A t'" 3SBSSSSSSSBSSSBBE Marine Biological Laboratory Library Woods Hole, Mass. w^% Presented ty John Wiley and sons,Inc. Aug, 195S o ■jg^Bjj □ LD 5=: □ T-" ru ^ o a m- gi ■ r-q 3 a " m : □ □ ^^^^™ Essays in Biochemistry %oua ?. Q^^ 5 Essays in Biochemistry SAMUEL GRAFF Editor NEW YORK • JOHN WILEY & SONS, INC. LONDON • CHAPMAN & HALL, LIMITED Copyright © 1956 BY John Wiley & Sons, Inc. All Rights Reserved This book or any part thereof must not be reproduced in any form unthoid the written permission of the publisher. Library of Congress Catalog Card Number: 56-7155 PRINTED IN THE UNITED STATES OF AMERICA Preface These essays were written in honor of Hans Thacher Clarke, on the occasion of his retirement as Professor and Chairman of the De- partment of Biochemistry, College of Physicians and Surgeons, Colum- bia University. Through their contributions to this volume, the authors, all former students or academic associates of Professor Clarke, sought to express their affection and esteem for him, and their gratitude for his generous aid and counsel. If merit be found in these essays, it is a consequence, in large measure, of the high standards of chemical scholarship to which the authors were exposed, early in scientific life, through their association with Professor Clarke. For this opportunity to learn from him not only some of the facts of sci- ence but something of its spirit as well they are deeply grateful. The essayists were accorded free rein in regard to subject and style. The articles, for the most part, are neither reviews nor experimental reports. Some are critical discussions of the status of a biochemical problem at the time of writing, whereas others are frankly speculative or deliberately provocative. The absence of editorial directive permits the discerning reader to glimpse the personalities behind the printed page and to contrast diverse approaches toward scientific goals. Nature sometimes yields her secrets upon the skillful use of but a few tools; frequently, the techniques of many scientific disciplines are needed. But always there is the excitement of the search. Eael A. Evans, Jr. Joseph S. Fruton Samuel Graff David Shemin New York November, 1955 Contents Some Metabolic Products of Basidiomycetes 1 Marjorie Anchel, Ph.D. Research Associate The New York Botanical Garden Heterogeneity of Deoxyribonucleic Acid (DNA) 14 Aaron Bendich, Ph.D. Associate Professor of Biochemistry Sloan-Kettering Division Cornell University Medical College Biosynthesis of Branched-Chain Compounds 22 Konrad Bloch, Ph.D. Higgins Professor of Biochemistry Department of Chemistry Harvard University The Biochemistry of Lysogeny 35 Ernest Borek, Ph.D. Associate Professor of Biochemistry The City College of New York Biochemist Brooklyn Veterans Administration Hospital The Development of a Plasma Volume Expander 56 Max Bovarnick, M.D., Ph.D. Chief of Investigative Medicine Brooklyn Veterans Administration Hospital Clinical Associate Professor of Medicine State University of New York, College of Medicine and Marianna R. Bovarnick, Ph.D. Lecturer in Medicine State University of New York, College of Medicine Biochemist Brooklyn Veterans Administration Hospital The Very Big and the Very Small: Remarks on Conjugated Proteins 72 Erwin Char gaff, Ph.D. Professor of Biochemistry 71161 viii Contents Cell Chemistry Laboratory Department oj Biochemistry Columbia University Unbalanced Growth and Death: A Study in Thymine Metabolism 77 Seymour S. Cohen, Ph.D. Professor of Biochemistry Professor of Pediatrics University of Pennsylvania School of Medicine The Childrens Hospital of Philadelphia Some Thoughts on the Biochemistry of the Steroid Hormones 85 Lewis L. Engel, Ph.D. Assistant Professor of Biological Chemistry Harvard Medical School Associate Biochemist Massachusetts General Hospital Tutor in Biochemical Sciences Harvard College The Biochemistry of the Bacterial Viruses 94 E. A. Evans, Jr., Ph.D. Professor and Chairman Department of Biochemistry University of Chicago The Biosynthesis of Peptide Bonds 106 Joseph S. Fruton, Ph.D. Professor of Chemistry Chairman, Department of Biochemistry Yale University On the Nature of Cancer 119 Samuel Graff, Ph.D. Professor of Biochemistry Columbia University The Francis Delafield Hospital Problems in Lipide Metabolism 133 Samuel Gurin, Ph.D. Professor and Chairman Department of Biochemistry University of Pennsylvania Tetrazoles as Carboxylic Acid Analogs 141 Robert M. Herbst, Ph.D. Professor of Chemistry Kedzie Chemical Laboratory Michigan Stale College The Structural Basis for the Differentiation of Identical Groups in Asymmetric Reactions 156 Hans Hirschmann, Ph.D. Associate Professor of Biochemistry Department of Medicine Western Reserve University Contents ix The Nitrogen-Sparing Effect of Glucose 175 Henry D. Hoberman, Ph.D., M.D. Associate Professor of Biochemistry Albert Einstein College of Medicine The Metabolism of Inositol in Microorganisms: A Study of Molecular Conformation and Biochemical Reactivity 181 Boris Magasanik, Ph.D. Silas Arnold Houghton Assistant Professor of BacU riology and Immunology Department of Bacteriology and Immunology Harvard Medical School The Biochemistry of Ferritin 198 Abraham Mazur, Ph.D. Associate Professor, Department of Chemistry City College of New York Assistant Professor of Biochemistry in Medicine Department of Medicine Cornell University Medical College The New York Hospital Some Aspects of Nitrogen Transfer in Biosynthetic Mechanisms 216 Sarah Ratner, Ph.D. Associate Member Division of Nutrition and Physiology The Public Health Research Institute of the City of New York. Inc. On the Bigness of Enzymes 232 David Rittenberg, Ph.D. Professor of Biochemistry College of Physicians and Surgeons Columbia University The Biosynthesis of Porphyrins 241 David Shemin, Ph.D. Professor of Biochemistry College of Physicians and Surgeons Columbia University The Role of Carbohydrates in the Biosynthesis of Aromatic Compounds 259 David B. Sprinson, Ph.D. Associate Professor of Biochemistry College of Physicians and Surgeons Columbia University On Determining the Chemical Structure of Proteins 270 William H. Stein, Ph.D. Member The Rockefeller Institute for Medical Research Glycogen Turnover 291 DeWitt Stetten, Jr., M.D., Ph.D. Associate Director in Charge of Research National Institute of Arthritis and Metabolic Dis< ases National Institutes of Health and Marjorie R. Stetten, Ph.D. Chemist, Section on Intermediary Metabolism x Contents National Institute of Arthritis and Metabolic Diseases National Institutes of Health The Veratrum Alkamines 308 Oskar W inter steiner, Ph.D. Director Division of Organic Chemistry The Squibb Institute for Medical Research The Chemical Basis of Heredity Determinants 322 Stephen Zamenhof, Ph.D. Assistant Professor of Biochemistry College of Physicians and Surgeons Columbia University Index 339 Some Metabolic Products of Basidiomycetes MARJORIE ANCHEL Investigation of the metabolic products of fungi has yielded results of considerable chemical and biological interest and practical impor- tance. The chemistry of the actinomycetes, penicillia, and aspergilli has been studied intensively, but the chemistry of the higher fungi, including the basidiomycetes, has received much less attention. The compounds discussed in this paper were all obtained from culture liquids of basidiomycete fungi. Their isolation was followed by microbiological assay, specifically, activity against certain bacteria. Had a chemical assay been used, it would be expected that the com- pounds isolated would have in common some particular chemical characteristic, although this might be only a small, even an insignifi- cant, part of their general chemical make-up. Since the assay was biological, a common denominator of biological activity must be ac- cepted, namely, in this case, that the compound interferes with the metabolism of the test organisms. Since it is to be expected that the same gross biological effect may be arrived at by different mechanisms, it is not surprising that compounds which have in common the prop- erty of antibiotic activity are frequently entirely unrelated chemically. Further, such compounds sometimes represent types which are unusual from the point of view of synthetic organic chemistry as well as little known from a biological standpoint. The antibiotic compounds isolated from the basidiomycetes investi- gated at the New York Botanical Garden offer one more illustration of these general conclusions, in that they represent diverse chemical types, some of which have no exact synthetic counterparts, and/or are biologically unfamiliar and, one might almost say, unexpected. On the other hand, certain rough groupings may be made into which fall the greater number of the compounds isolated. A surprisingly large 1 2 Essays in Biochemistry- group consists of polyacetylenic compounds, with which the major part of this paper will be concerned. Since the chemistry of most of the compounds discussed has not been completely elucidated and since their biological significance is entirely unknown, much that will be said is purely speculative. It is hoped that this will justify itself in some measure by suggesting further paths for exploration in relation both to chemical and to biological questions. For purposes of discussion, the compounds may be divided into groups comprising quinones, sesquiterpenes, larger polycyclic mole- cules, halogenated aromatic compounds, and polyacetylenes. Quinones This group, of which there are many examples among products of fungal origin, as well as among those produced by higher plants, is represented in our compounds by 5-methoxy-p-toluquinone * 1 and pos- sibly by fomecins A and B.2 Fomecins A and B are compounds isolated from culture liquids of Fomes juniperinus. Both have the molecular formula C8H805. Fomecin A has one acidic grouping. Its ultraviolet absorption spec- trum (in 95% ethanol) shows maxima at 241 nut {E — 10,760) and 305 in/* {E = 14,260), maxima not typical of quinones.3 When an aqueous suspension of fomecin A was treated with S02 the crystals dissolved, as would be expected on formation of a hydroquinone, but the orange-red color was not discharged nor could a hydroquinone be recovered. When the S02 was removed with nitrogen and the solution somewhat concentrated, crystals of fomecin A (identified by ultraviolet absorption spectrum and by analysis) were recovered. On the other hand, it reacted with 2,4-dinitrophenylhydrazine to yield a mono- 2,4-dinitrophenylhydrazone, Ci4Hi2N408, and with amines to yield deeply colored solutions which deposited colored crystalline deriva- tives. The aniline derivative analyzed for C14H13NO4 (C8H805 + CcH5NH2-H20). Fomecin B is isomeric with fomecin A (C8H805). It has absorption maxima in the ultraviolet at 262 mju, (E = ca. 20,000) and at 337 m/z, (E = ca. 7000). Its behavior toward 2,4-dinitrophenylhydrazine and amines is similar to that of fomecin A, but no crystalline derivatives have been isolated. If the fomecins are quinoidal, they apparently are not simple p-quinones. The compound 5-methoxy-p-toluquinone (I) was isolated from cul- * Unpublished work. Some Metabolic Products of Basidiomycetes 3 ture liquids of Copnnus similis and Lentinus degener. It fits into a series of oxygenated quinones, two closely related members of which, fumigatin (II) and spinulosin (III), were isolated from species of Penicillium and Aspergillus.4 Another closely related quinone, 2,5- dimethoxyquinone (IV) has been isolated from the basidiomycete, Polyporus jumosus.5 O CH3O^L J CH30 ,CH3 hOitSch, u HOljilCH3 [fi]0CH3 CH3olL^OH CH30IJ II II o o III IV Sesquiterpenes The compounds included here, solely on the basis of their molecular formulas, are illudin M and illudin S, isolated from culture liquids of Clitocybe illudens, and marasmic acid, from culture liquids of Marasmius conigenus. Illudin M, C15H20O3, and illudin S, Ci5H2o04 * (previously reported as C15H22O4),6 show striking resemblance in their chemical behavior to santonin, Ci5Hi803, and pseudosantonin, C15H20O4, which have been isolated from higher plants. All four com- pounds possess carbonyl groups. The santonins, and probably the illudins, possess a lactone ring. Pseudosantonin and illudin S yield acetates. Oxygen is lost on catalytic reduction of the illudins, but in the santonins straightforward saturation of the double bonds and reduction of the carbonyl group occur (also hydrogenolysis of the lactone ring of pseudosantonin).7 The behavior of the two groups of compounds towards reduction is not understood because the mechanism of oxygen loss in the illudins is unknown. All four compounds undergo acidic rearrangements which yield more than one product, and, in each case, one of the products is phenolic. Marasmic acid,8 C15H18O4,* like * Unpublished work. 4 Essays in Biochemistry the illudins, has a free carbonyl group, but unlike them it also has a free carboxyl group which can be esterified to yield a monoester. Larger Polycyclic Molecules This group includes ochracic acid and pleuromutilin. Ochracic acid was isolated from culture liquids of Corticum ochraceum. It has not yet been crystallized, but a pure freeze-dried preparation analyzed as a dicarboxylic acid of molecular formula CssH^-^Ot and was active against Staphylococcus aureus at a concentration of about 0.1 fxg. per ml. (about one-tenth the activity of penicillin in our assays). Pleuromutilin, C22H34O5, was isolated from culture liquids of Pleu- rotus mutilus, Pleurotus passeckerianus Pilat,9 and Drosophila suba- trata,10 and its structure was partly elucidated as that of a polycyclic compound containing two hydroxyl groups, a hindered carbonyl group, and, possibly, a lactone ring.11 Halogenated Aromatic Compounds This group is represented only by p-methoxytetrachlorophenol (V) ,12 a rather effective antifungal agent 13 obtained from culture liquids of Drosophila subatrata.10 OH !ci 0CH3 v Halogenated aromatic compounds have been reported previously as fungal products, but the mechanism of biological halogenation is still obscure.14 It is interesting, in this connection, that MacMillan 15 iso- lated the bromine analog of griseofulvin from culture liquids (of Penicillium griseofulvium and Penicillium nigricans) in which potas- sium bromide had been substituted for the chloride. Polyacetylenes The term polyacetylenes, as used here, indicates compounds with two or more triple bonds, in conjugation. This class of compounds has proved to be the largest group and comprises about as many compounds as are included in the other four groups combined. It includes nemotin, nemotinic acid,16 drosophilin C, drosophilin D,10,1T agrocybin,16,18 biformyne 1, biformyne 2,19,2° diatretyne amide,16,21 Some Metabolic Products of Basidiomycetes 5 diatretyne nitrile,16'22 and two newly isolated polyacetylenes from Coprinus variegatus.23 The polyacetylenes, intensively investigated by synthesis, were al- most unknown in biological material until quite recently. They have now been isolated from higher plants, from fungi {Basidiomycetes) 2i and from Actinomycetes.25 The probable isolation of propiolic acid from a bacterial culture has also been reported.26 Matricaria ester and its corresponding alcohol, originally isolated from higher plants,27 have been obtained from cultures of the basidiomycete, Polyporus anthro- cophilus Cke.28 Although polyacetylenes containing as many as eight conjugated triple bonds have been synthesized,* structural variety is better repre- sented in the biologically produced polyacetylenes, some of which contain groupings not yet synthesized. Most polyacetylenes of bio- logical origin contain double bonds as well as conjugated triple bonds; several contain an allene grouping, and some a benzene ring. Carlina oxide, a monoacetylenic compound, contains both a benzene ring and a furan ring.29 Among the functional groupings which have been found, alone or in combination, are: hydroxyl, carbonyl, carboxyl, ester, amide, lactone, and most recently, nitrile. An unbroken series of polyacetylenes with chain lengths C8 to C13 have been reported 2i in addition to the Ci7 and C18 compounds. Nemotin and nemotin A, the first of the polyacetylenes characterized in our laboratory, were not immediately recognized as such,f although they had been studied spectrophotometrically and the characteristic alkali conversion of nemotin to nemotin A had been examined in some detail.16 It was only after application of the too-little-known rule of Hausser, Kuhn, and Seitz 30 on spacings of the absorption maxima of polyacetylenes and polyethylenes, about 65/ for polyacetylenes and 37-47/ for polyethylenes, that these compounds, as well as the rest of the series, were placed in the polyacetylene group. The chemistry of nemotin has proved to be of considerable interest. The ultraviolet absorption spectrum of this polyacetylene indicated an endiyne system of unsaturation. On reduction, nemotin yielded undecanoic acid. Since nemotin itself was not acidic, it was postulated *The synthesis of diphenyloctaacetylene was reported from the laboratory of E. R. H. Jones, Nature, 168, 900 (1951). t A remark by Sorensen seems appropriate here. In discussing Semmler's pref- erence for an allenic rather than one of the possible acetylenie structures for carlina oxide, he says that Semmler, "led by an irrational aversion against the occurrence of the acetylenic compounds in nature . . . ," chose the wrong formula. 6 Essays in Biochemistry that it might contain an unsaturated lactone ring, which underwent hydrogenolysis on catalytic reduction to yield the reduced deoxy acid. Since it underwent a rearrangement strikingly similar to the myco- mycin to isomycomycin isomerization and since the product, nemotin A, also yielded undecanoic acid on reduction, it was further postulated that nemotin, like mycomycin, contained an allene grouping which was isomerized to an acetylenic linkage on treatment with alkali. These two postulates proved to be partially correct, since a lactone ring and an allenic system have been demonstrated by Bu'Lock, Jones, and Leeming.31 On the basis of the mycomycin-isomycomycin isomeriza- tion, which involves a shift of bonds as well as an allene-acetylene shift, the conversion might be pictured as follows: HC=C— C=C— C=C=C— C=C— C— C=0 vi o H3C— GheeC— c^c— C^C— C=C— C— COOH ViiiV=°hJ The hydroxyl might be lost on reduction of nemotin A 32> 33 or during the conversion of VI to VIII. The acid VIII was synthesized by F. Bohlmann 34 who very kindly sent us a sample for comparison with nemotin A. This compound proved to have less antibiotic activity than nemotin A against a number of bacteria and fungi and to have an ultraviolet spectrum in which the maxima, as compared to those of nemotin A, were shifted consistently 2 to 3 m/i, toward higher wave- lengths (Fig. 1). The explanation for both of these differences has been provided by the findings of Bu'Lock, Jones, and Leeming, who have demonstrated the conversion to be: HC^C— C=C— C=C=C— C— C— C— C=0 IX Nemotin O HC=C— C=C— C=C— C=C— C— C— COOH X Nemotin A That is, nemotin contains a saturated lactone ring, which opens on alkali treatment, with loss of water, providing a new double bond. No shift of acetylenic bonds occurs, only an allene to acetylene shift. The fact that nemotin A (X), as opposed to Bohlmann's acid (VIII), contains a terminal acetylenic carbon atom presumably accounts for Some Metabolic Products of Basidiomycetes 7 the difference in its ultraviolet absorption spectrum and also for its higher antibiotic activity. Drosophilins C and D, isolated from culture liquids of Drosophila subatrata, undergo the same characteristic allene to acetylene shift •C'l cm 4000 3500 3000 2500 2000 1500 1000 500 (1 n 0.0003% ( A n's acid • in H,0 | flv i / i n": * r. "*•..••" '...' V 200 225 250 275 300 Wavelength, m/i 325 350 Fig. 1. Ultraviolet absorption spectra of nemotin A and Bohlmann's acid. 8 Essays in Biochemistry as judged by spectrophotometric evidence, but the chemistry of these compounds has not yet been investigated. Agrocybin, isolated from culture liquids of Agrocybe dura, was char- acterized as XI.33 HOH2C— C=C— feC— C=C— CONH2 xi This compound bears interesting relationships both to biformyne 2 and to the diatretynes, discussed below. Biformyne 1 is a polyacetylene which on catalytic reduction yields a 9-carbon diol.20 The positions of the ultraviolet absorption maxima of the compound suggest a diendiyne system of unsaturation. The low extinction coefficients of the maxima, however, would speak against such a system, but, as was kindly pointed out to us by F. Bohlmann, they are in better agreement with those of symmetrical triyne diols.35 Oxidation of the reduction product with chromium trioxide in acetic acid yields octanoic acid,* suggesting adjacent hydroxyl groups, one of which is terminal, as in formula XII for the reduction product. H3C(CH2)6CHOHCH2OH " XII The presence of a 1,2-diol grouping is supported by the results of periodic acid oxidation of the reduction product, which yields formalde- hyde. Formula XIII is suggested for biformyne 1 by formula XII, its reduction product; by its absorption spectrum; and by the yield of an immediate precipitate with silver nitrate. HC^C— C^C— C^C— CH2CHOHCH2OH XIII Biformyne 2, like agrocybin, apparently occurs in a bound form in the culture liquid and is freed on boiling. The nature of the complex is unknown in both cases. The fact that both agrocybin and biformyne 2 possess hydroxyl groups suggests that this group may be involved in the attachment. The two polyacetylenes, diatretyne amide (XIV) and diatretyne nitrile (XV) , along with agrocybin (XI) form a rather interesting series. HOOC— C=C— C^C— feC— CONH2 xiv HOOC— C=C— C=C— C=C— C=N xv * Unpublished work. Some Metabolic Products of Basidiomycetes 9 Diatretyne amide differs from agrocybin only in that the former possesses a double bond in place of one of the triple bonds of agrocybin and a carboxyl group instead of the terminal hydroxyl group. It differs from diatretyne nitrile in that it possesses an amide group in place of the nitrile group. The rest of the molecule is, apparently, identical, but the possibility that XIV and XV differ also in their configuration (cis-trans) at the double bond has not been excluded. The effect of this difference, or these differences, on the antibiotic activity is rather striking. Diatretyne nitrile is active against Staphylococcus aureus at a concentration of about 0.1 jxg. per ml., whereas diatretyne amide is inactive at a concentration 8000 times as great. The antibiotic compounds isolated from basidiomycetes suggest prob- lems which are of interest both in their chemical and biological aspects. If any one tendency, from a chemical standpoint, appears to be common to this group of compounds, it is that they seem to possess, in general, an enhanced reactivity compared with other compounds of related structure. Thus, the illudins, which are notably similar to the santonins in their behavior, undergo the characteristic change with acid, described above, under considerably milder conditions. A large proportion of the basidiomycete polyacetylenes, in contrast to the synthetic ones and to those isolated from higher plants, are too unstable to be obtained in crystalline form but can be handled only in solution. Further, the allene to acetylene conversion of nemotin to nemotin A takes place under milder conditions than those reported for other simi- lar rearrangements.25-36 The behavior of other antibiotic compounds isolated, on the other hand, does not support the idea of enhanced reactivity. Pleuromutilin, ochracic acid, and diatretyne nitrile are relatively stable chemically, although all three of these compounds have a high degree of antibiotic activity. The basidiomycete compounds present other interesting chemical problems and shed further light on little-explored questions. The illudins, fomecins, and pleuromutilin exhibit what appears to be some- what unorthodox or at least unusual chemical behavior. The structures of the biformynes and of the precursors of biformyne 2 and agrocybin need clarification. The availability of compounds such as the nemotins and the drosophilins (along with mycomycin) provides models (not readily available synthetically) for study of the allene to acetylene shift as influenced by other structural features of the molecule. The behavior of agrocybin on catalytic reduction, using platinic oxide catalyst (loss of hydroxyl oxygen), serves as a reminder, particularly important in working with those polyacetylenes which are too unstable 10 Essays in Biochemistry to be analyzed, that caution is necessary in deducing the formula of a poly acetylene from that of its reduction products. Data useful for spectrophotometric correlations have been obtained in some instances. For example, the absorption maxima of diatretyne amide, in which the endiyne system is conjugated at either end (to a carboxyl carbonyl at one end and an amide carbonyl at the other end), are essentially the same as would be expected for an endiyne system conjugated only to a carboxyl group, and the spectrum of diatretyne nitrile resembles closely that of an entriyne system. Thus, the nitrile grouping as it occurs in this compound acts chromophorically like another acetylenic bond, whereas the conjugated carboxyl grouping again has no pronounced effect on the spectrum. The existence of polyacetylenic compounds of biological origin * poses many questions of general biological interest, and their isolation and structural elucidation pave the way for attacking some of the problems raised. From a metabolic standpoint, the existence of polyacetylenes of biological origin presents several problems. The first question which arises, naturally, is one concerning the origin of these compounds. By what series of reactions do these highly unsaturated compounds, of chain lengths as short as Cs and as long as Ci8, arise? What are their immediate precursors? Are they formed by combinations of shorter-chain unsaturated compounds or by dehydrogenation of com- pounds of similar chain length? Growth of poly acetylene-producing organisms using an isotopic carbon source may help to provide some of the answers. The fate of these compounds in the organism is also an unexplored problem. It seems reasonable to suppose that highly reactive compounds like the polyacetylenes may be not merely end products of metabolism but, rather, intermediates. By using labeled polyacetylenes produced either by the fungus or synthetically, it should be possible also to attack this problem. The general question of relation of structure to antibiotic activity arises also in connection with polyacetylenes. Neither the synthetic polyacetylenes nor those isolated from higher plants have been tested systematically for antibiotic activity. However, there are several facts * The term "naturally occurring" as applied to polyacetylenes isolated from fungal culture liquids was used in contradistinction to "synthetic" but is, perhaps, somewhat ambiguous in its implications. Nothing is known about the production of these compounds by fungi in nature since they have been isolated only under laboratory conditions. A term such as "biologically produced," suggested by Dr. Selman Waksman, or "of biological origin" might, therefore, be preferable. Some Metabolic Products of Basidiomycetes 11 which indicate that such activity is not a general property of con- jugated acetylenic linkages but depends, at least in part, on other groupings present in the molecule. Thus, among the basidiomycete polyacetylenes, compounds with similar unsaturated systems may show pronounced specificity in their bacterial and fungal spectra. A more striking demonstration of this fact is afforded by comparison of the compounds diatretyne nitrile and diatretyne amide discussed below. As mentioned previously, the latter is at least 8000 times as active against Staphylococcus aureus as the former. More specific questions are raised by the frequent occurrence of pairs of closely related compounds in a culture liquid. Diatretyne amide (XIII) and diatretyne nitrile (XIV) offer one of the most interesting examples of this. Are these compounds synthesized independently by the organism, or do they arise from a common, closely related precursor, or is one the precursor of the other? The likelihood of the last possibility is enhanced by the close structural similarity of the compounds: the difference between them consists only in possession of a nitrile grouping by one and an amide grouping by the other. If this last mechanism does represent the actual sequence, it would further suggest the possibility of the existence in the organism of an amide dehydrase. This would be a novel sort of enzyme, since reports of nitriles of biological origin are comparatively rare,37 and nothing is known of their close association with the corresponding amide. Diatretyne amide and diatretyne nitrile appear to be the first such pair reported, and diatretyne nitrile is the first reported polyacetylenic nitrile of biological origin. The possibility of taxonomic implications of basidiomycete products is interesting to consider. In higher plants, rather extensive investiga- tions have been made by Erdtman's group on the wood of Gymno- spermae (softwoods) and by Sorensen's group using plants of the Composite family. Erdtman has reported 3S that pinosylvin (3,5- dihydroxystilbene) or its methyl ether is present in the heartwood of most species of the genus Pinus, and that other genera of coniferous trees do not contain them. An instance in which "chemical taxonomy" has lent support to a taxonomic classification which apparently is in some question on morphological grounds is furnished by Sorensen's group. On the basis of their investigations of acetylenes of the Com- posite family, they state: 39 ''The power to synthesize acetylenic com- pounds thus separates Tripleurospermum distinctly from Matricaria." Although it appears, on examination of the basidiomycete compounds 12 Essays in Biochemistry from a taxonomic point of view, that these exhibit no relationships of taxonomic interest, it is, of course, too soon to draw conclusions since the number of basidiomycete compounds characterized is still relatively small. One facet of the group of findings is, perhaps, some- what more apparent than the others. The isolation of acetylenic compounds from basidiomycetes suggests their rather general biological occurrence since they have been shown to be produced by higher plants, by the basidiomycete group of fungi and by an actinomycete. The isolation of a propiolic acidlike compound from Escherichia coli, men- tioned above, is of further interest in this connection. From the findings and the discussion presented in this paper, it is apparent that no conclusions of a general nature can be drawn which would make a cohesive story of the particular corner of "polychem- ism" 40 observed in the compounds isolated in our laboratory from basidiomycetes. It is hoped, however, that some interesting trends have been pointed out and that possible paths have been suggested toward desirable goals. Some of the work described in this article was supported by a research grant (C-2308) from the National Cancer Institute of the National Institutes of Health, United States Public Health Service. The author is also indebted to Dr. William J. Robbins, Dr. Richard Klein, and Dr. Herbert Rackow for valuable suggestions on the manu- script, to Dr. Julian Wolff for preparation of Fig. 1, and to Drs. E. R. H. Jones and J. D. Bu'Lock for making available their findings prior to publication. References 1. M. Anchel, A. Hervey, F. Kavanagh, J. Polatnick, and W. J. Robbins, Proc. Natl. Acad. Sci. U. S., 84, 498 (1948). 2. M. Anchel, A. Hervey, and W. J. Robbins, Proc. Natl. Acad. Sci. U. S., 88, 655 (1952). 3. E. A. Braude, J. Chem. Soc, 1945, 490. 4. J. H. Birkenshaw, and H. Raistrick, Trans. Roy. Soc. London, B220, 245 (1931); W. K. Anslow and H. Raistrick, Biochem. J., 82, 2288 (1938); W. K. Anslow and H. Raistrick, Biochem. J., 82, 687 (1938). 5. J. D. Bu'Lock, J. Chem. Soc, 1955, 575. 6. M. Anchel, A. Hervey, and W. J. Robbins, Proc. Natl. Acad. Sci. U. S., 86, 300 (1950). 7. G. R. Clemo and W. Cocker, J. Chem. Soc, 1946, 30. 8. F. Kavanagh, A. Hervey, and W. J. Robbins, Proc Natl. Acad. Sci. U. S., 35, 343 (1949). 9. F. Kavanagh, A. Hervey, and W. J. Robbins, Proc. Natl. Acad. Sci. U. S., 87, 570 (1951). Some Metabolic Products of Basidiomycetes 13 10. F. Kavanagh, A. Hervey, and W. J. Robbins, Proc. Natl. Acad. Sci. U. S., 88, 555 (1952). 11. M. Anchel, J. Biol. Chem., 199, 133 (1952). 12. M. Anchel, J. Am. Chem. Soc, 74, 2943 (1952). 13. M. Anchel, A. Hervey, and W. J. Robbins, Mycologia, 47, 30 (1955). 14. J. H. Birkinshaw, Chemistry of the Fungi, Ann. Rev. Biochem., 1953, 383. 15. J. MacMillan, J. Chem. Soc, 1954, 2585. 16. F. Kavanagh, A. Hervey, and W. J. Robbins, Proc. Natl. Acad. Sci. U . S., 36, 1 (1950) ; H. Anchel, J. Polatnick, and F. Kavanagh, Arch. Biochem., 25, 208 (1950); M. Anchel, J. Am. Chem. Soc, 74, 1588 (1952). 17. M. Anchel, Arch. Biochem., 48, 127 (1953). 18. F. Kavanagh, A. Hervey, and W. J. Robbins, Proc. Natl. Acad. Sci. U. S., 36, 102 (1950). 19. W. J. Robbins, F. Kavanagh, and A. Hervey, Proc Natl. Acad. Sci. U. S., 38, 176 (1947). 20. M. Anchel, and M. P. Cohen, J. Biol. Chem., 208, 319 (1954). 21. M. Anchel, J. Am. Chem. Soc, 75, 4621 (1953). 22. M. Anchel, Science, 121, 607 (1955). 23. M. Anchel, Proc. Am. Soc. Exptl. Biol, 14, 173 (1955). 24. M. Anchel, Trans. N. Y. Acad. Sci., 16, 337 (1954). 25. W. P. Celmer and I. A. Solomons, /. Am. Chem. Soc, 74, 1870 (1952). 26. W. F. Lange, Proc Soc. Exptl. Biol. Med., 29, 1134 (1932). 27. N. A. Sorensen, Chem. Ind., 1953, 240. D. Holme and N. A. Sorensen, Acta Chem. Scand., 8, 34 (1954). K. S. Baalsrud, D. Holme, M. Nestvold, J. Pliva, J. S. Sorensen, and N. A. Sorensen, Acta Chem. Scand., 6, 883 (1952). 28. J. D. Bu'Lock, E. R. H. Jones, and W. B. Turner, Isolation of Matricaria, Chemistry & Industry, 24, 686 (1955). 29. A. S. Pfau, J. Pictet, P. Plattner, and B. Susz, Helv. Chim. Acta, 18, 935 (1935). 30. K. W. Hausser, R. Kuhn, and G. Seitz, Physik. Chem., 29B, 391 (1935). 31. J. D. Bu'Lock, E. R. H. Jones, and P. R. Leeming, in press. 32. E. Anet, B. Lythgoe, and S. J. Trippet, J. Chem. Soc, 1958, 309. 33. J. D. Bu'Lock, E. R. H. Jones, G. H. Mansfield, J. W. Thompson, and M. C. Whiting, Agrocybin, Chemistry & Industry, 32, 990 (1954). 34. F. Bohlmann and H. G. Viehe, Chem. Ber., 87, 712 (1954). 35. J. B. Armitage, C. L. Cook, E. R. H. Jones, and M. C. Whiting, J. Chem. Soc, 1952, 2010. 36. W. H. Caruthers and G. J. Berchet (E. I. du Pont de Nemours & Co.), U. S. pat. 2,136,178, C.A., 83, 1345.5 37. H. B. Henbest and E. R. H. Jones, J. Chem. Soc, 1953, 3796. 38. H. Erdtman, Progress in Organic Chemistry, edited by J. W. Cook, Butter- worth, 1951. 39. J. S. Sorensen, T. Braun, D. Holme, and N. A. Sorensen, Acta Chem. Scand., 8, 26 (1954). 40. J. W. Foster, Chemical Activities of Fungi, 6, Academic Press, 1949. Heterogeneity of Deoxyribonucleie Aeid (DNA) AARON BENDICH A great many explorations have led to the view, which few people now question, that the assertion of biological characters and their transmission from one generation of cells or organisms to the next require the intervention of DNA. In fact, there is a growing convic- tion that the actual genetic determinants of the cell are composed of DNA. This conclusion has received its most impressive support from the knowledge that DNA preparations from many microorganisms can carry out heritable transformations. There are, of course, a great many heritable factors in the total genetic complement of any cell. It would seem necessary, for the gene: DNA relationship to be valid, to postulate that the total DNA of the cell consists of a great many DNA molecules each concerned with one phenotypic expression and differentiated by a particular chemical structure. Or, alternatively, these macromolecules may be few in number but possessing structures that are heterogeneous along the chain. According to this idea, bio- logical characteristics are associated with specific regions on a DNA molecule; hence, a given molecule may be polyfunctional. It is not the purpose here to choose between these alternatives; indeed, both may be involved. Rather, this essay will deal with several lines of evidence which have revealed the heterogeneous nature of DNA. Perhaps the earliest published account x that is suggestive of this idea is the report that the great bulk of the DNA of isolated "chromo- somes" is soluble in M NaCl, but a little DNA was still detected in the insoluble "residual chromosomes." This residual DNA was thought 1 to be a contaminant, but, in the light of more recent developments, it would be good to re-examine this finding since it may reflect hetero- geneity in distribution of DNA within the cell. The same applies to 14 Heterogeneity of Deoxyribonucleic Acid (DNA) 15 the very brief report 2 of the solubility of only part (over 60%) of the DNA of liver nuclei in strong sucrose solution. Under defined conditions, the DNA's of calf thymus and of wheat germ yield dialyzable fragments and small non-dialyzable "cores" (about 7% of the total) following prolonged treatment with deoxy- ribonuclease.3, 4 It has not yet been decided whether this result is due to the presence in these DNA's of more than one DNA or whether the nucleic acids are composed of molecules possessing regions differing in susceptibility to the enzyme. The type of heterogeneity those ex- periments reveal may be elucidated if the same technique is applied to the various DNA fractions which are now available (vide infra). Heterogeneity with respect to the sites of binding of positively charged dyes to DNA has been described.5 These results may arise from identical molecules whose linear structure is discontinuous or may be due to mixtures of dissimilar molecules, each possessing its own set of binding characteristics. Evidence of an inhomogeneous distribution of DNA within the nucleus has also come from studies 6 on the susceptibility to deoxy- ribonuclease. Depending upon the species examined, only 40 to 89% of the DNA of isolated nuclei can be removed with the enzyme. The resistance of some of the DNA to the action of the nuclease is not due to the same phenomenon as the resistance of the non-dialyzable "cores" referred to above, since removal of the basic protein with dilute HC1 renders the resistant DNA fraction in nuclei susceptible to digestion.6 The DNA of Streptococcus faecalis is present in that organism in two forms ; one is soluble in dilute sodium hydroxide solution, whereas the other is not and is apparently bound to polysaccharide.7 These various findings are reminiscent of our early studies 8 on the heterogeneity of DNA in mammalian tissue. It was found that two gross fractions could be obtained by subjecting the total DNA (ex- tracted with strong salt solution) to high-speed centrifugation in 0.87% NaCl. In this fashion, an insoluble (DNAi) and a soluble (DNA2) fraction were obtained. With the exception of the normal livers, a number of organs of the adult rat yielded the two fractions in varying ratios9 depending upon the organ (Table 1). Normal liver contains DNA2, but little if any DNAi. During regeneration following partial hepatectomy, DNAi appears in large amounts but the DNA2 content remains constant. It is not known whether the high ratios for small intestine and for regenerating liver are due to the high mitotic activity of these tissues. 16 Essays in Biochemistry- Table 1. Base Composition of DNA Fractions from Various Organs of the Adult Rat * f DNA Fraction DNAi DNA2 Organ Thymine Guanine Cytosine Thymine Guanine Cytosine Intestine 0.84 0.79 0.72 1.03 0.74 0.83 Kidney 0.79 0.73 0.70 0.97 0.69 0.77 Spleen 0.84 0.74 0.68 0.98 0.78 0.88 Pancreas 0.65 0.80 0.67 0.82 0.74 0.82 Brain 0.76 0.67 0.68 Testis 0.77 0.77 0.68 * Molar ratios with adenine taken as 1.00. t J. R. Fresco, P. J. Russell, Jr., and A. Bendich, unpublished results. This possibility was explored, in part, by examining the livers of rats which were subjected to partial hepatectomy. The DNAi to DNA2 ratio remained high 3 to 23 days following the operation, but, thereafter, the ratio dropped rapidly. The regenerated livers 2 and 3 months after partial hepatectomy resembled normal liver in that little or no DNAx could be isolated.* The virtual absence of DNAX in the normal livers of adult rats has been confirmed. It was found, further, that the DNA content of rat liver increases as a result of treatment of rats with alloxan. This increase was attributed specifically to the formation of DNAi, whereas the DNA2 content was unchanged.10 The isolation procedure which permits the isolation of DNAi and DNA2 was originally based upon an arbitrary concentration of salt, 0.87% NaCl. Yet, unexpectedly, it does not seem to be so arbitrary after all inasmuch as it reflects the formation of DNAi in the rapidly growing tissue; i.e., the fractionation procedure appears to have a biological basis in fact in this differentiation of DNA. Analysis of the soluble DNA fractions for base composition reveals (Table 1) them to be different from the insoluble ones, and, further, that there are significant differences among tissues. f These results reveal a heterogeneity in the composition of DNA from various organs of the rat and in this regard are in sharp contrast to findings 21 from only a few organs of other mammalian species. The basis of this apparent conflict is not clear at present but may be related to the fact * Unpublished results, with J. R. Fresco. t Unpublished results, with J. R. Fresco and R. J. Russell, Jr. Heterogeneity of Deoxyribonucleic Acid (DNA) 17 that differences among DNA fractions may be obscured when the total DNA containing such fractions is analyzed. To compare the metabolic behavior of these two gross DNA frac- tions, and the metabolism of the DNA of various organs, isotopic formate was administered to a group of rats. Half the animals were examined 1 day after administration of the formate and the remainder 23 days later. The DNA fractions isolated from a number of organs were broken down to free bases and the percentage change in the isotope contents during the 23-day period determined (Table 2) ; the Table 2. Per Cent Apparent Retention of C14 during 23-Day Interval Following Administration of Labeled Formate to Adult Rats * DNA Fraction Base Small Intestine Spleen Regenerating Liver Normal Liver Pancreas Kidney Testis Thymine 4.1 20 92 52 66 85 DNAX Adenine 2.4 11 55 42 39 82 Guanine 2.1 15 56 40 65 68 Thymine 4.2 22 85 53 38 93 62 DNA2 Adenine 2.7 6 57 15 35 92 105 Guanine 3.6 15 82 33 42 116 89 * Taken from A. Bendich, P. J. Russell, Jr., and G. B. Brown, J. Biol. Chem., 203, 305 (1953). change is referred to as "apparent retention." These results 9 show that the two fractions are metabolically dissimilar in any one organ and that the metabolic picture is different from organ to organ. With the possible exception of the DNA2 fraction of pancreas, the "apparent retention" of isotope differs from base to base for the individual frac- tions. (An analogous result has been obtained in mouse- and rat-liver DNA in studies with P32.12) These studies suggest that the DNA fractions of various organs of the rat have a heterogeneous metabolic origin. Thus, in addition to differences between the two fractions insofar as solubility and chemical composition are concerned, they could be distinguished on the basis of their anabolic behavior. Further evidence for chemical heterogeneity of DNA has recently been obtained by three independent techniques. One may be termed fractional dissociation of calf-thymus nucleoprotein; the second in- volves chromatography on columns of histone, now itself found to be heterogeneous; the last to be described employs an anion exchanger that is ostensibly homogeneous. Calf-thymus nucleoprotein, in the form of a loosely packed gel with chloroform-octanol, yields a few nucleoprotein fractions upon extrac- 18 Essays in Biochemistry tion of the gel with NaCl solutions of increasing concentration.13 When freed of protein, corresponding DNA fractions are obtained which show decreasing proportions of guanine and cytosine and in- creasing amounts of adenine and thymine. (Similar fractions may be obtained " by successive extractions of the gels with salt solutions of constant concentration.) An analysis of the 5-methylcytosine con- tents reveals a disproportionate distribution of this pyrimidine among the fractions, and this 13 constitutes newer, but striking, evidence of the heterogeneity of DNA. Despite the fact that twenty-one preparations of calf-thymus DNA afforded by several procedures showed 13 very similar compositions, this similarity obscured the large differences in the constitution of the components of the mixture which is known as calf-thymus DNA. Almost simultaneously, work was described 15 in which histone (from calf nucleohistone) was immobilized on columns of kieselghur. Solu- tions of DNA from either calf thymus, Escherichia coli, or human white blood cells were placed upon such columns and then eluted with NaCl solutions of increasing concentration. Convincing evidence of the heterogeneous character of these DNA preparations was obtained from an examination of the fractions in the eluates. A gradation in the base composition was also observed in several fractions from calf- thymus DNA. Although it furnished a clue to the heterogeneous nature of DNA, the fractionation procedure which yielded the soluble and insoluble fractions DNAi and DNA2 could, at best, be described as crude. Better methods were therefore sought. If DNA were indeed composed of different individuals, a DNA preparation should consist, in neutral solution, of a number of polyelectrolytes anionic in character due to phosphate dissociation. Accordingly, columns of anion exchangers were employed in attempts * to effect a separation of individual poly- anions, or groups thereof. Initial attempts with commercially avail- able strong and weak base anion exchanger resins furnished a few encouraging results. In one experiment with the chloride form of the Amberlite IR-4B, over one hundred chromatographic fractions were obtained with calf-thymus DNA. But the experiments were difficult to repeat, and the various resins showed many undesirable properties. The success in the fractionation of proteins 16 by means of anion (and cation) exchangers prepared from cellulose prompted an investi- gation of their suitability for the fractionation of DNA. A cation * Unpublished experiments, with J. R. Fresco and H. R. Rosenkranz. Heterogeneity of Deoxyribonucleic Acid (DNA) 10 exchanger containing carboxymethyl groups attached to the cellulose was without any affinity for calf-thymus DNA at neutral pH. How- ever, one containing the basic diethylaminoethyl group removed DNA from solution, and fractions could be obtained by salt elution. A new anion exchanger, Ecteola-Cellulose, prepared by the treatment of alka- line cellulose with a mixture of epichlorohydrin and triethanolamine,17 has given the most provocative results. A diagram of the chromato- graphic behavior of highly polymerized calf-thymus DNA on a column of this anion exchanger is shown in Fig. 1. Many chromatographic fractions of DNA were obtained by continuous gradient elution with NaCl solution of increasing concentration followed by graded changes in ?)H. About 65% of the original DNA was recovered at neutrality, some 20% up to pH 9.9, and the remainder required strong alkali for complete removal. Analogous chromatographic patterns result from the application of a discontinuous salt concentration or pH gradient. Fractions of DNA obtained this way are non-dialyzable and retain their chromatographic properties on rechromatography ; the reproduci- bility of the patterns is indeed gratifying. With this method at hand, a solution of transforming DNA from pneumococcus * was passed through a column of Ecteola-Cellulose and all of the transforming activity as well as of the DNA was retained by the exchanger. Fractions of DNA exhibiting transforming activity were obtained at several places along the chromatogram following the elution; wherever the activity was detected it always coincided with the presence of DNA. Perhaps this can be taken as additional argu- ment that DNA can carry and transmit genetic information. At any rate, retention of biological (transforming) activity following its re- moval from solution by the exchanger and its subsequent elution indi- cates that little tampering with the integrity of the DNA preparation had resulted. The pneumococcal DNA preparation originally possessed four de- monstrable genetic properties; these included transformations to peni- cillin, streptomycin and sulfonamide resistance, and mannitol utiliza- tion.18 The DNA preparation, isolated from bacteria arising from a single clone, induced these heritable transformations at random essen- tially as single, independent events. However, a significant number of the mutant cells showed two properties, those of mannitol utilization and of streptomycin resistance which were acquired, not at random, but rather as related or connected events.18 It would appear, then, * Unpublished experiments, with S. Beiser, J. K. Fresco, and R. D. Hotchkiss. 20 Essays in Biochemistry x a. a ■3 u O X U w bli s n CO oi ^ *"* o X 0) ,d rl CO o X' a 3 m ft C CH3COCH2COOH -> CH3COCII3 + C02 CH3 CH3 \C0 + CH3COOH -> \c=CH— OOOH -> CH3 CH3 CH3 — CH2— C=CH— CH2— 24 Essays in Biochemistry Once the distribution of acetate carbon in the isooctyl side chain of cholesterol had been ascertained, it became immediately clear that the isotopic pattern observed in the sterol could be readily accounted for by the series of reactions shown above. It thus became a logical step to postulate the occurrence of parallel events at some stage of rubber and steroid biogenesis. The details of the above scheme, in particular the suggestion that branching was achieved by a biochemical equiva- lent of the Reformatskii reaction, have not withstood later scrutiny mainly because acetone fails to show the properties of a specific steroid precursor.5,6 In the systems investigated so far, i.e., in the intact animal and in the liver slice, acetone at best equals the efficiency of acetate, a result which reflects the rapid oxidation of this 3-carbon compound to acetate and "formate." Moreover, when Brady and Gurin found that l-C14-acetoacetate was incorporated into cholesterol apparently without fragmentation of the carbon chain,7 participation of acetone seemed clearly ruled out. However, it has lately become less certain that the results with labeled acetoacetate were interpreted correctly, and the possibility that acetone participates in a direct way cannot as yet be dismissed entirely. Searching for an alternative mechanism which would permit the assembly of acetate units to a branched chain intermediate, we have suggested /?-hydroxy-/?-methylglutaric acid (I) as an intermediate that could be formed from acetoacetate acid and acetate, or acetyl CoA.s This condensation would bear some resemblance to the formation of citrate (II) from acetyl CoA and oxaloacetate. The dihvdroxy-6- CH2COOH CH3CO + CH3COOH I CH2COOH CH3COH CH2COOH CHoCOOH HOOC— CO + CH3COOH I CH2COOH HOOC— COH I CH2COOH n COH— COOH CH3CO + CH3COOH I COH— COOH 1 CH3C— OH CH2COOH in carbon dicarboxylic acid (III) which has been implicated by the work of Tatum and Adelberg as the common precursor of valine and iso- leucine in Neurospora 9 is most likely formed by an analogous reaction, Biosynthesis of Branched-Chain Compounds 25 a more highly oxidized 4-carboo compound taking the place of aceto- acetate. It is improbable that this hypothetical precursor of branched amino acids in Neurospora (111) is a source of isoprenoid intermedi- ates, since precursors other than acetate enter into its formation. It was fortunate that concurrently with the interest in branched- chain intermediates hydroxymethylglutarate (HMG) was identified as a constituent of various plants.1"11 At the same time it was recalled that acids such as /?-dimethylacrylic acid (DMA) , isovaleric acid ( IV ) , and /^-hydroxyisovaleric acid (HIV) are to be found in nature, and hence the moment seemed propitious for an inquiry into their origin and possible relation to isoprene and steroid biosynthesis. With the aid of the now widely used carrier technique the formation of the above-mentioned acids from labeled acetate in animal tissues could be readily demonstrated.12,13 Moreover, the number and spacing of the carbon atoms of acetate in the branched molecules conformed, where investigated,1- with the pattern that had been predicted for the hypothetical steroid precursor. So far, the postulated initial step which leads to a branched compound, i.e., the condensation of acetate and acetoacetate (or their coenzyme derivatives) to a 6-carbon dicarboxylic acid has been elusive, although the reverse reaction has been shown to occur. Considerable light has been thrown on the interrelationship of the various branched-chain compounds by the discovery of Coon 14>15 that in the conversion of the 5-carbon acids to acetoacetate a fixation of carbon dioxide is an integral part of the overall process suggesting the sequence of steps shown in Fig. 1. Some of these reactions are closely H3C H CH3 CH3 OH \l \ \ 1 CCH2COOH - C= =CH— COOH - C- -CHoCOOH / iv / DMA / HIV CH3 CH3 CH3 i+co2 CH3 H3C OH \ T \| CO + CH3COOH ^± C— CH2COOH / / HMG CH2 OH2 I. I COOH COOH Fig. I. 26 Essays in Biochemistry- analogous to the events which are known to occur with the straight- chain fatty acids, and hence their reversibility may be anticipated. In this event the above scheme would provide for the formation of hydroxymethylglutarate by two pathways: (a) by a C2 + C4 con- densation and (b) by C02 fixation of a 5-carbon acid. The second reaction is perhaps not quantitatively important at least in animal tissues, since the supply of isovaleric acid would appear to be limited by the rate of oxidation of leucine which is an essential amino acid. At any rate reactions have now been shown to occur which afford branched 5- or 6-carbon acids, and assuming these acids to be inter- mediates rather than metabolic end products we may now consider their relation to terpene and steroid biogenesis, the more central issue of our discussion. When we first took up this problem in 1944 it seemed reasonable to pose the question whether the preformed chains of leucine or valine might serve as carbon sources for some portion of the steroid molecule, and in fact deuterioleucine and deuterioisovaleric acid proved to be efficient precursors of cholesterol. Later, with the finding that only acetate carbons make up the skeleton of ergosterol and cholesterol, it became clear that a carbon source which originated from an indis- pensable amino acid could not be an obligatory intermediate, at least in the animal body. Nevertheless, it appeared possible that the metabolism of leucine, for example, led to a product that was identical with one of the intermediates in the acetate-sterol conversion, and hence experiments with leucine or isovaleric acid seemed worth pursu- ing. With the aid of isotopic carbon Zabin 6 showed isovaleric acid to be several times more effective than acetate for cholesterol synthesis. This was true, however, only when the precursor was labeled in the isopropyl portion of the molecule. Carboxyl-labeled isovalerate un- expectedly gave results that were indistinguishable from those obtained with l-C14-acetate. More recently, we have tested additional branched-chain acids as cholesterol precursors with results that have been both encouraging and puzzling. HMG, HIV, and DMA when labeled at the tertiary carbon atom were incorporated into cholesterol, but only with DMA as the substrate was the transformation extensive enough to indicate specific conversion. On the other hand, partial degradation of the cholesterol samples from the three experiments indicated that in all cases C14 was present only at those six positions which one would expect to be labeled if the carbon chains of the acids had remained intact during condensation to the triterpenoid intermediate (Fig. 2). Biosynthesis of Branched-Chain Compounds 27 Metabolic breakdown of the precursors to acetate or acetoacctate would have caused the appearance of labeled carbon in twelve positions of the steroid molecule. Although the relatively low overall efficiency of HMG and HIV as cholesterol precursors might be ascribed to rate differences in the transformation of the free acids to activated deriva- tives (admittedly a restatement of the problem rather than an ex- planation) a more discordant result came to light when 1-C14 DMA was tested. If it is valid to assume that the intact DMA molecule enters into the condensation reaction, then the conversion efficiency C? 6X C - C Fig. 2. Postulated distribution of C14 in triterpenoid precursor formed from branched acids labeled at carbon 3. should be independent of the location of the carbon label in the pre- cursor. This was not true for 1-C14 DMA. Moreover, C14 from 1-C14 DMA entered the C-25 position of the cholesterol side chain at a level indicating that extensive breakdown to 1-C14 acetate had oc- curred. This result in fact confirms our earlier experiences with isovaleric acid and leads to the anomalous situation in which the branched-chain acids behave as if they were direct cholesterol pre- cursors only when they are labeled in the isopropyl group. The forma- tion of C2 units from carbon atoms 1 and 2 of isovalerate and of DMA is readily explained by cleavage of the molecule between carbon atoms 2 and 3, perhaps by the reactions suggested by Coon (Fig. 1), i.e., with concomitant C02 fixation. On the other hand, the route taken by the isopropyl portion of isovalerate or dimethylacrylate during conversion to steroids is not immediately apparent. Acetoacetate formation from the isopropyl portion as depicted in Fig. 1 cannot be the explanation because the observed isotope-distribution pattern in cholesterol16 dif- fers so markedly from what is found with acetate or acetoacetate 17 as precursors. 28 Essays in Biochemistry There is one further observation which should be mentioned before an interpretation of the relevant data is attempted. Labeled C02 when administered to rats is incorporated into cholesterol at a level that is barely detectable. The incorporation values can, however, be sub- stantially increased by the simultaneous feeding of normal isovaleric acid. At the same time the concentration of C14 in the acetate pool is raised only slightly as judged by the C14 content of the acetyl groups of N-acetylphenylaminobutyric acid. Hence the C02-fixing step does not give rise to labeled acetic acid. One is inclined to interpret this result in accordance with Coon's scheme, as reflecting a C5 + Ci con- densation to form a dicarboxylic acid which is subsequently trans- formed into the steroid precursor. The structure of the substrate that fixes C02 is, however, not defined except that it would appear to be a transformation product of isovaleric acid. Any scheme designed to harmonize the available experimental find- ings must be capable of accounting for the following observations: (1) the ease of conversion to cholesterol of DMA as compared to other branched acids, at least in the intact animal; (2) the preferential utilization of the isopropyl portions of DMA and isovalerate; (3) the C02 fixation into cholesterol, which is promoted by isovaleric acid. Some of the experimental facts enumerated here can be fitted into a scheme that is highly speculative to be sure, but since it is susceptible to experimental test it may prove to be of temporary value as a working hypothesis. It should be understood that at the present time the concept which visualizes terpene and steroid biogenesis as a process involving the multiple condensation of C5 (or perhaps C6) units has merely the status of an hypothesis. There are experimental observa- tions which tend to support it, but no proof exists that the basic principle, i.e., polymerization of monomeric units to polyisoprenoids, has biological reality. The speculations which follow are designed to reconcile some of the conflicting experimental data and thereby to strengthen the underlying hypotheses. As the initial step in the metabolism of DMA, using the 1-C14 com- pound for purposes of illustration, we wish to propose (Fig. 3) a shift of the double bond to the exomethylene position, a reaction which could either proceed directly or by way of /?-hydroxyisovaleric acid. Coon 15 regards the hydroxy acid as the substrate for C02 fixation because in his enzyme system DMA formed labeled acetoacetate only in the presence of crotonase. It is possible, however, that this finding is the result of a more complex set of events, such as the hydration of DMA to the ^-hydroxy acid followed by dehydration to the /?-y Biosynthesis of Branched-Chain Compounds 29 unsaturated acid, which in turn might fix carbon dioxide. This mecha- nism is preferred because on purely chemical grounds fixation of C02 by a hydroxy acid is improbable, whereas both chemical and bio- CHa CHa ±H20 CH3 CH3 CH— CH2C*OOH 0) C=CH— COOH ~ C— CH2O0()H CH, (2) / | ch3 on IV HIV \ t ±H20 j(3) CH3 (6) H2C S C— CH2C-OOH MVA T ±co2 1(4) CH3 CHa -co* C— CH3 <- CH DMA COOH CH3 CO +CH;!OOOH / CH2 I COOH Fig. 3. (5) C— CH2C«OOH (8) CH MCA COOH T ±H20 1(7) H3C OH \l C— CHoC'OOH / CH2 HMG I COOH IV: isovaleric acid. DMA: /3-dimethylacrylic acid. HIV: /3-hvdroxyisovaleric acid. MVA: /3-methylvinylacetic acid. MGA: /3-methylglutaconic acid. HMG: /3-hvdroxy-/3-metliylglutaric acid. chemical analogies exist for the carboxylation of ethylenic compounds. Moreover, it appears that crotonase catalyzes reversible reactions which would allow an equilibrium to be established between the x, /?, and /3,y unsaturated acids, and the /3-hydroxy acid. If the fi,y acid (methylvinylacetic acid) were indeed the acceptor for carbon dioxide, one of two geometric isomers of methylglutaconic acid would be the product. Their structure makes these acids attractive on several 30 Essays in Biochemistry counts as intermediates in terpene and steroid biogenesis. Decarbox- ylation, by removal of the carboxyl group which was originally present, would regenerate DMA with retention of four of the five original carbon atoms. The molecule becomes reoriented, and the carboxyl group which was newly introduced will now be linked to the opposite end of the original molecule. If the cyclic regeneration of DMA occurred at a sufficiently rapid rate, then not only would the original carboxyl carbons be lost entirely but at the same time the isopropyl carbons 4 and 4' would become equilibrated with carbon atom 2 (see Fig. 3, reactions 2 to 6). In this event the labeling pattern in the eventual product of isoprene synthesis, e.g., cholesterol, should not be specific, i.e., it should be identical with that given by 2-C14-acetate. This has actually been found to be the case with 4,4'-C14-isovaleric acid. It should be emphasized perhaps that in the conversion of acetate to terpenes and steroids, dimethylacrylic acid need not be a direct intermediate; possibly it joins the main synthetic path merely by virtue of its conversion to methylglutaconic acid, and normally this dicarboxylic acid is formed chiefly from acetate and acetoacetate by way of hydroxymethylglutarate. The obvious reason why the discus- sion has nevertheless centered around dimethylacrylate is its superi- ority over other branched-chain acids as a precursor of cholesterol. The arguments presented so far imply that DMA (or a coenzyme derivative) is the monomeric unit which enters into the synthesis of polyisoprenoid chains. This reaction would be analogous to the fi- ketoacyl condensation which is the well-established mechanism for the synthesis of the straight-chain aliphatic acids. One may question, however, the likelihood that isopropylidene groups are sufficiently re- CH3 \ C=CH— COR 4- H3C— C=CH— COR -> / I CH3 CH3 CH3 \ C=CH— COCHoC=CH— COR / "I CH3 CH3 active to enter into a condensation of this type. By contrast the methylglutaconic group possesses a methylene group which should be Biosynthesis of Branched-Chain Compounds 31 CH2COOH COOH I I CH3C=CH— COR + H2C— C=CH— COR -> I CH3 CH2COOH COOH CH3C=CH— CO— CH— C=CH— COR ^> I CH3 CH3 \ C=CH— CO— CH2C=CH— ( !OR / I CH3 CH3 more favorable for reaction in a Claisen-type condensation. In this event the same Cio unit will be formed, but decarboxylation would follow or at least not precede the condensation step. It will be noted that the carbon atoms removed by the decarboxylation step will again be the original carboxyl groups of DMA. This second mechanism therefore accounts equally well for the experimental data obtained with 1-C14 DMA or isovaleric acid, but cyclic regeneration of a 5- carbon compound would not be obligatory. The existence of methylglutaconic acid in two stereoisomeric forms is the second structural feature of particular interest for polyisoprenoid synthesis. The natural acyclic terpenes occur both in the cis (e.g., nerol and rubber) and trans forms (geraniol, gutta-percha, and squal- ene I , and it is pertinent to ask at what stage of the overall synthetic process the geometric configuration of the respective products becomes established. If DMA were the condensing unit, the stereospecific step would follow the condensation reaction. On the other hand, the cis and trans .isomers of methylglutaconic acid would constitute precursors which already possess the configuration of the end product. Present biochemical knowledge is too scarce to predict the nature of the stereo- specific steps, although at least one prototype exists. Dehydration of hydroxy acids (malic, citric, and /?-hydroxybutyric acids) is a well- established enzymatic reaction which affords geometric isomers, and hence the formation of methylglutaconic acid by elimination of water from hydroxymethylglutarate would not be without precedent. Which one of the two isomers will be formed cannot be predicted, nor can the possibility be ignored that the cis and trans forms are converted into each other enzymatically by a reaction analogous to the isomerization 32 Essays in Biochemistry of maleylacetoacetate to fumarylacetoacetate which Knox has de- scribed.18 It is particularly relevant to the present discussion that squalene has been shown to possess the all trans configuration.19 The present hypothesis would therefore anticipate trans-methylglutaconate as a precursor of the triterpene and of the steroid derived from it. Two observations have been made which are consistent with the role as- signed to the methylglutaconic acids. Rabinowitz and Gurin 20 have shown that hydroxymethylglutarate is dehydrated by liver prepara- tions, and they state that ^?w?s-methylglutaconic acid is the product. Second, the cis form of the same acid has been tested in our laboratory as a cholesterol precursor and found to be inactive. Neither of these two observations are conclusive by themselves, but, taken together with the behavior of DMA as discussed earlier, they strengthen the case for a pivotal role of the cis and trans isomers of the methyl- glutaconic acids in terpene and steroid biogenesis. Though the main object of this essay has been to indicate current lines of thinking and investigation in a selected area of terpene and steroid biogenesis, it has been the intent to look also briefly at the origin of the branched-chain amino acids. Comparison of the two areas of biosynthesis raises the interesting point why higher animals can synthesize one type of carbon chain, but not another closely related one. Acetyl units are not the precursors of the branched portions of valine and leucine.21 Instead, it appears that the carbon skeletons of these amino acids are formed from two molecules of pyruvate by a mecha- nism which may involve an acyloin condensation to acetolactate and a subsequent pinacol rearrangement to an isovaleric acid derivative.22'23 There is also evidence that one of the immediate valine precursors, after condensation with an acetyl unit and decarboxylation, furnishes the carbon chain of leucine.1'4 It is of particular interest for the pur- oh on 2CH3CO— COOH -> CH3C— COOH -^ (CH3)2C— CO— COOH Valine precursor CO I CH3 OH (CH3)2C— CO— COOH + CH3COOH - ^> Leucine precursor Biosynthesis of Branched-Chain Compounds 33 poses of the present discussion that in the synthesis of the carbon chains of valine and leucine pyruvic acid does not serve as a source of acetate (or acetyl CoA) but enters as such into the condensation reactions. Leaving aside details of the synthetic mechanisms, we may compare the structures of two of the 5-carbon compounds which have been implicated in steroid biosynthesis (IV and V), and hence are formed in the animal body as well as in the plant and microbial cell, with two precursors of the amino acid valine (VI and VII i : CH3 \ C=CH- / CH3 COOH H3C OH \l C— CH2COOH / CH3 IV V H3C OH OH \l 1 C — C— COOH / CH3 CH3 \ CH— CO— COOH / CH3 VI VII Clearly a transformation of IV or V, to VI or VII does not occur in the animal body, or else valine would not be an essential amino acid. The action of crotonase, which catalyzes the addition of water to x,fi unsaturated acids of type IV is evidently restricted to the formation of /^-hydroxy acids, because the a-hydroxy acid, if formed, should be readily oxidized to a-ketoisovalerate, a compound which is converted to valine in the animal body. One may argue then that as a result of the specificity of crotonase the synthetic pathways leading to isoprenoid intermediates are useless as far as valine and leucine syn- thesis are concerned, and that as a consequence a synthetic pathway for these amino acids has evolved that is distinct both with respect to the mechanism of condensation and with respect to the carbon sources which it uses. It appears to be equally true that the valine precursors VI and VII are not convertible to IV and V since in microorganisms such as N&uro- spora the sole carbon source for ergosterol is acetate, which is used for the synthesis of IV and V but not of VI and VII. Thus, even in the absence of the restrictions which exist in the higher animal, the separa- tion of synthetic pathways for two closely similar structures is rigidly maintained. 34 Essays in Biochemistry References 1. D. Rittenberg and R. Schoenheimer, J. Biol. Chem., 121, 235 (1937). 2. J. Bonner and B. Arreguin, Arch. Biochem., 21, 109 (1949). 3. E. C. Grob, G. D. Poretti, A. V. Muralt, and W. H. Schopfer, Experientia, 7, 21S (1951). 4. I. Harary and Iv. Bloch, unpublished. 5. T. D. Price and D. Rittenberg, J. Biol. Chem., 1S5, 449 (1950). 6. I. Zabin and K. Bloch, J. Biol. Chem., 185, 131 (1950). 7. R. O. Brady and S. Gurin, J. Biol. Chem., 189, 371 (1951). 8. K. Bloch, Harvey Lectures, 1,8, 68 (1952-53). 9. E. L. Tatum and E. A. Adelberg, ./. Biol. Chem., 190, 843, (1951). 10. H. J. Klosterman and F. Smith, J. Am. Chem. Soc, 76, 1229 (1954). 11. R. Adams and B. L. Van Duuren, /. Am. Chem. Soc, 75, 2377 (1953). 12. H. J. Rudney, J. Am. Chem. Soc, 76, 2595 (1954). 13. J. L. Rabinowitz and S. Gurin, J. Biol. Chem,., 208, 307 (1954). 14. M. J. Coon, /. Biol. Chem., 187, 71 (1950). 15. B. K. Bachhawat, W. S. Robinson, and M. J. Coon, J. Am. Chem. Soc, 76, 3098 (1954). 16. K. Bloch, L. C. Clark, and I. Harary, J. Biol. Chem., 211, 687 (1954). 17. M. Blecher and S. Gurin, J. Biol. Chem,., 209, 953 (1954). 18. W. E. Knox and S. W. Edwards, in Glutathione: A Symposium, Academic Press, New York, 1954. 19. N. Nicolaides and F. Laves, J. Am. Chem. Soc, 76, 2596 (1954). 20. J. L. Rabinowitz and S. Gurin, J. Am. Chem. Soc, 76, 5168 (1954). 21. C. Gilvarg and K. Bloch, J. Biol. Chem., 193, 339 (1951). 22. M. Stassmann, A. J. Thomas, and S. J. Weinhouse, ./. Am. Chem. Soc. 75, 5135 (1953). 23. I. R. McManus, ./. Biol. Chem., 208, 639 (1954). 24. P. H. Abelson, ./. Biol. Chem., 206 (1954). The Biochemistry of Lysogeny ERNEST BORER Hope, rather than experience, prompted the title of this essay. "Chemical gropings in lysogeny" would more accurately describe the writer's contribution to this engrossing biological phenomenon. The bold, perhaps brash, title thus merely serves to delineate an ultimate goal: an understanding of the lysogenic mechanism at the molecular level. Such a goal may remain asymptotic to several generations of biochemists; nevertheless, the writer believes that biochemists can already occupy themselves fruitfully with this phenomenon not only with the aim of contributing to an understanding of lysogeny but, equally, with the hope that a study of lysogeny will contribute to our store of knowledge of biochemistry. A paramount problem in bio- chemistry today is the elucidation of the structures of macromolecules and their correlation with biological function, including the replication of those macromolecules. The writer feels that such problems can be more fruitfully approached at present from a study of the biological functions of the simplest of the self-reproducing systems than from the application of the tools of the organic chemist to isolated fragments of cellular mechanisms. In other words, biological function can reveal chemical structure and mechanism, but the process, at the level of macromolecules, is well-nigh irreversible at present. If the reader demands evidence of this, let him consider how little biochemistry has contributed to a knowledge of the mechanism of genetics and how much microbial genetics has contributed to our under- standing of intermediary metabolism. Indeed, even in classical organic chemistry, function has been the key to structure and not the reverse. Long before the elucidation of organic structures from X-ray analysis was dreamed of, Kekule derived the structure of the benzene ring from its functions, from its behavior during substitution reactions. 35 36 Essays in Biochemistry Before some of the biochemical objectives in the field of lysogeny can be stated, the biological phenomenon should be described. Fortu- nately, the reader can be referred for the history of the problem and for a summary of the brilliant recent work, mostly by Lwoff and his school, to a masterly review by Dr. Lwoff himself.1 We need, therefore, give only a sketchy outline of the biological phenomenon. A wide variety of microorganisms carry, hereditarily, the seeds of their own and of other bacteria's destruction, either within, or closely associated to their genetic material. Under the influence of random metabolic or physical stimuli the metabolism of such organisms can undergo a profound shift, producing seemingly de novo bacterio- phages, which then emerge from the crumbling hulk of the host cell. From this stage on these phages of lysogenic origin apparently differ in no way from other free virulent bacteriophages: their cycle of replication takes place within susceptible host cells which they infect by invasion. No chemical investigation of the phenomenon was possible as long as we were limited to the spontaneous rate of occurrence of the lyso- genic phenomenon. Since the frequency of the occurrence of phage development and lysis in a lysogenic culture is low (1%, or less), any chemical study was precluded by the dilution of the object of the study by its stable, colony-mate cells. A profound discovery by Lwoff allowed a complete reversal of the above ratio in some lysogenic strains. If cultures of some lysogenic bacteria are exposed to small doses of ultraviolet or X radiation, or to some mutagenic or carcinogenic agent, the fraction of organisms in which the prophage * is induced to proliferate into mature bacterio- phage approaches unity. It should be emphasized that not every strain of lysogenic organism yields to the inducers listed above. The Lwoff effect is an almost startling phenomenon. If a milky culture of Bacillus megatherium containing 10s viable cells per milli- liter is exposed to a small dose of ultraviolet irradiation and then incubated in the dark, the suspension begins to clarify after about 60 minutes and within a few minutes the culture becomes water clear. At the same time, it can be demonstrated, by plating for plaques on a sensitive strain of bacteria, that concomitant to the lysis there is a large increase in the free phage titer. How can one begin to make a dent in a problem such as this with the tools of the chemist? The great Hopkins'-' gave sound advice on * Dr. Lwoff's term for the "form in which lysogenic bacteria perpetuate the power to produce phage." The Biochemistry of Lysogeny 37 this point. He said that biochemists should strive to be biologists as well as chemists to justify their special designation, for, whereas the chemist is best provided with the machinery for the cultivation of the borderland frontier between chemistry and biology, it is the biologist who best knows the lay of the land. The next best thing to becoming a biologist is to stick close to one. Lwoff observed very early that the physiological condition of the lysogenic organism at the time of irradiation determined the extent of the lysogenic response. Organisms which were on a glucose starvation regimen prior to the irradiation seemed to have acquired a resistance to the irradiation. Whereas in logarithmic growth phase only 1% of lysogenic Escherichia cull K12 will survive a given dose of irradiation as colony formers, about 75^ of the same organisms will survive the same dose if the organisms are deprived of glucose for 3 hours prior to the irradiation. The yield of free phage is, of course, proportionally diminished. The organisms have become "inapt." The acquisition of inaptitude by starvation is general for all inducible lysogenic organ- isms, nor is it restricted to glucose starvation. Nitrogen or specific amino acid starvation in the presence of ample glucose confer inapti- tude on lysogenic organisms as well.'1,4 A typical experiment on the development of inaptitude on methionine starvation in a lysogenic, methionine-requiring auxotroph, E. coli Ki2 W-6 (isolated by Dr. J. Lederberg), is given in Fig. 1. Inaptitude, though it is only a peripheral problem related to lysog- eny, can be submitted to chemical scrutiny. The known effects of starvation could be probed seriatim as possible sources of inaptitude. In turn, should the search for the mechanism of inaptitude be success- ful, the findings might not be without bearing on the mechanism of induction, the ultraviolet-irradiation-induced proliferation of the other- wise stable prophage. For often, an understanding of a block to a biochemical mechanism has been revealing of the mechanism itself. With this, perhaps naive, overall plan in mind, some experiments were devised. In the first place we had to explore, and preferably rule out, the unfruitful possibility that inaptitude is merely the result of the diminished metabolism of starvation. E. coli Ki^ in logarithmic growth phase was rapidly chilled to 3°C, kept at that temperature for as long as 20 hours, and then subjected to a normal inducing dose of irradiation at 3°. When they were returned to the warm room such cultures gave the usual lysogenic response. Cultures whose growth was inhibited by an antimetabolite, such as ethionine, also showed but little 38 Essays in Biochemistry change in aptitude. A diminished rate of metabolism at the time of irradiation, whether induced by chilling or by a metabolic inhibitor, was thus ruled out as a possible source of inaptitude. Next to be explored was the possibility that something accumulates during the aberrant metabolism of starvation which somehow inhibits the lysogenic response. To test this possibility the methionine-requir- i — + Methionine Minutes Fig. 1. The effect of starvation prior to irradiation on the lysogenic response. Curve 1 = E. coli K12 W-6 in logarithmic growth phase. Curve 2 = same after induction by ultraviolet. Curve 3 = the organisms irradiated after 3 hours of starvation of methionine. The amino acid was, of course, added, after the irradiation. ing auxotroph E. coli K12 W-6 at a concentration of 109 cells per milli- liter was subjected to methionine starvation for 3 to 4 hours. After the starvation, the starved organisms were eliminated from the medium by high-speed centrifugation or by sterile filtration. Wild type of E. coli K12 in logarithmic growth phase, i.e., with full aptitude, were harvested and suspended to a concentration of 108 cells per milliliter in the same sterile filtrate in which the auxotrophs had starved. Such a filtrate should have been an adequate culture medium for the wild type of organism. Indeed, that is precisely the medium on which the wild type of organism is grown. However, even after an immersion of only ") minutes, the minimum time dictated by the technique, the The Biochemistry of Lysogeny 39 aptitude of the organisms was diminished. After prolonged immersion, the inaptitude became more pronounced.5 The starvation medium was found to be at pH 5, but the hydrogen- ion concentration could not have contributed to inaptitude for it was found that the pH of cultures in logarithmic growth phase could be lowered to 5 with HC1 and, even though the organisms were kept at that pH for 3 hours, their aptitude was not suppressed. A variety of acidic metabolites which might have accumulated into the starvation medium were tested for their effect on aptitude. Fumaric acid at a concentration of 0.005 M suppressed aptitude, provided the pH was kept at 5. An assay of the starvation medium for fumaric acid by Racker's method indicated the absence of the acid at anywhere near the effective concentration. Nevertheless, the suppression of the apti- tude by fumaric acid was studied in some detail; it did not appear to consist merely of a screening of the irradiation, for it was highly pH dependent and quite specific: the higher homolog, glutaconic acid, was found to be ineffective." Derivatives of fumaric acid which could not dissociate, i.e., the diamide, the diglycyl, and the diglutamyl derivative, were prepared, and it was found that these suppressed aptitude at pH 6.5 as well. Thus the low pH needed for fumaric acid to be effec- tive as a suppressor of aptitude was necessary to repress the ionization of the acid rather than for any effect on the microorganism. It was found that all of the above fumaric acid derivatives are screening- agents against irradiation if the irradiation is passed through them, via a quartz dish, when they are out of contact with the microorganism, but they are particularly potent in contact with the organisms. The phenomenon thus seemed to resolve itself into a curiously effective screening by these compounds (by concentration on the bacterial cell?) — and, since it appeared to be not related to aptitude, work on it was shelved. On return to the study of the starvation medium itself, it was found that it absorbed very strongly at 260 m/x. A cursory examination by elution chromatography revealed the presence of a variety of nucleic- acid fragments. The only similar observation we could find in the literature was a brief posthumous note by the late Dr. Marjory Stephenson who found in a study of autolytic ribonuclease in E. coli that acid-soluble phosphorus accumulated in the buffer medium in which the organisms had been suspended.7 We therefore investigated the phenomenon in some detail to determine whether the excretion of nucleic acid fragments is a concomitant of all types of starvation, whether it is limited only to some bacterial species, or whether it is. 40 Essays in Biochemistry indeed, an excretion, or merely the oozings from dying cells and, finally, whether the phenomenon is an attribute of starving cells only. Since the technique of the experiments bears directly on the answers to the above questions, the salient experimental procedures must be given. For the starvation experiments bacteria were raised on a syn- thetic medium to logarithmic growth phase from small inocula. They were harvested by centrifugation when they reached a cell count of not higher than 2 X 108 cells per milliliter and were resuspended after sterile washing in a fresh medium lacking the nutrient of which they were to be starved. When cellular concentrations higher than 2 X 108 cells per milliliter were desired, the bacteria were resuspended to an appropriately smaller final volume. This precaution is essential to insure that the experiments are performed with a bacterial population which is preponderantly viable and which approaches physiological homogeneity. For the study of the kinetics of the excretion, aliquots of these aerobically incubating, starving, bacterial suspensions were taken at intervals; the bacteria were eliminated by high-speed cen- trifugation, and the absorption at 260 m^ of the supernatant fluid was measured. At the same time, aliquots of the bacterial suspension were diluted appropriately for plating for viable cells. No change in the number of viable cells could be detected during the first 6 hours of starvation. In Fig. 2 a study of the kinetics of the excretion of ultraviolet - absorbing substances by E. coli K12 W-6 on methionine and on glucose starvation is represented. Analysis of the excreted products indicated the presence of free bases and of nucleotides, but the relative quantities were different on the two types of starvation. There are several lines of evidence to indicate that the nucleic acid fragments which accumulate in the medium during the first 5 to 6 hours of starvation are excreted products rather than the accumulated debris from dying bacteria. The accumulation is considerably larger on methionine starvation than on glucose starvation, yet, on prolonged incubation, the bacteria remain fully viable for a longer time on methionine starvation. Moreover, comparison of the output of ultra- violet-absorbing material in a culture of 10° cells per milliliter with that in a culture of 108 cells per milliliter revealed that there was a greater output per cell at the lower concentration, but the measurable death rate, on prolonged incubation, was greater at the higher cell concentration. Finally, the kinetics of the excretion point either to an exhaustion of excretable products or to some equilibrium, for a The Biochemistry of Lysogeny 41 plateau is reached, a finding which could not be expected from a steady death rate of the bacteria. On the other hand, there is the possibility with respect to the last point that a number of bacteria, too small to be detected by our counting technique, die, and their oozed-out cell contents bring the optical density of the medium to its high level in the first 4 to 5 hours 1.5 - | 1.0 0.5 - 1 1 1 1 1 i i i„ « i ,.^-^" u) y^ - / / + — / y/m / + — ""'^ / x / / / / — / / -j/ - _ i i i i i 1 1 1 1 0 3 ■15 6 Hours 10 Fig. 2. The kinetics of the excretion of nucleic acid fragments by E. coli K12 W-6 on (1) methionine and (2) glucose starvation. Cell count 1:1.2X10° cells per ml.; 2:1.1 X 109 cells per ml. of starvation. The diminished accumulation from this time on could be visualized as resulting from the adaptation of the remaining viable bacteria to their debris-laden milieu, and temporary equilibrium might be reached between the rate of death and the rate of scavenging. The following findings rule out this possibility. The probable error of viable-cell counting in our hands is less than 10%. Therefore, a num- ber of bacteria, lO^c or less of the total, would, upon their death, have to yield an optical density approximating the levels of curve 1 in Fig. 2. Bacteria in logarithmic growth phase, at a cellular concentration of 1 X 108 per milliliter, were exposed to sonic vibration in a Raytheon 42 Essays in Biochemistry 9KC sonic vibrator for 10 minutes. Less than 0.5 % of the bacteria remained viable after this treatment. After centrifugation at 5000 r.p.m., the optical density of the fluids at 260 lm* was determined. It was 0.33, or, only about 20% of the optical density of the medium represented by curve 1, Fig. 2. Of course, this argument excludes consideration of the greater specific absorption of nucleic acid fragments after enzymatic depolymerization. Low glucose (0.5%) 9.20 8.80 JS, 8.40 .o^ 1.00 0.80 Fig. 3. The accumulation of nucleic acid fragments in the culture medium of E. coli K12 in logarithmic growth phase on low glucose. However, incubation at 38° of such sonically disintegrated cells for 5 to 10 hours produced only a 15 to 20% increase in optical density. The excretion of nucleic acid fragments during starvation was found to be not restricted to lysogenic organisms, for E. coli B and B/r repeated the same pattern. Since "abnormal" metabolism is often but an exaggeration of "nor- mal" metabolism, we next studied whether bacteria in logarithmic growth phase excrete any nucleic acid fragments into the culture medium. In Fig. 3 the growth curve of a bacterial culture with a limiting glucose concentration of 0.05%, from start through logarithmic phase to declining growth phase, along with the output of ultraviolet-absorb- ing substances (its ultraviolet-absorbing shadow, as it were), are given. The Biochemistry of Lysogeny 43 As the bacterial population reaches a stationary level upon the ex- haustion of the carbon source, the accumulation of ultraviolet-absorb- ing substances also diminishes. This again is not a characteristic of lysogenic bacteria alone, for E. coli B under similar conditions of cul- ture repeats this pattern. That the excretion of ultraviolet-absorbing material per bacterial cell in a culture in logarithmic growth phase is quite constant is appar- Low glucose (0.05J&) 9.20 3.40 0 12 3 4 5 6 7 Hours Fig. 4. Same data as Fig. 3 but plotting the log of O.D. ent from Fig. 4. The plot of the log of the increasing bacterial popula- tion and of the log of the optical density of the culture medium are parallel. The constancy of the excretion was also shown by a different method. Cultures of E. coli Ki2 were kept in extended logarithmic growth phase at a concentration of 108 cells per milliliter by the con- stant dilution of the culture with fresh medium at 37° at a rate which doubled the volume of the culture per generation time. Cell counts and the optical density of the cell-free supernatant fluid were determined for several hours. In such experiments, the optical density at 260 m/x remained 0.07 with 108 cells per milliliter. The nature of the substances which account for the ultraviolet absorption in the culture media of bacteria in logarithmic growth phase on low glucose was investigated. Paper chromatography of lyophilized concentrates revealed the presence of nucleic acid fragments and of 44 Essays in Biochemistry amino acids. However, the concentration of amino acids was too low to contribute significantly to the absorption at 260 ni/t. The origin of the nucleic acid fragments in the culture media is obscure. The accumulation may be the result of haphazard "leakage" or of the spillage of surplus synthesis. On the other hand, it may represent the microbial counterpart of the excretion of specific en- dogenous waste products of nitrogen metabolism in metazoa. Such High glucose (0.30%) 9.20 1.80 8.40 £ 8.00 7.60 , 1.00 7.20 0 Fig. 5. The accumulation of nucleic acid fragments in the culture medium of E. colt K12 in logarithmic growth phase on high glucose. an interpretation might be based upon an inversion of Dr. J. Monod's aphorism on comparative biochemistry: "What is true of coli is true of elephants — even more so." The final choice from among the above explanations must await the analysis of the excreted nucleic acid fragments from different micro- organisms. When the bacteria are cultured with the excess glucose concentration, of 0.3%, customarily used by bacteriologists, the pattern of accumula- tion in the culture medium is completely different (Fig. 5). As the declining growth phase in the culture is reached, the output of ultraviolet-absorbing materials is substantially increased until it reaches twice the level of the cultures with the low glucose. The dif- ferent levels of ultraviolet-absorbing substances in the two cases is a reflection of the different conditions which arrest the growth of the bacteria. With the lower glucose concentrations, growth stops upon The Biochemistry of Lysogeny 45 the exhaustion of the carbon source. The pH of such cultures even after 24 hours of incubation does not fall below 6.4. However, with the high glucose, the pH of the medium at the end of 7 hours is 5.5. The stationary population under these conditions is the result not only of diminished rate of cell division but of an increased death rate. The two methods of culture yield similar bacterial densities, but the environment of the bacteria differ. These findings together with those on starving cultures point to the necessity of examining the culture - W/ - so _ 60 Z 40 20 0 12 3 4 5 6 Hours Fig. 6. The kinetics of the development of inaptitude on methionine .starvation. Survivors were determined both in the original starvation medium (striped bar) and in fresh starvation medium (solid bar). fluid in radiobiological studies of microorganisms for, under some con- ditions of culture, the medium may be laden with nucleic acid frag- ments and may have variable screening potency. For example, if E. coli B in logarithmic growth phase is centrifuged, and the cells are resuspended in a sterile filtrate of optical density 1.5 obtained from the experiment represented by curve 1, Fig. 2, and then exposed to 30 seconds irradiation from a G.E. 15-watt germicidal lamp at 1 meter distance, the survivors will be 70% higher than in a control in which the bacteria are resuspended in fresh culture medium and irradiated. * However, screening by nucleic acid fragments in the culture medium has practically nothing to do with inaptitude in lysogenic organisms as is shown by the data presented in Fig. 6. In this quantitative study of the development of inaptitude on methionine starvation, the survival of the organisms from the same dose of irradiation in their original starvation medium and in fresh deficient medium was compared. Only after 5 hours of starvation were sufficient amounts of nucleic acid fragments accumulated in the medium to screen the radiation mens- 46 Essays in Biochemistry urably; there were 20% more survivors in an aliquot irradiated in its starvation medium than in one suspended in fresh deficient medium. Thus, under the conditions of these experiments, the culture of 108 cells per milliliter being washed before the start of the starvation, nucleic acid fragments in the medium contribute very little to inaptitude. In the experiments in which we first demonstrated protection by starva- tion media, 109 cells per milliliter were starved for 4 hours and, after elimination of the organisms by sterile filtration, cells in logarithmic growth phase were added to the filtrate to yield a final concentration of 108 cells per milliliter. The protection by screening was thus more pronounced. However, it should be emphasized that the protection by the starva- tion medium is not due merely to screening, for if such media are added to bacteria after the irradiation they still suppress the induction : the number of colony-forming survivors is increased five- to eightfold, and there is a decrease in the number of infectious centers. The starvation medium can be added, in a ratio of 1 part to 10, as late as 20 minutes after the irradiation, and a significant reversal of induc- tion is still measurable. The starvation medium has no effect on the non-lysogenic E. coli B/r after its irradiation. As yet we know no more about this induction-reversing agent than that it is dialyzable and is labile to ultraviolet irradiation. Since so much of nucleic acid fragments are excreted during starva- tion the question naturally arises whether inaptitude may not be the result of intracellular shading of radiation-sensitive loci by these frag- ments as they flow towards the periphery of the bacterial cell. To explore such a possibility, the nucleic acid content of the methionine deficient auxotroph was studied after two types of starvation, glucose and methionine. For these studies 100-ml. aliquots of either glucose- or methionine- starved organisms were centrifuged at intervals in a Sorvall angle centrifuge at 5000 r.p.m. for 20 minutes. A cell count in the superna- tant fluid revealed the presence of 1%, or less, of the original bacteria. This finding, together with the finding that there was no change in the number of viable, colony-forming bacteria during the course of the starvations, insured the uniformity of sampling at various intervals. The bacterial pellet containing 2 to 3 X 1010 colony-forming cells was washed once with 0.9% saline, centrifuged, and subjected to analysis for RNA, DNA, and dilute-acid-soluble fragments by the method of Ogur. The Biochemistry of Lysogeny 47 In Fig. 7, a typical experiment on the relative changes in RNA and DNA and acid-soluble fragments on glucose starvation of E. coli K12 W-6 are shown. Since the probable error of cell counting is 10'y , the changes cannot be considered as significant except for the decrease in RNA and the increase in dilute-acid-soluble fragments after 4 hours of starvation. The stability of the nucleic acid levels on glucose star- vation may be related in this mutant to the presence of excess methio- nine in the medium or to some metabolic products resulting from its genetic deficiency. 20 1:10 in C 10 S 0 10 Hi Acid wash ElRNAP EZ3DNAP 1L ^JA n Hours of starvation Fig. 7. Relative changes in nucleic acid content on glucose starvation of E. coli K12 W-6. In Fig. 8, the results of a typical experiment on methionine starvation of E. coli K12 W-6 are summarized. There was no detectable change in the number of viable cells throughout the 6 hours of starvation. There was about a 30% increase, however, in the turbidity of the cell suspensions, indicating either an increase in cell size or a change in the light-scattering property of the starved cells. The RNA, DNA, and acid-soluble fragments are represented in the figure as changes from the values found when the organisms were harvested from logarithmic growth phase and resuspended in the medium lacking methionine. No unequivocal interpretation can be offered for the 17% increase in total DNA during the first hour of starvation. It may represent increased DNA per cell, or it may be the result of a correspondingly increased cellular population, from the utilization of residual intracellular methi- onine, during the first 15 to 20 minutes of incubation in the me- thionine-deficient medium. The magnitude of the increase in DNA is not quite double the magnitude of the error of cell counting under the best conditions, and, unfortunately, cell counting during the fust 30 48 Essays in Biochemistry minutes after the resuspension of a centrifuged bacterial colony yields erratic results due to clumping of cells. However, the cell count re- mains constant from one-half hour to 6 hours, and longer, on methio- nine starvation, and, therefore, the increase in acid-soluble material and in RNA which continue significantly after the first hour represents 100 60 dn 40 20 KEY I Acid wash H3RNAP 0 DNAP 1 i Fig. 8. 12 3 4 5 6 Hours of starvation Relative changes in nucleic acid content on methionine starvation of E. coli K12 W-6. unequivocally an increase of these components within each starved cell. The large increase in nucleic acid material without increase in the number of viable cells in these starving cultures was confirmed by a simple independent method. A 10-ml. aliquot containing 2.8 X 108 cells per milliliter at the start of the starvation was centrifuged, washed, resuspended in 10 ml. of fresh medium, and the cell suspension was exposed to ultrasonic vibration for 10 minutes. The optical density at 260 in/*, of the resulting clear fluid was 0.8. After 4 hours of starva- tion an identically treated aliquot yielded an optical density of 1.2, or an increase of 50%. n The Biochemistry of Lysogeny 49 Electron-microscope photographs of this starved mutant revealed enlarged, often misshapen, cells loaded with electron-dense material. The increase in nucleic acid material on methionine starvation in E. coli K12 W-6 parallels similar changes in microorganisms when they are exposed to drastic shocks. Mitchell reported an increase in the free nucleotide content of bacteria after the "initial attachment of penicillin to growing cells." 10 Park and Johnson showed that in Staphylococcus aureus in presence of 0.1 of a unit of penicillin per milliliter there is about a 40^ increase in RNA in 65 minutes with no concurrent change in the cell count.11 Kelner has noted that there is a marked increase in RNA in E. coli B/r following a dose of ultra- violet irradiation which prevents 90c/c of the organisms from giving rise to visible colonies.12 However, the large increase solely in RNA content, upon starvation, seems to be unique to this auxotroph. The appropriate autotrophs of E. coli Ki2 when starved of histidine, tryptophane, and leucine, respectively, yielded values similar to those obtained upon glucose starvation of E. coli \\]> W-6. We also studied a methionine-requiring mutant of the W strain of E. coli which was kindly supplied, as were those listed above, by Dr. Bernard Davis. The W strain is lysogenic but, unlike the Ki2 strain, the frequency of the occurrence of the phenomenon in a given population cannot be increased by radiations or other inducing agents. The methionine-requiring auxotroph of the W strain, W 122-33, ap- pears to have a genetic block analogous to that of E. coli K12 W-6, as far as this can be determined by the probably not-too-discriminating technique of the determination of accumulated metabolic precursors. This putatively analogous mutant of the W strain did not accumulate nucleic acid material on methionine starvation. The unique ability of E. coli W-6 to synthesize RNA independently of DNA and of protein (for there was no increase in total protein content) may be put to use to study the relations among the syntheses of these three entities, but here we are concerned only in what this mutant may have contributed to an understanding of inaptitude. The two types of starvation, glucose and methionine, have parallel effects on inaptitude, but, by rare chance, they have divergent effects on the nucleic acid content of the starved cell. Intracellular screening by nucleic acids or fragments thus appears to be an unlikely mechanism for inaptitude. We are thus back where we started. We learned a bit about nucleic acid metabolism, but apart from the essentially negative contribution 50 Essays in Biochemistry that inaptitude is not caused by screening we know nothing more about it. However, the preoccupation with the mechanism of excretion led to the next working hypothesis. We considered the possibility that inaptitude is the result of the loss during starvation, either by excretion or by enzymatic disposal, of some radiation-sensitive locus within the Fig. 9(«). Infectious centers in E. coli K12 incubated with irradiated leucovorin at a dilution of~2 X 10" 4. cell. That there is a specific site, an Achilles' heel, as it were, upon which the inducers fall in a lysogenic organism has a compelling plausi- bility. How else could we account for the homogeneity of the lyso- genic response to induction? It must be recalled that close to 100% of an inducible organism will, in logarithmic growth phase, consistently respond to the small inducing irradiation by the proliferation of bac- teriophage. The phenomenon is quite different from the mutagenesis induced by irradiation. Irradiation for mutagenesis must be, by com- parison to that for induction, prodigious, killing over 99.9% of the The Biochemistry of Lysogeny 51 cells, and the surviving ''biochemical cripples" run the gamut of known and unknown genetic deficiencies. Of course, the captious may argue that there is an equally homogeneous response in such experiments, too, since 99% of the bacteria die. Most likely, however, the cause of their death is not as homogeneous as in induced lysogenic organisms. Fig. 9(b). Untreated control at a dilution of 10 — 4. Since the lysogenic response to induction is so uniform the develop- ment of the phage could be more plausibly visualized as resulting from some metabolic shift in a delicately adjusted equilibrium rather than from random lesions in the genetic material. (The profound change in nucleic acid metabolism induced by invasion of external phage, cessation of RNA and increase of DNA synthesis,13 does not take place during phage development in induced lysogenic cultures. Neither RNA nor DNA synthesis is interrupted after induction by irradiation.14) We first considered the possibility that inaptitude results from the loss during starvation of some radiation-sensitive metabolite which is 52 Essays in Biochemistry needed to mediate the metabolic effect of irradiation. We must admit that we know of no evidence for the existence of a radiation-sensitive cofactor of induction. The hypothesis is frankly rooted in heuristic opportunism: nothing much could be done chemically with an hy- pothesis postulating some subtle change in a macromolecule during starvation. We therefore started a search for a metabolite which, upon irradia- tion, might act as an inducer on unirradiated E. coli K12; either in log- arithmic growth phase, or starving. We irradiated strongly a variety of metabolites (purines, pyrimidines, nucleosides, nucleotides, vitamins, and coenzymes) and then incubated E. coli K]2 in logarithmic growth phase with the irradiation products. We tested the irradiation prod- ucts for their toxic effects on the organisms by counting surviving cells, and for their inducing effect by assaying on a sensitive strain for an increase in the normal background of infectious centers. We found that several metabolites upon strong irradiation are con- verted into products toxic to the bacteria. Some of these had been described in the literature; some had not. But of all the irradiated metabolites only leucovorin, and its derivative anhydroleucovorin, acted as inducers. Such an experiment was performed as follows: Five hundred micrograms of anhydroleucovorin or calcium leucovorin tetrahydrate per milliliter in the usual synthetic medium was irradiated in a 10-ml. lot in a quartz petri dish for 15 hours with a 15-watt G.E. germicidal lamp. E. coli K12 in logarithmic growth phase was centrifuged, and the cell clump was resuspended in the irradiated medium to a concentration of about 108 cells per milliliter. The culture was then incubated in the dark for 50 minutes, and then appropriate plating was performed for the determination of surviving colony formers and of infectious centers.15 In Fig. 9 the photographs of the result of such an experiment along with that of an untreated control at twice the concentration are given. The irradiation-elicited inducing potency resides in the pteridine moiety of leucovorin, since 2-amino-4-hydroxy-5-formyl-6-methyl- 5,6,7,8-tetrahydropteridine 16 (by analogy to folic acid a plausible first cleavage product of leucovorin 17) is converted by 5 hours of irradiation into an inducer, but para-aminobenzoylglutamic acid is not. In Table 1 characteristic data are presented. It should be empha- sized that the induction by the irradiated products is only partial. Full induction, by 100 seconds irradiation, would induce, under these conditions, about 60% of the cells. No higher induction could, as yet, Surviving Induced Cells Cells per < 'ells per i Large Plaques) Milliliter Milliliter after per Milliliter at Start 50 Minutes after .50 Minutes 2.0 X 108 3.3 X 108 2 X 105 2.0 X 108 2.1 X 108 6 X 10r' 2.0 X 10s 2.4 X 108 .">.l X 106 1.9 X 10s 2.7 X 108 1.1 X 107 The Biochemistry of Lysogeny 53 Table 1. The Inducing Effeet of Irradiated Leueovorin on E. coli K,., Num- ber Experiment 1 E. coli K-12 control 2 Exposed to 15 seconds direct irradiation 3 Incubated with 500 jig/ml. of irradiated leueovorin 4 Incubated with 500 pg/ml. of irradiated 2-amino-4-hydroxy-5-formyl-6-methyl- 5,6,7,8-tetrahydropteridine be achieved, because the photolytic products, unlike their unirradiated precursors, inhibit, above a concentration of 500 micrograms per milliliter, the bacterial respiration essential for phage development. Whether induction approximating the direct irradiation of the bacteria can be obtained with the photolytic products will depend upon whether a separation of the inducing and inhibitory potencies can be effected. The induction by the photolytic products, unlike the mutagenesis by irradiated complex media reported by Wyss 18 and collaborators, is not caused by accumulated hydrogen peroxide. The concentration of hydrogen peroxide after irradiation in the inducing solutions is too low- by a factor of 10 to affect E. coli K12. Moreover, induction by hydro- gen peroxide is, of course, completely suppressed by catalase, whereas solutions of the photolytic products of leueovorin are completely un- affected by this enzyme. Since attempts at the identification of the inducer are just starting, one can speculate freely on its nature, untrammeled by the intrusion of obstreperous facts. An organic peroxide is one possibility, since some peroxides are known to act as inducers in lysogenic organisms. The sequence of reactions during the photolysis of folic acid is well known: cleavage occurs at the 6-methylene link, the methylene carbon being oxidized to an aldehyde, yielding 2-amino-4-hydroxy-6-formyl- pteridine. On further irradiation this compound is oxidized to the corresponding 6-carboxylic acid.17 Leueovorin differs from folic acid in containing a reduced ring system and a formyl group in the 5 position. By, perhaps oversimplified analogy, one might expect both the 5 and 6 carbons in leueovorin to be involved in oxidation, possibly yielding a peroxide. At any rate, the presence of the formyl group of leueovorin is essential, for the reduced form of folic acid, tetrahydro folic acid, does not yield an inducer on irradiation; nor does 10-formyl folic acid. However, if the inducer is a peroxide it is an unusually stable one; 54 Essays in Biochemistry it retains moist of its activity at 38° for 24 hours, and it is not. inac- tivated by glutathione or ascorbic acid. Unfortunately, we could not determine unequivocally whether this organic inducer could, in contrast to direct irradiation, induce bacteria which had been rendered inapt by starvation. The photolytic products are effective only if they are in contact with the organism for 40 to 50 minutes. At the same time the lacking nutrient must be provided to enable the induced organism to elaborate phage and the organisms are known to regain aptitude under these conditions. Although we pay constant heed to the devil's advocate who per- sistently insinuates that the unique effect of irradiated leucovorin is an artefact, unrelated to true induction, there are circumstantial indi- cations which entice one to pursue the study of the phenomenon: leucovorin is heavily implicated in purine metabolism 19 and the photo- lytic product of folic acid, 2-amino-4-hydroxy-6-formylpteridine, is known to be an extremely potent inhibitor of xanthine oxidase.20 At any rate, the phenomenon may not be without value as a possible tool to explore induction: the effect of the photolytic products on isolated enzyme systems can be studied, and its possible concentration in some component of the cell can be explored. There are, of course, a host of possible mechanisms for the develop- ment of inaptitude other than the loss of a radiation-sensitive co- factor of induction. However, we shall refrain from listing any of these. The writer is shackled by his own injunction to overly imag- inative students: fruitful biochemical speculation must stand on a bifurcated root, one reaching into the biological phenomenon, the other into the store of immediately applicable methods of chemical explora- tion; otherwise such speculation is not biochemistry. Unfortunately, we have not been able to nourish the second root too well. We must finish this essay, as we started, on hope, a hope that the goal may be reached when a detailed series of reactions can be written for the chemical events that are precipitated when an inducible lyso- genic organism is exposed to a packet of inducing energy. The writer is conditioned to grope for such a goal, for he belongs to what he likes to call the Hudson River School of biochemistry. Should the name the Hudson River School evoke in the reader's mind an association with the school of painting of the same name, the writer would be neither surprised nor displeased. For it must be recalled that the painters of the Hudson River School painted every tree, every twig, every leaf into a landscape. So, too, the school of biochemistry founded by the man to whom this book is dedicated strives for the The Biochemistry of Lysogeny 55 precise determination, atom hy atom, of the molecular structure and the molecular function of the components of the living cell. Whereas slavish devotion to minutiae in a painting may be aesthetically ques- tionable, the inspired unraveling of molecular anatomy and molecular physiology by the students from Dr. Clarke's department has proved to be richly rewarding. The reader must find examples of the shining products of this school in other essays in this book, but it is hoped that this one may not be without some value as an example of how to — or how not to — grope in a brand-new field. This work was supported by grants from the John Simon Guggen- heim Memorial Foundation and from the Division of Research Grants of the National Institutes of Health, Public Health Service. References 1. A. Lwoff, Bacteriol. Revs., 17, 269 (1933). 2. Hopkins and Biochemistry, F. G. Hopkins, W. Heffer & Sons, Cambridge, 1949. 3. F. Jacob, Ann. inst. Pasteur, 82, 433 (1952). 4. E. Borek, Biochim. el Biophys. Acta, 8, 211 (1952). 5. E. Borek and J. Roekenbach, Arch. Biochem. and Biophys., Jfi, 223 (1952). 6. E. Borek, Federation Proc, 12, 180 (1953). 7. M. Stephenson and J. M. Moyle, Biochem. J., 45, VII, 1949. 8. E. Borek and P. Owades, Proc. 6th Intern. Congr. Microbiology, Rome (1953). 9. E. Borek. A. Ryan, and J. Roekenbach, Federation Proc, 13, 184 (1954); J. Bacteriol., 69, 460 (1955). 10. P. Mitchell, Nature, 164, 259 (1949). 11. J. T. Park and M. J. Johnson, J. Biol. Chcm., 179, 5S5 (1949). 12. A. Kelner, J. Bacteriol, 65, 259 (1953). 13. S. S. Cohen. Cold Spring Harbor Symposia Quant. Biol, 12, 35 (1947). 14. L. Siminovitch, Ann. inst. Pasteur, 84, 265 (1953). 15. E. Borek and J. Roekenbach, Biochim. et Biophys. Acta, 15, 140 (1954). 16. D. B. Cosulich, et al., J. Am. Chcm. Soc, 74, 3247 (1952). 17. 0. H. Lowry, O. A. Bessey, and E. J. Crawford, J. Biol Chcm., ISO, 389 (1949). 18. O. Wyss, et al., J. Bacteriol., 54, 767 (1947) ; ./. Cellular Comp. Physiol, 35, suppl. 1, 133 (1950). 19. T. H. Jukes and E. L. R. Stokstad, Vitamins and Hormones, 9, 1 (1951). 20. H. M. Kalckar, N. O. Kjelgaard, and H. Klenow, Biochim. et Biophys. Acta, 5, 586 (1950). The Development of a Plasma Volume Expander MAX BOVARNICK and MARIANNA R. BOVARN1CK The innate and apparently incurable predilection of man for the "letting of blood" in one form or another has presented to the medical fraternity one of its oldest therapeutic problems — the problem of find- ing a substitute for whole blood or its essential components. According to no less an authority on the matter than the biblical record of human history, within the first generation of his appearance on earth man had already succeeded in converting his plowshare into a sword with which he eliminated one-third of the male population of the period. This sort of thing has been going on in one form or another ever since, and constant improvement of human knowledge over the centuries has in no way diminished the magnitude of the problem. On the contrary, modern surgical practice requires the replacement of enormous amounts of blood, and the unbelievable improvement in the science of warfare has threatened to raise the above-mentioned annihilation rate of 33.3% to nearly lOO'/c. It is thought that blood or plasma volume replace- ment on sufficient scale may possibly help to reduce the latter figure. Therefore, although it is perfectly obvious that no substitute for blood is likely to be as good as blood itself, the imminent potential magnitude of the replacement problem has stimulated efforts to pro- vide a substitute for plasma in those instances where an insufficient volume of circulating blood threatens to effect vascular collapse and death and where the oxygen-carrying capacity of red blood cells is not needed. As a result of the extensive experience gained in World War II it has been possible to formulate a set of criteria for any satisfactory plasma substitute. These are as follows: The material must primarily be able on injection to expand plasma volume by remaining in the blood stream a suitable length of time and exerting an oncotic pressure 56 The Development of a Plasma Volume Expander 57 in the same manner as does the plasma protein. Although opinion may vary as to the optimal length of stay of the substitute in the blood stream, it is generally agreed that a half-life of 12-20 hours is satisfactory; i.e., 50^ of the injected plasma substitute solution is to remain after 12-20 hours. A corollary requirement is that the material must not remain in the blood stream indefinitely or be stored in the tissues indefinitely, as it is then liable to give rise to undesirable reactions. Having served its purpose for the time stipulated, it should be excreted or metabolized, preferably the latter if in the process of metabolism it can serve as a source of nutritive energy or as suitable building material for protein lost in the original trauma. The material must not be toxic or produce any undesirable physiologic response such as hypotensive action. It must be non-pyrogenic and non- antigenic. It must not interfere with the clotting and other hemostatic properties of the blood, and it must not be harmful to any of the formed elements of the blood. It should not materially increase the viscosity of the blood. In addition to these physiological criteria certain physical and chemical requirements are imposed by considerations of practicality. The material must be readily and cheaply available in large quantities. It must be easy to sterilize, stable to conditions of long storage and climatic extremes of heat and cold, must not gel at low temperatures, and if possible should be transportable in minimal bulk volume. During the course of two world wars many substances have been proposed and tried, which have met the above criteria with varying degrees of satisfaction. Though the subject is too familiar and elementary to warrant any detailed presentation of these materials, a brief review may serve to provide background for further advances in this field. Chemically most of the materials proposed and used fall into two general classes: they are either polysaccharides or proteins. Of the latter class human serum albumin or human plasma itself should obviously provide the best substitute therapy for loss of plasma. This, indeed, they do, but there are certain drawbacks connected with their large-scale massive use. The first of these is connected with the amount that might be required in military or mass disaster. It is estimated that an average of some 30-40 units (500 cc. per unit) would be required to treat the average case of severe burn. Requirements for other types of fluid loss or shock vary with the nature of the condition but are generally less than this amount. It is evident then that an enormous amount of human blood would be required to stock- 58 Essays in Biochemistry pile plasma or its derivatives for a major disaster. There is the further difficulty that there is a significant deterioration rate for stored plasma. A still further major difficulty relates to the problem of processing plasma. This is due to the fact that a certain percentage of the donor population are carriers of the virus responsible for producing hepatitis. It is obvious, then, that any pool of plasma enormously increases the chances of spreading this disease. Thus far no practical effective method has been found of inactivating this virus or of removing it from plasma or its fractions. At best it is possible to minimize the dissemination of this disease by processing and distributing plasma in single units rather than in pools. For these various reasons the prob- lem of finding effective plasma substitutes that could be used as plasma volume expanders has arisen. Among the various proteins that have been tried are bovine albumin, gelatin, isinglass, globin prepared from erythrocytes of man, and hemo- globin. Attempts to despeciate various non-human proteins, such as bovine albumin and isinglass, have not been successful in that effective despeciation can only be achieved by means which are so drastic as to cause extensive breakdown and loss of necessary molecular size. Human globin, although free of the objectionable antigenicity or other toxic properties, seems to be non-effective in the treatment of shock probably because of rapid excretion. Hemoglobin, similarly, is un- satisfactory. Gelatin has been accepted but is not entirely satisfactory for wide use because those preparations which are of proper molecular size in relation to retention in the blood stream are apt to gel at low tempera- tures. Oxypolygelatin, an oxidized polymerized gelatin, has been in- troduced by Pauling as a material of greater fluidity than gelatin but similar physiological properties. There have been reports stating that gelatin is antigenic in humans, but general experience with it has not demonstrated this to be a serious factor in connection with its use. Of the various polysaccharides that have been considered, and these have included pectin, methyl cellulose, and dextran, only the last has found wide acceptance. Hazards of antigenicity or deficiencies of staying power in those which have been degraded to remove anti- genicity are associated with all the polysaccharides. In addition, there is the difficulty that these materials may not be subject to the met- abolic processes of the human body and may therefore remain deposited in organs for undue lengths of time, leading to conditions such as cirrhosis of the liver, which resulted from the use of the gum acacia in the first world war. Although preparations of dextran have been The Development of a Plasma Volume Expander 59 obtained of such degree of purity and homogeneity that they are acceptable in many respects, prolonged trial has nevertheless demon strated certain unsatisfactory properties such as antigenicity and un- desirable prolongation of bleeding time due to an unknown effect on the hemostatic mechanism. A third class of substances, of which there is one outstanding repre- sentative, is that of the synthetic polymers. The polymer of vinyl pyrrolidine known as PVP was used extensively by the Germans in World War II, with considerable success. It has been used to some extent in this country but is felt to be undesirable because the material is taken up by the reticuloendothelial system and remains there indefinitely. Against this background attention can now be turned to the exami- nation of another substance or group of substances which have been proposed and which offer some uniquely interesting properties with reference to their use as plasma volume expanders. These are the peptides of glutamic acid and polymers of these peptides. These polymers of glutamic acid can be produced both synthetically and biologically, but the biologically produced material will receive major consideration here. Glutamyl polypeptide was first described and isolated by Ivanovics and Erdos,1 who observed that it was a component of the capsule of Bacillus anthrads and of Bacillus subtilis. These investigators, who first isolated the pure polypeptide, demonstrated that it was composed of glutamic acid residues linked in y linkage. Oui' interest in this material first arose in connection with the pos- sibility that this peptide might offer an opportunity of studying the enzymatic synthesis of the peptide linkage. The reason for this was twofold. The peptide was formed as an extracellular product in B. subtilis cultures, and it was hoped that a cell-free synthetic enzyme system might be obtained similar to the system successfully studied by Hehre - in the dextran-producing organism Leuconostoc mesen- teroides. Furthermore, it was felt that, since the peptide contained only one amino acid, the enzyme system involved in its biosynthesis might be less complex than those needed for other peptide syntheses, and thus offer a favorable starting point for attack on the general problem of peptide biosynthesis. It was first necessary to demonstrate that the material obtained from the strains we worked with was truly composed of only glutamic acid and to ascertain the nature of the peptide linkage. Fortunately for the purposes of both these experiments and subsequent develop- 60 Essays in Biochemistry merits in relation to the use of these materials as plasma volume extenders, the subtilis organism produced these glutamyl peptides on a simple medium, Santon's medium, the organic constituents of which are glutamic acid, glycerin, and citric acid. The organism grows as a mat on the surface of shallow layers of this medium and produces up to 1 gm. of peptide per liter. By incorporating into this medium heavy nitrogen, either in the form of ammonia or of amino nitrogen in the glutamic acid, it was possible to obtain pure peptide containing 4. 00% isotopic nitrogen excess, and this peptide on hydrolysis yielded pure glutamic acid containing 3.99% isotopic nitrogen excess, indicat- ing that this was the only amino acid present. We also confirmed the fact that the glutamic acid residues in the peptide we obtained were y linked.3 At this point in our studies the year of 1941 ground to a close leaving behind it many new and pressing problems, among them the need for a method of preserving blood or developing a substitute for plasma. It had become apparent to us that here was available an easily and cheaply produced polypeptide composed of a single non-aromatic amino acid linked in a physiologically acceptable peptide linkage and possessing a free carboxyl group for each peptide linkage. There was a reasonable possibility that this material might be non-antigenic, because existing theories, based on considerable evidence, postulated that only large peptides or proteins containing some aromatic amino acids were antigenic. A particularly attractive and provocative feature was that the presence of a large number of carboxyl groups would assure maximal Donnan effects, so that if the molecule were large enough to be non-diffusible from the blood stream it would have an unusually high oncotic efficiency. Under the circumstances, it was evident that the possibility of the use of this material as a plasma volume extender had to be explored. It was very quickly demonstrated that the material as isolated from B. subtilis filtrates by the methods of culture then employed was in many respects an ideal plasma volume extender. In extensive tests in small animals it was proved to be non-toxic and non-pyrogenic. Its osmotic potency in vitro was found to be quite high, and this osmotic potency was found to be additive to that of serum albumin (Table l4). The material was obtainable as a dry white powder which was highly water-soluble as the sodium salt, and solutions of the latter were stable to autoclaving. The material did not deteriorate or develop any unfavorable properties on standing. The ease of prep- aration from culture filtrate was quite remarkable. After removing The Development of a Plasma. Volume Expander 61 Table 1. Osmotic Efficiency of* Glutamic Acid Polypeptide in Serum Mix i in in Solutions Albumin GAP Concentration. Concentration, ( temotic gm/liter gm/liter Efficiency * 60 0.00 159 50 L.08 158 40 2.16 155 30 3.27 145 20 4.39 144 10 5.53 145 0 6.68 152 * The osmotic efficiency of a non-diffusable substance in a solution is denned as the number of cubic centimeters added to an infinite volume of the solution for each gram of substance when the osmotic pressure, the quantities of all other non-diffusable substances inside the membrane, and the concentration of each diffusable substance outside the membrane are kept constant. the organisms by filtration or centrifugation, the peptide was precipi- tated from solution by addition of copper sulfate which formed a green rubbery complex with it. This dissolved readily in citric acid solution, from which copper was removed by precipitation as sulfide or by other methods. On acidification and standing in the cold a white precipitate of peptide separated out. This peptide, after resolution in sodium hydroxide and reprecipitation, is pure enough for physiological work. Molecular-weight determination by an end-group method showed molecular weight of 12,000 to 15,000. By light diffraction and viscosity measurements it was determined that the average molecular length ranged from 150 A. to 200 A., which is of the order of magnitude of serum albumin, and the molecular diameter was 11 A., approximately one-third that of serum albumin.4 On physiological testing it was found that this material unfortunately was excreted very rapidly in the urine in normal dogs and humans. In a human the blood stream is cleared of peptide in approximately 5 hours after injection of 20 gm. in a liter of saline. In a dog rendered hypotensive by bleeding, the peptide solution promptly restored the blood pressure to the normal level; whereupon the dog resumed urination, excreted the peptide very rapidly, and urinated himself back into the hypotensive stage. It was obvious that, although otherwise apparently suitable as an extender, this material was too small and too readily excreted by the kidney to be effective for the desired length of time. Since the molec- ular length was approximately that of serum albumin it seemed reason- 62 Essays in Biochemistry able that a first attempt to improve the situation should be in the direction of increasing the diameter of the molecule. This obviously could be done by attaching to the free carboxyl groups further amino acids, peptides, sugars, or any biological substance with a replaceable hydrogen. Since the peptide itself had seemed to be pharmacologically acceptable and physiologically promising, side chains of either glutamic acid of glutamyl peptide were the first choice, as it was felt that they would retain all the Donnan effects and contribute nothing new in the way of possible disadvantageous pharmacological effects. After some exploration it was found that conversion of the peptide to a polyazide could be accomplished, and that this polyazide conju- gated readily in pyridine solution with side chains in the form of the pyridine-soluble polymethyl ester of the peptide. After saponification the polysodium salt of the conjugate was obtained. (These steps will be described in more detail later.) The preparation of two such con- jugates in small amounts was completed, and these were tested in dogs. Fortunately for this purpose we were able to develop a simple, rapid, and convenient method of microassay. It had been observed that the peptide formed highly insoluble precipitate with some cationic dyes. Dr. Joseph Victor, one of the group * who collaborated in bio- logical experiments with the peptide, had developed this observation into a quantitative method of peptide determination applicable to blood and urine. This has since been elaborated to make it applicable to the assay of peptide content of other tissues. Of a number of dyes tested, the results obtained with safranine 0 seemed to be the most reliable. With this material it was found that at pH 5.98 in the presence of excess safranine the peptide was pre- cipitated quantitatively if its molecular weight was 3000 or higher. The presence of excess peptide in the mixtures interestingly enough dissolved the precipitate. This is reminiscent of the behavior of antigen-antibody precipitates with the peptide behaving in a manner similar to a multivalent antigen. In the actual analysis precipitation is complete in 15 or 20 minutes, and the decrease in the concentration of dye remaining in the supernatant is proportional to the amount of peptide present in the solutions being analyzed. Since safranine gives no precipitate with any material in normal plasma or urine which has been diluted 1:1, the analyses can be performed directly on these mate- rials. Where tissue extracts are present, the extraneous safranine- * Drs. A. J. Patek, J. Victor, F. E. Kendall, A. Lowell, W. Bloom, and G. C. Hennig at the Goldwater Memorial Hospital, Columbia Research Service, Welfare Island, N. Y. The Development of a Plasma Volume Expander G3 precipitable material can be removed by trichloracetic acid precipita- tion."' Peptide and conjugate have been found to be soluble over short periods of time in the trichloracetic acid supernatant. After filtration of the TCA mixture, however, the trichloracetic acid must be extracted immediately with ether, both because it will eventually cause precipi- tation of the peptide and because it forms an insoluble precipitate with safranine. 'With these methods available, excretion rates and blood levels were easily followed. As expected the conjugates were retained in the blood stream much longer than the straight-chain peptides. At this interesting juncture it was decided that a "blood substitute" was no longer urgently needed, as the transportation of chilled whole blood to needed areas was practicable on a desirable scale. Further- more, glutamyl peptide had suddenly acquired a new importance. It seemed that the available conventional military, industrial, and scien- tific forces might not suffice to meet the insatiable demands of war. Man, with inhuman ingenuity, successfully met this challenge — some- what to his later consternation. Among the new offerings at the time was that of biological warfare, and the possibility arose that certain knowledge and skills hitherto devoted to the benevolent purposes of the healing art might be perverted to more destructive ends. It there- fore became necessary to prepare defenses not only against the nat- urally occurring epidemic concomitants of war, but also against the possibility of man-made pestilence. The next phase of our work on the peptide was to be under the auspices of Camp Detrick, Maryland, then the center of biological warfare activity. The anthrax organism had achieved the high distinction of early consideration as a suitable agent for biological warfare. Glutamyl peptide, as already mentioned, was known to be a capsular component of the anthrax organism and was thought in this capacity to be partly responsible for its virulence. On this basis it was reasoned that antibodies to this peptide should confer some degree of protection against infection by the anthrax organism, and the development of a vaccine containing the glutamyl peptide as the specificity-conferring component became desirable. It had long been known that protective antipeptide antibodies were not readily produced by anthrax vaccines prepared in the usual fashions, nor. for that matter, by actual anthrax infection and recovery. It was hoped, therefore, that a more potent, protective vaccine for humans might be prepared by attaching the peptide as a haptene group to human protein. In man this antigen should presumably produce anti- bodies specific for this peptide. 64 Essays in Biochemistry Antigens were accordingly prepared in which the peptide was linked to human globulin both via the azide method listed above and via a diazo coupling reaction achieved by making a p-nitrobenzoyl derivative of the peptide by reacting it with p-nitrobenzoyl chloride, reduction of this to the p-amino derivative, and conversion of this to the dia- zonium salt. Peptide was readily separated from protein peptide con- jugates by alcohol fractionation. The conjugates were tested for antigenicity in mice, guinea pigs, and rabbits, using both in vitro and in vivo methods for the detection of antibodies. The results of the work along this line can be briefly summarized as essentially negative in that extensive courses of immunization with these antigens, as well as with living and dead bacterial vaccines yielded at best questionable evidence of antibodies in rabbits and guinea pigs, and very dubious protection in mice. The rabbit serum thus produced gave a dubiously positive precipitin test with the peptide used, and the guinea pigs did not become sensitive to the peptide as evidenced by lack of any symptoms of an anaphylactic shock on injec- tion of the peptide. The important point in relation to the present subject of discussion is, however, the fact that under no circumstances were the peptide itself, polymers of the peptide with itself, or vaccines of the peptide-producing B. subtilis ever found to be able to stimulate the production of antipeptide antibodies in animals, and the same seems to be true in humans at the time of the present writing, although human trials are not yet complete. In the meantime the turbulent pattern of war had created new and more urgent problems, and the subsequent opportunities of peacetime allowed the pursuit of more interesting ones so that shortly after com- pletion of this immunological study our attention was diverted from glutamyl peptide to other fields until 1950. The initiation of work on glutamyl peptide was, however, destined to yield a rich reward indeed. Following the war, interest in the peptide had persisted ;it Camp Detrick, and investigations relating to its mechanism of bio- synthesis had been conducted by Drs. Housewright, Thorne, and Williams.6'7'8 In an excellent series of studies they had been able to obtain an enzyme preparation from cultures of B. subtilis which cata- lyzed a transamidation reaction in which the y-glutamyl radical of glutamine is transferred to n-glutamic acid and n-glutainyl-n-glutamyl peptides. Tri- and probably higher peptides were formed by the same system. During the course of these fundamental investigations these workers had developed much information on conditions of formation The Development of a Plasma Volume Expander 65 of peptide and especially on methods of peptide production in deep culture. When the possibility arose in 1950 that replacement of blood or plasma might again become an urgent problem, and this time on a scale far exceeding the practical supply of whole blood or its com- ponents, it seemed that once again expediency must prevail over inter- est, and our work on glutamyl peptide as a plasma volume expander was resumed. At the same time we proposed the use of the straight-chain peptide as a sedimenting agent to be used in the low gravity centrifugal sepa- ration of red cells from plasma in the project on blood fractionation and preservation that was underway at Harvard. In connection with this activity testing of the straight-chain peptide in humans was begun. (The material had been thoroughly tested in animals and found satis- factory, in 1941.) No unacceptable properties manifested themselves in these tests, and these findings provided further encouragement to proceed with preparation and testing of conjugates as plasma volume expanders. With increasing promise of eventual usefulness of the conjugate as an extender, the problem of peptide production in deep culture had become more and more important as the available shallow- culture method of production was economically unsatisfactory. In preliminary experiments we had recently determined that peptide could be produced in deep culture by aeration with COo-air mixtures, but results in different batches were irregular and generally not too satis- factory. With the aid of the group at Detrick peptide production by the deep-culture methods soon reached a highly satisfactory status. One of the first findings with this deep-culture peptide was that its molecular weight was much higher than that produced in shallow culture. The highest number average molecular weight we ever ob- served in the latter was approximately 28,000, whereas the former frequently ran as high as 100,000 or over. The availability of this material greatly increased our range and flexibility with respect to size and shape of conjugates, and the ease of production * of peptide in large quantity greatly increased the impetus of the project. With the availability of straight-chain peptide of molecular weight 120,000, which was presumably about 10 times as long as that previ- ously used, the question immediately arose and was put to the test as to whether this material would show increased retention in the blood stream. The results are of considerable interest in that they demon- * The collaboration of Merck & Co. in the production of peptide is gratefully acknowledged. 66 Essays in Biochemistry strate that mere increase in molecular length of peptide is not in itself sufficient to effect significant changes in length of retention in the blood stream of humans. Thus, increasing the size of the straight glutamyl peptide chain from 10,000 to 120,000, a twelvefold increase in length, gives no real increase in the blood-stream half-life, which is approxi- mately 1 hour for both, even though the large molecules are now many times the length of the serum albumin molecules. As can be seen in Fig. 1 the large peptide is excreted with almost the same rapidity o Dog 2.4 gm. peptide infused, mol. wt. 56,000 • Man 20 gm. peptide infused, mol. wt. 120,000 X Man 3 gm. peptide infused, mol. wt. 15,000 10 12 Hours post infusion Fig. 1. Plasma levels of straight chain peptide in man and dog. as is the small. A conjugate with molecular weight of the order of 120,000, having side chains with molecular weights of approximately 8000, shows a half-life at least 10 times that of straight-chain peptide of the same molecular weight. Strangely enough, the situation is different in dogs. In this animal straight-chain peptides of molecular weights of 60,000 and over have a half-life in the blood stream of 12 hours or longer and bring about the desired hemodilution while in the blood stream. This difference between the two species is of considerable practical interest as the dog is the most commonly used test animal for plasma volume extender tests. In this case, at least, the results with the dog wrere entirely misleading with respect to behavior of the straight-chain peptide in man. Whether or not this difference between humans and dogs relates to differences in kidney physiology or in the manner in which the peptide is metabolized in the two species has not yet been ascertained. It was quite evident in view of the above results that the investiga- tions of conjugates and polymers of glutamyl peptide must be resumed. The Development of a Plasma Volume Expander 67 Our two, and only, previous trials with conjugate had been conducted with dogs as test animals. Although the material injected had been an unfractionated solution of sodium salt of conjugate containing some unreacted peptide, two conclusions were drawn from the experiments: (a) the conjugate remained longer in the blood stream than did the straight peptide and (6) the conjugate we had made was not large enough to give the desired duration of blood-stream retention. The number average molecular weight of backbone in these conjugates 'had been 12,000-13,000 and that of the side chain had been smaller. The side chains had been obtained by methylation of the peptide by sus- pending the pure dry peptide of molecular weight 12,000 in dry methanolic HC1 for 24 hours, removing methanol and HG1 in vacuo, and precipitating the ester from methanol solution with ether. During this procedure the peptide was degraded, the extent of degradation varying with the precise conditions of reaction. Methods were therefore also developed for preparing side chains from the peptide with no degradation in chain length. By working under controlled conditions of time, temperature, and moisture, it was found possible to prepare a suitable polymethyl ester of peptide by adding ethereal diazomethane to an ethereal suspension of peptide. If meth- ylation is allowed to proceed to completion, there is generally methyl- ation of practically all of the terminal amino groups and loss of their availability for conjugation. Too little methylation gives a pyridine- insoluble product. The best compromise seemed to be a product methylated to the extent of 50-60%. This is usually 60-70%> soluble in anhydrous pyridine and retains 80-90% of the end groups as un- methylated amino groups. There is no degradation of the peptide in this process.9 The backbone polyhydrazide was prepared as previously by full methylation of another sample of peptide. This is easily achieved by methylation with diazomethane as above, except that a small amount of methanol is added to the reaction, and the methylation is allowed t<> go to completion, as evidenced by absence of free carboxyl on titra- tion of a small filtered aliquot, with dilute alkali in the presence of phenolphthalein indicator. Excess diazomethane is discharged by addition of ethereal formic acid, and the ester is filtered off and dried. On dissolving in 50%) methanolic hydrazine the ester is converted to polyhydrazide which is precipitated out by addition of methanol. This in turn is converted into polyazide by conventional methods at — 10°C. Polyazide, which is a white water-insoluble powder at this temperature, is soluble and 68 Essays in Biochemistry fairly stable in cold pyridine. This polyazide in pyridine solution is a backbone ready for conjugation with side chains. Conjugation is effected by addition of pyridine solution of side chain to pyridine solution of backbone azide in the presence of triethylamine ; the reaction mixture is maintained at 0° for 12 hours with stirring and at room temperature for another 24 hours and then precipitated by addition of ether. The solid is dissolved in water and saponified with alkali, and the sodium salt of the conjugate is separated from unreacted peptide by fractional precipitation with alcohol. Since there is generally a large difference in order of magnitude of molecular weights of the peptide and conjugate, fractionation is not too difficult. The degree of fractionation is easily ascertained by an end-group determination using Sanger's method. The pure conjugate should give no reaction, as it has no terminal amino groups. It is obvious that molecular dimensions of the conjugate should be subject to variation by alteration of the size of backbone and of side chains, and of the ratio of side chain to backbone. Methods of varying the size of side chain have been described above. The size of the backbone can be controlled by alcohol fractionation of the sodium salt of peptide to be used as backbone. The number of side chains on a given backbone can be controlled by varying the relative concentra- tions of side chain and polyazide groups in the conjugation mixture. In this connection it was thought desirable to have available a method for quantitative determination of azide concentration in the reaction mixture. A rapid simple method was obtained by using the iron hydroxamic method of Lipmann and Tuttle.10 In this determination it is important that the aliquot of pyridine solution of azide used should be no larger than 0.1 ml., as the presence of too much pyridine interferes with color formation, and that two drops of gum acacia solu- tion be added before addition of the ferric chloride solution. This latter helps stabilize the solution. A considerable number of these conjugates have been prepared, and various aspects of their chemistry and biology have been investigated. These include studies of their effects on formed elements of the blood in connection with their use in the blood preservation, of their effects on hemostasis, of their possible pathological effects on prolonged and repeated injection in animals, etc. The results of these studies have to date been negative in that no biologically unacceptable properties of the peptide or conjugate have been observed. Physicochemical studies have been carried out for purposes of characterization of the various polymers produced. The Development of a Plasma Volume Expander 69 Of considerable importance in connection with the use of these materials is the question of their metabolic fate in man. It has been ascertained that both the peptide and the glutamyl peptide conjugates are broken down by aqueous tissue extracts of practically all human tissues including red cells, kidney, liver, spleen, and brain. The major exception seems to be muscle. It is significant also that plasma does not break down either peptide or conjugate. The precise extent of breakdown that occurred in these tissues in vitro is not known. It has been observed that if peptide or conjugate solu- tions are exposed to homogenates of the organs, or hemolyzates, then 70 or SO/! of the safranine-precipitable conjugate or peptide disappears within 12 hours at room temperature. Since it is known that, under the conditions of assay used, safranine precipitates quantitatively pep- tides and conjugates of molecular weights of 3000 and higher, it is obvious that the material that has disappeared in this time must have been degraded to molecules of less than 3000 molecular weight. By microbiological assay and chromatographic analysis it was determined that after precipitation with trichloracetic acid there were present in the digestion mixture low molecular weight glutamyl peptides and some free glutamic acid. The free glutamic acid found was never more than 10^ of the conjugate or peptide originally present, but it is possible that some free glutamic acid produced may have been metabolized. This point of completeness of metabolism or excretion of the material is important and will have to be conclusively settled by isotopic anal- ysis. It is interesting that, under the same conditions used in obtaining the above data, synthetic a-linked glutamyl peptide is degraded ex- tremely slowly, if at all, by extracts of human kidney, liver, and red cells. The potentialities for control of molecular size of conjugate may be exemplified by the following. In three separate preparations where the size of the side chain and the ratio of side chain to azide were kept constant, but where the backbone varied in length in the approximate ratio of 1 :2:3, the half-lives of the material in the human blood stream were respectively 12, 18, and 24 hours. In another set of preparations when backbone size was kept constant it was found that enlargement by side chains consisting of glutamic acid (or various other amino acids i caused no significant increase in the half-life in the blood stream over that of the original backbone peptide, whereas side chains of molecular weight of a few thousand increased the half-life by many hours even though the number of amino acid side chains per backbone was higher than that of peptide side chains. In a third instance where 70 Essays in Biochemistry both backbone and side chain were kept constant with respect to size but varied in ratio, it was found that doubling the ratio of side chain per azide increased the sedimentation rate of the conjugate from 1.2 to 1.4 S units. The sedimentation rate of the backbone peptide used in this experiment was 1.0 S units. ; ' 1 1 1 Key . | - | ■ | i | i | 1 | 1 Plasma volume and per cent retention of injected volume 4000 _ Plasma conjugate levels -c 10° 1 ^~"~" t^K 3500 , c c 2 <-> u 50 "S V Ns 3000 — £ | N > n X 0i \ — 2500 ft , 1 , I . 1 . 1 . 1 , 1 , 1 , 1 ,\ 00 2 4 6 48 72 10 12 14 16 18 20 22 Hours Time Fig. 2. 1 liter of 2.6% conjugate infused in 1 hr. - 6«" - 2 P 0 JS CL, 7 days The experience gained in the work described above has led to the production of many preparations which, when clinically tested, have given satisfactory results with respect to blood-stream retention, plasma volume extension, and absence of unfavorable physiologic or hematologic effects. A typical example is shown in Fig. 2 and Table 2. Table 2. Clinical Test: Hematological Findings (1 liter of 2.6% conjugate infused in 1 hour) Post Infusion Basel me Test Preinfusion 1 hr. 6hr. 24 hr. RBC 5,040,000 4,460,000 4,310,000 5,110,000 WBC 5, 500 9,000 8,000 5,050 Differential: Band forms 1 2 1 5 Segmental forms 43 56 47 55 Lymphocytes 50 36 45 37 Eosinophils 2 3 4 Monocytes 3 2 2 3 Basophils 1 1 1 Platelets 297 360 250,200 258,600 332,150 Clotting time 8' 30" 14' 55" 9' 5" 15' 55" (Lee White) Hematocrit 46 Benzoyl- i>tyrosylgIycinamide + H-jO Benzoyl-L-tyrosine- + NH.i+ — > Benzoyl-L-tyrosin- amide + H2O Glycyl-L-phenylalanine* + NH4+ — > Glycyl- L-phenylalanmamide + H2O dynamics of peptide formation, the available information is sufficient to indicate the oversimplification inherent in the assignment of an arbitrary value of -f 3 kcal. per mole to the synthesis of the CO-NH bonds of proteins. It may in fact be expected that, if two peptides of moderate length were converted to a single long-chain peptide in a condensation reaction, the energy required would be much less than 3 kcal. Since thermodynamic data only can tell us what may happen, but give no information about what does happen in a living cell, the only general conclusion that can be drawn from the available AF° values is that, at pH 7, the formation of a peptide bond by a condensation reaction is an endergonic process. These data do not rule out the physiological occurrence of such reactions, especially if the union of two peptides of moderate chain length is considered. In the absence of experimental evidence to the contrary, it would seem premature to discard the possibility of condensation reactions, as has been suggested at various times. Of special relevance is the consideration that the living cell is not a homogeneous system in which the chemical reactants are present in equilibrium concentrations. Hence, it seems reasonable to envisage the enzyme-catalyzed formation of interior peptide bonds AF°, AH, Tempera- kcal. per kcal. per Equilibrium ture mole mole Constant 37 °C. + 3.3 0.005 25°C. + 0.4 + 1.5 0..5 25°C. + 5.8 25°C. + 6.2 108 Essays in Biochemistry in condensation reactions that are "pulled" by subsequent processes, among which some may be strongly exergonic.2 This type of energetic coupling has been clearly illustrated in model experiments. For exam- ple, in the synthesis of benzoyl-L-tyrosylglycinamide from benzoyl-L- tyrosine (0.025 M) and glycinamide (0.025 M), catalyzed by chymo- trypsin, equilibrium is attained when only about Y/c of the reactants has undergone condensation. On the other hand, if glycinamide is replaced by glycinanilide, the resulting benzoyl-L-tyrosylglycinanilide crystallizes from the solution in a yield of about 65%. The driving force in the formation of the anilide is its removal from solution, because, unlike the amide, the anilide has a solubility lower than 2.5 X 10~4 M. Thus, the endergonic synthesis of the peptide bond is coupled to the exergonic process of the removal of the peptide derivative from supersaturated solution. Model experiments of this kind, conducted with intracellular proteinases as catalysts, provide the principal support for the view that, in the synthesis of the peptide bonds of proteins, such "pull" mechanisms are operative. It should be emphasized, however, that there is no evidence from biochemical studies with intact cells or organisms that this type of energetic coupling is important in protein synthesis, although it must also be noted that no investigations have yet been conducted that permit an objective decision on this question. As mentioned before, the studies on the energy changes in the hydrolysis and synthesis of peptide bonds have called attention to the difference in the AF° values for condensation reactions involving peptides, as compared with condensation reactions involving free amino acids. Since the latter reactions are, in general, more endergonic in character, it has seemed plausible to assume that, instead of a "pull" type of coupled reaction, the biosynthesis of a CO-NH bond between two amino acid residues is "pushed" by an exergonic reaction, such as the cleavage of a pyrophosphate bond of ATP. In some of the recent discussion of possible mechanisms of protein synthesis it has been implied that these two types of coupling are mutually exclusive. In the face of the available experimental knowledge, it would appear more profitable to consider the working hypothesis that both types of mechanism are involved in the biosynthesis of proteins from amino acids, but at different stages of the overall process. It is implicit in this hypothesis that peptides are intermediates in protein synthesis, a view that has been challenged recently on the basis of experimental findings to be discussed later in this essay. The Biosynthesis of Peptide Hoi ids 109 Despite the attractiveness of the concept that the cleavage of pyro- phosphate bunds of ATP is linked to the biosynthesis of the peptide bonds of proteins, it is not possible at present to specify the chemical nature of the "reactive" form of the amino acids. Efforts to demon- strate a direct enzyme-catalyzed reaction between ATP and the a- carboxyl group or the a-amino group of amino acids have been incon- clusive thus far. Although carboxyl phosphates and phosphoamides of amino acids have been synthesized in the chemical laboratory, there is no evidence that they play a role in the biosynthesis of CO-NH bonds. Hence, in a discussion of the synthesis of proteins from amino acids, one cannot write even the first chemical reaction on the way to the completed protein, in the sense that one can specify that, in the biosynthesis of glycogen from glucose, the first step is the formation of glucose-6-phosphate in the glucokinase reaction. Attempts to identify the "reactive" form of a-amino acids in protein formation have involved the study of the biosynthesis of the CO-NH bonds of compounds such as acetylsulfanilamide. These studies have been of exceptional importance to biochemistry, since they led to the discovery of coenzyme A (CoA) and have shown that, in the presence of suitable enzyme systems, the cleavage of ATP is associated with the formation of acetyl CoA.:i Like other thiol esters, this acyl mer- captan reacts with amines such as sulfanilamide to form amides. Similarly, in the biosynthesis of hippuric acid, the acylating agent is benzoyl CoA, whose formation from benzoic acid and CoA requires the participation of ATP. However, in the instance of hippuric acid synthesis by animal tissues, the enzyme that catalyzes the reaction between benzoyl CoA and glycine appears to be specific for glycine as the amine. In view of the limited specificity of this enzyme, it cannot be assigned a general role in protein formation, although the synthesis of acylamino acids such as hippuric acid illustrates a bio- chemical mechanism whereby energy may be "pushed" into the forma- tion of a CO-NH bond. However, there is no evidence as yet for the enzymic conversion of the a-carboxyl groups of free amino acids or peptides to thiol esters similar to acetyl CoA or benzoyl CoA. In the search for alternative mechanisms that may be operative in peptide bond synthesis, attention has also been given to the formation of pantothenic acid from pantoic acid and ^-alanine. Here, CoA does not appear to be involved, ami it has been assumed that ATP is cleaved to adenosine monophosphate with the formation of a reactive "enzyme pyrophosphate" compound which reacts with pantoic acid to form a "pantoyl enzyme" compound; this, in turn, is believed to react with 110 Essays in Biochemistry /^-alanine to form pantothenic acid. Lipmann 4 has suggested that a mechanism of this type may be involved in the synthesis of the poly- peptide chains of proteins, but no experimental evidence for or against this possibility is available at present. To the two mechanisms of "amino acid activation" proposed on the basis of studies on hippuric acid and pantothenic acid must be added the conclusions of Bloch and his associates, from their important work on the biosynthesis of glutathione.5 The synthesis of the two CO-NH bonds proceeds in separate steps, in each of which one equivalent of ATP is required; all efforts to demonstrate a role for CoA appear to have been unsuccessful. In contrast to the synthesis of pantothenic acid, the formation of glutathione from the component amino acids is accompanied by the liberation of one equivalent of phosphate, and not pyrophosphate, per CO-NH bond formed, indicating that ATP is cleaved at different pyrophosphate bonds in the two processes. Furthermore, the studies of Speck and of Elliott have shown that, in the biosynthesis of glutamine from glutamic acid and ammonia, which also requires the participation of ATP, inorganic phosphate is formed, as in the synthesis of y-glutamylcysteine from glutamic acid and cysteine in glutathione formation. These studies on the biosynthesis of pantothenic acid, glutathione, and glutamine have all led to the working hypothesis that the role of ATP is to make possible the formation of a reactive form of an amino acid, as for example in a "pantoyl enzyme" or a ' y-glutamyl enzyme," where the acyl group is attached to the enzyme protein at a suitable site (e.g., the sulfhydryl group of cysteine, the imidazolyl ring of histidine, etc. ) . However, no general scheme for the role of ATP in the formation of the acyl enzyme compounds can be offered at present. Nevertheless, the hypothesis that such reactive acyl enzyme compounds are intermediates in the biosynthesis of the peptide chains of proteins is an extremely attractive one and finds support in work to be discussed later in this essay. In the face of the fragmentary knowledge currently available, it would seem desirable to extend the study of the synthesis of CO-NH bonds in simple compounds to an examination of the metabolism of a-amides of amino acids. Until recently, compounds of this type were not considered to occur in nature, but the demonstration of a terminal glycinamide in oxytocin and vasopressin 6 confers upon such a-amides increased biological interest. Calorimetric studies have shown that the A// of hydrolysis for the CO-NH2 bonds of acylamino acid amides is of the same order of magnitude as that found for inorganic pyro- The Biosynthesis of Peptide Bonds 111 phosphate.7 The a-amides of amino acids thus belong to a group of compounds that may be loosely termed "energy-rich" amides, in anal- ogy to the widespread use of the term "energy-rich" phosphate com- pounds. Systematic studies do not appear to have been made to determine to what extent such a-amides are present in tissues and biological fluids. Whatever may be the mechanism whereby the exergonic cleavage of ATP is coupled to the endergonic synthesis of proteins from amino acids, the elucidation of this energetic coupling is only a part of the total problem of protein formation. Of equal, if not greater, impor- tance is the task of describing the cellular apparatus responsible for the synthesis of a protein in terms of the chemical specificity that leads to the characteristic arrangement of amino acid residues in the intricate sequence revealed by the systematic degradation of the protein. The complexity of the structure of proteins makes the biosynthesis of these polymeric molecules a unique phenomenon, whose study is made for- midable by the difficulties encountered in demonstrating it in cell-free systems, as has been done for the biosynthesis of glycogen from glucose; here, the work of the Coris and their associates has clearly established the nature of the individual catalytic components of the multienzyme system involved in glycogen formation. Although there is general agreement that the biosynthesis of proteins is also an enzyme-catalyzed process, our ignorance of its details have led to widely conflicting speculations about its nature. Specifically, it is frequently argued that protein synthesis does not involve a multienzyme system which converts amino acids, via intermediates, to the peptide chains of a protein. The proponents of this view suggest that the "activated" amino acid units align themselves along a "template" (frequently con- sidered to be a nucleic acid), and the alignment is followed by the simultaneous formation of all the peptide bonds by an undefined cata- lytic process, with the subsequent release of the completed protein from the "template." It would seem that, if this is indeed the case, and if it should prove impossible to dissect the enzymic apparatus responsible for protein synthesis, the successes achieved in the analysis of other biochemical syntheses may not be attainable in this field. The question may be raised, however, whether this pessimistic view is justified by the experimental evidence now available. Support for the "template" hypothesis has come largely from studies on the incorporation of labeled amino acids into the proteins of intact animals and on the formation of bacterial enzymes. Work from several laboratories has demonstrated conclusively that, if one or more isotopic 112 Essays in Biochemistry amino acids are administered to an animal, and if discrete proteins are isolated from the tissues or body fluids, every residue of a given labeled amino acid in the peptide chain has the same isotope content. In a striking experiment, Simpson and Velick 8 administered to a rabbit five isotopic amino acids, and 38 hours later isolated from the muscle highly purified samples of two enzymes (aldolase and glyceraldehyde- 3-phosphate dehydrogenase). Upon analysis of 11 of the amino acids obtained after hydrolysis, these investigators found that the ratio of specific radioactivity of each of these amino acids in aldolase to the specific activity of the same amino acid in the dehydrogenase was constant. Subsequent work by Simpson, in which the animal was sacri- ficed 30 minutes after the administration of the labeled amino acids, gave essentially the same result. The logical conclusion drawn from these experiments, and from similar studies performed by others (Neu- berger et al, Work et al.), is that the systems involved in the syn- thesis of the proteins under investigation were drawing on a single "pool" of each of the amino acids. Although these experimental find- ings are in accord with the "template" hypothesis, it has been pointed out that they may also be interpreted in terms of either the rapid successive formation of peptide bonds without the release from the protein-synthesizing system of intermediates that accumulate, or the existence of very small "pools" of intermediate peptides that equilibrate rapidly with the amino acid "pools." Experiments on the induced formation of bacterial enzymes, espe- cially those reported by Monod and by Spiegelman, have also been interpreted to indicate that the synthesis of adaptive enzymes proceeds by a "template" mechanism, without the intervention of peptide inter- mediates. For example, it has been reported" that, in the induced formation of /}-galactosidase during the growth of Escherichia coli, the enzyme protein is derived from free amino acids in the medium rather than from products of the cleavage of pre-existent bacterial proteins. Here again, the results are consistent with the "template" mechanism, but it is difficult to accept them as proof for its existence, so long as alternative interpretations are not excluded. Other studies in bacterial systems, notably those of Gale,10 have focused attention anew on the role of nucleic acids in protein synthesis and their possible function in serving as "templates." That there is an intimate metabolic relation between cellular nucleic acids and cellular proteins cannot be denied, but it seems premature to interpret the available knowledge as favor- ing the role of nucleic acids as "templates" in protein synthesis, espe- cially since the "template" hypothesis itself is so insecurely founded. The Biosynthesis of Peptide Bonds 113 Of special importance in regard to the question whether peptides are intermediates in protein synthesis are the experiments of Anfinsen and his associates, who have studied the incorporation of labeled amino acids into ovalbumin by minced hen oviduct or into insulin and ribo- nuclease by beef pancreas slices." In contrast to the results of in vivo studies, the residues of a given amino acid were found to be unequally labeled along the peptide chain of each of these proteins, a finding consistent with the view that peptides are metabolic intermediates between amino acids and proteins. Anfinsen has suggested that the apparent discrepancy between the results of in vivo and in vitro ex- periments may be ascribed to differences in the rates of metabolic reactions in the two types of systems. It would seem inescapable that the removal of a tissue from the intact animal would lead to a pro- found alteration in the rates of those enzymic processes that are influ- enced by such factors as hormonal regulation and normal blood flow. Thus far, no one has reported comparative experiments, at several time intervals, in which a labeled amino acid was incorporated into the same protein in vivo and in vitro and the pattern of labeling along the peptide chain was determined in each case. Such studies are needed to clarify the discrepancy between the results obtained in the living animal and in tissue preparations. Clearly, the available data on amino acid incorporation into proteins do not permit one to draw any definite conclusions about the chemical pathways of protein synthesis. However, it would seem more likely that useful clues will come from the continued study of those biological systems in which non-uniform labeling of peptide chains is observed and where one may expect the occurrence of well-defined intermediates between amino acids and proteins. It is frequently stated that peptides (other than glutathione, carnosine, etc.) are not present to a measur- able extent in mammalian or bacterial cells, but the validity of this assertion may be questioned, since it cannot be reconciled with the numerous reports in the literature on the isolation or identification of peptide material from animals, plants, and microorganisms.1'- Where negative results have been obtained in the search for peptides as cellular constituents, the possibility must be considered that the meth- ods applied were not adequate; for example, because of failure to react with ninhydrin or because of tight binding to the cellular proteins. Also, the question of the variations in the peptide level, which may depend on the physiological state of the cell, may be of importance but cannot he properly assessed at present. A logical difficulty in the 114 Essays in Biochemistry view that peptides are absent from living cells is presented by the problem of the chemical mechanism of the intracellular degradation of proteins. If peptides are not intermediates in this process, is it suggested that protein degradation, like protein synthesis, occurs on a ''template"? In the face of the present situation in regard to the question of specificity in the biosynthesis of proteins, it seems profitable to employ model systems as objects of study as has been done in attempts to determine the role of ATP in peptide synthesis. The model systems that have proved to be of special interest have involved tissue pro- teinases as the catalytic agents and, as substrates, peptides or amino acid derivatives of well-defined structure. It has long been customary to speak of proteinases solely as catalysts of hydrolytic reactions, but the work of the past few years has shown clearly that this is too restrictive a formulation since proteolytic enzymes are now known to catalyze replacement (or transfer) reactions of the following type: PvCO— NHR' + NH2X ^ RCO— NHX + NH2R' Such reactions have been termed "transamidation" or "transpep- tidation" reactions.13 Extensive studies of the action of purified pro- teinases (e.g., chymotrypsin, papain, ficin, cathepsin C) in catalyzing the hydrolysis of peptide derivatives have defined many of the struc- tural features required in the substrate for enzymic action. All experi- ments performed thus far on the catalysis of transamidation by these proteinases have shown that the specificity with respect to the struc- ture of the compound containing the sensitive CO-NH bond is the same for transamidation as for hydrolysis. However, in the catalysis of replacement reactions, where water is no longer the common reactant, the nature of the amine that serves as the replacement agent also is important in definining the specificity of enzyme action. For example, in the reaction catalyzed by the plant proteinase papain, in which carbobenzoxyglycinamide is the component having the sensitive CO- NH bond (only the terminal CO-NH 2 bond is susceptible to enzyme action), and glycylglycine, L-leucylglycine, or D-leucylglycine is the replacement agent, only about 13% of the amide reacts with glycyl- glycine under conditions (pH 7.5, 37°) where about 65% of the amide reacts with L-leucylglycine, and about 8% with D-leucylglycine.14 The extent of transamidation observed in each of these three experiments is a measure of the relative efficiency of the replacement agent in compet- ing with water for reaction with the amide. In the case of the papain- catalyzed interaction of carbobenzoxyglycinamide and L-leucylglycine, The Biosynthesis of Peptide Bonds 115 R R' I I Cbzo-NHCH2CO— NH2 + NH2CHCO— NHCHCOOH ^± R R' Cbzo-NHCH2CO—NHCHCO— NHCHCOOH + NH3 the replacement agent is so much more effective than water at pH 7.5, despite its lower molar concentration, that the predominant reaction is one of replacement rather than hydrolysis. On the other hand, the enantiomer D-leucylglycine is much less effective in this competition, as is glycylglycine. Results of this kind show clearly that, in transamida- tion reactions, the specificity of enzyme action is more exacting than in hydrolysis. Studies on the mechanism of transamidation reactions have demon- strated that it is the uncharged amino group of the replacement agent that is reactive; the extent of transamidation at a given pH depends, therefore, on the pK' of the corresponding cation.15 For this reason, the extent of transamidation observed with the above dipeptides (pK'2 ca. 8 1 is much greater at pH 7.5 than at pH 5, the optimum for the hydrolytic action of papain. The reaction between carbobenzoxyglycinamide and L-leucylglycine illustrates the replacement of a small group ( — NH2) by a dipeptide unit, thus effecting the elongation of the peptide chain from the car- boxyl end of the sensitive substrate. In general, transamidation reac- tions in which an a-amide is converted to a peptide appear to be exergonic reactions, as suggested by the thermodynamic data presented in Table 1. The exergonic nature of enzyme-catalyzed elongation of peptide chains may be further illustrated by results obtained with the intracellular proteinase cathepsin C, purified from beef spleen. This enzyme only acts on derivatives of dipeptides composed of two a-amino acid residues (X may be either NH2, as in an amide, or OCHo, as in an ester). If cathepsin C is allowed to act on a dipeptide amide such R R' I I I NH2CHCO— NHCHCO— X as L-alanyl-L-phenylalaninamide at pH 7.5, the extent of hydrolysis is negligible and there separates from the solution a precipitate which, on chemical analysis, was found to be the amide of the hexapeptide ( 1. 1 Ala. ( l 1 Phe. ( li Ala. (D Phe. (l I Ala. (l) Phe.lfl 116 Essays in Biochemistry This product was obtained in a yield of 86% of the theory, based on the amount of NH4+ liberated. Clearly, what had occurred was the proteinase-catalyzed polymerization of the dipeptide amide by two successive transamidation reactions. It is extremely probable that at each step, an alanylphenylalanyl unit was added to the amino end of the growing peptide chain, in a manner analogous to the action of crystalline muscle phosphorylase, which catalyzes the successive addition of glucosyl units of glucose-1 -phosphate to the non-reducing end of a growing amylose chain. There can be no doubt, therefore, that the known intracellular proteinases can catalyze transamidation reactions in which peptide chains are lengthened and that these enzymes exhibit considerable specificity in these reactions. However, no information is available about the role of the cathepsins in the metabolism of intact animal cells, and nothing can be said at present about the actual occurrence of transamidation reactions in the biosynthesis of proteins. All that can be stated is that the intracellular proteinases are the only well- defined cellular catalysts known to exhibit the specificity to be expected of enzymes that could participate in protein synthesis. One may therefore propose, as a working hypothesis, that a multienzyme system, composed of a number of intracellular proteinases, is responsible for the specific elongation of the peptide chains of proteins. According to this view, the characteristic sequence of amino acid residues in peptide chains is the consequence of the coupled action of a series of proteinases that differ in specificity. An objection that has been raised against this proposal is the difficulty of imagining a sufficiently large number of proteinases that may be expected to be needed to make one type of protein and the much larger number required for the synthesis of all the different proteins. However, there appears to be no valid reason (other than the search for simplicity I against the participation of a series of proteinases in the synthesis of a protein. Also, there does not appear to be any a priori objection to the possibility that a given proteinase could participate in the synthesis of more than one protein. Certainly it would be more in accord with the knowledge gained from the study of other biochemical processes, involving the metabolic synthesis or degradation of molecules much simpler than the proteins, to assume that a multienzyme system is involved. Such an assumption is not incompatible with the experimental results that have led to the formulation of the "template" hypothesis. If the term "template" is considered to represent an organized cellular assembly of enzymic catalysts (plus other constituents such as nucleic acids I The Biosynthesis of Peptide Bonds 117 thai rapidly catalyzes a series of successive peptide syntheses without releasing into the cellular fluids appreciable amounts of peptide inter- mediates, and if the nature of this assembly is genetically controlled, then much of the difference in opinion becomes a difference in language. The significance of the proteinase-catalyzed transamidation reac- tions that have been studied thus far does not lie in demonstrating processes involving peptide derivatives known to occur in nature, but rather in suggesting new approaches to the study of the energetics and specificity of the enzymic synthesis of peptides. To convert amino acids into a form that is capable of reacting with other amino acid residues, the participation of ATP is required, as has clearly been shown in the studies on the biosynthesis of peptides such as glutathione. On the other hand, once the y-glutamylcysteine bond of glutathione has been formed, it is capable of reacting with other amino acids or peptides in transpeptidation reactions, as has been shown by Hanes17 and by Waelsch.3 - Similarly, in the synthesis of glutamine, ATP is required, but, once formed, the CO-NEP bond can react, in the presence of suitable enzyme preparations, with amines such as hydroxylamine or amino acids (Speck, Elliott, Thorne, and Williams). The most plausible interpretation of these findings has been the suggestion that a "y-glutamyl enzyme" is an intermediate both in the synthesis of glutamine from glutamic acid and in the transamidation reactions of glutamine. In the light of these important studies on the intimate relation between the direct synthesis of CO-NH bonds and transamidation reactions in which they may participate, it seems justifiable to extend the hypothesis drawn from the reactions of y-glutamylcysteinc and of glutamine to include the action of intracellular proteinases. This would involve the assumption that, in the catalysis by papain of the reactions of carbobenzoxyglycinamide, there is formed as an "acti- vated" intermediate, a "carbobenzoxyglycylpapain," that can react either with water or with an amine such as L-leucylglycine. Similar "acvl enzyme" compounds could be invoked in the case of other trans- amidation reactions studied. It may be expected that such "acyl enzyme" compounds would be formed more readily when the reactive carboxyl group is joined in a bond that has a relatively high free energy of hydrolysis. This appears to be the case with the a-amides of acylamino acids and peptides. As noted earlier in this essay, little is known about the enzymic mechanisms whereby free at-amino acids are "activated" in reactions involving ATP. Until direct evidence for such mechanisms is avail- 118 Essays in Biochemistry able, it seems plausible to assume, as was suggested some years ago,1'- La that the coupling between the energy-yielding processes of the cell and the synthesis of the peptide bonds of proteins may be funneled through a relatively small number of amides or peptides such as glutamine or glutathione. At best, this view can still be considered only a working hypothesis. However, it seems a useful concept in attempting to discern new experimental approaches to the difficult task of elucidating the chemical pathways in the biosynthesis of peptide bonds. References 1. A. Dobry, J. S. Fruton. and J. M. Sturtevant, J. Biol. Ch< m.. 195, 149 (1952). 2. M. Bergmann and J. S. Fruton, Ann. X. Y. Acad. Sci., 45, 409 (1944). 3. F. Lipmann, Science, 120, 855 (1954). 4. F. Lipmann. in W. D. McElroy and B. Glass, The Mechanism oj Enzyme Action, p. 599, The Johns Hopkins Press, Baltimore, 1954. 5. K. Bloch, J. E. Snoke, and S. Yanari, in W. D. McElroy and B. Glass, Phosphorus Metabolism, II, p. 82, The Johns Hopkins Press, Baltimore, 1952. 6. V. du Vigneaud, H. C. Lawler, and E. A. Popenoe, /. Am. Chem. Soc, 75, 4880 (1953). 7. N. S. Ging and J. M. Sturtevant, ./. Am. Ch< m. Soc, 76, 2087 (1954). 8. M. V. Simpson and S. F. Velick, ./. Biol. Chem., 208, 61 (1954). 9. D. S. Hogness, M. Colin, and J. Monod, Biochim.et Biophys. Ada, 16, 99 (1955). 10. E. F. Gale and J. P. Folkes, Nature, 173, 1223 (1954). 11. M. Vaughan and C. P. Anfinsen, J. Biol. Chem... 211, 367 (1954). 12. R. L. M. Synge, in G. E. W. Wolstenholme and M. P. Cameron, The Ch< mi- cal Structure of Proteins, p. 43. J. & A. Churchill, London, 1953. 13. J. S. Fruton. Yale J. Biol, and Med., 22. 263 (1950). 14. Y. P. Dowmont and J. S. Fruton, J. Biol. Chem., 197, 271 (1952). 15. R. B. Johnston, M. J. Mycek, and J. S. Fruton, J. Biol. Chem., 1S5, 629 (1950). 16. J. S. Fruton. W. R. Hearn, V. M. Ingram, D. S. Wiggans. and M. Winitz, ./. Biol. Chem., 204, 891 (1953). 17. C. S. Hanes. F. J. R. Hird. and F. A. Isherwood. Biochem. ./.. 51. 25 (1952). IS. P. J. Fodor, A. Miller, and H. Waelsch, J. Biol. Chem., 202, 551, 203, 991 (1953). 19. J. S. Fruton, R. B. Johnston, and M. Fried, J. Biol. Chem., 190, 39 (1951). On the Nature of Caneer SAMUEL GRAFF Cancer is a familiar clinical problem, but its definitive inherent characteristics are still obscure. Many experiments have attested to the similarities between cancerous and normal tissues, but few have revealed unique disparities on which therapy could be based. Under such circumstances imaginative speculation within the framework of evidence is a necessity unless we are to resign ourselves to blind empiricism in the search for the cure for cancer. The biochemist in cancer research soon finds himself in an ever-widening circle of excur- sions into tangential disciplines in his search for a clue to the intrinsic nature of cancer, for it is only with the aid of observations apparently far removed from formal biochemistry that an hypothesis leading to therapy can be derived. Cancer is a cellular aberration distinguished by autonomy and ana- plasia, that is, by disregard for normal limitations of growth and by loss of normal organization and function. In this essay we wTill try to clothe these dry definitive bones with experimental conclusions and with plausible speculations, parading our conceptions of the origin of cancer and of possible chemotherapy whenever they contribute to this end. In effect, we set forth a comprehensive hypothesis on the nature, the cause, and the cure for cancer. The wide prevalence of cancer among animals permits of controlled experimentation to a degree which is unusual among other degener- ative diseases. Carcinomas, sarcomas, lymphomas, leukemias, as solid tumors or as ascitic suspensions, are readily available in variety as regards both host and growth characteristics. Many of these tumors occur spontaneously, and practically all of them can be maintained by serial passage in appropriate hosts. Some workers have the view that it is immaterial which tumor is selected for research, that they 119 120 Essays in Biochemistry are all fundamentally identical, and that the eventual cure for one is the cure for all. It is the immediate experiment which dictates the choice of tumor or tissue. The many types of cancer enumerated, as well as those occurring in humans where experimental access is not as good, differ markedly in their transplantability, their anaplasticity, and their capacity to grow rapidly. Rapid growth alone, however, is neither a necessary nor a sufficient criterion for cancer; many cancers grow very slowly. Many normal tissues, on the other hand, are rela- tively hyperplastic or may become so under non-carcinogenic stimuli.'" Embryonic development too consists of rapid cellular multiplication in good part, at least in early phases. It may be instructive, therefore, to compare the growth of a cancer with that of a frog embryo. The latter is comprised of a closed system containing a reservoir of metabo- lites sufficient for development to proceed to a fairly advanced stage without external nutritional supply. Fertilization stimulates the egg to rapid and orderly cleavage, differentiation, and growth, some phases of which can be studied as discrete processes even though they are not sharply separated in time. Early cleavage is signaled by an increase in oxygen consumption which rises gradually throughout development. The fertilized egg continues to divide without apparent morphological differentiation through the blastula stage, the total mass remain- ing constant, and segmentation producing progressively smaller cells. Each new cell, however, has a nucleus and nuclear apparatus visibly equivalent to the original single-cell nucleus. It might appear to the microscopist that DNA is being made at a prodigious rate, but, in fact, there is no synthesis of DNA at all in this early phase of development. Nor is synthesis necessary, for it has been shown that the unfertilized egg already contains an enormous amount of DNA, many times, indeed, the amount of DNA which could be accommodated in one nucleus.1 This is consistent with the fact that the total purines in the embryo are quite constant in amount in the unfertilized egg, the neurula, blastula, and early gastrula stages through which nuclei have been replicated perhaps 40,000-fold.2 Early cleavage in the frog embryo, then, entails only the reorientation and redistribution of DNA which was preformed in the ovary during maturation. The embryo does not have to bear the burden of synthesis of DNA until comparatively late in development, that is, until the late gastrula stage when differentia- tion commences. Early cleavage, where no growth or synthesis occurs, * It is characteristic of carcinogens that the malignant change persists even after the stimulus is removed. On the Nature of Cancer 121 is not expensive in energy. Early respiration only adds to the store of high-energy intermediates which are reserved for the endcrgonic syntheses of growth and differentiation. Although the dividing embryo has a high respiration rate, the egg can divide just as well under anaerobic conditions.3 Some high-energy phosphate bonds are split during anaerobiosis indicating that at least a little energy might be required for cleavage, but glycolysis alone is sufficient to meet this moderate demand. No growth or synthesis takes place during the time when glycolysis alone can supply all of the energy needed.* The cancer cell, on the other hand, has an inordinate growth obliga- tion wThich the early dividing embryo does not have. Even so, it lacks not only the reserve of substances which appear to be indispensable for growth and differentiation but also the capacity to make them. The vaunted capacity of cancer cells for anaerobic glycolysis is indeed a drastically insufficient compensation for the inadequacy of their oxidative energy generation mechanisms. So sluggish is the cancer cell aerobically that its only alternative in the face of the excessive demands is to parasitize the host for the necessary energy cycle inter- mediates, and in this way impair the host further, by reduction of the liver catalase,4 or by damaging the heart,5 for example. This inertia of the cancer cell was demonstrated in striking fashion by the experi- ments of Busch and Baltrush,6 in which it was found that tumors are incapable of metabolizing either acetate or pyruvate. It has been shown also that the rate of growth of tumor implants is rigorously limited by vascular supply,7 a fact which also suggests that cancer growth rates depend on parasitic diversion of oxidatively generated factors. It has been observed clinically also that the most rapidly growing tumors are those in blood-rich areas. In tissue culture too it has been noticed that tumor cells will grow only in the vicinity of the surface, whereas normal cells will often grow throughout the medium, a fact which also can be interpreted to mean that innate glycolysis alone, however large, is insufficient for growth, and that the growing tumor cell must use oxygen. The usual tissue-culture practice involves a discontinuous process which results in alternately feasting and starving the cultures, these rigors being only meagerly compensated for by the frequency of transfer to fresh media. The cells in the center of a hanging-drop culture, for example, sooner or later begin to suffer from oxygen deprivation and intoxication * High rates of glycolysis are a feature of growing tissues irrespective of malig- nancy, hut no growing tissue can survive long under anaerobic conditions. 122 Essays in Biochemistry and become necrotic. This, also, is the way most tumors behave in the intact animal. Cancer cells, then, appear to be members of a race deficient in oxidative mechanisms.""' Their deficiency or immaturity is a permanent inheritable factor. Cytoplasmic inheritance controls differentiation. The mutant we call cancer might well be characteristically deficient in cytoplasmic high energy conversion enzyme systems which are required for differentiation. The different effects of aerobiosis and anaerobiosis could be a reflection of the relative efficiency of the two processes. A 50% difference in oxygen tension such as obtains at an altitude of 20,000 ft. can spell the difference between life and death to an unacclimatized individual. The Himalayan expeditions have brought into sharp focus the fact, previously established in the laboratory, that human beings as well as animals, adults at least, can be adapted by suitable acclimatization, not only to survive, but also to do hard work in the rarefied atmosphere of altitudes 20,000 ft. or higher. But when acclimatized animals were implanted with tumors and then returned to high altitude the growth of tumors was markedly inhibited.9 In many instances the tumors regressed completely, and the animals were permanently cured. It seems as though frustration of either division, growth, or differentiation must lead inevitably to cell death. There is no alternative to com- pletion of these developmental compulsions. It appears to be impos- sible for embryonic cells or tumor cells to revert to resting stages once chromosomal rearrangement for division or cytoplasmic mobilization for growth or differentiation have taken place. A cell is most vulner- able during developmental change, whereas resting adult cells are far more resistant. The widely accepted rule of thumb that cells under- going active mitosis are more sensitive to lethal agents such as X rays than are resting cells is a restatement of the proposition that a cell dies if its division under urgency is blocked. The energy from gly- colysis plus only a modicum of oxidatively generated energy might be sufficient for acclimatized resting host cells but hardly so for vora- cious growing cells. Cancers or other obligatorily growing cells cannot survive partial anoxia; they cannot be acclimatized as can resting cells. It does not follow from the above, however, that the cause of cancer is necessarily related to interference with normal oxidative behavior. Mutagenesis is a sine qua non of carcinogenesis. Mutagens are non- * This conclusion has been reached before from other considerations. Van R. Potter is well known as an early and active protagonist of the oxidative1 deficiency hypothesis in cancer.8 On the Nature of Cancer 123 specific in their effects. Cancer is only one of the many possible muta- tional responses to injury or other embarrassment of normal functional processes. Haddow has suggested that malignant transformation is an artifact of adaptation, that a carcinogen which cannot be lethal or very virulent provokes an adaptive response in a cell which is selective within its capacity to retain the carcinogen or in its suscepti- bility to interference by the carcinogen. The cell can evade for a long time, but under continued assault or with the passage of time it succumbs to give rise to a race of mutants which lack the ability to mature. The price of survival by mutation is high; the price is the loss of those functional endowments which are presumably pre- empted by carcinogens. The azo dye carcinogens, for example, have been described as depriving liver cells of those proteins required for growth regulation.10 Thereafter the descendants of the afflicted cells no longer have the wherewithal to differentiate like normal liver cells. Many of the chemical carcinogens are known to behave as mutagens in other biological systems as well. It appears reasonable to suppose that those mutagens which incite cancer should exist widely in nature rather than in the laboratory or in highly industrialized surroundings alone. We shall therefore pay particular attention to the naturally occurring tumors, rather than those produced by chemical agents, in tracing the cause of cancer. Among the animal tumors the mammary carcinoma of the mouse is of especial interest because it bears more than a superficial resemblance to mammary carcinoma in the human. This disease occurs at random among wild mice, but its incidence is truly devastating among certain highly inbred laboratory strains, occurring in 96% of the Bar Harbor C3H and the Paris R III strains in adult life. Crossbreeding experi- ments once made it appear that the disease was an inherited condition, but on closer scrutiny of the data it became necessary to postulate an extrachromosomal factor in the transmission of this form of cancer. From this postulate emerged one of the significant landmarks of cancer research. By the elegantly simple expedient of foster nursing Bittner in 1936 demonstrated that this extrachromosomal factor was trans- mitted with the milk of the nursing mother. AYhen newly born C3H (high cancer strain) females were foster-nursed on C57 (low cancer strain) foster mothers the incidence of mammary carcinoma in the subsequently matured C3H mice was reduced markedly. Conversely, when the newly born C57-strain females were nursed on C3H foster mothers cancer of the breast appeared in many of them on maturity.11 It is significant that in these foster-nursing experiments the C3H 124 Essays in Biochemistry foster mothers had not developed any visible or palpable sign of cancer at the time they nursed the young C57s. Indeed, the first tumor appeared only after the passage of several months. We can therefore be certain that there had been no transfer of cancer cells to the nurslings, since, if there had been, the transferred cells would have erupted into full-grown tumors in a matter of days rather than of Fig. 1. Electron micrograph of mouse mammary carcinoma virus as purified by convection-current electrophoresis. The drop preparation was shadowed at tan-1 12° with chromium. Magnification 17,500X (McCarty-Graff). months. The conclusion is inescapable; mammary carcinoma in mice, a disease of the adult female, is transmitted from mother to offspring by an infectious agent in the milk. Complete confirmation of this conclusion was provided by the isolation and characterization of the virus in the author's laboratory.12 The viral bodies pictured in Fig. 1 were isolated from high cancer strain milk. The inoculation of these particles is equivalent to the ingestion of "infected" milk. Now, Rous had already shown, as far back as 1911, that cell-free extracts of chicken sarcoma were able to reproduce the disease on inoculation into new hosts.13 The comprehensive implications of Rous and Bittner's discoveries are not widely appreciated, but subsequent On the Nature of Cancer 125 developments now compel serious consideration of the proposition that all forms of cancer are cellular reactions to infectious processes. Viruses have already been implicated in a number of other animal cancers, including leukemia, which also has a human parallel. The techniques employed in the animal-tumor experiments, however, are not applicable to human beings since cancer viruses like the cancers they cause are species specific and variable in form and behavior. Perhaps the only applicable property common to all viruses is that they propagate. Fortunately the intact animal host is not essential; suspensions of living cells, tissue cultures, can support the growth of viruses in many instances. It is important to distinguish between the cause of a particular form of cancer, a virus perhaps, and the cellular reaction to that virus infection, the cancer. Often both can be separately cultivated, transplanted, and observed. Tissue culture, now in the full bloom of its renaissance, appears to offer the means for elucidation of the role of viruses in human cancer and also of the metabolic deviations which are the inherent characteristics of cancer. Experiments are in progress wherein human secretions are admixed with tissue cultures on the expectancy that carcinogenic viruses, if present, will reveal themselves either by increase in number of particles or by malignant transformation of host cells. It is intrinsic to our virus hypothesis that the presence of the virus in the cancer cell is wholly fortuitous, since the mutation produced by virus aggression is self-perpetuating. Although the malignant trans- formation appears to be an escape mechanism the malignant cell itself is not necessarily unpropitious for viral existence. Porter14 was able to obtain the beautiful electron micrograph reproduced as Fig. 2 by tissue culture of mouse mammary carcinoma cells in which the virus was still present. This picture demonstrates the exuberant growth of the virus in a cell. Note the resemblance to the isolated particles of Fig. 1. Now, Bittner's informative foster-nursing experiments established a procedural pattern for other workers who were in possession of inbred strains of animals with high incidence of one form of cancer or another. MacDowell of the Carnegie Institution had been investigating the genetic aspects of leukemia which was endemic in his C58-strain mice when he undertook to test the possibility of milk transmission of the disease by reciprocal foster nursing with Storrs-Little strain which was free of leukemia. His first experiments were inconclusive, and so he stubbornly repeated the experiment, this time with larger numbers. In so doing, however, it was necessary for him to employ all of the 126 Essays in Biochemistry foster mothers he had available, including a number of older mice. The second experiment also failed to throw any light on the trans- missibility of the disease, but MacDowell did observe that the inci- dence of leukemia was somewhat diminished in those C58 mice which had been nursed on old Storrs-Little mothers. This observation was verified in a third experiment wherein old foster-mother Storrs-Littles Fig. 2. Electron micrograph of a preparation from a 4-day-old culture of a spon- taneous mammary tumor explant fixed over osmic acid shadowed with gold at tan-1 10°. Magnification 17.500X (Porter and Thompson14). were used exclusively.15 These experiments established that the milk of old Storrs-Little mice contains a substance which is either prevent- ative or curative of leukemia. Leukemia, like mammary carcinoma, is a disease of the adult mouse. The transmission of the causative agent, whatever it might turn out to be, occurs antenatally or in nursing, probably the former. Leukemogenesis takes place over a period of time. The milk inhibitor was given early in this period, that is, before the leukemic state. It can be inferred then that Mac- D< >well's substance is prophylactic rather than therapeutic. Experi- ments on the isolation and characterization of the substance are still in progress at this writing, but it seems reasonable to suppose that this substance acts by compensating for a deficiency produced by the leukemogenic mutagen. On the Nature of Cancer 127 The deficiencies of cancer, stressed in this article, contrast strongly with the assumed totipotency of cancer so implicit in the concept of autonomy. Chemotherapy would be a hopeless endeavor indeed if the normal cells were forced to compete with complete autotrophics which could avail themselves not only of simple substrates but also of nu- merous alternate metabolic pathways. The tolerance of adult differ- entiated host cells to most drugs is limited. Although they do not require as much energy as do growing cells they cannot adapt to radical changes in their chemical functional environment. Indeed, it may be just such adversity which produces cancer! Curiously enough it has been suggested and also actually demonstrated that weak or very poor carcinogens are inhibitors of cancer. Cancer prevention in man is so far restricted to the elimination of carcinogens from the diet and from the immediate environment. But if it should be found that viruses do in fact play a major role in the dissemination of cancer, then presumably antisera to specific viruses and as yet undeveloped antibiotics would find employment in diagnosis and prophylaxis. It is perhaps fortuitous that some of the experi- mental drugs against cancer, 8-azaguanine, for example, are also anti- viral. The practical aspects of the problem, however, limit therapy to the treatment of cancer after diagnosis has been made, and, indeed, without respect to the precedent chain of events. It must be presumed that diagnosis will not be made until the cancer is well established and even metastatic. Therapy then consists in selective attack on neoplastic tissue. Most of the effort in design and synthesis of compounds which will either destroy or modify cancer cells preferentially is devoted to growth antagonism. Preferential, because normal differentiated cells do not have to divide; cancer cells apparently do. Growth antagonism is often thought of as being synonymous with interference with nucleic acid metabolism. This easy assumption is probably an oversimplifica- tion, but it has given rise to many new compounds and a number of interesting biological applications. Animal cells make their nucleic acids from amino acid; there is no nutritional requirement for either preformed purines or pyrimidines. Adenine is exceptional among the nucleic acid bases, since it is the only one which is accepted and incorporated by the metazoans. Part of this adenine is converted to polynucleotide guanine. Guanine itself, on the other hand, is not incorporated at all. The protozoan Tetra- hymena geleii, however, has an obligatory requirement for guanine. Kidder, who had interested himself in the dietary requirements of this 128 Essays in Biochemistry organism, also found that 8-azaguanine was a growth inhibitor of T. geleii by competition with guanine.16 He then discovered that this same compound would inhibit the growth of certain mouse tumors. How or why this drug inhibits tumors is still unknown." Tumors inhibited by 8-azaguanine are not arrested metabolically ; their nucleic acid turnover is approximately the same as uninhibited tumors. Doses of the drug which might be large enough to effect cure cannot be tolerated by the host. The drug has absolutely no effect on some tumor types such as sarcoma 180. Thus all tumors are not the same, and the selection of tumors for experimentation is not entirely a matter of convenience or indifference. 8-azaguanine, nevertheless a good experimental carcinostatic agent, is useless for human therapy because of its limited solubility and because of its toxicity. These objectionable features led to a number of synthetic maneuvers in an effort to increase the solubility of the active groups, to decrease the toxicity, and above all to increase the activity. Acetylation of the OH group proved to be futile; the ester was inactive biologically, apparently because hydrolysis did not occur with sufficient speed. Neoazaguanine, the sulfoxalate (by analogy with neoarsphenamine) was also inactive. What was most disappointing, however, was the inertness of the two pyrazolopyrimidines, 5-amino-7- hydroxypyrazolo-4,3-d-pyrimidine and 6-amino-4-hydroxy-pyrazolo- 3,4-d-pyrimidine, both isomeric with guanine. The high expectations for the latter compounds arose from an overly naive reliance on the applicability of competitive inhibition to a com- pletely unknown metabolic system. A competitor is an agent which differs so subtly from a required metabolite that it is accepted without discrimination in the initial incorporative reaction. It is only in the events which follow the diversion of enzymes that the cell discriminates between the true and the false, and biosynthesis grinds to a halt. If the proffered compound differs so subtly from its natural analog, there may be no distinction at all and the cells may utilize the substi- tute with equal facility. If, on the other hand, the alteration is too gross the compound is completely ignored in the first instance. Even * One could now surmise that 8-azaguanine acts by inhibiting an oxidatively generated factor. If this were true, it would follow that the drug would act synergistically with hypoxia, and that its toxicity would be diminished after acclimatization of the host to hypoxia. Furthermore, Drs. Philip Feigelson and J. D. Davidson of this institution have adduced strong evidence that the car- cinostatic activity of 8-azaguanine depends on its capacity to inhibit xanthine oxidase which appears to be necessary for the biosynthesis of guanine. (Personal communication.) On the Nature of Cancer 129 when compounds do bring about biological effects they may do so by interference in such involved reactions as to preclude any facile idea of single competitive inhibition. Although 8-azaguanine, its analogs, and the compounds described in the following bear some resemblance on paper to the naturally occurring purines it would be foolhardy indeed at this time to ascribe their activity to structural similarity. Furthermore, there is even no conclusive evidence that simple metabo- lites like purines or pyrimidines play any role at all in the utilization of foodstuffs for biosynthesis. Nevertheless our failures with the guanine analogs urged us on to further synthetic efforts. Variations of the bicyclic rings such as benzimidazoles, benzotriazoles, and quinoxalines were prepared as shown in Fig. 3.1T In addition simpler test systems were sought wherein it was hoped that the finer nuances of biological reactivity could be observed in line with the hypotheses bared above. Microorganisms and Rcma pipiens embryos were employed to investigate the effects of the compounds. Several of the compounds did in fact prove to be inhibitors of Escherichia coli, but in a special way. They delayed growth, but only by prolonging the lag phase, after which growth was resumed at its normal rate. This behavior could be explained in either of two ways: either the compounds were detoxicated metabolically, or the organisms mutated to resistant types. Both phenomena were found to take place in the various tests. Four of the 35 compounds tested on E. coli were mutagenic. It is germane to Haddow's hypothesis alluded to earlier that these four mutagens were all inhibitors and that only toxic com- pounds were mutagenic. Mutation occurs, however, at concentration levels far below those necessary to produce inhibition, i.e., lengthening of the lag phase.* It would be an amiable gesture of nature if the bacterial mutagens 4-methoxy-6-nitrobenzotriazole and 6-hydroxy-4-nitrobenzimidazole, for example, should turn out to be carcinogens also. But this is only moderately to be hoped for, since carcinogens are thought to form more or less stable compounds with cellular components. Bonding need not be stable for mutagenic action on bacteria where the cells are bathed in the toxic medium, but combination is probably necessary in more highly differentiated systems where circulatory and detoxicat- ing mechanisms might play a greater part. Neither does it follow * Unpublished work of Sheldon Greer and Professor Francis J. Ryan, Depart- ment of Zoology. Columbia University. 130 Essays in Biochemistry that agents which affect one species or even one cell type in a species should exhibit parallel or similar effects in another species or another cell type. Thus the application of these substances to E. coli must be weighed as an illustration of general principles without expectancy of utility in higher forms. OH \ OH "sN-^ JL^ /°H HjN-L^ JL / HoN III NaOSO-CHo-N IV OH // H,N VI OCH O.N 09N OCH VII VIII IX Fig. 3. I. Guanine. II. 8-azaguanine. III. Acetylazaguanine. IV. Azaguanine sulfoxalate. V. 5-amino-7-hydroxypyrazolo-[4,3-d]-pyrimidine. VI. 6-amino-4- hydroxypyrazolo-l 3,4-d]-pyrimidine. VII. 4-methoxy-6-nitrobenzotriazole. VIII. 6-hydroxy-4-nitrobenzimidazole. IX. 5-methoxy-7-nitroquinoxaline. The particular value of experimentation with the frog embryo, how- ever, lies in the possibility of destruction between cleavage and dif- ferentiation. The new compounds described above are still to be explored in the light of their effects on energy cycles, but they have already been found to exhibit stage specificity.18 One group of com- pounds, mostly quinoxalines, 5-methoxy-7-nitroquinoxaline, for exam- ple, selectively inhibits and arrests embryos in early cleavage stages. On the Nature of Cancer 131 Another group represented by 4-methoxy-6-nitrobenzotriazole arrests only late, partially differentiated stages. The specificity in activity is remarkable; even % hour of exposure to 0.1 ing./ml. of the quinoxa- line brings about immediate arrest of the two-cell stage, whereas ex- posure of 48 hours or more is required for arrest at the tail-bud stage. Exposing two-cell stage embryos for as long as 30 hours to the same concentration of the benzotriazole still permits development to the blastula or early gastrula stage before death ensues. The selective response to the quinoxaline appears to bear some relation to the mitotic rate and resembles the response to radiation and to radio- mimetic drugs. The benzotriazole effects, on the other hand, are characterized by the absence of radiomimicry and by the heightened sensitivity of the embryo with increasing age and with differentiation. These compounds illustrate the possibility of synthesizing compounds which can selectively attack either growth or differentiation. The finding of such compounds may be a matter of chance and perseverance, since we cannot yet discern any relation between chemical structure and function in this or any other series of compounds so far applied to the problem. We have indicated not only the desirability but also the practicability of such specificity in the chemotherapy of cancer. If the cancer cell is a deficient mutant resulting from a virus infection, then selective chemotherapy is within the grasp of the biochemist who can pin-point the deficiencies and fashion drugs to exploit them. It appears reasonable at this time to suggest that therapy be directed toward energy-cycle mechanisms whether in the tumor or the host. Although the author alone is responsible for the ideas expressed herein he owes a debt of gratitude to his associates, Ada M. Graff, Morris Engelman, Horace B. Gillespie, and Kathe B. Liedke for carry- ing out so much of the work from which these ideas are derived. The author is also grateful to the Damon Runyon Memorial Fund, the American Cancer Society, and the United States Public Health Service for financial assistance in the prosecution of these researches. References 1. L. C. Sze, J. Exptl. Zool, 122. 577 (1953). 2. S. Graff and L. G. Barth, Cold Spring Harbor Symposia Quant. Biol., 6, 103 (1938). 3. L. G. Barth and L. Jaeger, Physiol. Zool., 20, 133 (1947). 4. J. P. Greenstein, The Biochemistry of Cancer, Academic Press, New York, 1947. 132 Essays in Biochemistry 5. W. Antopol, S. Glaubach, and S. Graff, Proc. Am. Assoc. Cancer Research, 2, 1 (1955). 6. H. Busch and H. A. Baltrush, Cancer Research, 14, 448 (1954). 7. (o) W. Antopol, S. Glaubach, and S. Graff, Proc. Soc. Exptl. Biol. Med., 86, 364 (1954). (6) P. C. Zamecnik, R. B. Loftfield, M. L. Stephenson, and J. M. Steele, Cancer Research, 11, 592 (1951). (c) T. H. Algire and F. Y. LeGal- lois, J. Natl, Cancer Inst., 12, 399 (1951). 8. V. R. Potter, G. A. LePage, and H. L. King, ./. Biol. Chem., 175, 619 (1948). 9. A. Barach and H. A. Bickerman, Cancer Research, 14, No. 9, 672 (1954). 10. S. Sorof and P. P. Cohen. Cancer Research, 11, 376 (1951). 11. J. J. Bittner, Science, 84, 162 (1936). 12. S. Graff, D. H. Moore, W. M. Stanley, H. T. Randall, and C. D. Haagensen, Cancer, 2, 755 (1949). 13. P. Rous, J. Am. Med. Assoc, 56, 198 (1911). 14. K. R. Porter and N. P. Thompson, /. Exptl. Med., 88, 15 (1948). 15. E. C. MacDowell, Cancer Research, 15, 23 (1955). 16. G. W. Kidder, V. C. Dewey, R, E. Parks, Jr., and G. L. Woodside, Science, 109, 511 (1949). 17. H. B. Gillespie, M. Engelman, and S. Graff, J. Am, Chem. Soc, 76, 3531 (1954). 18. K. B. Liedke, M. Engelman, and S. Graff, J. Exptl. Zool, 127, 201 (1954). Problems in Lipide Metabolism SAMUEL GURIN It is now half a century since Knoop x fed w-phenyl-substituted aliphatic acids to dogs and isolated as excretory products either hip- puric or phenaceturic acids, depending upon the number of aliphatic carbon atoms in the administered acids. Ever since that time, and with a great crescendo during the past decade, biochemists have sought for a definitive explanation of the way in which fatty acids are oxidized by living tissue. The problem could undoubtedly have been epitomized by a certain well-known writer of detective stories as "The Case of the Mysterious Two-Carbon Fragment." Had an imaginary writer begun his story in 1904, it is only now that he would be able to write his concluding chapter. Evidence that 2-carbon fragments arise during the biological oxida- tion of fatty acids has come from so many sources and in such over- whelming volume that there is very little point in belaboring the matter. The isolation of acetyl coenzyme A by Lynen and his group in 1951 - brought the field to a new turning point and settled conclusively the nature of the active 2-carbon fragment. It is fitting to pay tribute to the fine work of Lipmann, Nachmansohn, and others who uncovered the role of coenzyme A and ATP in the activation of acetate. That fatty acids must have their carboxyl groups free prior to their utilization was strongly indicated by Lehninger's work 3 with mito- chondria. Lynen's 4 hypothesis, that long-chain acyl CoA derivatives are produced before fatty acids can be oxidized, fitted beautifully with such a notion. It should of course not be forgotten that the demonstration by Drys- dale and Lardy 5 and Mahler 6 that oxidation of fatty acids could be achieved in mitochondrial extracts paved the way for the spectacular advances in this field. With an appropriate electron acceptor it was clearly established that fatty acids could be activated, oxidized, and 133 134 Essays in Biochemistry- converted by such particle-free extracts to acetoacetate. Furthermore, Drysdale and Lardy were able to demonstrate (1) the formation of acetohydroxamic acid in the presence of hydroxylamine and (2) the quantitative formation of citrate when oxalacetate was supplied. This latter reaction suggested quite strongly the involvement of acetyl CoA, since Stern and Ochoa ' had isolated the enzyme capable of condensing this active intermediate with oxalacetate to form citrate. Acetyl CoA + Oxalacetate ^ Citrate + CoA In view of the evidence presented by Stern, Coon, and Del Campillo 8 concerning the formation of acetoacetyl CoA from acetyl CoA: 2 acetyl CoA ^± Acetoacetyl CoA + CoA there could no longer be any doubt concerning the nature of the active 2-carbon fragment derived from the oxidation of fatty acids. At least three separate enzyme systems appear to be involved in the activation of fatty acids: (1) an enzyme system described by Korn- berg and Pricer," which is capable of activating acids containing 12 to 20 carbon atoms; (2) enzymes derived from beef liver capable of activating C4-C12 acids (Mahler et al.) ; and (3) the well-known acetate activating system. All three appear to require ATP and CoA as well as Mg++. The currently accepted mechanism of oxidation of fatty acids is schematically represented in Fig. 1. Major contributors in this area Fatty acid oxidation r ► RCH„CH2CO-CoA 1H» I RCH:CHCO-CoA 1 l+H20 RCHOHCH2CO-CoA 1 I II- 2H RCOCH2CO - CoA I +CoA 1 — RCO - CoA -* -*- +CH3CO-C0A (recyclized) (generated) Fig. 1. Generation of acetyl CoA by oxidation of activated fatty acids. have been Lynen, Ochoa, and Green. The details of this oxidative mechanism have been fully reviewed elsewhere.10 Problems in Lipide Metabolism 135 It is not surprising thai a number of questions are raised by these striking discoveries. One of these concerns itself with the manner in which the fatty acid moieties of phospholipidcs, triglycerides, and other naturally occurring esters are converted into activated fatty acids. In this connection, a very promising start has been made by Kornberg and Pricer who have demonstrated that acyl CoA derivatives can be enzymatically esterified with L-a-glycerophosphate to yield phosphatide acids. If this type of reaction should prove to be revers- ible (and thermodynamically this seems likely), a thiolytic cleavage would be clearly involved as the initiating step rather than a reaction involving the hydrolytic action of lipase. The detailed mechanism of the initial activation clearly needs further investigation. Although the reversibility of all of the chemical steps of the oxidative cycle seems likely, it has been difficult to establish this experimentally. Stansly and Beinert J1 have been able to demonstrate with purified enzymes the conversion of labeled acetyl CoA to butyryl CoA, pro- vided suitable hydrogen donors were present. Since the thiolytic cleavage of /?-ketoacyl CoA proceeds nearly to completion,8'12 some driving force is necessary to shift the equilibrium in favor of synthesis. Esterification may be an important reaction for this purpose even though this obviously does not provide a complete answer to the problem of lipogenesis. Before turning to problems of lipogenesis, a brief comment should be made about another aspect of the oxidation problem. The original reports 13 that two types of metabolically different 2-carbon fragments are produced during the oxidation of fatty acids have been amply confirmed. According to this concept, the bulk of the 2-carbon frag- ments are probably acetyl CoA (originally described as — CH;.— CO— groups). The terminal two carbons of the fatty acid being oxidized give rise to a fragment (originally described as CH3— CO— ) which preferentially contributes to carbons 3 and 4 of acetoacetate. In other words, not all of the 2-carbon fragments derived from fatty acids are symmetrically incorporated into the two halves of acetoacetate. Lynen 14 has invoked the concept of an acetyl-enzyme complex to describe the behavior of the terminal fragment, whereas Beinert and Stansly 15 picture it as an acetyl-CoA-enzyme complex. CH3CO*-enzyme (CH3CO— CoA-enzyme) + CH3CO— CoA ^ CH3CO*CH2CO— CoA + Enzyme Asymmetrically labeled acetoacetate would in either case be formed. Although these theories are not easy to prove, they are attractive and 136 Essays in Biochemistry- explain most of the phenomena that have been observed. Chaikoff and Brown 16 as well as Green 10 have written extensive reviews of this subject. With the, by now, conclusive evidence that acetyl CoA represents the major if not the sole product of this oxidative cycle, the physiolog- ical disposition of this central intermediate becomes of profound sig- nificance. It has already been pointed out that it may condense with oxalacetate to form citrate presumably for oxidation to C02 by way Fat w /RCO-CoA /~,, Ac -CoA Glucose v ' / II Fragments i ' Deacylase ueacyiase I CH 3 COCH 2 CO - CoA »► CH 3 COCH 2 COO " \ 1 \ CH3COCOOH ^ CH 3 CO -CoA L_^ CH 2 CO -CoA I I I I +co2 COCOOH I + ' •- CH.,COOH Acetylations Citrate CH3-C-OH CH2COO" \ Cholesterol I C02 + H20 Fig. 2. Metabolic pathways of acetyl CoA. of the tricarboxylic acid cycle. Its self-condensation to form aceto- acetate has also been mentioned. In addition to its function as an acetylating agent, it may also be utilized for the biosynthesis of fatty acids and cholesterol. It has, of course, been known for a long time that labeled acetate is incorporated into fat by rat-liver slices. Similar experiments have more recently provided evidence that short-chain fatty acids are to a considerable degree cleaved to "active acetate" prior to their in- corporation. With homogenates and now aqueous extracts of rat liver17 it has been possible to demonstrate the incorporation of pyru- vate or acetyl CoA into fatty acids. In view of the conclusive evidence that pyruvate is converted in large measure to acetyl CoA by means of pyruvic oxidase, it now becomes clear that glucose must be converted to fat primarily by way of pyruvate and acetyl CoA. The first major metabolic process in the transformation of carbohydrate to fat is therefore glycolysis. This is probably true for most tissues; among Problems in Lipide Metabolism 137 those tested that undoubtedly incorporate acetate into fat have been mammary gland, intestinal mucosa, lung, heart, spleen, testes, ovaries, adrenals, and adipose tissue. It should be pointed out that such posi- tive results simply indicate that the above-mentioned tissues contain enzyme systems capable of transforming precursor carbons into fatty acid carbon atoms. The results do not imply that net synthesis has occurred, nor do they provide much information concerning the driving forces necessary to reverse the oxidative pathway in the direction of lipogenesis. It is clear that lipogenesis from acetyl CoA is an endergonic and reductive process. Although oxidation of fatty acids may be demon- strated in mitochondria or mitochondrial extracts, it has not yet been possible to obtain incorporation of labeled pyruvate or acetyl CoA in such preparations. The soluble supernatant fluid is required in addi- tion to mitochondrial extracts before this result can be achieved. A study of the cofactor requirements of the complete water-soluble system 18 indicates that it requires all of the known cofactors of glycolysis, Mg++, ATP, and DPN. In addition, it has been possible to demonstrate a dependency upon CoA. Although there is no doubt that flavoprotein is also required, this has not been demonstrated in such systems. The addition of citrate in relatively high concentra- tions (0.1-0.2 M) is also necessary for maximal lipogenesis in these systems, whereas equivalent concentrations of glutamate, oxalacetate, and a-ketoglutarate have proved to be not nearly as effective. Al- though the reason for this stimulating effect of citrate upon lipogenesis is not clear, there is some evidence that the presence or absence of citrate influences profoundly the amount of labeled acetate incorpo- rated into acetoacetate and /3-hydroxybutyrate. Attempts to replace the supernatant fluid with a source of glycolytic enzymes (rabbit-muscle acetone powder) and suitable substrates have been unsuccessful. Although it is obvious that both high-energy com- pounds and hydrogen donors are essential for lipogenesis, DPNH and ATP have not proved satisfactory nor has it been possible to obtain lipogenesis by enzymatic generation of DPNH and ATP. Relatively inactive supernatant fluids are obtained following brief dialysis; they are also inactive when prepared from livers of previ- ously fasted or alloxanized rats. Mitochondria or mitochondrial ex- tracts obtained from such animals appear to be essentially normal and can be supplemented with supernatant fluid obtained from livers of normal well-fed rats. 138 Essays in Biochemistry It is of interest to speculate about the defect in such "diabetic" extracts which can readily degrade fatty acids but not accomplish the reverse process. The addition of glycogen, hexose diphosphate, or a number of glycolytic intermediates, produces a marked stimulating effect upon the incorporation of labeled acetate or pyruvate into long- chain fatty acids.19 In fact, the addition of hexose diphosphate and ADP restores such lipogenic activity nearly to normal. It is significant that glucose is ineffective, whereas fructose has some stimulating effect- Such results are consistent with the data obtained by Chernick et al.20 who demonstrated that fructose is well utilized by liver slices of dia- betic animals. It appears likely therefore that fructokinase and the remaining glycolytic enzymes are relatively intact in the diabetic state. One is therefore tempted to speculate that the diabetic and fasting states may involve a deficiency of glycolytic intermediates. If this is so, then the addition of phosphorylated intermediates of glycolysis to cell-free systems prepared from diabetic animals should be stimulatory. If this concept is correct, then the failure of the diabetic to synthesize fat 21 can in large measure be ascribed to diminished glycolysis result- ing from the failure to utilize glucose. It is attractive to attempt to explain physiological events in terms of chemical reactions, even though the field is still in a primitive state. The fatty livers associated with the diabetic state cannot obviously be ascribed to accelerated lipogenesis, since such livers cannot syn- thesize fat. Nor can this accumulation of fat be due to a diminished catabolism of fat by the liver since there is abundant evidence to the contraiy. It is apparent that this phenomenon must be explained primarily on the basis of accelerated mobilization and transport of fat from the periphery to the liver. The ketosis of diabetes is understandable if one pictures an over- production of acetyl CoA from fatty acids in the face of a limited supply of oxalacetate. Conversely the antiketogenic effect of carbo- hydrate can be explained if one assumes that there is a resulting increased production of malate and oxalacetate for condensation with acetyl CoA to form citrate. On this basis, pyruvate which forms acetyl CoA, and should therefore possess some ketogenic activity, fails to do so since it can also provide its own oxalacetate (reactions involv- ing fixation of CO2) . That acetoacetate is not readily metabolized in liver is consistent with the absence of an effective activating system in liver. The succinyl CoA transferase system occurs elsewhere 8 and is capable of effectively transferring CoA to acetoacetate. Another reaction of acetyl Problems in Lipide Metabolism 139 CoA which occurs readily in liver involves a condensation with aceto- acetate or acetoacetyl CoA yielding eventually /Miydroxy-£-methyl- glutaconic acid.--' This substance readily yields squalene and cho- lesterol in aqueous liver extracts.23 Although this particular segment of the field presents many uncertainties which need clarification, it is apparent again that an overproduction of acetyl CoA (and aceto- acetate?) provides a situation which is favorable for increased bio- synthesis of cholesterol. That labeled acetoacetate can be incorporated into cholesterol by liver slices has been well established.24 Our rapidly increasing knowledge of the chemical pathways involved in the oxidation and biosynthesis of fat has shown us where the railroad tracks go and something about the location of the switches in this complicated railway system. Turnover studies tell us something about the volume of traffic. We know also that, in certain situations, some switches are open and others are closed. The manner in which hor- monal and physiological regulation influence these biochemical path- ways will challenge the best efforts of biochemists and biologists for many years to come. It is manifestly impossible to do justice to the many outstanding contributions that have been made in this difficult field. Many have not been touched upon. Such problems as the nature of the linkages in complexes of protein and lipide, the mode of synthesis of the long- chain polyethenoid acids, the biosynthesis of carotenoids and various terpenoid substances will occupy the attention of biochemists for many years. An attempt has been made here only to highlight a few prob- lems that happen to interest the author. Excellent review articles have been published by Lehninger,10 Green,10 and Chaikoff.16 Current developments in enzymology have depended very heavily upon information obtained with stable and radioactive isotopes. The author was privileged to witness some of the early applications of these research techniques to the study of intermediary metabolism. It is therefore a source of gratification that, from these beginnings, there has developed a renaissance in intermediary metabolism which is now in full tide. References 1. F. Knoop, Beitr. chem. physiol. Pathol, 6, 150 (1904). 2. F. Lynen and E. Reichert, Angew. Chem., 63, 47 (1951). 3. A. L. Lehninger, ./. Biol. Chan., 154, 309 (1944); 157, 363 (1945); Wl, 413, 437 (1945); 162, 333 (1946). 4. F. Lynen, E. Reichert, and L. Rueff, Ann. Chem., 57/,, 1 (1951). 140 Essays in Biochemistry 5. G. R. Drysdale and H. A. Lardy, Phosphorus Metabolism, 2, 281 (1952); ,/. Biol. Chem., 202, 119 (1953). 6. H. A. Mahler, Phosphorus Metabolism, 2, 286 (1952). 7. J. R. Stern and S. Ochoa, J. Biol. Chem., 191, 161 (1951). , 8. J. R. Stem, M. J. Coon, and A. Del Campillo, Nature, 171, 28 (1953). 9. A. Kornberg and W. E. Pricer, J. Biol. Chem., 204, 329, 345 (1953). 10. A. L. Lehninger, Fat Metabolism, The Johns Hopkins Press, pp. 117-132, 1954; D. E. Green, Biol. Revs., 29, 330 (1954). 11. P. G. Stansly and H. Beinert, Biochim. et Biophys. Acta, 11, 600 (1953). 12. D. S. Goldman, J. Biol. Chem., 208, 345 (1954). 13. M. J. Buchanan, W. Sakami, and S. J. Gurin, Biol. Chem., 169, 411 (1947) ; S. Gurin and D. I. Crandall, Cold Spring Harbor Symposia Quant. Biol., 13, 118 (1948). 14. F. Lynen, Federation Proc., 12, 683 (1953). 15. H. Beinert and P. G. Stansly, J. Biol. Chem., 204, 67 (1953). 16. I. L. Chaikoff and G. W. Brown, Chemical Pathways of Metabolism, Aca- demic Press, 1, 277 (1954). 17. R. O. Brady and S. Gurin. ./. Biol. Chem., 199, 421 (1952); F. Dituri and S. Gurin, Arch. Biochem. and Biophys., 43, 231 (1953). 18. J. Van Baalen and S. Gurin, J. Biol. Chem., 205, 303 (1953). 19. W. Shaw and S. Gurin. Arch. Biochem.. and Biophys., 47, 220 (1953). 20. S. S. Chernick and I. L. Chaikoff, J. Biol. Chem., 188, 389 (1951). 21. D. Stetten, Jr., and G. E. Boxer, /. Biol. Chem., 156, 271 (1944) ; D. Stetten, Jr., and B. V. Klein, ./. Biol. Chem., 159, 593 (1946) ; 162, 377 (1946) ; R. O. Brady and S. Gurin, J. Biol. Chem., 187, 589 (1950). 22. J. L. Rabinowitz and S. Gurin, ./. Biol. Chem., 208, 307 (1954) ; H. Rudney, J. Am. Chem. Soc, 76, 2595 (1954). 23. F. Dituri, F. Cobey, J. V. Warms, and S. Gurin, Federation Proc, 14, 203 (1955). 24. R. O. Brady and S. Gurin, J. Biol. Chem., 1S9, 371 (1951) ; G. L. Curran, J. Biol. Chem., 191, 775 (1951); M. Blecher and S. Gurin, ./. Biol Chem., 209, 953 (1954). Tetrazoles as Carboxylic Acid Analogs ROBERT M. HERBST The acidic character of the tetrazole ring system in which the ring nitrogens are unsubstituted has been recognized since the first prep- aration of the parent compound by Bladin in 1892. Although Bladin had suggested the name tetrazole for the ring system in 1886, the parent compound was for many years referred to as "tetrazotsaure" because of its acidic character. The acidic nature of several 5-substi- tuted tetrazoles was similarly recognized by nomenclature such as "benzenyl tetrazotsaure" (5-phenyltetrazole) and "amidotetrazot- saure" (5-aminotetrazole). Although it wras recognized with the first preparation of these compounds that 5-substituted tetrazoles behaved as acids, it is only recently that a systematic study of the factors influencing the acid strength of these compounds has been undertaken. An attempt will be made to develop the analogy between the factors influencing the acidity of 5-substituted tetrazoles and carboxylic acids in the following. Until recently, with few exceptions, only the 5-aryltetrazoles have been easily accessible. These were usually prepared by a sequence of reactions due to Pinner that involved the conversion of a nitrile successively into an iminoester, a hydrazidine, an imide azide, and finally a tetrazole. It was also known that certain nitriles such as cyanogen, cyanogen chloride, ethyl cyanoformate, and cyanamide could be converted into the corresponding 5-substituted tetrazoles by inter- action with hydrazoic acid (scheme I, R = H, CN, CI, COOC2H5, R— C=NH HN- — C— R RCN-^ I -> I (I) N3 N N \ / N 141 142 Essays in Biochemistry NH2). The development of this reaction was aborted by von Braun's observation that its application to alkyl and aryl cyanides in the presence of a large excess of concentrated sulfuric acid led to rearranged products such as 1-alkyl- or l-aryl-5-aminotetrazoles (scheme II). Although the conditions under which von Braun's experiments were done were entirely different from those applied by earlier workers, von Braun did not hesitate to conclude that hydrazoic acid would not add to the cyanide group of alkyl or aryl cyanides. HSO R— N- -C— NH2 RCN + 2HN3 -^-4 | || + N2 (II) N N \ / N R = alkyl or aryl Recent investigations have shown that von Braun's conclusions are not tenable. As a result the addition of hydrazoic acid to nitriles has become the most generally applicable procedure 2 5 for the synthesis of 5-substituted tetrazoles. The symbol R in scheme I may be re- defined to include in addition to the groups mentioned alkyl and aryl, alkyl- or arylamino, and dialkyl- or diarylamino groups. It must be pointed out, however, that in one instance, when R is a monoalkyl- or monoarylamino group as in the monoalkyl- or monoarylcyanamides, cyclization of the imide azide may take place in either of two ways and the course shown in scheme III is followed exclusively.4 It is particularly interesting to note that the nature of the group R in the monosubstituted cyanamides appears to have little effect upon the course of the cyclization illustrated in scheme III. RNH— CN hn, RNH-C=NH > 1 ^ N3 RN=C— NH2 N3 R— N- — C— NH2 N N \ / N (III) With the accumulation of a group of 5-substituted tetrazoles the determination of their apparent acidic dissociation constants by po- tentiometric titrations could be undertaken. It soon became apparent that the 5-tetrazolyl group ( — CN4H) was quite comparable to the carboxyl group ( — C02H) in bestowing acidic character upon com- Tetrazoles as Carboxylic Acid Analogs 143 pounds built up around it. In a series of 5-alkyltetrazoles the apparent acidic dissociation constants were variously from one-tenth to one-half as large as those of the correspondingly substituted carboxylic acids. A few examples are cited in Table 1. It will be noted that the dissocia- Tahle 1. Apparent Dissociation Constants of 5-AlkyltetrazoIes and Aliphatic Carbovylic Acids R H CH3 C2H5 C3H7-" CsHv-iso C4H9-n CiHg-iso C5H11-W R~CN4H * K X 10fi at 25' 16.2 2.74 (2.43) t 2.5(3 2.47 2.80 (2.38) % (2.45) t (2.22) t R— C02H | A' X 10s at 25' 171.2 17.5 13.3 15 13.8 13.8 16.7 13.2 * Apparent dissociation constants were determined by potentiometric titra- tion in aqueous solution at 25°C. unless otherwise indicated.2 f Constants obtained from conductivity data in aqueous solution.6 J Apparent dissociation constants determined by potentiometric titration in 25% (by weight) methanol at 25°C.2 tion constants are influenced in roughly the same manner by the nature of the alkyl substituent in both series. The acidity of tetrazole itself is readily explained on the basis of the resonance concept, i.e., the resonance stabilization of the tetrazolyl anion formed upon dissociation of the proton attached to the ring nitrogens (scheme IV). Similarly, the acidity of the carboxylic acids HN- I N \ -CH II N H+ + 'N (IV) has been attributed to resonance stabilization of the carboxylate ion (scheme V). 144 Essays in Biochemistry o—i / (V) 0 0 0 / / / 1 _ ± H+ + HC <-> HC \ \ \ OH o- 0 From schemes IV and V it might be anticipated that resonance sta- bilization would be greater in the tetrazolyl anion than in the car- boxylate ion owing to the greater number of forms contributing to the resonance hybrid in the former instance. Although such a situation might suggest that tetrazole and the 5-substituted tetrazoles should be stronger acids than formic acid and the correspondingly substituted carboxylic acids, experimental observations do not support this sup- position. It is more likely that other factors, particularly the greater electronegativity of oxygen as compared with nitrogen, are the decisive factors in determining the relative strength of the tetrazoles as com- pared with the carboxylic acids of the aliphatic series. In general, factors influencing the electronegativity of the tetrazole system or the carboxyl group should have similar effects upon acid strengths in both series. Thus replacement of the hydrogen attached to the carboxyl carbon of formic acid by an alkyl group is assumed to cause a decrease in the electronegativity (increase in proton affinity) of the carboxyl group through the operation of an inductive effect with a resultant decrease in the apparent acidic dissociation of the carboxyl group. A similar effect is noted in the 5-alkyltetrazoles where replace- ment of the hydrogen in the 5-position on the tetrazole ring by an alkyl group causes a decrease in apparent acidic dissociation roughly com- parable to that noted in the carboxylic acid series. When aromatic groups are introduced as substituents in the 5- position of the tetrazole ring a different situation develops. Although the apparent dissociation constant of tetrazole is only about one-tenth as large as that of formic acid, the apparent dissociation constant of 5-phenyltetrazole is more than three times as large as that of benzoic acid. A similar relationship exists between a group of substituted 5-phenyltetrazoles and the correspondingly substituted benzoic acids (Table 2). Since the intrinsic inductive effect of the aryl group upon the 5-tetrazolyl and the carboxyl groups may be assumed to be of about the same order of magnitude, some other factor must be operating to cause the tetrazolyl group to exhibit a greater electronegativity than the carboxyl group in these instances. It will be noted that the phenyl group may participate in the resonance of 5-phenyltetrazole thus in- Tetrazoles as Carboxylic Acid Analogs 145 Table 2. Apparent Dissociation Constants of 5-Aryltetrazoles and Substituted Benzoic Acids * R— CN4H R— C02H R A' X 106 K X 106 CeHs 29 (13) f 8.0 o-CH3C6H4 15.2 9.3 ///-CH3C6H4 20 4.3 P-CH3C6H4 15.2 3.5 o-C1C6H4 57 (25) f 70.8 t w-ClC6H4 87 14.5 J P-C1C6H4 (32) f 10. Ot o-BrC6H4 60 70.8 t ??i-Bi'CeH4 92 (42) f 13.5 % p-BrC6H4 (30) f 9.3 J o-CH3OC6H4 1.2 6.5 p-CH3OC6H4 14 2.8 * Apparent dissociation constants determined by potentiometric titration at 25°C. in 50% (by volume) methanol unless otherwise indicated.6 t Apparent dissociation constants determined by potentiometric titration in 75% (by volume) methanol at 25°C.5 X Dissociation constants determined conductimetrieally in 50% methanol (by volume) at 18-20°C.7 creasing the number of forms contributing to the hybrid beyond those due only to the resonance of the tetrazolyl portion of the anion (scheme VI). It should also be emphasized that, in addition to forms involving HN- I N ^n> I \ / ^ H+ + N -rO N N \ / N N N \ S N N N- \ / N N C N N N r "tOH N N % / N (VI) a charge separation, a number of forms in which the benzene ring accepts a charge and effectively withdraws electrons from the tetrazole ring may be wrritten. Although a series of contributing forms may be 146 Essays in Biochemistry written for the benzoate ion (scheme VII), it will be noted that the phenyl group can participate in the resonance picture only when a charge separation is invoked. f\.. o r / — \ ° / — \ °~ CK--CK - C \ OH ^ H+ + (W "*0= ortho > para. Furthermore, in the aryltetrazole series the ortho and para substituted compounds are of approximately the same strength. One exception must be noted: 5-p-methoxyphenyltetrazole is about ten times as strong an acid as the ortho isomer, a reversal of the situation existing with the methoxy- benzoic acids. An explanation of this observation will be advanced later. Because of their insolubility in 50% methanol it was necessary to determine acidic dissociation constants of 5-p-chlorophenyl- and 5-p-bromophenyltetrazole in 75% methanol. However, dissociation constants determined in both solvent mixtures in several other instances indicated that the values in 50% methanol were about twice as large as those in 75% methanol. Consequently, it seems reasonable to double the values of the apparent dissociation constants measure in Tetrazoles as Carboxylic Acid Analogs 147 75% methanol when a rough comparison is made with the values for other compounds determined in 50^ methanol. The data in Table 2 indicate that introduction of the electronegative chlorine or bromine atoms as substituents on the benzene ring of 5-phenyltetrazole increases the apparent dissociation constant as com- pared with the parent compound, whereas introduction of the electro- positive methyl group decreases the apparent dissociation constant. This might be anticipated from the inductive effects associated with these substituent groups. However, from this point of view alone the decrease in apparent dissociation constant observed for the methoxyl- substituted compounds is not compatible with the slightly electronega- tive character of this group. Ordinarily the position of a substituent group should influence the extent to which its inductive effect is transmitted by the benzene ring; the greatest effect upon apparent dissociation constant might be expected when the substituent is nearest the point of attachment of the carboxyl group. On this basis electro- negative substituents should cause the apparent dissociation constants to decrease in the order ortho > meta > para; the opposite order might be anticipated for electropositive substituents. Since the observed order of dissociation constants for the 5-aryltetrazoles is meta > ortho > para, it is likely that factors other than inductive or field effects are also influencing the magnitude of the dissociation constants. A combination of the resonance concept and the inductive or field effects of substituent groups will serve to explain the greater strength as acids of the meta isomers as compared with the ortho and para isomers. If we consider the 5-tolyltetrazoles first, the methyl group may be assumed to exert its normal inductive effect which should cause the 5-tolyltetrazoles to be weaker acids than 5-phenyltetrazole. Furthermore, the electropositive methyl group would favor the forma- tion of a field of relatively high electron density around the carbon atom to which it is attached (scheme VIII). Such a field of high electron density in the ortho or para positions would oppose the devel- opment of a formal negative charge at these points and, in effect, reduce the contribution to the resonance hybrid of forms involving negative charges at these points. With the methyl group in the meta position, although the inductive effect is in the same direction, an increase in electron density at this point would offer less opposition to resonance phenomena involving formal ortho or para negative charges. Such an explanation is in accord with the observation that both o-tolyl- and p-tolyltetrazole are weaker acids than ///-tolyltetra- zole. Since resonance effects involving the benzene ring are not com- 148 Essays in Biochemistry CH3 I HN- — C— N N \ / N r\ (VIII) HN— — C- I II N N \ / N pletely eliminated, the 5-tolyltetrazoles are still appreciably stronger acids than the corresponding toluic acids. In the halogen-substituted 5-aryltetrazoles the inductive effects as- sociated with the electronegative halogen atoms favor electron displace- ments that should cause these compounds to be stronger acids than the parent 5-phenyltetrazole. The greater acid strength of the meta isomer as compared with the other two may be explained if we again assume that resonance phenomena are more pronounced with the meta isomer than with the ortho or para isomers. Possibly the development of a formal negative charge at the ortho and para carbon atoms is opposed by the high concentration of electrons in the outer shell of the halogen atoms attached at these positions. This would result in a decrease in apparent dissociation constant of the ortho and para isomers due to the lesser contribution of such forms to the resonance hybrid (scheme IX, X = CI or Br) . :X: i -N C-f~\ I \\\=J<-f N N /X N N N N Tetrazoles as Carboxylic Acid Analogs 149 Although the o-halobenzoic acids are many times stronger than the p-halobenzoic acids, the 5-o-halophenyl- and 5-p-halophenyltetrazoles are of about the same strength as acids. This probably represents a marked reduction of the strength of the orf/io-substituted compound rather than a large increase in the strength of the para isomer. Con- ceivably the relatively low dissociation constants of the 5-o-halo- phenyltetrazoles are manifestations of the bulk effect of the ortho substituents. The tetrazole ring structure may be sufficiently rigid to cause a moderately bulky group in the ortho position of the 5-phenyl substituent to interfere with the formation of structures in which the two rings are coplanar. Resonance forms involving both the tetrazole and the benzene rings simultaneously require coplanarity of the rings. The small dissociation constant of 5-o-methoxyphenyltetrazole as compared with the para isomer may be due to hydrogen bonding between the methoxyl oxygen and the acidic hydrogen of the tetrazole ring (scheme X). This effect is much more pronounced in the 5-aryl- N— N N— N (X) \ H 0 \ CH3 tetrazole series than in the benzoic acid series. In the undissociated forms of the 5-substituted tetrazoles the acidic hydrogen is held in the plane of the ring at a rather fixed angle in position 1 or 2 owing to the rigidity of the tetrazole ring structure. When the acidic hydro- gen is located at position 1, the rigidity of the ring would force the hydrogen into a position favorable for hydrogen bonding with a sub- stituent in the ortho position of a 5-phenyl group. Such hydrogen bonding would serve to increase the stability of the undissociated molecule and thus decrease the apparent dissociation constant of the compound. The effect of an amino group or a substituted amino group as sub- stituents in the 5-position of the tetrazole ring is particularly interest- ing. Such compounds are analogous to the carbamic acids in the carboxylic acid series. The instability of the latter precludes com- parisons. The substitution of an amino group in the 5-position of the tetrazole nucleus causes a marked decrease in acid strength (Table 3). Furthermore, it may be noted that the basic function of the amino 150 Essays in Biochemistry Table 3. Apparent Acidic Dissociation Constants of Some 5-amino- tetrazole Derivatives in 50% Methanol * R2N— CN4H Ka X 108 Ka X 108 NR2 25°C. NR2 25°C. Amino 36 (120) f Diallylamino 33 Dimethylamino 38 (120) f Methylamino 21 (87) f Diethylamino 11 (47) f Ethylamino 22 (76) f Diisopropylamino 6 Benzylamino 30 Di-n-butylamino 10 Phenylamino 320 Diisobutylamino 7 o-Nitrophenylamino 8300 Di-w-amylamino 8 m-Nitrophenylamino 1400 Diisoamylamino 7 p-Nitrophenylamino 4600 Dibenzylamino 36 Acetylamino (2950) t Benzylmethylamino 38 Nitramino t Benzylethylamino 25 * All values taken from refs. 3, 10, 11, 12, 13. f Determined potentiometrically in water at 25°C. $ The first dissociation is as a strong organic acid, and K\ is approximately 1 X 10_1; in the second dissociation, K* is 9 X 10_7.u Both in aqueous solu- tion at 25°C. group is also very weak. Both of these observations may be attributed to the type of resonance invoked for aniline.8 The free pair of elec- trons of the amino nitrogen may be involved in the resonance of the attached ring system. In such a resonating system, the electronega- tivity of the ring would be decreased and dissociation of a proton from the ring would become more difficult causing a decrease in acidity. Since charge separation imposes ammonium ion character on the amino group, the basicity of this group should also be decreased. Several forms of this type that may contribute to the resonance hybrid are illustrated in scheme XL H H \ / N N N= / C N— II =N H H N + H H \ / N+ C C / \ / \ N N— H N N— H I I - I! I N= =N N- — N~ (XI) The effect upon the acidity of replacing the hydrogens of the amino group with alkyl groups appears to be primarily steric in nature. The Tetrazoles as Carboxylic Acid Analogs 151 inductive effect noted when the alkyl groups were attached directly at the 5-position is not transmitted by the amino nitrogen. 5-Amino- and 5-dimethylaminotetrazole have the same apparent acidic dissocia- tion constants. Neither steric nor inductive effects are apparent in this instance. Other groups which may exert little steric influence likewise have essentially no effect on the magnitude of the dissociation constant. Thus, 5-dibenzylamino-, 5-benzylmethylamino-, and 5-dial- lylaminotetrazole are as strongly acidic as 5-aminotetrazole and 5- (liniethylaminotetrazole (Table 3). On the other hand, there is a marked decrease in acid strength on going from 5-dimethylaminotetra- zole to 5-diethylamino-, 5-diisopropylamino-, 5-di-n-butylaminotetra- zole, and other larger dialkylaminotetrazoles. Since the inductive effects of most alkyl groups are comparable, their steric influence apparently predominates. A similar decrease in acid strength accom- panies the change from 5-benzylmethylamino- to 5-benzylethylamino- tetrazole. There is a nice correspondence in the occurrence of a steric effect in this group with that observed in certain hindered carboxylic acids and described by Newman in terms of the "six-number." 9 Those compounds having the largest six-number are the weakest acids (Table 4). Table 4. Steric Factors and Apparent Dissociation Constants of Some 5-Dialkylaminotetrazoles H / R2N— C— N \ X N— N 1! Ka X lu8 Six-Number CH3 38 0 C2II6 1 1 6 C3H7-iso 6 12 C4H9-n 10 6 Apparently other factors play a part in determining the acid strength of the 5-monoalkylaminotetrazoles. 5-Methylaminotetrazole is a weaker acid than the corresponding dimethylamino compound. A similar relationship exists between 5-benzylamino- and 5-dibenzyl- aminotetrazole. On the other hand in 5-ethylaminotetrazole, where the steric effect of a single ethyl group should be appreciably less than 152 Essays in Biochemistry that of two ethyl groups as in 5-diethylaminotetrazole, the anticipated increase in acid strength is realized. As has been pointed out, the enhanced acidity of 5-phenyltetrazole arises from increased stabilization of the anion by virtue of the con- jugation of the tetrazole nucleus and the phenyl group. When the two groups are separated by an amino group, the resulting 5-phenyl- aminotetrazole is a weaker acid than 5-phenyltetrazole because the conjugation of the phenyl group and the tetrazole nucleus is interrupted by the amino group. (It is interesting to note that separation of the phenyl group and the tetrazole nucleus by a methylene group produces very nearly the same quantitative effect.) The fact that 5-phenyl- aminotetrazole is still more acidic than 5-aminotetrazole and its alkyl derivatives leads one to speculate that resonance forms such as those illustrated in scheme XII may be responsible. Such forms would be H H H C— N= + N— N / \ N— N H \ I C— N= N— N' / + (XII) expected to increase the electronegativity of the tetrazole nucleus in- ductively and result in a decrease in proton affinity. This speculation is given some weight by the relative acidities of the isomeric 5-nitro- phenylaminotetrazoles. The inductive effect just cited should be aug- mented in decreasing order by the ortho, meta, and para nitro groups. Superimposed on this effect is a resonance reinforcement in the case of the ortho and para isomers. This combination of effects should cause the o-nitrophenylaminotetrazole to be the strongest acid, the para intermediate in strength, and the meta the weakest. All of them should be distinctly stronger acids than 5-phenylaminotetrazole. This is also the observed order. In scheme XIII several possible contributing forms for the resonance hybrid of 5-p-nitrophenylaminotetrazole are illustrated. H H V[— N II 0" > \ 1 ryy C— N= <— > / + W \ N— N O N— N /- H C— N= o =N 0" (XIII) Tetrazoles as Carboxylic Acid Analogs 153 Acylation of the amino group causes some striking changes in the acidity of 5-aminotetrazole. Acetylation brings about a 25-fold in- crease in acid strength; 10 the resulting compound is almost twice as strong an acid as the unsubstituted tetrazole. Nitration, which in a sense may also be considered as acylation, causes at least a 10,000-fold increase in the acidity of 5-aminotetrazole.11 5-Acetylaminotetrazole still behaves as a monobasic acid in aqueous media, but 5-nitramino- tetrazole is a dibasic acid. The presence of the moderately electro- negative acetyl group might be expected to favor tautomerism of the lactam-lactim type or resonance forms involving the amino nitrogen and the carbonyl group (scheme XIV). Resonance of the aniline type involving the tetrazole ring and the amino group is likely to be repressed. Of the several resonance and tautomeric forms shown in scheme XIV those on the left would presumably make the smallest contribution to the hybrid. The forms on the right would cause the amino nitrogen to exert a strong inductive effect in such a direction as to increase the electronegativity and decrease the proton affinity of the ring; a stronger acid should result. OH I HN C=N— COCH., HN C— NHCOCH3 HN C— N=C— CH3 II ^ I II - I || N NH N N N N \ / \ / % / N N N / \ / \ (XIV) H H 0_ I I I HN C=N— COCH3 HN==C-N=C-CH3 I I + <— ■» I II + N N_ N N \ / \ / N N A large variety of tetrazole derivatives have been prepared for investigation of their pharmacological properties. Most of these com- pounds have carried a nitrogen substituent in addition to the carbon substituent. The group is of interest because pharmacological activity has been associated with the large majority of compounds studied. They have almost uniformly exhibited effects upon the central nervous system that varied from purely depressive to highly stimulatory types. In some instances certain centers of the central nervous system have been rather selectively affected. In addition new types of structures having highly bactericidal and fungicidal action have been encountered. The close analogy in acidic properties between the 5-tetrazolyl group and the carboxyl group suggests that analogies might be found in the physiological actions of compounds in which the carboxyl group is 154 Essays in Biochemistry replaced by the 5-tetrazolyl group. This thought suggests that a series of antimetabolites in which the 5-tetrazolyl group replaces the carboxyl group as an acidic function is conceivable. For instance, amino acid analogs in which the tetrazolyl group replaces the carboxyl group might serve as effective amino acid antimetabolites. Such an- tagonists could serve a useful purpose in the study of the effects of interference with the utilization of various amino acids normally occur- ring in plant and animal proteins. It is also conceivable that malfunc- tions in the utilization of specific amino acids produced by such anti- metabolites could lead to a better understanding of protein and amino acid metabolism. Furthermore, such antimetabolites could be useful in controlling pathological states associated with hyperactivity of amino acid and protein metabolism. At this time only one such com- pound is known, the 5-/?-aminoethyltetrazole which was prepared as a histamine analog by Ainsworth.14 It failed to exhibit either hista- minelike properties or histamine antagonism, observations that are not surprising since the compound should more properly be considered as a ^-alanine analog, an analogy which was not recognized at the time. Beyond this only simple aliphatic and aromatic carboxylic acid analogs have been studied. Although the 5-alkyltetrazoles have shown some interesting pharmacological properties as rather potent anticon- vulsant agents as measured by their antagonism towards Metrazole, no compounds have been prepared with long-chain alkyl groups which would make them comparable to fatty acids such as palmitic and stearic. Such long-chain compounds could be of interest in their rela- tionship to fatty acid metabolism in the living organism. Although the biological implications associated with the 5-substituted tetrazoles are purely speculative at this time, the preparation and study of such compounds in the future could lead to interesting organic and bio- chemical results. This discussion is based upon experimental work done by Drs. Joseph S. Mihina, William L. Garbrecht, and James A. Garrison, and Mr. Kenneth R. Wilson. The literature concerning tetrazoles has been reviewed by Benson 1 to whom reference should be made for the earlier work. References 1. F. R. Benson, Chem. Revs., 41, 1 (1947). 2. J. S. Mihina and R. M. Herbst, J. Org. Chem., 15, 1082 (1950). 3. W. L. Garbrecht and R. M. Herbst, J. Org. Chem.., IS, 1003 (1953). 4. W. L. Garbrecht and R. M. Herbst, J. Org. Chem., IS, 1014 (1953). Tetrazoles as Carboxylic Acid Analogs 155 5. K. R. Wilson, The Apparent Acidic Dissociation Constants of Some 5- Aryltetrazolcs, MS thesis, Michigan State College, 1955. 6. J. F. J. Dippy, Chem. Revs., 25, 151 (1939). 7. R. Kuhn and A. Wassermann, Helv. Chim. Acta, 11, 3, 31 (1928). 8. G. W. Wheland, The Theory oj Resonance and Its Application to Organic Chemistry, John Wiley & Sons, New York, 1944, p. 177. 9. M. S. Newman, /. Am. Chem. Soc, 72, 4783 (1950). 10. R. M. Herbst and W. L. Garbrecht, J. Org. Chem., IS, 1283 (1953). 11. R. M. Herbst and J. A. Garrison, J. Org. Chem., 18, 941 (1953). 12. W. L. Garbrecht and R. M. Herbst, /. Org. Chem., 18, 1022 (1953). 13. W. L. Garbrecht and R. M. Herbst, /. Org. Chem., IS, 1269 (1953). 14. C. Ainsworth, J. Am. Chem. Soc, 75, 5728 (1953). The Structural Basis for the Differentiation of Identical Groups in Asymmetric Reactions HANS H1RSCHMANN Classical organic chemistry defined asymmetric syntheses x as proc- esses which convert symmetrical molecules into optically active ones by the intermediate use of asymmetric agents, provided the methods employed take no recourse to processes of resolution. Numerous ex- amples have demonstrated the reality of the phenomenon. The asym- metric agent used to bring about this change can either be an "asym- metric form" of energy such as circularly polarized light or an asym- metric form of matter such as an asymmetric molecule. The success of the operation occasioned little inquiry into its structural require- ments since the essential features appeared to be trivial ones, i.e., the symmetry of the starting compound and the asymmetry of the product. Quite a different situation arose, however, with the discovery of closely related phenomena which could not be detected by testing the product for optical activity. If one of the substituents a in compound I is converted preferentially to a substituent d which is different from the three others, the product II is optically active. On the other hand b b b i i i i i i i i i ' i i i i i I I i c c c II I Til if the reaction involves the selective conversion of one group a into a group b identical with one already present, the product (III) has a plane of symmetry and hence is optically inactive. However, if one of the groups to be substituted is isotopically labeled, it is possible to demonstrate the asymmetry of the process in either case. The difference between these two types of sterically selective reac- 156 Structural Differentiation in Asymmetric Reactions 157 tions, therefore, appears not to be an intrinsic one but rather a reflec- tion of the diagnostic tool employed for demonstrating the asymmetry of the process. It can be shown that the selective reactivity of one of a set of identical substituents is possible in certain compounds, but not in others. It, therefore, becomes a problem of both theoretical and practical interest to determine the structural basis of such differ- entiation. Several criteria 2~4 have been proposed in answer to this question. However, one of these rules 3 is unreliable,4 and they all fail in providing information about situations that are likely to be encountered. It is the purpose of this discussion to propose a general criterion and to compare it with those which have been advanced by others. The possibility that two identical substituents a in a molecule Caabc can be distinguished from each other in a process catalyzed by an enzyme was postulated by Ogston.5 The reality of the phenomenon has been amply demonstrated in the case of citric acid and other sub- stances. As is well known, the mechanism envisaged by Ogston as- sumes an attachment of three groups of the substrate to three sites of the enzyme, of which only one is catalytically active. If the sites a' and a" can combine specifically with the groups * a' or a", and the sites /S and y with the groups b and c, respectively, if the reaction proceeds only if three groups are simultaneously engaged, and if only a' but not a" can catalyze the reaction, then an enzyme constituted as TVa or V will cause a reaction at a' but not at a". These stipula- * Throughout this discussion atoms or groups of atoms designated by a', a", a'" are defined as being identical and are designated by different symbols merely to facilitate discussion. Furthermore, unless stated otherwise, all substituents a, b, c, etc., are assumed to be symmetric. Bonds drawn in solid triangles are directed towards the observer; those in broken lines towards the rear. 158 Essays in Biochemistry tions have been made more stringent by Wilcox et al.,6 who pointed out that the Ogston scheme is strictly correct only if a fourth condition such as steric hindrance prevents the approach of the substrate from the opposite side, since an interaction as in TVb would place the group a" at the reactive site a'. Since the Ogston scheme provides a very satisfactory explanation for the complete asymmetry of most enzymatic processes, it apparently has been assumed by some investigators that a three-point attachment between the symmetric substrate and the asymmetric agent is essential for any differentiation of identical groups even if the selectivity is only partial. It is very doubtful that Ogston entertained such thoughts, and a clear statement to the contrary was made by Wilcox.2 A most emphatic denial can be found in a very lucid analysis by Schwartz and Carter,4 which may be summarized as follows: Although the sub- stituents a in Caabc are identical and although they are located in a symmetrical molecule, their locations within that molecule nevertheless are sterically non-equivalent. If one views from one substituent a the three other substituents of the central carbon atom, the groups a, b, c appear in a clockwise sequence; whereas a counterclockwise order prevails if one views from the other a group. If one orients the mole- cule in a predetermined way (e.g., as in I with groups b and c to the rear and with b above c), invariably the same group a will be directed to the right or to the left, respectively. Since the two a groups can be distinguished by inspection, their location cannot be sterically equivalent. A plane through the central carbon atom which bisects substituents b and c cuts the molecule into halves which are related to each other as object to mirror image but which are not superimposable as long as b and c are different. Since this situation resembles in some respects that prevailing in a meso compound which often can be cut into mirror- image halves that are not superimposable, Schwartz and Carter pro- posed the term meso carbon atom for one substituted with two identical and two different substituents (Caabc). It should be noted that the last argument for the steric non-equivalence of the two a groups must be used with caution, since such a molecule as Caaab in which all a groups are sterically equivalent also can be cut into mirror-image halves that cannot be superimposed upon each other. Another failure of this criterion was noted by the authors themselves.4 If we consider the approach of a molecule E to one or the other substituent a in Caabc leading to a one-point attachment (either by covalence, electrovalence, hydrogen bonding, or some other force) the Structural Differentiation in Asymmetric Reactions 159 result to be expected will depend on the symmetry or asymmetry of E. If E is symmetrical, the products Vila and VII6 are optical antipodes \ VI i b I I l ■ CM I I I c VI b E-a'- Vlla b I I I ■ » I I I c VII b a"-E and hence possess equal stability. Similarly, any approach of E to a' can be matched by a mirror-image approach to a" possessing equal probability. Hence no differentiation of a' and a" is possible. If E is asymmetric, the products Vila and VIIfc> are no longer optical antip- odes but diastereoisomers and hence would be expected to possess different stabilities. Even if the direction of motion of E towards a' is the mirror image of that of E towards a", the corresponding situa- tions during any stage of such an approach (e.g., Via and VIb) are related like diastereoisomers and not like mirror images and hence can be expected to possess different probabilities. The final proportion of Vila and of Vllb may be determined by the probability of a success- \ o— © Villa •V / o — © VHIb ful collision between E and a' or a" (kinetic control) , or by the relative stabilities (or solubilities) of the products (thermodynamic control) if the process is reversible. In either case the ratio of reaction products would be expected to be different from unity. If the reaction does not 160 Essays in Biochemistry involve some form of complexing between the a groups and the asym- metric E, differentiation of the a groups can still be expected. If the process consists in an attack of E on the central carbon atom which replaces either a' or a", the transition states 7 (Villa or VIII6, respectively) are not identical but related like diastereoisomers. Hence the heights of the energy barriers separating the starting com- pounds from the products as well as the energy contents of the latter would be expected to differ for the substitution of a' and of a". If the reaction of the a groups is initiated by some combination of E with the b group as in IXa, the steric relationship of a' to E is not identical with that of a" to E nor will it become identical by any rotation around a single bond as in 1X6. It is clear then that a sJX *%^' b b ! ! I I l I c c IXa IX b variety of reasonable mechanisms can be conceived which do not require a three-point attachment and which nevertheless permit the differentiation of identical substituents. Although this is to be expected if the optically active species E participates directly * in the steps which alter the a group in Caabc, it must be noted that this differentia- tion often is too small to be detected or too transitory to be preserved. Since the common feature of all these reactions is the diastereomeric character of some intermediate or product, it seems justified to neglect these mechanistic details and to seek the basis for the differentiation of identical groups in some structural characteristic of the substrate. To define it, the following rules have been proposed: 1. Wilcox: 2 "In a molecule which has a plane of symmetry or a point of symmetry, if one of the atoms which does not lie in any plane * This stipulation seems quite essential. If, for instance, the reaction can be represented by an ideal unimolecular substitution 7 in the a group by E, the rate- determining dissociation proceeds in the virtual absence of E and hence should proceed with equal facility in the a' and a" group. Similarly, if E attacks b changing it to the symmetric substituent d which subsequently in the absence of E causes a secondary change affecting the a groups, no differentiation is to be expected. Structural Differentiation in Asymmetric Reactions 161 or point of symmetry is replaced by an isotopic atom, the molecule becomes asymmetric with respect to the labelled atom. In any reaction with an asymmetric reagent, this labelled atom (or group) may react at a rate which is different from that of its counterpart through the plane or point of symmetry, and the difference in rates will be expressed in the distribution of the isotope in the products. This asymmetric behaviour would be superimposed on any difference in the rates of reaction which would result from the different masses of the isotopic atoms." 2. Racusen and Aronoff: 3 "Discrimination of identical groups or atoms by an asymmetric agent (enzyme, optical antipode, etc.) is possible only in molecules which do not possess a twofold (or greater) axis of symmetry." * 3. Schwartz and Carter: 4 "In any molecule containing one (or more) meso-carbon atoms, reaction of the two (a) groups with an asymmetric reagent will proceed at different rates, yielding unequal amounts of diastereoisomeric products." 4. We should like to propose the following criterion: A three-dimen- sional representation of a molecule containing two (or more) identical groups or atoms a designated as a' and a" is moved so that the position of a" will coincide with the original position of a'. If this can be done * Two kinds of axes of symmetry are being distinguished. An object is said to possess an n-fold simple axis of symmetry if a rotation around this axis through an angle of 360°/n yields an arrangement indistinguishable from the original. An object is said to possess an w-fold alternating axis of symmetry if rotation around this axis through an angle of 360°/n followed by a reflection in a plane perpendicular to this axis produces an arrangement indistinguishable from the original. It can readily be seen that a onefold alternating axis is equivalent to a plane of symmetry and that a twofold alternating axis is equivalent to a center (point) of symmetry. A compound which possesses no alternating axis cannot be superimposed on its mirror image and is termed asymmetric in the usual language of organic chemistry. However, such a compound may still possess a simple axis greater than one, and if it does it is often designated as dyssymmetric. A compound which has only onefold simple axes is not considered to possess symmetry of any kind and is termed asymmetric, since every object, no matter how irregular, has an infinite number of such axes. Numerous synonyms are in use. An alternating axis is also called a "rotation-reflection axis," "mirror axis," "improper axis," "axis of the second order," etc., whereas a simple axis has also been called a "rotation axis," "axis of the first order," or merely an "axis of symmetry." The last term, however, unless defined is ambiguous.8 A few ex- amples of axes of both kinds and a brief discussion of their significance will be given below, but for further information reference is made to the classical treatise by Schoenflies 9 and to accounts given by others.10-12 Reference 12 discusses the subject without the aid of mathematics. 1G2 Essays in Biochemistry in such a way that the second arrangement is indistinguishable from the first, the groups a' and a" cannot be differentiated from each other in any reaction. However, if such superposition is impossible, the groups a' and a" can react with an asymmetric reagent at different rates.* The validity of this rule can be demonstrated as follows: If the representation of the molecule A is indistinguishable from one showing a" in place of a', any product resulting from the change of a' to d must also be superimposable on the product resulting from the change of a" to d. This will be true regardless of the symmetry or asymmetry of d. The products must be superimposable even if the reaction is not confined to the a groups but involves additional changes at other substituents. Similarly, any approach of the reagent E (which again may be symmetric or asymmetric) towards a' or some other atom is indistinguishable from the corresponding approach of E towards a" or the corresponding other atom. Furthermore, if the products of the direct reaction at (or near) a' and a" are indistinguishable from each other, there can also be no differentiation in any successive process. If the reaction should involve several products, the same argument applies to each one of them. We conclude then that the a' and a" groups cannot be differentiated fijom each other in any process if the superposition specified is possible. If, on the other hand, such super- position is not possible, mechanisms exist which permit the differentia- tion of the a groups. For instance, the product resulting from the combination of a' with the optically active agent E cannot be super- imposed on that resulting from the combination with a''. These prod- ucts cannot be antipodes if E, as is ordinarily the case, retains its asymmetry during the reaction. Hence the thermodynamic properties of the two products are expected to differ. If the outcome of the reaction of E with A depends on other factors, analogous arguments apply as were outlined above for the specific case of Caabc. Since a possibility of differentiation exists whenever the a groups do not meet the superposition test, the validity of the rule is considered to be fully established. To illustrate its application and utility a few specific examples will be discussed. Example 1, Caabb (X). Rotation of Xa through 180° around an axis passing through the central carbon atom C and perpendicular to the plane of the paper results in X6 which is indistinguishable from Xa. Hence the two a groups cannot be differentiated from each other. The * The application of this rule to compounds which cannot be dealt with ade- quately by a single representation is illustrated in examples 8-10. Structural Differentiation in Asymmetric Reactions 1(33 same applies of course to the b groups. The molecule contains two planes of symmetry bisecting the a and b groups respectively, a twofold b' b" i i I i i i a'-^«C^^- a" a"-^«C^^- a' I I I I I I b" b' Xa Xb simple axis and no meso carbon atom. The result therefore is pre- dicted by rule 2 and not inconsistent with rules 1 and 3. Example 2, Caaab (XI). Successive rotations of XIa through 120° around the axis Cb convert it into XI6 and XIc which are indistin- c i\ 3.' 3/" A\ Clb XIc c a' a'" a" a" XIa guishable from the original and which contain either a" or a'" in place of a'. Hence the three a groups cannot be differentiated. All a groups lie in planes of symmetry, the molecule contains a threefold simple axis and no meso carbon atom. The result therefore is again predicted by rule 2 and is not inconsistent with rules 1 and 3. Example 3, Caabc. {XII). The two a groups can react at different rates with an asymmetric agent since it is impossible to superimpose simultaneously a" with a', b with b, and c with c (see, e.g., Xlla, XII6, b b c i 1 I i T i c c b Xlla Xllb XIIc and XIIc) . The molecule has one plane of symmetry but the two a groups do not lie in it; it possesses no axis greater than one and con- tains a meso carbon atom. Hence the result is predicted by rules 1 to 3. Example 4, Caa (-)-£>) ( — 6) (XIII). In this case the two b groups are assumed to be structurally identical asymmetric substituents that 164 Essays in Biochemistry are mirror images of each other and hence not superimposable. There- fore any attempt to place a" into the position of a' results in an arrangement different from the original (e.g., XHIa and XIII£>). The + b -b- e c f a'-^«C^^~ a' I I I I e C f a' T d XHIa -b +b f c e I I I I I I I f\ c e d Xlllb two a groups therefore can react at different rates with an asymmetric reagent. This result cannot be foreseen by applying rule 1 since both a groups lie in the plane of symmetry. It is predicted, however, by rule 2, since there is no axis greater than one. The definition given for the meso carbon atom did not refer specifically to such structures but could quite logically be interpreted to include them. If we do this we should realize, however, that we make a choice different from that usually made in defining an asymmetric carbon atom, since the struc- ture Ca( + b)(— b)c gives rise to two stereoisomers which are not optical antipodes. Example 5, Caa{ + b) ( + 6) (XIV). If the group +b is defined as in the preceding example, the compound is not superposable on its mirror image. In optically active compounds differential reactivity of identical substituents even in reaction with symmetrical reagents occurs with such frequency that this fact hardly requires comment. Example 5 shows, however, that this is not invariably the case and that even optically active compounds may possess substituents that cannot be differentiated from each other by any process. Rotation of structure XlVa through 180° around an axis which passes through the central carbon atom and is perpendicular to the plane of the paper yields XIV6. As both arrangements are indistinguishable, the two a groups cannot be differentiated from each other nor can the b groups. [This conclu- sion of course should not be construed to mean that the reactivity of the a groups in this compound, in reaction with an asymmetric reagent, Structural Differentiation in Asymmetric Reactions 165 would be the same as that of the a groups in the optical antipode ( !aa I — b) ( — b).] Previous discussions have not dealt with dyssym- e C— I I I I I I I I f c— XIV a > +b +b I I I I I I d XIV b metric structures. Rule 1 in particular is limited to symmetric com- pounds. It may be noted, however, that our finding is consistent with rule 2 since the compound, although non-superposable on its mirror image, possesses a twofold simple axis. As the central carbon is not a meso carbon atom the result is not inconsistent with rule 3. Example 6, Caab-Caab {XV) . Internal rotation around the central bond permits two symmetric arrangements (XVa and XVc) . If XVa is rotated 180° around an axis which is perpendicular to the plane w V/ c" C' a'" b a"" XVa a" b XVb a' b' a ' T N a"" b XVc \'V c' b' XVd 166 Essays in Biochemistry of the paper and which passes through the center of the C-C bond, XV6 results. Similarly, rotation of XVc around an axis which lies within the plane of the paper and which bisects and is perpendicular to the C-C bond yields XVd. The conclusions which can be drawn from these two operations are identical since both demonstrate the steric equivalence of C and C", of b' and b", of a' and a"", and of a" and a'", respectively. As no other motions produce indistinguishable arrangements we can conclude that the following pairs can be differ- entiated: a' from a", a' from a"', a" from a"", and a"' from a"". The steric inequalities could have been deduced also by considering example 6 a special case of example 3, which shows that a' and a" can be differentiated as well as a'" and a"". The remaining relation- ships between the a groups follow, then, from consideration of the steric equalities already established. The molecule arranged as in XVa possesses two planes of symmetry, as in XVc a plane and a center of symmetry. As the a groups lie outside these symmetry elements, the prevailing steric inequalities are predicted also by rule 1 which, however, gives no information about the groups which cannot be differentiated unless further symmetry considerations are applied. Structure XVa has a twofold simple axis, and XVc a twofold simple axis as well as twofold alternating axes. The results obtained, there- fore, are clearly inconsistent with rule 2. The compound contains two meso carbon atoms. Rule 3, therefore, predicts correctly two of the steric inequalities of the a groups but fails to disclose the two others. Moreover, it does not indicate the presence of identical substituents which cannot be differentiated. Example 7, Cabc-Cabc (XVI and XVII). Again, as in examples 4 and 5, the results to be expected depend on the configurations of the asymmetric centers. Application of our criterion to the meso compound XVI shows that all pairs of identical substituents as well as the two central carbon atoms can be differentiated, whereas this is impossible with the optically active forms since XVTIa on rotation a' be' a' b c' a" b" c* \!/ \i/ \i/ c' c c c" c" c b a" a" b c" a' b c' XVI XVIIa XVIIb Structural Differentiation in Asymmetric Reactions 167 gives XVII6, which is indistinguishable. The result with the meso compound is predicted by rule 1. However, as structure XVI has two- fold alternating axes, the differentiation of the substituents is an exception to rule 2 unless its authors intended to exclude alternating axes. The result is not predicted by rule 3 since the compound contains no meso carbon atoms but two enantiomorphic asymmetric carbon atoms. Our results present an interesting paradox which may be exemplified as follows. If tartaric acid is considered as an intermediate in an enzymatic process which results in differential labeling of the carboxyl groups, the symmetric meso compound can qualify as a pos- sible intermediate but the dyssymmetric optically active forms cannot. If one considers the types of structures which permit the differentia- tion of identical substituents and those which do not, one can think of innumerable reactions which link the two types in either direction. Since the conversion of a structure which does not permit differentiation by any process into one which admits this possibility might be con- sidered a contradiction, it is perhaps not superfluous to show that this is not the case. For example, the olefinic carbon atoms as well as their identical substituents in structure XVIIIa cannot be differentiated since the structure is indistinguishable from XVTII6 which results from it by rotation. However, if this substance undergoes cis addition of two c groups, a meso structure XIX results which as set forth in .a" a' I ,C'=C" c c' XVIIIa XVIII b ~v a' \ a" \ c XIX a XIX b example 7 permits the differentiation of the central carbon atoms and their substituents. Since arrangements XVIIIa and XVIIIb are in- distinguishable, there will be an equal chance that the molecule presents itself either way to an asymmetric reagent. Hence even a unidirec- tional vis addition (broken arrow) will produce the superposable struc- tures XlXtt and XIX6 in equal amounts. Therefore, even if a subse- quent reaction at the left of the two central carbon atoms in XlXa and XIXt> proceeds at a rate different from that at the right, this 168 Essays in Biochemistry cannot lead to a differentiation of the carbon atoms labeled C and C" as these have become randomly distributed over the two positions. Since even a single intermediate which does not permit the differ- entiation of identical substituents suffices to bring about this result in a long chain of reactions, it seems of particular importance to recog- nize the structural characteristics which prevent selective reactions of identical groups. To our knowledge this question has not been an- swered previously. Rules 1 and 3 describe structures which permit differentiation. Since we have shown that neither criterion covers all situations where this may occur, these rules clearly are not con- vertible and hence give no reliable information about structures which do not permit differentiation. Rule 2 tried to answer this question, but the criterion was found to have exceptions. One may conclude, therefore, that these three rules are no more than partial solutions of the problem. As one should expect to find the general principle that prevents differentiation of identical substituents in some structural regularity, we shall attempt to link rule 4 to molecular symmetry. Mathematical analysis has shown that two rigid objects, so related that the distance between any two points in one of them equals the distance of the corresponding points in the other, can be brought into coincidence by a combination of at most three operations, a trans- lational motion, a rotation around an axis, and a reflection in a plane perpendicular to this axis. If the two objects have one point in common, the rotation and the reflection will always suffice.9,10 A finite rigid object is said to possess symmetry if two or more indistinguishable arrangements exist that can be interconverted by these two types of operations.11 We therefore can distinguish two kinds of symmetry: If a rotation suffices to produce another indistinguishable arrangement, the object is congruent with itself in more than one way and the axis of rotation is termed a simple axis of symmetry. If the conversion to another indistinguishable arrangement is possible by a reflection or by a rotation and reflection, the object is congruent with its mirror image and the axis of rotation is termed an alternating axis of symmetry.* Organic chemistry has concerned itself almost exclusively with sym- metry of the second kind. It is quite clear, however, that this mirror- image symmetry has no bearing on our problem, since the operations considered in rule 4 are motions and not reflections and any super- position which cannot be achieved without a reflection is of no concern. * See footnote on p. 161. Structural Differentiation in Asymmetric Reactions 169 On the other hand, rotations around finite * simple axes through the angles specified by their multiplicity are motions which meet all criteria specified in rule 4. In fact, in the examples given, superposition of identical groups, if this was possible, was achieved by rotation around simple axes. This can be done for any rigid structure meeting this superposition test, since two identical objects which have one point in common n can always be brought into coincidence by means of a rotation.9-10 We have therefore two complementary situations. If a structure has an alternating axis of symmetry, it can be superposed on its mirror image and therefore cannot be resolved into optical antipodes. If a structure has a finite simple axis greater than one, it contains identical substituents which meet the superposition test specified in rule 4 and which therefore cannot be differentiated from each other in any reaction. It is clear, then, that the symmetry elements which prevent the differentiation of identical groups differ in kind from those preventing resolution into enantiomorphs, although both can be found in the same molecule. This dichotomy of symmetry elements is well illustrated by example 7 and fully resolves the paradox presented. * It is of interest to trace the reason for the failure of rule 2 in example 6, since this structure possessed a twofold simple axis. The compound contained four identical substituents, and rotation around this axis permitted the superposition of the members of two pairs of identical substituents but not the mutual superposition of all four. Hence, to exclude the possibility of differentiation of all identical groups in any structure, one cannot set a fixed limit for the multiplicity of simple axes but must relate in some manner the required number, multiplicity, and orientation of simple axes to the number and disposi- tion of identical groups. The so-called symmetry number, which equals the number of indistinguishable positions into which a molecule can be turned by simple rigid rotations,13 appears to be a useful character- istic for the general solution of this problem. This number has been related to the so-called symmetry group, which is determined by the combination of all symmetry elements which are found in a given structure. The symmetry number must equal, at least, the number of identical substituents to prevent their differentiation. Although the relationship between symmetry group and symmetry number is avail- able in tabular form, the operation of this criterion seems certainly * Infinite simple axes are found in linear molecules like hydrogen cyanide or acetylene and lie in the direction of their bonds. Rotation around such an axis does not achieve the superposition of different atoms. 170 Essays in Biochemistry no simpler than the application of rule 4. Moreover, the information so obtained is frequently less detailed. In example 6 we find a sym- metry number of two for either XVa or XVc, which is too low for the superposition of four identical groups. This result provides no in- formation about the disposition of the a groups which can or cannot be differentiated. Finally we shall find that the use of symmetry numbers encounters difficulties from yet another source. We have thus far assumed, quite unrealistically, that molecules can be represented by rigid bodies. The rotation around single bonds, in particular, has presented a problem also in classical stereochemistry, where it has been met by various devices, including the suggestions that the symmetry properties of a molecule be determined for a prop- erly selected conformation or for a time average of all conformations.12 The first of these devices has been used in this discussion thus far; the second can be applied to example 8. In examples 9 and 10, how- ever, it is no longer possible to derive the correct answer if we merely consider the symmetry properties of some rigid representation of the whole molecule. We shall still find it possible, however, to apply rule 4. ' Example 8, cyclohexane derivatives (XX). If one tests the chair form XXa for the steric equivalence of the two carbon atoms labeled a' and a", one obtains on rotation arrangement XXc which is obviously different. However, the original compound is accompanied by an equal amount of XX6 which yields XXd, when subjected to the XXa XXb XXe XXc XX d XX f corresponding motion. Since XXd can be superposed on XXa, and XXc on XXb, the actual compound which contains equal parts of XXa and XXb meets the superposition test even if the individual molecules do not. It follows that the two a groups or b groups, re- Structural Differentiation in Asymmetric Reactions 171 spectively, will react with equal rates under any condition which permits the equilibration of conformations. The chair forms, never- theless, possess no simple axis greater than one. If we consider, however, not the prevalent conformations but the averaged positions of their nuclei, we obtain the planar form XXe, which on rotation yields XX/. This representation, therefore, meets rule 4 and contains a twofold simple axis. Example 9, Caaa{-\-b) {XXI). This case differs from example 2 in having an asymmetric substituent -fb instead of the symmetric b. Such a molecule has no simple axis of symmetry >1. Unless rotation around the central C-C bond is severely restricted, every conformation (e.g., XXIa) is accompanied by two others (XXI6 and XXIe) which V + b •l\ s\\ s\\ a' a'" a" a" a' a"' a"' a" a' XXIa XXIb XXIc are equally stable and mutually superposable. The prevailing mixture, therefore, meets the superposition test jointly even if its individual members do not. The three a groups therefore cannot be differentiated from each other. Example 10, RCOOH (XXII). Tautomerism ordinarily refers to an equilibrium of non-equivalent structures which is of no concern in this discussion. If we consider, however, a reaction such as the prototropic shift of an acid, the question of the steric equivalence of the two oxygen atoms arises. A mixture of XXIIa and XXIIb will yield by XXII a R— C XXIIb R— C R-Cf „ XXIIc •0"H 0"H 0' 0 R-C^°, XXII d N0H prototropy the superposable mixture XXIIc and XXIId. Hence dif- ferentiation is not possible under ionizing conditions. An alternative and somewhat simpler way of analyzing situations arising from rotational isomerism or tautomerism (examples 8-101 is 172 Essays in Biochemistry- available if we recognize that the motions to be used for the super- position of identical substituents include not only those applicable to rigid bodies but also all internal motions of nuclei that proceed readily under reaction conditions. Obviously there is nothing in the derivation of rule 4 which would preclude such an interpretation of permissible motions.* As set forth above, numerous mechanisms can be conceived to explain the differentiation of identical substituents in suitably constituted molecules. The efficacy of these mechanisms, however, is often quite small, and the question arises whether any process besides three-point attachment to an asymmetric reagent can be expected to lead to a differentiation of a high order. Studies of relatively simple systems afford at least a tentative answer to this question. Only a few demon- strations of discrimination of identical substituents in symmetric mole- cules have been recorded in which an asymmetric reagent of known structure is employed. Although these experiments have clearly dem- onstrated the occurrence of the phenomenon, they have left some doubt as to the exact role of the asymmetric agent in the differentiation.4 The example reported most recently 4 appears to be no exception.8 In order to observe the differentiating powers of thermodynamic and kinetic factors separately, it seems best, therefore, to turn to related phenomena. If a racemic mixture of a compound containing a labile asymmetric center is permitted to interact with an optically active reagent, the 1:1 proportion of optical antipodes is frequently dis- turbed.1 The extent of such asymmetric transformations can be quite substantial even in homogeneous systems (62% excess in the case of chlorobromomethanesulfonic acid as a salt of ( — ) -hydroxyhydrinda- mine), but essentially complete conversion to one of the diastereo- isomers occurred only if the transformation was aided by selective precipitation. Although the asymmetry so induced is frequently lost again by racemization after removal of the optically active reagent, such transformations can be looked upon as a demonstration of the efficacy which thermodynamic factors can possess in asymmetric proc- esses. Good examples for kinetic control of asymmetric processes can be * Although the definition of symmetry number has also been adjusted to meet the problem of internal rotation (ref. 13, p. 510), such symmetry numbers are no longer applicable to our problem. For example, n-butane which has a "rigid symmetry number" of two and a "free-rotation symmetry number" of 18 contains six equivalent primary hydrogens but two non-equivalent pairs of secondary hydrogen atoms. ' Structural Differentiation in Asymmetric Reactions 173 found in certain addition reactions to carbonyl groups. An analogy between asymmetric additions and the differentiation of identical sub- stituents has been pointed out by Schwartz and Carter,4 who suggest that the "two bonds" linking C and a in a = Cbc (or perhaps more appropriately the two layers of high electron density occupied by the 7r electrons) could be considered to correspond to the two a groups of the ordinary meso carbon atom. The following example has been studied most thoroughly. The symmetric a-keto acid RiCOCOOH is esterified with the optically active alcohol H (OH) Cab; the resulting ester XXIII is treated with the Grignard reagent R2MgX and then j> -• OM.X A _A„/H A_/V/H COOH R. C C *- R, C C »-HO— C»-R, II / \ II / \ : O a b O a b R, XXIII XXIV XXV hydrolyzed to furnish the hydroxy acid XXV, which generally is found to consist of an unequal mixture of the two antipodes. Their propor- tion does not depend on the relative stabilities of the intermediate Grignard complexes XXIV, since an exchange of the alkyl groups between the keto acid and the Grignard reagent alters the sign of rota- tion of the resulting hydroxy acid. The steric result, therefore, is determined not by the nature of the product but by the mechanism of the addition reaction. Prelog et al.14 were able to relate the direct- ing influence of the alcohol H(OH)Cab to the bulk and orientation of the alkyl substituents a and b and therefore presumably to their differential ability to block the approach of the Grignard reagent to the keto group. Although the distance between the blocking group and the site of the reaction is rather large, the steric selectivity went as high as 69% excess. Even greater predominance of one isomer may result if the keto group and the directing asymmetric center are adjacent to each other.1'15 These results seem sufficiently encouraging to warrant the view that even a single linkage between an enzyme and its substrate might ex- plain a substantially complete differentiation of identical groups. This seems conceivable also in terms of a modified Ogston scheme if one assumes that the steric hindrance effects near the catalytic center are so graded and so distributed that only one of the two a groups can approach. As these mechanistic details may well be quite variable 174 Essays in Biochemistry and at the moment, at least, are largely a matter of conjecture, it seems fortunate that a knowledge of substrate structure alone suffices to foresee when differentiation is possible and when it is not. The author should like to acknowledge support by a grant from the National Institutes of Health, United States Public Health Service. References 1. P. D. Ritchie, Advances in Euzymol., 7, 65 (1947). 2. P. E. Wilcox, Nature, 164, 757 (1949). 3. D. W. Racusen and S. Aronoff, Arch. Biochem. and Biophys., 84, 218 (1951). 4. P. Schwartz and H. E. Carter, Proc. Natl. Acad. Sci. U. S., 40, 499 (1954). 5. A. G. Ogston, Nature, 162, 963 (1948). 6. P. E. Wilcox, C. Heidelberger, and V. R. Potter, J. Am. Chem. Soc, 72, 5019 (1950). 7. C. K. Ingold, Structure and Mechanism in Organic Chemistry, Chapter 7, Cornell University Press, Ithaca, N. Y., 1953. 8. M. L. Wolfrom, Proc. Natl. Acad. Sci. U. S., 40, 794 (1954). 9. A. Schoenfhes, Theorie der Kristallstruktur, Borntraeger, Berlin, 1923. 10. J. E. Rosenthal and G. M. Murphy, Revs. Mod. Phys., 8, 317 (1936). 11. F. M. Jaeger, Lectures on the Principle of Symmetry, 2nd ed., Elsevier, Amsterdam, 1920. 12. G. W\ Wheland, Advanced Organic Chemistry, 2nd ed., John Wiley & Sons, New York, 1949. 13. G. Herzberg, Molecular Spectra and Molecular Structure, II; Infrared and Raman Spectra of Polyatomic Molecules, Van Nostrand, New York, p. 508, 1-12, 1951. 14. V. Prelog, Helv. Chim. Acta, 36, 308 (1953); V. Prelog and H. L. Meier, ibid., 36, 320 (1953) ; W. G. Dauben, D. F. Dickel, O. Jeger, and V. Prelog, ibid., 36, 325 (1953). 15. R. Roger, Helv. Chim. Acta, 12, 1060 (1929); D. J. Cram and F. A. A. Elhafez, J. Am. Chem. Soc, 74, 5828 (1952). The Nitrogen-Sparing Effect of Glucose HENRY D. HOBERMAN The utilization of dietary nitrogen in the synthesis of body proteins is more efficient when starch, sucrose, or some other readily metabolized carbohydrate is present in the diet. This action of carbohydrates on the metabolism of nitrogen is commonly called the "nitrogen-sparing" effect. The retention of nitrogen induced in this way is a consequence of interactions in the intermediary metabolism of carbohydrates and amino acids and not of proteins. This follows from the fact that, when a small amount of N15-glycine or N15-aspartic acid is adminis- tered to fasting rats and to rats ingesting a solution of 30% glucose, 40% less N15-urea and 40% less N14-urea are excreted by the glucose- fed than by the fasting animals.1 The fall in the total urea-nitrogen output is quantitatively accounted for in terms of reactions at the amino acid level, for almost all of the excreted N15-urea was formed from the administered amino acid concurrently with and not after the incorporation of N15 into the body proteins. In similar experiments performed with N15-ammonia it was shown that the amount of N15 appearing in the urinary urea of animals receiving N15-ammonium citrate was 35% less when a solution of 30% glucose was ingested than when the animals were fasted.2 These results are interpreted to mean that in the sharing of common pathways of metabolism glucose sup- presses the oxidative deamination of amino acids and accelerates the synthesis of amino acids from ammonia, and/or that glucose inhibits the formation of urea by interfering with the operation of the Krebs- Henseleit cycle. In the following discussion these hypotheses will be examined in relation to our present knowledge of the reactions leading from the oxidative deamination of amino acids to the synthesis of urea. In mammalian liver the oxidative deamination of amino acids results from the activities of two enzymatic processes, i.e., direct oxidative 175 176 Essays in Biochemistry deaminations catalyzed by the flavoproteins, L-amino acid oxidase and glycine oxidase,3-4 and indirect oxidative deaminations catalyzed by aminophorases and DPN-dependent glutamic dehydrogenase.5 Where- as the deaminations brought about by the flavoproteins appear to be irreversible, this is not true of the reactions of the aminophorase — glutamic dehydrogenase system. Indeed with present evidence of the broad scope of transamination reactions 6 it may reasonably be as- sumed that in general the synthesis of amino acids from ammonia and a-keto acids occurs by way of the aminophorase-glutamic dehydro- enase system. Conditions are thus provided for a competition for ammonia between the glutamic dehydrogenase system and step I of the urea cycle, i.e., the citrulline-synthesizing system. This is indi- cated more clearly in the sequence of reactions shown. Step II Step I T Aspartic acid DPN+ L-Amino acids ;=± Glutamic acid v ^ Ammonia <— L-Amino acids DPNH In accordance with the fact that the specific enzymatic activity of crystalline beef-liver glutamic dehydrogenase is 10 times greater when the enzyme is present in the reduced than when in the oxidized form,7 it may be expected that a relatively small increase in the ratio of reduced to oxidized DPN may accelerate the formation of glutamic acid, diminish the concentration of hepatic ammonia, and thus reduce the rate of synthesis of urea. In the liver the synthesis and not the oxidation of glutamic acid predominates. Accordingly, the larger pro- portion of glutamic dehydrogenase is present in the reduced form and is maintained in this state by the operation of DPN-dependent coupled oxidations. Under conditions of low carbohydrate intake the coupling of the glutamic dehydrogenase system to endogenous oxidations, prin- cipally the oxidation of fat, may lead to a lower ratio of reduced to oxidized DPN than that which is established in the presence of exoge- nous glucose. Support for this hypothesis comes from the finding that fat, unlike glucose, does not evoke nitrogen retention in adult animals ingesting a protein meal.8 If the above concept is correct, we may explain the nitrogen-sparing effect of carbohydrates by assuming that, in the oxidative metabolism of ingested carbohydrates, the ratio, DPNH/DPN, coupled to the glutamic dehydrogenase system, is so The Nitrogen-Sparing Effect of Glucose 177 increased as to promote further than before the synthesis of glutamic acid from ammonia and a-ketoglutaric acid. The alternative hypothesis proposes that the nitrogen-sparing ac- tion of carbohydrates is a consequence of the inhibition of one or more reactions of the Krebs-Henseleit cycle. In step I of the urea cycle the synthesis of citrulline is accomplished in the enzyme-catalyzed reaction between ornithine and carbamyl phosphate, the latter com- pound being formed enzymatically from stoichiometric amounts of NH3, C02, and ATP.9 Siekevitz and Potter 10 have observed that the synthesis of citrulline by washed rat-liver mitochondria may be com- pletely inhibited in the presence of a sufficiently high concentration of hexokinase and glucose. Since it was also noted that the concentration of ATP in the medium declined to low levels, the authors concluded that the hexokinase reaction competes for ATP with the citrulline- synthesizing system. It is conceivable also that step II of the urea cycle is similarly blocked in the competition for ATP between the arginme-synthesizing system and the hexokinase reaction. Clearly, serious consideration must be given to the competition for ATP as a factor of physiological significance in regulating the rate of synthesis of urea. In rat-liver homogenates the synthesis of urea from ammonia and a-ketoglutaric acid is not as rapid as from glutamic acid.11 To account for this finding it has been suggested that a high concentration of a-ketoglutaric acid inhibits transaminations from glutamic acid, thus restricting the amount of aspartic acid available to the arginine- synthesizing system. This is still another way in which the formation of urea may be regulated by interactions between the Krebs-Henseleit cycle and the intermediary metabolism of carbohydrates. Referring once more to the sequence of reactions leading to the syn- thesis of urea from amino acids, it may be seen that, if the concept of interaction between the intermediary metabolism of carbohydrates and the aminophorase-glutamic dehydrogenase system is correct in accounting for the nitrogen-sparing effect of carbohydrates, it may be expected that the specific rate of utilization of ammonia in the forma- tion of amino acids will be increased in the presence of dietary carbo- hydrate and that the concentration of hepatic ammonia will decline. On the other hand, if nitrogen retention results from the blocking of step I of the urea cycle, it may be anticipated that the specific rate of utilization of ammonia in the synthesis of citrulline will decline and the concentration of ammonia in the liver will increase. In the event that the intermediary metabolism of carbohydrates invokes a 178 Essays in Biochemistry nitrogen-sparing action by inhibiting step II of the urea cycle, one may anticipate a decline in the specific rate of utilization of aspartic acid in the synthesis of urea and no change, or possibly a rise, in the concentration of hepatic ammonia. However, in view of the fact that the specific rate of step II would also decline in the event of the slow- ing of step I, only the anticipated effect on the ammonia concentra- tion is of interpretive value in this case. In order to arrive at the correct interpretation we will consider the kinetics of utilization of N15-ammonia in steps I and II of the urea cycle. In the scheme shown, a is the amount of N15 originally present as N15H3 ; Si is the fraction of N15H3 which, per unit time, is used in the synthesis of citrulline via step I; s2 is the fraction of N15, appearing in aspartic acid, which, per unit time, enters the urea cycle via step II; s3 is the fraction of N15H3 which, per unit time, is transformed to aspartic acid; and s4 is the fraction of N15 of aspartic acid which, per unit time, is incorporated into body proteins. U\ and c/2 are the amounts of N15 which are ultimately utilized in steps I and II, re- Step I Step II Ammonia > Aspartic acid > Proteins spectively. The assumed irreversibility of the utilization of ammonia in the formation of aspartic acid is in conformity with the view, pre- viously stated, that the reaction between ammonia and a-ketoglutaric acid stongly favors the formation of glutamic acid, as well as with the observation that aspartic-glutamic aminophorase of rat liver favors the formation of aspartic acid in the ratio of 2 to l.12 Theory shows that when all of the administered N15 is completely distributed between urea and the body proteins (neglecting the relatively slow return of isotope from labeled tissue proteins) : (U/a)Am = U\ + U2 s2Ss Si a (Si + s3)(s2 + s4) ' (sj +'s3) (1) Ui Si (2) a (sx + s3) U2 S2S3 (3) a (si + s3) (s2 + s4) The Nitrogen-Sparing Effect of Glucose 170 Since s2/{s2 + s4) is readily evaluated by measuring the fraction of isotopic urea formed from a given amount of N15 aspartic acid,13 the ratio s3/si may be calculated by substituting the appropriate data in the following equation: S3/S1 = (4) (I /a) Am - (U/a)As where (U/a)Am is the fraction of N15 given as ammonia which appears in the urinary urea in an arbitrary time (2 days) and (U/a)As is the fraction of N15 given as aspartic acid which appears in the urinary urea during an equal interval. In Table 1 are the results of experiments Table 1. Influence of Fasting and Glucose on the Utilization of N15-Labeled Precursors in the Synthesis of Urea and Amino Acids Rate of Excretion of Urinary Urea Conditions (U/a)Am ir/a)As S3/*l L'l/a Ui/a (mg. N/100 gm./hr.) Fasting 0.70 0.35 1 O.oO 0.18 2.2 Ingesting glucose 0.46 0.20 2 0.33 0.13 1.3 For definition of symbols see text. carried out, as indicated, on fasting rats and on animals ingesting a solution of 30% glucose. Attention is directed first to the results of calculations which show that glucose induces a substantial shift in the utilization of ammonia from step I of the urea cycle to the synthesis of amino acids. Theo- retically the proportion of amino acids formed from ammonia is S3 (S3 -f- s4). The calculated utilization of ammonia in the synthesis of amino acids in animals ingesting glucose is thus two-thirds, and in fasting animals one-half so that one-third more ammonia is trans- formed to amino acids and one-third less is used in the formation of urea. Although this accounts in large measure for the observed fall in the output of the total urea nitrogen of glucose-treated animals, the question which must be raised here is wrhether the increase in the ratio of s3/si resulting from the ingestion of glucose represents an increase in the absolute value of s3, or whether Si is decreased in relation to s3. In accordance with earlier considerations, an increase in the absolute value of s3 may be expected to lead to a decrease in the concentration of ammonia in the liver whereas a decline in the absolute value of s3 may be expected to have the reverse effect. It was found that the con- centration of ammonia in the liver of the glucose-fed rats was approxi- mately 30% less than in fasting animals. This was indicated by the 180 Essays in Biochemistry observation that the isotope concentration of the total urea nitrogen of the rats receiving glucose was 30% higher than that of the total urea nitrogen of fasting animals. The experimental results therefore support the view that the nitrogen-sparing effect of carbohydrates is a consequence of the coupling of the oxidation of carbohydrates to the aminophorase-glutamic dehydrogenase system. References 1. H. D. Hoberman, unpublished observations. 2. H. D. Hoberman and J. Graff, J. Biol. Chem., 186, 373 (1950). 3. M. Blanchard, D. E. Green, V. Nocito, and S. Ratner, J. Biol. Chem., 155, 421 (1944). 4. S. Ratner, V. Nocito, and D. E. Green, J. Biol. Chem., 152, 119 (1944). 5. A. E. Braunstein, Advances in Protein Chem., 3, 1 (1947). 6. P. S. Cammarata and P. P. Cohen, J. Biol. Chem., 187, 439 (1950). 7. J. A. Olsen and C. B. Anfinsen, J. Biol. Chem., 197, 67 (1952). 8. H. N. Munro, J. Nutrition, 39, 375 (1949). 9. M. E. Jones, L. Spector, and F. Lipmann, J. Am. Chem. Soc, 77, 819 (1955). 10. P. Siekevitz, and V. R. Potter, J. Biol. Chem., 201, 1 (1953). 11. S. Ratner, Advances in Enzymology, 15, 319 (1954). 12. P. P. Cohen, and G. L. Hekhuis, J. Biol. Chem., lJfi, 711 (1941). 13. H. D. Hoberman, J. Biol. Chem., 188, 797 (1951). The Metabolism of Inositol in Microorganisms A STUDY OF MOLECULAR CONFORMATION AND RIOCHEMICAL REACTIVITY BORIS MAGASANIK Inositol * was discovered in muscle extract by Scherer a little over 100 years ago and identified as a hexahydroxycyclohexane by Ma- i Scyllitol II myo - Inositol III neo ■ Inositol IV D - Inositol V L- Inositol VI epi - Inositol VII muco - Inositol VIII alio ■ Inositol IX Unknown Fig. 1. The configurations of the inositols. * Two reviews of the chemistry and biological activity of the inositols have appeared in recent years.1-2 In this paper the nomenclature proposed by Fletcher, Anderson, and Lardy (see ref. 2) is used. 181 182 Essays in Biochemistry quenne 37 years later. Bouveault pointed out that theoretically nine stereoisomers of inositol could exist, two of which would be optical enantiomorphs. Four of these isomers are known to occur in nature. Scherer's muscle sugar, now called mT/o-inositol, an ubiquitous cell component, has been recognized as a growth factor for certain yeasts and molds and as a vitamin necessary for the health of rats and mice. Scyllitol, originally discovered in the organs of fish, occurs in trees and in human urine. D-inositol and L-inositol are found as monomethyl ethers in a variety of plants. Similarly, two of the sixteen possible stereoisomer^ deoxyinositols have been isolated from plants. Of the remaining inositols, four have been synthesized, so that at present only one of the nine isomers remains unknown. The configurations of the inositols (Fig. 1) were determined by Posternak and by Dangschat and Fischer through the conversion of derivatives of the inositols to known saccharic acids. The studies described here were begun in collaboration with Dr. Erwin Chargaff at the Department of Biochemistry, Columbia Univer- sity, in 1946 and have their origin in the observation of Kluyver and Boezaardt 3 that Acetobacter suboxydans oxidizes mi/o-inositol (II) to a monoketone, subsequently identified by Posternak as 2-keto-myo- inositol (X).4 H,0, A. suboxydans The microorganism had thus singled out the central one of the three vicinal as-hydroxyl groups of M7/o-inositol for oxidation. The ability of A. suboxydans to carry out partial oxidations of polyhydroxy com- pounds to monoketones had long been known to be subject to certain steric limitations, which were formulated by Bertrand and by Hudson as the following rule: Only a secondary hydroxyl group located be- tween a primary hydroxyl group and another secondary hydroxyl group in cis position is oxidized. However, since this rule could obviously not be applied to cyclic compounds, it seemed of interest to study the specificity of the enzy- matic attack of A. suboxydans on the hydroxyl groups of cyclitols, par- ticularly as the rigidity of these cyclohexane derivatives, owing to the Metabolism of Inositol 183 lack of free rotation around carbon to carbon bonds, would permit a clearer correlation between reactivity and the spatial arrangement of the reactive groups. The measurement of the oxygen uptake of resting cell suspensions of .4. suboxydans with several isomers of inositol as substrates revealed that these compounds differed greatly in their susceptibility to en- zymatic attack. Afi/o-inositol and e/n-inositol were oxidized with the uptake of 1 gram atom of oxygen per mole and L-inositol, D-inositol, and L-2-deoxy-mt/co-inositol (d-quercitol) were oxidized with the up- take of 2 gram atoms of oxygen per mole, whereas scyllitol was not attacked at all.5 The oxidation product of epi-inositol was isolated and found to be a levorotatory ketoinositol which on reduction with sodium amalgam yielded m?/o-inositol, indicating that the hydroxy 1 group in either posi- tion 2 or 4 of epi-inositol had been oxidized.5 Posternak isolated the same keto compound and converted it by oxidation with permanganate to a mixture of D-talomucic and D-glucosaccharic acid.6 These re- actions identified the Acetobacter oxidation product as D-2-keto-e/n- inositol (XI). O 'AO, / \| Na-Hg A. suboxydans VI XI The final products of the oxidation of d- and of L-inositol were iso- lated by means of phenylhydrazine. Both products proved to be bisphenylhydrazones of diketoinositols; they had identical melting points, identical absorption spectra characteristic for osazones, and optical rotations equal but opposite in sign. The consumption of periodic acid corresponded to 3 moles of the oxidant per mole of bis- phenylhydrazone. The products of this reaction were cyclic deriva- tives of the dialdehydes expected in the periodic acid oxidation of bisphenylhydrazones of a-diketoinositols. The racemic mixture of the two enantiomorphic a-bisphenylhydrazones was found to be identical with the osazone obtained by the treatment of the phenylhydrazone of 2-keto-wi/o-inositol (X) with phenylhydrazine. These reactions, shown on the accompanying flow sheet, identified the oxidation prod- ucts of d- and of L-inositol as L-l,2-diketo-?m/o-inositol (XII) and D-l,2-diketo-mi/o-inositol (XIII), respectively.5 184 Essays in Biochemistry A. suboxydans + 2C6H5NHNH2 NNHC6H5 H5C6HNN NNHC6H5 H5C6HNN H H5C6-N^ ^C-N = NC6H5 I I N=C-CHO 3HI04 C6H5NHNH2 NNHCHc -H..0 HC-OH II C-N = NC6H, C = NNHC6HS CHO II CHO C = NNHC6H< I C = NNHC6H5 CHO Phenvlhydrazone of X It was possible to isolate the monoketoinositol which is the initial product of the action of A. suboxydans on D-inositol (IV). The com- pound was identified as L-l-keto-?ni/o-inositol (XIV) by its oxidation with resting cells of A. suboxydans to the diketone XII and by its reduction with hydrogen catalyzed by platinum to ??i?/o-inositol (II).7 The product formed by the enzymatic oxidation L-2-deoxy-muco- inositol (d-quercitol) (XV) was found to react with phenylhydrazine to form a bisphenylhydrazone which had an absorption spectrum char- acteristic of osazones and which reacted with periodic acid with the uptake of 2 moles of oxidant. These observations indicated that the keto groups were vicinal and that the three hydroxy! groups were located on adjacent carbon atoms. The choice between the two pos- sible structures was made by comparing the rate of oxidation by periodic acid of this a-bisphenylhydrazone and the one prepared from Metabolism of Inositol o IV XIV XII Pt, Hi D-inositol (XII). The deoxyinositol derivative was not attacked more rapidly than the D-inositol derivative in which all hydroxyl groups are in trans position. The oxidation product of XV was therefore identi- fied as D-2,3-diketo-4-deoxy-e?n-inositol (XVI), in which the three adjacent hydroxyl groups are in trans position.8 o, x = 0 A. suboxydans xv XVI The action of A. suboxydans on these cyclitols seemed at first to fit no easily discernible pattern. In the case of the inositol isomers, one hydroxyl group of a pair of adjacent cis hydroxyls had been attacked, suggesting a specificity of oxidation similar to the one found in straight-chain compounds, but the oxidation of the hydroxyl group on carbon 3 of L-2-deoxy-mwco-inositol (XV) was totally unexpected. However, it must be borne in mind that the structural formulas (Fig. 1) do not describe the actual position of the hydroxyl groups in space but represent merely planar projections based on the conventions intro- duced into stereochemistry by Emil Fischer. It had long been known that the cyclohexane molecule was not planar but could exist as a strainless ring in the "boat" and "chair" forms; later evidence ob- tained in studies using electron diffraction and infrared spectroscopy had led to the recognition of the chair form as the stable conformation of the cyclohexane ring (Fig. 2) . Inspection of a model of cyclohexane 186 Essays in Biochemistry in the chair form reveals that the carbon atoms are equidistant from a plane passing through the center of the molecule. Six of the free Fig. 2. The cyclohexane ring in the chair form. Equatorial bonds are shown as lines, north polar bonds as lines ending in solid circles, and south polar bonds by lines ending in open circles. n in 3D VII VIII Fig. 3. The conformations of the inositols. Equatorial hydroxyl groups are represented by lines, north polar hydroxyl groups by solid circles, and south polar hydroxyl groups by open circles. The hydrogen atoms are not shown. bonds on the carbon atoms are located in that plane projecting out- wards and are called equatorial bonds. The other six bonds are per- pendicular to the plane being directed alternately upwards (north Metabolism of Inositol 187 polar) and downwards (south polar).* Each carbon atom possesses one equatorial and one polar bond and can therefore carry substituents in either position. There are two possible chair forms, one being formed from the other by a movement of the ring atoms through the central plane which changes polar bonds to equatorials and vice versa. In substituted cyclohexanes, the substituent groups occupy preferen- tially the equatorial positions in which more space is available than in the crowded polar regions. Consequently, when the two chair forms are not identical, the molecule will be found to exist in the conforma- tion having the smallest number of polar substituents. On the basis of these considerations the conformations shown in Fig. 3 could be assigned to the inositol isomers. It was reasonable to expect that the susceptibility of the inositol molecule to chemical and enzymatic attack would depend on the actual shape of the molecule as shown by its conformation; and indeed, on correlation of the action of A. suboxydans on the inositols with their conformations (Table 1), Table 1. The Action of A. suboxydans on Inositols and Their Deoxy and Keto Derivatives Polar Action of A .. suboxydans Hydroxyls, Positional 02 Taken Up, Hydroxyls Oxidized, Compound Numbers * gram atoms per mole Positional Numbers * Ref. 1 Scyllitol None None 5 II myo-Inofiitol 2 1 2 4 III neo-Inositol 2, 5 None IV D-Inositol 2, 3 2 2,3 5 V L-Inositol 2,3 2 2, 3 5 VI epi-Inositol 2,4 1 2 5,6 VII njuco-Inositol 2,3,4 2 Not determined 9 XVIII 2-Deoxy-m?/o-i nositol None None 9 XXII D- 1 - Deoxy-m#o-inosi tol 2 1 2 10 XXI L-1-Deoxy-mj/o-inositol 2 1 2 10 XIX D-2-Deoxy-ep) ■inositol 4 None 9 XX L-2-Deoxy-epi- •inositol 4 1 4 9 XV L-2-Deoxy-mwco-inositol 3, 4 2 3,4 8 X 2-Keto-mj/o-inositol None None 4 XI D-2-Keto-epi-inositol 4 None 5 XVII L-2-Keto-epi-inositol 4 1 Not determined 5 XIV L-1-Keto-jm/o- inositol 2 1 2 7 * In meso compounds alternative numbering sequences are possible (see Fig. 1); for presentation in this table these compounds have been numbered clockwise. a striking regularity became at once apparent: only polar hydroxyl groups had been oxidized. *The use of the term "axial" instead of "polar" has been suggested to avoid ambiguity (D. M. R. Barton et al., Science, 119, 49 (1954). 188 Essays in Biochemistry However, it was also apparent that not all polar hydroxyl groups were attacked by the microorganism. Only one of the two polar groups of epi-inositol (VI) had been oxidized to yield the optically active D-2-keto-epi-inositol (XI), whose polar hydroxyl group in position 4 was resistant to further enzymatic oxidation. L-2-Keto-epi-inositol (XVII), on the other hand, could be attacked by A. suboxydans, and presumably converted to a diketone, as shown by the observation that racemic DL-2-keto-epi-inositol (XI -+- XVII) (produced from myo- inositol by oxidation with nitric acid) was oxidized with the uptake of 0.5 gram atoms of oxygen per mole. In order to define the requirements for oxidation by A. suboxydans with greater stringency, three deoxyinositols were prepared and sub- jected to the action of the microorganism. 2-Deoxy-ra?/o-inositol (XVIII), prepared by the catalytic reduction under acid conditions of 2-keto-?m/o-inositol (X), was not attacked. DL-2-Deoxy-epi-ino- sitol (XIX -f XX) , prepared in a similar manner from DL-2-keto-e7)i- inositol (XI -f- XVII), was oxidized with the uptake of 0.5 gram atoms of oxygen per mole. The isomer oxidized was identified as L-2-deoxy- epi-inositol (XX) by the isolation of unchanged XIX from the re- action mixture, as well as by the demonstration that XIX prepared from D-2-keto-epi-inositol (XI) was resistant to the action of A. sub- oxydans. The point of the enzymatic attack on XX was identified as the polar hydroxyl group in position 4 by the reduction of the resulting monoketone with sodium amalgam to 2-deoxy-m?/o-inositol (XVIII).9 XX XVIII A Inspection of the conformations of those pairs of enantiomorphs of which one member only is oxidized by A. suboxydans revealed that the isomers susceptible to attack all possess an equatorial hydroxyl group in position d relative to the location of the oxidizable polar hydroxyl group. Metabolism of Inositol 189 Replacement of this equatorial hydroxyl by a polar hydroxyl (VIb) ,* by oxygen (XI) or by hydrogen (XIX) prevents enzymatic oxidation; col-responding changes in position b are without effect. Not oxidized VI b* XI XIX Oxidized VI a* XVII XX The validity of this generalization was confirmed and its scope ex- tended by considering the other cyclitols that had been studied. Myo- inositol (XI) has five equatorial hydroxyls. The replacement of the equatorial hydroxyl in position a by a polar hydroxyl group (IV), by hydrogen (XXI), or by oxygen (XIV) does not interfere with the oxidation of the north polar hydroxyl group. Oxidized IV XIV Similarly, oxidation of the polar hydroxyl group occurs when the equatorial hydroxyl in position e is replaced by a polar hydroxyl (V) or by hydrogen (XXII). Oxidi v XXII The importance of the equatorial hydroxyl group in position c could at that time not be ascertained, as no compound without an equatorial * The structures are so arranged as to place the polar hydroxyl group under comparison at the top of the hexagon. For this reason epi-inositol (VI) is shown in two arrangements (Via and VIb). 190 Essays in Biochemistry hydroxyl in this position was available. Recently neo-inositol (III) which carries a polar hydroxyl group in position c was synthesized by Angyal ; 1X his kind gift of a small amount of this isomer permitted its use as a substrate for A. suboxydans. It was not attacked; appar- ently the presence of an equatorial hydroxyl group in position c is re- quired for oxidation by the microorganism. Not oxidized III The results of these studies can be summarized in three rules defin- ing the steric requirements for the oxidation of inositols, deoxyinosi- tols, and ketoinositols by A. suboxydans (Fig. 4). H 1 OH H H0~Vv ^\" - .""V V- ^^ A. suboxydans \ ho-1 H r HO-n H Fig. 4. The steric requirements for oxidation by A. suboxydans. 1. Only polar hydroxyl groups are oxidized. 2. The carbon in meta position to the one carrying the polar hy- droxyl group (in counterclockwise direction if north polar, clockwise if south polar) must carry an equatorial hydroxyl group. 3. The carbon in para position to the one carrying the polar hydroxyl group must carry an equatorial hydroxyl group. These structural requirements demonstrate three points of contact between substrate and enzyme and can account for the oxidation of epi- inositol (VI) to D-2-keto-epi-inositol (IX), an asymmetric synthesis which theoretically demands a three-point attachment of substrate to enzyme. The carbon atom carrying the polar hydroxyl group seems to be particularly susceptible to dehydrogenation, as shown by the recent report that platinum catalyzes the specific conversion of myo- inositol (II) to 2-keto-?/i?/o-inositol (X) with oxygen as hydrogen ac- ceptor.12 The initial attack of the A. suboxydans enzyme is presum- ably directed against the equatorial hydrogen atom located on the carbon carrying the polar hydroxyl group; the enzyme-substrate com- Metabolism of Inositol 191 plex may be held together by bonds between the enzyme or a metal ion associated with the enzyme and the two required equatorial hydroxy! groups which occupy the same plane as the hydrogen atom (Fig. 4). The rules which have been presented apply to hexahydroxycyclo- hexanes, pentahydroxycyclohexcmes, and their keto derivatives. Tri- hydroxycyclohexanes are oxidized by a different enzyme and one sub- ject to rules which have as yet not been elucidated; it is possible that this enzyme is identical with the one responsible fur the oxidation of straight-chain polyhydroxy compounds. The biological significance of the inositol dehydrogenase of A. sub- oxydans is not known. The enzyme is not adaptive but is present in cells grown in the absence as well as the presence of inositol. The organism obtains useful energy but no building blocks for the synthesis of its protoplasm by the incomplete oxidation of the inositols. In other species of microorganisms, however, ??it/o-inositol can be metabolized with the production of energy and building blocks. This can be in- ferred from the observation that seven bacterial species of a group of fourteen can grow on ?ni/o-inositol as the only source of carbon.13 One of these species, Aerobacter aerogenes, was chosen for a study of this extensive degradation of the inositol molecule. It was found that in addition to ?m/o-inositol (II), four other cycli- tols [D-inositol (IV), 2-keto-?m/o-inositol (X), L-l-keto-mi/o-inositol (XIV) and L-l,2-diketo-m(/o-inositol (XII) 1 could support the growth of capsulated strains of A. aerogenes as sole sources of carbon. Scyl- litol (I) was not attacked by the microorganism but inhibited specifi- cally and reversibly the dissimilation of these five compounds. The other cyclitols tested (V, VI, XIX, XX, XV, XI, XVII, and XIII) showed neither growth-supporting nor inhibitory activity.14 Suspensions of cells grown on glucose did not oxidize the five cyclitols immediately but only after a period of lag of about 1 hour, although suspensions of cells grown on ra?/o-inositol oxidized this compound as well as the other four cyclitols without lag. The process occurring during the period of lag could be shown to require energy by its sus- ceptibility to the inhibitory action of dinitrophenol. The energy was 192 Essays in Biochemistry apparently used for the biosynthesis of protein from small precursors, since the ability of glucose-grown amino acid-deficient mutants of this organism to attack ?m/o-inositol was stimulated by the amino acid required for growth.15 These experiments indicated that the attack on the cyclitols was mediated by adaptive enzymes whose synthesis could be induced by myoinositol. The validity of this assumption was confirmed by the demonstration that extracts of m?/o-inositol- grown cells possessed two enzymes (later identified as a dehydrogenase acting on inositols II and IV and a dehydrase acting on the ketoinosi- tols X and XIV), which were not found in extracts of glucose-grown cells.16 The enzymes could be separated by treatment with protamin sulfate. The dehydrogenase was precipitated and could be redissolved by the dissociation of the insoluble protamin salt with polymeth- acrylate. The dehydrogenase purified in this fashion was found to catalyze the reduction of diphosphopyridine nucleotide (DPN), but not that of triphosphopyridine nucleotide, by myoinositol, and more slowly by D-inositol, in the presence and absence of inorganic phosphate. The equilibria of these reactions were too unfavorable for dehydrogenation (even at a relatively high pH) to permit the isolation of the products; therefore the reverse reaction, the oxidation of DPNH by ketoinositols, was investigated. 2-Keto-myo-inositol (X) reacted with DPNH in the presence of the enzyme to give equimolar amounts of DPN and of ?m/o-inositol (II) (determined by bioassay with inositol-less Neuro- spora crassa) ; L-l-keto-myo-inositol (XIV) reacted with DPNH to give DPN but was itself not converted to myoinositol. These re- sults indicated that the enzyme catalyzes the conversions of myo- inositol (II) and of D-inositol (IV) to monoketones by removing the O Reaction la + DPN+ , * + DPNH + H" Reaction 1 b + DPN + , ' + DPNH + H + IV XIV Metabolism of Inositol 193 hydrogen atom from carbon atoms carrying a polar hydroxyl group (reactions la, b). The Aerobacter enzyme had thus singled out the same carbon atom of mt/o-inositol for dehydrogenation as the Acetobacter enzyme. Ap- parently, a carbon atom carrying a polar hydroxyl group is more sus- ceptible to dehydrogenation than one carrying an equatorial hydroxyl group; this concept is in good agreement with the observation men- tioned earlier that platinum similarly catalyzes specifically the attack on the carbon atom of ?m/o-inositol which carries the polar hydroxyl group. The inability of the Aerobacter enzyme to act on most of the cyclitols which are attacked by the Acetobacter enzyme shows that the steric requirements of the two enzymes other than their specificity for polar hydroxyl groups are not the same. A. suboxydans does not possess the enzymes necessary to carry the attack on mi/o-inositol beyond 2-keto-rayo-inositol. A. aerogenes, on the other hand, possesses an enzyme found in the supernatant fluid after treatment with protamin sulfate, which acts on 2-keto-myo- inositol (X) in the absence of added cofactors. The strong absorption band of the product {E, 4000) at 261 m^ and its reducing properties suggested it to be an a, ft unsaturated ketone. This assumption was confirmed by the isolation of this compound by means of phenylhydra- zine. The hydrazone was identified by its reaction with periodic acid and its absorption spectrum as the bisphenylhydrazone of 2,3-diketo-4- deoxy-e/w-inositol (XVI) ; the same compound had previously been obtained through the oxidation of L-2-deoxy-mwco-inositol (XV) by A. suboxydans. These results show the enzyme to be a dehydrase which converts 2-keto-mt/o-inositol to the enol XXIII by attacking one of the equatorial hydroxyl groups in meta position to the keto Reaction 2 a Reaction 2 b XXIII H XIV XX XVI 194 Essays in Biochemistry group (reaction 2a). A similar attack converts L-1-keto-wi^o-inositol to the same product at a slower rate (reaction 2b). The presence of the dehydrogenase and the dehydrase in extracts of ra?/o-inositol-grown cells indicates that the initial attack of A. aero- genes on m7/o-inositol is a dehydrogenation (reaction la), followed by dehydration (reaction 2a). The ability of inositol-adapted cells to attack 2-keto-wi/o-inositol is in agreement with its role as an interme- diate in this reaction sequence. The simultaneous adaptation to D-inositol and L-1-keto-m^o-inositol finds its explanation in the fact that these compounds are attacked by the same enzymes as myo- inositol and 2-keto-ra?/o-inositol and converted to a common product, the enol XXIII. The further steps in the degradation of rat/o-inositol have as yet not been demonstrated in a cell-free system. The rapid oxidation and fermentation of L-l,2-diketo-m7/o-inositol (XII) by suspensions of rai/o-inositol-grown cells suggests that this compound is an intermedi- ate in the degradation of m?/o-inositol. It could presumably arise by oxidation of the enol XXIII (reaction 3). Reaction 3 XXIII The broad outlines of the metabolic pathway leading to the com- plete degradation of ?m/o-inositol were elucidated by the study of the products of inositol dissimilation under aerobic and anaerobic condi- tions.17 The oxidation of w?/o-inositol, 2-keto-ra?/o-inositol, or 1,2- diketo-m?/o-inositol by resting cell suspensions of A. aerogenes yielded the same products as the oxidation of glucose. The amount of C02 produced and of 02 taken up was sufficient for the complete oxidation of one-half of the molecule to CO2 and H20. The other half of the molecule was apparently assimilated as material of the composition C3H60.3. In the presence of dinitrophenol, an agent known to inhibit oxidative assimilation, the oxygen uptake approached the theoretical values for complete oxidation. In the presence of As203, an inhibitor of pyruvate degradation, glucose was degraded largely to 3-carbon compounds (pyruvate and lactate) whereas w?/o-inositol and the keto- inositols were dissimilated to a mixture of 3-carbon compounds (pyru- vate and lactate), 2-carbon compounds (acetate and ethanol), and Metabolism of Inositol 195 C02. Similar patterns were observed in the anaerobic degradations of glucose and the cyclitols. Glucose yielded acid, but no C02; in the presence of As203 glucose was converted to 2 moles of lactic acid per mole. On the other hand, myo-'mositol, as well as its keto derivatives, was fermented with the production of 1 mole of C02 per mole of sub- strate. In the presence of As203 myoinositol was converted to an D- Glucose C, + C, myo- Inositol -2H 03 + 0;;+ Cj -« O 2 - Keto - myo ■ inositol L-l,2-Diketo- tnyo - inositol C.i H fiO: 6W3 + 3C02. CjHgO,-, + 3C02 Fig. 5. The proposed pathways of degradation of glucose and of myo-inositol in Material assimilated. A. aerogenes. C3H6°3 equimolar mixture of C02, ethanol, and lactic acid; apparently pyru- vate and acetate (or an active form of acetate) can serve as hydrogen acceptors in the dehydrogenation steps (reactions la and 3). Thus glucose and m^/o-inositol are metabolized by A. aerogenes by different pathways to the same final products (Fig. 5). The direct conversion of my o -inositol to glucose by cleavage of the bond between carbon atoms 3 and 4 which has frequently been postulated does not occur in this system. However, the production of pyruvate from myo- 'mositol by a pathway corresponding to the one described in A. aero- genes could well explain the antiketogenic effect of ra?/o-inositol, and the conversion of stably bound deuterium in ???i/o-inositol to stably bound deuterium in glucose which have been observed in the rat.2 The pathway of inositol degradation is superficially similar to the 196 Essays in Biochemistry "oxidative pathway" of glucose dissimilation, in that in both the carbon chain is split with the production of C02 and a 3-carbon compound. However, the inositol pathway has the distinguishing characteristic that the initial enzymatic attack is a dehydrogenation and not a phos- phorylation. The degradation proceeds at least as far as the diketone XII without the introduction of a phosphate group. It is not known by what mechanism the energy generated in the dehydrogenation steps (reactions 1 and 3) can be utilized by the cell. The observation that the microorganism can grow on myoinositol in the absence of oxygen suggests that a process other than oxidative phosphorylation can serve to harness the reactions of inositol degradation to the production of useful energy. The ability to initiate the degradation of a polyhydroxy compound with the dehydrogenation of a secondary hydroxy 1 group seems to be characteristic of certain microorganisms. A. suboxydans converts mi/o-inositol to 2-keto-myo -inositol and glycerol to dihydroxyacetone, but it does not possess the enzymes for the rapid degradation of these keto compounds. Capsulated strains of A. aerogenes can carry out not only the complete degradation of myoinositol but also that of glycerol.18 The direct dehydrogenation of glycerol by an enzyme whose synthesis is specifically induced by glycerol yields dihydroxyacetone which is rapidly dissimilated via pyruvic acid. On the other hand, acapsulated strains of A. aerogenes cannot attack w?/o-inositol at all and initiate the degradation of glycerol with its phosphorylation to L-a-glycerophosphate. The pathways of dissimilation in which dehydrogenation is the initial step of the enzymatic attack are as efficient as the classical Embden-Meyerhoff pathway in providing capsulated A. aerogenes with energy and building blocks for growth, since the growth rate and the total cell crop are nearly the same whether glucose, glycerol, or myo- inositol is the sole carbon source. However, the formation of adaptive enzymes, such as the biosynthesis of histidase, which is induced by histidine, proceeds during growth on m7/o-inositol or glycerol but not during growth on glucose.19 Thus the ultimate source of energy may have a profound influence on the enzymatic constitution of the bac- terial cell. References 1. H. G. Fletcher, Jr., Advances in Carbohydrate Chemistry, 73, 2917 (1951). 2. R. S. Harris et al., in W. H. Sebrell, Jr., and R. S. Harris, The Vitamins, II, Academic Press, New York, 1954, p. 322 ff. Metabolism of Inositol 197 3. A. J. Kluyver, and A. G. J. Boezaardt, Rec. trav. chim., 58, 956 (1939). 4. T. Posternak, Helv. Chim. Acta, 24, 1045 (1941). 5. B. Magasanik and E. Chargaff, /. Biol. Chem., 17',, 173 (1948). 6. T. Posternak, Helv. Chim. Ada, 29, 1991 (1946). 7. B. Magasanik and E. Chargaff, J. Biol. Chem., 175, 929 (1948). 8. B. Magasanik and E. Chargaff, J. Biol. Chem., 175, 939 (1948). 9. B. Magasanik, R. E. Franzl, and E. Chargaff, /. Am. Chem. Soc, 74, 2618 (1952). 10. T. Posternak, Helv. Chim. Acta, 83, 350, 1594 (1950). 11. S. J. Angyal and N. K. Matheson, J. Am. Chem. Soc, 77, 4343 (1955). 12. K. Heyns and H. Paulsen, Chem. Ber., 86, 833 (1953). 13. L. E. den Dooren de Jong, Dissertation, Rotterdam, 1926. 14. B. Magasanik, J. Biol. Chem., 205, 1007 (1953). 15. D. Ushiba and B. Magasanik, Proc. Soc. Exptl. Biol. Med., SO, 626 (1952). 16. J. M. Goldstone and B. Magasanik, Federation Proc, IS, 218 (1954), and unpublished observations. 17. B. Magasanik, J. Biol. Chem., 205, 1019 (1953). 18. B. Magasanik, M. S. Brooke, and D. Karibian, J. Bacteriol, 66, 611 (1953). 19. B. Magasanik, J. Biol. Chem., 218, 557 (1954). The Biochemistry of Ferritin ABRAHAM MAZUR An interpretation of biological phenomena in terms of chemical structure or chemical interaction remains a basic goal of biochemistry. Progress along these lines has been made with compounds of low molecular weight and relatively simple structure. Examples are the reactions involved in metabolism of carbohydrates, fatty acids, and amino acids; these reactions can now be written in great detail and their mechanisms have been revealed. Such advances have been aided in no small measure by the relative ease with which these compounds can be obtained in a pure state and by the simplicity of the criteria for their purity. The difficulties are greatly multiplied for macromolecules such as proteins, where the criteria for purity are themselves often in question. Progress in relating function to chemical structure is currently in evi- dence among one group of proteins, the enzymes; their activities are measured by methods which are relatively simple, quick, and quanti- tative. However, for proteins endowed with hormonal activity, our understanding is more limited. Because of the nature of the tests for physiological activity, the isolation of protein hormones in a pure state has been slow and studies relating their activity to structure are ham- pered by lack of sufficient amounts of the protein, lack of accuracy of the determination, and the length of time required to obtain an ade- quate measure of activity. An additional factor which complicates the study of protein hormones is the fact that the observed activity in an animal, organ, or tissue is usually the end result of a series of re- actions which occur between the administration of the protein and the final observation of functional activity, with no evidence concerning the nature of the initial or intermediate reactions. 198 The Biochemistry of Ferritin 199 Methods of study relating the activity of protein hormones to their chemical structure can be patterned after those used so successfully with enzymes. The early suggestions concerning enzyme-substrate interaction were brought to a focus by the Michaelis-Menten formula- tion of enzyme-substrate compound formation and subsequently to the implication in the reaction of specific groups on the enzyme surface. More difficult to study from a dynamic point of view are those cellu- lar proteins which we loosely term "structural" or "storage" proteins. Certainly such proteins have a biological function and can undergo chemical 'alterations in addition to those of biosynthesis and degrada- tion, but the problem of measuring such subtle changes or attempting to relate the structure of these proteins to function is obviously very difficult. The present report is concerned, nevertheless, with attempts to study the relationship of protein structure to biological activity where the protein is a "storage" protein, ferritin. Work in our laboratory, involving alterations in the circulation of animals in hemorrhagic shock led to the identification of the iron protein ferritin with a substance liberated in very small quantities into the circulation during this state. The physiological activities which our early work had associated with ferritin seemed unrelated at that time to its well-known iron-storage function. These activities, which might be termed hormonal if they could be shown to operate under physiological rather than pathological conditions, are twofold: (a) The intravenous injection into a normal rat of very small quan- tities of ferritin results in a temporary inhibition of the constrictor response to the topical application of adrenaline on the part of the muscular capillary blood vessels in the mesentery. This has been called its "vasodepressor" effect. (6) The injection of ferritin into the circulation of the hydrated rabbit or dog stimulates the neurohypophysis to the secretion of its antidiuretic hormone, which in turn acts on the kidney tubules to bring about an increased resorption of water, i.e., an oliguric or "anti- diuretic" effect. With these two activities, non-quantitative and time-consuming as their measurements may be, we were better able to investigate the relationship of the functional groups in ferritin to activity than if we had known only about its iron-storage activity. As will be seen, our present findings make it likely that all three physiological activities of ferritin are related to the same chemical groups and to similar alterations in their structure. 200 Essays in Biochemistry The Bulk of Ferritin Iron Ferritin is found mostly in the liver and spleen and to a lesser extent in bone marrow, kidneys, and placenta, and in much smaller quantities in skeletal muscle, testes, and pancreas. It has also been reported to be present in the intestinal mucosa of the anemic guinea pig in re- sponse to iron feeding.1 Table 1 lists quantitative data for the ferritin Table 1. Ferritin Content of Various Tissues Hemo- globin, Ferritin N, /tg./gm. wet tissue gm. per R.B.C. Bone Kidney Pan- Skeletal Cardiac Pla- Species cent X 106 Liver Spleen Marrow Cortex creas Muscle Muscle centa Dog 16.2 6.1 134 90 16 18 6 2 1 — Dog 14.1 6.2 65 99 18 — 8 5 1 — Dog 14.1 6.1 75 166 12 14 3 9 2 — Dog 14.1 6.1 115 42 — 22 7 2 1 6 10 Human Human Human 12 A. Mazur and E. Shorr, J. Biol. Chem., 182, 607 (1950); determined by the quantitative immuno- chemical method. content of dog tissues 2 and human placenta, obtained by the quantita- tive immunochemical method of Heidelberger. The best source for the isolation of crystalline ferritin is horse spleen, from which it is prepared by the method of Granick 3 using CdS04 for crystallization of the protein as first recommended by Laufberger.4 Concentrated ferritin solutions free of inorganic ions are obtained by dialysis and can be stored in sterile bottles in the refrigerator after filtration through a Seitz filter. Such ferritin preparations have a low cadmium content and are quite stable. The bulk of iron in ferritin appears to exist in the form of colloidal ferric hydroxide, since the visible absorption spectrum is identical at equivalent concentrations of iron with that of solutions of colloidal ferric hydroxide. The iron is tightly held by the protein but can be removed by prolonged dialysis of ferritin solutions in the presence of sodium hydrosulfite at pH. 4.6 in concentrated acetate buffer and a,a'-dipyridyl. This treatment results in the reduction of ferric to ferrous iron and the removal of the latter by complexing with di- pyridyl. After several such treatments and dialysis against water, the protein is undenatured and essentially colorless. Addition of CdS04 produces crystals of apoferritin, almost entirely free of iron, and identical in form with those obtained from ferritin. Ferritin is The Biochemistry of Ferritin 201 also associated with phosphate which is removed together with the iron during this procedure (Table 2). Granick has assigned (FeOOH)8- Table 2. Elementary Analysis of Ferritin and Apoferritin Total N, Total Fe, Total P, % % % Ferritin 11.0 20.7 1.29 Apoferritin 16.2 0.0 0.05 A. Mazur, I. Litt, and E. Shorr, J. Biol. Client., 187, 473, 1950; analytical data reported on the basis of dry weight. Various preparations of ferritin vary in their ratio of N:Fe:P. (FeOOP03H2) as the formula for the colloidal micelles in ferritin. As a result of measurements of the paramagnetic susceptibility of ferritin as well as of the ferric hydroxide-ferric phosphate prepared from ferritin by alkali treatment, Michaelis 5 reported that the iron atoms in this protein have an orbital arrangement corresponding to 3 unpaired electrons. Ferric iron may also exist with 1 or 5 unpaired electrons. This type of orbital arrangement in ferritin iron makes it unique among biological iron compounds and stresses the highly spe- cific nature of the iron-incorporation reaction which takes place during ferritin biosynthesis, since the iron compounds ingested with food rep- resent all types of iron with respect to paramagnetic susceptibility. It also points to a specific type of iron binding to the protein, the nature of which is still unknown. The ultracentrifugal pattern obtained with ferritin solutions 6 indi- cates that it is a mixture of molecules consisting of approximately 20 to 25% iron-free apoferritin together with a series of ferritin molecules of varying total iron content and presumably in varying states of aggregation. Apoferritin, however, behaves as a single component dur- ing ultracentrifugation with a sedimentation constant corresponding to a molecular weight of 465,000 (horse apoferritin). If considered as an ellipsoid, apoferritin has an axial ratio of 3:1. In contrast to its behavior in the ultracentrifuge, ferritin behaves as a single component on electrophoresis. Apoferritin has mobilities identical with those of ferritin over a range of pH from 4 to 8.6,2 indi- cating that the large quantity of iron in ferritin does not affect the surface charge density of the protein in solution at these pH's. Other evidences of similarities of these two proteins from the point of view of surface properties are: identical viscosities 7 calculated on the basis of nitrogen content, and identical quantitative immunochemical re- 202 Essays in Biochemistry actions of ferritin and apoferritin when added to the antibody directed against either of these proteins.2 These findings, together with direct electron microscopic studies by Farrant 8 appear to offer conclusive evidence that most of the iron lies inside the protein molecule. Little information can be offered concerning the nature of bonding of the iron micelles to the protein. However, differences between ferritin and apoferritin can be demonstrated with regard to ease of denaturation. As a rule globular proteins must be denatured before they can be digested by pepsin or trypsin. This is also true for ferritin and apofeiritin.9 Our studies show, in addition, that ferritin is less easily denatured by acid (at the same pH) than apoferritin (Table 3) Table 3. Effect of pH on Peptic Hydrolysis and Protein Denaturation Ferritin Apoferritin Hemoglobin Hydrol- Dena- Hydrol- Dena- pH ysis turation ysis turation Hydrolysis 1.6 67 69 99 99 96 2.0 16 35 79 47 89 2.5 3 3 17 17 19 3.0 0 0 4 4 18 azur, '. [. Litt, and E Shorr, J. Biol. Chem. 187, 473, 1950. Denaturation determined by degree of insolubility at the isoelectric point. and that the greater the iron content of ferritin the less easily it is denatured and, therefore, digested by pepsin (Table 4). After de- Table 4. Effect of Iron Content on Peptic Hydrolysis of Ferritin Ferritin mS- Fe Per cent Sample mg. N Hydrolysis Original 1.55 11 Fraction A 1.04 27 Fraction C 0.0 52 A. Mazur, I. Litt, and E. Shorr, /. Biol. Chem., 187, 473, 1950. Fractions A and C prepared from original ferritin by partial and complete removal of iron, respectively, by reduction with hydrolsulfite and dialysis. naturation by urea-alkali, both proteins are digested by trypsin. The protective effect of ferritin-bound iron against acid denaturation of the protein may be due to the existence of strong bonds between the iron micelles and groups within the protein which are necessary for maintenance of its native state. The Biochemistry of Ferritin 203 Amino acid analyses of apoferritin and ferritin indicate that no nitro- gen components are present other than amino acids. The content of dicarboxylic and basic amino acids (Table 5) agrees with the observed Table 5. Amino Acid Nitrogen Distribution in Apoferritin and Ferritin Apoferritin gm. per 100 gm. %of Ferritin, %of Amino Acid protein total N total N Ammonia N 10.0 Humin N 3.4 Glutamic acid 17.2 10.1 9.9 Aspartic acid 6.8 4.4 Lysine 7.8 9.2 Arginine 9.1 18.0 Histidine 4.8 8.0 Cystine 1.7 1.2 Methionine 1.9 1.1 1.0 Tyrosine 5.0 2.4 Phenylalanine 6.1 3.2 3.0 Leucine 19.1 12.6 12.5 Isoleucine 1.4 0.9 Glycine 3.4 3.9 Valine 4.3 3.2 3.1 Alanine 1.9 1.8 Threonine 4.3 3.1 Proline 1.5 1.1 Tryptophan 1.2 1.0 A. Mazur, I. Litt, and E. Shorr, J. Biol. Chem., 187, 473, 1950. Serine has also been shown to be present. Analyses reported above were done in most cases by the microbiological assay technique; several by specific colorimetric methods. The total S was 0.89%, of which 98% was accounted for by the cystine and methionine content. The amino N was 5.0% of the total N. isoelectric point for horse ferritin, 4.4, determined electrophoretically. Crystalline dog ferritin has an isoelectric point of 5.2; that from the human is 5.5. Evidence was obtained from quantitative immuno- chemical studies 2 that these three ferritins are related but not identi- cal; they cross-react, but the heterologous ferritin antigen reacts to a lesser extent than the homologous ferritin with its antibody. However, ferritins from different tissues of the same animal (liver and spleen) appear to be immunochemically identical. Sulfhydryl Groups and Ferritin Activity Our studies attempting to relate the structure of ferritin to its bio- logical activities made use of the rat test for vasodepressor effect10 204 Essays in Biochemistry and the antidiuretic action in hydrated dogs as indices of physiological activity.11 Inactivation of these ferritin activities by aerobic liver slices (rat, rabbit, or dog) and activation by anaerobic liver slices sug- gested the presence of groups in ferritin capable of undergoing revers- ible oxidation-reduction. Inactive ferritin could be activated by treat- ment with cysteine or reduced glutathione, whereas active ferritin was inactivated by treatment with iodoacetamide, o-iodosobenzoate or p- chloromercuribenzoate, all sulfhydryl-reacting reagents. These data appeared to fit the hypothesis that sulfhydryl groups were involved in these biological activities of ferritin. Chemical estimation of sulfhydryl groups yielded elevated values for active ferritin and decreased values for inactive ferritin. The choice of a method for measurement of SH content was a difficult problem because of the intense color of the protein. Amperometric titration with Ag+ or Hg++ gave poor results with ferritin as it does with some proteins. The method of Rosner using iodoacetic acid was modified so as to use iodoacetamide followed by precipitation of the ferritin with trichloracetic acid. The extent of reaction of sulfhydryl groups with iodoacetamide (for a 10-minute period at pH 7.4) was measured in the clear supernatant solution by oxidizing the HI formed, with H2O2, to yield free iodine. The iodine color was read in a photo- colorimeter before and after treatment with thiosulfate. If adequate blanks are used, this method gives reproducible values and is quite specific for sulfhydryl groups since addition of p-chloromercuriben- zoate prior to reaction with iodoacetamide reduces the extent of reac- tion of ferritin with iodoacetamide, under these conditions, to zero. The conclusion, based on the data described above, that sulfhydryl groups were directly related to ferritin activity, was brought into question by an experimental finding contradictory to this hypothesis. When inactive iodoacetamide-treated ferritin is incubated with cys- teine, reduced glutathione or ascorbic acid, the resulting ferritin be- comes quite active. Since the reaction of sulfhydryl groups with iodoacetamide is known to be irreversible, we had to consider the pos- sibility that another group capable of undergoing oxidation-reduction might be responsible for activity of this protein. Iron and Ferritin Activity In our earlier studies, iron was not seriously considered as related to ferritin activity for two reasons: (a) Apoferritin, essentially free of iron, is as active as ferritin con- taining 23% iron, when tested in the rat for vasodepressor activity The Biochemistry of Ferritin 205 and when administered intravenously to the dog for study of its anti- diuretic effect. (b) Ferritin can be separated into a number of fractions which have varying total iron: total nitrogen ratios, all of which, however, are equally active on the basis of nitrogen content. The first of these results could be explained without eliminating iron as a participant if it could be shown that apoferritin, on intra- venous injection, combines with iron in the plasma to yield active ferritin. The second could be explained if the various fractions ob- tained by fractionating ferritin could be shown to consist of two forms of iron, one of which was present in these fractions in a constant ratio to the nitrogen content. This second form of iron would fulfill the requirements of oxidation-reduction reactions if it were ionic and therefore capable of existence either in the ferric or ferrous state. Iodoacetamide-treated ferritin, with no measurable sulfhydryl con- tent was incubated with cysteine, ascorbic acid, or reduced glutathione. The reagents were then removed by dialysis and the ferritin analyzed for sulfhydryl groups. No sulfhydryl groups were regenerated by cysteine or ascorbic acid, but a definite though small increase appeared after treatment with glutathione. The new sulfhydryl groups were, however, part of the added glutathione which had been bound to the ferritin molecule. This was established by incubating iodoacetamide ferritin with S35-labeled glutathione followed by extensive dialysis against water. The radioactivity associated with the ferritin was found to be approximately equal to the sulfhydryl content as measured chemically by reaction with iodoacetamide and could be removed by treatment with trichloracetic acid. These results make it unlikely that ferritin activity is directly concerned with sulfhydryl groups. Active ferritin solutions were now treated with a,a'-dipyridyl and the ferritin precipitated by addition of an equal volume of saturated ammonium sulfate. The clear protein-free supernatant solution had the typical pink color of the ferrous-dipyridyl complex even when the reaction was carried out at pH 7.4. At a constant concentration of dipyridyl and varying concentrations of ferritin the data showed that, as the ferritin was diluted, a larger fraction of its total iron was bound by dipyridyl in the form of ferrous iron. At a constant concentration of both dipyridyl and ferritin the quantity of ferrous iron bound by dipyridyl increased with decrease in pH. These results were consistent with the hypothesis of a competition between dipyridyl and ferritin for its ferrous iron, which was therefore capable of some dissociation and probably at or near the surface of the protein. 206 Essays in Biochemistry Although this method does not yield data for the absolute amount of ferrous iron in ferritin it was found useful for comparative purposes. From this data it can be calculated that active ferritin can contain at least 0.2% of its total iron in the form of ferrous iron. Repetition of our earlier experiments to include ferrous iron analyses indicated that whenever sulfhydryl groups increased ferrous iron also increased, and vice versa (Table 6). Further, iodoacetamide ferritin Table 6. Effect of Liver Slices on Sulfhydryl Groups and Ferrous Iron of Ferritin SH Fe++ Ferritin Treatment yuM per 100 mg. ferritin N Original ferritin 25 . 0 1.7 (a) Ferritin + liver slices in N2 33 . 7 6.5 (b) (a) + liver slices in 02 14.9 2.3 A. Mazur, S. Baez, and E. Shorr, J. Biol. Chem.. March. 1955. after treatment with cysteine, ascorbic acid, or glutathione showed marked increases in ferrous iron content (Table 7 and 8). Finally, Table 7. Effect of Reducing Agents on Ferrous Iron of Ferritin Ferrous Iron Treatment Original ferritin Ferritin + GSH Ferritin + ascorbate Ferritin + cysteine Total Iron juM per mM 0.86 9.3 27.6 105.0 A. Mazur, S. Baez, and E. Shorr, J. Biol. Chem., March, 1955. Table 8. Effect of Reducing Agents on Iodoacetamide-Treated Ferritin SH Ferrous Iron Total N Total N Treatment (a) Original ferritin (b) Ferritin + iodoacetamide and dialyzed (b) + cysteine (b) + ascorbate (6) + glutathione fjM per 100 16.5 6.5 5.2 5.7 8.7 mg. total N 2.0 1.2 11.1 4.9 8.4 A. Mazur, S. Baez, and E. Shorr, J. Biol. Chem., March, 1955. The Biochemistry of Ferritin 207 after treatment of ferritin with iodoacetamide, o-iodosobenzoate, or p-chloromercuribenzoate, a decrease in ferrous iron content was noted. Since iodoacetamide and p-chloromercuribenzoate are known to react only with sulfhydryl groups and not with iron, the result may be explained if we assume that the function of the sulfhydryl groups in active ferritin is to stabilize the ferrous iron against autoxidation, a reaction which occurs spontaneously when ferrous iron is added to water or to most proteins at neutral pH. Alteration of the sulfhydryl groups would thus lead to autoxidation of ferrous iron and inactivation of ferritin. It should be noted that chelation of metals like ferrous iron or cuprous copper tends to stabilize the lower valence state of these metals. The finding, mentioned previously, of equal vasodepressor activity of ferritin fractions with varying total iron content was now reinvesti- gated. Fractions were prepared from ferritin by serial precipitation using increasing concentrations of ammonium sulfate. In this way a number of fractions were obtained with decreasing ratios of total iron : total nitrogen as the concentration of ammonium sulfate needed to precipitate these fractions increased (Table 9). The sulfhydryl con- Table 9. Relationship of Total Iron, Ferrous Iron, and SH groups in Ferritin Fraetions Total Iron Total N Ferrous Iron Total N SH Ferrous Iron Concentration Of (NH4V2SO4, % of saturation (Original) 0-27 27-31 31-34 34-50 Total N Total Iron (454) 549 454 361 251 ;uM per (0.8) 0.7 0.0 0.6 0.8 mM (4.1) 3.2 3.3 3.4 5.1 (1.7) 1.2 1.4 1.6 3.2 A. Mazur, S. Baez, and E. Shorr, ./. Biol. Chem., March, 1955. tent among these fractions was fairly constant with the exception of that fraction containing the least total iron; it had a higher sulfhydryl content and would correspond to a mixture of molecules relatively rich in apoferritin. In contrast, the ratio of ferrous iron: total nitrogen was constant for all fractions, a result to be expected if ferrous iron were more specifically associated with ferritin activity. The other difficulty with the argument for iron participation in ferritin activity is the equivalent activity of iron-free apoferritin. There is no direct evidence for the reaction of apoferritin with plasma 208 Essays in Biochemistry iron to form a ferrous apoferritin compound. However, some indirect evidence is available to indicate that apoferritin needs to be in contact with plasma before it can become active as an antidiuretic, whereas ferritin does not. A study of its antidiuretic action has shown that ferritin acts by stimulating the neurohypophysis. When injected into the circulation via the femoral vein ferritin is a potent antidiuretic in amounts of 150 to 250 /xg. ferritin 1ST per kg; the same is true of apoferritin. Both soon disappear from the circulation, presumably due to inactivation by the liver. However, when injected directly into the carotid artery in order to reach the neurohypophysis immediately, ferritin was found to be active in amounts of 10 to 30 fxg. ferritin nitrogen per kg., whereas at these concentrations apoferritin was not active at all. This result would be consistent with the apparent inabil- ity of apoferritin to combine with plasma iron in the brief interval between injection and arrival at the site of action. Our data therefore lead us to postulate the following structure for ferritin: a protein, apoferritin, containing varying quantities of col- loidal micelles or clusters of ferric hydroxide-ferric phosphate internally situated and held to the protein by unknown bonds. The iron is prob- ably in equilibrium with surface ionic iron which may exist in the ferric or ferrous state, the state being dependent on the presence of free sulfhydryl groups, which help to chelate the ferrous iron and thus stabilize it p gainst autoxidation. Blocking or oxidation of the sulf- hydryl groups leads to a spontaneous oxidation of ferrous to ferric iron, this change representing a change from a physiologically active to inactive ferritin. The activity of ferritin is primarily due to the presence of stably bound ferrous iron which nevertheless is capable of dissociation for combination with any avid iron-binding agent. Transport of Iron from Liver to Plasma Our earlier experiments with anaerobic liver indicated that all of the ferritin in such a liver exists in the sulfhydryl form. We were interested in determining whether the presence of an hypoxic liver would result in an increase in plasma iron derived from the more easily dissociable hepatic ferritin-ferrous iron. Dogs, subjected to prolonged hypotension by graded hemorrhage to the state of shock, are known to have hypoxic livers; that is, livers in which the ferritin is in the sulfhydryl state and can now be assumed to contain a maxi- mum of ferrous iron at its surface. Plasma samples were withdrawn from a series of such animals and analyzed for total iron and iron- The Biochemistry of Ferritin 209 binding capacity. From these two values the degree of saturation of the plasma iron-binding protein could be calculated. Fourteen dogs were bled by a standard procedure for inducing hemor- rhagic shock. Of these, 9 died within 24 hours after terminating the experiment by retransfusion of all the blood previously withdrawn. Analyses of the last blood sample gave the following values: (a) Total iron: 287 (173-534) % of the original. (b) Iron-binding capacity: 15 (6-25) % of the original. (c) Saturation of iron-binding protein: 90 (80-97) % as contrasted with 20 to 40 % for control values. Of the remaining 5 dogs in this series, one survived beyond 24 hours but died subsequently, with values for the last blood sample similar to those noted above. The remaining 4 dogs which survived showed less marked changes for final plasma samples: (a) Total iron: 171 (142-216) % of original. (6) Iron-binding capacity: 46 (38-58) % of original. (c) Saturation of iron-binding protein: 70 (55-84) %. In contrast to these alterations, changes were barely evident in another series of dogs pretreated with the adrenergic blocking agent Dibenzyline* and then subjected to hemorrhagic hypotension of an equivalent degree and duration. Dogs pretreated with Dibenzyline usually survive this hemorrhagic procedure. There is evidence that Dibenzyline exerts its protective effect by virtue of its blunting action on the intense peripheral vasoconstriction which normally accompanies hemorrhage, thus maintaining a better blood flow through the liver and other splanchnic organs. Six dogs treated in this manner had plasma values for the last sample as follows: (a) Plasma iron: 100 (50-150) % of original. (b) Iron-binding capacity: 86 (54-133) % of original. (c) Saturation of iron-binding protein: 40 (18-67) %. These results confirm the reality in vivo of the more labile ferrous linkage to ferritin and emphasize the fact that in the reduced state ferritin can liberate its iron for passage into the plasma, to be bound by the plasma iron-binding protein. This process was readily demon- strated in vitro by dialysis of partially purified plasma iron-binding protein against several ferritins, each treated in such a way that they contained differing quantities of ferrous iron (Table 10) . The greatest quantity of ferritin iron was transported for binding by the iron-binding * N-Phenoxyisopropyl-N-benzyl-/3-chlorethylamine, manufactured by Smith, Kline and French, Philadelphia. 210 Essays in Biochemistry Table 10. Extent of Binding of Ferritin Ferrous Iron by Plasma Iron-Binding Protein Fe Bound by Plasma Iron-Binding Protein SH Fe++ Content Content (c) /iM/100 mg. ferritin N 25.0 1.7 33.7 6.5 14.9 2.3 Ferritin Treatment aiM/100 mg. ferritin N /iM/100 mg. ferritin N Per cent of (c) Original Ferritin 25.0 1.7 0.8 47 (a) Ferritin + liver slices in N2 33.7 6.5 5.7 88 (6) (a) + liver slices in 02 14.9 2.3 0.2 9 A. Mazur, S. Baez, and E. Shorr, J. Biol. Cliem., March, 1955. protein from that ferritin which had been incubated under anaerobic conditions with rat-liver slices. The least ferritin iron was transported from that ferritin which had been treated aerobically with liver slices. Intermediate binding occurred with an untreated ferritin solution. Mechanism of Iron Transport Since ferritin occurs in the bone marrow we next determined whether marrow could convert ferric ferritin to ferrous ferritin. It was demon- strated that rabbit-bone-marrow suspensions were able under anaerobic conditions to increase the ferrous iron content of ferritin. As a result of these studies we can postulate that all three physiological activities of ferritin — its iron-storage and release property, its vasodepressor activity, and its antidiuretic activity — are related to the same func- tional groups in ferritin and that the same mechanisms operate for their alteration. A scheme is shown in Fig. 1 which attempts to demonstrate this idea. Although this scheme is not novel,12 it does give mechanisms which are substantiated by experimental data for almost all reactions: 1. Inactive ferric disulfide ferritin in the liver is changed to active ferrous sulfhydryl ferritin under hypoxic conditions. Glutathione, which is present in the liver in relatively high concentrations, can perform this reaction. The hypoxia can be mild and local under normal physiological conditions, thus allowing a small amount of iron to be present in the ferrous state. 2. Ferrous iron from reduced ferritin is transferred into the plasma to be bound by the plasma iron-binding protein. This reaction occurs in vivo as an acute response to severe blood loss and is probably a reflection of a less marked transfer of iron under physiological con- ditions. 3. The iron-binding plasma protein is presumed by many workers to contain ferric iron only, although Laurell provides evidence 13 which The Biochemistry of Ferritin 211 makes it likely that some of the iron is in the ferrous state and that the complex is dissociable in vivo. The iron could thus be made avail- able for transport to the bone marrow for storage as ferritin. 4. In response to lowered oxygen tensions, such as occur during anoxia or at high altitudes, some ferritin iron is reduced in the marrow to the ferrous state so that it can combine with protoporphyrin for Intestine (absorption) X Placenta (fetal absorption) H ''.. 7W<+r\ Aerobic Fet^lFe*— fFe °) " ^ H Anaerobic |[- (reduced ferritin) (oxidized ferritin) -H- Fe™^ + Iron-binding protein -«- (GSH?) „* IBP II- -44— ^^^ 7FJ*6\ AerotilC- Fe% Fe*- L Anaerobic //H Fe~"~ + Protoporphyrin » Heme Liver > (storage and release) Plasma (transport) Bone marrow )> (storage and heme synthesis) Fig. 1. Scheme illustrating the participation of ferritin in iron transport. heme synthesis. The latter reaction has been reported by Granick in hemolyzates of chick red cells under anaerobic conditions, after addi- tion of inorganic ferrous iron and protoporphyrin.14 5. The presence of ferritin in the placenta and the presence there as well of lowered oxygen tensions 15 makes this tissue an ideal one for storage of iron for purposes of transfer from maternal to fetal plasma. Ferritin in the placenta would exist in the ferrous state, and the transfer of iron would take place regularly across the placental barrier to the fetal plasma iron-binding protein. This hypothesis is supported by the observation 13 that in the fetus serum iron is normal or slightly increased and iron-binding capacity is low, whereas in the pregnant woman serum iron is somewhat low or normal and iron-binding capac- ity greatly increased. 212 Essays in Biochemistry Biochemical Activity of Ferritin The activity of ferritin in vivo as demonstrated by its "vasodepres- sor" effect suggests the possibility that it, or its iron, might react with endogenous adrenaline in the smooth muscle cells of the precapillary blood vessels and catalyze its oxidative inactivation. In this way the smooth muscle cells would now become less reactive to the topical application of a threshold concentration of exogenous adrenaline, an effect which occurs during the rat mesoappendix assay. Some of the preliminary results obtained from experiments concerned with the interaction of ferritin with adrenaline may be pertinent to the mecha- nism of action of ferritin in vivo. At pH 4.5 ferritin itself has little effect on adrenaline oxidation. The same is true for inorganic Fe++ or Fe+ + + , although the latter forms a colored complex with adrenaline. Such colored complexes are known to form with many o-dihydroxyphenols and their colors vary with pH. Adrenaline forms similar compounds with Fe+ + + which are green at pH 4.5, purple at pH 6.0, and wine red above 7.0. The different colors are due to varying ratios of Fe+ + + :phenol in the complex. On the basis of reports with other o-dihydroxyphenols it is likely that at pH 4.5 the ratio of Fe+ + + : phenol is 1:1 and that at pH 7.4 it is 1:2 or 1:3. At pH 4.5, in acetate buffer and in the presence of ferritin, the addi- tion of H202 results in a catalytic oxidation of adrenaline to a series of colored compounds, of which the N-methyl indole quinone, adreno- chrome, can be recognized. Adrenochrome then undergoes further oxidative changes which result in the formation of melaninlike pig- ments, a conversion which is relatively slow at acid pH. The addition of H202 to inorganic Fe++ also results in a catalytic oxidation of adrenaline to adrenochrome. Fe+ + + has a lower initial activity but does act eventually since it is reduced to Fe++ by the H202 present. These can best be interpreted in terms of the action of Fenton's reagent (Fe++ and H202) which produces the free radicals OH and 02H which oxidize adrenaline. A similar oxidation of adrenaline takes place in the presence of high-energy X rays which are presumed to act via the formation of similar free radicals. The pseudoperoxidase action of ferritin at acid pH cannot be used to suggest a mechanism for adrenaline oxidation in vivo. However, at a pH of 7.4, ferritin itself catalyzes the oxidation of adrenaline without the addition of H202. At this pH, adrenochrome which is formed is rapidly converted to the melanins, a reaction which is not The Biochemistry of Ferritin 213 affected by ferritin. The major difference between the reactions at 4.5 and 7.4 is the fact that, whereas at the more acid pH inorganic Fe+ + + is less active than Fe+ + , at pH 7.4 Fe+ + + is more active than Fe++ in catalyzing the oxidation of adrenaline. The activity of inorganic Fe+ + is due to its autoxidation, at pH 7.4, to Fe+ + +. One should keep in mind that the addition of inorganic iron salts to solutions at pH 7.4 would ordinarily produce insoluble and highly un- dissociated hydroxides were it not for the presence in the above systems of adrenaline, which forms soluble complexes with these ions. Ferritin, of course, serves the very useful purpose of carrying both Fe++ and Fe+ + + at pH 7.4 in a soluble and reactive state. A curious problem now arises with respect to (a) the biological activities of ferritin and (fc>) its adrenaline oxidation activity, since in (a) it is the ferrous iron which is associated with activity and in (b) it is the Fe+ + + which appears to be responsible for its activity. It is possible however to suggest an hypothesis for the behavior of ferritin in vivo which would satisfy both of these findings. The available evidence concerning the passage of iron across a cell- wall barrier — whether it be from intestine across the intestinal mucosa for iron absorption, from placenta across the placental membrane into fetal blood, or from liver ferritin stores across the liver cell wall into the plasma for transport — requires that iron be present in the ferrous state. Since the biologically active form of ferritin contains Fe++ in a dissociable state, these facts may hold the clue to the problem: Ferritin which is circulating in the plasma is active because it carries ferrous iron capable of passing across the muscle cell wall into its interior. Once inside the cell the Fe+ + would be oxidized to Fe+ + + , complex with adrenaline, and bring about its oxidation. The reaction sequence for the oxidation of adrenaline by iron derived from ferritin may be expressed along the lines suggested by the work of Nelson and Dawson16 on the oxidation of catechol by tyrosinase: HO— r-^N— CHOH-CH2 ° 0= HO^J I Th^ 0= — CHOHCHo 0H NH I CH3 214 Essays in Biochemistry 0=r"^— CHOHCH2 . 0=^ >— 0=1 J-OH CH3 Adrenochronie The end result of these reactions is the catalytic destruction of the physiological activity of adrenaline, brought about by the appearance in the circulation of the sulfhydryl ferrous form of ferritin in very small quantities. In addition, during the state of irreversible hemor- rhagic shock, because of the change within the liver cell of disulfide ferric ferritin to the reduced state, there is produced an increased quantity of ferrous iron available for transfer into the circulation. Such extra iron tends to saturate the iron-binding protein of the plasma and decreases its effectiveness in reacting with the ferrous iron which is being carried by circulating ferritin. Thus, ferritin is portrayed as the carrier, by virtue of its surface sulfhydryl groups, of the iron which, after penetration of the smooth muscle cell wall, will inactivate adrenaline. Finally, it is now possible to suggest a reason why the injection of relatively large doses of active ferritin does not bring about a propor- tionally greater effect on the precapillary blood vessels, an experimen- tal finding which has troubled us for some time. Just as there exists a "mucosal block" that sets a limit to the extent of iron absorption through the intestinal wall, there may also exist a similar block to the transfer of ferrous iron across the smooth muscle cell wall under normal physiological conditions. By this method a limit is soon reached to the intensity of the biological effect of circulating ferritin. Should the permeability of the cell wall be damaged (under patho- logical conditions), this block to iron transfer would be removed, with consequent deleterious effects on the cell. Gould this be the state in irreversible hemorrhagic shock? The experimental facts and hypotheses which have been presented serve to emphasize that the biological actions of ferritin — its vaso- depressor effect, its antidiuretic effect, and its iron-storage and iron- release properties — are affected by changes in oxygen tension which may be local and operative under normal conditions or may be general and acute under pathological conditions. In addition the biological effects which this molecule can exert may be limited by a factor such as permeability. In this respect, ferritin may be regarded as part of The Biochemistry of Ferritin 215 a homeostatic mechanism which can perform useful biological work. These findings also point to the fact that proteins which are usually thought of as relatively inert storage compounds may have wider physiological function. In the present case, all three activities are based on the peculiar ability of ferritin to carry iron in a variety of ways. This paper has presented some biochemical aspects of a broader study being carried out under the direction of Dr. Ephraim Shorr at Cornell University Medical College and the New York Hospital. Cur- rent contributors to this work are Drs. Silvio Baez and Saul Green. It has been aided by grants from The Josiah Macy, Jr., Foundation, Eli Lilly & Co., The National Institutes of Health, U. S. Public Health Service (Grant H-79), The Armour Laboratories, The Office of the Surgeon General, Department of the Army (Contract DA-49-007-MD- 388), and The Postley Fund. References 1. S. Granick, J. Biol Chem., 164, 737 (1946). 2. A. Mazur and E. Shorr, /. Biol. Chem., 182, 607 (1950). 3. S. Granick, and L. Michaelis, J. Biol. Chem., 147, 91 (1943). 4. V. Laufberger, Bull. soc. chim. biol., 19, 1575 (1937). 5. L. Michaelis, C. D. Coryell, and S. Granick, J. Biol. Chem., 14S, 463 (1943). 6. A. Rothen, J. Biol. Chem., 152, 679 (1944). 7. A. Mazur and E. Shorr, J. Biol. Chem., 176, 771 (1948). 8. J. L. Farrant, Biochim. et Biophys. Acta, 13, 569 (1954). 9. A. Mazur, I. Litt, and E. Shorr, /. Biol. Chem., 187, 473 (1950). 10. B. W. Zweifach, in V. R. Potter, Methods in Medical Research, I, p. 131, The Year Book Publishers, Chicago, 1948. 11. S. Baez. A. Mazur, and E. Shorr, Am. J. Physiol, 162, 198 (1950). 12. S. Granick, Chem. Revs., 88, 379 (1946). 13. C. B. Laurell, Acta Physiol Scand., 14, suppl. 46, 1 (1947). 14. S. Granick, Federation Proc, 13, 219 (1954). 15. J. Walker, and E. P. N. Turnbull, Lancet, 265, 312 (1953). 16. J. M. Nelson and C. R. Dawson, Advances in Enzymol, 4, 99 (1944). Some Aspects of Nitrogen Transfer in Biosynthetic Mechanisms SAHAH RATHER The origin of urinaiy urea has been one of the oldest concerns of biochemistry as well as a major problem of nitrogen metabolism. The attention it has received stems from the biochemist's great interest in mammalian, and particularly human, physiology and pathology. Dur- ing the last few decades, the novel concepts of Krebs were responsible for bringing the whole problem much closer to solution. By visualizing the formation of urea as a cyclic phenomenon, carried out through the agency of intermediates acting as nitrogen carriers, he opened the experimental approach to urea formation as a process of cellular me- tabolism. In the past few years our detailed understanding of the reactions which comprise the ornithine cycle of Krebs, along with general ad- vances in our knowledge of intermediary metabolism, have caused our attention to turn from the significance of these reactions in the forma- tion of urea to their significance in the formation of arginine. The individual reactions which lead to the formation of citrulline and arginine now hold considerable interest. It is growing more evident that these reactions are engaged in synthetic activities far beyond their connection with the formation of two amino acids. A broader meta- bolic network can be seen in which urea production assumes a place as but one aspect of arginine metabolism. Urea is formed by a cir- cuitous pathway, and it is only through the action of arginase that two nitrogens can be conveniently removed as urea if such a pathway is to be followed. The entire synthetic course, up to the arginase step, is of general use metabolically. From this point of view, it becomes economical for the living organism to employ elaborate mechanisms for the formation of urea when they serve more than one function. Scheme 1 represents our present concepts of the various ramifications of arginine metabolism with the original participants of the ornithine 216 Nitrogen Transfer in Biosynthetic Mechanisms 217 CO- + NH3 Ornitkim I 1 I Aspartic acid Urea Carbamylphosphate Citrulline ArgininosuccLnic Arginine Arginine phosphate T 4-. Aspartic acid Carbainylaspartic acid I' 1 i Carbainylaspartic Guanidino acid acetic acid I Pyrimidines i Pyrimidines Scheme 1 I Creatine Proteins — Glycine Creatine phosphate cycle shown in italics. Before pursuing the various relationships, it is well to examine the reaction mechanisms through which they are inter- linked. In 1932, Krebs found that the respiring liver slice can form large amounts of urea from C02 and NH3 in the presence of a small amount of ornithine. As an explanation of his observations, he proposed 1 the three-step cyclic mechanism shown in scheme 2. 0=C / \ NH, NH3 I Step 2 -> H*N— C / NH C02 + NH3 NH I R Step 1 NH2 NH2CO~NHR + H3P04 NH2CO— 0~PO(OH)2 + ADP -» NH2COOH + ADP~PO(OH)2 Conversion of Citrulline to Arginine ATP is also required in the formation of arginine. The generation of ATP accompanies the oxidative processes of respiration and, indeed, it was the recognition that phosphate-bond energy is the driving force of these reactions that finally permitted them to be made experimen- tally independent of oxidative metabolism.4,5,7 220 Essays in Biochemistry Arginine formation can be resolved into two enzymatically distinct steps. The first of these, reaction 3, involves the condensation of citrul- line with aspartic acid to form the relatively stable intermediate NH COOH II I C— OH + NH2CH + ATP -> I I NH CH2 (CH2)3 COOH sTH2CHCOOH Citrulline Aspartic acid NH COOH C-NH— CH NH CH2 I (CH2)3 COOH NH2CHCOOH Argininosuccinic acid + ADP + H3PO4 (3) argininosuccinic acid. This is followed, as shown in reaction 4, by a cleavage which liberates arginine by detaching fumaric acid. NH COOH NH COOH C NH CH NH CH2 C— NH, CH I + II NH CH (4) (CH2)3 NH2CHCOOH Argininosuccinic acid COOH (CH2)3 NH2CHCOOH Arginine COOH Fumaric acid The nitrogen atom acquired by citrulline can be donated only by aspartic acid and not by NH3 or glutamic acid, as the earlier experi- ments with slices and homogenates appeared to indicate, nor by any other amino acid. Argininosuccinic acid is a guanidine derivative, and the reaction by which it is formed is depicted as a condensation between aspartic acid and the tautomeric isourea form of citrulline, analogous to the chemi- cal synthesis of guanidines from amines and S-methylisothiourea or O-methylisourea. The structure of argininosuccinic acid has been sub- Nitrogen Transfer in Biosynthetic Mechanisms 221 stantiated in considerable detail, and the compound exhibits the chem- ical properties to be expected of a substituted guanidine. Though much more stable than the intermediate encountered in reaction lb, when subjected to heat or dilute acid it rapidly undergoes non- enzymatic conversion to a cyclic anhydride (amidine N to carboxyl C). The latter can be converted back to argininosuccinic acid only by exposure to dilute alkali. The two compounds behave like creatine and creatinine in this respect and in solution tend to form equilibrium mixtures governed by pH and temperature. The anhydride appears to be metabolically inert.18,19 Reaction 4 merely serves to convert a disubstituted guanidine to a monosubstituted one. The conversion is reversible and argininosuccinic acid can be readily formed from arginine and fumaric acid by this reaction. It involves but a small net change in free energy. The formation and cleavage of this type of C-N bond cannot therefore account for any significant part of the energy required for the synthesis of urea. The energy utilization is confined to the C-N bond estab- lished in reaction 3. The transformation of the ureide group to the guanidine level represents the actual synthesis of the amidine group. Although phosphate-bond energy is utilized in the condensation and orthophosphate is thereby liberated, it has not been possible to detect a free, phosphorylated intermediate formed prior to condensation. It may, of course, be formed transiently on the enzyme surface, or the equilibrium of the reaction may be too unfavorable to allow detection. For the present, the precise manner of phosphate-bond utilization is a matter of conjecture, and the analogy to chemical guanidination, mentioned above, is the basis for the hypothesis which now appears most attractive. The isourea configuration, represented below in the citrulline structure II, is the more reactive of two tautomeric forms. HoN H— N " \ \ C=0 ;=± C— OH / / H— N H— N Hi R, i ii -Ph H— N OH IT— N ^. / Aspartic ^. C— O— P=0 — » C— NHR9 / \ add / H— N OH H— N Ri Ri III IV 222 Essays in Biochemistry- Removal of this tautomer by phosphorylation would tend to displace an unfavorable equilibrium mixture of citrulline tautomers in the direction favorable to condensation. The energy of the pyrophosphate bond is perhaps utilized just to promote this tautomerization. Whether or not condensation can be further broken down to partial steps, reaction 3 is at present the only one of the four reactions that is not detectably reversible, although a number of procedures designed to detect reversibility have been carried out.20 Nitrogen Transfer in the Ornithine Cycle A condensed representation of the ornithine cycle, as it appears at present, is given in scheme 3. Compound X is replaced by carbamyl phosphate for convenience. Mention has already been made of the highly provisional status of phosphorylated citrulline. The individual o=c NH2 HoNC- ± ~ph -O— P(OH)2 < > H.NCOOH <-» NH3 + C02 NH2 -> RNH2 ±H3P04 / -> 0=C \ NHj NH I R NH H H NH "*" NH f I I S +~ph // H2NC < HOOC— C— N— C < HOC \ t I -H3PO4 \ NH 1 HOOC— CH NH f NH I HC— COOH I I HCNH2COOH | R II H R I R HC— COOH HC— COOH I H Scheme 3 (RNH2 = ornithine) steps include (1) the formation of carbamic acid, (2) the formation of carbamyl phosphate, (3) the transfer of the carbamyl group to the terminal nitrogen of ornithine to form citrulline, (4) the condensation of citrulline with aspartic acid to form argininosuccinic acid, (5) the removal of fumaric acid to form arginine, (6) the hydrolytic cleavage of arginine to form urea and ornithine. It may be seen that each transformation involves a different mechanism of C-N attachment or cleavage. The incorporation of these steps into a cycle by the action of arginase represents, in view of our present knowledge, a diversion from their more general functions toward a special requirement devel- oped by the mammalian liver in connection with nitrogen excretion. Nitrogen Transfer in Biosynthetic Mechanisms 223 Synthetic Mechanisms Associated with Arginine Formation As familiarity with these mechanisms increases, it appears that the stepwise manner in which "active" groupings are built up and trans- ferred not only adapts them to the utilization of a common supply of energy but also allows their participation in the synthesis of many compounds that possess the ureide, amidine, or guanidine structure. Arginine Synthesis in General The interest in arginine synthesis has in the past been primarily confined to the ornithine cycle. It is perhaps as a result of gaps in our knowledge of nitrogen metabolism that the significance of these mechanisms in providing arginine for cellular protein has been some- what neglected. The important implications of the isotope experi- ments carried out by Schoenheimer and his group with respect to the rates of protein synthesis and the origin of the incorporated amino acids are too well known to require discussion here.21 The incorpora- tion of N15 derived from NH3 or amino acids into the amidine group of tissue arginine is in accord with the operation of an ornithine cycle. Evidence of this type, moreover, obtained by "trapping" arginine in tissue proteins, offers the additional demonstration that the same arginine- forming mechanisms are drawn upon to supply the needs of protein synthesis. Many unicellular organisms that lack arginase are able to synthesize arginine, and, wherever it has been investigated, citrulline invariably appears to lie in the pathway of arginine synthesis from ornithine. Specific enzymes catalyzing each of the individual reactions have al- ready been detected in a number of these organisms, thus providing more detailed evidence that the same group of arginine- forming mecha- nisms which operate in the mammalian liver are widely distributed in nature. Carhamyl Group Transfer Just as the carbamyl group of compound X or carbamyl phosphate can be transferred to ornithine, there is evidence that this group can also be transferred to aspartic acid to form carbamyl aspartic acid, as shown in reaction 5. Here, as in scheme 3, compound X has been omitted for convenience. The reaction occurs in mammalian liver preparations with compound X, and in bacteria with carbamyl phos- phate.15-22-23 224 Essays in Biochemistry O O COOH H2N— C— O— P(OH)2 + H2N— CH -> I HC— COOH H Carbamyl phosphate Aspartic acid 0 COOH :2N C N CH + H3PO4 (5) H HC— COOH I H Carbamyl aspartic acid Considerable interest attaches to this reaction, for carbamyl aspartic acid (ureidosuccinic acid) can function as a precursor of pyrimidines, through orotic acid, according to scheme 4. Nil, COOH I " I 0=C CH2 I I UN CH I COOH Ureidosuccinic acid NH— C=0 I I 0=C CH9 I I HN CH I COOH Dihydroorotic acid NH— C=0 I I 0=C CH HN C COOH Orotic acid NH— C=0 0=C CH I II HN CH . I ribose — PO4 Uridylic acid Scheme 4 There is ample evidence for the mammalian and bacterial incorpora- tion of orotic acid into the pyrimidines of nucleic acid,24-26 and evi- dence has been obtained for the steps represented in scheme 4.27'28 According to this pathway, the cyclic ureido configuration, as well as the carbon skeleton incorporated in the pyrimidine structure, will have originated in carbamyl phosphate and aspartic acid. Nitrogen Transfer in Biosynthetic Mechanisms 225 The possibility exists that argininosuccinic acid may also lie in the pathway to pyrimidines since hydrolytic cleavage can theoretically occur so as to form ornithine and ureidosuccinic acid. Amidine Group Transfer Arginine assumes an important role in the formation of guanidino- acetic acid (glycocyamine) and creatine.29,30 The former compound is synthesized in the kidney and is subsequently converted to creatine by a methylation which occurs in the liver. Guanidinoacetic acid is formed by the interaction of arginine and glycine, according to reac- tion 6. The process represented in this reaction involves the transfer of the amidine group (originally formed in reaction 3) from arginine to glycine. The transfer proves to be reversible.31-32 H— N H \ I C— NHR + HN— C— COOH ^± / I I H2N H H Arginine Glycine H— N H \ I C— N— C— COOH + NHoR (6) / I I H2N H H Guanidino acetic Ornithine acid Aspartic Acid and Conversion of Inosinic Acid to Adenylic Acid The 6-NHo group of adenylic acid is replaced in intact animals more rapidly than the nitrogens of the ring.33 Although it is known that the deamination of adenylic acid to inosinic acid occurs hydrolytically and irreversibly, the mechanism of amination has been obscure. Both of the nucleotides are active in metabolism as the mono-, di-, and triphosphates, for inosine triphosphate, like ATP, acts as a donor in systems requiring high-energy phosphate. Progress in solving this problem has come from the isolation of adenylosuccinic acid, a new nucleotide that may prove to be an inter- mediate in the amination process. The compound is formed, reversibly, from fumaric and adenylic acid, as shown in the second part of reaction 7, by an enzyme found in yeast.34 Structurally, adenylosuccinic acid resembles argininosuccinic acid, for the 6-NHL> group of purines can be looked upon as part of a cyclic amidine configuration. The similarities 226 Essays in Biochemistry COOH OH 1 HN— CHCH2COOH 1 1 C / \ N C— R C / \ atp N C— R 1 II H— C C— R \ / N 1 II ^ arid1" H— C C— R \ / N Inosinic acid Adenylosuccinic acid NH2 i C / \ N C— R 1 II H— C C— R \ / N COOH 1 CH + 11 CH 1 COOH Adenylic acid Fumaric acid (7) in structure and in enzymatic detachment of fumaric acid suggest that a prior step occurs, similar to reaction 3, in which adenylosuccinic acid is formed by condensation of inosinic acid with aspartic acid. The mechanisms of group synthesis or transfer discussed in the preceding sections all function in the formation of compounds which possess the ureicle, guanidine, or amidine configuration (cf. scheme 1). It is interesting to find that the chemical lability associated with these groupings has a counterpart in their manifold metabolic activities and in what might be called their metabolic lability. Ammonia and Aspartic Acid When nitrogen excretion was intensively examined as a problem of evolutionary development and survival, urea formation came to be seen as a means of converting NH3 into a form which is relatively innocuous and easily excreted, well suited to the physiological limita- tions of many terrestrial forms of life. Our attention was drawn to this point of view before knowledge of nitrogen metabolism had been extended to its present scope and complexity. Most of the nitrogen is, of course, derived from amino acids, and it had been supposed that the amino acids were directly broken down with the liberation of NH3. Nitrogen Transfer in Biosynthetic Mechanisms 227 The whole metabolic burden of maintaining a low concentration of NH3 within the body was thus assigned to the ornithine cycle. This view of amino acid breakdown goes back to the early work of Neubauer, Dakin, and Knoop on the oxidative deamination of amino acids and was strengthened by the observations that many natural and unnatural amino acids can undergo oxidative deamination in liver slices and, later, by the isolation of fiavoproteins that arc capable of catalyzing the interaction with oxygen, as shown in re- action 8. NH2 O R— C— COOH + 02 + H20 -> R— C— COOH + NH3 + H20 (8) I H A second mechanism of deamination was later proposed by Braun- stein, consisting of transamination coupled to the dehydrogenation of glutamic acid (reactions 9 and 10). NH2 R— C— COOH + a-Ketoglutaric acid ^ I H O R— C— COOH + Glutamic acid (9) Glutamic acid + DPN + H20 ^± a-Ketoglutaric acid + NH3 + DPNH + H+ (10) Since both of these reactions are reversible, they can be used in the direction of reductive amination for the synthesis of amino acids. This combination of reactions accounts, in a more satisfactory way than has yet been possible, for the incorporation of N15-labeled amino acids and NH3 into other amino acids isolated from the proteins of intact animals. We now realize that nitrogen metabolism involves as much synthetic activity as catabolic breakdown. A process such as reductive amination can also be concerned with the intracellular removal of NH3. NH3 is an extremely toxic substance and cannot be tolerated by many organisms above a very low blood concentration. Teleologically, the conversion of NH3 to urea has the appearance of a detoxication mechanism and has been frequently interpreted as such. As pointed out in preceding sections, the nitrogen atoms of a number of bodily 228 Essays in Biochemistry- constituents are in part acquired from aspartic acid, including half of the urea nitrogen, one of the pyrimidine nitrogens, and possibly the 6-NH2 group of purines. Their origin in amino nitrogen actually circumvents the formation of NH3. Aspartic acid nitrogen has to come, ultimately, from the general pool of amino acids and can do so directly by two successive transaminations involving glutamic acid, as shown in reactions 11 and 12. Amino acids + a-Ketoglutaric acid ^ a-Keto acids + Glutamic acid (11) Glutamic acid + Oxalacetic acid ^ a-Ketoglutaric acid + Aspartic acid (12) None of these interconversions and transfers has the appearance of a detoxication mechanism, nor are they primarily concerned with pro- moting the excretion of nitrogen, yet they reduce the possibility of undue NH3 accumulation. Many metabolic processes, in effect, col- laborate to this end. As aspartic acid assumes an increasing share in the specific donation of nitrogen, the function of the highly active glutamic-aspartic transamination becomes much easier to understand. This transaminating pair appears as a link in the transfer of nitrogen from amino acids to other nitrogenous constituents of the body through aspartic acid. The formation of glutamine and asparagine, and their ability to undergo transamination with other amino acids,35 must also exert a control on NH3 levels. Unlike the amino acids from which they derive, glutamine, and presumably asparagine, are freely permeable to the liver cell, are present in high concentration in the blood, and undoubt- edly play an important role in the transport of nitrogen to the liver.30 Energetics of NH3 Formation and Nitrogen Transfer From an inspection of reactions 3 and 4, it may be seen that their combined effect is to bring about detachment of nitrogen from aspartic acid, incorporation of nitrogen in arginine, and liberation of fumaric acid. In order to estimate the energy change that occurs when one nitrogen atom is moved from the amino acid pool into urea, by way of aspartic acid and citrulline, the shortest pathway is considered. Of the seven reactions covered in scheme 5, several involve little change in free energy and can be neglected; they are the two trans- aminations, the detachment of fumaric acid from argininosuccinic acid, Nitrogen Transfer in Biosynthetic Mechanisms 229 a- + DPN a - Keto acids -*^ s"+~ Glutamic acid — ^ ^~ Oxalacetic acid * Malic acid Y ±H> Amino acids — ' ^— a - Ketoglutaric <* >^- Aspartic acid acid ±H20 Aspartic acid + Citrulline —+- Argininosuccinic — *- Arginine + Fumaric acid acid Urea + Ornithine Scheme 5 and the conversion of fumaric acid to malic acid. No utilizable energy is made available in the removal of urea. For the remaining two reactions, the DPN-linked oxidation of malic acid to oxalacetic acid permits three high-energy phosphate bonds (~ph) to be gained through phosphorylation coupled to the reoxidation of DPNH, and the synthesis of argininosuccinic acid utilizes l~ph. The net gain of the entire process will be +2-— .ph. It is of interest to compare this with the net gain obtained when another nitrogen atom comes to ornithine from the amino acid pool by way of NH3. Assume, for purposes of discussion, that the route of deamination is the most favorable energetically, and that NH3 is liberated by reactions 9 and 10, which allow energy to be gained from the reoxidation of DPNH. The detachment of nitrogen is then fol- lowed by the generation of 3~ph, and the carbamyl group attachment is associated with the utilization of 1^-ph. The two pathways are thus approximately equal in terms of cell economy, for each incor- porates a DPN-linked step. As far as we know, the energy liberated in flavoprotein catalyzed reactions, which link directly to oxygen, is not made available in a utilizable form. If oxidative liberation of NH3, according to reaction 8, were to precede the ornithine step, it would have the effect of lower- ing the energy gain from +2 to — l~ph. Nitrogen Equilibration In the early investigations of nucleic acid metabolism with N15- labeled nitrogen, the two ring nitrogens of pyrimidines and all but one of the four ring nitrogens of purines appeared to come from NH3. It now seems highly probable that half of the pyrimidine nitrogen comes from aspartic acid, and evidence is accumulating that positions 1, 3. and 9 of purines are derived from aspartic or glutamic acids and from the amide group of glutamine.37 The free amino group of adenylic 230 Essays in Biochemistry- acid, once thought to come from NH3 or amide nitrogen, is probably derived from aspartic acid, and, by analogy, the same may prove to be true of the amino group of cytidylic acid. The role of the nitrogen of aspartic and glutamic acids, and their amides, has been difficult to interpret for a number of reasons, some of which have become apparent during the investigation of arginine synthesis. In contrast to their amides, or to NH3, these two amino acids are relatively impermeable to liver cells and usually give the misleading impression of not being precursors when they are investi- gated in tissue slices or the intact animal. On the other hand, glutamic and aspartic acids can be rapidly synthesized within the respiring cell from NH3 and their respective keto acids (supplied by the operation of the citric acid cycle) through reactions 10 and 12. NH3 thus masks their participation in the donation of nitrogen. Difficulties in detecting a precursor, or in determining the actual pathway which a nitrogen atom has taken, also arise whenever N15-labeled NH3 is compared to glutamic and aspartic acids, or to their amide groups, because rapid isotope equilibration among all the nitrogens takes place through the same reversible reactions in conjunction with reversible amide forma- tion from NH:?. Further experimental difficulties are introduced by the dependence of the synthetic mechanisms on ATP. In respiring systems, just as the citric acid cycle, by facilitating the formation of glutamic and aspartic acids from NH3, tends to obscure the nitrogen source, it can also obscure the energy source by forming ATP. Our present problems of nitrogen metabolism come to us from the pioneer accomplishments of earlier decades, when the enormous syn- thetic potentialities of intracellular metabolism were clearly recognized by investigators who initiated the experimental approach to cellular activity, using surviving tissue or isotopes and the intact animal. Perhaps it will be said of this decade that we are exploring nitrogen metabolism at a higher level of magnification, where the enzyme is both subject and tool. References 1. H. A. Krebs and K. Henseleit, Z. physiol. Chem., 210, 33 (1932). 2. P. P. Cohen and M. Hayano, ./. Biol. Chem,., 166, 239, 251 (1946). 3. P. P. Cohen and M. Hayano, J. Biol. Chem., 172, 405 (1948). 4. S. Ratner, J. Biol. Chem., 170, 761 (1947). 5. S. Ratner and A. Pappas, J. Biol. Chem., 179, 1183, 1199 (1949). 6. S. Grisolia, S. B. Koritz, and P. P. Cohen, /. Biol. Chem., 191, 181 (1951). 7. S. Grisolia and P. P. Cohen, ./. Biol. Chem., 191, 189 (1951). 8. S. Grisolia and P. P. Cohen, ./. Biol. Chem., 19S, 561 (1952). Nitrogen Transfer in Biosynthetic Mechanisms 231 9. S. Grisolia, in Phosphorous Metabolism, I, p. 619, The Johns Hopkins Press, Baltimore, 1951. 10. S. Grisolia and R. O. Marshall, in Amino Acid Metabolism, p. 258, The Johns Hopkins Press, Baltimore, 1955. 11. V. A. Knivett, Biochem. J., 50, xxx (1951) ; 58, 480 (1954). 12. M. Korzenovsky, in Amino Acid Metabolism, p. 309, The Johns Hopkins Press, Baltimore, 1955. 13. H. D. Slade, in Amino Acid Metabolism, p. 321, The Johns Hopkins Press, Baltimore. 1955. 14. M. P. Stulberg and P. D. Boyer, J. Am. Chem. Soc, 76, 5569 (1954). 15. M. E. Jones, L. Spector, and F. Lipmann, J. Am. Chem. Soc, 77, 819 (1955). 16. H. A. Krebs, L. V. Eggleston, and V. A. Knivett, Biochem. J., 59, 185 (1955). 17. S. Ratner, in Adi>ances in Enzymology, XV, p. 319, Interscience, New York, London, 1954. 18. S. Ratner, B. Petrack, and O. Rochovansky, J. Biol. Chem., 204, 95 (1953). 19. S. Ratner, W. P. Anslow, Jr., and B. Petrack, J. Biol. Chem., 204, 115 (1953). 20. S. Ratner and B. Petrack, J. Biol. Chem., 200, 161 (1952). 21. R. Schoenheimer, The Dynamic State of Body Constituents, Harvard Uni- versity Press, Cambridge, Mass., 1942. 22. J. M. Lowenstein and P. P. Cohen, J. Am. Chem. Soc, 76. 5571 (1954). 23. P. Reichard, Acta Chem. Scand., 8, 795 (1954). 24. H. Arvidson, N. A. Eliasson, E. Hammarsten, P. Reichard, H. von Ubisch, and S. Bergstrom, /. Biol. Chem., 179, 169 (1949). 25. L. L. Weed and D. W. Wilson, /. Biol. Chem., 1S9, 435 (1951). 26. L. D. Wright, C. S. Miller, H. R. Skeggs, J. W. Huff, L. L. Weed, and D. W. Wilson, J. Am. Chem. Soc, 73, 1898 (1951). 27. I. Lieberman and A. Romberg, Biochim. et Biophys. Acta, 12, 223 (1953). 28. A. Romberg, I. Lieberman, and E. S. Simms, J. Am. Chem. Soc, 76, 2027 (1954). 29. K. Block and R. Schoenheimer, J. Biol. Chem., 184, 785 (1940) ; 138, 167 (1941). 30. H. Borsook and J. W. Dnbnoff, J. Biol. Chem., 13S, 389 (1941); 169, 247 (1947); 171, 363 (1947). 31. M. Fuld. Federation Proc, 13, 215 (1954). 32. S. Ratner, in Amino Acid Metabolism, p. 231, The Johns Hopkins Press, Baltimore, 1955. 33. H. M. Kalckar and D. Rittenberg, ./. Biol. Chem., 170, 455 (1947). 34. C. E. Carter and L. H. Cohen, /. Am. Chem. Soc, 77, 499 (1955). 35. A. Meister, in Amino Acid Metabolism, p. 3, The Johns Hopkins Press, Baltimore, 1955. 36. H. Waelsch, in Advances in Enzymology, XIII, p. 237, Interscience, New York, London, 1952. 37. J. M. Buchanan, B. Levenberg, J. G. Flaks, J. A. Gladner, in Amino Arid Metabolism, p. 743, The Johns Hopkins Press, Baltimore, 1955. On the Bigness of Enzymes DAVID RITTENBERC The action of enzymes seems to be dependent on two factors: one geometric, the other energetic. The geometric factor is determined by spatial relationships of the substrate and the corresponding enzyme. Although it is beyond our capabilities to alter the basic structure of an enzyme in any significant manner, it is relatively simple to prepare and to test the interaction of a great number of variants of the natural substrate with the enzyme. In this manner it has become clear that specificity is geometrical in nature. It seems at present that all the data can be explained if we assume that the substrate and the enzyme have such a geometrical configuration as to permit the substrate (generally the smaller of the two interacting particles) to approach closely a portion of the enzyme molecule (the active site). Were we able to visualize the relative positions of the individual atoms of the enzyme and of the substrate, we would see a mutual complementarity preserved by non-specific Coulombic and van der Waals forces. The Ogston hypothesis offers a simple explanation for stereochemical specificity.1 The same general concepts have been used by Nachmansohn and Wilson 2 in their synthesis of all the isolated facts concerning the interaction of choline esterase and its variously modified substrates and inhibitors. Neither the substrate nor the products of the enzymatic reaction should be bound too strongly to the enzyme since if either were the reaction would cease owing to poisoning either by the substrate or the products. Indeed, the explanation of the action of many catalytic poisons seems to be the fact that they are strongly bound to the enzyme surface, and that they do not permit access of the natural substrate molecules, e.g., the inhibition of ferrous iron enzymes by carbon mon- oxide. Since all enzymatic reaction takes place in water, the reaction between the substrate and the enzyme should be formulated as: E±-(H20)„ + S ^ E^CHjjOWS + 2/-H20 (1) 232 On the Bigness of Enzymes 233 in which we denote the enzyme by E± and the substrate by S. Not only does the substrate displace water molecules bound to the enzyme, but it also changes the organization of water molecules in the vicinity of the enzyme. The free-energy change of the reaction must be rela- tively small (of the order of RT) since appreciably larger values would lead to a self-poisoning, either by the substrate or by water. It would seem that in those reactions in which the substrate is a charged mole- cule the value of —AH should be large since there would be large interactions of charged groups of the substrate and the enzyme. On the other hand those reactions involving uncharged substrates should have a large value of AS since the substrate should partly destroy the ordered arrangement of water molecules around the polar enzyme. The enzymatic reaction as usually studied is the resultant of at least three reactions: S + E • (H20)n ^ S • E • (H20)„.y + N-VV 2 (3) -Methyl -3(2) - Ethyltartarimide o" ^r ^o H Hematinic acid (C 10, D 10) O^^NT^O H Methylethylmaleimide CH.j - CO - COOH 6 4 5 CH3-CH,-CO-COOH 9 8 3 2 CH, 6 — COOH — 4 + CO, CH,NH, 6 + co2 4 CO 2 6 CH3 - COOH 9 8 CH3NH, 9 + CO, CH,-CH,-COOH 9 8 3 + CO, CO, 9 CH,-CH,NH, 9 8 + CO, 3 Fia. 2. Protoporphyrin degradation. The letters and numbers designate positions of the carbon atoms. 243 244 Essays in Biochemistry positions located are the four methene bridges8,9 and one in each pyrrole8 (Fig. 3). It will be noticed that the carbon atoms in the pyrrole rings, derived from the a-carbon atom of glycine, are in the a-position under the vinyl and propionic acid side chains. This finding supported the suggestion of a common precursor pyrrole first being formed and led to the suggestion that the vinyl side chains arose from propionic acid side chains by decarboxylation and dehydrogenation. CH, CH, II " CH HC*<5 CH, CH., I " CH, I ' COOH H :C*- a 7 :C*- H CH CH, I " CH, I " COOH CH, II CH C'H/3 CH: NH,-C'H,-COOH Glycine -2 -C14 NH, - C*H, - CO - CH, - CH, - COOH 8 - Aminolevulinic Acid - 5 - C ' ' Protoporphyrin IX Fig. 3. The carbon atoms of protoporphyrin which arise from the a-carbon atom of glycine and from the 5-carbon atom of 5-aminolevulinic acid. Having accounted for eight carbon atoms of protoporphyrin, the origin of the remaining twenty-six carbon atoms remained to be deter- mined. It had been found by Bloch and Rittenberg 13 that, on admin- istration of deuterioacetic acid (CD3COOH) to a rat, the hemin isolated contained deuterium. This indicated that some of the side- chain carbon atoms, at least, were derived from the methyl group of acetate, since these are the only carbon atoms bonded to hydrogen. In order to determine the extent of utilization of acetate for por- phyrin synthesis and to locate all the carbon atoms which may be derived from acetate, duck blood was incubated separately with C14- methyl-labeled acetate and with C14-carboxyl-labeled acetate and the resulting C14-labeled-hemin samples degraded by the method men- tioned above. It was found that all the remaining twenty-six carbon atoms were derived from acetate.12 The composite C14-distribution pattern among all the labeled twenty- six carbon atoms derived from acetate is given in Fig. 4. Since all four The Biosynthesis of Porphyrins 245 pyrrole rings had the same C14-distribution pattern, support was ob- tained for the suggestion that a common precursor pyrrole is first formed. Furthermore, since both the methyl side of the pyrrole units and the vinyl propionic acid sides of the pyrrole units had the same C14-distribution pattern, it was concluded that each side of each pyr- role unit is made from the same compound. It would appear that the compound which condenses with glycine to form the pyrrole unit must be either a 3- or 4-carbon-atom compound. On examination of the structure of protoporphyrin and noting the quantitative distribution CH 3 C l4 OOH experiment C " H ., COOH experiment (lb) COOH (H70) (80) COOH @ I I CH2 © V\^VCH2 ® © H3C CH2 ® © H3C^>CH2 © ® feFf ® n 'If'' n H H Fig. 4. Average activities of comparable carbon atoms in all pyrrole units. The activities are given in parentheses. The pyrrole unit represented contains a car- boxyl group which is found only in rings C and D of protoporphyrin. of C14 among the carbon atoms in the experiments, it can be seen that a 3-carbon-atom compound would satisfy the data as the precursor of the methyl sides of the pyrrole units (carbon atoms 6, 4, and 5) and that the same compound would also be consistent with the data as the precursor of the vinyl sides of the pyrrole units (carbon atoms 9, 8, and 3), excluding carbon atom 2, which is derived from the a-carbon atom of glycine. However, it would appear that a 4-carbon-atom com- pound would be necessary as the precursor for the propionic acid sides (carbon atoms 10, 9, 8, and 3), again exclusive of carbon atom 2. If a 3-carbon-atom compound were utilized, subsequent carboxylations must have occurred to give rise to the propionic acid side chains in pyrrole rings C and D. On the other hand, if a 4-carbon-atom com- pound were utilized, decarboxylations must have occurred to give rise to the methyl and vinyl groups. It can be decided which of these two alternative mechanisms operates in the synthesis of protoporphyrin by comparing the data obtained in the experiments using methyl-labeled and carboxyl-labeled acetate. The C14 activities of the carboxyl groups (1170 c.p.m.) in protoporphyrin synthesized from carboxyl-labeled ace- 24G Essays in Biochemistry tate are equal to those found in the carbon atoms (1130) adjacent to these groups in the porphyrin synthesized from methyl-labeled acetate (Fig. 4). This equality, i.e., the same degree of dilution, demonstrates that the acetic acid enters as a unit and that the utilization of acetic acid for pyrrole formation is via a 4-carbon-atom unsymmetrical com- pound. Therefore the common precursor pyrrole originally contained acetic and propionic acid side chains in its (3 positions; the methyl groups in the porphyrin arose by decarboxylation of the acetic acid side chains, and the vinyl groups arose from decarboxylation and de- hydrogenation of propionic acid side chains. The data obtained in these experiments can readily be explained by assuming the participation of the tricarboxylic acid cycle in porphyrin formation. In the light of the relative distribution of the C14 activities among the carbon atoms of the porphyrin derived from acetate, it appeared that the acetate was converted to the 4-carbon unsymmetrical compound via this cycle. The entrance of methyl-labeled acetate in the citric acid cycle can give rise to a 4-carbon-atom compound, derived Table 1. Relative Distribution of C14 Activity in Carbon Atoms of a-Ketoglutaric Acid Resulting from Utilization of C14-Labeled Acetate in Tricarboxylic Acid Cycle * From C14-Methyl-Labeled Acetate (activity of methyl group = 10 c.p.m.) From C14-Carboxyl-Labeled Acetate (activity of carboxyl group = 10 c.p.m.) a-Ketoglutaric Acid Number of Cycles in Tricarboxylic Acid Cycle 1st 2nd 3rd OO 1st 2nd 3rd OO COOH 0 0 0 0 10 10 10 10 CH2 10 10 10 10 0 0 0 0 CH2 1 0 5 7.5 10 0 0 0 0 1 c=o 1 0 5 7.5 10 0 0 0 0 1 COOH 0 0 2.5 5 0 5 5 5 The results are expressed in counts per minute. The Biosynthesis of Porphyrins '247 from a-ketoglutarate, which would have a relative C14-distribution pattern similar to that found in the porphyrin synthesized from methyl- Labeled acetate. For example, if one starts with methyl-labeled acetate with a relative activity of 10 in the methyl group, the a-ketoglutarate formed on the first turn of the cycle would contain C14 activity only in the y-carbon atom and the relative activity would be 10 also (Table 1 i . On formation of symmetrical succinate, the activities of the methylene carbon atoms would be 5 and 5, and those of the oxaloacetate eventu- ally formed would contain half of the C14 activity of the y-carbon atom of a-ketoglutarate. The recycling of this newly formed oxaloacetate (F) (F) Tricarboxylic acid cycle - (F) — +~ a - Ketoglutarate — (B) Succinate - (D) I (C) + Glycine (E) Pyrroles +~ Protoporphyrin Fig. 5. The relationship of the citric acid cycle and protoporphyrin formation. with the labeled acetate would now result in a specimen of a-keto- glutarate having the relative activities shown in Table 1 for the second cycle. In Table 1 the relative activities found in a-ketoglutarate formed from methyl-labeled acetate are given after a number of cycles. It can be seen that a 4-carbon-atom compound arising from a-keto- glutarate after a finite number of cycles would have the same C14- distribution pattern as is found in the 4-carbon-atom unit in the por- phyrin synthesized from methyl-labeled acetate; three adjacent carbon atoms are radioactive, and the one arising from the y-carbon atom has the highest activity. The relationship between the citric acid cycle and porphyrin formation is shown in Fig. 5.12 Direct documentation of these conclusions was obtained by studying the utilization of C14-succinate,14 C14-a-ketoglutarate,15 and C14-labeled citrate 15 for porphyrin formation. In each case the predicted carbon atom of the porphyrin molecule contained C14. For example, a-keto- glutarate-5-C14 and primary carboxyl-labeled citrate produced the same labeling pattern as was found for carboxyl-labeled acetate. The studies with C14-labeled succinate furnished direct evidence for the participation of a 4-carbon compound in porphyrin formation and 248 Essays in Biochemistry also indicated that the succinyl intermediate is formed from succinate as well as from a-ketoglutarate; that is, reaction C occurs (Fig. 5). In Fig. 6 the labeling pattern which should be found in protoporphyrin synthesized from carboxyl-labeled succinate is given. On degradation of protoporphyrin synthesized from carboxyl-labeled succinate, it was C'OOH I CH, I : CH, I ' C'OOH C'OOH I CH, I CH2 C'OR + Glycine C'OOH C'OOH CH2 I I CH2 CH2 CHT C'OR I. ! C'OR CH2 COOH H2 10 C'OOH I 7 C'OOH 9CH, I I ' 6CH, 8CH, COOH * Radioactive carbon atoms positions 7 and 10 9CH, I 10 C'OOH C'OOH 10 Fig. 6. The position of succinate in protoporphyrin and the labeling pattern obtained in protoporphyrin synthesized from succinate-l,4-C14. found that the indicated 10-carbon atoms contained the C14.11 In order to demonstrate that reaction C occurs, a study of the utilization of both raethylene-labeled and carboxyl-labeled succinate for porphyrin formation in the absence and presence of malonate was carried out. Theoretically carboxyl-labeled succinate cannot produce labeled por- phyrin by entering the oxidative pathway of the citric acid cycle (reaction F, Fig. 5), for in this direction the 4-carbon-atom compound formed from a-ketoglutarate would no longer contain any of the orig- inal carboxyl groups of the labeled succinate. Therefore the formation of labeled porphyrin from carboxyl-labeled succinate would be evidence for the occurrence of reaction C. These theoretical considerations and The Biosynthesis of Porphyrins 249 conclusions can be tested experimentally by blocking reaction F with malonate. If these considerations are valid and if reaction C occurs, the degree of labeling found in protoporphyrin from carboxyl-labeled succinate should not be influenced by the presence of malonate. On the other hand, methylene-labeled succinate can produce labeled por- phyrin via reaction F. In this case the degree of labeling from methylene-labeled succinate should be lowered in the presence of malo- nate. It was found experimentally that the C14 activity of the porphy- rin synthesized from carboxyl-labeled succinate in the presence or absence of malonate was the same, whereas that of the porphyrin synthesized from methylene-labeled succinate was 50-6G"/o lower in the presence of malonate than in its absence.14 These experiments demonstrated that the succinate can be converted to "active" succinate via two pathways and that the malonate effect is a reflection of the positions in the succinate which contain C14. It then became of interest to find the mechanism by which the "active" succinate and glycine combine to form the pyrrole unit of the porphyrin. It was realized that in the initial condensation of glycine and succinate the whole molecule of glycine is involved, since in all experiments in which glycine-2-C14 was the substrate the carbon atom in the pyrrole ring and the methene bridge carbon atom (Fig. 3) had the same C14 activity and no derivative of the a-carbon atom of glycine (CH3OH, H2CO, HCOOH, CH3NH2) could substitute for glycine. These findings led us to the conclusion that the same deriva- tive of glycine was utilized for the pyrrole ring carbon atom and for the bridge carbon atom, even though the bridge carbon atom was no longer attached to the nitrogen atom of glycine as is the ring atom. On consideration of the possible method of condensation of succinate and glycine, which would give rise to a product from which a pyrrole could reasonably be made, the suggested mechanism must also account for a method by which the carboxyl group of glycine is detached from its a-carbon atom, subsequent to the initial condensation, for the carboxyl group of glycine is not utilized for porphyrin synthesis. The condensation of succinate on the a-carbon atom of glycine to form a-amino-/?-ketoadipic acid (Fig. 7) would appear to be in agreement with all the experimental findings and conclusions. The compound formed, being a /?-keto acid, could then readily decarboxylate and thus provide a mechanism by which the carboxyl group of glycine is de- tached from its a-carbon atom subsequent to the initial condensation of the whole molecule of glycine with succinate. Further, the product 250 Essays in Biochemistry Tricarboxylic Acid Cycle \ x^. Succinyl X / \f? Co/ Y ^M \ cv-K. Ureido group of purines, formate, etc. the a- carbon atom to [HOOC - CH, - CH2 - C - CHO] II 0 Ketoglutaraldehyde /fOJ HOOC - CH, - CH, - C - COOH " II o - Ketoglutarate Succinate\ I ^s^ + Glycine \ H Succinate - HOOC - CH , - CH 2 - C - C - COOH Glycine 1 1 | Cycle 0 NH2 a-Amino-/3-ketoadipic acid (I) / CO, HOOC - CH2 - CH, - C - CH,NH. * II O 6 -Aminolevulinic acid (II) Porphyrin a-Ketoglutaric acid Fin. 7. Succinate-glycine cycle: a pathway for the metabolism of glycine. COOH I COOH CH2 I I CH, CH, I I CH2 + C = O C = 0 C*H, CH, / | " H2N NH2 5- Aminolevulinic acid (II) + (ID -2H,0 COOH I COOH CH2 I I CH, CH9 Protoporphyrin C*H2 g NH, Precursor pyrrole Fig. 8. The mechanism for the formation of the monopyrrole, porphobilinogen, by condensation of two moles of 5-aminolevulinic acid. The carbon atoms bear- ing the closed circles (•) were originally the a-carbon atom of glycine. The Biosynthesis of Porphyrins 2m of decarboxylation would be 8-aminolevulinic acid, and condensation of two moles of the latter compound, by a Knorr type of condensation (Fig. 8) would give a reasonable mechanism for formation of a pyrrole in which the a-carbon atom of glycine would be distributed in the positions previously observed. To test this hypothesis, S-aminolev- ulinic acid was synthesized and its utilization for porphyrin synthesis studied.16-18 In the initial experiments, unlabeled 8- aminolevulinic acid was added to duck red blood cell hemolyzates along with either C14-labeled glycine or C14-labeled succinate. The radioactivities of the hemin samples iso- lated in these experiments were compared with those obtained from controls in which the unlabeled 8-aminolevulinic acid was omitted. The rationale for these dilution-type experiments is as follows: if 8-amino- levulinic acid is an intermediate formed from the condensation of glycine and succinate, any labeled 8-aminolevulinic acid formed from these labeled substrates will be diluted by the added unlabeled compound, and consequently this should be reflected in the lowered radioactivity of the hemin samples synthesized in the presence of un- labeled 8-aminolevulinic acid. It can be seen from Table 2 that the hemin samples made in the presence of unlabeled 8-aminolevulinic acid Table 2. Comparison of C14 Activities of Hemin Samples Synthesized from Glycine-2-C14 (0.05 mc./mM) or Succinic Aeid-2-C14 (0.05 mc./mM) in the Presence and Absence of Non-radioactive 6-Aminolevulinic Acid Experi- Substrates Isotop e Concentration in Hemin ment C14-Labeled N15-Labeled Unlabeled C14 N15 C.p.m. (Atom ' '( excess) 1 Glycine-2-C14 (0.05 mM) — — 12.3 Glycine-2-C14 (0.05 mM) — 6-Aminolevulinic acid (0.05 mM) 15 2 Glyeine-2-C14 '0.05 mM) — — 230 Glycine-2-C14 (0.05 mM) ^-Aminolevulinic acid (0.05 mM) * — 48 0.21 — Glycine (0.33 mM) * — 0.06 3 Succinate-2-C14 (0.1 mM) — — 660 Succinate-2-C14 (0.1 mM) ^-Aminolevulinic acid (0.1 mM) 180 * The isotopic concentrations of these samples were 34 atom % excess N'°. In each of the experiments the volume of the hemolyzed preparation was 30 ml. Unlabeled succinate (0.1 mM) was added to the flasks in which labeled glycine was the substrate, and unlabeled glycine (0.33 mM) was added to the flasks in which labeled succinate was the substrate. Each flask contained 1 m°. of iron (ferric). 252 Essays in Biochemistry contained less C14 than those of the controls made either from C14- labeled glycine or succinate.17 These results, which are in full agree- ment with the hypothesis, can also be explained, however, by the possi- bility that S-aminolevulinic acid is acting, not as a diluent, but as an inhibitor of heme synthesis. To rule out the latter possibility, the S-aminolevulinic acid added in experiment 2 (Table 2) was labeled with N15. It can be seen that, whereas the incorporation of C14 from the glycine was lowered, there was a comparatively large incorporation of N15 into the porphyrin, thus demonstrating that the lowered C14 activity of the hemin sample was due to dilution rather than inhibition. Further proof that ^-aminolevulinic acid is a result of the condensation of glycine and succinate was obtained by incubating red blood cell hemolyzates with glycine-2-C14 and unlabeled 8-aminolevulinic acid, and subsequently isolating the S-carbon atom. In such an experiment it was found that the formaldehyde liberated upon periodate oxidation of a crude fraction containing S-aminolevulinic acid was highly radio- active. More rigorous proof that S-aminolevulinic acid is indeed the pre- cursor for porphyrin synthesis was obtained by degrading a hemin sample synthesized from S-aminolevulinic acid-5-C14 and from S- aminolevulinic acid-l,4-C14. The S-carbon atom of the former com- pound should label the same carbon atoms of protoporphyrin as those which we have previously found to arise from the a-carbon atom of glycine, since according to the hypothesis the latter carbon atom is the biological source of the S-carbon atom of S-aminolevulinic acid. Furthermore, the S-aminolevulinic acid-l,4-C14 should label the same carbon atoms of protoporphyrin found to arise from the carboxyl Table 3. Distribution of C'4 Activity in Protoporphyrin Synthesized from 8-AminoIevuIinic Acid-5-C14 and from Glycine-2-C14 Molar Activity (%) in Fragments of Porphyrin Synthesized from 5-Aminolevulinic Fragments of Porphyrin Acid-5-C14 Gly cine-2-C14 Protoporphyrin 100 100 Pyrrole rings A + B (methylethylmaleimide) 24.5 24.6 Pyrrole rings C + D (hematinic acid) 25.2 25.3 Pyrrole rings A + B + C + D 49.7 49.9 Methene bridge carbon atoms 50.3 50.1 The Biosynthesis of Porphyrins 253 groups of succinate since i Fig. 7) these carbon atoms arise from succinate. It can be seen from Table 3 that the same C14-distribution pattern was found in protoporphyrin synthesized from S-aminolevulinic acid- 5-C14 as from glycine-2-C14; 50% of the C14 activity resides in the pyrrole rings and 50% in the methene bridge carbon atoms (see Fig. 3).17-18 Also it can be seen from Table 4 that the same (^-distribution pattern was found in protoporphyrin synthesized from 8-aminolevulinic Table 4. Distribution of C" Activity in Protoporphyrin Synthesized from 8-Aminolevulinic A.cid-1,4-C14 and from Siiccinate-1,4-C14 Molar Activity (%) in Fragments of Porphyrin Synthesized from 5- Aminolevulinic Succinic Fragments of Porphyrin Acid-5-C14 Acid-1,4-C14 Protoporphyrin 100 100 Pyrrole rings A + B (methylethylmaleimide) 38.0 39.4 Pyrrole rings C + D (hematinic acid) 61.5 59.5 Pyrrole rings A + B + C + D 99.5 98.5 Carboxyl groups 20.4 20.5 acid-l,4-C14 as from succinate-l,4-C14; ten carbon atoms are equally radioactive, 40% of the C14-activity resides in pyrrole rings A and B, 60% of the activity resides in pyrrole rings C and D, and the carboxyl groups contain 20% of the C14 activity (see Fig. 6).19 Thus all the carbon atoms of protoporphyrin are derived from 8- aminolevulinic acid. The role of 8-aminolevulinic acid in porphyrin synthesis was also actively pursued by Neuberger and Scott,2" and just subsequent to our initial finding they published a confirmatory paper; further confirmation was published by Dresel and Falk.21 Furthermore, it may be well to point out that the theoretical formu- lation of the structure of the precursor pyrrole 16 is the same structure which was determined for porphobilinogen22 by Cookson and Riming- ton,23 a compound excreted in the urine of patients with acute por- phyria. These findings make a-amino-/?-ketoadiptic acid an obligatory intermediate; we have found experimentally that this /3-keto is indeed an intermediate. Injection of 8-aminolevulinic acid or the diethyl ester of a-amino-/?-ketoadipic acid gives rise to the urinary excretion of porphobilinogen.24 254 Essays in Biochemistry The condensation of ''active" succinate and glycine to form 8-amino- levnlinic acid subsequently thus far appears to require the partially intact structure of the red blood cell. It has been found that, whereas 8-aminolevulinic acid can be converted to protoporphyrin in either an homogenized preparation or in a cell-free extract, the conversion of succinate and glycine to porphyrin takes place only with intact cells or with those cells which have been hemolyzed with water.18 Homog- enized preparations obtained in a blender are no longer capable of synthesizing protoporphyrin from succinate and glycine. It would appear that on homogenization the functional activity of only those enzymes of the system that are involved in the condensation of suc- cinate and glycine is lost. However, the finding that 8-aminolevulinic acid can be converted to protoporphyrin in a cell-free extract opened up the possibility that soluble enzymes, concerned with each of the steps in this conversion, could be isolated. Indeed, it was subsequently and independently found in three dif- ferent laboratories that a highly purified protein fraction from ox liver,-5 duck erythrocytes,26 and chicken erythrocytes 27 can convert 8-aminolevulinic acid to porphobilinogen. In our laboratory we ob- tained a highly purified fraction from duck blood which on incubation with 8-aminolevulinic acid-5-C14 produced labeled porphobilinogen. Since the porphobilinogen is presumably synthesized from two moles of 8-aminolevulinic acid (Fig. 8), its molar radioactivity should be twice that of the 8-aminolevulinic acid used as the substrate. The molar radioactivities of the substrate, 8-aminolevulinic acid, and of the product, porphobilinogen, were found to be 242 X 103 c.p.m. and 487 X 103 c.p.m. respectively. This finding demonstrates experimen- tally the utilization of two moles of 8-aminolevulinic acid for porpho- bilinogen formation. Further evidence that porphobilinogen is an intermediate in protoporphyrin synthesis was obtained by incubating equal volumes of the cell-free extract of duck erythrocytes with equi- molar amounts of 8-aminolevulinic acid (0.018 mc./mM) and with the enzymatically synthesized radioactive porphobilinogen (0.036 mc./mM I and subsequently isolating the hemin and determining its radioactivity. The radioactivities of the hemin samples synthesized from 8-amino- levulinic acid and from the porphobilinogen were 92 c.p.m. and 85 c.p.m. respectively, after a 2-hour incubation, and 350 and 336 c.p.m. respectively after a 15-hour incubation period.20 This latter result is in agreement with the findings of Falk, Dresel, and Rimington 2" and of Bogorad and Granick.29 The Biosynthesis of Porphyrins 255 Although no evidence has yet been obtained concerning the bio- logical mechanism of conversion of the monopyrrole to the tetrapyr- role structure, several suggestions have been advanced.-'1" We would like to suggest still another possibility which may explain the distri- bution of the a-carbon atom of glycine or the 8-carbon atom of 8-amino- levulinic acid in the porphyrin molecule of the I and III series. This Ac Ac Ac H B ■CM- NH,C"H., N " H C*H.. N' 'NH. h Ac A NH C*H -NH, Ac P Ac P J I „ I NH. H H JLC. t H H 1 | B S C#H,-|^ NHo y_ ss^C,H2NH2 Ac Ac Fig. 9. A mechanism of porphyrin formation from the monopyrrole. Ac = acetic acid side chain. P = propionic acid side chain. • = a-carbon atom of glycine and 5-carbon atom of 5-aminolevulinic acid. mechanism is based on the synthetic mechanism of dipyrrole and tetrapyrrole formation demonstrated by Corwin and Andrews,31 and by Andrews, Corwin, and Sharp.32 Condensation of three moles of the precursor pyrrole (porphobilino- gen), or of a closely related derivative, would lead to a tripyrryl- methane compound, as schematically represented in Fig. 9. The tri- pyrrylmethane then breaks down into a dipyrrylmethane and a mono- pyrrole. The structure of the clipyrrylmethane is dependent on the place of splitting. An A split would give rise to dipyrrylmethane A, and a B split would give rise to dipyrrylmethane B. Condensation of two moles of dipyrrylmethane A would give rise to a porphyrin of the I series, and condensation of a mole of A and a mole of B would give rise to a porphyrin of the III series. In the formation of the 256 Essays in Biochemistry porphyrin of the III series it can be seen from Fig. 6 that it is neces- sary to lose a 1 -carbon-atom compound since there are three amino- methyl side chains, and only two are required to condense the two dipyrroles to the porphyrin structure. If the mechanism similar to that outlined in Fig. 6 is concerned with porphyrin synthesis, it would appear that this 1-carbon-atom compound given off could well be formaldehyde. Consistent with this idea is our finding 17 that on the conversion of porphobilinogen to porphyrins either by heating under acid conditions 22 or by enzymatic conversion in cell-free extracts 18 formaldehyde was indeed formed. This was established by heating or incubating porphobilinogen, labeled with C14 in the aminomethyl group, and subsequently isolating radioactive formaldehyde as the dimedon derivative. It would appear that, on conversion of porphobilinogen to porphyrins, formaldehyde from the aminomethyl group is formed and that any postulated mechanism should take this into consideration. It is dif- ficult at present to establish the structure of the intermediate tetra- pyrrole compounds which are formed prior to the formation of proto- porphyrin. However, we would like to suggest that these intermediate tetrapyrrole compounds may be the more highly reduced state, con- taining methylene bridge carbon atoms rather than methene bridge carbon atoms, and consequently uroporphyrin and coproporphyrin are oxidized products of the intermediates. The biosynthetic pathway for porphyrin synthesis, given above, may, from a more general viewpoint, be looked upon as merely one aspect of glycine metabolism. The a-carbon atom of glycine besides being utilized for porphyrin synthesis is also known to participate in the synthesis of several other compounds: the ureido groups of purines, the /3-carbon atom of serine, methyl groups, and for formic acid. It would appear that these different compounds and porphyrins may be related via a metabolic pathway of glycine. If indeed these mentioned compounds and porphyrin synthesis are related through a series of reactions occurring with glycine, then an intermediate utilized for porphyrin synthesis may have the same metabolic pattern as is known for glycine. If the succinate-glycine cycle proposed in Fig. 7 16 were the pathway by which all the compounds are related, then specifically the S-carbon atom of 8-aminolevulinic acid should have the same metabolic spectrum as the a-carbon atom of glycine. In a study carried out in ducks and rats it was found that the S-carbon atom of this aminoketone is indeed utilized for the ureido groups of purines, for the /?-carbon atom of serine, for the methyl group of methionine, and The Biosynthesis of Porphyrins 257 is also converted to formic acid. Thus, it has been demonstrated that glycine is metabolized via this pathway.33 The condensation of glycine with "active" succinate provides a path- way whereby glycine can be oxidized to carbon dioxide and the inter- mediates produced in the cycle drawn off for the synthesis of other compounds. This is similar to the citric acid cycle, in which another 2-carbon compound is oxidized to carbon dioxide and intermediates are produced which can be drawn off for synthesis. In the succinate- glycine cycle, succinate is the catalyst instead of oxaloacetate. This work was supported by grants from the National Institutes of Health, United States Public Health Service (RG-1128(C6) ), from the American Cancer Society on the recommendation of the Committee on Growth of the National Research Council, from the Rockefeller Foundation, and from the Williams-Waterman Fund. References 1. D. Shemin and D. Rittenberg, J. Biol. Chem., 159, 567 (1945). 2. D. Shemin and D. Rittenberg;, /. Biol. Chem., 166, 621 (1946). 3. D. Shemin and D. Rittenberg;, J. Biol. Ch< m., 166, 627 (1946). 4. J. Wittenberg and D. Shemin, J. Biol. Chem., 17S, 47 (1949). 5. H. M. Muir and A. Neuberger, Biochem. J., 45, 163 (1949). 6. K. I. Altman, G. W. Casarett, R. E. Masters, T. R. Noonan, and K. Salomon, J. Biol. Chem., 176, 319 (1948). 7. N. S. Radin, D. Rittenberg, and D. Shemin, ./. Biol. Chem.. 184, 745 (1950). 8. J. Wittenberg and D. Shemin, /. Biol. Chem., 1S5, 103 (1950). 9. H. M. Muir and A. Neuberger, Biochem. J., 47, 97 (1950). 10. M. Grinstein, M. D. Kamen, and C. V. Moore, J. Biol. Chem., 174, 767 (1948). 11. D. Shemin, I. M. London, and D. Rittenberg, ./. Biol.Chem.. 173, 799 (1948) ; 183, 757 (1950). 12. D. Shemin and J. Wittenberg, ./. Biol. Chem., 192, 315 (1951). 13. K. Bloch and D. Rittenberg, J. Biol. Chem., 159, 45 (1945). 14. D. Shemin and S. Kumin, ./. Biol. Chem., 198, 827 (1952). 15. J. C. Wriston, Jr., L. Lack, and D. Shemin, Federation Proc, 12, 294 (1953) ; J. Biol. Chem., July, 1955. 16. D. Shemin and C. S. Russell, J. Am. Chem. Soc., 75, 4873 (1953). 17. D. Shemin, C. S. Russell, and T. Abramsky, J. Biol. Chem., in press (1955). 18. D. Shemin, T. Abramsky, and C. S. Russell, J. Am. Chem. Soc, 76, 1204 (1954). 19. E. Schiffman and D. Shemin, unpublished. 20. A. Neuberger and J. J. Scott, Nature, 172, 1093 (1953). 21. E. I. B. Diesel and J. E. Falk, Nature, 172, 1185 (1953). 22. R. G. West all. Nature, 170. 614 (1952). 23. G. H. Cookson and C. Rimington, Nature, 171, 875 (1953). 258 Essays in Biochemistry 24. I. Weliky and D. Shemin, unpublished findings. 25. K. D. Gibson, A. Neuberger, and J. J. Scott, Biochem. J., 58, xli (1954). 26. R. Schmid and D. Shemin, J. Am. Chem. Soc, 77, 506 (1955). 27. S. Granick, Science, 120, 1105 (1954). 28. J. E. Falk, E. I. B. Diesel, and C. Rimington, Nature, 172, 292 (1953). 29. L. Bogorad and S. Granick, Proc. Nat. Acad. Sci., 39, 1176 (1953). 30. G. H. Cookson and C. Rimington, Biochem. J., 57, 476 (1954). 31. A. H. Corwin and J. S. Andrews, /. Am. Chem. Soc, 59, 1973 (1937). 32. J. S. Andrews, A. H. Corwin, and A. G. Sharp, J. Am. Chem. Soc, 72, 491 (1950). 33. C. S. Russell, S. Gatt, G. L. Foster, and D. Shemin, unpublished observation. The Role of Carbohydrates in the Biosynthesis of Aromatic Compounds DAVID B. SPRINSON The diversity and importance of aromatic compounds in biological materials, from simple hydrocarbons to complex alkaloids, have been responsible for several theories of their biogenesis on the basis of struc- tural relationships or known laboratory reactions. Direct experimen- tal attack on this problem began with the discovery by B. D. Davis in 1950 that shikimic acid (SA) is an intermediate in the formation of the aromatic amino acids 1 in certain nutritionally deficient mutants of Escherichia coli. The studies on the formation of tyrosine and phenylalanine from labeled compounds 2~6 do not permit the two sides of the ring to be distinguished from each other, in contrast to those with SA. Accord- ingly, in collaboration with B. D. Davis,7 methods were developed for the isolation of SA from nitrates of E. coli mutant 83-24, and for its chemical degradation (Fig. 1). It had been observed that, when this organism was grown on a glucose-salts medium with the addition of NaHC1403, HC14OONa, acetate, labeled in either carbon atom, or pyruvate-2-C14, the activity incorporated into SA from these additions was negligible. The participation of the tricarboxylic acid cycle inter- mediates in the biosynthesis of tyrosine and phenylalanine in yeast had also been excluded by Gilvarg and Bloch,5 who showed that, al- though labeled acetate, in the presence of glucose, was incorporated into glutamate, aspartate, and alanine, no activity was observed in the aromatic amino acids. When glucose labeled in G-l, G-2, G-3, 4, or G-6 * was utilized for * The abbreviations S-l, S-2 ••• S-7 will be used to denote carbon atoms 1, 2 ••• and carboxyl of sbikimate; and G-l, G-2 •••to denote carbon atoms 1, 2 • • • of glucose. 259 (1) HO 2NaI04 HCOOH Kg" — co2 (S-4) COOH Cu, quinoline 237 ( ■*■ C02 (S-7) (3) COOH COOH COOH (S-l, 7) -*- + CHI3 (S-2) COOH HO-Uu HO NaOI "^-^^OH OH _ 4 NaI04 COOH I C = 0 CH2 CHO Formylpyruvic acid 4NaI04 ■*- 3 HCOOH -*- co2 (S-4, 5, 6) COOH 1 2,4-DNPH C = N-NH-C6H3-(N02)2 " CH2 HC = N-NH-C6H3-(N02)2 (S-7, 1,2,3) (4) HO" HOOC COOCH, COOH NaOH COOH (2,6) HOOC (3,5) (2,6) COOH (3,5) 180° • C02 (S-3,5) co2 - (S-3,5) CH20 NaIO< (S-2, 6) -« + COOH 1 c = o I CH2 £00; H COOH I .OH /< /OH H2C CH2 I COOH Os04 COOH H2C CH2 COOH Itaconic acid Fig. 1. Reactions used for the degradation of labeled shikimic acid. 260 Carbohydrates in Aromatic Compound Biosynthesis 261 SA synthesis it gave the distribution of activities summarized in Fig. 2. The value for each SA carbon atom is represented as the fraction of activity of the labeled atom of the glucose from which the SA had arisen. Except for S-l and S-5, the SA atoms were 80-90% accounted for by large contributions from glucose atoms 1, 2, 3, 4, and 6. In addition there were small incorporations which could not be deter- mined accurately. The large "deficiencies" in S-l and S-5 presumably arise mostly from G-5, the one carbon atom of glucose that could not be tested experimentally. 3 or 4 (0.86) COOH 2 (0.4) 5 (0.5) (0.25) 1 (0.60) 6 (0.22) 2 (0.6) 5 1 (0.4) 6 (0.5) 2 (0.24) 3 or 4 (0.59) 3 or 4 (0.9) Fig. 2. Major contributions of glucose carbon atoms to shikimate biosynthesis. It is clear that S-2 is derived almost entirely from G-l and G-6, S-l from G-2 and G-5, and S-7 from G-3, 4. The G-l and G-6 con- tributions are about equal, as are those from G-2 and G-5, and so it may be assumed that the G-3 and G-4 contributions are also about equal. From this distribution it may be concluded that S-2, 1, 7 is derived from a three-carbon intermediate of glycolysis, which as pointed out above cannot be pyruvate. The derivation of the 3, 4, 5, 6 portion of SA from glucose is more complex. S-6, S-5, and S-4 are derived 0.8-0.9 from G-l, 6, G-2, 5, and G-3, 4, respectively. Since the G-6/G-1 and the G-5/G-2 contri- butions are both present in a ratio of about 2.5/1 it seems reasonable to assume the same ratio for the G-4/G-3 contributions. S-3 arises from G-3, 4 and G-2, also in a 2.5/1 ratio. Such a distribution of label can be explained by assuming that S-3, 4, 5, 6 was derived from tetrose phosphate whose formation had involved, as shown in Figs. 3 and 4, the glycolytic as well as the pentose phosphate pathway. In reaction 1 (Fig. 3) fructose-6-phosphate (F-6-P) (I), derived directly from the glucose administered, exchanges under the influence of transaldolase with triose phosphate which has been equilibrated by triose isomerase. This exchange incorporates G-l, 2, 3 into the "bot- 262 Essays in Biochemistry torn" three carbon atoms of F-6-P, yielding II. In a steady state the proportion of I and II in the pool of F-6-P (III) will depend on the rates of supply of F-6-P (I) and triose phosphate and on the rate of exchange between them. (1) (i) 1C I 2C I 3C 4C (2) 3 = 4C I I 5C + 2 = 5C I I 6 COP 1 = 6 COP 1 = I -> FDP -> 2 = 5 C I 3 = 4 COP (ii) 1C I 2C I TA 3 C I 3 = 4C 4C 2 = 5 C + 5 C 1 = 6 COP 6 COP 6C 14-11 = (III) 1C 1 2C 3C 4 > 3C 5 >2C 6 > 1 COP FDP (IV) 1 = 6C 2 = 5C I 3 = 4 C 3 = 4C 2 = 5 C I 1 = 6 COP I 4- IV = (V) 1 > 6C I 2 > 5C I 3 > 4C I 4 > 3C I 5 >2C 6 > 1 COP Fig. 3. Possible effects of transaldolase (TA) and fructose diphosphatase (FDP) on the distribution of carbon in the fructose-6-phosphate (F-6-P) pool. Alternatively, in reaction 2 (Fig. 3), part of the F-6-P may arise by hydrolysis of fructose diphosphate (FDP), in a fraction of which G-l, 2, 3 and G-6, 5, 4 have been equilibrated through the action of aldolase and triose isomerase. The resulting F-6-P is given by IV and the F-6-P pool by V. If both reactions 1 and 2 occur, a different pool will be formed.* * The enzymes involved in reaction 1 are known to be present in E. coli, but the exchange reaction indicated for transaldolase in this sequence, though expected, Carbohydrates in Aromatic Compound Biosynthesis 263 From the F-6-P pool tetrose phosphate can then he formed by reactions 3, 4, and 5 (Fig. 4), in which pentose phosphate and sedo- heptulose-7-phosphate are intermediates." '- The net effect of these reactions is to convert one molecule of F-6-P and two molecules of 1C 2C 1 c 3C 3C 2C (3) J II 4>3C 3=4C 4>3C 3=4C I I TK 1 | 5>2C +2 = 5C > f>>2C + 2 = 5 C II II 6 > 1 COP 1 = 6 COP 6 > 1 COP 1 = 6 COP 1 C I 1 C 2C 2 C 3 C 3 C I I 1 (4) (a)3 = 4C + 4>3C 1C +4>3C (a) I I TK | | 2 = .5C 5 > 2 C > 2 C 5 > 2 C II II 1 = 6 COP 6 > 1 COP 1 C 6 > 1 COP I 2C I 1C 1C 3=4C 3=4C TK | | 2 C > 2 = 5 C + 2 = 5 C (6) I I I (fe) 3 = 4 C + 3 = 4 C 1=6 COP 1 = 6 COP 2C I 2=5C 2=5C I I 1 = 6 COP 1 = 6 COP 1C 2C 1 C 1C 2C (5) 2 C 2 C 1 C I I I 3 = 4 C 3=4C 3 = 4C 3 = 4C I | TA | | 2 = 5 C +2 = 5C > 2 = 5C +2 = 5C II II 1 = 6 COP 1 = 6 COP 1 = 6 COP 1 = 6 COP Fig. 4. Synthesis of tetrose phosphate via the pentose phosphate pathway from the pooled F-6-P of reaction 1, Fig. 3. triose phosphate to three molecules of tetrose phosphate. In two of these the "bottom" three carbon atoms are derived from the corre- sponding atoms of F-6-P, and in the third tetrose molecule they are derived from the triose phosphate. Iia> not been directly explored. The phosphatase required for reaction 2 has been demonstrated in preliminary experiments on extracts of this strain of E. coli. (P. R. Srinivasan and D. B. Sprinson, unpublished results. We are indebted to Dr. E. Racker for generous supplies of purified FDP and glucose-6-phosphate dehydrogenase required in this assay). 264 Essays in Biochemistry With the F-6-P pool (III) resulting from reaction 1, reactions 3 and 4a each yield a molecule of tetrose of the composition G-3, 4 > 3, 5 > 2, 6 > 1. Reaction 5 would then give rise to a molecule of the composition G-2, 3 = 4, 2 = 5, 1 = 6. Alternatively, the production of two molecules of the first kind of tetrose in reaction 3 can be followed by reactions 4b and 5, again yielding a molecule of the second kind of tetrose. In either case the ratio of G-3 to G-2 in carbon 1 of the pooled tetrose would be 2. The F-6-P pool (V) from reaction 2 would yield through the same process a pair of tetrose phosphates identical with the above in carbon atoms 2 to 4. In carbon atom 1, however, G-3 > 4 would replace G-3 and G-2 > 5 would replace G-2. It is clear that the isotope distribution observed in atoms 4, 5, 6 of SA is consistent with their origin from carbon atoms 2, 3, 4 re- spectively of tetrose phosphate formed as described above. In one- third of the tetrose molecules these atoms are derived from triose phosphate, which would account for a G-l, 2, 3/G-6, 5, 4 ratio of 1/6 in S-6, 5, 4. From the higher ratio observed ( 1/2.5 1, it can be calculated that about two-thirds of the pooled F-6-P was derived from the intact chain of glucose and one-third had incorporated triose phos- phate, possibly by direct exchange (reaction 1) or by hydrolysis of equilibrated FDP (reaction 2).* Similarly, the isotope distribution observed in S-3 (G-3, 4/G-2 in a ratio of 2.5/1) is consistent with the origin of S-3 from carbon atom 1 of tetrose phosphate. In the latter atom a ratio of 2/1 would be expected, as noted above, for the G-3/G-2 contribution derived from "unexchanged" F-6-P (I). The effect of reaction 1 on the F-6-P pool would not influence the isotopic composition of this atom of tetrose, whereas reaction 2 would cause small incorporations of G-4 and G-5. Theoretically the importance of reaction 2 could be evaluated from a precise determination of these incorporations, but such data are not available.! * Let a = fraction of G-l in position 6 of pooled F-6-P (III or V). Assume fraction of G-l in position 3 of pooled triose phosphate = 0.5. Assume also that S-3, 4, 5, 6 is derived from tetrose molecules of which two-thirds obtain then bottom carbon atoms from F-6-P and one-third from triose. Then: ((1-1) 0.25 (0.5 X 1) + 2n Fraction of G-l in S-6 = - = = 0.29 = — — (G-l) + (G-6) 0.85 3 a = 0.185 Fraction of F-6-P pool derived by exchange (reactions 1 or 2, Fig. 3) is 2a, or 0.37. t As a further consequence of reactions 3 to 5 (Fig. 4), the F-6-P regenerated in reaction 5 (which would amount to one-third of the glucose entering these Carbohydrates in Aromatic Compound Biosynthesis 265 A further consequence of these considerations is that attachment between the two intermediates utilized in SA formation involves car- bon 1 of the tetrose and carbon 3 of the triose. As will be pointed out later this does not correspond to the known ways of forming heptoses. A complete series of experiments on the biosynthesis of tyrosine, phenylalanine, the tryptophan from variously labeled glucose is un- available for comparison with the present studies. However, several relevant investigations are known. The incorporation of approximately 0.5 of an atom of G-l into carbons 2, 6 of tyrosine and phenylalanine in yeast 5 agrees with the presence of 0.65 G-l in S-2, 6. In agreement also are the incorpora- tions, in E. coli, of 1.0 to 1.1 atoms of G-6 into carbons 2, 6 of both tyrosine13 and SA. The earlier results of Ehrensvard and collab- orators2 on tyrosine formation from acetate-1-C14 (showing a high incorporation of label into carbons 4 and 3/5 and essentially none elsewhere in the ring) are in agreement too, if it is assumed that this compound is incorporated via glucose-3,4-C14. In contrast, subsequent results of Ehrensvard et al. have indicated that carbon 1 of acetate is extensively incorporated into at least three atoms of the ring of tyrosine,3,4 and into three consecutive atoms of the benzene ring of tryptophan.14 A similar incorporation of G-3, 4 into tryptophan has been observed by Rafelson.15 These results appear to be in conflict with the finding (Fig. 2) that G-3, 4 enters significantly into only two atoms of the ring of SA. The tryptophan data have led to the suggestion15-16 that glucose-3,4-C14 is converted to heptose- 3,4,5-C14,8'12 and that the intact carbon chain of the latter is cyclized to SA-4,5,6-C14. However, utilization of the intact carbon chain of such a heptose or of heptose formed by any known mechanism seems excluded by the results presented above, as well as by enzymatic studies to be discussed later. In order to test various intermediates of the glycolytic as well as the pentose phosphate pathways as precursors of SA, a cell-free test system was developed. In most cases the formation of 5-dehydro- shikimate (DHS) , the precursor of SA,17 rather than SA was studied, reactions) should yield the triose G-l, 2, 1 (or G-l > 6, 2 > 5, 1 > 6) and the tetrose G-l, 3 = 4, 2 = 5, 1 = 6 (or G-l > 6, 3 = 4, 2 = 5, 1=6). From the observed insignificant incorporation of these fragments into SA, it appears that under these experimental conditions the contribution of reactions 3 to 5 to the F-6-P pool is small. It has been assumed throughout that in triose phosphate G-6, 5, 4 = G-l, 2, 3. This is a simplifying assumption, since S-2, 1, 7, which are derived from a three- carbon intermediate of glycolysis, appear to show a small excess of G-6, 5, 4 over 1, 2, 3. 266 Essays in Biochemistry since the reduction 1S of DHS to SA required an additional substrate, isocitrate, and TPN for generating the required TPNH. Extracts of strain 83-24 of E. coli were able to convert hexose phos- phates, ribose-5-phosphate, or sedoheptulose-7-phosphate (S-7-P) to DHS in a yield of about 5%.19 S-7-P plus FDP did not yield significantly higher values. However, sedoheptulose-l,7-diphosphate (SDP) 20 was almost quantitatively converted to DHS.21 When SDP- 4,5,6,7-C14 was incubated under these conditions, the SA obtained had the same activity, and only in S-3, 4, 5, 6, showing that a four-carbon fragment from SDP is utilized directly in shikimate formation.21 Table 1. Synthesis of DHS from E-4-P + PEP and from SDP * | Per Cent Conversion Substrates and Additions E-4-P + 0.3 mM PEP + fluoride + 0.3 fiM PEP + fluoride + 0.5 nM 3-PGA + iodoacetate + 0.3 nM PEP + iodoacetate + 0.5 juM 3-PGA SDP + fluoride + fluoride + 0.5 /xM FDP + fluoride + 0.5 mM 3-PGA + fluoride + 0.5 nM pyruvate + fluoride + 0.3 MM PEP + iodoacetate + iodoacetate + 0.5 juM FDP + iodoacetate + 0.5 nM 3-PGA -f iodoacetate + 0.5 /xM pyruvate + iodoacetate + 0.3 ^M PEP * Cell-free extracts were prepared by subjecting cells of freshly harvested E. coli mutant 83-24 [B. D. Davis, J. Biol. Chem., 191, 315 (1951)] to sonic vibration. The incubation mixtures contained 0.1 ml. of extract (2 mg. of protein), 5 yM of MgCl2, 50 fiM of P04s buffer pH 7.4, 0.25 nM of E-4-P + 0.3 nM of PEP, or 0.25 yM of SDP, + additions (10 mM of KF, or 0.5 mM of iodoacetate) in a final volume of 1 ml. When iodoacetate was added, the solu- tion, 0.95 ml., was preincubated at 37° for 15 minutes prior to the addition of substrate. Following incubation at 37° for the indicated length of time aliquots were removed for the bioassay of DHS with Aerobacter aerogenes mutant A170-143S1 [B. D. Davis and U. Weiss, Arch. exp. Pathol, and Pharmakol, 220, 1 (1953)]. f Abbreviations: E-4-P, D-erythrose-4-phosphate; PEP, phosphoenolpyru- vate; SDP, sedoheptulose-l,7-diphosphate; FDP, fructuse diphosphate; 3-PGA, D-3-phosphogly eerie acid. 1 Hour 2 Hours 88 86 88 88 0 0 90 90 90 90 39 83 0 0 0 0 0 0 0 0 37 80 0 0 0 0 46 83 0 0 46 83 Carbohydrates in Aromatic Compound Biosynthesis 267 The high conversion of SDP to DHS cannot be due to direct cycliza- tion of the seven-carbon chain of SDP. Carbon atoms 7, 1, and 2 of SA were shown to arise from G- (3, 4) - (2, 5) - (1, 6) , respectively. Such a sequence of glucose carbon atoms in carbons 1, 2, and 3 of heptose has never been observed. The reverse order, i.e., G- (1, 6) - (2, 5) - (3, 4) , would be expected in carbon atoms 1, 2, and 3 of heptose.12'20 Since S-2, 1, 7 are derived from a three-carbon product of glycolysis, but not pyruvate, it was surmised that carbons 1, 2, and 3 of SDP are converted to phosphoenolpyruvate and then condensed with tetrose phosphate to a seven-carbon precursor of DHS. In support of this scheme, it was found (Table 1) that the conversion of SDP to DHS was completely inhibited by fluoride and iodoacetate but that it was completely restored by phosphoenolpyruvate.22 Moreover, synthetic D-erythrose-4-phosphate 23 and phosphoenolpyruvate were almost quan- titatively converted to DHS (Table 1). The efficient conversion of SDP to DHS may therefore be explained on the basis of the following reactions: SDP ^ D-Erythrose-4-phosphate + Dihydroxyacetone phosphate 20 Dihydroxyacetone phosphate ^ Phosphoenolpyruvate D-Erythrose-4-phosphate + Phosphoenolpyruvate — » — > DHS Although the details of the reactions are still unknown, it may be postulated at present that the pathway from glucose to DHS is in part as shown in the diagram below. ch2op I c=o HOCH Glycolysis CHOH I CHOH I CH2OP CH2OH I c=o HOCH CHOH I CHOH I CH2OP Reactions 1 to 5 Figs. 3 and 4 COOH I C— O— P II CH2 CHO CHOH I CHOH I CH2OP D-Erythrose- 4-phosphate COOH c=o | COOH CH2 HOCH — > ^ CHOH ^f < CHOH 1 CH2OP OH DHS 2-Keto-3-deoxy- 7-phospho- D-glucoheptonic acid OH It is of interest that several classes of alicyclic compounds, e.g., the carotenoids and steroids, in the biosynthesis of which an isoprene unit appears to be involved, are derived from acetate. On the other hand, 268 Essays in Biochemistry the biosynthesis of the aromatic amino acids, and presumably the alkaloids and flavonoids derived from them,24,25 is dependent on several reactions in the metabolism of carbohydrates which either precede, or are unrelated to, the production of acetate. "Intuitively" and on struc- tural grounds this division was foreseen many years ago.24 It is a pleasure to acknowledge the contributions, intellectual as well as experimental, of the investigators who have participated in this investigation in our laboratories, Drs. H. T. Shigeura, P. R. Srinivasan, M. Sprecher, and M. Katagiri, as well as the stimulating and pleasant collaboration with Dr. B. D. Davis. I am grateful to Dr. D. Rittenberg for his support and encouragement during the early period of this investigation. Without his generosity it could not have been under- taken. We are indebted to Dr. H. S. Isbell for the labeled sugars, and to Dr. G. Ehrensvard for part of the glucose-6-C14 used in the investiga- tion; to Dr. B. L. Horecker for gifts of sedoheptulose-7-phosphate and sedoheptulose-l,7-diphosphate and for an opportunity to prepare the latter compound in his laboratory; to Mr. W. E. Pricer, Jr., for phos- phoenolpyruvate; and to Dr. C. E. Ballou for D-erythrose-4-phosphate. This work was supported by grants from The American Cancer Society (on recommendation of the Committee on Growth of the Na- tional Research Council), the Lederle Laboratories Division of the American Cyanamid Company, the National Institutes of Health, United States Public Health Service, and the Rockefeller Foundation. References 1. B. D. Davis, J. Biol. Chem., 191, 315 (1951); B. D. Davis, in Amino Acid Metabolism, W. D. McElroy and B. D. Glass, eds., p. 799, The Johns Hopkins Press, Baltimore, 1955. 2. J. Baddiley, G. Ehrensvard, E. Klein, L. Reio, and E. Saluste, J. Biol. Ch< m.. 188, 777 (1950). 3. L. Reio and G. Ehrensvard, Arkiv Kemi, 5, 301 (1953). 4. G. Ehrensvard and L. Reio, Arkiv Kemi, 5, 327 (1953). 5. C. Gilvarg and K. Bloch. ./. Biol. Chem., 193, 339 (1951) ; 199, 689 (1952). 6. R. C. Thomas, V. H. Cheldelin, B. E. Christensen, and C. H. Wang, ./. Am. Chem. Soc., 75, 5554 (1953). 7. P. R. Srinivasan, H. T. Shigeura, M. Sprecher, D. B. Sprinson, and B. D. Davis, in press. 8. B. L. Horecker, P. Z. Smyrniotis, and H. Klenow, ./. Biol. Chem., 205, 661 (1953). 9. B. L. Horeeker, M. Gibbs, H. Klenow. and P. Z. Smyrniotis, ./. Biol. Chem., 207, 393 (1954). Carbohydrates in Aromatic Compound Biosynthesis 269 10. E. Racker, G. de la Haba, and I. G. Leder, Arch. Biochem. and Biophys., 48, 238 (1954). 11. G. de la Haba, I. G. Leder, and E. Racker, ./. Biol. Chem., 214, 409 (1955). 12. J. Bassham, A. A. Benson, L. D. Kay, A. Z. Harris, A. T. Wilson, and M. Calvin, J. Am. Chem. Soc, 76, 1760 (1954). 13. P. R. Srinivasan, M. Spreeher, and D. B. Sprinson, Federation Pr<><-., />', 302 (1954). 14. M. E. Rafelson, Jr., G. Ehrensviird, M. Bashford, E. Saluste, and C. G. Heden, J. Biol. Chem., 211, 725 (1954). 15. M. E. Rafelson, Jr., J. Biol. Chem., 213, 479 (1955). 16. G. Ehrensvard, Ann. Rev. Biochem., 24, 275 (1955). 17. I. I. Salamon and B. D. Davis, /. Am. Chem. Soc., 75, 5567 (1953). 18. H. Yaniv and C. Gilvarg. ./. Biol. Chem., 213, 787 (1955). 19. E. B. Kalan, B. D. Davis, P. R. Srinivasan, and D. B. Sprinson, in press. 20. B. L. Horecker, P. Z. Smyrniotis, H. H. Hiatt, and P. A. Marks, J. Biol. Chem., 212, 827 (1955). 21. P. R. Srinivasan, D. B. Sprinson, E. B. Kalan, and B. D. Davis, in press. 22. P. R. Srinivasan, M. Katagiri, and D. B. Sprinson, /. Am. Chem. Soc, 77, 4943 (1955). 23. C. E. Ballou, H. O. L. Fischer, and D. L. MacDonald, /. Am. Chem. Sue, 77, 2658 (1955). 24. R. Robinson, Proc. Univ. Durham Phil. Soc, S, Part 1, 14 (1927-1928). 25. R. B. Woodward, Nature, 162, 155 (1948). On Determining the Chemical Structure of Proteins WILLIAM H. STEIIN A decade ago an essay under this title would have seemed hopelessly visionary. That it is not so today testifies to the precipitate progress made in the intervening years by biochemistry in general and protein chemistry in particular. Commencing with the introduction of chro- matographic methods into amino acid chemistry by Martin and Synge, this progress has culminated in the brilliant studies of Sanger (cf. ref. 1 for references) that have revealed completely the arrangement of the amino acid residues in the insulin molecule. Despite this strik- ing success, determining the chemical structure of a protein is still far from the routine procedure that will be necessary if proteins as a group are to become susceptible to searching structural analysis. On the basis of current knowledge, however, it is possible to formulate in some detail the steps that will be required to determine the chemical struc- ture of a protein. It is also possible to apprehend the nature of some of the difficulties remaining, and, in a few cases, to suggest possible methods by which they may be overcome. This essay will attempt to discuss the problem from this point of view, using ribonuclease as a specific example.* For the purposes of the present discussion, the determination of the chemical structure of a protein will be limited to finding the sequence of amino acids in the peptide chain (or chains), ascertaining the nature and position of any cross linkages in the molecule, and determining the manner of attachment to the peptide chains of any non- amino acid * The author owes much to the many fruitful and stimulating discussions he has enjoyed with Dr. Stanford Moore and Dr. C. H. W. Hirs. It is particularly fitting that, the studies on the structure of ribonuclease carried out by Dr. Hirs form an important part of this essay, for he is a recent graduate of the Depart- ment of Biochemistry at Columbia University. Permission to include this work. some of it unpublished at the time of writing, is gratefully acknowledged. 270 On Determining the Chemical Structure of Proteins 271 moieties that may be present, such as carbohydrates, nucleic acids, or porphyrins. On the basis of this definition, determining the chemical structure of a protein can be broken down into the following sequence of operations: (a) Purification and isolation of a sample of a protein sufficiently pure to warrant detailed structural work. (6) Quantitative amino acid analysis of the purified protein, (c) End-group analysis to determine the number of peptide chains in the molecule, (d) Rup- ture of the disulfide bridges or other cross linkages in the molecule, (e) Separation of the peptide chains from one another, if more than one is present. (/) Partial hydrolysis of a peptide chain, (g) Frac- tionation of the mixture of peptide fragments formed on partial hy- drolysis, (h) Determination of the structure of the isolated fragments. It will be apparent that formulation of the problem in this way implicitly assumes that proteins have a definite and determinable chemical structure." It assumes that when a given bovine pancreas synthesizes ribonuclease, for example, the amino acid residues in the chains of the various protein molecules will be laid down in the same order from one molecule to the next. It does not preclude the possi- bility that more than one kind of molecule possessing ribonuclease activity may be synthesized by the pancreas, but it does assume that the number of such kinds of molecules will be small and that the individual species of proteins from a single organism are not popula- tions of molecules the organic chemical structure of which varies more or less continuously around a mean. In short, this conception of the proteins ascribes to them the attri- butes of well-defined chemical compounds similar in fundamental char- acter to, though more complicated in structure than, the other types of substances synthesized by living organisms. There is considerable evidence in favor of this conception, the most compelling of which is the fact that Sanger has been able to establish an unambiguous chemi- cal structure for insulin. Were beef insulin to consist of a family of closely related molecular species, it is difficult to see how a unique structure could have been derived. Recent results with ribonuclease point in the same direction. This conception also has the pragmatic advantage that it encourages investigators to attack the problem of protein structure from the point of view of the organic chemist. Organic chemistry has no techniques for handling complex structural problems involving heterogeneous populations of molecules. No matter * For a summary, with references, of the contrary point of view, cf. Colvin, Smith, and Cook.- 272 Essays in Biochemistry- how complex the structure, however, where there is homogeneity there is hope. If the hope is ill-founded and proteins really are families of molecules, detailed investigations of their structure are bound to reveal this fact. In finally deciding whether or not the biochemical events directing the synthesis of protein molecules are so ordered as to permit exact structural duplication from one molecule to the next, it will be neces- sary to decide not only whether a given protein preparation is hetero- geneous, but also whether heterogeneity, if found, has been imposed in the process of isolating the material from natural sources. Oppor- tunities abound for introducing heterogeneity where none originally existed. At the outset, the choice of source material may be crucial in this connection. For years, insulins derived from the ox, the pig, and the sheep were tacitly assumed to be identical by many investi- gators simply because each had identical effects on blood sugar. The recent amino acid analyses of Harfenist 3 and the structural studies of Sanger 4 have finally proved, however, that the insulins of these species are slightly different. This finding should not be too surprising in view of the abundant immunological evidence demonstrating the species specificity of proteins. Most investigators would agree that structural work on a protein preparation derived from several animal species would probably be a waste of time, but how many would agree on just what constitutes a species? What degree of genetic homogene- ity must one demand before molecular homogeneity can be expected? Does individuality extend to the molecular level? It does not seem inconceivable that in the case of species, such as man, possessing an unknown and uncontrollable genetic constitution, true homogeneity of some types of proteins may only be attainable in a preparation derived from a single individual. Even granted a suitable starting material containing a mixture of initially homogeneous proteins, however, the problems involved in iso- lating a single molecular species without introducing alterations in the molecule remain among the most intractable facing the protein chemist. For a long time, virtually the only preparative procedures available made use of fractional precipitation in some form. Fortunately, recent years have seen the introduction of multistage techniques of high re- solving power such as count ercurrent distribution, zone electrophoresis, and chromatography. Chromatography in particular would appear to possess several features that, in principle, should make it ideally adapted to work with proteins. The method is gentle, flexible, has high resolving power, and can be used for both analytical and preparative On Determining the Chemical Structure of Proteins 273 purposes. Nevertheless, despite intensive efforts in many laboratories (cf. Zittle,5 Moore and Stein,0 and Porter7 for references*, only a handful of proteins have been chromatographed successfully. Usually, however, successful chromatography has led to purer products and new information, indicating that if the method could be more widely em- ployed protein chemistry would benefit greatly. The difficulties in chromatographing proteins doubtless arise from the fact that proteins are large, fragile, polyvalent molecules. Because of their size, they cannot penetrate into the particles of most column packings the way small molecules do but must be bound largely at the surface. Unless the column packing has a large surface, therefore, the capacity may be so small that effective chromatography is impos- sible.* A limited capacity will also render the behavior of the column very sensitive to variations in the composition of the mixture being chromatographed. Displacement of one protein by another, and com- petition between proteins for a limited number of binding sites, may complicate the interpretation of the effluent curves. Under such cir- cumstances, rechromatography, determinations of enzymatic or other activities in the effluent, electrophoresis, or amino acid analyses of the individual peaks will all prove helpful as ancillary means of following the fractionation and will serve to minimize errors of interpretation. The fact that proteins are polyvalent molecules undoubtedly makes effective chromatography more difficult. Most proteins contain several residues of each of the amino acids; hence, no matter what types of linkage bind the protein to the column packing, these linkages are almost sure to be multiple. As has been pointed out before, notably by Tiselius,9 polyvalent molecules are likely to be all adsorbed or all eluted. The Rf is likely to be either one or zero and to change abruptly from one extreme to the other over a rather narrow range of experi- mental conditions. Satisfactory elution analysis, however, requires a reversible distribution of solute between stationary and mobile phases leading to Rf values intermediate between zero and one. It has seldom been possible to find such conditions with proteins, and for this reason an eluent of constantly changing pH or ionic strength or both (so- called "gradient elution") has of necessity been employed in several * The high capacity for proteins exhibited by XE-64, a ground form of IRC-50, is probably related to its large surface, inasmuch as a bead form of the same resin does not yield satisfactory chromatograms. It is not clear whether the high capacity possessed by the ion-exchange materials made by Sober and Peterson 8 from cellulose depends upon the surface of the cellulose or a loose gel structure that permits penetration of protein into the column packing. 274 Essays in Biochemistry- laboratories, including our own. In this procedure, the chromatogram is begun with an eluent incapable of eluting the protein in question (Rf = 0) , whereupon the composition of the eluent is gradually changed to one in which the protein has an Rf of 1. The exact point of emer- gence of the protein will thus be a function more of the rate of change of the eluent than of the length of the column. Although worth-while purifications can be effected in this manner,9'10 elution analysis of proteins has, in our experience, been found to be superior when it can be made to work properly. Satisfactory elution analysis has been possible with several proteins on both partition and ion-exchange columns. Despite the fact that these proteins are polyvalent molecules, test-tube experiments showed unquestionably that, in several instances, reversible distribution be- tween resin (IRC-50, in each case) and buffer existed. Moreover, the magnitude of this distribution coefficient could be changed in a pre- dictable fashion by altering the pH or the ionic strength of the buffer phase. In short, several proteins, among them ribonuclease,11 have been found to behave much like simple substances. It is pertinent to inquire why this should be so. From a logical point of view, the problem really is not why most proteins chromatograph unsatisfac- torily but why some behave well. Ribonuclease, for example, contains a minimum of fifteen possible cationic sites (cf. Table 1) which could exchange with Na+ on a buffered IRC-50 column. A clue may be furnished by the electrophoretic studies of Crestfield and Allen 12 that show that the isoelectric point of ribonuclease drops from pH 7.48 in dilute acetate buffers to pH 5.49 in 0.2 N phosphate buffers of the type used in chromatography. Apparently phosphate ions form a feebly dissociable complex with ribonuclease, and it may be that the existence of this complex facilitates chromatography on the ion exchange resin. Complexing agents have been utilized with success in other fields, notably in the case of the rare-earth elements by Speckling and his associates 13 and in the case of the sugars by Khym and Zill.14 Perhaps a systematic search for agents that form complexes with proteins might widen the scope of elution analysis. Whether or not chromatography is the best way to procure them, it is unquestionably true that a larger number of rigorously purified proteins remains one of the critical needs of the protein chemist. In considering which proteins were likely to be good subjects for detailed structural study, the fact that ribonuclease could be purified chromatographically was one of the most important considerations leading to its selection. Its small size (mol. wt., 14,000) and the On Determining the Chemical Structure of Proteins 275 absence of tryptophan from the molecule were also strongly in its favor. As a first step in the investigation, detailed amino acid analyses were performed so that an accurate balance sheet in terms of amino acids could be kept as the structural studies progressed. Although the quan- titative amino acid analysis of a protein hydrolyzate no longer presents serious problems, it has become apparent that the number of amino acids that may decompose on hydrolysis and the extent of the decom- position may vary from laboratory to laboratory and from protein to protein. Possibly this variation could be minimized if the tempera- ture at which the hydrolysis is conducted were precisely controlled. Some decomposition during acid hydrolysis has been noted at one time or another for serine, threonine, cystine, tyrosine, aspartic acid, glu- tamic acid, proline, methionine, histidine, lysine, and arginine.2-3-15"17 (Tryptophan, of course, decomposes nearly completely.) The variable nature of this decomposition requires that individual corrections must be worked out for each protein studied. Analysis after two times of hydrolysis, say 20 and 70 hours, permits extrapolation to zero time and seems to yield the most accurate results obtainable at present. The longer time of hydrolysis also allows some estimate to be made of those amino acids, such as valine or isoleucine, originally bound in peptide linkages resistant to acid hydrolysis. The amino acid com- position of ribonuclease given in Table 1 was determined in this manner.17 Corrections for decomposition had to be made for serine, threonine, cystine, tyrosine, aspartic acid, glutamic acid, proline, and arginine. Isoleucine was liberated slowly. Despite these corrections, however, the number of residues of each amino acid present to the extent of ten residues per mole or less is known with considerable assurance. For those amino acids present in larger quantities, a rea- sonable experimental error of ±4%, coupled with the uncertainties introduced by corrections, may cause an error of ± one residue in the final value. With proteins larger than ribonuclease, this limitation might extend to most of the amino acids. End-group analyses of ribonuclease were performed by Anfinsen, Redfield, Choate, Page, and Carroll.18 When the DNP technique of Sanger was employed, it was established that ribonuclease consisted of a single peptide chain bearing lysine as the amino-terminal amino acid, followed in turn by glutamic acid, threonine, and alanine. By the use of carboxypeptidase, the carboxyl-terminal amino acid was found to be valine, followed back along the peptide chain by phenyl- alanine, isoleucine, or leucine, alanine, tyrosine, and methionine in un- determined order. On the basis of this information and the analytical 276 Essays in Biochemistry- Table 1. Amino Acid Composition of Rihonuclease (The values for the stable amino acids are the average of six determinations) Residues per Mole To Nearest Amino Acid Found Integer Aspartic acid * 15.8 16 Glutamic acid * 11.8 12 Glycine 3.05 3 Alanine 12.0 12 Valine 8.95 9 Leucine 2.15 2 Isoleucine 2.85 3 Serine * 15.0 15 Threonine * 10.45 10 Half cystine f 8.15 8 Methionine 3.75 4 Proline * 4.79 5 Phenylalanine 2.97 3 Tyrosine * 5.87 6 Histidine 3.80 4 Lysine 10.0 10 Arginine * 3.97 4 Amide NH3 J 17.0 (17) § Total 126 Calculated molecular weight 13,895 Nitrogen recovery Weight 97% 99% * Extrapolated values for amino acids that show decomposition on hydrolysis. t The average of triplicate determination of cystine as cysteic acid. X Average of two determinations of amide NH3. § Amide groups not included in the summation of the total number of residues. data in Table 1, Anfinsen et al. proposed the structure for ribonuclease shown in Fig. 1.* The existence of four disulfide bridges in the molecule was assumed wdien attempts to detect sulfhydryl groups failed. Ledoux 10 has claimed, on the basis of experiments with oxidizing and reducing agents such as H202, HCN, and reduced glutathione, that sulfhydryl groups are essential for enzymatic activity. Davis and Allen 2n and Dickman and Aroskar,-1 on the other hand, could find no evidence for the presence of sulfhydryl groups in the molecule. This * At the time this structure was proposed, it was not appreciated that there are probably five and not four proline residues in the molecule. On Determining the Chemical Structure of Proteins 277 apparent contradiction may be resolved by the observation20--1 that traces of copper inhibit the enzyme strongly. It seems entirely possi- ble that HCN and sulfhydryl compounds could enhance enzymatic activity by sequestering inhibitory traces of copper that might be present in either the enzyme, the substrate, or some of the reagents used in the assay procedure. According to this view, copper could not inhibit enzymatic activity by reacting with sulfhydryl groups of the enzyme, which is the usual assumption advanced to explain inhibition by copper, but must be bound at some other site in the molecule. The LNH2] i^ya.vj 1 s s 1 [Pro] [Pro] s 1 s 1 1 S i s 1 [Pro] [Pro] 1 s s 1 IMnt i T>U„1 Vo Fig. 1. Generalized gross structure of ribonuclease (from Anfinsen, Redfield, Choate, Page, and Carroll18). imidazole ring of one or more of the four histidine residues in the enzyme seems a likely binding site, particularly in view of the findings of ^'eil and Seibles 2- that photooxidation of only one of the histidine residues in ribonuclease leads to a 74% decrease in enzymatic activity. If, in fact, a histidine residue is associated with the active center of the enzyme, it seems possible that the studies now in progress might reveal something of the structure associated with this active center. If disulfide bridges exist in a protein molecule, it is desirable that they be ruptured before partial degradation is attempted. The peptide chains, if there is more than one, should then be separable, as they were in the case of insulin." If, as in ribonuclease, only one chain exists, hydrolysis of peptide bonds will not necessarily lead to a reduc- * The peptide chains of insulin were separable on the basis of solubility. To separate the chains obtained from other proteins, techniques that have been found useful for the separation of large peptides, such as countercurrenl distribu- tion, or chromatography on IRC-50, or low-eross-linked polystyrene sulfonic acid resins, may prove advantageous. 278 Essays in Biochemistry tion in molecular weight unless the disulfide cross links have first been cleaved.18 Moreover, if hydrolysis by proteolytic enzymes is contem- plated, rupture of disulfide bonds is frequently necessary before enzy- matic attack can occur. Finally, Ryle and Sanger 23 have shown that a mixture of peptides of the structure R — S — S— R' and R" — S — S — R'" can rearrange easily to give R— S-S-R" + R'— S— S— R"'. It is obviously desirable to rupture disulfide bonds and eliminate this pos- sibility, inasmuch as the determination of the sequence of amino acids in a long peptide chain already presents sufficient problems without introducing rearrangements to complicate the picture still further! Two general methods for cleaving disulfide bonds exist, the oxidative and the reductive. Sanger employed the performic acid oxidation pro- cedure of Toennies and Homiller 21 in his work with insulin, by which means each mole of cystine yields two moles of cysteic acid. Schram, Moore, and Bigwood 25 and Mueller, Pierce, and du Vigneaud 26 have described conditions under which the conversion is virtually quanti- tative, and Hirs 27 has found that in ribonuclease methionine is also quantitatively transformed to the sulfone. With the definition by Mueller et al.,26 Hirs,27 and Thompson,28 of conditions under which tyrosine is fully stable, the performic acid oxidation procedure appears admirably suited for use with proteins, such as insulin and ribonuclease, that contain no tryptophan. With the large majority of proteins that contain tryptophan, however, difficulties will be encountered, for Wit- kop and his co-workers 29 have shown that tryptophan is unstable to oxidation. If a single product, such as formylkynurenine, could be obtained in good yield after oxidation, the instability of tryptophan would make little difference. Formylkynurenine containing peptides would then be formed on partial hydrolysis, and total acid hydrolysis of these peptides would yield kynurenine which should readily be determined chromatographically. It seems likely, however, that oxida- tion of a tryptophan-containing protein with performic acid, followed by partial hydrolysis, would result in the formation of a number of different fragments derived from each tryptophan residue, thus greatly complicating subsequent structural work. The difficulty with tryptophan might be circumvented if the di- sulfide bridges of a protein were cleaved by reduction instead of by oxidation. Sodium in liquid ammonia or lithium borohydride (on the unesterified protein) might be capable of reducing all the disulfide bonds in the molecule without attacking other vulnerable groups (cf. Roberts,30 for example, and Bailey31). It would then, of course, be necessary to cover the newly formed sulfhydryl groups by some reagent On Determining the Chemical Structure of Proteins 279 that would not react elsewhere in the molecule. One of the most specific reactants for an — SH group is another — SH group in the presence of oxygen. Whether protein— SH groups fonned by reduction could be induced to react completely in the presence of oxygen with an excess of another SH compound such as cysteine or thioglycolic acid remains to be determined. Under any circumstances, the question of how to deal with disulfide bridges * in proteins containing trypto- phan may prove to be one of the crucial bottlenecks in future work with protein structure. Once a single peptide chain free from cross bridges is obtained, three general approaches to the determination of its structure are avail- able. In one, the sequence of amino acids is derived by hydrolyzing the chain into relatively small fragments employing more or less random hydrolysis with strong acid. The large number of di-, tri-, and tetrapeptides fonned must then be separated and the structure of each determined. This is essentially the approach utilized by Sanger, but, despite his notable success, this method has serious draw- backs which he well appreciated.32 An enormous number of peptides are formed after such treatment, most of them in relatively poor yield. The problems of fractionating and isolating pure peptides from such a mixture are great. Although each individual peptide is of relatively simple structure, this advantage is offset by the large number of peptides requiring study. It is doubtful whether such an investigation could be completed successfully for peptide chains much longer than the thirty amino acids contained in the B chain of insulin. It does not appear to be a generally feasible approach for determining the structure of a large number of complex proteins. A second possible method of determining the sequence of amino acids in a peptide chain is by stepwise degradation starting at either the carboxyl or amino end. Promising techniques for accomplishing this are available, or on the horizon, and it may well be the approach of choice for small peptides. It seems doubtful, however, whether any stepwise procedure will be suitable for the degradation of long chains of amino acids such as occur in intact proteins. The losses involved in each step seem insupportable. If, for example, 1 gm. of a protein were degraded stepwise and a yield of 75% obtained at each * The existence of diester phosphate bridges in proteins, and methods for cleav- ing them, have been suggested by the work of Perlmann.33 Diester sulfate bridges are also a possibility after the finding of Bettelheim 34 that tyrosine-O-sulfate is a constituent of fibrinogen. 280 Essays in Biochemistry- step, only 15 mg. of material would remain after fifteen steps. Clearly, more information can be derived from 1 gm. of protein by other means. The most appealing approach to the determination of the sequence of amino acids in a peptide chain consists of specific hydrolysis of the chain at selected linkages, separation of the relatively few fragments formed, followed by further specific cleavage of the larger fragments when necessary, and direct structural analysis of the smaller fragments. Although specific chemical methods would be ideal, at the present time proteolytic enzymes appear to offer the greatest promise as ana- lytical reagents for the selective hydrolysis of peptide chains.* Some, such as trypsin, have an extremely sharp specificity. All operate under mild conditions of pH and temperature, so that after hydrolysis, amide linkages are still intact and non-amino acid moieties such as carbohydrates or porphyrins are likely to remain attached to that portion of the peptide chain to which they were originally joined. Sanger employed trypsin, pepsin, chymotrypsin, and papain in his work with insulin. The enzymes were used as ancillary tools, how- ever, to establish sequences of amino acids in the neighborhood of peptide bonds that were particularly labile to acid hydrolysis. More recently, Bell and his associates 35 used trypsin, pepsin, and chymo- trypsin as primary hydrolytic agents in their studies with ACTH. Trypsin has been employed by Tuppy and Bodo 36 in work with cyto- chrome c, by Gorini, Felix, and Fromageot who studied lysozyme,37 and by Monier and Jutisz 3S who isolated some basic peptides from protamines. When enzymes are employed as primary hydrolytic agents, it is, of course, essential to eliminate the possibility that the products isolated might have been formed as a result of rearrange- ments catalyzed by the enzymes. Unless this can be done with some assurance relatively early in the investigation, an enormous amount of labor could be wasted proving the structure of artifacts. In beginning partial degradation studies with ribonuclease, a tryptic hydrolyzate of the oxidized protein was chosen for the first investi- gations.27-39 From the classic specificity studies of Bergmann, Fruton, and Hofmann, taken in conjunction with the analytical results shown in Table 1 and the data of Anfinsen et al. on the N- and C-terminal amino acids, it would be anticipated that fourteen peptides should be found in a tryptic hydrolyzate. Nine of them should terminate in lysine, four in arginine, and one, corresponding to the C-terminal frag- ment, should be devoid of basic amino acids. Free lysine, from the * The possibility of using proteolytic enzymes in this way was clearly envisaged by Max Bergmann and stated in a Harvey Lecture delivered by him in 1935. On Determining the Chemical Structure of Proteins 281 N-terminal position, might or might not be present, because little is known as to whether trypsin can hydrolyze bonds of this type. The rate at which oxidized ribonuclease is split by trypsin is shown in Fig. 2. Amino acid analyses of the oxidized preparation employed showed that cystine had been converted quantitatively to cysteic acid and methionine to the sulfone, but that all the other amino acids were untouched. Moreover, end-group analysis by the DNP technique Time, hours Fig. 2. Hydrolysis of oxidized ribonuclease by trypsin at 7)H 7.0 and 25° ; ribo- nuclease concentration, 0.5%; trypsin concentration, 0.0025%. revealed that the peptide chain had not been cleaved to any significant degree during the oxidation. On the basis of the curve shown in Fig. 2, it was decided to investigate the peptides present after both 3 and 20 hours of tiyptic action. For this purpose, the hydrolyzate was fractionated on a column (150 X 2 cm.) of Dowex 50-X2 with the results shown in Fig. 3. The experiment was performed with about 200 mg. of ribonuclease so as to yield a sufficient quantity of each peptide for further structural investigation. As a first step, the quan- titative amino acid composition of each peptide was determined after acid hydrolysis by chromatography on columns of Dowex 50-X4.40 The buffer salts from the eluent used in the Dowex 50-X2 column usually did not interfere in this process and were not removed. The amino acid composition of each peptide obtained in pure form in the initial fractionation is shown on the curve in Fig. 3. The analyses 282 Essays in Biochemistry x X X i a. (•dood j^ui 's^ua^BAtnba aupnai) O © K CO << ."d o O 1 | ! ^ lO ^ lOI *r*^ 001 a lO ^s. c^ A be H < 0) JS a, ffi 2 - 3 3 J + CO -< S ^ s w o <33 (fc o lO S| lO IOI rH-l ib'i 22.' SS CO >> CO J bJD >> 5 < u o CO -M < d s" 0) < >-4 CO 3 ci fl) ft I/) a> cS ft a) — d o d o c X <+-! — O T3 -r OJ - o -q — i >. (N " O cu T3 V — 1 c3 d c > 0 0 ft be 3 3 0 A rr c 43 o CI +-> -;: L_ bl 3 CO td 03 ^ 0) o ~ fei — CI Q R o n C £. o CO o d - l=H '" On Determining the Chemical Structure of Proteins 283 gave whole numbers for each of the amino acids in the peptides with an accuracy of ±5 to 10%. The numbers in parenthesis give the yield of each peptide obtained after 3 and 20 hours of hydrolysis, on the assumption that each one of these fragments occurs only once in the molecule.* On the basis of the specificity of trypsin, the basic amino acid, lysine or arginine, is assigned to the C-terminal position. Eventu- ally it would be well to check this point experimentally, but for the present this assignment seems reasonable. It will be noted that peptide 10 contains two lysine residues. Inasmuch as DNP end-group analyses placed one lysine at the amino-terminal position, it seems extremely probable that this peptide represents the amino-terminal sequence. The first four amino acids of this sequence were determined by Anfinsen et al. to be Lys.Glu.Thr.Ala-, so that the finding of two more alanine residues and a lysine makes it possible to extend this sequence to seven amino acids without further structural work. The mixture of peptides labeled 2, 3, 4, occurring at the beginning of the chromatogram, and those at 9 and at 13 + 14 were rechromato- graphed on Dowex 50-X2 under slightly different conditions to give the results shown in Fig. 4. Three peptides were separated from the front part of the original chromatogram (a 20-hour tryptic hydrol- yzate), two containing lysine and one arginine. Both lysine peptides had the same 21 amino acid residues and probably differed only in their amide content. The fact that the faster moving of the two lysine peptides, the center peak in the top curve of Fig. 4, was present to only a very small extent after 3 hours of tryptic digestion is compatible with this supposition. The arginine peptide is also very large, con- taining nineteen amino acid residues. It is unstable, however, the yield decreasing from 55% in the 3-hour hydrolyzate to 15% in the 20-hour hydrolyzate. The reason for this instability is not certain at the moment, but possibly this fragment contains some linkage that is extremely susceptible to the action of chymotrypsin, a trace of which cannot be ruled out as a contaminant of the trypsin. Peptide 9 is very large and contains two lysine residues. The pres- ence of proline may explain why one of the lysyl bonds is not split by trypsin to yield two peptides. If the sequence -Lys.Pro- existed, the lysylprolyl bond might be very resistant to tryptic hydrolysis. In * The figures with rules under them were computed from the actual amino acid analyses of the peptides. The other yields for different times of hydrolysis were computed by comparing the area of the peak with the area of the corresponding peak for which the amino acid composition was known. 284 Essays in Biochemistry [Cys, Asp3(xNH2), Glu2 (;yNH2), Ala2, Met3, 4 Tyr, Ser6,Thr, His] Lys (50; 45%) \. Peptides 2, 3, and 4 1.5 \: 1.0 [Cys2, Asp3(xNH2), Glu2 CyNH2)Gly, lieu, Tyr2,Ser3,Thr3]Arg (55; 15%) \ 2 Met, 1? 0.5 J 1 0 .<><>C><>OOCKl£!e©ol><><><>0^^ -Qa^y Vx£ — Effluent liters 1.0 •Gradually increasing pH, 3.1 2.0 4.3, at constant [Na+], 35° ,2? 10 [Cys2, Asp2(xNH2), Glu3 CyNH2), Ala,, Val4, Leu, Pro, Phe, Ser2, Thr, His, Lys]Lys (30; 50%)\ ft 9 0.5 Peptide 9 Effluent liters 2.0 3.0 — | Gradually increasing pH and [Na+ ], (0.2 N pH 3.1— 0.5 N pH 5.1), 50° |— -3 1.0 0.5 - Peptides 13 and 14 [Cys, Asp-NH2, Ala, Tyr2 Pro] Lys (- 1 ;85%) [Cys Leu, 2, Asp3(xNH2), Pro, Phe, Ser2 Glu3CyNH2), Ala 2, Val4, Thr, His, Lys, Arg]Lys \ ; 20%) 13 rM W Effluent liters 2.0 3.0 — | Gradually increasing pH and [Na+ ], (0.2 N pH 3.1— 1.0 iST pH 5.1), 50° |— Fig. 4. Rechromatography of peptides from a tryptic hydrolyzate of oxidized ribonuclease on a 150 X 2 cm. column of Dowex 50-X2. The peptides fraction- ated are taken from the overlapping peaks shown in Fig. 3, except that peptides 2, 3, and 4 were obtained from a 3-hour tryptic hydrolyzate. On Determining the Chemical Structure of Proteins 285 ACTH, for example, a similar bond was found by Bell 35 not to be hydrolyzed by trypsin. Rechromatography of the next to last peak in Fig. 3 gave the two peptides shown at the bottom of Fig. 4. One is a heptapeptide contain- ing lysine found in good yield. The other, however, contains twenty- four amino acid residues including two lysine and one argininc residue and was obtained in low yield. Since all the lysine and argininc residues in the molecule have already been accounted for, this peptide must represent a combination of two other peptides. From its com- position, it could be formulated as aspartylarginine (peptide 7) added to peptide 9. Whether this peptide is an intermediate degradative product or a rearrangement product cannot be determined for certain at present, but, since the yield of aspartylarginine increases with time, the former possibility seems more likely. The peptides characterized thus far account for 105 of the 126 amino acid residues in ribonuclease. Missing are four residues of valine, three of proline and aspartic acid, two of alanine, isoleucine and histidine, and one of glutamic acid, glycine, half-cystine, phenyl- alanine, and tyrosine. Also missing is a peptide devoid of basic amino acids that should arise from the carboxyl-terminal end of ribonuclease. According to Anfinsen et al.,18 the last six residues in the molecule are -met, tyr, ala, leu or ileu, phe, and val. It has not been possible thus far to isolate such a carboxyl-terminal fragment from the Dowex 50-X2 columns. It may be that the "end piece" of ribonuclease is particularly susceptible to the action of traces of chymotrypsin, which, as has been noted above, may contaminate the trypsin. It is tempting to assume that all the amino acids listed above as not accounted for in the basic peptides should be assigned to the carboxyl-terminal end piece, but only future work can decide the validity of this assumption.* The balance sheet does prove, however, that isoleucine and not leucine is present in the end fragment, because both leucine residues have been found in peptides shown in Figs. 3 and 4. The existence of a residue of methionine near the carboxyl end cannot be reconciled with the analyses of the peptides reported above, because all four of the methio- nines supposed to exist in the molecule have been accounted for. This discrepancy will also have to be resolved by future work. Although the work on tryptic digests of ribonuclease is still far from elucidating the complete structure of the molecule, it has demonstrated * A large peptide devoid of basic amino acids and believed to represent the carboxyl-terminal segment has recently been isolated. Its composition is discussed in ref. 39. 286 Essays in Biochemistry several things in quite a clear-cut fashion. In the first place, it seems justifiable to conclude that under the proper circumstances, proteolytic enzymes will prove to be remarkably general and useful tools for determining the structure of proteins. The principal theoretical draw- back to their use, namely, that they catalyze rearrangements, seems not to be operative, at least in the case of trypsin.* It is difficult to conceive that the group of peptides shown in Figs. 3 and 4 could be artifacts. Practically all of them were obtained in yields of from 50 to 90% of theory after only 3 hours of tryptic digestion. It hardly seems credible that rearrangements could proceed so rapidly and so completely in homogeneous solution. The absence in most of the peptides of more than one residue of a basic amino acid also argues against transpeptidation. Of course it cannot be assumed that, because transpeptidation does not occur to a significant extent when trypsin acts on ribonuclease, it will never occur. By the use of quantitative procedures, however, it should be possible to insure against undetected rearrangements. When the peptides can be separated quantitatively and their amino acid composition determined quantitatively, there is far less likelihood of being misled. In the final analysis, proof for the validity of the results obtained with trypsin can only emerge from an examination of the products formed by the action of other enzymes. Such studies with ribonuclease are now under way. It can be said already that chymotrypsin splits oxidized ribonuclease into twenty to thirty fragments that can be separated on the Dowex 50-X2 columns. About fifteen of these fragments are found in major quantities. Pepsin also attacks oxidized ribonuclease but with the formation of more peptides in poorer yield. f The work with tryptic digests has also demonstrated quite clearly that columns of Dowex 50-X2 provide extremely effective means for separating mixtures of peptides. The resolving power is great, and even peptides containing twenty or more amino acid residues can be handled without difficulty. Moreover, such columns possess the great advantage that they can be scaled up readily to a size sufficient to permit the isolation of enough of a peptide for subsequent structural studies. In the present work, the columns have been operated in the sodium form. Doubtless quite similar separations could be achieved with the ammonium form of the resin and NH4OAc or formate as * The subject of the synthesis of peptide bonds by proteolytic enzymes has been elegantly summarized by J. S. Fruton in a Harvey Lecture given November 17, 1955. t J. L. Bailey. On Determining the Chemical Structure of Proteins 287 eluents, both of which can be removed by volatilization. Alternatively, eluents containing sodium can be desalted by passage over a small bed of NH4-Dowex 50, whereby Na is exchanged for the volatile ammonium. Once the amino acid composition of the various peptides obtained from peptic and chymotryptic digests of ribonuclease has been deter- mined and correlated with the data obtained by the use of trypsin, it should be possible to learn a great deal about the way the molecule is constructed even before sequence studies on the individual peptides have been performed.* Moreover, from the array of data thus pre- sented, it should be possible with some assurance to choose for detailed investigation those and only those peptides the amino acid sequence of which will contribute the most to an understanding of the final structure of the protein. It should also be possible to choose those peptides the size and composition of which are most suitable for study by existing techniques. In this manner the number of sequence deter- minations required will be reduced to a minimum. There is not space to go into the question of the actual determination of the sequence of amino acids in peptides. If the peptides chosen for such an investi- gation contain from four to ten amino acid residues, there is little doubt that a sequence for each can be determined by one or a com- bination of several existing methods. Structural analysis would be simplified greatly if there were avail- able more proteolytic enzymes possessing a specificity as sharp as that of trypsin, only directed towards other linkages, such as those involv- ing proline, or histidine, or the aromatic amino acids. Possibly the enzymologist will be able to come to the aid of the analyst and provide the necessary tools. If, however, enzymes of the requisite specificity cannot be found in nature, possibly they can be created artificially. Microorganisms seem to be able to do almost anything, given sufficient provocation. Perhaps with the aid of synthetic substrates they can be induced to synthesize enzymes capable of hydrolyzing the particular types of peptide linkages in which the protein chemist is interested. It should be emphasized that there is nothing extremely novel in * The separation and analysis of the peptides formed by the action of chymo- trypsin has recently been completed by Dr. Hirs. On the basis of his data, it has been possible to construct, as a working hypothesis, a tentative, partial structural formula showing how the peptides present in the tryptic and chymo- tryptic hydrolyzates might have originally been joined together in the molecule of oxidized ribonuclease [Hirs, Stein, and Moore, /. Biol. Chem., in press]. The expectations expressed in this paragraph have, therefore, been realized. 288 Essays in Biochemistry the approach to determining the structure of a protein summarized in this essay. Enzymatic hydrolysis with several different enzymes, fol- lowed by fractionation of the peptides thus produced and amino acid analysis of each before ultimate sequence determinations on a selected few, is the procedure used in essence by Bell and his associates in their studies on ACTH. It is also a simple and logical development of the approach the organic chemist has employed for years in the structural analysis of simpler compounds. The general use of quantitative tech- niques will, it is felt, simplify the task with proteins and make for less uncertainty in the final result. Obviously, a large number of routine quantitative amino acid analyses are required, partly as a substitute for, and always as a prelude to, detailed structural work. To render this approach truly feasible, therefore, these amino acid analyses must be made extremely simple and quick. It seems safe to say that none of the existing procedures completely fills this need. There are two obvious ways- of improving this situation. One is to render paper chromatography truly quantitative and more fully mechanized. The other is to render current column methods faster and more automatic, a subject under continued study in our laboratory. The amount of information now available on the structure of pro- teins is so limited that it is too early to expect any ironclad conclusions to emerge. Nevertheless, a few suggestive facts are worth noting. There seems to be some tendency for like types of amino acids to cluster together along the peptide chain. For example, in insulin we find the three sequences, -Glu.Glu-NH2-, -Glu-NHo.Leu.Glu.Asp- NH2-, and -Asp-NH2.Glu-NH2-, thus placing acidic amino acids and their amides near one another. In another part of the insulin molecule we find, in the A chain, three out of six half-cystine residues in the protein in the single sequence -Cys.Cys.Ala.Ser.Val.Cys-. There is also a cluster of three aromatic amino acids together in the sequence -Phe.Phe.Tyr-, and finally, the only two strongly basic amino acid residues in the molecule are tucked off at one end of the B chain separated by only seven amino acids. This clustering of like R groups is even more obvious in the ACTH molecule. There is a sequence of fourteen amino acid residues near the N-terminal sequence, of which seven are either lysine or arginine. At one point the sequence -Lys.Lys.Arg.Arg- occurs. At the opposite end of the chain there is a cluster of five acidic amino acid residues out of a sequence of eight. Although the information on ribonuclease is still meager, it is already apparent that the basic amino acids also tend to cluster in this molecule. Of the four arginine residues, two are separated from another basic On Determining the Chemical Structure of Proteins 289 amino acid (either lysine or arginine) by two amino acid residues and one arginine is separated by three. Of the ten lysine residues, seven are separated from another basic amino acid by six amino acid residues or less. In another part of the molecule we find three alanine residues in a row, and in another peptide seven out of twenty-one amino acid residues are serine. Recently, Davie and Neurath " have isolated the hexapeptide that is split off when trypsinogen is transformed to trypsin and proved it to have four aspartic acid residues joined together in the structure Val.Asp4.Lys. If, in fact, this apparent tendency of amino acids of like structure to group together turns out to be a general occurrence, it means that in the peptide chain there will be found areas in which the positive charge density is high, areas in which the negative charge density is high, and areas in which van der Waals forces pre- dominate. There is suggestive evidence based on studies of enzyme specificity and of the manner in which protein molecules interact with one another that is compatible with this idea. How these areas are disposed on the surface of the protein molecule will, of course, be a function of the manner in which the peptide chain is folded. At this stage of development, speculation is tempting, but the real need is many more facts. It appears that at last the protein chemist is about to possess tools adequate to the difficult task of accumulating these facts. References 1. F. Sanger, E. O. P. Thompson, and H. Tuppy, Symposium 4, on Protein Hormones and Protein Derivatives, p. 26, 2nd International Congress of Biochem- istry, Paris, 1952. 2. J. R. Colvin, D. B. Smith, and W. H. Cook, Chetn. Revs., 54, 687 (1954). 3. E. J. Harfenist, J. Am. Chem. Soc, 75, 5528 (1953). 4. F. Sanger, Nature, 164, 529 (1949). 5. C. A. Zittle, Advances in Enzymol, 14, 319 (1953). 6. S. Moore and W. H. Stein, Ann. Rev. Biochem., 21, 521 (1952). 7. R. R. Porter. Brit. Med. Bull, 10, 237 (1954). 8. H. A. Sober and E. A. Peterson, J. Am. Chem. Soc., 76, 1711 (1954). 9. A. Tiselius, Archiv Kemi, in press. 10. C. F. Crampton, S. Moore, and W. H. Stein, /. Biol. Chem., 215, 787 (1955). 11. C. H. W. Hirs, S. Moore, and W. H. Stein, /. Biol. Chem., 200, 493 (1953). 12. A. M. Crestfield and F. W . Allen. ./. Biol. Chem., 211, 363 (1954). 13. F. H. Spedding, Discussions Faraday Soc, 7, 214 (1949). 14. J. X. Khym and L. P. Zill. ./. Am. Chem. Soc, 73, 2399 (1951). 15. E. L. Smith and A. Stockell. ./. Biol. Chem., 207. 501 (1954). 16. E. L. Smith, A. Stockell, and J. R. Kimmel, J. Biol. Chem., 207, 551 (1954). 17. C. H. W. Hirs, W. H. Stein, and S. Moore, J. Biol. Chem., 211, 941 (1954). 290 Essays in Biochemistry 18. C. B. Anfinsen, R. R. Redfield, W. L. Choate, J. Page, and W. R. Carroll, J. Biol. Chem., 207, 201 (1954). 19. L. Ledoux, Biochim. et Biophys. Acta, 13, 121 (1954). 20. S. R. Dickman and J. P. Aroskar, Federation Proc, 14, 202 (1955). 21. F. F. Davis and F. W. Allen, Federation Proc, U, 200 (1955). 22. L. Weil and T. S. Seibles, Arch. Biochem. and Biophys., 54, 368 (1955). 23. A. P. Ryle, and F. Sanger, Biochem. J., 58, v (1954). 24. G. Toennies and R. P. Homiller, J. Am. Chem. Soc, 64, 3054 (1942). 25. E. Schram, S. Moore, and E. J. Bigwood, Biochem. J., 57, 33 (1954). 26. J. M. Mueller, J. G. Pierce, and V. du Vigneaud, J. Biol. Chem., 204, 857 (1953). 27. C. H. W. Hirs, Federation Proc, 13, 230 (1954); J. Biol. Chem., in press. 28. E. O. P. Thompson, Biochim. et Biophys. Acta, 15, 440 (1954). 29. A. Ek, H. Kissman, J. B. Patrick, B. Witkop, Experientia, 8, 36 (1952). 30. R. G. Roberts and C. O. Miller, J. Am. Chem. Soc, 58, 309 (1936) ; R. G. Roberts, D. M. Hilker, and A. Gasior-Russell, Proc. Soc. Exp. Biol. Med., 68, 466 (1948). 31. J. L. Bailey, Biochem. J., 60, 170 (1955). 32. F. Sanger, Advances in Protein Chem., 7, 1 (1952). 33. G. E. Perlmann, Biochim. et Biophys. Acta, 13, 452 (1954). 34. F. R. Bettelheim, J. Am. Chem. Soc, 76, 2838 (1954). 35. P. H. Bell, J. Am. Chem. Soc, 76, 5565 (1954). 36. H. Tuppy, and G. Bodo, Monatsh. Chemie, 85, 1024 (1954). 37. L. Gorini, F. Felix, and C. Fromageot, Biochim. et Biophys. Acta, 12, 283 (1953). 38. R. Monier and M. Jutisz, Biochim. et Biophys. Acta, 15, 62 (1954). 39. C. H. W. Hirs, W. H. Stein, and S. Moore, Abstr. Am. Chem. Soc, N. Y., 1954, p. 89C; C. H. W. Hirs, S. Moore, and W. H. Stein, J. Biol. Chem., in press. 40. S. Moore and W. H. Stein, /. Biol. Chem., 211, 893 (1954). 41. E. W. Davie and H. Neurath, /. Biol. Chem., 212, 515 (1955). Glycogen Turnover DEWITT STETTEN, Jr., and MARJORIE R. STETTEN Since the time of Claude Bernard it has been appreciated by bio- chemists that glycogen represents an animal storage form of glucose. Glycogen is well suited to this storage function for several reasons. Its high molecular weight and low intrinsic viscosity permit its accumu- lation in tissue fluids with but slight effect upon the osmotic pressure or the fluidity. The readily reversible phosphorylase reaction permits both the synthesis of glycogen from hexose phosphate and glycogen breakdown to glucose-1-phosphate to proceed with very small energy transfers. The breakdown of liver glycogen, under hormonal regulation by epinephrine and probably also by glucagon, serves as an elegant homeostatic device to sustain the blood glucose concentration under a wide variety of circumstances. The storage of glycogen in liver or in muscle is not a dead storage. It is, on the contrary, very much alive in that the constituent atoms of the glycogen molecule, even in the animal maintained in a balanced nutritional state, are continuously being regenerated de novo at the expense of a variety of precursors. Since, under such controlled experi- mental conditions, the quantity of glycogen in a given tissue remains approximately constant, coincident with new glycogen synthesis the occurrence of continuous glycogen breakdown must be postulated. The fact that, in the animal in nutritional balance, glycogen of liver and muscle is in a dynamic steady state, undergoing continuous "turn- over," was first observed in the study of the appearance of carbon- bound deuterium in glycogen isolated from animals whose body fluids were enriched with deuterium oxide.1 In these experiments glycogen was isolated from tissues of rats maintained for varying periods of time with their body water enriched with D20, and a progressive enrichment of glycogen with stably bound deuterium was noted. When glucose 291 202 Essays in Biochemistry was abundantly present in the diet, the deuterium concentration in glycogen approached asymptotically a maximal value of about 30% of that in the body water. However, when lactate served as the source of glycogen carbon, glycogen containing about 60% as high a concen- tration of deuterium as body water was recovered,2 and this value approximated the fraction of all hydrogen in glycogen which is carbon- bound. From these results it could be concluded that glycogen was under- going turnover at the expense of available glucose. This glucose was isotopically labeled insofar as it was synthesized in the D20-enriched medium, but was subject to dilution prior to glycogenosis by such glucose as might be derived from the diet. In order to translate the results of these experiments into meaningful quantities, certain assumptions were made and equations were derived designed to relate the change in isotope abundance as a function of time to the turnover rate of glycogen. It could be shown that, if: i = isotope concentration in glycogen at time / imax — maximal isotope concentration, achieved at t = <» then : -I tmax K = - In '■ hnux ~ where k is the turnover rate constant, the fraction of the total pool replaced per unit time. The turnover rate, as weight of glycogen replaced per unit time, could then be evaluated as the product of k and the total weight of glycogen in the tissue. By this method, esti- mates of a half-life of about 1 day for liver glycogen and 4 days for muscle glycogen of rats were made. The several assumptions upon which this treatment rests include the following: 1. That the pool size remains constant over the period of observa- tion; that the rates of entry into and departure from the pool be equal. 2. That these rates in turn be constant. 3. That each new molecule of glycogen entering the pool be enriched with deuterium at the same concentration, namely, imax- 4. That perfect mixing take place; that the deuterium concentration of each molecule of glycogen leaving the pool be the same as the average deuterium concentration of the pool at that time, namely, i. The first three of these assumptions appear to us to be entirely plausible. The last assumption, however, is much more troublesome. Glycogen Turnover 293 It implies perfect mixing of newly introduced with preformed mole- cules; it requires complete randomization of isotopic and non-isotopic molecules and random selection of material from this pool for destruc- tion. For want of a method of approach, this fourth assumption was not explored at that time. Our more recent studies have led to a realization that this assumption is incorrect. Since our more recent exploration of this assumption has depended upon the newer knowledge of glycogen structure and of the several enzymes related to glycogen breakdown, it may be well to digress sufficiently to review the pertinent aspects of these developments. It is generally agreed that glycogen is a branched polysaccharide in which the major glucosidic linkage is a-1,4' and the minor linkage is a-1,6'. The ratio of 1,4' to 1,6' linkages in typical glycogen samples range from 10 to 18.3 Of the three possible branching patterns that have come into serious consideration the arboreal pattern first advocated for amylopectin by Meyer 4 is almost certainly the dominant one. It is in the nature of an arboreally branched polyglucosidc containing n glucose residues that there must be, per molecule, one free reducing end and n — 1 glucosidic bonds. In glycogen the number of branch points should equal the number of a-1,6' glucosidic bonds and will be one less than the number of non-reducing ends. Discrepancies between experimentally secured values and the above expectations have resulted in the suggestion that the structure of glycogen may not be of a simple arboreal nature,5 but, for the purposes of the present discussion, the arboreal structure will be taken for granted. The experimental basis for the above assignment of structure rests largely upon the examination of products secured when glycogen is subjected to various enzymic degradations. AVhen exhaustively treated with /^-amylase, glycogen was found by Meyer to yield maltose and a limit dextrin which represented approximately 50% of the glucose residues initially present. This dextrin was totally refractory to further attack by /3-amylase.6 In Cori's laboratory this product was further studied, and, in addition, a limit dextrin was secured from glycogen by the action of purified phosphorylase and inorganic orthophosphate.7 This latter limit dextrin, LD (glycogen, phosphorylase), insensitive to further attack by phosphorylase, could be rendered phosphorylase- sensitive again by the action of a specific amylo-l,6-glucosidase, which yielded glucose by the selective hydrolytic cleavage of a-1,6' glucoside bonds of exposed glucose residues.8 The exposure of glycogen to simul- taneous attack by phosphorylase and by amylo-l,6-glucosidase yielded 294 Essays in Biochemistry a mixture of glucose- 1 -phosphate and glucose as the sole products, free glucose presumably arising uniquely from the hydrolysis of a-1,6' bonds. The ratio of these products was therefore taken as equal to the ratio of 1,4' to 1,6' links and the abundance of free glucose as a measure of the number of branch points.9 By the alternating attack R, reducing end of molecule. O, ©, and Q, glucose residues removed by first, second, and third degradation with phosphorylasc, respectively. #, glucose residues split off as free glucose by amylo-l,6-glucosidase. Fig. 1. The structure of glycogen (from J. Larner, B. IUingworth, CJ. T. Cori, and C. F. Cori, J. Biol. Chem., 199, 641, 1952). upon glycogen by phosphorylase and amyloglucnsidase, a series of limit dextrins could be secured differing one from another by the successive elimination of tiers of glucosidic residues. Regardless of whether gly- cogen or a susceptible limit detrin thereof served as precursor, the product secured after each treatment with phosphorylase represented a loss of 35-45% of the glucosidic residues initially present.10 A schematic representation of what was happening is shown in Fig. 1, from which it will be seen that the alternating exposure to phospho- rylase and amyloglucosidase has resulted in the peeling of the molecule, removal of successive peripheral layers, much as an onion might be peeled. Glycogen Turnover 295 In view of the lack of information concerning the metabolic homo- geneity of glycogen in tissues of the intact animal, it was considered of interest to re-explore the intimate nature of glycogen regeneration. Is glycogen turnover a process in which a molecule is broken down completely to be replaced by a newly formed one, or does it involve removal and replacement of a fraction of the glucosyl residues within a given glycogen molecule? The experiments designed to this end have taken advantage both of the newer knowledge of glycogen structure and of the specific enzyme activities discussed above. Glycogen ex- hibits a certain structural homogeneity, not shared by many of the mammalian macromolecular substances, in that on total hydrolysis all of its carbon is recoverable exclusively in glucose. The question of whether glycogen constituted, metabolically, a perfectly mixed pool of glucose residues wherein, after isotopic enrichment by one or an- other device, glucose residues, regardless of their location in the mole- cule, had an equal probability of being labeled could now be studied. The basic experimental design has been simple. Into normal rats and rabbits glucose uniformly labeled with C14 has been injected. After various intervals of time, animals were sacrificed and glycogen was isolated from livers and carcasses. After purification of these products, their radioactivities were determined. Samples were then subjected to one or more enzyme digestions, and the radioactivity of the products of such degradations were examined.11 Enzyme digestions, in the earlier experiments, were limited to treat- ment with barley malt /^-amylase. The commercially available prep- aration of this enzyme, although far from pure, is demonstrably free of a-amylase, amylo-l,6-glucosidase, and maltase activity. Glycogen subjected to exhaustive treatment with /^-amylase gives, in excellent yields, maltose and a limit dextrin, LD (glycogen, /3-amylase), which resists all further attack by this enzyme. In effect this enzyme bisects the glycogen molecule into two approximately equal portions, the peripheral cortex of unbranched termini, recoverable as maltose, and the central medullary portion, recoverable as the limit dextrin. When the concentrations of C14 in these two portions of a given sample of glycogen were compared, they were found, in general, to be unequal. Samples of glycogen were studied, which were obtained from rat livers and rat carcasses, from 3 to 48 hours after intraperitoneal injection of glucose-C14. Results are given in Table 1. In the earlier time intervals, the specific activity was in all cases higher in the maltose (periphery) than in the LD (glycogen, /^-amylase). Indeed, despite a relative rise in the specific activity of maltose and a relative fall in 290 Essays in Biochemistry- Table 1. Distribution of Radioactivity in Rat Liver and Carcass Glycogen after Intraperitoneal Injection of Glucose-C14 Relative Specific Activity * In Liver Glycogen In Carcass Glycogen Time, hours Periphery Limit Dextrin f Periphery Limit Dextrin f 3 178 36 127 56 6 146 51 129 56 6 — — 156 56 6 — — 148 59 6 129 74 157 56 G 134 71 150 61 12 92 101 137 65 24 84 105 133 72 24 79 122 124 81 48 64 107 117 89 * Expressed as per cent of specific activity of glycogen. f The limit dextrin here referred to is LD (glycogen, ^-amylase). that of the limit dextrin, even 48 hours after the injection of glucose-C14 the periphery of the carcass glycogen molecule was more radioactive than was the center. Liver glycogen was somewhat different in this regard in that by about the twelfth hour the radioactivity was found to be approximately uniformly distributed between maltose and LD (glycogen, /2-amylase), whereas in the experiments of longer duration the central core of the molecule (limit dextrin) was consistently more radioactive than the peripheral (maltose) tier. The distribution of isotope between the periphery and limit dextrin is plotted in Fig. 2 as a function of time after glucose-C14 injection. In the interpretation of these results it must be borne in mind that the glucose-C14 was all injected at the outset of the experiment and that therefore the specific activity of circulating glucose was constantly declining throughout the period of observation. On the basis of results of Feller et al.12 it may be estimated that the rate of this decline was such as to give a half-time of about 1 hour. Undoubtedly by the twelfth hour the specific radioactivity of circulating glucose available for glycogen synthesis was lower than the mean specific activity of the glycogen into which it was being incorporated. Whereas the earlier data reflect the incorporation of glucose-C14 into glycogen, the later data are a consequence of introduction of glucose-C12 into glycogen-C14. The data strongly suggest that, whereas circulating glucose serves as a fairly immediate precursor of the glucosidic residues situated periph- Glycogen Turnover 297 Fig. 2. Distribution of radioactivity in liver and carcass glycogen between periphery and limit dextrin. The total counts per minute in the maltose and in the limit dextrin (glycogen, j3-amylase) have been determined in each sample of glycogen after exhaustive treatment with /3-amylase. erally in the glycogen molecule, it is these glucosidic residues rather than circulating glucose which give rise to the glucosidic residues situ- ated in the central core of the molecule. This relationship between precursors and products may be expressed by the sequence: Circulating glucose — » Glucosidic residues at periphery of glycogen — > Glucosidic residues at center of glycogen It is interesting to compare the distribution of radioactivity within the molecule of liver glycogen derived from the fasted and from the fed rat. The experiments recorded in Table 2 represent such a com- parison. One rat was fasted for 24 hours prior to glucose-C14 injection ; another had continuous access to diet. Three hours after injection each rat was killed, and the distribution of isotope within glycogen was studied. The striking finding was that, whereas in the liver from the fed rat the major portion of the label was in the periphery (maltose) of the glycogen, the glycogen recovered from the liver of the fasted 298 Essays in Biochemistry- Table 2. Distribution of Radioactivity in Glycogen from Fasted and Fed Rats Killed 3 Hours after Administration of Glueose-C14 Liver Gly cogen Carcass Glycogen Fasted Rat Fed Rat Fasted Rat Fed Rat Specific activity: * Glycogen Periphery Limit Dextrin 33,500 36,600 31,600 366 651 131 3870 5190 2580 369 469 206 Per cent of total c.p.m. in: Periphery Limit dextrin 49.2 50.8 80.2 19.8 58.8 41.2 61.8 38.2 * Specific activity reported as c.p.m. per milliatom C. rat appeared to be labeled nearly uniformly. This equality of labeling of the maltose and limit dextrin shows that, in the liver glycogen recovered when glucose is given after a suitable period of fasting, virtually the entire molecule, center as well as periphery, has arisen from blood glucose. This is clearly a consequence of the depletion of hepatic glycogen which results from fasting. The specific activity of carcass glycogen was far higher when derived from the fasted than from the fed rat, probably because of less dilution of injected glu- cose-C14 by non-isotopic glucose in the fasted rat. Nevertheless, it is noteworthy that the distribution of isotope within this glycogen be- tween maltose and LD (glycogen, /3-amylase) was not greatly different from that in the corresponding sample from the fed animal. This latter finding reflects the well-established observation that muscle glycogen, in contrast to liver glycogen, is not depleted by fasting. It would appear, from these results that glycogen regeneration in tissues involves primarily the addition of glucoside residues derived from blood glucose to non-reducing ends of pre-existing polysaccharide molecules. The reactions whereby this occurs probably include those catalyzed by hexokinase, phosphoglucomutase, and phosphorylase. These reactions alone would account for the appearance, after glu- cose-C14 administration, of radioactivity in the maltose liberated from glycogen by the action of /3-amylase. The introduction of isotope into the limit dextrin (glycogen, ^-amylase) , although a slower process, takes place both in liver and in muscle and indicates the occurrence of another distinct process involving the establishment of new points of branching in the glycogen molecule. Glycogen Turnover 299 To explore this process more fully, it was deemed advisable not merely to bisect the glycogen molecule by treatment with /3-amylase but to dissect it, layer by layer, by the alternating exposure of glycogen to phosphorylase and amylo-l,6-glucosidase (debranching enzyme). To this end the methods devised in Cori's laboratory 3> 8' 10 were repro- duced and applied to glycogen samples which were obtained from animals after injection of glucose-C14.13 A preliminary experiment was conducted upon a sample of glycogen secured from rabbit muscle. This animal was killed 6 hours after intravenous injection of glucose-C14, and glycogen was isolated from muscle and analyzed for C14. Treatment of this polysaccharide with crystalline rabbit-muscle phosphorylase in a phosphate buffer yielded glucose-1-phosphate and a limit dextrin, LD-1- (glycogen, phospho- rylase). This product was assayed for radioactivity and was then treated with a crude rabbit-muscle amylo-l,6-glucosidase solution in the absence of inorganic phosphate. The enzyme was destroyed, and phosphorylase and inorganic phosphate were added, to yield, ulti- mately, a second limit dextrin, LD-2. Repetition of this procedure yielded a third product, LD-3. The extent of each digestion was measured by the sum of the glucose-6-phosphate plus glucose liberated at each step. Radioactivity measurements were performed upon each polysaccharide and, where feasible, also upon phenylglucosazone sam- ples prepared from the products of each successive degradation step. Excellent recoveries of isotope were secured from these combined analyses, so that in later studies only the polysaccharides were analyzed. The results secured are given in Table 3. Included also are the results of a single-step /^-amylase degradation performed upon the Table 3. Distribution of Radioactivity in Rabbit-Muscle Glycogen 6 Hours after Intravenous Injection of Glucose-C14 Relative Specific Per Cent of Glycogen Sample Activity Digested, Cumulative Glycogen 100 0 LD-1 (phosphorylase) 69.2 37.9 LD-2 47.3 63.3 LD-3 29.9 77.7 LD (/3-amylase) 61.3 44 . 0 same glycogen sample. From the values in this table, the specific activities of the glucose residues eliminated by each degradation step 300 Essays in Biochemistry have been evaluated and the results have been plotted in Fig. 3. The ordinates, in this and succeeding figures, represent counts per minute per milliatom carbon, corrected to 100 c.p.m. per milliatom C in the initial glycogen sample. The abscissas represent the percentages of glucosyl residues, initially present in glycogen, which have been elimi- nated by the successive digestions. With reference to the structure of glycogen (Fig. 1), R is the sole reducing end of the molecule which is approached from the non-reducing ends by the repeated enzyme 150 i- y 100 p-1 i — Amylase P-2 ^Average P-3 LD-3 o 2b 50 75 100 Per cent glucose residues Fig. 3. Distribution of radioactivity in rabbit-muscle glycogen 6 hours after in- jection of glucose-C14. treatments. The area in each block represents the total c.p.m. con- tained in that segment of the molecule. From the data in Table 3 it will be seen that each successive limit dextrin obtained was less radioactive than the parent substance from which it was secured. This finding confirms the metabolic inhomoge- neity of glycogen noted previously. From the schematic representation (Fig. 4) of the tierwise degradation of a glycogen molecule, together with the data plotted in Fig. 3, a clearer picture of the distribution of isotope in this sample of glycogen is obtained. Here it will at once be seen that the outer layer of the glycogen molecule (P-1) is more radioactive and the central core (LD-3) is less radioactive than the average of all the glucoside residues in the molecule. As one progresses from the non-reducing ends of this 6-hour glycogen sample centrally toward the reducing end, R, the enrichment of isotope diminishes in a fairly smooth fashion. Glycogen Turnover 301 — P - 3— |— P - 2-f-P - 1-| Fio. 4. Schematic representation of the tierwise degradation of a glycogen molecule. In order to gain insight into the time course of the distribution of isotope in glycogen, studied by this technique, samples of glycogen from rat carcass and liver, which had previously been studied by the /^-amylase method, were reinvestigated. These samples were derived from rats which were killed 6, 12, 24, and 48 hours after intraperitoneal injection of glucose-C14. The results of these studies are summarized in Tables 4 and 5. Table 4. Relative Specific Activity in LD (Phosphorylase) from Rat-Carcass Glycogen 0 Hours 12 Hours 24 Hours 48 Hours Glycogen 100 100 100 100 LD-1 74 75 92 91 LD-2 50 54 80 87 LD-3 45 52 73 70 Table 5. Relative Specific Activity in LD (Phospborvlase) from Rat -Liver Glycogen 6 Hours 12 Hours 24 Hours 48 Hours Glycogen 100 100 100 100 LD-1 74 104 101 115 LD-2 44 108 103 139 LD-3 23 98 117 149 At each point in time studied, rat-carcass glycogen bore certain similarities to rabbit-muscle glycogen described above. The relative 302 Essays in Biochemistry specific activity of each derivative was lower than that of the parent substance from which it was derived. In each sample of carcass glyco- gen (Fig. 5), the outermost tier of glucose residues (P-l) had a higher specific activity than did the central core (LD-3) and, in each case, LD-3 was the least radioactive portion of the molecule. With the passage of time, the difference in specific activity between the periph- eral and the central portions of the molecule diminished slowly. 200 150 > 100 '-3 M o 50 J3 'o Q. 0 (0 CD 150 > 1 100 6HR. 12 HR. - p-l £: 2 P-3 P-l A '-2 P-3 LD-3 LD-3 50 0 L P-l 24 HR. P-3 48 HR. P-2 LD-3 P-l P-2 P-3 LD-3 0 25 50 75 100 25 50 75 100 Per cent glucose residues R Fig. 5. Distribution of radioactivity in rat-carcass glycogen. Rat-liver glycogen, isolated 6 hours after injection of glucose-C14, resembled, in regard to isotope distribution (Table 5), the samples secured from rat carcass. Again each successive limit dextrin was less radioactive than its precursor, and the specific activity decreased reg- ularly as the molecule was entered from the non-reducing termini (Fig. 4) . However, with the passage of time, a striking change occurs which is clearly manifested in the 48-hour sample. Here we find a reversal of the sequence of specific activities of the serial polysaccharides, LD-3 > LD-2 > LD-1 > glycogen. This is reflected in the recon- struction of the glycogen molecule in Fig. 6, where it will be seen that the outermost tier (P-l) is now the least radioactive and that the radioactivity systematically increases as the reducing end of the mole- cule is approached. From these results it is clear that, in addition to a process which adds glucose residues to the non-reducing ends of preformed glycogen molecules, glycogen regeneration entails a second process which results in the introduction of glucosyl residues into inner tiers of the molecule. Glycogen Turnover 303 The presumed source of a glucosyl residue in inner tiers is a glucosyl residue in an outer tier, which hypothesis requires that new branch points be continuously established. A mechanism whereby this may happen has recently been elucidated by Larner,14 who has given the name amylo-(l,4 — » l,6)-transglucosidase to a branching enzyme which generates a-1,6' glucosidic bonds at the expense of a-1,4' bonds. He has demonstrated an effect in vitro which places a labeled glucosyl residue, initially in the outermost tier, into an inner tier of the glycogen 6HR. 150 100 50 0 L_ 150 100 50 0 12 HR. P-l P-2 ^P-3 ^P-3 LD-3 P-l P-2 LD-3 24 HR. P-3. P-l P-L' LD-3 48 HR. P-3 P-l P-2 LD-3 0 25 50 75 100 25 50 75 100 Per cent glucose residues R Fig. 6. Distribution of radioactivity in rat-liver glycogen. molecule. It must be supposed that it is this or some similar mecha- nism which is responsible for the effect which we have observed, namely, the gradual transfer of labeled glucose residues, initially most abundant at the periphery of glycogen, into more and more centrally situated laminae. From these studies a picture, or rather a motion picture, is presented of at least a part of the process of glycogen regeneration. It is quite clear that this is not a process wherein a newly synthesized glycogen molecule is created in place of another one which is destroyed. Rather, new glucose residues are introduced into the glycogen reservoir by a process of continuous accretion at the periphery of pre-existing glyco- gen molecules. By a second process, involving establishment of new branch points, glucose residues in the glycogen molecule become pro- gressively more remote from the outermost tier of glucosyl residues. Since, under most circumstances, including those of the present experi- 304 Essays in Biochemistry ments, the animal is in approximate nutritional balance and analyt- ically in an approximately steady state, the process of glycogen growth must be offset by a process of glycogen decay. The steps involved in this process are presumed to be the phosphorolytic uncoupling of peripherally situated glucosyl residues operating in conjunction with the hydrolytic cleavage of exposed glucosyl residues in a-1,6' linkage due to the action of the debranching enzyme, amylo-l,6-glucosidase. It should be noted that the description just given of glycogen turn- over may also serve to explain glycogen accrual and depletion. If this proves to be the case, it represents an unusual mode of storage in that no change in the number of glycogen molecules is postulated. If glycogen accrual results merely from the action of phosphorylase and branching enzyme, if depletion is the result simply of phosphorylase and debranching enzyme activity, then changes in glycogen content, in contrast to alterations in fat or protein content, are reflections of changes in mean molecular weight of glycogen, the number of molecules remaining constant. This would be a happy arrangement in that it would permit large fluctuations in the magnitude of the glycogen re- serve to occur with no accompanying changes in the colligative proper- ties, notably osmotic pressure, of tissue fluids. Regardless of whether the count of glycogen molecules in a given adult tissue is or is not constant, it is clear that during growth the number of glycogen molecules must at some stage increase. Within the limits of our present understanding of the actions of phosphorylase, amylo-l,6-glucosidase and amylo-(l,4 -^ l,6)-transglucosidase, no ex- planation for replication of glycogen molecules exists. Replication, when and if it does occur, may result from the a-amylolytic activity of tissues, generating a plurality of seeds for glycogen synthesis from a single glycogen molecule. It appears likely that much of what has been said about the nature of glycogen synthesis in the living animal may also apply, with very minor modifications, to the synthesis of the less highly branched plant polysaccharides, the starches. Certain differences exist, however, which may raise the question of whether polysaccharide synthesis occupies the same crucial position in animal as it does in vegetable economy. If attention is momentarily focused upon glycogen of liver, it will be recalled that this is considered to be a mobile reservoir of glucose and of energy. It is, however, a small reservoir, amounting calorically in the rat to some 3% of the total daily caloric requirement. Further- more, after it has been virtually eliminated, as by fasting or by injec- Glycogen Turnover 305 tion of epinephrine, the animal is not demonstrably sick and indeed under most circumstances its chance of survival has been but little diminished when compared with that of a litter mate whose liver is rich in glycogen. Lastly, the major process of glycogenesis, the phos- phorylase reaction, is the precise reversal of the major process of glycogenolysis. In other words the route in is the reversal of the route out ; hence glycogen is a sort of biochemical blind alley. In summary, the reserve of liver glycogen is small, the animal does quite well with- out it, and it is a cul-de-sac. It will be recalled that on these very grounds the vermiform appendix in man has been designated as a vestigial organ. Is liver glycogen then to be considered a biochemical vestige? The probable answer is in the negative, in view of the fact that under certain stresses of short duration, liver glycogen does serve as a useful source of extra blood glucose. The fact remains, however, that, whereas the well-nourished potato accumulates polysaccharide, the well-nourished rat accumulates characteristically not glycogen but lipid. The major elastic compartment for energy storage in adult mammals is the depot fat, not the carbohydrate compartment. This fact was drawn to our attention some years ago when we had occasion to compare rates of glycogenesis and lipogenesis in adult and fetal rats.15 Although the fetus is, until shortly before term, very poor in depot fat, its glycogenic capacity, per gram of tissue, is far greater than that of the adult. The possibility that this difference in fetal and adult habits might represent a sort of recapitulation of phylogeny suggests itself. In quest of a more satisfactory explanation for the lack of fat storage in fetal tisues, one is forced to consider the generally stated functions of subcutaneous depot fat. This adipose tissue serves as an energy reservoir, it functions as mechanical upholstery against physical trauma, and it acts as thermal insulation in homothermic species. For none of these functions does the mammalian fetus have any need. It is nourished not discontinuously but continuously by the maternal circulation. It is well protected against physical trauma from without by maternal tissues and amniotic fluid. It resides in a precisely thermoregulated environment. It is therefore not surprising that the fetus receives its investiture of subcutaneous adipose tissue only shortly before term, as if in anticipation of its ejection from its highly pro- tected environment. A further inspection of the difference in energy-storage habits of 306 Essays in Biochemistry plants and animals reveals certain interesting exceptions. Whereas plants in general accumulate polysaccharides and are poor in lipids, many important vegetable oils do occur. With few exceptions, how- ever, major lipid storage in plants is restricted to seed parts. Again, whereas it is the habit of animals to accumulate lipid rather than polysaccharide, in the tissues of molluscs one finds large amounts of glycogen and but little fat. These exceptions suggest the possibility that the habit of preferential lipid storage is an adaptation to motility. An obvious advantage of lipid storage over polysaccharide storage is that, per calorie stored, lipid weighs far less than carbohydrate. This is of little survival benefit to sessile forms of life such as the higher plants, but it may be of real importance to the mouse, who must evade his natural enemies, or the cat, who must capture the mouse. Of all the tissues of higher plants, it is uniquely in the seed parts that light weight may be considered to be of survival value. It is necessary both for the survival of the individual and of the species that seeds be disseminated, and, regardless of the vector, the less a seed weighs the higher the probability that it will be carried away from its source. Molluscs are undoubtedly motile, but of all animal species few are better protected against natural enemies and few have less need for motility in order to survive. What has been said of molluscs may also be said of the mammalian fetus. From the foregoing it will be seen that in the starch-glycogen group of polysaccharides nature has produced a system peculiarly and ele- gantly adapted to the function of energy storage. Each glycogen mole- cule acts as a minute and elastic glucose reservoir at a molecular level and is continuously acquiring and losing glycosyl residues at its pe- riphery. As an energy reservoir it suffers, however, in comparison with fat in that, per calorie stored, it adds far more to organism weight. Although this is unimportant to the survival of such forms as do not depend upon motility, such as the tuber, the tree, the mollusc, or the fetus, it may be of importance to the plant seed and the adult mammal. In these latter forms the habit of lipid storage has largely supplanted the habit of polysaccharide accumulation. Reverting to the question of whether mammalian-liver glycogen is a biochemical vestige, we are still inclined to the opinion that it is not. It is, however, regarded as entirely possible that a transfer from a predominantly glycogenic habit to a predominantly lipogenic habit has real survival value to many motile forms of life and that this change has, in the long course of evolution, actually occurred. Glycogen Turnover 307 References 1. D. Stetten, Jr., and G. E. Boxer, J. Biol. Chem., 155, 231 (1944). 2. G. E. Boxer and D. Stetten, Jr., J. Biol. Chem., 155, 237 (1944). 3. B. Illingworth, J. Lamer, and G. T. Cori, J. Biol. Chem., 199, 631 (1952). 4. K. H. Meyer and P. Bernfeld, Helv. Chim. Acta, 23, 875 (1940). 5. C. O. Beckman, Ann. N. Y. Acad. Sci., 57, 384 (1953). 6. K. H. Meyer and M. Fuld, Helv. Chim. Acta, 24, 375 (1941). 7. S. Hestrin, J. Biol. Chem., 179, 943 (1949). 8. G. T. Cori and J. Larner, J. Biol. Chem., 188, 17 (1951). 9. B. Illingworth, J. Larner, and G. T. Cori, J. Biol. Chem., 199, 631 (1952). 10. J. Larner, B. Illingworth, G. T. Cori, and C. F. Cori, ./. Biol. Chem., 199, 641 (1952). 11. M. R. Stetten and D. Stetten, Jr., J. Biol. Chem., 207, 331 (1954). 12. D. D. Feller, E. H. Strisower, and I. L. Chaikoff, J. Biol. Chem., 187, 571 (1950). 13. M. R. Stetten and D. Stetten, Jr., J. Biol. Chem., 213, 723 (1955). 14. J. Larner. J. Biol. Chem., 202, 491 (1953). 15. W. H. Goldwater and D. Stetten, Jr., J. Biol. Chem., 169, 723 (1947). The Veratrum Alkamines OSKAR WINTERSTEINER This essay attempts to give a brief account of the advances made in the structure elucidation of the veratrum alkamines, and in particu- lar to bring out how the peculiar dichotomy of skeletal structure which sets these alkaloids, as a group, apart from other families of steroidal bases came to be recognized. Naturally only the facts most directly relevant to the elaboration of the presently accepted structures can be presented here. The literature cited covers most of the more recent work, except that on cevine, for which the reader is referred to the bibliography in the comprehensive 1954 paper on the constitution of this alkamine by Barton et al.1 More detailed if slightly outdated resumes of the chemical and other aspects of the subject can be found in review articles.2 Of the more than two dozen veratrum alkaloids now known the majority is of the conjugated type (esters or glucosides) ; the much smaller group of unconjugated bases (alkamines) from which these are derived comprises only eight well-characterized members. They all contain twenty-seven carbon atoms and one nitrogen atom, which in all but two instances (jervine, C27H;{903N; veratramine, C27H39O2N) is tertiary. The tertiary bases include rubijervine and isorubi jervine (C27H43O2N) , and a group of highly oxygenated compounds (zygade- nine, C^H^C^N; * cevine, genuine, C27H43OsN; and protoverine, C27H43O9N), which occur in nature predominantly in combination with acids, i.e., as the alkamine moieties of the hypotensically active and hence medicinally important ester alkaloids. Much of our present knowledge of the chemistry of the alkamines we owe to the fundamental studies of W. A. Jacobs, who, together with L. C. Craig, began exploring this subject in 1937. In the initial phase * This alkamine, together with certain ester alkaloids derived from genuine, occurs in Zygadenus venenosus, which is not a member of the genus Veratrum. 308 The Veratrum Alkamines 300 of this work the prime objective was to gain insight into the nature of the underlying carbon-nitrogen skeleton by means of soda-lime distillation and selenium dehydrogenation. This approach proved to be immediately fruitful at least as far as the nitrogenous portion of the molecule was concerned. Particularly informative in this respect was the finding that 3-methyl-6-ethylpyridine (I) was formed on selenium dehydrogenation from all the alkamines then so investigated, and that the corresponding piperidine base was formed from cevine on soda-lime distillation. There could be no doubt then that the nitrogenous moiety was a substituted piperidine ring, joined, as it turned out later, to the rest of the molecule through the a-carbon of the ethyl group and, in the tertiary bases, also through the nitrogen atom. The structural significance of three other basic dehydrogenation products, 3-methyl-5-hydroxypyridine (II), from veratramine, a base which in all probability is 3-methyl-5-hydroxy-6-ethylpyridine (III), from jervine, and an isomer of jervine recently 1 assigned structure IV, from cevine, will become evident later. It will be noted that the carbon skeleton of I is that of the cholesterol side chain. It was then not unreasonable to adopt the working hy- pothesis that the remainder of the molecule was made up of the androstane skeleton. Strength was added to this supposition when it was found (Prelog and Szpilvogel, 1942; Craig and Jacobs, 1943) that solanidine, C07H43O, the main alkaloid of the potato plant (Solatium tuberosum) , then already known to be a steroid, likewise contained its nitrogen atom in a 3-methyl-6-ethylpiperidine moiety. Against it, however, stood the fact that painstaking fractionation of the neutral dehydrogenation products from jervine and cevine (up till then the only alkamine so investigated), had failed to disclose the presence of any phenanthrene derivatives. In fact, with the exception of a tricyclic compound from cevine, 1,2-cyclopentenonapthalene (V), none of the numerous physically and analytically well-characterized hydrocarbons obtained from these alkamines could be identified structurally, and some of the tetra- and pentacyclic members of the series exhibited ultraviolet characteristics which indicated a relationship to fluorene (VI) and 1,2-benzofluorene (VII) rather than to phenanthrene. For this reason Jacobs and Craig for a time gave consideration to a modi- fied steroidal nucleus with a five-membered ring B and a methyl group at C5. However, when in 1943 they shifted their attention to rubi- jervine, and later to its newly isolated isomer isorubijervine, it became increasingly evident that these two alkamines, at any event, must be normal steroids. Thus, rubijervine yielded on selenium dehydrogena- 310 Essays in Biochemistry tion an isomer of Diels' hydrocarbon, in all probability l'-methyl-l,2- cyclopentanophenanthrene (Villa), and isorubijervine, 1,2-cyclopen- tanophenanthrene (VIII6) itself. Furthermore, the presence of the usual 3-hydroxy-5,6- double-bond grouping in both compounds could be readily demonstrated by oxidation to the corresponding A4-unsatu- rated ketones and reduction of the ketones to the Rosenheim-positive allylic alcohols corresponding to allocholesterol. C..H CH HO CH. HO C,Hr CH, CH( CH,CH ^N OH IV VI VII Villa (R=CH, VIII b (R=H) The last vestige of doubt regarding the significance of these results vanished when it could be shown that both these alkamines were indeed derivatives of solanidine (IX), which differed from this base merely by the presence of a second hydroxy 1 group. With rubijervine (X), the conversion to solanidine was effected by selective oxidation of this group, for sound reasons assumed to occupy the 12 position and to be a oriented, and Wolff-Kishner reduction of the resulting 12-monoketone.3 In isorubijervine (XI) the additional hydroxyl group is primary (formation of a monoketomonocarboxylic acid C27H41O3N on oxidation of the 5,6-dihydro derivative 4 and resides on the methyl carbon atom 18. This site was first deduced 4 from the absence of The Veratrum Alkamines 311 methyl substitution in the cyclopentene ring of the dehydrogenatioE product VIII6, but a more cogent argument evolved from the qua- ternary base nature of the intermediates Xlla and XII6 (originally regarded as the normal 18-O-tosylate and 18-iodide 5 in the sequence by which isorubijervine was converted to solanidine; 5 7 clearly, of the methyl carbons in the vicinity of the basic group only C-18 fulfills the steric requirements for facile formation of a bond to the nitrogen atom. IX (R = H) Solanidine X |R = ---OH) Rubijervine CH..OH XI Isorubijervine Ci.iH.mO Na, C,HsOH IX Xlla (X = p-CH:i-C6H4-S03) Xllb (X = I| With the presence of a normal steroidal nucleus in these two bases assured, it was only natural that in the tentative expressions which had meanwhile been proposed for jervine 8 and cevine 9 these alkamines were also accorded the normal structure. Implicit in these formula- tions was the assumption that the "abnormal" fluorene-like hydrocar- bons formed from these bases in the selenium dehydrogenation were probably artifacts resulting from skeletal rearrangements such as some- times occur in this high temperature reaction. That, to the contrary, these products accurately mirror the original ring system was first 312 Essays in Biochemistry brought out in a reinvestigation of jervine, begun in 1950 in the labora- tory of the writer, which definitely established the perhydrobenz- fluorene structure XIII for this alkaloid. Much valuable information on jervine was already on hand from the work of Jacobs with Craig, and with Sato.8 Thus the evidence adduced by these authors for a normal A/B ring system with a hydroxyl group at C-3 and a 5,6 double bond was incontestable. The function of the other two oxygen atoms present could not be ascertained by simple chemical means. One of these was assumed to be oxidic and to link C-23 in the piperidine ring (corresponding to the position of the hydroxyl group in the dehydrogenation product III) to C-16 in a normal ring D. The second inert oxygen atom was revealed as part of an a,/?-unsaturated ketone grouping by the absorption spectrum {Xmax, 250 nut). In view of the complete lack of reactivity towards ketone reagents the ketonic carbonyl was placed in position 11 of a normal six-membered ring C, and this in turn necessitated accom- modating the conjugated double bond in 8, 9. That the latter proposition was incorrect became apparent from a study of the ultraviolet characteristics of diacetyl-7-ketojervine and its 5,6-dihydro derivative,10 which showed that the new a, ^-unsaturated ketone system created in the former compound by the introduction of the 7-keto group could not be contiguous with that pre-existing in jervine, and hence that ring C, if containing the unreactive keto group, could not be normally constituted. In the further pursuit of the problem the use of acetolyzing agents proved to be decisive. Thus, treatment of jervine with boiling acetic anhydride and zinc chloride yielded a nitrogen-free compound, C2.3H30O3, which contained the un- reactive keto group as part of a dienone system. Its structure XIV followed in essence from its oxidative degradation to acetaldehyde and the en-l,4-dione XV and the alkali-induced aromatization of the latter to the phenol XVIa, the absorption spectrum of which left no doubt as to the presence of an a-indanone (or a-tetralone) system carrying a phenolic hydroxyl group in the position indicated. The alternative structures XVIb and XVIc could be excluded on the grounds that the keto group which had survived the aromatization was completely un- reactive to ketone reagents.11-12 When milder acetolytic conditions were employed, ring E was opened without loss of the side chain and the reaction led through the inter- mediate XVII to the indanone XVIII, the structure of which was readily deducible from its spectrum and from the fact that on reduction it formed a product showing benzenoid absorption.11'13 The presence The Veratrum Alkamines 313 CH - CH, AcO XXII Veratramine 314 Essays in Biochemistry of a new, non-acetylatable hydroxy group in the precursor XVII of the indanone provided the information that the oxidic bridge of jervine was attached in ring D to the tertiary carbon atom 17. It is interesting, and perhaps of biogenetic significance, that this feature renders XVII, and generally "open" derivative of this type, prone to undergo re- arrangements catalyzed by alkali in which the nitrogen atom, after losing its acetyl group by N — > 0 migration to that hydroxyl group, adds to an activated and sterically favored site in ring D. Thus XVII on O-deacetylation with cold alkali rearranges to the sterically hin- dered and hence very weak tertiary base XIX,13 whereas the olefin XX, an acetolysis product of tetrahydrojervine, on hydroxylation with osmium tetroxide and treatment of the adduct with sodium sulfite yields, besides the normal 16,17-glycol, a tertiary amine of type XXI,14 reminiscent of rubijervine. A concurrent study of the secondary base veratramine 15 showed that this alkamine was closely related to the acetolysis product XVIII from jervine; in fact it differed from it structurally only by the absence of the 11-keto group. The presence of a preformed benzene aromatic ring, first deduced from the absorption spectrum (Jacobs and Craig, 1945), rests on solid chemical evidence,15 as does the allocation of the two secondary hydroxyl groups to positions 3 and 23, and of the double bond to the 5,6 position.15*16 Structural correlation with XVIII and hence with jervine was achieved through the 5,6-dihydro derivative of XVIII,17 which proved to be identical with an indanone-like com- pound found among the chromic acid oxidation products of triacetyl- dihydroveratramine.15 Veratramine is therefore XXII. The finding that on permanganate oxidation it afforded benzene-l,2,3,4-tetracar- boxylic acid,18 aside from disposing of an earlier expression advanced by Jacobs and Sato 16 which differed from XXII only by inclusion of C-18 in a six-membered ring C, made secure the allocation to posi- tion 17a of the methyl group representing that carbon atom. In the sequel Jacobs and Pelletier,19 through a careful re-evaluation of the absorption spectra of the "abnormal" dehydrogenation products and some of their partly hydrogenated derivatives, came to the con- clusion that the perhydrobenznuorene skeleton should be allowed not only to the two secondary bases but also to cevine and the other tertiary ester alkamines. Particularly indicative in this respect were two large basic fragments from cevine not previously mentioned here, namely, cevanthridine, C25H2-N, and veranthridine, C26H25N, now formulated, respectively, as XXIII and XXIV. The characteristic ease with which these compounds formed the corresponding 11 -ketones The Veratrum Alkamines 315 ^ XXX (R = CH3CH:C(CH3)CO-) XXIX (X = CH3-C— 0 ) O XXXIII Cevagenine XXXII (R = H) Veracevine XXXIIa (R = CH3CH:C(CH3)CO-) Cevadine XXXIIb (R = 3,4-(MeO)2C6H3CO-) Veratridine XXXIIc (R=CH3CO-) Cevacine 316 Essays in Biochemistry OAc XXXlVb OAc 7J O O HC0 C^O^^ XC^CH3)2 XXXIV a > 1 tert. OH XXXV Germine (9-fluorenones) on oxidation is quite in line with the assigned struc- tures.20 The contraction of ring A in XXIII may have its cause in the multiple substitution of this ring, in cevine, with oxygen (cf. below) . On the other hand, this explanation is not applicable to some of the dehydrogenation products of jervine which likewise show this feature. With the skeletal structure of cevine thus in essence defined by XXIV, there remained still the formidable task of determining the location of its eight oxygen atoms, all but one of which could be assumed to be hydroxylic. Studies towards this end had been in prog- ress since about 1951 in several laboratories, including those of Barton (London), Jeger and Prelog (Zurich), and Woodward (Harvard). Before long these investigators, by pooling their information, were able to advance in a joint communication 1 structure XXV for cevine. It is not possible to do justice here to the intricate and ingenious argument essential to the elaboration of this formula, nor to survey the host of experimental facts, some already on record from the work of Jacobs, and many others of more recent vintage, which could be marshalled The Veratrum Alkamines 317 in its support. A very brief and cursory exposition of the main lines of evidence will have to suffice. One of these developed out of a reconsideration of the properties and mode of formation of decevinic acid, C14H14O6, a degradation product already extensively explored by Jacobs and Craig and now shown to be XXVI. Decevinic acid is formed by pyrolytic dehydration of one of the chromic acid oxidation products of cevine, a lactone tricarboxylic acid of the composition Ci4HiSOs. The recognition of this lactone as XXVII, partly implicit in XXVI, and of another closely related oxidation product, a hexane- tetracarboxylic acid Ci0Hi4O8, as XXVIII showed these fragments to be derived from a normal A/B ring system substituted with oxygen at C-3, C-4, and C-9, and beyond that provided confirmatory evidence (preservation in XXVII of carbon atom 13 * in the oxidative scission) for ring C being five-membered ( Jeger, Prelog, Woodward) . At this point it was possible to accommodate in the positions men- tioned a masked secondary a-ketol system which had been disclosed by the work of Barton, and further to place tentatively an independent ditertiary glycol grouping (Stoll and Seebeck, Barton) at the C/D bridge atoms 13 and 14. More decisive evidence on this latter point, as well as on the location of one other hydroxyl group, derives from the properties and reactions of the "anhydrocevine" formed from cevine on vigorous acetylation followed by hydrolysis (Stoll and Seebeck). This compound is now shown to be a tritertiary orthoacetate (Barton) which must be formulated as XXIX, since other possible sites for the hydroxyl groups linked together in the orthoacetate grouping may be excluded on the strength of periodate titration data, as well as for other reasons (for instance, XXX below) . The presence of a secondary hydroxyl group at C-16 is deduced, inter alia, from the structure (XXX) assigned to a product which is formed from the ester alkaloid cevadine (XXXIIa) on chromic acid oxidation. This compound is undoubtedly a phenolic indanone and is thought to arise from the intermediary seco-13,14-diketone XXXI by transannular Claisen con- densation (C-13^C-15), followed by dehydration at C-9, and aro- matization of ring C (Jeger, Prelog) . The allocation of the remaining hydroxyl group, which must be tertiary, to C-20 rests in essence on the (as yet unproved) structure of the dehydrogenation product IV. It is now well established that an isomer of cevine, veracevine, and not cevine itself is the true native alkamine occurring in the ester * Numbering according to proposals for steroid nomenclature now before Union internationale de chimie pure et appliquee (cf. ref. 12). Barton et al.1 designate the carbon atoms here numbered 13 and 17a as 12 and 13, respectively. 318 Essays in Biochemistry alkaloids (Pelletier and Jacobs, Kupchan et al.). Veracevine, which can be obtained from the latter by methanolysis, is transformed into cevine under the conditions (vigorous alkaline hydrolysis) generally used for liberating the alkamine moiety. Yet another isomer, cevage- nine (Stoll and Seebeck) , is formed from veracevine on mild treatment with alkali. Cevagenine differs from cevine and veracevine in that it contains a ketonic carbonyl. These isomers are formulated by Barton et al. as shown in XXXII, XXXIII, and XXV, and accordingly the three known ester alkaloids, cevacine, veratridine, and cevadine, have to be expressed as XXXIIa, b, and c, respectively. It would lead too far afield to review here the evidence for the configurational assign- ments for C-3, C-5, and C-9 in these isomers and for the asymmetric centers in the remainder of the molecule. Genuine, a native alkamine which occurs in nine of the ester alka- loids presently on record, is under the influence of alkali analogously isomerized to the ketonic isogermine 21 and further to the non-ketonic pseudogermine,22 and before long evidence was at hand showing that this behavior is referable to the presence of an isomerizable a-ketol system similar to that of veracevine 23a but differing from it in that the terminus of the hemiacetalic ether bridge is C-7 and not C-9.23b This was deduced from: (1) The finding that all three genuine isomers form tetraacetates (which is possible only if the hydroxyl function at that terminus is primary or secondary). (2) The fact that the hemiacetalic ring is five-membered (formation of an aldehydo-y- lactone on periodate oxidation of genuine (or pseudogermine) ace- tonide.23a (3) The exclusion of the alternative non-tertiary y-positions 1 and 19 for that hydroxyl function on the grounds that germine, like cevine, can be degraded to the tetracarboxylic acid XXVIII which carries no oxygen at these carbon atoms. Finally Kupchan and Narayanan,24 on the basis of additional but perhaps somewhat less compelling evidence which cannot be given here in detail, came to ascribe structures XXXI Va and b, respectively, to the aforementioned germine acetonide aldehydo-y-lactone and its diacetate and thus to formulate germine itself as XXXV. Although structural information on the other two tertiary alkamines of this group, zygadenine and protoverine, is still scant, there can be little doubt that they, too, conform with cevine in regard to the nature of the skeleton and the presence of an isomerizable ketol system. Thus, except for what remains to be done on these two alkamines, the task of the organic chemist in this field has been largely accom- plished. With the coexistence in several species of two principal groups The Veratrum Alkamines 319 of alkaloids differing in their skeletal structure clearly established, it is legitimate to speculate on mechanisms by which the perhydrobenz- fluorene system, observed here in nature for the first time, might arise biogenetically from the normal steroid nucleus. The structures now before us, as such, are not informative in this respect, except perhaps insofar as it can be inferred from the relative prevalence of functional groups, or of equivalent unsaturation, in the C/D ring moiety that this portion of the molecule has lived through an eventful biogenetic past. However, a significant clue (though it is one derived from purely organic-chemical facts and hence, in the eyes of biogeneticists, perhaps not too trustworthy) is available from the work of R. Hirschmann, N. L. Wendler, and their colleagues on ring C/D rearrangements in certain sapogenins appropriately substituted with oxygen in ring C.25 XXXVI Rockogenin CH, XXXVII XXXIX 320 Essays in Biochemistry- It was shown by these investigators that rockogenin (12/3-hydroxy- tigogenin, XXXVI), when treated in the form of its 3-methylsuccinate- 12-mesylate with potassium £-butoxide in £-butanol, or with the alcohol alone, was transformed in fairly facile reaction into a mixture of the C-nor/D-homospirostene XXXVII and its 17,17a double-bond isomer. Similarly, the 12-toluene-p-sulfonylhydrazone derivative of 11-keto- hecogenin (XXXVIII) with alkali yielded the rearrangement product XXXIX, reminiscent of jervine. Since an ll/?-hydroxy-12-keto bile acid also undergoes this rearrangement, it is clear that the constitu- tional requirements are rather simple: a 12-keto or 12/J-hydroxyl group in an activated state such as supplied here by the tosyl group in the derivatives used. The authors suggest that the veratrum alkaloids possessing the modified nucleus may arise in the plant by rearrange- ments of this kind from similarly constituted precursors. Against this hypothesis can be held the fact that there is no evidence for the occurrence in the plant sources of rockogenin and hecogenin of products of the type obtained by Hirschmann et al. The question could be presumably settled by growing veratrum plants in the presence of 1-C14- and 2-C14-labeled acetate and determining the isotope distribu- tion in suitable degradation products of the more abundant "abnormal" alkaloids, but, aside from the great practical difficulties inherent in a project of this kind, it would obviously have to await the demonstra- tion of a constant distribution pattern of acetate carbons in several classes of plant steroids including normally constituted steroid alka- loids, since so far it is only a presumption that the pattern ascertained for the cholesterol nucleus holds true for all steroids. References 1. D. H. R. Barton, 0. Jeger, V. Prelog, and R. B. Woodward, Experientia. 10, 81 (1954). 2. O. Wintersteiner, Record Chem. Progr., 14, 19 (1953); J. McKenna, Quart. Rev. Chem. Soc, 7, 231 (1953) : V. Prelog and 0. Jeger, in The Alkaloids. Chem- istry and Physiology, edited by R. H. Manske, 7/7, p. 231, Academic Press, New York, 1953. 3. Y. Sato and W. A. Jacobs, J. Biol. Chem., 179, 623 (1949). 4. Y. Sato and W. A. Jacobs, J. Biol. Chem., 191, 63 (1951). 5. S. W. Pelletier and W. A. Jacobs, J. Am. Chem. Soc., 74, 4218 (1952). 6. F. L. Weisenborn and D. Burn, J. Am. Chem. Soc, 75, 259 (1953). 7. S. W. Pelletier and W. A. Jacobs, J. Am. Chem. Soc, 75, 4442 (1953). 8. W. A. Jacobs and Y. Sato, J. Biol. Chem., 181, 55 (1949). 9. A. Stoll and E. Seebeck, Helv. Chim. Acta, 36, 189 (1953). The Veratrum Alkamines 321 10. 0. Wintersteiner, M. Moore, J. Fried, and B. M. Iselin, Proc. Nat. Acad. Sci., 87, 333 (1951). 11. J. Fried, O. Wintersteiner, M. Moore, B. M. Iselin, and A. Klingsberg, J. Am. Chem. Soc, 87, 333 (1951). 12. J. Fried and A. Klingsberg, J. Am. Chem. Soc, 75, 4929 (1953). 13. 0. Wintersteiner and M. Moore, J. Am. Chem. Soc, 75, 4938 (1953). 14. 0. Wintersteiner, M. Moore, and B. M. Iselin, /. Am. Chem. Soc, 76, 5609 (1954). 15. C. Tamm and 0. Wintersteiner, /. Am. Chem. Soc, 74, 3842 (1952). 16. W. A. Jacobs and Y. Sato, J. Biol. Chem., 191, 71 (1951). 17. O. Wintersteiner and N. Hosansky, J. Am. Chem. Soc, 74, 4474 (1952). 18. 0. Wintersteiner, M. Moore, and N. Hosansky, /. Am. Chem. Soc, 75, 2781 (1953). 19. W. A. Jacobs and S. W. Pelletier, J. Org. Chem., 18, 765 (1953). 20. S. W. Pelletier and W. A. Jacobs, /. Am. Chem. Soc, 76, 2028 (1954). 21. H. Jaffe and W. A. Jacobs, J. Biol. Chem., 193, 325 (1951). 22. S. W. Pelletier and W. A. Jacobs, J. Am. Chem. Soc, 75, 3248 (1953). 23. S. M. Kupchan, M. Fieser, C. R. Narayanan, L. F. Fieser, and J. Fried, (a) J. Am. Chem. Soc, 76, 1200 (1954); (b) ibid., 76, 5259 (1954). 24. S. M. Kupchan and C. R. Narayanan, Chemistry and Industry, 1955, 251. 25. R. Hirschmann, C. S. Snoddy, Jr., C. F. Hiskey, and N. L. Wendler, /. Am. Chem. Soc, 76, 4013 (1954), and earlier papers. The Chemical Basis of Heredity Determinants STEPHEN ZAMENHOF Students of the history of science cannot fail to notice how often the issues most essential for our species are continuously avoided. Thus, the science of heredity which affects us more than astronomy was practically non-existent until the second half of the nineteenth century although Mendel's conclusions (1866) were actually much easier to arrive at and to accept than those of Copernicus (1530) or of Harvey (1628). As late as 1872 Spencer writes: "We are obliged to confess that Life in its essence cannot be conceived in physico- chemical terms." By the end of the first half of this century the foundations for chemi- cal explanation of several biological phenomena had already been laid. However, the phenomenon of heredity was not attractive to the chemist, perhaps because of a fear of the multitude of substances involved and their insurmountable complexity; indeed, in contrast to simpler bio- chemical functions, nothing less than the whole cell or at least the chromosomal apparatus seemed indispensable as heredity determinant. The remains of this Spencerian attitude still hamper modern research on the chemistry of the transmission of heredity. Heredity Determinants Although the hypothetical "working gene" may be, indeed, a very complex system, not all the elements of such a system need be essential for the determination of heredity: a few may be actual heredity determinants, and all the others merely auxiliary elements. This essay deals with the search for the decisive factors. If one approaches the problem with an unbiased mind one also has to test the possibility that the heredity determinants are not chemical 322 The Chemical Basis of Heredity Determinants 323 substances. A set of "genes" could be, for instance, a set of specific reactions going on, or a specific distribution of independent molecules; such suggestions were, indeed, made in the past. To dispose of these notions one has only to turn to the simplest living entities, the crystal- line viruses. Each of these consists of a single molecule of nucleopro- tein; there are no reactions going on, no distribution of independent molecules because only one is present. Since it is a matter of common sense (or definition) that each living entity carries its own heredity, one has to conclude that the only heredity determinants of these viruses are indeed chemical substances: either nucleoproteins as such or their components, nucleic acids * and/or proteins. One could argue that perhaps a different principle is involved in higher organisms; however, as will be shown below, the evidence there points in the same direction. The next more complex living entities studied in this respect are the bacterial viruses. The viruses of Escherichia coli consist mainly, but not exclusively, of nucleoprotein, about 40% of which may be nucleic acid (here deoxyribonucleic acid or DNA). As shown by Hershey and Chase,1 upon infection practically only the DNA of the virus reaches the inside of the host and is allowed to reproduce (deter- mine) the new virus particle; the DNA, then, must be the only heredity determinant of these species. In bacteria, the discovery of the transforming phenomenon (Grif- fith -) and of the nature of the transforming principle (Avery, Mac- Leod, and McCarty 3) made it clear that there, too, the DNA alone is capable of acting as a heredity determinant; however, the question as to whether the DNA is the sole or only one of many heredity deter- minants still remains debatable. The subject of transforming phe- nomena will be discussed in detail below. The situation in higher organisms is, naturally, mure complex, and evidence of the above-mentioned degree of validity has not yet been presented; however, there, too, all the existing evidence, though in- direct, points to DNA as a heredity determinant. The spermatozoa, which carry the entire heredity of the male, may contain over 90% of deoxyribonucleoprotein (referred to dry weight) ; the high deoxyribonucleoprotein content of somatic chromosomes, un- doubtedly concerned with the orderly transmission of heredity, is well realized. But, although the composition and gross structure of DNA seem to remain unchanged when the spermatozoon (and ovum) * The nucleic acid involved here is the ribonucleic acid (RNA) ; this RNA, however, seems to consist of giant molecules resembling more, in this respect, the cellular DNA than RNA. 324 Essays in Biochemistry changes into somatic cells of the same heredity (for a review see ref . 4) , the composition and structure of nuclear protein may undergo drastic changes: the change of protamines into histones is but one example. In addition, the study of the quantity of DNA and of protein (per single set of chromosomes) revealed that, although the quantity of DNA remains the same in all cells of a species,5 the quantity of protein varies a great deal. All these findings can be taken as an indication (though not as an absolute proof) that in higher species also the DNA serves as a determinant of heredity. This concept encountered difficulties, however. Workers in the field were still influenced by the authority of "old masters" (especially P. A. Levene 6) who, on the basis of erroneous chemical analysis, con- cluded that the DNA from all sources contains equimolar amounts of individual purines and pyrimidines; that these form subunits ("tetra- nucleotides") ; and that, in fact, all deoxyribonucleic acids are identical, being merely composed of identical "tetranucleotides." Lehmann- Echternacht even reported that he had actually isolated this (purely fictitious) substance. Of course, the identical and simple DNA mole- cules could not serve as highly specific heredity determinants. These erroneous views about the structure of DNA were slowly corrected, mainly through the work of Chargaff and his collaborators (for a review see ref. 4). It was found that the DNA has a highly complex asymmetrical structure; 7 that the individual purines and pyrimidines, as a rule, are not present in equimolar amounts; that the "tetranucleotide" unit simply does not exist; 8 that the composition of the whole DNA is different for each species but similar for different organs of the same species ; 9 and that the DNA of one type of cell is actually a mixture of different molecules.* 10,:l1 All these features arc necessary if the DNA molecules are to serve as heredity determinants ; but the finding of these features does not, of course, furnish a proof that the DNA molecules actually are heredity determinants. The Transforming Phenomenon The study of the nature of the transforming principle furnished the most acceptable proof that one heredity determinant is DNA. The transforming phenomenon will be discussed at greater length because it offers unique possibilities for studying the correlation between the * Assuming the molecular weight of DNA to be of the order of 5 X 106, the number of possible combinations of sequence of different nucleotides for the DNA molecules of just one composition is of the order of 109000. The Chemical Basis of Heredity Determinants 325 structure and the function of heredity determinants. The reasons for this will become clear from the description of the phenomenon and of the agent involved. In 1928, Griffith 2 reported that, from mice that had been injected with living non-encapsulated (R) pneumococci mixed with heat-killed encapsulated (S) ones, living S pneumococci were recovered. Of par- ticular interest was the fact that the specific type of S cells was the same as the heat-killed cells rather than the type from which the living R cells were derived. Once the new feature (production of the capsule of a specific type) was established, it was retained and reproduced in subsequent generations as if a new gene had been added to the genetic make-up of the receptor cells; indeed, such transformed cells, when heat-killed, could in turn induce the transformation of R cells exactly in the same way as did the original S cells. Alloway was the first to demonstrate that the presence of the whole "donor" (S) cells is not necessary for the phenomenon. This investi- gator prepared cell-free aqueous extracts of the heat-killed S cells, passed the extract through a bacterial filter and demonstrated that the filtrate is still capable of transforming the R cells into S cells. In 1944, Avery, MacLeod, and McCarty 3 purified the extract further and found that the responsible agent (the "transforming princi- ple") had all the properties of a highly polymerized deoxypentose nucleic acid. This single finding laid the foundation for the chemical study of the transforming principle. Species Susceptible to Transformation To date, reproducible transformation phenomena are known only in bacteria. A transformationlike phenomenon in viruses has been reported12 and confirmed; however, it is difficult to judge whether the nature of this phenomenon is similar to that of bacterial transforma- tion. The transformation of organisms higher than bacteria has not yet been reported. Of bacteria, the organism used originally by Griffith, the pneumo- coccus, is still most widely experimented upon, because of the large body of knowledge accumulated, the comparatively low pathogenicity of this organism, and, above all, the reproducibility of results. The disadvantage of this species is the presence in the bacterial cultures of an enzyme, deoxyribonuclease (DNAase), which tends to destroy the transforming principle. The transformation in several other bacteria has been reported. These are Escherichia coli, Shigella paradysenteriae, Proteus, Salmo- 326 Essays in Biochemistry nella, Staphylococcus, tubercle bacillus, Alcaligenes radiobacter, Phyto- monas tumefaciens, and Brucella. However, these reports have not as yet been confirmed by others and thus far cannot be used for a routine study of the transforming principle. Besides pneumococcus, the only two species in which the transforma- tion phenomenon can be repeated day after day with the same reliability are Hemophilus influenzae 13 and Neisseria meningitidis.1* Hemophilus influenzae lends itself particularly well to quantitative studies on the transforming principle because of the absence of DNAase from bacterial cultures. Features Transferable in the Transformation Phenomenon The feature transferred (induced) in the transformation phenome- non, as originally discovered by Griffith, was the production of capsules of type I, II, or III of pneumococcus; as is well known, each capsule contains polysaccharide specific for its type. Further studies of McCarty and Avery demonstrated that the production of capsules of types VI or XIV can also be induced by using the transforming prin- ciples from the cells of these types. It is probable that transformation could be demonstrated for every one of the 70-odd known types of pneumococcus. The feature transferred need not be limited to poly- saccharide however. Austrian and MacLeod demonstrated acquisition of a specific M protein through transformation in pneumococci. The six capsular substances a, b, c, d, e, and / produced after the trans- formation of Hemophilus influenzae 13 at first appeared to be polysac- charides, but closer chemical study 15, 16 revealed that most of them form a new class of immunologically active compounds, the polysugar phosphates. The substance of type b is of particular interest: it appears to consist of a polyribophosphate chain, as it exists in pentose- nucleic acids, in which the place of the purines and pyrimidines is occupied by a second similar chain, linked to the first in 1:1' glycosidic linkages. The substance of type a appears to be a polyglucophosphate and that of type c a polygalactophosphate. The transferable feature in Neisseria meningitidis is the production of capsular substances of type I or Ha which appear to be more complex than polysaccharides. Other transferable features include induction of fermentation of salicin, change in metabolism of glucose and lactic acid, production of mannitol phosphate dehydrogenase, and change in quantity of polysaccharide produced; in this last case it was suggested that the genes involved form indeed a series analogous to what is known in higher organisms as an allelic series. Still other transferable features include change The Chemical Basis of Heredity Determinants 327 from sensitivity to resistance to penicillin, streptomycin, and sulfanil- amide, and a change from resistance to sensitivity to streptomycin (Hotchkiss et al.). This class of transformation phenomena is of par- ticular interest because the gradations involved (one-step or multistep acquisition of resistance) resemble closely the gradations acquired by a natural process, i.e., spontaneous mutation. In summary, then, one can say that the phenomenon of transforma- tion involves a great variety of genetic characters, and, were it not for the nature of the phenomenon, it would closely resemble spontane- ous mutation. The Nature of the Transforming Phenomenon Although a considerable amount of work has been done in the field of bacterial transformations, the nature of the phenomenon itself re- mains far from clear. At present, the following conclusions appear logical: The molecules of the transforming principle are heredity determi- nants because their presence determines the presence of hereditary characters. However, it is not known whether all hereditary characters can be accounted for by the (joint) action of all the molecules of the transforming principle in the cell; it is still conceivable that the determination of the most fundamental features proceeds through an entirely different mechanism. The transforming principle seems to consist of DNA molecules. On successful transformation these molecules must reproduce (or be re- produced) because more of them are obtained. The reproduction must take place inside the cell because the DNA is never found outside, and because a thorough destruction of the cell is necessary to isolate the transforming principle. Furthermore, within 3 minutes after the transforming principle is added to the cells, it becomes completely protected against the action of added strong DNAase.13 Thus, it appears that on transformation the molecule of DNA must penetrate the cell. However, the mechanism of this hypothetical penetration remains unknown; the DNA is composed of giant molecules and ought to be stopped by a normal bacterial membrane; at neutral pH the DNA molecules are highly negatively charged and ought to be repelled by the majority of bacterial cells which are also known to be nega- tively charged, especially when coated with polysugar phosphates.15-16 Only a small proportion of receptor cells (about one per hundred in pneumococcus, 10 to 103 times less in H. influenzae) is actually sus- 328 Essays in Biochemistry ceptible to transformation; such cells must therefore be in a special physiological state.17 Upon entering the susceptible cell the DNA may become fixed to the genetical "locus," presumably on bacterial chromosomes. That some sort of "fixation" or precipitation is necessary has been postulated on the demonstration that the molecule of DNA of E. coli in solution is actually longer than the cell itself.18 The fact that the transforming principle as extracted from the cell seems to be in the form of a nucleoprotein 3*19 also indicates some binding of DNA. In such an hypothetical fixation or acceptability inside the cell, two or more transforming principles may compete for the substrate. A phenomenon of competition (a vs. b, or b vs. c) has indeed been demonstrated in H. influenzae ; 17 such competing transforming principles can be demon- strated to obey the law of mass action. The problem of hypothetical competition for the "locus" may be closely connected with the problem of possible DNA exchange. When an R (non-encapsulated) cell is transformed into an S (encapsulated) cell, one may suppose that a missing transforming-principle molecule has been added (to a bare locus?) ; however, one may also postulate that an inactive (or less active) molecule has been replaced by an active one. This latter explanation appears more probable for the following reasons: (1) Many gradations of "roughness" are known, corresponding to various amounts of polysaccharide produced, and in some cases introduction of a transforming principle may abolish the action of an already existing gene (compare "allelism" in higher organisms). (2) S cells of one type can be transformed directly into S cells of another type.20 The Transforming Principle From the foregoing discussion one can easily appreciate the poten- tialities of using the transforming principle for the study of the chem- istry of heredity determinants; since no other form or system of heredity determination offers such possibilities, most of the chemical work was indeed done on this substance. The transforming principle can be extracted from the cell, purified, chemically identified, and analyzed. It can be subjected in vitro to the action of physical and chemical agents stronger than those which can act on the living cell without killing it; after removal of excess reagent, the nature and the extent of changes induced in the DNA molecule can be estimated; and, finally, such changed transforming principle (DNA) can be re- The Chemical Basis of Heredity Determinants 329 introduced into the cell to study the relationship between the change in structure and the change in function. In short, this mode of attack offers possibilities of applying to the problem of heredity the same rational approach which yielded elucidations in so many fields of biochemistry. Although the above reasons for interest in the transforming principle appear to the author to be the most important ones, the biochemist's and biophysicist's interest may often come from the application of the biological activity of DNA to the study of DNA itself. In early chemical approaches the DNA was subjected to degradation, with complete disregard of the macromolecular nature of the substance. This error was compensated for in the last decade by extensive bio- physical studies of the giant molecules of DNA. However, the bio- physicist soon faced a dilemma as to whether his substance wTas indeed "native" or still degraded. The discovery that the DNA loses its transforming activity on the slightest degradation suggested a con- venient yardstick of "intactness"; for, although no one can say whether his DNA preparation is in the same state as the DNA that exists in the cell, the "functionally intact" unit seems important enough to warrant study and constant enough to serve as a standard. In such a study subtle reactions of certain agents such as mutagenic, carcino- genic, or carcinostatic agents with the DNA can often be demonstrated, by studying the loss of transforming activity, long before these reac- tions can be discovered by any physical or chemical method. In the following, the emphasis will be on the transforming principle itself and the transforming phenomena will be mentioned only in their role as detectors of activity. It must be remembered, however, that a transforming phenomenon may involve several steps, and at present it is not possible to decide which is or are responsible for inactivation. The Chemical Nature of the Transforming Principle In 1944 Avery, MacLeod, and McCarty found the purified trans- forming principle to have all the properties of a highly polymerized DNA. Their conclusion that the transforming principle is DNA was based on the following observations: (1) Elementary analysis of the transforming principle corresponded to that of DNA. (2) Chemical and physical tests revealed the presence of DNA as the only detectable substance. (3) Serological tests failed to detect the presence of any immunologically active substances (such as polysaccharides). (4) Of several enzymes tested only deoxyribonuclease was able to destroy the transforming activity. At the same time the physical study (vis- 330 Essays in Biochemistry cosity, ultracentrif ligation) revealed that the substance was in highly polymerized form. The conclusion that the transforming principle is DNA was criticized mainly on the following grounds: (1) The chemical methods were not sensitive enough to exclude the presence of an impurity (protein?) which might be responsible for the activity. (2) Even accepting the evidence that the destruction of DNA by DNAase destroys the activ- ity, it could still be that DNA is not active alone but merely in combination with such a hypothetical protein so that the destruction of either moiety results in inactivation. (3) The non-destruction of activity by a few proteolytic enzymes, by itself, does not prove the non-protein nature of the transforming principle, since proteins are known which resist many proteolytic enzymes. The methods of purification and analysis have undergone consider- able improvement in recent years in partial answer to the first point of criticism. The transforming principle of H. influenzae has been purified to the point where it contains less than 0.4% of protein, immunologically active substance, or ribonucleic acid.19*21 No loss of biological activity occurred during the gradual removal of impurities. The transforming principle of pneumococcus has now been purified to where it contains less than 0.02% of protein.22 The amount of DNA sufficient to transform one cell of H. influenzae is of the order of 10 ~8 /xg. according to one estimate.21 An impurity of 0.01% would correspond to about six molecules of molecular weight 105, or less than one molecule of molecular weight 106. Thus one approaches the situa- tion where the probability of the transforming principle being protein in nature can be excluded on purely analytical grounds. Other studies also offer further indications although not absolute proof that the transforming principle is DNA. Crystalline pancreatic DNAase in concentrations lower than 10~4 ^g./ml. produces 10-fold decrease of activity within 20 minutes.21 Upon heating, the tempera- ture at which the transforming principle begins to lose its activity (81°, 1 hour) is the same as the temperature at which the viscosity of the bulk of the preparation begins to decrease.21 Most known proteins cannot withstand these heating conditions. The pH values (on both acid and alkaline side) at which the activity begins to decrease are again the same as the pH values at which the viscosity begins to decrease. Thus, the active molecule of the transforming principle seems to behave like the average molecule of DNA. In summary, the evidence favors the view that the transforming principle is DNA ; no evidence to the contrary has ever been presented. The Chemical Basis of Heredity Determinants 331 The Heterogeneity A chemist who intends to analyze a purified preparation of the transforming principle will, of course, be concerned with the problem as to whether the preparation represents a single chemical species or a mixture of many. Both the biological and the chemical evidence indicates that the latter may be the case. If one performs a transformation experiment using the DNA from a donor carrying several transformable characters ("markers"), each of the resulting transformed cells carries as a rule * only a single marker. Doubly transformed cells occur as rarely as predicted from the probability of two independent particles hitting the same cell. Thus, the DNA of each cell seems to consist of many different mole- cules, each of which determines a different hereditary character. Chemical evidence has been furnished by the important discovery of DNA fractionation. It has been shown that DNA preparations from calf thymus or from E. coli can be separated into several frac- tions each differing in its proportions of individual purines and py- rimidines. The compositions of whole DNA preparations recorded in the literature are therefore not the compositions of individual molecules but the averages of compositions. The differences in the proportions of purines and pyrimidines (or their nucleotides) may be not the only manifestation of non-identity of individual molecules; other differences may involve a difference in sequence of nucleotides, in length of the molecule which is asymmetri- cal,7 or in some other unknown feature. The Molecule of the Transforming Principle If, as it seems, the molecules carrying different features are different, it becomes important to estimate how many molecules of one kind are necessary to transform one cell. Such an estimate has been made for H. influenzae. This species is more convenient than pneumococcus because the cultures of the latter contain DNAase, which tends to obscure quantitative study. The total amount of DNA necessary to transform one cell of H. influenzae was found to be 10~8 fig., which is five times more than the total amount of DNA per cell (2 X 10 _9 fig.) in this species.21 If each molecule of DNA is different, then the number of molecules of one kind necessary for transformation would be of the order of five. If this number could be further reduced, support would be gained for the hypothesis that practically all mole- * An interesting exception will be discussed on p. 332. 332 Essays in Biochemistry cules of one kind are active. If, in addition, every DNA molecule in the cell is assumed to be a potential transforming principle, then the physical and chemical behavior of the bulk of the DNA preparation are representative of the physical and chemical behavior of the active molecules. Obviously, more evidence is needed before such a view can be fully accepted, but even at the present status the "infectivity" of DNA particles is comparable to that of bacterial viruses. If one assumes an arbitrary molecular weight of DNA, the above estimates can be expressed in terms of numbers of molecules. For a molecular weight of the order of 5 X 106, the number of DNA molecules of all kinds necessary to transform one cell would be of the order of 1000 and the total number of molecules of DNA in one cell of the order of 200. Even if one assumes that all the molecules of DNA in the cell are functional, it still appears that there are too few of them to serve as determinants of all the hereditary characters of the cell. One is tempted to speculate whether the DNA molecules do not determine all the characters or whether one molecule of DNA deter- mines several characters. That the latter may be the case is sug- gested by the discovery of the multiple transformations in H. influ- enzae 23 and in pneumococcus.24 In H . influenzae a new strain ab has been obtained by exposing cells b to the transforming principle from cells a (TPa). The new type produces two capsules (a and b) and yields a new transforming prin- ciple TP0& capable of producing cells ab from any susceptible receptor cells. These results cannot be obtained by simply mixing TPa with TP6 in vitro. In pneumococcus, the exposure of sensitive cells to the transforming principle from cells bearing two genetic markers, mannitol utilization and streptomycin resistance, produces up to 15 times more cells bearing two markers than would be expected from randomly distributed inde- pendent transformations. Again, this result cannot be obtained by simply mixing the two transforming principles in vitro. At present, no evidence exists for the possibility that the two fea- tures reside in separate molecules connected by some link such as protein, since extensive deproteinization and autolytic proteolysis did not abolish the linkage; in addition, no evidence for a double molecular weight of the doubly transforming particles has been obtained. Thus it appears more likely that one molecule of DNA can indeed determine more than one genetical marker. Just how one molecule of DNA can reproduce faultily in the presence of another one (in the same locus?) remains entirely in the domain of speculation. The Chemical Basis of Heredity Determinants 333 The Resistance to Physical and Chemical Agents The literature on the effect of physical and chemical agents on DNA is rather voluminous. In most cases the starting material for these studies was prepared by the methods which are now known to give denatured and therefore less resistant DNA. The following discussion will be limited to studies in which the starting material had full trans- forming activity. 1. Heat and pH. Quantitative study of the effect of these factors has been made for the transforming principle of H. influenzae.21 Both the viscosity and the activity of the purified preparation remain un- changed after 1 hour of heating at temperatures up to 81° (in citrate buffer) or after incubation at 23° between pH 5 and 10. The remark- able stability to heat, which is higher than for most known proteins, is similar to the stability of human DNA and calf-thymus DNA when they are prepared by a similar method ; 25 this stability is much higher than the values reported in the literature for DNA prepared by the previous, somewhat injurious methods. However, it should be pointed out that the inactivation at the low range of pH could be due to a removal of purines and/or hypothetical hydrogen bonds to these purines. A study 21 of the amount of such "depurination" of the trans- forming principle reveals that a 100-fold inactivation occurs with the removal of less than two purines per thousand; thus, practically every purine may be necessary for the activity.* 2. Deoxyribonuclease . Mammalian and bacterial DNAases in mi- nute amounts destroy the activity of pneumococcal transforming prin- ciple.3 A quantitative study on the transforming principle of H. in- fluenzae 21 reveals that crystalline pancreatic DNAase in concentra- tions of less than 10-4 /xg./ml. causes a 10-fold decrease of activity within 28 minutes and complete inactivation within 140 minutes. On the other hand, the drop of viscosity at the beginning of inactivation is insignificant. The reason for this discrepancy is still not clear, but the initial change might involve the breaking of a few phosphate bonds, sufficient to destroy the activity but insufficient to cause any decrease in the size of the molecule still held together by hydrogen bonds. 3. Ionic strength. A quantitative study of the effects of exposure to various ionic strengths on activity and viscosity has been made for the transforming principle of H. influenzae.21 Previous exposure to lower or higher ionic strengths did not affect the viscosity (as measured *This may be true only on the assumption that the active molecule behaves like an average molecule of DNA. 334 Essays in Biochemistry in a standard buffer) ; on the other hand, the activities were irreversibly reduced by exposure to lower (but not to higher) ionic strength. This permanent damage could be due to the breakage of a few vital bonds, such as hydrogen bonds, during stretching caused by repulsion of anions in the DNA molecule in the absence of salts. 4. Deamination. The chemicals whose action on the transforming principle was studied 21 were chosen for two reasons: (1) because the nature and the extent of reaction with DNA could be determined; (2) because the agents themselves have important biological activity, either mutagenic, carcinogenic, or carcinostatic. Nitrous acid belongs to the first group of reagents. Incubation of the transforming principle with 2 M NaN02 at pH 5.3 (veiy mild deaminating conditions) results in a very rapid inactivation. However, the viscosity remains constant indicating that the average DNA mole- cule is but slightly altered. The extent of deamination corresponding to a 1000-fold decrease of activity is found to be of the order of 0.1%. Thus it seems that practically all the primary amino groups are essen- tial for activity. 5. Mutagenic agents. The agents in this very heterogeneous class are grouped together, although the mechanisms of their action may be entirely different; they all seem to prime a phenomenon of great importance and of unknown nature, namely, mutation. It has been suggested 26 that the processes leading to mutation and to death are essentially the same, with the exception that the latter is accompanied by more extensive molecular changes. If this is indeed so, then the inactivation of the transforming principle by these agents could be a demonstration of a "too strong mutation." 6. Ultraviolet irradiation. The ultraviolet irradiation of the trans- forming principle (transformation to type specificity) has been studied in H. influenzae.27 The inactivating dose was found to be of the same order of magnitude as the one necessary to inactivate bacterial viruses. The action of ultraviolet irradiation on DNA is now believed to be due to free radicals H- and OH-, and to peroxides formed during irradiation of organic molecules. McCarty 28 observed reversible in- activation of pneumococcal transforming principle by ascorbic acid and several other self-oxidizing agents. 7. Ferrous ion. Ferrous ion self-oxidizes even in the absence of H202. In concentrations as low as 10-5 M, Fe++ alone causes a 10-fold inactivation of the transforming principle of H. influenzae ; 21 again, no change in viscosity is observed. The exact nature of the damage is still The Chemical Basis of Heredity Determinants 335 unknown ; an oxidative deamination of a few nitrogenous bases and/or breakage of a few vital labile bonds might be postulated. 8. Mustards. Herriott 29 studied inactivation of pneumococcal trans- forming principle by di-(2-chloroethyl) sulfide (mustard gas). The inactivation of the transforming principle (type specificity) of H. in- fluenzae by various nitrogen mustards was studied by Zamenhof et al.27 In both experiments the inactivating concentrations were as low as 10 ~5 M. For various nitrogen mustards the order of inactivating power for the transforming principle seems to be the same as the order of carcinostatic power of these compounds, thus suggesting a correla- tion between these phenomena.27 Induction of Changes in the Active DNA Molecule The removal of less than 0.2% of the amino groups of the DNA molecule coincides with a total inactivation of the transforming prin- ciple. The treatment applied (deamination or depurination) is not known to act specifically on jew amino groups or purines of special importance; thus it appears that practically all the amino groups are necessary for the activity of the transforming principle or, perhaps, DNA in general. This view gained support when Watson and Crick 30 proposed their model of the DNA molecule. In this model all the amino groups are indeed essential for maintaining the integrity of the molecule through the hydrogen bonds. One might suspect that any change in the DNA molecule, whether it affects the pattern of electrical changes (in depurination and deami- nation) or not, results in a total inactivation or at least in a mutation. The study of the effect of various agents on the transforming principle in vitro thus far has not solved this problem because the extent of changes which can be induced in vitro before total inactivation occurs is rather small and because the nature of the changes is not always known. It has been shown in special cases that a drastic change in the chemical composition of the DNA molecule can be induced in vivo. When the cells of E. coli were grown on a medium containing 5-bromo- uracil, the highly polymerized DNA from these cells had up to half of the thymine molecules replaced by 5-bromouracil.31*32'33 5-iodoura- cil can also replace thymine in the DNA, but to a smaller extent, undoubtedly because it is sterically less similar to the methyl group in thymine. Neither of these groups bears any charge; neither partici- pates in the H-bond structure of the Watson and Crick model.30 This substitution did not result in any demonstrable changes in the pheno- 336 Essays in Biochemistry type or the genotype of the cells ; 32 for on a medium free of 5-bromo- uracil or 5-iodouracil, the substitution is reversed. Thus it appears that certain drastic changes in the DNA molecule may be without consequences, but, on the other hand, the maintenance of the original pattern of electrical changes or H-bonds may be essential for un- changed activity. As a working hypothesis one could even postulate that a mutation is a heritable change in such a pattern of electrical charges.34 DNA with bases deaminated or partially absent has never been found in nature. The change of pattern must therefore occur through some other mechanism such as change of sequence or proportion of nucleotides or change of length of the DNA molecule. At present any such process can be visualized only as a rare fault occurring during DNA reproduction. Such a process, then, might be the chemical basis of mutation. Unstable DNA The premise that mutation occurs only during DNA reproduction might appear inconsistent with the phenomenon of delayed mutations since in certain cases mutations have been found to occur long after the originally working mutagenic agent had been removed. It has been suggested 35 that in such cases, or perhaps in every case, the role of the mutagenic agent is to bring the gene into an unstable state from which it can either return to the previous stable state or change into a stable mutant gene. This secondary change might no longer require the presence of the mutagenic agent. It is of interest to note that the process of unstabilization of heredity determinants can also be demonstrated on the transforming principle in vitro. When the transforming principle of H. influenzae is subjected in vitro to the sublethal action of heat, H+ ion, deoxyribonuclease, ultraviolet irradiation, or nitrogen mustard, the surviving (active) molecules become very unstable to heat or even to storage in the cold, under conditions entirely harmless for intact molecules.36 Study of the kinetics of inactivation of the transforming principle by heat indi- cated that at least two reactions are involved: the unstabilization and the actual inactivation. When the transforming principle, made very unstable by heat treat- ment or by mustard treatment, was used for the transformation experi- ment, it was found to reproduce as completely stable. Thus, the injury to DNA (unstabilization) induced in vitro was not retained on repro- duction; the change was therefore not a "mutation in vitro." However, The Chemical Basis of Heredity Determinants 337 such unstabilization processes may be actually as important as muta- tions because the injury may determine the part of the molecule where the fault in the reproduction (i.e., mutation) is most likely to occur.34 The work described herein is part of a research project supported by grants from the National Institutes of Health, Public Health Serv- ice, by a grant-in-aid from the American Cancer Society upon recom- mendation of the Committee on Growth of the National Research Council, and by an institutional grant from the American Cancer Society to Columbia University. References 1. A. D. Hershey and M. Chase, J. Gen. Physiol, 36, 39 (1952). 2. F. Griffith, J. Hyg., 27, 113 (1928). 3. O. T. Avery, C. M. MacLeod, and M. McCarty, J. Exptl. Med., 79, 137 (1944). 4. S. Zamenhof, in Phosphorus Metabolism, 2, 301, Baltimore (1952). 5. A. Boivin, R. Vendrely, and C. Vendrely, C. R. Acad. Sci., 226, 1061 (1948). 6. P. A. Levene and L. W. Bass, Nucleic Acids, The Chemical Catalog Co., New York, 1931. 7. S. Zamenhof and E. Chargaff, J. Biol. Chem., 178, 531 (1949) ; 187, 1 (1950). 8. E. Vischer and E. Chargaff, J. Biol. Chem., 176, 715 (1948). 9. E. Chargaff, S. Zamenhof, and C. Green, Nature, 165, 756 (1950). 10. E. Chargaff, C. F. Crampton, and R. Lipshitz, Nature, 172, 289 (1953). 11. G. L. Brown and M. Watson, Nature, 172, 339 (1953). 12. G. P. Berry and H. M. Dedrick, J. Bacteriol, 81, 50 (1936). 13. H. E. Alexander and G. Leidy, Proc. Soc. Exptl. Biol. Med., 78, 485 (1950) ; J. Exptl. Med., 98, 345 (1951). 14. H. E. Alexander and W. Redman, J. Exptl. Med., 97, 797 (1953). 15. S. Zamenhof, G. Leidy, P. L. FitzGerald, H. E. Alexander, and E. Chargaff, Federation Proc, 11, 315 (1952); J. Biol. Chem., 203, 695 (1953). 16. S. Zamenhof and G. Leidy, Federation Proc, 13, 327 (1954). 17. H. E. Alexander, G. Leidy, and E. Hahn, J. Exptl. Med., 99, 505 (1954). 18. J. W. Rowen and A. Norman, Arch. Biochem. and Biophys., 51, 524 (1954). 19. S. Zamenhof, G. Leidy, H. E. Alexander, P. L. FitzGerald, and E. Chargaff, Arch. Biochem. and Biophys., 40, 50 (1952). 20. H. E. Alexander and G. Leidy, Proc Soc. Exptl. Biol. Med., 78, 625 (1951). 21. S. Zamenhof, H. E. Alexander, and G. Leidy, J. Exptl. Med., 98, 373 (1953). 22. R. D. Hotchkiss, Harvey Lectures, 49, 124 (1953-1954). 23. G. Leidy, E. Hahn, and H. E. Alexander, J. Exptl. Med., 97, 467 (1953). 24. R. D. Hotchkiss and J. Marmur, Proc Natl. Acad. Sci., Jfl, 55 (1954). 25. S. Zamenhof, G. Griboff, and S. Marullo, Biochim. et Biophys. Acta, 18, 459 (1954). 26. W. D. McElroy, Science, 115, 623 (1952). 27. S. Zamenhof, G. Leidy, and B. Reiner, Proc. Am. Assoc. Cancer Research, 1, 53 (1954). 338 Essays in Biochemistry 28. M. McCarty, J. Exptl. Med., 81, 501 (1945). 29. R. M. Herriott, J. Gen. Physiol, 82, 221 (1948). 30. J. D. Watson and F. H. C. Crick, Nature, 171, 737, 964 (1953). 31. D. B. Dunn and J. D. Smith, Nature, 174, 305 (1954). 32. S. Zamenhof and G. Griboff, Nature, 174, 306, 307 (1954). 33. S. Zamenhof, Abstracts Am. Chem. Soc, 126th meeting, New York, 42c (1954). 34. S. Zamenhof, Science, 120, 791 (1954). 35. C. Auerbach, J. M. Robson, and J. G. Carr, Science, 105, 243 (1947). 36. S. Zamenhof, G. Leidy, and E. Hahn, Records Genetics Soc. Am., 23, 75 (1954). Index Acclimatization to high altitudes, 122 Acetate metabolism, 133, 138 in porphyrin biosynthesis, 244 in rubber biosynthesis, 23 in steroid biosynthesis, 85 in terpine biosynthesis, 23 Acetoacetic acid, in amino acid bio- synthesis, 32 in cholesterol biosynthesis, 23 in rubber biosynthesis, 23 Acetoacetyl coenzyme A, 134 metabolism in liver, 138 Acetobacter suboxidans, 183, 187, 193 Acetohydroxamic acid, 134 Acetone in rubber biosynthesis, 23 Acetyl coenzyme A, 133, 135, 136, 109 Acetylenes, 4 Acetylsulfanilamide in synthesis of peptide bonds, 109 Active succinate, 249, 254, 256 Acyl enzyme, 117 Adenosine triphosphate, 133, 137, 219 in synthesis of glutamine, 110 in synthesis of 7-glutamylcystine, 110 in synthesis of peptides, 106, 108, 110, 114 Adenylic acid, 225 Adenylosuccinic acid, 225 Adipose tissue as energy store, 305 Adrenal cortex, 86 Aerobacter aerogenes, 191, 193 Agrocybin, 4, 8 L-Alanyl-L-phenylalaninamide, 115, 116 Aldolase, 112 Allocholesterol, 310 Alloxan, 137 Amino acid, activation of, 111 clustering of, in peptides, 288 free, in cells, 113 in T3, 96 incorporation of, 111 sequence in peptides, 270 a-Amino-/3-ketoadipic acid, 249, 253 5-Aminolevulinic acid, 250 Aminophorase-glutamic dehydrogenase, 176 Ammonia, detoxication of, 226 /3-Amylase, attack on glycogen, 293, 304 amylo-(l,4 -> l,6)-transglucosidase, 303 Amylo-l,6-glucosidase, 293 Amylopectin, 293 Androgens, 90 Antigen, glutamylpeptide as, 63 Antitumor agents, 82, 129 Apoferritin, 200 amino acid composition of, 203 Arginine, 219, 223 metabolism of, 216 Argininosuccinic acid, 220 Aromatic compounds, biosynthesis of, 259 Ascorbic acid as inducing agent, 101 Asparagine, 228 Aspartic acid, 225, 230 Asymmetric synthesis, 156 Axis of symmetry, 161 8-Azaguanine, 128 339 340 Index Bacillus megatherium, loss of lysoge- nicity in, 102 Lwoff effect on, 36 Bacillus subtilis, culture of, 64 Bacterial membranes, 97 N« labeling of, 97 Bacteriophage, 78, 94 attachment to host, 97 burst size of, 103 classification of, 95 DNA in, 95, 98 effect of temperature on, 102 genetic characteristics of, 99 ghosts, 96 5-hydroxymethylcytosine in, 96 labeled with N" and P32, 99 protein antigen, 98 protein synthesis in, 103 RNA in, 100 temperate virus X, 101 vegetative state of, 103 Basidiomycetes, 1 Benzene-l,2,3,4-tetracarboxylic acid, 314 Benzimidazoles as inhibitors, 129 1,2-Benzofluorine, 309 Benzotriazoles as inhibitors, 129 Benzoyl CoA in hippuric acid syn- thesis, 109 Benzoyl L-tyrosylglycinanilide, 108 Biformyne (1 and 2). 4, 8 Bittner factor, 123 Bohlmann's acid, 6 Bovine albumin as plasma expander. 58 Branched chains, 22 CO2 fixation in, 25 Branching enzyme, 303 5-Bromouracil, 80 in E. coli medium, 335 Burst size in virus-infected cell, 103 Butyryl coenzyme A, 135 Cancer. 83, 119 cause of, 122, 125 definition of, 119 steroid hormones in, 86 Carbamyl phosphate, 219, 223 Carbobenzoxyglycylpapain, 117 Carbon dioxide, fixation of, 25, 28 in cholesterol synthesis, 28 Carboxypeptidase as analytical tool, 275 Carcinogens, 120 as inhibitors, 127 as mutagens, 123 Carcinoma, mammary, in the mouse, 123 virus of, 124 Carlina oxide, 5 Carotene, 23 Catalysis, 234 Cathepsin C, 115 Cevacine, 318 Cevadine, 317 Cevagenine, 318 Cevanthridine, 314 Cevine, 308, 311 Charge density, in tetrazoles, 147 of peptide bonds, 237, 289 Chlorophyll, biosynthesis of. 241 Cholesterol, biosynthesis of, 23, 26, 136, 139 C02 in, 28 nucleus in steroids, 320 Chromatography, of deoxyribonucleic acid, 20 of proteins, 273 of ribonuclease hydrolyzates. 282 Chromosomes, solubility of. 14 Chymotrypsin, 114 effect on ribonuclease, 286 Citrate, formation, 136 in biosynthesis of porphyrins, 247 Citrulline, 218 Claisen condensation in isoprene, 31 Coenzyme A, 109, 133 Colicines, 105 Coliphage, see Bacteriophage inaptitude in, 37 Collagen disease, steroid hormones in, 86 Competitive inhibition, 128 Condensing enzyme, 134 Conjugation, of glutamyl peptides, 66 of proteins, 73 Cross linkage in peptides, 270 Crotonase, 28, 33 Index 341 Cyclitols, 182 anaerobic degradation of, 195 1 ,2-Cyclopentenonaphthalene, 309 Deamination, oxidative, 175 Decevinic acid, 317 Degradation, of glycogen, 300 of peptides, 279 of shikimic acid, 260 of steroid hormones, 88 5-Dehydroshikimate, 265 L-2-Deoxy-mttco-inositol, 183, 185, 193 Deoxyribonuclease, 100, 333 effect on transforming principle, 327, 330 Deoxyribonucleic acid (DNA), compo- sition of, 16 5-hydroxymethylcytosine, 96 in embryonic development, 120 non-antigenicity of, 98 of Streptococcus faecalis, 15 of virulent phage vs. temperate phage, 102 of viruses, 78, 96 synthesis in infected bacteria, 103 transforming activity of, 19, 323 Watson and Crick model, 335 Deuterium, as glycogen label, 292 Deuteroacetic acid in porphyrin syn- thesis, 244 Dextran as plasma substitute, 58 Diabetes. 138 Diacetyl-7-ketojervine, 312 Diastereomers, 160 Diatretyne amide, 4, 8 nitrile, 5, 8 Diel's hydrocarbon, 310 D-2,3-Diketo-4-deoxy-e7n"-inositol, 185 bisphenylhydrazone, 193 L-l,2-Diketo-myo-inositol, 183. 191 2.5-Dimethoxyquinone, 3 jS-Dimethylacrylic acid (DMA), 23, 25 Disulfide bridges, 277 Drosophilin (C and D). 4, 7 Ecteola-Cellulose, 19. 20. 273 Electron excitation. 235 Electron transfer, 238 Electron microscopy of viruses. 97. 100, 124. 126 Embryonic development, 120 End-group analysis, 275, 283 Energy, 106 levels in enzymes, 236 of activation, 234 of growth, 121 storage, in glycogen, 306 Enzyme-substrate interaction. 232 Enzymes, as analytical tools, 280, 286, 287 condensing, 134 in asymmetric synthesis, 157 induced in bacteria, 112 Epoxides as inducing agents. 101 Ergosterol, 33 Escherichia coli, bacteriophage, in, 96, 98 /3-galactosidase, in, 112 strain Ki2, 102, 37 strain 15T-, 78 thymineless mutant, 78 Estrogens, 89 Ethylene imine as inducing agent, 101 Fatty acids, 133 activation of, 134 biosynthesis of. 136. 137. 138 Ferritin, 198 function of, 212 structure of, 200 transport of, 212 Ferrous ion effect on transforming principle, 334 Ficin, 114 Flavoprotein, 229, 176, 137 Fluorene, 309 Fomecin (A and B), 2 Formylkynenurine, 278 Foster-nursing, 123 Free-energy change. 233. 106 Fructose-6-phosphate, 262 Fumarie acid. 228 Fumigatin, 3 ^-Galactosidasc, induced formation in E. coli, 112 Gelatin, as plasma substitute, 58 Geraniol, 23 Genuine. 308, 318 "Ghosts" of coliphage, 96 342 Index Glucose, anaerobic degradation, 195 in glycogen synthesis, 296, 116 in nucleic acid synthesis, 79 starvation in bacteria, 40 Glucosidic linkage, 293, 116 Glutamic acid, 227 Glutamine, 176, 228. 118 Glutamyl enzyme, 110, 117 Glutamyl polypeptides, 56 as antigen, 63 metabolism of, 69 Glutathione, 110, 117 as inducing agent, 101 Glyceraldehyde-3-phosphate, 112 L-a-Glycerophosphate, 135 Glycinamide in oxytocin and vasopres- sin, 110 Glycine in porphyrin synthesis, 241 Glycogen, 291 degradation of, 300 homogeneity of, 295, 300 synthesis from glucose. 111, 296 Glycogenesis vs. lipogenesis, 305 Glycolysis, 121 Gradient elution, of deoxyribonucleic acid, 18, 20 of proteins, 273 Growth, in cancer and frog embryo, 120 unbalanced, 81 Guanidoacetic acid, 225 Guanine in Tetrahymena gelii, 127 AH, of peptide bonds, 107 Hecogenin, 320 Hematinic acid, 252 Hemin, 244, 251 Heredity, 322 DNA as determinant of. 324 in carcinogenesis, 122 Hexahydroxycyclohexanes, 191 Hippuric acid, 133 Histidase, induction of, by myo-mos\- tol, 196 Histone in chromatography. 18 Homogeneity, of DNA, 14, 322 of glycogen, 291 of proteins, 272 Homospirostene, 320 Hormones, steroid, 85 Hydrolysis, of glycogen, 293 of protein, 275 /3-Hydroxyisovaleric acid (HIV), 25, 26, 29 Hydroxylamine, 134 reaction with glutamine, 117 5-Hydroxymethylcytosine, 78, 96. 98, 100 /3-Hydroxy-/3-methylglutaconic acid (HMG), 24, 25, 29, 139 Hydroxymethylglutarate (HMG), 25 Hypoxia, effect on tumors, 122 Illudin (M and S), 3 Inaptitude in bacteriophage, 37, 102 Inducing agents, 36, 101 Inosinic acid, 225 epi-Inositol, 183, 188 myo-Inositol, 182 neo-Inositol, 190 Inositol phenylhydrazones, 183 Inositols, 181 dehydrogenase of, 192 stereochemistry of, 190 Insulin, 113 disulfide bridges in, 277 Ion exchange, 18, 273, 274. 281 Ions in enzyme activation, 238 Iron-porphyrin complex, 235 Iron transport, 210, 211 Isocitrate, as shikimic acid precursor, 266 Isogermine, 318 Isoleucine, 22 Isomers, 156 Isoprene, 23 Isorubijervine, 308 Isovaleric acid (IV), 25 Jervine, 308 a-indanone system in, 312 oxide bridge in, 314 /3-Ketoacyl coenzyme A, 134, 135 Ketoglutaraldehyde, 250 a-Ketoglutaric acid, 227, 228, 247, 250 as transaminase inhibitor, 177 11-Ketohecogenin, 320 2-Keto-epwnositol, 183, 188 L-1-Keto-myo-inositol, 184, 191 Index 343 2-Keto-myo-inositol, 182, 190 Ketoinositol dehydrase, 192 Knorr condensation in porphyrin syn- thesis, 251 Krebs-Henseleit cycle inhibition, 177 Lactic acid as glycogen precursor, 292 Leucine, 22, 26, 32 Leucovorin, 52, 53 L-Leucylglycine, 114 Leukemia in mice, 125 Limit dextrin, 293 Lipides, 133 Lipogenesis vs. glycogenesis, 305 Lipoproteins, 75 Lwoff effect, 36 L3'sogeny, 35 Malic acid, 229 Marasmic acid, 3 Matricaria ester, 5 Membranes, bacterial, 97 Metals, role in catalysis, 234 p-Methoxytetrachlorophenol, 4 5-Methoxy-p-toluquinone, 2 l-Methyl-l,2-cyclopentanophenan- threne, 310 5-Methylcytosine, 18 3-Methyl-6-ethylpiperidine, 309 /3-Methylglutaconic acid (MGA), 29, 31 3-Methyl-5-hydroxy-6-ethylpyridine, 309 3-Methyl-5-hydroxypyridine, 309 /3-methyl vinylacetic acid (MVA), 29 Milk factor, 123 Mitochondria, 14, 133, 137 Nemotin, 4, 7 Nemotinic acid, 4 Neurospora crassa, inositol-requiring mutant of, 192 Nitrogen mustards, as inducing agents, 101 effect on transforming principle, 335 Nitrogen sparing, 175 Nitrogen transfer, 216 Nuclear synthesis, in bacteria, 79 in frog embryos, 120 Nucleic acid, 47, 323 fragments, 40 in cancer, 127 Nucleoproteins, 17, 74 Ochracic acid, 4, 9 Ogston hypothesis, 157, 232 Optical activity, 156 in electron excitation, 235 Ornithine cycle, 216 Orotic acid, 224 Osmotic efficiency, 61 Ovalbumin, 113 Oxalacetic acid, 229 in porphyrin synthesis, 297 Oxidation, by enzymes, 239 deamination, 227 deficiency in cancer, 122 Pantothenic acid, 109, 110 Papain, 114 Paramagnetism of enzyme-substrate complex, 239 Pentahydroxycyclohexanes, 191 Peptide bonds, 107 charge density of, 237 Peptides, free, 113 rearrangement of, 278 stepwise degradation of, 279 Perhydrofiuorene system, 319 Periodic acid oxidation, of inositols, 183 in shikimic acid degradation, 260 Peroxides, as inducing agents, 53, 101 Phenaceturic acid, 133 Phenanthrene, 309 Phosphate-bond energy, 218 Phosphorylase attack on glycogen, 293, 304 Pinosylvin, 11 Plasma substitutes, 56 Pleuromutilin, 4, 9 Polyacetylenes, 4 Polysaccharides as plasma substitutes, 57 Polysugar phosphates, in transforma- tion phenomenon, 320 Polyvinylpyrrolidone as plasma substi- tute, 59 Porphyrins, 241 344 Index Progesterone, 90 Prophage, 36, 101 Propiolic acid, 5, 12 Prosthetic groups, 73, 235 Proteinases, 114 Proteins, 72, 270 as antigens in phage, 98 as blood substitutes, 57 Protoporphyrins, 241 Protoverine, 308, 318 Pseudogermine, 318 Pseudosantonin, 3 Pyrimidine nitrogen, 229 Pyrophosphate, AH of, 110, 111 Quinones, 2 D-Quercitol, see Deoxy-?/iuco-inositol Radiation, 36, 122 as inducer, 101 effect on transforming principle, 334 of leucovorin, 52 "Reactive" amino acids, 109, 223 Rearrangement, mycomycin-isomyco- mycin isomerization, 6 of peptides, 278 of veratrum alkamines, 311 Reformatskii reaction, 24 Ribonuclease, amino acids in, 113, 276 hydrolysis of by trypsin, 281 ionic behavior, 271 Ribonucleic acid, in bacteria, 100, 103 in viruses, 94 Rockogenin, 320 Rous sarcoma, 124 Rubber biosynthesis, 23 Rubijervine, 308, 310, 311 Saccharic acids, 182 Safranine method for peptide determi- nations, 62 Salmonella ty-phimurium, transduction in, 105 Santonin, 3 Sapogenins, 319 Sarcoma, Rous, 124 Scyllitol, 182, 183, 191 Sedoheptuloso-7-phosphatp. 263 Selenium rlehydrogenation, 309, 311 Sesquiterpenes, 3 Slukimic acid, 259 degradation scheme, 260 Solanidine, 309, 310, 311 Sjiinulosin, 3 Squalene, 22, 23, 31, 32, 139 Stereoisomerism, 156 in terpenes and steroids, 31, 32 Steric selectivity, 156 Steroid hormones, 86 Steroids, 85, 320 biosynthesis of, 23, 24, 25 Structure of proteins, 270 Substrate bonding, 129, 232 see also Ogston hypothesis Succinate, "active," 249 in porphyrin synthesis, 241 Succinate-glycine cycle, 250 Sulfhydryl groups, in ferritin, 203, 204 in peptides, 276 Sulfonamide effect on bacterial growth, 81 Symmetry, 101, 169 S3^nchrony in bacterial multiplication, 81 D-Talomucic acid, 183 Tartaric acid in enzymatic process, 167 Temperate virus, 101 Temperature effect on temperate pro- phage, 102 "Template" hypothesis, 111, 75 Terpenes, 22, 23 Tetrapyrrole, 255 Tetrazoles, 141 Tetrose phosphate, 263 Thiol esters, 109 Thiolytic cleavage, 135 Thiomalic acid as inducing agent, 101 Three-point attachment, see Ogston hypothesis Thymidylic acid in E. coli, 82 Thymine metabolism, 77 Tissue culture, oxygen requirement for, 121 Transaldolase, 262 Transamidation, 115 Transduction in Salmonella typhimu- rium, 105 Transformation, multiple, 332 principle, 325 Index 345 Tricarboxylic acid cycle, 136 in porphyrin synthesis, 246 Urea, source of, 175 ureidosuccinic acid, 224 urinary, 216 Valine, 22, 32 Veracevine, 317, 318 Veranthridine, 314 Veratramine, 308, 309 Veratridine, 318 Veratrum alkamines, 308 oxidation of, 310 Viruses, see Bacteriophage as carcinogens, 123, 124, 126 Wolff-Kishner reduction, 310 X rays, see Radiation Zygadenine, 308 "- !