Progress in Nucleic Acid Research and Molecular Biology, Volume 17

PROGRESS IN Nucleic Acid Research and Molecular Biology Volume 77 This Page Intentionally Left Blank PROGRESS IN ...

Author: Waldo E Cohn

6 downloads 1205 Views 16MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

PROGRESS IN

Nucleic Acid Research and Molecular Biology Volume 77

This Page Intentionally Left Blank

PROGRESS IN

NucIeic Acid Research and Molecular Biology edited by

WALDO E. COHN Biology Division Oak Ridge National Laboratoy Oak Ridge, Tennessee

Volume

77

7976

ACADEMIC PRESS New York Sun Francisco London A Subsidiary of Harcourt Brace Jovanouich, Publishers

COPYRIGHT Q 1976, BY ACADEMIC PRESS,INC. ALL RIGHTS RESERVED. NO PART OF THIS PUBLICATION MAY BE REPRODUCED OR TRANSMITTED IN A N Y FORM OR BY ANY MEANS, ELECTRONIC OR MECHANICAL, INCLUDING PHOTOCOPY, RECORDING, OR ANY INFORMATION STORAGE AND RETRIEVAL SYSTEM, WITHOUT PERMISSION IN WRITING FROM THE PUBLISHER.

ACADEMIC PRESS, INC.

111 Fifth Avenue, N e w York. New York 10003

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

London NWI

LIBRARY OF CONGRESS CATALOG CARDNUMBER: 63-15847 ISBN 0-12-540017-9 PRINTED IN THE UNITED STATES OF AMERICA

Contents LISTOF CONTRIBUTORS

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

ix

. . . . . . . . . .

xi

DEDICATION-ERWIN CHARGAFF

. . . . . . . . . . FUTURE VOLUMES. . . . . . .

ABBREVIATIONS AND SYMBOLS.

xv

SOMEARTICLES PLANNED FOR

xix

The Enzymic Mechanism of Guanosine 5',3'-Polyp hosphate Synthesis FRITZLIPMANNAND JOSE SY I. Introduction . . . . . . . . 11. The in Vitro Reaction . . . . . . 111. Nonribosomal Synthesis of ppGpp and pppGpp IV. Conclusion References . . . . . . . .

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

. . . . .

. . . . .

. . . . .

. . . . .

1 2 9

14 14

Effects of Polyamines on the Structure and Reactivity of tRNA TEDT. SAKAIAND SEYMOURS . COHEN I. Introduction . . . . . . . 11. Roles of Polyamines in Biological Systems . 111. Chemical and Physicochemical Studies. . IV. Conclusions . . . . . . . References . . . . . . .

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

.

.

.

.

.

.

15 16 21 38 39

Information Transfer and Sperm Uptake by Mammalian Somatic Cells AARONBENDICH,ELLENBORENFFIEUND, STEVENS. W m , DELIABEJU, AND PAULJ. HIGCINS I. Introduction . . . . . . . . . . . . . 11. General Method for the Study of Interaction of Spermatozoa and Mam. . . . . . . . . . malian Cells in Culture. 111. RNase-Sensitive Endogenous DNA Polymerase Activity in Human Sperm . . . . . . . . . IV. Conclusions and Speculations. References . . . . . . . . . . . . . Note Added in Proof . * * * * *

-

-

V

-

43 45 65 71 72 75

vi

CONTENTS

Studies on the Ribosome and Its Components F’NINA SPITNIK-ELSON AND DAVID ELSON I. Introduction . . . . . . . . . . I1. Ribosomal Proteins . . . . . . . . . I11. Macromolecular Interactions in the Ribosome . . . IV. Ribosomal Conformation: Large Conformational Changes . V. Ribosomal Conformation: Restricted Changes . . . VI. Cation Specificity in the Ribosome . . . . . VII. Conclusion . . . . . . . . . . References

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

. . . . . . . .

. . . . . . . .

77 78 80 85 91

93 96 97

Classical and Postclassical Modes of Regulation of the Synthesis of Degradative Bacterial Enzymes BORISMACASANIK I. Introduction . . . . I1 Classical Mode . . . . 111. Postclassical Mode: Energy . IV. Postclassical Mode: Nitrogen V. Conclusion . . . . References . . . .

. . . . . . .

.

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

99 100 103 105 113 114

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

118 119 123 127 128 130 133 140 143 146 147

.

.

.

.

.

149 150

Characteristics and Significance of the Polyadenylate Sequence in Mammalian Messenger RNA GEORGEBRAWERMAN I. Introduction . . . . . . . . I1. Discovery . . . . . . . . . . . I11. Isolation of Poly( A )-Containing RNA . IV. Size of Poly(A) and Location in Messenger RNA V. Decrease in Poly(A) Size . . . . . VI. Steady-State Poly( A ) . . . . . . VII . Poly(A) Elongation . . . . . . . VIII . Poly(A) and Structure of Messenger RNA . . IX. Significance of Poly(A) . . . . . . X Conclusion . . . . . . . . References . . . . . . . .

.

Polyadenylate Polymerases MARY EDMONDS AND MARY ANN WINTERS I . Introduction I1 Historical

.

.

.

.

.

.

.

.

.

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

vii

CONTENTS

. . . . . . . .

I11. The Poly(A) Polymerase Reaction IV . Poly(A) Polymerase Proteins V. Multiple Poly(A) Polymerases VI. Cellular Location of Poly(A) Polymerases VII . Poly(A) Polymerases as a Subunit of RNA Polymerase VIII . Viral Poly( A ) Polymerases . . . . . . I X . Regulation of Poly(A) Polymerases . . . . X. Poly( A) Polymerases as Polyadenylylating Reagents XI . Summary References . . . . . . . . . .

. . . . . .

157 165 166 168 169 171 173 174 175 176

I. Introduction . . . . . . . . . . . . . I1. Life Cycle of tRNA . . . . . . . . . . . . I11. Secondary Structure of tRNA . . . . . . . . . . IV . Tertiary Structure of tRNA . . . . . . . . . . V. Correlation between the 3-D Structure and Physical and Chemical Data VI . Functional Implications of the 3-D Structure of tRNA VII . Concluding Remarks References . . . . . . . . . . . . .

182 182 184 187 199 207 212 213

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

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

. . . . . .

. . . . . .

Three-Dimensional Structure of Transfer RNA SUNG-HOUKIM

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

Insights into Protein Biosynthesis and Ribosome Function through Inhibitors SIDNEYPESTKA I. Introduction . . . . . . . . . . . . . I1. Summary of Protein Biosynthesis . . . . . . . . . I11. Translocation Inhibitors: Localization of Action through Nonenzymic . . . . . . . . and Enzymic Translocation . IV. Ribosomal States . . . . . . . . . . . . V . Erythromycin Binding to Ribosomes . . . . . . . . VI . Effect of Chloramphenicol on the Puromycin Reaction: Models of Ribo. . . . . . . . . . . . some Function . VII . The Sistrand: the Translational Unit . . . . . . . . References . . . . . . . . . . . . .

217 217 223 223 230 234 238 244

Interaction with Nucleic Acids of Carcinogenic and Mutagenic N-Nitroso Compounds W. LIJINSKY I. Introduction . . . . . . I1. Biological Activity of Nitroso Compounds

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

247 249

viii

CONTENTS

I11. Alkylation of Nucleic Acids . . . . . . . IV. Relation of Nucleic Acid Interactions to Biological Activity V. Conclusions . . . . . . . . . . References . . . . . . . . . .

. . .

.

. . .

.

. . . . . . .

. . . . . . . .

271 272 288 296 298 300 301 308

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

309

.

. . .

254 263 266 268

Biochemistry and Physiology of Bacterial Ribonucleases ALOKK . DAITA AND SALILK NIYOCI

.

I. Introduction . . . . . . . . . . . I1. A General Survey of Different Ribonucleases Present in Bacteria I11. Role of Ribonucleases in Cell Physiology . . . . . IV. Regulation of Ribonuclease Activity . . . . . . V. Physiology of Ribonuclease-Minus Mutants . . . . . VI . Concluding Remarks . . . . . . . . . References . . . . . . . . . . . Notes Added in Proof . . . . . . . . . SUBJECTINDEX.

CONTENTS OF PREVIOUS VOLUMES.

.

. . . . . . . . .

312

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

DELIABEJU (43), Laboratory of Viral Ultrastructure, Memorial SloanKettering Cancer Center, New York, New York AARONBENDICH (43), Laboratory of Cell Biochemistry, Memorial SloanKettering Cancer Center, New York, New York ELLENBORENFREUND (43) Laboratory of Cell Biochemistry, Memorial Sloan-Kettering Cancer Center, New York, New York GEORGE BRAWERMAN ( 117), Department of Biochemisty and Phamuzcology, Tufts University School of Medicine, Boston, Massachzrsetts SEYMOUR S. COHEN(15), Department of Microbiology, Uniuersity of Colorado Medical Center, Denver, Colorado ALOKK. DAITA(271), The University of Tennessee-Oak Ridge Graduate School of Biomedical Sciences and Biology Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee MARYEDMONDS (149), Department of Life Sciences, Faculty of Arts and Sciences, University of Pittsburgh, Pittsburgh, Pennsylvania DAVIDELSON( 77), Biochemistry Department, The Weizmunn Institute of Science, Rehovot, Israel PAULJ . HICGINS(43), Laboratory of Cell Biochemistry, Memmial SloanKettering Cancer Center, New York, New York SUNG-HOUKIM (181)) Department of Biochemistry, Duke University Medical Center, Durham, North Carolina W. LIJINSKY(247), Biology Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee FRITZLIPMANN( l ) ,The Rockefeller University, New York, New York BORISMACASANIK ( 99), Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts SALILK. NIYOGI(271), Biology Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee SIDNEY PESTKA (217), Roche Institute of Molecular Biology, Nutley, New Jersey TED T. SAKAI( 15), Department of Microbiology, University of Colorado Medical Center, Denver, Colorado PNINASPITNIK-ELSON ( 77), Biochemistry Department, The Weizmann Institute q! Science, Rehovot, Israel JOSE SY ( l ) ,The Rockefeller University, New York, New York MARY ANN WINTERS(149), Chemistry Department, Seton Hill College, Greensburg, Pennsylvania STEVENS. WITKIN(43), Laboratory of Cell Biochemistry, Memorial Sloan-Kettering Cancer Center, New York, New York )

ix

Dedication- ERWINCHARGAFF The composition of this volume is somewhat extraordinary, in that the first six contributions in it are, or are based upon, the presentations at a symposium honoring Erwin Chargaff on the occasion of his retirement from Columbia University after four decades of continuous service. Inasmuch as Chargaff is, in a way hinted at in the preface to Volume 15, the godfather of this serial publication as well as a major influence in the scientific area defined by its title (Nucleic Acids and Molecular Biology), it seemed appropriate that a volume containing the aforementioned presentations, as were the latter themselves, should be dedicated to him. Five of these contributions are from those for whom Chargaff filled directly the role of teacher, in that they were predoctoral graduate students with him. However, there are many, many more presently active scientists for whom he filled that role in a less direct way. His approach to his subject, the logic of his analysis, and the concise yet elegant exposition of his findings and interpretation-he is a past master in the art of self-expression-taught something to each reader of his many scientific papers. In a competitive age, he competed only with nature, all the while realizing that “even the most exact of our exact sciences floats above axiomatic abysses that cannot be explored” (1). His insistence upon proceeding stepwise from the known to the unknown, avoiding the alltoo-popular tendency to pile assumption upon assumption (or to avoid calling a spade a spade) may have cost him the priorities and the headlines cherished by many, but they have gained him the respect and the following of those concerned with the quest for truth. Perhaps the best way to learn of Chargaffs view of science and his place in it is to read his own perception of himself as “the outsider at the inside” in several essays he has written (1-6). The most recent ( 1 ) is autobiographical and deals with his youth and early career before he came to Columbia University in 1935 at the age of 30. His views of himself may not coincide entirely with the high regard in which he is held by others, especially by his colleagues and students. According to these recollections, Chargaff was deeply influenced in his early years by “Karl Kraus, the greatest satirical and polemical writer of our times . , , a fearless critic of the war [World War I ] and of the society that had given rise to it. . . , This apocalyptic writer . . . who made me sensitive to platitudes . . . to take care of words as if they were little children . . . to weigh the consequences of what I said . . . was truly my only teacher.” Chargaffs early education, “limited in scope, but on a very xi

xii

DEDICATION

high level,” encompassed languages, philosophy, history and mathematics, a little physics, but no chemistry. From that time on, even beyond the acquisition of a doctorate in chemistry at age 23, he thought of himself as a writer. More or less drifting into chemistry (“I chose chemistry for essentially frivolous reasons-knowing least about” it, and because it was “the only natural science offering hope of employment”), he continued to think of himself as a writer, having actually published some creditable items, and made a distinction between his profession (“what would feed and sustain me”) and “what one did with one’s head.” His own postdoctoral period, during which he “floated from one thing to the next,” and even his predoctoral associations, did not, in his mind, make him “the pupil . . . of any of the great establishment figures of the past. . I must say that I have not learned much from my teachers [for], in the strictest sense of the word, I have had none. During almost my entire life, I have been much more of a teacher than a pupil.”Greatness, he says, “can certainly not be transferred by what is commonly called teaching. What the disciples learn are mannerisms, tricks of the trade, ways to make a career, or perhaps, in the rarest of cases, a critical view of the meaning of scientific evidence and its interpretation. A real teacher can teach through his example-or, most infrequently, through the intensity and originality of his view, or his vision, of nature.” At the symposium, and referring to the accomplishments of his students, Chargaff remarked (as recorded on tape) : “I couldn’t help feeling that I really had very little to do with the whole thing. My definition of a teacher has always been that he’s a man who helps a student to find himself. If there’s nothing to find, of course, there’s trouble, and even the best teacher won’t make a good scientist out of a blockhead. I was really blessed in my career here at Columbia by having many, many students, almost all my students, whom I could help to find themselves. This is the only function that I see a teacher can have at a university.” Aaron Bendich (see p. 43) remarks, “His impact and influence, at the time I joined him in January 1940, were serious since he was, and is, a very serious scholar and an uncompromising investigator. His appreciation of precision and accuracy in describing biochemicals, especially of large molecular weight, left no room for ambiguities [see ref. 5 : What is DNA?]. “Hence, when the analytical composition of the nucleic acids from several tissues and animal species became clear, then, and only then, could he state that DNA appeared to reflect the species of origin, while RNA was characteristic of the tissue of origin. This was back in 1950-51,and is, in my judgment, the most important interpretation made until then regarding the chemical basis of differentiation and speciation.

. .

DEDICATION

...

Xlll

The intervening 25 years have provided numerous bases for confirmation of this fundamental concept, which guides so much current biochemical and molecular biological work and thought.” It was never my privilege to work with, or close to, Erwin Chargaff, but more than once I have found myself the recipient of direct, useful, and sharply worded advice that has been, or that might have been, if heeded, valuable. These need not be recounted here; rather, I would recall two among his many works that particularly impress( ed) me. One, which formed a critical link in the reasoning that (following the discovery here of 5’-nucleotides in RNA hydrolysates) led to the establishment of a 3’4’ phosphodiester link in RNA (versus the previously accepted 2’3’ structure), was the clever use of :‘.P, then more novel than now, to show that the appearance of isomeric glycerol phosphates upon alkaline hydrolysis of the phosphodiester of either one of them resulted from a migration of phosphorus via a cyclic intermediate. The other is, of course, the famous “Chargaff base ratios” that exist in most well-bred DNAs. These, as Bendich indicates above, were not only instrumental in the perfection of their model by Watson and Crick, but clearly foretold the spin-off (or “bonus,” as Crick describes it) that came out of the structural work: the chemical basis for the accuracy of replication and transcription. Much as one is tempted to speculate upon the origins of the concepts that led to these and others of Chargaffs contributions, or perhaps simply to list others, or to analyze the thought processes and reasoning that went into them, it is better that these be left to one who is better qualified to know and express them, namely, the man himself. Fortunately, exercising his early developed literary talents, he has set these down in writing for all to enjoy ( 1 - 6 ) . A self-styled “outsider at the inside of science,” he complies with his “own definition of a good teacher: he learned much, he taught more.” In the summarizing and para-scientific writings listed below (and quoted above), there is much that can teach all of us, not only about scientific facts but about scientific reasoning, about scientists, and about the larger world of which science is only a part. It is with admiration and affection, and perhaps a bit of trepidation lest we stray from his own well-defined standards of truth and reason, that this volume is dedicated to Erwin Chargaff by the Editor, the publisher, and by the contributors to the symposium, who speak for the many students, postdoctoral associates, and others who have been influenced by him.

W. E. C.

xiv

DEDICATION

REFERENCES 1. “A Fever of Reason: The Early Way.” Annu. Reo. Biochem. 44, 1 ( 1975). 2. “Bitter Fruits from the Tree of Knowledge: Remarks on the Current Revulsion from Science.” Perspect. B i d . Med. 16,486 ( 1973). 3. “Essays on Nucleic Acids.” Elsevier Amsterdam, 1963. 4. Quotation, in Science 190, 135 (1975). 5. “What Really in DNA? Remarks on the Changing Aspects of a Scientific Concept,.” This series, Vol. 8,297 ( 1968). 6. “Profitable Wonders: A Few Thoughts on Nucleic Acid Research,” The Sciences

15 (No.6)21 (1975).

Abbreviations and Symbols All contributors to this Series are asked to use the terminology (abbreviations and symbols ) recommended by the IUPAC-IUB Commission on Biochemical Nomenclature (CBN) and approved by IUPAC and IUB, and the Editor endeavors to assure conformity. These Recommendations have been published in many journals (I, 2 ) and compendia ( 3 ) in four languages and are available in reprint form from the NAS-NRC Office of Biochemical Nomenclature ( OBN), as stated in each publication, and are therefore considered to be generally known. Those used in nucleic acid work, originally set out in section 5 of the first Recommendations ( I ) and subsequently revised and expanded (2, 3), are given in condensed form (I-V) below for the convenience of the reader. Authors may use them without definition, when necessary.

1. Bases, Nucleosides, Mononucleotides 1. Bases (in tables, figures, equations, or chromatograms) are symbolized by Ade, Gua, Hyp, Xan, Cyt, Thy, Oro, Ura; Pur = any purine, Pyr = any pyrimidine, Base = any base. The prefixes S-, H2, F-, Br, Me, etc., may be used for modifications of these. 2. Ribonucleosides (in tables, figures, equations, or chromatograms) are symbolized, in the same order, by Ado, Guo, Ino, Xao, Cyd, Thd, Ord, Urd ( q r d ) , Puo, Pyd, Nuc. Modifications may be expressed as indicated in ( 1 ) above. Sugar residues may be specified by the prefixes r (optional), d ( =deoxyribo), a, x, 1, etc., to these, or by two three-letter symbols, as in Ara-Cyt (for aCyd) or dRib-Ade (for dAdo ) . 3. Mono-, di-, and triphosphates of nucleosides ( 5 ’ ) are designated by NMP, NDP, NTP. The N (for “nucleoside”) may be replaced by any one of the nucleoside symbols given in 11-1 below. 2’-, 3’-, and 5’- are used as prefixes when necessary. The prefix d signifies “deoxy.” [Alternatively, nucleotides may be expressed by attaching P to the symbols in ( 2 ) above. Thus: P-Ado = AMP; Ado-P = 3’-AMP.] cNMP = cyclic 3’:5’-NMP; Bt2cAMP = dibutyryl CAMP; etc.

II. Oligonucleotides and Polynucleotides 1 . Ribonucleoside Residues

( a ) Common: A, G , I, X, C, T, 0, U, \k, R, Y, N (in the order of 1-2 above). ( b ) Base-modified: sI or M for thioinosine = 6-mercaptopurine ribonucleoside; sU or S for thiouridine; brU or B for 5-bromouridine; hU or D for 5,6-dihydrouridine; i for isopentenyl; f for formyl. Other modifications are similarly indicated by appropriate lower-case prefixes (in contrast to 1-1above) (2, 3). ( c ) Sugar-modified: prefixes are d, a, x, or 1 as in 1-2 above; alternatively, by italics or boldface type (with definition) unless the entire chain is specified by an appropriate prefix. The 2’-0-methyl group is indicated by sufix m (e.g., -Am- for 2’-0-methyladenosine, but -mA- for N-methyladenosine). ( d ) Locants and multipliers, when necessary, are indicated by superscripts and subscripts, respectively, e.g., -m:A- = 6-dimethyladenosine; -s‘U- or -‘S- = 4-thiouridine; -ac‘Cm- = 2’-O-methyl-4-acetylcytidine. ( e ) When space is limited, as in two-dimensional arrays or in aligning homoxv

ABBREVIATIONS AND SYMBOLS

XVi

logous sequences, the prefixes may be placed ouer the capital ktter, the suffixes ouer the phosphodiester symbol. 2. Phosphoric Acid Residues [left side

. .

=

5’, right side = 3’ (or 2’)]

( a ) Terminal: p; e.g., pppN . is a polynucleotide with a 5’-triphosphate at one end; Ap is adenosine 3’-phosphate; C > p is cytidine 2’:3’-cyclic phosphate ( 1 , 2, 3 ) . ( b ) Internal: hyphen (for known sequence), comma (for unknown sequence); unknown sequences are enclosed in parentheses. E.g., PA-G-A-C( C,,A,U )A-U-GC >p is a sequence with a (5’) phosphate at one end, a 2’:3’-cyclic phosphate at the other, and a tetranucleotide of unknown sequence in the middle. (Only codon triplets are written without some punctuation separating the residues. ) 3. Polarity, or Direction of Chain

The symbol for the phosphodiester group (whether hyphen or comma or parentheses, as in 2b) represents a 3‘4’ link (i.e., a 5’ . . . 3’ chain) unless otherwise indicated by appropriate numbers. “Reverse polarity” ( a chain proceeding from a 3’ terminus at left to a 5’ terminus at right) may be shown by numerals or by right-toleft arrows. Polarity in any direction, as in a two-dimensional array, may be shown by appropriate rotation of the (capital) letters so that 5’ is at left, 3’ at right when the letter is viewed right-side-up. 4. Synthetic Polymers

The complete name or the appropriate group of symbols (see 11-1 above) of the repeating unit, enclosed in parentheses if complex or a symbol, is either ( a ) preceded by “poly,” or ( b ) followed by a subscript “n” or appropriate number. No space follows “poly” ( 2 , 5 ) . The conventions of 11-2b are used to specify known or unknown (random) sequence, e.g., polyadenylate = poly( A) or (A)“, a simple homopolymer; or (A&)“, a random copolypoly( 3 adenylate, 2 cytidylate) = poly( A&) mer of A and C in 3:2 proportions; poly( deoxyadenylate-deoxythymidylate)= poly[d( A-T)] or poly( dA-T) or (dA-dT). or d ( A-T)., an alternating copolymer of dA and dT; poly( adenylate, guanylate, cytidylate, uridylate ) = poly ( A,G,C,U ) or (A,G,C,U)., a random assortment of A, G, C, and U residues, proportions unspecified. The prefix copoly or oligo may replace poly, if desired. The subscript ‘‘n” may be replaced by numerals indicating actual size.

111. Association of Polynucleotide Chains 1 . AssocMted (e.g., H-bonded) chains, or bases within chains, are indicated by a center dot (not a hyphen or a plus sign) separating the complete names or symbols, e.g.: poly(A).poly(U) or (A)n*(U)m poly(A).2 poly(U) or (A)..2(U)m poly( dA-dC)*poly(d G d T ) or (dA-dC)..( dC-dT),.

xvii

ABBREVIATIONS AND SYMBOLS

2. Nonassociated chains are separated by the plus sign, e.g.:

+

2 [ p o l y ( A ) * p o l y ( U ) lpoly(A)-2 ~ poly(U) poly ( A ) or 2[A,.U,,] + A,.2UIn + A n . 3. Unspecified or unknown association is expressed by a comma (again meaning “unknown”) b&ween the completely specified chains. Note: In all cases, each chain is completely specified in one or the other of the two systems described in 11-4 above.

IV. Natural Nucleic Acids ribonucleic acid or ribonucleate deoxyribonucleic acid or deoxyribonucleate messenger RNA; ribosomal RNA; nuclear RNA “DNA-like” RNA; complementary RNA mitochondria1 DNA transfer (or acceptor or amino acid-accepting) RNA; replaces sRNA, which is not to be used for any purpose “charged” tRNA (i.e., tRNA‘s carrying aminoacyl residues); aminoacyl-tRNA may be abbreviated to AA-tRNA tRNA normally capable of accepting alanine, to form alanine tRNA or tRNA”’”, etc. alanyl-tRN A The same, with alanyl residue covalently attached. alanyl-tRNA or alanyl-tRNA”’“ [Note: fMet = fonnylmethionyl; hence tRNAfMet,identical with tRNAFet1 Isoacceptors are indicated by appropriate subscripts, i.e., tRNAt’*, tRNAt’*, etc. RNA DNA mRNA; rRNA; nRNA D-RNA; cRNA rntDNA tRNA

V. Miscellaneous Abbreviations Pi, PPI inorganic orthophosphate, pyrophosphate RNase, DNase ribonuclease, deoxyribonuclease melting temperature ( “ C ) t , (not T,) Others listed in Table I1 of Reference 1 may also be used without definition. No others, with or without definition, are used unless, in the opinion of the editors, they increase the ease of reading. Enzymes In naming enzymes, the 1972 recommendations of the IUPAC-IUB Commission on Biochemical Nomenclature (CBN) ( 4 ) , are followed as far as possible. At first mention, each enzyme is described either by its systematic name or by the equation for the reaction catalyzed or by the recommended trivial name, followed by its EC number in parentheses. Thereafter, a trivial name may be used. Enzyme names are not to be abbreviated except when the substrate has an approved abbreviation (e.g., ATPase, but not LDH, is acceptable).

REFERENCES’ 1. IBC 241, 527 (1966); Bchem 5, 1445 (1966); BJ 101, 1 (1966); ABB 115, 1 (1966), 129, 1 ( 1969); and e1sewhere.t

* Contractions for names of journals follow. f Reprints of all CBN Recommendations are available from the Office of Biochemical Nomenclature (W. E. Cohn, Director), Biology Division, Oak Ridge National Laboratory, Box Y, Oak Ridge, Tennessee 37830, USA.

xviii

ABBRFNIATIONS A N D SYMBOLS

2. EJB 15, 203 (1970); JBC 245, 5171 (1970); J M B 55, 299 (19771); and elsewhere.” 3. “Handbook of Biochemistry” (H. A. Sober, ed.), 2nd ed. Chemical Rubber CO., Cleveland, Ohio, 1970,Section A and pp. H130-133. 4. “Enzyme Nomenclature,” Elsevier Scientific h b l . Co., Amsterdam, 1973. 5. “Nomenclature of Synthetic Polypeptides,” JBC 247, 323 ( 1972);Biopolymrs 11, 321 ( 1972); and elsewhere.*

Abbreviations of Journal Titles Journals Annu. Rev. Biochem. Arch. Biochem. Biophys. Biochem. Biophys. Res. Commun. Biochemistry Biochem. J. Biochim. Biophys. Acta Cold Spring Harbor Symp. Quant. Biol. Eur. J. Biochem. Fed. Proc. J. Amer. Chem. SOC. J. Bacteriol. J. Biol. Chem. J. Chem. Soc. J. Mol. Biol. Nature, New Biology Proc. Nat. Acad. Sci. U.S. Proc. ‘SOC.Exp. Biol. Med. Progr. Nucl. Acid Res. Mol. Biol.

Abbreviations used ARB ABB BBRC Bchem BJ BBA CSHSQB EJB FP JACS J. ,Bact. JBC JCS IMB Nature NB PNAS PSEBM This Series

” Reprints of all CBN Recommendations are available from the Office of Biochemical Nomenclature (W. E. Cohn, Director), Biology Division, Oak Ridge National Laboratory, Box Y, Oak Ridge, Tennessee 37830,USA.

Some Articles Planned for Future Volumes Mechanisms in Polypeptide Chain Elongation on Ribosomes

E. BERMEK The Ribosome of

E. coli

R. BRIMACOMBE, K. H. NIERHAUS,R. A. GARRETT,AND H. G. WIITMAN Structure and Function of 5

S and 5.8 S RNA

V. A. ERDMANN Initiation of Protein Synthesis

M. GRUNBERG-MANAGO High Resolution Nuclear Magnetic Resonance of the Structure of tRNA in Solution

DAVID R. KEARNS Ribosomal tRNA Binding Sites

H. MATTHAEI Protein Synthesis

S. OCHOA Premelting Changes in DNA Conformation

E. PALA~EK Quantum Mechanical Investigation of the Electronic Structure of Nucleic Acids and Their Constituents

B. PULLMAN The Biochemical and Microbiological Action of Platinum Compounds A. J. THOMSONAND J. J. ROBERTS Structure and Functions of Ribosomal RNA

R. ZIMMERMANN

This Page Intentionally Left Blank

The Enzymic Mechanism of Guanosine 5’,3’-Polyp hosphate Synthesis

I

FRITZLIPMANNAND

JOSE

SY

The Rockefeller University New York, New York

I. Introduction . . . . . . . . . . . . 11. The in Vitro Reaction . . . . . . . . . . A. The Ribosomal System . . . . . . . . . B. Mechanism of Pyrophosphate Transfer . . . . . . C. Definition of the Position into Which the Pyrophosphate Is Transferred on the Guanosine Derivative . . . . . D. Role of the Ribosome in the Formation of ppGpp from ATP in the Transfer of Pyrophosphate from ATP to GTP and GDP . 111. Nonribosomal Synthesis of ppGpp and pppGpp . . . . . A. Pyrophosphate Transfer from ATP to GTP or GDP Catalyzed by the Stringent Factor . . . . . . . . . B. Reversibility of the Pyrophosphate Transfer between ppGpp and pppGpp and ATP . . . . . . . . . . IV. Conclusion . . . . . . . . . . . . References . . . . . . . . . . . .

7 8

9 9

13 14 14

1. Introduction It is a great pleasure to be asked to take part in this symposium in honor of Erwin Chargaff. He has done admirable work, in particular the part of it that provided the frame around which the double-helix of DNA was going to be wound. However, not only his contribution as a scientist, but his human, or one might say humanistic, qualities are unusual. In many of his lectures and in his book he has, by sometimes sharp but witty turns, provided a welcome diversion in the generally charmless, dry scientific literature. Our contribution to the symposium is in the field of nucleotides with which he has been foremostly concerned. It will deal with the two initially rather strange guanosine derivatives that were discovered by Cashel and Gallant (1) and were originally called “magic spot” compounds because for some time their chemistry and function were not well defined. However, they were soon reported by Cashel and Kalbacher (2) to be guanosine polyphosphates, a tetra- and a pentaphos1

2

FRITZ LIPMANN AND JOSE SY

phate,’ that, in addition to polyphosphate in the usual 5’-position, contained a pyrophosphate group that was assumed, since the compounds were not split by periodate, to be attached to either the 2’or the 3’-position on the ribose. These compounds were found to be formed by “stringent” strains of Escherichia coli, strains that are deficient in certain amino acid biosyntheses and, if deprived of the amino acid, not only cease to synthesize proteins but also stop synthesizing RNA. This is in contrast to “relaxed strains that, on omission of deficient amino acids, only stop synthesizing protein but continue with RNA synthesis.

II. The in Vitro Reaction A great deal of progress in understanding the mechanism by which these compounds are produced was obtained by the relatively recent work of Haseltine et al. (3), who found them to be synthesized by a phosphate transfer from ATP to GTP or GDP; the transfer was dependent on a stringent factor found in the 0.5-1.0 M NH,C1 ribosomal wash from stringent E . coli. The factor is not found on ribosomes of relaxed E . coli. In their experiments, a system composed of the ribosome, the stringent factor, ATP and GTP was used. It had been assumed, on what we felt was not quite convincing evidence, that pyrophosphate was transferred as a unit from ATP to GTP or GDP. Furthermore, the exact position in which the pyrophosphate was inserted still remained to be decided, although the assumption that it went into either the 2’- or the 3’-position seemed well founded. We had been interested for a long time in the role of GTP in polypeptide elongation, and the appearance of these new guanosine polyphosphates caused by amino-acid deficiency had created a great deal of speculation about their derivation during a modified protein synthesis. A tendency developed to rethink the role of GTP, and possibly to look for a function of ATP in peptide elongation. We felt an urge, therefore, to use the possibility provided by this work to analyze in detail the enzymic reaction responsible for the phosphate transfer from ATP to GTP.

A. The Ribosomal System We show in Fig. 1 the method (2, 3) by which guanosine polyphosphates, in particular ppGpp and pppGpp, are separated. Figure 1 shows The abbreviations are: ppGpp, guanosine 5’-diphosphate 3‘-diphosphate;

*

pppGpp, guanosine 5’-triphosphate 3’-diphosphate; an asterisk over a “p,” e.g., p, denotes a ”P-label in that position.

ENZYMIC SYNTHESIS OF CUANOSINE POLYPHOSPHATES

3

FIG. 1. Radioautograph of polyethyleneimine-cellulosechromatogram. For details, see Fig. 2.

a sample analysis of products obtained by reaction between GTP and ATP with crude stringent factor and E . coli ribosomes. At various time intervals in the experiments, samples were removed and protein was precipitated with formic acid. The formic acid extract was then applied to a polyethyleneimine-cellulose thin-layer chromatography plate and developed with 1.5 M KH,PO,, pH 3.4 ( 2 ) . In this system, phosphorylated guanosine polyphosphates migrate at rates that are the inverse of their phosphate content. Figure 2 shows an experiment of the type performed by Haseltine et al. ( 3 ) . In the presence of ATP, ribosomes and crude stringent-factor, there is a rapid breakdown of GTP by the EF-G-activated ribosomal GTPase. This causes the concentration of pppG to fall sharply and to remain low while ppG increases; therefore, ppGpp dominates as the reaction product. Because GTPase attacks GTP as well as pppGpp, the latter being converted to ppGpp, the figure indicates not only that this system produces ppGpp directly but also that, since the concentration of pppGpp always remains low and declines toward the end, some of the ppGpp is derived from the pppGpp formed initially. By using a purified fraction of the stringent factor from which most of the GTPase activity was removed, we eventually learned to retain pppGpp as a

4

FRITZ LIF'MANN AND JOSE SY

8

7

6

c

m

\

I

0

x 4

E a

\

u

3

0

2

I

p-I

1

20

6

Y

I

I

40

I

I

60

I

I

80

I

I

100

Time (min)

FIG.2. Kinetics of ppGpp and pppGpp synthesis. NHC1-washed ribosomes (145 pg, 2 A, units) and ribosomal wash (39 fig) were incubated at 37°C in 50 pI of 40 mM Tris-acetate (pH 7.8),2 mM dithiothreitol, 10 mM MgCh, 100 mM NHCl, 0.43 mM [~-*2P]GTP(23 Ci/mol), and 3.6 mM ATP. At various time points, 5 pl of reaction mixture was withdrawn and mixed with 5 ~l of 3.5% formic acid. After removal of the precipitated ribosomes by centrifugation, 2 pl of the supernatant fraction was applied to polyethyleneimine-celluloseplates, which were then dried, washed with absolute methanol, and developed with 1.5 M KH,PO, (pH 3.4) ( 2 ) . The developed plates were dried and radioautographed. Regions corresponding to GDP, GTP, ppGpp, and pppGpp were then cut out and counted in a Packard scintillation counter with Bray's solution ( 4 ) ,

5

ENZYMIC SYNTHESIS OF GUANOSINE POLYPHOSPHATES

product of the reaction between GTP and ATP. However, the experiments we discuss in the first part of this paper deal largely with ppGpp, which, in an experiment like that shown in Fig. 2, can be prepared containing various markers by transferring pyrophosphate from ATP specifically marked with (5).

B. Mechanism of Pyrophosphate Transfer The two problems of primary interest were ( a ) to determine exactly the manner in which pyrophosphate is inserted into the guanosine 5'polyphosphates, and ( b ) to find the position at which it is attached. Two different modes by which pyrophosphate is transferred from ATP are known. One was explored by Khorana et al. ( 6 ) , who studied the transfer of pyrophosphate from ATP to ribose 5-phosphate to form the 1-pyrophosphate; they found the pyrophosphate to be transferred as a unit: Rib(5)P

+ Ado-P-PB-PY + P-P@(l)Rib(5)P+ Ado-P"

(1)

The PO-PY moiety at ATP was transferred with the PO ending up attached to the ribose. Another mode of pyrophosphate ester generation had been found (7, 8) in the synthesis of mevalonyl pyrophosphate, an intermediary in steroid biosynthesis. In this case, the donor again is ATP; however, phosphate transfer occurs in two steps, first yielding a monophosphate that, in the second step, is phosphorylated by another molecule of ATP: Mevalonic acid Mevalonyl-PY

+ Ado-P-PB-PY

+

+

mevalonyl-PY ADP (a) A d o - P - P - P + mevalonyl-PY-PY ADP (b) +

+

(2)

Therefore, in guanosine polyphosphate synthesis, if pyrophosphate was transferred as a unit from [p-32P]ATP, mild acid hydrolysis of the 3'-linked pyrophosphate should leave radioactive phosphate on the ribose, the released inorganic phosphate being unlabeled. With a two-step transfer, the p-label would not become attached to the guanosine; only a 7-label transferred from the donor ATP would yield 32P1on hydrolysis (cf. Eq. 2). Accordingly, we prepared both the P- and 7-labeled ATPs and, after synthesis of ppGpp, analyzed for the localization of the label by an acid hydrolysis which we knew to yield ppGp. It appeared that, on acid hydrolysis, the p-label remained attached to ppGp but the 7-label was released (Fig. 3). This showed that the compound formed was obtained by pyrophosphate transfer as in Eq. 1, similar to the transfer

FRIlZ LIPMANN AND JOSE SY

6

* PPGP

1

A

CPm 3000 - P P ~ P

PGP

I

Pi

I

1

1000 -Origin t 1

J

I

I

I

x

Pi

3ooot

0

5

10

15 crn

FIG.3. Stepwise acid hydrolysis of variously labeled ppGpp. ppGpp preparations were labeled with "P [labeled and isolated as described ( 5 ) ] as follows: ( A )

*

ppG;p (23,000 cpm), and ( B ) ppGpp (27,000 cpm). kor acid hydrolysis they were incubated in 10 pl of 1 N HCI at 37°C for 30 minutes. Hydrolysis was stopped.by chilling in ice and the addition of 10 ~l of 1 M TrisCl (pH 10). The neutralized samples were applied to polyethyleneimine-cellulose sheets, and excess salts were washed away by immersion in absolute methanol (9). Thin-layer chromatography sheets were developed in 0.75 M KH2P0, (pH 3.4). The radioactivity of the sheets was scanned with a Varian aerograph Berthold radioscanner. Under the conditions of this experiment (1 N HCl, 37"C, 30 minutes), ppGpp is converted to ppGp; longer hydrolysis will further convert ppGp to pGp ( 5 ) .

7

ENZYMIC SYNTHESIS OF GUANOSINE POLYPHOSPHATES

from ATP to ribose 5-phosphate in the reaction that yields ribosyl pyrophosphate 5-phosphate.

C. Definition of the Position into Which the Pyrophosphate Is Transferred on the Guanosine Derivative

The 3’-nucleotidase found in extracts of ryegrass by Shuster and Kaplan (10) has been very useful for locating phosphates added to 5’mbstituted nucleotides. The enzyme was first used by Kaplan et al. (11) to show that the second phosphate in coenzyme A is in the 3’-position. Later, during the analysis of “active sulfate,” Robbins and Lipmann (12) used the enzyme to show again that the second phosphate is in the 3’-position. Another added phosphate, namely, that in NADP (11), is untouched by this 3’-nucleotidase, which led to the conclusion that it is in the 2’-position. When ppGpp is treated by the 3’-nucleotidase, no phosphate is released; however, the nucleotidase appears to react only with monophosphates, not with pyrophosphates. It was necessary to remove the second phosphate from the pyrophosphate link. Although this is easily achieved by mild treatment with acid, the 2’- and 3’-positions are isomerized by this treatment; hence, it was not practical here to use acid hydrolysis to remove the terminal phosphate. We remembered, then, that Kunitz’s yeast inorganic pyrophosphatase, if activated by zinc instead of magnesium ( 1 3 ) , attacks polyphosphate groups bound to nucleotides; it hydrolyzes all pyrophosphate links but leaves intact the phosphate directly bound to the nucleotide. Again choosing GDP as

*

phosphate acceptor and [jPP]ATP as donor for preparing ppGpp, the

*

latter was incubated with the zinc pyrophosphatase to produce pGp, which was then used for the test with the ryegrass enzyme.

*

As shown in Fig. 4, using pGp obtained enzymically from ppGpp,

*

the p was found to be nearly 100%split by 3’-nucleotidase. However,

*

when acid-produced pGp was assayed with the 3’-nucleotidase (lower curve), only half of the 32Pwas split owing to the expected acid-catalyzed 3‘-2’-isomerization. Thus, Fig. 4 indicates attachment of the pyrophosphate to the 3’-position, and incidentally confirms the specificity of the nucleotidase for 3’-phosphates. Although the analysis was carried out only on ppGpp, pppGpp is expected to be formed in a similar fashion since some of the ppGpp, as prepared (Fig. 2), actually should be derived from pppGpp during the reaction.

FRI'R, LIPMANN AND JOSE SY

8

90

-

80 70

-

60

-

50

-

--

60 80 T irne (rnin)

100

120

*

FIG. 4. Digestion by 3'-nucleotidase of pGp, prepared enzymically or by acid

*

hydrolysis. pGp ( 4000 cpm) obtained by inorganic pyrophosphatase digestion (0-0) and (5000 cpm) hydrolysis in HC1 (0-0) for 18.5 hours at 37°C (S),was incubated at 37°C in 100 a1 of 20 mM Tris (pH 7.5)containing 0.002 unit of 3'-nucleotidase. At various times, lO-pl aliquots were withdrawn, applied directly to polyethyleneimine-cellulose sheets, and air-dried. Chromatograms were developed with 0.75 M KHIPOI (pH 3.4)and radioautographed overnight. Areas corresponding to pGp and Pi were cut out and counted in a scintillation counter. The percentage

*

of PI released corresponds to the percentage of **Pcounts released from pGp.

D. Role of the Ribosome in the Formation of ppGpp from ATP in the Transfer of Pyrophosphate from ATP to GTP and GDP

In all the experiments discussed, the formation of ppGpp and pppGpp was followed in a ribosomal system with added stringent factor present. Haseltine et a2. (3) reported some effects of EF-G on this transfer reaction; however, on further investigation, these effects seemed to be an accessory phenomenon. Thiostrepton, which inhibits binding of the tRNA-Tu * GTP complex to the ribosome, consistently inhibits; this indicates a relation of guanosine polyphosphate synthesis to the binding of tRNA to the ribosome. Furthermore, in oivo experiments had been

9

ENZYMIC SYNTHESIS OF GUANOSINE POLYPHOSPHATES

TABLE I GUANOSINE POLYPHOSPHATE SYNTHESIS BY Bacillus brevis RIBOSOMEV Additions Ribosomes: 40 pg 102 Pg 204 rg Ribosomes (40 pg) Stringent factor

Percent GTP converted to PPGPP PPPGPP

+

3

+ stringent factor

5 12

62 0

The preparation of B. brevis ribosomes and the assay for guanosine polyphosphates were as described (16). Stringent factor was that of fraction I1 previously described (18).

interpreted by Cashel and Gallant in their early work (1) on the formation of these compounds to be caused by an “idling” reaction in which elongation fails because of the missing amino acid. Recently, Pederson et al. ( 1 4 ) observed that, using purified preparations of ribosomes and stringent factor in oitro, the reaction becomes dependent on the presence of uncharged tRNA in addition to mRNA. They also showed that synthetic mRNAs are quite suitable for complementation. In testing for specificity, it was noted early that, in the presence of added stringent factor, the ribosomes of relaxed strains are just as active as those from stringent ones. Furthermore, we assayed ribosomes obtained from a representative of the Bacillus species ( B . b r e d , ATCC 8185); they responded with a slight formation of ppGpp and pppGpp, which was increased to levels even higher than found with E . coli ribosomes by the addition of E . coli stringent factor (Table I ) . We also tested ribosomes from the cytoplasm and chloroplasts of Chlamydomom reinhardtii ( 15); the chloroplast ribosomes corresponding to their prokaryotic type reacted well, but only when supplemented with E. coli stringent factor. However, the cytoplasmic, i.e., eukaryotic, ribosomes were inactive. This had been generally recognized, but recently ppGpp has been reported (16, 17) to be formed in some eukaryote cells.

111. Nonribosomal Synthesis of ppGpp and pppGpp A. Pyrophosphate Transfer from ATP to GTP or GDP Catalyzed by the Stringent Factor

We have discussed the action of the ribosome in some detail in order to contrast it with the ribosome-independent reaction that has

10

FRITZ LIPMANN AND JOSE SY

been found (18). During purification of the stringent factor, we became aware of such an activity of the factor. It was rather small, but could be strongly stimulated by the addition of 20% methanol. In the absence of the ribosome, the factor became unstable and quite temperature-sensitive; it deteriorated rapidly at temperatures over 30°C. An experiment showing the effect of methanol addition as compared to the very slow reaction in the absence of methanol is shown in Fig. 5. In this and some of the following experiments, stringent-factor fractions of higher degrees of purification were used. They were obtained by taking a fraction from the 0.5 M NH,C1 ribosomal wash between 0 and 35%ammonium sulfate; further purification results when the precipitate obtained on dialysis against low-salt buffer is dissolved in 0.5 M ammonium chloride. This solution represents a stringent factor preparation free of, or very low in, GTPase and is useful for tests on the acceptor function

,

I

1

1

3

1

,

5

,

,

7

,

1

9

Hours

FIG.5. Time c a m e of ppGpp and pppGpp formation in the nonribosomal system. Reaction mixtures in 100 pl contained 20 mM Tris-acetate (pH 8 ) , 20 mM Mg(OAc)*, 4 mM dithiothreitol, 150 mM NH4Cl, 0.4 mM [or-"PIGTP (70Ci/mol), 4 mM ATP, 4.5 pg of fraction I NH4Cl wash, and 2m methanol (0-0) or no Incubations were at 28"C, and at various time points aliquots methanol (0-0). of 5 a1 were mixed with 5 pl of 1.76%formic acid. Precipitates were removed by centrifugation, and 2 d of the supernatant was assayed for ppGpp and pppGpp formation. The results are expressed as percentage of added [cr-**P]GTP converted to PPGPP and PPPCPP.

ENZYMIC SYNTHESIS OF GUANOSINE POLYPHOSPHATES

11

of guanosine triphosphate, which is quickly split with impure stringent factor as shown in Fig. 2. The rate of the nonribosomal reaction with increasing concentration of purified stringent factor is followed in Fig. 6. At low concentration, probably because of instability, activity is relatively low, but with rising concentration it increases proportionately up to a maximum level. Figure 6 shows that the nonribosomal reaction with high concentrations of a purified extract can be rather considerable since 40-50% of the added GTP is converted within a 90-minute period. To verify strictly a nonribosoma1 reaction, it seemed important to exclude the presence of any ribosomal remnants. Therefore, the stringent-factor preparation was centrifuged in a sucrose gradient and samples corresponding to various molecular weights were taken. It may be seen in Fig. 7 that activity was found only in the fraction appearing at the low-molecular-weight level, and none could be found in the region of either one of the two ribosomal subunits. This makes us feel sure that the observed effect is due to the factor, not to some particulate contamination. In another experiment, not shown here, the sucrose gradient analysis was used with the purified fraction of the stringent factor, and its activity was checked both with the ribosomal system and the nonribosomal system. In this experiment, at the region corresponding to a molecular

50 501 a

{

40-

+a

30-

n a

a

sa

I 10

/*/*

-

-/*

i

/*

tt

fl-0 I

5

I

10

1

15

I

20

pg protein

FIG. 6. Dependence of ppGpp and pppGpp formation on enzyme concentrations in the nonribosomal system with and without methanol. Reaction mixtures in 50 ~1 contained 20 mM Tris-acetate (pH 8 ) , 10 mM Mg( OAc),, 4 mM dithiothreitol, 100 mM NH,Cl, 0.4 mM [a-”PIGTP (20 Ci/mol), 4 mM ATP and various amounts of the enzyme preparation. Incubation was at 30°C for 90 minutes. ppGpp and pppGpp were assayed: 0-0, +20% methanol; 0-0, no methanol. Buffer containing 1 mg/ml of bovine serum albumin was used for enzyme dilution.

12

FRI'IZ LIPMANN AND JOSE SY

F

"'1 08

2000

07 a n

06

1500

2 8

+

8

05

'3 , a

E T

C ._

04

1000 0

03

500

02

01

Fractions

5%

10

Sucrose

20

20 %

FIG. 7. Sucrose density gradient centrifugation of NH,Cl wash. Crude NH,Cl wash (254a g ) in 20 mM Tris-acetate ( p H 7.8), 10 mM Mg( OAc)*, 250 mM NH,Cl and 2 mM dithiothreitol was layered on a 5 to 20% linear sucrose density gradient containing 40 mM TrisCl (pH 7.5), 25 mM NH,Cl, 2 mM dithiothreitol and 0.1 mM MgCL. Parallel gradients were loaded with dissociated ribosomes to serve as sedimentation standards. Sucrose gradients were spun in an International centrifuge with the SB283 rotor at 39,000 rpm at 18°C for 3 hours. The gradients were unloaded from the top with an ISCO gradient fractionator, and the ultraviolet absorption at 260 nm was continuously monitored (-). For activity measurement (lower part of figure), fractions of 0.5 ml were collected and, from each, 20 pl was diluted with reaction mixture to 50 PI, giving a final concentration of 20 mM Tris-acetate ( p H 7.8), 25 mM Mg(OAc), 150 mM NH,CI, 3 mM dithiothreitol, 4 mM ATP and 0.4 mM [a-"PIGTP (32 Ci/mol). Methanol was added to one set of reaction mixtures (0-0) to a final concentration of 20%, while the other set received no such addition (0-0). Reactions were done at 25°C for 18 hours. ppGpp and pppGpp were then analyzed.

weight of 75,000, the activities in both the ribosomal and nonribosomal systems peaked sharply and coincidentally. We consider this experiment as definite proof that the stringent factor is the transferring enzyme, and that the considerable speed-up and protection supplied by the ribo-

ENZYMIC SYNTHESIS OF GUANOSINE POLYPHOSPHATES

13

soma1 system is a combination effect whereby the stringent factor interacts with the ribosome in a manner that we do not fully understand; it requires for this stimulation, as mentioned, the presence of uncharged tRNA containing the matching anticodon.

B. Reversibility of the Pyrophosphate Transfer between ppGpp and pppGpp and ATP Recent experiments by Sy (19) showed that a reverse reaction occurs if these guanosine polyphosphates are incubated with AMP in a ribosomal or nonribosomal system. The effect of the concentration of AMP is assayed in Fig. 8; it indicates that an excess of AMP is required. pppGpp is a much better donor to AMP than ppGpp, but ATP is strongly inhibitory and blocks the reverse reaction at low concentrations. Thus, reversal is observed only when the concentration is tilted toward excess AMP; in view of the inhibitory effect of ATP, it is unlikely that an appreciable reversal would be expected in the living organism. The enzymic reactions between GTP or GDP and ATP are thus essentially reversible, and the overall effect may be written: G T P + A T P t , PPPGPP AMP.

+

5’-AMP ( n r n )

FIG.8. 5’-AMP concentration curve. The standard reaction mixture contained in 20 PI: 20 mM Tris-OAc ( p H 8.1), 0.1 mM dithiothreitol, 10 mM Mg(OA),, 1.5 x

*

lo-‘ M pppGpp (20 Ci/mol), 29 fig of ammonium chloride-washed ribosomes, 1 pg of fraction I1 stringent factor, 2 fig of poly( A,U,G), 5 pg of tRNA, and the indicated amounts of 5’-AMP. Incubations were at 30°C for 90 minutes, and the reactions were stopped by the addition of HCOOH. The resulting precipitates were centrifuged and 10 pI of the supernatant was spotted on polyethyleneimine-celluloseplates.

14

FRJTZ LIPMANN AND JOSE SY

IV. Conclusion In this paper, we have analyzed the enzymic synthesis of the two new guanosine polyphosphates. In the first approximation, it seemed essential to understand the mechanism of their biosynthesis and clearly define their structure. One cannot help being fascinated by the variety of unusual nucleotide phosphates that are being formed. Now that the excitement about the 3’: 5’-cyclic nucleotide monophosphates is ebbing somewhat, we again have another strange metabolic creation appearing. We have entirely omitted here the problem of how these compounds achieve the metabolic effect characteristic of the stringent response, namely, that their formation triggers the cessation of synthesis of the bulk of cellular RNA, the tRNA and rRNA. There is as yet no fully satisfactory explanation for the mechanism by which this stoppage is achieved. ACKNOWLEDGMENT The work reported here was supported by grant GM-13972 from the U.S. Public Health Service.

REFERENCES 1. M. Cashel and J. Gallant, Nature (London) 221, 838 ( 1969). 2. M. Cashel and B. Kalbacher, JBC 245, 2309 ( 1970/).’ 3. W. A. Haseltine, R. Block, W. Gilbert, and K. Weber, Nature (London) 238, 381 (1972). 4. G. A. Bray, Anal. Biochem. 1,279 ( 1960). 5. J. Sy and F. Lipmann, PNAS 70,306 ( 1973). 6. H. G. Khorana, J. G. Fernandes, and A. Kornberg, IBC 230, 911 (1958). 7. S. Chaykin, J. Law, A. H. Phillips, T. T. Tchen, and K. Bloch, PNAS 44, 998 (1958). 8. U. Henning, E. M. Moslein, and F. Lynen, ABB 83,259 ( 1959). 9. K. Randerath and E. Randerath, in “Methods in Enzymology,” Vol. 12: Nucleic Acids (L. Grossman and K. Moldave, eds. ), Part A, pp. 323447. Academic Press, New York, 1967. 10. L. Shuster and N. 0. Kaplan, JBC 201,535 (1953). 11 . T. P. Wang, L. Shuster, and N. 0. Kaplan, JBC 206,299 ( 1954). 12. P. W. Robbins and F. Lipmann, JACS 78,2652 ( 1956). 13. M. J. Schlesinger and M.J. Coon, BBA 41, 30 (1960). 14. F. S . Pedersen, E. Lund, and N. 0. Kjeldgaard, Nature N B 243, 13 (1973). 15. J. Sy, N. H. Chua, Y. Ogawa, and F. Lipmann, BBRC 56, 611 (1974). 16. C. Klein, FEBS Lett. 38, 149 ( 1974). 17. J. D. Irr, M. S. Kaulenas, and B. R. Unsworth, Cell 3, 249 ( 1974). 18. J. Sy, Y. Ogawa, and F. Lipmann, PNAS 70,2145 (1973). 19. J. Sy, PNAS 71,3470 (1974).

Effects of Polyamines on the Structure and Reactivity

of tRNA 1

TEDT. SAKAIAND SEYMOURS. COHEN Department of Microbiology University of Colorado Medical Center Denver, Colorado

1. Introduction

.

.

.

.

.

.

11. Roles of Polyamines in Biological Systems

.

.

.

.

.

.

.

.

.

. .

A. Some Biological and Biochemical Functions of Polyamines B. The Binding of Polyamines to Nucleic Acids and Other . . . . . . . . Cellular Components . C. Transfer RNA and Cation Binding . . . . . . 111. Chemical and Physicochemical Studies . . . . . . A. Reactions of 4-Thiouridine in Escherichia coli tRNA . . B. Cation Effects on the pK of s'U in tRNA . . . . C. Thermal Denaturation Studies . . . . . . . D. Chemical Reactions of 4-Thiouridine. . . . . . E. Reactions of Co'+-tRNA. . . . . . . . . F. Effects of Cations on p-Chloromercuribenzoate-Labeled tRNA G. Spin-Label Studies . . . . . . . . . H. Binding Studies . . . . . . . . . . IV. Conclusions . . . . . . . . . . . References . . . . . . . . . . .

. .

15 16

. . . .

17 19 21 21 23 24 25 28 29 30 34 38 39

.

.

. .

. .

. .

.

.

16

1. Introduction In presenting a joint paper on this occasion, the authors are indicating that thc influence of Erwin Chargaff will have some continuity, for several scientific generations at least, We think that the subject of our paper is one that he may appreciate particularly, in that we express our interest in all parts of the natural nucleates-their much neglected cations as well as the anionic polynucleotides. The topic may be appropriate also in that the structures of the organic cations with which we are concerned wcre elucidated just about fifty years ago (1,2 ) , at the time when Erwin Chargaff began his own professional career. Although crystals of spermine phosphate were first observed in human semen by Leeuwenhoek in 1678 ( 3 ) and rediscovered by Vauquelin in 1791 ( 4 ) , the biological importance of the polyamines has only re15

16

TED T. S A W 1 A N D SEYMOUR S. COHEN H,N(C%),NH,

H,N(CH,),NH(CH,),NH,

putrescine

spermidine

H~N(cH,)~NH(cH,),NH(c~)~NH,

spermine

FIG. 1. Structures of the common polyamines putrescine, spermidine and spermine.

cently been recognized. The common polyamines spermidine and spermine and the diamine putrescine (Fig. 1) occur in essentially all plant and animal cells; prokaryotic cells appear not to synthesize spennine ( 5 ) . In Escherichia coli, the most thoroughly studied organism, putrescine and spermidine are synthesized in large amounts in glucose-salts media, suggesting important metabolic functions for these amines. The polyamines are also found in DNA and RNA viruses of bacteria (S), plants and animals. The putrescine, spermidine and MgZ+present in bacteriophage T4 are sufEcient to neutralize a large part of the phage DNA (7). Turnip yellow mosaic virus (TYMV) and some other plant viruses have been shown to contain spermidine (8). Recently, we noted the presence of spennidine in the RNA isolated from R17 phage ( 9 ) and from TYMV (10). Infection of E. coEi by R17 leads to a marked accumulation of spermidine, which parallels that of RNA (11). In 1948, Herbst and Snell ( 1 2 ) identified putrescine as a factor essential for the growth of Haemophilus parainflzrenzae. Since that time, polyamines have been shown to be stimulatory to (or essential for) the growth of various organisms, from bacteria (13-15) and fungi (1 6 ) to the Jerusalem artichoke and cultured animal cells ( 5 , 1 7 ) ,The cellular levels of these cations are markedly increased in various stages of tissue development (18, 19) and of embryonic growth (20-22). In recent findings, it appears that patients with certain tumors excrete increased amounts of polyamines in the urine, and that these increases may serve to indicate tumor proliferation (23).

II. Roles of Polyamines in Biological Systems A. Some Biological and Biochemical Functions of Polyamines In addition to effects such as growth stimulation, better defined biological systems have been employed to help define sites of polyamine action. Dion and Cohen ( 2 4 ) showed that both multiplication of bacte-

EFFECTS OF POLYAMINES ON

tRNA

17

riophage T4D and DNA synthesis are inhibited severely in a putrescinedepleted strain of E . coli. Addition of either putrescine or spermidine to this system restores DNA synthesis and virus yield. Polyamines have a role in translation in this putrescine-deficient organism infected with the RNA phage f2 ( 2 5 ) . It has been suggested that aminoacylation of tRNA is the rate-limiting step in the growth of putrescine-depleted cells ( 2 6 ) . Polyamines have a stimulatory effect on normal RNA synthesis. Uridine incorporation is stimulated by the addition of polyamines to various organisms (27, 28) and organelles ( 2 1 ) .We have shown correlations between the synthesis of polyamines and of stable RNA in E . coli (5, 2 9 ) , suggesting a possible regulatory role of the polyamines in the systems studied. In in vitro systems, polyamines increase RNA synthesis with RNA polymerases isolated from several organisms ( 3 O 3 2 ) , both by enhancing the binding of polymerase to DNA templates and by displacing RNA products. Spermidine has differential effects on the selection of cistrons to be transcribed in some phage systems ( 3 3 ) . Although this paper is generally confined to a discussion of polyamine effects on RNA, it should be noted that polyamines are important in steps of DNA replication, as in the marked stimulation of T4 polynucleotide kinase (34) and in the activity of various DNA polymerases (35-37). In effects on protein synthesis, the polyamines help to maintain ribosomal structure ( 3 8 , 3 9 ) as well as to facilitate the assembly of ribosomal subunits ( 4 0 ) . Spermidine appears to be required for the function of ribosomes isolated from one organism (41) and is a component of the ribosomal association factor in another ( 4 2 ) . Also, polyamines can activate the formation of aminoacyl-tRNA in the absence of Mg2+ (43-46). Spermidine and spermine can stimulate the in vitro incorporation of amino acids into protein in numerous systems (43, 47, 48), and can replace substantial portions of the Mg2+requirements in such systems; e.g., by various combinations plus suboptimal concentrations of MgZ+, either spermidine or putrescine plus spermidine increase the p d y ( U)directed or R17-RNA-directed synthesis of protein to a level attained in the presence of optimal Mg" concentrations alone ( 9 ) .

B. The Binding of Polyamines to Nucleic Acids a n d Other Cellular Cornponents

Because of their cationic nature, polyamines bind strongly to nucleic acids. Tsuboi ( 4 9 ) and Liquori et al. ( 5 0 ) postulated a model for the interaction of polyamines with double-helical nucleic acids to explain the stabilization of nucleic acids by polyamines (51, 5 2 ) . In the model

18

TED T. SAKAI AND SEYMOUR S. COHEN

(Fig. 2), spermidine (or spermine) interacts with 3 (or 4) phosphate groups; the tetramethylene portion of the polyamine bridges the narrow groove of the helix, and the trimethylene portion( s ) bridges adjacent phosphate groups in one chain. Not only can polyamines stabilize DNA, as determined by increases in melting temperature, but the levels of polyamine required for this stabilization are much lower than the concentrations of MgZ+necessary for comparable effects (53).Similar observations have been made on the stabilization of synthetic polyribonucleotides by polyamines ( 54-56). Polyamines have been implicated in various systems that involve proteins and membrane components (57, 58) in addition to nucleic acid, suggesting that the cations may facilitate interactions between nucleic acids and other cellular components. Levy and co-workers described a nuclease from Citrobacter that is normally specific for poly( U ) , but that is altered in its specificity by the addition of spermidine (59, 60); their evidence indicates that one molecule of the polyamine is bound to a molecule of the enzyme. Fukuma and Cohen (9, 10) have shown that significantly more spermidine is bound to intact virions of phage R17 than is found on the isolated viral RNA and suggested that the

hc. 2. Tsuboi model for interaction of spermine with a double helical nucleic acid. (Courtesy of M. Tsuboi. )

EFFECTS OF POLYAMINES ON

tRNA

19

polyamine plays a role in “normal” virus assembly, such as in the assembly of cowpea chlorotic mottle virus (61).

C. Transfer RNA and Cation Binding We have been studying the interactions of polyamines with tRNA because the well-defined structure of tRNA suggested that such interactions would be more amenable to dissection at the molecular level than the more complex reactions of protein or nucleic acid synthesis. Spermidine is known to activate denatured tRNA (62) and is useful in the formation of crystals of tRNA for X-ray analysis (63). Spermidine also activates the transfer of an amino acid from an isolated amino acid. AMP. aminoacyl-tRNA-synthetase complex to tRNA (64). Other studies (44, 45) indicate that the polyamine acts by binding to tRNA. The fluorescence of the “Y base” in the anticodon loop of yeast phenylalanine tRNA and amino-acid acceptance by this tRNA are increased in a parallel fashion by the addition of spermidine (65). This result was interpreted as being due to the formation of an active conformation of the nucleic acid by the addition of the polyamine. The activation of tRNA methyltransferases by polyamines has been described and significant differences have been found between methylations stimulated by spermidine and Mg2+ (66, 67). Wildenauer et al. (68) have described the effects of the diamine cadaverine and Mg2+ on the methylation of a pure species of tRNA. These workers found that the presence of each cation alone results in different methylation products and these, in turn, differ from that obtained in the presence of both cations. Again, the results argue for the formation of different configurations of tRNA by different cations and that these modulations of conformation effected by the polyamines, as well as Mg2+,do affect the metabolism of tRNA. tRNA isolated at low ionic strength from E . coli harvested during the phase of synthesis of both tRNA and spermidine contains 2 moles of spermidine per mole of nucleic acid (69). The tRNA of mouse fibroblasts (70) and plants (71)after similar isolation procedures contains a mixture of spermidine and spermine. We have studied the binding properties of spermidine with tRNA and, because our working hypothesis suggests a high affinity of spermidine for double-stranded helical regions of nucleic acid, we have compared these properties with those of ethidium bromide, which forms fluorescent complexes with such regions (72-74). We have tried to determine whether spermidine and ethidium might bind to the same region of tRNA. As shown in Fig. 3,there is at least a superficial structural similarity between ethidium and spermidine, both bases containing two primary amino groups, each of which

20

TED T. SAKAI AND SEYMOUR S. COHEN

ethidium

bromide

HN’

Cl$Ct$-

CH2-CH2- NH2

I

c

H

3

m

c

1

quiwcrine

C?& I

spermidine

FIG.3. Comparison the structures of ethidium bromide and quinacrine with spermidine.

is separated similarly from the quaternary nitrogen in the former and from the secondary nitrogen in the latter. Another molecule that interacts with DNA, quinacrine, is also structurally similar to spermidine, and it seems possible that such similarities to polyamines may correlate with the ability of these dyes to interact with nucleic acids. Our studies (75) show that tRNA contains 2 or 3 relatively tight binding sites and 14 or 15 relatively weak binding sites for spermidine. It appeared of interest to attempt to localize the sites of tRNA that bind spermidine tightly. The primary sequences of tRNA from various sources have been determined (76,77) and a common secondary “cloverleaf” structure may be drawn from these. We have examined this structure from the point of view of the Tsuboi hypothesis and looked for regions of the tRNA molecule that might be stabilized by the binding of polyamines. In every tRNA sequenced to date, the arm of tRNA containing dihydrouridine (hU) contains only 3 or 4 complementary base-pairs (Fig. 4). We have asked whether spermidine is used to stabilize this arm, since the 3 or 4 base-pairs could not be sufficient to confer three-dimensional stability on such a large polynucleotide sequence ( 78). From studies of the melting of the various base-paired regions as analyzed by proton magnetic resonance, it appears that this region of the structure is the first to be disrupted by increasing temperature (79, 80). Corey-Pauling-Koltun (CPK) models of this arm show that there are insufficient base-pairs to allow spermidine to stabilize it in the strict sense of the Tsuboi model (see Fig. 2). However, it seems possible

EFFECTS OF POLYAMINES ON

21

tRNA A-3!0H

P

hU loop

-

0

anticodon arm

variable arm

anticodon

loop

FIG.4. Cloverleaf drawing of tRNA showing the various regions of the nucleic acid.

that stabilization of this region might occur by interaction with another region of the tRNA molecule. Indeed, recent X-ray crystallographic studies (81, 82) show that the hU arm is not isolated from the rest of the tRNA molecule and that there are also several tertiary interactions not evident from the cloverleaf structures. These X-ray studies and CPK models based on them suggest that spermidine-mediated interactions might be possible between the hU arm and the amino-acid-acceptor arm.

111. Chemical and Physicochemical Studies A. Reactions of 4-Thiouridine in Escherichia coli tRNA Many species of tRNA from E . coli possess a 4-thiouridine (s'U) residue at position 8 in the primary sequence at the juncture of the hU and amino-acid-acceptor arms of the nucleic acid (76, 77), and the tyrosine tRNA possesses two s4U residues, at positions 7 and 8 (83). These tRNAs have a characteristic absorption peak centered near 335 nm, which may be monitored to follow reactions of s4U. Reagents used to modify s4U residues in tRNA include ethyleneimine ( 8 4 ) , N-ethylmaleimide ( 85), cyanogen bromide ( 8 6 ) , p-chloromercuribenzo-

22

TED T. SAKAI AND SEYMOUR S. O H E N

R = ribor. phoaphote in t R N A R f luoroacont

FIG. 5. Formation and reduction of the 4-thiouridine-cytidine photoproduct in Escherichla coli tRNA.

ate ( 87), and the spin-label 3- ( 2-bromoacetamido ) -2,2,5,5,-tetramethyIpyrrolidinyl-l-oxyl ( 88). The s4U residue is fluorescent when it is in a tRNA molecule; however, the free nucleotide shows no fluorescence properties (89, go), indicating that the structure of tRNA determines the fluorescence of this residue. Spermidine is half as effective as Mg" in enhancing the fluorescence of s4U in denatured Mg"-free tRNA, but the two cations can act cooperatively to generate maximum fluorescence (91) . These observations suggested that both cations are important in restructuring the conformation of tRNA, particularly near s4U. Irradiation at 335 nm of tRNAs containing s4U links s4U and a noncontiguous cytidine residue, which, in all of these tRNAs, is at position 13. The photoproduct, which spans the postulated binding site for spermidine, is now reducible to a highly fluorescent derivative by sodium borohydride (Fig. 5 ) (89, 90). Pochon and Cohen (91) showed that the rate of photoreaction is greater in the presence of spermidine than of Mg2+.They also showed that the fluorescence of the reduced photoproduct, lost after dialysis to remove cations, is more readily restored by spermidine than by Mg2+(91, 92). These preliminary studies suggested that spermidine affects the conformation of tRNA, and the data are consistent with the hypothesis that some molecules of spermidine are involved in the organization of the three-dimensional geometry of the limited helical region close to sW. To test the hypothesis further, we undertook the additional chemical and physicochemical studies (92, 93), which are discussed in this paper. While several papers are concerned with the effects of Mg2+on the properties and reactivity of the s4U region of tRNA (87, 94, 9 5 ) , little work has been done on effects of polyamines on tRNA, nor has much attempt been made to characterize these cation effects in terms of the specificity of binding sites. Pal et al. (87) suggested, on the basis of

EFFECTS OF POLYAMINES ON

tRNA

23

their studies with p-chloromercuribenzoate and tRNA, that Mg2' may bind at s4U; however, the majority of studies have been made with Mg2+ in great excess, making interpretation of cation effects difficult. We have attempted to study various cation effects at low ionic strength, and at low cation concentrations under conditions in which cation binding to tRNA would be maximal (Sections 111, B-F) as well as under equilibrium conditions (Section 111, G and H ) .

B. Cation Effects on pK of s4Uin tRNA The effects of MgZt and spermidine on the pK of s4U in tRNA were examined spectrophotometrically by determining the ratio of optical densities at 335 and 310 nm (the wavelengths of maximum absorption of the neutral and anionic species of s4U, respectively) as a function of pH. Calculations indicate that, in 0.05 M buffer and in the absence of added cations, the apparent pK of s4Uin tRNA is 9.45 (92). The apparent pK for s4U increases to 9.60 in the presence of 10 spermidine and 9.70 in the presence of 10 Mg2+.As the ratio of spermidine to tRNA is increased to 30, the apparent pK of s4U remains relatively unchanged at 9.63, whereas it increases significantly (to 9.85) in the presence of 30 Mg". Under the same conditions, the free nucleoside has a pK of 8.30, and this value is not altered by addition of spermidine. We can postulate several possible reasons for these observations: ( a ) the cations bind near s4U and inhibit ionization by a direct physical or electrostatic interaction; ( b ) the cations alter the composition of the medium such that ionization is suppressed; ( c ) the cations may cause a conformational change in the tRNA molecule, altering the local environment near s'U, changing the apparent pK, perhaps by "squeezing out" water from the s4U site. It is difficult to believe that the observed increase in pK is due to ( b ) , and it appears more likely that ( a ) or ( c ) or both approach the true explanations. This is primarily based on the fact that the pK of s4U is increased from 8.30 in the free nucleoside to 9.45 in the tRNA molecule; the binding of cations to tRNA further enhances the effect. Based on the study of model compounds, Iwamura (952) postulated that s4U exists in a relatively hydrophobic environment in tRNA. Since spermidine does not affect the pK of the free nucleoside, the effect of cations on tRNA may involve a tightening of the tRNA structure and removal of water from the interior of the molecule, creating a more hydrophobic area about s4U. (It should be mentioned that all the effects are given in terms of cations added per mole tRNA and do not necessarily reflect the amounts of cation actually bound to tRNA under the various conditions studied.)

24

TED T. SAKAI AND SEYMOUR S. COHEN

C. Thermal Denaturation Studies Hyperchromicity for mixed tRNA in the presence of varying amounts of Mgz+and spermidine was measured as a function of increasing temperature and "melting temperatures" ( t,) were taken to be the temperature at which 50% of the maximum hyperchromicity occurred. Typical curves monitored at 335 nm, the absorption maximum of s4U, are shown in Fig. 6; t , values at varying levels of cation per tRNA are shown in Fig. 7 and indicate that spermidine and Mgz+ have slightly different effects on them. Mg2+increases the t , (335 nm) from 48°C in the absence of added cation to 68°C at Mg2+/tRNAratios equal to or greater than 20. Spermidine increases t , to 73°C at 20 spermidine/ tRNA. By comparison, t , values measured at 260 and 280 nm appear to be independent of cation up to 20 moles per mole tRNA and lie in the range 56"-58OC. However, measurements at the latter wavelengths were carried out at a sixteenth of the concentration needed for measurements at 335 nm, and the observed differences may reflect the difference in cation binding at the different concentrations. The hyperchromicity observed at 335 nm in the thermal denaturation studies indicates that the structure about s4U is being melted out and provides another means of measuring the sensitivity of this region to added cations. Wong et al. (95b) have shown by nuclear magnetic resonance ( N M R ) studies that the s4U residue in E. coli tRNAfMetis

t,"C

FIG. 6. Typical thermal denaturation curves at 335 nm of Escherichia coli mixed tRNAs in 0.05 M cacodylate buffer in the presence of Mg" or spermidine (spd) at the given ratio to tRNA. [tRNA] = 10 PM. Absorbanc- readings were made on a Gilford Model 2000 spectrophotometer equipped with a Haake circulating water Mg"/ bath. Heating rate was 0.5"C per minute. 0-0, Cation, 0; CI-0, spermidine/tRNA = 20. tRNA = 20; A-A,

25 80

70

E

-

/

O

-

O

A

N

A

v)

0

m

50

FIG. 7. Dependence of t,,, of Escherichia coli mixed tRNAs at 335 nm as a function of moles of cation per mole of tRNA (see text, Section 111, C ) . 0-0, Spermidine; A-A, Mgl'.

involved in a tertiary base-pair. Our t , results indicate that the cations are helping to maintain the tertiary structure of tRNA. The greater effectiveness of spermidine in maintaining tertiary structure may reflect the ability of spermidine to tighten structure by bridging phosphate groups in separate parts of the molecule.

D. Chemical Reactions of 4-Thiouridine 1. BOROHYDRIDE REDUCXIONOF PHOTOLINKED tRNA As discussed earlier (Section 111, A), a photochemical reaction can occur that links s'U and C-13 and the product now spans the postulated spermidine-binding region. The effects of the various cations on the reduction of photolinked tRNA by NaBH, are shown in Fig. 8. Reactions were run at 25°C in 0.05 M carbonate buffer, pH 9.4, to assure reasonable stability of borohydride during the course of reaction. In the absence of added cation, the reaction appears first-order with respect to both tRNA and initial borohydride concentration, and the observed rate constant, kobs,is related to the second-order rate constant k, by the expression kobs = k,[tRNA][BH,-] where [tRNA] and [BH,-] are the concentrations of tRNA and NaBH,. Under the conditions of the reactions, k, has the value 6.21 M-l min-', which compares favorably with the value obtained by Favre and Yaniv (96) of 6 M-' min-'. The data show that for a borohydride concentration of 0.03 M and tRNA concentration of 0.1 pM, Mg" affects the reaction to the greatest extent, the effect being a 30%reduction in rate for Mg2+/tRNAratios greater than

TED T. SAKAI AND SEYMOUR S. WHEN

26 0.125

1

-k

0.075 0

I

I

5

10

1

15

20

1

25

30

Cationr/tRNA

FIG.8. Effects of cations on the borohydride reduction of photolinked tRNA in 0.05 M carbonate buffer, pH 9.4, 25°C. [BH4-] = 30 mM; [tRNA] = 28 pM. 0-0, Putrescine; - - 0 , spermine; A-A, sperrnidine; 0-0, M e .

..

1CL15. Spermidine shows a similar effect although only to the extent of giving a 20%decrease in rate. Spermine shows a slightly smaller effect, and putrescine causes very little decrease in rate (92).

2. IODOACETAMIDE REACTIONS The reactions of tRNA and iodoacetamide at 25"C, as measured by loss of s4U absorbance at 335 nm at pH 9.4, are very similar to the borohydride reactions. The reaction is first-order with respect to both tRNA and alkylating agent for iodoacetamide concentrations in the range 0-5.0 mM, and the second-order rate constant for the reaction is 10.8 M-' min-'. At an iodoacetamide concentration of 4 mM and a tRNA concentration of 28 pM, the inhibitory effects of cations are similar to those observed in the reduction of photolinked tRNA; i.e., Mgz+has the greatest effect, followed by spermidine, spermine and putrescine. Again, all effects appear maximal beyond 10-15 cations per tRNA. The rates of alkylation were decreased by Mg2+,spermidine and putrescine by 50, 45 and 25%,respectively. The greater effect of cations in these cases than in the reduction reactions may reflect the larger size of iodoacetamide relative to borohydride (92). Plots of reaction rates vs iodoacetamide concentration at varying levels of MgZ+and spermidine are shown in Fig. 9. Cations decrease the second-order rate constant. The change in slope without a change in intercept indicates that the binding of the cations is competitive with the iodoacetamide reaction and that the binding of the cations is reversi-

EFFECTS OF POLYAMINES ON

tRNA

27

[ iodoace tami de], rn M

FIG. 9. Dependence of observed rate constant, k o b s , on iodoacetamide concentration at varying concentrations of Mg" and spermidine (spd) in 0.05 M carbonate buffer, pH 9.4, 25°C. [Iodoacetamide] = 4 mM; [tRNA] = 28 pM. x-x, no addition; -0, spd, 5; A-A, M e ,5; .-.a, spd, 10; A-A, Mg*, 10.

ble. This apparent competition between a reagent specific for s4U and the cations shows that the cations indeed alter the structure of tRNA about s'U and suggest strongly that the binding might be near s4U. 3. CYANOCEN BROMIDEREACI~ONS Cyanogen bromide reacts specifically with s4U residues in E . coli tRNA (86) to give the 4-thiocyanatouridine derivative, which in turn hydrolyzes readily to uridine in acid or base. The reaction is first-order with respect to both tRNA and BrCN for BrCN concentrations below approximately 0.50 mM at pH 7.4. The vaIue of the second-order rate constant under the conditions of the reaction is 3.79 x lo3 M-' min-l. The results of the reactions of tRNA (21 p M ) and cyanogen bromide (0.25 mM), summarized in Fig. 10, differ slightly from the reactions with iodoacetamide and borohydride. Spermidine and spermine exhibit inhibitory effects similar to those of Mg2+until a ratio of 10 moles/mole tRNA is reached. Beyond this ratio, polyamine effects appear to level off, suggesting that there is limited binding of the polyamines. Putrescine again shows the smallest effect on the reaction rate. The studies with borohydride and iodoacetamide indicate that polyamines and Mg" make the s'U region of tRNA less reactive. The choice

TED T. SAKAI AND SEYMOUR S. WHEN

28

0.5

-.-'E .t

0.3 0.2 0-0

0.1

I

I

I

I

I

I

c a t ion s/ t R NA FIG.10. Cation effects on reaction of s'U in tRNA with BrCN in 0.05 M cacodylate buffer, pH 7.4, 25°C [BrCN] = 0.25 mM; [tRNA] = 21 FM. 0-0, Putrescine; A-A, spermidine; a-H, spermine; 0 4 , Me.

of an alkaline pH seems somewhat questionable in that the polyamines probably are not fully charged at pH 9.4; also, the association constants for tRNA and polyamines are different from the values at physiological pH. Nonetheless, polyamines show effects of a magnitude similar to Mg". The reactions of cyanogen bromide also show that the cations are altering the reactivity of s4U to the reagent, presumably by reorganization of the tRNA structure.

E. Reactions of Co3+-tRNA Treatment of tRNA with Co2+in the presence of oxygen can result in a very tight linkage of CoS+to tRNA (97, 98). We have prepared CoS+-tRNAat various initial ratios of Coz+to tRNA as described by Danchin (97), using aeration at 4°C for 30 minutes at pH 9 followed by exhaustive dialysis. The effect of this stably bound CoS+on the reactivity of s4U in tRNA with iodoacetamide and cyanogen bromide was compared with that caused by equivalent amounts of Co2+.It can be seen in Table I that reactions of CoS+-tRNAwith iodoacetamide are inhibited as the amount of CoS+per tRNA increases. On the other hand, Co2+is not as inhibitory. Similar results are found if the effects of CoS+ and Co2+on the reaction of tRNA with cyanogen bromide are compared. The initial site of attachment of these cations is believed to be in the anticodon arm. However, as these studies show, the conformation about s4U must also be affected, suggesting that CoS+is also bound to regions other than the anticodon arm or that its site of attachment is near s4U in the three-dimensional structure of tRNA.

E F F E ~ SOF POLYAMINES ON

EFFECTSOF

tRNA

29

TABLE I PRESENCE OF Co3+ A N D Co*+ I N tRNA ON THE REACTIVITY s4U WITH IODOACETAMIDE AND CYANOGEN BROMIDE'

THE

OF

k (min-1)

x

Coa+/tRNA

Iodoacetamide Cyanogen bromide

10'

CoP+/tRNA

0

5

10

15

5

15

6.30 1.15

5.32 0.60

4.33 0.46

3.30 0.46

5.77 0.77

5.32 0.77

Reactions were run as described in the text (Section 111, D, parts 2 and 3). [iodoacetamide] = 5 mM; [cyanogen bromide] = 0.25 mM; [tRNA] = 33 pM.

F. Effects of Cations on p-Chloromercuribenzoate-labeled tRNA Pal et al. (87) reported that MgZ+can displace the p-mercuribenzoate group from s4U residues in tRNA, but not from the labeled free nucleoside (Fig. 11).These observations were interpreted to mean that there is a binding site for Mg2+in tRNA near s4U. We have compared polyamines and Mg2+ for their ability to release p-mercuribenzoate from s'U in tRNA. Using p-CIHgBzO--labeled tRNA (33 pM) (87) and 2 mM Mgz+, regeneration of the s4U absorbance at 335 nm was observed as reported. Nevertheless, 2 mM spermidine failed to regenerate the absorbance. If Mgz+was added to the spermidine-containing solution, the s4U peak was regenerated. Spermine or putrescine ( 2 mM) also had no effect. Ca2+( 2 mM) is about half as effective as MgZ+,while Ba2+,Fez+,Zn2+, Co2+ and AI3+ appear to have little or no effect and FeS+, CuZ+and MnZ+at 2 mM cause precipitation of the tRNA. At low Mg"-to-tRNA ratios (approximately 1-2 MgZ+/tRNA),no release of p-mercuribenzoate is observed. This indicates that the effect of Mg2+is not catalytic. These results are difficult to interpret. The observation (87) that Mg" effects the release of p-mercuribenzoate from s4U in tRNA but not from the labeled nucleoside suggests an effect on the tRNA structure,

FIG. 11. Reaction of p-chloromercuribenzoate with s'U in tRNA.

TED T. S A K A I AND SEYMOUR

30

S. COHEN

not a metal ion-sulfur interaction. Other studies (98u-98~) show that the reactivity of various Hg( 11) compounds with s4U in tRNA increases with decreasing size of the labeling compound and that small Hg( 11) compounds react with tRNA even in the presence of Mg2+,indicating that the structure of tRNA is determining the reactivity of s4U with p-ClHgBzO- as well as the reverse reaction. However, the inability of polyamines and metal ions closely related to Mg2+to remove the label suggests that tRNA conformation alone is not involved. The apparent specificity for Mgz+(and Ca2+to some extent) may reflect the importance of ion size or orbital geometry.

G. Spin-label Studies In order to explore the structure of tRNA about the postulated binding site for spermidine, we began studies on tRNA labeled at s'U residues with 3- ( 2-bromoacetamido) -2,2,5,5-tetramethylpyrrolidinyl-l-oxyl( Fig. 12). This spin-labeled tRNA has an electron paramagnetic resonance (EPR) spectrum reflecting the conformation of the s4U region of tRNA. Hara et al. (88) have shown that the reaction with s4U is *specificin tRNAs bearing this modified base and that the other major and minor components of tRNA do not react with the spin-labeling reagent under the conditions of the reaction. Typical spectra are shown in Fig. 13 for spin-labeled mixed tRNAs. The EPR spectra of the spin-labeled tRNAs in the absence of added cations are characteristic of partially immobilized spins; nevertheless, the absorption lines become broader and signal heights decrease as cation/ tRNA ratios are increased, indicating further immobilization of the spin. As shown by Hoffman et al. (99) and Kabat et al. ( 1 0 0 ) in other studies on spin-labeled tRNA, rotational correlation times T can be calculated from such spectra using the equations given by Stone et al. (101).The T values for spin-labeled mixed tRNAs and tRNAfMetare shown in Figs. 14 and 15. The data obtained from the EPR spectra are rotational correlation times T , which are a measure of the time it takes for a nitroxide radical to rotate an average of 40". Thus, an increase in the value of T indicates a decrease in the rotational freedom of the nitroxide radical or an increase in immobilization. While the absolute accuracy in the determination

FIG.12. Reaction of 3- ( 2-bromoacetamido)-2,2,5,5-tetramethyIpyrrolidinyl-l-oxyl with s'U.

EFFECTS OF POLYAMINES ON

tRNA

31

FIG. 13. Typical electron paramagnetic resonance spectra of spin-labeled mixed tRNAs in 0.05 M TrisCl buffer, pH 7.0. No additions (-); 25 moles of spermidine/ mole of tRNA ( - - ) ; spin-labeled 4-thiouridine ( * ). Spectra were determined using a Varian E-4 spectrometer operating at 9.1 GHz with 10 mW power and a field setting at 3242 G . [tRNA] = 30 aM.

. .

of T values may be questioned, such values, which have been used by several groups in analyzing the tertiary structure of tRNA (99,loo), show relative effects of the cations clearly and provide at least a crude quantitation of cation effects on spin-labeled tRNA. In mixed tRNAs, putrescine confers only a small effect on the rotational freedom of the radical in tRNA, and Mgz+ has only a slightly greater effect. Spermine gives a greater initial immobilization of the spin-label than spermidine, but spermidine shows a greater finaI effect. Spin-labeled 4-thiouridine ( Fig. 12) was prepared (88) and isolated by preparative layer chromatography on Avicel cellulose plates. The material has an R, of 0.85 in the solvent system isopropanol/methyl ethyI ketone/H20/NH,0H (4:3:2: 1 ) and has an ultraviolet spectrum very similar to that of the spin-labeled nucleotide ( 88). Its EPR spectrum is characteristic of a highly mobile spin, and the addition of a 50-fold molar excess of spermidine or Mg2+ has no effect on it. On the other hand, a spin-Iabel in tRNA without added cations shows a spectrum characteristic of a significantly immobilized spin with a T value of approximately lo-'" second. Spermidine, when added to tRNA at ratios greater than about 15,

TED T. SAKAI AND SEYMOUR S. WHEN

32

0.8 0

I

5

1

10

I I5

I

20

I

25

Cations/t R N A

FIG. 14. Effects of cations on the rotational correlation time 7 of spin-labeled Escherichia coli mixed tRNAs, calculated from spectra shown in Fig. 13. Spermidine; 0-0, spermine; A-A, M e ; 0-0, putrescine.

m,

appears to restrict rotation of the spin-label nearly 2-fold relative to tRNA without spermidine, 7 doubling from 8.5 x lo-" second to nearly 17 X lo-" second. With tRNAfMet,similar general effects are seen; however, the overall increase in 7 is only about 25% for spermine and spermidine with 7 increasing from 1.1 x 10-lo second in the absence of added cation to 1.4 x 10-lo second at about 20 moles of spermidine (or spermine) per mole of tRNA. Again Mg2+and putrescine show minimal effects. In the absence of added cation, spin-labeled tRNAValgives a T value of about 7 x second, indicating that each species of spin-labeled tRNA may behave differently and that although all s4U residues occur in the same position in the primary sequence of tRNAs bearing s4U, the tertiary structures may be significantly different. tRNAGLu was found to react with the spin-label under more rigorous conditions (24-36 hours under the alkylating conditions described above). This species of tRNA does not contain s4U but does contain an

EFFECTS OF POLYAMINES ON

33

tRNA

I

I

I

I

I

5

10

15

20

25

Cations/tRNA

FIG.15. Effects of cations on the rotational correlation time of spin-labeled Escherichia coli tRNA,Met;[tRNA] = 15 p M . 0-0, Spermidine; 0-0, sperMf; 0-0, putrescine. mine; A-A,

s2U derivative, 5-methylaminomethyl-2-thiouridine( 102), in the anticodon loop. Although the nature of the labeled nucleoside in tRNAG1" has not been established, it is possible that the sulfur atom of this rare nucleoside may have reacted with the spin-label reagent. The T value for tRNAQtlIin the absence of added cation is 1.1x 10-lo second, and the value increases somewhat in the presence of spermidine and Mg2+ (Table I I ) , although the increases are markedly smaller than those obtained with tRNAs labeled at s4Uresidues. Spermidine and spermine appear to close the structure of the s4U-containing tRNA molecule so that the nitroxide radical is more hindered in its rotation relative to the tRNA in the absence of added cation. Although the polyamines have definite effects on the spin-labeled tRNA, the assumption has been made that the conformations of normal and spin-labeled tRNA are similar. It i,s postulated, but not yet proved, that results from the spin-labeled tRNA may be extended to normal tRNA. That the two do have similar conformations is suggested by the retention

34

TED T. SAKAI AND SEYMOUR S. COHEN

TABLE I1 ROTATIONAL CORRELATION TIMES( T ) FOR SPIN-LABELED tRNAQ'u Cations/tRNA 0 5 Mg'+ 10 Mg'+ 20 Mg*+ 5 Spermidine 10 Spermidine 20 Spermidine a

T

(sec)

x

1Olo

1.11 1.12 1.16 1.20 1.17 1.19 1.26

Spectra were determined in 0.05 M TrisC1, pH 7.0 at 25°C (See t,ext, Section

111, G ) .

of activities such as amino-acid acceptance by the spin-labeled RN A (88). Spermidine appears to be most effective in causing restriction of the spin-label in mixed tRNAs; Mg2+ is only half as effective at the same levels per tRNA. Spermine is most effective at lower levels (less than 5-8 per tRNA) but is not quite as effective as spermidine at higher levels; putrescine has very little effect. The differences might be attributed to the ability of spermidine and spermine to bridge separate parts of the tRNA molecule to help close the structure. Similar results are seen for spin-labeled tRNAfMet,although the effects of spermidine and spermine differ slightly from those seen with mixed tRNAs. That tRNAoIumay be labeled with the nitroxide radical indicates that EPR studies may be a more versatile tool than originally thought for studying tRNA structure. The base( s ) alkylated remain to be determined although a likely candidate is the 2-thiouridine derivative, 5-methylaminomethyl-2-thiouridine. The observation that the rotation of this label in tRNAQIUis not as restricted by polyamines, as is the rotation of a label attached to s4U, suggests the relative lack of structure in the reion (presumably the anticodon loop) to which the label is attached. Singhal (102a) has shown that 5 ( MeNHMe)2Srd, as well as a C residue, in the anticodon loop of tRNAQIUare very susceptible to various reagents, indicating that this loop is exposed or lacks structure.

H. Binding Studies As described earlier (Section 11, C), tRNA from bacteria containing 2 moles of spermidine per mole of tRNA can be isolated. When an excess of spermidine is mixed with Mg2+-freetRNA and the mixture is fractionated on Sephadex G-100 (75), complexes of spermidine and

EFFECTS OF POLYAMINES ON

tRNA

35

tRNA may be isolated containing 10-11 moles of spermidine per mole of tRNA. Similarly, tRNA can bind 3 fluorescent ethidium molecules, and complexes with tRNA can be formed containing 10-11 moles of dye per mole of tRNA. Isoaccepting species behave similarly to mixtures. The limited number of spermidine and ethidium molecules bound per tRNA are consistent with the idea that these ligands are binding to specific regions of the tRNA molecule and perhaps holding together regions of the molecule that would otherwise become unfolded. In the isolated complexes, either cation will displace the other; i.e., addition of spermidine to fluorescent ethidium complexes discharges the fluorescence, whereas ethidium can attain full fluorescence in spermidine complexes. Although the first ethidium molecule bound to spermidine-free tRNA is entirely at a fluorescent site, the presence of a single spermidine in a tRNA complex significantly inhibits the fluorescence developed by the addition of a single molecule of ethidium. Thus, the molecules appear to compete for the initial binding site ( 75). Equilibrium binding studies were done to test these hypotheses further. The studies were done by equilibrium dialysis or by spectrophotometric titration (75, 103, 104). The data were anaryzed by Scatchard plots (105, 106), and the numbers obtained are the total number of binding sites (c), the number of sites in each class of binding sites (ni) and the association constant KR(M-*) for each class of binding sites. The values of D and ni are given as moles of ligand bound per mole of nucleic acid-phosphate and also as moles of ligand bound per mole of tRNA, assuming 80 nucleotides for an average tRNA molecule. 1. SPERMIDINE BINDINGTO tRNA Analysis of the binding of spermidine to yeast mixed tRNA (in 0.05 M potassium cacodylate, 0.05 M KCl, pH 7.0) shows that there are two classes of binding sites for spermidine and the equivalent of 1 6 1 7 moles of spermidine per mole of tRNA ( n = 0.221). The higher affinity site, corresponding to 2-3 molecules of spermidine per tRNA molecule ( n = 0.031), has a K , = 6.6 x lo' M-'. The other class ( n = 0.190) shows a K, = 7.6 X lo3 M-'. Very similar results are obtained with yeast phenylalanine tRNA. Again, there are two classes of binding sites; the higher d n i t y type of site has a K, = 6.1 x 104 M-I and binds the equivalent of about 3 moles of spermidine per mole of tRNA ( n = 0.041). The other type of site, which binds about 14-15 moles per m,ole of tRNA ( n = 0.184), has a K., = 5.4 X los M-I. A comparison of the values is given in Table 111. In 9 mM Mg2+,both mixed tRNA and phenylalanine tRNA show a single class of binding sites for spermidine. There appear to be about

36

TED T. SAKAI AND SEYMOUR S. COHEN

TABLE I11 COMPARISON OF THE BINDING PARAMETERS FOR THE BINDING OF SPERMIDINE TO YEASTMIXEDtRNA A N D PHENYLALANINE tRNA I N THE PRESENCE A N D I N THE ABSENCE OF M G ~ + "

Yeast mixed tRNA Phenylalanine t RNA Yeast mixed tRNA 9 mM Mg*+ Phenylalanine tRNA 9 mM Mgzf

+

+

0.221 0.225

0.031 0.041

-

-

-

-

6.6 6.1

0.190 0.184

0.76 0.54

-

0.190 0.170

0.060 0.059

-

a Parameters were determined by equilibrium dialysis (76) in 0.05 M K cacodylate, 0.05 M KCI, pH 7.0.

14-15 molecules of spermidine bound in mixed tRNA ( n = 0.190) with an average K, = 6.0 X lo2 M-l. In phenylalanine tRNA there are about 13 moles of spermidine bound per mole of tRNA ( n = 0.170) with an average K, = 5.9 )( lo2 M-' (Table 111). 2. ETHIDIUM BROMIDEBINDING AND COMPETITION BY POLYAMINES AND

Results similar to those found with spermidine binding are obtained for the binding of ethidium bromide to yeast mixed tRNA. In the absence of spermidine, there are approximately 16 total sites for ethidium binding to tRNA ( n = 0.206), approximately 3 of which ( n = 0.037) have a K, = 4.2 x lo5 M-l. With increasing spermidine concentrations, the affinity constants for both classes of sites decreases. The number of tight binding sites increases, as apparently does the total number of sites. However, the data obtained in the presence of 0.24 mM spermidine may also be interpreted as being due to the presence of a single class of binding sites, as found with the binding of spermidine to tRNA in the presence of Mg2+.The inability to obtain data at low values makes the plot difficult to interpret more clearly, but if a single class of sites is assumed, there are 17-18 molecules of ethidium bound ( n = 0.220) with a K, = 1.7 )( lo4 M-l. Similar results are obtained if the binding of ethidium is determined by spectrophotometric titration. A comparison of the data obtained by the two methods is shown in Table IV. Fluorometric analysis of ethidium binding in the presence of spermidine shows that there are 3-4 sites per tRNA ( n = 0.04-0.05) and these sites correspond fairly well to the tight binding sites obtained by equilib-

EFFECTS OF POLYAMINES ON

37

tRNA

TABLE IV COMPARISON OF T H E BINDING PARAMETERS FOR THE BINDING OF ETHIDIUM BROMIDE TO YEASTMIXEDtRNA AS DETERMINED BY EQUILIBRIUM DIALYSIS A N D SPECTROPHOTOMETRIC TITRATIONO

Equilibrium dialysis Ethidium bromide plus 0.06 mM 0.12 mM 0.24 mM spermidine

0.037 0.206 0.020 0.220 0.030 0.235 0.250 0.021 Assuming single class

Spectrophotometric litration 0.023 Ethidium bromide 0.237 0.235 0.032 plus 0.06 mM 0.021 0.221 0.12 mM 0.022 0.24 mM 0.250 spermidine Assuming single class

4.16 3.81 1.82 1.15

0.169 0.200 0.205 0.229 0.220

0.426 0.274 0.177 0.122 0.170

9.03 2.38 2.08 0.89

0.214 0.203 0.201 0.228 0.190

0.403 0.250 0.242 0.138 0.140

The number of binding sites is given as values of n in terms of moles of ethidium per mole of tltNA-phosphate. The values subscripted 1 are the values for the tight binding sites, those subscripted 2 are for the looser sites. The conditions are the same as described in Table 111.

rium dialysis. Addition of increasing amounts of spermidine (to 0.24 mM) has little effect on the number of binding sites for fluorescent ethidium, although the affinity of ethidium is decreased. Putrescine (0.12 m M ) has little effect on ethidium binding to either tight or loose class of sites. Spermine and Mg2+ appear to eliminate the loose class of ethidium binding sites, and the number of fluorescent binding sites decreases to 1 or 2 per tRNA in the presence of either of these cations. In general, the results are consistent with the data obtained under nonequilibrium conditions and support the initial hypotheses ( Section 11, C ) (75, 92) : ( a ) there are fairly specific binding sites for spermidine in tRNA; ( b ) there is a second, looser class of sites for binding spermidine; ( c ) ethidium may be formally analogous to spermidine and may occupy the same or similar sites as does spermidine; and ( d ) spermidine and ethidium appear to compete for at least the initial binding site. The data, when taken with other data, are consistent with a role for spermidine in which it plays a particular role in organizing tRNA structure. Also the hypothesis that ethidium is binding in a manner similar to spermidine is borne out, suggesting that ethidium may indeed be thought of as an analog of spermidine.

38

TED T. SAKAI AND SEYMOUR S. COHEN

It is interesting to note that the 16-17 moles of spermidine bound per mole of tRNA can neutralize approximately 50 phosphate groups of the tRNA backbone. The latter number correlates well with the number of nucleotides involved in complementary base-pairing in an average tRNA molecule, suggesting the possibility that these cations are binding to phosphate groups in double-stranded regions of the nucleic acid. Similarly, the number of total ethidium binding sites may be related to the number of base-pairs in helical regions; however, it appears that the intercalative (fluorescent) binding of ethidium is limited to 2 or 3 sites per tRNA, perhaps by the folding of the tRNA molecule.

IV. Conclusions Many of the recent data on polyamines and tRNA are consistent with the hypothesis that polyamines have specific effects on tRNA conformation. Our chemical and physicochemical studies of the effects of polyamines on tRNA suggest that the polyamines are binding in the vicinity of the dihydrouridine arm and perhaps stabilizing this region of tRNA. Takeda and Ohnishi (107) have recently reported that tRNA in aitm binds spermine and/or spermidine in preference to Mg2+,and that this differential binding may reflect what is occurring within a cell. However, based on the observations made in this study that spermidine and Mg" have similar effects in most systems, a cooperative role for the two cations in the activation or alteration of tRNA structure and conformation should be considered. Such a cooperative type of binding, however, does not rule out the possibility that there are unique sites for the cations to bind to tRNA. That such sites exist for the binding of ions has been suggested and is shown by a number of recent papers (see below). Fluorescence studies by Kayne and Cohn (108) and Wolfson and Kearns (109) indicate that Tb(II1) and Eu(II1) bind near s'U in E . coli tRNA and that the fluorescence of these ions is due to energy transfer from s4U to the ions. Jones and Kearns (110) in a PMR study showed that 4-5 Eu(II1) are bound tightly to tRNA and that the binding of the first ion shifts resonances in the hU and amino-acid-acceptor arms. They suggested that the ion binds between the two arms, at a site close to a s4U residue, if one is present, These reports suggest that the hU arm may in fact be a general binding site for many cations. Several reports by Schimmel and co-workers on the binding of Mg2' ( I l l ) , Mn2+(112)and polyamines (113)to tRNA point to a mechanism of binding that is cooperative for the first 4 or 5 ions bound. This cooperative binding is suggested by the authors to reflect the formation of tertiary structure in tRNA. There are other reports supporting such a

EFFECTS

OF POLYAMINES ON

tRNA

39

cooperative binding of ligands to tRNA (114);on the other hand, we (75) and others (115)find no indication of cooperativity. These differing results probably reflect the variation in conditions-ionic strength, buffers, pH etc.-used in the various studies, and they show the need for some standardization in such studies. Our findings (75 and Section 111, H ) that ethidium bromide has binding characteristics very similar to those of spermidine, and the suggestion by Schreier and Schimmel (113)that Mg2+,Mn2+and polyamines behave similarly indicate that ethidium may be a very useful tool not only for studying nucleic acid structure, but also for studying the interaction of many cations with nucleic acids. A recent NMR study by Jones and Kearns (116) showed that there is indeed a unique binding site for ethidium bromide in yeast tRNAPhelocated at the base of the amino acid acceptor arm, near the postulated site for spermidine (and ethidium) binding. Thus, it appears that our postulates, which were based on relatively simple concepts, may be correct.

ACKNOWLEDGMENTS This research was supported by a fellowship from the Damon Runyon Memorial Fund for Cancer Research (T.T.S.) and U.S. Public Health Service Grant No. AI10424. We would like to thank Dr. Kenneth E. Rubenstein of SYVA Research Institute, Palo Alto, California for help in determining the electron paramagnetic resonance spectra. We are grateful to Dr. A. D. Kelmers of the Oak Ridge National Laboratory, Oak Ridge, Tennessee for gifts of E . coli tRNAflIet, tRNA"" and tRNA""'.

REFERENCES 1. H. W. Dudley, 0. Rosenheim and W. W. Starling, BJ 20, 1082 (1926). 2. F. Wrede, H. Fanselow and E. Strack, Z. Prakt. Chem. 161, 66 (1926). 3. A. van Leeuwenhoek, Phil. Trans. Roy SOC. 12, 1040 (1678). 4. L. N. Vauquelin, Ann. Chim. 9, 64 ( 1791). 5. S. S. Cohen, "Introduction to the Polyamines." Prentice-Hall, Englewood Cliffs, New Jersey, 1971. 6. A. D. Hershey, Virology 4, 237 ( 1957). 7. B. N. Ames and D. T. Duhin, JBC 235,769 (1960). 8. S . V. Beer and T. Kosuge, Virology 40, 930 (1970). 9. I. Fukuma and S. S. Cohen, J . ViroZ. 16, 222 (1975). 10. I. Fukuma and S. S. Cohen, Proceedings of the International Symposium on the Biochemistry of Adenosylmethionine, in press. 1 1 . I. Fukuma and S. S. Cohen, 1. Virol. 12, 1259 ( 1973). 12. E. J, Herhst and E. E. Snell, JBC 176,989 (1948). 13. H. Kihara and E. E. Snell, PNAS 43, 867 (1957). 14. W. H. Martin, Jr., M. J. Pelczar, Jr, and P. A. Hansen, Science 116, 483 (1952). 15. A. Trauh, J. Mager and N. Grossowicz, J . Bact. 70, 60 ( 1955). 16. P. H. A. Sneath, Nature (London) 175,818 (1955).

40

TED T. SAKAI AND SEYMOUR S. COHEN

17. R. G. Ham, BBRC 14, 34 (1964). 18. W. G. Dykstra and E. J. Herbst, Science 149,428 (1965). 19. A. Raina, J. Janne and M. Siimes, BBA 123, 197 (1966). 20. A. Raina, Acta Physiol. Scund. 80,7 ( 1963). 21. A. S. Dion and E. J. Herbst, PNAS 58,2367 (1967). 22. D. H. Russell, S. H. Snyder and V. J. Medina, Life Sci. 8, 1247 ( 1969). 23. D. H. Russell, Nature N B 253, 144 (1971). 24. A. S. Dion and S. S. Cohen, PNAS 69,213 ( 1972). 25. D. V. Young and P. R. Srinivasan, J . Bad. 117,1280 (1974). 26. A. Pastuszyn and R. B. Loftfield, FP 34, 501 ( 1975). 27. A. Raina and S. S. Cohen, PNAS 55, 1587 (1966). 28. C. Barrios and G. Guidice, Exp. Cell Res. 50, 671 ( 1968). 29. S. S. Cohen, N. Hoffner, M. Jansen, M. Moore and A. Raina, PNAS 57, 721 (1967). 30. C. F. Fox and S. B. Weiss, JBC 239,175 ( 1964). 31. E. Fuchs, R. L. Millette, W. Zillig and G. Walter, E J B 3, 183 (1967). 32. W. W. Fredericks, U. Maitra and J. Hunvitz, JBC 244, 413 (1969). 33. P. Herrlich, E. Scherzinger and M. Schweiger, Mol. Gen. Genet. 114,31 ( 1971). 34. J. R. Lillehaus and K. Kleppe, Bchem 14, 1225 (1975). 35. W. Wickner, D. Brutlag, R. Schekman and A. Kornberg, PNAS 69, 965 (1972). 36. R. Schekman, A. Weiner and A. Komberg, Science 186, 987 ( 1974). 37. J.-F. Chiu and S. C. ,Sung, BBA 281, 535 (1972). 38. R. L. Weiss and D. R. Morris, B c h m 12,435 (1973). 39. B. W. Kimes and D. R. Morris, Bchem 12,442 ( 1973). 40. K. Hosokawa, Y. Kiho and L. Miigata, JBC 248,4135 (1973). 41. L.-M. Changchien and J. N. Aronson, J . Bact. 103,734 ( 1970). 42. M. Garcia-Patrone, N. S. Gonzalez and I. D. Algranati, BBA 299, 452 ( 1973). 43. Y. Takeda, BBA 182,258 ( 1969). 44. Y. Takeda and K. Igarashi, BBRC 37,917 (1969). 45. K. Igarashi and Y . Takeda, BBA 213,240 ( 1970). 46. K. Matsuzaki and Y. Takeda, BBA 308, 339 (1973). 47. A. Hershko, S . Amoz and J. Mager, BBRC 5,46 ( 1961). 48. R. G. Martin and B. N. Ames, PNAS 48,271 (1962). 49. M. Tsuboi, Bull. Chem. SOC. ]up. 37, 1514 (1964). 50. A. M. Liquori, L. Constantino, V. Crescenzi, V. Elia, E. Giglio, R. Puliti, M. deSantis Savino and V. Vitagliano, J M B 24, 113 (1967). 51. P. HoraEbk and I. J. Cemohorsky, BBRC 32,956 (1968). 52. L. Stevens, BJ 103, 811 (1957). 53. H. Tabor, Bchem 1, 496 (1962). 54. W. Szer, BBRC 22, 559 ( 1960). 55. W. Szer, Acta Biochim. Polon. 10,251 ( 1966). 56. J. C . Thrierr, V. Deubel and M. Leng, Biochimie 54, 1115 (1972). 57. J. W. Quigley and S. S. Cohen, JBC 244, 2450 ( 1969). 58. P. H. Ray and T. D. Brock, J . Gen. Microbiol. 66, 133 ( 1971). 59. C. C . Levy, W. E. Mitch and M. Schmulker in “Polyamines in Normal and Neoplastic Growth” (D. H. Russell, ed.), p. 91. Raven Press, New York, 1973. 60. C. C . Levy, P. Hieter and S. LeGendre, JBC 249,6762 ( 1974). 61. E. Hiebert and J. B. Bancroft, Virology 39, 296 (1960). 62. J. R. Fresco, A. Adams, R. Ascione, D. Hemley and T. Lindahl, CSHSQB 31, 527 ( 1966).

EFFECTS

OF POLYAMINES ON

tRNA

41

63. A. Hampel and R. M. Bock, Bchem 2, 1873 (1970). 64. H. G. Bluestein, C. C. Allende, J. E. Allende and G. P. Cantoni, JBC 243, 4693 (1968). 65. B. Robison and T. P. Zimmerman, JBC 246, 110 (1971). 66. P. S. Leboy, FEES Lett. 16,117 (1971). 67. D. V. Young and P. R. Srinivasan, BBA 238,447 ( 1971). 68. D. Wildenauer, H. Gross and D. Riesner, NARes. 1, 1165 (1974). 69. S. S. Cohen, S. Morgan and E. Streibel, PNAS 64, 669 (1969). 70. S. S. Cohen, Ann. N.Y. Acad. Sci. 171, 869 (1970). 71. N. Bagni, G. Stabellini and D. S. Fracassini, Physiol. Plant. 29, 218 (1972). 72. J. B. LePecq, P. Yot and C. Paoletti, C. R. Acad. Sci. 259, 1786 (1964). 73. J. B. LePecq and C. Paoletti, C. R. Acad. Sci. 260,7033 (1965). 74. M. J. Waring, BBA 114,234 (1966). 75. T. T. Sakai, R. Torget, J. I., C. E. Freda and S. S. Cohen, NARes. 2, 1005 (1975). 76. M. Staehelin, Erperientia 27, 1 (1971). 77. S. Nishimura in “Biochemistry of Nucleic Acids” (K. Burton, ed.), p. 289. University Park Press, Baltimore, Maryland, 1974. 78. I. Tinoco, 0. C. Uhlenbeck and M. D. Levine, Nature (London) 230, 362 (1971). 79. C. Hilbers, R. Shulman and S. Kim, BBRC 55,953 (1973). 80. C. Hilbers and R. Shulman, PNAS 71,3239 (1974). 81. S. H. Kim, F. Suddath, G. Quigley, A. McPherson, J. Sussman, A. Wang, N. Seeman, and A. Rich, Science 185,435 ( 1974). 82. A. Klug, J. Robertus, J. Ladner, R. Brown and J. Finch, PNAS 71,3711 (1974). 83. M. N. Lipsett and B. P. Doctor, JBC 242,4072 (1967). 84. B. Reid, BBRC 33, 627 (1968). 85. J. Carbon and H. David, Bchern 7,3851 ( 1968). 86. M. Saneyoshi and S. Nishimura, BBA 204,389 (1970). 87. B. C. Pal, L. R. Shugart, K. R. Isham and M. P. Stulberg, ABB 150, 86 (1972). 88. H. Hara, T. Horiuchi, M. Saneyoshi and S. Nishimura, BBRC 38, 305 (1970). 89. A. Favre, M. Yaniv and A. M. Michelson, BBRC 37, 266 (1969). 90. A. Favre, B. Roques and J. L. Fourrey, FEES Lett. 24, 209 (1972). 91. F. Pochon and S. S. Cohen, BBRC 47,720 ( 1972). 92. T. T. Sakai and S. S. Cohen in “Polyamines in Normal and Neoplastic Growth” ( D. H. Russell, ed.), p. 41. Raven Press, New York, 1973. 93. T. T. Sakai and S. S. Cohen, FP 32,546 (1973). 94. T. Seno, M. Kobayashi and S. Nishimura, BBA 174,71 ( 1969). 95. E. Ziff and J. R. Fresco, F P 26,871 ( 1967). 95a. H. Iwamura, BBA 308, 333 ( 1973). 95b. K. L. Wong, Y. P. Wong and D. R. Kearns, Biopolymers 14, 749 (1975). 96. A. Favre and M. Yaniv, FEES Lett. 17,236 ( 1971). 97. A. Danchin, Biochimie 54, 333 ( 1972). 98. A. Danchin, Biochimie 55, 17 ( 1973). 98a. H. I. Heitner, S. J. Lippard and H. R. Sunshine, JACS 94, 8936 (1972). 98b. H. R. Sunshine and S. J. Lippard, NARes. 1, 673 (1974). 98c. A. S. Jones, R. T. Walker and V. Young, BBA 299,293 (1973). 99. B. Hoffman, P. Schofield and A. Rich, PNAS 62, 1195 (1969). 100. D. Kabat, B. Hoffman and A. Rich, Biopolymers 9,95 (1970). 101. T. Stone, T. Buchman, P. Nordio and H. McConnell, PNAS 54, 1010 (1965).

42

TED T. S A M 1 AND SEYMOUR S. W H E N

102. Z. Ohashi, F. Harada and S. Nishimura, FEBS Lett. 20, 239 (1972). 102a. R. P. Singhal, Bchem 13, 2924 ( 1974). 103. M. J. Waring, J M B 13, 269 (1965). 104. A. R. Peacocke and J. N. H. Skerrett, Trans. Faraday SOC. 51,261 (1956). 105. G. Scatchard, Ann. N.Y. Acad. Sn'. 51,660 (1949). 106. I. M. Klotz and D. L. Hunston, Bchem 10, 3065 (1971). 107. Y. Takeda and T. Ohnishi, BBRC 83,611 (1975). 108. M. S. Kayne and M. Cohn, Bchem 13,4159 (1974). 109. J. M. Wolfson and D. R. Kearns, Bchem 14,1436 ( 1975). 110. C. R. Jones and D. R. Kearns, PNAS 71,4237 ( 1974). 111. D. C. Lynch and P. R. Schimmel, Bchem 13,1841 (1974). 112. A. A. Schreier and P. R. Schimmel, J M B 88,601 (1974). 113. A. A. Schreier and P. R. Schimmel, J M B 93, 323 ( 1975). 114. A. Danchin, Biopulymers 11,1317 (1972). 115. G. Rialdi, J. Levy and R. Biltonen, Bchem 11,2472 (1972). 116. C. R. Jones and D. R. Kearns, Bchem 14,2660 (1975).

information Transfer and Sperm Uptake by Mammalian Somatic Cells AARONBENDICH, ELLENBORENFREUND, STEVENS. WITKIN, DELIABEJU,' AND PAULJ. HICCINS Laboratory of Cell Biochemistry Memorial Sloan-Kettering Cancer Center New York, New York

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

I. Introduction 11. General Method for the Study of Interaction of Spermatozoa and Mammalian Cells in Culture . . . . . . . . . A. Living Sperm and Recipient Cells . . . . . . . B. Electron Microscopic Evidence of Uptake of Sperm . . . C. Autoradiographic Data . . . . . . . . . . D. Morphological Changes Resembling Oncogenic Transformation . E. Information Transfer and Expression of Fetal Gene Products . F. Uptake of Heat-Killed Sperm by Target Cells 111. RNase-Sensitive Endogenous DNA Polymerase Activity in Human Semen. . . . . . . . . . . . . . A. Isolation and Characteristics . . . . . . . . B.'Possible Role in Embryogenesis and Oncogenesis . . . . IV. Conclusions and Speculations . . . . . . . . . References . . . . . . . . . . . . . Note Added in Proof. . . . . . . . . . .

. . . . .

43 45 45 49

55 57 61 64 65 66 70 71 72 75

I. Introduction Sperm are endowed by their specific differentiation to interact with and to deliver their genetic components into ova for subsequent activation and function. Whether analogous genetic information can be transferred by sperm to mammalian somatic cells is unknown. Studies were begun (I, 2) to determine whether this could be achieved with cells growing in culture, The initial results ( 3 ) indicated that normal diploid Chinese hamster ( C H ) cells can be penetrated following simple admixture in uitro with spermatozoa obtained from the vasa deferentia or epididymides of the mouse. The interactions in these species and in Laboratory of Viral Ultrastructure, Memorial Sloan-Kettering Cancer Center.

43

44

AARON BENDICH ET AL.

other sperm/somatic cell systems were monitored by light microscopy on living or fixed and stained specimens, by scanning or transmission electron microscopy, by autoradiography using sperm whose DNA had previously been labeled with [3H]thymidine in uiuo and by immunochemical means (3-5). Abnormalities seen in the growth pattern of target cells mimicked those appearing in controls treated in parallel with chemical carcinogens. It is not yet clear whether the changes were due to a fertilization-type acquisition of genetic information from the sperm, or to a recently discovered (6) RNase-sensitive DNA polymerase/ endogenous template complex either within or intimately associated with the sperm head, and both possibilities are under active investigation. This article describes the in uitro sperm/somatic cell system and the isolation of the sperm DNA polymerase/ template complex, and explores their potential to reveal aspects of the molecular biological mechanisms operative during genetic information transfer, fertilization and oncogenesis. Although not very numerous, various possible consequences resulting from the in uiuo interaction of sperm with somatic cell tissues have appeared in the literature. In an early study ( 7 ) , it was claimed that sperm penetrate into the uterine tissue and previously formed dividing ova and blastulae of mammals. Since the sperm were introduced into the already pregnant female by subsequent coitus, it was suggested that a genetic trait absent in the parents could be passed on to the offspring in this manner (telegony). Entrance of sperm into somatic tissues of the female reproductive tract has proved to be a controversial subject involving claims and counter claims (8, 9). The most persuasive data, obtained by autoradiography utilizing I4C-labeled sperm, provide no evidence for the incorporation of spermatozoa into tuba1 or uterine cells of the mouse after natural mating ( 1 0 ) . Sperm were recently reported deep within the tissues of the uterine mucosa when rats were artificially inseminated 4-6 days after natural mating ( 1 1 ) ; this was regarded as a possible cause of infertility. The phagocytic uptake of sperm by immature squamous metaplastic cells in the human cervix has been postulated as a cause of cancer after a long latent period (12, 13). It has been reported that sperm interact with somatic cells of the epididymis, as evidenced by the formation of abnormal cells and granulomas (14, 15). Escape of sperm from the genitourinary tract may be the basis for the appearance of antisperm antibodies in normal males and in those with blockage or after vasectomy (16, 1 6 ~ ) . Using cell-to-cell fusion-enhancing agents such as Sendai virus or lysolecithin, rabbit sperm have been brought into Syrian hamster fibroblasts in culture (17, 1 8 ) ; the biological effect of this interaction has

INFORMATION TRANSFER AND SPERM/ SOMATIC CELL INTERACTION

45

not been described. Uptake and sequestering of sperm in the corona cells surrounding the ovum have been described after natural mating in the rabbit ( 1 9 ) and in human in uitro fertilization studies ( 2 0 ) , but it is not clear whether this represents a phagocytic process or has particular significance in fertilization. Phagocytosis would, ostensibly, lead to the destruction of ingested sperm (9, 21 ), but surprisingly little is known about this phenomenon ( cf. 16a). There has been little experimentation in the area of sperm/somaticcell interactions, perhaps because of a disinclination to consider that sperm could exert effects other than those seen in the fertilization of eggs. It has generally been taken for granted that somatic cells are not penetrated or otherwise affected by the male gamete. Sperm supposedly contributes its haploid genome to the genetics of the developing zygote, and nothing else.

II. General Method for the Study of Interaction of Spermatozoa and Mammalian Cells in Culture

Although apparent uptake of sperm occurs upon their addition to monolayer cultures of various somatic cells, counting, mixing and subsequent experimental manipulations and observation are more accurately and conveniently performed following coincubation in suspension. Target somatic cells growing in monolayer are harvested by standard trypsin treatment, washed and suspended in nutrient medium. The cells are then mixed with 1 to 15 times as many washed spermatozoa, collected aseptically from the vasa deferentia or epididymides of animals, or from humans by masturbation; the admixture is incubated at 37OC in a moist atmosphere of 5%CO, in air (3).The suspended mixtures, which contain from 2 to 5 X lo5 somatic cells per milliliter, are periodically examined in the living state in culture chambers ( 2 2 ) or placed in petri dishes containing cover slips, which are removed after various times for study, or passaged for further propagation. Most of the cell-culturing techniques and methods of examination employed have been described (23-25); details for particular applications are given in the legends to the figures, below.

A. Living Sperm and Recipient Cells Within 1 to 3 hours after mixing, living spermatozoa can be observed to associate intimately with the target somatic cells and either to penetrate actively or to be taken up passively at later times when the sperm have lost their motility. Illustrations of homologous and heterogenetic interactions are shown below. The first example, shown in Fig. 1, is

46

AARON BENDICH ET AL.

F'xc. 1. Living cocultures of mouse spermatozoa and Chinese hamster (CH)cells; ratio of sperm to somatic cells is 5: 1; phase contrast. ( a ) Two hours after admixture in suspension, most cells have not yet attached to the surface of the culture chamber and spread out. ( b ) After 24 hours, most CH cells have attached and flattened out and contain at least one spermatozoon. ( c ) After 40 hours, about half of the CH cells are binucleate. ( a ) ~ 3 1 0 ;( b ) and ( c ) ~ 3 9 0 .Figures ( a ) and ( c ) are from Bendich et d. ( 3 ) , reproduced by permission of the American Association for the Advancement of Science.

the mouse sperm/CH somatic cell system (1-3). More than half of the still partially suspended CH. cells appeared to contain at least one mouse sperm head 2 hours after mixing (Fig. l a ) , and nearly all had included one or more after 24 hours, by which time many of the target cells had flattened and spread out on the cover slip surface (Fig. lb), In this experiment, the ratio of sperm to CH cells was 5:l. The first recognizable alteration in these normally mononucleate recipients was the appearance of binucleates (Fig. l c ) in about half of the cells; this occurred by about the 40th hour after admixture. In another series of experiments, rat sperm were mixed with established, functional rat liver parenchymal cells ( 2 5 ); the photomicrographs (Fig. 2) show sperm within such cells 40-42 hours after admixture. This system is of special interest since the recipient epitheloid cells are from the liver of the Gunn rat, a strain with a genetic disease characterized by the absence of the liver enzyme bilirubin :uridinediphosphoglucuronate transferase ( 2 6 ) . Presumably, the disease is due to the absence or impairment of the pertinent genetic determinant( s). (The human counterpart of this anomaly is known as Criggler-Najjar syndrome.) The spermatozoa used were from a rat (Fischer strain) genetically normal in this regard, Offspring cells from this combination are under study to determine whether restoration of the enzymic activity

INFORMATION TRANSFER AND SPERM/SOMATIC

CELL INTERACTION

47

FIG.2. Phase contrast micrograph of living cocultures showing Fischer strain rat sperm in Gunn rat liver parenchymal cells (25) 4&42 hours after admixture in an original ratio of 10:1. ( a ) ~ 6 4 0 (; b ) ~ 8 0 0 .

has been effected; however, unexpected changes were observed (see below). Another homologous system investigated revealed the uptake of normal human spermatozoa by normal human foreskin fibroblasts, and is illustrated with specimens fixed 48-72 hours after interaction, and stained (Fig. 3). This is part of an ongoing study to learn whether various chemical and enzymic pretreatments of the sperm might induce alterations resembling the capacitation and the acrosome reactions seen after insemination (cf. 21, 27-29) and affect the extent of interaction and consequent biological effects. Such alterations are regarded as prerequisites for the fertilization of mammalian ova. Spermatozoa of the above eutherian mammals have a high cystine content in their keratinlike nuclear chromatin, and prior cleavage of 4-Sbonds is necessary for release of their DNA in soluble form ( 3 0 ) , and also probably for the decondensation of the chromatin in the formation of the male pronucleus after fertilization (31-33). In contrast, the sperm chromatin from noneutherians, such as the Australian opossum ( Trichosurus volpecula), contains very little cystine and undergoes decondensation by simple exposure to detergent. We are grateful to Dr. J. M. Bedford for an epididymis freshly removed from this marsu-

48

AARON BENDICH ET AL.

FIG.3. Interaction of human spermatozoa and human foreskin fibroblasts, ratio 15:1; fixed and Giemsa stained. ( a ) At 72 hours after mixture with untreated living sperm. ( b ) At 48 hours after mixture with living sperm in cocultures to which a fraction of human seminal fluid was added. ( c ) At 48 hours after mixture with sperm which had been preheated in phosphate-buffered saline at 80°C for 15 minutes. ( a ) and ( b ) x390; ( c ) ~ 2 4 0 .

FIG.4. Uptake of living Australian opossum sperm by Chinese hamster ( C H ) cells; ratio 10:1; fixed and Giemsa stained. Note numerous examples of swollen sperm heads within the recipient CH cells. In contrast, the heads of the sperm of this marsupial are normally very compact, and remain so when not taken up by the cells in these cultures. Coculture was for 48 hours in ( a ) and for 72 hours in ( b ) and ( c ) All x 390.

.

pial. The interaction of the highly motile excised opossum sperm with cells of several mammalian species was studied and examples of uptake by CH cells are shown (Fig. 4 ) . Many examples of decondensed sperm nuclei were seen after uptake.

INFORMATION TRANSFER AND SPERM/SOMATIC CELL INTERACTION

49

B. Electron Microscopic Evidence of Uptake of Sperm Light microscopy by itself is not an ideal means to determine actual uptake of sperm; scanning electron microscopy (SEM) at various periods after the initiation of an experiment provides more persuasive evidence. A Chinese hamster cell that has not yet completely flattened out is seen penetrated by a mouse spermatozoon 3 hours after admixture (Fig. 5 ) ( I ) . In a series of experiments in which homologous systems were studied, rat sperm were coincubated with established rat liver parenchymal cells, and an illustration of the penetration observed by SEM is shown (Fig. 6) 2 hours after mixing. The rat sperm are much larger

FIG.5. Scanning electron micrograph of mouse spermatozoon that had penetrated a Chinese hamster cell 3 hours after admixture in the living state, at an original ratio of 1:1, fixed in glutaraldehyde, and coated with gold; x7750. Reproduced from Bendich et 4Z. ( I ), with permission.

50

AARON BENDICH ET AL.

FIG. 6. Scanning electron micrograph showing normal Fischer rat spermatozoa that had penetrated Gunn rat liver parenchymal cells ( 2 5 ) , 2 hours after admixture in the living state at a ratio of 10:l. Preparation was fixed in glutaraldehyde, dried by the COz critical point method and gold coated. Arrow indicates sleeve protruding from recipient cell plasma membrane and enveloping part of the sperm midpiece (see Fig. 8 for an enlarged view of this phenomenon in a transmission electron photomicrograph). x 1700.

and more vigorous than are those of the mouse and on occasion a spermatozoon appears to pass directly through the target cell. In another homologous system, cells (SH-2 line) of a human breast adenocarcinoma were coincubated with normal human sperm, and an aliquot was removed 2 hours after admixture for SEM examination (Fig. 7). This illustration shows a spermatozoon whose head and adjacent midpiece have been engulfed by the upper SH-2 cell, microvilli of which are seen extending from the surface membrane of the engulfing cell.

INFORMATION TRANSFER AND SPERM/SOMATIC CELL INTERACTON

51

Atop the same upper SH-2 cell, a second spermatozoon is seen whose midpiece is just below the upper surface of the target cell and is enclosed by microvillus structures, which also interlace around the sperm head in a meshlike pattern. Further examination of SEM photographs such as these may reveal aspects of the mode of uptake of sperm that, taken together with the transmission electron micrographs below, indicate that a process akin to cell-cell fusion, analogous to what is seen in

FIG.7. Scanning electron micrograph showing the penetration (double arrows) of human sperm into a human breast adenocarcinoma cell [SH-2 line (33a)l (original sperm to cell ratio 15:l) 2 hours after in uitro coincubation. Preparation was fixed in glutaraldehyde, dried by the COz critical point method and gold coated. Another spermatozoon is shown whose proximal midpiece is superficially embedded ( heavy black arrow) within the SH-2 cell by microvilli, which extend over the sperm head . Note Added in Proof, p. 75.1 in a meshlike pattern. ~ 5 3 0 0[See

52

AARON BENDICH ET AL.

FIG.8. Transmission electron micrograph. Illustrating the penetration of a mouse sperm into a Chinese hamster (CH) cell 48 hours after coincubation; original sperm to cell ratio was 5:l. Small arrows show the terminus of sleeve, or protrusion, which is continuous with the plasma membrane of the recipient CH cell and envelops the proximal part of the sperm midpiece. The heavy arrow points to the perforatorium at the apical region of the sperm head, which has produced an indentation in the CH nucleus (N).The nuclear chromatin has undergone extensive decondensation. Compare this interaction of a sperm head at the nuclear membrane with those in the autoradiographs in Fig. 11. Specimen was fixed in glutaraldehyde postfixed with 0 ~ 0 4 ,and flat-embedded in epoxy resin; thin sections were stained with uranyl acetate and lead citrate for examination in a Siemens electron microscope. x6800. the penetration of ova (19, 20, 34-40), attends the interaction of sperm and somatic cells. The most direct evidence for the presence of sperm in somatic cells after coincubation is derived from transmission electron micrographs of a thin section; illustrations 2-3 hours after interaction have been pub-

INFORMATION TRANSFER AND SPERM/SOMATIC CELL INTERACITON

53

lished ( 3 ) . Additional examples, shown in Figs. 8 and 9, reveal cytological details of the entry 48 hours after admixture of mouse sperm and CH cells. Figure 8 shows a sagittal section of a sperm head, the chromatin of which has largely dispersed. The “perforatorium” at the apex of the head is easily discernible and is seen as having produced an indentation in the target cell nucleus; a considerable portion of the tail and its mitochondria are still intact. Of interest is the formation of a “sleeve” or protrusion continuous with the plasma membrane and covering a considerable segment of the sperm midpiece, suggesting that

FIG.9. Transmission electron micrograph showing a mouse sperm head ( S ) and sperm tails ( T ) within a Chinese hamster’(CH) cell 48 hours after coculture (original ratio of sperm to CH cells = 5:l). Note intact appearance of mitochondria within tail in center, the initiation of decondensation of the sperm head, and the absence of digestion vacuoles or their membranes around these spermatozoal organelles. For details of preparation, see legend to Fig. 8. ~37,000.

54

AARON BENDICH ET AL.

FIG. 10. Uptake of mouse sperm, preheated at 80°C for 15 minutes (in 0.14 M NaCl, 0.01 M phosphate buffer, pH 7.2), by Chinese hamster ( C H ) cells 72 hours after incubation at a ratio of 5: 1. The sperm heads ( S ) and tail ( T ) are

not enclosed in digestion vacuoles; N represents the nucleus. Polyspermy is also seen with unheated sperm. Details of preparation are given in Fig. 8. x 10,700.

the recipient cell itself was involved in an active process of uptake (see also Fig. 6 ) . A very similar protrusion from a sea urchin egg that surrounded a mussel sperm head was described in experiments in which heterologous fertilization had been attempted (41 ). There is no evidence of the presence of cell plasma membrane or of digestion vacuole membrane within this Chinese hamster “sleeve” or within those areas of the cell that surround the tail and sperm head (see also Fig. 9) as would be expected if the uptake were the result of a phagocytic or digestive process (21, 42, 43). [Spinach chloroplasts, organelles that contain DNA, are similarly taken up in the cytoplasm by mammalian cells in culture, and retain their viability for several days. However, if they are first heat-denatured, the chloroplasts are all observed in digestion vacuoles ( 4 3 ) . Mouse sperm heated at 80°C for 15 minutes are taken up by CH cells; an example is shown (Fig. 10) in which three sperm heads can be seen in the cytoplasm, and there is no evidence of digestive or phagocytic vacuole formation.] Examination of these and many other transmission electron micrographs do not reveal outer sperm membranes or acrosomes associated

INFORMATION TRANSFER AND SPERM/ SOMATIC CELL, I N T E R A ~ O N

55

with the spermatozoa1 nuclei within the recipient somatic cells, with the possible exception of a few ( 3 ) at very early times after admixture. The absence of the outer vestments and acrosomes from these nuclei and the appearance of the vesiculating and decondensing chromatin (Figs. 8-10) are closely analogous to what has been described in electron micrographs of sperm-egg fertilization (19, 20, 3 4 4 0 ) . The close association of the decondensing sperm head with the target cell nucleus (Fig. 8; other examples will be published elsewhere) may well be the origin of the nuclear label seen in autoradiographs, shown below. Taken together, these illustrations leave no doubt that sperm can enter somatic cells under culture conditions in uitro. Confirmation of this phenomenon has appeared in brief reports (44, 4 5 ) . It is important to learn whether any observed biological effects on the recipient cell can be ascribed specifically to the introduction of the sperm genome or to other factors still unknown. This problem is difficult to resolve, since sperm heads contain potent enzymes in their acrosomes, the tails contain mitochondria, and semen contains cells other than sperm ( 4 6 ) even though the proportion of such contaminants in normal specimens is small indeed. Furthermore, the genitourinary tract of normal adult animals and humans is not sterile since it contains viruses and mycoplasma, and these can adsorb to sperm ( 4 7 - 4 9 ) . These variables are probably not as relevant when sperm are excised aseptically from the vas or epididymis or when semen specimens have been heated at 80°C for 15 minutes. However, rigorous evaluation of data must take them into account.

C. Autoradiographic Data To follow the fate of donor DNA after admixture of sperm and recipient somatic cells, animals were injected intraperitoneally with ["]thymidine (20 pCi, 40 Ci/mmol) and the spermatozoa were harvested from the vas at the period expected to represent the peak of their maturation. Three to four injections spread over 2 days, given to adult Swiss mice, provided well-labeled sperm when excised aseptically 32 days later (50). Under these conditions, and depending on the experiment, 21-75% of the sperm were labeled ( 3 ) ,producing many silver grains in autoradiographs (cf. 24, 5 1 ) after about 7-13 weeks of development. Only the heads of the spermatozoa were labeled; there was no label at all in any of the few somatic cells (>0.5%) in the sperm preparations. In consequence, the silver grains in the autoradiographs shown here (Fig. 11) and elsewhere ( 3 ) cannot be ascribed to cells other than sperm. Very few of the recipient cells showed nuclear uptake of label until about 48 hours after admixture of the "-labeled sperm with CH cells

56

AARON BENDICH ET AL.

FIG. 11. Autoradiographic evidence of penetration of somatic cells by labeled sperm collected from mice 32 days after they were inoculated with ['Hlthymidine and fixed in Giemsa stain. Specimens were photographed with partial interference contrast optics to accentuate silver grains. For ( a ) and ( b ) , living 'H-labeled sperm were mixed at a ratio of 15: 1 with Chinese hamster ( C H ) cells (Dede strain, female, American Type Culture Collection), and for ( c ) to ( f ) with CH cells [CLM strain, male ( 5 2 ) ] at a ratio of 2:l; all specimens were fixed 3 days after coincubation. Arrows point to interaction of 'H-labeled sperm heads at nuclear membrane and labeling of nuclei in ( a ) , ( b ) , ( c ) , ( d ) and ( f ) suggesting direct transfer of label into nuclei. Note extranuclear label in ( e ) and the uniform labeling of the complex nucleus in ( d ) and of the nuclei of the binucleate cell in ( e ) and the trinucleate cell in ( f ) . ~ 7 3 0 .

of the Dede line (lung; female) (Fig, l l a , l l b ) or until about 48-72 hours with the CLM line ( 5 2 ) (bone marrow; male) (Fig. llollf). Between 3 and 10%of the recipient cells were labeled, depending on the experiment, and, with particular exceptions discussed below, the label was within the nuclear region rather than over the cytoplasm.

INFORMATION TRANSFER AND SPERM/ SOMATIC CELL INTERACTION

57

The labeling pattern is consistent with an intrucellular release of spermatozoal chromatin or DNA after sperm had been taken up by the target cells. If release had taken place outside the recipient cells, the label would have been diluted and distributed over many cells (24, 53, 5 4 ) . Of particular interest is the frequent presence of a labeled sperm head on or at the nuclear membrane (see arrows in Fig. 11) suggesting a direct transfer of 3H material into the nucleus. These and other illustrations show direct continuity of silver grains between the sperm head and the nucleus much as though expulsion of chromatin had occurred, perhaps in a manner analogous to that described after stallion sperm had been treated with a disulfide-cleaving agent (55; cf. 30-33). Although the resolution in these light microscope autoradiographs does not permit a more precise description of this phenomenon, transmission electron micrographs, such as Fig. 8 (others unpublished) indicate an intimate association and interaction of sperm heads with the nuclear membranes after uptake. These illustrations do not reveal the original level of complexity or nativeness of the labeled material that had entered and remained in the nucleus; this problem is under study. In several instances, small, apparently organized, packets of labeled material, not associated with recognizable sperm structures, were seen in the cytoplasm of cells containing two or more labeled nuclei (Fig. l l e ) . Although the only source of 3H in these experiments was the sperm itself, it is not yet clear whether the extranuclear isotope arose from partially dissociated spermatozoa1 remnants in the cytoplasm. It is quite possible that these DNA-containing cytoplasmic bodies were derived from the nucleus (cf. 56), since microinclusions such as these are found to contain label when 48-hour cocultures of unlabeled sperm and CH cells are subsequently exposed to traces of [ "1 thymidine and examined by autoradiography ( unpublishcd ) . Other examples of extranuclear DNA in mammalian cells have been described (57-59). If, on the other hand, the microinclusions were of direct spermatozoal origin, uptake of ]H"[ thymidine would suggest that they contain an active DNA-synthesizing system. The existence of a sperm-associated RNase-sensitive DNA polymerase endogenous template complex ( 6 ) is discussed below. It has thus far been difficult to obtain appropriately labeled rat sperm using the labeled-thymidine procedure described above. We are studying methods to calibrate the rat system so that inordinately long autoradiographic exposure periods can be avoided.

D. Morphological Changes Resembling Oncogenic Transformation Reference was made earlier regarding the very frequent formation of binucleates (Fig. l c ) about 40 hours after the coculturing of mouse sperm and recipicnt mononucleate CH somatic cells. The use of hetero-

58

AARON BENDICH ET At.

genetic systems permitted karyological (23, 24, 52) exploration of the possibility that the second nucleus contained chromosomes belonging to the invading sperm. There was no evidence in any of the heterologous systems studied that any of the chromosomes seen were of spermatozoa1 origin. Absence of donor species chromosomes also minimized the chance that progeny cells arose from fusion with somatic cell contaminants in the sperm preparations. The binucleates therefore appeared to arise from a cytokinetic block, and, accordingly, the entering sperm had interfered not with the first mitosis per se, but rather with orderly cell division. At later times, tri- and polynucleate cells frequently formed that did not occur in parallel, contemporaneous control cultures in the absence of sperm; in addition, cells with abnormally large, small, or distorted nuclei also appeared (Fig. lld, l l f and 12). This interference with orderly cytokinesis and mitosis was also seen in the homologous sperm/somatic cell combinations (see Figs, 2b, 3b and 13). Evidence that the bi- and polynucleation did not arise from fusion of the somatic cells, possibly sperm-induced, is derived from examination of autoradiographs of recipient cells which contain two (3) (see also Fig. l l e ) or more labeled nuclei ( Fig. l l f ) following coculture with SH-labeled sperm. The silver grains in these (and many unpublished) examples were always homogeneously distributed among the labeled nuclei in individual cells. It is therefore highly unlikely indeed that only those somatic cells whose nuclei were equally labeled would have undergone fusion. The morphological alterations resembled those ( 25) attending the malignant transformation of Gunn rat liver parenchymal cells exposed to the potent chemical carcinogen methylazoxymethanol acetate ( MAM ) ( cf. 61 ) . Accordingly, the effect of treatment of these cells with rat sperm, MAM and another carcinogen 3,4-benzo[a]pyrene was studied, and a comparison was made with the untreated liver cells carried as a parallel control. The results are summarized in Fig. 13. The mononucleate liver cells grow in regular cobblestone array as monolayers in culture (Fig. 13a) and show a contact-inhibited pattern of growth ( 62, 51 ) . Treatment with either carcinogen changed this morphology (Fig. 13b and 13c) so that polynucleation, loss of regular growth pattern and piling-up of cells on top of one another resulted (25). Very similar alterations resulted from interaction with rat sperm (Fig. 13d). The untreated cells continued to show a regular morphological pattern throughout the many weeks of growth, passage and observation while all 3 sets of treated cells maintained their abnormal, transformed pattern of growth and acquired the ability to grow as three-dimensional colonies in semisolid agar (63). Whereas the untreated cells and those treated with benzo-

INFORMATION TRANSFER AND SPERM/SOMATIC CELL INTERACTION

59

FIG. 12. Morphological alterations of somatic cells after coculture with living mouse spermatozoa. ( a ) Four hours and ( b ) 50 hours after admixture of thymidinekinaseless Syrian hamster cells, BHK-TK-, ( 6 0 ) with mouse sperm (ratio of sperm to cells, 4:l); photographed in living state with phase contrast optics. ( c ) At 48 hours and ( d ) 144 hours after coculture of Chinese hamster cells with 15 times as many Australian opposum spermatozoa; fixed and Giemsa stain. Note bi- and polynucleate cells. ( a ) , ( c ) and ( d ) ~ 4 4 0 (; b ) x510.

pyrene or sperm have not produced malignant tumors upon inoculation into thymusless nude mice ( 6 4 ) , the MAM-treated cells have acquired malignant potential since tumors have developed upon their injection into these mice (25). Perhaps the sperm-treated cells have a longer

60

AARON BENDIM ET AL.

FIG.13. Changes resembling oncogenic transformation. Specimens were fixed and Giemsa stained. ( a ) Monolayer culture of established Gunn rat liver parenchymal cells at passage No. 35. Continued growth and passage for an additional 30 weeks provided cultures with unchanged appearance (25). ( b ) Passage No. 35 cells 12 weeks after single treatment with the carcinogen methylazoxymethanol acetate. Note piling character, abnormal forms and loss of contact-inhibited type of growth pattern. (c) Passage No. 35 cells 12 weeks after single treatment with the carcinogen 3,4benzo[a] pyrene. (a) Passage No. 35 cells 6 weeks after coincubation with 10 times as many normal Fischer rat sperm. Note similarity of appearance to that in ( b ) and ( c ) . All ~180 .

INFORMATION TRANSFER AND SPERM/SOMATIC CELL INTERACTION

61

latent period or require the additional action of tumor promoters (65), such as phorbol ester (66), before they will undergo malignant transformation; this cocarcinogenic promoter treatment is prerequisite for the malignant transformation of benzopyrene-treated cells in culture ( 63). This possibility is under investigation. An original objective in the use of the Gunn liver cells was to learn whether their genetic deficiency (26) manifested by the absence of bilirubin: uridinediphosphoglucuronate transferase, an enzyme not yet isolated, could be corrected by their exposure to sperm from genetically normal rats. The unexpected changes observed (Fig. 13d) have introduced a potentially hazardous complication in this exercise in attempted genetic engineering. Restoration of the above enzymic activity has recently been reported (67) following transplantation of small liver segments of normal rats into the livers of Gunn rats. The prevailing assays for the determination of the enzyme are not as yet sensitive enough to be applied to the limited quantities of uncloned cells available from our culture systems. Development of pertinent micro tests for the enzyme is under study.

E. Information Transfer and Expression of Fetal Gene Products The characteristic appearance of embryonal or fetal gene products in tumors (cf. 68-72), and the obvious role of the male gamete in mammalian embryogenesis, prompted studies to see whether fetal antigen production could be initiated by sperm-somatic cell interactions. In our initial study ( 3 ) , mouse fetal antigen expression was observed (Fig. 14a) at a frequency of about 0.01% in the offspring CH cells [CLM line (52)] 21 days after interaction with mouse sperm. This was demonstrated by immunofluorescence methods ( 73, 2 5 ) utilizing antiserum, prepared in rabbits against extracts of 19-day mouse fetuses (69), and fluorescein-conjugated goat anti-rabbit IgG, which were appropriately absorbed 'so that control CH cells were negative. To determine whether the efficiency of this phenomenon could be enhanced, combinations of several species of sperm and homologous or heterologous target cells were compared in orienting experiments ( 4 ) . The interaction of sperm, from Wistar strain rats, and CH cells (Don line; male, lung) led to the highest yield of cells expressing fetal antigens of the sperm species (Fig. 14b, 14c). In this latter system, rabbit antisera were prepared against lyophilized aqueous extracts (69) of day 17-21 rat fetal tissue. The rabbit antiserum and the goat anti-rabbit IgG were absorbed with lyophilized extracts and whole tissue homogenates of adult CH organs; the fluorescein-conjugated goat anti-rabbit IgG was further absorbed with lyophilized extracts of adult rat liver. Neither the sperm gently

62

M R O N BENDICH ET AL.

FIG.14. Appearance and production of fetal antigens after sperm-somatic cell interaction. ( a ) Chinese hamster ( C H ) cells [CLM strain (52)] 21 days after admixture with 5 times 8s many mouse spermatozoa; methanol-fixed cells exhibit a positive indirect immunofluorescence in the ultraviolet after treatment with CH-absorbed rabbit antiserum to mouse embryo. Taken from Bendich et d. (3),reproduced by permission of the American Association for the Advancement of Science. ’( b ) CH Cells (DON strain, male, lung, American Type Culture Collection) 8 days after mixed culture with 15 times as many rat spermatozoa ( 5 ) . Cells fixed with methanol and treated with CH-absorbed rabbit antiserum to day 17-21 rat fetal tissue. Fluoresceinconjugated goat anti-rabbit IgG antiserum was absorbed with both CH extract and rat liver extract. Control CH(DON) and established adult Gunn rat liver cells as well as established Morris hepatoma cells were negative with this antiserum; rat embryo cells were positive. ( c ) CH( DON) cells 48 days after admixture with rat spermatozoa. Antisera used are those described in ( b ) . Cells were seeded lightly to obtain discrete colonies; all the cells in 1-2% of the colonies that arose thereafter showed positive indirect immunofluorescence ( 5 ) after treatment with CH-absorbed rabbit antiserum to rat fetal tissue. Photograph is taken from a positive colony. ( a ) and ( c ) x150; ( b ) x240.

extruded from the vas nor the CH cells, nor the rat sperm-somatic cell coculture at any time before day 8 after admixture, gave positive immunofluorescence reactions, suggesting that none of these cells contained detectable rat fetd antigen. On day 8 after coculture, 0.5-1.0% of the propagated CH cells showed a strong immunofluorescence (Fig. 14b) indicating that they had produced rat fetal antigens. Chromosomes of rat origin were not observed in the offspring cells at any time after coculture, nor was it possible to cultivate any rat cells from the original rat sperm preparation. Furthermore, cultured adult rat cells were negative in immunofluorescence tests with the antifetal serum. In consequence, it is most unlikely that the positively reacting cells (Fig. 14b, 14c) arose from a fusion of the CH cells and any somatic rat cell contami-

INFORMATION TRANSFER AND SPERM/ SOMATIC CELL INTERACTION

63

nants that may have been present in the rat sperm preparation; rather they most probably resulted from interaction of the rat sperm and CH cells. At days 14 and 36 after coincubation with rat sperm, the proportion of positively reacting cells had increased to about 3%and 4%,respectively. At day 39, cells were lightly seeded onto cover slips to obtain individual colonies (ZS),and, after 9 days of growth, all the cells in 1-211, of the resultant colonies showed the positive immunofluorescence reaction diagnostic for the presence of rat fetal antigen. Gunn rat liver cells, originally derived from adult animals were negative in this test, whereas cultured rat embryo cells were positive. On the other hand, the liver cells were positive when rabbit antiserum against whole adult rat serum was used as above, while the CH cells were negative when this latter test was applied to the CH cells before and after rat sperm treatment. The mechanism whereby these rat embryo antigen-producing CH cells acquired and expressed this heterologous genetic activity is not yet known. It is important to ascertain whether those CH cells that synthesized the rat antigen had in fact been penetrated by the rat sperm and whether the appearance of the antigen(s) was the result of direct gene transfer, induction by the sperm-associated RNase-sensitive DNA polymerase endogenous template complex, or a still unknown factor. This problem is under active study. The genes of one species of mammalian cells can be expressed and even maintained in those of another after cell-cell fusion (cf. 74-79), cell-cell or isolated DNA-cell interaction (24, 54, 80) and metaphase chromosome-cell coincubation ( 81-83). Acquisition of new genetic activities has resulted from such heterologous g,ene transfer. The restoration of a genetic function following the in vitro interaction of deficient target cells with isolated metaphase chromosomes from heterologous species that express this function occurred in a very low frequency (ca. lo-' to the acquired expression was transient in some instances (82), but permanent in others ( 8 3 ) . Specific selective culture media were used to obtain the altered progeny. The communications (81-83) of these findings did not include cytological evidence of actual uptake of metaphase chromosomes by the recipient cells (for examples of evidence of such uptake cf. 84-89), although much evidence was provided that donor chromosomes or fragments thereof were not seen in any of the selected offspring of the target cells. Consequently, the uctua2 molecular biological mechanism of the genetic information transfer involved in these experiments is still not known. It would indeed be remarkably fortuitous if gene transfer from the donor entities did not occur in the manner anticipated in the experiments, whether the donor gene-containing material was isolated DNA, spermatozoa, or isolated

64

AARON BENDICH ET AL.

metaphase chromosomes. However, it must be recognized that in all the various claims of genetic information transfer to recipient somatic mammalian cells cited, step-by-step rigorous proof of the mechanism by which genetic determinants were acquired and expressed is still lacking. It would be of interest to learn whether metaphase chromosomes from fetal cells as well as isolated sperm chromatin can be taken up by adult somatic cells to produce progeny in which fetal antigens reappear, and, indeed, if such cells acquire malignant potential. A difficult though not unsurmountable experimental stumbling block in such a study arises from the lack of an appropriate method to select those few cells that do express fetal gene activity and to separate them from those that do not. Another difficulty arises from a lack of understanding of the role and functions the embryonal and fetal gene products play in ontogenesis and oncogenesis. Recent findings implicate these gene products in phenomena concerned with the onset or control of mitotic activity (90,91).

F. Uptake of Heat-Killed Sperm by Target Cells Although the uptake, and in some cases expression, of exogenous genetic determinants by somatic cells is well known (24, 53, 54, 80, 92-94) (see also Section 11, E, above), mammalian cells also have a propensity to take up inanimate particulate matter of various kinds in oitro (95-97) and in viuo (98, 99). The inanimate particles [or heatdenatured organelles ( 4 3 ) ] are enclosed by the recipient cells in digestion or phagocytic membrane-lined vacuoles; this constitutes an efficient and clean method to prepare cell membranes (97). It was of interest to see whether heated sperm would also be engulfed by somatic cells; reference was made to this above (Fig. 3c, 10).The transmission electron micrograph (Fig. 10) provides unmistakable evidence of such uptake after mouse spermatozoa, 72 hours after having been heated at 80°C for 15 minutes in 0.14 M NaCl, 0.01 M phosphate buffer, pH 7.2 (buffered saline), were coincubated with CH cells, The absence of phagocytic or digestion vacuoles or membranes around the sperm nuclei is selfevident (in this and other, unpublished, micrographs) ; the appearance of the sperm heads and tail segments is indistinguishable from that obtained with living sperm. The abnormalities and polynucleation shown above in the various living sperm/cell systems (Figs. lc, 2b, 3b, 4a,c, lld-f, 12b-d, 13d and 14b) also develop when heated sperm are used (Fig. 15). These results disprove the possibility that these alterations arose from a fusion of the recipients with somatic cells in the sperm preparations

INFORMATION TRANSFER AND SPERM/SOMATIC CELL INTERACTION

65

FIG. 15. Morphological alteration of Chinese hamster ( C H ) cells [CLM strain ( S Z ) ] 6 days after coculture with 5 times as many preheated mouse sperm. Sperm were treated for 15 minutes at 80°C (in 0.14 M NaCI, 0.01 M phosphate buffer, pH 7.2), fixed and Giemsa stained. ~ 2 4 0 .

since such contaminants would be destroyed by the heating procedure. Although they are rapidly immobilized by the heating, mammalian spermatozoa are extraordinarily hardy ( 4 6 ) . The chromatin contained in the sperm head, especially of eutherian mammals, is highly condensed and cross-linked (see above). The DNA in mammalian chromatin has a much higher denaturation temperature than the protein-free DNA isolated therefrom (100, 101), and will remain undenatured in buffered saline at 80°. The chromatin in heated sperm would thus be potentially capable of exerting a genetic effect after uptake by somatic cells, or by fertilizable ova from which the coronal cells and zona pellucida were removed (102); perhaps such denuded ova could be fertilized by heated sperm.

111. RNase-Sensitive Endogenous DNA Polymerase Activity in Human Sperm Postfertilization development in amphibians and echinoderms is initiated by the unmasking of latent maternal messenger,RNAs that preexist in the ovum ( 103-105). In contrast, mammalian development appears to originate from the very early stages onward in the newly formed embryonic genome itself ( 106).Paternally derived gene products have been identified as early as the 8-cell stage in mouse embryos (107), but this does not occur until the time of hatching in Drosophilu (108)

66

AARON BENDICH ET AL.

or tailbud formation in frogs ( 1 0 9 ) . Apparently, a mechanism has evolved in mammals for activation (or derepression) of selective gene loci shortly after union of haploid paternal and maternal pronuclei. The substance( s ) responsible for the initiation of embryonic gene functioning, or the information for its biosynthesis, might be extrachromosomal in at least one of the mature gametes, Spermatozoa contain a complex with the potential to generate nonchromosomal genetic information. Sperm heads are intimately associated with a DNA polymerase/RNA template complex capable of synthesizing DNA when incubated with the four deoxyribonucleoside triphosphates plus Mgz+ ( 6 ); a similar complex was also found in seminal fluid. This section deals with the preparation and initial characterization of this complex from human sperm and its possible relevance to information transfer during fertilization or malignant transformation.

A. Isolation a n d Characteristics Human semen, obtained from apparently healthy donors by masturbation, was fractionated to yield cells and cell-free seminal plasma by centrifugation at 2000 g for 10 minutes, The cell fraction was resuspended in buffered saline containing 2 mM KC1 and centrifuged; the process was repeated for 4 cycles. In agreement with others ( 4 6 ) , we have consistently observed variable amounts of somatic cells in human ejaculates. In addition, mitochondria present in sperm tails possess a DNA-dependent DNA polymerase ( 1 1 0 ) , DNA and RNA ( I l l ) , which would also complicate the interpretation of subsequent experimental data. It was therefore imperative to obtain a purified population of sperm heads prior to biochemical analyses for components capable of participating in information transfer. The anionic detergent sarkosyl (sodium lauroyl sarcosinate) quantitatively cleaves mammalian sperm heads from tails at a specific region of the neck ( 1 1 2 ) . Accordingly, the 2000 g semen pellet was suspended in buffered saline containing 1.5% sarkosyl; after gentle agitation at 22°C for 60 minutes, all somatic cells were lysed and the sperm heads were separated from midpieces and tails ( 6 ) . Isolation of the dense sperm heads was then achieved by differential centrifugation in sucrose buffered at pH 7.5 ( 1 1 3 ) . The resultant pellet contained sperm heads with less than 1% tail contamination; no other cells were seen. Disruption of the isolated sperm heads must be accomplished without dissociating or inactivating any existing components that might function in information transfer. The rigidity of eutherian mammalian sperm head nuclei has been ascribed to extensive cross-linking of their cystine-rich protamines ( 3 3 ) . The addition of dithiothreitol, which specifically

INFORMATION TRANSFER AND SPERM~SOMATIC CELL INTERAC~ION

67

cleaves disulfide bonds, resulted in a labilization of the sperm heads. Suspension of the sperm heads in 0.02 M dithiothreitol at pH 7.5 and subsequent gentle Potter-Elvehjem homogenization led to their becoming markedly swollen, analogous to what has been described (32) when human sperm are treated with sodium dodecyl sulfate/ dithiothreitol at pH 9. Dithiothreitol treatment by itself without homogenization resulted in much less swelling; homogenization of untreated heads neither altered their morphology nor liberated polymerizing activity. Swollen sperm head preparations were assayed for their ability to synthesize DNA in the absence of exogenous templates. The assay reaction mixture added contained ["ITTP, dATP, dCTP, dGTP and MgCI,. Incubation was at 37OC. The reactions were terminated by the addition of yeast RNA and coprecipitation with trichloroacetic acid containing sodium pyrophosphate. Acid-precipitable radioactivity was collected on 0.45 pm pore membrane filters, which were then washed with 5%trichloroacetic acid, dried and assayed for radioactivity in a liquid scintillation counter. Swelling of the sperm heads was a prerequisite for the detection of the polymerization activity: unswollen heads were completely inactive. Maximal reactivity was obtained in the presence of 0.02 M dithiothreitol (Fig. 16A). Higher concentrations resulted in a lowered activity, owing perhaps to a cleavage of polymerase disulfide bonds or separation of enzyme from its template. Treatment of the swollen heads with the nonionic detergent Triton X-100 progressively decreased endogenous activity ( Fig. 16B), probably by dissociating the enzyme/ template complex. An initial fractionation of this endogenous activity was performed by subjecting the swollen sperm head preparations to differential centrifugation prior to analysis. Activity was demonstrated in a 10,000 g supernatant fraction, whereas the corresponding sediment was devoid of measurable polymerase. The active component was thus shown to be separable from the vast bulk of sperm chromatin. The endogenous polymerase activity was recovered from the sediment that resulted by centrifugation of the 10,000 g supernatant at 165,000 g for 60 minutes. The buoyant density of the endogenous complex was determined by centrifugation of the resuspended 165,000 g sediment through a linear 20 to 65% ( w / w ) sucrose gradient for 16 hours at 165,000 g. Fractions were collected by puncturing the bottom of the centrifuge tube, and aliquots of each were assayed for endogenous DNA polymerase activity in the usual deoxyribonucleoside triphosphate reaction mixture, which contained either [3H]dTTP or [3H]dCTP as labeled precursor. In both cases, a peak of acid-insoluble radioactivity was obtained that corre-

68

AARON BEIVDICH ET AL.

Dithiothreitol (molority)

Triton X-100 (%)

FIG. 16. The effect of dithiothreitol and Triton X-100 on human sperm head endogenous DNA polymerase activity. Purified sperm heads from 3 ejaculates were suspended in 0.05 M TrisC1, pH 7.5, mntaining either different dithiothreitol concentrations (A), or 0.02 M dithiothreitol plus graded amounts of the nonionic detergent Triton X-100 (B). After incubation for 20 minutes at O'C, all Triton-containing samples were brought to the same final detergent concentration and volume, and all samples were tested for endogenous DNA polymerase activity by incubation at 37°C for 10 minutes in a standard assay mixture containing dATP, dCTP, dGTP (20 nmol each), MgCL 1.875 pmol, KCl 2.5 pmol, EDTA 2.4 nmol, 5 PCi ['HIdTTP 53.7 Ci/mmol, 1650 cpm/pmol final concentration, TrisC1, pH 7.5, 0.12 pmol. The final volume was 0.125 ml. The reactions were terminated by the sequential addition of 10 pg of yeast RNA and 2 ml of 5% trichloroacetic acid containing 1%sodium pyrophosphate. 100%equals either 422 ( A ) or 627 ( B ) cpm.

sponded to a density of 1.16 g/ml (Fig. 17). The polymerization of [3H]dCTP by the 1.16 g/ml fraction occurred only in the presence of the other three deoxyribonucleoside triphosphates, and MgZ+(Table I). TABLE I REQUIREMENTS FOR SPERMHEADENDOGENOUS DNA POLYMERASE REACTION Incubation mix turea Complete -d ATP -dGTP -dTTP

Incorporation' 0.8 0.1 0.02 co.01

Aliquota (10pl) of the 1.16 g/ml density fraction obtained from a sperm head equilibrium centrifugation in sucrose (see legend to Fig. 16) were incubated at 37°C for 15 minutes in the [aH]dCTP assay mixture described in Fig. 17, from which the appropriate triphosphate was omitted. b Picomoles of dCTP converted to acid-insoluble product.

INFORMATION TRANSFER AND SPERM/ SOMATIC CELL INTERACTION

I

I

I

69

T-

1.3

7 xI

1.2

2

\

m

1.1 ,&ul

B 1.0

Fraction

FIG. 17. Buoyant density of sperm DNA polymerase-endogenom template complex. Purified sperm heads were treated for 180 minutes with 0.02 M dithiothreitol, homogenized, and centrifuged at 10,000 g. The 165,000 g pellet fraction obtained from the supernatant was suspended in 0.3 ml of 0.05 M TrisC1, pH 7.5 and layered onto 4.8 ml of a 20 to 65% ( w / w ) linear sucrose gradient in 0.05 M TrisC1, 1 mM EDTA and centrifuged in cellulose nitrate tubes for 16 hours at 45,000 rpm in an SW 50L rotor. Fractions were collected by puncturing the bottom of the tube, density was determined from the refractive indices ( x - - - x ) , and aliquots were incubated for 15 minutes at 37°C in the endogenous DNA polymerase assay mixture described in Fig. 16, modified by the omission of both KCI and EDTA. The three unlabeled deoxyribonucleoside triphosphates, 20 nmol each, were augmented by 5 PCi of either ["HldTTP (0-0) (17.5 Ci/mmol, 805 cpm/pmol final concentration) or [3HldCTP ( 0-0) (24.8Ci/mmol, 785 cpm/pmol final concentration).

Similar results using [3H]dTTP have been published ( 6 ) . Prior incubation with pancreatic ribonuclease A (EC 3.1.4.22) abolished the ability of the 1.16 g/ml fraction to synthesize DNA (unpublished). In earlier experiments, we reported that the main peak of endogenous sperm head complex banded in sucrose at a density of 1.22 to 1.25 ( 6 ) . In those instances, the purified sperm heads had been treated with 0.02 M dithiothreitol for only 20 minutes prior to fractionation, while in the present case treatment was for 180 minutes. Apparently, the longer incubation with dithiothreitol removed a component of the endogenous complex, which led to a lowering of density; in most instances, activity was increased by the prolonged incubation. An indication of the general occurrence of this complex in human

70

AARON BENDICH ET AL.

TABLE I1 RIBONUCLEASE-SENSITIVE ENDOQENOUS DNA POLYMERASE ACTIVITYI N HUMAN SPERMHEADS Donor"

Activity' 1.2 3.4 1.0 4.9 1.1 0.5 5.7 2.5

Each letter and subscript represents a single ejaculate from a particular donor. The 165,000 g pellet fractions (see legend to Fig. 17) of sperm heads were resuspended, banded in 20-65% sucrose gradients, fractionated and assayed for endogenous DNA polymerase activity as in Fig. 16, using I3H]dTTP. Activity is expressed in picomoles of dTTP inEorporated into acid-insoluble product in 10 minutes a t 37°C by ribonuclease-sensitive fractions per ejaculate. The activities listed for pooled specimens are average values. (I

'

sperm heads is shown in Table 11. Endogenous ribonuclease-sensitive DNA polymerase activity was fairly constant in the sucrose-banded material obtained from the sperm heads of ten individuals. This result argues against the possibility that the polymerization activity arises from adventitious genital tract contaminants and leads to the conclusion that the endogenous DNA-synthesizing complex is probably a normal sperm head component. Similar complexes were obtained from mouse sperm collected aseptically from the vas or epididymis; furthermore, an active polymerizing complex was recovered from human sperm heated at 80" in buffered saline for 20 minutes (unpublished).

B. Possible Role in Embryogenesis and Oncogenesis During fertilization, there is a fusion of the outer sperm and ovum membranes (see above), and the cystine-rich protamines of the sperm head dissipate and are lost (114);presumably, the DNA polymeraseendogenous template complex would be liberated and become active at some point during the ensuing nuclear dispersion and subsequent union of the gamete genomes. The DNA that would be synthesized via this polymerase could provide either amplification of selected information [perhaps for trophoblast formation (115) or for implantation of the early embryo (IlS)] or a product necessary for the activation or derepression of genetic determinants needed during early embryogenesis

INFORMATION TRANSFER AND SPERM/SOMATIC CELL INTERACTION

71

and without which normal development would be impaired ( 1 1 6 ) . The complex has properties comparable to oncogenic RNA viruses ( 1 1 7 ) and to those particles that have been observed in human neoplasia (118). Whatever the precise mechanism, the presence of an extrachromosomal genetic-information-generating system in sperm and its transmission to ova could provide an efficient mechanism to initiate or influence the programmed sequence of events that lead to implantation and development of the mammalian embryo. The penetration of somatic cells by sperm and subsequent release of the endogenous complex could result in a recapitulation of early embryonal gene functions and could lead to their becoming transformed. We have indicated in previous sections that penetration of somatic cells by sperm in vitro is followed by the appearance of embryonal gene products ( antigens ) and morphological cellular changes reminiscent of those seen in malignant transformation.

IV. Conclusions and Speculations Mammalian sperm interact with mammalian somatic cells as evidenced by actual penetration, the appearance of changes in morphology and growth characteristics, and the production of new gene products. These changes may be consequences of the direct acquisition and genetic expression of either the sperm chromatin itself or the reverse transcriptase-containing complex within the sperm heads. These possibilities are under investigation. It is tantalizing to contemplate the possibility that the interaction of sperm with somatic cells that leads to fetal antigen expression and to changes resembling oncogenic transformation may involve a recapitulation, at least in part, of the program of postfertilization gene expression. Viewed in this way, oncogenesis can be regarded as a form of embryogenesis that occurs at the wrong time and in the wrong place. Since the normal function of sperm is fertilization, the presence of its various components in somatic cells might very well induce changes analogous to those that occur in embryological and fetal development. It has been estimated (115)that the amount of unique-sequence DNA that is transcribed in the rabbit blastocyst but no longer in the midgestation embryo is enough to specify some 23,000 different proteins; the corresponding difference between normal and virus-transformed mouse cells is regarded as sufficient to specify some 200,000 different proteins (cf. 119). To sort out the many steps involved in embryogenesis, fetal development, and oncogenesis in the face of such complexity is a prodigious task indeed. It is hoped that the studies described in this article will also stimulate the curiosity of others to examine the sperm/ somatic cell system

72

AARON BENDICH ET AL.

and that this will provide a new basis to understand embryogenesis and “a lifetime of carcinogenesis” ( 120).

ACKNOWLEDGMENTS Previously unpublished work was supported in part by U.S.Public Health Service Grant CA07848 from the National Cancer Institute and the National Cancer Institute, Contract No. N01-CB-43904. We are grateful to M. Steinglass, G. Korngold, C. Ward and L. Rothman for excellent assistance, N. Lampen for excellent assistance with the scanning electron micrographs, and J. M. Bedford, E. de Harven and S. S. Sternberg for critical guidance and discussion.

REFERENCES 1. A. Bendich, E. Borenfreund, P. J. Higgins, S. S. Stemberg and E. H. Stonehill, in “Chemical Carcinogenesis” (P. 0. P. Ts’o and J. DiPaolo, eds.), F’t. B, pp. 515530. Dekker, New York, 1974 [from the World Symposium on Model Studies in Chemical Carcinogenesis, Johns Hopkins Univ., Baltimore, Md., Oct. 31-Nov. 3, 19721. 2. A. Bendich, E. Borenfreund and S. 5. Sternberg, FP 32, 615 (19’73). 3. A. Bendich, E. Borenfreund and S. S. Stemberg, Science 183, 857 (1974). 4. A. Bendich, E. Borenfreund and D. Beju, A d a Cytol. 18, 544 (1974). 5. P. J. Higgins, E. Borenfreund and A. Bendich, Nature (London) 257, 488 (19751. 6. S. S. Witkin, G. C. Korngold and A. Bendich, PNAS 72, 3295 (1975). 7. J. H. F. Kohlbrugge, Arch. EntwickZungsmech. Org. 35, 165 (1912). 8. C. R. Austin, Nature 183, 908 (1959); J . Endocrinol. 18, viii (1959). 9. C. R. Austin, J. Reprod. Fertil. 1, 151 (1960). 10. R. G. Edwards and J. L. Sirlin, Nature (London) 183, 1744 (1959). 11. R. Stein-Werblowsky, J. Fee#. 18,74 (1973). 12. B. L. Reid, Lancet I, 21 ( 1964). 13. B. Reid and B. Coppelson, in “Scientific Basis of Obstetrics and Gynaecology” (R.R. MacDonald, ed.), pp. 415-453. Churchill, London, 1971. 14. E. S. J. King, J. Pathol. Bacteriol. 70, 459 ( 1955). 15. A. D. Morgan, Brit. J. Urol. 36,95 ( 1964). 16. J. S. Cochran, Urology 4,367 ( 1974). 16a. N. J. Alexander, F P 34, 1692 ( 1975). 17. W. Sawiki and H. Koprowski, Erp. Cell Res. 66,145 (1971). 18. B. Gabara, B. L. Gledhill, C. M. Croce, J. P. Cesarini and H. Koprowski, PSEBM 143,1120 ( 1973). 19. J. M. Bedford, Amer. J . Anat. 133, 213 (1972). 20. P. Soupart and P. A. Strong, Fertil. SteriZ. 25, 11 ( 1974). 21. J. M. Bedford, J. Reprod. Fertil. 9,249 ( 1965). 22. A. Bendich, A. D. Vizoso and R. G. Harris, PNAS 57, 1029 ( 1967). 23. E. Borenfreund, M. Krim, F. K. Sanders, S. S. Sternberg and A. Bendich, PNAS 56,672 ( 1966 ) . 24. E. Borenfreund, Y. Honda, M. Steinglass and A. Bendich, J. Erp. Med. 132, 1071 ( 1970). 25. E. Borenfreund, P. J. Higgins, M. Steinglass and A. Bendich, J. Nut. Cancer Inst. 55, 375 ( 1975).

INFORMATION TRANSFER ANTI SPERM/SOMATIC CELL INTERACTION

73

26. R. Schmid, in “The Metabolic Basis of Inherited Disease” (J. B. Stanbury, J. B. Wyngaarden and D. S . Fredrickson, eds.), 3rd ed., pp. 1141-1178. McGraw-Hill, New York, 1972. 27. R. Yanagimachi, J . Erp. Zool. 170, 269 (1969); Biol. Reprod. 3, 147 (1970); J . Reprod. Fertil. 28, 477 ( 1972). 28. J. M. Bedford, Biol. Reprod. 2, Suppl. 2, 128 (1970). 29. B. G. Brackett, F P 32,2065 ( 1973). 30. E. Borenfreund, E. Fitt, and A. Bendich, Nature (London) 191, 1375 ( 1961). 31. H. I. Calvin and J. M. Bedford, 1. Reprod. Fertil. Sup$. 13, 65 (1971). 32. J. M. Bedford, M. J. Bent and H. Calvin, J . Reprod. F e d . 33, 19 ( 1973). 33. J. M. Bedford and H. Calvin, J. Erp. Zool. 188, 137 (1974). 33a. G. Seman, S. J. Hunter and L. Dmochowski, Abstr. Proc. 66th Annu. Meeting, Amer. Assoc. Cancer Res. 66, 59 (1975). 34. D. Szollosi and H. Ris,. J. Biophys. Biochem. Cytol. 10, 275 (1961). @. C. Barros, J. M. Bedford, L. E. Franklin and C . R. Austin, J. Cell Biol. 34, C1 (1967). 36. M. Stephanini, C. Oura and L. Zamboni, J . Submicrosc. Cytol. 1, 1 (1969). 37. L. Pik6, in “Fertilization” (C. B. Metz and A. Monroy, eds. ), vol. 11, pp. 325-404. Academic Press, New York, 1969. 38. R. Yanagimachi and Y. D. Noda, J. Ultrastruct. Res. 31, 465 (1970); Amer. J. Anat. 128,429 (1970). 39. F. J. Longo, A n d . Rec. 179, 27 ( 1974); Biol. Reprod. 9, 149 (1973). 40. E. Anderson, P. C. Hoppe, W. K. Whitten and G. S. Lee, J. Ultrastruct. Res. 50,231 (1975). 41. B. A. Afzelius, E r p . Cell Res. 72, 25 (1972). 42. E. Essner, J . Bwphys. Biochem. Cytol. 7, 329 (1960). 43. M. M. K. Nass, Science 165, 1128 ( 1969). 44. S. G. Phillips and D. M. Phillips, Abstr. 14th Annu. Meeting, Amer. SOC. Cell Biol., 1974, p. 269a. 45. L. C. Lau, Abstr. 19th Annu. Meeting Biophys. SOC., 1975, p. 58a. 46. T. Mann, “Biochemistry of Semen and of the Male Reproductive Tract.” Wiley, New York, 1964. 47. Y. M. Centifanto, D. M. Drylie, S. L. Deardourff and H. E. Kaufman, Science 178, 318 (1972). 48. D. Taylor-Robinson and R. J. Manchee, Nature 215,484 (1967). 49. D. Taylor-Robinson and P. M. FLIIT, Ann. N.Y. Acad. Sci. 225, 108 (1973). 50. A. J. Bateman and A. C. Chandley, Nature (London) 193, 705 (1962). 51. A. Bendich, E. Borenfreund, Y. Honda and M. Steinglass, Arch. Enuiron. Health 19, 157 (1969). 52. J. L. Biedler and H. Riehm, Cancer Rex 30, 1174 ( 1970). 53. E. Borenfreund and A. Bendich, J . Biophys. Biochem. Cytol. 9, 81 (1961). 54. J. C. Leavitt, L. M. Schechtman, P. 0. P. Ts’o, E. Borenfreund and A. Bendich, in “Chemical Carcinogenesis” (P. 0. P. Ts’o and J..DiPaolo, eds.), Pt. B, pp. 639-659. Dekker, New York, 1974. 55. T. E. Wagner, D. R. Mann and R. C. Vincent, J . Erp. Zool. 189, 387 (1974). 56. J. Koch and H. V. Pfeil, FEBS Lett. 24,53 (1972). 57. C. A. Smith and J. Vinograd, J M B 69, 163 (1972). 58. J. C. Rogers, D. Boldt, S. Komfeld, S. A. Skinner and C. R. Valeri, PNAS 69, 1685 ( 1972). 59. S. R. Pelc, Int. Rev. Cytol. 32, 327 (1972).

74

AARON BENDICH ET AL.

60. J. W. Littlefield and C. Basilico, Nature (London) 211, 250 (1966). 61. M. Zedeck, S. S. Sternberg and X. Yataganas, J. Nut. Cancer I&. 53, 719

(1974). M. Abercrombie, I n Vitro 6, 128 (1970). C. Lasne, A. Gentil and I. Chouroulinkov, Nature (London) 247,490 ( 1974). V. H. Freedman and S. Shin, Cell 3,355 ( 1974). I. Berenblum, in “Cellular Control Mechanisms and Cancer” (P. Emmelot and 0. Muhlbock, eds. ), pp. 25%267. Elsevier, Amsterdam, 1964. 66. E. Hecker, Z.Krebsforsch. 65,325 ( 1963). 67. A. B. Mukherjee and J. Krasner, Science 182,68 (1973). 68. G. K. Abelev, Aduan. Cancer Res. 14,295 (1971). 69. E. H. Stonehill and A. Bendich, Nature (London) 228, 370 (1970). 70. P. Alexander, Nature ( London ) 235, 137 ( 1972) . 71. J. H. Coggin and N. G. Anderson, Aduan. Cancer Res. 19, 105 (1974). 72. P. Medawar, Cancer Res. 34,2053 ( 1974). 73. R. C. Mellors, Cancer Res. 28, 1372 (1968). 74. B. Ephrussi, in “Hybridization of Somatic Cells.” Princeton Univ. Press, New Jersey, 1972. 75. A. G. Schwarts, P. R. Cook and H. Harris, Nature NB 230, 5 ( 1971). 76. B. Watson, I. P. Gormley, S. E. Gardiner, H. J. Evans and H. Hams, Exp. Cell Res. 75,401 ( 1972). 77. B. Backay, C. M. Crow, H. Koprowski and W. L. Nyhan, PNAS 70, 1998 (1973). 78. C. M. Crow, B. Bakay, W. L. Nyhan and H. Koprowski, PNAS 70, 2590 (1973). 79. S. Shin, R. Cavena, C. L. Schildkraut, H. P. Klinger and M. Siniscalco, Nature NB 241, 194 (1973). 80. A. Bendich, E. Borenfreund and Y. Honda, in “Informative Molecules in Biological Systems” (L. G. H. Ledoux, ed.), pp. 80-87. North-Holland Publ., Amsterdam, 1971. 81. 0. W. McBride and H. L. Ozer, PNAS 70,1258 (1973). 82. K. Willecke and F. H. Ruddle, PNAS 72,1792 ( 1975). 83. J. W. Burch and 0. W. McBride, PNAS 72,1797 ( 1975). 84. M. Chorgiy, A. Bendich, E. Borenfreund, 0. L. Ittensohn and D. J. Hutchison, J. Cell Biol. 19,71 ( 1963). 85. 0.L. Ittensohn and D. Hutchison, Exp. Cell Res. 55, 149 ( 1969). 86. T. Ebina, I. Kamo, K. Takahashi, M. Homma and N. Ishida, Exp. Cell Res. 62,384 ( 1970). 87. G. D. Burkholder and B. B. Mukheiee, Exp. Cell Res. 61, 413 (1970). 88. H. Kato, K. Sekiya and T. H. Yoshida, Exp. CeZZ Res. 85, 454 ( 1971). 89. T.Sekiguchi and M.-A. Yamada, Exp. Cell Res. 80,233 (1973). 90. M. Bertini, G. Forni and P. M. Comoglio, Clin. Exp. Immunol. 18, 101 (1974). 91. S. MBnard, M. I. Colnaghi and G. Della Porta, Brit. J . Cancer 30, 524 ( 1974). 92. G. P. Di Mayorca, B. E. Eddy, S . E. Stewart, W. S . Hunter, C. Friend and A. Bendich, PNAS 45, 1805 (1959). 93. P. M. Bhargava and G. Shanmugam, This Series 11,103 ( 1971). 94. L. Ledoux ( ed. ), “Informative Molecules in Biological Systems.” North-Holland Publ., Amsterdam, 1971; “Uptake of Informative Molecules by Living Cells.” North-Holland Publ., Amsterdam, 1972. 95. R. R. Dourmashkin and R. M. Dougherty, Exp. Cell Res. 25, 480 ( 1961).

62. 63. 64. 65.

INFORMATION TRANSFER AND SPERM/ SOMATIC CELL INTERACTION

75

96. R. H. Michel, S. J. Pancake, J. Noseworthy and M. L. Karnovsky, J. Cell Biol. 40, 215 (1969). 97. J. W. Heine and C. A. Schnaitman, J. Cell. Biol. 48, 703 (1971). 98. M. D. Schoenberg, P. A. Gilman, V. R. Mumaw and R. D. Moore, Brit. J. Exp. Pathol. 42, 486 ( 1961). 99. K. Wolff and K. Konrad, J . Ultrastruc. Res. 39,262 ( 1972). 100. K. Marushige and J. Bonner, J. Mol. Biol. 15, 160 (1966). 101. M. H. Meisler and R. H. McCluer, Science 154, 896 (1966). 102. A. Hanada and M. C. Chang, Biol. Reprod. 6,300 (1972). 103. E. H. Davidson, “Gene Activity in Early Development.” Academic Press, New York, 1968. 104. J. Brachet and P. Malpoix, Aduan. Morphogenesis 9,263 (1971). 105. M. Nemer, This Series 7,243 (1971). 106. C. J. Epstein, B i d . Reprod. 12, 82 (1975). 107. R. L. Brinster, Biochern. Genet. 9, 187 (1973). 108. D. A. Wright and C. R. Shaw, Biochern. Genet. 4,385 (1970). 109. D. A. Wright and S. Subtelny, Deoelop. Biol. 24, 119 ( 1971). 110. N. B. Hecht, J . Reprod. F e d . 41,345 (1974). 111. E. Premkumar and P. M. Bhargava, Indian J. Biochem. Biophys. 10, 239 (1973). 112. C. F. Millette, W. E. Gall and G. M. Edelman, FP 33, 1395 ( 1974). 113. H. I. Calvin, C. C. Tu and J. M. Bedford, Exp. Cell Res. 81, 333 (1973). 114. Y. KopeEny, Z. Zellfmsch. Mikrosk. Anat. 109, 414 (1970).\ 115. C. Manes, Cancer Res. 34, 2044 ( 1974). 116. C. F. Graham, B i d . Rev. 49,399 (1974). 117. M. Green and G. F. Gerard, This Series 14, 187 ( 1974). 118. S. Spiegelman, R. Axel, W. Baxt, D. Kufe and J. Schlom, Cancer 34, 1406 (1974). 119. L. J. Grady and W. P. Campbell, Nature NB 243,195 (1973). 121. W. A. Nelson-Rees and R. R. Flandermeyer, Science 191, 96 (1976). NOTE ADDEDIN PROOF The likelihood that the SH-2 cell line ( 3 3 a ) (Fig. 7 ) may be a strain of the (human) HeLa cell line has been raised ( 121 ).

This Page Intentionally Left Blank

Studies on the Ribosome and Its Components PNINASPITNIK-ELSON AND DAVID ELSON Biochemisty Depaztment The Weizmann lnstitute of Science Rehmot, Israel

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

. . . . . . .

I. Introduction. 11. Ribosomal Proteins 111. Macromolecular Interactions in the Ribosome IV. Ribosomal Conformation: Large Conformational Changes V. Ribosomal Conformation: Restricted Changes VI. Cation Specificity in the Ribosome. VII. Conclusion. References.

. . . . . . . .

. . . . . . . .

. . . . . . . .

77 78 80 85 91 93

96 97

1. Introduction In dedicating this contribution to Professor Erwin Chargaff on his retirement as Chairman of the Department of Biochemistry at Columbia University, the authors, both former members of the Chargaff Laboratory, do so with feelings of both personal esteem and scientific indebtedness. It was in his laboratory that we both learned about nucleic acids in a milieu where it was natural to believe that the study of the chemistry of biological macromolecules is a necessary pathway toward the understanding of their physiological function. Together with the studies of one of us on polyelectrolyte chemistry and interactions with the late Aharon Katchalsky, this influenced the direction of much of our subsequent work on bacterial ribosomes. We summarize a portion of this work here. When we began to work with ribosomes we did not expect to stay with the subject as long as we have. But it is difficult to leave the ribosome, since it leads into the problem of the relationship between chemical structure and biological function at a level of complexity that appears to be not quite beyond the possibility of solution. The bacterial ribosome is a noncovalent complex made up of three molecules of RNA, over fifty different protein molecules, and certain monovalent and bivalent cations, arranged in two interacting but separable subunits (1-3).Long before this catalog of components was compiled, the structural complexity required of the ribosome was evident from a consideration of its multiple functions. 77

78

PNINA SPJTNIK-ELSON AND DAVID ELSON

The ribosome is the site of protein synthesis. It carries at least two catalytic sites that participate in this process, and also serves as the focus and organizing center of the protein-synthesizing apparatus. As such, it interacts with other macromolecular components of the apparatus : messenger RNA, aminoacyl- and peptidyl-tRNAs, and the numerous protein factors that assist in peptide chain initiation, elongation and termination. Each of these ligands interacts with a specific binding site on the surface of the ribosome in a strictly controlled sequence: binding; acting on or being acted on by the ribosome; in some cases, being transferred to a different site; and eventually becoming detached. The image suggested by such a sequence of interrelated events is of a particle with a constantly changing surface on which functional sites are formed, deformed and replaced by others in a cyclic sequence, with the ligands possibly acting as allosteric effectors. If this is true, workers in the field must deal not only with the problem of how the RNA and protein molecules are fitted together in the functional ribosome, but also with the question of structural flexibility. In summarizing some of our work on tliese problems, we discuss experiments having to do with the nature of the ribosoinal proteins, protein-RNA interactions, conformation and conformational changes, and the specific cations required for the structural integrity of the ribosome.

II. Ribosomal Proteins All the ribosomal proteins of 'Escherichia coli have now been isolated and are being characterized with the techniques of conventional protein chemistry, a major achievement in ribosome research (see refs. 1-3 for reviews). Some years ago, however, the approach to this task was beset with serious difficulties. Among the foremost were the scarcity of information about the chemical properties of the ribosomal proteins, and their insolubility. When detached from the ribosomal RNA, the proteins precipitated. They could be redissolved, or precipitation prevented, only with denaturing solvents such as acid, alkali, detergents or urea, a fact that inhibited many protein chemists from entering the field. The solubility problem was substantially clarified during the course of some of our early experiments on the nature of the ribosomal proteins. Drawing a parallel with the behavior of natural and synthetic polyelectrolytes ( M ) it, was postulated that the ribosomes might contain both basic and acidic proteins that, being oppositely charged at neutral pH, would form insoluble electrostatic complexes ( 7). This was verified experimentally. When ribosomes were digested with ribonuclease in a neutral buffer of high ionic strength, to reduce electrostatic interactions, the liberated proteins remained completely soluble, the first time this

THE RIBOSOME AND ITS COMPONENTS

79

had been accomplished at neutral pH without recourse to detergents, urea etc. ( 7 ) . The solubilized proteins were fractionated by stepwise salting out ( 8) and ion-exchange chromatography (9). Electrophoretically distinct fractions were obtained, among them acidic and basic fractions that were soluble at neutral pH and low ionic strength when separate, but that precipitated when mixed. About 15% of the protein was estimated to be acidic or neutral, the rest being basic (9). During these experiments, anomalies in the behavior of certain protein fractions suggested the presence of nonelectrostatic aggregates among some of the basic proteins (9). This is also characteristic of the histones ( l o ) , which had been noted to show similarities to the ribosomal proteins (11). Accordingly, a fractionation technique that had been used to separate aggregating from nonaggregating histones ( 1 0 ) was adapted to the ribosomal proteins (12). Among others, there were obtained two highly water-soluble heterogeneous basic fractions that differed markedly in their electrophoretic behavior in gels. One fraction entered the gel cleanly and completely, giving the same pattern whether urea was present or not. The pattern of the other fraction was greatly affected by urea, with much material failing to enter the gel unless urea was present ( 1 2 ) .These experiments showed that, like the histones, the ribosomal proteins contain aggregating and nonaggregating basic fractions, and that there are nonelectrostatic as well as electrostatic interactions among the ribosomal proteins in solution. Taken together, these early studies supplied considerable information on the ribosomal proteins as a whole. They provided one of the early demonstrations that the proteins are highly heterogeneous, and showed that ribosomes contain acidic proteins in addition to the preponderance of basic proteins. The estimate of the amount of nonbasic protein was later shown to be quite accurate when, in other laboratories, the individual proteins were isolated and characterized (1-3). The troublesome solubility problems of the proteins could be understood in terms of the interactions shown to take place in mixtures of the total ribosomal proteins. This in turn suggested ways of solubilizing and proteins and understanding the solubilizing effects of high salt, extreme pH, ionic detergents, urea and group fractionation of the proteins and helped to bring the field to a point where the techniques of conventional protein chemistry could be applied. The solubilizing effect of concentrated neutral salt was later exploited in the fundamental experiments that led to the reconstitution of active ribosomes from isolated proteins and RNA ( 1 3 ) ; some of the group fractionation procedures developed have been used as preliminary steps in isolating the individual proteins (e.g., 14, 1 5 ) .

80

PNINA SPITNIX-ELSON AND DAVID ELSON

The extensive protein-protein interactions observed in solution seem likely to be relevant to the structure of the ribosome. When the proteins occupy their specific positions on the ribosomal RNA, the same types of protein-protein interaction would be expected to take place in an organized way, and may be required for the formation and stability of the functional conformation. Another point of interest is the resemblance between the ribosomal proteins and the histones. Both are heterogeneous, predominantly basic (and therefore different from the bulk of the cellular proteins), and can be separated into aggregating and nonaggregating species by the same procedure. The resemblance is probably not fortuitous, since both must condense on a backbone of nucleic acid to form a biologically active, and therefore structurally specific, complex. While the structures are different, it is likely that the same kinds of macromolecular interaction are used to make them. More than once we have found techniques of histone chemistry to be applicable to the ribosomal proteins.

111. Macromolecular Interactions in the Ribosome RNA-protein and protein-protein interactions constitute a major item in the study of ribosomal structure. In one group of experiments, we attempted to gain information on these macromolecular interactions and the chemical bonds involved in them by examining different methods for separating the ribosomal components. The detachment of proteins from the ribosome was followed, with particular attention paid to the effects of three types of treatment: ( a ) exposure to high concentrations of neutral salt, which primarily affects electrostatic bonds; ( b ) exposure to urea, whose primary effect is on hydrogen and hydrophobic bonds; and ( c ) variation in the concentration of Mg2+ ions, known to affect the overall ribosomal conformation, removal causing a large drop in sedimentation coefficient (16, 17). With a structure as complex as the ribosome, the total effect of a reagent cannot be ascribed to a single kind of chemical action, since the primary effect on one type of interaction may cause structural alterations with secondary effects on other kinds of interaction. We were interested, therefore, not only in trying to assess and compare the primary effects of different treatments, as far as that is possible, but also in observing any interplay and mutual influence among different kinds of interaction in the ribosome. High concentrations of neutral salt detach large amounts of protein from ribosomes ( 18-24), certain proteins being detached preferentially (23-26). The amount of protein detached increases with increasing NaCl concentration, beginning at a relatively low salt concentration (27)

THE RIBOSOME AND ITS COMPONENTS

TABLE I DETACHMENT OF RIBOSOMAL PROTEINS AS FUNCTION O F NACL CONCENTRATION^^^

A

Composition of particles

NaCl conc.

(MI 0.0 0.15 0.3 0.5 1 2 a

Protein RNA Protein (%I (%I RNA 36.5 35 32 30 25 25

63.5 65 68 70 75 75

0.57 0.54 0.47 0.43 0.33 0.33

Protein detached

(%I 0 5 18 25 42 42

From Spitnik-Elson and Atsmon (27).

* Ribosomes (1 mg/ml) were left overnight in ice in a medium containing 2 mM MgCl, 10 mM TrisCl (pH 7.4), and NaCl as indicated. The solutions were centrifuged 5 hours at 50,000 rpm, and the nucleoprotein pellets were analyzed for RNA and protein.

(Table I ) . Such observations showed the presence of electrostatic bonds, as expected in a complex containing both acidic and basic components. These nonspecific bonds, while probably helping to stabilize the ribosome, cannot confer on it the unique conformation it must have, and they presumably operate in conjunction with other, more specific bonds. Changes in the effectiveness of salt in detaching proteins may also be used as an indicator of structural alterations in the ribosome. The complete removal of Mg2+has far reaching effects on the structure of the ribosome, among them a drastic lowering of the sedimentation constant, termed “unfolding” ( 16, 1 7 ) . Circular dichroism studies indicate that the effect is exerted on the RNA moiety of the ribosome, with little or no visible change in the conformation of the proteins (28-30; see below). That is, Mg?+ ions appear to stabilize the ribosome by stabilizing the RNA conformation. As is seen in Table 11, a fixed concentration of salt removes progressively more protein as the MgZ+ concentration is lowered, i.e., as the ribosomal structure is progressively destabilized and eventually disrupted. These results indicate that changes in the conformation of the ribosomal RNA influence the binding of the proteins. Treatment with 6 M urea ako deranges the structure of the ribosome and causes a significant drop in its sedimentation coefficient but, in contrast to Mg“-depletion, its primary effect appears to be on the proteins, which are extensively altered (29, see below). Here, too, the proteins are detached by salt at lower ionic strength than when the ribosome

82

PNINA SPITNIK-ELSON AND DAVID ELSON

TABLE I1 OF RIBOSOMAL PROTEINS AS A FUNCTION DETACHMENT OF SALTA N D M G ~ CONCENTRATION + Composition of particles Mg'+ conc. (mM)

10 2 1 0.1 0 0 0

Salt

Salt conc. (M)

Original ribosomes NaCl 0.5 NaCl 0.5 NaCl 0.5 NaCl 0.5 NaCl 0.5 NaCl 1 LiCl 2

Protein (%)

RNA

(%I

Protein RNA

40 36 31 31 26.5 20 11 5.5

60 64 69 69 73.5 80 89 94.5

0.67 0.56 0.45 0.45 0.36 0.25 0.12 0.058

Protein detached

(%I 0 16 33 33 46 63 82 91

From Spitnik-Elson and Atsmon (27). Procedure as in Table I. Mg'+ concentration was reduced to 0 by a preliminary dialysis against EDTA. b

is intact. Table I11 illustrates this. It also shows that increasing concentra-

tions of salt release the various species of protein in a different order when urea is present than when it is absent. For example, proteins S7 and S8 are among the four proteins most resistant to salt detachment in the absence of urea, but among the earliest to be detached when urea is present. This shows that the binding of the ribosomal proteins is at least partly dependent on urea-sensitive specific bonds, whether protein-protein, protein-RNA, or intramolecular within protein molecules. Stronger evidence for the presence of these specific (i.e., nonelectrostatic) bonds was obtained with a newly developed solid-phase technique that was being tested for the fractionation of ribosomal proteins (33, 34). The technique was suggested by reports that, although native ribosomes can be adsorbed on columns of DEAE-cellulose and eluted with salt without impairing their activity (35,36), purified ribosomal RNA cannot be eluted except by deionizing the ion exchanger with aIkali (37). It was reasoned that if the native structure of the ribosome were disrupted, the less structured nucleoprotein complex might behave like RNA, and this was found to be the case (33). Ribosomes were unfolded by Mg2+-depletionand adsorbed on DEAE-cellulose. The proteins remained bound to the adsorbed RNA at low ionic strength and were quantitatively eluted with a salt gradient, while the RNA remained adsorbed until the column was washed with alkali. As in solution, 6 M urea reduced the salt concentration required to detach the proteins.

THE RIBOSOME AND ITS COMPONENTS

83

TABLE 111

30 s ~ t I n 0 S O M A L S U B U N I T EFFECT OF 6 M U R E . ~ ~

O F P R O T E I N S FROM T H E

I)ICT.\CHMI,:NT

IIY S A L T : THE

Proteins remaining attached to RNA Urea absentb

Proteind

LiCl (hl):

Urea present”

1 100 1 170 Mg(OAc)2 (mM). 1 NH,CI (mM): - 12 - 500 0.8 1.15 2 . 0 3 . 0 Ionicstrength: 0.01 0.02 0 . 3 1 0.51 0..52

s1 s2 53 s4 S5 S6 s7 S8 s9c s10 Sll’ 512 813 514 515 S16 S17 S18 s19 s20

+ + + + It

+ + + + It

R

I

+

-

t

-

-

+ + - - + + + + + + + + + + R R R -

+- + - + + + + -

-

-

-

-

~

-

+

-

-

-

t

R

R

-

-

-

R

-

t

-

+ + + + +

-

+

R

-

-

+ + + + + n

+ + R + + -

+ + + -

-

+

+

-

l

+ + + + + + + + + + + + + +

+ + + + + + +

521

-

+ I

- I - + - - -

+

+ + + + ~

-

+ + + +

-

+ + +

-

-

-

-

~

The table shows the “core” proteins, i.e., those not detached from the RNA. Symbols: present; 13, present in reduced amount; -, absent. * From Schreiner and Nierhaus (31). Medium: 10 mM Mg(OAc)2, 10 mM TrisCl (pH 7.4), LiCl as indicated. c From Rpitnik-Elson, Greenman and Abramovitz (99). Medium: 6 M urea, 10 mM TrisCl (pH 7.4), Mg(0Ac)l and NH,CI as indicated. Proteins S9 and S11 are not separated in the analytical system employed. d Proteins numbered according to Wittman et al. (32).

+,

An elution pattern is shown in Fig. 1. A partial fractionation of the ribosomal proteins was achieved in these preliminary experiments ( 33, 3 4 )* Under these conditions, where the negative potential of the RNA is largely neutralized by the positively charged groups of the ion exchanger, proteins are detached at ionic strengths much lower than those

84

PNINA SPITNIK-ELSON AND DAVID ELSON

TUBE NUMBER

FIG. 1. Elution of proteins from unfolded 30 S ribosomal subunits adsorbed on DEAE-cellulose. Recovery of protein: zone 1: 1 mM phosphate 6 M urea, 34%; zone 2, NaCl gradient in 1 mM phosphate 6 M urea, 63.5%;zone 3, 0.5 M NaOH, about 2%.The absorbance of the NaOH eluate at 230 nm was due to the RNA. There was no RNA in the earlier fractions. From Spitnik-Elson and Greenman ( 3 4 ) .

+

+

required when the unfolded ribosomes are free in solution (compare Fig. 1with Table 111). With the electrostatic “background reduced, it became easier to detect and demonstrate other types of interaction. A group of proteins (about one-third of the total, Fig. 1) was detached from the adsorbed ribosome by 6 M urea at low ionic strength, although they were not detached on adsorption in the absence of urea or by urea in the absence of the ion exchanger. This shows that under the conditions employed the binding of these proteins is directly dependent on the kinds of interaction influenced by urea (34). Taken together, the detachment experiments indicate that many factors are involved in ribosomal protein binding. Reagents whose primary action is to weaken either one or another type of interaction affect the overall ribosomal structure, and changes imposed on the structure make proteins more easily detached from the RNA. There appears to be a complex interplay of interdependent interactions-both electrostatic and nonelectrostatic-that form and stabilize the structure of the ribosome and are, in turn, dependent on it.

THE RIBOSOME AND ITS COMPONENTS

85

IV. Ribosomal Conformation: large Conformational Changes The preceding sections are concerned with the ribosomal proteins and their detachment from the ribosome by salt; they show that the proteins become less firmly bound when the overall ribosomal structure is disrupted, as by Mg2+-depletion(16, 17) or urea (29). In this section we summarize certain of our observations on the changes produced in the 30 S ribosome by these disruptive treatments. We attempted not only to follow the overall changes in the ribosome, but also to learn something of the behavior of its RNA and protein moieties during these changes. Several techniques were employed, among them the measurement of sedimentation velocity, optical rotatory dispersion, circular dichroism, and the intrinsic tryptophan fluorescence of the ribosomal proteins. As mentioned above, the removal of Mg2+ions causes a marked decrease in the sedimentation constant of the ribosome (IS,17). The decrease is largest at low ionic strength, where it may drop from 30 S to 5 S or even lower. Since little or no protein is detached under these conditions (IS), this behavior signals a conformational change without a decrease in mass. The sedimentation constant of such unfolded ribosomes is sensitive to changes in ionic strength. It increases with the salt concentration (even though some of the proteins are detached by the added salt), showing the residual particle to become more compact, thus exhibiting something of the behavior of a randomly coiled polyelectrolyte. To examine this behavior more closely and also to observe the RNA in parallel (28), ribosome samples were equilibrated with a series of buffers, one of which contained MgZ+while the others lacked Mgz+and were of diminishing ionic strength. Sedimentation constants and optical rotatory dispersion (ORD) spectra were determined, the first to follow changes in the overall shape of the ribosome, and the second to observe changes in the base-pairing and base-stacking of the RNA. The details and results are shown in Table IV and Fig. 2. Both parameters changed as Mgz+was removed and ionic strength decreased. The sedimentation constant fell, showing the ribosome to become less compact, and the ORD spectrum moved to longer wavelengths, indicating the disruption of base-paired regions in the RNA. There was little evidence of diminished base-stacking [the magnitude of the Cotton effect remained relatively constant (28)]. The changes were closely parallel and were not continuous, but occurred in two discrete steps. The first transition, from a compact to a partly unfolded form, took place when Mg2+was removed at constant ionic strength. After this there was relatively little change as the salt concentration was lowered until, at a low ionic strength,

PNINA SPITNM-ELSON AND DAVID ELSON

86

TABLE I V EFFECTOF DIFFERENT MEDIAO N SEDIMENTATION VELOCITY AND OPTICALROTATORY DISPERSION (ORD) O F THE 30 s RIBOSOMAL SUBUNIT5 Concentration (mM)

x Sample Mg(0Ac)l NHICl 1 2 3 4 5 6c

1 -

100 100 50 20 -

-

-

TrisCl EDTA (pH 7.4) (pH 7.4) 10 1 1 1 1 1

1 1 1 1 -

Ionic strength

a

0.111 0.107 0.057 0.027 0.007 0.0008

30 20

~

(S)

19

17 11 8.5

~ crossingb . ~

(nm) 265.6 266.7 267.0 267.0 270.0 270.0

From Eilam and Elson (88). *The wavelength at which optical rotation was zero. This is indicative of the position of the entire ORD curve. Previously dialyzed against the medium of sample 5. 30

-

P

-265

- 266

-

20

267

3

-268

g 0

a 0

-269

4

270

0

0.05

0.10

IONIC STRENGTH

FIG.2. 30 S ribosomes in different media: sedimentation constant (s) and wavelength of zero rotation ( A crossing) as functions of ionic strength and composition. Data taken from Table IV. From Eilam and Elson (28).

a further unfolding occurred. The results show the RNA of the 30 S ribosome to contain two classes of base-paired regions, one more stable than the other. The dissociation of each class is correlated with an overall conformational change in the ribosome.

87

THE RIBOSOME AND ITS COMPONENTS

The above experiment gave no information on protein conformation. This was obtained through measurements of circular dichroism ( C D ) ( 2 8 3 0 ) and tryptophan fluorescence (30). The CD spectrum of ribosomes is convenient for sorting out the contributions of the RNA and protein moieties, since it contains a positive region at above 250 nm which is due to the RNA, while at about 220 nm the contribution of the RNA is negligible and the negative response seen there is almost entirely due to the proteins (28, 29)-mainly, it is believed, to a-helical structures. The CD spectra shown in Fig. 3 confirm that the RNA conformation changes when the Mgz+-depleted ribosome unfolds, but show no significant difference in protein conformation when compact and unfolded ribosomes are compared. The information obtained from fluorescence measurements is of a different kind, since tryptophan fluorescence is sensitive to the immediate environment of the fluorophore. Tryptophan in aqueous solution has a characteristic emission spectrum with a maximum at about 350355 nm (38). In less polar solvents the maximum shifts to lower wavelengths (39, 40). Proteins in aqueous media typically show a tryptophan emission maximum below 350 nm, the exact position differing from protein to protein (39, 41). This variable blue shift is considered to result from the fact that different tryptophan residues are situated differently on or in the protein and are in different environments (30, 40, 42). The effect of Mg2+-depletionand unfolding of the 30 S ribosome on the fluorescence of its tryptophan residues is shown in Fig. 4 and Table V. Unfolding is seen to cause a spectral shift to longer wavelengths t6 *4

+2

0 -2 -4

-6 220

240

269

280

m

Wovelength (nm)

FIG. 3. Circular dichroism spectra of folded and unfolded 30 S ribosomes. ---, Folded ribosomes in 100 mM NHCI, 1 mM M ~ ( O A C )10 ~ , mM TrisCl (pH 7.4); , unfolded ribosomes in 10 mM Tris (previously dialyzed against EDTA); -, refolded ribosomes (unfolded ribosomes dialyzed against the medium of “the folded ribosomes”). From Spitnik-Elson et al. (30).

._._

88

PNINA SPITNIK-ELSON AND DAVID ELSON

t

t 01’

I

320

’

I

340

’

I

360

’

I

380

Emisson Wovelength (nm)

FIG.4. Intrinsic tryptophan fluorescence of folded and unfolded 30 S ribosomes. Media as in Fig. 3. The curves are normalized to the same maximum intensity. ---, Folded ribosomes; , unfolded ribosomes; -, refolded ribosomes. Excitation at 290 nm. From Spitnik-Elson et d.(30).

.-.-

TABLE V TRYPTOPHAN FLUORESCENC~ INTENSITY OF 30 S RIBOSOMES: EFFECTOF UNFOLDINCP-~

Parameter 82O,w(s)

Relative fluorescence intensity”

Ribosome C conc. A B Partly Expt. (pglml) Folded Folded unfolded 1 2

-

1 1 2 2

57.5 28.8 57.5 28.8

29 29

29 30

15

100

93 95 97 109

84 86 95 70

100 100 100

D D+A Unfolded Refolded 2.6 2.8 57 68 68 59

-

108 88 -

From Spitnik-Elson et al. (30). “edia: A: 100 mM NH,Cl, 1 mM Mg(OAc)2, 10 mM TrisCl (pH 7.4); B: 1 mM Mg(OAc)2, 10 mM Tris; C: 10 mM Tris (previously dialyzed against EDTA); 1): 1 mM Tris (previously dialyzed against EDTA. D -+ A: 1) dialyzed against medium A. c Maximum intensities normalized to value in medium A. 0

89

THE RIBOSOME AND ITS COMPONENTS

and a reduction in fluorescence intensity, indicating a change in the microenvironment of tryptophan residues. The nature of the change is consonant with, though not proof of, increased exposure to the aqueous medium. In summary, our observations indicate that the following structural changes take place when the 30 S ribosome is depleted of Mg2+ ions. Base-paired regions in the RNA become less stable and dissociate. There is a parallel change in the shape of the ribosome, the particle becoming less compact. These changes are not continuous, but take place in discrete steps as the ionic strength is lowered. No change was detected in the internal structure of the individual proteins. However, there was evidence of a change in the relationship of certain proteins to the other proteins and the RNA and solvent, which together constitute their microI

1

1

I

I

I

I

I

I

I

I

I

I

I

I

Wavelength I n m )

FIG.5. Circular dichroism spectra of 30 S ribosomes: effect of 6 M urea. 0, Native ribosomes in 5 mM Mg (OAc)Z, 10 mM TrisCl ( p H 7.4). A, Ribosomes h the same solvent plus 6 M urea. 0 , Ribosomes that had been in the same solvent plus urea, after dialysis to remove urea. 0 , Ribosomes in 1 mM Tris, 1 mM EDTA. The curves relevant to the text are 0 and A. From Spitnik-Elson et al. (29).

90

PNINA SPITNIK-ELSON AND DAVID ELSON

Emissm Wovelength (nm)

FIG. 8. Intrinsic tryptophan fluorescence of 30 S ribosomes, free ribosomal proteins, and tryptophan in absence ( A ) and in the presence ( B ) of urea. Media: ( A ) 350 mM NHJ.21, 20 mM Mg(OAc)*, 10 mM TrisCl (pH 7.4). ( B ) ,Same plus 8 M urea. Symbols and concentrations: Ribosomes (---): ( A ) 57.5 pg/ml; (B), 50 pglml; containing 23 and 20 pg protein/ml. 30 S ribosomal proteins (-): (A) 13.5 @ml; (B) 27 &ml. Tryptophan (.--): the free amino acid, 2.0 pg/ml. All curves are normalized. Excitation at 290 nm. From Spitnik-Elson et d. (30).

environment. It seems likely that the primary effect of Mg2+-depletion is the destabilization of the ribosomal RNA. As noted above, urea also affects the overall structure of the ribosome (29). Subunits that sediment at 29-30 S in 1 mM Mg( OAc)* 10 mM TrisCl ( p H 7.4) sediment at 16-17 S when the same medium is made 6 M in urea. In contrast to Mg--depletion, urea causes extensive conformational alterations in the ribosomal proteins (Figs. 5, 6). A major effect of 6 M urea on the CD spectrum (Fig. 5 ) is a large decrease in the negative response around 220 nm, indicating the destruction of a-helical conformation in the proteins (29). Figure 6 shows the effect of 6 M urea on the fluorescence spectrum of 30 S ribosomal proteins, both in the ribosome and free in solution (30). In the absence of urea (Fig. 6A), the emission spectrum of the free proteins is shifted considerably to the blue relative to the spectrum of free tryptophan, while that of the intact ribosome is shifted slightly farther to the blue and is narrower than the free protein spectrum, owing to lower emission at the longer wavelengths. This indicates that tryptophan residues in the proteins are in an environment less polar than the aqueous medium.

+

THE RIBOSOME AND ITS COMPONENTS

91

When they are incorporated into the structure of the ribosome, a further change takes place that causes an additional blue shift. The addition of 6 M urea (Fig. 6B) had no effect on the fluorescence of tryptophan but produced a large red shift in the emission spectrum of the proteins, whether free or in the ribosome, so that they became identical with that of the free amino acid. This was seen not only in the high salt medium of this experiment, where many proteins would be detached from the ribosome (Table 111), but also in low-salt media, which detach very little protein. These observations corroborate those made with CD, and show that 6 M urea extensively alters the ribosomal proteins, fully exposing their tryptophan residues to the external medium, as has been observed with nonribosomal proteins (39, 41 ). Very little protein is detached at low ionic strength, but the ribosomal structure is disrupted (29). These observations indicate that the structural stability of the ribosome depends on the structural integrity of both of its macromolecular components, RNA and protein. When either is disarranged-the RNA by Mg2'-depletion or the protein by exposure. to urea-the ribosomal conformation is disrupted ( 29).

V. Ribosomal Conformation: Restricted Changes In addition to the gross and sometimes destructive conformational changes discussed above, more restricted and subtle alterations can be induced in the ribosome, and reversed, under conditions consonant with ribosomal activity in uitro. It seems likely, for example, that the ribosome undergoes a series of conformational transitions during its functional cycle. Although not extensive, evidence for this exists, particularly with respect to the closely related phenomenon of ribosome-antibiotic interactions ( 4 3 5 3 ) . By relatively mild manipulation of the medium and temperature, ribosomes can be made to undergo reversible transitions that directly affect their functional activity and their ability to interact with a number of antibiotics (43, 44, 48, 54, 55). The phenomenon can be brieffy described as follows. When ribosomes or their individual subunits are depleted of K' and NH,',, they lose the ability to perform a number of ribosomal reactions and to bind certain antibiotics. The 30 S subunit also becomes inactive when the Mg2' ion concentration is reduced to about 1 mM or less. Activity can be restored by replacing the depleted cations, the process showing specific monovalent cation requirements. Either NH,', K+,Rb+ or Cs' ions preserve or restore activity; Na+ and Li+ do not. The rate of reactivation is strongly dependent on temperature

92

PNINA SPITNIK-ELSON AND DAVID ELSON

and the concentration of the activating cation, and may vary from almost imperceptibly slow ( O'C, low cation concentration) to almost instantaneous ( 40°C, high cation concentration). During the transition from active to inactive, or vice versa, structural changes take place in the ribosome. This has been shown by the demonstration that different sets of sulfhydryl proteins in active and inactive 30 S subunits are able to react with the sulfhydryl reagent N-ethylmaleimide (53). These observations point to a significant flexibility in the ribosome and suggest that it may be an inherent requirement for ribosomal function. We wished not only to try to obtain physical evidence to correlate conformational alterations with changes in activity, but to do so with a technique that would allow us to follow the transition while it takes place; an optical technique, for example. Preliminary experiments with ultraviolet absorption and optical rotatory dispersion did not yield promising results. However, fluorescence measurements did show a difference between active and inactive 30 S ribosomal subunits. The experiments were performed by inactivating 30 S ribosomes at low MgZ+concentration. The ribosomes were then diluted into a cold, high Mg" buffer that could support reactivation. Part of the solution was kept cold and inactive, while another part was heated to cause activation and was then chilled. This gave two ribosome solutions, one active and the other inactive, but otherwise identical. Measurement of the intrinsic tryptophan fluorescence showed the two ribosome solutions to differ, the emission intensity of the active subunits being lower by about 101% (30). A similar result was obtained with a different fluorophore, the covalently bound fluorescent sulfhydryl reagent N- (3-pyrene)maleimide ( 5 6 ) ; activation again caused a drop in fluorescence intensity of 1620% (57). In order to examine the activation process in greater detail, kinetic measurements were made during the heat treatment (Fig. 7). Two ribosomal parameters were monitored in parallel : aminoacyl-tRNA binding activity, and the fluorescence of the pyrene fluorophore (57). Each followed a different time course. The restoration of biological activity followed first-order kinetics from the beginning, as previously observed ( 5 5 ) . The fluorescence change was slower and more complex, showing an initial lag before assuming first-order kinetics. The kinetic differences have more than one possible explanation. For one, it must be remembered that the ribosomal population is divided into two groups with respect to aminoacyl-tRNA binding ability. At the highest activation achieved in this experiment, only about one out of each four ribosomes could bind the substrate. The rest were incapable of binding, although they still undergo a structural alteration on heating (53). This heterogeneity may have been reflected in the kinetics. How-

93

THE RIBOSOME AND ITS COMPONENTS

0

0

20

TIME (MIN 1

FIG.7. Fluorescence as a monitor of the heat activation of 30 S ribosomes. Ribosomes labeled with N-(3-pyrene )maleimide were inactivated by dialysis against 100 mM NHXl, 1 mM Mg( OAC)~,10 mM TrisCl ( p H 7.4), 6 mM 2-mercaptoethanol. The cold medium was changed to 300 mM KCI, 20 mM Mg(OAc)*, 10 mM Tris, 6 mM 2-mercaptoethanol. The mixture was incubated at 30"C, and aliquots were periodically removed, chilled at 0°C and taken for the measurement of fluorescence (0-0, F, excitation at 330 nm, emission at 375 nm) and activity [O-0, A, the binding of phenylalanyl-tRNA in the presence of poly(U)]. Subscripts: o, a t 0 time; a , after 60 minutes, none, at time t. First-order rate constants: activity, 0.14 per minute; fluorescence, 0.08 per minute. From Schechter and Elson (57).

ever, whether or not this is the case, it is probable that the fluorescence change actually reflects several conformational changes with different rates, not all of which are involved in re-forming the aminoacyl-tRNA binding sites. This is in accord with earlier evidence that heat activation is a complex process, with different activities restored at different rates (45, 48). These experiments have borne out our expectations that fluorescence might offer a sensitive physical technique to detect and monitor nondisruptive conformational changes in the ribosome. It may be possible to do this while a ribosomal reaction is taking place in the cuvette, and this may be of use in correlating conformational transitions with biological function.

VI. Cation Specificity in the Ribosome Since the earliest studies on isolated ribosomes, it has been recognized that bivalent and monovalent cations play a critical role in main-

94

PNINA SPITNIK-ELSON AND DAVID ELSON

taining the structure of the ribosome and modulating ribosomal interactions between subunits and with other macromolecules. It further became evident that a certain specificity exists, and in most work with ribosomes the cations Mg2+and K+or NH,' are used. The few systematic studies made have served to establish and partly define this cation specificity. A clear demonstration of monovalent cation specificity in maintaining ribosomal structure was provided by the previously mentioned experiments on the detachment of ribosomal proteins by salt (27). Table VI shows a comparison of the effects of NH4CI and NaCl. When Mg2+ ions were present and the ribosomes were structurally intact, high concentrations of NaCl removed large portions of the ribosomal protein. However, the same concentrations of NH4Cl detached none of the proteins (NH,*Clis commonly used at these concentrations to purify active ribosomes). This is clearly a specific cation effect. The specificity vanished after the ribosomal structure had been disarranged by Mg?+-depletion, whereupon NH,Cl became as effective as NaCl in detaching proteins. TABLE VI DETACHMENT OF RIBOSOMAL PROTEINS BY SALT: COMPARISON OF NaCl A N D NH4CIo Composition of particlesb Salt conc.

(MI

Salt

Protein Protein RNA Protein detached (%I (%I RNA (%)

Intact ribosomesc

-

-

NH4Cl NaCl

0.5 1. 0 0.5 1 .0

Unjolded ribosomesd NH&l 0.5 1.0 NaCl 0.5 1 .0 ~

~

~~~

37 37 37.5 30 2.5

63 63 62.5 70 75

0.59 0.59 0.60 0.43 0.33

0 0 0 27 44

40 24 12 19 12

60 76 88 81 88

0.67 0.32 0.14 0.24 0.14

0 52 79 64 79

~~~~

~

~

From Spitnik-Elson and Atsmon (27). *Procedure as in Table I. The particles contained the RNA plus the proteins that were not detached from it. Medium: 2 mM MgC12, 10 mM TrieC1 (pH 7.4), salt as indicated. The ribosomes were depleted of Mgo+. Medium: 10 mM Tris, salt as indicated.

THE RIBOSOME AND ITS COMPONENTS

95

Under these conditions, the detachment of ribosomal proteins appears to be a function only of ionic strength. However, in order for detachment to take place, the structure of the ribosome apparently must be destabilized, and this is where the cation specificity appears to be expressed. The intact ribosome requires NH,+, and Na' cannot substitute for it. Another example is found in Table 111, which shows that Mg2+and NH,+ salts, both of which contribute to the stability of the intact ribosome, strip off proteins when the ribosomal structure is deformed by urea (29). Mg" is somewhat more effective, but the major factor in determining the amount of protein detached is ionic strength and not the nature of the cation. Monovalent cation specificity is seen again in the inactivation-reactivation transitions ( 4 8 ) . As noted above, ribosomes absolutely require certain monovalent cations for activity and for the maintenance of a conformation that supports activity. When they are replaced by other monovalent cations, the conformation changes and activity disappears. The ions shown to maintain or restore activity are NH,', K', Rb+ and Cs'; Na' and Lit are ineffective. Morris and his colleagues have shown that a comparable specificity exists for bivalent cations ( 5 8 ) . Among other things, they investigated the effect on the ribosome of replacing Mg2+ions with MnZ+or bivalent alkaline-earth cations, none of which inhibit ribosomal activity when Mg"+ ions are also present. In the absence of Mg2+ions, some of these bivalent cations could support ribosomal activity, while others could not. In those cases where activity was lost, the ribosomes underwent physical changes, becoming more susceptible to ribonuclease digestion and showing a reduced sedimentation constant and an increased viscosity. These physical changes did not occur in the presence of those cations that preserved activity. This demonstrates a connection between conformation and activity, and shows them to be dependent on specific bivalent cations. Table VII summarizes some of our findings on monovalent cation specificity and those of Morris et al. (58) on bivalent cation specificity. In both cases, there is a clear group specificity. Each type of cation is divided into two groups, one that supports activity and appears necessary for the maintenance of a ribosomal conformation consonant with activity, and a second group that can support neither activity nor the conformation required for it. When members of the first group are removed or replaced by members of the second group, the ribosome undergoes a discernible conformational change and becomes inactive. There appears to be a correlation with the size of the cation (Table MI). The effective bivalent cations are those with a crystal radius of about 1

96

PNINA SPITNIK-ELSON AND DAVID ELSON

TABLE VII CATION SPECIFICITY I N SUPPORTINQ RIBOSOMAL ACTIVITY Activity" Ionic crystal radius Cation

(A)

Monovalent cations" Li+ 0.68 Na+ 0.97 K+ 1.33 NHI+ 1.43 Rb+ 1.47 cs+ 1.67 Bivalent cation& Mg*+ 0.66 Mn2+ 0.80 Ca2+ 0.99 Sr*+ 1.12 Bat+ 1.34

50 S 30 S subunit subunit -

+ + + + + + + f

-

-

-

+ + + + + + -*

-

See Zamir et al. (48, 66) and Miskin et al. (64).

* See Weiss el al. (68). +, Active; f,partly active; -, inactive.

A or less ( 5 8 ) ; the effective monovalent cations have crystal radii larger than 1A. It is di5cult to reconcile these properties with a scheme in which the cations act only by contributing to the ionic strength of the medium. Salts are undoubtedly required to neutralize and screen the charged groups of the ribosomal RNA and proteins, making it possible for nonelectrostatic forces to participate effectively in shaping the conformation of the ribosome and its macromolecular components. This would not, however, explain the sharp cation specificity and its apparent correlation with ionic radius. It is reasonable to consider seriously the idea that a certain number of the required cations may occupy specific sites in the ribosome and that their presence may be required for the maintenance of the functional conformation, in analogy to the metal-requiring proteins. If this is 'so, then specific mono- and bivalent cations may be considered to be essential structural components of the active ribosome, not less than the ribosomal RNA and proteins.

VII. Conclusion The functional ribosome is a flexible structure formed and stabilized by the delicately balanced interplay of various kinds of interaction

THE RIBOSOME AND ITS COMPONENTS

97

among its components, which include R N A , proteins and specific monovalent and bivalent cations. There appears to be a reciprocal conformational relationship between the total ribosome and its macromolecular components. Not only does the structure of the ribosome require the conformational integrity of the macromolecular components, but the functional conformation of at least some of these macromolecules appears to be formed and stabilized only when the molecule is incorporated into the ribosome. Some of the specific cations also appear to be essential structural components; as yet, virtually nothing is known of this aspect of ribosome chemistry.

ACKNOWLEDGMENT Partial support from the US.-Israel Binational Science Foundation (Grant No. 394) is gratefully acknowledged.

REFERENCES 1. M. Nomura, A. Tissieres and P. Lengyel, eds., “Ribosomes.” Cold Spring Harbor Lab., Cold Spring Harbor, New York, 1974. 2. H. G. Wittmann and G. Stoffler, in “The Mechanism of Protein Synthesis and Its Regulation ” ( L . Bosch, ed.), pp. 285-351. North-Holland Publ., Amsterdam, 1972. 3. C. G. Kurland, ARB 41, 377 (1972). 4. P. Spitnik, A. Nevo and A. Katchalsky, Bull. Res. Council Isr. 4, 318 (1954). 5. A. Katchalsky and P. Spitnik, Coll. Int. C.N.R.S., Sur Mmromol. 57, 103 (1955). 6. P. Spitnik, R. Lipshitz and E. Chargaff, JBC 215, 765 (1955). 7. P. Spitnik-Elson, BBA 55, 741 (1962). 8. P. Spitnik-Elson, BBA 74, 105 (1963). 9. P. Spitnik-Elson, BBA 80, 594 (1964). 10. E. W. Johns, BJ 92,55 (1964). 11. C. F. Crampton and M. L. Petermann, JBC 234,2642 ( 1959). 12. P. Spitnik-Elson and S. Zingher, BBA 133,480 (1967). 13. P. Traub and M. Nomura, JMB 40, 391 ( 1969). 14. E. Kaltschmidt and H. G. Wittman, Biochimie 54, 167 (1972). 15. I. Hindennach, E. Kaltschmidt and H. G. Wittmann, EJB 23, 12 (1971). 16. A. S. Spirin, N. A. Kisselev, R. S. Shukulov and A. A. Bogdanov, Biokhimiya 28, 920 ( 1963). 17. R. F. Gesteland, JMB 18, 356 ( 1966). 18. M. Meselson, M. Nomura, S. Brenner, C. Davern and D. Schlessinger, JMB 9, 696 (1964). 19. J. Marcot-Queiroz and R. Monier, Bull. Soc. Chim. B i d . 48, 446 (1966). 20. K. Hosokawa, R. Fujimura and M. Nomura, PNAS 55, 198 (1966). 21. T. Staehelin and M. Meselson, J M B 16, 245 (1966). 22. M. I. Lehrman, A. S. Spirin, L. P. Gavrilova and V. F. Golov, JMB 15, 268 (1966). 23. A. Atsmon, P. Spitnik-Elson and D. Elson, JMB 25, 161 (1967). 24. R. F. Gesteland and T. Staehelin, JMB 24, 149 (1967).

98

PNINA SPITNIK-ELSON AND DAVID ELSON

P. Traub and M. Nomura, JMB 34,575 ( 1968). T. Itoh, E. Otaka and S. Osawa, J M B 33, 109 (1968). P. Spitnik-Elson and A. Atsmon, JMB 45,113 (1969). Y. Eilam and D. Elson, Bchem 10, 1489 ( 1971). P. Spitnik-Elson, B. Greenman and R. Abramovitz, E J B 49, 87 ( 1974). P. Spitnik-Elson, R. Abramovitz and D. Elson, submitted for publication. G. Schreiner and K. H. Nierhaus, JMB 81,71 ( 1973). H. G. Wittmann, G. Stoffler, I. Hindennach, C. G. Kurland, L. Randall-Hazelbauer, E. A. Birge, M. Nomura, E. Kaltschmidt, S. Mizushima, R. R. Traut and T. A. Bickle, Mol. Gen. Genet. 111, 327 (1971). 33. P. Spitnik-Elson, FEES Lett. 7,214 (1970). 34. P. Spitnik-Elson and B. Greenman, FEES Lett. 17, 187 (1971). 35. M. Salas, M. A. Smith, W. M. Stanley, Jr., A. J. Wahba and S. Ochoa, JBC 240, 3988 ( 1965 ) . 36. A. V. Furano, J B C 241,2237 ( 1966). 37. R. Monier, M. L. Stephenson and P. C. Zamecnik, B B A 43, 1 (1960). 38. F. W. J. Teale and G. Weber, BJ 65,476 ( 1957). 39. F. W. J. Teale, BJ 76, 381 (1960). 40. R. F. Steiner, R. E. Lippoldt, H. Edelhoch and V. Frattali, in “Quantum Aspects of Polypeptides and Polynucleotides” ( M. Weissbluth, ed. ), pp. 355-366. Interscience, New York, 1964. 41. M. J. Kronman and F. M. Robbins, in “Fine Structure of Proteins and Nucleic Acids” (G. D. Fasman and S . N. Timasheff, eds.), pp. 271-416. Dekker, New York, 1970. 42. G. M. Barenboim, A. N. Domanskii and K. K. Turoverov, “Luminescence of Biopolymers and Cells.” Plenum, New York, 1969. 43. Z. Vogel, T. Vogel, A. Zamir and D. Elson, J M B 54,379 ( 1970). 44. Z. Vogel, T. Vogel, A. Zamir and D. Elson, J M B 60,339 ( 1971). 45. R. Miskin and A. Zamir, Nature NB 238,78 (1972). 46. R. Miskin and A. Zamir, JMB 87,121 (1974). 47. R. Miskin and A. Zamir, J M B 87, 135 ( 1974). 48. A. Zamir, R. Miskin, Z. Vogel and D. Elson, in “Methods in Enzymology,” Vol. 30: Nucleic Acids and Protein Synthesis (K. Moldave and L. Grossman, eds.), Part F, p. 406. Academic Press, New York, 1974. 49. M. I. Sherman and M. V. Simpson, PNAS 64,1388 ( 1969). 50. D. Chuang and M. V. Simpson, PNAS 68, 1474 ( 1971). 51. M. H. Schreier and H. Noll, PNAS 68, 805 (1971). 52. J. Waterson, M. Sopon, S. L. Gupta and P. Lengyel, Bchem 11, 1377 (1972). 53. I. Ginzburg, R. Miskin and A. Zamir, ] M E 79,481 ( 1973). 54. R. Miskin, A. Zamir and D. Elson, J M B 54,355 ( 1970). 55. A. Zamir, R. Miskin and D. Elson, JMB 60,347 ( 1971 ). 56. J. K. Weltman, R. P. Szaro, A. R. Frackelton, R. M. Dowben, J. R. Bunting and R. E. Cathou, JBC 248,3183 (1973). 57. N. Schechter and D. Elson, submitted for publication. 58. R. L. Weiss, B. W. Kimes and D. R. Morns, Bchem 12, 450 (1973).

25. 26. 27. 28. 29. 30. 31. 32.

Classical and Postclassical Modes of Regulation of the Synthesis of Deg rada tive Bacterial Enzymes’ BORISMAGASANIK Department of Biology Massachusetts Institute of Technology Cambridge, Massachusetts I. 11. 111. IV. V.

Introduction. . . . Classical Mode . . . Postclassical Mode: Energy. Postclassical Mode: Nitrogen Conclusion. . . . References. . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

99 100 103 105 113 114

I. Introduction Enteric bacteria, for example Klebsiclla aerogenes, have evolved in a manner that has made glucose their preferred source of carbon and energy, and ammonia their preferred source of nitrogen. However, they have the ability to use a large variety of carbon compounds as substitutes for glucose and a large variety of nitrogen compounds as substitutes for ammonia. Some compounds, like the amino acid L-histidine, can substitute for both glucose and ammonia: their degradation serves to ‘My experience as a graduate student of Erwin Chargaff between 1945 and 1948 has been, as I strongly believe, the major determinant in the course of my subsequent research. His example and the ideas he expressed in many discussions stimulated my interest in the interactions of the multitude of chemical reactions that enable microorganisms to grow in changing environments. My thesis topic was a problem in the intermediary metabolism of carbon compounds, the oxidation of the stereoisomers of inositol by Acetobactet suborydans. This led me to investigate the complete degradation of myo-inositol in Klebsiella aerogenes, an enteric organism. The enzymes responsible for this degradation were found to be “adaptive,” as it was called in the 1950s-we now say “inducible.” The use of auxotrophic mutants in the investigation of this problem led me to the study of the biosynthesis and degradation of histidine and to the study of the biosynthesis and interconversion of purine nucleotides. My colleagues and I became particularly interested in the regulation of biochemical pathways and the regulation of enzyme synthesis. It is a great pleasure to dedicate this essay dealing with the metabolic regulation of enzyme synthesis to Erwin Chargaff. 99

100

BORIS MAGASANIK

provide the cell with energy, as well as with carbon- and nitrogen-containing building blocks. The utilization of these alternative sources of energy, carbon and nitrogen, depends on enzymes not required by cells growing on glucose and ammonia. This essay deals with what I call the classical and postclassical modes of regulation of their synthesis. The result of these regulatory mechanisms is that the production of these enzymes occurs only when they are required for growth.

II. Classical Mode Our current concepts of the regulation of enzyme synthesis are based on the classical studies of the lac system of Escherichia coZi by Monod and Jacob ( I ) , who focused their attention on the highly specific events that result from the cell's response to the presence of lactose in the medium. This response requires that the cells contain sufficient p-galactosidase to convert some of the lactose to allolactose, the actual inducer ( 2 , 3 ) . The same enzyme also hydrolyzes p-galactosides, such as lactose and allolactose; it is thus simultaneously an inducer-forming and inducerdestroying enzyme. The inducer, allolactose reacts with the repressor and brings about an allosteric change leading to the release of repressor from the operator lac0, a stretch of D N A located at the end of the lac operon where transcription is initiated ( 4 ) (Fig. 1). The release of the repressor permits initiation of transcription, which proceeds through the structural genes for p-galactosidase, ZwZ, p-galactoside permease, ZacY and thiogalactoside transacetylase, ZucA, to produce unstable lac messenger RNA. Translation of this R N A yields the appropriate proteins ( 5 ) . The instability of the messenger R N A ensures that accumulation of the products of the lac operon ceases promptly when repression is reestablished by the withdrawal of the inducer. The structural gene for the repressor, lacl, though closely linked to the other lac genes, is not part of the operon (6). Thus the level of repressor in the cell, that is, the sum of operator-bound and free repressor, is not affected by the presence of inducer. We have studied in considerable detail the regulation of the hut (histidine utilization) system in Salmonella tzjphimurium (7-9) and K . aerogenes (10-12), enteric organisms related to E . coli; this system comprises four enzymes required for the conversion of L-histidine to glutamate, ammonia and formamide. These enzymes enable the cell to use histidine as source of energy (Fig. 2). Thus the physiological role of hut is in this respect similar to the role of lac: possession of these systems enables the cells to substitute histidine and lactose, respectively, for

101

REGULATION OF ENZYME SYNTHESlS

L

LACTOSE

GLUCOSE t GALACTOSE

AL LO L A C TO S E

I

P

O

Z

Y

I

A 1

FIG. 1. The regulatory mechanism of the lac operon of Escherlchia coli. Lactose is converted by 8-galactosidase ( Z ) , the product of the lac2 gene to glucose and galactose or to allolactose; this disaccharide can inactivate the repressor ( I ) to cause . structural genes lac1 (repressor), lacZ ( p-galacits release from the operator ( 0 ) The tosidase ), lacy ( @-galactoside pennease), and ZucA, thiogalactoside transacetylase are found linked in the order shown on the E . coli chromosome. The promoter P is the site of initiation of transcription of the lac operon. The operator 0 is the target of the repressor. HISTIDINE UTILIZATION

H

Histidine

-- U

Uroconate +NH,

IPA

I

G

FGA

Glutomate +HCONH,

HUTOPERONS

--j

U ,C iroconate

M

I

,

G

,

C

,P,R,Q

U

,

H

,

bio

V

FIG.2. The regulatory mechanism of the hut (histidine utilization) operons of Salmonella typhimurium. Histidine is degraded by S . typhimurium to glutamate, ammonia and formamide by the sequential action of four enzymes: L-histidine ammonia-lyase (histidase, EC 4.3.1.3); urocanate hydratase ( urocanase, EC 4.2.1.49); imidazolonepropionase ( IPA hydrolase, EC 3.5.2.7); and formiminoglutamase ( FGA hydrolase, EC 3.5.3.8).The genes that code for these enzymes, hutH, U, I, and G, respectively, are linked and lie between gal and bio on the map of S . typhimurium. Two operons are involved in the hut system; hutMZGC constitutes the left-hand operon and hut( P,R,Q)UH constitutes the right-hand operon. M represents the promoter-operator region of the left-hand operon, whereas P,R,Q represents the promoter-operator region of the right-hand operon. C, a member of the left-hand operon, codes for the repressor that, presumably, binds at the operator site of both operons. Urocanate, the first degradation product of histidine, is the inducer, and inactivates the repressor.

102

BORIS MAGASANIX

their usual energy source, glucose. The structural genes of the hut enzymes are organized as two closely linked operons (12, 1 3 ) . Both are repressed by the same repressor, a product of the hutC gene ( 1 4 ) . The two operators presumably have similar, but not identical, nucleotide sequences, since they differ in their affinity for the repressor. This repressor is released from its binding to the operators, when it reacts with urocanate, the product of the action of histidase, whose structural gene is hutH ( 7 , 1 4 ) . In these respects, the regulation of hut is very similar to that of lac: in both instances the physiological inducers, histidine and lactose, are converted by enzymes of the respective systems to the actual inducers, urocanate and allolactose. However, lac and hut differ in that in the former case the actual inducer is destroyed by the very same enzyme, p-galactosidase, that produced it, whereas in the latter case, the actual inducer is destroyed by an enzyme urocanase, the product of the hutU gene, distinct from the enzyme that produced it. For this reason the balance between the cellular levels of histidase and urocanase is essential for correct induction of the hut operons by histidine. An excess of urocanase over histidase prevents induction (15), and conversely, an excess of histidase over urocanase leads to apparently constitutive synthesis, actually due to induction of hut by endogenously produced histidine ( 7 , I I ) . The major difference in the regulation of hut and lac is that the structural gene of the hut repressor, hutC, is a member of one of the operons it controls (14, 16) (Fig. 2). In other words, the repressor is inducible. We can see that hut is very cautiously expressed: when the cells are growing in a medium free of histidine, sufficient repressor is formed to prevent almost completely the expression of the right-hand operon and to reduce expression of the left-hand operon, to which hutC belongs (16). Addition of histidine to the culture leads to production of urocanate by the small amount of histidase present in the cells. Urocanate, in turn, inactivates the repressor. The inactivation leads to increased expression of both operons, and thus to the production of more repressor. Unless sufficient urocanate is produced to inactivate the newly formed repressor, induction will not proceed. Full expression of the operons requires the presence of sufficient urocanate to inactivate the repressor produced by the fully derepressed left-hand hut operon. We see that hut and lac have, in common, negative regulation by a specific repressor protein, and induction by the product of one of the regulated enzymes. They differ from one another in that lac constitutes a single operon and hut two operons, and in that the expression of the structural gene for the repressor ZacZ is not regulated, whereas

REGULATION OF ENZYME SYNTHESIS

103

the expression of the structural gene for the repressor hutC is regulated by its own product. In addition, for both lac and hut, the sole role of the repressor is to prevent transcription of the operons: deletion of either repressor results in constitutive expression of the operons ( 6 , 1 6 ) . Considering now other systems, we find that induction by a product of the physiological inducer is not a general phenomenon. For example, the actual inducer of the gal (galactose utilization) system, is galactose itself (17, 1 8 ) . Similarly, arabinose is the actual inducer of the a m system ( 1 9 ) . In the latter case, the product of the regulatory gene, araC, not only prevents transcription of the ma operon, but, when combined with arabinose, is an essential activator of this transcription ( 1 9 ) . In summary, we may conclude that the essential feature of the classical control system discovered by Jacob and Monod in their studies of lac is the existence of the highly specific repressor protein. This repressor receives information from the environment by its ability to bind specifically a small molecule, the inducer, whose appearance in the cell is the signal that it may have need for the enzymes controlled by the repressor. The repressor transmits this information to the portion or portions of the bacterial chromosome containing the structural genes for the formation of these enzymes. The repressor accomplishes this task by its ability to recognize a sequence of nucleotides, the operator, near the site where transcription of these structural genes is initiated: it binds to the operator when free and prevents transcription; it is released from the operator by combination with the inducer.

111. Postclassical Mode: Energy It was recognized quite early that the presence of the inducer in the growth medium is a necessary, but not a sufficient, condition for the synthesis of inducible enzymes. Quite generally, inducible enzymes are not present at high level in cells of enteric organisms grown in a medium containing glucose ( 2 0 ) . Glucose as such is not responsible for this effect, but rather the rapid catabolism of glucose ( 2 1 ) . This realization led to the hypothesis of catabolite repression: those enzymes whose physiological role it is to supply the cell with energy are not produced by cells already supplied with a better source of energy ( 2 2 ) . Glucose is the energy source par excellence for enteric organisms; consequently, cells supplied with glucose are prevented from producing enzymes whose activity would only augment the already abundant supply of energy available from the degradation of glucose (21) . The physiological significance of this regulatory mechanism is easily understood. It ensures that the enzymes are

104

BORIS MAGASANIK

produced only when there is need for their activity. The classical control by induction ensures that they are only produced when their substrate is available. The effect of catabolite repression, like that of inducer antagonized repression, is inhibition of the initiation of transcription of the k c operon; nevertheless, neither the repressor specified by the h Z gene, nor the lac operator play a role in catabolite repression ( 2 3 ) . The target of catabolite repression in both lac and hut operons is a site very closely linked to the promoter site, where transcription is initiated by RNA polymerase ( 7 , 2 4 ) (Figs. 1 and 2). The clue that led to the present understanding of the mechanism of catabolite repression was the observation (25, 26) that addition of 3’:5’-cyclic AMP to the culture medium can overcome the repressive effect of glucose on catabolite-sensitive enzyme systems. Furthermore, two classes of mutants were isolated that had lost their ability to grow on any one of the many carbon compounds degraded by enzymes subject to catabolite repression. Addition of cyclic AMP restored the ability of one class of these mutants to grow on these compounds; these mutants were found to be defective in an enzyme, adenylate cyclase, required for the formation of cyclic AMP from ATP ( 2 7 ) . The mutants of the second class, whose ability to grow on compounds like lactose, histidine and arabinose could not be restored by addition of cyclic AMP, are defective in a protein, CAP (for cyclic-AMP-binding protein or catabolite-gene activator protein), capable of binding cyclic AMP ( 2 8 ) . The correct initiation of transcription of lac-specific ( 2 9 ) , gal-specific ( 1 8 ) , ara-specific (19) and hut-specific DNA (30) is greatly stimulated by the presence of CAP together with cyclic AMP. The presence of glucose in the growth medium results in a lowered level of intracellular cyclic AMP ( 3 1 ) . The result of the imposition of catabolite repression would thus be a decline in the intracellular concentration of CAP charged with cyclic AMP, and consequently a diminished transcription of catabolite-sensitive operons. So far, it has not been determined what specifically causes the drop in the level of cyclic AMP in cells subjected to conditions that result in the imposition of catabolite repression. It is also not known whether the total amount of CAP is constant or changes with the composition of the growth medium. It can be seen that the postclassical mode of regulation by catabolite repression is similar to the classical mode in that a specific protein, CAP, receives information from the environment by its ability to bind a small molecule, cyclic AMP, whose increased cellular level is a signal that the cell is energy-starved. CAP transmits this information to the many different portions of the bacterial chromosome where genes for

REGULATION OF ENZYME SYNTHESIS

105

energy-producing enzymes are located; these enzymes are the ones that permit the cell to degrade carbon compounds other than glucose. CAP charged with cyclic AMP activates the transcription of the many operons that comprise these structural genes. CAP resembles a classical repressor protein in its specificity for a single small molecule, cyclic AMP, and in its ability to bind specifically to certain nucleotide sequences on the bacterial chromosome. It differs from a classical repressor in that it acts on DNA when combined with the effector, and in that it stimulates, rather than inhibits, the initiation of transcription. However, the most important difference is the much greater specificity of the classical system. The nucleotide sequence recognized by the repressor is a rare one: a particular repressor recognizes only the operators of the single or few operons comprising the system, lac, hut, ara, etc., under its control. The nucleotide sequence recognized by CAP, on the other hand, must recur frequently on the bacterial chromosome. It must be found at the beginning of most sets of genes for enzymes capable of providing the cell with energy other than by degradation of its favorite energy source, glucose. Together, the two control mechanisms permit the cell to produce those enzymes only when they are needed and can be useful: that is, in cells lacking energy, but provided with the appropriate substrate.

IV. Postclassical Mode: Nitrogen The preferred source of nitrogen for enteric bacteria is ammonia. Substitution of another source of nitrogen for ammonia in a minimal medium with glucose as source of carbon almost invariably results in slower growth. Ammonia reacts with a-ketoglutarate produced by the degradation of glucose to form L-glutamate. Glutamate is the precursor of several amino acids, and furnishes the amino group of other amino acids by transamination. Finally, glutamate is a precursor of glutamine, whose amide group furnishes some of the nitrogen atoms of purine and pyrimidine nucleotides and of amino acids. Other nitrogen-containing compounds can substitute for ammonia in the growth medium to support the growth of cells equipped with enzymes capable of converting them to ammonia or glutamate. The problem of the regulation of enzyme systems whose function is to supply the cell with ammonia and glutamate by the degradation of nitrogen-containing compounds presented itself to us almost twenty years ago in the course of our studies of histidine degradation in K. aemgenes ( Fig. 2 ) . I have already mentioned the classical regulation of the hut system by the Hut repressor. This control ensures that tran-

106

BORIS MAGASANM

scription of the hut system will be initiated only when histidine is present in the medium. The hut system is also subject to postclassical regulation by CAP and cyclic AMP; consequently, the transcription of the hut system is not initiated even in the presence of histidine when the medium contains glucose and ammonia. Obviously, cells growing in such a medium do not require the Hut enzymes. But the cells can also be placed in a medium containing glucose and histidine, but no ammonia. In that case, their growth depends on their ability to produce the Hut enzymes, in spite of the catabolite repression exerted by glucose. We found that K. aerogenes does in fact grow on glucose histidine and produces the Hut enzymes (32). Our recent studies led to an explanation for this “escape from catabolite repression.” In the experiments I describe below, we used a mutant (cya-) with a defective adenylate cyclase and therefore incapable of producing any catabolite-sensitive enzyme unless cyclic AMP is added to the growth medium. In contrast to the parent strain with active cyclase, the mutant fails to produce histidase when grown in a chemostat with excess ammonia and limiting glucose unless cyclic AMP is added. It also fails to produce p-galactosidase in the absence of cyclic AMP in this condition. However, when the cya+ strain, or its cya- derivative is grown in a chemostat with excess glucose and limiting ammonia, each produces histidase readily in the absence of cyclic AMP. Neither produces /3-galactosidase in this medium ( 33). These results gave us the first indication that the expression of the hut operons, but not that of the 2ac operon, could be activated during ammonia starvation by a mechanism not dependent on cyclic AMP. Similar experiments with a mutant unable to produce active CAP proved this mechanism not to require CAP either (33). We have identified the protein capable of activating the transcription of hut as the enzyme glutamine synthetase ( E C 6.3.1.2) (34). This role of glutamine synthetase is convincingly demonstrated by a study of the transcription of hut D N A using highly purified RNA polymerase and the four nucleotide triphosphates. This transcription proceeds at a very low rate unless either CAP and cyclic AMP or highly purified glutamine synthetase is added (30). The glutamine synthetase must be in the unadenylylated form (see below), which predominates in the nitrogen-starved cells. An explanation for this unexpected role of glutamine synthetase emerges from recent studies by Tempest et aZ. (35, 36), who proved that, because of the unfavorable equilibrium of the reaction, the enzyme L-glutamate dehydrogenase ( E C 1.4.1.4), generally considered to be responsible for the synthesis of glutamate from a-ketoglutarate and am-

+

107

REGULATION OF ENZYME SYNTHESIS

monia (Eq. l ) , cannot perform this function for cells growing in a medium with a concentration of ammonia below 1 mM. NHI

+ a-ketoglutarate + NADPH + Hf + glutamate + NADPH+ + H20

(1)

They ascribed the assimilation of ammonia into glutamate under this condition to the combined action of glutamine synthetase and an enzyme newly discovered by them, glutamate synthase (E C 1.4.1.13, formerly 2.6.1.53) (Eqs. 2 and 3).

+

Glutamine

Glutamate ATP + a-ketoglutarate

+ NHa glutamine + ADP + Pi + NADPH + H+ 2 glutamate + NADP+ -+

+

(2)

(3)

It is immediately apparent that these two enzymes acting together bring about the same result as glutamate dehydrogenase acting alone: the synthesis of glutamate from a-ketoglutarate and ammonia. The hydrolysis of ATP is responsible for the favorable equilibrium of the reaction catalyzed by the combined action of glutamine synthetase and glutamate synthase. The role of glutamate synthase in glutamate formation has been confirmed by the isolation of mutants lacking this enzyme: they fail to grow in minimal medium containing ammonia at a concentration less than 1 mM (37, 38). Mutants lacking glutamate dehydrogenase as well as glutamate synthase require glutamate for growth, irrespective of the concentration of ammonia in the growth medium ( 3 9 ) . The discovery of this pathway, responsible for the assimilation of ammonia from a low concentration, established a new role for glutamine synthetase. In addition to providing glutamine for incorporation into protein and as a participant in certain biosynthetic reactions, the enzyme is responsible for the formation of glutamate when the concentration of ammonia is too low to allow it to be assimilated by the action of glutamate dehydrogenase. It is therefore not surprising to find both the activity and the formation of glutamine synthetase to be subject to regulation by the ammonia concentration of the growth medium. The level of the enzyme is 6- to 10-fold increased by growing cells in a chemostat in limiting ammonia or by substituting a poor nitrogen source, such as histidine, for ammonia ( 3 4 ) . Furthermore, as shown by Holzer and Stadtman et al., in cells growing with an excess of ammonia, the enzyme is present in large part in an adenylylated form that lacks the ability to catalyze the formation of glutamine (40, 41 ) (Eq. 3 ) . It is fortunate that both unadenylylated and adenylylated enzyme have transferase activity, that is, the ability to catalyze the synthesis of y-glutamylhydroxamate when incubated with glutamine and hydroxylamine in the presence of ADP and of arsenate. It is therefore possible to assav hoth forms of the enzyme ( 4 2 ) .

108

BORIS MAGASANM

The fact that the amount and form of glutamine synthetase change in response to the ammonia concentration of the medium offers a sensible

explanation for the role of glutamine synthetase in the regulation of the synthesis of enzymes capable of providing the cell with glutamate or ammonia. Ammonia deficiency leads to deadenylylation and to an increase in the level of glutamine synthetase. The glutamine synthetase then activates the transcription of operons like hut coding for enzymes whose activity would provide the cell with glutamate and ammonia. Transcription of hut also requires the removal of the Hut repressor, so that hut transcription will occur when histidine is available and glutamate and ammonia are required. Alternatively, transcription of hut can be activated in cells supplied with an excess of ammonia by CAP, but only when an increase in the level of cyclic AMP signals that the cell requires energy. The role of glutamine synthetase in the activation of hut transcription is reflected in the properties of mutants altered in the expression of glutamine synthetase. We can distinguish three phenotypes: those unable to produce glutamine synthetase ( Gln-), those producing glutamine synthetase at a high level even in the presence of an excess of ammonia ( GlnC-), and those producing no active glutamine synthetase, but producing glutamine synthetase antigen at a high level even in the presence of ammonia (Gln(AC)-) (34, 43).Gln- mutants fail to produce histidase during ammonia starvation; GlnC- and Gln( AC ) - mutants, conversely, produce histidine even when grown with an excess of ammonia. The existence of these mutants made it possible to examine other enzyme systems for their response to glutamine synthetase (Table I ) . The put system, involving the enzymes responsible for proline degradation, is TABLE I REQULATION OF SYNTHESIS OF ENZYMES OF NITROQEN METABOLISM I N Klebsiella Regulation by Enzyme(s)

Substrate

Products

Hut Put Nitrogenase Urease Asparaginase Tryptophan TA

Histidine Proline NI Urea Asparagine Trytophan

Glutamate, NHa Glutamate NHa COz, NHa Aspartate, NHs Glutamate

CAP. Glutamine Inducer - CAMP synthetase

+ +-

+ +-

-

-

Cyclic-AMP receptor protein or catabolite-gene activator protein.

+ + + + + +

109

REGULATION OF ENZYME SYNTHESIS

regulated just like the hut system, by induction, activation by CAP and cyclic AMP, and by glutamine synthetase (33, 3 4 ) . The put enzymes, like the hut enzymes, can provide the cell with energy or with glutamate. Other enzymes capable of providing the cell with ammonia or glutamate, but not with energy, such as nitrogenase (44, 4 5 ) , asparaginase ( 4 6 ) ) urease ( 4 7 ) and a tryptophan transaminase (48),are not subject to induction or catabolite repression. Their synthesis appears to be regulated entirely through activation by glutamine synthetase. Finally, the synthesis of glutamate dehydrogenase is also regulated by glutamine synthetase, but in an opposite sense. The level of this enzyme is high in cells growing with an excess of ammonia, and low in nitrogen-starved cells; it is low in GlnC- and Gln(AC)- mutants, and high in GlnA- mutants; apparently, glutamine synthetase represses glutamate dehydrogenase ( 3 8 ) . It is unknown whether this repression reflects direct inhibition by glutamine synthetase of the transcription of gdh, the structural gene for glutamate dehydrogenase, or stimulation by glutamine synthetase of the formation of a Gdh repressor. Having established regulation by glutamine synthetase, we turned to investigate regulation of glutamine synthetase. A discussion of this investigation requires a brief description of the exhaustive studies of the adenylylation and deadenylylation of glutamine synthetase by Stadtman et ul. ( 4 9 ) (Fig. 3 ) . The enzyme is a dodecamer composed of identical

yl” ADENYLYLATION

E y p i *

ATose GS Active

G S -AMP lnoct ive

PIIA

p n D -UMP

( gG l nS A )

-7-iAD P

GS-AMP

Pi

DEADENYLY LATlON

FIG.3. Adenylylation and deadenylylation of glutamine synthetase, adapted from Ginsburg and Stadtman ( 4 9 ) . GS = glutamine synthetase (EC 6.3.1.2); ATase = adenylylation enzyme (EC 3.1.4.15);UTase = uridylylation enzyme; UR = uridylylremoving enzyme; gln = gene controlling production of GS.

110

BORIS MAGASANIK

subunits and has a molecular weight of approximately 600,OOO. An adenylyl group may be attached to a tyrosine residue on each of the twelve subunits. The adenylylation, which results in the loss of the ability of the enzyme to catalyze the formation of glutamine, requires two proteins, called ATase ( adenylyl-[glutamine-synthetase hydrolase, EC 3.1.4.15) and P,. The same two proteins are also required for the removal of the adenylyl groups, which restores the biosynthetic function of glutamine synthetase. Whether ATase and PrI catalyze adenylylation or deadenylylation depends on the form of PI,. This protein is converted, by the attachment of uridylyl groups, from P I I A , which activates adenylylation, to PIID,which activates deadenylylation. The uridylylation of PIIis catalyzed by an enzyme called UTase. Another enzyme, termed UR, so far not studied in detail,'2catalyzes the removal of the uridylyl groups from PnD. The activity of UTase depends on a-ketoglutarate and glutamine: the former activates and the latter inhibits this enzyme. Conversely, glutamine activates and a-ketoglutarate inhibits adenylylation of glutamine synthetase by the ATase.PIIAcomplex. We can seethat in intact cells adenylylation and deadenylylation are controlled by the concentration of ammonia. When the ammonia concentration is low, a-ketoglutarate will accumulate and glutamine will be in short supply. Under this condition, adenylylation by ATase.PIIAis inhibited, and conversion of P I I A to PIID by UTase is stimulated: the result is conversion of glutamine synthetase to the active, nonadenylylated form. Conversely, when the ammonia concentration is high, the level of a-ketoglutarate is reduced by its conversion to glutamate, and the level of glutamine increases. Under this condition, the conversion of PIIA to PIIDis not activated, and presumably PIIDloses its uridylyl group and is converted to PIIA; in addition adenylylation by ATase * P I I A is stimulated: the result is conversion of glutamine synthetase to the inactive adenylylated form. We employed genetic and biochemical methods to elucidate the regulation of the synthesis of glutamine synthetase in K. aerogenes. These studies became possible when a method was developed that made K. aerogenes, previously only infrequently used in genetic studies, susceptible to transduction by the coliphage P1 (SO). Using this method, we found mutations resulting in the inability of the cell to produce glutamine synthetase in three unlinked sites on the chromosome, gZnA, gZnB and gZnD (43, 51) (Fig. 4 ) . Mutations in glnA and in gZnE, a site linked to gZn.B, restored the ability of a mutant in gZnB to produce glutamine synthetase (34, 43, 51 ). We identified glnA as the site of the structural gene for glutamine synthetase, by the demonstration that mutations in glnA could lead to * In fact, UTase and UR may be the same enzyme (49a).

111

REGULATION OF ENZYME SYNTHESIS

FIG.4. The location of some relevant genes on the chromosome of Klebsiellu aerogenes. For gZnA, gZnB, gZnD and gZnE, see Table XI, and for hut see Fig. 2. gdh, gene controlling glutamate dehydrogenase ( E C 1.4.1.4); asm, gene controlling glutamate synthase ( E C 1.4.1.13, formerly EC 2.6.1.53).

the production of enzymically inactive glutamine synthetase antigen or to a heat-labile enzyme (52). Examination of extracts of mutants in glnB, glnD and glnE revealed each to be defective in a different component of the adenylylation-deadenylylation system (51 ) (Table 11). The glnD mutation has resulted in the loss of UTase activity, and the gZnB mutation has altered PII so that it cannot assume the PIln form required for stimulation of deadenylylation. Either of these mutations should lead to the intracellular accumulation of glutamine synthetase in the adenylylated form: they unexpectedly result in the failure to produce glutamine synthetase. We postulate therefore that adenylylated glutamine synthetase represses the synthesis of glutamine synthetase. This view is strengthened by the finding that the mutation in gZnE has led to an alteration in ATase: the enzyme cannot catalyze adenylylation of glutamine synthetase, but catalyzes its deadenylylation whether combined with PIIDor PII.\. The gZnE mutant contains only unadenylylated glutaminc synthetase, and produces the enzyme at a high level even in the presence of an excess of ammonia ( GlnC- phenotype) : apparently, TABLE I1 EFFECTS OF gln MUTATIONS= Mutation

Phenotype

Enzymic defect

glnA gln I3 glnD gln E

GlnGlnGlnGlnC-

Glutamine synthetase inactive PII does not stimulate deadenylylation UTase inactive ATase does not adcnylylate

~~

~~

UTase = uridylylation enzyme; ATase = adenylylation enzyme (EC 3.1.4.15); PII = protein required for the action of ATasc.

112

BORIS MACASANIK

the lack of adenylylated glutamine synthetase prevents repression of glutamine synthetase. Most mutations suppressing glnB or glnD are located in the glnA site and also result in the GlnC- phenotype ( 4 3 ) . Some mutations in gZnA also result in the Gln( AC)- phenotype, production of glutamine synthetase antigen at high level even in the presence of ammonia ( 4 3 ) . The merodiploids, GlnA+/GlnA-, GlnA+/GlnC-, GlnA+/Gln(AC )-, all have the GlnA+phenotype. GlnC-/GlnA- has the GlnC- phenotype ( 5 3 ) . These results indicate that the GlnC- phenotype is the result of an alteration in the structural gene for glutamine synthetase that prevents the enzyme from assuming the form required for its own repression. In that case, introduction of a second functional glnA gene would restore repression. We find that the glutamine synthetase in the glnA mutant with the GlnC- phenotype can be adenylylated. We do not know whether adenylylation fails to proceed far enough to produce repressor, perhaps an enzyme in which all twelve subunits are adenylylated, or whether the allosteric change normally resulting from adenylylation fails to occur. We are still far from understanding the molecular mechanism of repression of glutamine synthetase by adenylylated glutamine synthetase. Recent experiments by Weglenski and Tyler ( 5 4 ) , who measured the intracellular level of gln-specific mRNA, suggest transcription to be the target of regulation. We can now summarize the fundamental aspects of postclassical regulation of enzyme synthesis by glutamine synthetase. It resembles postclassical regulation by CAP in that a single regulatory protein can activate the transcription of many different operons that have in common the same physiological function. In the case of CAP, this function is provision of energy; in the case of glutamine synthetase, provision of ammonia and glutamate. As I have mentioned above, we do not know how the message that energy is required is transmitted through CAP to the responsive systems other than that CAP becomes charged with cyclic AMP and activates their transcription. We have a much better understanding of the mechanism by which the message that ammonia is required is transmitted through glutamine synthetase to the responsive systems. This intricate and elegant cascade mechanism proceeds through three stages, each requiring glutamine synthetase in a different role. I illustrate the mechanism by considering cells growing in a medium containing glucose, histidine and ammonia. Initially, these cells use glutamate dehydrogenase to assimilate ammonia; they contain little glutamine synthetase, and that mostly in the adenylylated form and consequently unable to activate the transcription of hut. Let us now assume that

REGULATION OF ENZYME SYNTHESIS

113

these cells have used up most of the ammonia, so that its level has fallen below 1 mM. Now, glutamate dehydrogenase will no longer be able to function, and the cellular level of a-ketoglutarate will increase; glutamine synthetase, present largely in the inactive form, will not function effectively, and the cellular level of glutamine will decrease. The consequent increase in the ratio of a-ketoglutarate to glutamine will cause deadenylylation of glutamine synthetase, and thus remove the repressor of glutamine synthetase. As a result nonadenylylated glutamine synthetase will accumulate in the cell, repress glutamate dehydrogenase, and activate the transcription of the hut system. The degradation of histidine by the hut enzymes will provide glutamate as well as ammonia, which in turn will be converted to glutamate through combined action of glutamine synthetase and glutamate synthase. Let us now supply these cells, after several generations of growth on glucose and histidine, with an excess of ammonia. Because these cells are rich in active glutamine synthetase, this will result in an increase in the intracellular level of glutamine. The consequent decrease in the ratio of a-ketoglutarate to glutamine will lead to adenylylation of glutamine synthetase, removing the activator of hut transcription and the repressor of glutamate dehydrogenase, and providing the repressor of glutamine synthetase. As a result, the cells will cease to produce the hut enzymes and glutamine synthetase and begin to produce glutamate dehydrogenase. After several generations of growth in this medium, they will have acquired the enzymic constitution appropriate for growth in excess ammonia. It is a remarkable aspect of this postclassical system of regulation that a single protein, glutamine synthetase, can be an enzyme, an activator of transcription and a repressor. In contrast, CAP and the specific repressors are, as far as we know, proteins with but a single function. It is interesting to speculate whether evolution has selected the combination of functions in glutamine synthetase as a particularly effective mechanism of regulation, or has not proceeded long enough to distribute these functions among separate proteins.

V. Conclusion I have called regulation by a highly specific repressor the ‘‘classical mode” of regulation, alluding to the justly termed classical studies of Monod and Jacob ( I ) on the lac system of E . coli. But in another sense, too, this system may be called classical, because of its order and simplicity. A specific repressor reacts with a specific inducer and recognizes a specific operator; its own synthesis is not subject to regulation.

114

BORIS MAGASANIK

Some slight departures from this severely classical mode are encountered in other systems, such as the regulation of hut and ara by their specific repressors, but they still have the essential classical features of high specificity of inducer, repressor and operator. In the postclassical systems, these simple, direct relations are no longer observed. Here, a single protein, glutamine synthetase, plays the role of enzyme, repressor and activator and in each of these incarnations contributes to the regulation of synthesis of a large array of enzymes. Its conversion from one state to another requires the participation of other proteins, whose activity is in turn regulated through metabolites in whose formation glutamine synthetase itself plays a role. Stark simplicity is replaced by subtle intricacy. Finally, the term “postclassical” also properly expresses the debt of the investigators of these systems to those who, by their classical investigations, established the basic concepts of the regulation of enzyme synthesis.

ACKNOWLEDGMENTS The experimental work described in this essay was supported by U.S. Public Health Service research grants GM 07446 from the National Institute of General Medical Sciences and AM 13894 from the National Institute of Arthritis, Metabolism and Digestive Disease, and grant GB 03398 from the National Science Foundation.

REFERENCES 1. F. Jacob and J. Monod, J M B 3,318 ( 1961). 2. C.Burstein, M. Colm, A. Kepes, and J. Monod, BBA 95,634 ( 1965). 3. A. Jobe and S. Bourgeois, J M B 69,397 ( 1972). 4. W. Gilbert and B. Muller-Hill, in “The Lactose Operon” (J. R. Beckwith and D. Zipser, eds.), p. 93 Cold Spring Harbor Laboratory, Cold Spring Harbor, New York 1970. 5. G. Contesse, M. Crepin and F. Gros, in “The Lactose Operon” ( J . R. Beckwith and D. Zipser, eds.), p. 111 Cold Spring Harbor Laboratory, Cold Spring Harbor, New York 1970. 6. J. R. Beckwith, in “The Lactose Operon” (J. R. Beckwith and D. Zipser, eds.), p. 5 Cold Spring Harbor Laboratory, Cold Spring Harbor, New York 1970. 7. W. J. Brill and B. Magasanik, JBC 244,5392 ( 1969). 8. M. K. Meiss, W. J. Brill and B. Magasanik, JBC 244, 5382 (1969). 9. G. R. Smith, Y. S. Halpern and B. Magasanik, JBC 246, 3320 ( 1971). 10. B. Magasanik, P. Lund, F. C. Neidhardt and D. T. Schwartz, JBC 240, 4320 (1965). 11. S. Schlesinger, P. Scotto and B. Magasanik, JBC 240, 4331 ( 1965). 12. R. B. Goldberg and B. Magasanik, J. Bact. 122, 1025 (1975). 13. G. R. Smith and B. Magasanik, JBC 246,3330 ( 1971). 14. D.C. Hagen and B. Magasanik, PNAS 70,808 ( 1973). 15. D. C. Hagen, P. J. Lipton and B. Magasanik, J . B a d . 120, 906 (1974). 16. G. R. Smith and B. Magasanik, PNAS 68, 1493 ( 1971). 17. M.C. P. Wu and M. M. Kalckar, PNAS 55,622 ( 1966).

REGULATION OF ENZYME SYNTHESIS

115

18. J. S. Parks, M. Gottesman, K. Shimada, R. A. Weisberg, R. L. Perlman, and I. Pastan, PNAS 68, 1891 (1971). 19. E. Englesberg, in “Metabolic Regulation” ( A . Vogel, ed.), 3rd ed., Vol. 3, p. 257, 1971. 20. J. Monod, “Recherches sur la croissance des cultures bacteriennes.” Hennann, Paris, 1942. 21. F. C. Neidhardt and B. Magasanik, Nature (London) 178, 801 (1956). 22. B. Magasanik, CSHSQB 26, 249 ( 1961). 23. B. Magasanik, in “The Lactose Operon” (Beckwith and Zipser, eds.), p. 189, 1970. 24. A. E. Silverstone, R. R. Arditti, and B. Magasanik, PNAS 66, 773 (1970). 25. A. Ullmann and J. Monod, FEBS Lett. 2,57 (1968). 26. R. L. Perlman and I. Pastan, BBRC 30, 656 (1968). 27. R. L. Perlman and I. Pastan, BBRC 37, 151 (1969). 28. G . Zubay, D. Schwartz and J. Beckwith, PNAS 66, 104 (1970). 29. L. Eron and R. Block, PNAS 68, 1828 (1971). 30. B. Tyler, A. B. DeLeo and B. Magasanik, PNAS 71,225 (1974). 31. W. Epstein, L. B. Rothman-Denes and J. Merse, PNAS 72, 2300 (1975). 32. F. C. Neidhardt and B. Magasanik, J. Bact. 73,253 (1957). 33. M. J. Prival and B. Magasanik, IBC 246, 6288 ( 1971). 34. M. J. Prival, J. E. Brenchley and B. Magasanik, JBC 248, 4334 1973). 35. 1. L. Meers, D. W. Tempest and C. M. Brown, J. Gen. Microbiol. 64, 87 (1970). 36. D. W. Tempest, J. L. Meers and C. M. Brown, J. Biochem. 117,405 (1970). 37. M. Nagatani, M. Shimizu and R. Valentine, Arch. Microbiol. 79, 164 (1971). 38. J. E. Brenchley, M. J. Prival and B. Magasanik, JBC 248, 6122 (1973). 39. J, E. Brenchley and B. Magasanik, J . Bact. 117,544 ( 1974). 40. K. Wulff, D. Mecke and M. Holzer, BBRC 28, 740 ( 1967). 41. M. S. Kingdon, B. M. Shapiro and E. R. Stadtman, PNAS 58, 1703 (1967). 42. E. R. Stadtman, A. Ginsburg, J. E. Ciardi, J. Yeh, S. B. Hennig and B. M . Shapiro, Aduan. Enzyme Regul. 8 , 9 9 ( 1970). 43. S . L. Streicher, R. A. Bender and B. Magasanik, J . Bact. 121, 320 (1975). 44. S. L. Streicher, K. T. Shanmugam, F. Ausubel, C. Morandi and R. B. Goldberg, J . Bact. 120, 815 (1974). 45. R. S . Tubb, Nature (London) 251, 481 (1974). 46. A. D. Resnick and B. Magasanik, Abstr. Annu. Meeting Amer. SOC. Microbiol. 190 (1975). 47. B. Freidrich and B. Magasanik, Unpublished observation. 48. C. G . Paris and B. Magasanik, Unpublished observation. 49. A. Ginsburg nnd E. R. Stadtman, in “The Enzymes of Glutamine Metabolism” ( S . Prusiner and E. R. Stadtman, eds.), p. 9 Academic Press, New York, 1973. 49a. S. P. Adler, D. Purich and E. R. Stadtman, JBC 16, 6264 (1975). 50. R. B. Goldberg, R. A. Bender and S. L. Streicher, J. Bacteriol. 118, 810 (1974). 51. F. Foor, K. A. Janssen and B. Magasanik, PNAS (in press). 52. A. B. DeLeo and B. Magasanik, J. Bact. 121,313 (1975). 53. R. A. Bender and B. Magasanik, Unpublished observation. 54. P. Weglenski and B. Tyler, personal communication.

This Page Intentionally Left Blank

Characteristics and Significance of the Polyadenylate Sequence in M a m ma lia n Messenger RNA GEORGEBRAWERMAN Depaztment of Biochemistry and Phamcology Tufts University School of Medicine Boston. Massachusetts I . Introduction . . . . . . . . . . . . I1. Discovery. . . . . . . . . . . . A . DNA-like RNA Refractory to Phenol Extraction B. Cytoplasmic DNA-like RNA Rich in Adenylic Acid . . C. Poly(A) in the Cytoplasm of Rat Liver . . . . . D . Poly(A) as a Component of mRNA . . . . . . I11. Isolation of Poly(A)-Containing RNA . . . . . . . A . Conditions for Phenol Extraction . . . . . . . B. Absorption on Nitrocellulose Filters . . . . . . C. Other Adsorption Techniques . . . . . . . D . Assay for Labeled Poly( A ) I V . Size of Poly(A) and Location on Messenger RNA . . . . A. Sedimentation Analysis B. Alkaline Hydrolysis and End-Group Analysis . . . C. Digestion with Exonucleases . . . . . . . . V. Decrease in Poly(A) Size . . . . . . . . A . Comparison of Cytoplasmic and Nuclear Poly( A ) . . . . B . Size Reduction in the Cytoplasm .. . VI . Steady-State P o l ~A( ) A. Size Distribution of Long-Labeled Poly(A) . . . . . B. Total Poly( A ) as Detected by Annealing with Labeled Poly( U ) C . Distribution of Poly(A) Sizes in the Cytoplasm . . . VII . Poly( A ) Elongation . . . . . . . . . . A . Biosynthesis in the Nucleus . . . . . . . . B. Poly(A ) Synthesis in Actinomycin-Treated Cells . . . . C. Poly(A) Elongation in the Cytoplasm . . . . . . . . . . . D. Poly(A) Elongation in the Nucleus . E . Differences in Levels of Poly(A) Elongation . . . . . VIII . Poly(A) and Structure of Messenger RNA . . . . . . A . Poly(A) as Binding Site for Protein in mRNA . . . . B . Interaction of Poly(A ) with other Polynucleotide Sequences in Polysomes . . . . . . . . . . . . IX . Significance of Poly(A) . . . . . . . . . . A . The Widespread Occurrence of the Poly( A ) Sequence in mRNA B. Poly( A ) and mRNA Transport 117

. . . . . .

.

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

. .

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

. . . . . . .

118 119 119 120 121 121 123 123 125 126 127 127 127 127 128 128 128 128 130 130 130 132 133 133 134 137 138 139 140 140 141 143 143 144

118

GEORGE BRAWERMAN

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

C. Poly(A) and mRNA Stability. D. Poly(A) and mRNA Translation X. Conclusion, References,

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

144 145 148 147

1. Introduction Perhaps one of the more remarkable aspects of the poly( A ) sequence in eukaryotic mRNA is that it remained unnoticed for such a long period of time. The presence of this sequence, about 200 nucleotides long, in RNA preparations leads to a substantial excess of AMP in their overall nucleotide composition. With the sensitive analytical methods available, small deviations in base composition could be readily detected. There were numerous examples in the early literature of mammalian RNA fractions unusually rich in AMP (1-3). Moreover, it had been observed by several investigators that the residual RNA labeled in actinomycintreated organisms is particularly rich in this nucleotide (4-6). This writer owes his appreciation of the potential significance of subtle differences in nucleotide composition to his experience in Erwin Chargaffs laboratory. At that time, the central effort in the laboratory was directed toward understanding the regularities in base ratios, both in DNA and RNA. Deviations from normal or usual base compositions were greeted with both interest and concern. One particular instance is worth mentioning, as it led us eventually to the finding that cell organelles like chloroplasts possess their own protein-synthesizing machinery ( 7) : RNA preparations from Euglena gracilis grown in the light and in the dark yielded slightly different nucleotide compositions (Table I ). This photosynthetic microorganism develops chloroplasts only when grown in the light. This led to the suspicion that the differences might be due to the presence in the green organisms of a unique species of RNA associated with the photosynthetic apparatus (8). The RNA TABLE I OF RNA COMPONENTS OF Euglena gracilisO NUCLEOTIDE COMPOSITION

A G C U

Total cellular RNA

Ribosomal RNA

Cells grown Cells grown in darkness in light

Cytoplasm Chloroplasts

21.6 30.2 27.5 20.7

23.2 28.5 26.3 22.0

22.7 29.5 27.1 19.6

29.3 27.4 19.4 23.9

Data are from Brawerman and Chargaff (8) and Brawerman (9)

POLY( A ) IN MAMMALIAN MESSENGER

119

RNA

TABLE I1 PHENOL FRACTIONATION OF RAT LIVERRNAo

A G C U

RNA extracted a t pH 7.6

RNA extracted a t pH 8.5-9.0

18.0 32.8 31.2 18.0

25.6 26.9 26.0 21.5

a Rat liver preparation first extracted with phenol a t pH 7.6. After several extractions, the residue was reextracted with buffers of pH 8.5 to 9.0. Data are from Brawerman and Eisenstadt (14).

component responsible for the shifts in overall base composition was subsequently isolated and found to be part of ribosomal particles associated with the chloroplasts (9). The discovery of poly(A) as a natural component of rat liver also began with studies of nucleotide composition of RNA fractions. It had been reported by Georgiev and Mantieva that DNA-like RNA components of rat liver are refractory to phenol extraction in the cold, but can be recovered by reextraction at elevated temperatures (10). We developed a procedure for the separation of this material based on sequential phenol extractions at neutral and slightly alkaline pH ( 1 1 ). The base composition of the high pH RNA fractions did resemble that of DNA to some extent, but their adenine content was very high compared to that of uracil (Table 11).This anomaly could hardly be ignored by a disciple of the “Chargaff school,” and a study was soon undertaken to uncover its significance.

II. Discovery A. DNA-like RNA Refractory to Phenol Extraction

The first hint that mammalian mRNA might possess a unique physical feature came from studies by Sibatani and collaborators, who reported that phenol treatment of various tissues and cells fails to cause the release of a minor RNA fraction with a base composition and metabolic behavior distinct from that of the bulk of the RNA (12). It was subsequently shown that an RNA component of calf thymus nuclei not released by phenol deproteinization has a DNA-like base composition ( 1 3 ) . Georgiev and Mantieva developed a procedure for the recovery of DNA-like RNA components refractory to phenol extraction (10). Phenol treatment of rat liver homogenates in the cold at pH 6.0 led

120

GEORGE BRAWERMAN

to release of the ribosomal RNA into the aqueous phase. After several reextractions in the cold, the phenol residue was extracted at elevated temperatures (45"-60" ). The RNA components obtained under these conditions were DNA-like in base composition and heterodisperse with respect to sedimentation rate. They also became labeled more rapidly than the rRNA. The investigators chose to attribute the anomalous behavior of these components during phenol extraction to their localization in the nucleus.

B. Cytoplasmic DNA-like RNA Rich in Adenylic Acid In our own laboratory, we developed a procedure for the recovery of the DNA-like RNA components that involves their extraction with cold TrisCl buffers of pH 8.0-9.0 after exhaustive extraction of the rRNA at pH 7.6 (11, 1 4 ) . The starting material was a rat liver nuclear fraction heavily contaminated with cytoplasmic material. DNA-like RNA components similar to those of Georgiev and Mantieva were obtained, but they showed an unexpectedly high adenine content (Table 11).Further studies showed that the A-rich RNA components could be obtained from rat liver cytoplasm (15). This indicated clearly that the resistance to phenol extraction was not due to localization of the RNA in the nucleus. The cytoplasmic RNA component obtained in this manner was labeled far more rapidly than rRNA and was highly effective in stimulating polypeptide synthesis in cell-free extracts of Escherichia coli (Table 111). It was also heterogeneous in size, as judged by zone sedimentation (Fig. 1).All these characteristics suggested at that time that the high pH cytoplasmic RNA component might represent mRNA. TABLE I11 SPECIFIC RADIOACTIVITY A N D TEMPLATE ACTIVITY FOR POLYPEPTIDE SYNTHESIS OF RAT LIVERRNA FRACTIONS"

Property Yields (mg RNA/SO g liver) Radioactivity (cpmlpg RNA) Template activity (pmol Leu/mg RNA)

Nuclear

Cytoplasmic

pH 7.6 pH 8.3

pH 7.6 pH 8.3

1.5

47 360

1.4 490 1930

106

1.3 170

1.6 50 11.50

a Rat liver nuclei and cytoplasm from animal exposed to [14C]oroticacid for 40 minutes were subjected to phenol fractionation. RNA fractions were analyzed for total yield, for specific radioactivity, and for capacity to stimulate polypeptide synthesis in cell-free extract of Escherichia coli. Data are from Hadjivaasiliou and Brawerman (16).

POLY ( A ) IN MAMMALIAN MESSENGER RNA

121

FRACTION NUMBER

FIG.1. Sedimentation characteristics of rat liver cytoplasmic RNA fraction extracted at pH 8.3. Cytoplasmic fraction, obtained from liver of animal exposed to [“CJorotic acid for 40 minutes, was first extracted at pH 7.6; the residue was reextracted with TrisCl buffer of pH 8.3. Data are from Hadjivassiliou and Brawennan (15).

C. Poly(A) in the Cytoplasm of Rat liver To obtain more conclusive evidence for the identification of the high pH RNA as mRNA, it was essential to demonstrate its occurrence in polyribosomes. However, attempts to isolate these structures by differential centrifugation of rat liver cytoplasm led to disappearance of this RNA component. The fractionation procedure also led to degradation of the polysomes, probably by RNase released from lyzosomes. When the microsome fraction was dissolved in sodium deoxycholate and recentrifuged to sediment the ribosomes, the resulting supernatant fraction yielded RNA very rich in adenine (Table IV). The latter could be obtained only after phenol treatment at pH 9. Digestion of this RNA with pancreatic RNase under conditions that leave p l y ( A) intact yielded high-molecular-weight material that consisted almost exclusively of AMP residues (Table IV). Its sedimentation coefficient was about 10 S at pH 5.1. Since poly(A) is known to form aggregates at low pH, a precise molecular weight could not be assigned to our material. The finding of poly( A ) as a natural cellular component was not entirely without precedent. Edmonds and Abrams had reported the presence of this polymer in a poly( A ) polymerase preparation from thymus nuclei ( 17). Also, poly ( A ) -synthesizing enzymes had been found in mammalian tissues (18, 19) and in E. coli (20, 21).

D. Poly(A1 as a Component of mRNA The above studies showed that poly( A ) and a cytoplasmic RNA component with characteristics of mRNA had the same anomalous be-

122

GEORGE BRAWERMAN

TABLE I V NUCLEOTIDE COMPOSITION OF RNA FRACTIONS I~ECOVEHED A'r HIGHpH FROM THE CYTOPLASM O F RAT LIVER" Deoxy cholate-soluble fraction of microsomesb

A G C U

Whole cytoplasmic fraction pH 8.3 extract

pH 9 extract

RNase digestion

29.8 24.5 24.5 21.2

64.5 18.5 17.8 9.2

97.0 2.3 0 0.7

10 S component after

Data are from Hadjivassiliou and Brawerman (16).

* Microsomes were prepared from rat liver cytoplasm by sedimentation a t 100,000 g. The pellets were suspended in sodium deoxycholate and recentrifuged at 100,000 g. The supernatant fraction was subjected to phenol fractionation at pH 7.6 followed by pH 9.0. RNA recovered a t pH 9 was treated with pancreatic RNase under conditions that prevent hydrolysis of poly (A), and the digest was subjected t,o zone sedimentation. The 10 S zone was used for nucleotide analysis.

havior during phenol extraction. They failed, however, to establish a connection between the homopolymer and mRNA. The major obstacle to further investigation along these lines was the lack of an easy procedure for the detection and isolation of the poly(A) sequence. In 1969, Edmonds and Caramela introduced the use of oligo( dT) coupled to cellulose as a selection agent (22). Using ascites cells labeled with 32P, they isolated radioactive poly( A). However, they concluded that the polymer is localized primarily in the nucleus. Lim and Canellakis used polystyrene beads, an agent capable of binding purine-rich polyribonucleotides, in a search for poly( A) sequences (23). They observed that a labeled 9 S RNA component from rabbit reticulocyte polysomes was adsorbed on this resin. Treatment of this component with pancreatic RNase left a fragment rich in adenine still capable of binding. This work provided the first evidence for the possible occurrence of poly( A) in globin mRNA. Kates observed that vaccinia virus cores, which are capable of transcribing the viral DNA into RNA, also promote the synthesis of poly(A) sequences about 100 nucleotide long ( 2 4 ) . Much of the poly( A ) was linked covalently to the RNA transcripts. Using poly( U ) coupled to glass fiber filters for the binding of poly(A), he obtained evidence of the occurrence of this sequence in the vaccinia-specific RNA of infected cells. Some RNA from uninfected cells also seemed to be capable of binding to the filters.

POLY ( A ) IN MAMMALIAN MESSENGER RNA

123

TABLE V EFFECT OF RII~ONUCLEASES ON THE BINDING OF LABELED POLYSOMAL RNA T O MILLIPORIC FILTERS",~

Enzyme None Pan crea t.ic

T1

Uridine-labeled

Adenosine-labeled

AcidBound to insoluble Millipore

AcidBound to insoluble Millipore

1323 214 379

672 3 15

1490 379 910

827 397 382

_ _ _ _ _ _ _ ~

Data are from Lee et al. (26). Polysomes from mouse sarcoma ascites cells labeled either with [aHluridine or [3H]adenosine, in the presence of 0.04 pg of actinomycin 11 per milliliter to prevent ribosomal RNA synthesis, were extracted with phenol directly at pH 9. Enzyme digestions were carried out in presence of 50 mhf TrisCl pH 7.6, 50 mM KCl, and 1 m M hlgClr, to avoid hydrolysis of poly(A) by pancreatic RNase. b

The finding of poly( A ) as a component of mRNA came in our own laboratory through the use of Millipore filters ( 2 5 ) . We observed that labeled polysomal RNA from mouse ascites cells binds to the filters at high ionic strength, whereas ribosomal RNA does not bind under these conditions. Digestion of adenosine-labeled RNA with RNase left about 40%of the radioactivity still capable of binding to the filters (Table V ) . Synthetic poly ( A ) showed the same binding characteristics, and we soon realized that the component responsible for the binding of the polysomal RNA was a polyadenylate sequence. Edmonds et al. described similar findings using their oligo ( dT ) -cellulose as the binding agent (26). Darnel1 et al. also noted the occurrence in polysomal RNA of a sequence rich in AMP and resistant to pancreatic RNase (27).

111. Isolation of Poly(A1-ContainingRNA A. Conditions for Phenol Extraction During the course of our studies of RNA binding to Millipore filters, we observed that only the RNA components extracted with phenol at alkaline pH were capable of binding (Table VI). Phenol extraction of polysomes at neutral pH released a substantial amount of heterogeneous labeled RNA, but very little of it was retained on the filters. The RNA obtained after reextraction at pH 9 was adsorbed quantitatively on the filters, an indication that all the RNA chains recovered in this fashion have a poly(A) sequence. Thus it appeared that our phenol fractionation procedure separated RNA components on the basis

124

GEORGE BRAWERMAN

TABLE VI MILLIPORE-BINDINQ CHARACTERISTICS OF POLYSOMAL RNA FRACTIONS EXTRACTED WITH PHENOL AT NEUTRAL A N D ALKALINEpH0.b ~~

pH of RNA Extraction 7.6 9.0

~~

~

Radioactivity

(rg)

Total (cpm)

Bound to Millipore (% of total in fraction)

1152 38

54 ,500 79,250

2 100

Yield

Data are from Brawerman et al. (88). sarcoma ascites cells labeled in presence of 0.04 pg of actinomycin D per milliliter were extracted sequentially with TrisCl buffers of pH 7.6 and 9.0.

* Polysomes from mouse

of their poly ( A) content. Reconstruction experiments showed that poly ( A ) interacts with denatured proteins during the phenol treatment at neutral pH, thus causing the poly(A)-containing RNA to remain in the protein gel (Table VII), The poly( A)-protein interaction appears to be promoted by monovalent cations, and the TrisCl buffer at neutral pH used in the extraction procedure provided an abundant source of such cations. When lower concentrations of Tris buffer are used, the poly( A)-containing RNA is readily extracted at neutral pH (Table VIII). The alkaline pH seems to favor the release of this RNA species when conditions are otherwise unfavorable (Table VIII ) . Moreover, TABLE VII

INTERACTION OF THE POLY (A) SEQUENCE WITH PROTEINS DURING PHENOL EXTRACTION I N THE PRESENCE OF NEUTRAL TRIS BUFFER^^ Amount left in aqueous phase (% of total)

RNA component

No addition

Polysomal, pH 7.6 Polysomal, pH 9.0 Poly (A), synthetic

98 96

Polysomes added 68 19

3

Methylated albumin added 80 35 6

Data are from Brawerman et al. (88). Labeled RNA components were mixed with either unlabeled polysomes or methylated albumin and subjected to phenol treatment in the presence of 100 m M TrisCI, pH 7.6; the radioactivity left in the aqueous phase was measured.

POLY ( A ) IN MAMMALIAN MESSENGER

RNA

1%

TABLE VIII EFFECTO F VARIOUS AGENTS ON THE EXTRACTION OF LABELED RNA COMPONENTS FROM MOUSESARCOMA POLYSOMIW-* Tris p H 9.0, 100 m M

Tris pH 7.6, 10 m M RNA component extracted

No addition

RNA without poly(A) RNA with poly(A) Poly(A) segments

330 420

a

130

No EDTA+ addition CHCls

EDTA

CHCla

EDTA+ CHCl3

330

280

280

310

130 40

10

10 10

470

270 470

160

150

10

Data are from Brawerman (29).

* Polysomes from cells labeled with adenosine in the presence of 0.04 pg of actino-

mycin D per milliliter, to prevent ribosomal RNA synthesis, were subjected to phenol extraction a t 0"-4" in the presence of sodium dodecyl sulfate and other components listed in the table. Values represent the radioactivity recovered in the aqueous phase.

only a small proportion of the Tris molecules are in the protonated form at pH 9. The recovery of the poly(A)-containing RNA can be affected by a variety of agents, and many investigators may have missed this material because of inadequate extraction procedures. We have observed that the inclusion of either EDTA or chloroform during phenol extraction in the cold prevents the recovery of poly(A)-containing RNA (Table VIII). At room temperature, on the other hand, CHC1, favors the release of these RNA molecules (30).Extraction at elevated temperature in the presence of CHCI, or EDTA also favors poly(A) release (29). The latter procedure is generally used for nuclear RNA, because the high temperature tends to keep the DNA in the protein gel. Chloroform, however, promotes release of the DNA even when used at high temperature.

B. Adsorption on Nitrocellulose Filters The binding of poly(A) to Millipore filters provided the basis for a procedure for the isolation of RNA molecules bearing this sequence. Of this RNA, 30-40 pg can be adsorbed on a single filter (28). The adsorbed material can be eluted in a small volume of 100 mM Tris, pH 9, and 0.511: sodium dodecyl sulfate. Zone sedimentation of the material recovered from mouse sarcoma polysomal RNA showed a considerable enrichment in heterogeneous components sedimenting between 10 and 35 S (Fig. 2). Adsorption of rabbit reticulocyte polysomal RNA led to a preparation with a prominent 10 S component. Some ribosomal RNA, primarily the 28 S component, was still present. Filtration at

126

GEORGE BRAWERMAN

FIG.2. Sedimentation characteristics of unlabeled RNA components adsorbed on Millipore filters. Total polysomal RNA ( A ) was filtered through Millipore, and adsorbed material was eluted and subjected to zone sedimentation. Material at the top of tube represents impurity derived from Millipore filters. ( B ) Mouse sarcoma Millipore-bound RNA; ( C ) rabbit reticulocyte millipore-bound RNA. Data are from Brawerman et al. ( 2 8 ) .

18"-22" instead of 0" reduced the amount of ribosomal contamination, but the latter could not be eliminated entirely. Neither did a second cycle of binding to Millipore cause any further reduction in rRNA contamination. It is possible that the residual rRNA represents molecules complexed with mRNA. The use of more selective procedures, such as chromatography on oligo ( dT )-cellulose, cah also lead to preparations with ribosomal contamination (31 ) , One possible shortcoming of the Millipore-binding technique is the apparent failure to adsorb molecules with short poly( A) segments. The globin mRNA of rabbit reticulocytes has been fractionated by adsorption on Millipore into molecules with large poly(A) segments capable of binding, and others with shorter segments not retained on the filters ( 3 2 ) . On the other hand, the isolation of insect chorion RNA with very short poly(A) sequences by binding to Millipore has been reported ( 3 3 ) . We have determined that about 98%of the poly( A)-containing RNA of mouse sarcoma ascites polysomes can bind to Millipore (unpublished). The Millipore-binding technique has been used by various investigators to isolate specific mRNA components active in protein synthesis ( 3 4 ) . Ovalbumin mRNA has been purified to a considerable extent in this fashion ( 3 5 , 3 6 ) .

C. Other Adsorption Techniques It was observed by Sullivan and Roberts that material with the characteristics of mRNA can bind to cellulose powder ( 3 7 ) . This binding was subsequently attributed to a property of the poly(A) sequence (37, 3 8 ) . It appears to be due to interaction of the poly( A) with lignins present as contaminant in some cellulose preparations ( 3 7 ) . While the above procedures may provide convenient means for the isolation of mRNA, techniques based on specific interaction of poly ( A )

POLY ( A ) IN MAMMALIAN MESSENGER

RNA

127

with oligo( dT) coupled to cellulose or poly( U ) coupled to Sepharose are more rational and are preferred by most investigators (29).

D. Assay for labeled Poly(A) Studies of poly( A ) metabolism require the serial assay of many samples for radioactivity. We found the adsorption of the samples on Millipore followed by counting of the filters in scintillation fluid to be a most convenient procedure (39). Kates and collaborators developed another convenient method for the assay of labeled poly(A) ( 4 0 ) . This involves adsorption on glass-fiber filters to which poly( U ) has been coupled by UV irradiat‘ion.

IV. Size of Poly(A1 and location in Messenger RNA A. Sedimentation Analysis The first estimate for the size of the poly( A ) sequence was obtained by zone sedimentation of the labeled homopolymer released from polysoma1 RNA by ribonuclease treatment. We found that the poly( A ) segment from mouse sarcoma ascites cells behaves as a homogeneous component with a sedimentation coefficient very similar to that of transfer RNA ( 2 5 ) . The same result was obtained with the poly(A) segment from HeLa cells (27). Calibration curves relating poly( A) size to sedimentation coefficient have been published by Fresco and Doty ( 4 1 ) . Use of these yields a value of about 220 nucleotides for the size of the poly ( A ) segment ( 4 2 ) , but this value applies only to cytoplasmic poly(A) after a 2-hour labeling period, As shown below (Section V, B ) , the poly(A) segment changes in size as the mRNA ages in the cytoplasm.

B. Alkaline Hydrolysis and End-Group Analysis A different approach led to additional information on the size of the poly(A) sequence, and at the same time provided strong evidence for its localization at the 3’ end of the mRNA chain. We observed that alkaline hydrolysis of the isolated poly(A) yields small amounts of adenosine in addition to the nucleotide. The ratio of AMP to adenosine in the hydrolyzates was about 170 (39). The nucleoside could have been generatcd only from a polynucleotide with a free 3’ OH. Since the ribonuclease treatment used to release the poly( A ) segment generates phosphorylated 3’ termini, the free 3’ OH at the end of the poly( A ) segment must have been present as such in the intact mRNA. This led to the conclusion that the poly(A) sequence is localized at the

128

GEORGE BRAWERMAN

3’ end of the RNA chains. Since each poly(A) sequence would yield one adenosine residue, the yield of nucleoside provided a measure of the average length of this sequence. The value of 170 residues per poly(A) chain was close to the estimate obtained by sedimentation analysis. This value, however, was based on the assumption that the poly(A) chains were uniformly labeled. This turned out not to be the case, as discussed below (Section VII).

C. Digestion with Exonucleases The results from alkaline hydrolysis provided the first evidence for the localization of the poly( A ) sequence at the 3’ end of mRNA. Since the chain-length estimate obtained in this fashion was close to the one obtained by sedimentation analysis, it seemed likely that most, if not all, of the poly(A) chains are at the 3’ end (39). This conclusion was confirmed by Perry and collaborators through the use of highly purified exonucleases specific for the 3’ end of RNA chains (43, 44).Experiments with periodate oxidation of mRNA provided additional evidence for this conclusion ( 45 ) . V. Decrease in Poly(A1 Size A. Comparison of Cytoplasmic and Nuclear Poly(A) Poly(A) sequences had been detected in the nucleus as well as in the cytoplasm ( 2 6 ) .We observed that alkaline hydrolysis of the nuclear poly(A) also releases small amounts of adenosine in addition to AMP (39). The yields of adenosine from the nuclear poly(A) were similar to those obtained with the cytoplasmic component, which indicated that the poly ( A ) sequences in the nucleus are also located at the 3’ end of RNA chains. Thus it seemed that the precursors to mRNA in the nucleus become polyadenylated after transcription is completed, and that the portion of these molecules next to the poly( A ) segment contains the sequences destined to become the mature mRNA (39, 4 6 ) . While the sizes of nuclear and cytoplasmic poly(A) appeared to be very similar ( 2 6 ) , the yields of adenosine in alkaline hydrolyzates of the nuclear component were consistently lower (39). This finding suggested that the nuclear poly( A ) sequence might be somewhat larger than the cytoplasmic sequence. A careful comparison of the two poly( A ) components by polyacrylamide gel electrophoresis confirmed this difference in size (39). B. Size Reduction in the Cytoplasm The above results suggested that a reduction in the size of the poly(A) sequence might take place as the nuclear RNA is converted

POLY ( A ) IN MAMMALIAN MESSENGER

129

RNA

[l![&[ CYloPLASMIc

2 4 2

lo

20

10 20 FRACTION NUMBER

30.40

30

40

(+I

FIG.3. Polyacrylamide gel electrophoresis of poly( A ) preparations from mouse sarcoma ascites cells labeled in uiuo. Tritiated adenosine was injected into peritoneal cavity of infected mice. All preparations were mixed with [“C]poIy(A) from the cytoplasm of cells labeled in uitro for 2 hours. Values are plotted on relative scale of 10 for better comparison. Vertical dashed lines indicate position of average mobility of 2-hour labeled sample. Data are from Brawerman (47).

into mature mRNA. Our initial experiments on poly( A ) sizes in mouse ascites cells had indicated that the decrease might be gradual, since cytoplasmic poly(A) labeled for 1 hour seemed smaller than the same component labeled for 30 minutes (39). A more systematic study showed that the poly(A) sequence in mRNA becomes gradually shorter as the RNA ages in the cytoplasm (Fig. 3 ) . The distribution of poly( A ) sizes also becomes more heterogeneous ( 4 7 ) . This finding brought to mind the “ticketing hypothesis” for the control of mRNA lifetime in the cell ( 4 8 ) . Sussman had proposed that a special redundant sequence in front of the initiation site in mRNA loses a short segment each time a new polypeptide chain is initiated, complete loss of this sequence signaling the inactivation of the mRNA. Thus the longevity of the RNA would be determined by the length of this sequence. While the poly(A) sequence is located at the wrong end of the RNA chain, it seemed possible that it might function in this fashion. To verify this possibility, we examined the poly ( A ) sizes in mRNA not associated with polysomes. The latter showed the same size decrease as that of polysomal poly(A). Moreover, inhibition of protein synthesis by cycloheximide or by aminoacid deprivation led to the same initial changes in cytoplasmic poly( A) sizes as those occurring in cells actively engaged in protein synthesis. Thus it appears that the gradual decrease in poly(A) size associated with mRNA aging in the cytoplasm is independent of the translation process. Similar results were obtained with HeLa cell poly(A) ( 4 9 ) .

130

GEORGE BRAWERMAN

VI. Steady-State Poly(A1 A. Size Distribution of Long-Labeled PolytA) The gradual size decrease of the poly ( A ) sequence in mRNA seemed to provide a mechanism for the orderly aging of this RNA species. A uniform rate of erosion would eventually lead to the complete removal of this sequence and leave the mRNA vulnerable to exonucleolytic attack. However, the size distributions of poly(A) segments labeled for 6 hours and 24 hours are rather similar. Thus it seemed that the poly( A) size decrease in the cytoplasm is relatively rapid at first, and slows down considerably after a few hours. This suggested to us that the poly( A ) sequence reaches a terminal steady-state size that is still quite substantial. Measurements of the size of steady-state poly( A ) were first carried out on preparations labeled for periods of 24 hours or more. Greenberg and Perry (50) showed that the long-labeled poly(A) of mouse L cells is quite heterogeneous, with a substantial proportion of relatively large segments. The bulk of the poly ( A ) migrated more slowly than tRNA. Sheiness and Darnel1 (49) obtained similar results with HeLa cell poly(A), and further observed that the nuclear poly( A ) component remains large and relatively homogeneous in size. The apparent failure of the nuclear poly( A ) to undergo the initial stages of shortening had already been noted (39). B. Total Poly(A) as Detected by Annealing with Labeled Poly(U) For studies of steady-state poly( A), it seemed preferable to measure the total poly(A) population of unlabeled cells rather than the longlabeled components. This could be accomplished easily by annealing the unlabeled material with radioactive poly( U ) and destroying the excess poly( U ) with pancreatic RNase (42, 51). The sedimentation profile of total cytoplasmic poly(A) from mouse sarcoma ascites cells was more heterogeneous than the newly synthesized poly( A ) , and its average sedimentation coefficient was significantly smaller (3.4 S instead of 3.8 S ) (Fig. 4). The sedimentation profile of poly(A) from rabbit reticulocytes suggested a bimodal distribution of poly( A ) sizes, with sedimentation coefficientsconsiderably lower than that of mouse sarcoma poly ( A ) (2.4 S and 3.0 S ) . The poly(A) size distribution in rabbit reticulocytes had also been studied by subjecting long-labeled material to polyacrylamide gel electrophoresis, and two size classes had been observed here as well ( 32,52), Since the resolving power of polyacrylamide gel electrophoresis is far superior to that of zone sedimentation, we adapted this technique to the study of steady-state poly(A) (53).Preparations from unlabeled

POLY ( A ) IN MAMMALIAN MESSENGER

131

RNA

FRACTION NUMBER

FIG.4. Zone sedimentation analysis of steady-state polysomal poly( A) components from ( A ) mouse sarcoma ascites cells and ( B ) rabbit reticulocytes. RNA preparations were digested with pancreatic RNase, the digests were deproteinized, and samples were subjected to zone sedimentation. Poly( A) was assayed by annealing with ['H]poly( U). The mouse sarcoma RNA preparation was from cells exposed to ["Cladenine for 2 hours, and radioactive poly( A) was also measured (dashed line). The curve for mouse sarcoma steady-state poly(A) is reproduced in panel B (thin line). Data are from Jeffery and Brawerman ( 4 2 ) .

cells were subjected to electrophoresis, and the material extracted from gel slices was assayed with labeled poly( U ) . This revealed a broad distribution of poly(A) sizes for the cytoplasmic poly(A) of mouse sarcoma cells (Fig. 5). The extent of alteration in poly(A) sizes during mRNA maturation is clearly illustrated by comparison of the electropho-

I

c

NUCLEARSTEADY STATE

EAR ACTINOMKN

-

1 0 NUMBER

20

30

40

(+I

FIG.5. Polyacrylamide gel electrophoresis of poly( A ) components of mouse sarcoma cells. Upper panels represent steady-state ( A ) cytoplasmic and ( C ) nuclear poly( A ) ( 0-0 ), assayed by annealing with radioactive poly( U). Profiles of pulse-labeled ( 5 minutes) poly( A ) components (thin lines) are superimposed. Lower panels represent poly(A) labeled for 5 minutes in cells pretreated with high levels of actinomycin ( B, polysomal; D, nuclear). Data are from Brawerman and Diez (53).

132

GEORGE BRAWERMAN

retic profiles of newly synthesized and steady-state poly (A). The former is quite homogeneous and its mobility is far lower than that of the steady-state component. Yet a significant portion of the steady-state segments has the mobility of the newly synthesized component. The nuclear steady-state poly( A ) showed an electrophoretic profile closer to that of the newly synthesized polymer. It is possible that the poly(A) sequence remains essentially unchanged in the nucleus, and that the smaller components evident in the steady-state preparation are residual cytoplasmic contaminants.

C. Distribution of PolytA) Sizes in the Cytoplasm Precise knowledge of the poly( A ) sizes depends on the availability of reference components of known molecular weight. We used the data of Fresco and Doty relating poly( A ) sizes to sedimentation coefficients ( 4 1 ) . The sedimentation profiles shown in Fig. 4 were converted into size distributions by assigning a molecular weight value to each fraction.

I

1

I

I

I

I

I

I

50 K)o I 5 0 200 250 300 350 WLMA) CHAIN LENGW (NUMBER OF AMP RESIDUES)

FIG.8. Size distribution of poly( A ) chains. Calibration values of Fresco and Doty (41 ), relating sedimentation coefficients to poly( A ) sizes, were used to convert data in Fig. 4 into size distribution. Values were corrected so as to represent number of chains in each size class. ( A ) Rabbit reticulocyte; ( B ) Mouse sarcoma; 0-0, steady state;

-- - 0, 2-hour labeled.

POLY ( A ) IN MAMMALIAN MESSENGER

RNA

133

The relative number of chains in each fraction was also computed. The results indicate that the steady-state poly( A) segments range in length from 60 to 330 nucleotides (Fig. 6 ) . The distribution is very broad with a peak around 120 nucleotides. There is substantial overlap with the curve for the poly(A) labeled for 2 hours, which shows a peak around the 200 nucleotide size. The reticulocyte poly( A ) shows a narrower distribution of sizes, ranging from 40 to 170 nucleotides, with the major peak around 60 nucleotides. These data, however, must be considered with some reservation since the calibration curves of Fresco and Doty may not be strictly valid under the conditions used in this study. The data on size distribution indicate a lower size limit of approximately 60 nucleotides for the mouse sarcoma poly(A) and 40 for the component from rabbit reticulocytes. This suggests that the poly( A ) sequence reaches a limit size beyond which it remains stable until rapidly destroyed. It is possible, however, that segments of smaller size do occur in the cell but are not detected by the radioactive poly(U) assay procedure.

VII. Poly(A1 Elongation* A. Biosynthesis in the Nucleus Poly( A ) synthesis appears to proceed by stepwise addition of AMP residues to the 3' end of nuclear RNA chains ( 5 4 ) . It is a relatively rapid process as indicated by the uniform size of nuclear poly( A) after a 5-minute labeling period (see Fig. 5). One interesting feature of poly(A) synthesis is its high sensitivity to the adenosine analog cordycepin. Low levels of this drug, insufficient to affect the synthesis of the heterogeneous nucleoplasmic RNA, inhibit the appearance of mRNA in the cytoplasm ( 5 5 ) . This inhibition is directed specifically at the poly( A)-containing RNA (Table IX). It had been assumed that cordycepin exerts its inhibitory effect on poly( A ) synthesis by becoming incorporated into the growing chains and causing their premature termination, owing to the absence of the 3' OH in this analog. Our own data, however, indicate that the residual poly( A ) produced in the inhibited cells is still terminated by AMP (39). Moreover, the poly(A) segments labeled in the treated cells are nearly of normal size (unpublished results). It may be that cordycepin acts at the initiation stage of polyadenylation. Poly(A) synthesis is also .sensitive to actinomycin (22, 46), in spite of the fact that this process does not seem to be part of transcription.

* See article by Edmonds and Winters in this volume [Ed.].

134

GEORGE BRAWERMAN

TABLE IX EFFECTO F CORDYCEPIN O N LABELINQ OF NUCLEAR A N D POLYSOMAL RNA COMPONENTW~ Nuclei

Polysomes RNA

POlY (A) Total RNA Segment Control Cordycepin Percent inhibition

40,500 29,000 28

Poly(A) - Poly(A)

1010 210 79

1830 2230 -20 ~~~

4160 1130 73

+

POlY(A) Segment 1180 270 77

~~

Data are from Mendecki et al. (39). Nuclei and polysomes prepared from mouse sarcoma ascites cells were incubated in a medium containing 0.04 ag/ml actinomycin D, in the presence or absence of 25 pg/ml cordycepin, and labeled for 30 minutes with adenosine. Values represent radioactivity in various RNA components. Poly(A) - and poly(A) represent RNA components lacking and containing a poly(A) sequence, respectively.

+

The inhibition is apparently due to exhaustion of the nuclear RNA molecules that serve as primers for polyadenylation. Messenger RNA appearance into the cytoplasm continues for a while after exposure to the drug. We observed that some of the poly( A ) fails to leave the nucleus in the inhibited cells (Fig. 7 ) .

B. Poly(A1 Synthesis i n Actinomycin-Treated Cells Our studies of the fate of poly(A) during mRNA aging had led us to suspect that the extremity of this sequence might be subject to turnover. To investigate this possibility, we studied the labeling of poly( A ) in cells pretreated with actinomycin D (53,56). Since de nouo poly( A ) synthesis is blocked in these cells, we expected that any addition of labeled AMP residues to preexisting chains would be readily detectable. The uptake of radioactivity in the cytoplasmic poly(A) of the actinomycin-treated Chinese hamster cells was initially more rapid than in nuclear poly ( A ) . Moreover, the labeling in the cytoplasm leveled off within 10 minutes, while it continued linearly in the nucleus (Fig. 8 ) . This pattern of labeling is strikingly different from that observed under normal conditions, and does not fit the precursor-product relationship between nuclear and cytoplasmic poly( A). The size distribution of the poly( A ) labeled in the cytoplasm was highly heterogeneous, and its mobility in polyacrylamide gel was similar to that of the steady-state poly(A) (Fig. 5 ) . It is possible to determine the size of the labeled sequence provided it is located at the 3’ end of the poly(A) chains.

POLY ( A )

135

IN MAMMALIAN MESSENGER RNA B

I

KILY7A). MJCLEAR

10

20

33 TIME, MIN

FIG.7. Effect of actinomycin D on labeling of nuclear and cytoplasmic RNA components of mouse sarcoma ascites cells. Cell suspensions were incubated in the presence of ['Hladenosine and 0.04 a g of actinomycin per milliliter to prevent labeling of rRNA components. Five minutes after addition of label, a portion of the suspension was supplemented with actinoniycin at 10 pg/ml (AD). RNA lacking poly(A) represents the portion of labeled polysomal RNA that did not bind to Millipore filters. O---O,Cells subjected to high levels of actinomycin. Data are from Mendecki et d. (39).

.---.,

FIG. 8. Time course of poly(A) labeling in cells pretreated with high levels of actinomycin D. Cell suspensions were incubated with 5-10 p g of actinomycin per milliliter for 30 minutes before exposure to ['Hladenosine. ( A ) Chinese hamster nuclei. Data are cells; ( B ) Mouse sarcoma cells; 0-0, polysomes; from Brawerman and Diez (53)and Diez and Brawerman ( 5 6 ) .

136

GEORGE BRAWERMAN

Alkaline hydrolysis releases one labeled adenosine and a certain number of labeled AMP residues. The AMP/adenosine ratio showed that the poly(A) chains in the cytoplasm contained an average of 8 labeled AMP residues ( 5 6 ) . Hence it appears that a continuous process of AMP addition takes place at the 3' end of the poly( A ) sequences in the cytoplasm of actinomycin-treated cells. This is apparently coupled with exonucleolytic cleavage at the 3' end, leading to turnover of an average of 8 residues at the extremity of the RNA chains. This can explain the rapid leveling-off of incorporation into the cytoplasmic poly( A) segments. The synthesis in the nucleus also represents slow elongation of preexisting poly( A ) sequences. ApparentIy, there is no concomitant cleavage in this case. The average length of the radioactive sequence in the nuclear poly( A ) chains is about 40 residues after a 1-hour labeling period, and the size of the labeled segments is considerably greater than in normal cells (Fig. 5 ) . One interesting feature of the elongation process is its relative insensitivity to cordycepin (Table X). As mentioned above, de nooo poly( A ) synthesis is highly sensitive to this drug. We had concluded tentatively that cordycepin inhibits specifically the polyadenylating enzyme, and that it may operate at the level of initiation of poly( A ) synthesis. This may account for its lack of effect on the poly ( A ) chain extension process. Cytoplasmic polyadenylylation had already been reported in sea urchin embryos (57, 58). It was observed that fertilization induces a rapid and extensive elongation of this sequence in the cytoplasm, concomitant EFFECTOF CORDYCEPIN ON

TABLE X Novo SYNTHESIS AND PoLY(A) SEQUENCEQJ

DE

De novo synthesis (control cells) Addition None Cordycepin Percent inhibition

Nuclei Polysomes

ON

ELONGATION OF

THE

Elongation (actinomycin-treated cells) Nuclei

Polysomes

33,800 3,100

8000 2300

4200

1300

4050

91

71

3

840 35

Data are from Diez and Brawerman (68). Polysomes and nuclei prepared from Chinese hamster cells in suspension culture, incubated first in the presence or in the absence of 5 pg of actinomycin D per milliliter and exposed to labeled adenosine for 30 minutes in the presence or in the absence of 50 pg of cordycepin per milliliter. Radioactivity was measured in poIy(A) segments prepared from deproteinized RNA.

POLY ( A ) IN MAMMALIAN MESSENGER RNA

137

with the mobilization of the mRNA into polysomes. This latter process of poly ( A ) elongation is also insensitive to cordycepin.

C. Poly(A1 Elongation in the Cytoplasm The results of the studies with actinomycin-treated cells indicate that the poly(A) chains being elongated in the cytoplasm represent a heterogeneous population of segments with a small number of labeled AMP residues at the 3' end. Since these characteristics differ sharply from those of newly synthesized poly(A) derived from the nucleus, it seemed to us that the elongating population might be detectable even in the presence of the de nuuo poly( A ) in cells not treated with actinomycin. The proportion of radioactivity due to cytoplasmic elongation should be greatest at early times, and decrease gradually as new poly( A ) derived from the nucleus accumulates in the cytoplasm. This was reflected in the changing AMP/ adenosine ratios in alkaline hydrolyzates of cytoplasmic poly( A ) , as the duration of adenosine incorporation increased (Fig. 9). The ratio was very low at first, owing to the preponderance of elongating segments with short labeled sequences. It increased gradually as more and more fully labeled segments synthesized de m o o appeared in the cytoplasm. Measurements of the AMP/ adenosine ratio in poly( A ) fractions with different mobilities in polyacrylamide gel con-

0

a

40

PERIOD OF LABELING, MIN

FIG.9. Average sizes of labeled sequences in nuclear ( ----a)and cytoplasmic ( 0-0) poly( A ) segments of mouse sarcoma ascites cells. Cell suspensions were incubated with ['Hladenosine in the absence of any drug. Poly(A) preparations were hydrolyzed in alkali, and the amounts of radioactivity in AMP and adenosine were determined. The ratio of AMP to adenosine is a measure of the average number of labeled AMP residues at the 3' end of poly( A) chains. Data are from Brawerman and Diez ( 53 ) .

138

GEORGE BRAWERMAN

firmed the occurrence of the two populations. Fractions with high ratios (derived from the nucleus) were confined to a narrow zone of low mobility, while the components with low ratios migrated more rapidly ( Fig. 10). The fractions of highest mobility, presumably well separated from the labeled nuclear poly( A), yielded ratios of 6-8, similar to those obtained from the cytoplasmic poly ( A ) of actinomycin-treated cells. This provided a nice confirmation of the size of the sequence undergoing turnover in the cytoplasm.

D. Poly(A1 Elongation in the Nucleus Our studies of poly( A ) synthesis in actinomycin-treated cells led to the rather unexpected conclusion that the nuclear segments also undergo slow elongation. We had already observed that the poly(A) in the nucleus becomes unusually long in cells during starvation ( 4 7 ) . The segments in the nuclei of actinomycin-treated cells were also longer than usual (Fig. 5 ) , and slow nuclear poly(A) elongation was evident in these cells. This process could be observed as well in untreated cells. The AMP/adenosine ratio of nuclear poly(A) was rather low after a short exposure to labeled adenosine, and increased gradually with time (Fig. 9). Only after 30 minutes did it reach the maximum value of about 200. The components with shorter ratios were distributed among all size classes, unlike the situation in the cytoplasm (Fig. 10). Synthesis of the poly(A) sequence in the nucleus is a very rapid process, as indicated by the narrow size distribution of labeled material

FIG.10. Size distribution of nuclear ( B ) and cytoplasmic ( A ) poly( A ) segments with long and short labeled sequences at their 3’ terminus. Poly(A) preparations from cells labeled for 15 minutes were subjected to polyacrylamide gel electrophoresis, and the material extracted from gel slices was treated with alkali for AMP and adenosine measurements. Data are from Brawerman and Diez (53).

POLY ( A ) IN MAMMALIAN MESSENGER

139

RNA

after 5 minutes of incorporation (Fig. 5 ) . Thus the gradual rise in the AMPIadenosine ratio of nuclear poly( A ) must be due to a secondary process of slow elongation of the completed segments. Under normal conditions, the majority of the segments apparently escape this latter process by being rapidly transferred to the cytoplasm. The large size of the nuclear poly( A ) in actinomycin-treated cells can be explained by our earlier observation that a portion of these segments fail to leave the nucleus after new poly(A) synthesis has been blocked. Thus the remaining segments continue to be elongated, and are not being diluted by newly-synthesized poly( A ) of normal size.

E. Differences in Levels of Poly(A) Elongation Our original studies of poly( A ) synthesis after actinomycin treatment were carried out with Chinese hamster cells (56). The cytoplasmic elongation process proved more difficult to detect in the mouse sarcoma ascites cells as the labeling in the cytoplasm was very low. Moreover, the incorporation was greater in the nucleus than in the cytoplasm, even at very early times (Fig. 8 ) . This led us to suspect that cytoplasmic polyadenylation might be less active in the mouse ascites cells. A comparison of AMPIadenosine ratios in cytoplasmic poly( A ) showed that the Chinese hamster cells had lower ratios at early times of adenosine incorporation, an indication that they contained a greater proportion of elongating segments (Table XI). The ratios in nuclear poly(A) were also lower in these cells. While these results suggested that the Chinese hamster cells are more active in poly( A ) elongating activity, particularly in the cytoplasm, definitive conclusions could not be drawn because of lack of knowledge of the specific activities of precursors for poly( A ) synthesis, both in the nucleus and the cytoplasm. TABLE XI COMPARISON OF AMP/ADENOSINE RATIOSI N ALKALINE HYDROLYZATES OF LAnELED POLY (A) PREPARATIONS FROM MOUSE SARCOMA ASCITESCELLS A N D CHINESE HAMSTER CELLS

Nuclear poly(A) Cytoplasmic poly(A)

Mouse sarcoma

Chinese hamster

128 32

87 13

Data are from Brawerman and Dier (63). *Polysomes and nuclei prepared from mouse sarcoma ascites cells and from Chinese hamster cells in suspension culture were exposed to labeled adenosine for 10 minutes. Poly (A) preparations were subjected to alkaline hydrolysis, and AMP and adenosine were separated by paper chromatography. a

140

GEORGE BRAWERMAN

VIII. Poly(A1 and Structure of Messenger RNA A. Poly(A1 as Binding Site for Protein in mRNA It has been known for quite a while that mRNA isolated without deproteinization is associated with proteins that are tightly bound to the RNA chain. We had reasoned that these proteins must be located on noncoding portions of the mRNA chain, so as not to impede the translation process ( 59). The poly ( A) sequence, which is obviously not translated, seemed like an ideal site for this protein binding. Digestion of polysomes with RNase released a poly ( A)-containing particle that sedimented as a 12 S component (42, 51). Digestion of this structure with Pronase reduced the sedimentation coefficient to 4 S , which corresponds to that of free poly( A). Thus the 12 S structure was a poly( A)protein complex. When the steady-state poly ( A ) was measured instead of the newly synthesized material, a more heterogeneous population of poly( A)-protein complexes was evident ( Fig. 11).We also looked at the complex derived from rabbit reticulocyte ( 4 2 ) , and found it to be smaller and more homogeneous in size (Fig. 11).Thus the sedimentation characteristics of the poly( A)-protein complexes seem to be governed primarily by the nature of their poly( A ) components. At least part of the poly(A) segment in the complex is available for annealing with poly( U). This makes possible the assay of the steadystate complex. It is also possible to measure the steady-state mRNA as it exists in polysomes and in nucleoprotein particles (Fig. 12). However, the interaction between poly(A) and protein is such as to make the polynucleotide resistant to RNase T2 ( 5 1 ) . We have subsequently

*

I

IA

I

I

IE

I

i

3 2

8

loo I

3

50 5 0 1 5 2 0 5 D l 5 2 0 FRACTION NUMBER

FIG. 11. Zone sedimentation of poly( A)-protein complexes (A, mouse sarcoma; B, rabbit reticulocyte ) released from polysomes by ribonuclease treatment. Dashed lines represent steady-state complexes, assays by annealing with radioactive poly( U ) . Solid line with open circles (in A ) represents labeled complexes from mouse sarcoma cells incubated with ["Cladenosine for 2 hours. The thin line in panel B represents the mouse sarcoma steady-state complex. Data are from Jeffery and Brawerman (42).

POLY( A ) IN MAMMALIAN MESSENGER

5

10 15 x) 25

RNA

5

10

FRACTION NUMBER

141

15

20

FIG.12. Distribution of poly( A ) in unlabeled polysomes and in ribonucleoprotein particles released from polysomes. Preparations were subjected to zone sedimentation and gradient fractions were assayed by annealing with radioactive poly( U ). Ribonucleoprotein particles were released from reticulocyte polysomes by treatment with EDTA. Data are from Jeffery and Brawerman ( 4 2 ) .

observed that the whole length of the poly(A) sequence in the polysomes is resistant to snake venom diesterase, a 3’-exonuclease (unpublished). The poly(A) in the complex has also been shown to be protected from the action of polynucleotide phosphorylase ( 5 9 a ) . It is readily degraded, however, by an enzyme present in cytoplasmic extracts of mouse ascites cells. We have not been able so far to determine the nature of the protein or proteins in the complex. Blobel has observed that the poly( A ) in reticulocyte polysomes is associated with a single 72,000 daltons polypeptide (60).

B. Interaction of Poly(A) with other Polynucleotide Sequences in Polysomes In our search for functions for the poly(A) sequence, we had considered the possibility that it might influence the overall configuration of the mRNA by interacting with some other segment of this RNA. We assumed that such an association might be preserved if the RNA were fragmented by limited RNase treatment and deproteinized carefully. We undertook to isolate the poly( A ) segments from such digests, using mild conditions to avoid any dissociation between noncovalently

142

GEORGE BRAWERMAN

12

ro

xe

z t P z

$4

FRACTION NUMBER

FIG. 13. Chromatography on oligo( dT)-cellulose of poly( A)-containing mRNA fragments before and after heat dissociation. Mouse sarcoma polysomes from cells labeled with [3H]uridine and [3H]hypoxanthine were subjected to limited ribonuclease digestion, followed by deproteinization and two cycles of oligo( d T ) -cellulose chromatography. Portions of final preparations were rechromatographed after incubation at either 37°C (panel A ) or 70°C (panel B ) in 10 mM TrisCl pH 7.6, 10 mM KCI and 1 mM MgCh. Samples were applied in 50 mM TrisCl p H 7.6, 200 mM NaCl, 1 mM MgClz and 0.1% sodium dodecyl sulfate. After washing in same buffer, the concentrations of Tris and NaCl were reduced to 10 mM and 20 mM, respectively (first arrow). Poly(A)-containing material was eluted with 10 mM Tris and 0.1% sodium dodecyl sulfate (second arrow). Columns were operated at room temperature. Data are from Jeffery and Brawerman (61).

linked fragments (61). Repeated chromatography on oligo( dT)-cellulose left some polynucleotide material associated with the poly ( A ) segment, and heating caused the release of much of this material from its association with poly( A ) ( Fig, 13). The release occurred over a narrow range of temperatures, indicating a rather homogeneous duplex structure, and the T, value suggested a weak interaction (61). The same poly(A)containing structure could be obtained when the fragmentation by RNase was carried out directly on polysomes, followed by deproteinization. While the possibility of artificial associations taking place during the manipulations cannot easily be dismissed, these results provide tentative evidence for the occurrence of a poly ( A ) -polynucleotide interaction in polysomes. One interesting possibility would be that the 5’ end of the mRNA is somehow looped back and associated with the poly(A) sequence at the 3’ end. This arrangement would generate a circular structure, with the poly(A) sequence in the vicinity of the initiation site. The suggestion of polysomes being circular is not entirely novel, since such structures have been observed in electron micrographs of rabbit reticulocyte lysates ( 62).

POLY ( A ) IN MAMMALIAN MESSENGER

RNA

143

IX. Significance of Poly(A1 A. The Widespread Occurrence of the Poly(A) Sequence in mRNA In attempting to understand the physiological significance (if any) of poly(A), the extent to which this sequence is an essential feature of mRNA must be first considered. A near-universal occurrence would suggest that the poly(A) sequence has an important role to play in mRNA function or formation, Moreover, mRNA species devoid of this sequence might possess unique functional characteristics that might shed some light on possible poly(A) functions. The evidence available so far indicates that a large number of eukaryotic mRNA species, if not the great majority, contain a poly( A ) segment. This sequence has been detected in all but one of the well-investigated mRNA components (63). Most of the mRNAs of viruses that infect mammalian cells also contain a poly( A ) segment (24, 64, 65). Since the known species represent a tiny sample of cellular mRNA populations, other approaches are required to obtain more definitive information on this point. Various investigators have attempted to estimate the extent of mRNA polyadenylation by measuring the fraction of newly synthesized messenger-like RNAs that contain a poly( A ) region. Ribosomal RNA synthesis is highly sensitive to actinomycin D, and concentrations of the drug sufficient to block its labeling have relatively little effect on mRNA synthesis (63). Thus, the heterogeneous polysomal RNA component that becomes labeled in cells exposed to low levels of actinomycin is considered to represent mRNA. A substantial portion of this messenger-like RNA has been found to lack a poly( A ) sequence. There is the possibility, however, that part of the poly( A)-containing RNA may lose this segment during deproteinization ( 3 0 ) , or become fragmented. We observed that the poly ( A )-deficient messenger-like RNA of mouse sarcoma ascites cells has sedimentation characteristics similar to those of the poly ( A)-containing components, but has a distinct metabolic behavior (39). It appears in the cytoplasm without any apparent lag, and is not affected by cordycepin (Fig. 7, Table IX). The poly(A)-deficient RNA species of HeLa cells differ from the species containing poly(A) with respect to reiterated polynucleotide sequences (66). It remains to be determined whether the heterogeneous polysomal RNA species devoid of poly ( A ) are in fact mRNA components. This will require the demonstration that they contain information for the synthesis of unique proteins. The absence of poly(A) from the histone mRNAs (67, 68), and the possibility that a large number of mRNA components may lack this sequence, make it seem doubtful that the poly( A ) segment is essential for mRNA function in the cell. It may be significant that the histone

144

GEORGE BRAWERMAN

mRNAs possess some unique functional characteristics. They appear very rapidly in the cytoplasm, while the poly ( A ) -containing species are released more slowly from the nucleus (68). Also they are detected in the cell only during the period of DNA replication (69). The heterogeneous polysomal RNA devoid of poly(A) is also transferred more rapidly into the cytoplasm (39).

B. Poly(A) and mRNA Transport Some observations on the effect of cordycepin on mRNA metabolism led to the suggestion that the function of poly( A ) is to permit transfer of the mRNA from nucleus to cytoplasm. This drug can prevent the appearance of labeled mRNA in the cytoplasm, without affecting the synthesis of the presumed precursor to mRNA in the nucleus ( 5 5 ) . The drug was found subsequently to inhibit preferentially the synthesis of poly(A) in the nucleus (39, 46). Moreover, the inhibition of mRNA appearance in the cytoplasm was directed specifically at the poly ( A ) containing species (Table I X ) . Thus it was proposed that the nuclear RNA species destined to become cytoplasmic mRNA become polyadenylylated, and that the presence of this segment serves as a signal for release of the RNA molecule from the nucleus (39, 4 6 ) . It appears, however, that presence of a poly(A) segment is not an absolute requirement for release, since the histone mRNA, which lacks this sequence, is exported rapidly from the nucleus. It has also been suggested, on the basis of kinetic experiments, that not all the poly ( A ) made in the nucleus is transferred to the cytoplasm (70). Kinetic experiments of this kind, however, are difficult to interpret because of the possibility of poly(A) turnover in the nucleus and the cytoplasm (53). We have observed that a portion of the poly(A) fails to leave the nucleus in cells treated with actinomycin D, and becomes gradually elongated in this cell compartment ( 5 6 ) . It remains to be determined whether this failure to be exported represents an abnormal response to the drug, or whether there exists a class of polyadenylylated molecules that are not destined to be exported from the nucleus. While the evidence available at present does not support a specific role for poly ( A ) in mRNA transport, it remains possible that polyadenylylation is an essential component of the process of mRNA maturation in the nucleus.

C. Poly(A1 and mRNA Stability The recent studies of mRNA turnover in mammalian cells, which showed that this RNA species is far more stable than previously believed (63), prompted the search for a mechanism that would promote mRNA

POLY(A) IN

MAMMALIAN MESSENGER

RNA

145

stability. The poly( A ) sequence, located at the 3’ end of the RNA chains and covered with a protein or proteins that make it resistant to nucleases ( 51 ), could provide a barrier against degradation by 3’-exonucleases. Thus the poly( A)-protein structure could serve to prevent mRNA breakdown. The gradual decrease in poly( A ) size observed in the cytoplasm seemed at first to provide a mechanism for the orderly aging of mRNA molecules. The eventual loss of the sequence would leave the RNA chain vulnerable to exonucleases and signal the demise of the mRNA molecule ( 4 7 ) . It was pointed out, however, that mRNA decays in an exponential fashion, with every molecule (young and old) equally susceptible to inactivation (71 ). This seemed to be incompatible with an orderly aging process. The gradual reduction in poly( A ) size is rapid at first, but seems to slow down or cease altogether after a few hours, leading to still substantial steady-state poly( A ) sizes ( 4 2 ) . The latter could be destroyed in a random fashion. Such a scheme would be compatible with the observed stochastic decay of mRNA. Not all the available experimental evidence supports a role for poly(A) in promoting mRNA stability. Histone mRNA, which has no poly(A), is apparently quite stable in the cytoplasm ( 7 1 ) . Moreover, a short-lived class of poly( A)-containing RNA has been detected in HeLa cells ( 7 2 ) . It is conceivable that the metabolic stability of histone mRNA is determined by a different mechanism. It may be significant in this respect that the histone mRNAs do not seem to decay in a random fashion, but become rapidly inactivated after completion of the period of DNA synthesis ( 7 1 ) . As for the short-lived poly( A)-containing species, it is conceivable that their poly( A ) segment could be more exposed to nucleolytic attack. Experiments have been designed to determine the effect of poly( A ) on niRNA stability in uitro. Globin mRNA, from which the poly(A) sequence had been removed enzymically, was reported to function for shorter periods of time in cell-free systems for polypeptide synthesis (73, 7 4 ) . The effect, however, was rather small. A far more dramatic effect was observed when these RNA components were injected into frog eggs ( 7 4 ) .The latter result was interpreted by the authors as indicating a considerably reduced stability for the globin mRNA devoid of poly( A). Actual measurements of globin mRNA content after prolonged incubation of the injected eggs showed that the molecules lacking poly( A) are degraded far more rapidly ( 7 4 ~ ) .

D. Poly(A1 and mRNA Translation The primary function of mRNA is to serve as template for polypeptide synthesis. Is the poly( A ) sequence required for normal translation? Again the histone mRNAs show that this is not the case. These poly( A ) -

146

GEORGE BRAWERMAN

deficient RNA species are translated in uitro as well as in uiuo. Globin mRNA freed of the poly(A) sequence can also be translated in vitro (73-75). The studies of translation of globin mRNA injected into frog oocytes, however, revealed a striking difference between poly (A)-containing and poly ( A ) -deficient molecules. Translation with the polyadenylylated molecules seemed far more effective ( 7 4 ) . While this latter effect was interpreted in terms of a greater stability for the poly(A)containing molecules, a greater efficiency for the utilization of host ribosomes seems equally plausible. The above studies raise the possibility that the currently available cell-free systems may not be adequate to reflect the conditions for translation within the cells. Thus any subtle effect of the poly( A ) sequence might not be detectable in uitro. Some sort of participation of the poly( A ) sequence in the translation process was also suggested by studies with sea urchin eggs. Fertilization leads to the uptake of previously inactive messenger RNA into polysomes. This is accompanied by a large increase in the size of the poly(A)-sequence in the mRNA (57, 5 8 ) . It is unlikely, however, that this poly( A ) elongation is required for mRNA activation, since inhibition of this process by high doses of cordycepin does not prevent the conversion to active mRNA (76).

X. Conclusion The poly(A) segment in mRNA provides a dramatic example of an accessory polynucleotide sequence that is clearly not involved in the primary function of the RNA chain. All RNA types seem to possess extra sequences of this kind. The first ones to be clearly defined are the ribosomal binding sites (or initiation sites) on bacteriophage RNA ( 7 7 ) . The precursor molecules to ribosomal RNA, messenger RNA and transfer RNA also contain large polynucleotide segments that are eliminated during the maturation process (54, 78). It is quite likely that these segments play some as yet undefined function, or functions. What is unique about the poly( A ) segment is that it represents a homopolymeric sequence and that its size is subject to variation. This segment provides a highly specialized structure that identifies a class of RNA in the cell. It does not seem to allow for any diversity that would permit discrimination between different species of mRNA. This type of sequence is clearly useful to the biochemist, since it permits the detection and isolation of a unique RNA class. It is presumably also useful to the cell, although its prezise function remains elusive. It is likely that further work will reveal a variety of regulatory functions for the accessory polynucleotide sequences in the cell.

POLY(A) IN

MAMMALIAN MESSENGER

RNA

147

ACKNOWLEDGMENTS The studies in the author’s laboratory described in this article are the result of collaborations with Drs. Jerome Eisenstadt, Anthony Hadjivassiliou (then a graduate student), Jurgen Drews, Se Yong Lee, Velibor Krsmanovic, Jozef Mendecki, Sau Wah Kwan, William Jeffery and Julio Diez. The work was supported by grants from the U.S. Public Health Service, from the National Science Foundation, and from the Anna Fuller Fund.

REFERENCES B. H. Hoyer, B. J. McCarthy and E. T. Bolton, Science 140, 1408 (1963). N. P. Salzman, A. J. Shatkin and E. D. Sebring, J M B 8, 405 (1964). E. Henshaw, M. Revel and H. Hiatt, J M B 14,241 (1965). G. P. Georgiev, 0. P. Samarina, M. I. Lerrnan, M. N. Srnirnov and A. N. Severtzov Nature (London) 200, 1291 ( 1963). 5. L. Harel, J. Harel, A. Boer, J. Imbenotte and N. Carpeni, BBA 61, 34 (1964). 6. C. Lamar, M. Prival and H. Pitot, Cancer Res. 26, 1909 ( 1966). 7 . J. M. Eisenstadt and G. Brawerman, J M B 10,392 (1964). 8. G. Brawerman and E. Chargaff, BBA 31, 172 (1959). 9. G. Brawerman, BBA 72,317 (1963). 10. G. P. Georgiev and V. L. Mantieva, BRA 61, 153 ( 1962). 1 1 . G. Brawerman, BBA 76, 322 (1963). 12. A. Sibatani, K. Kimura, K. Yamana and T. Takahashi, Nature (London) 186, 215 (1960). 13. A. Sibatani, S. R . deKloet, V. G. Allfrey and A. E. Mirsky, PNAS 48, 471 (1962). 14. G. Brawennan, L. Gold and J. Eisenstadt, PNAS 50, 630 (1963). 15. A. Hadjivassiliou and G. Brawerman, Bchem 6, 1934 (1967). 16. A. Hadjivassiliou and G. Brawerman, J M B 20, 1 ( 1966). 17. M. Edmonds and R. Abrams, JBC 238, PC, 1186 ( 1963). 18. M. Edmonds and R. Abrams, JBC 235, 1142 ( 1960). 19. P. R. Venkataraman and H. R. Mahler, JBC 238, 1058 (1963). 20. M. E. Gottesman, Z. N. Canellakis and E. S. Canellakis, BBA 61, 34 (1962). 21. J. T. August, P. J. Ortiz and J. J. Hunvitz, JBC 237, 3786 ( 1962). 22. M. Edmonds and M. G. Cararnela, JBC 244, 1314 (1969). 23. L. Lim and E. S. Canellakis, Nature (London) 227, 710 ( 1970). 24. J. Kates, CSHSQB 35, 743 (1970). 25. S. Y. Lee, J. Mendecki and G. Brawerman, PNAS 68, 1331 (1971). 26. M. Edmonds, M. H. Vaughan Jr. and H. Nakazato, PNAS 68, 1336 (1971). 27. J. E. Darnell, R. Wall and R. J. Tushinski, PNAS 68, 1321 (1971). 28. G. Brawerman, J. Mendecki and S . Y. Lee, Bchem 11,637 (1972). 29. G. Brawerman. Methods Cell Biol. 7. 1 1973). 30. R. P. Perry, J. La Torre, D. E. Kelley and J. R. Greenberg, BBA 262, 220 (1972). 31. H. Aviv and P. Leder, PNAS 69, 1408 ( 1972). 32. J. Gorski, M. R. Morrison, C. G. Merkel and J. B. Lingrel, J M B 86, 363 (1974). 33. J. N. Vournakis, R. E. Gelinas and F. C. Kafatos, Cell 3,265 (1974). 34. G. C. Rosenfeld, J. P. Comstock, A. R. Means and B. W. O’Malley, BBRC 47, 387 ( 1972). 35. A. R. Means, J. P. Comstock, G. C. Rosenfeld and B. W. O’Malley, PNAS 69, 1146 (1972). 1. 2. 3. 4.

148

GEORGE BRAWERMAN

36. R. Palacios, D. Sullivan, N. M. Summers, M. L. Kiely and R. T. Schimke, JBC 248, 172 ( 1973). 37. N. Sullivan and W. K. Roberts, Bchem 12,2395 (1973). 38. P. A. Kitos, G. Saxon and H. Amos, BBRC 47, 1426 ( 1972). 39. J. Mendecki, S. Y. Lee and G. Brawerman, Bchem 11, 792 (1972). 40. R. Sheldon, C . Jurale and J. Kates, PNAS 69,417 (1972). 41. J. R. Fresco and P. Doty, JACS 79,3928 ( 1957). 42. W. R. Jeffery and G. Brawerman, Bchm 13,4633 ( 1974). 43. G. R. Molloy, M. B. Sporn, D. E. Kelley and R. P. Perry, Bchem 11, 3256 (1972). 44. R. Sheldon, J. Kates, D. E. Kelley and R. P. Perry, Bchem 11, 3829 (1972). 45. H. Nakazato, D. W. Kopp and M. Edmonds, JBC 248, 1472 (1973). 46. J. E. Damell, L. Philipson, R. Wall and M. Adesnik, Science 174, 507 (1971). 47. G. Brawerman, Mol. Biol. Rep. 1, 7 (1973). 48. M. Sussman, Nature (London) 225,1245 ( 1970). 49. D. Sheiness and J. E. Darnell, Nature N B 241,265 (1973). 50. J. R. Greenberg and R. P. Perry, BBA 287,361 ( 1972). 51. S.-W. Kwan and G. Brawerman, PNAS 69,3247 ( 1972). 52. J. N. Mansbridge, J. A. Crossley, W. G. Laynon and R. Williamson, E J B 44, 261 (1974). 53. G. Brawerman and J. Diez, Cell 5, 271 (1975). 54. R. A. Weinberg, ARB 42, 329 ( 1973). 55. S. Penman, M. Rosbach and M. Penman, PNAS 67, 1878 (1970). 56. J. Diez and G. Brawerman, PNAS 71,4091 (1974). 57. I. Slater, D. Gillespie and D. W. Slater, PNAS 70, 406 ( 1973). 58. F. H. Wilt, PNAS 70,2345 ( 1973). 59. S. Y. Lee and G. Brawerman, Bchem 10,510 (1971). 59a. H. Soreg, U. Nudel, R. Solomon, M. Revel, and U. Z. Littauer, JMB 88, 233 (1974). 60. G. Blobel, PNAS 70, 924 ( 1973). 61. W. Jeffery and G. Brawerman, Bchem 14, 3445 ( 1975). 62. A. P. Mathias, R. Williamson, H. E. Huxley and S. Page, J M B 9, 154 ( 1964). 63. G. Brawerman, ARB 43,621 ( 1974). 64. Y. Yogo and E. Wimmer, PNAS 69, 1877 (1972). 65. M. Soria and A. S. Huang, J M B 77,449 ( 1973). 66. C. Milcarek, R. Price and S. Penman, Cell 3, 1 (1974). 67. M. Adesnick and J. E. Darnell, J M B 67, 397 (1972). 68. J. R. Greenberg and R. P. Perry, J M B 72, 91 ( 1973). 69. E. Robbins and T. W. Borun, PNAS 57, 409 (1967). 70. R. P. Perry, D. E. Kelley and J. LaTorre, J M B 82,315 ( 1974). 71. R. P. Perry and D. E. Kelley, JMB 79, 681 ( 1973). 72. L. Puckett, S. Chambers and J. Darnell, PNAS 72, 389 (1975). 73. A. E. Sippel, J. G. Stavrianopoulos, G. Schutz and P. Feigelson, PNAS 71, 4635 (1974). 74. G . Huez, G. Marbaix, E. Hubert, M. Leclercq, U. Nudel, H. Soreq, R. Salomon, B. Lebleu, M. Revel and U. Z. Littauer, PNAS 71,3143 (1974). 74a. G. Marbaix, et al., PNAS 72,3065 ( 1975). 75. R. Williamson, J. Crossley and S. Humphries, Bchem 13, 703 (1974). 76. A. Mescher and T. Humphreys, Nature (London) 249, 138 ( 1974). 77. J. A. Steitz, Nature (London) 224, 957 ( 1969). 78. R. H. Burdon, This Series, 11, 33 (1971).

Polyadenylate Polymerases MARYEDMONDS Department of Life Sciences Faculty of Arts and Sciences University of Pittsburgh Pittsburgh, Pennsylvania AND

MARY ANN WINTERS Chemistry Department Seton Hill College Greensburg, Pennsylvania

. . . . . . . . . . . . . . . . Polymerase . . .

. . . . . B. 1960-1965 . , C. 1965-1970 . . D. 1970-1975 . . 111. The Poly( A ) Reaction A. Substrates . . B. Ion Requirements . . . C. Polynucleotide Requirement . D. Primer Specificity . . . E. Reaction Products . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F. Kinetics of Poly(A) Synthesis . . . . . IV. Poly( A) Polymerase Proteins . . . . . . V. Multiple Poly(A) Polymerases . . . . . . VI. Cellular Location of Poly(A) Polymerases . . . . VII. Poly(A) Polymerases as a Subunit of RNA Polymerase . VIII. Viral Poly( A) Polymerases . . . . . . . IX. Regulation of Poly( A ) Polymerases . . . . . X. Poly(A) Polymerases as Polyadenylylating Reagents . XI. Summary. . . . . . . . . . . References . . . . . . . . . . I. Introduction 11. Historical . A. 1955-1960.

.

. . .

. . . . . . . .

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

. .

.

. .

. .

. .

. . . . . . . . .

.

149 150 151 152 155 156 157 157 157 158 159 164 164 165 166 168 169 171 173 174 175 176

1. Introduction The discovery of poly( A ) sequences covalently bound to the 3’ ends of the rapidly labeled polydisperse RNA molecules of the nucleus (hnRNA), the messenger RNA (mRNA) molecules of polysomes of animal cells, and to the RNA of many animal viruses as well, immediately stimulated investigations of the mechanism of their biosynthesis. The experimental evidence accumulated at this time supports a nontranscriptional process for the biosynthesis of these poly( A ) sequences. Ac149

150

MARY EDMONDS AND MARY ANN WINTERS

tually, a reaction that might generate posttranscriptionally a poly ( A ) sequence on the 3’ end of an RNA molecule had been discovered in the late 1950s. During the early 1 9 6 0 ~several ~ laboratories described reactions in both animal cells and bacteria (1-7) that catalyze the addition of AMP residues from ATP to the 3’ ends of a variety of RNAs and oligonucleotides. The common characteristics of such reactions were a marked specificity for ATP as substrate and a requirement for an RNA primer to which the poly( A) was added. Interest in these enzymes subsided rather quickly since poly(A) sequences in nucleic acid were not known. It was usually assumed that these in uitro reactions had no cellular counterpart. Some support for this view came from the poly( A ) synthesis that occurs when ATP reacts with Escherichia coli RNA polymerase in the absence of other nucleoside triphosphates (8, 9). In this reaction, it was possible to generate in uitro poly(A) sequences of more than 60 nucleotides from synthetic oligomers of dTMP as short as 10 or 12 (10). Since it is difficult even now to imagine a role for such a reaction in the cell, it appeared that RNA polymerase, and perhaps poly( A ) polymerases as well, catalyzed reactions in uitro that probably did not occur in uiuo. Although sporadic reports appeared throughout the late 1960s (11-19), interest in these reactions were revived only when poly(A) sequences were found, in 1971, to be covalently bound to hnRNA and mRNA molecules of animal cells (20-22). The apparently universal distribution of poly(A) sequences and of poly(A) polymerases points to a significant but as yet unknown cellular function for poly( A ) sequences that makes it appropriate to examine the early work and the current status of the polymerases that catalyze their formations. We believe this to be the first comprehensive review of this class of enzymes, more accurately named “RNA terminal riboadenylate transferases,” referred to throughout this review by the trivial name “poly( A ) polymerase.”

II. Historical The modern era of investigations of nucleic acid biosynthesis began with the discovery of Grunberg-Manago and Ochoa in 1955 ( 2 3 ) of an enzyme activity in extracts of Azotobacter uinelandii that polymerized nucleoside diphosphates into phosphodiester bonds characteristic of those found in natural RNAs. Shortly afterward, Kornberg and co-workers reported a DNA polymerase activity in extracts of E . coli ( 2 4 ) . These experiments not only removed all doubts of the feasibility of studying polynucleotide biosynthesis with isolated enzyme systems, but they provided the basic experimental approach that became the prototype for subsequent investigations in the field: to measure the incor-

POLYADENYLATE POLYMERASES

151

poration of radioactivity from labeled mononucleotide into a polynucleotide readily separable from the precursor by its relative insolubility at low pH. The simplicity of this approach encouraged many biochemists to begin the study of RNA and DNA synthesis with extracts of both animal cells and bacteria. However, the studies of RNA synthesis at that time were severely hampered by a lack of knowledge of the variety of RNA species occurring in cells, such as transfer RNA and ribosomal RNA. This led to considerable confusion in the interpretation of ribonucleotide incorporation data, which was only gradually dispelled as the identification and characterization of natural RNA species was achieved in the late 1950s.

A. 1955-1960 It is likely that most of the RNA synthesis in animal cells reported during this period resulted primarily from the activity of tRNA terminal nucleotidyltransferases or of poly( A ) polymerases, or, in the case of bacterial extracts, of polynucleotide phosphorylases. The relative ease of solubilization of these enzymes and the limited availability at that time of labeled nucleotides other than those of adenine favored the detection of these activities rather than of the RNA polymerase activity actually sought by most investigators. Thus, although poly ( A ) polymerases probably were detected before RNA polymerases, there was some delay in reporting it until the properties of ,such novel reactions were well established. The first detailed description of a poly( A ) polymerase appeared in 1960 ( I ) at about the same time as reports of RNA polymerase activity in both rat liver nuclei (25) and E . coli appeared (26, 27). The crucial evidence that a long sequence of AMP residues was polymerized by partially purified extract,s of thymus nuclei was obtained from a nearest-neighbor analysis of the reaction product synthesized from [d2P]ATP. It was found that 99% of the incorporated 32Pwas in the dinucleotide A-A. The average length of the sequences had been estimated in a separate experiment as 50-100 nucleotides by end-group analyses of the product synthesized from W-labeled ATP (28). This activity had to be distinguished from other polynucleotide syntheses under investigation at the time, particularly polynucleotide phosphorylase. That the latter activity strongly influenced the early work on poly( A ) polymerases is evident in the use of ,adenosine diphosphate as substrate in the first measurements of polyribonucleotide synthesis in Ehrlich ascites cell extracts (29). A detailed analysis of the energy requirements for nucleotide incorporation eventually established that ATP is the true substrate for the reaction in both ascites cell extracts (29) and thymus nuclei ( I ) .In studies in animal cells, polynucleotide phosphorylasc presented few experimental difficulties, since there is no

152

MARY EDMONDS AND MARY ANN WINTERS

clear evidence that it is present. This is not the case in bacteria, however, and recent evidence suggests it may still complicate studies of the poly(A) polymerase of E . coli, particularly when it is present together with an ATPase (30). Another reaction discovered during this period was that of tRNA terminal nucleotidyltransferase, first detected by Heidelberg et al. (31 ), and subsequently shown by Hecht and her co-workers to be responsible for the C-C-A sequence at the 3' end of transfer RNA molecules (32). Since this reaction accounted for most of the ATP incorporated into RNA with extracts of Ehrlich ascites cells, a somewhat laborious assay to avoid it was devised, which measured AMP incorporated into internal positions of polynucleotides. It was soon clearly shown that a different enzyme system is responsible for the nonterminal incorporated of ATP into polyribonucleotides in ascites cells (29), and that it is present in calf thymus nuclear extracts as well. Since the latter were apparently free of tRNA nucleotidyltransferases, ascites tumor cell extracts were replaced by thymus nuclear extracts as the source of poly( A ) polymerase in subsequent investigations ( 1 ),

B. 1960-1965 Prior to 1965, ,several laboratories identified poly ( A ) polymerases in a variety of animal tissues and cells, as well as in E . coli (1-7),and clearly differentiated the reaction catalyzed by these enzymes from other well-recognized polynucleotide syntheses. Modest purifications of poly( A ) polymerases were reported from calf thymus nuclei ( I ) , chick embryo allantoic membranes ( 6 ) and rat liver cytoplasm ( 7 ) . The enzyme from E . coli was reported to have been crystallized ( 3 ) . Most preparations showed a marked preference for ATP as the substrate and none for deoxyribonucleoside triphosphates. The requirement for a divalent cation was filled in most preparations by Mg", although a mixture of Mg" and MnZ+was more effective for E . coli polymerase. However, a well-documented comparative study of the effects of Mg2+and Mn2+ with the chick chorioallantoic membrane enzyme ( 6 ) showed that in this case MgZt was essentially inactive relative to Mn". Replacing Mg'+ with Mn2+ also resulted in marked stimulations of poly(A) synthesis in the rat-liver cytoplasmic system ( 7 ) and by the calf-thymus nuclear enzyme (33). There was little effort to interpret these divalent cation effects at the time, although the existence of two different poly(A) polymerases in calf-thymus nuclear extracts was proposed on the basis of differences in properties of the reaction observed in the presence of Mg*+ or Mn2+ (33). The discovery that the substrate specificity of E . coli DNA polymerase could be altered to allow ribonucleotides to be

POLYADENYLATE POLYMERASES

153

incorporated into DNA by replacing Mg2+ with Mn2+ ( 3 4 ) fostered a cautious interpretation of divalent ion effects. A general description of the reaction mechanism as one involving addition of poly(A) to a primer rather than as being generated by a template-copying mechanism emerged in these early investigations. The RNA molecules most effective as primers could hardly be envisioned as templates for complementary base-pairing and, for several polymerase preparations, poly( U ) , the obvious candidate, was often a poor primer or an inhibitor (6, 7, 3 3 ) . The question of a template versus a primer mechanism was closely tied to that of initiation of poly(A) sequences de nooo. The earliest studies relied almost entirely on the analysis of isotope-labeled end groups to study this question. A de nooo synthesis would be expected to produce an equivalence of label in the nucleotides at the 5’ and 3’ ends of the product, while covalent attachment to a primer would label nucleotides at the 3’ ends only. It should be noted that at this time, when the reaction mechanisms for all polynucleotide syntheses were only beginning to be defined, alternatives that today might be considered highly improbable could not be excluded. For example, there was no reason then to exclude n priori the addition of the 3’ OH group of the mononucleotide to the 5’ phosphate end of a primer since the direction of RNA chain extension in uioo had not yet been established. The evidence, although clearly favoring the addition of AMPS to the 3’ ends of primers, was not entirely conclusive in all cases and in some preparations, most notably, E . coli, was apparently contradictory. In the latter case, evidence for an equal labeling of adenosine and 3’,5’bisphosphoadenosine released from a reaction product of several hundred AMP residues pointed to a de novo synthesis of poly(A) chains ( 3 ) . This did, however, seem to contradict an earlier brief report of the absence of labeled 5’ ends in the product isolated from an E. coli poly( A ) polymerase preparation ( 4 ) . It was well understood at the time that endonucleolytic cleavages occurring during poly (A) synthesis might account for label found in nucleotides of the type derived from the 5’ ends of polynucleotide products. Contaminating phosphomonoesterases might also account for a lack of label in such products. In 1966, a new study of the E. coli poly( A ) polymerase failed to detect evidence for initiation of the 5’ termini of the poly(A) sequences and showed that poly ( A ) chains were most probably covalently attached to the RNA molecules added to the reaction (12). Throughout this period, while the coding problem was being intensely investigated, it was natural to assign to poly( A ) a role in protein synthesis. Coding for polylysine for instance, was seriously considered. The tendency for the poly( A ) polymerase activity to be tightly bound

154

MARY EDMONDS AND MARY ANN WINTERS

to ribosomes in E . coli (3) offered support for such involvement until it was eventually shown that this association most likely arose during cell fractionation ( 1 2 ) . However, two types of experimental data that did appear to be of fundamental biological importance and that did much to sustain interest in continued research on poly( A ) synthesis were reported in these early studies. One was the fact that infection of E . coli with T2, T4 or T5 phage causes an immediate and marked loss of poly(A) polymerase activity, which was not observed for several other enzymes prominent in RNA metabolism in E . coli (35).The decline appeared to be caused by the rapid synthesis of a protein inhibitor of poly(A) polymerase. Although even today these results remain unexplained, at the time they were instrumental in suggesting a fundamental role for these reactions in macromolecule metabolism. Another stimulus to continued study of poly( A ) polymerase was the discovery of poly( A ) sequences in calfthymus nuclear extracts (36). This first evidence for the existence of AMP-rich sequences in cells resulted indirectly from studies aimed at detecting a primer requirement for the calf-thymus poly ( A ) polymerase. A boiled extract of this crude enzyme was able to restore the activity lost when the partially purified enzyme was treated with RNase. Dialysis of the extract, after extensive digestion with RNase A and RNase T1, left a nondialyzable RNA core that was 87% AMP and that retained full priming activity for poly(A) synthesis. This result was confirmed shortly thereafter when a poly( A ) fraction lacking other nucleotides was isolated from calf thymus nuclear extracts on oligo ( dT ) -cellulose columns ( 3 7 ) . A review of this early work would be incomplete without mention of the other specific homopolynucleotide polymerases often present in the same extracts with poly(A) polymerases (38-49). A poly(U) polymerase was separated from the poly(A) polymerase found in extracts of rat liver cytoplasm ( 4 1 ) , and a poly(C) polymerase was separated from the ATP polymerase of calf thymus extracts ( 4 2 ) . Since these polymerases have yet to be linked to the presence of poly( U ) or poly( C ) sequences in cells, they have not been as extensively studied, although investigations of these (43-47) and poly(G) polymerases as well (48, 49) continue. At the time they were discovered, it was logical to expect that polymerases specific for different nucleoside triphosphates occurring in the same tissue extracts might be functionally related, perhaps in some coordinated reaction. One experiment to test this idea was carried out by mixing the complete poly( A ) and poly( C ) polymerase systems in one experiment with [,x-~~P]ATP and unlabeled CTP as substrates and

POLYADENYLATE POLYMERASES

155

in another with [ W ~ ~ P I C T and P unlabeled ATP (50). Failure to detect label in the dinucleotide C-A in the first case, as well as in A-C in the second, demonstrated a lack of interaction between these two systems. All label was recovered instead in A-A in the first case and C-C in the second. Although these results discouraged further tests of interactions between specific polymerases at that time, such possibilities should perhaps be reconsidered once purified polymerases become available. It is of interest to note that rather highly purified preparations from yeast that polymerize AMP from ATP also appeared to polymerize significant quantities of CMP from CTP, but not UMP or GMP (51). It was suggested that this may be the activity of a single enzyme able to synthesize a copolymer from ATP and CTP.

C. 1965-1970 In the late 1960s, little new information was reported for poly(A) polymerases although other sources continued to be found (11-16, 18, 19). Some of these, such as nuclear extracts of sea urchin embryo (14) and intact rat liver nuclei (17,1 8 ) , are of current interest in the light of studies of the metabolism of poly( A)-containing RNA molecules in the nucleus and cytoplasm. Few improvements in the degree of purification from either animal cells or bacteria were reported. The marked and rapid inhibition of poly(A) polymerase activity accompanying T4 bacteriophage infection was confirmed for E . coli, as well as for Shigella dysenteriae by another group, who also observed the apparent synthesis of a protein inhibitor of poly(A) polymerase upon T4 infection (15).Shortly thereafter, an interesting attempt was made to learn whether the inhibitor was coded for by the phage genome (52). Sixteen different T4 amber mutants were tested for their ability to inhibit poly( A ) polymerase, but none differed significantly from normal T4 phage in this respect. The most significant work of this period was not directed at polynierases, but at the detection and characterization of poly( A) sequences in animal cells and viruses. Most of these results were by-products of studies of RNA metabolism in which AMP-rich polynucleotides turned up in RNA fractions from sources as diverse as rat liver cytoplasmic membranes (5.3), pea roots (54),and reovirus (55). In 1969, two of the laboratories that had participated in the early studies of poly( A ) polymerases reported investigations of the characterization of AMP-rich fractions in animal cells. A nuclear RNA fraction of Ehrlich ascites cells was found to contain close to 1.0%of a homogeneous population of poly(A) sequences about 200 nucleotides in length

156

MARY EDMONDS AND MARY ANN WINTERS

( 5 6 ) . In a study of RNA metabolism in mouse liver (57-59), an AMPrich fraction, presumed to be in mRNA, was found on polyribosomes (59). AMP-rich fractions were also detected in the nucleus in these experiments. This group also found an AMP-rich sequence in the globin mRNA fraction of rabbit reticulocytes (60). These were forerunners of the experiments that eventually clarified the significance of AMP-rich sequences in animal cells by establishing their covalent attachment to hnRNA and to mRNA (20-22).

D. 1970-1975 The interest in poly( A ) polymerases stimulated by the discovery of poly( A ) sequences in hnRNA and mRNA increased when they were found at the 3’ ends (61-63) and especially when evidence for their posttranscriptional addition was reported ( 6 4 ) . It becomes difficult to resist the conclusion that this addition represents the true cellular function of poly ( A ) polymerases. New sources of the polymerase activity continue to be found. Some, such as those from cultured animal cells (65-68), yeast (69, 70), mitochondria (71, 72), and animal viruses (73-79) offer better systems for relating these activities more directly to poly( A ) synthesis in uiuo, particularly those associated with animal viruses (see Section VIII). Recently, several laboratories have concentrated on increasing the purification of the enzyme to define the properties of the reaction more accurately (80, 81) . Two large-scale purifications from calf-thymus tissue have been reported (82, 83), and smaller, but significant, purifications of the activity from Krebs mouse tumor cells (65, 66) and cultured HeLa cells (67) have been achieved. An improved purification of the activity from E . coli has also been reported ( 3 0 ) .Other recent information has raised questions as to the number of different poly(A) polymerases in the cell and their intracellular location (67, 8486). From this historical review of a reaction discovered 15 years ago, it is apparent that our knowledge remains limited primarily to basic descriptions of poly( A) polymerase reactions from many cell types. Efforts have only recently been made to study in depth either the mechanism of the reaction or its biological function. This being the case, a review might be considered premature at this time, were it not for the intense current interest in the function of poly(A) sequences. A collation and critical evaluation of available data on the enzymic reactions presumed to carry out polv( A ) synthesis in the cell may stimulate new ideas of how to relate poly(A) polymerases to cellular poly(A) synthesis and may encourage more detailed study of the polymerase proteins.

157

POLYADENYLATE POLYMERASES

111. The Poly(A1 Polymerase Reaction Poly( A ) polymerases ( EC 2.7.7.19) are enzymes essentially specific for polymerizing the AMP of ATP into a polyadenylate sequence. Activities with this singular property, regardless of source or of any specific or unusual requirements for primers, templates, ions, etc., are included. RNA polymerases, polynucleotide phosphorylases, or preparations that incorporate significant amounts of nucleotides other than AMP are excluded by this definition. The overall reaction catalyzed by poly( A ) polymerases, with RNA' (or a polynucleotide) present, can be formulated as follows: RNA

+ n ATP

Mga+ (or Mnl+)

* RNA-(A).

+ n PPi

Although the properties of polymerases from most sources are compatible with this reaction, the consideration of evidence for it relies heavily on data reported for the most highly purified preparations, primarily those from calf thymus (81-83), Krebs ascites cells (65, 66) and E . coli ( 3 0 ) , since some assessment of their contamination with nucleases and polynucleotides has been reported. However, less purified preparations from other sources are not necessarily ignored, since they have often been studied in ways that give special insights into the many questions to be posed about poly(A) polymerases in this review.

A. Substrates A marked specificity for ATP characterizes poly( A ) polymerases from all sources. Even after relatively little purification, the utilization of other ribonucleoside triphosphates either singly or in the presence of ATP is usually less than 1%of that of ATP. The same is true for dATP and rADP. Two types of ATP analogs have been tested as substrates. The ATP analog adenosine 5'- ( p,y-methylene) triphosphate has been polymerized efficiently into poly( A ) with a preparation from quail oviduct (87). Cordycepin triphosphate ( 3'-dATP), which should terminate polynucleotide chains after a single 3'-dAMP residue is attached to the primer, has also been observed to be incorporated at the 3' terminus (81). However, it has not been shown that this is the mechanism by which cordycepin inhibits poly ( A ) synthesis in animal cells. Cordycepin triphosphate itself is a relatively poor competitive inhibitor of ATP for the poly ( A ) polymerases of HeLa cells and maize ( 88).

B. Ion Requirements The requirement for divalent cations can be filled by Mg" or Mn2+ or some combination of the two, depending on the source of the activity.

* See article by Brawerman in the volume [Ed.].

158

MARY EDMONDS AND MARY ANN WINTERS

The optimal divalent ion concentration is related stoichiometrically to ATP, being either equimolar or twice that of the ATP concentration. The relative effectiveness of Mg2+or Mn2+for difficult preparations (Table I ) turns out to be difficult to interpret for reasons discussed below (Section 111, b ) , but the data may be of considerable importance to the question of distinctive poly( A ) polymerases requiring either Mg" or Mn2+.Factors such as the ATP concentration, the nature of the primer and its concentration, as well as the presence of nucleases, need to be considered for evaluation of the relative effects of Mg2+ and Mn2+. A few examples best illustrate this point. An early study of the polymerase activity in calf thymus nuclear extracts (33) showed a marked stimulation of poly(A) synthesis at low ATP concentrations when MnZ+replaced Mgz+. At the higher ATP concentrations normally used in assaying Mg?+-activatedreactions, this stimulation disappeared. A poly( A ) polymerase highly purified from whole calf thymus extract is far more active in MnZ+than Mg2+when the reaction is primed with low levels of p(A), (81). As the primer concentration is raised, there is an increased ability of Mg2+to replace Mn2+.A kinetic analysis showed the K, for this primer to be 4 times lower in Mn?+,but V,,,,, was the same with either Mg2+or Mn2+.Until the natural primers for poly( A ) polymerases are identified and studied in similar fashion, the biological significance of these observations remains unclear. Divalent cations may also have different effects on contaminating nucleases, which may obscure the true divalent cation requirements of the polymerase. For example, the partially purified calf-thymus poly ( A ) polymerase from nuclei contained a Mg2+-activatednuclease particularly effective in degrading poly ( A ) , which was completely inhibited by Mn2+ either in the presence or in the absence of MgZ+(33). Poly ( A ) polymerases are particularly sensitive to ionic strength. Salt concentrations required to activate RNA polymerases ( 9042) usually inhibit poly(A) polymerases. It is usual for concentrations of NaCl or KC1 as low as 100 mM to inhibit most poly(A) polymerase preparations (71, 73,80, 89).

C. Polynucleotide Requirement Almost all poly ( A ) polymerizing systems require a polynucleotide for optimal activity, although it is usually evident only after endogenous nucleic acids have been removed by ion-exchange chromatography. It is now clear that the added polynucleotide functions as a primer for the covalent attachment of the poly( A ) sequences, although at one time a template mechanism was given serious consideration (see Section 11, B ).

POLYADENYLATE POLYMERASES

159

Since it was first shown ( 1 2 ) that the poly(A) synthesized by the E . coli poly(A) polymerase cosediments in sucrose gradients with the added RNA, both direct and indirect evidence has been accumulated, using preparations from animals, bacteria and yeast, to support the conclusion that poly ( A ) is covalently attached to the exogenous RNA. However, these early sedimentation experiments are open to the criticism that, since denaturing conditions were not used, noncovalent interactions between poly ( A ) and added RNAs were not excluded. A more direct approach to demonstrating a covalent link was shown with a poly( A ) polymerase in maize that uses transfer RNA as a primer ( 94). When a tRNA preparation specifically labeled in the 3’-terminal adenosine was added to this enzyme, label was gradually converted from this position to an internal one during the course of poly(A) synthesis. When the labeled product subsequently recovered from this experiment was heat-denatured, it electrophoresed with the added tRNA primer. A highly purified poly ( A ) polymerase from calf thymus attaches much longer poly(A) sequences to tRNAs added to the reaction ( 8 3 ) . In this case, the poly( A ) not only electrophoresed with the tRNA, but caused a marked reduction in the mobility of the tRNA as Poly(A) synthesis proceeded, as might be expected if a poly(A) tail of 50-100 nucleotides were attached to tRNA (Fig. 1 ) . Finally, simple, but less direct evidence for a covalent attachment has come from many studies of the effects of modifications at the 3’-terminal ribose group of different primers. The presence of a phosphate on this ribose blocks the priming activities of specific adenylate oligomers (19, 66, 69, 80, 83). Periodate oxidation reduces markedly the activity of tRNA as a primer for both E . coli (30) and Krebs ascites tumor cell polymerases (66). The oligo( A ) synthesis carried out by virions of reovirus (see Section VIII) is apparently an authentic case of an unprimed poly(A) [(or oligo( A)] synthesis. Unprimcd poly( A ) synthesis also occurs with a highly purified polymerase from E . coli (95). Although in this case exogenous RNA stimulated poly( A ) synthesis, the newly made poly( A) was not attached to the added RNA. Several other purified poly(A) polymerase preparations show a low level of activity without added primers (30, 66, 67, 72, 81). It will be difficult to determine whether this is an inherent activity of the polymerase or merely the result of contaminating polynucleotides.

D. Primer Specificity The most striking fact of primer specificity seems to be the lack of it. This does not mean that all poly(A) polymerase preparations

160

MARY EDMONDS AND MARY ANN WINTERS

,

I '

"

10

20 30 FRACTIONS

40

FIG.1. Covalent attachment of poly(A) sequences to an exogenous primer. Electrophoretogram of reaction product synthesized by poly ( A ) polymerase from [*H]ATP with a "P-labeled mixture of 4 S and 5 S RNA of HeLa cells as primer. B, "P before incubation; 0, 32Pand 0,'H after a 60-minute incubation (83).

use all classes of polynucleotides or even that any single polymerase does, but only that in no case has a striking dependence on sequence, size or secondary structure been demonstrated for polynucleotides priming poly(A) synthesis. This is apparent from the data of Table I, although it should be noted that we do not evaluate nuclease contamination of the polymerases or the purity or identity of the primers studied, since these were seldom reported. The primer requirement of the two highly purified preparations from calf thymus that are essentially free of nucleases has been examined in most detail (81, 83). The Mg?+-activated nuclear polymerase uses transfer RNAs, poly( A ) , and a mixture of small RNA fragments (about 4 S) from calf thymus equally well as primers on a molar basis. It does not use the 18 or 28 S ribosomal RNA of HeLa cells or MS-2 bacteriophage RNA, although a small but significant incorporation of AMP was always observed with these RNAs. This label was apparently associated with fragments or contaminating RNAs since label did not cosediment in sucrose gradients with 18 S or 28 S RNA or MS-2 RNA, which were recovered intact from the reaction. This only points up

POLYADENYLATE POLYMERASES

161

the problems of interpreting primer studies based solely on incorporation data. The priming function of homopolynucleotides is difficult to quantitate because of the size heterogeneity of these preparations, but the fact that poly( C ) and poly( I ) each stimulated the Mg?+-activatedpolymerase suggests that priming was not restricted to molecules terminating in AOH. This lack of specificity was confirmed ‘by a nearest-neighbor analysis of the product synthesized from [a-32P]ATPusing a mixture of thymus RNA fragments as primers. 32Pwas recovered from the product in about equal amounts in UMP, CMP and GMP. The total amount in these nucleotides relative to that in AMP was close to that expected on the basis of the length of the poly( A ) sequence calculated by endgroup analysis ( 83). Size is one property of the primer that significantly influences primer effectiveness for this preparation. Small oligomers of AMP were far less effective on a molar basis than was a poly( A ) of about 40 nucleotides. This effect of length was not seen for the Mn‘+ activated poly( A ) polymerase of calf thymus, where detailed studies have been carried out with short oligomers of AMP, primarily P ( A ) ~(81). Poly(A), however, was equally effective based on the number of 3’ hydroxyl ends present (80). Othcr natural RNA species were less effective, although this preparation has been used to add poly(A) tails to Qp viral RNA without destroying infectivity (96). A similar lack of .specificity characterizes the primer requirement for the purified E . coli polymerase (Table I ) . The specificity of the primer requirement for poly( A) polymerases has acquired new significance now that poly(A) sequences are known to be attached to the 3’ ends of mRNA and hnRNA, and particularly since mRNAs appear to have an untranslated sequence adjacent to the poly( A ) sequence. Evidence that this sequence might be similar in many species of mRNA was suggested when poly(A) sequences derived from RNase T1 digests of total HeLa cell mRNA as well as hnRNA appeared to have a common sequence of one or two CMP and two UMP residues at the 5’ end of the uninterrupted poly( A ) sequence (97). A 50-nucleotide sequence adjacent to the poly( A ) sequences of rabbit globin mRNA and mouse immunoglobulin light-chain mRNA contain regions of homology that include four nucleotides immediately adjacent to the poly( A ) sequence in each case (98). These 50 nucleotides can be folded into a double-hairpin structure that is remarkably similar for both mRNAs. If nuclear poly(A) polymerases do add AMP residues to hnRNA, it is possible that the specificity for recognition of appropriate molecules depends either on this 3-dimensional folded structure or on a short sequence attached to the poly( A ) sequence. It is interesting that the Mg2+activated poly( A ) polymerase of calf thymus thought to be responsible

F

TABLE I SYNTHESIS OF PoLY(A) SEQUENCES Source of Enzyme Escherichia coli Escherichia coli

Cell fraction Ribosome Supernatant of crude extract

Divalent cation Mg2+ Mnz+ (mW (mM) 25 16

-

Escherichia coli

Whole cell homogenate

Escherichia coli Escherichia coli

Cell debris Cell extract

10

Eschcrichia coli

Cell extract

Shigella dysmteriac Psewlomonm putida

Purified ribosomes Whole cell homogenate

0.8 1 20

Yeast Yeast Tobacco leaf Corn seedlings

Ribosomal pellet Nuclei Tissue homogenate Tissue homogenate

Krebs ascites cells Krebs ascites cells

Cytoplasmic homogenate Cell extract

3 0.62

Krebs ascites cells

Cell extract

1 .25

Lrtndschutz ascites cells Vaccinia virus

2 -

2 -

2.5 -

0.2 -

Primer

rRNA, E. coli tRNA, phage RNA rRNA, ribonucleotidylDNA, degraded D N A

E. coli t R N A

-

Ribosomes Virions

-

1

10 12.5

Product (no. average A M P units/sequence)

Poly(A) (25-35)

P O ~(A) Y All classes of natural RNA Poly(A) (300-several thousand) None Poly(A) (130)

0.8

-

c u

E. coli tRNA, E. colirRNA Poly(A) (50) R a t liver RNA, E. coli Poly (A) (100-200)

rRNA, tRNA, poly(C) P (A) 4 1 Yeast RNA 1 rRNA - Endogenous RNA 5 tRNA, rRNA, poly(U), ssDNA - Ascites rRNA, t R N A Ascites rRNA, tRNA, r a t rRNA, poly(A) 1 . 2 3 Ascites RNA, rat rRNA 2.0 - POlY (A)

10

Q,

Natural IINA, poly(dT), POlY (C)

POlY (A) Poly(A) (100) Poly (A) (10-20) Poly(A) (4-7) POlY (A) PO~Y(A)(30)

Poly(A) (30-40) Poly A Poly A (40-80)

Reference 18 3

4 110,111 30 '

g

95

!i

15 19

M

69 70 16 93,94,13%

i* 2

*

65,66

Z Z

5

32:

73, 74

a

z?

1-accinia virus T'accinia virus-infected HeLa cells T'csicular stomatitis virus L cells infected with vesicular stomatitis virus Sea urchin Chick chorioallantoic membrane Chick embryonic heart or liver Quail oviduct Guinea pig Rat liver R a t liver R a t liver R a t liver Rat liver

Virions Cell rytoplasm

-

Virion, ribonuclcoprotcin core Cell extract

5

1

-

Nuclei Particulate free extract Cytoplasmic extract

10 -

1 10

Nuclear extract 5 Neuronal, glia!, liver nuclei 10 Nuclear extract Nuclear extract Cell homogenate Mitochondria 5.7 Mitochondria -

0 ..i P(A)G,(A112 Endogenous KNA 2 P O l Y (U) Endogenous RNA 4 NTP's required - Endogenous RNA (qucstion need for primer)

Poly (A) (100) Poly(A) (4 S)

75

0.8 2 1

Poly(A), yeast RNA rRNA, poly(U), poly(A)

Poly(A) (3-4) Poly(A) (8-9)

14

2

Endogenous RNA, E . coli tRNA Oviduct RNA Endogenous RNA rRNA Nuclear RNA Poly(A), poly(U), rRNA Question need for primer Poly(A), mitochondria1 RNA P O ~(A) Y Cellular RNA, poly(A)

POlY (A)

-

2 0.6 5.5 2

2.8 0.5

Rat liver Hamster embryo fibroblast Calf thymus Calf thymus

Nuclear lysate Cell extract

-

-

4 4

Nuclear extract Cytoplasmic extract

8 -

0.5

HeLa cells Human lymphocytes

Cytoplasmic compartment Cytoplasmic

-

1 0.5

-

Endogenous RNA, poly (A) Any 3'-OH-terminated RNA chain tRNA, 18 S a n d 28 S RNA (N)12--16

P O ~(A) Y POlY(A) (12)

Poly(A) POlY (A) Poly(A) POlY (A) Poly(A) POlY (A) Poly(A)

2

2

(50) (2-14) (3-30)

(600)

Poly(A) (7-59) Poly(A) (20-50) (Poly (A) (100-200) Poly (A) Poly(A) POlY (A)

77 78, 79

87 17 18 85 7 71 72 105 68 1,28,82,85 81,82 67 86

164

MARY EDMONDS AND MARY ANN WINTERS

for polyadenylation of hnRNA in uiuo showed no specificity for the terminal nucleotide at the primer attachment site (83). This result along with the many others suggesting a low primer specificity for most poly(A) polymerases leads to a suspicion that the reaction in uitro may not closely resemble cellular polyadenylylation reactions. Other properties of the in uitro reaction also suggest this. The rate of the elongation reaction, for example, measured with the Mg2+activated thymus polymerase at saturating levels of primer, for example, is far slower than would be needed to synthesize poly(A) sequences in uiuo (83). Under conditions where poly( A ) synthesis was linear for at least 2 hours, chain lengths increased by only two residues per minute. The absence of some as yet unrecognized component required for the reaction, or of the appropriate biological primers suggested by these results should stimulate continued study of the requirements for optimal poly( A ) synthesis in uitro.

E. Reaction Products Since first shown for the E. coli poly(A) polymerase ( 3 ) , it has been assumed, although rarely demonstrated, that pyrophosphate is produced during poly ( A ) formation. The expected stoichiometry between AMP utilization and pyrophosphate production has recently been observed with the Mn2+ stimulated polymerase of calf thymus (81). As noted above, the poly(A) product is covalently attached to the primer. Lengths of poly( A ) would naturally be a function of the duration of the reaction and as few as two or more than a thousand AMPS have been reported in the product from different preparations, although most of the eukaryotic polymerases appear to be capable of synthesizing sequences similar in size to those found in the cell, namely 100-200 nucleotides (Table I). Except for the Mg2+-activatedcalf thymus enzyme noted previously, almost no studies of the rates of elongation or of maximum sizes attainable have been reported.

F. Kinetics of Poly(A) Synthesis In contrast to other polynucleotide polymerases, poly ( A) synthesis has not yet been separated experimentally into the stages of initiation, elongation and termination that generally characterize linear polymer formation. Lack of an experimental tool for segregating these reactions has prevented such studies, although initiation in the absence of elongation might be possible with cordycepin triphosphate if it can effectively replace ATP as a substrate. Kinetics of the overall reaction have been reported for the calf thymus and Krebs ascites polymerases where a pronounced lag was

POLYADENYLATE POLYMERASES

165

observed that, in the case of the mouse tumor polymerase (66) but not of the calf-thymus enzyme ( 8 2 ) , was abolished at higher enzyme concentrations. A preliminary incubation of the polymerase with or without added primer could not reduce the lag. Both preparations display a marked reduction in reaction rate at low enzyme concentration that was not overcome by raising the total protein concentration with added serum albumin (66, S2),

IV. Poly(A1 Polymerase Proteins Little is known of the proteins of poly(A) polymerases except for reports of molecular weights derived primarily from gel-filtration studies. Recently, however, important progress has been made with the Mn2+activated polymerase of calf thymus, which is reported to be close to homogeneous ( 81 ) . Although problems with protein aggregation have apparently not been completely solved, polymerase activity migrated with a single major protein band during gel electrophoresis. One major band was also observed after electrophoresis in dodecyl sulfate/polyacrylamide gels that corresponded to a 60,000-dalton polypeptide. The fact that this size corresponds closely to the size estimated for the enzymically active protein recovered both from a calibrated gel-filtration column and a sucrose-density gradient suggests that poly ( A ) polymerase is a single polypeptide. Similar results have been reported for the E . coli enzyme, although homogeneity has not been achieved (30). Electrophoresis of a highly purified preparation gave bands on gel of 50,000 and 10,000 daltons. When the individual fractions from a single peak of activity from a phosphocellulose column were electrophoresed, the intensity of both the 50,000- and 10,000-dalton bands corresponded to the amount of active enzyme. After gel filtration, poly( A ) polymerase activity was associated with a 58,000 (_t5000) polypeptide. However, there appeared to be a large loss of activity and it was suggested that the 10,000-dalton polypeptide removed during this procedure might be needed for maximal activity. The somewhat larger size calculated for the polypeptide by gel filtration was attributed to the presence of high concentrations of NaCl needed to stabilize the enzyme activity (30). Molecular weights of 80,000 were obtained by density-gradient sedimentation analysis for the Krebs mouse tumor (65) and of 120,000140,000 by gel filtration for the Mg?+-activatedcalf-thymus poly( A ) polymerase (82). Although few components were evident in either of these prcparations on gels, no attempt was made to associate enzyme activity with proteins separated by gel electrophoresis. A poly( A ) polymerase

166

MARY EDMONDS AND MARY ANN WINTERS

extensively purified from vaccinia virus cores by affinity chromatography on DNA-cellulose has a molecular weight of 80,000 as calculated by velocity gradient-sedimentation analysis. Two polypeptides of 51,000 and 38,000 daltons were recovered from all preparations of this purified enzyme electrophoresed in gels containing sodium dodecyl sulfate ( 7 4 ) . These preliminary results suggest that the poly( A ) polymerase activity in two preparations, at least, is associated with a 60,000-dalton single polypeptide, although some uncertainty about the number of components in the E . coli enzyme has been expressed (30). Two components are present in the vaccinia polymerase, but it has not yet been shown that both are required for activity. Data for other preparations are less complete, but proteins larger than 60,000 appear to be involved, although aggregation has not been ruled out. More structural data are needed, particularly since they may be necessary to establish the existence of different poly( A ) polymerases in the cell.

V. Multiple Poly(A1 Polymerases The existence of more than one poly(A) polymerase in a single tissue or cell type is receiving renewed attention although it was predicted for calf thymus nuclear extracts more than 10 years ago (33). As was the case then, much of the current evidence is based on strikingly different effects of Mn2+and Mgz+ on poly(A) synthesis. Caution has already been suggested in interpreting differential divalent ion effects in crude preparations as indicating multiple polymerases ( see Section 111). However, such evidence acquired new credibility when eukaryotic RNA polymerase I was clearly shown to prefer Mg2+,while RNA polymerase I1 was maximally activated by Mn'+ (91). In this case, however, the two enzymes proteins easily separated from a single extract show other differences, most significantly their sensitivity to a-amanitin ( 99, loo). The situation is not as clear-cut in the case of poly( A ) polymerases. Although two highly purified preparations from calf-thymus tissue that differ in their activation by Mg'+ and Mn2+have been described (81, 82), each has been purified and studied in a different laboratory. They have not been compared in similar assays, so that observed differences, in primer requirements for example, could be due to divalent ions rather than distinctive enzyme proteins. However, large differences in molecular weights have been reported for the two enzymes (see Section IV); if confirmed, these would certainly indicate the existence of separate enzymes in calf thymus tissue, although not necessarily in a single cell type. A somewhat puzzling aspect of the two calf-thymus polymerases

POLYADENYLATE POLYMERASES

167

is that they have never been separated and isolated from the same extracts, although a Mn2+activatcd poly ( A ) synthesis is always observed in the extracts from which the Mg?+-activated enzyme is purified. The Mn?+-stimulated activity, which is often 40 times that of Mg2+in crude extracts, is gradually lost during purification of the Mg2+-activated polymerase, most noticeably during phosphocellulose chromatography ( 82). However, a separate MI?+-stimulated poly( A) synthesis has never been detected in any other protein fraction released from this phosphocellulose column, or in any other fraction for that matter. Failure to detect a separate Mg?+-activated polymerase in the extracts from which the Mn?+polymerase is purified is less surprising, since it is initially present in far lower concentrations (81). It should be pointed out in this regard that the divalent ion requirement for each enzyme is not completely specific, since Mn2+can replace Mg?+ quite effectively in the most purified Mg?+-activated poly ( A ) polymerase, although Mg2+is less effective with the highly purified Mn?+-activatedpolymerase. Although separation of two polymerases has not been achieved with a calf-thymus preparation, it may have been in extracts of rat liver nuclei (85) and hamster embryo fibroblasts (68). Extracts from both sources have been fractionated on ion-exchange celluloses into multiple components with poly(A) polymerase activity. In the case of rat liver nuclear cxtracts, the separation of a Mg'+-activated polymerase depended on its tendency to aggregate in high salt, which was not shown by the Mil?+stimulated activity. However, the Mn"-activated fraction also retained activities for homopolymer synthesis from CTP, GTP or UTP, although it was claimed that the Mn"-activated poly( A ) synthesis had bee partially separated by gel filtration from these other polymerase activities (85).The three poly( A) polymerase fractions from the hamster embryo fibroblasts were all far more active with Mn2+ than Mg2+, but were assumed to be distinct, since each responded somewhat differently to divalent cations (68). Although the evidence for fractionation of poly ( A ) polymerase activities is clear-cut in both these cases, it cannot be assumed that they are diff ercnt enzyme proteins because separation of activities could be due to partial aggregations or incomplete removal of polynucleotides bound to polymerase. On balance, however, these observations support the existence of separate poly(A) polymerases. The results with the hamster embryo fibroblasts may be especially significant, since it is the first to provide evidence of more than one poly(A) polymerase in a single cell typc. Separate polymerases for different cell organelles was suggested to account for this multiplicity. This is a plausible assumption, and is compatible with the current knowledge of the location of poly( A ) sequences in animal cells.

168

MARY EDMONDS AND MARY ANN WINTERS

VI. Cellular Location of Poly(A1 Polymerases An early study of the kinetics of poly(A) synthesis in HeLa cells suggested that polyadenylylation is exclusively a nuclear event in which the poly( A ) ,synthesized in the nucleus is conserved and is transported to the cytoplasm as poly(A)-containing mRNA (101).It was reassuring, then, to find that poly(A) polymerases able to catalyze this reaction had already been purified from the nuclei of several animal tissues (see Table I). Interest in the cellular location of poly( A ) polymerases has intensified recently, as evidence suggesting that poly ( A ) synthesis may not be confined to the nucleus, but may also occur in the cytoplasm of animal cells, has been accumulating (102-104). A glance at Table I shows, in fact, that cytoplasmic extracts of several tissues have also served as a source of these polymerases (6, 7, 80, 81). Such observations say little about the actual location of poly(A) polymerases in the cell, however, since techniques used for their extraction were generally more suited for obtaining high yields of enzyme than for subcellular localization studies. The evidence for a nuclear enzyme derives from its extraction from nuclei of several different tissues that, in most cases, had been purified by sedimentation through dense sucrose to remove contaminating cytoplasm (1, 18, 70,85, 87, 105, 106). Two nuclear preparations contained sufficient Mg2+-activatedpolymerase to account for that present in an equivalent amount of whole tissue. More than four times as many enzyme units were extracted from calf-thymus nuclei as from an equivalent weight of whole tissue (89). A similar result was reported for a MgZ+activated poly( A ) polymerase of quail-oviduct nuclei (87).Nuclear extracts in both cases had specific activities 8-10 times that of whole tissue extracts. The activity extracted from calf thymus nuclei could not be distinguished from that extracted from whole tissue in its substrate, cation or primer requirement, nor in its chromatographic properties on DEAE- and phosphocellulose columns. It was concluded that identical polymerases had been isolated and purified (82). Polymerases have also been detected in the cytosol of isotonic sucrose homogenates prepared from a number of rat tissues. Only about 1% of this Mnz+-activated poly(A) synthesis was found in the nuclei of thymus, liver, kidney or spleen prepared from similar homogenates ( 81 ) . Similar results were obtained with human lymphocytes ( 86). Although little activity could be extracted from these nuclei, the detergent wash used to remove contaminating cytoplasm would most certainly have removed both inner and outer nuclear membranes and, in the process, any loosely bound enzymes (107).The poly( A ) polymerase recovered

POLYADENYLATE POLYMERASES

169

from a cytoplasmic extract of HeLa cells (67) is difficult to localize since the hypotonic swelling technique used to rupture the HeLa cells produces a marked swelling of the nucleus that would certainly favor the release of soluble proteins. A poly(A) polymerase that apparently differs from that found in liver cytosol has also been purified from rat liver mitochondria ( 72). While the usual reservations associated with problems of translocation of enzymes after disruption of cells in aqueous media must be kept in mind, it scems that poly(A) polymerases may be present in both nucleus and cytoplasm. However, to suggest that these are different enzymes would be premature, even though the preparations described here are activated to different extents by Mn2+ and Mg2+. It should be noted that nuclei also show both Mn2+-and Mg2+-activatedpoly(A) synthesis (33, 85) (see Section V ) . While a nuclear location can be more readily accepted because most of the poly( A ) in the cell is synthesized in the nucleus, the present uncertainty about cytoplasmic poly (A) synthesis makes it more difficult to dismiss the possibility that activity of cytoplasmic extracts results from a loosely bound nuclear polymerase released to the cytosol during cell fractionation. Similar arguments of course might be raised about the adsorption of cytoplasmic polymerases to nuclei. Techniques of cell fractionation superior to the usual aqueous extraction procedures used thus far are clearly needed to answer questions of the location of poly ( A ) polymerases in the cell.

VII. Poly(A1 Polymerases as a Subunit of RNA Polymerase The early speculations on a possible connection between poly( A) and RNA polymerases (108) was prompted by the rapid inhibition of poly( A ) polymerase observed immediately after T2 and T4 phage infection of E. coli, which precedes the cessation of host RNA synthesis (35). The inhibition of poly(A) polymerase in this case was traced to a protein inhibitor synthesized after infection. The fact that a partially purified preparation of the inhibitor also inhibited C. coli RNA polymerase suggested that the two polymerases may share a common structural component (108). The purified RNA and poly( A ) polymerases of E. coli did in fact show three common bands when electrophoresed on polyacrylaniide/dodecyl sulfate gels that included the a-subunit of RNA polymerase (108). More convincing evidence for a common structural component is the fact that antibodies prepared from each polymerase show cross-reactivity (108). It has been suggested that poly(A) polymerase might actually be one of the subunits of E . coli RNA polymerase (108). If true, this might explain the marked preference for

170

MARY FDMONDS AND MARY ANN WINTERS

the synthesis of poly( A) when single nucleoside triphosphates are substrates for RNA polymerases (8, 9,109). A direct test of this hypothesis was undertaken by another group (110),who reported that a smaller protein could be dissociated from E. coli RNA polymerase preparations inactivated by treatment with p-chloromercuribenzoate. This protein had poly ( A ) polymerase activity and restored RNA synthesis when added back to the inactivated RNA polymerase. The polypeptide with poly( A ) polymerase activity had the same electrophoretic mobility in dodecyl sulfate gels as the a-subunit of RNA polymerase. A separate purification of a poly(A) polymerase from E. coli cell debris was also carried out by this group (111).Its properties were indistinguishable from the protein dissociated from the purified RNA polymerase and it also reactivated the mercurial-inactivated RNA polymerase. These experiments, suggesting a coordination of poly ( A ) synthesis and RNA synthesis through some type of regulation of the multimeric RNA polymerase system, support an idea that continues to attract attention. It has recently been extended, for example, to eukaryotic RNA polymerases, where a mild heat treatment of the rat liver RNA polymerase I1 greatly reduced RNA synthesis, but increased poly( A ) synthesis (112). In addition, the poly(A) polymerase activity appearing on heating could be separated from RNA polymerase I1 by DEAE-cellulose chromatography. The idea is perhaps even more attractive now that polyadenylylation has been closely linked to transcription in eukaryotes and may extend to prokaryotes as well, since some of the messenger RNA of E. coli has also been shown to be polyadenylylated at the 3' terminus (113).Nonetheless, serious critickm of the results used to support this view of poly(A) polymerase as a subunit of RNA polymerase can be made, on the grounds that neither of the polymerases was of sufficient purity to rule out the presence of contaminating activities. It has been noted that RNA polymerase obtained from E . coZi by the procedure of Chamberlin and Berg ( 8 ) , as was done in the experiments noted above, is often contaminated by an ATPase and polynucleotide phosphorylase as well (114).A sequential reaction carried out by these two enzymes could be assayed as poly (A) polymerase: ATP-

ATPase

ADP

+ Pi

polynucleotide

ADP '

phosphorylase

,poly(A)

(1)

+ Pi

(2)

The marked inhibition of poly( A ) polymerase by low levels of inorganic phosphate noted in some cases may well arise from the phosphorolysis reaction characteristic of polynucleotide phosphorylase in the

POLYADENYLATE POLYMERASES

171

presence of inorganic phosphate. There is new evidence that a contamination with these two enzymes may have produced the poly( A)-synthesizing activity found in RNA polymerase of E . coli (A. Tsugita, personal communication). More direct approaches have also failed to establish the identity of the poly(A) polymerase and the a-subunit of the RNA polymerase of E . coli. None of the polypeptides from a highly purified poly(A) polymerase of E . coli migrated with the 40,000-dalton a-subunit of the E. coli RNA polymerase (30). Poly( A ) was not synthesized by a highly purified 0-subunit of RNA polymerase (115). Thus it now appears that little solid evidence exists for a polypeptide of RNA polymerase with properties of a poly(A) polymerase. It is still possible that the two polymerases are functionally related, but additional evidence and more highly purified enzymes will be needed to clarify this question.

VIII. Viral Poly(A1 Polymerases Animal virus studies have had a significant impact on developments involving poly ( A ) sequences since oligo ( adenylic acids) were first detected in reovirus virions (55, 116). I n contrast to studies of cellular poly( A ) sequences, which began with enzymology, detection of poly( A ) sequences and study of their relation to viral RNAs preceded the discovery of virus-associated poly ( A ) polymerases. Virion “transcriptases” had already been found in those viruses later shown to synthesize poly( A ) in uitro, i.e., vaccinia (117), reovirus (118, 119), and vesicular stomatitis virus (VSV) (120). Delay in recognition of a true viral poly(A) polymerase was due in part to early reports that poly(A) is apparently transcribed in vitro by disrupted vaccinia virions ( 121) . An early hint that poly(A) and RNA might be synthesized by separate enzymes was given by the observation that certain detergent treatments of vaccinia virions caused them to lose the ability to incorporate UTP, but not ATP, into an acid-insoluble product (122). Several groups eventually provided evidence for such a poly( A ) polymerase activity (73, 77, 123), which has now been established by purification of the enzyme from isolated virions free of transcriptase activity ( 1 2 4 ) .It should be noted, however, that the poly( A) syntheses observed in isolated virions of reovirus and VSV have not as yet been associated with a specific protein released from virions. These two activities also have unique properties not characteristic of poly ( A ) polymerases from vaccinia or cells in general. This includes, in the case of VSV, the fact that poly( A) synthesis appears to require concomitant RNA synthesis (75, 76, 125). The genome of VSV is a single-stranded RNA molecule, and the comple-

172

MARY EDMONDS AND MARY ANN WINTERS

mentary RNA synthesized from it after infection contains a poly(A) sequence of 50-150 nucleotides for which there is no complementary region in the virion RNA. Detergent treatment of isolated virions activates a transcriptase that synthesizes RNA in the presence of the four nucleoside triphosphates. This RNA also contains covalently bound poly(A) sequences similar in size to those found in the VSV RNA of infected cells. This poly( A ) is not synthesized unless all four nucleoside triphosphates are present. This would suggest that poly( A ) is also transcribed, were it not for the fact that a variety of poly(A) lengths are attached to the RNA product (126) and that the VSV genome lacks oligo( U ) sequences ( 1 2 7 ) . The apparent requirement for concomitant RNA synthesis could also indicate a need for an appropriate primer for poly( A ) attachment. The poly( A ) synthesis observed in isolated reovirus virions is quite unusual. Short oligomers of AMP are synthesized in uitro, as well as in uiuo, that are not attached to any of the ten double-stranded RNA segments that characterize the reovirus genome (128) or to the mRNAs transcribed from the genome ( 1 2 9 ) . These oligo( A ) sequences are unique among poly(A)s recovered from cells or enzyme reactions in that pyrophosphate groups are found at the 5' terminus (130, 131). This appears to be a rare example of an initiated poly( A ) synthesis. Although a poly( A ) polymerase has not been isolated from the purified virion, poly(A) synthesis can be observed in the absence of RNA synthesis and vice versa. Oligo(A) synthesis, but not RNA synthesis, occurs in intact purified virions. A limited chymotrypsin digestion of the virion is needed to activate RNA synthesis. A more drastic treatment with chymotrypsin produces reovirus cores that no longer jsynthesize oligo( A), but make RNA as usual (130, 131). Although separate polymerase activities are uncovered during this progressive disruption of virions, the two activities can also be observed independently, even when they coexist in a subviral particle such as that produced by limited chymotrypsin treatment. In Mn2+ (12 mM), only oligo( A ) is made in these particles with ATP alone or even with all four ribonucleoside triphosphates present, while only RNA is made in Mg*+ ( 6 mM) ( 1 3 0 ) . The fact that distinct enzyme proteins have not been isolated and the two activities share certain properties such as similar pH and an unusual temperature optima (45"), have led investigators of this system to propose that poly(A) polymerase activity is an altered form of the viral transcriptase ( 130, 131 ) . A similar view of the two activities in vaccinia virions was also invoked to account for the fact that transcriptase activity was lost by the disruption procedure used to release the poly (A) polymerase from the virion ( 7 3 ) . The linking of poly(A) polymerase activity with a

POLYADENYLATE POLYMERASES

173

subunit of E. co2i RNA polymerase undoubtedly provided a bask for these speculations (see Section VII ) . However, the successful purification of the poly( A ) polymerase from vaccinia virions and its characterization as a typical poly(A) polymerase would tend to favor the existence of separate polymerases. If similar enzymological approaches can be applied to reovirus and VSV, a separation of activities may also be achieved. Although vaccinia poly( A ) polymerase resembles most others, it has an unusual ability to use polydeoxyribonucleotides as primers in addition to a wide variety of polyribonucleotides. Covalent attachment of most of thc poly( A ) sequences to the 3’-dT end of poly( dA * dT) has been clearly shown ( 7 4 ) . This property was of particular interest in vaccinia since poly ( A ) had originally been reported to be synthesized by transcription. Now, however, the meaning of this finding is unclear since polyadenylylated DNA has not been reported in uiuo. It may merely reflect the general lack of primer specificity of these polymerases. There have been reports of DNA-primed poly( A ) synthesis with eukaryotic polymerses. For example, activity of maize seedling poly ( A) polymerase was stimulated both by synthetic poly( dA) and poly( dT) (132). However, a reported stimulation of the calf-thymus polymerase by heat-denatured calf-thymus DNA was most likely due to RNA contamination, since the effect was alkali labile (83). Poly(A) polymerase activity is increased 3- to 4-fold in HeLa cells 4 hours after infection with vaccinia virus (79). This increase in activity precedes the appearance of the viral RNA polymerase by about 2 hours. The activity has not been highly purified, but it did appear to attach poly ( A ) to polynucleotides characteristic of vaccinia-specific RNA. The relation of this activity to the highly purified virion polymerase, as well as to cellular poly( A ) polymerases is not known. The success in the purification of a poly( A ) polymerase from vaccinia virions should encourage further investigation of viruses, which provide simple clean experimental systems in which transcription and polyadenylylation must be closely linked.

IX. Regulation of Poly(A1 Polymerases In considering the role that poly( A ) synthesis undoubtably plays in the processing of hnRNA to mRNA it is natural to assume that either the synthesis or the activity of poly(A) polymerases could be points where mRNA production might be regulated. One effort to examine the role of hormones in regulating poly( A ) polymerase levels has been carried out with estrogen-treated quail. Since estrogens apparently exert their effects at the level of transcription, an increased number of transcripts of poly(A) polymerase structural genes might be detected as

174

MARY EDMONDS AND MARY ANN WINTERS

an increased specific activity of the enzyme in the target tissue after estrogen treatment. No such increase was detected, however, in oviduct tissue of quails subjected to a variety of estrogen treatments, although the specific activity of RNA polymerase I and I1 was increased in these experiments (87). On the other hand, lymphocytes stimulated to grow by phytohemagglutin showed a 3- to 6-fold-increased specific activity of the MnZ+stimulated poly ( A ) polymerase over resting lymphocytes ( 86). Another form of regulation of the activity of poly(A) polymerase may be indicated by the polyadenylylation of mRNA stored in the cytoplasm, which occurs after fertilization of sea urchin eggs ( 1 0 3 ) . Since this polyadenylylation occurs in the presence either of actinomycin D or ethidium bromide, fertilization may activate poly ( A ) polymerases by a structural modification. It is obvious from these scattered examples that little evidence now exists for any regulation of poly( A) polymerases. New experimental systems and techniques bearing on aspects of regulation may provide the clues now needed to assign a cellular function to poly( A ) sequences.

X. Poly(A1 Polymerases as PolyadenylylatingReagents An important use for poly( A ) polymerases developed when poly( A ) sequences were found on the 3’ ends of RNA molecules. Such RNAs become templates for RNA-dependent DNA polymerases (reverse transcriptases) when oligo( U ) or oligo( dT) are included in the reaction to provide the double-stranded primer region needed to initiate transcription of the single-stranded RNA. It was obvious that complementary DNA copies of RNA molecules not containing poly( A ) might also be made if poly(A) could be attached to 3’ ends. The marked specificity for ATP and the low specificity for RNA primers plus a limited reversibility of the polymerization reaction made poly( A ) polymerases highly suitable reagents for 3’ polyadenylylation of RNA. Possible restrictions on their effectiveness would, of course, be ribonucleases contaminating the enzyme and a lack of availability of the 3’ end of the RNA acceptor. One important class of messenger RNAs that is a prime target for an enzymic polyadenylation is that for histones. Complementary DNA copies would be especially valuable for quantitative assay of histone mRNA transcripts as well as for histone genes and their localization in chromosomes. Success has already been reported with a polymerase from maize that converts the histone mRNAs of HeLa cells into templates for Rous sarcoma virus RNA-dependent DNA polymerase by polyadenylylation (133).

POLYADENYLATE POLYMERASES

175

The E . coli poly( A ) polymerase has also been used successfully ( 1 3 4 ) to polyadenylylate a mouse globin mRNA that had been deadenylylated by treatment with polynucleotide phosphorylase ( 135). On readenylylation, the globin mRNA regained its template function in the reverse transcriptase reaction ( 1 3 4 ) . The same E. coli preparation was also used to polyadenylylate 18 S and 28 S ribosomal RNAs of mouse cells. After treatment, the binding of the 18 S, RNA to poly( U)-Sepharose was increased from 3 to 20%.The limited effectiveness in this case was ascribed to the ribonuclease activity remaining in the polymerase. Similar treatment did enhance the ability of 28 S RNA to serve as a template for the reverse transcriptase although the DNA copy was of low fidelity compared with that prepared from globin mRNA. Probably the most rigorous test of the feasibility of enzymic polyadenylylation has been met by the Mn*+-activated poly( A ) polymerase of calf thymus, which was able to polyadenylylate the RNA of Qp bacteriophage without loss of infectivity (96). In this experiment, at least 50% of the Qp RNA was sufficiently adenylylated to bind to poly( U ) Sepharose. Preparations with a few additional AMPS at the 3’ end ( a n average of 6 ) were equally effective in infecting E. coli spheroplasts as were preparations with 75 additional AMP residues. The unique properties that poly( A) sequences confer on RNA molecules will most certainly result in other applications of this enzymic polyadenylylation reaction.

XI. Summary Poly ( A ) polymerases, although studied in considerable detail many years ago, were generally ignored until poly( A ) sequences were found in both the mRNA and hnRNA of animal cells in 1971. Renewed interest in the enzymes likely to be responsible for synthesis of poly(A) sequences has led to new methods for obtaining preparations of high purity. This has allowed the first studies of the enzyme protein(s) to begin and has provided preparations of the polymerase that can safely be used to polyadenylylate non-poly ( A ) -containing RNA molecules. Evidence for multiple poly( A ) polymerases, in single-cell types in at least one case, has been accumulating. This could be related to the small amounts of poly(A) synthesis now being studied in the cytoplasm and mitochondria of animal cells. Localization of the enzyme within the cell has encountered the problems usual to such studies, but nuclear, cytoplasmic and mitochondria1 sites are all supported by experimental evidence. Several animal viruses also show poly ( A ) polymerase activity, and a highly purified enzyme free of RNA polymerase

176

MARY EDMONDS AND MARY ANN WINTERS

activity has been obtained from vaccinia virions. Claims for the identity of the E . coli poly(A) polymerase with a subunit of RNA polymerase have not been supported by more recent evidence. Not surprisingly, the regulation of poly( A) synthesis through control of either the activity or synthesis of poly( A ) polymerases is under consideration, but as yet little evidence is in, although this is likely to change in the near future.

ACKNOWLEDGMENTS The authors express appreciation to the American Cancer Society for support from Grant NP-130A for studies of poly( A) polymerases carried out in our laboratory and to the many colleagues who so willingly provided their manuscripts and data prior to publication.

REFERENCES 1. M. Edmonds and R. Abrams, JBC 235, 1142 (1960). 2. C. W. Chung, H. R. Mahler and M. Enrione, JBC 235, 1448 (1960). 3. J. T. August, P. Ortiz and J. Hunvitz, JBC 237,3786 (1962). 4. M. E. Gottesman, Z. N. Canellakis and E. S. Canellakis, BBA 61, 34 ( 1962). 5. R. H. Burdon, BBRC 11,472 ( 1963). 6. P. R. Venkataraman and H. R. Mahler, JBC 238, 1058 ( 1963). 7 . H. G . Klemperer, BBA 95, 251 (1965). 8. M. Chamberlin and P. Berg, PNAS 48, 81 (1962). 9. A. Stevens, JBC 239, 204 ( 1964). 10. A. Falaschi, J. Adler and H. G. Khorana, JBC 238,3080 (1963). 11. M. L. Giron and J. Huppert, Bull. SOC. Chirn. B i d . 47, 1677 (1966). 12. S. J. S . Hardy and C. G. Kurland, Bchem 5,3668 ( 1966). 13. E. Horton, S. L. Liu, E. M. Martin and T. S. Work, J M B 15, 62 (1966). 14. E. A. Hyatt, BBA 142, 254 (1967). 15. A. J. E. Colvill and M. Terzi, BBA 155,394 ( 1968). 16. R. C. Sachar, BBA 169,58 (1968). 17. T. Kato and M. Kurokawa, BJ 116,599 (1970). 18. H. Nakazato, Ph.D. Dissertation, Tokyo Univ., Tokyo, Japan (1970). 19. K. J. Payne and J. A. Boezi, JBC 245,3178 (1970). 20. J. E. Darnell, R. Wall and R. J. Tushinski, PNAS 68, 1321 (1971). 21. S. Lee, J. Mendecki and G. Brawerman, PNAS 68, 1331 ( 1971). 22. M. Edmonds, M. Vaughan and H. Nakazato, PNAS 68, 1336 ( 1971). 23. M. Grunberg-Manago and S. Ochoa, JACS 77,3165 (1955). 24. A. Kornberg, I. R. Lehman, M. J. Bessman and E. S. Simms, BBA 21, 197 (1956). 25. S . B. Weiss and L. Gladstone, JACS 81,4118 (1959). 26. A. Stevens, BBRC 3, 92 (1960). 27. J. Hunvitz, A. Bresler and R. Diringer, BBRC 3, 15 ( 1960). 28. M. Edmonds and R. Abrams, JBC 237,2636 (1962). 29. M. Edmonds and R. Abrams, BBA 26,226 (1957). 30. A. E. Sippel, EJB 37, 31 (1973). 31. C. Heidelberger, E. Harbers, K. C. Leibman, Y. Takagi and V. R. Potter, BBA 20, 445 ( 1956).

POLYADENYLATE POLYMERASES

177

32. L. I. Hecht, M. L. Stephenson and P. C. Zamecnik, PNAS 45, 505 (1959). 33. M. Edmonds and R. Abrams, in “Symposium on Nucleic Acids,” p. 1. Council of Scientific and Industrial Research, New Delhi 1965. 34. P. Berg, M. Chamberlin and H. Fancher. in “Informational Macromolecules” ( H . J. Vogel, V. Bryson and J. D. Lanipen, eds.), p. 467. Academic Press, New York, 1963. 35. P. Ortiz, J. T. August, M. Watanabe, A. M. Kaye and J. Hurwitz, JBC 240, 423 (1965). 36. M. Edmonds and R. Abrams, JBC 237,2636 ( 1962). 37. M. Edmonds and R. Abrams, JBC 238, 1186 (1963). 38. N. M. Wilkie and R. M. S. Smellie, BJ 109, 229 ( 1968). 39. N. M. Wilkie and R. M. S. Smellie, BJ 109, 485 (1968). 40. M. G. Page, G. R. Haynes and H. G. Klemperer, BJ 102, 181 (1967). 41. H. G. Klemperer, BBA 72,403 ( 1963). 42. M. Edmonds, JBC 240, 4621 (1965). 43. R. H. Burdon and R. M. S . Smellie, BBA 51, 153 (1961). 44. D. B. Straus and E. Goldwasser, JBC 236, 849 ( 1961). 45. A. K. Chakravorty and B. B. Biswas, JBC 240, 4406 (1965). 46. P. Mandel, A. R. Dravid and N. Pete, J . Neurochem. 14, 301 (1967). 47. J. G. Cory, A. W. Benson and A. J. Girgenti, BBRC 42, 778 ( 1971). 48. H. Fukasawa and T. Mizuochi, BBRC 58,405 (1974). 49. G. Burkard and E. B. Keller, PNAS 71, 389 ( 1974). 50. R. Abrams, M. Edmonds and B. B. Biswas, CoUoq. Znt. C.N.R.S. 106, 323 (1962). 51. L. A. Haff and E. B. Keller, JBC 250, 1838 ( 1975). 52. B. Goz, BBA 204, 267 (1970). 53. A. Hadjivassiliou and G. Brawerman, JMB 20, 1 (1966). 54. E. E. Ewing and J. H. Cherry, BBA 161,331 (1968). 55. A. R. Bellamy and W. K. Joklik, PNAS 58, 1389 ( 1967). 56. M. Edmonds and M. G. Caramela, JBC 244,1314 (1969). 57. L. Lim, Z. N. Canellakis and E. S. Canellakis, BBRC 34, 536 (1969). 58. L. Lim, Z. N. Canellakis and E. S. Canellakis, BBA 209, 112 (1970). 59. L. Lim, Z. N. Canellakis and E. S. Canellakis, BBA 209, 128 (1970). 60. L. Lim and E. S. Canellakis, Nature (London) 227, 710 ( 1970). 61. H. Nakazato, D. Kopp and M. Edmonds, JBC 248, 1473 ( 1973). 62. G. R. Molloy, M. B. Spom, D. E. Kelley and R. P. Perry, Bchem 11, 3256 (1972). 63. J. Mendecki, S. Y. Lee and G. Brawerman, Bchem 11, 792 (1972). 64. J. E. Darnell, L. Philipson, R. Wall and M. Adesnik, Science 174, 507 (1971). 65. M.-L. Giron and J. Huppert, BBA 287, 438 (1972). 66. M.-L. Giron and J. Huppert, BBA 287,448 (1972). 67. R. Mans and G. Stein, Life Sci. 14,437 (1974). 68. A. Pellicer, J. Salas and M. L. Salas, BBA 378, 107 (1975). 69. J. S. Twu and R. K. Bretthauer, Bchem 10, 1576 (1971). 70. L. A . Haff and E. B. Keller, BBRC 51,704 (1973). 71. S. T. Jacob and D. G. Schindler, BBRC 48, 126 (1972). 72. K. M. Rose, H. P. Morris and S. T. Jacob, Bchem 14,1025 (1975). 73. B. Moss, E. N. Rosenblum and E. Paoletti, Nature N B 245, 59 ( 1973). 74. B. Moss, E. N. Rosenblum and A. Gershowitz, JBC 250, 4722 (1975). 75. H. Galet and L. Prevec, Nature NB 243,200 ( 1973).

178

MARY EDMONDS AND MARY ANN WINTERS

76. L. P. Villarreal and J. J. Holland, Nature NB 246, 17 (1973). 77. M. K. Brown, J. W. Dorson and F. J. Bollum, J. Virol. 12, 203 (1973). 78. C. Brakel and J. R. Kates. J. Virol. 14, 715 (1974). 79. C. Brakel and J. R. Kates, J. Virol. 14, 724 ( 1974). 80. C. M. Tsiapalis, J. W. Dorson, D. M. deSante and F. J. Bollum, BBRC 50, 737 ( 1973). 81. C. M. Tsiapalis, J. W. Dorson and F. J. Bollum, JBC 250, 4486 (1975). 82. M. A. Winters and M. Edmonds, JBC 248,4756 ( 1973). 83. M. A. Winters and M. Edmonds, JBC 248,4763 (1973). 84. J. Niessing and C. E. Sekeris, Nature N B 243, 9 (1973). 85. J. Niessing and C. E. Sekeris, BBRC 60, 673 ( 1974). 86. M. S. Coleman, J. J. Hutton and F. J. Bollum, Nature (London) 248, 407 (1974). 87. W. E. G. Miiller, A. Totsuka, M. Kroll, I. Nusser and R. K. Zahn, BBA 383, 147 ( 1975). 88. G. Maale, G. Stein and R. Mans, Nature (London) 255,80 (1975). 89. M. A. Winters, Ph.D. Dissertation, Univ. of Pittsburgh, Pittsburgh, Pennsylvania, 1972. 90. C . Kedinger, P. Nuret and P. Chambon, FEBS Lett. 15, 169 (1971). 91. R. G. Roeder and W. J. Rutter, PNAS 65, 675 (1970). 92. S . B. Weiss, PNAS 46, 1020 (1960). 93. T. J. Walter and R. J. Mans, BBA 217, 72 (1970). 94. R. Mans and T. Walter, BBA 247, 113 ( 1971). 95. M. J. Modak and P. R. Srinivasan, JBC 248,6904 (1973). 96. C. Gilvarg, F. J. Bollum and C. Weissmann, PNAS 72, 428 (1975). 97. G . Molloy and J. E. Darnell, Bchem 12, 2324 (1973). 98. N. J. Proudfoot and G. G. Brownlee, Nature (London) 252, 359 (1974). 99. C. Kedinger, M. Gniazdowski, J. L. Mandel, Jr., F. Gissinger and P. Chambon, BBRC 38, 165 (1970). 100. T. J. Lindell, F. Weinberg, P. Morris, R. G. Roeder and W. J. Rutter, Science 170, 447 (1970). 101. W. Jelinek, M. Adesnik, M. Salditt, D. Sheiness, R. Wall, G. Molloy, L. Philipson and J. E. Darnell, J M B 75,515 (1973). 102. J. Diez and G. Brawerman, PNAS 71,4091 (1974). 103. I. Slater and D. W. Slater, PNAS 71, 1103 ( 1974). 104. R. P. Perry, D. E. Kelley and J. La Torre, J M B 82, 315 ( 1974). 105. A. R. Dravid, N. Pete and P. Mandel, J . Neurochem. 18,299 (1971). 106. J. Niessing and C. E. Sekeris, FEBS Lett. 22, 83 ( 1972). 107. R. P. Aaronson and G. Blobel, J. Cell Biol. 62,746 ( 1974). 108. M. Terzi, A. Cascino and C. Urbani, Nature (London) 226, 1052 ( 1970). 109. R. A. Salvo, P. R. Chakraborty and U. Maitra, JBC 248, 6647 (1973). 110. S. Ohasa and A. Tsugita, Nature N B 240,35 ( 1972). 111. S. Ohasa A. Tsugita and S. Mii, Nature NB 240, 39 (1972). 112. K. M. Rose and S. T. Jacob, BBRC 60,390 ( 1974). 113. H. Nakazato, :S. Venkatesan and M. Edmonds, Nature (London) 256, 144 (1975). 114. V. Paetkau and G. Coy, Can. J . Biochem 50, 143 (1972). 115. H. J. Gross and H. Alberty, BBRC 46, 1581 (1972). 116. A. J. Shatkin and J. D. a p e , PNAS 59,246 ( 1968). 117. J. Kates and B. R. McAuslan, PNAS 58, 134 ( 1967).

POLYADENYLATE POLYMERASES

179

A. J. Shatkin and J. D. Sipe, PNAS 61, 1463 (1968). J. Borsa and A. F. Graham, BBRC 33,895 ( 1968). D. Baltimore, A. S. Huang and M. Stampfer, PNAS 66, 572 (1970). J. Kates and J. Beeson, J M B 50, 19 (1970). J. H. Subak-Sharpe, T. H. Pennington, J. F. Szilagyi, M. C. Timbury and J. F. Williams, “The Effect of Rifampicin on Mammalian Viruses and Cells,’’ Lepetit Coll. Biol. Med., Vol. 1, p. 260. North-Holland Publ., Amsterdam, 1970. 123. R. Sheldon and J. Kates, J. Virol. 14, 214 (1974). 124. B. Moss and E. N. Rosenbhm, J. Virol. 14, 86 ( 1974). 125. A. K. Banerjee and D. P. Rhodes, PNAS 70,3566 ( 1973). 126. E. Ehrenfeld and D. F. Summers, 1. Virol. 10, 683 (1972). 127. S. Marshall and D. Gillespie, Nature N B 240, 43 (1972). 128. J. Nichols, A. R. Bellamy and W. K. Joklik, Virology 49, 562 (1972). 129. C. M. Stoltzfus, A. J. Shatkin and A. K. Banerjee, JBC 248, 7993 (1973). 130. C. M. Stolzfus, M. Morgan, A. K. Banejee and A. J. Shatkin, J. Virol. 13, 1338 (1974). 131. S . C. Silverstein, C. Astell, J. Christman, H. Klett and G. Acs, J. Virol. 13, 740 ( 1974). 132. R. Mans, BBRC 45, 980 ( 1971). 133. C. L. Thrall, W. D. Park, H. W. Rashba, J. L. Stein, R. J. Mans and G . S . Stein, BBRC 61, 1413 (1974). 134. M. Getz, G. D. Birnie and J. Paul, Bchem 13, 2235 (1974). 135. R. W. Williamson, J. Crossley and S. Humphries, Bchem 13, 703 (1974). 118. 119. 120. 121. 122.

This Page Intentionally Left Blank

Three-Dimensional Structure of Transfer RNA SUNG-HOU KIM Department of Biochemistry Duke University Medical Center Durham. North Carolina

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

I. Introduction I1. Life Cycle of tRNA . . . . . . . . . . . I11 Secondary Structure of tRNA . . . . . . . . . A . All tRNAs Have the Same Secondary Structure B Classification of tRNA . . . . . . . . . . C. Invariant and Semi-invariant Bases in tRNA Sequences D All Mature tRNAs Contain Many Modified Nucleosides IV . Tertiary Structure of tRNA . . . . . . . . . A . Optical Analogy of the Crystallographic Method B. General Remarks on X-Ray Crystallographic Studies on Yeast tRNA"h". . . . . . . . . . . . . C . tRNA is GShaped . . . . . . . . . . D . Most of the Invariant and Semi-invariant Bases Form Hydrogen Bonds among Themselves . . . . . . . . . E . Other Tertiary Interactions F There Are Two Extensive Base-Stacking Systems, Each Running Half of the Molecule G . Conforniation of Nucleosides in tRNA . . . . . . H . General 3-D Structure of tRNA . . . . . . . I. Crystallographic Studies of Other tRNAs V. Correlation between the 3-D Structure and Physical and Chemical Data . . . . . . . . . . . . . . A . Small-Angle X-Ray Scattering . . . . . . . . R . Fluorescence Energy Transfer . . . . . . . . C Complementary Oligonucleotide Binding . . . . . D . Low-Field Nuclear Magnetic Resonance (NMR) Spectroscopy E . NMR Studies on Methyl and Methylene Proton Resonances F. UV-Induced Cross-Link . . . . . . . . . G . Base-Specific Chemical Modification H . Tritium Exchange of Free Purines I . Conformation of tRNA in Solution and in the Crystal . . . VI . Functional Implications of the 3-D Structure of tRNA . . . A Protein-tRNA Recognition . . . . . . . . B. tRNA-Synthetase Interaction . . . . . . . . C Interaction among tRNA, mRNA and the Ribosome . . . VII. Concluding Remarks . . . . . . . . . . References . . . . . . . . . . . . . Note Added in Proof

.

. . . . . . . . . . .

. .

.

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

. . . . .

.

.

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

. .

. . . . . . . . . . 181

182 182 184 184 184 186 187 187 188 189 190 190 194 194 195 196 199 199 199 200 201 201 203 203 204 204 206 207 207 207 211 212 213 216

182

SUNG-HOU KIM

1. Introduction One of the most interesting aspects of transfer RNA (tRNA) is that it is a common “substrate” for many enzymes involved in protein synthesis, a process in which tRNA plays a key role in transmitting the genetic information from nucleic acids to proteins. Almost nothing is known about the mode of interaction by which tRNA is specifically recognized by more than a dozen different proteins in the protein-synthesizing system. Transfer RNA also interacts with messenger RNA and possibly with ribosomal RNA. To understand these highly specific protein-nucleic acid and nucleic acid-nucleic acid interactions, it is imperative to know the three-dimensional (3-D) structure of tRNA. It is likely that there will be some conformational changes occurring during the recognition process by the various enzymes and nucleic acids. Nevertheless, the starting point of the studies of these specific interactions with tRNA is a well-defined 3-D structure of tRNA. In recent years, significant progress has ‘been made in the determination of the 3-D structure of tRNA. We now know the 3-D structure of one tRNA molecule in enough detail to answer some of the important questions and to provide a basis for specific experiments to answer other questions. This article reviews primarily the structural features of tRNA, their correlation with the various experimental results, and functional implications. In the past few years, several review articles have been published covering various aspects of tRNA, such as genetics ( I ) , biosynthesis and processing (2), interaction with aminoacyl-tRNA synthetases (3, 4 ) , regulation ( 5 ) and chemical modification (6). The role of tRNA in basic cellular processes is a diverse one. tRNAs participate in the following activities: (i ) ribosomes-mediated protein synthesis, where most of the reaction steps are understood in detail (for review, see 7); (ii) cell-wall synthesis in bacteria through special tRNA species (8); (iii) nonribosomal transfer of amino acid (for review, see 9); (iv) regulation of synthesis of certain enzymes (for review, see, e.g., 5, 10);( v ) DNA synthesis as a primer for the reverse transcriptase (11,I l a ) . A tRNA-like structure has also been found in the 3’ end of many plant viral RNAs, the role of which is not known (12). Thus tRNA is one of the most versatile macromolecules in the cell.

II. life Cycle of tRNA To put tRNA in perspective, a simplified life cycle as we understand it at present (mostly from the bacterial systems) is shown in Fig. 1 and described as follows.

183

THREE-DIMENSIONAL STRUCTURE OF TRANSFER RNA

I. Paptide Transfer

2. Translocalton

Racycling

FIG.1. Simplified life cycle of a tRNA molecule.

1. An RNA polymerase transcribes a tRNA gene to produce tRNA precursor, which contains extra nucleotides on the 5’ as well as the 3’ end of mature-size tRNA (23). Sometimes the precursor contains more than one tRNA (14).Some nucleosides may be modified at the precursor stage. 2, Processing enzymes recognize tRNA precursors and cleave the extraneous nucleotide to give a mature-size tRNA. Enzymes involved in Escherichia coli are endonuclease P, exonuclease 11, and perhaps other nucleases (for a review, see 2). 3. Some processed tRNAs lack the C-C-A sequence at the 3’ end. tRNA nucleotidyl transferase recognizes these incomplete tRNAs and adds on the C-C-A sequence. 4. Nucleosides at various positions of tRNA are altered by many different tRNA-modifying enzymes, which in most cases recognize the specific tRNAs. In some tRNAs, as much as 16%of the nucleosides are modified (for example, see 1 5 ) . 5. The mature tRNA is aminoacylated by the cognate aminoacyltRNA synthetase through a specific recognition process (for review, see 3, 4 ) . 6. The elongation factor T u forms a complex with tRNA in the presence of GTP, and the complex delivers aminoacyl-tRNA to the “ A site of the ribosome. 7 . tRNA is then recognized by various ribosomal proteins and pos-

J

184

SUNG-HOU KIM

sibly ribosomal RNAs during the processes of peptide formation and translocation. 8. tRNA is released from the ribosome and recycled. 9. Finally, tRNA is degraded by various nucleases into mononucleotides, which then join the nucleotide pool. Thus a tRNA molecule experiences a variety of recognitions by many different enzymes throughout its life cycle. Some of them are very specific for each tRNA (steps 4 and 5), and some are common to all tRNAs (steps 2,3,6 and 7 ) .

111. Secondary Structure of tRNA A. All tRNAs Have the Same Secondary Structure

Utilizing chromatography on DEAE-Sephadex, Holley et al. ( 16) established the first nucleotide sequence of an RNA, yeast tRNA"'", and proposed the well-known "cloverleaf" pattern for the secondary structure of tRNA. Minor corrections of the sequence have since been m,ade ( 17, 1 8 ) . Since then, the two-dimensional electrophoresis technique developed by Sanger et a,?. (19) and its improvements have greatly accelerated tRNA sequencing studies. Over 60 different tRNAs have been sequenced, and all can be arranged into the cloverleaf secondary structure pattern (for a convenient compilation, see 1 5 ) . A generalized cloverleaf pattern is shown in Fig. 2 together with the "L" pattern, which is a rearrangement of the cloverleaf pattern to simulate the 3-D structure of tRNA. The nomenclature used is given in Table I. The common features of the secondary structure are as follows. Starting from the 5' end, all tRNAs have 7 base-pairs in the amino acid tem (AA stem), 3 or 4 base-pairs in the dihydrouracil stem ( D stem), 5 base-pairs in the anticodon stem ( AC stem), and 5 base-pairs in the TI@ stem ( T stem). The base-pairs in the stems are of the WatsonCrick type, with occasional exceptions. There are always two nucleotides between the AA and D stems, one between the D and AC stems, and no nucleotide between the AA and T stems. There are always 7 nucleotides in the AC and T loops, 7-10 nucleotides in the D loop and 4-21 nucleotides in the variable arm ( V arm).

B. Classification of tRNA Among the three variable regions ( D stem, D loop, and V arm), the D stem and V arm were found to be very intricately involved in shaping the 3-D structure of yeast tRNAPhe(20, 21 ). Based on these two regions, tRNAs can be classified into four classes ( 2 2 ) by a minor extension of

185

THREE-DIMENSIONAL STRUCTURE OF TRANSFER RNA

?."

F

F

.1 1.1

p.

1.! I

A A arm

.

1

,A ,!.

r . ,

FIG.2. Generalized representation of tRNA in the "cloverleaf" form (left) and in the "L" form ( right). The bases common to all tRNAs (participating in the peptide elongation) are indicated. Other symbols are R for a purine nucleoside and Y for a pyrimidine nucleoside in oll tRNAs, r for a purine nucleoside and y for a pyrimidine nucleoside in most tRNAs, p for (5') terminal phosphate, and OH for (3') terminal hydroxyl. The dotted lines are the regions in the chain where the number of nucleotides varies among tRNAs.

the classification proposed earlier by Levitt (23), which was based on only 14 tRNA sequences (see Table 11). Although these classifications have no known functional implication yet, they are convenient for understanding the generalized 3-D structure of tRNA. The names have the TABLE I NOMENCLATURE FOR THE GENERALIZED tRNA Name Arm

Stem/loop

Short notation used here

Amino acid arm

Amino acid stem

AA stem

Dihydrouracil arm

IXhydrouracil stem Dihydrouracil loop Anticodon stem Anticodon loop Variable stem Variable loop T-+C stem T-p-C loop

1) stem D loop AC stem AC loop V stem v loop T stem T loop

Anticodon arm Variable arm T-p-C arm

Other names Stem a, acceptor stem, CCA stem Stem b, hU stem Loop I, hU loop Stem c Loop I1 Stem d, extra stem Loop 111, extra loop Stem e Loop IV

186

SUNG-HOU KIM

TABLE I1 CLASSIFICATION OF tRNA

a

Class"

Class*

D,Vs DIVN Da-4V4 D,Vs

Class I Class I11 (Class 11) Class I1

Number of Number of base pairs nucleotides in D stem in V arm 4 3 3 or 4 3

5 N 4 5

Occurrencee 50 % 24 % 14%

12 %

Names used in this review.

* Names proposed in Levitt (9.9). Occurrence among the tRNA sequences known at this time.

general form D,V,, where n is the number of base pairs in the D stem and m is the number of nucleotides in the V arm; for example, class D4V5 tRNAs all have 4 base pairs in the D stem, 5 nucleotides in the V arm. The class D3VNmay be further subdivided in the future depending on the possible involvement of the V arm in forming the 3-D structure. These four classes are shown in Fig. 3. Among nearly 80 tRNAs sequenced so far, 50% belong to the class D4V5,24%to D3VN,14% to D,-,V,, and 122 to D3V5.

C. Invariant a n d Semi-invariant Bases in tRNA Sequences As shown in Fig. 2, certain bases are invariant among all tRNAs in all four classes: U between the AA and D stems; A at the beginning of the D loop; two Gs in the D loop; U in the AC loop; a G . C base pair in the T stem; the T-gC sequence and A in the T loop (the initiator tRNAs from mammalian sources have A-U-C in place of T-+-C); t" c

D4V5

I"

+O* E

'3V5

c

'

3-4v4

FIG.3. Four classes of tRNA. Symbols are the same as in Fig. 2.

t 0"

f

'3'N

THREE-DIMENSIONAL STRUCTURE OF TRANSFER RNA

187

and the C-C-A sequence at the 3’ end. There are other bases that are invariant only in a particular class. Similarly, there are semi-invariant bases. The term semi-invariant is used when a position in the sequence is always occupied by the same type of base and indicated by R (for purine) or Y (for pyrimidine) in Fig. 2. For example, the base after the sequence T - $ 4 in the T loop is always either G or A; i.e., a purine. Most of these invariants and the semi-invariants are located in the D arm and the T loop. One may safely say that these residues must play important roles in providing either the recognition surfaces common to all tRNAs or the basic architectural framework to build up the 3-D structure of tRNA, or both. Another possible (but not very likely) reason for some invariants may be that they are necessary to ensure correct sequential folding of tRNA precursors as they emerge from RNA polymerase. X-ray crystallographic studies on yeast tRNAPhereveal that the majority of these invariant and semi-invariant bases are involved in forming the tertiary structure of this tRNA (see Section IV, D).

D. All Mature tRNAs Contain Many Modified Nucleosides One of the characteristic aspects of tRNAs is that a significant portion of the nucleosides are modified. In some tRNAs, close to 16%of the nucleosides are modified, mostly methylation ‘on bases or at the 2’ hydroxyl oxygen of the riboses. A list of these “rare nucleosides” can be found in reference 15, and work on this subject has been reviewed (23u, 5 ) . The functional role of most of the modified nucleosides is not known. Some of these, T and $ in the T loop, are invariant elements in all except mammalian initiator tRNAs. In most cases, these minor nucleosides appear not to be essential for the aminoacylation reaction. Two correlations exist regarding the minor nucleosides: ( a ) when the first letter of the codon is either A or U, the base at the 3’ side of the anticodon in the corresponding tRNA is very extensively modified ( 2 3 ~ ) ; ( b ) the invariant sequence GT-q-C in the T loop probably is recognized by the 5 S ribosomal RNA in prokaryotic ribosomes (see Section VI, C ) . In three 5 S ribosomal RNA sequences ( 1 5 ) , the complementary sequence G-A-A-C occurs approximately at the same positions in the sequences.

IV. Tertiary Structure of tRNA This section covers a simple conceptual description of the crystallographic method and the results of X-ray crystallographic studies on

,

188

SUNG-HOU KIM

TABLE I11 CRYSTAL DATAFOR Two FORMS OF YEASTtRNAPhe CRYSTALS Crystal form: Space group: Cell parameters: Molecules per asymmetric unit: tRNA in Crystal (by volume) : Resolution limit: References:

Orthorhombic P21221 33.2 X 5 6 . 1 X 161.0 A a =p = = 90”

Monoclinic P21 56.0 X 33.4 X 63.0 p = 90.5°;a= y = goo

1

1

40 % -2 36, 36

K

one tRNA, yeast tRNAPlle,in two different crystal forms. The orthorhombic crystal form has been studied by groups at the Massachusetts Institute of Technology and Duke University, and the monoclinic form at the Medical Research Council Laboratories at Cambridge, England and the University of Wisconsin. The crystal data for the two crystal forms are given in Table 111. The crystal structures have been determined by the multiple isomorphous replacement method ( MIR) (for a review, see 24) for the monoclinic form ( 2 1 ) , and the MIR method (25, 2 6 ) plus the partial structure method ( 2 7 ) for the orthorhombic form. Both structures are being refined using the real space method as well as the reciprocal space method. A. Optical Analogy of the Crystallographic Method

The crystallographic method of structure determination of biological macromolecules has been reviewed many times (for example, see 2 4 ) . For those who are intimidated by the apparent complexity of the X-ray diffraction technique, a simple analogy of the method is given below. X-ray crystallography can be considered as microscopy using X-rays rather than visible light. In light microscopy, light hits the object and scatters in all directions. These scattered beams of light are “recombined by the objective lens to form the image, which is then magnified by the eyepiece lens for viewing, This is possible because the lens can bend the visible light. Similarly, when X-rays hit the crystal they scatter in all directions. However, there is no “X-ray lens” to recombine the scattered rays to form the image. Instead, the intensities of the scattered X-rays in all directions are measured, and one “recombines” them muthemutically with the estimated phases to form the image, which is usually called the electron density map. The crucial step in the mathematical “recombination” of the scattered X-rays is the estimation of the phases

THREE-DIMENSIONAL STRUCTURE OF TRANSFER RNA

189

of the scattered X-rays. The best-known procedure is the isomorphous replacement method. This method relies on the fact that when the crystal is soaked in a solution containing heavy atom ions, such as those of mercury or platinum, the heavy atom often replaces mother liquor at certain specific sites without altering the conformation of the molecule (thus “isomorphous” replacement) and the scattered X-rays from this “heavy atom derivative” crystal have slightly different intensities from those of the “native” crystal. These intensity differences are utilized to help the mathematical “recombination” of the scattered X-rays to calculate the electron density map. The quality of the electron density map depends on many factors, such as the quality of the “native” and the “derivative” crystals, the amount of data used, and the quality of isomorphism. Even after the electron density map is obtained, the interpretation of the map requires the knowledge of the conformation of monomer and oligomer units, constraints on the bond rotation in the backbone, and the base or aminoacid sequence. The term “interpretation” is used because in most cases one sees the peaks of electron densities in the map representing phosphates, ribose and bases rather than individual atoms, and 3-D atomic models are built to fit the electron density peaks. The principles and techniques of X-ray crystallography other than the simple conceptual description given above can be found in many review articles (see, for example, 2 4 ) .

B. General Remarks on X-Ray Crystallographic Studies on Yeast t R NAP”e The 3-D backbone structure of yeast tRNAPhewas first determined in the orthorhombic crystal form at 4 A resolution ( 2 5 ) , and the tertiary interactions between the bases have since been determined at 3 A resolution in the orthorhombic (24) and monoclinic (21) crystal forms. Since the initial success of obtaining microcrystals (28) and single crystals of tRNA suitable for X-ray crystallographic studies ( 2 9 3 2 ) , three factors have made important technical contributions to the eventual determination of the 3-D structure of yeast tRNAPhe:( a ) the application of the vapor diffusion method to tRNA by Hampel et al. (31, 33) to screen quickly the large number of crystallization conditions with a minimum amount of material, ( b ) discovery of the crystallization condition under which yeast tRNAPhrforms a crystal that gives high-resolution X-ray diffraction data (34-36), and ( c ) discovery of the first heavy-atom derivative, osmium trioxide ( pyridine) complex by Schevitz et al. (37). The 3-D structures of yeast tRNAPhein both crystal forms appear to be almost identical. This is understandable from the fact that the two

190

SUNG-HOU KIM

crystal forms have two almost identical axial lengths and a similar distribution of X-ray diffraction intensities (38, 39). The chronological development of the crystallographic studies up to 1974 on yeast tRNAPhe in the two crystal forms has been reviewed by Sigler (40).

C. tRNA is 1-Shaped In early 1973, from X-ray crystallographic studies at 4 A resolution ( 2 5 ) , the backbone structure of yeast tRNAPhewas found to have the following characteristics, differing from a dozen or so proposed models. 1. The molecule has the overall shape of the letter “L.” 2. The base-paired double-helical stems implied by the cloverleaf secondary structure are maintained in the 3-D structure, and they are all right-handed antiparallel helices. 3. The AA and T stems form one continuous double helix, and the D and AC stems form the other approximately continuous double helix. These two approximately continuous double helices form an “L.” 4. The C-C-A and anticodon termini lie at the two ends of the “L.” 5. The D and T loops form the corner of the “L.” 6. The molecule is flat with the thickness of about 20 A, which corresponds to the width of a double-helical RNA ( 41 ) , All the above features have since been reconfirmed by 3 A resolution studies on yeast tRNAPhecrystals in the orthorhombic form (20, 26) and in the monoclinic form (21 ) , D. Most of the Invariant a n d Semi-invariant Bases Form Hydrogen Bonds among Themselves

Figure 4 shows the nucleotide sequence of yeast tRNAPhe( 4 2 ) .Also indicated are the invariant and semi-invariant bases for the class D4V5, where this tRNA and 50% of all the tRNAs sequenced so far belong. The hydrogen bonds in the base-pairs in the stems are indicated by dots, and other H-bonds (tertiary hydrogen bonds) by lines connecting the bases. These hydrogen bonds are discussed below in sequence starting from the 5’ end and using the numbering system for the atoms shown in Fig. 5. 1. All the base-pairs in the stems except one are of the Watson-Crick type shown in Figs. 6(a ) and 6(b ) (20, 2 1 ) . 2. G4-U69 ( G at position 4 and U at position 69 in the nucleotide sequence) form a base-pair. In the monoclinic crystal, this base-pair is of the type shown in Fig. 6(c), and it does not distort the helical backbone ( 2 1 ) . However, in the orthorhombic crystal, even though the exact nature of the base-pair is not clear, the G-U pair appears to break the continuity of the helical backbone slightly without interrupting the continuous stacking of the bases (20, 2 7 ) .

191

THREE-DIMENSIONAL STRUcTURF, OF TRANSFER I W A

T arm

AA arm

. . .C G r n

uu

C$

A,G G C G P

b

FIG.4. Nucleotide sequence of yeast tRNAPhe( 4 2 ) ( a ) in “cloverleaf‘ form and ( b ) in “L” form. The bases involved in making tertiary base-base H-bonds are connected by lines. The only nonstandard symbol used is H for “hypermodified guanosine.” The invariant (boxes) and the semi-invariant or almost semi-invariant ( circles) residues are indicated. Note in ( b ) how these are clustered around the comer of the “L.”

3. U8 forms a base-pair with A14 [Fig. 6( d ) ] utilizing N7 of adenine (20-22). This type of base-pair has been observed previously ( 4 3 ) . 4. A9 forms a base-pair with A23, as was observed in the crystal

FIG.5. The numbering system for the bases and ribose.

192

SUNG-HOU KIM

structures of adenine hydrochloride ( M ) ,U-A (44u), and the fiber of poly(A) ( 4 5 ) ; A23, in turn, is base-paired to U12 [Fig. 6 ( e , ) ] (20-22). This type of base-pair involving three bases has been observed in the crystal structure of a complex between 2 adenines and 1 barbiturate (46),the barbiturate being the analog of uracil. 5. G15 and C48 m.ake a base-pair as shown in Fig. 6 ( f 2 ) (20-22, 4 7 ) . These five hydrogen-bonding features have been observed in both crystal forms of yeast tRNAPhe,and all of them involve the invariant or semi-invariant bases. The following hydrogen-bondings between the bases have been observed in the orthorhombic crystal form (20, 27) , but appear uncertain or uninterpretable in the monoclinic crystal form ( 2 1 ) at the current A

U

FIG.6. See legend on opposite page.

193

THREE-DIMENSIONAL STRUCTURE OF TRANSFER RNA

CH \3

46

FIG. 6 . Various types of H-bonding observed in the crystal structure of yeast tRNA""" and postulated for other tRNAs. The filled black circles indicate the C1' position of the riboses. In ( e2) and ( f ? ) , alternative positions for guanine are illustrated by dotted lines.

stage of refinements of the 3 A resolution data.' It is likely that most of these bonds will be proved to be correct. The additional features are: 6.T54 makes a base-pair with A58 [Fig. 6(i)] (20-22). 7. G18 is approximately on the same plane as $55 and probably makes one hydrogen bond ( N 2 of G18 to 0 4 of $55) (20). 8. G19 forms a Watson-Crick type of base-pair with C56 (20). 9. G45 appears'to form an H-bond to G10 (27). * See Note Added in Proof, p. 216.

194

SUNG-HOW K I M

E. Other Tertiary Interactions Besides the base-base H-bonding, other interactions among base, ribose, and phosphate have been inferred to be present in the 3-D structure at the current stage of refinement. Base-ribose interaction. Two likely such H-bonding interactions are between 02’ of U8 and N1 of A21 (21, 2 7 ) ; 02’ of A21 and 0 6 of G46. In both cases, 02’ of ribose is involved and both are at the region where a sharp bend of the polynucleotide chain is observed. Ribose-ribose interaction. The most common H-bond is between 02’ of one ribose and 01’of the following ribose in the sequence. There are many candidates for this type of H-bonding ( 2 1 , 2 7 ) . Phosphate-phosphate interaction. The closest interaction of this type is between two phosphates of U8 and C49 ( 2 7 ) and C48 and U50 ( 2 7 ) They are as close at 4-6 A apart. Metal-ion interaction. At the current stage of refinement, the positions of metal ions such as Mg2+ are difficult to locate unambiguously. F. There Are Two Extensive Base-Stacking Systems, Each Running Half of the Molecule One of the most spectacular features of the molecule is that almost all the bases are stacked. This is shown schematically in Fig. 7 [see also Figs. 8 ( a ) and 8 ( b ) l . There is one base-stacked column running horizontally and another vertically. Thus, at the end of each stem, the helix is augmented along at least one strand. Such an extensive base-stacking system was observed in both crystal forms (20-22) with one major difference: in the monoclinic crystal form, G26 is partially intercalated between A44 and G45 ( 2 1 ) ,while in the orthorhombic case it is not (20, 2 2 ) . This difference may be due to the difference in crystal lattice in the two crystal forms (see Table 11). If this is a real difference,* it means that the angle made between the D and AC stems is different in the two crystal forms. The implication then is that the AC arm containing the anticodon has some flexibility, which may be necessary for the interaction between tRNA and the ribosome. This extensive stacking interaction probably is the primary source of the stabilizing force in the 3-D structure of tRNA, and is expected to be a general feature for all tRNAs. The atomic model and a simplified artistic drawing of the 3-D structure of yeast tRNA (20, 27) are shown in Figs. 8 ( a ) and 8(b). Three stereo views are shown in Fig. 8( c). ‘See Note Added in Proof, p. 216.

THREE-DIMENSIONAL STRUCXURE OF TRANSFER RNA

195

FIG. 7. A schematic diagram illustrating the base-stacking interactions in yeast tRNA"h" in the orthorhombic crystal (20, 2 2 ) . The orientation of the molecule is similar to that of Fig. 4 ( b ) . Where adjacent stacked bases are connected by a ribosephosphate backbone, the linkage is indicated by a thin line. The thicker lines represent the bases, and the base-stacking interactions are indicated by the blocks between the thicker lines. The covalent connections of the nucleotides in the molecule can be followed by the base numbering. Four bases (16, 17, 20, 47) that are not involved in base-stacking are omitted from the figure.

G. Conformation of Nucleosides in tRNA All the riboses are in the 3'-endo conformation except those from residues 7, 9, 17, 19, 21, 46,48 and 60, which are in the 2'-enclo conformation and located either in loop regions or at the end of the stem. All the bases are in the anti conformation with respect to their riboses except the base of residue 19, which is in between anti and syn conformation, and the bases of residues 17, 60 and 76, the conformations of which are not certain. Assignmcnt of nuclcosides are not unambiguous at the present stage of refinement. However, it is certain that not all the riboses are in the 3'-endo conformation.

196

SUNG-HOU KIM

AA arm

k T 0 r m - I

+

20

i

10

i

0

antiiodon

FIG. 8a. A drawing of the 3-D structure of yeast tRNA”’“ in the orthorhombic crystal (20, 27). The phosphate-ribose backbone is indicated by a thin line, and the bases are indicated by slabs. The whole molecule is flat with an average thickness of about 20 A. The molecular plane is approximately parallel to the page. Secondary and tertiary H-bond interactions between the bases are shown by either fusing two slabs together or connecting the slabs with a line. From Kim ( 1 1 5 ~ ) .

H.

General 3-D Structure of tRNA Since most of the tertiary H-bonding between the bases involves the invariant or semi-invariant bases, one can safely assume that those

H-bondings will be preserved in all tRNAs (20-22, 4 7 ) . Base-pairs (two or more H-bonds between the two bases) expected to be common in class D,V, tRNAs are: 8-14 pair. I n all tRNAs, this pair is U * A and expected to have the base pairing scheme shown in Fig. 6( d ) . 9-23-12 triple. Either A*A-U or G - C - G , as shown in Figs. 6(e,) and ( e2). In both cases, the relative positions of the riboses are similar, thus replacing one triple with the other will not change the backbone structure significantly at this region in the molecule. These three positions are a “coordinated-invariant triple”. 15-48 pair. Either G * C or A*U as shown in Figs. 6 ( f , ) and (f2). Here again the replacement of one by the other will introduce only a slight change in the backbone conformation of that region, thus a “coordinated-invariant pair.” 1 9 5 6 pair. In all tRNAs, this pair is G - C as shown in Fig. 6( g ) .

THREE-DIMENSIONAL STRUCl’URE OF TRANSFER RNA

197

FIG. 8b. Photograph of the skeletal molecular model of yeast tRNAPhe in the orthorhombic crystal built to fit the electron density map at 3 A resolution (20). The supporting rods for the model have been removed from the photograph. The viewing direction is approximately perpendicular to the molecular plane, which is 20 A thick on the average. The amino-acid-accepting end is at the upper right, the anticodon at the bottom, and the T-$-C loop at the upper left. From Kim et al. (20).

13-2246 triple. Either C . G - G or U-A*A as shown in Fig. 6(h,) and ( h ? ) . Here again, the relative positions of the riboses in both triples are similar, another “coordinated-invariant triple.” 54-58 pair. This is T * A in all tRNAs except the initiator tRNA from mammalian cells. The observed base-pairing is shown in Fig. 6 ( i ) . These six types of base-pairs or -triples will provide a very specific

/’

FIG.8c. See legend on opposite page.

THREE-DIMENSIONAL STRUCI’URE OF TRANSFER RNA

199

architectural network, which can be further stabilized by the extensive nonspecific stacking of the bases, as was found with yeast tRNAPhe.Similar generalizations can be made for the other classes of tRNAs.

I. Crystallographic Studies of Other tRNAs Among other crystallographic studies of tRNAs, the following two are making good progress. Yeast initiator tRNAMethas been crystallized and the electron-density map prepared with 5.7 A resolution data using five heavy-atom derivatives ( 4 0 ) . However, the quality of the map is not good enough for full interpretation. Nevertheless, some structural information has been obtained. Two of the heavy atoms serve as markers in the electrondensity map because one is covalently bonded to C38 and the other to C73. The distance between these two points is about 54 A (@), which agrees well with 56 A observed in the yeast tRNAPhestructure (20). The 3-D structure of this tRNA will definitively answer the question of whether the initiating tRNAs have a 3-D structure different from other tRNAs. Independent crystallographic studies ( 4 8 ) on yeast tRNAPhein the monoclinic crystal yield results that so far agree with the structure described in Section IV, C.

V. Correlation between the 3-D Structure and Physical and Chemical Data

The 3-D structure of yeast tRNAPheis consistent with most of the results of experiments that were aimed at obtaining structural information. A. Small-Angle X-Ray Scattering

For particles in a nearly random arrangement, as in a gas or dilute solution, the angular dependence of the scattered X-ray intensity closely approximates that from a single particle averaged over all orientations. Usually the intensity decreases rapidly within a few degrees of scattering angles (thus “small-angle”). The shape of the curve, when scattered X-ray intensities are plotted against scattering angles on a log-log scale, FIG. 8c. Three stereo views of yeast tRNAr’”. The C1’-backbone structure is viewed along an axis perpendicular to the molecular plane ( 1 ) and along the pseudo twofold axis ( 2 ) . The complete atomic structure viewed along the same axis as in ( 1 ) is shown in ( 3 ) .

200

SUNG-HOU KIM

provides information about overall orientation-averaged size and, often, shape. The radii of gyration for the various tRNAs in aqueous solution have been measured to be around 23.5-25 A (49-53). The calculated value for yeast tRNAPheusing only the phosphate groups is 24 A (S. H. Kim, unpublished result),

B. Fluorescence Energy Transfer This method of estimating the distance is based on the observation that energy absorbed by a chromophore can be transmitted to another chromophore some distance away. In singlet-singlet transfer, the return of the energy donor from the lowest excited state to the ground state is coupled to the excitation of the energy acceptor from the ground state to a higher singlet level. This type of energy transfer occurs over the appreciable distance of tens of angstroms and is suitable for the distance estimation in biological macromolecules ( 53u), provided that there is a single donor and a single acceptor at specific sites on the molecule, that information on the relative orientation of the donor-acceptor pair is available, and that the distance at which singlet-singlet transfer is 50% efficient is comparable to the magnitude of the distance measured. Singlet-singlet energy transfer between the hypermodified base H37 and an energy acceptor such as acriflavin, proflavinyl acetic acid, hydrazide, or 9-hydrazinacridine at the 3' end of yeast tRNAPtlepermit an estimate of the distance between these two points to be greater than 4 0 A (54). Recently, similar experiments have been carried out with fluorescent groups attached on the 5' end, on position 8 (located between the AA and T stems), on dihydrouracil in the D loop, on t/, in the T loop, at the 3' terminus of E. coli tRNAeMet, and on s2U in the anticodon and at the 3' end of E. coli tRNAQ"' ( 5 5 ) . The distances between these points have been estimated to be 24 A between the 5' and 3' ends, 38 A between position 8 and the 3' end, 55 A between t/, and the 3' end, 36 A between t/, and D, and greater than 65 A between the anticodon and the 3' end. The corresponding distances in yeast tRNAPhe in the orthorhombic crystal (20, 27) are 16 A, 42 A, 52 A, 25 A or 20 A, and 67 A. The distances involving the 3' terminus require special consideration, because the single-strand stretch at the 3' end of a free tRNA is expected to be flexible in solution. In addition, the distance between the base at the 3' end and the fifth base on the same strand is likely to range between 15 A when the bases are stacked and 22 A when the backbone is in extended conformation. With this consideration in mind, one can say that the ex-

201

THREE-DIMENSIONAL STRUCTURE OF TRANSFER RNA

TABLE IV SOMEUSEFULDISTANCES (A) BETWEEN THE BASESI N YEASTtRNAPhe I N THE ORTHORHOMBIC CRYSTAL Base

U8

A35

C56

C72n

U8 A36 C56 C72

0 40 26 30

40 0 60 61

26 60 0 46

30 61 46 0

Since the single strand between A73 and A76 is expected to be flexible in free tRNA, C72, in the last base-pair of the AA stem, is chosen instead of A76. The distance between C72 and A76 can range between 15 for stacked and 22 for extended conformation. 0

perimental results of this type are approximately consistent with the 3-D structure of yeast tRNAPhein the orthorhombic crystal (20) and are likely to be so with the monoclinic form (21). Some useful distances in yeast tRNAP1lein the orthorhombic crystal are given in Table IV.

C. Complementary Oligonucleotide Binding In an aqueous solution of moderate ionic strength, a short oligonucleotide of a specific sequence, as small as a trinucleotide, can bind specifically and strongly to a free, single-stranded region of RNA that contains a sequence complementary to that oligonucleotide. Short oligonucleotides have been employed as “probes” to search for the singlestranded regions of tRNA that are not shielded by secondary or tertiary interactions (56, 57). Such experiments have been done on all regions for E . coli tRNATyr, tRNAfMet’; yeast tRNAflle, tRNALeu,and initiator tRNAMet(58-61). The results agree with the crystal structure as to the availability of the AC loop and the tetranucleotide at the 3’ end of tRNA. Four out of five tRNAs so examined have T loops not exposed (the exception is given in reference 50), and this agrees with the 3-D structure of yeast tRNAP1le(20, 21 ). The availability of the D loop and the V loop as studied by this method varies among different tRNAs, and this may mean that there is competition for these regions of the molecule by complementary oligonucleotides and other parts of the molecule (60).

D. Low-Field Nuclear Magnetic Resonance NMR) Spectroscopy Since the discovery by Kearns et aZ. (62) that the exchangeable, H-bonded hydrogens attached to the ring nitrogen ( N 1 of G and N 3 of U ) produce NMR resonance spectra shifted sufficiently far downfield from the huge water peaks, great progress has been made in NMR

202

SUNG-HOU KIM

spectroscopic studies on tRNA in aqueous solution. These spectra were found to be further resolved by the “ring current” of the adjacent bases or base-pairs that are stacked. Thus the low-field NMR (-11 to -15 ppm; standard reference compound 3-trimethylsilyl-1-propanesulfonic acid) provides a sensitive clue as to the presence or absence of stacking by the neighboring bases as well as the estimate of the total number of base-pairs involving exchangeable protons of ring nitrogens in tRNA and other structural information. The spectral assignments in this region are based on two lines of observations: ( a ) NMR spectra from tRNA fragments containing doublestranded stems produce resonance peaks in this region that correlate with the number of base-pairs in the stems ( 6 4 ) ; ( b ) a semiempirical rule can be derived ( 6 5 ) that can predict approximately the NMR spectral positions from the theoretical calculation ( 66) of “ring-current shifts,” assuming that the bases are stacked as in the 12-fold RNA-A’ double-helix ( 4 0 ) . The interpretations of the results are summarized below. 1. All the Watson-Crick base-pairs predicted from the cloverleaf secondary structures are present. Initially it was concluded that these are the only base-pairs in yeast tRNAPhe( 6 7 ) .Subsequent reinterpretation of the NMR spectra suggests that a few additional base-pairs are present (68),consistent with the 3-D structure of yeast tRNAPhedetermined by X-ray crystallographic methods (20-22). 2. G * U base pairs, which occur often in the AA stem, are consistently undetectable (67, 69), implying that the G - U pair is not present or the base-pair is formed with very weak H-bond(s) so that the proton exchange is too fast to be observed in the measured region. The ‘cwobble” G - U base-pair has been observed by X-ray studies in yeast tRNAPhe in the monoclinic crystal form ( 2 1 ) ,but is uncertain in the orthorhombic crystal form ( 2 0 ) . Recently, another region (-10 to -11 ppm) has been tentatively assigned to the resonance of the G - U pair ( 6 8 ) . 3. The helical parameters for the four stem regions of tRNA are very close to the 12-fold RNA-A’ type. X-ray studies indicate that three stems (AA stem, T stem, AC stem) are closer to the 11-fold RNA-A type ( 4 0 ) , and the D stem is different from the others. The difference between the two forms is very small. 4. The amino-acid stem is stacked on the T stem ( 7 0 ) .This is consistent with the X-ray results first observed at 4 A resolution of yeast tRNAPhe(25) and subsequently at 3 A resolution ( 2 0 , 2 1 ). 5. The ends of the helical stems are usually stacked by the adjacent bases in the single strand (70, 71). This again, in general, agrees with X-ray crystallographic res,ults (20, 21 ) .

203

THREE-DIMENSIONAL STRUCTURE OF TRANSFER RNA

E. NMR Studies on Methyl a n d Methylene Proton Resonances Yeast tRNAPhecontains 14 methyl groups and 4 methylene groups belonging to 12 modified nucleosides, and they are distributed throughout the molecule in all except the AA arm. Kan et al. (72) were able to assign these methyl and methylene resonances by careful comparison of the resonances of the intact tRNA with that of the mononucleoside, mononucleotide and appropriate oligonucleotide at various temperatures. By measuring the chemical shifts and line widths of these resonances as a function of temperature between 10°C and 80"C, they reached the following conclusions: 1. Methyl groups in the anticodon loop of yeast tRNAPhe (from Cm32, Gm34 and H37) are in a similar magnetic environment as those in mononucleotides and dodecamers containing nucleotides 31 42, the implication being that the whole anticodon loop does not associate with other parts of the molecule in either the absence or the presence of the Mg" ion (Ts'o and Kan, personal communication). 2. Methyl resonances of m5C40, m5C49, m2G10 and m'A58, even though these bases are in the double-helical stems, do not show significant chemical shift changes in the absence of Mg2+ions as the temperature is raised, again implying that these methyl groups are not near diamagnetic regions in the molecule. However, methyl resonances from mC40 and mT49 experience thermal transition in the presence of Mg2+ ions (Ts'o and Kan, personal communication). 3. Methyl resonances from miG26 and T54 and the methylene resonances from Dl6 and D17 go through dramatic chemical shifts as well as line-width changes on heating, implying that they are within strong diamagnetic regions of the molecule and not free. All these observations can be understood in terms of the 3-D structure of yeast tRNAPheexcept that of the thermal transition of methylene resonances from D16 and D17 (the methyl resonance of m'A58 in the presence of Mg" is uncertain). It is not clear at present whether these discrepancies are due to the difference between tRNA conformation of that region in solution and in the crystal, misinterpretation of NMR data or X-ray data, or different experimental conditions used. Nevertheless, the NMR study of methyl and methylene groups promises to be one of the very powerful methods of studying the change of conformation of tRNA in solution.

-

F. UV-induced Cross-Link Yaniv et al. (73) found that when E . coli tRNAV*' is exposed to UV light, the 4-thiouracil at position 8 becomes linked to cytosine at

204

SUNG-HOU KIM

position 13. This phenomenon has been observed in many other E. coli tRNAs. Furthermore, the photo cross-linked tRNAs still can be charged with amino acids by the cognate synthetase and recognized by nucleotidyltransferase (74) and Tu factor (75). This phenomenon is understandable from the 3-D structure of yeast tRNAphe,where U8 is basepaired to A14 by a “reverse Hoogstein scheme” [Fig. 6 ( d ) ] and stacked on C13, thus allowing the UV-catalyzed cross-linking to occur between the two stacked bases when U8 is 4-thiouracil (76).

G. Base-Specific Chemical Modification So far, this method has provided the highest resolution information as to the degree of exposure of certain bases (for review, see 77). The reagents and specific bases attacked are listed in Table V. The basic procedure is as follows: allow tRNA to react with one of the reagents in Table V; enzymically digest the reaction product to obtain oligonucleotides; identify those oligonucleotides that have altered mobilities compared to the unreacted oligonucleotide pattern by column or gel chromatography; and finally, identify which base is altered by further digestion of the altered oligonucleotides. Such experiments have been carried out with many pure tRNAs: E. coli tRNAy’”, tRNA&, tRNA‘Met, tRNALe“;yeast tRNAVa’,tRNAAta,tRNAPhe,tRNATyr,tRNAser; mammalian initiator tRNAMet(78-83, 92; for an earlier review up to 1970, see 6 ) . The assumption behind the interpretation of the selective chemical modification results is that a base that is not modified by a reagent specific for that base is inaccessible because it is in the doublehelical stems, or involved in a tertiary H-bonding, or stacked or intercalated, etc. A summary of these results for D,V, class tRNAs is shown in Fig. 9. As can be seen from the figure, most of the bases at the positions that are protected from chemical modification are involved in tertiary H-bonding in yeast tRNAPhe (see Fig. 4 and Section IV, D). Among the remaining protected bases, two bases (U59 and C60) are buried inside the molecule (20); the rest are either stacked on the adjacent base-pairs (20, 21, 27) or H-bonded to other moieties (see Sections IV, D and IV, E ) . In conclusion, the results from the base-specific chemical modification are in extremely good agreement with the 3-D structure at the individual base level (20, 21, 82, 8 4 ) .

H.

Tritium Exchange of Free Purines

The proton at the C8 position of purines is slightly acidic and can exchange with tritium unless the purines are protected by the secondary

205

THREE-DIMENSIONAL STRUCTURE OF TRANSFER RNA

TABLE V BASIC-SPECIFIC CHEMICAL RE.4GlCNTS Reagents

Hydroxyamine, methoxyamine

pH range for reaction

Base attacked

I

Cytosine

Reaction product"

I

I R

R +R"H

0

+WNH

o

k

R

NH Monoperphtalic acid

7

Adenine (cytosine)

I R

Kethoxal

7

Guanine

I Borohydride

6-10

Dihydrouracil

Bisulfite

5.0

Cytosine

a

R

NH2-C-N-CH2-C&-

b\A

A CH2-OH

As nucleoside in IlNA (It = ribosyl).

or tertiary structure. The protection pattern obtained by Gamble and Schimmel ( 8 5 ) in yeast tRNAPhecan be explained if one assumes that the molecule in solution has a 3-D structure similar to that of the crystal. The resolution obtainable from this type of experiment is presently not as good as that from base-specific chemical modification (Section V, G).

206

SUNG-HOW KIM

D Base attacked b Bare

not attacked

0Constant

nucleotide

I-

: ,1

Constant purine or pyrimidine

A4

A * U G,, m5C4, A ' Y Cm Ad

0

vGm

a

& :

350 A

FIG.9. Summary of the results of the base-specific chemical modification experiments on many D,Vs class tRNAs is shown in relation to the nucleotide sequence of yeast tRNAPhe. H = hypermodified guanosine. The circled bases are invariant, and the dashed circles indicate the semi-invariant bases in all tRNAs participating in peptide elongation (i.e., a G thus marked may be A, a C may be a U in other D4V6-classtRNAs ) . The accessibility of various bases to chemical modification is indicated by open or filled triangles. The half-filled triangles indicate the bases with possible partial exposure.

I. Conformation of tRNA in Solution and in the Crystal There is always some doubt as to whether the crystal structure determined is the same as the structure in solution, especially when one considers the molecular contacts necessary to build up the crystal lattice. The solution conformation as measured by circular dichroism ( C D ) is quite variable depending on the specific solvent conditions ( 86). However, the interpretation of CD change in terms of the quantitative extent of the change in conformation is difficult at the present time. Overall consistency between the 3-D tRNA structure in the crystal and the structure in aqueous solution as examined by the physical and chemical experiments mentioned above is quite good. Additional evidence for the similarity of the conformations in solution and in the crystal comes from the direct comparison of the fluorescence life-times, in crystal form

THREE-DIMENSIONAL STRUCTURE OF TRANSFER RNA

207

and in solution, of the hypermodified base that lies at the 3’ side of the anticodon of yeast tRNAPhe. The fluorescence properties of this base are very sensitive to the environment, i.e., to the conformation of the anticodon loop (87, 5 4 ) . The fluorescence life-time of this base for tRNA in the orthorhombic crystal is identical with that in the solution from which these crystals were grown, and very similar to that in the dilute solution of tRNA (88). Thus within the limits of the sensitivity of the method, the conformation of the anticodon loop around this hypermodified base is the same in the crystal and in solution. In conclusion, one can say that the conformation of free tRNA in solution is likely to be almost the same as that found in the crystal, especially since about 751%of the crystal volume is occupied by buffer as is the case with an orthorhombic crystal of yeast tRNAPhe(89).

VI. Functional Implications of the 3-D Structure of tRNA A. Protein-tRNA Recognition As pointed out in Section 11, tRNA goes through many different recognition processes with various proteins. Some proteins, such as EF-Tu factor and tRNA nucleotidyltransferase, presumably recognize those features common to all tRNAs, some recognize only certain specific tRNAs, and others behave in between. The first case fs easy to understand. The proteins in this category probably recognize the common phosphate-ribose backbone structure (see Section IV, H ) , or the invariant and semi-invariant bases (see Fig. 3 ) , or the combination of both in tRNA. Recognition of the second type is difficult to predict and is discussed below in terms of the interaction of aminoacyl-tRNA synthetase and its cognate tRNA. Examination of the 3-D structure of yeast tRNAPhereveals that most of the methyl groups in modified nucleosides appear not to be essential in maintaining the tertiary structure, thus implying a functional role for them as recognition markers for certain proteins.

B. tRNA-Synthetase Interaction The fidelity of the genetic information transfer depends primarily on the accuracy in ( a ) aminoacylation of each tRNA by its cognate aminoacyl-tRNA synthetase ( amino-acid:tRNA ligase) and ( b ) the codon-anticodon interaction. The tRNA-cognate synthetase interaction is likely to involve two types of recognition: a nonspecific recognition

208

SUNG-HOU KIM

over a large area to provide stability to the complex, and a specific recognition to distinguish between a cognate and a noncognate partner. Search for the recognition site(s) on tRNA by the cognate ligases has been going on for many years. So far, almost every part of tRNA except the D loop and T arm has been implicated. Since there are a few recent reviews on the subject (3, 4 ) , we here only summarize the existing hypotheses and introduce a new one that looks at the problem from a different point of view.

1. ANTICODON a. The chemical modification of the bases in the anticodon affects the charging property drastically for some tRNAs such as yeast tRNAraI (W), E. coli tRNAG1y(91),E. coli tRNAMet(92,93). b. Enzymic cleavage and subsequent removal of bases in the anticodon relinquish the amino-acid acceptor activity for yeast tRNAr” (94), but not for yeast tRNAAIa(95), tRNAPhe(96), and tRNATyr (97) although the acceptor activity is decreased for the latter group. c. The E. coli glycine ligase loses its affinity for tRNA drastically when either the second or third base of the anticodon (i,e,, the two bases other than the “wobble” base) of a suppressor tRNAQ’yis changed (98, 99). d. An E. coli suppressor tRNATrp can be charged in uiuo with glutamine when the anticodon of the tRNATrpis changed from C to U at the second position (100).

2. THE AA STEM Genetic manipulation of E . coli amber suppressor su+ tRNATyr resulted in a few altered su: tRNATYPspecies that then accepted glutamine and tyrosine (for most of them). All these mutant tRNAs had the base alterations in the AA stem near the 5’ end or 3’ end (101 ). The heterologous charging of nine E. co2i tRNAs plus yeast and wheat germ tRNAPheby yeast phenylalanine ligase indicates that the fourth base from the 3’ end may be involved in the recognition (102) (see also “The D stem” below). 3. THE D STEM a. Yeast phenylalanine ligase can charge phenylalanine not only to yeast tRNAPhebut also to wheat germ tRNAPlle,E . co2i tRNAPhe,tRNAYal, tRNAT& tRNAT?;, tRNAMet,tRNA1Ie,tRNA:la, tRNAf:”, and tRNArdrs with similar K, values but with wide variations in V,,,, (102). The nucleotide sequences of the first nine are known. Comparison of these

THREE-DIMENSIONAL STRUCTURE OF TRANSFER RNA

209

sequences indicates that the only common sequence besides the T loop is in the D stem, nucleotide number 9, and the fourth base from the 3’ end (102). However, under the appropriate conditions, almost all E . coli tRNAs could be charged by yeast phenylalanine ligase ( 1 0 4 ) or E . coli valine ligase ( 105,105a ), b. Ultraviolet light cross-links tRNAs and the ligases. The RNase-T1 digests of such cross-linked material have been analyzed for E . coli tRNATyr (106),tRNA1“ (107),and yeast tRNAPhe(108).The results are difficult to interpret until the specific bases rather than just the oligonucleotide fragments can be identified. However, this approach is very promising.

4. THEV ARM Chemical modification of a G in the V arm of E . coZi tRNArMeteliminates its interaction with the cognate ligase ( 109).

5. THEAC STEM In an E . coli suppressor tRNATyr,a base change by mutation in the AC stem affects the charging property considerably (110). It is clear from these observations that there are no unique specific recognition sites universal to all tRNAs. Besides, the molecular weights of the aminoacyl-synthetases range from 70,000 to 270,000, superficially implying that there is no unified scheme in the interaction between tRNAs and ligases.

6. SYMMETRY RECOGNITION HYPOTHESIS There are three general comments that one can make about tRNAs and their ligases. ( a ) An examination of the 3-D structure of yeast tRNAPhein the orthorhombic crystal ( 2 0 ) reveals that the molecule contains a pseudo 2-fold axis as shown in Fig. 10. This may be a general feature for all tRNAs. ( b ) When the regions implicated as ligase-binding sites (reviewed above) are identified in the 3-D structure, all except the V arm are located “inside” the “ L as shown in Fig. 10. ( c ) Some ligases contain an even number of the “small protomers” (molecular weight of 35,000-50,000) and others contain one or more subunits of the “large protomers” ( molecular weight equal or greater than about 70,000) (for review, see 3 and 4).It has been suggested that the “large protomer” is composed of two homologous sequences (111-113). This sequence homology suggests that the 3-D structure of such protomers is likely to contain two structural domains as in the crystal of the Fab’ fragment of a human immunoglobulin (114) (for review, see 115),

210

SUNG-HOU KIM

f-

FIG. 10. Two orthogonal views of yeast tRNAPhe in the orthorhombic crystal (11Sa). The large, medium and small circles represent the centers of the phos-

phates, the riboses and the bases, respectively. The pseudo 2-fold axes are indicated by + when on the plane of the page and 0 when perpendicular to it. The pseudo %fold symmetry-related backbones are shown in black. The aminoacyl-tRNA synthetase recognition sites implicated by various experiments are circled by dotted lines.

where the two structural domains in one chain are related to each other by a pseudoscrew axis. Therefore, an aminoacyl-synthetase can be considered as a multidomain enzyme consisting of an even number of such “unit domains,” where two domains in each pair are related to each other by a pseudosymmetry. The hypothesis is that the pseudo 2-fold symmetry of a tRNA is recognized by that of an aminoacyl-synthetase. This is shown schemati) that there is a unified scheme cally in Fig. 11. The proposal ( 1 1 5 ~ is by which tRNAs are so recognized and the specific recognition that discriminates between cognates and noncognates is achieved by the specific base or base-pair recognition within or near the symmetrically related contact areas. During such interactions, it is possible that small conformational changes will occur in both molecules.

THREE-DIMENSIONAL STRUCTURE OF TRANSFER RNA

211

FIG.11. Two orthogonal views of ( a ) tRNA, ( b ) two symmetry-related “unit domains” (hypothetical) in an aminoacyl-tRNA synthetase. ( c ) and ( d ) are two alternative hypothetical models for the tRNA-enzyme complex formed by coinciding the pseudo %fold axes of both molecules. The pseudo 2-fold axes are indicated by + when on the plane of the page and by 0 when perpendicular to it. From Kim (115a).

C. Interaction among tRNA, mRNA and the Ribosome Interaction of tRNA with mRNA and the ribosome is vastly more complicated than that between tRNAs and aminoacyl-synthetases, so much so that the 3-D structure of tRNA can provide only very limited information, as listed below. a. The 3-D structure of tRNA suggests that the conformations of tRNA and aminoacyl-tRNA are the same except possibly at the singlestranded 3‘ end. The distance between the anticodon and the 3’ end can be between about 46 A and 80 A (see Fig. 8 and Table IV) in yeast tRNAPtlein the orthorhombic crystal (ZO), depending on the orientation of the 3’ terminal A-C-C-A sequence, It is likely that tRNAs experience some conformational change on binding to ribosomes. However, examination of the 3-D structure of tRNA suggests that the distance cannot be very far from the above values. Therefore, the codon-anticodon recognition site and the peptide formation site are far apart (at least 40 A ) in the ribosome. b. The 3-D structure shows that the AC loop as well as the AC stem are free from contacts with other parts of the molecule (see Fig. S), suggesting that the AC arm may be flexible. The anticodon backbone

212

SUNG-HOU KIM

itself is in a pseudohelical rather than a helical conformation, the latter being more suitable to form a double helix of three base-pairs in the codon-anticodon complex. Thus the 3-D structure of tRNA suggests that the AC loop and possibly the AC stem go through conformational changes during the translational process in the ribosome, probably induced by mRNA and ribosomal components. c. All tRNAs except the mammalian initiator tRNAs contain G-T$-C-G (or -A) sequence in the T loop. Four out of five bases in this sequence form either secondary or tertiary hydrogen bonds and/or undergo base-stacking with bases from mostly the D loop in the 3-D structure (see Figs. 4 and 8). This extensive tertiary interaction of the T loop is in apparent contradiction with the observations that T-G-C-G inhibits the binding of Phe-tRNA to both P and A sites on the ribosome ( 116), implying the interaction between T-$-C-G and the ribosome, and that T-G-C-Ginteracts with the 5 S rRNA-ribosomal protein complex (117).In fact, three known prokaryotic 5 S RNA sequences ( 1 5 ) contain the C-G-A-A-C sequence, which is complementary to the common G-T$-C-G sequence of tRNAs, in virtually identical positions, i.e., starting at about the 43rd nucleotide from the 5' end. These experimental results can be explained only if one hypothesizes that the T l o o p D loop interaction found in the crystal structure of tRNA is broken to form a new T l o o p 5 S RNA interaction in the ribosome. The G-T-+-C-G sequence is located at the tip of the elbow of the "L" and is one of the easily accessible parts of the structure.

VII. Concluding Remarks The crystallographic method is the most powerful technique now available for obtaining the detailed 3-D structure of a molecule, especially when the molecule can be crystallized. The 3-D structures of yeast tRNAPhein two crystal forms (20, 21, 25) determined by this method are almost identical and consistent with most of the other physical and chemical experimental results. Thus, there is strong evidence that the structure in the crystal is almost identical to that of the free tRNA in solution, but there are many indications suggesting that the molecule changes its conformation to an unknown extent when complexing with other macromolecules. The crystal structure also explains in most cases the reason for the invariant, semi-invariant and coordinated invariant pairs and triples present in all tRNA. The 3-D structure determined by the X-ray crystallographic method is necessarily static, and no dynamic aspects of the molecule can be drawn directly. However, it is safe to assume that the crystal structure

THREE-DIMENSIONAL STRUCTURE OF TRANSFER RNA

213

of tRNA is in one of the most energetically stable conformational states if not the most stable state for free tRNA under the solution condition from which the crystals are formed. Therefore, the crystal structure of tRNA should serve as a starting point in designing further experiments to answer the questions regarding the various functional roles of tRNA (described in Section I ) regardless of whether the roles involve dynamic or static aspects of tRNA.

ACKNOWLEDGMENTS I wish to thank Paul Sigler and James Ofengand for sending me their manuscripts before publication, Paul Ts’o and Lou S. Kan for the discussion of their NMR work, and Marilyn Warrant for the preparation of the manuscript. The research from the author’s laboratory referred to in this review was supported by grants from the U.S. Public Health Service (CA-15802) and the National Science Foundation (GB-40814).

REFERENCES 1. 2. 3. 4.

J. D. Smith, Brit. Med. Bull. 29,220 (1973). K. P. Schafer and D. Sol], Biochimie 56, 795 (1974). L. L. Kisselev and 0. 0. Favorova, Aduan. Enzymol. 40, 141 ( 1974). D. Sol1 and P. Schimmel, Enzymes 10, 489 (1974). 5. U. Z. Littauer and H. Inouye, ARB 42,439 (1973). 6. F. Cramer, This Series 11, 391 (1971). 7. R. Haselkorn and L. B. Rothman-Denes, ARB 42,397 (1973). 8. T. S . Stewart, R. J. Roberts and J. L. Strominger, Nature (London) 230, 36 (1971). 9. R. L. Soffer, Aduan. Enzymol. 40,91 ( 1974). 10. R. Goldberger, Science 183, 810 (1974). 1 1 . J. E. Dahlberg, R. C. Sawyer, J. M. Taylor, A. J. Faras, W. E. Levinson, H. M. Goodman and J. M. Bishop, I. Virol. 13,1126 (1974). Ila. A. J. Faras, J. E. Dahlberg, R. C. Sawyer, F. Harada, J. M. Taylor, W. E. Levinson, J. M. Bishop and H. M. Goodman, J . Virol. 13, 1134 (1974). 12. P. Yot, M. Pinck, A. Haenni, H. Duranton and F. Chapeville, PNAS 67, 1345 ( 1970). 13. S. Altman and J. D. Smith, Nature N B 233,35 (1971). 14. W. H. McClain, C. Guthrie and B. G. Barrell, PNAS 69, 3703 (1972). 15. B. G. Barrell and B. F. C. Clark, “Handbook of Nucleic Acid Sequences.” Joynson-Bruvvers Ltd., Oxford, England, 1974. 16. R. W. Holley, J. Apgar, G. A. Everett, J. T. Madison, M. Marquisee, S. H. Merrill, J. R. Penswick and A. Zamir, Science 147, 1462 (1965). 17. C. R. Merrill, Biopolymers 6, 1727 ( 1968). 18. J. R. Penswick, R. Martin and G. Dirheimer, FEBS Lett. 50,28 (1975). 19. F. Sanger, G. G. Brownlee and B. G. Barrell, J M B 13, 373 (1965). 20. S. H. Kim, F. L. Suddath, G. J. Quigley, A. McPherson, J. L. Sussman, A. Wang, N. C. Seeman and A. Rich, Science 185,435 ( 1974). 21. J. D. Robertus, J. E. Ladner, J. T. Finch, D. Rhodes, R. D. Brown, B. F. C. Clark and A. Klug, Nature (London) 250,546 ( 1974).

214

SUNG-HOU KIM

22. S . H. Kim, J. L. Sussman, F. L. Suddath, G. J. Quigley, A. McPherson, A. Wang, N. C. Seeman and A. Rich, PNAS 71,4970 ( 1974). 23. M. Levitt, Nature (London) 224,759 ( 1969). 23a. S. Nishimura, This Series 12, 49 ( 1972). 24. D. Eisenberg, in “The Enzymes” (P. Boyer, ed.), 3rd ed., Vol. 1, p. 1. Academic Press, New York, 1970; K. Holms and D. Blow, in “Methods of Biochemical

Analysis” (D. Glick, ed.), Vol. 13, p. 113. Wiley, New York, 1965. 25. S. H. Kim, G. J. Quigley, F. L. Suddath, A. McPherson, D. Sneden, J. J. Kim, J. Weinzierl and A. Rich, Science 179,285 ( 1973). 26. F. L. Suddath, G. J. Quigley, A. McPherson, D. Sneden, J. J. Kim, S. H. Kim and A. Rich, Nature (London) 248,20 ( 1974). 27. J. L. Sussman and S. H. Kim, in “Environmental Effects on Molecular Structure

and Properties” (E. Bergmann and B. Pullman, eds.), 8th Jerusalem Symp. D. Reidel Publ., Dordrecht, Holland, 1975. 28. B. F. C. Clark, B. P. Doctor, K. C. Holmes, A. Klug, K. A. Marcker, S. J. Morris and H. H. Paradies, Nature (London) 219, 1222 (1968). 29. F. Cramer, F. van der Haar, W. Saenger and E. Schlimme, Angew. Chem., Int. Ed. Engl. 7, 895 ( 1968). 30. S. H. Kim and A. Rich, Science 162, 1381 ( 1968). 31. A. Hampel, M. Labanauskas, P. G. Conners, *L. Kirkegard, U. RajBhandary, P. Sigler and R. M. Bock, Science 162, 1384 ( 1968). 32. J. R. Fresco, R. D. Blake and R. Langridge, Nature (London) 220,1285 ( 1968). 33. A. Hampel and R. Bock, Bchem 9, 1873 (1970). 34. S. H. Kim, G. Quigley, F. L. Suddath and A. Rich, PNAS 68, 841 (1971). 35. T. Ichikawa and M. Sundaralingam, Nature (London) 236, 174 (1972). 36. J. E. Ladner, J. T. Finch, A. Klug and B. F. C. Clark, J M B 72,99 ( 1972). 37. R. W. Schevitz, M. A. Navia, D. A. Bantz, G. Cornick, J. J. Rosa, M. D. Rosa and P. B. Sigler, Science 177, 429 (1972). 38. G. J. Quigley, F. L. Suddath, A. McPherson, J. J. Kim, D. Sneden and A. Rich, PNAS 71,2146 (1974). 39. A. Klug, J. D. Robertus, J. E. Ladner, R. S. Brown and J. T. Finch, PNAS 71, 3711 (1974). 40. P. Sigler, Annu. Reo. Biophys. Bioeng. 4, 477 (1975). 41. S. Amott, D. W. L. Hukins and S. D. Dover, BBRC 48, 1392 (1972). 42. U. L. RajBhandary and S. H. Chang, JBC 243,598 (1968). 43. A. E. V. Haschmeyer and H. M. Sobell, PNAS 50, 872 (1963). 44. J. Broomhead, Acta Crystallogr. 1, 324 (1948). 44a. J. L. Sussman, N. C. Seeman, S. H. Kim and H. Berman, J M B 66, 403 ( 1972). 45. A. Rich, D. R. Davies, F. H. C. Crick and J. D. Watson, JMB 3,71 (1961). 46. S. H. Kim and A. Rich, PNAS 60,402 (1968). 47. A. Klug, J. Ladner and J. D. Robertus, JMB 89,511 ( 1974). 48. T. Brennan, J. Rubin, D. Stout, H. Mizuno, R. K. McMullan, T. Ichikawa, S. T. Rao and M. Sundaralingam, in “Structure and Conformation of Nucleic Acids and Protein-Nucleic Acid Interaction” (S. T. Rao and M. Sundaralingam, eds.), p. 25. Univ. Park Press, Baltimore, Maryland, 1975. 49. J. A. Lake and W. W. Beeman, J M B 31, 115 (1968). 50. W. R. Krigbaum and R. W. Godwin, Science 154,423 (1966). 51. P. G. Connors, M. Labanauskas and W. W. Beeman, Science 166, 1528 (1969). 52. J. Ninio, A. Favre and M. Yaniv, Nature (London) 223, 1333 (1969). 53. I. Pilz and 0. Kratky, EJB 15,401 (1970).

THREE-DIMENSIONAL STRUCTURE OF TRANSFER RNA

215

53a. L. Stryer, Science 162, 56 (1968).

K. Beardsley and C. R. Cantor, PNAS 65, 39 (1970). C. Yang and D. Soll, PNAS 71, 2838 (1974). 0. C. Uhlenbeck, J. Baller and P. Doty, Nature (London) 225, 508 ( 1970). G. Hogenauer, E J B 12,527 ( 1970). 0. C. Uhlenbeck, J M B 65,25 (1972). 0. Pongs and E. Reinwald, EJB 32, 117 (1973). 0. C. Uhlenbeck, J. G. Chirikjian and J. R. Fresco, J M B 89, 495 (1974). S. Freier and I. Tinoco, Jr., Bchem 14, 3310 ( 1975). D. R. Kearns, D. J. Patel and R. G. Shulman, Nature 229, 338 (1971). D. L. Lightfoot, K. L. Wong, D. R. Kearns, B. R. Reid and R. G. Shulman, J M B 78, 71 (1973). 65. R. G. Shulman, C. W. Hilbers, D. R. Kearns, B. R. Reid and Y. P. Wong, J M B 78, 57 ( 1973). 66. G. Giessner-Prethe and B. Pullman, J . Theoret. Biol. 27, 87 (1970). 67. Y. P. Wong, D. R. Kearns, B. R. Reid and R. G. Shulman, J M B 72, 725 (1972). 68. B. R. Reid, N. S. Ribeiro, G. Could, G. Robillard, C. W. Hilbers and R. G. Shulman, PNAS 72, 2049 (1975). 69. Y. P. Wong, D. R. Kearns, R. G. Shulman, T. Yamane, J. C. Chirikjian and J. R. Fresco, J M B 74, 403 ( 1973). 70. R. G. Shulman, C. W. Hilbers, Y. P. Wong, K. L. Wong, D. R. Lightfoot, B. R. Reid and D. R. Kearns, PNAS 70,2042 ( 1973). 71. D. M. Crothers, P. E. Cole, C. W. Hilbers and R. G. Shulrnan, J M B 87, 83 (1974). 72. L. S. Kan, P. 0. P. Ts’o, F. van der Haar, M. Sprinzle and F. Cramer, BBRC 59, 22 ( 1974). 73. M. Yaniv, A. Favre and B. G. Barrell, Nature (London) 223, 1331 (1969). 74. D. S. Card, G. Thomas and A. Favre, Biochimie 56, 1089 (1974). 75. M. Krauskopf, C.-M. Chen and J. Ofengand, JBC 247, 842 (1972). 76. D. E. Bergstrom and N. J. Leonard, Bchem 11, 1 (1972). 77. D. M. Brown, in “Basic Principles in Nucleic Acid Chemistry” (P. 0. P. Ts’o, ed. ), Vol. 11, p. 2. Academic Press, New York, 1975. 78. A. R. Cashmore, Nature NB 230, 236 (1971). 78a. Z. Kucan, K. A. Frende, I. Kucan and R. W. Chambers, Nature NB 232, 177 (1971). 78b. R. P. Singhal, JBC 246, 5848 ( 1971). 79. S. E. Chang, A. R. Cashmore and D. M. Brown, J M B 68, 455 (1972). 80. S. E. Chang, J M B 75,533 (1973). 81. S. Chang and D. Ish-Horowicz, J M B 84, 375 (1974). 82. D. Rhodes, J M B 94,449 ( 1975). 83. P. Piper and B. F. C. Clark, Nucleic Acid Res. 1, 45 (1974). 84. J. D. Robertus, J. Ladner, J. Finch, D. Rhodes, R. D. Brown, B. F. C. Clark, and A. Klug, Nucleic Acid Res. 1, 927 (1974). 85. R. Gamble and P. R. Schimmel, PNAS 71, 1356 (1974). 86. H. Prinz, A. Maelicke and F. Cramer, Bchem 13, 1322 ( 1973). 87. K. Nakanishi, N. Furutachi, M. Funamizu, D. Grunberger and I. B. Weinstein, JACS 92, 7617 (1970). 88. R. Langlois, S. H. Kim and C. R. Cantor, Bchem 14,2554 (1975). 89. S. H. Kim, G. Quigley, F. L. Suddath, A. McPherson, D. Sneden, J. J. Kim, J. Weinzierl and A. Rich, J M B 75, 429 (1973). 54. 55. 56. 57. 58. 59. 60. 61. 62. 64.

216

SUNG-HOW KIM

90. T. I. Jilyaeva and L. L. Kisselev, FEBS Lett. 10,229 ( 1970). 91. L. L. Kisselev, L. Frolova and M. Alexandrova, Mol. Bid. 1, 123 (1967). 92. J. P. Goddard and L. H. Schulman, JBC 247,3864 ( 1972). 93. L. H. Schulman and J. P. Goddard, JBC 248,1341 (1973). 94. A. D. Minabekov, D. Lastity, E. S. Levina and A. A. Bayev, Nature N B 229, 21 (1971). 95. N. Ianura, G. B. Weiss and R. W. Chambers, Nature (London) 222, 1147 (1969). 96. R. Thiebe, K. Harbers and H. G. Zachau, E J B 26, 144 ( 1972). 97. S. Hashinoto, M. Kawata and S. Takemura, J. Biochem. (Tokyo) 72, 1339 (1972). 98. J. Carbon and J. B. Curry, J M B 38,204 ( 1968). 99. C. Squires and J. Carbon, Nature N B 233,274 (1971). 100. M. Yaniv, W. R. Folk and P. Berg, J M B 86,245 (1974). 101. A. Ghysen and J. E. Cellis, J M B 83,333 ( 1974). 102. B. Roe, M. Sirover and B. Dudock, Bchem 12,4146 (1973). 103. D. M. Crothers, T. Seno and D. Soll, PNAS 69,3063 ( 1972). 104. D. Kern, R. Giegb and J.-P. Ebel, EJB 31, 148 (1972). 105. R. GiegB, D. Kern and J.-P. Ebel, Biochimie 54, 1245 (1972). 105a. R. Giegb, D. Kern, J.-P. Ebel, H. Grosjean, S. DeHenau and H. Chantrenne, EJB 45,351 (1974). 106. H. J. P. Schoemaker and P. R. Schimmel, J M B 84,503 ( 1974). 107. G. P. Budzik, S. S. M. Lam, H. J. P. Schoemaker and P. R. Schimmel, JBC 250, 4433 ( 1975). 108. H. J. P. Schoemaker, G. P. Budzik, R. Giegb and P. R. Schimmel, JBC 250, 4440 ( 1975). 109. L. H. Schulman, J M B 58, 117 ( 1971). 110. K. W. Anderson and J. D. Smith, J M B 69,349 (1972). 111. M.-R. Kula, FEBS Lett. 35, 299 (1973). 112. G. L. E. Koch, Y. Boulanger and B. S. Hartley, Nature (London) 249, 316 (1974). 1 1 2 ~ C. . J. Bruton, R. Jakes and G. L. E. Koch, FEBS Lett. 45,26 (1974). 113. R. M. Waterson and W. H. Konigsberg, PNAS 71,376 (1974). 114. R. J. Poljak, L. M. Arnzel, H. P. Avey, B. L. Chen, R. P. Phizacherley and F. Saul, PNAS 70,3305 ( 1973). 115. D. R. Davies, E. A. Padlan and D. M. Segal, ARB 44, 639 (1975). 115a. S. H. Kim, Nature (London) 258,679 ( 1975). 116. J. Ofengand and C. Henes, JBC 244,6241 (1969). 117. V. A. Erdman, M. Sprinzl and 0.Pongs, BBRC 54,942 (1973). Note Added in Proof The tertiary base-pairs described on p. 193 ( 6 to 9 ) are also found in the monoclinic crystal form (Ladner et al. PNAS 72, 4414, 1975). Their revised interpretation shows that G26 is not intercalated between A44 and G45 (see p. 194), but base-paired to A44 as seen in the orthorhombic form ( 2 0 ) . The three-dimensional structure of yeast tRNAPhcin both crystal forms is now practically identical.

Insights into Protein Biosynthesis and Ribosome Function through Inhibitors

I

SIDNEYPESTKA Roche lnstitute of Molecular Biology Nutley, New Jersey

.

-

I. Introduction * . . . * * 11. Summary of Protein Biosynthesis . . . . . . . . 111. Translocation Inhibitors: Localization of Action through Nonenzymic and Enzymic Translocation IV. Ribosomal States . . . . . . . . . . . V. Erythromycin Binding to Ribosomes . . . . . . . VI. Effect of Chloramphenicol on the Puromycin Reaction: Models of Ribosome Function . . . . . . . . . . . VII. The Sistrand: The Translational Unit . . . . . . . References. . . . . . . . . . . . .

. . . . . . . .

217 217 223 223 230 234 238 244

1. Introduction This account is a highly personal view of some selected experiences with antibiotic inhibitors of protein synthesis. It is not intended to be a comprehensive review of the literature or of my own work. It is intended to outline some interesting, perhaps even controversial, observations and interpretations that have emanated from studies of inhibitors of protein biosynthesis. In general, detailed knowledge of protein biosynthesis has enabled localization of the action of many antibiotics and other inhibitors. Nevertheless, examination of the effects of various inhibitors on steps of protein biosynthesis has uncovered intriguing and valuable insights into protein synthesis as well as into ribosome structure and function.

II. Summary of Protein Biosynthesis For reference, the steps of protein biosynthesis are summarized in Fig. 1. The details of these interactions are summarized in the legend to Fig. 1. For further details, a number of recent reviews may be consulted (I, 2). Inhibitors of protein synthesis can be classified in a number of ways (35). According to the classification presented in Table I, inhibitors are divided into groups according to their site of inhibition: supernatant factors, small subunit (30 S or 40 S ) or large subunit (50 S or 60 217

INITIATION

RECOGNITION OF INTERNAL CODONS

/.-" .

.

*-=

E

PEPTIDE BOND FORMATION. TRANSLOCATION

P

-

TERMIN"

am a o

n

P

INHIBITORS OF PROTEIN SYNTHESIS

219

S ) . Then they are classified according to their ability to inhibit these functional sites in prokaryotes only, in eukaryotes only, or in both proand eukaryotes. The classification is simplified in that inhibition of mitochondrial or chloroplast ribosomes is not included as a distinct entity. In general, although most of the prokaryotic inhibitors block protein synthesis in bacteria, blue-green algae, mitochondria, and chloroplasts, some show some specificity within this group. For example, erythromycin and lincomycin inhibit protein synthesis by bacteria, but not by rat liver mitochondria ( 6 ) . Likewise, some inhibitors exhibit some discrimination among various eukaryotes ( 7 ) . The puromycin reaction (summarized in Fig. 2 ) has been extremely useful in studying transpeptidation and even other steps of protein biosynthesis. In the presence of puromycin, which serves as an analog of aminoacyl-tRNA, donor peptides or peptide analogs on peptidyl-tRNA can be transferred to puromycin in the presence of ribosomes. The peptidyltransferase that catalyzes this reaction is an integral part of the ribosome structure. Donors for the puromycin reaction may be peptidyltRNAs of many sorts. For example, native peptidyl-tRNA may be used on polyribosomes isolated from intact cells. Synthetic peptidyl-tRNAs, such as polyphenylalanyl-tRNA, polylysyl-tRNA, N-acetylphenylalanyltRNA, or formylmethionyl-tRNA, may serve as suitable peptidyl donors FIG.1. Schematic summary of protein synthesis. The semilunar cap ( I ) represents the free 30 S subunit. Initiation of protein synthesis involves attachment of mRNA to a 30 S subunit ( I ) to form complex 11; this process requires Mg” as well as initiation factor IF-3. Subsequent attachment of fMet-tRNA, in response to initiation codon AUG, to form complex I11 requires GTP and initiation factors IF-I and IF-2. Junction of the 50 S subunit to complex 111 produces complex IV. Enzymic recognition of internal codons involves factors EF-Tu, EF-Ts, and GTP: the (EFTu) GTP * ( Ala-tRNA) complex binds to the ribosomes in response to the GCU codon to form complex V. Peptide bond formation occurs by transfer of the fMet (peptidyl) group to form fMet-Ala ( V I ) ; peptidyl transfer requires only ribosomes and K . Translocation involves several coordinate processes, such as: release of deacylated tRNArMetto form state VII; one-codon movement of mRNA and ribosome with respect to each other, precisely positioning the next codon, UCU, into position for translation; and coordinate movement of peptidyl-tRNA ( fMet-Ala ) from the “A” to “P” site, resulting in state VIII. By repetition of the codon recognition step, Ser-tRNA would enter the A site in response to the codon UCU. Complex IX represents a peptidyl-tRNA with a polypeptide almost completed. Transpeptidation and translocation produces complex X with a completed protein still attached to tRNA and a termination codon, UAA, in the next recognition site. In response to a release factor, the completed protein is released, and perhaps also tRNA after translocation (XI). Provided no further sistrands (see Fig. 22) are to be translated, the ribosome may be dissociated into 30 S and 50 S subunits with release of mRNA (XII). Alternatively, mRNA may be degraded prior to this stage. See reviews (1, 2 ) for further details.

220

SIDNEY PESTKA

TABLE I CLASSIFICATION OF INHIBITORS OF PROTEIN SYNTHESIS Supernatant Folk acid antagonists

Aminoalkyl adenylates Guanosine B’-(B,y-methylene)triphosphate Fusidic acid

Diphtheria toxin

30 S (40 S)

Prokaryotic inhibitors Aminoglycosides Streptomycin Dihydrostreptomycin Paromom ycin Neom ycin Kanam ycin Gentamycin Bluensomycin y m y c i n asugamycin Negam ycin Edeine Edeine A Edeine B

50 S (60 S)

Chloramphenicol Macrolides Niddamycin Carbomycin Spiramycin I11 Tylosin Leucomycin Erythromycin Chalcomycin Oleandomycin Lankam ycin Meth ym ycin Lincomycin Streptogramin A group Ostreogrycin A PA114A Streptogramin A Vernamycin A Mikamycin A Streptogramin B group Ostreogrycin B PA114B, etc. Viridogrisein (etamycin) Thiostrepton group Thiostrepton (bryamycin, thiactin) Siomycin A Thiopeptin B Multhiomycin Sporangiomycin A-59 Althiomycin Micrococcin Bottromycin A2

Both Prokaryotic and Eukaryotic Pactamycin Aurintricarboxylic acid

Eukaryotic Inhibitors Harringtonine alkaloids

Purom ycin 4-Aminohexose pyrimidine nucleosides Gougerotin Amicetin Blaaticidin S Plicacetin Bamicetin Sparsomycin Tetracyclines Chlortetracycline Glutarimides Cycloheximide (Actidione) Acetoxyc ycloheximide Streptovitacin A Ipecac alkaloids Emetine Anisomycin Trichodermin

221

INHIBITORS OF PROTEIN SYNTHESIS PEPTIDE

f!3

PUROMYCIN

PEPTIDYLTRANSFERASE

PEPTIDYLPUROMYCIN

FIG.2. Schematic illustration of the puromycin reaction with peptidyl-tRNA on ribosomes or polyribosomes.

( 3 ) . Furthermore, the oligonucleotides that comprise the ends of these aminoacyl-tRNAs also may be suitable donors. For example, C-AC-C-A ( AcPhe ) or C-A-A-C-C-A ( f Met ) , the aminoacyl-oligonucleotide termini of AcPhe-tRNA and fMet-tRNA, respectively, can serve as suitable donors (8). Similarly, acceptors of the puromycin reaction may be any aminoacyl-tRNA or a number of aminoacyl-tRNA analogs as shown in Fig. 3. Puromycin and the puromycin analogs are the smallest acceptors described. Analogs of aminoacyl-tRNA, such as C-A-C-C-A(Phe ) and similar compounds, also serve as suitable acceptors. Many puromycin analogs that shed light on the structural requirements of the acceptor have been made. For example, hydroxypuromycin (Fig. 3 ) can serve as an acceptor, which indicates that transfer can be made to a hydroxyl as well as an amino group (9, 10).In addition, two unique carbocyclic analogs of puromycin have been synthesized (Fig. 3 ) and their activity in various assays has been determined (10). Studies with these analogs have indicated that neither the sugar moiety CHs,

NH

I

OH

c=o I

PUROMYCIN

NH

I

OH

c=o I

JI-HYDROXY-

NH

I

c=o I

OH

HO

C , H3 N

NH

I

c=o I

CARBOCYCLIC P U R O M Y C I N A N A L O G S

P U R O M YC I N

FIG.3. Structures of puromycin, +hydroxypuromycin ( methoxy group is absent), and carbocyclic puromydn analogs I and I1 ( stereoisomers, with hydroxymethyl group absent).

222

SIDNEY PESTKA

nor the 5-hydroxymethyl group is necessary for puromycin to serve as a suitable acceptor. Furthermore, carbocyclic-puromycin analog I was 100-fold more active than analog 11. Further studies on these carbocyclicpuromycin analogs should be enlightening. An example of the effect of carbocyclic analog I on peptidyl puromycin synthesis with polyribosomes is shown in Fig. 4. It can be seen that the carbocyclic puromycin analog inhibits peptidyl-puromycin synthesis with a K , of 10 pM. As expected, the inhibition is completely competitive with puromycin, which under these same conditions has a K, of 4 pM. The similar reaction with hydroxypuromycin provides a K i of 200 pM for hydroxypuromycin. It is also noteworthy that puromycin has about 100-fold greater affinity for polyribosomes than for 1 M NH,C1-washed ribosomes as judged by the K, for the reactions (11). 10 9

8

7

I 4

6

5 4 3

2 I

0

2

4

6

0

1

0

[PUROMYCIN]-'X~O-~

[CARBOCYCLIC PUROMYCIN]xlOS

FIG.4. Competitive inhibition of peptidylpuromycin synthesis by Esckrichio cob polyribosomes with carbocyclic analog I of puromycin. Each 50-al reaction mixture contained 50 mM Tris-Ac ( p H 7.2), 100 mM KCI, 4 mM MgCl, and 2 Amounits of E. c d i polyribosomes ( 1 0 ) . [aH]Puromycin and the carbocyclic analog of puromycin were present at the concentrations indicated in the figure. Reaction mixtures were incubated at 24°C for 1 minute and assayed as described (10). In the left panel, a double-reciprocal plot of the data is presented. For each curve of the left panel, the concentration of the carbocyclic analog of puromycin used is given. The KI of 8-11 pM was calculated with the use of a K, of 4.45 pM for puromycin under these conditions represented by the data in the absence of the carbocyclic analog. In the right panel, a Dixon plot of the data is presented. The Ki in this case was determined to be 11 pM. The various amounts of ['H]puromycin used in the reactions are shown in the figure. The interaction coefficients obtained for carbocyclic analog I of puromycin range from 0.8 to 1.0; these were obtained from Hill plots of the data.

INHIBITORS OF PROTEIN SYNTHESIS

223

111. Translocation Inhibitors: Localization of Action through Nonenzymic and Enzymic Translocation Although the mechanism of nonenzymic translocation has not yet been delineated (12,13),studies on the effect of antibiotics on enzymic and nonenzymic translocation have proved to be simple and powerful techniques for localizing antibiotic action ( 12). That nonenzymic translocation is, in fact, distinct from enzymic translocation has been demonstrated (12,13). The essential difference between enzymic and nonenzymic translocation is the need for elongation factor EF-G for enzymic translocation, but not for the nonenzymic reaction. This was demonstrated in several ways. When proteins L7 and L12 were removed from ribosomes, enzymic translocation was inhibited, but nonenzymic translocation was not affected. Also, thiostrepton inhibited nonenzymic translocation from ribosomes devoid of L7 and L12 (12).Furthermore, p-chloromercuribenzenesulfonate, which inhibits elongation factor EF-G and thus is a powerful inhibitor of enzymic translocation, had no effect on or stimulated nonenzymic translocation (12,13). I n addition, guanosine 5'- ( p,y-methylene ) triphosphate, a competitive inhibitor of GTP, inhibited enzymic, but not nonenzymic translocation. Furthermore, when ribosomes and EF-G were prepared from an E . coli strain containing a temperature-sensitive EF-G, enzymic translocation was temperature sensitive, but nonenzymic translocation was not ( 12). Nevertheless, although the mechanism of nonenzymic translocation is not yet understood, it can be and has been used successfully to localize antibiotic action. For example, any agent that inhibits both enzymic and nonenzymic translocation must be a ribosomal inhibitor, for nonenzymic translocation requires only ribosomes, but no additional supernatant factors. In this way, thiostrepton, siomycin A and micrococcin were determined to be ribosomal inhibitors (12, 14, 15). Any agent that inhibits the enzymic, but not the nonenzymic process, must therefore be an inhibitor of elongation factor EF-G. In this way, fusidic acid was demonstrated to be an inhibitor of EF-G (16).These results are summarized in Table 11. Thus, by simple and direct comparison of the effect of an antibiotic on enzymic and nonenzymic translocation, one can localize the action of an antibiotic to either EF-G or to the ribosome. Mixing experiments with EF-G and ribosomes from antibiotic-sensitive and -resistant strains are therefore unnecessary. For more detailed studies of these inhibitors, recent reviews can be consulted (17,18).

IV. Ribosomal States During protein synthesis, ribosomes exist in a number of discrete functional states, each state containing the attached components differ-

224

SIDNEY PESTKA

TABLE I1

CLASSES OF INHIBITORS OF TRANSLOCATION‘

Antibiotic Fusidic acid Thiostrepton, siomycin A Micrococcin

Cell component

RiboNonsome: Enzymic enzymic mRNA GTP TransAAtranshypepti- tRNA trans- movelocation location ment drolysis dation binding

EF-G

+b

-

Ribosome Ribosome

+ +

+ +

+ + +

+ + k

-

-

-

+ +

a Although these substances inhibit various translocation reactions, thiostrepton, siomycin A and, perhaps, micrococcin can also inhibit aminoacyl-tRNA (AA-tRNA) binding, which may, in fact,, be the predominant effect on intact cells. Also, micrococcin may have some effect on GTP binding to ribosomes. b Inhibition; -, no effect; f,pmsible effect.

+,

Ia

EFO,

FIG.5. Ribosomal states and the ribosome cycle. The individual complexes I-VIII are described in the legend to Fig. 1. The initiation sequence comprises stages I-IV. For every addition of a single amino acid (i.e., elongation), the ribosome cycles from state V to VIII. State IV is formally equivalent to VIII. Thus, the elongation epicycle (V-VIII) is superimposed on the overall ribosome cycle and repeats itself on each amino-acid addition. The termination sequence is not given in detail, as it has not been precisely delineated. Dissociation factor ( D F ) bound to 30 S subunits ( l a ) may be identical to one of the initiation factors and may be involved in dissociation of 70 S ribosomes after termination. The symbol “f” represents fMet.

INHIBITORS OF PROTEIN SYNTHESIS

225

ently juxtaposed. It is very likely that each functional state corresponds to a conformational state as well. Ribosomes can exist as free subunits, as a 30 S subunit initiation complex and as various 70 S states depending on the location and presence of aminoacyl-tRNA, peptidyl-tRNA or deacylated-tRNA (states IV-VII) as illustrated in Fig. 5. During each elongation epicycle-that is, during the addition of a single amino acidthe ribosomes pass through states IV-VII. Each elongation epicycle takes place on the polyribosome. Although it is clear that the various ribosomal states exist, it was not clear until recently that antibiotics discriminate between various states. Inhibitors that discriminate between the various states have been designated topostatic agents (19). Topostatic agents are agents that inhibit or stimulate a reaction depending on the state of the components involved. For example, a number of antibiotics can inhibit peptide bond formation when AcPhe-tRNA is a donor. AS illustrated in Table I11 under the column “Synthetic Models for Peptide Bond Synthesis” and in Table IV, chloramphenicol, sparsomycin, the aminonucleosides ( amicetin, gougerotin and blasticidin S ), lincomycin, and macrolides (erythromycin, niddamycin, carbomycin, tylosin and spiramycin 111) as well as streptogramin A antibiotics (PA114A and vernamycin A ) were excellent inhibitors of synthetic models for peptide synthesis (3, 15, 20). In these assays, washed ribosomes and the synthetic donor AcPhe-tRNA was used. Althiomycin did not inhibit this reaction under the conditions studied ( 1 5 ) , but could inhibit acetylphenylalanylpuromycin synthesis when conditions were varied (21, 22). In striking contrast were the results when native polyribosomes were used for the assay of peptidylpuromycin synthesis (Tables I11 and IV). Chloramphenicol, sparsomycin, and the aminonucleosides were good inhibitors of peptidylpuromycin synthesis (23). They were good inhibitors of both the synthetic as well as the native reaction. However, lincomycin and the macrolides as well as the streptogramin A antibiotics were ineffective in inhibiting peptidylpuromycin synthesis. Furthermore, althiomycin which did not inhibit acetylphenylalanylpuromycin synthesis, was a strong inhibitor of peptidylpuromycin formation with native polyribosomes. Since transpeptidation was measured in both assays, the contrasting effects observed represent different interactions of the antibiotics with the discrete ribosomal states present in the two distinct assays. The reaction of puromycin with native polyribosomes is illustrated in Fig. 6. It can be seen that puromycin interacts with nascent peptides of 3 pM. As expected, one puroon polyribosomes with an apparent K:,, mycin molecule is involved in the release of each nascent peptide. Study of the kinetics of this reaction has been useful in understanding transpeptidation on polyribosomes as well as the interactions of antibiotics with ribosomes in various states.

226

SIDNEY PESTKA

TABLE I11 EFFECT OF ANTIBIOTICS O N TRANSPEPTIDITION ASSAY

Antibiotic Chloramphenicol Lincomycin Carbomycin Erythromycin Niddamycin Oleandomycin Spiramycin I11 Tylosin Streptogramin A PA114A Vernamycin A A1thiomycin Sparsom ycin Amicetin Gouger0 tin Blasticidin S Anisomycin Tenuazonic acid Trichodermin Cycloheximide

Synthetic models for peptide bond synthesis"

Native peptide bond synthesisb

+d

+

+ + + + + + + + + + k + + + + + + +?

-

+ + + + + + + + +

Species inhibitedc

P P P P P P P P P P P P B B B B E E E E

a Refers to assays where NH4C1-washed ribosomes are used as well as synthetic peptidyl donors, such as AcPhe-tltNA, fMet-tItNA, polylysyl-tItNA, C-A-C-C-A(AcPhe), C-A-A-C-C-A(fMet) and others, to form peptidylpuromycin. * Refers to the synthesis of peptidylpuromycin with native peptidyl-tltNA on polyribosomcs isolated from cells. P = prokaryotic; E = eukaryotic; B = both. Inhibition; -, no effect; +, partial inhibition.

+,

In addition, the effect of antibiotics on chloramphenicol binding to 70 S ribosomes, and polyribosomes could also be used as a probe for their interaction. Since chloramphenicol inhibits both acetylphenylalanylpuromycin and peptidylpuromycin synthesis, it was expected that it would interact with both washed ribosomes and the ribosome monomers of polyribosomes. In fact, this has proved to be the case ( 2 4 ) . Those antibiotics that inhibit acetylphenylalanylpuromycin synthesis, but not peptidylpuromycin synthesis ( the niacrolides, lincosamides and 'streptogramin A antibiotics ), inhibit chloramphenicol binding to NH,Cl-washed ribosomes. However, they are essentially ineffective in inhibiting chloramphenicol binding to polyribosomes (Tables V and VI). In addition, streptogramin B antibiotics and the aminonucleosides, which inhibit

227

INHIBITORS OF PROTEIN SYNTHESIS

TABLE IV EFFECT O F AKTII~IIJTICS O N SYNTHESIS OF ACKVYLPHENYLALINYLIJUROMYCIN .\ND PEPTIDYLPUItOMYCINa

Antibiotic Chloramphenicol Lincomycin Carbomycin Erythromycin Oleandomy cin Methymycin Niddamycin Tylosin Spiramycin 111 Vernamycin A PA114A A1thiomycin Sparsomycin

Acetylphenylalanylpuromycin synthesis

Peptidylpuromycin synthesis

10-631 l0-5M l0-4hhZ lO-3M

10-oM l0-6h'I 1 0 V M l0-3M

99 98 83 104

88 88 18 103

51 60 17 134

108

23

23

1 1 94 4

1 2 114 2

100 103 102 101 104 44

103 96 112 97

105 98 91 100 92 86 95 100 98 108 96 3.5 16

64 102 114 102 103 88 103 105 99 103 110 14 1

42 92 89 10t5 96 83 94 -

98 101 1

a The data in the table are taken from Pestka (15, 80, 23) and are presented as percentage of control reactions in the absence of antibiotics.

LOG [ PUROMYCIN]

R

h XIO-5

FIG.6. Hill plot and double-reciprocal plot for puromycin participation in peptidylpuromycin synthesig. Each 50-r(.l reaction mixtures contained the components indicated in the legend to Fig. 4. Puromycin concentration was vaned, and reactions took place at 24°C for 1 minute. The interaction coefficient, n, was found to be 0.996 and K , was determined as 3.3 p h l for puromycin (log K,,, = 5.48, the intercept of the ordinate). When u = V,1,ny/2,log[o/( V,,,,, - u ) ] equals 0, and K., = 3.0 fiM (pK,,, = 5.52), as determined from the intercept of the abscissa of the Hill plot. The K,,, was determined to be 3.2 pM from the double-reciprocal plot.

chloramphenicol binding to ribosomes, are ineffective in inhibiting chloramphenicol binding to polyribosomes ( 24 ) . Furthermore, sparsomycin and althiomycin are ineffective in inhibiting chloramphenicol binding to washed ribosomes, but are strong inhibitors of chloramphenicol binding to polyribosomes (Fig. 7 ) . It can be seen that sparsomycin is a potent inhibitor of chloramphenicol binding to polyribosomes. At lo-' M,

228

SIDNEY PESTKA

TABLE V EFFECT OF ANTIBIOTICS O N CHLORAMPHENICOL BINDING TO 70 S RIBOSOMES A N D POLYRIBOSOMES Antibiotic

Ribosomes Polyribosomes

+" + + +

Macrolides Lincosamides Streptogramins A Streptogramins B Aminonucleosides Sparsomycin A1thiomycin

f f

-

+

-

-

+ +

+, Strong inhibition; -, no inhibition; f,partial inhibition. ~~

~

TABLE VI EFFECTOF ANTIBIOTICSO N BINDINQ OF CHLORAMPHENICOL TO RIBOSOMES A N D TO POLYRIBOSOMEW Binding to ribosomes Antibiotic Erythromycin Carbomycin Spiramycin I11 Niddamycin Oleandomycin Methymycin Tylosin Lincomycin Celesticetin Vcrnamycin A PA114A Vernamycin Ba PA114B Sparsomycin Althiomycin C hloramphenicol Blasticidin S Gougerotin Amicetin

10-6

M 10-6 M 10-4 M 10-3 M

94 85 93 96 107 95 94 104 108 91 118 110 107 98 101 -

32 71 14 Ti8 33 .59 12 96 99 16 16 102 33 100 96 60 96 106 105

11 15 10 11 14 23 15 65 89 12 13 32 23 92 100 18 73 84 108

Binding to polyribosomes 10-6

M 10-KM10-4 M 10-3 M

99 100 81 92 go 89 94 93 101 89 100 97 94 53 98 -

-

93 87 93 87 87 66 94 94 96 93 88 95 90 20 47 54 101 102 102

78 89 92 93 74 39 92 80 87 86 90 88 93 15 22 17 97 97 114

The data in the table are taken from Pest,ka ($4) and are presented as a percentage of control reactions in the absence of antibiotics.

229

INHIBITORS OF PROTEIN SYNTHESIS

o

1.1

I

"0-7

I 10-6

I

I 10-5

I

I

0

10-4

SPARSOMYCIN [MOLARITY]

II

1.1

I

10-3 0-10-6

10-5

10-4

ALTHIOMYCIN [MOLARITY]

FIG. 7. Effect of sparsomycin and althiomycin on chloramphenicol binding to ribosomes and polyribosomes. Reactions were performed as indicated in reference 24. Antibiotic concentration is given on the abscissas. The data are plotted as a percentage of the control value for the binding of chloramphenicol to ribosomes or polyribosomes in the absence of any antibiotic. Left: This cuntrol value was 36.5 and 35.1 pmol of chloramphenicol for ribosomes [6.4 absorbancy ( A m o ) units] and polyribosomes (4.2 Amo units), respectively. Right: Control value was 45.2 and 47.0 pmol of chloramphenicol for ribosomes (6.4A, units) and polyribosomes ( 5.6 A,lo units), respectively. Symbols: chloramphenicol binding to ribosomes (0) and to polyribosomes ( 0 ) .

sparsomycin inhibition of chloramphenicol binding to ribosomes was detectable and at approximately M, inhibition of chloramphenicol binding to polyribosomes was greater than 50%. Nevertheless, at M (1000- to 10,000-fold higher sparsomycin concentrations than necessary to inhibit chloramphenicol binding to polyribosomes ), sparsomycin had little or no effect on chloramphenicol binding to washed ribosomes. The results with althiomycin were similar. Althiomycin inhibited chloramphenicol binding to polyribosomes but had little or no effect on chloramphenicol binding to washed ribosomes. These antibiotics are clear probes of the ribosomal states involved and illustrate the precise and subtle insights into the ribosomal states that they can provide. Since erythromycin inhibits transpeptidation with washed ribosomes but not with polyribosomes (3, 23), it was anticipated that this might reflect the inability of erythromycin to interact with ribosomes in polyribosomes. As shown in Table VII, erythromycin binds to ribosomes but relatively little to polyribosomes. However, if peptidyl-tRNA is removed from polyribosomes by treatment with puromycin, erythromycin binds to the resultant ribosomes. It is thus clear that the presence of peptidyl-tRNA on polyribosomes inhibits erythromycin binding and pre-

230

SIDNEY PESTKA

TABLE VII BINDING OF ERYTHROMYCIN TO RIBOSOMES AND POLYRIBOSOMES BEFORE AFTER REACTION WITH PUROMYCIN'

AND

~~

Molecules of erythromycin bound/ribosome Polyribosomes or ribosomes Polyribosomes 70 S ribosomes (native) 70 S ribosomes (washed)

Plus Minus puromycin puromycin 0.17 0.82 0.75

Ratio plus: minus puromycin

0.80 0.97 0.69

4.7 1.2 0.93

,, Data taken from Pestka (84).

sumably its interaction with the ribosome (24, 24u). It should be noted that these studies were carried out with riiosomes in a static state. Ribosomes actively involved in the dynamic cycles and various states of protein synthesis, however, may, in fact, not react identically with these antibiotics. It is possible that the various conformational and translocational movements involved during elongation of the peptide chain may allow these antibiotics to interact with ribosomes on polyribosomes despite our results using a static state, that is, using polyribosomes not actively involved in protein synthesis. Nevertheless, whether or not this proves to be true for the dynamic ribosomes of protein synthesis, these agents serve as useful probes into the static ribosomal states involved.

V. Erythromycin Binding to Ribosomes Erythromycin binding to ribosomes was examined for several reasons. In the first place, valid binding constants for erythromycin binding to E . coli ribosomes has not been reported, Second, we hoped to use the erythromycin binding site as a focus for affinity labeling of ribosomal proteins and other components. Furthermore, it would be useful to correlate the effects of erythromycin derivatives on erythromycin binding to ribosomes with their antibacterial activities. Since the ribosome is the target of erythromycin action, the affinity of an erythromycin derivative for ribosomes should reflect its antibacterial activity. Any discrepancies from this correlation should, therefore, be due to alterations in other parameters, such as metabolic conversion or cellular permeability to the derivative.

231

INHIBITORS OF PROTEIN SYNTHESIS ERYTHROMYCIN MOLARITY x 10'

246810

0

PICOMOLES ERYTHROMYCINADDED TO REACTION MIXTURE

FIG.8. Binding of erythromycin to ribosomes as a function of erythromycin concentration. Each 0.50-ml reaction mixture for the filter assay, 24°C (left panel) contained 5.6 Az, units of ribosomes and other components as described. The initial concentration of ["Clerythromycin added to the reaction mixture is given on the abscissa. Dissociation ( K d = 1.0 x 10.' M ) and association (K. = 9.9 x 10' M-') constants were computed from the Scatchard plot (inset). Equilibrium dialysis, 5°C (right panel) was performed at 46 and 70 hours ( 2 5 ) . The values at 70 hours are plotted directly and as a Scatchard plot (inset). At 46 hours, the slope of the Scatchard plot was similar, but the baseline intercept was 0.91. At 70 hours, the baseline intercept was 0.86. The dissociation ( K d = 1.4 x lo-' M ) and association (K. = 7.2 x 10' M-') constants were calculated from the Scatchard plot. A computer program was used to determine the line of best fit by the method of least squares. The data for the filter assay arc presented as solid circles ( 0 ) and the data for equilibrium dialysis by open circles ( 0).

Erythromycin binding to ribosomes was examined by equilibrium dialysis as well as by the filter binding technique (Fig. 8). Th'is association constant was determined to be 7.2 x lo7 M-' and the dissociation constant 0.014 PM by equilibrium dialysis. Equilibrium dialysis was performed at 5°C. Association (K,= 9.9 x lo7 M-') and dissociation ( Kd = 0.01 PM) constants were also determined by the filter binding technique at 24°C (25). The binding of erythromycin to ribosomes is reversible (Fig. 9). The forward reaction is extremely rapid and essentially completed by the time components are mixed. The reverse reaction rate is measurable. Thus, from the equilibrium constant and the reverse reaction rate, the forward reaction rate was calculated to be 1.7 X lo7liters mol-' min-'. Equations relating the effect of erythromycin derivatives on erythro-

232

SIDNEY PESTKA

20L

I

-I

MINUTES

FIG.9. Reversibility of erythromycin binding to ribosomes. Each 0.50-ml reaction mixture contained 5.6 A, units of ribosomes and other components as described in reference 25. The time course of [“Clerythromycin binding to ribosomes was followed as a function of time at 24°C. At 15 minutes, 10 ~1of a 10 mM solution of unlabeled erythromycin was added to some of the tubes as indicated (arrow) to produce a ratio of 200/1 for unlabeled to %-labeled erythromycin in these tubes. 0, No unlabeled erythromycin; 0, 0.1 pmol of unlabeled erythromycin added to each of these tubes at 15 minutes.

&UATlONS

where:

TABLE VIII RELATINQ ERYTHROMYCIN BINDING TO RIBOSOMES AND (Nu) INHIBITOR CONCENTRATION

[El = free erythromycin concentration [I] = 10 = input unlabeled antibiotic concentration [Ro] = concentration of total ribosomes potentially active in binding erythromycin [R] = concentration of free ribosomes potentially active in binding erythromycin [RE] = concentration of ribosome * erythromycin complex [RI,] = concentration of ribosome . inhibitor complex

INHIBJTORS OF PROTEIN SYNTHESIS

233

mycin binding to ribosomes are described in Table VIII. Thus, from the effect of an erythromycin derivative on erythromycin binding to ribosomes, it was possible to calculate the dissociation constant of that derivative as well as its interaction coefficient (26). This was done for approximately 50 erythromycin derivatives with the results shown in Figs. 10A and 10B. Many analogs were practically inactive in inhibiting erythromycin binding to ribosomes, and others were almost as active as erythromycin itself. When these data were plotted according to the equations presented in Table VIII, straight lines were generally obtained from which the association constants as well as the interaction coefficients for each derivative could be derived (Fig. 11). Examples of some of the erythromycin derivatives synthesized and examined are shown in Figs. 12-14. It can be seen that various portions of the erythromycin molecule can be modified with no substantial loss of binding. Specifically, the cladinose can be removed and replaced by various groups with retention of activity, and various substitutions can be made on the oxime as well as on the sugar moieties. There is a general correlation between the ribosomal binding of a derivative and its antibacterial activity (27). Analogous studies with erythromycin have been performed by others (28, 29). Similar studies were performed with leucomycin derivatives ( 3 0 ) , which complete with erythromycin binding to ribosomes as shown in Fig. 15. Binding of each leucomycin derivative to ribosomes correlated exceeding well with its antibacterial activity ( Fig. 16). With the knowledge of which sites on the erythromycin molecule could be chemically altered, a number of erythromycin derivatives with groups for covalent attachment to ribosomes have been synthesized (31 ). These erythromycin affinity analogs should help localize the erythromycin binding site on the surface of the ribosome. A fluorescent derivative of erythromycin has been prepared ( 3 1 ) for use in probing the erythromycin binding site and in fluorescence transfer measurements for determination of intraribosomal distances. Such energy transfer measurements between the bound fluorescein isothiocyanate derivative of erythromycylamine and fluorescent labeled L7 yielded a distance of approximately 70 A between these two moieties (R. Langlois, C. C. Lee, C. Cantor, R. Vince and S. Pestka, unpublished). Although the erythromycin binding site appears to bt: in the vicinity of the peptidyl portion of peptidyl-tRNA (24, 24u), further characterization and definition of the erythromycin binding site should result from affinity labeling studies with derivatives of erythromycylamine (31). After such localization, the distance measurements between the erythromycin binding site and the factordependent GTPase center involving L7/ L12 should have more relevance.

234

SIDNEY PESTKA

B

ANTIBIOTIC MOLARITY

FIGS.10A and 10B. Effect of erythromycin analogs (see Figs. 12-14) on erythromycin binding to ribosomes. Each 0.50-ml reaction mixture contained the following components: 0.1 M KCI; 0.01 M NH,CI; 0.004 M MgCL; 0.01 M Tris-chloride, pH 7.2; 1.1 pM ["Clerythromycin; 7.5 A, units of NRCI-washed ribosomes; and erythromycin analogs at the concentrations indicated on the abscissas. Reactions were started by the addition of ribosomes and incubated at 24°C for 30 minutes. In the case of analog 28, the reaction was incubated at 24°C for 15 minutes to minimize hydrolysis of the erythromycin derivative. Assays were performed by adsorbing ribosomes to cellulose acetate-cellulose nitrate filters (26).The data are presented as a percentage of the ["C]erythromycin bound to ribosomes in the absence of any nonradioactive erythromycin or erythromycin derivatives.

VI. Effect of Chloramphenicol on the Puromycin Reaction: Models of Ribosome Function Kinetic analysis of the effect of chloramphenicol and other antibiotics on peptidylpuromycin formation stimulated some interesting concepts. The effect of sparsomycin on peptidylpuromycin synthesis is shown in Fig. 17 as a control. Sparsomycin has been reported to be a competitive inhibitor of puromycin, and its activity as a competitive inhibitor of puromycin in the synthesis of peptidylpuromycin, by analysis with a double-reciprocal plot as well as with a Dixon plot, supports the validity of this interpretation (23). The results with chloramphenicol were different. In the usual double-reciprocal plot, chloramphenicol appeared to be a mixed inhibitor of the puromycin reaction for peptidylpuromycin synthesis (Fig. 18). When a Dixon analysis was performed (Fig. 19),

INHIBITORS OF PROTEIN SYNTHESIS

-7 -6 -5 -4

-7 -6 -5 -4

-7 -6 -5 -4

-7 -6 -5 4

LOG (ANTIBIOTIC MOLARITY)

FIG.11. Graph of y as a function of antibiotic concentration for computation of association and dissociation constants. (The names of the antibiotics are given in Figs. 12-14.) The equation log { ( K O [El ( [Ro] - [RE])/[RE]) - I} = n log [I] log KI = y (Table VIII) was used to calculate KI,the association constant for the binding of inhibitor to ribosomes; KO is the association constant for the binding of erythromycin to ribosomes (9.9 x 10' M-' was used for these calculations); n is the number of inhibitor molecules binding to ribosomes; [El is the free erythromycin concentration; [Ro] is the total concentration of ribosomes potentially active in binding erythromycin; and [RE] is the concentration of the erythromycin-ribosome complex. From the data of Figs. 10A and 10B, y was plotted as a function of log [I] (26). Some representative results are presented in this figure. A computer program was used to determine the line of best fit by the method of least squares. When y = 0, n log [I] = -log Kl; thus assuming that n = 1, log KI = -log [I] at the baseline intercept. The baseline intercept was used to determine Kr. When n = 1, the KI values determined by the intercept of the abscissa or the ordinate are theoretically equivalent. KI for erythromycin was determined independently to be 9.9 x 10' M-' and, as computed above with a nonradioactive competing sample, as 1oR M-I.

+

it appeared that chloramphenicol inhibited this synthesis in two distinct phases. A fraction of the peptidylpuromycin formed was very sensitive to chloramphenicol whereas another portion was relatively resistant ( 23). This was quantitatively indicated by the Dixon plot, from which the two dissociation constants were estimated as 70 pM and 2.2 mM. Similar results were obtained with other antibiotics, such as amicetin, blasticidin S and gougerotin.

236

SIDNEY PESTKA

c.Hi'*ko+3

"0-OESOSAMINE

'"o-OESOSAMINE

*. '0-CLADINOSE

o' cn3

'0-CLAOINOSE

cn3

CHI

I

ERYTHROMYCIN A

2

OESOSAMINE MODIFICATIONS OF ERYTHROMYCIN A:

-

On

0

9

-

CHI

p

0

O-

...oT...o-p+::o B

~

~

CHI

3

4

9c-c~~

On

N/CH3 3

~

cn3 5

C H ~ 6

FIG. 12. Structures of erythromycin and erythromycin analogs.

14

IS

17

R = R'= R"= H

ie

R = R'= R * = CH~CO-

R=H R=CH3CO-

19 R * R'. CH3CO20 R = R ' = H

FIG.13. Structures of erythromycin analogs.

237

INHIBITORS OF PROTEIN SYNTHESIS

27 28 29

30 31 32 33

34 35 36 37 38 39

40 41 42 43 44

FIG.14. Structures of erythromycin analogs.

In essence, this result does not appear to be consistent with the simple acceptor-donor site model for protein synthesis on the ribosome. The model indicates that there are two states with different sensitivities to chloramphenicol in which peptidylpuromycin can be synthesized. From the model of the donor-acceptor site hypothesis (Fig. 20), there should be only one state in which peptidyl-tRNA can react with puromycin. The chloramphenicol probe indicates that peptidyl-tRNA in fact exists in two states that can react with puromycin, one state being relatively sensitive to chloraniphenicol and the other relatively insensitive. A different model for ribosome function may be proposed to explain these results. In this model of the ribosome epicycle (Fig. 21), it is proposed that peptidyl-tRNA exists in two states or may enter the ribosome from two sides. From each side of the ribosome, the peptidyl-tRNA could enter a donor or peptidyl site, which is distinguished essentially by the end of peptidyl-tRNA. Similarly, aminoacyl-tRNA, the acceptor, would enter an acceptor area, which is reserved for the aminoacyl ter-

238

SIDNEY PESTKA

ANTIBIOTIC MOLARITY

FIG.15. Effect of leucomycin ( L M ) analogs on erythromycin binding to ribosomes. Experiments were performed as described in the legend to Figs. 10A and 10B.The data are taken from Pestka et al. (30).

minus of aminoacyl-tRNA. Nevertheless, both peptidyl- and aminoacyltRNA could enter these sites from either side. Hence, it is postulated that there may be two sites for chloramphenicol binding to ribosomes and these may affect peptidyl transfer from the two directions with different sensitivities. Alternatively, there may be one site for chloramphenicol binding that affects the peptidyl transfer from the two sides differentially. Equilibrium dialysis with NHaC1-washed ribosomes, in fact, indicates that there may be two sites for chloramphenicol binding, one with high affinity and one with low affinity for ribosomes (32).

VII. The Sistrand: the Translational Unit It has been known for some time that polycistronic messenger RNAs exist in bacteria. The mRNA transcribed from the lactose and histidine operons of Escherichia coli (33j 34) as well as from the RNA phages f2, R17, MS2 and Qp (35-39) represent such polyci,stronic messages.

239

INHIBITORS OF PROTEIN SYNTHESIS

1.5

-

1.0 -

-

6.0

5.5

50

4.5

4.0

3.5

FIG.16. Minimal inhibitory concentration (MIC) as a function of concentration for 50% inhibition of erythromycin binding to ribosomes for leucomycins and leucomycin derivatives. The log of the MIC (rg/ml) for each of the compounds as determined against Staphylococcus aureus is plotted as a function of the p L % for these same compounds numbered as follows: 1, leucomycin A;2, leucornycin A,; 3, leucomycin A,; 4, leucomycin A; 5, leucomycin As; 6, magnamycin B; 7, leucomycin &; 8, leucomycin U; 9, leucomycin A3 N-oxide; 10, demycarosyl leucomycin As; 11, 9-dehydro-18-dihydroleucomycin A3; 13, 2’-O-acetyl-3’-desdimethylamino-3’-oxoleucomycin A,. The line of best fit was determined by the method of least squares with a computer program. For determination of this line of best fit, only the filled circles were used. This line is described by the following equation: log (MIC) = 6.861 - 1.362 PI&%.Thus, for any pIGos, value the estimated MIC can be calculated. Data are from Pestka et al. ( 3 0 ) .

These RNA phage messages carry the information for three different cistrons and they are translated into three separate polypeptides. Poliovirus RNA is also polycistronic, leading to the production of a number of different proteins (40,41).However, the mechanism by which each of these mRNAs is translated is quite different. The mRNA for the RNA phages has three separate initiation and termination sites separated by nontranslated regions. Each peptide is produced separately from start to finish. On the other hand, the polio RNA has a single initiation and single termination site. During translation a single large precursor, which is subsequently cleaved into the individual virus proteins, is synthesized (40,41 ). The term “polycistronic” in no way distinguishes between these two mechanisms of translation and is a misnomer when applied to translation of mRNA since it is a genetic term. At present, there is no term describing the translational unit. We have proposed the term “sistrand,” derived from “single initiation single termination strand,’’ for

240

SIDNEY PESTKA

PUROMYCIN MOLARITY

do6

11s

10-5

FIG.17. Kinetics of sparsomycin inhibition of peptidylpuromycin formation. Peptidylpuromycin synthesis was determined at the various concentrations of sparsomycin. Reaction conditions are similar to those described in the legend to Fig. 4. 0, NO sparsomycin; 0, 1 pM sparsomycin; A, 3 gM sparsomycin. A reciprocal plot of the data of the left panel ( VmSx= 4.2) is presented in the right panel (puromycin, K,,,= 2.4 pM; sparsomycin, K I = 0.16 pM).

5 4

3 2 I

0

5

K

)

l

5

2

0

2

[3H]PUROMYClN MOURITY x106

5

-

2

0

2

4

11s

6

8

K

)

10-5

FIG. 18. Kinetics of chloramphenicol inhibition of peptidylpuromycin formation. Reaction conditions were similar to those of Fig. 4 (23 ), The concentration of puromycin was vaned as shown on the abscissa. A reciprocal plot of the data of the left panel is presented in the right panel. 0, No chloramphenicol; A, 0.1 mM chloramphenicol; 0, 0.5 mM chloramphenicol; A, 1 mM chloramphenicol.

the unit of translation ( 4 2 ) . The sistrand represents the translational unit whereas the cistron represents the genetic unit. The sistrand is that unit of mRNA that lies between an initiation and termination signal (Fig. 22). Thus, these phage mRNAs are polysistrandic, having three

INHIBITORS OF PROTEIN' SYNTHESIS

Ki2

Ki1

CHLORAMPHENICOLMOURITY

~ 0 4

I

FIG.19. Kinetics of chloramphenicol inhibition of peptidylpuromycin formation. Reaction conditions were similar to those of Fig. 4 (23).Chloramphenicol concentration was vaned as indicated on the abscissa at four different puromycin concentrations. Polynbosomes were added last to start the reactions. The data of the left panel are plotted and calculated in the right panel ( Ktl = 70 PM; Kt, = 2.2 mM) ac13.7 pM puromycin; A, 4.55 pM purocording to Dixon [BJ 55, 170 (1953)l. mycin; 0, 2.28 p M puromycin; A,1.14 p M puromycin.

e,

distinct initiation and termination sites, as well as polycistronic. Polio mRNA is monosistrandic, but polycistronic. For some time, it has been suspected that most, if not all, eukaryotic mRNAs are monosistrandic (40,43, 44), but the evidence was not definitive. With the use of specific inhibitors of initiation, it should be possible to block protein synthesis at initiation, and thus functionally fix ribosomes at the initiation sites of mRNA. Therefore, mRNAs with a single initiation site should appear in the monoribosome region, those with two in the

n

- CODON

TRANSPEP-

RECOGNITION

-

PEP

TIDATION

t

-

e

EC TRANSLOCATION

FIG. 20. Schematic illustration of the donor ( P ) and acceptor ( A ) site model of ribosome function.

CODON

RECOGNITION

-

c

b

= I

TRANSLOCATION

n

- TRANSEP-

CODON

TIDATION

RECOGNITION

f

e

d

FIG.21. The ribosome epicycle; a model of ribosome sites and ribosome function for codon recognition, transpeptidation and translocation. Major features of the model are as follows: ( a ) There are two or more sites for tRNA binding on the 50 S subunit. Although the model would be essentially similar for any number of sites greater than two, the minimum number of sites on the 50 S subunit is two. ( b ) The two sites on the 50 S subunit are functionally similar but not identical. Transpeptidation can occur in both directions as illustrated ( b , e ) . As a consequence of this, each site can contain aminoacykRNA, peptidyl-tRN.4 or deacylated tRNA. The direction of transpeptidation depends on the nature of the constituents of the sites rather than on the sites themselves. ( c ) There is only one site for tRNA binding on the 30 S subunit; this is the decoding site. By a conformational change (shown in the figure) or by a rotation of the subunit, the site can align with either 50 S site. Because of this separate relative movement over many angstroms, the subunits should be separate particles. ( d ) Translocation involves movement of mRNA along the 30 S subunit, realignment of the 30 S subunit decoding site with the second 50 S site, and removal of deacylated tRNA. It is probable that removal of tRNA is the step dependent on factor EF-G and GTP; realignment of the 30 S site with the second 50 S site may or may not be dependent on EF-G. It is possible that factor EF-Tu may have a role here. ( e ) In order for the active center of the peptidyltransferase to interact in identical ways with peptidyl-tRNA and aminoacyl-tRNA in the two sites, it may be necessary that the active site of the peptidyltransferase rotate with respect to the sites. However, since the C-C-A( peptidyl ) end of peptidyl-tRNA does not bind firmly to ribosomes, it may be possible for the 3’-terminal portion of peptidyl-tRNA to align identically with the catalytic center of the peptidyltransferase without rotation of the catalytic center. This may be possible because of free rotation around each of the bonds of the C C - A ( peptidyl) end. Analogously, although the C-C-A( amino acid) of aminoacyl-tRNA is the fixed moiety, the free rotation around each bond of this end may place the C-C-A(amino acid) from either 50 S site in the same stereochemical position with respect to the peptidyltransferase.

243

INHIBITORS OF PROTEIN SYNTHESIS INITIATION

TERMINATION

v

1 SISTRAND

( PE::iRgk v m

v m

3 SISTRANDS

(POLYCISTRONIC)

FIG.22. Schematic illustration of the sistrand, the translational unit; strand of mRNA between initiation and termination signal.

diribosome region, those with three in the triribosome region, and so forth. Although the usual initiation inhibitors were not sufficiently specific to allow performing such studies, we found a group of inhibitors that were useful ( 4 5 ) . Studies with the harringtonine alkaloids indicated that these agents are specific inhibitors of initiation in intact HeLa cells and that most ribosomes appear in the monosome region in cells exposed to harringtonine ( 4 5 ) . These results indicate that HeLa cell mRNAs may be monosistrandic. It is likely that many cellular messages from animal cells are polycistronic, but monosistrandic. It is of interest to ask why eukaryotic cells should translate mRNAs predominantly via monosistrandic messages and to consider what advantages, if any, such posttranslational controls offer. Since prokaryotes are essentially nutritionally oriented with minimal differentiation, the synthesis of proteins essentially as final products corresponds to their immediate needs, i.e., their proteins are synthesized and immediately utilized. Differentiation, in contrast, requires, and by definition means, that cells evolve specialized functions. A cell in one part of the organism synthesizes what a cell or area in another location requires. In other words, protein products from one part of the system may be required in another part of the system and thus must be transported through space and time within the system. A further requirement may be that the product be present in an inactive form until needed. A priori, there appears to be no reason why prokaryotic polysistrandic mRNA cannot accomplish this. However, if eukaryotes routinely synthesized precursor proteins that are inactive and that require activation prior to function, eukaryotes would have a built-in mechanism for differentiation of cellular functions. The blood coagulation system provides a good example of the synthesis of numerous protein precursors in the liver. These proteins are transported to a particular place (the bloodstream) and normally function only when required (bleeding), at which time they are activated. It is possible that the synthesis of all proteins as inactive precursors provides a mechanism for the evolutionary beginnings of cellular differentiation.

244

SIDNEY PESTKA

It appears that inhibitors of protein synthesis and ribosome function allow us a striking insight into ribosome function and soon, perhaps, into ribosome structure. Our knowledge of protein synthesis has allowed us to delineate the mode of action of many of these agents. The agents themselves have allowed us insights and views into ribosomes and protein synthesis not available by other means.

REFERENCES I. J, Lucas-Lenard and F. Lipmann, ARB 40,409 ( 1971 ). 2. R. Haselkorn and L. B. Rothman-Denes, ARB 42,397 ( 1973). 3. S. Pestka, Annu. Rev. Microbiol. 25,487 ( 1971). 4. B. Weisblum and J. Davies, Bacteriol. Reo. 32, 493 (1968). 5. S. Pestka, in “Methods in Enzymology” Vol. 30: Nucleic Acids and Protein Synthesis (L. Grossman and K. Moldave, eds.), Part F, p. 261. Academic Press, New York, 1974. 6. N. R. Towers, H. Dixon, G. M. Kellerman and A. W. Linnane, ABB 151. 361 (1972). 7. D. Vazquez, FEBS Lett. 40, S63 ( 1974). 8. R. E. Monro and K. A. Marcker, J M B 25,347 ( 1967). 9. S. Fahnestock, H. Neumann, V. Shashoua and A. Rich, Bchem 9,2477 (1970). 10. S. Pestka, R. Vince, S. Daluge, and R. Harris, Antimicrob. Ag. Chemother. 4, 37 (1973). 11. S. Pestka, PNAS 89, 624 ( 1972). 12. S . Pestka, in “Methods in Enzymology” Vol. 30: Nucleic Acids and Protein Synthesis (L. Grossman and K. Moldave, eds. ), Part F, p. 462. Academic Press, New York, 1974. 13. L. P. Gavrilova and A. S. Spirin, in “Methods in Enzymology” Vol. 30: Nucleic Acids and Protein Synthesis (L. Grossman and K. Moldave, eds.), Part F, p. 452. Academic Press, New York, 1974. 14. S. Pestka, BBRC 40, 667 (1970). 15. S. Pestka and N. Brot, JBC 248,7715 (1971). 16. S. Pestka, JBC 244, 1533 (1969). 17. S. Pestka and J. W. Bodley, in “Antibiotics” (J. W. Corcoran and F. E. Hahn, eds.), Vol. 111, pp. 551-573. Springer-Verlag, Berlin and New York, 1974. 18. S. Pestka, in “Antibiotics” (J. W. Corcoran and F. E. Hahn, eds.), Vol. 111, pp. 480-486. Springer-Verlag, Berlin and New York, 1974. 19. S. Pestka, Abst. Papers, 166th Nut. Meet. Amer. Chem. SOC., Chicago, Illinois, 1973. Carb #2. 20. S . Pestka, ABB 138, 80 (1970). 21. H. Fujimoto, T. Kinoshita, H. Suzuki and H. Umezawa, J . Antfbtot. 23, 271 (1970). 22. S. Pestka, in “Antibiotics” (J. W. Corcoran and F. E. Hahn, eds.), Vol. 111, pp. 323-326. Springer-Verlag, New York, 1974. 23. S. Pestka, JBC 247, 4669 (1972). 24. S. Pestka, Antimicrob. Ag. Chemother. 5, 255 (1974). 24a. N. L. Oleinick and J. W. Corcoran, Roc. Int. Congr. Chemother., 6th, 1969 p. 202 ( 1970). 25. S. Pestka, Antimicrob. Ag. Chemother. 8, 474 (1974).

INHIBITORS OF PROTEIN SYNTHESIS

245

25a. R. Harris and S. Pestka, JBC 248, 1168 (1973). 26. S. Pestka and R. A. LeMahieu, Antimicrob. Ag. Chemother. 6, 479 (1974). 27. S. Pestka, R. A. LeMahieu and P. Miller, Antimicrob. Ag. Chemother. 6, 489 (1974). 28. J. A. Wilhelm, N. L. Oleinick and J. W. Corcoran, Antimicrob. Ag. Chemother.1967 p. 236 (1968). 29. J. C.-H. Mao and M. Putterman, JMB 44,347 (1969). 30. S. Pestka, A. Nakagawa and S. bmura, Anttmicrob. Ag. Chemother. 6, 606 (1974). 31. R. Vince, D. Weiss and S. Pestka, Antimicrob. Ag. Chemother. 9, 131 (1976). 32. J. L. Lessard and S. Pestka, JBC 247, 6909 ( 1972). 33. Y. Kiho and A. Rich, PNAS 54, 1751 (1965). 34. R. G. Martin, CSHSQB 28, 357 (1963). 35. D. Nathans, M. P. Oeschger, K. Eggen and Y. Shimura, PNAS 56, 1844 (1966). 36. E. Viiiuela, I. D. Algranati and S. Ochoa, E J B 1, 1 (1967). 37. H. F. Lodish and H. D. Robertson, CSHSQB 34,655 (1969). 38. J. Argetsinger Steitz, CSHSQB 34, 621 (1969). 39. H. D. Voonna, FEBS Abstr. 6, 114 (1969). 40. D. Baltimore, M. F. Jacobson, J. Asso, and A. Huang, CSHSQB 34,741 (1969). 41. J. L. Saborio, S. S. Pong and G. Kocli, J M B 85, 195 ( 1974). 42. J. S. Tscherne and S. Pestka, Amer. SOC. MicroMol. Abstr., 75th Annu. Meet. K105, p. 164 (1975). 43. E. L. Kuff and N. E. Roberts, J M B 26,211 (1967). 44. N. S. Petersen and C. S. McLaughlin, J M B 81,33 (1973). 45. J. S. Tscherne and S . Pestka, Antimicrob. Ag. Chemother. 8,479 (1975).

This Page Intentionally Left Blank

Interaction with Nucleic Acids of Carcinogenic and Mutagenic N-Nitroso Cornpounds W. LIJINSKY Biology Division Oak Ridge National Laboratoy' Oak Ridge, Tennessee I. Introduction

.

.

.

.

.

.

11. Biological Activity of Nitroso Compounds A. Direct-Acting Mutagens and Carcinogens

. .

. . .

. . .

. . .

. . .

. . . . . . . . .

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

B. Indirectly Acting Mutagens and Carcinogens 111. Alkylation of Nucleic Acids . . . . . . . . . A. In Bacteria, Viruses and Chemical Systems. . B. In Tissue or Organ Cultures . . . . . . C. In Animals in Vivo . . . . . . . . IV. Relation of Nucleic Acid Interactions to Biological Activity. V. Conclusions. References

247 249 250 251 254 254 257 258 263 266 268

1. Introduction A very attractive hypothesis, that carcinogenic N-nitroso compounds have that particular activity by virtue of undergoing in uivo conversion to an agent that alkylates nucleic acids, was advanced some 15 years ago. At the time the idea was logical and simple and was supported by the rather scant experimental evidence then available. The idea arose from the work of Lawley and Wallick ( I ) with nitrogen mustards, known to be alkylating agents, which interact with the nitrogen-7 of guanine in nucleic acids, The first report of such a reaction taking place with a nitrosamine was that of Magee and Farber ( 2 ) ; it was based upon experiments carried out with dimethylnitrosamine ( I ) in rats in U ~ U O . Formation of such alkylated products of nucleic acids was also demonstrated in uitro using tissue slices, and the participation of enzymes in the reaction scheme was inferred ( 3 ) . At about the same time, long-term testing of several nitrosamines was being carried out both by Magee et al. and by Druckrey et al. The results led to the hypothesis that, in general, nitrosamines that can be converted into alkylating agents are carcinogenic, whereas those that cannot be so converted, such as diphenylnitrosamine (11) and Operated for the Energy Research and Development Administration by Union Carbide Corporation. 247

248

W. LIJINSKY

@?a

CH3-N-CH3

A0

NO

I -1

CH3-C- y - CH3 H36 NO

CH3

@J-

NO

H

I @N-N-NlCH3

Y

CH3

e N - C H 3

A0 H

FIG. 1. I, Dimethylnitrosamine; 11, diphenylnitrosamine; 111, t-butylmethylnitrosIV, N-methyl-N-nitrosoaniline; V, l,l-dimethyl-3-phenyltriazene; VI, N-methyl-N-nitrosocyclohexylamine,

amine;

t-butylmethylnitrosamine (111), are not carcinogenic ( 4 ) . However, a noteworthy exception was the carcinogenic methylnitrosoaniline ( IV) , which, like 111, should not be convertible into an alkylating agent ( 5 ) . Other deviations from the rule have continued to appear over the years. The suggestion now is that some alkylating species can be formed from, for example, IV, even though it cannot form a diazoalkane or alkylcarbonium ion; a phenyldiazonium ion has been suggested, and the same reactive species could be responsible for the carcinogenic action of phenyldimethyltriazine ( V ) , which, though not a nitroso compound, is closely related chemically (6). This proposal, too, raises another question. If methylnitrosoaniline (IV) acts through conversion to a diazonium ion, does methylnitrosocyclohexylamine ( VI ) also act through this mechanism, since these two nitroso compounds have almost identical carcinogenic action, both qualitatively and quantitatively ( 7 ) ? Furthermore, if VI, known to methylate nucleic acids in vivo (8, 9), acts through diazonium ion formation, it might be expected that all N-nitroso compounds act in a similar way. One would also assume that a mixture of a primary amine and nitrous acid would be biologically active (mutagenic and carcinogenic). There is an additional inconsistency in applying this hypothesis to IV and VI, namely, that the latter is converted into a methylating agent but not, as far as can be determined, into a cyclohexylating agent. On the other hand, IV must act as a phenylating agent. Since both compounds are equally effective in inducing esophageal tumors when fed to rats ( 7 ) , we must assume that they act by different mechanisms, or that both act by an as yet unknown mechanism common to both. The literature in this area is profuse with exceptions and deviations

CARCINOGENIC AND MUTAGENIC N-NITROSO COMPOUNDS

249

of this kind from the several hypotheses advanced to explain carcinogenesis of N-nitroso compounds in terms of alkylation of cellular DNA, SO much so that there are some who doubt the whole concept (10). Indeed, many experiments have been designed to show only positive correlations, while the essential negative correlations have been forgotten or ignored. The vital observation that what is true for carcinogens of this class is not true for noncarcinogens of this class has not been made. Until it is, alkylation of nucleic acids by nitroso compounds in uiuo remains an intercsting but unproved explanation of the mechanism of their carcinogenic activity. It is certainly possible that other types of interaction with nucleic acids (and perhaps proteins) might be important. Much attention has been given to the commonly used alkylating agents niethylniethanesulfonate and ethylmethanesulfonate, with which a great deal of experimental work has been done. These compounds are very potent mutagens in bacteria as well as in higher organisms. They alkylate nucleic acids extensively both in uitro and in uiuo, giving rise to sevcral alkylated bases, particularly 7-alkylguanine (11). Nevertheless, neither of these compounds appears to be carcinogenic or, at most, neither is more than a very weak carcinogen. However, most carcinogens do interact with nucleic acids, often after suitable activation, and N-nitroso compounds are no exception. Only a handful have been tested, owing to the expense and difficulty of labeling any but the simple compounds with a radioactive marker. In some cases, it has not been possible to localize the sites of interaction of the nitroso compounds on the nucleic acids, or to pinpoint the types of interaction that took place, since hydrolysis led to loss of the label ( 8 , 9). More important, two large questions remain: ( 1 ) Does the type and extent of reaction of a nitroso compound with nucleic acids-particularly DNA-correlate with its relative carcinogenicity or mutagenicity? ( 2 ) Is the type of intcraction we can measure, especially alkylation, relevant to carcinogenesis and mutagenesis, which ( according to conventional ideas) result from select and specific changes in small regions of the genome?

II. Biological Activity of Nitroso Compounds First wc' must consider the differences in biological activity between various types of nitroso compounds, and thereby obtain a basis for judging the relcvance of the biochemical studies described and discussed below. It is convenient, although not necessarily justified, to separate N-nitroso compounds into two types when considering their biological activity-those that act directly and those that require metabolic (or chemical) activation to be effective.

250

W. LIJINSKY

A. Direct-Acting Mutagens and Carcinogens The direct-acting nitroso compounds, referred to usually as nitrosamides, are derivatives of N-nitroso-N-alkylcarbamic acids ( VII ) . This group includes nitrosoalkylureas ( VIIb,c,d ), nitrosoalkylcarbamates (VIIa) and nitrosoalkylguanidines (VIII). As a class, they are much less stable than are nitrosamines. They tend to be especially unstable in water at pHs at or above neutrality (i.e., alkaline) and decompose with the formation of diazoalkanes. Several of the best-known ones were used for many years in the preparation of the valuable methylating agent diazomethane. One of the most studied of these compounds is methylnitrosonitroguanidine ( MeNNGdn), first prepared for use as a detonator, but now the most widely used chemical mutagen. As a carcinogen, it has not been very versatile, since its action is exclusively local, inducing tumors of the skin (12), glandular stomach ( 1 3 ) and large intestine of rodents and in the stomach of dogs (14)when applied in solution to the tissue. The compound has been extensively studied biochemically, the relation between biological activity and chemical interaction with proteins and nucleic acids being the main concern. In one comparative study, the ethyl analog appears to be a weaker carcinogen than the methyl prototype (12). A number of nitrosoalkylureas (VIIb-d) have been tested as mutagens and carcinogens. They appear to be less potent mutagens than MeNNGdn (in instances where comparison is possible) but quite effective carcinogens. Although methylnitrosourea and ethylnitrosourea (VIIb, R = Me or Et) are locally acting carcinogens, producing skin tumors in mice ( 1 5 ) , rats (16) and hamsters (16) by chronic skin painting, their carcinogenic action is much more widespread than that of MeNNGdn. They have induced tumors of the nervous system in rats and mice after several treatments with quite low doses (17),or even

Yo?

R-N-C-X

I

y0yH

R-N-C-N,

H ,

NO2

m FIG.2. Structures of nitrosamides. VII: alkylnitrosocarbamates ( R = alkyl or aryl); ( a ) X = OR, carbamate esters; ( b ) X = NR, monoalkylureas; ( c ) X = NHR’, dialkylureas; ( d ) X = N/R‘ \R”

8

trialkylureas.VIII: nitrosoalkylguanidines.

CARCINOGENIC AND MUTAGENIC N-NITROSO COMPOUNDS

251

after a single dose (18), and act transplacentally when injected into pregnant females, inducing tumors of the nervous system and of other organs in a large proportion of the offspring (19). They also induce tumors of many organs other than the nervous system after single or few injections, although no liver tumors appear in rats even when the compounds are injected into the hepatic portal vein ( 2 0 ) .n-Butylnitrosourea (VIIb, R = Bu) has been a potent leukemogen in rats following multiple treatments (21 ). It is significant that nitrosotrialkylureas ( VIId), which are very much more stable than nitrosoalkylureas ( VIIb), also give rise to nervous system tumors in rats after multiple treatments (22). Nitrosoalkylcarbamic esters ( VIIa ) are an interesting and possibly important group of nitrosamides, because many widely used insecticides are N-alkylcarbamate esters that may be converted into nitroso derivatives in the environment or in vim (23). Ethyl N-methyl-N-nitrosocarbamate ( VIIa, R = Me, R’ = E t ) (methylnitrosourethan) has been extensively studied as a mutagen and carcinogen and is highly effective as both (24,25);the corresponding N-ethyl compound has been less investigated, but seems to be an equally potent carcinogen ( 4 ) . In chronic tests in animals, methylnitrosourethan has induced tumors in the esophagus and forestomach of rats and mice and induced experimental tumors of the pancreas in guinea pigs (26). This was until recently the only good animal model of this common human cancer. An analog of methylnitrosourethan, N-methyl-l-naphthyl-N-nitrosocarbamate ( VIIa, R = Me, R’ = l-naphthyl), the nitroso derivative of the common insecticide Carbaryl, is a very potent mutagen in some systems [Haemophilus influenzae, Escherichia coli (27), yeast (28) 1, more potent than either MeNNGdn or methylnitrosourethan, but is a considerably weaker carcinogen than the latter (Lijinsky et al., unpublished data ) .

B. Indirectly Acting Mutagens and Carcinogens Nitrosamines are a very large class of compounds that have attracted interest because they comprise the most broadly acting group of carcinogens. Study of their chemistry remains neglected, although knowledge of them goes back to the beginnings of organic chemistry. This ignorance certainly hampers our understanding of the mechanism of their carcinogenic action. Approximately a hundred nitrosamines have been tested for carcinogenic activity thus far, the overwhelming majority in the rat, which seems to be the most sensitive species. As with other types of carcinogens, they are most effective when given as multiple small doses over an extended period, rather than as a single dose, and they are highly effec-

252

W. LIJINSKY

tive when given orally. They show organ specificity to an exceptional degree, a point first made by Druckrey et al. ( 4 ) , to whom we are indebted for most of our knowledge of the biological activity of these compounds, There is almost no organ in which one or another nitrosamine has not induced tumors in some species. Neither does there seem to be any species resistant to the carcinogenic action of nitrosamines, whereas such refractoriness is common with other types of carcinogen. Diethylnitrosamine (XX) has been tested in some 20 species, mostly mammals, but including fish, birds and amphibians, and has induced tumors of some type in all of them. No generalizations can be made about the carcinogenic action of nitrosamines, although some common factors can be adduced from the extensive studies made in rats. Thus, there is no essential difference between open-chain nitrosamines (aliphatic, e.g., I, XX) and cyclic nitrosamines ( alicyclic, e.g., IX-XIX), either qualitatively in the organs affected, or quantitatively in the dose of compound needed to induce tumors in a large proportion of animals in a reasonable time. For example, a daily dose of 100 pg of dimethylnitrosamine given for 30 weeks to rats induced tumors in the liver in 27%of them (29), whereas an equimolar dose, 160 pg, of nitrosomorpholine (XVII) per day for 30 weeks induced tumors of the liver in 40% (30). There does not seem to be any generalization that can be made about molecular size or the presence of functional groups as influences on the carcinogenicity of nitrosamines. In the aliphatic series, the smaller members tend to be more potent than the larger molecules, and the symmetrical ones generally induce liver tumors whereas the unsymmetrical ones tend to induce esophageal tumors [pointed out by Druckrey et al. ( 4 ) ] , but there are many exceptions. In the series of cyclic nitrosamines, in contrast, the smaller ones, nitrosazetidine (IX) and nitrosopyrrolidine (X), are much the weakest, while the 6-, 7-, and 8-membered ring compounds are the most potent; still larger rings are less potent. The types of tumor induced in rats by various cyclic nitrosamines are given in Fig. 3. The presence of substituents in either aliphatic or cyclic nitrosamines can have profound effects on carcinogenic activity. These effects might be steric or electronic and can affect both potency and organ specificity. For example, methyl groups on the carbon atoms alpha to the nitroso function reduce carcinogenic activity, so that diisopropylnitrosamine (XXI) is a much weaker carcinogen than is diethylnitrosamine (XX) [or di-n-propylnitrosamine ( 4 ) 1, 2,6-dimethylnitrosopiperidine (XXIII ) is very much weaker than nitrosopiperidine (XI1 ) ( 3 1) and 2,5-dimethylnitrosopyrrolidine ( XXII ) is very much weaker than nitrosopyrrolidine ( X ) (Lijinsky, Singer and Taylor, unpublished data).

CARCINOGENIC AND MUTAGENIC

N-NITROSO COMPOUNDS

Nitroso-

x

Azetidine CN-NO Pyrrolidine CN-NO

XI

EN-NO

IX

O r g a n s with tumors Liver, lung Liver (hepatocellular) Liver (hepatocellular)

Pyrroline

m rn

Piperidine CN-NO Tetrahydropyridine CN-NO Hexamethyleneimine CN-NO

Xp

0 - N O

Heptamethyleneimine

m

0 - N O

Octameth yleneimine

xpl~

O ~ N - N O Morpholine

Lung (squamous carcinoma), esophagus Liver ( hepatocellular 1

m

S ~ N - N O Thiomorpholine

Esophagus

H[

n

Piperazine

ON-NUN-NO

253

Esophagus, nasal cavity Liver (hemangioendothelial) Liver (hemangioendothelial), esophagus Lung (squamous carcinoma 1, esophagus

Esophagus, nasal cavity, liver ( hepatocellular)

FIG.3. Tumors induced in rats by the ingestion of cyclic nitrosamines.

CHB-CH~,

CH3-CH, N-NO

CH3- CHz

'

xx

CH3

)7

0

N-NO

w

CH3

XxE

N-NO CH3-CH' CH3

XXI

CH3 ON-N

t\

N-NO

F

CH3

xxr

FIG.4. XX, Diethylnitrosamine; XXI, diisopropylnitrosamine;XXII, 2,s-dimethylN-nitrosopyrrolidine; XXIII, 2,6-dimethyl-N-nitrosopiperidine;XXIV, 2,6-dimethyl-Nnitrosomorpholine; XXV, 2,6-dimethyl-N,N'-dinitrosopiperazine.

254

W. LIJINSKY

In some cyclic nitrosamines, methyl groups in the beta position to the nitroso function, on the other hand, greatly enhance carcinogenic activity, as in 2,6-dimethylnitrosomorpholine ( XXIV ) and 2,6-dimethyldinitrosopiperazine (XXV) ( 3 2 , 3 3 ) . In the aliphatic nitrosamines, hydroxyl groups reduce carcinogenic activity ( 4 ) , but they have no such effect in the cyclic ones ( 3 4 ) . Unsaturation in the dialkylnitrosamines, greatly reduces or eliminates carcinogenic activity, as does the presence of cyano groups ( 4 ) , whereas in the cyclic nitrosamines unsaturation seems, if anything, to increase the activity (35, 36). Halogens, especially chlorine, greatly enhance carcinogenic activity, both in the aliphatic ( 4 ) and cyclic ( 3 7 ) nitrosamines. Carboxyl functions, as free acids or esters, eliminate carcinogenic activity in aliphatic or cyclic nitrosamines in all cases but one, nitrososarcosine, which is a weak esophageal carcinogen in rats ( 4 , 3 8 ) . Although only a few quantitative comparisons have been made of the carcinogenic activity of closely related nitrosamines, there is enough information of a semiquantitative nature to lead to the conclusion that there are clues to the mechanism of action of nitrosamines in the steric and electronic effects of substitution. The key seems to be the activity of the hydrogens on the carbon atoms alpha to the nitroso function.

111. Alkylation of Nucleic Acids Starting with the observations of Magee and Farber ( 2 ) , there has been much activity in the area of modifications brought about in D N A and RNA structure and activity by insertion of alkyl groups (usually methyl or ethyl) at certain places in the chain. The alkyl groups can be on purine or pyrimidine bases, on sugar moieties or on phosphate groups. The connection of such changes with mutagenesis is well established, but with carcinogenesis and metazoan cell transformation it is less certain. Therefore, the three types of study of alkylation are considered separately, even though many of the methods and procedures used are similar in all the investigations.

A. In Bacteria, Viruses and Chemical Systems Since the discovery that the mutagenic action of nitrogen mustards in microorganisms is connected with isolation from them of nucleic acids containing 7-alkylguanine, a great deal of work along the same lines has been done with other compounds (39). In general, the correlation has been fairly good qualitatively, although the extent of alkylation varies considerably from one compound to another. Guanine and other bases are alkylated at other sites [reviewed recently by Singer (11);

CARCINOGENIC AND MUTAGENIC

N-NITROSO COMPOUNDS

255

cf. also 40,411, but it is still uncertain which of these, if any, is responsible for the mutagenic effect. Comparison is often made between the alkylating nitroso compounds and more obvious alkylating agents, such as dimethyl sulfate or methyland ethylmethanesulfonate. The latter two are quite potent mutagens in bacteria and other microorganisms and are mutagenic in mammals ( 4 2 ) . They are only very weakly carcinogenic or noncarcinogenic. Nevertheless, a large amount of work has been done on alkylation of bacterial nucleic acids by methyl and ethyl methanesulfonates, and many alkylated bases have been isolated and identified ( 4 3 ) . The mutagenicity of the two compounds has been compared in yeast (44),and the relative extents of alkylation of nucleic acids in bacteriophage ( 4 5 ) . Using these data as a base, the assumption has been made that nitroso compounds behave similarly, especially the nitrosamides ( e.g., VIIb ), which appear not to need metabolic activation. These compounds also react at near neutral pH with nucleic acids in bacteria (and, like methyl and ethyl methanesulfonates, with DNA and RNA in purely chemical systems), to form alkylated nucleic acid derivatives (41 ) . Alkylation by MeNNGdn is increased in the presence of thiols, but they do not have this effect on other nitrosamides ( 4 6 ) . From these nucleic acids, a variety of alkylated purine and pyrimidine bases have been isolated ( 4 7 ) . Lawley has described the isolation and identification of ten alkylated products in such an experiment. Included are two O-methylated bases, which have recently been considered to be more relevant than bases alkylated at other positions, because of interference with the participation of these methylated positions in normal hydrogen bonding, therefore causing miscoding in DNA replication ( 4 8 ) . Alkylation at other positions, such as the N-7 of guanine is said to lead mainly to destabilization of the glycosyl linkage and subsequent deletion of the alkylated bases ( 4 9 ) and perhaps some mispairing. The actions of several nitrosamides have been investigated, and alkylation at N-7 of guanine has been usually the index of alkylating activity. Alkylation of the O6 of guanine has often been observed when looked for, although this is a more difficult procedure than the isolation of 7-alkylguanines ( 4 1 ). Methylnitrosourea ( MeNur ) and ethylnitrosourea (EtNur) give essentially the same results, but differ somewhat quantitatively ( 4 0 ) . There is little doubt that other nitrosamides, such as methylnitrosourethan and ethylnitrosourethan, which are also potent mutagens in bacteria (50) would give similar results if the trouble were taken to carry out the experiment with appropriately labeled compounds. Alkylation by nitrosamides in uiuo (and in uitro) might be expected to follow a course similar to the breakdown of these compounds in

256

W. LIJINSKY

chemical systems-that is, through formation of a diazoalkane, which takes place at only slightly alkaline pH ( p H 8). This mechanism of alkylation through diazoalkane formation was supported somewhat by the stronger mutagenicity of MeNur and MeNNGdn at pH 8 than at pH 6 ( 5 1 ) . However, other work designed to demonstrate the intermediate formation of diazomethane as the methylating agent derived from MeNNGdn has shown the opposite, that diazomethane is not involved ( 52). By using MeNNGdn prepared from trideuteromethylamine and containing a methyl group in which all the hydrogen atoms are replaced by deuterium, a nucleic acid that contains methyl groups derived from the deuterium-labeled nitrosamide can be isolated. Hydrolysis of the nucleic acid, followed by purification of the (abnormal) alkylated bases gives materials suitable for mass spectrometric analysis. The mass spectrum will show the nominal mass of the methylated base (e.g., 7-methylguanine ) if it is not derived from the deuterated nitrosamide, 2 mass units higher if the alkyl group is derived from the deuterated nitrosamide via diazomethane, or 3 mass units higher if the methyl group is transferred intact, without intermediate formation of diazomethane. In such experiments, the alkylated base always has a mass corresponding to the transfer of an intact alkyl group. This eliminates the possibility that diazomethane or any other methylenic compound (-CH,X; X # H ) is involved in alkylation of nucleic acids by nitrosamides. While nitrosamides alkylate nucleic acids in uitro (or in purely chemical systems) as readily as do direct alkylating agents, such as methyl methanesulfonate and, therefore, probably without needing significant activation, the nitrosamines present a different story. In bacteria, nitrosamides are mutagenic, often potently so, whereas nitrosamines are invariably inactive alone (24, 53). However, they can be converted into mutagens by suitable activation. This can take the form of generalized activation in an animal, as in the host-mediated assay ( 5 4 ) , or more specific activation by tissue homogenates or microsomal preparation, usually from liver, as in the extensive work of Ames ( 5 5 ) . Not all organs and tissues will serve to activate the nitrosamines. Even those in which the nitrosamine is activated sufficiently to give rise to tumors in uiuo often will not activate the nitrosamine to form a mutagen (56). This suggests that the mammalian liver converts nitrosamines ( and possibly other carcinogens responding in the Ames test) into mutagenic detoxification products. The link to carcinogenesis might, therefore, be a myth. Nevertheless, it does seem that, when suitably activated by a mammalian system, nitrosamines are converted to a form that will interact with nucleic acids of bacteria or other microorganisms (neurospora or yeast ) and give rise to mutations (54, 57). Presumably the product of such

CARCINOGENIC AND MUTAGENIC N-NITROSO COMPOUNDS

257

activation is an alkylating species in the case of simple aliphatic nitrosamines such as Me,N-NO ( I ) or Et,N-NO (XX), but what it might be in the case of cyclic nitrosamines such as XI1 or XIX is not SO clear (58). However, it is quite unlikely that the ring of such compounds is opened to form an alkylating moiety similar to a carbonium (or carbenium) ion (10). No evidence of reactions of this type has been obtained in studies in mammals, and it is believed that this type of heterocyclic ring is fairly stable,

B. In Tissue or Organ Culture Comparatively few studies have been made of interactions of nitroso compounds with nucleic acids in tissue culture or organ culture. Much more attention has been given to the simpler system of mutagenesis in bacteria, or to pharmacological studies in whole animals, which is more closely related to carcinogenesis. However, some of the earliest studies of the mechanisms of action of nitrosamines involved the use of tissue slices and homogenates, in which the metabolism of, for example, dimethylnitrosamine ( I ) was investigated (2). The findings were that both DNA and RNA were alkylated, principally at the 7-position of guanine, and that this occurred in organs in which tumors arose on administration of the compound to animals (59). More recent experiments have been concerned with the direct alkylating agents, the nitrosamides ( VIIb), rather than nitrosamines requiring metabolic activation, because the experiments can be more clear-cut and simple. Thus, Lawley has investigated the interaction of MeNNGdn with cultured mammalian cells (hamster embryo) and has compared the alkylation of nucleic acids with that by Me,SO, ( 4 1 ) . The reactions of the two compounds have been related to their respective mutagenic and carcinogenic activities. The effects of thiol compounds in promoting decomposition of MeNNGdn to an active methylating species, and the parallel increase in mutagenicity and in methylation of nucleic acids have been contrasted with the lack of effect of thiols on methylation by Me,SO, and on the very weak mutagenicity of the latter compound (60). Methylation of guanine at the 6-position is particularly responsive to the presence of thiol compounds. Sun and Singer have examined the action of some ethylating agents in HeLa cell DNA, with special attention to O6 ethylation of guanine ( 6 1 ) . Further insight into the mechanism of action of MeNNGdn as a mutagen was obtained by comparing the extent of labeling of nucleic acids with guanidino-labeled and methyl-labeled MeNNGdn ( both with 14C). The labeling was much more extensive with the methyl-labeled MeNNGdn than with the compound labeled in the guanidino portion

258

W. LIJINSKY

of the molecule, and was mainly in the form of various methylated bases ( 6 2 ) .That there was incorporation of the latter label at all, albeit much smaller, indicates that there is some interaction of the entire molecule of MeNNGdn with nucleic acids and, presumably, with proteins. This form of reaction might be more important than is presently believed (63). Few nitrosamines have been examined for their effect on DNA in cultured cells. In particular there seem to be no studies with cyclic nitrosamines. This is partially due to the difficulty of preparing radioactively labeled material, but also largely to a lack of interest in other than the standard compounds, dimethyl- and diethylnitrosamine ( I and XX), about which we seem to know so much, and yet so little. This is part of the general lack of innovation in the study of the mechanisms of carcinogenesis and mutagenesis. There have been a few studies of the interaction of carcinogens with macromolecules in oitro, notably the experiments of Setlow and Regan, who examined the repair of lesions in DNA induced by a variety of carcinogenic agents, including nitroso compounds ( 6 4 ) . As would be expected, nitrosamines are inactive in this system, which uses epithelial cells that lack the necessary metabolic activating mechanism, but nitrosamides are quite active. In particular, nitroso derivatives of several insecticides that are esters of N-methylcarbamic acid (for example, Carbaryl) produce alkali-labile bonds in the DNA of these cells (65). The change in the DNA is no greater after 24 hours than after a few minutes of contact ( 6 6 ) . It appears that a very strong interaction of the nitrosamide with DNA takes place in the cultured cells and that this causes fragmentation of the DNA when placed on an alkaline sucrose gradient. That this interaction does not occur with isolated DNA implies that metabolic activation is needed even for these supposedly direct alkylating agents. This is supported by experiments in which nitrosocarbaryl labeled with tritium in the naphthalene ring was used; the methyl group was not labeled. A small, but significant, binding of radioactivity to the DNA was observed, and this radioactivity sedimented with the DNA ( 6 6 ) . This argues that the entire molecule interacts with DNA in oitm and that any inethylation of the DNA bases might be due to rearrangement and decomposition of the complex during analysis and workup. Other nitrosamides, such as nitrosomethylurethan and MeNNGdn, show the same behavior, although binding studies with the labeled compounds have not been done in these cases.

C. In Animals

in Vivo

The first studies of nitrosamine biochemistry in relation to carcinogenesis were carried out in whole animal experiments by Magee and

CARCINOGENIC AND MUTAGENIC

N-NITROSO COMPOUNDS

259

Farber ( 2 ) and similar experiments have continued throughout the intervening period without adding much to their basic findings. In other words, our understanding of the mechanism of carcinogenic action of nitrosaniines has not greatly advanced in the last 15 years in spite of the large amount of good work that has been done. Briefly, the concept has been that nitroso compounds are converted, either enzymically or chemically, into moieties able to alkylate nucleic acids, particularly DNA, at specific sites. This leads to faulty replication and thence to faulty transcription, thereby giving rise to cells with a heritable abnormality that manifests itself in neoplasia. It has been customary to carry out “alkylation studies” by injecting animals with large, slightly subacutely lethal doses of a nitroso compound labeled with 14C or 3H (usually at low specific activity because of expense), isolating the nucleic acids from various organs a few hours later (2). The nucleic acids are hydrolyzed with more or less vigor, and the presence of a radioactively labeled alkylated base is sought among the hydrolysis products ( 2 ) . The use of large doses of these toxic compounds causes a lot of cell death (commonly 601%of liver cells in the dimethylnitrosamine studies), and this type of treatment does not remotely resemble the treatment necessary to give rise to liver tumors (67). Large single doses of any carcinogen are usually ineffective in causing tumors, or at least much less effective than smaller repeated doses, and there is a suspicion that observations of the effects of nitroso compounds given in this way are related more to toxicosis than to carcinogenesis. Hence such observations must be treated with reserve, although hidden within them might be information truly relevant to carcinogenesis. One has the feeling, particularly in studying nitrosamines, that any reaction in which they participate that is very common among tissues and organs is not ips0 facto related to carcinogenesis, since these compounds have such organspecific tumorigenic effects, as has been described ( 4 ) . Furthermore, since tumorigenesis seemingly involves only a comparatively small number of cells and only a small portion of the DNA is damaged, any chemical alteration of DNA that could be detected in such whole animal experiments is rather unlikely to be other than an artifact in the context of carcinogenesis. However, the reactions that have been observed might be an index of the reactions that are relevant to carcinogenesis, since there can be little doubt that the differences we see in carcinogenic activity between rather similar molecules are related to the differences in chemical structure and chemical reactivity. The concept of direct interaction with DNA by carcinogenic nitroso compounds, or by-products of their metabolism, as being directly causative of neoplasia has engaged the attention of most workers in this area of cancer research. The large number of experiments carried out

260

W. LIJINSKY

in oiuo, mainly in rats and mice, has shown that several carcinogenic nitroso compounds do interact with nucleic acids, both in “target” and “nontarget” organs. Early experiments of Magee and his co-workers, studying dimethylnitrosamine ( I ), showed that the specific activities of RNA isolated from liver, kidney, pancreas and spleen of rats were not very different (68). However, after hydrolysis of the nucleic acids and analysis of the products there was a larger proportion of 7-methylguanine in liver and kidney nucleic acids than in nucleic acids from pancreas and spleen, organs in which tumors are not induced by dimethylnitrosamine. In these and subsequent experiments there was always a large amount of radioactivity associated with the pyrimidine nucleosides eluting in the first few fractions of the ion-exchange chromatograms; the nature of this association is not known. The alkylation of DNA in such experiments was somewhat higher than of RNA, but RNA was often preferentially worked up because it was easier to isolate and purify in quantity. Swann et al. found that when alkylation of guanine in nucleic acids by diethylnitrosamine (XX) and ethylnitrosourea ( VIIb, R = Et) is compared with that by ethyl methanesulfonate and Et,SO, ( 6 9 ) , the level of alkylation by the last two compounds was minute, consistent with their lack of carcinogenicity. In one study with XX, it appeared that N-7-guanine alkylation of nucleic acids in the lungs of rats was not noticeably different from that in liver, although this compound induces liver tumors in rats as readily as I, but does not give rise to lung tumors (70). Kriiger has extended these studies of aliphatic nitrosamines to di-npropyl- and di-n-butylnitrosamine. In each case he was able to isolate, from rats treated with these compounds, nucleic acids containing a very small amount of some alkylated base, which he identified tentatively as, respectively, 7-propylguanine and 7-butylguanine ( 71 ). The proportions of these alkylated bases formed was orders of magnitude smaller than that from dimethyl- or diethylnitrosamine, which is not in accordance with the closely similar carcinogenic potencies of the four compounds. In extending these investigations to a variety of other nitroso compounds, especially heterocyclic ones, the main obstacle has been the difficulty in obtaining radioactively labeled compounds of sufficient purity. Except in a few instances, it has been necessary to use tritium-labeled compounds (prepared by exchange) because of the difficulty and expense in preparing 14C-labeled compounds. Purification of the tritiumlabeled compounds has been difficult, and the presence of traces of impurities has led to publication of at least one erroneous report of methylation of nucleic acids by cyclic nitrosamines (72). In a later

CARCINOGENIC AND MUTAGENIC

N-NITROSO COMPOUNDS

261

report, the same authors ( 9 ) failed to detect alkylation of nucleic acids by IX, X, XI1 and XIX, using deuterium-labeled nitrosamines combined with mass spectrometry, as well as tritium-labeled compounds. There was a small amount of a labeled base, but insufficient to identify, in the liver nucleic acids of rats treated with nitrosomorpholine (XVII). A recent study by Stewart et al. (73) has demonstrated the formation of small amounts of several radioactive products in the nucleic acids of rats treated with [14C]nitrosomorpholine.One of these products has been tentatively identified as 7- ( 2-hydroxyethyl)guanine, indicating that ring opening of XVII had occurred at the oxygen bridge. Some alkylation of liver nucleic acids in rats by [14C]nitrosopyrrolidine ( X ) has been reported by Kruger et al. ( 7 4 ) , but the product, present in very small amount, has not been identified, although it is certainly not 7-methyl- or 7-ethylguanine. Nitrosohexamethyleneimine (XIV) has been labeled with 14C and its metabolism and interaction with nucleic acids have been studied ( 7 5 ) . As in earlier work with the tritium-labeled compound (8),there has been no report of isolation of labeled alkylated bases from rats treated with this compound, which is a liver carcinogen in rats of potency comparable with that of dimethylnitrosamine (76). In contrast with this lack of detectable alkylation of liver nucleic acids in rats by most of the hepatocarcinogenic cyclic nitrosamines, nitrosomethylcyclohexylamine ( VI ) gives rise to methylation of liver nucleic acids almost as extensive as that by dimethylnitrosamine ( I ) (8), although this compound induces no liver tumors in rats (7). Using deuterium- and tritium-labeled VI, 7-methylguanine was easily seen in hydrolyzates of both DNA and RNA as a peak in the optical-density trace from the ion-exchange chromatogram, which was also a radioactive peak and coincided with the position of 7-methylguanine. Mass spectrometry of the material showed it to be (CD,)guanine, indicating that alkylation had occurred by transfer of an intact methyl group rather than through formation of diazomethane; the same result has been obtained in all similar studies using deuterium-labeled nitroso compounds (77). Although so much methylation of rat liver nucleic acids by VI was seen, none whatsoever was seen in the nucleic acids of the esophagus and squamous stomach of rats in which this compound does induce a high incidence of tumors (7, 8). Because VI is hepatotoxic at high doses in rats, this result suggests that 7-alkylation of guanine by nitroso compounds in the liver of rats might be a consequence of the action of a rather nonspecific enzyme on nitroso compounds of the “right” structure, and that the reaction of the product with the nucleic acids is related to hepatotoxicity, but is unrelated to carcinogenicity. Similar

262

W. LIJINSKY

interpretations can be given to the failure of methyl- or ethylnitrosourea to give rise to liver tumors in rats, even though the former produces easily detectable 7-methylguanine in rat liver ( 20) ; ethylnitrosourea alkylates liver nucleic acids to a much lesser extent ( 69). More recently, it has been claimed that 7-alkylation of guanine in nucleic acids in uiuo is probably not related to carcinogenesis, because of inconsistencies such as those mentioned above, but that alkylation of nucleic acids at one of the several other sites that have been reported is the significant event as in the previously cited biological studies. The large discrepancy in alkylation at N-7 of guanine between methyl- and ethyl-donating compounds (which does not correspond with the near similarity of the biological activities of the two types of compound), does not appear when alkylation of OGof guanine is examined ( 7 8 ) . In fact ethylation at this site is often more extensive than methylation (40). In studies related to carcinogenesis, Swann et al. have recently shown that, following a single large dose of dimethylnitrosamine, O6 inethylation of guanine in liver DNA is initially higher than in kidney, but disappears much more rapidly from liver than from kidney, which might relate better to the tendency of this type of treatment to induce kidney tumors rather than liver tumors (79). On the other hand, the falling off of alkylation in the liver nucleic acids might be due to hepatotoxicity of the treatment, leading to rapid death of liver cells and regeneration. A better indication of the role of O6 alkylation of guanine in carcinogenesis is recent work of Goth and Rajewsky, who showed (80) that a single treatment of rats with ethylnitrosourea (which induced nervous system tumors within a year) led to isolation of OG-ethylguaninefrom both liver and brain; initial extents of ethylation in both organs was about the same. However, the “half-life” of the OO-ethylguanine in the liver nucleic acids was quite short (30 hours) compared with that in brain (220 hours), which correlated with the induction of tumors in the brain, but not in the liver, after this treatment. Experiments such as this are certainly more satisfying than the earlier ones in which 7-alkylguanines were the objects of concern. Nevertheless, apart from suggesting that O6 alkylation of guanine in DNA is likely to cause anomalous base-pairing during replication of DNA, the authors have no explanation of the role this plays in neoplasia. Until more compounds have been examined in this type of system, and satisfactory negative as well as positive correlations have been obtained, the relationship of 0“alkylation to carcinogenesis is just as tentative as was N-7 alkylation. It is notable that no O6 alkylation of guanine by any carcinogenic cyclic nitrosamine has been reported.

CARCINOGENIC AND MUTAGENIC

N-NITROSO COMPOUNDS

263

Although several nitroso compounds do not form a detectable amount of an identifiable alkylated base in nucleic acids in uiuo, all the compounds so far examined interact to some extent with nucleic acids in both target and nontarget organs, as measured by incorporation of radioactivity. In Table I are the results of experiments in which large, slightly sublethal doses of several (labeled) nitrosamines were administered to rats and the specific activities of nucleic acids and proteins in various organs were measured. The results are expressed as the proportion of the dose incorporated per milligram of macromolecule and show no pattern related to induction of tumors by the nitrosamines, although these large doses do not, in general, give rise to tumors anyway. It is possible that some type of interaction not yet considered is the important one in carcinogenesis. Biochemists have focused instead on transfer of an alkyl group to a nucleic acid base, which is comparatively easy to detect and to interpret. Other types of interaction, even other types of alkylation, such as that of sugars or of phosphate, which might be of great importance (49) have not been given much attention. We are hampered in knowing what to look for by lack of knowledge of the chemistry of nitroso compounds, and it is hoped that this deficiency will be remedied in the near future.

IV. Relation of Nucleic Acid Interactions to Biological Activity While there is a wealth of evidence supporting the concept that mutagenesis in microorganisms is related to interaction of the mutagen, or a metabolite of it, with DNA, such a relationship in carcinogenesis is far from proved. Indeed there is no certainty that the processes of mutagenesis and carcinogenesis are related, although it would not be surprising if such were proved. I t is unwise to translate findings in microorganisms to projections of mechanisms in higher organisms. Although many N-nitroso compounds are mutagenic in bacteria, but only after enzymic activation in the case of nitrosamines, there appears to be only one report of mutagenesis in mammals by nitroso compounds ( 4 2 ) . The failure to demonstrate mutagenesis in the dominant lethal test with other nitroso compounds might be a consequence of an efficient DNA repair system for certain types of damage in the affected cells. On the other hand, most nitroso compounds are carcinogenic in mammals and interact with nucleic acids (as well as other cell components) in both susceptible and nonsusceptible tissues and organs. However, the connection between the observed interaction with nucleic acids and neoplastic transformation is unclear. In bacteria, nitroso compounds alkylate nucleic acids at a variety

RADIOACTIVITY

TABLE I NUCLEICACIDS AND P R O T E I N OF RATS GIVENS H - L . k ~ ~ ~ ~ ~ INTRAPERITONEALLY NITROSAMINES

INCORPORATED I N

Nitrosamine Dimethylnitrosamine (I)

Dose administered per rat 9mg

=

11 pCi

Diethylnitrosamine

48 mg

=

580 pCi

Methylnitrosoaniline (IV)

66 mg

=

800 pCi

Methylnitrosocyclohesylamine (VI)

24 mg

=

250 pCi

Nitrosoazetidine (IX)

46 mg

=

360 pCi

Nitrosopyrrolidine (X)

180 mg

=

130 pCi

Nitrosopiperidine (XII)

40 mg = 120 pCi

Nitrosohexamethyleneimine (XIV)

40 mg = 550 pCi

Nitrosomorpholine (XVII)

90mg

=

130pCi

Dinitrosopiperazine (XIX)

40 mg

=

420 pCi

0

Specific activity' and proportion of doseb Liver DNA RNA Protein DNA RNA Protein DNA RNA Protein DNA RNA Protein DNA RNA Protein DNA RNA Protein DNA RNA Protein DNA RNA Protein DNA RNA Protein DNA RNA Protein

Liver 2200 (1100)" 800 (400)c 250 (130) 1900 (14)c 900 (6)" 1900 (14) 80 (0.4) 400 (2) 5300 (29) 500 (ll)c 700 (14)c 1100 (22) 230 (4) ' 280 (5) 490 (9) 600 (20) 110 (4) 550 (18) 370 (15) 190 (8) 490 (20) 1500 (12) 300 (2.5) 1900 (16) 150 ( 5 ) 120 (4)C 180 (6) 300 (3)

Kidney Lung 200 (100)" 100 (60)c 140 (70) 200 (1.4) 300 (2)" 360 (3)C 400 (3)

iz ${

Esophagus Intestine 270 (140) 270 (140) 160 (80) 900 (6) 600 (4) 1300 (9) -

-

190 (4) 390 (8) 230 (5)

-

-

250 (10) 250 (10) 450 (19) 320 (3) -

-

210 (4.4) 190 (4) 600 (12) 230 (4.2) 160 (3) 160 (3) 180 (8) 170 (7) 320 (13) 210 (2) 200 (2) 600 (5) -

-

-

70(0.7)

50(0.5)

Disintegrations per minute per milligram of macromolecule. ( X 10') of the administered dose per milligram of macromolecule. Alkylated base detected.

* The numbers in parentheses are the proportion

P

-

E:

u

z$

CARCINOGENIC AND MUTAGENIC N-NITROSO COMPOUNDS

265

of sites and at least some of these alkylations give rise to mutations; most of them lead to death of the cells although the toxicity of nitroso compounds varies considerably among bacterial species. The quantitative relation between alkylation and mutagenicity is not clear, mainly because of the small number of nitroso compounds studied. In the same system, several nitrosamides have had very different mutagenic activities, but it is not known to what extent the alkylation of nucleic acids by the compounds differs. There is a profound effect of pH of the medium on the mutagenicity of nitrosamides, but again it is not known to what extent alkylation of nucleic acids differs in the different conditions. Other forms of direct interaction of nitroso compounds or their metabolites with nucleic acids are possible (these compounds are quite reactive and decomposition to diazoalkanes is only one example of this reactivity). It is possible that the extent to which the compound itself reacts with nucleic acids is the determining factor in the differences in mutagenic activity between the nitrosamides. Some preliminary evidence has been obtained (R. Elespuru, unpublished results) that the uptake of radiolabeled nitrosamides ( MeNur, MeNNGdn, and nitrosocarbaryl ) by E. coli is quite different and roughly parallels the mutagenic potencies of the compounds in this organism. Regan et al. have shown that nitrosocarbaryl and nitrosomethylurethan and related compounds differ from other carcinogens examined in their interaction with human DNA in vitro ( 6 5 ) .Further investigation of this unusual type of interaction is obviously needed. The earlier correlations of N-7 alkylation of guanosine by nitroso compounds with carcinogenic activity seem no longer valid. Far too many exceptions have appeared, and it is obvious that the 7-position of guanine is merely a good receptor site for alkylating species that are formed in viuo, especially in nucleic acids of the liver. The same might be true of OR alkylation of nucleic acids in target organs. The grossness of the analyses, the variety of cell types from which the nucleic acids examined are derived, and the lcng interval between the alkylation event and the appearance of tumors, leave too large a gap in our knowledge. This gap must start to be filled before any of the alkylation proposals are more than suggestions. On the other hand, recent studies of DNA repair following treatment of animals with carcinogenic nitroso compounds suggest that this might be an important factor in the action of nitroso compounds. Both aliphatic and cyclic nitroso compounds in uiuo lead to single-strand breaks in liver DNA, and these breaks are repaired at different rates, depending on the treatment of the animal. The single-strand breaks induced by dimethylnitrosaniine ( I ) and methylazoxymethanol acetate were not

266

W. LIJINSKY

completely repaired in 14 days, whereas those induced by methylnitrosurea, which does not give rise to liver tumors in rats, were repaired within 7 days ( 8 1 ) . Cyclic nitrosamines, too, damage liver DNA in rats even at low doses, XVII inducing both single-strand and doublestrand breaks, the latter of which are quite quickly repaired; the singlestrand breaks induced by XVII are not completely repaired in 14 days (as with I ) , whereas those induced by the much weaker liver carcinogens XI1 and XIX are repaired in a little more than 6 days (82). Apart from XVII, there has been no thorough study of alkylation of nucleic acids in duo by cyclic nitrosamines, including XI1 and XIX, SO it is not known whether the DNA damage seen is caused by alkylation or by some other type of interaction of the nitroso compounds with the nucleic acids. As with the other theories, it is necessary to reserve judgment on the part played by slow or deficient repair of nitrosamineinduced DNA damage in the carcinogenic process. It must be shown, for example, that this is true in organs other than liver in which tumors are induced by the nitroso compounds. The phenomenon could relate to metabolism (detoxification) and toxicity of the nitroso compound in the liver, and peculiar to this organ. As a parallel, we find that formation of bacterial mutagens from diethylnitrosamine by rat liver microsomes, supposedly related to liver tumor induction, does not occur with hamster lung microsomes, although diethylnitrosamine does induce lung tumors in the hamster ( 5 6 ) .

V. Conclusions The question whether interaction of nitroso compounds with nucleic acids is related to carcinogenesis cannot be answered at present, whereas the relation with mutagenesis is almost certain. In the case of mutagenesis, it is likely that alkylation of DNA at certain sites leads to mutations, and at most other sites to variable degrees of toxicity. The important sites of alkylation have not been established, but probably N-7 of guanine is not one of them; O6 of guanine or N-3 of adenine are more likely. It is also possible, because of some recent results, that interaction of the whole molecule, following activation, is an initial reaction of nitrosamines, and possibly even of nitrosamides. Intramolecular rearrangements during replication of the DNA might then lead to fixation of the lesion as a mutation. The relation of the reaction of nitroso compounds with nucleic acids to carcinogenesis is less decisive, and it cannot be said that any good correlations, withstanding rigorous tests, have been established. That nitroso compounds interact with nucleic acids in d u o there is no doubt; however, there is no quantitative correlation between the extent of inter-

CARCINOGENIC AND MUTAGENIC

N-NITROSO COMPOUNDS

267

action of compounds of similar carcinogenic activity with nucleic acids of target and nontarget organs. Nor is there correlation, so far as it has been tested, between susceptible and relatively nonsusceptible species. When attention is focused on alkylation at specific nucleic acid sites, the correlation becomes even less certain. This is true whether the now less favored N-7 alkylation of guanine or the more favorable Oa alkylation of guanine is considered. Too few good comparisons have been made, and undue attention has been paid to dimethylnitrosamine and diethylnitrosamine, two compounds that are liver carcinogens and readily metabolized in the liver to a variety of reactive toxic and carcinogenic products, some of which have been easy to detect as, for example, alkylated nucleic acid bases. The difficulties in finding similar products from equally carcinogenic cyclic nitrosamines has been overlooked. The existence of a deuterium isotope effect in carcinogenesis by nitrosamines implies that one of the initial steps in carcinogenesis is loss of a hydrogen atom alpha to the nitroso function. This is explainable in terms of formation of a simple alkylating agent from dimethylnitrosamine ( I ) (29),but much more difficult to explain in this way for nitrosomorpholine (XVII) ( 3 0 ) . However, other types of interaction of the latter with nucleic acids (or other macromolecules) are certainly compatible with the lesser carcinogenicity of the deuterium-labeled compound. Such interactions could damage DNA, and differences in rates of repair that have been found are conceivable as responsible for the differential carcinogenic effects of various nitroso compounds. There are indications that there is a deuterium isotope effect in carcinogenicity of nitrosamides (Lijinsky et al., unpublished data), which suggest that these, too, require activation for carcinogenic activity and would rule out simple transfer of an alkyl group (by supposedly direct action) to nucleic acids as involved in carcinogenesis by them. Much remains to be done to establish a relationship between reaction of nitroso compounds with nucleic acids and carcinogenesis (if there is such a relationship ). Correlations must be made with organ specificities and with differences in biological effect in different species. Minor differences in rate of some reaction (such as disappearance of an alkylated base from DNA in several organs) must be related with sharp, all-ornothing differences in tumor response of one organ versus another. The latter makes it not unlikely that the differences in chemical reactions between susceptible and nonsusceptible organs are not marginal, but that the reactions leading to neoplastic transformation occur in susceptible organs and not in others. This will be revealed only by ensuring that the negative studies in organs in which tumors do not appear receive equal weight with the positive correlations.

268

W. LIJINSKY

REFERENCES 1. P. D. Lawley and C. A. Wallick, Chem. lnd. (London) 633 (1957). 2. P. N. Magee and E. Farber, BJ 83,114 ( 1962). 3. P. N. Magee and T. Hultin, BJ 83, 106 ( 1962). 4. H. Druckrey, R. Preussmann, S. Ivankovic and D. Schmiihihl, Z . Krebsforsch. 89, 103 ( 1967). 5. P. N. Magee and K. Y. Lee, BJ 9 1 , s ( 1964). 6. R. Preussmann, H. Druckrey, S. Ivankovic and A. von Hodenberg, Ann. N.Y. Acad. Scf. 183, 697 ( 1969). 7 . C . M. Goodall, W. Lijinsky, L. Tomatis and C. E. M. Wenyon, Toxicol. Appl. Pharmacol. 17, 426 (1970). 8. W. Lijinsky and A. E. Ross, J. Nat. Cancer Inst. 42, 1095 ( 1969). 9. W. Lijinsky, L. Keefer, 1. Loo and A. E. Ross, Cancer Res. 33, 1634 (1973). 10. M. F. Argus, J. C. Arcos and C. Hoch-Ligeti, J. Nat. Cancer Inst. 35, 949 (1965). 11. B. Singer, This Series 15,219 (1975). 12. S. Takayama, N. Kuwabara, Y. Azama and T. Sugimura, 1. Nat. Cancer Inst. 48, 973 ( 1971). 13. T. Sugimura and S. Fujimura, Nature (London) 218,943 ( 1967). 14. T. Sugimura, N. Tanaka, T. Kawachi, K. Kogure, S. Fujimura and Y. Shimosato, GANN 82, 67 ( 1971). 15. F. Hoffmann and A. Cra5, Arch Geschwulstforsch. 28,89 (1966). 16. A. Graffi and F. Hoffmann, Arch. Geschwulstforsch. 28, 234 (1966). 17. J. A. Swenberg, A. Koestner and W. Wechsler, Lab. lnoest. 28, 74 ( 1972’1). 18. W. Wechsler, P. Kleihues, S. Matsumoto, K. J. Ziilch, S. Ivankovic, R. Preussmann and H. Druckrey, Ann. N.Y. Acad. Sci. 159,360 (1969). 19. S. Ivankovic and H. Druckrey, Z . Krebsforsch. 71,320 (1968). 20. W. Lijinsky, H. Garcia, L. Keefer, J. Loo and A. E. Ross, Cancer Res. 32, 893 (1972). 21. S . Odashima, GANN 80,237 ( 1969). 22. W. Lijinsky and H. W. Taylor, 2.Krebsforsch. 83, 315 ( 1975). 23. R. Elespuru and W. Lijinsky, Fd. Cosmet. Toxicol. 11, 807 (1973). 24. L. Pasternak, Arzneimittel-Forsch. 14, 802 ( 1964). 25. R. Schoental and P. N. Magee, Brit. J. Cancer 18,92 ( 1962). 26. H. Druckrey, S . Ivankovic, J. Bucheler, R. Preussmann and C. Thomas, Z . Krebsforsch. 71, 167 (1968). 27. R. Elespuru, W. Lijinsky and J. K. Setlow, Nature (London) 247,386 (1974). 28. D. Siebert and C . Eisenbrand, Mutation Res. 22, 121 ( 1974). 29. L. Keefer, W. Lijinsky and H. Garcia, J. Nut. Cancer Inst. 51, 299 (1973). 30. W. Lijinsky, H. W. Taylor and L. Keefer, J. Nut. Cancer Inst. In press. 31. W. Lijinsky and H. W. Taylor, Int. J . Cancer 18,318 ( 1975 ). 32. W. Lijinsky and H. W. Taylor, Cancer Res. 35,2123 ( 1975). 33. W. Lijinsky and H. W. Taylor, Cancer Res. 35, 1270 ( 1975). 34. W. Lijinsky and H. W. Taylor, J . Nut. Cancer Inst. 55,705 ( 1925). 35. H. Garcia and W. Lijinsky, Z . Krebsforsch. 77,257 (1972). 36. W. Lijinsky and H. W. Taylor, in preparation. 37. W. Lijinsky and H. W. Taylor, Cancer Res. 35,3209 (1975). 38. H. Garcia and W. Lijinsky, Z . Krebsforsch. 79, 141 (1973). 39. P. D. Lawley, This Series 5, 89 ( 1966). 40. A. Loveless, Nature (London) 223,206 ( 1969).

CARCINOGENIC AND MUTAGENIC N-NITROSO COMPOUNDS

269

41. P. D. Lawley and C. J. Thatcher, BJ 116,693 ( 1970). 42. R. Parkin, H. B. Waynforth and P. N. Magee, Mutation Res. 21, 155 (1973).

43. P. D. Lawley and C. N. Martin, BJ 145,85 (1975).

44. N. Loprieno, R. Barale, C. Bauer, S. Baroncelli, G. Bronzetti, A. Cammellini,

A. Cinci, G. Corsi, C. Leporini, R. Nieri, M. Nozzolini and C. Serra, Mutation Res. 25, 197 (1974). 45. A. Loveless, Proc. Roy. SOC., Ser. B 150,497 (1959). 46. U.Schulz and D. R. McCalla, Can. J . Chem. 47,2021 (1969). 47. P. D. Lawley, D. J. O m and S. A. Shah, Chem-Bid. Interact. 4, 431 (1972). 48. L. L. Gerchman and D. B. Ludlum, BBA 308,310 (1973). 49. B. Singer, Bchem 11,3939 (1972). 50. R. Elespuru, unpublished data. 51. S. Neale, Mutatton Res. 14, 155 (1972). 52. R. Sussmuth, R. Heerlin and F. Lingens, BBA 269,276 ( 1972). 53. R. F. Gomez, M. Johnston and A. J. Sinskey, Mutation Res. 24, 5 (1974). 54. E.Zeiger and M. Legator, Mutation Res. 12,469 (1971). 55. B. N. Ames, W. E. Durston, E. Yamasaki and F. D. Lee, PNAS 70,2281 (1973). 56. H. Bartsch, C. MalaveiUe and R. Montesano, Cancer Res. 35, 644 (1975). 57. H. V. Malling and C. N. Franz, Mutation Res. 25, 179 ( 1974). 58. E.Zeiger, M. S. Legator and W. Lijinsky, Cancer Res. 32,1598 (1972). 59. P. N. Magee and J. M. Barnes, J. Puthol. Bacterial. 84, 19 (1962). 60. P. D. Lawley, Mutdjon Res. 23,283 (1974). 61. L.Sun and B. Singer, Bchem 14, 1795 ( 1974). 62. T.Saito and T. Sugimura, GANN 64,537 ( 1973). 63. V. M.Craddock, BJ 106,921 (1968). 64. J. D. Regan and R. B. Setlow, in “Chemical Mutagens, Principles and Methods for Their Detection” (A. Hollaender, ed.), Vol. 3, p. 151. Plenum Press, New York, 1973. 65. J. D. Regan, R. B. Setlow, R. D. Blevins and W. Lijinsky, Proc. Ames. ASS. Cunces Res. 16, 85 ( 1975). 66. J. D. Regan, R. B. Setlow, A. A. Francis and W. Lijinsky, Mutation Res. In press. 67. P. N. Magee and J. M. Barnes, Brit. J . Cancer 10, 114 (1956). 68. K. Y. Lee, W. Lijinsky and P. N. Magee, J. Nut. Cancer Inst. 32, 65 (1964). 69. P.F. Swann and P. N. Magee, BJ 110,39 (1968). 70. A. E. Ross, L. Keefer and W. Lijinsky, J . Nut. Cancer Inst. 47, 789 ( 1971). 71. F.W. Kriiger, Z . Krebsforsch. 76, 145 ( 1971). 72. K. Y. Lee and W. Lijinsky, J . Nut. Cancer Inst. 37,401 ( 1966). 73. B. W. Stewart, P. F. Swann, J. W. Holsman and P. N. Magee, Z. Krebsforsch. 82, 1 (1974). 74. F. W. Kruger, in “Topics in Chemical Carcinogenesis” (W. Nakahara, S. Takayama, T. Sugimura and 0. Odashima, eds.), p. 213, Univ. of Tokyo Press, 1972. 75. C.Grandjean, personal communication. 76. C . M. Goodall, W. Lijinsky and L. Tomatis, Cancer Res. 28, 1217 (1968). 77. W. Lijinsky, J. Loo and A. E. Ross, Nature (London) 218, 5174 (1968). 78. R.Goth and M. F. Rajewsky, Z. Krebsforsch. 82,37 ( 1974). 79. J. W.Nicoll, P. F. Swann and A. E. Pegg, Nature (London) 254,261 (1975). 80. R. Goth and M. F. Rajewsky, PNAS 71, 639 (1974). 81. I. Damjanov, R. Cox, D. S. R. Sarma and E. Farber, Cancer Res. 33, 2122 (1973). 82. B. W. Stewart and E. Farber, Cancer Res. 33,3209 ( 1973).

This Page Intentionally Left Blank

Biochemistry and Physiology of Bacterial Ribonucleases ALOKK. DATTA'AND SALE K. NIYOCI The University of Tennessee-Oak Ridge Graduate School of Biomedical Sciences and Biology , Division Oak Ridge National hboratoryl Oak Ridge, Tennessee

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

I. Introduction. 11. A General Survey of Different Ribonucleases Present in Bacteria A. E. coli Ribonuclease I. . . . . . . . . B. Polynucleotide Phosphorylase . . . . . . . C . E. coli Ribonuclease I1 . . . . . . . . D. E . coli Ribonuclease I11 . . . . . . . . E. Ribonuclease H . . . . . . . . . . F. Ribonuclease IV . . . . . . . . . . G. Ribonuclease P and Other Ribonucleases . . . . H. Oligoribonuclease. . . . . . . . . . I. Ribonucleases from Other Bacteria . . . . . . 111. Role of Ribonucleases in Cell Physiology . . . . . A. Messenger RNADegradation and Processing . . , . B. Ribosomal RNA Processing . . . . . . . C. Transfer RNA Processing . . . . . . . . D. Stable RNA Degradation under Conditions of Starvation , IV. Regulation of Ribonuclease Activity . . . . . . V. Physiology of Ribonuclease-Minus Mutants . . . . . VI. Concluding Remarks. . . . . . . . . . References . . . . . . . . . . . . Notes Added in Proof

.

.

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

271 272 272 277 279 28 1 282 284 284 285 287 288 289 293 295 296 296 298 300 301 308

1. Introduction One of the important branches of molecular biology at the present time is the study of the synthesis, structure and function of ribonucleic acid. During the past decade, enormous progress has been made in defining the enzymic events in RNA biosynthesis and its regulation (1-5). But in spite of all these advances in the area of RNA synthesis, there is still much uncertainty about the precise roles that various riboPostdoctoral Investigator supported by subcontract No. 3322 from the Biology Division of the Oak Ridge National Laboratory to the University of Tennessee. Operated by Union Carbide Corporation for the U.S. Energy Research and Development Administration. 271

272

ALOK K. D A m A AND SALIL K. NIYOGI

nucleases play within the cell, although it is obvious that these enzymes degrade different types of cellular RNAs. The greatest impetus in recent times to the study of RNA catabolism was the demonstration of the extremely unstable nature of mRNA in Eschedchia coli ( 6 ) .The high turnover rate of mRNA can be explained only by the involvement of specific enzymes. Subsequently, many workers discovered several enzymes and implicated them in mRNA degradation, but still there is lack of consensus as to which enzymes participate in this process. Furthermore, the involvement of nucleases in the posttranscriptional processing of rRNA and tRNA and in their degradation under conditions of starvation have made this field of research quite challenging. The purpose of this review is to examine the information presently available with respect to the various ribonucleases elaborated by bacteria. Attention is directed mainly to a single species, Escherichia coli, simply because most systematic studies have been carried out on this organism. However, the discussion is extended to other organisms whenever possible. Table I lists in a comprehensive form the different ribonucleases known to be present in E. coli. [An extensive list of different nucleophosphodiesterases of different organisms has been presented by Laskowski (7).]

II. A General Survey of Different Ribonucleases Present in Bacteria A. E. coli Ribonuclease I3 This endonuclease catalyzes the breakdown of ssRNAs to oligonucleotides terminating in a 3’-phosphate, and finally to 3’-mononucleotides with 2’:3’( cyclic) nucleotides as intermediate products. Elson (8) discovered t h i s ‘latent” RNase in ribosomes of E . coli and showed that the enzyme can be detected only when the ribosomes are treated with 4 M urea, salt, EDTA, or any reagent that causes disintegration of the ribosomes.‘ Under these conditions, the enzyme becomes active and rapidly degrades the rRNA ($11). The name “RNase I” was given to it, as the activity was the first of its kind found in E. c01i.~ Several observations support the view that the enzyme may be an integral part of the ribosome and thus responsible for the high turnover of mRNA (12, 13). However, an estimation of the amount of enzyme yielded ( 1 4 ) a value of 0.1 mol per mole of 70 S ribosomes (MW The prefix E . coli is necessary to distinguish this enzyme (ribonuclease 11, EC 3.1.4.23) from ribonuclease I (pancreatic RNase, EC 3.1.4.22). ‘See article by Spitnik-Elson and Elson in this volume.

BACTERIAL FUBONUCLEASES

273

taken as 2.8 X lo6), assuming the molecular weight of RNase I to be the same as that of pancreatic RNase. Another report indicated between 0.1 and 1 mol per ribosome ( 1 5 ) . Thus it could not then be argued convincingly that RNase I is an integral part of E. coli ribosomes unless the unlikely assumption was made that only 10%of the ribosomes are functionally active at any tim,e ( 1 5 ) . Later it was definitely proved that the association of RNase I with isolated ribosomes is an artifact occurring after cell disruption (16-20); RNase I, originally present in the periplasmic space, is adsorbed to the 30 S subunit of the ribosome. Neu and Heppel (16-18) observed that ( a ) washed ribosomes can adsorb a large amount of RNase I and ( b ) when spheroplasts are formed, or when cells are given cold shock treatment, RNase I and many other enzymes-namely, cyclic phosphodiesterase, 5’-nucleotidase and alkaline phosphatase, all present in the periplasmic space (21, 22)-are released into the outside medium leaving m.any other enzymes inside the cell ( 1 9 ) . It is interesting that the released enzymes are located in the same place as RNase I. External, nonpenetrating substrates for RNase I can therefore by hydrolyzed to nucleosides, which can then enter the cell and be utilized. RNase I is also released on treatment with a membraneactive polypeptide antibiotic that penetrates only up to the periplasmic space ( 2 3 ) . The precise cause of the inactivity (latency) of RNase I when associated with ribosomes has not been established with certainty. It is believed that polyamines (putrescine, cadaverine and spermidine ) ,5 present in relatively high concentrations of E. coli ribosomes (24, 2 5 ) , mask the enzyme activity by stabilizing the ribosomes ( 2 5 ) . The association of RNase I with the ribosomes is Mg2+-dependent ( 2 6 ) . Also, RNase I bound to the ribosomes in the presence of suitable amounts of Mg2+ remains in the latent stage ( 2 6 ) .When the Mg” concentration is reduced to very low levels (or better, when EDTA is present), RNase I becomes highly active and disrupts the ribosomal organization by breaking down the rRNA, the core material for the structural organization of ribosomal proteins (27, 2 8 ) . Ribosomal particles deficient in protein do not mask the RNase I activity, but when ribosomal particles are reconstituted with previously detached ribosomal proteins, the inhibitory property is regained ( 2 6 ) . These results suggest that overall ribosomal organization is essential for the Mg2+-dependentinhibition of RWase I (26). This ribosome-bound ribonuclease has been extensively purified from E. coli ( 1 4 ) and purified to homogeneity from Salmonella typhimurium (26, 29). The biochemical properties of the enzyme from both sources are almost identical. Although its physical properties have not been ‘See article by Sakai and Cohen in this volume.

274

ALOK K. DA'ITA AND S A L E K. NIYOCI

TABLE RIRONUCLEASEB OF Enzyme 1. E. coli RNase I

pH optimum

Divalent cation requirement

Monovalent cation activator

8.1

None

None

Naturnl or synthetic ssRNAs

8.1

None

Natural or synthetic ssRNAs

7-7.5

Usually Mg*+;in some mutants the altered enzyme requires Mn*+ instead Mg*+and Mn*+

K+ and NHI+

Nonhelical ssRNAs

>O

Mg*+ and Mn*+

Absolute requirement of K+ and NHI+

Natural or synthetic dsRNAs

None

None

R 1 7 viral RNA

(RNase II),b-c EC 3.1.4.23 2. Polynucleotide

phospborylase. EC 2.7.7.8 3. E. coli RNase I1

Substrate specificity

(Exoribonuclease (111s) EC 3.1.4.24 4. E. coli RNase 111

(Endoribonuclease Ill)' EC 3.1.4.24

5. RNase

IV

6

(single-stranded form) 6. RNase H (Hybrid

-

Hydrolyzes the RNA chain of DNA RNA hybrids

Mg*+

nuclease),' EC 3.1.4.34 7. RNase

P

8. RNase Pt

0. RNase PI11 (PI)

Mn'+, Mg'+

8

-

K+ and NH4+

-

-

8.2

K+,N H P , Me*+

7.5

Mg'+

NHa+

Mn*+

None

10. Colicin E3

11. RNase M

12. Oligoribonuclease

7-8

(oligonucleotidase),

EC 3.1.4.10 0

ss

-

single-stranded, ds

-

Precursor to tRNAs Multimeric precursor tRNA molecules

Precursor tRNA with extra nucleotides a t the 3' end Cleaves 16 S rRNA of ribosomes engaged in protein synthesis Precursor 16 S rRNA Specific for short oligoribonucleotides

double-stranded.

' Names in parentheses are thhse recommended in Enzyme Nomenclature. 1072.

Identical in action to RNase T2. Plant RNase (see Enzyme Nomenclature, 1972).

275

BACTERIAL RIBONUCLEASES

I Escherichia colia Mode of action

Products

Other information

15, 17-90 Located at. the periplasmic space of the cell b u t becomes attached specifically to the 30 S ribosomal subunit when cells are broken Located on the ribosomes 45-48,65, 70 and not necessary for cell function

Endonuclease, prodacing 2’: 3’(cyclic) nucleotides as intermediates

3‘-Mononucleotides

Processive exonuclease, hydrolyzing from 3’ toward 5’ end

Nucleoside 5’-diphosphates

Processive exonuclease, hydrolyzing from 3’ toward 5’ end

5’-Mononucleotides, including short, resistant oligonucleotides

Endonuclease (see Section 11, D)

Oligonucleotides of 10-25 nucleotide chain length

Endonuclease, cleaves a t a specific but unknown position

15 S fragment carrying 5’ terminus of the original molecule, and 21 S fragment lacking 5‘ end Physiological role uncertain; Produces mono- as well as suggested that it removes oligonucleotides carrying RNA primers of newly 5’-phosphate and 3’formed DNA chains hydroxyl termini Mature tRNAs when acts in Associated with ribosomes, but detaches in concenconjunction with RNase I1 trated salt solutions Acts in conjunction with Two products containing RNase P in the removal mature tRNA sequences of extra 5‘-sequence of the (I and 11) and two smsll precursor tRN.4 molecule oligonucleotides from the 41-nucleotide leader sequence of the precursor tRNA with matured 3’ end

Endonucleaae

Endonuclease. cleaves a t a specific position Endonucleolytic cleavage a t specific positions in tho spacer regions

Attached to the ribosomes, 80-82,85. 89 but less 80 than is RNase I; indispensable for cell survival 106,108 Physiological role not fully understood, but i t participates in the processing of T 7 “early messenger RNA“ for subsequent efficient translation; also involved in the processing of stable ribosomal RNAs Nature of the bonds cleaved 197 and physiological function unknown

-

Removes the extra nucleotides a t the 3’ end in a n unknown manner

Not known

Mature 16 S rRNA

Processive exonuclease, hydrolyzing from 3‘ toward 5’ end

5‘-Mononucleotides only

References

119-fB6, 126-169 157.254

258

262,663

Requires the 50 S ribosomal 160,161 subunit for action; may activate some nuclease on the ribosomes Associated with ribosomes; 186 separates during high salt ribosomal wash In cytoplasm; has been pre- 166-160 pared free of all the above nucleases

276

ALOK K. DATTA AND SALIL K. NIYOGI

studied in detail, it appears to be a low-molecular-weight basic protein with but a single subunit. It is relatively heat-stable, much like the classical pancreatic RNase ( EC 3.1.4.22). However, the ribosomal E . coli RNase I (EC 3.1.4.23) does differ from the pancreatic enzyme in its specificity. Unlike the latter, which is a phosphodiesterase specific for phosphoric esters of pyrimidine nucleoside 3’-phosphates, the E . C O Z ~ enzyme attacks all bonds in RNA endonucleolytically. In this respect, E . coli RNase I3 behaves like RNases found in other sources like ryegrass ( 3 0 ) , tobacco leaf (31, 32) and pealeaf ( 3 3 ) , which are classified as RNase I1 (EC 3.1.4.23) and known also as plant RNase and RNase T2. Since E. coli spheroplasts are practically free of RNase I activity, they were employed ( 3 4 ) to test whether RNase I has any effect on the rate of gross breakdown of pulse-labeled RNA; it was found that the rate of breakdown was the same as in intact cells. This was also supported by the experiments with mutant strains of E. coli deficient in RNase I ( 3 5 ) . Determinations of the active half-life of the mRNA of induced p-galactosidase in several strains of E. coli deficient in both RNase I and polynucleotide phosphorylase confirmed the earlier observations, indicating that RNase I is not required for the metabolic inactivation of mRNA (36, 3 7 ) . Later on, two kinds of enzymes very similar in their properties to RNase I, were discovered. One appears in the acid-soluble fraction when cells are treated with perchloric acid (38), and the other can be found in the sedimented cell debris (39). Both are very similar to E. coli RNase I but differ from it in substrate specificities and chromatographic properties, or both. But RNase I is actually located in three different cellular fractions: ( a ) ribosomes, ( b ) debris and ( c ) shock fluid ( 4 0 ) . The chromatographic behavior of the enzyme on the ion-exchanger CG-50 is dependent upon the procedures used in the purification, not upon the method of extraction from the cell. Since, in a mutant strain that lacks RNase I, the activities of these two enzymes were also undetectable, Gesteland ( 3 5 ) proposed that all three enzymes may well be controlled by a single gene, that is, they originate as one and the same gene product. In connection with the masking of RNase I by the 30 S subunit of ribosomes, it should be mentioned that not only ribonucleoprotein particles such as ribosomes, but also single-stranded and native DNAs bind to RNase I ( 4 1 ) . In this regard, E. coli RNase I behaves like pancreatic RNase (42, 4 3 ) . The nature of this binding may also be similar (42, 4 3 ) . It has been suggested (44)that this binding is due in part to electrostatic interaction and in part to an interaction of the enzyme with pyrimidine nucleotides. The inhibition of RNase I due to

BACTERIAL RIBONUCLEASES

277

its binding with DNA indicates that the binding site for DNA is perhaps at or close to the active site of the enzyme ( 4 4 ) . It has also been concluded that the same pyrimidine sites and the same electrostatic and hydrophobic interactions are involved in binding both RNA and DNA, although the enzyme hydrolyzes RNA, but not DNA.

B. Polynucleotide Phosphorylase After the discovery of this enzyme (EC 2.7.7.8) in Azotobacter agilis ( 4 5 ) , it was detected in E . coli ( 4 6 ) . It catalyzes the following reaction

The reaction is Mg“-dependent, but in certain mutants of E . coli the enzyme prefers Mn2+ over Mg2+for its divalent-cation requirement ( 47-49). However, under certain conditions, deoxyribonucleotides can also be incorporated into polyribonucleotide chains by polynucleotide phosphorylase ( 5 0 5 2 ) . Depending on the concentration of inorganic phosphate in the medium, the enzyme can act in both directions; i.e., when the concentration of Pi is high the reaction is phosphorolytic, whereas at lower concentrations of Pi, polymerization results. Two kinds of polynucleotide phosphorylases can be obtained from the same strain of Micrococcus Zuteus cells depending on the type of cells used ( 5 3 ) . This observation led to the discovery of two forms of the enzyme, primer-dependent (form T ) and prirner-independent (form I ) . Various treatments that change the conformation of proteins (urea, trypsin, guanidinemHC1) convert the form I to form T (53, 5 4 ) . The change is accompanied by the loss of approximately 15%of the original enzyme molecular weight (54, 5 5 ) . Furthermore, the single, smaller species (form T ) catalyzes de novo polymerization with great difficulty and is markedly stimulated by the addition of oligonucleotides as primers (56). The most highly purified preparations of polynucleotide phosphorylase also carry out an exchange reaction between nucleoside 5’-diphosphates and Pi under conditions where no polymer formation is detectable (46, 5 7 ) . The relationship of the exchange reaction to the mechanism of the polymerization or phosphorolysis reactions is not yet known. The formation of poly ( U ) is strongly inhibited in the presence of pancreatic RNase, whereas the rate of exchange of Pi with UDP is not. Pi-exchange also occurs rapidly with GDP under conditions where no accumulation of poly(G) can be detected. Exchange reactions also proceed without detectable phosphorolysis.

278

ALOK K. DATTA AND SALIL K. NNOGI

In addition to phosphorolysis, polymerization and exchange reactions, polynucleotide phosphorylase catalyzes the transfer of nucleoside 5’monophosphate units from one polynucleotide to another (58). Extensive reviews describe the mechanism of action of this enzyme in detail (59, 60). Since we are mainly concerned here with the degradative role of this enzyme, we confine our attention to its nucleolytic activities. Although there are two forms of polynucleotide phosphorylase, as described above, both forms of enzymes behave identically (61) in their degradative capacities. Polynucleotide phosphorylase is found mainly in the supernatant fraction of cell-free extracts, although there are reports of its association with ribosomal particulate fractions (62, 63). This enzyme was first thought to be responsible for RNA synthesis because of its efficiency in catalyzing the synthesis of polynucleotides, but it is now believed that its in uiuo role is probably catabolic (59). The arguments that favor a catabolic role are: ( a ) it is not found in most animal tissues and its reaction favors the phosphorolysis reaction at the intracellular levels of phosphate; ( b ) it has no informational mechanism to determine the specific nucleotide sequence in a polymer; ( c ) the K, values for the ribonucleoside diphosphate substrates is of the order of 0.02 M, which is much higher than the intracellular concentration. Polynucleotide phosphorylase prefers single-stranded RNAs for its phosphorolytic activity ( 6 4 ); double-stranded structures are not attacked ( 65). It catalyzes the phosphorolysis of both oligoribonucleotides and polyribonucleotides to yield ribonucleoside diphosphates. The mechanism of phosphorolysis is exonucleolyt:c, starting at an unesterified 3’-OH group (66-68). Polynucleotide phosphorylase from either E . coli (69) or M . luteus ( 7 0 ) degrades long polynucleotides by a processive mechanism; the enzyme tends to degrade a single chain to completion (nucleoside diphosphates and resistant dinucleotides ) before initiating phosphorolysis of another chain. On the other hand, the phosphorolysis of the oligonucleotide PA-A-A-U by the E . coli enzyme is not processive, and PA-A-A and ppU (UDP) are the major initial products (67). Somewhere between a chain length of 6 and 70, the mechanism of degradation shifts from the random to the processive mode ( 7 1 ) . From the kinetic analysis of both types of phosphorolysis, Chou and Singer (71 ) and Chou et al. (72) suggested that the phosphorolysis of oligonucleotides by polynucleotide phosphorylase proceeds with the so-called “rapid equilibrium random Bi-Bi” mechanism ( 73, 7 4 ) . Such mechanisms have also been proposed for the polysaccharide phosphorylases from E . coli (75) and from liver (76). It will be very interesting to see whether this mechanism is a general property of all polymer phosphorylases.

BACTERIAL RIBONUCLEASFS

279

I t has been postulated (77, 78) that the changeover from processive degradation of polynucleotides to nonprocessive degradation of oligonucleotides with the E . coli enzyme arises from the existence of two binding sites in the enzyme. The first, subsite I, is the catalytic center and binds either polynucleotides or oligonucleotides at the 3’-OH end. When the substrates are long, they can reach a second site, subsite 11, which is not involved in the phosphorolysis of oligonucleotides. This dual attachment of the substrate to the enzyme allows polymers to snap back to a reactive position after removal of one unit and thereby be degraded in a processive manner. Thus, the binding to subsite I1 is responsible for the marked enhancement in binding of long polymers compared to short oligonucleotides. Oligonucleotides, not being anchored at subsite 11, are lost into the medium after the reaction and released from subsite I to allow another substrate molecule to get into the active site, leading to nonprocessive mode of degradation.

C.

E. coli Ribonuclease 116 Wade et al. (79, 80) first suggested that E . coli ribosomes contain, in addition to RNase I, another RNA depolymerizing activity that, unlike the former, requires magnesium ion for its activity. When this RNase was actually found in E . coli, it was designated RNase IIB(81). Subsequently, it was purified extensively and its properties have been studied in detail (82). It requires divalent cation and is activated by K+ and NH,+. Similar activities have been purified from S. typhimuriurn ( 8 3 ) and Lactobacillus plantarum ( 8 4 ) . In contrast to E. coli RNase I, E. coli RNase I1 is found in both the high-speed supernatant and particulate fractions ( 8 5 ) . The enzyme activities purified from the two fractions are identical in such properties as ion requirement, p H optimum and mechanism of action. The modes of attachment of RNases I and I1 to ribosomes seem to be different. The binding of I1 to ribosomes has been shown to be very weak, and the enzyme is detached from the ribosomes on repeated washing ( 8 4 ) and during sucrose density-gradient ultracentrifugation of crude ribosomes (86). Unlike I, this association with ribosomes does not inhibit its activity. In contrast to RNase I, RiVase I1 is an exonuclease and hydrolyzes from the ?-end to 5’-mononucleotides (85). This degradation, unlike snake venom phosphodiesterase ( 87), takes place in a processive manner; i x . , a givcn enzyme molecule hydrolyzes a single RNA molecule repeatedly until its 5’-OH chain end is released as a small, resistant oligonucleotide ( n = 2-4); only then does it begin to hydrolyze another RNA

’ Exorihonuclease

(EC 3.1.4.20).

280

ALOK K. DAlTA AND SALK K. NNOGI

molecule (88, 89). Thus, the mode of degradation of substrate by RNase I1 differs from the classical Michaelis-Menten mode of enzyme action, which postulates that the enzyme should release its products-in this case, polymer and mononucleotide-and become a free enzyme after each catalytic step. In this respect, the mechanism of action of RNase I1 is identical to that of polynucleotide phosphorylase and differs from the random mechanism of those exonucleases that dissociate from the polynucleotide after hydrolysis of the terminal mononucleotide and then choose the next phosphodiester bond to be hydrolyzed more or less randomly from other molecules in the solution. Like pancreatic RNase and E. coli RNase I, E . coli RNase I1 is also inhibited by deoxyoligonucleotides (90). Eley and Crawford (91 ) found in extracts of E . coli an RNase activity that degrades singlestranded polyribonucleotides to 5’-mononucleotides and that is inhibited by DNA. These authors considered the activity as one different from RNase 11, but it was later demonstrated that RNase I1 was responsible for the activity (90). Venkov et al. (92) reported that E. coli RNase IT action on poly ( U ) is competitively inhibited by tRNA and noncompetitively by Na+, but whether the inhibition of tRNA arises from its secondary structure or not remains to be known. At low substrate concentrations, ATP was also found to be inhibitory, but it was later demonstrated that what was thought to be an inhibition of RNase I1 by ATP (93, 94) arose from the formation of less-soluble nucleoside diphosphates, as compared to mononucleotides, by a contaminating enzyme, adenylate kinase (EC 2.7.4.3) which converts product AMP to ADP in the presence of ATP (94). When ribosome function is blocked by antibiotics like chloramphenicol, mRNA synthesis continues but degradation of mRNA is prevented ( 9 5 ) . This led Kuwano et al. (96) to consider that mRNA breakdown might be coupled to continued protein synthesis. Working with this idea, they reported a new enzyme in a mutant of E . coli lacking RNase I activity and in which RNase I1 was presumably heat-labile (97). They claimed that this activity is coupled to protein synthesis and is responsible for mRNA degradation in uitro’ (96, 98), and named it RNase V. They also reported (99)that this enzyme degraded mRNA exonucleolytically from the 5’ end to produce 5’-mononucleotides and that the activity depends on both 30 S and 50 S subunits of the ribosome. The enzyme did not degrade ribosomal RNA and the activity required G and T factors, tRNA, K’, Mg2+,GTP and reducing compounds (100). However, reports from two other laboratories (101, 102) cast doubt on the existence of this postulated ribonuclease. According to these reports, the activity could be that of RNase 11, which becomes stable

BACTERIAL RIBONUCLEASES

281

to heat treatment in the presence of ribosomes. The earlier workers (96-98) had used a mutant of E . coli that has a thermolabile RNase I1 activity, and they had presumed that the activity was destroyed by heat treatment. So at present there is no evidence for an exoribonuclease in bacterial systems that degrades RNA chains from the 5'- to the 3'-direction. A quite different RNase, designated "RNase V" of E. coli, has been described ( 103). It hydrolyzes ribosomal RNA, 5 S RNA, poly( U ) , poly( C ) and R17 RNA, but not transfer RNA. Although no firm conclusions can be drawn on the involvement of E. coli RNase I1 in messenger RNA degradation, results from several laboratories suggest that RNase I1 is one of the enzymes responsible for mRNA degradation within the cell. In another section of this review (Section 111), the role played by RNase I1 in cell physiology is discussed in detail.

D. E. coli Ribonuclease Ill An enzyme from E . coli with endonucleolytic activity against doublestranded RNAs was reported first by Robertson et d. (104) and designated as RNase I11 (endoribonuclease 111, EC 3.1.4.24). Subsequently, it was found in other organisms (105). The activity of the enzyme remains attached to ribosomes within osmotically shocked cells. The activity sediments with the ribosomes in <0.2M NH,CI, but higher concentrations of salt detach it, thus affording a route to purification (106). RNase I11 shows an absolute requirement for divalent cations including Mg?+ and Mn2+,and for monovalent cations like NH,', K" and Na'; Mg" is essential for activity, but K+ is an activator (]Or), suggesting that conformational changes in the RNA substrate molecules are responsible for the lower activity at low salt concentration. The enzyme is somewhat stable in acidic media. Recently, purification of RNase I11 to homogeneity has been reported (108). It has also been demonstrated that purified RNase I11 has a molecular weight of about 44,000 with two identical subunits. The mechanism of action of this enzyme has been studied in detail by Schweitz and Ebel (109). The enzyme seems to be endonucleolytic in its action, and the internucleotide cleavage appears to yield a 3'-phosphate and a 5'-hydroxyl group without the intermediary formation of 2':3'-cyclic phosphates. The activity is unique in the sense that on exhaustive hydrolysis of double-stranded polyribonucleotides, such as poly( I - C ) and poly( G - C ) , it produces oligonucleotides of chain length of about 10-25, the average being 16. The mechanism of recognition of the secondary structure of the substrate by the enzyme is not yet fully understood. Base specificity and base-stacking interactions may

282

ALOK K. DATI'A AND S A L E K. NNOGI

play important roles (110, 1 1 1 ) . The fact that double-stranded DNA appears to be neither a substrate for RNase I11 nor a competitive inhibitor of double-stranded RNA suggests that RNase I11 can distinguish among various helical forms of double-stranded polyribonucleotides. Several workers ( 112-116) have postulated that the structure of double-stranded RNA is more closely related to the A-helix than to the B-helix form of DNA. The A-form of DNA was first described by Franklin and Goslin (117) as occurring at low relative humidities, in contrast to the B-form normally found at high relative humidities. Since all the natural substrates for RNase I11 may be in such A-type helices, it is possible that part of the specificity of RNase I11 resides in its ability to recognize structures similar to the A-helix. Regarding the mode of action of the enzyme on double-stranded RNAs, RNase I11 can act in two ways: single-hit mechanism, similar to that of acid DNase; or a double-hit mechanism, like pancreatic DNase. To determine which of the mechanisms is operative, Schweitz and Ebel (109) subjected (rI)". ( [3H]rC),, to mild RNase I11 hydrolysis. The hydrolyzate and control samples incubated without RNase I11 were analyzed by gel electrophoresis in the presence and in the absence of urea. In the absence of urea, migration of the radioactivity was faster for the treated sample than for the control, whereas in the presence of urea, migration speed was similar for both treated and untreated samples. It was concluded (109) that the breaks took place on both strands of the molecule, either simultaneously on both strands, i.e., by a "singlehit" mechanism, or successively on each strand at the same level. Although many reports have appeared recently on the biological function of this enzyme, its precise role in cell physiology still awaits further progress. Further knowledge of its substrate specificity might throw more light on this problem. This is discussed in Section I11 below.

E.

Ribonuclease H Initially, it was presumed that RNase I11 might hydrolyze doublestranded RNAs as well as the RNA chain of DNAeRNA hybrids. Then a report (118) claimed that E . coZi RNase I11 could also hydrolyze the RNA strand of DNAeRNA hybrids, leaving the DNA strand intact. Subsequently, it was shown that the activity responsible for this hydrolysis is not RNase I11 but a different activity that copurifies with RNase 111; this was called RNase H (hybrid nuclease, EC 3.1.4.34) (119, 120). The RNase I11 activity has been separated from the RNase H activity by DNA-agarose column chromatography ( 121) and purified about 2000fold from extracts of E . coli (122). It is worth mentioning that a similar type of activity was obtained earlier from animal sources and designated

BACTERIAL RIBONUCLEASES

283

RNase H (123, 124). The presence of RNase H activity in preparations of purified reverse transcriptase was also reported ( 1 2 5 ) and confirmed in several laboratories (126-129). The ubiquity of RNase H activity in virions of RNA tumor viruses has also been emphasized ( 1 3 0 ) . RNase H has an absolute requirement for divalent cations. Ammonium sulfate concentrations greater than 0.1 M are strongly inhibitory. Although omission of salt from the standard assay results in higher enzyme activity, RNase H assay is routinely carried out at 0.1 M (NH,)SO, to ensure integrity of the substrate and to discriminate against other nucleases. The enzyme also requires reduced sulfhydryl groups for its activity. Glycerol gradient-centrifugation indicates an s20,w of 3.3 S. If it is a globular protein, the molecular weight of RNase H is about 40,000-45,OOO ( 1 2 0 ) . However, a recent report suggests the molecular weight of RNase H to be around 21,000 with a single subunit (108). Escherichin coli RNase H degrades endonucleolytically all homopolymers in the presence of their complementary deoxyribopolymers to yield mono- and oligonucleotides that carry 5’-phosphate and 3’-hydroxyl termini (128, 129). This is in contrast to the avian myeloblastosis virus ( AMV) enzyme, which acts exonucleolytically in a processive manner ( 1 2 2 ) . RNase H activity from calf thymus is inhibited by S-adenosylmethionine ( 1 3 1 ) . The latter cannot be replaced by its deoxy analog or ATP or cyclic 3’: 5’-AMP. The physiological implication of this inhibition is not known. Several studies indicate a requirement for RNA synthesis in the initiation of DNA replication (132-136). Because the known DNA polymerases can only extend polynucleotide chains, these observations have been interpreted to indicate a requirement for the synthesis of a short RNA chain that exists as an RNA-DNA hybrid. Sugino et al. ( 1 3 6 ) have already shown that a short strand (So-100 nucleotides) of RNA is covalently attached to nascent “Okazaki” fragments and the RNA is removed before ligation of the DNA fragments. Escherichia coli RNase H may serve this purpose, to remove the RNA moiety from these fragments, which are apparently obligatory intermediates in DNA synthesis. Alternatively, RNase H, by generating 3’-hydroxyl groups, may provide a proper substrate for the initiation of replicntion. This enzyme could also be of great value in probing homopolymeric stretches in a DNA or RNA that, only after hybridization with suitable ribo- or deoxyribooligonucleotides, should become susceptible to RNase H. It should be mentioned in this connection that DNA polymerase I may also be involved in the removal of the RNA primer of a DNA chain ( 1 3 7 ) . So it is possible that when polymerase I activity is substantially reduced,

284

ALOK K. DA’ITA AND SALK K. NNOGI

e.g., in a Pol I- mutant, RNase H can perform this function and polymerase I1 or I11 may fill in the gap left by the excised ribonucleotides.

F. Ribonuclease IV Spahr and Gesteland (138) partially purified and characterized this enzyme from an RNase-I-deficient strain of E. coli and designated it RNase IV. This enzyme cleaves phage R17 RNA, yielding two fragments, one sedimenting at about 15 S and carrying the 5’ terminus of the original molecule, and another, sedimenting at 21 S, that lacks the original 5’ end. The cleavage takes place endonucleolytically at a specific site about one-third of the distance from the 5’ terminus. This enzyme requires no divalent or monovalent cation for its activity. Preliminary results (138) indicate that the 21 S RNA fragment directs the synthesis of the RNA synthetase of R17 phage, whereas the 15 S RNA fragment directs the synthesis of some coat-protein-like materials. Nothing is known about the physiological function of this enzyme because of the difficulty of its assay procedure. Nevertheless, it has been most useful in the elucidation of the structure of R17 RNA (139). Bellemare et al. showed the presence of a highly exposed region in E . coli 5 S rRNA structure by partial hydrolysis with ribonuclease IV and sheep kidney nuclease ( 140). A highly purified ribosomal preparation of E . coli 413 deficient in RNase I and polynucleotide phosphorylase undergoes degradation when kept at O°C for a few days (141).It was demonstrated that this degradation results in the complete disappearance of 23 S RNA with the concomitant accumulation of 16 S RNA in the larger subunit. However, when 50 S subunits are immediately resolved into core particles and free proteins in 5 M CsCI, they can be stored at 0°C for a long time. It was postulated that an endonuclease like RNase IV might be responsible for this degradation. The identity of this endonuclease remains to be elucidated.

G. Ribonuclease P and Other Ribonucleases It is now well established that both tRNAs and rRNAs in bacterial and mammalian cells are transcribed in units larger than the final functional products (142-148). The structural differences between the precursor molecules and mature RNAs are under intensive study (147-153). It is also believed that specific RNases are involved in the conversion of precursor polyribonucleotides to mature molecules, and an extensive search is already under way for such “maturation” enzymes. In E . coli, RNase I1 is responsible for the transformation of p16 (precursor of 16 S rRNA) to mature 16 S rRNA (154-156).

BACTERIAL RIBONUCLEASFS

285

Robertson et al. isolated a specific endonucleolytic RNase from E. coli extracts that cleaves only a single phosphodiester bond of the 129nucleotide tyrosine tRNA precursor molecule ( 157), and designated it RNase P. The 129-nucleotide precursor molecule contains 44 nucleotides in addition to the 85 normally present in the mature tyrosine tRNA. Of these extra 44 nucleotides, 41, including a 5’-terminal ribonucleoside triphosphate, are joined to the 5’-end of the mature tRNA sequence. The cleavage by RNase P removes the extra nucleotides present at tlie 5’-terminus of the precursor as a single 41-nucleotide fragment. The purified enzyme has no effect upon the extra nucleotides at the 3’-end of the tRNA precursor. It requires both monovalent and divalent cations for optimal activity and has a pH optimum of 8.0. None of the nucleotide modifications are required for RNase P action (158). RNase P-like activities have been shown to exist in a number of other organisms, like green-monkey kidney cells and Pseudomonas putidu ( 1 5 9 ) . In purifying RNase P activity from extracts of E . coli, Robertson et al. (157) observed a nucleolytic activity seemingly different from other ribonuclease activities; they called it an “obliterase activity.” It is found in relatively strong association with ribosomes, and, more significantly, it remains inactive until some type of salt-dependent activation occurs. Using tyrosine tRNA precursor and synthetic mRNAs as substrates, a potent nucleolytic activity in “obliterase” was shown (157), but purification could not be achieved. The authors speculated that more than one protein in the ribosomes could be responsible for “obliterase” activity. The requirement of gross ribosomal structure in nuclease action has also been noted by other workers (160, 161). Colicin E 3 (161a), a bacteriocin produced by a colicinogenic strain of E . coli, is known to cleave 16 S rRNA from E. coli into two unequal fragments (161h, 1 6 1 ~ )The . smaller fragment is about 50 nucleotides long and contains the 3’ end of the molecule (161b, c). Cleavage of the 16 S rRNA takes place in oiuo after infection of sensitive E . coli with colicin E3 (161b, 161c) or in uitro by incubation of ribosomes ( 1 6 1 d ) . The reaction requires the presence of both ribosomal subunits (160, 161). Although colicin E3 by itself has no nuclease activity, it seems to facilitate the action of an enzyme located on the ribosomes. It is possible that “obliterase” may be this activity, but a direct demonstration of the requirement of other proteins for “obliterase” activity remains to be shown.

H. Oligoribonuclease Stimulation of in vitro RNA synthesis by oligoribonucleotides has been observed by many workers ( 162-165). Only oligoribonucleotides

286

ALOK K. DAlTA AND S A L E K. NNOCI

complementary to the template are stimulatory, and they are preferentially incorporated into the 5’-phosphomonoester chain end of the product, hence acting as chain initiators (163, 165). While testing oligoribonucleotides for priming activity with RNA polymerase, Niyogi and Stevens (163) observed that crude enzyme preparations from E. coZi showed no priming activity, but when purified preparations of RNA polymerase were used, stimulation with complementary oligomers was observed. This discrepancy led to the detection of an enzyme that hydrolyzes short oligoribonucleotides ( 166). Very recently, details of the methods of purification of this enzyme from extracts of E . coli have been reported (167, 168). Using different mutants of E. coli deficient in various RNases, Niyogi and Datta (168) demonstrated that this enzyme is unique in its specificity in hydrolyzing short oligoribonucleotide chains, and designated it “oligoribonuclease” (oligonucleotidase, EC 3.1.4.19, in “Enzyme Nomenclature,” 1972). This nuclease, which is most probably located in the cytoplasm, has been purified over 200-fold. Its molecular weight is approximately 38,000 with no apparent subunits. The enzyme prefers Mnz+ over Mg2+in its divalent cation requirement. Monovalent cations do not have any effect on its activity. The activity is remarkably stable even at 65”C,and the optimum temperature is about 50°C. Studies on the substrate specificity and mode of action show that. the oligoribonuclease is absolutely specific for short single-stranded oligoribonucleotides (166-169). The enzyme prefers a free 3’-OH group and attacks exonucleolytically (from the 3’ end) to produce 5’-ribonucleotides. In this respect, it resembles snake venom phosphodiesterase (87) and leukemic cell phosphodiesterase ( 170) in rapidly hydrolyzing an oligoribonucleotide with a 5’-phosphomonoester end group. But it differs from the two diesterases in that ( a ) it also rapidly hydrolyzes oligoribonucleotides with no phosphomonoester end group, and ( b ) it has no activity on deoxyoligoribonucleotides. The mode of action is unique in the sense that the reaction rate is inversely proportional to the chain length of the substrate although the enzyme has a higher affinity for longer chains (169). It has also been demonstrated that it acts in a processive manner. In this respect, the mode of action appears to be identical to that of RNase I1 (88) except that the enzyme hydrolyzes only short oligoribonucleotides, whereas RNase I1 does not. The same type of processive degradation occurs with polynucleotide phosphorylase (70) and DNA exonuclease VII of E. coli (171, 172). The enzyme activity is markedly inhibited by secondary structure (169); oligoribonucleotides combined with complementary polyribonucleotides are at-

287

BACTERIAL RIBONUCLEASES

tacked efficiently above the melting temperature of the complex but are resistant below it. The activity of the enzyme depends on the sequence in which the different bases are arranged in the oligoribonucleotide substrates. Studies with dinucleoside monophosphates show highest reaction rates with pyrimidine sequences in the order: C-C > U-U > C-U > U-C. The presence of guanine at the 3’-end is strongly inhibitory; the reaction rates are C-G > U-G = A-G > G-G. Thus, the relative effects of sequence on the rate of hydrolysis, especially the inhibitory effect of guanine at the 3’-end, may prove useful for sequence analysis of short oligoribonucleotides. Nothing is yet known about the physiological function of this enzyme. Since RNase I1 leaves short oligoribonucleotides as resistant products ( 88), this oligoribonucleotidase could serve a very useful physiological function as a “finishing enzyme,” creating products that could be phosphorylated to ribonucleoside 5’-triphosphates for reutilization in RNA biosynthesis and other processes. The validity of this postulation will be proved correct only when a mutant strain deficient in this particular enzyme is isolated.

I. RNases from Other Bacteria So far the discussion has mainly centered on the ribonucleases in

E . coli. But besides E . coli RNases, which are found within the cells, certain RNases appear in the culture media of various other organisms. Although the physiology of most of these “extracellular” RNases is not known with any certainty, studies of them have been of great help in comparative protein and nucleic acid chemistry. Nishimura and co-workers ( 1 7 3 ) isolated an extracellular RNase of B. subtih strain H and purified it to crystallinity. This enzyme has a very complicated base specificity (174, 175). The 3’-phosphodiester bonds of purine nucleotides are cleaved faster than those of pyrimidine nucleotides, and those of 3’-nucleotides with 6 ( 4)-keto groups are cleaved faster than those with a 6( 4)-amino groups. Thus, when adjacent bonds are the same, the following results are obtained: -GN-

>

-A-N-

> -C-N-U-N-

Because of its complex specificity, it has not been possible to use the enzyme definitively in nucleotide sequence analysis of RNA, although in some cases it has been useful (176, 177), especially to digest the

288

ALOK K. DATI’A A N D SALIL K. NIYOGI

large fragments produced by RNase TI. Under these circumstances, it usually splits 3’-adenylyl and 3’-uridylyl bonds, but not 3’-cytidylyl bonds. This enzyme has also been used to produce large fragments from tRNA in the presence of Mg2+(177,178). Nishimura and Maruo (179)identified another RNase from B. subti2is strain H that is intracellular and differs from the extracellular enzyme. The enzyme is unique in the sense that the products of RNA digestion are exclusively the four nucleoside 2’ :3’-cyclic phosphates. The enzyme has been extensively purified from another strain of B. subtilis, and its properties have been fully investigated (180,181). ATP and dATP are strong inhibitors of the enzyme, and it appears that ATP may participate in the regulation of intracellular RNase activity (180, 181). Other RNases, such as the intracellular RNase of Azotobader ugilis (182) have also been reported. The enzyme from A. ugilis degrades RNA endonucleolytically to oligonucleotides with chain lengths ranging from two to nine, having 5’-phosphate termini (182).This enzyme has been very useful in preparing oligoribonucleotides of known chain lengths. An enzyme activity in B. subtilis that is responsible for the maturation of 5 S ribosomal RNA has been found; it was named RNase M5. Bacillus subtilis accumulates two precursors of 5 S RNA designated pSA (180 nucleotides) and pSB (150 nucleotides). Both are converted to mature 5 S rRNA, which in this organism is composed of 118 nucleotides. The enzyme RNase M 5 involved in this processing (183)has a single catalytic site that removes specific segments from both the 5’- and 3’-terminals of p5A and pSB. The enzyme is an endonuclease (183),5-6 S in size, corresponding to a molecular weight of 1 to 1.5 X lo5. Maximal activity of this enzyme requires the presence of one relatively low-molecularweight component, possibly a ribosomal protein (183).In E. coli, the precursor 5 S rRNA is converted to the mature molecule by an exonuclease (184,185) that is probably also responsible for the maturation of 16 S and 23 S rRNAs. An enzyme activity in extracts of E . coli that cleaves precursor 16 S RNA to its mature form has been reported (186).This activity, designated tentatively as RNase M (see Section 111, B ) , seems to differfrom all the other known ribonucleases (186).

111. Role of Ribonucleases in Cell Physiology In the preceding section, a broad survey was made of the different ribonucleases of bacterial systems. Naturally one wonders about the physiological advantage of having so many different types of nucleases in one cell. Unfortunately, no definite conclusions can yet be drawn on this subject. In this section, the available information on the tentative

BACERIAL RIBONUCLEASES

289

functions of various RNA depolymerizing enzymes in cells is presented, and the views of several groups of workers are discussed.

A. Messenger RNA Degradation and Processing 1. HOSTSYSTEMS

It was shown long ago that during exponential growth of cells mRNA appears to be metabolically very unstable (6). The half-lives of various mRNAs in different bacteria have been shown by a variety of techniques to be of the order of 90 seconds (187-190),leading to the question as to which enzymes are responsible for the rapid turnover of mRNA. From time to time, various enzymes have been postulated to be involved in this process. It has already been discussed (Section 11, A ) that RNase I, an enzyme present in the periplasmic space of the cell, is probably not responsible for mRNA degradation. Sekiguchi and Cohen (191) first implicated polynucleotide phosphorylase as the enzyme responsible for the degradation of the rapidly labeled RNA (mRNA) formed after infection of E . coli with T6 phage. Their conclusion was based on the following observations: ( a ) the reaction requires a certain amount of inorganic phosphate; ( b ) nucleoside 5’-diphosphate accumulates as the major product; ( c ) 6-aza-UDP, a known inhibitor of polynucleotide phosphorylase, has a strong inhibitory effect; and ( d ) externally added polynucleotide phosphorylase accelerates the reaction. They also noted that a kind of phosphodiesterase bound to ribosomes takes part in the degradation of mRNA in this system. A similar observation has been made on the degradation of mRNA in normal cells under conditions where mRNA was selectively attacked ( 192). Other experiments also suggest that polynucleotide phosphorylase and RNase I1 are involved in the breakdown of poly(U) messenger in a cell-free system during protein synthesis (193, 194). However, it has been observed that a mutant of E . coli deficient in polynucleotide phosphorylase can thrive without this enzyme, that mRNA half-life is not affected in the absence of this enzyme, and, most interesting, that the enzyme is not detectable in extracts of Lactobacillus arabinosus and most animal tissues. These findings contradict the argument that polynucleotide phosphorylase function is mandatory for cell survival. On the other hand, E . coli RNase I1 seems to be essential for cell survival ( 195). This conclusion comes from studies with hyperactive RNase I1 mutants of E . coli (196-198), where it has been demonstrated that tem,perature-sensitive mutant strains exhibit a remarkable increase in RNase activity when grown at nonpermissive temperatures. An increase in the extent of breakdown of pulse-labeled RNA and a

290

ALOK K. DA’ITA AND S A L E K. NNOGI

decrease in the functional lifetime of the mRNA of the Zac operon have also been noted during growth at nonpermissive temperatures. The involvement of RNase I1 in mRNA degradation has been suggested by other workers (199, 200). That a hydrolytic enzyme other than polynucleotide phosphorylase participates in the turnover of mRNA in E. coli was also indicated by the studies of Chaney and Boyer (201 ) . However, one encounters certain problems in accepting E . coli RNase I1 as the enzyme solely responsible for mRNA degradation because of the following experimental facts. ( a ) Both biochemical and genetic experiments show that the overall direction of mRNA degradation is from the 5’ end to the 3’ end of the molecule (200, 202) whereas E . coli RNase I1 degrades polyribonucleotide chains exonucleolytically from the 3’ to the 5’ end ( 8 5 ) . ( b ) mRNA is degraded exponentially and randomly; i.e., a newly synthesized molecule of mRNA has the same probability of being degraded as an old molecule that has already participated in many rounds of translation (203, 204). ( c ) In E . coli, each message has a unique functional decay rate, which can differ from that of another even on the same polycistronic mRNA molecule (205-207). In the case of lac mRNA, the temperature coefficients for decay of the first and last messages differ markedly ( 2 0 8 ) ,and at 37°C the thiogalactoside transacetylase (last) message decays two to three times faster than the p-galactosidase (first) message (206, 209, 210). ( d ) Degradation of mRNA is coupled to transcription-translation as suggested by Stent (211) and demonstrated by several other workers (212-215). In Stent’s hypothesis (211),the mRNA remains bound to a complex containing DNA and RNA polymerase during the process of transcription. The growing end of the mRNA chain, namely, the 3’-OH end group (216, 217) is presumably involved in this binding and would thereby be protected from degradation by RNase 11, which needs a free 3’-OH group for its activity. The binding of polysomes to DNA has already been shown by electron micrography [Miller et al. ( 2 1 8 ) ] ,and the protection of homopolymers by ribosomes from RNase I1 digestion has been demonstrated ( 199).The coupling theory further proposes that ribosomes begin to attach to mRNA chains, and translation is initiated even before transcription is completed. Finally, the ribosomes reach the 3’-OH end of the mRNA and the whole chain is freed from the DNA-RNA polymerase complex. At this stage, if RNase I1 has to attack the mRNA chain, the ribosomes from the 3’-OH end should be released. But studies with polarity mutants (219) suggest that a ribosome may remain bound to a message after reading a termination codon. These observations thus appear to eliminate degrading mechanisms involving exclusively either RNase I1 or a 5’-exonuclease.

BACTERIAL RIBONUCLEASES

291

Such observations led workers in this field to believe that each message is inactivated first by an endonucleolytic attack at one or a limited number of targets for that message, such as a ribosome loading site or an untranslated region just before its 5’ end ( 2 0 6 ) . The fragments may then be degraded exonucleolytically from the 3’ to the 5’ end by an exonuclease, which could be RNase 11. An endonucleolytic mode of decay is amply evident for the lac operon, where the last or operatordistal message is inactivated ( 2 2 0 ) faster, as discussed before, and where endonucleolytic cleavage of the mRNA has been demonstrated by analysis of the size distribution of the decaying molecules ( 2 2 0 ) . Evidence for endonucleolytic fragmentation of gal-operon-specific mRNA ( 221 ) has also been found, using identical techniques. Targets of endonucleolytic attack on Zac mRNA have also been reported recently ( 2 2 2 ) . Endonucleolytic cleavage of tryptophan mRNA has also been proposed ( 207, 223 ) . The postulated endoribonuclease could be either RNase I11 or RNase H or another endoribonuclease. Apirion ( 2 2 4 ) proposed a hypothesis by which RNase I11 or RNase H could act endonucleolytically on mRNA. According to his hypothesis, when mRNA is being synthesized and a ribosome does not become attached to it, or when a ribosome leaves the 5’ end of the message and another ribosome fails to attach to it, thy RNA could form double-stranded regions within itself, if there are appropriate stretches of nucleotides, and then be attacked by an enzyme that degrades double-stranded RNA, or pulled back.to it,s DNA template and attacked endonucleolytically by an enzyme that degrades RNA chains in DNAaRNA hybrids (e.g., RNase H ) . As soon as any endonucleolytic cleavage takes place, the 3’-exonuclease could continue the degradation of the RNA to nucleotides, thus resulting in the overall direction of degradation of the messenger from the 5’ to the 3’ end. The appealing feature of this hypothesis is that there is no need to have specific enzymes for the degradation of mRNA after a polar mutation, and the function of the SuA gene product can be explained if one considers the product to be responsible for controlling the activity of the endonuclease, as also suggested by Apirion ( 2 2 4 ) . However, a more recent report ( 2 2 4 4 suggests that rho is the product of the SuA gene, and it does not have any ribonuclease activity. Apirion’s hypothesis also faces a serious challenge when one tries to explain the effect of puromycin in a SUA mutant ( 2 2 5 ) . The effect of puromycin, namely, faster degradation of mRNA, can be antagonized by SuA mutation ( 2 2 5 ) . It is well known ( 2 2 6 ) that puromycin causes increased mRNA degradation by decreasing the polysome level and thus exposing mRNA to nuclease attack. Under these circumstances, it is

292

ALOK K. DATTA AND SALIL K. NNOGI

difficult to conceive how SuA mutation can cause a reversion of the puromycin effect, thus stabilizing mRNA ( 2 2 5 ) . These findings can be explained if we consider that SuA strains show lower levels of an endonuclease activity (227) (“endonuclease A ) that elutes at a specific position from DEAE-cellulose columns (228). Richardson et al. ( 2 2 4 4 speculate that rho could possibly be a component of this “endonuclease A.”’ Moreover, the experimental evidences of the involvement of RNase 111 as the endonuclease remain unresolved because of the conflicting reports from Silengo et al. (229) and Apirion and Watson (230) about the stability of mRNA in RNase I11 mutants. Thus, a satisfactory explanation of the complex process of mRNA degradation still awaits further investigation.

2. POSTTRANSCRIPTIONAL PROCESSING OF PHAGE SPECIFICmRNAs Within the last few years, evidence has accumulated suggesting that, like the host mRNAs, phage-specific mRNAs are also processed by specific ribonucleases. The observation that led to this finding came from the studies of Dunn and Studier (231,232),who found that transcription of T7 DNA by purified E . coZi RNA polymerase in uitro without added factors produces a long RNA molecule that starts near the left end of T7 DNA and terminates at the end of the “early” region. But when a “sizing factor” from uninfected E . coli is added to the in uitro incubation mixture, the long RNA molecule is cleaved specifically at five distinct sites to produce RNA transcripts that correspond essentially to the five in oiuo monocistronic messages of genes 0.3, 0.7, 1, 1.1 and 1.3 of the early T7 genome. This “sizing factor,” from both genetic and biochemical evidence, turned out to be RNase I11 (231, 232). Recently, Dunn and Studier (233) produced evidence that, like the early message, T7 late messages are also cleaved at specific sites. On analysis of the terminal nucleotide sequences of fractionated T7 early RNAs isolated from infected cells, Kramer et aZ. (234) observed certain specific nucleotide sequences both at the 5’ and 3’ termini. Structural analysis of the RNA fragments (produced by RNase I11 action on the in uitro-synthesized T7 RNA) showed (235) that the terminal sequences were identical to these. These observations, together with another recent report ( 2 3 6 ) , strongly substantiate the earlier finding (231, 232) that T7 early RNAs are indeed the direct products of RNase I11 cleavage of a long parent molecule, and that the recognition sites for the enzyme are sequencespecific. Bacteriophage T4 species-I RNA, a molecule 140 nucleotides in length with some structural features resembling a tRNA (237, 2 3 7 ~ ) , is cleaved by RNase 111 in uitro into segments of 19, 47 and 73 nucleo* See Note 1, p. 308.

BACTERIAL RIBONUCLEASES

293

tides with nucleotide 67 missing from the fragments (237b, 2 3 7 ~ )These . fragments are not found in uiuo and apparently T4 species-I RNA is quite stable. This suggests that the action of RNase I11 on T4 species4 RNA is influenced by in uitro assay conditions (237c). It has been suggested that the concentration of salt in the in uitro reaction mixture influences the structure of the substrate in such a way that it becomes susceptible to attack by RNase 111. Thus, the fact that the RNase I11 reaction is so sensitive to changes in reaction conditions opens the possibility of investigation of its detailed mode of action with a wide variety of substrates, thus leading to a further understanding of its specificity. Now the question arises: Why must T7 mRNA be cut at specific sites? Hercules et al. ( 2 3 8 ) ,using an in uitro protein-synthesizing system, have demonstrated that RNase I11 cleavage is required for efficient translation of the messages of genes 1 (RNA polymerase), 1.3 (ligase), and 3.5 (lysozyme) of the T7 genome, and for the S-adenosylmethioninecleaving activity of the T3 genome. However, most of the early T7 RNAs are translated with equal efficiency whether they are cleaved by RNase I11 or not ( 2 3 9 ) . The exception lies in the gene 0.3 protein and a few other late proteins, which seem to need endonucleolytic cleavage of the mRNAs for efficient translation. It has also been pointed out ( 2 3 9 ) that the strain used by Hercules et al. (238) is a male strain in which growth of T7 phage is restricted ( 2 4 0 ) . Thus, a number of unrelated factors could affect the in uitro synthesizing system, and the reasons for the posttranscriptional processing of T7 and T3 mRNAs remain to be elucidated.

B. Ribosomal RNA Processing During the past few years, evidence has accumulated suggesting that both ribosomal and transfer RNAs are transcribed in units larger than the final mature products. This maturation process obviously consists of two parts: nucleolytic cleavage of the precursor molecules (“tailo r i n g ) and enzymic formation of the modified bases (modification). Since the present review is on the role of ribonucleases in cell physiology, we confine ourselves here to the role of ribonucleases in RNA maturation. Extensive reviews have already appeared on the structure and regulation of rRNA synthesis ( 2 - 4 ) ,but very little is known about the posttranscriptional tailoring of rRNA and the role of enzymes responsible for the cleavage events accompanying stable RNA maturation. Escherichia coli N464, which has a thennolabile RNase I1 activity, cannot synthesize mature ribosomes or mature rRNA at elevated temperatures (l54-156).They also showed that purified precursor 16 S rRNA may be converted in uitro, by highly purified E . coli RNase I1 prepara-

294

ALOK K. DAlTA AND SALIL K. NIYOCI

tions, to products that move identically with mature 16 S rRNA during polyacrylamide gel electrophoresis, with the liberation of fragments at least 100 nucleotides in length (156). They concluded that RNase I1 mediates the conversion of precursor 16 S to mature 16 S rRNA. Since RNase I1 is an exonuclease, the above result is surprising, unless the RNase I1 preparation also contained an endonuclease. Subsequent reports suggest that the observed accumulation of precursor 16 S rRNA in E . coli N464 at restrictive temperatures is not a consequence of defective RNase 11, and thus RNase I1 may not be the agent of the conversion in d u o (241). It has recently been claimed that a different type of activity, called RNase M, in the extracts of E . coli (186), is most probably responsible, but no direct characterization of RNase M in this process has yet been made. RNase I11 participates in the processing of E . coli rRNAs in uiuo by cleaving the primary transcripts at a few sites (231, 232, 242-245). It is suspected that the strain used in this work [an RNase III-deficient host, AB105, isolated after heavy mutagenesis of parental A19 ( 2 4 6 ) ] contains several other secondary mutations ( at least seven) besides that in RNase I11 (247). Therefore, it is not certain that RNase I11 deficiency alone is responsible for all the effects on stable RNA synthesis observed in uiuo. When analysis of stable RNA was carried out with four pairs of isogenic strains of RNase 111' and RNase III-, it was observed that RNase I11 may not be responsible in the processing of ribosomal RNAs in uiuo (248). Some workers in this field believe that, in prokaryotes, maturational cleavage of precursor 16 S sequences from the tandem precursor of 16 S and 23 S RNA occurs during rather than after the process of transcription (249-251 ). It is not yet known, however, whether there exists in prokaryotes a mechanism that couples RNA transcription and maturation. Like 16 S and 23 S rRNAs, mature 5 S rRNA components of all prokaryotes are also fragments of larger precursor molecules. Precursor 5 S rRNA is accumulated in a pulse-labeled cell culture of E. coli that is not undergoing protein synthesis (144,184,252). When isolated, this precursor is a mixture of RNA molecules that are only one, two or maximally three nucleotides larger than the mature 5 S rRNA. All these additional nucleotides are at the 5' terminus (144, 252), and kinetic experiments have demonstrated that, during maturation, these are removed in a stepwise fashion until the mature 5 S rRNA is formed. Precursor 5 S rRNAs similar in size to those of E . coli have also been detected in S . typhimurium (253). Monier et al. surmise that maturation is achieved by an exonuclease rather than the specific endonuclease probably responsible for maturation of 16 S and 23 S rRNAs (184).

BACTERIAL RIBONUCLEASES

295

C. Transfer RNA Processing’ Like rRNAs, tRNAs are also not the immediate products of DNA transcription. Rather, precursors of the mature RNA species undergo a series of nucleotide modification and/or cleavage events. Only limited information is available concerning mechanisms and rationale of posttranscriptional processing of tRNAs. Altman et al. (147, 148) first showed the presence of a precursor of tyrosine tRNA ( ptRNATyr) species in E. coli. Subsequently, Robertson et al. (157) discovered RNase P, which specifically cleaves ptRNATyr endonucleolytically to its mature product. (The mechanism of action of this enzyme has been discussed in Section 11,G.) Subsequently, Ikeda ( 2 5 4 ) presented evidence from in uitro experiments that RNase I1 and RNase P act sequentially in the processing of ptRNATYr. That RNase P may be one of the activities responsible for processing of tRNA was suggested from the recent isolation of temperature-sensitive mutants of E . coli (255-257) in which, at restrictive temperatures, tRNA precursors accumulate. Later it was found that RNase P is not the only enzyme responsible for tRNA processing, and it might act in conjunction with some other activities (258, 259). When certain polycistronic precursors are treated in uitro with RNase P, they are either not attacked at all or are partially cleaved with the production of molecules of lower sizes ( 2 5 8 ) . In E. coli extracts, there is another endonuclease, called RNase Pa, that is capable of processing polycistronic precursors in intercistronic regions to products having 5’-monophosphate groups (258). (RNase P and RNase P, act endonucleolytically to remove the extra nucleotides from the 5’ end.) The trimming of the 3’ end may be done by the exonuclease RNase I1 ( 2 6 0 ) . In uitro, this enzyme can carry out such a trimming reaction on the ptRNATyr isolated from E . coli A49, which has a thermosensitive RNase P ( 2 5 6 ) ; extracts of E . coli that contain thermosensitive RNase I1 cannot carry out the trimming reaction at elevated temperature (260). Considering these results, Altman (261) proposed a hypothetical model involving RNases P, P, and I1 for the processing of large tRNA precursors. Although RNase I1 is capable of catalyzing the removal of the extra 3’-nucleotides from the Su:,, tRNATyrprecursor, the efficiency of the reaction is very poor (262, 263). Bikoff et al. (263) identified an RNase activity that catalyzes the removal of the extra 3’-nucleotides and designated the activity as RNase PIII.* They also showed that this activity is

’ See article by J. D. Smith in Volume 16 of this series. “ I n accordance with the previous nomenclature, we suggest that RNase PI11 be called RNase Pa.

296

ALOK K. DA'ITA AND SALlL K. NIYOCI

clearly separated from RNase I1 during the process of purification and that purified RNase I1 is unable to replace the requirement for RNase P, in the in oitro system for tRNA synthesis. Recently, two other tRNA processing activities, viz., RNase 0 and RNase Q, have been reported ( 2 6 3 4 . The functions of these nucleases are sequential in the trimming process with respect to that of RNase P. It has been claimed that RNase 0 acts prior to RNase P followed by RNase Q. The activities described have not been fully characterized and may be the same activities as described by other workers (258, 263).

D. Stable RNA Degradation under Conditions of Starvation rRNA and tRNA molecules are quite stable; under normal growing conditions, their turnover rate is much slower than the generation time of their host cell (264, 265), although the turnover rate is faster in eukaryotes than in prokaryotes ( 2 6 6 ) . However, under certain conditions the newly synthesized stable RNAs are degraded probably as fast as mRNA molecules, namely, when cells are growing with a very long doubling time (267-272 ), when stringent cells are allowed to grow under starved conditions ( 271-273), when temperature-sensitive ribosomal mutants are shifted from permissive to nonpennissive temperatures (274), and when cells are starved for some carbon source (275-277). Under certain other conditions, a fraction of the newly synthesized rRNA is degraded even during exponential growth ( 2 7 8 ) .It is interesting that, even under the limited growth conditions described above, the viability of the cells is not affected and cells can still form inducible enzymes when inducers are added to the medium ( 2 6 4 ) . The degradation of rRNA is also observed when cells are treated with mitomycin C or colicin E2 (279, 280). Various enzymes have been implicated in stable RNA degradation. An involvement of E . coli RNase I1 in the degradation of stable RNA during starvation has been suggested (281). Under conditions of starvation, the process of degradation starts with an endonucleolytic attack by RNase I (275-277). The small fragments thus created are further degraded to nucleotides by the two processive exoribonucleases, RNase I1 and polynucleotide phosphorylase. They also indicated that the rate and extent of stable RNA degradation during carbon starvation depends mainly on the genetic makeup of the strains, namely, with respect to genes that affect ribonuclease activities.

IV. Regulation of Ribonuclease Activity The most obvious function of an RNase is to degrade RNA, but this tells little about the actual physiological action of these enzymes.

BACTERIAL RIBONUCLEASES

297

Which RNase is involved, and what prevents indiscriminate breakdown of RNA molecules? These are questions that immediately come to mind. Although not much effort has been expended in the study of the functional modulation of different RNases, several possible mechanisms may be considered on the basis of in uitm experiments. Control could result from ( a ) the existence of specific inhibitors, ( b ) spatial compartmentalization, ( c ) protection of RNA by some cellular components, ( d ) specificity for one form of macromolecular structure and ( e ) concentration of the substrates, products and cofactors. In mammalian cells, the device for the control of degradation of RNA seems to be a combination of the alkaline RNase with an inhibitor (282-285). The inhibitor is a protein, and it has been purified and characterized (286, 287). Single-stranded mRNA from rat liver cells is also protected by this inhibitor (288). Alkaline RNase is bound to ribosomes only in their subunit form, and it is postulated that, in animal cells, mRNA is destroyed when it becomes attached to reassociated ribosomes of this type (285). Further, the breakdown of mRNA is randomly controlled depending upon the segregation of the alkaline RNase between the soluble inhibitor and the ribosomes (282-285). A general inhibitor like the one for mammalian RNase has not yet been found in bacteria, but a specific inhibitor that is protein in nature (MW 12,500) and inhibits specifically the extracellular RNase produced by B . subtilis has been purified from these bacterial cells. It is located inside the cells and probably acts as a protective agent against the extracellular RNase (289). The spatial compartmentalization of RNases might be one of the mechanisms that introduces certain restrictions on the degradation of intracellular RNA by RNases. It has been proposed (290) that under conditions of stress (phosphate deficiency, temperature shock, etc. ) RNase I, an enzyme present in the periplasmic space of the cell, enters the cells after an alteration or rupture in the membrane structure (274) and causes considerable breakdown of stable RNA, just like lysosomal RNases of higher organisms. Cellular RNA might also be protected from the action of RNases when it is complexed with ribosomal proteins. In an incubation mixture containing ribosomes and rRNA, the free rRNA is susceptible to RNase attack whereas RNA in the ribosomes is completely resistant (291). Moreover, mRNA is more susceptible to degradative enzymes, probably because it is not bound extensively to proteins. The assumption of a particular macromolecular structure also plays a major part in the protection of different RNAs from RNase action. All 480 phage derivatives carrying mutant tRNATYr genes (these mutants

-

298

ALOK K. DATTA AND SALIL K. NIYOGI

produce tRNATYr that differ structurally from the well-known cloverleaf form) produce much less tRNA in oioo than do phages carrying the wild-type or Sut,, genes. Moreover, their tRNA precursors have longer half-lives in oioo than does the wild-type precursor (147, 148). Altman and Smith (148) explained the phenomenon with the postulate that, for efficient cleavage to occur, RNase P needs a specific conformation of the tRNA molecule. If the conformation of the substrate is altered, the rate of processing is slowed and other intracellular enzymes have time to attack and degrade the improper precursor (261). Thus, substrate conformation may regulate the activity of RNases. Even though the chemical structure of the substrate remains the same, a particular conformation may lead to recognition and degradation by a different ribonuclease (261). Similar observations (292), show that a mutant producing conformationally altered tRNAserhas a slower rate of RNase P processing of both tRNAser and tRNAprosegments of the dicistronic T4 tRNA precursors (292). Inspection of known tRNA sequences also confirms that primary nucleotide sequence alone is not sufficient to specify the RNase P cleavage site (237~). The intracellular concentration of substrates, products and cofactors might also play some regulatory role in the controlling mechanism of apparent nuclease activity. In spore-forming B. subtilis, the enzymic breakdown of polyribonucleotides to nucleoside monophosphates and free nucleosides is modulated by the levels of nucleoside triphosphate concentration within the cell (293, 294). It has also been claimed (92) that the activity of RNase I1 in uiuo in E . coli might be regulated by Na+,ATP and tRNA levels.

V. Physiology of Ribonuclease-Minus Mutants In the preceding sections, discussions are confined mainly to the involvement of different ribonucleases in some of the steps of RNA metabolism. The conclusions were reached both from in oitro and in uiuo studies. The approach that proved to be of great use in establishing the role of various RNases in uioo has been the finding of mutants deficient in one or more of these enzymes. Studies of this kind were first initiated by Gesteland (35), who showed that a mutant strain of E. coli lacking E . coli RNase I activity is completely normal in its growth properties. This conclusion was further supported by the existence of a naturally occurring E. coli strain, MREBOO, that also lacks RNase I (295). Subsequently, isolation and characterization of such mutants in other organisms have also been reported (296, 297). The existence of such RNase I- mutants suggested

BACTERIAL RIBONUCLEASES

299

that the enzyme is nonessential for bacteria, and a physiological role of this enzyme thus could not be assigned. Mutants lacking polynucleotide phosphorylase also seem to have normal growth characteristics (298, 299). If polynucleotide phosphorylase were an essential enzyme, cells devoid of it could be isolated only as conditional lethals. Since this is not the case, it seems that cells can survive a lack of this enzyme also. Lennette and Apirion (300) isolated a class of temperature-sensitive mutants of E. coli that also show an increased sensitivity to carbon starvation. These mutants, designated as sts ( starvation-temperature sensitive) mutants, cannot recover from starvation and also show temperature sensitivity (growing at 30"C, but not at 43°C). Among this class, strains were also isolated that are altered in their protein synthetic capacity. While analyzing these strains for mRNase activity, it was found that one of these, designated N4752, had a modified RNase I1 activity when grown at the restrictive temperature. From genetic analysis of revertants and transductants, as well as location of sts mutants at a unique site, it was shown that these phenotypes of the strain are caused by a single-site mutation that lies between the genes ilv and rbs (301) on the E. coli map (302). Thus, survival of these strains at permissive temperatures suggests that this enzyme ( E . coli RNase 11) may be essential for cell survival.* The study of RNase mutants is usually complicated by the fact that there is no specific procedure for the isolation of mutants defective in RNase function. The usual procedure consists of treating the cells with high concentration of N-methyl-N'-nitro-N-nitrosoguanidine( nitrosoguanidine ), It is well known that nitrosoguanidine at high concentration causes multiple closely linked mutations (303). E. coli AB105 is one such strain isolated by Kindler et al. (246). This strain, in addition to a lesion in RNase 111 function, contains at least 7 or 8 additional mutations (247). These mutations are related to each other physiologically; that is, they are compensating in their functions (247). Thus, physiological analysis of such mutants becomes extremely difficult and sometimes erroneous as discussed before (Section 111, B). Recently, Kinscherf et al. (304) claim to have devised a new technique for the isolation of mutants affected in the ability to degrade cellular RNA. It has also been claimed that, using this mild technique, they have been able to isolate mutants altered in polynucleotide phosphorylase activity, and that the behavior of such a mutant is almost identical with that of the mutant isolated previously by Reiner (298, 299), using usual procedures. ' S e e Note 2, p. 308.

300

ALOK K. DAITA AND SALIL K. NNOCI

VI. Concluding Remarks In thi,s review, emphasis is placed on the properties of RNases and their physiological roles in the cell economy. A feature of these enzymes that has not been discussed is their use as biochemical tools in the study of the structures and nucleotide sequences of various RNA molecules. Polynucleotide phosphorylase and Azotobucter nuclease have been used by many workers to prepare polyribonucleotide and oligoribonucleotide chains, respectively, of defined lengths, base compositions and sequences. Recently, Jeppesen et al. ( 1 3 9 ) and Bellemare et al. ( 1 4 0 ) exploited the specific cleavage property of RNase IV in the determination of the structure of R17 phage RNA and 5 S rRNA, respectively. Kramer et al. ( 2 3 4 ) , and Rosenberg et al. ( 2 3 5 ) have also successfully used RNase I11 to determine the terminal sequences of the early messages of the T7 genome. It is quite obvious from the review that, although a considerable amount of information is available regarding the RNases of bacterial systems, the mechanism of action and physiological functions of the enzymes still remain to be elucidated. Moreover, the increasing number of enzymes whose functions are either to process or degrade polyribonucleotide chains poses a challenge to biochemists and molecular biologists in this area. There are questions concerning the mechanism and specificity of action of the enzymes already known. Most of our knowledge on the mechanism of action of the enzymes come from studies that involve the isolation of a particular enzyme and studies of its action with specific substrates in uitro, with subsequent extrapolation of these results to in uiuo conditions. While progress is being made at an accelerated pace, the answers to many questions of the actual in uiuo mechanisms still remain unclear. This is further complicated by contradictory reports from different laboratories. Recently, however, through the use of both genetic and biochemical approaches and in uitro processing systems, rapid progress has been made in elucidating individual steps involved in RNA maturation in a wide variety of systems. For example, the approaches taken by Dunn and Studier (231-233) and Bikoff et al. (263) are promising. The future for the biochemistry and physiology of RNA processing certainly augers to be an exciting one.

ACKNOWLEDGMENTS The authors are indebted to Dr. Waldo E. Cohn for his critical suggestions and incisive comments. Thanks are expressed to various investigators for sending us their preprints and unpublished observations.

301

BACTERIAL RIBONUCLEASES

REFERENCES E. P. Geiduschek and R. Haselkorn, ARB 38, 647 (1969). G. Attardi and F. Amaldi, ARB 39, 183 ( 1970). R. H. Burdon, This Series 11,33 (1971). N. R. Pace, Bacteriol. Rev. 37,562 (1973). 5. M. J. Chamberlin, ARB 43, 721 (1974). 6 . J. Mandelstam, BacterioZ. Reo. 24, 289 ( 1960). 7 . M. Laskowski, Sr., in “Handbook of Biochemistry” (G. Fasman, ed.), 3rd ed. Chemical Rubber Co., Cleveland, Ohio. 8 . D. Elson, BBA 27,216 ( 1958). 9. D. Elson, BBA 38, 362 (1959a). 10. D. Elson, BBA 36,372 ( 1959b). 1 1 . D. Elson and M. Tal, BBA 36, 281 ( 1959). 12. M. Tal and D. Elson, BBA 76, 40 (1963). 13. M. Artman and H. Engelberg, BBA 80,517 (1964). 14. P. F. Spahr and B. R. Hollingworth, ]BC 236, 823 ( 1961). 15. E. T. Bolton, R. J. Britten, D. B. Cowie, B. J. McCarthy, K. Mcquillen and R. B. Roberts, Carnegie Inst. Wash. Ann. Rep. p. 259 ( 1959). 16. H. C. Neu and L. A. Heppel, PNAS 51, 1264 ( 1964a). 17. H. C. Neu and L. A. Heppel, PNAS 51, 1267 (1964b). 18. H. C. Neu and L. A. Heppel, JBC 239,3893 ( 1 9 6 4 ~ ) . 19. H. C. Neu and L. A. Heppel, BBRC 17,215 (1964d). 20. L. A. Heppel, Science 158, 145 (1967). 21. M. Holden and N. W. Pirie, B] 60,39 ( 1955). 22. R. W. Brockman and L. A. Heppel, Bchem. 7,2554 (1964). 23. M. Teuber and G. Cemy, FEBS Lett. 8,49 (1970). 24. W. Zillig, W. Krone and M. Albers, Z. Physiol. Chem. 317, 131 (1959). 25. S. S. Cohen and J. Lichtenstein, JBC 235, 2112 (1960). 26. A. K. Datta and D. P. Burma, JBC 247,6795 ( 1972). 27. M. Nomura, Bacteriol. Reu. 34, 228 ( 1970). 28. A. K. Datta, Ph.D. Thesis, Calcutta University, India (1974). 29. K. Chakraburtty and D. P. Burma, JBC 243, 1132 (1968). 30. L. Schuster, H. G. Khorana and L. A. Heppel, BBA 33, 452 (1959). 31. K. K. Reddi, BBA 28,386 (1958). 32. W. Frisch-Niggemeyar and K. K. Reddi, BBA 26, 40 (1957). 33. R. Markham and J. L. Strominger, B] 64,460 ( 1956). 34. M. Artman and H. Engelberg, BBA 95,687 (1965). 35. R. F. Gesteland, J M B 16, 67 (1966). 36. T. Kivity-Vogel and D. Elson, BBA 138, 66 ( 1967). 37. D. Elson and T. Kivity-Vogel, in “Regulation of Nucleic Acid and Protein Biosynthesis” (V. V. Konningsberger and L. Bosch, eds.), p. 196. Elsevier, Amsterdam, 1967. 38. J. H. Anderson and C. E. Carter, Bchem. 4, 1102 ( 1965). 39. Y. Anraku and D. Mizuno, BBRC 18,462 (1965). 40. J. W. Abrell, ABB 142, 693 (1971). 41. A. K. Mukhopadhyay and D. P. Burma, BBA 190,232 (1969). 42. H. Sekine, E. Nakano and K. Sakaguchi, BBA 174,202 ( 1969). 43. G. Felsenfeld, G . Sandeen and P. H. von Hippel, PNAS 50, 644 (1963). 1. 2. 3. 4.

302

ALOK K. DATTA AND SALIL X. NIYOGI

44. E. V. Raju and N. Davidson, Biopolymers 8,743 ( 1969). 45. M. Grunberg-Manago and S. Ochoa, JACS 77,3165 ( 1955). 46. U. Z. Littauer and A. Kornberg, JBC 226, 1077 ( 1957). 47. M. N. Thang, D. C. Thang, and M. Grunberg-Manago, BBRC 28, 374 (1967). 48. W. T. Hsieh and J. M. Buchanan, PNAS 58,2468 ( 1967). 49. J. J. Castles and M. F. Singer, BBRC 32,715 (1968). 50. J. Y. Chou and M. F. Singer, JBC 246,7486 ( 1971). 51. J. Y. Chou and M. F. Singer, JBC 246,7497 ( 1971). 52. J. Y. Chou and M. F. Singer, JBC 246,7505 ( 1971). 53. C. B. Klee, JBC 242,3579 ( 1967). 54. C. B. Klee, jBC 244,2558 (1969). 55. C. B. Klee and M. F. Singer, BBRC 29,356 (1967). 56. C. H. Letendre and M. F. Singer, JBC 249,7383 ( 1974). 57. L. A. Heppel, M. F. Singer and R. J. Hilmoe, Ann. N.Y. Acad. Sci. 81, 635 (1959). 58. M. F. Singer, L. A. Heppel, R. J. Hilmoe, S. Ochoa and S. Mii, Proc. 3rd Cancer Res. Conf. p. 41 (1959). 59. M. Grunberg-Manago, in This Series 1,93 (1963). 60. I. R. Lehman, This Series 2, 86 ( 1963). 61. J. Y. Chou and M. F. Singer, JBC 245, 995 (1970). 62. Y. Kimhi and U. Z. Littauer, BuEl. Res. Council, lsr., Sect. A 11, 77 (1962). 63. Y. Kimhi and U.Z. Littauer, Bchem. 6,2066 (1967). 64. S. Ochoa, ABB 69, 119 (1957). 65. M. Grunberg-Manago, J M B 1,240 (1959). 66. M. F. Singer, JBC 232, 211 (1958). 67. M. F. Singer, R. J. Hilmoe and M. Grunberg-Manago, JBC 235, 2705 (1960). 68. R. J. Hilmoe, Ann. N.Y. h a d . Sci. 81,660 (1959). 69. M. N. Thang, W. Guschlabauer, H. Zachau and M. Grunberg-Manago, J M B 26, 403 ( 1967). 70. C. B. Klee and M. F. Singer, JBC 243,923 (1968). 71. J. Y. Chou and M. F. Singer, JBC 245, 995 (1970). 72. J. Y. Chou, M. F. Singer and P. McPhie, JBC 250, 508 (1975). 73. R. A. Alberty, JACS 75, 1928 ( 1953). 74. W. W. Cleland, BBA 67, 104 (1963). 75. J. Chao, G. F. Johnson and D. J. Graves, Biochemie 8, 1459 ( 1969). 76. V. T. Maddaiah and N. B. Madsen, JBC 241,3873 (1966). 77. T. Godefroy, E J B 14,222 ( 1970) , 78. T . Godefroy-Colburn and M. Grunberg-Manago, in “The Enzymes” (P. D. Boyer, ed. ), 3rd ed., Vol. 7, p. 533. Academic Press, New York. 79. H. E. Wade, BJ 78,457 (1961). 80. H. E. Wade and S. Lovett, BJ 81, 139 ( 1961). 81. P. F. Spahr and D. Schlessinger, JBC 238, 2251 ( 1963). 82. P. F. Spahr, JBC 239,3716 (1964). 83. R. K. Ray and D. P. Burma, BBA 212, 102 (1970). 84. D. M. Logan and M. F. Singer, JBC 243, 6161 ( 1968). 85. M. F. Singer and G. Tolbert, Bchem. 4, 139 (1965). 86. R. V. Krishna, T. S. KO, B. Meyhack and D. Apirion, FEBS Lett. 39, 105 (1973). 87. H. G. Khorana, in “The Enzymes” (P. D. Boyer, H. Lardy and K. Myrback, eds.), Vol. 5, p. 79 ( 1961). Academic Press, New York, 1961.

BACTERIAL RIBONUCLEASES

303

N. G. Nossal and M. F. Singer, JBC 243, 913 ( 1968). M. F. Singer and G. Tolbert, Science 145, 593 (1964). L. Gorelic and D. Apirion, BBRC 44, 1184 (1971). J. Eley and K. Crawford, BBRC 43, 1064 ( 1971). P. Venkov, D. Scblessinger and D. Longo, J. Bacteriol. 108, 601 (1971). R. Holmes and M. F. Singer, JBC 248,2014 (1973). T. S. KO,L. Gorelic and D. Apirion, JBC 248,3748 (1973). C. Gurgo, D. Apirion and D. Schlessinger, FEBS Lett. 3, 34 (1969). M. Kuwano, C. N. Kwan, D. Apirion and D. Schlessinger, PNAS 64, 693 (1969). 97. M. Kuwano, D. Schlessinger and D. Apirion, Science 168, 1225 (1970). 98. M. Kuwano, D. Apirion and D. Schlessinger, JMB 51,453 ( 1970). 99. M. Kuwano, D. Schlessinger and D. Apirion, Nature (London) 226, 514 (1970). 100. M. Kuwano, D. Apirion and D. Schlessinger, JMB 51, 75 (1970). 101. R. HoImes and M. F. Singer, BBRC 44,837 ( 1971). 102. A. L. M. Bothwell and D. Apirion, BBRC 44, 844 ( 1971). 103. F. Ben Hamida and F. R. William, Bull. SOC. Chim.Biol. 51, 1545 (1969). 104. H. D. Robertson, R. E. Webster and N. D. Zinder, Virology 32, 718 (1967). 105. A. K. Datta, A. K. Mukhopadhyay, T. Suryanarayana and D. P. Burma, Anal. Biochem. 51, 321 (1973). 106. H. D. Robertson, R. E. Webster and N. D. Zinder, JBC 243, 82 (1968). 107. M. Libonati, Bull. SOC.ltd. Biol. Sper. 44,786 (1968). 108. J. L. Darlix, EJB 51, 369 ( 1975). 109. H. Schweitz and J. P. Ebel, Biochirnie 53, 585 ( 1971). 110. H. Witzel and E. A. Bernard, BBRC 7, 295 (1962). 111. H. Follman, H. J. Weiker and H. Witzel, EJB 1, 243 (1967). 112. R. H. Langridge, and P. J. Gomatos, Science 141, 691 (1963). 113. R. H. Langridge, M. A. Billeter, P. Borst, R. H. Burdon and C. Weissmann, PNAS 52, 114 (1964). 114. T. Sato, Y. Kyogoku, S. Higuchi, Y. Mitsui, Y. Iitaka, M. Tsuboi and K. Miura, JMB 16,180 ( 1966). 115. K. Tomita and A. Rich, Nature (London) 201, 1160 (1964). 116. S. Arnott, F. Hutchinson, M. Spencer, M. H. F. Wilkins, W. Fuller, and R. Langridge, Nature (London) 211,227 (1966). 117. R. E. Franklin and R. G. Goslin, Nature (London) 171, 740 (1953). 118. H. D. Robertson, Nature (London) 229,169 ( 1971). 119. H. I. Miller, G. N. Gill and A. D. Riggs, FP 31,500 (1972). 120. H. I. Miller, A. D. Riggs and G. N. Gill, JBC 248,2621 (1973). 121. S. C. Weatherford, L. S. Weisberg, D. Achord and D. Apirion, BBRC 49, 1307 (1972). 122. J. P. Leis, I. Berkower and J. Hurwitz, PNAS 70, 466 (1973). 123. H. Stein and P. Hausen, Science 166, 393 (1970). 124. P. Hausen and H. Stein, EJB 14,278 (1970). 125. K. Molling, D. P. Bolognesi, H. Bauer, W. Busen, H. W. Planmann and P. Hausen, Nature (London) 234,240 (1971). 126. D. Baltimore and D. E. Smoler, JBC 247,7282 ( 1972). 127. J. P. Leis, I. Berkower and J. Hurwitz, in “Symposium of DNA Synthesis in Vitro” ( R . Wells and R. Innian, eds.), p. 287. University Park Press, Maryland, 1972.

88. 89. 90. 91. 92. 93. 94. 95. 96.

304

ALOK K. DATI’A AND SALIL K. NIYOGI

W. Keller, PNAS 69, 1560 ( 1972). W. Keller and R. Crouch, PNAS 69,3360 ( 1972). D. P. Grandgenett, G. P. Gerard and M. Green, J. V i r d . 10, 1136 (1972). J. G. Stavrianopoulos and E. Chargaff, PNAS 70, 1959 (1973). K. G. Lark, J M B 64, 47 (1972). P. Bazzicalupo and G. P. Tocchini-Valentini, PNAS 69, 298 (1972). W. F. Dove, H. Inokuchi and W. F. Stevens, in “The Bacteriophage Lambda” (A. D. Hershey, ed.), p. 747 Cold Spring Harbor Laboratory, New York. 1971. 135. D. Brutlag, R. Schekman and A. Kornberg, PNAS 68, 2826 (1971). 136. A. Sugino, S . Hirose and R. Okazaki, PNAS 69, 1863 (1972). 137. W. Goebel and H. Schrempf, Nature N B 245,39 ( 1973). 138. P. F. Spahr and R. F. Gesteland, PNAS 59,876 ( 1968). 139. P. G. N. Jeppesen, S. J. Argetsinger, R. F. Gesteland and P. F. Spahr, Nature (London) 226, 230 ( 1970). 140. G. Bellemare, B. R. Jordan and R. Monier, J M B 71, 307 (1972). 141. W. Szer, BBRC 35, 653 ( 1969). 142. N. B. Hecht and C. R. Woese, J. Bacteriol. 95, 986 ( 1968). 143. M. Adesnik and C. Levinthal, J M B 46, 281 (1969). 144. B. G. Forget and B. Jordan, Science 167,382 ( 1970). 145. J. E. Darnell, Jr., Bacteriol. Reu. 32, 262 (1968). 146. R. H. Burdon, B. T. Martin and B. M. Lal, J M B 28,357 ( 1967). 147. S. Altman, Nature N B 229, 19 (1971). 148. S. Altman and J. D. Smith, Nature N B 233, 35 ( 1971). 149. A. Dahlberg and A. C. Peacock, J M B 55,61 ( 1971). 150. M. Sogin, B. Pace, N. R. Pace and C. R. Woese, Nature N B 232,48 (1971). 151. G. G. Brownlee and E. Cartwight, Nature N B 232, 50 (1971). 152. C. V. Lowry and J. R. Dahlberg, Nature N B 232,53 ( 1971). 153. F. Hayes, D. Heyes, P. Fellner and C. Ehresmann, Nature N B 232, 54 (1971). 154. A. Yuki, J M B 56,435 ( 1971). 155. A. Yuki, J M B 62,321 (1971). 156. G. Corte, D. Schlessinger, D. Longo and P. Venkov, J M B 60, 325 (1971). 157. H. D. Robertson, S. Altman and J. D. Smith, JBC 247, 5243 (1972). 158. K. P. Schilfer, S. Altman and D. SOU, PNAS 70,3626 ( 1973). 159. B. Roe, M. Michael and B. Dudock, Nature N B 246, 135 (1973). 160. T. Boon, PNAS 69,549 (19‘72). 161. B. Meyhack, I. Meyhack and D. Apirion, PNAS 70, 156 ( 1973). 161a. M. Nomura, ARM 21,257 (1967). ,161b. C. M. Bowman, J. E. Dahlberg, T.Ikemura, J. Konisky and M. Nomura, PNAS 68, 964 ( 1971). 161c. B. W. Senior and J. B. Holland, PNAS 68,959 (1971). 161d. T. Boon, PNAS 68,2421 (1971). 162. F. Gros, J. Dubert, A. Tissieres, S. Burgeois, M. Michelson, R. Soffer and L. Legault, CSHSQB 28,299 ( 1963). 163. S. K. Niyogi and A. Stevens, JBC 240,2593 (1965). 164. K. M. Downey and A. G. So, Bchem. 9,2520 ( 1970). 165. D. J. Hoffman and S. K. Niyogi, PNAS 70,574 (1973). 166. A. Stevens and S. K. Niyogi, BBRC 29,550 (1967). 167. A. K. Datta and S. K. Niyogi, Fed. Proc. 34, 699 ( 1975). 168. S. K. Niyogi and A. K. Datta, JBC 250,7307 ( 1975). 169. A. K. Datta and S. K. Niyogi, JBC 250,7313 ( 1975). 170. E. P. Anderson and L. A. Heppel, BBA 43,79 ( 1960).

128. 129. 130. 131. 132. 133. 134.

BACTERIAL RIBONUCLEASES

305

171. J. W. Chase and C. C. Richardson, JBC 249,4545 ( 1974). 172. J. W. Chase and C. C. Richardson, JBC 249,4553 ( 1974). 173. S. Nishimura, in “Procedures in Nucleic Acid Research” (G. L. Cantoni and D. R. Davies, eds. ), p. 56. Harper & Row, New York, 1966. 174. P. €7. Whitfeld and H. Witzel, B B A 72, 262 ( 1963). 175. G . W. Rushizky, A. E. Grew, R. W. Hartley, Jr., and H. A. Sober, Bchem. 2, 787 ( 1963 ) 176. G . C. Brownlee, F. Sanger and B. G. Barrell, J M B 34, 379 (1968). 177. F. Harada, F. Kimura and S. Nishimura, B B A 195,590 ( 1969). 178. K. Oda, F. Kimura, F. Harada and S. Nishimura, B B A 179, 97 (1969). 179. S. Nishimura, and B. Maruo, B B A 40,355 ( 1960). 180. M. Yamasaki and K. Arima, B B A 139,202 (1967). 181. M. Yamasaki and K. Arima, BBRC 37,430 ( 1969). 182. A. Stevens and R. J. Hilmoe, JBC 235,3016 (1960). 183. M. L. Sogin and N. R. Pace, Nature (London) 252,598 ( 1974). 184. R. Monier, J. Feunteun, B. Forget, B. Jordan, M. Reynier and F. Varricchio,

.

CSHSQB 34, 139 (1969). 185. F. Galibert, P. Tiollais, F. Sanforche and M. Boiron, EJB 20, 381 (1971). 186. B. Meyhack, I. Meyhack and D. Apirion, FEBS Lett. 49, 215 (1974). 187. D. Nakada and B. Magasanik, J M B 8, 105 (1964). 188. H. Hiatt, F. Gros and F. Jacob, B B A 72, 15 ( 1963). 189. D. P. Fan, A. Higa, and C. Levinthal, J M B 8,210 (1964). 190. A. Kepes, B B A 76, 293 ( 1963). 191. M. Sekiguchi and S. S. Cohen, JBC 238,349 ( 1963). 192. T. Andoh and D. Mizuno, J . Biochem. (Tokyo) 54,237 ( 1963). 193. S. H. Barondes and M. W. Nirenberg, Science 138,810 ( 1962). 194. M. Futai, Y. Anraku and D. Mizuno, B B A 119,373 (1968). 195. T. Kivity-Vogel and D. Elson, BBRC 33,412 (1968). 196. E. T. Lennette, L. Gorelic and D. Apirion, PNAS 68, 3140 (1971). 197. M. Simon and D. Apirion, Nature N B 239, 79 (1972). 198. L. Gorelic and D. Apirion, FEBS Lett. 30, 133 (1973). 199. J. J. Castles and M. F. Singer, J M B 40, 1 ( 1969). 200. N. Morikawa, and F. Imamoto, Noture (London) 223,37 ( 1969). 201. S. G . Chaney and P. D. Boyer, J M B 64,581 ( 1972). 202. D. E. Morse, R. Mosteller, R. F. Baker and C. Yanofsky, Noture (London)

223, 40 (1969). 203. T. Schwartz, E. Craig and D. Kennell, J M B 54,299 (1970). 204. R. D. Mosteller, J. K. Rose and C. Yanofsky, C S H S Q B 35, 461 (1970). 205. B. L. Marrs and C. Yanofsky, Nature N B 234, 168 ( 1971). 206. M. Blundell, E. Craig and D. Kennell, Nature N B 238, 46 ( 1972). 207. J. Forchharnmer, E. N. Jackson and C. Yanofsky, J M B 71, 687 (1972). 208. D. Kennell and I. Bicknell, J M B 74, 21 ( 1973). 209. A. Kepes, B B A 138, 107 (1967). 210. D. Kennell and C. Simmons, J M B 70,451 (1972). 211. G. S. Stent, Proc. Roy Soc., Ser. B 164, 181 (1966). 212. D. H. Shin and K. Moldave, J M B 21,231 (1966). 213. 0. W. Jones, M. Dieckmann and P. Berg, J M B 31, 177 (1968). 214. J. C. Brown and P. Doty, BBRC 30,284 ( 1968). 215. M. Revel, M. Herzberg, A. Becarevic and F. Gros, J M B 33, 231 (1968). 216. U. Maitra and J. Hurwitz, PNAS 54, 815 (1965). 217. H. Bremer, M. W. Konrad, K. Gaines and G. S. Stent, J M B 13, 540 (1965).

306

ALOK K. DATTA AND SALIL K. NIYOGI

218. 0.L. Miller, Jr., B. A. Hamkalo and C. A. Thomas, Jr., Science 169, 392 ( 1970). 219. D. L. Engelhardt, R. E. Webster and N. D. Zinder, JMB 29, 45 (1967). 220. M. Blundell and D. Kennell, JMB 83,143 (1974). 221. D. Achord and D. Kennell, J M B 90,581 ( 1974). 222. L. W. Lim and D. Kennell, Fed. Proc. 34, 675 (1975). 223. J. E. Toivonen and D. P. Nierlich, Nature (London) 252, 74 (1974). 224. D. Apirion, Mol. Gen. Genet. 122, 313 (1973). 224a. J. P. Richardson, C. Grimley and C. Lowery, PNAS 72, 1725 (1975). 225. T. Carter and A. Newtcn, PNAS 68,2962 ( 1971). 226. E. Z. Ron, R. E. Kohler and B. D. Davis, PNAS 56,471 ( 1966). 227. M. Kuwano, D. Schlessinger and D. E. Morse, Nature (London) 231, 214 (1971). 228. P. Venkov, L. Waltschewa and D. Schlessinger, Funct. Units Protein Biosynth., Fed. Eur. Biochem. SOC. Meet., 7th, 1971 p. 379 (1972). 229. L. Silengo, N. Nikolaev, D. Schlessinger and F. Imamoto, Mol. Cen. Genet. 134, 7 (1974). 230. D. Apirion and N. Watson, Mol. Gen. Genet. 136, 317 (1975). 231. J. J. Dunn and F. W. Studier, PNAS 70, 1559 (1973). 232. J. J. Dunn and F. W. Studier, PNAS 70,3296 ( 1973). 233. J. J. Dunn and F. W. Studier, Brookhawen Symp. B i d . 26, 267 (1974). 234. R. A. Kramer, M. Rosenberg and J. A. Steitz, J M B 89, 767 (1974). 235. M. Rosenberg, R. A. Kramer and J. A. Steitz, J M B 89, 777 (1974). 236. H. D. Robertson and J. J. Dunn, JBC 250, 3050 (1975). 237. W. H. McClain, C. Guthriew and B. G. Barrell, PNAS 69,3703 (1972). 237a. T. C. Pinkerton, G. Paddock and J. Abelson, Nature NB 240, 88 (1972). 237b. G. Paddock and J. Abelson, Nature N B 246,2 ( 1973). 237c. G. V. Paddock, K. Fukada, J. Abelson and H. D. Robertson, Personal communication ( 1975). 238. K. Hercules, M. Schweiger and W. Sauerbier, PNAS 71, 840 (1974). 239. J. J. Dunn and F. W. Studier, Personal communication ( 1975). 240. 0. Makela, P. H. Miikela and S. Soikkeli, Ann. M e d . Exp. B i d . Fenn. 42, 188 ( 1964). 241. S . C. Weatherford, L. Rosen, L. Gorelic and D. Apirion, JBC 247, 5404 ( 1972). 242. N. Nikolaev, L. Silengo and D. Schlessinger, JBC 248, 7967 (1973). 243. N. Nikolaev, L. Silengo and D. Schlessinger, PNAS 70, 3361 (1973). 244. N. Nikolaev, D. Schlessinger and P. K. Wellauer, J M B 86, 741 (1974). 245. N. Nikolaev, M. Birenbaum and D. Schlessinger, BBA 395, 478 (1975). 246. P. Kindler, T. U. Keil and P. H. Hofschneider, 126, 53 ( 1973). 247. D. Apirion and N. Watson, Mol. Cen. Genet. 132,89 (1974). 248. P. Gegenheimer and D. Apirion, JBC 249,2407 ( 1974). 249. W. F. Doolittle and N. R. Pace, PNAS 68, 1786 ( 1971). 250. D. Pettijohn, Nature NB 235,204 (1972). 251. H. Bremer and L. Berry, Nature NB 234, 81 ( 1971). 252. J. Feunteun, B. R. Jordan and R. Monier, J M B 70,465 ( 1972). 253. H. A. Raue and M. Gruber, B B A 232,314 ( 1971). 254. H. Ikeda, Abstr. Mol. Biol. Meet., Osaka,Japan, P28, 1973. 255. T. G. Vickers and J. E. M. Midgley, Nature NB 233, 210 (1971). 256. P. Schedl and P. Primakoff, PNAS 70,2091 ( 1973). 257. H. Sakano, S . Yamada, T. Ikemura, Y. Shimqra and H. Ozeki, NARes. 1, 355 (1974).

BAaERIAL RIBONUCLEASES

307

258. P. Schedl, P. Primakoff and J. Roberts, Brookhauen Syrnp. Bid. 26, 53 (1974). 259. R. H. Hilderman and M. P. Deutscher, BBRC 54,205 ( 1973). 260. S. Altman and H. D. Robertson, Mol. Cell. Biochem. 1, 83 (1973). 261. S. Altman, Cell 4, 21 (1975). 262. E. K. Bikoff and M. L. Gefter, JBC 250, 6240 (1975). 263. E. K. Bikoff, B. F. LaRue and M. L. Gefter, JBC 250, 6248 (1975). 263a. H. Sakano and Y. Shimura, PNAS 72,3369 (1975). 264. F. C. Neidhardt, This Series 3, 145 ( 1964). 265. B. Pace, R. L. Peterson and N. R. Pace, PNAS 65, 1097 (1970). 266. H. T. Abelson, L. F. Johnson, S. Penman and H. Green, Cell 1, 161 (1974). 267. T. Andoh and D. Mizuno, J. Biochern. (Tokyo) 54,237 ( 1963). 268. T. Horiuchi, S . Horiuchi and D. Mizuno, BBA 31,570 (1959). 269. H. Maruyama and D. Mizuno, BBA 108,93 (1965). 270. S. Natori, R. Nozawa and D. Mizuno, BBA 114,245 (1966). 271. A. L. Koch, Aduan. Microb. Physiol. 6, 147 ( 1971). 272. R. A. Lazzarini and R. M. Winslow, CSHSQB 35,383 (1970). 273. P. Donini, J M B 72, 553 ( 1972). 274. R. Nozawa, T. Horiuchi and D. Mizuno, ABB 118,402 (1967). 275. R. Kaplan and D. Apirion, JBC 249,149 ( 1974). 276. R. Kaplan and D. Apirion, JBC 250,1854 ( 1975). 277. R. Kaplan and D. Apirion, JBC 250,3174 (1975). 278. S. Pederson and N. 0. Kjkldgaard, MoZ. Gen. Genet. 118, 85 (1972). 279. H. Kersten and H. M. Rauen, Nature (Londoti) 190, 1195 (1963). 280. K. Nose, D. Mizuno and H. Ozeki, BBA 119,636 (1966). 281. E. T. Lennette, B. Meyhack and D. Apirion, FEBS Lett. 21, 286 ( 1972). 282. J. S. Roth, in "Methods in Cancer Research" (H. Busch, ed.), Vol. 3, p. 154. Academic Press, New York, 1967. 283. A. D. Smith and H. Winkler, Nature (London) 207, 634 ( 1965). 284. N. S. Girija and A. Sreenivasan, BJ 98, 562 ( 1966). 285. T. Utsunoniiya and J. S. Roth, J . Cell Bid. 29,395 ( 1966). 286. K. Shortman, BBA 55, 88 (1962). 287. H. Priess and W. Z. Zillig, 2. Physiol. Chem. 348, 817 ( 1967). 288. G. Blobel and V. R. Potter, PNAS 55, 1283 (1966). 289. J. R. Smeaton and W. H. Elliot, BBA 145, 547 (1967). 290. D. Mizuno and Y. Anraku, Jap. J . Med. Sci. B i d . 20, 127 (1967). 291. H. C. Neu and L. A. Heppel, BBRC 14, 109 ( 1964). 292. W. H. McClain, C. Guthrie and B. G. Barrell, J M B 81, 157 (1973). 293. R. A. Felicioli, S . Sensi, F. Marmocchi, G. Falcone, and P. L. Ipata, Bchem. 12, 547 ( 1973). 294. S. Sensi, G. Falcone, P. L. Ipata and R. A. Felicioli, Bchern. 13, 5008 (1974). 295. K. A. Cammack and H. E. Wade, B j 96,671 (1965). 296. C. B. Bruni, M. M. Rechler and R. G. Martin, J. Bacteriol. 113, 1207 (1973). 297. C. T. Wehr, J. Bacten'ol. 14, 96 (1973). 298. A. M. Reiner, J. Bacteriol. 97, 1431 (1969). 299. A. M. Reiner, J . Bacteriol. 97, 1437 (1969). 300. E. T. Lennette and D. Apirion, J . Bacteriol. 108, 1322 ( 1971). 301. A. L. Taylor, Bacteriol. Reo. 34, 155 (1970). 302. A. L. Taylor and C. D. Trotter, Bacteriol. Reo. 31, 332 (1967). 303. E. Cerda-Olmedo, P. C. Hanawalt and N. Guerola, J M B 33, 705 (1968). 304. T. G. Kinsherf, Y. F. Lee and D. Apirion, NARes. 1, 1439 (1974).

308

ALOK K. DATTA AND S A L E K. NNOGI

305. D.Ratner, Nature (London) 259,151 (1976). 306. F. W. Studier, J . Bad. 124, 307 (1975). 307. D.Apirion and N. Watson, J. Bact. 124,317 (1975). The literature survey for this review was completed before September 30, 1975. NOTESADDEDIN PROOF 1. The observation of Richardson et d. (2240) that rho could be the product of the SuA gene has been confirmed by recent observations by Ratner (305).Affinity chromatography .of rho protein from different E. coli strains reveals that various SuA mutations affect rho activity in different ways: one class produces an unstable rho, a second class apparently increases the molecular weight of the polypeptide, while others, presumed to be missense mutations, overproduce rho. Making use of molecular weight differences between rhos isolated from different E. coli strains, Ratner shows that the rho gene is tightly linked to the SuA locus. ( See p. 292.) 2. The mutation that causes RNase I11 deficiency in strain AB 301-105 of Kindler ( 2 4 6 ) has now been mapped (306, 307). The mutation, called rnc 105 by Apirion and Watson (307), lies between nadB and purI on the genetic map of E. coZi. The precise mapping of the RNase I11 mutation could make possible the construction of isogenic RNase 111' and 111- strains of known genetic background and might make more precise the study of RNA metabolism in both E. cdi and bacteriophage T7. (See p. 299.)

Subject Index historical, 150-151

C

Chloramphenicol, puromycin reaction and, 234-238

E Enzyme ( s ) regulation, 99-100 classical mode, 100-103 postclassical mode: energy, 103-105 postclassical mode: nitrogen, 105-

113 Erythromycin, binding to ribosomes,

230-234 G Guanosine 5,3'-polyphosphate synthesis in uitro reaction definition of position of transfer, 7-8 mechanism of pyrophosphate transfer, 5-7 ribosomal system, 2 5 role of ribosome, 8-9 nonribosomal reversibilty of transfer and, 13 stringent factor and, 9-13

N

1955-1960, 151-152 1960-1965, 152-155 1965-1970, 155-156 1970-1975, 156 multiple, 166-167 proteins, 165-166 reaction ion requirements, 157-158 kinetics, 164-165 polynucleotide requirements, 158-

159

primer specificity, 159-164 reaction products, 164 substrates, 157 as polyadenylating reagents, 174-175 regulaton of, 173-174 ribonucleic acid polymerase and,

169-171 viral, 171-173 Polyadenylate sequences decrease in size comparison of cytoplasmic and nuclear, 128 cytoplasm and, 128-129 discovery as component of mRNA, 121-123 cytoplasmic, 120-121 rat liver cytoplasm, 121 refractory to phenol extraction,

119-120

Nitroso compounds biological activity, 249 direct acting, 250-251 indirect acting, 251-254 nucleic acid interactions and, 263-

elongation actinomycin-treated cells, 134137 biosynthesis in nucleus, 133-134 in cytoplasm, 137-138 differences in levels, 139 in nucleus, 138-139 isolation adsorption on nitrocellulose filters,

268

Nucleic acid( s ) alkylation animals in uiuo, 258-263 bacteria, viruses and chemical systems, 254-257 tissue or organ cultures, 257-258

125-126

assay, 127 conditions for phenol extraction,

123-125

other adsorption techniques, 126-127 messenger RNA structure and interaction with polysome polynucleotides, 141-142

P Polyadenylate polymerase( s ) cellular location, 168-169

309

3 10

SUBJECT INDEX

protein binding sites, 140-141 significance in messenger RNA stability and, 144-145 translation and, 145-146 transport and, 144 widespread occurrence in, 143-144 size and location on mRNA alkaline hydrolysis, 127-128 exonucleases and, 128 sedimentation analysis, 127 steady-state size distribution in cytoplasm, 132133 size distribution of long-labeled, 130 total by annealing with polyuridylate, 130-132 Polyamines chemical and physicochemical studies binding studies, 34-38 cation effects on p-chloromercuribenzoate tRNA, 29-30 cation effects on 4-thiouridine, 23 chemical reactions of 4-thiouridine, 25-28 reactions of 4-thiouridine, in tRNA, 21-23 reactions of Co'+-tRNA, 28-29 spin-label studies, 30-34 thermal denaturation studies, 24-25 roles of, binding to cellular components, 17-19 biolo&cal and biochemical functions, 18-17 transfer RNA and cation binding, 19-21 Protein biosynthesis, summary of, 217-222

R Ribonuclease(s ) bacterial Escherichio coli ribonuclease I, 272-277 Escherichio coli ribonuclease 11, 279-281 Escherichia coli ribonuclease 111, 281-282 oligoribonuclease, 285-287 others, 287-288 polynucleotide phosphorylase, 277279

ribonuclease IV, 284 ribonuclease H, 282-284 ribonuclease P and others, 284-285 mutants lacking, physiology of, 298299 re lation of,298-298 r o c i n cell physiology, 288-289 messenger RNA and, 289-293 ribosomal RNA and, 293-294 stable RNA degradation and, 298 transfer RNA processing, 295-296 Ribosome( s ) cation specificity, 93-96 conformational changes large; 85-91 restricted, 91-93 function, models of, 234-238 macromolecular interactions, 80-84 proteins, 78-80 states of, 223-230

S Semen endo enous DNA polymerase, 65-86 is0 ation and characteristics, 66-70 possible role, 70-71 Sperm u take by somatic cells generi method for study autoradiographic data, 55-57 electron microscopic evidence, 49-55 heat-killed sperm uptake, 64-85 information transfer and fetal gene products, 61-84 living sperm and recipient cells, 45-48 morphological changes and oncogenesis, 57-81 Sistrand, the translational unit, 238-244

f

T Transfer ribonucleic acid( s ) classification of, 184-186 correlation between structure, physical and chemical data chemical modification, 204 complementary oligonucleotide binding, 201 conformation in solution and crystal, 206-207 fluorescence energ transfer, 200-201 methyl and methy ene proton resonance, 203 nuclear magnetic resonance spectroscopy, 201-202

r

311

SUBJECT INDEX

small angle X-ray scattering, 199200 tritium exchange, 204-208 UV-induced cross-link, 203-204 functional implications of 3-D structure interaction with other RNA's, 211-212 protein recognition, 207 synthetase interaction, 207-21 1 invariant and semivariant bases in, 186-187 life cycle, 182-184 mature, modified nucleosides in, 187

secondary structure, 184 tertiary structure, 187-188 base-stacking s stems, 194-195 conformation olnucleosides, 195-196 general 3-D structure, 196-199 hydrogen bonding, 190-193 L-sha e, 190 optica analog of crystallographic method, 188-189 other crystallographic studies, 199 other tertiar interactions, 194 X-ray CrystaGography, 189-190 Translocation inhibitors of, 223

P

Contents of Previous Volumes Volume 1 "Primer" in DNA Polymerase Reactions

F. J. BOLLUM The Biosynthesis of Ribonucleic Acid i n Animal Systems

R. M. S. SMELLIE The Role of DNA i n RNA Synthesis JERARD

HURWITZAND J. T. AUGUST

Polynucleotide Phosphorylase

M. GRUNBERG-MANAGO Messenger Ribonucleic Acid

FRITZLIPMANN The Recent Excitement in the Coding Problem

F. H. C. CRICK Some Thoughts on the Double-Stranded Model of Deoxyribonucleic Acid

AARONBENDICHAND HERBERTS. ROSENKRANZ Denaturation and Renaturation of Deoxyribonucleic Acid

J. MARMUR,R. ROWND,AND C. L. SCHILDKRAUT

Some Problems Concerning the Macromolecular Structure of Ribonucleic Acids

A. S. SPIRIN The Structure of DNA as Determined by X-Ray Scattering Techniques

V ~ R I LUZZATI O Molecular Mechanisms of Radiation Effects

A. WACKER AUTHORINDEX-SUBj ~ c rINDEX Volume 2 Nucleic Acids and Information Transfer

LIEBE F. CAVALIERIAND BARBARAH. ROSENBERC Nuclear Ribonucleic Acid

HENRY HARRIS

312

CONTENTS OF PREVIOUS VOLUMES

Plant Virus Nucleic Acids

ROY MARKHAM The Nucleases of Escherichia coli

I. R. LEHMAN Specificity of Chemical Mutagenesis

DAVID R. KRIEG Column Chromatography of Oligonucleotides and Polynucleotides

MA~THYSSTAEHELIN Mechanism of Action and Application of Azapyrimidines

J. SKODA The Function of the Pyrimidine Base in the Ribonuclease Reaction

HERBERT WITZEL Preparation, Fractionation, and Properties of sRNA

G. L. BROWN

AUTHORINDEX-SUBJECTINDEX Volume 3 Isolation and Fractionation of Nucleic Acids

K . S. KIRBY Cellular Sites of RNA Synthesis

DAVIDM. PRESCOTT Ribonucleases in Taka-Diastase: Properties, Chemical Nature, and Applications

FUJIOEGAMI,KENJI TAKAHASHI, AND TSUNEKO UCHIDA Chemical Effects of Ionizing Radiations on Nucleic Acids and Related Compounds

JOSEPH J. WEISS The Regulation of RNA Synthesis in Bacteria

FREDERICK C. NEIDHARDT Actinomycin and Nucleic Acid Function

E. REICH AND I. H. GOLDBERC De Novo Protein Synthesis in Vitro

B. NISMANAND J. PELMONT

313

314

CONTENTS OF PREVIOUS VOLUMES

Free Nucleotides in Animal Tissues

P. MANDEL AUTHORINDEX-SUBJECX INDEX Volume 4 Fluorinated Pyrimidines

CHARLES HEIDELBERGER Genetic Recombination in Bacteriophage

E. VOLKIN DNA Polymerases from Mammalian Cells

H. M. KEIR The Evolution of Base Sequences i n Polynucleotides

B. J. MCCARTHY Biosynthesis of Ribosomes in Bacterial Cells

SYOZOOSAWA 5-Hydroxymethylpyrimidines and Their Derivatives

T. L. V. ULBRIMT Amino Acid Esters of RNA, Nucleotides, and Related Compounds

H. G. ZACHAU AND H. FELDMANN Uptake of DNA b y living Cells

L. LEDOUX AUTHORINDEX-SUBJECX

INDEX

Volume 5 Introduction to the Biochemistry of D-Arabinosyl Nucleosides

SEYMOUR S. COHEN Effects of Some Chemical Mutagens and Carcinogens an Nucleic Acids

P. D. LAWLEY Nucleic Acids in Chloroplasts and Metabolic DNA

TATSUICHI IWAMURA Enzymatic Alteration of Macromolecular Structure

P. R. SRINIVASAN AND ERNEST BOREK Hormones and the Synthesis and Utilization o f Ribonucleic Acids

J. R. TATA

CONTENTS OF PREVIOUS VOLUMES

315

Nucleoside Antibiotics

JACKJ. Fox, KYOICHIA. WATANABE, AND ALEXANDER BLOCH Recombination of DNA Molecules

CHARLES A. THOMAS, JR. Appendix I. Recombination o f a Pool o f DNA Fragments with Complementary Single-Chain Ends

G. S. WATSON, W. K. SMITH,AND CHARLES A, THOMAS, JR. Appendix II. Proof That Sequences o f A, C, G, and T Can Be Assembled to Produce Chains o f Ultimate length, Avoiding Repetitions Everywhere

A. S. FRAENKEL AND J. GILLIS The Chemistry of Pseudouridine

ROBERT WARNER CHAMBERS The Biochemistry o f Pseudouridine

EUGENE GOLDWASSER AND ROBERTL. HEINRIKSON AUTHORINDEX-SUBJECX INDEX Volume 6 Nucleic Acids and Mutability

STEPHENZAMENHOF Specificity in the Structure of Transfer RNA

KIN-ICHIROMIURA Synthetic Polynucleotides

A. M. MICHELSON, J. MASSOULI~, AND W. GUSCHLBAUER The DNA of Chloroplasts, Mitochondria, and Centrioles

S. GRANICK AND AHARONGIBOR Behavior, Neural Function, and RNA

H. HYD$N The Nucleolus and the Synthesis o f Ribosomes

ROBERTP. PERRY The Nature and Biosynthesis o f Nuclear Ribonucleic Acids

G. P. GEORGIEV Replication of Phage RNA

CHARLES WEISSMANN AND SEVERO OCHOA AUTHORINDEX-SUBJECX INDEX

316

CONTENTS OF PREVIOUS VOLUMES

Volume 7 Autoradiographic Studies on DNA Replication in Normal and leukemic Human Chromosomes

FELICEGAVOSTO Proteins of the Cell Nucleus

LUBOMIRS. HNILICA The Present Status of the Genetic Code

CARLR. WOESE The Search for the Messenger RNA of Hemoglobin

H. CHANTRENNE, A. BURNY,AND G. MARBAIX Ribonucleic Acids and Information Transfer in Animal Cells

A. A. HADJIOLOV Transfer of Genetic lnformction during Embryogenesis

MARTINNEMER Enzymatic Reduction of Ribonucleotides AND PETERREICHARD ACNELARSSON

The Mutagenic Action of Hydroxylamine

J. H. PHILLIPSAND D. M. BROWN Mammalian Nucleolytic Enzymes and Their localization

DAVID SHUCAR AND HALINASIERAKOWSKA AUTHORINDEX-SUBJ E INDEX ~ Volume 8 Nucleic Acids-The

First Hundred Years

J. N. DAVIDSON Nucleic Acids and Protamine in Salmon Testes

GORDONH. DIXONAND MICHAELSMITH Experimental Approaches to the Determination o f the Nucleotide Sequences of large Oligonucleotides and Small Nucleic Acids

ROBERTW. HOLLEY Alterations of DNA Base Composition in Bacteria

G. F. GAUSE Chemistry of Guanine and Its Biologically Significant Derivatives

ROBERTSHAPIRO 4x1 74 and Related Viruses ROBERTL. SINSHEIMER

Bacteriophage

CONTENTS OF PREVIOUS VOLUMES

The Preparation and Characterization of l a rg e Oligonucleotides

GEORGE W. RUSHIZKYAND HERBERT A. SOBER Purine N-Oxides and Cancer

GEORGE BOSWORTHBROWN The Photochemistry, Photobiology, and Repair of Polynucleotides

R. B.

SETLOW

What Really Is DNA? Remarks on the Changing Aspects of a Scientific Concept

ERWINCHARGAFF Recent Nucleic Acid Research in China

TIEN-HSICHENCAND ROY H. Do1

AUTHOR INDEX-SUB JECT INDEX

Volume 9 The Role of Conformation in Chemical Mutagenesis

B. SINGER AND H. FRAENKEL-CONRAT Polarographic Techniques in Nucleic Acid Research

E. P A L E ~ K RNA Polymerase and the Control of RNA Synthesis

JOHNP. RICHARDSON Radiation-Induced Alterations i n the Structure of Deoxyribonucleic Acid and Their Biological Consequences

D. T. KANAZIR Optical Rotatory Dispersion and Circular Dichroism of Nucleic Acids

JENTSI YANC

AND

TATSUYA SAMEJIMA

The Specificity of Molecular Hybridization in Relation to Studies on Higher Organisms

P. M. B. WALKER Quantum-Mechanical Investigations of the Electronic Structure of Nucleic Acids and Their Constituents

BERNARD PULLMAN AND ALBERTEPULLMAN The Chemical Modification of Nucleic Acids

N. K. KOCHETKOVAND E. I. BUDOWSKY AUTHOR INDEX-SUB JECT INDEX

317

318

CONTENTS OF PREVIOUS VOLUMES

Volume 10 Induced Activation of Amino Acid Activating Enzymes by Amino Acids and tRNA

ALAN H. MEHLER Transfer RNA and Cell Differentiation

NOBORUSUEOKAA N D TAMIKO KANO-SUEOKA N"- (A'-lsopentenyl) adenosine: Chemical Reactions, Biosynthesis, Metabolism, and Significance to the Structure and Function of tRNA

Ross H. HALL Nucleotide Biosynthesis from Preformed Purines in Mammalian Cells: Regulatory Mechanisms and Biologicai Significance

A. W. MURRAY,DAPHNE C. ELLIOTT; AND M. R. ATKINSON Ribosome Specificity of Protein Synthesis in Vitro

ORIO CIFERRIAND BRUNOPARISI Synthetic Nucleotide-peptides

ZOE A. SHABAROVA The Crystal Structures of Purines, Pyrimidines and Their Intermolecular Complexes

DONALD VOET AND ALEXANDERRICH AUTHORINDEX-SUBJECT INDEX Volume 11 The Induction of Interferon by Natural and Synthetic Polynucleotides

CLARENCE COLBY,JR. Ribonucleic Acid Maturation in Animal Cells

R. H. BURDON Liporibonucleoprotein as an Integral Part of Animal Cell Membranes

V. S . SHAPOTAND S. YA. DAVIDOVA Uptake of Nonviral Nucleic Acids by Mammalian Cells

PUSHPA M. BHARCAVA AND G. SHANMUCAM The Relaxed Control Phenomenon

ANN M. RYANAND ERNESTBOREK Molecular Aspects of Genetic Recombination

CEDRIC I. DAVERN

319

CONTENTS OF PREVIOUS VOLUMES

Principles and Practices of Nucleic Acid Hybridization

DAVIDE. KENNELL Recent Studies Concerning the Coding Mechanism

THOMAS H.

JUKES AND

LILAGATLIN

The Ribosomal RNA Cistrons

M. L. BIRNSTIEL,M. CHIPCHASE, AND J. SPEIRS Three-Dimensional Structure o f tRNA

FRIEDRICH CRAMER Current Thoughts on the Replication of DNA

ANDREWBECKERAND

JERARD

HURWITZ

Reaction of Aminoacyl-tRNA Synthetases with Heterologous tRNA's

K. BRUCEJACOBSON O n the Recognition o f tRNA by Its Aminoacyl-tRNA Ligase

ROBERTW. CHAMBERS AUTHOR INDEX-SUB JEGT INDEX Volume 12

Ultraviolet Photochemistry as a Probe of Polyribonucleotide Conformation

A. J. LOMANTAND

JACQUES

R. FRESCO

Some Recent Developments in DNA Enzymology

MEHRANGOULIAN Minor Components in Transfer RNA: Their Characterization, Location, and Function

SUSUMUNISHIMURA The Mechanism o f Aminoacylation o f Transfer RNA

ROBERTB. LOFTFIELD Regulation of RNA Synthesis

EKKEHARD K. F. BAUTZ The Poly(dA-dT) of Crab

M. LASKOWSKI, SR. The Chemical Synthesis and the Biochemical Properties of Peptidyl-tRNA

YEHUDA LAPIDOT AND NATHAN DE GROOT SUBJECTINDEX

320

CONTENTS OF PREVIOUS VOLUMES

Volume 13 Reactions of Nucleic Acids and Nucleoproteins with Formaldehyde

M. YA. FELDMAN Synthesis and Functions of the -C-C-A Terminus of Transfer RNA

MURRAYP, DEUTSCHER Mammalian RNA Polymerases

SAMSONT.

JACOB

Poly( adenosine diphosphate ribose)

TAKASHI SUGIMURA The Stereochemistry of Actinomycin Binding to DNA and Its Implications in Molecular Biology

HENRYM. SOBELL Resistance Factors and Their Ecological Importance to Bacteria and to Man

M. H. RICHMOND Lysogenic Induction

ERNESTBOREKAND ANN RYAN Recognition in Nucleic Acids and the Anticodon Families JACQUES

NINIO

Translation and Transcription of the Tryptophan Operon

FUMIOIMAMOTO lymphoid Cell RNA's and Immunity

A. ARTHURG O ~ I E B SUBJECT

INDEX

Volume 14 DNA Modification and Restriction

WERNER AREER Mechanism of Bacterial Transformation and Transfection

NIHALK. NOTANIAND

JANE

K. SETLOW

DNA Polymerases II and 111 of Escherichia coli

MALCOLML. GEFTER The Primary Structure of DNA

KENNETHMURRAYAND ROBERTW. OLD RNA-Directed DNA Polymerase-Properties and Functions in Oncogenic RNA Viruses and Cells

MAURICEGREENAND GRAYF. GERARD SUBJECXINDEX

CONTENTS OF PREVIOUS VOLUMES

321

Volume 15 Information Transfer in Cells Infected by RNA Tumor Viruses and Extension to Human Neoplasia

D. GILLESPIE,W. C. SAXINGER,AND R. C. GALLO Mammalian DNA Polymerases

F. J. BOLLUM Eukaryotic RNA Polymerases and the Factors That Control Them

B. B. BISWAS,A. GANGULP,AND D. DAS Structural and Energetic Consequences o f Noncomplementary Base Oppositions in Nucleic Acid Helices

A. J. LOMANTAND

JACQUES

R. FRESCO

The Chemical Effects o f Nucleic Acid Alkylation and Their Relation to Mutogenesis and Carcinogenesis

B. SINGER Effects of the Antibiotics Netropsin and Distamycin A on the Structure and, Function of Nucleic Acids

CHRISTOPH ZIMMER SUBJECX INDEX

Volume 16 Initiation of Enzymic Synthesis o f Deoxyribonucleic Acid by Ribonucleic Acid Primers

ERWINCHARGAFF Transcription and Processing of Transfer RNA Precursors JOHN

D. SMITH

Bisulfite Modification of Nucleic Acids and Their Constituents

HIKOYA HAYATSU The Mechanism of the Mutagenic Action of Hydroxylamines

E. I. BUDOWSKY A B C 0

6 7 8 9

E O

F 1

C 2

H 3

1 4 J 5

Diethyl Pyrocarbonate in Nucleic Acid Research

L. EHRENBERG, I. FEDORCS~K, AND F. SOLYMOSY

SUBJECT INDEX

This Page Intentionally Left Blank