Advances in Protein Chemistry: v. 7

ADVANCES IN PROTEIN CHEMISTRY VOLUME VII

This Page Intentionall Left Blank

ADVANCES IN PROTEIN CHEMISTRY EDITEDBY

M. L. ANSON

KENNETH BAILEY

Research Division; Lever Brothers Co. Edgewater, New Jersey

University of Cambridge Cambridge, England

JOHN T. EDSALL Harvard Medical School Boston, Massachusetts

VOLUME VII

1952 ACADEMIC PRESS INC., PUBLISHERS NEW YORK, N. Y.

Copyright 1952, by ACADEMIC PRESS INC. 125 EAST 2 NEW YORK

3 STREET, ~ ~ 10, N . Y.

All Rights Reserved No part of this book may be reproduced in any form, by photostat, microfilm, or any other means, without written permission from the publishers. Library of Congress Catalog Card Number: 44-8853

PRiNTED I N THE UNITED STATES OF AMERICA

CONTRIBUTORS TO VOLUME VII

RICHARD S . BEAR,Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts

G. H. BEAVEN, Medical Research Council Spectrographic Research Unit, London Hospital, London, England E. R. HOLIDAY, Medical Research Council Spectrographic Research Unit, London Hospital, London, England HILDEGARD PORTZEHL, Physiological Institute, University of Tubingen, Germany K. M. RUDALL, Department of Biomolecular Structure, The University, Leeds, England F. SANGER, Sir William D u n n Institute of Biochemistry, University of Cambridge, Cambridge, England G. B. B. M. SUTHERLAND, Physics Department, University of Michigan, Ann Arbor, Michigan HANSH. WEBER, Physiological Institute, University of Tubingen, Germany

V

This Page Intentionall Left Blank

CONTENTS CONTRIBUTORS TO VOLUME VII . . . . . . . . . . . . . . . . . . . . . .

v

The Arrangement of Amino Acids in Proteins

BY F. SANGER.Sir W i l l i a m D u n n Institute of Biochemistry. University of Cambridge. Cambridge. England I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 I1. Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 5 111. Determination of the Position of Individual Residues in Proteins . . . . IV. Methods for the Degradation of Proteins . . . . . . . . . . . . . . . 11 V. Fractionation of Peptides . . . . . . . . . . . . . . . . . . . . . . 29 VI . Determination of Peptide Structure . . . . . . . . . . . . . . . . . 41 VII . Results of Investigations on Various Proteins . . . . . . . . . . . . . 44 i7111. General Conclusions . . . . . . . . . . . . . . . . . . . . . . . . 59 The Structure of Collagen Fibrils

BY RICHARDS. BEAR.Department of Biology, Massachusetts Institute of Technology, 1. I1. 111. IV. V. VI .

Cambridge, Massachusetts Introduction . . . . . . . . . . . . . . . . . Identification and Distribution of Collagens . . . Colloidal Structure of Collagen Fibrils . . . . . The Collagen Protofibril . . . . . . . . . . . . The Collagen Molecule . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. 69 . . 76 . . 97 . 112 . . . . . . . . . . 135 . . . . . . . . . . 150

Muscle Contraction and Fibrous Muscle Proteins

BY HANSH . WEBERA N D HILDEGARD PORTZEHL, Physiological Institute, University of Tubingen, Germany

I . Introduction . : . . . . . . . . . . . . . . . . . . . . . . . . . I1. The Contractile Models . . . . . . . . . . . . . . . . . . . . . . I11. The Proteins of the Myofibril and Their Reactions . . . . . . . . . . . IV . The Proteins of the Myofibril and the Fine Structure of Skeletal Muscle . V. General Conclusion . . . . . . . . . . . . . . . . . . . . . . . .

162 163 193 237 246

The Proteins of the Mammalian Epidermis

BY K . M . RUDALL, Department 0.f Biomolecular Structure, Leeds, England I. Introduction . . . . . . . . . . . . . . . . . . . . I1. Properties of the Epidermis as a Whole . . . . . . . 111. The Extraction of Epidermal Proteins . . . . . . . . I V. Some Physical Properties of Epidermal Proteins . . . . V. Infrared Absorption Studies . . . . . . . . . . . . . VI . Summary . . . . . . . . . . . . . . . . . . . . . vii

The University,

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

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

253

. . 254 . . 264 . . 270 . 279 . 286

...

Vlll

CONTENTS

Infrared Analysis of the Structure of Amino Acids. Polypeptides and Proteins

BY G . B . U . M . SUTHERLANI). I’hysics Department. University of Michigan. Anri Arbor. Michigan I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. Amino Acids . . . . . . . . . . . . . . . . . . . . . . . . . . I11. Amides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Polypeptides . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Some General Observations . . . . . . . . . . . . . . . . . . . . .

2!)1 297 299 304 306

312

Ultraviolet Absorption Spectra of Proteins and Amino Acids BY G . H . BEAVENAND E. R . HOLIDAY, Medical Research Council Spectroyraphic Research Unit. London Hospital. London. England I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 320 I1. Methods and Experimental Aspects . . . . . . . . . . . . . . . . . 321 I11 Absorption Constants of the Aromatic and the SuIfnr-(:ontaimirig Amino Acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323 I V . The Vibrational Fine Structure of thc Ahsorption Spectra of the Aromatic Amino Acids and of Proteins . . . . . . . . . . . . . . . . . . . 329 V . Low Temperature Spectra of Amino Acids and Proteins . . . . . . . . 331 VI . Fine-Structure Shifts in Protein Spectra-Structural Implications . . . . 336 VII The Absorption Spectra of the Aromatic Amino Acids and Proteins in Strongly Alkaline Solution (pH 12-13) . . . . . . . . . . . . . . 343 VIII . The Ultraviolet Absorption Spectrum of the Peptide Bond and of the Polypeptide Fabric . . . . . . . . . . . . . . . . . . . . . . . . . 352

.

.

AUTHORINDEX . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

387

SUBJECT INDEX . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

400

The Arrangement of Amino Acids in Proteins BY F. SANGER Sir W i k m D u n n Institute of Biochemistry. [Tniversity of Cambridge. Cambridge. England

CONTENTS I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Polypeptide Chains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . a. Open Polypeptide Chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

P'age

............................. . . . . . . . c. Branched Chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 . Terminal Residues and Peptides. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I11. Determination of t h e Position of Individual Residues s. . . . . . . . 1. The Dinitrophenyl (DNP) Method . . . . . . . . . . . . . . . a . Fractionation of D N P Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . h N-Terminal Residues of Proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . c . N-Terminal Peptides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 . The Phenylthiocarbamyl Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. The Thiocarbamate Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. The Use of S-Methylisothiourea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 . C-Terminal Residues ........................................ a. Carboxypeptidase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . h Reduction to B-Amino Alcohols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV . Methods for the Degradation of Proteins . . . . . . . . . . . . . . . . . . 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . a. Non-Specific Degradation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . b . Specific Degradation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 . General Aspects of Partial Hydrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . a . Yield of Peptides., . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . b . Complexity of Partial Hydrolyzates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . c. The Question of Rearrangement of Peptide Sequences during Hydrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 . Hydrolysis in Concentrated Acid., . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . a . Electrostatic Effects . . . . . . . . . .

.

.

.

.

4. 5. 6.

7. 8.

2 3 3

3 3 4 4

5 5 5 6 0

7 8 9 9 9 9

10 11 11 11 11 12 12 13

15 18 18 19 ............................................ c . Bonds Involving the Amino Groups of the Hydroxy Amino Acids. . 21 Hydrolysis in Dilute Acid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Hydrolysis in Alkali . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Hydrolysis with Proteolytic Enzymes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Rates of Hydrolysis of Proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 27 Non-Hydrolytic Methods of Degradation . . . . . . . . . . . . . . . . . . . . . . . . . 1

2

F. SANOER

Page 27 29 29 29 30 31 34 35

a. Splitting the Disulfide Bridges.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . b. Splitting by Radiation, ......................... .. ...... c. Other Methods,. , . , . , , ................................... V. Fractionation of Peptides, . . , ........................ 1. Ionophoretic Methods, , , , ....... ... .... . ...... ... .... . ... ... 2. Ion Exchange Methods, , , ... ....... . ...... . .... 3. Adsorption Chromatography. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Partition Chromatography. . . . . . . . . . . . . . . . , . . . . . . . . . . . . . . . . . . . . . a. Paper Chromatography. . . , , . . . , . . . . . , . . . . . . . , , , . . . , . . . . , . . b. Starch Chromatography. . , , , . . . . . . . . . . . . . . . . . . . . . . . . . . . , . . . c. Other Partition Chromatography, . . , , , , . . . . . . . . . . . . , , . . . . . . . 5. Detection of Peptides from Columns. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 . Countercurrent Distribution. , , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . , 7. Lysine Peptides.. . . . . . . . . . . , . . . . , . . . . , . . . . . . . . . . . . . 8. Cystine Peptides.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ,

9. Other Methods of Fractionation ............................... 10. Conclusions. , , . . . , . . . . . , . . . . . VI. Determination of Peptide Structure.. . , . . . . . . . . . . . . . . . . . . . . , . . . . . . . , . 1. Identification and Estimation of Amino Acids.. . , , , . . . , , , , , . , , . . , . 2. Amino Acid Sequence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..._ 3. Estimation of Peptides. . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . , . . . , . . . . . ........................ . The Pipsyl Method. . . . . . . . . . . . . . . . s . .. . . . . . . . . . . . . . . . . . . . . . . VII. Results of Investigations on Various Pro 1. Silk Fibroin (Bombyz M o r i ) . . . . . . . . . . . . . . . . , . . . . . . . . . . , . . . . . . . . , 2. Protamines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Wool Keratin , . . . . . . . . . , . . , . . . . . , . . . . . . . . . . . . . . . . , . . . , , . . , , . . . , . 4. Insulin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . a. N-Terminal Peptides. , . . , . , . b. Amino Acid Sequence in the Phenylalanyl Chains.. . . . . . . . , . . . . . . ....................... 5. Ovalbumin.. . . . . . . . . . . . . . . . . . . 6. 7-Globulin.. . . . . . . . . . . . . . . . . . , 7. Hemoglobin., , . , , . , . . . . . . . . . . . 8. Gelatin.. . . . , , . . . . . . . . . . . . . . . VIII. General Conclusions, , , . . . . . . . . . . . References ,

38 38 39 40 40 40 41

41 42 43 43 44 44 47 48 50 54

57

62

I. INTRODUCTION A comprehensive review of the earlier literature on the partial hydrolysis products of proteins was given by Synge in 1943. Up to that time only a few simple peptides had been clearly identified from proteins by the classical and rather laborious methods of organic chemistry and Synge concluded that ‘ I the main obstacle to progress in the study of protein structure by the methods of organic chemistry is inadequacy of technique!” Probably the greatest advance that has been made recently in this field was the development by Martin and Synge (1941) of the entirely new technique of partition chromatography. The great problem in peptide chemistry has always been to find methods of frac-

THE ARRANGEMENT OF AMINO ACIDS I N PROTEINS

3

tionating the extremely complex mixtures produced by the partial degradation of a protein. Older methods of fractional crystallization and precipitation with various reagents were as a rule inadequate to deal with these mixtures, and countercurrent methods of high resolving power, which could fractionate non-volatile, water-soluble substances, were needed. Partition chromatography, especially in the form of paper chromatography (Consden et al., 1944), is such a method, so that it has already been possible to identify as breakdown products of proteins more peptides using this technique than had previously been identified by the classical methods of organic chemistry. During the last few years, work in this field has centered largely on the development of methods, so that this review will be more a consideration of techniques and their uses than a discussion of results, which are still rather few. As an initial working hypothesis it will be assumed that the peptide theory is valid, in other words, that a protein molecule is built up only of chains of a-amino (and a-imino) acids bound together by peptide bonds between their a-amino and a-carboxyl groups. While this peptide theory is almost certainly valid (see Vickery and Osborne, 1928; Pauling and Niemann, 1939; Synge, 1943), it should be remembered that it is still a hypothesis and has not been definitely proved. Probably the best evidence in support of it is that since its enunciation in 1902 no facts have been found to contradict it. It is to be expected that investigations of the types described in the present article will throw further light on the accuracy of the peptide theory and on the possible existence of nonpeptide bonds in proteins.

11.

NOMENCLATURE

Some of the terms to be used are new and require definition. 1. Polypeptide Chains

Three types of polypeptide chains are possible, open, cyclic and branched. a. Open Polypeptide Chain. A chain of amino acids joined together by peptide bonds between the a-amino and a-carboxyl groups with a free amino group at one end of the chain and a free carboxyl group at the other end is the most usually considered type of polypeptide chain. The number of these chains in a protein may be estimated by the number of a-amino or a-carboxyl groups. b. Cyclic Chain. A cyclic chain may be derived from an open polypeptide chain by peptide bond formation between the two terminal residues, and contains no free a-amino or a-carboxyl groups. The anti-

4

F. MANGER

biotics “gramicidin S ” (Sanger, 1946) and tyrocidine (Christensen, 1945) have been shown to possess cyclic structures and the absence of any free a-amino groups in ovalbumin and certain muscle proteins suggests that they too are built up from cyclic chains. c. Branched Chain. The presence of two carboxyl groups in glutamic and aspartic acids and of two amino groups in lysine suggests the possibility that branched chains may occur in proteins. For instance, such a branched system could be formed by the formation of a peptide bond between the free a-amino group of one open chain and a y-carboxyl group of a glutarnic acid residue in another chain. For all proteins that have been studied by the dinitrophenyl (DNP) method (see below) it was found that all the c-amino groups of the lysine residues were free, indicating that these groups are not involved in any bond formation. Thus it is unlikely that there is very much branching of chains from the lysine residues. If the branching points are very few, however, it is just possible that some of the c-amino groups thus masked have escaped detection. No evidence is available as t o whether hranahing can occur from the w-carboxyl groups of aspartic or glutamic acids. The presence of the y-peptide linkage of glutamic acid in glutathione, glutamine and the capsular substances from B. Anthracis (Hanby and Rydon, 1946) suggests that it may also occur in proteins. Some proteins are built up of two or more polypeptide chains held together by a stable bond other than the peptide link. Such a bond will be referred to as a “cross linkage.” The only one that is definitely bridge of cystine. Here two cysteine known to exist is the -S-Sresidues are joined together through their side chains.

2. Terminal Residues and Peptides There are two types of terminal residues, those with a free amino group and those with a free carboxyl group. In previous publications (Sanger, 1945, 194913) the expression “terminal residue’’ has been used to denote only that residue which carries a free amino group. However, it seems that a distinction should be made. Fox (1945) has suggested that those residues containing a free amino group and a bound carboxyl group be referred to as terminal amino acids and those with a free carboxyl group as terminal amino acids. Although clear on paper, this distinction is rather difficult to make in conversation. Following a suggestion by Dr. K. Bailey, it is proposed to use the term N-terminal residue for the residue having a free amino group and C-terminal residue for that having a free carboxyl group. The same nomenclature will apply to the terminal peptides.

5

T H E ARRANGEMENT O F AMINO ACIDS I N P R O T E I N S

3. Abbreviations Throughout this review the abbreviations for the amino acid residues suggested by Brand and Edsall (1947) are used. Cysteic acid is abbreviated CySOSH. Their method of writing empirical formulae of proteins and peptides is also adopted, e.g., Gly2vAlallValzl . . . etc., the subscripts representing the number of residues of the amino acid per protein molecule. I n this type of formula the order in which the amino acids are given has no significance. Where the order of residues is known, as in the description of a peptide, the symbols are joined by a period. Thus, whereas Gly,Ala signifies a dipeptide containing glycine and alanine, Gly.Ala represents glycylalanine. Gly. (Ala,Leu) indicates a peptide containing glycine, alanine and leucine with glycine as the N-terminal residue, and the order of the alanine and leucine is unknown. As is customary, the first residue is the N-terminal residue and the last the C-terminal.

111. DETERMINATION OF THE POSITION OF INDIVIDUAL RESIDUES IN PROTEINS The terminal residues of proteins differ from other residues in the chain, since they contain free amino or free carboxyl groups, and this fact may be used to identify them. Fox (1945) has reviewed the earlier literature on the study of terminal residues. At that time the position of only one amino acid in one protein was known. This was the presence of phenylalanine as an N-terminal residue in insulin. It was identified by Jensen and Evans (1935) who isolated the phenylhydantoin of phenylalanine from a hydrolyzate of insulin that had been treated with phenylisocyanate. More recently a general method has been worked out for the study of N-terminal residues, and preliminary investigations have been carried out on three methods of stepwise degradation which promise t o be of great use in the future. 1. The Dinitrophenyl ( D N P ) Method

No20

The principle of this general method (Sanger, 1945; Porter and Sanger, 1948) for the identification and estimation of the N-terminal residues of proteins may be summarized by the following formulae: +NH2iHCO-prot. +JN /h,(

NO2 1 :2 :4-Fluorodinitrobenaene (FDNB)

Protein

k

NHCH.CO-prot. +NOJz/,

k

NH.CH.COOH

IICl

N? 2. 2: 4-Dinitrophenyl(DNP)-protein

NOz DNP-Amino acids

+

Amino acids

6

F. SANQER

The FDNB reacts with the free amino groups of the protein under mild (slightly alkaline) conditions where the peptide bond is quite stable. On hydrolysis of the protein the N-terminal residues are liberated in the form of DNP amino acids. These are bright yellow compounds that can be extracted with an organic solvent, fractionated chromatographically, and estimated colorimetrically. The accuracy varies somewhat with the particular amino acids involved, due to differences in the stability of the DNP derivatives. I n most cases the N-terminal residues of proteins and peptides may be estimated to within 10-15%. a. Fractionation of D N P Derivatives. Several methods of fractionation have been suggested, all depending essentially on partition chromatography. Originally (Sanger, 1945; Porter and Sanger, 1948; Porter, 1950c) a scheme was worked out for separating the D N P derivatives of all the known amino acids using silica gel saturated with water as the stationary phase and various organic solvents as the mobile phase. This method was found to give satisfactory and reproducible separations in the author’s laboratory; other workers (Consden et al., 194710; Blackburn, 1949), however, have found difficulty in obtaining suitable gels. These may be prepared from any type of sodium silicate by appropriate modification of the method of preparation (Desnuelle et al., 1950), and should be rather strongly adsorbent in order to hold the relatively insoluble D N P derivatives in the stationary phase. Other methods of fractionation have been suggested in which the stationary phase is a buffer, adsorbed on kieselguhr (Bell et al., 1949) or silica (Blackburn, 1949; Middlebrook, 1949). The DNP-derivatives are then partially ionized and are thus rendered more soluble in the aqueous phase. Such systems may prove of more general use, though R values are still not reproducible on different batches of silica (K. Bailey, unpublished observation). The use of paper chromatograms using buffer solutions has also been suggested (Blackburn and Lowther, 1950; Monnier and Penasse, 1950) though details have not been reported. Recently Partridge and Swain (1950) have obtained excellent separations using butanol adsorbed on rubber as the stationary phase and buffer solution as the mobile phase. Such a system with an aqueous moving phase would be expected to be more satisfactory for compounds which are more soluble in organic solvents than in water. b. N-Terminal Residues of Proteins. I n Table I arelisted the N-terminal residues of a number of proteins as determined by the DNP-technique. Besides reacting with the a-amino groups of proteins, FDNB also reacts with the eamino groups of the lysine residues, and an estimation of the e-DNP-lysine in the hydrolysate indicates how many of these amino groups are free in the intact protein. For all proteins studied reasonable

7

THE ARRANGEMENT OF AMINO ACIDS I N PROTEINS

agreement is found between the e-DNP-lysine and the total lysine content of the protein, if the protein is first denatured. It thus seems unlikely that any branching of peptide chains occurs through the €-amino TABLEI N-Terminal Residues of Proteins Protein

Assumed mol. wt.

Insulin" (Ox,pig, sheep)

12,000

Hemoglobinb (horse, donkey) Hemoglobinb (ox, sheep, goat)

66 ,000 66,000

Hemoglobinb (human adult) Hemoglobinb (human fetal) Myoglobinb (horse) Myoglobinc (whale) B-Lactoglobulind OvalbumineJ ?-Globulin' (rabbit) Edestino

66,000 66,000 17,000 17,000 40,000

Salmineh ClupeinA Myosini Tropomyosini Lysozyme j Pancreatic Trypsin Inhibitork Serum albumin' (human, horse, ox) ~-

-

160,000 300,000 -

-

14,000 9 ,000 69,000

N-Terminal residue

Number per molecule

Phenylalanine Glycine Valine Valine Methionine Valine Valine Glycine Valine Leucine None Alanine Glycine Leucine Proline Proline None None Lysine Arginine Aspartic acid

2 2 6 2 2 5 2-3 1 1 3 1 6 1 ? ? -

-

< 1 1 1

~~~

Sanger (1945). b Porter and Sanger (1948). c Schmid (1949). d Porter (1948). 0 Deanuelle and Cmal (1948). f Porter (1950a). 0 Sanger (1949d). * Felix et al. (1950). Bailey (1951). 1 Green and Schroeder (1951). k Green and Work (1951). 1 van Vunakis and Brand (1951). 0

groups of the lysine residu s. In ertain native proteins (e.g., p-lactoglobulin) some of the e-amino groups do not react, indicating that they are in some way masked in the native molecule (Porter, 1948). c. N-Terminal Peptides. The DNP technique may also be used to identify and estimate N-terminal peptides (Sanger, 194933). Thus if a DNP-protein is only partially hydrolyzed, one obtains DNP-peptides ~~

8

F. SANGER

which may be extracted into an organic solvent and fractionated on suitable chromatograms. Complete hydrolysis of these purified DNPpeptides then reveals the nature of the amino acids and the N-terminal residue present, and amino acid arrangement may be determined by a second partial hydrolysis. In this way it is possible to identify the residues that occupy positions in the polypeptide chains near to the N-terminal residues. So far it has been possible to identify peptides up to about four residues long. Longer DNP-peptides are more difficult to separate from the other unsubstituted peptides and to fractionate, though the possibilities have not yet been fully explored. For an illustration of the use of these methods the reader is referred t o the section on insulin (p. 50). 2. The Phenylthiocarbamyl Method

Recently Edman (1950) described a method for determining the sequence of amino acids in a polypeptide chain, by splitting off one residue at a time starting from the N-terminal residue. The principle of the method is illustrated by the following equations:

0’ NCS

NHzCHCO-pep.

--+o

NHCSNHCHCONH-pep.

Pyridine

---o

Phenylisothiocyanate Peptide

Anhydrous

IICl

-N---cS

CO i

‘CL R Phenylthiohydantoin

I

R

Phenylthiocarbarnyl (PTC)-peptide

-I- NH2-pep. NHI. alkali

R

I

NH2-CH-COOH N-Terminal amino acid

The formation of the phenylthiohydantoin does not require the presence of water, as does the hydrolysis of peptide bonds. Thus by heating the PTC-peptide with anhydrous HC1 in nitromethane i t is possible to break off the N-terminal residue as a phenylthiohydantoin without splitting other bonds in the peptide. The hydantoin dissolves in the nitromethane and is then separated and hydrolyzed to the amino acid which may be identified by paper chromatography. The rest of the peptide with the N-terminal residue removed is insoluble in nitromethane and the process may be repeated. The second residue is thus split off and identified. This method has given excellent results with synthetic peptides and it will be interesting to see how far it may be applied to a

THE ARRANGEMENT OF AMINO ACIDS I N P R O T E I N S

9

protein. Theoretically, it should be possible to determine the complete structure of a pure single chain polypeptide or protein. At least the method should be extremely valuable for working out the structure of smaller peptides.

3. T h e Thiocarbamate Method Another method for the stepwise degradation of a polypeptide chain has been suggested by Levy (1950). Here the N-terminal residue is split off as a 2-thiothiaaolid-5-one derivative as follows: R

R

AHCONH-pep. ""f, AHCONH-pep.

I

Alkaii

NH2

1

NHCSS-

R pH 3-4 --+

I

CH-CO NI H S1

+ NHt-pep.

'CL Peptide

Thiocarbamate

2-Thiothiazolid-5-one

Here again the process may be repeated on the peptide chain containing one residue less. This method has not yet been worked out on a small scale, but was found to give satisfactory results with synthetic peptides. The reaction of C S z with amino groups has also been used by LBonis (1948) as the basis for a titration method for estimating the total a-amino groups as well as c-amino, imino, and thiol groups. CS2 reacts about ten times as rapidly with a-amino groups as with e-amino groups, so that the two may be distinguished. 4. T h e Use of S-Methylisothiourea Christensen (1945) has used S-methylisothiourea to study the free amino groups of tyrocidine. The amino groups are converted to guanidine groups which may be estimated by the Sakaguchi reaction. This method is especially useful for detecting the free amino group of ornithine, which is converted to arginine and thus may be estimated using arginase. In the case of tyrocidine the Sakaguchi reaction was negative after the action of arginase, indicating that the only free amino groups were the b-amino groups of ornithine, and that tyrocidine was thus a cyclopeptide. It is doubtful if this method could be used to identify the N-terminal residue of more complex proteins. 5. C-Terminal Residues

a. Carboxypeptidase. Specificity studies (reviewed in Neurath and Schwert, 1950) indicate that carboxypeptidase attacks only those peptide bonds that are adjacent to a free a-carboxyl group. It appears that all

10

F. SANGER

such bonds with the possible exception of those involving glycine are attacked to some extent although the rate of hydrolysis varies greatly with the nature of the residue involved. Thus if a protein or peptide is treated with carboxypeptidase the first amino acid to be liberated in the free form is the C-terminal residue. Lens (1949) has applied the method to insulin. Samples of the digest were removed at various intervals, ultrafiltered and the ultrafiltrate analyzed by paper chromatography. Free alanine was clearly liberated first and was therefore present in insulin as a C-terminal residue. With a single pure polypeptide chain it should theoretically be possible to determine the complete sequence of residues by following the rate of liberation of different amino acids under the action of carboxypeptidase. No such experiments have been described, but it would be interesting to know how far this method could be applied in practice. Clearly it would be impossible to draw any conclusions beyond the C-terminal residues if more than one peptide chain were present, as in the case of insulin. b. Reduction to /3-Amino AZcohoZs. Fromageot et al. (1950) and Chibnall and Rees (1951) have independently worked out techniques for the identification of C-terminal residues by reduction to 8-amino alcohols. I n the former method lithium aluminum hydride is used to reduce the protein. After hydrolysis, the p-amino alcohols are extracted into ether, identified by paper chromatography and estimated by reaction with periodate. Some reduction of peptide bonds was observed, and the yields were rather low. In the method of Chibnall and Rees, the carboxyl groups are first esterified with diazomethane and then reduced with the less violent reagent, lithium borohydride. The amino alcohols are separated from the amino acids in the hydrolyzate by electrodialysis, and treated with periodate, which decomposes them according to the equation :

A determination of the extra formaldehyde and ammonia produced in this reaction gives an estimation of the number of free a-carboxyl groups, and the residue on which they are located is identified from the nature of the amino alcohol and of the aldehyde RCHO. Chibnall and Rees have also used this technique to determine the distribution of the protein amide groups between asparagine and glutamine. Residues of aspartic or glutamic acid which have a free w-carboxyl group are destroyed by the above treatment, while those in amide form remain intact and can be estimated after hydrolysis of the protein.

THE ARRANGEMENT OF AMINO ACIDS I N PROTEINS

IV. METHODSFOR

THE

11

DEGRADATION OF PROTEINS

1. Introduction

At present the only residues to which definite positions in the protein chain can be assigned, are those at or near the terminal positions, and these residues constitute only a small part of the molecule. The position of other residues in the chain can only be determined in the first place relative t o one another by the identification of products of partial breakdown. The absolute location of any one residue can be decided only when the relative positions of all the residues are known and when the complete structure of the protein is thus worked out. Thus in determining the amino acid sequence of a protein or polypeptide the main task would seem t o be to identify as many degradation products as possible. There are essentially two ways of approaching this problem, the first requiring non-specific methods of degradation and the second specific methods. a. Non-Specific Degradation. One method of determining amino acid sequence is to degrade the polypeptide or protein directly to a mixture of small peptides whose structure can be determined and to work out a unique sequence by fitting together the degradation products. It is possible to determine directly the structure of di- and possibly tripeptides. To obtain a unique solution for the amino acid sequence of a polypeptide it is necessary to obtain as many different small peptides as possible. This may best be achieved by using unspecific methods of degradation to give a mixture of maximum complexity and using reagents having different specificities. If we consider a hypothetical polypeptide whose structure may be written as A.B.C.D.E.F.G.H. where the letters represent different amino acid residues, degradation to the dipeptides AB,BC,CD,DE,EF,FG,GH would give a unique solution. If only AB,DE,FG and GH are obtained, various structures are possible, but if a different type of hydrolysis yields EF and the tripeptide BCD then the structure is determined. This method will generally be applicable to simpler peptides. I n the case of larger peptides, the complexity of the mixture will render fractionation difficult and a proportion of larger peptides will be required to work out an unambiguous result. b. Specific Degradation. Most proteins contain a t least 100 residues. It seems unlikely that a complete solution of their amino acid sequence will ever be obtained solely by the non-specific method and it will probably be necessary t o approach the problem by the second method, involving a gradual specific breakdown into a small number of large fragments which can be purified and again broken down to smaller

12

F. SANGER

products. Thus if the peptide A.B.C.D.E.F.G.H. could be split into two products A.B.C.D. arid E.F.G.H. which could be separated, their structure could be determined much more readily than that of the original peptide. In the choice of a suitable method of degradation three main factors are to be considered: 1. The reagent should cause a minimum of side reactions such as destruction of the constituent amino acids, or if such reactions do occur they should occur quantitatively to produce known products. 2. The reagent should exhibit the desired specificity. Specific methods are required for the initial breakdown and several relatively non-specific methods are needed for the final elucidation of sequence. 3. There should be a minimum of synthesis or rearrangement of peptide bonds under the influence of the reagent. The first two points will be considered in connection with the different methods of hydrolysis, and the last will be discussed separately on page 15. 2. General Aspects of Partial Hydrolysis By far the most important method of breaking down proteins is by hydrolysis of the peptide bonds. All the peptide bonds in a protein are susceptible to hydrolysis but there are great differences in the stability of the different bonds depending on the nature of the residues involved. a. Yield of Peptides. If we assume that 1 mole of a polypeptide A.B.C.D.E.F is hydrolyzed to such an extent that the mole fractions of the bonds A-B, B-C, C-D etc. that are split are b,c,d etc. respectively, the yield of the amino acid C will be cd moles, since only those molecules of the polypeptide in which the bonds B-C and C-D are broken will give rise to C. T o obtain the dipeptide CD the bonds BC,DE must be broken and CD unbroken, so that the yield will be c(l - d)e. In general the yield F of a peptide A1.A2.A.3 . . . A,. is given by:

where a1,az . . . a, are the mole fractions of the bonds involving the amino groups of A1, Az, . . . A,,, respectively that are split, and F is expressed in moles of peptide. This treatment assumes that the rate of hydrolysis remains constant as the reaction proceeds, which is not entirely true in all cases. Equation 1 indicates that the yield of a peptide depends not only on the lability of the bonds involved in its terminal residues but also on the stability of the bonds within the molecules. It is also evident that in

T H E ARRANGEMENT OF AMINO ACIDS IN PROTEINS

13

general the yield will be greater if the peptide is a terminal one since a1or a,+l is already a maximum ( = 1). Various workers (Kuhn, 1930; Montroll and Simha, 1940; Warner, 194213; Myrback, 1949) have considered the mathematical treatment of the breakdown of high molecular chains assuming that all bonds were broken a t the same rate, i.e., that a1 = a2 = a, etc. Equation 1 then becomes : F = a2(1 - a)"-' (2) Assuming the chain t o be of infinite length or cyclic, the fraction of the original chain appearing as peptides containing n residues is given by:

F,

=

na2(1 - c ~ ) ~ - - l

(3)

It is difficult to know how far this type of treatment can really be applied t o proteins where there is such great variation in the susceptibility of various bonds. Presumably in a large molecule where the intrinsic rates of hydrolysis are fairly evenly distributed about the mean, equation (3) would apply. It is interesting to note that for any value of a in equation (2), F is a maximum if n = 1 and decreases as n increases, in other words the molar yield of smaller peptides is always greater than the yield of larger peptides. On the average this is also true for equation (l), which expresses more closely the situation present in a protein, since F is never greater than alan+l;however, as each a is different it will not apply to every case and the yield of certain higher peptides may be greater than the yield of certain smaller ones. Though the values for a1,az, etc. in equation (1) can of course be expressed in terms of the hydrolysis constants for the separate bonds, the treatment becomes extremely complicated as a different constant will be required for each polypeptide in which the particular bond occurs. In other words, the constant will vary as the reaction proceeds. Thus Kuhn et al. (1932) found that a three line formula was required t o express the rate of hydrolysis of tetraglycine. Thus it does not seem that any rigid mathematical approach can be given a t present, and we shall have to be content with a few generalizations derived largely from experimental observations. b. Complexity of Partial Hydrolyzates. When a protein is partially hydrolyzed, a very complex mixture of peptides is produced, the exact complexity of which is difficult to assess. If we consider an open polypeptide chain consisting of N residues, complete hydrolysis will give rise t o N amino acids, many of which may

14

F. SANGER

be identical. If it is partially hydrolyzed, the number of possible dipeptides is N - 1, of tripeptides N - 2 and of n-peptides (ie., pep1, very few of which are likely t o be tides with n residues) N - n identical. The total number of possible peptides is N ( N 1)/2. Clearly the shorter the time of hydrolysis the greater will be thenumber of higher peptides present in significant amounts. From equation (1) (p. 12) it was inferred that the yield of the higher peptides is always less than that of the smaller ones, so that the shorter the time of hydrolysis, the greater will be the complexity of the mixture. When a protein is hydrolyzed, there is only one molecular species t o start with. The number of species present in significant amounts then increases rapidly t o a maximum and gradually falls off. The initial rise in complexity will depend on what one considers to be a significant amount. Some splitting of each bond starts immediately with the instantaneous production of small amounts of all the N ( N 1)/2 different peptides. If one defines the “significant amount,” then for a while the only species will be the original protein, and others will gradually be added to it. It is clear however that there is really no phase during the hydrolysis when the complexity of the mixture is increasing, and that it is impossible to make use of this apparently simple composition during the initial phase of hydrolysis to obtain a mixture in which only a few polypeptides are present (‘in significant amount,” as there will be too many others present in insignificant amounts. As an example we may consider a protein such as ovalbumin which is probably a cyclic polypeptide of 400 residues. If the hydrolysis were such that all bonds were split a t the same rate, then during the initial phase of hydrolysis there will be 160,000 different chemical species produced, and this number will gradually decrease to the 20 free amino acids. I n the presence of any agent, such as acid or alkali, which splits all bonds t o some extent, all the 160,000 will be produced, but some in negligible amounts depending on the specificity of the reagent. Bull and Hahn (1948) have suggested that when ovalbumin is hydrolyzed by strong acid about 50 bonds are readily broken and that the rest are broken more slowly, say a t one tenth of the rate. Consider first only the initial phase in which the 50 labile bonds are broken. When they are all broken they should give rise to 50 peptides of average length 8 residues. At the beginning of hydrolysis 502 = 2500 different combinations of these peptides will-be produced the number gradually falling to the 50 octapeptides. At the same time, however, all the other bonds within these octapeptides have been subject t o hydrolysis a t one tenth of the rate, so that 3i0 of these octapeptides are broken down and each will 1) = 36 different degradation products. The give rise t o Y$(S

+

+

+

+

T H E ARRANGEMENT O F AMINO ACIDS IN P R O T E I N S

15

hydrolysis will thus have the following composition. One-fifth will be in the form of 50 octapeptides and the remaining four-fifths in the form of 50 X 36 = 1800 smaller peptides. I n reality the situation is even more complex since there are not two types of bonds but bonds with every type of stability. However, the above example does make it clear that at no stage in the hydrolysis, except near the end, will the composition of the hydrolyzate be sufficiently simple to justify investigation. In the case of enzymic hydrolysis, a large proportion of bonds are unattacked and here again the simplest mixture and the most profitable one to investigate is the complete hydrolyzate. At earlier stages in the hydrolysis the mixture will be increasingly complex. The only other means of obtaining a relatively simple hydrolyzate would arise if there was a very sharp break in the hydrolysis curve, that is to say, if there are two types of bonds with very different labilities. For instance, if the 50 labile bonds in ovalbumin were hydrolyzed 100 times as rapidly as the other bonds, one would at a certain stage obtain a mixture in which the 50 octapeptides are present to the extent of 90%. However, it is unlikely that any one of the commonly-used enzymes could bring about such a sharp differentiation. Since methods of fractionation a t present available may be capable of separating 100 short peptides but not 1000 long peptides, it would seem advisable to confine our attention t o the later stages of hydrolysis by any agent. Special emphasis is laid on this point as it appears to be quite a common practice for protein chemists to attempt to determine the nature of proteins by splitting them to a few large peptides by partial hydrolysis, on the assumption that by starting with one compound and ending with 20, a t some stage there must be only two or three compounds. This is true only when the first molecule in the solution is split. The chance that the second molecule will split in the same place is rather remote. c. The Question of Rearrangement of Peptide Sequences during Hydrolysis. It is clear that if any synthesis or rearrangement of peptide bonds takes place during the course of hydrolysis of a protein, the amino acid sequences identified in products of partial hydrolysis may not be the actual sequences that were present in the original protein, so that this approach to the problem would be useless. The hydrolysis of the peptide bond is reversible, so that the theoretical possibility exists that any peptide bond may be synthesized. The free energy of formation of a peptide bond is probably about 3000-4000 calories per mole for a dipeptide (Huffman, 1942) and 2000 calories per mole for a peptide bond within a protein (Haugaard and Roberts, 1942) so that the equilibrium will be very much on the side of hydrolysis for most peptide bonds, and a direct reversal of hydrolysis would seem rather unlikely.

16

F. SANGER

Synthetic reactions are more likely to take place if conditions are such that the product of the reaction is rapidly removed. Bergmann and his colleagues have shown that in the presence of certain proteolytic enzymes such conditions may be obtained either when the synthetic product is insoluble and is removed from the solution by crystallization (Bergmann and Fraenkel-Conrat, 1937, 1938; Bergmann and Behrens, 1938; Bergmann and Fruton, 1938) or else when it is removed by rapid hydrolysis to other products (Behrens and Bergmann, 1939). The former possibility may be avoided by keeping the reaction mixture always in solution. However the latter possibility, is rather more difficult to eliminate. Such a reaction may be formulated as follows: A.B

+ C.D % A.B.C.D + A.B.C + D

+

If the reaction A.B.C.D -+A.B.C D is very rapid it will shift the equilibrium of the other reaction to the right and thus bring about the Such a reaction will only occur to an synthesis of the bond B-C. appreciable extent if the hydrolytic reagent has a much greater affinity for the bond C--D in the peptide A.B.C.D than in C.D, that is to say if its specificity is determined not only by the residues involved in the susceptible bond, but by residues further removed from it. This would seem to be possible in the case of hydrolysis by proteolytic enzymes but less likely when acid or alkali are used (Sanger, 1949~). Fruton (1950) has recently pointed out the possibility that “transpeptidation )’reactions may occur in the presence of proteolytic enzymes. Such reactions may be represented as follows: R’ RCO-NH

A

HCO-NHR”

R’

+ NH,R”’ % RCO-NH&HCO--NHR”’ + NHJ-l”

It was originally shown by Bergmann and Fraenkel-Conrat (1937) that benzoylglycylanilide may be synthesized more rapidly by papain from benzoylglycinamide than from benzoylglycine, thus indicating that a direct transformation of the amide to the anilide occurs without intermediate formation of the acid. Following up this work Johnston et al. (1950) showed that if papain is allowed to act on benzoylglycinamide in the presence of NH, containing N16, a small amount of the isotope is introduced into the amide. Clearly, if this type of reaction which involves very little change in free energy occurs to any great extent during a partial hydrolysis of a protein it may lead to rearrangement of peptide sequences. Another type of rearrangement that might occur is through the amino acid anhydrides or diketopiperazines. The formation of these stable six-membered rings presumably involves a smaller free energy of forma-

THE ARRANGEMENT O F AMINO ACIDS IN PROTEINS

17

tion than the synthesis of most other peptide bonds and takes place readily under various conditions. If such conditions are present the following equilibrium will exist in the hydrolyzing solution

A where A.B represents a dipeptide and [ ] the corresponding anhydride. B The relative rates of the different reactions may be such as to cause the partial or complete inversion of the order of the residues in the dipeptide, even if the anhydride intermediate never appears in the solution. Anhydrides are very much more labile both to alkali (Levene et al., 1930, 1932; Kuhn et al., 1932) and to acid (Abderhalden and Mahn, 1927,1928), than the corresponding dipeptides, presumably since they contain no free charged group. This property tends to accelerate any possible inversion of a dipeptide. The formation of diketopiperazines seems to be associated with high temperatures. Brig1 (1923) and Abderhalden and Komm (1924a, b) ware able to convert dipeptides to anhydrides in quite high yield by heating in water or dilute acids a t temperatures of 150-250’. No conversion could be demonstrated when the strength of acid was greater than 1 M . Small amounts of anhydrides mere formed when the dipeptides were refluxed in water for several days. No reaction could be demonstrated under the action of 70% HzS04 or concentrated HC1 a t room temperature. This does not necessarily prove that no anhydrides were formed, as they may have been broken down as fast as they were formed. However these results do at least indicate the possibility of anhydride formation and inversion under certain conditions. Clearly the question of whether rearrangements do actually take place in partial hydrolysis experiments can only be solved by experience. If any such reactions occur it would be impossible to interpret results in terms of a unique sequence of amino acids unless the syntheses are completely specific and quantitative, which is most unlikely. I n fact, using concentrated acid at low temperatures it has been possible to work out a unique sequence for “gramicidin S ” (Consden et al., 194713 from the peptides identified, and there was no evidence of any peptide that did not fit this sequence. Similarly a unique structure could be determined for the phenylalanyl chains of insulin (p. 54). It thus seems unlikely that any rearrangement occurs under the action of this type of reagent. On the other hand, syntheses have been definitely shown to occur in the presence of proteolytic enzymes. The formation of plastein by the action of pepsin or trypsin on concentrated peptide mixtures has clearly been shown in certain cases t o be accompanied by a decrease in amino

18

F. SANGER

nitrogen, which can only be interpreted as a net synthesis of peptide bonds (reviewed by Wasteneys and Borsook, 1930; Virtanen et al., 1950) and the results with synthetic subst,rates are even more clear-cut, although in no case have naturally-occurring peptides been used as substrates. I n conclusion it may be said th at while the theoretical possibility exists that synthetic reactions may occur during the partial hydrolysis of proteins, it is the opinion of the reviewer tha t future research will show tha t such syntheses are insignificant and will not interfere with the interpretation of data derived from partial hydrolysis experiments in terms of protein structure. 3. Hydrolysis in Concentrated Acid Acid hydrolysis is the most generally used method of degrading proteins, and i t is almost universally employed when amino acids are to be isolated or estimated, since it leads t o complete hydrolysis with a minimum of destruction. The only amino acid th a t is extensively destroyed is tryptophan, the destruction of which may be largely due to the presence of traces of heavy metals during hydrolysis and may be reduced by using very pure HC1 and quartz vessels (Jacobsen, 1949; Monnier and Jutisz, 1950). Slight destruction of serine and threonine (Rees, 1946) also takes place, but in partial hydrolyzates this would be almost negligible. No synthetic reactions or rearrangements have been shown to take place under the action of strong acids. The relative rate of hydrolysis in acid of any peptide bond and hence the yield of a given peptide is determined mainly by the number of hydrogen ions that can approach the bond. While the rate probably depends on a number of different factors, we may consider two which probably play a major role, namely, electrostatic effects and steric effects. a. Electrostatic Efects. The presence of any charged groups in the neighborhood of a peptide bond will clearly affect the approach of hydrogen ions. In strong acid all the carboxyl groups on the proteins will be uncharged but all the basic groups (amino, imidazole, and guanidyl) will be fully charged and will oppose the approach of the similarly charged hydrogen ions. From their studies of the course of hydrolysis of various proteins b y concentrated acid Gordon et al. (1941) were able to calculate the ratio of free basic amino acids in the hydrolyzate to total basic residues (free and in peptide combination). This ratio gives a n average estimate of the stability of basic peptides. In general, the values found were slightly lower than the corresponding figures for the neutral residues, indicating

THE ARRANGEMENT O F AMINO ACIDS I N PROTEINS

19

that the basic peptides were slightly, though not very much more stable, than the average peptide. Thus the ratio obtained for wool a t a certain stage of hydrolysis was 0.33 and the corresponding figure for the neutral residues 0.39. Probably the charged groups that are most effective in stabilizing peptide bonds are the a-amino groups, which are closer to a peptide bond than the e-amino, imidazole and guanidyl groups. A bond involving an N-terminal residue should thus be relatively stable to acid. An important result of this is the stability of dipeptides which leads to their accumulation a t a certain stage of hydrolysis (Gordon et al., 1941). This stability is evident from the results of Stein et al. (1944), who followed the course of hydrolysis of silk fibroin in concentrated HCl a t 40" using the van Slyke nitrous acid method for estimating the rate of liberation of free amino groups and the ninhydrin method (van Slyke et al., 1941) for free amino acids. From these results it is possible to calculate the average length of the peptides excluding in the average the free amino acids. After 43 hours hydrolysis 60% of the peptide bonds were split, and the hydrolyzate contained 25% of its nitrogen in the form of free amino acids and 75% in the form of peptides whose average length was 2.05 residues; in other words almost the whole of the N in peptide linkage was assignable to dipeptides. From equation (3) (p. 13) it may be calculated that if a completely random splitting had occurred the yield of amino acids would have been 36%) of dipeptides 29%, of tripeptides 17%) etc. Similar conclusions may be drawn from the rates of hydrolysis of gramicidin (Synge, 1945; Christensen and Hegsted, 1945) the effect being more marked a t 37" than a t the boiling temperature. On the contrary, however, it was found that the yield of free amino acids during the hydrolysis of ovalbumin with 1 N HC1 was almost exactly theoretical (Warner, 1942b). b. Steric E$ects. Apart from the effect of positively charged groups, probably the most important factor influencing the rate of hydrolysis of a peptide bond is the effective size of the amino acid side chains on either side the bond, preventing the approach of hydrogen ions by steric means. The effect will be expected to depend on the actual size of the side chain and its position relative to the bond in question. Synge (1945) has made a kinetic study of the hydrolysis of a number of simple peptides by a mixture of equal volumes of 10 N HC1 and glacial acetic acid a t 37". His results are shown in Table 11. The most stable peptides appear to be those containing valine. The bulky CH3CHCH3group is close t o the main peptide chain and effectively prevents the approach of hydrogen ions from a fairly wide angle.

20

F. SANGER

I n leucine the CH,CHCH3 is slightly further from the peptide hond, so that leucyl peptides are less stable than the corresponding valyl peptides, and peptides containing alanine and glycine are still more labile. ,4 side chain seems to be less effective in stabilizing the peptide bond if it, is on the residue whose amino group forms part of the bond (Levene et al., 1932). Val.Gly for instance, is more stable than Gly.Va1, though it should be noted that these generalizations are derived only from a study of peptides of glycine. TABLErr Hydrolysis of Dipeptides in Strong Acid (Sy nge , 1945)

Peptide Gly.Gly Gly.Ala Ala.Gly Gly.Leu Gly.Try Gly.Va1 Leu. Gly Leu.Leu IAeu.Try Val.Gly

Relative velocity of hydrolysis (G1y.Gly = 1)

1 0.62 0.62 0.40 0.35 0.31 0.23 0.048 0.041 0.015

The marked stability of valine peptides has frequently been noted. Thus Christensen (1943) was able to isolate Val.Va1 in 1.5% yield from gramicidin after boiling for 24 hours with 16% HC1, and in 5-6%) yield after 6 hours hydrolysis. Synge (1944) could find no free valine in partial hydrolyzates of gramicidin that had been treated with 5 N HCI for 10 days. The stability of valyl peptides was also apparent from the work on the partial hydrolysis of insulin (Sanger and Tuppy, 1951a). Thus tripeptides containing valine as the central residue were present in higher concentrations than other tripeptides. A lability of bonds involving the carboxyl groups of glycine was also noted. Peptides containing proline also appear to be unusually stable presumably due to steric factors (Consden et al., 1947b). In this connection the unusual stability of cyclic structures should be mentioned. In the case of carbohydrates this was clearly demonstrated by Swanson and Cori (1948) and by Myrbiick (1949), who showed that the cyclic Schardinger dextrins are considerably more stable than corresponding open chain polysaccharides. Similarly Consden et al. (1947b) found the cyclopeptide “gramicidin S” to be unusually

21

T H E ARRANGEMENT O F AMINO ACIDS I N P R O T E I N S

resistant t o acid hydrolysis. It has been suggested by Synge (personal communication) that immobilization of the peptide bond from rotation may be an important factor in inhibiting hydrolysis. Such a n effect might also partly account for the relative stability of native proteins t o enzymic hydrolysis (p. 26) and possibly for the stability of proline peptides. c. Bonds Involving the Amino Groups of the Hydroxy Amino Acids. While the electrostatic and steric effects mentioned above probably play an important part in determining the stability of a given peptide bond, they are certainly not the only factors concerned and the course of hydrolysis of a protein cannot wholly be explained in this simple manner. In fact i t has been observed th at the first bonds in a protein t o be split by strong acid are those involving the amino groups of the serine and threonine residues. Abderhalden and Bahn (1935) made use of this lability t o prepare seryl peptides from the corresponding anhydrides. Under suitable conditions only one of the bonds in the anhydrides split. Gordon et al. (1941) hydrolyzed a number of proteins with strong acid and followed the rate of liberation of the free amino groups of serine and threonine residues by means of the periodate reaction. Only those residues with a free hydroxyl and amino group react with periodate t o give ammonia. The serine and threonine residues could be differentiated by estimating the acetaldehyde produced by periodate from threonine. It was found that the free amino groups of these hydroxy amino acids were liberated much more rapidly than the average amino group (estimated by the van Slyke method). Similar results were obtained by Christensen and Hegsted (1945). More recently Desnuelle and Casal (1948) followed the relative rates of liberation of the amino groups of the different amino acids using the D N P method. During the initial stages of hydrolysis with 10 N HCl a t 30" there appeared to he a very specific breakdown of the bonds involving the amino groups of serine and threonine. Thus after one hour about 20-30% of these bonds were split whereas very few other amino groups were liberated. There was no difference between serine and threonine. I n order t o account for this effect Desnuelle and Casal have suggested that under the action of concentrated HCl an intermediate oxazoline ring is formed which breaks a t the amino group: R I -NHcHcoNHcHcoHdH2

R

R

I

-NHcHc=N-cHco-

d-CH,

I

NH2

--h'HLHCo

I

&co-

O--CH2

I

I n this way the peptide chain is considered to migrate from the amino t o the hydroxy group. The ester bond then formed will be fairly

22

F. SANGER

rapidly broken down. Such reactions have been shown to take place with simple compounds such as benzoyl serine (Bergmann and Miekeley, 1924) under the action of chlorinating agents, e.g., thionyl chloride. I n support of the above mechanism it was found that the rate of liberation of the free amino groups of the hydroxy amino acids was greater when estimated by the DNP method than when estimated by the periodate method, thus suggesting that the hydroxy group was being simultaneously masked. To sum up, when a protein is hydrolyzed with strong acid we may expect to find an initial rather specific hydrolysis liberating the free amino groups of the hydroxy amino acids followed by a more random breakdown a t different bonds, the relative rates depending largely on the nature of the residues involved, and there is likely to be a slowing up of hydrolysis and accumulation of dipeptides towards the end of the reaction. In general, the specificity of hydrolysis will be greater a t lower temperatures, since the activation energies for hydrolysis of the various bonds will be expected to show greater differences relative to the mean thermal energy of the molecules. This is evident in the results of Christensen and Hegsted (1945), who found a more random splitting at higher temperature. Desnuelle and Casal (1948) also found that the liberation of hydroxy amino residues was much more specific a t lower temperatures. 4. Hydrolysis in D i l u t e Acid While concentrated acid is usually preferred as a hydrolytic agent, it may be advantageous in certain cases to use dilute acid, which seems to exercise a rather different kind of specificity. Thus for instance very poor yields of the N-terminal glycyl peptides of insulin were obtained after hydrolysis in concentrated HC1, whereas much higher yields were obtained by boiling in 0.1 M HCl (Sanger, 194913). This difference in specificity may to some extent be ascribed to differences in the charged groups. In sufficiently dilute acid the acidic groups may become negatively charged, so that they will attract hydrogen ions and labilize any nearby peptide bonds. An attempt t o utilize this effect t o obtain a specific hydrolysis near the cysteic acid residues of oxidized insulin gave no clear-cut results (Sanger, 1949b). A very specific hydrolysis has, however, been demonstrated by Partridge and Davis (1950). They found that when a protein was boiled with acetic or oxalic acid and the hydrolyzate investigated by paper chromatography, free aspartic acid was liberated much more rapidly than any other amino acid. For most proteins during the first 8 hours of hydrolysis the only ninhydrin positive spot was that due t o aspartic acid. Free glutamic acid was the next

THE ARRANGEMENT O F AMINO ACIDS IN PROTEINS

23

amino acid to appear, but a t a very much slower rate. While it is possible that other bonds besides those next to the aspartic acid residues are also split, this type of hydrolysis seems to exhibit a marked specificity and may prove a useful method for obtaining larger polypeptides from proteins. Partridge and Davis suggest further that the presence of a free carboxyl group should labilize the C-terminal residues of a peptide chain and it may thus be possible to effect a stepwise liberation of residues from the C-terminal group. The main disadvantage of using dilute acid for hydrolysis is the possibility of anhydride formation, already referred to (p. 16). Another type of hydrolysis that may exhibit a different specificity is that catalyzed by long-chain anionic detergents. Steinhardt and Fugitt (1942) showed that the rate of hydrolysis of proteins by acid was dependent, not only on the concentration of hydrogen ion, but also on the nature of the anion present. For a homogeneous series of long chain sulfonic acids the rate of hydrolysis was proportional to the chain length of the anion. Thus the hydrolysis proceeded about a hundred times as rapidly in dodecylsulfonic acid as in HC1. It seems that the detergent is adsorbed on the protein and the presence of the -SO3- group labilizes the peptide bonds. At low concentrations of detergent the amide groups are especially labile to this type of hydrolysis. Being slightly stronger bases than the peptide bonds, they preferentially bind the anions (Steinhardt, 1941). Thus it was possible to remove amide groups with very little destruction of peptide bonds. This suggests a different type of specificity from other methods of hydrolysis. The lability of amide groups, however, seems to be a characteristic of low temperature hydrolysis in dilute acids, since Virtanen and Hamberg (1947) hydrolyzed zein at pH 1.5-1.8 and 37" for 4 months and found that 50% of the amide groups were split off while very little amino N was liberated. An obvious advantage of using this method of catalyzed hydrolysis is that there will be a minimum of destruction of the amino acids or of any bonds other than the peptide bonds. A disadvantage of the method is the practical difficulty of working with solutions of these strongly surface-active reagents. 5. Hydro&& in Alkali The main disadvantage of alkaline hydrolysis is that much more destruction of amino acids occurs than with acid. The chief amino acids to be affected are cysteine, serine, threonine and arginine. Wieland and Wirth (1949) have used paper chromatography to study the effect of strong alkali on certain amino acids. They found that serine broke down to give appreciable quantities of glycine and alanine, threonine

24

F. SANGER

gave glycine, alanine and a-amino butyric acid and cysteirie gave alariine. While relatively violent conditions were required for these reactions, they were shown to take place more readily if the amino acids were in peptide form. Hellerman and Stock (1938) and Warner (1942a) showed that, arginine is broken down by strong alkali to give ornithine and citrulline. Sanger and Tuppy (1951a) studied a n alkaline hydrolyzate of a n oxidized insulin fraction. Although the conditions used would not have been expected t o cause any destruction of free serine or threonine, it was evident that most of these residues in the protein had been broken down. Thus, for instance, Gly.Pro was found in the alkaline hydrolyzate whereas Thr.Pro was present in the acid hydrolyzate. The arginine residues were also converted to ornithine (or citrulline). Extensive racemization of the amino acids also occurs in the presence of alkali, which may complicate the results of a partial hydrolysis experiment (Levene and Bass, 1928, 1929). I n spite of these obvious disadvantages i t may, in certain cases, be worth while to employ alkaline hydrolysis, which may exhibit somewhat different specificities from those found for acid hydrolysis. Synge (1945) compared the relative rates of hydrolysis in acid and in alkali of a number of simple dipeptides of the monoamino acids, and found them to be similar in both reagents. I n these experiments the relative stabilities of the peptide bonds were probably determined largely by the steric effects produced by the side chains of the residues, which inhibit the approach of hydroxyl ions, and these effects will be similar for both acid and alkali hydrolysis. However, where the effects of charged groups are concerned one may expect to find a somewhat different specificity. I n strongly alkaline solution the basic groups are uncharged, whereas the carboxyl groups will be negatively charged and will stabilize bonds in their neighborhood. Similarly, dipeptides would be stable in alkaline solution due to the presence of the charged a-carboxyl group. It is also probable that the peptide bonds involving the amino groups of the hydroxyamino acids are more stable in alkali than in acid. Abderhalden and Bahn (1935) showed that whereas anhydrides containing serine were split by acid at the bond involving the amino group of the serine residue, they were split at the other bond by alkali. It was also found by the D N P method (Sanger, unpublished) that the relative rate of liberation of the amino group of serine from fraction A of oxidized insulin was slower in alkali than in strong acid. Alkaline hydrolysis offers a possible advantage for the investigation of tryptophan peptides since tryptophan itself is more stable in alkali than in acid (Lugg, 1938; Brand and Kassell, 1939). The possibility that long chain bases, such as hexyl trimethyl ammo-

T H E ARRANGEMENT O F AMINO ACIDS I N P R O T E I N S

25

nium, will show catalytic effects similar to those found for long chain anions is suggested by the work of Steinhardt and Zaiser (1950), who showed that they are also bourd t o protein and cause anomalous titration effects. It may be that the mild conditions necessary for such a n hydrolysis may make it possible t o avoid the undesirable side reactions that occur during alkaline hydrolysis.

6. Hydrolysis with Proteolytic Enzymes The action of proteolytic enzymes on proteins and synthetic pcptides has been extensively studied and reviewed (see Bergmann and Fruton, 1941; Bergmann, 1942; Neurath and Schwert, 1950; Linderstrgm-Lang, 1949) so t ha t the subject will not be dealt with in great detail here. One obvious advantage of using proteolytic enzymes is th a t they are very unlikely t o cause any destruction of amino acid residues, since they act under very mild conditions. Another important advantage is their specificity. Not only do they exhibit an entirely different specificity from t ha t shown b y acid and alkali but a specificity which is much more limited and exacting. Enzymes should therefore be useful for the initial specific splitting of proteins into large peptides. The finding of synthetic substrates of known structure that are hydrolyzed by the endopeptidases has made it possible to predict what bonds will be susceptible to particular enzymes. This subject has been reviewed in detail by Bergmann and Fruton (1941) and by Neurath and Schwert (1950). I n brief it may be said that trypsin splits those bonds in which the carboxyl groups of arginine or lysine are involved, chymotrypsin those involving the carboxyl groups of tyrosine, phenylalanine, tryptophan or methionine, and pepsin those involving the amino groups of the aromatic amino acids. It seems t ha t the rate at which synthetic substrates are split is considerably slower than the rate a t which peptide bonds in proteins are split (Northrop et al., 1948). The question therefore arises as to whether bonds of the type present in the synthetic peptides are the most labile in a protein or whether there are other types of bonds that are broken down. I n the case of pepsin, there is evidence that other types of bonds are also split. Thus Harington and Pitt-Rivers (1944) found that pepsin would hydrolyze the peptide Tyr.CySH, in which the bond involves the carboxyl group and not the amino group of tyrosine. Recently Desnuelle et al. (1950) have studied the action of pepsin on ovalbumin and on horse globin using the D N P technique. If the above specificity were the only one, the only free amino groups liberated should be those of the aromatic residues. I n fact this was not the case. For ovalbumin no specificity could be detected, and free amino groups of almost all amino acids were liberated simultaneously. With globin the bonds split first were those

26

F. SANGER

involving the amino groups of alanine, phenylalanine, leucine, and serine ; during a second slower phase of hydrolysis no specificity was apparent. I n a study of the action of proteolytic,enzymes on a n oxidation product (fraction B, p. 54) of insulin, Sanger and Tuppy (1951b) found that other bonds besides those adjacent to aromatic residues were split by pepsin, including those of Leu-Val, Ala-Leu, and Glu(-NH&His. It thus seems that in the case of pepsin at least there is much to be learnt about its specificity when proteins act as substrate. Trypsin and chymotrypsin were found t o split oxidized insulin with the same specificity as was found for synthetic peptides, and it seems probable th a t this specificity may be shown in their action on other proteins. Clearly a knowledge of the exact mode of action of these enzymes would greatly help in the elucidation of protein structure just a s advances in our knowledge of protein structure must throw light on the behavior of the endopeptidases. The possibility of rearrangement of sequences of amino acids under the action of proteolytic enzymes has already been mentioned (p. 15) but this danger would not seem sufficiently great to offset the advantages to be gained by using them. Nevertheless, the results must be interpreted with caution. It has already been emphasized (p. 14) that the best stage of hydrolysis a t which t o attempt the fractionation of peptides is at the point where enzyme action will proceed no farther or when there is a very sharp break in the hydrolysis curve. The disadvantage of a long incubation period is of course that the danger of rearrangements increases (see Linderstrpim-Lang and Ottesen, 1949). It may be that proteolytic enzymes act more specifically on native than on denatured proteins. The work of Linderstrgm-Lang and coworkers (reviewed by Linderstrom-Lang, 1949) has indicated th a t a t least in the case of the action of trypsin on 0-lactoglobulin, the initial step in proteolysis is some type of denaturation. Nevertheless, certain hydrolyses proceed without denaturation of the substrate. Examples are the formation of plakalbumin from ovalbumin (p. 57) a n d the splitting of globulin molecules by papain (Petermann and Pappenheimer, 1941; Petermann, 1942), which appear to be rather specific re:tctions. Presumably only a few susceptible sites are available to the enzyme on the native protein, and if these can be split without denaturing the protein and exposing the other susceptible sites, a specific type of hydrolysis may take place. 7 . Rates of Hydrolysis of Proteins I n experimental work it is very often desirable t o know how far a protein will be hydrolyzed under a given set of conditions. The course

THE ARRANGEMENT O F AMINO ACIDS I N PROTEINS

27

of hydrolysis of a number of proteins in various concentrations of acid and alkali have been recorded in the literature, though it is often difficult to find the particular and relevant data one is interested in. I n Table I11 are listed some references where such data may be found. Since protein hydrolysis follows no known kinetic laws it is impossible to define the rate in terms of any constant, so th at it is best t o refer to the original work. I n the last column are listed the approximate half-lives of the proteins in the various reagents. This is defined as the time at which 50% of the peptide bonds in the protein are broken. 8. Non-Hydrolytic Methods of Degradation

Critics of the peptide theory have claimed th a t too much emphasis has been laid on studies in which hydrolysis is used for degrading proteins, and that other methods of degradation such as oxidation and reduction should be employed. In the earlier years of protein chemistry many attempts were made to study proteins in this way but very few recognizable products could be obtained, and with those th a t could be identified it was difficult to know what was their relationship t o the original protein. This was undoubtedly due to the great complexity of the problem, the different side chains and bonds in the proteins each reacting t o give a variety of products of unknown origin. Clearly, in order to obtain any recognizable breakdown products, one must use reagents that attack only one or a very few types of bond or residue. Hydrolytic reagents, which attack almost exclusively the peptide bond, are the obvious first choice. I n the search for other reagents for degrading proteins i t would seem more profitable to consider the nature of proteins first and t o choose a reagent for a particular purpose rather than to take any reagent off the shelf and see what i t does to the protein. There is obviously a great use for reagents th at will attack a protein in a specific manner. If i t were possible to destroy one type of residue exclusively and split i t out of a protein chain, it would be a valuable step in the degradation of proteins. a. Splitting the Disulfide Bridges. Probably the only covalent linkage t ha t occurs in proteins other than the peptide bond between the amino acid residues, is the disulfide bridge of cystine; this may be split by oxidation t o sulfonic acid groups. Toennies and Homiller (1942) studied the action of performic acid on a large number of amino acids and found t ha t the only acids attacked were cystine, methionine, and tryptophan. Cystine was quantitatively converted to cysteic acid, and methionine t o the corresponding sulfone. I n the case of tryptophan the products were not identified. Sanger (1949a) used performic acid to split the disulfide bridges of insulin, which contains no tryptophan or methionine (see p. 51). This method could probably be used as a n

28

F. SANGER

TABLE I11 Rates of Hydrolysis of Proteinse

Ileferencc

Proteins studied

Reagents used

Temp. (“C.)

Ahderhaltien and Mahn. 1928 Acher et al., 1950 Bull and Hahn, 1948

Gelatin 15,38, 50,70 1 N HCI 37 Lysozyme 10 N HC1 30 Ovalbumin 7.95 N HCI 7.95 N HC1 45 7.95 N HCl 60 2.5 N HC1 Boiling Desnuelle and CasaI, 1948 Casein 37 10 N HCl 10 N HCI 30 Silk fihroin 10 N HCl 30 3.6 N HzSOa Casein Dunn, 1925 Boiling Gordon et al., 1‘341 Wool 10 N HCI 37 Edestin 10 N HC1 37 Gelatin 10 N HCI 37 Levene and Bass, 1928 Casein 125 5 N HC1 0.5, 1.0, 5 N NaOH 25 Casein Pittom, 1934 5.7 N HCI Boiling Boiling Ovalbumin 5.7 N HCI Silk fibroin 12 N HCl 40 Stein et al., 1944h Vickery, 1922 Gliadin 0.1, 0.2 N HCI 93-94 0.5 N HCI 93-94 1.0 N HC1 94-95 94-95 2.0 N HCl 4.0 N HCI 98-1 04 20% HCI 102-1 10 0.2 N HZSOa 93-94 4.0 N HzS04 96-98 0.2 N NaOH 93--94 1.0 N NaOH 93-94 0.2 N Ba(OH)2 93-94 Warner, l942b Ovalburnin I N HCl Boiling I00 20% H&O, 1.4, 4.3 N NaOH 100, 68, 35 0.43 N NaOH 100 0.2 N NaOH 68 2.3, 3.7 N Ba(OH)z 100 ‘I

b

For other earlier references see Vickery, 1922. In these experiments 1 g . protein was hydrolyzed with only 2 rnl. IICl.

Approx. half-life of protein (hours)

96 195 42 9 2 . :3 1 I5 192 144 5 95 120 35

17

35 17 7 I .8 < 1 0

16 36 4.5

THE ARRANGEMENT OF AMINO ACIDS IN PROTEINS

29

initial specific method of degrading other proteins that contain disulfide bridges, though the presence of tryptophan might lead to side reactions. Disulfide bridges can also be split specifically by mild reduction, but the thiol groups formed tend to reoxidize and cause polymerization of the products (Miller and Andersson, 1942). b . Splitting by Radiation. If a protein is irradiated with ultraviolet light of a suitable wavelength, the only residues that absorb energy are those of the aromatic amino acids phenylalanine, tyrosine, and tryptophan, and it has been suggested that the photochemical energy may be sufficient to split the peptide bonds adjoining these residues (Carpenter, 1940; McLaren, 1949; Mandl et al., 1950). Rideal and Mitchell (1937) showed that substances such as stearic anilide could be split. Here the aromatic residue is directly adjoining the peptide bond. However it was also shown that cleavage could take place when the peptide bond and aromatic residue were separated by a number of -CH2groups, since substances such as stearyl benzylamine were also split (Carpenter, 1940). Propionylphenylalanine and phenylpropionylalanine, which resemble more closely a natural peptide were both split with equal efficiency (Mandl et al., 1950). In each case the main nitrogenous end product was ammonia rather than the amino acid, indicating that deamination had also occurred. No liberation of free amino acids from Tyr.Leu or Leu.Tyr could be detected but this was probably due to deamination. However it has been shown (Carpenter, 1941; Kaplan et al., 1950) that when insulin is irradiated free tyrosine is liberated into the solution. Clearly the effects of light on a protein may be expected to be rather complex (McLaren, 1949) but it would be extremely valuable if conditions could be found for a specific photolysis near the aromatic residues. c. Other Methods. Fodor and coworkers (see Fodor, Fodor and Kuk-meiri, 1947) used anhydrous glycerol a t 130-140” to break down proteins. The products formed, which were termed “acropeptides,” appeared to be large cyclic polypeptides. Using a similar method Uchino (1934) and Tazawa (1949) reported the formation of large amounts of diketopiperazines from ovalbumin. Troensgaard (1947) has used reduction with sodium and amyl alcohol to split proteins and has isolated a number of piperazines and pyrrole derivatives. The relationship of these products to structures i n the original protein is not clear a t present.

V. FRACTIONATION OF PEPTIDES

It has already been emphasized that a partial hydrolyzate of a protein is a complex mixture of closely related compounds, so that very sensitive

30

F. SANGER

methods of fractionation are required to separate pure peptides. I n recent years countercurrent methods, especially those depending on partition effects have been found to be most useful for this purpose. Before applying such methods it is often desirable to carry out preliminary group separations of the hydrolyxate into fractions containing fewer peptides than the original mixture. For instance, not more than about twenty peptides can be satisfactorily fractionated by paper chromatography. Thus if a partial hydrolyzate of a protein were applied directly t o such a chromatogram there would probably be considerable overlapping of the spots and interpretation of the results would be difficult. Separation into a number of relatively simple groups would make fractionation much easier. We may thus distinguish between two classes of methods of fractionation, those for preliminary group separations and general methods involving the countercurrent principle. The distinction is somewhat arbitrary as most methods may be used to some extent for both purposes. The most useful methods for group separation are those which give the most clear-cut fractionations with a minimum of overlapping, the ideal being th at each peptide should be present in only one fraction. Such an ideal is rarely achieved but it is sometimes possible, as in the case of the cystine peptides, to make use of a unique property of one amino acid, t o separate out all peptides in which it is involved. It is clearly an advantage to use for group separations methods which depend on different properties of the peptides than those on which the final general fractionation depends. Thus, whereas the methods of general fractionation usually depend on differences in the partition coefficients of the peptides, methods of group separation more often depend on differences in ionophoretic mobility or in the adsorption coefficients. 1. Ionophoretic Methods Since amino acids and peptides contain several differently charged groups, they can be fractionated by methods which make use of differences in isoelectric point or electrophoretic mobility. The various techniques available for such separations have been comprehensively reviewed by Svensson (1948). A simple compartment type of apparatus may be used to separate a peptide mixture into basic, neutral and acidic fractions (Gordon et al., 1941, 1943; Sanger and Tuppy, 1951a). Since the simplification of the mixture is usually more important than the yield of peptides, it is often advisable to repeat the ionophoresis on each fraction, and in this way clear cut fractionations may be obtained with very little “overlapping.” This method is especially useful for separating the basic peptides. By

THE ARRANGEMENT O F AMINO ACIDS IN PROTEINS

31

choosing suitable conditions of p H etc. it is also possible to carry out more extensive group separations using this type of apparatus; thus, by working a t a n alkaline p H for instance it is possible to obtain a fraction containing only arginine peptides (Sanger and Tuppy, 1951a). For more extensive separations it is best to use methods th a t depend on differences in mobility. The method that has been used most is th a t of Consden, Gordon and Martin (1946) for ionophoresis in silica jelly. It was successfully used to separate the acidic peptides from a partial hydrolyzate of wool into fractions of different mobilities (Consden et al., 1949; Consden and Gordon, 1950). This type of apparatus is less suitable for peptides containing aromatic or basic residues, due t o adsorption on the silica, which causes ((tailing” and difficulties in eluting the peptides. The use of agar-agar instead of silica may prove more suitable for such peptides (Gordon et al., 1939). Svensson and Brattsten (1949) and Grassmann (1950) have described a method of ionophoresis in which the separations are carried out in a box of glass powder. The solution is allowed to flow down the box while a voltage is applied horizontally, so that the direction of flow of each solute depends on its mobility; fractions are collected from outlet tubes a t the bottom of the box. This apparatus should prove especially useful for longer peptides and proteins, since adsorption on the stabilizer (glass powder) is unlikely. A similar continuous method using filter paper was described by Grassmann and Hannig (1950). A number of other workers have devised methods for ionophoresis on filter paper. Haugaard and Kroner (1948) have combined ionophoresis with paper chromatography by threading electrodes down the side of a paper chromatogram, which was first soaked in buffer solution. A potential was applied during the development of the chromatogram with phenol. Wieland and Fischer (1948) have used an apparatus in which a potential is applied to the ends of a strip of filter paper which is soaked in buffer and held in a small glass chamber. The ends of the strip are outside the chamber and dip into the electrode vessels. Good separations with amino acids and proteins (Turba and Enenkel, 1950) were reported. A similar principle is employed in the methods reported by Biserte (1950), Durrum (1950) and Cremer and Tiselius (1950). The method of the latter authors was designed specially for the fractionation of proteins. The whole paper is immersed in a bath of chlorobenzene, which prevents heating and loss of solvent by evaporation. 2. Ion Exchange Methods

Another method of fractionating peptides according to their charge These have frequently been

is by the use of ion exchange materials.

32

F. SANGER

used for the group separation of amino acids (reviewed by Turha, 1945; Block, 1949; Martin and Synge, 1945). I n most work the amino acids have merely been separated into two groups in each experiment, those which are adsorbed, and those th at are not adsorbed, the latter being eluted with a different solvent. I n this way it is possible to obtain quite sharp separations by choosing suitable adsorbents and conditions of pH etc. Thus, for instance, basic amino acids may be adsorbed on basic A l z 0 3 ” (Wieland, 1942), silica gel (Schramm and Primosigh, 1944), permutite (Felix and Lang, 1929) or suitable synthetic resins (Block, 1942; Wieland, 1944). Acidic amino acids may be adsorbed on “acid Alz03” (Wieland, 1942; Jutisz and Lederer, 1947; Turba and Richter, 1942) and synthetic anion exchange resins (Cannan, 1944; Tiselius et al., 1947). The synthetic resins are probably to be generally preferred because of their greater capacity and because their properties are more reproducible from batch to batch. There are also a large number of different ones now commercially available, which ex tends their use. It is probable that simple group separations of this type may also be carried out with peptide mixtures (qf. Waldschmidt-Leitz and Turba, 1941) though in such mixtures of substances with a much more scattered range of isoelectric points, considerable overlapping may be expected. Also large peptides and those containing aromatic residues will probably be held on the exchangers by ordinary adsorption which may cause difficulties. Jutisz and Lederer (1947) and Lederer and Tchen (1947) have devised a method for the group separation of neutral amino acids and peptides by making use of differences in the apparent pK of the amino groups in the presence of formaldehyde. Neutral amino acids and peptides are not adsorbed on “acid AlzOa” from water but in the presence of formaldehyde they acquire acidic properties and those with the lower pKa‘ values are adsorbed. Table IV shows the approximate pKa’ values for the three groups in 10% formaldehyde, and their behavior on columns of “acid A1203”in the presence of 10% and 1%formaldehyde, respectively. In this way three clearly separated groups may be obtained. Presumably scryl, threonyl, and cysteinyl peptides will behave as the glycyl peptides. Final elution from the columns is effected with hot water which dissociates the formaldehyde complexes. This method, which gives rather a different group separation from other techniques may prove iiseful for the initial simplification of peptide mixtures. Under suitable conditions it is possible to obtain separations on an ion exchange column by elution analysis. The p H of the developing solution should be such th at the solutes are distributed between the resin and solvent so that they move down the column as definite bands.

33

THE ARRANGEMENT O F AMINO ACIDS I N PROTEINS

I n this way Consden et al. (1948) and Drake (1947) separated glutamic and aspartic acid on columns of Amberlite IK-4 (polyamine anion exchanger) maintained a t p H 2.5, and this type of column has been used to separate acidic peptides from a wool hydrolyzate (Consden et al., 1949). More recently Stein and Moore (1949) have obtained excellent fractionations of amino acids on columns of Dowex-50 (cation exchanger, sulfonic acid groups) developed with 1.5-4 N HCl. The separations obtained are almost certainly due to differences in adsorption affinity for the resin, as well as to charge effects. At the p H used the only partially TABLEIV Group Separation of Amino Acids and Peplides according to Jutisz and Lederer (1 347) Group Monoamino acids other than those in group 2 Glycine Serine Threonine Cysteine Simple dipeptides other than those in group 3 Glycyl peptides

Approx. pKa’ Adsorption on “acid A1,03” in 10% H.CHO in 10% H,CHO in 1% H.CHO

7.0

-

-

6.8

+

-

4.2

+

+

ionized groups will be the sulfonic acid groups of the resin, the amino acids having either one or two positive charges. This technique appears to be one of the most efficient methods of fractionation and it will be interesting t o see how far it may be used for the separation of peptides. In contrast t o most other methods depending on adsorption (see next section) the bands exhibit no tailing, but give elution curves showing sharp well-defined peaks. Separations may also be carried out on ion exchangers by the principle of displacement chromatography. If the ionizing groups of a resin are completely saturated with an amino acid, and a second amino acid, having a greater affinity for the resin is introduced onto the column, the first amino acid will be displaced and will move down the column. I n this way columns may be obtained from which the amino acids are displaced in the order of their isoelectric points. Partridge (1949a, h, c ; Partridge and Westall, 1949) has developed this principle using the sulfonic acid resin “Zeo-Karb 215,” for the separation of amino acids. Elution is carried out with ammonia which has a greater affinity for the resin than most of the amino acids. This method, which is best carried

j4

F. SANGER

out on a relatively large scale, may prove useful for the group separation of peptides, but no such results have yet been reported.

3. Adsorption Chrornatography The possibilities of fractionating amino acids, peptides and proteins by adsorption chromatography have been fairly extensively explored, especially by Tiselius and coworkers (reviewed in Tiselius, 1947; Turba, 1948; Martin and Synge, 1945). Active carbon is probably the most effective adsorbent for amino acids, being one of the few on which they are retained. Aromatic amino acids are strongly adsorbed and may be separated from the other amino acids (Schramm and Primosigh 1943; Jutisz and Lederer, 1947; Partridge, 1949b) by adsorption from acetic acid solution. They can be eluted then with phenol or ethyl acetate. A similar type of group separation may also be used to separate aromatic peptides from a partial hydrolyzate (Synge and Tiselius, 1949; Sanger and Tuppy, 1951a) though larger peptides and peptides containing basic amino acids, are rather strongly adsorbed, and may appear in the “aromatic” fraction. Considerable losses are often involved in this type of fractionation, as it is difficult to elute the substances completely. Attempts t o use adsorption on charcoal for a more detailed chromatographic fractionation of peptides have not met with great success. Adsorption isotherms are usually of the Langmuir type and the only method that is generally applicable is the technique of frontal analysis. This may be used as an analytical method (Moring-Claesson, 1948; Synge and Tiselius, 1947, 1949), but does not give appreciable fractionation. Usually the bands tail too much to make separations possible by elution analysis. The most satisfactory method for separating compounds having this type of adsorption isotherm is the displacement method, though unfortunately it is not a general method and has only been used in a few special cases. One of the disadvantages of displacement analysis is that there is no intermediate zone between two consecutive bands of adsorbed substances, so th at a clear separation is almost impossible. This difficulty has been very neatly overcome by Tiselius and Hagdahl (1950) by the addition of a volatile substance having an adsorption affinity intermediate between those of the two substances to be separated. For instance when a mixture of methionine, Leu.Gly.Gly and n-butanol was subjected to displacement chromatography, the bands were eluted in the following order: methionine, n-butanol, Leu.Gly.Gly. By cutting in the middle of the n-butanol fraction it was possible t o obtain complete separation of the methionine and Leu.Gly.Gly. It is likely that adsorption methods may be useful for fractionating larger peptides, which cannot readily be separated b y other methods.

THE ARRANGEMENT OF AMINO ACIDS I N PROTEINS

35

Thus Synge and Tiselius (1947) were able to fractionate the components of tyrocidin, both by elution and displacement methods. I n a study of the partial hydrolysis of ovalbumin Moring-Claesson (1948) was able to separate by adsorption the unchanged protein from the breakdown products. The former was adsorbed much more strongly on alumina and less strongly on carbon than the smaller peptides and amino acids. On most of the more commonly used adsorbents, peptides and proteins are only weakly adsorbed from aqueous solution. Tiselius (1948; Shepard and Tiselius, 1949) has shown that in the presence of a high concentration of salt, substances are much more strongly adsorbed, thus making i t possible to use adsorbents such as paper or silica for the chromatography of proteins (see also Mitchell et al., 1949). This technique which is known as “salting out adsorption” may prove useful for the fractionation of larger peptides. Hamoir (1945) was able to fractionate amino acids into four groups by adsorption on silver sulfide. I n general those amino acids th a t form the Ieast soluble silver salts were the most strongly adsorbed. The order in which the amino acids are adsorbed on the column, is rather different from the order on other columns, so th a t a different type of group separation may be obtained. 4. Partition Chromatography Partition chromatography was originally introduced as a method for the fractionation of acetamido acids (Martin and Synge, 1941) and was early applied t o the separation of acetamido peptides from a partial hydrolyzate of gelatin (Gordon, Martin and Synge, 1943). I n this method the fractionation was carried out on columns of silica gel and the bands were located by incorporating an indicator in the aqueous phase of the chromatogram. The mixture studied was rather too complex to give really satisfactory resolution of the peptides and since much better results are obtained by direct fractionation of peptides on paper or starch chromatograms, the technique has not been used further. a. Paper Chromatography. Undoubtedly the most satisfactory procedure for fractionating amino acids and smaller peptides is paper chromatography (Consden et al., 1944) which is now familiar to most workers. The method has been the subject of a number of reviews (e.g. Consden, 1948; Martin, 1950; Jones, 1949) and its application to peptides has been described by Consden et al. (1947a). I n this method samples containing about 1 mg. of the peptide mixtures to be analyzed are fractionated on two-dimensional chromatograms. The position of the peptide spots is determined either by spraying with ninhydrin, which causes only a small amount of destruction of the peptide (see Woiwod, 1949) or from the

36

F. SANGER

fluorescence which is produced if the paper is heated (Phillips, 1948). The exact nature of this latter reaction is still obscure but it seems t o depend both on the peptide and on the paper (Patton et al., 1949). It is extremely valuable for locating spots with a minimum of concomitant destruction. The peptides may then be cut out and eluted by running a small amount of water through the ((cut.” Microtechniques for identifying the constituent amino acids and the N-terminal residues of these peptides have been described by Consden et al. (1947a) (see p. 48). The choice of suitable solvents for running the chromatograms will depend on the nature of the particular mixture to be analyzed. Probably the most generally useful ones are phenol, collidine and butanol-acetic acid mixtures (Partridge, 1948). The latter are especially useful for the larger peptides that “tail” badly on other solvents (Jones, 1948, 1949). Thus Phillips (1949a) found that a mixture of peptides from insulin which could not be satisfactorily fractionated on phenol or collidine could be separated into well-defined spots using butanol-acetic acid. By varying the acetic acid content, mixtures with different properties can be prepared. It is not possible a t present to predict the exact position of a peptide spot on a chromatogram from a knowledge of its composition, but a n approximate determination of its RF values may be obtained as follows. Martin (1949a) has shown theoretically th at the partition coefficient of a dipeptide divided by the product of the partition coefficients of the constituent amino acids is a constant for any given phase pair, ie., that for a peptide A.B ~ A . B ~- constant

O A ~ B

where aA,as and a A . B are the respective partition coefficients. Consden et al. (1944) showed th at the RF values are related to the partition coefficients by the equation

AL -

As

is a constant for any solvent system so that

RFAB

=

K(i

-

RF~RF~ RFA)(l- RFB) R F ~ R F ~

+

THE ARRANGEMENT O F AMINO ACIDS I N PROTEINS

37

Thus the Rr value of a dipeptide may be calculated in terms of the RF values of the constituent amino acids and of a constant K , which is best determined experimentally for each solvent system. The relationship is not absolutely accurate, since, for instance, peptides containing the same amino acids in different order may frequently be separated. It has, however, been found to apply satisfactorily in most cases and to be a useful check on the identity of a peptide. b. Starch Chromatography. Starch chromatography, originally introduced by Elsden and Synge (1944; Synge, 1944) has been developed into a very accurate method of amino acid analysis by Moore and Stein (1948, 1949; Stein and Moore, 1948). Amino acids and simple peptides move as sharp well defined bands and excellent resolutions may be obtained. Synge (1944, 1949) fractionated a partial hydrolyzate of gramicidin on starch and was able to identify a number of peptides. Amino acids and peptides on starch columns behave much as they do on paper chromatograms ; but the use of starch makes possible separations on a larger scale. The efficiency of separations is probably rather less on starch and it is not possible of course to use the two-dimensional technique. Starch chromatograms run extremely slowly and it is necessary to run each column for several days to obtain satisfactory fractionations. c. Other Partition Chromatography. A number of other types of partition chromatograms have been suggested and may be useful in the separation of peptides. Paper chromatography is only applicable on a micro scale and several attempts have been made to extend it to a larger scale. Mitchell and Haskins (1949) have described a " chromatopile " for such fractionations. This consists of a pile of filterpapers which are used as the column. Jones (1949) suggests the use of thick paper which may best be run by ascending chromatography. Recently there has become available preparations of powdered cellulose which are suitable for use in columns. Diatomaceous earth (Kieselguhr) may be used as an inert support for the aqueous phase of a partition chromatogram (Martin, 1949a; Bell et aE., 1949). Being only a very weak adsorbent, it is likely to be useful for larger peptides which may tail badly on other chromatograms due to excessive adsorption. Certain hydrophobic peptides are difficult to fractionate as they all tend to run fast on the usual chromatograms in which water is the stationary phase (Synge, 1949). Recently systems have been described using adsorbents such as rubber, in which an organic solvent is held as the stationary phase and the column is developed with water or a buffer solution (Boldingh, 1948; Howard and Martin, 1950).

38

F. SANGER

5 . Detection of Peptides from Columns I n any of the chromatographic methods described above it is necessary t o have some method for locating and if possible, for estimating the colorless peptides on the column or as they are eluted from it. I n paper chromatography they are located on the paper but where a column arrangement is used it is usually simpler to identify the bands as they are eluted. This is most usually done by collecting the effluent in a large number of small fractions. A number of automatic fraction collectors have been described that may be used for this purpose (Stein and Moore, 1948; Randall and Martin, 1949; Phillips, 194913) and some are commercially available. A suitable test may then be applied to each fraction. For this purpose the ninhydrin reaction is most generally used (Moore and Stein, 1948). Drake (1947) has described an automatic arrangement for spotting aliquot drops of the effluent on to filter paper, which can be developed by a suitable reagent. This could probably be used in conjunction with an automatic fraction collector. Several methods have been described for continuously recording the concentration of solute in the effluent from a column. Tiselius and Claesson (1942) have observed changes in the refractive index of the solution using a special interferometer. This method has proved very useful where adsorption chromatography has been used, but is not suitable for use with partition chromatograms as any slight changes in the composition of the solvent will cause changes of refractive index and will interfere with the recording. A small conductivity cell attached to the bottom of a column may also be used t o locate the bands of amino acids and peptides, which will cause changes in the conductivity of the solution (Randall and Martin, 1949). Recently Drake (1950) has described a polarographic method for following the chromatography of proteins. Only substances containing cystine or cysteine will be detected but it may be useful where large peptides are being studied. The method is unaffected by changes in the solvent or salt concentration.

6. Countercurrent Distribution The development of the method of countercurrent distribution is largely due t o the work of Craig and his associates (see Craig et al., 1949). The method is applicable to the separation of any substances tha t can be reversibly distributed between two immiscible solvents, and these, of course, include peptides. Several different types of apparatus

THE ARRANGEMENT O F AMINO ACIDS IN PROTEINS

39

have been devised for this purpose (Craig, 1944; Kies and Davis, 1950; Rometsch, 1950), the most efficient probably being the all-glass apparatus recently designed by Craig and Post (1949). Compared with chromatographic methods, this countercurrent distribution method is probably less efficient, the apparatus required is very much more complicated, and the labor involved is considerably greater; nevertheless, it possesses several definite advantages over chromatographic methods. The chief of these is the absence of any solid phase, which may act as an adsorbent for the solutes. In partition chromatographic methods this adsorption may often lead to distortion of the bands and render fractionation very inefficient. The behavior of a solute on a countercurrent distribution depends only on its partition coefficient, which in most cases is constant, so that it is possible to calculate the exact theoretical distribution curve and this may be used as a very sensitive test for purity. The method is especially useful for the fractionation of larger polypeptides, which cannot be fractionated easily by other techniques. Thus, for instance, Gregory and Craig (1948) studied crystalline gramicidin by this method, and found it to consist of at least three components, although it had previously been thought to be pure. Other naturally occurring polypeptides have similarly been purified (Craig et al., 1949; Barry et al., 1948; Livermore and du Vigneaud, 1949). 7. Lysine Peptides

The DNP method has been used for the separation and identification of peptides containing lysine (Sanger, 1949b). When a DNP-protein is partially hydrolyzed, the only colored products present are the DNP derivatives of the N-terminal peptides and of those peptides which contain lysine residues. The DNP-terminal peptides are extracted by an organic solvent as already described (p. 7). This procedure can be regarded as a very specific type of group separation in which only a few special peptides are separated out. The peptides containing c-DNP-lysine, which remain in the aqueous hydrolyzate solution mixed with other unsubstituted peptides, can be separated by adsorption on talc from acid solution, since the D N P group is held strongly on this adsorbent. Elution can be effected with acid ethanol and the peptides subsequently fractionated on suitable partition chromatograms. FDNB also reacts with tyrosine and histidine residues to give colorless products, which are likely to be retained by the talc. I n the case of insulin no such interference was observed, but it is doubtful if the method could be applied successfully to other more complicated proteins.

40

F. SANGER

8. Cystine Peptides Consden and Gordon (1950) have described an elegant method for investigating peptides involving cystine residues. After removal of acidic peptides from a partial hydrolyzate using an ion exchange column (Amberlite IR-4), the remaining peptides are oxidized with bromine, which converts the cystine residues to cysteic acid residues. The acidic peptides so formed can be separated from the remaining neutral peptides on another ion exchange column. Only peptides which originally contained cystine or cysteine linked with neutral amino acids appear in this fraction as cysteic acid peptides. This simplified mixture may then be fractionated by ionophoresis and paper chromatography.

9. Other Methods of Fractionation So far we have considered only the newer countercurrent methods for fractionating peptides. These will probably play a predominant role in the future, though the classical methods of fractional crystallization and precipitation should not be forgotten. They are still the most effective methods of fractionating proteins and probably larger polypeptides, such as the oxidation products of insulin (Sanger, 1949a). The aromatic sulfonic acids, which were developed by Bergmann for the specific precipitation of amino acids, have also been used for the separation of peptides from a partial hydrolyzate of silk (Stein et al., 1944). Synge and Tiselius (1950) have recently described a method for fractionating substances according to their molecular weight by electrokinetic ultrafiltration which may be applicable to the group separation of peptides. A number of ingenious methods of fractionation have recently been suggested by Martin (1949b).

10. Conclusions Many of the methods considered above have not been extensively used for the fractionation of peptides, so that it is impossible to know how far they may be applied and which are the most effective methods. Also it is to be expected that considerable improvements in these techniques will take place and that other new methods will be devised in the near future. The properties of amino acids and small peptides render them suitable to fractionation by methods employing partition chromatography. Paper chromatography is especially to be preferred for the final fractionation of a simplified peptide mixture because of the good resolutions obtained, the ease and rapidity of technique, and the possibility of using

THE ARRANGEMENT O F AMINO ACIDS IN PROTEINS

41

the two-dimensional method. The chief disadvantage is that only a very small amount of each peptide can be obtained from a chromatogram, though it is usually possible to obtain sufficient material to determine the structure of di- and tripeptides. When larger amounts of peptides are required, several methods are available; chromatography on starch, cellulose powder or ion exchange resins, or countercurrent distribution. Probably none of these gives such good resolution as paper chromatography and the work involved is greater. For the preliminary separation of a complex protein hydrolyzate into simpler peptide mixtures, ionophoretic methods are probably the most generally useful. Aromatic peptides may be separated by adsorption on charcoal and cystine peptides by oxidation. One of the main requirements at present is for methods that will efficiently fractionate large polypeptides with molecular weights between say 1,000 and 10,000. At present the best method for dealing with them is by countercurrent distribution and possibly by certain ionophoretic methods. While techniques for separating this class of substances would be extremely useful, it seems that the chief difficulty is to find suitable degradation procedures for producing a simple mixture of large polypeptides from a protein.

VI. DETERMINATION OF PEPTIDE STRUCTURE 1. IdentiJcation and Estimation of Amino Acids The first stage in the determination of the structure of a peptide is to identify the amino acid residues present in it, and the obvious technique for this is paper chromatography. By no other method is it possible to identify completely the amino acids present in a mixture so simply and so rapidly. Unfortunately no solvent has yet been found on which it is possible to separate all the naturally occurring amino acids and it is necessary to run each hydrolyzate on a two-dimensional chromatogram or else to, run aliquots on two separate one-dimensional chromatograms. Where a large number of peptides is to be analyzed, the latter technique is usually preferable (Consden et al., 1949). It is often desirable to carry out an approximate amino acid analysis to determine whether there are one or two residues of a particular amino acid in a peptide. This may be done with sufficient accuracy by carrying out the paper chromatography in a semi-quantitative manner (Polson, 1948; Consden et al., 1949). Several modifications of greater accuracy have been described (Martin and Mittelmann, 1948; Wieland and Fischer, 1948; Woiwod, 1949; Fowden and Penney, 1950; Boissonnas, 1950). The question as to how accurately it is possible to determine an amino

42

F. SANGER

acid on a paper chromatogram has been frequently discussed (Jones, 1949; Gordon, 1949; Martin, 1950) and is probably still unsettled. If careful control experiments are carried out the above methods should give results to within 5-10 %. Other methods (starch chromatography, microbiological methods, or the pipsyl method) may be used to obtain an accurate analysis but are considerably more laborious. An elegant method for determining the optical configuration of amino acids on paper chromatograms using the enzyme n-amino acid oxidase has been described by Synge (1949).

2 . Amino Acid Sequence The determination of the amino acid sequence in peptides is essentially a question of identifying terminal residues. Thus if the N-terminal residue of a dipeptide is known its structure is determined. The sequence in a tripeptide would be determined by identification of the N- and C-terminal residues. The structure of larger peptides may be worked out from their terminal residues and by degrading them to di and tripeptides. Thus, for instance, if a tetrapeptide has residue A as an N-terminal residue and on hydrolysis gives the dipeptides A.B and C.D, its structure must be A.B.C.D. An alternate method of determining the amino acid sequence is by a method of step-wise degradation from the terminal residues (p. 8). The method of Edman (1950), which has already been applied to synthetic peptides would seem especially suitable for this purpose. Consden et al. (1947a) have developed a rapid micro method for the deamination of the N-terminal residues of peptides eluted from paper chromatograms. A sample of the peptide is completely hydrolyzed and the amino acids present are identified. Another sample is then deaminated with nitrosyl chloride which destroys the N-terminal residue. The amino acids remaining are then identified after hydrolysis of the deaminated peptide. Although the reaction rarely goes completely smoothly and quantitatively, and a certain amount of destruction of the non-terminal residues usually takes place, it is nevertheless pQssible in most cases to identify the N-terminal residue. The DNP method (p. 5) may also be used for the identification of the N-terminal residues in peptides. It is rather more laborious than the above deamination procedure, but the results are usually more clear-cut. Two methods are available. Either the terminal DNP amino acid is identified by chromatography or the remaining unsubstituted amino acids are identified by paper chromatography. The former method has the advantage that it may be carried out quantitatively but requires rather more material than is usually available from the elution of one paper chromatogram.

THE ARRANGEMENT O F AMINO ACIDS IN PROTEINS

43

Bowman (1950) has shown that amino acids and peptides may be quantitatively converted to their dimethyl derivatives by reductive alkylation in the presence of formaldehyde, and has used the method to detect the N-terminal residues of peptides. A micro-modification was described by Ingram (1950) who identified the amino acid which is absent from a hydrolyzate of the dimethyl peptide.

3. Estimation of Peptides

It is clear that the estimation of the amount of a particular peptide in a partial hydrolyzate may in certain cases yield considerably more information concerning the structure of the protein than its mere identification. Unfortunately it is impossible to determine directly the total amount of a particular peptide sequence in a protein molecule. Only a minimal value can be obtained from the composition of a partial hydrolyzate. I n the case of the N-terminal peptides of insulin it was possible to estimate certain sequences in the protein from the yields of the peptides by allowing for the rate of breakdown of the bonds involved (p. 5 2 ) . Such an approach may clearly be useful in the future where the yield of a peptide can be determined. Usually only small amounts of peptide are available after chromatographic fractionation of a partial hydrolyzate, so that micro methods of estimation are required. Consden et al. (1949) obtained an approximate estimate of peptides from a paper chromatogram by the color intensities of the amino acid spots produced on hydrolysis. Clearly careful control analyses with synthetic substances should accompany any estimaitons by this type of technique to allow for losses during fractionation. The use of isotopic methods for the estimation of peptides appears to offer considerable advantages of accuracy and general applicability. I n the original “isotope dilution” technique of Rittenberg and Foster (1940) it was necessary to isolate a pure sample of the compound to be estimated, from a mixture of comparable amounts of other similar substances. In the case of peptides this would clearly be a formidable task, and has never been attempted. The Pipsyl Method. Using radioactive isotopes Keston, Udenfriend and Cannan (1946, 1949; Keston and Udenfriend, 1949) have developed an accurate micro method, which theoretically could be applied to the estimation of any amino acid or peptide in a mixture. Two different techniques, the carrier technique and the indicator technique, have been worked out. In the former the peptide mixture to be analyzed is treated with p-iodophenylsulfonyl chloride (pipsyl chloride) containing radioactive 1 1 3 1 , which reacts quantitatively with the amino groups givingpipsyl derivatives. To this is added a great excess of the non-isotopic pipsyl

44

F. SANGER

derivative of the peptide to be estimated, so diluting the corresponding isotopic derivative in the hydrolyzate. A pure sample of this pipsyl peptide is then isolated from the mixture, and from its isotope content, the amount of peptide in the original mixture may be calculated. The chief difficulty lies in the isolation of the pipsyl derivative in an absolutely pure state free from any other radioactive pipsyl derivatives of closely related compounds which tend to form mixed crystals. In many cases a large number of recrystallizations are necessary before a constant isotope concentration is obtained. To overcome the above difficulties the indicator technique was developed for the estimation of amino acids (Keston et al., 1947, 1950). The amino acid mixture is again treated with pipsyl chloride containing I1al. This mixture can then be fractionated by countercurrent distribution and paper chromatography and if all the derivatives are completely separated, the amount of amino acid can be determined from the isotope content of the spot. By adding an indicator to the mixture before fractionation it is possible to obtain an estimate even if the spots are not completely resolved. For instance, if alanine is to be estimated the indicator added would be pipsyl alanine containing S36,which may be estimated independently of the The ratio of 113' to S36in a part of the pipsyl-alanine spot may be used to estimate the original alanine content. This is only true for those parts of the alanine spot which contain no other pipsylderivatives (containing Il3'). I n such parts the to S36is constant, and this is used as the test of purity. This ratio of method has not yet been applied to peptides, but there seems no reason why it should not be, though rather complex fractionation procedure might be necessary to obtain a part of the peptide spot in a pure form. OF INVESTIGATIONS ON VARIOUSPROTEINS VII. RESULTS

In this section we shall consider the results obtained since 1943, which provide information concerning the arrangement of amino acids in proteins. Similar studies on the naturally occurring polypeptides will not be dealt with here. 1. Silk Fibroin (Bombyx Mori) Assuming a molecular weight of 30,000 for silk fibroin (Coleman and Howitt, 1946) its composition is given by the following formula: (~ly16gAlalo,Ser40Tyr lgAsp6Glu4Leu6Va18PronPhe,Thr4Arg,LyslHis1 (Levy and Slobodiansky, 1949; Tristram, 1949)

It may be noticed that this formula tends approximately but not exactly to the formula (Gly3Ala2X2),where X represents any residues

THE ARRANGEMENT O F AMINO ACIDS I N PROTIGINS

45

other than glycine or alanine, and this unusually simple composition suggests that the amino acid arrangement may also be simple. By the classical methods of fractionation the peptides Gly.Ala (Fischer and Abderhalden, 1907) Ala.Gly (Abderhalden, 1909a) and Gly.Tyr (Abderhalden , 1909b) were isolated in considerable yield from the silk fibroin of Bombyx Mori and unambiguously characterized. Several longer peptides were also isolated in small yield and fairly convincing evidence for their structure was presented, though they were not finally identified by synthesis. Thus Abderhalden and Bahn (1932) benzoylated a fraction from a hydrolyzate of fibroin which had been obtained by the action of 1 N NaOH a t 37", and obtained several benzoyl peptides. One of these contained the amino acids glycine, serine, tyrosine, and proline in the ratio 1:1:1:2 and on partial hydrolysis with 10% HzS04 gave hippuric acid, Ser.Pro and Tyr.Pro. In an experiment with synthetic peptides it was found that benzoyl Gly.Ser was completely broken down under the conditions of hydrolysis in 10% H2S04 whereas benzoyl Gly.Tyr was rather stable. From this it was concluded that the only possible structure for the pentapeptide was G1y.Ser.Pro.Tyr.Pro. A second fraction was similarly identified as Ser.Pro.Tyr.Pro. A third peptide (Abderhalden and Bahn, 1933) containing the amino acids tyrosine, serine, and proline in the ratio 2 :1: 1 was isolated from a similar hydrolyzate. On treatment with trypsin (presumably a crude preparation) this gave tyrosine, Ser.Pro and a tripeptide containing one molecule of the three amino acids. This tripeptide was then treated with benzylamine and phenylisocyanate, according to the method of Abderhalden and Brockman (1930) and from the hydrolyzate the phenylhydantoin of tyrosine and Ser.Pro benzylamide were isolated. This established the structure of the tripeptide as Tyr.Ser.Pro, and it was concltxded that the tetrapeptide was Tyr.Ser.Pro.Tyr. The presence of tyrosine as a N-terminal amino acid was assumed since the tetrapeptide was precipitable with mercuric sulfate, as were tyrosine and other dipeptides containing a free tyrosine carboxyl group, whereas the tripeptide was not precipitable. While the evidence for the structure of these peptides is fairly good, it is possible that they may have been mixtures of peptides having slightly different structures. At least they do show that there are tetrapeptide sequences in fibroin that contain no glycine or alanine. Stein et al. (1944) have made use of their method of specific precipitation with aryl sulfonic acids to isolate peptides from an acid hydrolyzate of silk fibroin. Gly.Ala was isolated in 5.597, yield using 2:5-dibromobenzene sulfonic acid and Ala.Gly in 6.0% yield using 2,6-diiodophenol4-sulfonic acid. A more precise estimation of these two peptides present at various

46

F. SANGER

stages of hydrolysis was obtained by Levy and Slobodiansky (1949) with the pipsyl carrier technique. Samples of fibroin that had been hydrolyzed with 12 N I-ICl a t 39” for 16, 24, and 48 hours respectively were analysed for glycine, alanine, Ala.Gly, Gly.Ala and Gly.Gly. The results of one such experiment are shown in Table V. It was calculated that if the arrangement were completely random, the maximum possible yield of Gly.Gly would be 18.2% and of Ala.Gly and Gly.Ala 12.2%. The above figures, especially those for Ala.Gly, which are over twice the theoretical value show th a t this is not the case. Th e relatively small yield of Gly.Gly suggests that this sequence does not occur t o any great extent whereas the yield of Ala.Gly accounts for about half of the alanine present in the protein. Levy and Slobodiansky have pointed out that these results would be expected if the minimum TABLEV Analysis of Partial Hydrolyzate of Silk Fibroin (Levy and Slobodiansky, 1949) (Yields expressed in terms of N as % of total N) Time of hydrolysis in 12 N HC1 at 39” Glycine Alanine Ala.Gly Gly.Ala Gly.Gly

16 hr.

24 hr.

48 hr.

4.4

7.1 6.3 23.3 9.0

12.9 10.5 27.0 8.3 1.8

3.1 16.9 5.4 0.1

-

repeating peptide sequence were of the type: X.Ala.Gly.Ala.Gly.X.Gly. While this is probably the simplest explanation of the results, there are certainly many other more complicated ones. It does not account for the tri- and tetra-peptides isolated by Abderhalden and Bahn which contain no glycine or alanine or for the results of Drucker and Smith (see below). The results of Goldschmidt et al. (1933), and of Grant and Lewis (1935) referred to by Synge (1943) also suggest an uneven distribution of residues throughout the molecule. At least, it may be said that the results of Levy and Slobodiansky make it likely that the sequence X.Ala.Gly.Ala.Gly.X occurs frequently in silk fibroin. Silk fibroin may be dissolved in a solution of cupri-ethylenediamine (Coleman and Howitt, 1947). On dialysis part of the material remains in solution and is said t o be “renatured.” Drucker and Smith (1950) treated this material for a short period with trypsin and obtained a precipitate, which had a n average molecular weight of about 7000 and contained only the amino acids glycine, alanine, and serine. The remaining amino acids were all left in solution. It, was suggested on the basis of this evidence

THE ARRANGEMENT OF AMINO ACIDS I N PROTEINS

47

that the silk fibroin molecule is built up of five parts, three of which have molecular weight 7000 and contain only the above three amino acids, the other two parts having molecular weight 5000 and a more complex composition. These results are in agreement with the suggestion of Meyer et al. (1940), t ha t silk fibroin is composed of two parts, an amorphous part of complex amino acid composition and a crystalline part built u p simply of Gly.Ala units linked together. This theory was based on the observation t ha t the X-ray data could be fitted by a unit cell containing four parallel Gly.Ala residues. It was considered th a t some of the alanine residues in the crystalline part could be replaced by serine since its molecular dimensions are similar. Clearly the idea of a long chain containing glycine and alanine alternately is untenable in view of the results of Levy and Slobodiansky since in such a case the yields of Ala.Gly and Gly.Ala should be equal. However such a sequence could be broken u p to a certain extent by serine residues. Abderhalden (1940, 1943) has recently renewed his investigations on the isolation of diketopiperazines from silk fibroin and has isolated considerable quantities of the anhydrides of Gly.Ala, Gly.Tyr, and Ala.Ser from partial hydrolyzates obtained b y the use of strong acid or proteolytic enzymes. No evidence could be obtained of anhydrides containing only one amino acid, making it unlikely th at such sequences occur to any great extent in fibroin. Abderhalden believes that these diketopiperazines are derived from some cyclic structure in the protein. It is possible however t ha t they may have been formed from peptides during the isolation procedure since certain quite mild conditions are known to catalyze their formation (Agren, 1940; Huang and Niemann, 1950). Hot ethanol was used to extract them, and this might have brought about ring formation. Control experiments were carried out to show that dipeptides did not undergo ring closure under the conditions of hydrolysis, but the other obvious control, boiling a dipeptide with ethanol, was not reported.

2. Protamines Little progress has been made concerning the chemistry of the protamines since Synge reviewed the subject in 1943. However it would seem that much could be learnt from the application of the newer methods of peptide chemistry to these proteins, which from their amino acid composition and molecular weight appear to have a rather simple structure (see Tristram, 1949). Porter and Sanger (1948) showed the presence of proline as a N-terminal residue in salmine, but were unable to estimate it due to the instability

48

F. SANQER

of DNP-proline. Felix and Mager (1937) had already suggested that proline was the N-terminal residue in clupein on the basis of titration data. This was confirmed using the DNP method, and it was suggested there was also a small amount of N-terminal serine (Felix et al., 1950). The presence of arginine as a C-terminal residue was indicated by Dirr and Felix (1932). If dibenzoyl arginine is treated with acetic anhydride, a n acetyl derivative is formed which on hydrolysis gives benzoyl acetyl urea and p-benzoylamino-a-piperidone. This is readily split with acid to give ornithine. It was found that ornithine was produced after benzoylation, acetylation, and hydrolysis of clupein. The above series of reactions could only take place if the a-carboxyl group was not involved in peptide linkage so that it must have originated from a C-terminal arginine residue. Since two-thirds of the residues in clupein are arginine it follows that the sequence Arg.Arg must occur frequently in the molecule, and the dipeptide has often been isolated. Felix and Schuberth (1942) have described a method for the preparation of Arg.Arg and obtained 12.5 g. of the pure diflavianate from 50 g. clupein. The fractionation of partial hydrolyzates of clupein by adsorption chromatography on filtrol-neutrol (Waldschmidt-Leitz and Turba, 1940, 1941) arid on the cation exchange resin Wofatite C (Rauen and Felix, 1948) have been studied. Considerable fractionations could be obtained, but no definite peptides were identified. 3. Wool Keratin Considerable doubt exists as to whether wool can be regarded as a single protein or whether it is a mixture of different insoluble proteins, so that it is perhaps not one of the most suitable proteins to study by partial hydrolysis methods. However, because of its practical importance the Leeds workers have applied their newer methods to it, and a number of peptides have been identified. Consden et al. (1949) studied a n acidic fraction of a partial hydrolyzate of wool. Preliminary group separation was carried out on an ion exchange column (Amberlite IR-4) to obtain an acidic fraction, which was separated into ten fractions by ionophoresis in silica jelly. Each fraction was then subjected to two dimensional paper chromatography using phenol and collidine. One of these ionophoretic fractions contained only one peptide which was identified as Glu.GIu. This was present in much greater amount than any other acidic peptide and accounted for about 10% of the total glutamic acid of wool (estimated from the total N of the ionophoretic fraction). I n Table VI are listed the peptides that were considered to be probably present. About an

THE ARRANGEMENT O F AMINO ACIDS I N PROTICINS

49

equal number of peptides were given as being possibly present. Interpretation of the results was rendered difficult by the extreme complexity of the mixture, so that only a few of the spots appeared to contain a singli: peptide. Where no difference in composition was found before and after deamination of a dipeptide spot, it was assumed that both possible peptides were present. It is doubtful if this is entirely justified, since some dipeptides are extremely stable to deamination. Clear evidence was obtained for the presence of Glu.Glu, Ala.Glu, Glu.Ala TABLEV I Acidic Peptides Identijied from Wool (Consden et al., 1949) Peptide

Yield *

(ilii.Asp Glu. G1u GIu.Gly Glu.Ala Glu .T y r Glu.I,eu Glu. Cy8 Asp.Glu Ser.Glu Gly.Glu Ala.Glu Tyr .Glu Val.Glu Leu. Glu CyS.Glu Asp.Va1 Asp.Leu Ser.Asp Leu.Asp

4 600

* mg. N of peptide per 100 g. N of

5 2!J

20 21 4

4 17

8 27 27 19 51 4 11 11

5

19

wool.

and Asp.Leu, and the peptides listed in Table VI are probably present also. The approximate yields of the peptides were estimated from the strength of ninhydrin color of the amino acids produced on hydrolysis. These yields are listed in Table VI. Consden and Gordon (1950) have studied the cysteic acid peptides derived from the cystine peptides of wool as described on p. 40. After a preliminary group separation of the cysteic acid peptides by ionophoresis, each fraction was fractionated on paper chromatograms with phenol and collidine. In Table VII are listed the peptides considered to be probably present and their approximate yield. The results were more clear-cut in this case than with the aspartic and glutamic acid

F. SANGER

50

peptides, largely due to the fact that it was possible t o separate by ionophoresis or chromatographically any two dipeptides containing the same amino acids in different order. This separation was made possible b y the low p K of the amino group in cysteic acid peptides. It is clear from the results that the structure of wool is extremely complex. Almost all of the monoamino acids occur linked t o both sides of glutamic and cysteic acids. The small number of aspartic acid peptides identified is probably due to the lower content of aspartic acid in wool so t ha t the peptide spots would probably be faint and would not show on the chromatograms. TABLEV I I Cysteic Acid Peptides from Wool (Consden and Gordon, 1960)

* m g . N of

Peptide

Yield*

Asp.CySOaH Glu. Cy SOaH Ser .Cy SOSH Gly .CySO3H Thr.CySOaH Ala.CySO3H Leu.CySO3H CyS0aH.Gly CySOaH.Thr CyS03H.Ala CyS03H.Val CyS03H.Leu CySOaH.Phe

2 2 60 24 20 30 8 10 20 14 20 10 4

peptide per 100 g . N of wool.

Middlebrook (1949) investigated the N-terminal residues by the DNP method and found rather a complex mixture, which would seem t o be a further indication of the heterogeneity of wool. 4. Insulin

Insulin has been studied in rather more detail than most other proteins due t o the interest arising from its physiological properties and to the fact that it is one of the few proteins that can be obtained in a reasonably pure form. It possesses a relatively simple structure, being built up of fairly short open polypeptide chains. For the purposes of the following discussion a value of 12,000 (Gutfreund, 1948) will be assumed for the molecular weight. The most recent analytical figures (see Tristram, 1949) indicate the following composition :

51

T H E ARRANGEMENT O F AMINO ACIDS I N PROTEINS

Gly7Ala~Va18LeulzIleu3Pro3Phe6( CyS-) 1zArg~HislLyszAsp~Glu6 (Asp-NHz)4(Glu-NH2)8Ser6ThrzTyrs The DNP method showed the presence of four N-terminal residues two of which were glycine and two phenylalanine (Sanger, 1945). From a study of the action of carboxypeptidase on insulin Lens (1949) found that free alanine was liberated before any other free amino acids, indicating the presence of alanine in the C-terminal position. Using their respective methods of reduction of the free carboxyl groups to alcohol TABLE VIII Properties of the Fractions of Oxidized Insulin (Sanger, 1949a)

Yield from insulin N-terminal residue Amino acids absent

Mol. wt. 1. From estimation of N-terminal residues 2. By ultracentrifugation (Gutfreund and Ogston, 1949)

A

B

30-40 % Glycine Lysine, arginine, histidine, phenylalanine, threonine, proline

25 % Phenylalanine Isoleucine

2900

3800

2900

7000

groups (p. 10) Fromageot et al. (1950) and Chibnall and Rees (1951) both identified alanine and glycine as the C-terminal amino acids. These results indicate that insulin is built up of four open polypeptide bridges of cystine. chains, which seem to be held together by -S-SThese are present to the extent of six residues per molecule. By oxidation with performic acid (p. 27) it was possible to split these bridges and thus to liberate the separate peptide chains (Sanger, 1949a). The oxidized insulin was fractionated by precipitation methods and two main fractions were obtained, which appeared to represent the whole of the oxidized insulin. The properties of these two fractions, designated A (acidic) and B (basic), are summarized in Table VIII. The most probable composition of fraction A is, GlyzAlalValzLeu Jleu 1(CyS) (AspzGlu4Ser~ T y (-NHz) r

4

and of fraction B, Gly3AlazVa13Leu4Pro 1Phe3(CyS) zArglHiszLyslAsplGlu 3Ser1Thr1Tyr2 (--NH2)2

Butler and coworkers have studied the action of various proteolytic enzymes on insulin. From a chymotryptic hydrolyzate they obtained

52

F. SANGER

two fractions, one precipitable by trichloracetic acid and one soluble (Butler et al., 1948). The latter fraction consisted of relatively small peptides with an average chain length of seven residues. These could be separated satisfactorily on paper chromatograms using butanol-aretic acid (Phillips, 1949a) but they have not yet been investigated further. The trichloracetic acid precipitate, which was referred to as the core” had a mean molecular weight of about 5000 and a high proportion of glycyl (approximately 3 equivalents) and a smaller amount of valyl TABLE IX U N P Deravatzves .from Fractaon B of Oxidized Insulan (Sanyer, 1949b)

Derivative

Bl €52

B3

B4

B5

Products of complete hydrolysis

Products of partial hydrolysis

Yield* from DNPinsulin

DNP-phenylalrtnine 1) N P-phenylalaninr Valine DNP-phenylalanine Valine Aspartic acid DNP-phcnylalanine Valine Aspartic acid Glutaniic arid

-

13 10

13

BI, B2

13

12

€31, l32, R3

30

55

B1, B2, B3, B4

20

-

Yield* from

IM

14

-

Total

* Moles of

92

04

peptide as % of the total N-terminal phenylalanyl residues.

and other N-terminal residues (Butler et al., 1950). On oxidation with performic acid the core was mostly converted to a fraction corresponding in properties to the fraction A of the oxidized insulin. a. N-Terminal Peptides. The N-terminal peptides of the two fractions from the oxidized insulin were determined by subjecting their DNI’ derivatives to partial hydrolysis (Sanger, 1949b). I n the case of fraction B four main bands (Bl-B4) were identified in an ethyl acetate extra(% of a partial hydrolyzate. Their properties are summarized in rl’ablc I X . I t is clear that all these derivatives are derived from the one N-terminal sequence Phe.Val.Asp.Glu. Other fainter bands (B5) were also present ; on partial hydrolysis these gave rise t o B3 and B4 and were therefore higher peptides from the same peptide sequence. The aspartic and glutamic acid residues are probably in the form of asparagine and glutamine residues in the intact protein, since other bands containing the

53

THE ARRBNGEMENT OF l M I N O ACIDS I N PROTEINS

same amino acids as H3, B4 and probably therefore amides, were also present when a shorter time of hydrolysis was used. The yields of the different peptides from a partial hydrolyzate of DNP-insulin are shown in column 4 of Table IX. It can be seen that virtually all the N-terminal phenylalanyl residues may be accounted for in terms of this one sequence. This is confirmed by comparing the yields of the DNP peptides from DNP-insulin with the yields from a sample of peptide B4 which had been hydrolyzed under similar conditions (column 5, Table IX). AgreeTABLE S Peptides Containing e-D,VP-lysine from DiVP-insulin (Sanger, 1949b)

*

Prptidc

Arnino acids present

N-tcrnrirral rcsidiic

Products uf partial hydrolysis

Structure

Yicld fronr Yield* DNP- from insulin L4

L1

c-DNP-Lysinc c-DNP-Lysine Alaninc Threonine Proline c-DNP-Lysine Threonine Proline c-DNP-Lysine Alanine

r,z

L3

1.4

-

-

c-DNP-Lysinc

-

c-DNP-Lys t-DNP-Lys.Ala

14 19

23

Thrconine

L1

Thr.Pro.(r-DN1')Lys

32

32

23

21

Thrconinc

L1. L2, L3 Thr.Pro.(a-DNP) Lys.

Ala

Other unidentified bands

6

Total

* Moles of peptide as

14

94

% of the total lysinc.

ment between the two sets of figures makes it clear that both the N-terminal phenylalanyl residues of insulin are present in the form of this one sequence, Phe.Val.Asp.Glu, and it was therefore concluded that the two phenylalanyl chains of insulin, which contain the same N-terminal tetrapeptides are in fact identical. It was also possible to separate and identify the peptides containing t-DNP-lysine from a partial hydrolyzate of the DKP derivative of fraction B. Here again four main colored bands were present and their properties are summarized in Table X. These all fit into the sequence Thr.Pro.Lys.Ala and the yields from DNP-insulin and from peptide L4 make it clear that both the lysyl residues of insulin are present in this single tetrapeptide sequence. When the DNP-derivative of fraction A of the oxidized insulin was partially hydrolyzed, four DNP derivatives were produced, all of which fitted into the sequence DNP-Gly.Ileu.Va1.Glu. When strong HCl was

54

F. SANGER

used for hydrolysis the peptides were produced in very small yields. Because of the great lability of the bond involving the carboxyl group of the glycyl residue most of the DNP-glycyl residues were present as DNP-glycine itself and as larger peptides which could not readily be fractionated. However when dilute acid was used for hydrolysis the yield of these peptides was raised and another band which appeared to be Gly.Ileu.Val.Glu.Glu was obtained. The yields of these peptides indicated that both the N-terminal glycyl residues are combined in the one amino acid sequence and it was concluded that the two glycyl chains were also identical. b. Amino Acid Sequence in the Phenylalanyl Chains. The results with the N-terminal peptides showed for the first time that the fractions A and B of the oxidized insulin were each an essentially homogeneous preparation of a polypeptide chain of 20 and 30 residues respectively. It was therefore considered worth while to investigate the peptides present in their partial hydrolysates by the methods of Consden et al. (1947b). Fraction B was subjected to hydrolysis in conc. HCI andthe hydrolyzate separated into a number of fractions by ionophoresis, charcoal adsorption, and adsorption on an ion exchange resin (Sanger and Tuppy, 1951a). The resulting peptide mixtures were then fractionated on two-dimensional paper chromatograms and their structures investigated. To illustrate the methods used a fraction containing peptides of cysteic acid will be considered. This was obtained by adsorption on an ion exchange resin (Amberlite IR-4B a t pH 2.6; Consden et at., 1948). A chromatogram of this fraction on phenol/butanol-acetic acid revealed eight peptide spots, whose compositions are given in Table XI. The N-terminal residues were determined by hydrolysis of the DNP derivatives. The structures of the peptides, as far as they can be deduced from the data in the table are given in the last column. Three dipeptides were identified: CySO,.H.Gly, Val.CyS03H, and Leu.CyS03H. Fraction B contains only two cysteic acid residues so that any peptide containing cysteic acid which is not the N-terminal residue must either contain the sequence Val.CyS0,H or Leu.CySO,H. Peptide 3 must therefore be Val.CySO,H.Gly and peptide 5 Leu.CySO,H.Gly. Thus both cysteic acid residues are joined through their carboxyl groups to a glycine residue. Peptide 8 which contains no glycine but has leucine as the N-terminal residue can only be Leu.Val.CySO,H and peptide 7 Leu.Val.CyS03H.Gly. These peptides establish therefore the presence of the sequences Leu.Va1.CySOaH and Leu.Val.CyS0,H.Gly in fraction B. In the upper part of Table XI1 are listed the various peptides identified in this work and the sequences which were deduced from them as

.

THE ARRANGEMENT O F AMINO ACIDS IN PROTEINS

55

being present in fraction B. Hydrolyzates obtained by the use of dilute acid and of alkali were also studied but only a few new peptides were found. These are also included in Table XU. From the results it was possible t o deduce five sequences as being present in fraction B, accounting for all the amino acids present except for one leucine, one tyrosine and TABLEXI Cysteic Acid Peptides from Fraction B of Oxidized I n s u l i n (Sanger and T u p p y , 1961a ) Strength of amino acid after hydrolysis of Peptide 1

2 3

4

5

6

7

8

Amino acids present Cysteic acid Glycine Aspartic acid Glutamic acid Cysteic acid Glycine Valine Cysteic acid Valine Cysteic acid Glycine Leucine Cysteic acid Leucine Cysteic acid Glycine Valine Leucine Cysteic acid Valine Leucine

Peptide

DNP-Peptide

xxxx xxxx

xxx

-

Structure CySOaH.Gly (ASP, Glu)

X X

xx

X

X

X

xx xxx xxx

xx

-

X

X

X

?

X

xxx xxx xx xx xx xx xxxx xxxx xxxx

Val. (CyS03H, Gly) Val.CyS03H Leu. (CyS03H, Gly)

xxx

Leu. CySO3 H

xx xx xx

Leu.(CyS03H, Gly, Val)

xxx xxx

Leu.(CyS03H, Val)

-

-

two phenylalanine residues. It was not, however, possible to fit these five sequences together into a single unique structure. This was partly due to the lability of certain bonds t o hydrolysis, especially those involving the amino groups of the serine and threonine residues. Thus no peptide was identified which contained these bonds intact. Another difficulty was experienced in separating the less polar peptides containing phenylalanine, leucine, valine, and tyrosine, since these residues are grouped together in the polypeptide chain and give rise to a large number of similar peptides. The action of the proteolytic enzymes pepsin, trypsin, and chymotrypsin on this fraction was next investigated (Sanger and Tuppy,

TABLEX I Peptides Identified in Hydrolyzates of Fraction B of Oxidized I n s u l i n (Sanger and T u p p y , 19510, b ) Dipeptides from acid Phe.Va1 Glu.His CySO3H.Gly His.Leu Glu.Ala CySOaH.Gly Arg.Glg Leu.Va1. Ala.Leu Leu.Va1. Gly.Glu Gly.Phe Val.Asp His.Leu and alkaline hySer.His Val.Glu Val.CySO3H Glu.Arg Asp.Glu Leu.CySOaH drolyeates Leu.CySOaH.Gly Val.Glu.Ala Tyr.Leu.Va1 Gly.Glu.Arg Tripeptides from acid Phe.Val.Asp and alkaline hyGlu.His.Leu Ser.His.Leu Val.CySO3H.Gly drolyeates Val.Asp.Glu Leu.Val.Glu Leu.Va1.CySOaH His.Leu.CyS08H Ala.Leu.Tyr Higher peptides from Phe.Val.Asp.Glu Ser.His.Leu.Va1 Tyr.Leu.Val.CySO3H acid and alkaline His.Leu.CySOsH.Gly Leu.Va1 .Glu.Ala Leu.Val.CySOaH.Gly hydrolyzates Phe.Val.Asp.Glu.His Ser.His.Leu.Val.Glu Glu.His.Leu.CySO3H His.Leu.Val.Glu Ser.His.Leu.Val.Glu..41s Sequences deduced Phe.Val.Asp.Glu.His.Leu.CySOtH.G1y Tyr.Leu.Val.CyS01H.Gly from above pepSer.His.Leu.Va1.Glu.Ala Gly.Glu.Brg.Gly tides Peptides identified in Phe.Val.Asp.Glu.His.Leu.CySO~H.Gly.Ser.His.Leu Leu.VaLCyS03H.Gly.Glu.Arg.Gly.Phe peptic hydrolyzate Val .Glu.Ala.Leu

Lys.Ala Thr.Pro Pro.Lys.Ala

Thr.Pro.Lys.Ala

Thr .Pro .Lys .Ala

Tyr.Thr.Pro.Lys.Ala

His.Leu.CySOaH.Gly.Ser.His.Leu Tyr.Thr.Pro.Lys.Ala PeDtides identified in Phe.Val.As~.Glu.His.Leu.C~SO~H.Glv.Ber.His.Leu.Val.Glu.Ala.Leu.Tpr chymotryptic Leu.Val.CyS01H.Gly.Glu.Arg.Gly.Phe.Phe hydrolyzate Peptides identified in G1y.Phe.Phe.Tyr.Thr.Pro.Lys. tryptic bydrolyeate Ala Phe.Va1.Asp.Glu.His.Leu.(CyS-) .Gly.Ser.His.Leu.Val.Glu.Ala.Leu.Tyr.Leu.Val.(CyS-) .Gly.Glu.Arg.Gly.Phe.Phe.Tyr.Thr.Pro.Lys. Structure of the phenylalanyl chain Ala of insulin

THE ARRANGEMENT OF AMINO ACIDS I N PROTEINS

57

1951b). The larger peptides present, which contained up to about fifteen residues, could be satisfactorily separated on paper chromatograms and with the knowledge obtained from the lower peptides it was possible to deduce the structure of many of them from their amino acid composition and N-terminal residues and to work out a unique sequence for fraction B. These results are also shown in Table XII, where the structure of those peptides which played a major part in the elucidation of the sequence are given. The structure of fraction B was worked out as the only possible sequence which would fit all the experimental results, assuming that it was a single polypeptide chain of about thirty residues. The fact that it was possible to find a unique sequence was regarded as proof that such an assumption was correct, and that there is only one type of phenylalanyl chain in insulin. 5 . Ovalbumin Using the DNP method it has been shown that ovalbumin has no N-terminal residue (Desnuelle and Casal, 1948; Porter, 1950a). Either the amino groups at the end of the chains are masked by the carbohydrate moiety or the protein contains one or more cyclopeptide units. A very specific type of hydrolysis occurs when ovalbumin is incubated with a proteolytic enzyme prepared from B. subtilis. A new crystalline protein, plakalbumin, is produced (Linderstrgm-Lang and Ottesen, 1949), which differs from ovalbumin in crystalline form and in solubility and has a somewhat lower molecular weight (Guntelberg and Linderstrgm-Lang, 1949). A t pH values below 7.0 it has a slightly different electrophoretic mobility which is concordant with the loss of two acidic groups per molecule (Perlmann, 1949). The conversion is accompanied by the liberation of six atoms of non-protein nitrogen per molecule of ovalbumin (Eeg-Larsen et al., 1948). This fraction contains two free amino groups, 3.6 free carboxyl groups and 4 atoms of peptide bond N per six nitrogen atoms; it contains no free amino acids but on hydrolysis yields alanine, glycine, valine and aspartic acid (Villee et al., 1950). The results suggest that about two small peptides are present. Since the ratio of the rate of formation of plakalbumin to the rate of formation of nonprotein N was not constant, it was evident that the reaction takes place in a t least two stages, probably the rupture of two peptide bonds. On further incubation with the enzyme, the plakalbumin is broken down slowly to a mixture of products. Clearly this is a case where it is possible to obtain characteristic degradation products at the early stages of an hydrolysis. It would seem that about two peptide bonds are split about ten times as rapidly as any others.

58

F. SANGER

6. 7-Globulin

Porter (1950a) found rabbit y-globulin to have one N-terminal alanine residue per molecule of molecular weight 160,000, which suggests that it is probably a single long chain of over a thousand residues. IIt: also determined the N-terminal peptide sequence. All the DNP-peptides found fitted into the one sequence DNP-Ala.Leu.T’al.Asp and the fifth residue was probably glutamic acid. There were no other D N P peptides present and the whole of the terminal alanine residues could be quantitatively accounted for in peptides fitting the above sequence. In contrast t o the results of physicochemical studies, these findings suggest that the protein is chemically homogeneous, since it would seem unlikely that two different proteins would have the same terminal tetrapeptide sequence. It is difficult t o be absolutely certain about this question, since nothing is known about the principles that govern the order of amino acid residues in proteins, but it may be pointed out that 011 a purely random basis there are 1g4 possible terminal tetrapeptide sequences. The fact t ha t various different residues have been found to occupy the N-terminal positions in proteins indicates that there is no generd principle that defines closely the amino acid that occupies a particular position in a protein, so that the presence of a single N-terminal peptide sequence in a preparation can best be explained on the basis of chcmiral homogeneity. Porter also compared the N-terminal peptides of normal 7-globulin and of purified antivoalbumin, which was studied in the form of a specific precipitate with ovalbumin. No difference could be found, again suggesting the chemical similarity between the antibody and the normal y-globulin, from which it is formed. An alternative explanation is th a t only a small part of the polypeptide chain, “the active center,” has a different amino acid sequence in the antibody. In an attempt to identify the “active center” of the antiovalbumin molecule, Porter (I 950b) studied the action of proteolytic enzymes on it. By the action of papain a molecule about a quarter the size of the original 7-globulin was split off, which acted as a specific inhibitor in the antibody reaction. Since it had a N-terminal alanine group it appeared to come from the N-terminal quarter of the y-globulin molecule. On further hydrolysis all activity was lost and it was not possible t o obtain an active molecule small enough for chemical investigation. These results illustrate the possibilities of applying chemical methods to biologically active proteins, most of which have been studied so far only by physicochemical techniques.

T H E ARRANGEMENT O F AMINO ACIDS I N PROTEINS

59

7. Hemoglobin Porter and Sanger (1948) determined the N-terminal residues of hemoglobins from a number of different animal species. These results are summarized in Table I. Considerable species differences were evident both in the nature of the N-terminal residues and in the number of open polypeptide chains present. I n the case of horse hemoglobin which was studied in more detail, there are six valyl N-terminal residues and therefore six open polypeptide chains. It is interesting to note th a t the cystine content is not more than three residues, so that some of the chains must be held together by another type of cross-linkage. An investigation of the N-terminal peptides by the DNP-technique showed the presence of the following N-terminal sequences : Val.Leu, Val.Glu (--NH2) .Leu Thus there are a t least two different types of polypeptide chains present in the molecule (Sanger, 1948).

8. Gelatin Gordon et al. (1943) attempted to fractionate the acetyl derivatives of peptides from gelatin on silica gel chromatograms. The hydrolyzate was separated into neutral and basic fractions by ionophoresis in a three compartment cell. The neutral peptides were then fractionated on a silica chromatogram using ethyl acetate, and each fraction was refractionated using butanol-chloroform. I n general the fractionation was not sufficient t o give clear-cut results though the presence of the following peptides was suggested : Leu.Gly, Gly.Leu, Gly.Pro, a dipeptide containing proline and alanine, and a t tripeptide containing proline, alanine and glycine. These investigations are largely of historical interest, since they mere carried out before the introduction of paper chromatography. The amino acids present in the peptides were identified and estimated as their acetyl derivatives on silica chromatograms. VIII. GENERALCONCLUSIONS It is clear from the above that considerable progress has been made in recent years in the development of methods for investigating the arrangement of amino acid residues in proteins. These methods have not yet been extensively applied and a vast amount of work is still required in this field. Such work may be expected t o be rather unrewarding a t first. The separation of a few peptides from a protein is not likely to

60

F. SANGER

make possible the formulation of any general theory of protein structure or to explain the physiological action of a protein. Only by the accumulation of a large amount of experimental evidence can such objectives be attained. It does not appear, however, a t present th a t there is any easy short cut t o the solution of the problem of protein structure and action, Probably only when more is known about the exact chemical structure of proteins will it be possible to understand the unique part played by proteins in the living organism. Every peptide identified will contribute towards this ideal, even though it may appear to have no significance in itself. By the application of the methods described in this review the structure of a pure polypeptide containing thirty residues has been determined (p. 54) and there seems no reason why it should not be possible to work out the complete amino acid sequence in proteins which are as simple as insulin. How far it will be possible t o apply these techniques t o more complex proteins is difficult to say. The larger the polypeptide chains in a protein, the greater the necessity of isolating larger peptide breakdown products. Probably the chief need in this field is for techniques for the specific breakdown of proteins into larger peptides and for the fractionation of such peptides. Most of the more commonly studied proteins contain more than 300 residues but it is possible that some of them, when studied in greater detail may be found to have a simpler structure than is a t present believed. The relative simplicity of insulin may be merely apparent as insulin has been studied in more detail than have other proteins. It has frequently been suggested that proteins may not be pure chemical entities but may consist of mixtures of closely related substances with no absolute unique structure. The chemical results so far obtained suggest that this is not the case and that a protein is really a single chemical substance, each molecule of one protein being identical with every other molecule of the same pure protein. Thus it was possible to assign a unique structure to the phenylalanyl chains of insulin. Each position in the chain was occupied by only one amino acid and there was no evidence that any of them could be occupied by a different residue. Whether this is true for other proteins is not certain but it seems probable that it is. The N-terminal residues of several pure proteins have been determined (Table I) and this position is always found t o be occupied by a single unique amino acid. These results would imply a n absolute specificity for the mechanisms responsible for protein synthesis and this should be taken into account when considering such mechanisms. It is impossible with the small amount of experimental evidence a t present available to form any general theory of protein structure or to

THE ARRANGEMENT OF AMINO ACIDS IN PROTEINS

61

formulate any principles that govern the arrangement of amino acids in proteins, though several such theories have been put forward on very much less evidence. Certainly there is no simple periodic arrangement of residues along the chains of the type suggested by Bergmann and Niemann (1936). The presence of peptides such as Glu.Glu in a hydrolyzate of wool (Consden et al., 1949) precluded this possibility and no periodicity whatsoever was evident in the structure of the phenylalanyl chains of insulin (Table XI). Although they are not immediately apparent, it still seems probable that there may be certain principles which determine amino acid sequences. The mechanisms of protein synthesis, about which almost nothing is known, would be expected to have their limitations so that one might at least expect to find certain sequences that occur more frequently than others. The results at present available do suggest that this may be the case. Thus, for instance, in the phenylalanyl chain of insulin there are three dipeptide sequences (His.Leu, Leu.Va1, CyS.Gly) that occur twice in the chain of thirty residues. One of these (CyS.Gly) also occurs in glutathione. Other dipeptide sequences in this chain were Thr.Pro, which was also detected in the antibiotic actinomycin (Dalgliesh et al., 1950) and Ala.Leu, which was found in gramicidin (Synge, 1949). The sequence Glu.Glu occurs both in wool (Consden et al., 1949) and in insulin (Sanger, 1949b) and is probably also present in gliadin, since Nakashima (1927) obtained a peptide fraction that contained tyrosine and glutamic acid in the ratio 1 :3. It may be that this sequence has a special significance since glutamic acid and glutamine residues frequently occur linked together in natural products, such as folic acid, the capsular substance from B. Anthracis (Hanby and Rydon, 1946) and in the derivative of triglutamine isolated by Dekker et al. (1949) from a marine alga. In this connection Woolley (1949) has suggested that the same amino acid sequence occurs both in insulin and in trypsinogen, since two similar fractions were obtained from tryptic digests of the two proteins. How far these results do reflect a general principle of protein structure will only be known when considerably more experimental evidence is available. The results obtained with wool (p. 48) on the contrary suggest that almost every possible dipeptide containing glutamic acid or cystine is produced on hydrolysis of this protein and that there are therefore no obvious limitations to the type of sequences that can occur. It is certain that proteins are extremely complex molecules but they are no longer completely beyond the reach of the chemist, so that we may expect to see in the near future considerable advances in our knowledge of the chemistry of these substances which are the essence of living matter.

62

F. SANGER

ACKNOWLEDGMENTS I wish to express my thanks to Dr. A. C. Chibnall, Dr. R. L. M. Synge, and Dr. H. Tuppy for valuable discussions and criticisms which have greatly helped in the preparation of the present review.

REFERENCES Abderhalden, E. (190%). 2.physiol. Chem. 62, 315. Abderhalden, E. (1909b). 2.physiol. Chem. 63,401. Abderhalden, E. (1940). 2.physiol. Chem. 266, 23. Abderhalden, E. (1943). Z. physiol. Chem. 277, 248. Abderhalden, E., and Bahn, A. (1932). 2. physiol. Chem. 210, 246. Abderhalden, E., and Bahn, A. (1933). 2. physiol. Chem. 219, 72. Abderhalden, E., and Bahn, A. (1935). 2. physiol. Chem. 234, 181. Abderhalden, E., and Brockmann, H. (1930). Biochem. 2. 226,386. Abderhalden, E., and Komm, E. (1924a). 2. physiol. Chem. 134, 121. Abderhalden, E., and Komm, E. (1924b). Z. physiol. Chem. 139, 147. Abderhalden, E., and Mahn, H. (1927). 2. physiol. Chem. 169, 196. Abderhalden, E., and Mahn, H. (1928). 2. physiol. Chcm. 174,47. Acher, R., Jutisz, M., and Fromageot, C. (1950). Biochim. et Biophys. Acta 6,493. Agren, G. (1940). Arkiu. Kerni Mineral. Geol. 14B,No. 21. Bailey, K. (1951). Biochern. J . 49, 23. Barry, G.T., Gregory, J. D., and Craig, L. C. (1948). J . Biol. Chem. 176, 485. Behrens, 0. K., and Bergmann, M. (1939). J . B i d . Chem. 129, 587. Bell, P.H., Hone, J. F., English, J. P., Fellows, C. E., Howard, K. S., Rogers, 31. M., Shephard, It. G . , and Winterbottom, R. (1949). Ann. N . Y . Acad. Sci. 61,8!17. Bergmann, M. (1942). Advances in Enzyrnol. 2, 49. Bergmann, M., and Behrens, 0. K. (1938). J . Bid. Chem. 124, 7. Bergmann, M.,and Fraenkel-Conrat, H. (1937). J . Biol. Chem. 119, TOT. Bergmann, M., and Fraenkel-Conrat, H. (1938). J . Bid. Chem. 124, 1 . Bergmann, M., and Fruton, J. S. (1938). J . Biol. Chem. 124,321. Bergmann, M., and Fruton, J . S. (1941). Advances in E’nzywiol. 1, 63. Bergmann, M., and Miekeley, A. (1924). 2. physiol. Cheni. 140, 128. Bergmann, M., and Niemann, C. (1936). J . B i d . Chem. 116,77. Biserte, G. (1950). Biochim. el Biophys. Acta 4, 416. Blackburn, S. (1949). Biochem. J . 46, 579. Blackburn, S.,and Lowther, A. G. (1950). Biochem. J . 46, sxvii. Block, R. J. (1942). Proc. SOC.Exptl. Biol. Med. 61,252, Block, R. J. (1949). I n Ion Exchange, edited by Nachod, F. C. Academic Press, New York, p. 295. Boissonnas, R. A. (1950). Helv. C h i m Acla 33, 1975. Boldingh, J. (1948). Experientia 4, 270. Bowman, R. E. (1960). J . Chem. SOC.1349. Brand, E., and Edsall, J. T. (1947). Ann. Rev. Biochem. 16, 224. Brand, E., and Kassell, B. (1939). J . Biol. Chem. 131,489. Brigl, P. (1923). Ber. 66, 1887. Bull, H. B., and Hahn, J. W. (1948). J . Am. Chern. SOC.70, 2128. Butler, J. A. V., Dodds, E. C., Phillips, D. M. P., and Stephen, J. M. L. (1948). Biochem. J . 42, 116.

THE ARRANGEMENT O F AMINO ACIDS I N PROTEINS

63

Butler, J. A. V., Phillips, D. M. P., Stephen, J. M. I,., and Creeth, J. M. (1950). Biochem. J . 46, 74. Cannan, R. K. (1944). J. Biol. Chem. 162,401. Carpenter, D.C. (1940). J . Am. Chem. SOC.62,289. Carpenter, D. C. (1941). J . Franklin I n s t . 232, 76. Chibnall, A. C., and Rees, M. W. (1951). Biochem. J . 48, xlvii. Christensen, H. S. (1943). J . Biol. Chem. 161,319. Christensen, 11. S. (1945). J. Biol. Chern. 160, 75. Christensen, I-I. N., and Hegsted, D. M. (1945). J . Biol. Chem. 168, 593. Coleman, D., and Howitt, F. 0. (1946). Fibrous Proteins. Society of Dyers and Colourists, &adford, p. 144. Coleman, D., and Howitt, F. 0. (1947). Proc. Roy. SOC.London A190, 145. Consden, R. (1948). Nature 162, 359. Consden, It., and Gordon, A. H. (1950). Biochem. J . 46, 8. Consden, R., Gordon, A. H., and Martin, A. J. P. (1944). Biochem. J . 38, 224. Consden, R., Gordon, A. H., and Martin, A. J. P. (1946). Biochem. J . 40, 33. Consden, R., Gordon, A. H., and Martin, A. J. P. (1947a). Biochem. J . 41, 590. Consden, It., Gordon, A. H., and Martin, A. J. P. (1948). Biochem. J . 42, 443. Consden, R.,Gordon, A. H., and Martin, A. J. P. (1949). Biochem. J . 44, 548. Consden, R., Gordon, A. H., Martin, A. J. P., and Synge, R. L. M. (194713). Biochem. J . 41, 596. Craig, L. C. (1944). J. Biol. Chem. 166, 519. Craig, L.C., Gregory, J. D., and Barry, G. T. (1949). Cold Spring Harbor Symposia Quant. Biol. 14,24. Craig, L. C., and Post, 0. (1949). Anal. Chem. 21, 500. Cremer, H. D., and Tiselius, A. (1950). Biochem. Z. 320, 273. Dalgliesh, C.E., Johnson, A. W., Todd, A. R., and Vining, L. C. (1960). J . Chern. SOC.2946. Dekker, C. A., Stone, D., and Fruton, J. S. (1949). J . Biol. Chem. 181, 719. Desnuelle, P., and Casal A. (1948). Biochim. et Biophys. A d a 2, 64. Desnuelle, P., Rovery, M., and Bonjonr, G. (1950). Bzochim. et Biophys. Acta 6, 116. Dim, K., and Felix, K. (1932). 2. physiol. Chem. 206, 83. Drake, B. (1947). Nature 160,602. Drake, B. (1950). Acta Chem. Scand. 4, 554. Drucker, B.,and Smith, S. G. (1950). Nature 166, 196. Dunn, M. S. (1925). J. Awl. Chem. SOC.47,2564. Durrum, E. L. (1950). J . Am. Chern. SOC.72, 2943. Edman, P. (1950). Acta Chem. Scancl. 4, 283. Eeg-Larsen, N., Linderstr@m-Lang, I N-H . . . 0 = C 50Rh

0.003b

0.005i 0 .85k 1.6gk 2.5k

l . l d

1.6d 2.7d

isotonici ~~

0

Portzehl (1950b).

c

A. Weber (1951). Gasser and Hill (1924).

Strobe1 (1951). Hill (1950b). 1 Ramsey (1947). 0 Hill (1951). d 8

h

Hill (1950~).

i Ebner

(1916).

i Josenhans (1949). 1

Fischer (1947). Ralston et al. (1947. 1949). Haxton (1944).

inactive uncontracted or inactive contracted state (cf. curves 1 and 3 of Fig. 4). The elastic after-effect following a stretch is only about half as great for the fiber model as for living muscle (cf. Figs. 5 and 6 ; and Table 11, column 6; also Buchthal et al., 1947).

176

HANS H. WEBER AND HILDEGARD PORTZEHL

The great rigidity and stiffness of the models must be properties of the contractile substance, since they disappear in presence of ATP (Fig. 6; Szent-Gyorgyi, 1949), and since the actomyosin thread, in any case, is composed only of contractile material. The difference between the contractile protein of the models and that of muscle becomes even greater if one attributes the elastic resistance of resting living muscle entirely to the sarcolemma and the connective tissue (Ramsey and Street, 1940, 1941; A. V. Hill, 1949a, 1950a). The con-

RELATIVE LENGTH, 1 0 0 L / L o

FIG.9. Variation of tension with length on progressive release of the stretched model fiber, the stretched actomyosin thread and the stretched rabbit psoas muscle. Curves 1 and 2: from A. Weber (1951); curve 3: from Portzehl (1950b); curves 4 and 5, from Szent-Gyorgyi (1949).

tractile substance must then be regarded as almost completely plastic in muscle, and exceedingly rigid in both kinds of model. We turn now to a comparison of the active state of the model with the active state of the muscle. I n the active state the models become less rigid, If by working at a low temperature (OOC.) the tension developed by contraction is kept low, the fiber model can be stretched to 145% of its resting length without breaking (Szent-Gyorgyi, 1949). The rise of tension with increasing length is smaller than in the inactive model and the elastic after-effect greater (Fig. 6 ; column 3 of Table 11; Engelhardt and Ljubimova, 1939; Buchthal et al., 1947; column 6 of Table 11). The contractile material of living muscle, on the other hand, passes from a very extensible to a more rigid state as the active phase sets in (A. V. Hill,

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

177

1950a; but compare Buchthal 1942 and Sichell935). For this reason the elastic resistance to sudden stretch becomes greater (Fig. 10). Thus in the active state no marked difference is demonstrable between the elastic properties of the models and those of living muscle. ATP decreases the pronounced rigidity of the inactive model by giving a greater freedom of movement to the “frozen” actomyosin particles. This is one action of ATP on the models which is not demonstrable in living muscle. The other action of ATP is to produce contraction. Tension, shortening, redevelopment of tension after release, temperature dependence of tension, and a structural rearrangement of the contractile molecules are

I I I

I

0

20

40

60

I

80

TIME AFTER STIMULATION, rn. sec.

FIG.10. Increase of tension of living muscle on stretch. Top curve: stretched during latent period; middle: during rest; bottom: difference between the two (from A. V. Hill, 1950~).

all properties in which the contracted models resemble contracted muscle. The differences between model and muscle are not greater than those shown by different kinds of living muscle, ranging from the slow smooth muscle to the fast striated type. There is the possible exception of the low tension developed by the actomyosin thread. Some of the contractile properties of the models, however, differ markedly from those of the particular muscle from which they have been prepared (Table 11, columns 1, 5 and 7). One of the most important of these differences, perhaps, is the slowness with which the models redevelop their tension after a release; the time in the case of living muscle is short and characteristic of the speed of contraction (A. V. Hill, 1926; Gasser and Hill, 1924). It is hardly surprising that an actomyosin thread from a striated muscle no longer has the same short recovery time as that of the muscle (Table 11, column 5), for the actomyosin from both slow and fast muscles appears to be much the same (Hamoir 1949). Actin and 1.-myosin from quite different animals combine to give actomyosins with the characteristic properties of the natural ones (Cigada

178

HANS H. WEBER A N D HILDEGARD PORTZEHL

et al., 1948), and all known actomyosins behave identically or similarly with ATP (Guba, 1943; Roth, 1946; Csapo, 1949). It is therefore questionable whether there are “fast” and “slow” actomyosins. The difference between fast and slow muscles may rather be due t o the fact that actomyosin is built into structures of varying degrees of perfection. Much more of the “perfect” structure of the rabbit psoas muscle is retained when it is extracted with glycerol-water than when an actomyosin thread is prepared from it; and correspondingly, this model prepared by glycerol-water extraction in an isometric ATP-contraction develops as high a tension as the living muscle in a tetanus, and the redevelopment of tension, though 200 times as slow as in living muscle, is twice as fast as in the actomyosin thread (Table 11, column 5). Now i t is evident that even quite small structural disturbances suffice t o reduce the speed of contraction of skeletal muscle t o that of smooth muscle. When an undamaged muscle fiber is forced t o contract actively t o 7 5 4 0 % of the initial length, it passes into the so-called delta state (Itamsey and Street, 1940; A. V. Hill, 1949b), and when the muscle fiber is in this state, shortening is very slow: “When shortening is extensive, it is very slow” (Ramsey, 1947). In a similar way, the models contract extensively and rather slowly, and the speed of contraction depends upon the degree of shortening (Table 11, Fig. 3). The contraction of the models thus seems t o resemble that of muscle in the delta state more closely than that of undamaged skeletal muscle. The precision and speed of the mechanical changes in contraction may be represented in a series increasing in complexity and development : the actomyosin thread smooth muscle < model fiber delta state of living skeletal muscle < undamaged skeletal muscle. In such an unbroken series of contraction types, it is unlikely that a t any one stage the fundamental process should suddenly change in character. Probably, therefore, contraction is due in all cases to the same process as occurs in the models-the interaction of actomyosin and ATP. This raises two questions. How does ATP abolish the rigidity of the models, and how are they brought into an active, contracted state?

-

-

4. Thermodynamics of Contraction in the Models a. I s Contraction a Steady State or a New Thermodynamic Equilibrium? The small rise in tension with temperature of a resting muscle,

an inactive fiber model or an inactive actomyosin thread is almost certainly due t o shift in a reversible thermoelastic equilibrium. But is the same true of the much greater temperature dependence of tension in tetanized muscle and in the models when brought into the active state? If so, the very old theory that a new elastic state is created (E. Weber,

MUSCLE CONTRACTION

AND FIBROUS MUSCLE PROTEINS

179

1846) would be confirmed. With the models this possibility can be tested, for here the conditions inducing contraction are relatively clear cut. If a shift in the position of a temperature-dependent equilibrium is involved, then, in the active state, the tension a t a given temperature should be the same whether that temperature is reached from above or below. This condition is fulfilled very exactly (Szent-Gyorgyi, 1949;

TEMPERATURE,

OC.

FIG.11. Attainment of equilibrium tension on changing the temperatures Upper half; model fiber (A. and H. H. Weber, 1951); lower half: actomyosin thread. (Portzehl, 1951). -0-0temperature falling; -0-0temperature rising; equilibrium tension reached from above, from below.

I

see also Fig. l l ) , but this fact does not suffice to show the reversibility of the energy changes with temperature, since the models split ATP very considerably (Korey, 1950; Heinz and Holton, 1950; Portzehl, 1950b). If the energy of decomposition is the cause of contraction (see Section 11, 4d and e ) , then the temperature dependence of the tension could be solely due to the fact that ATP is split more slowly a t the lower temperatures. In this case the energy change would only be a question of magnitude and not of sign. The changes in tension would not be thermodynamically reversible, but would be due to shift in a steady state, the position of which is temperature-dependent. b. Energy Requirement and Availability in the Case of Equilibrium and of Steady State. Considering first an equilibrium, the models can be

180

H A N S H. W E B E R AND HILDEGARD PORTZEHL

treated according to the elasticity equation of Wiegand and Snyder (1 934) :

K = &)T

+ T (g)L

where K is the elastic force, U the internal energy, L the length and T the absolute temperature. The application of this equation to the

RELATIVE LENGTH, IOOL/Lo

FIG.12. Changes on shortening in the thermoelastic force (curves 1 and la), the potential-elastic force (curves 2 and 2a) and the mechanical force (curves 3 and 3a). Left, model fiber (A. Weber and H. H. Weber, 1951); right, actomyosin thread (Portzehl, 1950b).

tension-temperature diagram (Fig. 8) shows that the thermoelastic force

[

($$)L

*

T] is ten to forty times greater than the mechanical force

(Fig. 12). This would mean that the internal energy

yg)T

increases

on shortening because cohesional forces which tend to hold the contractile particles in the extended position are overcome. These cohesional stretching forces should really become greater as the length decreases, because the system is moving away from the point a t which their value is zero. In fact, however, they become smaller in both models (Fig. 12, curves 2 and 2a). This is the first argument against treating the temperature dependence of the tension as an equilibrium reaction.

MUSCLE CONTRACTION A N D FIBROUS MUSCLE PROTEINS

181

On the other hand, calculation shows that the thermokinetic forces supply more than enough energy for the contraction, and the same applies to the energy provided by the splitting of ATP. A t an ATP concentration of 5 X mol of ATP per M , 1 ml. of fiber model splits 1.2 X minute, which gives 0.14 cal. per minute. Since complete contraction in these experiments takes at least an hour, not less than 8 cal. are available from the decomposition. Complete contraction yielding maximal tension cal. It is thus impossible on requires a work equivalent of 3-4 X the basis of energy requirement to differentiate between the alternatives of a thermokinetic elasticity and a stationary state. c. Connection between A TP Concentration and Temperature Dependence of the Tension for the Case of Thermodynamic Equilibrium. According to Wohlisch (1926; 1940), Meyer et al. (1932); Kuhn (1936a, b), thermoelastic shortening takes place when thread-like elementary particles are (i) flexible enough to coil up, and (ii) not hindered from doing so by cohesional forces. When such part,icles are held extended by external sources, a tension is produced which increases by 4 4 7 3 for a temperature increase of 1°C.; i.e., the thermal coefficient of the temperature p, is 1 -1. -AK =where K = force and T = temperature. K AT 273 In the active state of the models, ,8 is in fact very much greater than >473, and on the basis of the thermokinetic theory this would mean that a rise of temperature increases not only the force exerted by individual particles, but, by virtue of a sort of melting of the stretched crystallites, increases also the number of freely moveable particles. Since, however, the tension rises on increasing not only the temperature but also the ATP concentration (up to an optimum-see Section 11, 4e) ATP must, on the basis of this theory, be regarded as a substance which lowers the melting point of the crystallites, reversibly diminishing the cohesive forces. As there is some evidence in favor of such a theory it is tempting t o assume the diminution of cohesive forces by ATP to be the cause of contraction (cf. Sections 11,3g and II,4h). All effects of ATP could then be explained by the same mechanism. These effects are: the dissociation of the actomyosin complex in solution; the shrinking of unoriented gels; the contraction of the models; and their subsequent gelation or freezing when ATP is removed (see Sections 111, 5a, 11, 2 and 3g). But one consequence of this concept is not confirmed experimentally. If ATP works by lowering the melting point, then the temperature range over which melting takes place-taken as that over which the temperature coefficient of the tension is especially high-should shift towards higher temperatures as the ATP concentration decreases. This is by no means borne out in actual fact (Fig. 13). The whole rise in tension

182

HANS H. W'EBER AND HILDEGARD PORTZEHL

between 0 and 20" becomes smaller and is concentrated in the range 0-10' as the ATP concentration decreases. This is the second argument against the concept that ATP contraction is due t o thermokinetic equilibrium. There is another less direct way, however, in which heat could be the energy source of contraction. It could be consumed in an endothermic chemical reaction which in turn leads, not to a statistical coiling, but to a regular folding of threadlike molecules (Varga, 1946; Szent-

10 20 TEMPERATURE, OC.

0

FIG. 13. Comparison of tension increase in the ranges 0-10" and 10-20" for Al different ATP concentrations. Curve 1: 0.67 X 10-4 M ATP; curve 2; 5 X ATP; curve 3: 24 X 10-4 M ATP. Material, model fibers (A. and H. H. Weber, 1951).

Gyorgyi, 1947, 1949) ; rather like the supercontraction theory of Astbury and Dickinson (1940). (See also H. H. Weber, 1934a and b.) It is possible to treat the temperature-dependence of this reaction as follows : the tension of the contraction is determined by the saturation of the contractile protein with the reaction ~ a r t n e r a; ~t full saturation, the a The reaction partner need not be ATP itself (cf. Section 11, 4g); it need only be in reversible equilibrium with ATP, so that its concentration is determined b y that

of ATP.

Degree of saturation s

=

K

i (ATP)

where K = "apparent

constant (Michaelis, 1922) of the ATP-actomyosin complex.

"

dissociation

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

183

tension is maximal. The degree of saturation is dependent upon the ATP concentration and the affinity constant of protein and reaction partner, which varies with temperature. The three saturation curves for 0, 10, and 20" respectively (left of Fig. 14) fulfil this condition. For these three temperatures three values of saturation are given for each of the three ATP-concentrations A, B, and C on the left side of Fig. 14 and three corresponding values of tension on the right side. For each concentration of A TP one finds three proportional values of tension. These latter three curves show that the range of temperature in which tension

FIG. 14. Temperature dependence of tension, calculated assuming an endothermic reaction, whose extent is determined by the ATP concentration (from A. and H. H. Weber, 1951). A, B, and C are concentrations of ATP for which the degrees of saturation a t 0", lo", and 20" are compared, and curves A, B, and C show the increase of tension for ATP concns. A, B and C in the ranges 0-10" and 10-20".

is rising steeply is shifted to higher temperatures the lower the AT P concentration (right hand side of Fig. 14). Tha t in actual fact the opposite is observed is an argument against Szent-Gyorgyi's theory th a t contraction results from an endothermic reaction proceeding to equilibrium. d . Inhibition of Adenosine Triphosphatase Activity and Contraction. It appears that when the ATP-splitting activity of myosin is inhibited there is a n accompanying inhibition of those reactions between myosin and A T P which result in colloidal changes in the actomyosin complex (e.g., contraction, shrinking, superprecipitation), and in general the diminution in the two activities even run parallel on a roughly quantitative basis (Table 111). Only one exception is recorded. Buchthal et al. (1947), found t hat actomyosin threads whose ATPase activity had been

r

M

TAESLE I11

rp

The Parallelism between A TPase Znhibition and the Znhibition of Various Actomyosin Reactionsa ATPase Molarity Inhibition

of

by

reagent 4 X 3X 2 X 3.5 X 1X

HzOz Iodine

1

o-Iodoso benzoate Iodacetamide Salyrganb

{

pChloromcrcuribenzoate Benzaldehyde Ageing (20 days at pH 7.4) Ageing (77 days) Cupric glycinate Dialysis against HzO at pH 6.2

Svstem employed

lo-'

Lmyosin4 10-1 L-myosind 10-1 L m y o s i n 4 10" L-myosin" 10-8 L-myosine

1 X 10-2 Glycerol fiber

4 X 10-2 L-myosin< 1 x 10-1 >2 X 10-4 Actomyosinf > 1 . 5 X 10-4 L-myosin

{

2 X 10-4 L-myosin< 2 X 10-3 Actomyosin/

Glycerol fiber L-myosins 5 X 10-4

Actomyosino Actornyosinb

Inhibition ( %)

Action on colloidal phenomena In absence of A T P In presence of ATP Inhibition Inhibition ( %) Character of reaction ( %) Character of reaction

Viscosity increase with F-actin'

90

80

-

-

100

F-actini Viscosity increase with F-acting

100

99

Strong Strong

-

Viscosity increase with F-actin' G-actin -+ F-actiw

-

-

80

70

100 80

Contraction (glycerol thread)e Shrinkage (thread)i

65

\'iscosity increaee with

100

-

85

9&100

-

97

Strong

-

-

100

-

-

-

-

"1

-

-

Shrinkage (thread)i Superprecipitationf .i

Superprecipitationl Contraction'

Superprecipitationo Shrinkage (threadp

+ +

100

With lo-* M Cysteine With Cysteine, 80-100 70

100

None None

-

None

100

Strong 0

* The action of inorganic ions is not given in this table. since they affect the properties of colloids in a general and not in a specific manner.

* Salicyl-(h-hydroxymercuri-~-methoxypropyl)amide-O-acetate.

Bailey and Perry (1947). d Ziff (1944). 0 Korey (1950). f Kuschinsky and Turba (19%). s Turba ct d. (1950). k Buchthal eta. (1947). i Kuschinsky and Turba (1950b). j Godeaux (1945). c

P

With 1 X 10-1 M Cysteine With 1 X 10-2 M Cysteine By dialysis, n o n e

+

+

None

F 2

None None None None

++ 100

Contraction'

Revcrsibility

$ P

1:

u

$u

M

ti P

a

U

Cd

0

a

z

i

r

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

185

largely destroyed4 by dialysis at pH 6 (according to the procedure of Singher and Meister, 1945), in presence of ATP were still capable of shrinking as much as normal threads.6 Since, however, this shrinking does not take place against a tension, its energy requirement is small and difficult to evaluate; large changes in the energy available from decomposition could lead only to small changes in shrinking. The parallelism between ATPase activity and colloidal changes might be used as an argument for the view that the ATPase activity is the cause of the colloidal changes, but the parallelism does not in reality prove as much; for when ATPase activity is inhibited so are those reactions of the contractile proteins which are not dependent upon ATP-the formation of actomyosin from actin and L-myosin and the polymerization of G-actin (Table 111, columns 5 and 6). It is possible, therefore, that the chemical factors affecting ATPase activity and contraction work side by side. e. Optimum ATP Concentration for Splitting and Contraction. The rate of ATP breakdown and the extent of contraction can, however, be modified without the use of inhibitors, for there are optimal concentrations of ATP for both processes; above these concentrations, the values for tension and for breakdown again decrease. This is the case both for the fiber and for the thread, and true also for the splitting and shrinkage which occurs in the superprecipitation of actomyosin (A. and H. H. Weber, 1951; see also Biro and Szent-Gyorgyi, 1949; Portzehl, 1950b). The optimal concentration of ATP for tension (A. and H. H. Weber, 1951) and breakdown (Biro and Szent-Gyorgyi, 1949; Heinz and Holton, 1950; Hasselbach, 1950) and also, apparently, for superprecipitation (Biro and Szent-Gyorgyi, 1949), all decrease as the temperature is Iowered, and for an exact comparison of the optima1 ATP concentrations for contraction and splitting, two conditions must be fulfilled when the measurements are made: (i) the rate of ATP breakdown must represent the initial velocity rate; (ii) the contractile system must be in the same state of shortening (cf. Section II,4f) or of shrinkage in the case of superprecipitation, for the two types of measurements. These conditions are fulfilled exactly in the measurements of A. and H. H. Weber (1951) and of Heinz and Holton (1950), both of whom used strongly contracted model fibers. At 2”, the optimum ATP concentrations are 10-2.26M for the development of tension and 10-2.aM for the 4 Mommaerts (1947) also found it impossible to destroy the ATPase activity of actomyosin completely by acidification-even after precipitating three times at pH 5.2. 5 Szent-Gyorgyi, however, finds that contractility disappears when the ATPase activity falls below 50%. He explains this as the effect of removing his “ATP-cprotin” by slightly acid pH (Szent-Gyorgyi, 1947, pp. 120-121).

186

HANS H. WEBER AND HILDEOARD PORTZEHL

splitting of ATP; at 20" the values are M and 1O-l.' M , respectively (Fig. 15). The rise and fall of the curves on either side of the optima are also quite similar. At the higher temperature however, the tension below the optimum concentration diminishes rather more quickly than the rate of breakdown (cf. curves 1 and l a of Fig. 15). This may be due to the fact that at lower ATP concentrations the diffusion barrier in the interior of the fiber is greater in the contraction experiments than in the enzymatic, since in the latter case the fibers are broken down by treatment in a blendor for several minutes.6 At an ATP concentration of 10-2.4-10-2.6 the hydrolysis rate (cf. curves l a and 2a of Fig. 15) and the tension (Fig. 8; Section 11, 3f)'show a twofold increase when the temperature is raised from 2 to 17". A comparison of the results of Biro and Szent-Gyorgyi (1949) with those of Korey (1950) reveals a similar parallelism between breakdown and structural changes in the protein. In these experiments, it is true, the measured hydrolysis rate does not represent an initial velocity, but the second of the conditions under (ii) above is fulfilled, for the preparations were as free to contract in the contraction experiments as in the enzymatic. The optimum ATP concentration found for shortening (Korey, 1950) and for breakdown (Biro and Szent-Gyorgyi, 1949) are both for extracted fiber bundles a t 20°C. The fact that this value is higher than for the isolated model fiber may be only an apparent discrepancy, for the effective concentration of ATP in the interior of the fiber bundle is appreciably lower than that in the bath. Finally, Biro and Szent-Gyorgyi also find a parallelism between the superprecipitation of actomyosin and the hydrolysis rate of ATP a t four different temperatures in the range 5-30°C. The optimal ATP concenThis interpretation contradicts the calculation (Table I, Section 11, 2) that the ATP concentration in a single fiber of the psoas is uniform throughout. I n this calculation the constant for free diffusion of ATP is used, and it is doubtful whether this is correct, since ATP breakdown is greater and the diffusion path shorter when the fiber is broken up into fibrils by long treatment in a blendor. Experiments are in progress to compare the diffusion constants of ATP in free solution and in t h e fiber model. The tensions a t the optima in Fig. 15 cannot be used to compare the relative magnitudes of the two optima, because (a) they are determined with different single fibers, (b) the number of experiments is small, and (c) different fibers under the same conditions develop very different tensions according to the extent they are denatured. In the experiments of Fig. 8, the tension is measured on the same fibers a t different temperatures, and, moreover, on a greater number of fibers. In any case, the correspondence between the temperature coefficients of the tension and hydroIysis rate has not very strict significance, since the one is measured on extended fibers and the other on freely suspended (Le., contracted) fiber particles (cf. 11, 4f).

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

187

tration for both processes shifts from 10-2JJto M when the temperature is raised from 5 to 30°C. The hydrolysis rate apparently has an ATP concentration optimum only when breakdown is accompanied by contraction or superprecipitation. According to Biro and Szent-Gyorgyi, pure L-myosin a t 20" splits ATP at the same rate for all concentrations between and l O - l . 9 6 M . Thus the parallelism between ATP breakdown and ATP contraction is not only close but also specific. These facts, together with the difficulties encountered when ATP contraction is treated as a purely elastic phenomenon, make it difficult not to regard the breakdown of ATP as the cause of contraction (cf. Section 11, 4b and c). f. ATP Optimum for Contraction and the Extent of Shortening. The ATP optimum for contraction is different for extended fibers under tension and for shortened ones. A t 0", fibers a t 40% of the equilibrium length or less develop their maximum tension in 10-2.26M ATP, a t 80-100 % in M ; at 20", the corresponding values are M and 10-2.2M respectively. g. The Explanation of the Diferent Positions of the A T P Optima. Assuming that the ATP concentration in the whole cross-section of the fiber equals that of the bath, one must conclude from the different positions of the ATP optima that the affinity of actomyosin and ATP varies under different conditions. Making this assumption, A. and H. H. Weber calculated from the temperature dependence of this optimum the heat of combination of ATP and actomyosin. They derive the curiously high value of about 10,000 cal. The assumption that the ATP concentration in the center of the model equals that of the bath can only be made if the diffusion constant for diffusion inside the model is not very different from the constant for diffusion in solution. There is no evidence for this assumption, and it is doubtful whether it is correct (compare footnote 6). If the diffusion constant in the model is much smaller, it is possible to explain the different position of the optima by different rates of decrease of the ATP concentration from the bath to the center of the model. This rate is given Ar2 by the Meyerhof formula 40 ( A = rate of breakdown, D = diffusion constant, r = radius), where A is changing with temperature and r with shortening. The interpretation of the different positions of the ATP optima in terms of affinity therefore seems not yet to be justified. h. Analysis of the Contraction of the Models and the Contraction Cycle of Muscle. At the close of Section 3 of the present review two questions were put: (i) how does ATP remove the rigidity of the models, and (ii)

188

HANS H. WEBER AND HILDEGARD PORTZEHL

how does it cause them to contract? If the answer to the second question is that active contraction is caused by ATP breakdown, then it necessarily follows that the answer to the first is that the actual combination with undecomposed ATP destroys the rigidity. This conclusion is reached by considering the temperature dependence of ATP breakdown. It is possible by changing the temperature to change the rate of breakdown without altering the level of ATP, but when ATP is washed out both the rate and the level fall to zero, and since the model becomes rigid the tension remains “frozen” in the structure. When, however, the active contracted model is cooled from 20 to O”, only the breakdown rate falls t o one-third to one-quarter of its former value, and with it the tension; the model does not become rigid (cf. Section 11, 39) and the tension is not “frozen in” because the ATP level and the amount of ATP combined are unaltered. An attempt will now be made to see t o what extent the types of contraction found in living muscle and in the models can be treated on a common basis, making the assumption that the active state is due to ATP breakdown and physiological extensibility to the presence of bound ATP. I n the model, ATP is broken down whenever it is present, and the models thus pass from the rigid into the active contracted state as soon as ATP is added, and stay thus as long as any ATP remains. When the ATP in removed they become inactive again; i e . , they lose the ability to contract and to develop tension anew after release; but they do not relax because they become so rigid that they remain “frozen” in the previously attained state of shortening and tension. In living muscle, the presence of ATP is not necessarily accompanied by breakdown. The resting muscle differs from the models in possessing a mechanism for inhibiting ATP breakdown, and is thus able to remain very extensible. When the inhibition is lifted, the muscle passes into the active contracted state for as long as breakdown continues, and if this is so long that the supply of ATP is exhausted, the muscle becomes frozen in the contracted state. I n this manner, a muscle poisoned with iodoacetate (compare Sandow and Brust 1946, Crepax and Herion 1950) or other metabolic poisons passes from a tetanus into a contracture and thence into rigor because ATP resynthesis is prevented. I n physiological contraction, however, only part of the ATP is ever decomposed, so that the muscle remains extensible when breakdown ceases. Whether or not the muscle lengthens again depends on mechanical conditions-the load, the elasticity of the sarcolemma, or connective tissues, etc. The study of the models appears to show that relaxation is not an active process, but is due solely to the fact that the breakdown of ATP

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

189

ceases although ATP is still present. The following observations fit into this general picture: (1) Dubuisson in 1937-9 found that a t the end of the latent period there was a release of acid groups, which, from their amount and dissociation constants, could be considered to arise by the splitting of ATP (see Dubuisson, 1950a). (2) Lundsgaard (1950) observed that muscles which were caused to contract by plunging in liquid air, and thus frozen in the contracted condition, had decomposed an appreciable amount of ATP. (3) In 1949 Hill showed that during relaxation there is no heat production which might be identified with a chemical reaction. Finally, the same author (1949b), using several very sensitive methods, found that a shortened muscle does not lengthen during relaxation if no external force is operating. He assumes therefore that relaxation is not an active process. In this picture of the contraction cycle, the only fundamental difference between muscle and model is the ability of the muscle to limit ATP breakdown to a short period after excitation. i. Rigor Mortis. Following the earlier work of Erdos (1943), BateSmith and Bendall (1947, 1949) showed that after death ATP is a t first broken down quite slowly, until, after a variable period of time, rapid breakdown begins. The onset of rigor always coincides with this rapid phase, and it is accompanied by contracture only when the rapid phase occurs at a pH above 6, as in muscles of low glycogen content. The contracture becomes frozen in when ATP breakdown is complete. With actomyosin threads, too, a change of pH from 6.8 to 6.1 causes a reversible fall of tension, usually to one half, sometimes even to zero (Portzehl, 1950b). The stiffness of rigor mortis is evidently due t o the disappearance of ATP (see Bate-Smith and Bendall, 1947, 1949), and whenever the rapid phase of ATP breakdown (which is the cause of the disappearance of ATP) takes place in conditions (pH > 6) under which actomyosin is capable of contracting a t all, rigor is accompanied by contracture. The conceptions developed here (in Sections 11, 3 and 4) still require proof; some of their consequences can be tested on the models. APPENDIX:THE EFFECT

OF ATP ON THE CONTRACTION CYCLE OF THE

ACTOMYOSIN SYSTEMS

Since this review was written, direct evidence has been obtained that the contraction of the actomyosin systems is dependent on the splitting

190

HANS H. WEBER AND HILDEGARD PORTZEHL

of ATP. Relaxation always occurs when the splitting of ATP ceases, provided that the muscle or the model is prevented from becoming rigid (H. H. Weber, 1951; H. Portzehl, 1952).

Observations 1. All polyorthophosphates prevent actomyosin systems from becoming rigid. Therefore, we call them plasticizers (" Weichmacher "). The plasticizing effect is determined quantitatively by stretching the model to the same extent in the presence and absence of the plasticizer. The ratio of the moduli of elasticity with and without plasticizer is a measure of the magnitude of the effect. Plasticizing effect is indicated by a ratio less than one; the greater the effect, the more nearly the ratio approaches zero. The plasticizing effect of ATP is the greatest; that of sodium triphosphate and sodium pyrophosphate is appreciable smaller. I n order to make this comparison, one must inhibit, or at least greatly diminish, the contractile action of ATP by working a t low temperature. The effects of sodium triphosphate and pyrophosphate are equal at equal molarities. Saturated solutions of benzaldehyde give a considerable but irreversible effect (Table IIIA). TABLEIIIA Plasticizing Effect of Some Polyphosphates and Poisons (from Portzehl, 1961)

Substance

Conc. mol/l.

ATP NaJ'zO7 Nad'tOlo Salyrgan

7.0 x 1.8 X lo-*

Benzaldehyde

Saturated

1 .o

1.3

x x

10-2 10-4

Ap/Ae with plasticizer Ap/Ae without plasticizer

Number of experiments

0.1 0.3 0.4 0.8

3 2 2 3

*0.15

1

Model Fiber Thread Thread Thread Fiber Fiber

2. If the ATP is washed out from a contracted thread- or fiber model with pyrophosphate or triphosphate rather than with a buffer solution, the model does not remain contracted but relaxes. The tension falls to a value between zero and 30% of the initial tension as a result of removing the ATP while the actomyosin remains in a plastic condition due t o the presence of the polyphosphate ions. The reason that the tension does not always fall to zero, as it does during relaxation of the living muscle, is due to the smaller plasticizing effect of the inorganic phosphates compared with ATP. If the model is kept in a solution of polyphosphate before the addition

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

191

LOG [ATP]

FIG.15. ATP concentration optima for tension and breakdown. Upper half: amol. ATP split/g. protein N/min.; Lower half: tension. Curves 2 and 2a at 2”; curve 1 at 20”and l a at 17“ (from A. Weber and H. H. Weber, 1951, Heins and Holton, 1950).

80 -

Washing

TIME, min.

M pyrophosphate FIG. 15A. Contraction cycle in the presence of 1.5 X (thread model). Ordinate, tension; abscissa, time. First contraction with 1.6 X M ATP; second contraction with 4.4 X 10-3 M ATP (from Portzehl, 1952).

of ATP, and if the polyphosphate concentration is held constant during the course of the experiment, contraction occurs upon addition of ATP and relaxation after removal of the ATP. This cycle can be repeated (Fig. 15A). The pyrophosphate is present throughout the entire cycle; thus there is nothing in the experimental evidence to suggest that it provides the energy for an “active relaxation.” Relaxation is the thermo-

192

HANS H. WEBER AND HILDEGARD PORTZEHL

dynamically spontaneous phase of the cycle while contraction occurs only when free energy is supplied to the contractile mechanism by ATP. From these facts, we may conclude that contraction is not an entropy effect. 3. Relaxation occurs in the presence of ATP when its breakdown is inhibited. If the ATP is washed out from a contracted model system and the actomyosin poisoned by Salyrgan (Hg-salicyl-allylamid-sodium acetate), the tension remains nearly the same (Fig. 15B). If now ATP is added again, the system relaxes due to its plasticizing effect (Fig. 15B). Now upon addition of cysteine, the model contracts because the splitting

f

300'

Washing

A'

Cystein

Ib

2b

o;

40

5"o

60

40

FIG.15B. Relaxation of the contracted and Salyrgan-poisoned fiber model upon addition of ATP. Ordinate, tension in g./cm.*; abscissa, time. First contraction with 1.7 X lo-* M ATP. After 16 minutes, removal of ATP; after 25 minutes, addition of 6.6 X M ATP; M Salyrgan; after 28 minutes, addition of 1.7 X after 52 minutes, addition of 6.7 x lo-* M cysteine from Portzehl (1952).

of ATP is restored. During this relaxation the tension falls only t o about 50% of the original value because Salyrgan also poisons the plasticizing effect of ATP t o a certain extent. On addition of Salyrgan to a solution still containing ATP, the tension of the contracted model falls immediately to between zero and 20% of the original value. Apparently the poisoning of the plasticizing effect by Salyrgan requires more time than the poisoning of the ATPase. Thus if ATP and Salyrgan are present at the same time, relaxation is nearly complete before the plasticizing effect has been diminished. The actomyosin-ATPase can be inhibited by blocking the amino groups as well as the SH groups (Kuschinsky and Turba, 1950a, b) ;for instance, with an emulsion of freshly distilled benzaldehyde (compare 11, 4 4 . Benzaldehyde causes an immediate and complete relaxation of the contracted models. However, this observation is not quite conclusive because the relaxation due to benzaldehyde is irreversible. These new observations seem to confirm the conception of the rela-

MUSCLE CONTRACTION

AND FIBROUS MUSCLE PROTEINS

193

tionship between ATP and actomyosin in living muscle which is given in Sections 11, 39, 11, 4h, and 11, 4i.

MYOFIBRIL AND T H E I R REACTIONS The mechanism by which ATP affects the elasticity and activity of 111. THE P R O T E I N S

OF THE

the contractile proteins can be investigated in two ways: i. D a t a can be collected which give information about the molecular and micellar structure of the contractile systems, from myosin gel to living muscle, without breaking down these systems any further. Under this heading are grouped X-ray diffraction, birefringence, light scattering, and fine structure as revealed by the electron microscope; under certain circumstances also the absorption of ultraviolet light in the normal, stretched, and contracted states. ii. The individual proteins which make up the myofibril can be isolated; their chemical and physical properties can be determined, whether in the solid state or in solution; their shapes ascertained and any changes in shape induced by interaction with ATP or other polyphosphates. Neither of these modes of approach has yet led to an explanation of the ATP-actomyosin mechanism, nor is i t likely that an explanation will be found until the two lines of investigation are pursued together. X-ray diffraction and electron micrographs can only be interpreted in terms of the individual components of the contractile system when the behavior of these to each individual method has been ascertained. I n this section, therefore, the individual proteins and their reactions mill be described, and in a later section an attempt will be made to correlate the data with those obtained on muscle itself by means of X-rays, polarized light, electron microscope, etc.

1. Historical and Nomenclature I n 1930, Edsall, and v. Muralt and Edsall, extracted minced muscle with salt solutions of high ionic strength, and purified them. These extracts had all the characteristics ascribable to solutions of fibrous molecules, including the property of flow birefringence. Since the protein of the A band is birefringent, the authors correctly supposed th a t this protein was present in their extracts. Since, however, the flow birefringence was too small and variable to account for the birefringence of the A bands, the authors also supposed, this time incorrectly, th a t their extract contained other proteins. In 1934 Weber showed that the “muscle globulin’’ of v. Muralt and Edsall, when transformed into myosin threads, possessed a birefringence high enough to account for both the intrinsic and form birefringence of the A bands, provided the myosin gel was completely oriented, Edsall’s protein has therefore

194

HANS H. WEBER AND HILDEGARD PORTZEHL

generally been known as “myosin.” By 1939 it seemed entirely probable from studies of the swelling, X-ray diffraction and elastic properties of myosin threads and films, th at the A band of muscle is composed of myosin (Boehm and H. H. Weber, 1932; H. H. Weber, 1934a, 1939; Astbury and Dickinson, 1935, 1940). I n the same year, Engelhardt and Lj ubimova discovered that myosin preparations possess ATPase activity, and t ha t the modulus of elasticity of myosin threads is lowered by ATP. Soon afterwards, J. Needham and coworkers (1941) found th a t A T P diminishes reversibly both the viscosity and flow birefringence of myosin solutions. Then for the first time it was shown by Schramm and Weber (1942) that myosin solutions are polydisperse, containing a slowly sedimenting component with a low birefringence of flow (L-myosin) and several rapidly sedimenting components with high flow birefringence (S-myosins). The connection between these varied results was explained by Szent-Gyorgyi and his pupils in 1942 and in the years which followed. The Szeged school confirmed the results of Needham’s group, and extended them by showing th at ATP reversibly influences also the solubility of myosin and the light scattering of the solution. The most important finding of this group was that A T P exerted these effects only when the solution contained actomyosin (the S-myosin of Schramm and Weber). It has no effect on the so-called “ crystalline” myosin (L-myosin of Schramm and Weber, A-myosin of Szent-Gyiirgyi). The end of this analytical phase came with the discovery by Szent-Gyorgyi’s collaborator, F. B. Straub (1942), th at actomyosin is in reality a complex of two fibrous proteins, actin and L-myosin. The diminution in viscosity, flow birefringence and light scattering, and the increase in solubility of actomyosin nearly to the values corresponding to L-myosin, show th a t ATP dissociates actomyosin into its two components. This conclusion has been questioned only once (Jordan and Oster, 1948) and has been confirmed by many independent methods. Recently, still more fibrous proteins have been isolated from muscle tropomyosin, discovered b y Bailey in 1946 (1946a), and paramyosin, discovered by Bear in 1944. It seems certain that neither of these proteins participates in the contraction process (Bailey 1948; Astbury 1948; Schmitt et al., 1947). From 1945 on, Dubuisson has examined electrophoretically in the Tiselius apparatus extracts of miiscle prepared with extracting media of differing salt content. By this means, he found three myosin components, a, 0 and y. It is almost certain th at &myosin is identical with L-myosin and a-myosin with actomyosin (Table V). y-Myosin seems to be identical with contractin (cf. Section IV, 3). T h e Y-protein may

195

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

TABLE IV T h e I d e n t i t y of Szent-G'yorqyi's

Properties

M y o s i n a n d Actomyosin w i t h the L- a n d S - M y o s i n s of S c h r a m m a n d Weber

Myosin proper of Szent-Gyorgyi

-

Salting-in begins a t ionic 0.04(pH 7.0Ja strength Salting-in complete at 0.3a ionic strength Precipitation aa doubly- Possibleb refracting threads (socalled crystallization) Double refraction of flow Snin11~ Weak. Light scattering 7.2C.d 4 0 0.2b - 0.22'J \ Zn Small, linear o n velocity gradient None On the gel Effeot of None tion

Actomyosin (Szent-Gyorgyi)

S-myosin

0.04(pH 7.0)'

- 0 . 3 ( p H 6.8)n

0.3(pH 6.8)i

0.3(pH 7.0)'

B 0.35(pH6.8)0 2

L-myosin

-

-

0.35(pH 6.8)i

Possible'

Not possible

Not possible

Smalli Weak; 7.1* 0.2' Small. linear

Pronouncede Strong. 2 50d.4m 2 0,34',fJ Strong, nonlinear

Pronouncedi Strong'

None

Shrinking or Shrinking or contraction",' contraetion',P nrel falls to that of L-myosin,n.o and s;,(c = 0 ) to the values obtained for L-myosin and a c t i d '

-

None

0.2b+ >0.34q 0.2 b 0 . 3 4 b s$ of L-myosin -+ s& of actornyosind,i

2 9Oim 0.34' Strong, nonlinear =

-

-

Szent-Gyorgyi (1947). Szent-Gyorgyi (1943). c Snellman and Erdos (1948a). d Johnson and Landolt (1950). ' Mommaerts (1945). 0

6

I The viscosity number 2, is calculated from authors' data by the formula 2, =

2.3 log ~

?re1

(see

C

Section 111, 5a). I Balenovic and Straub (1942). * Portzehl et al. (1950). 1 Schramm and H. H. Weber (1942). Portzehl (1950a). I Snellman and Tenow (1948). * The difference in s:o between actomyosin and S-myosin is due not to differences in the experimental values, but to different extrapolation methods. " Buchtbal et al. (1947). 0 H. H. Weber (1950a). p Portzehl (1951). q Jaisle (1951).

also be a fibrous protein, for it is extractable only in solutions of ionic strength greater th an 0.5, and it appears to have a very low diffusion constant. Its mobility is little less than that of P-myosin, and it is a normal component of unfatigued muscle. Nothing further is known about it. It is certain th at the so-called "crystalline" or "water-soluble " myosin is identical with L-myosin (Table IV). Since the terms "crystal-

196

HANS H . WEBER AND HILDEGARD PORTZEHL

line” and “water-soluble” might give the wrong impression of the properties of myosin, and since the word “myosin” without further qualification is generally used to denote unfractionated mixtures of this protein and actomyosin, it will always be referred to here as L-myosin. TABLEV O n the Apparent Identity of SzentCyorgyi’s Myosin Fractions with Those of Dubuisson (Supplemented f r o m Dubuisson, 1960d) Myosin proper of Seent-Gyorgyi

Myosin p of Dubuisson

+

+

Transparent gel, F B Clouding of gel a t 0.003 M KC1 Dissolves 0.5 M KC1, F B clear solution myosin in Combines with actin of stroma in long extractions Transformed into long extractions Contains no lipid Contains 3 % lipid material Precipitated as regular threads Precipitated as regular threads (Crystals?) Abundant yield in short extractions 75 % yield in short extractions Fully dissolved by 0.25 M KCI, Fully dissolved by 0.30 M KClb p H 7.0d Electrophoretic mobility, ionic strength 0.3, p H At ionic strength 0.4, p H 7.15: 2.5 cm,2/V./sec: 7.14:2.6 cm.2/V./sec.c

In water, transparent gel, FBD Pptd. 0.04 M KCl Dissolves in 0.5 M KC1, no FB; clear solution

+,

(Y

Actomyosin (Szent-Gyorgyi)

Myosin

01

(Dubuisson)

Very viscous solutions, strong persistent F B Similar 90 % yield in long extractions Abundant yield in long extractions Similar Ppt. becomes very insoluble on ageing Fully dissolved in KCl ionic strength 0.35 and p H Fully dissolved in buffered NaCl, ionic strength 0.35 and 6.5-6.7b pH 7.4’ FB = flow birefringence. Szent-Gyorgyi (1947). 0 Erdiis and Snellman (1948). d Dubuisson (1948a). * Dubuisson (1950b). Harnoir (1947). 0

b

’

There is no longer any doubt that actomyosin is identical with the S-myosins ; these actin-containing complexes will always be referred to as actomyosins. General data on these proteins e.g., solubility, viscosity, birefringence, will be entered in the Tables as L-myosin and actomyosin, but footnotes will be added for data obtained on a and myosins; electrophoretic data will be discussed using Dubuisson’s nomenclature, though for the sake of clarity “actomyosin” will be added in parentheses after a-myosin, and “L-myosin ’’ after p-myosin.

MUSCLE CONTRACTION

197

AND FIBROUS MUSCLE PROTEINS

2. Solubility, Colloidal Properties, and Crystallinity

All the known fibrous proteins of muscle are globulins which are insoluble a t the isoelectric point (I.P.) in absence of salt (Table VI). This is true also of L-myosin which is sometimes described as “watersoluble.” I n salt free solutions, Donnan effects are very marked as the pH is moved away from the I.P., and the protein swells and finally dissolves. TABLEVI Isoelectric Precipitation Zone and Isoelectric Point of the Fibrillar Proteins Isoelectric point Protein

Animal

Actomyosinb

Carp

Actomyosinb

Carp

ActornyosincBd

Rabbit

L-m yosine

Rabbit

Actinf

Rabbit

Tropomyosinr

Rabbit

Method

pH

Isoelectric precipitation zone

Salte

Cataphoresis 5.4 Phosphate (0.015) Swelling 5 . 4 Phosphate (0.015) minimum Precipitation 5 . 6 KC1 optimum (0.5) Electro5 . 4 KCl(O.l-0.5) 0.05 M phoresis K veronal acetate Precipitation 4.8 Acetate (0.01M ) optimum Precipitation 5 . 1 NaCl (0.01) optimum

+

PH

-

Salt

8.0-4.5 Phosphate (0.015)

7.5-4.8 KCl(O.15) 7.0-4.0 KCl(0.3) 6.0-2.0 KCl(0.6) 6.5-4.8

0

4.5-6.5

0

* I n parenthesis, ionic strength. * Roth (1946). Hamoir (1947). d Found for u-myosin. 8 Erdos and Snellinan (1948). Straub (1942). 0 Bailey (1948).

When the Donnan potential and the excess osmotic pressure within the gel are sufficiently reduced by additional salt, the swelling and dissolution can be prevented; but salt may also exert a “salting-in” effect which can compensate for the abolition of the electric and osmotic Donnan effects. When the salting-in threshold is very low, therefore, proteins caused to dissolve by shifting the pH away from the isoelectric point are not precipitated by addition of salt. This is true at p H 7 of tropomyosin, a relatively soluble protein for which the salting-in limit is particularly low and the salting-out limit is particularly high (Table VII).

5; 00

TABLEVII Salting-In and Salting-Out Ranges of the Fibrillar Prota'ns Ionic strength for salting-in

Protein

Animal

pH

Actomyosin

Rabbit Rabbit Carp

5.6 6.5-6.7 7.0

Rabbit

7.6

Beginning

Maximal e5ect

Ionic strength for salting-out

Salt

Maximal effect

pH

Beginning

5.4-5.5 6.2-6.4

2.95a.b 2.95a."

3.39'6 3.4a.n

-

-

-

?

4.494.25a."

4.77"' 4.75a.n

2'

-

Salt

Fz m

L-Myosin Actin

Rabbit Rabbit Rabbit

Tropomyosin

Rabbit

Paramyosin

Mussel (adductor muscles)

Data f o r a and fl myosins respectively. Hamoir (1947). 0 Portzehl et al. (1950). Mommserts and Parrish (1951). 0 Roth (1946). f H. H. Weber and Meyer (1933). Kamp (1941). * Seent-Gyorgyi (1947). 1 Straub (1942). i Bailey (1948). k Schmitt et al. (1947). 1 Dubuisson (1948s). * Snellman and Gelotte (1950). Dubuisson (1946s). 0 Straub (1943). a

b

6.54.9 7.0 6.0-7.0 7.0 6.0-7.0(?)

0.5a.b -0.3c.d

-

KCI 2 0.35c,h KCI 0.24e 2 0.3< KCI 0.43n.b < 0.850,') I Acetate = 2 I Phosphate 0.04c.f.o.h .% 0.3' KC1 > 0.25' KCl Fully soluble in water and diKCI lute salt' Fully soluble in water and diNaCl lute salti 2 0.45k 2 0.6' KCI

-

-

6.2-6.4 7.0

-

7.0

5.45i

-

-

-

7.3 -

(NHr)zS01

(NH~)~SOI

-

(NH4)zSOn (NH~)ZSOI KC1

(NH~)PSOI

-

m

3 W

M

m

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

199

When, however, the salting-in limit is appreciably higher than the ionic strength necessary t o abolish Donnan effects, the addition of salt first produces a shrinking and precipitation, followed later b y dissolution. This is the case a t p H 7 with L-myosin, which is less soluble than tropomyosin, the salting-in limit being higher and the salting-out limit lower (Table VII). The decrease and subsequent increase in solubility as the salt concentration is raised are not due to ion binding or to charge effects, as is shown by the electrophoretic control experiments of Erdos and Snellman (1948). Actomyosin is the least soluble of the muscle proteins; the threshold for salting-in is by far the highest, and for salting-out the lowest. Actomyosin gel is not usually soluble in water a t p H 7 (but compare also Szent-Gyorgyi, 1943), and the state of the gel alters little on addition of salt until it begins to dissolve a t an ionic strength of about 0.3. It is not possible to include paramyosin and actin in this series of decreasing solubility (cf. Table VII), for there are no salting-out data for the former and the data on salting-in refer not to purified paramyosin but only to the paramyosin fibrils, which have a limited distribution in the adductor of some mussels (Schmitt et al., 1947). I n the case of actin, the solubility relations are obscured by the reversible transformation of G into F-actin (see Section 111, 5 b ) . I n the solid state, only tropomyosin forms crystals (Bailey, 1948), and it is the only protein of the k-m-e-f group which has yet been crystallized. All the fibrous proteins of muscle do, however, readily form gels which though amorphous macroscopically show some structure in the electron microscope-due to their inherent ability to aggregate side-byside and end-to-end t o form fibers. That a gel can be built out of such fibers has been shown by Perry et al. (1948) for actomyosin and by Erdos and Snellman (1948) for salt-free L-myosin a t p H 7. Since L-myosin in true solution does not give a network of fibers but a homogeneous protein film (Snellman and Erdos, 1948b), i t would seem that L-myosin in water a t pH 7 is not in true solution. The points a t which L-myosin particles begin t o align themselves in parallel fashion enlarge very easily t o give microscopically visible threads (Szent-Gyorgyi, 1943; H. H. Weber, 1947), which when short are called “crystals” by Szent-Gyorgyi (1943). The electron microscope shows, however, that they are really short fibers and not crystals (Rozsa and Staudinger, 1948; Snellman and Erdos, 194810). F-Actin, too, which is insoluble a t p H values of 6 and below (Straub, 1942), is found to consist of fibers between p H 5 and 6 (Jakus and Hall, 1947). Under conditions in which the proteins are macroscopically soluble, aggregation into fibers often occurs as the forerunner of precipitation ;

200

H A NS H. WEBER AND HILDEGARD PORTZEHL

actin, for instance, transforming to the F-form (Straub, 1942) and tropomyosin in salt solutions near pH 7 (Bailey 1948). The process is detectable in solution by a viscosity increase and appearance of flow birefringence; after drying, the particles appear in the E.M. as threads, both single and in networks (Astbury et aE., 1947, 1948; Rozsa et al., 1949). Aggregation does not take place in more strongly alkaline solutions (pH 8 for actin, pH 12 for tropomyosin), and in the case of actin, one assumes that specific factors are also involved (Straub and Feuer, 1950; cf. Section 111, 5b). 3. Electric Charge There is not much information on the charge of the fibrous muscle proteins. Older work on the titration curve, alkali binding and I.P. (Hollwede and H. 11. Weber, 1938; Dubuisson, 1941 ; Dubuisson and Hamoir, 1943) is not as valuable as it might be, since unfractionated myosin was used. It is not certain even whether the I.P. of L-myosin and actomyosin are identical (cf. Table VI). More fac,ts are available concerning the electrophoretic mobility of the main proteins between p H 7 and 7.5 (Table VIII). The mobility of F-actin is exceptionally high and is a little smaller after depolymerization in sodium iodide, i.e., for G-actin. (Indeed G-actin should have a smaller negative charge a t pH 7 than F-actin, because polymerization is accompanied by the release of H ions (Dubuisson, 1 9 5 0 ~ ;Dubuisson and Mathieu, 1950).) That P-myosin and L-myosin are identical is confirmed by a comparison of their mobilities a t p H 7. There is present also a protein of low mobility (y-myosin or contractin) but it is not yet known whether this belongs to the actomyosins or to L-myosin. Only in the case of L-myosin is it known how the mobility changes with ionic composition over a wide p H range; it is not affected perceptibly by Na, K, C1 or phosphate, veronal and acetate for values of ionic strength 0.15-0.55 (cf. Table VIII and Fig. 16). The dependence of mobility on pH is much greater on the acid than on the alkaline side of the I.P., as is true of most proteins. That the electrophoretic mobility is independent of the nature and concentration of the above ions, and depends solely on pH, is almost conclusive proof that the charge on L-myosin particles is determined exclusively by the binding (and release) of H ions. This means that the other ions, including sodium and potassium, are not bound, but are free in solution as “gegenions.” The agreement between the alkali content of salt-free myosin gel and the amount of protons given up on the alkaline side of the I.P. leads to the same conclusion (Hollwede and Weber, 1938) ; and finally, the pH-mobility curve found by Erdos and Snellman (1948)

MUSCLE CONTRACTION AND F I B R O U S MUSCLE P R O T E I N S

201

TABLEVIII Electrophoretic Migration Velocities of the Fibrillar Proteins

Protein

Animal

a-Myosin (actomyosin)

&Myosin

L-Myosin

r-Myosin ("contractin")

F-Actin

G-Actin

Tropomyosin

0

b

I

Rabbita

0.35

Rabbit6.c

0.4

Salt

+ +

0 . 2 M NaCl 0.052 M phosphate 0 . 2 5 M NaCl 0.052 M phosphate

-

pH

Velocity (cm.Z/V./ sec.) ( X 106)

7.4

-3.0

I-hr. extract

7.35

-3 1

1-hi-. extract

Details of extraction procedure

Rabbitd

0.4

7.15

-2.7

Rabbitd

0.4

-

7.15

-2.7

Snaild (foot muscle)

0.4

-

7.1

-2.8

As above, pptd. from 1-hr. extract

Rabbit"

0.36

-2 . 8

I-hr. extract

Rabbitb

0.4

Rabbitd

0.4 0.4

-

7.15

-2.5

Carp'

0.35

._

7.1

-2.9

Snaild (foot muscle)

0.4

-~

7.1

-2.6

Rabbit,

0.15

4.5

$3.85 \

4.98 5.75 7.14

+1.21 -1.51 - 2 . 6 } Once crystallized

Rabbit, Rabbitf Rabbit,

I

Ionic strength

)::00 . 3 0.15. 0.55 0.3

+ +

0.2 M NaCl 7.4 0.052 M phosphate 0 . 2 5 M NaCl 7.35 0,052 M phosphate 7.15

+ +

0 . 1 M KCl 0.05 M K veronal acetate 0 . 5 M KC1 0.05 M K veronal acetate 0.25 M KC1 0.05 M K veronal acetate

+

+

0.25 M NaCl 0.052 M phosphate

Rabbit'

0.4

Rabbitd

0.4

Rabbitd

0.4

Snaild (foot muscle)

0.4

Rabbit9 Rabbit,

0.4 0.15

0 . 2 5 M NaCl 0,054 M phosphate

Rabbit9

0.4

0.25 M NaCl 0.054 M phosphate

Rabbit*

0.4

Rabbit9 Rabbit'

-

-

-2.9

1-hr. extract

-2.5

Whole myosin pptd. from 1-hr. extract As above, pptd. from 24-hr. extract Isolated &myosin, 10min. extract Whole myosin pptd. from I-hr. extract

I

-2.32 -2.9

7.35

-2.25

7.15

Whole myosin pptd. from 1-hr. extract -2.1 As above, pptd. from 24-hr. extract - 1 . 9 5 As above, pptd. from 1-hr. extract

7.1

+

7.4

-6.3 -9.3

7.4

-4.6

-

7,4

-4.55

0.15

-

7.6

-6.4

0.4

-

7.4

-5.6'

(1950).

1-hr. extract

-2.1

7.6

+

pptd.

7.5 7.8

7.15

-

Dubuisson (1948b) Dubuisson 11950bl

Dubuisson (1950e).

. .-

Whole myosin

from 1-hr. extract A s above, pptd. from 24-hr. extract

KCl activated extract from acetone powder

{

F-actin depolymerized by K I Myosin first extd. (3x with buffered KC1) and residue extd. with 0.6 M K I F-actin depolymerized by K I Solution according to Bailey

202

HANS H . WEBER AND HILDEGARD PORTZEHL

1.,

8

-4'

I

1

3

5

I

7

I

9

I

1

I 1

P"

FIG. 16. Electrophoretic mobility and Hf binding of L-myosin. p H titration curve from Dubuisson, 1941; mobility from Dubqisson (194613, 194813, and 1950b) (cf. Table VIII); 0 mobility from Erdos and Snellman (1948).

+

-

,

0 PH

FIG. 17. Electrophoretic mobility-pH curve of L-myosin in varying concentrations of calcium and magnesium chlorides (0.03 - 0.24 M ) . 0 < 0.1 ICI salt; > 0.1 M salt (after Erdos and Snellman, 1948).

MUSCLE CONTRACTION

AND FIBROUS MUSCLE PROTXINS

203

agrees closely with that expected from the H bound as determined by Dubuisson (1941) (see Fig. 16). The alkaline earth metals, calcium and magnesium, behave in a special way (Fig. 17). They raise the mobility (towards the cathode) by about cm.2 X volt-' X set.-' a t pH 3 and by about 4.3 X loF6cm.2a t 2X pH 7.5, so that a t the latter pH the mobility is changed from one of 2.9 X towards the anode to one of 1.4 X loF5 in the reverse direction. Calcium and magnesium ions are tightly bound by L-myosin,s the amount being less on the acid side of the I.P. than on the alkaline, though still unusually high a t a pH as low as 2.5, and the I.P. is shifted from 5.4 t o over 9.0 (Fig. 17). Calcium and magnesium, moreover, are the two ions which profoundly influence the ATPase activity of L-myosin and its actin complexes (see Section 111, 5d). Magnesium also increases the contraction of the actomyosin and fiber models (see Section 11, 3a). 4. Particle Xize and Shape a. L-Myosin. The data on the size and shape of L-myosin particles are relatively numerous and well confirmed. Portzehl, Schramm and H. H. Weber (1950) investigated in the ultracentrifuge twenty-five preparations of L-myosin over a period of five years, and found with all preparations that the sedimentation constant was a linear function of the protein concentration over a range of 0.03-0.95%, the slope of the curve being small. Extrapolation to zero concentration gives s&, 7.1 (cf. Fig. 23). Snellman and Erdos (1948a) found sio to be 7.2, though this was derived from only one curve of six points, and Johnson and Landolt (1950) have reported the same value in a preliminary communication. The value of 6.7 obtained by Mommaerts and Parrish (1951) for some unexplained reason deviates more considerably from the values obtained by other investigators. When L-myosin is denatured a second peak appears, whose area increases with increasing denaturation while that of the original peak decreases (H. H. Weber, 1950a and Fig. 18). Since the sedimentation rate of the original peak remains unaltered, the denaturation process does not involve a continuous change of structure but passes through a series of sharply defined stages, of which several are sometimes present in the same solution (Fig. 23, curve l a and Section 111, 6). The denaturation products have always a higher sedimentation constant than 8 Since calcium and not potassium ions appear to be bound by L-myosin, it is difficult to understand why excess of the latter should decrease the effect of the former on the charge on the alkaline side of the I.P. It seems desirable therefore that these results of Erdos and Snellman should be systematically reinvestigated on a wider basis.

204

HANS H. WEBER AND HILDEGARD PORTZEHL

L-myosin itself, and that with the value s ; ~ about 15 appears to be especially stable (cf. Fig. 23, curve la). The diffusion constant of L-myosin is difficult to measure, because this determination, unlike that of sedimentation constant, is spoiled by the denaturation products. These arise very easily in the purification of L-myosin (cf. Section 111, 6b), and, most important, spontaneously, even in completely homogeneous preparations kept at O”, from about

” ’

90 80

2oL 10

5.9

DISTANCE FROM AXIS,

5

cm.

FIG. 18. Denaturation of L-myosin. The peak on the left in each case is Lmyosin (& = 7.1). The right-hand peaks (b) and (c) are denatured L-myosin (s,”, = 12 in (b) and > 15 in (c)). (a) is once “crystallized” 4 days post mortem; (b) is twice “crystallized” 8 days post mortem; (c) is twice “crystallized” and once reprecipitated, 9 days post mortem (from Portzehl et al., 1950).

the 10th to the 14th day onward (Portzehl et al., 1950); a t 20°, they appear in only a few days. The diffusion must therefore be measured in a cell which allows completion of the determination in 6-8 hours. The homogeneity of the preparation at the end of the run can be tested by sedimentation, and the diffusion gradient curve is itself a further check. For a strictly monodisperse preparation, this should be a symmetrical binomial curve (Fig. 19), which can be tested by plotting the square of the width against the log of the height a t which the width is measured (Fig. 20). Such curves have so far only been obtained by Portzehl (1950a) (cf. also H. H. Weber, 1950b), using the diffusion cell of Bergold (1946) (cf. Figs. 19 and 2O).9 and they show that L-myosin 9 The experiments of Snellman and Erdos (194%) lasted 5-7 days a t 20”. The curves of the diffusion gradient are therefore quite unsymmetrical, and the values of diffusion constant and particle weight so derived are not entered in Table IX.

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

205

diffuses as a strictly monodisperse substance. A change in protein concentration from 0.7 to 0.04% increases the diffusion constant from 0.55 to 0.84 X lo-', the extrapolated value at zero concentration being 0.87($_0.03)X

-0.1

-0.2

0

0.1

0.2

x , cm.

FIG. 19. Diffusion gradient curves of L-myosin. Curves 1, 4, 7, and 11 after 150, 330, 450 and 780 minutes respectively. X I etc. mark the points of inflection (from Portaehl, 1950a). I

-

H

FIG. 20.

4

I .o

1.4 LOG H

1.8

Monodisperse diffusion of L-myosin. B = width of gradient curves, B2 on the left of the symmetry axis, 0 on the right (from Portsehl;

= height; x = 1950a).

The sedimentation diagrams of Fig. 21 are also indicative of a monodisperse substance. The curve with open circles represents the theoretical spreading of the peak between x1 and X I O due to diffusion, as calculated from the diffusion constant at that particular concentration (0.64 X 1 0 ~

206

HANS H. WEBER AND HILDEGARD PORTZEHL

for c = 0.25%) (according to the methods of Bergold and Schramm, 1947). The right side coincides with the experimental curve, whereas the left side lies further away from the symmetry axis. The experimental curve is thus not broader, but actually narrower than the calculated. The effect is due to the dependence of sedimentation on concentration; the particles which are left behind by diffusion sediment with an s20 value of 7. Those at the point of maximal concentration

I

FIG. 21. Monodisperse sedimentation of L-myosin. Sedimentation gradients observed, unbroken line; z l 0calculated from z1 and Dso, broken line (from Porteehl, 1950a).

sediment at a value of 6.2. This represents a sedimentation difference of 14%, and the curve shows in fact that the foot of the left-hand side has travelled about 10% faster than that of the calculated (Fig. 21). The narrowing of the experimental curve can thus be accounted for in a quantitative manner within the experimental error. The strict monodispersity shown by L-myosin both in diffusion and sedimentation shows that the protein particles are identical both in size and in shape. The particle weight calculated from D and X is 858,000 f 30,000. The axial ratio from the frictional coefficient is 98 k 4, neglecting hydration. This strict monodispersity makes it possible to check the data osmotically (Portzehl, 1950a). Again using a method which avoids denaturation (H. €1. Weber and Portzehl, 1949) the plot of PIC against C leads to a particle weight value of 840,000(f33,OOO). The axial ratio calculated from the slope of the PIC curve according to the method of Schulz (1947) is 128.

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

207

The values of particle weight from osmotic and sedimentatioii agree very well, those of axial ratio fairly well (Table IX). Taking the osmotic value as the more correct (diffusion measurements being less accurate), and the axial ratio from the frictional constant, the myosin particle appears t o be a rod 22-24 A. thick and 2200-2400 A. long. Whilst the particle weight can be taken as definitely established, the length of the axes is less certain since hydration has not been allowed for. As the Schula formula is influenced in a similar way by hydration the good agreement between the osmotic and sediment,ation result is no evidence for the correctness of the value. The viscosity of all myosin preparations varies with concentration according t o the Arrhenius formula log vrel = K * c (Edsall, 1930; H. H. Weber, 1947; cf. also Fig. 24). The viscosity numberlo 2, is therefore

Since the viscosity of L-myosin varies very little with velocity gradient (Mommaerts, 1945; see also Fig. 25), the values obtained b y different authors at quite different gradients agree well. At gradients greater than 1000, Z, is about 0.22. The apparent absorption coefficient due to light scattering, determined on a 1 % solution of L-myosin is given as 0.05 cm.-' (Portzehl et al., 1950). It seems to increase, as the solution is diluted. For the angular dissymmetry of the light scattering, Mommaerts (1951b) finds a value of 1.57 for blue and 1.42 for green light, from which by Oster's method (Oster et al., 1947), the length of the particle is 1500 A. (Table I X). No new X-ray data have become available since the summary in this journal by Bailey in 1944. The X-ray examination of actin (Astbury et al., 1947) has shown indirectly th at the wide angle diagram of myosin (actomyosin-L-myosin mixture) is that of the L-myosin component. The relation between the X-ray data and amino acid analysis was also thoroughly discussed by Bailey (1944) in the same review. Few modifications in the amino-acid composition have been made since th a t time, and the present position (cf. Bailey, 1948) is given in Table X. L-Myosin has a phosphorus-containing prosthetic group. The total P content of 0.0440.07% (Bate-Smith, 1938; Bailey, 1942; Lajtha, 1948) has been further subdivided by Buchthal et al. (1948, 1949) :I1 ortho10 2, is calculated as grams per liter and is thus ten times greater than the intrinsic viscosity [TI. 11 The values of Buchthal et al. are referred not to the weight of myosin but to the weight of wet L-myosin threads. It is assumed that the threads contained the same amount of total P/g. of L-myosin as those of other authors-and this would be the case if the threads contained 1 % of protein, an entirely probable value.

f.3

TABLEIX

0

00

Physical Constants Relating to the Particle Weight and Shape of the Fibrillar Proteins Particle weight 6'20

Protein Natural actomyosin Artificial actomyosin L-Myosin

( X 1013)

> 90-

>> 2800

>> 280s 7.2b.C 7 .l a

DQio

G-Actin

Tropomymin dimer (I = 0.27) Tropomyosin monomer (urea and acid)

Value

-

-

-

-

-

-

-

-

-

0.87'1

858,000'1

Method

8 and D 840,000'1 Osmotic

-

> 650 3.44 > 4-5.

2.40

(c = 0 . 6 % )

-

-

* Uncorrected for hydration.

** Corrected for hydration. Porteehl d a!. (1950). * Snellman and Erdiia (194%).

Johnsohn and Landolt (1950). Mommaerts and Parrish (1951). Mommaerts (19518). f Snellman and Gelotte (1950). *Bailey d d.(1948). * Porteehl (1950a). ' Straub, quoted by Seent-GyBrgyi (1947). j Tsao et d. (1951). k H. H. Weber (1950b). 1 Mommaerts (1945a). m Mommaerta (1951b). Sakus and Hall (1947). 0 Rozsa d aE. (1949). P Ardenne and Weber (1941). c

d

98'1 100b.A 128k

-

> 50.4

2.60

-1P

( X 10')

6.76 F-Actin

Axial ratio*

-

-

70,OOOi Minimal, from tryptophan content 93,0000 s a n d D 88,0000 Osmotic 53, OOOi Osmotic

-

Derivation

Dimensions (A,) Length 12.000' 0 . 5 to > 5 p

and D 1,500"' and osmotic 2,200 -+ 2,400'1 mol wt. Osmotic mol

Thickness

8

8

wt.

-

1 to >5p*.*

-

-

Method

50+ 2 5 0 " ~Light scattering 5 0 - 250 E.M. 22 + 24h

-

1

POlY

Light scattering s, D, and osmotic mol w t . }Homo

10Ov E.M.

-

Dispersity

Poly

-

M

1

Homio

560

-

sand D

26**.i Viscometric

385i

-

15i

Axial ratio, mol diagram

2m

] Homio

0

*

Z

U

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

209

TABLEX Analysis of Rabbit Myosin and Tropomyosin (after Bailey, 1948) (Results calculated on N contents of 16.7 %)

Tropomyosin (Residues/100 g.)

Myosin (Residues/100 g.)

0.0063 0.0188 0.0172 0.0988 0.0267 0.1190 0.0279 0.0113 0.0417 0.0244 0.0055 0.0448 0.1074 0.2236 0.0684 0.0636 0.8418

0.0117 0.0228 0.0188 0.0039 0.0253 0.0730 0.0221 0.1190 0.0262 0.0167 0.0412 0.0429 0.0155 0.0423 0.0814 0.1503 0.0669 0.0857 0.7800

115.6}Mean 116.4 117.2 859

115.8}Mean 115.5 115.3 866

26.6}Total 18.4 45.0

16.1 18.0*}Total 34.1

35.2

35.7 57.2 11.9 9.9

Cystine/2 Methionine Tyrosine Tryptophan Glycine Alanine Valine Leucine Phenylalanine Proline Serine Threonine Histidine Arginine Lysine Glu tamic acid Aspartic acid Amide N Totals Average residue wt. : (1) From N partition (2) From wt. of residues Residues/lOb g. protein As % of total residues: Free-acid groups Base groups Nonpolar groups Polar groups Hydroxyl groups Amide groups

62.8 9.7 7.4

*In the calculation for myosin the free-acid groups found by Dubuisson and Hamoir (1943) from titration data (0.156 residues/100 9.) have been used.

phosphate, 30 %; pyrophosphate, 30 %; and organic phosphate, difficultly hydrolyzable, 40 %. The same authors find that the P/adenine/ribose ratio on a mole basis is 1.65/1/1. Bailey (1946) finds that part of the phosphous at least is present as nucleic acid. Summarizing, it can be said that particles of L-myosin are homogeneous in size and shape, and also in electrical charge if L-myosin is considered to be identical with P-myosin (Erdos and Snellman, 1948; Dubuisson, 1946a) (see, however, Mommaerts and Parrish, 1951). It

2 10

HANS H. WEBER AND HILDEGARD PORTZEHL

appears also t o be chemically homogeneous, since Bailey (1951) finds tha t L-myosin, like tropomyosin, is composed of cyclic peptides chains, and only traces of open chain polypeptide are detectable. 6 . Tropomyosin. L-Myosin and tropomyosin have no terminal amino groups, are similar in amino-acid composition (Bailey, 1948, 1951 ; Table X) and they both belong to the k-m-e-f-group in Astbury’s X-ray classification (see Astbury et al., 1948). I n strong urea solutions, L-myosin is depolymerized (H. H. Weber and Stover, 1933; Greenstein and Edsall, 1940), and this is an irreversible process (Ardenne and H. H. Weber, 1941; Snellman and Erdos, 1948a; Szent-Gyorgyi and coworkers, 1944). The depolymerized protein by osmotic pressure measurements has a particle weight of about lo6 (Weber and Stover, 1933). The true elementary units of tropomyosin are much smaller, but they polymerize as the ionic strength diminishes to particles of this order of magnitude (see below). I n water, this’polymerization becomes very marked, the solutions possess birefringence of flow, and the electron microscope reveals an aggregation into fibrils (Astbury et al., 1948). These aggregates are also depolymerized in urea, and the protein is somewhat modified by this treatment, for it can no longer be crystallized (Tsao et al., 1951). The depolymerization, however, unlike th a t of L-myosin, is reversible ; and while L-myosin in urea is polydisperse (Snellman and ErdBs, 1948a) tropomyosin is probably monodisperse (see below). L-Myosin is thus not merely a polymer of tropomyosin, but the latter could conceivably be a subunit (cf. Section 111, 6a) or a physiological precursor of L-myosin (Bailey, 1948). The molecular data on tropomyosin, too, are known quite accurately. At a n ionic strength of 0.27 (pH 6.5) the values of ~ ~ ~ ( 2 . and 6 5 )D20(2.4 X cm.2 sec.-l) do not vary very much with concentration between c = 1.2 and 0.6%; the corresponding particle weight is 90,000 and the same value was obtained osmotically (Bailey et al., 1948). The axial ratio b y Svederg’s method is 56, whereas that calculated from the O.P. measurements b y the method of Schulz (1947) is 111. I n concentrated urea solution and a t p H values below 2.8 the particle weight determined osmotically falls to 53,000, three times the minimal value from the histidine content. Since the effect of acids is fully reversible-the tropomyosin crystallizes as well afterwards as beforethis is evidently the weight of the true monomer. The particle weight at I = 0.27 and p H 6.5 thus corresponds to an average degree of polymerization of about 2. By means of very careful viscosity measurements and taking into account the velocity gradient, the relaxation time and the hydration, Tsao et al. (1951) calculate an axial ratio of 25 for the hydrated mono-

MUSCLE CONTRACTION A N D FIBROUS MUSCLIC PROTEINS

21 1

mer.12 Since tropomyosin gives an X-ray diagram of a-type, the axial ratio which can be calculated, assuming three amino acid residues in the fold, is consistent only with a model containing two a-chains side-by-side. c. Actin. I n salt-free solutions a t pH values above 6, actin exists in a globular form called G-actin (Straub, 1943a), but below p H 6 in absence of salt, or below p H 8 in presence of salt, polymerizes into the fibrous form, termed F-actin (Straub, 1943a; Jakus and Hall, 1947). G-Actin a t p H about 7 sediments with a somewhat diffuse boundary, and the sedimentation velocity a t a concentration of 0.2% varies from one preparation to another. The values 3.2 (Snellman et al., 1949)) 3.7 (Portzehl et al., 1950)) give a particle weight of the order of 70,000, which is said t o be the minimum molecular weight from the tryptophan content (Straub, quoted in Szent-Gyorgyi, 1947). G-Actin cannot be resolved in the electron microscope (,Jakus arid Hall, 1947) except when it aggregates during the drying into spherical clumps (Astbury et al., 1947; Snellman and Gelotte, 1950). F-Actin sediments a t p H 7 with a velocity (50-658) which is fairly reproducible in solutions of varying ionic strength (Portzehl et al., 1950; Johnson and Landolt, 1950; Mommaerts, 195la). The electron microscope reveals long threads about 100 A (80-140) thick and of varying length (I->> 5 p)(Jakus and Hall, 1947; Rozsa et al., 1949). The relative sharpness and reproducibility of the sedimentation are evidently due to the uniform thickness of the primary threads, for the sedimentation velocity does not vary appreciably with length and depends only on thickness. l 3 The great and variable length of the fibers accounts for the high and variable viscosity; 2, =

(I.cGE)

c-+ 0

varies from preparation to

la The axial ratio calculated from the O.P. data by the Schulz method is 136. This value is undoubtedly too high, even for a single alpha-folded peptide chain of mol wt. 53,000. On the other hand, the viscometric behavior does not quite follow the theory of Simha (1940). In the case of L-myosin (Table IX) and serum globulin (Schulz, 1947), the axial ratios agree within l0-25% with those obtained by the Svedberg method. It may be advisable t o check the asymmetry of the tropomyosin monomer by the latter method. 13 According to Svedberg, s = (1 - V p ) M / f = 0 . 2 0 M / f , wheref, the frictional constant, is a function of the surface area. For very long thin particles, the cross sectional area of the two ends is negligible compared with the rest, and when such particles polymerize end t o end, both the particle weight and the cylindrical (or prolate) surface increase n-fold:

-

s2

=

0.20Mz/f2 = 0.2QnMJnfl

=

0.20Ml/fl =

SI

If, however, they aggregate side by side, the area a t the sides increases by only a factor of &: s2 =

0.20M2/f2 = O . Z O n l l f I / &

f1

=

0.20

z/n M l / f l

=

SI

212

H A N S €1. WEBER AND IIILDEGARD PORTZEHL

preparation between 0.19 and 0.34 (Jaisle, 1951). Because of their enormous length, the fibrous particles of F-actin exhibit flow birefringence a t very low shear rates (Straub, 1943a). At a sufficiently high magnification in the electron microscope, they show spacings of about 300 A. along the fiber axis (Hozsa et aZ., 19-19). The meridional periodicity in the X-ray diffraction pattern is at least 5-1 A. more, probably 108 A., which is one-third that found in the electron microscope. There are present

FIG.22. X-Ray f i l m photograph given tiy F-actin film, photographed with the beam pnrallel to the surface of t h r film. CuIC, radiation; collimator 50 X 0.25 mm; original film-to-specimen distance 4 cm. (from Astbury et al., 1947).

also small angle reflections which have not been resolved in the large angle photographs available until now (Astbury el al., 1947; see Fig. 22). A more detailed interpretation must await small angle diffraction studies. In the polymerization of G-actin, the particles evidently come together to form groups 300 A. in length and 100 A. in width, and these in turn arrange themselves in rows to give long fibers. When a G-actin film is polymerized hy KC1 in situ, it is observed that the threads of F-actin do not form independently of each other, but align in parallel fashion to give a cross striated structure of periodicity 300 A. It thus seems th a t the secondary uriitsin the polymerization process, i.e., those 300 x 100 A.,

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

213

tend to aggregate regularly not only end to end, but also to a certain extent side by side (Rozsa et al., 1949). This fact is of interest in connection with the electron microscope picture of muscle fibrils (cf. Section IV). d. Actomyosin. Sedimentation studies point to the existence of several natural actomyosins, each sedimenting with a well defined but different velocity (Fig. 23) ; often these components occur side-by-side in

0

0.2

0.4

0.6

0.8

PROTEIN CONCENTRATION, '1'0

1.0

.,

FIG. 23. Sedimentation curves of L-myosin and actomyosin. Points: 0 , pure, homogeneous L-myosin; V, denatured, homogeneous L-myosin; A, components from mixtures of pure and denatured L-myosin; 0 4 homogeneous natural natural actomyosins with two components; A, actomyosin from actomyosins; actin and L-myosin. Curve 1: L-myosin; curve la, denatured L-myosin of s&, = 15; curves 2, 3, and 4:actomyosin. The broken curve 2 is extrapolated by means of the 1 1 formula - = 8 . 5 ~(from Portzehl et aE., 1950).

+,

820

+

co+

the same solution (Portzehl et al., 1950). For monodisperse preparations, the sedimentation velocity varies according t o the formula

where c = concentration and K a constant which, even at quite different sedimentation velocities, has always been found t o be about 8. If the extrapolation to zero concentration is done by means of this formula, sio is found to vary from rather more than ninety to several hundred Svedberg unitsI4 (cf. Fig. 23; H. H. Weber, 1950a and b ; Portzehl et al., 1950). According t o Johnson and Landolt (1950) siois > 60.

2 14

HANS H. WEBER AND HILDEGARD PORTZEHL

It has not been possible thus far to obtain good diffusion-concentration curves (Bergold el al., 1945). This may be due to the large variation in length of the long actomyosin particles, a variation which in fact is confirmed by electron microscope studies: the length of the threads varies from a few thousand angstroms to several micra, and the thickness from 50 to 250 A. (mostly from 120 to 150 A.) (Hall et aZ., 1946; Ardenne and Weber, 1941). Natural actomyosin particles are so long that here too sio depends only on the thickness and not on the length. The .sioof an F-actin particle is about the same as that of the slowest sedimenting actomyosin, confirming the electron microscope result that the statistical average in thickness of natural actomyosin particles is no higher, or not appreciably higher, than that of F-actin particles. The sio of natural actomyosin confirms that the fibers seen in the electron microscope are not artefacts due to aggregation during drying, for the value is at least ten times greater than that of L-myosin. This means that the diameter must be ten times greater and the cross sectional area 100 times greater; and since the diameter of L-myosin is about 20 A., that of natural actomyosin would be about 200 A. The thickness of the fibers is important in considering the fine structure of the actomyosin complex. It is striking that the scatter in the thickness of the filaments as seen in the electron miscroscope (50-250 A.) does not show itself in the sedimentation velocity, and it seems probable that the particles do not sediment independently of each other, even in very dilute solution, because of their huge length, so that the solution appears to be more homodisperse than it really is.16 The comparison of electron optical and sedimentation studies makes it probable that for F-actin and actomyosin (i) the fibers seen in the electron miscroscope are not artefacts, and (ii) that the resolving power of the latter is superior to that of the ultracentrifuge in this case. “Artificial” actomyosins made from actin and L-myosin appear to sediment faster than the natural ones (Fig. 23; H. H. Weber, 1947; Snellman and Erdos, 1949), and in the electron microscope the fibers never appear so fine (Jakus and Hall, 1947; Perry et al., 1948; Snellman and Erdijs, 1948b)-even when the same authors have made the comparison (Jakus and Hall, 1947). The very variable length and the 14By means of straight line extrapolation to zero concentration over a region devoid of points, i t is possible, from s values a t high protein concentration, to obtain sio = 50 (Snellman and Tenow, 1948). The procedure mentioned in the text seems better justified (Portzehl et al., 1950). 1 6 Signer and Gross (1934) have analyzed such apparent monodispersity by mixing different homogeneous linear polymers.

MUSCLE CONTRACTION AND F::BROUS MUSCLE PROTEINS

215

tendency to anastomose (as shown by the electron microscope) are reflected in the very high and variable viscosity values, and a t high concentration in a structural viscosit;y; this is true both for natural (Portzehl et al., 1950) and artificial (Jaisle, 1951) actomyosins. The concentration dependqnce (Fig. 24) follows accurately the Arrhenius relation log qrel = Kc. 2, is then 2.3 log qrel/c.

PROTEIN CONCE:NTRATION, g J I ,

FIG. 24. Variation of viscosity with concentration of actomyosin and L-myosin. Actomyosin: from Jaisle (1951); 0 , from Balenovic and Straub (1942); A , from Mommaerts (1945). L-myosin: 0 , from Guba and Straub (1943); A, from Mommaerts (1945).

+,

VELOCITY GRPDIENT,

FIG. 25. Dependence of viscosity number on velocity gradient. Open circles: homogeneous L-myosin; others: actomyosins. The figures 2 and 4 refer to the same preparations as in Fig. 23 (from Portzehl et al., 1950).

The viscosity c (per liter) and 2, depend much more upon the velocity gradient than is the case with L-myosin (Fig. 25; Mommaerts, 1945). For a gradient of 2,000 the 2, values for natural actomyosin lie between 0.3 and 0.5 (Portzehl et al., 1950), while artificial actomyosins, containing optimal amounts of the components, give values from 0.45 t o 1.0 (Jaisle, 1951). There is no proportionality between s&, and 2,; a preparation with a high sio may have a low 2, and vice versa. Thus the viscosity too gives only qualitative information about particle size and shape. e

216

HANS H. WEBER AND HILDEGARD PORTZEHL

The flow birefringence of actomyosin is appreciably higher than that of L-myosin (Szent-Gyorgyi and coworkers, 1942; Schramm and Weber, 1942), but has not yet been measured under quantitatively defined conditions; greater too are the turbidity due to light scattering (Portzehl et al., 1950) and the “angular dissymmetry’’ (Jordan and Oster, 1948; Mommaerts and Parrish, 1951). The turbidity of a 1% solution is about 0.5 cm.-l X-Ray studies have introduced no new concepts since the discussion by Bailey (1944). Chemical changes take place when actomyosin films are stretched. Schauenstein and Treiber (1950) and Burgermeister and Schauenstein (1949) attribute changes in ultraviolet absorption and electrical conductivity to enolization and formation of H bonds between neighboring peptide chains, forming an energy-conducting system (see Wirtz, 1947). Even a t extremely low concentrations, actomyosin in combination with fluorescent dyes exhibits phosphorescence, which decreases when the actomyosin is dissociated by ATP (Szent-Gyorgyi, 1947). This phenomenon can be explained in several ways.

5. Reactions of the Fibrous Proteins

a. The Interaction of L-Myosin and Actin. When solutions of F-actin and myosin are mixed the viscosity (Straub, 1942, 1943; Bailey and Perry, 1947; H. H. Weber, 1950a and b; Jaisle, 1951) and sedimentation constant (H. H. Weber, 1947; Snellman and Erdos, 1949; Portzehl et al., 1950; Johnson and Landolt, 1950) rise t o values which are usually higher than those of the natural actomyosins (cf. 111, 4 4 , and the amount by which they increase depends upon the relative proportion of the reactants. For a total protein concentration Z 0.2% ’ the maximum apparent viscosity is reached with three parts of L-myosin to two parts of F-actinI6 (Straub, 1942; Jaisle, 1951). Such mixtures show, over and above the true viscosity, a structural viscosity which is greater the higher the protein concentration. This shifts the maximum in the apparent viscosity towards a higher proportion of F-actin, i.e., to an L-myosin/ actin ratio of 2:3 (Jaisle, 1951). For protein concentrations below 0.2%, the true and apparent viscosity seem t o be the same, though this is not 16 Mommaerts (1951a) asserts that less than 50 % of the protein in actin solutions prepared by Straub’s method poIymerizes to F-actin on addition of salt. Correcting for this, the L-myosin-actin ratio for maximum viscosity would be 3: 1. The purity of actin preparations can also be tested, however, by making use of the ability to combine with myosin, the impurities remaining in solution when the actomyosin complex is precipitated (A. Weber, 1949). By this test, our actin preparations are 60-90 % pure, and impurities were allowed for in the rFtios reported by Jaisle.

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

217

quite ~ e r t a i n . ' ~Whether the maxima must be considered false is important for two reasons: (i) the ratio of L-myosin and actin in muscle is nearer 3 : l than 3 : 2 (Hasselbach and Schneider, 1951; cf. also Section IV) and (ii), Snellman and Gelotte (1950) state that in the ultracentrifuge the actomyosin peaks, unaccompanied by others, appear only when the ratio is 3 : 1; in other mixtures, the peaks of L-myosin and F-actin also appear. Johnson and Landolt (1950) report similar though less precise findings. These facts indicate that, from the chemical point of view, there is only a single actomyosin complex of composition three parts of myosin to one of actin and that mixtures in any other proportion contain one component in excess. If this is so, the maximum in true viscosity should be at the 3 : 1 ratio, a point which should be tested in the Couette viscometer. Since artificial actomyosin solutions contain several very rapidly sedimenting fractions (H. H. Weber, 1947; Snellman and Gelotte, 1950; Johnson and Landolt, 1950), and since their viscosities are very variable (Jaisle, 1951;see Section III,4d), it would have to be assumed that the same 3 : 1 complex forms threads of very variable length and thickness. This may be so, but the problem requires further elucidation. It is possible that actomyosins with different physical properties represent sharply defined stages in the interaction of the two components. When actin and myosin have once combined to give actomyosin, it is not possible by any known method to separate them completely on a preparative scale. There is no doubt, however, that natural actomyosin is really a complex of actin and myosin, (a) because Straub (1942) obtained in small yield from actomyosin the same actin as obtained from the dry acetone powder of muscle, and (b) natural and artificial actomyosins react with ATP in the same typical manner (Section 111,5 4 . It can therefore be concluded that complex formation is thermodynamically irreversible, for by repeated fractional precipitation a preparation can be obtained from muscle extracts in which no free L-myosin can be detected by methods at present available. The ultracentrifugal peak of L-myosin reappears, however, when the actomyosin in solution by its history and its properties, e.g., disappearance of ATP-sensitivity, may be regarded as denatured (Portzehl el' al., 1950; see also Johnson and Landolt, 1950). The dissociation of the actomyosin lcomplex is evidently possible only when the mutual affinity of the two components is decreased, either spontaneously or by chemical agents {such as ATP). The decrease is 17 The constancy of the viscosity in repeated viscometer runs could also be due to a balanced process between the destruction of structural viscosity during movement and a building up in the resting parts of the liquid.

218

HANS H. WEBER AND HILDEGARD PORTZEHL

reflected in the lower viscosity and in lower ATP sensitivity (Jaisle, 1951), and takes place quite quickly when the ionic strength is high (> 1) or the reaction fairly alkaline (pH > 7.5) (Guba, 1943; Portzehl et al., 1950). It is partially reversed when by dialysis the ionic strength and pH are lowered ( I = 0.6 and pH 7), but even a t this ionic strength the affinity grows less, though more slowly and apparently irreversibly. The decrease in affinity is more rapid when the L-myosin and F-actin solutions are stored separately than when they are combined (Spicer, 1949; Jaisle, 1951). The affinity of L-myosin and actin can be destroyed by a large number of chemical reagents. Most of these have been given in Table 111, and only copper glycinate and oxarsan need be added here (Turba et al., 1950). All these reagents either oxidize or substitute sulfhydryl groups. These groups appear to be necessary for the interaction, and are more sensitive to chemical attack when the components are separate than when they are combined. Treatment of L-myosin with oxarsan inhibits subsequent complex formation, though the reagent does not split preformed actomyosin. Many of the reagents in Table I11 may be expected to behave similarly, Salyrgan and copper glycinate, however, not only inhibit complex formation, but also split the actomyosin complex once it is formed (Turba et al., 1950; Kuschinsky and Turba, 1950b). As Bailey and Perry (1947) have shown, reagents which destroy the affinity of 1,-myosin for actin also destroy the ATPase activity of the preparations, and in the spontaneous changes which occur on storage, the ATPase activity and the ability to form actomyosin decrease in a strictly parallel fashion, while the number of oxidized SH groups increases correspondingly. Spontaneous changes are thus traceable to destruction of SH groups, and this process is relevant only in the myosin partner, for the destruction of SR groups in the actin component does not affect complex formation (Bailey and Perry, 1947; Kuschinski and Turba, 1950b). It is thus the SH groups of L-myosin which interact with somc unknown group or groups in F-actin. The effect of a series of organic and inorganic ions which diminish the actin-myosin interaction (Edsall and Mehl, 1940) has already been discussed by Bailey (1944) in this publication. The formation of filaments of natural actomyosin during the extraction of minced muscle takes place to a greater extent when the extraction is prolonged; the more slowly extractable actin is thus given a chance t o combine with the more quickly extractable L-myosin, as in the 24 and 48 hour extract of Szent-Gyorgyi (1942) and all later workers. The filaments of natural actomyosin evidently build up by much the same process as those of artificial actomyosin; nevertheless, the former are much

MUSCLE CONTRACTION AND FIBROUS

MUSCLE PROTEINS

219

thinner, as thin in fact as those of pure F-actin (111, 4c). One cannot escape the conclusion that the filamenw of F-actin formed direct in the extract are thinner than those made by the polymerization of G-actin. A certain degree of caution is thus necessary in the mental transposition of artificial actomyosin as seen in the electron microscope to the state of actomyosin in muscle. When G-actin is mixed with an excess of L-myosin, the viscosity does not increase, but if ATP is then added, the increase is observed as soon as the ATP is broken down by the ATPase activity of the L-myosin. This means that G-actin is quantitatively bound by the L-myosin such that it cannot polymerize even in presence of salt. Polymerization takes place, however, when the G-actin is dissociated from the myosin on addition of ATP. As soon as the latter is enzymatically decomposed, L-myosin recombines with the actin, now present in the fibrous form, to give the highly viscous type of actomyosin (Straub, 1943). A distinction must therefore be made between G- and F-actomyosin. The former cannot yet be differentiated with certainty from L-myosin particles by physical methods, though Snellman and Gelotte (1950) state that L-myosin in the ultracentrifuge becomes polydisperse when G-actin is added. b. The Reaction of G-Actin Particles with One Another. G-actin particles are capable of free existence at pH 7 only in the complete absence of salt and of L-myosin. When salt is present, and L-myosin absent, the particles polymerize (Straub, 1943), and perhaps nothing more than salt is necessary for the process. The rate of polymerization depends markedly on the nature and concentnttion of the ions present (Feuer et al., 1948). For the alkali metal chlorides it is greatest between I = 0.1 and 0.15, when the half-time value is about 15 minutes; a t higher ionic strength, the rate decreases sharply. Magnesium and calcium ions give half-time values of 15 minutes in concentrations as low as M . Whereas magnesium tons in Fresence of 0.1 M alkali salt accelerate polymerization, Ca ions and those of the alkali salts mutually oppose each other (Feuer et al., 1948). The extent of the inhibition depends on the ionic ratio of (>aand K, and it is smallest when this is 1:50 (Straub et al., 1948). The actin preparations of these authors do actually contain traces of Mg, which are indispensible for the polymerization by alkali salts; the process does not occur if Mg is removed by Calgon. Magnesium is not necessary, however, for that part of the polymerization process in which long artin fibers are formed. When G-actin is treated with 2.5 x M Mg++, the viscosity does not change a t first, but when this Mg+f, together with that originally present, is removed with calgon,

-

220

HANS H. \\%BER

AND HILDEGARD PORTZEHL

and then KC1 is added to a final concentration of 0.1 M , polymerization and its accompanying viscosity increase occur a t once. Evidently the action of Mg ions is merely preparatory t o the polymerization into viscometrically recognizable threads; perhaps i t makes possible the aggregation into the Szent-Gyorgyi units of 300 X 100 A. (111, 4c); magnesium is certainly not necessary for the further progress of polymerization. T hat the calgon really does remove magnesium ions is shown by the complete absence of any polymerization when calgon is added simultaneously with Mg in the first states (Feuer et uZ., 1948). The action of ions in promoting polymerization suggests th a t the ionogenic groups of actin itself are involved. Dubuisson (1950~)does in fact find t ha t the polymerization of actin in a solution a t p H 8 (bicarbonate M ) is accompanied by a considerable release of hydrocontent 7 x gen ions; when 0.1 M KC1 is employed to catalyze the process, the p H falls to 7.2, and with 0.05 144 CaC12, 6.8. If the same initial solution is equilibrated with an atmosphere containing 5% COZ, and thereby mmol. H+ acidified to pH 7.2, then 1 g. of actin sets free 45 X (measured as COZ) in presence of 0.05 M calcium salt, but only 2.5 X mmol. with 0.1 M potassium chloride. It remains to be decided whether this considerable discrepancy is due to the limited duration of the experiments, since the rates of polymerization in the two cases are so different (see above). The release of H+ ions during polymerization can explain the higher negative charge and mobility of F-actin as compared with G-actin (cf. 111, 3). The ability of G-actin to polymerize is slowly lost spontaneously (rapidly in absence of ATP-see Section 111, 5c). It is destroyed also by a number of reagents, almost all of which are either oxidizing reagents or -SH reagents (Table XI, column 1). Substances which protect against their action are either reducing compounds or contain -SH groups (Table XI, column 2). It seems therefore th a t intact --SH groups of G-actin are necessary for the polymerization of F-actin, just as those of L-myosin are necessary for the formation of actomyosin. The spontankous loss of polymerizing ability may also be due to the oxidation of -SH groups, for the change is retarded by ascorbic acid (Straub and Feuer, 1950);i t is retarded also by ATP (111,5c), and as we have seen the binding of ATP appears to involve -SH groups (cf. sections 111, 5a and d). c. The Reaction of Aetin Particles with ATP. The polymeriz at’ ion and depolymerization of actin represents a profound alteration in structure of one component of the contractile protein complex, and it is worth while t o enquire whether ATP is connected with the process. Although some of the findings are controversial, others seem well established:

MUSCLE CONTRACTION A N D FIBROUS MUSCLE P R O T E I N S

22 1

(1) G-Actin solutions contain small, variable amounts of ATP (Straub and Feuer, 1950; Dubuisson and Mathieu, 1950; Laki et al., 1950; Mommaerts, 1951a). After deproteinization, ATP has been determined (a) by its effect on actomyosin solution, (b) elementary analysis of the isolated substance (Straub and Feuer, 1950), (c) by the determination of labile phosphate (Dubuisson and Mathieu, 1950), (d) by ultraviolet absorption (Laki et al., 1950; Mommaerts, 1951a), and (e) by reduction of coenzyme II.’* TABLEXI Factors Influencing the Transformation of G-Actin to F-Actin Factors preventing the transformation or reversing i t Factor Author SalyrganQ Cupric glycinate Oxarsanb Cystine Methylene blue KMnOa KI > 0.5 M Dialysis

I

5

6

Kuschinsky and Turba (1950b) Turba et al. (1950) Turba et al. (1950)

Substances antagonistic to those in Column 1 Substance Author Cysteine

Turba el al. (1950)

Ascorbic acid

Straub and Feuer (1950)

Feuer et al. (1948) Straub (1943a) Straub and Feuer (1950)

See Table 111. m-Amino-p-oxyphenylarsenic oxide.

(2) ATP protects G-actin from inactivation. The polymerizing ability is lost when an isoelectric actin precipitate is washed repeatedly with ATP-free solutions, or when a G-actin solution is exhaustively dialyzed; and this loss is not reversed by subsequent addition of ATP. When ATP is present in the wash or dialysis liquor, the ability t o polymerize is retained (Straub and Feuer, 1950; Laki et at., 1950; Mommaerts, 1951b). (3) The protective action of ATP shows with certainty that part a t least of the variable amount of ATP in G-actin is bound to the protein, and the binding probably involves -SH groups (see below). The answer to the question whether ATP is necessary for the polymerization is less certain. Two kinds of data are available: (1) The removal of ATP from an actin solution does result in a, less 18 The reduction is effected by the oxidation of hexose-6-phosphate in presence of Warburg and Christian’s “Zwischenferment ” (1932). The initial esterification is brought about by the action of hexokinase on glucose and ATP.

222

H A N S H. W E B E R AND HILDEGARD P O R T Z E H L

complete inactivation of actin if the removal is effected quickly by enzymes. Addition of salt then promotes some degree of polymerization, which is considerably increased by addition of ATP, to about one half that of untreated actin (Laki et al., 1950). These results are not unambiguous. They could mean that ATP has partially reversed a n incomplete process of inactivation, but i t could also mean that ATP itself is involved in the polymerization mechanism. (2) Data on the splitting of ATP in the polymerization of G-actin are contradictory. Straub and Feuer (1950) state th a t in the polymerization 40-807, of the ATP in G-actin is split into ADP and phosphate, the course of splitting and of polymerization running parallel. Laki et al. (1950) find that the difference in AT P content between G- and F-actin is small (10-20%) but nevertheless real. Duhuisson and Mathieu (1950) do not find any ATP breakdown during p ~ ly me riz a tio n .'~ If there is a splitting of ATP it is probably connected with polymerization. Straub and Feuer assume that the energy-content of F-Actin is higher than that of G-Actin. Therefore the energy necessary for polymerization is supposed t o be provided by the ATP-splitting. They formulate a n energy-cycle assuming th at the energy liberated during depolymerization is used for building up ATP. T o prove this point, they assume furthermore that in the acetone dried power actin exists as F-Actin. During extraction i t is dissolved only so far as i t is transformed into G-Actin. They find t ha t in such solutions of G-Actin the concentration of A T P is increasing immediately after extraction. No reasons are given why this reaction follows extraction instead of accompanying it. Summarizing, it seems almost certain that a dissociahle compound of ATP and G-actin exists, and that after dissociation G-actin denatures unusually rapidly. No further conclusions as to the role of ATP in the polymerization and depolymerization of actin can be made without more experimental evidence. d. The Interaction of L-Myosin and Actomyosin with ATP. Although ATP has no specific action on the colloidal properties of L-myosin, i t profoundly influences those of actomyosin either in sol or gel form. Both actomyosin and L-myosin break down ATP. ATP BREAKDOWN BY ACTOMYOSIN AND L-MYOSIN. Myosin preparations which have been sufficiently purified (in the usual way) split off only the terminal phosphate group of A T P ; i.e., they have only ATPase and not ADPase activity (Engelhardt and Ljubimova, 1939; D. M. 1 9 Even if all ATP were split during polymerization, this would not be sufficient to account for all the H ions released (Section 111, 5 b ) ; in the presence of Ca ions the breakdown of ATP to the extent of one-third the weight of the actin would be necessary (Dubuisson and Matthieu, 1950).

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

223

Needham, 1942; Bailey, 1942; Szent-Gyorgyi and coworkers, 1942, 1943).

No conclusive separation of the ATPase activity from myosin without simultaneous denaturation has ever been achieved (Szent-Gyorgyi, 1947, 1951; Polis and Meyerhof, 1947; Mommaerts, 1948). Polis and Meyerhof (1947) succeeded in fractionating myosin by lanthanum precipitation so that one of the fractions gave an activity some four times greater than the original myosin. This important finding suggests that the ATPase groups are not uniformly distributed over the myosin particle. The groups responsible for ATPase activity are exceptionally sensitive t o heavy metals, -SH reagents and oxidants, and, like the polymerizing groups of G-actin, are usually reactivated by reducing agents, -SH compounds, and substances which form complexes with heavy metals. They too therefore contain SH groups (Szent-Gyorgyi, 1942; Bailey, 1942; Polis and Meyerhof, 1947; see Table 111). The ATPase activity of F-actomyosin and L-myosin both in the sol and gel states, is increased by salt up to an optimum concentration (Banga, 1942). In the case of L-myosin the ATPase activity and the activation by salt are independent of the colloidal state; whether as a sol (Bailey, 1942) or as a gel (Banga, 1942) i t is not appreciably activated by Mg, and in each case, Mg inhibits the activation by KC1 (cf. Banga, 1942 with Banga, 1943). The sol (Bailey, 1942) and probably the gelzD (Banga, 1942) is activated far more strongly by Ca than by any other ion so far investigated. The curve of activity against KC1 concentration shows no discontinuity at the point where L-myosin is salted out. This has been found by all investigators, even when the form of the curves was quite differenteZ1 In the case of actomyosin, the ATPase activity is quite different in the sol and gel states. As a sol i t appears t o be the same as that of L-myosin, and is thus barely activated by Mg ions, which strongly antagonize the activation by other ions (Banga, 1942; Banga and SzentGyorgyi, 1943). Ca ions activate both forms t o the same extent. Moreover, according to Biro and Szent-Gyorgyi (1949), the activity of actoZo It is not quite certain whether L-myosin is in the sol or gel state a t the experimental concentration of 0.01 2M CaC12. I n precipitation experiments, Szent-Gyorgyi (1947), Table I, p. 5 ) states that L-myosin is not completely dissolved until a CaCl? or MgClz concentration of over 0.1 M is reached. In ATPase experiments, however, he states t h a t even F-actomyosin is completely soluble in M/100 MgC12 (see Fig. 26). The salting-in limit for the alkaline earth metals is thus uncertain; 1,-myosin should be in the gel state a t ionic strength 0.03. 2 1 Banga and Szent-Gyorgyi (1943) and Mommaerts and Seraidarian (1947) find a sharp optimum a t about 0.3 M KC1; Biro and Szent-Gyorgyi (1949) on the other hand find t h a t the activation is independent of KCl concentration between 0.1 and

0.4 M .

224

HANS H. WEBER AND HILDEGARD PORTZEHL

myosin in solution a t p H 7 is the same as th at of the L-myosin i t contains, whereas that of the gel is two to three times greater. The correspondence in the enzymatic activities of L-myosin and actomyosin in solution is plausible, since the latter is split into its components in presence of ATP. As a gel, actomyosin appears to split ATP faster than as a sol, but the statements in the literature are contradictory (Biro and Szent-Gyorgyi, 1949; Banga and Szent-Gyorgyi, 1943; Banga, 1942; Mommaerts and Seraidarian, 1947). The gel is activated strongly by Mg (Fig. 26), and

MgCLZ(M)

FIG.26. Activation of ATPase of actomyosin gel by magnesium chloride. Activation is expressed as percentage of the maximal, as obtained with 5 X 10-3 M CaC12. Curve, ATPase activjty; X superprecipitation; o solution of the gel (from Ranga, 1942).

magnesium ions do not inhibit, but rather increase, the activating influence of other ions (Banga 1942). The contradictory results of experiments with Ca allow of no definite conclusion, but Ca ions do appear to increase further the activity of KC1-activated gel (Banga and SzentGyorgyi, 1943). The ATPase activity of L-myosin, and of actomyosin in solution and in gel form, are given in Table XII. There appear to be no data on the p H dependence of the ATPase activity of pure L-myosin. Banga (1942) finds a n optimum a t p H about 6 for actomyosin (B myosin) in mixtures of alkali salts and M MgC12. The total salt concentration or ionic strength is not specified here nor in the experiments of Engelhardt (1946), in which the p H optimum of the myosin (L-myosin/actomyosin mixture) in presence of Mg++ was around pH 9, while in the presence of Ca++ there was a n additional, much weaker, optimum a t 6.3. The activity and the optimum a t pH 9 are very much lower with Mg++ than with Ca++. Mommaerts and Seraidarian (1947) give an optimum a t p H 6.5 for actomyosin (B-myosin) in alkali salts at ionic strength 0.15-0.17. I n the presence

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

225

of 0.05 M CaClz and alkali salts (total ionic strength 0.17) the same authors find an exceedingly sharp maximum a t about p H 9.0.22 Under these conditions, F-actomyosin is certainly in solution, so that the optimum at p H 9.0 can be attributed to L-myosin. The shift of the optimum t o pH 9 in presence of Mg and Ca ions corresponds t o the influence of these two ions upon the isoelectric point (cf. Section 111, 3). Whether the optimum a t pH 6.3 is that of L-myosin in absence of Ca++ TABLEXI1 Activation of A T P a s e by salts at p H Y (ajter Banga, 1942) ATP added (mg. labile P)

No salt added

Phosphate split (mg.) In KCl In MgClz I n CuC12 I n KCl (0.01 M) (0.1 M ) (0.001 M ) (0.01 M) f MgClz (0.001 M)

1.4 2.8 4.2 5.5

0.003 0.006 0.019 0.038

0.037 0.078 0.104 0.134

1.4 2.8 4.2

0.002 0.004 0.022

0.023 0.044 0.066

ACTOMYOSIN 0,074 0.104 0.162 0.168

0.055 0,100 0.140 0.162

0.074 0.120 0.134 0,180

0.064 0,109 0.165

0.028 0.037 0.044

L-MYOSIN 0.004 0.008 0.026

and Mg++ cannot be inferred. It could be that of actomyosin, since a t p H 6 this complex is insoluble and therefore undissociated up t o an ionic strength of 0.2. Taken as a whole, the influence of ions on the ATPase of the myofibril stands more in need of further work under well defined experimental conditions than any other aspect of the colloid chemistry of muscle proteins. INFLUENCE OF ATP ON ACTOMYOSIN IN SOLIJTION. The effect of A T P on the colloidal state of actomyosin solutions is quite different from th a t on actomyosin gels. In solutions, ATP lowers (i) the viscosity, (ii) the flow birefringence, (iii) the light scattering, (iv) the phosphorescence, (v) the salting-in limit, and (vi) the sedimentation velocity (Needham et al., 1941; Szent-Gyorgyi and coworkers, 1942, 1943, 1947, 1951; H. H. Weber, 1947; Snellman and Tenow, 1948). The first four of these changes show that under the influence of ATP, actomyosin in solution loses some of its fibrous character, and the lowering of the salting-in limit best fits this interpretation. 22 This optimum is absent in 0.07 M veronal buffer under otherwise identical conditions. Polis and Meyerhof (1947), however (also with carefully purified ATP), find equal activity a t p H 9 in glycine, borate, and veronal.

226

HANS H. M’EBER AND HILDEGARD PORTZEIIL

Fibrous particles can become shorter in two ways; (i) by contracting or coiling, or (ii) by dissociating into smaller particles. Jordan and Oster (1948) from the angular dependence of light scattering decide in favor of coiling. Even qualitatively, this theory is untenable, since the sedimentation constant falls to a few per cent of its original value and a decrease in assymmetry without change in mass would produce an increase. The quantitative consideration of other methods shows quite definitely t ha t ATP does not change only the shape of the particles. The following considerations are relevant; (1) However large sio may be before the addition of ATP-whether i t is too large t o be measured (the gel-like component of Snellman and Erdos (1949) and of Johnson and Landolt (1950), or >> 280, or approximately 90 (Portzehl et al., 1950)-after the addition there appears consistently only the component with &, = 7.1 (H. H. Weber, 1947; Portzehl et al., 1950; Snellman and Erdos, 1949; confirmed also by Mommaerts, 1 9 5 1 ~ ) ;and 7.1 is the sedimentation constant of pure L-myosin. Snellman and Erdos (1949) mention th a t they have sometimes found also the F-actin peak, which is much more difficult to observe. (2) When two fibrous colloids are mixed, log vrel of the mixture is additive, provided they do not react with each other (Kaumans, 1949).23 Thus the plot of log vrel against the composition of the mixture is a straight line joining the values of log vrS1for the two original solutions. The values of log vrel for mixtures of F-actin and L-myosin lie far above the line, but after the addition of ATP they fall exactly upon it (Fig. 27; H. H. Weber, 1950a), as would be expected for mixtures of free L-myosin and free F-actin (confirmed by Mommaerts, 1951~). (3) Finally, the close similarity in the ATPase activity of actomyosin and L-myosin solutions is evidence that, in presence of ATP, L-myosin is free. When bound to actin, its ATPase activity differs from th a t in the free form (cf. Section 111, 5 4 . It can thus be taken as definitely established th a t ATP causes a dissociation of actin and L-myosin and not merely a change in the shape of actomyosin particles. The higher solubility of actomyosin in the presence of A T P can then be explained by its conversion into relatively soluble L-myosin and water soluble F-actin. The question again arises whether the effect of ATP is due merely to its presence and to its attachment to the actomyosin, or is due t o the process of breakdown. This question is more easily and certainly answered for the dissociation of the two proteins in solution than for the 23

qspcc itself

vrel =

2.303 log

is additive only when it is so small t h a t the approximation qrcl liolds closely enough.

?

I

= ~In ~

~

~

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

227

shrinkage or contraction of the gel (see Section 11, 4 4 , for the dissociation effect is less specific and is brought about by a number of organic and inorganic polyphosphates and by substances which form complexes with -SH groups, most of which are not split by F-actomyosin (cf. columns 6 and 8, Table XIII). The dissociation of actomyosin in solution is due, therefore, not t o breakdown but to combination. Unlike the synaeresis or contraction of the gel, the dissociation of actomyosin is completely reversible when the causal agent is removed. When ATP or ITP have been broken down, F-actin and L-myosin com-

,

2

.

0

ACTIN,ml. 12 MYOSIN,ml. 0

8 4

4 8

0 12

FIG.27. Viscosity of artificial actomyosins. Curve 1: before, curve 2 after, ATP addition. Actin solution 0.385 %, L-myosin 0.701 % (from H. H. Weber, 1950a).

bine once more, as is shown by sedimentation, viscosity and flow birefringence (Table XIII). There is no spontaneous return, however, when the agent is not split, even in the case of A T P itself if breakdown is largely or completely inhibited by addition of Mg ions (cf. columns 7 and 8, Table XIII). Thus, unlike dissociation, spontaneous recombination is dependent upon breakdown, and implies th a t the action of ATP on actomyosin in solution has no connection with the fundamental process of contraction (cf. Section 11, 4e and h ) , b u t may be related t o the second fundamental action of ATP on the contractile models; for since its action is t o diminish the cohesional forces of the actomyosin system, i t may be connected more with the destruction of rigidity, the increase of extensibility, and the production of some degree of plasticity both in oriented and unorierited actomyosin gels (cf. 11, 3g and 11, 4h). It has not yet been established to what extent ADP causes dissocia-

228

H A N S H . WEBER A N D H I L D E G A R D PORTZEHL

TABLEXI11 Reagents Causing Dissociation of Actomgosin Effect ___. Experimental conditions for an Sponta0.55 and .~ pII 7 ionic strength neous -~ hlgClz Temp Dissocia- recoin"C. Method tion ( % ) bination

-

Concentration (MI

Substance

Enzymatic splitting

~~

x

10-4

ATP

2.5 X

lo-'

ATI'

5.4

x

10-2

ITI'a ? ?

ITPG

?

ITP Na triphosphate< Na triphosphatc' Na trlphosphate? Na triphosplratec Na triphosphstec Na triphosilhateh Na pyrophosp h a ted Na pyrophosphated Na pyrophosphateh Na pyrophosphatc': Salyrgan' Chpric glycinatef

10-4 10-4 10-3 10-3

10-3

3

x

10-2

9 6 X 10-5

x

63

10-3

3

x

10-2

5

x

10-4

1

- 5.10-2 ATP

4 X 10-3 3

0 Viscosity 2o {Viscosity< , Sedunentationf Flow birefrin20 genceb Sedimentationq 20 1"low birefringence 20 Viscosity 20 Viscosity 0 Viscosity 18 Viscosity

1 . 5 x 10-4 5 x 10-4

ITP

100 100

100

75 0 100 100 0

+

None in 30 niiri

+ + -

+? +?

-

+

Very small

+ +* +' +' +' Nonch

0

Viscosity

100

None

None

18

Viscosity

80

Noiic

None

0

Viscosity

50

None

None

18

Viscosity

0

-

None

20

Flow birefringence Viscosity

0

-

None

0

0

0

23

0

20

10-2 ? ?

Viscosity

100

None

None

0

None

None

-

None

17

Flow hirefringence Viscosity

100

None

None

20 20

Viscosity Viscosity

100 100

None None

None None

0

" Szent-GyGrgyi and coworkers (1942, 1943).

' Kleinzeller (1942). Dainty ef al. (1944). Tnrba el al. (19.50). Moniiriaerts (1948). d Straub (1943h). a Kuschinsky and Turba (19501~). f Snellnian and Erd& (1949). y Wpher (1950a). Bailey (1944) finds that inorganic triphosphate is not split. Dainty et al. (1944) found a slight decomposition. b u t assumed that i t was due to a n enzymatic impurity. h

(.

tion of actomyosin in solution. Dainty et al., (1944) and Mommaerts (1948) do find small effects, but consider their actomyosin preparations to be too impure to give a final answer. All the organic and inorganic polyphosphxtes (apart from ATP) which produce dissociation work better a t 0" than a t a higher temperature in the concentrations employed (Straub 1943b; hiommaerts 1948). This could mean that their combina-

229

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

tion with F-actomyosin is appreciably exothermic, though such a view requires proof. The dissociation b y substances which form complexes with SH groups (Turba et al., 1950), and th at b y polyphosphates at favorable temperatures (Mommaerts, 1948), takes an appreciable time. Since ionic equilibria are reached immeasurably quickly, this may- indicate that slow secondary reactions are also involved. As regards the mechanism by which actomyosin is dissociated, i t is found t ha t reagents which form complexes with S H groups and which imitate the dissociating action of ATP (salyrgan, copper glycinate; see Table XIII),a t the same time inhibit all other effects of A T P (Table 111). It would appear therefore th at these SH reagents, A T P and actin all combine with the same SH centers of the L-myosin molecule. SzentGyorgyi’s earlier conception is thus extended and made more precise by the chemical definition of the active center. The affinity increases in the order actin < ATP < the SH reagents mentioned above. It cannot yet be explained how the SH groups bind ATP. The obvious assumption th at esterification occurs (Binkley, 1945) can hardly be correct, since the technically elegant measurements of Fabry-Hamoir (1950) show that there is no pH change when A T P is added to actomyosin. Even if the binding of A T P by actomyosin were confined to that portion of the SH groups, which, according t o Singer and Barron (1944) are involved in ATP breakdown, the p H change should be 0.01, whereas the standard error of the measurements was only 0.05 mV or 0.003 pH units. This difference is very small, b u t the excellent technique for the measurement of small pH changes developed by Dubuisson and his school make it probable t h a t no such difference has escaped detection. There is, of ,course, a p H change produced by A T P breakdown from the time of mixing onwards, b u t the first measurement was made 15-20 seconds after mixing, and the pH-time curves were extrapolated hack t o zero time. INFLUENCE OF ATP ON ACTOMYOSIN GEL. F-Actin and L-myosin, which under the influence of ATP are dissociated in solution, combine t o give F-actomyosin with the formation of a gel when the ionic strength drops below 0.2 (Szent-Gyiirgyi, 1947,1951). The supernatant liquid contains no protein, although free F-actin is water soluble and the precipitation of free L-myosin is not complete until the ionic strength reaches 0.03 to 0.05. I n neutral solutions of alkali salts, A T P ceases to act as a dissociating agent a t ionic strengths below 0.2 or lower, though it may still diminish the cohesional forces in the gel (cf. Sections 11, 39, 11, 4h and 111, 5 4 . At ionic strengths of 0.15 or less, the second effect of ATP is observable, the microscopic superprecipitation and macroscopic shrinking of

-

230

HANS H. WrlSBICR A N D HILDEGARD PORTZEHL

the gels (main Section 11). The gel volume can shrink to 1/20th of the original (Szent-Gyorgyi, 1942; Porteehl et al., 1950), whereas in physiological contraction the volume remains practically the same (SeentGyorgyi and coworkers, 1942; Buchthal et al., 1947). Shrinkage and contraction appear to depend upon A T P breakdown (cf. Section 11, 4d and e ) . I n the shrunk or the contracted state, actomyosin appears to be undissociated, for the characteristics of the ATPase activity are those of F-actomyosin gel and not of free L-myosin (cf. Section 111, 5 4 . The changes which occur in the minute structure of actomyosin on shrinkage are not known. Perry et al. (1948) found th a t the wide angle X-ray diagram was of a-type, both before and after shrinking. The latter authors in the electron microscope also observed th a t on addition of ATP there was an enhanced tendency to fiber formation by side t o side aggregation, whereas Snellman and Erdos (194%) found th a t thread-like particles gave rise to cluster formation. The photographs of Perry et al. are of natural actomyosin, which contains an excess of L-myosin, whereas Snellman and Erdos used an artificial actomyosin containing myosin and actin in a 3/1 ratio. Finally, Buchtal et al. (1949) investigated whether chemical changes take place in F-actomyosin when the threads shrink. Threads of F-actomyosin and also of L-myosin were found to contain, after treatment with 2 X M ATP followed by 7-12 washings, a content of phosphate, adenine and ribose three to five times greater than before. The relative amounts of the different phosphate reactions (cf. Section 111, 4a) are not appreciably altered from those already present. The effect is as specific for ATP as shrinkage and contraction, but i t probably has no direct connection with the fundamental process of contraction since i t also occurs when actomyosin and L-myosin are in the dissolved state (cf. Section 111, 5 4 . APPENDIX:OTHERENZYMATIC

ACTIVITIES

OF MYOSIN

Myosin sufficiently purified in the usual way does not dephosphorylate any of the numerous phosphate compounds of living muscle other than ATP, and perhaps ITP, nor can it transphosphorylate. Menne (1943) finds that myosin, unlike the other main fractions of muscle, can convert arginine, histidine, glycocyamine, and choline into creatine. The myosin used, however, was only reprecipitated once and subsequently washed, and i t is possible that the enzyme activity might be lost on further precipitation. After fractionation and precipitating three times, myosin possesses an appreciable adenylic deaminase activity (Hermann and Josepovits, 1949 ; Summerson and Meister, 1944).

MUSCLE CONTRACTION AND FIBROUS

231

MUSCLE PROTEINS

6. Isolation of the Fibrous Proteins of Muscle

a. Extractability. Since none of the fibrous proteins a t p H 6 or higher has a salting-in limit greater than ionic strength 0.6, i t might be expected that such a solution would be capable of extracting all of them from fresh muscle brei. This is not the case, however. Tropomyosin has not been found in such extracts (see Bailey, 1948; cf. also Dubuisson’s electrophoretic diagrams 1950d with those of 1950c), and certainly L-myosin can be exhaustiveIy extracted at p H about 6 without extracting at the same time an appreciable amount of actin. Evidently, extractability is not solely determined by solubility, and this is hardly surprising, for it would indeed be strange if the dissolution of F-actin or of F-actomyosin threads, several micra long, were not seriously impeded in a purely mechanical way by the insoluble components of muscle. It has long been known that these solutions when filtered will quickly block the filter paper. When the actomyosin threads are made t o dissociate b y A T P or inorganic pyrophosphate or KI, L-myosin is rendered extractable (Szent-Gyorgyi and coworkers, 1943; Dubuisson, 1950f; Hasselbach and Schneider, 1951). I n this case, it is not even necessary tomince the muscle a t all (Amberson et al., 1949). F-Actin threads become extractable when they are converted into undenatured G-actin b y Straub’s method (1942 and 1943a; see also Szent-Gyorgyi, 1947, 1951), or in the denatured condition after extraction with K I (Dubuisson, 1950f; Dubuisson and Fabry-Hamoir, 1950). Failing this, the rate of extraction of F-actin depends on the state of the enclosing insoluble muscle 6 is structures, which become denser as the isoelectric point p H approached. The muscle structure can be mechanically destroyed t o varying degrees. When muscle is coarsely minced, F-actin is not extractable a t all a t p H < 6 (Hasselbach and Schneider, 1951), and is given up very slowly a t p H 7.5, the extraction being still incomplete after 24-28 hours. The product is the myosin B of Szent-Gyorgyi, (Banga and Szent-Gyorgyi, 1942), which is relatively rich in actin. If coarsely minced muscle is treated in the Waring blendor for about 4 minutes, F-actin is extracted rapidly and completely a t both p H 6 and p H 7.5 (Hasselbach and Schneider, 1951). If L-myosin has not previously been extracted, actomyosin is formed in the extract.24 The proportion of

-

2’ Treatment in the blendor not only destroys the structures which hinder diffusion, but t o a greater or lesser degree very probably breaks up the long F-actin threads. This cannot be directly proved, since such threads regenerate spontaneously, as shown electron optically by Rozsa and Staudinger (1948) in the case of threads destroyed by ultrasonic vibration.

232

H A N S H. W E B E R AND HILDI’GARD

PORTZEHL

F-actin in the extract is greater the longer the homogenization. This is shown in Fig. 28, in which ATP-sensitivity of a 20-minute extract a t p H 6 is plotted against the duration of blending. When finally the surrounding structure is entirely eliminated by isolating the actual fibrils (Schick and Hass, 1949), these latter are dissolved immediately in 0.6 M KC1 to give a typical actomyosin solution. It is unnecessary, therefore, to invoke a chemical combination of actin with the insoluble structural components to explain the difficulty of its extraction (Straub, 1942; Szent-Gyorgyi, 1947). The view that a solution of L-myosin is a particularly good extractant because its affinity for

TIME OF BLENDING, minutes

FIG.28. Dependence of thc ATP-sensitivity of a 20 minute extract pH = 6 on the duration of bleridor treatment. ATP sensitivity expressed as %2%E’ x 100 ZVATP

(see Section 111, 6, Appendix) (from Hasselbach and Schneidcr, 1951).

actin is greater than that of the insoluble components (Straub, 1942, 1943a has not been confirmed. Exact comparative experiments show that such L-myosin solutions are no better than salt solutions of the same ionic strength (Hasselbach and Schneider, 1951; cf. also Csapo, 1950). On these concepts, the fact that an increase of ionic strength as well as the state of comminution increase the rate of extraction (Dubuisson, 1947) can be explained on two grounds: (i) the ionic strength inside the muscle fragments more rapidly reaches th at required t o effect solution because of the steeper concentration gradient. This factor certainly plays a part in 20-minute extractions. (ii) When the A T P level of muscle is diminished, as by fatigue or contracture, very high salt concentrations, in the absence of specific dissociating ions, (cf. Section 111, 5d and Table XIII), induce a certain degree of dissociation in the actomyosin complex (111, 5a), and this is a prerequisite for extraction.

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

233

It thus seems true to say that the extractability of L-myosin and actin depends solely on the mutual combination of these proteins, and the hindrance t o diffusion by the surrounding muscle structures (cf. Dubuisson, 1947). The case of tropomyosin is more complicated. This protein seems to have been identified only in saline extracts of muscle residue which has been dried with ethanol and ether (Bailey, 1948), or with acetone, CS2 and light petroleum (Dubuisson, 1950e). The failure t o find it could be due to its small amount, for in extracts of dried powder its proportion is much greater when most of the other proteins have been rendered insoluble. Tropomyosin is water soluble, but it can nevertheless be obtained from the dried residue even when the water soluble proteins have been removed by repeated extraction. Moreover, for the extraction from the residue, water alone does not suffice, and salt solution of optimal strength 1 = 1 is necessary (Bailey, 1948; Dubuisson, 1950e). Thus, tropomyosin really does undergo some change of state during extraction. If before the residue is dried the L-myosin and the globular proteins are extracted, the yield of tropomyosin from the dried powder is diminished by an amount which is greater the higher the ionic strength employed for the extraction of L-myosin; for a n ionic strength of 1, the yield falls t o 20% of the normal (Bailey, 1948). This fact, and the close relationship between the two proteins (see Section 111, 4b) suggest that tropomyosin might be one of the subunits of 1,-myosin which is split off under the influence of organic solvents and of strong salt solutions. This concept is a t variance, however, with the fact that such salt solutions do not extract appreciable amounts of protein from pure myosin preparations dried in ethanol and ether (Bailey, 1948). Paramyosin has not yet been extracted in the usual sense. It is obtained from certain mollusc muscles by grinding in Edsall-Weber solution, destroying the gross structure, and bringing much of the protein into solution. I n such extracts the paramyosin occurs in the form of intact fibrils, which dissolve a t ionic strength 0.6 (Schmitt et al., 1947). b. Fractionation and PuriJication. The differences in extractability make i t possible to begin purification a t the earliest stage by fractional extraction. This process has been exploited most in the purification of actin by Straub (1942-3). Most of the L-myosin and the globular proteins are removed from fresh brei b y initial extraction with salt solutions ( I = O . G ) , and the residual sacroplasmic proteins are removed by repeated washing with water a t alkaline pH; the unextracted L-myosin is then denatured with acetone. When the dried residue is extracted

234

H A NS H. WEBER AND HILDEGARD PORTZEHL

with COz-free water, only actin goes into solution, as has been checked electrophoretically by Dubuisson (1950c), tropomyosin not being soluble in water alone. Denaturation is avoided to a large, though variable, extent (Dubuisson, 1 9 5 0 ~ Mommaerts, ; 1951a) by the shortness of the process and the arrest of post-mortem changes by the use of acetone.26 Part of the extracted G-actin, however, always appears to have lost its ability to polymerize (Mommaerts, 1951a) and to combine with L-myosin (Jaisle, 1951). The undenatured portion can be converted into F-actin, spun down in the ultra-centrifuge, and redissolved as G-actin in water containing ATP (cf. Section 111, 5c; Mommaerts, 1951a). Short extraction a t low pH (6-6.5) gives an L-myosin largely, sometimes entirely, free of actin and actomyosin (Portzehl et al., 1950; Jaisle, 1951). This is the principle of the “ A ” extraction, which yields the A-myosin of Szent-Gyorgyi. Such extracts are suitable material for the preparation of pure L-myosin, which tends, however, to denature during purification and ageing. The danger is particularly great in the so-called ‘ I crystallization’’ process of Szent-Gyorgyi (Table XIV; Snellman, and Erdos, 1948a). A worth-while procedure has proved to be that in which the sarcoplasmic proteins are first removed by precipitating all the myosin at ionic strength 0.04 and pH 6.5, dissolving the precipitate and fractionating at ionic strength 0.3, when actomyosin is precipitated and L-myosin remains in solution. Reprecipitation can be repeated at least four times without causing the appearance of denaturation products, provided the total time taken does not exceed five days (TableXIV). These findings have been confirmed by Mommaerts and Parrish (1951).26 The actomyosin fraction can be freed from L-myosin by further reprecipitation. It too tends to denature on repeated precipitation and on ageing, and although the actual viscosity does not change much, the ATP sensitivity decreases at first slowly, and then more rapidly. When 25 Actin can be extracted by itself from fresh muscle brei provided the L-myosin has been exhaustively removed (cf. IV, 1). This is done b y repeatedly extracting relatively coarsely-minced muscle with 0.6 M KC1 at pH 6, but the process takes so long that an appreciable amount of actin becomes denatured (Hasselbach and Schneider, 1951). The significance of t h e acetone treatment is t h a t it renders L-myosin insoluble, breaks its combination with actin, changes F-actin into G-actin, and a t the same time arrests denaturation processes. 2G These workers carry out the last two precipitations of the aIready purified L-myosin as in the “crystallization” procedure of Szent-Gyorgyi (1943). They avoid damage by working quickly (15 minutes) and a t 0”, b u t the purity of the preparation is not increased. The “recrystallization” can be considered t o some extent a control on the purity, since impure L-myosin does not so easily form paralleloriented threads.

-

MUSCLE CONTRACTION

AND FIBROUS MUSCLE PROTEINS

235

denaturation is extensive the sedimentation peaks of L-myosin and of its denaturation products become more and more pronounced. For the isolation of natural actomyosin i t is best to start with 24 hour extracts (pH 7-7.5) which are rich in actomyosin. TABLEXIV Denaturation of L-Myosin from Sedimentation Experiments (from Portzehl el al., 1950) Denatured L-myosin Exp. PreparaNo. __

1 2

3 4 5 6

7 8 9 10 11 12 13 14 15 16 17 18 19

I

Treatment of preparation

Extract

Once pptd. Once pptd. and fractionate Once crystallized

Fractionated and twice ppttl.

}Twice crystallized Twice crystallized and once pptd. Fractionated and 3 times pptd. Fractionated and 4 times pptd.

Age postProportion mortem of L-myosin (days) (s,", = 7.1)

Sedimentation

1'roL;ortion

1 14 15 16 5 8

100 77 82 76 100 100

0 23 8 24 0 0

4

100

13

0 I3 0 0 0 17 15 16 19 15 33

15

87 100 100 100

19

68

8

84

12

66

9

67

5 5

100 65 100

0

47

4 6

13

0 35 0 39 14

Tropomyosin extract prepared according to Bailey ( I 048) contains proteins of the myogen group and denatured C-actinZ7 (Dubuisson, 1950e). For complete purification a series of precipitations and crystallization are necessary (Bailey, 1948). 27 The G-actin in the residue which is not denatured becomes F-actin, since the extracting solution contain3 salt; but the long F-nctin threads cannot escape through the surrounding muscle structure.

236

HANS H . W’EBER AND HILDEGARD PORTZEHL

APPENDIX:“ACTIVITY”(STRAUB,1942) AND (Portzehl et al., 1950)

“ ATP-SENSITIVITY

”

Since viscometry is convenient and rapid, it is very much used in the characterization of the fibrous muscle proteins. The viscosity number 2, (Section 111, 4a) gives some information on the purity of L-myosin and on the polymerization of actin. The viscosity change on adding AT P is a sensitive test for the presence of undenatured actomyosin, and a large effect is taken to mean a high actomyosin content. Straub assumes further that the magnitude of the ATP effect can be used to derive the proportions of F-actin and L-myosin present, believing th a t it depends only on the proportions and on the protein concentration (Straub, 1942). This assumption has not been borne out in practice (cf. 111, 4 4 , and the ATP effect does not give quantitative information on the F-actin content. It does, however, provide a means of following the denaturation of an actomyosin preparation. I n order t o obtain information on the L-myosin-actin ratio without determining the total protein content, Straub (1942) converts the A T P effect into an “activity.” This relates the observed AqSpe.. to that of a. normal preparation (AllBpec. of such concentration that its viscosity in presence of ATP is the same as that of the solution under investigation. Aqspeo. X 100. Aqswc. normal

The “activity” is -~

AoSpec.is obtained by measurement,

The ATP effect on a normal preparation is th at corresponding to 100% activity. The characterization of actomyosin solutions by means of “ATPsensitivity” is simpler (Portzehl et al., 1950). Here a constant independent of concentration, the viscosity number, is employed. The 2, - ZqATP ATP-sensitivity ” is defined as where 2, and ZVATP are the A~speo.,rormal is read off from a diagram in Straub (1942).

Z,ATP

viscosity numbers before and after addition of ATP respectively. If the ATP is added in so small a volume of liquid that the protein concentration is effectively unaltered, this formula then becomes log __ f r e l . - log 7]re1. ATP log v r e l . ATP

x

100,

since by the Arrhenius equation log 9 is proportional to Z , (Section 111, 4a and d ) . The expression thus gives the percentage difference between 2, without and with ATP. The ratio of L-myosin to F-actin can be estimated quantitatively only from the concentration gradient curves in the ultracentrifuge or Tiselius apparatus (Dubuisson, 1946b).

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

237

IV. THE PROTEINS OF THE MYOFIBRIL AND THE FINESTRUCTURE OF SKELETAL MUSCLE The study of the contractile model (section B) has shown th a t the contraction, shrinkage and superprecipitation of F-actomyosin are apparently due to the same fundamental process which occurs in the contraction of living muscle. The study of the fibrous proteins and their reactions (Section 111) has shown that this process occurs only when F-actomyosin is in the gel state, but it has not thrown light on the nature of the change in protein structure t o which it is due. The question arises whether this gap can be bridged by analysis of the fine structure of muscle itself, for considerably more work of a physical nature (polarized light, X-rays, and particularly the electron microscope) has been carried out on whole muscle than on actomyosin gels. This work will be discussed in so far as i t is of significance in the identification of actomyosin in muscle.

1. The Proportion of F-Actin and L-Myosin in the Whole Muscle Protein T o decide what parts of the muscle structures are composed of actomyosin, the proportion of F-actin and L-myosin in the whole muscle protein must first be known. These amounts could not formerly be determined, since the successive extracts obtained on exhaustive extration mostly contained both L-myosin and F-actin, and the F-actomyosin formed from them could not again be separated completely into its components (Straub, 1942). Hasselbach and Schneider (1951) have found, however, th at fractional extraction of the two proteins is possible. At ionic strength 0.6 and p H 6, L-myosin and the sarcoplasmic proteins can be exhaustively extracted from coarsely minced muscle, and the L-myosin can then be isolated and estimated. If now the residue is homogenized in the Waring blendor, only actin goes into solution, and it too can be determined. From the residue only 2-5% of protein is extracted by strong urea solutions, and this figure represents the maximal amount of L-myosin and actin left behind. According to GerendAs and Matoltsy (1948), this urea-extractable protein should actually be N-protein, claimed by them to possess negative flow birefringence, but neither Dubuisson and Fabry-Hamoir, (1950), nor Hasselbach and Schneider (1951) in our laboratory, have observed this property in the urea fraction. Such a fractionation of rabbit muscle leads t o values of 38% of L-myosin (as per cent of the total protein), and 13-15% of actin, giving a ratio 2.5-3 for the two proteins. The total F-actomyosin is some 52%

-

238

HANS H . WEBER AND HILDEGARD PORTLEHL

of the muscle protein, or 56 % if the urea-extractable protein is also regarded as actomyosin. The figure of 38% for L-myosin is nearly as high as that obtained by Weber and Meyer (1933) (cf. Table XV) for myosin. This latter on our present knowledge might be thought to include the actin as well, and thus to represent the actomyosin content. Evidently, however, it does not, for the reason that extraction was performed at a pH (8-9) a t which, as we now know, the actomyosin becomes dissociated (Guba, 1943; TABLE XV Protein Fractions of Muscle (as Percentage of Protein N ) Author

Material

Weber and Meyer Rabbit (1933) (white muscle) Bate-Smith (1937) Rabbit (white muscle) Hasselbach and Rabbit Schneider (1951) Bailey (1939) Torpedo Reay and Kuchel Haddock (1936) Dyer, French, and Cod Snow (1950)

Sarcoplasmic proteins

L-Myosin Actin

Stroma

44

39

17

27

57

16

28

38

12 30

68 67

20 (16 after urea extraction) 10 3

21

76

3

14

52

Portzehl et al., 1950), and the actin subsequently destroyed; a t least, it does so in absence of ATP. It can be assumed, therefore, that in these estimations a large part of the actin depolymerized and was included in the sarcoplasmic fraction. The agreement between the actomyosin value of Bate-Smith (1937) and the sum of the values for L-myosin and actin obtained by Hasselbach and Schneider (1951) is satisfactory.28 For the further discussion of rabbit muscle, the newer values of Hasselbach and Schneider will be taken as a basis. The actomyosin content of fish muscle, some 70%, appears to be appreciably greater than that of mammals. The difference may well reside in the muscle as a whole rather than in the fiber content. If it is assumed that the stroma protein consists substantially of sarcolemma, connective tissue, and blood vessels, then the difference would be due to 28 Bate-Smith’s results are quoted differently by various authors (cf. Bailey, 1944; Dyer et al., 1950; Hasselbach and Schneider, 1951), because the author gives not only the values determined directly, but also those incorporating various corrections. The values in Table XV are taken from the critical selection by Bailey (1944).

MTJSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

239

-

the lesser development of these auxiliary tissues in fish. But of the 67 % in the rabbit total extractable protein, actomyosin comprises and 70% in fish. The tropomyosin content in the case of rabbit muscle is about 6 % of the total protein (Bailey, 1948), and it is not improbable that in the scheme of fractionation given above i t is included in the L-myosin value.

-

2. ACTOMYOSINAND

THE

FINESTRUCTURE OF MUSCLE

a. The Fibrous Fine Structure of Resting Muscle. One might expect that the following characteristics of F-actin and L-myosin would be recognizable in the whole muscle: (1) X-ray diffraction pattern; (2) threadlike shape ; (3) birefringence. While the large-angle X-ray pattern of muscle is that of well-oriented L-myosin (a-type), the observations of MacArthur (1943) and Astbury (1948) make it probable that most of the small-angle spacings found by Bear (1945) are those of F-actin (cf. Section 111, 4c). Both of these proteins exist, evidently, as filaments well oriented with respect to the fiber axis. It is not known whether all the F-actin is thus oriented, or only part, localized perhaps in certain parts of the sarcomere. In the A bands of the muscle the filaments are oriented parallel to the fiber, for these bands show positive intrinsic and rod birefringence (cf. Table 11), and for a long time it was thought certain that they were composed of actomyosin, and probable that they contained all the actomyosin (H. H. Weber, 1934a; 1939). This last assumption is now hardly tenable in view of the recent data on the actomyosin content of muscle (see above). The A bands occupy some 60% of the volume of the fibril in most vertebrates (Hurthle, 1930; Holz, 1932; Buchtal et al., 1936; Lundi, 1944), and the total volume of the fibrils themselves is a t most 60430% that of the fibers. All the actomyosin could therefore be accommodated in the A bands only by assuming a protein concentration greater than in the rest of the muscle. Moreover, the excellent electron micrograms show filaments aligned in the direction of the fiber axis and running through both the A and I bands. These look very like the threads of pure F-actin and F-actomyosin. (Wolpers, 1944; Hall et al., 1945, 1946; Draper and Hodge, 1949; Rozsa et al., 1950; HoffmannBerling and Kausche, 1950). To the filaments composing the myofibril varying thicknesses have been assigned: 40 A. in the frog (Hoffman-Berling, and Kausche, 1950); 100 A. in the rabbit (Rossa et al., 1950); 160 A. in the toad Bufo marinus (Draper and Hodge, 1949). On the other hand, Hall et aE. (1946) consider the filaments are not of uniform width, but vary from 50 t o 250 A., with a statistical weight average of about 130 A. The differences are

240

H A NS H. \VEBER A N D HILDEGARD PORTZEHL

partly due t o the fact th at in metal-shadowed preparations the thickness is taken as the distance from the middle of one filament to that of the next (Draper and Hodge, 1949; Rozsa et al., 1950), whereas in stained specimens this distance is further subdivided into the thickness of the filament itself and the interstitial space. Thus Hoffman-Berling and Kausche find the latter t o be 70 A., which together with the filament diameter of 40 A. amounts to 110 A. Similar considerations may apply for the smallest diameter (50 A.) observed in the stained preparations of Hall, Jakus, and Schmitt. It would thus appear th a t for bundles of filaments a transverse periodicity of rather more than 100 A. is the most common; one of 115 A. was given by Bear (1945) from small-angle diffraction patterns. It is still possible, however, that the ultimate filament is much thinner. When viewed in bundles, the filaments appear t o have longitudinal repeating units of 400 A. for the frog, rabbit and toad. (Hall et al., 1946; Draper and Hodge, 1949; Rozsa et al., 1950). Hoffmann-Berling and Kausche, however, give for the same repeating unit a value of 230300 A. with a sharply defined weight-average value of 250 A., and in the rabbit, Rozsa et al. believe that in favorable specimens they can see this smaller spacing as well as the larger repeat of 400 A. It is doubtful to what extent these longitudinal spacings can still be detected in single filaments torn from the bundle. The opinions of the above authors are somewhat indefinite and various. Bear (1945) however, has deduced from X-ray diffraction photographs a longitudinal periodicity of ahout 400 A. in several different muscles. What fraction of the total volume of the fibril is occupied by the filaments is a controversial matter. Apart from the uncertainty as to how much of the transverse spacing of 115 A. is taken up by the filament and how much is interstitial space, there is no agreement as to the extent to which the fibril is filled by bundles of filaments a t all. Draper and Hodge (1949) regard the fibril as a sort of tube of which only a thin wall contains the axially aligned filaments (cf. Pease and Baker, 1949). This view is based on the excellent electron optical transparency, which makes it possible t o distinguish each individual filament even in relatively broad fibrils. Other authors assume either definitely (Rozsa et al., 1950) or tentatively (Hall et al., 1946; Hoffmann-Berling and Kausche, 1950) that the whole cross-section is occupied by filaments. These data can now be compared with those on the pure proteins of the actomyosin complex, for the length of the muscle filaments makes it plausible t o regard them either as threads of F-actin or of F-actomyosin. I n comparing dimensions obtained in the electron microscope, the type of preparation must first be taken into account. Small diameters

MUSCLE CONTRACTION

AND FIBROUS

MUSCLE PROTEINS

241

(50 A.) have been obtained both for muscle filaments and natural actomyosin threads (Section 111, 4d) only when the preparations are stained (see above), and in metal-shadowed preparations thicknesses less than 100 A. have never been found (Draper and Hodge 1949; Rozsa et al, 1950; Jakus and Hall, 1947). A comparison of purified material with muscle filaments is therefore justified only when the preparations have been subjected to the same treatment. Metal-shadowed threads of artificial F-actin obtained by polymerization of G-actin are 100 A. thick (Jakus and Hall, 1947; Rozsa et al., 1949), and threads of natural actomyosin appear to be about as thick, for they have a very similar sedimentation constant and a similar thickness in metal-shadowed preparations. From its thickness, the muscle filament could thus be identified with the F-actin or actomyosin thread. The significance of the longitudinal spacing is doubtful. The 400 A. period has never been found either in F-actin or in F-actomyosin, and Rozsa et al. (1950) do not believe that it is intrinsic in the structure of the filament, but is superimposed by bands of material which run transversely over the filaments. Draper and Hodge (1949) take the opposite view. It would seem desirable to investigate whether the spacing can be derived by small-angle X-ray diffraction from actomyosin threads or oriented films. Since artificial actin threads show a very marked longitudinal repeat of 300 A. (Section 111, 4c), Rozsa et al. (1950) are inclined to regard the muscle filaments as threads of actin. The situation is thus a little obscure, but two alternatives can be formulated if the amounts of L-myosin and actin are considered. (1) Taking into account that only 15% of the muscle protein consists of actin, the filaments themselves could represent F-actin threads if it is accepted, as Draper and Hodge (1949) believe, that they occupy just a small portion of the fibril. In such a case, the myosin must be distributed in a form which is not resolvable in the electron microscope. (2) The muscle filaments can be identified with a c t o m y ~ s i nif~they ~ are assumed to fill the whole fibril, for actomyosin comprises some 60-70% of the fiber content. (Section IV, 1.) These two possibilities lead to different consequences in explaining the double refraction of muscle. The A bands are ten times more birefringent than the I bands (Schmidt, 1934) and the first alternative would make possible the assumption that the L-myosin between the 29 If the filaments are built of a continuous F-actin thread with L-myosin attached (8eent-Gyorgyi, 1947; Roam et al., 1950), t h a t part of the filament consisting of actin could not have a diameter greater than about 50 A.; and since there is three times as much myosin as actin (cf. Section IV, 1) the former must account for three quarters of the cross sectional area of the whole filament.

242

HANS 13. WEBER AND HILDEGARD PORTZEHL

filaments is oriented parallel t o the axis only in the A bands, giving rise to strong intrinsic and form birefringence. It would not then be difficult to regard the rods of 2000 A. in length (cf. Section 111, 4a), lying parallel to each other, as the cause of form birefringence, and the 8-10 peptide chains in each as the cause of intrinsic birefringence. Such a structure would not be resolvable in the electron microscope. The second alternative would make it necessary to invoke auxiliary hypotheses, to explain how the positive birefringence due to actomyosin, uniformly distributed throughout the sarcomere, is compensated in the I bands by a negative birefringence almost as great as the positive birefringence of the A band. According t o Gerendh and Matoltsy (1948), the compensation is due to N-protein, a fibrous protein with negative intrinsic and positive form birefringence. No other worker, however, has found such a protein (cf. Section IV, 1). Furthermore, if in the aqueous medium of the muscle a negatively birefringent protein is compensating the positive birefringence of the actomyosin, then after extraction of the latter, the I bands in aqueous medium should be as strongly birefringent in the negative sense as the A bands formerly were in the positive. This has never been observed, even by GerendAs and his ~ollaborators.~0 Hoffman-Berling and Kausche (1950) assume that the positive intrinsic birefringence of the actomyosin is compensated in the I bands by negative form birefringence due t o the transverse structures mentioned above, which overlie or connect the protein filaments a t distances of 300-400 A. The A bands also possess these structures, but the authors consider that here the interstitial fluid has so high a protein concentration that its refractive index is the same as that of the filaments. All form birefringence is thus eliminated, and only the positive intrinsic birefringence remains. This suggestion covers the facts for muscle in aqueous medium, but does not consider the fact that positive form as well as intrinsic birefringence have been demonstrated in the A bands (No11 and Weber, 1934; Fischer, 1944). Any theory involving compensation must take into account the fact that the total birefringence (form plus intrinsic) of all oriented actomyosin systems varied greatly with the refractive index of the medium (H. H. Weber, 1934b), whereas the I bands remain practically isotropic

-

a. These workers do, however, claim that after extraction of actomyosin the I bands, which in water are almost isotropic, become strongly negatively birefringent when 1.5. This observation cannot explain the soaked in fluids of refractive index isotropy of normal actomyosin-containing I bands in their aqueous environment, and at best it only explains the isotropy in media of refractive index 1.5. Moreover, despite considerable effort, it has been impossible to confirm the observation itself in our laboratory (Hasselbach and Schneider, 1951).

-

-

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

243

in all media. The compensating factor would therefore need t o have a negative birefringence of a magnitude which remains about equal to that of the actomyosin a t all refractive indexes. Such a factor seems improbable. It may be remarked that the second of the assumptions about the structure of the fibrils by no means makes it necessary t o invoke a theory involving compensation. It is certain that the formed elements of the muscle and the models can be functionally changed both in their intrinsic and form birefringence (cf. Sections 111, 3e and IV, 2b), involving evidently a change in the internal structure of the actomyosin complex. It is thus quite possible that the same actomyosin filament could exist in a positively birefringent state in the A bands and in an isotropic state in the I bands. b. The Fibrous Fine Structure of Contracted Muscle. The discrepancy between information given by the electron microscope and studies with polarized light exists also in contracted muscle. The total birefringence of living muscle decreases in contraction (v. Muralt, 1932; Table II), and for the models, it can be shown that the decrease takes place both in intrinsic and form birefringence. Electron microscope pictures, however, show that the protein filaments run just as straight and parallel to the fiber axis in contracted as in relaxed muscle, and the structures responsible for most of the birefringence are not those resolvable in the electron miscroscope, but smaller ones. The only change visible in the electron microscope according to Draper and Hodge (1949) is a shortening of the longitudinal periodicity, which is accurately proportional to the shortening of the whole sarcomere. This would mean that the colloidal change which is responsible for contraction involves particles which are part of the individual filaments and which are too small to be visible in the electron microscope. The situation becomes still more puzzling when the X-ray results are considered. Astbury (1947) has shown, from the excellent diffraction photographs in all stages of contraction, that the normal a-diagram hardly changes a t all during shortening, even when the contracted length is less than half the original. The very small increase in angular dispersion confirms the electron microscope studies that the protein structures in question largely retain their axially parallel alinement, and the fact that the intramolecular a-pattern remains the same is evidence against the folding of individual protein chains. Astbury thus reaches the conclusion that the decisive molecular event in contraction does not take place a t the level of structures which give rise to the large-angle pattern. On the changes occurring in the small-angle pattern there are no very definite data.

244

HANS H. WEBER AND HILDEGARD PORTZEHL

The region of high electron density shifts in contraction from the A bands into the I bands as far as the Z disc, i.e., i t moves about 1p in the space of no more than 0.05 seconds. If the substance were L-myosin, and supposing it to be moved by electrophoresis, the potential difference between the M and Z discs would need t o be of the order of ten millivolts or 100 V./cm. (Table VIII). Diffusion is out of the question since i t is too slow by several factors of ten (Table IX). What exactly does shift from the A bands to the Z discs forming the so-calIed stripes is not known. It is possible that nothing actually moves, and that the differences in electron density in the two states may be due to an alteration in the affinity of the protein for the electron optical stain, the affinity being greatest in the A bands in resting muscle, and greatest near the Z discs in contracted muscle. 3. Changes in the Fibrous Muscle Proteins in Contracture and Fatigue The proteins of the fibril give a different electrophoretic spectrum when they are extracted from muscle in contraction or contracture instead of muscle in the resting state; the peaks of @-myosin(L-myosin), a-myosin (actomyosin) and the y-protein are then almost entirely absent, and instead there is an increase in the y-myosin peak (contractin), which is barely seen in extracts of resting muscle (Table XVI). The same effect is seen in rigor, in halogenoacetate contracture, and when a tetanus is frozen in liquid air (Dubuisson, 1948c, 1950d; Crepax and HBrion, 1950; Crepax et al., 1950). TABLEXVI Variation in the Amounts o j Electrophoretic Components According to the State of the Muscle* (pH 7.1 0.3511 and 40 minutes extraction time) Amount of component as % extracted protein State of muscle Resting Contracted

B 12 0

a

7 4.5

Y

2 13.5

* After Crepax, Jacob, and Seldeslachts. 1950. The disappearance of the a! and 0 peaks is probably due to the inability of the actomyosin t o dissociate when the muscle ATP is exhausted (Section 111, 4a VIa).31 31 Deuticke (1932) first found t h a t the protein content of muscle extracts at pH 7 is lower in fatigue and in contracture. Kamp (1941)showed, and Dubuisson (1947) confirmed t h a t the decrease takes place in the myosin fraction, and the studies of Erdos (1943) indicate that a strict parallelism exists between the magnitude of the effect and the disappearance of ATP.

MUSCLE CONTRACTION

AND FIBROUS MUSCLE PROTEINS

245

Dubuisson, it is true (see Dubuisson and Mathieu, 1950), regards this explanation as insufficient; a- and &myosins do not appear in extracts which are made without delay after the ATP has first been washed out, but they do appear in the usual amounts when ATP is added immediately. The a- and 0-peaks in extracts of fatigued muscle, on the other hand, cannot be augmented in this way (Table XVll). The difference can, however, be explained by assuming nothing more than a denaturation of the actomyosin when the ATP content is low, leading t o a loss of dissociability when A T P is again added. (See Section 111, Bb.) TABLEXVII Effect of ATP on the Yield of Myosin* ( r / 2 = 0.5; 20 minutes extraction time) State of muscle ~

Pretreatment

Extracted (g. myosin/lO g. muscle)

None ATP washed out ATP washed out and fresh added None ATP added

0 43 0 0 4 4 0 08 0 54 0 24 0.24

~~~

Resting Resting Resting Fatigued Fatigued

* After Dubuisson

and Mathieu, 1950.

The appearance of y-myosin, however, cannot so readily be explained.

It could be due to a change in the charge of one of the myosin components; or t o the liberation of a preformed component which in resting muscle is too strongly bound to be extractable; or to the formation of a new substance from other proteins. At present, it is not possible t o decide between these possibilities, but any one of them would indicate clearly that contraction involves protein reactions of which we have little cognizance, and which are not brought out in electron microscope studies. In summary, we may say that the filaments visible in the electron microscope are the same in the A and I bands, but th a t the two differ in birefringence. Since the proportion of the fibril volume occupied by the filaments is uncertain, i t is not yet possible to say how the known amounts of 1,-myosin, actin and actomyosin are distributed amongst the formed elements of the sarcomere. The ultimate structural units of L-myosin and F-actin are not as resolvable in the electron microscope, and the same is probably true of natural actomyosin (cf. Section 111, 4 4 . The functional units of the filaments are certainly not visible in the electron microscope, for on contraction the filaments show no change other than shortening. They are not, however, self-contained contractile complexes, for side by side with contraction there occurs outside, but

246

HANS H . %<EBER A N D HILDEGARD PORTZEHL

between the filaments, a movement of material from the A band towards the 2 disc. The rate of movement is compatible with the movement of protein by an electrical mechanism, but not by simple diffusion. Electrophoresis of extracts of muscle in rest and in contracture shows differences indicative of processes which find no counterpart in electron optical studies. V. GENERAL CONCLUSION In contraction, different kinds of muscle show differences with respect to the amount of tension developed, the maximum shortening, the rate of shortening, and the fuel requirement. These differences not only reflect the varying levels of evolutionary development, but also a considerable adaptation to the performance of special functions. The crossstriated musculature of vertebrates and arthropods is evidently specialized to give in particular a high tension and velocity of shortening, and less to achieve a large degree of shortening. Its especially complicated structure appears to serve exactly this functional specialization, for if by suitable means a living muscle fiber is brought into the highly shortened “delta state’’ then hand in hand with a certain degree of structural disorganization the maximal tension which can be developed falls to about half its former value, contraction becomes very much slower, and the extent of shortening increases from 50 to 80% of the original length (Ramsey, 1947). The skeletal muscle fiber thus comes to resemble closely the smooth retractor penis muscle of the dog, which was investigated by Winton (1926). When the crystalloids and most of the globular proteins and enzymes are removed from the skeletal muscle fiber by extraction with glycerolwater, the tension developed in ATP-contraction remains as high as before, but the maximal shortening is now comparable to that of a muscle fiber in the delta state, and the velocity of shortening is even less. And finally, the thread model of purified, oriented actomyosin develops only a low tension of a few hundred grams per square centimeter, which is of the same order as that for some smooth muscles. The possibility of imitating, by progressive simplification of the striated fiber, all the main stages in the development of muscular contraction, makes it practically certain that the fundamental process is the same throughout. Since the actomyosin thread contracts only on addition of ATP, one is led to the conclusion that the fundamental process is based upon the interaction of actomyosin and ATP. ATP has two effects upon the models. It causes contraction and imparts that degree of plasticity and extensibility which is a prerequisite both for active and passive changes in length. It does not seem possible

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

247

to treat the ATP-contracted state of the models as a new elastic thermodynamic equilibrium, in the way suggested by E. Weber (1846). The general parallelism between ATP breakdown and tension suggests rather th a t contraction is due t o the actual breakdown of ATP. The softening action of ATP, however, is due solely t o the combination of the A T P with actomyosin, for it persists when breakdown and tension are reduced by two-thirds, as when the system is brought to a lower temperature. Cessation of A T P breakdown when ATP is still present is thus the explanation of passive relaxation. I n the unoriented, as distinct from the oriented, actomyosin gel, shrinking (superprecipitation) occurs instead of contraction, but it appears t o be due to the same fundamental process, for it shows the same dependence on experimental conditions. When the actomyosin gel is in solution, however, this fundamental process no longer takes place. The dissociation of actin and myosin is then due to the actual presence and not t o the breakdown of ATP, and moreover, it is affected not by A T P exclusively, as is the contraction, but also by all substances which have a sufficiently high affinity for the -SH center of the myosin. The action of A T P in dissociating the complex in solution, and thus in diminishing the cohesional forces between the two proteins, may rather represent the softening action of ATP discussed above. The path t hat leads to an understanding of the fundamental process of contraction ends a t the syneresis effect given by unoriented actomyosin in gel. Neither the study of the individual purified proteins of the myofibril, nor the brilliant investigations on the fine structure of the fibril, has led t o any well-founded theory as to the nature and mechanism of the structural changes which take place in the contractile particles.

REFERENCES Abbott, C., and Ritchie, J. M. (1948). J . Physiol. 107,5. Amberson, W., Erdos, T., Chinn, B., andLudes, H. (1949). J. Biol. Chem. 181,405. Ardenne, M. v., and Weber, H. H. (1941). Kolloid-2. 97, 322. Astbury, W. T. (1941). Nature 147, 696. Astbury, W..T. (1947). Proc. Roy. SOC.London B134, 303. Astbury, W.T. (1948). Proc. 6th Intern. Congr. Exptl. Cytol. 234. Astbury, W. T., and Dickinson, S. (1935). Nature 136,95 and 1765. Astbury, W. T., and Dickinson, S. (1940). Proc. Roy. SOC.London Bl29, 307. Astbury, W. T., Perry, S. V., Reed, R., and Spark, L. C. (1947). Biochim. et Biophys. Acta 1, 379. Astbury, W. T., Reed, R., and Spark, L. C. (1948). Biochem. J . 43,282. Bailey, K. (1939). Biochem. J. 33, 255. Bailey, K. (1942). Biochem. J . 36, 121. Bailey, K. (1944). Advances i n Protein Chem. 1, 289. Bailey, K. (1946a). Nature 167, 368.

248

HANS H . WEBER A N D HILDEGARD PORTZEHL

Bailey, K. (1946b). Progr. Repts. Chem. SOC.43, 283. Bailey, K. (1948). Biochem. J . 43, 271. Bailey, K. (1951). Biochem. J . 49,23. Bailey, K.,and Perry, S. V. (1947). Biochim. el Biophys. Acta 1, 506. Bailey, K., Gutfreund, T. H., and Ogston, A . G. (1948). Biochem. J. 43, 279. Balenovic, K., and Straub, F. B. (1942). Studies Inst. Med. Chem. Univ. Szeged 2, 17. Banga, 1. (1942). Studies Inst. Med. Chem. Univ. Szeged 1, 27. Panga, I. (t943). Studies Inst. Med. Chem. Univ. Szeged 3, 64. Banga, I., and Szent-Gyorgyi, A. (1942). Studies Inst. Med. Chem. Uniii. S z q e d 1, 5.

,. 3, Ranga, I., and Rzent-Gyorgyi, A. (1943). Studies Inst. Med. Chem. U n i ~ Szeged 72.

Bate-Smith, E. C. (1937). Proc. Roy. SOC.London Bl24, 136. Bate-Smith, E. C. (1938). Rept. Food Invest. Bd. Dept. Sci. I n d . Research (Brit.) 22.

Bate-Smith, E. C., and Bendall, J. R. (1947). J . Physiol. 106, 177. Bate-Smith, E. C., and Bendall, J. R. (1949). J . Physiol. 110,47. Bear, R. S. (1944). J . A m . Chem. Sac. 66,2043. Bear, R.S. (1945). J . A m . Chem. Sac. 67, 1625. Bergold, G. (1946). Z . Naturforsch. 1, 100. Bcrgold, G.,and Schrumm, G. (1947). 2. Naturforsch. 2b, 108. Bcrgold, G.,Portzehl, H., and Weber, H. H. (1945), unpublished. Binkley, F. (1945). Science 102,477. Biro, N. A., and Szent-Gyorgyi, A. G. (1949). Hung. Acta Physiol. 2, I . Boehm, G., and Weber, H. H. (1932). Kolloid-2. 61,269. Borbiro, H., and Szent-Gyorgyi, A. (1949). Biol. Bull. 96, 162. Buchthal, F. (1942). Kgl. Danske Videnskab. Selskabs Biol. Med. 2, 1. Buchthal, F., Deutsch, A., Knappeis, G., and Petersen, A. (1947). Acta Physiol. Scand. 13, 167. Buchthal, F., Deutsch, A., Knappeis, G., and Munk-Petersen, A. (1948). Nature 162,965. Buchthal, F., Deutsch, A., Knappeis, G., and Munk-Petcrscn, A. (1949). Acta Physiol. Scand. 16,326. Buchthal, F., Knappeis, G., and Lindhard, J. (1936). Scand. Arch. Ph,ysiol. 73, 162. Biirgermeistcr, E., and Schauenstein, E. (1949). Monatsh 80, 310. Cigada, M., Cittcrio, P., Ranzi, S., and Tosi, 12. (1948). Experientia 4, 480. Crcpax, P., Jacob, J., and Seldcslachts, *J. (1950). Riochim. et Riophys. Acta 4,410. Crepax, P., and Hbrion, A. (1950). Biochim. et Biophys. A d a 6,54. Csapo, A. (1949). Nature 164, 102. Csapo, A. (1950). A m . J . Physiol. 160, 40. Dainty, M., Kleinzellcr, A., Lawrence, A. S. C., Miall, M., Ncedham, D., Needham, J., and Shih-Chang-Shen, (1944). J . Gen. Physiol. 27, 355. Deuticke, H. J. (1932). 2. physio2. Chem. 210, 97. Draper, M. H., and Hodgc, A. J. (1949). Australian J . Ex@. Bid. Med. Science 27, 465. Dubuisson, M. (1941). Arch. Intern. Physiol. 61, 133. Dubuisson, M. (1946a). Ezperientia 2, 412. Dubuisson, M. (19461~). Ezperientia 2, 258. Dubuisson, M. (1947). Experientia 3, 372. Dubuisson, M. (1948a). h a d . roy. BeEg. bull. class sci. Ser. 6, 34, 978.

MUSCLE CONTRACTION

AND FIBROUS

MUSCLE PROTEINS

249

Dubuisson, M. (194813). Experientia 4, 437. Dubuisson, M. ( 1 9 4 8 ~ ) . Proc. 6th Intern. Congr. Exptl. Cytol. 257. Dubuisson, M. (1950a). Proc. Roy. Soc. London B137, 63. Dubuisson, M. (1950b). Biochim. et Biophys. Acta 4, 25. Dubuisson, M. (1950~). Biochim. et Biophys. Acta 6,426. Dubuisson, M. (1950d). Bid. Revs. 26, 46. Dubuisson, M. (1950e). Experientia 6,269. Dubuisson, M. (1950f). Biochim. et Biophys. Acta 6,489. Dubuisson, M., and Fabry-Hamoir, C. (1950). Experientia 6, 102. Dubuisson, M., and Hamoir, G. (1943). Arch. Intern. Physiol. 63, 308. Dubuisson, M., and Mathieu, L. (1950). Experientia 6, 103. Dyer, W.J., French, H. V., and Snow, J. M. (1950). J . Fisheries Research Board Can. 7 , 585. Ebner, V. von. (1916). Pflugers Arch. ges. Physiol. 163, 179. Edsall, J. T. (1930). J . Biol. Chem. 89,289. Edsall, J. T., and Mehl, J. W. (1940). J . Biol. Chem. 133,409. Engelhardt, W. A. (1946). Advances in Enzymol. 6, 147. Engelhardt, W.A., and Ljubimova, M. N. (1939). Nature 146, 668. Erdos, T. (1943). Studies Znst. Med. Chem. Univ. Szeged 3, 51. Erdos, T., and Snellman, 0. (1948). Biochim. et Biophys. Acta 2, 642. Fabry-Hamoir, C. (1950). Biochim. et Biophys. Acta 4, 445. Fenn, W.0. (1923). J . Physiol. 68, 175. Fenn, W.0. (1924). J . Physiol. 68,373. Feuer, G., Molnar, F., Pettko, E., and Straub, F. B. (1948). Hung. Acta Physiol. 1, 150. Fischer, E. (1944). J . Cellular Comp. Physiol. 23, 113. Fischer, E. (1947). Ann. N . Y . Acad. Sciences 47, 783. Gasser, H., and Hill, A. V. (1924). Proc. Roy. SOC.London B96, 398. Gerendtts, M. (1942). Studies Znst. Med. Chem. Univ. Szeged 1, 47. Gerendhs, M., and Matoltsy, A. G. (1948). Hung. Acta Physiol. 1, 124, 128. Godeaux, J. (1945). Bull. soc. roy. sci. Liege 100,216. Greenstein, J. P., and Edsall, J. T. (1940). J . Bid. Chem. 133, 397. Guba, F. (1942). Studies Znst. Med. Chem. Univ. Szeged 2, 4. Guba, F. (1943). Studies Inst. Med. Chem. Univ. Szeged 3, 40. Guba, F., and Straub, F. B. (1943). Studies Znst. Med. Chem. Univ. Szeged 3, 49. Hall, C. E., Jakus, M. A,, and Schmitt, F. 0. (1945). J . Applied Phys. 16, 459. Hall, C. E., Jakus, M. A,, and Schmitt, F. 0. (1946). Biol. Bull. 90,32. Hamoir, G. (1947). Ezperientia 3, 498. Hamoir, G. (1949). Bull. SOC. chim. b i d . 31, 118. Harris, E. J. (1950). Personal communication by A. V. Hill. Hasselbach, W. (1950). Not yet published. Hasselbach, W., and Schneider, G. (1951). Biochem. 2. 321,462. Haxton, H. A. (1944). J . Physiol. 103, 267. Heinz, E., and Holton, F. (1950). Unpublished. Hermann, V. S., and Josepovits, G. (1949). Nature 164, 845. Hill, A. V. (1926). Proc. Roy. SOC.London B100, 108. Hill, A. V. (1938). Proc. Roy. SOC.London B126, 136. Hill, A. V. (1939a). Proc. Roy. SOC.London Bl27, 434. Hill, A. V. (1939b). Proc. Phys. SOC.London 61, 1. Hill, A. V. (1949a). Proc. Roy. SOC.London B136, 399 Hill, A. V. (1949b). Proc. Roy. SOC.London B136, 420.

2 50

HANS H. WEBER AND HILDEGARD PORTZEHL

Hill, A. V. (1949~). Proc. Roy. Soc. London B136, 210. Hill, A. V. (1950a). Nature 166,415. Hill, A. V. (1950b). Proc. Roy. Soc. London B137, 268. Hill, A. V. '(1950~). Proc. Roy. Sac. London B137, 320. Hill, A. V. (1951). Nature 167, 372. Hill, D.K. (1949). J. Physiol. 108, 292. Hoffmann-Berling, H., and Kausche, G. A. (1950). Z. Naturforsch. 6b, 139. Hollwede, W., and Weber, H. H. (1938). Biochem. Z. 296, 205. Hole, B. (1932). Pflugers Arch. ges. Physiol. 230, 246. Hiirthle, K. (1930). Pflugers Arch. ges. Physiol. 223, 685. Jacob, J. (1945). Bull. soc. roy. sci., Liege 3, 100. Jaisle, F. (1951). Biochem. 2. 321,451. Jakus, M. A., and Hall, C. E. (1947). J . Biol. Chem. 167,705. Johnson, P., and Landolt, R. (1950). Nature 166, 430. Jordan, W.K., and Oster, G. (1948). Science 108, 188. Josenhans, W. (1949). 2. Biol. 103, 61. Kamp, F. (1941). Biochem. 2. 307,226. Kaumans, H. (1949). Doktordissertation, Universitat Tubingen. Kleinzeller, A. (1942). Biochem. J. 36, 729. Knappeis, G. (1948). Abstr. Commun. 6th Scand. Physiol. Congr. Oslo. Korey, S. (1950). Biochim. et Biophys. Acta 4,58. Kuhn, W. (1936a). Kolloid-Z. 76,258. Kuhn, W. (1936b). Naturwissenschaffcn 24, 346. Kuschinsky, G.,and Turba, F. (1950a). Erperientia 6, 103. Kuschinsky, G.,and Turba, F. (1950h). Naturwissenschaften 37, 425. Lajtha, A. (1948). Hung. Acta Physiol. 1, 134. Laki, K.,Bowen, W., and Clark, A. (1950). J. Gen. Physiol. 33, 437. Levin, A., and Wyman, J. (1927). Proc. Rov. SOC.London B101, 218. Lundi, G. (1944). A d a Physiol. Scand. 7, 86. Lundsgaard, E. (1950). Proc. Roy. SOC.London B137, 40. MacArthur, I. (1943). Nature 162, 38. Menne, F. (1943). 2.physiol. Chem. 279, 105. Meyer, K. H., Susich, G. v., and Valko, E. (1932). Kolloid-Z. 69,208. Meyerhof, 0. (1923). Pjlugers Arch. ges. Physiol. 196, 22. Meyerhof, 0. (1924). Pjlugers Arch. ges. Physiol. 204, 295. Meyerhof, 0. (1930). Die chemischen Vorgange im Muskel. J. Springer, Berlin. Michaelis, L. (1922). Die Wasserstoffionenkonzentration I. J. Springer, Berlin. Michaelis, L., and Menten, M. L. (1913). Biochem. Z. 49, 333. Mommaerts, W. F. H. M. (1945). Arkiv Kemi Mineral. Geol. 19, No. 17,l. Mommaerts, W. F. H. M. (1948). J. Gen. Physiol. 31, 361. Mommaerts, W. F. H. M. (1950). Biochim. et Biophys. Acta 4, 50. Mommaerts, W. F. H. M. (1951a). J . Biol. Chem. 188, 559. Mommaerts, W. F. H. M. (1951b). J. Biol. Chem. 188, 553. Mommaerts, W. F. H. M. (1951~). Exptl. Cell Research 2, 133. Mommaerts, W. F. H. M., and Parrish, R. G. (1951). J . B i d . Chern. 188, 545. Mommaerts, W. F. H. M,, and Seraidarian, K. (1947). J . Gen. Phylsiol. 30, 401. Muralt, A. von. (1932). PfEugers Arch. ges. Physiol. 230, 299. Muralt, A. von, and Edsall, J. T. (1930). J . Biol. C'hem. 89, 315 and 351. Needham, D. R4. (1942). Biochemical y.'36, 113. Needham, J., Shih-Chang-Shen, Needham, D. M., and Lawrence, A. S. C. (1941). Nature 147, 766.

MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS

251

Noll, D., and Weber, H. H. (1934). PJugers Arch. ges. Physiol. 236, 234. Oster, G., Doty, P. M., and Zimm, B. M. (1947). J . A m . Chem. SOC.69, 1193. Pease, D. C., and Baker, R. F. (1949). A m . J . Anat. 84, 175. Perry, S. V., Reed, R., Astbury, W. T., and Spark, L. C. (1948). Biochim. et Biophys. Acta 2, 674. Polis, D., and Meyerhof, 0. (1947). J . Biol. Chem. 169, 389. Portzehl, H. (1950a). Z. Naturforsch. 6b, 75. Portzehl, H. (1950b). Unpublished. Portzehl, H. (1951). 2. Naturforsch. 6b, 355. Portzehl, H. (1952). 2. Naturforsch. 7b, 1. Portzehl, H., and Weber, H. H. (1950). Z. Naturforsch. 6b, 123. Portzehl, H., Sehramm, G., and Weber, H. H. (1950). Z . Naturjorsch. 65,61. Ralston, H. Y., Inman, V. T., Strait, L. A,, and Shaffrath, M. W. (1947). Am. J . Physiol. 161, 612. Ralston, H. Y., Polissar, M. J., Inman, V. T., Close, J. R., and Feinstein, 13. (1949). J . Appl. Physiol. 1, 526. Ramsey, R. W. (1947). Ann. N . Y . Acad. Sci. 47, 110. Ramsey, R. W., and Street, S.F. (1940). J . Cellular Comp. Physiol. 16, 11. Ramsey, R. W., and Street, S. F. (1941). Biol. Symposia 3, 9. Reay, G. A., and Kuchel, C. C. (1936). Dept. Sci. Ind. Research (Brit.) Uepts. Food Invest. 93. Roth, E. (1946). Biochem. Z. 318, 74. Rozsa, G., and Staudinger, M. (1948). Makromolek. Chem. 2, 66. Rozsa, G., Szent-Gyorgyi, A., and Wyckoff, R. W. G. (1949). Biochim. et Biophys. Acta 3, 561. Ilozsa, G., Szent-Gyorgyi, A,, and Wyckoff, R. W. G. (1950). Exptl. Cell Research 1, 194. Sandow, A. (1944). J . Cellular Comp. Physiol. 24, 221. Sandow, A. (1947). Ann. N . Y . Acad. Sci. 47, 895. Sandow, A. (1949). Ann. Rev. Physiol. 11, 297. Sandow, A,, and Brust, M. (1946). Proc. SOC.Exptl. Biol. Med. 63, 462. Sehafer, H., and Gopfert, H. (1941). PJEugers Arch. ges. Physiol. 246, 60. Sehauenstein, E., and Treiber, E. (1950). J . Polymer Sci. 6, 145. Schick, A. F., and Ham, G. M. (1949). Science 109, 486. Schmidt, W. J. (1934). 2. Zelljorsch. 21, 224. Schmitt, F. O., Bear, R. S.,Hall, C. E., and Jakus, M. A. (1947). Ann. N . Y . Acad. Sci. 47, 799. Schramm, G., and Weber, H. H. (1942). Kolloid-2. 100, 242. Sehulz, G. V. (1947). 2. Naturforsch. 2a, 348. Sichel, F. J. (1935). J . Cellular Comp. Physiol. 6, 21. Signer, R., and Gross, H. (1934). Helv. Chim. Acta 36, 726. Simha, It. (1940). J . Phys. Chem. 44, 25. Singer, T. P., and Barron, E. S. G. (1944). Proc. SOC.Exptl. Biol. Med. 66, 120. Singher, H. O., and Meister, A. (1945). J . Biol. Chem. 169, 491. Snellman, O., and Erdos, T. (1948a). Biochim. el Biophys. Acta 2, 650. Snellman, O., and ErdBs, T. (1948b). Biochim. et Biophys. Acta 2, 660. Snellman, O., and Erdos, T. (1949). Biochim. et Biophys. Acta 3, 523. Snellman, O., Erdos, T., and Tenow, M. (1949). Proc. 6th Intern. Congr. Exptl. Cytol. 247. Snellman, O., and Gelotte, B. (1950). Exptl. Cell Research 1, 234. Snellman, O., and Tenow, M. (1948). Biochim. et Biophys. Acta 2, 384.

252

BANS

n.

WERER AND HILDBGARD PORTZEHL

Spicer, S. (1949). Arch. Biochem. 26, 369. Straub, F. B. (1942). Studies Inst. Med. Chem. IJniv. Szeged 2, 3. Straub, F. B. (1943a). Studies Inst. Med. Chem. IJniv. Szeged 3, 23. Straub, F. B. (1943b). Studies Inst. Med. Chenk. Univ. Szpged 3, 38. Straub, F. B., and Feller, G. (1950). Biochim. el Biophys. Acta 4, 455. Strauh, F. B., Feuer, G., and Lajos, I. (1948). Nature 162, 217. Strobel, G. (1952). 2. Naturforsch. 7b, 102. Summerson, W. H., and Meister, A. (1944). Ahtracts Div. Biol. Chern., 108t)h Meeting, Am. Chem. SOC., 42B. Szent-Gyorgyi, A. (1942). Studies Inst. Med. Chern. Univ. Sreged 1, 17. Szent-Gyorgyi, A. (1943). Studies Inst. Med. Chem. CJniv. Sreyed 3, 76. Szent-Gyijrgyi, A. (1947). Chemistry of Muscular Contraction. Academic Press, New York. Szent-Gyorgyi, A. (1949). Biol. Bull. 96, 140. Szent-Gyorgyi, A. (1951). Chemistry of Muscwlar Contraction, 2nd ed. Academic Press, S e w York. Szent-Gyorgyi, A., and coworkers. (1942). Studies Inst. Med. Chem. Univ. Szeged 1 arid 2. Szent-Gybrgyi, A . , and coworkers. (1943). Studies Inst. Med. Chem. Univ. Szeged 3. Szent-Gyorgyi, A., and coworkers. (1944). Studies Inst. Med. Chem. Univ. Szeged. Tsao, T. C., Bailey, K., and Adair, G. S. (1951). Biochem. J . 49, 27. Turba, F., Kuschinsky, G., and Thornann, H. (1950). ~VatzLrwissrrLschaften37, 453. Varga, L. (1946). Hung. Acta Physiol. 1, 1 . Warhurg, O., and Christian, W. (1832). Biocherrr. Z. 264, -138. Weber, A. (1949). linpublished. Webcr, A. (1951). Biochim. el Biophys. Acta 7, 214. Weber, A,, arid Weber, H. H. (1950). 2. Naturforsch. 6b, 124. Weber, A., and Weber, H. H. (1951). Biochim. et Biophys. Acta 7, 339. Weber, E. (1846). “ Muskelbewegung ” Handwiirterbuch der Physiologie. Bd. 3, Teil 2, 110, Braunschweig. Weher, H. H. (1934a). Ergeb. Physiol. 36, 103. Weher, H. H. (1934h). PfEugers Arch. ges. Physiol. 236, 205. Wefier, H. €1. (1939). ~ ~ t ~ r ~ ~ ~ e n s c27, h a33. ften Weher, H.H. (1947). FIAT Review (Naturw. und Mediz. in Deutschland 19391946) Ilietrich’sche Verlagsbuchlmdlung Wieshaderl. “ Physiologie,’’ Teil 3 , Abschn. “ Muskel ” Seite 1 . Weher, IT. H. (1950a). Biochim. rt Biophys. A d a 4, 12. Weher, H.H. (1950b). Proc. Roy. Soc. London, B137, 50. Weber, 11. H. (1!350c). 16th Intern. Physiol. Congr. (Copenhagen),62. Weher, H. H., and Meyer, K. (1933). Biochem. 2. 266, 137. Weber, H. H., and Portzehl, €1. (1949). Makroniolek. Chem. 3, 132. Weher, 1%.H., and Sttiver, R. (1933). Biochem. Z. 269, 269. Wiegand, W. H . , and Snyder, I. W. (1934). Trans. Inst. Iiubber Ind. 10, 234. Winton, F. R. (1926). J . Physiol. 61, 868. Wirtz, K. (1947). Z.Naturforsrh. 2b, 94. Wohlisch, 15. (1926). C’erhandl. phys. nted. Ges. Wurzburg N.F. 61 and 63. Wohliuch, E:. (1940). Naturiuissenschuften 28, 305 and 326. Wohlisch, E., and Gruning, W. (1943). PJlugers Arch. ges. Physiol. 246, 469. Wolpers, C. (1944). Deut. med. Wochschr. 29/30, 495. Ziff, J. (1944). J . B i d . Chem. 26, 153.

The Proteins of the Mammalian Epidermis BY K . M . RIJDA1.T. 1)epartiiieiil of Bioniolecrdar Structure. The 1'nii.ersil.y. Leeds. England

CONTENTS Page 253 11. Properties of the Epidermis as a Whole . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254 1 General Structure . . . . . . . . . . .......................... 254 2 . Histochemistry of Sulfur-Con o Acids . . . . . . . . . . . . . . . . . . . . 255 ......................... 258 3 . X-ray Absorption and Sulfur 4. Ribonuclcic Acid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259 5 . Distribution of Cell IXvision., . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259 ............................................... 259 udies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260 8. X-Ray 1)iffractiori Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 9 . Temperature and Stability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262 111. The Extraction of Epidermal Proteins. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264 1 . Solubility of Cell Structure Proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264 2. Extraction of Epidermal Protein . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266 3 . Purification of Epidermal Protein . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267 4 . Molecular Weight and Asymmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268 5. Sulfur Content of the Isolated Proteins ......................... 269 IV . Some Physical Properties of Epidermal Proteins. . . . . . . . . . . . . . . . . . . . 270 1 . X-Ray Diffraction of the Fibrous Protein.. . . . . . . . . . . . . . . . . . . . . . . . 270 2 . X-Ray Diffraction of the Nonfibrous Protein., . . . . . . . . . . . . . . . . . . . . . . 272 3 . The Effect of Temperature on Various Preparations of Epidermal Protein 272 4. Cross p Form of a-Type Proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274 5 . Reversibility of Cross B Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277 6. Elastic Properties of Epidermin Films and Fibers . . . . . . . . . . . . . . . . . . . . . 277 V . Infrared Ahsorption Studies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279 1 . General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279 2 . Oriented Films of 01 Epidermin., . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280 3 . Transformation to 0 Epidermin by Stretching . . . . . . . . . . . . . . . . . . . . . . . . 282 283 4 . Heat Denaturation and Infrared Absorption . . . . . . . . . . . . . . . . . . . . . . . . . 5. The Oriented Cross 6 Structure., . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285 VI . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288

I . IntroductioIi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.

1. INTRODUCTION In studying the proteins of muscle one has always in the background the question-how does this new knowledge help t o explain the mechanism of muscular contraction? With the proteins of nervous tissue the 253

254

K . M. RUDALL

question always at hand is-are me learning any more about the conduction of nerve impulses:? The study of epidermal proteins is not sustained by such dramatic or popular questionings. Nevertheless the problems to be answered are interesting and important. Many of these concern the structure of proteins, a matter which W. T. hstbury has said may make a man rather quiet, with the rest we seek to know more ol what goes on during the development of cells and in their various stages of differentiation and decline. As in other epithelia serious afflictions affect the epidermis and the mood of our studies might be to overwhelm our ignorance as well as to gratify our curiosity. We are a long way from being overwhelming but we can start t o ask simple questions. A first question about the epidermis is-what is the nature of and the condition of the proteins in the three principal levels, i.e., the level where cell division is a maximum, the level where mitosis is a minimum, and the layers of dead cornified cells. Given the correct answers we should know the mechanical properties and role of the cell structure proteins and how these change upon cell differentiation. We should be able t o understand the variation of epidermal texture over the surface of a n individual and that occurring in the various classes of animals. As a product of our efforts we should like to gain some very broad ideas on the relationships of epidermal proteins with those of other cell types, first of all with muscle proteins because these are comparatively well known, but in the course of time with as full a range as possible. From our concepts about protein molecules i t seems likely that there are a few basic types which are modified t o the needs of each tissue. So the interest that lies in a study of epidermal proteins is concerned with the comparative structure and organization of cell proteins rather than with specially enhanced properties such as contraction or conduction. I n this article we begin by considering that knowledge of the epidermis as a whole which is likely to be helpful in our main quest. Then we describe the extraction of the principal cell proteins with an investigation of some of their chief properties. We are probably a long way from uncovering the best secrets; nevertheless, some first shape is given t o important aspects of the subject.

11. P R O P E R T I E S

O F THE

EPIDERMIS AS

A WHOLE

1. General Structure The epidermis is a very simple tissue composed entirely of cells. Although dermal papillae do penetrate into it from below it is rcadily (or potentially) detachable from these connective tissue elements.

THE: PROTEINS O F THE MAMMALIAN EPIDERMIS

255

Almost the whole is composed of the one kind of unit, i.e., the true epidermal cells, while there is a small number of dendritic cells which may or may not produce pigment (18). The tissue is thus simple in the lack of a variety of component cells and because it can be detached from connective tissue elements. The epidermis is complex, however, from the chemical viewpoint because of the different states in which its cells exist a t different levels. Three main levels may be considered which we can designate a s inner, intermediate, and outer. The inner level contains the cells which are rapidly dividing and synthesizing new cytoplasmic protein, the intermediate level represents a state where these generative processes are less conspicuous and where changes are taking place in preparation for the formation of the outer layer. The outer layer itself is the finally stabilized or cornified layer consisting of dead cells, i.e. the stratum corneum. The first two layers together constitute the stratum mucosum. The chemicasl complexity concerns the different state of the structure proteins a t the various levels arid also changes in the reactivity of the intracellular metabolic systems. 2. Histochemistry of Sulfur-Containing Amino Acids

A guide t o what we have to discover about epidermal proteins is given by the principal histochemical findings. Most would prefer t o regard intracellular reactions as explained only when the neat series of test tube experiments have given a well checked and repeatable series of results. Hut the histological studies indicate what is t o be sought for and above all give the best answer t o the question, just where is there chemical reactivity of a certain kind. The best known chemical activity in the epidermis is t ha t of the oxidation of cysteine t o cystine. Following studies of the importance of cystine in hair keratin, the main outlines of the similar situation in the epidermis were established by Giroud and his coworkers (34, 35). Using the classical nitroprusside test they found the stratum spinosum t o stratum granulosum region t o be rich in free -SH groups while the stratum corneum gave a negative reaction. This was interpreted as the formation of cystine bridges from two cysteine residues, thus joining together polypeptides by strong covalent linkages, and seemed to explain very satisfactorily the hardness and durability of the stratum corneum compared with the softness and instability of the mucosum. The nitroprusside reaction is noteworthy for being highly specific for -SH groups (if carried out in faintly alkaline solutions) but notoriously difficult for defining detail of distribution. The color developed fades quickly; also fairly thick sections are required t o give adequately robust

256

K.

M.

RUDALL

coloring. It is not useful for examining fine detail as a t transition zones where oxidation is taking place. These disadvantages of the otherwise excellent nitroprusside reaction are not present in the case of the Prussian blue method which has been applied to the same problems by Ch h remo n t and Frederic (25). However, the Prussian blue method -’is not specific for -SH groups, but

FIG. 1 .

Verticd section of cow’s nose epidermis. Freeze-drying preparation. Prussian blue reaction throughout the mucosum.

reveals the presence of any reducing groups. I t s success as a test for -SH groups depends on the fact that these are often the only powerful reducing groups present. Caution and many controls are necessary before safe conclusions can be reached. A main objection is the absence of any defined endpoint, so that one has to he suspicious of reducing action orcurring after periods of 30-00 minutes, depending on the kind of material and its pretreatment. Chevremont and Frederic (25) have drawn attention to the great convenience of the Prussian blue reaction in defining the detailed distribution of free -SH groups. Their new discovery in this field was the description of the classical keratohyalin granules as sites of intense -SH activity, and this is of note because in cornifying tissues we have to keep a special watch on the fate of sulfur

T H E P R O T E I N S O F T H E MAMMALIAN E P I D E R M I S

257

atoms. On the other hand reactive -SH groups are commonly associated with cytoplasmic granules in a variety of tissues (23). The Prussian blue reaction was tested on the thick epidermis of the nose of the cow, which is the chief source of epidermal proteins so far studied. Following short period fixation (12 hours) in 5Y0 formaldehyde a t ca. pH 6 and rapid embedding in parafin, only the weakest reaction

FIG. 2. Cow’s nose epidermis, formaldehyde fixation. Reduction in thioglycolic acid. Prussian blue reaction in stratum corneum.

with ferricyanide was obtained (cf. 42). Thus the procedure using this formalin fixation was regarded as a failure. But where paraffin sections were prepared following freeze-drying, a rapid and well defined Prussian blue reaction was obtained. The main features of this are the relatively intense reaction throughout the stratum mucosum with a slightly more intense reaction in several cell layers a t the outer limits of the mucosum. The change t o the stratum corneum is abrupt and confined to individual cells. It appears as though the intercellular region or the cell membrane retains reducing action for some time, while the interior of the cells is

258

K. M. RUDALL

non-reacting. Higher up in the corneum isolated cells or groups of cells frequently show considerable reducing activity while their neighbors arc not reactive. Most of these features are fairly obvious in Fig. 1. The nitroprusside reaction on thicker sections frequently shows that regions of the stratum corneum do have free -SH groups, and th a t there is intensification of the reaction at the outer regions of the stratum mucosum. It is desirable to know whether there are also some -S-Sgroups in the mucosum and whether the disappearance of -SH groups in the corneum is due to the formation of -S-Sgroups there. Some of these matters can be resolved by blocking -SH groups and reducing the -S-Sgroups. Sections from formalin-fixed material in which only the feeblest -SH activity remained were reduced in thioglycolic acid a t pH 7 for 3 hours and thoroughly washed according t o the procedures of Patterson et al. (4).An intense Prussian blue reaction was obtained in the stratum corneum only, with a very feeble reaction in the mucosal layers, Fig. 2. These observations satisfactorily demonstrate the -S-Scondition in the stratum corneum. Some quantitative aspects of these -SH reactions remain puzzling. For example, the appearance of the reduced corneum suggests that there is more -SH activity between the cells or a t the cell surface than there is internally in the cell. Again, there is a n apparent increase in reducing activity a t the outer levels of the untreated mucosum but this may be due t o a decreasing water content there, or may represent in this parakeratotic epidermis the equivalent of the keratohyalin granules which are so obvious in normal epidermis.

3. X-Ray Absorption and Sulfur Content Engstrom and his colleagues have used epidermal material in several of their X-ray absorption studies. General absorption, in a vacuumdried section, is greater in the stratum corneum than in the mucosal layers, leading t o a mass ratio of 1.4 : 1 for these two regions. But when measuring the sulfur by the study of absorption edges a sulfur content ratio oE nearly 8 : 1 is found (32). This seems t o be a n excessively high ratio even allowing for the mucosal layers containing a large amount of connective tissue in the form of dermal papillae-an error easily avoided. If we examine the X-ray absorption data for the corneum, where there is no connective tissue inclusion, we can test whether the order of the results is comparable with that obtained by direct chemical analysis. Accepting the section thickness as a t least near to the 15p given (33), the mass of the corneum tissue works out as 7.8 X 10-10/1500p3 and the sulfur as ca. 0.78 X 10-lo g./150Ops. These give a sulfur content of about 10% which is approximately ten times the measured values for similar material (58). An arithmetical error somewhere may account

T H E PROTEINS OF THE MAMMALIAN EPIDERMIS

259

for this. Then how would the low sulfur mass in the mucosum be accounted for unless by assuming the water here t o have been about 90% before vacuum drying? However, the mass ratio 1.4: 1 of corneum to mucosum does not indicate a very high water-content difference. 4. Ribonucleic Acid

Studies of cytoplasmic basophilia using ribonuclease and methyl green pyronin staining have informed us of the approximate distribution of ribonucleic acids in the different layers of the epidermis. The pioneer observations of Brachet (22) revealed intense basophilia throughout the stratum mucosum, but showing a decreasing gradient from the basal layer t o the stratum corneum. Later studies b y Nolte (45) confirmed this picture and provided detailed illustrations in the case of human toe epidermis. Very noticeable is the comparative absence of basophilia in the outer layers of the stratum mucosum, i.e., in those regions where cell division and protein synthesis have ceased or been greatly reduced. 5. Distribution of Cell Division Earlier workers maintained and some maintain nowadays (24) that mitosis in the epidermis is restricted to the single layer of basal cells in contact with the dermis. Thuringer (54, 55) presented many photographs and an extensive series of measurements which seemed to show that mitosis is widely distributed throughout the stratum mucosum. According t o these results the lower half or two-thirds of the mucosum is a region of comparatively high cell division rate, while the outer third or half is markedly different, mitosis being relatively infrequent. We have t o bear these differences in mind when considering possible changes in the condition of the cell structure proteins a t different levels of the mucosum. Related to this gradient of cell division rate there is an increasing apparent size of the cells as we pass from the basal layer to the outer levels of the mucosum. This size change has often been figured (21) but no measurements seem to have been made. 6. Amino Acids

Active studies of the basic amino acids in the epidermis in the 1930’s sought t o make a distinction between the hard keratin (eukeratin) of hair, horn, nails etc., and the soft keratin (pseudokeratin) of the epidermis, horse burrs, and whalebone (20). The results of independent workers like Block (19) and Eckstein (29) were in agreement in the case of pepsin- or trypsin-treated epidermis where the ratios of histidine: lysine :arginine residues were approximately 1 :3 :3. These results were in marked contrast t o the earlier work of Wilkinson (57) who found a

2G0

K . M. RUDALL

ratio of 1:6 : 15 in the case of exfoliative dermatitis material which had not been pretreated with proteolytic enzymes. Wilkinson’s results were characteristic. of those obtained from the harder keratins such as hair or horn. We do not know whether Wilkinson’s analyses were incorrect or whether it is true, as his figures suggest, that hair keratin and epidermal keratin can have a very similar content and ratio of basic amino acids. In more recent times the keratin and the carcinomas derived from rabbit papillomas (Shope) have been shown to have approximately the 1:3 :3 ratio characteristic of pseudokeratins (51). At the time these estimates of the basic amino acids seemed a useful contribution t o make. A classification of proteins as closely or distantly related is still a desirable goal, but it is fairly certain th a t proteins having basic amino acid ratios of 1:3 :3 and 1 :6 : 15 can be more closely related than two proteins which may both have the same ratio. The efforts did show t ha t the enzyme-resistant protein in the epidermis differed from the similarly resistant protein of the hard keratins in containing relatively much less arginine. They also showed th a t the ratios for the whole structure did not differ significantly from the ratio for the very small residue left after enzyme treatment. A modern approach would attempt a complete analysis of all amino acids and preferably only on well defined protein species within these structures.

7 . Sulfur Content Studips By far the most interesting chemical facts about the epidermis are those giving the sulfur content and approximate percentages of cystine and methionine. Measurements for a variety of epidermal tissues are listed in Table T. Most of the later values for the sulfur content of true epidermis lie in the range 1.1-1.2%. The value found for the corneum of the cow’s nose is rather higher, but there is always the possibility of short hairs having been included in the samples. However, such a high value was not obtained in four determinations made in connection with this present study, viz. (1.12, 1.14) and (1.14, 1.18), the material being the whole rorneum together with a few adhering cells of the outer mucosum. The figures suggest th at a considerable proportion of the sulfur may he present as methionine, the epidermis being comparable with fibrinogen where cystine and methionine are about equal in quantity (56). Though the measurements are as yet few it is probably very important to pay close attention to the relative quantities of cystine- and methioninesulfur in different types of epidermis. Compared with the hard ketatins like hair, wool, etc., the epidermal keratins show three main differences. The cystine and arginine contents

20 1

THE PROTEINS O F T H E MAMMALIAN EPIDERMIS

TABLEI Sulfur, Cystine and Methionine in the Epidermis

Material and treatment

Cystine Methionine Total % % S % Ref. and date

Human epidermis 1.8-2.3 Horse burrs, eorneum miirosum 2.31 Human epidermis, normal 3.80 pepsin / t r y p i n 3.40 pepsin 2.38 normal 2.51 pepsin/trypsin Cow’s nosc, corneiini 4.80 mucosiim 3.60 Cow’s nose, eorneum few outer layers of mucosum Same, extracted in 6 M urea 23 days

0.70 0.50 0.49

2.47 2 GO

+

1.09 1.I0 1.41 1.15 1.13 1.16

1927 (59) 1834 (34) 1934 (57) 1935 (29) 1935 (19) 1939 (58) 1937 (28)

1950 (Author unpublished)

are very much lower, while the methioriirie seems to be much more abundant. All these measurements are of limited meaning until we can find the cwnstitution of the component molecular species. 8. X-Ray Diflraction Studies

I n the early 1930’s Astbury and his collaborators defined the principal features of the molecular structure of essentially all mammalian hard keratins (4, 6). This work introduced the concept of the regularly folded a-protein chain. Within this scheme of things the hard keratins of birds and reptiles stood out, for they gave highly characteristic diffraction patterns which could only be interpreted in terms of a special type of p or nonfolded chain (5). Nevertheless, much of the epidermal protein in birds and reptiles has the same a-type structure as is always found in the case of mammalian epidermal tissue. The special interest of these observations was t o reveal a widespread common type of molecular structure in the principal fibrous proteins of vertebrate epidermis. This common type is not quite universal because of the seeming mutation in the hard keratins of birds and reptiles, which defines the unique relationship of these groups a t the molecular level (5, 50). For mammalian epidermal tissues Giroud and Champetier (36) and Derksen and Heringa (27) found the a-diffraction pattern to be characteristic of the stratum corneum and also of the mucosum which produces it, and both related the diffraction pattern t o the observed tonofibrils. Derksen and Heringa succeeded in transforming the a-pattern of their material (cow’s nose epidermis) t o the ,&pattern by stretching in hot

262

I

ern.-'

FIG.14. Infrared absorption spectra of films of epidermin. A, stretched (200% extension from cast film). a-form. U, same as A but contracted by 42% in cold water; lesser orientation of a-form. C, partial conversion to p-form hy high extension in saturated ammonium sulfate. _ _ Electric vector parallcl to axis of orientation. -... . . Electric vector perpendicular to axis of orientation.

those of other natural a-type proteins (2, 31). Even in the stretched film, Fig. 14A, there is no certain contribution due t o the fi form absorption a t 1630 cm.-l If there is some development of fi chains in this extended film the only evidence for it is the very slight parallel dichroism of the XH deformation about 1520 cm.-'

282

K. M. RUDALL

The dichroic ratio of the main N H absorption a t 3280 crn.-l in Fig. 14A is 1.8/1.* For the contracted epidermin, Fig. 14B, the corresponding dichroic ratio is 1.5, so that in an extension of 70% there is a change of dichroic ratio from 1.5 to 1.8. This amount of change has no clear meaning until we know more about the bonds th a t contribute to these figures, i.e. whether they all have a similar orientation or whether there are several differently oriented bonds per unit of the fiber structure. But we may suspect that a main cause of this reversible 70% extension is due to the formation of p chains. This, however, must involve relatively few chains as there is no very good evidence for j3 structure, either on the grounds of dichroic effects or changes of frequency. This draws attention t o a possible limitation of the infrared absorption studies compared with the study of elastic properties. I n biological systems we may often be concerned with a! -+ changes involving less than 1% of all chains present, which nevertheless have a large effect on the properties of the whole material. Apart from the difference in dichroic ratio, the total absorption of the N H and C=O stretching bands of the contracted film is less than th a t of the stretched film. This change is almost certainly due to a closer parallelism of the molecular chains with the plane of the film surface i n the stretched specimen. It has been noted above th a t the contracted films appear t o be thicker as judged from the amount of absorption a t the C H bands, and this would most likely be due to chains becoming less nearly parallel with the plane of the surface.

3. Transformation to p Epidermin by Stretching As in the case of fibers of epidermin, thin films can be stretched to the form in hot saturated ammonium sulfate. Films first oriented by stretching in cold water were dried and stretched t o 92% extension of the oriented length, i.e. a total extension from the original cast film of nearly 500%. They were fixed in formaldehyde, prior to washing away the ammonium sulfate, in order to prevent 0 -+ a! change. The infrared spectra for these highly stretched films are shown in Fig. 14C. The dichroism has changed so that maximum absorption of the main NII stretching bands is now obtained with the electric vector vibrating perpendicular t o the fiber axis. This suggests th a t the chains are mainly in the /3 configuration, but it is unlikely that the dichroisms of a! and 0 absorption are equal and opposite. Of particular interest is the appearance of a second C=O stretching * T h e dirhroic ratio is the ratio of the optical density with the electric vector along the fiber axis to the optical density with the electric vector at right angles to the fiber axis.

T H E P R O T E I N S O F T H E MAMMALIAN E P I D E R M I S

283

band at 1630 cm.-', in addition to that a t 1655 cm.-' The band at 1655 cm.-l shows parallel dichroism (associated with the a configuration) and the band a t 1630 cm-' shows perpendicular dichroism (associated

cm.-l

FIG. 15. Infrared absorption spectra (unpolarized radiation) of epidermin films showing the progress of heat denaturation. Films were heated in hot water for various times: A, not heated; B, 68"C., 2 minutes; C, 80"C., 2 minutes; D, lOO"C., 2 minutes; E, lOO"C., 30 minutes; F, lOO"C., 270 minutes. Curves G is spectrum of epidermin highly stretched in saturated ammonium sulfate.

with the fl configuration) (31). In Fig. 1 4 c the NH deformation band shows changes in dichroism and frequency which are associated with the 6 form. While there is a considerable amount of /3-type absorption in Fig. 14C (see the spectrum using unpolarized radiation-Fig. 15G) the a-type appears to be the major component. The X-ray spectrum shows

284

K. M. RUDALL

almost complete p structure (cf. Fig. 6). However, the amount of absorption a t 1630 cm.-l gives an erroneous impression of the amount of p present. The main expression of this is the rather low absorption of the NH stretching vibration of Fig. 14C compared with that of Figs. 14A and B. For bands with parallel dichroism the absorbing bonds all lie nearly in the plane of the film (which is mounted perpendicularly t o the radiation). Where bands have perpendicular dichroism the absorbing bonds are not oriented in the plane of the surface, b u t a t all angles in a plane nearly perpendicular to the surface. Thus only some of the N H and C=O bonds of the p configuration are in a position t o give maximum absorption. The consideration of such orientations is important in comparing optical densities, whether polarized radiation is used or not. Besides confirming the types of infrared absorption associated with a and p forms, which have been described for other proteins and polypeptides (2, 31) the spectra of Fig. 14C demonstrate a n a-type phase which is more difficult to transform by stretching than the crystalline phase. It corresponds to the K3 phase of wool keratin (6); the 6-type absorption a t 1630 cm.-l arises from the crystalline phase Kz and possibly some Kl phase which transforms along with or before the crystalline material. 4. Heat Denaturation and Infrared Absorption

Prior to the study of oriented cross 0 films, a n investigation was made of the effect of temperature on unoriented films. A series of pieces of one large uniform film of purified epidermin were mounted on stainless steel frames and fixed down over an aperture measuring 25 X 4 mm. The infrared spectra were recorded for such films, not heated in water, heated in water at 68, 80 and 100°C. for 2 minutes, and at 100°C. for 30 and 270 minutes. The hot water treatments at 68, 80 and 100°C. were chosen because scarcely any a --+ p transformation could be detected by X-ray means a t 68"C., while a t 80°C. there is substantial p formation and a t 100°C. the a - + p transformation is complete according t o the diffraction patterns. The amount of p formed a t 100°C. should be not less than the amount of crystalline material and might, of course, be due to the transformation to p of the entire structure. The principal spectra are shown in Fig. 15 for the region 1750-1450 cm.-l Optical densities are plotted in the ordinate from vertically displaced zero lines marked A, B, C, etc. Intensities have been adjusted t o give roughly comparable thickness of film. There is no evidence for the appearance of a second C--V vibration at 1630 cm.-l in A and B, which are the spectra for untreated film and film heated in water a t 68°C. for 2 minutes; the X-ray diagrams of these

THE PROTEINS O F THE MAMMALIAN EPIDERMIS

285

films were entirely in the a form. I n C, heated at 80°C. for 2 minutes, there is a small but perceptible shoulder a t 1630 cm.-', and the X-ray photograph of this film showed a substantial development of @-pattern. At 100°C. for 2 minutes the X-ray transformation to p is complete, and a stronger peak a t 1630 cm.-l is seen in D. More and more of the material gives absorption a t 1630 cm.-' in E (100°C. for 30 minutes) and F (100°C. for 270 minutes). I n F the film was very brittle. For comparison with this series A-F, the spectrum for epidermin stretched in ammonium sulfate a t 105°C. so as t o give a practically complete @ diffraction pattern, is shown in the interrupted curve G. The spectrum G corresponds very closely to that for the unstretched film heated in water at 100°C. for 30 minutes, E. It is of interest t o note that the stretching in hot saturated ammonium sulfate took approximately this length of time (30 minutes). The point of greatest interest from the present infrared studies is the conclusion that the spontaneous a -+ @ transformation, caused b y heating in water, takes place very much more readily in the crystalline material than in the non-crystalline. This result, though surprising, is seemingly identical with the classical observation of Astbury and Woods (6) th a t in stretched hair or wool the @ diffraction pattern is set after about 2 minutes steaming, while the rest of the structure is very much more difficult to set. We can picture the events during heating as follows. If the chain molecules are agitated by heat they unfold, the organized regions become set in the @ form while the unorganized (non-crystalline) regions return t o the a! form or adopt some other configuration which is not p-like. 5. The Oriented Cross R Structure An attempt was made to study the oriented cross @ structure by infrared means. Oriented epidermin film in the same condition as th a t giving the spectrum of Fig. 14B was heated, while held just extended, in water a t 100°C. for 2 minutes. After cooling it was stretched b y 100% in order t o produce the oriented cross @ diagram. As is evident from Fig. 15D the amount of @ formed is very small for this degree of heat treatment and the peak a t 1630 cm.-l is too ill-defined to give very satisfactory differences with the two orientations of polarized infrared. However, using polarized radiation it was observed th a t the cross @ structure is oriented so that the C=O stretching vibration is approximately a t right angles t o its direction in the parallel @ structure. This is satisfactory in so far as it shows that the infrared and X-ray results are in good agreement with respect to the orientations and the change of vibration frequency associated with p structure. Most of the infrared absorp-

286

K . M. RUDALL

tion in the oriented, heat-treated film corresponds t o the a-type with respect to the band a t 1655 em.-' and the marked parallel dichroism a t the NH and C=O stretching bands.

VI. SUMMARY The chief value of this study is to throw further light on the fibrous protein which forms the greater part of the epidermal cell. The protein is highly asymmetric, it has a typical a structure, and in solution in urea/bisulfite it has a molecular weight of the order of 60,000. When reprecipitated and dried the extracted protein shows a lack of cohesion between its particles, while within the tissue much of the protein behaves as if the particles were extensively linked together. There are outstanding differences in the behavior of the protein in the various layers of the cow's nose epidermis. I n the lower parts of the mucosum most of the protein is readily dispersed in 6 M urea, while in the outer parts much of the cell structure does not disperse; in the stratum corneum the main effect of urea is to cause separation of the cells as if an intercellular cement was being dissolved. A measure of the different conditions of the proteins a t the principal levels is given by the various types of thermal contraction which are obtained. The protein which is extracted from the mucosum in 6 M urea shows differences in behavior according to whether it is obtained from the outer or the inner levels. It gradually dissociates into two water soluble fractions which can be separated a t pH values corresponding to their different isoelectric points. The major fraction is the characteristic a-type fibrous protein, while the lesser fraction is non-fibrous and gives a @-type diffraction pattern. These fractions differ also in the relatively low sulfur content of the fibrous protein and the relatively high sulfur content of the nonfibrous protein. The relative proportions of the two proteins is of the right order to account for the total sulfur content of the intact stratum corneum. The condition of the sulfur is, of course, of outstanding importance in determining the stability of the epidermal structure. The whole mucosum is rich in free -SH groups, the corneum containing the oxidized disulfide form. The approximate measure of the difference in stability of the protein fiber structure is that the a! pattern changes spontaneously to @ when heated a t 65°C. in the -SH zone, and not unless zone. heated to about 85°C. in the -S-SImportant advances have been made in methods of producing the oriented 0 form of soluble a-type proteins. By stretching in hot saturated ammonium sulfate very well oriented and nearly perfect p patterns are obtained and this has been particularly useful in studies on infrared absorption; and it should also help in distinguishing two superposed

THE PROTEINS O F THE MAMMALIAN EPIDERMIS

287

diffraction patterns in certain important cases, e.g., the fibrinogen-fibrin system. The cross p diffraction pattern which is associated with the thermally contracted condition of labile a-type proteins, has been further studied. The earliest appearance of this pattern occurs on heating epidermal protein t o 65°C. and shows that a is directly transformed to cross p and not by way of parallel p. By the study of fibers thermally contracted, re-oriented and thermally contracted a second and a third time, evidence is produced that the cross /3 change is associated with just a part of the whole thermal contraction, i.e. that taking place above 55°C. The disorientation or the superfolding, which is responsible for the contraction between 30 and 55"C., is easily re-oriented or unfolded by stretching in cold water. The cross p pattern is reconvertible to the normal a pattern by means of strong urea solution. The conversion of p + a by urea solutions is a specially important phenomenon which should be borne in mind when considering the various changes, or lack of changes, which urea causes in corpuscular or fibrous proteins. The principal elastic properties of films of epidermal protein are seen in a study of the relationship between extension and contraction. Stretching of cast films in cold water proceeds readiIy until extensions of 200-300% are reached, when almost all samples break. The curves relating extension to contraction show that a t 200-3000/, extension a maximum is reached in the degree of contraction. This maximum is associated with a highly oriented condition and a point where further extension involves a-P transformation of the crystalline regions. The maximum contraction is 50% and is noteworthy as it corresponds to the 100% extension proposed for the a -+ p transformation. At much higher extensions of up to 600% (stretching in saturated ammonium sulfate) there is progressive a -+ p transformation and the maximum contraction lies between 50 and 60%. This type of relation (Fig. 13), if it is associated with the length changes involved in reversible a-p transformation, should be a characteristic of a-type proteins in general. A number of infrared absorption studies on the purified epidermal protein are described. I n combination with the X-ray diffraction data, they entirely confirm the recent advances in the definition of absorption spectra, which are characteristic of a- and p-type proteins and polypeptides, in terms of the kinds of dichroism and the frequency changes associated with the p form. It is pointed out that as the dichroism changes with the a + p change it is not possible to read off the relative amounts of a and p from the absorption spectra because of the differences in the orientation of the absorbing bonds with reference to the plane of the film. I n the absorption spectra of highly stretched epidermin, where

288

K. M. RUDALL

the X-ray diffraction pattern is practically completely p, there is a very considerable amount of material which still shows the a-type absorption, with a-type dichroism. We conclude, therefore, th a t the epidermal protein, like the hair fiber, contains a K, phase which is not crystalline and is more difficult to transform than the crystalline or Kf phase. Perhaps the outstanding result of the infrared studies concerns the progress of heat denaturation. Surprisingly, the first effect of heat is t o cause a spontaneous a + @ change in the crystalline regions, for the infrared change closely parallels the change of X-ray diffraction pattern. This phenomenon is similar to the rapid setting in the fl form of the crystalline phase in stretched wool. Prolonged boiling in water of epidermin caused a n extensive but slow conversion to 0,but it is not certain th a t the complete conversion would be achieved by more prolonged boiling. Two principal comments should be made on the infrared work. First, the elastic and other properties frequently suggest th a t a -+ fl transformation is taking place, while the infrared spectra have given no clear evidence of it. Possibly a very refined reading of the spectra may tell us more, but with present technique we can scarcely hope to record small changes which, nevertheless, have a large effect on elastic properties. Secondly, while the X-ray and infrared results have agreed very closely, one wonders whether all p configurations give the frequency changes a t 1630 cm.-', or only the set p configurations. If there is no change in the C=O stretching frequency from 1655 1630 cm.-' a moderate amount of ,f3 could not be detected unless by very careful consideration of the dichroic ratios. The studies add more, perhaps, in the field of protein structure, than t o knowledge of the epidermis or of epithelia in general. B u t what contribution there is of the latter kind is of importance, since our knowledge of the epidermis a t the molecular level is so very small. --f

ACKNOWLEDGMENTS I am indebted to Prof. W. T. Astbury for many facilities, opportunities and discussions. Dr. S. E. Darmon has generously helped with the infrared studies, and their interpretation, while my colleagues Drs. L. Lorand and W. R. Middlebrook have given valued information on matters of biochemical interest.

REFERENCES 1. Alexander, P., and Earland, C. (1950). Nature 166, 396. 2. Ambrose, E. J., Elliott, A., and Temple, R. B. (1949). Nature 163, 859. 3. Ambrose, E. J., Bamford, C. H., Elliott, A., and Hanby, W. E. (1951). Nature 167, 267. 4. Astbury, W. T., and Street, A. (1931). Trans. Roy. Sac. London A230, 75. 5. Astbury, W . T., and Marwick, T. C. (1932). Nature 130, 309. 6. Astbury, W. T., and Woods, H. J. (1933). Trans. Roy. Sac. London A232,333.

THE PROTEINS OF THE MAMMALIAN EPIDERMIS

289

Asthury, W. T., Dickinson, S., and Bailey, K. (1935). Riochem. J. 29, 2351. Astbury, W. T., and Dickinson, S. (1940). Proc. Roy. Soc. London B129, 307. Astbury, W. T. (1947). Proc. Roy. Soc. London B134, 303. Bailey, K., Astbury, W. T., and Rudall, K. M. (1943). Nature 161, 716. Bailey, K. (1944). Advances in Protein Chem. 1, 289. Bailey, K. (1948). Biochem. J. 43, 271. Bailey, K., Gutfreund, H., and Ogston, A. G. (1948). Biochem. J. 43, 279. Bailey, K., Bettelheim, F. R., Lorand, L., and Middlebrook, W. R. (1951). Nature 167, 233. 15. Banga, I., and Szent-Gyorgyi, A. (1940). Science 92, 514. 16. Bamford, C. H., Hanby, W. E., and Happey, F. (1950). Nature 166, 829. 17. Bekker, J. G., and King, A. T. (1931). Biochem. J. 26, 1077. 18. Billingham, R. E., and Medawar, P. B. (1948). Heredity 2, 29. 19. Block, R. J. (1935). Proc. SOC.Ezptl. Biol. Med. 32, 1574. 20. Block, R. J. (1937). J. Biol. Chem. 121, 761. 21. Boeke, J., De Groodt, A., and Heringa, G. C. (1931). Leerboek Der Bijzondere Weefselleer 11. Oostkoek, UTRECHT. 22. Brachet, J. (1942). Arch. Biol. Liege 63, 207. 23. Brachet, J., and Jeener, R. (1944). Enzymologia 11, 196. 24. Bullough, W. S. (1950). Nature 166, 672. 25. ChBvremont, M., and Frederic, J. (1943). Arch. Biol. Lidge 64, 589. 26. Cohn, E. J., and Edsall, J. T. (1943). Proteins, Amino Acids and Peptides, Reinhold, New York. 27. Derksen, J. C., and Heringa, G. C. (1936). Szymonowcz Festchr., Polska Gaz. Lekarska 16, 532. 28. Derksen, J. C., Heringa, G. C., and Weidinger, A. (1937). Acta Neerland. Morphol. 1, 31. 29. Eckstein, H. C. (1935). Proc. SOC.Exptl. Biol. and Med. 32, 1573. 30. Edsall, J. T. (1930). J. Biol. Chem. 89, 289. 31. Elliott, A., and Ambrose, E. J. (1950). Nature 166, 194. 32. Engstrom, A., and Lindstrom, B. (1949). Ezperientia 3, 191. 33. Engstrom, A., and Lindstrom, B. (1950). Biochim. et Biophys. Acta. 4, 351. 34. Giroud, A,, Bulliard, H., and Leblond, C. P. (1934). Bull. Histol. appl. physiol. et path. et tech. Microscop. 11, 129. 35. Giroud, A., and Bulliard, H. (1935). Arch. anal. microscop. 31, 271. 36. Giroud, A,, and Champetier, G. (1936). Bull. SOC. chim. biol. 18, 656. 37. Jacobsen, C. F., and Christensen, L. K. (1948). Nature 161, 656. 38. Lindley, H. (1948). Biochem. J. 42, 481. 39. Longley, J. B. (1949). Ph.D. Thesis, Cambridge University, England. 40. Lorand, L. (1950). Nature 166, 694. 41. Mercer, E. H., and Oloffson, B. (1951). J . Polymer Research 6, 261. 42. Middlebrook, W. R., and Phillips, H. (1947). Biochem. J . 41, 218. 43. MihBlyi, E. (1950). Acta Chem. Scand. 4, 344. 44. Neurath, H., and Saum, A. M. (1939). J . Biol. Chem. 128, 359. 45. Nolte, A. (1947). 2. Naturforsch. 2B, 295. 46. Patterson, W. I., Geiger, W. B., Mizell, L. R., and Harris, M. (1941). J. Research Natl. B u r , Standards U.S. 27, 89. 47, Perutz, M. F. (1949). Proc. Roy. SOC.London A196, 474. 48. Ripa, O., and Speakman, J. B. (1950). Nature 166, 570. 49. Rudall, K. M. (1946). Symposium on Fibrous Proteins, J . Soc. Dyers and Colourists p. 15. 7. 8. 9. 10. 11. 12. 13. 14.

290 50. 51. 52. 53. 54. 55. 56. 57. 58. 59.

K. M. RUDALL

Rudall, K. M. (1947). Biochim. et Biophys. Acta 1, 549. Scherp, H. W., and Syverton, J. T. (1949). Cancer Research 9, 12. Steinhardt, J. (1938). J. Biol. Chem. 123, 543. Straub, F. B. (1942). Studies Inst. Med. Chem. Univ. Szeged 2, 3. Thuringer, J. M. (1924). Anat. Record 28, 31. Thuringer, J. M. (1929). Anat. Record 40, 1. Tristram, G . R. (1949). Advances in Protein Chem. 6 , 83. Wilkinson, V. A. (1934). J . Biol. Chem. 197, 377. Wilkinson, V. A,, and Tulane, V. J. (1939). J . Biol. Chem. 129, 477. Wilson, R. H., and Lewis, H. B. (1927). J . Biol. Chem. 73, 543.

Infrared Analysis of the Structure of Amino Acids, Polypeptides and Proteins

BY G. B. B. M. SUTHERLAND Physics Department, University of Michigun, A n n Arbor, Michigan

CONTENTS Page I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291 1. General Principles of the Method.. . . . . . . . . . . . . . . . . . . . . . . . . . 292 2. Identification of Groups through a . Hydrogenic Stretching Frequencies ....................... 293 b. Multiple Bond Frequencies. . . c. Hydrogenic Deformation Freq d. Skeletal Frequencies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Experimental Techniques. . . . . . . . 11. Amino Acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297 ........................................................ 299 1. T h e 3 p B a n d . . . . ................................. 300 2. The 6 p Band.. . . ....................... . . . . . . . . . . . . . . 302 3. The6.4pBand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302 IV. Polypeptides. . . . . . . . . . . . . . . . . . . . . . . . . . . . 304 2. Protein Denaturation.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Contraction of Muscle Protein.. . . . . . . . . . . . . . . . . . . . . . . . 4. Other Work on Proteins.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. Some General Observations. . . . . . . . . . . . . . . . . . . . . . . . . . . .

309 311

I. INTRODUCTION The purpose of this article is to review the knowledge which has been obtained on the structure of amino acids, peptides and proteins from their vibration spectra. Of the two methods employed to investigate vibration spectra, viz., infrared absorption and Iight scattering, the former has in general proved to be the more powerful and as it is likely t o continue t o be so, considerably more attention will be given in what follows t o infrared spectra than to Raman spectra. The reason for the superiority of the infrared method is that, in general, the technical difficulties in obtaining Raman spectra of small samples (especially of solid materials) are rather formidable whereas the infrared spectrum of 29 1

292

G. B. B. M. SUTHERLAND

almost any material can be observed, given a sample of to 10-7 g. This does not mean that work on Raman spectra of proteins and their constituents should be neglected. I n certain cases (e.g. in aqueous solutions) Raman spectra are frequently much easier to observe than infrared spectra and the Raman method gives many of the low frequency vibrations, which necessitate especially difficult and tedious techniques in infrared absorption. Ideally, one should obtain both the infrared and the Raman spectrum of each compound, as these two spectra are never wholly identical arid are frequently largely complementary for highly symmetrical molecules and for symmetrical groups within larger unsymmetrical molecules. 1 . General Principles of the Method

The use of vibration spectra to determine molecular structure is based on the assumption that each distinct type of molecule will have a characteristic vibration spectrum and that rules exist, or can be formulated, which make it possible to deduce a structure from the corresponding spectrum. I n the case of small polyatomic molecules (e.g. H20, NH3, C2H4) which can be investigated in the gaseous state and which have a fairly high degree of symmetry, strict “selection rules” can be derived by quantum mechanics which determine a unique interpretation of the infrared and Itaman data in favor of one particular configurat,ion of the atomic nuclei. For large polyatomic molecules, such as will be discussed here, it is a t present impossible to give more than a partial interpretation of the observed data, and in no case can the positions of all the nuclei be determined as in the X-ray method of analysis. There are several reasons why the spectroscopic method (which certainly yields the most precise irif ormation available on simple polyatomic molecules) cannot uniquely determine the structure of a complex molecule. First of all, a non-linear molecule with n atoms possesses 3n - 6 fundamental vibration frequencies; thus as n increases, so does the complexity of the spectrum, and th6 mathematical problem of analyzing the corresponding vibration patterns becomes unmanageable. Secondly, large molecules, in general possess a much lower degree of symmetry than the smaller ones, with the result that the “symmetry selection rules” which are so powerful in establishing a structure for small molecules are usually nonexistent for large unsymmetrical molecules. I n the few cases where such rules might be used, they are not strictly obeyed, since interactions between neighboring molecules in the liquid or solid phase modify them in a complex and often unpredictable manner. Nevertheless, much valuable structural information can be derived from the partial interpretations now possible, and as the experimental

THE STRUCTURE OF AMINO ACIDS, POLYPEPTIDES AND PROTEINS

293

and interpretational techniques develop, it is certain that vibration spectra will become an increasingly important structural tool. Already it is possible to settle very quickly certain structural questions which cannot be answered nearly so easily, if at all, by alternative methods. The basic guiding principles of the method can be stated very simply, but many difficulties arise in applying them to particular problems. Some of these will be illustrated in the course of what follows. 2. IdentiJication of Groups through Characteristic Frequencies

A fair proportion of the most intense frequencies observed in the infrared and Raman spectra of a polyatomic molecule may be correlated directly with vibrations localized in small groups of 2, 3, or 4 atoms (e.g. CH, CHZ, CH,). Several of these "group frequencies" are so constant in magnitude and intensity in a wide variety of molecules that they may be used as tracers for the identification of such groups in molecules of unknown structure. It should be realized that correlation rules for the identification of chemical bonds and groups by vibration spectra are partly empirical, being based on observations of the spectra of many compounds containing the groups in question, and have not the same foundation as the symmetry selection rules used in the interpretation of the spectra of smaller molecules. However, theoretical justification can be given in a general manner for many of the assignments, especially those involving hydrogenic frequencies. Within the next few years, no doubt, the precision with which deductions can be made will be considerably improved. Several compilations (Barnes et. al., 1948; Thompson, 1948; Randall et al., 1949; Colthup, 1950) have been made of characteristic frequencies for the identification of groups by vibration spectra, but it may be useful to indicate the four main classes of group frequencies. For convenience, the positions of bands will generally be given both in wavelength ( p ) and in frequency units (cm.-') as there is no agreed convention in this matter. a. Hydrogenic Stretching Frequencies. These frequencies consist mainly in motions of hydrogen atoms along the conventional chemical bonds by which they are attached to other atoms in the molecule. The frequencies in this class of special importance in this article are:

-C-H -N-H -0-H -S-H -p-H

occurring between 3.2 and 3.5 p or 3125 and 2860 cm.-' " 2.8 and 2.9 p or 3570 and 3450 cm.-' 6' " 2.7 and 2.85 p or 3700 and 3510 cm.-' (6 " 3.85 and 4.0 p or 2600 and 2500 crn.-' " 4.1 and 4.25 p or 2440 and 2350 cm.-' '(

((

294

G . B. B. M. SUTHERLAND

Some of these frequencies can be modified considerably by hydrogen bonding thus: -N-H . . may appear between 3.0 and 3.28 p or 3330 to 3050 cm.-l (L " 2.9 and 3.4 p or 3450 to 2940 cm.-l -0-H . . . " " -S-H.. . " 3.9 and 4.05 p or 2560 to 2470 cm.-l It will be noticed that these all occur in the very near infrared, i.e., between 2.5 and 4.5 p . Moving to longer wavelengths reveals the next class of group frequencies. b. Multiple Bond Frequencies. These involve mainly the expansion and contraction of double or triple bonds between atoms such as C, N and 0. These are found between 4.8 /I and 6.5 p , the principal ones which will concern us here are: C=O occurring between 5.5 and 6.0 p or 1820 to 1670 cm.-' C=N (' " 5.9 and 6.1 p or 1690 to 1640 cm.-' 'I " 6.0 and 6.25 p or 1670 t o 1600 cm.-* Of these, only the C=O frequency has so far been observed to be appreciably affected by hydrogen bonding, being lowered by about 50 cm.-' (0.2 PI. c. Hydrogenic Deformation Frequencies. These, involving mainly motions of hydrogen atoms perpendicular to the conventional chemical bonds, lie between 6 and about 16 p or 1670 to 625 cm.-l These frequencies are much more variable in position than the corresponding stretching frequencies and much work still has to be done t o establish them for reliable structural work. Only in the hydrocarbons have they been sufficiently thoroughly investigated to be classified for structural diagnosis. Moreover the effect of hydrogen bonding on NH and OH deformation frequencies is imperfectly understood (Sutherland, 1951), although in general the greater restoring force imposed by the hydrogen bond will have the effect of raising such frequencies by appreciable percentages over their unperturbed values. It should be noted that hydrogen bonding has the opposite effect on the stretching frequencies in (a) and (b). d. Skeletal Frequencies. These frequencies may be described as those in which the motion is not localized in a single chemical bond, but involves changes in bond lengths and bond angles between several non-hydrogenic atoms linked by single bonds. Strictly speaking, all of the frequencies of a polyatomic molecule have some skeletal character, but it is usually justifiable to ignore it in the case of the first three classes. The skeletal frequencies overlap the hydrogenic deformation frequencies and extend to much lower values, having a range between about 8 and 50 p or 1250 t o 200 em.-' Up to the present, these have been of minor value in structural work, although extremely valuable for analytical work and

.

((

c=c

THE STRUCTURE OF AMINO ACIDS,

POLYPEPTIDES

AND PROTEINS

295

“finger printing” a chemical compound. The reason is, of course, that the assignment of these frequencies to specific modes of vibration of the whole molecule has been too difficult a theoretical problem except in the case of very small molecules. Skeletal frequencies which are localized in small groups in a larger molecule, however, are gradually being identified and classified, e.g., those characterizing the COO- ion in amino acids.

3. Experimental Techniques The vibration spectrum of a large polyatomic molecule extends over a very wide range of wavelengths. Absorption bands may occur anywhere from just inside the red end of the visible spectrum (0.75 p ) , all through the infrared region (0.75 to 1000 p ) and into the microwave domain of radio waves (1000 p = 1 mm. to 10 cm.). Like Gaul, this may be divided into three parts, but for somewhat different reasons. The division, which is shown in Table I, is based on the origin of the absorption and each part is further subdivided according t o the method of observing the absorption. TABLEI Regions in p

Range

Dispersing element

Detector element

Materia for cell window

Origin of absorption

I

1

Very short

Intermediate

Very long

* “Mixed

0.75-1.25 Glass or 1.25-2.50 quartz 2.50- 6.0 6 . 0 -15.0 15.0 -28.0 28.0 -38.0

Photographic Glass or Overtones and plate, thermo- quartz combinations couple

Lif or CaFz Thermocouple, LiF Fundamental bolometer or NaCl NaCl frequencies of KBr KBr Golay cell intramolecular KRS 5* KRS 5* vibration

38-100,000 See Randall (1938), Williams (1948), Intermolecular and Faraday Society Discussion on vibrations Dielectrics (1946). crystal” of thallium bromide and iodide.

In the shortest range of wavelengths (0.75 to 2.5 p ) , absorption is by the overtone and combination frequencies of the higher fundamental modes of internal vibration of the individual molecules. Such absorption is weak in intensity, generally requiring a path length of the material varying between 0.1 and 2.5 mm. The amount of information which can be derived from spectra in this region is strictly limited and correlation with molecular structure often less certain than in the succeeding region of fundamental frequencies. Moreover, in the case of proteins, the

296

G . B. B. M. SUTHERLAND

scattering of such short wavelength radiation by the specimen under study frequently makes it impossible t o obtain satisfactory spectra. Nevertheless, valuable work has been done in this region (e.g., Ellis and Bath, 1938) and from the experimental standpoint, it has certain real advantages viz.:

1) Glass can be used for the prism of the spectrometer and for the windows of any absorption cells required. 2) It is not necessary to prepare the extremely thin sections of the sample (2 t o 10 microns) frequently required in the region 2.5 to 38 p, where the absorption due t o the fundamentals is much more intense. 3) Polarized radiation may be produced rather more easily th a n a t longer wavelengths, b y the use of a Glan Thompson prism or, over part of the range, by Polaroid. 4) The absorption due t o water is not quite so troublesome as a t longer wavelengths. For the photographic range i.e. up to 1.25 p , any conventional spectrograph with a resolving power of about 500 may be used; for the range beyond 1.25 p, any of the spectrometers mentioned in the next paragraph bu t one may be used, modified where necessary by the introduction of a quartz or glass prism or a grating to give a resolving power of about 250. I n the iritermediate range of wavelengths, the region between 2.5 and 15 1.1 contains a t least four fifths of the absorption bands due t o fundamental frequencies of intramolecular vibration and is by f a r the most important, since the majority of the information on molecular structure is obtained from these fundamentals, as will become clear from the succeeding sections of this article. It can be most conveniently surveyed in about 15 minutes by using a rock salt prism as the dispersing element. Although this is adequate for most work, the dispersion of rock salt is rather poor in the region between 2 and 6 p , for which it should be replaced by lithium fluoride for the best results. I n some respects calcium fluoride is a better substitute, since it can be used out to 9 p and is not greatly inferior t o lithium fluoride between 2.5 and 6 p ; it is then possible t o dispense with rock salt and use potassium bromide from 9 t o 28 p. This will mean some sacrifice of resolving power between 9 and 15 p where rock salt has its best performance. The region between 28 and 38 p demands a special prism made of a “mixed crystal” of thallium bromide and iodide (sometimes called KRS 5) which is very expensive and tends to deform after some time. The possibilities of cesium bromide as an alternative to KRS 5 are now being investigated by Dr. E. K. Plyler a t the Bureau of Standards and the Harshaw Chemical Co. of Cleveland, Ohio. Throughout the whole of the intermediate

THE STRUCTURE OF AMINO ACIDS,

POLYPEPTIDES AND PROTEINS

297

range, the detector may be either a thermocouple, a bolometer or a Golay cell. Polarized radiation can be most conveniently produced by transmission through a small pile of silver chloride plates, set a t the correct angle t o the incident beam. Many of the manufacturers of infrared spectrometers listed below will supply such polarixers, made u p ready for use. Infrared spectrometers for this range may be divided into two classes, called “single beam” and “double beam” instruments. I n the former, the absorption due t o atmospheric carbon dioxide is troublesome near 4.3 and 14 p and more especially, th at due to water vapor near 2.8 p and between 5 and 8 p . Furthermore, percentage absorptions have to be computed from the record obtained. The most commonly used single beam instruments are made by Beckman (Pasadena, California), Grubb Parsons (Newcastle on Tyne, England), Hilger (London, England), and Perkin Elmer (Norwalk, Conn.). In the double beam instruments, a compensating beam passes through the same atmospheric path and the troublesome atmospheric absorption bands are eliminated from the final spectrum. These instruments are also extremely convenient for work with substances in solution since the solvent may be put in the compensating beam and its absorption spectrum similarly eliminated, except of course where it is very intense, and insufficient energy is left in the two beams t o make operation reliable. The most commonly used double beam spectrophotometers are manufactured by Baird Associates (Cambridge, Mass.), Beckman, Hilger and Perkin Elmer. There are several ways of achieving the desired compensating effect in a double beam instrument. The merits and demerits of the three main methods have recently been discussed by Williams (1951), who has also published a very valuable review on infrared instrumentation (Williams, 1948). I n the range of very long wavelengths, the absorption of radiation by polyatomic molecules in the liquid or solid state is generally due to intermolecular vibrations, i.e. hindered translation or hindered rotation in the case of liquids and lattice oscillations in the case of crystals. So far as the writer is aware, no work has been done on proteins in this field and no discussion will therefore be given of the experimental techniques which are fairly elaborate and expensive. Observations in this region might be of considerable interest in relation to hydrogen bonding in proteins but it must be remembered th at absorption of liquid water is very intense over the whole range.

11. AMINOACIDS The first systematic work on the vibration spectra of amino acids in relation t o their structure was done by Edsall and his coworkers (1936, 1937, 1938, 1940, 1943, and 1950) who examined the Raman spectra

298

G. B. B . M. SUTHERLAND

of a wide variety of amino acids and related compounds, generally in aqueous solution. Edsall was able to provide strong spectroscopic evidence for the dipolar ion structure of the amino acids. The following were two crucial arguments. Edsall showed that in a fatty acid, ionization of the COOH group caused the CO frequency of the COOH group, near 1730 cm.-', to disappear, proving that this is not a characteristic frequency of the COO- ion. In the free amino acid, no Raman line is found near 1730 cm.-', although when the cation, +NHgR.COOH, is formed, this frequency is immediately evident. Secondly, he showed that the unionized amino group has a characteristic frequency near 3320 cm.-', which is present in the sodium salts of glycine and alanine but is absent in the spectra of the free amino acids. It will be noted that these arguments are essentially arguments against the structure NH2CHRCOOH, but positive evidence was also found from Raman spectra for the presence of NHa+ and COO- groups in free amino acids although this was not quite so clear-cut (Edsall, 1938). The reason is that the key Raman frequencies of these groups appear to be close to certain CH2and CHI frequencies from which it is not too easy to differentiate them. In infrared absorption, the evidence for the dipolar ion structure is equally strong. Klotz and Gruen (1948) showed that glycine, valine, and norleucine all possess a strong absorption band, peaked near 6.35 p, (1575 cm.-l) which is still present when the sodium salts of these are examined, but which is modified or disappears when the hydrochloride is formed. In the hydrochloride an absorption band appears near 5.85 p, (1710 cm.-l) which is well known to be characteristic of the CO link in a COOH group. Independent evidence exists that an absorption near 6.39 p (1565 cm.-l) is to be associated with the presence of carboxylate ions (Davies and Sutherland, 1938). Infrared spectra for most of the common amino acids were first recorded by Wright (1937, 1939) but no attempt was made to assign the bands. More recently Thompson et al. (1951) have also recorded the spectra of several amino acids and confirm the interpretation of a band common to all of them near 6.3 p (1587 cm.-') as due to the COO- ion. These investigators also suggest that another characteristic COO- infrared frequency lies near 1400 cm.-l and that weak absorption noted in many amino acids near 2100 cm.-' may be attributed to NH frequencies in the NH3+ ion. There is previous evidence from Edsall's work that similar frequencies in the Raman spectra should also receive this interpretation. An extensive investigation of the infrared spectra of amino acids was made by Darmon and Sutherland between 1945 and 1948 but the results havemot yet been published in an accessible form (Darmon, 1948) as it was hoped that further work would

THE STRUCTURE OF AMINO ACIDS,

POLYPEPTIDES AND PROTEINS

299

elucidate some of the many puzzling features of these spectra. Our results and conclusions are in general agreement with the published work just reviewed, but a detailed discussion of the many anomalies would be out of place in this article. An interesting fact, first noticed by Wright (1937, 1939) is that the infrared spectrum of the DL-form of an amino acid is usually markedly different from the spectrum of either the D- or the L-form of the same acid when each is examined in the solid state. Wright attributes this to “compound formation’’ between the D- and L-forms. Darmon et al. (1948) have confirmed this observation, which is extremely important if infrared methods are to be used for the analysis of mixtures of amino acids, e.g., the estimation of leucine: iso-leucine ratios in protein hydrolyzates. In this connection, Gore and Petersen (1949) have reported differences between the spectra of L-threonine and D-threonine when examined in the solid state. They point out that this might arise from a polarization effect in the spectrometer. 111. AMIDES

A study of the spectra of amides and of polypeptides is clearly an essential preliminary to any investigation of protein structure by vibration spectra, and a considerable amount of work has already been done on such compounds. I n the infrared field, a very thorough study of the amides and of the simple -CO-NH(peptide) link was made as part of the infrared analysis of penicillin carried out in several laboratories in Britain and U.S.A. between 1943 and 1945 (Thompson et aZ., 1949). From this work, it was well established that all compounds of the type R-CO-NH-R1, where R and R1 are alkyl groups, exhibit strong absorptions near 3, 6.0, and 6.4 p which have been assigned respectively to the stretching vibration of the NH group, the stretching vibration of TABLEI1 Characteristic Absorption Frequencies of the -CO.NHMain position of absorption Wavelength Frequency in p in cm.-’

3.0 6.0 6.4

3330 1670 1560

Group

Assignment

Remarks

N-H stretching C=O stretching N-H deformation

Has anomalous features

the CO group, and the deformation vibration of the NH group (Table 11). There is no doubt whatever about the first two of these assignments, but

300

G . B. B. M . SUTHERLAND

the third presents certain difficulties (Randall et al., 1949 and Sutherland, 1951) and has been the subject of controversy (Lenormant, 1951). These bands require some detailed consideration in view of the fact that they are not always single and shift in position under certain circumstances, e.g., changes are found for the amide group in dilute and concentrated solution, and in the liquid and solid state. Their behavior will therefore be considered separately in the succeeding three subsections, since a proper understanding of such phenomena must precede the effective use of these frequencies in the interpretation of protein spectra. 1. The 3 p Band

The changes in the absorption spectrum of a monosubstituted amide in the neighborhood of 3 p when the amide is dissolved in solvents such as CCl, or CHCl, have been investigated by Buswell and various collaborators (Buswell et al., 1938, 1940a) and later by Lecomte and Freymann (1941), Richards and Thompson (1947), Darmon and Sutherland (1949), Tsuboi (1949a, 1949b), and Mizushima and others (1950). The idea underlying all these investigations was that since hydrogen bonding of the type -OH . . . 0 in alcohols and carboxylic acids can be studied in the infrared through variations in the position of the OH frequency, hydrogen bonding of type -CO . . . HN- in the amides might be followed in a similar manner through variations in the NH frequency. Thus in very dilute solutions, where virtually no interaction occurs among the amide molecules, one might expect to observe a single absorption band. This is indeed the case and the unperturbed NH frequency has been established independently by all investigators as occurring near 3440 cm.-' (2.9 p ) . As the concentration is increased, this sharp band gradually disappears, while a new broader band appears, which grows in intensity very rapidly. This new band has a maximum close to 3330 cm.-' (3 p ) . As the concentration is still further increased, a second new band appears, which is also rather broad and has its maximum intensity near 3070 cm.-' (3.26 p ) . The spectrum of the pure liquid is very similar to that of a concentrated solution in this region (i.e. for acetyl methylamine) except that the maximum of the 3330 cm.-' (3 p ) band appears to have moved to about 3275 cm.-l (3.05 p ) . These effects are illustrated in Fig. 1. We are therefore led to make the following correlations for N H frequencies in monosubstituted amides: N-H (unbonded) 3440 cm.-'-2.9 p 3300 cm.-'-3.0 p N-H (hydrogen bonded) 3275 cm.-'-3.05 p 3070 cm.-'-3.26 p

...

THE STRUCTURE OF AMINO ACIDS,

POLYPEPTIDES AND PROTEINS

301

The reasons for the different frequencies observed in the case of the bonded N-H . . . are not clear. Buswell et al. (1940a) suggested that the 3.26 p band arose from cyclic dimers (I),and that the 3.0-3.05 p

(11)

(1)

band was due to single links (11)in the formation of polymers. It will be noted that (I) and (11) correspond to cis and trans configurations of the CO and NH groups about the C-N link. The study of cyclic amides

FIQ.1. Change in the NH and CO stretching frequencies in an amide in passing from dilute solution (A) to concentrated solution (B) in carbon tetrachloride.

in which only the cis configuration is presumed present was undertaken independently by Darmon and Sutherland (1949) and Tsuboi (1949a, b). The former reported NH association bands for cyclic amides at 3160 cm.-l (3.16 p) and 3070 cm.-' (3.26 p ) , of which the higher frequency band was the more intense. Since the 3300 cm.-1 (3.0 p ) band was never observed with cyclic amides, Darmon and Sutherland assigned this band to association in the trans configuration, attributing the 3160 cm.-l (3.16 p ) band to polymeric association in the cis configuration as shown in (111). Tsuboi (1949a, b) also concluded that the 3300 cm.-' band was

(111)

to be assigned to association in the trans configuration, but regarded only the 3160 cm.-l band as characteristic of association in the cis configuration

302

G. B. B . M. SUTHERLAND

and offered no interpretation for the 3070 crn.-' band, which is found in the association spectrum of both cyclic and noncyclic nonsubstituted amides. It has t o be admitted th at an unequivocal assignment of the various N H bands occurring near 3 p in amides cannot yet be given. 2. The 6 p Band

Here again the effect of dilution has been studied by Richards and Thompson (1947) , by Miaushima and others (1 950), and by Darmon and Sutherland (unpublished). The effects found were similar to those observed on the C-0 frequency in carboxylic acids (Davies and Sutherland, 1938). Thus in concentrated solutions, in pure liquids, or in the solid state where hydrogen bonding is bound to be prevalent, this band lies a t a markedly lower frequency than for dilute solutions; the difference in this CO frequency being about 40 cm.-' (Fig. 1). It should be noticed that this is less than the change in the N H stretching frequency, which was several hundred cm.-' Typical values for this frequency in a simple monosubstituted arnide are:

C=O C=O

...

(unbonded) 1695 cm.-' (5.9 k ) 1655 cm.-' (6.04 p ) (bonded)

It has been noted that the absolute value of this frequency varies in different amides and is probably affected by the degree of electronegativity of neighboring groups (Richards and Thompson, 1947). I n the solid state, however, this band is frequently double and the cause of this doubling is not understood. I n certain cases it may indicate that there are both bonded and unbonded CO groups present in the solid state but it is possible that the doubling is due t o an interaction effect other than hydrogen bonding. It should be remarked, that there seems to be general agreement th a t the exceptionally low value of the CO frequency in a peptide link, viz. 1695 cm.-' (cf. 1740 cm.-' in ketones) can be attributed t o some resonance contribution of the form -0C--NH+ in the electronic structure of this group. 3. The 6.4 p Band The main arguments in favor of assigning this absorption band t o the

NH deformation frequency are as follows: 1. It is absent in disubstituted arnides, i.e. when the N H group becomes NR, where R is, for instance, a n alkyl group. 2. It moves towards shorter wavelengths (by about 35 cm.-l) in the transition from dilute t o concentrated solution in nonpolar solvents.

THE STRUCTURE OF AMINO ACIDS,

3. 4. 5. 6.

POLYPEPTIDES AND PROTEINS

303

This is to be expected if it is a deformation frequency being affected by hydrogen bonding. It has the appropriate numerical value for such a frequency (cf. 6.5 p in amines). It exhibits the correct polarization properties in oriented polypeptides. It is considerably reduced in intensity when the amide is partially deuterated. It is virtually unaffected when N16is substituted for N14 in acetylglycine (Darmon and Sutherland, unpublished).

The main arguments against this assignment are as follows:

7 . It is absent in the spectra of cyclic amides and of lactams. 8. When partial deuteration takes place, this band would be expected to weaken (as observed) and a new band corresponding t o the NU deformation frequency would be anticipated near 9 p. Although a weak new absorption is observed at 8.9 p the most intense new band is observed near 6.75 p (1480 cm.-l). 9. It is inactive in the Raman spectra of monosubstituted amides. The foIlowing alternative assignments have been considered : a. The stretching frequency for the C-N link. This explanation seems to be inconsistent with (2) and (6) and although it would possibly explain (8), gives no help in accounting for (7) and (9). Moreover the numerical value seems much too high. b. The stretching frequency for the C=N link in an enolic form of the amide coexisting with the keto form. However no evidence has been found by any observers for an OH frequency required by this explanation.

[

c. A stretching frequency of the group -C