ADVANCES IN PROTEIN CHEMISTRY VOLUME VII
This Page Intentionall Left Blank
ADVANCES IN PROTEIN CHEMISTRY EDITEDBY
...
14 downloads
766 Views
23MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
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<., and Ottesen, M. (1948). Arch. Biochem. 19,340. Elsden, S. R., and Synge, R. I,. hl. (1944). Biochem. J. 38, ix. Felix, K., Fischer, H., Krekels, A , , and Rauen, H. M. (1950). 2. physiol. Chem. 286, 67. Felix, K., and Lang, A. (1929). Z. physiol. Chem. 182, 125. Felix, K., and Mager, A. (1937). Z.physiol. Chem. 249, 111. Felix, K., and Schuberth, H. (1942). Z. physiol. Chem. 273,97. Fischer, E., and Abderhalden, E. (1907). Ber. 40, 3544. Fodor, A., Fodor, P. J., and Kuk-meiri, S. (1947). Enzymologia 12, 101. Fowden, L., and Penney, J. R. (1950). Nature 166, 846.
64
F. SANGER
Fox, S. W. (1945). Advances i n Protein Chem. 2, 155. Fromageot, C., Jutisz, M., Meyer, D., and Pknasse, L. (1950). Biochim. et Biophys. A d a 6, 283. Fruton, J. S. (1950). Yale J . Biol. Med. 22, 263. Goldschmidt, S., Freyss, G., and Strauss, K. (1933). Ann. 606, 262. Gordon, A. H. (1949). Discussions of the Faraday SOC.7, 128. Gordon, A. H., Keil, B., and Sebesta, K. (1949). Nature 164, 498. Gordon, A. H., Martin, A. J. P., and Synge, R. L. M. (1941). Biochem. J . 36, 1369. Gordon, A. H., Martin, A. J. P., and Synge, R. L. M. (1943). Biochem. J . 37, 92. Grant, R. L., and Lewis, H. B. (1935). J . Bid. Chem. 108, 667. Grassmann, W. (1950). Angew. Chem. 62, 170. Grassmann, W., and Hannig, K. (1950). Nat,uru,issenschajten 37, 397. Green, F. C., and Schroeder, W. A. (1951). J . Am. Chem. SOC.73, 1385. Green, N. M., and Work, E. (1951). Biochem. J . 49, xxxvii. Gregory, J. D., and Craig, L. C. (1948). J . Biol. Chem. 172, 839. Guntelberg, A, V., and LinderstrZm-Lang, K. (1949). Compt. rend. lab. trav. Carlsberg. Sdr. chim. 27, 1. Gutfreund, H. (1948). Biochem. J. 42, 544. Gutfreund, H., and Ogston, A. G. (1949). Biochem. J. 44, 163. Hamoir, G. C. M. (1945). Biochem. J . 39, 485. Hanby, W. E., and Rydon, H. N. (1946). Biochem. J . 40, 297. Harington, C. R., and Pitt-Rivers, R. (1944). Natiire 164, 301. Haugaard, G., and Kroner, T. D. (1948). J . Am. Chem. SOC.70, 2135. Haugaard, G., and Roberts, R. M. (1942). J . Am. Chem. Sor. 64, 2664. Hellerman, L., and Stock, C. C. (1938). J . Biol. Chem. 126, 771. Howard, G. A., and Martin, A. J. P. (1950). Biochem. J . 46, 532. Huang, H. T., and Niemann, C. (1950). J . Am. Chem. SOC.72, 921. Huffman, H. M. (1942). J . Phys. Chem. 46, 885. Ingram, V. M. (1950). Nature 166, 1038. Jacobsen, C. F. (1949). Compt. rend. lab. trav. Carlsberg. Shr. chini. 26, 455. Jensen, H., and Evans, E. A. (1935). J . Bid. Chem. 108, 1. Johnston, R. B., Mycek, M. J., and Fruton, J. S. (1950). J . Biol. Chem. 186, 629. Jones, T. S. G. (1948). Biochem. J . 42, xxxv. Jones, T. S. G. (1949). Discussions of the Faraday Sor. 7, 285. Jutisz, M., and Lederer, E. (1947). Nature 169, 445. Kaplan, E.H., Campbell, E. D., and McLaren, A. D (1950). Biochim. et Biophys. Acta 4, 493. Keston, A. S., and Udenfriend, S. (1949). Cold Spring Harbor Symposia Quant. Biol. 14, 92. Keston, A. S., Udenfriend, S., and Cannan, R. K. (1946). J . Am. Chem. SOC.68, 1390.
Keston, A. S., Udenfriend, S., and Cannan, It. K.
(1049). J . Anz. Chem. SOC.71,
249.
Keston, A. S., Udenfriend, S., and Levy, M. (1947). J . A m . Chenz. SOC.69, 3151. Keston, A. S., Udenfriend, S., and Levy, M. (1950). J . A m . Chem. Soc. 72, 748. Kies, M. W., and Davis, P. L. (1950). Federation Proc. 9, 189. Kuhn, W. (1930). Ber. 63, 1503. Kuhn, W.,Molster, C. C., and Freudenberg, K. (1932). Ber. 66, 1179. Lederer, E., and Tchen, P. K. (1947). Biochim. et Biophys. A d a 1, 35. Lens, J. (1949). Biochim. et Biophys. A d a 3, 367.
THE ARRANGEMENT O F AMINO ACIDS IN PROTEINS
65
LBonis, J. (1948). Compt. rend. lab. trav. Carlsberg. Sdr. chim. 26, 315. Levene, P. A., and Bass, L. W. (1928). J . Biol. Chem. 78, 145. Levene, P. A., and Bass, L. W. (1929). J . Biol. Chem. 82, 171. Levene, P. A., Rothen, A., Steiger, R. E., and Osaki, M. (1930). J . Biol. Chem. 86, 723. Levene, P. A., Steiger, R. E., and Rothen, A. (1932). J . Biol. Chem. 97, 717. Levy, A. L. (1960). J . Chem. SOC.404. Levy, M., and Slobodiansky, E. (1929). Cold Spring Harbor Symposia Quant. Biol. 14, 113. Linderstrgim-Lang, I<. (1949). Cold Spring Harbor Symposia Quant. Biol. 14, 117. Linderstrflm-Lang, K., and Ottesen, M. (1949). Compt. rend. lab. trav. Carlsberg. Sbr. chim. 26, 403. Livermore, A. H., and du Vigneaud, V. (1949). J . Biol. Chem. 180, 365. Lugg, J. W. H. (1938). Biochem. J . 32, 775. McLaren, A. D. (1949). Advances in Enzymol. 9, 75. Mandl, L., Levy, B., and McLaren, A. D. (1950). J . Am. Chem. SOC.72, 1790. Martin, A. J. P. (1949a). Biochem. SOC.Symposia 3, 4. Martin, A. J. P. (194913). Discussions of the Faraday SOC.7, 332. Martin, A. J. P. (1950). Ann. Rev. Biochem. 19, 517. Martin, A. J. P., and Mittelmann, R. (1948). Biochem. J . 43,353. Martin, A. J. P., and Synge, R. L. M. (1941). Biochem. J . 36, 1358. Martin, A. J. P., and Synge, R. L. M. (1945). Advances in Protein Chem. 2, 1. Meyer, K. H., Fuld, M., and Klemm, 0. (1940). Helv. Chim. Acta 23, 1441. Middlebrook, W. R. (1949). Nature 164, 501. Miller, G. L., and Andersson, K. J. I. (1942). J . Biol. Chem. 144, 465. Mitchell, H. K., Gordon, M., and Haskins, F. A. (1949). J . Biol. Chem. 180, 1071. Mitchell, H. K., and Haskins, F. A. (1949). Science 110, 278. Monnier, R., and Jutisz, M. (1950). Bull. SOC. chim. biol. 32, 228. Monnier, R., and Penasse, L. (1950). Compt. rend. 230, 1176. Montroll, E. W., and Simha, R. (1940). J . Chem. Phys. 8, 721. Moore, S., and Stein, W. H. (1948). J . Biol. Chem. 176, 367. Moore, S., and Stein, W. H. (1949). J . Biol. Chem. 178, 53. Moring-Claesson, I. (1948). Biochim. et Biophys. Acta 3, 389. Myrback, K. (1949). Arkiv. Kemi 1, 161. Nakashima, R. (1927). J . Biochem. (Japan). 7, 441. Neurath, H., and Schwert, G. W. (1950). Chem. Revs. 46, 69. Northrop, J. H., Kunitz, M., and Herriott, R. M. (1948). Crystalline Enzymes, 2nd ed., New York, p. 73. Partridge, S. M. (1948). Biochem. J . 43, 238. Partridge, S. M. (1949a). Discussions of the Faraday SOC.7, 296. Partridge, S. M. (194913). Biochem. J. 44, 521. Partridge, S. M. (1949~). Biochem. J. 46, 459. Partridge, S. M., and Davis, H. F. (1950). Nature 166, 62. Partridge, S. M., and Swain, T. (1950). Nature 166, 272. Partridge, S. M., and Westall, R. G. (1949). Biochem. J . 44, 418. Patton, A. R., Foreman, E. M., and Wilson, P. C. (1949). Science 110, 593. Pauling, L., and Niemann, C. (1939). J . Am. Chem. SOC.61, 1860. Perlmann, G. E. (1949). J . Am. Chem. Sac. 71, 1146. Petermann, M. L. (1942). J . Biol. Chem. 144, 607. Petermann, M. L., and Pappenheimer, A. M. (1941). J . Phys. Chem. 46, 1.
66
F. SANGER
Phillips, D. M. P. (1948). Nature 161,53. Phillips, D.M. P. (194%). Biochim. et. Biophys. Acta 3, 341. Phillips, D. M. P. (1949b). Nature 164, 545. Pittom, W.W. P. (1914). Biochem. J. 8, 157. Polson, A. (1948). Biochim. et Biophys. Acta 2, 575. Porter, R. R. (1948). Biochim. et Biophys. Acta 2, 105. Porter, R. R. (1950a). Biochem. J. 46, 473. Porter, R. R. (1950b). Biochem. J. 46, 479. Porter, R. R. (1950~). Methods in Medical Research 3, 25G. Porter, R. R., and Sanger, F. (1948). Biochem. J. 42, 287. Randall, S. S., and Martin, A. J. P. (1949). Biochem. J. 44, ii. Rauen, H. M., and Felix, K. (1948). 2. physiol. Chem. 283, 139. Rees, M. W. (1946). Biochem. J. 40, 632. Rideal, E.K., and Mitchell, J. S. (1937). Proc. Roy. SOC.London A169, 206. Rittenberg, D., and Foster, G . L. (1940). J . Biol. Chem. 133, 737. Rometsch, R. (1950). Helv. Chim. Acta 33, 184. Sanger, F. (1945). Biochem. J. 39, 507. Sanger, F. (1946). Biochem. J. 40, 261. Sanger, I?. (1948). Nature 162,491. Sanger, F. (1949a). Biochem. J. 44, 126. Sanger, F. (1949b). Biochem. J. 46, 563. Sanger, F. (1049~). Cold Spring Harbor Symposia Quant. Biol. 14, 153. Sanger, F. (1949d). Hiochem. Soc. Symposia 3, 21. Sanger, F., and Tuppy, H. (1951a). Biochem. J. 49, 463. Sanger, F.,and Tuppy, H. (1951b). Biochem. J. 49, 481. Schmid, K. (1949). Helv. Chim. Acta 32, 105. Schramm, G.,and Primosigh, J. (1943). Ber. 76, 373. Schramm, G., and Primosigh, J. (1944). Ber. 77, 417. Shepard, C. C., and Tiselius, A. (1949). Discussions of the Faraday SOC.7,275. Stein, W. H., and Moore, S. (1948). J . Biol. Chem. 176, 337. Stein, W. H., and Moore, S. (1949). Cold Spring Harbor Symposia Quant. Biol. 14, 179. Stein, W. H., Moore, S., and Bergmann, M. (1944). J . Biol. Cheni. 164, 191. Steinhardt, J. (1941). J. Biol. Chem. 141,996. Steinhardt, J., and Fugitt, C . If. (1942). J . Research Natl. Biir. Standards 29, 315. Steinhardt, J., and Zaiser, E. M. (1950). J . Biol. Chem. 183, 789. Svensson, H. (1948). Advances in Protein Chem. 4, 251. Svensson, H.,and Brattsten, I. (1949). Arkiv Kemi 1, 401. Swanson, M. A., and Cori, C. F. (1948). J . Riol. Chem. 72, 797. Synge, R. 1,. M. (1943). Cheni. Revs. 32, 135. Synge, R. L. M. (1944). Biochem. J. 38, 285. Synge, R. L. M. (1945). Biochem. J . 39, 351. Synge, R. 1,. M. (1949). Biochem. J . 44, 542. Synge, R. L. M., and Tiselius, A. (1947). Acta Chem. Scand. 1, 749. Synge, R. 1,. M., and Tiseiius, A. (1949). Acta Chem. Scand. 3, 231. Synge, It. L. M., and Tiselius, A. (1950). Biochem. J. 46,xli. Taeawa, Y. (1949). Acta Phytochim. (Japan) 16,73. Tiselius, A. (1947). Advances i n Protein Chem. 3, 67. Tiselius, A. (1948): Arkiv Kemi Mineral. Geol. 26B, No. 1. Tiselius, A , , and Claesson, S. (1942). Arkiv Kemi Mineral. Geol. 16B,No. 18.
T H E ARRANGEMENT O F AMINO ACIDS I N P R O T E I N S
67
Tiselius, A., Drake, B., and Hagdahl, L. (1947). Experientiu 3, 651. Tiselius, A,, and Hagdahl, L. (1950). Acta Chem. Scand. 4,394. Toennies, G., and Homiller, R. P. (1942). J . Am. Chem. SOC.64,3054. Tristram, G. R. (1949). Advances in Protein Chem. 6, 83. Troensegaard, N. (1947). Acta Chem. Scand. 1, 672. Turba, F. (1948). 2. Vitamin-, Hormon- u. Fermentforsch. 2, 49. Turba, F., and Enenkel, H. J. (1950). Naturwissenschuften 37, 93. Turba, F.,and Richter, M. (1942). Ber. 76,340. Uchino, T. (1934). J. Biochem. (Japan) 20, 65. van Slyke, D. D., Dillon, R. J., MacFadyen, D. A., and Hamilton, P. (1941). J. Biol. Chem. 141, 627. van Vunakis, H., and Brand, E. (1951). Abst,racts, 119th Meeting, American Chemical Society, April, 1951, p. 28c. Vickery, H. B. (1922). J . Biol. Chem. 63,495. Vickery, H.B., and Osborne, T. B. (1928). Physiol. IZevs. 8,393. Villee, C. A,, Linderstrbm-Lang, Ti., and Ottesen, hl. (1950). Federation Proc. 9, 241. Virtanen, A. I., and Hamberg, U. (1947). Acta Chem. Scand. 1, 847. Virtanen, A. I., Kerkkonen, H., Hakala, M., and Laakonen, T. (1950). Xatwwissenschaften 37, 139. Waldschmidt-Leitz, E., and Turba, F. (1940). J . prakt. Chem. 166, 55. Waldschmidt-Leitz, E., and Turba, F. (1941). J . prakt. Chem. 168, 72. Warner, It. C. (1942a). J. Biol. Chem. 142,705. Warner, R.C. (1942b). J . Biol. Chem. 142,741. Wasteneys, H.,and Borsook, H. (1930). Physiol. Revs. 10, 110. Wieland, T. (1942). 2. physiol. Chem. 273,24. Wieland, T. (1944). Ber. 77, 539. Wieland, T., and Fischer, E. (1948). Naturwissenschaften 36, 29. Wieland, T., and Wirth, 1,. (1949). Ber. 82, 468. Woiwod, A. J. (1949). Biochem. J . 46, 412. Woolley, D. W. (1949). J . Riol. Chem. 179, 593.
This Page Intentionall Left Blank
The Structure of Collagen Fibrils
BY RICHARD S. BEAR Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts
CONTENTS I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Significance of Collagen.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Hierarchies of Structural Element., . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Methods.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Identification and Distribution of Collagens. . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. General Characteristics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Physical Criteria for Classification.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Types of Mesenchymal Collagen. . . . . . . . . . . . . . . . . . . ............. 4. The “Secreted” Collagens.. . . . . . . . . . . . . . . . . . . . . . . . ............. 5. The Chemical Significance of the Physical Classification.. . . . . . . 111. Colloidal Structure of Collagen Fibrils. . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Optical Evidence 2. A Fibrillar Mode 3. Diffraction Theory for 4. Electron Optical Evide IV. The Collagen Protofibril. . . ......... 1 . General Characteristics 2. Pseudoperiodic Axial Structure of Small Size.. . . . . . . . . . . . . . . . . . . . . . . . 3. Structure Transverse to Fibril Axes; Hydration.. . . . . . . . . . . . . . . . . . . . . 4. Protofibril Extensibility, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Protofibril Contractility; Thermal Contraction. . . . . . . . . 6. The Content per Unit of Protofibrillar Length.. . . . . . . . . . . . . . . . . . . . . . . V. The Collagen Molecule.. ......................................... 1. Description of the Molecule., . . . . . . . . . . . . . . . 2. Distribution of Side Chains. . . . . . . . . . . . . . . . 3. Swelling Phenomena. . . . . . . . . . . . . . . . . . . . . . . ................
Page 69 69 71 73 76 76 78 84 90
112 115 117 119 127 135 146
........................ . . . . . . . . . . . . . . . . 150 2. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
I. INTRODUCTION 1. SigniJicance of Collagen
The collagenous tissues of animals are in some respects similar to the cellulosic structures of plants. As a major fibrous constituent of skins, 69
70
RICHARD S. BEAR
bones, tendons, fascia, and other types of loose and dense connective tissue occurring in practically all body organs (149) collagen is undoubtedly the most prevalent single organic constituent of animal organisms. Medical concern with collagen is derived from its importance as a constituent of tissues intimately involved in aspects of the aging process, in neoplastic growth, in diseases connected with vitamin deficiencies or hormonal imbalances, and, indeed, in many general systemic disorders involving the connective tissues in a diffuse or inflammatory way (124). As the principal constituent of animal skins, collagen is manipulated in the tanning of leather. Gelatin and glue are well known degradation products, both of which find wide spread industrial use, the former, for example, being indispensable to the production of photographic emulsions. The immunological inertness of collagen gives rise to its usefulness in surgical sutures. h recent bibliography of the literature of collagen (37) lists o \ w a thousand references. h’otwithstanding this volume of activity the state of fundamental concepts in the field is unsatisfactory. The present review summarizes some of the basic information supplied by X-ray diffraction, electron microscopy, and chemical studies, developing a viewpoint regarding the structure of the ultramicroscopic element of all collagenous substances, the collagen fibril. I n thus narrowing the present subject many medically and commercially important aspects will not be immediately served, for these often depend on tissue organization, which involves interrelation between collagen and other components. A perusal of modern literature dealing with the physiology and pathology of connective tissues (GO, 124) leaves the general impression that the fibrous components, collagen and elastin, are proving of less immediate importance in problems of crucial medical concern than are the mucopolysaccharide-containing “ground substance l 1 and the cellular elements. Kevertheless, the ubiquitous (204) collagen fibrils can scarcely fail to affect in fundamental ways the avenues of movement and growth open to cells, as well as to act as barriers or filters screening the metabolites passing between individual cells, the blood and lymph. Lloyd (138) has pointed out that, on the basis of the rough correlation between the water contents of tissue elements and their biological or biochemical activity, the collagen fibril is one of the more moribund of tissue elements, being chemically designed, in fact, for stability in normal physiological environments. As the present discussion advances it becomes clear that a collagen fibril is essentially a long, thin, single “crystal”; as such its formation or dispersion, synthesis or degradation, is expected t o be largely at the mercy of its surroundings. Nevertheless,
THE STRUCTURE OF COLLAGEN FIBRILS
71
it is capable of more variation than is met with in simpler crystalloids, so th a t when methods for the examination of its chemical and physical structure are refined, these variations may furnish information concerning the physiological milieu from which it was derived. I n order that these practical aims may be served it is necessary first to develop in as much detail as possible the structure of normal collagen fibrils. A wide variety of these objects are available. They provide particularly suitable material for structural studies, which should furnish important bases from which t o attempt extrapolation to protein systems of more distinct biological activities.
2. Hierarchies of Structural Element
To make clear the aspects of structure herein treated, a few definitions are of importance a t the outset. Figure 1 indicates the major structural elements found in connective tissues, as well a s the correspondence between these elements and the methods applicable for their resolution and study. The terminology adopted is similar to th a t employed by Kuntzel (131), who described as fibers the microscopically visible aggregates with diameters of 100 to 200 p in tendon, 20 to 40 u in skin. Smaller components are primitive fibers ( 2 to 10 u ) . The latter may be further subdivided by mild mechanical or chemical means into fibrils whose diameters range from figures close to the resolution of light microscopes down to hundreds of Angstroms and require the electron microscope for adequate resolution. While the possibilities for longitudinal cleavage do not stop with the fibril, it is generally a fairly coherent unit whose properties provide the ultramicroscopic basis for collagenous tissue elements. I n fact, since the grosser fibers may contain also cellular, polysaccharide, and elastic components, it is with the fibril that one first expects t o approach something like a “pure” collagen. When thinner, subfibrillar elements are discernible, these are termed generally $laments, and, as Schmitt et a l . (192) further suggest, it is useful to consider a hypothetical thinnest unit filament, the protojibrill which carries the essential chemical and configurational structure of collagen. A collagen molecule would be smaller, though perhaps not thinner, than the protofibril. It is suggested below th at the protofibril is no wider than a very few polypeptide chains (probably one, diameter ca. 12 to 17 A), so that chain termini (defined as the points where primary valence connections between atoms become lacking or weak) will mark off molecular boundaries along the protofibril. The classical term, micelle, is no longer very useful in describing for collagen the volume over which the quasi-crystalline arrangement of intrafibrillar structure extends to produce the oriented X-ray diffraction
72
RICHARD S. BEAR
FIQ.1. See opposite page for legend.
T H E STRUCTURE O F COLLAGEN FIBRILS
73
effects that are observed. The fibril turns out to have diffracting structures-within-structures, of sizes ranging from hundreds to a few Angstroms. The micelle for large features of structure would be the entire fibril, or much of it, while the micelle of the smaller details would approach classical significance, including a length much less than th a t of the molecule but a width which is that of a few clustered protofibrils. There are certain descriptive terms which apply to intrafibrillar volumes whose boundaries cut across protofibrillar or molecular limits. These terms are useful, particularly in describing the appearance of fibrils in electron micrographs. Therein one observes bands, which are disk-like enlargements of fibrillar diameter with greater than average power to deflect or absorb electrons and to take u p electron stains. T h e bands form a characteristic pattern of cross striations which periodically repeat along the fibril axis. The fibrillar levels remaining between bands may be called interbands. The discussion which follows deals with the means of recognizing fibrils of the collagen type and with the description and significance of the subfibrillar entities just enumerated.
3. Methods For present purposes electron microscopy is particularly useful because it is the only available method for directly observing individual fibrils with any degree of adequacy. Th e X-ray diffraction method offers the next most direct approach, since, although the results are obtained from relatively large fiber bundles (diameters of about 1 mm.), the diffraction effects are the summation of the effects of individual fibrils or subfibrillar components. One takes precautions t o see th a t the fibrils are as nearly parallel as possible, and if individuals are not too different in structure, the entire diffraction pattern becomes essentially t h a t of a single fibril. The diffraction method, while suffering from certain inherent limitations of interpretation, offers the greatest resolving power for the elucidation of configuration th at has been applied to collagen as yet. The X-ray information is obtained from diffraction patterns which FIG.1. Illustrating the various hierarchies of element involved in present considerations. A t the top are shown the distinctive components of connective tissue. The ground substance, which contains the mucopolysaccharides, hyaluronate and chondroitin sulfate, has been studied chemically by Karl Meyer (150, 151) and by K. H. Meyer (158),among others, and electron optically by Gross (87), b u t this component does not receive significant further attention herein. Occasional reference is made to elastin, but the text considers principally the collagenous elements, shown in steps of increasing magnification.
74
RICHARD S. BEAR
represent a peculiar sort of inverted image of structure. Since many readers will not be accustomed to analysis of these diffraction images, Fig. 2 is given in explanation of some of the unavoidable technical terms.
FIG.2. Illustrating the X-ray diffraction method as applied to collagenous fibers. Electrons, e, emitted hy an incandescent filament, F , are driven by *. pofmtiLt1 of about 40,000 volts onto the target of copper, Cu, which yields the X-radiation that emerges from the X-ray tube through a window of beryllium, Be. A thin foil of nickel, N i , transmits principally the characteristic Cu Ka radiation (wavelength A = 1.54 A . ) to the collimating system, which consists of fine pinholes cut in lead disks, Pb. Beyond the collagen fibw a t C emerge the undiffracted beam lf and various diffracted or deviated beams represented by the one marked D. The relatively intense ccntral beam is attenuated by the beam trap, Y,whose shadow is wen overlaying the center of each of the diffraction patterns of Figs. 3 through 20. The final diffraction pattern, as registered on a photographic film consists of layer lines (vertical dashed lines) and row lines (horizontal dotted lines) of spots. Prominent features of the collagen wide-angle pattern are indicated by M , the important 2.86 A arc on the meridian; E, the 10 to 12 A. equatorial spot; and H , the diffuse “half-halo” (4.6 A.) referred to in the text. The small-angle layer lines a t S generally require finer pinholes and longer collimators and specimen-to-film distances to permit their resolution. In some cases collimating apertures are horizontal slits, which diffuse the diffraction spots along layer lines but retain layer-line sharpness vertically and permit more rapid registration of patterns.
There are two general size ranges which the diffraction method can cover. One includes sizes approximating those of atoms, atom groups, amino-acid residues, and the thinner dimension of collagen protofibrils and molecules (20 A. or less) ; these quantities are investigated at wide diffraction angles, just as the large apertures of high-power objectives in
THE STRUCTURE O F COLLAGEN FIBRILS
75
ordinary microscopy are employed for the resolution of small details. The second size range (20 to 1000 A.) overlaps with that of the electron microscope in indicating large fibrillar or subfibrillar features of structure; the diffraction angles are small (up t o about 4" with Cu K a radiation of wavelength 1.54 A.), corresponding to the small apertures ordinarily used by lower-power microscope objectives. On either type of diffraction pattern the individual diffraction spots can be used t o calculate spacings between planes of equivalent structure in the specimen. The spots often occur in row or layer lines as shown in Fig. 2 . Particular importance attaches to the layer lines, which are expected t o cross the central row line (meridian) a t regular intervals and may be numbered (indexed) according to their distance from the central layer line (equator). The separation between layer lines usually permits unequivocal establishment of the size of the axial period of the fibril's structure, i.e. the distance between equivalent locations along the fibril. Row lines can be used to determine structure transverse to the fibril axis, but unless a large number of row lines are available the study of transverse structure is limited. I n collagen investigations the true fibrillar period is gained at small angles. Wide-angle patterns exhibit a smaller, apparent or pseudoperiod, indicating t ha t within the large pattern there are smaller sub-patterns. I n diffraction by such fibrous systems other unusual effects (21) are encountered because the fibrils are not ideal three-dimensional crystals, or because all fibrils of a massive specimen may not be perfectly parallel. In the latter case individual spots show arcing about a circle of constant radius, the full circle being completed when orientations are entirely random. Factors such as distortion of internal structure or limitation in external dimensions of fibrils cause the spots to spread along row or layer lines, and the manner in which this diffusion is exhibited indicates the nature of the cause. I n collagen each small-angle layer line consists of a single, meridionally centered spot which acquires appreciable length along the layer line because of these effects. At wide angles non-meridional spots are observed, but they become particularly diffuse as they depart from the meridian. Ordinary optical, organic chemical and physical chemical approaches also offer information which is indispensable for the present purposes. These commonly suffer from the difficulty th a t they may include extraneous, noncollagenous tissue elements in the analysis, but they furnish useful auxiliary information about molecular composition, size, shape and orientation. Some of these molecular characteristics are particularly difficult t o derive from the methods of electron microscopy and X-ray diffraction directly.
76
RICHARD S. BEAR
11. IDENTIFICATION AND DISTRIBUTION OF COLLAGENS 1. General Characteristics
The determination of a set of properties which can be used to form a definition of collagen is as difficult as this substance is complex. Collagen fibers are often recognized histologically b y their ability to swell i n acids and alkalies; by rather non-specific affinities for some dyes, such as acid fuchsin or aniline blue, particularly when mordanted; by lack of extensibility comparable with that of elastic tissue components; hy an immunological inertness and resistance to degradation by enzymes ; by a pronounced shrinkage in length at elevated temperatures or in the presence of certain electrolyte solutions; and by a transformation into gelatin in boiling aqueous environments. These general characteristics have sufficed to permit investigators to identify collagenous fibers in practically all parts of vertebrate animals, and, indeed, in all animals except those of small size or cell number (Protozoa). It is desirable, however, t o possess more definite means of establishing relationships between these diverse substances. The types of difficulty facing the more common means of identifying collagms may be illustrated by a few examples. Histologically visible fibers of collagenous nature are complex objects (see Fig. 1). They may, on occasion, be expected t o show chemical and physical properties characteristic of any of the major connective tissue components, even though they may most often resemble collagen predominantly. Reliance on the microscopic appearance of collagen fibers has often led to difficulty, such as th at involving the relation between reticulin and collagen. The former occurs as very fine fibers which commonly stain with silver much more strongly than do the relatively massive, typical collagen fibers. A similar distinction may be made with Schiff reagent after periodic acid oxidation, a method which stains many tissue components containing polysaccharides and also colors reticular fibers strongly (137). Embryonic skins, involving reticulin and almost no typical collagen, have also been reported to contain abnormally high sulfur content, which was judged to have been responsible for variations in physical properties from those normal for collagen-containing material (100). I n view of the well known fact that embryonic skins are high i n polysaccharide content (204) it would seem possible th a t the peculiar properties and high sulfur contents of reticular tissues may be ascribed to polysaccharide which has been in part adsorbed on or occluded within the fine reticulin fibrils. Partridge (166) has recently suggested that
THE STRUCTURE OF COLLAGEN FIBRILS
77
chondroitin sulfate plays an important role in the organization of collagen in developing connective tissue. I n any event, further discussion below definitely shows that reticulin is basically a collagen, as has long been suspected (162). Factors such as fiber diameter, compactness, and the difficult solubility of collagen would seem of importance in determining such properties as resistance to hydrolytic attack by enzymes. This inertness is probably not entirely derived from the chemistry of collagen but depends also on physical factors which prevent an enzyme from gaining access to its substrate. Pepsin, acting under acid conditions on swollen fibrils, is able to cause degradation (195), and increased temperature or physical maceration facilitates even the attack by trypsin in neutral or slightly alkaline environments (194, 200). On the other hand, collagenase capable of hydrolytically attacking, rather specifically and at neutral pH, both gelatin and collagen has been reported to be produced by cultures of certain anaerobic bacteria (various members of the genus Glostridium, 29, 113). This enzyme has not been more than partially purified, which is an important point, for it will be interesting to determine whether and how a single enzyme species can attack the bulky collagen fibril, overcoming its insolubility and causing hydrolysis simultaneously. This problem does not arise to the same degree with gelatin, but even here the particular mode of hydrolytic action is not yet demonstrated. Certain of the physical properties commonly recognized as characteristic of collagen are not infallible guides. I n ordinary microscopy extensibility has been found limited for collagen fibers, but this has not proved the case for individual fibrils observed electron optically (216). Mammalian and other collagens are normally transformed into gelatin by action of hot water, but the elastoidin of elasmobranch fishes, otherwise much like typical collagens, is resistant to this transformation (69). Elastoidin also shows, in its thermal shrinkage properties, a degree of reversibility which many collagens exhibit only after formalin fixation. These are just a few of the differences which otherwise clearly related collagenous substances exhibit, and along with other variations they are considered more fully a t appropriate places below. Chemical criteria for distinguishing collagens are also not satisfactory. This is in part because present evidence is not available in sufficient quantity to determine whether significant chemical definitions can be formed. As is elaborated below, however, it seems likely that these adequate chemical criteria will be elusive for some time. I n any case, in dealing with substances as complex as collagen, one will not immediately wish to demand chemical identity of all collagens, for
78
RICHARD S . BEAR
physical configuration of molecule and fibril may be as important as constitution. Required are methods which reach the basic structure of the ultramicroscopic collagen fibrils and are sensitive enough t o show the features which are fundamental. At the same time they must be sufficiently insensitive t o overlook minor variations such as are encountered in passing from tissue to tissue in the same organism, or possibly even from animal to animal, A method which meets these requirements has been available in X-ray diffraction for some time, and somewhat more recently electron microscopy has been found similarly useful. 2. Physical Criteria for Classification The first X-ray investigations of collagen-containing tissues or derivatives involved wide-angle diffraction (99, 104, 186) and were carried out in 1920 t o 1926. Early techniques failed in some cases to yield very significant results, but there soon followed a recognizable description of the typical wide-angle diffraction of native collagen fibers (104) and a demonstration that even the degradation product, gelatin, yields, upon stretching, the same oriented diagram as tendon (82, 120). It was evident quite early that collagens from a number of mesenchymal tissues exhibit essentially identical patterns, differing only in the degree of orientation indicated. Gone11 and Kratky (84) were among the first clearly t o contrast the collagenous tissues with the keratinous ones on the basis of their diffraction, but Astbury (7) finally crystallized the views regarding the wide-angle diffraction classification of the fibrous proteins in 1938. Astbury pointed out that it was possible to recognize common structural features in a number of protein fibers which yield practically identical diffraction patterns in spite of diverse origins. Distinction could readily be made in this way between members of a collagen class and those of another broad and considerably more heterogeneous set, now termed the keratin-myosin-epidermis-fibrinogen (k-m-e-f) group, comprising as the name suggests fibers from hair, muscle, epidermal and blood sources (12). Although the significance of Astbury’s classification has been questioned (154), there is as yet no convincing reason for discarding it. The diffraction patterns originally used for this classification were of the wide-angle type. I n 1936 Clark, Parker, Schaad, and Warren (54) and Wyckoff, Corey and Biscoe (217) discovered th a t toward the center of the collagen diffraction pattern, in a region hitherto obscured by bulky traps for the undiff racted beam, occur a number of diffractions indicating structures along the fibril axes with periods of possibly several hundred
THE STRUCTURE OF COLLAGEN FIBRILS
79
A. These were resolved a few years later (18, 128) t o reveal a large fibrillar period of 640 A., which was almost simultaneously photographed directly with electron microscopes (96, 207). The lc-m-e-f group of fibers also show small-angle diffraction (19, 58, 147)) and corresponding large structures may be seen in electromicrographs (59, 68, 97). Criteria useful for distinguishing the two great groups of protein fibers in the realm of large structure have been suggested (19, 148). The structural significance of the X-ray diffraction and electron optical observations is discussed in further detail below. At the moment these physical observations may be regarded as empirical means of segregating or distinguishing members of a class. Those aspects of the physical observation which are important for identification and classification of collagens now follow. Figures 3 through 5 show a number of the variations of the wide-angle collagen diagram that are obtained from several dried mesenchymal tissues. In all of these cases an attempt has been made, by means of proper selection of t,issue for purity and natural orientation, plus application of tension during drying, to secure as near an approach as possible to the best wide-angle diagram of collagen, typified by Fig. 3. However, in examining the many diverse tissues which contain members of the class the desired purity and orientation are not always secured. I n the worst instances one has to contend with the contribution by impurities of overlying diffuse background, which is most likely to obscure the diffraction areas corresponding to spacings in the neighborhood of 5 A. Lack of orientation throws the normal spots or arcs into rings, wiping out or running together the fainter and more diffuse details. Nevertheless, even under poor conditions dry specimens yield the 10 to 12 and the 2.86 A. diffraction rings, which normally are equatorially and meridionally accentuated, respectively. The collagens also show characteristic wideangle diffraction when moistened, as shown in Figs. 7 and 8, with differences in detail (Figs. 9 and 10) which are discussed further in Section IV, 3. While all types of collagen fibril diffract in nearly identical fashion a t wide angles, this is not the case at small angles, where individual differences become apparent. Figures 11 through 20 present a number of small-angle diffraction diagrams of members of the collagen class, serving to illustrate the points: (a) that the fundamental large structures of the different members studied to date are all of essentially the same size (600-660 A. normally in the dry state), as is evidenced by the fair constancy of the separations between consecutive lines of each pattern, and (b) that individual members of the class do, however, exhibit differences
80 RICHARD 6. BEAR
T H E STRUCTURE O F COLLAGEN FIBRILS
81
in the way in which they distribute diffracted intensity among the several lines of the separate patterns. Figures 11 and 15 show that the diffraction patterns of different members become more nearly alike when the specimens are moistened, whereupon the axial period also becomes somewhat enlarged (665 to 680 A.) and there is a distinct tendency for many of the odd layer lines to be more intense than the alternating even ones. By appropriate manipulation the small-angle diagrams of a single type of collagen may be varied, with results of the kind shown in Figs. 12, 13, 14, and 16, whose significance is considered further below. The small-angle diffraction method of identifying and comparing collagens has not often been applied as yet (148), though the cases available include material of quite widely separated origins and offer illuminating indications as to what may be expected of a general survey. Added t o the difficulties of the type mentioned above for the wide-angle examination of collagenous tissues, are the generally increased technical troubles of small-angle work, which particularly requires massive, fairly pure, well oriented specimens in order to facilitate registration of the diagrams during reasonable exposure times. Also fibrous protein specimens lose their ability to exhibit diffraction effects after physical or chemical manipulation more readily for the small- than for the wide-angle range (18), which reduces the permissible variety of pretreatments of material, designed for purification or orientation. The electron optics should furnish structural criteria for collagen classification nearly equivalent to those of small-angle diffraction. Electron micrographs disclose the large periodic structure, running along fibril axes, as a sequence of bands and interbands, seen in various aspects illustrated by Figs. 21 through 27. The over-all size of the sequential pattern corresponds in magnitude to the fundamental large spacing revealed by the small-angle diffraction line separations, while the details of different band and interband patterns correspond to the variFIGS.3 THROUGH 8. Wide-anglc diffraction patterns of collagens. All are a t the same magnification, which is about three quarters of the original size as obtained at a specimen-to-film distance of 5 cm. (rf. Fig. 2). Individual patterns were obtained from the following specimens: 3. Dry kangaroo tail tendon (KTT). 4. Dry shark-fin ceratotrichium (elastoidin). 5. Thyone body wall, dried under tension. 6. Dry chrome-tanned KTT. 7. Moist KTT. 8. Moist elastoidin. FIGS.9 A N D 10. Enlarged views of the central equatorial streaks of patterns 7 and 8, respectively. See Section IV, 3.
82
RICHARD S. B E A R
FIGS.11
THROUGH
20.
See opposite page for legends.
THE STRUCTURE OF COLLAGEN FIBRILS
83
ous distributions of diffracted intensity among the small-angle lines. Although the electron microscope has not as yet been used for systematic identification and comparison of widely different collagens, this correspondence between image characteristics and small-angle diffraction suggests that electron optics will disclose relatively constant large periods but quite different band and interband structures within them. The electron optical method suffers, however, from several defects. The most important of these is that its specimens are necessarily subjected to more or less drastic pretreatments in order to separate fibrils for individual examination. The result is often a considerable variation between the individual fibrils relative to size of period and the details of appearance of the band pattern (164, 191), a variability greater than would seem indicated b y the diffraction effects observed in massive samples which have received minimal manipulation (see Section 111, 2 ) . Consequently, conclusions as to the representative condition of the fibrils of a given population must be reached after statistical treatment of many observations (90, 192). Because of various technical limitations, it is not generally possible to demand of the electron optical studies clear resolution of structures with size less than about 50 A. in organic substances of this type. Consequently, the possibility arises that a given structure, departing appreciably from what is usually observed for collagens, may be difficult t o classify (cf. the case of “neurotubules,” 190). The X-ray method offers the possibility of employing both wide- and small-angle criteria, in effect establishing that certain structural details as small as 2.86 A. are those of a collagen. For this reason X-ray examination, where applicable, furnishes more definitive classification. FIGS.11 THROUGH 20. Small-angle diffraction patterns of collagens. All are a t the same magnification, which is about 5.4 times the original size yielded a t a specimen-to-film distance of 15 cm. Patterns 11 through 16 were obtained from a camera with resolving power for small angles much better than that of the one used for patterns 17 through 20, which came from poorer diffractors. (The cameras were those described as having “ f d ” values of 800 and 400 A,, respectively, by Bolduxn and Bear, 34.) Layer lines whose indices are multiples of 3 are indicated by dots to the left of the upper halves of each pattern. Specimens are as follows: 11. Moist K T T . 12. Dry (“unfanned”) KTT. 13. Dry (“fanned”) KTT. 14. Dry K T T with ca. 1.5 % phosphotungstate stain. 15. Moist elastoidin. 16. Dry chrome-tannrd KTT. 17. Oven dried carp swim bladder (ichthyocol). 18. Thyone body wall, dried undcr tension. 19. Dry elastoidin. 20. Dry decalcified Ralticina axial stalk (cornein).
84
RICHARD S. BEAR
There are two aspects to the identification problem: (a) in the first, one is concerned with determining the characteristics of collagens which are found a s components somewhere within massive samples, possibly originating from digerent tissues; but (b) one might also desire to localize and distinguish the several components of a single tissue. The second problem, which involves employing the physical methods discussed above as adjuncts to histochemical studies, is technically more difficult than the first. With X-ray diffraction, the limitations arise largely from the weakness of the diffraction effects which can be obtained from the small amount of collagen localized in a small area of a thin tissue section. With available techniques and X-ray sources, the wide-angle and small-angle diagrams can be obtained in feasible exposure times (a few hundred hours) from reasonably pure and well oriented volumes of the order of ml., corresponding t o short segments of fibers tens of microns in diameter. Feitelberg and Kaunitz (76) have used wide-angle X-ray diffraction in this way t o study changes with age in the alignment of collagen in human chorda tendineae. I n electron microscopy individual fibrils may be photographed, but the great magnification normally employed makes difficult the localization of groups of fibrils relative to structure at the ordinary microscopic level. Nevertheless, with improvement of sectioning methods for electron microscopy, this procedure will undoubtedly prove to be the one of choice for this purpose. Punching of small, microscopically located portions of tissue sections, followed by electron optical study of the isolated bits, is an alternative method which may be used in studying the state of collagen in localized areas of tissue sections.
3. T y p e s of Mesenchymal Collagen Collagen fibrils are generally expected in mesodermal or mesenchymal tissues, or in the mesoglea when typical mesoderm is not developed, as in the Coelenterata and Porifera. Table I summarizes the varieties of collagen which have been identified in sources of this kind and b y the physical methods described immediately above. As is easily understandable, most of the available information deals with the mammalian tissues. In this class of vertebrate animal the small-angle diffraction evidence suggests that all common forms of collagen are basically similar. For example, their fundamental fibrillar periods are in the neighborhood of 640 A. for dry specimens, and they show prominent intensification of such diffraction lines as the sixth, ninth, and eleventh (18). Examples examined to date are too numerous t o mention individually in the table, and, indeed, they furnish much of the
THE STRUCTURE O F COLLAGEN FIBRILS
FIGS.21
THROUGH
24.
See page 87 for legends.
85
86
RICHARD S . BEAR
FIGS.25
THROUGH
28. See page 87 for legends.
THE: STRUCTURE O F COLLAGEN FIBRILS
87
wide-angle diffraction and electron optical evidence regarding collagen structure cited throughout the present paper. The small-angle diff raction investigations include tendons of various sources (rat, beef, kangaroo), decalcified bone (guinea pig), skin (human, calf), cornea (beef), intestinal submucosa (sheep surgical gut), and demineralized mammoth tusk. Beef spleen was examined after concentration of the collagenous elements by partial digestion with trypsin (18). This organ is usually considered to provide a good source of reticulin, and the small-angle diffraction evidence indicated the presence of typical mammalian collagen. Since histologically recognizable ordinary collagen was present, this may not have constituted conclusive proof th a t reticulin and collagen are structurally similar. Nevertheless, electron optical study places reticulin definitely in the collagen class, since it differs from normal collagen chiefly in the matter of fibril diameter a s far as appearance in electron micrographs can determine (89). Wide-angle diffraction evidence serves only to make identification definite and is able to make no distinctions except a s to general purity and orientation. Heringa and Weidinger (100) reported a n exception to the rule, namely, that reticulin from the skin of newborn rats exhibits a shortened equatorial spacing, 9.7 A., instead of the usual value of about 10 to 12 A., observed with other d ry collagens. Experiments of the present author with Dr. Jerome Gross (89), using similar material, ~-
~
FIGS.21 THROUGH 28. Electron micrographs of collagenous preparations as follows: 21. Rat tail tendon (RTT) fibril stained with phosphotungstic acid (PTA), reprinted from Hall (95). The stained bands are labelled after Schmitt and Gross (191). Dots mark bands a to e, reading up. 22. RTT fibril frayed by swelling in 0.1 % pepsin and 0.5 % acetic acid, stained with 0.1 % ’ PTA. Unpublished photograph of Schmitt, Hall, and Jakus (192). 23. Chromium-shadowed collodion replica of moist human skin fibrils, with two dried fibrils adhering. Reproduced from Gross and Schmitt (90). 24. Similar shadowed replica of parallel human skin fibrils whose periods are in lateral register. Also after Gross and Schmitt. 25. Unstained R T T fibril, greatly stretched by accidental breaking of supporting collodion film. Reprinted from Schmitt, Hall and Jakus. 26. Fibril from chromc-tanned calfskin, otherwise unstained. Reprinted from Schmitt and Gross. Dots mark bands b l , b2, c, d, e and a, which form three pairs of bands whose centers are equally spaced. 27. Part of a human skin fibril left embedded in a formvar replica and shadowed with Cr. One portion, a single periodic segment in length, has been displaced, leaving a clean gap. Unpublished photograph of Gross and Schmitt. 28. Long spacing fibrils precipitated by dialysis from a mixture of plasma mucoprotein and an acetic acid solution of ichthyocol. Stained with PTA. Prepared from a photograph of Highherger, Gross and Schmitt (110).
88
RICHARD S. B E A R
indicated t ha t the observation of an anomalous value for the reticulin spacing is in error, and th at Heringa and Weidinger were confused b y a background of diffuse diffraction supplied b y the voluminous “ground substance” present in immature skin. Again it seems clear th a t reticulin is basically a collagen, since it possesses normal wide-angle diffraction. The elastic fibrous component of connective tissue, elastin, has often been examined by use of wide-angle diffraction techniques, applied t o ligaments as sources of high elastin content. Earliest studies showed collagen-type diffraction (104) , and Kolpak (126) found that stretching TABLEI Mesenchynial and Mesogleal Collagens Identified by Physical Criteria Phylum Vertebrate.
Class Mammalia
Aves Pisces Teleostomi Teleostomi Elasmobranchii Mollusca Echinodermata
Name
Common sources
Collagen
Connective tissues, tendons, bones, skin corium
Reticulin
Embryonic skin, spleen
Collagen
Tendons
Ichthyocol
Skins, tendons, swim bladders Scales Ceratotrichia of shark fins Squid connective tissue Thvone body wall Arbacia peristonie Asterias body wall
lchthylepidin Elastoidin
Cephalopoda Holothuroida Echinoida Asteroida
I
Coelenterata
Anthozoa
Cornein
Axial stalks, acontia
I’orifera
Demospongiae
Spongin
Skeletal fibers
Identification* W.A., S.A., E.M.
W.A. (89, 100) S.A. (18) E.Al. (89) W.A. (121)
I {
W.A. (49, 98, 108) S.A. (18) W.A. (84, 104) W.A. (16,47, 84, 104) S . A . (this paper) E.AI. (192)
W.A., S.A. (148) W . A . (9,50, 148) S.A. (118) W.A. (104, 148)
* Identification by means of W.A.. wide-angle X-ray diffraction; S.A., sniall-anple diffraction; E. M., electron microscopy. References to original papers are given where not too numerous.
beef liqamentum nuchae beyond 75 % was effective in developing oriented collagen diffraction against the normal background of diffuse, unoriented halos. Astbury included elastin in the collagen family (7, 8) but recognized that the collagen diffraction of ligament is probably due to typical collagen, which is always present in small amounts (9). He suggested that elastin itself, which apparently cannot be induced t o yield sharp, oriented diffraction, may he a collagen whose thermal shrinkage temperature (see Section IV, 5) is abnormally low, below room temperature. Small-angle diffraction disclosed a faint collagen-type diffraction in beef ligament (18), such as would result from a small amount of collagen impurity included with the elastin. Elastin has also been studied eler-
THE STRUCTURE O F COLLAGEN FIBRILS
89
tron optically (78,88,208). It seems clear that elastin must be excluded from the collagen family. Of the other possible collagens of vertebrate animals, those obtained from fish tissues have received most attention, ichthyocol and elastoidin in particular. Small-angle diffraction yields for ichthyocol a normal axial period but line intensities which are similar but not identical with those of mammalian collagen (Fig. 17). Chemical and physical properties, often cited below, also suggest that ichthyocol differs appreciably from other vertebrate collagens. Unpublished diffraction studies have produced evidence that carp swim bladder tunic requires more intensive drying than does mammalian collagen to develop the strong sixth, ninth, and eleventh small-angle layer lines. Elastoidin, however, shows still more definite departure of its large structure from that of mammalian collagen, as is clearly seen in Fig. 19. The nature of these variations appears in more detail below, but suffice it at the moment that elastoidin shows diffraction phenomena typical of a fibrillar structure which is considerably distorted internally. The over-all spacing (dry) is reduced to 600 A., and the small-angle line intensities and “shapes ” are quite distinctly altered; for example, the sixth line is almost lacking and most lines show greatest strength off the meridian. It becomes difficult to observe the small-angle diffraction because the distortion makes the fibril into a poor diffractor. Wideangle diffraction shows the evidence for unusual distortion only in subtle ways, being for the most part normal. Of the collagens from invertebrate animals listed in Table I only those of the Echinodermata and Coelenterata have received more than incidental attention, Three echinoderm classes are represented, whose collagens tended to show the usual marked intensity of the sixth smallangle diffraction, but an abnormal depression of the ninth line’s intensity. Other normally (i.e., as in mammalian collagen) weak lines, such as the fifth and tenth, often showed moderate expression. Among the anthozoan axial stalks examined, Balticina yielded the best small-angle diagram, which provided a result quite different from all other collagens: only the fifth and fourteenth small-angle lines had appreciable intensity, although the sixth still showed some persistence. The specimens from these two phyla usually exhibited somewhat larger periods (650-675 A., dry) than normally encountered, but not beyond reason. I n several instances of Table I only wide-angle diffraction is used to identify the members as collagen-type fibers. Spongin, however, is the sole one of these cases for which the large spacings have actually been sought so far. No conclusive result was obtained (148). Negative evidence is difficult to assess because tissue impurity or internal structural
90
RICHARD S. B E A R
distortions can easily render the small-angle diffractions difficult to observe. Spongin is, however, phylogenetically farthest removed from the collagens of typical mesenchymal origin and may represent a n early evolutionary stage of the collagen type of fiber. Systematic coverage of all possible varieties of mesenchymal collagen has obviously not been secured as yet. Nevertheless, certain general trends seem apparent in the available evidence. Over wide phylogenetic excursions the members of the family show a remarkable degree of constancy in the magnitudes of their large and small structural spacings. The variations which occur involve chiefly those structural features th a t develop the alterations in small-angle line intensities, whose significance is considered subsequently. These variations show some relation to phylogeny: witness, for example, the fact th a t the elastoidin, obtained from the most primitive vertebrates listed in Table I, shows greatest departures from the properties of the other vertebrate cases, and th a t in moving t o other phyla, the echinoderms and coelenterates, in each instance one encounters distinctly new characteristics. Members of the collagen class of fibrous proteins form a more homogeneous group structurally than do those of the keratin-myosin class. While one must make appreciable steps across or between phyla in order to discover distinctly new collagens, with the keratin-myosin group one encounters, within the same organism, tissues (hair and muscle, for example) whose short spacings are comparable but whose large structures are very different (19). However, a given tissue type (muscle is the best known example, 189) shows fibrous components with perhaps even greater phylogenetic stability than the collagens possess, suggesting that it may be more proper t o compare separate sub-groups of constant tissue type. Although the wide-angle diffraction serves a useful function in permitting the classification of fibrous proteins according to certain features of fundamental structure, it is now clear that the larger size ranges covered by small-angle diffraction and electron microscopy are also necessary in obtaining deeper insight into the meaning of such classification. 4. The “Secreted” Collagens Not all of the fibrous proteins which have been reported to yield the collagen-type wide-angle diffraction diagram are listed in Table 1. Several others are segregated in Table I1 because it is possible th a t they may prove of special significance. Among these are two substances which Champetier and FaurB-Fremiet (48) included under the term ‘‘ secreted keratins.” Their results showed, however, that designations such as “ovo-keratin” and “bysso-keratin ” of the older literature are inappropriate according to diffraction evidence.
THE STRUCTURE O F COLLAGEN FIBRILS
91
It is widely considered that all collagens are produced extracellularly, although cells are actively engaged somehow, from the fibroblasts of mammalian tissues to the spongioblasts of Porifera. Until the mechanism of formation is better understood, it remains possible that all collagens are essentially secreted, but in the sense applied to the present special group of “secreted’’ collagens, this designation refers t o the rather more obvious secretory origin of typical members. The special group would scarcely be recognized as collagens except for the diffraction evidence. They arise not from obviously mesenchymal tissue but from epithelial cells or glands, sometimes of non-mesodermal origin. I n a sense, therefore, they may not deserve the term “collagen,” TABLE I1 “Secreted” Collagens
Phylum
Special name
Source
Reference
Vertebrata Ovokeratin Skate egg capsule (48) Mollusca Byssokeratin Bivalve byssus threads (45, 46, 48) Annelida Earthworm cuticle (11, 174, 182) Nematoda Ascaris cuticle (74, 174) Echinodermata “Ejected filaments” of sea cucumber (8, 9)
although this term has no particular biological significance except through custom. Since the wide-angle criteria indicate close similarity to the more typical collagens, these special members of the class need not be excluded. To date none of the secreted collagens have been recognized as such except through wide-angle diffraction evidence. Small-angle diffraction or electron optical evidence for large structure is lacking, either for the reason that no attempt has been made or because of failure to secure evidence when sought. The situation in this regard may be summarized as follows: The diffraction studies cited in Table I1 were confined to wide angles. In the author’s laboratory preliminary small-angle attempts were at one time unsuccessful with byssus threads. Recently more exhaustive efforts were expended on the skate egg capsule’s ovokeratin. The origin, histology and biochemical and physical properties of this material have been the subject of extensive investigation by Faur6-Fremiet and collaborators (17,70,72,73). The fibrous layers of the capsuleprovidedpreparations which seemed macroscopically well oriented but which yielded, rather reluctantly and with disappointing diffraction orientation, the typical collagen wide-angle pattern. Small-angle methods, satisfactory
92
RICHARD 5. BEAR
for the most refractory cases of Table I, failed to produce clear evidence of large spacings, though faint indications were sometimes suspected. Reed and Rudall(l82) studied the cuticle of the earthworm (AZZoZobophora longa), securing by an ingenious stripping technique electron micrographs of layers of the crossed collagen-type fibrils secreted by columnar epithelial cells. Most important for the present discussion they reported inability to observe in their material the banded appearance characteristic of the usual connective tissue fibrils. There may be other substances related to the secreted collagens. Champetier and FaurB-Fremiet (48) mention also the membrane immediately within the shell of a bird’s egg and a fibrous secretion (koilin) of the avian gizzard. Neither of these, however, clearly diffract in detail like collagen at wide angles. Fraenkel and Rudall (77) point out that the collagenous cuticles of annelids and nematodes are replaced by chitinous ones in arthropods. Insect cuticles were found to have a layered structure composed of chitin and a protein belonging to the P-subgroup of the k-m-e-f family. Chitin, a polymer of acetylglucosamine, is chemically related to the hyaluronate and chondroitin sulfate of connective tissue “ground substance.” The question as to whether the collagenous and chitinous tissues are chemically homologous or analogous provides an interesting problem. The secreted collagens, if indeed they possess significance as a group, at present pose more questions than they illuminate. Are they truly without large structure? Since, as is developed subsequently, the wideangle diffraction of collagen arises largely from a certain type of polypeptide chain configuration, is this the only common feature possessed by the secreted and mesenchymal collagens? Gelatin, a degraded collagen, is unable to form banded fibrils and does not exhibit the diffraction evidence for large structure (18, 5 5 ) . Are the secreted collagens, like gelatin, incomplete collagens, produced by feeble synthetic mechanisms or resulting from degradation of complete collagens? Do they represent separate evolutionary developments of collagen-like polypeptides, accomplished by epithelial cells?
5. The Chemical Significance of the Physical Classification
It is customary to apply the term collagen most properly t o the limited group of mammalian members of the class, although evidence is available for the presence of collagenous substances in all but the simplest animals. Nevertheless, recognizing that it may be improper to include all these chemically diverse substances under one name, investigators have invented the separate ones shown in Tables I and 11.
THE STRUCTURE O F COLLAGEN FIBRILS
93
The diffraction evidence shows clearly that there is a basic structural similarity connecting all of the collagens and provides justification for use of a term to cover the whole class. From the biochemical point of view, however, it may be feared that this action carries along with it an unwarranted assumption of chemical identity. This is not the case, for the physical methods can not be presumed to indicate even that two closely related mammalian collagens are chemically identical. The situation is analogous to the phenomenon of isomorphism encountered in simpler substances, wherein chemically similar but not identical molecular species may replace one another in almost identical crystal lattices. I n the collagens, amino-acid residues may be presumed to be the replaceable units within fibrils which retain a basic type of organization. While chemical constitution will undoubtedly be found eventually a t the basis of collagen configuration, the chemical evidence now available is disappointing. The difficulties are, of course, great, since adequate results would require securing of sufficiently massive and pure specimens, the latter being hard to obtain in the absence of definite chemical criteria of purity. Excellent summaries have been prepared concerning the results of past attempts to arrive at the amino-acid contents of collagens and the gelatins derived therefrom (31, 38). Table I11 collects together the information suitable for a comparison of the class members of predominant present interest. It is desired to consider to what extent the chemical data parallel the results of the physical comparisons of the collagens. There are, of course, many studies of the composition of gelatins derived from mammalian sources and from fish skins. I n Table 111the relatively recently available results on the whole collagens have been cited. Mammalian collagen is represented largely by the analyses summarized by Bowes and Kenten (40), performed on a preparation of ox hide. The data account for nearly 100% of the total substance and have been found in reasonable agreement with acid-base titration curves. This set of values may be used to define a standard collagen with which other substances may be compared. The general finding of the physical methods, namely, that mammalian collagens are similarly constructed, seems in agreement with the existing chemical information. Neuman (163) observed .that a number of collagens from different tissues and from several mammals, including also an avian source, were scarcely distinguishable chemically. A. Pirie et al. (175) obtained paper partition chromatograms of hydrolysates of collagens from ox vitreous humor and cornea, finding them very similar to those yielded by a gelatin hydrolyzate. Bowes and Kenten (42) used
94
RICHARD S. BEAR
TOLE I11 The Chemical Composition of Collagens and Elastin (Grams of residues or of N in 100 g. of protein)
Bovine collagen
Bovine elastin
Glycine Alanine Phenylalanine Leucines Valine Proline Hydroxyproline Glutamic acid Aspartic acid Arginine Lysine Hydroxylysine Hist dine Serine Threonine Methionine Cystine Tyrosine Tryptophan
(40) 19.9 7.6 3.7 4.8 2.9 12.7 12.1 10.0 5.5 7.9 4.0 1.1 0.7 2.7 2.0 0.7 0.0 1.3 0.0
(85, 163, 196) 22.5 15.1 4.4 10.1 12.5 13.4 1.7 2.4 0.35 0.88 0.39
-
-
Total Total N Amide N
99.6 18.6 0.65
87.2 16.9
Amino Acid
0.04 0.68 0.87 0.18 0.28 1.4 0.0
0.04
Ichthyocol (28, 163) 14.9 9.5 1.9 5.2 2.0 13.3 4.0 8.6 5.1 8.4 4.4 0.32 0.91 4.2 2.9 2.2 0.09 0.39 0.0
Herring scales Cornein Spongin (30, 33) (30, 53) 10.9 11.8 0.2 3.3 6.7 6.1
(32) 15.7 2.6 4.7 2.4 7.2
6.0 3.2
5.2 2.6
4.3 0.0 14.6 3.5 4.6 3.0
2.2 6.2 2.4 2.5 0.60 2.5 0.28
0.4
Lo.0
5.5 12.8 0.0
2.6 1.5
6.3
-
-
-
88.3 18.3 0.47
64.8 16.0
44.0 14.7 ~
0.0 54.6 14.6 0.6
~~
Original sources are indicated a t column heads and include only the most recent reports needed t o secure as complete cover'age as possible for these significantly different preparations. All data have been reduced to expression as grams of residues or of N per 100 g. of ash- and moisture-free protein. Bovine collagen and elastin are those of purified hide and ligamentum nuchae, respectively. Ichthyocol samples were derived from fish swim bladders and skins, whose analyses are sufficiently similar to be combined for present purposes. Similarly, available analyses of the corneins, gorgonin of Gorgonia flabellum and plexaurin of Plexaurella dichotoma, have been combined, as have also results derived from the spongins of Hippospongia equina and of the bath sponge. Analyses of herring scales present the most complete available information for this sort of material, which is complex, though all isolated components have presented similar collagen-like patterns of residue composition according to chromatographic studies.
T H E STRUCTURE O F COLLAGEN FIBRILS
95
paper chromatograms and titration curves to show that elastins differ from normal collagen, but that reticulins are essentially collagens. Data of Table I11 agree in some detail with the physical conclusion that elastin is not a collagen. The former is relatively low in residues with polar side chains : hydroxyproline, glutamic and aspartic acids, arginine, lysine, serine, and threonine. It is particularly rich in glycine, alanine, the leucines and valine, residues whose side chains are nonpolar. Elastin and collagen are similar only in their moderately high proline contents and in their low proportions of histidine, cystine, tyrosine, and tryptophan. Just as the small-angle diffraction suggested, one still finds ichthyocol recognizably similar to mammalian collagen according to the chemical results. Here, however, there begin to be indications of significant differences. Neuman (163) pointed out that fish-skin collagen contains appreciably greater quantities of serine, threonine and methionine than does mammalian collagen, and the earlier results of Beveridge and Lucas (28) on hake swim bladder indicated the same thing. It should also be noted that according to the latter authors the hydroxyproline content of swim bladder is relatively low, so much so that the over-all content of hydroxyl-containing residues (serine, threonine, and hydroxyproline) may be significantly lower in ichthyocol than in collagen. It is unfortunate that beyond this in the direction of the more primitive collagens the chemical data are disappointing. Figures for these cases are open to the serious criticism that they are incomplete and were often obtained from impure specimens after dubious pretreatments. Glycine, alanine, serine, and hydroxyproline have been identified qualitatively but not estimated quantitatively in elastoidin (63). I n view of similarities between actinotrichia of teleosts and the ceratotrichia of the elasmobranchs (go), plus their general physical resemblance to teleostean fish scales, the analyses for herring scales of Table I11 may possibly provide the best available indication of what may be expected for these substances. It is probable, however, that these collagens, as usually studied, are far from being pure substances. They contain possibly two proteins in appreciable proportion (80, 32), one soluble and the other a more resistant collagen-like component, and considerable amounts of aminopolysaccharide (63), according to available descriptions of individual cases. If the analyses of the corneins and spongins have any significance it would appear that certain aspects of composition which have often been considered typical of collagen, judging from mammalian examples, may be questionable. For example, cystine which is practically absent in mammalian collagens is said to be represented appreciably in the corneins
96
RICHARD S. BEAR
and spongins; the glutamic acid content of spongin is reported much greater than t ha t of normal collagen; hydroxyproline and proline, often regarded as the distinctive components of familiar collagens, are either absent or diminished in spongin; and both gorgonin and spongin are reported as rich in tyrosine. I n these instances the tyrosine and possibly other residues are halogenated considerably with iodine and bromine (1, 183). Very little is known regarding the chemical composition of the secreted collagens. Analyses of several show reasonable nitrogen values for protein and higher sulfur content than is typical of mammalian collagen (0.9 t o 1.2% in the selachian ovokeratin, 72). According to Baudouy (17) a precursor of ovokeratin (prokeratin) contains sulfhydryl groups which become oxidized in the final stabilization of the selachian egg capsule. One gains the general impression th at the relatively primitive collagens often contain more sulfur than do mammalian collagens, though whether the sources are methionine and carbohydrate sulfates, in contrast to cystine, is not always clear. Lloyd (140) pointed out that in swelling behavior normal collagen, elastoidin and ovokeratin, progressively richer in sulfur, also show increasing resemblance to keratin, whose resistance to swelling is attributed to stabilization of fibrillar structure by disulfide bridges between polypeptide chains. The chemical evidence is disappointing just at the points where it could be most illuminating. When available, it will be most definitive in suggesting the underlying causes for the structural variations of members of the collagen class. This problem is twofold, requiring (a) indications as t o the chemical features common t o all collagens and permitting the molecular chains to adopt the small-spacing arrangement typical of all, and (b) information relative to variations in chemical composition which could account for the larger structural features recognized by small-angle diffraction and electron optics. Astbury’s original definition of the collagen class anticipated only part (a) of the problem, but part (b) seems also essential if the differences between members of the class are t o be understood as well as the similarities. The chemical results to date are at least in general qualitative agreement with the facts disclosed b y small-angle diffraction. Both approaches indicate that t o find collagens which are significantly different from the more common varieties it will be necessary to study a wider range of animal phyla than has been the general practise. Until this is accomplished in more detail than a t present the definitive chemical features which link the collagens will remain elusive. The remaining sections use the best understood collagens to provide
THE STRUCTURE O F COLLAGEN FIBRILS
97
type-models for collagen fibrils. These models illuminate further the nature of the structural problems involved and provide working hypotheses in the light of which the future chemical and physical evidence may be considered in attempting to arrive eventually at complete structures.
111. COLLOIDAL STRUCTURE OF COLLAGEN FIBRILS 1. Optical Evidence
Before the application of the small-angle X-ray diffraction and the electron optical methods of study to collagen, there existed a largely unfilled gap between the histological or microscopic size range and the molecular or chemical level. This intermediate region, which may be termed colloidal, was then represented by extensions of ordinary light optics beyond true resolution, i.e. by the indirect optical methods of ultramicroscopy and polarization optics. The wide-angle X-ray diffraction entered the scene with details of structure so far below those clearly discernible by the light-optical methods, that a great deal of imagination had to be exercised in drawing together the information available a t the lowest levels with that discovered by optical means. The wide-angle diffraction method also has, a t least with collagen, a critical blindspot, in th a t its information (Section IV, 1) deals with only a portion of the structure of the collagen fibril. For these reasons it seems desirable to introduce the present discussion of fibril organization with a consideration of structure in the colloidal size range before progressing to the finer details, even though the historical development reversed this order in some respects. While ultramicroscopy and polarization optics have not offered all that might be desired, they did contribute much to early ideas. The darkfield investigations supplied knowledge of the existence of collagen fibrils, estimates of their diameters, and indications of how they swell (65). Studies of collagen double refraction were more illuminating, although even now many of these observations remain incompletely explained. Relatively recent data of Pfeiffer (172) are used in Fig. 29 to show th a t collagen fibers, like most fibrous proteins and polysaccharides, possess both form and intrinsic types of double refraction. Under all normal conditions the sign of the first kind of double refraction in collagen is positive relative to a n optic axis paralleling the fiber axis. According to optical theory of colloidal systems (79, 188) this means that the macroscopic collagen fiber contains submicroscopic particles of anisodiametric shape with long dimensions aligned with the fiber axis.
98
RICHARD S. BEAR
The identification of the particles is not immediately clear, however. It has been the fashion to speak of these as rodlet micelles, but further discussion below indicates that the only elements which might be involved here are the fibril and the protofibril. The latter are closely packed into fibrils and offer limited and peculiar possibilities for an interpenetration of the immersion fluids used to demonstrate the form double refraction. Perhaps the fibrils supply the major form effects.
I l 4
FIG.29. Double refraction of collagen fibers produced by fibroblast cultures of chick embryo explants, after Pfeiffer (172). Retardation, r, in mp is plotted, as ordinate, against the refractive index, n, of the immersion medium. Curve A is for fibers fixed in Kelly’s fluid (containing formalin), and curve B shows the result of fixation in tannic acid.
The intrinsic double refraction expresses the essentially anisotropic or internally crystalloidal character of the aligned particles themselves. Since formalin-treated fibers show positive intrinsic double refraction relative to the fiber axis (curve A of Fig. 29) this is probably the normal condition. Because of colloidal variability of native collagen fibers the analysis of form and intrinsic components by means of immersion methods requires prior fixation. Von Ebner (61, 187) long ago observed that certain phenols, phenolic acids and higher aldehydes exert a profound effect on collagen double refraction, reversing its sign. Kuntzel (130) demonstrated that the same effect is accomplished by treatment with several pyrogallic tanning
THE STRUCTURE O F COLLAGEN FIBRILS
99
agents (curve B of Fig. 29) and that fibers of this sort owe their reversal largely to a change of the intrinsic double refraction from positive to negative. Formalin, chrome tan, and other reagents do not accomplish the reversal, although many of them cause a decline in the positive magnitude. Kuntzel also observed anomalous immersion curves, with double refraction continually rising along with refractive index of the immersion medium, encountered in fibers treated with pine bark extract and sulfite cellulose. The normal positive sign of the intrinsic double refraction indicates that " intra-micellar " (intrafibrillar) molecular chains run parallel to the fibril axis. The variations produced by the substances described in the last paragraph have been ascribed to intramicellar penetration by the added agents (130), either with oriented absorption (188) or accompanied by chemical reaction (79). The last word has probably not been written about the complex optical properties of collagen. The situation is undoubtedly so difficult that it will be clarified only after more detailed models for fibril structure appear from other directions. At the moment one should conclude from the optical studies that in collagen fibers there are oriented long particles (fibrils), that within these are also regularly arranged thinner elements (protofibrils or molecular chains), and that the particles are penetrable in optically indeterminable ways by chemical agents whose molecules can be of some size. 2. A Fibrillar Model Derived from Small-Angle X-Ray Difraction
After 1942 the diffraction techniques were improved to the point where resolution of consecutive small-angle layer lines became possible (18, 128). Success was secured by the use of finely collimated beams, large specimen-to-film distances, and long exposure times. Economy of radiation was improved by using slits to form ribbon beams, which because of cross-fire of great aperture orthogonal to the fiber axes of specimens, contributed adequate intensity to meridionally separated points of the registering film (see Fig. 2). These slit-camera patterns were capable of exhibiting excellent angular resolution, corresponding to spacings of thousands of Angstroms, along one direction of the resultant pattern, but lacked resolution in the direction paralleling the large dimensions of the slits. Consequently, although the early results were capable of separating the layer lines, they were utterly inadequate for the demonstration of true line shapes, i e . , the manner of distribution of diffracted intensity along individual layer lines. Since collagen appeared to exhibit only a single set of diffractions,
100
RICHARD 5. BEAR
Bolduan and Bear (35) suspected that the collagen fibril was effectively a one-dimensional diff ractor at small angles. They looked for certain theoretical consequences by means of which order deficient systems (i.e., those lacking the full three-dimensional lattices of normal crystalline materials, 21) may be recognized. It became clear that the large structure of the collagen fibril was indeed one-dimensional, with periodic regularity extending solely along the fibril axis. Certain unexpected diffraction behavior was observed, however, which suggested that it would be desirable to view line shape. In the meantime experimental methods employing efficient pinhole diffraction cameras capable of accomplishing this purpose had been developed (34). Preliminary studies (35), using combinations of results obtained with pinhole- and slit-camera photographs, furnished comparisons between the diffraction effects of collagen and those expected of simple models for one-dimensional diffractors (21). After these investigations it became possible t o obtain the full, rather complex small-angle diagram of collagen and to develop a structural model for its general explanation. A brief summary of these studies is available (20). Specimens of collagen kept wet during examination show smallangle diagrams like Figs. 11 and 16. These are particularly simple, but because of overwhelming influence of water (see Section IV, 3) they are not as directly related to the intrafibrillar distribution of protein as are the patterns of dry material. Even the latter often yield patterns such as Figs. 12, 17, 18, and 20, which are not as detailed in diffraction phenomena as are the patterns of Figs. 13, 14, 16, and 19. The latter exhibit a characteristic “fanning” of the layer lines, which also show maxima and minima of intensity. The development of this fanned appearance often depends on the individual collagen source. It was readily obtained in certain kangaroo tendon and elastoidin samples, and many other dry preparations showed at least some tendencies in the same direction. Apparently, the excellence of the development of fanning is a function of variables involving the previous history of a sample, the degree of hydration permitted prior t o final drying under tension, the tension then applied, etc. (23). The dry fibril is, as demonstrated below, a statistical structure, and the fanning results from this fact. It is not surprising, therefore, that to obtain a massive sample in which the fibrillar structure is statistically constant throughout may not always be a matter that is readily controlled, At any rate, these fortunate cases are particularly valuable and furnish the diffraction patterns now described. Note that in a pattern such as Fig. 13 all lines are straight, extending normal to the pattern meridian, and they are equally spaced along
THE STRUCTURE O F COLLAGEN FIBRILS
101
the meridian. The straightness indicates that the fibrils of which the specimen was composed were all effectively rendered parallel by the tension which was applied to the massive sample during preparatory drying to accomplish this purpose. The separation between lines is reciprocal in magnitude to the large axial structure of the fibril, which is found from evidence of this sort to be usually near 640 A. in air-dried specimens. No evidence is observed for the existence of periodic structure across the fibril, i.e., no true row lines are present. One concludes that as far as structure revealed at small angles is concerned the fibril shows only one-dimensional periodic structure, along its axis. There is, however, wide-angle evidence of transverse periodic structure, as developed below, but this is of another smaller order of magnitude. The spectacular fanning feature of the layer lines is seen in their progressive lengthening as the index increases. Line lengths transverse to the meridian are normally expected to be, with one-dimensional diffractors, inversely proportional to the diameter of the fibril over which coherent diffraction to the line occurs, hence constant for all lines yielded by a well organized, cylindrical fibril (21). A continually changing line length can only mean that as smaller structural details, corresponding to lines of greater index, are brought into consideration, a smaller portion of the fibril cross section is able to contribute coherently to each line. The line spreading phenomena clearly indicate internal distortion within each fibril, and the unit of cross sectional area which can be removed by distortion from coherence with the remainder would seem small, since there are no large increments in length from line to line. The above major characteristics of the small-angle diffraction suggested the model for fibrillar structure shown in Fig. 1 (22, 23). The fibril is regarded as composed of much thinner filaments or protofibrils, each of which contains along it various specifically ordered structural details (chemical variations), which are considered alike in all protofibrils. The diffraction characteristics which establish collagen fibrils as onedimensional diffractors with respect to large structure require that the protofibrils be adjusted axially so that equivalent structural details of neighboring ones match transversely. Entire fibrillar cross sections are thus characterized by a degree of constancy of chemical composition and electron density. The forces of attraction between adjacent protofibrils are not everywhere of sufficient strength to maintain rigid matching, however. Consequently, along the fibril one has a periodic variation in degree of matching, corresponding to the periodic alteration in the chemical composition of protofibrillar levels. Wherever matching is maintained one may say
102
RICHARD S. BEAR
that a cross-section (interband) of perfect structure is present. At those levels where matching is relaxed, displacements of protofibril elements across and along the fibril cause development of bands in which distortion or imperfection is present. In attempts to calculate the diffraction expected of the model (22) one of the difficult problems involved the particular manner in which the distortions were to be introduced. A rather general type of statistical displacement of elementary protofibril sections was found satisfactory. Corresponding elements of length along adjacent protofibrils were assumed to be randomly shifted by small amounts, both along the fibril axis and normal to it. Neither the axial (longitudinal) nor transverse (radial) distortion alone accounted for the line shapes observed, a fact which provides possible evidence for longitudinal interconnection of intrafibrillar volume elements into protofibrils. It is difficult with a filamentous structure to displace a small segment normal to the axis of the filament without a t the same time causing some axial distortion, and vice versa. The diffraction effects of the perfect interbands and of the two kinds of distortion a t bands may be separately recognized: (a) perfection accounts for the short line components closely applied to line centers and seen most purely a t the fourth and seventh lines of Fig. 13; (b) longitudinal imperfection is related to the components which are also maximal at line centers but are more extensively spread along the lines, as a t the ninth line; and (c) radial imperfection is responsible for the non-central maxima observed particularly clearly on the eighth and eleventh lines. Only the diffraction components of imperfection expand in length with increase of line index. All of these effects are well accounted for by the application of diffraction theory to the model. The particular way in which the bands and interbands are distributed along the axis of the typical fibril is related to the observed manner of distribution of various types of diffraction component among the several layer lines. It is not possible, however, to derive these features of fibril structure directly from their corresponding diffraction effects because of the well known difficulty of determining the relative phases of the several diffraction lines or line components. The electron optical methods determine the band and interband locations more directly. In spite of these limitations of the diffraction method one may use the various characteristics mentioned above (line separation, length, and shape) to determine size of periodic structure, fibril diameter, and the types of perfection and imperfection present. The collagen small-angle layer lines are remarkable for their sharpness, i e . , their small breadth in the meridional direction, and their large
THE STRUCTURE O F COLLAQEN FIBRILS
103
number. Only the first dozen lines reproduce well in Fig. 13, but on the original films it may be seen that lines continue with increasing length out through those with indices of about twenty. The components of line shape due to imperfection become, however, quite faint and almost indistinguishable at these large indices, with major emphasis then placed on the central components of perfection. Beyond an index of 20, direct pinhole registration has not been accomplished, but slit-camera patterns have shown in some instances lines with indices as high as 32. These facts indicate that the collagen fibrils in the population of an intact tendon are predominantly of the same sized period. It may easily be calculated that consecutive diffraction lines of two different fibrils should fuse near the kth layer line if approximately l / k is the fraction of period by which their fundamental structures differ in size. The same fraction represents also a generously maximal deviation from average intraperiod positions permitted for the structural details of an individual fibril if every detail is to contribute coherent scatter to a kth line. Since the highest k value observed is 32, the fraction representing faithfulness of structure, both within and among fibrils, is about 3%. In absolute size (20 A.), however, this is still large compared to single amino-acid residues. Kaesberg et al. (114) determined the absolute line breadth of the first layer line of the collagen small-angle diagram to be less than 20 seconds radiation), which places a minimal value of roughly of arc (with Fe KLY thirty periodic segments on the effective lengths of the fibrils for coherent diffraction to the first line (calculated from the theory of line breadth of reference 21). Although this result refers to moist specimens, which may be structurally more perfect than dry ones (IV, 3), this estimated length seems consistent also with the number of lines which have been observed with dry specimens. If, for example, within a single fibril the statistical variations in axial position of comparable intrasegment structure reduce the diffraction coherence between adjacent periods a t a rate such that in thirty periods the coherence for diffraction to the first layer line is destroyed, then one would expect that diffraction could not be observed much beyond the thirtieth line. Electron micrographs show that collagen fibrils are much longer than the effective lengths indicated here from the diffraction observations, so that intrafibrillar distortion is clearly necessary to reconcile these facts. Examination of the model suggests other consequences which have not been given quantitative expression as yet. First of all, the fibrillar period (640 A.) may be regarded as a quantity which shows the constancy of magnitude indicated above, from one unit segment to the next, largely as a result of a statistical summation of the projections of protofibrillar
104
RICHARD S . BEAR
elements upon the axis, not because each consecutive segment’s corresponding elements are identically contorted. Increase in the general degree of distortion should shorten the over-all period, while decrease in imperfection will lengthen it. One may also expect that bands, i.e., portions of the fibril which possess distortion, will have protofibrils less densely packed in the fibrillar cross sections than would be the case if perfection existed there. The actual density at a given fibrillar level is a function also of the particular chemical content, but unless protofibrils have compensating variations in diameter at different levels, the bands will be distinguishable as locations at which fibril diameter is greater than average. Because of the poorer packing of protofibrils at band locations, these should be more readily invaded by foreign substances. Large invading molecules may actually increase the distortions of protofibrils at bands or actually “pry open” the structure to increase the spread of distortion along the fibril. On the other hand, molecules which are small compared to protofibril diameters may serve to “smooth out the kinks” and increase perfection, or perhaps to “fill in the holes” and reduce the electron-density fluctuations a t the imperfect cross sections, thus causing similar reduction in the diffraction evidence of imperfection. Chemical attack on the fibril should most easily be accomplished at band locations. The general, qualitative properties just described as expected of the fibrillar model are used to illuminate various aspects of problems which arise subsequently. 3. Digraction Theory for the Model
A complete understanding of the structural significance of the smallangle diffraction diagrams of collagens is difficult to secure without more detailed reference to the diffraction theory applicable to the model developed in the last section. This model presents a kind of structure which, though of a rather general type, as yet is not well known or often encountered. It should apply to other collagens as these are discovered, or as known ones are made to produce the fanned diagrams. Elastoidin, in particular, has already yielded (Fig. 19) an extreme version of this sort of diffraction, whose meaning would have been difficult to determine without prior experience with kangaroo tail tendon. In determining the diffraction expected of structures of the type of the model shown in Fig. 1, parameters Ub and ua were used to express the root mean square longitudinal and radial displacements, respectively, experienced between corresponding points of any pair of adjacent protofibrils (22). In terms of these parameters the fibrillar structure may be described
THE STRUCTUEE OF COLLAGEN FIBRILS
105
as follows: in any cross section a t a given axial level of the fibril the parameters remain constant, though not necessarily equal for the two kinds of distortion. They vary with axial level, being periodic functions of position along the fibril. One distinguishes bands from interbands by arbitrarily letting the former comprise levels at which the u’s are appreciable and the latter those where the a’s are zero. Without entering into details of derivation, the final useful result may be written where IO may be regarded as a proportionality constant connecting the intensity, I k t , observed at the kth layer line at the angular separation [ from its center, with the other experimentally or theoretically interesting quantities. The respective cross sections of total fibril and of individual protofibril are r R 2 and aRo2. The Fk’s are structure factors for the kth layer line, distinguished with subscript p , in the case of the factor involving only axial levels of perfection, and with b and a, respectively, for those concerned with regions of longitudinal and radial imperfection. The 8’s are functions of [ indicating the theoretical line-component shapes, while B is a factor which largely determines the general effect of the degree of distortion on the intensities of both types of imperfection component. The theoretical shape functions (8’s) turn out to be much like those observed experimentally. Expressions for them are :
8, E
[Ji(2rGR5)] Lybk
Sak [k
+ exp [ - ( ~ f i R [ ) ~ ] .
1/(1 fk2/(1
tk2)3
+
tk2)3
(%OuOa/kUOb)
l.
I n these equations i is the reciprocal of the X-radiation wave length and bo is the magnitude of the fibrillar axial period (ca. 640 A.). The c o b and uoaare mean values for the respective displacement parameters averaged over the fibrillar levels a t which distortion occurs. The symbol J 1 signifies the first-order Bessel function whose argument is included after it in parentheses. The S , shape function is independent of k, and if R is large (1000 A. for kangaroo tail tendon, see 35) it falls rapidly with increasing 5 to produce the short line components typical of diffraction by fibrillar perfection. S b k has greatest value at f or t k zero, and declines as t. increases, a t a rate determined by (i%OuOa/kuOb), which becomes less as k increases. Sob has maximum intensity a t 4 = (Icuob/t/Z fboaoa) but falls to zero at
106
RICHARD S. BEAR
6 = 0 (the line center) or a t large 4. These various shape characteristics are easily identified in the experimental observations. The structure factors are unknown a priori and provide the greatest barriers t o attainment of complete analysis of fibril structure. They contain the distributions of matter and of distortion along the fibril axis. Each has the form
where uo is the average distortion and a the actual distortion a t level v along the fibril, the latter expressed in fractions of the axial period (bo). Both of these a’s are further specified b y subscripts b or a, depending upon which type of distortion the structure factor concerns. For perfection, (uo/u)is always unity, and then variations in N ( v ) , the number of electrons per unit cross sectional area of the fibril, provide the only structural details involved. Integration is restricted to the portions of the unit fibrillar period a t which imperfection or perfection exists, whichever is involved. The F’s are complex numbers whose phases are not encountered (because of squaring) in the intensity expression. Consequently, the functions u and N can not be experimentally determined, and in this way the usual limitation of X-ray studies is encountered in the present case. Nevertheless, certain qualitative consequences of the structure factors can be realized. Note that the structure factors are functions of k and will not be the same for all layer lines, explaining why different lines have not the same relative expression of the components of perfection and imperfection. The components of imperfection will lose particularly the contributions to scatter from fibril elements whose a’s are unusually large relative t o u o . The factor B of the intensity expression is given by b o 4 / 2 ( k ~ ~ 0 band )4 is a potent one in determining the influence of the general degree of distortion upon the expression of the diffraction components of imperfection. Note that increase in (kuoa)rapidly diminishes this factor, showing that increased distortion can rapidly wipe out diffraction evidence of imperfection, or that components of imperfection should be lost a t layer lines of high index. The latter effect has been mentioned before in describing the collagen small-angle diagram. The factor B is a large one, if, as is expected, (Tab is a small fraction (0.01) of the size of bo. However, the total coefficient for the diffraction components of imperfection contains also r R o 2 . The corresponding coefficient for the component of perfection is nR2 alone. Since with kangaroo tendon R is about two hundred times Ro ( 2 2 ) , the net effect of
THE STRUCTURE O F COLLAGEN FIBRILS
107
these quantities is to favor, for layer lines above about the fourth, expression of only the components of perfection. One actually observes the components of imperfection, however, hence it must be concluded that normally in dry collagen the structure factors of perfection are small in comparison to those of imperfection, as was indeed assumed in deriving the form of intensity expression given. The theory provides for direct measurement of the ratio of uOa/u06 from the positions of the non-central intensity maxima observed on layer lines with outstanding components of radial imperfection. If trnis the position along the hth line at which the intensity maximum occurs, the equation go. = k
fiGbotm provides means for calculating the ratio of mean displacements. One cannot, however, obtain absolute values of the go’s, except as these may eventually be estimated from the general rate a t which the components of imperfection fade, relative to the components of perfection, as layerline index increases. Because the mathematical treatment did not permit allowing full freedom of range to all the parameters involved, the usefulness of the given intensity expression is limited. Implicit in the theory, but not well expressed in the formulation given, is the possibility that components of perfection may dominate the entire small-angle pattern under two extreme conditions: (a) when the preponderant portion of the fibril lacks distortion, or (b) when the imperfect regions are so heavily distorted as to lose capability of coherent scatter. One may attempt to distinguish these cases by examining the values of bo, which, as mentioned above, should be large when the fibril is relatively perfect and decrease when imperfection becomes serious. This method can not, however, be expected to correlate in all instances with diffraction evidence for degrees of perfection. It is possible that certain fibrillar regions will be so badly distorted as not to contribute much to the structure factors of imperfection while markedly decreasing bo, and other parts of the fibril may remain slightly distorted, contributing to imperfection structure factors without great change in bo. The former kind of distorted level can cause variable bo without leaving an impression on observed diffraction intensities. The theory also presumes both types of imperfection to be present together, and can not be used in the form given when one desires to pass continuously to a situation in which either or both types of imperfection vanish. Nevertheless, one is able to discuss all cases encountered thus
108
RICHARD S. BEAR
far in terms of three types of situation: (a) occasions in which chiefly the diffraction components of perfection are observed, when essentially the first term of the intensity equation applies; (b) instances which show the complete set of phenomena shown in Fig. 13, accounted for by the complete equation; and (c) the case of elastoidin (Fig. 19), whose dry specimens show largely the diffraction components of imperfection, requiring chiefly the part of the theory dealing with distortion. I n effect; because they are not immediately determinable except by observation, the squared structure factors of the intensity equation may be given any relative positive values (including zero) that may be required by a given situation. The case of elastoidin is particularly interesting because it presents a structure in which distortion is very prevalent. Figure 19 shows that, except for the first three layer lines, practically all of the diffraction components are those recognized as typical of radial imperfection alone. One can not conclude, however, that relatively little longitudinal imperfection is present, because the noncentral intensity maxima are even more widely separated from line centers than in kangaroo tendon diagrams. The latter are consistent with a ratio of uOa/uObof about 1.85, whereas the same ratio for elastoidin is 0.77. One concludes that elastoidin fibrils are so badly distorted axially that no diffraction evidence of this type of imperfection remains, except possibly a t the lowest layer lines. The excessive degree of general distortion in dry elastoidin also shows up in its unusually low bo (600 A.) and in the fact that several times the normal exposure to X-radiation is required to yield its small-angle diagram. The shape factors for components of imperfection must be regarded as approximate, though qualitatively correct and becoming more exact at small [ and large k, because of certain approximations of the theoretical treatment. On actual diagrams, such as that of Fig. 13, each line has non-meridional end locations beyond which diffracted intensity of all sorts becomes reduced to practically nothing. The envelopes of all these “cut-offs ” give the impression of linear boundaries for diffraction, extending from the pattern center at about 12” to the meridional direction in the case of kangaroo tendon. The fairly abrupt line cut-offs are not well explained by the theory, but this is a matter which is also influenced by the particular statistical distribution of the distortions assumed over fibril cross sections. Gaussian distributions were employed which failed to take complete account of the steric hindrances to distortion exerted by one protofibril on another or the interconnection of consecutive protofibrillar elements. Nevertheless, the general significance of collagen small-angle diffraction seems capable of being accounted for in fair detail by the theory in its present form.
THE STRUCTURE O F COLLAGEN FIBRILS
109
Hess and Kiessig (107) have suggested that the large axial periods of fibers of polyamides (nylons) and polyesters are related to structural patterns of well ordered regions (the present perfect interbands) alternating with regions of disorder (bands). The diffraction theory developed for collagen should, therefore, have some relation to the synthetic fibers also. Arnett et al. (3) have, indeed, reported for nylon 610 and polythene fibers small-angle diffraction resembling that of collagen, particularly elastoidin, in that non-meridional intensity maxima were observed. They concluded, however, that a second translational period transverse to the fiber axis is present in their material, since it failed to show a certain diffraction consequence (a shift of layer line along the meridion with tilt of the specimen in the X-ray beam), characteristic of one dimensional diffractors (21). Their own published pinhole photographs suggest that this may be a matter of experimental difficulty and not proof of the presence of a transverse spacing. The collagen small-angle diffraction is, however, much more highly developed in terms of the number and sharpness of layer lines than is that of the synthetic fibers. While in the latter the large spacings are a sort of crystallization artifact, with the collagens the large structures are clearly related to complex chemical structure.
4. Electron Optical Evidence Many of the features which appear in the model for the collagen fibril and are designed to account primarily for small-angle diffraction, can be photographed directly in the electron microscope. Indeed, much of the model development was facilitated by the availability of the electron optical evidence. Perhaps the chief difficulties preventing the latter method from immediately arriving at the same result as the former are derived from the fact that electron micrographs are not reliable, with objects of organic composition, for details much below 50 A. in size. The objects themselves are lacking in contrast and often undergo appreciable alteration or deterioration during preparation for examination and while under the action of the electron beam in the microscope. Consequently, the electron optical images can not furnish clear views of some of the details of fibril structure which have been mentioned. Nevertheless, the electron micrographs do yield direct images of fibrils, whose band and interband distributions can be seen in rough outline a t least. Inferences of finer structure can often be gained by noting physical changes resulting from accidents of manipulation. The following paragraphs collect together the observations made in this way and believed to bear structural significance.
110
RICHARD S. BEAR
First of all, the large periodic structure is readily observed in a number of ways. The fibrils, examined with minimal pretreatment (Fig. 25) show alternating dark and light segments designated, respectively, a s A and B by Schmitt et al. (192) and as D and H by Wolpers (207). Two of these, one of each kind, form a single period. Staining methods reveal th at the gross divisions just described contain fine structure (Figs. 21 and 26). Wolpers, using chiefly osmic acid, at first described (209) two subcomponents, the 61 and 6 2 disks, which develop from the broad D level by interposition of a lighter region, designated y . Later he observed the disks each to split into two lamellae (210) but believed the lamellar state to be abnormal, a sign of structural weakness. Wolpers also saw an occasional lightly staining area a t the H region. Schmitt et al. (193) used phosphotungstate to develop essentially the same five stained levels per period. Nutting and Borasky (164) obtained a similar number in formalin-tanned material stained with uranyl acetate, but also indicated that in other cases the number of stained levels can reach as high as six. Schmitt and Gross (191) found it convenient to specify five primary staining levels, a, b, c, d and el of which the first four comprise the original A or D division, while e is at the B or H region. Both b and e show a tendency t o split into two components, so that as many as seven staining components have been observed. Higher numbers are not excluded and the present limit may be largely a n experimental one. The states showing the maximum number of staining components, contrary t o Wolpers, are probably not abnormal developments, since similar numbers of levels characterized by greater than average diameter can be demonstrated by shadowing techniques on fibrils with minimal handling (90). Also, the shadowing methods, which demonstrate unevenness of fibrillar surface by overlay of metal (Cr, Pt, Au) atoms, directed onto the fibrils at a n angle to their supporting film, demonstrate that the levels which can stain most heavily also have greatest diameter (90, 164, 191), with or without prior introduction of stain. At this point, one may identify the levels which take up stain and possess enlarged diameter as the bands of the model. It was pointed out above that imperfection a t band levels should most readily permit entry of stain and cause the less compact organization leading to enhanced diameters. X-Ray evidence exists th at phosphotungstate, commonly used t o bring out the electron optical bands, does in fact choose the imperfections for entry. Graded sorption of this substance by kangaroo tendon specimens caused eventual fading of the diffraction components of imperfection arising from the bands as required b y the diffraction theory when distortion is increased sufficiently by addition of stain.
THE STRUCTURE O F COLLAGEN FIBRILS
111
The band regions attract attention in electron micrographs because of staining and enlarged diameter, but by default leave the intermediate regions of small or non-existent staining and depressed diameter to be interpreted as the more perfectly organized interbands. The size of period disclosed by the electron optics is in agreement with that of small-angle diffraction, except that allowance must be made for a degree of variability observed with the former method. While there seems to be a fair constancy of period along each individual fibril, those of a popu1at)ionmay show considerable variation (90, 164, 191). Schmitt, Hall and Jakus reported individual fibrils of rat tail tendon whose periods measured between 400 and 1000 A., although the average was 644 A. This spread is so wide that it seems doubtful from the sharpness of smallangle diffraction lines that the distribution could be as great in intact tissues. One presumes that it is at least partly due to the manipulations designed to separate fibrils for observation. A regrettable consequence of inadequate electron optical resolution has been the rise of a number of descriptive terms which give an erroneous impression of band and interband structure. For example, Wolpers (210) writes of “disks” and “lamellae.” Porter and Vanamee (177), failing to resolve the detailed band structure, described 270 A. “beads” along fine collagen fibrils, implying that they may be closely related to particulate collagen macromolecules. Wyckoff (215) prefers to call the band elevations “rings,” although alternatively they are termed “disks ” or “threads,” the latter being presumed to run transverse to fibril axes. These descriptions are suitable as long as it is realized that in most instances elements with bounding surfaces normal to fibril axes are only apparent and not basic. (One exception will be cited below in Section V, 1.) Diffraction evidence favors the assumption that the primary structural cleavage of collagen fibrils is longitudinal, continuing down to elements of protofibrillar thinness, resolvable only by means of wideangle diffraction. The nearest electron optical equivalent of this evidence is now summarized. In some collagen fibrils (rat tail tendon) there is a marked tendency for cleavage to develop along fibrils (Fig. 22), so that mechanical damage causes them to fray into finer filaments which show the original banded structure of the parent fibril (192). Others are more tightly integrated (human skin) and are rarely observed to split longitudinally (90). Diameters of collagen fibrils show a continuous range of sizes. They have been observed as small as the resolution of the microscope (specimens frayed by treatment with acetic acid, fine fibrils developed in tissue cultures, 177) up through diameters of several hundred Angstroms (beef spleen and cornea, 192) to quite large thicknesses, thousands of Angstroms
112
RICHARD S. BEAR
(mammalian skin and tendons, 192). Some tissues contain fibrils with a great diversity of diameters (tendons), and those of others show relatively constant widths (skin, 90). Observations of the type briefly covered in the last two paragraphs demonstrate that fibrils must be constituted of thin protofibrils and that depending on circumstances of fibril formation a continuous range of diameters may be achieved. Occasional reports have appeared in which thickenings or coagulations of protofibrillar elements have arisen to provide a punctate appearance across the transverse fibrillar bands or interbands (210) or to develop intrafibrillar columns or filaments (164, 191). These are readily explained as a local consequence of the same type of cleavage between protofibrils which permits their extensive fraying. This incipient cleavage occurs even in skin fibrils (90) in which the lateral forces between protofibrils are unusually strong. The apparently continuous cross striations seen in most cases indicate that the unresolved protofibrils are placed in the fibril in such a way as to align corresponding chemical features normal to the fibril axis, as was assumed for the small-angle diffraction model. This tendency toward transverse matching can be demonstrated in another way. Fibrils contain along their surfaces the same chemical array as do thc! protofibrils, so that even fibrils often occur in parallel bundles with a marked degree of correspondence between locations of band levels (90). This is shown in Fig. 24. The electron optical observations reviewed to this point have most direct bearing on those features of fibrillar structure which also can be derived from small-angle X-ray diffraction. Other important disclosures are more conveniently covered in relation to special problems below. IV. THE COLLAGEN PROTOFIBRIL The evidence discussed above seems consistent with the view that collagen fibrils are composed of very thin elements whose long dimensions are parallel to the fibril axis and whose corresponding periodic chemical features match transversely with varying degrees of perfection. None of the information hitherto cited has been capable of demonstrating the absolute size of the thin dimension of these elements. In anticipation of the conclusion soon reached below, namely, that the elements are as thin as individual polypeptide chains, they have been termed protofibrils, since they represent the thinnest structural prototype capable of carrying the characteristics of collagen. Other protofibrillar properties, such as the content of matter per unit length, the possibilities for extension, and ability to be subdivided into collagen molecules, are also derived in the next few sections, which
113
THE STRUCTURE O F COLLAGEN FIBRILS
make use principally of various types of physical and physical chemical evidence relating to these matters. 1. General Characteristics of Collagen Wide-Angle X-Ray Difraction
The wide-angle patterns yielded by collagens have been the subject of numerous investigations, none of which, however, have led t o proposal of reliable models of collagen structure. This is due t o the fact that organic fibrous systems yield faint and diffuse diffraction phenomena. The maintenance of the X-ray apparatus has often been tedious or expensive, leading the impatient or impecunious scientist t o sacrifice quality of pattern for faster and cheaper results. Even when adequately investigated, collagen will not yield a t wide angles the sharpness and wealth of diffraction detail which is desirable for orthodox crystallographic procedures. One must adopt special methods of approach and be satisfied with but rough outlines of structure. On the other hand, the proposition can be defended that a given diffraction pattern, if analyzed in all details, will eventually yield the structural order that is permissible for the material involved. Mammalian tendons are readily available and provide the best degree of fibrillar orientation, Patterns obtained from them (Figs. 3 and 7) permit observation of the maximum number of diffractions. Instead of tracing the historical development of the subject, this discussion is based on data obtained from the best patterns available t o the author. Table IV shows the average measurements for the positions of individual spot TABLEIV Small Spacings of Collagen (Kangaroo Tail Tendon) k
0 2 4 5 7
Meridian A.
Pseudoperiod
Prominent row line, A.
Secondary row line, A.
Other
A.
(a)
19.1
11.6 11.6
6.7
4.6
20.
12.
6. 6. 6.
4.
9.66 (5.0) 3.97 2.86
19.9
20.
A.
The bold figures represent the stronger wide-angle diffractions commonly observed. The column headed “meridian” indicates the spacing components of t h e layer lines parallel with the fiber axis and, except for the fourth line, most accurately measured by means of the corresponding meridional spots. The “pseudoperiod” column indicates the product of the layer-line index, k , and the related meridional spacing. Other columns give the spacing components of non-meridional diffractions transverse to the axis, on the appropriate layer lines. The equatorial diffractions are those on the line with index zero.
114
RICHARD S. B E A R
or arc centers, compiled from several air-dried specimens whose smallangle patterns had previously proved th at their fibrils were accurately parallel. Bold figures indicate diffractions whose existence has often been recognized because of sufficient intensity, while a few additional but rather faint and diffuse spots of subsequent usefulness are given without emphasis. I n spite of the good jibrillar orientation known t o exist in the specimens, the diffractions show considerable arcing, such as is generally attributed t o imperfect orientation of constituent elements. This indicates that intrafibrillar elements (protofibrils) causing the wide-angle diffraction are not as well oriented as the fibrils. Disregarding the arcing, one notices that the spots are sharpest on and in the direction of the pattern meridian and become increasingly diffuse as they depart from the meridian. Qualities of this general sort are evident on all diffraction patterns of the fibrous proteins, both a t wide and small angles. They are expected of fibrous structures because the individual diffracting elements (fibrils or aggregatcs of protofibrils) are larger and also probably less distorted in the direction of the axis than normal thereto. Indeed, one suspects that in collagen the intrafibrillar units may be packed into fibrillar cross sections with a degree of regular arrangement intermediate between that of molecules in a liquid and that in normal crystals. I n the early days of wide-angle study the dimensions of collagen and gelatin diffractions were used to estimate micellar dimensions (83, 106, 133) on the assumption th at the diffuseness was due entirely to a limitation of the volume domains over which regularity of crystalline structure was maintained. The subsequent experience with collagen small-angle diffraction, as outlined above, indicates widespread distortion within fibrils. One learns from this experience th a t distortion is a relative matter: that which does not greatly influence the length of the first small-angle layer line becomes more obvious a t higher ones. It follows that a distortion sufficient t o influence the large structures in evidence a t any small-angle layer line should have considerably diffused and deteriorated diffraction to wide-angles. The wide-angle diffraction by collagen, diffuse as it may be, must largely arise from the more perfect interbands of the fibril, since the bands contain excessive distortion. The diffuseness arid incomplete orientation of the wide-angle diagram shows that even in interbands appreciable imperfection is present, even though not of a degree to have prevented coherent diffraction by the interbands t o the small-angle line components of perfection. The interband structure, however, shows some of the diffraction characteristics of crystalline substances, since rudimentary layer and row lines are distinguishable in the wide-angle pattern.
THE STRUCTURE OF COLLAGEN FIBRILS
115
2. Pseudoperiodic Axial Structure of Small Size I n this section are considered the layer lines, from which one may obtain estimations of any small sized pseudoperiodicities extending along protofibrils at interbands. Historically the recognition of a sufficient number of layer lines for this purpose with collagen has been difficult. This is because most patterns yield only two clearly evident layer lines: those cited in Table IV as having meridional spacing components and 9.55 A. The strong meridional arcs at spacings of 3.97 and 2.86 A. were not clearly ascribable to layer lines of simple relationship to the others. Failing to recognize more than the two distinct layer lines, investigators have in the past, when they have considered the problem at all, usually adopted one of two procedures. Some have assigned index one to the 9.55 A. line, arriving at values for the period between 9.3 and 9.8 A. (102, 125, 133, 201). Others essentially have assigned index three to the strong 2.86 A. arc, thereby obtaining in the neighborhood of 8.58 A. for the period (generally cited as 8.4 A., 117, 152). The difficulty of correlating the distinct layer lines with the positions of the meridional arcs, led K. H. Meyer (153) t o suggest that the collagen wide-angle pattern may be in reality a dual pattern, resulting from the superposition of two independent diffraction systems, representing two quite different types of structure located a t various situations within the collagen fibrils. In the light of the modern small-angle diffraction and electron optical results, these two kinds of structure might arise from interbands and bands, or alternatively from different interbands, if it be agreed that wide-angle diffraction arises largely from interbands. It is, however, not entirely impossible to relate the distinct layer lines and the prominent meridional arcs. Heraog and Gonnell (105) long ago pointed out that a period of approximately 20 A. is capable of accounting for the collagen wide-angle layer lines. In terms of the measurements of Table IV the basis for a 20 A. pseudoperiod may be stated as follows: The two strong meridional arcs are not hyperbolic intersections of the flat film with widely extended reciprocal disks ( Z l ) , but their circular arc shape indicates clearly that their expression is caused by disorientation. That layer lines exist at these levels is indicated by the fainter diffuse lines with approximately the same axial spacing components. In addition, there is evidence for another layer line at about 5.0 A. These, along with the distinct ones, can be assigned indices 0, 2, 4, 5, and 7 to very nearly account for their relationships. Astbury on one occasion (9), although not citing measurements or attempting index assignments, published an idealized rotation diagram for collagen wide-angle diffraction. It is difficult t o conclude whether
116
RICHARD S. BEAR
all of the spots there shown are reliable, but they very nicely fall on the layer lines for a 20 A. period, within the probable accuracy. Indices of the layer lines there shown would be 0, 2, 3, 4, 5, and 7. Astbury has generally disregarded the desirability of arriving at a period and emphasized instead an interpretation of the strong meridional arc at 2.86 A. (discussed further in Section IV, 6). The 20 A. pseudoperiod has been largely ignored, and the reason may be the following: although the spacings of the more diffuse diffractions of Table IV are not measurable with sufficient accuracy to deny or affirm the index assignment conclusively, the stronger and sharper spots near or on the meridian are not exactly in agreement. It does not seem possible to account for these discrepancies by recourse to any of the conceivable sources for accidental or systematic error. The difficulty involves something like a 5% difference between the period (19.1 A.) calculated for the second layer line and that derived from the seventh (20.0 A .) . Nevertheless, the layer lines so nearly support the 20 A. value that one is inclined to accept it and suggest that these small variations are not unreasonable in the light of the complexities of fibril structure disclosed by the small-angle diffraction and electron optics. Using a variation of Meyer’s idea, one may suppose that the several interbands, or different parts of these, contribute in different degrees to the various layer lines, but that all such localities possess only slightly different pseudoperiods instead of the very different ones implied by Meyer’s original suggestion. This view recognizes that wide-angle diffraction represents only the smaller details of a very large structure. Because of this fact the wideangle layer lines need not be indexed in the exact proportions 0, 2, 4, 5, and 7, but could be given very large indices with nearly these relative values in the small-angle system. Although this cumbersome procedure could be adopted to satisfy crystallographic ideals, it is not a practical one. The wide-angle layer lines are so broad in the meridional direction that they could include something like ten or more layer lines of the large period. Instances have been observed where the large fibril period varies as much as 20 yo or more without accompanying measurable alteration of the wide-angle layer lines (18). Cameras capable of resolving the small-angle layer lines never resolve the wide-angle ones into several components. For these reasons, tendencies to assign large indices to the wide-angle layer lines (10, 127, 128) seem unnecessary, and it would appear more in line with real conditions t o consider the wide-angle system of diffraction as practically independent of the small-angle one. At the same time one must realize that the wide-angle pattern in this case may have
THE STRUCTURE O F COLLAGEN FIBRILS
117
special characteristics because it is the result of several, effectively independent portions of the large fibrillar structure. 3. Structure Transverse to Fibril Axes; Hydration An examination of the wide-angle diffraction row lines indicated in Table I V shows that, in addition t o the meridian, an outstanding one occurs a t a spacing of 11.6 A. transverse to the fibril axis. Still another, which may be a second order of the one just mentioned, is faintly and diffusely represented at 5.7 A. on the equator and a t higher layer lines. The prominent 11.6 A. row line, represented most obviously by its equatorial spot, is famous in the historical development as providing an outstanding example of the “intramicellar l 1 type of swelling (117, 133). When collagen is moistened this row line moves closer to the pattern meridian, corresponding to an increase of the spacing. Detailed studies have disclosed that the prominent row line’s position is a continuous function of humidity; the spacing changes from about 10.4 A. in very dry material t o an upper limit of 15 t o 16 A. with moist collagen (133) and about 17 A. in gelatin (83, 102, 118). The diffractions a t the meridian are not appreciably displaced by hydration, except perhaps for a small decrease in the spacing of the second layer line (to about 9.3 from 9.55 A.). Only the stronger layer lines, with indices 2, 5, and 7, remain visibly represented a t the meridian. The over-all picture presented by collagen wide-angle diffraction, and its alteration by hydration, is that of a structure composed of parallel, thin elements whose longitudinal structure is relatively constant during introduction of water, but whose separation, transverse to their common axial direction, is increased by the hydration. These elements meet the specifications of the protofibrils postulated in earlier sections. The 10 to 15 A., hydration sensitive, spacing furnishes a measure of the distance between planes passing through their centers (not necessarily between their centers directly, cf. Section IV, 6). There are other significant alterations to collagen wide-angle diff raction caused by hydration. Figures 7 and 8 show that the spots of the prominent row line occurring a t the equator and second layer line almost fuse with extensions of their meridional neighbors, the result being an approach t o the diffraction which might be expected of isolated protofibrils. The latter would provide solely meridionally centered, continuous layer lines, such as are expected of long, thin periodic structures (21). A difference between hydrated kangaroo tendon and elastoidin is illustrated in Figs. 9 and 10, which shows the central equatorial streaks extending out t o the location of the normally prominent 15 A. equatorial
118
RICHARD S. BEAR
diffraction. Elastoidin exhibits only the streak, while kangaroo tendon patterns possess the additional bulbous terminal spot corresponding to the spacing mentioned. This equatorial streaking was previously described for elastoidin by Astbury and Lomax (16), who cited i t as evidence of a certain amount of breakdown of regular structure transverse t o fibril axes. Apparently elastoidin, which also showed, a t small diffraction angles, evidence of unusual development of distortion in the dry condition (Section 111, 3), also undergoes extraordinary relaxation of attachment between protofibrils when hydrated. It is of interest to consider the concomitant changes introduced by hydration into the structure seen a t small diffraction angles. Figures 11 and 16 show t ha t a t small angles wet specimens lack diffraction evidence of imperfection and exhibit longer periods (665 t o 680 A.) than when dry (GOO t o 645 A. See also 18, 213). This behavior suggests that the protofibrillar distortions have been partially smoothed out by introduction of water. Empirical support for this view is added b y the observation that the lines with even indices below about ten, which often have the strongest components of imperfection with dry kangaroo tendon specimens, are the ones which lose most of their total intensity after hydration (23). Evidence t hat fibrils, when hydrated, lose to a certain extent the differentiation between band and interband regions is furnished also by electron optical experiments in which surface replicas of moist fibrils were compared with the same fibrils after drying, the latter remaining attached t o the replicas (90). The distinctly corrugated appearance of the dry fibrils is not as evident in the replicas of the moist fibrils (Fig. 23). The wide-angle and small-angle views of the effects of hydration a t first sight seem inconsistent. For example, the relaxation of protofibrillar attachment, which takes place during hydration, scarcely seems compatible with the increased perfcrtion suggested b y t)he small-angle observations. It is possible, however, th at the great amount of water introduced between and around protofibrils largely overbalances the contributions of protein to small-angle diffraction. This seems indicated by the fact t hat substances as different as are kangaroo tendon and elastoidin, when dry at small-angle diffraction (Figs. 13 and 19), become more alike in this diffraction region after moistening (Figs. 11 and 15). The type of perfertion introduced by hydration would seem t o be one which straightens protofibrillar kinks a t bands to provide the longer over-all period. The wide-angle pseudoperiod is not correspondingly enlarged, because it is related to the structure at interbands. At the latter, hydration largely influences protofibrillar separation. Additional non-meridional wide-angle diffraction is very diffuse,
THE STRUCTURE O F COLLAGEN FIBRILS
119
much of it intrinsically unresolvable into separate spots. Most peculiarly characteristic of collagen is the exceedingly diffuse half-halo, which extends above and below the equator at a radius corresponding to a spacing of about 4.6 A., nearly enveloping the equatorial spot (5.7 A.) on its inner margin. Hydration does not greatly displace the half-halo, but it becomes considerably less prominent. It is also sharpened and somewhat resolved into a diffuse row line of spacing about 4.5 A. Two possibilities seem available for this row-line: either (a) it represents finer details of structure within protofibrils, or (b) it indicates that protofibrils have a second transverse direction of approach which allows them to occur in planes as close together as 4.5 A. Either view takes cognizance of the fact that this spacing is essentially independent of changes in the separation of protofibrils brought about by hydration. Relatively large, equatorially oriented spacings, verging into the small-angle range, have occasionally been reported for collagen. One of the most persistently encountered of these was measured a t 23 A. by Gerngross, Herrmann and Abitz (83, l02), but Clark and Schaad (55) found this to be an artifact arising because of the continuous general radiation and the bromine absorption edge of the registering film. The spacings 19.9, 30.0, and 47.6 A. reported by Corey and Wyckoff (58) are also of doubtful validity, since spacings of this sort often arise in tissues because of lipid impurity. In any event, these anomalous cases are weak and by no means constantly represented on all tendon diagrams (18). Other remarks regarding the significance of wide-angle diffraction by collagen and relative to the phenomena of swelling and hydration appear more conveniently in Section V, 3. At the moment it may be said that wide-angle diffraction does not supply sufficient information for direct application of orthodox crystallographic procedures. Nevertheless, successful models for collagen fibril and protofibril organization must account for: (a) the approximately 20.0 A. axial pseudoperiod, and the attachment of particular intensity to the second, fifth and seventh layer lines of this period; (b) the spacing of protofibrils transverse to axes in planes separated by 10.4 to 15 A., depending on degree of hydration; and (c) the peculiar equatorial half-halo or very diffuse row line, whose spacing remains relatively constant at 4.5 to 4.6 A. Unfortunately, most of this information appears to apply to interband structure, and little can be derived regarding details of band organization. 4. Protofibril Extensibility
Massive fibers of collagen are capable of being extended only a little beyond normal relaxed length before rupture occurs. This fact has been widely used as an indication that the thinnest elements (polypeptide
120
RICHARD S. BEAR
chains) are alao nornially in a fairly inextensible condition (9). Electron optical studies have, however, made it abundantly clear that collagen fibrils are far from being as limited in this respect as are the grosser Jibers. For example, in Section 111, 4 it was mentioned that fibrillar periods commonly vary between something like 400 and 1000 A., the latter figure being more than 50% greater than the normal average value of 640 A. It is obvious that massive fibrils will break down when interfibrillar bonding collapses, which may occur before sufficient tension has been communicated t o individual fibrils to cause them to become very greatly stretched (109). The only way which has arisen for experimentally observing the result of application of great stress upon single fibrils has been that of relying upon accidents of electron optical study, wherein the chance breaking of a supporting film occasionally causes individual fibrils t o become caught between retracted parts of the film. An outstanding example is that described by Schmitt et al. (192) and reproduced in Fig. 25. I n this case more than a ninefold increase in the over-all length of the periodic band pattern was observed. However, when the period exceeded 4000 A. the boundaries between A and B regions became indistinct. The fine details of band pattern were not under examination a t this time. A fibril is still massive in diameter compared to a protofibril, and it is the extensibility of the latter which is of fundamental interest. One can expect with an elongating fibril to encounter at least two phenomena in addition t o simple protofibrillar lengthening: (a) a slippage of protofibrils past each other, and possibly (b) the breaking of protofibrils a t weak levels and their retraction therefrom (see further Section V, 1). The second effect will probably be encountered most seriously under the highest extensions, so that i t is desirable t o avoid this by considering only the lower and intermediate degrees of stretching. If protofibrils do not themselves extend but only slip past one another, then unless the slipping elements are very short (which is inconsistent with the markedly filamentous nature of collagen) the band pattern will have been obliterated. Since Schmitt and Gross (191) cite detailed band pattern measurements for fibrils whose periods ranged from 550 t o 1310 A., i t seems inescapable that the protofibrils must be capable of concerted extension. The question arises as t o whether all levels along a fibril or protofibril are capable of equal extension. It has been reported that the H (or B) regions rupture more easily and become thin or lengthen more readily than do the D (or A) regions when a fibril is stretched (209,219), but that lengthening of the period proceeds at expense of material drawn from the
THE STRUCTURE O F COLLAGEN FIBRILS
121
A (or D) regions (192). Finer details relative t o individual bands and interbands are difficult to secure. Although the several kinds of fibrillar or protofibrillar level may conceivably yield variously t o tension a t different stages of extension, it seems likely that all levels are not too different in content. The suggestion of Lelli and Marotta (134) that molecular chains are doubled back on themselves a t bands would place more material there than seems possible. The band elevations in unstretched fibrils are of the order of 15% or less of fibrillar diameter (164). The longitudinal cleavage phenomena indicate t h a t as many protofibrils course through cross section a t bands as a t interbands. Consequently, in dealing with protofibril extensions one should not count on more than about 30% greater extensibility from bands than from interbands. This rough calculation indicates that one might expect much of the normal variability of the large period to come from alterations in the protofibrillar distortion at bands, b u t it would seem that quite another degree of extensibility is required of both bands and interbands t o explain the greatly extended fibril of Schmitt, Hall and Jakus cited above. This must be several fold, since some evidence of band pattern could be seen when the period was as high as 4000 A., although by then some protofibrillar slippage and breakage had undoubtedly occurred. 5. Protofibril Contractility; Thermal Contraction Collagen fibrils are also capable of undergoing considerable shortening. Not infrequently they are observed with axial periods as low as 400 A., and down to a t least 550 A. the details of the band pattern have been discerned (191). There is, however, an even more extensive contraction possible, which has been recognized for almost 150 years as an outstanding characteristic of connective tissue (for the early history see 206). This is the so-called thermal shrinkage, which resides in the collagen component. When collagen fibers are heated to GO'C., more or less, depending on previous history and environmental conditions, they shorten to something like a third or a quarter of their initial length. Various hypotheses have been advanced to account for the thermal contraction. Some of the main lines of thought are briefly described below. An early suggestion was provided by the coagulation theory of Hermann (101), who argued from the similarity of the conditions which cause protein coagulation (denaturation) and collagen shrinkage that the two phenomena are closely alike. Subsequent thermodynamic studies pave shown why this is so: both denaturation (67, 159) and the shortening hrocess (203) require preliminary, rate-determining, activation processes
122
RICHARD S. BEAR
involving large increases in heat content (AH”) and in entropy (AS*). Because of large AH*, the temperature coefficients are also large and the reactions proceed with an apparent but not actual abruptness near a temperature I”, which is about 60°C. under minimal swelling conditions in aqueous environments. Then TAS* nearly balances AH* to yield small enough free energy change, AF*, to permit appreciable activation to occur. From the kinetic standpoint the large AH* is accounted for by assuming that the large molecuIes invohed require for activation the breaking of a number of relatively weak bonds which normally serve to hold the protein in its native configuration. In the case of the collagen fibril this would mean the breaking of links between the neighboring protofibrils, and possibly also the release of intra-protofibrillar bonds tending t o stiffen them individually, in sufficient number to permit the subsequent relaxation. The large AS* corresponds to the increased freedom of motion permitted the protofibrillar segments after this release. Weir (203) gives typical values for the thermodynamic quantities involved in the shrinkage of kangaroo tail tendon in water a t 60°C. : AH* = 141 kcal./mole, AS* = 349 cal./mole-degree, and AF* = AH* TAS* = 24.7 kcal./mole. These figures apply to a “molecule” whose weight is not known with certainty. On the assumption that the subsequent contraction process is adiabatic (AH = 0) and that the AH* of the activation process is the only heat change involved in a reaction whose over-all heat absorbed has been measured to be 12.5 cal./g., Weir calculates a “molerular weight” of 11,300 for collagen. This need not be the weight of a complete collagen molecule or protofibril, but can represent the size of segment which is required to be activated a t one time. The assumption that the contraction process is ideally adiabatic, while perhaps not entirely permissible practically, seems indicated by modern theory of the behavior of molecular chains, which pictures these as undergoing, when freed of restraints, a sort of segmental diffusion, much like the adiabatic expansion of an ideal gas into a vacuum (155). I n the case of the molecular chain, it diffuses to the most probable, randomly coiled configuration, which is much less asymmetric, hence shorter, than a n initially extended chain. Because rubber most nearly presents this ideal behavior, those fibers which develop increased tension (a measure of the tendency toward assumption of the contracted form) when held isometrically under conditions of increasing temperature (favoring the “diffusion”) are said t o be “rubber-like.” Most normal elastic solids upon stress are strained from some stable structure and relax as the temperature is raised. Protein molecular chains often show more nearly the normal elastic
THE STRUCTURE OF COLLAGEN FIBRILS
123
behavior, presumably because of a tendency to prefer particular folded configurations. This is the case with members of the k-m-e-f group of fibrous proteins a t moderate temperatures and swelling (212), but after increased swelling and above a sort of “rubbery melting point,” which depends on the conditions, the preferred or quasicrystalline configurations “melt” and rubberlike behavior ensues (62, 141, 156). The collagens have not been as extensively studied in these respects as might be desired, but some evidence is available. Wohlisch and de Rochemont (206) believed contraction t o be a reversible transformation, a t constant temperature, between two states I and 11, corresponding, respectively, to the solid or crystalline state and the melted or dissolved condition of ordinary crystalloids. Somewhat earlier it had been suggested that during contraction collagen is transformed into gelatin. Working initially with beef tendon, Wohlisch did not immediately identify collagen I1 with gelatin, though he did so after experience with the more easily gelatinized rat tail tendon (205). Cherbuliez, et al. (51) believed it possible to distinguish the reversible “melting l 1 of the polypeptide chains from the irreversible hydrolysis which produces gelatin. These authors compared pig achilles tendon, elastoidin, and ichthyocol in these respects, finding that ichthyocol most readily yields gelatin, while elastoidin does so only with difficulty (see also 69). Elastoidin furnishes particularly suitable material for study of thermal shrinkage because it provides an approach t o a reversible transformation between states I and I1 without gelatinization complications. Native elastoidin (state I) shows many characteristics of a collagen. It has a quasicrystalline structure yielding the usual wide-angle diffraction (16, 47) and a small-angle diagram which, while distinctive, as discussed in Section 111, 3, is still recognizably that of a collagen. Some of the other physical and chemical properties, such as the positive uniaxial double refraction (69), the resistance to tryptic hydrolysis (69) and a thermoelastic behavior like that of normal solids, with negative temperature coefficient (173) also indicate that elastoidin I has a n ordered structure. Transformation t o state I1 occurs in water near 62”C., but a restraining tension raises the temperature required and swelling media lower it (75). The elastoidin 11, which is produced and is stable a t temperatures above the transition point, has quite new properties. It is contractile and rubbery, shortening spontaneously to 27y0 of the original length (69) ; its thermoelastic coefficient is large and positive, attesting t o its rubberlike nature (173). In shortening it loses its ordered structure, yielding a single wide-angle diffraction halo characteristic of a n amor-
124
RICHARD S. BEAR
phous state (47). The double refraction diminishes and is said to change sign from positive to negative (69), though this behavior is inconsistent with a n amorphous condition and may result from internal strains set up in an accompanying non-collagen matrix (see further below). The resistance t o trypsin is markedly diminished (69). When cooled, elastoidin I1 shows signs of reversion t o state I (69). It regains spontaneously over half of the initial length at 20"C., the positive double refraction and the normal wide-angle diffraction return, and there is even some regain of resistance t o trypsin. Elastoidin differs from the general run of collagens in respect to these reversionary tendencies. Gerngross and Katz (81), in a n early study of tanned and untanned avian tendons, found that the untanned preparations shorten and remain shortened, losing their normal wide-angle diffraction (except for the diffuse halo characteristic of the amorphous condition) unless forcibly re-extended. Formalin-treated, or lightly chrome-tanned specimens, on the other hand, showed the behavior, originally described by Ewald (66), of spontaneous partial regain of length, and, correspondingly, the collagen wide-angle pattern also returned. A similar formalin-fixed rat tail tendon has been carried through the shortening and spontaneous lengthening cycle, with examination of the ability of the final preparation t o diffract a t small angles (18). A diagram was obtained consisting solely of the first three layer lines of a period which had been reduced to 550 A., corresponding closely t o the macroscopically observed net loss of 13% in length. Normally, without the formalin stabilization, the tendon loses all discrete small-angle diffraction, even after forcible re-extension. This result throws light on the meaning of the formalin stabilization, which must link the protofibrils of each fibril so that they do not become hopelessly disarranged during the shortening phase and can return on lengthening t o reconstitute, in part a t least, the original fibrillar structure. The native stabilization of elastoidin is probably produced in another fashion; as mentioned in Section 11, 5 , elastoidin is impure, and the fibrillar stabilization may result from confinement in a n amorphous matrix which tends to keep the protofibrils of a fibril together. The stabilized collagens show the effect of restraints; for example, Gerngross and Katz found their untanned tendons could shorten 89% at 60 t o 65°C. but the formalin-treated ones contracted only about 60% at 85 t o 90°C. In elastoidin the stabilization is not intrafibrillar, as seen by its normal shortening temperature (62"C.), but it is restrained extrafibrillarly from shortening more than 73%. After the first cycle of shrinkage and re-extension, these cases may be carried with more com-
THE STRUCTURE O F COLLAGEN FIBRILS
125
plete reversibility through subsequent cycles, though the transformation temperature is lower than initially and the condition in state I is never that of the native collagen. The essentially cyclic nature of the contraction and re-extension process has been emphasized because it illustrates particularly clearly the facts that (a) the cyclic phenomena are primarily protofibrillar properties, the fibril stabilization being only the means by which the massive sample behaves as do the individual protofibrils; and th a t (b) the normal state of protofibrils, ie. a t ordinary temperatures in the absence of agents producing excessive swelling, is not the contracted, random form, but the extended condition, possessing the specific configuration which yields the collagen wide-angle diffraction pattern when several protofibrils are arranged properly with their long axes parallel. I n other words, the amorphous state I1 is not the normally stable condition; the ordered configuration of state I is preferred. The situation is very much like that in gelatin sols and gels. Katz, Derksen and Bon (116, 119) have shown that a freshly prepared gelatin solution has amorphous structure, which on cooling transforms into a gel yielding the collagen wide-angle diffraction diagram. I n this sense, the earlier investigators were correct in comparing the thermal contraction of collagen t o a transformation into gelatin, even though the actual production of gelatin may not be demonstrable. An early rival of Hermann’s coagulation theory was the swelling theory of Engelmann (64), and, indeed, the problem of thermal contraction can not be dismissed without consideration of the relation of swelling to the phenomenon. It has long been known th a t a marked swelling of the collagen is not obviously involved in contraction, although there is a small increase in volume. Nevertheless, prior swelling of the collagen is quite generally effective in reducing the temperature required t o effect it (123, 135, 199). Perhaps the most elegant way of expressing the effects of the swelling reagents is by way ‘of the changes in the thermodynamic quantities involved. Weir (203) gives data of this sort for kangaroo tail tendon, showing t ha t acids, alkalies and neutral salts generally decrease A H * , AS* and AF*, except th at concentrated salts increase AF*. Tans, which generally oppose swelling, increase AF*, but in ways dependent on the tan: most of them reduce AS* more than AH*, but chrome increases both of these markedly. Lowered AF*, of course, means a lesser temperature required for activation leading t o contraction, and decreased AH* corresponds t o a broader temperature interval for measurable contraction rates. The interpretation of the thermodynamic quantities in terms of
126
RICHARD S. BEAR
detailed kinetic pictures is not easy, however, since the activation process involves not only collagen but the water and other reagents present. The thermodynamic quantities represent the net result of energy and entropy changes in all components of the system involved in activation. This may be made clearer by a consideration of the quantities cited above for contraction of tendon in normal aqueous environment. According to the chemical analyses of mammalian collagen (Table VI) the average residue weight is 93. If a protofibrillar segment of weight 1 1,300 is involved in the activation process, then about 121 residues participate. On the assumption that the activation energy, AH* = 141 kcal., is largely used to break weak hydrogen bonds of 5 kcal. each within the protein structure, 28 bonds are involved, or one for every four to five residues. In actual fact, however, a rupture of more than this number is probably required before the structure is sufficiently weakened. While these are being broken within the protein framework, others between protein and water are undoubtedly being formed, so that AH* is the result of heat absorbed in releasing intraprotein links and heat evolved as hydration occurs. It is well known that removal of water greatly increases the shrinkage temperature. Thermodynamic properties of the contraction by relatively anhydrous collagen would come closer t o furnishing information directly interpretable in terms of what is happening to the protein itself. At different places in a massive sample of collagen the activation necessary to initiate contraction is not always the same. Thus, shortening can arise at local regions while others remain unaffected. The local contraction probably also distorts neighboring fibrils to reduce their activation requirement, so that the contraction spreads from the initial loci. This phenomenon is illustrated particularly clearly by Weir (203). Small-angle diffraction studies disclose that as shortening progresses, during stages short of completion the sample still exhibits fading normal diffraction rather than continuous other change (18, 214). The contracted portions lose ability to diffract at small angles while the uncontracted ones retain it, and the proportion of the latter diminishes gradually. Contributions of electron optics to this problem have been disappointingly meager, possibly because of difficulties in properly arranging for observation of contracted individual fibrils. Nutting and Borasky (164) briefly indicated that heating produces fibrils with greatly enlarged diameters and probably shortened lengths, though banding is lost. The clearest indication of shortening was seen in fibrils with a spiral contra(.tion of a central undispersed core produced by shortening of the outer layers. Wolpers (211) also found a tendency for band levels to disperse
THE STRUCTURE O F COLLAGEN FIBRILS
127
transversely t o fibril axes, and when the fibrils were sufficiently shortcned all visible structure was lost. 6. The Content per Unit of ProtoJLibrillarLength I n the last two sections compelling evidence has been cited t o indicate that collagen protofibrils, while normally in a configuration characterized by the 640 A. period when dry, may be forcibly extended t o several times this length, or they can spontaneously contract to a fraction of their normal length when thermally activated. Models for the detailed protofibrillar structure must contain enough polypeptide chain per unit of length t o allow, in particular, the remarkable extensions observed. The wide-angle diffraction data do not clearly enough indicate the exact interpretation of protofibrillar dimensions normal t o the long axis to permit one to derive unequivocally the amount of protein which density considerations permit to exist per unit axial projection. Nevertheless, a consideration of density limitations in relation t o other protofibrillar properties produces some interesting suggestions, as now follow. The most widely circulated views relative to this matter are those of Astbury. His suggestions are as follows, revised to expression in terms of magnitudes employed in this paper. The argument essentially starts with the assumption th a t the 10.4 and 4.6 A. spacings of collagen, transverse t o fibril axes (Section IV, 3) are analogous, respectively, t o the “side-chain l 1 and “back-bone ” spacings of &keratin structures (13). I n this way two dimensions of a protofibril (polypeptide chain) would be established normal t o the long axis. Assuming that these dimensions may not be strictly orthogonal, but include a n unknown angle p, one readily calculates th a t a single residue of average weight 93 would satisfy a typical protein density of 1.35 if it occupied an axial length of about 2.4 sin ,8 A. (15). Thus, the length per residue is of the order of 2.86 A., the prominent meridional spacing of collagen, which has, therefore, been regarded as having this significance. Proceeding in the reverse direction, one calculates a density of 1.12/sin @ for collagen or gelatin if the average residue occupies a volume 10.4 X 4.6 x 2.86 sin ,B A.3 (see 13). While this density value is low, uncertainties regarding p and the value of the very diffuse 4.6 A. spacing could allow correction upward. According t o the present suggestions that wide-angle diffraction may be related to only a part of the total structure, it is conceivable that the average residue may not be involved, and heavier than average residues at interbands could also increase the calculated density. As subsequent considerations show, however (see
128
RICHARD S . BEAR
Section V, 2) i t seems likely that interbands contain lighter than average residues. A maximum extension of about 3.64 A. per residue is permitted (57) for polypeptide chains arranged as shown in Fig. 30a. Silk yields the highest observed value of 3.5 A. (157), and normally the /3-keratins show still lower values, 3.1 to 3.3 A. (14). While it is possible to suppose th a t silk and p-keratin have polypeptide chains which are somewhat crumpled and contracted from maximum extension, collagen would require rather special treatment for explanation of a residue extension as low as 2.86 A. /
OC\
T-!
HN\
To
"
R-CH
7.0
OC\ HN
\ a
b
C
FIG.30. Polypeptide chain configurations.
Astbury has from time t o time proposed different reasons for this low value in collagen. His most recent model (9) is shown in Fig. 30b. A partial cis configuration of the main chain causes shortening, at the same time allowing the prevalent pyrrolidine (of proline and hydroxyproline) and glycine (hydrogen) side-chains (see Section V, 2) to be on the same side of the main chain, opposite the other bulky ones, t o avoid side-chain congestion. Attempts to stretch this model are said to cause it to coil upon itself, which is believed to explain the supposed inability of collagen polypeptide chains to be extended. There are several objections which might be made to the Astbury
THE STRUCTURE O F COLLAGEN FIBRILS
129
model. The extension per residue is not invariable but only averages 2.86 A., and one would expect the distance between pyrrolidine residues (3 X 2.86 = 8.58 A.), rather than 2.86 A. itself, to be heavily represented in the wide-angle diagram. The meridional spacing nearest t o 8.6 A. is observed a t 9.5 A.; indeed, no serious attempt appears to have been made to consider what fiber pseudoperiod may be present in collagen and how the model could explain it. Huggins (112) has also made more general types of objection to the models proposed by Astbury. The main polypeptide atom chain of the Astbury model does not possess a screw axis of symmetry. The unbalanced forces on opposite sides of such a chain would tend to prevent it from readily adopting a straight course, bending it continuously in some direction. Huggins also believed it likely that corresponding atoms along the main chain should be surrounded with similar configurations of atoms (except possibly for differences in side chains), which the Astbury model does not accomplish. The high content of pyrrole ring side chains in mammalian collagen may obviate the validity of the second objection, but the first is less easily escaped. Huggins offered the coiled model for a polypeptide chain shown in Fig. 30c, believing this suitable for collagen. Such chains were placed side by side in a thick plane, joined by hydrogen bonds such as > N-H . . . 0 = C <. The distance between chain centers in the layer thus formed corresponded to 4.5 A., or essentially the spacing of the wide-angle diagram’s diffuse equatorial half-halo. The layers were regarded as stacked, with a separation of 11 A. determined by the side chains which extend above and below the layer planes. This corresponds to the spacing of the moisture-sensitive, prominent row line of the diffraction pattern. The most serious objection to the Huggins model is found in the axial pseudoperiod of 5.8 A., which represents the projected length of two residues. While half this value is close to the 2.86 A. meridional spacing, the discussion of Section IV, 2 shows that other prominent layer lines are present which bear no simple relation to 5.8 A. Ambrose and Elliott (2) have recently suggested that a slightly modified form of the Huggins model can account for the facts regarding collagen wide-angle diffraction and infrared dichroism. The objections applicable to the Huggins model may be made again here. The modifications were designed largely to satisfy the infrared data, which suggest an apparent preponderance of orientation of carbonyl and imino bonds of the main chain normal to the fiber axis. The Astbury, Huggins and Ambrose-Elliott models give essentially the same interpretation to the spacings 11, 4.6, and 2.9 A., differing only in the manner in which the main chain is contracted to present one residue
130
RICHARD S. B E A R
per 2.86 A. of axial projection, or in other details. Full extension could produce a t the most only about 28% increase beyond normal protobrillar length. The evidence presented in Section IV, 4 seems, therefore, t o exclude these models. Similar objection can be raised with regard t o the models of Zahn (218), which are essentially the result of exploration attempting to preserve the 2.86 A. spacing as the projected length of a single residue. The interpretation pf the 2.86 A. spacing as the length of a single residue in collagen has been fixed so firmly in the thinking of investigators that a number of recent, otherwise provocative trends of thought have become seriously confused. For example, Mustacchi (160) realizing the significance of the great extensibility of collagen fibrils and protofibrils (micelles, to him), was led, in part by a misunderstanding of the significance of X-ray diffraction data, to postulate unnecessarily complex deductions regarding the numbers of chain molecubs in collagen " micelles." Pauling and Corey (170), following lines of thopght initially very similar t o those developed below, were also induced by a belief in the essential inextensibility of collagen molecular chains, t o devise a model for collagen which fails to account for the actual extensibility of collagen protofibrils. It seems, therefore, desirable to attempt a n entirely new departure in arriking a t a conception of the protofibrillar configuration. The discussion of small-angle diffraction and electron optical evidence of pdqeding sections has disclosed that there is every reason to believe that fibrils are cylindrical objects, with reasonably- good integration over entire cross sections (particularly a t interbands whfch yield. the wide-angle diffraction) , and that longitudinal cleavage does-not produce laminar but essentially cylindrical filaments or protofibrils, as far as can be seen. For the new approach, suppose that the protofibril, like thedbril, is essentially cylindrical. I n the fibril these thin cylinders will achieve closest packing if they adopt a hexagonal arrangement of centers a t cross sections as shown in Fig. 31. The spacing between planes through centers is assumed t o be the 10.4 A. spacing of dry collagen. This corresponds t o an actual separation between protofibrils of 10.4/cos 30" or 12.0 A. The cross sectional area available to a single protofibril is 124.8 A.2 It is then easily determined that the average residue of weight 9 3 accounts for a density of 1.35 if the residue has an axial projection of about 0.91 A. According t o a model of this sort, a meridional spacing of 2.86 A. would correspond more nearly to the projected length of three residues. at 0.95 A. per residue, rather than t o that of a single residue. Indeed, examination of the measured layer-line spacings of Table I V discloses that a variability of residue extension between 0.91 and 0.95 A., with the lower value for the second layer line and the higher extreme for the
THE STRUCTURE OF COLLAGEN FIBRILS
131
10.44
20.0A
FIG.31. Illustrating helical models for the collagen polypeptide chhin (protofibril). Wn the left is the skeleton of the helix, with the locatioq of atoms filled in for the bottom turn. The numbefed carbon (solid black) atoms of the skeleton represent the 21 plaqes (CY carbon atoms) at which residue side chains are attached in one full period of the helix. The number of turns per period illustrated is 4, indicated in general a s n. The diameter of the helix skeleton is d. The upper right part of the drawing indicates cross sectional plans for three helices and their mode of packing in the collagen fibril. I n one helix the atoms of a single turn are shown; in another, lines directly joining the Q carbon atoms of the residues are given for a full period; and in .the third, their general circular locus of projection is indicated. Dotted lines show rough domains within which penetration of side-chains from outside each helix is largely prohibited by van der Waals repulsions of the helix skeleton. At the lower right is shown one unrolled coil, with one residue indicated in detail. Stippled circles represent nit,Eogen atoms, -small open circles are hydrogen, large open circles oxygen, and shaded circles show the general direction for emergence of R groups (side chains).
132
RICHARD 6. BEAR
seventh, indicates that 21 residues would be accommodated in the 19.1 to 20.0 A. pseudoperiod. To account for small residue projections of this sort a polypeptide chain must clearly be coiled. When this is done according to the above specifications one obtains a model which is capable of being extended maximally almost fourfold, though somewhat less will be realizable if the pyrrolidine residues prevent full extension (see 167). Coiled models for polypeptide chains have been proposed with increasing frequency in recent years. Even before Huggins, Taylor (197) had suggested a coiled model suitable for a keratin, with three residues per turn. Bragg et al. (43), in particular, have explored a wide variety of possible folded chain configurations, emphasizing the usefulness, in providing fold stabilization, of hydrogen bonds between intrachain > NH and >CO groups, separated in position along the chain but by folding or coiling brought into the proximity required for bonding. Pauling and Corey (169, 171) believe that, beyond the widely recognized, extended chain model of silk or p-keratin, it is possible t o narrow the conceivable coiled configurations to forms in which the bond distances and angles are rigorously those indicated by studies of the crystal structures of amino acids and dipeptides. They call attention to two spiral models, one with 3.7 residues per turn at 1.47 A. axial projection per residue, which is suggested for @-keratin, and another with 5.1 residue per turn at 0.96 A. per residue, which is proposed for supercontracted keratin. The Pauling-Corey models have the attractive features that >NH groups are located with respect to > CO groups of adjacent coil turns so that all of these may stabilize the coil internally by hydrogen bonding, and that the side chains may be attached t o the coiled main chain in such a way as to “bristle out” from the coil, normal to its axis. A pyrrolidine residue may easily be inserted into the coil, with only the absence of a > NH * * 0 = C < hydrogen bond becoming necessary. The second of the simpler Pauling-Corey helical models is easily adaptable to collagen with minor adjustments. Thus, a t 0.95 A. per residue, 21 residues would form four full coil turns of 5.25 residues each to account for the 20.0 A. pseudoperiod of collagen. Because this particular coil satisfies the bond distances and angles encountered in the amino acids and dipeptides it has been used in Fig. 31 as a particular structure for consideration. Pauling and Corey (170) developed, however, a special model for collagen involving an intertwined coil of three chains, each of which has a relatively steep pitch to account for the supposed inextensibility of collagen. The studies of the infrared dichroism of collagen by Ambrose and Elliott (2) may furnish additional requirements for the arrangement of
THE STRUCTURE OF COLLAGEN FIBRILS
133
atoms in collagen helices. As this information now stands it favors the three-chain helix of Pauling and Corey and is against the particular one-chain (7)helix favored above. Only the former provides carbonyl and imino bonds normal t o the helix axis. One must, however, be cautious in translating optical properties of gross fibers into properties of ultramicroscopic fibrils or subfibrillar parts. Theoretical and experimental difficulties result from the marked anisotropy of surface distribution in fibers. Also, the models are derived to account for interband structure primarily, and the considerable band portions of the fibrils may involve distorted other arrangements. At the moment it seems wise not to regard details of possible helical models as too firmly fixed. It may be expected, however, that the number of residues per turn in helical models, if such be applicable t o collagen, cannot be chosen without upper and lower limits. Table V is designed to indicate some dimensions of coils with three to six turns per pseudoperiod. It is impossible to entertain a projection per turn less than 4 to 5 A. The rough coil diameters, calculated according to Fig. 31, also cannot be much less than 5 A. Consequently, as columns 2 and 4 show, the coil with six turns per pseudoperiod may be excluded. There must also be sufficient space between adjacent coils, whose separation was calculated above to be 12.0 A. in dry collagen, to accommodate the side chains. Column 5 shows that the coil with three turns offers an excessively tight squeeze in this respect (see Section V, 2), so that all but the models with about four to five turns per pseudoperiod may undoubtedly be eliminated . Qualitative inspection of the coiled models suggests that they will not be in obvious disagreement with collagen wide-angle diffraction. With regard to the distribution of intensity between layer and row lines the following remarks seem pertinent. The electron density originating from the coiled main chain alone projects rather evenly along the protofibril axis, particularly if the coil is not perfectly constructed and is distorted in minor ways, because, for example, of strains imposed on it by differences in the side chains appended along it. Consequently, the intensity distribution along the meridian of the wide-angle diffraction pattern will be largely a result of the way in which side chains distribute their own matter along the coil. For example, the outstanding strength of the 2.86 A. meridional arc rather suggests that every third side-chain may be particularly massive or unusually light. Since little is known (see, however, Section V, 2) about the details of distribution of side chains along the collagen main chain, no conclusive examination of the model relative t o the meridional distribution of wide-angle diffraction is possible a t the moment.
134
RICHARD S. BEAR
If the protofibrils, i.e., the coiled polypeptide chains with their bristling side chains, do indeed pack in hexagonal or pseudohexagonal fashion over fibrillar cross sections, as illustrated in Fig. 31, this should be most obvious in the production of diffraction row lines with characteristic spatial relat,ionships. Unfortunately, the intra- and interprotofibrillar distortions, which are not unexpected because of side-chain variability, are sufficiently bad to cause the rapid deterioration in diffraction encountered as the distance from the meridian increases. Thus, the number of observed row lines is insufficient to test the hexagonal packing features of the proposed structure. TABLE V Approximate Dimensions of Polypeptide Chain Coils No. of turns per pseudo period
n 3 4 5 6
Axial projection No. of residues Main-chain coil Available per turn per turn diameter0 between coilsh 20 n 6.67 A. 5.00 4.00 3.33 -
21 n 7.00 5.25 4.20 3.50
d
10 - d
7.50 A. 5.63 4.50 3.75
2 . 5 A. 4.4 5.5 6.3
* The distance from the point of attachment of a side chain to an adjacent helix skeleton, allowing about 2 A. for the van der Waal's exclusion space of the latter. These distances are immediately appropriate for comparison with the side-chain lengths of Table VI. Projection of the electron density of coil and appended side chains upon a fibrillar cross section does not offer sufficiently sharp concentrations of density to lead to an expectation that much more than the 10.4 A. spacing between the major planes through protofibrillar centers will be strongly evident. The observed 4.6 A. spacing may well represent roughly the distances between atoms across the main chain coil, which are rendered variable in position and diffract almost independently in the very diffuse fashion observed, because of the irregularity of coil mentioned above. It is clear that any reliance on a specific coiled protofibrillar configuration can not be based at the moment on the wide-angle diffraction alone, but must be gained from a synthesis of all pertinent protofibrillar properties, such as has been attempted above.
THE STRUCTURE O F COLLAGEN FIBRILS
135
V. THE COLLAGEN MOLECULE The unit of connective tissue structure is the fibril, and up to this point the intrafibrillar unit has been regarded as the protofibril. The latter elements are t o be conceived as many periods (i.e., many times 640 A.) in length. Two possible views may be taken of the significance of their periodic structure: either the primary valence connection of the protofibril is interrupted at regular points t o provide what may be described as a linear string of molecules, or the periodic structure is that of a repeated pattern of amino-acid residues extending along a n indefinitely long polypeptide chain. There is some evidence that the former of these views is correct, as will now be shown. It must be remembered, however, th a t with complex substances, such as are collagens, one will not immediately require the identity of composition that is required in defining simpler organic species, demanding only that certain general characteristics, such as the number and relative arrangement of constituent residues, shall be fairly constant. One may hope t o maintain the original requirement of all molecules, namely, t ha t the atoms of each individual be connected with primary valence bonds, although failing in demonstration of this difficult criterion, one may ask that the molecule be at least a particle of fairly constant size or weight. In view of the possibility th a t there may be a few covalent links occasionally set up between normally separable units, one will also wish t o disregard these accidental unions in defining the simplest molecular unit. 1. Description of the Molecule The simplest concept of a molecule suggested by the model for the collagen fibril presented here is th at it comprises a portion of a protofibril which includes the entire width of the latter, i.e., the cross section of a coiled single polypeptide chain, and a length which is either equal t o the full period of 640 A. (in dry fibrils) or some simple fraction of this. The general lack of longitudinal symmetry in fibrils (see Section V, 2) makes it difficult t o conceive, on simple crystallographic grounds, th a t the molecular length could be less than a full period unless there is more than one variety of molecule present, each occupying some characteristic fraction of the period in a single protofibril. A number of electron optical observations furnish rather convincing indications that once in each fibrillar period there is a weak level a t which protofibrillar continuity may be interrupted under sufficient stress. For example, fibrils have been observed t o break off squarely, normal to
136
RICHARD 9. BEAR
axes; indeed, in the case of human skin fibrils, whose longitudinal cleavage is infrequent, fibrillar fracture most often occurs in this manner (90). I n rare instances, with replicated skin fibrils that remained attached t o their replicas, portions of segments one period long were observed to be disIocated from the remainder, as reproduced in Fig. 27. Highly stretched fibrils show exaggerated differences in band elevation (209), with particularly pronounced thinning a t B regions and less near the A band group. The result is an appearance resembling nodose strings (see Fig. 25). In favorable instances one observes that the thinnest locations are between bands e and a. Isolated statements regarding the a band make it of interest in this connection. It is said t o appear occasionally as a doublet (90) and, while often a t one of the thickest levels of the fibril, it may be notched (191). It is also near or at the a band that the square fractures mentioned above occur. The hypothesis may be offered that near or a t the a band may be located the weak level which these phenomena indicate t o be present once in every fibrillar .period. The excessive thinness of stretched fibrils a t locations toward e from a may result from a retraction in this direction of parted protofibrils (see Section IV, 4), which thus add matter t o the A region and remove it from the B locality. In subsequent considerations it is useful t o keep in mind the weight and degree of polymerization of a molecule which is of protofibrillar width and one full period in length. Section IV, 6 developed the concept of a protofibril with residues of average weight 93 projecting 0.91 t o 0.95 A. each along the axis a t interbands. While the complete period contains bands as well, unless a t these the fibril is very different with respect t o diameter and content per unit of axial length, one can estimate that about 670 to 700 residues would be found in the full molecule, corresponding t o a total weight of 63,000 to 65,000. One might expect minimally degraded gelatins t o have molecular weights of this order of size. A wide variety of molecular weights have been reported for gelatins, many of them being unreliable. In a recent osmotic pressure study of carefully demineralized gelatin Pouradier et al. (179) concluded that gelatin molecules, a t least in dilute solution near the isoelectric point (pH 4.7) and above 30°C., are molecularly dispersed, since relatively constant molecular weights in the range 61,000 to 67,000 were obtained in various environments. The values measured were a number average, however, of a population from which fractions with weights between 15,000 and 250,000 were drawn. The picture of this gelatin population, according t o present suggestions, would be that of protofibrillar fragments ranging from those smaller than the collagen molecule t o several times it, but averaging about the size of one.
THE STRUCTURE O F COLLAGEN FIBRILS
137
Scatchard et al. (185) in studying the progressive acid degradation of gelatin, proposed that collagen contains molecular chains along which there are unusually weak, acid-labile bonds. An ideal “parent gelatin” molecule is prepared from collagen when these links are broken, and further, less rapid, hydrolytic scission of the parent gelatin molecules accounts for the polydisperse molecular populations of degraded gelatins. Sedimentation studies were consistent with a molecular chain diameter of 17 A., approximately independent of degradation, and viscosity studies suggested a length of 800 A. for the undegraded parent gelatin molecule containing something like 1170 residues. A more direct approach to collagen molecules is secured by studying the solutions of collagen obtained by subjecting tendons or connective tissue to weak organic acids. Nageotte (161) generally employed rat tail tendon and dilute acetic acid t o obtain clear, filterable solutions which gel when dialyzed against pure water but reform coarser fibrils when salt is added (94). According to Wyckoff and Corey the recrystallized fibrils diffract in the same way as do native ones to both wide and small angles. Electron optical studies also show typical collagen band patterns in the reconstituted fibrils (191, 192). The dispersed molecules themselves are difficult t o examine, presumably because their thin dimension is below the resolution of electron microscopes. Collagens from various sources can be dispersed in differing degrees. Leplat (136) examined different animal tendons, employing formic, propionic, citric and even hydrochloric acids for their partial solution. FaurB-Fremiet and Baudouy (71) found ichthyocol completely soluble in 0.0004 N formic acid and formamide, the latter also dispersing mammalian tendons partially. Salo (184) has reported molecular shapes and sizes, similar to those of Scatchard et al. for parent gelatin, from viscosity studies of ichthyocol dispersed in warm dilute acetic acid solutions containing some salt. Orekhovich and colleagues (165, 176) have been studying intensively in recent years a soluble collagen, unfortunately termed “procollagen,” which is widely distributed in animal connective tissues. A typical source is fresh ox skin, which is extracted with dilute organic acids, notably citrate buffered at pH 3 to 4. Dialysis of the extracts against water yields excellent fibrils. The amino-acid content of ox procollagen resembles that of normal mammalian collagens (52). Bresler et al. ( 4 4 ) calculated from sedimentation and diffusion measurements that the molecular weight of procollagen is 70,000 3500. Further, they estimated that the molecule has an asymmetry corresponding t o that of a cylinder of diameter 16.7 A. and length 380 A. The degree of polymerization was calculated to be about 600,’which at an
138
RICHARD S. BEAR
assumed (excessive) length per residue of 4 A. yielded a total polypeptide chain length of ca. 2400 A. These authors suggested th a t the polypeptide must normally be coiled to give the observed length of the dispersed molecule. Evidence was cited for two types of aggregation of these asymmetric molecules, caused by addition of salt. They attach end t o end or by aposition of their long sides. Highberger et al. (110) studied, electron optically, dialyzed citrate extracts from rats, calf and steer skins and rat-tail tendons, obtaining two types of fibrils, some with structure similar t o native collagen fibrils and others with interesting, hitherto unobserved features. The latter are like normal collagen fibrils in exhibiting solely a n axially periodic structure but are distinguished by having an unusually large period of 2000 t o 2600 A. (see Fig. 28). The type of reconstituted fibril obtained by dialysis can be controlled by adding varying amounts of serum mucoprotein, the long spacing variety being favored by increasing amounts of mucoprotein. ’ The correspondence of the very large periods with the length of the fully extended polypeptide chain of the procollagen molecule, a s calculated by Bresler ~t at., suggests that the new form is one in which the normally coiled configurations have, for some reason, become unrolled, This possibility is in agreement with calculations based on the type of coil suggested by X-ray diffraction and density considerations given above, for the collagen molecule made of 670 t o 700 residues would be capable of attaining a length of about 2300 to 2500 A. when stretched t o a n axial projection per residue of 3.5 A. The various determinations and calculations of molecular size and shape for parent gelatin, procollagen and collagen molecules cited above are, a t present, only in rough agreement. They are, however, suggestive of interesting possibilities which may be tested by further investigation. A major difficulty at the moment stems from lack of knowledge of the factors influencing the homogeneity of molecular populations with respect to molecular or particle weight and to polypeptide chain configuration. While valuable information can be gained from study of collagen and gelatin dispersions, there are indications that these are not identical. In addition t o the possibility of hydrolytic degradation of the size of gelatin molecules relative t o the original collagen protofibrils or molecules, gelatin configurations are generally more random, possibly because of subjection t o environments (thermal and chemical) which have destroyed in part the more highly anisodiametric shapes of the parent particles. Kuhn (129) considered gelatin molecules in warm aqueous solutions to be randomly contracted to near spherical shape, and t o yield weak streaming double refraction because of stretching under the shearing
THE STRUCTURE O F COLLAGEN FIBRILS
139
action of their flowing environment. Thaureaux (198) has more recently compared acid solutions of ichthyocol and of rat tail tendon collagen, before and after heating, with respect to double refraction of flow and optical rotation. The unheated preparations showed relatively ready streaming orientation as well as high and relatively stable optical activity, compared to the same solutions after heating. The latter exhibited the phenomenon of mutarotation, which Katz, Derksen and Bon (116, 119) linked to the gradual recrystallization of gelatin in the transformation from sol t o gel (see Section IV, 5 ) . Gelatin molecules in solution, while possibly more or less randomly coiled after heating, probably undergo, on cooling and aging, at least partial return to the regularity of coiling characteristic of the state which is capable of yielding the collagen wide-angle diffraction (Section IV, 6). Hess and Trogus (108) emphasized the fact that gelatin films do not require stretching to produce the oriented collagen diagram, which can be obtained merely by passing the incident X-ray beam parallel with the film surface. Astbury (5) corrected an error in their statement of the natural orientation in these films, finding that the protein chains are parallel with the surface. Pouradier and colleagues presented evidence that gelatin molecules are chemically linear chain molecules rather than nets (178, 179), but that depending on ionic strength and pH of the solutions these chains are contracted or extended. For example, at the isoelectric point the specific viscosity is at a minimum and increases two to three times when the pH is shifted toward acid or alkaline reactions.
2. Distribution of Side Chains One of the important facts regarding collagen fibril structure disclosed by electron optical studies is that the periodic band-interband patterns are polarized (193). I n other words, as can be seen in Fig. 21, the fibrils possess no cross sectional planes of symmetry; each has “head” and “tail ” directions which are distinguishable. In the light of the description for intrafibrillar structure which has been presented above, the polarity has the following consequences. Since polypeptide chains, however folded in a regular manner, also possess similar polarity, the fibrillar polarity must arise because all constituent protofibrils are parallel and similarly directed. The model postulates also that the neighboring protofibrils or molecules are located axially so that corresponding features of chemical structure match across the fibril to form the bands and interbands. The important result emerges that the periodic band-interband pattern of the fibril is a direct indication,
140
RICHARD S. B E A R
however imperfect, of the linear distribution of chemical variations along the coiled molecules. In the coiled chain proposed in Section IV, 6, except where the imino acids (pyrrolidines) introduce minor variations, the main-chain coil is essentially uniform. The side-chain appendages bristling from the coil furnish the features which distinguish levels along the molecule. It is of interest to attempt derivation of the ways in which these are distributed. As discussed earlier in Section 11, 5 , only for typical mammalian collagen is chemical evidence reasonably complete. Table VI adds to the chemical data summarized by Bowes and Kenten (40)several sidechain properties which are useful in considering these appendages in relation to space filling characteristics and ability to scatter X-rays. Consider the meaning of the figures in Table VI in terms of the model for the collagen molecule herein proposed. The column showing mole fractions of residues may also be interpreted as approximating figures for the fractions of the 640 A. molecular length along which the several residues are found, if each residue has nearly the same axial projection. The residues are also grouped in the Table according t o types in relation t o expected capabilities of their side chains for forming unions joining neighboring molecules (142). Thus, the “nonpolar” side chains, largely hydrocarbon in nature, are capable only of the van der Waals’ attractions possessed by all matter. The hydroxyl-containing side chains are possible hydrogen bond formers, as are also the acidic, amide, and basic ones, although the acidic and basic side chains can also enter into salt, i.e. ionic, types of linkage. A few of the residues occur in outstanding proportions. Thus, some 64% of the total belong to the nonpolar classification, of which glycine, alanine, and proline provide the greater part. These three are not entirely similar in their roles in protein structures, since, for example, glycine has a negligible side chain and proline has the unusual pyrrole ring. Nevertheless, their side chains are alike in being relatively inert chemically and in possessing lengths and densities, with respect to both mass and electrons, which are smaller than the average for the whole collagen population. The nonpolar side chains are so prevalent that it seems unlikely that any very large segment of the molecule can escape having a fair proportion of them. Glycine, in particular, occupies one-third the axial length of the molecule, and it has been suggested that every third residue along the collagen chain may be glycine (26, 27). This has ,received experimental support from the isolation from a gelatin hydrolysate of the tripeptide lysylprolyglycine by Grassman and Riederle (86). As reference to Table VI shows, this evidence could only refer to a fraction of
THE STRUCTURE O F COLLAGEN FIBRILS
141
about 9 % of the molecular length (three times the mole fraction of lysine). If one may presume that the same distribution with respect t o glycine occurs everywhere, the strength of the 2.86 A. wide-angle diffraction (Section IV, 6 ) would be accounted for, since glycine is outstandingly different from all other residues in the low mass and electron content of its side chain. The hydroxylic residues account for about 16% of the molecular length. Hydroxyproline and serine are predominant among these. On the whole, this group has about average side-chain lengths and somewhat larger than average densities. The acidic, amide and basic residues supply the remaining 20% of side chains, which are generally unusually long and, with the exception of lysine, also distinctly above average in mass and electron densities. Reference to column 5 of Table V shows that the acidic, amide, and basic side chains would find difficulty in fitting between the coiled main chains of adjacent protofibrils, since only 4 to 6 A. seems available in dry collagen. Consequently, the possibility is suggested that the fibrillar bands, with distorted protofibrillar organization, are levels at which these residues with extraordinarily long side chains are located. Interbands would then contain predominantly the hydroxylic side chains of nearer average length, which can fit more easily between protofibrils. At both bands and interbands the polar side chains would be diluted by the prevalent nonpolar ones. In this connection, note that the distance between centers of protofibrils in wet specimens is about 15/cos 30" or 17 A. Consequently, the hydrated fibril has something like 9 to 15 A. between main chain coils. At interbands this allows actually two of the average hydroxylic side chains to separate the main chains, but little more than one of the longer basic side chains if full extension is to be maintained. During drying, the interband hydroxylic side-chains have some chance of allowing reasonably regular packing of coils to be maintained, but the dehydration makes the band levels uncomfortably restricted for the longer side chains. Distortion at band levels then results to yield the small-angle diffraction and electron optical evidences of imperfection. The factors governing the band and interband distributions, hence the placement of various types of side-chains along the molecule's main chain, are not known as yet. The periodicity theory, originally suggested by Bergmann and Niemann (26), has not received successful support for proteins in general, although Kratky (127, 128) believes that with collagen the strong diffraction layer lines at small angles bear simple relationship to the fractions of the period separating individual residues of the same kind. Thus, according to the mole fraction data
TABLEVI Properties of Collagen Side Chains
Density
Amino Acid Glycine Prolie Alanine Leucines Valine Phenylalanine Methionine Nonpolar Hydroxyproline Serine Threonine Tyrosine Hydroxylic
Side Chain
Mole fraction Weight Electrons
Molal volume Mass cc. g./cc.
0.325 0.122 0.100 0.039 0.027 0.023 0.005
1.o 42.1 15.0 57.1 43.1 91.1 75.1
24 9 33 25 49 41
3.3 43.7 19.6 68.5 52.2 80.9 64.7
0.30 0.96 0.77 0.83 0.83 1.13 1.16
0.30 0.55 0.46 0.48 0.48 0.61 0.63
2.3 4.2 3.4 5.9 4.6 7.2 7.6
&
0.641
20.0
12
22.8
0.56
0.41
3.4
p
-CHZ-CHOH-CH~-CHzOH -CHOH-CHj -CHz-CsH,-OH
0.100 0.030 0.018 0.007
58.1 31 . O 45.1 107.1
32 17 25 57
47.1 21.9 38.2 83.2
1.23 1.42 1.18 1.29
0.68 0.78 0.65 0.69
5.3 4.4 4.6 8.4
Total
0.155
53.6
29
1.26
0.70
5.2
-H -CHz-CHz-CH?-CHI -CaHo -CH = (CH3)2 -CHz-CsH6 -(CHz)z-S-CH3
Total
__
1
Acids and Amides
-(CHZ)z-COOH -CHz-COOH -(CHz)I-CO-N Hz -CH z-CO-N Hz
Total
x
W
42.8 ~
Glutamic acid Aspartic acid Glutamine Aaparagine
Electron Length electrons/cc. A.
~
~~
~
0.072 0.044 (0.043)
73.1 59.0 (72.1) (58.0)
39 31 (39) (31)
51.7 35.4 (53.3) (37.6)
1.41 1.67 (1.35) (1.54)
0.75 0.88
(0.73) (0.82)
6.9 5.7 (6.9) (5.7)
0.116
67.8
36
45.5
1.51
0.80
6.4
Arginine Lysine OH-lysine Histidine
-(CHz)a-NH-C(NHp) -(CH2)4NH2 -(CHz) sCHOH-NH2 -CH2--CaN zH 3
=NH
0.047 0.029 0.007
0.005
100.1 i2.1 88.1 81.1
55 41 49 43
76.1 72.1 74.4 58.9
1.32 1.oo 1.18 1.38
0.72 0.57 0.66 0.73
0.088
88.9
49
73.7
1 20
0.67
8.8
1.ooo
36.8
21
33.0
0.83
0.52
4.5
~
Bases Totals all residues
Total
~~
9.4 8.3 8.3 7.1
~~~~~~
~
Mole fractions are calculated from data summarized by Bowes and Kenten (40),assuming analysis complete. Mold volumes are apparent values given by Cohn and Edsall (56). Lengths are estimated for fully stretched side chains from connection with main chain to end, including van der Waal’s radii for end atoms; proline and hydroxyproline side chains are portions of pyrrole rings, effective “lengths” being thereby shortened. “Weights” and “electrons” are sums, respectively, of atomic weights and numbers for the atoms of the side chain. “Densities” are calculated from appropriate other columns. Group and final totals are sums of mole fractions or of other properties weighted within the group according to mole fractions; the property totals are, therefore, essentially averages for the groups. Amides are excluded from the totals. The addition of 56, the weight of the -CO-CH-XHatoms common to all residues (except the pyrrolidines), to the average side-chain weight (36.8), yields 92.8, very nearly the average residue weight.
144
RICHARD 8 . BEAR
of Table VI, Iysine and hydroxylysine account for 3.6% of the residues of a molecule, whicah Kratky believed t o contain 220 residues, each a t 2.86 A. of axial projection. In this molecule there would be about 8 residues which, if distributed evenly along the 640 A. period, would contribute diffracted intensity to something like the eighth small-angle layer line. The case of the lysines has been cited because evidence is available showing the effect of the removal of their c-amino groups, by deamination, on the intensity of small-angle diffraction. (Actually, the examination was made by tagging the basic groups with hexametaphosphate before and after deamination.) The sixth rather than the eighth line was notably changed, being reduced by the deamination from its normally outstanding intensity (3G). Kratky's theory is better upheld if the lysine alone is considered, whereupon the sixth line is expected to be involved. This agreement cannot, however, be taken as valid evidence for his views, since the theoretical predictions depend on the molecular size assumed. With a molecule containing 670 to 700 residues, as proposed herein, the lysine residues would be expected to contribute to about the twentieth layer line. In this case the experimental evidence rather suggests that lysine groups may be bunched at about the band levels (see further below) rather than being evenly distributed along the structure. The difficulties facing theoretical attempts to relate intense layer lines of small index with the distribution of individual amino-acid residues are similar to those which are encountered in calculations of apparent protein molecular weights from chemical composition. In both cases reliance is largely placed on the least frequent residues, which are the ones whose data are, in fact, least reliable. While it may not as yet be possible t o locate polar residues very precisely, their rough division between the two categories of fibrillar level, characterized by hydroxylic side chains or by the larger other polar groups, seems capable of some support. A few indications of the type of available evidence follow. The interbands, which in the model are pictured as the more compact and stable fibrillar levels, provide the main ties between protofibrils. Gustavson (91) has pointed out that introduction of 0-naphthalene sulfonic acid, by combining with basic groups and thus eliminating possibility of salt links, nevertheless does not greatly reduce the hydrothermal stability of mammalian collagen. The shrinkage temperature is decreased only 10-12"C., whereas urea, generally expected to break hydrogen bonds, lowers this temperature about 40°C. Hydrogen-bond formation is, therefore, important in determining fibril stability. The predominant location of these at interbands seems appropriate.
THX STRUCTURE O F COLLAGEN FIBRILS
145
Similar arguments relative t o the importance of hydrogen bonding in collagen fibrillar structure are also derived from a comparison of the shrinkage temperatures and compositions of mammalian and fish skin collagens. The latter, which possesses (see Section 11, 5) fewer hydroxylic side chains than does the former, also has less hydrothermal stability (92). Band structure is more open and accessible than that of interbands to chemical invasion, as long as fibrils remain intact. Consequently, such experiments as the introduction of P-naphthalene sulfonic acid without destruction of fibrillar structure, indicated by the lack of great reduction in shrinkage temperature, provide presumptive evidence that the basic side chains are accessible a t band localities. More directly, the heavy anions of phosphotungstic acid can be seen in electron micrographs to locate a t bands, possibly because of attraction to and accessibility of the basic side chains. Diffraction evidence that this stain and also hexametaphosphate enter bands almost exclusively was also obtained (23). It is not the intention t o review here in detail the great body of evidence which studies of tanning and other phases of collagen chemistry may offer regarding the distribution of side chains. Gustavson (93) and Bowes and Kenten (41) have recently commented in illuminating fashion regarding the accessibility of the reactive groups of collagen t o tanning agents, acids and bases, esterifying agents, etc. While the reactions of tanning agents are undoubtedly complex and involve t o some degree all types of reactive groups of the collagen, it is worthwhile t o note how generally the importance of the acidic and basic side chains has been indicated. Formaldehyde is said t o involve principally the lysine +amino groups. Vegetable tannins, which contain numerous phenolic and a few ionic groups, impart hydrothermal stability to collagen mostly by interaction with basic groups. The active forms of chrome in tanning processes are believed to be the cationic complexes. The carboxyls of the acidic side-chains are particularly able t o coordinate within the complexes and be thus linked with the chain-like chrome aggregates. It would seem of some significance that those substances which are recognized as useful tanning agents should show particular ability t o provide intrafibrillar cross linking through the acidic and basic sidechains. The latter would be most available if located a t bands, SO that present structural hypotheses provide a satisfactory solution a t once of the accessibility problem and of the general qualities which are necessary in a reagent to provide successful tanning action. According to studies of Schmitt and Gross (191) electron micrographs show clearly that chrome tan accumulates particularly strongly at bands
146
RICHARD 8. DEAR
(Fig. 26), though some general light penetration throughout the fibril is apparent. In addition the relative locations of bands in chrometanned fibrils are particularly simple: they appear to be six in number, grouped as three pairs whose centers divide the axial period into thirds. This is of interest in relation to the small-angle diffraction evidence for similarly tanned kangaroo tendon (Fig. 16), whose third, sixth and ninth lines are unusually prominent, so th at alterations which particularly affect these lines may be associated with events occurring a t bands. Similar diffraction effects were noted with hexametaphosphate, formaldehyde and vegetable-tanned specimens, though to a less notable degree (36). Wide-angle diffraction also offers suggestion that tanning agents predominantly enter hands and less into interbands. For many years it has been a puzzling fact th at the wide-angle diffraction patterns of collagens are often not greatly influenced by tanning, unless the tanning is excessive (139), whereupon a destruction of structure akin to that obtained by mechanical disintegration may occur (111). Herzog (103) early expressed the view th at a tan is introduced into “intermicellar” spaces and does not interfere with the more crystalline (micellar) parts of the collagen. In terms of the present model, the intermicellar spaces are, in part, the hands and the micelles correspond to interbands (20). This view accounts particularly well for the fact that tanned collagen can still absorb water normally a t low humidities (115), and th a t the wide-angle diffraction evidence of “intramicellar ” swelling is still obtainablc in tanned specimens (132). Specimens of moderately chrome-tanned kangaroo tendon showed, by yielding small-angle diffraction patterns like Fig. 16, which is clearly altered from t hat of untreated material (Fig. 13), th a t there had been fibrillar penetration by the tan (36). Nevertheless, the wide-angle diagram had not been greatly affected, beyond a reduction in the degree of arcing of some of the sharper diffractions. This result is consistent with the view that the tans enter bands most easily and do not immediately influence greatly the interband structure, except as there may be penetration a t interband boundaries where protofibrillar orientation is less perfect. Excessive tanning may well cause destructive action at interbands by wedging apart tlhe fibrillar structure, starting a t bands. 3. Swelling Phenomena
When the uptake of water exceeds that normally associated with native collagen fibrils the phenomena of hydration discussed in Section IV, 3 are no longer adequate to describe the situation. Under these exaggerated conditions of swelling the large amount of water present
THE STRUCTURE O F COLLAGEN FIBRILS
147
makes it difficult to employ the diffraction and electron optical methods significantly. One is forced t o rely largely on deductive reasoning from macroscopic and microscopic observations to obtain concepts of the detailed events. Nevertheless, the model which has been developed from the physical methods can be examined t o see whether it is consistent with the swelliiig properties of collagen. It is customary to divide the swelling phenomena into two types (144) : (a) the swelling in acids and bases, which is largely understandable as being associated with Dolman effects arising from ionizable acidic and basic groups according to the original suggestions of Proctor and Wilson (180); and (b) the swelling induced by neutral salt solutions, which may be described as a lyophylic or Hofmeister effect. Much of the observed swelling behavior of collagen is understandable on the basis of the theoretical expectations in these two very different cases. For example, the Donrian theory explains the minimum in swelling in salt free media a t the neutral isoelectric range; the rise of swelling in acid and alkaline environments t o maximal values, followed by decline a t very high or low p H; and the effects of salts in decreasing acid or alkaline swelling. It is not able t o account for the fact th a t in weak acids or strong bases a t very low or high pH, respectively, somewhat more swelling is observed than expected (39). It seems likely that the latter phenomena are related t o relaxation of the intrafibrillar structural links which oppose swelling in any event and without which the fibrils would completely disperse. It has been suggested that these additional effects are in large part caused by superimposed lyophylic swelling, similar t o that described below (144). The Donnan or osmotic swelling involves chiefly the acidic and basic groups. On the other hand, the lyophylic swelling is capable of involving both of the major types of cross link. Ions mask the charges of the ionized groups carried by the protein, relaxing the normal salt linkages (93, 145). Neutral molecules or hydrated ions can also break hydrogenbond cross links between protofibrils, much as they do in causing swelling of the polymeric carbohydrates (122), whose unions are largely of this type (4). I n the above brief description of the swelling phenomena little has appeared which has much t o do with the steric relationships of the ionizable and hydrogen-bond-forming side chains of collagen fibrils. A possible exception may be found in the requirement that if salt links are to be of significance in collagen the acidic and basic groups must be located a t the same localities and not distributed a t random through the fibril. The present model places these together at fibrillar bands. A consequence of swelling which does depend on structure may be
148
RICHARD S . BEAR
stated as follows. Donnan swelling, which influences only ionic groups, and lyotropic swelling, which influences both kinds, should differ in the external shapes which they cause a collagen fibril or fiber t o develop. Kuntzel (131, 133) has emphasized that in Donnan or osmotic swelling the fibers shorten and thicken, while in neutral swelling they chiefly thicken and may even increase slightly in length ( 5 % ) . Because of this difference, tension applied t o a swelling fiber opposes the Donnan swelling but not the lyotropic swelling (144). X-Ray diffraction or electron optical evidence of the changes in shape of individual fibrils caused by swelling is meager, but Nutting and Borasky (164) reported th at acid-swollen fibrils become thickened and exhibit axial periods shortened to 540 A, Small-angle diffraction, on the other hand, shows in neutral aqueous environments (Section IV, 3) an increase of fibrillar period from 640 to about 670 A., though the effects of salt solutions have not been studied. Kuntzel objectedto the fringedmicellar model of Herrmann et al. (102) because it seemed incapable of accounting for the lengthening accompanying neutral swelling. When pure water is applied to collagen it causes limited increase in the distance between polypeptide chains a t ‘‘ micelles” (the present interbands) and also enters the “ intermicellar” spaces (bands, spaces between fibrils, etc.). According to the fringe model the iiitermicellar space could only be enlarged by increasing the fibril or fiber thickness, a t the same time causing shortening. The prcsent model accounts for the lengthening, since the water which penetrates between protofibrils a t bands allows greater space for the longer side chains, which then do not distort the main chain coil from its preferred straight condition. According t o the views of Lloyd (144) neutral lyophylic swelling of collagen may be largely interfibrillar, since the swollen samples do not take on the transparency encountered under acid and alkaline conditions. The present model does not distinguish the structure of the fibrillar surface from t hat of individual protofibrils in the interior, except th a t the reactive groups between fibrils are more easily available to the swelling environment. If Lloyd was correct, then one must assume th a t the hydrated lyotropic ions do not readily penetrate the fibrils. The fact that the shrinkage temperature is reduced under such conditions (123) is against this conclusion, however. In any event, the neutral salt environment should not produce such pronounced changes in length as do acid or basic swelling, since in the lyotropic swelling a t all locations, band or interbands, internally or externally, there is more widespread relaxation of interprotofibrillar or inter-fibrillar unions. Figure 32 serves t o illustrate diagrammatically how the present model would account for the changes in fibril shape under the two different
,
149
THE STRUCTURE O F COLLAGEN FIBRILS
swelling conditions. Undoubtedly, when bands are highly enlarged as in Donnan swelling, they will cause considerable rupture of interband structure. The most persistent links within interbands will determine the extent to which the structure can swell or dissolve.
a
b
C
FIG.32. Diagrammatic representation of the difference between a, a dry fibril; b, a fibril swelling in water a t neutrality; and c, the result of acid swelling. Only polar side chains are shown, with open-circled heads representing uncharged side chains, and - signs designating correspondingly charged heads or ions, and H indicating hydrogen ions. The long charged side chains at bands normally distort the vertical main chain helices from a straight course. Neutral water (not shown) penetrates bands and interbands, separating main chains to a n extent limited by hydrogen bonds between polar heads a t interbands, and simultaneously more room becomes available for the charged side chains a t bands, which now permit straightening of the main chains. Addition of acid discharges the negative side chains by means of hydrogen ions, and the equal number of free negative ions required t o remain a t the bands produce local osmotic swellings, which contract the structure axially.
+
While there seems some agreement that salt links are, under aqueous conditions, of relatively minor importance in integrating fibrillar structure, and that hydrogen bonding is effective in determining thermal contraction and dispersion phenomena, there seems to be some disagreement as t o the extent of rupture of the latter a t various stages in fibrillar
150
RICHARD S . BEAR
disintegration. This is undoubtedly a difficult question, and perhaps the most helpful suggestion a t the moment is derived by noting that there are several kinds of hydrogen-bond-forming side chain available for interprotofibril linking: (a) the acidic, amide, and basic groups, presumed herein, however, to be found relatively rarely a t the stable interbands; (b) the primary alcoholic hydroxyls of serine; and (c) the secondary alcoholic groups of hydroxyproline and threonine. Acidic and primary alcoholic groups are particularly potent hydrogen bond formers (168). Asaf et al. (4) have contrasted the relative abilities of primary and secondary alcoholic groups t o stabilize the polymeric carbohydrates, cellulose and starch. The former owes its greater resistance t o swelling agents to the fact that its crystalline structure is integrated largely by primary alcoholic groups, whereas the latter employs secondary groups principally and is correspondingly easily gelatinized. Indeed, starch granules undergo a transformation much like that of collagen thermal contraction and under similar conditions (25). The few primary alcoholic groups which prevent complete dispersion require lyophylic swelling or some 30°C. additional temperature (autoclaving) t o cause the complete release required for dispersion. Similar relationships could be envisaged for collagen, with the additional complication that occasional accidental interprotofibrillar unions of some strength (peptide, ether or ester cross linkage), about which little is known, may complicate the picture. One has also to reckon with the complex structures of connective tissues, whose reticular fibers often confine the more massive fibers (143, 146). VI. CONCLUSION 1. General Remarks
I n many respects the present models for the structure of the collagen fibril and its components -show the members of the collagen class to be structurally unique. Other native fibrous proteins, such as those of the k-m-e-f group, do not as clearly possess the same long thin types of molecular unit. Instead, as the case of feather keratin seems t o show particularly well (24), the latter fibers have more than a single large structural periodicity, two, in fact, which can be accounted for by assuming more or less globular molecules arranged as two-dimensional nets in ribbon-like fibrils (see also 97). Thus, the keratin-myosin fibrils are probably intermediate in organization between the collagens and the crystallizable globular proteins. The simple equivalence of the array of chemical properties along the
THE STRUCTURE O F COLLAGEN FIBRILS
151
collagen fibrils, the constituent protofibrils, arid the molecules, causes even the relatively large fibrils t o exhibit unusually directly the consequences of the sequential arrangement of residues along the individual polypeptide chains. This simplifies considerably the task of relating the finest chemical details t o the grosser characteristics of fibrils, fibers or even fiber bundles at the histological level. The type of organization adopted in collagen fibrils is particularly interesting from the biological point of view. The tendency for the individual polypeptide chains (protofibrils or molecules) t o arrange themselves with corresponding chemical features matching transverse t o their long axes calls to mind such problems as chromosomal pairing and duplication. One of the difficult features t o comprehend regarding such phenomena is the process by which like can attract like, rather than its complement. Consequently, one is inclined t o think of intermediate “templates” or “lock-and-key ” devices, but in the collagen example this does not seem required. To be sure, the details of the analysis do not as yet exclude the possibility th at different polypeptide or other chains of complementary local configuration may be present. This feature is not obviously necessary, however, since a single coiled chain, with many side chains emerging therefrom, as in the present model, could have one side complementary to the other, so that adjacent molecules can adjust themselves rotatiorially about their long axes to satisfy their needs in this respect. Collagen fibrils seem ideally constructed t o provide simple experimental models which may be used, by observation of swelling, contraction, and extension as furictions of chemical or physical environments, t o simulate the possible roles of protein fibrous systems in tissues or t o learn how the environmental factors influence protein chain configuration in general. Pryor (181) has already offered suggestions of this sort in connection with problems of muscle contraction. One of the important requisites for such study, however, is a n ability to follow the actual behavior of individual fibrils and protofibrils. Physical met,hods, such as electron optics and X-ray diffraction, are particularly useful in this regard, as has been amply demonstrated in preceding sections. It is highly desirable that these methods, which reduce the size range over which extrapolation must be made to link the macroscopic phenomena with the molecular, be more widely used in connection with observations of the colloidal properties of collagen, such as swelling. While collagen offers favorable material for this extrapolation, experience of the past has shown th at it must still be made cautiously in order t o avoid mistaken conceptions. The present development has attempted to relate a wide variety of
152
RICHARD 8. BEAR
collagen properties to fibrillar structure, using as few elements as possible: protofibrils, coiled molecules, bands and interbands, main chains and side chains. The roles of each of these have been considered t o the extent that seems desirable a t this time. Collagen is, in fact, still more complex than the model. For example, all bands are probably only approximately similar, and instead of the use of only two terms, band and interband, t o describe all fibrillar levels, it may eventually be necessary t o distinguish several types of level with varying degrees of distortion. Details of main-chain coil and of side-chain distribution, beyond those suggested, are not discernible as yet. The important question as t o the manner of link connecting molecules a t their ends into protofibrils has not been considered. Nevertheless, it is hoped that the model proposed will serve as a framework upon which future modifications and amplifications can be based, offering clarification of some points which have been obscure. 2. Summary
A review of available information, concerning collagen and featuring in particular X-ray diffraction, electron optics and certain phases of chemical evidence, has been made. A point of view is developed, covering the characterization of various members of the collagen class of fibrous protein and a derivation of the principal structural features of the unitary element of connective tissues, the collagen fibril. 1. Collagens are recognized most unequivocally by certain criteria derived from X-ray diffraction and electron optics, indicating a degree of constancy in the fundamental structure of all members of the class. Typical mesenchymal or mesogleal collagens are found in many animal phyla (Table I) arid other members are derived from secretions of epithelial tissues (Table 11). The latter thus far lack indications of the larger structures described below. I n general, one must go far across or between phyla to discover differences in the larger structural features, though certain finer details are common t o all. The differenccs between collagens resemble those of isomorphous simpler substances ; in the various collagens there are differences in the content and distribution of amino-acid residues within similar sized frameworks. More chemical evidence regarding the collagens of lower vertebrate and invertebrate animals is needed to make the meaning of these differences clearer. 2. Judging from evidence drawn largely from vertebrate animal sources, collagen fibrils are composed of thinner fundamental units, protofibrils. The latter are parallel and axially arrailged so that corresponding particulars of chemical structure more or less match transverse to the fibril axis. Each protofibril carries a periodic pattern of chemical content along it (period ca. 640 A.). The matching varies in perfection
THE STRUCTURE OF COLLAGEN FIBRILS
153
a t different fibrillar levels, giving rise to bands of distorted structure alternating with more perfectly constituted interbands, the two types of level providing a periodic fibrillar pattern of the same size as that of individual protofibrils. 3. The protofibril is of indefinite large length and packs into the fibrillar cross sections so that there is a 12 to 17 A. separation between protofibrillar centers a t interbands, depending on the degree of hydration. At interbands there is also a pseudoperiodic axial structure of about 20 A. in length. Band structure probably also approximates the same dimensions. Models which place 21 amino-acid residues per 20 A. of axial length, or about 0.95 A. per residue, account for satisfactory fibrillar density and extensibility. The protofibril is pictured as normally a single, regularly coiled polypeptide chain, with about 4.20 t o 5.25 residues per turn, stiffened by inter-turn hydrogen bonds arising from the imino and carbonyl group of the polypeptide main chain. Protofibrils are also capable of considerable individual contraction when freed thermally from linkage with neighbors and internally activated. 4. Collagen molecules are isolated when certain weak links occurring periodically along protofibrils are broken. The molecule is normally about 640 A. long and contains 670 t o 700 residues in coiled linear sequence. The molecular weight is calculated t o be 63,000 t o 65,000. The collagen molecule, like the protofibril, can become extended or randomly contracted. Gelatin molecules are aggregates or fragments of collagen molecules which'have been thermally altered and randomized t o some degree. 5. Since the collagen molecules are matched a t band and interband levels of the fibril, the pattern of these features as seen in fibrils is a more or less direct indication of the gross features of the sequence of residues along individual molecules. Bands, a t which tanning reagents largely enter, probably represent levels rich in the larger acidic, amide and basic side chains bristling out from main-chain coils. These bulky side-chains give rise to the protofibrillar distortions at bands. Interbands contain the, average sized, hydroxyl-containing side-chains, such as possibly hydroxyproline and serine. The prevalent (64 %) nonpolar side chains (largely glycine, proline and alanine) are probably distributed through both types of level. Corresponding t o the chemical contents, intrafibrillar stabilization a t bands is accomplished largely by salt-type linkages, and by hydrogen bonding a t interbands. Similarly, Donnan or osmotic swelling in acid or alkaline solutions is largely a t bands, causing a fibrillar shortening. In neutral environments the protofibrillar separation is increased a t all levels, removing the necessity for main chain crumpling at bands and causing a slight lengthening of the fibril. Figures 1, 31, and 32 illustrate the models involved.
154
RICHARD S. B E A R
ACKNOWLEDGMENTS The author is pleased to acknowledge an indebtedness to his colleagues, Drs. J. Gross, C . E. Hall, M. A. Jakus and F. 0. Schmitt, who contributed the electron micrographs reproduced in Figs. 21 through 28. Many helpful discussions with them have contributed materially to the contents of this review, although they are in no way accountable for any errors or excesses which may he found herein. Much of the matter dealing with the small-angle X-ray diffraction was accomplished in collaboration with Dr. 0. E. A. Bolduan and Mrs. Mildred Marks Siegel, with biological and chemical cooperation from Drs. C. H. Blake and T. P. Salo, respectively. Dr. E. FaurB-Fremiet of Paris contributed the shark elastoidin and skate ovokeratin samples, concerning which the results of small-angle diffraction study appear here for the first time.
REFERENCES 1. Ackerman, D., Muller, E., and Burchard, C. (1941). 2. physiol. Chem. 269, 146; 271, 183. 2. Amhrose, E. J., and Elliott, A. (1951). Proc. Roy. Soe. London A206, 206. 3. Arnett, L. M., Meibohm, E. P. H., and Smith, A. F. (1950). J . Polymer Sci. 6, 737. 4. Asaf, A. G., Haas, R. H., and Purves, C. B. (1944). J. Am. Chem. Soc. 66, 66. 5. Asthury, W. T. (1934). Kolloid-Z. 69, 340. 6. Asthury, W. T. (1936). Chem. Weekblad. 33, 778. 7. Asthury, W. T. (1938). Compt. Tend. trav. lab. Carlsberg. Sdr. chim. 22, 45; Trans. Faraday Soe. 34, 378; Kolloid-Z. 83, 130. 8. Asthury, W. T. (1939). Ann. Rev. Biochem. 8, 113. 9. Asthury, W. T. (1940). J . Intern. SOC.Leather Trades Chemists 24, 69. 10. Asthury, W. T. (1943). Advances in Enzymol. 3,'63. 11. Asthury, W. T. (1944). Proc. Roy. Znst. G. Brit. 33, 140. 12. Asthury, W. T. (1947). Proc. Sizth Intern. Congress Ezptl. Cytology. Exptl. Cell Research, supplement 1, 234. 13. Asthury, W. T., and Atkin, W. R. (1933). Nature 132, 348. 14. Asthury, W. T., and Bell, F. 0. (1939). Tabulae Biologicae 17, 90. 15. Asthury, W. T., and Bell, F. 0. (1940). Nature 146, 421. 16. Asthury, W. T., and Lomax, R. (1935). J. Chem. Sac., 846. 17. Baudouy, C. T. (1938). Compt. rend. 207, 1008. 18. Bear, R. S. (1942). J . Am. Chem. SOC.64,727; (1944) 66, 1297. 19. Bear, R. S. (1943). J. Am. Chem. Sac. 66, 1784; (1944) 66, 2043. 20. Bear, R. S. (1951). J. Am. Leather Chemists' Assoc. 46, 438. 21. Bear, R. S., and Bolduan, 0. E. A. (1950). Acta Crystallographica 3,230,236. 22. Bear, R. S., and Bolduan, 0. E. A. (1951). J . Applied Phys. 22, 191. 23. Bear, R. S., Bolduan, 0. E. A., and Salo, T. P. (1951). J. Am. Leather Chemists Assoc. 46, 107. 24. Bear, R. S., and Rugo, H. J. (1951). Ann. N . Y . Acad. Sci. 63, 627. 25. Bear, R. S., and Samsa, E. G. (1943). I n d . Eng. Chem. 36, 721. 26. Bergmann, M., and Niemann, C. (1937). J . Rial. Chem. 118, 301. 27. Bergmann, M., and Stein, W. H. (1939). J. Biol. Chem. 128, 217. 28. Beveridge, J. M. R., and Lucas, C. C. (1944). J . Biol. Chem. 166, 547. 29. Bidwell, E., van Heyningen, W. E., and Charlwood, P. A. (1948). Biochem. J. 42. 140.
THE STRUCTURE O F COLLAGEN FIBRILS
155
30. Block, R. J., and Bolling, D. (1939). J . Biol. Chem. 127, 685. 31. Block, R. J., and Bolling, D. (1951). The Amino Acid Composition of Protein and Foods. Charles C Thomas, Springfield, Illinois. 32. Block, R. J., Horwitt, M. K., and Bolling, D. (1949). J. Dental Research 28, 518. 33. Block, R. J., and Vickery, H. B. (1931). J . Biol. Chem. 93, 113. 34. Bolduan, 0. E. A., and Bear, R. S. (1949). J . Applied Phys. 20, 983. 35. Bolduan, 0. E. A., and Bear, R. S. (1950). J. Polymer Sci. 6, 159; (1951) 6, 271. 36. Bolduan, 0. ELA., Salo, T. P., and Bear, R. S. (1951). J. Am. Leather Chemists Assoc. 46, 124. 37. Borasky, R. (1950). Guide to the Literature on Collagen. Bureau of Agricultural and Industrial Chem., Agricultural Research Administration, U. S. D. A. Eastern Regional Research Lab., Philadelphia. 38. Bowes, J. H., and Kenten, R. H. (1947). Summary of Published Results (1900-1946) on Animo Acid Composition of Gelatin and Collagen. Brit. Leather Manufacturers' Research Assoc., London. 39. Bowes, J. H., and Kenten, R. H. (1947). Nature 160, 827. 40. Bowes, J. H., and Kenten, R. H. (1948). Biochem. J. 43, 358. 41. Bowes, J. H., and Kenten, R. H. (1948). Biochem. J. 43,365; (1949) 44, 142. 42. Bowes, J. H., and Kenten, R. H. (1949). Biochem. J. 46, 281. 43. Bragg, W. L., Kendrew, J. C., and Perutz, M. F. (1950). Proc. Roy. Soe. London A203,321. 44. Bresler, S. E., Finogenov, P. A., and Frenkel, S. Y. (1950). Doklady Akad. N a u k S. S. S. R. 73, 555. 45. Brown, C. H. (1947). Proc. Sixth Intern. Congress Exptl. Cytology. Ezptl. Cell Research, Supplement 1, 351. 46. Centola, G. (1936). Gazz. chim. ital. 66, 71. 47. Champetier, G., and FaurB-Fremiet, E. (1937). J . chim. phys. 34, 197. 48. Champetier, G., and Faur6-Fremiet, E. (1938). Compt. rend. 207, 1133. 49. Champetier, G., and FaurB-Fremiet, E. (1938). J. chim. phys. 36, 223. 50. Champetier, G., and FaurB-Fremiet, E. (1942). Compt. rend. 216, 94. 51. Cherbuliez, E., Jeannerat, J., and Meyer, K. H. (1938). Z. physiol. Chem. 264, 241. 52. Chernikov, M. P. (1949). Doklady Akad. Nauk 67, 345. 53. Clancey, V. J. (1926). J. Biol. Chem. 20, 1186. 54. Clark, G. L., Parker, E. A., Schaad, J. A., and Warren, W. J. (1935). J. Am. Chem. SOC.67, 1509. 55. Clark, G. L., and Schaad, J. A. (1936). Radiology 27, 339. 56. Cohn, E. J., and Edsall, J. T. (1943). Proteins, Amino Acids and Peptides. Reinhold, New York. 57. Corey, R. B., and Donohue, J. (1950). J . Am. Chem. SOC.72,2899. 58. Corey, R. B., and Wyckoff, R. W. G. (1936). J. Biol. Chem. 114, 407. 59. Draper, M. H., and Hodge, A. J. (1949). Nature 163, 576; Australian J . Exptl. Biol. Med. Sci. 27, 465. 60. Duran-Reynals, F. (1950). Ann. N . Y . Acad. Sci. 62, 943-1196. 61. Ebner, V. von. (1894). Sitzber. Akad. Wiss. W i e n . Math-naturw. Klasse, 103, 162. 62. Elod, E., and Zahn, H. (1949). Kolloid-2. 113, 10. 63. Engeland, R., and Bastian, A. (1938). Compt. rend. 207, 945.
156 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83.
84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102.
RICHARD S. BEAR
Engelmann, T. W. (1873). Arch. ges. Physiol. (Pflziger’s)7, 155; (1874) 8, 95. Ettisch, G., and Svegvari, A. (1927). Protoplasma 1, 214. Ewald, A. (1919). Z. physiol. Chem. 106, 135. Eyring, H., and Stearn, A. E. (1939). Chem. Revs. 24, 253. Farrant, J. L., Rees, A. L. G., and Mercer, E. H. (1947). Nature 169, 535. FaurC-Fremiet, E. (1936). J . chim. phys. 33, 681; (1937) 34, 125. FaurC-Fremiet, E. (1938). Arch. anat. microscop. 34, 23. FnurC-Fremiet, E., and Baudouy, C. (1937). Bull. SOC. chim. biol. 19, 1134. Faur6-Fremiet, E., and Baudouy, C. (1938). Bull. SOC. chim. biol. 20, 14; (1939) Arch. anat. microscop. 36, 77. Faur6-Fremiet, E., and Garrault, H. (1938). Bull. SOC. chim. biol. 20, 24. Faur6-Fremiet, E., and Garrault, H. (1944). Bull. biol. France Belg. 78, 206. FaurC-Fremiet, E., and Woelfflin, R. (1936). J. chim. phys. 33, 666, 695, 801. Feitelberg, S., and Kaunitz, P. E. (1949). Biochim. et Biophys. Acta 3, 155. Fraenkel, G., and Rudall, K. M. (1947). Proc. Roy. SOC.London B134, 1 1 I . Franchi, C. M., and de Robertis, E. (1951). Proc. SOC.Exptl. Biol. Med. 76, 515. Frey-Wyssling, A. (1948). Submicroscopic Morphology of Protoplasm and Its Derivatives. Elsevier, New York. Garrault, H. (1936). Ann. physiol. physzco-chim. b i d . 12, 291. Gerngross, O., and Katz, J. R. (1926). Kolloidchem. Beihefte 23, 368. Gerngross, O., and Katz, J. R. (1926). Kolloid-Z. 39, 181. Gerngross, O., Herrmann, K., and Abitz, W. (1930). Biochem. Z. 228, 409; (1931) Collegium 53. Gonnell, H. W., and Kratky, 0. (1931). Handbuch der physikaliucheti rind technisrhen Merhanik ,Band IV, 2 Hiilfte, 317. J. A. Barth, Leipzig. Graham, C. E., Waitkoff, H. K., and Hier, S. W. (1949). J. Bid. Chern. 177, 529. Grassman, W., and Riederle, K. (1936). Biochem. 2. 284, 177. Gross, J. (1948). J. Bid. Chem. 172, 511; (1950) Ann. N. Y. Acad. Sci. 62, 964. Gross, J. (1949). J. Exptl. Med. 89, 699. Gross, J. (1950). J . Gerontol. 6, 343. Gross, J., and Schmitt, F. 0. (1948). J . E’rptl. Med. 88, 555. Gustavson, K. H. (1942). Biochem. Z. 311, 347. Gustavson, K. H. (1942). Svensk Kem. Tid. 64, 74. Gustavson, K. H. (1949). Advances in Protein Chem. 6, 354. Guyon, L. (1934). Compt. rend. 198, 975. Hall, C. E. (1947). Am. Annual of Photography 61, 37. Hall, C. E., Jakus, M. A., and Schmitt, F. 0. (1942). J . Am. Chem. SOC.04, 1234. Hall, C. E., Jakus, M. A., and Schmitt, F. 0. (1945). J. Applied Phys. 16, 459; (1946) Biol. Bull. 90, 32. Halla, F., and Tandler, R. (1931). Z. physik. Chem. Bl2, 89. Heringa, G. C., and Kolkmeijer, N. H. (1926). Verhandel. Koninkl. Akad. Wetenschap. Amsterdam, Ajdeel. Natuurk. 29, 1092. Heringa, G. C., and Weidinger, A. (1940). Nederland Tijdschr Geneesk. 84, 5113; (1946) Acta Neerland. Morphol. 4, 291. 7, 417; (1874) 8, 275. Hermann, L. (1873). Arch. ges. Physiol. (Pfluger’s) Herrmann, K., Gerngross, O., and Abitz, W. (1930). 2. physik. Chem. B10, 371.
T H E STRUCTURE O F COLLAGEN FIBRILS
157
103. Herzog, R. 0. (1927). Kolloid-2. 41, 277. 104. Herzog, R. O., and Gonnell, H. W. (1924). Naturwissenschaften 12, 1153; (1925) Ber. B68,2228; (1926) Collegium 189. 105. Herzog, R. O., and Gonnell, H. W. (1931). Z. physik. Chem. B12, 228. 106. Herzog, R. O., and Jancke, W. (1926). Ber. B69, 2487. 107. Hess, K., and Kiessig, H. (1944). Z. physzk. Chem. 193, 196. 108. Hess, K., and Trogus, C. (1933). Biochem. 2. 262, 131. 109. Highberger, J. H. (1947). J . Am. Leather Chemists Assoc. 42, 493. 110. Highberger, J. H., Gross, J., and Schmitt, F. 0. (1950). J. Am. Chem. SOC. 72, 3321; (1951) Proc. Natl. Acad. Sci. U . S. 37, 286. 111. Highberger, J. H., and Kersten, H. J. (1938). J . Am. Leather Chemists Assoc. 33, 16; (1939) Nature 143, 1067. 112. Huggins, M. L. (1943). Chem. Revs. 32, 195. 113. Jennison, M. W. (1945). J. Bact. 60, 369. 114. Kaesberg, P., Ritland, H. N., and Beeman, W. W. (1948). Phys. Rev. 74, 71; Bull. Am. Phys. SOC.23, No. 4, 6. 115. Kanagy, J. R. (1950). J . Am. Leather Chemists Assoc. 46, 12. 116. Katz, J. R. (1932). Rec. trav. chim. 61, 835. 117. Katz, J. R. (1934). Die Rontgenspektrographie als Untersuchungsmethode. Abderhalden’s Handbuch der biologischen Arbeitsmethoden, Abt. 2, Teil 3, 3589-3600. Urban and Schwarzenberg, Berlin. 118. Katz, J. R., and Derksen, J. C. (1932). Rec. trav. chim. 61, 513. 119. Kate, J. R., Derksen, J. C., and Bon, W. F. (1931). Rec. trau. chim. 60, 725. 120. Katz, J. R., and Gerngross, 0. (1925). Naturwissenschaften 13, 900. 121. Katz, J. R., and Gerngross, 0. (1926). Kolloid-2. 40, 332; (1931) Collegium, 67. 122. Katz, J. R., Muschter, F. J. F. Jr., Weidinger, A., and Seiberlich, J. (1933). Biochem. 2. 267, 385, 397; 269, 76; 261, 15, 433; 262, 355; 263, 323; (1934) 271, 54; (1938) 297, 412; 298,320, 323. 123. Katz, J. R., and Weidinger, A. (1933). Collegium 85, 290; Biochem. 2. 269, 191, 223; 263, 421. 124. Klemperer, P. (1950). Am. J. Path. 26, 505. 125. Koisi, S. (1941). Mitt.med. Akad. Kioto 31, 809. 126. Kolpak, H. (1935). Kolloid-2. 73, 129. 127. Kratky, 0. (1947). Monatsh. 77, 224; (1948) J . Polymer Sci. 3, 195. 128. Kratky, O., and Sekora, A. (1943). J . makromol. Chem. 1, 113. 129. Kuhn, W. (1932). 2.physik. Chem. A161, 427. 130. Kuntzel, A. (1925). Collegium 623; (1926) 176; (1929) 207. 131. Kuntzel, A. (1934). Collegium 1; (1941) Kolloid-2. 96, 273. 132. Kuntzel, A. (1936). collegium 625. 133. Kuntzel, A., and Prakke, F. (1933). Biochem. 2. 267, 243. 134. Lelli, G., and Marotta, U. (1950). Rend. ist. super. sanitb 13, 503, 513, 518. 135. Lennox, F. G. (1949). Biochim. et Biophys. Acta 3, 170. 136. Leplat, G . (1933). Compt. rend. soc. biol. 112, 1256. 137. Lillie, R. D. (1947). J . Lab. Clin. Med. 32, 910. 138. Lloyd, D. J. (1932). J . SOC.Chem. Ind. 61, 141T. 139. Lloyd, D. J. (1935). J . Intern. Soc. Leather Trades’ Chemists 19, 336; (1938) 22, 558. 140. Lloyd, D. J. (1941). J . SOC.Dyers Colourists 67, 281. 141. Lloyd, D. J., and Garrod, M. (1946). Soc. Dyers Colourists, Symposium on Fibrous Proteins, 24.
158
RICHARD S. B E A R
Lloyd, D. J., and Garrod, M. (1948). Trans. Faraday SOC.44, 441. Lloyd, D. J., and Marriott, R. H. (1935). Proc. Roy. SOC.London B118, 439. Lloyd, D. J., and Marriott, R. H. (1936). Trans. Faraday SOC.32,932. Lloyd, D. J., Marriott, R. H., and Pleass, W. B. (1933). Trans. Faraday SOC. 29, 554. 146. Lloyd, D. J., and Stockall, G. (1937). Stiasny Festschrift, 232. Roether, Darmstadt. 147. MacArthur, I. (1943). Nature 162,38. 148. Marks, M. H., Bear, R. S., and Blake, C. H. (1949). J . Ezptl. Zool. 111, 55. 149. Maximow, A. A., and Bloom, W. (1943). Textbook of Histology. Saunders, Philadelphia. 150. Meyer, Karl. (1945). Advances in Protein Chem. 2, 249; (1947) Physiol. Revs. 27, 335; (1950) Ann. N . Y. Acad. Sci. 62,961. 151. Meyer, Karl, and Rapport, M. M. (1951). Science 113, 596. 152. Meyer, K. H. (1929). Biochem. 2. 214, 265. 153. Meyer, K. H. (1942). Natural and Synthetic High Polymers. Interscience, New York, p. 456. 154. Meyer, K. H. (1949). Nature 164,33. (Reply by Astbury follows.) 155. Meyer, K. H., and Ferri, C. (1936). Arch. ges. Physiol. (Pfldger’s)238, 78. 156. Meyer, K. H., and Haselbach, C. (1949). Nature 164,33. (Reply by Woods follows.) 157. Meyer, K. H., and Mark, H. (1928). Ber. 61, 1932. 158. Meyer, K. H., Odier, M. E., and Siegrist, A. E. (1948). Helv. Chim. Acta 31, 1400. 159. Mirsky, A. E., and Pauling, L. (1936). Proc. Natl. Acad. Sci. U . S . 22, 439. 160. Mustacchi, P. 0. (1951). Science 113, 405. 161. Nageotte, J. (1927). Compt. rend. 184, 115; Compt. rend. SOC.bid. 96, 172, 464, 828, 1268; 97, 559; (1928) 98, 15; (1930) 104, 156; (1933) 113, 841, 1398, 1401; (1931) Ann. anat. path. 8, 1. 162. Nageotte, J., and Guyon, L. (1930). Am. J . Path. 6, 631; (1931) Arch. biol. (Liege) 41, 1; (1933) Compt. rend, soc. biol. 113, 1398; (1934) Compt. rend. assoc. anat. 408. 163. Neuman, R. E. (1949). Arch. Biochem. 24, 289. 164. Nutting, G. C., and Borasky, R. (1948). J. Am. Leather Chemists Assoc. 43, 96. 165. Orekhovich, V. N., Tustanovskii, A. A., Orekhovich, K. D., and Plotnikova, N. E. (1948). Biokhimiya 13, 55; Doklady Akad. Nauk S. S.S.R . 60, 837. 166. Partridge, S. M. (1948). Biochem. J . 43,387. 167. Pauling, L. (1940). J . Am. Chem. SOC.62, 2643. 168. Pauling, L. (1945). The Nature of the Chemical Bond. Cornell Univ. Press, Ithaca, p. 333. 169. Pauling, L., and Corey, R. B. (1950). J . Am. Chem. SOC.72, 5349: (1951) Proc. Natl. Acad. Sci. U.S. 37, 235. 170. Pauling, L., and Corey, R. B. (1951). Proc. Natl. Acad. Sci. U.S. 37, 272. 171. Pauling, L., Corey, R. B., and Branson, H. R. (1951). Proc. Natl. Acad. Sci. U.S.37, 205. 172. Pfeiffer, H. H. (1943). Arch. e x p t l . Zelljorsch. Gewebezucht. 26, 92. 173. Picken, L. E. R. (1937). J . chim. phys. 34,764; (1940) B i d . Revs. Cambridge Phil. SOC.16, 133. 174. Picken, L. E. R., Pryor, M. G. M., and Swann, N. M. (1947). Nature 169, 434.
142. 143. 144. 145.
THE STRUCTURE O F COLLAGEN FIBRILS
159
Pirie, A., Schmidt, G., and Waters, 3. W. (1948). Brit. J. Ophthalmol. 321. Plotnikova, N. E. (1949). Doklady Akad. Nauk 66, 1145. Porter, K. R., and Vanamee, P. (1949). Proc. Soc. Exptl. Biol. Med. 71, 513. Pouradier, J., and Abribat, M. (1949). Science et inds. phot. 20, 444. Pouradier, J., Roman, J., and Venet, A. M. (1949). Compt. rend. 229, 1325; (1950) J. chim. phys. 47, 11, 391, 887. 180. Proctor, H. R., and Wilson, J. A. (1916). J. Chem. SOC.109, 307. 181. Pryor, M. G. M. (1950). Proc. Roy. SOC.London B137, 71. 182. Reed, R., and Rudall, K. M. (1948). Biochem. et Biophys. Acta 2, 7. 183. Roche, J., and Lafon, M. (1948). Compt. rend. SOC. biol. 142, 1200; (1949) Bull. SOC. chim. biol. 31, 147. 184. Salo, T. P. (1949). J. Am. Chem. SOC.71,2276. 185. Scatchard, G., Oncley, J. L., Williams, J. W., and Brown, A. (1944). J . Am. Chem. SOC.66, 1980. 186. Schemer, P. (1920). Zsigmondy’s Lehrbuch der Kolloidchemie. Auflage 111, p. 408. 0. Spamer, Leipzig. 187. Schmidt, W. J. (1924). Die Bausteine des Tierkorpers in polarisiertem Lichte. Friedrich Cohen, Bonn, p. 279. 188. Schmidt, W. J. (1934). Polarisationsoptische Analyse des submikroskopischen Baues von Zellen und Geweben. Abderhalden’s Handbuch der biologischen Arbeitsmethoden, Abt. 5, Teil 10, 435. Urban and Schwarzenberg, Berlin. 189. Schmitt, F. O., Bear, R. S., Hall, C. E., and Jakus, M. A. (1947). Ann. N. Y. Acad. Sci. 47, 799. 190. Schmitt, F. O., and Geren, B. B. (1950). J. Exptl. Med. 91, 499. 191. Schmitt, F. O., and Gross, J. (1948). J . Am. Leather Chemists Assoc. 43, 658. 192. Schmitt, F. O., Hall, C. E., and Jakus, M. A. (1942). J . Cellular Comp. Physiol. 20, 11. 193. Schmitt, F. O., Hall, C. E., and Jakus, M. A. (1945). J. Applied Phys. 16, 263. 194. Sizer, I. W. (1942). Anat. Record 84, 256; (1945) Ann. Surg. 121, 231. 195. Sizer, I. W. (1949). Enzymologia 13, 288. 196. Stein, W. H., and Miller, E. G. , Jr. (1938). J. Biol. Chem. 126, 599. 197. Taylor, H. S. (1941). Proc. Am. Phil. SOC.86, 1. 198. Thaureaux, J. (1945). Bull. SOC. chim. biol. 27, 327. 199. Theis, E. R. (1946). Trans. Faraday Soc. 42B,244. 200. Thomas, A. W., and Seymour-Jones, F. L. (1942). J . Am. Chern. SOC.46,1515. 201. Trillat, J.-J. (1930). Compt. rend. 190, 265. 202. Weir, C. E. (1948). J . Research Null. Bur. Standards 41,279; (1949) J. Am. Leather Chemists Assoc. 44, 79; (1950) J . Research Nail. Bur. Standards 46, 468. 203. Weir, C. E. (1949). J. Research Natl. Bur. Standards 42, 17; J. Am. Leather Chemists Assoc. 44, 108. 204. Wislocki, G. B., Bunting, H., and Dempsey, E. W. (1947). Am. J. Anat. 81,l. 205. Wohlisch, E. (1928). 2. Biol. 87, 353; (1932) Biochem. 2. 247, 329. 206. Wijhlisch, E., and de Rochemont, R. D. (1927). Beitr. path. Anat. 78, 233; 2. Biol. 86, 406. 207. Wolpers, C. (1943). Klin. Wochschr. 22, 624. 208. Wolpers, C. (1944). Klin. Wochschr. 23, 169. 209. Wolpers, C. (1944). Arch. path. Anat. Physiot. (Virchow’s) 312, 292; Deut. med. Wochschr. 70, 435. 210. Wolpers, C. (1948). Makromol. Chem. 2, 37. 211. Wolpers, C. (1948). Biochem. 2. 318, 373. 175. 176. 177. 178. 179.
160
RICHARD 6. R E A R
212. Woods, H. J. (1946). J . Colloid Sci. 1, 407. 213. Wright, B. A. (1948). Nature 162, 23. 214. Wright, B. A , , and Wiederhorn, N. M. (1949). Abstracts of Sept. Meeting, Am. Chem. Soc., 17C. 215. Wyckoff, R. W. G. (1949). Electron Microscopy. Interscience, New York, p. 199. 216. Wyckoff, R . W. Q., and Corey, R. B. (1936). Proc. SOC.Exptl. Biol. Med. 34, 285. 217. Wyckoff, R. W. G., Corey, R. B., and Biscoe, J. (1935). Science 82, 175. 218. Zahn, H. (1948). Kollozd-Z. 111, 96. 219. Za'Ides, A. L., and Pupko, 8. L. (1950). Kolloid. Zhur. 12, 275.
Muscle Contraction and Fibrous Muscle Proteins* BY HANS H. WEBER
AND
HILDEGARD PORTZEHL
Physiological Institute] University of Tsiibingen,Germany
CONTENTS Page I. Introduction. . . . . . . . . . . . . . . . . . . . . , . . . . . . . . . . . . . 162 11. The Contractile Models.. . . . . . . . . . . . . . . . . . . . . . . . . 163 1. The Various Kinds of Model.. . . . . . . . . 2. Properties of the Models not Dependin 3. Properties of the Models with Respect to Contraction.. . . . . . . . . . . . . . . . 166 a. Tension and Shortening.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 b. The Quick Release Phenomenon.. . . . . . . , . . . . . . . . . . . . . . . c. Reversibility and the Active S t a t e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 d. Quick Stretch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 e. Changes in Birefringence and the Degree of Shortening in Fibrils and Micelles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 f. Temperature Dependence of the Contraction.. . . . . . . . . . . . . . . . . . . , . 173 g. Comparison of Contraction States in the Models and in Muscle.. . . . . 174 4. Thermodynamics of Contraction in the Models.. . . . . . . . . . . . . . . . . . . . . . 178 a. Is Contraction a Steady State or a New Thermodynamic Equilibrium? 178 b. Energy Requirement and Availability in the Case of Equilibrium and of Steady State 179 c. Connection between ATP Concentration and Temperature Dependence of the Tension for the Case of Thermodynamic Equilibrium. . 181 d. Inhibition of Adenosine Triphosphatase Activity and Contraction. . . . 183 e . Optimum ATP Concentration for Splitting and Contraction. . . . . . . . . 185 f. ATP Optimum for Contraction and the Extent of Shortening.. . . , . . . 187 g. The Explanation of the Different Positions of the ATP Optima. . . . . . 187 h. Analysis of the Contraction of the Models and the Contraction Cycle of Muscle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187 .. . ... . . . . _ _ . . . _ .189 .
1 . Historical and Nomenclature. . . . . . . ............................ 2. Solubility, Colloidal Properties, and Crystallinity. . . . . . . . . . . . . . . . . . . . , 3. Electric Charge.. . . . . , , . . . , , . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Particle Size and Shape.. , . . . . . . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . a. L-Myosin . . . _ .. . . . . . . , . . . , , . , , . . . . . . . . . . . . . . . . . . , . . . . . . . . . . . b. Tropomyosin. , , . . , . . . . . ......,_......,..... ....... ,
,
193
194 200 203 203 210
* Translated from the German by F. R. Bettelheim and Kenneth Bailey, University of Cambridge, England. 161
162
HANS H. U-EBER AND HILDEGARD PORTZEHL
Page ................................................... d. Actomyosin . . . . . . . . . . . . . . . . . ........................... 5. Reactions of the Fibrous Proteins ........................... a. The Interaction of L-Myosin and Actin. . . . . . . . . . . . . . . . . . . . . . . . b. The Reaction of G-Actin Particles with One Another.. . . . . . . . . . . . . .
c. The Reaction of Actin Particles with A T P . . . . . . . . . . . . . . . . . . . . . . . . d. The Interaction of L-Myosin and Actomyosin with ATP.. . . . . . . 1. ATP-breakdown by Actomyosin and L-Myosin. . . . . . . . . . . . . . . . . 2. Influence of ATP on Actomyosin in Solution., . . . . . . . . . . . . . . . . . 3. Influence of ATP on Actomyosin Gel.. . . . . . . . . . . . . . . . . . . . . . . . . Appendix: Other Enzymatic Activities of Myosin. . . . . . . . . . . . . . . . . . . . . 6. Isolation of the Fibrous Proteins of Muscle.. . . . . . . . . . . . . . . . . . . . . . . . . a. Extractability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... ... b. Fractionation and Pur ......................... Appendix: “Activity ” (Straub, 1942) and “ ATP-Sensitivity ” (Portzehl, Schramm and Weber, ......................... ucture of Skeletal Muscle.. . IV. The Proteins of the Myofibr 1. The Proportion of F-Actin and L-Myosin in the Whole Muscle Protein. . 2. Actomyosin and the Fine Structure of Muscle.. . . . . . . . . . . . . . . . . . . . . . . a. The Fibrous Fine Structure of Resting Muscle. . . . . . . . . . . . . . . . . . . . b. The Fibrous Fine Structure of Contracted Muscle.. . . . . . . .
211 213 216 216 219 220 222 225 220 230 231 231 233 236 237 237 239 239
3. Changes in the Fibrous Muscle Proteins in Contracture and F V. General Conclusion, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . ..................
I. INTRODUCTION Muscle works by alternating between contraction and relaxation. Mechanical work is done during contraction, and the relaxation phase enables the work to be repeated. Contraction and relaxation involve alterations of colloidal state of opposite sign; they are linked by a chain of reactions which begins at excitation and lasts until the end of the recovery phase. The change in shape is brought about by particles of contractile protein, whose mechanical state begins to change about the middle of the latent period (A. V. Hill, 1949a; 1950a, b, c; D. K. Hill, 1949; Abbot,t and Ritchie, 1948; Sandow, 1944,1947 ;Schafer and Gopfert, 1941). This change in shape appears t o be over by the end of the relaxation phase. The electrical changes, however, last into the recovery period, since as lactic acid accumulates, the protein binds hydrogen ions exothermically, and gives them up, absorbing heat, as the acid disappears (Meyerhof, 1923, 1924, 1930). The relaxation and contraction phases will obviously be easier to study when it becomes possible t o isolate them experimentally. For the contractile phase, such isolation now appears possible by progressively breaking up the whole muscle into simpler components. We may think of this breaking u p as going through five stages: in the first stage we have
MUSCLE CONTRACTION
AND FIBROUS
MUSCLE PROTEINS
163
the whole structure of fibrous proteins intact; in the second, we have the purified actomyosin complex in its oriented state; in the third, unoriented actomyosin; in the fourth, actomyosin in solution; and finally, each of the components, actin and L-myosin,‘ in solution. The present review will deal with the properties of these five systems and their behavior on the addition of adenosine triphosphate (ATP), adenosine diphosphate (ADP), inosine triphosphate (ITP), inorganic triphosphate, and pyrophosphate. The really important question is how far the interaction phenomena with ATP can be regarded as a model for contraction in living muscle-a model for contraction as a whole in the case of the earlier of these stages, and a model of the molecular mechanism in the case of the later stages. For the relaxation phase, no model is yet known (cf. however, Section 11, 4h and appendix). 11. THE CONTRACTILE MODELS 1. The Various Kinds of Model
The author introduced in 1934 the oriented (acto-)myosin thread as a model with which to explain the mechanical properties, X-ray diffraction and birefringence of muscle (1934a, b). I n 1942-1943 Szent-Gyorgyi reported that he had observed undried and therefore unoriented actomyosin threads to shrink very greatly on addition of ATP. At the same time, Gerendhs (1942) showed th at actomyosin threads could be oriented by stretching without the necessity of drying if certain heavy metal ions were added. When copper ione were used, the capacity of the threads to react t o the addition of ATP was completely destroyed, and with zinc ions largely so. The residual activity gave rise, however, not to a n isodimensional shrinkage but to an actual shortening. Szent-Gyorgyi and his school ever since have called the shrinkage of these threads a “contraction.” I n 1947, Buchthal and coworkers succeeded in making actomyosin threads extensible by partial drying and to some degree they were thus able t o orient them (Buchthal, Deutsch, Knappeis, and Petersen). When unloaded, these threads showed a n actual shortening on immersion in a solution of ATP (2 X 10-3 M ) , but after loading an increase in length was observed, which under some conditions was quite considerable. Buchthal was therefore disinclined to see a correlation between “contraction” in actomyosin threads and the contraction of muscle, and several other workers have taken this attitude (Perry et al., 1948; Jacob, 1945; Sandow, 1949; Astbury, 1947). In this review, the contractile protein complex is called actomyosin, and its components actin and L-myosin. To avoid ambiguity, the word myosin is used only for preparations which may contain actomyosin and L-myosin, i.e., for all preparations made before the discovery of actin, and for the A and B myosins of Szent-Gyorgyi.
164
HANS H. WEBER A N D HILDEGARD PORTZEHL
I n the meanwhile Varga (1946) had found in Szent-Gyorgyi's laboratory that microtome sections of muscle, about 1OOp thick, after extraction with water and freezing, shortened to 30% of their original length when placed in a bath of ATP. It was not possible to measure the tension they developed because they tore too easily. Later, Szent-Gyorgyi (1949) substituted for the microtome sections bundles of fibers 0.2-0.5 mm. thick from the psoas muscle of the rabbit, and he found th a t these fiber bundles developed tensions of up t o 2 kg./cm.2 on addition of ATP. The same tension was developed without immersion in ATP when fresh muscle fiber bundles of any thickness, which had been very rapidly frozen, were allowed to thaw. Since this tension developed by the thicker bundle of fibers shows the same temperature dependence as th a t of the thin fiber bundle, Szent-Gyorgyi assumes th a t the process is due t o the ATP which has remained in the muscle itself (Szent-Gyorgyi, 1949; Borbiro and Szent-Gyorgyi, 1949). I n spite of some differences from the ATP-induced contraction of other models, for example, two contractions following one another and the disappearance of tension after a few minutes, this assumption may well be correct. If i t is, then a macroscopic model is now available for certain purposes. I n 1950, Weber and collaborators reported the preparation of actomyosin threads and of glycerol-extracted muscle preparations with which it could be shown that the fundamental process of contraction in the two models was the same. Moreover, the conditions for the contraction and its character could be so sharply defined that very exact comparisons with the contraction of living muscle could be made (A. and H. H. Weber, 1950, 1951; Portzehl and H. H. Weber, 1950; H. H. Weber, 1950c; A. Weber, 1951).
2. Properties of the Models not Depending o n Contraction I n order t o make actomyosin threads contract with the development of tension, the protein concentration must be regulated; if the concentration is too low, the cohesive forces are too small; if too high, the movement of the protein particles is too restricted. The"optima1 concentration of protein lies between 6 and lo%, as opposed to the low value of 1-2.5% when the thread is freshly extruded. If the thread is dried in air for a few minutes (cf. Buchthal et al., 1947) the increase in protein concentration is much greater near the periphery than in the center, which is also true for the orientation and birefringence on stretching. This inhomogeneity of drying can be avoided by first soaking the thread in a glycerol-water mixture and then drying a t 1°C. in an atmosphere almost saturated with water vapor. Thus, the drying process takes many hours, the center and periphery remaining in water equilibrium.
MUSCLE CONTRACTION
AND FIBROUS MUSCLE PROTEINS
165
The final protein concentration is determined by the proportion of glycerol in the glycerol-water mixture, and must not be exceeded anywhere in the fiber (Weber and Portzehl, 1950; Portzehl, 1951). If glycerol-extracted muscle fibers are to develop their maximal tension under defined conditions, the preparation must not be too thick, because the ATP is rapidly broken down (Korey, 1950; Biro and SzentGyorgyi, 1949; Heinz and Holton, 1950) and its concentration therefore decreases as it passes inside the fibers. Even if the diffusion of ATP were as rapid inside the preparation as in free solution, the concentration would be virtually zero in the center of a cylinder of radius 250 p. Only TABLEI Decrease of ATP Concentration from the Outside to the Center of the Fiber, Calculated according to Meyerhof (1930)* from A . Weber (1961)
ATP concentration in bath ( M x 104)
Temperature, "C.
Fall in ATP concentration between outside and center ( M x 106)
7 0
16
* C _ A . 4 0
1.3 1.7
1.7
4.0
0.7
1.4 15.5 13.9
0.9 1.6
10 60
20
100 200
Fall in ATP concentration outer concentration % 1.9
0.7
2
'
f
J
where C = concentration in the bath, D = diffusion constant (40 X 10-7 cm.' sec.-l), r = radius, approx. 2 5 p ; J > 0, A = rate of ATI' breakdown.
with single fibers is the diffusion path so short th a t the A T P concentration is practically constant throughout (see Table I, taken from A. Weber, 1951) and this would be true even though the diffusion of A T P were five t o ten times as slow as in free solution. Extraction of muscle fibers with water or glycerol-water mixtures removes 20% of the muscle protein, consisting for the most part of the sarcoplasmic fraction (Szent-Gyorgyi, 1949). The concentration of actomyosin is thereby increased from a value of about 50% in the case of living muscle (Hasselbach and Schneider, 1951) to one of about 65%. Of the salts originally present, only a few cations, mostly potassium, remain behind as protein cations, and these are mostly replaced b y sodium ions when the extracting solution is buffered with sodium salts (E. J. Harris, personal communication).
166
HANS H. WEBER A N D HILDEGARD PORTZEHL
Actomyosin threads made from actomyosin which has been repeatedly precipitated contain no other proteins, and the salt concentration depends only on the composition of the bath. Those contractile properties which are common both to the extracted muscle fiber (fiber model) and the actomyosin thread (thread model) can therefore be ascribed t o the actomyosin. Both fiber and thread models can contract in presence of sodium or potassium ions when these are the cations in the medium (Buchthal et al., 1947; A. Weber, 1951; Portzehl, 1950b). The fiber model contracts with ATP and ADP but not with inorganic pyrophosphate (A. Weber, 1951; Korey, 1950). The thread model, on the other hand, contracts only with ATP, and not with ADP, ITP, inorganic triphosphate or pyrophosphate (Portzehl, 1950b; Buchthal et al., 1947). The difference in behavior with ADP is presumably due t o the presence of myokinase in the fiber model which converts ADP t o AT P and adenylic acid.
3. Properties of the Models with Respect to Contraction a. Tension and Shortening. A model fiber, fastened in a suitable solution (0.01 M sodium phosphate; p H about 6.9; 0.001 M magnesium chloride) so t ha t it is extended without tension, develops a t 18°C. tensions of up t o 4 kg./cm.2 as soon as the ATP concentration of the bath is brought t o 5-8 X M (curves of Group 2, Fig. 1). This tension is of the order of the maximal tension reached by muscles of warm-blooded animals. The maximal tension which can be developed by a model fiber is limited b y its tensile strength (Fig. 1). The tension decreases proportionately with increasing shortening, and becomes zero a t 15-20 % of the starting length. This starting length is defined as the equilibrium length of the resting fiber after removal of the insertions, and is 80% of the resting length in situ (A. Weber, 1951). The actomyosin thread under the same conditions develops a tension of about 200 g./cm.2, which becomes zero a t 30-40% of the starting length. The maximal tension is limited here also by the tensile strength, which is thus some twenty times smaller than for the fiber (curves of group 2, Fig. 2). If the actomyosin composing the thread is not directly isolated from the muscle extract, but is reconstituted from pure mondisperse L-myosin (Portzehl, 1950a; Portzehl et al., 1950) and actin (Straub, 1942; Szent-Gyorgyi, 1947), mixed in the proportion of 3.8:1, the maximal tension is much smaller, though the maximal shortening is the same (Fig. 2, curve 3). All these values refer to tensions developed under conditions in which the ATP concentration is optimal. The threshold ATP concentration for the fiber model is about 2X M (A. Weber, 1951)-some three times lower than th a t for
MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS
167
RELATIVE LENGTH, 100 L/L,
FIG.1. ATP contraction of the model fiber (from A. Weber, 1951). Relative length L / L o X 100 where LOis the equilibrium length without ATP. Group 1: three length-tension diagrams of fibers shortening from the stretched state, without ATP. Group 2: two diagrams of fibers shortening in ATP-contraction. X, *, torn fibers.
+,
200
-
160
-
N
E
9
-
c5 120-
2-
0 In z
w I-
-
1
80-
40 0-
'
=
40
50
60
70
80
90
100
RELATIVE LENGTH, 100 L / L o
FIG.2. ATP contraction of actomyosin threads (from Portzehl, 1950b). Groups 1 and 2: natural actomyosin. Curve 3: actomyosin from L-myosin and actin (3.8:1). Group 1 in absence, groups 2 and curve 3 in presence of ATP.
168
H A N S H . IV’EBER A N D HILDEGARD PORTZEHL
-
the fiber bundles which Korey (1950) describes; and for actomyosin 5 X lov6 M (Portzehl, 1950b). threads it is of the same order, I n both models the decrease of tension with shortening is much smaller if the tension is produced in the presence of A T P than if i t is produced in absence of ATP by stretching the models beyond their resting length (cf. groups 1 and 2 in Figs. 1 and 2; also Szent-Gyorgyi, 1949). The tensions which have been discussed were always recorded as the model was allowed to shorten progressively. If after shortening the model is stretched again, the tension corresponding to a given length is higher, and the difference between the values obtained by the two procedures is greater the more rapidly the alterations in length are made (A. Weber, 1951; Portzehl, 1951). This effect, which is observed both in the presence and in the absence of ATP, can be obtained both with contracting muscle (Levin and Wyman, 1927) and with resting muscle (Josenhans, 1949). Levin and Wyman assumed that, if sufficient time were allowed, the tension a t any length would reach a unique equilibrium value intermediate between those obtained by stretching and allowing t o shorten. I n the case of a model fiber in ATP, this state is not attained even over a period of eleven hours (A. Weber, 1951). b. The Quick Release Phenomenon. If a frog muscle in isometric tetanus is suddenly released so that it can shorten a t least lo%, the tension falls t o zero and then rises again. The time taken to reach the equilibrium value corresponding to the new length is 0.15 second after a shortening t o 90% of the initial length, and 0.4 second after a shortening t o 60%.2 This is the “quick release” phenomenon of Gasser and Hill (1924). Both models when under the influence of ATP show this effect if they are suddenly released, and a shortening of only 5 % suffices to abolish the tension completely. Their development of tension, however, takes longer than for skeletal muscle, especially in the case of the actomyosin thread (Fig. 3). Thus, for the model fiber, the new tension after a release from 72 t o 67% is reached in about 60 seconds, and after a release from 55 to 50% in about 150 seconds; for the actomyosin thread the new tension after a release from 94 to 91 % is attained in approximately 240 seconds, and from 82 t o 79% in as much as 320 seconds. The quick release phenomenon is characteristic of the contracted state. The tension produced by stretching an uncontracted muscle or model falls after release only a little below the new equilibrium value, which is finally reached by a small elastic after-effect (see curves 1, 2, 3, and l a and 2a of Fig. 4).
* Gasser and Hill do not state the length of the muscle; it is assumed here to be 30 mm. (frog sartorius).
MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS
169
IW
*-.*-
0
80
160
240 0 TIME, seconda
80
160
240
320
FIG. 3. Quick-release-recovery phenomenon. Upper half: model fiber (from A. Weber, 1951). Lower half: actomyosin threads (from Portzehl, 1950b). Tension in millimeters deflection of light spot. Curve 1: release form 67 to 65 % of relative length; Curve 2, from 55 to 50 %; curve la, from 94 to 91 %; Curve 2a, from 82 to 79 %.
RELATIVE LENGTH, 100 L/La
FIG.4. Redevelopment of tension in presence of ATP, and elastic after-effect in the absence of ATP, following release. Curve 1: stretched fiber without ATP; Curve 2: unstretched fiber with ATP; Curve 3: fiber after washing out A T P (from A. Weber, 1951). Curve la: stretched actomyosin thread without ATP; curve 2a; the same, unstretched, with ATP (from Portsehl, 1950b).
170
HANS H. WEBER AND HILDEGARD PORTZEHL
c. Reversibility and the Active State. If the ATP is washed out of the models when contracted, 7 0 4 0 % of the tension and the whole of the shortening remain. The model, however, now behaves as if it were stretched beyond the resting length; the tension falls very steeply when further shortening is allowed, there is no quick release effect (Fig. 4, curve 3) and the temperature dependence of tension is very small (Fig. 5; see Section 11, 3f). These models evidently remain “frozen” in the
* I
I
1
I
I
20
10 TEMPERATURE,
OC.
FIG.5 . Temperature dependence of the tension of the fiber model before and after washing out ATP. Before: curves 1 and 2; after: curves l a and 2a; - - - - - fall of tension during the washing (from A. and H. H. Weber, 1951).
condition they happened t o be when the ATP was removed. I n this state they may provide a useful model for muscle in rigor. The model in presence of ATP is described as being in the “active” state. The model which keeps its isometric tension or its short length after the removal of ATP is referred to as being in the “contracted” but “inactive” state. The effects of contraction by ATP, i.e., the residual tension or the final short lengths, are considered irreversible, since they cannot be abolished by removing the ATP. But all the other characteristics of the active state, the redevelopment of tension after a release, the shortening over a range of 80-85 % of the initial length, the tempera-
MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS
171
ture dependence of tension, are bound t o the presence of ATP and can thus be made to appear or disappear reversibly. d . Quick Stretch. A quick stretch increases the tension of the inactive model more than that of the active model, showing that the inactive fiber is less extensible than the active one (Fig. 6). Further, after the stretch is stopped, the tension decreases much less in an inactive model than it does in the active one or in muscle. Thus the pronounced rigidity of the inactive fiber is shown also by its lack of elastic after-effects (see Fig. 6; Section 11, 39). 600 500 -
2
I 400
-
300 -
---
-----
tt
t
FIG. 6. Tension changes in model fiber on stretch and subsequent release. Curve 1: with ATP; curve 2, after washing out ATP a t time intervals worked b y arrows. Initial tension - - - - (from A. Weber, 1951).
If the model fiber after stretch is released t o its initial length, then in the presence of ATP the tension returns t o its original value; in absence of ATP (Fig. 6) however, the tension falls below the original value and remains there. Thus, for a fiber without ATP, a sudden stretch of 2-3 % suffices to exceed the elastic limit. e . Changes i n Birefringence and the Degree of Shortening in Fibrils and Micelles. Birefringence studies (H. H. Weber, 1934a, b) and X-ray investigations (Astbury and Dickinson, 1940) have shown that passive changes in length of (acto)myosin systems are due to passive changes in length of the individual protein micelles and the peptide chains which they contain. One may enquire whether the same applies when change of length are actively induced. In birefringence studies, it is necessary to know how much of the whole change is due t o intrinsic and how much to form birefringence. The birefringence of muscle decreases only very slightly in isometric contraction (von Muralt, 1932), but very considerably when shortening takes place (von Ebner, 1916), and its value is determined overwhelm-
172
HANS H. WEBER AND HILDEGARD PORTZEHL
ingly by the change in length and only insignificantly by the transition from the inactive to the active state. The same applies to the fiber model. I n isometric contraction, the birefringence remains substantially
2
1.4
1.5
1.6
1.7
1.3
1.4
1.5
1.6
1.7
REFRACTIVE INDEX OF MEDIUM
FIG. 7. Total and form birefringence of uncontracted (left) and contracted (right) model fiber. Unbroken curves represent total birefringence: left, group 1, maximal and minimal values from six fibers; on the right, group 2, average values for five fibers of relative length 2 8 4 3 % ; below, group 3, negatively birefringent disk produced by contraction. Broken curves are corrected for form birefringence; A 0 0 points corresponding to different fibers. Values are calculated to 15% protein concentration-for intrinsic birefringence according to Weber (193413) and for form birefringence according t o Strobel (1952).
the same, but falls by some 60% (from 2.7 to 1.1 X when a fiber with a protein content of 15 % shortens t o 25% of its initial length. For in vivo contraction i t decreases by only 45% (von Ebner, 1916) b u t the degree of shortening is here not more than two-thirds th a t of the model.
MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS
173
The total fall in birefringence is due in fairly equal proportions to intrinsic and form birefringence (Fig. 7), and this is to be expected if the protein micelles are displaced from their exact alignment along the fiber axis. Such displacement does apparently occur to some extent, for in strongly contracted model fibers the optic axis is no longer exactly parallel to the fiber axis. The side to which it deviates varies from one segment to another, and it is evident that the submicroscopic protein chains run a slightly winding course, remaining in spite of this remarkably parallel to each other. The decrease in birefringence is not due only to disorientation, however, but probably also to a shortening of the protein particles. The birefringence of the actomyosin thread tells us nothing important about the structural changes associated with contraction. The birefringence of the stretched thread is less than that of the model fiber (Buchthal et al., 1947; Portzehl, 1950b), due partly to the lower protein concentration and partly to incomplete orientation. At the same protein concentration and with complete orientation, the birefringence of the thread model ought to be about twice as great as that of the fiber model and of living muscle because the thread does not possess isotropic bands, and this high value can actually be attained with noncontractile threads (H. H. Weber, 193413). The birefringence increases both on stretching and in isometric contraction due t o loss of water and increase in orientation. On shortening it decreases as in the model fiber (Buchthal et al., 1947; Portzehl, 1950b). When the thread first shortens, and then by further contraction develops tension, the two effects are combined, the birefringence first falls and then increases again (Buchthal et al., 1947; Portzehl, 1950b). f. Temperature Dependence of the Contraction. The tension of a resting, slightly stretched frog muscle increases slightly with temperature (Fig. 8, curve 5; see also Wohlisch and Gruning, 1943; Knappeis, 1948). If the muscle is stimulated into a tetanus so as to minimize relaxation effects, the increase of tension with temperature is now five times greater (Fig. 8, curve 4). Similarly, the tension of the inactive stretched model fiber rises very little with temperature (curve 3, Fig. 8), but that developed both by the model fiber (Szent-Gyorgyi, 1949; A. and H. H. Weber, 1950; A. Weber, 1951) and by the actomyosin thread (Portzehl, 1951; Portzehl and H. H. Weber, 1950) in isometric contraction shows a marked temperature dependence (Fig. 8, curves 1, l a and 2). The temperature dependence becomes larger with increasing ATP concentration up to the concentration in which the model develops a maximal tension a t 10'. At this optimal concentration, the temperature dependence is considerably greater than in muscle itself (Fig. 8, curve 1). Just as the tension shows
174
HANS H. WEBER AND HILDEGARD PORTZEHL
such a marked dependence upon temperature, so too does the shortening (Varga, 1946). g. Comparison of Contraction States in the Models and in Muscle. Comparing first the inactive state of the models with th a t of living muscle, we find the following similarities; the actomyosin particles are extended and lie parallel to one another (Table 11, birefringence properties, column 8), the dependence of tension upon temperature is small (column 7) and all actomyosin systems are able t o pass from the uncontracted to the contracted state. -The elastic properties of the model and living
TEMPERATURE,
OC.
FIG.8. Temperature dependence of tension. Curve 1: model fiber (A. Weber, 1951); l a , actomyosin thread (Portzehl, 1950b), both in 3 X 1 0 - 3 M ATP; curve 2: model fiber in 4 X 10-4 M ATP; curve 3, model fiber without ATP (A. Weber, 1951); curve 4: muscle in tetanus (from A. V. Hill, 1951); curve 5: resting muscle (from Josenhans, 1949).
muscle, on the other hand, are very different. A sudden stretch of 2-3 %, stretches the models beyond the elastic limit (Fig. 6), and sometimes tears them (Szent-Gyorgyi, 1949). If the stretching is carried out very slowly, the model fiber can be stretched by 30-40% without any macroscopic sign of breaking, but even then the elastic limit is far exceeded and on release the amount of shortening is only about 10% (Fig. 9). Moreover, this kind of stretching makes the fiber inhomogeneous to polarized light, showing that the structure of the contractile portions has been damaged. I n the case of living muscle, the elastic limit and breaking point are not reached until the fibers are stretched b y 50-100% (for rabbit psoas see Szent-Gyijrgyi, 1949).
175
MUSCLE CONTRACTION AND FIBROUS MUSCLE PROTEINS
The rise of tension during stretch is 4-5 times greater for the fiber model and actomyosin thread than for resting muscle (cf. curves 1, 2, 3, with 4 and 5 of Fig. 9). This is true whether the models are in the TABLEI1 Comparison of Several Properties of the Thread and Fiber Models with Those of Living Muscle, in the Active and Inactive State Inactive state
Active state Property
Maximal tension Maximal shortening
Slope of tension with shortening Quick release
Fiber model
Myosin thread
2-5 kg.'am 4 kg.b 80-85 %* 50-60 %" (80% in A-state),
0 . 2 kg." 60-70 %"
I
Change of tension following stretch p between 0-5" Intrinsic Total
28 sec.0
1
-
lo
1
0. OBb
0.53d 0. 56d l.ld
Limited fall of tension and only small aftereff ecta.6
0.08 sec.c
16 secb
Rise during stretch is immediate 67%b >50%h
0.08"
I 0
(2)
plete recoverya*bnc Half-time for redevelopment of tension
0 0
After stretch
7 Y Fall of tension to zero and com(4)
7
Skeletal muscle
Skeletal muscle
0.0270 -
-
1.4
33%b
>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<. M. RUDALL
water. They remark, however, that 150% extension of their materials in cold water lead only t o orientation without intramolecular transformation. I n our own studies we have always found th a t both corneum and mucosum show obvious a-/3 transformation when fresh material is stretched 100-150% in cold water. It is important to know this because Derksen and Heringa’s statement on this point would suggest a lesser cohesion in the structures while we believe the cohesion of intact epidermal proteins is quite high and certainly sufficient to bring about intramolecular unfolding when stretching occurs in the cold. By coiitrast, fresh muscle tissue shows no intramolecular transformation when highly stretched at, room temperature (9). These differences are a s yet imperfectly understood but must be related to the relative strength of inter- and intramolecular forces. 9. Temperature and Stability
I t was, of course, of interest t o know to which class of fibrous proteins the main constituents of the epidermis belonged as judged by X-ray means. Within the group of a-type proteins the most important further criterion (in the absence of any observable small angle diffraction) is the stability of the structure again judged according to changes in the diffrttction pattern caused by temperature, acids, etc. The first of such observations were those of Astbury and Dickinson (8) who found that myosin, though like hair in many aspects of its diffraction, showed the instability of the a-pattern in water a t temperatures above 40-50°C., while hair keratin is quite stable a t 100°C. At the same time as these molecular changes were taking place oriented myosin fibers became shorter, i . e . , show thermal contraction. The epidermis, like myosin, is unstable in hot water and this is the principal physical distinction between hair keratin and epidermal keratin. The thermal instability of the epidermis varies a t different levels of the tissue structure. This variation is best illustrated by thermal contraction curves from a series of layers parallel to the surface of a thick epidermis (49). I n cow’s nose material, the stratum corneum is stable in water a t temperatures up to about 80°C. (a, Fig. 3). Bu t the inner layers of the mucosum show contraatioii at much lower temperatures, i.e. a t about 65°C. (c, Fig. 3). The significance of intermediate types of thermal contraction, a s in curve 0 , prohably depends on the change in solubility of part of the cell in the outer levels of the mucosum. The curves certainly express a real diffcreiiche i n the internal stabilization of the various layers of the epidermis. Direct observation by X-ray means has shown th a t where mucaosal layers (rewetted after drying) were immersed in hot water a t 70°C. for 2 minutes, the a pattern was slightly changed to /3, while a t 80°C. there
was very extensive development of the p pattern. On the other hand, dried corneum layers, rewetted and immersed in hot water for 2 minutes showed no change at 70 or 80°C. and only the slightest development of p pattern a t 85°C'. We can thus define a rise of nearly 20°C. in the temperature of transformation in passing from the mucosal layers t o the corneum; this increase in stability is associated with the change from the zone of free -SH groups t o that of disulfide groups.
20
40
60
80
1
TEMPERATURE, "C
FIG.3. Thermal contraction curves of various layers of the epidermis: a , stratum corneum; b, outer mucosum; c, inner mucosum. Curve for film of cpidermal protein, precipitated from urca solution, is shown a t d.
The thermal contraction curves and the temperature a t which the a-pattern changes to the &form give the best indication of the extent of the internal linkages. Were these structures obtainable as well oriented fibers, of uniform thickness and molecular configuration, load-extension curves and related measurements would give a more complete picture of the kinds of internal bonding. Nevertheless, with the thermal contraction data we can compare various types of epidermis and stratified epithelia, normal and pathological, and obtain useful clues about the internal stability. In Fig. 4 we compare the stratum corneum of the human leg, a, raised as a sunburn blister, and the normally shed stratum corneum of the amphibian, Triturus pyrrhogaster, b. The corresponding
2G4
K. M. RIJDALL
curve for rabbit, myosin is included for comparison ( c ) . While the behavior of the human corneum is almost identical with th a t of the (*ow's nose corneum, Fig. 3a, the amphibian material is markedly less stable t o temperature. Longley (39) noted in studies of epidermal material, using the Prussian blue test, th a t the stratum corneum of amphibian material was rich in reducing groups which he interpreted as free -SH groups. This would give a simple explanation of the main 0
@ 10 Z-
0 l0
a U I-
Z
0 0
20
TEMPERATURE, "C.
0
FIG.4. Thermal contraction curves for: a, human leg stratum corncum; 6 , amphibian stratum eorneum; c, film of rabbit myosin.
difference in stability of the two types of corneum as illustrated in Fig. 4. The stability of the stratum corneum of the amphibian, Fig. 4b, differs little from t ha t of the mammalian mucosum, Fig. 3c. It may be th a t in the mucosum of the amphibian material the stability is still less and may be more like that for myosin, Fig. 4c (50). 111. THE EXTRACTION OF EPIDERMAL PROTEINS 1. Solubility of Cell Structure Proteins . Concerning the nature of cell proteins, there is not much profitable information t o be gained from the chemical analysis of the epidermis as a whole, or for that matter of any whole epithelial tissue. The line of
THE PROTEINS OF THE MAMMALIAN EPIDERMIS
265
progress is very obviously to divide such tissues into their constituent molecules and proceed t o define the constitution of each molecular species. Methods of dissolving such tissues have first to be discovered, as well a s fractionation procedures. The classical example of dissolving epithelial tissues is of course represented by the long series of studies on muscle proteins. Here primary solution is readily achieved b y means of weak neutral salt solutions (30, 11). A variety of “newer” processes, discovered quite empirically, have allowed the fractionation of watersoluble myosin, actin (53) and tropomyosin (12). Other processes may reveal other new components. I n the case of the epidermis the obvious first step was to examine the solubility of the mucosum in neutral salt solutions. For this purpose the pulversized or freeze-dried material was extracted with 0.5 M KC1 a t p H 7, but nothing passed into solution which could be characterized as myosin-like when tested for by the addition of ammonium sulfate u p t o half saturation. These experiments were made a number of years ago, and what mas thrown away then may be of interest in relation t o what is now known. The concept th at the solubility of the proteins in epidermal cells was possibly like that within muscle was abandoned. However, primary solution of epidermal proteins is readily achieved by urea solutions, some dispersal of the proteins taking place in 1 M urea, but really effective solution is obtained only in stronger solutions such as 6 M urea. This indicates that epidermal proteins are far less soluble than muscle proteins, the srale of the difference being seen by analogy with the fibrinogenfibrin system. Fibrinogen is soluble in weak neutral salt solutions while fibrin is not; but fibrin is readily soluble in urea solutions (43, 40). Furthermore Lorand (40) has distinguished two main types of fibrin, viz. those which are soluble in 30% (w/v) urea and those which are not, the difference between the two being th at the urea-insoluble fibrin has been allowed t o react with a factor present in the serum. Three conditions of the fibrinogen proteins are thus apparent, i.e. fibrinogen, urea-soluble fibrin and urea-insoluble fibrin. These different products have occurred as a result of reactions which are specific and controlled as distinct from general denaturation. The analogy of the fibrin system with the epidermal protein system is made closer by the following. Whereas in the stratum mucosum of the cow’s nose epidermis the bulk of the fibrous protein is soluble in 6 M urea, the fibrous proteins from the mucosum of rabbit papillomas (Shope) are quite insoluble in urea. These different states correspond in a general way t o urea-soluble and urea-insoluble fibrin. I n tackling the problems of new proteins like those of the epidermis it is only right to draw upon
266
K. M. RUDALL
the many decades of experience with classical objects like myosin and fibrinogen. And after all these and the proteins of the epidermis give similar X-ray diffraction patterns and are comparable a t least in amino acid pattern (10). Thus in the study of intracellular protein in epithelial cells i t seems profit,able to seek answers to the following questions. Is there a readily-soluble phase of epidermal protein in the lower layers of the epidermis, i.e. a phase analogous to fibrinogen and soluble in weak neutral salt solutions or water soluble? In the development of the intracellular fibrous proteins which are not soluble in salt solution but are readily soluble in urea, i.e. proteins analogous t o fibrin, do enzymes act similarly t o thrombin, with the production of new end-groups in the polypeptide system (14) ? Finally, are other factors involved which lead t o urea-insoluble states like plasma-fibrin? These are possibly general concepts for many types of epithelial cells. Szent-Gyorgyi and Banga (15) have already divided cell structure proteins into two types, i.e. those soluble in 30% urea/KCl and those soluble in alkaline urea/ KCl. The latter could be likened t o the urea-insoluble fibrin of Lorand (40). I n a search for an understanding of the range of solubility of intracellular fibrous proteins we see that the principal types are displayed in the fibrinogen-fibrin system, where, of course, stJudy of the factors involved is fairly well advanced, 2. Extraction of Epidermal P r o k i n
There is no epidermal material which is so readily available in quantity as the cow’s nose epidermis. In countries where horse slaughtering is common the “burrs” or “chestnuts” have been studied (34) and doubt,less form an excellent source of material of the same general class a s tlie true epidermis. The cow’s nose epidermis is practically hairless and also comparatively free of lipid substances, which are abundantly incorporated in other types such as human epidermis, Each head yields a suitable strip of skin about 7 X 2 cm. while the epidermis is 1-1.5 mm. thick, of which about 200p consists of the insoluble stratum corneum (see Fig. 1). If vertical freehand or frozen sections are placed in 6 M urea the lower part of the mucosum rapidly disintegrates into spherical bodies which increase in size from the basal layer upwards. The upper third of the mucosum swells but behaves differently from the lower twothirds in that i t remains intact for a much longer period. Thus if a whole strip of skin, cut across just beneath the epidermis, is placed in 6 M urea solution for a n hour, gentle scraping from below separates most of the mucosum, leaving the entire corneum together with the resistant outer
THE PROTEINS O F THE MAMMALIAN EPIDI!XMIS
267
layer of the mucosum. This outer layer of the mucosum can in turn he removed by further scraping. The character of the solutions of epidermal proteins in urea varies in the following way. Where the whole mucosum is dispersed in 6 M urea an opaque solution is obtained after several hours and this is comparatively viscous. The viscosity rapidly falls and is very low after 24 hours. Under similar conditions 6 M urea in 1 M KCl solution yields a clear solution which is markedly more viscous and this viscosity falls much less rapidly than in urea alone. On dialyzing the urea extracts against NC1 or water a marked difference in precipitation occurs. The bulk of the protein in the urea/KCl extract rapidly precipitates, that in the urea only remains in solution for a much longer time. It is not intended to describe these differences any further a t present, but merely t o indicate that the effect of KCl in the extraction medium appears t o retard the dissolution of the protein complexes, which proceeds more rapidly and probably to a greater extent in urea without salt. At the present stage of the work a constant behavior of the extracts cannot be described. But concentrating upon the study of extracts in urea solution only, we can define two main conditions of the protein. The extract when dialyzed against distilled water a t neutral p H may form a clear coherent clot indistinguishable from that of fibrin. This is normally obtained when the extraction lasts for only 24 hours. I n other cases the extract may remain in solution after thorough dialysis. This behavior is common in extracts made for periods of one week or longer periods, and is more readily achieved with extracts from the outer layers of the mucosum than with those from the lower layers. I n the long run these properties may be traceable to specific interactions of the cell proteins. 3. Purification of Epidermal Protein From fresh epidermis, rinsed in distilled water, layers were cut by hand. These may be described as follows: the outer level consists of the corneum together with the outermost regions of the mucosum, the middle level consists of most of the outer half of the mucosum, and the inner level consists of the lower half of the mucosum. The layers were extracted separately in 6 M urea for periods of 24 hours a t room temperature, after which the supernatants were poured off and new extraction fluid added. After 7 days most of the extractable material had been obtained. The combined extracts of the outer level gave comparatively clear solutions and the middle and inner levels cloudy solutions. The former filtered readily, but the latter two extracts were very difficult t o filter because of abundant minute particles just visible in the microscope. The well-
268
K . M. RUDALL
filtered solutions were dialyzed against several changes of distilled water, after whivh the material from the outer levels remained in solution, while th a t from the middle and inner levels clotted in the sac. The dialyzed protein from the outer level was filtered and acidified with 1 N HC1. Maximum precipitation occurred at about p H 5.5, leaving a clear supernatant fluid. The precipitate was washed with water and redispersed in 6 M urea. The clear supernatant liquid a t p H 5.5 was filtered and further acidified. A cloudy precipitate forms a t about p H 5, the precipitation attaining a maximum near p H 4.5. This settles out on standing. Thus the primary solution of the epidermal proteins in 6 M urea is divided into two main fractions-a coarse, flocculent precipitate at pH 5.5 and a fine, granular precipitate a t p H 4.5. The first precipitate can be dissolved in urea, dialyzed and reprecipitated a t p H 5.5 many times. For the present purposes the processes were repeated three times. The precipitate a t pH 4.5 can be redissolved by adjusting the p H t o 7 with NaOH; these processes were also repeated three times. The materials from the middle aiid inner levels which had clotted in the sac were shaken and the clear fluid poured away from the retracted clots. This fluid gave no precipitate a t p H 5.5, but cloudy precipitates a t about pH 4.5. These were dissolved and reprecipitated three times as above. The main clot-like precipitates were redispersed in 6 M urea, filtered and dialyzed; these processes were repeated three times. I n this way two main fractions were prepared for the middle and inner levels as was done in the case of the outer levels. With each successive reprecipitation of the main precipitate less and less of the fraction precipitating a t p H 4.5 is obtained. The most important point about these epidermal proteins is thus the recognition of two main fractions which occur in all levels of the mucosum. After only 24 hours extraction in 6 M urea the dissolved material gives a firm clot upon dialysis. After 7 days extraction the dissolved material is dissociated into two water soluble proteins in the outer levels, and much the samc is true of the middle aiid inner levels; the main precipitate becomes more water soluble during the sucvessive re-solution in urea and dialysis. 4. MolPcular Weight and AsymmPtry The two main conditions of the soluble protein for which we should like t o have measurements of molecular weight and particle asymmetry are: a) the primary solution in urea which yields a tough clot on dialysis, and b) the materials which have become water soluble on longer extraction with urea. Using the ultracentrifuge, measurements have been made on the
TlIIC P1~OTE:Ih'H O F TIIIS MAMMALIAN EPIDEKMIS
269
primary solution of epidermal protcins in urea for 24 hours (41). The scdirnentation velocities show a wide range, but there is a main group with S = 1.3-1.6, which itself tends to split into two subgroups in some cases. From the measured diffusion constant Mercer and Olofsson calculated a molecular weight of 60,000 and a frictional ratio of 3.5, which would indicate a very asymmetric particle; the authors point out the similarity in shape arid size with the data for tropomyosin (13). In some of these studies of epidermal protein in urea, bisulfite was added and the sedimentation showed a remarkable narrowing of the range of particle sizesthere appeared just one group with S = 2. The main conclusion was that reduction of disulfides brought about the disaggregation of the larger particles, leading to a comparatively uniform particle size. These observations enable us to accept a t least a preliminary picture of molecular size and they confirm the asymmetric shape which was deduced from X-ray studies (40). 5. Sulfur Content of the Isolated Proteins
No studies of amino acids have been made on these proteins as yet; that is best left until we are more certain about the purity of the fractions. There is no suggestion that the proteins contain any nucleic acid as the phosphorus content is nil except in one case (inner level, nonfibrous protein of Table IT). But a number of determinations of sulfur content have been made, and these are of obvious interest. Of the two main fractions, that precipitating a t p H 5.5 is referred t o as the fibrous protein or epidermin, and that precipitating a t pH 4.5 as the nonfibrous protein. TABLEI1 Sulfur Content of Epidermal Proteins Level of epitlcrinis Outer Middle Inner
8 in nonfibrous protein (%)
S in fibrous protein (%)
2.86 2.24 I . 13
0.8 0.9 1.08
The figures of Table I1 have been corrected for ash which, in the case of the fibrous proteins, varies from practically zero to 0.8%. The ash content of the stratum corneum f outer mucosum layers is 1.6-1.796, The quantities of Table IT should be compared with those for the intact epidermis which are given in the last two rows of Table I, uiz., 1.13 and 1.16%. To a first approximation a ratio of 10 parts of the fibrous protein t o 1 part of the nonfibrous protein gives the order of the sulfur con-
270
li. M. HUDALL
tent of the intact epidermis. This is about the ratio of the two fractions weighed after the purification processes. The striking feature of Table I1 is the comparatively high sulfur content of the norifibrous protein and this tends to increase as we pass from the inner to the outer layers of the mucosum. It is of course possible t o read many things into these results, but interpretation should be postponed until the measurements are extensively repeated. The values for the middle and inner levels do support observations made during the purification procedures that in these levels it was more difficult t o dissociate the complex present in the primary solution. Whether this dissociation is a mild hydrolysis which occurs during urea extraction, or even an enzyme action during dialysis, are suggestions which yet have t o be examined. However the separation comes about, the main point of interest is the relatively low sulfur content of the intrinsic fibrous protein and the rather high sulfur of the lesser non-fibrous fraction. Insufficient analyses of epidermal keratins are available t o judge whether sulfur contents are subject to wide variation. Bu t of course very wide variations are known in the case of the hard keratins of various hairs, quills, and horn (17, 38) and even a gross variation in this respect from fiber to fiber within a single fleece or staple (48). The basis of these wide variations seems likely to be due t o a two component system with the main fibrous protein having a very low sulfur content, perhaps even less than 1% as in epidermis. Possibly the first word on this subject was said long ago by Giroud, Bulliard and Leblond (34) who in studies of horn found 3.4% S in the horn layers and only 1%S in the horn mucosum. The mucosum has the main fibrous structure of the horn layer and the suggestion is th a t i t lacks as yet a component which is rich in sulfur. From another viewpoint we may consider the dissolution of wool keratin into two main fractions. By oxidizing the abundant cystine linkages with peracetic acid, Alexander and Earland (1) have found a low sulfur-content fibrous fraction and a very high sulfur-content amorphous fraction. These findings strengthen the concept th at in the development of the whole epidermal system there are two principal types of structure protein, the relative proportions of which are important in determining the sulfurcontent of the whole tissue.
IV. SOMEPHYSICAL PROPERTIES OF EPIDERMAL PROTEINS 1. X-Ray Diffractionof the Fibrous Protein
The protein fraction obtained by isoelectric precipitation a t pH 5.5 is generally prepared by drying solutions as films on silicone-coated glass.
THE PROTEINS OF T H E MAMMALIAN E P I D E R M I S
271
These films are made fairly thick (loop) for diffraction and elasticity studies. Narrow strips of film, when stretched in cold water three t o four times their initial length, give strong a-type diffraction patterns very similar indeed t o those of a-keratin and a-myosin. Very rarely does the stretching in cold water lead t o any visible /3 diffraction patternoccasionally the merest trace is observed. The a pattern (Fig. 5) is characterized by a very strong meridional arc of spacing 5.1 A. (plus a very weak one a t ca. 3.0 A.) and strong equatorial spots of spacing 9.6 A. Unresolved diffuse scattering is present in the background. While stretching in cold water does not give rise t o any significant amount of /3 diffraction, stretching in hot water a t 65-80°C. shows the relatively great effect of temperature on the ease of transformation. I n hot water there is, however, great loss of orientation, and well defined parallel /3 diffraction patterns cannot be obtained (49). But if oriented fibers of epidermin are stretched in 50 % ethanol in the cold a significant amount of well-oriented /3 component is produced. To obtain the maximum amount of this component, stretching was carried out in ethanol at higher temperature, i.e. 70, 80, 90, and 95% a t 78-80°C. I n 70 and 80% hot ethanol the transformation to /3 was essentially complete when oriented fibers were stretched b y loo%, though there was still a tendency t o produce some disoriented /3, especially in 70% ethanol. I n 90 and 95% ethanol the tendency to produce disoriented /3 was negligible, but even a t 100% extension there was still a small amount of a remaining. Upon relaxing in cold water the /3 component formed by stretching in 90 or 95% ethanol was reconverted t o a, but that produced in 70 or 80% ethanol was largely set. For a number of purposes we wish t o obtain complete and reversible a+ /3 transformation, but stretching in hot ethanol does not provide the desired conditions. The most perfect reversible a + /3 transformation has been obtained by stretching in saturated ammonium sulfate a t 90-100°C. Fibers can be stretched quite easily in this solution and show little tendency t o contract when they break. The ammonium sulfate hydrates and lubricates air-dried fibers t o just the right extent to facilitate transformation. There is no tendency to produce disoriented /3, but a difficulty arises during the washing away of ammonium sulfate. With the extended fibers the internal structure relaxes and /Ireturns to a, so th a t washed but stretched fibers show a mixture of a and /3 patterns. This has been overcome t o a considerable extent as follows: After stretching in hot ammonium sulfate and cooling, the specimens are transferred to cold saturated solution containing about 5% formaldehyde. Left in this solution for 36-48 hours, the ammonium sulfate can be washed away,
2 72
I<. M. RIJDALL
leaving the 0 pattern fixed. Fibers stretched in saturated ammonium sulfate a t 90-100°C. contract on washing in water and show no residual p form. At higher t,emperatures such as 105°C. a corisideraltle amount of 0 is set and contraction is less. The ammonium sulfate procedure offers the best prospect of complete and reversible a -+ 0 transformation. The 0 diffraction pattern of epidermin stretched in saturated ammonium sulfate a t 105°C. is shown in Fig. 6. There is just a trace of a pattern remaining; arid there is some spreading of spots along the first layer line (6). The prinvipal spaciiigx are 3.33, 4.66 arid 9.8 A for the main chai ti, backbone and side-chaiii periods respectively.
FIG.5
FIG.
6
FIG.5 . X-ray fitier photograph of epidermal protein (rpiderrniri), a-form. FIG.6. 8-form of epidermin.
2 . X-Ray Dij’raction of the NonJibrous Protein The prot)ein fraction precipitating a t pH 4.5 was dried off as a film from neutral solution. The film is very fragile and no fiber can be drawn from it. I n X-ray studies the diffraction pattern is of the 0 type arid shows no orientation with reference to the plane of the film. The film is not again soluble in water after drying. In addition t o the chemical differences arid the isolecatrici points, the nonfiber-forming protein is sharply distinguished from the fiber-forming protein (epitlermin) by its diffraction pattern. 3. The &feet of Temperature o n Various Preparations of Epidormal Protein
The effect of incbreased temperature on the intact epidermis, whether corneum or mucosum, is t o cause contraction of the tissue specimen and a spontaneous a +/3 transformation. Isolated epidermal proteins show the same effect. Fibers precipitated from the urea solution with ammonium
273
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
sulfate have behaved very similarly to the lower layers of the mucosum with respect to the kind of thermal contraction they show (Fig. 3c, d ) . However, thermal contraction of purified epidermal protein occurs a t much lower temperatures and has given reason t o believe that there may be complex formation of two or more protein types in the intact epidermis and also in the primary solution in urea. Thermal contraction begins, in the case of purified epidermin, between 30 and 40°C. and is nearly complete before the transformation, a -+,8,
0 c
I
I
20
I
1
40
I
I
60
1
I
80
I
TEMPERATURE, "C.
FIG. 7.
Repeated thermal contraction of the same fiber of epidermal protein. first contraction; B , second contraction; C, third contraction.
1 I A,
becomes detectable a t 65°C. Hence much of the thermal contraction is independent of the visible molecular transformation; this independence is further illustrated in the following way. I n Fig. 7 the curve A is for epidermin fiber thermally contracted by nearly 50% as a result of 1 minute periods of immersion in water a t each 10" interval up t o 100°C. In this process all the crystalline material had been fixed in th e p form. The fiber was then stretched highly to orient (or unfold) the structure again, arid relaxed in cold water to constant length. A second thermal contraction, R, was then induced; this shows that most of the contraction occurs a t the same temperature range as in A. Very noticeable in B is the considerable loss of ability t o contract a t the higher temperature
274
K. M. RUDALL
ranges, i.e. where the CY --$ /3 transformation occurred in A. The conclusion would be that the set fi fraction does not contribute to the contraction, while this same fraction when it was in the a form caused a considerable proportion of the Contraction taking place above 55°C. The lesser amount of total contraction may be due in part to a lesser initial orientation in B than in A. Another cause may be th a t the first heating t o 100°C. affects the contractility of all the chain molecules t o a certain extent, for example by causing cross-linkage. This latter effect does not seem t o be very great because of the results for a third thermal contraction of the same fiber, shown a t C, which is closely similar to B with only a slight further loss in total contraction. These studies show the possibility of relating a particular part of the thermal contraction to the formation of cross /3, i e . , that part occurring above about 55°C. According t o the infra-red studies, described later, there is no evidence tha t the thermal contraction a t 30-55'C. involves cross @ formation. And because this region of contraction is repeatable after successive reorientation, it does not involve chains which are set in a @-likeconditioil. 4. Cross @ Form of a-Type Proteins The p structure produced by the thermal contraction of intact epidermis, or of isolated epidermin, is unusual and t o be distinguished from the normal parallel @ form produced by stretching, where the main chains lie parallel to the axis of stretching. If thermally contracted material is stretched SO-lOO% and dried, the /3 pattern is oriented so as to show that the main chains lie for the most part across the axis of the fiber i.e. a t right angles t o the direction of stretching. The diffraction pattern is here referred t o as cross (3 (49). The two main conditions for the production of cross @ are a high temperature (above 65°C.) and a sufficient degree of swelling with water. The degree of swelling has a very wide range, since cross @ formation takes place when aqueous solutions of the protein are heated, and also with fibers in 70% ethanol a t 80°C. The chain molecules must be sufficiently free t o move as a result of thermal agitation. The stretching of such contracted systems reveals th at crystalline regions are still asymmetric in shape but become oriented with the main chains a t right angles to the direction of stretching. The process of stretching from a contracted state is not necessary to reveal this change. Oriented fibera held a t constant length and heated in water show the transformation from a to cross /3 though there is always some small loss of orientation in the process. Thus, during isometric contraction, thermally produced, the molecular chains move from an orientation parallel to the fiber axis t o one where they lie approximately
THE PROTEINS OF THE MAMMALIAN EPIDERMIS
275
a t right angles to it. In this process do the a chains transform to parallel chains which subsequently become oriented across the fiber axis? The earliest observed production of 0 pattern occurred after 2 minutes treatment in water at 65°C. in a condition of isometric contraction. The fiber was then extended by 100 % ’ to thoroughly align asymmetric units in the
FIGS.8-11. Progressive formation of cross p from a by heating epidermin fibers in hot water for 2 minutes at various temperatures: Fig. 8, 65°C.; Fig. 9, 70°C.; Fig. 10, 80°C.; Fig. 11, 100°C. Fiber axis vertical in each case.
structure. Although only a trace of 0 pattern is observable, this is all oriented as cross 0 showing the direct transformation of a to cross P (Fig. 8). The progressive increase of cross /3 which occurs with 2 minute treatments a t 70,80, and 100°C. is shown in Figs. 9, 10 and 11. Had the process of cross 0 formation occurred by the intermediate production of parallel 0 then it should have been observed in the earliest stages of these changes. The isometric contractions described above are of oriented fibers,
K. M. RUDALL
276
i.e., films stretched by about 200%, dried, and relaxed in cold water to constant length giving a contraction of 40-50% (see Fig. 13). If the isometric contraction is performed on the stretched fibers (200 % extension from the original film) the diffraction pattern is a mixture of cross p and parallel /3. If the particular orientation of the pattern depends upon the degree of crystallization along the a axis (backbone direction), we should expect a continuously varying asymmetry for the shape of p micelles. The superposed cross p and parallel /3 diagrams shorn th a t the
I A
B
FIG.12. Possible forms of polypeptide chains in cross p configuration. A, supcrfolding of a single long chain; B, aggregation of a niirnher of shorter chains.
asymmetry is in two distinct groups, i . ~ .relatively , elongated in the a direction arid relatively short in the a direction. This is consistent with the view t ha t cross p and parallel /3 diagrams are due to two separate phenomena and are not dependent on one continuous function such as the degree of crystallization along an axis. The special interest in cross p patterns is that they could he interpreted as a superfolding of polypeptide chains as in Fig. 12A. The only indication t ha t this is so depends on their association with the process of thermal contraction in systems with long fibrous molecules, e.g., fibrinogen, myosin, epidermin, and the fact th at by high stretching cross /3 systems can be extended t o the parallel /3 form (49). The alternative explanation is that they may he due to aggregation of shorter polypeptide chains in the backbone direction as in Fig. 12R (7). Probably the best means of arriving at the correct answer is by the study of the frirtional
THE PROTEINS OF THE MAMMALIAN EPIDERMIS
277
ratio and particle weight of cross p produced from a! in aqueous solutions or in t ha t concentration of urea, less than 3 M , which allows the cross p form t o be produced. 5. Reversibility of Cross p Form Cross p, formed in the intact epidermal layers or in isolated epidermin, is readily reconverted to the a structure by means of saturated urea and to a large extent by means of 6 M urea. These epidermal proteins have been ideally suited for demonstrating this phenomenon (49). In principle the same can be demonstrated with myosin but in that case urea has a very marked depolymerizing effect and a destructive action on the a diffraction pattern (26, 49). It should be very interesting to follow this p + a change, which is caused by urea, in some other proteins. For example, in the classical rase of the urea denaturation of serum albumin (44) is it a superfolded p structure (cf. Fig. 12A) which is being unfolded to ail a structure? Or is the negligible effect of urea on the frictional ratio of hemoglobin (52) due to the predominantly a-type structure proposed for this molecule (47) ? IJrea reverses the cross p structure formed by heating in water. It is interesting t o note th at on heating epidermal protein in strong urea no p structure is formed. In weaker solutions, i.e. 9 and 16% urea, heating a t 100°C. for 2 minutes gave complete a + p transformation. But in 23% urea a t 100°C. only the merest trace of p is formed, while a t higher strengths no a --+ 0 transformation is apparent. Urea thus opposes the effect of heat in causing chain unfolding. This phenomenon may throw light on the negative temperature coefficient of urea denaturation which is observed in a number of proteins (37). Here a n effect on protein structure is observcd in which addition of urea revcrses the effect of increased temperature. 6. Elastic Properties of Epidermin Films and Fibers
Films of epidermal protein, as cast upon a glass surface, have a natural orientation with the long fibrous molecules lying approximately parallel t o the surface, but there is no orientation with respect t o a n axis in the plane of the surface. These conclusions are deduced from diffraction patterns taken with the X-ray beam parallel and perpendicular t o the surface of such films. When strips of film about 2-mm. wide are stretched gently in cold water the most frequent point of breakage occurs a t extensions between three and four times the initial length. A considerable number of measurements on the relationship between the extension and contraction of films are grouped together in Fig. 13. Films were stretched in cold water (ca. 15OC.) 25, 50, 100, 150% etc. of
278
K. M. RUDALL
the initial length and after each extension they were allowed to contract in cold water for 4 hours. The percentage contraction is given in the vertical ordinate. When stretching in cold water over the range of 200-300% extension the amount of contraction tends t o a maximum of about 50% (full line curves). This maximum contraction is very similar to the 100% extension proposed by Astbury and Woods (6) for the full a -+ fl transformation. It is conceivable that a curve similar to th a t of Fig. 13 should be obtained for all soluble a-type proteins, and it would be 70
se 50 z
0 k
0 a
30 0 0
10 I
I
I
100
200
300
I
400 EXTENSION, % ’
I
I
500
600
FIG.13. Relation hetween extension and contraction of stretchrd films of epi-stretching in rold water; - - - - stretching in saturated ammonium
dermin. sulfate.
especially interesting to know whetjher there is continued flattening of the curve a t about 50% contraction for much higher extension. Some figures for the contraction occurring at higher extensions were obtained by stretching in cold or hot saturated ammonium sulfate and in these cases there is obvious a-fl transformation as seen in parallel X-ray studies. The contractions obtained by stretching in ammonium sulfate and washing away the salt for 24 hours are given in the dotted line curve. These contractions range between 50 and 60%. While supercontractiori phenomena cannot be ruled out as a possible explanation of the increased total contractions, it does not seem unlikely t h a t these contractions are, for the most part, related to the amount of folding taking place during the conversion of p- t o a-chains. * Where some fl pattern is set, as by stretch-
* Principal fiber axis periods are respectively 3.33 A. and 1.5 A. for @ and 01 epidermin, and if we take these as the corresponding repeat of the amino arid residues then the contraction from @ --f (Y should be ca. 55%.
THE PROTEINS O F THE MAMMALIAN EPIDERMIS
2 79
ing in ammonium sulfate a t 105"C., the contractions obtained are much less than 50%. The curves of Fig. 13 give a clue as to why films break when extended in cold water by 200-300%. This is the region where maximum contraction occurs when the stretched film is released. The phenomena suggest t ha t a maximum has been reached in the orientation of continuous phases in the fiber, and that the cohesion of the structure is insufficient to allow further extension by unfolding of chains. Where greater cohesion obtains, as in saturated ammonium sulfate, unfolding of oriented chains proceeds continuously. There is also a continuous slipping, but with each process of slipping some new chains seem to be in a position to be unfolded. In contrast to natural hair fibers, where many of the chain molecules are linked in parallel and in series (and therefore there is no significant slipping) (61, unstretched epidermin behaves as if this linking of molecules was very slight. But as stretching is continued in saturated ammonium sulfate there seems to be an increasing linking in parallel, and it may even be possible by continued stretching t o transform practically all the fibrous molecules into the fully extended form.
V. INFRARED ABSORPTION STCJDIES 1. General
Our ideas on protein fiber structure are mainly due to Astbury. The long range reversible transformation of a -+ fi forms, and the recognition of crystalline and less crystalline or amorphous phases are among the principal discoveries of Astbury and his colleagues. These were very difficult advances to make, and there was no independent or previous study which gave anything like a quantitative picture of the folding of polypeptide chains. I n recent years, infra red spectroscopy has been able to build on these fundamental observations in a very welcome way. I n the main i t confirms what was already known, but is particularly useful in enabling studies to be made in non-crystalline regions (regions of no regular diffraction) where X-ray and elastic studies give a less adequate picture of the structure. It is most pleasing of all in th a t it may lead t o a recognition of mistakes in interpretation that can not be ruled out of the pioneer's work. Using polarized infrared radiation, a team of workers a t Courtaulds Ltd. Research Laboratory, Maidenhead, England, have paid particular attention t o the orientation of the hydrogen-bonded link between the carbonyl and imino groups of the polypeptide main chain. In the a-form of keratin, myosin, and tropomyosin they find that the CEO and N-H bonds are directed mainly along the fiber axis, while in sundry
2 80
K. M. RIJDALL
p-forms these bonds are directed a t right angles to the fiber axis (2, 31, 1 6 , 3 ) . This is strong confirmation of the view that the main chain in the a-form is folded so as to make segments of the chain lie more or less along the transverse axis of the fiber. Bu t the infrared work has discovered many other things such as the apparent absence of unbonded -NH or -C=O groups; infrared is absorbed alike in crystalline and '' amorphous " regions and reveals no difference in chain configuration between them. One of the most remarkable observations is the recognition of a characteristic frequency difference between the carbonyl vibrations of folded and unfolded polypeptide chains (31). All these successful studies recall an older battle where many thought th at the investigation of fibrous molecules had no contribution to make t o the structure of the protein molecule and protein crystals. I n the following sections are described infrared studies of the principal fiber-forming rlpidermal protein. For these results and their main interpretation I am indebted to my colleague Dr. S. E. Darmon. 2 . Oriented Films of a Epidwmin
Uniform clear films of a! epidermin, freed from the non-fibrous /3 vomponent, were chosen of thickness about 2 p . Pieces were mounted on stainless steel frames, the initial area being about 1 X I cm. After wetting in water these films were stretched a little more than three times their initial length, which is about the limit th a t can be achieved safely with such delicate films. According to Fig. 13 extensions of this order give nearly the maximum contraction on release. The infrared spectra of such a film with 200y0 extension are shown in Fig. 14A for the ranges 3600-2600 cm.-' and 1750-1450 cm.-l A neighboring film, stretched and dried a t the same time, was allowed to contract in cold water and reached a steady length showing a contraction of 42%. The film was dried a t this length and the infrared spectra are shown in Fig. 14R. This may be referred to as contracted epidermin, while the data of Fig. 14A are for stretched epidermin. The infrared spectra of the air-dried films were recorded with the electric vector parallel to the fiber axis, perpendicular t o the fiber axis, and runs were also made with n o polarization of the radiation. The amount of absorption is plotted as optical density to make the peak heights directly proportional t o the specimen thickness or concentration of absorbing bonds. There are differences in the thickness of the films, the contracted film showing a greater absorption a t the CH frequencies around 2900 ern.-' As the CH band a t 2925 em.-' shows n o dichroism its absorption is used as a n approximate measure of film thickness. The optical densities of Fig. 14 have been adjusted with reference to a constant absorption at
THE PROTEINS O F THE MAMMALIAN EPIDERMIS
281
2925 cm.-' For both Figs. 14A and €3 the preferred orientations of the NH stretching bands at 3280 ern.-' and 3060 cm.-', the C=O hand a t 1655 cm.-' and the NH deformation a t 1545 cam.-' are erit,irely similar to I
I
C
3400
3000
,
Z
I
1700
I
I
1600
1500
; $ ,'>
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
<
1 - 7
which might
NH
be regarded as analogous to the carboxylate ion. Lenormant (1948), who made this proposal, considers that the ionic form coexists with the generally accepted keto form in the monosubstituted amide but that in cyclic and disubstituted amides, only the keto form is present. This explanation is hard t o reconcile with (2), (4) and (6) although it would account for ( 7 ) and (9) and could possibly be consistent with (8).
304
G . B. B. M. SUTHERLAND
It would appear wiser, therefore, to rbgard the assignment of this frequency t o a deformation motion of the NH group as still open to question and until further research has clearly established the origin of the 6.4 absorption band, deductions regarding structure based on it should be treated with some reserve. IV. POLYPEPTIDES The infrared spectrum of one of the first synthetic polypeptides made by Woodward and Schramm (1947) was investigated by Darmon and Sutherland (1947) and found t o exhibit intense absorption bands a t the positions expected for the peptide link from work on the amides vix., 3330 cm.-l, 1670 cm.-' and 1560 cm.-l About twenty synthetic polypeptides since investigated b y the same workers all show these same three characteristic absorption bands, which may be regarded as well established key bands for the identification of a polypeptide structure (cf. Fig. 8 in Katchalski, 1951). The next question which naturally arises is whether the side groups RIRz . . . can be identified spectroscopically in a polypeptide by infrared methods. This has been investigated by Dalgleish, Darmon, Davidsson and Sutherland, but only a few of the results have as yet been published. The general method has been t o study the spectra of polymers of individual amino acids and compare these with the spectra of copolymers of two or more amino acids. It appears th a t certain amino acid residues can be characterized by specific absorption bands, which may presumably be attributed to the hydrocarbon part of the amino acid. I n this way, the glycine and alanine present in silk fibroin can be identified spectroscopically in the intact silk fiber (Astbury et al., 1948). Similarly, any amino acid containing phenyl groups can be picked up in a mixed polypeptide containing an amino acid of this type e.g., phenylalanine. It is found t ha t the spectrum of poly-L-leucine differs markedly from that of poly-D-leucine. Certain frequencies are common to all polypeptides viz., 1230 cm.-', 1160 cm.-' and 720 cm.-' and presumably arise from vibrations of the backbone. There is some evidence that the 1160 cm.-l frequency may be due to another deformation vibration of the N H group. The extent to which these findings could be extended t o provide a n independent method of protein analysis for amino acids is, however, strictly limited. Only proteins built from a few amino acids could be tackled with much hope of success and the method would be useless for investigating the amino acid composition, although it might give information on the repeat pattern. However, in cases where one or two amino acids dominate a protein structure, it may be possible, when the
POLYPEPTIDES
THE STRUCTURE OF AMINO ACIDS,
AND PROTEINS
305
protein takes part in a reaction, t o see whether these abundant amino acids are being removed from the protein preferentially or are being altered structurally. There would therefore appear to be potential applications of infrared analysis in the study of the metabolism of certain proteins and in following the course of model experiments with synthetic polypeptides, as a guide to the chemical behavior of proteins towards various compounds. Some of the most interesting results which have come from a study of the infrared spectra of synthetic polypeptides, have been obtained by investigating the polarization properties of the N H and CO bands in an oriented film and correlating these results with X-ray diffraction studies on the same films. Ambrose and Hanby (1949) showed in this way that the polymer (IV) formed by the polymerization of the N-carbonic anhydride of L-glutamic 7-methyl ester could not have a fully extended structure such as nylon. In (IV) the N H and CO bands at 3 p and 6 p had polarization properties indicating that these groups must be oriented mainly parallel to the direction of extension (orientation produced by
[ r,'7
COOCHi
I CH2 I
CHP
I XH*-CH-CO--
-NH-CH-CO
COOCH,
1
CHZ
n
tH2
--NH-
H-COOH
IV
stretching), whereas in nylon (where X-ray methods have shown th a t the CO and N H groups are essentially perpendicular to the direction of stretching) the opposite polarization characters occur. They proposed a folded structure of the type V to account for these results. This structure will be discussed in the following section on proteins. R I
C
0
H
---\ //
\N/
C
I
N
\A/
/ \ H
€3
/I\
I C /Xo
H \ // C
I
N H / \
O 0
H
\ /-N
I H /No C
\y I
I
R
R V
I n the work of Ambrose and Hanby (1949) attention was concentrated mainly on the N H bands near 3 p. Ambrose and Elliott (1950 and 1951) studied the CO band a t 6 p in a DL-phenylalanine-L-glutamic methylester
306
G. B . B . M. SUTHERLAND
copolymer which X-ray studies had indicated could be obtained mainly in an Q (folded) form, when prepared from m-cresol solution, but in a ,8 (extended) form, when prepared from formic acid solution. They concluded that in the a form the CO frequency is a t 1658 cm.-l (6.03 p ) whereas in the /3 form this frequency is lowered t o 1629 cm.-’ (6.14 p ) . Concurrently, they considered that a change also occurs in the N H deformation frequency to which they assign a value of 1550 cm.-’ (6.45 p ) in the a form and 1527 cm.-’ (6.55 p ) in the form, but found no corresponding shift in the N H stretching frequency near 3 p. They point out that this may provide a method of distinguishing between a and /3 forms of a polypeptide chain in unoriented or amorphous materials.
V. PROTEINS The first attempt to apply infrared analysis to the structure of proteins was made by Vles and Heintz (1937) who had previously recorded the spectra of a number of proteins and amino acids (Heintz, 1935; Vles and Heintz, 1935) in the region between 2 and 8 p. Unfortunately, the spectrometer employed had insufficient resolving power to reveal all the bands now known to be present and by restricting attention t o wavelengths below 8 p, much valuable information was missed. Vles and Heintz (1937) claimed that the spectrum of serum albumin was practically identical with a “synthetic ” curve obtained by ‘ I adding ” the absorption spectra in the same region of the constituent amino acids, viz., leucine, lysine, glutamic acid, arginine, cystine, histidine and aspartic acid. In the light of the knowledge we now have, the agreement between the observed and calculated absorption curves cannot be regarded as significant. The spectra of a few proteins in the range 1-15 p had been recorded previously by Stair and Coblentz (1935), but no attempt was made by these workers t o interpret the absorption bands. Buswell et al. (1940b) examined the spectra of about twenty proteins in the range 2.5 t o 3.5 p . All exhibited a strong band at 3.0 p , a weaker band of variable intensity near 3.22 p and another strong band a t 3.35 p. The first two bands were assigned to N H stretching frequencies, perturbed by hydrogen bonding, the former to single N H . . . 0 links and the latter to cyclic dimers (cf. p. 301). The 3.35 p band was of course assigned t o CH stretching vibrations. These investigators noticed that the 3.22 p band was stronger in proteins, where denaturation was more likely t o have occurred, and suggested that the removal of water might facilitate the formation of the ring dimers they presumed were the cause of that band. However, subsequent work has not confirmed that denaturation can be correlated with any increase in the 3.22 p band (cf. p. 309).
THE STRUCTURE OF AMINO ACIDS, POLYPEPTIDES AND PROTEINS
307
Darmon and Sutherland (1947) showed that the spectra of high molecular weight synthetic polypeptides and of proteins of the keratin type are each dominated by bands characteristic of the peptide link, vix., 3300 cm.-l (NH), 1650 cm.-l (CO), and 1550 cm.-l (NH?) providing very strong evidence in favor of the classical polypeptide picture of protein structure. Klotz and Griswold (1949) showed that the same was true for a native globular protein such as bovine serum albumin. Subsequent work by various investigators has confirmed these findings for a wide variety of proteins and polypeptides. 1. Polypeptide C h a i n Folding
The dominating feature of the peptide link in protein structure having been satisfactorily confirmed by infrared analysis, the next step is to investigate whether information can be obtained about the configuration of the polypeptide chains in various proteins. This involves the use of polarized radiation to determine the direction of the CO and N H links through observations on the dichroism of the associated absorption bands in proteins where a considerable degree of order exists, i.e., in crystalline or highly oriented proteins. Important results in this field have been obtained by Ambrose et al. (1949) who showed that in oriented films of P-keratin (swan feather) the NH groups must be predominantly perpendicular to the direction of extension of the polypeptide chain, as determined by X-ray analysis. This is in agreement with the picture of the fully extended polypeptide chain proposed for these proteins by Astbury and Street (1931), the chains being cross linked through CO . . . HN hydrogen bonds as in nylon. However, for stretched films of myosin, tropomyosin, and for a-keratin (porcupine quill), the dichroic properties of the N H bands near 3 p are exactly reversed showing that the maximum change in electric moment for the NH vibration takes place parallel to the polypeptide chain (cf. Fig. 10 in Katchalski, 1951). On the basis of these results, Ambrose, Elliott, and Temple proposed a model structure of the type V for the folding of the polypeptide chain in the a form of these proteins, i.e., essentially the same as that given earlier for the glutamic ester polypeptide by Ambrose and Hanby (1949). Similar seven membered rings for folded polypeptide chains had been suggested earlier by Huggins (1943), Zahn (1947) and Simanouti and Mizushima (1948) although with different arrangements of the side chains. Bamford, Hanby, and Happey (1949) came to identical conclusions from X-ray investigations of some of the same polypeptides and proteins as those investigated by Ambrose, Elliott, and Temple. However, these conclusions have been criticized by Astbury (1949) and by Bragg et al. (1950).
308
G. B. B. M . SUTHERLAND
Darmon and Sutherland (1949) criticized the conclusion of Ambrose, Elliott and Temple as premature on the following grounds. 1. The interpretation of the bands found in a protein in the 3 p region is by no means complete. Although the main band near 3280 cm.-’ is an N H frequency, in which the hydrogen is bonded to a CO group, there is a band near 3060 cm.-’ the origin of which is not a t all clear (cf. p. 301). There is evidence (Astbury et al., 1948) t ha t this band is affected by the transition of a protein from the CY t o the @ form and similarly for polypeptides (Ambrose and Elliott, 1951). There is evidence from work on the spectra of simple monosubstituted amides in solution th a t this band may arise from hydrogen bonding between two peptide links in the cis configuration. (Darmon and Sutherland, 1949 and other unpublished work by the same authors.) However, variations in hydrogen bonding configurations are not the only possible cause of multiplicity of NH frequencies. Interaction between hydrogenic frequencies in particular crystalline states of a molecule can cause two bands to appear, where otherwise, only one would be anticipated. There is also the possibility of multiplicity arising from what is known as “Fermi resonance,” a phenomenon familiar to spectroscopists in the interpretation of the spectra of small molecules. It has still to be established decisively to what extent these various factors are operative in the rather complex pattern which more recent studies show exists in the absorption of a protein near 3 p. 2. The dichroic ratios* observed in a-keratin, myosin and tropomyosin are extremely small (cf. Goldstein and Halford, 1949) and until an independent method is available for estimating the degree of orientation in the polypeptides and proteins being investigated spectroscopically, no proper quantitative proof can be given for any model, although some models can be shown to be more probable than others. Indeed it appears from more recent work by Ambrose and Elliott (1951) that as their synthetic polypeptides become more highly oriented, the dichroic ratio of the CO band is too high to fit the seven-membered ring fold. Furthermore, although the N H frequencies near 3280 cm.-l and 3060 cm.-l
* The dichroic ratio is defined as the ratio of the optical density with the electric vector of the incident radiation parallel to the defined axis of orientation to the optical density with the electric vector perpendicular to that axis. With conventional log I d I r notation this is written ~’ log Io/Ifl
THE STRUCTURE OF AMINO ACIDS, POLYPEPTIDES AND PROTEINS
309
show similar polarization properties, the dichroic ratio of the 3060 cm.-' band appears to be different from that of the 3280 crn.-I band.
It seems to the reviewer that more fundamental experimental and theoretical work must be done on the interpretation of the spectra of the NH frequencies in amides and polypeptides, especially in relation to hydrogen bonding and crystalline structure, before categorical statements can be made regarding protein structure by infrared analysis. 2. Protein Denaturation
It was mentioned earlier in this review (p. 306) that Ambrose and Elliott (1950) had noted that the CO frequency near 1640 cm.-' appeared to change appreciably in a synthetic polypeptide according t o whether the polymer was prepared in the a form (1660 cm.-') or in the /3 form (1630 cm.-l). Elliott et al. (1950) have applied this result in a spectroscopic study of the denaturation of insulin to see whether evidence could be obtained for the unfolding of polypeptide chains in the denaturation process. They found that insulin denatured by boiling in dilute HCl solution showed a maximum CO absorption near 1640 cm.-l with a minor maximum near 1670 cm.-', whereas renatured insulin, precipitated from aqueous phenol, showed a wide single maximum at 1660 cm.-l This may prove to be a most significant result in a region of protein chemistry in which much controversy has raged, indicating a new line of attack on an old and very difficult problem. In this connection, Uzman and Blout (1950) have studied the infrared spectrum of native, denatured, and regenerated pepsin films in the region of 3 p and noted on denaturation a marked increase in the intensity of the 3270 ern.-' NH frequency relative to the CH bands near 2950 cm.-' This increase is partially reduced by regeneration. They suggest that the increase in the intensity of the principal hydrogen bonded N H frequencies on denaturation is consistent with the greater possibilities of intermolecular bonding, if the denaturation process consists of an uncoiling of the polypeptide chain of the native protein. Whatever the final interpretation of these results may be, it is important that spectroscopic changes have been observed to occur with denaturation for these must eventually yield some information about this very puzzling phenomenon. 3. Contraction of Muscle Protein The exact mechanism by which the protein in muscle fiber contracts and extends has been the subject of much research but it is still far from clear. As might be expected, some attempts have been made to see
3 10
G. B. B. M. SUTHERLAND
what information can be gleaned by the application of infrared methods. Lecomte et al. (1942) first showed that an infrared spectrum with reasonably sharp bands could be obtained from dried muscle fibers. Their observations covered the region between 6 and 15 p, where they observed bands at 11 (weak and broad), 9.6, 8.1, and 6.35 p. They found no difference between the spectrum of a resting muscle fiber and that obtained from a fiber stimulated isometrically. Changes were found when the muscle had been contracted (before freezing and drying) in that the intensities of all the bands, except that at 6.35 p, appeared to diminish considerably. They also examined the spectrum of myosin which was found to have three bands (viz., 9.6, 8.1, and 6.35 p ) coinciding with those of muscle. No interpretation of these results was offered. More recently Morales and Cecchini (1951) have looked into this same point but they report no very significant differences between the spectra of resting and of contracted muscle fiber, apart' from small possible alterations in bands near 8.33 and 10.75 p. It should be noted that here the method of freezing and drying the muscle was not the same as that used by Lecomte, Dubuisson, and Monnier and that this may account for the different result. These observers also examined the spectrum of actomyosin, recording some 20 distinct bands between 3500 and 900 em.-'; of particular interest were bands found a t 3455 and 3368 em.-* which, as they suggest, could arise from N H or OH groups which are not hydrogen bonded. Morales and his coworkers in a second paper (1951) found no evidence for SH groups in actin, although there were indications of an SH frequency in myosin, actomyosin and insulin similar to one detected by them in cysteine and glutathione. It must be remembered that the SH band is generally rather weak in intensity and failure to find it in the first instance should not be regarded as finally excluding the presence of SH groups in actin until quantitative intensity measurements have been made on the SH band in a variety of compounds. Morales and Cecchini (1951) found that the main difference between the spectrum of muscle fiber and that of actomyosin consisted in the former having more intense absorption at frequencies below 2000 em.-' and suggested that this might arise from the presence of nucleotides or glycogen in the muscle fiber. However, work by Wood (1951) indicates that the concentration of glycogen in muscle is probably too low to account for much of this absorption, at least in wet muscle. The work of Wood (1951) is of particular interest since (I) he used a reflecting microscope and so was able to study the spectrum of a single muscle fiber and (2) he kept the fiber alive in Ringers solution, using a compensating cell full of the solution to enable him t o cancel out the strong absorption due to water. Wood examined single fibers from
THE STRUCTURE OF AMINO ACIDS, POLYPEPTIDES AND PROTEINS
311
various muscles of a particular animal (frog) and also from various animals. He found that certain bands are common to all muscle fibers (6.5, 6.85, 7.10, 7.65, and 8.00 p ) while others (in particular, a band a t 9.71 p ) vary greatly in intensity in different fibers, including those from different muscles of the same animal. There appeared to be no obvious correlation between the appearance of the band at 9.71 p and the structure or function of the muscle from which it came. In the light of these preliminary investigations, what conclusions can be drawn about the use of infrared methods on the problem of muscular contraction? In the first place more experimental work is required to establish with certainty the spectroscopic changes which take place in contraction and whether these occur in bands common to all muscle fibers or in bands which vary in intensity from muscle to muscle. In particular, changes in the dichroic properties of bands must be carefully sought for in connection with theories of folding of polypeptide chains. Morales and Cecchini (1951) found little or no dichroism in their spectra but such effects have been reported (Thompson, quoted by Morales and Cecchini, 1951). The next question is whether any changes observed can be correlated with chemical changes postulated during contraction. Here some work of Morales and his coworkers (1951) in which they have followed the dephosphorlyation of ATP by actomyosin by watching the decrease in the intensity of a band at 911 cm.-l (attributed to ATP) and the appearance of a band at 930 cm.-l (attributed to ADP) may be very significant in the polymerization of G actin to F actin although Wood's work would indicate that the chances of detecting the minute amounts of ATP and ADP present in an u n d e h y d T ~ ~muscle e ~ fiber are rather poor.
4. Other Work o n Proteins In the preceding sections we have dealt with work on the infrared spectra of proteins in relation to certain specific problems, viz. polypeptide chain folding, denaturation and muscular contraction. There are a few recent investigations on proteins which do not fall into these categories and which should be mentioned. Buswell and Gore (1942) examined the spectrum of salmine between 2.5 and 4.5 p comparing the spectrum obtained with those they found for arginine and proline in the same range. As might be expected, no very definite conclusions could be established from spectra covering such a short range. Klotz et al. (1949) have investigated the spectra of tyrocidine, gramicidin, salmine, polylysine, cytochrome c, lysozyme, pepsin, bovine serum albumin, and complexes of some of these proteins with sodium dodecyl sulfate, in the range 2.5 to 10 1.1. They have attempted to correlate certain absorption maxima in the spectra with the presence of particular groups known to predominate
312
G . B. B . M. SUTHERLAND
in these proteins from amino acid analysis (cf. Fig. 9, Katchalski, 1951). They noted some small differences between the spectra of the proteins and those of the complexes with sodium dodecyl sulfate but were unable t o demonstrate interaction between cationic groups because of the strong absorption due to the sulfate ion. Blout and Mellors (1949) have examined the spectra of thin sections of normal and cancerous tissue and found that in addition to the usual protein bands a t 3.0, 6.0, and 6.5 p there were differences in the bands beyond 6.5 p in the various materials. The potentialities of the reflecting microscope in obtaining the spectra of proteins were first illustrated in the work of Barer et al. (1949). A fuller discussion of the use of the reflecting microscope for infrared work has been given by Blout et al. (1950). Combined infrared and X-ray analysis of chitin and related compounds by Darmon and Rudall (1951) groups of have demonstrated hydrogen bonding between -CO-NHadjacent acetamido side chains to form linked piles of chitin chains within crystalline regions. Elliott and Ambrose (1951) have obtained some evidence that in a single haemoglobin crystal the NH bonds are directed more along than across the a axis.
VI. SOMEGENERALOBSERVATIONS
It should be realized th at the application of infrared analysis t o the structure of large molecules is still little beyond the exploratory stage. At present only a fraction of the observed absorption bands can be interpreted with certainty and therefore only a fraction of the potential information can be derived even from the existing spectra. The results obtained to date should be regarded as merely giving some indication of the potentialities of the method, yet these results are by no means inconsiderable. The preceding survey shows that through infrared analysis i t has been possible (a) t o obtain an independent proof of the dipolar ion structure of the amino acids, (b) t o establish a new physical test for the identity of each amino acid, including in many cases a distinction (in the solid state) between the m-form and the D or L form, (c) t o recognize certain amino acids when incorporated into polypeptides and proteins, (d) t o obtain an independent proof of the dominating role of the peptide link in the structure of proteins, (e) t o establish an independent proof that in certain synthetic polypeptides and certain proteins the essential structure consists of
THE STRUCTURE OF AMINO ACIDS, POLYPEPTIDES AND PROTEINS
313
polypeptide chains linked by hydrogen bonds between the CO and NH groups a t right angles t o the chains, (f) t o obtain some information on the orientation of the CO and N H groups in polypeptides and proteins in which the polypeptide chains are folded, (g) t o show that in certain cases of denaturation, changes occur in the CO and NH groups which are possibly connected with changes in hydrogen bonding and changes in folding of the polypeptide chains. There are indications that each protein has a characteristic infrared spectrum and so infrared analysis may prove to be a valuable analytical tool in rapidly distinguishing similar from dissimilar proteins, and also similar proteins from one another. Infrared spectroscopy may also be a useful technique for studying the reactions of polypeptides and proteins with other chemical compounds. Further progress in the application of infrared analysis to the protein field depends on several factors, which fall into two main categories. I n the first place the techniques for obtaining the spectra of proteins can be improved. Most of the work reviewed has been done with single beam spectrometers over a very restricted range of the infrared, viz. 2-8 p (5000-1250 cm.-l), and frequently only as fa r as 3.5 p (2860 cm.-'). Spectra should be extended to a t least 20 p (500 cm.-l) and preferably to 150 p (66 cm.-l). The potentialities of the reflecting microscope have still t o be exploited in the examination of single fibers and single crystals. The use of low temperatures as a means of sharpening diffuse spectra should be investigated. Almost no work has been done in aqueous media, the natural environment for the majority of the interesting proteins. Although the strong absorption bands of water near 3 and 6 p and beyond 8 p are a great handicap here, the intervening regions can yield some information and the work of Wood (1951) has demonstrated that beyond 8 p the absorption of water is not too great to obscure the characteristic bands of muscle fibers between 8 and 12 p. Moreover Gore et al. (1949) have shown that by using DzO in place of H20 the 3 and 6 p regions can be made accessible for the study of amino acids. The use of this technique of course raises the question of the rate and extent of exchange of D with H in polypeptides immersed in DZO. The preparation of well oriented synthetic polypeptides (Elliott and Ambrose, 1951) will assist greatly in obtaining more precise information on the orientation of CO and NH bonds from measurements on infrared dichroism with polarized radiation. The second set of factors that will govern progress in this field are
314
G. B. B. M. YUTHERLAND
related to the general problem of the theoretical interpretation of the infrared spectra of large polyatomic molecules. The absorption bands which occur between 1 and 6.3 p can almost always be readily assigned to stretching vibrations of CH, NH, OH, SH, PH, C=O, C=C, C=N or N=N bonds but those which occur beyond 6.3 p may either be hydrogenic deformation frequencies of the first five of these bonds or may be skeletal vibrations. The latter will present a big problem, but by studying simple amides, which have been deuterated in known positions, it should be possible to identify the hydrogenic deformation frequencies fairly readily, while the use of heavy isotopes of nitrogen and carbon may lead to the assignment of some of the skeletal frequencies. Possibly of greater immediate importance is the problem of interpreting small shifts and certain fine structure frequently encountered in the assignable bands occurring between 1 and 6.3 p. These may be due to changes in the internal environment or in the external environment of the vibrating group. The former includes effects due to changes in the bond character (e.g., the CH stretching frequencies are generally higher in aromatic than in aliphatic hydrocarbons) and interaction of the group frequency with identical or nearly identical frequencies from other parts of the molecule. The latter includes all interaction effects arising from intermolecular forces, of which the most important is hydrogen bonding. Although much work has already been done on all these effects, our understanding of them is very incomplete and the interpretation of the spectra of large molecules is correspondingly vague on many important details. Thus although the sensitivity of infrared spectra to small interaction forces is at present a disadvantage because of the consequent uncertainties in making precise deductions, there is a good prospect that these minor changes in position and intensity of absorption bands will eventually make infrared analysis a still more powerful tool for structural research in the biological field. ACKNOWLEDGMENTS I am very much indebted to Dr. A. Elliott, Dr. M. F. Morales, and Dr. D. L. Wood for letting me have manuscript copies in advance 01 publication of some of their most recent work discussed in this paper.
NOTEADDEDIN PROOF Since the above was written, a most important series of papers have appeared (Pauling and Corey, 1951) in which an entirely different structure has been proposed for the a form of proteins such as keratin and of certain synthetic polypeptides. It appears that the X-ray data are consistent with a apiral configuration of the polypeptide chain in which there are 3.7 residues per turn of the helix. All the CO and NH groups
THE STRUCTURE OF AMINO ACIDS, POLYPEPTIDES AND PROTEINS
315
are hydrogen bonded and the directions of the CO and NH bonds are nearly parallel (within 12") to the axis of the helix. Clearly this structure is also consistent with the infrared dichroism of the NH and CO bands a t 3.0 p and 6.0 p , which indicates that these bonds are highly oriented along the fiber axis. A difficulty encountered with the folded structure ( V ) of Ambrose and Hanby (1949) is that the CO dichroic ratio in the case of poly-L-glutamic methyl ester is higher than that predicted by this model (Ambrose and Elliott, 1951) in a highly oriented specimen of this polypeptide. The helical structure overcomes this difficulty as the predicted ratio is considerably higher than that observed. This is understandable since perfect alignment of the polypeptide helices cannot be expected to occur. On the other hand, it has been pointed out by Ambrose and Elliott (1951) that the dichroic ratios for the CO and NH bands are very different for many synthetic polypeptides in the a form. This observation is not easily explained on the helical structure. Attention should also be drawn to two very interesting papers by Ambrose and Elliott (1951a and b). In the first paper, further observations are given on the spectra of silk, porcupine quill, elephant hair, gelatin, and certain synthetic polypeptides, indicating that the extended and folded configurations found in the proteins have marked spectroscopic resemblances to the corresponding configurations found in the polypeptides. It is also shown here that there is infrared evidence for a folded structure for collagen very similar to one originally proposed by Huggins (1943). In the second paper, a more detailed description is given of the work on denaturation mentioned above (Elliott, Ambrose, and Robinson, 1950). Results similar to those reported earlier for insulin have also been found for chymotrypsinogen and glyoxalase. Observations have been made on an insulin crystal and models are suggested for the structure of insulin to account for the change from the native to the denatured state in terms of the change from a seven membered ring ( V ) to a fully extended polypeptide chain. Two other papers from this laboratory should also be mentioned. Ambrose, Elliott, and Temple (1951) have studied the infrared spectrum of a single crystal of diketopeperazine and obtained excellent confirmation of the structure worked out by Corey (1938) for this molecule by X-ray methods. Of particular interest is the conclusion that the three bonds of the nitrogen atom are coplanar in this molecule. The other publication (Ambrose, Bamford, Elliott, and Hanby, 1951) concerns the spectra of silk rendered soluble in water by treatment in a concentrated solution of lithium bromide. The soluble silk appears to be in the a (folded) configuration and becomes insoluble when it (or part of it) goes over to the 0 (extended) configuration.
316
G. B. B. M. SUTHERLAND
It-should also be added that Pauling and Corey (1951) have put forward an alternative structure for the (extended) protein molecule, which appears to give better agreement with X-ray data in certain cases. The CO and NH bonds in this structure are not quite perpendicular t o the chain axis, but infrared data are as yet insufficiently precise to distinguish between this structure and the conventional one of Astbury arid Street (1931) in which the angle is exactly 90”.
REFERENCES Ambrose, E. J., Bamford, C. H., Elliott, A., and Hanby, W. E. (1951). Nature 167,264. Ambrose, E. J., and Elliott, A. (1950). Nature 166, 921. Ambrose, E.J., and Elliott, A. (1951). Proc. Roy. SOC.London A206,47. Ambrose, E. J., and Elliott, A. (1951a). Proc. Roy. SOC.London A206, 206. Ambrose, F,. J., and Elliott, A. (1951b). Proc. Roy. SOC.London A208, 75. Ambrose, E. J., Elliott, A., and Temple, R. B. (1949). Nature 163, 859. Ambrose, E. J., Elliott, A4.,and Temple, R. B. (1951). Proc. Roy. SOC.London A206, 192. Ambrose, E. J., and Hanby, W. E. (1949). Nature 163,483. Astbury, W. T. (1949). Nature 164,439. Astbury, W. T., Dalgleish, C. E., Darmon, S. E., and Sutherland, G. B. B. M. (1948). Nature 162, 596. Astbury, W. T., and Street, A. (1931). Phil. Trans. Roy. SOC.230, 75. Bamford, C. H,, Hanhy, W. E., and Happey, F. (1949a). Nature 164, 138. Bamford, C. H., Hanby, W. E., and Happey, F. (1949b). Nature 164, 751. Barer, R., Cole, A. R. H., and Thompson, H. W. (1949). Nature 163, 198. Barnes, R. B., Gore, It. C., Stafford, R. W., and Williams, V. Z. (1948). Anal. Chem. 20, 402. Blout, E. R., and Mellors, R. C. (1949). Science 110, 137. Blout, E:. R., Bird, G. R., and Grey, D. S. (1950). J . Optical SOC.Am. 40, 304. Bragg, W. L., Kendrew, J. C., and Perutz, M. F. (1950). Proc. Roy. SOC.London A203, 321. Buswell, A. M., Downing, J. It., and Rodebush, W. H. (1940a). J . Am. Chem. SOC.62, 2759. Buswell, A. M., and Gore, R. C. (1942). J . Phys. Chem. 46, 575. Buswell, A. M., Krebs, K. F., and Rodebush, W. H. (194Ob). J . Phys. Chem. 44, 1126. Buswell, A. M., Rodebush, W. H., and Roy, M. F. (1938). J. Am. Chem. SOC.60, 2444. Colthup, N. B. (1950). J . Optical SOC.Am. 40, 397. Corey, R . B. (1938). J . Am. Chem. SOC.60, 1598. Darmon, S. E. (1948). Ph. D. Dissertation, Cambridge. Darmon, S. E., and Rudall, K. (1951). Faraday SOC.Discussion 9 “Spectroscopy and Molecular Structure.” Darmon, S. E., and Sutherland, G. B. B. M. (1947). J . Am. Chem. SOC.69, 2074. Darmon, S. E., and Sutherland, G . B. B. M. (1949). Nature 164,440. Darmon, S. E., Sutherland, G. B. B. M., and Tristram, G. R. (1948). Biochem. J . 42, 508.
THE STRUCTURE OF AMINO ACIDS, POLYPEPTIDES AND PROTEINS
317
Davies, M. M., and Sutherland, G. B. B. M. (1938). J . Chem. Phys. 6, 755. Edsall, J. T. (1936). J . Chem. Phys. 4, 1. Edsall, J. T. (1937a). J . Phys. Chem. 41, 133. Edsall, J. T. (193713). J . Chem. Phys. 6, 225 and 508. Edsall, J. T. (1938). Cold Spring Harbor Symposia Quant. Biol. 6, 40. Edsall, J. T., and Scheinberg, H. (1940). J . Chem. Phys. 8, 520. Edsall, J. T. (1943). J . Am. Chem. Soc. 66, 1767. Edsall, J. T., Otvos, J. W., and Rich, A. (1950). J . Am. Chem. Soc. 72, 474. Elliott, A., and Ambrose, E. J. (1951). Faraday Society Discussion 9 “Spectroscopy and Molecular Structure.” Elliott, A., Ambrose, E. J., and Robinson, C. (1950). Nature 166, 194. Ellis, J. W., and Bath, J. (1938). J . Chem. Phys. 6,723. Goldstein, M., and Halford, R. S. (1949). J . Am. Chem. Soc. 71, 3854. Gore, R. C., Barnes, R. B., and Petersen, E. (1949). Anal. Chem. 21, 382. Gore, R. C., and Petersen, E. M. (1949). Ann. N . Y . Acad. Sci. 61,924. Heintz, E. (1935). Compt. rend. 201, 1478. Huggins, M. (1943). Chem. Revs. 32, 195. Katchalski, E. (1951). Advances in Protein Chem. 6, 123. Klotz, I. M., and Gruen, D. M. (1948). J . Phys. Colloid Chem. 62,961. Klotz, I. M., Griswold, P., and Gruen, D. M. (1949). J . Am. Chem. SOC.71, 1615. Klotz, I. M., and Griswold, P. (1949). Science 109,309. Lecomte, J., and Freymann, R. (1941). Bull. soc. chim. 8,601. Lecomte, J., Duhuisson, M., and Monnier, A. M. (1942). Arch. Internat. Physiol. 62, 408. Lenormant, H. (1948). Bull. SOC. chim. 16, 33. Lenormant, H. (1951). Faraday Society Discussion on “Spectroscopy and Molecular Structure.” Miaushima, S., Simanouti, T., Nagakura, S., Kuratani, K., Tsubio, M., Baba, H., and Fujioka, 0. (1950). J . Am. Chem. SOC.72,3490. Morales, M. F., and Cecchini, L. F. (1951). J . Cellular Comp. Physiol. 37, 107. Morales, M. F., Laki, K., Gergely, J., and Cecchini, L. P. (1951). J . Cellular Comp. Physiol. 37,477. Pauling, L., and Corey, R. B. (1951). Proc. Null. Acad. Sci. L‘.S. 37, 204-285 (7 papers). Randall, H.M. (1938). Rev. Modern Phys. 10,72. Randall, H. M., Fowler, R. G., Fuson, N., and Dangl, J. R. (1949). Infrared Determinations of Organic Structures. 1st. edition, Van Nostrand, New York. Richards, R. E., and Thompson, H. W. (1947). J. Chem. SOC.1248. Simanouti, T., and Mizushima, S. (1948). Bull. Chem. Sac. Japan 21, 1. Stair, R., and Coblentz, W. W. (1935). J . Research Natl. Bur. Standards U.S. 16, 295. Sutherland, G. B. B. M. (1951). Faraday SOC.Discussion 9 “Spectroscopy and Molecular Structure.” Thompson, H. W. (1948). J . Chem. SOC.328. Thompson, H. W., Brattain, R. R., Randall, H. M., and Rasmussen, R. S. (1949). Chapter XI11 in “TheChemistry of Penicillin,” by Clarke, H. T., Johnson, J. R., and Robinson, R. R., Princeton University Press, Princeton, New Jersey. Thompson, H. W., Nicholson, D. L., and Short, N. L. (1951). Faraday SOC.Discussion 9 “Spectroscopy and Molecular Structure.” Tsuboi, M. (1949a). Bull. Chem. SOC.Japan 22, 215.
318
G. B. B. M. SUTHERLAND
Tsuboi, M. (1949b). Bull. Chem. SOC.Japan 22, 255. Uaman, L. L., and Blout, E. R: (1950). Nature 166, 1029. Vles, F., and Heintz, E. (1935). Compt. rend. 200, 1927. Vles, F., and Heinta, E. (1937). Compt. rend. 204, 567. Williams, V. 2. (1948). Rev. Sci. Znstruments 19, 135. Williams, V. 2. (1951). Science 113, 51. Wood, D. L. (1951). Science 114,36. Woodward, R. B., and Schram, R. (1947). J . Am. Chem. SOC.69, 1551. Wright, N. (1937). J . B i d . Chem. 120, 641. Wright, N. (1939). J . B i d . Chem. 127, 137. Zahn, H. (1947). 2. Naturforsch. 26, 104.
Ultraviolet Absorption Spectra of Proteins and Amino Acids BY G. H. BEAVEN
AND
E. R. HOLIDAY
Medical Research Council Spectrographic Research Unit, London Hospital, London, England
CONTENTS Page I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320 11. Methods and Experimental Aspects. . . . . . . . . . . . 111. Absorption Constants of the Aromatic and the . . . . . . . . . . . . . . . . . 323 . . . . . . . . . . . . . . . . . 323 b. Tryptophan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325 c. Phenylalanine . . , . . .......................... 326 .......................... 326 2. Cystine, Cysteine, Me IV. The Vibrational Fine Structure of the Absorption Spectra of the Aromatic Amino Acids and of Proteins.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329 V. Low Temperature Spectra of Amino Acids and Proteins.. . . . . . . . . . . . . . . 331
2. Effect of Peptide Combination.. . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Native Proteins..
. . . . . . . . 341
1. General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. The Spectrophotometric Estimation of pK Values of Tyrosine
346 tives and Peptides.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Alkali-Labile Structures in Globular Protei . . . . . . . . . . . . . . . . . . .347 4. Time Effects of High p H on Absorption Sp f Globular Proteins.. 349 5. Additional Evidence of Tyrosine Phenolic Hydroxyl Group Binding in . . . . . . . . 351 Native Proteins. . . . . . . . . . . . . . . . . . . . . he Peptide Bond and of the PolyVIII. The Ultraviolet Absorption Spectru peptide Fabric. . . . . . . . . . . . . . . . ............................. 352 1. General . . . . . . . . . . . . . . . . . . . . . . ....................... 2. Absorption Due to the Peptide Fabric of Globular Proteins. . . . 3. Absorption Due to the Peptide Fabric of Fibrous Proteins.. . . . . 4. Comments on the Evidence for Selective Absorption by the P Fabric. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
320
G. H. B E A V E N AND E. R. HOLIDAY
Page
5. The Absorption of the Isolated Peptide Linkage. . . . . . . . . . . . . . . . . . . . 364 6. The Absorption Characteristics of Living Tissues.. . . . . . . . . . . . . . . . . . . 367 IX. Analysis of the Absorption Spectra of Proteins in Terms of Tyrosine and Tryptophan Content . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369 1. General.. ....... ............ 369 2. Method of Mixtu a h . . . . . . . . . . . . . . . . . 370 3. Errors in Estimates Introduced by Errors in Spectrophotometric Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371 4. Recommended Method for Estimation of Tyrosine and Tryptophan in Proteins ..................................................... 375 5. Comparison of Results Obtained by Spectrophotometric Analysis with those Obtained by Chemical Methods.. . . . . . . . . . . . . . . . . . . . . . . . . 377 6. Comments on the Problem of Analysis.. . . . . . . . . . . . . . . . . . . . . . . . . . . . 378 X. Other Analytical Applications, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 380 1. Serum Analysis.. .................... . . . . . . . 380 2. Analysis of Immu tates and Immune Pro . . . . . . . 382 ...................................... 382
I. INTRODUCTION The majority of proteins are colorless in solution. The few colored proteins are conjugated proteins in which the protein carrier is colorless. This transparency of protein solutions extends into the ultraviolet region of the spectrum and many proteins do not absorb radiation of longer wavelength than 2500 A. The essential protein fabric, consisting as i t does of a peptide chain in various forms, is not responsible for absorption a t longer wavelengths than this in the case of globular proteins in spite of what has been argued t o the contrary. I n the case of fibrous proteins there is some evidence that the peptide fabric is responsible for absorption in the region 2500-3000 A. Many proteins absorb selectively in the region 2500-3200 A., and i t was suggested as far back as 1883 b y Soret t ha t this absorption is due to the aromatic amino-acids present in the protein. This has been amply confirmed by later workers and with the advent of quantitative methods of spectrophotometry has been made the basis of a method of determining tyrosine and tryptophan in proteins. It should be emphasized that the striking property of proteins is their transparency, indicating a high degree of electronic saturation. The configurational stability of the protein molecule depends almost entirely on extravalerice forces and not on unsaturation which would result in high absorption in the ultraviolet. All attempts to give a n explanation of the form-rigidity of proteins on the basis of unsaturation (e.g., cyclol hypothesis) must, in spite of their ingenuity and formal simplicity, be said t o have failed (Pauling and Niemann, 1939). One must conclude th a t it is the absence of such rigidifying bonds which endows the protein with its unique characters of plasticity or flexibility while the number-
ULTRAVIOLET ABSORPTION SPECTRA
321
sequence of side chains gives it its chemical constancy. These two properties allow these molecules t o be arranged in large polymorphic masses t o form a matrix fabric of recurrent pattern in media which are essentially aqueous. Recently certain anomalies in the ultraviolet absorption spectra of aromatic amino acids brought about by their combination in proteins have given evidence which, in conjunction with evidence from other sources, may throw some light on the structure of proteins (see Section VI). Apart from this the interest and application of ultraviolet spectra of proteins are analytical. On a microscale the absorption spectrum may be the simplest and best evidence for the recognition of a protein. It is possible that, with care, i t will be the best means of obtaining a n estimate of tyrosine and tryptophan in a protein. The instability of tryptophan under the conditions required for protein hydrolysis gives weight in favor of a method such as the spectrophotometric which allows a direct determination of tryptophan t o be made (on a protein) without hydrolysis. On a submicro scale, techniques of the kind developed by Caspersson (1936 and 1940) and his school, as also those of Cole and Brackett (1940) and Thorell (1947), are essentially extensions and developments of macrospectrophotometric methods. Criticisms of such techniques must arise from doubts as to the validity of Beer-Lambert laws of absorption in small inhomogeneous objects in uncollimated beams (Commoner, 1949, Commoner and Lipkin, 1949). However these techniques when applied to certain selected objects, e.g. crystals, are free from this criticism of inhomogeneity of material and may be used in much the same manner a s the macrotechniques, with the added advantage th a t the sample may be solid and the effect of molecular orientation studied in polarized radiation. The study of the ultraviolet dichroism of substances such as proteins is still in its infancy but may be expected to yield evidence bearing on the orientation of the aromatic amino acids in the protein molecule (Butenandt et al., 1942; Stokes et al., 1950; Perutz et al., 1951; Schauenstein, 194913, 1950). The material reviewed here is principally th a t derived from the study of homogeneous absorbing systems or, more correctly, of absorbing systems in which the inhomogeneity is finer in grade by several orders than the dimensions of the exploring light beam. No account is given of loss b y scatter as this has already been covered in this series (Doty and Edsall, 1951).
11. METHODSAND EXPERIMENTAL ASPECTS Proteins are in general amenable t o examination by the usual methods of spectrophotometry. They are fairly stable to irradiation b y ultra-
322
G. H. BEAVEN AND E. R. HOLIDAY
violet light of the intensity and wavelengths used in such methods. Harris (1926) and Allen et al. (1937) have shown that prolonged irradiation of amino acids and proteins results in oxidation, especially of the aromatic amino acids. The extensive literature on the photochemistry of proteins, including the photochemical inactivation of enzymes and viruses, has been ably reviewed by McLaren (1949b). In quantitative work difficulties are met with, some common to quantitative manipulations of proteins in general and some peculiar to their spectrophotometry. Under the first heading comes that of assessing the dry weight of the protein to which intensity of absorption is t o be related. The nitrogen contents of a few well-characterized proteins have been determined and where possible reference should be made to these values, using Kjeldahl nitrogen values as intermediates. Where no nitrogen to protein dry weight ratios are available, the convention of expressing amino acid contents relative to 16.0% total N on a dry and ash-free basis (Block and Bolling, 1945) is recommended. As it is often desired to know the molar ratio of tyrosine to tryptophan it is preferable to express contents in terms of moles per unit weight of protein. Many proteins may be insoluble or only partially soluble in a particular solvent. Most proteins are soluble in dilute alkali and this solvent is fortunately desirable on other grounds for the analytical application of spectrophotometry to proteins. In some cases it may be necessary to subject the protein to a mild enzymic hydrolysis in order to solubilize it. In other cases solution in a hydrogen-bond breaking solvent such as lithium bromide or urea solutions may be resorted to. If these resources fail it may be possible to produce thin films of insoluble proteins by drying down suspensions on quartz plates (Loofbourow et al., 1949). Many protein preparations taken down to dryness redissolve to give a hazy solution. The best method of drying is from the frozen state. Solutions to be examined quantitatively should be as clear as possible. If available, a high speed angle centrifuge (11,000-19,000 rpm) provides the safest and most economic means of clearing solutions. Adsorption on filters, contamination of solutions by filters and frothing with resulting denaturation are all drawbacks of filtration methods. Where filtration is adopted it should be done under positive pressure (air pressure or centrifugal force) to avoid frothing. It is perhaps needless to state that nitrogen estimations should be made on solutions only after processing in this way. A practice to be recommended is always to measure the absorption of a protein solution well to the long wave side of the absorption band in order to provide a reference base line for the assessment of scattered radiation (Section IX, 4). Such precautions t o obviate haze
323
ULTRAVIOLET ABSORPTION SPECTRA
are only necessary where strictly quantitative measurements of specific extinction coefficient are required, or where measurement is being made of small changes in absorption with change of p H or with the addition of reagents which may alter the haziness, 111. ABSORPTIONCONSTANTSOF THE AROMATICAND &CONTAINING AMINOACIDS
THE
SULFUR-
1. Aromatic A m i n o A c i d s
There are several recorded determinations of the absorption curves of the aromatic amino-acids. Most of these were obtained with photographic methods of spectrophotometry which have been superceded by more accurate photoelectric methods. It will be shown t h a t in the spectrophotometric analysis of tyrosine and tryptophan in proteins, the photometric error is magnified in the final estimate of tyrosine and tryptophan contents. This fact is inevitably bound up with the form of the equations of mixture analysis. It is therefore important th a t the absorption constants be measured as accurately as possible. It is evident from the published data and from data privately collected th a t considerable variations exist between the values of these constants as determined in different laboratories. This applies particularly t o tryptophan. The question is discussed separately for each amino acid but it is relevant here to comment generally on the accuracy of spectrophotometric methods. Among physical methods photometry is one of the less accurate. An accuracy of one part in 1000 is difficult to achieve. The added complications of spectrophotometry introduce systematic errors the aources of which are still t o be traced (see Goldring et al., 1951). A recent statistical analysis (Gridgeman, 1951) of 75 photoelectric spectrophotometers has revealed an overall error among these instruments having a coefficient of variation of 1.6%. This is less than the error found for the photographic method (Holiday and Irwin, 1946) which was shown t o be 2.5%. Even so these errors probably compare favorably with those arising in microchemical methods in general and especially those depending on a final colorimetric comparison. a. Tyrosine. In Fig. 1 and Table I are given the absorption data of tyrosine in N/10 acid and N/10 Alkali. The p K of the phenolic hydroxyl group has been determined spectrophotometrically as 10.05-10.1 (Crammer and Neuberger, 1943). Since the ionization of the phenolic group of tyrosine, first observed by Stenstrum and Reinhardt (1925), is chiefly responsible for the change in spectrum on changing the p H of a protein solution i t is necessary th at spectrophotometric measurements be made
324
Q. H. BEAVEN AND E. R. HOLIDAY
0.2
x lo4
cngl
0.1
FIG. 1. Absorption curves of tyrosine in N/10 HCl NROH----.
-------.--and
in N/10
TABLE I Molecular Extinction Coeflcients (€,,,,,I) of the Aromatic Amino Acids Compound Xm,
Tyrosine
A.
Acid A.
Xmin
2745 2450 2230 (1940)
Tryptophan
2875 2780 2420 2180
Phenylalanine
2671 2634 2603 2575 2516 2465 2415 2360
emol
X lo-’ , , ,X
1.34 0.17 8.20 (35.0) 4.55 5.55 1.93 33.5 .092 .152 .144 .195 .154 .115 .081 .063
A.
AIkali A.
Xmin
2935 2695 2400 2880 2805 2440 2215 2676 2640 2610 2580 2522 2468 2415
I)
elnol
X 10+
2.33 1.00 11.05 4.60 5.43 1.90 34.6 .124 .160 .176 .206 ,172 .128 .084
2358
.062
ULTRAVIOLET ABSORPTION SPECTRA
325
at a pH of < 8.0 or > 12.0 to ensure virtual homogeneity of ionic species in the solution. The change of pK of the hydroxyl group brought about by combination of tyrosine in a protein is discussed in Section VII, 2. The curves in Fig. 1 do not include the short wave band observed by Smith (1929) a t 1940 A. in acid. The intensity of this band is uncertain owing to the experimental difficulty of excluding stray radiation in the spectrophotometer. Smith gave the molar extinction coefficient of this band as 3.5 X lo4.
FIQ.2. Absorption curves of tryptophan in NIIO HCL ------------ and in N/IQ NaOH I n Table VI are collected values of molar extinction coefficients, both published and privately communicated, of tyrosine in alkali over the range of spectrum which is used in the analysis of protein spectra. b. Tryptophan. In Fig. 2 and Table I are given the absorption curves of tryptophan in 0.1 N acid and alkali. Much greater discrepancies exist in the published values for the molar extinction coefficients of tryptophan as compared with those of tyrosine. The present authors have found considerable variations in the values for different commercial samples. Now that synthetic tryptophan is commercially available a t a reasonable price it is recommended that samples for spectrophotometric standards should be recrystallized two or three times from hot 70% methanol-water. Considerable loss. occurs but the product gives reproducible spectrophotometric values and also a theoretical Folin Color value, taking tyrosine as standard. Nitrogen content is of little value in assessing purity.
.-
32 6
G. H. B E A V E N A N D E. R. HOLIDAY
As in the case of tyrosine, data from the published literature and also privately communicated data have been collected in Table VI to cover the range of spectrum of value in the application of two-component analysis to the absorption curve of a protein. c. Phenylalanine. Although the spectrum of phenyfalanine is much more complex than that of tyrosine or of tryptophan, the molar extinction coefficient of phenylalanine is low. Except in the case of gelatin and polymyxin, it has not been found in a protein or naturally-occurring polypeptide free from tyrosine and tryptophan. P
I
-
;
FIG.3. Absorption curves of phenylalanine in N I l O HC1 NaOH .
7
and in N/10
It may be readily detected spectrographically in proteins, even in presence of tyrosine and tryptophan, by means such as the moving-plate method (Holiday, 1937, 1950a). A source giving a continuous spectrum is essential to show up the fine structure bands which reveal its presence. The surprising feature of the absorption spectrum of phenylalanine is that it is sensitive to acid and alkali. This indicates that an appreciable change in the vibrational levels of the benzyl chromophor occurs, due to inductive effects brought about by ionization of the carboxyl or amino group. As will be shown later, such an influence on the vibrational levels can be shown to occur when an aromatic amino acid is combined in peptide linkage (Section VI, 2). The absorption curves of phenylalanine in acid and alkali are shown in Fig. 3. 2. Cystine, Cysteine, Methionine The ultraviolet absorption of these three sulfur-containing amina
acids is of interest since that of cystine is appreciable in the 2500-3100 A,
ULTRAVIOLET ABSORPTION SPECTRA
327
region of the collective aromatic amino acid absorption. Cystine occurs in sufficiently high proportions in many proteins to make an appreciable contribution to the total absorption in this wavelength range. For this reason the calculated absorption due to the cystine content of a protein is often added to the calculated absorption due to the aromatic amino acids when a comparison is made between the observed absorption and that to be expected from independent (ie., non-spectrophotometric) estimations of the tyrosine and tryptophan contents. An example of this procedure is provided in the work of Fromageot and Schnek (1950) on lysozyme which contains 5 , 3, and 6 molecules of cystine, tyrosine, and tryptophan respectively, in the molecule of molecular weight 14,700. Dean (1949) included the absorption of cystine in his calculations of the absorption curves of the groundnut (peanut) protein arachin at neutral and alkaline pH values. There are some discrepancies in the literature concerning the absorption spectrum of cystine. The early measurements (Ley and Arends, 1932; Bednarczyk and Marchlewski, 1937) reported a poorIy-resolved falling to negligible values band with Amax. at about 2500 A., ~,,,.300, around 3000 A. and rising steeply below 2500 A. Anslow et al. (1930) found only smoothly-rising absorption, with no maximum or inflection, starting at about 3000 A. In a later paper Anslow and Foster (1932) repeated their work on cystine and reported a band with X,.2510 A., r,,,.250 or 500. They stated that the two values for emax. correspond to two “varieties” of cystine described in their earlier work which were obtained by crystallization from either strongly or weakly acid solution, and having qualitatively similar but quantitatively different absorption characteristics. These two “varieties” of cystine are not recognized by other workers (cf. Dunn and Rockland, 1947). Loofbourow and coworkers (1951) examined cystine in a rigid medium of acidified glycerol water at -196°C. and failed to detect either the maximum at A2500 A. reported by other workers, or even a shoulder, though they observed that it was the only one of several aliphatic amino acids examined by them which had appreciable absorption above 2500 A. Bruigom (1950) examined cystine a t p H 6 and pH 13, but did not observe the maximum at 2500 A., as his observations extended only to 2600 A. (at which emax. = 240). His results at pH 13 show an increase in absorption intensity of about 20% over the values a t pH 6 and corresponding wavelengths. Fromageot and Schnek (1950) reported measurements down to 2300 A. in 0.1 N acid and alkali. In acid they observed an inflection at ca. 2400 A. (E ca. 300) falling more or less smoothly to 2950 A. (at which is ca. 50). I n alkali the absorption is higher, with a shallow, but the curve falls smoothly clearly-resolved maximum at 2490 A., r,,.340; to 2900 A. (E = 100).
328
G. H. BEAVEN AND E. R. HOLIDAY
The reviewers’ data for cystine (Fig. 4) are in fair qualitative and quantitative agreement with the results of Fromageot and Schnek. Cysteine: The published data for cysteine (Anslow and Foster, 1932; Loofbourow et al., 1951) show negligible absorption above 2500 A., rising t o high values in the 2000 A. region. Fromageot and Schnek (1950)
FIG.4. Absorption curves of: 1. Cystine in N/10 NaOH 2. Cystine in N/10 HC1 3. Methionine in N/10 HC1 4. Cysteine in N/10 HC1
examined cysteine in 0.1 N acid with essentially similar results; their curve rises smoothly from 2600 A. (B = 10) t o 2300 A. (B = 60) and the authors’ results (Fig. 4) are in general agreement with their results. Methionine: The absorption of this compound is also negligible above 2500 A. and is lower than that of cysteine down t o the limit of observation (Fig. 4). It may be concluded from the results discussed above that the absorption characteristics of cystine, cysteine, and methionine are consistent with the behavior t o be expected for compounds containing the disulfide, thiol, and alkyl sulfide chromophors. I n this connection, lanthionine, which may be formed by the action of alkali cyanides or hydroxides on
ULTRAVIOLET ABSORPTION SPECTRA
329
cystine or on cystine-containing proteins, can be grouped spectrally with methionine, as it also contains the alkyl sulfide chromophor. The effect of alkali on the various sulfur-containing chromophors is of interest as proteins are frequently examined at high pH. The thiol group of cysteine is weakly acidic (pK 10.3) and will therefore exist in strongly alkaline solution as the anion, which is very easily oxidized by atmospheric oxygen; its absorption spectrum is not known with any certainty. The disulfide bond of cystine is susceptible to alkali, the primary products of fission being -SH and -S.OH (sulfinic acid) groups, which are both unstable in alkali. The sulfinic acid may decompose t o give free sulfur (or hydrogen sulfide) giving, in the first instance, an aldehyde group a t the original point of attachment of the sulfur atom to the remainder of the amino acid residue. These changes are recognized as part of the consequences of alkaline denaturation, and are probably involved in the alterations in absorption spectra which occur when some proteins are examined at pH 13 (Section VII). There is abundant evidence (see Neurath et al., 1944, for review) that in many native proteins a large proportion of the cysteine and cystine residues are not available to specific reagents for these groups, and that their presence can only be demonstrated after denaturation. The absorption contribution of the cystine residues of native proteins in neutral solution is, of course, not subject to this complication. Finally it may be pointed out that accurate data for the molecular extinction of the cystine residue in peptide combination do not seem to have been determined. The authors’ data for simple cystine peptides suggest, however, that it is very similar to that of free cystine.
FINESTRUCTURE OF THE ABSORPTION SPECTRA IV. THE VIBRATIONAL OF THE AROMATIC AMINOACIDS AND OF PROTEINS With the gradual introduction of refinements into spectrophotometric techniques it came to be recognized that the absorption spectra of the aromatic amino acids in solution all show more or less fine structure. For phenylalanine this is so marked that it is revealed by the crudest methods. In the case of tryptophan and tyrosine the fine structure is not easily detectable except under special conditions, the chief of which is the use of a source of radiation giving a continuous spectrum. With such a source and a spectrograph of moderate dispersion, the spectrogram obtained from a solution of tyrosine or tryptophan will, if the exposure has been correctly chosen, reveal the fine structure in the absorption bands of these amino acids. It is not always possible with one exposure to obtain a spectrogram showing all the fine-structure bands if these differ considerably in intensity, but a set of graded exposures will over-
330
G . H. BEAVEN AND E. R. HOLIDAY
come this difficulty. Coulter et al. (1936) and Lavin and Northrop (1935) were the first to investigate amino acids and proteins in this way. They made the observation that in the spectrogram of a protein solution the fine-structure bands of the aromatic amino acids were present but were all shifted to longer wavelengths by some 15-30 A. Later, one of the present authors (Holiday, 1937, 1950a) described the moving-plate method in which, instead of making a set of graded exposures as described above, the plate holder of a spectrograph is moved continuously a t an increasing rate to give a continuous variation of exposure of the plate. In this way there could be found at some level on the plate the optimum exposure required to show up fine-structure bands, within a range of intensity governed by the design of the mechanism moving the plate. A similar device was independently described by Brdicka and Pavlik (1930)) in which the thickness of the solution varied a t the same time as the plate was moved. These methods, for reasons which are outside the present discussion (see Holiday, 1950b, for a general treatment) are by far the most sensitive means of detecting departures from smoothness of the envelope of an absorption curve. By means of the moving-plate method the findings of Coulter, Stone, and Kabat and of Lavin and Northop have been confirmed and extended (Beaven et al., 1950). In Fig. 5 are shown the spectrograms of solutions of tyrosine and tryptophan in 0.1 N acid and of phenylalanine in 0.1 N acid and alkali, as obtained by the moving-plate method, and in Table I1 the positions of the fine structure bands are listed. It will be seen that all the spectra are complex. The error of placing of the fine structure bands by this method varies with the width of the band and the dispersion of the spectrograph. An instrument such as the Hilger Medium quartz spectrograph has a convenient dispersion and with this instrument the bands may be located with an error of k 1 A. in the case of the sharper bands and k 2-5 A. for the broader bands. If different protein solutions be examined by the moving-plate method the fine structure due to the constituent aromatic amino acids is evident on the spectrograms but, as shown by Coulter et aE. (1936) and Lavin and Northrop (1935), these fine features are displaced by 15-30 A. to longer wavelengths. I n Table I1 are listed the wavelengths of these features for a number of proteins, and some moving-plate spectrograms of proteins are shown in Fig. 6. The spectra of all three aromatic amino acids are displaced, but there are no regularities in the displacements observed for different proteins, for the different amino acids or for the individual fine-structure bands of a given amino acid. The sensitivity of detection of the individual aromatic amino acids is well illustrated bv Dhenvlalanine. which can be readilv detected in a
ULTRAVIOLET ABSORPTION SPECTRA
33 1
protein in which the number of residues of phenylalanine in one mole of protein exceeds half the sum of the number of tyrosine and tryptophan residues. At this relative concentration the intensity of absorption of
FIG.5. Moving-plate spectrograms of amino acids. a. Phenylalanine, pH 6.0 6 . Phenylalanine, pH 12.0 c. Tyrosine, pH 6.0 d. Tryptophan, pH 6.0 Wavelength scales of spectrograms are in A. X 10-2.
phenylalanine is only one-thirtieth of that of the combined absorption of tyrosine and tryptophan.
V. LOW-TEMPERATURE SPECTRAOF AMINOACIDSAND PROTEINS The effect of low temperature in sharpening the vibrational finestructure bands of the absorption spectra of both organic and inorganic compounds has been investigated b y various workers. Kronenberger
w w
TABLE I1 Wavelength ( A . ) of Vibrational Fine-Structure Bands of Amino Acids, Peptides and Proteins
N
AMINO ACIDS AND PEPTIDES Phenylalanine pH 2.0 Phenylalanine pH 12.0 Tyrosine pH 6.0 Tryptophan pH 2.0 Tryptophan pH12.0 Tyrosyl glycine Glycyl tyrosine Leucyl tyrosine Tyrosyl tyrosine Tyrosine anhydride Tryptophan ethyl ester Glycyl tryptophan pH 2.0 Glycyl tryptophan pH > 12.0 Alanyl tryptophan ethyl ester pH 2.0 Alanyl tryptophan ethyl ester pH > 12.0 Alanyl phenylalanine ethyl ester pH 2.0 Polymyxine pH 2.0
(2350-66) 2416 k 1 2466 k 1 516-7 ......... 2415 2468 522
......................... ......................... ........................ ......................... ......................... .........................
25754 2580
603 610
....... .......... ..... ....... .......... ..... .......... ..... ....... .......... ..... ....... . . . . . . . . . . ..... .......
..... .....
2635 2640
......
....... .......
......
2745
2710 2715
.......
......
1742 2747 2750 2745 2746
2673-4 2676 . . . . . . . . . . . 2670
$2715)
................ ................. . . . . . . . . . . . 2670-8 . . . . . . . . . . . 2680 . . . . . . . . . . . 2660
2706
.......
2715
.......
2804
2884
2810
2890
3
2802
2881
M
2805
2889
......
....... .......... ....... . . . . . . . . . . ....... .......... ....... ..........
..... .....
................. ................. ........... ......
.........................
. . . . . . . . . . . . . . . . . .....
. . . . . .I
......
.........................
. . . . . . . .......... .....
. . . . . . . . . . .I
2720
.........................
....... .......... .....
.................
2710
.....................
....... .......... .....
.................
2717
. . . . . . . . . 2410 2464 . . . . . . . . . . . . . . . . . 2466
I I 1 I
514 515
2574 2574
602 603
.......... .......... ..........
2818 2794 2803 2817 2816 2822 2818 2816 2790
......................... ......................... .........................
1
.......... ..........
......
...... ......
2877 2889
..........
.......... ..........
.......... .......... 2879
2634 2634
2635 2645
%it 2630
m W M
i
2 P
2671 2673
...... ......
....
.......
..........
.......... ..........
~ ~%~
2672 2680
2672
m
m E
0 U
k+t
PROTEINS AND PROTEIN DIGESTS
Gelatin (sol) ......................... 2515 2603 2577 Gelatin (dry sheet) ................................ 2618 2586 Ox serum albumin ......................... 2525 ....... 2585 Human globulin ............................... ....... 2580 Human globulin (digest) pH 2.0 .................................................... Adult human hemo. . . . . . . . . . . . . . . . . . . . . . . . . 2531 k 3 2588 k 3 2621 i- 1 globin Fetal human hemoglobin ......................... 2533 f 3 2588 f 5 2618 i 2
?
...... . 2748 ......
I2726 . .. .. .
. 2760
2820 2849 2850 2832
. 2743
2816
. 2779 2781
...........
-
........... 2900 2910 2894
2649 k 2 2687 f
2910 It 1
2649 f 3 2686 f
2898 f 1
Egg albumin
..................
Solubilized wool keratin Insulin Insulin (digest) pH 2.0 Lysoeyme Lysozyme (digest) pH 2.0 Casein Casein (digest) pH 2.0 Trypsin Chymotrypsin Pepsin
......................... ......................... ......................... .........................
. 2532 2523 2526 2517 2521
................................ ......................... 2522 ......................... 2522 ................................
.........................
2529
................................
2591 2580 2585 2576 2581
2622
2581 2586 2581 2583 2581 2590
.......
.......
.......
....... .......
2653 2641 2642 2640 2635
2691 2681 2685 2676 2689
not detected 2641 ....... 2680 2640 ....... 2674 ....... 2648 2687 2640 1612 2680 . _ .. . . . . . . . -2655
2743
2803
2854 2833 2835 2819 2823
. . . . . . . . . 2759 . . . . . . . . . 2755
. . . . . . . . .2745 2741
........
2732
....
......... 2764 . . . . . . . . . 2751
................. ................. . . . . . . . . .I . . . . . . . .
-
2816 2834 2819 2832 2827 2840
2914
.... .......... 2911 2898 2910 2896 2907 2911 2916
AMINO ACIDS A N D P E P T I D E S I N GELS Phenylalanine in gelatin
. . . . . . . . . . . . . . . . . . . . . . . . 2515 sol 2577 Phenylalanine in gelatin gel . . . . . . . . . . . . . . . . . 2473 2528 2588 Glycyl tryptophan in gelatin sol ............................................ Glycyl tryptophan in gelatin gel ............................................
.......
2635
2672
.......
.......
2650
2690
....... . . . . . . . . . . . . . . . . . . . . . . . . . .
.........................
.......
.........................
....... . . . . . . . .
2719
2802
2890
2831
2917
w w
0
334
G. H. B E AVE N AND E. R. HOLIDAY
(1930) and Arnold and Kistiakowsky (1932) showed the effect on the spectrum of benzene, and this compound was later studied quantitatively by Mayneord and Roe (1937). In the case of proteins, Hartridge as early
FIG.6.
Moving-plate spectrograms of polypeptides and proteins. a. Gramicidin in EtOH b. Gelatin sol c. Polymyxin d. Bovine serum albumin Wavelength scales of spectrograms are in A. X 10-2.
as 1920 showed th at the visible bands of hemoglobin were temperature sensitive, shifting towards the red on warming from 8 t o 40°C. Keilin and his coworkers, for work at low temperature on the visible absorption spectra of cytochromes and other hCmoproteins, have made elegant use of a microspectroscopic ocular and have reviewed this aspect of the sub-
ULTRAVIOLET ABSORPTION SPECTRA
335
ject (Keilin and Hartree, 1949). They have demonstrated the very marked sharpening of the visible band systems of these compounds at the temperature of liquid air. The application of low-temperature techniques to the investigation of protein spectra in the ultraviolet region was initiated by Lavin and Northrop (1935) who investigated the ultraviolet absorption spectra of pepsin, serum albumin, and ovalbumin in glycerol, and showed that the fine structure of the protein spectrum was enhanced a t -100°C. Preliminary reports of similar- work have been published by Randall and Brown (1949) on thin films of sublimed tryptophan and phenylalanine at 90°C., and by Sinsheimer et al. (1949) for tryptophan a t 77.6"K. Loofbourow and his coworkers (Sinsheimer et al., 1950) have begun publication of a series of papers reporting much more comprehensive work on the influence of low temperature on the spectra of amino acids and proteins in thin films and in solid solution. Beaven et al. (1950) have reported a few results on thin films of the aromatic amino acids. The aim of this type of investigation has been the better characterization of protein spectra, in terms of their constituent aromatic amino acids, by increasing the degree of resolution of the fine structure of the latter and hence making their identification and differentiation more .certain. This effect of low temperature results from the stabilization of internal Stark and Zeeman effects, which are the chief temperaturesensitive factors determining the frequency spread of absorption bands. Sinsheimer et al. (1950) give a brief review of the subject in relation t o organic compounds. There is no doubt that some very useful empirical distinctions between the visible absorption spectra of different colored proteins have already been made by means of low-temperature techniques. This applies especially to the technique of Keilin and Hartree (1949). These methods for the study of visible absorption spectra in small amounts of unpurified biological material seem to offer advantages of simplicity and rapidity. By contrast the methods used for investigating ultraviblet absorption spectra a t low temperature suffer from considerable complications and the results so far obtained on proteins do not warrant great optimism as to the value of the new information that may be gained from these methods. It appears that in complex macromolecules of the type of protein or nucleic acid the increase in resolution of the vibrational band systems of the individual chromophors is much less than might have been expected simply from the behavior of the free absorbing units under similar conditions. In the authors' opinion the information obtained by low temperature U.V. techniques corroborates but does not significantly extend that
336
G . H. BEAVEN AND E. R. HOLIDAY
obtained at room temperature by the moving-plate method. The wavelengths of the fine-structure bands of ,proteins are given b y both methods. The absence of absorption due to the peptide chain is confirmed. No fine structure is revealed in the spectrum of alkaline tyrosine a t low temperatures. On the other hand the effect of p H on the spectrum of phenylalanine, which is clearly revealed by the moving plate method, was not detected a t low temperature although more bands were evident. There is, however, a possibility that certain differences between the spectra of different proteins a t low temperatures may be of interest. It appears from the work of Loofbourow and coworkers (1951) that, for instance, insulin and serum albumin, two proteins containing tyrosine and phenylalanine but virtually no tryptophan, show great differences in spectra a t 77°K. The observation th at the tyrosine bands are well resolved in the spectrum of insulin at low temperature, but hardly at all in t ha t of serum albumin, is of interest as indicating different milieux for this amino acid in the two proteins. For deductions t o be made as t o the state of the aromatic amino acid residues in proteins, the conditions under which the amino acids and the proteins are examined must be similar. Thin-film data on crystalline material are scarcely comparable with those obtained from the study of the same compounds when combined in macromolecules in solution, although they may have intrinsic interest of their own. Similarly, caution must be observed in comparing data obtained a t low temperatures with those at room temperature where there is a possibility tha t ionic dissociation is involved. The meaning of p H a t low temperatures in solid media and the values of the relevent dissociation constants are scarcely possible of definition. Keilin and Hartree (1949) have demonstrated the decolorization of certain dyes and indicators on cooling, which they ascribe t o decrease of ionization. For instance, a red phenolphthalein solution a t p H 10.0 becomes colorless on cooling t o - 15°C. VI. FINE-STRUCTURE SHIFTSIN PROTEIN SPECTRA-STRUCTURAL IMPLICATIONS 1. General
It is well known that for aromatic chromophors generally the introduction of non-chromophoric and non-auxochromic substituents usually results in some loss of resolution of the vibrational fine structure and a shift of the absorption bands to longer wavelengths. These effects are greatest when the substituents are directly bonded to the original chromophoric structure, but also occur, though t o a lesser extent, with the progressive increase in size of the substituent group, e . g . from methyl to
ULTRAVIOLET ABSORPTION SPECTRA
337
larger alkyl groups, the so-called weighting effect. The reduction in resolution of the fine-structure bands which results from the introduction of either non-chromophoric or auxochromic substituents may be most simply considered as arising from the reduction in symmetry of the original chromophore, and this view gains some support from the fact that symmetrically-substituted compounds, e.g., hexamethylbenzene, tend t o retain in large measure the well-resolved fine structure of the parent compound. It is also well known that environmental and notably solvent variations have an appreciable effect both on the degree of resolution of the fine structure and on the wavelength of electronic absorption bands. Thus for phenol, in which the parent benzene chromophor is modified by the introduction of a hydroxyl group, the characteristic benzenoid fine structure is retained, though with reduced resolution, in non-polar hydrocarbon solvents. I n polar solvents, on the other hand, such as ethanol, the fine structure is largely suppressed and the whole band is shifted t o longer wavelengths by some 30 A. (Brode, 1939, gives other examples). The influence of polar solvents on electronic absorption spectra in solution may be considered as arising from interaction of a dipolar nature between solute and solvent molecules, and in some cases from more specific interactions, e.g., hydrogen-bonding between solvent molecules and appropriate substituent groups of the solute.
2. Efect of Peptide Combination In the case of the aromatic amino acids, peptide combination involves the carboxyl and a-amino groups which are separated from the chromophoric groupings by a saturated carbon atom. This provides a large measure of insulation to the aromatic chromophor from changes in the functional groups, but there is some recent evidence (Braude, 1950) that the insulating power of a single CH2 group is not complete. Thus, for the aromatic amino acids, changes in the state of ionization of the amino and carboxyl groups attached to the a-carbon atom cause small but quite definite changes in both the positions and intensities of the fine-structure bands (see Table IT) and also alter the degree of resolution of the finestructure bands slightly. A good example of the latter effect is found in phenylalanine which, in acid and in neutral solution, shows a very faint inflection at 2603 A . ; in alkali this feature is shifted t o 2610 A. and is very clearly resolved on moving plate spectrograms (Fig. 5, a and b). In the light of these general effects of substitution and solvent changes on electronic absorption spectra it is to be expected that simple peptide combination will have some influence on the absorption spectra of the aromatic amino acids. These changes have been studied with simple
338
G . H . BEAVEN A N D E. R. HOLIDAY
di- and tripeptides by several workers (Fruton and Lavin, 1939; Barkdoll and ROSS,1944) and both longwave shifts and loss of fine-structure resolution have been observed. I n the opinion of the authors the moving-plate method or one of its variants is t o be preferred for such measurements and some typical results obtained in this way for dipeptides of tyrosine and tryptophan are given in Table 11. Quantitative spectrograms of the same peptides determined on an automatic recording spectrophotometer a t small spectral bandwidths (of the order of 10 A.) also show that the degree of resolution of the fine-structure features is appreciably lower in the dipeptides. From Table I1 the magnitudes of the longwave shifts due to peptide combination are seen t o be of the order of 4 A. for the tyrosine inflection and maximum and 9 A. for tryptophan fine-structure band.
3. Native Proteins Comparison of these results with the data for the fine-structure band positions in proteins (Table 11) indicates that peptide combination alone is not sufficient to account for the longwave shifts found for proteins, which in some cases, e.g. pepsin, may exceed 30 A. for the tryptophan fine structure maximum. There is some evidence that the protein macromolecule itself constitutes an environment which also influences the position of the fine-structure features. The observations in this connection for proteins are discussed below, but the following results (Beaven et al., 1950) for the free amino acids and a peptide are also suggestive. 4. A m i n o Acids and Peptides Examined in Gels With these compounds a longwave shift can be obtained by examining them in a solid film; gelatin has been used as a typical example of a protein gel but polysaccharide films (dextrin, gum arabic) are also effective. Both types of film may be considered to interact with the functional groups of the amino acids and peptide by similar processes of hydrogen-bond formation involving the hydroxyl groups of the polysaccharides and the peptide linkages and side-chain polar groups of gelatin. The results for phenylalanine (Table 11) show that gelatin in the sol state has little effect on the positions of the fine-structure band, but that gel formation and more particularly the dried film causes an appreciable longwave shift. For glycyl-tryptophan, in which the position of the longwave fine-structure maximum can be measured with high accuracy (k1 A.) its position in a gelatin film is comparable with that in a native protein of high molecular weight e.g., ovalbumin (Table 11). It is well known of course that in the crystalline state the fine structure of aromatic chromophors is located a t different wavelengths than
ULTRAVIOLET ABSORPTION SPECTRA
339
for the same compounds in solution, even in non-polar solvents (see Loofbourow et al., 1951 for data on the aromatic amino acids), and X-ray diffraction studies indicate a high degree of interaction involving both van der Waals’ forces and hydrogen bonds between the molecules. The interactions between the solute and the medium in a gelatin film are clearly much less highly ordered than in the crystalline state. I n the latter the high degree of three-dimensional order results in an increased resolution of the fine structure, as a consequence of the nonrandom character of the intermolecular forces; the opposite is true for the aromatic chromophors in the gelatin film environment. Despite the poorlydefined character of the protein-protein chromophor interaction suggested above it seems to be a factor of some importance in determining the positions of the fine-structure bands not only in simple peptides but also in proteins proper. 5. Adult and Fetal Hemoglobin The protein that has been most intensively studied in this connection is human hemoglobin. The globin moiety of this conjugated protein contains (Schroeder et al., 1950) appreciable proportions of phenylalanine (7.7%) tyrosine (3.0%)and tryptophan (1.5%) and the tryptophan finestructure peak a t 2910 A. is a highly characteristic feature of movingplate spectrograms of adult human hemoglobin. It was found by Jope (1949), using the moving plate technique, that in fetal blood, the position of this peak was a t a shorter wavelength (2898 A.) than in adult blood (2910 A,). This difference between adult and foetal types was not found for sheep and rat blood, and for several other adult hemoglobins the peak was in the same position as for human adult type. It was also shown that the difference between the adult and fetal types of human hemoglobin was detectable in intact red cells by observations on thin layers of settled cells. Jope pointed out that this small but quite definite difference between human adult and fetal haemoglobins had the great advantage for cytological research that it was a property that could be studied without disturbing the hemoglobin (or even the intact cells) chemically, and suggested that it provided a possible tool for,thestudy of red cell generation in late fetal and early post-natal life. Further investigation of this subject has confirmed Jope’s prediction and provides an example of the value of fine-structure measurements in protein biochemistry. Beaven et al. (1951) found that in fetal human hemoglobin the tryptophan longwave fine-structure feature is a resolved maximum while in the adult type it is an unresolved inflection, irrespective of the spectral band width used in obtaining the absorption curves. It was also found that the position of this feature, as measured by the
340
G . H. BEAVEN AND E. R . HOLIDAY
moving plate method, moved progressively from the fetal to the adult position during the first few months of infant life. Since parallel electrophoretic studies made it possible to estimate the relative proportions of fetal and adult hemoglobins, it was possible to use the spectroscopic data to construct. a calibration curve relating the wavelength of the finestructure maximum to composition (Fig. 7). Over the composition range 0-50% adult type, the curve is relatively flat and of no practical value for estimating the proportion of adult type in early fetal samples. Over the range 50-100 % adult type, however, the experimental precision 2915 2910
22905
2
2900 2895
0 100
50 50
100 % Adult 0 % Fetal
FIQ.7. Position of tryptophan fine-structure band in mixtures of adult and fetal hemoglobin.
with which the position of the band can be measured is sufficient for estimating the proportion of adult type within lo%, and the results so obtained are in good agreement with the electrophoretic figures. By combining the two methods it was possible to show that: (a) Adult-type hemoglobin is detectable at 20 weeks fetal age. (b) At birth the proportion of adult type is 20%. (c) The progressive reduction in the proportion of fetal-type hemoglobin is almost complete 4 months after birth. This period is comparable with other estimates of the maximum life of a red blood cell (Jope et al., 1946). From the fact that in adult-type hemoglobin the fine-structure feature is an unresolved inflection while in the fetal type it is a resolved maximum, it was suggested that the differences between these two forms of human hemoglobin may arise from slightly different environmental influences of the globin molecule as a whole in the two types. I n the fetal type the majority of the tryptophan residues may be considered to
ULTRAVIOLET ABSORPTION SPECTRA
34 1
be in a uniform environment, so that the position of the maximum, for all the residues, is more or less identical. In the adult type on the other hand, it is conceivable that some of the residues are exposed to an altered environment, arising from a different configuration of part of the globin molecule. A shift of the fine-structure band in a proportion of the residues from such a cause would be expected to reduce or even abolish the resolution of the maximum by “filling in” the minimum on its shortwave side. In this connection there is considerable evidence, which has been reviewed by Jope and O’Brien (1949), that adult and fetal human hemoglobins also differ with respect to crystal form, solubility, aminoacid composition, oxygen aanity, resistance t o alkaline denaturation (see also Jonxis, 1949) and immunological behavior. Despite this wide variety of differences only electrophoretic mobility studies under conditions of large boundary anomalies (Hoch, 1950), the moving-plate spectrogram method for mixtures containing 50% or more of the adult type, and the measurement of alkaline denaturation rates are of service for quantitative studies. 6. Cooley’s Anemia Hemoglobin
Examination of a small number of blood samples from patients with various diseases involving blood abnormalities have not shown any of these conditions to be associated with a fetal-type hemoglobin absorption spectrum with the exception of Cooley’s anemia, a congenital disease which is common in certain Mediterranean countries. Liquori (1951) has found that blood from cases of this type of anemia has a slow alkali denaturation rate, comparable with that observed for fetal-type hemoglobin by Jonxis (1949), which enables it to be clearly distinguished from the normal adult type, which is denatured much more rapidly under the same conditions. Specimens of blood from an authentic Italian case of Cooley’s anemia supplied by Liquori showed the tryptophan fine-structure band a t a position (2902 f 1 A.) intermediate between those of human fetal and adult types, and corresponding to a composition of 50% adult, 50% fetal types (both f lo%), in good agreement with Liquori’s estimate on the same sample by denaturation rate measurements. Tests on samples of blood from carriers of Cooley’s trait (ie., parents with one gene) and from an alleged case of Cooley’s anemia of British origin showed no significant deviation of the tryptophan fine-structure band from the normal adult position. In view of difficulties of diagnosis, more observations are required on definitely established cases of true Cooley’s anemia. In the course of this work it has also been established that in the remarkable condition known as sickle cell anemia (Pauling et at., 1949,
342
G. H. BEAVEN AND E. R. HOLIDAY
1950), in which the hemoglobin can be crystallized in the red blood corpuscle when in the reduced state and can be detected as such by dichroism measurements (Perutz and Mitchison, 1950), the tryptophan fine-structure band is in the normal adult position. 7. Trypsin-Trypsin Inhibitor
A further example of a fine-structure shift resulting from a specific interaction between proteins is provided by the trypsin-trypsin inhibitor combination (Northrop et ad., 1948). Working with crystalline enzyme and inhibitor, both isolated from cattle pancreas, it was found (Beaven and Green, 1951) th at a t p H 7 the tryptophan band of trypsiri was shifted from 2906( +_2)A. t o 2914(+_2)A.; changing the pH to 2, which causes the enzyme-inhibitor combination t o dissociate, shifts the finestructure band t o 2907( +_2)A,, compared with 2907 A. for trypsin alone at pH 2. It does not seem necessary t o argue th a t the shift on combination indicates that the tryptophan residues of trypsin are specifically involved in the combination with inhibitor (which is itself free from tryptophan) ; the environmental effects discussed above seem t o provide a n adequate explanation if it is considered th at the inhibitor, which is presumably bound to the enzyme by a combination of hydrogen bonds, salt linkages, and van der Waals’ forces, provides an altered environment for at least a part of the trypsin molecule containing a sufficient number of tryptophan residues to make their altered absorption characteristics observable. 8. Enzymic Digestion of Proteins A final, very direct example of fine-structure shifts in relation to changes in the degree of complexity of native proteins is provided b y the effect of enzymic digestion. Here the resulting reduction in molecular size and loss of specific configuration of the native protein is accompanied by shifts t o shorter wavelengths. This effect is to be expected, as the products of proteolysis approximate to simple polypeptides and the environmental effects of the protein molecule as a whole may be expected to be greatly reduced. These shortwave shifts are clearly seen in Figs. 8 and 9 for the peptic digestion of an albumin and a globulin, and indicate that the entire absorption band is involved. Data for the fine-structure bands, obtained by the moving-plate method, are given in Table I1 for the peptic digestion of casein, 7-globulin, insulin, and lysozyme; the displacements occur for all three aromatic amino acids, and are of the order of 15-20 A. for the tyrosine and tryptophan bands and rather less, up t o 10 A., for the phenylalanirie bands. It is of interest to note th a t the 2640 A. band of phenylalanine seems to be unaffected by enzymic
ULTRAVIOLET ABSORPTION SPECTRA
343
hydrolysis. In general the fine structure of this aromatic amino acid is very little affected by combination in simple dipeptides or even in low molecular weight polypeptides (e.g. polymyxin, Table II), though it exhibits definite longwave shifts in proteins of higher molecular weight (e.g. trypsin inhibitor, insulin, serum and egg albumins). Here again
FIG.8. Serum globulin in water, pH 6.0. Before peptic digestion ---- ; after peptic digestion ------------
.
however the band near 2640 A. seems to be less susceptible than the other bands t o peptide combination shifts.
VII. THE A~SORPTION SPECTRA OF THE AROMATIC AMINOACIDSAND PROTEINS IN STRONGLY ALKALINE SOLUTION (pH 12-13) 1. General
For the reason discussed below, the spectrophotometric estimation of tryptophan and tyrosine in proteins is carried out in strongly alkaline solution, usually 0.1 N KOH. The effects of this medium on the absorption spectra of the aromatic amino acids and of cystine are therefore of interest.
344
G . H. BEAVEN AND E. R. HOLIDAY
As shown by Fig. 2, the absorption spectrum of tryptophan is only slightly affected by pH. Comparing values found a t pH 13 with those at pH 1-3, there are slight decreases in the extinction coefficients of the 2700-2900 A. band and of the minimum near 2400 A., and a slight increase in intensity of the shortwave band near 2200 A. ;these changes are accompanied by a slight shift to longer wavelengths of both absorption bands.
FIQ.9. Serum albumin. Curves 1 and 2 in acid; 3 and 4 in alkali. Curves 2 and 4 before peptic digestion; 1 and 3 after peptic digestion.
The spectra in acid and alkali also differ in the degree of resolution of the fine structure, which is less at pH 13. The other data listed in Table 1 show the same general differences, but
ULTRAVIOLET ABSORPTION SPECTRA
345
structure maximum in alkali. Neither 8-phenylalanine nor tryptophan would be expected to show large acid-alkali effects in their UV absorption spectra, as the changes in ionization with pH are confined to groups which are effectively insulated from the aromatic chromophores by two saturated C atoms:
For tyrosine the absorption spectrum is not very sensitive to pH changes over the range 1-8. There is a slight longwave shift with increasing pH and a slight fall in the intensity of the band (Fromageot and Schnek, 1950). From pH 8 upwards, the absorption alters progressively, the change being complete at pH 12. Quantitatively the changes are completely consistent with an ionization process associated with a pK of ca. 10, and are identified with the ionization of the phenolic hydroxyl group.
(1)
(11)
(111)
The differences between the absorption spectra of un-ionized phenols (I) and their anions (11) were observed in very early studies of simple phenols by Baly and Ewbank (1905) and Ley (1920), but there are few systematic investigations of the effect in the more recent literature. An excellent example in the biochemical field is due to Callow (1936) who found that the phenolic hydroxyl groups in ring A of estrone, estradiol and estriol showed the same changes in alkali as simple phenols. The longwave shifts of both absorption bands and the increase in intensity can be explained by the increased number of low-energy resonance structures possible for the anion (111) compared with the neutral molecule and the consequent increase in stability of its first excited state (Pauling, 1942). The alterations in the absorption spectra of proteins a t high pH values were first ascribed to ionization of the tyrosine residue by Stenstrom and Reinhard (1925). However this explanation seems t o have been accepted with some reserve, as the arguments for it have been frequently restated, most recently by Sizer and Peacock (1947) who give references
346
G. H. B E A V E N AND E. R. HOLIDAY
t o earlier papers on the subject (see also Beaven et al., 1950). Ultraviolet spectrophotometry is now firmly established as a method for the determination of the pK values of phenolic hydroxyl groups, including tyrosine and its derivatives, tyrosine peptides, and, in principle at least, the tyrosine residues in proteins. I n this respect it now appears (see following sections) that information may be obtained on specific forms of bonding of the tyrosine residues in native proteins. From the data in Fig. 1 it can be seen that in the completely ionized form (i.e. at pH 12 or higher) the absorption maximum of tyrosine shifts to 2930 A. and the molar extinction coefficient increases by some 70%. This increase in intensity reduces the large difference between the peak absorption intensities of un-ionized tyrosine and tryptophan, and thereby increases the precision of the two-component analysis (Holiday, 1936). 2. T h e Spectrophotometric Estimation of pK Values of Tyrosine
Derivatives and Peptides Some values which have been obtained by this method are listed in Table 111 and are of considerable interest to the present discussion. For tyrosine itself the values obtained by Crammer and Neuberger (1943) and Fromageot and Schnek (1950) are in good agreement, after TABLE I11 Spectrophotometric Values of p K for the Phenolic Hydroxyl Group of Tyrosine and Some Derivatives and Peptides Compound TyrosineO Tyrosined L-Tyrosinee DbTyrosine' 3 :5-Diiodotyrosinee DLTyrosine ethyl ester" Tyramine' Chloroacetyl-L-t yrosinea Chloroacetyl-L-tyrosine ethyl esterd Gly cyl-L-tyrosinee L-Glutaminyl-btyrosinee Tyrosyl cysteined Leucyl tyroshed Lysozymed Insulinc pK' Observed. pKo Corrected to zero ionic strength. 0 Crammer and Neuberger (1943). d Frornageot and Schnek (1950). e Bruigom (1950). 0
b
pK'a 10.05 10.1 10.04 10.04 6.42 9.98 9.77 10.23 10.06 9.99 10.22
10.5 10.2 10.8 10.G-11.8
pKob
APK
10.15 10.15 10.15 10.09 9.88 10.34 10.17 10.07 10.33
-3.63 -0.06 -0.27 +0.19 +o .02 -0.08 +0.18
-
(+0.4)
-
(+O.l) (+0.7)
-
ULTRAVIOLET ABSORPTION SPECTRA
347
correction to zero ionic strength by the simplified Debye equation, which has also been applied by Bruigom (1950) to his spectrophotometric values. From these figures the effect of peptide combination alone is seen to increase the pK significantly; for lysozyme ApK is +0.7 but it may be seen that the electrostatic effects of other charged groups in the molecule are the most important factor in determining the tyrosine pK in a protein (see also Tanford (1950) for a detailed discussion and references). For tyrosine ethyl ester ApK is small (+0.06), similarly for glycyl tyrosine ( - O . O S ) , but the effects of the ionizable groups show up in glutamyltyrosine and more so in tyrosyl-cysteine. The effect of substitution of iodine in the phenyl nucleus is noteworthy; spectrophotometric data for iodinated human serum albumin are given by Hughes and Straessle (1950). On the basis of the pK value for lysozyme, it might be anticipated that for proteins, values of the same order would be found, leading to complete ionization of the tyrosine residues at about pH 12. The evidence that this is not the case will be discussed below. 3. Alkali-Labile Structures in Globular Proteins Working with ovalbumin Crammer and Neuberger (1943) found that its absorption spectrum did not alter greatly over the pH range 9.5-12, but was similar to that at pH 4.9. Calculations of the expected molecular extinction coefficients at selected wavelengths were in fair agreement with the observed figures, using the chemically determined tyrosine and tryptophan contents (Chibnall, 1942) and assuming that no phenolic hydroxyl groups were ionized at pH 9.5. At pH 13 the spectrum changed very markedly to the type associated with ionization of the tyrosine hydroxyl groups and the calculated extinction coefficients agreed with the observed figures if this ionization process was assumed t o be complete. If, after exposure of the ovalbumin to pH 13, the solution was then adjusted to pH 12, the spectrum differed from that of the same protein which had been exposed only to pH 12. The difference corresponded to a higher degree of ionization of the tyrosine hydroxyl groups in the former case. It was shown that this irreversible effect of previous exposure to alkali at pH 13 on the spectrum of the protein at pH 12 resulted from denaturation. The same altered spectrum a t p H 12 could be produced by acid, heat, and urea denaturation treatment at much lower pH values, followed by adjustment to pH 12. Conversely, exposure of the protein to pH 12 for 3 days at room temperature produced no denaturation (based on precipitate formation at pH 5 and positive nitroprusside reaction) and no increase in absorption; a t pH 13 denaturation appeared to be instantaneous (but see below). For denatured
348
G. H. BEAVEN AND E. R . HOLIDAY
ovalbumin Crammer and Neuberger estimated the pK of the phenolic hydroxyl groups to lie between 11.4 and 11.8, which was comparable with the values found by them for insulin. Of the total of ten tyrosine residues in the ovalbumin molecule (M 43,000) it was estimated that only about two were ionized in the native protein at pH 12, and the remaining eight only after denaturation. The hypothesis that in the native protein some of the phenolic hydroxyl groups have pK values greater than 12.5, due to the specific configuration of the molecule, was rejected, on the ground that it implies a highly asymmetrical charge distribution. It was also considered unlikely that the interior of the native protein molecule is inaccessible to the hydroxyl ion. The explanation favored by Crammer and Neuberger was that those tyrosine hydroxyl groups not free to ionize at pH 12 in the native protein are bound in some form of linkage which is broken by denaturation. They further suggested that the bonding was hydrogen-bonding to appropriate groups, such as COO-, in adjacent parts of the folded native protein molecule, and that this hydrogen-bonding assisted in stabilizing the configuration of the native protein. If this explanation is accepted-and it seems to be consistent with present views on the agencies involved in maintaining the specific configurations of native proteins and with existing knowledge of the denaturation process (Neurath et al., 1944)-it follows that an irreversible change in the absorption spectrum of a protein at pH 12 due t o previous exposure to pH 13 can be used to demonstrate the existence of hydrogen bonding of tyrosine hydroxyl groups. In this connection it may be noted that Schauenstein (1949b) detected appreciable dichroism in the selective tyrosine absorption in oriented silk fibroin films, indicating a considerable degree of orientation of the planes of the phenyl rings of the tyrosine residues, relative to the axis of the peptide chains; this finding is not inconsistent with the involvement of tyrosine hydroxyl groups with appropriate groups in adjacent parts of the molecule in hydrogenbonding. In studies of the recombination of heme with globin Jope, Jope and O’Brien (1949) have used the concept of bound tyrosine hydroxyl groups as a criterion of native character in globin, based on the spectrophotometric study of the alkaline ionization process. They were able to show that globin preparations which were native according to this criterion could be classified as denatured on the basis of the position of their tryptophan fine-structure bands when recombined with heme. These results led them to suggest that the process of denaturation could be separated into several stages, even by spectroscopic techniques alone (see also Jope, 1949).
ULTRAVIOLET ABSORPTION SPECTRA
349
The effects of alkali up to 0.1 N on the absorption spectrum of lysozyme are fully reversible, according to Fromageot and Schnek (1950); it is of interest to note that this enzyme is a protein which is remarkably resistant to denaturation (Fraenkel-Conrat, 1950). Insulin behaves similarly, and it may be surmised that in both these proteins the tyrosine hydroxyl groups are not bound. The pK value of 10.8 found by Fromageot and Schnek (1950) for lysozyme supports this view. The pK data for insulin are more complicated and have been discussed at length by Crammer and Neuberger (1943). For human serum albumin Tanford (1950) found by spectrophotometry that the ionization of the tyrosine hydroxyl groups was completely reversible up to pH 12. Measurements at the wavelength of the tyrosine anion maximum (2930 A.), uncorrected for the small tryptophan contribution, gave a pK of 11.7 for this process. Both the ultraviolet absorption and titration data for this protein could be quantitatively interpreted on the basis of complete freedom of all the 18 tyrosine hydroxyl groups in the molecule to ionize. In this respect human serum albumin thus resembles insulin and not ovalbumin. The only quantitative spectrophotometric study of the bonding of tyrosine hydroxyl groups that has been reported to date is the original work by Crammer and Neuberger (1943) on ovalbumin. Some evidence for the existence of this phenomenon in wool keratin solubilized by peracetic acid oxidation (Alexander et al., 1950; Alexander and Earland, 1950) has recently been obtained (Beaven, 1951). The time effects shown by many globular proteins in strongly alkaline solution (see following section) indicate that many other examples may be found. Experimental difficulties arise in the spectrophotometry of strongly alkaline solutions and the control and measurement of pH in the range 12-14 is not easy. A further complication is that the liberation and ionization of bound tyrosine hydroxyl groups is only a part of the more complex process of alkaline denaturation, which may be associated with changes in the absorption spectrum of proteins. Oxidation of proteins may also occur in strongly alkaline solution and gives rise t o marked qualitative changes in their absorption spectra (Fromageot and Schnek, 1950; Beaven and Holiday, 1950; Giese and McLean, 1950). 4. Time Effects of High p H on Absorption Spectra of Globular Proteins Holiday (1936) did not observe any time effects during the spectrophotometry of globular proteins in 0.1 N alkali, though the Spekker photometer used in this work is admittedly not well suited for the measurement of rapid changes in optical density at selected wavelengths. Such
350
G. H. BEAVEN A N D E. R. HOLIDAY
measurements can be carried out, however, with modern photoelectric spectrophotometers. Goodwin and Morton (1946), using the Beckman instrument, did not report any changes in optical density of their test solutions with time, but such effects have now been detected by several workers. Finkelstein and McLaren (1949) found that some of the tyrosine residues in chymotrypsin were not ionized until the pH had been raised to 13 (cf. preceding section) and that a t this pH the absorption was changing significantly during measurement. They associated this time effect with the “finite rate of denaturation-inactivation.” Havinga (1950) disclosed that a similar effect has been observed by Bruigom (1950) for proteins, though it was not discussed in his thesis. Shugar (1951) found that for ribonuclease in 0.1 N alkali a t least 30 minutes were required for the absorption spectrum to attain its final value. It was suggested that the phenolic hydroxyl groups were not free to ionize in the intact molecule. Time effects have also been observed for bovine plasma albumin, horse serum albumin and rabbit r-globulin (Beaven and Holiday, 1950). The behavior of bovine plasma albumin was most striking; the intensity of the absorption in 0.1 N alkali increased steadily over a period of ca. 3 hours a t room temperature; the light-scattering properties of the solution, as shown by its apparent absorption on the long-wave side of the absorption band proper, also increased. Of the few proteins which were examined, only lysozyme showed no time effect. With trypsin the change in absorption was small and complete in a few minutes a t pH 13, but the solution then became visibly turbid. In the reviewers’ opinion, these time effects are due to the breakdown of hydrogen bond structures involving tyrosine hydroxyl groups as envisaged by Crammer and Neuberger (1943), on which are superposed other more drastic irreversible changes associated with alkaline denaturation. This process involves, among many other reactions, the opening of S-S bonds which may easily be detected in, e.g., bovine plasma albumin a t pH 13. Apart from the intrinsic interest that the phenomenon of bound tyrosine residues has in relation to the stabilization of the configuration of native proteins, further work will be required to ascertain if the time effects in strongly alkaline solution are of common occurrence and of sufficient magnitude to invalidate the spectrophotometric estimation of tyrosine and tryptophan in native proteins. The results listed in Table VII suggest that under the conditions recommended for this estimation, the possible complications that might arise from the presence of bound tyrosine residues do not in fact result in serious error.
ULTRAVIOLET ABSORPTION SPECTRA
351
5 . Additional Evidence of Tyrosine Phenolic Hydroxyl Group Binding in Native Proteins The spectrophotometric evidence reviewed above for the binding of a proportion of the phenolic hydroxyl groups of the tyrosine residues of native proteins is supported by work on the action of tyrosinase on proteins. Sizer (1946) found that this enzyme oxidizes the tyrosine residues in native trypsin, pepsin, chymotrypsin, casein, peptone, insulin, and hemoglobin. Native ovalbumin, human and bovine serum albumin, tobacco mosaic virus (nucleoprotein), human y- and bovine @-globulins, and bovine fibrinogen are not susceptible to tyrosinase, but become so after tryptic digestion. It was shown (Sizer, 1947) that for the proteins which are oxidized by tyrosinase in the native state, the observed reaction does indeed occur with the intact proteins and does not require preliminary degradation to tyrosine peptides or free tyrosine. The kinetics of the oxidation of tyrosine by tyrosinase have been studied spectrophotometrically (Mason, 1948 etc.). Further studies on proteins have been carried out by Haas et al. (1951a) using lom-temperature techniques to improve the resolution of the fine-structure. These methods made it possible to recognize the absorption maximum of dopa (3 :4-dihydroxyphenylalanine)a t 2820 A. which appears in the spectrum as a progressive enhancement of the tyrosine fine-structure maximum at the same wavelength; dopa is apparently a characteristic intermediate in the oxidation of combined tyrosine or tyrosine derivatives, though not of free tyrosine. Serum albumin was found to be completely resistant to tyrosinase, insulin was only slightly attacked and pepsin was oxidized, though probably only after denaturation at pH 7. Partial hydrolyzates of pepsin and insulin were oxidized relatively rapidly. Of the tyrosine derivatives studied, methoxytyrosine was resistant and the ethyl ester, N-formyl, N-formyl ethyl ester derivatives and glycyl-tyrosine were oxidized, though a t differing rates. It was suggested that the reduced reactivity of the tyrosine residues in proteins towards tyrosine is “related to the fact that they do not have free phenolic tyrosine hydroxyl groups,” together with additional effects due to the a-amino and carboxyl groups being involved in peptide linkage. It appears from these results that the reactivity of tyrosine residues in proteins towards tyrosinase is not entirely parallel to the indications of the spectrophotometric studies of alkaline ionization discussed above. The study of tyrosinase activity has the great advantage, however, that proteins may be studied a t low pH (c.g. 6-7) where the complications of
352
G . H. BEAVEN AND E. R . HOLIDAY
alkaline denaturation are absent. For this reason alone, the approach of Haas, Sizer, and Loofbourow clearly merits further investigation. Supplementing their work on tyrosinase the same authors (1951b) have also studied spectrophotometrically the acid permanganate oxidation of the aromatic amino acids and of proteins. They found phenylalanine t o be relatively resistant t o this reagent while tyrosine and tryptophan were oxidized, leading to changes in fine-structure which could also be obtained in pepsin and insulin. Evidence was obtained for the formation of dopa groups in these two proteins by permanganate oxidation. Supporting evidence of a non-spectroscopic character on the binding of tyrosine phenolic hydroxyl groups in native proteins comes from titration studies, some of which may be briefly noted here. For example, Brand and coworkers (1942, 1945), who have reported the complete amino-acid composition of p-lactoglobulin, regard all the tyrosine hydroxyl groups of the native protein as being bound, on the basis of titration data (Cannan, Palmer and Kibrick, 1942), and ascribe this behavior t o intramolecular hydrogen bonding. They have used their analytical findings to calculate the numbers of polar ionic (i.e. titratable), polar non-ionic (i.e. non-titratable) and non-polar side-chain groups in this protein, and discuss the role which such polar groups may play in maintaining the configuration of native proteins and in the binding of water.
VIII. THEULTRAVIOLET ABSORPTION SPECTRUM OF THE PEPTIDE BONDAND OF THE POLYPEPTIDE FABRIC 1. General Two groups of workers have presented evidence in support of their views that the absorption of proteins in the wavelength range 25003000 A. is not due exclusively to the three aromatic amino acids and cystine, but t hat i t includes a contribution arising from the selective absorption of the peptide bonds, a t least t o the extent th a t these are involved in interchain hydrogen-bonded systems extending throughout the “fabric” or “polypeptide grid” of the entire molecule. The two alternative suggestions are supported by rather different lines of evidence and will be separately discussed; they are clearly of the greatest importance t o the subject of the present review, as their acceptance would necessitate a considerable revision of previous interpretations of the ultraviolet absorption spectra of proteins. 2. Absorption Due to the Peptide Fabric of Globular Proteins
Anslow and Nassar (1941) suggested tha t the 2800 A. absorption maximum of proteins was due primarily t o the peptide linkage itself, and
ULTRAVIOLET ABSORPTION SPECTRA
353
only secondarily to the aromatic chromophors in the side-chain groups of the three aromatic amino acids. Their measurements on proteins were interpreted on the basis of earlier studies of simple aliphatic compounds, including aminocarboxylic acids and substances containing peptide bonds, such as glutathione (GSH). Thus Anslow and Lyman (1941) found four absorption maxima in GSI-I which were ascribed t o the following groupings: 2510 A. S-S (formed from GSH by irradiation during measurement and recombination of GS radicals). 2800 A. CO-NH peptide bond. 3250 A. S-H of GSH. 5000 A. The glutamic acid residue of GSH. These assignments of the 2500 and 5000 A. bands were in turn based on measurements by Anslow and Foster (1932) on glutamic acid and cystine. The results obtained by Anslow and coworkers for cystine and cysteine have been reviewed above (Section 111, 2) and need not be considered further in the present connection. For alanine and butyric acid no selective absorption was found, the molar extinction coefficients rising smoothly from negligible values around 2500 A. t o ca. 2000 in the 20002200 A. region. For aspartic, glutamic, and succinic acids, however, bands were observed with maxima at ca. 5000 A. and associated molar extinction coefficients of 400-1250. For glycine, Anslow et al. (1933) found the absorption t o be similar t o th at of alanine, with a small shift t o longer wavelengths in alkali. I n these and other papers by Anslow and coworkers the frequency a t which absorption is “complete,” is used t o calculate the energy required for the presumed related dissociation process ( p . g . of a proton from a carboxyl group, fission of the S-S bond in cystine), using an equation relating the mass of the ion produced to the frequency of “complete” absorption (see also Anslow, 1932a, b). As the frequency a t which complete dissociation commences has been obtained with complete disregard for the well-established selective absorption of saturated carboxylic acids and aminocarboxylic acids in the vacuum ultraviolet below 2000 A. (see Platt and Klevens, 1944, for a review of the available data), and the more easily accessible short wave bands of the aromatic amino acids (see Section 111, l ) ,and as the theoretical basis of the equation concerned is open t o serious criticism, this particular aspect of the work will not be considered further. I n the study of proteins, Anslow and coworkers have utilized their earlier work on glutathione to support the view th a t the peptide band in proteins contributes to the absorption in the 2800 A. region, Thus for
354
G . H. BE.IVEN
AND E . R. HOLIDA4Y
ovalbumin at pH 4.8 (Anslow and Nassar, 1941), fine-structure bands at 2890 and 2790 A. were ascribed to tryptophan, at 2804,2755, and 2680 A. (inflection) to tyrosine, a group of six bands on the shortwave side of the main maximum to phenylalanine, and two a t 2740 and 2840 A. t o the peptide linkage. Anslow and Lyman (1941) had previously reported the 2800 A. peptide band of glutathione to be double, with a separation of ca. 100 A. in water, ethanol, and alkali. The results for gelatin at pH 4.5 were discussed on the assumption (based on analytical data listed by Schmidt, 1938), that this protein contains no tryptophan or tyrosine. The phenylalanine fine-structure bands were observed and identified as such in agreement with many other workers (Coulter et al., 1936; Beaven et al., 1950; etc.). Two other fine-structure bands were assigned t o the peptide linkage in the assumed absence of tyrosine. The same peptide bands were also identified in strongly alkaline solutions of gelatin, ovalbumin and lactalbumin, and further evidence of their occurrence was adduced from previous work of Fruton and Lavin (1939) on tyrosine peptides and of Lavin et al. (1937, 1939) on papain and tobacco mosaic virus. In these materials the peptide bands were distinguished from the aromatic amino acid bands, as in the interpretation cited above for ovalbumin. The work of Allen et al. (1937) was quoted as showing peptide absorption in simple anilides, in which a band a t 2800 A. was regarded as originating not in the aromatic chromophor but in the CO-NH group. The objections to Anslow’s view that the work reviewed above constitutes a demonstration of the selective absorption of the peptide fabric are given in some detail below. It may be pointed out here that the discovery of a band at ca. 5000 A. in saturated carboxylic acids is in complete contradiction to the accepted absorption properties of such compounds (see e.g. Platt and Klevens, 1944; Braude, 1945; for reviews). The simple carboxyl group is regarded as a chromophor with its first absorption band occurring outside the usual range of photographic or photoelectric spectrophotometry, i.e. in the 2000 A. region. The claim to have observed bands at ca. 5000 A. in colorless materials containing no chromophoric groups exhibiting absorption in the visible and near ultraviolet is quite unacceptable. It seems clear that these particular results of Anslow and coworkers are vitiated by the use of impure compounds and faulty technique. It may be pointed out that more recent studies from the same laboratory (Anslow, 1950) on a variety of simple amides and alcohols also report absorption features which are totally a t variance with the established absorption characteristics of such compounds, and which have been criticized (Holiday, 1950b; Ketelaar, 1950; Davies, 1950) on the same grounds as the work on carboxylic acids.
ULTRAVIOLET ABSORPTION SPECTRA
355
These criticisms do not apply to the results for alanine and butyric acids (Anslow and Foster, 1932). Finally it is of interest t o note th at the alleged absorption of the protein polypeptide fabric a t ca. 2800 A. has been interpreted b y Anslow (1945) a s arising from structures of the type \ / HO-C-N
/
RCH \
'CHR
/
/N-CToH and hence as providing support for the cyclol theory of globular protein structure proposed by Wrinch (1937).
3. Absorption Due to the Peptide Fabric of Fibrous Proteins Schauenstein and his coworkers have been particularly concerned with the structural implications of the ultraviolet absorption characteristics of oriented fibrous proteins, especially silk fibroin and the various muscle proteins. I n a preliminary note on oriented silk fibroin Fix1 et al. (1949a) reported that the characteristic tyrosine absorption is obtained if the (electric) vector of plane-polarized U.V. radiation is parallel t o the direction of the chain axis of the fibroin, and th a t if the vector makes an angle of 25-65" with the chain axis, additional absorption is observed which shows a maximum near 2470 A. In a second note (194913) the same authors studied thin slices (40 p ) of collagen from Achilles tendon by the same technique, after swelling in p H 9.4 phosphate buffer. For this material the absorption was greatest when the electric vector was parallel t o the fiber axis. At 45" t o the axis the anisotropy was negligible. Since collagen contains only traces of tyrosine (see also Loofbourow et al., 1949, but compare Block and Bolling, 1945), they postulated the existence of an enolizable bond chromophor associated with interchain linkages : H....
\
\
denotes H-bonding to adjacent chains
I n a third note Schauenstein and Stanke (1949) correlated the absorption in the 2500 A. region of tendon collagen with the dimensional changes on stretching and contraction. It was assumed th a t a t acid p H the native protein contained no peptide linkages in the enol form and th a t the enolization process was complete in strong alkali. From the absorption a t 2500 A., after correction for Tyndall scattering, the proportion of
356
G. H. BEAVEN AND E. R. HOLIDAY
enolized peptide linkages in water-swollen tendon was estimated a t 19%, and 14% for cold-stretched material; the intensity of the 2500 A. absorption was thus related to the degree of orientation of the protein, a t least with respect to the formation of intermolecular hydrogen bonds. These observations and their interpretation were discussed a t greater length by Schauenstein (1949a), and glycyltyrosyl anhydride was used as a model compound t o demonstrate the selective absorption of the enolized peptide linkage. It was also shown th at actomyosin exhibited additional absorption in the 2500 A. region in excess of th a t required by its tyrosine content, especially a t p H 11-13, and th at the magnitude of this absorption was increased by orienting the actomyosin film through stretching. These effects were further investigated b y measurements with polarized ultraviolet radiation (Schauenstein, 1949b) for films of silk fibroin. In this paper, in which all the preceding work was summarized, it was stated tha t the 2500 A. “additional absorption” of the C=N enolized peptide linkage or ‘(peptenol ” chromophor had been observed in stretched-rolled oriented silk fibroin, collagen, actin, gelatin and tobacco mosaic virus nucleoprotein (Butenandt et al., 1942). The peptenol chromophor was envisaged as resulting from the formation of pseudo-diketopiperazine structures by H-bonding, such as those postulated t o account for the infrared absorption spectra of simple H-bonded N-alkyl amide dimers: R
From the work with polarized ultraviolet radiation it was concluded th a t the additional peptenol absorption was due to a chromophor containing ?r electron bonds and the ultraviolet dichroism of crystalline p-benzoquinone was studied as a model compound on the assumption th a t the C=O groups of this compound were analogous chromophors to the C=N structures postulated in proteins. From the ultraviolet dichroism a t 2500 A. of oriented silk fibroin films i t was inferred that the axes of the C=N peptenol chromophors were inclined a t angles of ca. 45” t o the stretching axis of the films, which was also taken t o be the direction of the extended polypeptide chains. For actomyosin the additional absorption mas largely eliminated by treatment with proteolytic enzymes, which was taken as further evidence of its origin being in the peptide linkage. It was considered th a t the available evidence was not sufficient t o decide if the formation of the
ULTRAVIOLET ABSORPTION SPECTRA
357
peptenol chromophor was confined solely t o aromatic amino acid residues (which also showed ultraviolet dichroism in silk fibroin films th a t exhibited peptenol absorption), or if saturated amino acid residues also behaved in a similar manner. I n this paper, the importance of applying corrections for Tyndall scattering in the spectrophotometry of proteins was emphasized, and methods for obtaining both empirical and exact corrections were outlined, the latter involving the experimental evaluation of the wavelength exponent n in a n expression of the form
in which n is not necessarily 4, as i t is in Rayleigh’s scattering equation. This exponent was discussed in more detail by Treiber and Schauenstein (1949), who determined it for several scattering systems including proteins (compare the review by Doty and Edsall, 1951). Schauenstein and Treiber (1950) discussed the ultraviolet absorption spectrum of rabbit actomyosin in more detail, and correlated the increase in the absorption contribution of the peptenol chromophor with stretching of the sample in terms of increased enolization or ‘‘ alkalization” of the peptide linkage due t o the formation of interchain hydrogen bonds. From the pH-dependence of the 2500 A. “additional” absorption, the change in absorption for 5-10% extension was estimated t o be equivalent t o an increase in p H of about 1.5 units, and this degree of extension was regarded as transforming the proteins into a n energeticallyfavored quasi-native state, which could be regarded as the initial state for greater extensions. Schauenstein et al. (1950) correlated the spectroscopic work on silk fibroin films with X-ray diffraction studies; it was concluded th a t alkali treatment increased both the enolization of the peptide linkages t o give the C=N peptenol chromophor, and th at part of the X-ray diffraction pattern which was due t o structural order in the plane of the peptide grid. The necessary condition for the development of the peptenol chromophor was stated to be the formation of extended systems of interchain hydrogen bonds between @-typechains. The general properties of the peptenol chromophor absorption band of fibrous proteins have recently been restated b y Schauenstein (1950) in the following general terms: (a) It has a very flat contour with a “smeared-out” maximum a t CU. 2500 A. and a n intensity a t this wavelength comparable with th a t of proteins containing aromatic amino acids. It does not interfere with the spectrophotometric estimation of tyrosine and tryptophan, which are based on measurements a t about 2850 A. and longer wavelengths.
358
G . H. BEAVEN AND E. R. HOLIDAY
(b) The band disappears if the proteins are subjected to alkaline or enzymic hydrolysis. (c) The intensity of the band is reversibly pH-dependent; i t is observable a t low p H values but increases with increasing pH. (d) The intensity of the band increases on stretching a protein specimen and decreases on contraction. This effect is reversible (e.g. collagen) or irreversible (silk fibroin, actomyosin), according as the deformation is reversible or not. (e) Measurements with polarized ultraviolet radiation on wellcrystallized silk fibroin specimens show the peptenol chromopher t o be a t an angle of ca. 40” t o the fiber axis. I n collagen the main axis of absorption is in part parallel to the fiber axis and in part a t an angle of about 50“ t o it. (f) The band is associated with the existence of interchain hydrogen bonds. Cleavage of these bonds leads to a decrease in absorption, arid the pH-dependence disappears t o the same extent. It is concluded from these observations th at the results indicate the enolization of the peptide groups t o form the C=N chromophor. It is considered that this structure is not necessarily present in all proteins, but that it can be formed i f extended systems of hydrogen bonds render the enolization possible (reviewers’ italics). 4. Comments o n the Evidence for Selective Absorption by the Peptide Fabric
The evidence presented by Anslow and Schauenstein and their respective co-workers in support of the view th at selective absorption of the peptide fabric makes a significant contribution to the absorption of proteins in the 2500-3000 A. region must be taken into consideration in any discussion of the analytical spectrophotometry of proteins. Furthermore because of the structural implications which arise from the suggestion t hat peptide linkages exist, a t least in part, in the enol form as the so-called “peptenol” chromophor (OH)C=N it seems necessary t o state as explicitly as possible the available evidence. This, in the reviewers’ opinion, is contrary to the concept of peptide fabric absorption. Some of the evidence has been indicated very briefly elsewhere (Beaven and Holiday, 1950). (1) Several proteins can be cited (see Table VII) in which the agreement between the spectrophotometric figures for tyrosine and tryptophan and the analytical data obtained by other methods is sufficiently close t o justify the conclusion that in these proteins there cannot be any significant contribution t o the absorption spectrum b y the peptide fabric, in the 2700-3100 A. region. Any such contribution would have seriously affected the accuracy of the spectrophotometric analysis.
ULTRAVIOLET ABSORPTION SPECTRA
359
(2) A protein which contains no aromatic amino acids should still exhibit the selective peptide fabric absorption a s a band with a maximum a t ca. 2800 A. (Anslow and Lyman, 1941) or ca. 2500 A. (Schauenstein, 1950), free from interference by the intense bands of tyrosine and tryptophan. Clupeine, a polypeptide of molecular weight ca. 4000, satisfies this requirement. Its absorption spectrum (McLaren, 1949) in concentrated solution a t p H values of 7, 10.5 and 12 shows no selective absorption from 3000 down t o 2600 A., a t which point the curves rise steeply a t the longwave side of the absorption bands of the saturated amino-acid residues in the far ultraviolet. The curves fall smoothly a t wavelengths longer than 3000 A. and are practically flat over the range 2600-2800 A. The observed non-selective absorption corresponds t o a nominal molar extinction coefficient of approximately 50, and as clupein contains about 40 amino-acid residues, the apparent molar extinction coefficient per peptide linkage would be of the order of 1. Even if it is granted th a t in such a relatively small molecule the peptide fabric chromophor is not developed t o the maximum possible degree, this figure is not comparable with the molar extinction coefficients of the model compounds containing the C=N chromophor cited by Schauenstein (1949a, b). It may be pointed out, moreover, th at compounds are known containing either the single C=N or even the double N=C-C=N chromophors, which are reported t o have no maxima in the 2500 A. region but only a t much shorter wavelengths (see Braude, 1945 for examples). Furthermore, three of the model C=N compounds cited by Schauenstein as having maxima in the region of 2500 A. are actually the anions of oximes in which the chromophoric group has the structure C=N-0rather than -C(OH)=Nor -C(O-)=’Nas envisaged for the peptenol chromophor. The flat nonselective absorption observed b y McLaren for clupein may be ascribed with some confidence to light-scattering, in view of its obvious extension t o long wavelengths outside the limits of the protein absorption band proper, though confirmation of this explanation is clearly desirable. For a less concentrated solution of very pure clupein i t was observed in this laboratory (Beaven et al., 1950) that the absorption from ca. 3300 A. down to the onset of the short ultraviolet bands was practically negligible and flat. (3) Proteins or peptides containing either tyrosine or tryptophan, but not both, give absorption curves which resemble very closely, both qualitatively and quantitatively, the curves of the appropriate free aromatic amino acid, after allowing for the longwave shift and loss of resolution which results from peptide combination (Section VI, 2). Typical substances which satisfy this condition are insulin (12 % tyrosine) and gramicidin (39% tryptophan). (See Table VII and Fig. 11.)
360
G. H. BEAVEN AND E. R . HOLIDAY
(4) Polypeptides and low molecular weight proteins which can be obtained in high purity (e.g., by crystallization) and free from traces of biological pigments give absorption curves in which the protein band ends relatively sharply a t approximately 3200 A. (in neutral or acid solution) with no sign of any contribution from a peptide fabric absorption band extending from 2800 A. (Anslow) to longer wavelengths. Typical examples are lysozyme (M 13,700) and gramicidin (M 3000), in both of which longwave light scattering appears to be negligible at the concentration used in spectrophotometry. ( 5 ) The enzymic degradation of proteins (Haurowitz and Astrup, 1939; Lerner and Barnum, 1946; Sizer and Peacock, 1947; Beaven et al., 1950) causes very little alteration in the over-all intensity of the absorption band, although it results in the expected shortwave shift of the finestructure bands towards the positions th at they have in simple lowmolecular weight polypeptides. The reduction in molecular size, which is the essential feature of the degradation process, clearly reduces the total number of individual peptide links in the protein, and also greatly reduces the possibility of long-range order extending through the original undegraded molecule through the agency of the peptide fabric chromophor in any of the various forms which have been suggested for it. Such degradation would be expected to reduce the absorption contribution of the fabric chromophor, and hence the over-all intensity of the total protein absorption band. (6) The evidence obtained by Anslow and coworkers for the existence of a band a t ca. 2800 A. in simple saturated compounds containing the peptide linkage is completely a t variance with the general body of spectroscopic knowledge of the ultraviolet absorption spectra of such compounds. The evidence in support of the fabric absorption derived from the absorption spectrum of gelatin is based on the assumption th a t this material is completely and consistently free from tyrosine. This assumption is quite unjustifiable, as numerous chemical analysis all show th a t gelatin contains variable but significant proportions of tyrosine; the reported values range from zero t o 0.4 g./100 g. (Block and Bolling, 1945). The tyrosine content of gelatin depends on the method used t o prepare i t from collagen (Loofbourow et al., 19491, and the phenylalanine content is also variable (1.2-2.3 g./100 g.). All analytical results agree in indicating the absence of tryptophan. (7) Any interpretation (Anslow and Nassar, 1941) of the finestructure bands of proteins which does not take into account the longwave shifts due t o peptide combination and environmental effects is likely t o result in the erroneous identification of aromatic amino-acid bands as arising from other presumptive chromophors. In the work of
ULTRAVIOLET ABSORPTION SPECTRA
361
Schauenstein and coworkers, in which the peptenol chromophor is assigned t o the 2500 A. region, longwave shifts associated with the steeply-rising absorption of shorter wavelength are likely to be involved but these have not yet been studied in detail. (8) Any comparison between the observed absorption of a high molecular weight protein and the calculated absorption due t o its chemically-determined content of tyrosine and tryptophan (and possibly also cystine and phenylalanine) must take into account the possible contribution t o the observed density arising from light losses due t o scattering. The form and magnitude of the correction function depends, inter alia, on the size and shape of the protein molecule. An extensive theoretical background is available (see Oster, 1948, and Doty and Edsall, 1951, for reviews) for the computation of scattering functions, and has been applied with some success in the case of a few proteins and polymers for the measurement of molecular weight and dissymmetry. The scattering of light by protein solutions in the ultraviolet has been studied by Custers et al. (1933) for gelatin, and for silk fibroin, gelatin, actomyosin and Fand G-actins by Schauenstein and coworkers (see especially Treiber and Schauenstein, 1949). Their results show that the wavelength exponent of Tyndall scattering is invariably less than the value of 4 required by the Rayleigh theory, and is best evaluated empirically from the slope of the plot of log density versus log wavelength in the region of pure scattering outside the region of true absorption, from the visible down t o about 3100 A. (see also the work of Doty and Steiner, 1950). As this method of evaluating the exponent of the scattering function is admittedly rather complicated, Schauenstein and Treiber (1950) have also used a simplified method based on the exposure time found necessary t o equal the plate blackening due to scattering outside the region of true absorption. The application of either this empirical correction, or of the “exact” correction based on the measured value of the wavelength exponent, to the absorption curves of, e.g., actomyosin, certainly bring the curves into much closer correspondence with the free aromatic amino-acid curves by eliminating the longwave scatter “pedestal” on which the true absorption is superimposed. It would appear th a t exact corrections for light scattering in protein solutions are not easy t o evaluate. For this reason it seems unsatisfactory to place too much emphasis on the difference between the calculated and observed absorption curves of a protein, when the magnitude of this difference is largely determined by the nature and magnitude of the scattering corrections. It is clear from their papers th at Schauenstein and his coworkers have investigated such corrections in some detail, and the methods th a t they have devised for their evaluation are probably adequate for the routine spectropho-
362
G . H. BEAVEN AND E. R. HOLIDAY
tometry of proteins solutions. But it appears to the reviewers th a t some attempt should be made t o demonstrate the existence of peptenol chromophor absorption in proteins or polypeptides of low molecular weight in which light scattering corrections are small or negligible. (9) A part of the evidence for the existence of peptide absorption, either of the peptide fabric (Anslow) or of the peptenol chromophor (Schauenstein) , is based on measurements made in solutions of high p H (> 12) in which the tyrosine residues are more or less completely ionized. As pointed out elsewhere in this review (Section VII, 2) the quantitative study of the ionization of the tyrosine residues in proteins is complicated, due to the change in pK of the tyrosine hydroxyl groups a s a result of peptide combination, and the possibility th a t these same hydroxyl groups may be involved in hydrogen-bonded structures with groups in adjacent peptide chains. The change in p K necessitates the use of p H values greater than 12 in order t o ensure complete ionization of the tyrosine residues, and hence complete development of the tyrosine anion spectrum. It is clear from the absorption curves given in some published work that this condition has not always been satisfied (cf. Sizer and Peacock, 1947). As shown above (Section VII, 4) the breakdown of the hydrogen-bonded structures in some proteins is a slow reaction and merges into the more complex process of alkaline denaturation, which may be associated with the development of turbidity, and further changes in absorption intensity. Because of these complications, the complete interpretation of the absorption spectra of proteins a t high p H values is still open t o discussion and does not provide the best experimental conditions for the demonstration of the peptenol chromophore. As in the case of the comments made above on light-scattering corrections, i t seems that experiments on polypeptides and proteins of low molecular weight, in which the above complications may be expected to be of less importance, are likely to provide more convincing information. Lysozyme, which is stated t o be particularly resistant t o alkaline denaturation, may be noted as a possible test substance. Finally, i t may be pointed out that the formation of the peptenol chromophor envisaged by Schauenstein, which involves complete enolization of a peptide link: 0
OH
k would be expected to have far-reaching effects on the allowable configurations of the polypeptide chain, because of the restrictions imposed by the
ULTRAVIOLET ABSORPTION SPECTRA
363
introduction of the C=N double bond. The recent work of Elliott and his associates (1949, 1950, 1951), however, provides much evidence that the conventional bond angles and interatomic distances, which have often been used in discussing the geometry of the fibrous proteins, are satisfactory for the construction of model /%type fully-extended chains and also for a type of a-folding which satisfies both X-ray and infrared dichroism criteria. Other proposals for polypeptide chain models, however, make use of C-N bonds with appreciable double-band character. Thus from X-ray diffraction studies on glycine and diketopiperazine Corey (1940) found that the C-N bond distance (1.33 A.) was decidedly less than the expected value for a pure C-N single bond. The shortened bond was ascribed t o resonance involving a double bond ionic structure. H
0-
\+
/
,N=c\
Corey used his X-ray data to construct a polypeptide chain model with 1.33 A. C-N links of similar partial double-bond character. More recently Pauling et al. (1951) have employed further X-ray diffraction data on simple crystalline amino acids and related compounds t o construct two hydrogen-bonded helical configurations of the polypeptide chain involving a C-N distance of 1.32 A., corresponding to nearly 50% double-bond character, and a planar configuration for each peptide residue. In these models every nitrogen atom forms a hydrogen bond (energy 8 Kcal./mol) with an oxygen atom of another residue, which satisfies appropriate conditions for the N-H . . . 0 distance and for the angle between the N-H direction and the vector from the N atom to the hydrogen-bonded 0 atom (X30"). The helical polypeptide chain structures devised by Pauling, Corey and Bransom are not identical with the structures, including those of a helical character, proposed by earlier workers. They are stated to be applicable to the fibrous proteins (in their a forms) and also t o globular proteins and synthetic polypeptides. The importance attached to a C-N bond of partly double-bond character might be regarded as providing some support for the C(OH)=N peptenol chromophor suggested by Schauenstein, though in this connection the following difficulties may be noted: (a) The helical polypeptide configuration is based on bond lengths and angles derived from X-ray diffraction studies on crystalline materials. The application of these data t o protein molecules in solution may be questioned.
364
G. H. BEAVEN AND E. R. HOLIDAY
(b) The involvement of every peptide N atom in a n interchain hydrogen bond, and hence in partial enolization, is scarcely consistent with the relatively low absorption intensities that appear t o be associated with the peptenol chromophor. (c) The implications of the helical configurations with respect to ultraviolet and infrared dichroism have not yet been considered. I n the reviewers’ opinion the dichroism associated with the “ additional ” absorption of the peptenol chromophor near 2500 A. is the best evidence in support of Schauenstein’s proposals, as most of the contrary indications are based on observations made on globular proteins. For this and other reasons ultraviolet dichroism studies on oriented proteins are becoming an important new field of research. It may be noted, however, that Perutz et al. (1950) were unable t o detect any significant ultraviolet dichroic effect in several fibrous proteins of the at-keratin type. The validity of absorption measurements on oriented material has been exhaustively criticized by Commoner (1949), though Wilkins and coworkers (1950) have pointed th at both intrinsic and form dichroism are involved, and suggest th at some caution should be exercised in the interpretation of dichroism measurements. As a final comment on the apparently close connection between the alleged absorption of the hydrogen-bond modified peptide link and the various configurations which have been proposed for folded a-type polypeptide chains, it may be pointed out th at the support adduced by Anslow (1945) for the Wrinch cyclol hypothesis of protein structure is susceptible t o the massive weight of criticism th a t has been directed against this hypothesis (Pauling and Niemann, 1939). 5. T h e Absorption of the Isolated Peptide Linkage Absorption due to the peptide linkage has been sought a t much shorter wavelengths by Saidel and coworkers (1950; 1951a, b). From observations on bovine albumin, glycylglycine and diglycylglycine, using a vacuum spectrograph (cf. Platt and Klevens, 1944), it was concluded tha t a n absorption maximum occurring a t ca. 1900 A. in these substances was associated with the peptide linkage. More extensive measurements on a variety of proteins and amino acids were made a t 2050 and 2010 A., on the longwave side of this 1900 A. band, using a Beckman photoelectric spectrophotometer, and applying corrections for stray light (Saidel et al., 19 51~) . From the nearly constant specific extinction coefficients a t 2050 A. of several proteins of known molecular weight, their molecular extinctions a t the same wavelength were calculated, and also the average contribution per peptide bond to this value. The calculated peptide bond contributions were then corrected by subtraction of the contribu-
365
ULTRAVIOLET ABSORPTION SPECTRA
tions of the free constituent amino acids a t 2050 A. (calculated in accordance with the composition of the protein), to allow for absorption arising from the side-chain groups. The results obtained in this manner gave a n average figure for the molar extinction coefficient of the peptide bond a t 2050 A. of ca. 2800 500, which is several times greater than that of saturated carboxylip acids and amides a t this wavelength. I n connection with what has been said above (Section VII, 3) about the possible existence of the C(OH)=N “peptenol ” chromophor in oriented proteins, it is interesting t o note that Saidel and coworkers suggest that a peptide linkage absorption of this magnitude might be associated with some double-bond character in the C-N bond, as is indicated by X-ray diffraction studies on simple amino acids and a dipeptide (Pauling et aZ., 1951)) and quoted in this connection a value for e of ca. 8500 a t ,A, 1900 A. th a t has been given for the average absorption characteristics of the C=N chromophor (Harrison, Lord, and Loofbourow, 1948). From the variations in for various dipeptides, Saidel et al. suggested that it was affected by steric and electrostatic factors arising from the nature of the side-chain groups and their relative positions in the peptide chain, and th a t the average value of the unit peptide bond absorption in proteins would be determined by the operation of these factors. The identification of the ca. 1900 A. absorption maximum as a peptide linkage band by Saidel et al. is not inconsistent with what is known about the absorption characteristics of saturated carboxylic acids and amides in the 2000 A. region, and provides a reasonable explanation for the familiar steep rise in absorption of proteins a t wavelengths shorter than 2500 A. which is observed even in the absence of the aromatic amino acids. It is these acids, together with arginine, cystine, methionine and histidine, which require the greatest side-chain absorption corrections in calculating the absorption of the peptide linkages only at 2050 A. Thus for trytophan and phenylalanine, the molar extinction coefficients at 2050 A. are 20,400, and 8600 respectively, compared with ca. 2800 for the peptide linkage, and even for cystine, e2050 is 2200. Thus it would be necessary t o know the amino acid composition of a protein in some detail before its average eitzide could be determined, and small variations in this value due t o steric and electrostatic effects of the side chains evaluated. A possible source of error in the measurement of arises from the fact that the measurements are made on the side of a band where the absorption is changing very rapidly with wavelength. Any longwave shifts of the side-chain absorption bands due to peptide combination, such as are known t o occur in the 2500-3100 A. absorption bands of the aromatic amino acids (Section VI, 2), would be expected t o havea large
€;:side
366
G . IS. BEAVKN A N D E. R. HOLIDAY
effect on their absorption a t 2050 A . ; for this reason the side-chain absorption corrections might best be evaluated on simple peptides of known structure rather than on free amino acids. The errors associated with stray radiation effects in spectrophotometry have been examined in some detail in recent years (Saidel et al., 1951c; Beaven and Holiday, 1950b; Hogness et al., 1937)) and it is possible to evaluate the necessary corrections, though these may not remain constant with time for a given spectrophotometer and light source. It is probably true to say th a t with currently available equipment, accurate spectrophotometry below 2200 A. is still difficult and laborious, and further instrumental advances are highly desirable. Another possible source of error in spectrophotometry a t 2050 A., is the absorption due to impurities in the sample, which may be expected to increase appreciably with decreasing wavelength, even though it may be negligible in the 2500-3100 A. collective aromatic amino-acid absorption band. The same spectral region has also been studied by Setlow and Guild (1951). From a comparison of the absorption of glycine, L-leucine and glycyl-L-leucine a t wavelengths shorter than 2700 A., they concluded th a t below 2300 A. the peptide bond represents the major absorbing structural unit of a protein, e.g. to the extent of 85% for ovalbumin a t 2100 A., with tryptophan and tyrosine accounting for 11 % and 453, respectively. From Setlow and Guild’s data for glycyl-L-leucine, it appears that the peptide bond absorption is measurable a t wavelengths as long as 2700-2800 A. There appears t o be a quantitative discrepancy between Setlow and Guild’s results for glycyl-L-leucine at 2050 A., from which ~::6,”,~, is ca. 4000, and those of Saidel et al. (1951b)) which indicate this value to be ca. 2800 500. Setlow and Guild did not apply stray light corrections to their measurements, which were carried out on a new Beckman spectrophotometer th at was assumed to be free from stray radiation; their curves above 2500 A. suggest the presence of some weakly-absorbing impurities. The possibility of absorption by the peptide bond at 2800-2700 A. is of importance in studies on the photochemical decomposition of proteins, as energy transfer from the strongly-absorbing aromatic amino acids t o the peptide bond may then be envisaged. Such a transfer of absorbed energy from aromatic side-chain groups to the adjacent peptide linkages was discussed by Mitchell and Rideal (1938) in an investigation of photochemical reactions in protein monolayers. More recently, however, Rideal and Roberts (1951) have concluded from a comparison of the absorption of several proteins below 2550 A. with that of their constituent amino acids, t hat radiation is absorbed directly by the peptide linkage, and estimate t hat a t 2537 A. some 5 % of the total absorption occurs in
ULTRAVIOLET ABSORPTION SPECTRA
367
this manner. Their calculation of the quantum efficiency of the photochemical aggregation process by 2537 A. radiation makes use of this figure, and does not consider the much larger proportion of such radiation absorbed by the aromatic side-chain groups as being involved in the photochemical opening of the peptide linkage. Rideal and Roberts also discuss the additivity of absorption of the amino acids in proteins and conclude that quantitative agreement between the observed and calculated absorption curves of proteins is poor, even a t wavelengths longer than 2600 A. where peptide linkage absorption is absent. They suggest that longwave shifts and changes in the extinction coefficients of the aromatic amino acid residues, both due t o peptide combination, may be responsible for such discrepancies. From the work reviewed in this section, it appears that the selective absorption of the peptide linkage is, in principle a t least, accessible to measurement in the 2000-2200 A. region by using commercial photoelectric spectrophotometers and applying stray light corrections. The maximum of the peptide linkage absorption band, which is located a t ca. 1900 A. (Saidel et at., 1950b) is outside the range of such instruments. It appears that peptide linkage absorption falls rapidly with increasing wavelength, though i t is still appreciably a t 2537 A. and Setlow and Guild consider that it extends as far as 2700-2800 A. The photochemical significance of direct absorption by the peptide linkage in the 2500 A. region has been discussed in detail by Rideal and Roberts, Setlow and Guild suggest that energy might be transferred to the peptide bond following absorption b y the aromatic amino acid side chains, the absorption bands of which overlap the peptide absorption band, and argue that existing data on the quantum efficiency of photochemical enzyme inactivation (McLaren, 1949b) are consistent with this view, which is not utilized however by Rideal and Roberts. It appears that peptide linkage absorption does not extend to sufficiently long wavelengths t o interfere with the spectrophotometric estimation of tyrosine and tryptophan in proteins. The discrepancies in the additivity of amino acid residue absorption in proteins to which Rideal and Roberts draw attention requires further study, however. The possible value of the peptide linkage absorption band in the study of protein structure has been most clearly envisaged by Saidel and coworkers (1951b), and the approach which they have developed is of great interest.
6. The Absorption Characteristics of Living Tissues Although this review is not concerned with spectromicroscopy itself, some comment seems necessary on the controversy regarding the alleged differences in the absorption of protein and nucleic acid in living and
368
G . H. BEAVEN A N D E. R . HOLIDAY
nonliving cells. Some statements on this subject have appeared in a recent book (Brachet, 1950) which require comment. Brachet quotes the results of Larionow and Brumberg (1946)) who found no selective absorption a t 2600 A. (due to nucleic acid) in either the nucleus or nucleolus of living cells. When the cells were killed by prolonged exposure t o ultraviolet radiation or by chemical agents, the expected absorption was found, in agreement with Caspersson’s results. Brachet suggests that the results of Larionov and Brumberg “indicate, if correct, that thymonucleic acid in the living cell is present in a different form than previously recognized. Undoubtedly (reviewers’ italics), the purine and pyrimidine bases may be combined in such a manner th a t they do not absorb U.V.” In support of this view Brachet erroneously cites Crammer and Neuberger’s demonstration th at a proportion of the phenolic hydroxyl groups of the tyrosine residues in native ovalbumin are not free t o ionize t o the anion without previous denaturation. This work has been discussed above (Section VII) and it is only necessary t o point out here th at the anomaly refers to the conditions required for the ionization process and not t o the absorption of the molecular species involved. Both in the neutral and ionized forms, the absorption characteristics of the tyrosine residues in ovalbumin are qualitatively and quantitatively in agreement with its total tyrosine content. Thus this work provides no support for Brachet’s suggestion. Brachet also quotes the earlier work of Vl6s and Gex (19341, who reported that the living sea urchin egg did not show the characteristic absorption spectrum of protein, but th at this appeared after cytolysis. No confirmation of this observation seems to have been made, but the nucleic acid and protein absorption of living chick fibroblast nuclei has been studied quantitatively by Walker and Davies (1950). They found that the nuclei of these cells, examined without previous irradiation, had appreciable optical densities (ca. 0.1-0.15) a t 2650 A., which were sometimes greater and sometimes less than th at of the surrounding cytoplasm. They confirmed Larionow and Brumberg’s findings to the extent that they also found the optical density of the nuclei, not only a t 2650 A., but over the whole nucleic acid-protein absorption region (ca. 2500-3100 A.), increased progressively during irradiation in the course of extended observations. They were also able to show th at the increase in optical density was due t o a decrease in nuclear area during irradiation (as originally suggested by Larionow and Brumberg), while calculation of the total integrated absorption of the whole nucleus indicated that the total amount of absorbing material did not increase but actually decreased on irradiation; this effect may possibly be explained either by loss of nucleotides through the nuclear membrane or by chemical changes due t o irradiation. Brad-
ULTRAVIOLET ABSORPTION SPECTRA
369
field (1950) found th at the changes in the ultraviolet absorption characteristics of living cells caused by irradiation differ for various types of cell. H e obtained all three possible results, uix., increase, decrease and no change in absorption. He also found th a t a reduction in absorption on irradiation could be prevented b y the presence of reagents known t o destroy sulfhydryl groups, suggesting that the stability of nucleic acids towards irradiation may be affected by the state of the sulfhydryl groups in the associated proteins. There is therefore no reliable evidence th a t the selective absorption properties of nucleic acid and the aromatic amino-acids residues of proteins in living structures differ from those of the same materials in nonliving tissue, though effects due t o orientation of the absorbing material and light losses by scattering may seriously complicate the quantitative interpretation of microspectrophotometric measurements on biological systems. IX. ANALYSISOF THE ABSORPTION SPECTRA OF PROTEINS IN TERMS OF TYROSINE AND TRYPTOPHAN CONTENT 1. General
Although i t had been very early suggested th a t the selective absorption of proteins was due t o the contained aromatic amino acids it was not until the advent of quantitative methods of spectrophotometry in the form of the Twyman (1910) sector photometer th a t this could be studied more closely. Several workers measured the absorption constants of the amino acids and it was Stenstrom and Reinhardt (1925) who first noted the marked change in the spectrum of tyrosine on passing from acid t o alkaline solution. One of the present authors (Holiday, 1936; Holiday and Ogston, 1938) studied several proteins starting with the hypothesis that the absorption curve of a protein is the simple sum of those of its contained amino acids. If this were the case then the absorption curve of the protein could be analyzed by the method of mixtures (Twyman and Allsopp, 1934) and the concentration of the contained amino acids determined. The possibility of performing such an analysis depends on several factors discussed below. One of these is the possibility of finding a region of the spectrum where relatively few amino acids absorb. Such a favorable region exists between X2800-3150 A. where only tyrosine and tryptophan absorb. I n proteins and polypeptides containing no tyrosine or tryptophan, e.g. gelatin and polymyxin, the absorption in the region A2400 to 2800 A. should be due only to phenylalanine and cystine. Analysis of the absorption spectrum of a protein in the region X2800-3200 A. did indeed lead t o figures for tyrosine
370
G. H. BEAVEN AND E. R. HOLIDAY
and tryptophan in fair agreement with those obtained by chemical analysis (Section IX, 5 ) . 2. Method of Mixtures as Applied to Absorption Data
The method is strictly applicable only when the Beer-Lambert law holds, that is, when the optical density (Dx) a t wavelength X of a solution is porportional t o thickness of layer (1 cm.) and concentration (c). Concentration is conventionally expressed as gram-moles per liter or grams per cent. I n the case of the aromatic amino acids the Beer-Lambert law is obeyed over the concentration ranges met with in practice and the following arguments are therefore confined t o such cases. Cases where the law is not obeyed have been discussed by Brattain et al. (1943). If, for a protein, we define K , the extinction coefficient, as
where c is expressed in grams per liter and 1 is in centimeters, 1 0 and I having the usual signification. Also if we define molar extinction coefficients of the amino acids as
where C is in gram-moles per liter, then it caii be shown that in a protein or mixture of amino acids,
where M 1 and Mz are the gram-moles of amino acids 1 and 2 in 1 g. of protein and a and b are the wavelengths a t which measurements are made. The applicability of the above analysis t o a protein depends on the following considerations : (1) Only a small region of the spectrum of a protein between A2700 and 3150 A. is available for analyzing the absorption of a protein as th a t of a two-component system of tyrosine and tryptophan. (2) The value of the analysis depends on the assumption th a t the absorption by these two amino acids is unaffected by their incorporation into the protein. It has now been shown that this is not strictly true (Section VI). (3) If there is interfering absorption by other material in the protein
ULTRAVIOLET ABSORPTION SPECTRA
371
then error will be introduced unless this can be allowed for. The same applies t o interference by scattering. (4) The solution of the above equations gives the molar concentration of these amino acids in one gram of protein. To determine K X for a protein the concentration of the protein must be known. The Kjeldahl method is the most convenient. For some important proteins the nitrogen content of the dry ash-free protein has been determined. I n these cases the content of tyrosine and tryptophan may be calculated in th e dry ashfree protein. I n other cases it is probably best t o follow the convention of relating the content of tyrosine and tryptophan t o a dry weight basis ol protein assuming a nitrogen content of 16 g. of nitrogen per 100 g. of protein. (5) Where the concentration of the protein is unknown, useful information from a n analysis of its absorption spectrum may still be obtained, since from equation (3) and (4) the molar ratio (or weight ratio) of the two amino acids may be obtained, for
This ratio is very characteristic for certain proteins as may be judged from Table VII and Fig. 11. Since it is more useful t o know the molar than weight ratios of amino acids in proteins they have been expressed in this form. An example of the use of this formula is t o be found in a paper by Snellman et al. (1951). (6) The choice of wavelength a t which measurements are made is important in minimizing errors. 3. Errors in Estimates Introduced by Errors in Spectrophotometric Measurement
The manner in which errors of measurement are transferred t o errors in estimates, in the method of mixtures, is not simple and cannot be generally dealt with here. For a complete treatment reference should be made t o textbooks on the subject (Whittaker and Robinson, 1944). Since however the matter is of some importance if the spectrophotometric method for analysis of tyrosine and tryptophan in proteins is to be applied in the best possible manner, some aspects of the problem are dealt with here. I n equations (3) and (4)the concentration of one of the components ( M ) is seen t o depend on six measurements, viz.: four measurements of molar extinction coefficients (e) and two of optical density. Each of these measurements is subject t o error. I n theory we can reduce random
372
G. H . BEAVEN AND E. R . HOLIDAY
errors in e t o any desired value by repeated measurements on pure solutions of tyrosine and tryptophan. The purity of the solutions will have t o be assured by some test, e.g., color reaction or recrystallization t o constant e. It mill also have t o be ascertained th a t the spectrophotometer does not suffer from a systematic error, for such error will not be reduced by repeated measurements. (Random error refers t o reproducibility, systematic error to accuracy.) The estimates of K will probably be derived from duplicate measurements on the mixture (or protein). Usually the estimates of t and of K are made on the same instrument. If tyrosine and tryptophan curves are obtained on the same instrument employed for the study of the mixture of the two (or for the protein) then it can be seen from inspection of equations 3 and 4 th a t systematic instrumental error will tend t o cancel out (see Table IV). It might be argued from this that it is better not t o rely on published standard curves for tyrosine and tryptophan but for each individual t o prepare his own. Although this procedure may be expedient it must be considered a retrograde step. The values given in Table VI are certainly not t o be accepted as final, but it is t o be recommended that every precaution should be taken t o assure chemical purity of the amino acids and t o check the accuracy of the spectrophotometer before attempting t o establish alternative standards. If i t were possible t o find a wavelength region at which only one of the amino acids absorbed, then measurement a t this wavelength would yield directly the concentration of that amino acid. Inspection of the absorption curves of tyrosine and tryptophan (Fig. 10) shows that there is a small portion of the “alkaline” tyrosine curve which extends to longer wavelength than any part of the tryptophan curve. The absorption of tyrosine here is however so low that it would be very sensitive to interference from more than one cause. It is therefore hardly possible t o use this for estimating tyrosine. It can be shown on general grounds th a t the next best wavelengths at which to measure are those where in one case the absorption of one component (let us say, component A) is high relative t o the other (let us say, component B) and in the other case the absorption of component B is high relative to A. (Philpot, 1936). Such considerations led t o the choice of wavelengths A2800 and 3050 A. (Holiday, 1936), measured on solutions in N/10 alkali. Later Goodwin and Morton (1946) selected wavelengths 2800 and 2944 A., the latter being the point of equal absorption for equimolar solutions of tyrosine and tryptophan in N/10 alkali. This choice can be criticized, for measurement a t this point gives no information as to the relative contributions t o absorption by the two amino acids. Morton and Goodwin argued that 3050 A. was on a steep part of the curve of tyrosine and that this made it
ULTRAVIOLET ABSORPTION SPECTRA
373
unsuitable for measurement since it would be too subject t o errors of wavelength setting. The argument may be tested as follows. Suppose the estimates of e have so small a n error as to be negligible and the estimates of K have an error of & 1%. Then a t the two chosen wavelengths the error in K may be either 1% too high or 1% too low. We can therefore tabulate
FIG.10. Plot of mean values of
emo1 of tyrosine and tryptophan in N / l O alkali in the region 2750-3000 A.
four possible extreme conditions of error. I n the case of a mixture of 3 moles tyrosine and 1 mole tryptophan the error introduced in these four cases into estimates of tyrosine and tryptophan can be calculated. In Table I V are shown the results of this when using the Holiday formula and when using the Goodwin and Morton formula with the values of e chosen in Table VI. It can be seen th at the errors in the estimates of tryptophan and tyrosine concentrations are subject t o less error when using the original method. It should be noted th at the large error in tryptophan is due to the small proportion present. As its proportion in a mixture increases
374
G . H. BEAVEN AND E. R . HOLIDAY
so does the error of estimation decrease. will increase.
Conversely that of tyrosine
TABLEIV Percent Error in Estimated Concentration of Tyrosine and Tryptophan Resulting from 1 % Error in Measurement of Optical Density. TyrosinelTryptophan Ratio = S:1 Holiday, Xz = 3050 A.
Case 1 2 3
4
DzsuU +1 -1 +1 -1
Dxz
Tryptophan
Tyrosine
-1
+3.2 -3.2 $1.0 -1.0
-1.6 +l.6 +1 .o -1.0
+l
+I
-1
Goodwin and Morton, Xz = 2944 A. Tryptophan Tyrosine +4.2 -4.3 +1.0 -1.0
-2.7 $2.7 +1.0 -1.0
It is also of interest to note that when the errors a t both wavelengths have the same sign then there is no magnification of error of estimation. Such a situation exists in practice where there is background absorption or a systematic instrumental error. (3) I n the case of proteins another form of error is introduced by the fact that the spectra of tyrosine and tryptophan are shifted to longer wavelengths. By reference t o Fig. 10 i t can be inferred th a t the shift in wavelength of tyrosine and tryptophan will hardly affect and ~ ~ ~ ~ 0 T ~~~~~~y~ ~ y p . will be slightly affected, but ~ ~ 0 and ~ ~ ~ ~ will y p be considerably affected. Inspection shows that the Goodwin and Morton formula introduces less error from this type of deviation than does the Holiday formula. This is illustrated in Table V. TABLEV Error Introduced into Estimates of Tyrosine and Tryptophan by a 10 A . shift in Wavelength of the Spectra of Tyrosine and Tryptophan in a Mixture Containing T h e m in a Molar Ratio of 3/1
Tyrosine Tryptophan
Holiday Xz = 3050 A.
Goodwin and Morton Xz = 2944 A,
+13.8% -16.0%
-12.0%
+8.0%
It is evident that shifts in the spectra would result in errors in estimates of tyrosine and tryptophan. Goodwin and Morton (1946) had noted that the original method of Holiday when applied to proteins tended t o give high figures for tyrosine and low for tryptophan. They explained this as due to interference by extraneous absorbing material in the protein. It seems however that the explanation lies partly in the wavelengths shift which occurs in proteins.
375
IJLTRAVIOLET ABSORPTION SPECTRA
4. Recommended Method for Estimation of Tyrosine and Tryptophan in Proteins
It is recommended that the Goodwin and Morton formula (1946) be used.
This entails measurement of optical density of a solution of
TABLEVI Collected Values for Molar Extinction Coeficients of Tyrosine and Tryptophan in Alkali in the Region 2750-3100 A. X mp
280
285
290
2944
295
300
1.58 1.54 1.50 1.S8 1.51 1.51 1.54
1.96 1.80 2.02 1.90 1.92 1.97
2.38 2.24 2.15 2.35 2.24 2.25 2.32
1 536
1.93
2.275
5.43 5 43 5.43 5.25 5.40 5.25
4.87 4.57 4.65 4.67 4.55
4.10 4.13 4.35 4.13 4.20 4.30
2.29 2 17 2.15 2.18 2.00 2.26
0.93 1.01
5.34
4 64
4.21
2.18
1-10
305
Author
TYROSINE IN N/10 ALPALI, rmoi X 10-8 2.39
2.40
2.36 2.26 2.30 2.39 2.32 2.34 2.35
1.89 1.97 2.02 1.98 1.93 1.97
1.25 1.14 I . 25 1.22 1.25 1.15 1.23
Goodwin and Morton (1946) Holiday and Oaston (1938) Crammer and Neuberger (1943) Drake (1949) Beaven and Holiday (1951) Froinageot and Schnek (1950) Vandenbelt (1346)
2.33
1.96
1.23
Mean
______
_
TRYPTOPHAN IN N / 1 0 ALKALI,rmoi X 10-8 25 1.14 1.10
0.52 0.50 0.45 0.42
Crainnier and Neuberger (1943) Holiday (1936) Holiday and Ogston (1938) Goodwin and Morton (1946) Beaven and Holiday (1951) Vandenbelt (1946)
0.47
Mean
protein in NIlO alkali at wavelengths 12800 and 2944 A. These values, together with the nitrogen or dry weight content of the solution, are used for calculating the K'S for substituting in the equations.
where ~ ~ and ~ 4~ 42 are 8 the ~ ~ extinction coefficients of the protein in N/10 alkali a t A2944 and 2800 A. and and M,,,, are the gram moles of tyrosine and tryptophan, respectively, in one gram of protein. When the nitrogen or protein content is unknown we can calculate
where D is the optical density of the solution a t wavelengths 2800 and 2944 A. This molar ratio of tyrosine t o tryptophan is a useful char-
376
G . H. BEAVEN AND E. R. HOLIDAY
acteristic of a protein (Table VII). As pointed out above (Section IX, 3) errors in the estimates of tyrosine and tryptophan are inevitably introduced by this simple form of analysis. The tyrosine estimate will tend t o be too high and that of tryptophan too low. Some suggestions have been made in the preceding sections on possible ways of reducing these errors. The chief practical sources of error are those arising from haze in the solution or from pigment adsorbed to the protein. The solution should be as clear as possible. A sensitive indication of haze is absorption in the region 3200-3600 A. It is always advisable t o measure optical densities in this region on any solution of protein. If the absorption is appreciable it may be reduced by high speed centrifugation or filtration as indicated in Section 11. If the absorption is not reduced by this treatment the presence of absorbing material other than protein in the solution is indicated. The nature of the protein and the amount of it available will govern the possibilities of removing interfering absorbing material. Where extraneous absorption has been reduced to a minimum some form of correction must be made for this before calculating the specific extinction coefficients of the protein. Such methods of correction have been suggested (Holiday and Ogston, 1938; Goodwin and Morton, 1946). From experience it seems t o the authors th at the simplest correction is t o measure a t points between A3200 and A3600 A. and t o project the line through these points as far as A2800 A. This forms the base line from which optical densities are measured. This base line should therefore be obtained for each solution on which measurements are made. An inner check is available for testing whether the base line corrections have been properly made. It can be seen from Figs. 1 and 2 th a t isobestic points exist in the absorption curves of tyrosine in acid and alkali a t A2768 A. and in those of tryptophan a t A2800 A. under the same conditions. It can be concluded therefore that in a mixture of the two the curves in acid and alkali (0.1 N ) will intersect between A2768 and 2810 A. This rule should apply also t o a protein in acid and alkali. Many curves of proteins are published showing no such point. Such a finding should not be accepted before the conditions of spectrophotometry, e.g., absence of scatter, adequacy of control solution etc. have been checked. For example, it is always safer to measure alkaline solutions of protein against controls containing the same amount of alkali whenever possible. If such checks reveal no factors t o account for the anomaly, then it must be ascribed t o the presence of absorbing material other than tyrosine and tryptophan and consequently the analysis in terms of these two will be partially invalidated.
377
ULTRAVIOLET ABSORPTION SPECTRA
5 . Comparison of Results Obtained by Xpectrophotometric A n a l y s i s with Those Obtained by Chemical Methods I n Table VII are collected values of tryptophan and tyrosine content of various proteins obtained by chemical and spectrophotometric methods. TABLEVII Tyrosine and Tryptophan Content of Proteins Determined by Chemical and Spectroscopic Methods (Grams Amino Acid per 100 g. Protein)
Tyrosine Chem. Spec. Human serum albumin 4.70 5.2 Human serum globulin 6.7' 7.4 Insulin 12.3" 12.7 Pepsin 8.5C 8.3 7.gd 7 . 6 Trypsin 3.7" 4 . 7 Lysozyme 8.OC 8.5 Ribonuclease 13 .Of 1 3 . 5 Tyrocidin 0.00 0.00 Gramicidin 5.5* 8.0k Actoni yosin 4.3i Edestin 4.5 3.6i 4.6 @-Lactoglobulin
Tryptophan Chem. Spec. 0.2a 0.2 2.9" 2 . 8 0.0b 0.0 2.4c 2.4 3.6d 3.7 8.3" 8 .I 0.oc 0.0 16.3' 1 6 . 0 39.60 39.30 1.4" 1.3k 1 .4i 1.3 2.lf 2.2
Molar ratio Tyrosine/Tryptophan Chem. Spec. 26.1 2.6
28.8 3.0
3.8 2.4 0.5
3.9 2.3 0.6
0.9
1 .o
4.4 3.4 2 .o
6.9 3.9 2.3
Brand, Kassel, and Saidel (1944). Brand (1946). c Brand (1948), quoted by Northrop et al. (1948). ' d Kunitz (1947). Fromageot and Schnek (1950). f Christensen, Edwards, and Piersma (1941). 0 Edwards (1049). h Jordan (1947). i Chibnall (1946). j Stein and Moore (1949). k Schauenstein and Treiber (1950). I Hotchkiss (1941). All other results were obtained in the authortl' laboratory. @
In order to arrive a t satisfactory values for substitution in equations (3) and (4) averages have been taken of values for molar extinction coefficients of tyrosine and tryptophan published or privately communicated. These values are given in Table VI and plotted as curves in Fig. 10. The choice of wavelengths 2800 and 2944 A. adopted by Goodwin and Morton (1946) is recommended as being less sensitive to the protein wavelength shift than wavelengths 2800 and 3050 A. recommended by Holiday (1936) and Holiday and Ogston (1938).
378
G . H. BEAVEN AND E. R . IIOLIDAB
6 . Comments on the Probtem of Analysis Since in many respects chemical analysis of the proteins for their tyrosine and tryptophan content is not very satisfactory, especially in the case of the latter, it would seem worthwhile to explore the limits of applicability of the spectrophotometric method. As a sequel to the finding of the shift in the spectra of amino acids in proteins without evidence of change of intensity it should not be impsssible to allow for this by certain modifications of procedure. Firstly, it is evident that the free aromatic amino acids are not the correct standards from which to determine the contributions to protein absorption. The same can be said of peptides of the aromatic amino acids. Although peptide combination causes some shift in the spectra of the aromatic amino acids it is not sufficient to account for the shift occurring in proteins (Section VI, 2). However it may be an improvement to use the peptides for comparison rather than the free amino acids, for Edwards (1949) found a much better fit to gramicidin curves by using glycyl-tryptophan as a standard. A drawback to the use of peptides is that they are more difficult t o purify than the amino acids and some are difficult to synthesize. One should ideally employ a sandwich tripeptide with the aromatic acid in the middle, but no reports of the synthesis of such peptides have come to the authors’ notice.. Secondly, if it were possible t o measure precisely the shift in the spectra it should be possible to allow for this in the calculations. For instance if the shift were estimated as Ah it would be possible to measure the absorption of the protein at X I Ah and h2 Ah or alternatively to measure the amino acids a t XI - Ah and h2 - Ah and the protein at X I and AS. Inspection of the absorption curves of proteins in alkali shows that the estimate of shift should be possible. In Fig. 11 are shown for comparison curves of various mixtures of tyrosine and tryptophan in N/lO alkali together with curves of some proteins under the same conditions. It can be seen at once that a fair estimate of the tyrosine/tryptophan ratio may be obtained by inspection of the head of the band. By the moving-plate method the position of the tryptophan fine-structure band, which forms one maximum of the curve can be estimated to &- 1 A. By standard photoelectric spectrophotometry it can be estimated to 2 -5 A. The difficulty would be to determine whether the shift applies t o the spectra of both tyrosine and tryptophan. However in practice this difficulty would not result in great uncertainty, for, as already noted (Section IX, 3), it is the shift in the tryptophan spectrum which is chiefly responsible for the error in the estimates of tyrosine and tryptophan. Thirdly, it is inevitable that, of the wavelengths chosen at which to measure the values of K for insertion in equations (3) and, (4),two at
+
+
ULTRAVIOLET ABSORPTION SPECTRA
379
least will lie on the rather steeply sloping sides of the absorption curves of either tyrosine or tryptophan. Errors in wavelength settings of the spectrophotometer will therefore contribute significantly to other errors of estimation. I n the Goodwin-Morton formula this will apply particularly t o the 2944 A. setting. One method of eliminating such error would be t o use a source giving a line spectrum. The mercury arc pro-
1
PROTEINS
MIXTURES OF TYROSINE AND TRYPTOPHAN IN
3.7
3.
3.3
3.1
FIG. 11. Comparison of absorption spectra of alkaline solutions of proteins and of mixtures of tyrosine and tryptophan. The curves, obtained with a n automatic recording spectrophotometer, are displaced along the vertical axis to avoid overlapping.
vides two strong lines a t A2804 and 2976 A. with a weaker line a t 2926 A. which might well serve for this purpose. Fourthly, it has been shown (Haurowitz and Astrup, 1939; Lerner and Barnum, 1946; Sizer and Peacock, 1947; Beaven et al., 1950) that the anomalous spectral shift is reduced but not abolished by digestion of a protein with pepsin or trypsin (Section VI, 8). After digestion the shift corresponds closely t o that which is observed in the spectrum of aromatic amino acids in peptide combination (Figs. 8 and 9 and Table 11). A preliminary short digestion
380
G. H . BEAVEN AND E. R. HOLIDAY
of the protein before measurement of its absorption would be expected t o give better values of K for insertion in the equations (3) and (4). The considerations discussed in this section though applying particularly t o the problem of protein analysis, have general bearing on any application of the equations of mixtures to complex molecules containing several relatively independent chromophors, e.g., nucleic acids in which a similar shortwave shift appears following digestion with ribonuclease. Kunitz (1946) has used the rate of decrease in optical density on the long-wave side of the nucleic acid band, e.g., a t 3000 A,, t o estimate the activity of this enzyme. It cannot be said that the analysis of protein spectra has yet been developed to its limits but the discussion given here indicates further lines by which the spectrophotometric method might be made a means of estimating at least one amino acid (tryptophan) for which the chemical methods are notoriously unreliable.
X. OTHERANALYTICAL APPLICATIONS 1. Serum Analysis Of the several protein components of serum, albumin is peculiar in its almost negligible content of tryptophan. This is reflected in the comparatively low specific extinction coefficient, relative to nitrogen content, of albumin a t the maximum of its absorption band as compared with those of other serum components. Edsall (1950) has collected data for amino acid content of the four main serum components derived from the work of Brand et al. (1944), Brand (1946) and Shemin (1945). These show clearly the unique deficiency of tryptophan in serum albumin. The major contribution (90%) of the albumin to the osmotic pressure of the blood has for many years given clinical significance t o variations in the albumin content of whole blood. Since neither corpuscles nor fibrinogen contribute appreciably to the osmotic pressure of blood, knowledge of the albumin content of serum is sufficient for clinical appraisal of the osmotic properties of whole blood. I n earlier times emphasis was laid on the albumin-globulin ratio as clinical evidence of a disturbance of blood proteins balance. Later, it became recognized that the absolute concentration of each separate component of serum or plasma had a special pathological significance. Both the ratio and the concentration of albumin in serum are readily estimated by absorption spectroscopy, provided the term globulin is defined in this context as meaning serum protein containing tryptophan and albumin as the serum protein not containing tryptophan. The value obtained for albumin by the spectroscopic method gives a truer estimate than that given by methods depending on separation of albumin by salting out since in the latter methods the separation of albumin from other proteins is never complete. By
381
ULTRAVIOLET ABSORPTION SPECTRA
reason of a fact, probably quite fortuitous, that there is a virtual absence of tryptophan in the only serum protein which is appreciably responsible for the osmotic pressure of the blood, it is possible very simply and rapidly t o estimate this contribution by absorption spectroscopy. To estimate albumin in serum spectroscopically it is necessary to know the specific extinction coefficient of pure albumin and of the non-albumin moiety of serum. The absorption spectra of many pure protein components are not yet available. Lerner and Barnum (194Ga) have however determined the spectra for human serum albumin and yglobulin, and reference to the data collected by Edsall (1950) shows that the tyrosine and tryptophan content of the other globulins relative to nitrogen content is very similar to that of globulin. We can assume that the spectra of other globulins will therefore be very similar to that of yglobulin. An approximate estimate of mean specific extinction coefficient may therefore be obtained for total serum globulins as suggested by Lerner and Barnum (1946b). Measurements at A2776 A . in 0.1 A' HCl
E*;: : Albumin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Globulin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Totalglobulin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1 13.9 15.0
I
* EiTm is defined as E:Fm = __ . log 2 c x 1 I where c is in grams per 100 ml. and I is in centimeters. This is a frequently used convention. It follows that Ei:m = IOK, defining
K
as in Section IX, 2.
F. C. Smith in 1929 suggested a method of estimating albumin/ globulin ratio in serum by measuring the extinction of the serum against nitrogen content as determined by Kjeldahl. Lerner and Barnum (1946) have used a very similar method. If E, be the extinction coefficient of a serum a t A2775 A. in 1 cm. layer, T, A and G be the concentration in serum of total protein (albumin plus globulin) in grams per 100 ~ r n . ~E,; and E, the E:,", of albumin and globulin a t A2775 A., respectively, then AE, ( T - A ) E , Eo = E ,
+
+
where E o is nonprotein absorption. I n the case of serum E o has been shown by Smith (1929) and Lerner and Barnum (1946) t o be chiefly due to uric acid which may account for as much as 5 % of the absorption a t this wavelength. The latter workers make a constant allowance of E , = 3.1 in equation (1). With values of E, = 5.0 and E , = 15.0 this transforms to A = 1.5T - O.lE, 0.3 G = T - A
+
3 82
G . H . BEAVEN AND E. R. HOLIDAY
The total concentration of protein may be measured b y nitrogen, or by specific gravity or refractive index. The method is generally applicable t o sera and t o serum protein solutions. Care should be taken t o check the possibility of absorption by nonprotein constituents of serum which in some diseases such as cirrhosis of the liver may be considerable. The presence of such substances is shown by a marked fall in absorption of the serum after dialysis. It should be noted that this simple form of analysis is possible because there are only two unknown quantities viz., total concentration of protein and the concentration of one component. The nitrogen content of the solution and measurement of absorption a t one wavelength provide the two necessary parameters for determining the two unknowns. I n the case of the determination of tyrosine and tryptophan in a protein there are three unknowms-the concentration of the protein solution, the Concentration of tyrosine, and the concentration of tryptophan. The nitrogen content of the solution and measurements of absorption a t two wavelengths are necessary t o provide three parameters for solving for the three unknowns. 2, Analysis of I m m u n e Precipitates and I m m u n e Proteins Eisen (1948) and Gitlin (1949) have applied the above type of analysis t o immune precipitates with the twofold aim of estimating antigenantibody ratio and of differentiating antibody protein from other serum proteins. I n the zone of excess antibody or in the case where the antigen shows negligible absorption, the absorption spectrum of the precipitate may be taken a s th at of the antibody and this may be directly estimated. Their results show that the spectra of antibodies resemble closely those of the 7-globulins, agreeing with the findings of Smith and Coy (1946). However in the case of antibodies from colostrum the last workers found considerable variation from the globulin type of spectrum. It would appear from the spectrum of their T-globulin obtained from colostrom, taken together with the figures for the tryptophan content of this protein of 2.60-2.88% given by Smith et al. (1946), th a t its content of tyrosine must be much lower than th at of other globulins. For instance Smith and Coy give EtFm for T-globulin as 10.5 at A2800 A. and Smith et al. (1946) give its tryptophan content as 2.64%. This content of tryptophan would contribute 7.2 to the EiFm a t 2800 A., leaving 3.3 a s the contribution of tyrosine. Calculation shows th a t this would correspond to a tyrosine content of just over 4% compared with nearly 7% in globulins and immune proteins of the globulin type. REFERENCES Allen, A. J., Steiger, R. IC., Magill, M. A., and Gibson, It. G. (1937). Biockem J. 31, 195. Alexander, P., and Earland, C.
(1050). Nature 166, 345.
ULTR.4VIOLET ABSORPTION SPECTRA
383
Alexander, P., Hudson, R. F., and Fox, M. (1950). Biocheni. J . 46, 27. Anslow, G. A. (1945). J . Applied Phys. 16, 41. Anslow, G. A. (1932a,b). Phys. Rev. 40, 115, 639. Anslow, G. A. (1950). Faraday Soc. Discussion No. 9, p. 299. Anslow, G. A., and Foster, M. I,. (1932). J . Riol. Chem. 97,37. Anslow, G. A., Foster, M. L., and Barnes, D. (1930). J . Biol. Chem. 89, 665. Anslow, G. A., Foster, M. I,., and Klingler, C. (1933). J . Biol. Chem. 103, 81. Anslow, G. A., and Lyman, E. R. (1941). J . Optical SOC.Am. 31, 114. Anslow, G. A., and Nassar, S. C. (1941). J . Opticul Soc. Am. 31, 118. Arnold, 1,. B., and Kistiakowsky, G. B. (1932). J . Am. Chem. Soc. 64, 1713. Baly, E. C. C., and Ewbank, E. K. (1905). J . Chem. Soc. 1347. Barkdoll, A. E., and Ross, W.F. (1944). J . Am. Chenr. SOC.66,951. Beaven, 0. H . (1951). Unpublished work. Beaven, G. H., and Green, M. (1951). Unpublished work. Beaven, G. H., Hoch, H., and Holiday, 11:. R. (1951). Biochem. J . 49, 374. Beaven, G. H., and Holiday, I<. R. ( 1 9 5 0 ~ ) . Faraday Society Discussion No. 9, p. 492. Beaven, G. I€., and IIoliday, E. R. (19501)). Photoelectric Spectrometry Group, Bulletin No. 3, 53. Beaven, G. H., Holiday, E. R., and Jope, EL M. (1950). Faraday Society Discussion No. 9, pp. 406, 492. Beaven, G. H., and Holiday, E. R. (1951). Unpublished work. Bednarczyk, W., and Marchlewski, L. (1937). Bull. intern. acud. polon. sci., Classe sci. math. nat., Series A, 140. Block, R. J., and Bolling, D. (1945). The Amino Acid Composition of Proteins and Foods. Thomas, Springfield, Illinois. Brachet, J. (1950). Chemical Embryology. English translation by Lester G. Barth. Interscience Publishers, New York and London, p. 234. Bradfield, J. R. G. (1950). Faraday Society Discussion No. 9, p. 481. Brand, R. (1946). Ann. N . Y . Acad. Sci. 47, 487. Brand, E. (1946). Ann. N . Y . Acad. Sci. 47, 187. Brand, E.,Kassel, B., and Saidel, L. J. (1944). J . Clin. Invest. 23, 437. Brand, E., and Kassell, B. (1942). J . Biol. Ckern. 146, 365. Brand, E., Saidel, L. J., Goldwater, W. €I., Kassell, B., and Ryan, F. J. (19451. J . Ant. Chem. Soc. 67, 1524. Brattain, R. R,., Rasmussen, R. S., Cravath, A. M. (1943). J . Applied Phys. 14, 418. Braude, E. A. (1945). Ann. E e p t s . Chem. Sac. 43, 105. Braude, E.A. (1960). J. Chenz. SOC.379. Brdicka, R., and Pavlik, M. (1930). Bull. intern. acad. sci. Boheme 1. Brode, W. R. (1939). Chemical Spectroscopy Wiley, New York; Chapman and Hall, London. Bruigom, E. S. (1950). Eigenschappen van Tyrosine als Bouwsteen van Eiwitten en Peptiden. Thesis, Leiden (in Dutch, with a summary in English). Butenandt, A., Friedrich-Ficksa, H., Hartwig, S., and Scheibe, G. (1942). 2. physiol. Chem. 274, 276. Callow, R. K. (1936). Biochem J . 30,906. Cannan, R. K., Palmer, A. H., and Kibrick, A. C. (1942). J . Riol. Chem. 142, 725. Caspersson, T. (1936). Skand, Arch. Physiol. 73,suppl. 8. Caspersson, T. (1940). J . Roy. Microscop. Soc. 60, 608. Chibnall, A. C. (1942). Proc. Roy. Soc. B131, 136. Chibnall, A. C. (1946). Proctor Memorial Lecture.
384
G. H. BEAVEN AND E. R. HOLIDAY
Christensen, H. N., Edwards, R. R., and Piersma, H. D. (1941). J. B i d . Chem. 141, 187. Cole, 1’. A., and Brackett, F. S. (1940). Rev. Sci. Instruments 11,419. Corey, R. B. (1940). Chem. Revs. 26, 227. Commoner, I3. (1949). Science 110,31. Commoner, B.,and Lipkin, D. (1949). Science 110, 41. Coulter, C. B., Stone, F. M., and Kabat, E. A. (1936). J . Gen. Physiol. 19, 739. Crammer, J. L., and Neuberger, A. (1943). Biochem. J . 37, 302. Custers, J. F. H., De Boer, J. H., and Dippel, C. J. (1933). Rec. trav. chim. 62,195. Davies, M. (1950). Faraday Society Discussion No. 9, p. 332. Dean, A. C. It. (1949). Trans. Faraday SOC.46, 487. Drake, B. (1949). Private communication. Dunn, M. S., and Rockland, L. B. (1947). Advances in Protein Chem. 3, 296. Doty, P., and Edsall, J. T. (1951). Advances in Protein Chem. 6, 37. Vol. VI. Doty, P., and Steiner, It. F. (1950). J . Chem. Phys. 18, 1211. Edsall, J, T. (1950). Ergeb. Physiol. Diol. Chem. exptl. Pharmakol. 46, 308. Edwards, B. G. (1949). Arch. Biochem. 21, 103. Eisen, €1. N. (1048). J . Zmmunol. 60,77. Elliott, A., Anihrose, E. J., and Temple, R. B. (1949). Nature 163,859. Finkelstein, P., and MrLaren, A. 1). (1949). J . Polymer Sci. 4, 573. Fixl, J. O., Kratky, O., and Schauenstein, E. (1949a). Monatsh. 80, 153. Fixl, J. O., Kratky, O., and Schauenstein, E. (194913). Monatsh. 80, 439. Fraenkel-Conrat, H. (1950). Arch. Biochem. 27, 10. Fromageot, C.,and Schnek, G. (1950). Biochem. et Biophys. Acta 6, 113. Fruton, J. S., and Lavin, G. I. (1939). J . B i d . Chem. 130,375. Giese, A . C., and McLean, D. J. (1950). J . Biol. Chem. 187,537. Gitlin, D. (1949). J . ZnLmunol. 62, 437. Goldring, L. S., Coryell, C. D., Hawes, R. C., Hare, G. H., Beckman, A. O., and Stickney, M. E. (1951). Anal. Chem. (in press). Goodwin, T. W., and Morton, R . A. (1946). Biochem. J . 40, 628. Gridgeman, N. T. (1951). Report on the 2nd Collaborative Test of Photo-electric Spectrophotometers by the Photoelectric Spectrometry Group (Hon. Sec., W. H. Storey, Arbury Works, Cambridge, England). Bulletin No. 4. 67. Haas, W. J., Sizcr, I. W., andLoofbourow, J. R. (1951a). Biochem. et Biophys. Acta 6, 589. and Loofhourow, J. R. (1951b). Biochem. et Biophys. A d a Haas, W. J., Siaer, I. \I7., 6, 601. Hartridge, H. (1920-21). Proc. Physiol. Soc. 64, 128. Harris, D. T. (1926). Biochena. J . 20, 288. Harrison, G. It., Lord, R. C., and Loofbourow, J. R. (1948). Practical Spcctroscopy. Prentice-Hall, New York. Haurowitz, F., and Astrup, T. (1939). Nature 143, 118. Havinga, E. (1951). Personal communication. Hoch, H. (1950). Biochem. J . 46, 199. Hogness, T.R., Zscheile, F. P., and Sidwell, A. E., Jr. (1937). J . Phys. Chem. 41, 379. Holiday, E. R. (1936). Bzoehem. J. SO, 1795. Holiday, E. R. (1937). J . Sci. Instruments 14, 166. Holiday, E. R. (1950a). Zn Analytical Absorption Spcctroscopy. Edited by M. G. Mellon, Wiley, New York; Chapman and Hall, London. Holiday, E. R. (1950b). Faraday Society Discussion No. 9, p. 332. Holiday, E. R., and Irwin, J. 0. (1946). Quart. J. Pharm. Pharmacol. 19, 155.
ULTRAVIOLET ABSORPTION SPECTRA
385
Holiday, E. R., and Ogston, A. G. (1938). Biochem. J . 32, 1166. Hotchkiss, R. D. (1941). J . Biol.Chem. 141, 171. Hughes, W. L., Jr., and Straessle, R. (1950). J . Am. Chem. SOC.72, 452. Jonxis, J. H. P. (1949). I n Haemoglobin. Edited by F. J. W. Roughton, and J. C. Kendrew, Butterworth, London ; Interscience Publishers, New York. Jope, E. M. (1946). Brit. J . Znd. Med. 3, 136; see also Proc. Roy. Soc. Med. (1946). 39, 755. Jope, E. M. (1949). I n Haemoglobin. Edited by F. J. W. Roughton, and J. C. Kendrew, Butterworth, London; Interscience Publishers, New York. Jope, E. M., Jope, H. M., O’Brien, J. R. P. (1949). Nature 164, 622. Jope, H. M., and O’Brien, J. R. P. (1949). I n Haemoglobin. Edited by F. J. W. Roughton, and J. C. Kendrew Butterworth, London; Interscience Publishers, New York. Jordan, P. (1947). Eiweissmolekule. Stuttgart, 1947. Keilin, D., and Hartree, E. F. (1949). Nature 164,254. Ketelaar, J. A. A. (1950). Faraday Society Discussion KO. 9, p. 333. Kronenberger, A. (1930). 2. Physik. 63,494. Kunitz, M. (1946). J. Biol. Chem. 164, 563. Kunitz, M. (1947). J . Gem. Physiol. 30, 291 and 311. Larionov, L. P., and Brumberg, E. M. (1946). Compt. rend. acad. sci. U.R.S.S. 64, 267. Lavin, G. I., and Fruton, J. S. (1939). J . Bid. Chem. 130,375. Lavin, G. I., and Northrop, J. H. (1935). J . Am. Chem. Soc. 67,874. Lavin, G. I., and Stanley, W. M. (1937). J . Biol. Chem. 118,269. Lerner, A., and Barnum, C. P. (1946a). Arch. Biochem. 10,417. Lerner, A. B., and Barnum, C. P. (1946b). Arch. Biochem. 11, 505. Ley, H. (1920). 2. physik. Chem. 94,405. Ley, H., and Arends, B. (1932). 2. physzk. Chem. B17, 177. Liquori, A. M. (1951). Nature, 167,950. Loofbourow, J. R., Gould, B. S., and Sizer, I. W. (1949). Arch. Biochem. 22, 406. Loofbourow, J. R. (1951). Unpublished work. McLaren, A. D. (1949a). Acta Chem. S c a d 3, 648. McLaren, A. D’. (1949b). Advances in Enzymol. 9,75. Mason, H. S. (1948). J . Bid. Chem. 172,83. Mayneord, W.V., and Roc, E. M. F. (1937). Proc. Roy. Sac. London A169, 634. Mitchell, J. S., and Rideal, E. K. (1938). Proc. Roy. SOC.London, A167, 342. Northrop, J. H., Kunitz, M., and Herriott, R. M. (1948). Crystalline Enzymes. 2nd edition, Columbia Univ. Press, New York. Neurath, H., Greenstein, J. P., Putnam, F. W., and Erikson, J. 0. (1944). Chem. Revs. 34, 157. Oster, G. (1948). Chem. Revs. 43, 319. Pauling, L. (1942). The Nature of the Chemical Bond. Cornell Univ. Press, Ithaca, New York. Pauling, L., Corey, R. B., and Bransom, H. R. (1951). Proc. NatZ. Acad. Sci. U.S. 37, 205. Pauling, L., Itano, M. A., Singer, S. J., and Wells, I. C. (1949). Science 110, 543. Pauling, L., Itano, M. A., Wells, I. C., Schroeder, W. A,, Kay, L. M., Singer, S. J., and Lorey, R. B. (1950). Science 111, 459. Pauling, L., and Niemann, C. (1939). J . Am. Chem. Soe. 61, 1860. Perutz, M. F., and Mitchison, J. M. (1950). Nature 166,677. Perutz, M. F., Barer, R., Jope, E. M. (1951). Faraday Society Discussion No. 9, p. 423.
386
G . H. HICAVEN AND E. R . HOLIDAY
Philpot, J. S t . I,. (1936). (Quoted by Holiday 1936.) Platt, J. R., and Klevens, H. B. (1944). Revs. Modern Phys. 16, 182. Randall, J. T., and Brown, S. L. (1949). Nature 163, 209. Rideal, E. K., and Roberts, R. (1951). Proc. Roy. Soc. London A206, 391. Saidel, L. J., and Goldfarb, A. It. (1050). Science 114, 156. Saidel, L. J., and Goldfarh, A. It. (1‘351a). Federation Proc. 10, 241. Saidel, L.J., Goldfarh, A. It., and Mosovich, E. (1951h). b. Biol. Chem. 193, 397. Saidel, L. J., Goldfsrh, A. R., and K d t , IT.B. (1951~). Scieme 113, 683. Schauenstein, E. (1949a). Monatsh. 80, 820. Schauenstein, E. (194913). Monatsh. 80, 843. Schaucnstein, E. (1950). Faraday Society Discussion No. 9, p. 491. Schauenstein, E., Kratky, O., and Sekora, A. (1950). Nature 166, 1031. Schauenstein, E., and Treiher, E. (1950). J . Polymer Sci. 6, 145. Schauenstein, E., and Stanke, D. (1949). Monatsh. 80, 870. Schmidt, C. L. A. (Editor). (1938). The Chemistry of the Amino Acids and Proteins. Thomas, Springfield, Illinois, p. 217. Sehormullcr, J. (1949). Pharmazie 4, 105; Chem. Abstracts 43, 5665. Schroeder, W. A., Kay, I,. M., and Wells, I. C. (1950). J . Biol. Chem. 167, 221. Setlow, It. B., and Guild, W. R. (1951). Arch. Biochem. Biophys. 34,223. Shemin, D. (1045). J. Biol. Chrm. 169, 430. Shugar, D. (1951). Personal communication. Sinsheimer, R. L., Scott, J. F., Loofhourow, J. It. (1949). Nature 164, 796. Sinsheimer, R. L., Scott, J. F., and Loofhourow, J. R. (1950). J. Biol. Chem. 167, 299, 313. Sizer, I. W. (1946). 1.Biol. C h e w 163, 145. Sizer, I. W. (1947). J . Biol. Chem. 169, 303. Sizer, I. W., and Peacock, A. C. (1947). J . Bid. Cheni. 171, 767. Smith, $2. L., and Coy, N. €1. (1946). J . Biol. Chem. 164, 37. Smith, E. L., Green, R. D., Bartner, E. (1946). J . Biol. Chem. 164, 359. Smith, F. C. (1929). Proc. Roy. SOC.London B104, 198. Snellman, O.,SylvBn, B., and JulBn, C. (1951). Biochim. et Biophys. Acta 7, 96. Soret, J. L. (1883). Arch. sci. phys. nnt. 10, 429. and Moore, S. (1949). J . Biol. CI8ern. 178, 79. Stein, W., Stenstrom, W., and Reinhard, M. (1925). J . B i d . Chem. 66,819. Stokes, A. R., Seeds, W. E., and Oster G. (1950). Nature 166, 127. Tanford, C. (1950). J . Am. Chem. Sac. 72, 441. Thorell, B. (1947). Acta Med. Scand. Shppl cc. Treiher, I?., and Schauenstein, E. (1949). 2.f. Naturforsch. 4b, 252. Twyman, F. (1910). See Howe, H. E., Phys. Rev. 8, 674, 1916. Twyman, F., and Allsopp, C. B. (1934). The Practice of Absorption Spectrophotometry with Hilger Instruments. Hilger, 1934. Vandenbelt, J. M. (1946). Private communication. Vli?s, F., and Gex, M. (1934). Bull. inst. ocdanog. No. 658; Chem. Abstracts 29, 2184 (1935). Walker, P. M. B., and Davies, H. G. (1950). Faraday Society Discussion No. 9, p. 461. Whittaker, E., and Robinson, G. (1944). The Calculus of Observations. 4th ed. Blacltie & Sons. London & Glasgow. II’ilkins, M. H. F., Stokes, A. R., Seeds, W. E., and Ostcr, G. (1950). Nature 166, 127. Wrinch, D. (1937). Proc. IZoy. SOC.London A160, 59; A161, 505.
Author Index Numbers in parentheses are reference numbers and are included to assist in locating references in which the authors’ names are not mentioned in the text. Names in parentheses indicate coauthors. Numbers in italics indicate the page on which the reference is listed. Example: Abitz, W., 114(83), . . . 256, means that this author’s article is reference 83 on p. 114 and is listed on p. 156 at the end of the article.
A
Astrup, T., 360, 379, 384 Atkin, W. R., 127(13), 154
Abbott, C., 162, 247 B Abderhalden, E., 17,21, 24, 28,45,46,47, 68 Abitz, W., 114(83), 115(102), 117(83, Baba, H., 300, 302, 317 (see Mizushima) 102), 119, 148(102), 166 Bahn, A., 21, 24, 45, 46, 62 Bailey, K., 4, 6, 7, 62, 184, 194, 197, 198, Abribat, M., 139(178), 169 199, 200, 207, 208, 209, 210, 216, 218, Acher, R., 28, 62 223, 228, 231, 233, 235, 238, 239, Ackerman, D., 96(1), 164 247, 248, 252, 265(11, 12), 266(14), Adair, G. S., 208, 210, 262 (see Tsao)) 269(13), 276(7), 289 Agren, G., 47, 62 Baker, R. F., 240, 261 Alexander, P., 270, 288, 349, 382 Balenovic, K., 195, 215, 248 Allen, A. J., 322, 354, 388 Baly, E. C . C., 345, 383 Allsopp, C. B., 369, 386 Bamford, C. H., 280(3, 16), 289, 307, 315, Amberson, W., 231, 247 316 Ambrose, E. J., 129, 132, 166, 280(2, 3, 31), 281(2, 31), 283(31), 284(2, 31), Banga, I., 223, 224, 231, 248, 266, 289 288, 289, 305, 307, 308, 309, 312, 313, Barer, R., 312, 316, 321, 385 (see Perutz) 315, 316, 317, 363, 384 (see Elliott) Barkdoll, A. E., 338, 383 Barn&, D., 327, 383 (see Anslow) Anderson, K. J. I., 29, 66 Anslow, G. A., 327, 328, 352, 353, 354, Barnes, R. B., 293, 313, 316 (see Gore), 355, 358, 359, 360, 362, 364, 383 31r Barnum, C. P., 360, 379, 381, 385 Ardenne, M. v., 208, 210, 214, 247 Barron, E. S. G., 229, 861 Arends, B., 327, 385 Barry, G. T., 38, 39, 62, 63 (see Craig) Arnett, L. M., 109, 154 Bartner, E., 382, 386 (see Smith) Arnold, L. B., 334, 383 Bass, L. W., 24, 28, 65 Asaf, A. G., 147(4), 150, 164 Astbury, W. T., 78(12), 88, 91(8, 9, l l ) , Bastian, A., 95(63), 165 115, 116(10), 118, 120(9), 123(16), Bate-Smith, E. C., 189, 207, 238, 248 127(13, 15), 128(9, 14), 129, 139, 164, Bath, J., 296, 317 163, 171, 182, 194, 199, 200, 207, 210, Baudouy, C. T., 91(17, 72), 96(72), 137, 164, 166 211,212, 214, 230,239, 243, 247, 251 (see Perry), 254, 261, 262, 272(6), Bear, R. S., 79(18, 19, 148), 81(18, 148), 276(7), 278, 279(6), 284(6), 285,288, 84(18), 87(18), 88(18), 89(148), 889,304, 307, 308, S f 6 90(19, 189), 92(18), 99(18), 100, 387
388
AUTHOR INDEX
lOl(21, 22, 23), 102(22), 103(21), 104(22), 105(35), 106(22), 109(21), 115(21), 116(18), 117(21), 118(18, 23), 119(18), 124(18), 126(18), 144 (36), 145(23), 146(20, 36), 150(24, 25), 154, 155, 158, 194, 198, 199, 233, 239, 240, 248, 251 (see Schmitt) Beaven, G. H., 330, 335, 338, 339, 342, 346, 349, 350, 354, 358, 359, 360, 366, 375, 379, 583 Beckman, A. O., 323 (see Goldring), 384 Bednarczyk, W., 327, 585 Beeman, W. W., 103(114), 167 Bekker, J. G., 270(17), 289 Bell, F. O., 127(15), 128(14), 154 Bell, P. H., 6, 37, G2 Bendall, J. R., 189, 248 Bergmann, M., 16, 19, 22, 25, 28, 40, 45, 61, 62, 66 (see Stein), 140(26, 27), 141, 164 Bergold, G., 204, 206, 214, 248 Bettelheim, F. R., 266(14), 289 Beveridge, J. M. R., 95, 154 Bidwell, E., 77(29), 154 Billingham, R. E., 255(18), 289 Binkley, F., 229, 248 Bird, G. R., 312, 516 (see Blout) Biro, N. A., 165, 185, 186, 187, 223, 248 Biscoe, J., 78, 160 Biserte, G., 31, 62 Blackburn, S., 6, 62 Blake, C. H., 79(148), 81(148), 89(148), 158
Block, R. J., 32, 62, 93(31), 95(32), 155, 259, 289, 322, 355, 360, 585 Bloom, W., 70(149), 168 Blout, E. R., 309, 312, 316, 518 Boehm, G., 194, 248 Boeke, J., 259(21), 289 Boissonnas, R. A., 41, 62 Boldingh, J., 37, 62 Bolduan, 0. E. A,, 83, 100, lOl(21, 22, 23), 102(22), 103(21), 104(22), 105 (35), 106(22), l09(21), 115(21), 118(23), 144(36), 145(23), 146(36),
Borasky, R., 70(37), 83(164), 110, 111 (164), 112(164), 121(164), 126, 148, 155, 158
Borbiro, H., 164, 248 Borsook, H., 18, 67 Bowen, W., 221, 222, 250 (see Laki) Bowes, J. H., 93(38), 140, 143, 145, 147 (39), 155 Bowman, R. E., 43, 62 Brachet, J., 256(23), 259, 189, 368, 583 Brackett, F. S., 321, 384 Bradfield, J. R. G., 369, 383 Bragg, W. L., 132, 165, 307, 316 Brand, E., 5, 7, 24, 62, 67, 352, 377, 380, 383 Branson, H. R., 132(171), 168, 363, 365, 385 (see Pauling) Brattain, R. R., 299,317 (see Thompson), 370, 385 Brattsten, I., 31, 66 Braude, E. A., 337, 354, 359, 385 Brdicka, R., 330, 585 Bresler, S. E., 137, 165 Brigl, P., 17, 62 Brockmann, H., 45, 61 Brode, W. R., 337, 383 Brown, A., 137(185), 159 Brown, C. H., 91(45), 155 Brown, S. L., 335, 586 Bruigom, E. S., 327, 346, 347, 350, 583 Brumberg, E. M., 368, 586 Brust, M., 188, 251 Buchthal, F., 163, 164, 166, 173, 175, 176, 177, 183, 195, 207, 230, 239, 248 Burgermeister, E., 216, 248 Bull, H. B., 14, 28, 61 Bulliard, H., 255, 261(34), 270, 189 Bullough, W. S., 259(24), 289 Bunting, H., 70(204), 76(204), 169 Burchard, C., 96(1), 164 Buswell, A. M., 300, 301, 306, 311, 316 Butenandt, A., 321, 356, 583 Butler, J. A. V., 51, 52, 62, 65 C
16dt 155
Bolling, D., 93(31), 95(32), 156, 322, 355, 360, 385 Bon, W. F., 125, 139, 157 Bone, J. F., 6, 37, 62 (see Bell) Bonjour, G., 6, 25, 63 (see Desnuclle)
Callow, R. K., 345, 983 Campbell, E. D., 29, 64 (see Kaplan) Cannan, R. K., 32, 43, 63, 64 (see Keston), 352, 583 Carpenter, D. C., 29, 63
389
AUTHOR INDEX
Casal, A., 7, 21, 22, 28, 57, 63 Caspersson, T., 321, 383 Cecchini, L. F., 310, 311, 317 Centola, G., 91(46), 155 Champetier, G., 90, 91(48), 92, 123(47), 124(47), 155, 261, 289 Charlwood, P. A., 77(29), 1,54 Chhvremont, M., 256, 289 Cherbulisz, E., 123, 155 Chernikov, M. P., 137(52), 155 Chibnall, A. C., 10, 51, 63, 347, 377, 383 Chinn, B., 231, 247 (see Amherson) Christensen, €1. N., 4, 9, 19, 20, 21, 22, 63, 377, 384 Christensen, I,. K., 277(37), 280 Christian, W., 252 Cigada, M., 177, 248 Citterio, P., 177, 248 (see Cigada) Claesson, S., 38, 67 Clancey, V. J., 156 Clark, A., 221, 222, 250 (see Laki) Clark, G. L., 78, 92(55), 119, 155 Close, J. R., 175, 251 (see Ralston) Coblentz, W. W., 306, 3 i 7 Cohn, E. J., 143, 155, 277(26), 289 Cole, A. R. H., 312, 316 (see Barer) Cole, P. A,, 321, 384 Coleman, D., 46, 63 Colthup, N. B., 293, 316 Commoner, B., 321, 364, 384 Consden, R., 3, 6, 17, 20, 31, 33, 35, 36, 40, 41, 42, 43, 48, 49, 50, 54, 61, 63 Corey, R. B., 77(216), 78, 79(58), 119, 128(57), 130, 132(169, 170, 171), 137, 155, 158, 160, 314, 315, 316, 317, 363, 365, 384, 385 (see Pauling) Cori, C. F., 20, 66 Corycll, C. D., 323 (see Goldring), 384 Coulter, C. B., 330, 354, 384 Coy, N. I€., 382, 386 (see Smith) Craig, L. C., 38, 39, 62 (see Barry), 63, 64 (see Gregory) Crammer, J. L., 323, 346, 347, 348, 349, 350, 375, 384 Cravath, A. M., 370, 383 (see Brattain) Creeth, J. M., 51, 52, 62, 63 (see Butler) Cremer, H. D., 31, 6.9 Crepax, P., 188, 244, 248 Csapo, A,, 178, 232, 248 Custers, J. F. H., 361, 384
D Dainty, M., 228, 948 Dalgliesh, C. E., 61, 63, 304, 308, 316 (see Astbury) Dangl, J. R., 293, 300, 317 (see Randall) Darmon, S. E., 298, 299, 300, 301, 302, 303, 304, 307, 308, 312, 316 Davies, H. G., 368, 386 Davies, M., 354, 384 Davies, M. M., 208, 302, 317 Davis, H. F., 22, 23, 66 Davis, P. L., 39, 64 Dean, A. C. R., 327, 384 De Boer, J. H., 361, 384 (see Custers) De Groodt, A,, 259(21), 289 Dekker, C. A., 61, 63 Dempsey, E. W., 70(204), 76(204), 15,9 Derksen, J. C., 117(118), 125, 139, 167, 261(27, 28), 262, 289 de Robertis, E., 89(78), 156 de Rochemont, R. D., 123, 159 Desnuelle, P., 6, 7, 21, 22, 25, 28, 57, 69 Deuticke, H. J., 244, 248 Deutsch, A., 163, 164, 166, 173, 175, 176, 183, 207, 230, 248 (see Buchthal) Dickinson, S., 171, 182, 194, 2.47, 262, 276(7), 289 Dillon, R. J., 19, 67 (see van Slyke) Dippel, C. J., 361, 384 (see Custers) Dirr, K., 48, 63 Dodds, E. C., 51, 52, 62, 63 (see Butler) Donohue, J., 128(57), 155 Doty, P. M., 207, 251 (see Oster), 321, 357, 361, 384 Downing, J. R., 300, 301, 516 (see Buswell) Drake, B., 32, 33, 38,63, 67 (see Tiselius), 375, 384 Draper, M. H., 79(59), 155, 239, 240, 241, 243, 248 Drucker, B., 46, 63 Dubxiisson, M., 189, 194, 196, 198, 200, 201, 202, 203, 209, 220, 221, 222, 229, 231, 232, 233, 234, 235, 236, 237, 244, 245, 248, 249, 310, 317 (see Lecomte) Dunn, M. S., 28, 63, 327, 384 Duran-Reynals, F., 70(60), 155 Durrum, E. L., 31, 63
390
AUTHOR INDEX
d u Vigneaud, V., 39, 65 Dyer, W. J., 238, 649
E Earland, C., 270, 689, 349, 582 Ebner, V. von, 98, 155, 171, 172, 175, 249 Eckstein, H. C., 259, 261 (29), 689 Edman, P., 8, 42, 63 Edsall, J. T., 143, 155, 193, 207, 210, 218, 649, 250, 265(30), 277(26), 289, 297, 298, S l 7 , 321, 357, 361, 380, 381, S84 Edwards, B. G., 377, 378, S84 Edwards, R. R., 377, 384 Eeg-Larsen, N., 57, 63 Eisen, H. N., 382, 384 Elliott, A., 129, 132, 155, 280(2, 3, 31), 281(2, 31), 283(31), 284(2, 31), 288, 689, 305, 307, 308, 309, 312, 313, 315, 316, S17, 363, 384 Ellis, J. W., 296, 317 filod, E., 123(62), 155 Elsden, S. R., 37, 63 Enenkel, H. J., 31, 67 Engeland, R., 95(63), 155 Engelhardt, W. A., 176,194,222,224,949 Engelmann, T. W., 125, 166 English, J. P., 6, 37, 62 (see Bell) Engstrom, A , , 258(32, 33), 289 Erdos, T., 189, 195, 196, 199, 200, 201, 202, 203, 204, 208, 209, 210, 211, 214, 216, 217, 226, 228, 230, 231, 234, 244, $49 (see Amberson), ,951 (see Snellman) Erikson, J. O., 329 (see Neurath), 348, 985
FeIix, K., 7, 48, 63, 66 Fellows, C. E., 6, 37, 66 (see Bell) Fenn, W. O., 649 Ferri, C., 122(155), 168 Feuer, G., 200, 219, 220, 221, 222, 249, ,953
Finkelstein, P., 350, 584 Finogenov, P. A., 137(44), 166 Fischer, E., 31, 41, 45,48,63,67, 175, 242, 949 Fischer, H., 7, 48, 63 (see Felix) Fixl, J. O., 355, 384 Fodor, A., 29, 6 s Fodor, P. J., 29, 63 Foreman, E. M., 36, 65 (see Patton) Foster, G. I., 43, 66 Foster, M. L., 327, 328, 353, 355, S83 Fowden, L., 41, 69 Fowler, R. G., 293, 300,517 (see Randall) Fox, M., 349, 58s (see Alexander) Fox, S. W., 4, 5, 64 Fraenkel, G., 92, 166 Fraenkel-Conrat, H., 16, 62, 349, 384 Franchi, C. M., 89(78), 156 Frederic, J., 256, 289 French, H. V., 238, 649 Frenkel, S. Y., 137(44), 165 Freudenderg, K., 13, 17, 64 (see Kuhn) Freymann, R., 300, 317 Freyss, G., 46, 64 (see Goldschmidt) Frey-Wyssling, A., 97(79), 99(79), 156 Friedrich-Ficksa, H., 321, 356, 583 (see Butenandt) Fromageot, C., 10, 28, 51, 62 (see Acher), 64, 327, 328, 345, 346, 349, 375, 377, 384
Ettisch, G., 97(65), 156 Evans, E. A., 5, 64 Ewald, A,, 124, 166 Ewbank, E. K., 345, 583 Eyring, IT., 121(67), 166
F Fabry-Hamoir, C., 201,229,231, 237,249 Farrant, J. L., 79(68), 1.56 FaurB-Fremiet, E., 77(69), 90, 91(48, 70, 72, 73, 74), 92, 96(72), 123(47, 69, 75), 124(47, 69), 137, 155, 156 Feinstein, B., 175, 2661 (see Ralston) Feitelberg, S., 84, 166
Fruton, J. S., 16, 25, 61, 62, 69 (see Dekker), 64 (see Johnston), 338, 354, S84, 385 (see Lavin) Fugitt, C. H., 23, 66 Fujioka, O., 300, S17 (see Mizushima) Fuld, M., 47, 65 (see Meyer) Fuson, N., 293, 300, S l 7 (see Randall)
G Garrault, H., 91(73, 74), 95(80), 166 Garrod, M., 123(141), 158(142), 167, 168 Gasser, H., 168, 175, 177, 849 Geiger, W. B., 257, 289 Gelotte, B., 198, 208, 211, 217, 661
391
AUTHOR INDEX
Geren, B. B., 83(190), 159 GerendBs, M., 163, 237, 242, 249 Gergely, J., 310, 311, 317 (see Morales) Gerngross, O., 78(82, 120), 114(83), 115 (102), 117(83, 102), 119, 124, 148 (102), 156, 157 Gel;, M., 368, 386 Gibson, R. G., 322, 354, 382 (see Allen) Giese, A. C., 349, 384 Giroud, A., 255, 261(34, 36), 270, 289 Gitlin, D., 382, 384 Godeaux, J., 184, 249 Gopfert, H., 162, 251 Goldfarb, A. R., 364, 365, 366, 367, 386 (see Saidel) Goldring, L. S., 323, 384 Goldschmidt, S., 46, 64 Goldstein, M., 308, 317 Goldwater, W. H., 352, 383 (see Brand) Gonnell, H. W., 78(104), 88(104), 89 (104), 115, 156, 157 Goodwin, T. W., 350, 372, 374, 375, 376, 377, 384 Gordon, A. H., 3, 6, 17, 18, 19, 20, 21, 28, 30, 31, 33, 35, 36, 40, 41, 42, 43, 48, 49, 50, 54, 59, 61, 63 (see Consden),
Gruen, D. M., 298, 311, 317 (see Klotz) Griining, W., 173, 252 Cuba, F., 178, 215, 218, 238, 249 Guild, W. R., 366, 367, 386 Guntelberg, A. V., 57, 64 Gustavson, K. H., 144, 145(92, 93), 147 (93), 156 Gutfreund, H., 50, 64 Gutfreund, T. H., 208, 210, 248 (see Bailey), 269(13), 289 Guyon, L., 77(162), 137(94), 156, 158
H
Haas, R. H., 147(4), 150(4), 154 Haas, W. J., 351, 352, S84 Hagdahl, I,., 32, 34, 67 (see Tiselius) Hahn, J. W., 14, 28, 62 Hakala, M., 18, 67 (see Virtanen) Halford, S., 308, 317 Hall, C. E., 71(192), 79(96, 97), 83(192), 87, 90(189), llO(192, 193), 111, 112 (192), 120(192), 121(192), 137(192), 139(193), 150(97), 156,159, 194, 198, 199, 208, 211, 214,233,239, 240, 241, 249, 250, 251 (see Schmitt) Halla, F., 156 64 Hamberg, V., 23, 67 Gordon, M., 35, 65 (see Mitchell) Gore, R. C., 293, 299, 311, 313, 316 (see Hamilton, P., 19, 67 (see van Slyke) Hamoir, G., 177, 196, 197, 198, 200, 201, Barnes), 317 209, 249 Gould, B. S., 322, 327, 328, 3G0, 385 (see Hamoir, G. C. M., 35, 64 Loofhourow) Hanby, W. L., 4, 61, 64, 280(3, 16), 288, Graham, C. E., 156 289, 305, 307, 315, 516 Grant, R. I,., 46, 64 Hannig, K., 31, 64 Grassmann, W., 31, 64, 140, 156 Happey, F., 280(16), 289, 307, S16 Green, F. C., 7, 64 Hare, G. H., 323, 384 (see Goldring) Grcen, M., 342, 383 Harington, C. R., 25, 64 Green, N. M., 7 , 64 Harris, D. T., 322, 384 Green, R. D., 382, 986 (see Smith) Greenstein, J. P., 210, 249, 329, 348, 385 Harris, E. J., 165, 249 Harris, M., 257, 289 (see Neurath) Gregory, J. D., 38, 39, 62 (see Barry), Harrison, G. R., 365, S84 Hartrce, E. F., 335, 336, 385 63 (see Craig), 64 Hartridge, H., 334, 384 Grey, D. S., 312, 316 (see Blout) Hartwig, S., 321, 356, 383 (see ButenGridgeman, N. T., 323, 384 andt) Griswold, P., 307, 311, 317 (see Klota) Haskins, F. A., 35, 37, 65 (see Mitchell) Gross, H., 214, 251 Gross, J., 73,83(90, 191), 87(89), 110, 111 H a s , G. M., 232, 261 (90, 191), 112(90, l n l ) , 118(90), 120, Hasselbach, C., 123(156), 158 121(191), 136(90, 191), 137(191), 138 Hasselbach, W., 165, 185, 217, 231, 232, 234, 237, 238, 242, 249 (110), 145, 156, 157, 159
392
AUTHOR INDEX
Haugaard, G., 15, 31, 64 Haurowitz, F., 360, 379, 384 Havinga, E., 350, 384 Hawes, R. C., 323, 384 (see Goldring) Haxton, H. A., 175, 249 Hegsted, D. M., 19, 21, 22, 63 Heinta, E., 306, 318 Heinz, E., 165, 179, 185, 191, 248 Helleman, L., 24, 64 Heringa, G. C., 76(100), 78(99), 87, 88,
Huggins, M. I,., 129, 157 Hughes, W. L., Jr., 347, 385
I Ingram, V. M., 64 Inman, V. T., 175, 261 (see Ralston) Irwin, J. O., 323, 384 Itano, M. A,, 341, 386 (see Pauling)
156, 261(27, 28), 262, 289
HBrion, A., 188, 244, 248 Hermann, L., 121, 156 Hermann, V. S.,230, 249 Herriott, R. M., 25, 65 (see Northrop), 342, 377, 385 (see Northrop) Herrmann, K., 114(83), 115(102), 117(83, 102), 119, 148, 156 Herzog, R. O., 78(104), 88(104), 89(104), 114(106), 115, 146, 157 Hess, K., 109, 139, 157
Hier, S. W., 156 IIighberger, J. H., 87, 120(109), 138, 146 ( l l l ) , 157 Hill, A. V., 162, 168, 175, 176, 177, 178, 189, 249, 250 Hill, D. K., 162, 250 Hoch, H., 339, 341, 383 (see Beaven), 384 Hodge, A. J., 79(59), 155, 239, 240, 241, 243, 248 Hoffmann-Berling, H., 239, 240, 242, 250 Hogness, T. R., 366, 384 Holiday, E. R., 323, 326, 330, 335, 338, 339, 346, 349, 350, 354, 358, 359, 360,366, 369, 372, 375, 376, 377, 379, 383 (see Beaven), 984 Hollwede, W., 200, 250 Holton, F., 165, 179, 185, 191, 249 Ilolz, B., 239, 250 Homiller, R. P., 27, 67 Horwitt, M. K., 95(32), 155 Hotchkiss, R. D., 377, 384 Howard, G. A., 37, 64 Howard, K. S., 6, 37, 62 (see Bell) Howitt, I. O., 46, 63 Huang, H. T., 47, 64 Hudson, R. F., 349, 383 (see Alexander) Hiirthle, K., 239, 260 Huffman, H. M., 15, 64 Huggins, M., 307, 315, 317
J Jacob, J., 163, 244, 248 (see Crepax), 260 Jacobsen, C. F., 18, 64, 277(37), 289 Jaisle, F., 195, 212, 215, 216, 217, 218, 234, 250
M. A,, 71(192), 79(96, 97), 83(192), 87, 90(189), llO(192, 193), 111, 112(192), 120(192), 121(192), 137(192), 139(193), 150(97), 156, 159, 194, 198, 199, 208, 211, 214, 233, 239, 240, 241, 249 (sec Hall), 250,
Jakus,
$51 (see Schmitt) Jancke, W., 114(106), 157 Jeannerat, J., 123(51), 156 Jeener, R., 256(23), 289 Jennison, M. IT.,77(113), 11;7 Jensen, H., 5, 64 Johnson, A. W., 61, 63 (see Dalgliesh) Johnson, P., 195, 203, 208, 211, 213, 216,
217, 226, 250
Johnston, R. B., 16, 64 Jones, T. S. G., 35, 36, 37, 42, 64 Jonxis, J. H. P., 341, 385 Jope, E. M., 321, 330, 335, 338, 339, 340, 346, 348, 354, 359, 360, 379, 983 (see Beaven), 385 (see Perutz) Jope, H. M., 341, 348, 385 Jordan, P., 377, 385 Jordan, W. K., 194, 216, 226, 250 Josenhans, W., 168, 175, 260 Josepovits, G., 230, 249 JulBn, C., 371, 386 (see Snellman) Jutisz, M., 10, 18, 32, 33, 34, 51, 62 (see Acher), 64
K Kabat, E. A., 330, 354, 384 (sce Coulter) Kaesberg, P., 103, 157
393
AUTHOR INDEX
Kalt, W. B., 364, 365, 366, 386 (see Saidel) Kamp, F., 198, 244, 250 Kanagy, J. R., 146(115), 157 Kaplan, E. H., 29, 64 Kassell, B., 5, 24, 62, 352, 377, 380, 383 (see Brand) Katchalski, E., 304, 307, 312, 317 Kata, J. R., 78(82, 120), 115(117), 117 (117, 118), 124, 125, 139, 147(122), 148(123), 156, 157
Kaumans, H., 226, 250 Kaunita, P. E., 84, 156 Kausche, G. A., 239, 240, 242, 250 Kay, L. M., 339, 341, 385 (see Pauling), 386 (see Schroeder) Keil, B., 31, 64 (see Gordon) Keilin, D., 334, 335, 336, 385 Kendrew, J. C., 132(43), 155, 307, 316 (see Bragg) Kenten, R. H., 93(38), 140, 143, 145, I
147(39), 155
Kerkkonen, H., 18, 67 (see Virtanen) Kersten, H. J., 146(111), 157 Keston, A. S., 43, 44, 64 Ketelaar, J. A. A., 354, 385 Kibrick, A. C., 352, 383 Kies, M. W., 39, 64 Kiessig, H., 109, 157 King, A. T., 270(17), 289 Kistiakowsky, G. B., 334, 383 Kleinaeller, A., 228, 248 (see Dainty), 250 Klemm, O., 47, 65 (see Meycr) Klemperer, P., 70, 157 Klevens, H. B., 353, 354, 364, 385 Klingler, C., 353, 383 (see Anslow) Klota, J. M., 298, 307, 311, 317 Knappeis, G., 163, 164, 166, 173, 175, 176, 183, 207, 230, 239, 248 (see Buchthal) , 250 Koisi, S., 115(125), 157 Kolkmeijer, N. H., 78(99), 166 Kolpak, H., 88, 157 Komm, E., 17, 62 Korey, S., 165, 166, 168, 179, 184, 186, 250
Kratky, O., 78, 79(328), 99(128), 116 (127, 128), 141, 156, 157, 355, 357, 384 (see Fixl), 386 (see Schauenstein) Krebs, K. F., 306, 316 (see Buswell) Krekels, A,, 7, 48, 63 (see Felix)
Kronenberger, A., 331, 334, 685 Kroner, T. D., 31, 64 Kuchel, C. C., 238, 251 Kuntacl, A,, 71, 98, 99, 114(133), 115 (133), 117(133), 146(132), 148, 157
Kuhn, W., 13, 17, 64, 138, 157, 181, 250 Kuk-meiri, S., 29, 63 Kunita, M., 25, 65 (see Northrop), 342, 377, 380, 385 (see Northrop) Kuratani, K., 300, 302, 317 (see Miaushima) Kuschinsky, G., 184, 192, 218, 221, 228, 229, 250, 252
L Laakonen, T., 18, 67 (see Virtanen) Lafon, M., 96(183), 159 Lajos, I., 219, ,952 (see Straub) Lajtha, A., 207, 250 Laki, K., 221, 222, 250, 310, 311, 317 (see Morales) Landolt, R., 195, 203, 208, 211, 213, 216, 217, 226, 250
Lang, A., 32, 69 Larionov, L. P . , 368, 385 Lavin, G. I., 330,335,338,354, S84,386 Lawrence, A. S. C., 194,225,228,248(see Dainty), 250 (see Needham) Leblond, C. P., 255, 261(34), 270, 289 Lecomte, J., 300, 310, 317 Lederer, E., 32, 33, 34, 64 Lelli, G., 121, 157 Lennox, F. G., 125(135), 157 Lenormant, H., 300, 303, 317 Lens, J., 10, 51, 64 Leonis, J., 9, 65 Leplat, G., 137, 157 Lerner, A,, 360, 379, 381, 385 Levene, P. A., 17, 20, 24, 28, 66 Levin, A., 168, 250 Levy, A. L., 9, 46, 47, 65 Levy, B., 29, 65 (see Mandl) Levy, M., 44. 64 (see Keston) Lewis, H. B., 46, 64, 261(59), 290 Ley, H., 327, 345, 385 Lillie, R. D., 76(137), 157 LinderstrZm-Lang, K., 25, 26, 57, 63 (see Eeg-Larsen), 65, 67 (see Villee) Lindhard, J., 239. 849 (see Buchthal) Lindley H., 270(38), 289
394
AUTHOR INDEX
Lindstrom, B., 258(32, 33), 289 Lipken, D., 321, 384 Liquori, A. M., 341, 386 Livermore, A. H., 39, 66 Ljubimova, M. N., 176, 194, 222, 249 Lloyd, 1). J., 70, 96, 123(141), 140(142), 146(140), 147(144, 145), 148(144), 150(143, 146), 167, 158 Lomax, R., 118, 123(16), 164 Longley, J. B., 264, 289 Loofbourow, J. R., 322, 327, 328, 335, 336, 339,351,352, 355,360,365, 384 (see Haas), 385 Lorand, L., 265, 266, 289 Lord, R. C., 365, 384 Lorey, R. B., 341, 385 (see Pauling) Lowthee, A. G., 6, 62 Lucas, C. C., 95, 154 Ludes, H., 231, 247 (see Amherson) Lugg, J. W. H., 24, 6c5 Lundi, G., 239, 250 Lundsgaard, E., 189, 250 Lyman, E. R., 353, 354, 359, 383
M MacArthur, I., 79(147), 158, 239, 250 MacFadyen, D. A., 19, 67 (see van Slyke) McLaren, A. D., 29, 64 (see Kaplan), 65, 322, 350, 359, 367, 384, 585 McLean, D. J., 349, 384 Mager, A., 48, 6 3 Magill, M. A., 322, 354, 582 (see Allen) Mahn, H., 17, 28, 62 Mandl, L., 29, 65 Marchlewski, L., 327, 383 Mark, H., 128(157), lti8 Marks, M. H., 79(148), 81(148), 89(148), 158 Marotta, V., 121, 157 Marriott, R. H., 147(144, 145), 148(144), 150(143), 158 Martin, A. J. P., 2, 3, 6, 17, 18, 19, 20, 21, 28, 30, 31, 32, 33, 34, 35, 36, 37, 38, 40, 41, 42, 43, 48,49, 54, 59, 61, 63 (see Consden), 64:(seeyGordon), 66, 66 Marwick, T. C., 261(5), $88 Mason, H. S., 351, S85 Mathieu, L., 200, 221, 222, 245, 849
Matoltsy, A. G., 237, 242, ,049 Maximov, A. A., 70(149), 168 Mayneord, W. V., 334, 385 Medawar, P. B., 255(18), 289 Mehl, J. W., 218, 249 Meibohm, E. P. H., 109(3), 164 Meister, A., 185, 230, 651, 252 Mellors, R. C., 312, 316 Menne, F., 230, 250 Menten, M. L., 250 Mercer, E. H., 79(68), 156, 269(41), 289 Meyer, D., 10, 51, 64 (see Fromageot) Meyer, K., 198, 238, 252 Meyer, Karl, 73, 158 Meyer, K. H., 47, 65, 73, 78(154), 115 (152), 122(155), 123(51, 156) 128 (157), 158, 181, 250 Meyerhof, O., 162, 165, 223, 225, 250 Miall, M., 228, 248 (see Dainty) Michaelis, L., 182, 260 Middlebrook, W.R., 6, 50, 65, 256(42), 266(14), 289 Miekeley, A., 22, 62 MihBlyi, E., 265(43), 289 Miller, E. G., Jr., 169 Miller, G. L., 29, 65 Mirsky, A. E., 121(159), 158 Mitchell, H. K., 35, 37, 65 Mitchell, J. S., 29, 66, 366, 386 Mitchison, J. M., 342, 364, 386 (see Perutz) Mittelmann, R., 41, 65 Mizell, L. R., 257, 289 Mizushirna, S., 300, 302, 307, 31 7 Molnar, F., 219, 220, 221, 249 (see Fcuer) Molster, C. C., 13, 17, 64 (see Kuhn) Mommaerts, W. F. H. M., 185, 195, 198, 203, 207, 208, 209, 211, 215, 216, 221, 223, 224, 226, 228, 229, 234, 250 Monnier, A. M., 310, 517 Monnier, R., 6, 18, 65 Montroll, E. W., 13, 66 Moore, S., 19, 28, 33, 37, 38, 40, 45, 66 (see Stein), 377, 386 Morales, M. I., 310, 311, 317 Moring-Claesson, I., 34, 35, 66 Morton, R. A., 350, 372, 374, 375, 376, 377, 384 Mosovich, E., 365, 366, 367, 386 (see Saidel) Miiller, E., 96(1), 164
395
AUTHOR INDEX
Munk-Peterson, A., 175, 207, 848 (see Buchthal) Muralt, A. von, 171, 193, 243, 250 Muschter, F. J. F., Jr., 147(122), 157 Mustacchi, P. O., 130, 158 Mycek, M. J., 16, 64 (see Johnston) Myrback, K., 13, 20, 66
N Nagakura, S., 300, 302, 317 (see Mieushima) Nageotte, J., 77(162), 137, 158 Nakashima, R., 61, 65 Nassar, S. C., 352, 354, 360, 383 Needham, D., 194, 223,225,228, 248 (see Dainty), 260 (see Needham, J.) Needham, J., 194, 225, 228, 248 (see Dainty), 260 Neuberger, A., 323, 346, 347, 348, 349, 350, 375, 384
Neuman, R. E., 93, 95, 168 Neurath, H., 9, 25, 65, 277(44), 289, 329, 348, 385
Nicholson, D. L., 298, 317 (see Thompson) Niemann, C., 3, 47, 61, 66, 64, 65, 140(26), 141, 164, 320, 364, 385
Noll, D., 242, 251 Nolte, A., 259, 689 Northrop, J. H., 25, 65, 330, 335, 342, 377, 386 Nutting, G. C., 83(164), 110, 111(164), 112(164), 121(164), 126, 148, 168
0 O'Brien, J. R. P., 341, 348, 385 Odier, M. E., 73(158), 158 Ogsten, A. G., 208, 210, 248 (see Bailey), 269(13), 289, 369, 375, 376, 377, 584
Oloffson, B., 269(41), 289 Oncley, J. L., 137(185), 169 Orekhovich, K. D., 137(165), 158 Orekhovich, V. N., 137, 158 Osaki, M., 17, 65 Osborne, T. B., 3, 67 Oster, G., 194, 216, 207, 226, 260, 251, 321, 361, 385, 586 (see Stokes) Ottesen, M., 26, 57, 63 (see Eeg-Larsen), 66, 67 (see Villee) Otvos, J. W., 297, 317 (see Edsall)
P Palmer, A. H., 352, 58.5 Pappenheimer, A. M., 26, 65 Parker, E. A., 78, 165 Parrish, R. G., 198, 203, 208, 209, 216, 234, 250
Partridge, S. M., 6, 22, 23, 33, 34, 36, 66, 76, 158
Patterson, W. I., 257, 289 Patton, A. R., 36, 65 Pauling, L., 3, 65, 121(159), 130, 132(167, 169, 170j 171), 150(168), 158, 314, 316, 317, 320, 341, 345, 363, 364, 365, 385 Pavlik, M., 330, 385 Peacock, A. C., 345, 360, 362, 379, 386 Pease, D. C., 240, 251 PBnasse, L., 6, 10, 51, 64 (see Fromageot), 65 (see Monnier) Penney, J. R., 41, 6s Perlmann, G. E., 57, 65 Perry, S.V., 163, 184, 199, 200, 207, 211, 212, 214, 216, 218, 230, 247 (see Astbury ) ,248,251 Perutz, M. F., 132(43), 155,277(47), 289, 307, 316 (see Bragg), 316, 321, 342, 364, 585 Petermann, M. L., 26, 65 Petersen, A., 163, 164, 166, 173, 176, 183, 230, 248 (see Buchthal) Petersen, E., 313, 317 (see Gore) Petersen, E. M., 299, 317 Pettko, E., 219, 220, 221, 249 (see Feuer) Pfeiffer, H. H., 97, 98, 158 Phillips, D. M. P., 36, 38, 51, 52, 62, 63 (see Butler), 66 Phillips, H., 256(42), 689 Philpot, J. St. L., 372, 385 Picken, L. E. R., 91(174), 123(173), 168 Piersma, H. D., 377, 384 Pirie, A,, 93, 169 Pittom, W. W. P., 28, 66 Pitt-Rivers, R., 25, 64 Platt, J. R., 353, 354, 364, 986 Pleass, W. B., 147(145), 158 Plotnikova, N. E., 137(165, 176), 158,159 Polis, D., 223, 225, 261 Polissar, M. J., 175, 261 (see Ralston) Polson, A., 41, 66 Porter, K. R., 111, 169
396
AUTHOR INDEX
Porter, R. R., 5, 6, 7, 47, 57, 58, 59, 66 Portzehl, H., 164, 165, 166, 167, 168, 169, 173, 175, 176, 179, 180, 189, 190, 191, 192, 195, 198, 203, 204, 205, 206, 207, 208, 211, 213, 214, 215, 216, 217, 218, 226, 230, 234, 235, 236, 238, 248, 651 Post, O . , 39, 63 Pouradier, J., 136, 139, 158 Prakke, F., 114(133), 115(133), 117(133), 138(133), 157 Primosigh, J., 32, 34, 66 Proctor, H. R., 147, 159 Pryor, M. G. M., 91(173), 151(181), 168, 159 Pupko, S. L., 120(219), 160 Purves, C. B., 147(4), 150(4), 154 Putnam, F. W., 329, 348, 385 (see Neurath)
R Ralston, H. Y., 175, 251 Ramsey, R. W., 175, 176, 178, 246, 251 Randall, H. M., 293, 295, 299, 300, 317 (see Thompson) Randall, J. T., 335, 385 Randall, S. S., 38, 66 Ranzi, S., 177, 248 (see Cigada) Rapport, M. M., 73(151), 168 Rasmussen, R. S., 299, S f 7 (see Thompson), 370, 383 (see Brattain) Rauen, H. M., 48, 63, 66 Reay, G. A., 238, 261 Reed, R., 91(182), 92(182), 159, 163, 199, 200, 207, 210, 211, 212,214,230, 247 (see Astbury), 261 (see Perry) Rees, A. L. G., 79(68), 166 Rees, M. W., 10, 18, 51, 63, 66 Reinhard, M., 323, 345, 369, 386 Rich, A., 297, 51 7 (see Edsall) Richards, R. E., 300, 302, 317 Richter, M., 32, 67 Rideal, E. K., 29, 66, 366, 367, 585 Riederle, K., 140, 156 Ripa, O., 270(48), 289 Ritchie, J. M., 162, 247 Ritland, H. N., 103(114), 157 Rittenberg, D., 43, 66 Roberts, It., 366, 367, 385 Roberts, R. M., 15, 64 Robinson, C., 309, 315, 517 (see Elliott)
Robinson, G., 371, 586 Roc, E. M. F., 334, 585 Rochc, J., 96(183), 169 Rockland, L. B., 327, 384 Rodebush, W. H., 300, 301, 306, 316 (see Buswell) Rogers, M. M., 6, 37, 62 (see Bell) Roman, J., 136(179), 139(179), 169 Rometsch, It., 39, 66 Ross, W. F., 338, 383 Roth, E., 178, 197, 198, 251 Rothen, A., 17, 20, 65 (see Levene) Rovery, M., 6, 25, 63 (see Desnuelle) Roy, M. F., 300, 516 (see Buswell) Rozsa, G., 199, 200, 208, 211, 212, 213, 231, 239, 240, 241, 251 Rudall, K. M., 91(182), 92(182), 159, 261(50), 262(49), 264(50), 269(49), 271(49), 274(49), 276(49), 277(49), 289, 290, 312, 316 Rugo, H. J., 150(24), 154 Ryan, F. J., 352, 385 (see Brand) Rydon, H. N., 4, 61, 64 S Saidel, L. J., 352, 364, 365, 366, 367, 377, 380, 383 (see Brand), 386 Salo, T. P., 100(23), 101(23), 118(23), 137, 144(36), 145(23), 146(36), 154, 155, 159 Samsa, E. G., 150(25), 154 Sandow, A., 162, 163, 188, 261 Sanger, F., 4, 5, 6, 7, 16, 20, 22, 24, 26, 27, 30, 31, 34, 39, 40, 47, 51, 52, 53, 55, 56, 59, 61, 66 Saum, A. M., 277(44), 289 Scatchard, G., 137, 159 Schaad, J. A., 78, 92(55), 119, 166 Schilfer, H., 162, 251 Schauenstein, E., 216, 248, 261, 321, 348, 355, 356,357, 358, 359, 361,362, 363, 377, 384 (see Fixl), 386 Scheibe, G., 321,356,383 (see Butenandt) Scheinberg, H., 297, 317 (see Edsall) Scherp, H. W., 260(51), 290 Scherrer, P., 78(186), 159 Schick, A. F., 232, 251 Schmid, K., 7 , 66 Schmidt, C. L. A., 354, 586 Schmidt, G., 93(175), 159
AUTHOR INDEX
Schmidt, W. J., 97(188), 98(187), 99 (188), 159, 241, 251 Schmitt, F. O., 71, 79(96, 97), 83(90, 190, 191, 192), 87, 90(189), 110, 111(90, 191), 112(90, 191, 192), 118(90), 120, 121(191, 192), 136(90, 191), 137(191, 192), 138(110), 139(193), 145, 150 (97), 166, 157,159, 194, 198, 199, 214 233, 239, 249 (see Hall), 661 Schneider, G., 165, 217, 231, 232, 234, 237, 238, 242, 249 Schnek, G., 327, 328, 345, 346, 349, 375, 377, 384 Schormuller, J., 386 Schram, R., 304, 318 Schramm, G., 32, 34, 66, 166, 194, 195, 198, 203, 204, 206, 207, 208, 211, 213, 214, 215, 216, 217, 218, 226, 230, 234, 235, 236, 238, 248,251 (see Portzehl) Schroeder, W. A., 7, 64, 339, 341, 386 (see Pauling), 386 Schuberth, H., 48, 63 Schulz, G. V., 206, 211, 251 Schwert, G. W., 9, 25, 65 Scott, J. F., 335, 386 (see Sinsheimer) Sebesta, K., 31, 64 (see Gordon) Seeds, W. E., 321, 364, 386 (see Stokes and Wilkins) Seiberlick, J., 147(122), 157 Sekora, A., 79(128), 99(128), 116(128), 141(128), 157,357,386 (see Schauenstein) Seldeslachts, J., 244, 248 (see Crepax) Seraidarian, K., 223, 224, 250 Setlow, R. B., 366, 367, 386 Seymour-Jones, F. L., 77(200), 159 Shaffrath, M. W., 175, 251 (see Ralston) Shemin, D., 380, 386 Shepard, C. C., 35, 66 Shephard, R. G., 6, 37, 62 (see Bell) Shih-Chang-Shen, 194, 225, 228, 248 (see Dainty), 250 (see Needham) Short, N. L., 298, 317 (see Thompson) Shugar, D., 350, 386 Sichel, F. J., 177, 251 Sidwell, A. E., Jr., 366, 384 (see Hogness) Siegrist, A. E., 73(158), 158 Signer, R., 214, 251 Simanouti, T., 300, 302, 307, 317 (see Mizushima) Simha, R., 13, 65, 211, 251
397
Singer, S. J., 341, 385 (see Pauling) Singer, T. P., 229, 251 Singher, H. O., 185, 251 Sinsheimer, R. L., 335, 586 Sizer, I. W., 77(194, 195), 159, 322, 327, 328, 345, 351, 352, 355, 360, 362, 379, 384 (see Haas), 385 (see Loofbourow), 386 Slobodiansky, E., 46, 47, 65 Smith, A. F., 109(3), 164 Smith, E. L., 382, 386 Smith, F. C., 325, 381, 386 Smith, S. G., 46, 63 Snellman, O., 195, 196, 198, 199, 200, 201, 202, 203, 204, 208, 209, 210, 211, 214, 216, 225, 226, 228, 230, 234, 249, 261, 371, 386 Snow, J. M., 238, 249 Snyder, I. W., 180, 252 Soret, J. L., 320, 386 Spark, L. C., 163, 199, 200, 207, 210, 211, 212, 214, 230, 247 (see Astbury), 251 (see Perry) Speakman, J. B., 270(48), 289 Spicer, S., 218, 252 Stafford, R. W., 293, 316 (see Barnes) Stair, R., 306, 317 Stanke, D., 355, 386 Stanley, W. M., 354, 385 (see Lavin) Staudinger, M., 199, 231, 251 Stearn, A. E., 121(67), 156 Steiger, R. E., 17, 20, 65 (see Levene), 322, 354, 382 (see Allen) Stein, W., 377, 586 Stein, W. H., 19, 28, 33, 37, 38, 40, 45, 65, 66, 140(27), 154, 159 Steiner, R. F., 361, 384 Steinhardt, J., 23, 25, 66, 277(52), 290 Stenstriim, W., 323, 345, 369, 386 Stephen, J. M. L., 51, 52, 62, 63 (see ButIer) Stickney, M. E., 323 (see Goldring), 384 Stock, C. C., 24, 64 Stockall, G., 150(146), 168 Stover, R., 210, 252 Stokes, A. R., 321, 364,386 (see Wilkins) Stone, D., 61, 65 (see Dekker) Stone, F. M., 330, 354, 384 (see Coulter) Straessle, R., 347, 385 Strait, L. A., 175, 651 (see Ralston)
398
AUTHOR INDEX
Straub, F. B., 166,194, 195, 197, 198, 199, 200, 208, 211, 212, 215, 216, 217, 219, 220, 221, 222, 228, 232, 233, 236, 237, 248, 249 (see Feuer), 251, 265(53), 290
Strauss, K., 46, 64 (see Goldschmidt) Street, A., 261(4), 288, 307, 316 Street, S. F., 176, 178, 251 Strobel, G., 172, 175, 252 Summerson, W. H., 230, 252 Susich, G. v., 181, ,950 (see Meyer) Sutherland, G. B. 13. M., 294, 298, 299, 300, 301, 302, 303, 304, 307, 308, 316, (see Darmon), 317 Svegvari, A., 97(65), 156 Svensson, H., 30, 31, 66 Swain, T., 6, 65 Swann, N. M., 91(174), 158 Swanson, M. A., 20, 66 S y l v h , B., 371, 386 (see Snellman) Synge, R. L. M., 2, 3, 18, 19, 20, 21, 24, 30, 32, 34, 35, 36, 37, 40 ,42, 47, 54, 59, 61, 63 (see Consden), 64 (see Gordon), 65, 66 Syverton, J. T., 260(51), 290 Szent-Gyorgyi, A., 163, 164, 165, 166, 168, 173, 174, 176, 179, 182, 185, 186, 187, 194, 195, 196, 198, 199, 200, 208, 210, 211, 212, 213, 216, 218, 223, 224, 225, 228, 229, 230, 231, 232, 234, 239, 240, 241, 248, 251 (see Rozsa), 252, 266, 289
T Tandler, R., 156 Tanford, C., 347, 349, 386 Taylor, H. S., 132, 159 Tazawa, Y., 29, 66 Tchen, P. U., 32, 64 Temple, R. B., 280(2), 281(2), 284(2), 288, 307, 315, 316, 363, 384 (see Elliott) Tenow, M., 195, 211, 214, 225, ,851 (see Snellman) Thaureaux, J., 139, 159 Theis, E. R., 125(199), 159 Thomann, H., 184, 218, 221, 228, 229, 252 (see Turba) Thomas, A. W., 77(200), 159
Thompson, H. W., 293, 298, 299, 300, 302, 312, 316 (see Barer), 317 Thorell, B., 321, 386 Thuringer, J. M., 259, 290 Tiselius, A., 31, 32, 34, 35, 38, 40, 63, 67 Todd, A. R., 61, 63 (see Dalgliesh) Toennies, G., 27, 67 Tosi, L., 177, 248 (see Cigada) Treiber, E., 216, 251, 357, 361, 377,386 Trillat, J.-J., 115(201), 159 Tristram, G. R., 47, 50, 67, 260(56), 290, 299, 316 (see Darmon) Troensegaard, N., 29, 67 Trogus, C., 139, 157 Tsao, T. C., 208, 210, 252 Tsubio, M., 300, 301, 302, 317 (see Mizushima), 318 Tulane, V. J., 258(58), 261(58), 290 Tuppy, H., 20, 24, 26, 30, 31, 34, 55, 56, 66
Turba, F., 31, 32, 34, 48, 67, 184, 192, 218, 221, 228, 229, 250, 252 Tustanovskii, A. A., 137(165), 168 Twyman, F., 369, 386
U Uchino, T., 29, 67 Udenfriend, S., 43, 44, 64 Uzman, L. L., 309, 318
V Valko, E., 181, 250 (see Meyer) Vanamee, P., 111, 159 Vandenbelt, J. M., 375, 386 van Heyningen, W. E., 77(29), 154 van Slyke, D. D., 19, 67 van Vunakis, H., 7, 67 Varga, L., 164, 174, 182, 252 Venet, A. M., 136(179), 138(179), 158 Vickery, H. B., 3, 28, 67, 155 Villee, C. A., 57, 67 Vining, L. C., 61, 63 (see Dalgliesh) Virtanen, A. I., 18, 23, 67 VlBs,'_F.,306, 518, 368, 586
W Waitkoff, H. K., 166 Waldschmidt-Leitz, E., 32, 48, 67
399
AUTHOR INDEX
Walker, P. M. B., 368, 386 Warburg, O., 66.2 Warner, R. C., 13, 19, 24, 28, 67 Warren, W. J., 78, 156 Wasteneys, H., 18, 67 Waters, J. W., 93(175), 159 Weber, A., 164, 165, 166, 167, 168, 169, 170, 171, 173, 175, 176, 179, 180, 182, 183, 185, 191, 216, 256 Weber, E., 247, 252 Weber, H. H., 163, 164, 165, 166, 170, 171, 172, 173, 179, 180, 182, 183, 185, 190, 191, 193, 194, 195, 198, 199, 200, 203, 204, 206, 207, 208, 210, 211, 213, 214, 215, 216, 217, 218, 225, 226, 227, 228, 230, 234, 235, 236, 237, 238, 239, 242, 247,648, 650,dril (see Portzehl), 252
Weidinger, A., 76(100), 87, 88, 125(123), 147(122), 148(123), 156, 167, 261 (28), 689 Weir, C. E., 121(203), 122(203), 125, 126, 159
Wells, I. C., 339, 341, 385 (see Pauling), 386 (see Schroeder) Westall, R. G., 33, 65 Whittaker, E., 371, 386 Wiederhorn, N. M., 126(214), 160 Wiegand, W. B., 180, 662 Wieland, T., 23, 31, 32, 41, 67 Wilkins, M. H. F., 364, 386 Wilkinson, V. A., 258(58), 259, 260, 261 (57, 58), 290 Williams, J. W., 137(185), 159 Williams, V. Z., 293, 295, 297, 316 (see Barnes), 318
Wilson, J. A., i47, 159 Wilson, P. C., 36, 65 (see Patton) Wilson, R. H., 261(59), 290 Winterbottom, R., 6, 37, 62 (see Bell) Winton, F. R., 246, 252 Wirth, L., 23, 67 Wirtz, K., 216, 252 Wislocki, G. B., 70(204), 76(204), 159 Wohlisch, E., 123(205),159,173, 181,252 Woelfflin, R., 123(75), 156 Woiwod, A. J., 35, 41, 67 Wolpers, C., 79(207), 89(208), 110, I l l , 112(210), 120(209), 126, 136(209), 159, 259, 652
Wood, D. L., 310, 311, 313, 318 Woods, H. J., 123(212), 160, 261(6), 272(6), 278, 279(6), 284(6), 285, 288 Woodward, R. B., 304, 318 Woolley, D. W., 61, 67 Work, E., 7, 64 Wright, B. A., 118(213), 126(214), 160 Wright, N., 298, 299, 318 Wrinch, D., 355, 386 Wyckoff, R. W. G., 77(216), 78, 79(58), 111, 119, 137, 155, 160, 200, 208, 211, 212, 213, 239, 240, 241, 251 (see Rozsa) Wyman, J., 168, 250 Z
Zahn, H., 123(62), 130, 155, 160, 307,318 Zaiser, E. M., 25, 66 Zddes, A. L., 120(219), 160 Ziff, J., 184, 252 Zimm, B. M., 207, 251 (see Oster) Zscheile, F. P., 366, 384 (see Hogness)
Subject Index A Acropeptidcs, 29 Actin, 211-213, 265 as component of actomyosin, 163, 194 components of, 211 in muscle protein, 238, 241 purification, 233-234 reaction with ATP, 220-222 solubility, 198, 199 sulfhydryl groups in, 310 F-Actin, conversion to G-actin, 221 electrophoretic mobility, 200, 201 extractability, 231, 232, 233 flow birefringence, 212 interaction with L-myosin, 216-217 particle size, weight and shape, 208, 214 sedimentation velocity, 21 I solubility, 199 G-Actin, 200, 211 ATP and, 221-222 conversion of F-actin to, 221 electrophoretic mobility, 201 molecular weight, 211 particle size, weight and shape, 208 polymerization, 212, 219-220 adenosinetriphosphatase and, 185 SH-groups and, 220 reaction with L-myosin, 219 Actinomycin, dipeptide sequences in, 61 Actinotrichia, 95 Actom yosin, 213-21 6 action of ATP on, 163, 178, 190, 193, 194, 222, 225, 226-230 heavy metals and, 163 mechanism of, 181-182 of fluorescent dyes on, 216 of magnesium on, 203 of polyorthophosphates on, 190-193 adcnosinetriphosphatase activity of, 183-185, 223-225, 226 400
artificial, 214 particle size and shape, 214-215 viscosity, 227 birefringence, 216, 242, 243 chemical changes on stretching, 216 components of, 163, 194 dephosphorylation of ATP by, 311 dissociation of, 228-229 extractability, 231 formation, 217, 218, 219 adenosinetriphosphatase and, 185 chemical agents affecting, 218, 221 infrared absorption spectrum, 310 in muscle, 238, 239-244 muscular contraction and, 163 a-myosin and, 194 S-myosin and, 194, 195, 196 oriented, 163 particle size, weight and shape, 208, 214 purity tests, 236 sedimentation velocity, 213 SH-groups in, 310 shrinkage, 230 solubility, 198, 199 superprecipitation, 185, 186 threads, 163ff. action of phosphates on, 166 birefringence, 173, 175 contraction of, 166, 167, 168 as model of muscle, 163, 164 properties of, 164-166, 175 ultraviolet absorption spectrum, 357, 358 viscosity, 215 ATP, see under Adenosinctriphosphate Adenosinetriphosphatase, activation by salts, 225 activity in actomyosin, 183-185, 223225, 226 in L-myosin, alkali earth metals and, 203 muscle contraction and, 183-185 Adenosinetriphosphate,
401
SUBJECT INDEX
G-actin and, 221-222 actomyosin and, 163, 178, 182, 190, 193, 194, 222, 225, 226-230 breakdown, 222-223 muscle contraction and, 187, 188, 189, 192 contractile proteins and, 193 effect on muscle proteins, 164, 177, 246-247 L-myosin and, 222ff. optimum concentration for muscle contraction, 185-187, 191 Albumin, serum, binding of tyrosine hydroxyl groups in, 349 infrared spectrum of, 306, 311 molecular weight, 7 N-terminal rcsidues of, 7 ultraviolet spectrum of, 332, 333, 335, 336 Alkali, effect on sulfur-containing chromophors, 329 on oxidized insulin, 24 on ultraviolet spectra of amino acids, 344ff. Amides, infrared spectra of, 299-304 steric configuration and, 301 Amino acids, action of performic acid on, 27 of prolonged irradiation on, 322 arrangement in proteins, 1-67 determination of sequence in peptides, 42-43 fractionation by paper chromatography, 35-38 histochemistry of sulfur-containing, 255-257 infrared spectrum, 298 optical configuration and, 299, 312 in mammalian epidermis, 259-260 raman spectra of, 297-298 structure, 298, 312 infrared analysis of, 291, 298-299 vibration spectra and 291,292,297ff. ultraviolet spectra of, 319, 321ff. effect of alkali on, 343-347 of gel state on, 338-339
of peptide combination on, 337-338 of temperature on, 331, 334-336 fine structure of, 329-331, 332 p-Amino alcohols, reduction of C-terminal residues to, 10 Antibodies, ultraviolet absorption spectra, 382 Antiovalbumin, 58 action of papain on, 58 active center, 58 y-globulin and, 58 Aspartic acid, liberation from proteins, 22
B Bacillus anthracis, capsular substance, glutamic acid in, 61 glutamine in, 61 peptide linkages in, 4 Benzene, ultraviolet spectrum, effect of temperature on, 334 Birefringence, of actomyosin, 173, 175, 216, 242, 243 of fibrous proteins, 193 Byssolreratin, 90 source, 91
C Carboxypeptidase, effect on peptide bonds, 9-10 Casein, rate of hydrolysis, 28 ultraviolet spectrum, 333 effect of enzymic digestion on, 333, 342 Cell division, in mammalian epidermis, 259 Ceratotrichia, 95 Chitin, 92 nature, 92 structure, 312 Chondroitin sulfate, collagen and, 77 Chromatography, determination of peptide structure b y paper, 41-42 fractionation of proteins by, 34-39
402
SUBJECT INDEX
partition, in protein studies, 2, 3 separation of peptides by, 40-41 Chromophors, sulfur-containing, of proteins, 328, 329 effect of alkali on, 329 Chymotrypsin, binding of tyrosine hydroxyl groups in,
.
350
effect on peptide bonds, 25 on oxidized insulin, 26 ultraviolet absorption spectrum, 338 Chymotrypsinogen molecular stnicture, 315
sources of, 84 types of, 84-90 molecule, 135-150 configuration, 139 preparation, 153 side-chains in, 153 size and shape, 137-138, 153 structure, 315 weight, 153 occurrence, 76, 88, 92, 152 origin, 91 polypeptide chain arrangement in, 128129, 131, 132
Clostridium, production of collagenase by, 77 Clupeine, molecular weight, 359 in protamines, 48 IV-terminal residues of, 7 ultraviolet spectrum, 359 Collagen (9) , action of enzymes on, 77 chemical composition, 94, 126 chondroitin sulfate and, 77 classification, 78 chemical significance of physical, 92-97
physical criteria for, 78-84 contractility, kinetics, 122, 125-126 conversion to gelatin, 77 effect of tanning agents on, 124, 145146
elastin and, 89, 95 elastoidin and, 124 electron microscopy in the study of, 76, 78, 83, 84, 87
fibrils of, 69-1 60 crystalloid nature of, 70 models of, diffraction theory for, 104-109
electron micrographs of, 109-112 optical properties of, 97-99 structure of, 97-112, 152 X-ray diffraction patterns, 99-104 ichthyocol and, 95 identification, 76 industrial application of degradation products, 70 mesenchymal, and secreted, 92
protofibrils, 112-134 contractility, 121-127, 153 hypotheses of thermal, 121ff. diffraction patterns, 1136. extensibility, 119-121 properties, 112 structure, 115-119, 152-1 53 effect of hydration on, 117-119 reticulin and, i 6 , 77, 87, 88 role in ageing, 70 in pathologic processes, 70 in tanning, 70 secreted, 90-92 cornpositSon, 96 mesenchymal and, 92 origin, 91 sources, 9 1 X-ray diffraction pattern, 91-92 side chains in, 139-146 amino acid composition of, 142, 143 properties of, 142-143 significance, 69-76 structure, 150-152 biological aspects, 151 sulfur content, 96 swelling of, 146-150 groups involved in, 147, 148, 150 structure and, 147ff., 150 types of, 147 ultraviolet spectra of, 355-356, 358 X-ray diffraction patterns of, 74-75, 78, 81-83, 84-92, 113-114
Collagen 11, gelatin and, 123 Collagenase, 77 action on collagcn, 77 on gelatin, 77
SUBJECT INDEX
Copper glycinate, effect on actomyosin formation, 218 on transformation of G-actin t o F-actin, 221 Corneins, composition, 94, 95-96 physical methods for identification, 88 sources, 88 X-ray diffraction pattern, 89 Countercurrent distribution, fractionation of peptides by, 38-39, 41 in the study of gramicidin, 39 Cross &structure, of epidermal proteins, 274-277, 285-286 Cysteine, oxidation in epidermis to cystine, 255257 ultraviolet absorption spectrum, 328 Cystine, action of performic acid on, 27 in mammalian epidermis, 260, 261 oxidation in epidermis of cysteine to, 255-257 ultraviolet spectrum, 327 Cystine peptides, separation of, 40 Cytochrome C, infrared spectrum of, 311 Cytochromes, effect of temperature on ultraviolet spectrum, 334 D DNP, see under Dinitrophenyl Diketopiperazine, structure, 315 Dinitrophenyl method, in the study of proteins, 4, 5-8 Dopa (3,4-dihydroxyphenylalamine), formation on permanganate oxidation of proteins, 352 ultraviolet absorption spectra, 351
E Echinodermata, collagen from, physical methods for identification of, 88 X-ray diffraction pattern, 89
403
Edestin, molecular weight, 7 N-terminal residues, 7 rate of hydrolysis, 28 Elastin, 70 chemical composition, 94, 95 collagen and, 89, 95 electron microscopy in t h e study of, 89 sources, 88 X-ray diffraction in the study of, 88 Elastoidin, 77 collagens and, 89, 124 composition, 95 conversion to elastoidin 11, 123 diffraction theories, 108 effect of formalin on, 124 physical methods for ident,ification, 88 sources, 88 thermal shrinkage, 123 X-ray diffraction pattern, 89 Elastoidin 11, conversion of elastoidin to, 123 properties, 123-124 Electron microscopy, in the study of collagens, 76, 78,83,84 of elastin, 89 Enzymes, see also under names of individual enzymes action on collagen, 77 on oxidized insulin, 26 protein hydrolysis with, 25-26 Epidermin, films of, effect of heat on unorientcd, 284-285 oriented, 280-282 thermal contraction of purified, 273 transformation to p-form by stretching, 282-284 X-ray fiber photograph, 272 Epidermis, see also under Epidermin amino acids in, 259-260 cell division in, 259 of cow’s nose, 257, 258, 266-267 dendritic cells in, 255 general structure, 254-255 layers of, 255 properties of, 254-264 proteins of, 253-290 conversion of a- t o @-type,261-262, 263
404
SUBJECT INDEX
cross 8-form of a-type, 274-277 effect of temperature on, 272-274 extraction, 266-267 fibrinogen and, 266 molecular weight, 268-269 myosin and, 266 particle size and shape, 269 physical properties, 270-279 purification, 267-268 solubility, 265 sulfur content, 269-270 thermal contraction, 273-274 thermal stability, 262 X-ra.y diffraction, 261-262 ribonucleic acid in, 259 sulfur content, 258, 260, 261 X-ray absorption and, 258-259 thermal contraction, 263 thermal stability, 262-264 X-ray diffraction studies on, 261-262
F Fibers, 71 diffraction patterns of protein, 78 fibrinogen (k-m-e-f) group of keratinmyosin-epidermis, 78 diffraction pattern, 79 i n insect cuticles, 92 Fibrils, of collagen, 69-160 Fibrin, solubility, 265 Fibrinogen, solubility, 265 Folic acid,!61 Formalin, effect on collagen, 124 on elastoidin, 124
G Gelatin, 70 action of collagenase on, 77 collagen I1 and, 123 configuration, 139 conversion of collagen to, 77 molecular weight, 136 peptides in, 59 phenylalanine in, 326 rate of hydrolysis, 28 ultraviolet absorption spectrum, 332
Gliadin, rate of hydrolysis, 28 Globulin, action of papain on, 26 denaturation, 348 pepsin and horse, 25 ultraviolet spectrum, 332 effect of enzymic digestion on, 332, 342 7-Globulin, 58 homogeneity, 58 molecular weight, 7, 58 peptides in, 58 ultraviolet spectrum of rabbit, 350 Glue, 70 Glutamic acid, liberation from proteins, 23 L-Glutamic 7-methyl ester, polymerization products of N-carbonic anhydride of, 305 Glutathione, dipeptide sequences in, 61 ultraviolet spectrum, 353, 354 Glycerol, action of anhydrous, on proteins, 29 Glycine, infrared spectra, 298 Glyoxalase, molecular structure, 315 Gramicidin, countercurrent distribution in the study of, 39 degradation products, 20 peptides in, 61 infrared spectrum, 311 molecular weight, 360 ultraviolet absorption spectrum, 360 Gramicidin S, stability, 20 structure, 4 “Ground substance,” of connective tissue, 70, 72, 92 mncopolysaccharides in, 70
H Hemoglobin, molecular weight, 7 N-terminal residues of, 7 structure, of crystalline, 312
405
SUBJECT INDEX
species differences in, 59 ultraviolet spectrum, 332 of Cooley’s anemia, 341-342 effect of temperatures on, 334 of human, 339-341
I Ichthylepidin, identification by physical methods, 88 sources, 88 Ichthyocol, chemical composition, 94 collagen and, 89, 95 identification by physical methods, 88 optical properties, 139 solubility, 137 sources, 88, 95 X-ray diffraction, 89 Immune precipitates, spectrophotometric analysis, 382 Infrared absorption spectra, of amides, 299-304 of amino acids, 297-299 configuration and, 299 of polypeptides, 304-306 optical configuration and, 304 structure and, 306 of proteins, 306-311 denaturation and, 306, 309 in epidermis, 279-286, 287, 288 polypeptide chain folding and, 307309 steric configuration and, 301 Insulin, 50-57, 61 action of irradiation on, 29 of performic acid on, 27 binding of tyrosine hydroxyl groups in, 349 composition, 51ff. degradation, 20, 22, 309 molecular weight, 7, 50 N-terminal residues of, 7 oxidized, action of proteolytic enzymes on, 26, 51-52 alkali and, 24 fractions of, 51-57 N-terminal peptides in, 52-54 properties of, 51
molecular weight, 51 structure, 50, 51, 315 of phcnylalanine chains in, 54-57, 60, 61 sulfhydryl groups in, 310 p-Iodophenylsulfonyl chloride, see under Pipsyl Ionophoresis, fractionation of peptides by, 30-31, 41
K Keratin(s), 96 hair, thermal stability, 262 infrared spectra, 279, 307 structure, 261, 314, 315 ultraviolet absorption spectra, 333, 349 wool, binding of tyrosine hydroxyl groups in, 349 composition, 48-50 homogeneity, 48, 50 a-Keratin, structure, 307, 308 Koilin, 92
L 6-Lactoglobulin, binding of tyrosine hydroxyl groups in, 352 effect of trypsin on, 26 molecular weight, 7 N-terminal residues of, 7 Lanthionine, 328 Light scattering, of protein solutions in the ultraviolet, 36 1 Lysine peptides, separation of, 39 Lysozyme, binding of tyrosine hydroxyl groups in, 349 infrared spectrum, 311 molecular weight, 7, 360 N-terminal residues of, 7 rate of hydrolysis, 28 resistance t o alkali, 362 ultraviolet spectrum, 333, 360 effect of alkali on, 349 of enzymic digestion on, 333, 342
406
SUBJECT INDEX
M Magnesium, action on actomyosin, 203 Metals, action of heavy, on actomyosin, 163 alkaline earth, effect on L-myosin and its actin complexes, 203 Methionine, action of performic acid on, 27 in mammalian epidermis, 260, 261 ultraviolet absorption spectrum, 328329 S-Methylisothiourea, in the study of tyrocidine, 9 Microscope, reflecting, 312 Models, contractile, of muscles, 163-193 Molecules, structure, spectrometric determination, 292-295, 314 Molluscs, physical methods for identification of collagen in, 88 Muscle ( s ) , A band of, 193, 194 action of ATP on, 164 birefringence of, 171-173, 241, 242, 243 comparison of properties of muscle models and, 174-178 contraction, adenosinetriphosphatase and, 185 adenosinetriphosphate breakdown and, 187, 188, 189 analysis of, 187-189 heat as energy source of, 182-183 mechanism of, 162-163 optimum ATP concentration for, 185-187, 191 role of fibrous muscle proteins in, 161-252 temperature and, 173, 174 fibers, see also under Muscle models and Actomyosin threads action of phosphates on, 166 contraction of, 166-168 myokinase in, 166 properties of, 175 tine structure of, 239-244 models of, see also under Muscle fibers
and Actomyosin threads, 163-193 action of ATP on, 177, 178ff. contractile of, 170, 171, 176, 177 contraction, analysis of, 187-189 thermodynamics of, 178-190 models, plasticizing effect of benzaldchydr on, 190, 192 of polyphosphatrs on, 19Off. of Salyrgan on, 190, 192 properties, 174-178 types of, 163-164 proteins, fibrous of, 231ff. contraction of, 244-245, 309-31 1 effect of fatigue on, 245 fractionation a n d purification of, 233-235 infrared methods in the study of, 310, 311 isolation, 231-233 proportion of F-actin and L-myosin in, 237-239 quick release phenomenon in, 168, 169 relaxation, mechanism of, 162, 163 Myo fibril, proteins of, 193-236 fine structure of skeletal muscle and, 237-246 Myoglobin, molecular weight, 7 N-terminal residues of, 7 Myokinase, in muscle fibers, 166 Myosin, see also L-Myosin, 194, 265 amino acid composition, 209 components, 194 crystalline, 194, 196 L-myosin and, 194, 195 enzymatic activity of, 194, 230 infrared spectrum, 279, 360 N-terminal residues of, 7 structure, 307, 308 thermal stability, 262 a-Myosin, actomyosin and, 194 @-Myosin, electrophoretic mobility, 201 L-myosin and, 194, 196, 200 ?-Myosin, . contractin and, 194
407
SUBJECT 1NDEX
A-Myosin, 234 Myosin B, 231 L-Myosin, 194, 195, 203-211 in actomyosin, 194 adenosinetriphosphatase activity of, 203, 219, 223-225, 226 amino acid composition, 207 chemical changes in shrinkage of, 230 colloidal state, 199 as component of actomyosin, 163 crystalline myosine and, 194, 195 denaturation, 203, 204, 234, 235 depolymerisation, 210 diffusion constant, 204, 205 electrophoretic mobility, 200, 201, 202 extractability, 231, 233 homogeneity, 210 interaction with F-actin, 216-217 with G-actin, 219 with ATP, 222 molecular data, 203-209, 213 monodispersity of, 205, 206 in muscle protein, 238 p-myosin and, 194, 196, 200 particle size, shape and weight, 207, 208, 209 prosthetic group, 207 purification, 234 role in muscular contraction, 163 solubility, 197, 198, 199 structure, 210 tropomyosin and, 210 viscosity, 215 S-Myosin, 194 actomyosin and, 194, 195, 196
N Norleucine, infrared spectrum, 298 Nylon, structure, 305, 307 0 Ovalbumin, 7 binding of tyrosine hydroxyl groups in, 349 conversion to plakalbumin, 57 degradation, 14-15, 19 formation of diketopiperazines from, 29
hydrolysis, 35 rate of, 28 pepsin and, 25 structure, 14, 57 ultraviolet absorption spectrum, 354 effect of low temperatures on, 335 of p H on, 347 Ovokeratin, 90 composition of, 96 precursor of, 96 sources, 91 Oxarsan, effect on formation of actomyosin, 218 on transformation of F-actin t o G-actin, 221
P Papain, action on antiovalbumin, 58 on globulin, 26 Paramyosin, 194 isolation, 233 solubility, 198, 199 Penicillin, infrared analysis of, 299 Pepsin, effect on peptide bonds, 25, 26 infrared spectrum, 309, 311 denaturation and, 309 ultraviolet spectrum, 333, 338 effect of low temperatures on, 335 Peptide bonds, ultraviolet absorption spectrum of, 352-355, 364-367 Peptides, determination of amino acid sequence in, 42-43 of N-terminal, 7-8 estimation in protein hydrolyzates, 43 fractionation of, 2 9 4 0 by chromatography, 4 0 4 1 by ion-exchange methods, 31-34 by ionophoretic methods, 30-31 spectrophotometric estimation of p K values of, 346-347 stability, 19-21 structure, determination of, 41-44 ultraviolet spectrum, 332 effect of gel state on, 338-339
408
SUBJECT INDEX
Performic acid, action on amino acids, 27 Phenols, effect of ionization of hydroxyl groups on absorption spectra of, 345 Phen ylalanine spectrographic determination in proteins, 326, 330-331 ultraviolet absorption spectrum, 324, 326, 329, 332 effect of alkali on, 344 of enzymic digestion on, 342 of low temperature on, 335 of peptide combination, 343 DL-Phenylalanine-L-glutamic, methylester, structure of copolymers of, 305-306 Plicnylthiocarbamyl method, in the study of polypeptides, 8 Pipsyl method, for estimation of peptides, 43-44 Plakalbumin, 57 formation, 26, 57 Polylysine, infrared spectrum, 31 1 Polymyxin, 343 phenylalanine in, 326 Polypeptides, chains of, 3-4 infrared spectra, 299, 304-306 nomenclature, 3-10 structnre, infrared analysis of, 291ff., 305, 306, 312, 313 synthetic, infrared spectra of high molecular weight, 307 structure, 314, 315 types of, in proteins, 3-10 ultraviolet absorption spectra, 360, 362 Polyphospha tes, action on actomyosin, 190 Polysaccharides, in embryonic skins, 76 Procollagcn, 137 molecular weight, 137 Prokeratin, 96 Protamines, structure, 47-48 Proteins, see also under names of individual proteins action of ATP on contractile, 193 of tyrosinase on native, 351-352
alkali-labile structure in globular, 347-349 arrangement of amino acids in, 1-67 binding of tyrosine hydroxyl groups in, 349-352 color in solution, 320, 321 degradation, 11-29 denaturation, 313 effect of prolonged irradiation, 322 epidermal, 253-290 asymmetry of, 286 a-p transformation of, 261, 262, 287, 288 cross p-forms of a-,274-277, 278 polypeptide chains in, 276-277 production, 274 reversibility of, 277 effect of temperature on, 272-274 elasticity, 277-279 extraction of, 266-267 infrared spectra, 279-286 heat denaturation and, 288 molecular weight, 268-269, 286 particle size and shape, 269 physical properties, 270-279 origin and, 286 purification, 267-268 structure, 261-262, 263, 270-277, 279ff., 286 sulfur content, 269-270, 286 urea and, 277 X-ray diffraction, 277 fibrous, diffraction patterns, 78 intracellular, solubility, 264-266 structure, 261-264 of muscle, aggregation, 199-200 birefringence, 193 changes in, due to fatigue, 245 colloidal state, 199 electric charge, 200 extraction, 193, 194, 231-235 globulin nature of, 197 muscular contraction and, 161252, 309-311 nomenclature, 194, 196 reactions of, 216-230 solubility, 191-199
409
SUBJECT INDEX
ultraviolet spectra, 355ff. of myofibril, 193-236 homogeneity, 60 hydrolysis of, 12-27 in acid, 18-23 in alkali, 23-25 electrostatic effects, 18-19 enzymic, 15 general aspects of, 12-18 mechanism of, 21-22 nature of partial hydrolyzates, 13-15 rate of, 18, 26-27, 28 factors determining, 23 rearrangement of peptide sequences during, 15-18 steric effects, 19-20 yield of peptides, 12-13 immune, spectrophotornetric analysis, 382 infrared spectra of, 306-31 1,313 complex formation with sodium dodecyl sulfate and, 311, 312 denaturation and, 306, 309 polypeptide chain folding and, 307309 light scattering in the ultraviolet by solutions of, 361 methods for determination of residues in, 5-10 spectrophotometric determination of tyrosine and tryptophan in, 369 structure, 3 infrared analysis of, 291, 306-311, 312, 313 vibration spectra and, 291 types of polypeptides in, 3-10 ultraviolet spectra of, 319, 358-364 effect of alkali on, 345-347 of enzymic digestion on, 332, 333, 342-343 of peptide bonds on, 352-355 of temperature on, 334-336 fine structure of, 332, 333-336, 337 time effect of high pII, 349-350 vibration spectra of, 295-296 Protofibril, 71
R Radiation, action on peptide bonds, 29
Raman spectra, 291, 292 of amino acids, 297-298 infrared spectra and, 292 Residues, terminal of proteins, determination of position, 5-10 of C-terminal, 9-10 N-terminal, 6-7 nomenclature, 4 reduction of C-terminal, 10 Reticulin, collagen and, 76, 77, 87, 88 identification by physical methods, 88 sources, 87, 88 Ribonu clease, time effect of high p H on ultraviolet spectra of, 349 Ribonucleic acid, in mammalian, epidermis, 259 Rigor mortis, 189 A T T a n d , 189
S Salmixie, infrared spectrum, 311 N-terminal residues of, 7 Salyrgan, effect on actomyosin formation, 218 on transformation of G-actin t o F-actin, 221 plasticizing effect on actomyosin systems, 190, 192 Scales, composition of herring, 94, 95 Schardinger dextrins, stability, 20 Serine, destruction by acid, 18 Serum, spectrophotornetric analysis, 380-382 Silk, structure, 315 Silk fibroin, composition, 44-47 molecular weight, 44 rate of hydrolysis, 28 spectroscopical identification of amino acids in, 304 ultraviolet spectrum of oriented, 355 structural implications, 355, 358 X-ray diffraction pattern, 357-358
410
STIRJECT INDEX
Skin, embryonic, polysaccharides in, 76 sulfur in, 76 Spectrophotometers, infrared, 297 Spectrophotometry, accuracy, 323 protein analysis by, 321ff. Spleen, reticulin in beef, 87 Spongin, chemical composition, 94, 95 collagen and, 90 Sulfhydryl groups, free, in epidermis, 255-257, 264, 286 in insulin, 310 polymerization of G-actin and, 220 Sulfonic acids, separation of peptides with, 40 Sulfur, in embryonic skin, 76 in epidermis, 258-260, 261
T Tanning agents, effect on collagen, 124, 145-146 Temperature, effect on epidermal proteins, 262 on epidermis, 262 Tendons, avian, contractility, 124 mammalian, contractility, 122, 126 diffraction patterns, 113 Thiocarbamate method, in the study of polypeptides, 8 Threonine, destruction by acid, 18 infrared spectrum, optical configuration and, 299 Tissues, connective, structural elements of, 71, 72, 73 infrared spectra of normal and cancerous, 312 ultraviolet absorption characteristics of living, 367-369 Tropomyosin, 194, 210-213, 265 amino acid composition, 200 crystallinity, 199
depolymerization, 210 electrophoretic mobility, 201 extractability, 231, 233 infrared absorption studies on, 279 molecular data, 210 L-myosin and, 210 N-terminal residues of, 7 particle size, shape and weight, 208, 210 polymerization, 210 purification, 235 in rabbit muscle, 23 solubility, 197, 198, 199 structure, 307, 308 Trypsin, effect on @-lactoglobulin,26 on oxidized insulin, 26 on peptide bonds, 25 ultraviolet spectrum, 333 effect of combination with trypsin inhibitor on, 342 time effect of high pN on, 350 Trypsin inhibitor, effect of combination with trypsin on ultraviolet spectrum, 342 pancreatic, molecular weight, 7 N-terminal residues of, 7 Trypsinogen, 61 Tryptophan, action of performic acid on, 27 spectrophotometric determination in proteins, 321, 343, 369-380 accuracy, 371-376 chemical determination, 377-378 limits of, 378-380 stability, 18, 24 ultraviolet spectrum, 324ff. effect of alkali on, 344 of temperature on, 335 fine structure of, 329, 332 Tyrocidine, determination of freeamino groups in, 9 hydrolysis, 35 infrared spectrum, 31 1 structure, 4 Tyrosinase, action on native proteins, 351-352 Tyrosine, hydroxyl groups of, binding in proteins, 349-352 spectrophotometric determination in
411
SUBJECT INDEX
proteins, 321, 323, 343, 369-380 accuracy, 371-376 chemical determination and, 377-378 limits of, 378-380 spectrophotometric estimation of pK values of, and derivatives, 346-347 ultraviolet spectrum, 323-325 effect of alkali on, 345 fine structure of, 329, 332
of ovalbumin, 335, 347, 352-355, 364367 of pepsin, 333, 335, 338 of phenylalanine, 324, 326, 329, 332, 335, 342-344 of trypsin, 333, 342, 350 of tryptophan, 324, 325-326, 329, 332, 335,344 LJrea, epidermal proteins and, 265, 277
U
V
Ultra filtration, separation of peptides by, 40 Ultraviolet absorption spectra, of albumins, 332, 333, 335, 336 of amino acids, 319, 321ff., 329-331, 332, 334-336, 337-338, 339, 343347 effect of alkali on, 344ff. of antibodies, 382 of benzene, 334 of casein, 333, 342 of chymotrypsin, 333 of clupeine, 359 of collagen, 355-356, 358 of cysteine, 328 of cystine, 327 of cytochromes, 334 of dopa, 351 of gelatin, 332 of globulin, 332 effect of enzymic digestion on, 332, 342 of glutathione, 353, 354 of hemoglobin, 332, 334, 339-342 of insulin, 333, 336, 342, 349 of keratin, 333, 349 of lysozyme, 333, 342, 349, 360 of methionine, 328-329
Valine, infrared spectrum, 298 Vibration spectra, 291 determination of molecular structure by, 202ff. experimental techniques, 295-297 group frequencies in, 293-297 of large polyatomic molecules, 295 of proteins, 295-206
W Wool, dipeptide sequences in, 61 rate of hydrolysis, 28
X X-Ray absorption, sulfur content of epiderniis and, 258259 X-Ray diffraction studies, on collagens, 73, 74-75, 78, 81-83, 8497, 113ff. on cornein, 89 on elastin, 88 on elastoidin, 89 on epidermis and epidermal proteins, 261-264,270-272 on ichthyocol, 89
This Page Intentionall Left Blank