNA (UAGGAAGGCGU)(Riedel et al., 1982). From the initiation codon there is one open reading frame 3760 nucleotides long. The 3‘ untranslated region in the SFV 26 S RNA is 264 nucleotides long, not including the poly(A) tail which is 60-70 nucleotides in length (Clegg and Kennedy, 1974; Kaariainen and Soderlund, 1978). The gene order on the 26 S RNA has been established by a number of methods, and is now conclusively known from the nucleotide sequence (Clegg, 1975; Lachmi et al., 1975; Garoff et al., 1982a).The gene nearest the 5’ end is the capsid gene followed by the genes for E3, E2, and El. All the other genes are contiguous except for E2 and E l . These have a segment of 180 nucleotides between them coding for a polypeptide 60 amino acids long, which has been named the 6 K peptide in SFV (Welch and Sefton, 1980; Garoff et al., 1980b). In Sindbis virus it is 55 amino acids long (Welch and Sefton, 1979; Rice and Straws, 1981). This 6K

SEMLIKI FOREST VIRUS

107

peptide has been detected in infected cells, and presumably functions as a signal peptide for the E l protein (see Section III,B,2). 1. Formation of the Nucleocapsid About 1.5-2 hours after infection the first viral structural proteins can be detected in cells efficiently infected by SFV (see Kaariainen and Soderlund, 1978). By 3-4 hours the synthesis of host cell proteins is shut off. The mechanism of host protein shutoff is not yet known, but studies by van Steeg (1982) suggest that it is the viral capsid protein itself which is responsible for the selective inhibition of host protein synthesis. The capsid protein seems to reduce the activity of the initiation factors e 1 F4B and CAP binding protein below levels necessary for the formation of the 80 S initiation complex from host mRNA. Translation of the viral 26 S RNA is, however, unaffected. More studies are clearly needed to substantiate this interesting possibility. After the shutoff of host protein synthesis, the cell has essentially been converted into an assembly line for the production of new viral particles. To assemble new nucleocapsids only two components are needed: capsid proteins and 42 S RNA. The capsid protein is the first to be translated from the 26 S RNA and is cleaved from the nascent polypeptide chain soon after it has been completed (Clegg, 1975; Garoff and Soderlund, 1978; Garoff et al., 1978). The proteolytic cleavage may be catalyzed by the capsid protein itself (Aliperti and Schlesinger, 1978). The newly synthesized capsid protein first associates with the large ribosomal subunit before it binds to the 42 S to form nucleocapsids (Glanville and Ulmanen, 1976; Ulmanen et al., 1976; Soderlund and Ulmanen, 1977). This process is fast and efficient (Soderlund, 1973). Only completed nucleocapsids can be detected in the infected cell (see Kaariainen and SiSderlund, 1978). Neither empty capsids nor partially completed aggregates of RNA and capsid proteins have been identified. Apparently, not only protein-protein interactions, but also RNA-protein interactions play a decisive role not only in initiating the encapsidation process but also in stabilizing the nucleocapsid. Ribonuclease treatment leads to contraction of the nucleocapsid, and, in combination with EDTA treatment, the structure of the SFV nucleocapsid is destroyed (Kaariainen and Soderlund, 1971; Soderlund et ul., 1975). Wengler et al. (1982) were able to assemble nucleocapsids of the correct density and size from 42 S RNA and from isolated capsid proteins. Surprisingly, the interaction between the nucleic acid and the capsid protein was found to be fairly unspecific since it was possible to substitute the viral RNA with RNA and DNA molecules ranging in size from

108

KAI SIMONS AND GRAHAM WARREN

100 to 6000 nucleotides (e.g., tRNA and fdDNA). In the infected cell there must be some mechanism to prevent the 26 S mRNA from being encapsidated. This in vitro assay may make it possible to study the specificity and the mechanism of the assembly process in detail.

2. Assembly of the Viral Glycoproteins in the Endoplasmic Reticulum The ribosome that translates the capsid protein continues synthesis of the E3, E2, and E l polypeptide chains. Studies of infected cells show that these proteins are assembled in the membrane of the ER (Wirth et al., 1977; Garoff and SiSderlund, 1978). They are also glycosylated during (or soon after) synthesis (Sefton, 1977). The glycosylation is performed by host enzymes. The biosynthesis of N-glycosidic oligosaccharides involves the en bloc transfer of a preformed glycan from an oligosaccharide diphosphate dolichol intermediate to the nascent polypeptide chain (see Kaariainen and Pesonen, 1982; Staneloni and Leloir, 1982). The oligosaccharide chain consists of (Glc)S(Man)g(GlcNAc)*. The assembly of the virus proteins into the ER membrane can be studied in more detail in vitro. Early translation studies in vitro usually showed only capsid protein and small amounts of the membrane proteins with aberrant molecular weights (Cancedda and Schlesinger, 1974; Simmons and Strauss, 1974; Glanville et al., 1976b; Clegg and Kennedy, 1975a,b). Only after the introduction of the in vitro translation system, supplemented with microsomal vesicles (Blobel and Dobberstein, 1975a,b),did it become possible to translate the 26 S RNA into authentic products representing the structural proteins of the virus (Garoff et al., 1978; Bonatti and Blobel, 1979; Bonatti et al., 1979). By using an HeLa cell-free system together with microsomes from dog pancreas, four proteins can be made from the 26 S RNA: the capsid protein, the E l protein, a protein with an M , of 62,000 (the p62 protein), and small amounts of a large protein, the 97K protein (M,= 97,000) (Garoff et al., 1978). Exactly the same proteins are seen in infected cells labeled with a short pulse of radioactive amino acids. The p62 protein is a precursor for the E3 and the E2 proteins (Schlesinger and Schlesinger, 1972, 1973; Simons et al., 1973b; Garoff et al., 1974), whereas the 97K protein contains the sequences for all the viral membrane proteins (Lachmi et al., 1975). If the microsomes are left out of the translation system, only two proteins are made: the capsid protein and the 97K protein, In this laboratory the 97K protein has been found to be nonglycosylated both in vitro and in vivo (Garoff and Schwarz, 1978; Garoff et al., 1978). This apparently represents an aberrant product which is not processed to form the authentic membrane proteins. In cells infected with Sindbis virus there have been claims that the equivalent of the 97K

SEMLIKI FOREST VIRUS

109

protein, the B protein (Schlesinger and Schlesinger, 1972; Simmons and Strauss, 1974)is glycosylated (Hakimi and Atkinson, 1982). However, no evidence has been presented showing that the product analyzed was indeed the 97K protein and not a dimer of the E l protein or another protein (see Kaluza and Pauli, 1980). With the in uitro system the time course of assembly can be followed. The proteins are translated sequentially; capsid protein is followed by the p62 protein and then by the E l protein (Garoff et al., 1978). The assembly of the polypeptide chains can be monitored by protease treatment (Wirth el al., 1977). Microsomal vesicles are impermeable to proteases, and only structures on the outside (the cytoplasmic side) are accessible to proteolytic degradation. Most of the p62 and all of the E l polypeptide chains are inaccessible to protease as soon as they can be detected, whereas the capsid protein and the 97K protein are completely degraded. It is the carboxyl-terminal end of the p62 protein that is degraded as expected since the E2 part spans the membrane (Garoff and Soderlund, 1978). Transfer of the p62 and El chains through the ER membrane takes place concomitantly with translation. If microsomes are added to the in vitro system after translation of the polypeptide chains is completed, no transfer takes place. By synchronized translation experiments modeled according to Rothman and Lodish (1977), it can be shown that the microsomal membranes must be added before about 100 amino acids of the p62 chain have been translated if subsequent assembly of the protein into the membrane is to occur (Garoff et al., 1978). Thus the signal peptide responsible for the initiation of transfer to the ER must be located in the amino-terminal end of the p62 protein in the E3 part. From these experiments it is not possible to find out whether the E l protein has its separate signal peptide, mainly because of the low yields of the E l protein in the synchronized translation experiments. However, evidence for such a peptide has come from studies with a temperaturesensitive mutant of SFV in which the cleavage between the capsid and the p62 protein is blocked (Hashimoto et al., 1981).In cells infected with this mutant at the restrictive temperature, the E 1 protein is assembled in the correct orientation into the membrane of the ER. The uncleaved protein containing the capsid and the p62 sequences (M,= 87,000) is left in the cytoplasm. These findings suggest that the E l protein has its own signal peptide which might be located in the 6K protein. Studies with VSV and SFV were the first to show that membrane glycoproteins make use of the same mechanisms used by secretory proteins to become segregated into the ER lumen (Katz et d., 1977; Rothman and Lodish, 1977; Wirth et al., 1977; Lingappa et al., 1978;

110

KAI SIMONS AND GRAHAM WARREN

Toneguzzo and Ghosh, 1978; Garoff et al., 1978; Bonatti et al., 1979). A signal peptide at the amino-terminal end of the nascent polypeptide to be segregated is recognized by what we now know is a cytoplasmic signal recognition protein which binds the ribosome. This halts translation until the complex binds to a docking protein in the ER membrane so that transfer of the polypeptide chain across the membrane can then take place (Walter and Blobel, 1982a,b; Meyer et aL, 1982). However, unlike most secretory and membrane proteins the signal peptide is not removed by a signal peptidase from the E3 protein (Bonatti and Blobel, 1979; Garoff et al., 1982a). Ovalbumin is another exception (Braell and Lodish, 1982; Meek et al., 1982). The putative signal peptide for El, the 6K peptide, is unusually large for a signal peptide (60 amino acids) (see Kreil, 1981),and exactly when and how it is excised during translation is not yet clear. In contrast to secretory proteins which are delivered into the lumen of the ER, membrane glycoproteins are assembled into the ER membrane. Membrane anchorage is a function of the hydrophobic peptide segments at the carboxyl-terminal end of the proteins (see Sabatini et al., 1982). The polypeptide chain is apparently transferred into the lumen of the ER until the hydrophobic spanning segment prevents further transfer. There is also another class of plasma membrane glycoproteins and viral glycoproteins which are attached to the membrane by their amino-terminal ends, which are presumed to be their signal peptides (Blok et al., 1982; Desnuelle, 1979; Fields et al., 1981; Hauri et al., 1982). Those features of the carboxyl-terminal hydrophobic segment needed to attach the SFV proteins to the membrane have been studied by Garoff et al. (1983). They constructed a series of deletion mutants from the cDNA (copied from the 26 S SFV RNA), in which the DNA sequences coding for the carboxyl-terminal end of the E2 protein have been shortened, and the 6K and the E l regions deleted. These DNA molecules containing the genes for the capsid and the shortened p62 protein have been inserted, under control of the early SV40 promoter, in vectors designed for expression of cDNA molecules in animal cells. The DNA molecules have been introduced into the nucleus of BHK-21 cells by microinjection, and the expression of the p62 protein is studied by immunofluorescence using antibodies specific to the E2 protein. Their results show that a shortened gene coding for only three of the amino acids of the internal (cytoplasmic) domain of the E2 protein (together with seven extra amino acids provided by a stop-linker nucleotide) is expressed normally, assembled into the ER, and transported to the cell surface. However, when the deletion extends into the region coding for

SEMLIKI FOREST VIRUS

111

the hydrophobic domain, the E2 protein precursor appears not to become membrane bound but is instead secreted into the ER lumen. The spike protein complex of the alphaviruses appears to be assembled in the ER. Cross-linking studies of Triton X- 100-solubilized viral proteins have shown that the p62 and E l proteins are linked together to form a complex in the ER (Ziemiecki et al., 1980).The topology of these complexes is the reverse of that in the viral particle; the spikes are within the lumen of the ER and the internal domain of E2 is on the cytoplasmic side of the ER membrane. Glycosylation of the p62 ahd E l proteins is not needed for correct assembly in the ER, since in tunicamycin-treated cells, the assembly of these nonglycosylated proteins proceeds normally (Garoff and Schwarz, 1978). The drug tunicamycin blocks the assembly of the dolichol-linked oligosaccharide intermediate so that no transfer of oligosaccharides to nascent protein is possible (Tkasz and Lampen, 1975). C. Intracellular Transport of the Viral Glycopoteins

I . Posttranslational Modijications It takes about 30-60 minutes for the spike protein complex to reach the cell surface (Scheele and Pfefferkorn, 1969; Green et al., 1981) and during intracellular transport the viral glycoproteins become modified-the carbohydrate units are trimmed and extended (Kaariainen and Pesonen, 1982), fatty acid acylation occurs (Schmidt and Schlesinger, 1980), and the p62 protein is cleaved to form the E3 and the E2 proteins (Fig. 3) (Schlesinger and Schlesinger, 1972; Simons et al., 1973b). The topology of the spike proteins appears to remain the same as in the ER; only the internal domain of the E2 chain is exposed on the cytoplasmic side whereas the spike proteins project into the lumen of those intracellular compartments through which the proteins pass during transport to the cell surface (Ziemiecki et al., 1980). a. Carbohydrate Processing. Most of the data for the biosynthesis and processing of the carbohydrate side chains for membrane glycoproteins have been derived from studies of viral glycoproteins (Hubbard and Ivatt, 1981; Staneloni and Leloir, 1982); VSV has been used for most of these studies (Kornfeld et al., 1978). For the alphaviruses the evidence is not as complete, but it seems to correspond to the pattern established for VSV. After transfer of a primary high-mannose-type oligosaccharide [(Glc)s(Man)~(GlcNAc)2] to the polypeptide chain during translation, the glucose residues are removed. This is followed by the removal of the four a-(1--* 2)-linked mannose residues within 20-30 minutes after trans-

112

KAI SIMONS A N D GRAHAM WARREN Synthesis Core glycosylotion t-a 1

ER

t Fotty ocid ocyloti on (EX31 Trimming Terminol sugor addition

t-al

mediol

GOLG I STACK

trons

I

P 62 cleovoge

PM FIG.3. Intracellular transport route of the SFV spike glycoproteins from the endoplasmic reticulum (ER),over the Golgi apparatus, to the plasma membrane (PM).The cis cisternae do not react positively for acid phosphatase or thiamin pyrophosphatase, and do not label with ricin in thin frozen sections. The medial cisternae do not react positively for thiamin pyrophosphatase or acid phosphatase, but label with ricin. The trans cisternae are positive for all of these markers.

fer (Robbins el al., 1977; Hubbard and Robbins, 1979). Two or more mannose residues are then removed, but only after an N-acetylglucosamine residue has been linked to the internal pentasaccharide (Kornfeld et al., 1978; Harpaz and Schachter, 1980a,b). Terminal carbohydrate residues are then added to construct the complex oligosaccharide side chains. The conversion of high-mannose type to complex oligosaccharides can be monitored conveniently using the enzyme endo-P-Nacetylglucosaminidase H which cleaves the bond between the two N-

SEMLIKI FOREST VIRUS

113

acetylglucosamine residues in the high-mannose-type oligosaccharide (Tarentino and Maley, 1974; Tarentino et al., 1978). Complex oligosaccharides are resistant to this enzyme. The alphavirus glycoproteins become resistant to endo-/3-N-acetylgIucosaminidaseH about 20-30 minutes after synthesis (Robbins et d.,1977; Green et d.,1981). b. Fat9 Acid Acylation. Both Sindbis virus and SFV El and E2 proteins have been shown to contain covalently attached fatty acids (Schmidt et al., 1979; Schmidt, 1982).These are detected by labeling infected cells with [3H]palmitate. Either one or two molecules of palmitate are attached to the El polypeptide and five or six molecules to E2 in Sindbis virus. These are located in the carboxyl-terminal hydrophobic stubs obtained after treating Sindbis virus with chymotrypsin to remove the external spikes (Rice et al., 1982). The fatty acids are believed to be linked to serine and threonine residues through an ester bond. These bonds are labile to transesterification and to hydroxylaminolysis, but in no instance has the amino acid to which the fatty acid is attached been directly identified. A number of other glycoproteins both of cellular and viral origin are now known to be acylated (Magee and Schlesinger, 1982). The addition of the fatty acids to the alphavirus polypeptide chain occurs some 20 minutes after synthesis, but before the oligosaccharide chains are processed into the complex form (Schmidt and Schlesinger, 1980; Quinn et al., 1983). c. Cleavage of the p62 Protein. Four proteolytic cleavages are needed to produce the viral structural proteins of the alphaviruses from the polypeptide chain translated from the 26 S RNA (Fig. 2). The cleavages releasing the capsid protein, the p62 protein, and the El protein occur during and not after synthesis (Garoff et al., 1978). The enzymes that excise the 6K peptide are unknown. It could be that the cleavage between the 6K peptide and the amino terminus of the E l protein is effected by the signal peptidase (see Kreil, 1981).The required specificity, if the bond to be cleaved is X-Y, is that X must be an amino acid with a short side chain. The carboxyl terminus of the 6K peptide is alanine. The p62 protein is cleaved to generate the E3 and the E2 proteins about 30-35 minutes after translation, shortly after the carbohydrate side chains have been processed to their complex form (Jones et al., 1974; Kaluza, 1976; Ziemiecki et al., 1980; Green et al., 1981). The cleavage site is Arg-Arg-Ser in SFV (Garoff et al., 1980b) and LysArg-Ser in Sindbis virus (Rice and Strauss, 1981). In SFV the last arginine is excised leaving the penultimate arginine residue as the carboxyl terminus of E3 (Kalkkinen, 1980). Such a combined action of a trypsinlike endopeptidase with a carboxypeptidase B-like exopeptidase has been found to be involved in the intracellular processing of a large

114

KAI SIMONS AND GRAHAM WARREN

number of prohormones and proproteins into their mature forms at a late stage of secretion (see Docherty and Steiner, 1982).The proteases responsible for the dibasic cleavages have not been characterized but different enzymes seem to be involved because the same hormone precursor can be processed differently by different cells. d. Antigenic Changes. Antibodies have also been used to detect changes in the SFV glycoproteins during intracellular transport. Kaluza et al. (1980)have shown that it is possible to obtain antibodies which recognize immature forms of the p62 and El proteins. This was done by absorbing polyclonal antisera against the SFV glycoproteins. The irnrnature forms of the proteins recognized by these antibodies disappear about 10-15 minutes after translation. Burke et al. (1983)have obtained monoclonal antibodies against the E2 protein that recognize the reverse event-the appearance of a new antigenic determinant in the E2 part of the p62 protein resulting from modifications taking place during intracellular transport. Bonatti and Cancedda (1982)have found that the apparent molecular weight of the El protein of Sindbis virus increases shortly after synthesis. This effect was seen in SDS-polyacrylamide gels, but not in those run in the presence of urea, suggesting a change in protein conformation. The precise nature of all of these modifications is still unknown but it would be useful to identify their trigger since changes in protein conformation might well be important in intracellular transport. e. Other Modifications. The glycoproteins of Sindbis virus are also known to become sulfated (Pinter and Compans, 1975)and phosphorylated (Tan and Sokol, 1974;Waite et al., 1974)during intracellular transport. Where and why these modifications occur are not known.

2. The Intracellular Transport Route Secretory glycoproteins are known to move from the ER to the Golgi complex where their carbohydrate side chains are trimmed and further modified (see Palade, 1975; Tartakoff, 1980; Farquhar and Palade, 1981). In most secretory cells the proteins are then concentrated into condensing vacuoles which store the secretory proteins until they are discharged by exocytosis through a fusion reaction between the vacuolar membrane and the plasma membrane. In other secretory cells, like plasma cells, proteins are continuously secreted and they appear to leave the Golgi complex within vesicles without being concentrated before exocy tosis. The central feature of the Golgi complex is a stack of flattened cisternae which has, in many secretory tissues, a clearly recognizable polarity with the cis side facing the nucleus (see Farquhar and Palade, 1981).

SEMLIKI FOREST VIRUS

115

Secretory proteins seem to enter at the cis side and leave the Golgi stack from the trans side, but the route taken through the stack has not been delineated. For membrane proteins there was not until recently any direct evidence that they took the same route from the ER as secretory protein. However, since they carried carbohydrate side chains similar to those present on secretory proteins, it was generally assumed that they would have to pass through the Golgi complex where the enzymes responsible for carbohydrate processing were known to be located. Autoradiographic evidence at the ultrastructural level (Hall et al., 1969; Fambrough and Devreotes, 1978) and immunofluorescence studies with the light microscope (Kaiiriainenet al., 1980; Saraste et al., 1980b) suggested that this was indeed the case, but the resolution was not high enough to show that the membrane glycoproteins passed through the stack. Direct evidence for the involvement of the stacks of Golgi cisternae in the intracellular pathway to the cell surface for membrane glycoproteins came once again from studies with viral glycoproteins (Bergmann et al., 1981; Green et al., 1981). The most extensive studies have been carried out with SFV. BHK-21 cells infected with SFV have been treated with cycloheximide to stop further synthesis of the viral proteins (Green et al., 1981). The viral glycoproteins can then be localized at different times in thin frozen sections of the cells using antibodies against the spike protein labeled indirectly with ferritin or gold particles. This immunocytochemical method allows precise and quantitative localization of the antigens at the ultrastructural level (Tokuyasu, 1980; Griffiths et al., 1983a).Before the addition of cycloheximide the spike proteins are found throughout the membranes of the ER, in all cisternae of the Golgi stacks, and at the cell surface. After the addition of cycloheximide the spike proteins move from the ER through the Golgi stack to the cell surface. Membrane carrier vesicles between the ER and the Golgi, and between the Golgi and the cell surface, could not be identified with certainty. The spike proteins spent about 15 minutes in the ER after the cycloheximide block, and another 15 minutes in the Golgi stack before being routed to the cell surface. Parallel biochemical studies show that many of the oligosaccharides in the viral spike proteins are modified to the complex forms at the same time that the proteins pass through the Golgi stacks. Cell fractionation studies reveal the same pattern; the proteins pass from the ER to the plasma membrane via a vesicle fraction. This fraction must be derived at least in part from the Golgi complex, because it was isolated according to its content of two Golgi markers, galactosyltransferase, an enzyme involved in the formation of complex oligosaccharide chains, and an antigen ( M , = 135,000) specifically localized in the Golgi complex (Green et al., 1981; Louvard et al., 1982). Further studies have shown

116

KAI SIMONS AND GRAHAM WARREN

that the Golgi stack in BHK-21 cells can be divided into at least three distinct parts, the cis, medial, and trans compartments, each comprising one and at most two Golgi cisternae (Griffiths et al., 1982, 1983b; Quinn et al., 1983).These three compartments have been defined by cytochemical and immunocytochemical criteria (Fig. 3). The pattern of cytochemical labeling is not changed during SFV infection. About 5 hours after infection, rod-shaped structures covered with viral nucleocapsids appear in the infected cells. These have been observed earlier and have been called cytopathic vacuoles I1 (CPV 11) (Acheson and Tamm, 1967; Grimley et al., 1968). In cross section the rods appear as membrane vesicles with nucleocapsids around their outer surface. These membranes are always labeled with ricin and with antibodies to the spike protein. Their function is unclear but they could be aberrant products caused by the massive transport of spike proteins through the cell. These structures are usually found on the trans side of the Golgi stack, and have been used to define the polarity of the stack, which is otherwise difficult to do in BHK-21 cells (Griffiths et al., 1982). When cycloheximide is added to infected cells to stop viral protein synthesis, the labeling with ricin of the trans cisternae decreases by about 50% after the spike proteins have been shown to leave the Golgi stacks. Since ricin labels carbohydrate side chains only after they have been trimmed and galactose has been added, these data suggest that the spike proteins acquire their galactose residues in the trans part of the Golgi in which galactosyltransferasehas recently been found to be localized in other cells (Roth and Berger, 1982). The viral proteins in the cis part of the Golgi are not labeled with ricin; their carbohydrate side chains have thus not yet been processed to the complex type. These results are in keeping with the movement of the spike proteins from the cis to the trans side of the Golgi stack. 3. Inhibition of Intracellular Transport A successful tool in the early studies of metabolic pathways was blocking the pathway at some specific point. This could be done by the use of either mutants or inhibitors. Schekman et al. have isolated a number of yeast mutants with blocks in their secretion pathway (Schekman, 1982). It is not yet known which proteins these mutations affect, but this is clearly a most promising approach for identifying those components involved in transport. In animal cells there are no cellular mutants with blocks in the intracellular transport of protein from the ER to the cell surface. There are, however, genetic diseases which affect the routing of lysosomal enzymes to the lysosomes (Neufeld et al., 1975; Sly and Fischer, 1982). For viruses it has been possible to isolate temperaturesensitive mutants in which a mutation in the viral glycoprotein arrests

SEMLIKI FOREST VIRUS

117

the protein in different cellular locations at the nonpermissive temperature (see Pfefferkorn, 1977; Saraste et al., 1980a; Zilberstein et al., 1980). A number of inhibitors have also been found that inhibit intracellular transport at different intracellular sites (see Tartakoff, 1980). a. Temperature-SensitiveMutants. Burge and Pfefferkorn (1966a,b) were the first to characterize a number of temperature-sensitive mutants of Sindbis virus. A mutant was found in which the viral glycoprotein appears to accumulate in the ER at the nonpermissive temperature, and, after shifting down to the permissive temperature, is transported to the cell surface (Bell and Waite, 1977; Smith and Brown, 1977; Saraste et al., 1980a).KaiPriPinen and co-workers have characterized a number of SFV mutants (Keranen and Kaariainen, 1974; Saraste et al., 1980b; Pesonen et al., 1981). In one mutant, ts-1, the glycoproteins are arrested in the ER at the restrictive temperature. The mutation is reversible; transport of the spike glycoproteins to the cell surface occurs when the temperature is lowered to 28°C. Spike glycoproteins from ts-1 mutants have oligosaccharide side chains exclusively of the high-mannose type at 39"C, but after a shift to 28"C, about 35% of the oligosaccharides are converted to complex glycans in keeping with a transport defect which arrests the proteins in the ER. Saraste et al. (1980b) have postulated the existence of "transport signals" carried by the viral glycoproteins (see also Blobel, 1980). These signals should be recognized by the cellular mechanisms responsible for sorting proteins from the ER to the Golgi complex, and from the Golgi complex to the cell surface. Since the cytoplasmic domain of the SFV spike glycoproteins can be essentially deleted (Garoff et al., 1983) without affecting the transport of the protein to the cell surface, one would assume that the signals for sorting are localized either in the transmembrane domain or in the hydrophilic portion of the spike. However, the basic cluster at the cytoplasmic end (see Section II,B, 1) still remained in these deletion mutants, and it could, of course, be involved in transport. b. Effect of Inhibitors. i. Tunicamycin. In the presence of tunicamycin, the N-glycosylation of nascent proteins is efficiently inhibited. This drug blocks the glycosylation of the p62 and the El proteins (Leavitt et al., 1977; Schwarz et al., 1976; Garoff and Schwarz, 1978). The nonglycosylated viral proteins are not transported to the cell surface. Instead, they appear to aggregate in the ER and cannot be extracted with Triton X-100, in contrast to their glycosylated counterparts. Tunicamycin does not always block intracellular transport in this way. A number of other nonglycosylated membrane and secretory proteins are transported to the cell surface normally (see Gibson et al., 1981). Thus, the lack of glycosylation in the case of the alphavirus glycoproteins changed their

118

KAI SIMONS AND GRAHAM WARREN

solubility in such a way that they aggregated. The carbohydrate side chains seem not to be a necessary requirement for intracellular transport to the cell surface, and cannot therefore be an essential target for the sorting mechanisms responsible for transport of the protein from the ER and from the Golgi apparatus. ii. Uncoupling agents. Transport of secretory proteins both from the ER to the Golgi complex and from there to the cell surface is blocked by inhibitors or uncouplers of oxidative phosphorylation (see Palade, 19’15). Kaariainen et al. (1980) have used the uncoupler carbonyl cyanide-p-trifluoromethoxyphenylhydrazone(FCCP) to show that the intracellular transport of the SFV glycoproteins is also energy dependent. iii. Drugs affecting the cytoshleton. Microfilaments and microtubules do not play a decisive role in cellular secretion (see Tartakoff, 1980). Inhibitors affecting these cytoskeletal components do not block secretion, which continues though often at a slower rate. The same results have been obtained using SFV (Richardson and Vance, 1978a,b; KaariPinen et al., 1980). Cytochalasin B, which disrupts actin-containing microfilaments, does not affect spike glycoprotein transport. Colchicine and vinblastine, which cause the disappearance of microtubules, decrease the rate of transport of the spike glycoproteins to the cell surface by about 50%. iv. Zonophores. Monensin, the Na+ and K+ ionophore, inhibits the intracellular transport of secretory as well as membrane glycoproteins (see Tartakoff, 1980). This is also true for the alphavirus glycoproteins (Johnson and Schlesinger, 1980; Kaariainen et al., 1980; Pesonen and Kaariainen, 1982).Monensin appears to block transport at some point in the Golgi complex but the precise site has been difficult to localize because monensin destroys the characteristic morphology of the Golgi stacks; the flattened cisternae become swollen and separated from each other. Interestingly, in cells infected with SFV or Sindbis virus, some of the swollen Golgi cisternae are found to be covered with nucleocapsids bound to the cytoplasmic face. Apparently, the accumulation of viral spike proteins in these swollen cisternae leads to the binding of numerous nucleocapsids. This observation has made it possible to determine the site at which monensin blocks the transport of SFV glycoproteins in BHK-21 cells. Observations from a variety of cytochemical and immunocytochemical experiments suggest that this site is located between the medial and the trans cisternae in the Golgi stack (Griffiths et al., 1983b; Quinn et al., 1983). In infected BHK-21 cells treated with monensin there was no significant trimming of the high-mannose residues or conversion of the carbohydrate side chains to their complex forms. These functions therefore presumably reside in the trans cisternae, although in

SEMLIKI FOREST VIRUS

119

another cell type it has been claimed that mannose trimming occurs proximal to the trans cisternae (Dunphy et al., 1981). Cleavage of the p62 protein in E3 and E2 is also blocked by the drug (cf. Oda and Ikehara, 1982). However, fatty acid acylation of the El and the p62 proteins continues during monensin treatment, suggesting that this modification might be a cis or a medial Golgi function. Earlier studies have suggested that acylation does not occur in the ER (Schmidt and Schlesinger, 1980). Due to the increased density of the medial cisternae caused by the binding of the dense nucleocapsids, it is possible to separate them from the cis and the trans Golgi cisternae membranes by density gradient centrifugation (Quinn et al., 1983). If the nucleocapsids are then detached using conditions known to disrupt the interactions between spike proteins and nucleocapsid (pH 8,O.l M NaCl; see Section II,B, 1,a), the membranes lose the bound nucleocapsids and regain the density of Golgi membranes from cells not treated with monensin. Although monensin appears to block the transport of the SFV glycoproteins at a fairly specific site in BHK-21 cells, the findings cannot be generalized. Different results have been obtained in other cells. Monensin seems to inhibit transport at a number of points along a number of pathways and its effects differ depending on the cell type (Johnson and Schlesinger, 1980; Smilowitz, 1980; Basu et al., 1981; Tartakoff et al., 1981). As a specific example, we can consider chick embryo fibroblasts infected with SFV (Pesonen and Kaarihinen, 1982). In these cells, monensin does not inhibit p62 cleavage, and some conversion of simple to complex oligosaccharides is found to take place (and some intermediate forms are found that are not present normally), but the appearance of the spike proteins on the cell surface is efficiently blocked. In these cells the monensin block would appear to be distal to the medial cisternae, perhaps even after the Golgi stack. Previously the p62 cleavage has been thought to occur at the cell surface (Bracha and Schlesinger, 1976;Jones et al., 1977; Ziemiecki et al., 1980). The results using monensin show that at least in chicken embryo fibroblasts the cleavage can take place intracellularly. The evidence that the p62 cleavage is a cell surface event is based on the finding that antibodies to the spike glycoproteins applied externally block the cleavage. However, since these antibodies may enter the cell by endocytosis and exert their effect intracellularly, these experiments do not rule out an intracellular cleavage. The p62 cleavage takes place about 5 minutes after the viral spike glycoproteins become resistant to endoacetylglucosaminidase H (Green et al., 1981). This timing would be compatible with the cleavage occurring either in the trans Golgi or on the post-Golgi pathway to the cell surface.

120

KAI SIMONS A N D GRAHAM WARREN

D. Budding 1. Assembly of the Viral Particle at the Cell Surface

The first spike proteins can be detected at the cell surface about 2 hours after infection (Birdwell and Strauss, 1974; KaariBinen et al., 1980). It takes about 1 hour more before mature viral particles are released extracellularly. The virus is released from the cell by a budding outward of the cell membrane. In this process the nucleocapsid binds to the plasma membrane which wraps around. the nucleocapsid and the bud is expelled from the cell (Acheson and Tamm, 1967). The lipids of the viral envelope are derived from the plasma membrane (see Section II,B,2), but practically all of the host proteins are excluded from the bud. At least 99.5% of the protein in the alphavirus particle is viral (capsid, El, E2, and E3) (Strauss, 1978). The central problem in budding at the molecular level is to understand how the nucleocapsid and the viral spike proteins recognize each other at the cell surface (Simons and Garoff, 1980).The realization that the SFV spike glycoproteins span the membrane led to the proposal that the binding of the spike proteins to the nucleocapsid was the major driving force in the budding process (Garoff and Simons, 1974).Another interaction facilitating the assembly may be the formation of lateral contacts between the spike proteins (McCarthy and Harrison, 1977; von Bonsdorff and Harrison, 1978). The alphavirus spike glycoproteins move by lateral diffusion after insertion into the cell surface (Birdwell and Strauss, 1974).Their diffusion coefficients have been measured by fluorescence ,photobleaching recovery experiments (Johnson et al., 198l), and are approximately 5 X 10-locm2 sec-', which is in the range reported for other cell surface glycoproteins (Peters, 1981).However, the longer infection proceeds, the larger the fraction of the spike proteins that become immobile on the time scale of the measurements; 7 hours after infection 14% are mobile and after 10 hours only about 1%. The immobile fraction may be due to the formation of spike protein aggregates in the membrane plane (see however Johnson et al., 1981,for an alternative explanation). There is evidence that budding may occur in patches especially at the cell periphery (Brown et al., 1972;Birdwell et al., 1973), whereas other areas of the cell surface are devoid of budding figures when examined in the electron microscope. However, it should be pointed out that in other cells there seems to be no clustering of budding figures. Budding is probably initiated by the viral nucleocapsid binding to a cluster of spike proteins at the cell surface. The binding must be mediated by the cytoplasmic domain of the E2 protein attaching to a capsid

SEMLIKI FOREST VIRUS

121

protein in the nucleocapsid. Wrapping of the plasma membrane around the nucleocapsid proceeds when more spike proteins move in and become attached to the underlying nucleocapsid. When the nucleocapsid is completely enclosed by the modified plasma membrane, it may pinch off simply from the strain of the curvature imposed by the nucleocapsidspike protein interactions. The resealing of the disrupted lipid bilayer would occur spontaneously. In keeping with this self-assembly model for budding, the process seems to be independent of metabolic energy (Waite and Pfefferkorn, 1970;Waite et al., 1972).If cells infected with Sindbis virus are treated with low salt (ionic strength 0.105)further budding is arrested; the nucleocapsids bind to the plasma membrane but do not bud. Treatment of arrested cells with metabolic inhibitors does not affect the subsequent release of viral particles after restoring the cells to normal ionic conditions. The budding process can also be inhibited by antibodies to the spike proteins (Bracha and Schlesinger, 1976). In thin section electron micrographs of such cells, cross-linked clusters of spike proteins can be seen on the external cell surface apposed to clusters of nucleocapsids on the opposite side of the membrane (Smith and Brown, 1977). The reason that host proteins are excluded from the plasma membrane segment enclosing the nucleocapsid is due probably to their lack of affinity for the capsid protein (Garoff and Simons, 1974).The apposition of the spike proteins on the external side of the bilayer and the close proximity of the nucleocapsid to the cytoplasmic face of the bilayer may effectively prevent host proteins from becoming included in the viral particle. If self-assembly is the mechanism for budding of alphaviruses, why does budding occur mainly at the cell surface, and not intracellularly? The cytoplasmic domain of the spike protein should be available for interaction with the nucleocapsid during intracellular transport. This is due probably to the low concentration of spike protein intracellularly which may not allow the formation of spike protein clusters. Saraste et al. (1980b)studied ts-7,the reversible mutant of SFV. In cells infected with ts-7,the spike proteins accumulate in the Golgi complex at the restrictive temperature. After lowering the temperature to 28"C,the spike proteins presumably resumed their native conformation. Within 10 minutes, binding of nucleocapsids to intracellular membrane was observed, followed by budding into an intracellular membrane compartment, probably the Golgi. By 60 minutes later, budding was seen mostly at the cell surface, and no longer inside the cell. Intracellular budding can be observed also in monensin-treated cells (Johnson and Schlesinger, 1980; Pesonen and Kaariainen, 1982;Griffiths et al., 1983b).In these cells, the

122

KAI SIMONS AND GRAHAM WARREN

spike proteins accumulate in the Golgi stack, and budding can frequently be observed into the swollen Golgi cisternae. These results suggest that the affinity of the nucleocapsid for the spike proteins is fairly low, and a critical concentration of spike protein is required before assembly can proceed. BHK-21 cells infected with SFV are synthesizing about 100,000 spike proteins per minute between 4 and 6 hours after infection. Quantitation in thin, frozen sections shows that the spike proteins spend on average about 15 minutes in both the ER and in the Golgi compartments (Green et al., 1981). Hence at any time there should be about 1,500,000 spike proteins in each of these membrane systems. The surface areas of the ER and the Golgi membranes in infected cells have been determined morphometrically so that the density of spike proteins can be calculated. These amount to about 85 and 800 spike proteins/pm2 for ER and Golgi membranes, respectively. Since a typical concentration of spanning proteins in many biological membranes is about 20,000/pm2, it is clear that at no time during intracellular transport do the spike proteins constitute more than 1 in 250 endogenous ER or 1 in 25 endogenous Golgi proteins. At the cell surface, the spike proteins accumulate and completed viral particles contain about 30,000 spike proteins/pm2. The low concentrations of spike proteins in transit probably ensure that budding does not take place intracellularly. Other factors may also be operative. The cytoplasmic domain of the E2 chain may not be available for interactions with the nucleocapsid at all stages of intracellular transport. Also the cleavage of the p62 protein which takes place shortly before budding may facilitate the molecular interactions involved in the budding process.

2. Analogies between Viral Budding and Protein Sorting during Intracellular Transport The problem we have not yet touched upon is how components can specifically move from one cellular component to another. Both the entry and the exit of SFV spike proteins are dependent on a number of such cellular processes. The newly synthesized spike proteins move from the ER to the Gold complex and then to the cell surface. The cell surface membrane is continuously retrieved by endocytosis into endosomes. From here the endocytosed membrane components probably recycle back to the cell surface, but some components may also be channeled into lysosomes for degradation. Especially in cells with secretory activity, the recycling pathway from the cell surface also includes the Golgi complex (see Farquhar and Palade, 1981). With the exception of coated vesicles, which endocytose surface membrane into endosomes, little is known about the membrane vesicles medi-

SEMLIKI FOREST VIRUS

123

ating traffic between the different cellular components (see Farquhar and Palade, 1981). There also are no data on how proteins move through the Golgi stack. Coated vesicles have also been implicated as membrane carriers of viral glycoproteins from the ER to the Golgi stack and from the Golgi to the plasma membrane (Rothman and Fine, 1980). However, because at any time a very small fraction of the protein undergoing intracellular transport is present in the carriers shuttling components from one organelle to another, identification of the carriers has proven to be exceedingly difficult. It is clear that the membrane carriers cannot remove membrane components randomly from one compartment and move them to another. There has to be some sorting device involved which leaves the proteins belonging to the organelle behind. When a membrane vesicle is formed from one organelle and transported to fuse with another, there must also be replenishment of membrane lipids lost. This cannot in most cases be due to the synthesis of new lipids, but is probably due to a compensating backward traffic of membrane. The membrane vesicles mediating the traffic in both directions must recognize their target membranes, and carry a fusion mechanism for delivery of membrane from one organelle to another. The difficulty in studying sorting at the molecular level is the present lack of assays. In vitro systems need to be worked out, and there are already some promising beginnings in this direction (Fries and Rothman, 1980; Rothman and Fries, 1981). Since molecular studies of cellular sorting processes still seem some way off, it might be useful to consider whether there are any other analogous processes more amenable to experimental study. In principle, SFV budding is such a sorting process. The nucleocapsid is the sorter; the affinity between the capsid protein and the spike protein enables the capsid to sort the viral glycoprotein into a membrane vesicle, the viral particle. The topology of the viral vesicle is reversed compared to intracellular membrane carrier vesicles; the viral vesicle has the cytoplasmic side of the membrane toward the inside, whereas the carrier vesicle has it on the outside. In the viral particle the nucleocapsid functions as a scaffold from the inside, whereas in coated vesicles the clathrin coat forms a polyhedral basket on the outside (Pearse and Bretscher, 1981). In viral budding the specificity of the sorting process is tight. Essentially no host proteins are included and only if a cell is infected with two different alphaviruses, e.g., Sindbis virus and eastern equine encephalitis virus, does the viral particle contain spike glycoproteins from both viruses (Burge and Pfefferkorn, 1966~). The budding process of more complicated enveloped viruses such as

124

KAI SIMONS AND GRAHAM WARREN

VSV bears even more resemblance to intracellular sorting. In this virus there is a layer of protein composed of the M protein. This layer lies between the nucleocapsid and the lipid bilayer (which contains one species of a spanning glycoprotein, the G protein) (see Simons and Garoff, 1980). The budding of VSV occurs at the cell surface and is most likely driven by an interaction between the G protein and the internal M protein, although an interaction between the G protein and the nucleocapsid proteins cannot be excluded at present. However, unlike alphaviruses, there is no precise stoichiometry between the G and the M proteins (Lodish and Porter, 1980a). The ratio between them can vary by a factor of about six whereas the ratio between the M protein and nucleocapsid proteins stays constant. In addition, the specificity is not as strict as with the alphaviruses. Mixed phenotypes, that is viruses produced from doubly infected cells and containing the spike proteins of two viruses but only one nucleocapsid, can easily be produced (Zavada, 1982). These contain the VSV nucleocapsid, the VSV M protein, and VSV G proteins mixed with the spike glycoproteins of the other virus used for double infection. Viral particles containing the nucleocapsid and M proteins from the other virus, with a mixture of spike proteins, may also be formed. VSV can form mixed phenotypes with alphaviruses (but not the reverse), RNA tumor viruses, influenza viruses, and parainfluenza viruses. Cellular glycoproteins are excluded from the viral envelope (see, however, Lodish and Porter, 1980b).The inclusion of the foreign viral spike glycoproteins could be due to an interaction either directly with the VSV M protein or with a critical amount of VSV G protein through which association with the M protein could occur (see Witte and Baltimore, 1977). The VSV M protein plays an essential role in the budding process and it can be considered analogous to the capsid protein in SFV, but very little is known of how it functions. A closer study of the VSV M protein in viral spike glycoprotein interactions might be of considerable interest now that methods based on cDNA technology are available which could be used for this purpose. IV. PERSPECTIVES Studies of the alphavirus life cycle have revealed how heavily the virus relies on cellular processes for replication. The paucity of functions that seem unique to the virus is striking. The binding of the virus to the cell surface and the fusion of its membrane intracellularly depend on the viral spike glycoproteins. RN A-dependent RNA polymerases specific for the virus catalyze the replication of the viral RNA. Exit from the cell requires the interaction of the viral spike proteins with the viral capsid

SEMLIKI FOREST VIRUS

125

protein. Close scrutiny of some of these processes may disclose that they are not unique after all. They may, in fact, have evolved from cellular counterparts. The heavy reliance on the normal function of the host cell is the key to the use of viruses as tools to study the molecular mechanisms of the animal cell. In cell biology the use of viruses as tools has only started. A number of new approaches are emerging. Polarized epithelial cells infected with enveloped viruses distribute the spike glycoproteins to either the apical or the basolateral membrane domain of the cell surface (Rodriguez et al., 1980). There are other enveloped viruses that bud intracellularly ; bunyaviruses probably assemble in the Golgi apparatus (Bishop and Shope, 1979; Madoff and Lenard, 1982; Pesonen et al., 1982), whereas coronaviruses mainly bud into the ER (McIntosh, 1974). Their glycoproteins can probably be used as models for the assembly of these organelles. There are viruses like rabies and herpes which enter neurons at peripheral nerve endings and move into the central nervous system probably by retrograde axonal transport (Wolinsky and Johnson, 1980). These examples are by no means exhaustive. There is a plethora of different viruses each of which has adapted to a combination of cellular functions for its own selfish purposes. ACKNOWLEDGMENTS We wish to thank Henrik Garoff, Ari Helenius, Carl-Henrik von Bonsdorff, and Marja Pesonen for a critical reading of the manuscript. One of us (K. S.) would like to express a special word of thanks to Hilkka Virta for unfailing and devoted technical support through all stages of the Semliki Forest virus studies both in Helsinki and in Heidelberg.

REFERENcEs Acheson, N. H.,and Tamm, I. (1967). Virology 32, 128-143. Aliperti, G.,and Schlesinger, M. J. (1978).Virology 90, 366-369. Anderson, R. G.W., and Kaplan, J. (1983).In “Modern Cell Biology” (B. Satir, ed.), Vol. 1, pp. 1-52. Plenum, New York. Basu, S. K., Goldstein, J. L., Anderson, R. G. W., and Brown, M. S. (1981).Cell 24,493502. Bell, J. R., and Strauss, J. H. (1981).J.Bwl. Chem. 256, 8006-8011. Bell, J. R., Hunkapiller, M. W., Hood, L.E., and Strauss, J. H. (1978).Proc. Natl. Acad. Sci. U.S.A. 75, 2722-2726. Bell, J. R., Rice, C. M., Hunkapiller, M. W., and Strauss, J. H. (1982).Virology 119, 255267. Bell, J. W., Jr., and Waite, M. R. F. (1977).Virology 21, 788-791. Bergmann, J. E.,Tokuyasu, K. T., and Singer, S. J. (1981):Proc. Natl. A c d . Sci. U.S.A. 78, 1746- 1750.

126

KAI SlMONS AND GRAHAM WARREN

Birdwell, C. R., and Strauss, J. H. (1974).J. Virol. 14,366-374. Birdwell, C. R., Strauss, E. G., and Strauss, J. H. (1973).Virology 56, 429-438. Bishop, D. H. L., and Shope, R. E. (1979).In “Comprehensive Virology” (H. FraenkelConrat and R. R. Wagner, eds.), Vol. 14,pp. 1-186. Plenum, New York. Blobel, G. (1980).Proc. Natl. Acad. Sci. 0.S.A. 77, 1496-1500. Blobel, G.,and Dobberstein, B. (1975a).J.Cell Biol. 67, 835-851. Blobel, G., and Dobberstein, B. (1975b).J.Cell Biol. 67, 852-862. Blok, J.. Air, G. M.,Laver, W. G., Ward, C. W., Lilley, G. G., Woods, E. F., Roxburgh, C. M., and lnglis, A. S.(1982).Virology 119, 109-121. Boege, U., Wengler, G., and Wittmann-Liebold, B. (1981).Virology 113, 293-303. Bonatti, S.,and Blobel, G. (1979).J.Biol. Chem. 254, 12261-12264. Bonatti, S., and Cancedda, F. D. (1982).J . Virol. 42,64-70. Bonatti, S., Cancedda, R., and Blobel, G. (1979).J. Cell Bwl. 80, 219-224. Bordier, C. (1981).J.Biol. Chem. 256, 1604-1607. Bracha, M.,and Schlesinger, M. J. (1976).Virology 74,441-449. Braell, W. A., and Lodish, H. F. (1982).J.Biol. Chem. 257,4578-4582. Brand, C. M., and Skehel, J. J. (1972).Nature (London),New Biol. 238, 145-147. Brown, D. T., Waite, M. R. F., and Pfefferkorn, E. R. (1972).J.Virol. 29, 546-554. Brown, F., Smale, C. J., and Horzinek, M. C. (1974).J.Gen. Virol. 22,455-458. Burge, B. W., and Pfefferkorn, E. R. (1966a).Virology SO, 204-213. Burge, B. W., and Pfefferkorn, E. R. (1966b).Virology 30, 214-223. Burge, B. W., and Pfefferkorn, E. R. (1966~). Nature (London) 210, 1397-1399. Burke, B., Walter, C.. Griffiths, G., and Warren, G. (1983).Eur.J. Cell Bwl. 31,315-324. Burke, D., and Keegstra, K. (1976).J. Virol. 20, 676-686. Burke, D., and Keegstra, K. (1979).J.Virol. 29,546-554. Cancedda, R., and Schlesinger, M. J. (1974).Proc. Natl. Acad. Sci. U.S.A. 71, 1843-1847. Cancedda, R., Villa-Komaroff, L., Lodish, H. F., and Schlesinger, M. J. (1975).Cell 6,215222. Chanas. A. C., Could, E. A., Clegg, J. C. S.,and Varma, M. G. R. (1982).J.Gen. Virol. 58, 37-46. Chapman, D., Gomez-Fernandez, J. C., and Goni, F. M. (1979).FEBS Lett. 98, 211223. Clegg, J. C. S.(1975).Nature (London) 254,454-455. Clegg, J. C. S., and Kennedy, S. I. T. (1974).FEBS Lett. 42, 327-330. Clegg, J. C. S.,and Kennedy, S. I. T. (1975a).Eur.J. Biochem. 53, 175-183. Clegg, J. C. S.,and Kennedy, S. I. T. (1975b).J.Mol. Biol. 97, 401-41 1. Clegg, J C. S.,Brzeski, H., and Kennedy, S. I. T. (1976).J.Gen. Virol. 32, 413-430. Compans, R. W. (1971).Nature (London) New Biol. 229, 114-116. De Duve, C., De Barsy, T., Poole, B., Tronet, A., Tulbens, P., and Van Hoof, F. (1974). Bio~hem.Phanno~ol.23,2495-2531. Desnuelle, P. (1979).Eur. J . Biochem. 101, 1-1 1. Dintzis, H. M. (1961).Proc. Natl. Acad. Sci. U.S.A. 47, 247-251. Docherty, K., and Steiner, D. F. (1982).Annu. Rev. Physiol. 44,625-638. Dunn, W. A., Hubbard, A. L., and Aronson, N. N. (1980).J. Biol. Chem. 255, 59715978. Dunphy, W.G., Fries, E., Urbani, L. J., and Rothman, J. E. (1981).Proc. Natl. Acad. Sci. U.S.A. 70, 7453-7457. Fambrough, D. M., and Devreotes, P. N. (1978).J.Cell Biol. 76, 237-294. Farquhar, M.G.,and Palade, G. E. (1981).J.Cell Biol. 91, 77s-103s. Fields, S., Winter, G., and Brownlee, G. G. (1981).Nature (London) 290, 213-217. Fiers, W., Contreras, R., Haegeman, G., Rogiers, R., Van de Woorde, A., Van

SEMLIKI FOREST VIRUS

127

Heuverswyn, H., Van Hereweghe, J., Volckaert, G., and Ysebart, M. (1978). Nature (London) 273, 113-120. Freeman, R., and Leonard, K. (1981).J. Microsc. (Oxford) 122, 275-286. Friedman, R. M., Levin,J. G., Grimley, P. M., and Berezesky, I. K. (1972).J. Virol. 10,504515. Fries, E., and Helenius, A. (1979). Eur.J. Biochem. 97, 213-220. Fries, E., and Rothman, J. E. (1980). Proc. Natl. Acad. Sci. U.S.A. 77, 3870-3874. Gahmberg, C. G., Utermann, G., and Simons, K. (1972). FEES Lett. 28, 179-182. Garoff, H., and Schwarz, R. T. (1978). Nature (London) 274,487-490. Garoff, H., and Simons, K. (1974). Proc. Natl. Acad. Sci. U.S.A. 71, 3988-3992. Garoff, H., and Stklerlund, H. (1978).J. Mol. Bwl. 124, 535-549. Garoff, H., Simons, K., and Renkonen, 0. (1974). Virology 61,493-504. Garoff, H., Simons, K., and Dobberstein, B. (1978).j.Mol. B i d . 124, 587-600. Garoff, H., Frischauf, A.-M., Simons, K., Lehrach, H., and Delius, H. (1980a).Proc. Natl. A d . Sci. U.S.A. 77, 6376-6380. Garoff, H., Frischauf, A.-M., Simons, K., Lehrach, H., and Delius, H. (1980b). Nature (London) 288,236-241. Garoff, H., Kondor-Koch, C., and Riedel, H. (1982a).Curr. Top. Mirrobiol. Immunol. 99, 150. Garoff, H., Riedel, H., and Lehrach, H. (1982b). Nuclek Acids Res. 10,675-687. Garoff, H., Kondor-Koch, C., Pettersson, R., and Burke, B. (1983).J. Cell. Biol. 97,652658. Gibson, R., Kornfeld, S.,and Schlesinger, S. (1981). Trends Biochem. Sci. Pen. Ed. 5,290293. Glanville, N., and Ulmanen, I. (1976). Biochem. Biophys. Res. Comrnun. 71, 393-399. Glanville, N., Ranki, M., Uomala, P., Kiiariiiinen, L., and Smith, A. E. (1976a).Proc. Natl. A d . Sci. U.S.A. 73, 3059-3063. Glanville, N., Morser, J., Uomala, P., and Kaiiriainen, L. (1976b).Eur.J. Biochem. 64, 167175. Goldstein, J. L., Anderson, R. G. W., and Brown, M. S. (1979). Nature (London) 279,679685. Green, J., Griffiths, G., Louvard, D., Quinn, P., and Warren, G. (1981).J.Mol. Biol. 152, 663-698. Griffiths, G., Brands, R., Burke, B., Louvard, D., and Warren, G. (1982).J. Cell Biol. 95, 78 1-792. Griffiths, G., Simons, K., Warren, G., and Tokuyasu, K. T. (1983a). In “Methods in Enzymology” (Sidney Fleischer and Becca Fleischer, eds.), Vol. 96, pp. 466-485. Academic Press, New York. Griffiths, G., Quinn, P., and Warren, G. (1983b).J. Cell Biol. 96, 835-850. Grimley, P. M.. Berezesky, I. K., and Friedman, R. M. (1968).J. Virol. 2, 1326-1338. Grimley, P. M., Levin, J. G., Berezesky, I. K., and Friedman, R. M. (1972).J. Virol. 10,492503. Hakimi, J., and Atkinson, P. H. (1982). Biochemkty 21, 2140-2145. Hall, M. D., Bok, D., and Bacharach, A. D. E. (1969).J. Mol. Biol. 45,397-406. Harpaz, N., and Schachter, H. (1980a).J . Biol. Chm . 255,4885-4893. Harpaz, N., Schachter, H. (1980b).J.Bwl. C h . 255,4894-4902. Harrison, S . C.. David, A., Jumblatt, J., and Darnell, J. E. (1971).J. Mol. Biol. 60,523-528. Hashimoto, K., Erdei, S.,Keriinen, S., Saraste, J., and Kiiiiriiiinen, L. (198l).J. Virol. 38, 34-40. Hauri, H.-P., Wacker, H., Rickli, E. E., Bigler-Meier, B., Quaroni, A., and Semenza, G. (1982).J. Biol. Chem. 257, 4522-4528.

128

KAI SIMONS AND GRAHAM WARREN

Helenius, A., and Kartenbeck, J. (1980).Eur. J.Biochem. 106,613-618. Helenius, A., and Simons, K. (1975).Biochim. Biophy. Acta 415, 29-79. Helenius, A.,and SBderlund, H. (1973).Biochim. Biophy. Acta 307, 287-300. Helenius, A., Fries, E., Garoff, H., and Simons, K. (1976).Biochim. Biophp. Acta 436,319334. Helenius, A., Morein, B., Fries, E., Robinson, P., Schirrmacher, U., Terhorst, C., and Strominger, J. L. (1978).Proc. Natl. Acad. Sci. U.S.A. 75,3846-3850. Helenius, A., Kartenbeck, J., Simons, K., and Fries, E. (1980).J. Cell Biol. 84, 404-420. Helenius, A., Marsh, M., and White, J. (1982).J.Gen. Virol. 58, 47-61. Hirschberg, C. B., and Robbins, P. W. (1974).Virology 61,602-608. Hosaka, Y., and Shimizu, K. (1977).In “Virus Infection and the Cell Surface” (G. Poste and G. L. Nicholson, eds.), Vol. 2,pp. 129-156. Elsevier, Amsterdam. Huang, R. T. C., Rott, R., and Klenk, H.-D. (1981).Virology 110, 243-247. Hubbard, S . C., and Ivatt, R. J. (1981).Annu. Rev. B i o c b . 50, 555-584. Hubbard, S. C., and Robbins, P. W. (1979).J.Biol. Chem. 254,4568-4576. Israelachvili, J. N.,and Mitchell, D. J. (1975).Biochim. Biophys. Acta 389, 13-19. Jacrot, B.,Cruiell, M., and Stklerlund, H. (1983).In preparation. Johnson, D. C., and Schlesinger, M. J. (1980).Virology lOS,407-424. Johnson, D. C., Schlesinger, M. J., and Elson, E. L. (1981).Cell 23,423-431. Jones, K. J., Waite, M. R. F., and Bose, H. R. (1974).J.Virol. 13, 809-817. Jones, K. J., Scupham, R. K., Pfeil, J. A., Wan, K., Sagik, B. P., and Bose, H. R. (1977).J. Virol. 21, 778-787. JBrnvall. H. (1975).J.Theor. Biol. 55, 1-12. KPPriPinen, L. and Pesonen, M. (1982).In “The Glycoconjugates” (M. I. Horowitz and W. Pigman, eds.), Vol. 4,pp. 191-242. Academic Press, New York. KPPriPinen, L., and Renkonen, 0. (1977).In “Cell Surface Reviews” (G. Poste and G. Nicolson, eds.), Vol. 4,pp. 791-801. Elsevier, Amsterdam. KPPriBinen, L., and SBderlund, H. (1971).Virology 43,291-299. KUriPinen, L., and SBderlund, H. (1978).Curr. Top. Microbiol. Immunol. 82, 15-69. WiriPinen, L.,Hashimoto, K., Saraste, J., Virtanen, I., and Penttinen, K. (1980).J. Cell BWl. 87,783-791. Kalkkinen, N. (1980).FEES Lett. 115, 163-166. Kalkkinen, N., Jornvall, H., Saderlund, H., and KaPriPinen, L. (1980).E w . J . Biochem. 108, 31-37. Kaluza, G. (1976).J.Virol. 19, 1-12. Kaluza, C.,and Pauli, G. (1980).Virology 102, 300-309. Kaluza, G.,Rott, R.,and Schwarz, R. T. (1980).Virology 102,286-299. Kaplan, J. (1981).Science 212, 14-20. Katz, F. N., Rothman, J. E., Lingappa, V. R., Blobel, G., and Lodish, H. F. (1977).Proc. Natl. Acad. Sci. U.S.A. 74,3278-3282. Keegstra, K., and Burke, D.(1977).J.Supramol. ~Lrucl.7,371-379. Keegstra, K., Sefton, B., and Burke, D. (1975).J.Virol. 16,613-620. Kennedy, S. I. T. (1976).J.Mol. Biol. 108,491-511. Kennedy, S. I. T. (1980).I n “Togaviruses” (R. W. Schlesinger, ed.), pp. 351-396. Academic Press, New York. Keranen, S., and KPPriPinen, L. (1974).Acta Pathol. Microbiol. Scand. Sect. B 82, 810-820. Kitamura, N., Semler, B. L., Rothberg, P. G., Larsen, C . R., Adler, C. J,, Dorner, A. J., Emini, E. A., Hanecak, R., Lee, J. J., van der Werf, S . , Anderson, C. W., and Wimmer, E. (1981).Nature (London) 291,547-553. Kondor-Koch, C., Burke, B., and Garoff, H. (1983).J.Cell. Biol. 97, 644-651.

SEMLIKI FOREST VIRUS

129

Kornfeld, S., Li, E., and Tabas, I. (1978).J.Eiol. Chem. 253, 7771-7778. Kozak, M. (1981).Curr. Top. Microbiol. Immunol. 93,81-123. Kreil, G . (1981).Annu. Rev. B i o c h . 51, 317-348. Lachmi, B., and KPPriainen, L. (1976).Proc. Natl. Acad. Sci. U.S.A. 73, 1936-1940. Lachmi, B., Glanville, N., Kerinen, S., and Kalriainen, L. (1975).J.Virol. 16, 1615-1629. Laine, R., Hettunen, M.-L., Gahmberg, C. G., KaPriPinen, L., and Renkonen, 0. (1972).J. Virol. 10, 433-438. Laine, R., Sderlund, H., and Renkonen, 0. (1973).Intervirology 1, 110-118. Leavitt, R., Schlesinger, S., and Kornfeld, S. (1977).J.Eiol. C k m . 252, 9018-9023. Lehtovaara, P., Ulmanen, I., KPPriainen, L., Keranen, S., and Philipson, L. (1980).I3ur.J. E i o c h . 112,461-468. Lenard, J. (1978).Annu. Rev. Eiophys. Eweng. 7 , 139-165. Lingappa, V. R., Katz, F. N., Lodish, H. F., and Blobel, G . (1978).J.Eiol. C h . 253,8667-

8670. Lodish, H. F., and Porter, M. (1980a).J. Virol. 33, 52-58. Lodish, H. F., and Porter, M. (1980b).Cell 19, 161-169. Lonberg-Holm, K., and Philipson, L. (1974).Monogr. Virol. 9, 1-149. Louvard, D., Reggio, H., and Warren, G. (1982).J. Cell Eiol. 92,92-107. Luukkonen, A., Kaariiinen, L.,and Renkonen, 0. (1976).Eiochim. Biophys. Acta 450, 109-

120. McCarthy, M., and Harrison, S. C. (1977).J.Virol. 23, 61-73. McIntosh, K. (1974).Curr. Top. Microbwl. Immunol. 63,85-129. Madoff, D. H., and Lenard, J. (1982).Cell 28, 821-829. Maeda, T., and Ohnishi, A. (1980).FEES Lett. 122, 283-287. Magee, A. I., and Schlesinger, M. J. (1982).Eiochim. Eiophys. A d a 694,279-289. Marsh, M., and Helenius, A. (1980).J . Mol. Eiol. 142,439-454. Marsh, M., Bolzan, E., and Helenius, A. (1983).Cell 32, 931-940. Matlin, K., Reggio, H., Helenius, A., and Simons, K. (1981).J. Cell Eiol. 91,601-613. Matlin, K., Reggio, H., Helenius, A., and Simons, K. (1982).J. Mol. Eiol. 156,609-631. Mattila, K., and Renkonen, 0. (1978).Virology 91,508-510. Mattila, N., Lunkkonen, A,, and Renkonen, 0.(1976).Biochim. Biophys. Acta 419,435-444. Meek, R. C., Walsh, K. A., and Palmiter, R. D. (1982).J. Biol. Chem. 257, 12245-12251. Meyer, D. I., Krause, E., and Dobberstein, B. (1982).Nature (London) 297, 647-650. Milstein, C., Brownlee, G. G., Harrison, T. M., and Mathews, M. B. (1972).Nature (London), N ~ EWl. u 239, 117-120. Murphy, F. A. (1980).I n “Togaviruses” (Schlesinger, R. W., ed.), pp. 241-316. Academic Press, New York. Mussgay, M., Enzmann, P.-J., Horzinek, M. C., and Weiland, E. (1975).Prog. Med. Virol.

19,257-323. Neuberger, A,, Gottschalk, A., Marshall, R. D., and Spiro, R. G. (1972).I n “The Glycoproteins: Their Composition, Structure and Function” (A. Gottschalk, ed.), pp. 450490. Elsevier, Amsterdam. Neufeld, E. F., Lim, T. W., and Shapiro, L. J. (1975).Annu. Rev. Eiochem. 44,357-376. Oda, K., and Ikehara, Y. (1982).Eiochem. Eiophys. Res. Commun. 105,766-772. Oldstone, M. B. A., Tishon, A., Dutko, F. J., Kennedy, S. I. T., Holland, J. J., and Lampert, P. W. (1980).J. Virol. 34,256-265. Osterrieth, P. M. (1965).LifeSci. 4, 1227-1231. Palade, G. (1975).Science 189,347-358. Pascher, I., and Sundell, S. (1977).C h . Phys. Li$idr 90, 175-191. Pastan, I. H., and Willingham, M. C. (1981).Science 214, 504-509.

130

KAl SIMONS AND GRAHAM WARREN

Patzer, E. J., Wagner, R. R., and Duboi, E. J. (1979).CRC Crit. Rev. Biochem. 6, 165-2 17. Pearse, B. M. F., and Bretscher, M. S. (1981).Annu. Rev.Biochem. 50, 85-101. Pesonen, M., and KBPriainen, L. (1982).J. Mol. Biol. 158,213-230. Pesonen, M., and Renkonen, 0.(1976).Biochim. Biophys. A c h 455,510-525. Pesonen, M., Haaktela, K., and Renkonen, 0. (1979).Biochim. Biophys. Acta 588, 102-1 12. Pesonen, M., Saraste, J., Hashimoto, K., and KilPriainen, L. (1981).Virology 109, 165-173. Pesonen, M., Kuismanen. E., and Pettersson, R. F. (1982).J. Virol. 41, 390-400. Peters, R. (1981).Cell Biol. Int. Rep. 5, 733-760. Pfefferkorn, E. R. (1977).In “Comprehensive Virology” (H. Fraenkel-Conrat and R. R. Wagner, eds.), Vol. 9,pp. 209-238. Plenum, New York. Pfefferkorn, E. R., and Hunter, H. S. (1963).Virology 20,446-456. Pinter, A., and Compans, R. W. (1975).J.Virol. 16, 859-866. Poole, B., and Ohkuma, S . (1981).J.Cell B i d . 90, 665-669. Porter, A. G., Barber, C., Carey, N. H., Hallewell, R. A., Threlfall, G., and Emtage, J. S. (1979).Nature (London) 282,471-477. Quigley, J. P., Rifbin, D. B., and Reich, E. (1971).Virology 46, 106-1 16. Quinn, P., Griffiths, G,, and Warren, G. (1983).J.CeU Biol. 96, 851-856. Ranki, M., and Kgbiiinen, L. (1979).Virology 98,298-307. Rasilo, M.-L., and Renkonen, 0. (1979).Biochim. Biophys. A c h 582,307-321. Renkonen, O.,KMriPinen, L., Simons, K., and Gahmberg, C. G. (1971).Virology 46,318326. Rice, C. M., and Strauss, J. H. (1981).Proc. Natl. Acad. Sci. U.S.A. 78,2062-2066. Rice, C. M., and Strauss, J. H. (1982).J.Mol. Biol. 154, 325-348. Rice, C. M., Bell, J. R., Hunkapiller, M. W., Strauss, E. G.,and Strauss, J. H. (1982).J.Mol. Biol. 154,355-378. Richardson, C. D., and Vance, D. E. (1978a).J.Virol. 28, 193-198. Richardson, C. D., and Vance, D. E. (1978b).J.Biol. Chem. 253,4584-4589. Richardson, C. D.,Scheid, A., and Choppin, P. W. (1980).Virology 105,205-222. Riedel, H., Lehrach, H., and Garoff, H. (1982).J.Virol. 42,725-729. Robbins, R. W., Hubbard, S. C., Turco, S. J., and Wirth, D. F. (1977).Cell 12,893-900. Rodriguez, A., Boulan, E. J., and Pendergast, M. (1980).Cell 20,45-54. Roth, J., and Berger, E. G. (1982).J.Cell Biol. 99,223-229. Rothman, J. E. (1981).Science 213, 1212-1219. Rothman, J. E., and Fine, R. E. (1980).Proc. Natl. Acad. Sci. U.S.A. 77, 780-784. Rothman, J. E.,and Fries, E. (1981).J.Cell Biol. 89, 162. Rothman, J. E., and Lodish, H. F. (1977).Nature (London) 269, 775-780. Sabatini, D. D., Kreibach, G., Morimoto, T., and Adesnik, M. (1982).J.CellBzol. 92, 1-22. Saraste, J., von Bonsdorff, C.-H., Hashimoto, K., Kerhen, S., and Kaariainen, L. (1980a). Cell Biol. Int. Rep. 4, 279-286. Saraste, J., von Bonsdorff, C.-H., Hashimoto, K., Kiiriainen, L., and Keranen, S. (1980b). Virology 100,229-245. Sawicki, D. L., and Sawicki, S. G. (198O).J.Virol. 34, 108-118. Scheele, C. M., and Pfefferkorn, E. R. (1969).J. Virol. 3,369-375. Schekman, R. (1982).T r d Biochem. Sci. Pers. Ed. 7,243-246. Schlesinger, M. J., and KBBriilinen, L. (1980).In “Togaviruses” (R. W. Schlesinger, ed.), pp. 371-392. Academic Press, New York. Schlesinger, M. J., and Schlesinger, S. (1973).J.Virol. 11, 1013-1016. Schlesinger, R. W.(ed.) (1980).“Togaviruses.” Academic Press, New York. Schlesinger, S., and Schlesinger, M. J. (1972).J. Virol. 10,925-932. Schmidt, M. F. G . (1982).Virology 116,327-338.

SEMLIKI FOREST VIRUS

131

Schmidt, M. F. G., and Schlesinger, M. J. (198O).J.Biol. Chem. 255, 3334-3339. Schmidt, M. F. G., Bracha, M., and Schlesinger, M. J. (1979).Proc. Natl. Acad. Sci. U.S.A. 76, 1687-1691. Schwarz, R. T., Rohrschneider, J. M.,and Schmidt, M. F. G. (1976).J.Virol. 19,782-791. Seelig, J., and Seelig, A. (1980).Q. Rev. Biophys. 19, 19-62. Sefton, B. M. (1977).Cell 10,659-668. Sefton, B. M., and Gaffney, B. J. (1974).J.Mol. B i d . 90,343-358. Sefton, B. M., and Keegstra, K. (1974).J.Virol. 14,522-530. Segrest, J. P., and Feldmann, R. J. (1974).J. Mol. Biol. 87,853-858. Shinnik, T. M., Lerner, R. A., and Sutcliffe, J. G. (1981).Nature (London) 293, 543-548. Silverstein, S.C., Steinmann, R. M.. and Cohn, Z. A. (1977).Annu. Rev. B i o c h . 46,669722. Simmons, D. T., and Strauss, J. H. (1974).J. Mol. Biol. 86,397-409. Simons, K., and Garoff, H. (1980).J . Gen. Virol. 50, 1-21. Simons, K., Helenius, A., and Garoff, H. (1973a).J.Mol. Biol. 80, 119-133. Simons, K., Keriinen, S., and KPiriiinen, L. (1973b).FEES Lelt. 29, 87-91. Simons, K., Garoff, H., and Helenius, A. (1977).In “Membrane Proteins and Their Interactions with Lipids” (R. Capaldi, ed.), Vol. I , pp. 207-234. Dekker, New York. Simons, K., Caroff, H., Helenius, A., and Ziemiecki, A. (1978).In “Frontiers of Physicochemical Biology” (B. Pullman, ed.), pp. 837-907. Academic Press, New York. Simons, K., Garoff, H., and Helenius, A. (1980).In “Togaviruses” (R. W. Schlesinger, ed.), pp. 317-341. Academic Press, New York. Skehel, J. J., and Waterfield, M. D. (1975).Proc. Natl. Acad. Sci. U.S.A. 72,93-97. Skehel, J. J., Bayley, P. M.,Brown, E. B., Martin, S. R., Waterfield, M. D., White, J. M., Wilson, I. A., and Wiley, D. C. (1982).Proc. Natl. Acad. Sci. U.S.A. 79,968-972. Sly, W. S.,and Fischer, H. D. (1982).J.Cell Biochen. 18,67-85. Smilowitz, H. (1980).Mol. Pharmacol. 16,202-214. Smith, J. F., and Brown, D. T. (1977).J. Virol. 22, 662-678. M e r l u n d , H. (1973).Intervirology 1, 354-361. S6derlund, H., and Ulmanen, I. (1977).J. Virol. 24, 907-909. StMerlund, H., miriainen, L., von Bonsdorff, C.-H., and Weckstram, P. (1972).Virology 47,753-760. W e r l u n d , H., KiiPriPinen, L., and von Bonsdorff, C.-H. (1975).Med. Biol. 53,412-417. M e r l u n d , H., von Bonsdorff, C.-H., and Ulrnanen, I. (1979).J. Gen. Virol. 45, 15-26. Staneloni, R. J., and Leloir, L. F. (1982).CRC Crit. Rev. Biochem. 12,289-326. Stollar, V., Stollar, B. D., Koo, R., Harrap, K. A., and Schlesinger, R. W. (1976).Virology 69, 104-1 15. Strauss, E. G. (1978).J.Virol. 28,466-474. Strauss, J. H., and Strauss, E. G. (1977).In “The Molecular Biology of Animal Viruses” (D. P. Nayak, ed.), pp. 1 1 1-166. Dekker, New York. Talbot, P. J., and Vance, D. E. (1980).Can.]. B i o c h . 58, 1131-1137. Talbot, P. J., and Vance, D. E. (1982).Virology 118,451-455. Tan, K. B., and Sokol, F. (1974).J.Virol. 13, 1245-1253. Tarentino, A. L.,and Maley, F. (1974).J.B i d . Chem. 249, 811-817. Tarentino, A. L.,Trimble, R. B., and Maley, F. (1978).In “Methods in Enzymology” (V.Ginsburg, ed.), Vol. 50,pp. 574-580. Academic Press, New York. Tartakoff, A. M.(1980).In:. Rev. Exp. Pathol. 22, 228-251. Tartakoff, A. M., Hoessli, D., and Vassali, P. (1981).J.Mol. Biol. 150,525-531. Tkasz, J. S., and Lampen, J. 0. (1975).B i o c h . Biophys. Res. Commun. 65,248-257. Tokuyasu, K. T.(1980).Hirtochem.]. 12,381-403.

132

KAI SIMONS AND GRAHAM WARREN

Tomita, M., and Marchesi, V. T. (1975).Proc. Natl. Acad. Sci. U.S.A. 72,2964-2968. Toneguzzo, F., and Ghosh, H. P. (1978).Proc. Natl. Acad. Sci. U.S.A. 75, 715-719. Tycho, B., and Maxfield, F. R. (1982).Cell 28,643-651. Ulmanen, I., W e r l u n d , H., and KMriPinen, L. (1976).J.Virol. 20, 203-210. Utermann, G., and Simons, K. (1974).J.Mol. Biol. 86, 569-587. VMnLnen, P., and KMrilinen, L. (1979).J. Gen. Virol. 43,593-601. VMnilnen, P., and KMri&nen, L. (1980).J. Gm. Virol. 46,467-475. VaPnhen, P.,Gahmberg, C. G., and KiiLriainen, L. (1981).Virology 110, 366-374. van Meer, G., Simons, K., and Op den Kamp, J. (1981).Biochemirty 20, 1974-1981. van Steeg, H. (1982).Ph.D. Thesis, University of Utrecht, Netherlands. von Bonsdorff, C.-H. (1973).Cmnrentat. Biol. 74, 1-53. von Bonsdorff, C.-H., and Harrison, S. C. (1975).J.Varol. 16, 141-145. von Bonsdorff, C.-H., and Harrison, S. C. (1978).J.Virol. 48,578-583. Waite, M. R. F., and Pfefferkorn, E. R. (1970).J.Virol. 5,60-71. Waite, M. R. F., Brown, D. T., and Pfefferkorn, E. R. (1972).J.Virol. 10,537-544. Waite, M. R. F., Lubin, M., Jones, K., and Bose, H. R. (1974).J.Virol. 13, 244-246. Walter, P., and Blobel, G. (1982a).J.Cell Baol. 91,557-561. Walter, P., and Blobel, G. (1982b).Nature (London) 299,691-698. Warren, G. B. (1981).I n “Comprehensive Biochemistry” (B. Finean and R. Mitchell, eds.), Vol. 1, pp. 215-257. Elsevier, Amsterdam. Welch, W. J., and Sefton, B. M. (1979).J.Virol. 29, 1186-1195. Welch, W. J., and Sefton, B. M. (1980).J.Virol. 33, 230-237. Welch, W. J., Sefton, B. M., and Esch, F. S. (1981).J.Virol. 38,968-972. Wengler, G.,and Wengler, G. (1976).Virology 73, 190-199. Wengler, G., Boege, U., Wengler, G., Bischoff, H., and Wahn, K. (1982).Virology 118, 401-4 10. White, J., and Helenius, A. (1980).Proc. Natl. Acad. Sci. U.S.A. 77,3273-3277. White, J., Kartenbeck, J., and Helenius, A. (198O).J.Cell Biol. 87,264-272. White, J., M a t h , K., and Helenius, A. (1981).J.Cell Biol. 89,674-679. Wiley, D. C., and von Bornsdorff, C.-H. (1978).J.Mol. Biol. 120,375-379. Wilson, I. A., Skehel, J. J., and Wiley, D. C. (1981).Nature (London) 289, 366-373. Wirth, D. F., Katz, F., Small, B., and Lodish, H. F. (1977).Cell 10,253-263. Witte, 0.N.,and Baltimore, D. (1977).Cell 11, 505-511. Wolinsky, J. S., and Johnson, R. T. (1980).I n “Comprehensive Virology” (H. FraenkelConrat and R. R. Wagner, eds), Vol. 16,pp. 257-296. Plenum, New York. Zavada, J. (1982).J . Gen. Virol. 63, 15-24. Ziemiecki, A.,and Garoff, H. (1978).J.Mol. Bwl. 142, 259-269. Ziemiecki, A,, Garoff, H., and Simons, K. (1980).J. Gen. Virol. 60, 1 11-123. Zilbentein, A., Snider, M. D.,Porter, M., and Lodish, H.F. (1980).Cell 21,417-427.

MICROTUBULE ASSEMBLY: A REVIEW OF PROGRESS, PRINCIPLES, AND PERSPECTIVES By DANIEL L. PURICH and DAVID KRISTOFFERSON‘ Department of Chemlotry, Unlvenlty of Callfornla, Sante Barbara, Callfornla

. . . . . . . . . . . . . . . . . . . A. Tubulin . . . . . . . . . . . . . . . B. Microtubule-Associated Proteins . . . . , . . . 111. Nudeation or Initiation of Microtubule Assembly . . . . A. Theoretical Considerations. . . . . . . . . . B. Experimental Findings. . . . . . . , , . . IV. Microtubule Elongation . . . . . . . , . . . A. Theoretical Considerations. . . . . . . , . . B. Experimental Findings . . . . . . . . . . . I. Introduction.

11. Biochemical Properties of Microtubule Proteins

. . . . . .

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

,

,

,

. . . . . .

V. Protomer-Polymer Equilibria and Critical Concentration Behavior . . A. Theoretical Considerations. . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . , . . , . Protomer Flux with Assembled Polymers (Microtubule “Treadmilling”) A. Theoretical Considerations. . . . . . . . . . . . B. Experimental Findings. . . . . , . . . . . . . Concluding Remarks . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . B. Experimental Findings.

VI. Microtubule Length Redistribution VII. VIII.

. , . .

. .

.

133 137 138 153 158 159 162 168 169 172 182 183 185 190 194 194 200 206 207

I. INTRODUCTION Until the past decade, the cytoplasm was widely considered to be structurally unorganized with the main division of labor at the organellar level. Certainly, relatively little was known about the nature of the cytoskeleton (with the notable exception of the mitotic apparatus and striated muscle), and the dynamics of cytoplasmicbehavior were conceptualized vaguely in terms of sol-gel transitions without a sound molecular foundation. Substantial improvements in electron, light, and Buorescence microscopy, as well as the isolation of discrete protein components of the cytoskeleton, have led the way to a much better appreciation of the structural organization of the cytoplasm. Indeed, the lacelike network of thin filaments, intermediate filaments, and microtubules in nonmuscle cells is as familiar today as the organelles identified Present address: Department of Biochemistry and Biophysics, University of California Medical Center, San Francisco, California 94143.

I33 ADVANCES IN PROTEIN CHEMISTRY, Val. 36

Copyright 0 1984 by Academic Press Inc. All rights of reproduction in any form reserved. ISBN 0-12-034236-7

134

DANIEL L. PURICH AND DAVID KRISTOFFERSON

after the advent of biological electron microscopy. While we are still at a loss to explain the underlying molecular details of cytoskeletal interactions, a major means for characterizing and comprehending the emerging principles of cytoskeletal action has been to isolate and define the properties of various constituents. In this article, we consider one such component, the microtubule, from the perspective of the theoretical and experimental efforts to elucidate its assembly and disassembly properties. For this purpose, we will concentrate on the in vitro self-assembly of microtubules derived from neural systems. The organization of this article is based in part on the Oosawa theory which states that biopolymer self-assembly is thought to proceed sequentially through steps involving nucleation, elongation, protomer-polymer equilibration, and polymer length redistribution. Yet, it is prudent to consider first some general aspects of microtubule research, as well as a description of the biochemical properties of microtubule components. We will close the article with an analysis of the steady-state opposite-end assembly-disassembly reactions of microtubules. General aspects ofmicrotubule research. The first studies of a microtubulelinked process were probably the observations of Leeuwenhoek who reported the existence of cilia in a communication to the Royal Society of London in 1676. Although the Spanish investigator Cajal first visualized with silver staining the structures now recognized as microtubules, the significance of this finding remained unclear until Keith Porter and others applied electron microscopy to the study of ciliary structure, fixed nerve axons, and the mitotic apparatus (see the special issue of The Journal of Cell Biology entitled “Discovery in Cell Biology” for an illuminating discussion of the milestones in the characterization of microtubule systems). In any case, the analysis of microtubule function over the last few decades has resulted in a greater appreciation of the generality of microtubule participation in eukaryotic cell function (see Table I). Microtubules are an example of that class of fibrous protein assemblies which are derived from the polymerization of globular subunits. The main component protein of microtubules is termed tubulin (Mohri, 1968),and it is the target or major receptor site for a number of pharmacological agents (Fig. 1). (Although we shall not elaborate on the detailed molecular action of these drugs, their importance in defining the properties of microtubule systems cannot be overestimated. Furthermore, they attest to the significance of understanding the dynamics of microtubule action as a means for developing and understanding clinically useful substances.) There is also ample evidence to suggest that microtubules form in what Lauffer (1975) calls an entropy-driven assembly reaction, but not all microtubules exhibit the same stability. Indeed,

MICROTUBULE ASSEMBLY

135

TABLE I Some Roles of Microtubules in Cellular Processes Function

Example($

1. Maintaining cellular anisometry

Nerve axons and retinal rod cells depend on the microtubule network to support their distinctive cell morphology The activation of platelets and the action of polymorphonuclear leukocytes depend on changes in the microtubule cytoskeleton Both doublet (or fused) microtubules and singlet microtubules are essential components of flagella and cilia. Singlet microtubules and centrioles, both composed largely of tubulin, are essential structural elements in eukaryotic mitosis and related processes. Granule relocation in melanocytes and axonal transport appear to require intact microtubules The so-called patching and capping for distributing cell surface determinants appear to be influenced by the state of microtubule assembly The arthropod mechanoreceptor is composed of specialized cilia acting in sensory-transducing processes

2. Promoting cell shape changes

3. Providing the superstructure for cell motility machinery

4. Distributing specific molecules on cell surfaces 5. Propagating signals in modified cilia

there are both stable and labile classes of tubule structures formed from microtubules, and the ability to distinguish such stability differences is frequently evaluated on the basis of the ability of the particular structure to withstand exposure to cold temperatures, chemical agents such as colchicine, and changes in solution variables. For example, the mitotic apparatus (excluding the centrioles) and the microtubule cytoskeleton of the axon are both readily disassembled upon cooling; yet, the microtubule network of the cilium or flagellum is relatively unaffected by such mild perturbations. It is rather generally accepted that factors other than the tubulin protomer are significant in imparting this differential stability, because tubulin isolated from a wide spectrum of sources behaves in a surprisingly similar fashion. Microtubules in the cytoskeleton and mitotic apparatus are also in a state of dynamic equilibrium and flux with unpolymerized tubulin, and tubulin appears to be an excellent example of the proteins which Pauling (1953) postulated to exist as globular protomers or as insoluble, fibrous, supramolecular structures akin to unpolymerized and polymeric hemoglobin S. The current view of the microtubule cytoskeleton in nondividing cells comes from the development of tubulin-specific antibodies for indirect immunofluorescent localization of microtubules (Fuller et al., 1975; Weber et al., 1975). The general structural features of such cyto-

136

DANIEL L. PURICH AND DAVID KRISTOFFERSON

0 OCH3

Colchicinr

CH3O

Grireofulvin

OCH3 OCH3

Podophyllotoxln

Vlnbloitinr

Tax01

Maytanrinr

FIG. 1. Structures of several prominent chemotherapeutic agents that interact with tubulin or microtubules.

skeletons are illustrated in Fig. 2. The tubules originate at the centrioles, which liejuxtaposed just outside the cell’s nuclear envelope. Such microtubule networks extend throughout the cytoplasm of the cell with the distal ends interacting in an unknown fashion with other filamentous or membranous anchor points. Neurons, on the other hand, bear cytoskeletal networks which are not so haphazardly arranged; instead, the tubules are colinear with the long axis of the axon, and they appear to promote the vectorial growth of neurites. In some cases, microtubules determine the general morphology of the cell, and the addition of various drugs results in the rounding up and loss of their elongated and frequently anisometric shape. In the course of normal cell function, however, microtubule depolymerization must occur, as in response to the global changes in cell structure achieved by switching from the resting cytoskeletal structure to the characteristically different mitotic apparatus. These changes most likely reflect the action of microtubule-associated enzymes which modulate various assembly and disassembly

MICROTUBULE ASSEMBLY

137

FIG. 2. Microtubule network in fibroblast cells as visualized by use of antibodies directed against tubulin and indirect irnrnunofluorescence techniques. (Courtesyof Dr. William C. Thompson.)

reactions of the tubulin protomer either directly or through associated proteins and structures. Terry and Purich (1982) have considered some of the nucleotide-dependent enzymes associated with microtubule systems. This very brief overview provides only a glimpse of the complexity of microtubule systems, but the necessity to understand the underlying protein chemistry of microtubule components should be evident. 11. BIOCHEMICAL PROPERTIES OF MICROTUBULE PROTEINS

There has been an intensive examination of the biochemical properties of microtubule proteins over the past 15 years, and most of the work has focused on proteins derived from neural systems. For convenience, we will deal with the molecular properties of tubulin first and then collectively consider the so-called microtubule-associated proteins (MAPS).

138

DANIEL L. PURICH AND DAVID KRISTOFFERSON

A. Tubulin Of all of the proteins associated with microtubules, tubulin is the principal structural component forming the characteristic structure of the tubule. The basis for this discovery was the report by Inoue (1952) who interpreted the action of colchicine as a mitotic spindle poison in terms of the disruption of spindle fibers and a shift in the subunit-polymer equilibrium to favor the disassembled state. Taylor (1965) placed these notions on a firmer molecular foundation by using labeled colchicine to examine the kinetics of colchicine action. The use of radiolabeled colchicine as a means for identifying the microtubule structural protein flourished during the mid- 1960s, and several groups reported the binding of colchicine to the protein now called tubulin (Shelanski and Taylor, 1968; Wilson and Friedkin, 1967; Borisy and Taylor, 1967; Weisenberg et al., 1968). The identification of the drug receptor as a 110,000 molecular weight protein accorded well with other investigations of ciliary microtubules (Renaud et aL, 1968). Since that time, several thousand reports on this protein have appeared.

I . Isolation The current methods for isolating microtubule protein are based on the early work of Weisenberg (Weisenberget al., 1968; Weisenberg and Timasheff, 1970; Weisenberg, 1972). Two basic approaches are employed: (1) ion exchange chromatography on DEAE-Sephadex to take advantage of the necessity of using 0.8 M sodium chloride to elute tubulin; and (2) recycling protocols based on the sequential assembly and disassembly of tubules in the warm and cold, respectively, with intervening centrifugation. The latter method is based upon the germinal finding by Weisenberg (1972) of conditions favoring the in vitro reassembly of microtubules. He observed that combination of rat brain extracts with the calcium ion chelator EGTA in the presence of magnesium ions, GTP (or ATP), and elevated temperature (35°C) resulted in rapid polymerization of tubulin. The structures formed in these experiments were helical in nature, strongly resembling the tubules observed in many eukaryotic cell types. Shelanski et al. (1973) introduced a modification of the Weisenberg protocol by including glycerol in the polymerization medium, and this procedure has been widely applied to the large-scale purification of tubulin and MAPS. Although the yields of tubulin based on the amount of tubulin in the cell extract are sufficient for most workers, more characterization could help to explain why the greatest fraction of tubulin remains unpolymerized in the cell extracts. The number of methods currently employed in microtubule protein

MICROTUBULE ASSEMBLY

139

isolation reflects far more than a consideration of ease of application or even yields. In part, the different viewpoints of investigators are reflected in the choice of the method. Some believe that pure tubulin (devoid of MAPs) should be the primary object of investigation, thereby eliminating the need to deal with possible complications arising from the presence of the MAPs. Because tubulin assembly is generally studied in the protein concentration range of 1-5 mglml, there is in fact the added task of sorting out the activities of contaminating enzymes which are copurified with whole microtubule protein (i.e., tubulin plus MAPs). Furthermore, there is reason to question whether the copurification of a protein through cycles of assembly and disassembly proves the intracellular association of such a protein with microtubules. The logic of this has been persuasively presented by Lee et ad. (1978b), and methods for isolating pure tubulin dimers are presented by Lee (1982). At the same time, other investigators, while recognizing this limitation on proving any such association, have set about to characterize proteins which are likely modulators of tubule assembly. These investigators generally welcome the opportunity to explore the possible roles played by proteins operationally defined as MAPs on the basis of their copurification properties (see Section 11,B). Another concern about the method of isolation relates to the altered assembly properties or other structural changes of tubulin attending high glycerol concentrations. Detrich et al. (1976) demonstrated that glycerol and tubulin do combine with rather high affinity, with 2 mol glycerol bound per 110,000 molecular weight. Moreover, glycerol affects the number of free thiol residues on the tubulin protomer (Mellon and Rebhun, 1976), and it influences the yield of MAPs copurified with tubulin (Scheele and Borisy, 1976). These findings have motivated the development of glycerol-free microtubule purification methods (Borisy et al., 1975; Asnes and Wilson, 1979; Murphy, 1982). Karr et al. (1979a, 1982) have also presented a microtubule purification protocol to deal with the above-mentioned enzyme contamination problem; in particular, their finding of mitochondria1 marker proteins in microtubule protein led them to a sucrose extraction scheme to minimize osmotic shock and disruption of mitochondria and synaptosomcs. For special purposes, tubulin may be prepared by vinblastine-induced paracrystal formation and subsequent centrifugal manipulation (Wilson et al., 1976). Dimethyl sulfoxide has been useful in promoting microtubule assembly (Himes, 1977), and deuterium oxide is also effective as a promoter of tubule assembly (Borisy et al., 1975). An example of the latter is provided by the isolation of microtubule proteins from neuroblastoma cells (Olmsted and Lyon, 1981). For separating the MAPs fraction from tubulin, one may apply the

140

DANIEL L. PURICH AND DAVID KRISTOFFERSON

phosphocellulose chromatography method of Weingarten et al. ( 1975) or the method of Williams and Detrich (1979) to allow for the loss of magnesium ion during passage over this cation exchanger. Alternatively, one may use DEAE-Sephadex chromatography (Murphy et al., 1977). The former eliminates the need for desalting the tubulin fractions prior to use, but does require the desalting of MAPS-enriched fractions which elute at 0.6 M potassium chloride. In this regard, the DEAE method requires desalting of both MAPS and tubulin fractions.

2. Structural Properties of the Tubulin Dimer By electrophoretic analysis, tubulin has been characterized as a dimer composed of two polypeptide chains, generally referred to as the a and P chains. On the basis of early cyanogen bromide cleavage experiments (Ludueiia and Woodward, 1973), distinctions between the sequences of these two polypeptide chains from embryonic chick brain protein became evident. Lu and Elzinga (1978) reported on the sequence at the carboxyl terminus of bovine brain tubulin, and their studies along with those of Ponstingl et al. (1979) show the remarkable sequence identity of the (Y subunit. Recently, Ponstingl et al. (1981) elucidated the complete amino acid sequence of this same subunit from porcine brain. Their studies indicate that the first 40 residues on the C terminus contain 16 glutamyl and 3 aspartyl residues, and work on the structure of this region will be interesting to follow. Furthermore, there was heterogeneity in the sequence at six positions concentrated near residue 270, and they presume that four different polypeptide chains are present in the porcine a subunits isolated from brain. This observation and the finding of microheterogeneity in calf brain tubulin (Lee, 1982) suggest the possibility of multiple tubulin pools akin to the earlier proposal by Fulton and Simpson (1976). Valenzuela et al. (198 1) determined the nucleotide sequences (and deduced the corresponding amino-acid sequence) of both the a-and Ptubulin messenger RNAs (mRNAs)isolated from embryonic chick brain. Figure 3 shows this deduced amino acid sequence for the complete P and a chains (except the 38 N-terminal amino acid residues of the latter). Several additional conclusions were drawn from these investigations. First, the sequence homology between the a-and P-tubulin subunits is substantial, suggesting a common ancestral origin. Second, the chick mRNA sequence data agree basically with the inferences of some sequence homology, although not great, between limited regions of t u b y . lin and actin. Third, the mRNA sequence demonstrates that the tyr ine at the C terminus of the a subunit [which is subject to e n z y m d y r o sination and detyrosination, as reviewed by Thompson (1982)j isalready

'

MICROTUBULE ASSEMBLY

141

specified at the gene level. In this regard, metabolism related to the turnover of this terminal tyrosino group probably requires detyrosination of the nascent a subunit. Finally, there is evidence of four separate a and p genes in both chicken and Drosophilu (Cleveland et al., 1980; Sanchez et d., 1980). Interestingly, a comparison of the primary structure derived by the German and American workers for porcine and chick brain tubulin confirms the notion that there is remarkable sequence homology, and this suggests that there are stringent requirements on the structure of tubulins over a reasonably great phylogenetic range. Virtually all of what is known about the secondary and tertiary structure of tubulin has been gleaned from a limited number of spectroscopic and chemical modification studies. Failure to obtain crystals of suitable quality for X-ray diffraction studies likely results from both heterogeneity in the subunits and the propensity of tubulin to polymerize into many polymorphs (see Section 111). George et al. (1981) reported that as many as 17 distinct protein peaks may be discerned after isoelectric focusing of purified tubulin. Timasheff’s laboratory has provided some of the best information on the secondary and tertiary structure through the judicious application of circular dichroism (CD) measurements (Ventilla et al., 1972; Lee et al., 1978a). Lee (1982) summarized the current status of these investigations. Briefly, the CD spectrum of tubulin is dominated by an absorption band near 275 nm which may be attributed to tryptophanyl residues and to the dihedral bond structure of disulfide linkages. The CD behavior in the region of 190-240 nm is, of course, especially valuable as an indicator of changes in secondary structure upon perturbation by solution variables (i.e., pH, ionic strength, buffer composition, etc.), ligand binding, or changes in temperature. Lee et al. (1978a) found that at pH 7 calf brain tubulin maintains its structure over the range from 5 to 37°C. Magnesium ion (up to 16 mM) and calcium ion (up to 0.1 mM) both fail to change the spectral properties; hence, these ions are not thought to have much general impact on tubulin structure. Above pH 7, there is evidence of a progressively greater conformational flexibility in the molecule, with especially prominent qualitative changes above pH 10 signaling more gross structural alteration. The single disulfide bridge in brain tubulin (Lee et al., 1973, 1975; Eipper, 1974b) may in fact contribute to the structural integrity of tubulin. It is also rather interesting that the addition of either daunomycin or vinblastine does not perturb the CD spectrum in the 190- to 250-nm range, suggesting that these inhibitors of self-assembly fail to evoke any major structural changes in the protein. In regard to the secondary structure of tubulin subunits, it is interest-

..

?b?t Arq

Glu Ile V a l Hi. '0

Met Arg G l r Scr I l e

.

.

20

phe T r p Glu Val I l e Ser Asp Glu H i s Gly I l e Asp I l e G l n Ala Gly Glu Cys Gly Asn Glu I l e Gly Ala 0 0 Glx Leu Tyr Ser . I l e H i s V a l Ru G l n Ala llu Val Gln I l e llu Asx Ala Ser

288

Ser

.

.

Ro llu G l y

. . . . . . . .

Ser

Tyr Hi. Gly

. . . .

llu T y r Qs I l e Asp Asn Glu Ala Leu Tyr App I l e Cys Pne Arq Ru Leu Lys Leu Ru 'Ihr 0 O.... O . . . . 0 Ala Rp R t l V a l Asp Asn Glu Ala I l e Tyr Asp I l e CyslArq Arq Asn Leu Asp I l e Glu A q l P r O Ihr T y r ] h r AsnlLeu Asn f i g Leu IleIGly

.

Arg

Ala Ru: pct Ser Gly Val Ru Ru Cys 0. G l n Ile Val Ser Ser Ile Ru Ala Ser

.

Ws Leu Ala Val Asn pct Val Pro Rr Pro

Ru Glu Phe Gln Ru Asn Leu Val

Pro 'IW Pro

288

.

388

Asp Ser Lys Asn pct pct Ala

0

Glu proAmp

Glu G l n

*

Asn G l n

pct

pct

o

Val Lys

Leu Am Val 0

Ala

G k

0

A.n Ala Ala Ile Ala Ru Ile Iqs

. .

3€0

.

.

388

Leu - Lys Wet Ser - Ala llu Phe -Ile Gly Asn Ser Ru Ala Ile G l n Glu Leu Phe Lya Aq Ile Ser Ars GlY 0 1 0 ..1 .1 . 0 Ru Val Val Pro Gly Gly Asp Leu Ala LyS Val G l n Arg Ala Val Cys Mt?t Leu SerIAm Ru llu Ala Ile]Ala[Glu Ala TrplAla Arg Leu Asp

.1

....

..

Gly Glu Gly I4t Asp Glu Wet Glu Phe Ru G l u Ala Glu H i s Lys Rp AsplIar M e t

.

.

Leu Val

0

.

wIAlaILyS Arg Ala Phe Val H i s T r p TyrjVallGly Glu Gly Met Glu GlU]Gly[GlU

0..

Ser Asn

Phe Ser Glu Ah]Arg G l u Asp

428

ser Glu

pct

~w Asp

pct

0. Ala Ala Leu Glu ~ y ~ s s p

Tyr

FIG.3. Comparison of the amino acid sequences predicted from the a- and p-tubulin mRNA sequences. Open circles indicate homologous, conservative substitutions, and boxed regions indicate areas of high homology. For completeness, the sequence of the first 25 amino acids of chick brain a-tubulin, deduced by the work of Ludueiia and Woodward (1973), is also included. Note that the asterisk indicate identical amino acids common to both polypeptide chain 289,650-655.1 sequences. [Reproduced from Valenzuela at al. (1981). Nature (L&)

144

DANIEL L. PURICH AND DAVID KRISTOFFERSON

ing to note that the primary structural data of Valenzuela et al. (198 1) permitted these investigators to predict the secondary structure on the basis of the approaches of Garnier et al. (1978) and Chou and Fasman (1974). The C-terminal regions are both predicted to be highly a-helical, but the authors state that other levels of structural homology between the two subunits were not obvious. Both these conclusions and the earlier work by Lee et al. (1978b) certainly suggest that tubulin is a rather typical protein when viewed in terms of secondary structure. The quaternary structure of tubulin became evident in the early efforts to characterize the colchicine-binding protein (Borisy and Taylor, 1967; Wilson and Friedkin, 1967), and the molecular weight was nearly 110,000. From sedimentation studies in the presence of 6 M guanidine * HC1 and 0.10 M 2-mercaptoethanol, Weisenberg et al. (1968) found the subunit molecular weight to be about 57,000. [The reader will note that the primary structural data now indicate that 50,000 is a better value for either subunit (Valenzuelaet al., 1981).] The data suggested that tubulin in its native form was a dimer with several possible forms: aa,ap, or pp. The current view that the tubulins assume an a@ heterodimer structure in solution is based on the studies of Luduefia et al. (1977). Using dimethyl-3,3'-(tetramethylenedioxy)dipropionimidateto cross-link soluble chick brain tubulin, these workers analyzed the frequency of a x a,a X p, or p X p dimers with gel electrophoresis. About 60-90% of the cross-linked dimers were of the a X /3 type. The relative amounts of a X a and /3 X /3 homodimeric species increase with reaction time, and Luduefia et al. (1977) attribute this to cross-linking of aggregated species. Of course, the lack of complete heterodimer cross-linking always raises the possibility that the heterodimer is the predominant species only in the in vitro studies. Nonetheless, the a/3 heterodimer model has gained rather broad acceptance, and it also explains the regular periodicity of protomer structure observed in image reconstruction experiments (Amos and Klug, 1974; Amos et al., 1976).

3. Properties of the Nucleotide Binding Sites Weisenberg et al. (1968) and Berry and Shelanski (1972) provided clear evidence for the existence of two, nonidentical guanine nucleotide sites on each tubulin dimer. These sites are distinguished on the basis of their ability to exchange with nucleotide added to the medium: the exchangeable- (or E-) site readily exchanges with labeled GDP or GTP whereas the nonexchangeable site (or N-site) exhibits no tendency toward exchange on the time scale of several hours. Terry and Purich (1982) discussed the possibility that the heterodimer structure is the cause of this asymmetry, but this idea was anticipated in a number of

145

MICROTUBULE ASSEMBLY

earlier studies. Indeed, Geahlin and Haley (1977, 1979) investigated the topology of labeling following photochemical activation of 8-azido-GTP, and their data strongly point to the p subunit as the principal labeled species. One might then propose that the a and p subunits are so configured as to allow the N-site nucleotide to reside at the a/3interface; then, using the finding of Geahlin and Haley (1977, 1979), the structure would be aN+ PE.In the microtubule, the subunits are arranged in a protofilamentous structure (see Section II,A,4); this would be CXN + * a N -+ * (YN -+ P E * (YN -+ & --,using the above notation. Such an array readily explains the observation that the E-site nucleotide becomes nonexchangeable in the transition from protomer to polymer (Weisenberg et al., 1976).The other consequence of this configuration is that the similarity in the primary sequence of the two subunits would be reflected in the manner of N- and E-site location in the heterodimer. The exchangeable nucleotide site has been characterized through a number of studies, and while the preference for guanine nucleotide binding is obvious, other ligands will bind under suitable conditions. The rough order of affinity is GTP [& = 2.2 X lo-' M (Zeeberg and Caplow, 1979)]> GDP [& = 6.1 X lO-'M (Zeeberg and Caplow, 1979)] > dGTP [& = 2 X M (Penningroth and Kirschner, 1978)] > 5'guanylylimido diphosphate * 5'-guanylmethylene diphosphonate [& = 3 X 10-6M (Karr and Purich, 1978)] > ITP [& = 10-5M (MacNeal and Purich, 1978b)l = GMP(CH2)PP[ K d = 1.3 X loW5M (Sandoval and Weber, 1980)] > chromium(II1) GTP [& = M (Purich and MacNeal, 1978)] = 5'-adenylylimido diphosphate [& = 4 X M (Penningroth and Kirschner, 1978)] = 5'-adenylylmethylene diphosphonate [Kd = 3 X M (Purich and MacNeal, 1978)l. Interestingly, the more weakly bound ligands will not bind to any appreciable extent in the presence of GTP or GDP; this necessitates removal of the E-site nucleotide by charcoal treatment (Penningroth and Kirschner, 1978) or alkaline phosphatase treatment (Purich and MacNeal, 1978). The work of Kirsch and Yarbrough (1981) has helped to show that GTPyS, an analog containing sulfur in the terminal phosphoryl moiety, will also bind to tubulin and promote assembly, but very little of the nucleotide was converted to GDP in the assembly process. These investigators also examined the properties of E-site-directed ligands fluorescently labeled via the y-thio group. They found that an analysis of the steady-state and dynamic fluorescence anisotropies shows that the fluorophore of such ligands experiences little rotational freedom, and that the observed depolarization of fluorescence is dominated by the rotational tumbling of the protein-ligand complex (Kirsch and Yarbrough, 1981). Until quite recently, it was thought that the off-rate constants for

..-

146

DANIEL L. PURICH AND DAVID KRISTOFFERSON

desorption of GTP and GDP could be fully evaluated by coupled enzyme assays such as those described by Terry and Purich (1979). While these investigators obtained correct apparent off-rate constants (0.005 and 0.011 sec-I for loss of GTP and GDP, respectively) at a fixed level of tubulin, the true off-rate constants were determined by Brylawski and Caplow (1983). These later investigators utilized a radioisotope-trapping protocol, starting with tubulin-bound radiolabeled GDP, to dilute the radiospecific activity of the labeled GDP after it desorbed from tubulin and mixed with unlabeled GDP. This trapping protocol offers the advantage that reversal of the label desorption process is no longer competitive, in a kinetic manner, with the pyruvate kinase-mediated conversion of labeled GDP to GTP for subsequent analytical determination of the rate of ligand desorption. A lower limit for the first-order rate constant for GDP dissociation was thus estimated to be 0.14 sec-', and this corresponds to a half-life of about 5 sec. From the off-rate constant and the dissociation constant for the Tb * GDP complex, Brylawski and Caplow (1983) estimate a bimolecular association rate constant of 2.2 X lo6 M-I sec-I, suggesting that this ligand-protein reaction proceeds at a rate near the diffusion limit. This also would suggest that the combination of GDP with tubulin may be a one-step process. Finally, these new experiments foreshadow the possibility that tubulin interactions with GTP may be faster than earlier proposed. Karr and Purich (1978) reported that GDP and GTP behave differently with respect to the intrinsic fluorescent properties of tubulin. Binding of GTP, GMPP(NH)P, and GMPP(CHz)Pto tubulin-GDP displaces the GDP and evokes the same final degree of fluorescence quenching, although the three ligands had slightly different apparent dissociation constants. These data suggested the possibility that the assembly-inhibiting properties of GDP and the assembly-promoting properties of GTP and its analogs might be related to the stabilization of two different conformations. The strongest evidence against a coupling of events at the tubulin Nsite to assembly was provided by Spiegelman et al. (1977). These investigators isolated tubulin from Chinese hamster ovary cells which had received a pulse of [s5S]methionineand [s2P]Pito label both protein and nucleotide during the course of biosynthesis of new tubulin molecules. After multiple cycles of assembly/disassembly with brain tubulin as a nonradioactive carrier, the state of phosphorylation of the N-site nucleotide could be evaluated by thin-layer chromatography of nucleotides freed from tubulin by denaturation. This protocol has the advantage that any radiolabel at the E-site is lost during the multiple recycling

MICROTUBULE ASSEMBLY

147

process, and the turnover of tubulin within the cultured cells could be determined by the radiospecific activity of 35S-labeledproteins. These measurements placed the half-lives for the N-site GTP/GDP label and the protein at 33 and 45 hours, respectively. Thus, their conclusion that the N-site nucleotide does not undergo significant exchange with cellular nucleotide pools seems quite well justified. From studies of the stoichiometry of nucleotide hydrolysis during assembly and the level of Pi produced, MacNeal and Purich (1978a) reached a similar conclusion with in vitro polymerization (see Section IV,B). 4. Structural Properties of Microtubules There has been a continuing effort to understand the tubulin polymerization process through the characterization of microtubule substructure and of the arrangement of tubulin promoters in polymorphic species such as ring and sheet structures. Indeed, microtubules can be described best as hollow cylindrical structures composed of longitudinally arranged protofilaments, each of which is itself a threadlike structure composed of cup protomers arranged in a head-to-tail manner. These gross structural features are almost immediately evident upon inspection of transmission micrographs which are generally obtained from heavy metal staining experiments. Actually, such micrographs contain a wealth of additional structural data for the partially collapsed microtubules, and the ongoing challenge has been to devise appropriate theoretical and experimental approaches for unraveling the structure from transmission electron micrographs which by their nature have structural information superimposed for the two layers or walls of the tubule. Nonetheless, much of the detailed spatial arrangement of the subunits can be determined through the application of the techniques of optical diffraction and optical image filtering that have become the mainstay of methods for defining supramolecular structure. These approaches have yielded valuable structural information about the parallel or antiparallel arrangement of protofilaments, the number of protofilaments in a tubule, the arrangement of subunits relative to the long axis and radius of the tubule, and the nature of interprotofilament interactions. For example, one can utilize such approaches to distinguish between homologous and heterologous interactions between the tubulin protomers in neighboring protofilaments in a microtubule. Amos (1979) has presented a detailed account of the current views on microtubule structure, and the discussion presented here is only a summary and cannot supplant Amos’ excellent treatment and analysis. Rather, our description is offered as a means to orient the reader with respect to the

148

DANIEL L. PURICH AND DAVID KRISTOFFERSON

potential connections between the gross morphological features and the mode of tubule assembly.

....-p-por*p-p.-p

....-p-pa+p-p.-p

Homologous Interactions

........-p. * p -8. ......... a-p. -p*

a-+p -p .... a-p. H p -

Heterologoua Interactions

From thin sectioning of various microtubule-containing tissues and embedded tubule samples, microtubules are known to exhibit a tubular structure of constant radius of about 12 nm along the exterior. The outside diameter of 24 nm and the inside diameter of 14 nm are most commonly observed for tubules containing 13 protofilaments. Yet, these values are to be contrasted with corresponding dimensions of 30 and 14 nm, respectively, obtained with hydrated brain tubules (Mandelkow et al., 1977). The 5 nm discrepancy in outside diameter apparently reflects the effects of dehydration during sample preparation for electron microscopy, and it serves to underscore the need to establish means for comparing the detailed solution with anhydrous structures of microtubules. In terms of length, reassembled brain microtubules tend to be distributed over a range from a fraction of a micron up to about 100 pm. The average tubule length obtained by reassembly depends on the conditions applied (Sloboda et al., 1976; Johnson and Borisy, 1977; Karr et al., 1980a; Terry and Purich, 1980). On the other hand, axonemal.tubules within flagella and cilia are limited to the length of the flagellum or cilium from which they are derived. Even so, the axoneme can serve as a template or microtubule organizing center in vitro (see Section IV,B,4) with singlet microtubules growing from the “A-type” fiber of the axonema1 doublet. As noted earlier, the protofilament is the most dominant and distinctive substructure of the microtubule. The number of protofilaments in microtubules from a variety of organisms has been determined. Although most tend to have 13 wall protofilaments (Tilney et al., 1973), microtubules of crayfish nerve cord axons have only 12, and glial cells associated with these axons have the usual 13 (Burton and Hinkley, 1974; Burton et al., 1975). Cockroach epidermal cells have 15 protofilaments per tubule (Nagano and Suzuki, 1975), demonstrating that there is a range of protofilament numbers found in various organisms. It is noteworthy that reassembled microtubules display a type of polymorphism with respect to the number of protofilaments. This was first reported by Pierson et al. (1978) who worked extensively with purified bovine brain tubulin subjected to reassembly under a number of conditions, The basic notion underlying their investigation was that the solu-

MICROTUBULE ASSEMBLY

149

tion conditions might be at least partly responsible for specifying or influencing the protofilament count. With each successive cycle of reassembly, the number of microtubules with 13 protofilaments decreased, while those containing 14 and other numbers of filaments increased. After three cycles, those with 14 predominated, although tubules with 11,12, 16, and 17 protofilaments were still observable. The buffer which maintained the fidelity of protofilament number best was 20 mM 2-(Nmorpho1ino)ethane sulfonate with 70 mM sodium chloride. In recent unpublished work, Himes (private communication) has found that microtubule protein isolated by the sucrose-extractionmethod of Karr et d. (1979a) displayed the greatest tendency to maintain the 13-protofilament count. A continuing series of studies of the lattice structure of microtubules has provided important information on the nature of subunit arrangement in assembled microtubules. The helical lattice of the A type or complete singlet tubule extending from an axonemal doublet tubule could be described as a 13-protofilament structure staggered slightly along the tubule axis (Grimstone and Klug, 1966; Amos and Klug, 1974; Linck and Amos, 1974). Furthermore, the A-tubule is arranged in a pattern best described as a left-handed three-start helix (i.e., 5,where the number refers to the multiplicity of helical starts and the overbar is used to distinguish the left-handed helices from other right-handed helical structures). Although the 4-nm spacings are most obvious in the optical diffraction patterns, there is also a weaker 8-nm layer-line corresponding to the length of the individual, somewhat oblong, protomer which appears as a dumbbell-shaped subunit tilted slightly with respect to the long fiber axis. Likewise, most of the images have a 5-nm repeat on the equatorial line, and this corresponds to the width of each protomer within a protofilament of both brain and flagellar tubules (Erikson, 1974; Amos and Klug, 1974). Most of these results along with alternative ways of representing the A-type flagellar singlet tubule lattice are presented in Fig. 4. While the details of intersubunit bonding still remain obscure, Amos (1979)indicates that the grooves between protofilaments are much more pronounced on the outside of the tubule wall than in the inner-wall grooves. This suggests that the bonding interactions between adjacent filaments probably lie closer to the radial center than to the center of each protomer. McEwen and Edelstein (1980) have carried out a further analysis of the substructure of reassembled porcine brain tubules to learn more about the relationship of the arrangement of subunits within the lattice and the number of protofilaments in a particular tubule. The initial work of Erickson (1974) showed a close correspondence between the 13-

150

DANIEL L. PURICH AND DAVID KRISTOFFEKSON

n.13

9

= 5(or 10)

FIG.4. Lattice structure of an “A-type’’ singlet microtubule from ciliary axonemes. [Redrawn from Amos (1979). In “Microtubules” (J. Hyams, ed.), pp. 2-64. Academic Press, New York.]

protofilament brain tubule and the A-type axonemal tubule. Both had a characteristic shallow three-start helix, but the distinction between homologous and heterologous subunit interactions in reassembled brain tubules remained unclear. As with the studies of bovine brain tubules noted earlier, McEwen and Edelstein (1980) observed that the filament count fell over a range: 12 protofilaments (1% frequency); 13 (13%); 14 (84%); 15 (1%); and 16 (1%). The basic problem that they identified relates to the maintenance of homologous (or heterologous) registration throughout the tubule without any discontinuity arising after each complete turn of the multistart helix structure. These investigators have pointed out that the “B-type” lattice (i.e., the lattice observed for the incomplete or C-shaped tubule in doublet microtubules) requires an even-start shallow helix irrespective of the number of protofilaments in the tubule wall. Thus, if the 14-protofilament tubules had a three-start A-type lattice, the microtubule subunit arrangement would be thrown out of registry, forming a discontinuous pattern of both homologous and heterologous bonding interactions. With the B-type lattice structure, the pattern of subunit interactions, whether fully homologous or completely heterologous, would maintain the continuity of subunit-subunit interactions by virtue of its intrinsic design features. In this regard, it is most interesting that McEwen and Edelstein (1980) have obtained

MICROTUBULE ASSEMBLY

151

evidence for a mixed lattice in microtubules that have been reassembled in vitro. Moreover, their studies emphasize that in vitro reassembly conditions are insufficient not only with respect to maintaining the protofilament number but also with regard to the regularity of the tubule lattice structure. This only underscores the need to uncover the factors which maintain the 13-protofilament count and lattice registry in brain tissues. That microtubules exhibit an intrinsic polarity (i.e., with protofilaments running parallel and tubulin protomers regularly arranged in a head-to-tail manner) is obvious from experiments other than those involving image reconstruction. This intrinsic polarity was observed by Rosenbaum and Child (1976) and Witman (1975), who demonstrated biased tubulin addition to microtubules in vitro. This biased addition is a consequence of the nature of tubule-tubulin interactions at the two ends of the microtubule:

where the transition states for protomer-polymer reactions need not be identical. Thus, the corresponding activation energies for the reactions at the two ends will in general be different, and the corresponding rate constants will then be different. Moreover, in vitro self-assembly with axonemes serving as microtubule-organizing centers is known to occur at the distal ends (Olmsted et al., 1974; Dentler et al., 1974). This is also true for basal bodies (Snell et al., 1974) and centrioles (McGill and Brinkley, 1975; Telzer and Rosenbaum, 1979), both of which show biased elongation. This characteristic polarity of microtubules can be determined by two important methods of direct visualization. The first is based on the interaction of flagellar dynein with assembled microtubules (Haimo et al., 1979; Haimo, 1982). Electron microscopy of longitudinal sections of microtubules decorated with dynein reveals that this protein is linked to tubules by its thin stalk. The main globular head of the dynein molecule tilts at an angle of 55" in the direction of the end of preferred growth, frequently called the (+) or assembly end. In transverse sections, microtubules decorated with dynein possess a distinctive chirality or handedness in radial arrangement when viewed consistently from the same end of the microtubule polymer. The dynein arms or hooks assume a clockwise disposition when the end that corresponds to

152

DANIEL L. PURICH AND DAVID KRISTOFFERSON

the distal or plus end of axonemal-organized singlet tubules is below the plane of the micrograph being examined. The stage was set for the development of a second independent check on microtubule polarity through the work of Burton and Himes (1978) and Mandelkow and Mandelkow (1979). They found in vitro assembly conditions which lead to the decoration of microtubule walls with additional protofilaments along the length of the microtubule. Such extra protofilaments align with each other assuming the appearance of a hook when viewed in transverse section, and their clockwise or anticlockwise configurations can be used as indicators of tubule polarity. This approach has been gainfully exploited to characterize tubules from a number of cellular regions (Heidemann and McIntosh, 1980; Euteneuer and McIntosh, 1980; Heideman and Euteneuer, 1982). Nevertheless, it would appear to us that the dynein labeling may not be sufficiently specific in its tubule interactions to distinguish between tubules with a regular A-type lattice and the discontinuous (or mixed) A-B-type lattice observed by McEwen and Edelstein (1980). Indeed, Haimo et al. (1979) found that dynein binding to preassembled microtubules in vitro gave the same center-tocenter spacing of dynein arms [e.g., 24.2 & 1.2 nm ( S D ,n = 255)] as that observed in the intact outer doublet microtubules of Chlamydomonas axonemes. What is needed is a more detailed examination of the lattice of the reassembled microtubules used in the study of Haimo et al. (1979) and a determination of the ability of dynein to affect the lattice structure of tubules formed in the presence of the dynein protein. Before ending our brief description of the structural organization of the microtubule, it is useful to recall the experimental evidence for the existence of a superhelix which may serve as the point of attachment for microtubule-associated proteins. From early work on the nature of the interaction of high-molecular-weight (HMW) microtubule-associated proteins, Borisy et al. (1975) observed that microtubules appear to become saturated with these HMW proteins, and Murphy and Borisy (1975) recognized, in both transverse and longitudinal sections of reassembled microtubule pellets, that these proteins formed projections from the external tubule wall at regular intervals of 32.5 2 9.4 nm. From studies of unfixed tubules of sufficiently regular structure, Amos (1977) obtained direct evidence for a unique symmetrical superlattice with an axial repeat length along each protofilament of 96 nm, corresponding to the distance spanned by 12 tubulin protomers. One such helix is partially illustrated in Fig. 4 by using a dotted line and small filled circles to - superhelix which represents one of the 2-start designate this left-handed helical family (i.e., n = 2). Finally, we note that microtubule protofilaments also participate in the

MICROTUBULE ASSEMBLY

153

formation of sheet structures, loops, rings of various sizes, and even double-walled tubules. The structures of some of these polymorphs are described by Amos (1979), but more work will be needed to uncover the exact morphology of many of these aberrant structures. The presence of such polymorphs in cells is still the subject of some dispute, especially the occurrence of so-called opened-out sheets which frequently appear in micrographs of microtubules. That a variety of structures can be formed reminds us of the unusual plasticity of the tubulin protomer with regard to assuming a number of shapes and expressing an organized array of intersubunit bonding interactions which may be of little or no importance within functioning cells.

B . Microtubule-Associated Proteins When one considers the wide variety of microtubule functions (see Table I), the temporal, spatial, and metabolic controls on microtubule self-assembly must be staggeringly great. In this regard, it is not surprising that there should be a large number of proteins (in addition to tubulin itself) required to regulate microtubule functions. Nonetheless, there are no uniform criteria for judging whether a particular protein has such a role. In some cases, the association of a protein or enzyme with tubulin or microtubules is obvious from cell ultrastructure, as is the case of the axonemal dynein ATPase which causes outer doublet microtubules to slide past each other to generate the motive force for cellular propulsion by cilia or flagella. In other situations, the association is remarkably obvious because tubulin serves as the substrate for an enzymatic interconversion or covalent modification; even so, the role in regulating microtubule function may remain equally remarkably elusive. Such is the case for the tubulin tyrosine ligase and corresponding carboxypeptidase which add and remove tyrosine, respectively, from the carboxyl terminus of the tubulin a subunit (Thompson, 1982). Here, though the precise site of modification on the tubulin molecule is known, the role of tyrosination/detyrosination cycling is baffling, made even more so by the observation that the mRNA sequence already specifies the code for the tyrosine at the C terminus of the a chain (Valenzuela et al., 1981). From the perspective of identifying proteins which copurify along with microtubule protein as microtubule-associated proteins, there appear to be a number of candidates for service as regulatory proteins. The so-called high-molecular-weight (or HMW) proteins, which are frequently referred to as MAPI and MAP2 to designate them as the highest and next highest molecular weight species of microtubuleassociated proteins, fall into this category. Both of these proteins behave as “quantitative” (or Q) MAPS by virtue of their propensity to remain

154

DANIEL L. PURICH AND DAVID KRISTOFFERSON

associated in a definite molar ratio to tubulin through multiple assembly/ disassembly cycles. Though an exhaustive survey of these microtubule-associated proteins lies beyond the objectives of this article, summary statements about several MAPs are presented in Table 11. From the comments listed in that table, the reader will note that there are few, if any, common features shared by these proteins. Without exception, however, the nature of MAP interactions with microtubules remains vague, and there is no comprehensive review of their properties. Instead, the reviews of Kirschner (1978) and Timasheff and Grisham (1980) cover some of what is known about these proteins, and Terry and Purich (1982) have considered the status of investigations on the many of the nucleotidedependent enzymes associated with microtubule systems. For our purposes, we shall concentrate on MAP, and MAP,; the importance of the tau proteins is surveyed by Kirschner (1978). Suffice it to say that many of the earlier suspicions that the tau proteins are proteolytic breakdown products of the HMW proteins have been dispelled by the recent studies of Cleveland et al. (1979). Their findings strongly suggest that the tau proteins are a separate class of microtubule-associated proteins, and these proteins merit greater attention to reveal their mode of action. The high-molecular-weight proteins, MAPI and MAP*, are intriguing proteins in several respects. First, they both exhibit extraordinarily large M, values of 300,000-330,000 and 270,000-300,000, respectively. To the extent that SDS-gel electrophoresis methods have revealed the true M, values and that these proteins are not processed by cross-linking reactions into such large species, MAPl and MAP2 are quite possibly two of the largest polypeptides known. Second, the proteins in the HMW fraction may not represent two distinct proteins; rather, inspection of published electropherograms nearly always reveals some degree of microheterogeneity with regard to electrophoretic mobility. Third, these proteins both appear to stimulate tubulin assembly in vitro, and the report by Sloboda et al. (1976) remains one of the most comprehensive investigations of this stimulatory effect. They found that MAPs reduced the critical concentration and that the average polymer length decreases markedly with increased MAP/tubulin ratios. Both observations are consistent with enhanced elongation and nucleation, and these investigators demonstrated that the stimulatory action on elongation and nucleation could be distinguished. Indeed, storage of MAPs at 4°C leads to proteolytic fragmentation of MAPs, and such proteolysis decreases the assembly-initiating potential without affecting the extent of tubule assembly. Murphy et al. (1977) reached similar conclusions, and they present additional evidence for the presence of at least 33 other nontubulin

TABLE I1 Proteins and Enzynres Associated with Microtubule Sytcrnr

Name or designation Brain adenosinetriphosphatase

Calmodulin

Cholera toxin

Dynein ATPase

HMW, (or MAPI)

Relationship to Microtubules This enzymatic activity is persistently associated with brain microtubules even after multiple cycles of warm-induced microtubule assembly, centrifugation to separate protomer and polymer, cold-induced disassembly, and subsequent centrifugation to remove coldstable aggregates (White et al., 1980). The enzyme hydrolyzes both GTP and ATP, and recent work by Tominaga and Kaziro (1982) indicates that there are two distinct ATPases, one that is of low M , (around 33,000) and tubulin dependent in the presence of calcium ion, and the other of larger size and associated with membrane vesicles This !ow-molecular-weight protein appears to depolymerize microtubules in the presence of calcium ion, but its ex& role is unclear due to conflicting assertions about whether its action is directed toward tubulin andlor toward microtubule-associated proteins (Kakiuchi, 1982; Kurnagai et al., 1982; Lee and Wolff, 1982) This protein catalyzes the covalent insertion of an ADP-ribosyl moiety derived from NAD+ into an unidentified site of tubulin, based on SDS-gel electrophoresis and peptide fragmentation patterns (Hawkins and Browning, 1982). Because only about 0.1% of the tubuli dimers appear to become modiied, the association with microtubule control remains tenuous This force-producingenzyme utilizes the free energy of ATP hydrolysis to power ciliary and flagellar motility (Gibbons, 1963). The extraction of flagellar or ciliary axonemes with high salt concentrations liberates the so-called outer arms which join microtubule doublets (Gibbons and Fronk, 1972). This observation has stimulated in vitro examination of the kinetics of MgATPP- hydrolysis and mode of vanadate inhibition (Anderson and Purich, 1982). Moreover, dynein binds to brain singlet tubules in a manner which reveals microtubule polarity (Haimo, 1982) This high-molecular-weight protein remains associated with microtubule protein during isolation from brain by the recycling method (Borisy et aL, 1975; Sloboda et al., 1976; Weingarten et al., 1975). See text for additional comments on this protein (Continued)

TABLE I1 (Codnued) Name or designation

This high-molecular-weight protein remains associated with microtubule protein during isolation from brain by the recyding method (Borisy et al., 1975;Sloboda ef al., 1976; Weingarten et al., 1975).HMW, undergoes cydic AMP-stimulated phosphorylation (Sloboda d al., 1975).See text for additional comments on this protein This protein or group of proteins of M, ranging from 28,000to 30,000has been isolated LMW protein(s) along with taxol-stabilized microtubules from brain gray matter (Vallee, 1982). No functional role has been established Neurofilament proteins This group of three proteins comprises the so-called 1O-nrn filaments, or intermediate filaments, which are part of neural cytoskeletal systems. Recent evidence suggests that ATP induces the formation of an associated complex between microtubules and neurofilaments (Runge d al., 1981b),but the role for this associated complex is not established This enzyme catalyzes the rephosphorylation of GDP to GTP, and may play a role in Nudeosidediphosphate b a s e regulating the assembly properties of tubulin by controlling the CTFGDP poise in vivo [SeeTerry and Punch (1982)for a review] Protein kinase (stimulated by 3',5'-cydic This enzyme phosphorylates HMWp (or MAP2) in a manner which affects the association of actin and microtubule networks (Sloboda et al., 1975;Pollard et a l , 1982) AMP) Protein k i ~ (stimulated ~ e by calcium ions) This enzyme phosphorylates tubulin in a calcium ion-activated process (Burke and DeLorenzo, 1981). The enzyme is very unstable and must be isolated rapidly after death from rat brain. The role of tubulin phosphorylation remains to be established HMW2 (or MAP2)

UI

o,

Relationship to Microtubules

Stable-tubule-only proteins (STOPS)

Tau protein

Tubulin-tyrosine ligase

Tyrosine-tubuli carboxypeptidase L

5

71K and 215K proteins

This group or complex of proteins may be isolated from cold-stable microtubules. Their ability to stabilize microtubules appears to be diminished by the combined action of calcium ion and calmodulin (Job d al., 1983).The role in controlling or affecting microtubule stability at physiological temperatures remains to be established This protein or group of proteins is characterized as four to five bands on SDS-plyacrylamide gel electropherograms in the range from 58,000to 65,000,and by their ability to recycle with tubulin and stimulate microtubule assembly. [See pp. 33-37 of the review by Kirschner (1978)for an excellent overview of this protein and its properties] This enzyme catalyzes the covalent insertion of a tyrosine at the C-terminal glutamate of the tubulin a subunit to effect the posttranslational synthesis of a peptide bond (Flavin ct al., 1982;Thompson, 1982).The role of this modification reaction remains to be established This enzyme removes C-terminal tyrosine residues from the a subunits of tubulin (Hallak ct al., 1977;Argarana et al., 1978;Martensen, 1982). The role of this modification reaction remains to be established These proteins can be isolated along with taxol-stabilized microtubules (VaUee, 1982), and they appear to be microtubule-associated proteins which are specific to differentiated neuroblastoma cells (Olmsted and Lyon, 1981).Their role(s) remain to be established

158

DANIEL L. PURICH AND DAVID KRISTOFFEKSON

proteins associated with recycled tubulin to various degrees. Fourth, MAP2 serves as a substrate for protein kinase action (Sloboda et al., 1975),and it may have important interactions with both actin and neurofilament networks (Pollard et al., 1982; Runge et al., 1981b). Finally, while MAPl and MAP2 can be separated (Kuznetsov et al., 1981), very little is known about MAPI. From the work of Vallee (1982), it is clear that MAP1 has a rather even distribution in both gray and white matter of the cerebral cortex, whereas MAP2 is far more abundant in gray matter. In this regard, the very recent development of monoclonal antibodies to MAP1 have permitted investigations of the subcellular locale of this protein (D. Asai, W. C. Thompson, C. F. Dresden, L. Wilson, and D. L. Purich, unpublished research). While these workers found diffuse nuclear staining in several cell types, the most intriguing result, obtained with fibroblasts, was that the actin stress fiber network, but not microtubules, could be visualized by indirect immunofluorescence methods using anti-MAPl. This raises the possibility that MAPl may be an actinassociated protein as well as a microtubule-associated protein. Even more striking is the observation (T. L. Karr, D. L. Purich, and B. M. Alberts, unpublished research) that these same anti-MAP1 monoclonal antibodies recognize a site localized in the kinetochore region of Drosophila KC cells. Punctate nuclear staining is most evident, and the distinctive one-to-one correspondence of labeled sites and chromosomes is distinctive during metaphase and anaphase. Aside from the labeling pattern, these experiments reveal that MAP1 has an antigenic site that is well conserved over organisms as diverse as arthropods and mammals. Altogether, there is much evidence linking the high-molecular-weight proteins to microtubules and demonstrating some degree of specificity. Nonetheless, lingering far behind such observations is an understanding of the molecular details. 111. NUCLEATION OR INITIATION OF MICROTUBULE ASSEMBLY Caspar and Klug (1962) made an important distinction between two fundamental types of assembly processes. True self-assembly was conceptualized as a series of reactions relying on the propensity of subunits to condense and form assembled structures strictly as a result of the information encoded in the architecture of the components. On the other hand, template-directed assembly may be considered as a process depending on the presence of a separate template that imparts structural constraints on the pathway for constructing the final assembled structure. True self-assembly is observed, for example, in the formation of many oligomeric proteins. Indeed, Friedman and Beychok (1979) have re-

MICROTUBULE ASSEMBLY

159

viewed the current status of efforts to define the subunit assembly and reconstitution pathways in multisubunit proteins, and all of the several dozen examples cited in their review represent true self-assembly. Polymeric species are also formed by true self-assembly, and the G-actin to Factin transition is an excellent example (Oosawa and Asakura, 1975). By contrast, there are strong indications that ribosomal RNA species play a central role in specifying the pathway to and the structure of ribosome particles. And it is interesting to note that the assembly of the tobacco mosaic virus (TMV) appears to be a two-step hybrid mechanism: the coat protein subunits first combine to form 34-subunit disks by true selfassembly from monomeric and trimeric components; then, the assembly of the virus particle is directed by single-stranded RNA at a specific site in the sequence of the polynucleotide. One characteristic of true self-assembly is the ability of the protomers to engage in nucleation, a series of cooperative reactions leading to the formation of polymerization nuclei. Once formed, such nuclei promote further assembly because the rate constants for the subsequent protomer addition steps favor elongation. These rate constants, of course, need not be identical because the first few steps beyond nucleation may be necessary to create the regular lattice for elongation reactions to become independent of polymer length. In this respect, the nuclei represent self-assembly intermediates with unfavorable equilibria leading to them and far more favorable equilibria beyond them. As noted earlier in this article, pure tubulin contains all of the structural information required to form microtubules, and there has been considerable interest in characterizing the initiation process. A. Theoretical Conriderations

The polymer self-assembly theory of Oosawa and Kasai (1962) provides valuable insights into the nature of the nucleation process. The polymerization nucleus is considered to form by the accretion of protomers, but the process is highly cooperative and unfavorable. Indeed, this is strongly suggested by the observation that thousands of actin or tubulin protomers are found in F-actin and microtubule structures; if nucleation of self-assembly were readily accomplished and highly favorable, the consequence would be that many more fibers of shorter polymer length would be observed. The Oosawa kinetic theory for nucleation permits one to obtain information about the size of the polymerization nucleus if two basic assumptions can be satisfied in the experimental system. First, the rate of nuclei formation is assumed to be proportional to the 20th power of the protomer concentration with io representing the number of protomers required to create the nucleus. Second, the treat-

160

DANIEL L. PURICH AND DAVID KRISTOFFERSON

ment deals only with the reaction period over which the rate of polymerization greatly exceeds the rate of protomer loss from the polymers (i.e., the initial stage of polymerization when the protomer concentration is the highest). Actually, Wegner and Engel (1975) extended the analysis by using the steady-state approximation to solve the rate equations without neglecting the depolymerization rate, and they have applied their extended theory to actin polymerization. Yet there is considerable evidence suggesting that microtubule structure is sufficiently more complex than the formation of the two-start F-actin helix, and the probability that such simplified theory applied to the nucleation of tubule formation is more remote. Nevertheless, the Oosawa theory has guided the design of many experiments aimed at defining the nucleation mechanism, and an acquaintance with this approach remains a means for meaningfully assessing the polymerization nucleation experiments. Oosawa and Kasai (1962) assumed that nucleation involved the formation of a higher order “closed” structure from an “open” structure with the same number of protomers. In particular, they dealt with the case in which the nucleation step was the conversion of an unstable linear trimer to a closed helical trimer, and the greater stability of the closed structure would be accountable in terms of additional interprotomer bonding. By assuming that the rate of nuclei formation was proportional to the third power of the protomer concentration, cI3, the following equation results:

dEp dt = (k+cl - k - ) /:Acl3 dt

=

&I -dt

where cp is the total concentration of protomer in the polymer form, k+ and li- are the assembly and disassembly rate constants, and A is yet another constant which incorporates the rate constants for nucleation. Basically, the integral in Eq. (1) calculates the polymer number concentration at time t, and the net rate of polymer formation, dc,/dt, is simply the number concentration times the rate of polymer formation for a The last part of the equality reflects the single polymer (i.e., k+cl - L). definition that polymer formation parallels the rate of loss of protomer concentration. Equation (1) is soluble only if one adds a further restriction that k+cl greatly exceeds 8 - , and under this condition Oosawa arrived at the following expression:

} = 3[(2/3)k+A]”*c0~~t

MICROTUBULE ASSEMBLY

161

Here, co is the total protein concentration, and a plot of polymer formed versus reaction time can be made by substituting decreasing values of cI (recalling that cp equals co minus cI) into Eq. (2) and solving for t. Oosawa and Asakura (1975) presented a more general solution for a nucleus composed of io protomers, and the cooperativity will depend on the number of protomers engaging in nucleation in the manner shown in the following expression:

Note that k+* is the forward rate constant for the nucleation event. If one chooses a particular value of cI/co (e.g., one-half), then the left-hand side of the equation can be held constant, and one may measure the time, t, required to obtain this particular cl/co ratio for different values of co. Indeed, Eq. (3) can be rearranged under these conditions to obtain the following form: t =

(constant)cii0’2

(4)

or log t = -(20/2) log co + log(constant)

(5) A plot of log t versus log CO, where t is the time required to achieve a final cl/co ratio, yields the number of protomers in the nucleus by determining the slope. Although Oosawa originally used a value of cl/coequal to one-half, we have carried out detailed computer simulation work and find that the k+cl % k- assumption introduces significant error in the estimation of io by the time that the reaction proceeds to the point that c I / co is at the value of one-half. Thus, one must actually work under conditions where a smaller cl/co value obtains. Aside from this practical constraint, the reader should recognize that the above theory is akin to the Hill Equation for cooperative ligand binding in the sense that the nucleation step is assumed to be of high reaction order without the significant accumulation of intermediates having fewer than io protomers. Another difficulty with applying this theory successfully is the need to establish that in the initial condition there is only protomer in solution. With microtubule protein, there are many cold-stable oligomeric polymorphs that might affect the experimental outcome. Wegner and Engel (1975) extended the Oosawa analysis by dropping the assumption of irreversibility in polymerization, and these workers applied a steady-state approximation. Their approach focuses on a dimeric nucleus, and there is no easy way to extend their theory to much more cooperative systems and still obtain tractable rate expressions.

162

DANIEL L. PURICH AND DAVID KRISTOFFERSON

Nonetheless, their steady-state solutions agreed well with those obtained by direct numerical integration of the rate laws for two.different sets of dimerization rate constants, and their analysis provides a rather satisfying view of the actin polymerization process. B. Experimental Findings In his original report detailing the conditions favoring in vitro reassembly of rat brain tubulin, Weisenberg (1972) suggested that microtubule reassembly is a spontaneous process which does not require the addition of nucleating centers. On the other hand, Borisy and Olmsted (1972) reported that tubule assembly from porcine brain homogenates clearly depended on the presence of particulate structures of higher molecular weight than the tubulin protomer. Indeed, extracts obtained by high-speed centrifugation formed few tubules although they were rich in 6 S tubulin. By electron microscopy, characteristically diskshaped structures were observed in the low-speed supernatant fluids, and they were found to disappear as the reassembly of microtubules progressed. The existence of such disks or ring structures and their disappearance during tubule formation do not, of course, establish a causal relationship between ring structures and the true nucleating tenters for microtubule reassembly. In the strictest sense, one must reserve the designation “nucleation center” for the most simple oligomeric structures which immediately promote elongation reactions to form microtubules. Nonetheless, the observations of Borisy and Olmsted (1972) did serve to stimulate a rather widespread interest in the role of ring structures in the assembly of microtubules. It is helpful to consider some of the possible fates of ring structures during microtubule assembly for the purpose of organizing what has become an imposing area of investigation with many seemingly conflicting reports. First, there is the possibility that ring structures serve as direct intermediates in microtubule assembly, much the same way that T M V protein disks stack to form rodlike particles. Second, rings may only serve as true nucleating centers to which tubulin protomers (and possibly oligomers) add to create microtubules, and one would presume that the disk acts as a template which controls the geometrical arrangement of the protomers during assembly. Third, the rings may break open into linear structures resembling protofilaments or they might open slightly to form “lock-washer” structures with a helical configuration. If protofilaments were formed, these could interact with each other laterally or bind protomers to form structures with an incipient helical nature favoring eventual closure to assume a microtubulelike arrangement. If lock-washer structures were formed, these might direct the

MICROTUBULE ASSEMBLY

163

helix formation by any of a number of pathways. Fourth, the possibility exists that the ring structure per se is not the true nucleating center but that oligomeric species issued from rings serve as the nucleating center or isomerize to form the actual nucleation centers. Finally, ring structures may only represent a tubulin polymorph whose formation is favored under certain assembly/disassembly conditions but without any function in the reassembly process. In this case, the ring structures would be a depot or reservoir of tubulin protomers rather than serving as an organizing center for tubule reassembly. The reader should note that nearly all of these possibilities have appealed to various investigators, and the summary in Table 111 describes some of the proposed pathways for microtubule polymerization based upon the possible fates of ring structures. Kirschner (1978) has reviewed much of the early literature concerning many of these alternative pathways, and the interested reader is referred to his fine analysis of the findings reported prior to 1977. We shall concentrate on the recent literature dealing with ring structures. Nonetheless, the views of Timasheff and Grisham (1980) should also be considered at this point in the discussion of the pathways for tubule assembly. First, they state that the various mechanistic proposals need not be mutually exclusive. Second, solvent composition,both in terms of ligands binding to tubulin and factors affecting the strength of interprotomer bonding, could influence the pathway of reassembly. And third, the ring structures, while interesting as structural isomers of oligomeric species, may have no direct role in assembly. Johnson and Borisy (1977) examined the lag and elongation phases associated with porcine brain microtubule assembly. The lag phase in the plot of turbidity (i.e., polymer weight concentration) versus time accounted for only 5-10% of the entire amplitude obtained upon completion of the polymerization process. By fitting the elongation phase to a single exponential process, these investigators came to the conclusion that the number concentration of tubules undergoing elongation remains relatively stable after the first few minutes following warming of cold-depolymerized microtubule protein. Furthermore, the constancy of microtubule number concentration during elongation has also been observed with bovine brain microtubule assembly. This was demonstrated directly by electron microscopy using samples fixed by dilution into microtubule stabilizing buffer under conditions which disfavor further nucleation (D. Kristofferson and D. L. Purich, unpublished findings). These and other findings of a similar nature strongly suggest that nucleation no longer occurs after the lag phase is over. The above findings also provide the basis for an interesting computation relating to the apparent reaction order for nucleation. We may consider the case where we assume that all of the protein is exclusively

164

DANIEL L. PURICH AND DAVID KRISTOFFERSON

TABLE 111 Summary of Proposed Mo&h for Microtubule Assembh from Depolymerized Microtubule Protein Olmsted el al. (1974) Single rings

Stacking

(540 A)

Stacked ring or Isomeriring, “tubule” (340 A) Adding 6 s protomers

’microtubules

Adding protofilaments and protomers

’microtubules

Small tubules” (250 A)

Elongated

Erickson (1974) Double rings (480 A)

Opening and aligning

Small sheets

Isomenzing

’of filaments

Small tubules (250

A)

Elongated

Kirschner and co-workers (Kirschner and Williams, 1974; Kirschner et al., 1975)

Rings

Opening and aligning

Curled bundles

’of protofilaments

Forming, sheets

Sheets Closing

t;uy:s

Adding protofilaments and protomers

Elongated

’microtubules

Jacobs et al. (1974) Double rings Adding of protomers to a Elongated template of inner ring tubules (480 A)

’

Borisy et al. (1976) Rings Opening and Curled ribbons with aligning ’5- and 8-start helices

h s of ring at ends

Elongated

’microtubules

-

Small tubules

Adding 6 S protomers

Elongated

’ microtubules

Oligomer model (Engelbroughs et at., 1980; Karr and Purich, 1980; Mandelkow et al., 1980)

Rings

Breaking down tooligomen

* Oligomers

Adding of other oligomers and/or protomers

Elongated

’microtubules

a The term “small tubules” is applied here to describe those structures having all of the characteristics of a microtubule.

protomeric at the very beginning of the experiments mentioned above. (Obviously, the presence of a substantial amount of ring structures in cold-depolymerized microtubule protein nullifies this assumption, but the kinetic argument is still persuasive.) Now then, after the polymerization reaction reaches 5- 10%of its maximal amplitude, the spontaneous formation of nucleation centers becomes sufficiently infrequent so as not to change the number concentration over the remaining course of elon-

MICROTUBULE ASSEMBLY

165

gation. Thus a reduction of the protomer concentration from about 20 to 18 phi reduces the apparent extent of nucleation by a factor of about 20, such that the polymer number concentration is constant to within 5 % throughout elongation. If nucleation were viewed as a one step cooperative event, then the rate of nucleation would be proportional to the ith power of the protomer concentration if i protomers cooperatively form the polymerization nucleus: i[Protomer] = [Nucleus]

(6)

The equilibrium constant for this process would be given as follows.

K

= [Pr~tomer]~/[Nucleus]

(7)

From this type of analysis, one would conclude that i must be approximately 28 for a 10% reduction in protomer to cause a 95% reduction in the nucleus concentration. This is a rather startling apparent reaction order even assuming infinite cooperativity between protomers. It is recalled that Hofrichter et al. (1974) found from a similar analysis of the rate of nucleation of human hemoglobin S (HbS) at 30°C that the apparent reaction order for the nucleation of HbS aggregation was about 32. Of course, such analyses are not fully justifiable because one may not assume ideality in the solution properties of biopolymers at high concentrations, particularly at 200 mg/ml in the case of hemoglobin. The computation for the case of tubulin polymerization does, nonetheless, emphasize that nucleation would be an especially cooperative event if only tubulin, and not ring structures, played the active role in nuclei formation. To gain a better perspective of the role of rings in nucleation, Engelbroughs et al. (1977) carried out a kinetic analysis with the porcine brain protein utilizing exposure of rings to 0.8 M potassium chloride as a means for dissociating these oligomeric structures. Apparently, after such treatment, the salt concentration can be reduced to 0.1 M without reformation of ring structures to any major extent, as inferred from the much slower rate of assembly of such diluted protein as compared with diluted protein not receiving the high-salt treatment. Furthermore, their studies revealed that only the untreated microtubule protein contributed to nucleation, whereas both treated and untreated microtubule protein participated in elongation (termed propagation in their paper). Engelbroughs et al. (1977) offered the interpretation that nucleation starts by the association of rings (probably broken open). Because the salt treatment obviates the need to fractionate the MAPSfrom the tubulin, these studies provided a means for analyzing the system at constant

166

DANIEL L. PURICH AND DAVID KRISTOFFERSON

MAPdtubulin ratios, and this fortifies the strength of their arguments. These results are also compatible with the view that nuclei for assembly issue forth from rings. Karr and Purich (1980) have also offered evidence that ring structures are themselves not microtubule assembly intermediates. Their analysis is based on the idea that ring structures are enthalpically stabilized whereas microtubule structures, particularly interprotofilament bonding domains, appear to be entropically stabilized. They attribute the characteristic drop in turbidity preceding the warm-induced assembly of cold-depolymerized microtubule protein to the rapid loss of ring structures. These workers estimated that a 1-2 mg/ml sample of whole microtubule protein would contain around lo-’ M ring structures, based upon the finding that nearly 40% of the total cold-depolymerized protein elutes with the ring fraction upon gel filtration. Yet, the microtubule number concentration during self-assembly is about M, and one would expect a much higher concentration of tubules if ring structures were fully active intermediates in self-assembly, especially because the free tubulin concentration would be in considerable excess over the critical microtubule protein concentration. Seeking another means for estimating the rate of ring breakdown upon warming of cold-depolymerized microtubule protein, Karr and Purich (1980) used 1 mM GDP to inhibit microtubule elongation and followed the time course of ring disintegration after warming samples to 30°C. By making the unproven assumption that the change in absorbance at 350 nm is a linear measure of the protein weight concentration in the ring fraction, they found that the loss in ring structure concentration followed an exponential decay (half-life = 73 seconds, at 0.84 mg/ml microtubule protein). Thus, ring dissociation appeared to be sufficiently long-lived to permit the production of smaller oligomers which could serve as initiating centers for elongation reactions. Corroborating evidence for this conclusion was provided in the studies of Terry and Purich (1979) who carried out pulse-chase measurements to demonstrate nearly complete equilibration of radiolabeled E-site nucleotide of tubulin warmed in the presence of unlabeled nucleotide. It should be noted, however, that the microtubule protein concentrations used in these experiments were in the range of 1-2 mg/ml; at higher concentrations, elongation may be sufficiently rapid so as to incorporate more tubulin oligomers into microtubule polymer. Pantaloni et al. (1981) investigated the mechanism of initiation of tubulin assembly, and the role of MAPS in the stabilization of microtubules. For their work they used microtubule protein carried through two cycles of assembly/disassembly, followed by treatment with 1 M sodium chloride for 1 hour, and subsequent centrifugation to remove the 70,000-, 168,000-, and 2 10,000-molecular weight proteins commonly

MICROTUBULE ASSEMBLY

167

called the neurofilament triplet. With ultracentrifugation experiments, they determined the fraction of microtubule protein present as ring structures as a function of temperature over the range from 0 to 37°C. They found approximately 60% was in the ring form over .the 0-20°C range with substantial dissociation to 6 S dimers occurring between 20 and 37°C. These investigators also studied the rate of exchange between tubulin dimers in the protomer and ring pools. They state that the measured apparent rate constant for the radioactivity exchange from protomer to ring forms at 0°C was 0.35/hour (half-life = 92-120 minutes). They modeled the exchange by the following scheme: %a

k

6 Tb27 + Tbi k-

(8)

where [Tbz,] Q [Tbz~]and the latter is a fully structured ring. From an Arrhenius plot of the rate of tubulin exchange, a linear relationship with an activation energy of about 12 kcal/mol was found to describe the kinetics as a function of temperature. They conclude that the half-life for such exchange is in the range of 1 minute at 37°C. These observations suggested that a study of the incorporation of tubulin from ring and protomer fractions into microtubules at 22°C was technically feasible, because the exchange between tubulin and rings is not too fast. In one such experiment, trace levels of '251-labeledtubulin were added at time zero to a solution of tubulin and rings at 4 mg/ml at O"C, and the radiotracer permitted them to follow the course of protomer entry into microtubules during warm-induced assembly. The results indicated that there was a preferential incorporation of unlabeled tubulin (i.e., protein from ring structures) during the early moments after warming to 22°C. Yet, at the full extent of assembly, the average specific activity of tubulin in the microtubules was about 80% of the radiospecific activity of the protomer pool at time zero. From their analysis, only about 8-10% of the tubulin incorporated during the first 2 minutes of assembly originated in the protomer pool, and the remaining 90% issued forth from the ring structures. The authors indicate that the situation at 37°C may be somewhat different because ring structures are substantially more labile, as demonstrated by Engelbroughs et al. (1980). Of all the published experiments dealing with the role of ring structures in tubule assembly, the studies of Mandelkow et al. (1980) are probably the most direct. These workers relied on the use of 1-A-wavelength X-rays from the DESY synchrotron at Hamburg, Germany to obtain time-resolved scattering data during cycles of assembly and disassembly controlled by temperature shifts between 4 and 36°C. Smallangle scattering theory served as the basis for obtaining structural information about the fraction of microtubule protein in various aggregated states of assembly. For the case of independent scattering from all parti-

168

DANIEL L. PURICH AND DAVID KRISTOFFERSON

cles in the solution, the theory relates the intensity of scattering to other parameters as follows:

where h is (4rsin 8)/A with 8 the angle of scattering and A the wavelength of X-radiation. The sum, z x k ( t ) is taken over all aggregates k; xk(t) is the fraction of tubulin promoters in an aggregate k; pk(t) represents the degree of polymerization (i.e., the number of protomers in a given aggregate); and ik(h) is the scattering function characteristic of an aggregate k normalized to unity at h = 0. Because x k ( t )is the fraction of subunits in a particular aggregate, Z xk(t) represents a conservation term. The ik (h) term contains structural information about each aggregate. Mandelkow et al. (1980) treat the rings as having apparent dimeters in solution of around 375 A and the tubules as having a mean diameter of about 350 A. The high X-ray flux available from the synchrotron allowed them to record sufficient intensity data over time frames of 15 seconds during the assembly and disassembly events, and the angular dependence could be evaluated over the range from 0 to about 30 mrad. They found that ring structures break down to small oligomers virtually completely within 1 minute following warming to 36°C. They indicate that such loss of ring structures leads to a condition wherein the main contribution to the scattering intensity can be interpreted in terms of a solution containing tubulin protomers and oligomers for the most part. These workers also followed the time course of cold-induced depolymerization, and they concluded that the depolymerization pathway is not simply the exact reverse of the assembly events. Thus, it would appear that microtubules are formed from tubulin oligomers of dimensions much smaller than ring structures and tubulin protomers.

IV. MICROTUBULE ELONGATION Once the polymerization nucleus has reacted with additional protomers to form a regular microtubule structure, the polymerization reaction enters the elongation phase. In the general case, the life span of nucleation may be very short or quite long, and the nucleation and elongation phases can appear as temporally separate or overlapping processes. There is even the possibility for continuous nucleation; although as indicated in the previous section, nucleation (or initiation) is quite cooperative and should diminish rapidly as the unpolymerized protomer pool falls in concentration. The reader will also recognize that

MICROTUBULE ASSEMBLY

169

elongation may be studied with preexisting microtubules and even with microtubule organizing censors such as flagellar axonemes. In this section, we treat the theoretical and experimental aspects of elongation. A. Theoretical Considerations Microtubule elongation may be conceptualized as the sequential addition of protomers to the growing points at the ends of the polymer. Several simplifying assumptions can be made for such a model: (1) that all of the growing points are identical, and thus are described by the same rate constants for protomer addition or loss from polymer ends; (2) that the addition of a protomer to a growing point creates a new growing point of identical reactivity, and thus the rate constants for addition or loss are independent of polymer length; (3) that continuous nucleation does not occur and the polymer number concentration is essentially constant; and (4) that other pathways for elongation (such as the addition of oligomeric species or a variety of addition sites) are negligible. In such a case, the elongation reactions can be written simply as follows: MTi + T b & MTi+l k k-

(10)

where MTi is a microtubule containing i protomers, Tb is a tubulin protomer, and k+ and k- are the corresponding rate constants for protomer addition and loss, respectively. The rate equation for this process is aklldt = k-m - k+mcl = (k-

- k+cl)m

(1 1)

where c1 is the concentration of free protomers, and m is the constant concentration of polymer ends. If one assumes that only a rather small fraction of the initial protein concentration (co) is depleted to create the nuclei during the nucleation phase, then c1 at the onset of elongation is virtually equal to CO. This also assumes that any preexisting oligomers (e.g., ring structures and protofilaments) comprise only a small part of the total protein at the start of the elongation phase. The rate equation given above describes a single exponential process, and under the constraints of these assumptions the following integrated rate law describes the time evolution of the elongation process: cl(t) = k J k +

+ [co - (k-/k+)]exp(-k+mt)

The equilibrium value of c1 is A J k + or KD,and the equilibrium constant should be independent of the total protein concentration in the polymerization reaction. Because all of the protein is considered to be in either the protomer or polymer forms, we may use the relation that cp =

170

DANIEL L. PURICH AND DAVID KRISTOFFERSON

co - cI,where cp is the total concentration of protomers in the polymeric state. This relationship can be substituted into Eq. (12) to yield a result for the time course of polymer formation: cp(t) =

- (k-/k+)][l - exp(-ktmt)]

(13) The quantity cp(t) can be measured experimentally by several techniques, but turbidimetric procedures are generally the preferred means (see Section IV,B). We may note that, if the polymer is to form, co must be greater than k-lk, . Thus, the Oosawa model predicts that there is a critical protomer concentration of k-lk, (or KD) below which polymer cannot exist (see Section V,A). The above theory can be extended to deal with other more complex cases. For example, the two ends of a biopolymer need not behave identically, and microtubules, as noted earlier, are helical polymers of asymmetric protomer units. Thus, two sets of on- and off-constants might be necessary. In other cases, such as in the polymerization of tubulin in the presence of tubulin-colchicine complex (Sternlicht and Ringel, 1979; Sternlicht et al., 1980), there may be the need to consider copolymerization possibilities. We believe that it is also appropriate to consider the kinetics of microtubule depolymerization in this section, because depolymerization is generally studied under conditions where one can probe the rate constants for elongation more directly. Indeed, depolymerization studies generally obviate many of the difficulties arising from overlapping of the nucleation and elongation phases during assembly. It is interesting to note that there are several methods for obtaining the elongation rate constants. The rate constant for polymerization, k, , can be determined from assembly kinetics data and a knowledge of the polymer number concentration, m, by plotting ln{[cp(t)- cp(eq)]/cp(eq)} versus time. The slope of this plot is (-ktm), as can be readily determined by rearranging Eq. (13). The value of m can be determined by microtubule length distribution measurements using electron microscopy (Kristofferson et al., 1982a) or dark-field microscopy with glutaraldehyde cross-linked tubules (Sternlicht and Ringel, 1979). In particular, the average polymer length is determined and subsequently divided into the total polymer length. The latter quantity is estimated from the amount of protein in the polymer form and the number of protomers per unit length of polymer. Once k, is evaluated, one may obtain k- with a knowledge of the critical concentration, KD. A variation on this theme is provided by utilizing seeded assembly with a known number of microtubule seeds prepared by mechanical shearing and measuring the initial rate of assembly at various concentrations of protomer. Alternatively, one may [cg

171

MICROTUBULE ASSEMBLY

resort to depolymerization experiments to get k- under conditions where the polymers are induced to disassemble irreversibly. This may be accomplished by rapidly dropping the temperature of the polymer solution or by rapidly reducing the protomer concentration to a value far below the prevailing critical concentration. The latter has the advantage that the rate constants refer to the temperature at which other tubule elongation reactions are most typically examined (Karr et al., 1980b; Purich et al., 1982a). The mechanism of irreversible disassembly may be written as follows: MT. LMT,,-~

- h-

1-

MT,,-*

(14) where the longest microtubule in solution contains n protomers, and the rate equations for the nth and ith steps are

&,ldt = -k-c,

etc.

dcildt = k-(ci+, - c i )

and

(for i < n) (15)

The solution to this set of differential equations is illustrated in the report by Kristofferson et al. (1980), and it gives a value for cn-, where n is defined above a n d j is an integer.

The polymer weight concentration can then be calculated by multiplying the polymer concentrations by the number of protomers per polymer and then by the protomer molecular weight. Thus we have n-m

w(t)= j = O

n-m

M W ( ~ - - ~ ) C= , - ~MW exp(-k-t)

C [(n - j )

j=O

i i=O

[(k-t)'-'/(j-i)!]c,0-i]

where W ( t ) is the polymer weight concentration as a function of time and MW is the protomer molecular weight (e.g., for tubulin, 100,000). Note that the series is cut off at a lower limit with polymer species c,, and cn-j reaches this value wheneverj = (n- m).This species is the minimum polymer length contributing significantly to polymer weight in the experimental histograms (see Section IV,B). The initial concentrations [c:-~] of the polymer species prior to the onset of disassembly can be approximated by electron microscopy of negatively stained microtubule samples (Kristofferson et aL, 1982a). An arbitrary value of k- can be chosen and then used in the last equation above with values for [ct-,] to calculate W ( t ) . By suitable scaling of this calculated depolymerization curve, plotted as W ( t )versus time, to fit the observed experimental depoly-

172

DANIEL L. PURICH AND DAVID KRISTOFFERSON

merization curve, one obtains the conversion factor by which the arbitrary k- can be multiplied to obtain the actual depolymerization rate constant. Karr et al. (1980a) applied this method to get k-, and this value fixes k+ with knowledge of KD. This theory clearly predicts that the shape of the polymer length distribution curve determines the shape of the time course of depolymerization. For example Kristofferson et al. (1980)were able to show that apparent first-order depolymerization kinetics arise from length distributions which are nearly exponential. It should also be noted that the above theory helps one to gain a better feeling for the time course of cytoskeleton or mitotic apparatus disassembly upon cooling cells to temperatures which destabilize microtubules and effect unidirectional depolymerization. Likewise, the linear depolymerization kinetic model could be applied to the disassembly of bacterial flagella, muscle and nonmuscle F-actin, tobacco mosaic virus, hemoglobin S fibers, and other linear polymers to elucidate important rate parameters and to test the sufficiency of the end-wise depolymerization assumption in such cases. The above theory allows one to estimate the off-rate constant for biopolymer disassembly, but it does not take into account the possibility of having distinctly different rate constants at each end of the polymer. Thus, one obtains the sum of the off-rate constants rather than the exact constants for each end. Another limitation is that one does not know how many growing points there are at each microtubule end, and from the discussion of the multiplicity of helical starts (Amos et al., 1976), the observed off-constant should represent the product of n (the number of growing points) and the actual microscopic rate constant for each of these growing points. As will be seen below, axoneme-promoted assembly of microtubules (Bergen and Borisy, 1980) may help to obviate the former limitation, but the latter requires further characterization of the true growing points.

B. Experimental Findings As noted earlier, microtubule elongation has been characterized largely with respect to the involvement of guanine nucleotides and the modes of drug inhibition of microtubule formation. There have also been a number of important studies on the influence of microtubuleassociated proteins and solution variables on the kinetics and thermodynamics of microtubule self-assembly. Of these, the characterization of the so-called mitotic spindle poisons has been particularly complex because of the variety of agents and the diversity of systems studied. For this reason, we shall concentrate on the other factors affecting the elongation process.

173

MICROTUBULE ASSEMBLY

1. Role of Guanine Nucleotides

Weisenberg’s finding that either GTP or ATP was essential for efficient assembly of microtubules (Weisenberg, 1972) stimulated great interest in the nature of nucleotide involvement in tubule elongation. Indeed, the elongation phase was probably the most readily explored aspect of the assembly process, particularly because the greatest part of the time course of turbidity development can be analyzed in terms of a first-order elongation process (Johnson and Borisy, 1977). The possibility that GTP hydrolysis plays an important role in elongation was probably first noted by Berry and Shelanski (1972) who found that GTP was cleaved to GDP and orthophosphate during the aggregation of tubulin protomers into paracrystals in the presence of vinblastine. The identification of two classes of guanine nucleotide binding sites (see Section II,A,3) led to a series of experiments aimed at determining the role(s) of the exchangeable and nonexchangeable guanine nucleotides in promoting assembly. Jacobs et al. (1974) proposed that both sites played a role in the elongation process on the basis of double-labeling experiments with [3H]GTP and [y9*P]GTPto follow the stoichiometry of nucleotide hydrolysis. They suggested that the N-site may become phosphorylated by added GTP and then undergo hydrolysis during the microtubule elongation process. Indeed, this conclusion at first appeared to be verified by the work of MacNeal and Purich (1977) who used acetate kinase and acetyL3*Pto label tightly bound guanine nucleotide which was assumed to correspond to the N-site. The conclusions from both of these studies were incorrect, however, as a result of the ineffectiveness of Sephadex G-25 gel filtration in freeing the exchangeable nucleotide site of bound GTP or GDP (Purich et d., 1982b). Kobayashi (1975) and Arai and Kaziro (1976) suggested that only the exchangeable nucleotide was hydrolyzed during elongation, and this conclusion was further substantiated in other investigations (Spiegelman et al., 1977; MacNeal and Purich, 1978a). The first comparison of the kinetics of E-site GTP hydrolysis and microtubule elongation was carried out by MacNeal and Purich (1978a). Their results indicated that the hydrolysis and elongation reactions were coupled because there was only a very slight lag in the hydrolytic process at the concentrations of microtubule protein employed. These data do not distinguish between two mechanistic alternatives: TbEGTP T~EGT+ P MT.

+ MT,

binding

during b binding step

T ~ E G T *PMT.

MTn+i + Pi

hydrolysis

MT.+I

(18)

+ Pi

(19)

174

DANIEL L. PURICH AND DAVID KRISTOFFERSON

[It should be noted that E-site GDP formed during the elongation steps is retained by the assembled microtubule in a nonexchangeable form (Weisenberg et al., 1976),and the above reactions are written to emphasize that Pi, but not GDP, is released during elongation (Weisenberg et al., 1976; MacNeal and Purich, 1978a).]While the latter mechanism had been considered the most appealing mechanistic pathway, it remained for Carlier and Pantaloni (198 1) to provide a more definitive demonstration of this two-step process. When microtubule protein concentrations are sufficiently high, the pathway with two consecutive first-order reactions becomes evident, and Carlier and Pantaloni (1981) obtained a rate constant of 0.25 min-’ for the hydrolysis of GTP in the TbEGTP MT, complex. These investigators state that their model is different from the one presented by Karr et al. (1979b) who first proposed that there is a boundary near the ends of microtubules which may be stabilized by the presence of TbGTP complex. Nonetheless, the number of TbCTP molecules at the ends of microtubules may be below the detection limits of currently available techniques, and it is possible to reconcile the TbCTP boundary hypothesis with the kinetic data of Carlier and Pantaloni (1981) in the following manner. If TbGTP binding to a microtubule end induces a conformational change in the penultimate TbCTP (which itself was previously the TbCTP boundary), then the penultimate TbCTP may assume a conformation akin to the other “interior” protomers, and become susceptible to hydrolysis. Nonetheless, the study by Carlier and Pantaloni (1981) provides many new insights into the coupling of hydrolysis and assembly. There is now little question that the assembly-induced GTPase is an activity carried out at the exchangeable nucleotide site, but the necessity of such hydrolysis in tubule elongation has been best addressed through the use of analogs of GTP. Although GMPP(CH2)P and GMPP(NH)P first appeared to be ineffective in promoting microtubule elongation, the sufficiency of these agents became evident during the mid-1970s (Penningroth et al., 1976; Arai and Kaziro, 1976; Weisenberg et al., 1976). Indeed, Penningroth et al. (1976) demonstrated that the methylene diphosphonate analog could support assembly for up to l l cycles of warm-induced assembly and cold-induced disassembly. The failure of earlier studies to demonstrate such action by GMPP(CH2)P and GMPP(NH)P may again be traced to the high affinity of tubulin interactions with both GTP and GDP. It is also interesting to note that MacNeal and Purich (1978b) examined the substitution-inert chromium(II1) complexes of GTP as promoters of tubule elongation. The rate of assembly with Cr(II1) GTP was found to be faster than that observed with MgCTP; however, from the use of Cr(II1) [14C]GTP,it was possible to

-

MICROTUBULE ASSEMBLY

175

demonstrate only partial hydrolysis of the scissile bond of GTP. Isolated microtubules contained 55% Cr(II1) GTP, 30% GDP, and 15% Cr(II1) GDP. Moreover, while these tubules had the usual morphology as judged by negative staining and electron microscopy, they were far more resistant to calcium ion-induced depolymerization. To evaluate further the role of GTP in the assembly process, Weisenberg and Deery (1976) used dilution-induced disassembly of GTP- and GMPP(NH)P-microtubules to explore the notion that GTP hydrolysis may be essential in processing the assembled tubules into a conformation essential for depolymerization at a later time. (See Section V,B for a discussion of their experiments and subsequent studies of the protomer-polymer equilibrium as influenced by hydrolyzable and nonhydrolyzable guanine nucleotides.) It is now evident that GMPP(NH)P-tubules are kinetically more resistant to dilution-induced disassembly than are GTP- and GMPP(CH2)P-tubules (Karr et al., 1979b); nonetheless, that GMPP(CH2)P-tubulesare readily susceptible to dilution-induced disassembly provides direct evidence against the notion that hydrolysis is essential for disassembly of microtubules. The action of GDP in the elongation reactions has drawn considerable interest in recent years. Partly, this reflects the keen interest in the energetics of microtubule assembly and in the potential roles of nucleotides in controlling microtubule interactions, but the overriding stimulus has been to resolve the seemingly contradictory findings reported from several laboratories. Indeed, conflicting hypotheses have been advanced to allow for several mechanisms of GDP action: (1) that the tubulin-GDP complex is capable of direct participation in microtubule elongation (Carlier and Pantaloni, 1978; Karr, et al., 1979b); (2) that only very limited polymerization is possible with GDP (Zackroff et al., 1980; Zeeberg and Caplow, 1981); and (3) that the tubulin-GDP complex is incompetent to serve in net elongation reactions (Jameson and Caplow, 1980). Evidence in favor of the direct participation of GDP was provided by Karr el al. (1979b) who observed that GDP inhibits the initiation but not the elongation phase of assembly. If GDP is added in the cold to microtubule protein and then warming is used to induce assembly, they observed the potent inhibitory effects of GDP. In contrast, if GDP is added only after warming of the microtubule protein samples, then elongation continues to occur with reasonably good efficiency. The shift from tubulin-GTP to tubulin-GDP upon the addition of a 10-fold molar excess of GDP over GTP resulted in about a 2.5-fold decrease in elongation rate. These results conflict with those of Weisenberg et al. (1976) who found that the amplitude of the polymerization depended largely on the extent of assembly prior to GDP addition. Lee et al. (1982)

176

DANIEL L. PURICH AND DAVID KRISTOFFERSON

reinvestigated this matter to learn whether the source of disagreement might be related to the protocol used in preparing the microtubule protein. Their experiments revealed that one may reproduce either of the above mentioned results, depending on the mode of microtubule protein isolation. Substantial differences in the assembly-disassembly properties were observed with three types of purification approaches: (1) isolation with glycerol present after hypotonic extraction of whole brain tissue; (2) isolation in the absence of glycerol after hypotonic extraction; and (3)isolation with glycerol after nearly isotonic extraction of cerebral cortical tissue. It would now appear that components found in the microtubule protein isolated by these approaches are in fact different, and more recent experiments have demonstrated that the neurofilament triplet of 68,000, 168,000, and 210,000 M, can impart partial stability to microtubules under conditions of GDP concentration which would otherwise lead to nearly complete depolymerization (Lee and Purich, unpublished findings). Moreover, the results of Lee et al. (1982) can also be interpreted as demonstrating that there may not be an allinclusive single mechanism for GDP participation; yet, conversion of GTP to GDP promptly aborts elongation, suggesting that GTP is needed for polymerization (or copolymerization) reactions. 2. Other Factors Affecting Microtubule Elongation Reactions Among the other agents which can promote microtubule elongation reactions are ATP and UTP. Although it was thought that the exchangeable nucleotide was somehow involved in the action of nucleosidediphosphate kinase (NDPK), several key questions remained unanswered until recently. Two possible mechanisms for NDPK action included (1) direct transphosphorylation of the tubulin-bound GDP to yield a tubulin. GTP complex capable of participation in assembly; and (2) a multistep pathway involving release of GDP from the tubulin exchangeable nucleotide site, conversion of unbound GDP to GTP, and rebinding of GTP to tubulin to form the primary assembly-reactive form, tubulin-GTP. The first pathway demands the ability of NDPK to recognize and act upon the tubulin-GDP complex, and such a sequence of steps would indicate the presence of a unique NDPK involved perhaps exclusively with microtubule control. As discussed by Terry and Purich (1979, 1982), the weight of the available evidence strongly favors the second pathway described above. Thus, to the extent that the concentrations of intracellular ATP, UTP, and other nucleoside tri- and diphosphates do exert an effect on the state of phosphorylation of the tubulin exchangeable nucleotide, there is every reason to anticipate that this reflects the generalized

MICROTUBULE ASSEMBLY

177

action of NDPK on the intracellular nucleotide pools and not on tubulin directly. Glycerol is another agent capable of promoting microtubule self-assembly, and this substance has been used in numerous laboratories to act as a “thermodynamic booster” through a nonspecific general thermodynamic solvent effect on the chemical potential of the tubulin protomer (Lee et al., 197813, 1979). These investigators stress that an agent can serve to drive a polymerization reaction without itself acting as a ligand for the protein undergoing polymerization. Thus, a particular protein can undergo self-association using differential exclusion of agents such as glycerol by the dissociated and associated states of a protein as the driving force for the reaction. Such preferential interaction terms also enter into equations relating reaction equilibrium constants to the concentration of ligands as shown in the linked-function treatment of Wyman (1964). 3. Kinetic Probes of Microtubule Elongation and Dzsclssembly Three basic approaches have been developed to examine the kinetic properties and mechanism of microtubule elongation reactions. The first, the rapid-dilution-induced disassembly approach, allows one to circumvent any kinetic complexity arising from the overlapping nucleation and elongation phases by concentrating on the reverse elongation reaction (i.e., disassembly of preexisting tubules). The second method involves the use of axonemes as templates for elongation, and it draws its advantage from the ready morphological distinction of the (+)- and (-)ends of the tubule, as described above. The third method relies on the use of preformed microtubules to measure the kinetics of elongation, most generally by monitoring turbidity changes. This method offers the advantage that one may limit and control the number of elongating species under conditions that minimize spontaneous nucleation. To analyze end-wise depolymerization by the rapid-dilution-induced disassembly route, one must satisfy several criteria: (1) the system must admit to treatment as a series of step-wise disassembly reactions at the polymer ends; (2) the off-rate constant should be independent of microtubule length during the course of depolymerization; (3) the on-rate must be zero, or nearly so, and this is achieved by working substantially below the critical concentration required for assembly; (3)turbidity must serve throughout the course of the measurement as a reliable means for estimating the remaining polymer weight concentration; (4) length distribution data, generally obtained by electron microscopy, must be available to define the initial reaction condition; and ( 5 ) computer simula-

178

DANIEL L. PURICH AND DAVID KRISTOFFERSON

tion techniques, based on such initial polymer length distribution data, should be of sufficient accuracy to allow comparisons of the theoretical and experimental depolymerization curves. Karr et al. (1980a) were the first to show that one could satisfy essentially all of these criteria to permit the examination of the sufficiency of the end-wise depolymerization model and the estimation of the rate constants for protomer release. The time course for dilution-induced depolymerization and the corresponding polymer length distribution prior to disassembly are shown in Fig. 5. By comparing the experimental data points with the theory line, one can readily note the excellent agreement. There are several additional points of interest. First, the only parameters used to fit the equation are the initial microtubule length distributions and a single value for the off-rate constant. Second, the model fits over the entire disassembly process, showing that the depolymerization process is uniform throughout and that Berne’s theory that turbidity is an adequate measure of polymer weight concentration is rather well satisfied over this range (Berne, 1974). Third, from the inset to this figure one can observe that the depolymerization process is essentially exponential over the first 90% of the amplitude; from theoretical considerations, Kristofferson et al. (1980) demonstrated that distributions that resemble an exponential will yield depolymerization time courses that are also somewhat exponential. Fourth, these data were obtained with MAP-depleted microtubule protein, and such good agreement is not observed with MAP-rich microtubule protein. Fifth, one finds that this same approach can be used to characterize cold- and calcium-induced depolymerization so long as adequate care is taken to ensure that the temperature and calcium concentration jumps are sufficiently rapid relative to the time scale of the depolymerization (Karr et al., 1980a,b).In this regard, the earlier investigations by Johnson and Borisy (1977) focused on demonstrating the end-wise depolymerization model using cold-induced disassembly, but their study was of only limited success as a result of relatively slow temperature jumps and the use of an approximate model. Finally, Karr et al. (1980a) found that mechanical shearing could be utilized to demonstrate a nearly quantitative rate dependence on polymer number concentration (see Table IV). The second approach for studying microtubule assembly and disassembly by use of axonemes was developed by L. Bergen and G. G. Borisy (Bergen and Borisy, 1980; Borisy and Bergen, 1982).This approach has several attractive features. First, the polarity of microtubule elongation and the kinetics of such polar growth can be visualized through the use of electron microscopy, and this approach might even be extended to solution kinetic determinations using the dark field approach described

179

MICROTUBULE ASSEMBLY

t ""., 2

4

-<6

8 10 limo

-12

14

16

18

2 Average Longth ~ 5 . p2 m

15

30

45

MicrotubuleLength (yn)

FIG.5. Kinetics of brain microtubule depolymerization following rapid dilution. (A) Time course of the disassembly reaction with experimental data represented by the data points and the theoretical progress curve indicated by the solid line. (The inset to A shows that the process can be fitted to a simple decaying exponential for part of the depolymerization reaction.) (B) Microtubule length distribution for the sample subjected to rapid dilution in A. (Reproduced from Karr ct 01. (1980).J. Bid. Chm. 455,8560-8566.)

180

DANIEL L. PURICH AND DAVID KRISTOFFERSON

TABLE IV Rafe Constants and Polymer Number Concentration Dcpmdmce for Dilution-. Cold-, and Calcium Ion-Induced Microtubulc Disassemblya

Protocol Rapid dilution Cold Calcium ion

Increase in tubule Increase in depolymeri- Off-rate number zation constant' concenrate (sec-I) tration 3.12

-

3.18

2.36

2.27

-

1 I3 166 860

On-rate constantc (M-'

Sec-1)

Reference

(2 X 107)ln Karr et al. (1980a) (3 X 10')ln Karr et al. (1980a) Karr et al. (1980b)

Values apply to recycled microtubule protein that contains about 75%tubulin and 25% microtubule-associated proteins by weight. 'The off-rate constant is the value corresponding to the sum of all sites capable of addition or loss of protomers on both ends of the microtubule. The value of the on-rate constant for a particular growth site will depend on n, the number of growing points on both ends of a tubule. For example, if the tubule is treated as a three-start helix with each helix starting point serving as a distinct growth site, then n will be six, and the on-rate constant will be about 3 x lo6 M - l sec-'. Such a value would be about a factor of 10 below the theoretical limit for diffusion-controlledprocesses involving macromolecules.

by Summers and Kirschner (1979) for untethered brain microtubules. Likewise, this approach can be used to examine the polarity of drug effects more directly. Second, the axoneme can serve as an effective organizing center, and Scheele et al. (1980) have shown that the A-subfiber and central pair microtubules, which consist of 13 protofilaments, nucleate the formation of microtubules that also have the identical protofilament count. Greater than 90% of the tubules were found to have 13 protofilaments, whereas 14 protofilaments appear to arise preferentially in self-nucleated brain microtubule self-assembly. Finally, the technique is quite conservative with regard to the amount of sample required, and a single experimental point requires only 10-20 pl. The basic approach with the axoneme-based analysis is to combine tubulin and axonema fractions and then to quench the reaction with glutaraldehyde. Samples of sufficient axoneme concentration are then added directly onto Formvar-coated sample grids for staining and electron microscopy. In some cases where the axoneme count is too low, samples may be sedimented onto grids by the method of Gould and Borisy (1977). With the methodology perfected by Borisy and Bergen (1982) samples can be taken as frequently as every 20 seconds, and the

MICROTUBULE ASSEMBLY

181

corresponding length determinations use about 2n length measurements per point to achieve adequate counting statistics. The utility of this approach is illustrated by the estimated rate constants and formation constants presented in Table V. Indeed, it is noteworthy that these data provide a clear indication of polar reaction kinetics. The off-rate at the (+)-end is actually higher than the corresponding rate at the (-)-end; nonetheless, the (+)-end is more stable than the opposite end on the basis of the observed formation constants. From the values for the equilibrium constants at each end, we estimate that the free energy difference between the two ends is only a fraction of a kcal/mol, suggesting that the differential coupling of GTP hydrolysis to assembly/disassembly events at the two ends is really rather small compared to the 7-8 kcal/ mol value for the free energy of GTP phosphoanhydride bond hydrolysis (see Section VII on the polymer-protomer flux at steady state). One might ask why the axoneme-based protocol has not superceded the usage of rapid-dilution methods. There are several compelling reasons for further development and testing of both techniques, and we offer the following discussion in an effort to stimulate further consideration of this issue. First, the off-rate constants determined by both methods do not agree. The Bergen and Borisy (1980) report yields an average off-rate constant of 12 sec-', whereas values that are 10-fold greater have been obtained from both initial-rate and full time-course analyses of rapid dilution data (Karr el al., 1980a). Likewise, Zeeberg et al. (1980) applied the initial rate method with porcine brain tubulin to obtain a value of 119 sec-' which is only slightly higher than those reported by Karr et al. (1980a). Second, the need to suppress spontaneous nucleation in the axoneme-based method requires the absence of microtubule-associated proteins. In this regard, Cote and Borisy (1981) suggested that the disagreement between their new rate parameters and those of Bergen and Borisy (1980) might be explained on the basis of this factor. Nonetheless, MAPs appear to have a stabilizing effect on the later stages of dilution-induced disassembly (Karr et al., 1980a), and the discrepancy between the off-constants cannot be attributed to the effects of MAPs. Third, the axoneme-based approach leads to the analysis of tubules with a homologous A-type lattice, whereas reassembled brain tubules are of the mixed lattice type. While the use of 13-protofilament tubules might be an advantage, there is as yet no evidence to show that the same lattice with respect to homologous versus heterologous interactions is attained with both cytoskeletal and ciliary tubules in viva In this regard, both methods may be deficient in providing a representation of the dynamic properties of brain cytoskeletal tubules. Finally, the use of axonemes in

182

DANIEL L. PURICH A N D DAVID KRISTOFFERSON

other experiments such as “treadmilling” measurements (see Section VII) is not feasible because the polymer number concentration is far less than that required in radiometric measurements. The third method for evaluating the kinetics of elongation takes advantage of the development of conditions used to suppress spontaneous nucleation while still allowing the use of preformed microtubules as “seeds” for further assembly. There are several approaches for minimizing changes in tubule number concentration resulting from nucleation, and these include removal of microtubule-associated proteins as ring-MAP complexes by centrifugation or by the addition of agents such as heparin sulfate to inhibit nucleation. Until quite recently, the experiments of Engelbroughs et al. (1977) could be regarded as definitive in their demonstration that the observed rate of elongation was a linear function of the concentration of preformed tubule seeds added to a solution of microtubule protein. Nonetheless, Carlier (1982) demonstrated that the elongation rate strongly deviates from a linear dependence on microtubule seed concentration, especially at much higher levels of tubule seeds. The limiting rate constant for elongation was found to be independent of the concentration of elongation sites, and Carlier (1982) proposes that a conformational isomerization step involving the tubulin protomer may be the cause of the nonlinear dependence of rate on microtubule number concentration. Although this behavior appears to be most obvious at very high levels of tubule seeds, the existence of a preassembly conformation, one requiring further conformational change to become polymerization competent, may also be of’some regulatory advantage. Those solution factors that are able to shift the poise of this newly observed equilibrium could, for instance, serve to regulate the elongation rate and affect the values of the elongation rate constants derived by other kinetic experiments. In particular, the practice of calculating the true on-rate from the ratio of k,b,/microtubule number concentration could lead to error under other experimental conditions.

V. PROTOMER-POLYMER EQUILIBRIA AND CRITICAL CONCENTRATION BEHAVIOR After elongation has proceeded for a sufficient period of time, the protomer and polymer will reach concentrations corresponding to the equilibrium constant KD,which from our earlier discussion may be represented as k - l k + . This constant represents the critical concentration for polymer assembly, and protomer concentrations which fall below the magnitude of this constant will not spontaneously assemble. In this sec-

183

MICROTUBULE ASSEMBLY

tion, we treat the theory for such protomer-polymer equilibria and some of the experiments aimed at defining these parameters. A. Theoretical Considerations Following the general approach of Oosawa and Kasai (1962), one may assume that the following set of equilibria define the polymerization scheme of tubulin protomers: Tbl + Tbl = Tbi (204 Tbn + Tbl = Tbs (20b) Tb, + T b l =

T b d

where Tbi represents a microtubule containing i protomers. Let ci be the concentration of Tbi, and let K be the equilibrium constant for every step (i.e., K is independent of polymer length). We then have K = c2/c12

c2 = K c I 2 =

or

K-I(KC')~

Likewise,

K = CS/(CIC~) or cs = Kclc2 = K-'(Kc# (2 1b) In general, the result for ci is K-'(Kcl)', and the total protein concentration is simply m

m

c0= 2 ici = K-' 2 i(~c~)i i= 1 i= 1

The above summation may be evaluated using the well-known formula for a geometric series, m

2 xi = x/(l - x) i= I

(1x1 < 1)

Taking the derivative of both sides and multiplying by x yields:

c ix' m

= x/(l

- x)2

(24)

i= 1

This expression may be used to evaluate Eq. (22) with the following result : = K-'(Kci)/(l

-

KC^)^ = cl/(l

-

KC^)^

(25) The value of co can increase from zero without bound, but c1 cannot exceed K - . It is a useful exercise to plot c1 versus co for any arbitrary value of K-' CO

184

DANIEL L. PURICH AND DAVID KRISTOFFERSON

(e.g., 5). By choosing cI values approaching K-l as a limit (in this case, 0.2), one calculates CO. Plotting these data reveals that although c1 approaches the predicted limit of K-' or KD, there is no sharp transition point above which protomer concentrations in excess of the magnitude of KD will be converted to polymer. In other words, the assumption of a single equilibrium constant for every step does not lead to critical concentration behavior. The failure of the above process to show a discrete or sharp critical concentration behavior can be obviated by having one or more of the initial equilibrium constants much smaller than K . For example, one may change the dimerization reaction equilibrium constant to KN , where KN is much less than K, the constant for all remaining polymerization steps. Now ci will be defined by the following expression: Ci

= ( K N / P ) [ K C ~ ] ' ( i 2 2)

(26)

We can obtain an equation for co similar to Eq. (22), except that Eq. (26) cannot be used for i = 1. Therefore, we rewrite the expression for co as follows:

To evaluate the sum, Eq. (27) must be modified by adding and subtracting ( K N / K * ) ( K Cto~the ) right-hand side. Then, by using Eq. (24) we have

If K N is much less than K, then (1 - K N / K )will be approximately unity, and Eq. (28) simplifies to co = c1

c1 + KN K (1 - Kc1)2

Plotting c1 versus co as suggested above (e.g., letting K equal 5, c1 equal 0.2, and KN equal it will be noted that when co is less than K-I or K D ,co = cl. At co equals KD, an abrupt transition occurs, and any further increases in co leave c1 unchanged at the value of KD (i.e., any protomer in excess of the critical concentration, KD , becomes incorporated into polymer). For co less than K D , the concentration of polymer will be zero.

MICROTUBULE ASSEMBLY

185

It should be noted that the value of K N does not change the value of the critical concentration; it only affects the sharpness of the protomerpolymer transition as co exceeds K D . B . Experimental Findings The usual procedure for determining the critical concentration for microtubule protein involves the measurement of the polymer weight concentration as a function of the total protein concentration. By extrapolating the experimental data to zero polymer weight concentration, the value of co corresponding to the intercept is taken to be KD .The polymer weight concentration, cp, can be determined by a variety of methods including turbidity and sedimentation of polymer (Timasheff, 1981). Gaskin et al. (1974) were among the first to apply such methods to study microtubules, relying on the theory of Berne (1974) to justify the linear relationship between turbidity and the polymer weight concentration of rods of lengths greater than the wavelength of light used in the experiments. The relation between cp and co is given by a conservation equation (cp = co - cl). At protomer-polymer equilibrium, we have

cp(eq) = CO - KD (30) As noted above, turbidity and polymer weight concentration can be directly related (cp = a ~where , T is the turbidity), and the proportionality constant may be determined experimentally (cf. Zackroff et al., 1980). Microtubule protein preparations, however, usually contain a fraction of protein that does not contribute to polymer formation, and the most likely interpretation is that this fraction is composed of assembly-incompetent tubulin and nontubulin protein contaminants (Gaskin et al., 1974; Zackroff et al., 1980). Note that Eq. (30) is based on the assumption that GO represents active, polymerization-competent protomer. If only a fraction y , less than one, is active, this equation must be corrected to give cp = ar = yco - K D

(31)

or = ( Y / ~ ) co KD/a

(32)

Note that denatured protein results in a lower-value of y and a correspondingly lower slope, y l a , in plots of turbidity versus total protein concentration. The intercept on the co axis equals KDly (see Fig. 6). Unfortunately, investigators tend to ignore the need for the above correction, instead equating K Ddirectly to the co intercept value. Yet, y can

186

DANIEL L. PURICH AND DAVID KRISTOFFERSON

$ A /

v

I

i/ d I

I

&-

‘Slope.

At Slope

1-Y

I

Total ProteinConc.,C.

z I

Total ProteinConc.,C.

FIG. 6. Determination of the critical protein concentration. (A) Plot of protein in the supernatant fluid after quantitatively sedimenting polymer from a polymerized solution of tubules and tubulin at steady state. The critical concentration, K D ,is determined from the value of they axis intercept, and the fraction of active protein, y, from the slope. (B) The conventionally used experimental method for estimating the critical concentration. Note that the x axis intercept is actually KDIy, instead of Kn. Interpretation of the slope from such plots requires knowledge of the ratio of polymer weight concentration to turbidity (g;iven here as a).Data from experiments such as those in A may be used in conjunction with this plot to obtain the critical concentration, and this can serve as an internal test for self-consistencyof the data.

fall into the range of 0.7 to 0.8 in typical experiments, and this can lead to errors amounting to 25-43% in the determination of KD. Another note of caution regarding the unpolymerized protein fraction is necessary, Because the co axis intercept includes the inactive protein fraction, the concentration of protein in the supernatant fluid after centrifugation will not be independent of the total protein concentration. Instead, we have Csupernatant

= co

(CO

< KD/Y)

= KD + (1 - Y ) C O

(33)

KD/Y) (34) where cSuFrnatant is the protein concentration in the supernatant fluid. This last equation provides a direct means for determining K D and y from a plot of cSupernatant versus CO. Once the value of y is determined, a Csupernatant

(CO

2

MICROTUBULE ASSEMBLY

187

can be estimated from the slope of a plot of turbidity versus total protein concentration. The co intercept of this plot at r equal to zero will be KD/y; thus, KD can be determined without determining a.This turbidimetric determination of KD should agree with the value obtained by the aforementioned centrifugation-pelleting protocol, provided that turbidity is an adequate measure of polymer weight concentration. Only a few investigators have rigorously analyzed their critical concentration determinations in the manner described above, and the reader should consult the reports by Sternlicht and Ringel (1979) and Zackroff el al. (1980) for examples of such a careful treatment. Another shortcoming in the current determinations of the critical protomer concentration in the case of tubulin polymerization is that there is no universally accepted “standard state” of solution variables or microtubule protein composition to which one may reference measurements of the KD values. Many investigators tend to utilize solution conditions that are somewhat similar (e.g., 1 mM GTP, 1 mM magnesium ion, 1 mM EGTA, and a “Good‘s”buffer, usually MES or PIPES, at about pH 6.8), but detailed comparisons are limited by the lack of standardized conditions. Further confusion arises from the widespread usage of different purification methods for obtaining tubulin, and the presence of MAPs and contaminating enzymes which may hydrolyze GTP pose particular problems. With so-called Shelanski microtubule protein, prepared by a modification of the in vitro reassembly protocol of Weisenberg (Shelanski et al., 1973), the critical concentration is typically 0.2 to 0.3 mg/ml. With microtubule protein isolated by a sucrose extraction method (Karr et al., 1979a), more MAPl and more MAPpare present in the protein, and the critical concentration is in the 0.08 to 0.1 mg/ml range. Purified tubulin in 3.4 M glycerol forms microtubules in the absence of MAPs with a critical protomer concentration of about 0.8 to 1.0 mg/ml (Lee and Timasheff, 1975). Because very few investigators have corrected for assembly-incompetent protein as described above, these values of KD must be viewed as apparent equilibrium constants. Critical concentration data have been useful in probing the effects of GTP hydrolysis on the assembly and disassembly properties of tubulin. Nonetheless, there is still controversy concerning the interactions of nonhydrolyzable nucleotides in the polymerization reactions. For example, Weisenberg and Deery (1976) examined the susceptibility of GTP tubules and GMPP(NH)P to dilution-induced depolymerization. They reasoned that the hydrolysis of GTP might induce a conformational change essential for subsequent depolymerization. In other words, the hydrolysis step might yield free energy, part of which could be stored as conformational energy in the microtubule lattice structure and released

188

DANIEL L. PURICH AND DAVID KRISTOFFERSON

in a subsequent disassembly event. In their experiments, microtubules were formed in the presence of GTP or GMPP(NH)P and then diluted to various total protein concentrations to examine the degree of polymerization after 5 minutes. They observed that the two types of tubules behaved differently with regard to dilution-induced disassembly. A positive co axis intercept was obtained with GTP and this corresponded to generally observed critical concentration behavior; yet, with GMPP(NH)P-tubules, a negative intercept was obtained. This type of behavior suggested that hydrolysis is essential for depolymerization. On the other hand, the studies of Penningroth et al. (1976) indicated that tubule assembly/disassembly,as controlled by thermally induced assembly at 37°C and depolymerization at 4"C, was readily accomplished with either GTP or GMPP(CH2)P. Karr et al. (197913) reinvestigated this unusual behavior with GTP, GMPP(NH)P, and GMPP(CH2)P,and recognized that one should distinguish between kinetically and thermodynamically determined behavior by allowing microtubule samples longer periods to equilibrate. Indeed, the results of Weisenberg and Deery ( 1976) were reproduced in 5-minute incubation experiments, except that GMPP(CH2)P did not behave like the imido diphosphonate analog. GMP(CH2)P-tubules exhibited a critical concentration value slightly above that obtained with GTP-tubules upon isothermal dilution. Interestingly, Karr et al. (1979b) found that dilution-induced disassembly of GMPP(NH)P-microtubules required approximately 45 minutes, and that the critical concentrations for both GTP and the imido diphosphonate analog were nearly identical if sufficient time was allowed for equilibration. From the earlier discussion about corrections for assembly-incompetent protomer and other proteins, the reader should note that both the slopes and intercepts for the GTP- and GMPP(NH)Ptubules are practically identical, suggesting that the behavior with regard to the apparent critical concentration value and the need for corrections to obtain the true KD values are surprisingly similar. These data thus suggest that there is no remarkable difference imparted by nucleotide hydrolysis in terms of the protomer-polymer equilibrium; however, kinetic differences do exist, and these might be of importance in adjusting the rapidity of effector-induced disassembly of tubules in living systems. At the same time, there are instances in which critical concentration determinations still do not offer the experimenter the opportunity to reach a definitive conclusion about important aspects of protomer-polymer interactions. The lack of agreement about the polymerization properties of the Tb * GDP complex is an excellent example. Several investigators have claimed that the tubulin GDP complex can elongate microtubules (Carlier and Pantaloni, 1978; Karr et al., 1979b); others,

-

189

MICROTUBULE ASSEMBLY

however, have suggested that this complex can elongate tubules only to a limited extent (Zackroff et al., 1980; Zeeberg and Caplow, 1981); even so, others suggested that no elongation was possible with the tubulin * GDP complex (Jameson and Caplow, 1980). In theory, critical concentration measurements provide an opportunity to distinguish between the first and third possibilities. Unfortunately, however, if Tb * GDP polymerizes much more slowly than Tb * GTP, a distinction between this case and the total inability to elongate becomes difficult to achieve by critical concentration measurements alone. Indeed, inherent errors in the determination of critical concentration values prohibit the straightforward discrimination between the above mechanistic alternatives (Kristofferson et al., 1982b). If Tb GDP cannot elongate tubules, the presence of GDP will tend to inactivate that portion of tubulin to which it is liganded. In this case, let Cz represent the critical concentration when only GTP is present (i.e., all the tubulin protomers are present as Tb GTP). The observed critical concentration, designated as CFh, in the presence of both GTP and GDP will be given by

-

-

+

CFbs = C," [TbGDP]

(35) This means that the total steady-state' protomer concentration, C:bs, includes the amount necessary to equilibrate the polymers with Tb * GTP (i.e., C,") plus the amount sequestered in the inactive form of Tb GDP. If the nucleotide dissociation constants, KCDp and KGTp, are known, then [Tb * GDP] is determined.

-

[Tb * GDP] = (S/e)[Tb * GTP]

(36) where S and E are defined to be the reduced concentrations [GDP]/KGDP and [ G T P ] / K G ~respectively. ~, Combining the last two expressions yields the relationship = C,"[1

+ (NE)]

(37) because Cfis defined as Tb GTP. For the case of Tb * GDP elongation, the CEb"value will be less at any particular ( W E )ratio than that predicted by the last equation. While Eq. (36) still holds, we find that [Tb * GTP] < Cz by the following argument. At steady-state assembly, the opposing reaction rates are equal, and c:bs

-

kl[Tb * GTP]

+ k2[Tb

GDP] = k3

(38) where kl and k2 are the on-rate constants for Tb GTP and Tb * GDP and k3 is the off-rate constant. As Tb * GTP is hydrolyzed to Tb . GDP upon assembly, only one constant, kS , should be necessary to describe the off process. [This, of course, presumes that there is no significant *

190

DANIEL L. PURICH AND DAVID KRISTOFFERSON

-

boundary of T b GTP protomers at the ends of microtubules as first proposed by Karr et al. (1979b).]Substituting Eq. (36) for Tb GDP into the last equation yields

-

[Tb . GTP] = k 3 / [ k l

+ kp(6/~)]

(39)

Likewise, the total protomer concentration becomes C:b* = [Tb * GTP] = {k3/[k,

+ [Tb - GDP] = [Tb - GTP][l + (NE)]

+ k2(6/E)]} [ l + ( 6 / E ) ]

(40)

Note, if GDP is absent, C:bg becomes Cf. For nonzero values of ( 6 / ~ ) the , term in braces in Eq. (40) is less than Cf. Comparing Eqs. (39) and (40) for the inactive and active T b * GDP models, one concludes that C:bs in the former is always greater than the same term in the latter whenever GDP is present. Thus, one may in principle distinguish between the two models. Nonetheless, such clear-cut distinctions are yet to be achieved in practice, and this method of distinguishing these mechanistic alternatives is not reliable, given the current experimental accuracy. Finally, one should recognize that determinations of the critical concentration depend wholly on the validity of the equilibrium or steadystate assumptions. If a stable end point for protomer-polymer coexistence is not attained, then kinetic factors affect the observed behavior. With the well recognized tendency of tubulin to lose its ability to engage in assembly reactions upon storage even at low temperature, and with the presence of various nucleotide hydrolases and transphosphorylases in microtubule protein, such kinetic effects are a serious problem.

VI. MICROTUBULE LENGTHREDISTRIBUTION Polymer length redistribution is the last phase of the assembly process as viewed by the Oosawa theory. To understand the relationship of polymer length redistribution to the other polymerization events, it is helpful to recognize that the in. vitro assembly process is characterized initially by its kinetically controlled nature and subsequently by what are thermodynamically dominated interactions. Polymerization is usually studied by the use of turbidity, viscosity, or centrifugal pelleting procedures (Timasheff, 198l), and the process of polymer length redistribution occurs after the attainment of a stable polymer-protomer equilibrium which is evidenced as a time-independent plateau. At the beginning of the assembly process, we have noted in Section I11 that there are a number of tubulin polymorphs (e.g., rings, small oligomers, ribbons, and protofilaments) in addition to tubulin protomers, and the

MICROTUBULE ASSEMBLY

191

assembly process can be greatly affected by the kinetics of both their breakdown to tubulin protomers and their interconversion reactions. The distribution of microtubules with respect to polymer length at the onset of the polymer-protomer equilibrium typically has a pronounced peak (see Fig. 7). The most reasonable and widely accepted explanation for this type of distribution is that the initiation phase of polymerization is confined to the short period following warming microtubule protein solutions to induce assembly. If the formation of nuclei occurs only during that short lag period, then the nuclei will continue to grow at similar rates because they are all experiencing similar elongation reaction conditions. The polymer length redistribution reactions provide a mechanism by which the kinetically established length distribution can evolve in time to the thermodynamically most stable condition. In this respect, it is worthwhile to consider the theory and the modest experimental information about the dynamics of the redistribution events. The equilibrium distribution of polymer length can be determined by recalling [see Eqs. (25) and (29)] that Kcl asymptotically approaches unity from below as co approaches infinity. The equilibrium between MT, and MTi+l is K = (ci+d/(Cl)(ci) or K C I= ( c i + l ) / ( c i ) (41) Thus, ci+l/c,will approach the above stated limit. In practice, asymptotic convergence is rapid, and Kcl is indistinguishable from one; thus, c,+~ equals ci at equilibrium. This would mean that the equilibrium length distribution will be uniform rather than peaked as shown in Fig. 7.

-C.

. C

J

FIG. 7. Steady-state microtubule length distribution under idealized conditions. The concentrationsof microtubule species at the dip in the distribution,at the maximum, and at the longest point are indicated by GI, c., and c,,, respectively.[Reproduced from Kristofferson and Purich (1981). Arch. B i o c h . Blophys. 411, 222-226.1

192

DANIEL L. PURICH AND DAVID KRISTOFFERSON

Because Kcl is, however, slightly less than unity, and even though no measurable difference is discernible between the concentrations of two adjacent polymer species (e.g., MT, and MT,+l),an overall exponential decline should be observed over a sufficiently broad range of polymer length. This is clearly evident when, for ci+llci equal to A (where A is K c l ) , one considers the case for A slightly less than unity. From the model for Eq. (26), we then obtain

where A equals e-" and A < 1. Equation (42) demonstrates mathematically that the thermodynamic distribution can be expressed as a decaying exponential. Oosawa and Asakura (1975) likened the polymer length redistribution reactions to a diffusional process. By this, they meant that the kinetically established length distribution gradually broadens and collapses to the equilibrium distribution via a random or diffusionlike redistribution of protomers among the polymers in solution. The driving force for the broadening phenomenon is the concentration gradient among polymer species as represented by the peak in the kinetic distribution. To appreciate this more clearly, consider the idealized distribution in Fig. 7. Broadening of this distribution can be demonstrated by calculating the net rate of conversion of (i - 1)-mers to i-mers: net rate = k+clci-l

- k-ci

(43)

During length redistribution, Kcl is essentially very close to unity; thus K+cl will be very nearly equal to k- . This simplifies the above equation to the following form: net rate = L(ci-1

- ci)

(44)

To the left of the peak where the tubules have shorter lengths, ci-l is less than 4; so the net flux is from i-mers to (i - 1)-mers.To the right of the peak, the distribution of polymer length falls off, and ci-1 is greater than ci. Therefore, the net flux will be in the opposite direction. The combined action of these fluxes will result in the broadening of the peak distribution. Eventually, the peak will completely disappear due to the relationships among the concentrations of each polymer species. In this respect, the initial polymer-protomer equilibrium is maintained by the balanced rates of protomer addition and loss from polymer ends, and the protomers will scramble or diffuse from one polymer to another. Indeed, even after the polymer length redistribution reaches its thermo-

MICROTUBULE ASSEMBLY

193

dynamically controlled state, there will be a relentless series of reactions involving protomer adsorption and protomer desorption from the polymers in the solution. The net flux will, however, cease to be important upon complete redistribution to the equilibrium condition. With regard to the importance of polymer length redistribution to the dynamics of microtubules, Kristofferson and Purich (198la) attempted to gauge the time scale of the redistribution process. There are serious experimental obstacles preventing a full analysis of the time course of this phase of the Oosawa polymerization model. Chief among these is the instability of microtubule proteins, and effects resulting from denaturation will lead to incorrect estimates of the reaction kinetics. Moreover, because nucleotide hydrolysis occurs with whole microtubule protein samples, changes in the GTP/GDP ratio can affect the protomer-polymer equilibrium. Extremely large sample sizes are also required to overcome problems arising from statistical fluctuations in length distribution measurements. Nonetheless, it was possible to demonstrate that some tubule redistribution, although very far from being complete, is evident in 4 hours (see Fig. 8) in the presence of a GTPregenerating system. These results suggest that microtubule redistribution is relatively unimportant during the course of most studies limited to a few hours or less.

Polymer Length cmetersb1O6

FIG.8. Experimental length distributionchanges for bovine brain microtubule protein at 1 and 4 hours after initiation of assembly. Plots A and B correspond to unsheared samples, and give average length values of 11.3 and 12.1 pm, respectively. Plots C and D correspond to the sheared samples at 1 and 4 hours after assembly was induced, and polymer length values averaged 3.1 and 5.2 pm, respectively. [Reproduced from Kristofferson and Purich (1981). Arch. B i o c h . Bz@hys. 211,222-226.1

194

DANIEL L. PURICH AND DAVID KRISTOFFERSON

VII. PROTOMER FLUXWITH ASSEMBLED POLYMERS (MICROTUBULE“TREADMILLINC”)

One of the most intriguing properties of assembled microtubules is their ability to participate in a biased addition of tubulin protomers at one end of a tubule and biased loss from the opposite end. This phenomenon was first studied with actin (Wegner, 1976), and the term “head-to-tail” polymerization was applied to describe its nature. Later, Margolis and Wilson (1978) provided the earliest evidence for this phenomenon with microtubules, and they have used the term “treadmilling” to emphasize the mechanical analogy. These workers and others have used the terms “opposite-end assembly/disassembly” or steady-state tubulin flux to describe the process. Each of these terms has limitations because there is a failure of each to describe the essential mechanistic features of the exchange reactions. As we shall elaborate below, both ends of the microtubules are active in protomer uptake and release, and the net addition of protomers to one end reflects the bias imparted by the differential thermodynamic stability of the two ends. In this regard, the process is rather more accurately described as “biased protomer flux” or with microtubules as “biased steady-state tubulin flux.” Nonetheless, the use of the term treadmilling has become rather popular, and as long as one does not take the mechanical analogy too literally, the term is useful. A. Theoretical Cmiderations Biased flux was first hypothesized by Wegner (1976) to explain the rapid uptake of radioactive actin protomers (or Gactin) into actin polymers (F-actin) after the protomer-polymer equilibrium was already achieved. Wegner observed that the label uptake was significantly faster than one would estimate from the protomer-polymer equilibrium exchange. He went on to demonstrate that ATP hydrolysis accompanying G-actin addition to F-actin allows the polymers to treadmill and can account for the rapid label incorporation. Because both actin and tubulin hydrolyze nucleotides in an assembly-induced fashion, Wegner’s arguments about the F- and G-actin exchange process have been of great interest to those investigating the dynamic properties of microtubules. His head-to-tail polymerization theory consists of two basic arguments relating to the polymerization kinetics for protomers which do and do not hydrolyze nucleotides upon polymerization. They also apply to other systems which may rely on other free energy-yielding reactions to promote polymerization. In the former case, the disassembly reaction is not merely the reverse of the assembly process because the protomer is

MICROTUBULE ASSEMBLY

195

SCHEMEI.

-

-

released as tubulin GDP in disassembly but it was bound as tubulin GTP in assembly reactions. We now proceed to review how this difference is a crucial point in the Wegner model. In Scheme I, we present a kinetic scheme for tubulin polymerization in the absence of hydrolysis. The rate constants are written such that the (+) signs refer to the association steps and the (-) signs refer to the dissociation steps. The rates of growth at each end may be written as follows:

- k-1 = k + l [ c l - (k-l/ktl)] Rate at end-2 = K+2cl - k 2= k+2[c1 - (k-2/k+2)]

Rate at end-1 = ktlcl

(45) (46)

As noted earlier, it is conceivable that the rate constants at each end are different; however, Wegner noted that in the absence of nucleotide hydrolysis the assembly and disassembly reactions are in fact reversible. The rate constants are directly related to the equilibrium constants for protomer-polymer interactions at each end:

K1

= (ktl/k-l)

(47)

K2 = (kt4k-1) (48) Furthermore, although the rates of addition and loss may differ at each end (i.e., the transition states for the individual association-dissociation reactions may depend on the polarity of protomer-polymer interactions), the initial state (protomer + MT,) and the final state (MT,,I) are independent of the pathway for protomer addition. This result is true because addition or loss of a protomer at either end leads to structurally indistinguishable polymers. Thus, the energetics of the reactions at each end are the same, and K 1equals K2 for the case of no nucleotide hydrolysis accompanying assembly. We may then reexpress the rate equations for protomer addition or loss to the two ends in the following manner:

196

DANIEL L. PURICH AND DAVID KRISTOFFERSON

Rate at end-1 = ktl(cl - Ki')

(49)

- Ki')

(50)

Rate at end-2 = k+2(c1

The quantity [cl - K;'] is either positive, negative, or zero depending on whether c1 is greater than, less than, or equal to K i',respectively. If it is greater than KL', polymer growth will occur; if less than Ki',protomer loss from the polymer will occur; and if the terms are equal, no net growth or loss can take place. Furthermore, both ends operate simultaneously under the same constraints, leading to bidirectional growth, loss, or stabilization; thus, it is impossible for one end to grow at the expense of protomer loss from the opposite end. Wegner also treated the case wherein assembly is coupled to nucleotide hydrolysis. Here, we consider a slight modification of his model to deal with the microtubule process. Normally, the concentration of GTP is maintained by use of a GTP-regenerating system (MacNeal et al., 1977), and the system at the steady-state plateau of assembly can be described as in Scheme 11. Under these conditions, the assembly-disassembly reactions are no longer reversible, and the primed rate constants are used to emphasize that we are dealing with a different case. The rate equations for the two ends are now given as:

- kl_l Rate at end-2 = k+pc, - k'g Rate at end-1 = ktlcl

(51) (52)

where in this case virtually all of the free protomers bind GTP because of the efficiency of nucleotide diphosphate kinase (NDPK) relative to the much slower treadmilling rate (i.e., cI = [Tb * GTP]). If Tb GTP were not hydrolyzed during the assembly reactions, then one could still set up equilibrium relationships such as k+l/k-l equals k + 2 / k - 2 . But, because the system is no longer reversible, the two ends may have different ratios

SCHEME 11.

197

MICROTUBULE ASSEMBLY

of rate constants for protomer-polymer reactions. In this respect, the relative stability of the two ends will depend on the differential release of the free energy of nucleotide hydrolysis at the two polymer ends, and k + l l k l _ lwill in the general case be different than k+2/kL2. Thus, the above rate equations cannot be coupled together as done in the case for which no hydrolysis occurred. The protomer concentration will equilibrate when the sum of the rates at both ends is zero. Therefore,

(k+] + k + * ) C * - [k’,

+ kI-21 = 0

(53)

and solving for c1 gives C]

=

[kL1

+ k!.2]/[k+I + k + 2 ]

(54)

Indeed, this critical concentration, cl, depends on the rate constants at both ends when GTP hydrolysis is linked to the assembly process. By substituting this relationship into Eqs. (51) and (52) we obtain the rates of protomer addition or loss from the two ends: Rate at end-1 =

[kLIkL2

rateat end-2 = [kLlkL2

(2

- 3 ] / [ k + 1 + k+2]

(2- &)]/[A+] -

kl_l

+ K+2]

(55) (56)

If k+l/kLI is greater than k+2/k!.2, end-1 will grow, and end-2 will lose protomers at an equal rate. Only if these two quotients were exactly equal, would one observe no biased flux or exchange from one end to the other. The microscopic mechanism generating this difference between k+ A’, and k+2/kl_2 is still unknown. This difference implies that the free energy released during hydrolysis of GTP at one end somehow differs from that released in a similar process at the opposite end. The difference may result from steric effects on the neighboring groups which comprise the constellation of reactive centers facilitating the hydrolysis. It is also possible that Pi release to the solution could be more or less sterically blocked at one or the other end as a result of the nature of the polar interaction of protomer and polymer at the two ends. Nonetheless, Terry and Purich (1980) first demonstrated that the treadmilling process requires GTP hydrolysis by showing that GMPP(CH2)P and GMPP(NH)P could not sustain the treadmilling process, a finding which has been confirmed in other laboratories (Margolis, 1981; Cote and Borisy, 1981). One should also note that Thompson (1982) has described a model for the cyclical tyrosination/detyrosination of tubulin,

198

DANIEL L. PURICH AND DAVID KRISTOFFERSON

and such mechanisms could provide yet another means for introducing the possibility for biased flux of protomers and polymers. Wegner (1976) compared his rates of labeled actin incorporation to those estimated for a nontreadmilling process, and he introduced the parameter “s” to evaluate the commitment of a particular system to biased flux: s = [k+,Cl

- ALI]/[kLI

+ kL*]

(57) Thus, s represents the net uptake at the assembling end divided by the total dissociation rate from both ends. The denominator term can be determined by isothermal dilution (Karr and Purich, 1979; Zeeberg et al., 1980; Purich et al., 1982a). Note also that cI in the above expression corresponds to that given in Eq. (54). The s value arises in Wegner’s comparison of the time needed to incorporate n labeled protomers via equilibrium exchange, l(n),with the time needed for the same extent of labeled protomer exchange via treadmilling, At(n), where

i(n)lAt(n) = ns (58) Thus, if s equals 0.5, the equilibrium exchange process requires twice as long a reaction period to incorporate 4 subunits as does treadmilling. Wegner determined’for his actin system that s was 0.25. The reader should note that Eq. (58) was derived using two approximations. First, i(n) was assumed to be equal to n2/k- where k- is the dissociation rate constant of a nontreadmilling polymer at one end (Kasai and Oosawa, 1969). Second, Wegner let k- be the sum of kLl and k L 2 . It would appear that a better approximation would be that 2k- equals this sum, because one is comparing the release rate at one end of a nontreadmilling polymer to the ends of a treadmilling polymer. Thus, i(n)lAt(n)would equal n(s/2).Of course, Wegner’s own approximation might be better in the extreme “one-hit” case when treadmilling would be 100% efficient (i.e., when kLl and k+* are both zero). The first approximation of Kasai and Oosawa (1969) has been improved by the efforts of two research groups. Zeeberg et al. (1980) employed an elegant approach with difference equations to the equilibrium exchange problem, and, through the use of a Taylor’s series approximation as well as Stirling’s approximation, obtained the following solution: Lj(t) = (2/7r)’n[n,]1/2 ( 2 = 1, 2) = (2/7r)”2[(k-i+ i,I)t]’”

(59)

where L j ( t )is the average label incorporated at the end i during time t,

MICROTUBULE ASSEMBLY

199

and l+iequals k+,cl. For nontreadmilling polymers, k iequals i + ithus, ; the following will be true:

L,(t) = (2/?T)’/2[2kit]”2 (60) To compare this result with the Kasai and Oosawa equation for the time required for the uptake of n labeled protomers [i(n) = n 2 / k - ] ,let Li(t) equal n, and rearrange Eq. (59) to give t = (7d4)n2/k-, (61) Because (d4)is about 0.79, Kasai and Oosawa overestimated the time for such an extent of exchange by about 27%. Note also that this result, together with the earlier stated factor of two, changes Wegner’s equation to the following: i(n)/At(n)= (7r/8)ns (62) It should be noted that Wegner’s experimental determination of s was made by another protocol which does not employ such approximations. At any rate, the discrepancy in the equilibrium exchange rate versus the treadmilling rate may be a factor of (d8)less than previously believed. Kristofferson and Purich (1981b) also solved the equilibrium exchange problem, but these workers used an exact computational algorithm for labeled protomer incorporation. The computed results agree very well with those of Zeeberg et al. (1980), except at small n values where the approximations used to obtain Eq. (59) break down. Nonetheless, for those values of n which are experimentally accessible (i.e., for reaction periods corresponding to large n values), the results of Zeeberg et al. (1980) are well within 1% of the exact values. Zeeberg et al. (1980) also solved the difference equations for treadmilling polymers. Their solution involved approximating their difference equations with a differential equation, and the reader should consult the original report for details. Kristofferson and Purich (unpublished results) programmed the difference equations of Zeeberg et al. (1980) to obtain an exact iterative solution, and determined that the approximations employed in the original solution do not significantly affect the results. The diffusional components to Eqs. (16)-( 18) in the Zeeberg et al. (1980)report were affected by the approximations when compared to the exact computer solution. However, this error was negligible in the overall label uptake because the diffusional component was always small compared to the directional component at any experimentally accessible time. An exact solution to this problem has been recently advanced by Tsuchiya and Nagai (1983).The approximate Zeeberg et al. (1980) diffu-

200

DANIEL L. PURICH AND DAVID KRISTOFFERSON

sional uptake limits should be ql(pl - q l ) - ’ and p2(q2 - p2)-’ instead of ( p l - q l ) - * and (42 - p$*, respectively. These modified limits can be derived from Equation (1,l)of Tsuchiya and Nagai (1983).The resulting change is small and does not affect any experimental conclusions in the treatment of Zeeberg et al. (1980). Finally, the theory for the bioenergetics and kinetics of microtubule assembly and disassembly of microtubules has been extended by Hill and Kirschner (1983).They consider the coupling of nucleotide hydrolysis in terms of the energetics of the [GTP]/[GDP][Pi] mass action ratio, the possible effects of force imparted by attachment of tubules to barriers on the rate constants, and other intriguing aspects of protomerpolymer exchange kinetics and thermodynamics. Unfortunately, much of their theory remains to be tested, and an evaluation of its importance in revealing the subtleties of assembly/disassembly remains for future investigations. B . Experimental Findings The asymmetry in rate constants for microtubule end interactions with tubulin protomers is reflected in the kinetics of labeled tubulin protomer uptake by unlabeled tubules or labeled tubulin protomer release from labeled microtubules. The original discovery of microtubule treadmilling (Margolis and Wilson, 1978)involved an experimental approach based on factors other than a direct comparison to equilibrium exchange rates. Instead of using radioactively labeled protein, Margolis and Wilson (1978)monitored the process by following the time course of [SH]GTPincorporation, relying on the inaccessibilityof E-site nucleotide to exchange when tubulin is in the assembled tubule. A tracer level of labeled nucleotide was added after attaining a stable protomer-polymer equilibrium, and the reaction was followed by pelleting assembled tubules through a sucrose “cushion” to fractionate polymerized tubulin from tubulin protomers. Alternatively, tubules could be assembled in the presence of labeled nucleotide, and subsequently the dynamics of tubulin protomer release could be measured by addition of unlabeled GTP which served as a trap for labeled nucleotide that became exchangeable. The cumbersome use of centrifugation has recently been replaced (Wilson et al., 1982)by using a modification of the Mqccioni and Seeds (1977)glass-fiber-filter method. In both uptake and release experiments, the reactions were linear with time, and the reaction rates were nearly identical. The uptake was blocked by the addition of podophyllotoxin, an antimitotic poison which acts in an end-wise manner. Margolis and Wilson (1978)also performed a pulse-chase experiment by adding

MICROTUBULE ASSEMBLY

20 1

labeled GTP to steady-state tubules for 1 hour, followed by a 1 hour chase in the presence of excess nonradioactive (or carrier) nucleotide. Their observation that only 4.4% of the incorporated label was lost during the subsequent chase led them to conclude that the treadmilling model was very efficient compared to equilibrium exchange. Together, these data formed the experimental basis for the idea that the treadmilling reaction was an essentially unidirectional reaction: T b L - TbTbTbTbTbTbTbTbTbTbTbTbA Tb Primary Primary Assembly Disassembly End End

(63)

In this mechanism, the protomer addition and release steps are exclusively taking place at opposite ends of the tubule. Though the above mechanism had value in describing the steady-state tubulin flux, Karr and Purich (1979) pointed out that a number of association-dissociation steps can occur for each net protomer addition. They found that the rate of isothermal dilution-induced disassembly of microtubules proceeded at rates 500- to 1000-fold greater than the observed treadmilling rate which itself corresponded to a turnover of about 7% of the tubulin pool/hour (Margolis and Wilson, 1978). Karr and Purich ( 1979) also proposed that the unidirectional treadmilling model be amended to account for the rapidity of on-off events at the polymer ends, and they offered the alternative mechanism presented here. Tb + MT(,,-I)

Primary Assembly End

____

MT(rn-11+ Tb Primary Disassembly End

MT(.)

(64)

In this case, MT,,-,) and MT(,) represent two states of polymerization, and there are association/dissociationevents occurring at each end. Indeed, Karr and Purich (1979) recognized that thermodynamic considerations dictate that there must be a reverse reaction rate, but they had no way to evaluate the relative magnitude of this step and used a dashed line instead. Nonetheless, the combined equilibrium-treadmilling model, while still correct as a minimal mechanism for the steady-state flux, was based on several simplifying assumptions which are now recognized as incorrect. First, the polarity of podophyllotoxin action on the treadmilling reaction was assumed to be exclusively at the primary assembly end (Margolis and Wilson, 1978), but there is now ample evidence that both colchicine and podophyllotoxin can inhibit the addition

202

DANIEL L. PURICH AND DAVID KRISTOFFERSON

of dimers to both ends of a microtubule (Sternlicht and Ringel, 1979; Borisy and Bergen, 1982). In this regard, the slow rate of disassembly observed after the initial rapid disassembly phase attending podophyllotoxin addition cannot be attributed to the rate of dimer release from the (-)-end of the microtubule. And second, the rates of protomer addition to both ends are roughly comparable. In Table V, data for the axoneme experiments of Bergen and Borisy (1980), for instance, show that the on-rates differ only by a factor of three. Thus, the two ends have roughly similar numbers of association-dissociation cycles with only a slight imbalance favoring one end. The importance of the need to treat the reactions at both ends of microtubules as reversible processes was recognized by Zeeberg et al. (1980) who applied the most general model and developed a clear framework for the quantitative analysis of the treadmilling process. They showed that the processes must be viewed as a diffusional incorporation of tubulin protomers taking place at both ends of the tubule (see Section VI1,A for details of their theoretical approach). In this regard, the kinetics of net protomer uptake by a microtubule depend on the relative size of four rate constants, and the bias is relatively small. For example, Zeeberg et al. (1980) found that porcine-brain-microtubule treadmilling progressed at a rate corresponding to 1.5 net incorporation stepdhundred cycles of protomer addition and loss. They also found that a model with only 2% directionality was sufficient to explain the kinetics of the pulse-chase experiments of Margolis and Wilson (1978) who found that only 4.4% of the labeled protomers could be chased from the tubules upon addition of unlabeled nucleotide serving as the TABLE V Summay of the Kinetic and Equilibrium Constunts CharaxterizingMicrotubule Elongation from Chlamydmnas Flagellar Axonemes" Parameter (the on-rate constant for the a end) (the off-rate constant for the a end) (the on-rate constant for the b end) (the off-rate constant for the b end) (the formation constant for the a end) (the formation constant for the b end) (the macroscopic formation constant) (the commitment to steady-state head-to-tail polymerization)

Approximate value and units 7.2 x 106M-' sec-' 17 sec-' 2.3 X lo6 M - ' sec-l 7 sec-' 4.2 x 1 0 5 ~ 4 3.3 x 105 M - 1 4.0 x 105 M - 1 0.07

Estimated from the values reported by Bergen and Borisy (1980).

MICROTUBULE ASSEMBLY

203

incorporation probe. Thus, Zeeberg et al. (1980) concluded that the treadmilling process is a rather inefficient process, and their analysis has raised important questions about the nature and role of treadmilling. Bergen and Borisy ( 1980) used the axoneme-based approach to explore the commitment to treadmilling, and they also found that the efficiency was quite low. An s value of 0.07 0.04 was obtained, corresponding to the net addition of 1.6 0.8 tubulin protomers/second. Interestingly, their estimates of the dissociation constants for the two ends were 2.2 f 0.6 and 3.2 f 0.6 pikf for the assembly and disassembly ends, respectively. We calculate that this corresponds to only about 0.2 kcal/mol. In a more recent investigation from the same laboratory (Cote and Borisy, 1981), porcine tubulin was found to exhibit an s value of about 0.26, and the rate of the tubulin flux was about 28 protornerd second. The authors of the latter study suggested that the discrepancy might be accounted for in terms of the need to use MAP-depleted tubulin with the axoneme system to prevent self-nucleation, or in terms of the temperature differences in the two studies. It is also interesting to recognize that Sandoval and Weber (1980) found that cY,P-methylene GTP blocked the steady-state tubulin flux. Although the earlier studies with P,y-nonhydrolyzable analogs (Terry and Purich, 1980; Margolis, 1981; Cote and Borisy, 198 1) demonstrated such behavior, the work of Sandoval and Weber (1980) is particularly fascinating because the scissile bond of GTP is not modified. This suggests that some steric interactions might be important or that the formation of the metal-nucleotide complex may be sufficiently altered so as to preclude the flux reactions. Margolis (1981) suggested that the value of s may depend inversely on the GTP concentration. Instead of measuring the uptake by the above methods, Margolis chose to measure the downward drift of the turbidity following the rapid phase of podophyllotoxin-induced tubule disassembly. Although no explanation for the rapid phase is given, the analysis of the slower phase is based on the notion that podophyllotoxin will block only the assembly end, leaving the disassembly end free to proceed at the treadmilling rate. By comparing this result to the kinetics of disassembly after converting all GTP in the solution into GDP with phosphofructokinase, Margolis (1981) reached the conclusion that the s value varies from 0.1 at high GTP concentrations to 0.8 at lower nucleotide levels. The experimental approach and theoretical framework for this particular analysis appears to be limited in several important respects. First, the action of podophyllotoxin is to cap both ends of the microtubule (Zeeberg and Caplow, 198 1). Second, any downward drift in the turbidity tracings is open to alternative explanations other than that presumed

*

*

204

DANIEL L. PURICH AND DAVID KRISTOFFERSON

in the Margolis (1981) analysis. Third, the kinetics of GTP conversion to GDP are necessarily superimposed on the kinetics of microtubule disassembly, and one may find the deconvolution of these two overlapping kinetic processes gives no clear or direct relationship to the microscopic dissociation constants. Finally, if the endogenous GTPase activity results, even partially, from the action of tubule-associated nucleotide phosphohydrolase (White et al., 1980),then the GTP/GDP ratio may change with consequential effects on the apparent turbidity plateau. This could explain the unusual dependence of the s value on nucleotide concentration. From these considerations, it therefore appears to be unlikely that Margolis (198 1) has modeled the dynamics of this treadmilling process. Finally, the report of Caplow et al. (1982) serves as a useful vehicle for closing this discussion of microtubule treadmilling. These investigators have reexamined the efficiency of the treadmilling phenomenon by using radioactively labeled tubulin protomers from both porcine and dogfish shark brain to evaluate quantitatively the protomer incorporation into microtubules under steady-state conditions. Their analysis took into account that diffusional uptake of labeled protomers occurs at both ends of the tubule, and that pulse labeling at steady-state results from uptake at both ends as well. Using pseudo-first-order rate constants (i.e., k; and k; corrected for the macroscopic critical concentration such that they represent kl[Tb],it and k2[TbIcit, respectively), these investigators show that the treadmilling-associated uptake of tubulin protomers by both ends is described by the following relation with Tb* representing the value of label uptake: Tb*(t) = [(k-l

+ k ; ) / ( k ; - k-I)] + (k; - k-l)t

(65) Here, a plot of subunit uptake versus time after pulsing will be linear, with an associated intercept corresponding to the reciprocal of the Wegner s value, and a slope corresponding to the so-called treadmilling rate constant. Nonetheless, prior to this linear phase, one may anticipate the occurrence of a diffusional process with the extent of uptake proportional to the square root of the time period of exchange. As noted earlier, the extent of diffusional exchange depends on the values of the off-rate constants (Kasai and Oosawa, 1969; Zeeberg and Caplow, 1981; Kristofferson and Purich, 1981b). In their studies, Caplow et al. (1982) obtained values of about 5000 (porcine brain) and 2500 (dogfish shark brain) tubulin protomers exchangedhecond at the two microtubule ends. It is noteworthy that the sum of these off-rate constants for the two tubule ends would be about 20 to 40 times lower than that obtained by rapid dilution techniques (Karr et al., 1980a; Zeeberg et al., 1980). To reconcile this disparity, Caplow et al. (1982) suggest that protomers, as

205

MICROTUBULE ASSEMBLY

tubulin GTP complexes, add to and are lost from tubule ends that are capped with a region of tubulin GTP protomers corresponding to about a thousand or so tubulin subunits. These protomers, containing nucleotide at the triphosphate state of phosphorylation, are thought to undergo rapid and facile exchange, whereas the protomers nearer the tubule interior are thought to be present as the tubulin * GDP complex. The latter are viewed as more slowly exchanging tubulin protomers. Such boundary models have been considered earlier (Karr et al., 1979b) but the tubulin GTP association with the tubule end was assumed to be more stable. With the boundary concept, Caplow et al. (1982) presume that the off-rate constant determinations by the rapid-dilution technique may fail to observe the more rapid initial burst corresponding to about 5-10% of the total tubulin content of the microtubule. Nonetheless, one is still left with the notion that the on-rate constants for tubulin addition to the microtubule must be correspondingly greater (i.e., k; and k; will be 20- to 40-fold greater), and the association rate constants, after allowing for the critical concentration of tubulin in dynamic equilibrium with the microtubules, will be at or beyond the bimolecular rate constants likely for macromolecule-macromolecule interactions. Furthermore, the much slower rate constants for the “interior” Tb * GDP protomers should limit the rates of extensive diffusional exchange, and further confirmation of this prediction will be a useful check on this rapidexchange boundary model. Another interesting consequence of the recent kinetic measurements of Caplow et al. (1982) is that the Wegner s value must be revised downward to a value less than 0.001, because the off-rate constant values factor into the estimation of the s value. The significance of this revision, however, is only technical in its nature, and consideration of an analogous situation in enzyme kinetics helps one to appreciate this perspective. Indeed, the Wegner s value is akin to the parameter termed the “commitment-to-catalysis”in enzyme rate processes. A very low commitment-to-catalysis is said to characterize the situation of Michaelis-Menten behavior wherein the enzyme and substrate associate and dissociate many times relative to the rate of catalytic turnover (i.e., there is a rapid preequilibration of enzyme and substrate). In contrast, there is said to be a high commitment-to-catalysis whenever the rate of enzyme substrate turnover to product is competitive with the rate of substrate desorption from the enzyme * substrate complex. Thus, while measurement of the commitment-to-catalysis reveals the number of unproductive and productive enzyme-substrate excursions, the parameter btcharacteristically defines the catalytic efficiency, and even a low commitment-tocatalysis does not diminish the significance of the observed k,,, value. By

-

-

206

DANIEL L. PURICH AND DAVID KRISTOFFERSON

analogy, the redetermination of the s value, while providing mechanistic insight into the nature of tubulin interactions with the tubule ends, does not change the rate of net subunit additions observed per tubule end per unit time. The new values of 4.2-6.3 subunits/end/second are in fact quite analogous to other literature values. For example, Zeeberg and Caplow (1981) reported a value of 1.8 protomers/end/second; Terry and Purich (1980) found 0.3 protomers/end/second; and the original net rates of Margolis and Wilson (1978) lie in this general range. Thus, from the foregoing discussion, these newer estimates do not alter the net rate of treadmilling in any major way.

REMARKS VIII. CONCLUDING The reader who has persevered to the end of this presentation on microtubule assembly and disassembly may feel somewhat disappointed by the scant consideration of the connections among the structural and dynamic properties of both tubulin and microtubules. While both tobacco mosaic virus and hemoglobin S fiber assembly, for example, have provided many powerful insights into the mechanisms of self-assembly of supramolecular structures, progress toward similar goals with microtubules still remains painfully slow. In spite of the numerous investigations of microtubule assembly reactions and the valuable framework of the Oosawa protein polymerization theory, there still remain significant lacunae in our understanding of nearly all phases of tubule assembly. Even in the case of elongation, the phase which is presently best understood, we still lack an understanding of the interplay between the types of protomer-protomer interactions which characterize the ends of microtubules. Until we understand such interactions at the molecular level, our appreciation of pathway(s) for assembly and disassembly reactions will remain too vague and unsatisfying. Nonetheless, few would quarrel with the assertion that we have already witnessed remarkable growth and maturation in an area of biochemical investigation that is less than 15 years old. Finally, we offer our apologies to those investigators whose work on microtubules was not included in this article. The explosive growth of microtubule research over the past few years imposes limitations on any reviewer who seeks to provide adequate coverage of the field. In this regard, the valuable reviews of Olmsted and Borisy (1973), Kirschner (1978), and Timasheff and Grisham (1980) should be consulted by those interested in gaining a broader perspective of the literature concerning the biochemical and biophysical properties of microtubule systems.

MICROTUBULE ASSEMBLY

207

REFERENCES Amos, L. A. (1977).J. Cell Biol. 69, 642-654. Amos, L. A. (1979).In “Microtubules” (J. Hyams, ed.), pp. 2-64. Academic Press, New York. Amos, L.A., and Klug, A. (1974).J. Cell Sci. 14,523-549. Amos, L. A., Linck, R. W., and Mug, A. (1976).In “Cell Motility” (R. Goldman, T. D. Pollard, and J. L. Rosenbaum, eds.), Book C, pp. 847-867. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Anderson, S. A., and Purich, D. L. (1982).J.Biol. Chem. 257,6656-6658. Arai, T., and Kaziro, Y. (1976).Bwchem. Biuphys. Res. Cummun. 69,369-376. Argarana, C. E., Barra, H. S., and Caputto, R. (1978).Mol. Cell. Biochem. 19, 17-21. Asnes, C. F., and Wilson, L. (1979).Anal. Biochem. 98,64-73. Bergen, L., and Borisy, G. G. (1980).J.Cell Biol. 84, 141-150. Berne, B. J. (1974).J. Mol. Biol. 89, 756-758. Berry, R. W., and Shelanski, M. L. (1972).J.Mol. Biol. 71, 71-80. Bonsy, G. G., and Bergen, L. G. (1982).Methods Cell Biol. 24, 171-188. Borisy, G.G.,and Olmsted, J. B. (1972).Science 177, 1196-1 197. Borisy, G. G., and Taylor, E. W. (1967).J. Cell Biol. 34,525-548. Borisy, G.G.,Marcum, J. M., Olmsted, J. B.. Murphy, D. B., and Johnson, K. A. (1975). Ann. N.Y. Acad. Sci. 253, 107-132. Borisy, G. G., Johnson, K. A., and Markum, J. M. (1976).In “Cell Motility” (R. Goldman, T. Pollard, and J. Rosenbaum, eds.), Book C, pp. 1093-1108. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Brylawski, B. P.,and Caplow, M. (1983).J.Biol. Chem. 258, 760-763. Burke, B. E., and DeLorenzo, R. J. (1981).Proc. Natl. Acad. Sci. U.S.A. 78,991-995. Burton, P. R., and Himes, R. H. (1978).J.Cell Biol. 77, 120-133. Burton, P. R., and Hinkley, R. E. (1974).J.Submicrosc. Cytol. 6,311-326. Burton, P. R., Hinkley, R. E., and Pierson, G. B. (1975).J.Cell Biol. 65,227-233. Caplow, M., and Zeeberg, B. (1981).J.Biol. Chem. 256,5608-5614. Caplow, M., Langford, G . M., and Zeeberg, B. (1982).J. Biol. Chem. 257, 15012-15021. Carlier, M. F. (1982).J.Biol. Chem. 257, 2415-2420. Carlier, M. F., and Pantaloni, D. (1978).Biochemistry 17, 1908-1915. Carlier. M. F.,and Pantaloni, D. (1981).Biochemistry 20, 1918-1924. Caspar. D. L. D., and Klug, A. (1962).CoM Spring Harbor Symp. Quant. Biol. 27, 1-24. Chou, P. Y., and Fasman, G. D. (1974).Biochemishy 13,222-245. Cleveland, D. W., Spiegelman, B. M., and Kirschner, M. W. (1979).J. Biol. Chem. 254, 12670-12678. Cleveland, D. W., Lopata. M. A., MacDonald, R. J., Cowan, N. J., Rutter, W. J., and Kirschner, M. W. (1980).Cell 20, 95-105. Mol. Biol. 150,577-602. Cote,-R. H., and Borisy, G. G. (1981).J. Dentler, W. L., Grannett, S.,Witman, G. B., and Rosenbaum, J. L. (1974).Proc. Natl. Acad. SCi. U.S.A. 71, 1710-1714. Detrich, H. W., Berkowitz, S.A., Kim, H., and Williams, R. C., Jr. (1976).Biochem. Biophys. Res. Cummun. 68,961-968. Eipper, B. A. (1974a).J. Bwl. Chem. 249, 1398-1406. Eipper, B. A. (1974b).J.Biol. Chem. 249, 1407-1416. Engelbroughs, Y.,DeMaeyer, L. C. M., and Chrerbergh, N. (1977).FEBS Lett. 80,81-85. Engelbroughs, Y., Robinson, J., and Ide, G. (1980).Biophys. J . 32,440-443. Erickson, H. P. (1974).J . Cell Biol. 60, 153-167.

208

DANIEL L. PURICH AND DAVID KRISTOFFERSON

Euteneuer, U., and McIntosh, J. R. (198O).J.Cell Biol. 87, 509-515. Flavin, M.,Kobayashi, T., and Martensen, T. M. (1982).Methods Cell Biol. 24,257-263. Friedman, F. K., and Beychok, S. (1979).Annu. Rev. Biochem. 48, 217-250. Fuller, G. M., Brinkley, B. R., and Boughter, M. J. (1975).Science 187,948-950. Fulton, C., and Simpson. P. A. (1976).In “Cell Motility” (R. Goldman, T. Pollard, and J. Rosenbaum, eds.), Book C, pp. 1051-1064. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Gamier, J., Osguthorpe, D. J., and Robson, B. (1978).J. Mol. Biol. 120,97-120. Gaskin, F., Cantor, C. R., and Shelanski, M. L. (1974).J.Mol. Biol. 89,737-758. Geahlin, R. L., and Haley, B. E. (1977).Proc. Natl. Acad. Sci. U.S.A. 74, 4375-4377. Geahlin, R. L., and Haley, B. E. (1979).J.Biol. Chem. 254, 11982-1 1987. George, H. J., Misra, L., Field, D. J., and Lee, J. C. (1981).Biochemittty 20, 2402-2409. Gibbons, I. R. (1963).Proc. Natl. Acad. Sci. U.S.A. 50, 1002-1006. Gibbons, I. R., and Fronk, E. (1972).J . Cell Biol. 54, 365-374. Gould, R., and Borisy, G . G. (1977).J.Cell Biol. 73,601-615. Griffith, L. M., and Pollard, T. D. (1978).J.Cell Biol. 78,958-965. Grimstone, A. V., and Klug, A. (1966).J. Cell Sci. 1, 351-356. Haimo, L. T.(1982).Methodr Cell Biol. 24, 189-206. Haimo, L. T., Telzer, B. R., and Rosenbaum, J. L. (1979).Proc. Natl. Acad. Sci. U.S.A. 76, 5759-5763. Hallak, M. E., Rodriguez, J. A., Barra, H. S., and Caputto, R. (1977).FEBS Lett. 73, 147150. Hawkins, D. J., and Browning, E. T. (1982).Biochemistty 21,4474-4479. Heidemann, S . R.,and Euteneur, U.(1982).Methods Cell Biol. 24, 207-216. Heidemann, S . R., and McIntosh, J. R. (1980). Nature (London) 486, 517-519. Hill, T. L., and Kirschner. M. W. (1983).Int. Rev. Cytol. 78, 1-125. Himes, R. H. (1977).J. Biol. C h . 252, 6222-6228. Hinz, H.-J., Gorbunoff, M. J., Price, B., and Timasheff, S . N. (1979).Biochemirty 18, 3084-3089. Hofrichter, J., Ross, P. D., and Eaton, W. A. (1974).Proc. Natl. Acad. Sci. U.S.A. 71,48644868. Inouk, S. (1952).Ex$. Cell. Res. Suppl. 2, 305-3 18. Jacobs, M., Smith, H., and Taylor, E. W. (1974).J.Mol. Biol. 89,455-468. Jameson, L., and Caplow, M. (1980).J . Biol. C h . 255, 2284-2292. Job, D.. Rauch, C. T., Fischer, E. H.. and Margolis, R. L. (1983).Proc. Natl. Acad. Sci. U.S.A. 80,3894-3898. Johnson, K. A., and Borisy, G. G. (1977).J.Mol. Biol. 117, 1-31. Kakiuchi, S. (1982).In “Biological Functions of Microtubules and Related Structures” (H. Sakai, H. Mohri, and G. G. Borisy, eds.), pp. 83-90. Academic Press, New York. Karr, T. L., and Purich, D. L. (1978).B i o c h . Biophys. Res. Commun. 84, 957-961. Karr, T.L., and Purich, D. L. (1979).J.Bwl. Chem. 254, 10885-10888. Karr. T.L., and Punch, D. L. (1980).B i o c h . Biophys. Res. Commun. 95, 1885-1889. Karr, T.L., White, H. D., and Purich, D. L. (1979a).J.Biol. Chem. 254, 6107-6111. Karr, T.L.,Podrasky, A. E., and Purich, D. L. (1979b).Proc. Natl. Acad. Sci. U.S.A. 76, 5475-5479. Karr, T. L., Kristofferson, D., and Purich, D. L. (1980a).J.Biol. Chem. 255,8560-8566. Karr, T.L., Kristofferson, D., and Purich, D. L. (198Ob).J.Biol. Chem. 255, 11853-1 1856. Karr, T. L., White, H. D., Coughlin, B. A., and Purich, D. L. (1982).Methods Cell Biol. 24, 5 1-60. Kasai, M., and Oosawa, F. (1969).Biochim. Biophys. Acta 172,300-310. Kim, H., Binder, L. I., and Rosenbaum, J, L. (1979).J.Cell Biol. 80,266-276.

MICROTUBULE ASSEMBLY

209

Kirsch, M., and Yarbrcugh, L. R. (1981).J.Biol. C h . 256, 106-111. Kirschner, M. W. (1978).Int. Rev. Cyfol. 54,l-71. Kirschner, M. W. (1980).]. CeU Bwl. 86,330-334. Kirschner, M. W.. and Williams, R. C., Jr. (1974).j.Supamol. Struct. 2,412-428. Kirschner, M. W., Honig, L. S.,and Williams, R. C., Jr. (1975).J.Mol. Biol. 99,263-276. Kobayashi, T. (1975).J.B i o c h . (Tokyo) 77, 1193-1197. Krauhs, E., Little, M., Kempf, T., Hofer-Warbinek, R., Ade, W., and Ponstingl, H. (1981). PYOC.Natl. Acad. Sci. U.S.A. 78, 4156-4160. Kristofferson, D.,and Purich, D. L. (1981a).Arch. B i o c h . Biophys. 211, 222-226. Kristofferson, D., and Punch, D. L. (1981b).J.Them. Biol. 92,85-96. Kristofferson, D., Karr, T. L., and Purich, D. L. (1980).J.Biol. C h .255, 8567-8572. Kristofferson, D.,Karr, T. L., Malefyt, T. R., and Purich, D. L. (1982a).Methods Cell Biol. 24, 133-144. Kristofferson, D., Lee, S.-H., Terry, B. J. (1982b).In “Biological Functions of Microtubules and Related Structures (H. Sakai, H. Mohri, and G. G. Borisy, eds.), pp. 49-60. Academic Press, New York. Kumagai, H., Nishida, E.,and Sakai, H. (1982).In “Biological Functions of Microtubules and Related Structures” (H. Sakai, H. Mohri, and G. G. Borisy, eds.), pp. 73-81. Academic Press, New York. Kuznemv, S. A., Rodionov, V. I., Celfand, G.I., and Rosenblat, V. A. (1981).FEBS Len. 135,237-240. buffer. M. A. (1975).“Entropy-Driven Processes in Biology.” Springer-Verlag, Berlin and New York. Lee, J. C. (1982).Methodr Cell Bwl. 24,9-30. Lee, j. C., and Lee, L. Y. (1979).S k h i s h y 18,5518-5526. Lee, J. C.,and Timasheff, S. N. (1975).Biochakt~y14,5183-5187. Lee, J. C.,and Timasheff, S. N. (1977).Biochemktry 16, 1754-1764. Lee, J. C., Frigon, R. P., and Timasheff, S. N. (1973).J. Biol. Chem. 248, 7253-7262. Lee, J. C., Frigon, R. P., and Timasheff, S. N. (1975).Ann. N.Y. Acad. Sci. 253,284-291. Lee, J. C., Corfman, D., Frigon, R. P., and Timasheff, S. N. (1978a).Arch. Biochem. Biophys. 18594-14. Lee, J. C., Tweedy, N., and Timasheff, S. N. (1978b).B i o c h i s b y 17,2783-2791. Lee, J. C., Gekko, K., and Timasheff, S.N. (1979).In “Methods in Enzymology” (C. H. W. Hirs and S. N. Timasheff, eds.), Vol. 61,pp. 26-49. Academic Press, New York. Lee, S.-H., Kristofferson, D., and Purich, D. L. (1982).B i o c h . Biophys. Res. Commun. 105, 1605-1610. Lee, Y. C., and Wolff, J. (1982).J.Bbl. C h . 257,6306-6310. Linck, R. W., and Amos, L. A. (1974).J.Cell Sci. 14,551-559. Lu,R. C., and Elzinga, M. (1978).Biochim. Bwphys. Actg. 537, 320-328. Ludueiia, R. F., and Woodward, D. 0. (1973).Proc. Natl. Acad. Sci. U.S.A. 70,3594-3598. Ludueiia, R. F., Shooter, E. M., and Wilson, L. (1977).J. Biol. C h . 252, 7006-7014. Maccioni, R., and Seeds, N. W. (1977).Proc. Natl. Acad. Sci. U.S.A. 74,462-466. McEwcn, B., and Edelstein, S.J. (1980).J. Mol. Biol. 139, 123-145. McGill, M., and Brinkley, B. R. (1975).J.Cell Biol. 67, 189-199. MacNeal, R. K., and Purich, D. L. (1977).J.Biol. Chem. 252,4440-4442. MacNeal, R. K., and Purich, D. L. (1978a).J . Bwl. C h . 253,4683-4687. MacNeal, R. K., and Purich, D. L. (1978b).Arch. B i o c h . Biophys. 191,233-243. MacNeal, R. K.,Webb, B.C., and Purich, D. L. (1977).Biochem. Biophys. Res. Cmmun. 74, 440-447. Mandelkow, E., Thomas, J., and Cohen, C. (1977).Proc. Nail. Acad. Sci. U.S.A. 74,33703374.

210

DANIEL L. PURlCH AND DAVID KRISTOFFERSON

Mandelkow, E.-M., and Mandelkow, E. (1979).J.Mol. Eiol. 129, 135-148. Mandelkow, E.-M., Harmsen, A., Mandelkow, E., and Bordas, J. (1980).Nature (London) 487,595-599. Margolis, R. L. (1981).Proc. Natl. Acad. Sci. U.S.A. 78, 1586-1590. Margolis, R. L., and Wilson, L. (1978).Cell 13, 1-8. Martensen, T. M. (1982).Methods Cell Eiol. 24,265-269. Mellon, M. G., and Rebhun, L. 1. (1976).J.Cell Eiol. 70,226-238. Mohri, H. (1968).Nature (London) 217, 1053-1054. Murphy, D. B. (1982).Methods Cell Eiol. 24, 31-50. Murphy, D. B., and Borisy, G. G. (1975).Proc. Natl. Acad. Sci. U.S.A. 72, 2696-2700. Murphy, D. B., Vallee, R. B., and Borisy, G. G. (1977).Biochemistry 16,2598-2605. Nagano, T., and Suzuki, F. (1975).J. Cell Eiol. 64, 242-245. Nishida, E., Kotani, S.,Kuwaki, T., and Sakai, H. (1982).In “Biological Functions of Microtubules and Related Structures” (H. Sakai, H. Mohri, and G. G. Borisy, eds.), pp. 285-295. Academic Press, New York. Olmsted, J. B., and Borisy, G. G. (1973).Annu. Rev. Biochem. 44,507-540. Olmsted, J. B., and Lyon. H. D. (1981).J.Eiol. C h . 256, 3507-351 1. Olmsted, J. B., Marcum, J. M., Johnson, K. A., Allen, C., and Borisy, G. G. (1974).J. Supramol. Struct. 2,429-450. Oosawa, F., and Asakura, S. (1975).“Thermodynamics of the Polymerization of Protein.” Academic Press, New York. Oosawa, F., and Kasai, M. (1962).J.Mol. Eiol. 4, 10-21. Pantaloni, D., Carlier, M. F., Simon, C., and Batelier, G. (1981).Eiochmktry 20, 47094715. Pauling, L. (1953).Discuss. Faradq SOC.13, 170-176. Penningroth, S. M., and Kirschner, M. W. (1978).Eiochemistry 17, 734-740. Penningroth, S. M., Cleveland, D. M., and Kirschner, M. W. (1976).In “Cell Motility” (R. Goldman, T. Pollard, and J. Rosenbaum, eds.), Book C, pp. 1233-1257. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Pierson, G. B., Burton, P. R., and Himes, R. H. (1978).J.Cell Biol. 76,223-228. Pollard, T. D., Selden, S. C., and Griffith, L. M. (1982).In “Biological Functions of Microtubules and Related Structures (H. Sakai, H. Mohri, and G. G. Borisy. eds.), pp. 31 13 19. Academic Press, New York. Ponstingl, H., Little, M., Krauhs, E., and Kempf, T. (1979).Nature (London) 282,423-424. Ponstingl, H., Krauhs, E., Little, M., and Kempf, T. (1981).Proc. Natl. Acad. Sci. U.S.A. 78, 2757-2761, Purich, D.L., and MacNeal, R. K. (1978).FEES Lett. 96,83-86. Purich, D. L., Karr, T. L., and Kristofferson, D. (1982a).In “Methods in Enzymology” (D. W. Frederikson and L. W. Cunningham, eds.), Vol. 85,pp. 439-449. Academic Press, New York. Purich. D. L., Terry, B. J., MacNeal, R. K.. and Karr, T. L. (1982b).I n “Methods in Enzymology’’ (D. W. Frederikson and L. W. Cunningham, eds.), vol. 85, pp. 416433. Academic Press, New York. Renaud, F., Rowe, A., and Gibbons, I. (1968).J. Cell Biol. 36, 79-90. Rosenbaum, J. L., and Child, F. M. (1976).J. Cell Eiol. 34,345-364. Runge, M. S.,El-Maghrabi, M. R.. Claus, T. H.,Pilkis, S . J., and Williams, R. C. (1981a). EiOChedtry 20, 175- 180. Runge, M. S.,Laue, T. M., Yphantis, D. A., Lifsics, M. R., Saito, A., Altin, M., Reinke, K., and Williams, R. C., Jr. (1981b).Proc. Natl. Acad. Sci. U.S.A. 78, 1431-1435. Salmon, E. D. (1975).Science 189,884-886.

MICROTUBULE ASSEMBLY

21 1

Sanchez, F., Natzle, J., Cleveland, D. W., Kirschner, M. W., and McCarthy, B. J. (1980). Cell 22,845-854. Sandoval, I. V., and Weber, K. (198O).J.Biol. C h . 255,6966-6974. Scheele, R. B., and Borisy, G. G. (1976).Biochem. Biophys. Res. Commun. 70, 1-7. Scheele, R., Bergen, L., and Borisy, G. (1980).J. Cell Bwl. 87, 2499. Shelanski, M., and Taylor, E. W. (1968).J.Cell Biol. 38, 304-315. Shelanski, M., Gaskin, F., and Cantor, C. R. (1973).Proc. Natl. Acad. Sci. U.S.A. 70, 765768. Sloboda, R. D., Rudolph, S. A., Rosenbaum, J. L., and Greengard, P. (1975).Proc. Natl. Acad. Sn‘. U.S.A. 71,4472-4476. Sloboda, R. D., Dentler, W. L., and Rosenbaum, J. L. (1976).B i o c h i s h y 15,4497-4505. Snell, W. J., Dentler, W. L., Haimo, L. T., Binder, L. I., and Rosenbaum, J. L. (1974). S k e 186,357-360. Spiegelman, B. M.,Penningroth, S. M., and Kirschner, M. W. (1977).Cell 12, 587-600. Steams, M. E.,and Brown, D. L. (1979).FEES Lett. 101, 15-20. Sternlicht, H., and Ringel, I. (1979).J.Biol. C h . 254, 10540-10550. Sternlicht, H., Ringel, I., and Szacz, J. (1980).J . Biol. Chem. 255, 9138-9146. Summers, K., and Kirschner, M. W. (1979).J.Cell Biol. 83, 205-217. Taylor, E. W. (1965).J.CellBiol. 25, 145-160. Telzer, B. R., and Rosenbaum, J. L. (1979).J.Cell Biol. 81,484-497. Terry, B. J., and Purich, D. L. (1979).J.Biol. C h . 254,9469-9475. Terry, B.J., and Purich, D. L. (198O).J.Biol. C h . 255,10532-10536. Terry, B.J., and Purich, D. L. (1982).Adu. Enzymol. Re&. Areas Mol. Bbl. 53, 113-161. Thompson, W. C.(1982).Methods Cell Biol. 24, 235-255. Tdney, L. G.,Bryan, J., Bush, D. J., Fujiwara, K., Mooseker, M. S., Murphy, D. B., and Snyder, D. H. (1973).J . Cell Bzol. 59, 267-272. Timasheff, S. N. (1981).In “Protein-Protein Interactions” (C. Frieden and L. W. Nichol, eds.), pp. 315-336. Wiley, New York. Timasheff, S. N., and Grisham, L. M. (1980).Annu. Rev. Biochem. 49,565-591. Tominaga, S., and Kaziro, Y. (1982).In “Biological Functions of Microtubules and Related Structures (H. Sakai, H. Mohri, and G. G. Borisy, eds.), pp. 11-22. Academic Press, New York. Tsuchiya, T., and Nagai, Y. (1983).Bi0phys.J. 41, 275-281. Valenzuela, P., Quiroga, M., Zaldivar, J., Rutter, W. J., Kirschner, M. W., and Cleveland, D. W. (1981).Nature (London) 289,650-655. Vallee, R. B. (1982).J . Cell Biol. 92, 432-442. Ventilla, M., Cantor, C. R., and Shelanski, M. (1972).Biochemishy 11, 1554-1561. Voter, W. A., and Erickson, H. P. (1982).J.Ultrastrut. Res. 80,374-382. Weber, K., Pollack, R.,and Bibring, T. (1975).Proc. Natl. Acad. Sci. U.S.A. 72,459-463. Wegner, A. (1976).J.Mol. Bwl. 108, 139-150. Wegner, A., and Engel, J. (1975).Biophys. C h . 3,215-225. Weingarten, M. D.,Lockwood A. H., Hwo, S.,and Kirschner, M. W. (1975).Proc. N ~ t l . A d . Sci. U.S.A. 72, 1858-1862. Weisenberg, R. (1972).Science 177, 1104-1 105. Weisenberg, R., and Deery, W. J. (1976).Nature (London) 263, 792-793. Weisenberg, R., and Timasheff, S. N. (1970).Biochemistry 9,4110-4116. Weisenberg, R., Borisy, G. G., and Taylor, E. W. (1968).Biochemistry 7, 4466-4479. Weisenberg, R., Deery, W. J., and Dickinson, P. J. (1976).Biochemistry 15, 4248-4254. White, H. D., Coughlin, B. A., and Purich, D. L. (1980).J.Biol. C h . 255, 486-491. Williams, R. C.,and Detrich, H. W. (1979).Biochemirtty 18, 2499-2503.

212

DANIEL L. PURICH AND DAVID KRISTOFFERSON

Wilson, L.,and Friedkin, M. (1967). B i o c k k h y 6,3126-3135. Wilson, L.,Anderson, K., and Chin, D. (1976). In “Cell Motility” (R. Goldman, T. Pollard, and J. Rosenbaum, eds.), Book C, pp. 1051-1064. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Wilson, L.,Snyder, K. B.,Thompson, W.C., and Margolis, R. L. (1982). Methods Cell Bwl. 24,159-170. Witman, G. B. (1975).Ann. N.Y.Acad. Sn‘. 468, 178-191. Wyman, J., Jr. (1964). Adv. Protein C h . 19, 223-285. Zackroff, R. V., Weisenberg, R. C., and Deery, W. J. (1980).J. Mol. Biol. 189,641-677. Zeeberg, B., and Caplow, M. (1979). B k W f t y 18,3880-3886. Zeeberg, B., and Caplow, M. (198l).J. B i d . C h . 266, 12051-12057. Zeeberg, B., Reid, R., and Caplow, M. (1980).J. Biol. C h . 466,9891-9899.

SPECIES ADAPTATION IN A PROTEIN MOLECULE’1* By M. F. PERUTZ MRC Laboratory of Molecular Blology, Cambridge, England

. . . . . . . . . . . . . . . . . . . . . . . Experimental Tests of Theory of the Root Effect. . Amphibia . . . . . . . . . . . . .

I. Introduction.

11. Fish Hemoglobins 111.

IV. V. VI. VII. VIII.

Reptiles . . . . . . . Birds. . . . . . . . Mammals . . . . . . Species Adaptation in Enzymes Discussion . . . . . . References . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

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

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

213 217 225 225 227 232 233 235 236 240

The allosteric properties of hemoglobins, especially their responses to ligands other than oxygen, vary widely in different classes of vertebrates. Knowing the stereochemistry of the cooperative effects in human hemoglobin, it is possible to infer the stereochemical basis of these variations from the changes in amino acid sequence. The results indicate that the tertiary and quaternary structures of deoxy- and oxyhemoglobin have remained almost invariant during vertebrate evolution, and that most of the amino acid replacements between species are functionally neutral. Adaptations leading to a response to new chemical stimuli have evolved by only a few (1-5) amino acid substitutions in key positions. Once such a response has become superfluous, it may be inactivated, not necessarily by a reversal of one of the original substitutions, but by any other that happens to inhibit it.

I. INTRODUCTION Species adaptation at the molecular level is a virgin field. In 1978 Lewontin complained that “it has proved remarkably difficult to get compelling evidence for changes in enzymes brought about by selection, not to speak of adaptive changes” (1). Such evidence has recently been This article has been previously published in Molecular Biology and Evolution, Volume 1, No. 1, pp. 1-28, copyright 1983 by the University of Chicago. It is reproduced here by permission. * This paper is dedicated to Dr. John T. Edsall, the great pioneer of protein chemistry, on his eightieth birthday. 213 ADVANCES IN PROTEIN CHEMISTRY, Val. 36

ISBN 0-12-034236-7

2 14

M. F. PERUTZ

gathered for hemoglobin whose response to different chemical stimuli varies widely in vertebrates living in different environments. How did this adaptation come about? Is it the result of changes in tertiary and quaternary structure? Has it been brought about by the gradual accumulation of minor mutations, each producing a small shift in chemical affinity, or by a few amino acid substitutions in key positions? Have similar changes in chemical affinity in different species been brought about by the same amino acid substitutions or by different ones? What selective advantage does possession of multiple hemoglobin genes offer to a species? Are most amino acid substitutions between species functionally significant and have they evolved by Darwinian selection (2),or are they caused by random fixation of neutral or nearly neutral mutations (3)? We can now at least begin to answer these questions. The hemoglobins of two primitive classes of vertebrates, lampreys and hagfish, are monomeric when oxygenated and oligomeric when deoxygenated. They contain only one type of chain. Hemoglobins of the higher vertebrates are tetramers made up of two a and two p chains, each containing between 141 and 147 amino acid residues. Each chain is linked to one heme. The a chain contains seven and the p chain eight helical segments, designated A to H; they are interrupted by nonhelical segments marked AB, BC, etc. A short nonhelical segment, called NA, precedes the first helix and another, called HC, follows the last (Fig. 1). The four chains are arranged tetrahedrally around a twofold symmetry axis that runs along a water-filled cavity. Uptake and release of oxygen is accompanied by small changes in tertiary structure of the segments surrounding the hemes and by a large change in quaternary structure involving a rotation of one ap dimer relative to the other by 15", together with a relative shift of 1 A. The rotation widens the cavity between the two p chains so that the cationic groups that form its lining can bind organic phosphates. The two quaternary structures will be referred to as the oxy or relaxed (R)and the deoxy or tense (T) structures (4). Most vertebrate hemoglobins react cooperatively with molecular oxygen. (An exception discovered recently is that of the reptile Sphenodon.) Their oxygen equilibria are influenced by various chemical factors, known as heterotropic ligands, and by temperature. In human erythrocytes, these ligands are H+, C1-, COs, and ~-2,3-bisphosphoglycerate (DPG); they all reduce the oxygen affinity of hemoglobin in a physiologically advantageous manner. In addition, the uptake of protons on release of oxygen, known as the alkaline Bohr effect, facilitates the transport of CO:, from the tissues to the lungs in the form of HCOs-. In 0.1 M C1-, the uptake of protons vanishes at pH 6; below that pH, protons are released on release of oxygen. This is known as the acid or reverse Bohr

SPECIES ADAPTATION IN A PROTEIN MOLECULE

215

FIG. 1. Tertiary structure of globin chain, showing helical and nonhelical segments and the position of the heme.

effect; it does not seem to have any physiological function in mammals, but is dominant, even at physiological pH, in certain fish and amphibia. In humans, a minor part of the CO:, is carried from the tissues to the lungs in the form of carbamino ion bound to hemoglobin. DPG is an allosteric regulator that lowers the oxygen affinity and thereby facilitates the release of oxygen to the tissues. The reaction of human hemoglobin with oxygen is exothermic, so that the oxygen affinity decreases with increasing temperature (5-7). The effects of heterotropic ligands arise because the T structure has a low affinity for oxygen and a high one for H+, C1-,organic phosphate, and Cop. In the R structure, these relative affinities are reversed. The transition between the two structures gives rise to the cooperativity of oxygen binding. Hill’s coefficient n provides a measure of the cooperativity. In terms of allosteric theory, the equilibrium curve is described by the oxygen dissociation constants KT and K Rof the two alternative structures, and by the equilibrium constant L = [T]/[R] in the absence of oxygen. Heterotropic ligands strongly influence K T and L,

216

M. F. PERUTZ

/

COO-----NH,+-CS

cDZPl Glu-COO------Gur+-FG4

a1

Lys

a1 Are

FIG.2. Salt bridges in human deoxyhemoglobin (6). For details of the stereochemistry see ref. 4.

but exercise only a weak influence on KR (6,7). (KTand K R are usually expressed in Torr-I). Structural analysis of human hemoglobin has shown that all heterotropic ligands lower the oxygen affinity because they stabilize the T structure. They do so by forming salt bridges within and between the subunits (Fig. 2). The binding sites of the heterotropic ligands are as follows: DPG binds in the cavity between the two /3 chains to Val NA1(1)fl,9 His NA2(2)p, Lys EF6(82)P, and His H21(143)P (Fig. 3). Bohr protons are bound by certain cationic groups that have their pK values raised in the R + T transition because they are free in the R structure, but form salt bridges with carboxylates, CI-, or phosphate in the T structure. In the absence of DPG, Bohr protons bind mainly to Val NA1( 1)a, Lys EF6(82)& and His HC3( 146)p, forming the hydrogen bonds shown in Fig. 2. In the presence of DPG, Bohr protons are also bound to His NA2(2)p and His H21(143)P. In vim, all these residues therefore contribute to the alkaline Bohr effect. Carbon dioxide forms carbamino compounds with Val NAl( l)a and p. Chloride is bound in the central cavity by Val N A l ( l)a, Lys EF6(82)P, and, probably, at other minor sites. In the absence of DPG, His H21(143)P contributes to the acid Bohr effect (6, 6a, 44).

'

In this notation, the first number gives the position of the residue within the helical or nonhelical segment, and the second number, in parentheses, gives its position in the chain as a whole.

217

SPECIES ADAPTATION IN A PROTEIN MOLECULE

His

IX

FIG.3. Binding site of 2,3-bisphosphoglycerate (DPG) between the two /3 chains in human deoxyhemoglobin. On transition to the oxy structure, the a-amino groups move apart and the EF segments close up, so that the complementarity of the binding site to DPG is lost (103). 11.

FISHHEMOGLOBINS

Many teleost (bony) fish use hemoglobin for both respiration and secretion of oxygen into the swim bladder and the eye (8-11). Each organ possesses an elaborate vascular network, the rete, which converts a gradient of increasing lactic acid concentration into a gradient of decreasing oxygen concentration by countercurrent circulation, thus causing oxygen to be discharged. This acid-activated discharge is known as the Root effect, but it is best considered as an enhanced version of the alkaline Bohr effect that causes so drastic a drop in the oxygen affinity at low pH that several hundred atmospheres of oxygen pressure fail to saturate the hemoglobin with oxygen (12). The effect arises because the response of teleost fish hemoglobins to heterotropic ligands is markedly different from that of most mammalian hemoglobins. Whereas in human hemoglobin Hill's constant n remains at or just below 3.0 independent of pH, in fish hemoglobins it drops to unity, or even below unity,

218

M. F. PERUTZ

near pH 6. This disappearance of cooperativity is due to inhibition of the allosteric transition from the T to the R structure at acid pH. Alkaline pH tends to inhibit the transition from the R to the T structure instead, so that n rises to a maximum of only just above 2.0 near neutral pH and in many, but not all fish, falls again at alkaline pH. Both the span of oxygen affinities covered by the pH range 6-9 and the magnitude of the alkaline Bohr effect (Alog p5,JApH) are larger in fish than in mammalian hemoglobins. In carp hemoglobin, for instance, K, , the association constant of the first oxygen molecule to be taken up, drops 100-fold between pH 9 and 6.5, compared to eightfold in human hemoglobin; over the same pH range K 4 , the association constant of the last oxygen, drops 10-fold in carp compared to 1.2-fold in human. On oxygenation, human hemoglobin in deionized water releases no more than one proton per tetramer, whereas carp releases 3.6. On the other hand, the enhancement of the alkaline Bohr effect by C1- or phosphate is smaller in carp than in human hemoglobin (13-20). Though the oxygen dissociation constants of the a and P subunits of human hemoglobin differ but slightly, those of fish hemoglobins at acid pH may differ by as much as 100-fold which allows one pair of subunits to secrete oxygen into the swim bladder and the eyes against hydrostatic pressures as high as several hundred atmospheres. (For details of the properties summarized here, see refs. 13-22.) What makes the allosteric properties of these fish hemoglobins with Root effects so different from those of mammalian hemoglobins?Seeing that the amino acid replacements between, say, human and carp hemoglobin number over 140, this seems a forbidding problem. But systematic examination on my atomic model of human hemoglobin led me to the conclusion that the Root effect is caused mainly by a single amino acid replacement, that of cysteine F9P in mammals by serine in fish. My reasoning was as follows. In the T structure of all hemoglobins that exhibit an alkaline Bohr effect, the imidazole of the C-terminal histidine HC3P forms a salt bridge with Asp or Glu FGlP; this raises the pK of the histidine so that protons are bound. The C-terminal carboxyl of the same histidine forms a salt bridge with the invariant Lys C5a, whether the hemoglobin shows a Bohr effect or not. In mammals, the sulfhydryl of Cys F9P is in van der Waals contact with that oxygen of the C terminus that is not bonded to Lys C5a, but it forms at most a very weak hydrogen bond with that oxygen. The - O H of serine, on the other hand, forms strong hydrogen bonds. The atomic model shows that the Ser F9P -OH is so placed that it can donate a hydrogen bond to the free terminal oxygen atom of His HC3P and accept a hydrogen bond from the peptide -NHof His HCJP (Figs. 4 and 5 , Table I). These additional

a

C

GH GH

FG 1 d GH His HC3

Glu FG 1

C5a

FIG.4. C-terminal salt bridges and the role of the residue F9P in the T and R structures of mammalian and fish hemoglobins. (a) Human T structure; (b) human R structure; (c) carp T structure; (d) carp R structure. The letters F, G, and H denote helical segments: FG, GH, and HC denote nonhelical segments. The same notation is used in Fig. 5 (23).

FIG.5. Stereodiagram showing the bonds at the C terminus of the P chain in the T structure of teleost fish hemoglobin. The amino group of Lys C5a donates a hydrogen bond to one of the carboxylate oxygens of His HCSP, and the - O H of Ser F9P donates a hydrogen bond to the other oxygen. The imidazole donates a hydrogen bond to the of His HC3 donates a hydrogen bond to carboxylate of Glu FGl. The main chain -NHthe lone pair electrons of the -OH of Ser F9. Note the direct link from the heme iron via His F8 and Ser F9 to the C terminus, which would therefore feel the movement of the iron toward the porphyrin that occurs on oxygen binding (23).

Irnportunt Amino Acid R~~~

Position of amino acid residue" and characcerirtics

TABLE I and Some Charactoistia Distinguishing Mammalian. Amphibian. and Fish Hemoghbinv

Terrestrial frogs

Human

Trout IV

Trout I

Glu

Ser Glu Arg Gln

ASP LYS Val Ser Glu Arg Gln

Carp

Aquatic frog Xcnopus

Tadpole of

Xmopuc

R. esm-

R. cotcslcnta be@M (Europe) (America)

Tadpole of R.&-

Shark (Hctcro-

POT&-

Lungfish (Lepidosiren

beia~

jachsoni)

paradma)

His LYS Glu

His

LYS LYS

His LYS His Ser Glu

clonhrs

/3 Chain

NA2 EF6 F6 F9 FG1 H21 HC 1 HC3 a Chain NA 1

c3 Root effect Allosteric effector Bohr effect References

Phe

GlY LYS LYS Ser Glu LYS G1Y His

His LYS Thr Ala Glu His GlY Phe

Ac-Ser Gln

Ac-Ser Gln

Leu LYS

? ?

+

+

-

+

DPG Normal

ATP Strong

ATP Strong

DPG Normal

?

?

Absent

?

Weak

105

24

26, 131

106, 130

46, 132

50

His LYS Glu CYS Asp His LYS His

Glu LYS Val

His

His

Val Thr

Ac-Ser Gln

-

For meaning of the symbols see Figs. 1-6.

Leu Glu

Ala

Asn Ser

Ala Asn

Ser His

Glu LYS Glu His

Ac-Ser Gln

Ala

His

GlY ?

? ?

Ala

Ac-Ser

Glu His Met-Arg GlY -

?

?

Weaknormal 107, 108 109

? Reversed

Weak

ATP Normal

110,111

33,112

57

?

SPECIES ADAPTATION IN A PROTEIN MOLECULE

22 1

hydrogen bonds would stabilize the C-terminal salt bridges in the T structure, thereby raising the allosteric constant L and the pK of His HC3/3, and lowering the oxygen binding constant KT, especially of the /3 chains (23). Experimental tests of the theory are described below. Fish hemoglobins exhibit at least one other feature that is likely to shift their allosteric equilibria toward the T structure. In the R structure of mammalian hemoglobins, the terminal carboxyl of His HC3P forms an external salt bridge with Lys HC1 of the same /3 chain. In teleost fish hemoglobins, this external lysine is replaced by a glutamine, which would not bind the carboxyl in an aqueous environment, so that less energy is needed to move His HC3P toward Asp or Glu FGlP on going from the R to the T structure (Figs. 4 and 5). The constellation of polar groups described here has been found in the hemoglobins of carp (24), goldfish (25), trout IV (26), and spot (Catostomzls clurkii) (27), all of which exhibit strong Root effects. Mammals and frogs use DPG as an allosteric effector, but teleost fish use ATP, GTP, or inositol pentaphosphate and, probably, also lactate (28, 29). Whereas in mammals DPG lowers the oxygen affinity more than ATP, the opposite holds in teleost fish (30). Mammalian hemoglobins whose oxygen affinity is regulated by DPG have a hydrogen donor side chain in position NA2/3 (His, Gln, or Asn) and they have His in position H21/3 (Fig. 3). Teleost fish have either Glu or Asp in position NA2P and Arg at H21/3. Substitution of those side chains in the atomic model of human deoxyhemoglobin produces a constellation of charged groups stereochemically complementary to strain-free models of ATP or GTP (Figs. 6 and 7). The model suggests that, when ATP is bound, in the physiological range of pH, the carboxylate of either Glu or Asp NA2/31 accepts a hydrogen bond from the N-6 amino group of adenine; the amino group of Val 1/32 and the guanidinium group of Arg H21/31 each donate a hydrogen bond to the y phosphate, while Lys EF6/31 and p2 donate hydrogen bonds to either of the other two phosphates, thus neutralizing the four negative charges of ATP (23). Since this structure was first proposed, Braunitzer and co-workers have determined the amino acid sequence of rhinoceros hemoglobin (23a). Its allosteric effector site shows only a single substitution compared to that of human hemoglobin-His NA2P + Glu-yet ATP lowers its oxygen affinity more than DPG, and GTP lowers it more than ATP, just as in teleost fish (R. Baumann, unpublished observations). This observation supports the hydrogen bond between N-6 of the adenine and Glu NA2 proposed in Fig. 6; in fact it can hardly be explained without that bond. The stereochemistry of ATP and GTP in carp deoxyhemoglobin has

222 L

M. F. PERUTZ R

FIG. 6. Stereodiagram of the suggested ATP binding site of fish hemoglobins.

recently been determined by nuclear magnetic resonance using the timedependent transferred nuclear Overhauser effect (1 15, 116). This experiment showed the purine to be in the anti conformation with respect to the ribose in both trinucleotides; the ribose pucker is 3' endo; the 5'0-5'-C-4'-C-3'-C torsion angle is trans, the a-P-5'-0-5'-C-4'-C angle is coplanar trans, a-P-0-p-P and /3-P-09-P are as in the free nucleotide, probably all trans, so that electrostatic repulsion between the phosphates is minimized. The results confirm the structure of ATP in carp deoxyhemoglobin proposed by Perutz and Brunori (23), except for a small change in the angle of tilt of the purine (Fig. 6). T o convert the structure of ATP into that of GTP, the entire nucleotide has to be turned by roughly 180" about its long axis. This leaves the y phosphate bound to Val l a of subunit and allows N-2 of the guanine to donate a hydrogen bond to Glu NA2 of subunit p2. In addition, the a-amino group of Val NAl& can donate a hydrogen bond to the 2'-0 of the ribose (Fig. 7). This extra hydrogen bond could contribute the 1.6 kcal of binding energy needed to account for GTP lowering the oxygen affinity of carp hemoglobin twice as much as ATP (23c). It is thought that during fast movement, the pH at the gills may drop too low for efficient oxygen uptake by those hemoglobins that exhibit the Root effect. T o ensure a continued oxygen supply, trout and some

SPECIES ADAPTATION IN A PROTEIN MOLECULE

223

FIG. 7. Suggested stereochemistry of GTP binding by fish hemoglobins.

other fast-swimming fish have two kinds of hemoglobin, those that respond to heterotropic ligands and those that do not. In trout, the hemoglobin that exhibits the Root effect has been assigned the number IV, while the two hemoglobins that fail to respond to heterotropic ligands are called I and I1 (12). Component 111 makes up only 3% of the total and its ligand binding properties are not known. In trout I, all the polar residues involved in both the alkaline and acid Bohr effects and in phosphate binding are replaced by neutral ones: Lys EF6P is replaced by Leu, Ser F9P by Ala, Glu FGlP by Am, Arg H21P by Ser, and His HC3P by Phe (Table I). The sequences of trout hemoglobins I and IV show many other replacements, but none of these is of a kind likely to have a significant effect on the affinity for heterotropic ligands. The properties just described are not typical of all fish. Among elasmobranchs, skate hemoglobin has no appreciable Bohr effect and C1raises rather than lowers its oxygen affinity. Like trout I, torpedo hemoglobin fails to respond to organic phosphates (31, 32). These properties cannot yet be interpreted due to lack of amino acid sequences. Hemoglobin of the shark Heterodontus portusjacksoni, a cartilagenous fish without a swim bladder that must swim continuously in order not to sink, exhibits a weak alkaline Bohr effect (33). Position F9P is occupied by Ala; residue FGlP is Glu, which could produce a normal alkaline Bohr effect by forming a strong salt bridge with His HC3P if it were not for a Lys in position F6P, one turn of the IT helix away, competing with His HC3P for that salt bridge. The South American lungfish Lepadosiren paradoxa

224

M. F. PERUTZ

also lacks a swim bladder but keeps afloat by snapping up air to fill its lung; its hemoglobin exhibits no Root effect and has an alkaline Bohr effect similar in magnitude to that of human hemoglobin (34). It does have Ser at F9P and Glu at FGlP, but, as in Xenopus and in the shark, its salt bridge with His HC3P is weakened by competition from residue F6P, in this instance a His. Lungfish hemoglobin has an additional residue, a Met, at the amino end' of the a chain (Table I). This may contribute to the alkaline Bohr effect by binding C1- jointly with Arg HC3 of the opposite a chain, as Val la does in mammals. Most of the fish discussed so far are cold-blooded, but fast-swimming sharks can maintain their bodies 7-10°C and tuna (Thunnus thynnus) up to 15°C above the water temperature (35, 36); cooling of the muscles is minimized by a counter current exchange system between the arterial and venous circulations that transfers metabolic heat from the veins to the cold blood arriving in the arteries from the gills. In most species, the oxygen affinity of hemoglobin drops with rising temperature because the reaction of heme with oxygen is exothermic; if this were true also in tuna, then heating of the cold arterial blood would cause some of its oxygen to dissociate and be transferred to the nearby veins unused. T o reduce this waste, tuna have evolved a hemoglobin in which the reaction with oxygen is endothermic (36, 37). This reversal arises because the intrinsic heat of oxygenation of the hemes is exceeded by the heat absorbed in the T + R transition. In human hemoglobin, the transition absorbs about one-half of the -66 kcall4 mol 0 2 released on oxygenation; in trout IV it absorbs about two-thirds. T o produce a reversed temperature coefficient, the T-* R transition in tuna absorbs about 100 kcal (37). Where do the extra 30 kcal come from? The main source of heat in the allosteric transition comes from the formation of hydrogen bonds. Taking the enthalpy of hydrogen bond formation in the dimerization of N-methylacetamide in benzene of 3.6 kcaYmol as a measure (38), the T structure would have to contain four additional and the R structure four fewer hydrogen bonds to achieve the desired result. Taking the enthalpy of 1.5 kcaYmol for the disruption of a CO-HN bond in water as a measure (39), these numbers would rise to the improbable value of 10. Another enthalpy sink could be generated if the sum of the van der Waals interactions in the R structure were less than in the T structure, so that heat of fusion would be absorbed in the T + R transition (40).This is known to be true in human hemoglobin. Lest hemoglobin be regarded as indispensable to vertebrate life, let me draw attention to the existence of three species of large antarctic fish devoid of erythrocytes or blood pigment (41). The amount of dissolved oxygen their very cold colorless blood can carry is only one-eighth of

SPECIES ADAPTATION IN A PROTEIN MOLECULE

225

that of two other antarctic fish with red blood, but that seems to be adequate for the sluggish existence of these large fish. Experimental Tests of Theory of the Root Effect When X-ray analysis had indicated that a salt bridge formed by His HC3P in the T structure may be responsible for a major part of the alkaline Bohr effect in horse and human hemoglobins, Kilmartin set out to test that prediction by preparing a hemoglobin from which that Cterminal histidine had been enzymatically cleaved. X-Ray analysis showed that this cleavage had no significant effect on the structure of the rest of the hemoglobin molecule. Nevertheless, it halved the alkaline Bohr effect (42, 43). This observation and many subsequent ones supported the dominant contribution of this one specific residue (44). The latest is the discovery of an abnormal human hemoglobin with a drastically reduced alkaline Bohr effect, due to the substitution Asp FGlP + Asn which neutralizes the anionic arch of the salt bridge, just as the substitution Glu FGlP --* Asn does that occurs between trout IV and I hemoglobins (1 17). If the Root effect in fish hemoglobins is due largely to the extra stabilization conferred by Ser F9P on the salt bridges of His HCSP, then it should be inhibited by enzymatic cleavage of His HC3. Parkhurst, GOSS, and Perutz tested that prediction by preparing des-His carp hemoglobin (45). This halved the alkaline Bohr (Root) effect, lowered L, raised K T , and also changed the kinetics of ligand binding. On approach to pH 6, native carp hemoglobin shows a sharp rise in the rate of oxygen dissociation and fall in the rate of CO recombination characteristic of the Root effect. Both the rise and fall are much diminished in des-His carp hemoglobin. All these observations supported the theory. A further test became possible when the hemoglobin of the South African frog Xenopus was found to have a serine in position F9P (46). According to the theory, this hemoglobin should exhibit a Root effect at low pH, and this was confirmed by Perutz and Brunori (23).They found that it has a much lower oxygen affinity than human hemoglobin, consistent with a more stable T structure; its Hill constant tends toward unity at pH 6, and its kinetics of CO recombination at low pH are similar to those of hemoglobins exhibiting a Root effect. However, its Root effect is not as strong as that of carp hemoglobin. 111. AMPHIBIA

To produce a strong alkaline Bohr effect and a Root effect at acid pH, Xenopls hemoglobin has conserved the essential Ser F9P, Glu FGlP, and

226

M. F. PERUTZ

His HC3/3, and it has Gly rather than Lys in position HClP to prevent anchoring of the C terminus in R structure (Table I). On the other hand, it has a Lys at F6P, one turn of the T helix away from the Glu FGlP and capable of competing with His HC3P for the essential salt bridge with that Glu (Figs. 4 and 5). This competition may weaken its alkaline Bohr effect compared to that of carp. Xenopw hemoglobin is exceptional among amphibians for its strong alkaline Bohr effect; in 0.1 M C1- other amphibian hemoglobins show either weak alkaline or only weak acid Bohr effects. The major hemoglobin component of the large American bullfrog Rana catesbeiana, when stripped and in dilute solution, shows a Bohr effect of Alog p~dApH= -0.16, compared to -0.56 in Xenopls (47). Position F9P is occupied by Ser, as in Xenofiw, but its influence is counteracted by the substitution of Gly for Glu in position FGlP. In consequence, His HC3P cannot have its pK raised in the T structure by the usual strong salt bridge with Glu FGl& but can only interact more weakly with Glu F6P which is one turn of a T helix away from residue FGlP (Table I, Figs. 4 and 5). The two major fractions of the adult hemoglobin of the European frog Rana escuhta also exhibit only weak alkaline Bohr effects (48). The sequence of the major component of the P chain is similar to that of R . catesbeiana; position FGlP is occupied by Asn rather than Gly, which again inhibits the strong salt bridge with His HC3P. His H21& which contributes to the acid Bohr effect in human hemoglobin A, is replaced by Lys in all three adult frog hemoglobins (Fig. 3), and their acid Bohr effects are weak or absent. Tadpole hemoglobin of R. catesbeiana, in the absence of phosphates, has only an acid Bohr effect (49). Here, Ser F9P is replaced by Ala and Glu FGl by Asn, and Ser la, the other major source of the alkaline Bohr effect, is acetylated, so that the alkaline Bohr effect is completely inhibited. On the other hand, His H21P is present, consistent with its role in the acid Bohr effect of human hemoglobin. His H21P is also present in Xenopls tadpole hemoglobin; this has Phe in position HC3P, which would inhibit most of the alkaline Bohr effect contributed by the P chains (50). The amino acid sequence of the a chain is not yet known. If its a-amino group were acetylated, as in the tadpole of R. catesbeiana, then Xenopls tadpole hemoglobin would also exhibit only an acid Bohr effect. The /3 chains of both tadpole species carry the same residues at the phosphate binding site as does adult human hemoglobin; organic phosphates, if present, would therefore be expected to overcome the acid Bohr effect and produce a weak alkaline one instead. In another aquatic frog, Amphiuma, the adult hemoglobin has an acid Bohr effect at physiological pH, converted in vivo to a weak alkaline one

SPECIES ADAPTATION IN A PROTEIN MOLECULE

227

by organic phosphate (51). The adult hemoglobin of the crested newt, Triturus cristatus, also exhibits an acid Bohr effect at physiological pH, but in this amphibian it is maintained in the hemolysate; even so, the higher concentration of organic phosphates in the red cell may convert it to a weak alkaline Bohr effect in vivo ( 5 2 ) .Blood of the frog Telmatobius c u h that lives in Lake Titicaca at an altitude of 3800 m has a higher oxygen affinity than that of any other frog (53), but the oxygen equilibrium curves of its purified hemoglobin are still unknown. It will be interesting to analyze the stereochemicalbases of all these unusual properties once amino acid sequences become available. Why is the alkaline Bohr effect of the bloods of the Port Jackson shark, of the newt, of Amphiuma, and of the anuran tadpoles so much weaker than that of mammals? In the blood of air breathers the partial pressure of C 0 2is about 40 Torr and a substantial Bohr effect is needed for its transport from the lungs to the tissues, but in the blood of water breathers the partial pressure of CO2 is only 2-4 Torr, because it is 28 times more soluble in water than is oxygen and it diffuses out of the body through the gills and the skin. Having no swim bladder and no rete to pump oxygen into the eye, the shark and the tadpoles can therefore get along with only a weak alkaline Bohr effect. The same is true of the aquatic frog Amphiumu, which is regarded as a permanent larva, even though the adult has lost its external gills and shows other evidence of partial metamorphosis (51, 54-56). Judging by the strong acid Bohr effect, the switch from tadpole to adult hemoglobin may not have occurred in this amphibian. On the other hand, the adult terrestrial bullfrog R. catesbeiana loses only a small part of its COP through the skin, even when submerged, and relies on ventilation through its lungs. Its blood therefore needs a larger alkaline Bohr effect than that of Amphiurn (55). The same appears to be true of L. purudoxu which lives in hot, stagnant water and surfaces to breathe (34,57). Xenofius is anomalous. It is an aquatic frog with a low blood [COs],capable of respiration through the skin or even of switching to anaerobic metabolism (56). At this stage, I have not found any satisfying explanation for the large alkaline Bohr effect of its hemoglobin at physiological pH, nor for its Root effect at acid pH. IV. REPTILES Crocodilians are able to stay under water for as long as an hour without coming up to breathe, but they lack the high concentrations of myoglobin that provide diving mammals and birds with large stores of oxygen. Instead, crocodilians reduce oxygen consumption by shutting

228

M. F. PERUTZ

off the circulation to their muscles so that oxygen supply is restricted to their brain and viscera. Even so, they need to use as much as possible of the oxygen stored in their lungs and their blood. This appears to be accomplished by the unusual allosteric properties of their hemoglobin, which hardly responds to the normal heterotropic ligands (protons, C 0 2 , C1- or organic phosphates), but responds strongly only to bicarbonate ion (Fig. 8). Crocodilian hemoglobins bind two equivalents of bicarbonate ion per tetramer which lower the oxygen affinity as much as Pg-inositol (inositol hexaphosphate) does in human hemoglobin (5860). Model building suggested that the bicarbonate binding sites lie in the cavity between the two P chains where organic phosphates or carbamino C02 are bound in other species; each HC03- is bound by Lys EF6 and Glu HCl of one chain together with the N-terminal residue of its partner chain (Fig. 9). In caiman hemoglobin, the N-terminal sequence is Ser-Pro-Phe-, compared to Val-His-Leu- in human hemoglobin (Table 11). The human tripeptide is straight (Fig. 3), but the caiman tripeptide is forced to turn a corner at the proline. The turn brings the N-terminal serine within reach of the bicarbonate ion, so that one of the bicarbonate oxygens can form a salt bridge with the a -NHs+ and can also accept a hydrogen bond from the serine -OH. The second bicarbonate oxygen can form a salt bridge with Lys EF6P and the hydroxyl can donate a hydrogen bond to one of the carboxylate oxygens of Glu HClP. In the hemoglobins of the Nile crocodile and the Missis-

80 90

70

1 1 .

.

I

. . .

I

'

*

'

.

1

'

.

1

-

6050

-

40

-

30-

20 10

-

-Caiman

-

-----Horse I

.

.

.

.

I

a

.

1

FIG. 8. Oxygen equilibrium curves of caiman and horse hemoglobin with and without COs.pCOs = 40 Torr, 25OC, I = 0.2 M,pH 7.2 for caiman (60),pH 7.4 for horse (104).

SPECIES ADAPTATION IN A PROTEIN MOLECULE

\A

229

I A

FIG. 9. Stereo drawing of proposed bicarbonate binding site between the two chains of caiman deoxyhemoglobin. The central sign marks the dyad symmetry axis. Bicarbonates and their binding residues are underlined. Capital letters mark helical and interhelical segments (6 1). Residues are marked in sequential, rather than structural notation. The following List gives the structural numbers with the sequential ones in parentheses: Ser NA1 (1); Pro NA2 (2); Phe NA3 (3); Ser A1 (4); Ala A2 ( 5 ) ; His A3 (6); Lys EF6 (82); Glu HCl (144); Tyr HC2 (145); His HC3 (146); all 8.

sippi alligator, where the N-terminal sequence is acetyl-Ala-Ser-Phe-, of acetylalanine could donate a hydrogen bond to 0-1 of the a -NHthe bicarbonate at the same distance as the a -NHs+ does in caiman, provided that the chain bends at Ser 2 at the same angle as it does at Pro 2 in caiman. Consider now the loss of allosteric inhibition by organic phosphates, carbamino C02, and C1-. The loss of affinity for organic phosphates is accounted for by the replacements His NA28 + Pro or Ser and His H218 --* Ser (Fig. 3). In human deoxyhemoglobin, carbamino COP bound to Val la accepts a hydrogen bond from Ser H14a. In the three crocodilian hemoglobins this is replaced by Ala. This replacement may also inactivate one of the strong C1- binding sites which probably coincides with the COP site. In human deoxyhemoglobin, another pair of C02molecules competes with organic phosphates for binding to Val 18; when bound, each carbamino Cop is stabilized by a salt bridge to Lys 828. In caiman hemoglobin, the bend in the chain forced by Pro 2 inhibits that bridge; in the other two crocodilian hemoglobins, the blocking of Val 18 inhibits binding of C 0 2 (61).

TABLE I1 Amino Acid rep^^ for Allostnic Control in Crocodilian Hemoglobins"

SpeCieS

Other bony vertebrates

Nile Mississippi Caiman crocodile crocodile

Position

Human

NAlg

Val

Val or Gly

Ser

Ac-Ala

Ac-Ala

In human and many other vertebrate deoxy-Hbs, the a-aminogroups bind either organic phosphate or carbamino COP,which in turn forms a salt bridge with Lys 82 of the same fl chain; in crocodile Hb, blocking of a-NHs' or Pro in position 2 inhibits these functions

NA28

His

Gln, Asn, Glu, His, Asp, Met

Pro

Ser

Ser

In human Hb, His, and, in many other vertebrates, Gln or Asn bind organic phosphates

EF6g

LYS

LYS

Lys

Lys

LYS

Lys binds diffusible anions in the internal cavity of deoxy-Hbs; invariant in all vertebrate Hbs except trout I which lacks all oxygen-linked functions and where it is replaced by Leu

Lys, Arg, Ser, His

Ala

Ala

Ala

His, Lys, or Arg form salt bridges with organic phosphates

Arg, Ala, Ser, Gln, LYS

Glu

Glu

Glu

In human Hb, Lys is free in deoxy and bound to His HC3 in oxy; Glu can accept an H-bond from HCOswhose charge is compensated by Lys 82

Very variable

Ala

Ala

Ala

In human Hb, - O H of Ser forms an H-bond with carbamino COPor CI- bound to Val la, but this bond is absent in many species

H14a

Ser

~

a

For full sequences see ref. 113.

~

~____

Functions

SPECIES ADAPTATION IN A PROTEIN MOLECULE

23 1

The decrease in oxygen affinity brought about by the interaction of crocodilian hemoglobins with bicarbonate ensures that oxygen is released from the blood to the tissues at a relatively high partial pressure of oxygen. If the hemoglobin were insensitive to bicarbonate, the venous popwould be only 7 Torr; interaction with bicarbonate raises this to 27 Torr, thus creating a large enough pressure head for the flow of oxygen from the blood to the tissues (59). In human hemoglobin, the same relative rise in pop from 19 to 40 Torr can be brought about only by the combined effects of C o pand DPG. It is surprising that the crocodilian hemoglobins’ simple and direct reciprocating action between oxygen and one of the end products of oxidative metabolism has not been adopted by other vertebrates. These results show that an entirely new function can evolve in a protein by no more than three amino acid substitutions (Val NAlP --* Ser, His NA2P+ Pro, and Lys HClP+ Glu), requiring only four nucleotide base changes; just two more amino acid substitutions (His H21 + Ala and Ser H 14a + Ala) or three base changes are needed for inhibition of the old functions (oxygen-linked phosphate, carbamino C 0 2 , and C1binding). Most of the other 100 substitutions that distinguish crocodilian from human hemoglobins are conservative and would have little, if any, effect on the oxygen equilibrium. There are some significant substitutions in the heme pockets and at the subunit contacts, but none of these is unique nor could they play any role in the allosteric control by bicarbonate ions. All the residues which are essential for the formation of the characteristic T and R structures are either conserved or have been replaced by ones that can serve the same purposes equally well. A recent paper suggests that the hemoglobins of another class of diving reptiles, the sea turtles, have allosteric properties resembling those of the crocodilians (1 14). Until recently, investigators have searched in vain for an intermediate between the primitive dimeric hemoglobins of the lamprey and hagfish and the a2Pp tetramers of the higher vertebrates. Such an intermediate has now been found in the reptile Sphenodon punctatus that inhabits islands off the shore of New Zealand and is a survivor from the Triassic period of the ancient order of “beakhead” reptiles. It closely resembles the rhynchocephalians which lived 200 million years ago. The reptile’s blood and its stripped hemolysate both exhibit hyperbolic oxygen equilibrium curves, the former with a $50 = 19 Torr and the latter with a P50 = 1.8 Torr. The alkaline Bohr effect was small: Ap501ApH= -0.16. The erythrocytes contain 1.2 mol ATP per mol hemoglobin tetramer, and addition of ATP restored P50 of the hydrolysate to 14 Torr (129). These properties pose a riddle. Organic phosphates have never been

-

232

M. F. PERUTZ

observed to influence the oxygen equilibrium curve of any hemoglobin other than 4 2 tetramers that undergo an allosteric transition, yet Sphenodon hemoglobin looks as though it responded to ATP without such a transition, because its reaction with oxygen is devoid of cooperativity. The hemoglobin is tetrameric, but it is not clear yet whether the tetramer is made up of pairs of a- and P-like subunits. Conceivably it consists of two pairs of P-like chains, each pair capable of combining with ATP as in Fig. 6. The ATP would constrain the chains in their tertiary deoxy structure. Further study of the hemoglobin may provide important clues to the evolution of cooperative oxygen binding. V. BIRDS

Bird hemoglobins are functionally similar to mammalian ones but they use P5-inositol (inositol pentaphosphate) in place of DPG as an allosteric effector. The residues at the phosphate binding site are the same as in mammals except that position H2 16 is occupied by Arg rather than His. Two further replacements in the internal cavity are H13P Ala 4 Arg and H17P Asn --* His (62-65). These two basic residues may be too far removed from the P5-inositol binding site to bond to its phosphates directly, but they would help to neutralize its negative charges and thus contribute indirectly to the binding energy. One remarkable instance of adaptation has been reported in the hemoglobins of two related species of goose. The greylag goose (Anrer anser) lives in the plains, and the oxygen affinity of its blood is normal; the bar-headed goose (Anser i n d i m ) migrates across the Himalayas at an altitude of 9000 m, and the oxygen affinity of its blood is abnormally high (66). In laboratory experiments, the bar-headed goose proved more resistant to hypoxic stress that did the Canada goose or the Pekin duck, neither of which flies at such high altitudes (67). It was then found that the oxygen affinities of the purified hemoglobins of the two species differed only slightly: at pH 7.2 in 0.1 M NaCl at 37"C, the hemoglobin of the greylag goose has a p 5 0 of 5.8 Torr and that of the bar-headed goose one of 4.5 Torr. Their affinities for P5-inositol were identical. How was this to be reconciled with the large differences between the oxygen affinities of their bloods? Rollema and Bauer demonstrated that, by lowering the oxygen affinities of both hemoglobins 10-fold, P5-inosito1 also amplified the differences between them 10-fold, raisingp50to 50 and 37 Torr, respectively (68). The amino acid sequences of the two hemoglobins differ by only four substitutions, of which only one is unique among the bird sequences determined so far (65,66). This is H2a Pro (greylag goose) 4Ala (bar-

SPECIES ADAPTATION IN A PROTEIN MOLECULE

233

headed goose). Pro H2a is invariant in nearly all other species, probably because its y-C forms an important van der Waals contact at the alpl boundary: in mammals with Met D6a and in birds with Leu D6P. Studies of abnormal human hemoglobins have shown that the loss of any interatomic contact at subunit boundaries is liable to loosen the constraints of the T structure, thereby lowering L and raising KT (4). The loss of this van der Waals contact in the bar-headed goose hemoglobin may therefore explain its high oxygen affinity. VI. MAMMALS H. F. Bunn has shown that mammalian hemoglobins can be divided broadly into two groups: the great majority have an intrinsically high oxygen affinity, which is lowered in the red cell by DPG, while those of ruminants and cats (Cervidae, Bovidae, Felidae) and of one primate, the lemur, have an intrinsically low oxygen affinity that is little, or not at all, lowered by DPG (69-73). Typically, hemoglobins with high intrinsic oxygen affinity have P 5 0 values of between 4 and 6 Torr, and those with low affinity have $50 values of between 10 and 20 Torr (measured in stripped Hb solutions in 0.05 M bis-tris, 0.1 M NaCl, pH 6.5-7.5, at 2025°C). The low affinity is due to lower values of K T and higher values of L, while KR remains the same in the two groups of species (74). Nearly all the amino acid replacements that distinguish the high and low oxygen affinity hemoglobins are conservative and external, but consistent differences are found in position NA2P which is occupied by hydrophilic residues (His, Gln, or Asn) in hemoglobins with high affinity, and by large hydrophobic residues (Leu, Met, or Phe) in those with low affinity. What role could residue NASP play in lowering the oxygen affinity? In the human T structure, DPG forms salt bridges with Val N A l and His NA2; these salt bridges pull the two helices A toward the center of the molecule so that they become locked more tightly to neighboring segments of the polypeptide chain (Fig. 3). In bovine, sheep, and deer hemoglobins, residue N A l is missing and residue NA2 is Met, which almost abolishes their affinity for organic phosphates. In these hemoglobins, the hydrophobic side chain of the N-terminal Met probably plays the same part in locking helix A tightly in place and thereby stabilizing the T structure as organic phosphates do in human and other hemoglobins with intrinsically high oxygen affinity. In cat hemoglobin, this part would be played by Phe and in lemur hemoglobin by Leu NA2 (74). Several mammals living at high altitude have bloods with high oxygen affinities. They include the llama, the golden-mantled ground squirrel

234

M. F. PERUTZ

(Citellus lateralis), and the yellow-bellied marmot (Marnota flaviventris). For example, at 35 Torr C 0 2 and 37”C, p 5 0 was about 20 Torr for the marmot, 30 Torr for the squirrel, and 50 Torr for a lowland rodent, the rat (75-77). Amino acid sequences have been reported only for llama hemoglobin (78, 79), which has a low oxygen affinity in the absence of phosphate, overcompensated by a low affinity for DPG due to the replacement NA2P His + Asn (80). The side chain of asparagine is shorter than that of His or Gln. Figure 3 shows that this shortening places the amino groups of the asparagines further from the phosphates of DPG than -NHof the histidine or the amino group of a glutamine. In consequence, the affinity for DPG is weakened. Asn NA2P occurs also in the hemoglobin of the elephant (81), whose blood shows a high oxygen affinity for an animal that lives at low and medium, rather than at high, altitudes ( p 5 0 = 22.4 Torr at 39°C) (76); perhaps this helped the elephants who carried the supplies for Hannibal’s army across the Alps in 218 B.C. In fact, there exists a somewhat weak correlation between body size and oxygen affinity of the blood, of which the elephant with p 5 0 = 23 Torr and the mouse with P 5 0 = 46 Torr are two extreme examples (70). A similar adaptive mutation is found in human fetal hemoglobin. This has a lower oxygen affinity than the adult in the absence of phosphate, but it also has a lower affinity for DPG,due mainly to the replacement His H21(143)p + Ser. As a result, the oxygen affinity of fetal blood exceeds that of the adult, which helps the transfer of oxygen from the maternal to the fetal circulation across the placenta (5). On going to high altitudes, humans show an increased DPG concentration in their erythrocytes, which lowers the oxygen affinity. This used to be regarded as an adaptive response, lower oxygen affinity allowing a larger fraction of the oxygen carried to be discharged in the tissues. Theoretical and experimental studies have shown that this may be true at moderate altitudes (<3000 m), but at higher altitudes a raised oxygen affinity is more advantageous, for the following reason. Suppose we mark the arterial and the venous oxygen pressure on the oxygen equilibrium curve and connect the two points by a line. Efficiency of oxygen transport will be greatest when the line lies on the steepest slope of the sigmoid curve. At moderate altitudes a shift to lower oxygen affinity was found to move the line to a steeper part of the curve, while at higher altitudes the reverse was true (82, 83). The mole (Talpa europaea) lives in its burrows under hypoxic conditions, to which it is adapted by having a blood with a high oxygen affinity, a higher concentration of hemoglobin per unit volume, and a low body temperature (84). The high oxygen affinity of its blood is reported to be due to a low affinity of its hemoglobin for DPG,but how the amino acid sequence effects this remains unclear (85).

SPECIES ADAPTATION IN A PROTEIN MOLECULE

235

VII. SPECIES ADAPTATION IN ENZYMES I know of only two instances of adaptation of protein molecules other than hemoglobin that have been interpreted in atomic detail. They are the enzymes ferredoxin and glyceraldehyde-phosphate dehydrogenase in thermophile bacteria. In these heat-resistant enzymes an entire range of adaptive mechanisms has been encountered, from 10 amino acid substitutions in the 55 residue chain of ferredoxin to create six more external salt bridges (86), to a multitude of replacements in glyceraldehyde-phosphate dehydrogenase of Bacillus stearothermophilus to produce extra salt bridges and van der Waals contacts (87). These enzymes have had many billions more generations to evolve than those of vertebrates, but, even so, their tertiary and quaternary structures have remained closely similar to those of mesophiles. New enzymatic activities have evolved in the laboratory in bacteria subjected to selective pressures. For instance, growth of Klebsiella aerogenes on xylitol in place of its “natural” substrate ribitol led to a point mutation in the gene for ribitol dehydrogenase that increased the enzyme’s affinity for xylitol (88). Clarke has studied the evolution of an aliphatic amidase in Pseudomonas aeruginosa (89). The wild-type strain grows well on acetamide, but not on longer chain amides. Under selective pressure a single point mutation that substituted a serine for a phenylalanine in the enzyme’s polypeptide chain produced a strain that grew on butyramide. A further single mutation produced a strain that grew on phenylacetamide and had lost the ability to grow on acetamide. Hall has studied the evolution of a minor P-galactosidase in a strain of Escherichia coli from which the gene for the major P-galactosidase had been deleted. He found spontaneous point mutations that altered the specificity of the enzyme, so that it hydrolyzed lactose and other sugars on which the original strain would not grow (90). The structures of the three enzymes used in these studies are unknown, so that we cannot unravel the stereochemical mechanisms underlying the evolution of new catalytic activities. But the fact that they did evolve by one or two point mutations is consistent with the conclusion from my hemoglobin studies that such evolution can be accomplished by very few amino acid substitutions. . There exists a large literature on enzyme polymorphism and species adaptation (91), but none of it can as yet be interpreted in stereochemical terms. The best-studied species is Drosophilu melanogaster, in which the frequency of the two dominant alleles of alcohol dehydrogenase varies with latitude in several continents; one of these alleles has a consistently lower Michaelis constant for alcohols than the other (92). The two enzymes have been found to differ by the single substitution of a lysine for

236

M. F. PERUTZ

a threonine at position 192 of the polypeptide chain (93). Unfortunately, the stereochemical meaning of the substitution remains unknown, because there is no sequence homology between the Drosophila enzyme and the horse heart enzyme whose structure has been determined (94). But it is noteworthy that, here again, enzyme adaptation is the result of a single substitution. Little is known as yet about the chemistry of other polymorphous enzymes. VIII. DISCUSSION We may now reconsider the questions raised in Section I. Does adaptation involve changes in the tertiary and quaternary structure of hemoglobin? I have no firm answer, because the only structures that have been accurately determined are those of human and horse hemoglobin. Their tertiary and quaternary structures are the same within experimental error, despite 42 amino acid substitutions between them. The number of amino acid substitutions between human and fish hemoglobins is about 140, or one-half of the total number of amino acids; those between human and caiman number 110. Few of these lie at heme contacts or at contacts between the subunits. The near constancy of the amino acids at these contacts and the presence of certain other invariant residues in key positions indicate that the tertiary and quaternary structures have remained almost constant throughout vertebrate evolution. I am confident that X-ray analysis will bear this out. Additions and deletions have occurred in nonhelical segments: for instance in the Port Jackson shark, which is regarded as a living fossil, the a chain contains 144 residues instead of the 142 in other fish, and the fl chain contains only 141 instead of the 147 in other fish. The additional 2 residues in the a chain are at the amino end and can be tucked away in the internal cavity without change of structure. The deletions in the fl chain occur in the CD segment which is nonhelical and external; they require only small local adjustments in tertiary structure (Fig. 1). My proposal that the tertiary and quaternary structure of the hemoglobins has been conserved throughout evolution from fish to mammals has been met with disbelief among biologists and others, but it comes as no surprise to protein crystallographers who have found that homologous proteins in distant species have closely similar structures. A few examples are glyceraldehyde phosphate dehydrogenase from B. stearothemnophilus (1 18) and lobster (1 19); phosphoglycerate kinase from yeast (120) and horse (121); bovine and fungal catalase (122, 123); human and hen lyoszyme (124, 125); iron superoxide dismutase from Pseudamonas ovaC (126) and E. coli (127), and the cytochromes c (128).

SPECIES ADAPTATION IN A PROTEIN MOLECULE

237

The tertiary structures of vertebrate myoglobins, of the monomeric lamprey hemoglobin, of invertebrate hemoglobins, and of leguminous plant hemoglobins all resemble those of the a and P chains of vertebrate hemoglobin to the extent necessary to preserve the vitally important geometry of the heme pocket, but they vary in detail. The angles between helical segments differ by up to 30" and the points of contact between them by up to 7 A. Many different combinations of side chains are found to produce helix interfaces that are comparably well packed, as if the tertiary structure had been conserved by a patchwork of improvisations (95). In fact, the only invariant residues common to all globins are the proximal His F8 and Phe CDl which wedges the heme into its pocket. What maintains the tertiary structure is an invariant set of sites that are always occupied by nonpolar residues (96). Has adaptation been brought about by the gradual accumulation of minor mutations, each producing a small shift in,chemical affinity, or by a few amino acid substitutions in key positions? In the instances analyzed so far, new chemical functions appear to have evolved by only a few amino acid substitutions in key positions. Consider the evolution from cartilaginous fish to bony fish. The hemoglobin of the Port Jackson shark has only a weak alkaline Bohr effect and has Ala in position F9. It seems that the Root effect in teleost fish has evolved primarily by the substitution of this Ala by Ser. In the crocodilians, affinity for bicarbonate evolved by no more than three amino acid substitutions, requiring only four nucleotide base changes; just two more amino acid substitutions, or three base changes, were needed for inhibition of the affinity for the usual heterotropic ligands (carbamino Cop, organic phosphate, and Cl-). Five amino acid substitutions probably suffice to inhibit in trout I hemoglobin all the heterotropic interactions exhibited by trout IV hemoglobin. Conversion from an ATP to a DPG binding site needs only one amino acid substitution or one base change; modulation of the affinity for DPG in the human fetus, the llama, and the elephant also a single base change. In amphibia that do not need a Root effect, the alkaline Bohr effect is weakened by a variety of single substitutions: Ser F9 + Ala in the tadpoles of Xenopus and R. cutesbeianu; Glu FGlP + Asn or Gly in adult frogs and in one of the tadpoles; His HC3 Phe in another tadpole. It looks as though once a function is no longer needed any mutation that inhibits the function becomes fixed (97). What selective advantage does possession of multiple hemoglobin genes provide? In fish that possess two types of hemoglobin, one responsive and the other unresponsive to heterotropic ligands, such polymorphism offers a clear advantage. One of the hemoglobins assures the

238

M. F. PERUTZ

supply of oxygen to the swim bladder and the eye, while the other assures the fish of a continued supply of oxygen when low pH inhibits the uptake of oxygen by the other hemoglobin (12). Primates and ruminants need a fetal hemoglobin with a higher oxygen affinity than the adult to ensure oxygen transfer across the placenta, but other species, among them the horse, manage with the same hemoglobin in both fetus and adult (98). Studies of dogs, rats, and rabbits suggest that, in species lacking fetal hemoglobins, the oxygen affinity of the fetal blood is raised relative to that of the adult by a lower concentration of 2,3-DPG. This is achieved by an increased activity of pyruvate kinase, which helps to metabolize 1,S-DPG, the precursor of 2,3-DPG (99-101). Human adults possess two a and two P globin genes. The two a chain genes code for identical amino acid sequences and are both efficiently expressed, so that humans have the advantage of being effectively tetraploid for a globin. This benefits individuals born with a deletion or malfunction of one of the a chain genes (e.g., a-thalassemia); like sickle cell heterozygotes, such individuals are healthy and also have a better chance of surviving malaria than individuals with wild-type hemoglobin genes. The two polymorphous adult P chains differ at 10 positions and are functionally indistinguishable, but one of them, known as the 6 chain, is made in only 1/40of the quantity of the other, so that possession of the 6 gene confers no significant advantage to either P-thalassemia or sickle cell anemia patients. It may be a “fossil” of some past stage of evolution. Many vertebrates have multiple a and P genes, some coding for functionally equivalent and others for functionally different proteins. It is not clear whether their possession offers any selective advantage or whether they have merely survived from selectively neutral gene duplications. H. F. Bunn has suggested to me that the presence of multiple hemoglobins might allow a higher total hemoglobin concentration to be maintained in the red cell than if there were only a single one. On the whole, the evidence supports Kimura’s view that “in many cases polymorphism has no visible phenotypic effect and no obvious correlation with environmental conditions” (102). How many of the amino acid substitutions between the hemoglobins of different species are functionally significant? The great majority of amino acid differences between hemoglobins of different species consist of conservative or semiconservative replacements of external polar and nonpolar residues and internal nonpolar ones. Comparison of normal and abnormal human hemoglobins, or of the hemoglobins of closely related species, show that such replacements have little, if any, influence on the functional properties of hemoglobin. For example, human and horse hemoglobins differ in sequence at 42 out of 287 positions, yet their

SPECIES ADAPTATION IN A PROTEIN MOLECULE

239

functional properties are the same. Only a minority of the 400 or so different amino acid replacements found in abnormal hemoglobins affect any of the hemoglobin functions. It could be argued that even an amino acid replacement that produces only a very small shift in the oxygen equilibrium curve may give an animal a selective advantage that would prove decisive over thousands of generations; against this could be held that homeostatic mechanisms allow organisms to compensate efficiently for quite large shifts in the curve. Studies of abnormal hemoglobins have shown mutations that merely change surface charges to be without significant effect on the respiratory functions of hemoglobins, but recent work has revealed that such mutations may influence the rate of assembly of the a and /3 subunits into a& tetramers by electrostatic effects arising from their different isoelectric points (aA= 8.0; PA= 6.5). At physiological pH free a subunits exist as monomers and dimers carrying a net positive charge, while free p chains exist as tetramers, dimers, and monomers carrying a net negative charge. The effect of these charges on the assembly is illustrated by the following example. The sickle cell mutation Glu A4(6)@--* Val reduces the negative charge on the p subunits thus reducing electrostatic repulsion in the p tetramer and electrostatic attraction between aAand ps subunits. In consequence both the dissociation of p tetramers and the association of pS and aA subunits is slower than the dissociation of tetramers and the association of PAand aAsubunits. That difference in rates appears to account for the ratio of hemoglobin A to S in sickle cell heterozygotes being about 3 : 2 rather than 1 : 1. The higher ratio of hemoglobin A contributes to the fitness of the heterozygotes. It seems important, therefore, that the net charges of the subunits in an oligmeric protein be maintained (133). With that proviso, the structural evidence suggests that most of the amino acid replacements between species are neutral or nearly so, that they are caused by random drift of selectively equivalent mutant genes, and that adaptive mechanisms generally operate by a few replacements in key positions. Here again, the evidence supports Kimura’s view that “adaptive mutations are much less frequent than selectively neutral or nearly neutral substitutions caused by random drift” (97). R.Lewontin has pointed out to me that a mutation adapting a species to a new environment is likely to have preceded occupation of that environment. For example, a mutation that raised the oxygen affinity of the llama’s blood would have occurred before llamas discovered that they were able to graze at altitudes barred to competing species. W. Bodmer suggested that once a large change in chemical affinities produced by one mutation had enabled a species to occupy a new envi-

240

M. F. PERUTZ

ronment, its effect might have been refined by later adaptive mutations, each contributing minor shifts, over a long period of time. The structure of glyceraldehyde-phosphate dehydrogenase from B. stearothemzophilus (87), referred to earlier, suggests that the heat stability of this enzyme may have evolved by such a process. This might be the molecular equivalent of punctuated equilibria in the evolution of species.

ACKNOWLEDGMENTS I appreciate the helpful suggestions made by Dr. Alexander Rich, Prof. Christian Bauer, Dr. Walter Bodmer, Drs. Joseph and Celia Bonaventura, Dr. Sydney Brenner, Prof. Maurizio Brunori, Dr. H. Frank Bunn, Prof. Patricia H. Clarke, Dr. Thomas E. Creighton, Dr. Richard E. Koehn, Dr. Arthur M. Lesk, Dr. Richard C. Lewontin, Dr. Robin N. Perutz, and Dr. Graham Shelton concerning this paper. T o all these colleagues I wish to express my gratitude.

REFERENCES R. C. Lewontin, Sci. Am. 239, 156 (1979). M. Goodman, Prog. Biophys. Mol. Biol. 37, 105 (1981). J. L. King and T. H. Jukes, Science 164, 788 (1969). G. Ferrni and M. F. Perutz, “Haemoglobin and Myoglobin, Atlas of Biological Structures 2.” Oxford Univ. Press (Clarendon), London and New York, 1981. 5. H. F. Bunn, B. J. Forget, and H. M. Ranney, “Human Hemoglobins.” Saunders, Philadelphia, 1977. 6. Haernoglobin: Structure, Function and Synthesis. Br. Med. Bull. 32(3) (1976). 6a. M. F. Perutz, J. V. Kilmartin, K. Nishikura, J. H. Fogg, P. J. E. Butler, and H. S . Rollema, J. Mol. B i d . 158, 649 (1980). 7. K. Imai, “Allosteric effects in haemoglobin.” Cambridge Univ. Press, London and New York, 1982. 8. J. B. Steen, in “Fish Physiology” (W. S. Hoar, D. J. Randall, and J. R. Brett, Eds.), Vol. 4. Academic Press, New York, 1979. 9. J. B. Wittenberg and B. A. Wittenberg, Biol. Bull. 146, 116 (1974). 10. J. B. Wittenberg and R. L. Haedrich, Biol. Bull. 146, 137 (1974). 11. R. L. Ingerrnan, Respir. Physiol. 49, 1 (1982). 12. M. Brunori, Cum. Top. Cell. Regul. 9, 1 (1975). 13. M. Brunori, J. Bonaventura, C. Bonaventura, B. Giardina, F. Bossa, and E. Antonini, Mol. Cell. Eiochem. 1, 189 (1973). 14. R. W. Noble, L. J. Parkhurst, and G. H. Gibson,]. Biol. Chem. 245,6628 (1970). 15. The Alpha Helix Expedition to the Amazon for the Study of Fish Bloods and Hemoglobins. Comfi. Biochem. Physiol. 624 (1) (1979). 16. A. L. Tan, A. DeYoung, and R. W. Noble, J. Biol. Chem. 244, 2493 (1972). 17. K. Imai and T. Yonetani,]. Biol. Chem. 250, 2227 (1975). 18. J. C. Chien and K. H. Mayo, J. E d . Chem. 255, 9790 (1980). 19. I. C. Binotti, B. Giovenco, B. Giardina, E. Antonini, M. Brunori, and J. Wyman,Arch. Biochem. Biophys. 142, 274 (1971). 20. R. J. Gillen and A. Riggs,]. Eiol. Chem. 247,6039 (1972). 21. R. J. Morris, W. S. Neckameyer, and Q. H. Gibson,]. B i d . Chem. 256,4598 (1981). 1. 2. 3. 4.

SPECIES ADAPTATION IN A PROTEIN MOLECULE

24 1

22. R. J. Morris and Q. H. Gibson, J. Biol. Chem. 257, 4869 (1982). 23. M. F. Perutz and M. Brunori, Nature (London) 299,421 (1982). 23a. G. Mazur, G. Braunitzer, and P. G . Wright, Hoppe-Seyler’s Z. Physiol. Chem. 363, 1077 (1982). 23b. C. E. Bogg and U. Thewalt, Biochem. Biophys. Res. Commun. 37, 623 (1969). 23c. R. E. Weber and G. Lykkeboe, J. Comp. Physiol. 128, 127 (1978). 24. B. Grujic-lnjac, G. Braunitzer, and A. Stangl, Hoppe-Seyler’s Z. Physiol. Chem. 361, 1629 (1980). 25. G. Braunitzer and K. Rodewald, Hoppe-Seyler’s Z. Physiol. Chem. 361, 587 (1980). 26. F. Bossa, D. Barra, M. Coletta, F. Martini, A. Liverzani, R. Petruzelli,J. Bonaventura, and M. Brunori, FEBS Lett. 64,76 (1976). 27. D. A. Powers and A. B. EdGundson, J. Biol. Chem. 47,6686 (1972). 28. R. J. Gillen and A. Riggs, Arch. Biochem. Biophys. 183,678 (1977). 29. R. E. Isaacks, H. D. Kim, G . R. Bartlett, and G. R. Harkness, Life Sci. 20,987 (1977). 30. G. R. Gillen and A. Riggs, Comp. Biochem. Physiol. B 38, 585 (1971). 31. A. Riggs, in “Fish Physiology” (W. S. Hoar and D. J. Randall, Eds.), Vol 2, p. 209. Academic Press, New York, 1970. 32. K. Johansen and R. W. Weber, in “Perspectives in Biology” (P. S. Davis, Ed.), Vol 1, p. 219. Pergamon, Oxford, 1976. 33. A. R. Nash, W. K. Fisher, and E. 0. P. Thompson, A w t . J . Biol. Sci. 29, 73 (1976). 34. C. Phelps, M.Farmer, H. J. Fyhn, U. E. H. Fyhn, R. L. Garlick, R. W. Noble, and D. A. Powers, Cmp. Biochem. Physiol. A 62, 139 (1979). 35. F. G. Carey and J. M. Teal, C m p . Biochem. Physiol. 28, 199 (1969). 36. F. G. Carey, J. M. Teal, J. W. Kanwisher, K. D. Lawson, and J. S. Beckett, Am. Zool. 11, 137 (1971). 37. Q. H. Gibson and F. G. Carey, Adv. Exp. Med. Biol. 148,49 (1982). 38. M. Davies and D. K. Thomas, J. Phys. Chem. 60, 763 (1956). 39. W. Kauzmann, Adv. Protein Chnn. 14, 1 (1959). 40. J. Bello, J. Theor. Biol. 68, 139 (1977). 41. J. T. Rund, Nature (London) 173,848 (1954). 42. J. V. Kilmartin and J. F. Wootton, Nature (London) 228, 766 (1970). 43. M. F. Perutz and L. F. Ten Eyck, Cold Spring Harbor Symp. Quant. Biol. 36, 295 (1971). 44. J. V. Kilmartin, J. H. Fogg, and M. F. Perutz, Biochemistry 19, 3189 (1980). 45. L. J. Parkhurst, D. J. Goss, and M. F. Perutz, Biochemistry 22, 5401 (1983). 46. J. G. Williams, R. M. Kay, and R. K. Patient, Nwleic Acids Res. 8,4247 (1980). 47. S. J. Aggarwal and A. Riggs,J. Biol. Chem. 244, 2372 (1969). 48. M. Brunori, E. Antonini, J. Wyman, L. Tentori, J. Vivaldi, and S. Carta, Comp. Biochem. Physiol. 24, 519 (1968). 49. D. H. Atha, A. Riggs, J. Bonaventura, and C. Bonaventura, J.Biol. Chem. 254,3393 (1979). 50. D. K. R. M. Banville, R. Harris, and J. G. WilliamsJ. Biol. Chem. 258, 7924 (1983). 51. J. Bonaventura, C. Bonaventura, B. Sullivan, and S. Bourne, Nature (London) 265, 474 (1977). 52. S . J. Condo, B. Giardina, M. Lunadel, A. Ferracin, and M. Brunori, Eur. J. Biochem. 120, 323 (1981). 53. V. H. Hutchinson, H. B. Haines, and G. Engbretson, Respir. Physiol. 27, 115 (1976). 54. H. G h n , Respir. Physiol. 1, 1-12 (1966). 55. C. Lenfant and K. Johansen, Resfiir. Physiol. 2, 247 (1967).

242

M. F. PERUTZ

G. Shelton and R. G . Boutilier,J. Exp. Eiol. 100,245 (1982). K. Rodewald and G. Braunitzer, Hoppe-Seyler’s Z. Physwl. Chem., in press. C. Bauer and W. Jelkmann, Nature (London) 209,825 (1977). W. Jelkmann and C. Bauer, Comp. Eiochem. Physiol. A 65, 331 (1980). C. Bauer, M. Forster, G. Gros, A. Mosca, M. Perella, H. S. Rollema. and D. Voge1,J. Eiol. Chem. 256, 8429 (1981). 61. M. F. Perutz, C. Bauer, G . Gros, F. Leclercq, C. Vandecasserie, A. G. Schnek, G. Braunitzer, A. E. Friday, and K. A. Joysey, Nature (London) 291,682 (1981). 62. H. Takei, Y. Ota, K. Wu, T. Kihara, and G. Matsuda, J. Eiochem. 77, 1345 56. 57. 58. 59. 60.

(1975). 63. G. Braunitzer and J. Godovac, Hoppe-Seyler’s Z. Physiol. Chem. 363, 229 (1982). 64. W. Oberthiir, W. Voelter, and G. Braunitzer, Hoppe-Seyler‘s Z. Physiol. Chem. 361,969 (1980). 65. W. OberthUr, G. Braunitzer, and S. Kalas, Hoppe-Seyler’s 2. Physwl. Chem. 362, 1101 (1981). 66. D. Petschow, I. Wiirdinger, R. Baumann, J. Duhm, G . Braunitzer, and C. Bauer,J. Appl. PhysM1.: Respir., Environ. Exercise Physiol. 42, 139 (1977). 67. C. P. Black and S. M. Tenney, Respir. Physiol. 39, 217 (1980). 68. H. S. Rollema and C. Bauer,]. Eiol. Chem. 254, 12038 (1979). 69. H. F. Bunn, Science 172, 1040 (1971). 70. A. F. Scott, H. F. Bunn, and H. A. Brush,]. Exp. Zool. 201, 269 (1977). 71. C. Bonaventura, B. Sullivan, and J. Bonaventura, J. Eiol. Chem. 249, 3768 (1974). 72. M. N. Hamilton and S.J. Edelstein,J. Eiol. Chem. 249, 1323 (1974). 73. F. Taketa. Ann. N . Y . Acud. Sci. 241, 524 (1974). 74. M. F. Perutz and K. Imai,J. Mol. Eiol. 136, 183 (1980). 75. F. G. Hall, D. B. Dill, and E. S. C. Barren,]. Cell. Comp. Physiol. 8, 301 (1936). 76. H.Bartels, P. Hilpert, K. Barbey, K. Betke, K. Riegel, E. M. Lang, and J. Metcalfe, Am. J. Physiol. 205, 33 1 (1963). 77. R. W. Bullard, C. Broumand, and F. R. Meyer,]. Appl. Physiol. 21, 994 (1966). 78. G. Braunitzer, B. Schrank, and A. Stangl, Hoppe-Seyler’s Z. Physiol. Chem. 358, 409 (1977). 79. G. Braunitzer, B. Schrank, A. Stangl, and C. Bauer, Hoppe-Seyler’s Z. Physwl. Chem. 358,921 (1977). 80. C. Bauer, H. S. Rollema, H. W. Till, and G. Braunitzer, J . Comp. Physiol. 136, 67 (1980). 81. G . Braunitzer, W. Jelkmann, A. Stangl, B. Schrank, and C. Krombach, Hoppe-Seyler’s 2.Physiol. Chem. 363, 683 (1982). 82. Z. Turek, F. Kreuzer, and L. J. C. Hoofd, Pjuegers Arch. 342, 185 (1973). 83. Z. Turek, F. Kreuzer, and B. E. M. Rignalda, Pfluegers Arch. 376, 7 (1978). 84. T. A. Quilliam, J. A. Clarke, and A. J. Salsbury, Comp. Ewchem. Physiol. A 40, 89 (1971). 85. T. Kleinschmidt, W. Jelkmann, and G . Braunitzer, Hoppe-Seyler’s Z. Physiol. Chem. 362, 1263 (1981). 86. M. F. Perutz and H. Raidt, Nature (London) 255,256 (1975). 87. J. E. Walker, A. F. Carne, M. J. Runswick,J. Bridgen, and J. I. Harris, Eur.1. E i o c h . 108,549 (1980). 88. B. S . Harley, I. Altosaar, J. M. Dothie, and M. S . Neuberger, in “Structure-Function Relationships of Proteins” (R. Markham and R. W. Horne, Eds.), p. 191. NorthHolland Publ., Amsterdam, 1976. 89. P. H. Clarke, Proc. R. Soc. London, Ser. E . 207, 385 (1980).

SPECIES ADAPTATION IN A PROTEIN MOLECULE

243

90. B. H. Hall, Biochishy 20,4042 (1981). 91. R. K. Koehn, A. J. Zera, and J. G. Hall, in “Evolution of Genes and Proteins” (M. Nei and R. K. Koehn, Eds.). Sinauer, Sunderland, Massachusetts, 1983. 92. J. F. McDonald, S . M. Anderson, and M. Santos, Genetics 95, 1013 (1980). 93. D.R. Thatcher, Biochem.J. 187, 875 (1981). 94. C.-I. Branden, H. Eklund, B. NordstrBm, T. Bowie, G . Soderlund, E. Zeppezauer, I. Ohlsson, and A. Akeson, Proc. Natl. Acad. Sci. U.S.A. 70, 2439 (1973). 95. A. Lesk and C. Chothia,]. Mol. Biol. 136, 225 (1980). 96. M. F. Perutz, J. C. Kendrew, and H. C. Watson,]. Mol. Biol. 13,669 (1965). 97. M.Kimura, in “Evolution of Genes and Proteins” (M. Nei and R. K. Koehn, Eds.). Sinauer, Sunderland, Massachusetts, 1983. 98. A. Stockell. M. F. Perutz, H. Muirhead, and S. G. Glauser,]. Mol. Biol. 3,112(1961). 99. P. A. Mueggler and J. A. Black, Am. 1.Physiol. 242, H500 (1982). 100. W.Jelkmann and C. Bauer, Pjliiegers Arch. 389,61 (1980). 101. W.Jelkmann and C. Bauer, Pjliiegers Arch. 375, 189 (1978). 102. M. Kimura, Sn‘. Am. 239,94 (1979). 103. A. Arnone, Nature (London) 237, 146 (1972). 104.J. V. Kilmartin and L. Rossi-Bernardi, Nature (London) 242, 1243 (1969). 105. M. 0.Dayhoff, “Atlas of Protein Sequence and Structure.” National Biomedical Research Foundation, Washington D.C., 1972. 106. F. Bossa, D. Barra, R. Petruzelli, F. Martini, and M. Brunori, Biochim. Biophys. Acta 556, 298 (1978). 107. J. P. Chauvet and R. Acher, Biochist*y 11,916 (1972). 108. L. Tentori, G. Vivaldi, S.Carta, S. Velani, and R. Zito, Biochirn. Biophys. Acta 133, 177 (1967). 109. T.0.Baldwin and A. Riggs,]. Biol. Chem.249,6110 (1974). 110. T. Maruyama, K. W. K. Watt, and A. Riggs,]. Biol. Chem. 455, 3285 (1980). 1 1 1. K. W.K. Watt, T. Maruyama, and A. Riggs,]. Biol. Chem. 255, 3294 (1980). 112. W. K. Fisher, A. R. Nash, and E. 0. P. Thompson,]. Biol. Chem. 30, 387 (1977). 113. F. Lerclecq, A. G. Schnek, G. Braunitzer, A. Stangl, and B. Schrank, Hee-Sqrler’s 2. Physiol. Chem. 362, 1151 (1981). 114. R. E. Isacks, D. R. Harkness, and J. R. White, Hemoglobin 6,549 (1982). 115. A. M. Gronenborn, G. M. Clore, M. Brunori, B. Giardina, G. Falcioni, and M. F. Perutz, unpublished.

116. G. M. Clore and A. M.Gronenborn,]. M a p . Reson. 53,423 (1983). 117. P. F. Como, D. Kennett, T. Wilkinson, and H. Kronenberg, Hemoglobin 7,413 (1983). 118. G. Biesecker, J. I. Harris, J. C. Thierry, J. E. Walker, and A. J. Wonacott, Nature (London) 266,328 (1977). 119. M. Buehner, G. C. Ford, D. Moras, K. W. Olsen, and M. G. Rossman,]. Mol. Biol. 90, 25 (1974). 120. T. N. Bryant, J. C. Watson, and P. L. Wendell, Nature (London) 247 (1974). 121. R. D. Banks, C. C. F. Blake, P. R. Evans, R. Haser, D. W. Rice, G. W. Hardy, M. Merrett, and A. W. Phillips, Nature (London) 279, 773 (1979). 122. M. R. Murthy, T. J. Reid, 111, A. Sicignana, N. Tanakas, and M. G. Rossmann,J. Mol. Biol. 154,465 (1981). 123. B. K. Vainshtein, W. R. Melik-Adamyan, V. V. Barynin, A. A. Vagin, and A. I. Grebenko, Nature (London) 293,411 (1981). 124. P. J. Artymiuk and C. C. F. Blake,]. Mol. Biol. 152, 737 (1981). 125. T. Imoto, L. N. Johnson, A. C. T. North, D. C. Phillips, and J. A. Rupley, in “The Enzymes” (P. Boyer, ed.), Vol 7,p. 665. Academic Press, New York, 1972.

244

M. F. PERUTZ

126. D. Ringe, G . A. Petsko, F. Yamakura, K. Suzuki, and D. Ohmori, Proc. Natl. Acad. Sci. U.S.A. 80, 3879 (1983). 127. W. C . Stallings, T. B. Powers, K. A. Pattridge, J. A. Fee, and M. L. Ludwig, Proc. Natl. Acad. Sci. U.S.A. 80, 3884 (1983). 128. R. E. Dickerson, Sci. Am., 226, April, p. 58. (1972). 129. R. M. G . Wells, V. Tetens, and T. Brittain, Nature (London) 306, 500 (1983). 130. E. Barra, R. Petruzzelli, F. Bossa, and M. Brunori, Biochim. Biophys. Acta 742, 72 (1983). 131. R. Petruzelli, E. Barra, F. Bossa, M. Coletta, and M. Brunori, Biochim. Biophys. Acta, in press. 132. R. M. Kay, R. Harris, R. K. Patient, and J. G. Williams, Nucl. Acids Res. 11, 1537 (1983). 133. H. F. Bunn and M. J. McDonald, Nature (London) 306,498 (1983);

PROTEINS AT WORK: “STOP-ACTION” PICTURES AT

.

By PIERRE DOUZOU’ and GREGORY A PETSKOt

.

*Instltut do Blologlr Physlco.Chlmlque. U 128 INSERM. Parls. France and tDepartment of Chemistry. Masaachusetts Institute of Technology. Cambridge. Massachusetts

I . Introduction . . . . . . . . . . . . . . . . I1. Trapping of Enzyme-Substrate Intermediates in Solution . . . A. Enzyme-Substrate Intermediates (Stop-Action Intermediates) . B. Validity and Exploitation of Data in Mixed Solvents . . . . C . The Problem of the Determination of Conformational Changes Accompanying Enzyme Catalysis . . . . . . . . . I11. X-Ray Studies at Subzero Temperatures: Physical-Chemical Basis of Cryoprotection of Proteins in Solution and in the Crystalline State and Related Problems . . . . . . . . . . . . A. Cryoprotection of Crystalline Proteins . . . . . . . . B. Dielectric Constant: Dielectric Shock and Solutions Isodielectric with Water . . . . . . . . . . . . C. Cosolvent and Temperature Effects on Protein Interactions . . D. Control of Protonic Activity . . . . . . . . . . . E. Effects of Cosolvent and Temperature on Donnan and Electrostatic Parameters of Protein Crystals . . . . . . F Possible Supercooling of Protein Crystal; . . . . . . . G Strategies for Trapping Crystalline Enzyme-Substrate Complexes IV. Devices. . . . . . . . . . . . . . . . . . A. Low-Temperature Equipment for X-Ray Diffraction . . . . B. The Flow Cell . . . . . . . . . . . . . . . C. Equipment for Low-Temperature Spectroscopy . . . . . D Solving and Refining Low-Temperature Protein Structures . V Results . . . . . . . . . . . . . . . . . . A. Radiation Damage and Crystal Lifetime . . . . . . . B. Resolution, Disorder, and Related Problems . . . . . . C Productive Intermediates . . . . . . . . . . . . D Unstable Oxidation States . . . . . . . . . . . E. Analysis of Protein Flexibility . . . . . . . . . . VI Prospects and Problems . . . . . . . . . . . . . A . Mapping Enzymatic Reaction Pathways . . . . . . . . B. Large Conformational Changes . . . . . . . . . . C. Application to Other High-Resolution Techniques . . . . D. Time-Resolved Pictures of Functioning Proteins . . . . . References . . . . . . . . . . . . . . . . . I

. .

.

.

.

. .

.

245 ADVANCES IN PROTEIN CHEMISTRY. V d. 36

. . . . . .

246 247 247 267

. .

275

. . . .

280 280

. .

283 292 295

. .

. .

. .

. . . .

. . . .

. .

. . . . . .

. .

. . . .

. .

. . . .

. .

. .

. .

. . . . . .

301 317 320 321 321 324 326 327 328 328 329 329 342 345 353 353 355 356 357 358

Copyright 0 1984 by Academic Press. Inc. All rights of reproduction in any form reserved. ISBN 0-12-0342967

246

PIERRE DOUZOU AND GREGORY A. PETSKO

I. INTRODUCTION Although conformational changes are essential features of proteins, the conformational basis of protein activity is not yet understood at the molecular and atomic levels. It is generally assumed that the mechanism of enzyme-catalyzed reactions would be defined if all the intermediates and transition states between the initial and final stages, as well as the rate constants, could be characterized. But in spite of constant progress in such characterization, most enzymatic mechanisms are not understood in terms of physical organic chemistry and enzyme activity is still regarded as a “miracle” as compared to classical catalysis. X-Ray studies could reveal the mechanism of enzyme action at the atomic level if they could provide information about the structure of proteins during the course of their reactions. Up to now, crystallographic studies have provided information about the static structure of uncomplexed enzymes, and, more generally, about the structures of proteins before and after they change conformation. On the other hand, since the method gives information about the time- and space-averaged structure, with the time usually measured in hours and the space enclosing millions of molecules, X-ray data often give “fuzzy” pictures and it is difficult to determine subtle changes in protein conformation as well as the structure of transient intermediates. While new data acquisition techniques under development may significantly reduce the time and increase the precision of recording diffraction patterns, it is obvious that X-ray diffraction techniques will be restricted to the study of conformations and intermediates which are stable for periods that exceed the normal half-life of these transient species. It is therefore necessary to increase the lifetime of such species so that their three-dimensional structures may be determined in the same manner as a native enzyme. The present article deals first with the temporal resolution of enzyme reactions normally occurring in a few seconds or minutes but which are sufficiently reduced in rate by lowering the temperature to permit enzyme-substrate intermediates to be stabilized in fluid mixed solvents (Douzou, 1977a). Such investigation in solution is an absolute prerequisite for studies in the crystalline state to determine optimal conditions for reactions in mixed solvents, to ascertain the average lifetime of any stabilized intermediate, and last, but not least, to check that the actual reaction pathway is identical to that occurring under normal conditions of medium and temperature. Since there is always an unavoidable reversible effect of the organic cosolvent used as antifreeze on the K, and/ or V,, values of the reaction under investigation, analysis of such effects

PROTEINS AT WORK

247

is necessary to decide whether cosolvents induce artifacts. We will show that solvents appear to act like the “physiological”modifiers of enzyme activity, a result demonstrating that the low-temperature procedure should not be surrounded by an “unnecessary mystique” and by a “condemnable caution,” in the words of R. P. Cox (1978). All these results are encouraging for investigators planning to use Xray diffraction in mixed solvents at subzero temperatures and the rest of the present article will be devoted to a discussion of methods and preliminary results in this field. The methodology for cryoprotection of protein crystals, its physical-chemical basis, and the specific problems raised by the crystalline state, as well as the devices used to collect data at subzero temperatures, will be described. Limitations and perspectives of the procedure will be discussed critically. First attempts to determine the structure of “productive” enzyme-substrate intermediates through “stopaction” pictures will be described, as well as investigations showing that X-ray diffraction at selected normal and subzero temperatures can reveal protein structural dynamics. 11. TRAPPING OF ENZYME-SUBSTRATE INTERMEDIATES IN SOLUTION

A, Enzyme-Substrate Intermediates (Stop-ActionIntermediates)

Work in solution is an absolute prerequisite for further studies of enzyme-substrate intermediates in the crystalline state. According to the Arrhenius relationship, K = A exp(-EA/RT), which relates the rate constant k to the temperature, reactions normally occurring in the second to minute ranges might be sufficiently decreased in rate at subzero temperatures to permit intermediates to be stabilized, and occasionally purified by column chromatography if reactions are carried out in fluid solvent mixtures. Therefore, the first problem is to find a suitable cryoprotective solvent for the protein in question. A great variety of aqueous-organic mixtures can be used. Most of them are listed in Table I with their respective freezing point and the temperature at which their bulk dielectric constant ( D ) equals that of pure water. These mixtures have physicochemical properties differing from those of an aqueous solution at normal temperature, but some of these differences can be compensated for. For example, the dielectric constant varies upon addition of cosolvent and cooling of the mixture in such a way that cooled mixed solvents can be prepared which keep D at is original value in water and are isodielectric with water at any selected temperature (Travers and Douzou; 1970, 1974).

248

PIERRE DOUZOU AND GREGORY A. PETSKO

TABLE I Cyoprotective Binary Solvents Solvendwater ratio (v1v) Solvent”

10190 20180

EGOH1water Freezing point (“C) D = 80 atb

-4

MeOH/water Freezing point (“C) D = 80 atb

-

PGOH1water Freezing point (“C) D = 80 atb

30170 40160

50150

60140

70/30

80120

90/10

-10 5

-17 0

-26 -10

-44 -20

-69 -30

-100

-83 -55

-50

-40

-

-

-

-40 -20

-49 -30

-67 -45

-85 -70

-100 -85

-

-35

-8

-17

-38 -20

SC‘ -35

SC -50

-70

SC -90

SC

SC -25

SC -40

-55

SC -75

SC

SC -10

-38 -30

-

-83 -60

-100

10

10

0

-10

Glycine1water Freezing point (“C) D = 80 at’

-3 10

-8 5

-14.5 -5

-29

DMSO1water Freezing point (“C) D = 80 atb

-3 20

-12

-19

-41

DMFIwater Freezing point (“C) D = 80 atb

-25 15

-7

15

10

15 -13 5

-10

10 -25

-10

-20

SC 10 -40 -20

5 -62 -40

SC

SC

-80

-

SC

-

-7

-90 -

a EGOH, Ethylene glycol; MeOH, methanol; PGOH, polyethylene glycol; DMSO, dimethyl sulfoxide; DMF, dimethylformamide. equals that of water (80). Temperature (“C) at which bulk dielectric constant (0) SC, Supercooling.

Perhaps the most important parameter involved in aqueous-organic mixtures is their effective protonic activity (denoted by pH* or paH). This parameter has been measured for most commonly used buffers in all selected mixtures down to their freezing point (Hui Bon Hoa and DOUZOU, 1975; Douzou et al., 1976). Values of pH* depend on solvent and temperature in a way that varies for different buffers, but with the data available a medium of known pH* under any desired condition may be prepared. An example of the effect of solvent and temperature is provided by Tris-HC1 buffer; a solution of this at pH 8.0 in water at 20°C will be pH* 10.5 in 50% (v/v) ethanediol at -40°C (Douzou e l al., 1976). On the other hand, “neutral” buffers’such as phosphate undergo

249

PROTEINS AT WORK

a change in pH* of about 1 unit on addition of an equal volume of ethanediol (pH 7.0 in water, pH* 8.0) at 20°C and a small decrease (0.1 unit) at -40°C. The pK values of groups on the protein may also be affected by temperature and solvent, so adjusting the pH* to a value equal to that at normal temperatures may not be sufficient. Therefore, once the internal environment (pH*, ionic strength, substrate concentrarion) of any enzyme system is set up in unusual conditions of medium and temperature, it is necessary to record the pH-activity profile to check whether its shape is unchanged and to determine any shift of the profile due to solvent and temperature effects on the dissociation constant of ionizing groups which are involved in catalysis. Due to shifts of the pH-activity profiles, pH values often must be adjusted to obtain the optimal activity of the enzyme under investigation (Maurel and Douzou, 1975). When the above requirements are fulfilled, there is always a residual effect of the cosolvent on enzyme activity. In most cases, such an effect is small compared to the effect of lowering temperature. It must be checked that the effect is instantaneous upon addition of the solvent, independent of time, and fully reversible by infinite dilution or dialysis. If these conditions are not met, one should suspect denaturation. Finally, it is necessary to record reaction kinetics as a function of temperature to determine whether the enzyme system follows the Arrhenius relationship, indicating that activation energies and, presumably, reaction mechanisms remain unchanged in the temperature range investigated. Once these investigations have been completed, low-temperature spectroscopy can be used to dissect the reaction mechanism by trapping normally unstable intermediates. Temporal Resolution of Enzyme-Catalyzed Reactions

An increasing number of enzyme-catalyzed reactions normally occurring in seconds to minutes have been successfully investigated in mixed solvents at subzero temperatures and have been resolved in time, step by step. Some examples are presented in the following sections. a. Horseradish Peroxidme. Over 10 years ago (Douzou et al., 1970) the first successful attempt to obtain the temporal resolution of a multistep enzyme-catalyzed reaction was reported. This was a peroxidatic reaction, i.e., the oxidation of substrates by hydrogen peroxide catalyzed by horseradish peroxidase (HRP): 2AH

+ H202-

Enzyme HRP

2&did

+ 2H20

250

PIERRE DOUZOU AND GREGORY A. PETSKO

It had been shown, chiefly by B. Chance (1943) using fast kinetic techniques, that this reaction can be formulated as follows:

-

E(HRP oxidized iron * Fe3+)

+ AH Compound I1 + AHCompound I

+ H202

Compound I

(1)

+ Aaxdird E(HRP) + 2H2P + Aa.idked

(2)

Compound I1

(3)

The enzyme horseradish peroxidase is a hemoprotein and the region of the Soret band exhibits large differences between the position and extinction coefficients of the uncombined and combined forms. Both forms were first studied by spectrophotometry, but the E-S complexes were so labile that they could not be examined extensively by any other spectroscopic method. Using rapid-scanning spectrophotometry and rapid mixing, Chance was able to distinguish the spectra of compound I and I1 and determine the various rate constants of the multistep reaction with rather poor precision. Investigating the same reaction in fluid mixed solvents in a range of subzero temperatures, Douzou and associates were able to obtain the “temporal” resolution of the reaction, step by step, by means of coolingheating cycles (Douzou et al., 1970). As an example, reactions carried out in a mixture of ethylene glycol and buffer, pH 6.5 (1 : 1 v/v), at -40°C gave the following result:

+ H202Compound I + AH2HRP

Compound I1

+ AH*-

+ H20 compound I1 + AH FepJ+ + H 2 0 + AH

compound I

(4) (5)

(6)

where AH2 is ascorbic acid. At -4O”C, compound I, obtained practically “pure” (98%),was stable indefinitely. With an increase in temperature of about 20°C there was a progressive formation of compound 11, which was then stabilized at -40°C and finally transformed into Fe:+ H20 by heating to 0°C. Thus cooling-warming cycles, suitably induced, allowed temporal resolution of the reaction, step by step. The absorption, optical rotary dispersion, and electron spin resonance (ESR) spectra of pure compounds I and I1 were then recorded and found to be similar to those obtained under fast-reaction conditions (Douzou et al., 1970; Douzou and Leterrier, 1970). Additional interesting observations were made possible by the low-temperature technique; for instance, instead of making the time scale of reactions feasible experimentally by increasing the substrate concentrations as in fast techniques, reactions at low temperatures could be performed with stoichiometric concentrations of enzyme. Such con-

-

PROTEINS AT WORK

25 1

centrations, often found in nature, are necessary for the performance of sophisticated experiments that permit direct observation of the sequential formation and decomposition of intermediates at the steady-state rates. Moreover, the formation of compound I with stoichiometric concentrations of peroxidase and hydrogen peroxide confirms that one molecule of peroxide combines with one atom of iron from the hemoprotein and that, in this respect, substrate binding is similar to the binding of inhibitors such as hydrogen sulfite and hydrogen cyanide. Also, hydrogen peroxide indeed can be used at low temperatures in place of the slow-reacting analogous alkyperoxides often employed at normal temperature. Compound I can be thermally activated and converted into compound 11; thus, true substrates can be used like abortive compounds undergoing only part of the catalytic cycle, which then can be continued as desired by warming. Compounds I and I1 are quite stable at low temperature and therefore can serve as pure “reactants” to study the mechanisms of peroxide oxidase processes (Douzou, 1971a,b). When compound I1 reacted with indole 3-acetate, this compound was immediately regenerated without the appearance of any other intermediary compound. Moreover, indole 3acetate in large excess induced the conversion of compound I11 into compound 11. A study of reaction mechanisms of indole 3-acetate degradation by various peroxidases was recently carried out by Ricard and Job (1974) using low-temperature spectroscopic techniques. They obtained new data that made it possible to propose electronic mechanisms of reactions less speculative than those dependent upon data obtained under normal conditions of temperature. b. Hydroylating Multienzyme Systems. Work has been carried out in Douzou’s laboratory to investigate well-known multienzyme systems performing drug hydroxylation in liver endoplasmic reticulum (Peterson et al., 1973)or camphor hydroxylation in a Pseudomonasputida strain (Gunsalus et al., 1974). These investigations provide an example of the potentialities, the innocuousness, and the analytical advantages of the fluid solvents-low temperature technique. The mechanism of hydroxylation by the bacterial monooxygenase has been resolved to show four main steps in the sequence of cytochrome P-450 reactions (Fig. 1).

1. Binding of the substrate camphor (RH) by the free ferric cytochrome (Fe9+):

252

PIERRE DOUZOU AND GREGORY A. PETSKO

- "1 I

I

Fe3+\

RH

?

i

Fes I RH

-_--I

H FIG. 1. The cycle of reactions catalyzed by the bacterial cytochrome P-450.

2. Reduction of the E-S complex by an iron-sulfur protein (putidaredoxin, Pd): Fe3+ RH

+ e-

-

Fez+ RH

Step (2) 3. Binding of molecular oxygen to give a ternary oxyferrous compound: Fez+ RH

+ 02 =(02-FeZ+). RH

Step (3) 4.Uptake of a second electron initiating the reactions leading to the hydroxylated substrate (ROH) with restoration of the free enzyme (Fe3+): (02-Fe2+)

RH

+ e-

-

Fe3+ + ROH

+ HzO

Step (4)

Cytochrome P-450has been characterized in four stable states [Fe3+, Fe3+* RH, Fe2+* RH, (02-Fe2') * RH (metastable)] during its oxygenase reaction cycle. In the complete native system a flavoprotein and a redoxin (putidaredoxin) act as electron donors but also as effectors that complement the cytochrome. In the more complex microsomal system the sequence and intermediates are less well defined; the electron-transfer chain contains two flavoproteins and one cytochrome, whose interactions with cytochrome P-450are still the subject of great controversy.

253

PROTEINS AT WORK

Electron-transfer reactions, especially step (4), are too fast and the intermediates too fleeting to be properly characterized and discriminated; they are, moreover, “contaminated” by recycling of the free ferric cytochrome (Fe3+)and by spontaneous nonhydroxylating decay of the oxyferrous compound. The reaction cycle lacks any consistent knowledge of the nature and reactivity of unStable intermediates occurring beyond the metastable oxyferrous compound (Op-Fe) RH. Single steps ( l ) , (2), (3), and (4) of the reaction cycle are quite temperature sensitive; thus work at subzero temperatures allows their temporal resolution, provided this procedure does not affect the enzymes nor alter the reaction sequence (Lange et al., 1977). Cytochrome P-450 is solubilized in polyol-water mixtures, the mixture generally used being ethylene glycol and buffered water (1 : 1 v/v), which freezes at about -50°C. Cytochrome P-450 is stable at -30°C in this still fluid mixture and thus may be kept in its native form without repetitive freezingthawing cycles. This is even more important for the iron-sulfur protein, putidaredoxin, which is very sensitive to thawing and moreover decays quite rapidly when stored in fluid solution at 4°C. (These results turn out to be a general feature of the now widely used polyol-water mixtures for storage of enzymes in the cold; the protective effect is better if freezing is avoided, the solution remaining homogeneous. Reversible perturbations of protein structure can occur and the activity be affected or suppressed in the range of normal temperature. Therefore, tests must be carried out to check enzyme activity and specificity in the mixed solvents. We have found repeatedly that the presence of high concentrations of poly01s and cooling per se do not alter the solubility and/or native structure of single-chain proteins, but freezing and uncontrolled changes in paH, ionic strength, and bulk dielectric constant often lead to precipitation and/or denaturation.) The various redox states of cytochrome P-450 (Fe3+,Fe3+ RH, Fe2+ RH)as well as the metastable oxyferrous compound [(02-Fe2+) * RH] are obtained in ethylene glycol-water mixture; their absorption spectra and formation rates are similar to those recorded in pure aqueous media. These identical spectra demonstrate that the intermediates obtained in the mixed solvent at normal and subzero temperatures are similar to those found in the productive enzyme pathway under normal conditions. This is an essential observation since the low-temperature procedure permits one to stabilize and accumulate intermediates and offers the opportunity of obtaining structural information about such intermediates-a result unattainable by classical fast-kinetic techniques. So far, observations under these conditions fit well with results obtained under normal conditions of medium and temperature. The de-

-

-

254

PIERRE DOUZOU AND GREGORY A. PETSKO

tailed pathway of the cycle is unaltered in suitable buffered polyol-water mixtures. Intrinsic temperature effects on single steps are then obtained by cooling-heating cycles. The most interesting effects of cooling are the following. 1. The stabilization and isolition of the metastable oxyferrous compound (02-Fe2+) RH, which can then be used as a pure reactant. 2. The possibility of starting reactions from the preformed dienzyme complex Pd--FeS+ RH to obtain the complex:

-

-

Pd--Fez+

/OZ

\

RH

- - --

and to “uncouple” its decomposition from subsequent recycling: Fe9+

Fez+ RH

FeJ+ RH

(Op-Fez+)

RH

and from spontaneous nonhydroxylating decay: (O2-Fe2+)

RH

Fe9+ RH

+ 0 2 + e-

Thus, the reaction can be viewed through only one turnover, a situation not possible in usual conditions of temperature. c. Serine Proteases. Different members of this class of enzymes involved in protein breakdown have different specificities, but all possess a similar catalytic mechanism involving a reactive serine residue at the active site, which is generally agreed to be involved in the formation of a covalent acyl-enzyme intermediate (Kraut, 1977). Fink and collaborators have made a series of studies of chymotrypsin at temperatures down to -90°C in media containing 65% (v/v) dimethyl sulfoxide. At low temperatures the acyl enzyme could be formed, but the rate of deacylation was so slow that the intermediate could be separated from excess substrate by gel filtration on Sephadex LH-20 at -40°C (Fink, 1973a,b). This provided direct evidence for the participation of a covalent acyl enzyme in the catalytic pathway with a specific substrate. Even lower temperatures have been used to study possible intermediate stages in the formation of the acyl enzyme. A tetrahedral intermediate (with a covalent bond between the substrate carbonyl carbon atom and the oxygen atom of the active site serine) (Fig. 2) had been suggested by analogy with nonenzymatic reactions. With rapid reaction techniques, spectrophotometric evidence has been obtained for an additional intermediate before the acyl enzyme in the case of chromophoric substrates. By using first the protein fluorescence emission (Fink and Wildi, 1974)

255

PROTEINS AT WORK

VI

I

C

I

C

H

h/y,

/ \cF=C

II

0

0 '

,/H'

/"'

V

m

FIG. 2. The generally accepted mechanism for the hydrolysis of peptide substrates by the serine proteases. The precise locations of the protons are still moot; their positions here are taken from Steiu and Shullman (1982). I, Michaelis complex; I1 and V, tetrahedral intermediates; 111 and IV, acyl-enzyme; VI, product complex.

256

PIERRE DOUZOU AND GREGORY A. PETSKO

and then the absorbance changes during the reaction with N-acetyl-Lnitroanilide (Fink, 1976a,b), it was possible at subzero temperatures to detect a number of intermediates before the formation of the acyl enzyme under conditions where deacylation was negligible. From the spectrophotometric studies, Fink (1976a,b) proposes four reactions before acylation. Formation of the Michaelis enzyme-substrate complex is very fast even at -9O"C, too fast to study even with the double-syringe device used to fill the spectrophotometer cuvette with a rapidly mixed solution. Two conformational changes follow, with kinetics that can be easily measured at -5O"C, and then a third pH-dependent reaction appears to involve the histidine residue adjacent to the active-site serine. (This histidine residue, together with the adjacent "buried" carboxylate group of an aspartate residue, is involved in a hydrogen-bonded system, which can be regarded as providing a protonbinding site during the formation of the acyl enzyme.) A fourth reaction was interpreted as formation of the acyl enzyme; deacylation was too slow to be measured under these conditions. It is interesting that no step with the properties corresponding to those expected for the formation of the tetrahedral intermediate could be detected. The hydrolysis of peptide bonds catalyzed by the serine proteases has been the reaction most extensively studied by low-temperature trapping experiments. The reasons for this preference are the ease of availability of substrates and purified enzymes, the stability of the proteins to extremes of pH, temperature, and organic solvent, and the existence of a well-characterized covalent acyl-enzyme intermediate. Both amides and esters are substrates for the serine proteases, and a number of chromophoric substrates have been synthesized to simplify assay by spectrophotometric techniques. N-Carbobenzoxy-L-alanine-p-nitrophenyl ester is a specific substrate for elastase in which the rate-limiting step is deacylation, that is, hydrolysis of the acyl-enzyme intermediate. In 70% methanol over a reasonable temperature range the energy of activation of the turnover reaction, that is, deacylation, is 15.4 kcal mol-'. In the pH* 6-7 region in this cryoprotective solvent, the turnover reaction can be made negligibly slow at temperatures of -50°C or below. Under such conditions p-nitrophenol is released concurrent to acyl enzyme formation in a 1 : 1 stoichiometry with active enzyme in the presence of excess substrate. In other words, even at low temperatures, the acylation rate is much faster than deacylation and the acyl enzyme will accumulate on the enzyme. The rate of acyl-enzyme formation can be monitored by following the rate of p-nitrophenol release, and thus the concentration of trapped acyl enzyme may be determined. This calculation has been carried out and

PROTEINS AT WORK

257

indicates, within experimental error, complete conversion of enzyme to acyl enzyme. At pH* 5.7 at -7O"C, only 0.4% of the acyl enzyme breaks down in 24 hours (at -5O"C, 11% is lost per day). In the pH* 6-7 region at -5O"C, the formation of the maximal concentration of the intermediate is complete in 3 hours. Consequently, by use of cryoenzymological methods, the covalent intermediate in serine protease catalysis of ester hydrolysis can be formed quickly in high yield and stabilized for days (Fink and Ahmed, 1976). Significant rate enhancements, mostly in the kcat term, result from increasing the number of amino acids in synthetic elastase substrates (about fivefold enhancement per additional residue up to a pentapeptide). Consequently, acetyl-Ala-Pro-Ala-p-nitroanilide is a specific amide substrate for elastase. Richards and co-workers (Hunkapiller et al., 1976) carried out stopped-flow studies of the reaction of this substrate with elastase in aqueous solution at room temperature, and observed an initial faster reaction prior to the turnover reaction. For anilide substrates the rate-limiting step in the overall enzymatic reaction is acylation, so the faster phase was ascribed to the formation of a pre-acyl tetrahedral intermediate. Using methanol-water cryosolvents at pH* around 9, Fink and Meehan (1979) were able to detect, stabilize, and characterize pre-acyl intermediates in the reaction of elastase with a number of p-nitroanilide peptide substrates at subzero temperatures without the necessity for fast-reaction techniques. For the tripeptide substrate used by Richards, the maximum absorption of the substrate occurs at 318 nm, that of the p-nitroaniline product at 381 nm, and that of the pre-acyl complex at 359 nm (Fink and Meehan, 1979).Turnover, corresponding to the collapse of the putative tetrahedral intermediate to the acyl enzyme (which deacylates extremely rapidly), can be made negligible at temperatures below -50°C. The maximum concentration of this intermediate is obtained at pH* 9.4, and corresponds to less than 100% of the enzyme concentration, even under conditions of excess substrate. This effect is probably due to concurrent existence of a preceding EStype intermediate. Lower temperatures favor the tetrahedral adduct at the expense of the preceding intermediate whereas lower pH* favors the intermediate preceding the tetrahedral adduct. With succinyl-AlaAla-Ala-p-nitroaniline and acetyl-Pro-Ala-p-nitroaniline, 100% of the tetrahedral intermediate may be accumulated at pH* 9-10 at temperatures below -70 and -50"C, respectively. The intermediate preceding the putative tetrahedral adduct is characterized by a spectrum with essentially the same wavelength of maximum absorbance as the substrate, but with a higher extinction coefficient. This intermediate is assumed to be a noncovalent Michaelis complex in which substrate binding

258

PIERRE DOUZOU AND GREGORY A. PETSKO

has induced changes in the position and solvation of active site groups as well as in the substrate itself. Thus, the use of subzero temperatures in cryosolvents has allowed a series of intermediates in the elastase reaction to be identified, characterized, and stabilized. Temporal resolution of the significant steps in this reaction was difficult to achieve by any other method. Analogous results with other serine proteases suggest that these results are general for this class of enzymes. The extreme stability of the acyl enzyme of serine proteinase at low temperature can be exploited, as shown below, in X-ray studies. d. Lysoryme. The three-dimensional structure of crystalline hen egg white lysozyme has been determined to atomic detail by Phillips (1967), and investigations were carried out to find the structure of a reactive complex between lysozyme and its polysaccharide substrate in the hope that the active residues in the enzyme could be recognized and their function understood. Because of the velocity of the catalyzed hydrolysis of polysaccharide molecules and the necessity of a rather long exposure of crystals to X-rays, it was not possible to use an oligosaccharide filling the active site in any X-ray diffraction study. Simple amino sugar molecules, including the monosaccharide N-acetylglucosamine (NAG) and a trisaccharide comprising three NAG units (tri-NAG), which act as competitive inhibitors, were used as substitutes for substrates since they bind to the enzyme in the same sites that the true substrate does. X-Ray crystallographic studies of lysozyme-inhibitor complexes have been rewarding (Phillips, 1967; Banyard, 1973),but it would be desirable, particularly in view of proposals for the catalytic mechanism which involve substrate strain, to obtain the structure of a true enzyme-substrate complex. One possible method of stabilizing active enzyme-substrate complexes would be to reduce the temperature. Tests allowing precise determination of the conditions to protect the soluble protein, and of the temperature at which the reaction was slow enough for X-ray data collection, were sought. To ascertain the best conditions for the determination of the structure of a productive lysozyme-substrate complex, the hydrolysis of bacterial cell walls and oligosaccharides was investigated both in high-salt solutions and in mixed solvents. Since use of increasing amounts of cosolvents as “antifreeze” could perturb the conformation and thus the activity of lysozyme, a number of experiments were carried out to try to determine conditions for investigating lysozyme reactions and lysozyme-substrate intermediates in cooled mixed solvents as a preliminary to similar investigation by X-ray diffraction on crystals. Work began with an estimate of the solubility of

PROTEINS AT WORK

259

the enzyme and of several oligosaccharides in various mixtures as a function of temperature. The assay of lysozyme activity was carried out by recording the cleavage of the polysaccharides of bacterial cell walls of Micrococcus lysodeikticus. The important part of these cell walls, as far as lysozyme is concerned, is made up of long polysaccharide chains containing the glucoselike amino sugars N-acetylglucosamine and N-acetylmuramic acid (NAM), occurring alternately in the chains and connected by bridges that include an oxygen atom (glycosidic linkage) between carbon atoms 1 and 4 of consecutive sugar rings. Lysozyme has been shown to break the linkages of carbon-1 in NAM to carbon4 in NAG but not the other linkages. This process brings about the dissolution of bacterial cell walls, which can be measured turbidimetrically. e. Kinetics of Lysozyme-Catalyzed LysL in Mixed Solvents at Room and Subzero Temperatures. Different solvents can be used to investigate this reaction (Douzou et al., 1975). With a methanol-water mixture of volume ratio 40 :60, the freezing point is about -35°C. When adding the 40% methanol to a solution of buffer, the protonic activity increases, and the difference ApH* = pH* - p H ~ = ~ 0.5 0 and 0.8, respectively, in the case of acetate and phosphate buffers. At -5O"C, the increase of pH* does not exceed 0.22 for acetate and phosphate buffers in 50% methanol; the extrapolated enthalpy of ionization of these buffers in 40% methanol can be estimated as AH"i = 0.7 and 0.9 kcal mo1-l respectively, for acetate and phosphate buffers. So the increase of pH* is very small at low temperatures and can be neglected. Another solvent is 2-methyl-2,4pentanediol-water with volume ratios of 50 :50 and 70 :30 and freezing points around -30 and -90°C. A solution of 2-methyl-2,4-pentanediol (MPD) has an absorption band at 275 nm and is not transparent below 260 nm; near its freezing point this solvent is very viscous, which decreases its usefulness. The last two solvents used are solutions of high salt concentration: 7 M ammonium nitrate (NH4NOs)and 5 M sodium chloride (NaCl). Their freezing points are about -20°C. For spectroscopic studies, 5 M NaCl solution is preferred because of its transparency in the UV region. The ammonium nitrate solution absorbs strongly in the UV. Mixed solutions are prepared by progressive addition of the organic solvent to aqueous buffered solution of lysozyme cooled to 0°C to prevent any possible denaturation. The enzymatic activity of lysozyme can be determined turbidimetrically by following the decrease in optical density at 71 1 or 500 nm of a mixture of M . lysodeikticus (0.1 mg ml-l) and lysozyme (8 pg ml-I); at low temperature the induction phase can take a few minutes. A plot of the logarithm of optical density against time gives a straight line, indicating that under the experimental conditions the reaction is first

260

PIERRE DOUZOU AND GREGORY A. PETSKO

order. In such conditions it has been found that lysozyme and oligosaccharides were soluble up to 0.2 mg ml-' in 50% by volume of methanol and of MPD down to -15°C. In 70% MPD, each component is still soluble separately, but their concentration must be lowered to 0.01 mg ml-' when mixed in order to be tested down to -20°C. The overall rate constant of hydrolysis of bacterial cell walls in aque= 17.8 X sec-' ous phosphate buffer (2 X lo-* M, pH 6.0)is for A = 500 nm at 20°C.In the methanol-water mixture (40:60),when 40% methanol is added to aqueous solutions buffered by phosphate at pH 6.0,the proton activity increases to pH* 6.8,and the rate constant k ~ ~ 0 ~ , 4 0= 9 67.8 X lo-' sec-' at A = 500 nm at 20°C.The maximum of the pH* profile (PH*,,,~~) is shifted toward alkaline values (7.5)and the sec-I. corresponding k M v l e 0 H becomes 1 1.5 x In the MPD-water (50 :50)mixture, in pH conditions similar to those above, k ~ ~ ~ =, 4.6 ~ 0 X% low3sec-' at A = 500 nm and 20°C.The pH*rate profile indicates a maximum value of the rate constant (kmaX)of 7.9 X sec-I when the aqueous buffered solution is at pH 7.0.In the MPD-water (70: 30)mixture, there is no detectable hydrolysis at 20"C, mainly due to the action of MPD on Micrococcus. There is a noticeable inhibition of the hydrolysis, insensitive to pH changes, in salty solutions.

1. For 7 M ammonium nitrate: the rate constants observed at pH 4.7, 6.0,and 8.0 are, respectively, 1.13 X 1.9 X and 1.3 X sec-I at 20°C. 2. For 5 M sodium chloride, the same results are obtained and, e.g., k = 4.2 X sec-' at pH 4.7.(pH values were estimated using a pH meter with glass and calomel reference electrodes. Values recorded are independent of the salt concentration between 1 and 7 M, but depend on the nature of the buffer.) In the above media the hydrolysis of bacterial cell walls has been studied between +20" and -30°C. The rate constants lie on a straight line in the Arrhenius plot [loglo k =f(l/T)]. Activation energies are 16.9 kcal mol-' in 40% methanol, and 20 kcal mol-l in 50% MPD. The velocity in salty solutions is much more sensitive to salt action (inhibition) than to temperature. In the presence of 7 M NH4N03, the rate factor ksalt/kHpo 9 X and the calculated activation energy is -5.3 kcal mol-'. The inhibitory effect of the solvent and of high salt concentration was found to be entirely reversible by dilution; lysozyme can be stored in these solutions for hours at 0°C (and for days below 0°C) and then diluted without any loss of activity. On the other hand, lysozyme-cata-

-

PROTEINS A T WORK

26 1

lyzed lysis of bacterial cell walls is practically quenched in mixed solvents at about -25°C. By warming up, the rate constant and activation energy values observed during cooling at selected temperatures are restored. Thus, investigation of the kinetics of lysozyme-catalyzedlysis of bacterial cell walls has shown that the presence of an organic solvent decreased the overall reaction rate reversibly with the ratios of the maximum rate constant of lysis in water and in mixed solvent given by

It has been shown previously that there is no macroscopic conformational change of the lysozyme in the presence of 40% methanol, and that there is no change of the levorotation at 436 nm and just a slight increase in the extinction coefficient at 292 nm in the presence of 50% MPD. The reversible reduction of reaction rates in mixed solvents is small compared to that obtained in salt solutions, where k H 2 0 I R N H 4 N o 3 , 7 M = 110. The temperature effect on the rates of hydrolysis, recorded between +20 and -25"C, shows that the rate constant R is reduced by a factor of 200 in the presence of 40% methanol over this range of temperatures, and by a factor of 510 in the presence of 50% MPD (such reductions correspond, respectively, to activation energies of 16.7 and 20 kcal mol-I). The first numerical value is similar to that obtained in water for the hydrolysis of the penta-NAG by lysozyme (E = 17 kcal mol-I), and such a similarity suggests that the mechanism of breakdown of the complex E-S to form product is identical in both media. The reduction of the rate of hydrolysis due to lowering the temperature 45°C in the solution containing 7 M N H 4 N O 3 is k H 2 0 / k N H q N O 3 , 7 M = 4.78, with an activation energy of 5 kcal mol-I. This solvent is not suitable for low-temperature studies of the lysozyme reaction. Observations carried out on lysozyme in mixed solvents as a function of temperature demonstrate that lysozyme-catalyzed lysis can be performed under such abnormal conditions and that the reaction can be quenched at subzero temperatures and resumed by heating. The problem is now to check whether one can obtain an enzyme-substrate complex (in this case lysozyme-oligosaccharide) stabilized at low temperature. Such a complex can be detected by differential absorption spectroscopy. Difference spectra in the UV region (240-320 nm) were recorded. The reference cell contained a solution of lysozyme (1.39 x M) and the sample cell contained a solution of lysozyme at the same concentration plus substrate, here hexa-NAG (1.66 X M). The buffer used is acetate at pH 5 plus organic solvent. Also, difference spectra have been determined in the presence of the small-chain sac-

262

PIERRE DOUZOU AND GREGORY A. PETSKO

charides di-NAG (4.83x loc4M ) and tri-NAG (3.27X M). These substrate and inhibitor concentrations are sufficient to nearly saturate the enzyme. An equimolar solution of lysozyme and hexa-NAG, in acetate buffer at pH 4.7 (corresponding to the pH of maximum activity), temperature 20°C, recorded by difference absorption spectrophotometry against a . solution of lysozyme (see Fig. 3) shows 3 positive peaks (277,287,294 nm) and 2 negative peaks (280,291 nm). The largest peak (at 294 nm, with a shoulder at 298 nm) occurs during mixing of the enzyme and substrate. This confirms previous results obtained by Hayashi et al.

+ c

0

+

A h )

FIG.3. Difference spectra of the complex of lysozyme plus hexa-N-acetylglucosamine (hexa-NAG) vs native lysozyrne at 20°C in various solvents.

263

PROTEINS AT WORK

(1963) with enzyme-glycol-chitin complex. The spectrum is relatively time independent for almost 30 minutes. It can be seen in Fig. 3 that similar spectra are obtained in salt solutions at neutral pH, and that spectra are only slightly modified in mixed solvents, probably due to modifications in pK. Note the transformation of the shoulder at 298 nm into a peak in methanol (40%), MPD (50%), and a much more pronounced perturbation in 70% MPD. At -20°C in mixed solvents and at the same concentrations as above, similar difference spectra are obtained (Fig. 4), except that there is a real peak at 298 nm and that the magnitude of OD294 is slightly reduced.

250

r

do

?

- 18.5.C

+

4

\

. _.l

iao is0

3

1

FIG. 4. Difference spectra of the complex of lysozyme plus hexa-NAG vs native lysozyme at subzero temperature in several cryoprotective solvents.

264

PIERRE DOUZOU AND GREGORY A. PETSKO

Obviously, the spectra are more sharply defined than at 20°C. They are quite stable over several hours. Difference spectra have been recorded in 40% methanol between pH* 2.0 and 10.6 at -20°C and are shown in Fig. 5. There is a real peak at 298 nm between pH* 4.7 and 6.1, which could correspond to carboxyl group ionization. The amplitude AOD = OD294- OD291is pH* dependent and the profile of substrate binding in 40% methanol at -20°C shows maximum binding at about pH 4.7, which compares well to results obtained in water by Banerjee and Rupley (1973). Thus, difference spectra obtained on mixing lysozyme and hexa-NAG in mixed solvents at room as well as at subzero tempera-

A(nm)

FIG.5. pH-Dependence of the (lysozyme + hexa-NAG)-lysozyme difference spectrum in 40% methanol at -20°C.

PROTEINS AT WORK

265

tures indicate the formation of an enzyme-substrate complex and it appears that such a complex is indefinitely stabilized at -25°C. The shape of the difference spectra attributed to the lysozyme-hexaNAG complex at low temperature (-20°C) is pH* dependent, and should depend on the ionization of chromophores near the binding site. The maximum amplitude of these spectra corresponds to the known pH of the maximum activity of lysozyme in the hydrolysis of hexa-NAG. These preliminary studies clearly show the following. 1. It is possible to obtain a lysozyme-catalyzed lysis of substrates in mixed solvents. There is a partial and reversible inhibition showing that any organic solvent acts as a modifier of the enzyme specific activity. 2. Lowering the temperature decreases the reaction rates markedly, and the reaction is quenched at the level of the E-S complex between -20 and -25°C. 3. Conditions of pH corresponding to the maximum of catalytic activity can be chosen to obtain maximum affinity between E and S.

These results suggest that the crystallographic determination of the structure of a productive enzyme-substrate complex is feasible for lysozyme and oligosaccharide substrates. They also provide the information of pH, temperature, and solvent effects on activity which are necessary to choose the best conditions for crystal structure work. The system of choice for human lysozyme is mixed aqueous-organic solvents at -25”C, pH* 4.7. Data gathered on the dielectric constant, viscosity, and pH* behavior of mixed solvents (Douzou, 1974) enable these conditions to be achieved with precision. f. Ribonuclease A . Bovine pancreatic ribonuclease A (RNase A) catalyzes the hydrolysis of single-stranded RNA molecules, with preferential cleavage at phosphodiester linkages in which the 3’ residue is a pyrimidine. The mechanism of this enzyme is known in some detail, and may be described as a two-stage process (see Fig. 19, Section V,C,2).The first stage involves an intramolecular displacement by the 2’ -OH group of the ribose on the 5’ hydroxymethyl group of the adjacent nucleotide to give a 2’,3‘-cyclic phosphate intermediate. The second stage involves hydrolysis of this intermediate by enzyme-activated water to give a terminal 3’ monophosphate ester. Both displacements at phosphorous are known to procede by an in-line mechanism. Ribonuclease has been studied extensively by a variety of physical and chemical techniques, but it is well suited for a cryoenzymology study for several reasons. First, the enzyme is extremely stable even in high concentrations of organic solvent. Second, the reaction is slow overall, and thus susceptible to tempo-

266

PIERRE DOUZOU AND GREGORY A. PETSKO

ral resolution. Third, substrate, product, and cyclic phosphate intermediate are all stable compounds which are commercially available. Fink and associates (Fink et al., 1984) have undertaken a cryoenzymological study with the following objectives in mind: (1) to find a suitable cryosolvent system for RNase A, (2) to find conditions of pH* and low temperature in which the turnover reaction can be effectively quenched, and (3) to determine the conditions necessary to prepare any intermediate in the crystalline state so that its detailed structure can be determined using X-ray diffraction. Based on thermal denaturation curves the enzyme is in the native conformation in high concentrations of methanol and ethanol, provided the temperature is suitably low. So once again the indicated procedure is to couple a progressive addition of organic solvent to a progressive lowering of the temperature. The 360 MHz NMR spectra of the enzyme in these cryosolvents have been recorded and show that the folding and tertiary interaction of the protein in aqueous-organic mixtures are very similar to those found in aqueous solution. This conclusion was supported by the effects of methanol on the intrinsic fluorescence emission of RNase. Smooth monotonic changes in the quantum yield and emission maximum were observed when the methanol concentration was increased from 0 to 70% at 0°C. The CD spectrum of RNase A in 50% methanol, pH* 6 at -20°C has been found to be very similar to that in pure water at 20°C. The effects of methanol on the catalytic properties for the hydrolysis of mono- and dinucleotide substrates are also consistent with the absence of any adverse effects of the cosolvent. Significant methanolysis occurs in the presence of methanol. The pH* dependence of k,,lK, in cryosolvents agrees well with the profile obtained in aqueous solution. Linear Arrhenius plots were observed in methanol-water cosolvent mixtures over the temperature range - 10 to -30°C; the activation energy calculated for the hydrolysis of cytidine 2',3'-cyclic phosphate was 12.4 kcal mol-I for k,,lK,, with approximate values of 13 kcal mol-I fork,, and 4 kcal mol-l for K,. These values extrapolate well to values obtained in pure water at elevated temperatures, indicating that the essential catalytic mechanism of RNase A is not perturbed under cryoenzymological conditions. The turnover reaction of hydrolysis of 2',3'-CMP could be made negligibly slow at temperatures below -60°C at pH* 3-6 in 70% methanol, and below -35°C at pH* 2.1. The rate of the catalytic reaction using crystalline enzyme was found to be 50-fold slower than that of dissolved enzyme for cyclic phosphate hydrolysis, and 200-fold slower for dinucleotide hydrolysis (presumably the greater reduction for the larger substrate reflects increased diffusional hindrance by the small solvent chan-

PROTEINS AT WORK

267

nels in the crystal). Severe competitive product inhibition was observed in both aqueous and methanolic solvents. A log-linear relationship was found between K, and methanol concentration, due to a hydrophobic partitioning effect on substrate binding caused by the hydrophobic solvent. The results of Fink et al. (1983) demonstrate that in 70% methanol at sufficiently low temperatures, the binding of 2‘,3’-CMP to RNase A still occurs relatively rapidly, but its turnover to product can be made negligible. This determination of the experimental conditions necessary to accumulate and stabilize this noncovalent enzyme-bound intermediate allows crystallographic studies of its structure to be carried out. These studies are described in Section V,C,2. Several other successful applications of the low-temperature procedure to the thermal control and analysis of multistep enzyme reactions could be described. We prefer to cite appropriate papers (Douzou, 1974, 1977a,b; Fink, 1976a) and to discuss two important problems raised by the present procedure, namely the validity of data obtained in such bizarre media and the necessity of obtaining suitable data on the conformational changes in proteins during their reaction pathways. B . Validity and Exploitation of Data an Mixed Solvents As pointed out in the preceding section, there are always unavoidable effects on enzyme specific activity of cosolvents used as antifreezes. The parameters Km,appand/or V,, are reversibly and significantly modified, casting doubts in the mind of biochemists about the real value of results obtained under such nonphysiological conditions. A series of suitable tests permits checking whether detected intermediates in mixed solvents are involved in the actual catalytic pathway observed in normal conditions by fast-kinetic techniques, but in spite of encouraging results, significant changes in the values of substrate affinity, heat of formation, and activation energies raise the problem of possible changes in substrate partitioning between the active site and the bulk solvent as well as in protein conformation. While changes in K,”,? and V,, values might be responsible for changes in the rate-determining steps, such that some intermediates could become more fleeting and some others could accumulate, such modifications raise the problem of possible artifacts and could lead investigators, as stated by R. P. Cox in a recent review (1978), to a “condemnable caution.” Recent work (Lange et al., 1984) sheds light on this important problem and clearly demonstrates that cosolvents used as antifreeze obviously act as reversible modifiers similar to physiological modifiers, including ligands of regulatory significance. Part of this work has been carried out

268

PIERRE DOUZOU AND GREGORY A. PETSKO

with a single-chain enzyme, bacterial cytochrome P -450. This molecule, which is a heme protein, shows a high spin c)low spin equilibrium when in its ferric form. This can be shifted almost at will by various structurally unrelated compounds. Addition of increasing amounts of camphor, the specific substrate of bacterial cytochrome P-450,initiates the conversion low spin --* high spin and, next, a reverse conversion when its concentration is further increased well above saturation (Fig. 6). A possible explanation of such a bell-shaped response is that camphor acts as a substrate, interacting with the active site, and also as a ligand, binding at a subsite of the active site or at an allosteric site thus inducing a conformational change favorable to the low spin state of camphor-bound cytochrome P-450.Thus, the substrate induces conformational changes in the protein and is able to stabilize two kinds of new structures depending on its binding at the active site or at an allosteric site. Under conditions of saturating substrate concentrations, other compounds initiate similar shifts of the high spin c) low spin equilibrium; monovalent cations, especially K+, cause a significant shift. Cosolvents, such as primary alcohol and polyols, induce a similar shift, shown in Table 11.Thus, there appears to be a unifying similarity between conformational changes induced by physiological modifiers and by organic

1.o

c

0.8

0.6 In

x'

0.4

0.2

t 01

I

I

I

10

102

103

[Camphor] (IN)

FIG.6. Effect of the concentration of camphor and the ionic strength on the fraction of bacterial cytochrome P-450 in the high-spin state (XHS)in aqueous solution at pH 7 in 25 mM sodium cacodylate buffer at 18°C. a, No added KCI; 0, 50 mM KCI.

269

PROTEINS AT WORK

TABLE I1 Factors I@uenCing the Spin State of Fenic-Substrate Bound Cytochrome P-450 (Fes+,R H ) Factor

Spin state

Increase in T Ethylene glycol (to 50%) Methanol (to 20%) Increase in paH (from 5 to 7) Monocations (K+, Na+) (mM concentration) Polycations (polyarginine, polylysine) ( p M concentration range)

High High High High High High

Decrease in T Hydrophobic solvents (n-butanol, t-butanol) (to 2%) Substrate (mM concentration range) Decrease in paH Pd"

Low spin Low spin

spin spin spin spin spin spin

Low spin Low spin Low spin

solvents. All operate by stabilizing a new protein conformation which affects the high spin-low spin equilibrium at the active site. The simultaneous presence of modifiers and cosolvents determines changes in the equilibrium. Each modifier (or cosolvent) involves a specific enthalpy change (Table 111). It is obvious that one can create a mutual complementary interaction leading to different final conformations, and the different combinations of modifiers (including cosolvents) TABLE I11 The Modulation of AH of the K E = [HS]/[LS]Induced by Physicochemical Parameters Constant parameters

PH

LW (kJ/mol)

5 6

28.5 22.2 21.1 23.4

~

Aqueous solvent Na+ buffer concentration = 25 mM [KCI] = 100 mM [Camphor] = 1.5 mM

7 8

Change of Initial parameters

Change of parameters

AH (kJ/mol)

Mixed solvent =7 Cacodylate "a+] = 100 mM [KCI] = 0 [Camphor] = 0.7 mM

Mixed solvent to water

31.4+ 26.8

[KCI]: 0- 100 mM [Camphor]: 0.7 + 3 mM

3 1.4 + 23.8 31.4 +-39.7

P H

270

PIERRE DOUZOU AND GREGORY A. PETSKO

lead to the same equilibrium value, through different enthalpy changes. Thus, cosolvents inducing reversible perturbations of the protein conformation behave like other physiological modifiers and, in the present case, the high spin * low spin equilibrium shift normally induced by a cosolvent can be corrected by suitable concentrations of effectors. The above observations provide a clear demonstration that cosolvents in selected ranges of concentration create reversible perturbations of protein similar to those induced by other modifiers. The reversibility of the cosolvent effect is a prerequisite to cosolvent use and will depend on the concentration of cosolvent, which in turn will vary markedly with the type of solvent used. For instance, polyols can be used at concentrations up to 8 M while methanol at 3 M causes the appearance of a new absorption band (410 nm) and, after further increases in concentration, an irreversible conversion of cytochrome P-450 into P -420. Other aliphatic alcohols cause denaturation at much lower concentrations. A quite interesting observation is that long-chain aliphatic alcohols (t-butanol, butanol), i.e., highly hydrophobic cosolvents, induce a shift qualitatively identical to that produced by the protein effector putidaredoxin, the properties of which have been described in Section II,A. Once again, there appears to be a unifying similarity between the conformational changes induced by the physiological modifier at a subsite of the active site or at an allosteric site, and by some organic solvents. However, it can be seen in Fig. 7 that there is a large difference in specificity between the two species, and experiments are now under way to understand the mechanism of each species and gain information about putidaredoxin binding. Observations of cosolvent “mimicking” the effects of other molecules on protein conformation can be extended to many other regulatory systems. These systems have been characterized by their ability to be activated or inhibited by specific modulators without damage to catalytic activity. A striking example is provided by glycogen phosphorylase, an enzyme which is at the center of all regulatory mechanisms controlling glycogen degradation. The molecular basis of regulatory effects on phosphorylase activity has been interpreted in terms of conformational changes, the more active state (phosphorylase a, tetramer) being stabilized either by covalent phosphorylation or through AMP binding (Graves and Wang, 1972).While the molecular mechanism by which the enzyme is allosterically activated is not known, there is evidence favoring the hypothesis of a shift of an allosteric equilibrium from a T to an R state by 5’-AMP.Organic solvents produce a similar shift, resulting eventually in a bypass of the AMP requirement for activity (Dreyfus et al., 1978).This unexpected result is shown in Fig. 8.It can be seen that long-

27 1

PROTEINS AT WORK

1

; ethylene glycol

C

.-0c

\

1

2

3

% Organic solvent

t e

-e c 0

x'

-1

.-C .s c .-L

'\

>"

'\

[Pd"] addition

\

-2

-3

I

1

10

20

30

I

I

I

40

50

60

t

[Pd'l(CuM)

FIG. 7. The effect of various substances on the spin-state equilibrium of bacterial cytochrome P-450.The fraction of protein in the high-spin state is plotted (in arbitrary units) against the concentration (in p M ) of the natural effector putidaredoxin (Pd") or the concentration in percent (v/v) of the organic solvents ethylene glycol or n-butanol. It is apparent that butanol induces a shift in spin state which is similar to that induced by the protein effector.

chain aliphatic alcohols provide a high degree of activation in the absence of AMP at moderate concentrations (for instance, 35% of the reference level for 12% v/v or 1.3 M t-butanol). Kinetic parameters of the phosphorylase reaction associated with AMP and t-butanol activation are reported in Table IV. These parameters are affected in two different ways: (1) the maximal velocity is markedly increased; and (2)

272

PIERRE DOUZOU AND GREGORY A. PETSKO

n- butanol

I

I

I

I

1

I

I

5

10

15

20

25

30

Cosolvent concentration (% v/v) FIG.8. The activity of muscle phosphorylase b (as a logarithm percentage of the activity of the fully AMP-activated enzyme in water) as a function of concentration (% v/v) of added organic cosolvent in the absence of AMP. It can be seen that long-chain aliphatic alcohols provide a high degree of activation.

the glucose 1-phosphate requirement for half-maximal activity is greatly reduced. Moreover, it is obvious that the kinetic response with respect to the glucose 1-phosphate ligand is noncooperative, as with phosphorylase a. In fact, all the kinetic properties of phosphorylase b in the presence of both AMP and alcohol are strikingly similar to those of phosphorylase a, suggesting a similar conformation. This observation is confirmed by the fact that in the presence of AMP (1 mM) and t-butanol (10% vlv), phosphorylase 6 sediments as a tetramer as does phosphorylase a, whereas in the presence of any of these effectors alone it remains essentially dimeric. These different results confirm that selected concentrations of organic solvents act as some modulators do, and that combination of the actions of these modulators and cosolvents can produce changes in reaction

273

PROTEINS AT WORK

TABLE IV Kinetics Parameters of Phosfihorylase Reaction Variable K, or (mM) vm, ligand (%Hid (prnol Pi rnin-1 rng-1)

Solvent Phosphorylase b +5% t-butanol

Glc 1 P Glycogene

0.5

Phosphorylase b + 10%t-butanol

Glc 1P Glycogene

2.3 (1.6) 0.4

12

Phosphorylase b +1 mM AMP

GlclP Glycogene

2.4 (1.4) 0.05

25

Glc 1 P Glycogene

1.25 0.04

35

Phosphorylase a

Glc 1 P Glycogene

1.2 0.08

25

Phosphorylase a +0.1 mM AMP

Glc 1 P Glycogene

1 .o 0.035

30

Phosphorylase b + 1 rnM AMP

+ 10%t-butanol

-

1 1 (1.55)

6

kinetics (cooperative noncooperative kinetic responses). Thus, the capacity of a regulatory enzyme to be activated or inhibited by its specific modulator could be abolished, and in some cases some modulators could be bypassed by suitable concentrations of organic solvents without damaging the catalytic activity, but with changes in the characteristic kinetic behavior when the enzyme is "desensitized." Such possibilities may also exist at the level of more highly organized regulatory systems, where they could be exploited to investigate the underlying mechanisms in a step-by-step fashion. As a matter of fact, cosolvents such as primary alcohols, polyols, dimethylformamide and dimethyl sulfoxide are now almost routinely used to perturb the overall reactions and elementary equilibria or rate processes of the highly organized systems carrying out DNA, RNA, and protein synthesis. However, in spite of the fact that such systems respond well and in a reversible way to these perturbations, cosolvent effects remain relatively poor probes of reaction mechanisms (Hamel, 1972; Voigt et al., 1974; Ballesta and Vasquez, 1973; Crepin et al., 1975; Nakanishi et al., 1974; Brody and Leautey, 1973). The most common result reported upon addition of increasing amounts of cosolvents is a bell-shaped curve; equilibria and rate processes are first stimulated and

274

PIERRE DOUZOU AND GREGORY A. PETSKO

next inhibited. The rising and falling portions of the curve express conflicting effects roughly interpreted as perturbations of polynucleotides, of proteins, or of both, but up to now it has been impossible to correlate these data with physical-chemical cosolvent-induced alterations in reaction mechanisms. The fact that structurally unrelated compounds produce similar effects, resulting eventually in a bypass of the requirement of specific modulators for activity, is both an unexpected result and an observation possibly leading to an explanation of the molecular mechanisms by which regulatory systems are activated by physiological modulators. A striking example of such a possibility is provided by investigation of uncharged (vacant) ribosomal particles in the presence of selected amounts of various cosolvents; a dynamic equilibrium between subparticles 30 S and 50 S, and the 70 S complex, is formed in specific ranges of ionic concentration with divalent alkali-earth cations (primarily Mg2+), ensuring the association of 30 S and 50 S subparticles. An increase in the Mg2+concentration increases the association and the stability of 70 S particles, whereas a decrease in Mg2+concentration induces dissociation of the complex. 30s

tMg?*

+ 50s--Mg4* 70 S

This behavior is in accordance with a model in which the principal effect of divalent ions is that of neutralizing, nonspecifically, the charge of the sugar phosphate backbone of ribosomal RNA (this does not exclude the possibility of further sites, involving the ribosomal proteins, in which other divalent metal ions can replace Mg2+). Cosolvents partially replace divalent cations, i.e., induce subunit association at much lower concentrations of Mg2+,presumably by interacting with hydrophobic groups of the proteins and stabilizing an active conformation of the subunits. The fact that cosolvent can partially replace divalent cations led to the postulate that Mg2+,in addition to playing a nonspecific role by neutralizing the negative charge of the sugar phosphate backbone and thus decreasing electrostatic repulsion, binds to the ribosomal proteins and stabilizes an active conformation of the subunits. Thus, cosolvents can be used as perturbing tools and as probes for conformation change after careful, comparative analysis of their effects compared to those induced by various physiological agents, such as protons, monovalent and divalent cations, and other ligands which act to regulate equilibria and rate processes.

PROTEINS AT WORK

275

C. The Problem of the Determination of Conformational Changes Accompanying Enzyme Catalysis While it is tempting to explain regulatory and cosolvent effects on the basis of conformational changes favorable or unfavorable to enzyme activity, it is much more difficult to demonstrate the actual involvement, amount, and structural details of such changes. Experimental evidence consists in most cases of bits and pieces provided by techniques such as absorption and fluorescence spectroscopy, circular dichroism, and magnetic circular dichroism. These tools work in solution (and, when desired, at subzero temperatures) to investigate not simply “empty” enzymes but enzyme-substrate intermediates. However, even with this information, the conformational basis of enzyme activity remains more postulated than demonstrated at the “ball and stick” level, and in spite of data about the number and sequence of intermediates, definition of their approximate nature, rate constants, and identification of the types of catalysis involved, full explanation of any particular reaction cannot be given and rests on speculative hypothesis. Subtle changes in protein conformation as well as more extensive, coordinated atom readjustments are often hard to detect and are even harder to quantify. The amount of conformational change needed to affect enzyme specific activity is unknown and should vary from one enzyme to another, but it may be assumed that only a fraction of an angstrom change in the orientation of catalytic groups at the active site, as well as conformational relaxations of the entire protein molecule between different substates of slightly different structures, could produce significant changes in reaction rates. At the present time the only estimate of the magnitude of conformational changes in proteins in solution comes from the change in enthalpy and entropy (Lumry and Rajender, 1970), although differences in heat capacity (AC, = dAH/dT) can provide additional information about the magnitude of such changes. As pointed out by Lumry and Rajender (1970), thermodynamics provide a simple theoretical basis for expecting conformational changes whenever solution composition, temperature, or even applied pressure (in an ultracentrifuge, for example) is changed; this thermodynamic basis rests on the generalized Gibbs-Duhem relationship. The change in energy of a protein at fixed concentration is given by Eq. (7), in which

dE

= TdS

- PdV

+ W w d V , + W p d V p+ S dA

(7)

,’&l is the infinitesimal amount of work required to change the total free volume of water V , by dV, at S, V, V,, and A constant, W , is the same

276

PIERRE DOUZOU AND GREGORY A. PETSKO

relationship for a change dV, in total protein volume, and 6 is the infinitesimal amount of work required to change the total interfacial surface area A by dA at V , S, V , , and V , constant, i.e., (dW/dA)S V * V , * V,. Then according to the Gibbs-Duhem equation S dT - V d P

+ V , dW, + V , dW, + A d6 = 0

(8)

and any change in T , P, Wp , or 6 must be accompanied by changes in one or more of the other terms. When T and P are held constant, Eq. (8) becomes A d6 = -Vw dW, dG = -S dT

- V , dWp

(9)

+ V dP + W, dV, + W, dV, + 6 dA

(10)

-

We can replace Eq. (8) by Eq. (10) so that W, = (dG/aV,) T * P * V , A , W , = (dG/aV,)T P P,A, and 6 = (dG/dA)T * P * P,V,. One can write that (aG/dV)T * P V , A = p-' in which p represents the compressibility of the total solution, and Ww and may be written as pW-land Pp-', respectively. pwIs the compressibility of water with respect to change in its free volume; pp is the compressibility of the protein. These quantities can be expressed as

-

- -

wp

They define an effective pressure resisting reductions in the free volume of the water and an effective pressure resisting protein compression. Then Eq. (9) can be written as

The quantity 6 is the specific interfacial free energy. Any change in 6 will cause changes in A, V,, and Vp according to Eq. (1 1). Thus, any change in the independent variables P , T , pp, p,, and 6 would produce a change in the volume and surface area of the protein; substrates which either bind on the surface of the protein or bind within the protein and change p,, as well as cosolvents which have an effect on P,, can be included in this treatment. As pointed out by Lumry and Rajender (1970),it is not known a priori whether the above changes are large or whether they will make large contributions to the total enthalpy, entropy, and free energy. It may be assumed that in a number of proteins the magnitude of the resulting changes may prove to be so small that the effects may be ignored. How-

PROTEINS AT WORK

277

ever, when protein surfaces derive their polar character more from the distribution of ionized groups than from that of polar, un-ionized groups, changes in ionization could produce large changes in protein geometry (since high polarity means low interfacial free energy). Since changes in enzyme conformation are often accompanied by large changes in volume (Koshland and Neet, 1968), particularly when the enzyme-substrate ground-state complex is activated to the transition state (activation volume, AW), an experimental approach to the study of conformational changes is provided by investigations under high electrostatic pressures. The pressure dependence of catalytic rates, determined by Laidler (1950), can be expressed by a basic equation relating reaction velocity to pressure: kp =

ko exp(-P AVVRT)

(12)

where kp is the velocity at a gauge pressure P (expressed in atmospheres), ko is the corresponding velocity at 1 atm, and R is the gas constant. Activation volumes or volume changes, i.e., the volume of the activated enzyme-substrate complex ( VEs*) minus the volume of the enzyme system containing the ground state complex (VEs), can be computed from the slope of a plot of the logarithm of the maximal velocity (V,,) of the reaction as a function of pressure (Low and Somero, 1975a,b). This procedure is formally identical to the Arrhenius plot technique for determining activation energies as shown by Eq. (13), in

AV* = RT(1n k2

- In kl)l(P2 - PI)

which k2 and kl are the reaction velocities at pressure P2 and P I , respectively. Thus, volume changes associated with the activation of enzymesubstrate ground-state complexes (E-S + E-S*) or with the regulation of allosteric enzymes by positive and negative effectors, and with the interaction of different low-molecular-weight solutes (ions, polyatomic molecules, organic solvents), can be determined. The energetic costs of such volume (conformational) changes can be derived from similar experiments as a function of temperature (Low and Somero, 1975b;Johnson and Eyring, 1970). It is obvious that an understanding of the structural basis for such volume changes might markedly improve our knowledge of the conformational changes occurring during reaction pathways. Adaptation of pressure cells to subzero temperatures should permit investigation of volume changes at the level of elementary steps, i.e., during interconversion of two consecutive intermediates or upon interaction of some intermediates with protein effectors.

278

PIERRE DOUZOU AND GREGORY A. PETSKO

Protein-protein interactions can be modified by pressure changes according to the basic equation relating equilibrium interactions to pressure: K, = KO exp(-P AVIRT)

(14)

where K, is the association constant at a gauge pressure P,and KO is the corresponding association constant value at 1 atm pressure. Since the equilibrium constant is a function of pressure, the higher the AV value, the higher the change in Kp as a function of pressure. On the other hand, in accordance with Le Chatelier’s principle, Eq. (14) shows that a pressure increase will drive the equilibrium back to subunits. Thus, investigation under pressure might modify a number of equilibrium processes and at the same time contribute to the knowledge of the structural and energetic characteristics of these processes. A convenient form of Eq. (14) for computing the volume change between the initial and final states of an equilibrium is

V = 2.303RT [10g10(KP2/KP1)](P2 - PI)

(15)

Other independent variables (changes in temperature or medium composition) should produce effects analogous to pressure and therefore induce phase changes and shifts in association equilibria. Combination of these different variables could be used to investigate proteinsubunit interactions and conformational changes, to determine the fundamental physical-chemical parameters of these changes. Preliminary investigations of enzyme systems (Johnson and Eyring, 1970) lead to the hypothesis that amino acid side chains and peptide linkages located on or near the enzyme surface change their exposure to bulk water during conformational events in catalysis, with accompanying large volume and energy changes. On the other hand, high- and lowmolecular-weight solutes (including cosolvents)can modify such changes by influencing the degree to which water can organize around surface amino acid side chains and peptide linkages. When one considers the vast number of such protein groups (Klotz, 1970), it becomes clear that changes in various independent variables will significantly contribute to regulation of catalytic rate, leading to better understanding of the structural basis (through A W ) and energetic consequences (through AC *) of conformational changes. Perturbation of the fundamental thermodynamic variables pressure and temperature can thus be used to obtain the temporal resolution of every kinetically significant step in an enzymatic reaction. Such perturbations, combined with pH-dependence studies and several different spectroscopic tools, will detect conformational changes if they occur dur-

PROTEINS AT WORK

279

ing the reaction. These studies will also permit the characterization of the volume and energy changes accompanying each change in structure. But in spite of such data, it is impossible by solution experiments alone to establish what these conformational changes are at the atomic and molecular levels. Analysis carried out on the high spin c) low spin equilibrium of cytochrome P-450 under the influence of various physical and chemical parameters described above clearly shows the multiplicity of protein conformations. The mutual influencesof ligands bound to cytochrome P450 are incompatible with the idea of unique, exclusive conformations. As pointed out by Weber (1979),one can explain the dynamic properties of proteins in solution without introducing any principles other than those involved in the case of much simpler molecules. Weber stresses that proteins are composed of a number of semiindependent domains in which the stabilizing influences do not extend beyond the nearest neighbors, and that if the number of these domains is at all large the probability of finding all of them at once in the average conformation becomes very low. Each state is therefore an average state of unknown composition involving multiple structures (a situation that does not differ in principle from that in a crystal: lattice vacancies are continuously being created and filled, molecular groups undergo rotations and translations, and there is permutation of hydrogen bonds without the crystal losing its identity or changing its properties). These considerations clearly stress that proteins are fluctuating systems moving rapidly from one conformational substate to another at sufficiently high temperature, and also able to undergo triggered conformational changes by binding of substrates and ligands. These two types of mobility involve, of course, different ranges of free energy values and different magnitudes of conformational changes. As stated by Weber, all these complex equilibria contain a wealth of possibilities for the explanation of enzyme-catalyzed processes. The ideal way to investigate both triggered changes in the average conformation and fluctuations about that average should be by X-ray diffraction of protein crystals. We will see later that X-ray diffraction at different selected temperatures provides a map of the dynamic features of proteins. Similar work on proteins interacting with effectors normally involved in the regulation of their functions should reveal new dynamic features, shed light on the molecular basis of the action of these effectors, and confirm our conclusions about cosolvent effects. Thus, investigation of enzyme-catalyzed reactions in cooled solutions provides a wealth of data which should be, when possible, complemented by similar investigation on protein crystals after suitable adapta-

200

PIERRE DOUZOU AND GREGORY A. PETSKO

tion of the low-temperature procedure to the crystalline state. With this technique, much information can be gathered about the kinds of conformational rearrangements which proteins undergo to perform their specific catalytic functions.

TEMPERATURES: PHYSICAL-CHEMICAL 111. X-RAY STUDIESAT SUBZERO BASISOF CRYOPROTECTION OF PROTEINS IN SOLUTION AND IN THE CRYSTALLINE STATEAND RELATED PROBLEMS A. Cryoprotection of Crystalline Proteins

The requirements for a cryosolvent system for crystallographic use are somewhat different from those for solution studies. The requirements also depend on whether or not the system is designed for substrate-binding studies or merely for crystal stabilization at subzero temperatures. If no binding studies are contemplated, the mixed solvent does not need to be fluid at low temperature, and glass-forming organic solvents such as glycerol, ethylene glycol, and 2-methyl-2,4-pentanediol are permissible. Otherwise, the primary requirements for a cryoprotective mother liquor are miscibility with water, low freezing point, low viscoscity even close to the freezing point, and mildness to both the protein and the crystal lattice. The latter requirement means that the protein must not be even slightly soluble in the cryosolvent at low temperature, or the crystal will slowly dissolve. Petsko (1975) has discussed the finding of suitable cryoprotective mother liquors in detail and a further discussion has been provided by Fink and Petsko (1981). In view of the requirements mentioned above and the general large effects on catalysis of high ionic strengths, solvents whose cryoprotection is based on high salt concentration have limited utility. The normal ammonium sulfate or polyethylene glycol mother liquor of most protein crystals will depress the freezing point of the liquid around a mounted enzyme crystal to about - 1 7 " C , a temperature low enough to stabilize some unstable oxidation states of metalloenzymes, for example. Also, radiation damage will be considerably reduced at this temperature over its room temperature rate (see below). For this reason, it is routine practice in our laboratory to collect all room temperature structural data at around -12°C. This policy is recommended for all protein crystallographers. Most protein crystals are not grown from naturally cryoprotecting solvents (even when an alcohol is the precipitant, its concentration is usually well below that required to prevent freezing below -50°C) and therefore must be transferred into such solvents before cryoenzymologi-

PROTEINS AT WORK

28 1

cal experiments can be done. The procedure for stabilizing the crystal depends on the composition of the normal mother liquor. Petsko (1975) identified two cases: proteins crystallized from aqueous-organic mixtures and proteins crystallized from high-salt solutions. More recent developments suggest that this classification needs to be expanded by the addition of two new categories: proteins crystallized from low-salt and proteins crystallized from polyethylene glycol (PEG) solutions. Protein crystals grown from aqueous-organic mixtures are the easiest case. Usually the only problem is to increase the concentration of the alcohol without damaging the crystals. This is best done by coupling a progressive addition of alcohol to a progressive lowering of the temperature. The reasons for this procedure are discussed in detail in Section II1,B. Briefly, it is employed to prevent dielectric shock caused by a sudden increase in organic solvent and to allow for the possible lowering of the denaturation temperature of the protein caused by increasing alcohol. The exact procedure may involve the crystal in a small vial containing mother liquor, with addition being done by pipette and temperature changes being carried out by different baths, or it may involve a crystal mounted in a flow cell (Section IV,B) on the diffractometer, ready for data collection. Examples of the successful application of this method, which is described in detail by Petsko (1975), are ribonuclease A and pepsin. Proteins crystallized from very low salt concentrations (examples are carboxypeptidase A and elastase) can often be treated exactly like proteins crystallized from alcohol-water mixtures. Their low solubility in water allows them to be transferred from their normal mother liquor to a distilled water solution or to a solution of low (10-20%) alcohol concentration without disorder. It is advisable to carry out this transfer at near 0°C to further decrease the protein solubility. From this stage it is trivial to add alcohol while cooling, as described above. Complications arise, however, when the salt employed as a precipitant in the native mother liquor is insoluble in alcohols. The solution to this problem is to replace the salt by ammonium acetate at equivalent or higher ionic strength. Ammonium acetate is soluble up to 1 M in pure methanol, and is very soluble in nearly all alcohol-water mixtures, even at low temperature. It therefore provides a convenient substitute for salts such as sodium sulfate or sodium phosphate. The use of PEG as a precipitant has grown in popularity in the past 10 years. It is normally employed in concentrations ranging from 5 to 20% by weight, and with average molecular weights of 3000 to 20,000, with 4000 or 6000 being the most successful. It is not a good cryoprotectant, and solutions in which it is present at high concentration become very

282

PIERRE DOUZOU AND GREGORY A. PETSKO

viscous at subzero temperatures. It is, however, somewhat soluble in alcohol. We have been able to transfer crystals of yeast triosephosphate isomerase which were grown from 20% PEG-4000 and stored in 35% PEG-4000 directly from their storage mother liquor into 70 :30% methanol-water at 4°C (Alber et al., 1981b). They will dissolve in this alcoholic medium at 4°C in a few days, but are stable at -35°C for months. Another approach which has been used successfully with PEG-grown crystals is to replace the PEG by a relatively high-molecular-weightorganic solvent as an intermediate or final step. For example, MPD at high concentration will often stabilize crystals which have been grown from PEG-3000 or -6000.The reason that PEG-grown crystals seem to be relatively unaffected by such treatment is probably that PEG at the molecular weights used does not enter the crystal lattice. Unfortunately, most protein crystals are still grown from high-salt solutions, and this is the category which is most difficult to stabilize in a cryoprotective solvent. Not only does salt enter the lattice, often making specific interactions with surface residues in the protein, but most common salts are insoluble in alcohol. Salt precipitation always leads to complete destruction of the crystal. The method suggested by Petsko (1975) of direct transfer to MPD solutions of high MPD concentration is often unsuccessful, and even when successful requires many trial-and-error experiments (all fatal to the crystals when they fail) before the exact conditions of temperature and percentage of MPD are found. To make matters worse, MPD is extremely viscous at low temperatures, rendering molecular diffusion almost impossible. Other suggestions that have been made, such as the use of ammonium acetate as an intermediate mother liquor (Fink and Petsko, 1981), have also had very limited utility. We believe that the best answer lies in an approach suggested many years ago by Haas (1968) but never implemented. This method involves the cross-linking of the crystal with glutaraldehyde to render it insoluble in virtually all mother liquors, even distilled water. Once cross-linked, the crystal may be transferred to salt-free mother liquor at the appropriate pH and then treated as though it had been crystallized from low salt. Cross-linking requires very low concentrations of glutaraldehyde (we commonly use 0.1 %) and relatively short exposures to the reagent (overnight in the cold room seems to work well). The objection to this technique is that the intermolecular cross-links may hinder essential conformational changes that occur on substrate binding. The behavior of crystalline ES complexes we have observed argues against this being a problem, but until the approach has been tried extensively this fact cannot be determined. We urge adoption of this method on a trial basis as the best available way of permitting studies on protein crystals grown from high salt.

PROTEINS AT WORK

283

Finally, we note that all transfers to alcohol-water mixtures or additions of alcohol to crystal mother liquor involve changes in the proton activity of the solution. Care must be taken to ensure that the pH* does not change too much, or the crystal may be disrupted. Worse still, the enzymatic activity may be abolished. Control of proton activity in mixed solvents is discussed in Section II1,D. If dielectric effects are controlled and pH* is properly adjusted, the microenvironment of a crystalline protein will correspond closely to that of aqueous solution at room temperature. Such correspondence is essential for temporal resolution of individual steps in a catalytic reaction.

B . Dielectric Constant: Dielectric Shock and Solutions Isodielectric with Water We have seen that crystals can be safely transferred to mixed solvents and that the percentage of organic solvent may often be increased to any desired,level provided that its gradual addition is coupled with a gradual reduction in temperature so as to keep the dielectric constant of the medium as near as possible to the value for the normal mother liquor. Such a result deserves explanation and comment about the behavior of the dielectric constant in mixed solvents as a function of temperature. The dielectric constant is the electrostatic expression of the interaction of atoms and molecules with macroscopic electric fields rather than with the exceedingly strong fields of individual atoms and molecules. The interaction between the homogeneous outside field and electrically asymmetrical (polar) molecules results in a finite effect, since in these molecules the contributions of positive and negative charges do not cancel. Water, hydroxy compounds, and a number of aqueous-organic mixtures are highly dipolar, and in most cases the effect of the individual dipoles is greatly intensified through their connections by hydrogen bonds to form more extended oriented structures. The unique structure of water is expressed by a high value of the dielectric constant and mixing with any miscible polar solvent perturbs its oriented structure and lowers this value. Since the dielectric constant is a fundamental physicochemical property, it is necessary to determine its numerical value in mixed solvents at any temperature. The behavior of any molecule under the influence of an outside electrical field is represented b!/ rigidly attaching to the molecule an imaginary vector. The fact that electric fields cause a net dipole with various media gives rise to their properties as dielectrics, characterized by the dielectric constant D, which, experimentally, is a measure of the relative effect a medium has on the force with which two oppositely charged plates attract each other and is expressed as a unitless number.

284

PIERRE DOUZOU AND GREGORY A. PETSKO

The dielectric constant D is related to the dipole moment p by the Debye expression ( D - 1)/(D + 2 ) = b[ao + (p2/3kT)]n

(16)

with a. = electronic polarizability and n = number of molecules per unit volume. The molar polarizability, P, has the dimension of volume and is given by Eq. (17), in which N is Avogadro’s number.

P =bNa0

+ (p2/3RT) = [ ( D - 1)/(D + 2)](N/n)

(17)

The molar dielectric polarizability of a mixture of two solvents is expressed as P12and is given by Eq. (18), in whichfi andf2 denote mole Pi2 = [ ( D + 2)][(fiM1+ fN2)/d] (18) fractions, M1 and M2 molar weights, and d density of the mixture. This equation is meant only for media of nonassociated compounds in nonpolar solvents, not for molecules like water and the organic solvent used as an antifreeze which are strongly associated by hydrogen bonding and which exhibit other abnormalities in behavior. However, it has been shown by several authors (Wyman, 1936; Timmermans et ad., 1955; Akerlof, 1932) that there is a linear variation of the polarizability (often termed, less appropriately, polarization) with the mole fraction of one of the polar solvents present. Polarizability data for the aqueous-organic mixtures used in our low-temperature method are now known: The points all lie on a straight line when plotted against the organic solvent concentration. The calculation of the dielectric constant of such mixtures from the known polarization of the pure solvents would, assuming this linear variation, give values of low accuracy. It is best to measure the dielectric constant. Even if many organic solvents act as very weak bases, this does not offer any particular difficulties for one can select a method of measurement for which the results would be influenced as little as possible by the conductance of mixed solvents. Large increases in dissociation constant with increasing temperature give less reliable data for dielectric constant even at ordinary temperatures.

I . Dielectric Constant QS a Function of Solvent Concentration Measurements of D carried out at a selected temperature (20°C) as a function of the concentration of organic solvent from 10 to 100% have been published elsewhere (Douzou, 1977a,b).There is a marked proportional decrease in the dielectric constant as the concentration of organic solvent is increased. This effect is very pronounced in the case of methanol and MPD and less pronounced for dimethylformamide and polyols. Adding 50% of most of the organic solvents selected decreases the di-

PROTEINS AT WORK

285

electric constant of water between 10 and 30 units, depending on the solvent, except for dimethyl sulfoxide, the dielectric constant of which is 76.0 (instead of 80 for water at the same temperature).

2. Dielectric Constant and Temperature Since our solvents are permanent dipoles that develop an orientation under the influence of an outside electric field activity in opposition to the disordering influence of thermal agitation, such an orientation process is governed by the Boltzmann distribution law and results in the dielectric constant being strongly dependent on the absolute temperature. Thus, as the systems become cooler, the random motion of their molecules decreases and the electric field becomes very effective in orienting them; the dielectric constant increases markedly with reduction in temperature at constant volume. Molar polarizability, P , can be evaluated from measurements of D by Eq. ( 1 8 ) ,and a critical test of the Debye theory (Debye, 1954) is provided by a plot of P against 1/T, which gives a straight line of slope dPld( 1/T) = 41rNp~I9k

(19)

and intercept 41rNc~o= Po

A vast amount of data based on dielectric constant measurements of substances in dilute solution in nonpolar solvents indicate that the theory is correct, but it looses validity and breaks down completely in the case of strongly polar media. It is interesting to notice that such a breakdown can be shown by the concept of the Curie point, introduced in connection with studies of magnetism but directly applicable to the case of dielectric constants. The dielectric constant can be related to the molar polarizability P and the molar concentration c by Eq. (21) and the value of P itself is given by Eq. (22). It is obvious that P increases as T diminD = (2Pc + 1 ) / ( 1 - Pc) P = (41rN/3)[ao+ (p2/3kT)]

(21)

(22) ishes, mainly because of the presence of 1/T as a multiplier of p2, and that, at a certain critical temperature, P will become equal to unity and D should go to infinity. This critical temperature, if it exists, is known as the Curie point. The sudden increase of D toward infinity at a certain critical value of p2/3kT has been referred to as the 41rII3 catastrophe arising from the presence of such a term in the expression for the electric field strength

286

PIERRE DOUZOU AND GREGORY A. PETSKO

+

(F = E 4 d 3 ) . In fact, no Curie point can be observed with polar liquids as a function of 1/T and the Debye theory is therefore not valid for our strongly polar mixtures. A much more refined theory than that of Debye, which is due to Kirkwood (1939), suffices to account for the general character of the dielectric constant of strongly polar liquids and its changes as a function of temperature. It is not the purpose of this article to give a full account of this theory. a. Dielectric Constant at Subzero Temperatures. In 1932 Akerlof published experimental values for dielectric constants of some aqueousorganic mixtures as a function of temperature between 0 and 80°C. Curves for the logarithm of D of pure solvents (water, methanol, ethylene glycol, glycerol) were plotted against T and gave straight lines obeying Eq. (23), in which a and b are empirical constants and T is absolute

D = ab-bT or log D = a - bT (23) temperature. This relationship has great value for interpolation purposes and therefore is well worth testing over as large a temperature range as possible. It was tested using a number of previous measurements for ethyl bromide and chlorobenzene from -52 to + 126°C and for dimethylpentane from -120 to +80"C and was found valid within experimental error over a temperature range of at least 150°C. Measurements were performed in DOUZOU'S laboratory (Travers and Douzou, 1970, 1974; Douzou et al., 1976) on various aqueous-organic mixtures as a function of the volume ratios at constant temperature and also for selected volume ratios as a function of temperature, between room temperature and the freezing point of the mixtures. The principle results are reported in Fig. 9. b. Preparation of Cooled Media Isodielectric with Water. It can be seen that, for solvent concentrations up to SO%, the dielectric constant of any mixture reaches the value 80, that is, the dielectric constant of pure water at 20"C, well before the freezing point, and that the lower the organic solvent concentration the higher the temperature at which each mixture reaches D = 80. For instance, in the case of ethylene glycol, for the concentrations 10, 20, 30, 40, 50, 60, 70, and 80%, D reaches 80 at 10,5,0, - 10, -20, -30, -40, and -5O"C, respectively. Accordingly, one need only to synchronize addition of organic solvent and lowering of temperature to keep the dielectric constant of a mixed cooled solution at the value 80, that is, at the original value of the aqueous solution. This is an easy procedure for obtaining cooled solutions of nucleic acids and proteins starting from aqueous media at 4°C. Moreover, many problems of precipitation observed upon addition of large volumes of organic solvent at 4°C can be avoided with the synchronized procedure since

287

PROTEINS AT WORK

I 9 J

t

5

8 .0

L

-He

6

I8

I

I

20

I

I

0

I

I

-20

I

I

-40

I

I

-60

I

I

-80

I

I

-100

I

1

1

-120

Temperature ("C)

FIG.9. Logarithm plots of the dielectric constant of water (7), methanol (I), and various aqueous-organic mixtures (2-6, Methanol-water at 80:20,70 :30,60 :40,50: 50, and 40: 60 v/v, respectively, and 8, ethylene glycol-water at 50 :50 v/v) vs temperature. As the concentration of the organic component increases, the dielectric constant decreases, but this effect is reversed by lowering the temperature.

media of high dielectric constant retain their ionization properties. Finally, conformational changes and denaturation due to dielectric constant variations are often avoided by the procedure. In his pioneering experiments, Freed (1965) injected a solution of an enzyme in water cooled to O'C, by means of a micropipette, as a fine spray into a previously cooled aqueous-organic mixture. The spray froze at once as floating particles of ice in which the enzyme was dissolved, and then the solution became clear within seconds to hours, depending on the temperature and the concentration of organic solvent. Such a procedure is hazardous and cannot be applied to a large number of enzyme systems and to the study of reaction kinetics. The synchronizing procedure described above appears to be superior. It might be useful in some cases to raise the dielectric constant of mixed solvents by addition of suitable substances and it is known that dipolar molecules such as amino acids do so in pure water. These amino acids are virtually insoluble in nonpolar solvents but they dissolve readily in aqueous salt solutions and in most mixed solvents according to their highly polar structure. Most of what is known about their dielectric behavior concerns aqueous solutions, in which they were studied up to concentrations near saturation.

288

PIERRE DOUZOU AND GREGORY A. PETSKO

For glycine, it is known that the dielectric constant D of water increases rapidly and linearly with the concentration of the amino acid, reaching a value of about 135 at a concentration of 2.5 mol liter-’ at 25°C. D is given by D = 78.54 + 22.58C, where 78.54 is the measured value of D for pure water, 22.58 is the numerical value of the “dielectric increment,” and C is the concentration in moViiter. This great increase of D reflects the extremely large moment of glycine as a dipolar ion, and the linearity of the relationship represents the proportionality between D and polarizability that is characteristic of strongly polar media. Essentially similar behavior is shown by other a-amino acids, since the numerical value of the slope of the curves dDldC is very nearly the same for all (because all have virtually the same dipole moment). Thus, for a given amino acid, the effect of D at a selected concentration is essentially independent of the nature of the solvent and its numerical value of D . We have said that the quantity dDldC is a direct reflection of the polarizability of molecules and the polarizability is given by Eq. (24), P2

= #(lo006

+ DIVZ)

(24)

derived from the Kirkwood equation, in which subscript 1 is associated with the solvent and subscript 2 with the amino acid. For dipolar ions the dominating term on the right of the above equation is 10006. In the case of the glycine in water, when DlV2 is about 3500, 10006 is about 22,580. The value of 6 varies from 22.58 in pure water to 20.4 in the presence of 60% ethanol at 25°C for glycine in the concentration range from 0 to 0.133 M. Similar results are obtained for the other mixed solvents used to carry out cryobiochemical investigations (Douzou, 1977b). Equation (24) renders intelligible the behavior of the dielectric constant of dipolar ions in polar solutions. It explains the linear increase of D with concentration, since changes in partial molar volumes, only slightly dependent on concentration, can only affect the D1V2 term. It also explains the nearly identical values of D of the amino acids of the same moment, and the fact that D of a given amino acid is insensitive to changes in the dielectric constant of the solvent, for the change of solvent can directly affect 6 only through the term D1V2. The addition of an amino acid to mixed solvents at selected temperatures can be a means to compensate even partially for the decrease of dielectric constant due to the solvent addition. Limitations are imposed by the solubility of the amino acid in such mixtures; for instance, there is a salting-out effect in methanol-water 50 :50 at 25°C when the concentration of glycine is about 0.5 M (6 20). Dielectric constant values listed in Tables V-VII permit safe use of the stepwise procedure for preparing mixed solutions without changing the

-

TABLE V Dichctric Constant ( D ) and Freezing Point of Ethylene Glycol-Water Mixtures"

Temperature ("C)

w solvent

+20

+10

0

-10

0 10 20 30 40 50 60 70 80 90 100

80.4 77.7 75.1 72.0 68.1 64.5 61.1 56.9 53.0 47.5 41.9

84.2 81.4 78.4 75.7 72.1 68.4 64.6 60.0 55.6 50.5 44.7

88.1 85.3 82.5 79.5 76.3 72.4 67.9 63.4 58.8 53.5 47.6

-

a

86.9 84.0 80.2 76.5 72.0 67.5 62.3 56.8 50.3

-20

-30

-40

-

84.4 80.7 76.3 71.3 66.2 60.2

From Travers and Douzou, (1974).

-

-

85 80.8 75.3 70.0 63.8

89.3 85.3 79.8 74.2 67.8

-50

-60

-

-

-

-

-

-70

-80

-90

-100

90.1 95.7 84.5 89.5 94.6 100.0 78.5 83.2 88.1 93.0 - 72.0 -

* The evolution of D as a function of temperature obeys Akerlof's law: log D = a - bT, with T in "C.

ab

bb x 103

1.945 1.931 1.916 1.912 1.880 1.860 1.832 1.803 1.770 1.728 1.675

2.00 2.02 2.15 2.20 2.30 2.35 2.41 2.47 2.50 2.52 2.54

Freezing points ("C)

0 -4 - 10

-17 -26 -44 -69 <-lo0 -83 -50 - 12.5

TABLE VI D i e l e c h Constant and Freezing Point of Methanol-Water Mixtures"

Temperature ("C)

k

10

a

0

bb

solvent

+20

+10

0

-10

0 40 50 60 70 80 100

80.4 63.8 60.3 55.1 46.3 43.7 33.6

84.2 67.7 64.0 58.7 49.4 46.4 35.4

88.1 71.9 67.8 62.5 53.0 49.5 37.9

-

-

75.6 71.2 66.0 56.6 52.3 40.6

79.5 75.5 70.5 60.1 55.4 42.7

a

-20

-30

-40

-50

-60

-70

-80

-90

-100

-

-

-

-

-

-

-

87.9 83.9 78.2 66.8 61.9 48.3

-

-

83.5 79.2 73.8 63.5 58.6 45.4

82.2 70.9 65.7 51.3

-

-

-

86.7 74.9 79.2 83.5 69.3 73.5 77.8 82.8 54.6 58.0 62.0 66.5

-

88.4

-

From Travers and Douzou (1970). The evolution of D as a fundon of temperature obeys Akerlof's law:log D = a - bT, with T in "C.

a'

1.945 1.855 1.830 1.790 1.730 1.695 1.580

x

lo5

2.00 2.20 2.30 2.50 2.45 2.50 2.65

Freezing points ("C) 0

-40 -49 -67 -85 <-lo0 -96

TABLE VII Dielectric Constant and Freezing Point of 2-Methy1-2-4-Pent.anediol-Water Mixture9 Temperature ("C)

% solvent 0 10 20 Es

2

30 40 50 60 70 80 90

100

+20

+10

0

-10

-20

-30

-40

80.4 75.4 70.8 65.7 59.6 53.7 47.7 41.3 35.6 30.6 26.3

84.2 79.2 74.7 69.7 63.7 57 50.6 43.8 37.8 32.6 28.0

88.1 83.2 78.8 73.5 67.3 60.5 53.5 46.8 40.3 34.7 29.9

-

-

-

-

78.0 71.3 64.5 56.9 49.4 42.9 36.9 31.9

75.0 68.3 60.5 52.6 45.3 39.1 31.0

-

-

72.7 64.1 55.7 48.4 41.9 36.2

-100

ab

bb x 103

-

-

-

-

1.945 1.920 1.895 1.866 1.829 1.783 1.728 1.668 1.605 1.540 1.476

2.00 2.13 2.33 2.48 2.50 2.60 2.59 2.60 2.71 2.72 2.80

-50

-60

-70

-80

-90

-

-

-

-

-

-

-

-

77.1 68.1 59.4 51.6 44.6 38.9

-

-

-

-

72.0 62.6 54.9 47.3 41.0

-

-

-

76.5 66.8 58.9 50.7 44.4

81.2 70.6 62.1 54.3 46.9

86.2 75.6 66.5 57.4 50.3

-

From Travers and Douzou (1974). The evolution of D as a function of temperature obeys Akerlof's law: log D = a sc. Supercooling.

-

-

91.7 79.6 70.9 60.9 53.5

97.5 84.7 75.1 64.7 57.4

- bT, with T in "C.

Freezing points ("C) 0 -1.5 -5 - 10.5 -26 -48 scc

sc sc

sc sc

292

PIERRE DOUZOU AND GREGORY A. PETSKO

dielectric constant value from that of pure water. More data are available in the literature (Travers and DOUZOU, 1970, 1974; DOUZOU, 1977a,b). Transfer of a protein crystal in one step from its normal salty mother liquor to a mixed solvent of high enough cosolvent concentration to prevent dissolution often leads to crystal damage as a possible consequence of “dielectric shock.” However, it has been found that once a stepwise transfer has been successfully done in the cold, some crystals may be warmed slowly to room temperature and are stable there, a result indicating that it is not the intrinsic dielectric constant value per se which causes crystal destruction (the case of proteins crystallized from aqueous-organic mixtures is another proof of this), but rather sudden, large changes in the value. Moreover, as we will see below, other parameters could, along with the dielectric constant, induce specific effects in a protein crystal requiring synchronization of cosolvent addition and cooling. C . Cosolvent and Temperature Effects on Protein Interactions The dielectric constant is not the only parameter able to influence crystals, even though its variations could change the electrostatic interactions between positive and negative charges near the contact points between proteins, and thus create disorder or cell dimension changes. Similar alterations could result from cosolvent and temperature effects on hydrophobic interactions which are involved in most protein-protein interactions, as well as on the electrostatic forces at the regions of contact between these proteins (hydrogen bonds, salt links) as a consequence of changes in the pK values of ionizing groups. Forces between protein molecules in crystals are weak. The weakness arises from the fact that each molecule touches neighbors at only a few places (six in sperm whale myoglobin type A crystals and four in horse cytochrome c crystals), each region consisting of several amino acids. The interactions at these few regions of contact may be hydrogen bonds, salt links, or van der Waals forces between nonpolar side chains. Away from the contact points, the forces between molecules are attenuated by the intervening solvent of high dielectric constant, and could be alternatively increased and decreased upon addition of cosolvent and cooling, respectively. Cosolvent and temperature effects on various types of noncovalent forces involved in protein-protein interactions are now well documented. These effects have been intensively studied in Douzou’s laboratory through their impact on protein fractionation (Douzou and Balny, 1978). Mixed solvents at carefully controlled concentration and temperature variations in the range of normal and subzero temperatures ap-

PROTEINS A T WORK

293

pear to be useful tools to modify fractionation patterns as well as probes to investigate protein interactions. Cosolvents which reduce polarity (such as polyols, dioxane) are expected to change interactions of proteins bound by hydrophobic forces and/or hydrogen bonds. As stated by Porath (1974), mixed solvents may be used as hydrogen bond breakers at carefully selected concentrations. Dissimilar cosolvents such as polyols, dimethylformamide, and dioxane were successfully used to remove hydrophobic substrates or ligands from proteins (Debey et al., 1976; Balny et al., 1979) as well as to favor the elution of some proteins from gel matrices. In some cases, cosolvents were unable to improve elution of proteins expected to be bound by hydrophobic forces, presumably because other types of forces were also involved. Thus, organic solvents at selected concentrations can be used to improve elution patterns as well as to detect the intervention of electrostatic bonding. In the last case, controlled variations of pH and ionic strength can reinforce or suppress the bonding and therefore change elution (Anderson et al., 1978). In view of these findings, it is not surprising that crystal transfer to mixed solvents at room temperature often induces alterations of crystal structure or complete disorder. On the other hand, it may be assumed that cosolvents in high concentration (30-60% v/v) which behave as antifreeze agents and allow reduction in temperatures down to -50 to -80°C without freezing will influence markedly any type of bonding. We have found when carrying out the purification of elastase (Balny et al., 1979) that lowering the temperature in the presence of an ethylene glycol-water mixture dissociates the enzyme-substrate complex, which was formed and stabilized on a gel matrix, as a probable consequence of “pure” hydrophobic bonding between the enzyme and its substrate. When polar as well as apolar forces are involved in the adsorption of a protein to a column, subzero temperatures can be used to influence electrostatic interactions and/or hydrogen bonding and then to dramatically modify elution patterns. Such interactions involve large enthalpy values, in general of negative sign, and should therefore be markedly strengthened when temperature is dropped, according to the relationship In K = -AH/RT

+ AS/R

in which K is the association constant. Moreover, it is known that temperature variations modify the pK of ionizing groups and the involvement of these groups in electrostatic and hydrogen bonding. Such changes in pK values in the range of subzero temperatures have been systematically investigated in Douzou’s labora-

294

PIERRE DOUZOU AND GREGORY A. PETSKO

tory on a variety of buffers and ionizing groups of amino acid residues (Douzou et al., 1976), and some implications in noncovalent bonding have been studied. Drawing schematic representations of dissociation curves involved in salt bridges and hydrogen bonds and taking into account the effect of temperature variations on pKs allow us to predict the effects on bonding (Fig. 10). The equilibrium state of any cationic or neutral acid group will be affected by lowering temperature according to the value of its enthalpy of ionization (which might itself vary in mixed solvents). It can be seen in Fig. 10 that, depending on the pH of the medium, lowering temperature can suppress the binding (a) or induce it (b). Accurate choice of buffers with stable or varying pKs as a function of temperature enables the experimenter to preserve binding or to suppress it, respectively.

a

BH+ e! B+H+

AH

E

c

A-

I

+ H+

t

PH

FIG. 10. Changes in pK values as a function of temperature and their possible consequences for electrostatic interactions (a) and hydrogen bonding (b).

PROTEINS A T WORK

295

The behavior of hydrogen bonds depicted in Fig. 10 has been explored for the equilibrium between the dimer and monomer forms of P-lactoglobulin (Hui Bon hoa, et al., 1973)and we have shown that, due to a pK change of 1.85 between 20 and -4O"C, in mixed solvents the dimer form (involving hydrogen bonds) is stable in the pH range where it would be dissociated at room temperature. Thus, hydrogen bonding should be induced or strengthened at subzero temperatures to an extent depending on the enthalpy of ionization of the donor group. It is obvious that temperature variations in the range +20 to -40°C can be used as tools to modify and improve bioaffinity methods and as probes to investigate protein-protein interactions. Preliminary trials have been carried out recently. It has been found that retention of proteins on simple support matrices was obtained in ethylene glycol mixtures at -20°C and the elution was obtained by mere warming to +20"C. Stepwise elution of different proteins (ovalbumin, chymotrypsinogen A, chymotrypsin) on substituted agarose occurred at selected temperatures between -15 and +20"C (Le Peuch and Balny, 1978). Interaction of dinitrophenol (DNP) with an anti-DNP antibody has been found to be markedly enhanced at -30°C and high elution yields have been obtained by warming to +30°C, suggesting the involvement of hydrogen bonding in this antigen-antibody complex in which hydrophobic interactions were considered predominant (Anderson et al., 1978). Thus, cosolvents and subzero temperatures can be used not only for separation purposes, but also to study the modes of interaction of proteins and to gain understanding of the behavior of protein crystals that are to be investigated in unusual conditions of medium and temperature. Concluding this brief survey of the effects of cosolvents and temperatures on noncovalent binding forces between proteins, we may assume that while the dielectric constant may play a role in the cryoprotection of protein crystals, changes in interaction forces may confer protection or in some cases be responsible for crystal destruction. However, we must bear in mind that hydrogen bonds and salt links involved in the regions of contact between proteins will be strengthened and/or stabilized at low temperatures within certain limits of paHvalues, which should aid in the cryoprotection of protein crystals.

D. Control of Protonic Activity While sequential transfer of crystals in cooled mixed solvents is a prerequisite to maintain crystal integrity, it has been found that changes in pH must also be taken into account and that buffers must be selected whose pH values show little change with temperature. Precise control of

296

PIERRE DOUZOU AND GREGORY A. PETSKO

pH (and, in fact, pH) is essential in avoiding crystal alterations on warming as well as on transferring crystals from mother liquors to mixed solvents of very low freezing point. Buffers must be prepared in aqueous solutions at pH values such that, at the temperature of transfer and the cosolvent concentration required, the paHwill equal that of the high-salt mother liquor. First, we describe the principle and measurements of paH values. Acid-base equilibria are quite sensitive to the presence of any organic solvent, but since the interpretation of measured pH values is limited to pure water solution, it has been both necessary to define a proton activity (termed paH) for any mixture and to use conveniently modified electrodes to carry out the potentiometric determinations of paH in the range of normal and subzero temperatures. Measurements of PUH values The potentiometric determinations have been achieved by the use of a glass-calomel electrode assembly described elsewhere (Lanoque et al., 1976). The potential of a glass electrode, reversible with respect to H + , immersed in a solution whose proton activity is aH is given by

Eglass= EOg + ( R T / 9 )In U H

(25)

Similarly, the calomel electrode, reversible with respect to C1-, has a potential

Ecd = E k

+ (RT19)In a c ~

(26)

In these relationships, Eog and E k are, respectively, the standard potentials of the glass and calomel electrodes, 9 is the Faraday constant, and acl is the activity of solvated C1-. If we let aHbe the protonic activity of a solution to be tested and E be the emf measured between the electrodes, then

E

=E,

- E,

+ Ej

(27)

where Ej is the sum of both asymmetry and junction potentials, and p H =

-1ogaH

+

= (E9/2.303RT)- log U C ~ [(E" - Ej)9/2.303RT]

(28)

with E" = E k - E O g . In practice E" cannot be measured and therefore absolute paH determinations are not possible with this method. However, the term -log acl + [(E" - Ej)SF/2.303RT]is a constant for a selected temperature so that

PROTEINS AT WORK

297

relative paH measurements can be performed using a standard medium of reference [termed (st)] under these conditions. P H =

paH(st)

+ [ [ E - E(st)8]/2.303RT]

(29) To perform paH measurements at subzero temperatures, the inner reference solution of the electrode must stay fluid in the whole temperature range investigated. For this purpose we use a solution of M HC1 in an aqueous-organic solvent mixture. The organic solvent must be the same as that present in the test solution. Its volume proportion is arbitrarily chosen as 50%, so very low temperature may be reached without freezing. Under such conditions, HCl has been shown to be fully dissociated (Maurel et al., 1975),so that the inner solution reference paH could be calculated by the Debye-Huckel formula. An empty glass electrode (Tacussel TB-HS) is filled with an HCl, hydroorganic solution (one electrode for each particular organic solvent). As the inner reference electrode, an Ag,AgCl element (Tacussel) is immersed in this internal liquid and the glass electrode is sealed with araldite. When not used, the electrode is stored in a M HCl aqueous-organic solution of exactly the same nature and composition as that contained in the electrode. Before measurements, the glass electrode is equilibrated with a lo-* M HCl aqueous-organic solution of exactly the same composition as that in which measurements are to be made. Such a procedure allows us to reach reasonably low equilibrium times. Basically, the calomel electrode consists of mercury, mercurous chloride (calomel), and chloride ion. The concentration of potassium chloride is 0.1 M in an aqueous-organic solvent (50 : 50) of the same nature as that contained in the solution to be investigated. The junction with the test solution is realized either with a capillary or a porous stone. When the capillary is used, a small hydrostatic pressure is maintained inside it in order to avoid any electrode contamination by the test solution. In the main part of our investigation, the porous stone junction was used. Moreover, the calomel electrode is thermostatted at 20°C, and temperature variations of this electrode giving appreciable emf variations involve uncertainty on the paHdetermination on the order of 0.2-0.3 paH unit/ 10°C. The measuring cell is made from Pyrex. The operational volume of the test solution is 10 ml. The cell is carefully stoppered, and a slow flow of dry N2is maintained inside it. This cell is enclosed by a thermostatic chamber that controls the temperature of the sample. This chamber is isolated from the surrounding atmosphere by a vacuum jacket. The sample, which is continuously stirred magnetically, can be thermostatted from +50 to -45 2 0.05% with a liquid thermostat. The temperature is

298

PIERRE DOUZOU AND GREGORY A. PETSKO

directly measured in the sample with a Chromel-Alumel thermocouple. In all paH measurements at very low temperature (T < - 10°C),in order to avoid thermal shock on the glass electrode (which is itself cooled or warmed very slowly), solutions to be tested are cooled in advance to the temperature of the electrode. Under these conditions the electrical equilibrium time is reasonably short, on the order of 5 minutes or less. Finally, the assembled glass-calomel electrodes and measuring cell are placed into a Faraday box. The ~ U Hresponse of a glass electrode is defined as so that the ideal paH response is, from Eq. (29),

Re = 2.303RTB = 0.198T

(31)

It appears in this formula that Re is only a function of absolute temperature. In practice, no glass electrode has this theoretical response in all types of samples and over the whole paHscale. Nevertheless, if either too acidic or too basic paH values are excluded, it is generally found that the departure from ideal behavior can be neglected. The ~ U Hresponse is tested by means of the paH values as determined by the indicator method: the electromotive force of the cell immersed in buffer solutions whose paH is known is measured and the paH is spectrophotometrically determined and then plotted against E (Fig. 11). It can be seen that for this ethylene glycol-glass electrode the practical response is in good agreement with the theoretical one between paH2 and 9 and for +2 1, 1, and - 19°C. The reproducibility of the determinations, estimated by the use of two different assemblies of electrodes] is better than 2 1.0 mV and the uncertainty of the paH determination is estimated at 20.1 paH unit. The results presented above show that with conveniently modified electrodes] potentiometric determinations of paH can be performed in aqueous-organic solvents at normal and subzero temperatures. In particular] it must be emphasized that these modified electrodes actually demonstrate the pH response theoretically expected. Once the electrodes have been prepared for a given aqueous-organic solvent, the pH determinations can be made at each temperature, either graphically or by direct reading on commercial pH meters calibrated for pH measurements. In this procedure the pH meter is used as a millivoltmeter. A solution A M HCl in the aqueous-organic solvent considered) is selected as the standard reference solution, its paH being calculated for any temperature according to the Debye-Huckel formula. After the electrodes have been immersed in this solution, the

+

PROTEINS AT WORK

299

pOH 10

-

'-

R'= 58.2 mv/po,

mV

mV

FIG.11. A plot of proton activity pH as a function of electromotive force (mV) for the ethylene glycol-glass electrode at t21"C and - 19°C. The response of the electrode (Re)is given by the slope of the line. The concentration of ethylene glycol is 50% by volume. The points are the experimental results in different buffer systems (a, chloroacetate; b, acetate; c, cacodylate; d, Tris): the straight lines represent ideal behavior.

corresponding emf is determined. In the diagram of PUH versus emf as shown in Fig. 11, this reference point is plotted and the theoretical straight line is drawn, the slope of which is calculated from Eq. (31). A solution B to be investigated is then placed into the measuring cell and the emf is determined. Using the diagram of pH versus emf, PuH(B) can be determined graphically. The pH meter is used classically here and calibrated according to the normal procedure using two standard solutions. The only difference with respect to the usual conditions is that these two solutions are solution A M HCl in the hydroorganic solvent), whose PUH has been calculated, and solution B, the PUH of which was determined graphically. Under these conditions the pH meter is calibrated for paH measurements in the aqueous-organic solvent considered. Experimental Results. Experimental values of PUH have been obtained with the potentiometric method as well as originally with the indicator method using colorimetry of acid-base indicators. i. Solvent effects. The addition of increasing concentration of any or-

300

PIERRE DOUZOU AND GREGORY A. PETSKO

ganic solvent to buffer solutions at +20"C can decrease or increase the paH according to the buffer. Upon addition of cosolvent there is a marked increase in the case of neutral buffers (cacodylate, phosphate) and a slight decrease in the case of cationic buffers. There is a relationship between the dissociation constants of acids and bases and the decrease in their dissociation by addition of organic solvent. The lower the dissociation constant, the higher the decrease in the dissociation of buffers. Variations in dielectric constant should alter the relative strength of acids of different charge types, since the amount of electrical work involved in a proton-transfer reaction must vary with the dielectric constant of the medium. Since a lowering in dielectric constant increases the work required to separate the ions (for example, to ionize an uncharged acid one must create an anion and a cation), any addition of organic solvent should lead to an increase of pKa values. Calculations have been carried out by Edsall and Wyman (1958) for dioxane-water solutions (v/v 70 :30), the dielectric constant of which is 18. The increase in the pKa between this medium and water should correspond to 5.3 units at 25°C. The observed pKa values for acetic acid are 4.756 in water and 8.321 in the above medium, that is, a difference in pKa of 3.565. This value is of the same order of magnitude as that calculated from the electrical effects using a very crude model (the anion and the cation resulting from the ionization of acetic acid are treated as two charged spheres). The difference could readily be adjusted by a moderate increase in the assumed values of the radii of the ions. However, when the pKa values in different mixed solvents, such as dioxane-water and methanol-water, are plotted against 1/D, the resulting graph is not linear, as the theory would predict, but concave to the 1/D axis. Thus, the model of charged spheres is too crude, and nonelectrostatic effects due to the solvent itself are presumably also involved. It must be stated again that the composition of the solution around the ions does not correspond to that in the bulk of the solution, so that pKa values do not depend entirely on the recorded macroscopic dielectric constant. With such observations in mind, we must try to gather as many data as possible on the pK values as a function of the concentration of each miscible organic solvent and, for every mixture, as a function of temperature. ii. Temperature effects. Cooling mixed buffered solutions produces an additional variation in paH. The variation is linear as a function of temperature. The ApaH with temperature of Tris buffer can be more than 2 units, whereas the variation for neutral buffers is very slight. The temperature coefficients of log K and log y (where y is the activity

PROTEINS A T WORK

30 1

coefficient), in terms of which the temperature coefficient of paH is expressed, are readily estimated for many buffers from data in the literature. Some general conclusions can be drawn regarding the effect of temperature on pH values. In dilute solution the variation of the activity coefficient with temperature is negative and uniform over the whole pH range and tends to lower dpaHldTfor strong bases, salts, and weak acid buffers with negative 2, and to raise it for strong acids and buffers containing weak bases with positive 2. However, this effect of the altered activity coefficient is small and most often overshadowed by d log KIdT. Thus, changes in concentration are frequently without noticeable effect on temperature coefficients. For these reasons, paH versus temperature curves are of much the same type as the curves of pK versus temperature. At constant pressure we have from the Van't Hoff equation: d log KIdT = AHo/RT2

(32)

where R is the gas constant and AHo is the change of enthalpy for the dissociation of 1 molecule of the acid or base in the standard state. Plots of -1ogK (pK) as a function of temperature are roughly parabolic in form, and pK for many weak electrolytes has a minimum value in the range 0-60°C. Hence d log KIdT may be either positive or negative at room temperature. In general, the characteristic minimum is shifted toward higher temperatures as the acid strength decreases, so that pKa increases with temperature for the stronger acids and higher temperatures and decreases with temperature for weaker acids and with rising temperatures. The paH, like pK, often passes through a minimum within the experimental range, and the effect of the term d log yldT is to shift the minimum to a temperature lower or higher than that characteristic of the dissociation constant alone. Thus, the temperature coefficient of pH may appear small at temperatures in the vicinity of the minimum. Tables VIII-XI show examples of paH variations of several buffers. With such tables, it is easy to adjust any desired paH value in mixed solvents at any selected temperature or in a given range of temperatures. We will see in Section II1,E how these values are essential to investigate safely both crystal structure and productive enzyme-substrate complexes in the crystalline state.

E . Effects of Cosolvent and Temperature on Donnan and Electrostatic Parameters of Protein Crystals Any crystal of a globular protein can be regarded as an ordered and open array of compact molecules that make nearly minimal contact with

302

PIERRE DOUZOU AND GREGORY A. PETSKO

TABLE VIII Protonic Activity of Some Usual Buffersll ~ ~ _ _ _ _ _ _

~

p4H in 50 :50 EGOH :water at given temperature ("C)

PH water Buffer (10-2 M ) Chloroacetate Acetate

Cacodylate

Phosphate

Tris

1

at

t2O"C

t20

0

-10

-20

-30

-40

2.35 2.55 2.85 3.65 4.10 4.40 4.55 4.75 5.4 5.8 6.2 6.4 7.0 6.0 6.5 7.0 7.5 7.0 7.5

2.90 3.25 3.55 4.25 4.65 4.90 5.05 5.25 5.90 6.30 6.65 6.85 7.30 6.60 7.10 7.60 8.10 7.20 7.65 7.95 8.45

2.95 3.30 3.60 4.35 4.75 5.00 5.15 5.35 6.00 6.40 6.75 6.95 7.40 6.75 7.25 7.75 8.25 7.95 8.40 8.70 9.20

2.95 3.30 3.60 4.40 4.80 5.05 5.20 5.40 6.05 6.45 6.80 7.00 7.45 6.80 7.30 7.80 8.30 8.35 8.80 9.10 9.60

3.00 3.35 3.65 4.45 4.85 5.10 5.25 5.45 6.10 6.50 6.85 7.05 7.50 6.90 7.40 7.90 8.40 8.75 9.20 9.50 10.00

3.05 3.40 3.70 4.50 4.90 5.15 5.30 5.50 6.15 6.55 6.90 7.10 7.55 7.00 7.50 8.00 8.50 9.20 9.65 9.95 10.45

3.05 3.40 3.70 4.55 4.95 5.20 5.35 5.55 6.25 6.65 7.00 7.20 7.65 7.10 7.60 8.10 8.60 9.70 10.15 10.45 10.95

8.0 8.5 a

From Hui Bon Hoa and Douzou (1973), Maurel

et

al. (1975).

interstitial regions that comprise about half of the crystal volume and are filled with solvent that is similar to the bulk liquid. In order to gain information about the behavior of crystals in mixed solvents, it is appropriate to approach protein crystals as if they were highly ordered polyelectrolyte gels (Rupley, 1969). The potentiometric titration curves of gels, which relate the pH of the exterior solution to the degree of ionization of the gel, resemble the titration curves of monofunctional acids or bases. However, the dissociation constants differ, often by two orders of magnitude, from the expected value for the functional group, and the slope of the curves is not the usual one. Addition of neutral salt changes the picture markedly and brings the curves closer to expectation. In the case of weak or medium

303

PROTEINS AT WORK

TABLE IX Protonic Activity of Some Usual BufmP

Buffer (lo-*M ) Chloroacetate Acetate

Cacodyla te

Phosphate

Tris

PH water at +2O"C

2.35 2.55 2.85 3.65 4.10 4.40 4.55 4.75 5.4 5.8 6.2 6.4

7.0 6.0 6.5 7.0 7.5 8.0 7.0 7.5 8.0 8.5 9.0

paHin 50 :50 MeOH :water at given temperature ("C)

+20

0

2.90 3.25 3.60 4.50 4.85 5.10 5.30 5.45 6.30 6.60 6.95 7.10 7.55 7.20 7.65 8.10 8.50 8.90 7.25 7.70 7.95 8.40 8.85

2.95 3.30 3.65 4.55 4.90 5.15 5.35 5.50 6.30 6.60 6.95 7.10 7.55 7.25 7.70 8.15 8.55 8.95 7.80 8.25 8.50 8.95 9.40

-10

-20

-30

-40

2.95 3.00 3.05 3.10 3.30 3.35 3.40 3.45 3.65 3.70 3.75 3.80 4.55 4.60 4.60 4.65 4.90 4.95 4.95 5.00 5.15 5.20 5.20 5.25 5.35 5.40 5.40 5.45 5.50 5.55 5.55 5.60 6.35 6.35 6.40 6.40 6.65 6.65 6.70 6.70 7.00 7.00 7.05 7.05 7.15 7.15 7.20 7.20 7.60 7.60 7.65 7.65 7.25 7.30 7.30 7.35 7.70 7.75 7.75 7.80 8.15 8.20 8.20 8.25 8.55 8.60 8.60 8.65 8.95 9.00 9.00 9.05 8.15 8.55 8.90 9.35 8.60 9.00 9.35 9.80 8.85 9.25 9.60 10.05 9.30 9.70 10.05 10.50 9.75 10.15 10.50 10.95

~

From Hui Bon Hoa and Douzou (1973).

polyacid gels the pH at half neutralization is always much higher than that of the monofunctional group, while in the presence of salt the pH is depressed and comes closer to that characteristic of the group. This has been shown with poly(methacry1ic acid) ion exchangers. The behavior of hydrogen ions in the gel is very similar to that in polyelectrolyte solutions and the same theoretical treatment may be applied. The pH of a polyelectrolyte solution is dependent on the intrinsic dissociation constant (pK") of the functional group-which is normally that of a monomer-on the degree of ionization (a), and on the potential

304

PIERRE DOUZOU AND GREGORY A. PETSKO

TABLE X Protonic Activity of Some Uswl Buffe7J"

paH in 70 :30 MeOH :water at given temperature ("C)

Buffer

PH water at

(lo-*M )

+20°C

+20

2.35 2.55 2.85 3.65 4.10 4.40 4.55 4.75 5.4 5.8 6.2 6.4 7.0 6.0 6.5 7.0 7.5 8.0 7.0 7.5 8.0 8.5 9.0

3.20 3.70 4.05 5.0 5.35 5.55 5.75 5.95 6.65 7.0 7.35 7.50 8.0 7.75 8.25 8.60 9.10 9.45 7.30 7.70 8.00 8.55 9.10

Chloroacetate Acetate

Cacodylate

Phosphate

Tris

0

-10

3.20 3.20 3.70 3.70 4.05 4.05 5.10 5.10 5.45 5.45 5.65 5.65 5.85 5.85 6.05 6.05 6.70 6.70 7.05 7.05 7.40 7.40 7.55 7.55 8.05 8.05 7.85 7.90 8.35 8.40 8.70 8.75 9.20 9.25 9.55 9.60 7.95 8.40 8.35 8.80 8.65 9.10 9.10 9.65 9.65 10.20

-20

-30

-40

3.20 3.70 4.05 5.15 5.50 5.70 5.90 6.10 6.75 7.10 7.45 7.60 8.10 7.95 8.45 8.80 9.30 9.65 8.80 9.20 9.50 10.05 10.60

3.20 3.70 4.05 5.20 5.55 5.75 5.95 6.15 6.80 7.15 7.50 7.65 8.15 8.00 8.50 8.85 9.35 9.70 9.25 9.65 9.95 10.50 11.05

3.20 3.70 4.05 5.25 5.60 5.80 6.00 6.20 6.80 7.15 7.50 7.65 8.15 8.10 8.60 8.95 9.45 9.80 9.80 10.20 10.50 11.05 11.60

~~

From Hui Bon Hoa and Douzou (1973).

(JI) of the molecular surface. Equation (33) describes the binding of hydrogen ions to a polyacid, pH = pK" - log[(1

- a)/.] + (0.4343~$/kT)

(33)

and a corresponding equation for polybases is given in Eq. (34). pH = pK" - lOg[p/(l

- p)] - (0.4343e$/kT)

(34)

Protein gels show the combined effect of both acid and basic groups; in the acid range they behave as polybase gels and in the alkaline range

305

PROTEINS AT WORK

TABLE XI Protonic Activily of S u m UsualBuffersll

PH water Buffer

pHin 50 :50 MPD at given temperature ("C)

at

M)

+20°C

+20

0

-20

-25

Chloroacetate

2.2 2.6 3.0 3.4 3.8 4.3 4.8 5.3 5.5 6.0 6.5 7.0 6.5 7.0 7.5 8.0 8.0 8.5 9.0 9.5 9.0 9.5 10.0 10.5

2.55 3.24 3.76 4.20 4.55 5.05 5.56 6.04 6.08 6.77 7.34 7.80 7.36 8.05 8.57 8.90 8.10 8.54 8.90 9.20 9.70 10.67 11.16 11.61

2.55 3.24 3.76 4.20 4.59 5.09 5.60 6.08 6.09 6.78 7.35 7.81 7.47 8.16 8.68 9.01 8.64 9.08 9.44 9.74 10.0 10.97 11.46 11.91

2.55 3.24 3.76 4.20 4.65 5.15 5.66 6.14 6.12 6.81 7.38 7.84 7.78 8.47 8.99 9.32 9.39 9.83 10.19 10.49 10.50 11.47 11.96 12.41

2.55 3.24 3.76 4.20 4.66 5.16 5.67 6.15 6.12 6.81 7.38 7.84 7.87 8.56 9.08 9.41 9.60 10.04 10.40 10.70 10.65 11.62 12.11 12.56

Acetate

Cacodylate

Phosphate

Tris

Carbonate

From Hui Bon Hoa and Douzou (1973).

as polyacid gels (Katchalsky, 1954). Equation (35) describes the binding of hydrogen ions to a charged protein gel:

where pK" is the intrinsic pK of an ionizing group, a the fraction ionized, A- a univalent anion distributed between the solution and protein phases, and B an electrostatic interaction constant for the solid macromolecule of net charge 2. For an anionic protein the Donnan and electrostatic corrections [the third and fourth terms, respectively, on the right-hand side of Eq. (35)] are each positive and lead to a higher appar-

306

PIERRE DOUZOU AND GREGORY A. PETSKO

ent pK. Although at low ionic strengths these effects can be several pH units in size, they are strongly ionic-strength dependent. The Donnan term is swamped when the free ion concentration is several times that of the net fixed charge within the solid, and the electrostatic term becomes negligible at ionic strength 0.5 to 1.0. Thus, there should be no particular difficulty in studying enzyme reactions in the crystalline state, provided the ionic strength is reasonably high, which is the case for protein crystals in their usual mother liquors. 1 . Cosolvent and Temperature Effects When the salt concentration is high enough (> 0.1 M ) some of the salt will precipitate in proportion as the cosolvent concentration is increased and the temperature of the mixture decreased. While in such conditions the precipitating effect of the cosolvent toward proteins will increase (Edsall, 1947; Singer, 1962) and a new equilibrium of ions between the crystals (Zin) and the bulk solvent medium (I""') will be established, modifying both the Donnan and electrostatic terms. In the presence of salt (say sodium chloride), the distribution of the hydrogen ions is correlated with that of the chloride ions by the Donnan relationship: ag$/ai;;+= a & - / a z i

(36)

which gives, when we take negative logarithms and substitute pH for -log aH pH""' = pHin

+ log(a~!./a&-)

(37)

Introducing into Eq. (37) the expression for pHin from Eq. (33) and for the anion activities (aod;L/a&-) = ( a $ i / a $ ) , we obtain for the titration of a polyacid gel in the presence of salt pH""' = pK" - log[(l - (Y)/(Y]+ lOg(C$/C~l-) -I-(0.4343/kT)(dFe/dV)~

(38)

While simplified equations, neglecting the potential term (dF,/dv)K, have been used in the analysis of gel titrations, it must be stressed that at low salt concentrations this term may contribute up to two pH units and its neglect may lead to significant errors. Equation (38) gives an explanation for the curves reported in different salt solutions: At high salt concentration the potential goes to zero and c$ approaches c&- , so that both terms diminish and the titration curve approaches that of a monofunctional acid. At low salt concentrations the potential is high and the Donnan distribution requires C ~ >L

PROTEINS AT WORK

307

c $ - . Hence, the last two terms on the right-hand side of Eq. (38) make a positive contribution to the pH. An inspection of the equation leads to the conclusion that in the case of polybasic gels the pH should exhibit the opposite behavior. Such changes in the values of electrostatic potentials in protein crystals have essential implications in both crystal structure and enzyme activity in the crystalline state. In recent years, it has been shown (Douzou and Maurel, 1976) that some proteins can behave as polyanions or polycations, and the stability of their solid state might be endangered at lower salt concentration due to repulsive forces between protein molecules. Much more important is the problem of enzyme activity in crystals suspended in cooled mixed solvents as a consequence of cosolvent- and temperature-induced changes in salt concentration and therefore in electrostatic potentials.

2. Catalytic Implications of Electrostatic Potentials a. Theoretical Aspects. Very few studies have been devoted to the influence of electrostatic potentials developed by polyelectrolytic environments on enzyme catalysis. In 1957, McLaren et al. reported that achymotrypsin functioning at the interphase waterlsolid (kaolinite) exhibited a pH profile different from that obtained with the free enzyme in solution. Later, Goldstein et al. (1964) and Goldstein (1972), studying the kinetic behavior of enzymes attached to a polyelectrolytic matrix (trypsin and a-chymotrypsin covalently bound to copolymers of maleic acid and ethylene), introduced a treatment of the catalytic implications of the polyelectrolyte theory. More recently, Engasser and Horvath (1975) made an extension of this treatment in which they took into account the influence of substrate, products, and/or inhibitors in the cases where these agents significantly affect the ionic strength of the medium. Consider here the simplified treatment developed by Katchalsky and co-workers (Goldstein et al., 1964). This treatment is based on the following restrictions: (1) the enzyme is maintained within the polyelectrolytic environment through a binding which does not affect its catalytic functions; (2) the electrostatic potential generated within the environment is assumed to be homogeneous and pH independent (i.e., the pK of the ionizable groups responsible for this electrostatic potential is considered as either very low or very high); (3) it is supposed that the electrostatic potential developed by the polyelectrolyticcarrier does not influence the structure of the latter nor the structure of the enzyme (in other words, the treatment does not take into account any structure-function relation introduced by the electrostatic potential of the environment); (4) the possible influence of the high density of the electric charges on the

308

PIERRE DOUZOU AND GREGORY A. PETSKO

structure of water in the environment is not considered; and ( 5 ) the contribution to the total ionic strength, brought about by substrates or inhibitors when these are charged molecules, is assumed to be negligible. This treatment actually concerns only the influence of the electrostatic potential prevailing within the environment on the local concentration of protons and substrates. In the environment of any polyelectrolyte, an electrostatic potential JI prevails, This electrostatic potential is a result of not only the high density of permanent charges pp borne by the polyelectrolyte, but it also results from the density of the mobile charges pmbrought about by the ionic species present in the solution, which form an ionic atmosphere around the polyelectrolyte. According to the Boltzmann distribution law, the local concentration of any ionic species X contributing to pmis given by Eq. (39), in which Zx is the algebraic value of the charge borne

[XI,,,

= [XI0 exp(-ZxWT)

(39)

by X, e is the unit electric charge, and kT has its usual significance. Charge density and electrostatic potential are related through the fundamental Poisson-Boltzmann equation [Eq. (40)], in which A is the Laplace operator, and D the dielectric constant of the medium.

A$

= (4M/D)(pp + P m )

(40)

If we consider a system polyelectrolyte-enzyme, the catalytic implications of the polyelectrolyte theory become evident when Eq. (39) is applied to protons, substrates, and inhibitors (as long as these are charged molecules). Equation (39) shows that the average concentration of protons in the polyelectrolytic environment is different from that of the bulk medium: Theoretically, JI is a decreasing function of the distance. However, for the sake of clarity and simplicity, we shall consider here that JI within the environment has an average value denoted 4. It follows from this relation that any ionizable group submitted to the influence of I$ has an apparent pK($) given by Eq. (42), in which pK" is the pK of the group in pK(I$) = pK" - 0.43t@kT

(42)

the absence of any electrostatic potential. Thus, the pKs of all the ionizable groups of the enzyme (or only of part of them, depending on whether the enzyme is completely or partly exposed to the electrostatic

PROTEINS AT WORK

309

potential of the environment) are modified according to Eq. (42). This, of course, includes the pKs of those groups apparently involved in the active site. Therefore, as a first consequence of the presence of the electrostatic potential, the pH profile of the enzyme located inside the polyelectrolytic environment is different from that observed with the free enzyme in solution: it appears to be shifted by an amount 0.43~I$/kT. The direction of the shift depends on the nature of the environment. In the case of a polyanionic environment surrounding the enzyme (I$ < 0, and accordingly pK($) > pK"), the pH optimum of the reaction will be displaced toward alkaline pHs. The reverse shift will be observed, i.e., toward acidic pHs, in a polycationic environment ($ > 0). Most natural substrates of enzyme systems are charged molecules. Their local concentration within any polyelectrolytic environment is given by Eq. (43). Therefore, as a second consequence of the presence of the electrostatic potential, the apparent Michaelis constant of the reac-

tion within the polyelectrolytic environment is different from that obtained with the free enzyme in solution [Eq. (44)]. We can see that in a polyanionic environment positively charged substrate molecules will tend to concentrate and the resulting K, will be lower than K," obtained

in the absence of any electrostatic potential. The same effect can be seen concerning inhibitor binding constants when the inhibitor is a charged molecule. Ionic control of enzyme reactions at the level of polyelectrolytic microenvironments provided by biological organization (biological membranes, nucleic acids, ribosomes, etc.) can therefore be expected and understood. This ionic control is based upon the following points.

1. A solution of any polyelectrolyte-enzyme system might be considered as consisting of two phases in equilibrium, although, within limits, the polyelectrolyte is soluble. Owing to the local electrostatic potential, the inner polyelectrolytic phase, or environment, possesses its own local physicochemical parameters which differ from those of the bulk phase. 2. The enzyme activity in the system is therefore controlled by the local electrostatic potential. 3. Any change in this electrostatic potential affects the equilibrium between the inner and the bulk phase, and, as a result, the enzyme

310

PIERRE DOUZOU AND GREGORY A. PETSKO

activity of the system changes. Those agents able to modulate the electrostatic potential of the system are obviously ionic species.

6. Experimental Aspects. Consider ionic strength as a modulator since it is the only one which is dealt with in both qualitative and quantitative terms by the polyelectrolyte theory (Maurel, 1976). The catalytic impli-

cations of this theory can be detected experimentally at three levels: pH profile of the enzyme reaction in the polyelectrolyte-enzyme system at various ionic strengths (or various concentrations of modulators), activity-ionic strength plots at various pHs, and ionic strength effect on the Michaelis constant of the enzyme. The ionic atmosphere created in the surrounding of the polyelectrolyte by the ionic species present in solution, whose distribution is given by Eq. (33),tends to minimize the electrostatic potential through the socalled electrostatic screening effect. Its action on the electrostatic potential is reasonably accounted for by the following semiempirical relation: &I,

=

&+O)

- A log1

(45)

where absolute magnitude of the potential is considered, A is a positive constant, and Z the ionic strength. Obviously, this relation holds only over a limited range of ionic strength, which, nevertheless, roughly coincides with the most frequently used conditions ( 10-4-10-1M ) . Now, combining Eqs. (45)and (42)it can be predicted that as ionic strength increases in the medium, 4 decreases, and the pH optimum of the enzyme reaction inside the environment tends, linearly with log I , toward the pH optimum of the same reaction with the enzyme free in solution. An alternate representation of this interdependence is a plot of enzyme activity (as V,,,,,,for instance) against ionic strength for various values of the pH, as shown in Fig. 12, for a polyanionic environment. Such a plot is frequently reported in the literature. Bell-shaped curves are commonly observed. This representation is actually the direct consequence of the pH profile shift (Fig. 12) as ionic strength increases: the lower the pH, the higher the ionic strength giving the optimum. However, in contrast to what is shown in Fig. 12, ionic strength in a real system cannot be decreased to zero. A certain amount of salt, Zmin, has to be kept in the solution to avoid denaturation and to maintain active conformations. Thus, ionic strength increases not from zero but from a minimum value, Zmin. Accordingly, a part of the curve is missing on the experimental activity-ionic strength plots (Fig. 13) and two kinds of curves are then generated: type I curves, bell-shaped; and type I1 curves, monotonically decreasing, depending on the pH at which the experiment has been performed. It can be seen that, if this pH is lower

31 1

PROTEINS AT WORK

a

I

l

I

l

C

< I , < I,

PH

FIG. 12. Plots of the V,, of a typical enzyme reaction as a function of pH for various conditions of ionic strength, for a polyanionic environment. For details see text.

312

PIERRE DOUZOU AND GREGORY A. PETSKO a

4

PH

b

I

I

FIG.13. The effect of the minimum ionic strength, I,, on the pH-rate profile for a typical enzymatic reaction. Two types of curves are generated: Type I, bell shaped; type 11, monotonically decreasing, depending on the pH of the experiment. Graphing the pH behavior as a function of ionic strength (a and b show the transformation) and applying the I, cut-off (c), it can be seen that, if experimental pH at I, is lower than the pH optimum, a type I curve is obtained. If the experimental pH is greater, a type I1 is obtained.

than the pH optimum, at Zmin a type I curve is generated. A type I1 curve is obtained when the pH is higher than the pH optimum. On the other hand, combinations between Eqs. (44) and (45) show how, via modification of the electrostatic potential, ionic strength affects K , (and KI, inhibitor constants), and thus the enzyme activity. A striking example of the catalytic implications of electrostatic poten-

PROTEINS AT WORK

313

tials is provided by the study of the lytic activity of lysozyme on lowmolecular-weight molecules, among which the p( 1 + 4)-linked hexasaccharide of N-acetylglucosamine and the corresponding hexasaccharide extracted from cell wall digests have been the most widely investigated. A great deal of work has been carried out with these small substrates and the mechanism of the reaction is now well established. In particular, the pH profile for the hydrolysis of oligosaccharides has a pH optimum at 5.2, resulting from the presence of two ionizable groups in the active site of the enzyme: a deprotonated carboxyl of Asp-52 and a protonated carboxyl of Glu-35, both identified by X-ray diffraction studies (Blake et al., 1967). Moreover, the pH profile is independent of ionic strength (Maurel and DOUZOU, 1976). Because of its physiological importance, many experimental studies were done on the reaction of lysis of bacterial cell walls (Imoto et al., 1972). Most of these studies show that, in contrast to the reaction with oligosaccharides, the lytic activity of lysozyme presents a number of peculiar characteristics which we can summarize as follows. 1. At low ionic strength (0.01) the pH profile of the lytic reaction has an optimum at about 9.0-10.0, i.e., in a range where both carboxyl groups of the active site would be deprotonated and consequently the enzyme should be inactive. 2. As ionic strength increases, the pH optimum of the reaction shifts toward acidic pHs; at sufficiently high ionic strength (0.16) it seems to be stabilized near 5.5. (Fig. 14). Moreover, it appears that pH optimum is a linear function of the logarithm of ionic strength. 3. As lytic activity is plotted against ionic strength at constant pH, the reaction is first activated, reaches a maximum, and then is inhibited; the lower the pH at which this plot is obtained, the higher the ionic strength giving the maximal activity. 4. Finally, K,(app) of the reaction increases strongly as ionic strength increases. Indeed, these results are qualitatively similar to those depicted in Figs. 12 and 13, and it is therefore tempting to ascribe them to the presence, on the bacterial cell walls, of polyanionic environments surrounding the sites where lysozyme cleaves the polysaccharide network. However, the point is that, although several authors have concluded from various experimental observations that the cell walls of bacteria such as Escherichiu cola and Micrococcus luteus are predominantly negatively charged (Kateralsky et al., 1953; Salton, 1964; Davies et al., 1969), the complexity of the bacterial cell wall architecture means that little is known about the

3 14

PIERRE DOUZOU AND GREGORY A. PETSKO

I

I

I = 0.06

20

15

& c

X

1

.Y

10

5(

PH

FIG. 14. The effect of ionic strength on the pH-rate profile of the lysozyme-catalyzed hydrolysis of M. lysodeihticus cell walls. As ionic strength increases, the pH optimum shifts toward acidic pH values.

315

PROTEINS AT WORK

arrangement of its ionizing groups and, accordingly, its electrical configuration. We therefore planned a seties of experiments to ascertain the existence of polyanionic sites on the cell walls of M. luteus, a finding which would strongly support the interpretation of the previous results in terms of the polyelectrolyte theory. These experiments are reported in Fig. 15. They are based on a well-known property of these cell walls: their aggregation in the presence of polycations such as poly(Lys) in the suspending medium (Maurel and DOUZOU, 1936). This aggregation, monitored by the increase in turbidity of the suspension, is specifically produced by polycations; polyanions such as poly(G1u) are inefficient in this process. As the concentration of polycations increases in the medium, the turbidity of the suspension increases first linearly, then reaches a plateau, indicating that the aggregation is at its maximum. If lysozyme is simultaneously added to the suspension and the rate of reaction determined, a strong inhibition is observed as a function of the polycation concentration. Figure 15 shows that the plots of turbidity change against poly(Lys) concentration and of percentage inhibition against poly(Lys)concentration are symmetrical. These data suggest that

1.a

0.1

t>

5

0 0

0.5

0.05

t (seconds) I

300 0

t

-0

0 I

I

2

1

I

4

I

I

I

6

I

8

I

I

10

Poly(Lys), poly(Glu) (10-8M)

FIG. 15. Inhibition of the lysis of Micrococcus luteus by lysozyrne, klko (ko = activity in the absence of inhibitor) (A, 0),and optical density increase of an M. luteus suspension at 500 nm, both as a function of polyelectrolyteConcentration.Poly(Lys),0;Poly(Glu),0 ;pH 8.5, I = 0.01, lysozyme 0.55 X lO-'M, M . luteus 100 mg liter-', +20°C. Inset: Recording vs time of the optical density increase of an M. luteus suspension on addition of poly(Lys)at 2.5 X lo-" M, at +20°C.

316

PIERRE DOUZOU AND GREGORY A. PETSKO

lysozyme and poly(Lys)compete for the same sites of the cell wall. Thus, it can be reasonably concluded that the bonds susceptible to lysozyme are located within (or in the immediate proximity of) polyanionic regions. Further supporting evidence for this conclusion is brought about by the experiments depicted in Fig. 16. In water at normal temperature, lysozyme activity is monitored by the decrease in turbidity which follows the enzyme addition to the bacterial suspension. Now, considering the facts that (1) lysozyme is, at neutral pH, a polycationic protein (isoionic point 11.2) and (2) very low temperature would stop the lytic reaction according to its high energy of activation, we expected that the addition of

0.32 .

P

0

0

0

200

400

000

t (seconds)

FIG. 16. Recordings versus time of optical density of an M. luteus suspension on addition of lysozyme. At +20"C one observes the lytic reaction leading to a decrease in OD. At -30°C in 40% methanol, I = 0.01, pH 6.5, lysozyme induces aggregation of the cells without lysis (lysozyme M , M luteus 50 mg liter-'). Inset: Optical density increase at -30°C as a function of lysozyme concentration.

PROTEINS A T WORK

317

lysozyme to a supercooled fluid cell wall suspension would produce an increase in turbidity, reflecting the aggregation induced by the polycationic lysozyme molecule. This was observed in a water-methanol mixture (60:40 v/v) at -30°C. Furthermore, we could show that, as with poly(Lys)under normal conditions, the aggregation reached a plateau as the concentration of lysozyme increased. It is therefore concluded from these experiments that the p(1 --* 4) linkages in the polysaccharide network of the cell walls of M. luteus are located within polyanionic microenvironments. The positively charged molecules of lysozyme are electrostatically attracted by these environments, inside of which the lytic reaction proceeds under physicochemical conditions (especially proton concentration) different from those of the bulk medium. The kinetic observations and the K,(app) variations with ionic strength can therefore be satisfactorily understood in terms of the catalytic implications of the polyelectrolyte theory. c. The Crystalline State. As pointed out above, the effects of electrostatic potentials in protein gels could be several pH units in magnitude at very low ionic strength, but are negligible in protein crystals at ionic strength 0.5 to 1.0. Under such conditions, which correspond to those created by salty mother liquor, the potentials go to zero and there should be no particular difficulty in studying enzyme-catalyzed reactions in the solid state. At low salt concentrations (i.e., with cryoprotective salt-free aqueous-organic mixtures), it may be difficult to compare solutions and solid-phase reactions unless paH values are consequently adjusted by using Tables VIII-XI, reported in the previous section. Due to the tight interdependence between pH and ionic strength values on enzyme activity, it will be possible to adjust pH values to compensate for the effects of salt precipitation and then to obtain solid-phase reactions presenting the characteristics of solution reactions.

F. Possible Supercooling of Protein Crystals Although there are numerous observations that suggest that the freezing point of the mother liquor is the critical temperature below which protein crystals cannot be safely cooled, there is no reason to expect that a normal ice phase is formed within the crystal lattice; protein crystals present an open structure in which a large part of the volume is occupied by water and they are composed of channels that are spacious enough to accommodate solutes of low molecular weight. However, the extremely small diameter of such channels suggests that the water in the interstices might be quite different from bulk water and might have quite different freezing properties. One of us assumed earlier (Petsko, 1975)that careful removal of bulk liquid from around the crystal might

318

PIERRE DOUZOU AND GREGORY A. PETSKO

enable some protein crystals to be supercooled without altering the mother liquor and then investigated at very low temperatures. Indeed, supercooling of water is a fascinating and useful property and recent experimental data gathered in Petsko's laboratory about the conditions to obtain such a state might help in the design of accurate experiments to test the above assumption. While it has been known for years that water droplets in the micrometer size range can supercool down to -40°C (Fletcher, 1962; Rasmusse et al., 1973), very few attempts have been carried out on water droplets in the nanometer range, which are obtained with micromicellar solutions of water in a number of nonpolar solvents of very low freezing point. Such solutions are homogeneous and of low viscosity; they can remain perfectly colorless and therefore optically transparent at very low temperature (2-60°C) and can be used as media to investigate enzymecatalyzed reactions. The structure of water droplets (often termed water pools) entrapped in spherical or near-spherical associations of amphiphilic surfactants is quite different from that of bulk water: their polarity, microviscosity, and behavior as a function of temperature reflect the uniqueness of such media. Supercooling is to be expected in emulsions in which the water phase is highly dispersed into spherical droplets whose size distribution is very narrow, the mean diameter being 3-5 pm depending on emulsion procedure. Calorimetric determinations as well as fluorescence can be used to investigate supercooling and freezing. We have used 1,&anilinonaphthalene sulfonate (ANS) as a fluorescent probe in studies about supercooling and freezing of water in oil emulsions. This compound displays a strong affinity for water droplets and its properties, e.g., quantum yield, lifetime, and position of the fluorescence maximum, are extremely sensitive to the polarity of the microenvironment. This behavior may be used to test the effective polarity of water droplets and their supercooling and freezing. The very low fluorescence quantum yield recorded in liquid solutions changes dramatically upon freezing, and fluorescence recordings so obtained resemble the differential scanning calorimetry (DSC) thermogram. The slope of the fluorescence enhancement curve at subzero temperatures gives a clear indication of the homogeneity of the emulsion, as well as a good estimate of the supercooling range. The inflection at -38.8"C is close to what is usually regarded to be the homogeneous nucleation temperature of water, i.e., the temperature at which water freezes spontaneously in the absence of impurities which normally act as nuclei for freezing. Recordings of fluorescence intensity as a function of time at selected subzero temperatures permit a check on the stability of the

PROTEINS AT WORK

319

supercooled state. Data collected during numerous trials clearly show that with the techniques employed it is impossible to predict a freezing point for any given emulsion. For these reasons, tests should be carried out with each emulsion system prior to using it in a kinetic run. Such fluorimetric recordings are easy to perform on very small aliquots (200300 ~ 1 of) the emulsions. It has been established that the results are in good agreement with those obtained by DSC, permitting determination of the range of temperature and time over which supercooling occurs. Similar investigations have been carried out on water in oil “microemulsions.” A microemulsion is a clear, transparent, and stable system consisting of essentially monodisperse oil in water (O/W) or water in oil (W/O) droplets with diameters generally in the range of 10-200 nm. Microemulsions are transparent because of their small particle size, they are spherical aggregates of oil or water dispersed in the other liquid, and they are stabilized by an interfacial film of one or more surfactants. Micelle formation in nonaqueous solution (polar solvents), although recognized for some time, has been investigated systematically only in recent years both experimentally and theoretically. Amphiphilic surfactants, characterized by possessing in the same molecule a hydrophilic group which tends to be water soluble and hydrocarbon insoluble, and a lipophilic group which tends to be hydrocarbon soluble and water insoluble, are miscible with both water and hydrocarbons. Suitable amphiphilic surfactants dissolved in a hydrocarbon are able to solubilize water which, in the absence of these compounds, is insoluble in the hydrocarbon. In such systems, water and hydrocarbon can be regarded as solubilized with amphiphile, which acts as cosolvent. The polar groups on the amphiphile aggregate and form an aqueous core in the presence of water. Such micelles are termed inverse, reverse, or inverted micelles. It appears from a survey of the literature that the essential properties of micelles in nonpolar solvents are understood, namely their stability and variations of size, the dissociation behavior, and their solubilizing capacities. Reverse micelles can dissolve relatively large amounts of water (1-10% w/v depending on emulsion formula) as well as polar solutes and, of course, water-soluble compounds. Consequently, they can be used as media for a number of reactions, including enzyme-catalyzed reactions. Very few attempts to investigate such reverse micelles at subzero temperatures are known, in spite of the fact that hydrocarbon solutions present very low freezing points. Due to the average micellar dimensions obtained (10-200 nm), trapped water should supercool at subzero temperatures in the transparent region, as did water droplets of much larger size obtained with insoluble surfactants. Investigations with the ANS fluorescent dye did

320

PIERRE DOUZOU AND GREGORY A. PETSKO

not show any clear-cut changes down to -40 to -45”C, temperatures at which such changes occurred in W/O emulsion with insoluble surfactants. Thus, we did not record any sign of freezing in the temperature range investigated. The microviscosity of water entrapped in such micelles is much higher than that of normal “bulk” water and increases markedly as temperature is dropped. Such behavior might be indicative of the difficulty that substrates or ligands could encounter in diffusing into protein crystals at subzero temperatures. On the other hand, our belief is that, due to the probable supercooling, careful removal of bulk liquid from around a crystal will enable crystallographic investigation at very low temperatures without altering the mother liquor. This hope has now been realized with the demonstration that crystals of sperm whale metmyoglobin, which are grown from nearly saturated ammonium sulfate, can be “flash-frozen”in liquid propane and then directly cooled to liquid nitrogen temperature without disorder, provided the external liquid is removed (Hartmann et al., 1982). Of course, such supercooled protein crystals cannot be used for substrate binding studies, because diffusion of small molecules into the lattice requires external mother liquor. However, these crystals are ideal for studies of protein dynamics by temperature-dependent X-ray diffraction (see Section V,E) and have been used successfully for that purpose. G. Strategies for Trapping Crystalline Enzyme-Substrate Complexes 1. Preliminary Experiments To summarize, the results of a wide variety of solution experiments have proven that temporal resolution of individual steps in an enzymatic reaction is possible, and that some protein crystals can be stabilized at sufficiently low temperatures for this resolution to be achieved over the time required to collect high-resolution X-ray diffraction data (i-e.,several days). Before such experiments are carried out on any new systems, it is essential to perform solution experiments analogous to those described above. Failure to do so exposes the crystallographer to the risk of many artifacts, including those arising from changes in the catalytic mechanism of the enzyme! Fink and Petsko (1981) discuss this point in detail. Proper preliminary experiments define the best solvent system for the desired structural studies; they also delineate the exact conditions of temperature and proton activity required to stabilize the various intermediates in the reaction for sufficient time periods. In the absence of these data, crystallographic studies may be done under conditions where turnover leads to a mixture of bound species in the crystal. High-

PROTEINS AT WORK

32 1

resolution structure determination demands the presence of one major species. Preliminary experiments in solution are necessary to achieve this goal. 2. The Problem of Dqfwion Protein crystals have channels which are often comparable in size to the substrate which is to be diffused into them. Under these conditions, diffusion times may be extremely long, on the order of hours or days. Fink and Petsko (1981) have provided a graphical treatment of this problem which may be used to analyze any experimental system. Conceptually, such analysis is essential because very long diffusion times lead to special experimental problems. If turnover is not zero, product formation will begin as soon as the substrate binds to enzyme molecules on the surface of the crystal. If diffusion times are comparable to turnover rate at low temperature, product will replace substrate in a layer near the surface before the entire crystal can be saturated with substrate. One possible way out of this difficulty is to continuously flow substrate over the crystal during data collection (such continuous flow is recommended for all cryocrystallography experiments for this very reason) since fresh substrate will displace product and the average structure determined will still be the desired one. However, if product binds more tightly than substrate, even continuous flow will not prevent the EP complex from accumulating in the crystal during data collection. The lesson from this discussion is that some large substrates cannot be studied by diffusing them into the crystal at low temperatures, and that protein crystals in which the active site faces a very narrow solvent channel are not suitable for temporal resolution experiments. Whenever a new crystalline protein is to be investigated by low-temperature crystallography, the size of the active-site-adjacent channel should be determined, and the thickness of the crystal used for diffusion experiments should be chosen according to the analysis given in Fink and Petsko (1981). Failure to carry out these preliminary evaluations can lead to disaster [see “The Story of SAP’ in Fink and Petsko (1981) for an example].

IV. DEVICES

A. Low-Temperature Equipment for X-Ray Diffraction

There are three methods of collecting high-resolution X-ray diffraction data: diffractometry, photographically, and by electronic area detector. Each method has advantages and disadvantages for a particular crystalline protein, but for very accurate data acquisition beyond 2

322

PIERRE DOUZOU AND GREGORY A. PETSKO

resolution most crystallographers will use diffractometer or area detector. However, the type of low-temperature device most simply employed will depend on the particular data collection hardware, as seen below. No matter how the data are collected, the overwhelming favorite as far as type of apparatus is concerned is the gas-stream apparatus. Protein crystals, even at low temperatures, are normally kept in an atmosphere of high relative humidity. This is accomplished by sealing the crystal inside a thin-walled quartz or glass capillary tube. If substrate bonding studies are to be performed, the tube may be anchored to a brass yoke or pin and have polyethylene tubing protruding from the top and bottom (see Section IV,B). Any device which cools the protein sample must allow for the presence of the tube or flow cell, cause minimal loss of data (i.e., by restricting the number of reflections that can be measured due to interference or by reducing the intensities of the incident and diffracted beams), and be both easy and economical to use. The gas-stream apparatus meets all of these criteria, and is the simplest to manufacture and maintain. In such a system, a stream of cold gas (usually dry nitrogen or air), generated by boiling a liquified gas or passing warm dried gas through a heat exchanger in a cold bath, is directed over the sample. If boiling is used, relatively large quantities of cryogen are consumed per hour. This disadvantage is largely offset by recently developed mechanically refrigerated Dewars and probes which can be used to cool a permanent cold bath for heat exchange. The problem of frost formation is overcome by one of a number of approaches: the outer edge of the cold stream can be heated by a ring around the tip of the delivery nozzle thus preventing mixing of the warm moist room air and the cold dry air of the stream; a concentric, dry warm stream can be produced by a second supply of uncooled gas; or a cryostat may be used to enclose the specimen. The technology of lowtemperature X-ray diffraction has given rise to a bewildering array of gadgets, but there is now a useful guide for the inexperienced investigator (Rudman, 1976). Photographic methods of data collection and, to nearly the same extent, area detectors require simple delivery systems and nozzle design because the stage on which the goniometer head is mounted moves through very small angles during reflection acquisition. A simple flexible delivery tube of silicone rubber, insulated with Armaflex or some other foam compound and placed so as to blow coaxially over the capillary tube, is usually sufficient. The large complex angular motions produced by the omega and chi circles of a diffractometer demand a more sophisticated apparatus if the gas stream is to remain coaxial with the capillary throughout data collection (and such a requirement is necessary to pre-

PROTEINS AT WORK

323

vent large temperature gradients along the tube). Two popular solutions are to bring the cold stream up through the base of the goniometer head or to provide flexiblejoints in the delivery stream which allow the nozzle to track the movement of the omega and chi circles. Rudman (1976) has given complete descriptions of these and other pieces of equipment. For routine work at around 0°C on any form of data collection device, a simple low-temperature attachment may be constructed from a copper coil, a refrigerated ethanol bath, a heatless air drier, a compressor, and a flexible hose (Marsh and Petsko, 1973). Such an apparatus costs less than $2000 to build and has essentially no operating costs. It fulfills the most important criterion for a successful low-temperature device: that it be so easy to operate as to encourage its use. Its disadvantage is that very low temperatures are not attainable without more sophisticated-and expensive- hardware. The authors’ personal choice for routine high-resolution data collection at temperatures ranging from 0 to - 100°C is a four-circle diffractometer equipped with a Nicolet LT-1 low-temperature device. The LT1 has been modified to operate with a heat exchange bath consisting of 40 liters of ethanol cooled to -110°C by a Neslab CClOOF Cryocool refrigerated coolong probe. Dry nitrogen gas is provided by the normal boil-off from large tanks of liquid nitrogen. Delivery of the cooled gas is through an evacuated Dewar line of silvered glass, with Teflon ball-andsocket joints that allow for tracking of the movement of the crystal. Ice formation is retarded (but not prevented at below -50°C during humid weather) by a heating mantle around the edge of the nozzle, which warms the periphery of the cold stream. A diffractometer is chosen as the data collection method because of the high precision and accuracy of the data that can be measured by this technique, and because the high background produced by flow cell materials is less serious for this instrument than for photographic recording of diffracted intensities. We are exploring the utility of an area detector, which combines the signal-tonoise ratio of a diffractometer with the multiple-reflection recording power (and consequent speed of data collection) of photographic film. It seems quite likely to us that area detectors will be the preferred technology in the near future. Cryostats have been much used in small-molecule crystallography because they allow rapid cooling to very low temperatures with minimal cold gas consumption and offer great advantages in the area of frost prevention. Unfortunately, most designs employ beryllium shrouds or other nontransparent material, and are of a size which does not lend itself to crystals mounted in capillary tubes and flow cells. A recent advance is the description of a Mylar cryostat specifically designed for

324

PIERRE DOUZOU AND GREGORY A. PETSKO

low-temperature protein crystallography with provision for flow cell work as well (Walter et al., 1982). Such a device is not needed for work down to about -2O"C, but represents a great boon to experiments which require colder temperatures. It should be widely copied.

B . The Flow Cell The problem of introducing a substrate, inhibitor, or other small molecule into the lattice of a protein crystal which is mounted for data collection was solved almost 20 years ago by Richards, Wyckoff, Tsernoglou, and colleagues (Wyckoff et al., 1967). They described the design and construction of a flow cell which immobilized the crystal despite continuous flow of mother liquor at a rate of up to several milliliters per day. A brief description of the flow cell may be given: a quartz capillary tube is attached to a brass yoke by epoxy cement, and a tightfitting polyethylene tube is cemented into the bottom of the capillary. The tip of the polyethylene tubing serves as a platform on which a bed of pipe cleaner fibers, 1-2 mm thick, is placed. The capillary is then filled with mother liquor and the protein crystal is introduced at the top and allowed to sink onto the bed of fibers. Additional fibers are gently packed down on top of the crystal to immobilize it, and the top of the capillary is then cut flush with the top of the yoke. Another polyethylene tube is placed into the capillary at the freshly cut end and epoxy cement is used to seal thisjoint. The result is a cell containing, in order, polyethylene tube, fibers, crystal, fibers, polyethylene tube. The free ends of the polyethylene tubing may be connected to reservoirs of mother liquor; the direction and rate of flow of liquid past the crystal are determined by adjusting the relative heights of the reservoirs. In such a cell the crystal remains fixed in position and may be lined up for data collection in the usual way. Substrate may then be introduced into the source reservoir and allowed to reach the crystal at any desired rate: we usually prefer a flow of 0.5 muday. If a sufficiently sensitive assay is available, turnover may be checked by assaying for the presence of product in the sink reservoir. Binding may be followed by periodic monitoring of the intensities of reflections which are sensitive to the presence of electron density in the active site; the reflections of choice may be determined by previous studies with inhibitors, or by calculation from modeling. Alber et al. (1976) give an example for substrate binding to crystalline elastase at -55°C (Fig. 17). Flow cell work is recommended for all mother liquor replacement experiments once crude studies have eliminated gross destruction of the crystal. The ability to simultaneously monitor the diffraction pattern and add alcohol greatly reduces the number of dead-end experiments. Com-

PROTEINS AT WORK

325

26.5

25.0

8 .-

23.5

-E X

P

--Y L

22.0

-e r ! l

20.5

B .-3

-EE

19.0

17.5

16.0

I

I

3

6

9

,

I

I

I

12

15

18

21

I

24

Time (hours)

FIG.17. Plots of the changes in intensity with time of two high-resolution reflections from an elastase crystal as substrate is added in a flow cell. The solid curves are calculated assuming a first-order rate constant of 5.4 x lo+ sec-1.

bining flow cell technology with a low-temperature device on the diffractometer, camera, or area detector allows the entire study, from transfer of the crystal to a cryoprotective solvent by simultaneous cooling and alcohol addition through substrate binding to data collection, to be carried out on a mounted specimen with continuous observation of diffraction quality. The one major difficulty with use of a flow cell is that the combination of pipe cleaner fibers and a large volume of mother liquor around the crystal greatly increases the background scatter of X-rays, particularly in the 3 8, resolution region. For diffractometric data collection-and, probably, for area detector work as well-this problem is not serious, but

326

PIERRE DOUZOU AND GREGORY A. PETSKO

for photographic data collection, where background measurements are difficult under good conditions and where the fog level of the film presents an inherent background problem of its own for weak reflections, considerable difficulties may arise. We are not aware of any detailed flow cell study in which film data collection was used. It is well to mention, however, that there is an advantage to flow cell work which partially compensates for the problem of elevated background: the cylindrical distribution of liquid around the crystal virtually eliminates the crystal-shape-dependent absorption of X-rays, which is difficult to correct for in a normal capillary mount, and causes all reflections to suffer essentially the same absorption. C. Equipment for Low-Temperature Spectroscopy Although this article concentrates on the technology which permits application of cryoenzymology to crystallography, we have also tried to stress the importance of adequate preliminary experiments in solution. Such experiments usually rely on spectroscopic methods for the detection of turnover and identification of intermediates. We believe that these preliminary experiments must be done in the appropriate cryosolvent at low temperature; it is not sufficient to extrapolate from studies at room temperature. The obvious reason for this caveat is the possible perturbing influence of the organic solvent and/or the subzero temperature, but there are other reasons as well. A steady-state kinetics experiment at room temperature essentially yields only two parameters, k,,, and K,, both of which are usually composite functions of several microscopic rate constants. In addition, under steady-state conditions, many intermediates are present at very low concentrations and have lifetimes of milliseconds or less, which make their study difficult. Pre-steady-state kinetics experiments, which involve rapid-mixing methods, can provide information about the elementary steps in the reaction, but even with these techniques intermediates may not be detected, either because the rate-limiting step of the reaction precedes one or more intermediates prior to the terminal, more stable complex, or because the lifetime of the intermediates may be too short for the detection methods employed. Relaxation techniques, such as temperature jump, suffer from some of the same limitations despite their inherent ability to detect species with very short lifetimes; relaxation techniques also require the system to be at (pseudo) equilibrium and to have reporter groups of adequate sensitivity and fortunate location. The purpose of this discussion is to point out that, even when crystallography cannot be applied to a particular enzymatic reaction, low-tem-

PROTEINS A T WORK

327

perature kinetics and spectroscopic studies are a valuable tool for dissecting the elementary steps of the catalytic pathway. They provide a series of “stop-action” pictures as well, but with much lower resolution than attainable by X-ray diffraction. The instrumentation needed for solution studies of enzyme structure and catalysis at subzero temperatures is not difficult to build and operate. Detailed discussion of the technology is beyond the scope of this article, but a number of recent treatments provide all necessary information JFink and Geeves, 1979; DOUZOU, 1974, 1977a,b).

D. Solving and Refining Low-Temperature Protein Structures Ideally, X-ray cryoenzymology involves the binding of substrates to existing native protein crystals at low temperatures. Unless there is a massive conformational change in the enzyme on binding or formation of the desired intermediate (Section VI,B), the structure of the enzymesubstrate complex is expected to be isomorphous with that of the native enzyme. Under these conditions, it is not necessary to determine the phases of the scattered structure factors from the crystal of the complex & novo. One can make the simplifying assumption that the phase angles for the native structure-which presumably have already been solved and refined at high resolution-are applicable to the reflections from the crystal of the complex. A difference electron density map may then be calculated using as amplitudes the difference in structure amplitude between the complex and the native enzyme, with native phase angles. This map will show positive features wherever electron density has been added to places devoid of it in the native structure, and negative features wherever electron density no longer occupies a position it did in the uncomplexed protein. Interpretation of such a map is usually straightforward, particularly with modern computer graphics systems for displaying the map and building structural models into it. High-resolution data greatly improve the interpretability. Once the substrate has been fitted to its new electron density in the difference map, and any large conformational changes in the enzyme have been ascertained and the protein coordinates adjusted accordingly, it is desirable to subject the structure of the complex to least-squares refinement. Refinement improves the fit of model to density, adjusts the structure to allow for any small conformational changes that were not built in on the graphics system, and corrects the entire structure for changes in atomic mobility. We have used the restrained least-squares refinement method of Konnert and Hendrickson (1980) with good success; other refinement programs are also generally available.

328

PIERRE DOUZOU AND GREGORY A. PETSKO

V. RESULTS A. Radiation Damage and Crystal Lifetime

Precise determination of bond lengths and angles in a refined enzyme-substrate complex crystal structure requires the collection of veryhigh-resolution data, to 1.5 8, or beyond if such data exist. Unfortunately, the number of reflections that must be collected increases as the reciprocal of the cube of the resolution. Over 2.3 times as many measurements must be made for a l .5 %, resolution structure determination as for a 2-8,structure. In addition to the problem of keeping the enzyme-substrate complex stable for an increased period of time, the crystallographer faces the problem of increased exposure of the crystal to the X-ray beam. Protein crystals suffer time-dependent radiation damage on irradiation, leading to eventual loss of crystalline order and diffracting power. For some proteins this effect is so severe at room temperature that the useful lifetime of the specimen in the X-ray beam is less than 1 day. High-resolution data collection normally requires at least 1 week of exposure. Area detectors hold out the promise that this time may be reduced to 1-2 days, but even under these conditions some proteins decay too rapidly for a complete set of measurements to be made with one crystal. Of course, multiple crystals could be used and the data from them scaled together; indeed, this is common practice in native structure determination. However, given the difficulties in crystal transfer, flow cell mounting, and all of the other problems inherent in a low-temperature crystallographic study of an enzymatic reaction, it is clearly preferable to obtain all of the data for a given complex from one crystal (this will also eliminate the problem of scaling together data sets from crystals of possibly differing substrate occupancy). Given this requirement, it is pleasant to observe that low temperatures greatly increase the lifetime of most protein crystals in the X-ray beam. A good example is the iron-containing superoxide dismutase from Pseudomom ovalis. Crystals of this enzyme will not survive more than 24 hours exposure at room temperature without experiencing at least 50% loss in diffracted intensity, but at - 15°C a complete set of data to 2.9 A resolution, over 10,000reflections, may be collected over a 5-day period with only 15%decay (Ringe et al., 1983).The temperatures required for cryoenzymological work prolong useful crystal lifetime even further: at -50°C we have been able to obtain over 50,000 measurements from one elastase crystal with less than 10% loss in diffracting power. For this reason alone it is standard procedure in our laboratory to collect all native data at -10°C. We have seldom used more than one crystal per data set under these conditions. An additional factor to note is that the

PROTEINS AT WORK

329

stability of protein crystals is often enhanced by maintenance of a constant temperature, regardless of its magnitude. Employment of a lowtemperature device eliminates fluctuations in temperature at the crystal due to changes in the room environment.

B . Resolution, DLorder, and Related Problems The most common question asked by people who have not used very low temperatures for collection of X-ray data is “Can the highest resolution observable for a given crystalline protein be improved by cooling to very low temperatures?” The answer depends on the particular protein crystal in question. Radiation damage effects the high-resolution reflections at a more rapid rate than the low-resolution ones, and so reducing the rate of crystal decay by cooling will improve one’s ability to measure these reflections. However, the question is probably addressed more to the intrinsic order in a crystal. Cooling below the freezing point of the mother liquor without great care will certainly decrease the order in the specimen due to damage caused by ice formation, but careful cooling without freezing, or supercooling, or “flash-freezing”under the right conditions, will often improve the scattering power of the crystal. Effects we have observed include a decrease in the peak width of the reflections, possibly due to “annealing” of the mosaic blocks in the specimen, and increase in the intensity of high-resolution reflections relative to those at low resolution. The latter effect is almost certainly due to the reduction in thermal motion of the atoms of the protein (see Section V,E). Under the right conditions, the improvement in resolution can be dramatic. Ribonuclease A gives monoclinic crystals which diffract to beyond 1.2 i% resolution at room temperature, but at -100°C reflections can be observed beyond 0.8 i% resolution! Such behavior is commonly observed for small molecule crystals, where freezing is not a problem because no bulk liquid exists either within the lattice or around the crystal. Unfortunately, the magnitude of the improvement possible for a protein crystal will depend on the extent to which the fall-off in diffracting power with resolution is due to simple thermal vibrations. If these motions dominate the temperature factor for a given crystal, then a marked improvement with reduction in temperature is likely. But if the weakness of highresolution data arises from either static disorder or a large-scale conformational disorder, then cooling may yield only a slight improvement in the diffraction pattern, or none at all. C . Productive Intermediates A detailed understanding of the mechanism of enzyme catalysis and of the reasons for enzyme catalytic efficiency requires knowledge of the

330

PIERRE DOUZOU AND GREGORY A. PETSKO

structures, kinetics, and thermodynamics of all the intermediates and transition states along the productive catalytic pathway (Fink and Petsko, 1981). As we have discussed, efforts in the past have been handicapped by the short lifetimes of enzyme-substrate complexes and intermediates, as well as those of the transition states. In this article we have shown how it is possible to circumvent these problems to a great extent by the use of low temperatures to stabilize the enzyme-substrate complex. It is further possible, by the use of a transition state analog, to mimic the transition state structure of the substrate so as to allow the structural elucidation of all key species along the reaction path. Thus far, two enzymatic reactions have been studied by this procedure, although additional studies are in progress in several laboratories. The first successful application of the principles set forth in this article to a crystalline enzyme-intermediate complex was the demonstration of the stabilization of an acyl-enzyme intermediate in the hydrolysis of ester substrates by porcine pancreatic elastase (Alber et al., 1976). Unfortunately, the size of the data collection problem presented by this system combined with the relatively short lifetime of the intermediate, even at low temperature, limited the elastase study to 3.5 A resolution. Ribonuclease A, a smaller enzyme with excellent diffracting power and long-lived intermediates, proved a more tractable system for high-resolution studies, and recently the complete three-dimensional structures of every kinetically significant step in the catalytic reaction of this enzyme have been determined (Fink et al., 1984). 1. Elastase

The presence of a covalent acyl-enzyme intermediate in the catalytic reaction of the serine proteases made this class of enzymes an attractive candidate for the initial attempt at using subzero temperatures to study an enzymatic mechanism. Elastase was chosen because it is easy to crystallize, diffracts to high resolution, has an active site which is accessible to small molecules diffusing through the crystal lattice, and is stable in high concentrations of cryoprotective solvents, The strategy used in the elastase experiment was to first determine in solution the exact conditions of temperature, organic solvent, and proton activity needed to stabilize an acyl-enzyme intermediate for sufficient time for X-ray data collection, and then to prepare the complex in the preformed, cooled crystal. Solution studies were carried out in the laboratory of Professor A. L. Fink, and were summarized in Section 1I,A,3. Briefly, it was shown that the chromophoric substrate N-carbobenzoxy-L-alanyl-p-nitrophenylester would react with elastase in both solution and in crystals in 70 :30 methanol-water at pH* 5.2 to form a productive covalent complex. These

PROTEINS AT WORK

33 1

conditions were used for the X-ray experiment, since the lifetime of the complex at -55°C was demonstrated to be at least 1 week (Fink and Ahmed, 1976). For the crystallographic study, it was first established that elastase crystals, which were grown from very-low-salt sodium sulfate solutions, could be transferred to mother liquor high in methanol concentration. Successful transfer was accomplished by coupling the progressive addition of methanol to a progressive lowering of the temperature. This transfer was carried out on the diffractometer by means of a flow cell, using a flow rate of 0.5 myday. The entire equilibration process required about 6 hours. Once the crystal had been stabilized in 70% methanol at -55"C, a complete set of native intensity data was measured to 3.5 %, resolution. Then the p-nitrophenyl ester substrate was added to the mother liquor at a concentration in excess of its K , and allowed to flow over the crystal for many hours. During this time, several selected X-ray reflections were remeasured every hour. Selection of the reflections was made on the basis of their demonstrated sensitivity to the presence of inhibitors in the active site of the enzyme; they thus served as monitors of the binding of substrate in the specificity pocket. The change of intensity of these reflections was monotonic with time and could be interpreted in terms of a first-order process with an overall rate constant of 5.4 X sec-', in excellent agreement with the acylation rate constant measured for crystalline elastase by low-temperature spectroscopy (Fink and Ahmed, 1976). After 24 hours of substrate flow, the intensity changes had ceased, indicating saturation of the crystal with substrate (Fig. 17). At this time, a second complete set of 3.5 %, data was collected. To prove that any complex which formed at the low temperature was both productive and covalent, two additional experiments were carried out. First, an attempt was made to wash the substrate out of the enzyme at low temperature. The crystal was held at -55°C and substrate-free 70% methanol was flowed over it for 4 days. There was no change in the substrate-sensitivereflections, which were monitored every 8 hours during this period, and when another data set was collected at the end of the wash, it revealed the substrate still bound in the active site. However, when the crystal was allowed to warm up to - 10°C, the monitor reflections immediately began to change in intensity, back to the values they had for the native enzyme. In less than 20 hours all of them had returned to these values, and a final set of data was collected; as expected, on processing it showed an empty active site and a native elastase structure. These two control experiments indicated that the structure that formed when elastase was exposed to the ester substrate was covalent, and that the covalent intermediate would undergo hydrolysis (presum-

332

PIERRE DOUZOU AND GREGORY A. PETSKO

ably to normal product, although no analysis was done) at elevated temperature (Alber et al., 1976). After these careful checks throughout the experiment, it is not surprising that the structure of the complex, determined by difference Fourier synthesis at 3.5 A resolution, was in accord with that expected for a covalent acyl-enzyme intermediate (Fig. 18).There was a large flat feature of electron density in the active site pocket suggestive of the carbobenzoxy group, and no indication of any portion of the p-nitropheno1 moiety. The alanyl residue occupied the specificity pocket in the active site, and was close enough to the electron density of the catalytic Ser-195 for a covalent interaction. In short, the structure was consistent with formation of an acyl-enzyme on hydrolysis of the p-nitrophenol group, further hydrolysis being prevented by the low temperature. This study was the first example of the use of subzero-temperature crystallography to provide a “stop-action’’picture of an enzyme-substrate intermediate along the productive catalytic pathway. Unfortunately, the size of the crystallographic problem presented by elastase coupled with the relatively short lifetime of the acyl-enzyme indicated that higher resolution X-ray data would be difficult to obtain without use of much lower temperatures or multidetector techniques to increase the rate of data acquisition. However, it was observed that the acyl-enzyme stability was a consequence of the known kinetic parameters for elastase action on ester substrates. Hydrolysis of esters by the enzyme involves both the formation and breakdown of the covalent intermediate, and even in alcohol-water mixtures at subzero temperatures the rate-limiting step is deacylation. It is this step which is most seriously affected by temperature, allowing the acyl-enzyme to accumulate relatively rapidly at -55°C but to break down very slowly. Amide substrates display different kinetic behavior: the slow step is acylation itself. It was predicted that use of a p-nitrophenyl amid6 substrate would give the structure of the pre-acyl-enzyme Michaelis complex or even the putative tetrahedral intermediate (Alber et al., 1976),but this experiment has not yet been carried out. Instead, over the following 7 years, attention shifted to the smaller enzyme bovine pancreatic ribonuclease A. 2. Ribonuclease A

Using the principles outlined in this article, the crystal structures of the following complexes of RNase A have been determined: the free enzyme, both with and without a sulfate ion in the active site, the enzyme-dinucleotide complex, the enzyme-cyclic phosphate intermediate complex, the enzyme-transition state complex, and the enzyme-product complex, all at or near atomic resolution. This structural informa-

333

PROTEINS AT WORK

v'

v

'.?.-

s

0

0

31

u

FIG. 18. The active site region of the electron density difference map between N carbobenzoxy-L-alanine-elastase at -55OC and native elastase at the same temperature. The resolution is 3.5 A. The bilobed feature is consistent with the binding of the alanyl portion of the substrate to the oxygen of the catalytic serine, with weak interaction of the carbobenzoxy group to the surface of the enzyme.

334

PIERRE DOUZOU AND GREGORY A. PETSKO

tion, in conjunction with low-temperature kinetics and other data, permits the formulation of a detailed chemical and structural description of the events in ribonuclease catalysis and the understanding of the structural basis for the catalytic power of the enzyme. In recent years attention has focused on the role of intrinsic binding energy and entropic factors as major contributors to enzyme catalytic efficiency (Page and Jencks, 1971;Jencks, 1975, 1981). The ribonuclease mechanism conforms to expectations based on these ideas. In particular, distortion occurs to raise the ground state of the substrate in the ES complex, and the bound substrate interacts with the enzyme in a manner such that the enzyme becomes complementary to the transition state of the reaction during the catalytic cycle. The crystal structure of RNase A has now been refined at 1.5 %i resolution at temperatures ranging from +22 to -32°C (Campbelland Petsko, 1983; Gilbert et al., 1983). The molecule is kidney shaped with a deep active site cleft that, in the enzyme as purified and crystallized, contains a sulfate ion which acts as a competitive inhibitor. Nucleotide inhibitors bind in the cleft with the pyrimidine ring buried in a deep pocket, the ribose making few interactions with the protein, and the phosphate lying in a groove on the surface of the enzyme in proximity to many positively charged side-chains, including His- 12 and His- 119 and Lys-4 1 and Lys7. The active site in the sulfate-free enzyme contains these residues, with a hydrogen-bonded network of bound water molecules occupying the space where the inhibitors bind. In addition to these charged amino acids, the active site cleft contains the side-chains of Thr-45 and Asp121. These six residues have formed the focus of most of the discussion of the structural basis of RNase catalysis. Despite considerable biochemical work, high-resolution crystal structure determination of native RNase A and S, and some medium-resolution studies of RNase A-inhibitor complexes, a number of questions existed concerning the details of the catalytic mechanism and the role of specific amino acids. Study of the low-temperature kinetics and threedimensional structures of the significant steps of the ribonuclease reaction was designed to address the following questions. 1. Is the favored “in-line” mechanism suggested by the geometry of the enzyme and the substrate in both stages of the reaction? 2. Are both His-12 and His-1 19 involved in both stages of the reaction? 3. What is the role of Lys41? 4. What movements occur in the enzyme and the substrate, especially the phosphate group, during the reaction?

PROTEINS AT WORK

335

5. Is there any ground-state destabilization of the substrate by bondangle strain or binding in a high-energy conformation? 6. Is the transition state stabilized more than the ground states due to extra bonding from the enzyme? 7. Is Lys-7 involved in the reaction? The experimental objective of the study was to obtain a series of “stopaction” photographs of ribonuclease A at work at atomic resolution. The strategy for such a program has been considered in detail by Fink and Petsko (1981), who treat such subjects as diffusional constraints and turnover rates, and in the preceding sections of this article. The ribonuclease reaction has a series of well-characterized, stable species which can be purchased, and crystals of the enzyme are large, well ordered, catalytically active (Fink et al., 1984), and have as their natural mother liquor a cryoprotective solvent (Petsko, 1975). RNase thus represents the ideal system for a step-by-step analysis of an enzymatic catalytic pathway by the methods outlined above. The experimental details for the preparation and analysis of the various complexes are too complex for detailed summary; the reader is directed to the individual papers (Gilbert and Petsko, 1984a-d). Here we review the overall philosophy and the specific problems that were met. The first, and essential, step was the determination of the conditions of solvent, pH, and temperature necessary to stabilize the enzymesubstrate complex for the time required to collect high-resolution X-ray diffraction data (7-10 days) (Fink et al., 1984).Discussion of this series of solution experiments was given in Section II,A,5. The section step was the solution and refinement of the structure of the native enzyme at room temperature; fortunately, this had already been accomplished by other workers (Kartha et al., 1967; Carlisle et al., 1974; Wlodawer, 1980; Borkakoti et al., 1982; Wlodawer and Sjolin, 1982). The third step was the demonstration, by crystallographic refinement at high resolution, that the structure of the native enzyme (and particularly, the conformations and relative positions of the residues in the active site) is unchanged on going from room temperature to subzero temperatures (Gilbert et al., 1983). From that point on, each step in the catalytic pathway (Fig. 19) could be stabilized and studied, but each step also presented its own particular problems, as discussed in the following Sections V,C,B,a-e. a. The True Native Enzyme. Good structural data were already available for native RNase A at room temperature, but mechanistic interpretation was complicated by the presence in both structures of a sulfate ion bound at the active site. Sulfate is a strong competitive inhibitor of the

336

PIERRE DOUZOU AND GREGORY A. PETSKO

JI

41

iis

J FIG.19. The in-line mechanism proposed for RNase A by Mathias and Rabin (Findlay ct al., 1962) as modified by Roberts et ~ l(1969). . Only the first stage, formation of the cyclic

intermediate, is shown. The second stage is the reverse of the first, but with R = H.The residues implicated in catalysis by this study are shown. This is the mechanistic proposal most consistent with the structural data summarized in this article.

enzyme and could produce rearrangements of the active site residues relative to their active conformations. This problem was solved by determining the crystal structure at 1.5 A resolution of sulfate-free native ribonuclease A (Campbell and Petsko, 1983). This structure determination was carried out at - 10°C to reduce radiation damage to sufficiently low rates to allow the complete set of high-resolution data to be collected on one crystal. Removal of the sulfate caused no significant changes in

PROTEINS AT WORK

337

the active site geometry of the enzyme. For comparison with low-temperature structures of E-S complexes, the structures of the sulfatecontaining enzyme at - 10 and -32°C were determined (Gilbert et al., 1983). These were found to be identical to that at room temperature. 6. The Enzyme-Dinucleotide ( E S )Complex. Long single-stranded RNA molecules are too large to diffuse into the channels in the crystal, but RNase also hydrolyzes tri- and dinucleotides whose 3’ base is a pyrimidine. Dinucleotides will enter the lattice (Fink et al., 1984),but only very slowly. To avoid turnover before the entire crystal is saturated with substrate (necessary because product binds even more tightly than the substrate), the crystal must be cooled to below - 100°C. Experimentally, this was not possible. However, the absolute requirement for a 2’ -OH on the sugar ring of the substrate makes dideoxynucleotides very good substrate analogs. Cocrystallizationof RNase with d-CpA, the DNA analog of the best dinucleotide substrate, yielded large crystals of a stable enzyme-inhibitor complex (Gilbert and Petsko, 1983a). Data collection on this complex was carried out at - 10°C to allow the complete 1.5 A sphere of data to be obtained with one crystal. c. The Enzyme-Cyclic Phosphate Intermediate (EZ) Complex. Ribonuclease is a very favorable enzyme for the trapping of catalytic intermediates because the longest lived intermediate of the reaction, a 2’,3’-cyclic monophosphate of a pyrimidine nucleoside, is itself a chemically stable compound which is commercially available. This intermediate in the hydrolysis of RNA is also the substrate for the second half of the overall enzymatic reaction. Spectroscopic data reviewed above suggested that, at -70°C at pH* 5.5, turnover of this species by crystalline RNase was negligible, permitting substrate diffusion into the crystal followed by collection of X-ray data (Fink et al., 1984). A native RNase crystal was mounted in a flow cell (Wyckoff et al., 1967) and cooled to -70°C on the diffractometer (Alber et al., 1976). Cytidine 2’,3’-cyclic monophosphate was introduced into the mother liquor at a concentration well in excess of its K, and allowed to flow over the crystal at a rate of 0.5 myday. The intensities of several selected reflections (sensitive to the presence of pyrimidine in the specificity site) were measured repeatedly for 48 hours; after this time, their values showed no further changes, indicating that saturation had been reached. The crystalline enzyme-substrate complex was held at -70°C while a complete set of data to 1.9 A resolution was collected as rapidly as possible (Gilbert and Petsko, 1983b). d. The Enzyme-Transition State Complex. Transition states are, by definition, too unstable to be observed crystallographically, even at very low temperatures. However, the expectation that enzymes promote transition state formation by binding the activated complex more tightly than

338

PIERRE DOUZOU AND GREGORY A. PETSKO

the substrate or product (Pauling, 1946; Wolfenden, 1969; Lienhard, 1973)has led to the proposal that compounds which resemble the transition state in stereochemistry and charge configuration should be extremely tight-binding competitive inhibitors. Such compounds are called transition state analogs (Wolfenden, 1972). For the RNase reaction, the transition state for both the transphosphorylation step and the hydrolysis of the cyclic phosphate intermediate is thought to be a pentacovalent phosphate with trigonal bipyramidal geometry and substantial negative charge delocalization on the equatorial oxygens. Lienhard ( 1973) has pointed out that vanadium tends to form five-coordinate oxygen complexes and suggested that a complex of uridine and vanadate might mimic the transition state for the hydrolysis of uridine 2’,3‘-cyclicphosphate. This complex will form in aqueous solution and is the best competitive inhibitor of RNase A, with a Ki that is several orders of magnitude tighter than the binding constant of either the corresponding cyclic phosphate substrate or the 3’-uridine monophosphate product. The uridine-vanadate complex was used as a transition state analog to gain structural information about the activated complex of the RNase reaction. Diffusion of this complex into RNase crystals resulted in crystals of the enzyme-analog complex which were then subjected to data collection to 1.5 A resolution at - 10°C; again, subzero temperature was used to permit very-high-resolution data collection with only one crystal. This insistence on one crystal per data set eliminates the necessity for scaling together data collected with different crystals where the occupancy of the desired complex may be different. e. The Enzyme-Product ( E P ) Complex. A number of strategies are possible for obtaining crystals of the enzyme-product complex in the case of an enzyme like RNase,A of which the product is also a good competitive inhibitor. One can diffuse product into the crystals and examine the resulting complex, one can cocrystallize the enzyme with the product, or one can add substrate to the enzyme at elevated temperatures and allow the E-P complex to form in the crystals by turnover, followed by subsequent lowering of the temperature to eliminate the possibility of any significant back reaction. To evaluate the catalytic competency of crystalline RNase A, and to ascertain that crystallization did not alter the geometry of the active site or prevent an essential conformational change on product binding, all three experiments were performed for the 3’-uridine monophosphate product of the RNase reaction. The electron density maps of all three E-P complexes were identical at 1.9 8, resolution, and the one formed by cocrystallization was selected for full refinement and analysis at 1.5 8, resolution at - 10°C (Gilbert and Petsko, 1983d). The three-dimensional structures of the active site of RNase A at

PROTEINS AT WORK

339

every kinetically significant step in its reaction are illustrated in Fig. 20. This information, combined with data from low-temperature kinetic studies of the ribonuclease reaction and with information already available from chemical nodification and other earlier work, provides a detailed description of the structural changes in both enzyme and substrate that occur during catalysis. First, the active site of the native, sulfate-free enzyme displays the essential catalytic histidines in the orientation needed for optimum substrate binding and acid-base catalysis. These residues move very little throughout the reaction. Lysine-41 is not in its preferred position to assist the reaction, but rather is dynamically disordered and samples a number of conformations, of which the active one is presumably a member. The enzyme binds its dinucleotide substrate in a strained conformation in which the normally tetrahedral phosphodiester linkage is distorted toward trigonal bipyramidal geometry. The torsion angles of the phosphodiester linkage in the bound substrate are gauche,trans instead of the preferred trans,trans seen in solution. This configuration is in accord with the principle of stereoelectronic control of phosphodiester hydrolysis. The orientation of the two catalytic imidazole residues, His-12 and His-119, is as expected for in-line attack at phosphorous as first proposed by Mathias and Rabin (Findlay et al., 1962) and by Roberts et al. (1969), and later confirmed by Usher and associates. The interactions of the pyrimidine ring with the enzyme explain the specificity of the reaction in terms of steric exclusion of purines and specific hydrogen bonding to the ring from the side-chain hydroxyl and main-chain amide of Thr-45, as first observed by Richards and Wyckoff and associates (Richards and Wyckoff, 1971). The cyclic phosphate intermediate is similarly strained prior to its hydrolysis in the second half of the reaction. In this complex as well, the two histidine residues are geometrically disposed to favor in-line attack at phosphorous. Lysine-4 1 remains disordered in both the enzyme-dinucleotide and enzyme-intermediate complexes. The structure of the complex of ribonuclease with its transition state analog uridine-vanadate has allowed the geometry of the activated complex to be defined and also given information about the role of Lys-4 1 in the reaction. The uridinevanadate complex is a pentacovalent species with distorted trigonal bipyramidal geometry, almost identical in bond lengths and angles to the predictions of Holmes and associates for the structure of the transition state for cyclic phosphate hydrolysis (Holmes, 1976). As the transition state forms, Lys-41 becomes ordered by charge coupling to the developing negative charges on the equatorial oxygens of the pentacoordinate phosphorous. The positive charge on the end of the Lys-41 side chain stabilizes the transition state, and provides strong bonding to this species

1

41

HIS 119

LYS 7 ASP I.2

d HIS 119

LYS 7 ASP

1

FIG.20. Frame-by-frame series of "stop-action'' pictures of the catalytic mechanism of RNase A at atomic resolution. Only the essential active site residues and the substrate (filled bonds) are shown. Frame 1, The native enzyme. The sulfate ion which binds to the active site is shown. Frame 2, The Michaelis E-S complex with the dinucleotide CpA. The 2' oxygen which is deprotonated by His-12 is blackened. Frame 3, The transition state for 340

4

LYS 7 ASP 1

HIS 119

5

THR 45

ASP

HIS 119

L,YS 7 ASP 1

HIS 119

stage 1, the transphosphorylation reaction. The 2' oxygen is blackened, while the leaving group 5' oxygen is striped. The geometry is that of a distorted trigonal bipyramid with the entering and leaving groups axial. Frame 4, The 2',3'-cyclic phosphate intermediate. The water molecule which is about to act as a nucleophile is shown as the isolated black circle. Frame 5 , The second transition state, that for hydrolysis of the cyclic phosphate intermediate. The nucleophile, the activated water, is blackened while the leaving group 2' oxygen, which is being protonated by His-12, is striped. Frame 6, The enzyme-product complex. 34 1

342

PIERRE DOUZOU AND GREGORY A. PETSKO

that is not found in the E-S and E-P complexes, where the side chain is highly mobile. The mobility of Lys-4 1 is essential for efficient catalysis by ribonuclease A. During catalysis, the phosphorous atom of the substrate moves by almost 2 A. The largest difference is observed between the phosphorous position in the initial dinucleotide ES complex and the cyclic phosphate intermediate. This difference would not have been observed without the use of subzero temperatures to stabilize the enzyme-intermediate complex. The movement of the phosphorous atom results in corresponding movements of the phosphate oxygens, which are “tracked” by the two histidines through small single-bond rotations. The structure of the enzyme-product complex is similar to that of the initial enzyme-substrate complex. As expected, Lys-4 1 has become disordered once again. During the entire reaction, Lys-7 remains where it is observed in the native structure, over 6 A from any of the substrate atoms and in no position for any direct contribution to the catalytic mechanism. No large conformational changes occur in the enzyme during catalysis, but many small movements take place. The structural basis for the catalytic power of ribonuclease thus resides in several different features: tight, specific binding of a strained conformation of the substrate, general acid-base catalysis by His-12 and His-1 19, and preferential stabilization of the transition state by ionic interactions with Lys-41. The preceding summary and Fig. 20 present a frame-by-frame account of the pathway for ribonuclease catalysis, based predominantly on knowledge of the structures of the various intermediates and transition states involved. The ability to carry out such a study is dependent on three critical features: (1) crystals of the enzyme which diffract sufficiently well to permit structural resolution to at least 2 A; (2) compatibility of the enzyme, its crystals, and its catalytic kinetic parameters with cryoenzymology so as to permit the accumulation and stabilization of enzyme-substrate complexes and intermediates at subzero temperatures in fluid cryosolvents with crystalline enzyme; and (3) the availability of suitable transition state analogs to mimic the actual transition states which are, of course, inaccessible due to their very short lifetimes. The results from this investigation demonstrate that this approach is feasible and can provide unparalleled information about an enzyme at work. D. Unstable Oxidation States As mentioned above, the normal mother liquor of most protein crystals contains a high enough concentration of salt, polyethylene glycol, or other precipitant to depress the freezing point to below -15°C. Substrate binding studies will nearly always require temperatures lower than

PROTEINS A T WORK

343

this if turnover is to be prevented (hence the necessity for transfer to a cryoprotective solvent), but - 17"C, though not low by cryoenzymological standards, is nonetheless 40°C cooler than room temperature and 55°C lower than physiological temperature. Metastable species are often sufficiently stable at around - 15°C for detailed crystallographic studies without further cooling or recourse to a cryosolvent. For example, Lipscomb carried out studies of the binding of poor substrates to crystals of carboxypeptidase A at 4°C in part because their hydrolysis by the enzyme, already extremely slow at room temperature, was slowed severalfold further at the reduced temperature thus permitting collection of diffraction data (Steitz et al., 1976). Therefore, there is considerable scope for studies of metastable species by low-temperature crystallography near the freezing point of the normal crystal mother liquor. Recently, a number of examples of this approach have appeared that demonstrate its power. They fall into two classes: systems in which the complex is nearly stable enough at room temperature and cooling provides the extra stabilization needed, as in the case of carboxypeptidase, and systems in which X-ray irradiation would destroy the particular complex too quickly at ambient temperature. In all cases, we are dealing with particular oxidation states of a protein involved in redox chemistry.

I , Oxymyoglobins Myoglobin is a small, heme-containing protein involved in the reversible binding of oxygen for storage in muscle and other tissue. As normally purified, the protein is found in the met state, with a high-spin ferric iron atom to which a water molecule is coordinated. Physiologically, the protein shuttles between two ferrous states, one low spin and one high spin. Deoxymyoglobin has five-coordinated iron and can be formed by treatment of the met protein with a suitable reducing agent such as dithionite. Oxymyoglobin is not sufficiently stable at room temperature for crystallographic studies, so until recently structural data only existed for the met and deoxy forms of the protein. Controversy regarding the geometry of the bound oxygen molecule stimulated two studies of the oxy protein which made use of low temperature to stabilize this form. Phillips ( 1978) converted crystalline deoxymyoglobin to the oxy form and then collected high-resolution diffraction data at around -15"C, at which temperature the lifetime of oxymyoglobin is at least several weeks. Petsko et al. (1978) used myoglobin in which a cobalt atom had been substituted for the iron to render the oxy form of the protein stable enough for direct crystallization at 4°C followed by high-resolution data collection at near -15°C. In both studies the oxygen was observed to coordinate to the metal in a bent, end-on fashion, making substantial van der Waals contacts with the heme pocket-lining residues

344

PIERRE DOUZOU AND GREGORY A. PETSKO

Val-68 and His-64, as well as a possible hydrogen bond to the imidazole of the latter residue. The structures of deoxy-and oxymyoglobin represent a set of stop-action pictures of the two principle states of this protein at work. Carbon monoxide also binds to myoglobin to form a stable complex. Spectroscopic studies had suggested more than one orientation for the bound CO, but earlier diffraction studies had been limited to too low a resolution to confirm this prediction. The limitation arose from the conversion of CO-myoglobin to metmyoglobin in the X-ray beam, a redox reaction presumably catalyzed by the absorption of X-rays by the iron atom. Such processes are quite temperature dependent, and Petsko and Wiltz (1984) employed a temperature of around - 15°C to stabilize COmyoglobin for sufficient time to collect a complete set of 1.5 %i resolution diffraction data on a single crystal. Following data collection, the crystal was dissolved and assayed spectrophotometrically. The protein was found to be 90% in the CO form, in contrast to a control crystal irradiated for the same length of time but at room temperature, in which less than 20% CO-myoglobin remained. Analysis of the high-resolution structure of CO-myoglobin shows two alternative positions for the bound carbon monoxide, in excellent agreement with the spectroscopic data. 2. Flawodoxin

Clostridial flavodoxin is a small flavoprotein which can exist in three oxidation states: reduced, semiquinone, and oxidized. The three-dimensional structures of the oxidized and semiquinone forms of the protein were easily determined at high resolution due to the stability of these species, but the reduced structure, of great interest for comparison with the oxidized because of the ability of the protein to modify the normal redox potential of the flavin, was impossible to measure under ordinary conditions. It was possible to prepare a crystal of the reduced protein by mounting the crystal in an inert atmosphere in the presence of dithionite, but X-ray irradiation at room temperature caused rapid conversion to an oxidized form, as was easily observed by change in color of the crystal. To solve this problem, both the low-temperature device and flow cell were used. The crystal was mounted in a flow cell and subjected to continuous flow of deoxygenated, dithionite-containing mother liquor during data collection at - 15°C. No color change was observed while a complete set of 1.9 %i data was obtained. The absorption spectrum of the dissolved crystal showed that at least 80% of the protein molecules were still fully reduced after irradiation. Determination of the structure of reduced flavodoxin showed the cofactor to be bound to the protein in a

PROTEINS A T WORK

345

geometrically strained conformation, consistent with the observed change in redox potential (Smith et al., 1978). The structures of reduced, semiquinone, and oxidized flavodoxin represent a set of stop-action pictures of the redox chemistry of this molecule to atomic resolution. E . Analysis of Protein Flexibility Restrained least-squares refinement of the atomic model of a protein against high-resolution X-ray diffraction permits the fitting of an individual isotropic B-factor (also called temperature factor) for every atom. The B-factor may be viewed in real space as the attempt to fit a Gaussian distribution to the spread of electron density about the average position of that atom. Anything which produces a spreading of electron density will contribute to the B-factor, and consequently to the mean-square displacement of the atom which may be estimated from the B-factor by the Debye-Waller model of thermal motion (Willis and Pryor, 1975). Atomic motion occurring over the course of the collection of the X-ray data, whether that motion is individual or part of a collective mode, will certainly contribute. It is important to emphasize that motions producing electron density whose magnitude is at or below the noise level in the structure determination will not be reflected in the final refined value. Experience suggests that this level is about 25% occupancy for most high-resolution structures. However, thermal-driven dynamic disorder in the atom’s position is not the only possible kind of disorder. The crystal may be heterogeneous with respect to the position of that atom: different molecules in the crystal may have the atom in different places, and these positions may not interconvert because of large potential energy barriers. This is equivalent to saying that the protein has folded into a number of distinct conformations, at least at the site in question. This static conformational disorder will also cause a spreading of the electron density if the different positions are closer together than the resolution of the structure refinement, because the measured X-ray data represent an average over all unit cells in the lattice. If the different positions are far apart, and there are only a few of them so that each one is more than about 25% occupied, it may be possible to discern the discrete conformations and use as the model the entire static distribution. It is important to realize that there may be some temperature at which enough kinetic energy exists to convert this static disorder into a dynamic one without denaturing the protein. The distinction between a static and a dynamic disorder in the conformation of a protein is only meaningful at a particular, stated temperature. Another kind of static disorder which can exist in a protein crystal and which is different from a distribution of discrete conformations is lattice

346

PIERRE DOUZOU AND GREGORY A. PETSKO

disorder. If every molecule in the crystal is not in the same position, relative to the origin of its unit cell, as every other molecule, there will be a spreading of electron density around the average position of every atom which is caused by this imperfection of crystal packing. The magnitude of the lattice disorder will vary from crystal to crystal and from protein to protein; in unfavorable cases it may completely dominate the observed mean-square displacements, which will be misinterpreted as arising from atomic motion. It is therefore desirable to obtain an estimate of the magnitude of the lattice disorder contribution for any protein before using the Debye- Waller model to interpret individual atomic B-factors in terms of the motion of the molecule. To separate the effects of static and dynamic disorder, and to obtain an assessment of the height of the potential barrier that is involved in a particular mean-square displacement (here abbreviated (x2)), it is necessary to find a parameter whose variation is sensitive to these quantities. Temperature is the obvious choice. A static disorder will be temperature independent, whereas a dynamic disorder will have a temperature dependence related to the shape of the potential well in which the atom moves, and to the height of any barriers it must cross (Frauenfelder et al., 1979). Simple harmonic thermal vibration decreases linearly with temperature until the Debye temperature TD; below TD the meansquare displacement due to vibration is temperature independent and has a value characteristic of the zero-point vibrational (x2). The hightemperature portion of a curve of (x2) vs T will therefore extrapolate smoothly to 0 at T = 0 K if the sole or dominant contribution to the measured (x2) is simple harmonic vibration ((x2)”). In such a plot the low-temperature limb is expected to have values of (x2) equal to about 0.01 A* (Willis and Pryor, 1975). Departures from this behavior indicate more complex motion or static disorder. Proteins are expected to undergo motions other than just vibration in a harmonic potential with weak restoring forces. An example of a more complicated motion would be the propellerlike rotation of side-chain methyl groups, or large-scale vibrations of aromatic rings. Any motion with a collective character will have a potential well different from a simple thermal vibration, and will require more energy to occur. There will thus be some temperature of relaxation, T R ,below which the motion will be “frozen-out’’and a static distribution of conformations will exist. In the nomenclature of Frauenfelder et al. (1979), each member of this distribution is called a conformational substate. If the barriers between substates are small, the distribution can still be dynamic even at temperatures as low as 80 K (Hartmann et al., 1982). However, evidence from Mossbauer scattering suggests that the average value of TR is about

PROTEINS AT WORK

347

180 K for myoglobin (Parak et al., 1981).Thus, measurements of (x2) at temperatures below this value should show a much less steep temperature dependence than measurements above, if nonharmonic or collective motions (whose mean-square displacement is denoted (x2),) are a significant component of the total (x2). Figure 21 illustrates the expected behavior of (x2),, (x2),, and their sum for a simple model system in which a small number of substates are separated by relatively large barriers. In practice, the relative contributions of simple harmonic vibrations and collective modes will vary from residue to residue within a given protein.

Conformational substates

FIG.21. (a) The local minima in a hypothetical conformational potential V,, given by the solid line, with three conformational substates. The bottoms of the wells are parameterized by power laws, V,, proportional to x three values of a are shown in dashed lines. (b) The temperature dependence of the vibrational ( ( x 2 ) , ) and collective motion ((x2),) contributions to the observed mean-square displacement ( ( x ~ ) ~from ” ) X-ray diffraction. It is assumed that there is no contribution from lattice disorder. *’O;

348

PIERRE DOUZOU AND GREGORY A. PETSKO

Measurements of (x2) as a function of temperature should also establish whether or not static disorder dominates the apparent dynamics in a crystalline protein. If the lattice disorder ((x2)ld) term is the major contributor to the observed (x2) there should be little or no temperature dependence for the individual atomic (x2) values. It should be clear, however, that a conformational disorder which is static at room temperature cannot be distinguished from disorder in the crystal packing by lowering the temperature, since both will show no decrease in (x2). Small molecule crystallographers are familiar with these concepts, since it is routine to measure data at low temperature to improve precision by reduction of thermal motion, and structures are often done at multiple temperatures to assess the origins of disorder in atomic positions. Albertsson et al. (1979) have reported the analysis of the crystal structure of D(+)-tartaric acid at 295, 160, 105, and 35 K. Figure 22 shows the individual isotropic B-factors for the atoms in the structure at each of these temperatures: the smooth variation of B with T is apparent. Below 105 K, B is essentially identical for all atoms and is also temperature independent; the value of B = 0.7 A2 agrees well with the expected zero-point vibrational value. However, even for this simple structure, not all of the atoms show B vs T behavior at high temperature which extrapolates to 0 A2 at 0 K. The first study of the temperature dependence of the B-factors in a protein was carried out by Frauenfelder et al. (1979), who showed that measurements on sperm whale metmyoglobin crystals at four temperatures from 220-300 K were incompatible with the traditional notion of a rigid macromolecule, or of a crystal dominated by lattice disorder. The dominant contribution to the observed B-factors was found to be intramolecular motion. The protein could be described as having a condensed core around the heme, with more mobile, semiliquid regions toward the outside of the molecule. There was a strong correlation between the secondary and tertiary structure interactions of an atom and its observed mean-square displacements. Unfortunately, the temperature range available was not extensive enough to permit full exploitation of the power of the method. However, two studies of protein crystal structures at very low temperatures have now reported refined B-factors. These investigations on myoglobin and trypsinogen reveal the richness of protein flexibility and demonstrate that crystallography at subzero temperatures can provide a complete map of the spatial distribution of the high-frequency motions in a macromolecule.

I. Trypsinogen Huber and associates have refined the structure of trypsinogen in methanol-water, initially at 213 K (Singh et al., 1980) and subsequently

PROTEINS AT WORK

349

FIG.22. Molecular geometry and thermal ellipsoids (75% probability) of D( +)-tartaric acid at 295, 160, 105, and 35 K.

350

PIERRE DOUZOU AND GREGORY A. PETSKO

at 173 and 103 K using X-ray data collected with synchotron radiation (Walter et al., 1982).The purpose of this study was to ascertain if the socalled activation domain, a region of 15% of the structure which is disordered in the native crystal structure at room temperature (Huber, 1979), became ordered at the low temperature. Cooling to 103 K reduced the overall B-factor of the trypsinogen structure from 16.1 to 11.5 A2, and one or two of the residues at the N-terminal region became more clearly visible, but the majority of the amino acids in the activation domain remained disordered. There are three possible explanations for this effect: either the domain is still mobile at 100 K, the disorder is static at all temperatures, or a dynamic disorder has been frozen into a static disorder in which a number of conformations are of nearly equal energy and all are nearly equally populated, but none to such an extent that it is visible above the noise level in the electron density map. A number of elegant chemical labeling (Walter et al., 1982) and physical measurements (Butz et al., 1982)have now been carried out by Huber and coworkers, who conclude that the disorder in the activation domain of trypsinogen is dynamic in solution at room temperature with conformational transitions occurring with a reorientation correlation time of l l ns. One interesting aspect of the trypsinogen low-temperature crystallographic study is that cooling to 173 K reduced the overall B-factor to 11.6 from 16.1 A*, so that no further change in overall B occurred on cooling to 103 K (Walter et al., 1982).This may reflect the relaxation temperature believed to exist for proteins at around 180 K, below which many of the collective motions are frozen out, or it may indicate that increasing lattice disorder on extreme cooling compensated for reduction in mobility. 2. My oglobin

In a colliborative study, Parak, Ringe, Frauenfelder, Petsko, and coworkers have determined the structure of metmyoglobin to 2 A resolution at 80 K (Hartmann el al., 1982).Data collection at this temperature was achieved by a “flash-freezing”technique that did not involve the use of a cryoprotective mother li uor. The overall B-factor decreased from 14 A2at 300 K ( ( x 2 ) = 0.17512)to 5 Hi2at 80 K ( ( x 2 ) = 0.063A2). The striking drop in ( x 2 ) with cooling observed in this study and that of Huber’s group on trypsinogen are the most convincing evidence that proteins are flexible in the crystalline state and that static disorder does not dominate the observed B-factors. When the 80 K data were combined with those observed earlier in a study of a more restricted range of temperatures (Frauenfelder et al., 1979),it was seen that only 46 of the 153 residues in myoglobin had average B-factors which extrapolated to 0

35 1

PROTEINS AT WORK

at 0 K. The temperature dependence of the remainder of the protein was inconsistent with simple harmonic vibration, but consistent with the notion that conformational substates could be frozen out at sufficiently low temperatures. Fifty-one residues can be modeled with a linear temperature dependence with (x2) = 0.04 A2 at 0 K. The remaining 56 amino acids in myoglobin do not show a good linear fit of (x2) vs T. Figure 23 shows the average (x2) values for some residues in myoglobin at the four temperatures for which measurements have been made to date. Figure 24 shows a plot of the average B-factor at both 300 and 80 K for the backbone atoms in myoglobin vs residue number. From these data it can be seen that, despite differences in crystal, data collection method, absorption correction, and other parameters, the overall correlation of (x2) with structure is the same at the two temperatures. The two curves also show that residues with large displacements tend to have large temperature dependence, while residues with small B-factors show little change on cooling. Although the observed data do not fit a simple harmonic motion model, this behavior is quasi-harmonic. Huber has noted similar correlation of the temperature dependence of B with its magnitude of trypsinogen (Walter et al., 1982). As mentioned above, spectroscopic data on a number of proteins suggest that there is a relaxation temperature at about 180 K. Unfortunately, the present data for myoglobin cluster in points far removed from this value. The data in Fig. 23 could be fitted by a single straight

"'1 100 200 Temperature (K)

300

FIG. 23. Temperature dependence of (x2) of some selected amino acids (main chain averages) in metmyoglobin. +, Gly-121 (GH3); X , Ma-19 (AB1); A, Glu-83 (EF6); 0,Phe138 (H14); 0,Leu-11 (A9); D, Val-68 (E9); 0 ,Ala-71 (E14); A, Asp-126 (H2); 0 ,average of all main-chain non-hydrogen atoms.

352

PIERRE DOUZOU AND GREGORY A. PETSKO

Residue

FIG.24. Average backbone (N, a C, carbonyl C) (x4) values for myoglobin versus residue number. 0, 80 K; 0, 300 K. An (x4)ld of 0.045 AP has been subtracted from the observed (x4) values in both structures.

line with high temperature scatter or by two lines which intersect at just below 200 K, one of which is nearly temperature independent. Although the latter interpretation is in accord with expectation, it is clearly important to determine the exact form of the temperature dependence experimentally. Parak and associates are currently measuring myoglobin data at 165 and at near 4 K;these two points should allow a proper determination. One remarkable observation from the comparison of the myoglobin structures at 300 and 80 K was an apparent reduction in the volume of the protein at low temperature. The unit cell volume of the crystal is 4.5% smaller at 80 K than at room temperature (Hartmann et al., 1982), and this is a reflection of a general shrinkage of the protein by approximately the same amount (Frauenfelder et al., 1984). Of course, proteins should have a thermal expansion coefficient, but because of the nonuniformity of interactions within the molecule, the expansion coefficient might be expected to be anisotropic. An analysis of the spatial distribution of the shrinkage of myoglobin on cooling has been carried out, and it is found that some regions of the molecule move much more than others. In particular, an external loop of charged amino acids, the socalled CD corner, moves in toward the center of the protein considerably more than any other region. Some of the a helices in myoglobin contract on cooling, while others are unchanged. Taken together with the temperature dependence of the (x2) values, this study of the thermal expansion of myoglobin provides the most detailed picture yet available of the physical chemistry of a protein. i t

PROTEINS AT WORK

353

should be noted also that Huber observed a change in the center of mass of trypsinogen on cooling to 100 K, consistent with an anisotropic contraction of the structure (Walter et al., 1982). Since myoglobin is somewhat atypical, being an all-helical protein with a large prosthetic group, analysis of the thermal expansion of other crystalline proteins is awaited with interest. Finally, we note that low-temperature crystallographic studies have been carried out on one nucleic acid, the B-DNA dodecamer whose room-temperature structure was solved in Dickerson's laboratory (Dickerson, 198 1). Refinement at 16 K revealed a large overall drop in B, but some of the atoms in the molecule still had very large B-factors even at this very low temperature. These large residual mean-square displacements were interpreted as demonstrating the presence of static disorder; however, by analogy with the results on myoglobin, a disorder which is dynamic at room temperature but becomes frozen into a static distribution at low temperature is also consistent with the observations. It is also possible that the disorder in these atoms is dynamic even at 16 K; this point has been considered by Hartmann et al. (1982). VI. PROSPECTS AND PROBLEMS A . Mapping Enzymatic Reaction Pathways

The ultimate objective of an X-ray cryoenzymological study is the mapping of the structures of all kinetically significant species along the reaction pathway. In the case of ribonuclease A this has been largely achieved, as described above. Other enzymatic reactions now await application of the same techniques. Unfortunately, not all crystalline enzymes lend themselves to study by this method. In some cases it may be impossible to find a suitable cryoprotective mother liquor; in others, the reaction may occur too rapidly at ordinary temperature. A reaction with kcat of seconds and an activation enthalpy of -6 kcal mol-' will not be quenched even at -75°C.The approach we have described in this article can be applied to only a small number of enzymes. T w o likely candidates for successors to ribonuclease are the enzymes yeast triosephosphate isomerase and porcine pancreatic elastase.

I . Triosephosphate Isomerase The glycolytic enzyme triosephosphate isomerase (TIM) catalyzes the interconversion of the sugar phosphates dihydroxyacetone phosphate (DHAP) and D-glyceraldehyde 3-phosphate (GAP). Although the reaction is extremely fast, the enzyme is an attractive candidate for mapping

354

PIERRE DOUZOU AND GREGORY A. PETSKO

the catalytic pathway because there is only one substrate and one product. For a hydrolytic enzyme, if one diffuses substrate into active crystals at room temperature, substrate will bind to the catalytic site, be converted into product, and the products will diffuse away. For an isomerase, in contrast, substrate will bind and be transformed into product, but product is just the substrate for the back reaction, which procedes at an appreciable rate. Therefore, if the substrate concentration is given in excess of the K,, the system will settle to equilibrium and the most stable bound species will saturate the crystal. If the equilibrium constant for the bound species is near unity, the structure observed will be an equal mixture of enzyme-substrate and enzyme-product complex. Isomerizations and mutations are general classes of enzymatic reaction for which low temperatures are not needed for direct crystallographicobservation of productive Michaelis complexes. Of all the crystalline enzymes from these two classes, TIM is the most attractive for such a study. The reaction is the simplest in all of metabolism, and the yeast enzyme gives stable crystals which are catalytically active and diffract to high resolution. Further, Knowles and associates have determined the complete free energy profile for the reaction, permitting calculation of the favored species under any given conditions (Knowles and Albery, 1976).The structure of native yeast TIM has been solved and refined (Alber, 1981), and the structure of the productive TIM-DHAP complex has also been determined at 3.5 A resolution (Alber et al., 1981a). Extension of this structure to 1.9 A resolution is underway. To further characterize the mechanism of action of TIM, the structure of a complex of the enzyme with the transition state analog inhibitor phosphoglycohydroxamate has just been solved at 2.8 A resolution and is currently under refinement at 1.9 8, resolution (R. C. Davenport and D. Ringe, unpublished). Preliminary information from the structures of both TIM-DHAP and the transition state analog complex indicates a large conformational change in a flexible loop of 10 amino acids on substrate binding (Alber et al., 1981a). An interesting sidelight of this study is that semi-low temperature had to be used (- 10°C) to inhibit an unwanted side reaction which occurs with this enzyme due to the instability of an intermediate in the reaction. The approach used here to examine an enzyme-substrate complex directly without cryocrystallographic techniques can be applied to a crystalline isomerase or mutase.

2. Serine Proteases Elastase was.the first enzyme for which a productive enzyme-intermediate complex was observed crystallographically by subzero-temperature

PROTEINS AT WORK

355

techniques. As explained earlier, that investigation did not procede beyond 3.5 8, resolution because of the amount of data required for higher resolution structure determination. Recent developments in area detector technology make it possible once again to examine productive elastase-substrate complexes. The enzyme is attractive for such studies because it is readily crystallized, will tolerate high concentrations of methanol, and has essentially full catalytic activity in the crystalline state. The structure of the acyl-elastase intermediate described above should be extended to at least 1.9 resolution and full refinement carried out. The structure of the pre-acyl-intermediate, which may be the tetrahedral adduct, could be investigated by use of p-nitroanalide synthetic substrates. As the turnover time for such a complex is faster than for an acyl-intermediate, it will be necessary to either lower the temperature to -75°C or use a very small substrate which will diffuse rapidly into the crystal, or both. Finally, the enzyme-product complex should be easy to form and will be stable enough for direct structure determination without low temperatures. The same protocol could be followed with other serine proteases such as trypsin or chymotrypsin, but elastase is the member of this class most easily transferred to a cryosolvent.

B . Large Conformational Changes If a large conformational change in the enzyme is triggered by substrate binding or by the formation of one or more of the transition states or intermediates during the reaction, the crystal lattice may not be able to withstand the movements that occur. In such cases the crystal cracks or dissolves on exposure to substrate. Cross-linking with glutaraldehyde may stabilize the crystal sufficiently to permit data collection, but there is always the risk that the cross-linkers may inhibit the very motions that one wishes to observe. Nevertheless, we believe that cross-linking has not been exploited as often as it should be. It is usually possible to measure the activity of the crystalline protein with and without cross-linking to test for such adverse effects. Of course, if the region of the protein that is involved in the conformational change is not involved in an intermolecular contact in the lattice, the movements of the atoms in question may proceed unhindered and the crystal may remain intact. If the crystal is destroyed on substrate addition and cross-linking is not possible, the only solution is to look for a different crystal form. It may be possible to crystallize the substrate-bound conformation of the enzyme by binding an inhibitor to the active site in solution, crystallizing the complex, and then diffusing the inhibitor out or exchanging it for substrate. All of these experiments are just searches for conditions under which there is no physical obstruction to the machinery of catalysis.

356

PIERRE DOUZOU AND GREGORY A. PETSKO

It is worth mentioning that cryoenzymological experiments may fail in the crystalline state, without such gross evidence as crystal destruction, if there is a large conformational change required for binding (“induced fit”). If the packing interactions are stronger than the driver force for the conformational change, and if the region involved in the change is also involved in the lattice contacts, then it is possible that the lattice may stabilize the native conformation of the protein and prevent any change from taking place. In such a situation, the crystal remains intact but substrate is not bound, despite an accessible active site. Again, the only solution for this problem is to find another crystal form. C . Application to Other High-Resolution Techniques

I , NMR at Subzero Temperatures Recent advances in solid-state NMR techniques offer hope that the power of this spectroscopic method can be linked with low-temperature biochemistry. Historically, subzero-temperature NMR, even in fluid cryosolvents, was not possible because the increased line widths of the resonances due to reduced mobility prevented resolution of signals. New superconducting magnets have given rise to a generation of high-field instruments that are capable of resolving resonances in a small (MW 20,000) protein at -20°C in a methanol-water mixture (Fink et al., 1984). Two-dimensional techniques for improving resolution, the possibility of preparing specifically labeled proteins by microbiological means, the ability of magic-angle spinning to provide interpretable signals from solid samples all suggest that NMR at low temperatures will be possible in the near future. Since this spectroscopic method is ideal for seeing protons, which are in general not visible to the protein crystallographer, the possible complementarity of the two techniques is most promising. 2 . Raman and FT-IR Spectroscopy Two other spectroscopic techniques that could be combined with the use of low temperatures to provide information complementary to X-ray diffraction are laser Raman spectroscopy and Fourier transform infrared spectroscopy. Both techniques are suited to solid as well as fluid samples, and give information about structure and motions of certain characteristic groups in biological macromolecules. The principle of using subzero temperatures to resolve the individual steps in an enzymatic reaction is not limited to protein crystallography. Although crystallography is unrivaled in its ability to provide detailed atomic resolution pictures of proteins, not all interesting systems can be crystallized, and information about the temporal domain is missing from X-ray results.

PROTEINS AT WORK

357

The combination of crystallography with spectroscopy is the ideal, but spectroscopy alone is also of great value.

D. Time-Resolved Pictures of Functioning Proteins All of experimental science may be viewed as an attempt to overcome the limitation of the human senses. If the eye possessed the ability to see things infinitely far away and infinitely small, there would be no need for telescopes and microscopes. But the dimension of distance is not the only one in which the physical world operates. Of equal importance is the dimension of time. Consider a camera which is capable of recording images of individual atoms. With such a camera, the structures of proteins may be observed directly (X-ray crystallography, of course, is our closest analog to this miraculous camera). Now let us use this camera to try to photograph an enzyme at work. Unless we have a shutter with speeds in the submicrosecond region, despite extraordinarily fast film, all we will see is a blur. This is equivalent to trying to take a picture of a running horse with a conventional camera and slow film; it required stroboscopic photography to “freeze” a horse in midgallop and prove that all four of its legs were off the ground at one time. As we mentioned, X-ray crystallography is our closest analog to the camera of our dreams. But it is a camera with a shutter that is essentially open all the time (averaging over many days of data collection), which negates its rather fast film. We do not, at present, have a means of making the shutter fast enough for viewing functioning proteins directly, although the combination of area detectors and high-intensity X-ray sources such as synchrotrons may eventually enable us to collect complete data sets in minutes or less. Our only recourse is to slow down the reaction, to make the time of transformation of one intermediate to another so long that each intermediate is essentially “frozen” in time and can be viewed as a stable state. By stepping through all of the stable species, one sees, not the actual interconversion of species, but rather a series of “stop-action” pictures of the most highly populated complexes along the reaction pathway. No dynamic information is obtained, but much can be deduced. This article was concerned with the philosophy and techniques behind the use of subzero temperatures to provide temporal resolution of all the significantlystable species in an enzymatic reaction, so that the camera of X-ray crystallography could photograph their structures and mobilities at atomic resolution. Such work is still in its formative stages, and it is not easy. But the reward is great: to see a protein at work, step-by-step without the possible artifacts caused by studying the binding of inhibitors or by building models of enzyme-substrate complexes without in-

358

PIERRE DOUZOU AND GREGORY A. PETSKO

formation about possible conformational changes. The results of such a study are like separate frames in a motion picture. With sufficient complementary data from other techniques, it may be possible to fill in the missing frames and watch, in smooth continuous motion, the chemistry of life.

REFERENCES Akerlof, G. (1932).J. Am. Chem. Soc. 54,4125-4130. Alber, T. (1981).“Structural Origins of the Catalytic Power of Triose Phosphate Isomerase.” Ph.D. thesis, Massachusetts Institute of Technology, Cambridge, Massachusetts. Alber, T., Petsko, G. A., and Tsernoglou, D. (1976).Nature (London) 263,297-300. Alber, T., Banner, D. W., Bloomer, A. C., Petsko, G. A., Phillips, D. C., Rivers, P. S., and Wilson, I. A. (1981a). Philos. Trans. R . SOC.London B 293, 159-171. Alber, T., Hartman, F. C., Johnson, R. M., Petsko, G. A., and Tsernoglou, D. (1981b).J. Biol. Chem. 256, 1356-1361. Albertsson, J., Oskarsson, A., and Stahl, K. (1979).J. Appl. Crystullogr. 12, 537-544. Andersson, K., Benyamin, Y., Douzou, P., and Balny, C. (1978).J. Immunol. Methods 23, 17-21. Ballesta, J. P., and Vasquez, D. (1973).Biochemitty 12, 5063-5068. Balny, C., Le Doucen, C., Douzou, P., and Bieth, J. G. (1979).J.Chromutogr. 168, 133-138. Banerjee, S. K., and Rupley, J. A. (1973).J. Biol. Chem. 248, 21 17. Banyard, S. M. (1973).Ph.D. Thesis, University of Oxford, Oxford, England. Blake, C. C. F., Johnson, L. N., Mair, G. A., North, A. C. T., Phillips, D. C., and Sharma, V. R. (1967).Proc. R. Soc. London 167, 378-388. Borkakoti, N., Moss, D. A., and Palmer, R. A. (1982).Acta Ctystullogr. A 38, 2210-2217. Brody, E., and Leautey, J. (1973).Eur. J. Biochem. 36, 347-361. Butz, T.,Lerf, A., and Huber, R. (1982).Phys. Rev. Lett. 48, 890-893. Campbell, R. L., and Petsko, G. A. (1984).Submitted. Carlisle, C. H., Palmer, R. A., Mazumdar, S. K., Gorinsky, B. A., and Yeates, D. G. R. (1974).J.Mol. Biol. 85, 1-18. Chance, B. (1943).J.Biol. Chem. 151, 533. Cox, R. P.(1978).Biochem. Soc. Trans. 6,689-697. Crepin, M.,Cukier-Kahn, R., and Gros, F. (1975).Proc. Natl. Acad. Sci. U.S.A. 72,333-337. Davies, R.C., Neuberger, A., and Wilson, B. P. (1969).Biochim. Biophys. Acta 178,294-305. Debey, P., Balny, C., and Douzou, P. (1976).FEBS Lett. 69, 231-235. Debye, P.(1954).‘Collected Papers of P. Debye,” p. 335. Wiley (Interscience), New York. Dickerson, R. E. (1981).J.Mol. Biol. 149. 761-786. Douzou, P. (1971a).Biochimie 53, 307. Douzou, P. (1971b).Biochimie 53, 1135. Douzou, P. (1974).Methods Biochem. Anal. 22,401. Douzou, P. (1977a).Adv. Enzymol. Relut. Areas Mol. Biol. 45, 157.-272. Douzou, P. (1977b).“Cryobiochemistry, an Introduction.” Academic Press, New York. Douzou, P., and Balny, C. (1978).Adu. Protein Chem. 32, 77-185. Douzou, P., and Leterrier, F. (1970).Biochim. Biophys. Acta 220, 338. Douzou, P., and Maurel, P. (1976).C . R . Hebd. Seances Acad. Sci. Ser. B. 282, 43-46. Douzou, P., Sireix, R., and Travers, F. (1970).Proc. Natl. Acud. Sci. U.S.A. 66,787.

PROTEINS AT WORK

359

Douzou, P., Hui Bon Hoa, G., and Petsko, G. A. (1975).J. Mol. Biol. 96, 367. Douzou, P., Hui Bon Hoa, G., Maurel, P., and Travers, F. (1976). Handbooh Biochem. Mol. Biol. 3rd Ed., Vol. 1, p. 520. Dreyfus, M., Vandenbunder, B., and Buc, H. (1978). FEBS Lett. 95, 185-189. Edsall, J. T. (1947). Adv. Protein Chem. 3, 384. Edsall, J. T., and Wyman, J. (1958). “Biophysical Chemistry,” Vol. 1, p. 472-476. Academic Press, New York. Engasser, J. M., and Horvath, C. (1975). Biochem.]. 145,431-435. Findlay, D., Herries, D. G., Mathias, A. P., Rabin, B. R., and Ross, C. A. (1962). Biochem.J. 85, 152-153. Fink, A. L. (1973a). Arch. Biochem. Biophys. 155,473-474. Fink, A. L. (1973b). Biochemisty 12, 1736-1742. Fink, A. L. (1976a). Biochemistry 15, 1580-1586. Fink, A. L. (1976b). J. Theor. Biol. 61,419-445. Fink, A. L., and Ahmed, A. I. (1976). Nature (London) 269, 294-297. Fink, A. L., and Geeves, M. A. (1979). In “Methods in Enzymology” (D. L. Purich, ed.), Vol. 63, p. 336. Academic Press, New York. Fink, A. L., and Meehan, P. (1979). Proc. Nutl. Acad. Sci. U.S.A. 76, 1566-1570. Fink, A. L., and Petsko, G. A. (1981). Adv. Enzymol. Rekzt. Areas Mol. Biol. 52, 177-246. Fink. A. L., and Wildi, E. (1974).J. Biol. Chem. 249, 6087-6089. Fink, A. L., Kar, D., and Kotin, R. (1984). Submitted. Fletcher, N. H. (1962). “The Physics of Rain Clouds.” Cambridge Univ. Press, London and New York. Frauenfelder, H., Petsko, G. A., and Tsernoglou, D. (1979).Nature (London) 280,558-563. Frauenfelder, H., Hartmann, H., Karplus, M., Kuntz, I. D., Kuriyan, J., Parak, F., Petsko, G. A., Ringe, D., Connoll, M., Max, N., Tilton, R. F. (1984). Science, submitted. Freed, S . (1965). Science 150,576-584. Gilbert, W. A., and Petsko, G. A. (1984a). Submitted. Gilbert, W. A., and Petsko, G. A. (1984b). Submitted. Gilbert, W. A., and Petsko, G. A. (1984~).Submitted. Gilbert, W. A., and Petsko, G. A. (1984d). Submitted. Gilbert, W. A., Petsko, G. A., Wlodauer, A., and Sjolin, L. (1984). Submitted. Golstein, L. (1972). Biochemistry 11,4072-4084. Goldstein, L., Levin, Y.,and Katchalsky, E. (1964). Biochemistry 3, 1913-1919. Graves, D. J., and Wang, J. H. (1972). I n “The Enzymes” (P. Boyer, ed.), Vol. 7, pp. 435482. Academic Press, New York. Gunsalus, I. C., Meeks, J. R.,Lipscomb, J. D., Debrunner, P. G., and Miinck, E. (1974). In “Molecular Mechanisms of Oxygen Activation” (D. Hayaishi, ed.), pp. 559-613. Academic Press, New York. Haas, D. J. (1968). Acta Crysfullogr.B 24, 604. Hamel, E. (1972). Biochemistry 11, 3933-3938. Hartmann, H., Parak, F., Steigemann, W., Petsko, G. A., Ringe Ponzi, D., and Frauenfelder, H., (1982). Proc. Natl. Acad. Sci. U.S.A. 79, 4967-4971. Hayashi, K., Imoto, K., and Funatsu, M. (1963).J. Biochem. 54, 381. Holmes, R. R. (1976). Znt. J. P e p . Protein Res. 8, 445-462. Huber, R. (1979). Trends Biochem. Sn’. (Pers. Ed.) 4, 271-276. Hui Bon Hoa, G., and Douzou, P. (1973).j. Biol. Chem. 248,4649. Hui Bon Hoa, G., Guinand, S., Pantaloni, C.,and Douzou, P. (1973). Biochimie 55, 269274. Hunkapiller, M. W., Forgac, M. D., and Richards, J. H. (1976).Biochemistty 15,5581-5586.

360

PIERRE DOUZOU AND GREGORY A. PETSKO

Imoto, T., Johnson, L. N., North, A. C. T., Phillips, D. C., and Rupley, J. A. (1972). In “The Enzymes” (P. D.Boyer, ed.), 3rd ed., Vol7, p. 665. Academic Press, New York. Jencks, W. P. (1975). Adv. Enzymol. Rekzt. A r e a Mol. Biol. 43, 219-410. Jencks, W. P. (1981). In “Concepts of Chemical Recognition in Biology.” Springer-Verlag, Berlin and New York. Johnson, F. H., and Eyring, H. (1970). In “High Pressures Effects on Cellular Processes” (A. M. Zimrnerman, ed.), pp. 1-44. Academic Press, New York. Kartha, G., Bello, J., and Harker, D. (1967). Nafure (London) 219, 862-865. Katchalsky, E. (1954). Prog. Biophys. Biophy. Chem. 4, 1-26. Katchalsky, E., Bichowsky-Dlomnistzki,L., and Volcani, B. E. (1953). Bi0chem.J. 55,677680.

Kirkwood, J. G. (1939).J. Chem. Phys. 7,911-928. Klotz, I. M. (1970). Arch. Biochem. Biophys. 138, 704-706. Knowles, J. R., and Albery, W. J. (1976). Biochemistry 15, 5627-5631. Konnert, J. H., and Hendrickson, W. A. (1980). Acta Crystallogr. A 36, 344-349. Koshland, D. E., and Neet, K. E. (1968). Annu. Rev. B i o c h . 37, 359-410. Kraut, J. (1977). Annu. Rev. Biochem. 46, 331-358. Laidler, K. J. (1950). Arch. B i o c h . 30, 226-236. Lange, R., Hui Bon Hoa, G., Debey, P., and Gunsalus, I. C. (1977). Eur. J. B i o c h . 77, 479-485. Lange, R. et al. (1984). In preparation. Larroque, C., Maurel, P., and Balny, C., and Douzou, P. (1976). Anal. B i o c h . 73,9-14. Le Peuch, C., and Balny, C. (1978). FEBS Lett. 87,232-234. Lienhard, G. E. (1973). Science 80, 149-154. Low, P. S., and Sornero, G. N. (1975a). Proc. Natl. Acad. Sci. U.S.A. 9,3305-3309. Low, P. S., and Sornero, G. N. (1975b). Proc. Natl. Acad. Sci. U.S.A. 8, 3014-3018. Lumry, R., and Rajender, (1970). Biopolymers 9, 1125-1127. McLaren, A. C., and Esterman, E. F. (1957). Arch. Biochent. Biophys. 68, 157-160. Marsh, D. J., and Petsko, G. A. (1973).J. AHI. Crystallogr. 6,76-80. Maurel, P. (1976). Thesis, University of Paris. Maurel, P., and Douzou, P. (1975).J. Biol. Chem. 250, 2678. Maurel, P., and Douzou, P. (1976).J. Mol. Biol. 102, 253-264. Maurel, P., Hui Bon Hoa, G., and Douzou, P. (1975).J. Biol. Chem. 250, 1376-1382. Melin, M., and Taylor, H. L. (1946).J.Am. Chem. SOC. 68, 459-490. Nakanishi, S., Adhya, S., Gottesman, M., and Pastan, I. (1974).J. Biol. Chem. 249, 40504056. Page, M. I., and Jencks, W. P. (1971). Proc. Natl. Acad. Sci. U.S.A. 68, 1678-1683. Parak, F., Frolov, E. N., Mossbauer, R. L., and Goldanskii, V. I. (1981).J.Mol. Biol. 145, 825-833. Pauling, L. (1946). Chem. Eng. News 24, 1975-1377. Peterson. J. A., Ishimura, Y., Baron, J., and Eastabrook. R. W. (1973). In “Oxidases and Related Redox System” (T. E. King, H. S. Mason, and M. Morisson, eds.), pp. 565581. Univ. Park Press, London. Petsko, G. A. (1975).J.Mol. Biol. 96, 381-392. Petsko, G. A., and Wilz, S. (1984). Proc. Natl. Acad. Sci. U.S.A., in press. Petsko, G. A,, Rose, D.,Tsernoglou, D., Ikeda-Saito, M., and Yonetani, T. (1978). In “Frontiers of Biological Energetics” (P. L. Dutton, J. S . Leigh, and A. Scarpa, eds), Vol. 2, p. 101 1. Academic Press, New York. Phillips, D. C. (1967). Proc. Natl. Acad. Sci. U.S.A. 57, 484. Phillips, S. E. V. (1978). Nature (London) 273, 247-248.

PROTEINS AT WORK

36 1

Porath, J. (1974).In “Methods in Enzymology” (W. B. Jakoby, ed.), Vol. 34, pp. 13-29. Academic Press, New York. Rasmusse, D. H., McKenzie, A. P., Angell, C. A., and Tucker, J. C. (1973).Science 131, 342-344. Ricard, J., and Job, D. (1974).Eur..; Biochem. 44,359. Richards, F. M., and Wyckoff, H. W. (1971).In “The Enzymes” (P. D. Boyer, ed.), 3rd Ed., pp. 647-806. Academic Press, New York. Ringe, D., Petsko, G. A., Yamakura, F., Suzuki, K., and Ohmori, D. (1983).Proc. Nutl. Acud. Sn’. U.S.A. 80, 3879-3883. Roberts, G. C. K., Dennis, E. A., Meadows, D. H., Cohen, J. S.,and Jardetsky, 0. (1969). Proc. Nutl. Acud. Sci. U.S.A. 64, 1151-1158. Rudman, R. (1976).“Low Temperature X-ray Diffraction.” Plenum, New York. Rupley, J. A. (1969).In “Protein Structure in the Crystal and in Solution” (S.N.Timasheff and G. D. Fasman, eds.), pp. 251-348. Dekker, New York. Salton, M. R. I. (1964).Proc. 3rd Symp. F h i n g ’ s Lysozyme, Milan, p. 5. Singer, S.J. (1962).Adu. Protein Chem. 17, 1-68. Singh, T. P., Bode, W., and Huber, R. (1980).Acta Cyskdlogr. E 36, 621-627. Smith, W. W., Ludwig, M. L., Pattridge, K. A., Tsernoglou, D., and Petsko, G. A. (1978).In “Frontiers of Biological Energetics (P. L. Dutton, J. S. Leigh, and A. Scarpa, eds.), Vol. 2,p. 957. Academic Press, New York. Steitz, T. A., and Shullman, R. G. (1982).Annu. R m . Eiophys. Eioeng. 11,419-444. Steitz, T. A., Ludwig, M. L., Quiocho, F. A., Lipscomb, W. N. (1976).J. Mol. Eiol. 444, 4662-467 1. Timmermans, J., Piette. A. M., and Phillipe, R. (1955).Bull. SOC.Chim. Eelg. 64, 5-9. Travers, F., and Douzou, P. (1970).J . Phys. Chem. 74,2243-2244. Travers, F., and Douzou, P. (1974).Eiochimie 56,509-514. Travers, F., Douzou, P., and Pederson, T., and Gunsalus, I. C.(1975).Eiochimie 57,43-48. Usher, D. A., Richardson, D. I., and Eckstein, F. (1970).Nulure (London) 248,663. Voigt, J., Sander, G., Nagel, K., and Parmeggiani, A. (1974).Biochem. Etophys. Res. Commun. 57, 1279-1286. Walter, J., Steigemann, W., Singh, T. P., Bartunik, H. D., Bode, W., and Huber, R. (1982). A ~ t oC y k d b p . B 38, 1462-1472. Weber, G. (1979).Biochemisty 11, 872. Willis. B. T. M., and Pryor, A. W. (1975).“Thermal Vibrations in Crystallography.” Cambridge Univ. Press, London and New York. Wlodawer, A. (1980).Actu Cytullogr B 36, 1826-1831. Wlodawer, A., and Sjolin, L. (1982).J.Eiol. Chem. 457, 1325-1332. Wolfenden, R. (1969).Nature (London) 443, 704-705. Wolfenden, R. (1972).Acc. Chem. Res. 5, 10-18. Wyckoff, H. W., Doscher, M., Tsernoglou, D., Inagami, T., Johnson, L. N., Hardman, K. D., Allewell, N. M., Kelley, D. M., and Richards, F. M. (1967).J.Mol. Biol. 127,563578. Wyckoff, H. W., Tsernoglou, D., Hanson, A. W., Knox, J. R., Lee, B., and Richards F. M. (1970).J . Biol. C h m . 445,305-328. Wyman, J. (1936).J.Am. C h . SOC.58, 1842-1844.

This Page Intentionally Left Blank

AUTHOR INDEX Numbers in italics refer to the pages on which the complete references are listed.

A Abduraschidova. G. G., 40,53, 54 Abramowiu, R., 44.52, 53 Acher, R., 220(107), 242 Acheson, N. H., 116, 120, 125 Ade, W., 209 Adesnik, M., 110, 130 Adhya, S., 273,360 Adler, A. J., 10, 48 Adler, C. J., 83, 128 Agganval, S. J., 226(47), 240 Ahrned, A. I., 257,331,359 Air, G. M., 110, 126 Akerlof, 284,286,358 Akermann, I., 13, 14,50 Akeson, A., 235(94), 242 Alakhov, Y.B., 6, 7, 21, 23, 74, 78 Alber, T., 282, 324, 330, 332, 337, 354, 358 Albertsson, J., 348,358 Albery, W. J., 360 Allen, C., 151,210 Allen, G., 8, 14, 19, 20, 23, 34, 39, 40, 43, 44, 48, 61, 77 Allewell, N. M., 324, 337, 361 Alperti, G., 107, 125 A h , M., 156, 158,210 Altosaar, I., 235(88), 242 Amons, R., 44,51 Amos, L. A., 144, 147, 149, 150, 152, 153, 172,207,209 Anderegg, J. W., 29, 49 Anderson, C. W., 83, 128 Anderson, K., 139,212 Anderson, R. G. W., 80, 100, 119,125, 127

Anderson, S . A., 155,207 Anderson, S . M., 235(92), 242 Anderson, K., 293, 295, 358 Angel, C. A., 318,361 Antonini, E., 218(13, 19), 226(48), 239, 240 Appelt, K., 15, 48 Arad, T., 32, 33,50 Arai, K., 9, 48 Arai, T., 173, 174,207 Arfsten, U., 6, 7, 8, 14, 20, 23, 68, 72, 75, 77, 78 Argarana, C. E., 157,207 Arnone, A., 217(103), 242 Aronson, N. N., 101,126 Artymiuk, P. J., 236( 124), 243 Asakura, S., 159, 161, 192,210 Asnes, C. F., 139,207 Atha, D. H., 226(49), 240 Atkinson, P. H., 127 Aune, K. C., 18, 21, 22, 44, 48, 49, 52, 53 Avital, S., 44, 52, 53

B Baca, 0. G., 40,48 Bacharach, A. D. E., 115, 127 BPumert, H. G., 40,48 Bald, R., 2, 32,50,53 Baldwin, T. O., 220(109), 242 Ballesta, J. P., 273, 358 Balny, C., 292, 293, 295, 296,358, 360 Baltimore, D., 124, 132 Banerjee, S. K., 264,358 Banks, R. D., 236(121), 242 Banner, D. W., 354,358 363

364

AUTHOR INDEX

Banville, D., 220(50), 227(50), 240 Banyard, S. M., 258,358 Baranov, V. I., 30, 31,54 Barber, C., 103,130 Barbey, K., 233(76), 241 Barbieri, M., 32, 48 Baron, J., 251,360 Barra, D., 220(26, 106), 221(26), 240, 242 Barra, E., 220(130, 131), 243 Barra, H. S., 157, 207, 208 Barrell, B., 25, 48 Barron, E. S. C., 233(75), 241 Bartels, H., 233(76), 241 Bartlett, G . R., 221(29), 240 Bartunik, H. D., 324, 350, 351, 352,361 Barynin, V. V., 236(123), 243 Bassel, A. R., 27, 49 Basu, S. K., 119, 125 Batelier, G., 166, 210 Bauer, C., 228(58, 59, 60), 229(61), 231(59, 61), 233(79, 80), 237(100, 101), 241,242 Baumann, R., 232(66), 241 Bayley, P. M.,103, 131 Beckman, A., 44,50 Behlke, J., 14, 19, 49 Bell, J. R., 83, 92, 94, 113, 125, 130 Bell, J. W.,Jr., 117, 125 Bello, J., 225(40), 240, 335,360 Benyamin, Y.,293, 295,358 Berezesky, I. K., 105, 116, 127 Berger, E. G., 116, 130 Bergen, L. G., 172, 178, 180, 181, 202, 203,207 Bergmann, J. E., 81, 115, 125 Berkowitz, S. A., 139,207 Berne, B. J., 178, 185,207 Berry, R. W.,144, 173,207 Betke, K., 233(76), 241 Beychok, S., 158,208 Bibring, T., 135, 211 Biesecker, G., 236(118), 242 Bieth, J. G., 293,358 Bichowsky-Dlomnistzki, L.,313,360 Bigler-Meier, B., 110, 127 Binder, L. I., 151,208, 211 Binotti, I. C., 218(19), 240 Birdwell, C. R.,120, 126 Bischoff, H., 107, 132 Bishop, D. H. L., 125, 126

Bitar, K. G., 78 Black, C. P., 232(67), 241 Black, J. A., 237(99), 242 Blake, C. C. F., 236(121, 124), 242, 243, 313,358

Blobel, G., 79, 80, 92, 108, 109, 110, 117, 126, 128,129, 132

Blok, J., 110, 126 Bloomer, A. C., 354,358 Bode, W.,324, 348, 350, 351, 353,361 Bodley, J. W.,40, 48 Boege, U.,83,92, 107,126, 132 Bogdanov, A. A,, 27,28,51,52 Bogg, C. E., 240 Boileau, G., 38, 39, 53 Bok, D., 115,127 Bolzan, E., 100, 101, 102, 129 Bonatti, S., 92, 108, 110, 114,126 Bonaventura, C., 218(13), 226(49), 227(51), 233(71), 239, 240,241

Bonaventura, J., 218(13), 220(26), 221(26), 226(49), 227(51), 233(71), 239,240,241 Bordas, J., 164, 167, 168, 210 Bordier, C., 91,126 Borisy, G. G., 138, 139, 140, 144, 148, 151, 152, 154, 155, 156, 162, 163, 164, 172, 173, 178, 180, 181, 197, 202,203,206,207,208,210,211 Borkakoti, N.,335,358 Bose, H. R.. 119,128, 132 Bossa, F., 218(13), 220(26, 106, 130, 131), 221(26), 239,240, 242, 243 Bosserhoff, A., 8, 14, 20, 22, 58, 77 Boublik, M., 27, 30, 31, 48, 52 Boughter, M. J., 135,208 Boulan, E. J., 125, 130 Bourne, S., 227(51), 240 Boutilier, R. G., 227(56), 241 Bowie, T., 235(94), 242 Bracha, M., 113, 119, 121,126, 131 Bradaczek, H., 21, 49 Bradbury, E. M., 13, 14,51 Braell, W.A., 110, 126 Brand, C. M., 103, 126 Branden, C.-I., 235(94), 242 Brands, R., 116, 127 Branlant, C., 25, 48 Brauer, D., 5, 6, 8, 9, 20, 22, 34, 38, 39, 40, 44,48, 54, 58, 77

365

AUTHOR INDEX Braunitzer, G., 220(24, 57), 221(23a, 25), 227(57), 229(61), 230(113), 231(61), 232(63, 64, 65, 66), 233(68, 79, 80, 811, 234(85), 240, 241,242 Breiman, A., 44,52 Bretscher, M. S., 99, 101, 123, 130 Bridgen, J., 234(87), 239(87), 242 Brimacombe, R.,2, 25, 26, 28,40, 41, 42, 43, 44, 48, 49, 50, 51, 52, 53, 54, 55 Brinkley, B. R., 135, 151,208,209 Brittain, T., 214(129), 244(129), 243 Brochon, J X . , 12, 48 Brody, E., 273,358 Bronde, N. E., 40,53, 54 Brosius, J.. 6, 7, 23, 25, 34,48,51, 54, 56, 69, 71, 72, 76, 78 Broumand, C., 233(77), 241 Brown, D. L., 211 Brown, D. T., 117, 120, 121,126,131, 132 Brown, E. B., 103,131 Brown, F., 92, 126 Brown, M. S., 100, 119, 155, 125, 127, 208 Brownlee, G. G., 25, 48, 79, 110, 126, 129 Brunori, M., 217(12), 218(13, 19), 219(23), 220(26, 106, 130, 131), 221(23, 26), 222, 223(12), 225, 226(48), 237(12), 239, 240, 241, 242, 243 Bryan, j.,148,211 Bryant, T. N., 236(120), 242 Brylawski, B. P., 146,207 Brzeski, H., 105, 126 Buc, H., 359 Budowsky, 40,53,54 Buehner, M., 236(119), 242 Bullard, R. W., 233(77), 241 Bunn, H. F., 215(5), 233(69), 234(5), 239, 241, 243, 244(129) Burge, B. W., 117, 123,126 Burke,B., 110, 114, 116, 117,126,127, 128, 156,207 Burke, D., 94,95,126,128 Burton, P. R., 148, 152,207,210 Bush, D. J., 148,211 Bushuev, V. N., 14, 18, 19, 21, 22,49,52, 53 Butler, P. J. E., 215(6a), 216(6a), 239 Butz, T., 350,358

Byers, B., 32,48 Bystrova, T. F., 28, 40,52, 53, 54

C Campbell, R. L., 334, 336,358 Cancedda, F. D., 114,126 Cancedda, R., 105, 108, 110,126 Cantor, C. R., 19, 28, 37, 38, 44, 48, 49, 50, 53,54, 138, 141, 185, 187, 208, 211

Capasso, R., 40, 48 Caplow, M., 145, 146, 175, 181, 185, 187, 189, 198, 199, 202, 203, 204, 205, 206,207,208,212 Caputto, R., 157,207, 208 Carbon, P., 25, 48, 53 Carey, F. G., 224(35, 36, 37), 240 Carey, N. H., 103, 130 Carlier, M. F., 166, 174, 175, 182, 188, 207,210 Carlisle, C. H., 335,358 Carne, A. F., 234(87), 239(87), 242 Carta, S., 220(108), 226(48), 240, 242 Caspar, D. L. D., 158,207 Castleman, H., 44,51 Chanas, A. C., 103,126 Chance, B., 358 Chang, F. N., 44, 48 Chang, J. Y., 5, 48 Changchien, L. M., 44, 45, 48, 50 Chapman, D., 96,126 Chauvet, J. P., 220(107), 242 Chen, R., 6, 7, 8, 14, 20, 23, 34, 39, 47, 56, 61, 68, 69, 71, 72, 75, 76, 77, 78 Chen-Schmeisser, U., 6, 8, 20, 23, 68, 77 Chien, J. C., 218(18), 240 Child, F. M., 151,210 Chin, D., 139,212 Choppin, P. W., 103, 130 Chothia, C., 236(95), 242 Chou, P. Y., 144,207 Chu, Y. G., 19, 48 Clark, B. F. C., 9, 48 Clark, M. W., 32, 48 Clarke, J. A., 234(84), 241 Clarke, P. H., 235,242 Claus, T. H., 210 Clegg, J. C. S., 92, 103, 105, 106, 107, 108,126

366

AUTHOR INDEX

Cleveland, D. M., 174,210 Cleveland, D. W., 140, 141, 143, 144, 153, 154,207,211 Clore, G. M., 221(115, 116), 242 Cohen, C., 148,209 Cohn, 2.A., 80, 100. 131 Coletta, M., 220(26, 131), 221(26), 240, 243 Como, P. F., 225(117), 242 Compans, R. W., 90, 114, 126, 130 Condo, S. J., 227(52), 241 Contreras, R., 83, 126, 127 Cooper, S., 18, 21, 22, 44,52 Corfman, D., 141,209 Cote, R. H., 181, 197, 203, 207 Cotter, R. I., 10, 48 Coughlin, B. A., 139, 155, 204,208, 211 Cover, J. A., 39, 49 Cowan, N. J., 141,207 Cox, R. P., 247, 267, 358 Craven, G. R.,44,45,48, 50 Crawford, N.,25,54 Crepin, M., 273,358 Crichton, R. R., 29, 31, 44, 49, 53 Cruiell, M., 82, 83, 96, 128 Cukier-Kahn, R., 273, 358

D Dahlberg, A. E., 44,53 Darnell, J. E., 82, 96, 127 Daune, M., 10, 12, 48, 49 David, A., 82, 96, 127 Davies, M., 224(38), 240 Davies, R. C., 313,358 Davis, J. P., 38, 49 Dayhoff, M. O., 220(105), 242 De Barsy, T., 101, 126 Debrunner, P. G., 251,359 Debye, G., 253,360 Debye, P., 293,358 De Duve, C., 101, 126 Deery, W. J., 145, 174, 175, 185, 187, 188, 189,211,212 Delius, H., 81, 83, 92, 94, 95, 97, 105, 106, 113,127 De Lorenzo, R. J., 156,207 De Maeyer, L. C. M., 165, 182,207

Dennis, P. P., 9, 52 Dentler, W. L., 148, 151, 154, 155, 156, 207,211 Desnuelle, P., 110, 126 Detrich, H. W., 139, 140,207, 211 Devreotes, P. N., 115, 126 De Wolf, B., 31, 53 DeYoung, A., 218(16), 240 Dickerson, R. E., 353, 358 Dickerson, R. E., 236(128), 243 Dickinson, P. J., 145, 174, 175,211 Dijk, J., 2, 10, 13, 14, 15, 21, 22, 23, 38, 44, 48, 49, 50t 51, 53 Dill, D. B., 233(75), 241 Dintzis, H. M., 91, 126 Dobberstein, B., 79, 80, 81, 107, 108, 110, 126, 127, 129 Docherty, K., 114, 126 Dodd, J. A., 27, 53 Dognin, J. M., 5, 54 Dognin, M. J., 6, 14, 21, 23, 40, 43, 56, 70, 78 Dohme, F., 25, 45, 49, 51 Donner, D., 15,50 Dorner, A. J., 83, 128 Doscher, M., 324, 337,361 Dothie, J. M., 235(88), 242 Doty, P., 10,52 Douzou, P., 246, 247, 248, 249, 250, 25 1, 259, 265, 267, 284, 286, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 302, 303, 304, 305, 307, 313, 315,327,358,359,360,361 Dovgas, N. V., 6, 7, 21, 23, 74, 78 Dreyfus, M., 359 Duboi, E. J., 96, 130 Duffy, L., 9, 48 Duhm, J., 252(66), 241 Dull, T. J., 25, 48 Dunn, W. A., 101,126 Dunphy, W. G., 119,126 Dutko, F. J., 99, 129 Dzionara, M., 11, 49, 54

E Eastabrook, R. W., 251,360 Eaton, W. A,, 165,208 Ebel, J. P., 25, 40, 43, 48, 51, 53, 55

AUTHOR INDEX Edelstein, S. J., 149, 150, 152,209, 233(72), 241 Edlind, T. D., 27, 49 Edmundson, A. B., 221(27), 240 Edsall, J. T., 300, 306,359 Edwards, K., 25,28,48, 49 Ehresmann, B., 25, 40, 48, 51 Ehresmann, C., 25, 43,53, 55 Ehrke, G., 6, 7, 8, 23, 68, 76, 77, 78 Eipper, B. A., 141,207 Eklund, H., 235(94), 242 El-Maghrabi, M. R., 210 Elson, D., 30, 44, 50, 52, 53 Elson, E. L., 120, 128 Elzinga, M., 140,209 Emini, E. A., 83, 128 Emtage, J. S., 103, 130 Engasser, J. M., 307,359 Engbretson, G., 227(53), 241 Engel, J., 160, 161,211 Engelbroughs, Y.,165, 167, 182, 207 Engelman, D. M., 18, 34, 37.51, 52 Enzmann, P.-J.,99, 129 Epe, B., 38, 49, 53 Epp, 0.. 15, 48 Erdei, S., 109, 127 Erdmann, V. A., 27, 32, 33,50,51,53,55 Erickson, H. P., 149,207,211 Erikson, S., 15, 22, 23, 50 Esch, F. S., 92, 132 Estafieva, A. G., 28,52 Esterman, E. F., 307,360 Euteneur, U., 152,208 Evans, P. R., 236(121), 242 Expert-Benzancon, A., 40,49 Eyring, H., 277, 278,360

F Fahnestock, S. R., 40,55 Fairclough, R. H., 37, 38, 48, 50 Falcioni, G., 221(115), 242 Fambrough, D. M., 115,126 Farmer, M., 224(34), 227(34), 240 Farquhar, M. G., 80, 114, 122, 123,126 Fasman, G. D., 10,48, 144,207 Fasold, H., 40, 49, 52 Faulaki, K., 9,52 Fee, J. A., 236(127), 243

367

Feldmann, R. J., 93, 131 Fellner, P., 43, 55 Fermi, G., 216(4), 232(4), 239 Ferracin, A., 227(52), 241 Fessenden, R. J., 29, 49 Feytons, V., 6, 23, 72, 78 Field, D. J., 141, 208 Fields, S., 110, 126 Fiers, W., 83, 126, 127 Findlay, D., 336,359 Fine, R. E., 123, 130 Fink, A. L., 254, 256, 257, 267, 280, 282, 320, 32 1, 327, 330, 331, 335, 337, 356,359 Fink, G., 40, 49, 52 Fischer, E. H., 157,208 Fischer, H. D., 116, 131 Fisher, W. K., 220(33, 112), 224(33), 240, 242 Flaks, J. G., 44, 48 Flavin, M., 157,208 Fletcher, N. H., 318,359 Fogg, J. H., 215(6a), 216(6a, 44), 225(44), 239,240 Ford, G. C., 236( 119). 242 Forgac, M. D., 257,360 Forget, B. J., 2 15(5), 234(5), 239 Forster, M., 228(60), 241 Foulaki, K., 8, 20, 23, 34, 38, 43, 57, 77 Franz, A., 18, 21, 22, 23, 49, 51 Frauenfelder, H., 320, 346, 348, 350, 352, 353,359 Freed, S., 287, 359 Freeman, R., 82, 127 Freund, A. M., 10, 18, 22, 49 Friday, A. E., 229(61), 231(61), 241 Friedkin, M., 138, 144,212 Friedman, F. K., 158,208 Friedman, R. M., 105, 116, 127 Fries, E., 81, 89, 99, 100, 101, 102, 119, 123,126, 127, 128, 130 Frigon, R. P., 141,209 Frischauf, A.-M., 81, 83, 92, 94, 95, 97, 105, 106, 113,127 Frolov, E. N., 347,360 Fronk, E., 155,208 Fujiwara, K., 148,211 Fuller, G. M., 135,208 Fuller, S., 125, 127 Fulton, C., 140, 208

368

AUTHOR INDEX

Funatsu, G., 8, 9, 14, 19, 20, 23, 34, 39, 40, 43, 56,62,63, 77 Funatsu, M., 8, 14, 19, 20, 23, 34, 39, 43, 63, 77, 262,263,359 Fyhn, H . J., 224(34), 227(34), 240 Fyhn, U. E. H., 224(34), 227(34), 240

0 Gaffney, B. J., 90, 131 Gahmberg, C. G., 90,96, 103,127,129, 130, 132 Garlick, R. L., 224(34), 227(34), 240 Gamier, J., 144,208 Garoff, H., 80, 81, 82, 83, 84, 87, 88, 89, 90,91,92, 93, 94,95, 97, 103, 105, 106, 107, 108, 109, 110, 111, 113, 117, 119, 120, 121, 124, 127. 128, 130,131,132 Garrett, R., 18, 22, 51 Garrett, R. A., 13, 22,43, 44, 49,51, 55 Gaskin, F., 138, 185, 187,208, 211 Geahlin, R. L., 145,208 Gekko, K., 177,209 Gelfand, G. I., 158, 209 Georgalis, Y., 21, 22, 49 George, H. J., 141,208 Gerard, D., 12,49 Gewitz, H. S., 32,33,54 Ghosh, H. P., 110,132 Ghosh, N., 44,49 Giardina, B., 218(13, 19), 221(115), 227(52), 239, 240, 241, 242 Gibbons, I., 138,210 Gibbons, I. R., 155,208 Gibson, G . H., 218(14), 239 Gibson, Q. H., 218(21, 22), 224(34), 240 Gibson, R., 117, 127 Gilbert, W. A., 334, 335, 337, 338,359 Gillen, G . R., 221(30), 240 Gdlen, R. J., 218(20), 221(28), 240 Ginzburg, I., 44, 49 Giorginis, S., 6, 23, 47, 71, 78 Giovenco, B., 218(19), 240 Giri, L., 19, 21, 22, 23, 49 Glanville, N., 92, 105, 106, 108, 127, 129 Glauser, S. G., 237(98), 242 Glitz, D. G., 28,50,52,53 Glotz, C., 25, 28, 49

Godovac, J., 232(63), 241 Gogia, Z. G., 10, 54 Gogia, 2. V., 14, 18, 19, 22, 49, 50, 52 Goldansltii, V. I., 347,360 Goldstein, J. L., 100, 119, 125, 127 Goldstein, L., 307, 359 Golinska, B., 40,51 Gomez-Fernandez, J. C., 96, 126 Goni, F. M., 96, 126 Goodman, M., 214(2), 239 Gorbunoff, M. J., 208 Gorelic, L., 40, 49 Gorinsky, 8. A., 335,358 Goss, D. J., 225(45). 240 Gottesman, M., 273,360 Gottschalk, A., 95, 129 Gould, E. A., 103, 126 Gould, R., 180,208 Graffunder, H., 5,54 Grannett, S., 151,207 Gratzer, W. B., 10, 48, 50 Graves, D. J., 270,359 Grebenko, A. I., 236(123), 243 Green, J., 81, 111, 113, 115, 119, 122, 127 Greengard, P., 156, 158,211 Greenfield, N. J., 10, 48 Greeves, M. A., 327, 359 Grenader, A. K., 31,52 Greuer, B., 7, 8, 9, 20, 22, 23, 34, 39, 40, 43, 44, 56, 58, 73, 74, 76, 77, 78 Griffith, L. M., 156, 158,208, 210 Griffiths,G., 81, 111, 113, 114, 115, 116, 118, 119, 121, 122,126,127,130 Grimley, P. M., 105, 116, 127 Grimstone, A. V., 149, 208 Grisham, L. M., 154, 163, 206, 211 Groene, A., 44,51 Gronenborn, A. M., 221(115, 116), 242 Gros,F., 273,358 Gros,G., 228(60), 229(61), 231(61), 241 Grujic-Injac, B., 220(24), 240 Gualerzi, C., 9, 40, 48, 52 Gudkov, A. T., 14, 21,49, 53 Guinand, S., 295,359 Gulik, A., 18, 22, 49 Gunsalus, I. C., 251, 253,359, 360, 361 Gupta, R., 25,51,54 Gutell, R. R., 25,51, 54

AUTHOR INDEX

Haaktela, K., 94, 130 Haas, D. J,, 282,359 Haas, J., 29, 31,53 Haedrich, R. L., 217(10), 239 Haegeman, G., 83,126, 127 Haimo, L. T., 151, 152, 155,208,211 Haines, H. B., 227(53), 241 Hakimi, J., 127 Haley, B. E., 145,208 Hall, B. H., 235(90), 242 Hall, F. G., 233(75), 241 Hall, J., 38, 49 Hall, J. G., 235(91), 242 Hall, M. D., 115, 127 Hallak, M. E., 157,208 Hallewell, R. A., 103, 130 Hamel, E., 273,359 Hamilton, M. N., 233(72), 241 Hampl, H., 3, 54 Hanecak, R., 83, 128 Hanson, A. W., 361 Hardesty, B., 38, 51, 53 Hardman, K. D., 324, 337,361 Hardy, G. W., 236(121), 242 Hardy, S.J. S.,22, 44,52 Harker, D., 335,360 Harkness, D. R., 231(114), 242 Harkness, G. R., 221(29), 240 Harley, B. S., 235(88), 242 Harmsen, A., 164, 167, 168,210 Harpaz, N., 112, 127 Harrap, K. A., 99, 131 Harris, J. I., 234(87), 236( 118), 239(87), 242

Harris, R., 220(50, 132), 240,243 Harrison, S. C., 82, 96, 120, 127, 129, 132

Harison, T. M., 79, 129 Hartman, F. C., 282,358 Hartmann, H., 320, 346,350,352,353, 359

Haser, R., 236(121), 242 Hashimoto, K., 109, 115, 117, 118, 120, 121,127,128,130

Hauri, H.-P., 110,127 Hawkins, D. J., 155,208 Hayashi, K., 262, 263,359

369

Hayes, D., 40, 49 Hearst, J. E., 28,53, 54 Heidemann, S. R., 152,208 Heiland, I., 6, 7, 15, 38, 40, 50, 73, 77, 78

Held, W. A., 45, 46,51 Helenius, A., 8 1, 87, 88, 89, 90, 9 1, 92, 93, 97, 99, 100, 101, 102, 103, 127, 128,129,131,132 Hellmann, W., 27, 48 Hendrickson, W. A., 327,360 Hettunen, M.-L., 96, 129 Hill, T. L.,200,208 Hill, W.E., 27, 29, 49, 53 Hilpert, P., 233(76), 241 Himes, R. H., 139, 148, 152,207, 208, 210 Hinkley, R. E., 148,207 Hinz, H.-J., 208 Hirschberg, C. B., 128 Hitz, H., 5, 8, 14, 19, 20, 34, 40, 43, 44, 49, 60, 77 Hoessli, D., 119, 131 Hofer-Warbinek, R., 209 Hofrichter, J., 165,208 Hogan, J. J., 25,54 Holland, J. J., 99, 129 Holmes, R. R., 339,359 Honig, L. S., 164,209 Hood, L. E., 92, 125 Hoofd, L. J. C., 234(82), 241 Horvath, C., 307,359 Horzinek, M. C., 92, 99, 126, 129 Hosaka, Y.,99, 128 Huang, K., 37,48 Huang, R. T. C., 103,128 Hubbard, A. L., 101,126 Hubbard, S.C., 111, 112, 113,128, 130 Huber, R., 324, 348, 350, 351, 353,358, 359,361 Hui Bon Hoa, G . , 248,249,253,259, 286, 294, 295, 297, 302, 303, 304, 305,313,315,358,359,360 Hunkapiller, M. W., 92, 94, 113, 125, 130, 257,360 Hunter, H. S., 95, 130 Hutchinson, V. H., 227(53), 241 Hwo, S., 140, 155, 156,211

370

AUTHOR INDEX

I

K

Ibel, K., 29, 31,53 Ide, G., 167,207 Ikeda-Saito, M., 343,360 Ikehara, Y.,119, 129 Imai, K., 215(7), 216(7), 218(17), 233(74), 239,240,241 Imoto, K., 262, 263,359 Imoto, T., 236(125), 243, 313,360 Inagami, T., 324, 337,361 Ingerman, R. L., 2 17(1 I ) , 239 Inglis, A. S.,110, 126 InouL., S., 138,208 Isaacks, R. E., 221(29), 240 Isacks, R. E., 231(114), 242 Ishimura, Y.,251,360 Isono, K., 8, 9, 20, 23, 34, 38, 43, 49, 52, 57,77 Israelachvili,J. N., 96, 128 Ivatt, R. J., 111, 128

KaPriiiinen, L., 80, 81, 82, 83, 92, 94, 95, 96, 103, 104, 105, 106, 107, 108, 109, 111, 115, 117, 118, 119, 120, 121, 127,128, 129,130, 131, 132 Kahan, L., 35,54 Kakiuchi, S.,155,208 Kalas, S., 232(65), 241 Kalkkinen, N., 83, 92, 95, 113, 128 Kaltschmidt, E., 4, 49 Kaluza, G., 109, 113, 114, 128 Kamp, R., 7, 8, 9, 14, 21, 23, 34, 56, 62, 75, 77, 78 Kamp, R. M., 6, 8, 21, 23, 40, 69, 77 Kanwisher, J. W., 224(36), 240 Kaplan, J., 80, 100, 125 Kar, D., 330, 335, 337, 356,359 Kar, E. G., 21, 22, 49 Karr, T. L., 139, 146, 148, 149, 166, 170, 171, 172, 173, 174, 175, 178, 179, 181, 187, 188, 190, 198, 201, 204, 205,208,209,210 Kartenbeck, J., 81, 92, 99, 100, 101, 102, 103,128,132 Kartha, G., 335,360 Kasai, M., 159, 160, 198, 204, 208, 210 Kastner, B., 2, 30, 31, 32, 49, 50, 53 Katchalsky. E., 304, 307, 313, 359, 360 Katz, F. N., 80, 93, 108, 109, 128, 129, 132 Kauzmann, W., 225(39), 240 Kay, R. M., 220(46, 132), 240, 243 Kaziro, Y.,9, 48, 155, 173, 174,207, 211 Keegstra, K., 94, 95, 126, 128, 131 Kelley, D. M., 324, 337,361 Kemph, T., 140,209,210 Kendrew, J. C., 236(96), 242 Kennedy, P. L., 25,48 Kennedy, S. I. T., 81, 99, 105, 106, 108, 126, 128,129 Kennett, D., 225(117), 242 Kenny, J. W., 38, 39,53 KerBnen, S.,92, 105, 106, 108, 109, 111, 115, 117,127,128, 129,130,131 Khavitch, G . , 32, 33, 55 Khechinashvili, N.,N., 14, 18, 22,50, 52 Kihara, T., 232(62), 241 Kilmartin, J. V., 215(6a), 216(6a, 44), 225(42, 44), 228(104), 239, 240, 242

J Jacobs, M., 173,208 Jacrot, B., 82, 83, 96, 128 Jameson, L., 189,208 Jelkmann, W., 228(58, 59), 231(59), 233(81), 234(85), 237(100, 101), 241, 242 Jencks, W. P., 334,360 Job, D., 157,208, 251,361 Johansen, K., 224(32), 227(55), 240, 24 1 Johnson,D.C., 118, 119, 120, 121,128 Johnson, F. H.,277, 278,360 Johnson, K. A., 139, 148, 151, 152, 155, 156, 163, 164, 173, 178, 207, 208. 210 Johnson, L. N., 236(125), 243, 313, 324, 337,338,360,361 Johnson, R. M., 282,358 Johnson, R. T., 125,132 Jones, K., I32 Jones, K. J,, 113, 119, 128 Jones, M. D., 9, 48 Jdmvall, H., 83, 92, 95, 128 Joysey, K. A., 229(61), 231(61), 241 Jukes, T. H.,214(3), 239 Jumblatt, J., 82, 96, 127

37 1

AUTHOR INDEX Kim, H. D., 139,207, 208, 221(29), 240 Kime, J. M., 13, 14,50 Kimura, M., 6, 8, 9, 15, 20, 23, 34, 38,43, 57, 66, 67, 71, 50, 52, 77, 78 King, J. L., 2 14(3), 239 Kirkwood, J. G., 286,360 Kirsch, M., 145,208 Kirschner, M. W., 140, 141, 143, 144, 145, 146, 153, 154, 155, 156, 157, 163, 164, 173, 174, 180, 200, 206, 207,208,209,210,211 Kitamura, N., 83,128 Kleinschmidt, T., 234(85), 241 Klenk, H.-D., 103,128 Klotz, I. M., 278, 360 Klug, A., 144, 149, 158, 172,207, 208 Knowles, J. R., 360 Knox, J. R., 361 Kobayashi, T., 157, 173,208, 209 Kobylov, A. M.,25, 51 Koch, M. H. J., 29, 31, 53 Koehn, R. K., 235(91), 242 Kohls, H., 5,54 Kojouharova, M. S., 52 Kondor-Koch, C., 81, 84, 94, 103, 105, 110, 117,127, 128 Konnert, J. H., 327, 360 Koo, R., 99, 131 Kop, J., 25,51, 54 Korn, A. P., 30, 50 Kornfeld, S., 111, 112, 117, 127, 129 Koshland, D. E., 277,360 Kossel, H., 25, 28, 48, 49 Kotani, S.,210 Koteliansky, V. E., 14, 27, 31,50, 52, 54 Kotin, R., 330, 335, 337, 356,359 Kozak, M., 105, 129 Krarner, G., 38,51, 53 Kratky, O., 16, 50 Krauhs, E., 140,209, 210 Krause, E., 110, 129 Kraut, J., 254,360 Kreibach, G., 110, 130 Kreil, G., 110, 129 Kress, Y., 32, 50 Kreuzer, F.,234(82, 83), 241 Krick, J., 38, 55 Kristofferson, D., 148, 170, 171, 172, 176, 178, 181, 189, 193, 198, 199, 204, 208,209,210

Krol. A., 25, 48 Krombach, C., 233(81), 241 Kronenberg, H., 225(117), 242 Kiihlbrandt, W., 32,50 Kuismanen, E., 125, 130 Kumagai, H., 155,209 Kurland, C. G., 15, 38, 39, 40, 44, 48, 50, 51

Kussova, K. S., 40, 53, 54 Kuwaki, T., 210 Kuznetsov, S. A., 158, 209

1 Labischinski, H., 19, 21, 22, 49, 50 Lachmi, B., 92, 105, 106, 108, 129 Laidler, K. J., 277, 360 Laine, R., 81, 96, 129 Lake, J. A., 2, 30, 31, 32, 35, 48, 50, 54 Lam, M. K. T., 44,50 Lambert, J. M., 38, 39, 50, 53 Lampen, J. O., 110,131 Lampert, P. W., 99, 129 Lang, E. M., 233(76), 241 Lange, R., 253, 267, 360 Langer, J. A., 37, 51 Langford, G . M., 204, 205,207 Langlois, R., 38, 50 Larroque, C., 296,360 Larsen, G . R., 83, 128 Laue, T. M., 156, 158,210 Lauffer, M. A., 134, 209 Laughrea, M., 13, 19, 21,50, 51 Laursen, R. A., 6, 8, 9, 14, 15, 20, 22, 23, 29, 38, 40, 43, 44, 48, 56, 70, 77, 78 Laustriat, G., 12, 49 Laver, W.G., 110, 126 Leautey, J., 273,358 Leavitt, R., 117, 129 Leclercq, C., 229(61), 231(61), 241 LeDoucen, C., 293,358 Lee, B.,361 Lee, C. C., 38, 50 Lee, J. C., 139, 140, 141, 144, 177, 187, 208,209 Lee, J. J., 83, 128 Lee, L.Y., 209 Lee, S.-H., 176, 189, 209 Lee, Y. C., 155,209 LeGall, J. Y., 40, 51

372

AUTHOR INDEX

Lehmann, A., 8, 20,22,23,40,43,77 Lehrach, H.,81,83,92,94,95,97,105, 106, 113,127,130 Lehtovaara, P., 105,129 Leijonmarck, M., 15,22,23,50 Leloir, L. F., 94,95, 108, 1 1 1 , 131 Lemieux, G., 12,49 Lenard, J., 80,125, 129 Lenfant, C.,227(55),241 Leonard, K. R., 32,33,48,50,82,127 Leopold, H.,16,50 LePeuch, C., 295,360 Lerclecq, F., 230(113),242 Lerf, A., 350,358 Lerner, R. A., 83,131 Lesk, A,, 236(95),242 Leterrier, F., 250,358 Levin, J. G., 105,127 Levin, Y.,307,359 Lewis, H., 9,52 Lewontin, R. C., 213(1),229(1),239 Li, E., 111, 112,129 Lienhard, C . E.,338,360 Lietzke, R., 23,34,37,51 Lifsics, M. R., 156, 158,210 Liihrmann, R., 2, 32,50,53 Liljas, A., 44,51 Liljas, H., 15,22,23,50 Lilley, G. G., 110,126 Lim, T.W.,116,I29 Lim, V. I., 14, 19,49 Linck, R. W., 144, 149, 172,207,209 Lindemann, H., 8, 19,43,63,77 Lingappa, V. R., 80, 109. 128,129 Lipscomb, J. D., 251,359 Lipscomb, W. N., 361 L'Italien, J., 9,48 Litman, D. J., 44,50 Little, M., 140,209,210 Littlechild, J. A., 2, 3, 9,10, 13, 14, 18, 19,21, 22,38,49,50,51, 54 Liverzani, A., 220(26),221(26),240 Lockwood, A. H.,140, 155, 156,211 Lodish, H.F., 80, 93, 105, 108, 109, 110, 117,124,126,128,129, 130,132 Lonberg-Holm, K., 99,129 Lopata, M. A,, 141,207 Lorenz, S., 27,32,33,53,55 Louvard,D.,81, 111, 113, 115, 116, 119, 122,127,129

Low, P. S., 277,360 Lu, R. C., 140,209 Lubin, M., 132 Luduena, R. F., 140,143, 144,209 Ludwig, M. I., 236(127),243, 345,361 Luer, C. A., 50 Lumry, R., 275,276,360 Lunadel, M., 227(52),241 Lunkkonen, A., 94,95,129 Lutter, L. C.,38,39,50 Luukkonen, A., 96,129 Lykkeboe, C., 222,240 Lyon, H.D.,139, 157,210

M Maassen, J. A., 38,40,50,55 McCarthy, B.J., 141,211 McCarthy, M., 120,129 Maccioni, R., 200,209 McDonald, J. F.,235(92),242 McDonald, M. J., 243,244(129) MacDonald, R. J., 141,207 McEwen, B.,149, 150, 152,209 McGill, M., 151,209 Machatt, M. A., 25,48 McIntosh, J. R., 152,208 McIntosh, K., 125,129 McKenzie, A. P., 318,361 Mackie, G. A., 43,55 McKuskie-Olsen, H., 28,50 McLaren, A. C., 307,360 MacNeal, R. K.,145, 147, 173,174, 196, 209,210 McPherson, A., 32,51 McPhie, P., 10,50 McWhirter, P. D., 40,52 Madoff, D.H., 125,129 Maeda, T.,103,129 Magee, A. I., 113, 129 Magrum, L. J., 25,54 Mair, G . A.,313,358 Malcolm, A. L., 13, 14,50 Malefyt, T.R., 170, 171,209 Maley, F., 113,131 Maly, P., 25, 26,40,41,42,43,44,48,50, 54 Mandekow, E., 148, 152,164, 167,168, 209,210

373

AUTHOR INDEX Mandekow, E.-M., 152, 164,167, 168,210 Marchesi, V. T., 93,132 Marcum, J. M., 139, 151, 152, 155,156,

164,207,210 Margolis, R. L., 157,194, 197,200,201,

202,203,204,206,210,212 Markova, L. F., 6,7,8,9,20,21, 23, 34,

56,67,74,77, 78 Marsh, D. J., 323,360 Marsh, M., 100, 101, 102,128, 129 Marshall, R. D., 95,129 Martensen, T.M., 157,208,210 Martin, S. R., 103,131 Martini, F., 220(26, 106),221(26),240,

242 Maruyama, T., 220(110, 111).242 Marzinzig, E., 7,8, 20,22,23,40,43,73,

75,77, 78 Mathews, M. B., 79,129 Matlin, K., 103,125,127,129,I32 Matsuda, G.,232(62),241 Mattila, K.,94,129 Mattila, N., 94,95,129 Maurel, P.,248,249,286,294,296,297,

307,313,315,358,359,360 Maxfield, F. R., 101,132 May, R. P.,23,34,37,51 Mayo, K. H.,218(18),240 Mazumdar, S. K.,335,358 Mazur, G.,221(23a),240 Mednikova, T. A., 7,21,23,74, 78 Meehan, P.,257,359 Meek, R. C.,110,129 Meeks, J. R., 251,359 Meinke, M., 40,43,52,54 Melik-Adamyan, W.R., 236(123),243 Melin, M., 360 Mellon, M. G.,139,210 Mende, L.,6,7,8,9, 14, 15,23,34,43,

Milstein, C., 79,129 Misra, L.,141,208 Mitchell, D. J., 96,128 Mochalova, L. V.,27,5I,52 Mohri, H.,134,210 Mliller, K.,40,51 Mliller, W.,6,8,9,14, 15,20,22,23,29, 38,40,43,44,50,51, 55,56,70,77,

78 Moore, P. B., 13, 14, 18, 19,21, 31, 34, 37,44,49,50,51,52 Mooseker, M. S., 148,211 Moras, D.,236(1 19),242 Morein, B., 99,128 Morgan, J., 43,5I Morimoto, T.,110,130 Morinaga, T.,8, 14, 19,20,23,34,39,

43,63,77 Morris, R. J., 218(21,22), 240 Morrison, C.A., 13, 14,5I Morser, J., 108,127 Mosca, A., 228(60),241 Moss, D.A,, 335,358 Mossbauer, R. L.,347,360 Muchamedganova, E. V.,40,53,54 Mueggler, P.A.,237(99),242 Muirhead, H.,237(98),242 Muranova, A. V., 6,8, 9,20,23,34,56,

67,77 Muranova, T.A.,6,8,9,20,23,34,56,

67,77 Murphy, D.B., 139, 140,148, 152,154,

155,156,207,210,211 Murphy, F. A., 82,83,129 Murthy, M. R., 236(122),242 Mussgay, M., 99,129 Muto, A., 43,55

50,66,70,75,77, 78 Merrett, M., 236(121),242 Metcalfe, J., 233(76),241 Meyer, D. I., 110,129 Meyer, F. R., 233(77),241 Michalski, C.J., 44,45,51 Miiller, R., 44,49 Miinck, E., 251,359 Miissig, J., 32,33,54,55 Miller, D. L.,9,48 Millon, R., 40,51

N Nagai, Y.,199,200,211 Nagano, T., 148,210 Nagarkatti, S., 9,48 Nagel, K.,273,361 Nakanishi, S.,273,360 Nash, A. R., 220(33, 112),224(33),240,

242 Natzle, J., 141,211 Neet, K.E., 277,360

374

AUTHOR INDEX

Neuberger, A., 95, 129, 313,358 Neuberger, M. S., 235(88), 242 Neufeld, E. F., 116, 129 Newcomer, M. E., 15,50 Newton, J., 44, 51 Nickameyer, W. S., 218(21), 240 Nielsen, K. M., 9, 48 Nierhaus, K. H., 3, 23, 27, 34, 37, 43,44, 45, 46, 47, 49, 51,52,54

Nishida, E., 155, 209,210 Nishikura, K., 215(6a), 216(6a), 239 Noah, M., 36,51, 53 Noble, R. W.,218(14, 16). 224(34), 227(34), 239, 240

Noller, H. F., 25, 28, 40, 44, 45, 48, 51, 52,54

Nomura, M., 9, 45, 46. 51, 52, 53 Nordstrbm. B., 235(94), 242 North, A. C. T., 236(125),243, 313,358,

Ovchinnikov, Y.A., 6, 7, 8, 9, 20, 21, 23, 34, 56, 67, 74, 77, 78 Overbergh, N.,165, 182,207

P Page, M. I., 334,360 Palade, G. E., 80, 114, 118, 122, 123, 126, 129

Palmer, M. L., 25, 48 Palmer, R. A., 335,358 Palmiter, R. D., 110, 129 Pannenbecker, R., 7,23, 76, 78 Pantaloni, C., 295, 359 Pantaloni, D., 166, 174, 175, 188,207, 210

Paradies, H. H., 18, 21, 22,51, 54 Parak, F., 320, 346, 347, 350, 352, 353, 359,360

360

Nowotny, V.,23, 34, 37,51

0 Oberthur, W., 232(64, 65), 241 OBrien, L., 32, 51 EBrien, S., 18, 21, 22, 44,52 Oda, K., 119, 129 Odom, 0. W., 38,51,53 Oechsner, I., 6, 8, 20, 23, 29, 34, 66, 77 Ohgi, K., 28, 53 Ohkuma, S., 101,130 Ohlsson, I., 235(94), 242 Ohmori, D., 236(126), 243, 328,361 Ohnishi, A., 103, 129 Okon, M. S., 14,53 Oldstone, M. B. A,, 99, 129 Olins, P. O., 9, 51 Olmsted, J. B., 139, 151, 152, 155, 156, 157, 162,206,207,210

Olomucki, M., 40,51 Oosawa, F., 159, 160, 161, 192, 198, 204, 208,210

Olsen, K. W., 236(119), 242 Op den Kamp, J., 132 Osguthorpe, D. J., 144,208 Oskarsson, A., 348,358 Osterberg, R., 18, 19, 22, 44,51 Osterrieth, P. M., 90, 129 Ota, Y.,232(62), 241

Parfait, R., 29, 31,53 Parkhurst, L. J., 218(14), 225(45), 239, 240

Parmeggiani, A., 273,361 Pascher, I., 96, 129 Pastan, I. H.,100, 129, 273, 360 Paterakis, K., 13, 14, 50 Patient, R. K., 220(46, 132), 240, 243 Pattridge, K. A., 236(127), 243, 345,361 Patzer, E. J., 96, 130 Pauli, G., 109, 128 Pauling, L.,135,210, 338,360 Pearse, B. M. F., 99, 101, 123, 130 Pederson, T., 361 Peeters, P., 8, 22, 43, 77 Pendergast, M., 125,130 Penningroth, S . M., 145, 146, 173, 174, 210,211

Penttinen, K., 115, 118, 120, 128 Perella, M., 228(60), 241 Perutz, M. F., 215(6a), 216(4, 6a, 44), 219(23), 221(23), 222, 225, 229(61), 232(4), 233(74), 234(86), 236(96), 237(98), 239, 240, 241, 242 Pesonen, M., 94. 95, 108, 111, 117, 118, 119, 121, 125,128, 130 Pestka, S., 38, 50 Peters, R., 120, 130 Petersen, T. E., 9, 48 Peterson, J. A., 251,360

375

AUTHOR INDEX Petruzelli, R., 220(26, 106, 130, 131), 22 1(26), 240, 242,243 Petschow, D., 232(66), 241 Petsko, G. A., 236(126), 243, 259, 280, 281, 282, 317, 320, 321, 323, 324, 328, 330, 332, 334, 335, 336, 337, 338, 343, 344, 345, 346, 348, 350, 352,353, 354,358,359,360,361 Pettersson, I., 22, 44,51, 52 Pettersson, R. F., 125, 127, 130 Pfefferkorn, E. R., 95, 111, 117, 120, 121, 123,126, 130,132 Pfeil, J. A., 119, 128 Phelps. C., 224(34), 227(34), 240 Philipson, L., 99, 105, 129 Phillipe, R., 284, 361 Phillips, A. W., 236( 121), 242 Phillips, D. C., 236(125), 243, 258, 313, 354,358, 360,361 Phillips, S . E. V., 343,361 Piefke, J., 32, 33, 54 Pierson, G. B., 148,207, 210 Piette, A. M.,284, 361 Pilkis, S. J., 210 Pinter, A., 114, 130 Podrasky, A. E., 174, 175, 188, 190, 205, 208 Politz, S. M.,28, 40, 52 Pollack, R., 135,211 Pollard, T. D., 156, 158, 208, 210 Pon, C., 9,52 Ponstingl, H., 140, 209, 210 Poole, B., 101, 126, 130 Porath, J., 293,361 Porter, A. G., 103, 130 Porter, M.,117, 124, 129, 132 Post, L. E., 9, 52 Pouyet, J., 10, 25, 48, 49 Powers, D. A., 221(27), 224(34), 227(34), 240 Powers, T. B., 236(127), 243 Prakash, V., 44, 52 Price, B., 208 Privalov, P. L., 14, 52 Provencher. S. W., 10,49 Pryor, A. W.,345, 346,361 Purich, D. L., 137, 139, 144, 145, 146, 147, 148, 149, 150, 151, 154, 155, 166, 170, 171, 172, 173, 174, 175, 176, 178, 179, 181, 187, 188, 189,

190, 193, 196, 197, 198, 199, 201, 203, 204, 205, 206,207, 208, 209, 210,211

Q Quaroni, A., 110,127 Quigley, J. P., 96, 130 Quilliam, T. A., 234(84), 241 Quinn, P., 81, 111, 113, 115, 116, 118, 119, 121, 122,127,130 Quiocho, F. A., 361 Quiroga, M., 140, 143, 144, 153,211

R Rahn, H., 227(54), 241 Raidt, H., 234(86), 241 Rajender, 275,276,360 Ramakrishnan, V. R., 18, 34, 37, 51, 52 Ranki, M., 105, 127, 130 Ranney, H. M., 215(5), 234(5), 239 Rasilo, M.-L., 95, 130 Rasmusse, D. H., 318, 361 Ratcliffe, R. G., 13, 14, 50 Rauch, C. T., 157,208 Rebhun, L. I., 139,210 Reggio, H., 103, 115, 127, 129 Reich, E., 96, 130 Reid, R., 175, 181, 185, 187, 189, 198, 199,202, 203,204,212 Reid, T. J., 236(122), 242 Reinbolt, J., 8, 14, 18, 20, 22, 34, 38, 39, 40, 43, 44, 58, 60, 77 Reinhardt, R., 15, 48 Reinke, K., 156, 158,210 Reithmeier, R. A. F., 8, 34, 43, 63, 77 Renaud, F., 138,210 Renkonen, O., 80, 81, 82, 94, 95, 96, 108, 127,128,129,130 Rheinberger. H. J., 32, 33, 54 Ricard, J., 251,361 Rice, C. M.,81, 83, 88, 91, 92, 94, 95, 105, 106, 113, 125,130 Rice, D. W.,236(121), 242 Richards, F. M., 324, 337, 339,361 Richards, J. H., 257,360 Richardson, C. D., 91, 103, 118, 130 Rickli, E. E., 110, 127

376

AUTHOR INDEX

Riedel, H., 81,84,92,94, 105, 106, 110, 127, 130 Riegel, K., 233(76),241 Rifbin, D. B., 96,130 Riggs, A.,218(20),220(109, 110, 1 1 l), 221(28,301, 224(31),226(47,49), 240,242 Rignalda, B. E. M., 234(83),241 Ringe, D.,236(126), 243,328,361 Ringel, I., 170, 187,202,211 Ringe Ponzi, D.,320, 346, 350,352, 353, 359 Rinke, J., 40,43,44,50, 51, 52 Ritter, E.,7, 14,21,23,78 avers, P. S., 354,358 Robakis, N.,27,48,52 Robbin, D.,38,551 Robbins, R. W.,112,113,128, I30 Roberts, G . C. K., 361 Robinson, J., 167,207 Robinson, P., 99,128 Robinson, S. M. L., 11, 49, 54 Robson, B.,144,208 Rodewald. K.,220(57),221(25),227(57), 240,241 Rodionov, V. I., 158,209 Rodriguez, A,, 125,130 Rodriguez, J. A,, 157,208 Roerning, R., 8,20, 22,34,39,44,58,77 Rogiers, R., 83,126, 127 Rohde, M. F., 18,21,22,44,52 Rbhl, R., 46,47,52 Rohrschneider, J. M.,113,131 Rollema, H.S.,215(6a),216(6a),228(60), 232,293(80),239,241 Rombauts, W.,6,8, 14,19,22,23,43,64, 72,77, 78 Rommel, W.,40,49,52 Rose, D.,343,360 Rosenbaum, J. L., 148, 151, 152,154, 155,156, 158,207,208,210,211 Rosenbaum, R. M.,32,50 Rosenblat, V. A., 158,209 Ross, P. D.,165,208 Rossi-Bernardi, L.,228(104),242 Rossman, M.G., 236(119, 122),242 Roth, H.E.,43,46,47,52 Roth, J., 116,130 Rothberg, P. G.,83,128

Rothman, J. E.,80, 109, 119, 123, 126, 127,128,130 Rott, R., 103, 114,128 Rowe, A., 138,210 Roxburgh, C.M.,110,126 Roy, C.,8,34,39,43,44,63,77 Rudrnan, R., 322,323,361 Rudolph, S.A., 156, 158,211 Rund, J. T.,225(41),240 Runge, M. S., 156, 158,210 Runswick, M.J., 234(87),239(87),242 Rupley, J. A., 236(125),243, 264,302, 313,358,360,361 Rutter, W.J., 140, 141,143, 144,153, 207.21I

S Sabatini, D. D., 110,130 Sagik, B. P., 119,128 Saito, A., 156, 158,210 Sakai, H., 155,209,210 Salinonova, 0.M.,27,30,31,54 Salmon, E. D.,211 Salsbury, A. J., 234(84),241 Salton, M.R. I., 313,361 Sanchez, F., 141,211 Sander, G., 273,361 Sandoval, I. V., 145,203,211 Sanger, F., 25,48 Santos, M.,235(92),242 Saraste, J., 109, 115, 117, 118, 120, 121, 127,128,130 Sarkar, P. K.,10.52 Sawicki, D.L., 105,130 Sawicki, S. G.,105, 130 Schachter, H.,112,127 Schaefer, D.,8, 14, 19,20,34,40,43,44, 60,77 Schiifer, D., 5, 49 Schechter, N.,44,52 Scheele, C.M.,111, 130 Scheele, R., 180,211 Scheele, R. B., 139,211 Scheid, A., 103,130 Schekrnan, R., 116,130 Schiltz, E.,6, 7,8, 14, 18,20,22,23,34, 38,39,40,43,44,58,69,72,77, 78

AUTHOR INDEX Schindler, D. G., 18, 34, 37,51,52 Schirrmacher, U., 99, 128 Schlesinger, M. J., 81, 105, 107, 108, 109, 111, 113, 118, 119, 120, 121, 125, 126,128,129, 130, 131 Schlesinger, R. W., 81, 99,130, 131 Schlesinger, S., 81, 108, 109, 1 1 1, 1 17, 127,129,130 Schmiady, H., 27,53 Schmidt, M. F. G., 113, 117, 119, 130, 131 Schnek, A. G., 229(61), 230(113), 231(61), 241,242 Schnier, J., 8, 9, 20, 23, 34, 38,43,52, 57, 77 Schoenborn, B. P., 37,51 Schop, E. N., 40,50 Schrank, B., 230(113), 233(78, 79, 81), 241,242 Schulze, H., 3 , 23, 34, 37,51, 54 Schwarz, R. T., 108, 111, 114, 117, 127, 128,131 Seeds, N. W., 200,209 Seelig, A., 96, 131 Seelig, J., 96, 131 Sefton, B. M.. 90, 92, 94, 95, 106, 108, 128,131,132 Segrest,J. P., 93, 131 Seib, C., 7,47, 78 Selden, S. C., 156, 158,210 Sells, B. H., 44, 45,51 Semenza, G., 110,127 Semler, B. L., 83, 128 Serdyuk, I. N., 14, 18, 19, 22, 29, 31,49, 52 Shapiro, L. J., 116,129 Sharma, V. R., 313,358 Shatsky, I. N., 27, 28, 51, 52 Shatsky,J. N., 40,53, 54 Shelanski, M. L., 138, 141, 144, 173, 185, 187,207,208,211 Shelley, K., 32,51 Shelton, G.,,227(56), 241 Shimizu, K., 99, 128 Shinnik, T. M., 83, 131 Shooter, E. M., 144,209 Shope, R. E.. 145,126 Shpungin, J. L., 29,52 Shullman, R. G., 255,361

377

Sicignana, A., 236(122), 242 Sieber, G., 27, 45,52 Siegel, R. B., 25,54 Sillers, LY.,18, 34, 37, 51, 54 Silverstein, S. C., 80, 100, 131 Simmons, D. T., 108, 109,131 Simon, C., 166, 210 Simons, K., 80, 81, 82, 83, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 99, 100, 101, 102, 103, 105, 106, 107, 108, 109, 110, 111, 113, 115, 119, 120, 121, 124, 125, 127, 128, 129, 130, 131,132 Simpson, K., 23, 34,51 Simpson, P. A., 140, 208 Singer, S. J., 81, 115, 125, 306,361 Singh, T. P., 324, 348, 350, 351, 353, 361 Sireix, R., 249, 250, 358 Sjdberg, B., 18, 19, 22, 44,51 Sjalin, L., 334, 335, 337, 359, 361 Skehel, J. J., 103, 126, 131, 132 Skdld, S. E., 40, 48 Slobin, L. I., 39,52 Sloboda, R. D., 148, 154, 155, 156, 158, 211 Sly, W. S., 116, 131 Smale, C. J., 92, 126 Small, B., 93, 108, 109, 132 Smilowitz, H., 119, 131 Smith, A. E., 105, 127 Smith, H., 173,208 Smith,J. F., 117, 121, 131 Smith, W. S., 29, 52 Smith, W. W., 345,361 Snell. W. J., 151, 211 Snider, M. D., 117, 132 Snyder, D. H., 148,211 Snyder, K. B., 200,212 Sliderlund, G.. 235(94), 242 Sliderlund, H., 81, 82, 83, 91, 92, 93, 95, 96, 97, 104, 105, 106, 107, 108, 109, 127,128, 129,131, 132 Sokol, F., 114, 131 Somero, G . N., 277,360 Sommer, A., 38,52 Spiegelman, B. M., 146, 154, 173,207, 211 Spiess, E., 30, 52

378

AUTHOR INDEX

Spirin, A. S., 14, 18, 19, 22, 29, 30, 31, 49, 52, 54 Spiro, R. G., 95, 129 Spitnik-Elson, P., 30, 44, 50, 52, 53 Stabinger. H., 16, 50 Stahl, D. A., 25, 51, 54 Stahl, K., 348,328 Stallings, W. C., 236(127), 243 Staneloni, R. J., 94, 95, 108, 111, 131 Stangl, A., 220(24), 230(113), 233(78, 79, 81), 240, 241, 242 Steams, M. E., 211 Steen, J. B., 217(8), 239 Steigemann, W., 320, 324, 346, 350, 351, 352, 353,359, 361 Steiner, D. F., 114, 126 SteinhPuser, K. G., 38,49,53 Steinmann, R. M., 80, 100, 131 Steitz, T. A., 255, 361 Sternlicht, H., 170, 187, 202,211 Stiege, W., 28, 53 Stiegler, P., 25, 43, 53, 55 Stockell, A., 237(98), 242 StOffler, G., 2, 3, 30, 31, 32, 35, 36, 38, 40, 43, 48, 49, 50, 51, 53 StOffler-Meilicke, M., 30, 31, 36, 49, 50, 51 Stollar, B. D., 99, 131 Strauss, E. G., 81, 94, 105, 113, 120, 126, 130,131 Strauss, J. H., 81, 83, 88, 91, 92, 94, 95, 105, 106, 108, 109, 113, 120, 125, 126,130,131 Strominger, J. L., 99, 128 Strycharz, G. D., 9, 36, 44, 50,52, 53 Stuhrmann, H. B., 23, 29, 31, 34, 37, 51, 53 Subramanian, A. R., 8, 9, 18, 19, 20, 21, 23, 34, 38, 43, 49, 50, 51, 52, 53, 57, 77 Sullivan, B., 227(51), 233(71), 240, 241 Summers, K., 180,211 Sundell, S., 96, 129 Sunhuesa, S., 15, 48 Sutcliffe, J. G., 83, 131 Suzuki, F., 148,210 Suzuki, K., 236(126), 243, 328,361 Szacz,J., 170,211

T Tabas, I., 111, 112, 129 Taddie, C., 32, 5? Takahashi, K., 9, 48 Takei, H., 232(62), 241 Taketa, F., 233(73), 241 Talbot, P. J., 103, 131 Tam, M. F., 27,53 Tamm, I., 116, 120,125 Tan, A. L., 218(16), 240 Tan, K. B., 114, 131 Tanakas, N., 236(122), 242 Tardieu, A., 29,53 Tarentino, A. L., 113, 131 Tartakoff,A. M., 114, 117, 118, 119, 131 Taylor, E. W., 138, 144, 173,207, 208, 211

Taylor, H. L., 360 Teal, J. M., 224(35, 36), 240 Telzer, B. R., 151, 152,208, 211 Ten Eyck, L. F., 225(43), 240 Tenney, S. M., 252(67), 241 Tentori, L., 220(108), 226(48), 240,242 Teraoka, H., 3, 54 Terhorst, C., 6, 8, 9, 14, 15, 20, 22, 23, 29, 38, 40, 43, 44,51, 56, 77, 78, 99, 128

Terry, B. J., 137, 144, 148, 150, 151, 154, 156, 166, 173, 176, 189, 197, 203, 206,209,210,211 Tesche, B., 2, 27, 32, 33, 45, 50, 52, 53, 55 Tetens, V., 214(129), 244(129), 243 Thammana, P., 28, 5?, 54 Thatcher, D. R., 235(93), 242 Thewalt, U., 240 Thierry, J. C., 236(1 18), 242 Thomas, D. K., 224(38), 240 Thomas, J., 148,209 Thompson, E. 0. P., 220(33, 112), 224(33), 240,242 Thompson, J. D., 29, 49 Thompson, J. F., 28,54 Thompson, W. C., 140, 153, 157, 197, 200,211,212 Threlfall, G., 103, 130 Till, H. W., 233(80), 241 Tilney, L. G.,148,211

AUTHOR INDEX

379

Timasheff, S. N., 138, 139, 141, 144,154, Usher, D. A., 361 163, 177, 185, 187, 190,206,208, Utermann, G.,90,127,132 209,211 Timm, B., 9,50 V Timmermans, J., 284,361 Vaininen, P., 103,132 Tindall, S. H., 44.53 Tischendorf, G. W., 2. 32,53 Vachette, P., 12, 18,22,29,48, 49,53 Vagin, A. A., 236(123),243 Tishon, A.,99,129 Vainshtein, B. K., 236(123),243 Tkasz, J. S., 110,131 Valenzuela, P.,140, 143, 144,153,211 Tokimatsu, T., 44,53 Vallee, R. B., 140, 154, 156,157, 158, Tokuyasu, K.T., 81, 115,125, 127,131 210,211 Tolberg, W.R.,29,53 Vance, D. E., 91,103, 118,130, 131 Tominaga, S.,155,211 Vandecasserie, C., 229(61),231(61),241 Tomita, M., 93,132 Toneguzzo, F.,110,132 Vandekerckhove, J., 8, 14, 19,22,43,64, 77 Towfighi, J., 32,51 Vandenbunder, B., 359 Traub, F.,45,53 van der Werf, S., 83,128 Traut, R. R.,38,39,49,50,52,53 Van de Woorde, A., 83,126,127 Travers, F., 247,248, 249,250, 286,289, Van Hereweghe, J., 83,126,127 290,291,292,294, 358,359,361 Van Heuverswyn, H., 83,126, 127 Trempe, M. R., 28,53 Van Hoof, F., 101,126 Trimble, R. B., 113, 131 van Meer, G.,132 Tritsch, D.,14,20,22,34,39,40,43,60, van Steeg, H., 107,132 77 Varma, M. G.R.,103,126 Tronet, A., 101,126 Vasiliev, V. D., 27, 28,29,30,31,51,52, Tsernoglou, D.,282,324,330,332,337, 54 343,345,346,348,350,358,359, Vasquez, D., 273,358 360,361 Vassali, P., 119,131 Tsuchiya, T., 199,200,ZlI Velani, S.,220(108),242 Tucker, J. C., 318,361 Velmoga, I. S.,6,78 Tulbens, P., 101,126 Ventilla, M., 141,211 Tumanova, L.G.,14,53 Venyaminov, S. Y.,10,21,49,54 Turchinsky, M. J., 40,53,54 Venyaminov, S. Yu.,14, 18, 19,22,49,52 Turco, S.J., 112, 113,130 Villa-Komaroff, L.,105,126 Turek, L.,234(82,83),241 Vince, R.,38,50 Turner, S., 28,54 Vinokurov, L. M.,6,7, 21,23, 74,78 Tweedy, N., 139,144, 177,209 Virtanen, I., 115, 118, 120,128 Tycho, B., 101,132 Vivaldi, G.,220(108),242 Vivaldi, J., 226(48),240 Voelter, W., 232(64),241 U Volcani, B. E., 313,360 Ulmanen, I., 104, 105, 107,127,129,131, Vogel, D., 228(60),241 132 Voigt, J., 273,361 Ulmer, E.,43,50 Volckaert, G.,83,126, 127 Ungewickell, E.,43,55 von Bonsdorff, C.-H., 82,92,97,104, Unwin, P. N. T., 32.50 107, 115, 117, 120,121,130,131, Uomala, P., 105, 108, 127 132 Urbani, L.J., 119,126 Voter, W. A., 21 1

380

AUTHOR INDEX

W Wabl, M. R., 32,54 Wacker, H., 110,127 Wade, M.,9,48 Wagner, R. R., 96,130 Wahl, P., 12,48 Wahn, K., 107,132 Waite, M. R. F., 117, 120, 121,125,126, 132 Walker, J. E., 234(87),236(1 IS),239(87), 242 Wall, J. S.,27,48 Walsh, K. A., 110,129 Walter, C.,114,126 Walter, J., 324,350,351,353,361 Walter, P.. 110,132 Wan, K., 119,128 Wang, J. H.,270,359 Warren, C.,81,94, 111, 113, 114,115, 116,118, 119, 121, 122,126, 127, 129,130,132 Waterfield, M. D., 103,131 Watson, H.C.,236(96),242 Watson, J. C.,236(120),242 Watt, K.W. K., 220(110, 1 1 l),242 Webb, B. C.,196,209 Weber, G.,279,361 Weber, K., 135, 145,203,211 Weber, R. E., 222(23c),240 Weber, R. W., 224(32),240 Wegner, A., 160,161, 194, 198,211 Weiland, E., 99,I29 Weingarten, M. D.,140, 155, 156,211 Weisenberg, R. C.,138, 144, 145,162, 173, 174, 175,185, 187, 188,189, 211,212 Welch, W. J., 92,95, 106,132 Wells, R. M.J., 214(129), 244(129),243 Wendell, P.L.,236(120),242 Wengler, G.,83,92, 105,107,126,132 Wengler, G., 105, 107,132 Wheaton, V.,25,51 White, H . D., 139,149, 155, 187,204, 208,21I White, J. M., 101, 102, 103,128,131, 132 White, J. R., 231(114),242 White, S. W.,15,48 Wildi, E., 254,359 Wiley, D. C.,97,103,131,132

Wilkinson, T., 225(117),242 Williams, J. G.,220(46,50, 132),227(50), 240,243 Williams, J. P., 14,50 Williams, R. C.,140,210,211 Williams, R.C.,Jr., 139,156, 158, 164, 207,209,210 Williams, R. J. P., 13, 14,50 Willingham, M.C., 100,129 Willis, B. T.M., 345,346,361 Wills, B. D., 38,50 Wills, P., 22,49 Wils, S.,344,360 Wilson, B. P., 313,358 Wilson, I. A.,103,131, 132,354,358 Wilson, K. S., 15,48 Wilson, L., 198,139, 144,200,201,202, 206,207,209,210,212 Wimmer, E., 83,128 Winkelmann, D. A., 35,54 Winter, G., 110,126 Wirth, D. F., 93, 108, 109, 112, 113, 130, 132 Witman, G. B., 151,207,212 Witte, 0.N., 124,132 Wittenberg, B. A,, 217(9),239 Wittenberg, J. B.,217(9, lo),239 Wittmann, H. G., 2,3,8,9,10,14, 16, 19,20,23,32,33, 34,37,38,39,40, 43,44,48,49,50,53,54,55, 63,64, 65,77 Wittmann-Liebold, B., 2, 3, 5,6,7,8,9, 10, 11, 14, 15, 16, 19,20, 21, 22,23, 29,34,38,39,40,43,44,47,48,49, 50,52,54, 57,58,60,61,62,63,64, 66,67,69,70,71,72,73,74,75,76, 77, 78,83,92,126 Wittner, M., 32,50 Wlodauer, A.,334,335,337,359,361 Woese, C. R., 25,51,54 Wolfenden, R., 338,361 Wolff, J., 155,209 Wolinsky, J. S., 125,132 Wollenzien. P., 28,53 Wollenzien, P. L.,28,54 Wonacott, A. J., 236(1 18),342 Wong, K.-P., 21, 22,50, 54 Woodward, D. O.,140,143,209 Woolley, P., 38,49,53 Wootton, J. F., 225(42),240

38 1

AUTHOR INDEX

Wower, I., 43, 54 Wower, J., 40, 43, 44,54 Wright, P. G., 221(23a), 240 Wu, K., 232(62), 241 Wiirdinger, I., 232(66), 241 Wurmbach, P., 23, 34, 37,51 Wyckoff, H. W., 324, 337, 339,361 Wyman, J., 218(19), 226(48), 240, 284, 300,359,361

Wyman, J., Jr., 177,212 Wystup, G., 3, 54

Y Yabuki, S., 18, 34, 37,51, 52 Yaguchi, M., 8, 9, 14, 20, 22, 34, 38, 39, 40, 43, 44, 56, 62, 63, 64, 65, 77

Yamakura, F.. 236(126), 243, 328,361 Yang, J. T., 10, 52 Yarbrough, L. R., 145,208 Yeates, D. G. R., 335,358 Yonath, A., 15, 32, 33, 48, 50, 54, 55 Yonetani, T., 218(17), 240, 343, 360 Yphantis, D. A., 156, 158,210 Ysebart, M., 83,126, 127 Yuki, A., 43, 44,55

z Zaccai, G., 19, 22, 31,52 Zackroff, R. V., 175, 185, 187, 189,212 Zaldivar, J., 140, 143, 144, 153,211 Zalite, 0. M., 27, 54 Zamir, A., 44, 49 Zantema, A., 38,55 Zavada, J., 124,132 Zeeberg, B., 145, 175, 181, 185, 187, 189, 198, 199, 202, 203, 204, 205, 206, 207,212 Zeichhardt, H., 32, 38, 50, 53 Zeppezauer, E., 235(94), 242 Zera, A. J., 235(91), 242 Ziemiecki, A., 81, 88, 111, 113, 119, 132 Zilberstein, A., 117, 132 Zimmermann, R. A., 40,43,55 Zinn, A., 53 Zito, R., 220(108), 242 Zobawa, M., 4 4 , H Zook, D. E., 40,55 Zuker, M., 25,53 Zwieb, C., 25, 26, 28, 40, 41, 42, 43, 48, 49, 50, 53,55

This Page Intentionally Left Blank

SUBJECT INDEX

A

stop-action intermediates, 247-267 validity and exploitation of data in mixed solvents, 267-274 Escherichia coli, ribosomal proteins, primary structure of, 56-76

Amphibia, hemoglobins of, 226-228

B Birds, hemoglobins of, 231-232

F

C

Fish, hemoglobins of, 217-225 experimental test of theory of Root effect, 225-226 Flavodoxin, unstable oxidation state, 344345 Flow cell, equipment for, 324-326 Fluorescent energy transfer, between ribosomal proteins, 37-38

Crystal lifetime, radiation damage and, 328-329

D Dielectric constant, dielectric shock and solutions isodielectric with water, 283-286

E

G

Elastase, productive intermediates, 330332 Enzymatic reaction pathways, mapping of serine proteases, 354-355 triosephosphate isomerase, 353-354 Enzyme(s), species adaptations in, 234235 Enzyme-catalyzed reactions, temporal resolution of horseradish peroxidase, 249-25 1 hydroxylating multienzyme systems, 25 1-254 kinetics of lysozyme-catalyzed reaction, 259-265 lysozyme, 258-259 ribonuclease A, 265-267 serine proteases, 254-258 Enzyme-substrate intermediates, problem of determination of conformational changes accompanying enzyme catalysis, 275-280

Glycoproteins, of Semliki Forest virus, 86-95 Guanine nucleotides, microtubule elongation and, 173-176

383

H Hemoglobin(s) of amphibia, 226-228 of birds, 23 1-232 fish, 217-225 experimental test of theory of Root effect, 225-226 of mammals, 232-234 of reptiles, 228-23 1 Horseradish peroxidase, temporal resolution of reaction, 249-25 1

I Immune electron microscopy, of ribosome, 32-34

384

SUBJECT INDEX

K Kinetic probes, of microtubule elongation and disassembly, 177- 182

L Lipids, of Semliki Forest virus, 95-98 Lysozyme, temporal resolution of reaction, 258-259 kinetics of, 259-265

M Mammals, hemoglobins of, 232-234 Microtubule(s) length redistribution, 190-193 protomer flux with assembled microtubules experimental findings, 200-206 theoretical considerations, 194-200 protomer-polymer equilibria and critical concentration behavior, 182190

experimental findings, 185-190 theoretical considerations, 183- 185 Microtubule assembly, nucleation or initiation, 158-168 experimental findings, 162- 168 theoretical considerations, 159- 162 Microtubule elongation, 168-182 experimental findings, 172- 182 theoretical considerations, 169-172 Microtubule proteins, biochemical properties of, 137-158 microtubule-associated proteins, 153158

tubulin, 138-153 Multienzyme systems, hydroxylating, temporal resolution of reaction, 251254

Myoglobin, analysis of flexibility, 350-353

N Neutron scattering, ribosome and, 34-37 Nucleocapsid, of Semliki Forest virus, structure, 82-90

0 Oxidation states, unstable, 342-343 flavodoxin, 344-345 oxymyoglobin, 343-344

Oxymyoglobin, unstable oxidation state, 343-344

P Productive intermediates, 329-330 elastase, 330-332 ribonuclease A, 332-335 cyclic phosphate intermediate, 337 dinucleotide complex, 337 product complex, 338-342 transition state complex, 337-338 Protein(@ analysis of flexibility, 345-348 myoglobin, 350-353 trypsinogen, 345-348 crystalline, cryoprotection of, 280-283 crystals effect of cosolvent and temperature on Donnan and electrostatic parameters of, 301-317 possible supercooling of, 3 17-320 interactions, cosolvent and temperature effects on, 292-295 low-temperature structures, solving and refining, 327 microtubule-associated, biochemical properties of, 153- 158 ribosomal preparation of, 2-4 primary structure, 4-9 secondary structure, 9- 12 shape, 15-23 tertiary structure, 12-15 stop-action pictures, 245-361 Protonic activity, control of, 295-301

R Reptiles, hemoglobins of, 228-23 1 Ribonuclease A productive intermediates cyclic phosphate intermediate, 337 dinucleotide complex, 337 product complex, 338-342 transition state complex, 337-338 temporal resolution of reaction, 265267

Ribonucleic acid, ribosomal primary and secondary structures, 2326

spatial arrangement in situ, 27-28

385

SUBJECT INDEX Ribosome topography of assembly of subunits, 45-47 chemical reactivity, 44-45 cross-linking, 38-40 crystals of protein, 32 fragments of ribosomal particles, 43-

44 immune electron microscopy, 32-34 neutron scattering, 34-37 protein-binding sites on RNA, 40-43 protein complexes, 44 singlet-singlet fluorescent energy transfer between proteins, 37-38 size and shape of subunits, 28-31

S Semliki Forest virus life cycle of budding, 120- 124 infection, 98- 104 intracellular transport of viral glycoprotein, l 11-1 19 synthesis, 104- 1 1 1 structure, 8 1-82 nucleocapsid, 82-90 viral envelope, 86-98 Serine proteases mapping pathway of, 354-355 temporal resolution of reaction, 254-

258 Species adaptations in enzymes, 234-235 in hemoglobins of amphibia, 226-228 of birds, 23 1-232 of fish, 217-226 of mammals, 232-234 of reptiles, 228-23 1 Spectroscopy, low-temperature, equipment for, 326-327 Stop-action pictures of proteins analysis of protein flexibility and, 345-

353 application to other high-resolution techniques, 356-357

large conformational changes and, 355-

356 productive intermediates and, 329-342 resolution, disorder and related problems, 329 time-resolved pictures of functioning proteins, 357-358 unstable oxidation states and, 342-345

T Triosephosphate isomerase, mapping pathway of, 353-354 Trypsinogen, analysis of flexibility, 348-

350 Tubulin, biochemical properties of isolation, 138-140 properties of nucleotide binding site,

144- 147 structural properties of dimer, 140-144 structural properties of microtubules,

147-153

W Water, preparation of cooled media isodielectric with, 286-292

X X-ray diffraction, low-temperature equipment for, 321-324 X-ray studies, at subzero temperatures control of protonic activity, 295-301 cosolvent and temperature effects on protein interactions, 292-295 cryoprotection of crystalline proteins,

280-283 dielectric constant, 283-292 effect of cosolvent and temperature on Donnan and electrostatic parameters, 301-317 possible supercooling of protein cryst al~,3 17-320 trapping of crystalline enzyme-substrate complexes, 320-32 1

This Page Intentionally Left Blank

CONTENTS OF PREVIOUS VOLUMES Volume 1

Lipoproteins ERWINCHARGAFF Structural Proteins of Cell and Tissues 0. SCHMITT FRANCIS Some Contributions of Immunology to the Study of Proteins HENRY P. TREFFERS The Interaction between the Alkali Earth Cations, Particularly Calcium, and Proteins DAVID M. GREENBERG

Terminal Amino Acids in Peptides and Proteins W. Fox SIDNEY The Copper Proteins C. R. DAWSON AND M. F. MALLETTE Mucoids and Clycoproteins KARLMEYER The Reactions of Formaldehyde with Amino Acids and Proteins DEXTER FRENCH AND JOHNT. EDSALL Wheat Gluten M. J. BLISH

The Purification and Properties of Certain Protein Hormones BACONF. CHOW

Protein Denaturation and the Properties of Protein Croups M. L. ANSON

Soybean Protein in Human Nutrition DONALD S. PAYNE A N D L. S. STUART

X-Ray Diffraction and Protein Structure L. FANKUCHEN

Nucleoproteins JESSEP. GREENSTEIN

AUTHOR INDEX-SUBJECT INDEX

The Proteins of Skeletal Muscle KENNETHBAILEY AUTHOR INDEX-SUBJECT INDEX

Volume 2

Volume 3

Transamination and the Integrative Functions of the Dicarboxylic Acids in Nitrogen Metabolism ALEXANDER F. BRAUNSTEIN Ferritin and Apoferritin LEONOR MICHAELIS

Analytical Chemistry of the Proteins A. J. P. MARTINAND R. L. M. SYNGE

Adsorption Analysis of Amino Acid Mixtures ARNETISELIUS

The Microbiological Assay of Amino Acids E. SNELL ESMOND

Spread Monolayers of Protein HENRY B. BULL

The Amino Acid Compositon of Food Proteins RICHARD J. BLCCK The Relationship of Protein Metabolism to Antibody Production and Resistance to Infection PAULR. CANNON

Films of Protein in Biological Processes ALEXANDER ROTHEN The Chemical Determination of Proteins PAULL. KIRK Reactions of Native Proteins with Chemical Reagents ROGERM. HERRIOTT 381

388

CONTENTS OF PREVIOUS VOLUMES

The Amino Acid Requirements of Man ANTHONY A. ALBANESE

Biological Evaluation of Proteins JAMES B. ALLISON

The Use of Protein and Protein Hydrolyzates for Intravenous Alimentation ROLBERT ELMAN The Preparation and Criteria of Purity of the Amino Acids MAXS. DUNNAND LOUIS8.ROCKLAND The Plasma Proteins and Their Fractionation JOHNT. EDSALL

Milk Proteins THOMAS L. MCMEEKIN AND B. DAVID

AUTHOR INDEX-SUBJECT INDEX

Proteins, Lipids, and Nucleic Acids in Cell Structure and Functions ALBERT CLAUDE

Volume 4

Protein Gels JOHND. FERRY The Interactions of Proteins and Synthetic Detergents FRANK W. PUTNAM Protein of Pathogenic Bacteria A. M. PAPPENHEIMER. JR. The Plasma Proteins in Disease B. GUTMAN ALEXANDER Preparative Electrophoresis and Ionophoresis HARRY SVENSSON Stereochemistry of Amino Acids A NEUBERGER X-Ray Studies of Amino Acids and Peptides ROBERTB. COREY Heme Proteins JEFFRIES WYMAN, JR. AUTHOR INDEX-SUBJECT INDEX

Volume 5

The Synthesis of Peptides JOSEPH S. FRUTON Amino Acid Composition of Purified Proteins G. R. TRISTRAM

POLS

Plant Proteins J. W. H. LUGG Synthetic Fibers Made from Proteins HAROLD €? LUNDGREN Some Protein-Chemical Aspects of Tanning Processes K.H. GUSTAVSON

AUTHOR INDEX-SUBJECT INDEX

Volume 6

The Electron Microscopy .of Macromolecules RALPHW. G . WYCKOFF Light Scattering in Protein Solutions PAULDOTYAND JOHNEDSALL Poly-a-Amino Acids EPHRAIM KATCHALSKI Egg Proteins HARRY L. FEVOLD Natural and Artifical lodoproteins JEANROCHEAND RAYMOND MICHEL Glutamic Acid and Cerebral Function HEINRICH WAELSCH Cross Linkages in Protein Chemistry JOHAN BJORKSTEN The Relation of Protein Metabolism to Disease HERBERT POLLACK AND SEYMOUR LIONEL HALPERN AUTHOR INDEX-SUBJECT INDEX CUMULATIVE SUBJECT INDEX FOR VOLUMES 1-5

CONTENTS OF PREVIOUS VOLUMES Volume 7

The Arrangement of Amino Acids in Proteins F. SANGER The Structure of Collagen Fibrils RICHABD S. BEAR Muscle Contraction and Fibrous Muscle Proteins HANSH. WEBERAND HILDEGARD PORTZEHL The Proteins of the Mammalian Epidermis K. M. RUDALL

V O I U9 ~

The Metabolism of Clycine H. R. V. ARNSTEIN The Digestion of Protein and Nitrogenous Compounds in Ruminants MARGARET 1. CHALMERS AND R. L. M. SYNGE The Resolution of Racemic a-Amino Acids JESSEl? GREENSTEIN Naturally Occurring Trypsin Inhibitors M. LASKOWSKI AND M. LASKOWSKI, JR.

Infrared Analysis of the Structure of Amino Acids, Polypeptides, and Proteins G. B. B. M. SUTHERLAND

The Formation, Composition, and Properties of the Keratins WILFRED H. WARDAND HAROLD I? LUNDGREN

Ultraviolet Absorption Spectra of Proteins and Amino Acids G. H. BEAVEN AND E. R. HOLIDAY

The Molecular Structure of Simple Substances Related to Proteins SAN-ICHIRO MIZUSHIMA

AUTHOR INDEX-SUBJECT INDEX

Protein-Protein Interactions DAVIDF. WAUGH

Volume 8

Naturally Occurring Peptides E. BRICAS AND CL. FROMAGEOT

389

Physiochemical and Biological Aspects of Proteins at Interfaces D. F,CHEESMAN AND J. T.DAVIES AUTHOR INDEX-SUBJECT INDEX

Peptide Bond Formation HENRY BORWOK Bacteriophages: Nature and Reproduction FRANK W. F’UTNAM Assimilation of Amino Acids by GramPositive Bacteria and Some Actions of Antibiotics Thereon ERNEST F. GALE Peanut Protein: Isolation, Composition, and Properties JETTC. ARTHUR, JR. Rotational Brownian Motion and Polarization of the Fluorescence of Solutions GREGORIO WEBER Zone Electrophoresis ARNETISELIUS AND PER FLODIN AUTHOR INDEX-SUBJECT INDEX

Volume 10

The Nature of Phosphorus Linkages in Phosphoroproteins GERTRUDE E. PERLMANN Metabolism of the Aromatic Amino Acids C. E. DALGLIESH Hydrogen Ion Equilibria in Native and Denatured Proteins JACINTO STEINHARDT AND ETHEL M. ZAISER Fish Proteins G. HAMOUR The Sea as a Potential Source of Protein Food LIONEL A. WALFORD AND CHARLES G. WILBER

390

CONTENTS OF PREVIOUS VOLUMES

Zinc and Metalloenzymes BERTL.VALLEE

and Lactogenic Hormones CHOHHAOLI

AUTHOR INDEX-SUBJECT INDEX

The Activation of Zymogens HANSNEURATH

Volume 11

Protein Structure in Relation to Function and Biosynthesis CHRISTIAN B. ANFINSENAND ROBERTR. REDFIELD Hormones of the Anterior Pituitary Gland Part I. Growth and Adrenocorticotropic Hormones CHOHHAOLI

The Chemical Nature of Antibodies HENRYC.ISLIKER The Synthesis of Peptides MURRAYGOODMAN AND G. W. KENNER AUTHOR INDEX-SUBJECT INDEX

Volume 13

The Use of Immunochemical Methods in Studies on Proteins RERRE GRABAR

Column Chromatography of Peptides and Proteins Protein-Carbohydrate Complexes STANFORD MOOREAND WILLIAMH.STEIN F. R. BETTELHEIM-JEVONS Countercurrent Distribution in Protein The Silk Fibrions Chemistry F. LUCAS,J. T.B. SHAW, A N D S. G. P. VON TAVEL AND R. SIGNER SMITH Complex Formation between Metallic Cations and Proteins, Peptides, and Amino Acids FRANK R. N. GURDAND PHILIP E.

Synthesis and Chemical Properties of Poly-a-Amino Acids EPHRAIM KATCHALSKI AND MICHAEL SELA AUTHOR INDEX-SUBJECT INDEX

WILCOX

Measurement and Interpretation of Diffusion Coefficients of Proteins Louis J. GOSTING AUTHOR INDEX-SUBJECT INDEX

CUMULATIVE SUBJECT INDEX FOR VOLUMES 6-10

Volume 12

Volume 14

Some Factors in the Interpretation o f Protein Denaturation W. KAUZMANN Zone Electrophoresis in Starch Gels and Its Application to Studies of Serum Proteins 0. SMITHIES

The Fibrinogen-Fibrin Conversion HAROLD A. SCHERAGAA N D MICHAEL LASKOWSKI, JR.

The Specificity of Protein Biosynthesis MARTHAVAUGHANAND DANIEL STEINBERG

X-Ray Analysis and Protein Structure F. H.c. CRICK AND J. c. KENDREW

Structural Aspects of Tobacco Mosiac Virus H. FRAENKEL-CONRAT AND L. K. RAMACHANDRAN

The Human Hemoglobins: Their Properties and Genetic Control HARVEY A. ITANO Hormones of the Anterior Pituitary Gland. Part 11. Melanocyte-Stimulating

The Serum Proteins of the Fetus and Young of Some Mammals R. A. KEKWICK

CONTENTS OF PREVIOUS VOLUMES The Sulfur Chemistry of Proteins R. CECILAND J. R. MCPHEE

391

Volume 17

The Properties of Proteins in Biological Properties of Poly-a-Amino Nonaqueous Solvents Acids S. J. SINGER MICHAEL SELAAND EPHRAIM KATCHALSKI The Interpretation of Hydrogen Ion AUTHOR INDEX-SUBJECT INDEX Titration Curves of Proteins CHARLES TANFORD Volume 15

Protamines KURTFELIX Osmotic Pressure D. W. KUPKE

Regularities in the Primary Structure of Proteins F. SORMAND B. KEIL Cross-Linked Dextrans as Molecular Sieves JERKER PORATH

Elastin Protein Malnutrition in Man S. M.PARTRIDGE A N D JOAN J. C. WATERLOW, J. CRAVIOTO M. L. StEPHEN Ultraviolet Spectra of Proteins and Amino Acids Rective Sites and Biological Transport D.B. WETLAUFER HALVOR N. CHRISTENSEN Crystallized Enzymes from the Myogen of Rabbit Skeletal Muscle R. CZOKAND TH. BUCHER AUTHOR INDEX-SUBJECT INDEX

Volume 16

AUTHOR INDEX-SUBJECT INDEX

Volume 18

Recent Studies on the Structure of Tobacco Mosaic Virus ANDERER F. ALFRED

The Structure of Collagen and Gelatin A N D PETER H. WILLIAM F. HARRINGTON VON HIPPEL

Assembly and Stability of the Tobacco Mosaic Virus Particle D. L. D. CASPAR

The Proteins of the Exocrine Pancreas I! DESNUELLE AND M. ROVERY

The Dissociation and Association of Protein Structures F. J. REITHEL

Enzyme Fractionation by Salting Out: A Theoretical Note C. WEBB MALCOLM DIXONAND EDWIN Nonenzymatic Methods for the Preferential and Selective Cleavage and Modification of Proteins B. WITKOP The Viscosity of Macromolecules in Relation to Molecular Conformation JENTSI YANG Optical Rotation and the Conformation of Polypeptides and Proteins PETER URNESAND PAULDOTY AUTHOR INDEX-SUBJECT INDEX

The Amino Acid Composition of Some Purified Proteins G. R. TRISTRAM AND R. H. SMITH AUTHOR INDEX-SUBJECT INDEX

Volume 19

The Hemoglobins K. HILSE,V. RUDLOFF. G. BRAUNITZER, A N D N. HILSCHMANN Hemoglobin and Myoglobin ALESSANDRO Ross1 FANELLI, ERALW AND ANTONIO CAPUTO ANTONINI,

392

CONTENTS OF PREVIOUS VOLUMES

Linked Functions and Reciprocal Effects in Hemoglobin: A Second Look JEPP~UES WYMAN. JR. Thermodynamic Analysis of Multicomponent Solutions EDWARD F. CASASSA AND HENRYK EISENBERG AUTHOR INDEX-SUBJECT INDEX

Volumo 22

Covalent Labeling of Active Sites S. J. SINOER Milk Proteins H. A. MCKENZIE Crystal Structure Studies of Amino Acids and Peptides RICHARDE. MARSHAND JERRY DONOHUE Crystal Structures of Metal-Peptide Complexes HANSC. FREEMAN

Volumo 20

Thrombosthenin, the Contractile Protein from Blood Platelets and Its Relation to Other Contractile Proteins M. BETTEX-GALLAND AND E. F. LUSCHER Hydrolysis of Proteins ROBERTL. HILL The Unusual Links and Cross-Links of Collagen JOHNJ. HARDING The Chemistry of Keratins W. G. CREWTHER, R. D. B. FRASER, F. G. LENNOX, AND H. LINDLEY AUTHOR INDEX-SUBJECT INDEX

AUTHOR INDEX-SUBJECT INDEX

Volumo 23

Relaxation Spectrometry of Biological Systems G. HAMMES GORDON The Preparation of Isomorphous Derivatives C. C. F. BLAKE Protein Denaturation CHARLES TANFORD Conformation of Polypeptides and Proteins G. N. RAMACHANDRAN AND v. SASISEKH ARAN AUTHOR INDEX-SUBJECT INDEX

Volume 21

Naturally Occurring Peptides S. 0. WALEY

Cytochrome c E. MARG~LIASH AND A. SCHECTER Hydrogen Exchange in Proteins AASEHVIDT AND SIGURD 0. NIELSEN Selenium Derivatives in Proteins J. J~JREGUI-ADELL AUTHOR INDEX-SUBJET INDEX

CUMULATIVE AUTHOR INDEX FOR VOLUMES 1-21 CUMULATIVE TITLE INDEX FOR VOLUMES 1-21

Volumo 24

Protein Denaturation Part C. Theoretical Models for the Mechanism of Denaturation CHARLES TANFORD Selective Cleavage and Modification of Peptides and Proteins T. F. SPANDE, B. WITKOP, Y. DEGANI, AND A. PATCHORNIK Recent Developments in Chemical Modification and Sequential Degradation of Proteins GEORGE R. STARK

CONTENTS OF PREVIOUS VOLUMES Partition of Cell Particles and Macromolecules in Polymer Two-Phase Systems PER-AKEALBERTSSON

393

Rate of Conformational Transitions in Biological Macromolecules and Their Analog HERBERTMORAWETZ

Insulin: The Structure in the Crystal and Analytical Gel Chromatography of Its Reflection in Chemistry and Biology Proteins TOMBLUNDELL, GUY DODSON, DOROTHY GARYK. ACKERS HODGKIN,AND DANMERCOLA Nuclear Magnetic Resonance AUTHOR INDEX-SUBJECT INDEX Spectroscopy of Amino Acids, Peptides, and Proteins G. C. K. ROBERTSAND OLEGJARDETZKY AUTHOR INDEX-SUBJECT INDEX

Volume 27 Volume 25

Carboxypeptidase A: A Protein and an Enzyme FLORANTE A. QUIOCHOAND WILLIAM N. LIPSCOMB The Structure of Papain J. DRENTH,J. N. JANSONIUS, R. KOEKOEK.AND B. G. WOLTHERS

Structural Aspects of the Fibrinogen to Fibrin Conversion R. F. -LITTLE The Structure and Chemistry of Keratin Fibers J. H. BRADBURY The Elongation Reactions in Protein Synthesis PHILIPLEDER

Protein Malnutrition in Children: Advances in Knowledge in the Last Ten Peptide Chain Termination Years C. T. CASKEY J. C. WATERLOW AND G.A. 0 . ALLEYNE Structure of Bacterial Ribosomes The Chemistry and Structure of AND H. G. WITTMANN ROGERA. GARRETT Collagen GIycoproteins WOLFIETRAUB AND KARLA. PIEZ ROBERTG. SPIRO AUTHOR INDEX-SUBJECT INDEX

AUTHOR INDEX-SUBJECT INDEX

Volume 26

Glutamine Synthetase of Escherichin coli: Some Physical and Chemical Properties ANNGINSBURG The History of the Discovery of the Amino Acids. 11. A. Review of Amino Acids Described Since 1931 as Components of Native Proteins HUBERTBRADFORD VICKERY Interferons: Physiochemical Properties and Control of Cellular Synthesis MUNH.NG AND JANVILCEK

Volume 28

Phosphoproteins GEORGETABORSKY The Mechanism of Interaction of Red Cell Organic Phosphates with Hemoglobin RUTHE. BENESCHAND REINHOLD BENESCH Hydration of Proteins and Polypeptides 1. D. KUNTZ,JR., AND W. KAUZMANN

394

CONTENTS OF PREVIOUS VOLUMES

Molecular Orbital Calculations on the Conformation of Amino Acid Residues of Proteins PULLMANAND ALBERTE BERNARD PULLMAN AUTHOR INDEX-SUBJECT INDEX

Volume 29

Energetics of Ligand Binding to Proteins GREGORIO WEBER Avidin N. MICHAELGREEN Carbonyl-Amine Reactions in Protein Chemistry ROBERT E. FEENEY, GUNTER BLANKENHORN. AND HENRY B. F. DIXON Experimental and Theoretical Aspects of Protein Folding C. B. ANFINSEN AND H. A. SCHERAGA AUTHOR INDEX-SUBJECT INDEX

Volume 30

Thin Film Dialysis KENTK. STEWART Tobacco Mosaic Virus Protein Aggregation and the Virus Assembly €? JONATHAN, G. BUTLER, AND ANTHONY C. H. DURHAM The Plasma Lipoproteins JAMES C. OSBORNE. JR.. AND H. BRYAN BREWER, JR. Nerve Growth Factor C. SERVER AND ERICM. SHOOTER ALFRED AUTHOR INDEX-SUBJECT INDEX

CUMULATIVE AUTHOR INDEX FOR VOLUMES 22-31 CUMULATIVE TITLE INDEX FOR VOLUMES 22-31

Volume 32

The Structural Basis of Antibody Complementarity ELVINA. KABAT

Protein Fractionation at Subzero Repressors Temperatures AND MAGNUS PFAHL SUZANNE BOURGEOIS PIERRE Douzou AND CLAUDE BALNY Bovine Liver Glutamate Dehydrogenase Antifreeze Protein from Fish Bloods HENRYK EISENBERG, ROBERT JOSEPHS. AND ROBERTE. FEEKNEY A N D YIN Y E H EMILREISLER Protein at Interfaces The Thermodynamic Basis of the F. MACRITCHIE Stability of Protein, Nucleic Acids, and AUTHOR INDEX-SUBJECT INDEX Membranes HAROLD EDELHWH A N D JAMES C. OSBORNE, JR. Membrane Receptors and Hormone Action PEDRO CUATRECASAS AND MORLEY D. HOLLENBERG AUTHOR INDEX-SUBJECT INDEX

Volume 31

The r-(y-Clutaniy1)lysine Crosslink and the Catalytic Role of Transglutaminases J. E. FOLK AND J. S. FINLAYSON

Volume 33

Activation of the Complement System by Antibody-Antigen Complexes: The Classical Pathway R. R. PORTER AND K. B. M. REID Motions in Proteins FRANK R. N. GURDAND T. MICHAEL ROTHGEB Stability of Proteins €? L. PRIVALOV

CONTENTS OF PREVIOUS VOLUMES Peptides of the Central Nervous System WERNERA. KLEE AUTHOR INDEX-SUBJECT INDEX

Volume 34

The Histones RUTHSPERLING AND ELLEN J. WACHTEL Folding of Protein Fragments DONALD B. WETLAUFER The Theory of Pressure Effects of Enzymes EDDIEMORILD The Anatomy and Taxonomy of Protein Structure JANE S. RICHARDSON AUTHOR INDEX-SUBJECT INDEX

395

Volume 35

Stability of Proteins: Proteins Which Do Not Present a Single Cooperative System P. L. PRIVALOV New Perspectives on c-Type Cytochromes T. E. MEYERAND M. D.KAMEN Calmodulin CLAUDEB. KLEEAND THOMAS C. VANAMAN Parathyroid Hormone: Chemistry, Biosynthesis, and Mode of Action JOHN T. POTTS,JR., HENRYM. KRONENBERG. AND MICHAEL ROSENBLATT AUTHOR INDEX-SUBJECT INDEX

This Page Intentionally Left Blank