PROGRESS IN
Nucleic A c i d Research
and M o l e c u l a r Biology Volume 22
This Page Intentionally Left Blank
P...
3 downloads
205 Views
18MB 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
PROGRESS IN
Nucleic A c i d Research
and M o l e c u l a r Biology Volume 22
This Page Intentionally Left Blank
PROGRESS IN
Nucleic Acid Research and Molecular Biology edited by
WALDO E. COHN Biology Division Oak Ridge National Laborutory Ouk Ridge, Tennessee
Volume 22 7 979 ACADEMIC PRESS New York
Sun Francisco
London
A Suhsidiury of Harcourt Bruce ]ovnnovich, Publishers
COPYRIGHT @ 1979, BY ACADEMIC PRESS, INC. ALL RIGHTS RESERVED. NO PART OF THIS PUBLICATION MAY BE REPRODUCED OR TRANSMITTED IN ANY 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.
1 1 1 Fifth Avenue, New York,New York 10003
United Kingdom Edition published by ACADEMIC PRESS, INC. (LONDON) LTD. 24/28 Oval Road, London NW17DX
LIBRARY OF CONGRESS CATALOG CARD NUMBER: 63-15847 ISBN 0-1 2-540022-5 PRINTED IN W E UNITED STATES O F AMERICA
79808182
9 8 7 6 5 4 3 2 1
Contents LIST OF CONTRIBUTORS .......................................................
ix
ABBREVIATIONS AND SYMBOLS ................................................
xi
.............................
XV
SOME ARTICLES PLANNED FOR FUTURE VOLUMES
The -C-C-A End of tRNA a n d Its Role in Protein Biosynthesis Mathias Sprinzl and Friedrich Cramer I . Introduction .......................................................... I1. Structure of the -N-C-C-A Terminus ..................................... I11. Structure of Aminoacyl-tRNA .......................................... IV. Enzymic Modification of the 3’ End of tRNA ........................... V. Aminoacylation of tRNA ............................................... VI Positional Specificity of Aminoacylation and Chemical “Proofreading” ....................................................... VII . Binding of Aminoacyl-tRNA to Ribosomes .............................. VIII . Ribosomal Peptidyltransferase Center .................................. References ...........................................................
.
2 3 10 14 23
39 46 59 63
The Mechanism of Action of Antitumor Platinum Compounds J . J . Roberts and A . J . Thomson I . Introduction .......................................................... 71 I1. Chemical Features of Platinum Drugs ................................. 75 I11. Biological Effects of PIatinum Coordination Complexes Indicative of Reactions with DNA ..................................... 85 IV. Biochemical Effects of Platinum Complexes Indicative of Reactions with DNA ................................................ 88 V. Interaction of Platinum Compounds with DNA ......................... 94 VI Repair of DNA Damage Induced by Platinum Complexes in Vivo ....... 110 VII . Concluding Remarks .................................................. 128 References ........................................................... 129
.
D N A Glycosylases. Endonucleases for Apurinic/ Apyrimid inic Sites. a n d Base Excision- Repair Tomas Lindahl I . Introduction .......................................................... I1. Models for Excision-Repair of DNA .................................... V
135 136
vi
CONTENTS
I11. DNA Clycosylases .................................................... IV. Endonucleases for Apurinic/Apyrimidinic Sites (AP Endonucleases) ...... V Repair of Apurinic Sites in DNA by Alternative Pathways ............... References ...........................................................
.
145 173 187 188
Naturally Occurring Nucleoside a n d Nucleotide Antibiotics
.
Robert J Suhadolnik Introduction .......................................................... I . Inhibitors of Protein Synthesis ......................................... I1 Inhibitors of RNA Synthesis ........................................... 111. Inhibitors of DNA Synthesis. DNA Viruses. and’ RNA Viruses ............ IV. Inhibitors of Adenosine Deaminase and Effectors of t h e Immune Response .............................................. V. Inhibitors of Cell-Wall Synthesis and Antifungal Agents ................. VI . Inhibitors of Purine and Pyrimidine Interconversions ................... VII . Hyperesthetic and Hyperemic Agents .................................. VIII Cyclic-AMP Phosphodiesterase Inhibitors .............................. IX . Miscellaneous Naturally Occurring Nucleosides ........................ References ...........................................................
.
.
193 196 209 239 245 249 266 269 269 270 272
Genetically Controlled Variation in the Shapes of Enzymes George Johnson I . Introduction .......................................................... I1. Internal Standardization ............................................... I11. The Interaction of Protein Size and Charge during Electrophoresis ....... IV. Detecting “Hidden” Variation in Shape ................................ V. The Nature of Cryptic Variants ........................................ VI . Major Unresolved Issues: Posttranslational Modification ................. VII . Novel Approaches to Electrophoresis .................................. References ...........................................................
293 294 295 303 305 315 321 325
Transcription Units for mRNA Production in Eukaryotic Cells and Their DNA Viruses James E . Darnell. Jr
.
I. Introduction .......................................................... I1 Techniques for Defining Transcription Units ........................... I11. Definition of Transcription Units for mRNA ............................
.
327 329 331
vii
CONTENTS
IV. The Precursor Relationship of Large Nuclear Primary Transcripts to mRNA ............................................................. V. Recent Evidence on the Formation of mRNA from Primary Transcripts . . . VI . Models for the Regulation of Eukaryotic Gene Expression ............... VII Conclusion ........................................................... References ........................................................... Note Added in Proof ..................................................
.
337 339 344 349 350 353
............................................................
355
CONTENTS OF PREVIOUSVOLUMES..........................................
358
SUBJECTINDEX
This Page Intentionally Left Blank
List of Contributors Numbers in parentheses indicate the pages on which the authors' contributions begin.
FRIEDRICH CRAMER(l), Max-Planck-Znstitut fiir experimentelle Medizin, Abteilung Chemie, Hermann-Rein-Strasse 3, 0-3400 Gottingen, Germany JAMES E. DARNELL, JR. (327), The Rockefeller University, New York, New York 10021 GEORGEJOHNSON (293),Department of Biology, Washington Uniuersity, S t . Louis, Missouri 63130 TOMAS LINDAHL( 135),Department of Medical Chemistry, Uniuersity of Gothenburg, 400 33, Gothenburg, Sweden J. J. ROBERTS (71),Znstitute of Cancer Research, Royal Cancer Hospital, Pollards Wood Research Stution, Nightingales Lane, Chalfont Street, Giles, Bucks., United Kingdom MATHIAS SPIUNZL(l),Max-Planck-Znstitut fur experimentelle Medixin, Abteilung Chemie, Hermann-Rein-Strasse 3, 0-3400 Gottingen, Germany ROBERT J. SUHADOLNIK(193),Department of Biochemistry, Temple University School of Medicine, Philadelphia, Pennsylvania 19140 A. J . THOMSON (71), School of Chemical Sciences, University of East Anglia, Norwich, Norfolk, United Kingdom
ix
This Page Intentionally Left Blank
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 (1,2) and compendia(3) in four languages and are available in reprint form from the 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 (1) 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. I. 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 (Yrd), 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 b y any one of the nucleoside symbols given in 11-1below. 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; BtlcAMP = dibutyryl CAMP, etc. 11. Oligonucleotides a n d Polynucleotides 1. Ribonucleoside Residues
(a) Common: A, G, I, X, C , T, 0, U, ‘P, R, Y, N (in the order of 1-2 above). (b) Base-modified: sI or M for thioinosine = Bmercaptopurine ribonucleoside; SU or S for thiouridine; brU or B for Sbromouridine; h U or D for 5,Bdihydrouridine; i for isopentenyl; f for formyl. Other modifications are similarly indicated by appropriate lower-case prefixes (in contrast to 1-1 above) (2,3). (c) Sugar-modified: prefixes are d, a, x, or 1as in 1-2 above, alternatively, by italics or boldface type (with definition) unless the entire chain is specified by an appropriate prefix. The 2’-O-methyl group is indicated by sufix m (e.g., -Am- for 2’-0methyladenosine, but -mA- for Bmethyladenosine). (d) Locants and multipliers, when necessary, are indicated by superscripts and subscripts, respectively, e.g., -mfA- = Bdimethyladenosine; -s4U- or -4S- = 4 thiouridine; -ac4Cm- = 2‘-O-methyl-4acetylcytidine. (e) When space is limited, as in two-dimensional arrays or in aligning homologous sequences, the prefixes may b e placed ouer the capitol letter, the suffixes over the phosphodiester symbol. xi
xii
ABBREVIATIONS AND SYMBOLS
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; A p is adenosine 3‘-phosphate; C > p is cytidine 2‘ : 3‘-cyclic phosphate (1, 2, 3); p < A is adenosine,3’ : 5’-cyclic phosphate. (b) Internal: hyphen (for known sequence), comma (for unknown sequence); unknown sequences are enclosed in parentheses. E.g., pA-C-A-C(Ct,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’-5’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-to-left arrows. Polarity in any direction, as in a two-dimensional array, may b e 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; poly(3 adenylate, 2 cytidylate) = poly(A&) or (A3,C2)”,an irregular copolymer of A and C in 3: 2 proportions; poly(deoxyadeny1ate-deoxythymidylate)= poly[d(A-T)] or poly (dA-dT) or (dAdT), or d(A-T),, an alternating copolymer of d A 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, e.g., A, . dT,2-18. 111. Association of Polynucleotide Chains 1. Associated (e.g., H-bonded) chains, or bases within chains, are indicated by a center dot (not a hyphen or aplus sign) separating the complete names or symbols, e.g.: or A, * U, poly(A) . poly(U) A, * 2U, or poly(A) . 2 poly (U) A poly(dA-dC) . poly(dG-dT) or (dA-dC), . (dG-dT),. 2. Nonassociated chains are separated by the plus sign, e.g.:
or
2[poly(A) . PO~Y(U)] + poly(A) * 2 PoIY(U)+ PO~Y(A) 2[A, . U,] + A, . 2U, + A,.
3. Unspecified or unknown association is expressed by a comma (again meaning “unknown”) between 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.
xiii
ABBREVIATIONS AND SYMBOLS
IV. Natural Nucleic Acids RNA DNA mRNA; rRNA; nRNA hnRNA D-RNA; cRNA mtDNA tRNA
ribonucleic acid or ribonucleate deoxyribonucleic acid or deoxyribonucleate messenger RNA; ribosomal RNA; nuclear RNA heterogeneous nuclear RNA “DNA-like’’ RNA; complementary RNA mitochondria1 DNA transfer (or acceptor or amino-acid-accepting) RNA; replaces sRNA, which is not to b e used for any purpose aminoacyl-tRNA “charged” tRNA (i.e., tRNA’s carrying aminoacyl residues); may b e abbreviated to AA-tRNA alanine tRNA or tRNA normally capable of accepting alanine, to form tRNAAia,etc. alanyl-tRNA alanyl-tRNA or The same, with alanyl residue covalently attached. alanyl-tRNAAia [Note: fMet = formylmethionyl; hence tRNAfMe‘,identical with tRNAP‘] Isoacceptors are indicated by appropriate subscripts, i.e., tRNAfia, tRNAtLa,etc. V. Miscellaneous Abbreviations
pi, ppi inorganic orthophosphate, pyrophosphate RNase, DNase ribonuclease, deoxyribonuclease melting temperature (“C) t m (not T m ) Others listed in Table I1 of Reference 1 may also b e used without definition. No others, with or without definition, are used unless, in the opinion of the editor, 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. JBC 241,527 (1966); &hem 5,1445 (1966);BJ 101,1(1966);ABB 115,1(1966),129, 1 (1969); and elsewhere.! 2. EJB 15, 203 (1970);JBC 245, 5171 (1970);JMB 55, 299 (1971); and elsewhere.! 3. “Handbook of Biochemistry” (G. Fasman, ed.), 3rd ed. Chemical Rubber Co., Cleveland, Ohio, 1970, 1975, Nucleic Acids, Vols. I and 11, pp. 3-59. 4. “Enzyme Nomenclature,” Elsevier Scientific Publ. Co., Amsterdam, 1973, and Supplement No. 1, BBA 429, (1976).
* Contractions
for names of journals follow.
!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.
xiv
ABBREVIATIONS AND SYMBOLS
5. “Nomenclature of Synthetic Polypeptides,” JEC 247, 323 (1972); Biopolymers 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 Nucleic Acid Research Proc. Nat. Acad. Sci. U.S. Proc. SOC.Exp. Biol. Med. Progr. Nucl. Acid Res. Mol. Biol.
Abbreviations used ARB ABB BBRC Bcheni Bj BBA CSHSQB EJB FP JACS J. Bact. JBC JCS JMB Nature NB NARes 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 Chromatin Structure a n d Function
P. CHAMBON Functional Aspects of the Interaction o f Chemical Carcinogens with Nucleic Acids
D. GRUNBERGER AND I. B. WEINSTEIN Ribonucleotide Reductase
F. D. HAMILTON Mechanism o f Interferon Action
P. LENGYEL AND G. SEN Mitochondria1 Nucleic Acids of Yeast
M. RABINOWITZAND J. LOCKER Ribosome Turnover in Eukaryotic Cells a n d Tissues
J. F. SCOTT Patterns of N ucleic Acid Synthesis in Physorum polycephalum
G. TURNOCK Structure a n d Functions of Ribosomal RNA
R. ZIMMERMANN
xv
This Page Intentionally Left Blank
The -C-C-A End of tRNA and Its Role i n Protein Biosynthesis MATHIAS SPRINZLAND FRIEDRICH CRAMER
’
Max-Planck-Institutfur experimentelle Medizin Abteilung Chemie Gottingen. Germany
I . Introduction .................................................... I1 . Structure of the -N-C-C-A Terminus .............................. A . X-Ray Diffraction Studies .................................... B . Chemical Modification ....................................... C . Physicochemical Studies ..................................... I11. Structure of Aminoacyl-tRNA .................................... A Positional Isomers of Aminoacyl-tRNA ........................ B . Conformation of the Aminoacyl Residue ....................... C . Influence of the Aminoacyl Residue on the Conformation oftRNA ..................................................... IV. Enzymic Modification of the 3’ End of tRNA ..................... A . Shortened tRNAs ............................................ B. Incorporation of Modified Nucleotides with ATP(CTP) :tRNA Nucleotidyltransferase ...................... C . Incorporation of Modified Nucleotides with Polynucleotide Phosphorylase ................................ V. Aminoacylation of tRNA ......................................... A . Substrate Properties of Modified tRNAs ....................... B. Role of the 3’-Terminal Adenosine during the Interaction of tRNA and Aminoacyl-tRNA Synthetawe ..................... C . Site of Aminoacylation of tRNA ............................... VI Positional Specificity of Aminoacylation and Chemical “Proofreading” .............................................. A . Site of Aminoacylation and Mischarging ....................... B. “Proofreading” by Aminoacyl-tRNA Synthetases, a .......... Chemical Event at the N-C-C-A End . . . . . . . VII . Binding of Aminoacyl-tRNA to Ribosomes . . . . . A . Formation of (EF-Tu) . GTP . (Aminoacyl-tRNA) Complexes .... B. (EF-Tu)-Dependent Binding of Aminoacyl-tRNA to Ribosomes . C . Transacylation from 2’ to 3’ Controls the Selection Process ..... VIII . Ribosomal Peptidyltransferase Center ...... , ..................... A Peptide-Bond Formation ..................................... B. Interaction with the -C-C-A End .............................. References .................. .............................. 1
.
.
.
Progress in Niicleic Acid Research and Molecular Biology. Vol . 22
2 3 4 4 6 10 10 11 12 14 14 16 22 23 23
28 29 39 39
45 47 51 57 59 59 60 63
.
Copylight @ 1979 by Academic Press Inc. All tights of reproduction in any form reserved. ISBN 012-540022-5
2
MATHIAS SPRINZL AND FRIEDRICH CRAMER
1. Introduction’ The adaptor role of tRNA requires that it have at least two functionally different sites-one for the specific interaction with the messenger, and one for accepting and transferring the particular amino acid to the growing polypeptide chain. The codon recognition mechanism of tRNA is explained in principle by the codon-anticodon interaction, including the wobble hypothesis (]), and has been confirmed by sequencing many tRNAs. The second function is much less clearly understood. Apart from the long-known fact that the -N-C-C-A end, common to all tRNAs, carries the amino acid in ester linkage (2 ), the problem of the specific attachment of an amino acid to its corresponding tRNA (the “recognition” problem) is still not solved ( 3 ) . Clearly, the -N-C-C-A end is not responsible for the specificity of the attachment; there must b e other features in the molecule or in the enzymic mechanism of aminoacylation responsible for this high degree of specificity. However, these characteristics-whatever they are-must be seen in relation to the -N-C-C-A end, since this part of the tRNA carries the amino acid. Thus the presence of an intact -N-C-C-A end is a general prerequisite for the aminoacylation and for amino-acid transfer. In this article, we discuss the structural and the chemical requirements for the functioning of the -N-C-C-A terminus. [In this connection, it should be mentioned that some viral RNAs can be aminoacylated. They also carry an -N-C-C-A end (4).2] In the “cloverleaf’ structure, the -C-C-A end always remains unpaired. The nature of the nucleotide prior to the -C-C-A terminus has been discussed in connection with a preselecting code that would Abbreviations: tRNAPhe= tRNAPhe-A-C-C-A= phenylalanine transfer RNA Phe-tRNAPhe-A-C-C-A= phenylalanyl-tRNAphe tRNAPhe-A-C-s*C-A, tRNAPhe-A-C-i5C-A, tRNAPhe-A-C-C-F,etc., are tRNAPhespecies containing, respectively, 2-thiocytidine, 5-iodocytidine, formycin, etc. tRNAPhe-A-C-C-dA,tRNAPhe-A-C-C-A(2‘NH2), etc., are tRNAphespecies containing, etc. respectively, 2’-deoxyadenosine, 2’-amino-2’-deoxyadenosine, tRNAPhe-A-C-C-Aox, and tRNAphe-A-C-C-Aoxf-rd represent tRNAPh‘after periodate oxidation, and after subsequent borohydride reduction, respectively tRNA-N-C-C-A and corresponding formulas represent unfractionated mixtures of tRNAs Aminoacyl-oligonucleotide nomenclature is as follows: C-A-Phe indicates C-A bearing a phenylalanine residue on the 2’ or 3’ hydroxyl group; C-A(Z’Phe)H and C-A(2’H) Phe indicate C-A bearing phenylalanine residues at the 2’- and 3’-position, with hydrogen in positions 3’ and 2‘, respectively. * See article by Waters and Mullin in Vol. 20 of this series [Ed.]. I
THE
-C-C-A END
OF
tRNA
3
govern the lipophilic or hydrophilic character of the amino acid to be attached (5). However, this hypothesis no longer seems tenable in view of the many new sequences that have been obtained during recent years. Although the -N-C-C-A end does not seem to be involved in the secondary (or tertiary) base-pairing, it might nevertheless exist in a defined physical conform,at’ion. The -N-C-C-A end does not merely serve as a spacer between the body of the tRNA molecule and the amino-acid residue, providing a sufficiently exposed position for the amino acid during the interaction with the aminoacyl-tRNA synthetase and on the ribosomal sites. Rather, it plays some highly specific roles: in the interaction with aminoacyl-tRNA synthetase, in the selection mechanism for the correct amino acid, and in the specific attachment to the ribosome. An earlier review in this series by Deutscher (6) on the acceptor end of tRNA emphasized mainly the biosynthesis of the -N-C-C-A end and the function of the ATP(CTP) : tRNA nucleotidyltransferase. The present article summarizes recent results on the function of the -N-C-C-A end of tRNA as a reactive site during aminoacylation and ribosomal protein biosynthesis, with information about the structure of -N-C-C-A end and changes in its function after its modification.
I I . Structure of the -N-C-C-A Terminus All tRNAs sequenced to date can be presented in a “cloverleaf” structure in which the 3’-terminal -N-C-C-A is not base-paired. The general existence of such a structure is evident from a comparison of the sequences of different tRNAs (7) and has been proved directly by high-resolution nuclear magnetic resonance (NMR) studies ( 8 ) .An important feature of this structure is the invariant number of nucleotides between the T-q-C sequence and the terminal adenosine: the terminal adenosine is always separated from the first nucleoside in the T-q-C loop (ribothymidine or uridine) by 21 nucleosides (7). This invariance underlines the functional importance of the single-stranded T-q-C loop and the 3’-terminal sequence (9). Several chemical and physicochemical studies seemed to indicate that the -N-C-C-A terminus of the tRNA in solution may be involved in tertiary interactions within the tRNA molecule ( 3 ) . I n view of the results that have accumulated in recent years, these interpretations are most probably not valid. However, the structure of the 3’ terminus in free and aminoacylated tRNA with respect to stability, stacking interactions, and conformations still remains to be determined.
4
MATHIAS SPRINZL AND FRIEDRICH CRAMER
A. X-Ray Diffraction Studies Successful interpretation of X-ray diffraction data of orthorhombic (10)and monoclinic (11,12)crystals of tRNAPhefrom yeast up to a 2.5 A resolution led to the elucidation of a three-dimensional crystal structure of this tRNA (for recent reviews, see 13-15). Atomic coordinates and torsional angles have been published for the corresponding three-dimensional models (16, 17). The -A-C-C-A terminus in these models is attached as a single strand to a long double-stranded helical region formed by the base pairs of the T-9-C and the aminoacyl stems. There are no additional tertiary interactions in which the -A-C-C-A end is involved. Unfortunately, the data concerning this part of the molecule are not sufficiently satisfactory to allow an unambiguous interpretation of the conformation of the -A-C-C-A end in the crystals (13).This could be due to a conformational perturbation of this singlestranded region or by an inhomogeneity of the terminal sequence with respect to the presence of the terminal adenosine. In the orthorhombic crystals, the 3'-terminal residues are helically stacked onto the aminoacyl stem except for the terminal adenosine-76, which is not stacked with respect to the penultimate cytidine-75 (16). In the monoclinic cell, the CT5and A,6 residues are in a more extended form (1 7 ) .
B. Chemical Modification One of the earliest approaches to elucidation of the structure of tRNA in solution was its chemical modification by base-specific reagents (18-37), followed by sequence analysis of the product. Using this approach, the -N-C-C-A terminus and some other single-stranded regions of tRNA showed a high reactivity toward certain reagents. This was demonstrated by the reaction of tRNA with bisulfite (18-26), monoperphthalic acid (27, 28), methoxylamine (29-32, 35), sodium periodate (33, 34), and hydroxylamine (36),and by tritium exchange experiments (37). Treatment of yeast tRNApheby monoperphthalic acid rapidly converts the terminal adenosine-76 to adenosine 1-oxide (27), but the unpaired adenosine-73 does not react. The same reaction has been used to modify yeast tRNAVa'and tRNAPhe(28).From the rates of the reaction of particular nucleosides at the 3' end of both tRNAs, it appears that after the modification of the 3'-terminal adenosine there is N-oxidation of the two penultimate cytidine residues. When the three terminal bases of the -A-C-C-A end are finally oxidized, the fourth unpaired base becomes reactive. This sequential step-by-step oxida-
THE
-C-C-A
END OF
tRNA
5
tion does not occur in single-stranded oligonucleotides, such as C-A-C-C-A or A-A-A-U-C-A-C-C-A. Similar observations were made during the investigation of the rate of conversion of cytidine to uridine in yeast tRNAPhe(38) and E . coli tRNAGIY (23)by NaHS03. Out of six cytidine residues in the unpaired regions of yeast tRNAPhe,only the two at the -A-C-C-A end were reactive. The relative rates of modification of each individual reactive residue in tRNAPhe,compared to those of single nucleotides and of the random copolymer poly(Ct,U,), are shown in Fig. 1. This demonstrates that the cytidine residues in the -A-C-C-A end, not being basepaired, react at a lower rate than does CMP or the random copolymer. This implies that the tRNA sites are involved in some kind of ordered structure that is probably determined by stacking interactions. Varying degrees of modification of the two cytidine residues were also observed in some other cases using bisulfite (18)or hydroxylamine (36), although in these investigations the rates of modification of particular residues were not determined.
PC
1pC-U -U -U-U-U-U-I random
161
FIG.1.Modification ofcytidine with sodium bisulfite (38).The numbers in the boxes give the half-times (in hours) of the reaction converting cytidine to uridine for CMP, for the irregular copolymer poly(C,,UJ, for the cytidine residues in a dodecanucleotide excised from the anticodon of tRNAPhefrom yeast, and for the cytidine residues in intact yeast tRNAPhe.Reaction conditions were as in Chambers et al. (18).
6
MATHIAS SPRINZL AND FRIEDRICH CRAMER
The results obtained by enzymic degradation of the 3’-terminus are similar. The action of polynucleotide phosphorylase (39) as well as that of snake venom phosphodiesterase (40) on tRNA results in stepwise degradation of the 3‘ end of its polynucleotide chain. Again the rate of enzymic degradatidn decreases as the reaction progresses from the terminal adenosine to the penultimate cytidine and further from the 3’ end. The removal of the 3’-terminal phosphate from tRNA-NC-Cp is much faster than from tRNA-N-Cp ( 4 1 ) . Different rates of degradation of the nucleotides at the 3’ terminus can also be explained by lower steric accessibihty of these residues near the doublestranded region of the aminoacylation stem of tRNA. However, although such a steric hindrance could be the reason for the observed differences in the enzymic treatment of the -N-C-C-A end of tRNAs, where relatively large molecules have to approach the particular site, it is unlikely to be the case for chemical modifications with small reagents, such as bisulfite, hydroxylamine, or monoperphthalic acid. A different approach in determining the structure of the regions of tRNA not involved in the base-pairing interactions, which proved to be very useful in study of the solution conformation of tRNA, is equilibrium dialysis of tRNA with oligonucleotides complementary to its primary sequence (42, 43, and references therein). Single-stranded regions that are free in the three-dimensional model of tRNAPhe,as elucidated by X-ray crystallography, are also accessible to the binding of complementary oligonucleotides (42).The binding of the tetranucleotide complementary to the 3’-terminal -A-C-C-A region was even stronger than the binding of the appropriate tetranucleotides to the anticodon region, again indicating the accessibility and structural organization of the 3’ end. The comparison of the binding of U-G-G, complementary to the -C74-C75-A76 of tRNAPhe,with the binding of G-G-U, complementary to the -A73-C,4-C75sequence, is of special interest: the association constant of the G-G-U is about seven times larger than that of U-G-G (42).This finding is in agreement with the results obtained b y chemical modifications of the 3’ terminus of tRNAPhe-A-C-C-Adiscussed above (28), which indicate an involvement of residues 73, 74, and 75 in stacking interactions. As shown b y the modification of Cmazof tRNAPhe-A-C-C-A(32),the stacking interactions decrease the chemical reactivity of the bases involved, but they increase the ability of the particular region to enter into hydrogen bonding with complementary nucleotides (42). C. Physicochemical Studies Great progress has been made in the last few years in measurements and interpretations of the NMR spectra of tRNA molecules.
THE
-C-C-A END
OF
tRNA
7
In principle, two approaches have been used for the elucidation of secondary or tertiary structure: (a) in the low-field region of the NMR spectrum, the resonances of the hydrogen-bonded NH-hydrogen atoms have been identified and assigned to particular secondary (8 )and tertiary (44-47) interactions; (b) the high-field signals arising from the resonances of methyl and methylene groups of the modified bases have been assigned ( 48, 49) . No additional N H resonances that could arise from interactions of the -N-C-C-A nucleotides with other parts of the tRNA have been found. Since the -N-C-C-A end contains no modified base, high-field NMR has so far not been applied to studies of this region. However, it is possible to incorporate modified methylated nucleotides or nucleotides containing other suitable nuclei into the 3' terminus by enzymic methods as discussed below (Section IV) and then to measure a particular newly evolved resonance. Using this approach, the 31Presonance of the last phosphodiester linkage of tRNAphe-A-C-CgA(50)was easily assigned, owing to the large differences between the chemical shifts of 31Pin the phosphodiester and in the phosphorothiodiester bond (51).Similarly, the high-field NMR spectra of aminoacyl-tRNA may provide information about conformation and mobility of the aininoacyl residue on the aminoacylated tRNA species. The P-methylene group of the phenylalanine attached to yeast tRNAPhecan easily be assigned (49).The CH, resonance in phenylalanine is not broadened at low temperature. This indicates that the -C-C-A end of Phe-tRNAPheis not associated with any other part of the molecule in the native form. Taking advantage of the fact that the -N-C-C-A end of tRNA can be specifically modified, several investigators have attached a "spinlabel" to this part of the molecule. Rich et al. acylated the a-amino group of Val-tRNAVa'-A-C-C-Aor Phe-tRNAPhe-A-C-C-Afrom E . coli by a spin-label reagent (52,53);later, Sprinzl et al. (54)used alkylation of the 2-thio group of the residue 75 of tRNAPhe-A-C-s'C-A(55)for spinlabeling. Caron and Dugas (56) introduced a spin-label into unfractionated tRNA from Escherichia coli in which the terminal adenosine had been oxidized b y periodate, by reaction of an amino derivative of a nitroxyl radical. At low temperatures, all spin-labeled tRNAs showed spectra characteristic of moderate immobilization of the nitroxyl radical. When the label was bound directly to the -N-C-C-A end (54, 57), the rate of its tumbling, as monitored b y ESR spectra, changed abruptly during the temperature-induced melting at temperatures about 10-30°C lower than the UV-monitored temperature transition, measured under the same conditions. This is an indication that the bases at the 3' terminus participate in some kind of an
8
MATHIAS SPRINZL AND FRIEDRICH CFUMER
ordered structure that is disrupted at temperatures where the conformations of other regions of the molecule are still intact. This melting does not reflect a disruption of the secondary structure of the aminoacyl stem, which, in the case of yeast tRNAPhe,begins to melt only at significantly higher temperatures (58).It is also interesting to note that in the case where unfractionated tRNA, spin-labeled at the 3’ end, was used for the melting experiments (54),a very sharp transition at 65°C was still observed. This behavior must therefore be correlated with a structural feature common to all tRNAs tather than to the structure of the aminoacyl stem, which is unique for each species (7). An increase in the mobility of a spin-label need not, in principle, depend on a local perturbation, but could also reflect the melting of a tertiary structure leading to a greater freedom of movement of the stems of the tRNA (54). I n the case where the nitroxyl spin-label was attached to the amino acid of aminoacyl-tRNAs, the ESR-monitored temperature transition occurred at the same temperature as the UV-monitored melting (52, 53). However, these experiments were performed in the absence of Mg2+,under which conditions the secondary structure melts at lower temperatures, so that the melting of the aminoacyl stem and of the -N-C-C-A structure could occur simultaneously. Valuable information about the local structure of the 3’-terminus of tRNA has been obtained by fluorescence measurements, using fluorescence labels attached to this part of the molecule. Formycin, a fluorescent analog of adenosine, has been incorporated enzymically into the 3’-terminal position of mixed rat liver tRNAs (59),resulting in a strong decrease in the quantum yield of the formycin fluorescence (60).This, together with the fact that a temperature-dependent transition observed in the fluorescence-monitored melting curves of the tRNA-NC-C-F occurred at a temperature lower than the t , based on absorbance, was interpreted as a possible melting of the interaction of the 3’-terminal formycin with other parts of the tRNA molecule. This also suggests an ordered involvement of the 3’ terminus in the threedimensional structure of tRNA (60). Maelicke et al. reinvestigated the spectroscopic properties of a tRNAPhefrom yeast with 3’-terminal formycin (61). By comparison of the quantum yields of the tRNAPhe-A-C-C-Ffluorescence with those of the oligonucleotide C-A-C-C-F fluorescence, which are almost identical, it became apparent that the quenching of formycin fluorescence results only from the stacking interactions between the formycin and adjacent nucleosides. No difference was found in the fluorescence vs. temperature profiles of FMP, C-A-C-C-F, and tRNAPhe-A-C-C-F.The minor anomalies in the melting curves (60) and pK values of FMP in
THE
-C-C-A
END OF
tRNA
9 ?
9
:..
FIG.2. Models for the structure of the -N-C-C-A terminus of a tRNA. Left: stacked form; middle: terminal nucleoside destacked as in the case of tRNAPhe-A-C-C-F,j-,M; right: extended form with no stacking interactions.
free form and as a part of tRNA (61) therefore probably result from increased stacking interactions in the -A-C-C-F sequence due to the adjacent double-helical region of the acceptor stem. This stacking interaction can be decreased by periodate oxidation of the 3’-terminal ribose residue of tRNAPhe-A-C-C-F(61). Obviously, opening of the terminal ribose ring increases the confonnational mobility of the sugar-phosphate backbone and consequently causes destacking of the -A-C-C-F end, as shown in Fig. 2. A comparison of the kinetics of the temperature-induced conformation change of the tRNAPhe-A-CC-F monitored by changes in the fluorescence of the wyosine3 in the anticodon loop and of the FMP residue in the 3’ terminus yielded data consistent with a structure in which the 3‘-terminal nucleotides are free and not involved in tertiary interactions (62). An important observation about the contribution of the 3’-terminal sequence to the total structure of tRNA was made by Beltchev et al. (63),who compared the thermal melting curves of tRNAPhefrom yeast lacking different numbers of nucleotides from the 3‘ end. The t , was shifted to a significantly lower value when two terminal nucleotides were missing, becoming more pronounced with a broad transition when three to five terminal nucleotides were removed. Removal of a single terminal nucleotide from yeast tRNAPhedoes not lead to such an effect (62). Summarizing the information concerning the structure of the -N-C-C-A end of an unaminoacylated tRNA, it seems apparent that this single-stranded part of the molecule possesses an ordered structure 3
[Ed.]
Wyosine (W) is the nucleoside of the base wye (originally called “base Y”) (7).
10
MATHIAS SPRINZL AND FFUEDFUCH CRAMER
that is stabilized by vertical stacking interactions through the aminoacyl stem (Fig. 2). Such a structure may be important for determining the steric arrangement of the 3’-terminal adenosine. I n addition, it would facilitate the interaction of the -N-C-C-A end with other macromolecular partners (such as aminoacyl-tRNA synthetases, elongation factors, and ribosomes) during the functional cycle of tRNA. With minimal changes in the physical environment, the -N-C-C-A end could then expand or contract, and thus serve as a mobile instrument able to be accommodated in its different binding sites (Fig. 2).
111. Structure of Aminoacyl-tRNA A. Positional Isomers of Aminoacyl-tRNA In aminoacyl-tRNA, the amino acid is attached by an energy-rich ester bond (AG = -7 to -8 kcal) (64)to the 2‘,3’-cis-diol system of the terminal adenosine residue. Alkaline hydrolysis or a nucleophilic attack on the carbonyl group of this ester bond is facilitated by the presence of a neighboring hydroxyl group (65, 66). The rates of hydrolysis of a particular aminoacyl residue from different monomeric pyrimidine or purine nucleotides or from aminoacyl-tRNA are similar (67, 68), indicating that neither the terminal base nor the polynucleotide chain of the aminoacyl-tRNA essentially influences the reactivity of this ester bond. On the other hand, there are significant differences in the rates of alkaline hydrolysis for different amino acids from aminoacyl-tRNAs (69-71 ). The presence of the free cis-vicinal hydroxyl group allows a migration of the aminoacyl residue between the 2’- and 3’-hydroxyl group (Fig. 3).The rate of this isomerization in
t1+.2.10-~ sec *
4 0, Lc=o kH-NHz
k
~0%~ QH
@yJ “QA
0 0
FH-NHz R
‘C‘
~d F-NH~ R
FIG.3. Migration of the aminoacyl residue of aminoacyl-tRNA between the 2’- and 3‘-position of the terminal adenosine (66).
THE
-C-C-A END OF tRNA
11
aqueous solution at p H 7 and 37°C is lo5 times the rate of hydrolysis (66). The half-life of equilibration of a specifically aminoacylated adenosine (66, 72) or appropriate model substances (73) has been determined. The most conclusive study (66) measured the rate of equilibration of 3’-O-formyladenosine by NMR spectroscopy. The half-life found was 1.8 x low4 sec, which can be taken as an approximate value for the half-life of equilibration of an average aminoacyltRNA in a neutral buffered medium at 37°C. Under these conditions, the ratio of 2’-aminoacyl-tRNA to 3’-aminoacyl-tRNA is about 1: 2 (68, 73-75). The physiological significance of the aminoacyl migration became apparent only in recent years from studies on the specificity of the enzymes involved in protein biosynthesis toward 2’- or 3’aminoacyl-tRNA.
B. Conformation of the Arninoacyl Residue There is little information in the literature about the conformation of the aminoacyl residue of the aminoacyl-tRNA and its possible intramolecular interactions with the polynucleotide chain. The main limitations in making such investigation are the great instability of the ester linkage and the relatively large amounts of substance required for physical measurements. Several authors have approached this problem by the investigation of analogs of aminoacyl-adenosine (7679). Sundaralingam et al. made X-ray (76, 77) and conformational (78) analyses of puromycin. This analog of aminoacyl-tRNA can accept a peptide residue during peptide-bond formation on ribosomes. The conformation of the aminoacyl residue of puromycin is therefore probably similar to that of aminoacyl-tRNA during this particular step. The adjacent molecules in the unit cell of puromycin . 2 HBr crystals exhibit stacking interactions between the alternating base rings and the aromatic tyrosyl rings. This structure may provide a model for a possible interaction of the terminal adenosine residue of the tRNA with the side-chain of aromatic amino acids. Comparison of the circular dichroism (CD) spectra of C-A(2’Phe)H, C-A(2’H)Phe and C-A-Phe (80) revealed that in the second case, where the phenylalanine is bound to the 3’-position of the adenosine residue, a larger base-base overlap takes place than when the phenylalanine is bound to the 2’-position, or when it can migrate between both 2’- and 3’-hydroxyls. Identical CD intensities were observed for both C-A(2‘H)Phe and C-A(S’H)Gly, indicating that this influence on the stacking of the bases is not dependent on the kind of amino acid attached to the 3’-adenosine. Similar conclusions can b e drawn from the measurements of the hypochromicity of the above models at 260 nm (80).
12
MATHIAS SPRINZL AND FRIEDRICH CRAMER
C. Influence of the Aminoacyl Residue on the Conformation of tRNA
An attractive speculation about a functionally dynamic structure of tRNA (81) led several investigators to study the differences between the conformations of free and aminoacylated tRNA. A large variety of physical and biochemical methods including X-ray scattering (82), NMR (83, 84), Raman spectroscopy (85), electron spin resonance (ESR) (86) and fluorescence spectroscopy (87),CD and optical rotatory dispersion (ORD) (8844), laser light-scattering (95), binding of complementary oligonucleotides (9648), binding of intercalating agents (99, loo), sedimentation velocity (101, 102), tritium exchange (103, 104), column chromatography (105), and susceptibility to nuclease digestion (106) were used in attempts to solve this problem. The results are conflicting. In some cases, a conformational difference was deduced from the observations (82,83,85,86,93,95,97-!39,101,102, 105), whereas in others there appeared to be no influence of the aminoacyl residue on the physical or chemical properties of the tRNA (84,87-92,96,103,104,106). Procedures designed to detect changes in the secondary structure, such as ORD, CD (88-92), tritium exchange (103, 104), and NMR measurements of NH resonances (84), gave negative results. Similarly, there were no differences in the susceptibilities of free and aminoacylated tRNAs to nuclease digestion (106), indicating an identical secondary structure for both. Minor changes in the three-dimensional structure of tRNA upon aminoacylation cannot be excluded. Although a conformational change in the T-T-C and hU loops of tRNA was predicted to take place during the interaction of aminoacyltRNA with the ribosomal A site (9), there is no direct evidence for the unfolding of these regions upon aminoacylation. If the interactions between the hU and T-T-Cloops were absent from the aminoacyltRNA, the binding of oligonucleotides complementary to these loops should be comparable with the binding of oIigonucleotides to other single-stranded regions of tRNA, such as the anticodon loop or the -N-C-C-A end. In two cases (96,107), the binding of oligonucleotides complementary to this region does not differ significantly between nonaminoacylated and aminoacylated tRNA species. Contrary to these results, a strong binding of oligonucleotides complementary to the T-Vr-C-G region and to the G-G-G-A sequence in the hU loop of PhetRNAPhe from yeast was found (98), which is not present in nonaminoacylated tRNA (42). Therefore, studies of the binding of oligonucleotides to aminoacyl-tRNA give conflicting results. The reason for this may lie in the different method of preparation of Phe-
THE
-C-C-A END
OF
tRNA
13
tRNAPhe. If a conformational difference between Phe-tRNAphe and tRNAPheexists, it is not understood how it may be maintained when the tRNA is free in solution. It is possible that the thermodynamic barrier between the two conformations is so small that the particular native form of Phe-tRNAPheis not preserved during the preparation and isolation procedure. It is likely that the changes upon aminoacylation observed b y X-ray laser-light scattering (95),sedimentation behavior (101, scattering (82), 102),and an NMR study of manganese binding sites (83) are due to perturbations in the shell of counterions. It is possible that the aminoacylation causes a minor rearrangement of the structure that alters the divalent cation binding sites and the distribution of counterions (95). It appears that at least some conformational changes of the tRNA upon aminoacylation reported in the literature (85,97) may take place at the 3' terminus and may be related to the stacking geometry of the 3'-terminal -N-C-C-A sequence. Although the stacking interactions at the -N-C-C-A end are probably increased by the presence of an amino acid (80),this single-stranded region of the tRNAPhe after aminoacylation is less susceptible to the binding of complementary oligonucleotides than is the free tRNA. This led to a suggestion that the amino group of the attached amino acid may interact with the 5'-terminal phosphate of the tRNA, leading to a steric hindrance of the -N-C-C-A nucleotides toward oligonucleotide binding (96). A similar suggestion was made on the basis of a chemical modification of tRNA (108),showing that the presence of the 5'-terminal phosphate group, although not required for the enzymic aminoacylation reaction, is
FIG. 4. Schematic representation of the acceptor stem region of aminoacyl-tRNA showing the possible interaction of the a-amino group of the amino acid with the 5'phosphate.
14
MATHIAS SPFUNZL AND FRIEDRICH CRAMER
necessary for the recognition of aminoacyl-tRNA by elongation factor Tu. Since it is assumed that this factor checks precisely the presence and conformation of the aminoacyl residue, it is possible that the 5’terminal phosphate plays some role in this process. An aminoacyltRNA bearing a reactive residue on the amino group of the amino acid can be intramolecularly crosslinked with the 5’-phosphate group (109).This finding supports the above-mentioned model (Fig. 4).
IV. Enzymic Modification of the 3’ End of tRNA A. Shortened tRNAs Chemical modification of biologically active molecules and investigation of the properties of such altered species is a widely used approach in biochemical research. Direct and specific modification of the -N-C-C-A end of tRNA by a chemical reagent is possible only by periodate oxidation of the 3‘-terminal ribose of the tRNA. Even this reagent reacts in some tRNA species with queue4 (110,111)and also with certain other residues. All other reagents discussed in Section 11, B that react with the -N-C-C-A end also alter other parts of the tRNA. Such unspecific modifications are therefore of only limited value in the investigation of the role and function of the -N-C-C-A end. An unambiguous modification at the -N-C-C-A end of tRNA can be accomplished b y enzymic methods. Those tRNA species having a partially or completely missing -C-C-A of the -N-C-C-A end can be regenerated enzymically to tRNA-N-C-C-A by incorporation of CMP and AMP residues. Using analogs of AMP or CMP, tRNA species with a modified 3’ end can thus be prepared. The most suitable enzyme for this purpose is the ATP(CTP) :tRNA nucleotidyltransferase (EC 2.7.7.21 and 2.7.7.25) (6).This enzyme, isolated from various sources, incorporates AMP and CMP into shortened tRNA using ATP and CTP as substrates: tRNA-N + 2 CTP
+ ATP atRNA-N-C-C-A+ 3 PP,
Whereas tRNA lacking part or all of its -C-C-A sequence cannot participate in protein synthesis, biological activity is regained in full after such regeneration. There are many analogs of AMP and CMP that can be incorporated into tRNA by means of this enzymic reaction (112). Another method for the preparation of a tRNA with a modified 3’ terminus is the incorporation of altered nucleotides b y a stepwise “Queue” is the name proposed for the base of nucleoside Q (queueosine)(7). [Ed.]
THE
-C-C-A END
OF
tRNA
15
addition catalyzed by polymerizing enzymes, such as polynucleotide phosphorylase (1 13-115). Most recently, a stepwise synthesis of oligonucleotides by T4 pol ynucleotide ligase was reported (107). The low specificity of this latter enzyme with respect to the donor nucleotide may allow incorporation of various nucleotides into the 3' end of tRNA. I n all cases it is necessary to use a shortened tRNA with a well characterized and unique 3' terminus in order to obtain a modified tRNA species in which a single modified nucleotide is placed in a defined position of the -N-C-C-A end. T h e shortened tRNA can be achieved by a limited enzymic degradation of the -N-C-C-A by a 3'exonuclease, such as snake venom phosphodiesterase. Although the rate of degradation decreases significantly on proceeding from the first to the next nucleotide (40), it is not possible to obtain a uniquely shortened tRNA species by this method. Usually a mixture of tRNA-N, tRNA-N-C, and tRNA-N-C-C is obtained in which the ratio of components depends on the conditions of the preparation (116).Such a mixture can be used as a starting material for the preparation of tRNAN-C-C b y incorporation of the missing CMP residues using ATP(CTP) : tRNA nucleotidyltransferase in the absence of ATP. Since an excess of the enzyme leads under these conditions to incorporation of more than two CMP residues ( 6 ) ,a chromatographic separation of tRNA-N-C-C from tRNA-N-C-C(C), is usually necessary to obtain a highly active tRNA-N-C-C (112). tRNA-N and tRNA-N-C can then b e prepared from tRNA-N-C-C by a stepwise degradation involving sodium periodate oxidation, base treatment of the oxidized tRNA, and alkaline phosphatase hydrolysis (41, 117), followed by chromatographic purification of the products on an ion-exchange column (112). Abbreviated tRNAs with a unique sequence can be prepared in this way only from those species that are resistant to sodium periodate at all nucleoside residues except the 3'-terminal cis-diol. Treatment of mixed E . coli tRNAs with periodate decreases the biological activity of certain tRNAs (41,118).The probable site of the side reaction leading to this deactivation may be the rare nucleoside queuosine4(110,111). Other tRNAs, such as yeast tRNAPhe,can be processed b y this method with full retention of biological activity. Yeast tRNAPhe-A,tRNAPhe-AC, and tRNAPhe-A-C-Chave been prepared in a pure form and their end nucleosides have been identified by chromatographic analysis ( 1 1 1 , 116) and gel electrophoresis (119). These abbreviated tRNAPhe species can be aminoacylated by phenylalanyl-tRNA synthetase to the same extent as the native tRNAPhe-A-C-C-Ain a reaction mixture containing CTP, ATP, and ATP(CTP) :tRNA nucleotidyltransferase (55,
16
MATHIAS SPEUNZL AND FRIEDRICH CRAMER
112, 116, 120). Preparation of shortened tRNAs from E . coli by pyrophosphorolysis of the -A-C-C-A end using this enzyme has been reported (121), but a complete removal of the 3' end cannot be achieved by this method.
B. Incorporation of Modified Nucleotides with ATP(CTP) :tRNA Nucleotidyltransferase
The most extensively utilized method for the preparation of tRNAs with an altered 3' end is the incorporation of modifications of the natural nucleotides, AMP and CMP, into shortened tRNAs by ATP(CTP) :tRNA nucleotidyl transferase. The substrate properties of ATP(CTP) : tRNA nucleotidyltransferase with respect to its specificity for triphosphates are also discussed in the review of Deutscher ( 6 ) ,but a large amount of new experimental material has accumulated in recent years. In Table I are listed CTP and ATP analogs that have been investigated as substrates for this enzyme. Adenosine derivatives having a substituent on position 1, 2, 6, or 8 of the heterocyclic ring are not substrates for the yeast enzyme (112, 126). Any tRNA containing 8-thioadenosine, 8-azidoadenosine7or 1:N6-ethenoadenosine (127)at its 3' end would be especially useful for X-ray crystallographic, photoaffinity labeling, or fluorescence spectroscopic experiments, but all attempts to prepare such tRNAs with the transferase were unsuccessful (1 12).Guanosine and inosine 5'-triphosphates are not substrates for the enzyme from rabbit liver (128) or yeast (112); neither are the triphosphates of nucleosides derived from 2-aminopurine or 2,6diaminopurine (129).Thus a free amino group on position 6 of the purine seems to be essential for the incorporation of AMP at the terminal position of tRNA. On the other hand, there is a report on the incorporation of 7-deazanebularin 5'-phosphate into tRNA by the rabbit liver enzyme (130).In this analog of adenosine, the 6-amino group as well as the endocyclic nitrogen-7 are missing. The substrate properties of the 5'-triphosphates of the nucleoside antibiotics tubercidin, toyocamycin, and sangivamycin5 have been investigated (129),using the rat liver enzyme. The fact that these nucleotides are incorporated may make feasible the introduction of various groups into the 3' terminus of tRNA via substitution of the position 7 of 7-deazaadenosine. Formycin, a fluorescent analog of adenosine, was incorporated into tRNAs using enzymes from rat liver (59)or yeast (61, 62), and tRNAs bearing a 3'-terminal formycin have been extenSee article by Suhadolnik in this volume. [Ed.]
THE -C-C-A END OF
17
tRNA
b
a
15
LT
N NH2 V R
AN
O
S
I
e
N
I
d
C
R/S
f
NI
y5 O
N I
h
9
FIG.5. Modifications of the base moieties of ATP and CTP used for elucidation of the substrate abilities of ATP(CTP) :tRNA nucleotidyltransferase: (a) formycin; (b) R = -H, -CN, or -CONHI: tubercidin, toyocamycin, or sangivamycin, respectively; (c) 1 :N6-ethenoadenosine (€-adenosine); (d) R = -Br, -1, -SH, or -Nz: 8-bromo-, 8-iodo-, 8-thio-, or 8-azidoadenosine, respectively; (e) R = -Br, or -1: 5-bromo-, or 5-iodocytidine, respectively; (r) 2-thiocytidine; (g) alkylated 2-thiocytidine, = -CHICONHI or -CI+-CO-NH
-c
N-0; (h) 4-thiouridine.
sively used for spectroscopic (60-62) and enzymic (131) studies (Fig. 5). There are conflicting reports about the substrate properties of deoxyadenosine 5‘-triphosphate (Fig. 6). 2‘-Deoxyadenosine 5’-
HO
H
a
HN , OH
d
H OH
b
HO NH2 C
-OwC -OwC HO H
HO NH2
e
f
FIG. 6. Modifications of the ribose residue of ATP and CTP used for elucidation of the substrate abilities of ATP(CTP) :tRNA nucleotidyltransferase: (a) 2‘deoxyadenosine (dA); (b) 3’-deoxyadenosine (d3A);(c) 2‘-amino-2’-deoxy adenosine; (d) 3’-amino-3’-deoxy-adenosine; (e) 2’-deoxycytidine (dC); (0 2’-amino-2’-deoxycytidine. All have been incorporated into yeast tRNAPhe.
TABLE I SUBSTRATE PROPERTIES OF ANALOGS OF ADENOSINEAND CYTIDINE FOR ATP(CTP):tRNA NUCLEOTIDYLTRANSFERASE Incorporation into position" S'-Triphosphate of
Source of enzyme
2-Chloroadenosine ZBromoadenosine 1:N6-Ethenoadenosine 1:N6-Ethenoadenosine 1-N-Oxoadenosine EBromoadenosine 8-Iodoadenosine EThioadenosine EAzidoadenosine Formycin Tubercidin Toyocam ycin Sangivamycin 7-Deazanebularin Guanosine Inosine 7-(~-~RibofuranosyI)-Z-~inopuriiie 7-(P-~-Ribofuranosyl)-2,6-diaminoPurine 2'-Deoxyadenosine
Yeast Yeast Yeast Yeast Yeast Yeast Yeast Yeast Yeast Yeast, rat liver Rat liver Rat liver Rat liver Rat liver Rabbit liver, yeast Rabbit liver, yeast Rat liver Rat liver Yeast, E. coli
2'-Deoxyadenosine
E. coli, rabbit muscle, rabbit liver
n-2
n-1
n
Reference
-
112 112 145 112,126 112 112 112 112 112 59,61,62 129 129 129 130 128,112 128,112 129 129 112,118,120, 134,136,137 128,132-135
-
+ -
-
+ + + + + -
+
-
9
Z
U
3'-Deoxyadenosine 3'-Deoxyadenosine
2'-Amino-2'-deoxyadenosine 3'-Amino-3'-deoxyadenosine 7-(p-~-Arabinofuranosyl)adenii1e 7-(p-~-Xylofuranosyl)adenine 2'-O-Methyladenosine 2'-O-Methyladenosine 3'-O-Methyladenosine 3'-O-Methyladenosine Adenosineoxi-& Adenosine 5'-(a-thio)triphosphate (A-form)b Adenosine S'-(a-thio)triphosphate ( B-form)b Adenosine-S'-(a, y-dithio)triphosphate 5-Bromocytidine 5-Iodoc ytidine 5-Formyluridine Uridine
4-Thiouridine ZThiocytidine 2'-Deox ycytidine Z'-Amino-2'-deoxycytidine
2'-Azido-2'-deoxycytidine 2'-Chloro-2'-deoxycytidine
E . coli Yeast E. coli, yeast E. coli, yeast Yeast Yeast Yeast E. coli Yeast E. coli Yeast Yeast Yeast Yeast Rabbit liver Yeast Yeast E. coli, yeast, rat liver, rabbit liver Yeast Yeast Yeast Yeast Yeast Yeast
+ + + +
+
+
+
-
+ + + + +
n is the number of the 3'-terminal nucleotide in the tRNA sequence; e.g., for yeast tRNAPhe,n = 76. Adenosine-5'-(a-thio)triphosphateexists in two diastereoisomeric forms, A and B (143).
-
+ -
+ -
-
134,137 118,120,134 138,139 125,138,139 112 112 112,134 134,141 112,134 134,141 142 50,143 143 112 149 116,150 112 112,128,135 148 55 112 151 112 112
20
MATHIAS SPRINZL AND FRIEDRICH CRAMER
phosphate (dAMP or dA) was not incorporated into tRNA by enzymes from E . coli (132-134), rabbit liver (128),and rabbit muscle (135),but incorporation of it into yeast tRNA-N-C-C by the yeast enzyme was later successfully accomplished (120, 134, 136, 137). Similar positive results were obtained using the heterologous system of tRNA-N-C-C from E . coli and the transferase from yeast (118,136). Lack of incorporation of dAMP by the E. coli enzyme (132-134) is probably due to the particular enzyme preparation, since it has recently been demonstrated that highly purified ATP(CTP) : tRNA nucleotidyltransferase from E. coZi incorporates dAMP into yeast tRNA-N-C-C with the same efficiency as with the yeast enzyme (112). Similarly, 3‘deoxyadenosine 5’-triphosphate (3‘-dATP) is a substrate for both the yeast (118,120, 134, 136,137) and E . cali (112) transferases, although the opposite was reported originally for the E . coZi enzyme (134). In a further modification of the ATP cis-diol function, replacement ‘of either hydroxyl by an amino group is possible without interfering with incorporation by ATP(CTP) :tRNA nucleotidyltransferase. The 5‘-phosphate of 2’-amino-2’-deoxyadenosinehas been incorporated into the terminal position of tRNA-N-C-C by the E . coli (138) and b y the yeast enzyme (139); so has the 3‘ isomer (3’-amino-3’deoxyadenosine 5’-triphosphate) (125, 138-140). Replacement of the 2’- or 3’-hydroxyl b y a methoxy group causes a loss of substrate properties in the case of the yeast enzyme (112),whereas the incorporation of 2‘-O-methyladenosine or 3‘-O-methyladenosine into tRNA-N-C-C by the E . coli enzyme has been reported (141).ATP, after periodate oxidation and borohydride reduction (ATP,xi-red)is not a substrate for the yeast transferase (142). Finally, modification of the phosphate groups of ATP by replacement of the a-0x0 group of the triphosphate side-chain by a sulfur atom has been investigated. Adenosine 5‘-(a-thio)triphosphate is a substrate for yeast ATP(CTP) : tRNA nucleotidyltransferase, and a tRNA having the 3’-terminal adenosine bound to the penultimate cytidylic residue by a phosphorothioate linkage has been prepared (50).The transferase accepts only one of the two diastereoisomeric forms of adenosine 5’(a-thio)triphosphate (143).The absolute configuration of this isomer is the same as that of the one that is a substrate for DNA-dependent RNA polymerase (144).Adenosine 5’-(a,y)dithiotriphosphateand adenosine 5’-(y-thio)triphosphateare not substrates for the yeast enzyme (112). Analogs of cytidine can be incorporated into tRNA-N or tRNA-N-C. In all cases investigated, no difference was observed in the ability of analogs to replace cytidylic acid in one of the two penultimate positions. Incorporation of uridine into both positions has been investi-
THE
-C-C-A END OF tRNA
21
gated extensively, and tRNAs terminating in uridine have been prepared using E. coli (121, 146), rabbit liver (128), rat liver (147), and rabbit muscle (135) ATP(CTP) :tRNA nucleotidyltransferase. These tRNAs could not be extended with either an adenylic residue or b y cytidylic and adenylic residues, and thus remained inactive in protein synthesis. Similarly, 4-thiouridylic acid was incorporated into the 3’ end of tRNAPhefrom yeast giving tRNAPhe-A-C-s4U,but attempts to prepare tRNAPhe-A-C-s4U-Awere unsuccessful (148). Cytidylic derivatives bearing a substituent on position 5 of the heterocyclic ring are good substrates for ATP(CTP) : tRNA nucleotidyltransferase. The 5’-phosphates of 5-bromocytidine (149) and 5-iodocytidine (116, 150) have been incorporated into tRNA. Sprinzl et al. (116) reported the incorporation of the latter into both terminal cytidine positions of yeast tRNAPhe,whereas Pasek et al. (150) prepared a tRNAMe‘from yeast in which only the second cytidine position was so occupied. Similar phenomena were observed during the preparation of yeast tRNAphe containing 2-thiocytidine (55); the second s2CMP residue was incorporated more slowly than the first. It was therefore possible to prepare a mixture of tRNAPhe-A-s2Cand tRNAPhe-A-s2C-s2C and separate the two chromatographically (38).All tRNAs containing a cytidylic residue modified in the heterocyclic ring could accept the missing AMP, or CMP and AMP. In this way, biologically active tRNAPhe-A-i5C-i5C-A,tRNAPhe-A-C-i5C-A, tRNAPhe-As’C-C-A, tRNAPhe-A-s2C-s2C-A, and tRNAPhe-A-C-s2C-A became available. 5-Formyluridine 5’-phosphate is not incorporated into tRNA, but is an inhibitor of the yeast enzyme (212). CTP can also b e modified on the ribose moiety without losing its substrate properties for ATP(CTP) :tRNA nucleotidyltransferase. Thus, 2’-amino-2’-deoxycytidylic acid (151) and 2’-deoxycytidylic acid (112) have been incorporated into tRNAPhefrom yeast. tRNAPheA-C-C(Z’NH,) obtained in this way could be further converted to tRNAPhe-A-C-C(2’NH2)-A whereas tRNAPhe-A-C-dCwas a competitive and 2’inhibitor for the enzyme (112). 2‘-Azido-2’-deoxycytidine chloro-2’-deoxycytidine 5’-triphosphates are not substrates for the yeast enzyme (112). In general, the K , values for modified nucleoside 5’-triphosphates are significantly higher than those for the usual substrates ATP and CTP and the rates of their incorporation are usually slower (Table 11). As a consequence, a large excess of modified substrates over tRNA must be used to achieve complete modification. Highly purified enzyme, free of any nucleolytic activity, is also a necessary prerequisite for a successful preparation. Failure to consider this in the earlier
22
MATHIAS SPRINZL AND FRIEDRICH CRAMER
TABLE I1 VALUESOF NATURALAND MODIFIEDNUCLEOTIDES FOR THE ATP(CTP) : TRNA NUCLEOTIDYLTRANSFERASE FROM BAKER’S YEAST (1 12)
K,
AND Vm,
Nucleotide 5‘-Triphosphateof: Adenosine 2’-Deoxyadenosine 3‘-Amino-3’-deoxyadenosine Adenosine 5’-(a-thio)triphosphate A form B form 5’-Triphosphateof: Cytidine 2’-Deoxycytidine Wridine 5-Formyluridine
Km
V max
(mM)
(rel.)
0.6 1.8 1.9
Kf
1.8 = 1.2
0.2 1.7 8.8 K = 0.46
100 11 14
90
-
74 15 6
-
investigations may be the reason for the failures to incorporate certain modified nucleotides into tRNA. In some cases, for example in the synthesis of tRNA-N-dC or tRNA-N-C-dC, the course of incorporation of a modified nucleotide can be complicated by the fact that the product of reaction is an inhibitor of the transferase. C. Incorporation of Modified Nucleotides with Polynucleotide Phosphorylase
A method for a stepwise incorporation of nucleotides into the 3‘ end of a polynucleotide chain using 3’-protected nucleoside 5’diphosphates and polynucleotide phosphorylase was developed by Gilham et al. (113-115), and was applied to the preparation of tRNA-N-C-C-Am (134, 152), tRNA-N-C-C-dA and tRNA-N-C-C-d3A (134,141). For the starting material, unfractionated tRNA-N-C-C from E . coli and the corresponding nucleoside 5’-diphosphates were used. Free 3’-hydroxyl groups were blocked by 1-methoxyethyl groups in order to prevent extensive polymerization (113,114). The experiments were carried out with polynucleotide phosphorylase from Micrococcus luteus. Incorporation of 2‘-0-methyl-3’-0-( 1-methoxyethy1)adenosine 5’-phosphate could be achieved by this method, but 3’-0-methyladenosine 5’-phosphate was not incorporated (152). Preparations of tRNAs terminating with 2’- or 3’-deoxyadenosine residues have been accomplished; however, the yields were low (only about 30%) (134).
THE
-C-C-A END OF tRNA
23
The incorporation of modified nucleotides into the 3’ end of tRNA b y polynucleotide phosphorylase should be improved. At present, this method is appropriate only where the nucleotides to b e incorporated are not substrates for ATP(CTP) : tRNA nucleotidyltransferase, as, for 5‘example, in the case of 3’-O-~-phenylalanyl-2’-deoxyadenosine triphosphate. Preparation of a tRNA terminating with this aminoacylnucleotide was reported by Hecht et al. (134).Although the yield of incorporation was again very low, the tRNA-N-C-C-dA(3‘Phe) could be separated from the starting material, tRNA-N-C-C, by RPC-5 or DEAE-cellulose chromatography (153). Using a similar approach, McCutchan et aE. (254) recently reported the preparation of tRNAs 1:N6-ethenoadenosine, 8-bromoadenosine, terminating with guanosine, and 8-azidoadenosine, Some of these, which cannot be prepared with ATP(CTP) : tRNA nucleotidyltransferase, can b e utilized for spectrofluorometric studies or for affinity-labeling experiments.
V. Aminoacylation of tRNA A. Substrate Properties of Modified tRNAs Enzymic esterification of the hydroxyl groups of the 3’-terminal adenosine of tRNA may be accomplished by an aminoacyl-tRNA synthetase (EC 6.1.1.1-6.1.1.22)specific for the appropriate tRNA and amino acid. Several recent reviews summarize the literature concerning the mechanisms of aminoacylation, interaction of tRNA and aminoacyl-tRNA synthetase, and isolation and characterization of the synthetases (155-257). Hence, the following discussion concentrates on the role of the -N-C-C-A end of tRNA in the interaction of enzyme and tRNA, and especially on the site where covalent bonds are made and broken during aminoaoylation, namely, on the cis-diol group of the terminal adenosine. A complete -N-C-C-A end is an absolute requirement for aminoacylation by a synthetase; the shortened species (tRNA-N-C-C, tRNA-NC, and tRNA-N) cannot accept amino acids. These tRNA species have been used extensively as inhibitors of aminoacyl-tRNA synthetases in investigations of the mechanism of aminoacylation. It is interesting to note that the mode of inhibition b y tRNA-N-C-C is different for particular aminoacyl-tRNA synthetases (158). Whereas, for example, tRNATYr-A-C-Cfrom yeast is a competitive inhibitor of tyrosyl-tRNA synthetase (159),tRNAphe-A-C-Cfrom yeast is a weak noncompetitive inhibitor of the phenylalanyl-tRNA synthetase (131). The different be-
24
MATHIAS SPRINZL AND FRIEDRICH CRAMER
havior of various synthetases toward inhibition by a specific tRNAN-C-C suggests that the mode of the synthetase interaction may differ between cognate pairs. Previous reports on the aminoacylation of ratliver tRNA-N-C-A, which lacks a penultimate cytidylic residue (160), could not be verified in other systems. E . coli tRNAPhe-A-C-Aand tRNAVa'-A-C-Acould not be aminoacylated (121), and a rat liver tRNA-N-C-A was also inactive in this reaction (41).Similar results were reported for tRNAPhe-A-C-Afrom yeast (161). The 3' terminus of tRNA can be elongated by one extra CMP residue. Incubation of tRNA-N-C-C and CTP with ATP(CTP) :tRNA nucleotidyltransferase in the absence of ATP leads to the formation of tRNA-N-C-C-C, which can then be converted to tRNA-N-C-C-C-A b y incorporation of ATP in the absence of CTP (161,162).Such modified unfractionated tRNA from rabbit liver reportedly accepts 12 amino acids (162). However, the extent of aminoacylation was lower than with unmodified tRNA and some amino acids were not incorporated. Rether et al. (161) recently observed that a preparation of yeast tRNAPhethat contained an -A-C-C-C-A terminus was aminoacylated by the corresponding aminocyl-tRNA synthetase to an extent of 80% compared with the native tRNA; the kinetic data of the reaction were not reported. A reinvestigation of the substrate properties of yeast tRNAPhe-A-C-C-C-Atoward its synthetase indicates that this tRNA is phenylalanylated very slowly, if at all (163).This is in agreement with an earlier report (121 ) that the aminoacylation activity of tRNAPhe-AC-C-C from E . coli is not restored by incorporation of a terminal adenylic acid. Clearly, this problem requires further investigation. Some tRNA species containing nucleosides other than adenosine in the normal terminal position are either not aminoacylated, or the kinetics of aminoacylation are significantly altered. tRNA-N-C-C-C, where the native terminal AMP position is occupied by a CMP residue, cannot be aminoacylated (121, I S ] ) , but there is no further information on the'interaction of such tRNAs with the synthetases. In view of the finding that the presence of the terminal adenosine yeast tRNAPhe-A-C-C-Atriggers the specific interaction with the synthetase (131), it would be of interest to investigate the properties of tRNAPhe-A-C-C-Cin this system. The acceptor activities of tRNAs from rat liver containing either toyocamycin, tubercidin, sangivamycin, or formycid at the 3' end has been investigated (59,129) using a mixture of synthetases. The incorporation of phenylalanine into tRNAs containing tubercidin was lower compared to the normal tRNA-N-C-C-A (129).Those tRNAs with terminal sangivamycin, toyocamycin (1291, and formycin (59) were
THE
-C-C-A END OF tRNA
25
aminoacylated, but at rates below that of the unmodified tRNA. The rate and extent of aminoacylation of tRNAPhe-A-C-C-Ffrom yeast was later determined (61) using purified phenylalanyl-tRNA synthetase. The K , values of the tRNA are not affected by this modification, but the maximal velocity of aminoacylation is reduced to one-fiftieth of that of native tRNAPhe. The terminal ribose of this tRNA can b e cleaved by periodate oxidation and the hydroxyl groups restored by reduction with borohydride. The tRNAPhe-A-C-C-Foxi--red obtained by this procedure is not aminoacylated and is a competitive inhibitor of the synthetase, having a K i of 0.45p M . Similar treatment of the native tRNA leads to tRNAPhe-A-C-C-AOxi-,,,which is aminoacylated by the yeast synthetase (33)with the same K , but about half of the maximal velocity of the native species. Comparison of the activities of the latter two modified tRNAs demonstrates that a small structural change at the 3'-terminal end can result in large differences in the substrate properties (Table 111).It was suggested that stacking interactions, which may be different for foimycin and adenosine residues (601,are important in determining the necessary conformation for enzymatic attachment of the amino acid (61). In the same way as the inhibitory activity of tRNA-N-C-C varies among synthetases, the substrate properties of tRNA-N-C-C-AoXi-re, also differ for particular tRNAs (Table IV). Whereas yeast tRNAPheA-C-C-Aox,-,edis fully aminoacylated (33),the extent of aminoacylation by the E. coli enzyme is of purified E. coli tRNAPhe-A-C-C-Aox,-red much lower than that of normal E. coli tRNAPhe(164).Phenylalanine, tyrosine, and methionine incorporation into mixed oxidized and reduced E. coli tRNAs is reduced to about half compared to native tRNA-N-C-C-A, but no aminoacylation with Ile, Val, Asp, Trp, Arg, His, and Ser took place ( 4 1 ) (Table IV). Despite the observation that periodate oxidation may cause some deactivation of tRNA, probably due to a modification of other parts of the molecule, not only of the terminal ribose (118), these results ( 4 1 ) indicate that the same modification on the invariable 3' end of tRNA results in very different effects on the aminoacylation properties. The activity of purified E. coli tRNAVa1-A-C-C-AOxi-,,din the aminoacylation reaction was also investigated using purified valyltRNA synthetase (165).To achieve complete aminoacylation, a 4000fold excess of enzyme was required in the aminoacylation assay compared to the normal tRNAVal.The K , values of the normal and the modified substrates are identical, whereas the maximal velocity of aminoacylation of tRNAz?/-,ed is reduced by a factor of lo4.It is therefore possible that the other tRNAs listed in Table IV that did not accept
TABLE I11 AMINOACYLATIONPROPERTIES OF YEAST T R N A ~ MODIFIED ~~s AT THE 3' TERMINUS
tRNAPhe- "
Maximal aminoacylation (pmol/A,, unit tRNA)
A-C-C-A A-s2C-C-A A-C-s'C- A A-C-acms*C-A' A-G( SL)s2GA' A-C-C-A(2'N HZ) A-i5C-iSC-A A-C-PC-A A-C-C-dA A-C-C-d3A A-C-C-Aox, A-C-C-AOxi-,, A-C-C-F A-C-C-Fox, A-C-C-F,, A-C-CSA A-GCp A-C-C
1500 1430 1390 1420 20 1380 970 1240 20 1420 20 1500 1530 20 20 1500 20 20
.Vmm (rel.)*
1.0 2.80 2.80 2.02 Weak inhibitor Not determined 1.1 1.0 KI = 2.16 competitive inhibitor 2.86 Weak mixed-type inhibitor 1.0 1.05 Weak mixed-type inhibitor K , = 0.25 competitive inhibitor 1.0 Weak mixed-type inhibitor Weak mixed-type inhibitor
References
100 64 45 40
61,131,169 38 55
-
151
78 50
116 116 169 169 130 33
64
-
45 2
54 168
61 131 61 50 131 131
z
a-
4
2 %
z3 r 9
2 U
2!2 U
" For formulas, see Figs. 5 and 6. acm = aminocarbonylmethyl (-CH,CONH,); SL = spin-labeled. V,,, is given as a percentage of the V,,, of native yeast tRNAPhe-A-C-C-A. Thiocytidine substituted with iodoacetamide and a spin label, respectively (compare Fig. 5).
THE
27
-C-C-A END OF tRNA TABLE IV OF Escherichiu coli T R N A - N - C - C - A . ~ ~(41,118) -~~ AMINOACYLATION ~~~~
~~~~
~
Aminoacylation (% activity remaining)" Amino acid Phenylalanine Tyrosine Methionine Isoleucine Valine Aspartic acid Tryptophan Arginine Histidine Serine
Controlb
81
45 86 78
tRNA-N-C-C-A,i-,,j
55 44 44
65
6 1
11 77
0
84 45 32
1 0 0 0
" Aminoacylation is given as a percentage of activity compared to native tRNAN-C-C-A. Control tRNA was treated with sodium periodate, the terminal nucleotide was removed (41), and the tRNA was restored by incorporation of AMP (118). Reduction with borohydride has no significant influence on the extent of aminoacylation (41).
an amino acid using crude synthetases would do so under appropriate conditions, although with a drastically reduced velocity. There seem to be differences between the substrate properties of yeast and E . coli tRNA-N-C-C-A,,i-,,,+ As shown with purified yeast systems, oxidized and reduced tRNAPheand tRNASer can be aminoacylated, whereas tRNATYr,tRNAThr,tRNAVal,and tRNA"" do not accept amino acids (51), or do so only at much lower rates. Comparison of these results with data in Table IV indicates that the activities of seryl- and tyrosyl-tRNA synthetases toward oxidized and reduced tRNAs are different in the two systems. This indicates that the interaction of a particular synthetase with the 3' end of tRNA is different not only for synthetases of different specificity from the same organism, but in some cases also for synthetases of the same specificity from different sources. There is loss of acceptor activity of yeast tRNApheafter oxidation of the terminal adenosine with monoperphthalic acid (27).Replacement of the terminal adenosine by inosine, accomplished b y enzymic deamination of tRNA, does not affect aminoacylation (166).A tRNA containing 1:N6-ethenoadenosine at the 3' end has been prepared and enzymically aminoacylated (154, 168). Modifications of the penultimate cytidine residues have little effect on the ability of tRNAPhe from yeast to accept its amino acid.
28
MATHIAS SPRINZL AND FRIEDRICH CRAMER
tRNAPhe-A-i5C-i5C-A, tRNAPhe-A-C-i5C-A (116),tRNAPhe-A-s2C-C-A, and (55) are aminoacylated to the same extent as native tRNAPhe-A-C-s2C-A tRNAPhefrom yeast. Similarly, rabbit liver tRNA in which the penultimate cytidine residues are replaced by 5-bromocytidine can be aminoacylated (149).Even the attachment of an acetamide residue to the 2-thio function of the s2C nucleotide of tRNAPhe-A-C4C-A (Fig. 5 ) does not interfere with tRNA : enzyme recognition (54) (Table 111).If, however, the alkyl substituent on this position carries a bulky spinlabel (Fig. 5), enzymic aminoacylation is not possible (167). Conversion of both CMP residues of the -N-C-C-A end to UMP by NaHSO, does not deactivate tRNATYrfrom E . coli (24) or E . coli tRNAAm(19). On the other hand, tRNAmet-A-C-U-A from E . coli prepared b y the same method was not chargeable (26).Reaction of the penultimate cytidine residue with chloroacetaldehyde leading to a -A-C-rC-A terminus (127) does not cause a deactivation of tRNAmetfrom E . coli toward aminoacylation (168). Despite the fact that in some cases it is possible to modify the nucleobases of the 3' terminus of tRNA without destroying the aminoacylation activity, the kinetics of aminoacylation are usually altered. By examination of the K , and V,,, values for aminoacylation of some yeast tRNAPhespecies altered at the 3' end (Table HI), it became apparent that, although the K , values of modified tRNAs remain unchanged (e.g., 169), the velocity of aminoacylation is lowered upon modification. This effect seems to be more pronounced where the modification is closer to the 3'-terminal nucleoside. The most significant changes in the rates appear in those tRNAs altered at the terminal adenosine residue (Table 111).
B. Role of the 3'-Terminal Adenosine during the Interaction of tRNA and Aminoacyl-tRNA Synthetase
Since the -N-C-C-A terminus is common to all tRNAs, it cannot be responsible for the recognition of a tRNA by a specific aminocyl-tRNA synthetase. I n agreement with this assumption, it is possible to modify extensively the cytidine residues of the -N-C-C-A end without altering the substrate properties of the particular tRNA (Table 111). I n contrast, modifications of the terminal adenosine have a large effect on the interaction of synthetase and tRNA, although the observed kinetics of aminoacylation or the inhibitory effects of tRNAs modified at the 3'terminal adenosine are not uniform for the different tRNAs. Cramer et al. have investigated in detail the effect of alterations of the residue of yeast tRNAPheon the aminoacylation reaction. The most surprising result, evident from Table 111, is that a simple modification such as removal of the terminal adenosine or AMP, lead-
THE
-C-C-A END
OF
tRNA
29
ing to tRNAPhe-A-C-Cpor tRNAPhe-A-C-C,causes a conversion of a substrate to a weak mixed-type inhibitor. The affinity of these modified tRNAs Phe for the phenylalanyl-tRNA synthetase is considerably less than that of tRNAPhe-A-C-C-A.For example, to achieve a 50% inhibition of aminoacylation, as much as a 10-fold excess of tRNAPhe-A-C-Cover the substrate tRNAPhe-A-C-C-Awas needed (131). The influence of the absence of the terminal adenosine of tRNAPheon the tRNA-synthetase interaction could also be shown by thermodynamic and kinetic measurements (170). It is very unlikely that the intrinsic binding increment of the terminal adenosine accounts for the observed differences between tRNAPhe-A-C-C-A and tRNAPhe-A-C-C. Therefore, the 3’-terminal adenosine of tRNAphe cannot be solely a passive acceptor of phenylalanine, but must play a more active role in the interaction o f tRNAPheand phenylalanyl-tRNA synthetase. It has been suggested (131),therefore, that the 3’-adenosine triggers a conformational change of this enzyme during the sequence of the recognition processes, which leads to a tighter binding of tRNA to the synthetase. In this hypothesis, the 3’-terminal adenosine would play a role analogous to an enzyme effector by the action of which the catalytically incompetent tRNA synthetase complex is converted to a catalytically competent one. Similar effects of the terminal adenosine on the efficiency of binding of tRNAPheare observed in other reactions. For example, tRNAPhe-A-C-Cdoes not inhibit the action of tRNAPhe-A-C-C-Ain the ribosome-dependent synthesis of guanosine tetraphosphate (171). Furthermore, a correct codon-anticodon interaction during the binding of tRNA to ribosomes is dependent on an accurate fit between the -N-C-C-A terminus of tRNA and the ribosome. Thus poly(U) stimulates the binding of tRNAPhemodified at the 3’-terminal adenosine to a much lesser extent than does the binding of native tRNAPhe(172). Since there are large differences in the properties of tRNAs of different specificities and of different origins after the same modification of the 3’-terminal adenosine (Tables 111-V), the results obtained from yeast tRNAPhe cannot be directly applied to other tRNAsynthetase interactions. Although there may be such a general function for the terminal adenosine during aminoacylation, this must b e proved independently for each particular pair of tRNA and aminoacyl-tRNA synthetase. C. Site of Aminoacylation of tRNA In principle, either of the two hydroxyl groups of the 3’-terminal adenosine can be esterified during enzymic aminoacylation. Although the aminoacyl-tRNA synthetase might be expected to b e specific for
30
MATHIAS SPRINZL AND FRIEDRICH CRAMER
TABLE V AMINOACYLATION OF TRNAs
Substrate
MODIFIEDON THE TERMINAL mBOSE RESIDUE
Source
Maximal aminoacylation (pmol/Azao unit)
tRNAPhe-A-C-C-A tRNAPhe-A-C-C-dA tRNAP"-A-C-C-d3A tRNAPh9-A-C-C-A(2'NHz) tRNAPhe-A-C-C-A( 3'NH*)
Yeast Yeast Yeast Yeast Yeast
1460 20 1420 1520 1580
tRNATYr-A-C-C-A tRNATYr-A-C-C-dA tRNATYr-A-C-C-d3A tRNATYr-A-C-C-A(2'NHz) tRNATYr-A-C-C-A(3'NHJ
Yeast Yeast Yeast Yeast Yeast
1630 1460 1420 40 1480
tRNAiIe-A-C-C-A tRNAile-A-C-C-dA tRNAi'e-A-C-C-d3A
Yeast Yeast Yeast
1600 40 1600
tRNAT"-A-C-C-A tRN AThr-A-C-C-dA tRNAThr-A-C-C-d'A tRNAThr-A-C-C-A(2'NHZ) tRNAThr-A-C-C-A( 3'NH2)
Yeast Yeast Yeast Yeast Yeast
tRNASer-G-C-C-A tRNASer-G-C-C-dA tRNASer-G-C-C-d3A tRNASer-G-C-C-A(2'NHz) tRNASer-G-C-C-A(3'NHz)
K, (PM) 2.80 Ki
= 2.16
V,, (rel.)n 100
-
2.86 3.80 2.20
64 40 85
1.7 4.5 1.7 NCI 2.22
100 18 75
4.3 Ki
= 2.3
-
63
100
-
4.04
21
1420 1120 20 1400 20
0.83 5.5 NCI 4.5 NCI
100 76
Yeast Yeast Yeast Yeast Yeast
1530 1490 20 1020 20
0.91 3.1 NCI ND ND
100 100
tRNAVal-A-C-C-A tRNAVa'-A-C-C-dA tRNAVal-A-C-C-d3A
Yeast Yeast Yeast
1600 20 1600
6.6 NCI 3.6
100
tRN APhe-A-C-C-A tRNAPhe-A-C-C-dA tRNAPhe-A-C-C-d3A tRNAPhe-A-C-C-A(2'NH,) tRNAPhe-A-C-C-A(3'NH,)
E. coli E. coli E. coli E. coli E. coli
1500 80 1450 60 1450
0.5 NCI 0.5 NCI 0.5
100
tRN A A'a-A-C-C-A tRNAAla-A-C-C-A tRNAAia-A-C-C-dA tRNAAia-A-C-C-d3A
E. coli E.colib E. coli
-
E . colib
-
-
1.66 1.64 1.67 1.68
-
2.4
-
-
-
9.2
References 169 169 169 38 140
173 173 173 159 159 51, 173 51,173 51, 173 51 51 51 51 51 51,173 51,173 51,173 38 38 51,173 51,173 51,173
125
38 38 38 38 38
100 100 83 22
176 176 176 176
0.125
THE
31
-C-C-A END OF tRNA TABLE V (Continued) ~
Substrate tRNALY'A-C-C-A tRNALYs-A-C-C-A tRNALY"A-C-C-dA tRNALYs-A-C-C-dA tRNALYs-A-C-C-dSA tRNALYs-A-C-C-d3A
Source
E. coli E . coli
E.coli E . coli
E.coli E . coli
Maximal aminoacylation (prnol/A,,, unit)
1450 1550b 1420 20b 30 1510b
~~
~
K, (PM)
6.4 5.5 8.3 NCI
K i = 13.2 2.1
~
~
V,, (re1.y
100 100 6.8
-
16.8
~
_
References
38 38 38 38 38 38
" ND, not determined; NCI, no competitive inhibition.
* V,,,
is given as a percentage of V,,, of the particular native tRNA. In these experiments, yeast phenylalanyl-tRNA synthetase was used for aminoacylation of modified tRNAsLYS and tRNAs*'" from Escherichia coli.
one of the hydroxyl groups, the site of the primary attachment of the amino acid with respect to the 2'- or 3'-hydroxyl groups cannot be determined b y a chemical analysis of the products of the reaction because of the rapid migration of the amino acid residue between the cis-vicinal hydroxyls (66).It is not possible to trap the primary product of aminoacylation b y chemical methods, and investigations using this approach (68, 73, 7 4 ) have not led to unambiguous conclusions. A more recent approach to this problem involved the use of modified tRNAs having 3'-terminal nucleosides in which the cis-diol function is missing (Fig. 7), thus precluding isomerization of the amino acid residue. Thus if, for example, the 2'-hydroxyl group is missing (Fig. 7) and the tRNA is still aminoacylated, it is obvious that the aminoacylation must have occurred on the 3'-hydroxyl. If the K , and V,,, of aminoacylation of such a deoxy species are reasonably close to those obtained for the corresponding unmodified tRNA, it can be considered that the initial position of aminoacylation of the unmodified tRNA is the one aminoacylated in the modified tRNA. This approach was first applied for determination of the site of aminoacylation of tRNAPhefrom yeast (169).tRNAP"-A-C-C-d3A (d3A3'-deoxyadenosine) is a substrate for yeast phenylalanyl-tRNA synthetase, and phenylalanine is attached to the 2'-hydroxyl group of this tRNAPhe.The K , value of tRNAphe-A-C-C-d3Ais identical to the native tRNAPhe,whereas the rate of aminoacylation is reduced to about twothirds of that observed for native tRNA. tRNAPhe-A-C-C-dA(dA-2'-
_
_
32
MATHIAS SPRINZL AND FRIEDRICH CRAMER
I
OH
yo YHNH,
i0
CHNH2
R
y z ? F0 fHNH2
?o+y 2
O H
H O
R
R a
b
OH YH
d
C
YH OH
$0
$0
FHNH2
FHNH2
R
R e
FIG. 7. Structure of the 3' end of nonisomerizabie aminoacyl-tRNAs obtained by enzymic aminoacylation of (a) tRNA-N-C-C-d3A,(b) tRNA-N-C-C-dA,(c) tRNA-N-C-CAoxl-red,(d) tRNA-N-C-C-A(2'NH2),(e) tRNA-N-C-C-A(3'NH2).
deoxyadenosine), where the 2'-OH is missing, is an inhibitor of the synthetase and is not aminoacylated (Table V). From these results, it was concluded that the site of aminoacylation of tRNAphefrom yeast is the 2'-OH group of the terminal adenosine. Later, the same approach was extended to the determination of the site of aminoacylation of other yeast tRNAs (1 73), in which aminoacylation of purified "deoxy" tRNAs was determined with purified enzymes. With tRNAValand tRNA"", only those species bearing a 2'-OH group were aminoacylated; the absence of a 3'-OH did not influence the reaction. On the other hand, tRNASer-G-C-C-dAand tRNAThr-A-C-CdA were chargeable while the corresponding d3A species were not. Therefore, tRNAThrand tRNASerfrom yeast are 3' acceptors (173).In the case of tRNATY',both the 2'-deoxy and 3'-deoxy species were enzymically aminoacylated but the species lacking the 2'-OH group reacted at significantly lower rates. The tyrosyl-tRNA synthetase from yeast, although less specific, is therefore probably also a 2'-hydroxyl esterifying enzyme. The kinetic parameters of aminoacylation (K,, V,,J are in some cases significantly altered when the hydroxyl group that is not an esterification site is removed, or is replaced by an amino group. For is seven times higher than that example, the K , of tRNAThr-A-C-C-dA
THE
-C-C-A
END OF
tRNA
33
of native tRNAThr-A-C-C-A, and the rate of aminoacylation of tRNAThr-A-C-C-A(2‘NH2) is decreased by a factor of 40.These effects are not uniform for all synthetases (Table V). It is likely that such differences are due to different conformations of the sugar residue in the deoxy and amino analogs of ribose, which are not tolerated by synthetases to the same extent. Despite these facts, the changes in the substrate properties of tRNAs from substrates to nonsubstrates or inhibitors after modification of one hydroxyl group can be considered as evidence in determining the site of esterification. A similar method can be used for the elucidation of the site of aminoacylation of a particular tRNA in a mixture of tRNAs from a given organism. Sprinzl and Cramer (118) incorporated 2‘deoxyadenosine and 3’-deoxyadenosine, respectively, into mixed E . coli tRNAs and compared the extent of aminoacylation of the resulting tRNA-N-C-C-dA and tRNA-N-C-C-d3A species with that of the normal tRNA-N-C-C-A, using a mixture of E. coli enzymes and one particular radioactive amino acid. I n the majority of cases, a complete loss of activity was observed after removal of the accepting O H group whereas the isomeric “deoxy” tRNA still accepted the given amino acid. From the results of these experiments, the E . coli tRNAs can be divided into three groups: (a) if only tRNA-N-C-C-dA is aminoacylated by a particular amino acid, the synthetase is 3’-specific; (b) if only tRNA-N-C-C-d3A is esterified, the synthetase is 2’-specific; (c) if both “deoxy” tRNAs are aminoacylated (as is the case with tyrosine and cysteine), there is relatively little specificity of the corresponding synthetase (118). These observations, together with the results obtained b y aminoacylation of “deoxy” tRNAs from yeast using purified enzymes (173),show that the site of aminoacylation is not, as suggested previously (74, 141, 174), uniform for all tRNAs. Neither the fact that the 2‘-OH group of adenosine is more reactive in chemical reactions (1 75) nor the finding that the 3’-aminoacylated tRNA is in excess over the isomeric 2’-derivative after aminoacylation (74) are relevant to the problem of the primary attachment of amino acid. Taking into account the expected similarity in the three-dimensional structure of tRNAs (13, 14) as well as the fact that the aminoacylation takes place on that part of the molecule that is the same for all tRNAs, the site of primary attachment of the amino acid with respect to the 2‘- or 3’-position is probably determined b y the architecture of the active site of the particular synthetase and not by a feature of the tRNA (118). This was also proved by direct experiments (38, 176). Phenylalanyl-tRNA synthetase from yeast is a 2’-esterifying enzyme
34
MATHIAS SPRINZL AND FRIEDRICH CRAMER
since only tRNAPhe-A-C-C-d3Ais aminoacylated. This enzyme can mischarge tRNALyS from E . coli leading to Phe-tRNALys.Although the lysyl-tRNA synthetase, being a 3‘-esterifying enzyme, aminoacylates only the tRNALYs-A-C-C-dA, the yeast phenylalanyl-tRNA synthetase but not the tRNALYsphenylalanylates the E . coli tRNALYs-A-C-C-d3A A-C-C-dA (Table V). A similar change in the site of attachment was observed in the case of tRNAALa from E . coli. Aminoacylation with the cognate synthetase takes place on the 3’-hydroxyl whereas under misacylation conditions by yeast phenylalanyl-tRNA synthetase, the 2’hydroxyl is aminoacylated. Thus the site of attachment of the amino acid must be governed b y the synthetase, not b y the tRNA. This again implies that synthetases are not a uniform class of enzymes, and that their active sites and probably also the mechanisms of their catalytic activity are different. Comparing the 2’ vs. 3’ specificities of yeast and E . coli synthetases (118, 173), it appears that this specificity has been retained in the evolution of prokaryotes to eukaryotes. More recently, Hecht and Chinault (136) investigated the 2‘ vs. 3‘ specificities of yeast and E . coli synthetases using mixed (unfractionated) isomeric “deoxy” tRNAs from yeast and E . coli, respectively. As before (118),the aminoacylation was performed with a partially purified mixture of synthetases from the particular organism. Again, the 2’ vs. 3’-specificity of the synthetases was the same, regardless of the source of the enzyme. A further extension of such studies to calf liver synthetases (177)did not disprove this rule. The only case in which the positional specificity might have been changed during evolution is that of tRNATrP.Here tRNA-N-C-C-d3A was a substrate for the E . coli enzyme (118, 136) whereas tRNA-N-C-C-dA was utilized by yeast and calf liver enzymes (136, 177).The data on the site of aminoacylation of particular tRNAs as determined b y the aminoacylation of “deoxy” tRNAs are summarized in Table VI. The rate of aminoacylation of “deoxy” tRNAs may be significantly reduced for some species. Whereas yeast tRNAPhe-A-C-C-d3A is aminoacylated at almost the same rate as the normal species, the E . coli tRNAPhe-A-C-C-d3Ais aminoacylated at 1/800th the rate of the native E . coli tRNAPhe(Table V). With a mixture of synthetases and tRNAs (118, 177), it might therefore be difficult to observe a slow aminoacylation of some modified tRNAs. This is probably the reason why no aminoacylation was observed of either tRNA-N-C-C-dA or tRNA-N-C-C-d3A with aspartic acid or glutamine in an E . coli system, with glutamic acid in a yeast system, and with glutamine, glutamic acid, methionine, or proline in a calf-liver system (Table VI).
THE
-C-C-A END
OF
tRNA
35
There is also the possibility that in those cases where an aminoacylation of only one "deoxy" tRNA was observed, both can in fact b e aminoacylated, but at different rates. For example, the tRNAArgfrom E . coli, originally identified as a 2'-acceptor (118),was later found to belong to the group where both 2' and 3'-deoxys can b e aminoacylated (177).The same may be true for the tRNATrpfrom E . coli (177),which would explain the only discrepancy between the site of aminoacylation of tRNAs from different sources. It is therefore clear that the data obtained using mixtures of 3'-modified tRNAs and mixtures of synthetases can be taken only as a preliminary indication of the 2' vs. 3' specificity of synthetases and that this must be confirmed, at least in ambiguous cases, with purified components. A different approach to the determination of the site of aminoacylation was used by Fraser and Rich, who incorporated 2'-amino-2'deoxyadenosine (138) (Fig. 6) or 3'-amino-3'-deoxyadenosine(125, 138) (Fig. 6 ) into mixed E . coli tRNAs and determined the extent of aminoacylation of these modified tRNAs. With the assumption that the aminoacyl residue can be transferred enzymically only from aminoacyl-AMP to a hydroxyl group of the 3'-terminal adenosine, these modified tRNAs could be used for the determination of 2' vs. 3' specificity (138).For instance, a 2'-specific synthetase would attach the amino acid only to tRNA-N-C-C-A(3'NH2) via the 2'-OH group. Aminoacylation to such a tRNA would then result indirectly in 3'-Naminoacyl-tRNA (Fig. 7e; X = tRNA-N-C-C-) since the amino acid migrates to the 3'-position, forming a stable amide bond. The opposite should be true for 3'-specific synthetases where aminoacylation of the 3'-OH group of tRNA-N-C-C-A(2'NHz) would result in a stable 2'-Naminoacyl-tRNA (Fig. 7d). As demonstrated in Table V this assumption is valid for yeast tRNAThrand tRNASer,which are 3' acceptors and whose derivatives, tRNA-N-C-C-A(2'NHz), are therefore chargeable, whereas the tRNA-N-C-C-A(3'NHz) derivatives are not. Similarly in the case of tRNATYr,which is preferentially aminoacylated on the 2'-OH group ( 1 73),tRNATYr-A-C-C-A(3'NHJis aminoacylated whereas tRNATYr-A-C-C-A(2'NHz)is not. However, yeast tRNAPhe,which is a 2'-acceptor, clearly demonstrates that a direct aminoacylation of the 2'NHz group is also possible since both tRNAPhe-A-C-C-A(2'NHz) and tRNAPhe-A-C-C-A(3'NH.Jare aminoacylated, exhibiting K,s and Vmaxsvery similar to those of native tRNA (Table V). T h e larger number of "2'- and 3'-esterifying" unspecific synthetases predicted (138) compared to other reports (118, 177) can be explained by this finding. It is interesting to note that E . coli phenylalanyl-tRNA synthetase does not show this type of behavior. As
w
m
INITIAL SITE OF
TABLE VI ESTERIFICATION OF Escherichia COli, YEAST, AND CALF LIVERTRNAs TO THE 2'- OR 3'-HYDROXYL GROUP, AS DETERMINED BY ENZYMIC AMINOACYLATION OF "DEOXY" TRNAs Attachment of amino acid
E. coli
AminoacyltRNA synthetase specificity
Positied
Methodb
Alanine Arginine Asparagine Asparagine Aspartic acid Cysteine Glutamine Glutamic acid Glycine Histidine
3' 2' 2' 2',3' NCC 2',3' NC 2' 3' 3'
Mix Mix Mix Mix Mix Mix Mix Mix Mix Mix
177 177 118 177
118,177 118,177 118,177 177 118,177 118,177
Position"
3' 2' 2',3' 2',3' 2',3' 3' NC 3' 3'
5
Calf liver
v)
Methodb
References
Position"
Methoda
References
Mix Mix Mix
177 177 177
3' 2' 2',3'
Mix Mix Mix
177 177 177
Mix Mix Mix Mix Mix Mix
177 177 177 177 177 177
3' 3' NC NC 3' 3'
Mix Miu Mix Mix Mix Mix
177
-
-
5
3
Yeast References
z
-
-
-
177
;;; 177 177
2z N
p
$ 2 8 P
Isoleucine Leucine Lysine Methionine Phenylalanine Proline Serine Threonine Tryptophan Tyrosine Valine
2' 2' 3' 2' 2' 3'
3' 3'*
2'' 2',3' 2'
Mix Mix Purified Mix Purified Mix Mix Mix Mix Mix Mix
118,177 118,177 38 118,177 38 177 118,177 177 118,177 118,177 118,177
2' 2' 3' 2' 2' 3' 3' 3' 3' 2',3'f 2'
Purified Mix Mix Mix Purified Mix Purified Purified Mix Purified Purified
173 177 177 177
2' 2'
169
2'
177 173 51 177 173 173
NC 3' 3' 3'
3' NC
2',3' 2'
Mix Mix Mix Mix Mix Mix Mix Mix Mix Mix Mix
177 177 177 177 177 177 177 177 177 177 177
NC, uncertain. If the method of determination is given as "Mix," the specificity was determined using a mixture of tRNA-N-C-C-dA or tRNAN-C-C-d3A aminoacylated with a crude synthetase preparation and a particular amino acid. In the cases indicated as "purified," purified tRNA was modified by incorporation of a particular deoxyadenosine and then aminoacylated using purified synthetase. In the cases where both methods were used, reference is made only to the case using the purified system. The site of attachment is the 3'- or both the 2'- and 3'-OH group (177). * T h e originally reported site of aminoacylation, 2' (118),is not valid, as was shown in later investigations (38, 177). The site of attachment of tryptophan to E. coli tRNA might also be 3' as determined with a partially purified system (1 77). 'The preferential site of attachment is 2' (173);see also Table V. (I
4
5 W
5 2
h
38
MATHIAS SPRINZL AND FRIEDRICH CRAMER
could be predicted from experiments with deoxy tRNAs, tRNAPhe-AC-C-A(3’NH2) is aminoacylated whereas tRNAPhe-A-C-C-A(2‘NH2)is not (Table V). Although the determination of the site of aminoacylation using “amino” tRNAs (125, 138) is more complicated than that using the isomeric “deoxy” species for the above mentioned reasons, the nonisomerizable products of aminoacylation of tRNA-N-C-C-A(2‘NH2) or tRNA-N-C-C-A(3‘NH2)(Fig. 7d,e) are very usefuI for the study of 2‘ vs. 3‘ specificity during the other steps of protein biosynthesis (125,
140,172). The site of enzymic acylation of “oxidized-and-reduced” tRNA, tRNA-N-C-C-Aoxi-ped, has been investigated in detail by Ofengand et al. tRNAPhe-A-C-C-A,xr-,dfrom yeast can be enzymically aminoacylated (33), but owing to the absence of the cis-diol group of the terminal adenosine, it is not expected that the aminoacyl residue would since transacylate to the remaining free OH group (Fig. 7c) (163,178), stereochemically highly unfavored intermediates have to be postulated for such isomerization. Investigation of the activity of PhetRNAPhe-A-C-C-Aoxi-red(Fig. 7c, X = tRNAPhe-A-C-C- from yeast) revealed that during the enzymic aminoacylation, phenylalanine is attached to the 2’-OH group of tRNAPhe-A-C-C-Aoxl-,d (173). This was deduced from the finding that such an enzymically aminoacylated tRNA could not act as an acceptor of a peptidyl residue on the ribosomes whereas (2’,3‘)O-phenylalanyladenosineobtained by unspecific chemical phenylalanylation of was active in this reaction. This indirect evidence was confirmed b y NMR determination of the structure of adenosine,,i-,,d-Phe (Fig. 7c) obtained by enzymic phenylalanylation of tRNA-N-C-C-Aoxl-,e~ foIlowed by the removal of the terminal nucleotide (164).In this work it was shown that the phenylalanyl-tRNA synthetases from E . coli, yeast, or rat liver catalyze the attachment of phenylalanine onto the OH group of tRNA-N-C-C-Aox,-,d that is in the vicinity of the glycosyl bond of the terminal adenosine residue, the former 2’-OH group. In connection with the aminoacylation of oxidized and reduced tRNAs, it is interesting to note that each tRNA-N-C-C-AOxi-,d listed in Table IV that can be enzymically aminoacylated is a 2’-acceptor, probably because this OH group of the oxidized-reduced tRNA is held in a position more similar to that of the native tRNA. Owing to its position at the end of the open Smembered chain, the “3’-hydroxyl” of the tRNAoxl-red does not have these conformational restrictions. However, the finding that tRNASer-G-C-C-Aoxi-redfrom yeast, which is a 3’acceptor (173), can be aminoacylated (51), and the fact that 3’-0phenylalanyl-adenosineoxi-red acts as an acceptor of the peptidyl-
THE
-C-C-A
END OF
39
tRNA
residue indicate that even after cleavage of the C2’-C3’ bond the original vicinal hydroxyl groups can adopt a riboselike conformation during enzymic reactions. I n view of the rapid isomerization of the amino acid between the 2’- or 3’-sites, it could be argued that there is little biological significance in the specific aminoacylation of tRNAs to the 2’ or 3‘ terminal hydroxyls, since the speed of protein synthesis is such that the isomer needed for a particular step can be generated b y isomerization (66). However, the finding that the aminoacyl-tRNA synthetases can b e divided into two groups with respect to their specificity for the 2‘- or 3’-sites shows that these enzymes can catalyze at least two types of reaction. This has to be considered in any attempt to describe a general reaction mechanism for the synthetases, which have previously been thought to be a coherent group of enzymes. The biological significance of the 2’ vs. 3’ specificity of the synthetases for the specificity of the aminoacylation reaction, as it appears at present, is discussed in the next section.
VI. Positional Specificity of Arninoacylation and Chemical ”Proof reading” A. Site of Aminoacylation and Mischarging The aminoacylation reaction of tRNA, occurring at the -N-C-C-A end, is the key reaction in distinguishing between the 20 amino acids. After the amino acid is attached to tRNA, there is no other discriminatory step (180). Therefore, this reaction must occur with a high degree of precision. There are several cases known in which the first step of aminoacylation of tRNA, the activation of the amino acid by ATP, is not very specific. In several cases, the discrimination is only in the second step during the transfer of the amino acid from the aminoacyl-AMP to the tRNA. The best-known case in this connection is the (mis)activation of valine b y isoleucyl-tRNA synthetase (181) (Table VII). In the complex between isoleucyl-tRNA synthetase and valyl-AMP, yalylAMP is immediately hydrolyzed on addition of tRNA’Ie-A-C-C-A(182) (Scheme 1). There seems to exist a correction mechanism that prevents misacylation of tRNA1le-A-C-C-A. Isoleucyl-tRNA synthetase belongs to the class of 2’-attaching synthetases (Table VI), and the 3’-OH group is therefore not required for the aminoacylation reaction. It has recently been found that tRNA1le-A-C-C-d3A,which can be aminoacylated with isoleucine as well as can tRNA1Ie-A-C-C-A,can be completely mischarged with va-
40
MATHIAS SPRINZL AND FRIEDRICH CRAMER
TABLE VII
MISAMINOACYLATION OF NATIVEAND MODIFIEDTRNAiteFROM YEAST WITH VALINE USING
ISOLEUCYL-TRNA SYNTHETASES FROM YEAST (183) Aminoacylation with (pinoVA26,unit tRNA)
tRNA species
Isoleucine
Wine
tRNAiie-A-C-C-A tRNA""-A-C-C-d3A
1470 1600
70 1530
line (183) (Scheme 1).Thus the presence of the nonaccepting 3'-OH group is a deciding factor in charging or mischarging. The only possible conclusion from this experiment is that the 3'-OH group causes the enzyme to be hydrolytic for the incorrect aminoacyl-AMP, or that the 3'-OH itself possesses a hydrolytic function. This important observation could be extended. It is known that synthetases possess a certain hydrolytic capacity (156). In other words, they hydrolyze aminoacyl-tRNA even in cognate systems at a rate that is of course much below the synthetic rate; otherwise, products would not be formed. This hydrolysis is normally not observed in the aminoacylation reaction because a large excess of ATP is usually added and any amino acid released from the tRNA is immediately fed back into the aminoacylating system. Concerning the mischarging of a tRNA"" with valine and the correction step involving the intact -A-C-C-A end (Scheme l),it is obvious that for each correction step 1 mol of ATP is consumed. Therefore, one can measure the endogenous hydrolysis of the enzyme (ATPIPP-independent hydrolysis) by the ATP consumption or the AMP formation, provided that one does not add too great an excess of ATP (184). As seen in the cognate homologous system of yeast tRNAphe,for example, the endogenous hydrolytic capacity of the enzyme is also completely abolished if the
Val + AMP + E1le+ tRNA1Ie-C-C-A
Val-tRNA1le-C-C-A(3'd)+AMP+ €Iie
SCHEME 1. Effect of modification of the 3'-ribose of tRNA on the chemical proofreading (183).In the absence of a 3'-OH group, niisacylation with valine occurs.
THE
-C-C-A END
OF
41
tRNA
nonaccepting hydroxyl, in this case 3'-OH, is missing (Table VIII). With tRNAPhe,the rate of AMP formation versus that of aminoacylation is one to thirty, pointing to the fact that every thirtieth activated phenylalanine is sacrificed by the action or co-action of the nonaccepting hydroxyl. Why is it sacrificed? As is evident from Table VIII, in all cases studied, the nonaccepting hydroxyl is required for this hydrolysis since the native tRNAs are active in this reaction, whereas the chargeable "deoxy" species are not. I n the case of misactivation and mischarging of valine, hydrolysis predominates when the wrong amino acid is recognized. This hydrolytic action of the synthetases is therefore interpreted as part of a general correction mechanism. The synthetases thus seem to be hydrolytic enzymes that hydrolyze the wrongly charged tRNA faster than the correctly charged one. It appears that part of the correctly charged tRNA is sacrificed for the sake of precision of aminoacylation. This mechanism can be considered as a general "proofreading" mechanism and demonstrates directly that energy in the form of ATP is required for specificity. It is interesting to note that tyrosyl-tRNA synthetase, described as nonspecific with respect to 2'TABLE VIII TURNOVER NUMBERSFOR AMINOACYLATION AND AMINOACYL-TRNA SYNTHETASE CATALYZED HYDROLYSIS OF AMINOACYL-TRNA USING VARIOUS 3'-TERMINAL RIBOSE-MODIFIED TRNAs FROM YEAST (184) Specificity
of
Turnover numbers (rnin-l)
synthetase
Amino acid
Phe
Phe
Val
Val
Ser
Ser
Ile
Ile
Ile
Val
TYr
TY
tRNA species tRNAPhe-A-C-C-A tRNAPhe-A-C-C-d3A tRNA""'-A-C-C-A tRNAVa'-A-C-C-d3A tRNASer-G-C-C-A tRNASer-G-C-C-dA tRNA'Ie-A-C-C-A tRNA"'-A-C-C-d3A tRNA'Ie-A-C-C-dA tRNA""A-C-C-A tRNA"'-A-C-C-d3A tRNA"'-A-C-C-dA tRNATYr-A-C-C-A tRNATyr-A-C-C-d3A tRNATYr-A-C-C-dA
Aminoacylation
AMPformation
300 240
10.05 0.70 1.00 0.08 0.70 0.05 17.00 0.05 0.05 8.30 0.05 0.05 0.04
49 13 16.3 8.5 62 45 0 0 2.9
0 537 482 32
0.05 0.04
Amino acylation: hydrolysis 30: 1 50: 1
25: 1 4:l 1:2
-
42
MATHIAS SPRINZL AND FRIEDRICH CRAMER
or 3’-charging, completely lacks such a hydrolytic property. The aminoacylated tRNATYr-A-C-C-A as well as the two deoxy tRNAsTY’are completely stable in presence of the enzyme (Table VIII). What precisely is the role of the nonaccepting hydroxyl in this hydrolysis? One possible interpretation is that the amino acid in its accepting site is stable, but if it migrates to the nonaccepting hydroxyl (e.g., in the case of phenylalanine, from the 2‘ to the 3’),it finds itself in a hydrolytic “editing” site and the aminoacyl ester bond is hydrolyzed. This hypothesis is reasonable since it is obvious that the correct amino acid would be held more tightly in its substrate binding site, which must be close to the accepting hydroxyl. The incorrect amino acid would therefore migrate more rapidly to the 3’-position. This nonaccepting site must have a hydrolytic capacity that brings about the hydrolysis in a way similar to that of the hydrolysis of an ester bond through the serine 119 in chymotrypsin (Fig. 8). However, this phenomenon is apparently not that simple. There is evidence that valine, primarily attached to the 2‘-hydroxyl of tRNA”“, is hydrolyzed from that same hydroxyl only if the 3’-hydroxyl is present (51). In this case, an indirect hydrolytic mechanism of the kind depicted in Fig. 9 must be assumed. Meanwhile, the misactivation/ mischarging of threonine by valyl-tRNA synthetase has been studied in detail (185). Here Thr-tRNAVal-A-C-C-A,where the tRNA is a 2‘ acceptor, could not be isolated, whereas Thr-tRNAVal-A-C-C-d3A was isolated in moderate yield and tRNAVa’-A-C-C-A( 3’NH2) was fully aminoaeylated with threonine by the same enzyme (Table IX). The hydrolysis of Thr-tRNAVa*-A-C-C-A by valyl-tRNA synthetase as measured by determination of ATP consumption during amino-
-
tRNA
A
QJ
\n N : H O
-’G4Lo /
H
I
0 !I
CHNHz
I R’
FIG. 8. Geheral mechanism of hydrolytic action in aminoacyl-tRNA synthetases by migration of the aminoacyl residue to the nonaccepting site and subsequent hydrolysis.
THE
-C-C-A
END OF
43
tRNA
FIG. 9. Diagrammatic representation of the proposed role of water in the hydrolytic correction of aminoacyl-tRNA by isoleucyl- or valyl-tRNA synthetase. Left: Val-tRNA"" as proposed by von der Haar and Cramer (184);righf: Thr-tRNAVa'showing activation of H,O via the threonine O H (open arrows) and via the ribose 3'-OH (dotted arrows) mechanistic pathways (185).
acylation is about 30 times faster than the corresponding breakdown of the homologous system (Table X). However, a certain hydrolytic activity is also observed when Thr-tRNAVa'-A-C-C-d3Ais treated with the enzyme. If the threonine is modified to 0-methylthreonine, no hydrolysis at all takes place. This indicates that O-methylthreonyl-tRNAVa'-A-C-C-Ais only very slowly removed even in the presence of the 3'-hydroxyl of the ribose. Removal of the methoxy group of 0-methylthreonine to form a-aminobutyric acid increases the rate of hydrolysis; a-aminobutyryl-tRNAva' is unstable under aminoacylating conditions. These data again suggest a slightly different mechanism of correction. It is probably necessary that a water TABLE IX AMINOACYLATION OF YEAST TRNA'"' MODIFIED ON THE QI-TERMINALmBOSE THREONINE USINGYEAST VALYL-TRNASYNTHETASE (18.5) tRNA species
Attachment of threonine (pmoliA zBo unit tRNAVal)
tRN A'"'-A-C-C-A tRNA'"'-A-C-C-dA tRNAVn1-A-C-C-d3A tRNAval-A-C-C-A(3'NHz)
52 500 1600
13
WITH
44
MATHIAS SPRINZL AND FRIEDRICH CRAMER
TABLE X TURNOVER NUMBERS FOR HYDROLYSIS OF VARIOUS DERIVATIVES OF YEAST AMINOACYL-TRNA~~‘ BY VALYL-TRNA SYNTHETASE (1851 Turnover number for hydrolysis (min-’) Amino acid Valine Threonine O-Methylthreonine a-Aminobutyric acid
tRNAVa’-A-C-GA tRNAva’-A-C-C-d3A
0.85 5.2 0.28 11.0
0.08 2.15 0.01 3.3
molecule be the vicinity of the ester bond. This water molecule is attracted by an adjacent hydroxyl group in the ribose and is further stabilized by the hydroxyl group of threonine. Thus there seem to be different mechanisms of corrective hydrolysis, which are depicted in Fig. 9. I n the same system, valyl-tRNA synthetase brings about a hydrolytic editing of threonine (186). This has been measured by quenched flow experiments in which a transient Thr-tRNA Va‘-A-C-C-A was observed. What could these mischarging and hydrolytic phenomena mean? These experiments reveal a novel role for the -N-C-C-A end of tRNA. The nonaccepting terminal hydroxyl catalyzes a corrective hydrolysis, which we call “chemical proofreading.” I n this respect, the -N-C-C-A end acts like the catalytic group of an enzyme and can be considered as a coenzyme in the chemical proofreading step. What meaning does chemical proofreading have? At the beginning of this chapter we said that the aminoacylation of tRNA requires an especially high precision of the enzymic process. Apparently this precision cannot be brought about in a single step. Therefore, the process is divided into two processes. Let us assume that the first step, the activation of the amino acid to form the aminoacyl-AMP, has a selectivity for the correct amino acid over the wrong one of a thousand to one. This might perhaps be sufficient for a normal enzymic process, but in protein biosynthesis the mistake is multiplied by the length of the protein. At an average mistake of one in a thousand, every second protein with a chain length of five hundred would be wrong. Therefore nature has devised a second mechanism, chemical proofreading. As we can see from the hydrolytic rates of the wrong versus the correct amino acid, this chemical proofreading causes an increase in specificity of a factor of about a hundred, and therefore the chance of
THE
-C-C-A END
OF
45
tRNA
incorporating a wrong amino acid becomes one in lo5,probably sufficient for protein biosynthesis. B. "Proof reading" by Ami noacyl-t RNA Synthetases, a C hemica I Event a t the -N-C-C-A End
The hydrolytic action of the nonaccepting hydroxyl at the -N-C-C-A end of tRNA described above is a defined chemical reaction catalyzed by the enzyme in cooperation with the -N-C-C-A end of tRNA. In 1974, Hopfield proposed a scheme, called "kinetic proofreading'' (187), which would increase enzyme specificity, especially in the case of aminoacyl-tRNA synthetases capable of selecting one substrate from a group of chemically similar amino acids. In such a family of compounds, the differences in the intrinsic free energy of enzyme-substrate complex formation might not be sufficient to account for the very high aminoacylation specificity, which is certainly better than one mistake in 3000 (188). The kinetic proofreading scheme makes use of the differences in the kinetics of enzymesubstrate binding, as well as an irreversible step whereby the wrong intermediate is released from the enzyme faster than the correct one and subsequently destroyed. For the case of misactivation of valine this would be described by Scheme 2. Chemical and kinetic proofreadings have recently been compared (189). In chemical proofreading as described above, the mischarged Val-tRNA"" is an obligatory intermediate (184). Rapid quenching techniques also show the existence of a covalent intermediate, ThrtRNAVa',in the misactivation of threonine by valyl-tRNA synthetase (186). When tRNAVal-A-C-C-A was replaced by tRNAVa'-A-C-CA(3'NHz),threonine was transferred to tRNAValby valyl-tRNA synthetase, thus a nonhydrolyzable Thr-tRNAVa'-A-C-C-A(3'NH.J was formed (185)(Table IX). Ile+Val+ATP+E1le=
[E1le~Ile.ATPI+ [EILe.Val.ATP]
Val
+
Ile +AMP
SCHEME2. Kinetic proofreading according to Hopfield (187) by fast release of ValAMP and its subsequent irreversible hydrolysis.
46
MATHIAS SPRINZL AND FRIEDRICH CRAMER
Kinetik proofreading potentiates the intrinsic differences in free energy between similar substrates and a single enzyme. Hence in a population of substrates with Kdiss.4
< KdissB < KdlssC < KdlssD
kinetic proofreading will always be selective for the substrate with the lowest Kdiss. However, it is doubtful that, even with such a mechanism of subsequent kinetic selection steps, the enzyme could differentiate between valine and isoleucine, since their binding equilibria are equal within experimental limits (190). Chemical proofreading depends on a defined enzyme-catalyzed reaction based on specificity criteria entirely independent of the free energy of complex formation. Therefore, the above-mentioned basic restriction of kinetic proofreading to select always for the substrate with Iowest Kdiss does not exist. One can easily imagine a case in which, by chemical proofreading, a substrate with a higher Kdiss is selected against a second substrate with lower Kdiss. This would not be very efficient with respect to product formation if both substrates are present in about equimolar amounts. However, such a case might be advantageous if a weakly bound substrate present in a high concentration must be selected against a strongly bound substrate present in only small amounts. The actuality of “chemical proofreading” has been firmly established in two cases (184-186),whereas the real existence of a kinetic proofreading mechanism has not been shown experimentally.
VII. Binding of Aminoacyl-tRNA to Ribosomes The high specificity of the aminoacylation reaction discussed above would be meaningless for the precision of the translation process if the degree of specificity of the codon-dependent binding of aminoacyltRNA to programmed ribosomes were lower than the specificity of aminoacyl-tRNA synthetase interaction. The process of selection of a particular aminoacyl-tRNA from the pool of several species during ribosomal binding must therefore be considered as the second keystep in the determination of the fidelity of translation. Whereas during aminoacylation of tRNA by the synthetase, in principle, any structural feature of tRNA may be utilized for recognition, during the interactions of tRNA with ribosomes, except for the anticodon region, only invariant features in the tRNA structure are candidates as sites of interaction. It is therefore very likely that the -N-C-C-A sequence of tRNA also plays some role in these processes.
THE
-C-C-A
END OF
47
tRNA
A. Formation of (EF-Tu)
*
GTP
-
(Aminoacyl-tRNA) Complexes‘
A generally accepted scheme for the elongation process during protein synthesis on ribosomes (191) involves the formation of a ternary complex between a protein (an elongation factor), GTP, and an aminoacyl-tRNA. Formation of this complex is observed in bacterial as well as in eukaryotic systems. Owing to the greater stability of the elongation factor derived from bacteria, investigations concerning the mode of interaction of elongation factors with aminoacyl-tRNA were performed with pure elongation factor Tu (EF-Tu) from E. coli. Since a very recent review (192) summarizing this work is available, only data concerning the interaction of the -N-C-C-A end of tRNA with the elongation factor and the specificity of EF-Tu for positional isomers of aminoacyl-tRNA are discussed here. The presence of an aminoacyl residue on the tRNA is an absolute requirement for the formation of (EF-Tu) * GTP * (aminoacyl-tRNA) complexes. Uncharged tRNA (2 93, 194), N-acetylaminoacyl-tRNA (195) or fMet-tRNAmet(296) are not substrates for EF-Tu. It appears therefore that the a-amino group of the attached amino acid is one part of the aminoacyl-tRNA that is recognized b y the protein. However, an isosteric substitution of this amino group with a hydroxyl group does not influence the ternary complex formation (197).The inability of the nonformylated E . coli Met-tRNAmet to form a complex with (EFTu) * GTP was ascribed to the absence of the last base-pair in the aminoacyl stem of this tRNA, where the fifth nucleoside from the 3’ end is C and the first nucleoside in the polynucleotide chain is A. By conversion of this C residue to U with sodium bisulfite, an A * U basepair was generated; the modified Met-tRNAfMetwas able to form a teniary complex (108).This experiment is indicative of the necessity of a stabilized -N-C-C-A elid confoiiiiatioii for interaction with
E F-TU. It was suggested by Schulman et al. (108)that the aminoacyl residue of aminoacyl-tRNA may have a conformation that allows an ionic interaction between the protonated a-amino group of the amino acid and the 5’-phosphate group of the first nucleotide ’ i n the polynucleotide chain (Fig. 4). This suggestion was based on the observation that an aminoacyl-tRNA from which the 5’-phosphate was removed by phosphatase treatment did not form a ternary complex. The decreased binding of oligonucleotides complementary to the -A-C-C-A end of Phe-tRNAPhe-A-C-C-A(3’NH2)as compared to unSee article by Berinek in Vol. 21 of this series. [Ed.]
48
MATHIAS SPRINZL AND FRIEDRICH CRAMER
charged tRNA (96)is in support of this model. Similarly, the length of the 3’ end of tRNA may play a role in this interaction, since PhetRNAPhe-A-C-C-C-A,having an extra nucleotide in the 3’ end, does not form a stable ternary complex (198). The presence of nucleotides other than -A-C-C-A at the 3’ terminus has no substantial effect on ternary complex formation. Thus PhetRNAPhe-A-iSC-i5C-A, Phe-tRNAPhe-A-C-i5C-A,Phe-tRNAphe-A-C-s2C-A (199) and Phe-tRNAPhe-A-C-C-F(200) from yeast are active in elongation-factor-dependent binding to ribosomes from rabbit reticulocytes. However, an aminoacyl-tRNA with uridine at the 3’ end was inactive in ternary complex formation, which suggests a role for the 4-NH, group of the penultimate cytidine residue in this interaction (108). The 2-0x0 group of this cytidine is not involved, as its replacement with a sulfur atom or a thioalkyl group (Fig. 5) does not affect participation in ternary complex formation (167). The rate of tumbling of a nitroxyl radical attached to position 2 of the thiocytidine-75 residue of yeast tRNAPhe-A-C-s2C-A as measured by ESR spectroscopy, does not change if this Phe-tRNAPheis free or in ternary complex with (EF-Tu) * GTP (167).This shows that the 2-0x0 group ofthe cytidine-75 residue in unmodified Phe-tRNAPhe-A-C-C-Ais also free when the tRNA is in a ternary complex, and it should therefore be able to interact with the ribosomal A-site as suggested b y several authors (201,202). Recognition of aminoacyl-tRNA by E . coli EF-Tu is strongly affected by modification of the ribofuranosyl residue to which the amino acid is attached. Ofengand and Chen investigated the substrate prop(Fig. 7c; X = tRNAPhe-C-C-from erties of Phe-tRNAPhe-A-C-C-A,xi-red yeast) (178,203) and found that this tRNA was inactive with respect to ternary complex formation. The reason for this loss of activity was later ascribed to the conforinational disturbances at the 3’ end arising from the opening of the ribose ring rather than to the inability of the phenylalanyl residue of this tRNA to migrate between the separated 2‘- and 3‘-positions (204). The specificity of the elongation factor Tu for 2’- or 3’-aminoacyl isomers of charged tRNA was investigated by Ringer and Chlidek (205) using synthetic aminoacylated fragments of tRNA: C-A-Phe, C-A(2’H)Phe, and C-A(2‘Phe)H, where A-Phe, A(2’H)Phe, and S‘-O-~-phenylA(2‘Phe)H are 2‘(3’)-O-~-phenylalanyladenosine, alanyl-2’-deoxyadenosine and 2’-O-~-phenylalanyl-3’-deoxyadenosine, respectively (Fig. 743; X = Cp). Formation of complexes between (EF-Tu) GTP and these aminoacylated oligonucleotides was demonstrated by the ability of oligonucleotides to prevent
-
THE
-C-C-A END OF tRNA
49
the binding of (EF-Tu) * GTP to nitrocellulose filters. A similar effect is exhibited b y native aminoacyl-tRNAs, whereas nonacylated tRNAs do not affect the binding of (EF-Tu) * GTP to nitrocellulose filters (206). In this assay, the oligonucleotide C-A(2‘Phe)H showed the same activity as C-A-Phe, whereas C-A(2’H)Phe was inactive, indicating that the aminoacyl residue must b e bound to the 2’-position in order to bring about an interaction of aminoacylated dinucleotides with EF-Tu. Unfortunately, the complex formation of the isomeric oligonucleotides with EF-Tu could not be demonstrated by more direct gel filtration experiments. It is known that aminoacylated singlestranded fragments of tRNA, such as U-C-C-A-C-C-A-Ala, do not form complexes with EF-Tu sufficiently stable for isolation by gel filtration (207). More recently the 2’ vs. 3‘ specificity of EF-Tu was investigated using “nonisomerizable” aminoacyl-tRNAs (depicted in Fig. 7), which can b e obtained by enzymic aminoacylation of modified tRNAs (139, 208). Thus Tyr-tRNATYr-A-C-C-d3A and Phe-tRNAPhe-A-C-C-A(2‘NH2) from yeast (Fig. 7a,d) have the amino acid attached to the 2’-position while Tyr-tRNATfl-A-C-C-dA and Phe-tRNAPhe-A-C-C-A(3’NH2) from yeast (Fig. 7b,e) have it in the 3’-position. Both Tyr-tRNATYr-A-C-C-dA and Tyr-tRNATYr-A-C-C-d3Aformed ternary complexes with (EFTu) . GTP that could be isolated by gel filtration (139). Both tRNAs to which the amino acid was attached via an amide bond, regardless of the position of the attachment (Fig. 7d,e), were inactive and did not form ternary complexes with EF-Tu. It is, therefore, not surprising that puromycin, in which the aminoacyl residue is also attached via an amide bond, is also inactive in this interaction (207).5Although Hecht et al. observed (208) no difference between the 2‘ and 3‘ isomers ofE. coli tRNA in their binding to (EF-Tu) . GTP, a significant preference for the 2’-aminoacyl analog of tRNA was detected b y an experiment in which Tyr-tRNATYr-A-C-C-dAand Tyr-tRNATYr-A-C-C-d3A were competing for a limiting amount of elongation factor (139). Elongation factor Tu from E. coli can, therefore, recognize both the 2‘- as well as the 3‘-aminoacyl derivatives of charged tRNA. Because the site of attachment of the amino acid during aminoacylation of tRNA is not the same for all tRNAs (118, 134) (Table VI), and isomerization can occur spontaneously when the aminoacyl-tRNA is not bound to proteins (66),it is reasonable to conclude that (EF-Tu) * GTP recognizes both isomers. The difference in affinity of the two analogs is probably not enough to cause a selection of the stronger binding 2’ isomers in the in vivo binding process, since the concentration of EF-Tu in the prokaryotic cell (209) is significantly higher than the
50
MATHIAS SPFUNZL AND FFUEDFUCH CRAMER
binding constant of either isomer (192). However, after the ternary complex is formed, the conversion of the aminoacyl residue to the more favored 2’-position probably takes place. It was therefore suggested (139) that EF-Tu recognizes both the 2‘ and 3’ isomers of aminoacyl-tRNA and fulfills the function of producing a single population of (2‘-aminoacyLtRNA) (EF-Tu) . GTP complex, which is needed for the next step of the elongation process (138, 204, 210) (Scheme 3). The fact that both the 2’- and 3’-O-aminoacyl-tRNAs interact with EF-Tu allows one to speculate about the stereochemistry of the interacting sites during ternary complex formation. It is reasonable to assume that the a-amino group of the aminoacyl-tRNA is one of such interacting sites (193-196). If, therefore, this a-amino group is recognized by EF-Tu, it must be possible to bring it to the same steric position regardless of the point of attachment of the amino acid to the ribose. By examination of a molecular model of 2’(3’)-0aminoacyladenosine it appears that this is possible only if the carbonyl group of the ester bond is distorted from its optimal conformation (78, 79) and, even then, only one reasonable conformation of the amino acid residue can be found that fulfills this requirement (204) (Fig. 10). It is possible that the rigid amide bond restricts the conformation of the amino acid residue required for ternary complex formation. This could explain the inability of Phe-tRNAPhe-A-C-C-A(3’NH.J and PhetRNAPhe-A-C-C-A(2‘NHz)to interact with EF-Tu. There is a further consequence of the model presented in Fig. 10, namely, that EF-Tu probably does not interact directly with the side chains of the amino
I m-tRNA
1
aa-tRNA
7
~
7OSrib mRNA aa-tRNAIA”) EF-TU GTP
%acn
pep-tRNA(P)
&!$ti,
70Srib
-
mRNA aa-tRNA(A) pep-tRNA(P1 I
SCHEME 3. Role of transacylation during the binding of aminoacyl-tRNAs to ribosomes.
THE
-C-C-A E N D
OF
51
tRNA
b
Q
C5'
CS
M'
HN
C
d
FIG. 10. Schematic drawing of the aminoacyl residues of the modified aminoacyltRNAs. (a) Torsion of the 0-CO bonds leading to a conformation where the amino group of both "ester" analogs adopts an identical position, irrespective of whether the position of attachment is the 2' or 3'-hydroxyl group. (I)) T h e rigid conformation of the "amido" analogs prevents the accommodation of the amino groups to such identical positions. (c) Model (a) viewed from above. (d) Structure of puromycin.
acids because these are placed, depending on the position of attachment of aminoacyl residue, on sites opposite to each other.
B. (EF-Tu)-Dependent Binding of Aminoacyl-tRNA to Ribosomes The aminoacyl-tRNA binding or codon recognition step results in the addition of the appropriate aminoacyl-tRNA to the (peptidyltRNA) . mRNA ribosome complex (191,211). It is somewhat difficult to investigate this process in an in vitro system with a single purified aminoacyl-tRNA, since for such experiments significant simplifications must be made. (a) The natural messenger must be replaced b y a synthetic species that allows an investigation of only one specific aminoacyl-tRNA; usually poly( U) in combination with Phe-tRNAPheA-C-C-A or poly(A) in combination with Lys-tRNALYs-A-C-C-Aare used. (b) During the in vivo elongation process, when an internal codon of mRNA is expressed, the ribosomes already carry a nascent
52
MATHIAS SPFUNZL AND FRIEDRICH CRAMER
peptidyl-tRNA in the P-site. This is usually not the case in in uitro experiments. However, the presence of peptidyl-tRNA in the P-site influences the enzymic binding of aminoacyl-tRNA to the A-site (172,212)with respect to rate and specificity. (c) Binding experiments must be performed on isolated ribosomes. However, ribosome activity is strongly dependent on the method of preparation, and there are still doubts about the integrity of such ribosomal preparations. Since an appropriate assay that closely mimics the situation during the in viuo elongation process is not available, it is difficult to measure exactly the binding of aminoacyl-tRNA to ribosomes. Consequently, the mechanism leading to the high specificity in codon-anticodon interactions and to the insertion of the proper amino acid into a polypeptide chain is not fully understood (213). Data in the literature (191,211) are consistent with a model in which the ternary complex, (aminoacyl-tRNA) * (EF-Tu) GTP, binds codon-specifically to the A-site of programmed posttranslocational ribosomes. At this stage, the aminoacyl-tRNA is still not able to react with the peptidyl-tRNA to form a new peptide bond. Peptide bond formation takes place only after an “accommodation” step involving cleavage of GTP and release of (EF-Tu) @DP from ribosomes (214218).The mechanism that leads from this “unreative” to a “reactive” aminoacyl-tRNA is not clear. The use of aminoacyl-tRNA species in which the terminal adenosine is modified may help solve this problem. Especially useful are those species in which the aminoacyl residue is so bound that 2‘,3‘-isomerization is not possible. Using the “nonisomerizable” species shown in Fig. 7, an attempt was made to answer the following questions: (a) Which positional isomer is used during the binding of tRNA to ribosomes? (b)What is the function, if any, of migration of the amino acid between the 2’-and 3‘-positions? Because of the limitation that mainly poly(U) is necessary for such investigations, only Phe-tRNAPhe-A-C-C-d3A, Phe-tRNAPhe-A-C-CA(2’NHz), and Phe-tRNAPhe-A-C-C-A(3‘NHz),obtained by enzymic aminoacylation of the modified tRNAphespecies (Table V), and PhetRNAPhe-A-C-C-dA,prepared by incorporation of 3’-O-phenylalanyl2‘-deoxyadenosine into tRNA (141),were utilized in the previous studies. The results of these investigations are summarized in Table XI. Most recently Lys-tRNALYS-A-C-C-dA and Phe-tRNALYs-A-C-C-d3A from E . coli were also used for the investigation of poly(A)-dependent enzymic binding to the ribosomal A-site (38). Chinali et al. (204) studied the properties of Phe-tRNAPhe-A-C-Cd3A where the amino acid is attached at the 2’-position. This tRNA
-
4
2 t
9 9 >
zU
TABLE XI ACTIVITY OF “NONISOMERIZABLE” T R N A SPECIES ~ ~ ~ IN RIBOSOMAL PEPTIDE SYNTHESIS USING SYSTEMa. AND YEAST T R N A ~ ~ % Position of Phe-tRN APhe-A-C-C aminoacyl residue -A --dA
-d3A -AOXi-red
-A(3‘NH,) -A(2’NH,)
2‘(3’) 3’ 2’ 2‘ 3’ 2‘
(aa-tRNA) . (EF-Tu) . GTP complex formation
+ + (139,208) + (139,208) - (I 78) - (139) - (139)
(EF-Tu)-dependent binding
+ - (38)‘ + (140,204) - (178,204) - (140) ND
0
Escherichin coli CELL-FREE
AN
Acceptor activity
+ + (141) (141,204) - (178,204) + (125) ND -
Donor activity
+ ND
- (204) - (204)
- (125) ND
References are given in parentheses; ND, not determined. Only activities comparable in rate and extent with native Phe-tRNAPheare considered. Determined using Lys-tRNALYS-A-C-C-dA from E. coli; binding is much less stimulated than with Phe-tRNALYs-A-C-C-d3A (T. Wagner and M. Sprinzl, unpublished results).
7
%
z *
54
MATHIAS SPFUNZL AND FRIEDRICH CRAMER
shows an (EF-Tu)-dependent binding to a poly(U) * (AcPhe-tRNAPheA-C-C-A) ribosome complex, which is similar with respect to extent as well as rate to that of native Phe-tRNAPhe-A-C-C-A,Although the binding of this Z'-aminoacylated tRNA triggered the hydrolysis of GTP to GDP, only a very slow formation of AcPhe-Phe was observed. Furthermore, Phe-tRNAPhe-A-C-C-d3A bound to ribosomes can easily be replaced by native Phe-tRNAPhe-A-C-C-Aif the ribosomal complex with prebound 2'-aminoacyl-tRNA is incubated with native (PhetRNAPhe-A-C-C-A)* (EF-Tu) GTP. This observation is very similar to the case when using a comparable assay in which Phe-tRNAPheA-C-C-A - (EF-Tu) and the nonhydrolyzable GMPP(CH2)P is bound to ribosomes (214, 215, 218). Here, too, the aminoacyl-tRNA cannot participate in peptide bond formation and can be displaced by the (Phe-tRNAPhe-A-C-C-A)(EF-Tu) - GTP complex(214). It is, therefore, likely that whereas in the second case the lack of GTP hydrolysis and release of (EF-Tu) * GDP prevented the incorporation of the amino acid into the dipeptide, with Phe-tRNAPhe-A-C-C-d3A it is the position of the attachment of amino acid that holds it in an unreactive form. Using native tRNAPhe-A-C-C-Ain a similar binding experiment it is not possible to isolate a ribosomal complex, after the release of (EFTu) * GDP, in which aminoacyl-tRNA is in the A-site and peptidyltRNA is in the P-site. A rapid peptide bond formation takes place (211, 215), leading to a situation where peptidyl-tRNA is in the A-site and uncharged tRNA is in the P-site (pretranslocation complex). Although the release of (EF-Tu) - GDP from ribosomes after binding of PhetRNAPhe-A-C-C-d3A was not directly demonstrated, it is likely that the (EF-Tu) * GDP is released, as hydrolysis of GTP to GDP was observed during the (EF-Tu)-dependent binding of this modified tRNA. Therefore, binding of Phe-tRNAPhe-A-C-C-d3A results in (EF-Tu) . GDP-free ribosomes containing aminoacyl-tRNA in the A-site and peptidyltRNA in the P-site, and does not lead to spontaneous formation of the peptide bond because the aminoacyl residue is trapped in the unreactive 2'-position of the terminal adenosine (204). Such a bound aminoacyl-tRNA can still be removed from ribosomes by the action of an excess of (EF-Tu) * GTP (aminoacyl-tRNA) (204). Unfortunately, Phe-tRNAPhe-A-C-C-dA,bearing the amino acid in the 3'-position, cannot be prepared by enzymic aminoacylation, and therefore no comparable study has been performed on the purified system using this isomer. Hecht et al. (141) prepared a mixture of tRNA-N-C-C-d3Aaminoacylated with phenylalanine, in which only tRNAphe(about 2% of the total tRNA) carried phenylalanine. PhetRNA-N-C-C-dA was prepared through the elongation of a tRNA-C-C
THE
-C-C-A END
OF
tRNA
55
mixture with 3’0-phenylalanyladenosine 5’-diphosphate using polynucleotide phosphorylase and resulting in a population of tRNAs with heterogeneous anticodons but homogeneously phenylalanylated 3‘ termini. These two preparations were compared in their inhibitory activities for (EF-Tu)-dependent binding of Phe-tRNAPhe. Both species showed about 50% inhibition of the (EF-Tu)-dependent binding, indicating that both 2‘ and 3’ isomers can interact with the ribosomal A-site. Later the activity of Phe-tRNAPhe-A-C-C-A(3’NHz) in cell-free poly(Phe)-synthesis was investigated and it was found that phenylalanine can be transferred from such a tRNA to a dipeptide (125).Although this is evidence that a 3’-aminoacylated tRNA can act as a peptide acceptor, the attachment of this tRNA to the ribosome was probably not mediated by EF-Tu, since yeast Phe-tRNAPhe-A-C-CA(3‘NHz) did not interact with this elongation factor (139). It is more likely that a nonenzyinic binding of this modified tRNA to ribosomes led to the observed results. There is evidence that in the absence of elongation factor Phe-tRNAPhe-A-C-C-A(3’ NH,) has a higher affinity for ribosomes than does the native Phe-tRNAphe( 1 72). Recently LystRNALYS-A-C-C-dAand Phe-tRNALYs-A-C-C-d3Afrom E . coli were prepared by enzymic aminoacylation (Table V). These two nonisomerizable species make possible the investigation of the 2’,3’ specificity during (EF-Tu)-dependent binding to ribosome by a direct comparison using poly(A) as messenger. The preliminary results of this work indicate that during the residence time of (EF-Tu) . GTP on ribosomes, the aminoacyl residue is exclusively in the 2’-position (38). Inhibition of (EF-Tu)-dependent binding of Phe-tRNAPheto E . coli ribosomes by aminoacylated tRNA fragments C-A-Phe, C-A(2’H)Phe, and C-A(2’Phe)H was investigated by Ringer e t al. (210). Both “nonisomerizable” aminoacylation fragments inhibited the attachment of Phe-tRNAPhe-A-C-C-Ato ribosomes. Similarly the transfer of a peptidyl residue to puromycin is inhibited when 2‘- or 3‘phenylanylated adenosine derivatives are present (219).The observation that these inhibitory effects are similar regardless of the position of the 2’ or 3‘ attachment of the amino acid led to the conclusion that the ribosomal A-site contains two binding loci, one for the amino acid attached to the 2’-position and one for the amino acid attached to the 3‘-position. Similar conclusions can also confidently be drawn from experiments performed with modified tRNAs (125,136,204). The function of the 3(-site is obvious from an analysis of the data in the literature, which show that only tRNAs or tRNA fragments on which the amino acid is attached to the 3’-position of the terminal
56
MATHIAS SPRINZL AND FRIEDRICH CRAMER
adenosine act efficiently as acceptors of the peptidyl residue. Thus puromycin (220) Phe-tRNAPhe-A-C-C-A( 3'NH2) (125, 172), C-A(2'H)Phe, C-A(2'OMe)Phe (219,221) give rise to dipeptide formation if bound to ribosomes containing peptidyl-tRNA in the P-site. On the other hand, aminoacyl-tRNAs or appropriate fragments on which the amino acid is linked to the 2'-position, as in the 2'-analog of puromycin (222), Phe-tRNAPhe-A-C-C-A,,i-red(178, 179, 203, 204), Phe-tRNAPhe-A-C-C-d3A (141,204), C-A(2'Phe)H (219,221 ), are inactive as acceptors or accept the peptidyl residue at a considerably lower rate (204). Krayevsky et al. in their recent summarizing article about the activities of different aminoacylated fragments in the ribosomal peptidyl transfer reaction reached the same conclusion (223).Thus it is clear that the aminoacyl residue of the aminoacyl-tRNA is bound to the 3'-OH group of the terminal adenosine during the process of peptide bond formation. Therefore the 3'-locus of the ribosomal peptidyltransferase center has the function of binding the aminoacyl residue in an optimal position for peptidyl transfer. The role of the 2'-aminoacyl binding locus of the peptidyltransferase center, the existence of which is apparent from experiments using both 2'-"nonisomerizable"-aminoacyl fragments (210, 219) and tRNAs (141,204), becomes clear when the mechanism of the enzymic binding of aminoacyl-tRNA is considered. (EF-Tu) * GTP, although able to bind both 2'- or 3'-aminoacyl-tRNAs (139,208) shows a preference for the 2'-aminoacylated species (139, 205) and in the ribosomal binding process utilizes only the 2'-O-aminoacyl derivative of tRNA. The experiments with nonisomerizable aminoacyl-tRNA and nonisomerizable aminoacylated dinucleotides strongly support this suggestion (38, 204, 210). This means that in the ribosomal complex involving ribosome, mRNA, aminoacyl-tRNA, EF-Tu, GTP, and peptidyl-tRNA (Scheme 3),the aminoacyl residue is on the 2'-position and positioned in the 2'-locus of the peptidyltransferase center. At this stage, the aminoacyl residue is not reactive and not available for peptide bond formation. Therefore, a 2' -+ 3'-transacylation must take place prior to peptide-bond formation (210).This probably occurs after GTP hydrolysis and release of (EF-Tu) * GDP, which constitutes at least part of the ribosomal 2'-binding locus of aminoacyl-tRNA, from the ribosomes. The transacylation to the 3'-binding site, which is situated entirely on the ribosomes, can then occur spontaneously (Fig. 11).Only by this 2' -3 3'-transacylation can the aminoacyl residue of aminoacyl-tRNA become reactive in the peptidyl transfer reaction. In this model, the elongation factor Tu acts as an agent that prevents the transacylation of the aminoacyl residue. This suggestion is
THE
-C-C-A END
OF
57
tRNA
2-complex
T
/
@
3-complex
‘AA-NH,
Peptide
FIG. 11. Elongation factor Tu (EF-Tu) retains the aminoacyl residue of aminoacyltRNA in its unreactive 2’-position during the codon-anticodon selection process.
strongly supported by the observation that in native Phe-tRNAphe, where there is no structural hindrance for transacylation, no peptidyl transfer can take place if GTP hydrolysis and (EF-Tu) * G D P release is prevented (214, 215). This can easily be understood if one assumes that the aminoacyl residue of aminoacyl-tRNA is held, during the presence of (EF-Tu) . GTP on the ribosomes, in its unreactive 2’-site. Clearly, if the GTP hydrolysis, necessary for EF-Tu release, is blocked b y a nonhydrolyzable GTP analog, GMPP(CH.JP (214) or GMPP(NH)P (216), the required transacylation from the 2’- to the 3’-position cannot occur. It can therefore be concluded that the interaction of an amino acid with the 2’-locus of the A-site of the peptidyltransferase center and the following transacylation are necessary steps in the process of (EF-Tu)-catalyzed binding of aminoacyl-tRNA to ribosomes (210).This, however, is not the case for the nonenzymic binding of aminoacyl-tRNA or aminoacylated fragments of the 3’ end of tRNA, under conditions that do not require EF-Tu. C. Transacylation from 2‘ to 3’ Controls the Selection Process
The aim of this section is to provide a possible explanation for the data concerning the 2’ vs. 3‘ specificity of the A-site of ribosomal peptidyltransferase and for the role of the 2‘ -+ 3’ migration of the aminoacyl residue during the ribosomal decoding process. If the a-amino group of the amino acid attached to the tRNA is
58
MATHIAS SPRINZL AND FRIEDRICH CRAMER
brought into the vicinity of the carbonyl group of the ester linkage by which the peptidyl residue is bound to peptidyl-tRNA, a very fast reaction, leading to the formation of the new peptide bond, should be expected. Assuming that the stereochemical arrangement of these two reacting groups on the peptidyltransferase center is optimal, such a reaction must occur with a half-time of less than second. In addition, at 37°C and pH 7, no activation of the reacting group is necessary for the reaction to take place. This can be deduced from the half-life of the migration of the amino acid on aminoacyl-tRNA, where a similar optimal steric arrangement facilitates the transfer of the amino acid (66). If the vicinal hydroxyl group is replaced by the more nucleophilic amino group and a new amide bond is formed, the rate of aminoacyl transfer could be expected to be even higher. An interesting model for such a reaction is tRNAPhe-A-C-C-A(3’NH2), where the 2’ ester is first formed by enzymic aminoacylation and then isomerizes rapidly to the 3’ amide. If a sufficiently large pool of aminoacyl-tRNAs and sufficient time are available, any protein could therefore be elongated by a spontaneous process. The role of the ribosome is to produce the particular protein coded by a specific messenger and to do so with an error frequency close to zero. In the process of binding aminoacyl-tRNA to ribosomes, therefore, two principal functions must be accomplished: (a) the a-amino group of the amino acid must be brought to the position that permits reaction with the ester carbonyl of the peptidyl-tRNA; (b) codonspecific selection of the appropriate aminoacyl-tRNA from the pool of all species must take place. Considering the expected spontaneous peptide bond formation if the first function is fulfilled, it is clear that the two processes must be coupled during the binding interaction. In other words, an efficient selection of the proper aminoacyl-tRNA can take place only under conditions where there is enough time for the noncognate aminoacyl-tRNA to dissociate from the ribosome during the course of the selection. The significance of such an interplay between the speed and error rate in protein synthesis was recently pointed out by Kurland et al. (213, 224). According to this scheme, peptide-bond formation, which can be considered an irreversible process, must take place at a slower rate than the series of reversible reactions leading to the selection of the correct aminoacyl-tRNA (224). Since, however, the transfer of the peptidyl residue is expected to be fast, there must be a mechanism that prevents this reaction during the codon-specific selection of aminoacyl-tRNA. The model presented in Scheme 3 provides a possible pathway for such a mechanism. Aminoacyl-tRNA synthetase releases the
THE
-C-C-A
END OF
tRNA
59
aminoacyl-tRNA as a 2'- or 3'-aminoacyl species (Table VI) and remains as a mixture if free in solution (66). In the formation of the ternary complex, both isomers can be utilized (139), and (EF-Tu) . GTP (2'-aminoacyl-tRNA) is formed (step I), which is bound to the ribosomes. At this stage, the amino acid is in its unreactive 2'-position and the whole ternary complex can still easily dissociate from the ribosomes (204) if the codon-anticodon interaction is not correct (selection step 11). In the case where a cognate aminoacyl-tRNA is bound, hydrolysis of GTP and dissociation of (EF-Tu) . GDP from the ribosomes takes place. The mechanism leading to hydiolysis of GTP and release of (EF-Tu) GDP is not yet understood. Since, during the (EF-Tu)-dependent binding of Phe-tRNAPhe-A-C-C-d3A, hydrolysis of GTP was observed, it is assumed that this takes place prior to 2' -+ 3' transacylation when native aminoacyl-tRNA is used. The release of (EF-Tu) . GDP must therefore be triggered by proper codon-anticodon matching and the subsequent processes involved in aminoacyltRNA selection (224). Dissociation of (EF-Tu) . GDP is the final step in the aminoacyl-tRNA selection process, after which the aminoacyl residue moves to the reactive position by 2' -+ 3' transacylation (step 111). According to this model, there is a sufficient time lag available during the residence time of EF-Tu on the ribosome for selection of the proper aminoacyl-tRNA before irreversible peptide bond formation takes place. After the binding and transacylation is completed, peptide bond formation (step IV) occurs, leading'to the pretranslocational ribosomes with peptidyl-tRNA in the A-site. The elongationfactor-controlled migration of the aminoacyl residue of aminoacyltRNA from the nonreactive 2'- to the reactive 3'-position during the binding to ribosomes is thus necessary for an error-free translation.
-
VIII. Ribosomal Peptidyltransferase Center A. Peptide-Bond Formation As discussed above, during peptide-bond formation the peptidyl residue is transferred to the aminoacyl-tRNA to which the amino acid is bound at the 3'-position of the terminal adenosine. It was suggested (225)that the peptide residue may be transferred in a two-step process by the intermediate formation of a 3'-aminoacyl-2'-peptidyl-tRNAat the A-site. This possibility was later eliminated b y experiments showing that the absence of the 2'-hydroxyl group or its protection by a methyl group does not lead to loss of acceptor activity. Thus GA(2'H)Phe and A(2'0Me)Phe (219, 221) are active as acceptors.
60
MATHIAS SPRINZL AND FRIEDRICH CRAMER
Which positional isomer is required for the donor activity of peptidyltRNA is not clear at the present time. It was demonstrated that both AcPhe-tRNAPhe-A-C-C-d3A(141, 172, 204) and the isomeric AcPhetRNAPhe-A-C-C-dA(141)interact with the P-site of E . coli ribosomes, as they strongly inhibit the binding of AcPhe-tRNAPhe-A-C-C-A. However, neither of these nonisomerizable analogs was active as a peptide donor. More recently, A(2’OMe)fMet and A(2’fMet)OMe were tested for donor activity; both isomers were inactive (223). Obviously, the presence of the free vicinal 2’(3’)-hydroxyl group is important for donor activity. It remains to be elucidated whether the function of this adjacent free hydroxyl is to activate the ester bond of the peptidyl residue, to participate in the interaction with ribosomes, or to serve as an intermediate acceptor via transpeptidation. Binding of Gly,PhetRNAPhe-A-C-C-d3A was investigated in more detail. It is known that the peptidyl-tRNA bound to the ribosomal P-site stimulates the knzymic binding of aminoacyl-tRNA to the A-site (212). AminoacyltRNA or uncharged tRNA does not show this effect. Gly,PhetRNAPhe-A-C-C-d3Awas as active as the Gly,Phe-tRNAPhe-A-C-C-A using this assay (172), indicating that the absence of the 3’-hydroxyl does not influence the specificity of binding of the peptidyl-tRNA to the P-site. B. Interaction with the -C-C-A End The 3’-terminal’nucleotides of tRNA play a major role in the interaction of both aminoacyl- and peptidyl-tRNA with ribosomal donor and acceptor sites. This was demonstrated by early investigations (for reviews, see 6,226) of the acceptor and donor activities of 3’-terminal fragments of tRNA carrying aminoacyl or peptidyl residues. The minimal requirement for the acceptor activity is the presence of aminoacylated adenosine in the A-site (222). However, in this case the properties of the amino-acid side chain also play an important role. An aminoacyl-adenosine carrying a lipophilic, aromatic amino acid, such as phenylalanine or tyrosine, is a better acceptor than one bearing lysine, methionine, alanine, leucine, valine, serine, or proline, whereas A-Gly and A-Trp do not posses acceptor activity (233 and references therein). Whether this effect is due to interaction of the amino-acid side chain with a specific hydrophobic or hydrophilic pocket of the ribosomes, as suggested by Harris and Symons (227), or by an effect of the amino-acid side chain on the conformation of adenosine (223) is not yet clear. Making the reasonable assumption that a certain conformation of the aminoacyl-adenosine residue of the aminoacyl-tRNA is required for effective interaction with the
THE
-C-C-A END OF tRNA
61
ribosomal A-site, the presence of an aromatic side-chain could stabilize such a conformation by stacking interactions. NMR and CD measurements of some 3‘-aminoacylated adenosine derivatives reveal that the distribution of conformers is dependent of the properties of the amino acid side-chain (228). Although some 2’(3’)-aminoacyl-adenosinespecies possess acceptor activity, the efficiency of their interaction with ribosomes is increased if cytidylic residues are attached to the 5’-hydroxyl of the adenosine. Thus A-Gly is not an acceptor, whereas C-A-Gly is (229). The effect of the cytidylic residue can be explained by occupation of an additional C-specific binding site on the ribosomal peptidyltransferase (227) or again by an influence of the cytidine residue on the conformation of the neighboring aminoacyl-adenosine. The cytidylic residue in C-A-Gly is involved in stacking interactions (80), which may explain its influence on the acceptor activity of this fragment. The existence of stacking interactions between the bases of the -C-C-A end and their importance for the activity of tRNA has also been demonstrated (Section 111). A-Phe, I-Phe, and C-Phe are acceptors, whereas U-Phe and G-Phe are not (6, 223). The activity of 2’(3’)-phenylalanyl-purine ribonucleosides is not substantially influenced by a substitution at the 6-position, provided that the basic electronic configuration of the purine ring is not altered by the substitution (230). However, electronic factors appear to have less importance than do conformational changes. Thus 2‘(3’)-phenylalanyl-8-bromoadenosine, with the purine residue in a s p conformation (231) is, unlike A-Phe, a substrate of only moderate activity (230). Isosteric replacement of the N-7 of the purine ring b y CH does not lead to loss of acceptor activity, as 2’(3’)phenylalanyl-tubercidin can accept a peptidyl residue (230).5Similarly, a substitution at N-1 of the purine ring, as in phenylalanyl-1 :N6ethenoadenosine, does not affect acceptor activity (202). It appears, therefore, that the aminoacyl terminus of aminoacyl-tRNA does not interact with the A-site of the ribosomal peptidyltransferase center by Watson-Crick base-pairing. Such nucleic acid-nucleic acid interaction has been suggested on the basis of affinity labeling experiments (201). It is more likely that the 3’ end of the aminoacyl-tRNA interacts with ribosomal protein(s) that are part of the peptidyltransferase center (232). Stacking interactions, which stabilize the conformation of the terminal aminoacyl-adenosine, are important in this process. Direct involvement of the -C-C-A terminus of tRNA in the interaction with the P-site of the peptidyltransferase center of E . coli ribosomes was demonstrated by Monro et al.: under special reaction
62
MATHIAS SPRINZL AND FFUEDRICH CRAMER
conditions, C-A-A-C-C-A-Met, A-C-C-A-fMet and C-C-A-fMet are equally active as donors of the peptidyl residue, while C-A-fMet and A-fMet are inactive (233).These results indicate that the entire -C-C-A terminus is required for reaction at the P-site. However, pA-fMet can also donate the peptidyl residue if incubated with aminoacyl-tRNA and E . coli ribosomes or their 50 S subunit under appropriate conditions (234).Whereas the C-A-C-C-A-fMet fragment acts as a donor at 10+ M, M pA-fMet is necessary to observe the peptidyl transfer. The donor activity of pA-fMet is stimulated significantly by CMP (235).The donor activity of pA-fMet in the presence of CMP is close to that of C-A-met. At the same time CMP does not stimulate the donor activity of C-A-Met (236). These results led to a model proposing a specific binding site for the terminal adenosine and for the penultimate cytidine residue at the P-site of the ribosomal peptidyltransferase center (223).According to this model, the second cytidine residue from the 3‘end of tRNA does not have such a specific binding site, the occupancy of which would influence the peptidyltransferase reaction. However, using a bisulfite-modified fMet-tRNAmet-A-U-C-Afrom E . coli, Sundari et al. demonstrated that the conversion of C,, to U75deactivates this tRNA with respect to both the initiation-factor-dependent binding to ribosomes and donor activity (25).After removal of the -U-C-A end and reconstitution of -C-C-A, the tRNA regained its activity as peptide donor. This indicates that the nucleotide sequence at the 3’ terminus is important for formation of a stable (aminoacyl-tRNA) * ribosome complex and for the correct positioning of the initiator tRNA, especially its 2’(3’)-formylmethionyladenosineend, in the P-site. The conversion of C,,to U,, may lead to a disruption of the confonnational integrity of the 3’end making it difficult to position the reactive site of tRNA into the correct location on the peptidyltransferase. Using their novel fragment reaction, Cerniiet al. (237)investigated the donor activities of different synthetic 2’(3’)-formylmethionyl-nucleoside5’phosphates. The activity of pA-fMet, resembling the 3’ end of the natural substrate, was the highest followed by pI-fMet and pG-Met. No evidence could be obtained for transfer of M e t from pU-fMet or from pC-Met. An interesting result was obtained b y testing 2’(3‘)formylmethionyl-8-bromoadenosine 5’-phosphate in this assay. Despite the fact that the adenine ring of this analog is preferentially in a syn conformation (231), it possesses donor activity and the formylmethionyl residue is transferred to the aminoacyl-tRNA at the A-site (238). Comparing this finding with the lack of acceptor activity of 2’(3‘)-phenylalanyl-8-bromoadenosineindicates that the A- and
THE
-C-C-A END
OF
63
tRNA
P-sites of the peptidyltransferase center may have different requirements for the conformation of the 3‘-terminal adenosine of tRNA. The P-site of the peptidyltransferase center probably contains an additional binding element that interacts with the peptidyl residue of the peptidyl-tRNA. This was most clearly demonstrated by experiments showing that the interaction of peptidyl-tRNA with the P-site is dependent of the length of the peptide chain, reaching an optimum when four amino-acid residues are attached (239). In addition, the presence of the peptidyl-tRNA in the P-site influences the (EF-Tu)dependent binding of aminoacyl-tRNA to the A-site. This effect is again dependent on the length of the peptide moiety (212). ACKNOWLEDGMENTS We would particularly like to acknowledge the close cooperation with our colleagues at the Abteilung Chemie ofthe Max-Planck-Institut fur experimentelle Medizin in Gottingen as well as with our colleagues outside the institute who made this contribution possible. We thank Dr. D. Gauss and Dr. G. Igloi for their help during the preparation ofthe manuscript and Drs. F. von der Haar, A. A. Krayevsky, 0. Uhlenbeck, and S. M. Hecht for communication of their results prior to publication.
REFERENCES 1 . T. H. Jukes, Curt-. Toji. Microhiol. Zmmunol. 49, 178 (1970). 2. H . G. Zachau,Angew. Chem. 81, 645 (1969);Angeto. Chem., Znt. Ed. Etigl. 8, 711 (1969). 3 . F. Cramer, This Series 11, 391 (1971). 4 . A. L. Haenni, A. Prochianz and P. Yot, in “Energy, Regulation and Biosynthesis in Molecular Biology” (D. Richter, ed.), p. 264. d e Gruyter, Berlin, 1974. 5. D. M. Crothers, T. Seno and D. G. Still, PNAS 69, 2063 (1972). 6 . M. P. Dentscher, This Series 13, 51 (1973). 7. B. G. Barrel1 and B. F. C. Clark, “Handbook of Nucleic Acid Sequences.” Joynson-Bruvvers Ltd., Oxford, 1974. 8. D. R. Kearns, This Series 18, 91 (1976). 9. M. Sprinzl, T. Wagner, S . Lorenz and V. A. Erdmann, Bchem. 15, 3031 (1976). 10. G. J. Quigley, A. H. J. Wang, N. C. Seeman, F. L. Suddath, A. Rich, J. L. Sussman and S. H. Kim, PNAS 72,4866 (1975). 1 1 . J. E. Ladner, A. Jack, J. D. Robertus, R. S. Brown, D. Rhodes, B. F. C. Clark and A. Klug, PNAS 72, 4414 (1975). 12. A. Jack, J. E. Ladner and A. Klug,jMB 108, 619 (1976). 13. A. Rich and U. L. RajBhandary, ARB 45, 805 (1976). 14. S . H. Kim, This Series 17, 182 (1976). 15. P. B. Sigler,Annu. Reu, B i o j h / . s . Bioeng. 4, 477 (1975). 16. G . J. Quigley, N . C. Seenian, A. H.-J. Wang, F. L. Suddath and A. Rich, NARes 2, 2329 (1975). 17. J. E. Ladner, A. Jack, J. D. Robertus, B. S. Brown, D. Rhodes, B. F. C. Clark and A. Klug, NARes 2, 1629 (1975).
64
MATHIAS SPRINZL AND FRIEDRICH CRAMER
18. R. W. Chambers, S. Aoyagi, Y.Furukawa, H. Zawadzka and 0. S. Bhanot, JBC 248, 5549 (1973). 19. K. Chakraburtty, NARes 2, 1793 (1975). 20. T. Seno, FEBS Lett. 51, 325 (1975). 21. J. P. Goddard and L. H. Schulman,JBC 247,3864 (1972). 22. H. Hayatsu, This series 16, 75 (1976). 23. R. P. Singhal, Bchem 13,2924 (1974). 24. Z . Kukan, K. A. Freude, I. Ku&n and R. W. Chambers, Nature NB 232,177 (1971). 25. R. M. Sundari, H. Pelka and L. H. Schulman,JBC 252,3941 (1977). 26. L. H. Schulman and J. P. Goddard,JBC 248, 1341 (1973). 27. F. von der Haar, E. Schlimme, V. A. Erdmann and F. Cramer, Bioorg. Chem. 1,282 (1971). 28. R. Solfert, F. von der Haar, H. Sternbach, M. Sprinzl and F. Cramer, HoppeSeyler’s 2. Physiol. Chem. 356, 1811 (1975). 29. A. R. Cashmore, FEBS Lett. 12, 90 (1970). 30. A. R. Cashmore, D. M. Brown and j. D. Smith,JMB 59,359 (1971). 31. S. E. Chang,JMB 75,533 (1973). 32. D. Rhodes,JMB 94,449 (1975). 33. F. Cramer, F. von der Haar and E. Schlimme, FEBS Lett. 2, 136 and 354 (1968). 34. J. Preiss, P. Berg, E. J. Ofengand, F. H. Bergmann and M. Dieckmann, PNAS 45, 319 (1959). 35. T. I. Jilyaeva and L. L. Kisselev, FEBS Lett. 10, 229 (1970). 36. Y. Kawamura and Y. Mizuno, BBA 227,323 (1972). 37. R. C. Gamble, H. J. P. Shoemaker, E. Jekowski and P. R. Schimmel, Bchem 15, 2791 (1976). 38. M. Sprinzl, unpublished results. 39. M. N. Thang, B. Beltchev and M. Grunberg-Manago, EJB 19, 184 (1971). 40. G. Zubay and M. Takanami, BBRC 15,207 (1964). 41. J. Tal, M. P. Deutscher and U. Z. Littauer, EJB 28,478 (1972). 42. 0. Pongs, R. Bald and E. Reinwald, EJB 32, 117 (1973). 43. S. M. Freier and I. Tinoco, Bchem 14, 3310 (1975). 44. R. Romer and V. Varadi, PNAS 74, 1561 (1977). 45. B. R. Reid and G. T. Robillard, Nature 257, 287 (1975). 46. B. R. Reid, N. S. Ribeiro, G. Gould, G. Robillard, C. W. Hillers and R. G. Shulman, PNAS 72,2049 (1975). 47. W. E. Daniel and M. Cohn, PNAS 72,2582 (1975). 48. L. S. Kan, P. 0. P. Ts’O, F. von der Haar, M. Sprinzl and F. Cramer, BBRC 59,22 (1974). 49. L. S. Kan, P. 0. P. Ts’O, M. Sprinzl, F. von der Haar and F. Cramer, Bchem. 16, 3143 (1977). 50. E. Schlimme, F. von der Haar, F. Eckstein and F. Cramer, EJB 14, 351 (1970). 51. F. von der Haar, personal communications. 52. P. Schofield, B. M. Hoffman and A. Rich, Bchem 9,2525 (1970). 53. B. M. Hoffman, P. Schofield and A. Rich, PNAS 62, 1195 (1969). 54. M. Sprinzl, E. Kramer and D. Stehlik, EJB 49, 595 (1974). 55. M. Sprinzl, K.-H. Scheit and F. Cramer, EJB 34, 306 (1973). 56. M. Caron and H. Dugas, NARes 3, 19 (1976). 57. M. Caron and H. Dugas, NARes 3 , 3 5 (1976). 58. J. Levy, G . Rialdi and R. Biltonen, Bchem 11, 4138 (1972). 59. D. C. Ward, A. Cerami, E. Reich, G. Acs and L. AltwergerJBC 244,3243 (1969).
THE
-C-C-A END
OF
tRNA
65
60. D. C. Ward, E. Reich and L. Stryer,JBC 244, 1228 (1969). 61. A. Maelicke, M. Sprinzl, F. von der Haar, T. A. Khwaja and F. Cramer, EJB 43,617 (1974). 62. S. M. Coutts, D. Riesner, R. Romer, C. R. Radl and G. Maass, Biophys. Chem. 3, 275 (1975). 63. B. Beltchev, M. Yaneva and D. Staynov, EJB 64,507 (1976). 64. C. Kitzinger and T. Benzinger, Z . Naturforsch. B 10, 375 (1955). 65. H. G. Zachau, Chern. Ber. 93, 1822 (1960). 66. B. E. Griffin, M. Jarman, C. B. Reese, J. E. Sulston and D. R. Trentham, Bchem 5, 3638 (1966). 67. H. G. Zachau and W. Karau, Chem. Ber. 93, 1830 (1960). 68. C. S. McLaughlin and V. M. Ingram, Bchem 4, 1442 (1965). 69. F.Schuber and M. Pinck, Biochimie 56,383 (1974). 70. F. Schuber and M. Pinck, Biochimie 56,391 (1974). 71. F. Schuber and M. Pinck, Biochimie 56, 397 (1974). 72. C. S. McLaughlin and V. M. Ingram, Bchem 4, 1448 (1965). 73. R. Wolfenden, D. H. Rammler and F. Lipmann, Bchem 3,329 (1964). 74. H. Feldmann and H. G . Zachau, BBRC 15, 13 (1964). 75. J. Sonnenbichler, H. Feldmann and H. G. Zachau, Hoppe Seyler’s Z . Physiol. Chern. 341,249 (1965). 76. M. Sundaralingam and S. K. Arora,JMB 71, 49 (1972). 77. M. Sundaralingam and S. K. Arora, PNAS 64,1021 (1969). 78. N. Yathindra and M. Sundaralingam, BBA 308, 17 (1973). 79. S. T. Rao and M. Sundaralingam,JACS 92, 4963 (1970). 80. A. M. Bobst and S. Chlkdek, NARes 3, 63 (1976). 81. F. Cramer, V. A. Erdmann, F. von der Haar and E. Schlimme,]. Cell. Physiol. 74, Suppl. 1, 163 (1969). 82. J. Ninio, V. Luzzati and M. YanivJMB 71, 217 (1972). 83. M. Cohn, A. Danchin and M. Grunberg-Manago,JMB 39, 199 (1969). 84. Y. P. Wong, B. R. Reid and D. R. Kearns, PNAS 70,2193 (1973). 85. G. J. Thomas, M. C. Chen, R. C. Lord, P. S. Kotsiopoulos, T. R. Tritton and S. C. Mohr, BBRC 54, 570 (1973). 86. M. Caron, N. Brisson and H. Dugas,JBC 251, 1529 (1976). 87. L. Beres and J. Lucas-Lenard, Bchem 12,3998 (1973). 88. P. S. Sarin and P. C. Zamecnik, BBRC 20, 400 (1965). 89. H. Hashizume and K. Imahori,J. Biochem. (Tokyo) 61, 738 (1967). 90. A. J. Adler and G. D. Fasman, BBA 204, 183 (1970). 91. A. Bernardi and G. L. Cantoni,JBC 244, 1468 (1969). 92. E. Wickstrom, BBRC 43,976 (1971). 93. K. Watanabe and K. Imahori, BBRC 45,488 (1971). 94. G. Melcher, D. Paulin and W. Guschlbauer, Biochimie 53,43 (1971). 95. R. Potts, M. J. Fournier and N. C . Ford, Nature 268, 563 (1977). 96. 0. Pongs, P. Wrede, V, A. Erdmann and M. Sprinzl, BBRC 71, 1025 (1976). 97. A. Danchin and M. Grunberg-Manago, FEBS Lett. 9, 327 (1970). 98. D. J. Dvorak, C. Kidson and R. Chin, JBC 251, 6730 (1976). 99. R. Chin and C. Kidson, PNAS 68, 2448 (1971). 100. T. R. Tritton and S. C. Mohr, Bchern 12, 905 (1973). 101. H. Kaji and Y. Tanaka, BBA 138,642 (1967). 102. S. K. Chatterjee and H. Kaji, BBA 224, 88 (1970). 103. R. R. Gantt, S. W. Englander and M. V. Simpson, Bchem 8,475 (1969).
66
MATHIAS SPRINZL AND FRIEDRICH CRAMER
104. J. J. Englander, N. R. Kallenbach and S. W. Englander,JMB 63, 153 (1972). R. Stern, L. E. Zutra and U. Z. Littauer, Bchem 8,313 (1969). U. J. Hanggi and H. G . Zachau, EJB 18,496 (1971). 0. Uhlenbeck, personal communication (1977). L. H. Schulman, H. Pelka and R. M. Sundari,JBC 249, 7102 (1974). V. V. Vlasov, N. I. Grineva and G. G . Karpova, Mol. Biol. 8, 752 (1974). H. Kasai, Z . Ohashi, F. Harada, S. Nishimura, N. J. Oppenheimer, P. F. Crain, J. G. Liehr, D. L. von Minden and J . A. McCloskey, Bchem 14,4198 (1975). 1 1 1 . T. Ohgi, T . Goto, H. Kasai and S. Nishimura, Tetrahedron Lett. p. 367 (1976). 112. M. Sprinzl, H. Sternbach, F. von der Haar and F. Cramer,EJB 81,579 (1977). 113. J. K. Mackey and P. T. Gilham, Nature 233,551 (1971). 114. G. N. Bennett, J. K. Mackey, J. L. Wiebers and P. T. Gilham, Bchem 12, 3956 (1973). 115. J. J. Sninsky, G. N. Bennett and P. T. Gilham, NARes 1, 1665 (1974). 116. M. Sprinzl, F. von der Haar, E. Schlimme, H. Sternbach and F. Cramer,EJB 25, 262 (1972). 117. J. X. Khym and M. Uziel, Bchem 7, 422 (1968), and reference therein. 118. M. Sprinzl and F. Cramer, PNAS 72, 3049 (1975). 119. M. Sprinzl, D.-I. Wolfrum and V. Neuhoff, FEBS Lett. 50, 54 (1975). 120. M. Sprinzl, K. H. Scheit, H. Sternbach, F. von der Haar and F. Cramer, BBRC 51, 881 (1973). 121. A. N. Best and G . D. Novelli,ABB 142,539 (1971). 125. T. H. Fraser and A. Rich, PNAS 70, 2671 (1973). 126. H. G . Zachau and H. S. Hertz, EJB 44,289 (1974). 127. J. A. Secrist, J. R. Barrio, N. J. Leonard and G. Weber, Bchem 11,3499 (1972). 128. M. P. Deutscher,JBC 247,459 (1972). 129. S. C. Uretsky, G. Acs, E. Reich, M. Mori and L. Altwerger,JBC 243, 306 (1968). 130. B. Brdar and E. Reich,JBC 247, 725 (1972). 131. F. von der Haar and E. Gaertner, PNAS 72, 1378 (1975). 132. D. S. Cane, S. Litvak and F. Chapeville, BBA 224, 371 (1970). 133. J. Preiss, M. Dieckmann and P. Berg,JBC 236, 1748 (1961). 134. A. C. Chinault, J. W. Kozarich, S. M. Hecht, F. J. Schmidt and R. M. Bock, Bchem 16,756 (1977). 135. M. P. Deutscher,JBC 247, 450 (1972). 136. S. M. Hecht and A. C. Chinault, PNAS 73,405 (1976). 137. S. M. Hecht, Acc. Chem. Res. 10,239 (1977). 138. T. H. Fraser and A. Rich, PNAS 72, 3044 (1974). 139. M. Sprinzl, M. Kucharzewski, J. Hobbs and F. Cramer, EJB 78,55 (1977). 105. 106. 107. 108. 109. 110.
140. M. Sprinzl, G . Chinali, A. Parmeggiani, K.-H. Scheit, A. Maelicke, H. Sternbach, F. von der Haar and F. Cramer, in “Structure and Conformation of Nucleic Acid and Protein-Nucleic Acid Interactions” (M. Sundaralingam and S. T. Rao, eds.), p. 293. Univ. Park Press, Baltimore, Maryland, 1975. 141. S . M. Hecht, J. W. Kozarich and F. J . Schmidt, PNAS 71, 4317 (1974). 142. F. von der Haar, E. Schlimme, M. GBmez-Guillen and F. Cramer, EJB 24, 296 (1971). 143. F. Eckstein, F. von der Haar and H. Sternbach, Bchem 16, 3429 (1977). 144. F. Eckstein, V. W. Armstrong and H. Sternbach, PNAS 73, 2987 (1976). 145. H. S. Hertz and H. G . Zachau, EJB 37,203 (1973). 146. D. Carrk and F. Chapeville, Biochimie 56, 1451 (1974).
THE
-C-C-A END
OF
tHNA
67
147. V. Daniel and U. 2. Littauer,JBC 238, 2102 (1963). 148. F. Cramer, R. Sprinzl, N. Furgac, W. Freist, W. Saenger, P. C. Manor, M. Sprinzl and H. Sternbach, BBA 349, 351 (1974). 149. R. L. Soffer, S. Uretsky, L. Alhverger and G . Acs, BBRC 24, 376 (1966). 150. M. Pasek, M. P. Venkatappa and P. B. Sigler, Bchetn 12, 4834 (1973). 151. H. Sternbach, M. Sprinzl, J. B. Hobbs and F. Cramer, EJB 67, 215 (1976). 152. S. M. Hecht, S. D. Hawrelak, J. W. Kozarich, F. J. Schmidt and R. M. Bock, BBRC 52, 1341 (1973). 153. T. F. McCutchan, P. T. Gilham and D. Siill, NARes 2, 853 (1975). 154. T. McCutchan, R. Wetzel and D. Siill, F P 35, 1735 (1976). 155. R. B. Loftfield, This Series 12, 87 (1972). 1S6. D. Sol1 and P. R. Schimmel, in “The Enzymes” (P. D. Boyer, ed.), 3rd ed., Vol. 10, p. 489. Academic Press, New York, 1974. 157. L. L. Kisselev and 0. 0. Favorova, Adu. Enz!jrnot. 40, 141 (1974). 158. K. L. Roy and G. M. Tener, Bchem 6, 2847 (1967). 159. H. Faulhammer, Ph.D. Thesis, Technical University of Braunschweig (1977). 160. V. Daniel and U. 2. Littauer,JMB 11, 692 (1965). 161. B. Rether, J. Gangloff and J.-P. Ebel, EJB 50, 289 (1974). 162. A. H. Kirschenbaum and M. P. Deutscher, BRAC 70, 258 (1976). 163. H. Sternbach and M. Sprinzl, unpublished. 164. J. Ofengand, S. ChlBdek, G. Robillard and J. Bierbaum, Bchem 13, 5425 (1974). 165. M . Uziel and K. B. Jacobson, BBA 366, 182 (1974). 166. C. Li and J. Su, BBRC 28, 1068 (1967). 167. M. Sprinzl, G. Siboska and S. A. Pedersen, NARes 5, 861 (1978). 168. L. H. Schulman and H. Pelka, Bchem 16, 4256 (1977). 169. M. Sprinzl and F. Cramer, Nature NB 245, 3 (1973). 170. G. Gauss, D. Riesner and G . Maass, Hoppe Seyler’s Z. Physiol. Chem. 358, 265 (1977). 171. M. Sprinzl and D. Richter,EJB 71, 171 (1976). 172. E. Baksht, N. de Groot, M. Sprinzl and F. Cramer, Bchem 15, 3639 (1976). 173. F. Cramer, H. Faulhammer, F. von der Haar, M. Sprinzl and H. Sternbach, FEBS Lett. 56, 212 (1975). 174. P. C. Zamecnik, BJ 85, 257 (1962). 175. Y.Ishido, N. Nakazaki and N . Sakairi, Chem. Comnaun. p. 832 (1976). 176. B. Alfert and S. M. Hecht, personal communication (1977). 177. A. C. Chinault, K. H. Tan, S. M. Hassur and S. M. Hecht, Bchein 16, 766 (1977). 178. J. Ofengand and C.-M. Chen,JBC 247, 2049 (1972). 179. 2. Hussain and J. Ofengand, BBRC 50, 1143 (1973). 180. F. Chapeville, F. Lipmann, G . von Ehrenstein, B. Weisblum, W. J. Ray, Jr. and S. Benzer, PNAS 48, 1086 (1962). 181. F. H. Bergmann, P. Berg and M. Dieckmann,JBC 236, 1735 (1961). 182. R. B. Loftfield and E. A. Eigner,JBC 240, 1482 (1965). 183. F. von der Haar and F. Cramer, FEBS Lett. 56,215 (1975). 184. F. von der Haar and F. Cramer, Bchem 15,4131 (1976). 185. G . L. Igloi, F. von der Haar and F. Cramer, Bchern 16, 1696 (1977). 186. A. R. Fersht and M. M. Kaethner, Bchem 15,3342 (1976). 187. J. J. Hopfield, PNAS 71,4135 (1974). 188. R. B. Loftfield and D. Vanderjagt, BJ 128, 1353 (1972).
68
MATHIAS SPRINZL AND FRIEDRICH CRAMER
F. von der Haar, FEBS Lett. 79,225 (1977). H.-J. Hinz, K. Weber, J. FLossdorfand M.-R. Kula, EJB 71, 437 (1976). P. Leder,Adu. Protein Chem. 27,213 (1973). D. L. Miller and H. Weissbach, in “Molecular Mechanisms of Protein Biosynthesis” (H. Weissbach and S. Pestka, eds.), p. 323. Academic Press, New York, 1977. 193. J. Gordon, PNAS 58, 1574 (1967). 194. A. H. Lockwood, S. Hnttman, J. S. Dubnoff and U. Maitra, JBC 246,2936 (1971). 195. J. M. Ravel, R. L. Shorey and W. Shive, BBRC 29, 68 (1967). 196. Y. Ono, A. Skoultchi, A. Klein and P. Lengyel, Nature 220, 1304 (1968). 197. S. Fahnestock, H. Weissbach and A. Rich, BBA 269,62 (1972). 198. M. N. Thang, L. Dondon, D. C. Thang and B. Rether, FEBS Lett. 26,145 (1972). 199. E. Baksht, A. Gal, N. de Groot, A. A. Hochberg, M. Sprinzl and F. Cramer, NARes 4, 2205 (1977). 200. E. Baksht, N. de Groot, M. Sprinzl and F.Cramer, FEBS Lett. 55, 105 (1975). 201. P. Greenwell, R. J. Harris and R. H. Symons, EJB 49,539 (1974). 202. S. Chlldek, D. Ringer and E. M. Abraham, NARes 3, 1215 (1976). 203. C.-M. Chen and J. Ofengand, BBRC 41, 190 (1970). 204. G. Chinali, M. Sprinzl, A. Parmeggiani and F. Cramer, Bchem 13,3001 (1974). 205. D. Ringer and S. Chlldek, PNAS 72, 2950 (1975). 206. J. Ofengand, in “Methods in Enzymology” Vol. 29, p. 661. Academic Press, New York, 1974. 207. M. Kawakami, S. Tanada and S. Takemura, FEBS Lett. 51, 321 (1975). 208. S. M. Hecht, K. H. Tan, A. C. Chinault and P. Arcari, PNAS 74, 437 (1977). 209. A. V. Furano, PNAS 72,4780 (1975). 210. D. Ringer, S . Chlldek and J. Ofengand, Bchem 15,2759 (1976). 211. P. Lengyel and D. Sol], Bacteriol. Reu. 33,264 (1969). 212. E. Bakhst and N. de Groot, Mol. B i d . Rep. 1,493 (1974). 213. C. G. Kurland,ARB 46, 173 (1977). 214. R. L. Shorey, J. M. Ravel and W. Shive,ABB 146, 110 (1971). 215. J. Lucas-Lenard, P. Tao and A.-L. Haenni, CSHSQB 34,455 (1969). 216. T. Girbes, D. Vazquez and J. Modolell, EJB 67, 257 (1976). 217. H. Yokosawa, N. Inoue-Yokosawa, K.-I. Arai, M. Kawakita and Y. Kaziro, JBC 248, 375 (1973). 218. A. Skoultchi, Y. Ono, J. Waterson and P. Lengyel, CSHSQB 34, 437 (1969). 219. D. Ringer, K. Quiggle and S. Chlidek, Bchem 14,514 (1975). 220. R. J. Suhadolnik, “Nucleoside Antibiotics,” p. 3. Wiley (Interscience),New York, 1970. 221. S . Chlldek, D. Ringer and K. Quiggle, Bchem 13, 2727 (1974). 222. D. Nathans and A. Neidle, Nature 197, 1076 (1963). 223. A. A. Krayevsky, M. K. Kukhanova and B. P. Gottikh, NARes 2,2223 (1975). 224. C. G. Kurland, R. Rigler, M. Ehrenberg and C. Blomberg,PNAS 72,4248 (1975). 225. H. Neumann, V. E. Shashoua, J. C. Sheehan and A. Rich, PNAS 61, 1207 (1968). 226. S. Pestka, in “Molecular Mechanisms of Protein Biosynthesis” (H. Weissbach and S. Pestka, eds.), p. 467. Academic Press, New York, 1977. 227. R. J. Harris and R. H. Symons, Bioorg. Chem. 2, 286 (1973). 228. A. V. Azhayev, S. V. Popovkina, N. B. Tarussova, M. P. Kirpichnikov, V. L. Florentiev, A. A. Krayevsky, M. K. Kukhanova and B. P. Gottikh, NARes 4,2223 (1977). 229. I. Rychlik, S. Chlidek and J. iernlicka, BBA 138, 640 (1967). 230. J. BemliEka, S . Chlldek, D. Ringer and K. Quiggle, Bchem. 14, 5239 (1975). 189. 190. 191. 192.
THE
-C-C-A END OF tRNA
69
231. S . S. Tavale and H. M. Sobel1,JMB 48, 109 (1970). 232. 0.Pongs, K. H. Nierhaus, V. A. Erdmann and H. G. Wittmann,FEBS Lett. 40,S28
(1974). R. E. Monro, J. c e r n i and K. A. Marcker, PNAS 61, 1042 (1968). J . Cerna, I. Rychlik, A. A. Krayevsky and B. P. Gottikh,FEBS Lett. 37,188(1973). J. c e r n i , FEBS Lett. 58, 94 (1975). A. A. Krayevsky, L. S. Victorova, V. V. Kotusov, M. K. Kukhanova, A. D. Treboganov, N. B. Tarussova and B. P. Gottikh, FEBS Lett. 62, 101 (1976). 237. J. cerna, I. Rychlik, A. A. Krayevsky and B. P. Cottikh, Acta B i d . Med. Ger. 33, 233. 234. 235. 236.
877 (1974). 238. A. A. Krayevsky, personal communication (1977). 239. N. de Groot, A. Panet and Y. Lapidot, EJB 15, 215 (1970).
This Page Intentionally Left Blank
The Mechanism of Action of Antitumor Platinum Compounds
J. J. ROBERTS
1
Znstitute of Cancer Research: Royal Cancer Hospital Pollurds Wood Research Station Chalfont, S t . Giles, Bucks., United Kingdom
I
A. J. THOMSON
1
SchooI of Cheniicul Sciences Uiiioersity of East Anglio Normicli, Norfolk, United Kingdom
1 1
1
I. Introduction.. ................................................... 11. Chemical Features of Platinum Drugs ............................ 111. Biological Effects of Platinum Coordination Complexes Indicative of Reactions with DNA .......................................... A. Filanient Fomiation in Bacteria ............................... B. Induction of Lysogeny ........................................ C. Mutagenic Propelties of Neutral Platinum Complexes .......... D. Reaction with Viruses and Transforming DNA ................. IV. Biochemical Effects of Plntinuni Complexes Indicative of Reactions with DNA ............................................. A. Inhibition of DNA Synthesis: the Primwry Biochemical Lesion Induced b y Platinum Coniplexes ............................. B. Enzyme Studies ............................................. C. Other Proposed Mechanisms of Actions ........................ V. Interaction of Platinum Compounds with DNA .................... A. Zti Vioo Studies Indicative of DNA iis the Target Molecule ...... B. I t 1 Vitro Reactions of Neutral Platinuni Complexes with Nucleic Acid Components ............................................ VI. Repair of DNA Damage Induced by Platinum Complexes in Vioo . . A. General Coninieiits ........................................... B. Excision Repair .............................................. C. Postreplication Repair (Replication Repair) .................... VII. Concluding Remarks ............................................. References ......................................................
71 75 85 85 86 87 87 88 88 89 93 94 94 104 110 110 110 116 128 129
1. Introduction In 1969, Roseliberg et u2. showed that the platinum coordiiiat'1011 compounds cis-diamminedichloroplatiiiu~n(1I) and cis-diamminetetrachloroplatinuin(1V) have potent antitumor activity against Sar71 Progress in Niicleic Acid Research and Molecular Biology, Vol. 22
Copylight @ 1979 by Academic Press. Inc. All tights of reproduction in any form reserved. ISBN 012-540022-5
72
J . J. ROBERTS AND A. J. THOMSON
CI
3N
(=)
01.
$I
(d)
"f
L71:/' H2
CI
CI
FIG, 1. Structures of the platinum complexes originally described by Rosenberg et (I). (a) cis-diaiiiininedichlo~opl~~tiii~ii~~ (II), c ~ . ~ - [ P ~ ( I I ) C I , ( N H ~ ) ~ ~ . (b) cis-dichloro(ethylenediamine)platinum(II),cis- [Pt(II)Cl,(en)]. (c) cis-diamminetetrachloroplatinum(Iv),cis- [Pt(IV)Cl,(NHJ,]. (d) cis- tetrachloroplatinuin(IV)ethylenediamine, cis- [Pt(IV)Cl,(en)].
coma 180 and Leukemia L1210, whereas the trcins geometrical isomers of these compounds are ineffective ( I , 2). It was subsequently shown that these coinpouiids are also effective against a virus-induced reticulum cell sarcoma in mice ( 3 ) , the Dunning ascitic leukemia, Walker 256 carcinosarcoma ( 4 ) , and a dimethylbeiizanthraceneinduced maminary carcinoma ( 5 ) in rats. These compounds, whose structures are given in Fig. 1, introduced an entirely new elass of antitumor compounds. Since their discovery, numerous laboratories have undertaken extensive programs to synthesize a wide range of analogous metal complexes and to investigate the underlying mechanism of action. Following the clear demonstration of activity against a broad spectrum of animal tumors, clinical trials were initiated on cis-Pt(11)' in many centers. A review of the clinical status of cis-Pt(I1) in cancer chemotherapy (6) of some early phase-I clinical trials, generally on extremely advanced disease, indicated promising anticancer activity. The tumors most sensitive to cis-Pt(I1) have so far proved to be testicular carcinoma (7),head and neck cancer (8),squamous cell carcinoma, malignant lymphoma, and endometrial carcinoma (9, l o ) ,and ovarian adenocarconoma ( 1 1 ) . cks-Pt(II) is an abbreviation of cis-[PtCl,(NH,),], platinurn(I1).Similarly for trans-Pt(I1).
cis-diamminedichloro-
ANTITUMOR PLATINUM COMPOUNDS
73
These early reports stressed the limitations imposed by severe toxicity to the kidney and bone marrow. Nausea and vomiting and sometimes audiotoxicity were also severe. However, subsequent studies showed that the renal toxicity of cis-Pt(11) could be dramatically decreased without inhibiting the therapeutic effects of the drug by the induction of diuresis with mannitol (12). Moreover, with further extended clinical trials of cis-Pt(II), there has been confirmation of the early promise against ovarian carcinoma, either when used alone and not necessarily in high-dose therapy (13, 14), or in combination with adrianiycin (now doxorubicin) and chlorambucil (24). Some instances of a complete response against advanced ovarian carcinoma (141, proved b y "second-look" surgery, have been produced b y platinum alone and with the platinum-adriamycin combination. The initially observed extreme sensitivity of testicular tumors to cis-Pt(I1) has also been confirmed (15).When cis-Pt(I1) was used in combination with vinblastine and bleomycin,2 complete and overall response rates of 75% and loo%, respectively, were obtained against this particular tumor (16).A summary of the cumulative response rate by tumor type (Table I), compiled by Rozencweig et a1 (17) for the Cancer Chemotherapy Evaluation program, notes the lack of effect of cisPt(I1) on all types of leukemias and colorectal tumors. It also emphasizes the need for further studies of the toxic effect ofcis-Pt(I1) and for a continuing evaluation of its possible effect on tumors, such as those of head and neck, prostrate, and lung, where suggestions of activity had been noted earlier. It is now clear that cis-Pt(I1) has established a place in the clinic and its use will undoubtedly grow. An enhanced therapeutic effect of cis-Pt(I1) against early and advanced L1210 is found when cis-Pt(I1) is used in combination with either cyclophosphamide (18, 19) or 1,2bis(3,s-dioxo-l-piperaziny1)propane (ICRF 159)(18).Similarly, combination therapy with methotrexate and also with adriamycin (doxorubicin) leads to greater than additive effects against L1210 leukemia in DBA mice and against S180 sarcoma, respectively. An encouraging and interesting new development has come from the discovery (20) that cis-Pt(1I) could act as a radiation sensitizer in the Bacillus megatherium bacterial spore, preferentially, but not exclusively, in the absence of oxygen and presumably by the sequestration of the hydrated electron, which would prevent its reaction with a hydroxy radical. This finding prompted studies on the effects of combined radiation and platinum treatment on mammalian tumors. Although no See article on bleomycin in Vol. 20 of this series.
74
J. J. ROBERTS AND A. J. THOMSON
TABLE I CUMULATIVE RESPONSE RATE BY TUMOR TYPECOMPILED BY CANCER THERAPY EVALUATION PROGRAM, NATIONALCANCERINSTITUTE" Cumulative results with cis- [Pt(II)Cl,(NH,)zI
No. of responding patients
Response rate
Tumor type
No. of evaluable patients
Testicle Acute leukemia AML ALL Colorectum Ovary Head and neck Bladder Lungb Sarcomac Lymphomab Neuroblas toma Melanoma Prostate Endometrium Hepatoma Thyroidd
89 66 2 24 39 38 31 28 27 21 14 14 8 6 2 2 1
54 1 0 0 0 11 5 12 2 4 3 3 1 2 2 1 1
61
(%)
2
-
-
0 29 16 43 7
a No response was obtained in cumulative series offewer than 14 patients with the following tumor types: brain, breast, kidney, mesothelioma, myeloma, pancreas, stomach. Data compiled by Rosencweig et uE. (17). * Insufficient information available regarding activity of cis-[Pt(II)Cl,(NH&] by histologic subtypes. Includes reticulum cell sarcoma. Additional responses were reported, but without mention of the total number of evaluable patients.
effect was observed in the transplantable mouse mammary adenocarcinoma, a pronounced synergistic effect of cis-Pt(11) in combination with local X-irradiation was seen in the rat brain tumor system (21). Dose- and cell-cycle-dependent enhancement of the effects of radiation on Chinese hamster ovary cells in culture by either of the analog compounds cis-dichlorobis(cyclopentylamine)platinum(II) (22) or cis - dichloro -trans - dihydroxo -cis - bis(isopropylamine)platinum(IV) (23) is seen, but it is not ascribed to the action of platinum drugs as radiation sensitizers. It is this encouraging and rapid progress in the clinical develop-
ANTITUMOR PLATINUM COMPOUNDS
75
ment of platinum drugs that has been the main impetus behind studies of the mechanism of action. The site of the primary lesion in cells that results in toxicity and hence tumor destruction is now generally agreed to be DNA in preference to other macromolecules, such a s RNA or protein. Modification of the DNA template results in the selective inhibition of DNA synthesis and, hence, unbalanced growth. T h e primaiy biochemical effect in treated cells thus resembles that induced by a number of other cytotoxic and antitumor agents. Therefore, the extent to which such drugs elicit selective effects on cells may be a reflection of differences in their ability to handle damage to the DNA template. Much effort has therefore been applied to elucidating the nature of the chemical interaction of platinum compounds with DNA, and of mechanisms for circumventing such damage during DNA replication, and of mechanisms for eliminating the damage by specific DNA excision-repair processes.*"Variations inthe levels of such repair processes among cell populations could therefore provide a partial explanation for the evolution of drug-resistant cells during the chemotherapeutic treatment of tumors. Before reviewing the detailed evidence that supports these views, a short description is given of the significant chemical features of the platinum drugs and of the information obtained from structure analog studies. The latter help to define the chemically important features of the molecular structure and in this way are important in suggesting mechanisms of action.
II. Chemical Features of Platinum Drugs The only well-characterized oxidation states of platinum in aqueous solution are 2+ and 4+,and both states dictate a strict stereochemical disposition of the groups or ligands that surround the metal ion. In the case of Pt(II), only a square planar geometry is found, whereas in Pt(1V) only octahedral coordination of six ligands is possible. Interconversion between the two oxidation states is facile, reduction from 4+ to 2+ usually occurring at a potential of about zero volts on the hydrogen scale, although the exact redox potential clearly depends upon the nature of the ligands about the platinum ion. The chemistry of both oxidation states is dominated by the ability of the compounds to undergo ligand substitution or exchange reactions. Since the reactions of square planar Pt(I1) compounds are much the more thoroughly investigated, we concentrate discussion on them for the purposes of illustration. z0
See article by Lindahl in this volume. [Ed.]
76
J. J. ROBERTS AND A. J. THOMSON
Generally, both oxidation states of platinum form their most stable complexes with the more polarizable atoms. The following order of thermodynamic stability usually holds: S > I > Br > C1 > N - O > F
If the incoming ligand can bind to the metal ion at more than one site, the thermodynamic stability is greatly enhanced. This is commonly referred to as the “chelate effect.” The kinetic stability of ligands bound to platinum(‘I1)varies over a very wide range. Furthermore, the rate of substitution of a given ligand is strongly influenced by the nature of the ligand opposite it in the square planar array. This is called the trans effect. Generally, the more strongly bound the ligand, the more effective it will be at stabilizing ligands trans to it. This effect can be very large and, indeed, is the whole basis for the preparation, in very high yield, of either the cis or the trans isomer of a square planar platinum(I1) compound. A consequence of these facts is that cis-Pt(I1)’ is a bifunctional reagent with both chloride ligands open to substitution by incoming groups that are either in large excess or form thermodynamically more stable links to platinum. The ammine groups are both kinetically and thermodynamically inert to substitution. Thus the remaining two position in the square plane are blocked to substitution by incoming groups. However, if the chloride ions were replaced by ligands with a very strong trans-directing influence, it is possible that the ammines would become labilized and replaced. Similarly, in the trans isomer, the chloride ions are readily substituted, whereas the ammine groups are difficult to replace. Hence both compounds are bifunctional reagents, but with very different requirements for the stereochemical disposition of two incoming groups (24, 25). It is a most important feature of the chemistry of square planar platinum(I1) compounds that substitution reactions proceed with retention of configuration. Isomerizations are very rare although light can catalyze the process. During extensive structure-activity studies, no examples of activity in the trans has been discovered. In aqueous solution, both chloride ions are slowly lost from the coordination sphere of the Pt(I1) ion, and water or hydroxide ion becomes bound. Thus a distribution of species is set up involving the presence of unhydrol yzed and of partially and fully hydrolyzed species. However, this equilibrium is labile so that, if the chloride ion concentration is raised to that of, say, isotonic saline, the majority species in solution will be unhydrolyzedcis-Pt(I1)(26).Recently it has
ANTITUMOR PLATINUM COMPOUNDS
77
been shown that, under certain conditions of pH and metal ion concentration, hydroxy-bridged polymeric species can be formed. Both dimeric and trimeric species have been isolated from solution, and their structures have been determined by X-ray crystallography (27) (Fig. 2). It is not clear whether such species could be formed in the high chloride levels of plasma, although it seems unlikely at this stage. Extensive series of analogs have now been screened against experimental tumor cells both in animals and in culture (26, 28, 29). The results lead to interesting conclusions. The natureofthe amine group(am),in Pt(am),Cl,, has aprimary influence on the antitumor activity or, more specifically, on the selectivity, by varying the toxicity (LD5,,)of the compounds. The activities of many such compounds against the S-180 and ADJ/PC6 plasma cell tumors in mice have been reported ( 3 0 3 2 ) . For the straight-chain amines (Table 11), it is to be noted that potency increases to a maximum with the n-butylamine derivative, while toxicity decreases markedly. Primary amines with an alicyclic substituent (Table 111)
FIG.2. (a) The structure of the centrosyinmetric ion [{Pt(NHJ2(0H)}J2+found in the nitrate salt. (b)The structure of the cation [{Pt(NH3),(OH)},IS+found in the nitrate salt (27).
TABLE I1 EFFECTOF VARIATIONS IN THE LENGTHOF ALKYL CHAIN IN [Pt(II)(am),C1,] COMPOUNDS AGAINST THE ADJlPC6 MOUSE PLASMACYTOMA
NHx CH3NH2 n-GH5NHZ n-C3H7NH2 n-C, HgN H2 n-C5H11NH2 II-C~H~~NH~ n-C,HISNH, n-C8HlrNHz
1.6 12 12 12 10 37 1500 900 200
13.0 18.5 26.5 26.5 110 92 1000 900 200
8.1
-
2.2 2.2 11 2.5
TI (therapeutic index) = LD5dIDw.
ANTITUMOR
TABLE 111 ACTIVITY AGAINST THE ADJIPC6 MOUSE PLASMACYTOMA AND AQUEOUS AND LIPID SOLUBILITY OF CiS-Pt(am)&1,"
m,
13.0
1.6
57
2.3
90
480
0-
>3200
NH,
1000
NH,
0-
2650
8.9
<0.008
25
610
1.6
C0.007
2.9
31
87
0.21
0.05
2.4
200
5.6
0.013
0.86
>267
1.9
0.0040
4.3
0.71
0.0014
8.2
0.29
0.00056
16.7
12
7.7
230
8.1
130
2.9
TI (therapeutic index) = LDJIDgo. From Braddock et al. (31).
79
ANTITUMOR PLATINUM COMPOUNDS
yielded two compounds, the cyclopentylamine and cyclohexylamine complexes, with very high therapeutic indexes against the ADJ/PC6 tumor. However, the cyclohexylamine complex was inactive against all other tumors, and the cyclopentylamine, while active against the Walker tumor and L1210 leukemia, was found to be too insoluble for intravenous administration and therefore not particularly suitable for clinical use. Marked variation within a series of cyclic secondary amines was noted, the three-membered ring (aziridine) complex being the most active. Unfortunately again, the compound was inactive against other animal tumors. Considerable variation in activity also exists within the branched-chain series of primary amines (Table IV). The isopropylainine complex, in addition to being active against the PC6
TABLE IV
EFFECTOF BRANCHED-CHAIN SUBSTITUENTS AND THE DISTANCE OF THE BRANCHOR RING FROM THE -NH, GROUPON THE TOXICITY, A N T ~ U M OACTIVITY R (ADJ/PC6 PLASMACYTOMA), AND AQUEOUS SOLUBILITY OF COMPLEXES OF THE
TYPEcis- (Pt(am),C1,)" Aqueous solubility
ID,,
LDm (mg/kd
(mdkd
TI
Pdml
33.5
0.9
37
84
730
27.5
26.5
83
6.2
13.4
i -
89
34
1150
132
2.6
5.8
132
~
~~
mM 0.22
0.41
0.00087
50
0.12
0.27
0.00056
198
0.56
0.0012
-
0.07
0.00012 ~~
" am is a primary amine with an alicyclic substituent (cf. Table 11). From Braddock et ul. (31).TI (therapeutic index)
= LD,dIDw.
80
J. J. ROBERTS AND A. J. THOMSON
tumor, also had marked activity against the Walker tumor and L1210 leukemia and is likely to be one of the “second generation” clinically useful compounds. However, the introduction of a second branchedchain seems to remove the antitumor activity completely. Aromatic heterocyclic and alicyclic amine complexes have shown little or no activity at the levels at which they have been tested. Again the insertion of a small substituent, like a methyl group, into the cyclohexylamine ring can greatly modify both the toxic and tumorinhibitory activity. However, the same substituents inserted into the o-phenylenediamine moiety can have quite opposite effects. It should be noted that the o-phenylenediamine ligand is the only example of an aromatic amine that forms antitumor platinum complexes. The pyridine complex shows some activity against the Ehrlich ascites carcinoma (33).This compound possesses actions similar to the wholly inorganic parent compound, cis-Pt(II), but is somewhat less potent. The methylamine congener, cis-dichlorobis(methy1amine)platinum(II), while also lacking the potency of the parent compound, interacts with cells and molecules in a fashion similar to that of either the pyridine analog or cis-Pt(I1) itself (34). The reasons for these major variations in biological activity with minor modifications to the amine structure are by no means obvious. Attempts have been made, without success, to explain them in terms of the kinetic, steric, or solubility properties of the various platinum complexes (32). It does not seem that kinetic effects alone can explain these differences. Possibly hydrogen-bonding interactions between the amine ligands and natural macromolecules could play a role in stabilizing the receptor drug complex. At one time it appeared (32) that there might be a relationship between the chloroform-water distribution coefficient and the effectiveness of the complexes toward the plasma cell tumor, with an optimum distribution coefficient of 1. The solubilities and distribution coefficients of the ci~-[Pt(am)~C1~1 complexes3 varied as dramatically with minor changes in the amine substituents as did .the toxicities and the potencies, but the variation was not always in parallel. Clearly the solubilities of the complexes are only a partial explanation for their activities. Of all the amines synthesized and tested by Tobe, Connors and their co-workers (30),only the isopropylamine derivative stands out as being more effective than the original ammonia complex prepared by Rosenberg et aE. ( I ) . Unlike so many of the other complexes, it proved to be effective against all three animal tumors investigated; the Walker am = RNH2, whete R is an alicyclic radical.
81
ANTITUMOR PLATINUM COMPOUNDS
tumor, the PC6 plasma cell tumor, and L1210 leukemia. Complexes in which the ligands are not donors of nitrogen show no activity, although only a few have been tested. Alternative donor ligands could be oxygen or strongly labilizing neutral groups, such as sulfur and phosphorus. Investigations of the effect of changing the leaving group X in compounds of the type PtA,X, and PtA'X,, where A is a monodentate and A' a bidentate ligand, reveal two main classes that give active compounds: (a) when A is NH3 and X is a monodentate anionic ligand of intermediate leaving ability-mainly C1- but also Br- (Table V); and (b) when X is a bidentate ligand, such as oxalate and malonate (Table VI).Various model studies indicate the leaving abilities of certain monodentate ligands in particular structures [for discussion, see Cleare (26)]. The screening of analogs of cis-Pt(I1) and ethylenediamine complexes (Table V) reflect this same general leaving order and show that the optimum monodentate leaving groups are chloride and bromide (30).Less labile groups are inactive and relatively nontoxic, but labile groups give extremely toxic compounds. Chloride has the further advantage of being a normal body constituent. Variations in the leaving group can sometimes confer considerable increases in aqueous solubility. As oxalate and malonate are not conTABLE V
VARIATION OF
x I N cis-[Pt(II)(NH3),X2]A N D ACTIVITY AGAINST SARCOMA 180"
X
Solvent
NO,N 0,HZO' ClBr-
Water Saline Water Saline NaBr (0.04 M) Saline Water slurry Saline Slurry
Br-
ISCNNO,
Dose range Dose Toxic level (nigkg) response (mgkg) 6-12 2.5-12 2-20 0.5-20 5-20
2-@ 10-25
-
+ -
+ + + -
5-lo@
-
5-100
-
Dose (mg/kg)
7d
54
11
8
6 10
5d 9 15
1 30
8 14
13 110
5 10-25 20-35 5-100
5-6c >25 50 > 100
-
TIC
70 99
Taken fiom Cleare and Hoeschele (29).T/C = (weight of tumor from test animal/ weight of tumor from control animal) x 100. * Slurry at higher concentrations. Daily injections for 9 days. Highly toxic; convulsions. e Cationic complex ion (2+).
TABLE VI: ACTIVITY OF VARIOUSPLATINUM AMINE COMPLEXES CONTAINING OXALATE AND MALONATELIGANDSAGAINST SARCOMA 180"
Complex
Solvent
Dose range (mgkg)
Dose response
Toxic level (mg/kg) T/C
Dose (mgkg)
W
5-80
+
20-3ob
21
30
W
80-180
+
180
20
120-180
W
10-80
f
20-40
21
10
W
45-90
+
65
9
60
W
20-80
+
90
28
60-80
W
0.25-16
-
3'
75
0.25-2
W
5-80
+
45-60
18
40
W
30-90
4
90
51
90-120
amine)platinum (11) HF-
p-
Malonatobis(methy1amine)platinum(II)
Malonato-l,&propylenediamineplatinum(I1)
Malonato- l,%propylenediamineplatinum( 11)
Oxalatoethylenediamineolatinum(I1)
Malonatoethylenediamineplatinum( 11)
>90
t
Methylmalonatoethylenediamineplatinum(I1)
I
ws
40-120
f
> 120
Ethylmalonatoethylenediamineplatinum( 11) a From Cleare and Hoeschele (29).T/C = (weight of tumor from test animavweight of tumor from control animal) x 100. Only 50% survivors. Highly toxic; convulsions.
ANTITUMOR PLATINUM COMPOUNDS
83
sidered to be good leaving groups, the activity of compounds containing these ligands is surprising. It may be that these ligands are removed inside the cell by some metabolic process. Lack of solubility in water has been a difficulty for the introduction into the clinic of analogs other than cis-Pt(I1). Therefore there have been some recent attempts to improve solubility by nieaiis either of mixed amine complexes of the type cis-[Pt(NH,)A”CI,], where A” is a substituted monodentate amine, or of Pt(1V) complexes of the type cis- [Pt(A”),Cl,I and cis-trans- [Pt(A;’)(OH),Cl,] (32).Although no clear pattern relating structure to solubility emerged, the compound cistrans-[Pt(NH,),(OH),CI,] proved to be only a tenth as toxic as cis[Pt(NH&I,I and to have a higher therapeutic index. Other promising structures appear to be the isopropylamine and cyclopentylamine congeners containing the trans-hydroxy groups. It is possible that Pt(1V) compounds are reduced to Pt(I1) compounds in vivo with the loss of the two axial ligands. If so, the trans-dihydroxy species are, in effect, a novel variant in the method of administering the corresponding Pt(I1) complex. It was inevitable that, during extensive preparation of analogs, some compounds of totally unexpected activity and chemistry would be discovered. One such is (1,2-diaminocyclohexane)sulfatoplatinum (11),which is formulated as in Fig. 3, that is, with the sulfate group acting as a bidentate ligand (35).However, the exact structure of this compound has not yet been confirmed by X-ray data. A bidentate sulfate group attached to a single metal center would be unique in platinum chemistry, although an example of a sulfate group bridging two platinum ions is known (36). Other examples of totally unexpected compounds turning up are the so-called platinum blues and purples. The diaquo hydrolysis product of cis-Pt(I1) and other amines react slowly with thymine, uracil, and related compounds to give water-soluble complexes of either deep blue or purple color. Since this color is most unusual among platinum compounds, attention was directed to the exact chemical nature of these materials (37, 38). Some of the materials were
FIG. 3. (1,2-Diaminocyclohexane)sulfatoplatinum(II),shown to have activity against experimental and human tumors (35).
84
J. J. ROBERTS AND A. J. THOMSON
,
FIG.4. Structure of blue [Ptg(NH3)4(C5H40N)0]2(NOs)5. The metal-metal distances are Pt 1-Pt 2, 2.779 A and Pt 2-Pt 2', 2.885 A. C,H40N is the a-pyridone anion (40).
observed to have activity against sarcoma 180 tumor in Swiss mice superior to that shown by cis-Pt(II), as well as activity against the Rauscher leukemia, Ehrlich ascites, and ADJ/PCGA tumors (37). Investigation of the exact chemical nature of these species has been hampered by inability to crystallize them. Thus it has proved difficultto purify them and to obtain structural characterization. However, it is now clear that both the blue and the purple compounds are polymeric cationic species containing several platinum ions with nonintegral or mixed oxidation states. The blue compounds are paramagnetic, with an average platinum oxidation state slightly above 2 (39), whereas the purple compounds are diamagnetic, with an average platinum oxidation state of about +3.6 (38). Recently, Barton et al. (40) succeeded in crystallizing a blue compound using a-pyridone in place of a thymine or uracil derivative. The structure shows (Fig. 4)a complex ion containing four platinum ions of average oxidation state 2.25. The a-pyridone ligand bridges pairs of platinum ions. The compound gives an electron paramagnetic resonance signal. This derivative appears then to mimic closely the polymeric platinum blue complexes. The uncertainty of the structures of platinum blues and the consequent failure to reproduce some of their biological activities has limited progress in establishing the clinical utility of these initially promising compounds.
85
ANTITUMOR PLATINUM COMPOUNDS
111. Biological Effects of Platinum Coordination Complexes Indicative of Reactions with DNA A. Filament Formation in Bacteria Probably the first observation of an effect of a platinum coordination complex in a biological system, and one that gave a clear indication of its biochemical mode of action, came during experiments on the effect of an electric current on growing bacteria (41).It was noticed that, when a low alternating current was passed through platinum electrodes to growing gram-negative bacteria in nutrient media, cell division was inhibited and the bacteria grew into long filaments. Subsequently it was discovered that some of the platinum dissolves under these conditions to give, first, the ionic species ammonium hexachloroplatinate. This compound can itself, at high concentration, inhibit cell division, but aged solutions were found to be far more efficient in producing filaments, but only if exposed to visible light. The photochemical change that occurred in a solution of hexachloroplatinate giving rise to a more active agent involved the replacement of the chloride ligands by NH3 with the loss of one negative charge per replacement to give finally a stable neutral species. The new species was shown convincingly to inhibit cell division but not growth, in contrast to the parent ionic species, which was a bacteriocide and not a bacteriostat. The difference between the properties of the charged and uncharged platinum species are shown in Table VII (42).Salts of other group-VIIIB metals, such as rhodium and ruthenium, have been TABLE VII SUMMARY OF EFFECTSOF PLATINUM COMPLEXES ON Escherichia coli GROWTHO Type of complex Doubly negative, [PtCIJSingly negative, [PtCIdNHJNeutral, [PtCI,(NH,),Io
Effects on cell growth
Effects on cell division
Inhibitory (bacteriocide) in low concentrations (>1 PPm) No inhibition
No detectable effect
No effect at low concentrations (<4 ppm); slight inhibition at higher concentrations Not tested in sufficient concentration
Inhibitory (filamentous growth) in low concentrations (>2 pprn)
Slight inhibition*
Not tested in sufficient concentration
~
From Rosenberg et al. (42). The slight inhibition observed may b e due to lability of the complex.
86
J. J. ROBERTS AND A. J. THOMSON
shown similarly to produce filamentous growth in E . coli, but in all cases a much higher concentration was required than with platinum complexes. (43). Filamentous growth in bacteria is almost certainly indicative of the ability of an agent to react with DNA, leading to a selective inhibition of DNA synthesis with no accompanying effect on other biosynthetic pathways, such as RNA or protein synthesis. A variety of agents, such as UV and X-irradiation and cytotoxic alkylating agents, can also elicit this response because of their common ability to damage DNA. Other evidence supporting this mechanism for the induction of filamentous growth by the platinum complexes came from tracer studies that compared the distribution of platinum ions within E. coli after the induction of filaments with cis-Pt(II)(NH&l, and after growth inhibition with PtCl,*- (44). In the filamentous cells, platinum ions were associated not only with the metabolic intermediates, but also with cytoplasmic proteins and nucleic acids, whereas in the cells inhibited by PtC1,Z- the platinum was combined only with the cytoplasmic proteins. The accumulation of areas of strikingly enhanced electron density within platinum-induced filaments of E . coli, thought (45) to be aggregates of ribonucleoprotein that had lost its usual distribution pattern but had retained some degree of biochemical integrity owing to its unimpeded rate of synthesis, was therefore also consistent with this proposed biochemical mechanism of filament formation. B. Induction of Lysogeny Further important evidence for direct attack on DNA was provided by the results of Reslova (46),who investigated the ability of platinum compounds to induce the growth of phage from lysogenic strains ofE. coli bacteria. The release of the phage DNA to direct synthesis of new phage is normally a rare event. However, agents reacting with D N A can cause the phage DNA to be released and phage particles to be released with consequent cell lysis. There is an excellent correlation between the antitumor activity of platinum compounds and their ability to induce lysogenic E. coli to enter the lytic cycle (46). The important question of whether viruses are similarly induced in mammalian cells by platinum compounds has not been fully resolved. Induction of viruses and subsequent cell lysis has been proposed as a possible explanation for the rapid disappearance of an S 180 sarcoma following administration of cis-Pt(I1) (47).Epstein-Barr virus (EBV) can be demonstrated in cells of human lymphoblastoid origin either by electron microscopy or by the immunofluorescence test (IF test) and the proportion of such EBV-positive cells, as detected by im-
ANTITUMOR PLATINUM COMPOUNDS
87
munofluorescence, has been shown to increase after treatment with cis-Pt(I1) (48).However, it was not clear whether virus particles or only virus-associated antigens are f o p e d in positive cells. Subsequently electron microscopy (49)revealed a 4fold increase in viruslike particles following treatment with the drug. However, the increase in particle count was mainly due to the appearance of small particles, and these were detected in cells that contained E B virus and in those in which electron microscopy failed to demonstrate EBV. The nature of the small particle has not been clarified. These particles are either immature forms of the virus or the product of aberrant cisPt(11)-influencedmaturation. Alternatively, the small particles are unrelated to EBV and their number increased because of the effect of the platinum compound on the synthesis of a different virus. The further possibility exists that the positive human sera used in the IF test also contained antibodies to this type of particle and that the increase in the percentage of IF-positive cells was a consequence of their reaction with their antigens.
C. Mutagenic Properties of Neutral Platinum Complexes
The radiomimetic nature of platinum compounds and the importance of the geometrical arrangements of ligands for biological effect also emerges from studies on the mutagenic properties of these agents in a number of systems (5053).The cis derivatives were in all cases appreciably more mutagenic than the corresponding trans isomers. D. Reaction with Viruses a n d Transforming DNA The interactions of platinum compounds with viruses have further indicated the relatively greater importance of reactions with DNA as against those with protein in producing biological effects. Kutinova et al. (54) demonstrated the inactivation of the infectious activity of extracellular papovavirus SV40 by cis-Pt(I1). An indication of the mechanism of inactivation was derived by following the capacity of the inactivated virus to induce either tumor or viral antigens. The cap&ity of the virus to induce the tumor antigen was less sensitive to cis-Pt(I1) than viral antigen formation or the infectivity of the virus. The slower rate of inactivation of the tumor antigen by cis-Pt(I1) thus corroborated studies in which SV40 or polyoma virus were inactivated by radiation or hydroxylamine, which revealed that the capacities to induce tumor antigen, thymidine kinase or transplantation immunity, and the transformation activity of the virus, were inactivated at a slower rate than the infectivity or viral antigen-inducing capacity ( 5 5 5 8 ) .These
88
J. J. ROBERTS AND A. J. THOMSON
findings indicate that the viral DNA, not the protein coat, is the primary target for both radiation and the platinum compound. The inactivation of Bacillus subtilis transforming DNA by platinum compounds likewise indicated the effect of these agents on the biological function of DNA (59). Moreover, by examining the changes in transformation frequency of three. unlinked genetic markers, the positions of which on the B . subtilis chromosome were known, some indication of specificity in the reaction of the platinum complexes with regions of DNA or with individual DNA bases was obtained. The adenine marker, for example, is close to the origin of replication and is rich in guanine cytosine base-pairs. This was found to be appreciably more sensitive than the methionine marker, located at the end of the B . subtilis chromosome. It is of interest that the hydrolysis products of cis-Pt(I1) in this system were more reactive than the parent compounds.
IV. Biochemical Effects of Platinum Complexes Indicative of Reactions with DNA A. Inhibition of DNA Synthesis: the Primary Biochemical lesion Induced by Platinum Complexes
The above clues as to the likely basis for the antitumor action of platinum complexes were soon confirmed by studies on cellular biosynthesis in mammalian cells. Simultaneously and independently, it was found that cis-Pt(I1) selectively and persistently inhibited the rate of DNA synthesis as compared with effects on RNA and protein synthesis in human AV3 cells in culture (60) (Fig. 5 ) and in Ehrlich ascites cells in vivo (61).These observations were confirmed in HeLa cells in culture and extended to show that such selective inhibition of DNA synthesis occurs with low doses of drugs, which showed only minimal cytotoxicity as measured by effects on colony-forming ability (62-64).Harder and Rosenberg further showed that those compounds effective against S-180 and also causing filament formation displayed similar effects, whereas the inactive compounds showed no effects until very high dose levels were employed (Fig. 6). Selective inhibition of DNA synthesis has also been demonstrated in phytohemagglutin-stimulated human peripheral lymphocytes (65), in the folate-stimulated kidney, and in the intestinal mucosa of normal and tumor-bearing rats (66). Inhibition of DNA synthesis was also demonstrated in a novel manner by Kara et al. (67),who showed that cis-Pt(I1) not only irreversibly blocked Rous sarcoma virus-stimulated
ANTITUMOR PLATINUM COMPOUNDS
89
DNA synthesis in infected chick embryo cells, but also inhibited their subsequent trans foim ation. The likely basis for this selective biochemical effect on DNA synthesis came fi-omthe observations that the inhibition of DNA synthesis was persistent and progressive with time after removal of the drug (Fig. 7). This latter effect was at first thought to reflect the possible conversion of the drug to other, more reactive species. However, it is now clear, particularly by comparison with analogous effects produced by direct-reacting agents such as mustard gas (68),that both effects are consistent with the view that the primary chemical lesion is in the DNA of the cell, which is then inhibited as a template for DNA replication. Thus modifications to the DNA template will block DNA replication but will not affect transcription or translation. Under conditions of low cell-killing, the selective inhibition of DNA synthesis, but not of RNA or protein synthesis, by platinum compounds leads to the formation of giant cells, a feature observed in cells treated with a variety of agents also known to block DNA replication selectively. N o marked variations in the sensitivity of cells to platinum compounds throughout the cell cycle have been noted (69,70,22). The GI phase is, as with alkylating agents (68), slightly more sensitive than other phases of the cell cycle, and not as a result of any decreased amount of reaction with cell DNA (H. N. A. Fraval and J. J. Roberts, unpublished). In this respect, therefore, platinum compounds resemble other non-cell-cycle-specific drugs, such as alkylating agents and other radiomimetic agents that also act by inactivating the DNA template for DNA replication. The effect of cis-Pt(I1) on the bone marrow has also been likened to other effects of X-irradiation (71). Convincing support for the proposed mechanism of action of platinum compounds has come from other studies on the extent of reaction with DNA and other macromolecules at measured levels of cell survival, as discussed in detail below (Section V, A). The above studies on the inhibition of macromolecular synthesis in cultured cells were based on the rate of incorporation of labeled precursors into macromolecules. Other indications of effects on DNA synthesis based on the size of the DNA synthesized at various times after treatment with cis-Pt(II), are discussed in detail in Section VI, C. B. Enzyme Studies The alternative possibility that DNA synthesis is inhibited because of the inactivation of enzymes involved in DNA replication seem contraindicated not only by the failure of cis-Pt(I1) to block protein synthesis but also by its failure to inactivate DNA polymerase in vitm
90
J. J. ROBERTS A N D A. J. T H O M S O N
a
/
0 2 4 6
10
control
24 hr
incubation Time in[’H] Thyrnidine and b -Pt(a)(NHJ&lz Media
-
?0
-
16-
0 2 4 6
10
24 hr
Incubation Time in &s-Pt(II)(NHI),Clt [‘HI Uridine Media
and
FIG.5. Effect ofcis-[Pt(II)CI,(NH,), on (a) DNA, (b) RNA, or (c) protein synthesis in human AV3 cells as measured by the uptake of tritiated thymidine, uridine, or leucine, respectively. The radioactive precursors and the platinum conipound were applied to the cells at zero time (60).
91
ANTITUMOR PLATINUM COMPOUNDS
t
Incubation Time in Pialinuh Media Incubation flme in Platinum Yedla
I. I I.o
c
@
I
C
09
0
z 0.8 L 0.7 -I
n
u-
0
g 0.6 c 0
g a5
O L ’ ” 0 2 4 6
’
10
24hr
Incubation Time in Platinum Media
FIG.6. Effect of some active antitumor platinum compounds (see Fig. 1) and the inactive congeners trans-[Pt(II)CI,(NH,),] and -[Pt(IV)CI,(NH,),] on (a) DNA, (b) RNA, and ( c )protein synthesis in human AV3 cells (60).
92
J. J. ROBERTS AND A. J. THOMSON
01 -4
I
'
0 2
'
6
'
10
1 24 hf
Incuballon Time after Removal of Experimental Media FIG.7. The relative effect of a 4-hour pretreatment of human AV3 cells with various concentrations of cis-[Pt(II)Cl2(NH,),] on DNA synthesis as measured by uptake of r3H]thymidine into DNA. Medium containing the platinum compound was added at time -4 hours and removed at zero time, and the cells were washed before applying fresh medium. Cells were then labeled for only the last 2 hours of incubation for each data point. The thymidine uptake into DNA is expressed as a fraction of that in control cells (60).
except with very high concentrations (72).Other studies on the reversible interaction of cis and trans platinum coordination complexes with a variety of enzymes also make it unlikely that the inhibition of enzymes is involved in the mechanism of action of these agents (73,74). The extent of binding to the enzyme by either cis or trans isomer was more dependent on the enzyme than on the configuration of the complex. However, the recent proposal (75) that some form of crosslinking between DNA and a DNA polymerase could account for the selective block to DNA replication possibly merits further consideration. When calf thymus DNA modified by dichloro(ethylenediamine) platinum4 was used as substrate for DNA or RNA polyrnerases, Abbreviated [PtCl,(en)]; cis is redundant and unnecessary.
ANTITUMOR PLATINUM COMPOUNDS
93
results at variance with those found in vivo were obtained, since both RNA and DNA synthesis were reduced similarly (76).On the other hand, an analogous study using salmon sperm DNA treated with cisand truns-Pt(I1) clearly suggested that the basis for the selectivity of the cis isomer may reside in its ability to react in a specific molecularly defined configuration with DNA (72).
C. Other Proposed Mechanisms of Actions 1. MODIFICATIONO F THE IMMUNE RFSPONSE
A main objective of this review is to analyze the evidence that implicates DNA as the vital target for platinum compounds in mammalian cells. Thus studies of the extent of their reaction with cellular DNA, their effects on DNA synthesis, and their effects on cells that differ in DNA repair capacity (see Section VI), all indicate that cellkilling occurs as a result of a selective block to DNA replication. However, Rosenberg (47) felt that such a conclusion does not adequately explain a number of other “facts”, particularly the disappearance of large solid tumors in many systems following treatment with dose levels apparently not cytotoxic to the tumor. Nor did such a mechanism of action explain the apparent selective killing of certain tumor cells. Rosenberg therefore argued that specificity is inherent in the immune response of the host animals and is a dominant component in the selective destruction of tumor cells. On the other hand, it is equally not easy to envisage how the platinum compounds can stimulate a specific immune attack upon the tumor when almost all antitumor agents in general and certain of the platinum compounds in particular have themselves been shown to be immunosuppressors. A possible resolution of this apparent conflict was afforded by the observation that certain platinum blues stained the membrane of tumorigenic cells more readily than nontumorigenic cells (77).Further studies suggested that the stained patches could be DNA bound to the surface of cells through association with neuraminic acid (77). Transformed cells produce new antigens at the cell surface, which can generate a host immune reaction and result in the destruction of cells. Rosenberg (47)proposed that in some transformed cells these “strong” antigens are masked by the relatively much “weaker” antigens, such as nucleic acids. He therefore further proposed that the reaction of platinum drugs with the surface DNA leads to disruption of what he called the “antigen-masking” action of the DNA in tumor cells and consequently leads to the exposure of the “strong” antigens on the cell surface to give the enhanced antigenicity discussed above. However, this appealing and highly original concept does not seem to be sup-
94
J. J. ROBERTS AND A. J. THOMSON
ported by subsequent studies. On examination of many more cell types, both normal and malignant, the initially observed correlation between tumorigenicity and cell-surface-associated nucleic acid does not appear to hold (78). However, it is still conceivable that the mitotically inhibited enlarged cell is more antigenic than a normal cell.
2. MUTATIONAS A MECHANISMOF CELL DEATH The possible involvement of the extranuclear 06-position of guanine as a site of reaction in DNA for the platinum compounds (79) and the known mispairing that can occur on bases alkylated in this position during DNA replication (80,81) led to yet another proposal for the mechanism of action of platinum compounds. Reactions on the O6 of guanine could produce multiple mutations in DNA, and these, it is proposed (82), would eventually lead to the death of the cell. Such a mechanism offers an explanation for the apparent considerable delay in cell-killing by these agents in some systems (82). Again, it is not immediately obvious how to reconcile this proposal with the wealth of evidence discussed in this chapter indicating that an effect on the rate of synthesis of DNA is critically important in determining toxicity. Certainly all lesions in DNA including those promutagenic lesions on the OB (83, 84) are likely to interfere with DNA replication. Moreover, it has been found previously that one round of replication of DNA on a template damaged by Nmethyl-N-nitrosourea, which would contain some 06-methylguanine adducts, amplifies that damage into lesions that are a block to further DNA replication. Such a process, if it occurred in response to platinum damage, would clearly contribute to some of the apparent delayed cell-killing by these agents.
V. Interaction of Platinum Compounds with DNA A. In Vivo Studies Indicative of DNA as the Target Molecule
1. QUANTITATIVE ASPECTS OF REACTIONS WITH MACROMOLECULES The biological and biochemical studies discussed so far clearly suggest that a modification of the DNA template by platinum compounds leads to its inactivation for DNA replication. They further indicated that an enhanced inactivation by a specific type of reaction occurs in the case of the cis isomers. Confirmation of these notions was convincingly obtained from a comparison of the effects of cis- and trans-Pt(I1) on the survival of HeLa cells in vitro and their ability to bind to macromolecules.
95
ANTITUMOR PLATINUM COMPOUNDS
The marked difference in the effect of the two isomers on tumors is similarly reflected in their effect on cells in culture (Fig. 8), indicating that this in vitro system is a valid model for mechanistic studies. Differences in concentrations of cis- and trans-Pt(11) required to produce equivalent effects in HeLa cell curvival need not necessarily be indicative of the true amount of reaction occurring with cell constituents, but could merely reflect differences in the relative ease of penetration of the two isomers into cells. However, Pascoe and Roberts (63) showed that the trans isomer binds to cell macromolecules quite as effectively as the cis isomer (Fig. 9). To assess the possible importance of DNA, RNA, and protein as primary targets for platinum(I1) compounds, these binding data (expressed as mol/g macromolecule) were used to construct 'curves of (logarithm of) survival against the amount of drug bound to each type of macromolecule. The resulting graphs were then characterized in a similar way to a curve of cell survival versus dose of drug given to the cells. The shoulder width of the binding curve was given by the value
DO .55 PU
DO = 50 !AM
Treatment for 2hr
LO
only
80 120 160 Concn. of agent ( u M )
200
FIG.8. The effect of cis- and trciti.s-[Pt(II)CI,(NH,),] on the suivival of HeLa cells. Various concentrations of the platinum compounds were applied to HeLa cells in Me,SO for a period of 2 hours, and survival was determined by the ;ibility of the cells to forin colonies within 14 days (63).
96
J. J. ROBERTS AND A. J. THOMSON
0.5-
0.4-
0.3
Y
1.0
o
-
0.2-
0.4
TRANS
n m I 0.1-
CONCN. OF AGENT (YM)
FIG.9. Binding of cis- and trans-[Pt(II)C1,(NH3),I to HeLa cell macromolecules. Cells were treated for 2 hours with the platinum compounds before isolation of DNA (e),RNA (0),cytoplasmic protein (A),and iiuclear protein (0).Essentially the same binding to cellular components occurs with both isomers (63).
B , and the slope of the straight-line portion b y B,. The binding coefficients, B , and B,, for the binding of each isomer to DNA, RNA, and protein are given in Table VIII. For both cis- and trans-Pt(II), the binding coefficients were higher for RNA than for DNA. However, the true significance of these binding coefficients can be appreciated only if account is taken of the molecular weights of the molecules concerned. If one assumes no selectivity in the binding to any particular RNA or proteitl molecule (and there is no evidence for such selectivity in any study with these agents), it is possible to calculate the number of platinum atoms bound to each macromolecule at a given toxic dose. Table IX shows the approximate molecular weight of DNA, RNA, and protein and the results of such a calculation performed at the concentration that reduced the surviving fraction from f to 0.37 f. Theoretically this is the concentration just required to kill one cell. There are significantly more molecules bound to DNA than to either RNA or protein at this concentration of cis- or trans-Pt(11), clearly indicating that DNA is the most sensitive cellular target for both cis- and transPt(I1). The binding data shown in Table IX further indicate that only one molecule of protein out of 5000 will have undergone reaction with platinum. Unless there is considerable specificity in the reaction of platinum drugs with a particular enzyme molecule, this level would be too low to inactivate all the enzyme present. Moreover, the level of reaction with rRNA or tRNA or mRNA would not be expected, again in
97
ANTITUMOR PLATINUM COMPOUNDS
TABLE VIII TO MACROMOLECULES* BINDINGOF [Pt(II)C12(NH,)z] COMPOUNDS Binding to macromolecules
Bo (/.mol/g) (slope of curve)
B , (wnollg) (shoulder width) Compound ~is-[Pt(I1)Cl~(NH,)~1 trans-[Pt(II)Cl~(NH,),]
DNA
RNA
Protein
DNA
RNA
Protein
0.045 0.170
0.300 0.300
0.002
0.0225 0.125
0.030 0.650
0.00675 -
-
a Binding coefficients obtained from curves resulting from a combination of the lethal effects of various dose of cis- and trans-[Pt(II)CI,(NH,),] (Fig. 8) and the expected binding to DNA at these doses obtained from separate studies relating binding to DNA and the dose of platinum compound (Fig. 9). Bo represents the slope of the straight portion of the curves, i.e., the binding to DNA associated with a reduction in survival from f to 0.37 f, while B , represents the binding to DNA when this portion of the curve is extrapolated to 100%(63). More recent data in which the binding to DNA was obtained at the same time as the measurement of cell survival have given lower Bo and B , values (124).
TABLE IX EXTENTOF REACTIONWITH DNA, RNA, AND PROTEIN ON A MOLARBASISAT THE Bo AND B , LEVELSOF BINDINGAS OBTAINEDFROM TABLEVIIIa
Molecule
(piioVg)
1 X 10” 4 x 106 0.5-1 x lo6 2.5 x lo4 1 x 105
0.0225 0.030 0.030 0.030 0.00675
~
DNA mRNA rRNA tRNA Protein
Bo
Approx. mol. wt.
Pt bound when surviving fiaction reduced from f to 0.37 f (mol/mol)
Bo (pmol/g)
Pt bound when surviving fraction reduced from f to 0.37 f (mol/mol)
~
22 PUDNA 1 PUS mRNA 1 PU30 rRNA 1 Pt/1500 tRNA 1 Pt/1500 protein
0.125 0.650 0.650 0.650
-
125 Pt/DNA 2.5 PtfmRNA 1 Pt/2 rRNA 1 Pt/70 tFiNA -
” From Pascoe and Roberts (63,64). the absence of any selectivity of reaction, to inactivate all such molecules and lead to interference with protein synthesis.
2. ROLE OF CROSS-LINKING REACTIONS a. Mammalian Cells. The structural requirement for difunctionality and the principal biochemical effects of the platinum compounds,
98
J. J. ROBERTS AND A. J. THOMSON
as discussed above, soon suggested a parallel between the platinum drugs and the classical bifunctional alkylating agents, such as the nitrogen mustards. The latter have been thought for some time to produce an inhibition of DNA synthesis by their ability to introduce crosslinks into the DNA of mammalian cells. However, it has been a matter of contention whether the principal lesion is a crosslink between strands of the DNA helix or crosslinks between bases on one strand of DNA or possibly even between DNA and protein (85).It soon became a matter of some interest to determine whether platinum compounds were similarly able to introduce crosslinks into cellular DNA and whether such reactions contributed to cell-killing. That the two types of agents act by a similar mechanism was indicated by the fact that a Walker carcinoma with an acquired resistance
LIGHT NORMAL DNA
\= AND RADIOACTIVE
rdU CONTAINING DNA
INTERSTRAND CROSS LINKS
-.'
-
wavy
Llghl
C5Cl Densily gradienl
',
-
---------
Hmvy Hybrld, i L g h l ' * ,
C s t l O e ~ i l ypmdient
FIG.10. Method used for quantitating the percentage of cellular DNA crosslinked in uiuo. Cells were grown for 3 hours in a medium containing by ~is-[Pt(11)Cl,(NH,)~ BrdU and L3H1thymidine to produce a proportion of labeled hybrid DNA. After treatment with platinum compounds, the isolated DNA was separated into single-strandedheavy, crosslinked hybrid, and single-stranded light DNAs by means of isopycnic alkaline cesium chloride gradient centrifugation. The proportion of the single-stranded heavy DNA present as the hybrid species gives a measure of the amount of crosslinking (62) .
99
ANTITUMOR PLATINUM COMPOUNDS
(b) CONTROL
froction number
froction number
CONCN.OF AGENT()IMI
FIG. 11. (a) Quantitation of crosslinking of HeLa cell DNA by cis- and truns[Pt(II)CI,(NH,),] (63). (b) Formation of crosslinked DNA in HeLa cells following treatment with mustard gas or C ~ ~ - [ P ~ ( I I ) C ~ ~ ((62). NH,)~]
100
J. J. ROBERTS AND A. J. THOMSON
to melphalan (L-phenylalanine mustard) is cross-resistant to cis-Pt(I1). Pascoe and Roberts therefore asked the questions: Does cis-Pt(I1) form interstrand crosslinks in DNA in vivo? Are the number of crosslinks in the DNA of HeLa cells at a measured level of cell survival comparable to the number of sulfur-mustard-induced crosslinks present in the genome after treatment with an equitoxic dose of sulfur mustard (86)?In order to estimate crosslinks in the DNA, one strand of DNA was given a density and radioactivity label by growing cells in the presence of [3H]bromodeoxyuridine. Crosslinking between a "labeled-heavy" strand and a "light-unlabeled" strand produced a "labeled-hybrid" species, and these species could be separated in an alkaline cesium chloride gradient (Fig. 10).Not only was crosslinking of DNA by cis-Pt(I1) demonstrated by this technique (Fig. ll), but it could also be calculated, from a knowledge of the overall extent of platination of DNA at a dose producing a measured number of crosslinks in the DNA of estimated molecular weight, that DNA interstrand crosslinking is a relatively rare event compared with mustard-induced crosslinks. Thus whereas approximately 1 in 8 sulfur-mustard reactions gave rise to a crosslink in DNA, in the case of cis-Pt(11)-treated DNA, the figure approximated 1 crosslink in 400 platination reactions. This calculation assumed that the DNAs isolated after the two treatments have the same molecular weight. A further indication of the possible role of the crosslinking reaction in determining the cytotoxic action of platinum drugs was obtained from a study aimed at answering the question: Does the extent of interstrand crosslinking of DNA correlate with the cytotoxic activity of a range of platinum compounds (63, a)? The relative toxicities of the cis and trans isomers of the platinum(I1) and platinum(1V) neutral complexes can be defined by the slopes of the survival curves (Do) obtained by treating HeLa cells in culture (Table X). The relative abilities of the various compounds to induce crosslinking in vivo or in vitro is also given in Table X. Comparison of these two sets of values initially suggested that the relative abilities of cis- and trans-Pt(11) compounds to crosslink DNA in vivo (but not in vitro) was related to their cytotoxic action. Thus the relative ability of cis- and trans-Pt(I1) to kill cells, measured either on the basis of dose [ D p / D $ * = 181 or DNA binding (Table VIII) [ B p / B $ * = 5.61 is of the same order as the 12-fold difference in the doses required to produce equal amounts of crosslinking with the two compounds. However, in the case of the platinum(1V) compounds, despite an even greater difference in the relative toxicities of the cis and trans compounds [Dd'"""/D$* = 431, there was very little difference in their capacities to induce crosslinks
101
ANTITUMOR PLATINUM COMPOUNDS
TABLE X RELATIONSHIP BETWEEN THE CONCENTRATIONS OF PLATINUM COMPOUNDS REQUIREDTO PRODUCED MEASUREDEFFECTSON CELL SURVIVAL AND THOSE REQUIREDTO PRODUCE 40% CROSSLINKING OF COMPLEMENTARY STRANDS OF DNA EITHERin Vitro OR in Vivo (63, 64)
Compound
(PM)
(PM)
Dose required to produce 10% crosslinking in vitro (PM)
cis-[Pt(II)CI,(NH,),J trans- [Pt(II)CI,(NH3),] cis-[Pt(IV)CI,(NH~)J trans- [Pt(IV)CI,(NH,),]
3 55 1.5 65.0
1 50 1.0 37.5
0.5 1.0 33.7 67.0
DO
DQ
Dose required to produce 10% crosslinking in vivo (PM)
150 1900 420 570
in DNA either in vitro or in whole cells. It would seem therefore that interstrand crosslinking is not an important cytotoxic event in the case of the platinum complexes (87). Other evidence supporting this conclusion was obtained from studies on bacteriophage (see below). b. Bacteriophage. A study by Shooter et al. (88)on the inactivation of bacteriophage also indicated that DNA interstrand crosslinking reactions are unlikely to be important cytotoxic events. These authors compared cis- and trans-Pt(I1) and [PtC12(en)14for their abilities to inactivate T7 bacteriophage. They also carried out parallel studies on the extent of interstrand crosslinking of phage DNA with all agents (Fig. 12) and the extent of overall reaction of the ethylenediamine derivative with double-stranded T7 phage DNA. It was possible to calculate that at the dose of labeled drug that reduces survival to 37% (i,e., the mean lethal dose, Do ) there are 5 molecules bound to each T7 phage particle. However, it could be shown that at a dose that induces an average of one crosslink into the phage DNA, there were 35 molecules of platinum drug bound to the phage. Crosslinking is therefore a relatively rare event under these conditions of treatment. Moreover, since there were only five platination reactions with the phage particle at the mean lethal dose, four of which were with the nucleic acid, it follows that the crosslinking reaction could not be an inactivating event. In other words, when one has a level of reaction that can inactivate all phages, only a small proportion of them contain a crosslink in their DNA. A further indication of the lack of importance of a DNA interstrand crosslink as an inactivating event came from the finding that while the
102
J. J. ROBERTS AND A. J. THOMSON ch-compound
36s i3s
tram-compound
36s 33s
FIG.12. Band sedimentation profiles of denatured T7 DNA observed after incubating native DNA in phosphate buffer at 37°C for 2 hours with cis- or trans [Pt(II)CI, (NH,),]. Under these conditions, the sedimentation coefficient of single chains of DNA is 36 S and of the crosslinked chains 53 S . From measurements under the areas of the two peaks, the proportion of crosslinked DNA molecules can b e calculated. These values were plotted against dose of platinum compounds to give the relationships shown in Fig. 13,from which were calculated the concentrations required to give on average one (A), (O), and crosslink in phage DNA with C ~ ~ - [ P ~ ( I I ) C ~ , ( N H ~truns-[Pt(II)Cl~(NH,),I )~] cis- [Pt(II)Cl,(en)l (0).
two cis platinum compounds were appreciably more effective than the trans isomer in inactivating the bacteriophage, all three compounds crosslink DNA with approximately equal efficiency (Figs. 12 and 13). Reasoning analogous to that described above indicated that crosslinking of DNA to protein also contributed little to the inactivating process (88). On the other hand, an indication of what was likely to be an inactivating event in this system and for the inactivation of the singlestranded RNA phages R17 and p2 was gained from a comparison of the inactivation of phages by platinum compounds with their inactivation by mono- and difunctional alkylating agents (89, 90). The extent of binding of [PtClz(en)] and its sensitivity toward the T7 and R17 phages are more comparable to those of the difunctional alkylating agent than to those of the monofunctional alkylating agent: this would indicate that a bifunctional attack on the DNA or RNA by either type of agent is biologically more effective than a monofunctional one. Calculations of the numbers of reactions that occur with the phage nucleic
ANTITUMOR PLATINUM COMPOUNDS
103
pg Pt compoundlml
FIG.13. Relationships between crosslinking of bacteriophage T7 and concentrations of [Pt(II)Cl,(en)] ( 0 ) ;cis- (A) and trans- (0)[Pt(II)CI,(NH,),]. The slopes of these curves give the concentrations required to give on average one crosslink in phage DNA. These values were appreciably higher than the mean lethal doses for inactivation of bacteriophage T7, suggesting that crosslinking of the DNA is not a major inactivating reaction (88).
acids at the mean lethal doses of the mono- and difunctional alkylating agents, combined with a knowledge of the extent of crosslinking with the bifunctional compound and of the products of reaction with DNA, indicated that, as with the platinum compounds, interstrand crosslinking was not likely to be the major inactivating event with difunctional alkylating agents. However, since both single-stranded (R17 and p2) and double-stranded (T7) phage molecules were inactivated more readily by difunctional than by monofunctional aklylating agents, it was suggested that the main contribution to inactivation came from crosslinking of neighboring bases on the same nucleic acid chaih, a reaction known to occur with the difunctional alkylating agents. It would therefore appear, from these comparative studies between alkylatiiig agents and platinum compounds, that some form of crosslinking on one strand of the nucleic acid molecule is responsible for the inactivation of both single- and double-stranded phage molecules by the platinum(I1) compounds.
104
J. J. ROBERTS AND A. J. THOMSON
B. In Vifro Reactions of N e u t r a l Platinum Complexes with Nucleic Acid Components
1. WACTION WITH BASES Changes in the ultraviolet absorption spectrum of salmon sperm DNA after reaction with either cis- or trans-Pt(II)(NH&Cl, provided conclusive evidence that both platinum compounds bind to the organic bases of DNA (Fig. 14) (91, 92). Spectrophotometric studies further confirmed that guanosine, adenosine, and cytidine all react with both isomers, the rate of reaction with guanosine being faster than with the other two (92).The [PtC12(en)]similarly reacted preferentially with guanosine as compared with adenosine and cytidine (93). A very slow reaction occurs with thymine or uracil (92)to give the blue and purple complexes of uncertain structure (38).This evidence indicates that guanine is the base in DNA most liable to react. This view was supported by studies on the reaction of cis-Pt(I1) to DNAs of varying (G + C)/(A + T) ratios (94), which showed that the extent of binding increased with the (G + C)-content. Similarly, the binding of a radioactively labeled platinum compound to the purine bases in DNA was preferentially to guanine (95).
REACTION ON BASES By blocking the various possible binding sites in the purine bases by either methylation or protonation, Mansy et al. (92) defined the sites most likely to be involved in reaction with either cis- or transPt(I1). They concluded that the cis isomer forms a bidentate chelate 2.
SITES OF
FIG. 14. The absorption spectra (1 cm light path) of solutions of salmon sperm DNA in 0.1 M NaCfO, in the presence of cis- and tmns-[Pt(II)Clz(NH,l,l after equilibrium had been achieved (r = Pt : P ratio),which indicate reactions with bases in DNA by both isomers (92).
ANTITUMOR PLATINUM COMPOUNDS
105
with either the 6-NH2 and N-7, or the 6-NH2 and N-1 of adenine, and the 4 N H 2 and N-3 of cytosine. The truns isomer, on the other hand, interacts monofunctionally at the N-7 and N-1 of adenine and the N-3 of cytosine. Both isomers react monofunctionally with the N-7 of guanine and hypoxanthine. Robbins (93) used a similar approach and blocked the N-7, N-9 or both positions of guanine or its derivatives, and concluded that the N-7 position of guanine is a primary point of attachment for cis-Pt(II), but that reaction probably occurs also at a second site. More recently, X-ray diffraction studies of the complexes formed between cis-Pt(NH,),X, and various bases have confirmed some of the features obtained from the early spectrophotometric study. The product of the reaction between inosine and [Pt(en)12]consists of two hypoxanthine rings bound to the platinum ion via the N-7 positions, as in Fig. 15 (96).A very similar structure results from the interaction of [PtC12(en)lwith guanosine (97, 98). Again, the N-7 position becomes occupied by the metal. However there is no interaction of the platinum ion with the 0 - 6 group of guanine. In a recent structure determination, two truns-dichlorobis(diisopropylsu1foxide-S) platinum(I1) complexes are bound to $methyladenine via the N-1 and N-7 positions. The same platinum derivative will also bind to the N-3
FIG.15. The structure ofcis-[Pt(NH& (B'-IMP),]"-. Note the close proximity of the phosphate groups to the amino groups bound to platinum (96).
106
J. J. ROBERTS AND A. J. THOMSON
of 1-methylcytosine (27). Likewise, the N-3 position of cytidine monophosphate is bound by platinum ion (99). There is no evidence yet from crystallographic studies that the 0 - 6 position of guanine, the 6-NHz group of adenine, or the 4-NHz group of cytosine can be occupied by platinum(I1) ions. However, the structure of the blue complex formed with a-pyridone (40) is of interest as the first example of a bridging reaction employing a neighboring ring nitrogen atom and its en01 oxygen group:
Q-----pt O---- Pt
However, there is still no crystallographic evidence for bidentate binding of this type to a single metal center as, for example,
q-;-;y 0'
Such a bidentate binding reaction would be specific to a cisplatinum compound; for this reason, such a binding mode, especially to guanine between the N-7 and 0-6 positions, is an attractive possibility to account for the difference between the biological effectiveness of the cis and trans isomers. Evidence for the type of binding undergone by trans-Pt(I1) is rather sparse. However, this compound was used recently as a heavyatom label in the solution of the structure, by X-ray diffraction, of the phenylalanine tRNA (100).The results show that the trans isomer lost one chloride ion and became bound to the N-7 position of guanine-34, as in Fig. 16. Interestingly, one of the ammine groups makes a hydro-
FIG. 16. The binding site of truns-[PtC1,(NHS),] to the guanosine-34 residue of phenylaIanine tRNA. One chloride ion has been replaced by the N-7 of guanosine. The amino groups are hydrogen-bonded to the phosphoric residue and the 0 - 6 atom of guanosine (100).
107
ANTITUMOR PLATINUM COMPOUNDS
gen bond to the 0-6 position of the same base whereas the ammine group trans to it can make three hydrogen bonds to the phosphate groups. Thus no clear-cut reason for the unique biological behavior of the cis compounds emerges so far from these studies. Although the cis compounds bind two guanine residues, the required stereochemical disposition of the two is unlikely to arise in native DNA without distortion of the structure. The case of bidentate binding to a single base has not yet been identified unambiguously, although evidence for it is claimed from analysis of electronic binding energies, using electron spectroscopy for chemical analysis (101, 79).
3. CROSSLINKING REACTIONS As discussed elsewhere, cis-Pt(I1) can induce the formation of X-links in DNA in vivo (62). There is, however, no direct in vitro evidence indicating which of the many possible binding sites discussed above are involved in such a reaction. A possibility suggested from an examination of a model of DNA is that crosslinking could occur between the &amino groups of adenines in opposing strands of DNA in a dA-dT sequence (25). These groups would be 3.5 A apart, which approximates 3 A, the distance between the cis leaving-groups in cis-Pt(I1). Evidence has been obtained that cis-Pt(I1) can link two NH2 groups in this way in a simple nucleotide (102,103). Another method of investigating DNA crosslinks induced by platinum drugs is illustrated in Fig. 17 (104, 105). SAminoacridine and other planar aromatic molecules intercalate into DNA; in doing so, they increase the interplanar base separation from 3.4A to 6.7 A at the intercalation site. The introduction .of crosslinks into DNA will therefore prevent the intercalation of Saminoacridine. Roos and Arnold (104) studied this inhibition of binding as a function of (G c)content or base-sequence of DNAs or copolymers. It was concluded from such studies that the binding of Pt(en)CI, is both monofunctional and bifunctional. The extent of crosslinking, both inter- and intrastrand, is about 30% for E . coli DNA, and the crosslinking increases with (G C)-content of the DNA. The A-T sequence is also a crosslinking site, but not the A-A (104), a finding consistent with the earlier chemical evidence favoring crosslinking between the 6 N H 2 groups of adenine (92). The affinity of platinum compounds for the N-7 position of guanine, the preferential drug binding by G G sequences of DNA (106), and various other findings led Kelman et al. (107) to argue strongly in favor of a crosslinking reaction between neighboring
+
+
108
J. J. ROBERTS AND A. J. THOMSON
Intercalated 9AA
htercalatim prevented by Pt cross link
P G P t 2 P
Intercalation not prevented
FIG.17. Scheme showing how intercalation of DNA by a dye can be prevented by platinum-induced crosslinks (105).
guanine bases in DNA. For such a reaction to occur, local perturbation of the double helix must occur and, indeed, some evidence for such distortion is available (95, 108, 109). The binding of platinum to the guanine N-7 could weaken the G - C hydrogen bonding, which, it is suggested, makes the N - l position of guanine available for further reaction (207). Goodgame et al. (96) also argued that the crosslinking of guanine moieties via the N-7 position would require considerable distortion of the DNA structures and therefore be unlikely to occur in uiuo. They therefore proposed that a firm binding at N-7 of guanine would initially be formed, with possibly a second weaker bond with the 0-6 position in the same guanine molecule. However, during DNA replication, which would involve separation of the strands of DNA, it could be envisaged that the weak link to the 0-6 could be broken and a new one established with the N-7 of another guanine moiety, Similar reasoning led Goodgame et d. (96) to consider that the crosslinked species present in the DNA of cells treated in vivo with cis-Pt(I1) (62) and visualized by isopycnic gradient centrifugation of alkalidenatured DNA could possibly have been formed during the denaturing process. However, if this were the mechanism for the observed in vivo and in vitro crosslinking, cis and trans compounds should be equaIly effective.
ANTITUMOR PLATINUM COMPOUNDS
4. REACTIONSWITH
SUGAR
109
MOLECULES
Reactions of platinum compounds with other components of nucleic acids, such as the deoxyribose moiety, must also be considered because of the formation of complexes between D-mannitol and cisPt(I1) (110). A yellow and a green complex were isolated; both had a Pt:mannitol ratio of 2 : 1. Studies with 13C NMR showed that the platinum binds with the three nonequivalent carbons in D-mannitol. This reaction is of particular interest in view of its possible relevance to the protective action of D-mannitol against cis-Pt(11)- induced renal toxicity. It is of further interest in view of the contention regarding the ease of formation and stability of platinum-oxygen bonds in DNA as, for example, on the 0-6 position of guanine.
5. REACTIONS WITH PHOSPHATE GROUPS It has emerged from some studies that platinum compounds can interact with phosphate groups, although at the outset an inorganic chemist would surely have argued that any binding would be weak and rather labile. Mention has already been made of the observation of the hydrogen bonding of ammine groups bound to platinum to the phosphate group of RNA (100). One of the intriguing findings to emerge from the structure-activity studies of Connors et a2. (30)was that a hydrogen atom on the nitrogen atom bound to the metal ion was a requirement for activity. Thus tertiary ammines do not give effective drug analogs. Also, in the structure of a complex formed between [PtC12(en)]and cytidine 5’-phosphate, the N-3 of the latter replaces one of the chloride ions, whereas a direct platinum-to-phosphateoxygen link is formed in place of the second chloride ion (99).Another interesting observation was that cis-Pt(11) forms a bridged complex with pyrophosphate. The crystal structure of this compound shows an interesting phosphate-bridged platinum dimer reminiscent of bridged sulfate platinum compounds, as in Fig. 18 ( 1 1 1 ) .
FIG.18.The structure of cis-[Pt,(NH,),P,O,], showing the direct Pt-O-P tion and the short Pt-Pt distance of 3.22 8, (111).
interac-
110
J. J. ROBERTS AND A. J. THOMSON
VI. Repair of DNA Damage Induced by Platinum Complexes in Vivo A. General Comments Some cells are known to be able to remove or circumvent damage to their DNA that has been modified by numerous agents, using various cellular repair processes. This is an area of high current interest and may well be important for understanding further aspects of drug action. The fact that cells can acquire resistance to drugs is an important limitation to their clinical usefulness. It is now necessary to ask whether tumor cells develop this resistance to platinum compounds by use of DNA repair mechanisms. Model studies using prokaryotic and eukaryotic cells in vitro indicate that this is a distinct possibility and these findings strongly support the notion that DNA is indeed the target for these cytotoxic agents. Investigations of the modifying influence of cellular repair processes on the lethal effects of radiation and chemically induced damage in both microbial and mammalian cells have revealed basically two different repair mechanisms. In one of these the DNA-bound adducts are recognized and removed by one of two excision processes. If, however, the damage to the DNA is not excised before the DNA is used as a template for DNA replication, it appears that the cell can circumvent the damage either by a mechanism involving recombination, or by a process called postreplication repair (or replication repair). Both excision- and postreplication-repair processes facilitate the recovery of cells from DNA damage introduced by a variety of physical and chemical agents. These processes have been studied extensively in UV-irradiated cells [for review, see Cleaver (112)l.
B.
Excision Repair
1. BACTERIALSTUDIES The main photoproduct in the DNA of UV-irradiated E . coli or mammalian cells is a “pyrimidine dimer.” The distortion created in the DNA by the dimer is thought to be recognized by an endonuclease that inserts a “nick” (single-strand cleavage) in the DNA adjacent to the dimer. Subsequent steps in this excision-repair process involve removal of the dimer attached to an oligonucleotide and resynthesis of the removed section of the DNA. The contribution this excision-repair process makes to the ultimate survival of bacterial cells treated with various agents can be assessed by determining the sensitivities of strains of E . coli carrying mutations in genes known to code for steps
ANTITUMOR PLATINUM COMPOUNDS
111
in this repair pathway. It was concluded from such studies that damage introduced into DNA by certain bifunctional agents, such as nitrogen mustards, mitomycin C, and psoralen plus visible light, as well as the damage introduced by certain bulky monofunctional carcinogenic agents such as 4-nitroquinoline 1-oxide and 7-bromomethylbenz[a lanthracene, was also eliminated by enzymes encoded in genes already known to code for the enzymes required for the repair of UV-induced thymine dimers. From analogous studies on the sensitivities of such DNA-repair-deficient E. coli mutants to cis-Pt(I1) it was similarly concluded that excision repair processes contribute to only a small extent to the recovery of strains of E . coli from the DNA-damaging effects of this agent (113, 114). Thus Drobnik et al. (113) studied the effects ofcis-Pt(I1) on the colony-forming abilities of strains ofE. coli mutated at thefil, exr, and hcr loci and compared their different sensitivities with their known abilities to survive X- or UVirradiation or treatment with an alkylating agent. An approximate sensitivity-increasing factor attributed to mutations at various loci is given in Table XI. The results indicate that the effects of mutations to the individual loci for the survival of colony-forming ability are different after treatment with cis-Pt(I1) from those for treatment with N-methyl-N’-nitro-N-nitrosoguanidine(MeNNGdn) or for UV- or X-irradiation. Mutation of the j i l marker had a pronounced effect on the growing culture when it affected both filament formation and the preservation of colony-forming ability. Some kinds of damage to the DNA of T-odd bacteriophages can be repaired by the enzymic excision-repair system of the host bacterium (host-cell repair). The role played by the hcr locus is important for the survival of cells after UV irradiation, but not after treatment with MeNNGdn and only rpinimally so after treatment with cis-Pt(11).This relatively minor importance of the hcr locus for the inactivation of E . coli by cis-Pt(I1) was confirmed by the observation that the bacteriophages T3 or T 4 Bol (Brenner) after treatment with cis-Pt(I1) give the same inactivation curves in both Hcr+ and Hcr- strains of the indicator bacteria (115). Similarly it was found that platinum-treated transforming DNA does not appear to be more sensitive when assayed in a strain of H . influenzue, which carries the uwrl mutation and hence lacks the activity of the specific UV-endonuclease that incises UV-irradiated DNA (116). The role of the exr locus in determining the sensitivity ofE. coli to various agents (Table XI) are discussed in Section VI, C in connection with other pathways of DNA repair. Recently, evidence has accumulated for the existence of a second excision repair process in both bacterial and mammalian cells, which
112
J. J. ROBERTS AND A. J. THOMSON
TABLE XI EFFECTOF MUTATIONIN INDIVIDUAL LOCI ON THE RELATIVESENSITIVITY OF COLONY-FORMING AEILITY TO uv, X-IRRADIATION, N-METHYL-N’-NITRON-NITROSOGUANIDINE OR cis- [Pt(II)CI,(NH,),I Phenotype tested Fil-IFil+
Exr+/Exr-
Hcr+/Hcr-
Other phenotypes Hcr-Exr+ Hcr+Exr+ Hcr+Exr+ Hcr-Exr+ Hcr+Exr+ Fil-Hcr+ Fil+Hcr+ Fil+Hcr+ Fil+Hcr+ Fil+Hcr-
Inactivating treatment“
uv uv uv MeNNGdn cis- Pt(11)
uv uv uv MeNNGdn MeNNGdn X
Fil+Hcr+ Fil+Hcr+
cis-Pt(I1) cis- Pt(I1)
Fil-Exr+ Fil-Exr+ Fil+ExrFil-Exr+ Fil+Exr+
uv uv uv MeNNGdn MeNNGdn
Fil+ExrFil+ExrFil+ExrFil-Exr+
M eN NGdn cis-Pt(I1) cis-Pt(I1) cis-Pt(II)
Sensitivity increase factor
Referenceb
11-12 8-12 4.0 3.1 1.0
2 2 1 2 1
2.7-4.5 3 6-( 18.3)’ 3 3.6 1.65-1.93
2 2 1 2 2
13.5 23
1 1
8-15 16.6 (1.4)’-4.3 1 1
2 1 1 2 2
1.2 1.9 1.8 5
2 1 1 1
MeNNGdn = iVV-methyl-N’-nitro-N-nitrosoguanidine; cis-Pt( 11) = cis-[Pt(II)Cl,(NHAzI. Table compiled by Drobnik et al. (113).(1) Data of Drobnik et al. (113);( 2 ) Data of Witkin (153);(3) Data of Rorsch et nl. ( 1 5 4 ) . Taken from initial steep part of the curve.
is able to remove alkyl groups from DNA (possibly those on sites involved in hydrogen bonding) (117), as well as abnormal bases in DNA like uracil (118). The first step involves the removal of a free base by the action of an N-glycosylase? The apurinic or apyrimidinic residue so produced is then a substrate for a specific apurinic/apySee article by Lindahl in this volume.
ANTITUMOR PLATINUM COMPOUNDS
113
rimidinic endonuclease? which incises the DNA. Subsequent steps in this excision process are possibly the same as those for the excision process previously described, namely degradation of a section of the DNA, followed by resynthesis and resealing to the original DNA strand. As a consequence of these various steps in either of the excision processes, it could be envisaged that DNA modified by physical and chemical agents will be reduced in size by the action of the various endonucleases and, after completion of the other steps in the repair process, be restored to its original size. Possibly, depending on the rates and extent of these various processes, such changes could lead to a modification of the molecular weight of cellular DNA and be visualized by physicochemical techniques.
2. MAMMALIANCELLSTUDIES Alkaline sucrose gradient sedimentation of prelabeled cellular DNA following treatment of cells with cis-Pt(I1) revealed no accumulation of low-molecular-weight DNA (119).From a knowledge of the extent of reaction of the platinum compound with DNA at the concentration employed, it could be concluded either that lesions were not generally recognized by an endonuclease that inserted “nicks” into DNA, or, alternatively, if the lesions were recognized by an endonuclease, completion of the later stages of the excision-repair processes lead to the rapid restoration of high-molecular-weight DNA. Moreover, since any apurinic sites in DNA would be converted into DNA single-strand breiiks (“nicks”) under these alkaline conditions, there was no obvious evidence from these studies for the removal of substituted purines by means of a N-glycosylase. On the other hand, the time-dependent changes in the sedimentation profiles of fully labeled DNA observed following a pulse treatment of Chinese hamster cells with cis-Pt(I1) could be interpreted as being due to the initial formation of DNA-interstrand or DNA-protein crosslinks that were subsequently removed during several hours, presumably by a DNA excision-repair process (120). In these experiments, mammalian cells were lysed under conditions that released labeled DNA, which sedimented to positions in an alkaline sucrose gradient corresponding to sedimentation coefficients of 700 S and (a lesser proportion) 400-650 S (Fig. 1%). Prior treatment with cis-Pt(I1) consistently resulted in a dose-dependent increase in the proportion of counts recovered in the 700 S region of the gradient at the expense of the DNA sedimenting in the 400-650 S region of the gradient (Fig. 19a). Hence treatment of cells with these concentrations of cis-Pt(11)hinders the release of DNA from the 700 S “complex.”
114
J.
1. ROBERTS AND A. J. THOMSON
(b)
Total counts 5-
0-
1
I
I
10. Oh
Tot01 carnts S’
0
0
010 Total counts 5
O
1
K)
20
30
Fractions No.
FIG.19. Alkaline sucrose gradient sedimentation profiles of [‘Clthymidine- labeled DNA from Chinese hamster celIs. (a) Immediately after 2-hour treatment with cis[Pt(II)Cl,(NH,),] Inset: the relationship between dose of cis-[Pt(II)C1,(NH3),] and proportion of radioactivity recovered in fractions 1 and 2 (“700 s’’region). (b) At 21 hours after a 2-hour treatment. Greater than 92% recovery of applied radioactivity was observed in all gradients (1 19).
It is probable that this effect is a consequence of crosslinking complementary strands of DNA or of crosslinking DNA to protein, which hinders its denaturation under alkaline conditions. [cis-Pt(I1) increases the amount of protein cosedimenting with DNA, and inhibits the deproteination of DNA by phenol and its extraction into p-aminosalicyclic acid. (J. M. Pascoe and J. J. Roberts, unpublished results).] The subsequent changes in this sedimentation profile that occurred by 24 hours after treatment, when the sedimentation value of the DNA (or complex) was reduced from 700 S to 350 S, may reflect the
I
40
ANTITUMOR PLATINUM COMPOUNDS
115
enzymic production of breaks in the DNA at platinum-damaged sites (Fig 19b). While these results cannot be fully interpreted at present because structures with such high sedimentation values are not fully defined, they do indicate that changes in the molecular weight of DNA or a DNA-protein complex do occur under physiological conditions compatible with the concept that platinum-induced damage in DNA can be repaired. Whether these changes contribute to the recovery of cells from the toxic action of the drug is still not clear. Cellular studies did not indicate any rapid recovery from the damaging effect of cisPt(I1) comparable to the classic split-dose radiation-recovery experiment normally regarded as indicative of repair. In fact, lymphoma cells exposed to a second dose of cis-Pt(I1) after an equal “priming” dose suffered more killing than that produced by the double dose given at one time (69).Synergism was also noted when cis-Pt(I1) was given with camptothecin or carmustine (BCNU), but it is not known if these agents can act at the molecular level to inhibit a cellular repair process in platinum-treated cells. The rare skin condition xeroderma pigmentosum (XP) is characterized by extreme sensitivity to sunlight and a predisposition to skin cancer. Cells taken from persons suffering from this condition are more sensitive to UV-irradiation than normal cells and are deficient in excision repair of UV-induced damage. These same cells are also sensitive to other DNA damaging agents, such as hydrocarbon epoxides, 4nitroquinoline l-oxide, and 7-bromomethylbenz[a]anthracene,and sensitivity is again associated with decreased levels of various manifestations of DNA excision repair [for review, see Roberts (121)l. Moreover, these findings were originally thought to indicate that the repair system that excises thymine dimers is also able to excise certain types of chemical damage. It has now been found that these repairdeficient XP cells are also more sensitive than normal fetal lung cells to cis-Pt(I1) when the lethal effects of the drug are expressed as a function of reaction with DNA rather than as a function of dose of reagent. It could therefore be reasoned that this increased sensitivity of X P cells is similarly due to their decreased ability to excise cisPt(I1)-induced DNA damage (122-125). If this conclusion is correct, the additional finding that an extract from Micrococcus luteus that can incise UV-irradiated DNA does not similarly incise cis-Pt(11)-treated DNA would indicate that different mechanisms are likely to be involved for the repair of the two types of damage. It now appears that platinum is indeed lost, with a half-life of about 3 days, from the DNA of stationary-phase Chinese hamster cells treated with cis-Pt(I1). Since the DNA-bound adducts were stable in
116
J. J. ROBERTS AND A. J. THOMSON
vitro, the removal of the platinum can be assumed to occur by an excision-repair mechanism. Removal of platinum from the genome enabled the cells to survive when normal cell growth was resumed. Moreover, cell survival was directly related to the platinum content of the DNA at the time that exponential growth was resumed (H. N. A. Fraval and J. J. Roberts, unpublished results). C. Postreplication Repair (Replication Repair) 1. BACTERIALSTUDIES
From the previous discussion, it could be conjectured that the majority of DNA-platinum products, those involving one strand of a double helix, are chemically stable and refractory to enzymic removal. If this is indeed so, it could be supposed that persistent lesions in DNA are circumvented during DNA replication by an alternative repair process analogous to that facilitating the survival of excision-defective bacteria after UV-irradiation. Evidence has been adduced for the existence of processes that permit DNA replication to proceed on a radiation or chemically damaged DNA template in both microbial and mammalian cells. DNA synthesized on a template containing unexcised UV-induced thymine dimers has been shown to be of smaller molecular weight than that synthesized in control cells. During the subsequent incubation of treated cells, the newly synthesized DNA increases in size and eventually attains the size of the DNA in control cells. Evidence has been adduced that gaps are initially left in the daughter DNA opposite thymine dimers in the template strand of DNA (126); these gaps are subsequently filled by a process involving recombination controlled by the rec genes (127) and/or de novo DNA synthesis, probably controlled by the exr, (now called Zex) gene. It can be seen in Table XI that mutation of Exr+ to Exr- resulted in an increase of 3-6 times in the sensitivity of E . coli toward UV and MeNNGdn and less than a %fold increase in the sensitivity to the X-irradiation. On the other hand, the increase in sensitivity of colonyforming ability to cis-Pt(1I) due to this mutation is 13-23 times. Beck and Brubaker (114) reached a similar conclusion with regard to the important role of this particular repair pathway from a comparison of the effect of treatment with UV light and cis-Pt( 11) on cell survival and filamentation in recombination-repair-deficient mutants of E . coli K- 12. The recombination-repair-deficient mutant recA13 and the double mutant uurA6 l e d , which is known to undergo extensive autodegradation of DNA after treatment with UV-irradiation,were particularly sensitive to cis-Pt(I1).
ANTITUMOR PLATINUM COMPOUNDS
117
Additional evidence implicating attack on DNA by platinum compounds came from the observation that degradation of DNA in recA mutants occurs following either UV-irradiation or treatment with cisPt(I1). The platinum compound also promoted the release of trichloroacetic-insoluble fragments of DNA from growing cells. The nature of these products is not known, but they may be analogous to the repair products that arise after treatment of bacteria with UV and certain radiomimetic chemicals.
2. MAMMALIANCELLSTUDIES a. General Comments. It now seems certain that mammalian cells also possess varying capacities to replicate their DNA on a template containing unexcised damage, which may be regarded as indicative of different levels of some form of so-called postreplication repair capacity. However, it is also apparent that the mechanism of any such repair process differs from what is thought to occur in bacteria. Newly synthesized DNA in some UV-irradiated mammalian cells is initially smaller than that in control cells, and it has been proposed that gaps are left in new DNA opposite thymine dimers. The subsequent increase in the molecular weight of nascent DNA during posttreatment incubation of cells (and defined operationally as postreplication repair) was, as for bacteria, thought to involve sealing of gaps. However, no evidence for recombinational exchanges was immediately forthcoming. Instead it was proposed that so-called gaps are filled by de no00 DNA synthesis (128). [However, some unconfirmed recent findings do suggest that recombinational exchanges may also occur to a limited extent in irradiated mammalian cells (129, 130).] On the other hand, others have argued that gaps are not formed in newly synthesized DNA but that there is simply a delay in the rate of synthesis at the site of each lesion (131).Thus it was found that the newly synthesized DNA in cis-Pt(I1)-treated cells attained the same size as that in control cells if the posttreatment incubation time was extended to allow for the same amount of overall DNA synthesis in treated and control cells, It has further been proposed that the lesion can be circumvented during DNA replication by a mechanism involving strand displacement (132, 133). However, irrespective of the mechanism involved in synthesizing past radiation or chemically induced lesions in DNA, it has been found that the process in some cells is amenable to inhibition by the trimethylxanthine caffeine. Thus it has been shown that the rate of ligation of newly synthesized DNA in UV-irradiated (134) and in cells treated with 2-[(N-acetoxy-Nacetyl)amino]fluorene (135), N-methyl-N-nitrosourea (136), or
118
J. J. ROBERTS AND A. J. THOMSON
7-bromomethylbenz[a]anthracene (137) was dramatically impaired in the presence of caffeine. As a consequence of this inhibition, many cell lines, competent in this replicative bypass repair, are rendered extremely sensitive to the lethal effects of either UVirradiation (138) or chemical damage (137,139-142) by posttreatment incubation in the presence of nontoxic concentrations of caffeine. There is now ample evidence indicating that UV- or X-irradiation or chemically induced cell-death is a function of the amount of chromosome damage, which can be observed at the first or second mitosis after treatment (143,144). Posttreatment incubation in the presence of caffeine enhances dramatically the chromosome-damaging effects of UV-irradiation and chemicals in both plant and animal cells (145,146). Further evidence that the potentiating effects of caffeine are the result of inhibition of the above replicative bypass process by inhibiting ligation of newly synthesized DNA came from the S-phase specificity of these caffeine effects (140, 147). b. Effects of Caffeine on cis-Pt(l1)-Induced Cell Death and Chromosome Damage. The various cellular effects of cis-Pt(I1) and their modification by caffeine suggests that lesions introduced in DNA by platinum compounds are circumvented by a caffeine-sensitive repair process.
J
1
tb
I
1
20
30
Concn of
l0
d0
cAsPt[IIlOJM)
FIG.20. Survival curves (based on colony-forming ability) for treatment of Chinese hamster V79-379A cells in suspension culture with C~~-[P~(II)CI,(NH,)~] alone (0)or followed by growth in the presence of 0.75 mM caffeine (A). Cells were exposed to cis[Pt(II)CI,(NH,)z] for 2 hours at 37°C (120).
119
ANTITUMOR PLATINUM COMPOUNDS
The effect of a 2-hour treatment with cis-Pt(I1) and the potentiation of this effect by caffeine are shown in Fig. 20. It can be seen that the shoulder on the survival curve is completely abolished by posttreatment incubation in a nontoxic concentration of caffeine (119). The potentiating effect of caffeine on cis-Pt(11)-induced lethality persists for approximately 12 hours in asynchronously growing Chinese hamster cells and during the first S phase only after treatment during the GI phase of synchronously growing Chinese hamster cells (122). In this respect, therefore, the response of cis-Pt(I1)-treated cells resembles that of UV-irradiated (147) or sulfur-mustard-treated cells (140). Cytological studies have been made on Chinese hamster cells at various times after treatment with either cis-Pt(I1) alone or in combination with posttreatment with a nontoxic concentration of caffeine, conditions that reduced ultimate cell survival to 50% and 4%, respectively. Four hours after treatment with 15 p M cis-Pt(II), the number of metaphases containing visible chromosomal abnormalities was not significantly above control level, and this proportion was unaffected by caffeine (Fig. 21). By 14 hours after treatment, 60% of the cells treated only with platinum contained chromosomal aberrations, and this proportion declined to less than 5% by 45 hours after treatment. Posttreatment incubation of cells in a medium containing 0.75 mM caffeine dramatically increased the number of cells containing chromosome damage, only 4% of the metaphases being classified as 100,
Hours after Treatment
FIG.21. Relationship between appearance of chroinosomal aberrations and tiine after treatment with cis-Pt(I1). 0-0: 15 p M cis-[Pt(II)CI,(NH,),]; A-A: 15 p M cis[Pt(II)CI,(NH,),] followed by growth in the presence of 0.75 mM caffeine. Cells were exposed to ci.s-[Pt(II)C1,(NH3)Z]for 2 hours at 37°C (120).
120
J. J. ROBERTS AND A. J. THOMSON
normal by 14 hours after treatment. Less than 1% of cells exposed to caffeine alone exhibited any chromosome abnormalities during the course of the experiment. Caffeine not only increases the number of cis-Pt(11)-treated cells containing chromosomal aberrations, but it also enhances the severity of the damage observed. The most dramatic effect was a marked increase in the number of cells containing shattered chromosomes and those with numerous chromatid deletions and exchanges. The delayed appearance of chromosome abnormalities after cis-Pt( 11)-treatment also suggests that DNA replication is necessary for their formation; in this respect, cis-Pt(I1) resembles UVirradiation and alkylating agents rather than X-irradiation. The proposal has therefore been made that inadequate replication of DNA on a DNA-damaged template is responsible for both cell death and chromosome damage, and that posttreatment incubation of cells in media containing caffeine enhances these two effects of DNA damage by inhibiting a process that would permit replication to proceed past the lesions. Support for this notion has come from studies on both the rate of DNA synthesis and the size of DNA synthesized in both asynchronous and synchronized populations of cis-Pt(I1)-treated cells in the presence and in the absence of caffeine. A study of chromosome damage in Chinese hamster ovary cells treated with another platinum drug revealed that gaps and breaks are found in the first mitosis after treatment, whereas chromatid exchanges are present only in cells at the second mitosis (148), a response that, according to Bender et al. (149), is characteristic of compounds that produce lesions repaired by a postreplication repair mechanism. c. Eflects of Caffeine on the Rate of DNA Synthesis in cis-Pt(ZZ)Treated Asynchronous and Synchronous CeZZs. The dose-dependent depression in rate of synthesis in cis-Pt(11)-treated asynchronous Chinese hamster cells (Fig. 22) can be seen as a dose-dependent delay in the peak rate of DNA synthesis (mid-S) in a synchronous populations of cells treated in GI (Fig. 23) (122).As a consequence of the dose-dependent extension of the time for passage through the S phase, cells were correspondingly delayed in the time at which they underwent cell division. The overall amounts of DNA synthesized after treatment with three different doses of cis-Pt(II), which resulted in a wide range of cell survivals, were not markedly different (i.e., the areas under the S phase peaks were similar). However, the amount of DNA synthesized in treated cultures at much later times after treatment (-20 hours) was very much less than that in control cultures, being decreased in a dose-dependent manner. Part of the decreased amount of DNA synthesis was due to a decrease in the proportion of
121
ANTITUMOR PLATINUM COMPOUNDS
0 1
2
1
c
3
5
FIG.22. The effect of cis-[Pt(II)Cl,(NH,),] on DNA synthesis in Chinese hamster V79-379A cells and the modifying influence of caffeine. After a 2-hour treatment with 30 p M (A, A) or 50 p M (0,0 ) .cis-[Pt(II)CI,(NH,),], cells were resuspended i n fresh medium in the absence (0,A) or in the presence (0,A) of caffeine (0.75 mM), and DNA synthesis measured at the times shown by the incorporation of [3H]dT during 20 minutes into acid-insoluble material (150).
treated cells undergoing cell division and passage into the DNA synthetic phase of the following cell cycle as compared to untreated control cells. In addition, it appears that those cells that did pass into the next DNA synthetic phase in the following cell cycle synthesized DNA at a reduced rate relative to that in control cells. A possible
n
'r 3
L
treatment time
10
n
/ /
M
vA
40
Time after harvest (hours)
FIG.23. The effect of treatment with cis-[Pt(II)Cl,(NH,),] during 1hour ofthe early GI phase of the Chinese hamster cell cycle on subsequent DNA synthesis (measured as in Fig. 22) (122).
122
J. J. ROBERTS AND A. J. THOMSON
explanation for this observation is that the DNA synthesized on a damaged template during the first cell cycle was an inadequate template for further DNA replication in the succeeding cell cycle. The apparent impaired G1 + S transition reported to occur in Chinese hamster ovary cells treated with cis-dichlorobis(cyc1opentylamine)platinum(II) (148) may well be a manifestation of this depressed rate of DNA synthesis as discussed above. It has been found that posttreatment incubation in media containing a nontoxic concentration of caffeine rapidly reverses the cis-Pt(I1)induced inhibition of DNA synthesis in asynchronous populations of cells, and this reversal was most rapid in cells previously treated with the higher dose of the platinum compound (Fig. 22) (150).Posttreatment incubation in the presence of caffeine of G1-treated cells therefore leads to a reversal of the cis-Pt(I1)-induced delay in the peak rate of DNA synthesis (Fig. 24). Under these conditions of cis Pt(I1) and caffeine treatment, the peak rate of synthesis now approximates that in the control cells in time of appearance.
d . Eflects of Cafieine on the Size of Newly Synthesized (Nascent) DNA in cis-Pt(1Z)-Treated Cells. The immediate, dose-dependent selective and persistent inhibition of DNA synthesis induced in cisPt(11)-treated Chinese hamster cells as measured by the decreased uptake of [3H]dT into DNA and as discussed above (Fig. 22) can also be visualized as a dose-dependent decrease in the size of pulselabeled newly synthesized DNA in treated cells (150). However, if compensation is made for the reduction in the rate of DNA synthesis
o-a control w 5pM &-FtII
.-. T
"
+ caffeine
treatment time
/
Time after harvest Ihoursl
FIG.24. The effect of treatment during 1 hour of the GI phase of the cell cycle with 5 p M cis-[Pt(II)Cl,(NH,),] alone and after posttreatment incubation in the presence of caffeine on subsequent DNA synthesis (measured as in Fig. 22) (122).
ANTITUMOR PLATINUM COMPOUNDS
123
by increasing the labeling period in cis-Pt(I1)-treated cells, the alkaline sucrose gradient sedimentation profile of labeled DNA in treated cells is very similar to that of DNA in control untreated cells. From such studies, it was concluded that the replicating machinery is delayed at the site of platinum-induced lesions in the template strand, but with sufficient time it can circumvent the lesions without forming discontinuities (gaps) in the newly synthesized DNA. Alternatively, if gaps are first formed opposite platinum reaction sites in DNA, they must be rapidly filled and are too transitory for detection. In this respect, therefore, replication of DNA in cells treated with cis-Pt(11) differs from that in some UV-irradiated cells (128) in which so-called “gaps” can be detected, but resembles that in cells treated with N-acetoxy-2-acetylaminofluorene (135) or 7-bromomethylbenz[a]anthracene (2.37); in the latter studies, no evidence was adduced for the presence of discontinuities (so-called “gaps”) in newly synthesized DNA. The size of newly synthesized DNA in cis-Pt(11)-treated cells may be contrasted with the size of such DNA in cells treated similarly with cis-Pt(I1) and labeled with [3H]dT in the presence of a nontoxic concentration of caffeine. Under these conditions, the size of nascent DNA was markedly reduced as compared with that in untreated control cells or in cells treated only with cis Pt(I1). The decrease in size of DNA was not the result of a decrease in overall rate of DNA synthesis since, as indicated earlier, the rate of DNA synthesis in cis-Pt(I1)treated cells is faster in the presence of caffeine than in its absence. The size of the DNA synthesized in 4 hours in the presence of caffeine in cis-Pt(I1)-treated cells was dependent on the initial dose of cisPt(I1) (Fig. 25). It thus appears that caffeine, in some as yet inexplicable manner, interferes with the mechanism by which the cell replicates DNA past lesions on the DNA template. Some support for this notion was obtained from a comparison of the distance between platinum-induced lesions on the template strand of DNA and the size of the newly synthesized DNA in cells treated with various doses of cis-Pt(11)and postincubated in the presence of caffeine. The distance between platinum atoms on one strand of DNA was calculated from atomic absorption measurements of the platinum bound to DNA isolated from cis-Pt(11)-treated cells, and this was found to correspond closely to the size of the newly synthesized DNA (Table XII). It was concluded that all platination reactions are normally circumvented during DNA replication by a caffeine-sensitive so-called DNA repair process. A model depicting the effect of platinum lesions in DNA on the size and rate of synthesis of nascent DNA is shown in Fig. 26.
I
-
i
10
30
20
40 f-
10 20 30 FRACTION NVMBER
1
40
10
20
30
10
0 10
30
20
'
- 0 40
c-FRACTDN N W R d
;0.5
F
c
,,,/ ?
0'
I
.I
I
I
I
I
50 @'ti111
I
1
1
1 100
Concn..HM
FIG.25. (a) Alkaline sucrose gradient sedimentation profiles of DNA synthesized during a 2-hour period in the presence of 0.75 mM caffeine in Chinese hamster cells treated for 2 hours with various concentrations of cis- [Pt(II)CI,(NH,),]. Sedimentation analyses were performed after a 30-minute incubation period in the absence of labeled precursor but in the presence of 0.75 mM caffeine. Sedimentation is from right to left. (b) Relationship between number-average-molecular weight (M,) of DNA synthesized in the presence of 0.75 mM caffeine in Chinese hamster V79-379A cells treated with cis[Pt(II)Cl,(NH3)2]and the initial dose of ~is-[Pt(11)Cl~(NH,)~], The M, values were derived from the sedimentation profiles shown in (a) (150).
Lo
125
ANTITUMOR PLATINUM COMPOUNDS
TABLE XI1 RELATIONSHIP BETWEEN DOSE OF cis-[Pt(II)CI,(NH,),] LEVELOF BINDINGTO DNA, AND MOLECULAR WEIGHTOF DNA SYNTHESIZED IN THE PRESENCEOF 0.75 MM CAFFEINE~
Dose of cis-Pt(II)b (PM)
Binding to DNA (WnoVg)
Calculated spacing between DNA platinations (daltons x lo-’)
30 50 100
0.040 0.076 0.111
5 (4.3) 2.6 (2.3) 1.8 (1.55)
M, of DNA synthesized (daltons x lW’)
3.5
1.85 0.95
fl Figures in parentheses are the previously published estimates of distance between platinations, which assumed smaller and variable molecular weights for DNA at the different doses ofcis-[Pt(II)Cl;(NH,),1 and which therefore allowed for an “end” platination. Revised estimates were based on DNA of infinite length. From Van den Berg and Roberts (150). cis-Pt(I1) = cis- [Pt(II)CI2(NH3)Z].
* +
WITH CAFFEINE
WITHOUT CAFFEINE
T
=+
+tar,+
I
II
I
I
’
A
I
A
h
A
h
-
-D I 8
I
81
T
’
-
DI I
8
I
I
I
0
I
A
I I A
I
1 A
A
- ---I
I
h
i
_I
’ Initiation of DNA Synthws bo&d
DNA Synthesis inhibited os o result of delay ot site of lesions
lesions, leaving gaps No reduction in rate d DNA Synthesis
Cornpenqte tor deloy
No gaps detectde
‘
I
I
Gaps fillod
FIG. 26. DNA strand elongation in Chinese hamster cells treated withcis-[Pt(II)Cl,(NH3),] and the effects of caffeine. In the absence of caffeine (left-hand panel) DNA synthesis is delayed at the site of DNA-platinum products (A). If the delay is compensated for, prior to sedimentation analyses, by increasing the length of the labeling period, no discontinuities in the daughter DNA strand are detectable. In the presence of caffeine (right-hand panel) the DNA replicating machinery is no longer delayed at the site of DNA-platinum products (Figs. 22 and 23), but DNA synthesis is reinitiated beyond them with the forniation of gaps presumably opposite the DNA-platinum products (Table XII). While caffeine is present, these gaps persist, but they are sealed on removal of caffeine from the incubation mixture. Key: T = template DNA; D = daughter DNA; A = DNA-Pt product; a = DNA synthesis initiation site.
126
J. J. ROBERTS AND A. J. THOMSON
It is envisaged that DNA synthesis in hamster cells is delayed at the site of DNA-platinum adducts. If, prior to sedimentation analysis of newly synthesized DNA, this delay is compensated for by increasing the length of the labeling period, no discontinuities are detectable in the daughter DNA after treatment with low doses of cis-Pt(I1). The absence of detectable gaps in the daughter DNA of cells exposed to concentrations of cis-Pt(I1) allowing some degree of cell survival suggests that either gaps are not formed at all or are too transitory for detection. The absence of gaps under the above conditions is interpreted as representing the successful operation of a postreplication repair system that allows the synthesis of a continuous daughter DNA molecule on a template containing unexcised damage. Caffeine inhibits this process by a mechanism as yet unknown. In so doing, it causes the DNA replication machinery to proceed at a faster rate, but to leave gaps in the newly synthesized DNA. These gaps seem to occur opposite lesions in the template strand of DNA (Table XII). It has been proposed that the cellular processes that synthesize high molecular DNA on a DNA template damaged by UV-irradiation or [2-(N-acetoxy-N-acetyl)amino]fluorene[AcON(Ac)Fln] (151) may be inducible. The ability to ligate newly synthesized DNA in Chinese hamster cells treated with UV (134)or AcO(AcNH)Fln (135)is also inhibited in the presence of caffeine. Conceivably, therefore, caffeine could inhibit the induction of DNA replication enzymes required for this postulated inducible process. e . Comparisons between HeLa and Chinese Hamster Cells. A comparison of the concentrations of cis-Pt(I1) required to achieve equal killing of HeLa and Chinese hamster cells (63,120,122,123) indicated that HeLa cells are approximately three times as sensitive as hamster cells. A recent reexamination of the survival of these cells accompanying measured Ievels of binding of platinum to their DNA confirmed the greater sensitivity of HeLa cells by a factor of nearly 2 on the basis of binding to their DNA rather than on the basis of dose of agent administered to the cells under equitoxic conditions (124).Previous studies using these two cell lines revealed no indications of major differences in their abilities to excise products from their DNA (68, 140),and it was proposed that hamster cells possess a caffeinesensitive replication repair process either not present in HeLa cells or present, if at all, only at a reduced level and then not amenable to inhibition by caffeine (141).A comparison of the effect of a number of agents including cis-Pt(11)on DNA synthesis in synchronized populations of HeLa and hamster cells appears to support this conclusion (141).It can be seen in Fig. 27 that, after the treatment of synchronous
127
ANTITUMOR PLATINUM COMPOUNDS
t
Time after harvest (hcursl
FIG.27. Effect of treatment during the G , phase of synchronized HeLa cells on subsequent DNA synthesis. There is no delay in the time of appearance of the peak rate of DNA synthesis relative to that in the control culture (as in hamster cells, Fig. 23), but there is nevertheless a marked close-dependent effect on overall DNA synthesis.
HeLa cells in GI phase with cis-Pt(II), the subsequent rate of DNA synthesis is reduced in a dose-dependent manner relative to that in control cells, but there is no delay in the time of appearance of the peak rate of DNA synthesis, as was observed in similarly treated Chinese hamster cells (Fig. 24). The delay in peak rate of DNA synthesis in chemically treated Chinese hamster cells had previously been interpreted as a reflection of the operation of some forni of replication repair process that circumvents lesions in the template DNA. Delays in the peak rate of DNA synthesis have been noted after treatment of synchronous Chinese hamster cells with other agents that produce less lethality in Chinese hamster cells as compared with HeLa cells. The partial reversal by caffeine of this chemically induced delay in DNA synthesis was cokrespondingly interpreted as a manifestation of its ability to inhibit this particular repair process (137,140).Possibly, owing to the speeding up of the rate of DNA replication in the presence of caffeine, insufficient time is permitted for the replication machinery to circumvent lesions in the template strand, with the result that unligated or gapcontaining” nascent DNA molecules are produced. Caffeine fails to modify DNA synthesis in cis-Pt(I1) HeLa cells in the same manner as in Chinese hamster cells, as would be expected on the above interpretation of its effects (Fig. 28). “
128
J. J . ROBERTS AND A. J. THOMSON 4
Time after harvest lhoursl
FIG.28. Effect of treatment with cis-[Pt(II)CI, (NH3),] (0.25 pM) with and without posttreatment incubation in the presence of caffeine on DNA synthesis in synchronous HeLa cells.
Consistent with these observations was the failure of caffeine to potentiate the lethal effects of cis-Pt(1I) in HeLa cells.
VII. Concluding Remarks This review has compiled a wealth of evidence clearly indicating that DNA is the principal target molecule for neutral platinum complexes in a variety of biological systems. Currently the most convincing mechanism for the cytotoxic action of these agents on cells in culture is that reactions with DNA impair its function as a template for further DNA replication. Alternative mechanisms proposed have not been supported by subsequent studies. A variety of reactions with DNA have now been described, but it is not yet known whether all or only some of these are important in inactivating the DNA template. Not only are most lesions in DNA recognized and removed by an excision-repair process, but in Chinese hamster cells all seem to be recognized by a caffeine-sensitive process that facilitates the ability of the replicating machinery to synthesize past them. Inability to synthesize past lesions is associated with mitotic-delay chromosome damage and eventually cell death. It remains to be determined whether all other cytotoxic platinum compounds act by a similar mechanism.
ANTITUMOR PLATINUM COMPOUNDS
129
The finding that cells can differ in their response to platinum compounds, due either to a decreased ability to excise lesions, as in the case of xeroderma pigmentosum cells, or to a decreased ability to circumvent lesions during DNA replication by some form of replication repair as in the case of HeLa cells, leaves one hopeful that selective attack on tumor cells is feasible. Caffeine can potentiate the antitumor activity of cyclophosphamide and nitrogen mustard in mice (152), probably as a consequence of inhibition of the above replication repair system. Conceivably, therefore, caffeine could be used in humans to achieve selective sensitization of tumor cells to the neutral platinum compounds. With the discovery that enforced diuresis can dramatically decrease the kidney toxicity induced by these agents, one of the initial obstacles to their clinical application seems to have been overcome. On the basis of their present clinical status, these agents clearly have a permanent place in the arrnamentarium of the clinician. Finally, the structure-activity relationships discussed in this review indicate that platinum complexes superior to those initially described will become available for clinical use. ACKNOWLEDGMENTS Both authors wish to record their indebtedness to Barnett Rosenberg and the late Professor Sir Alexander Haddow, who first interested them in the study of platinum compounds as cytotoxic agents. Professor R. J. P. Williams gave invaluable discussions and advice on aspects of the inorganic chemistry. Rustenberg Platinum Mines Ltd., and Johnson Matthey Company Ltd., have provided generous financial support.
REFERENCES I. B. Rosenberg, L. Van Camp, J. E. Trosko and V. H. Mansour, Nature 222, 385 (1969). 2. B. Rosenberg and L. Van Camp, Cancer Res. 30, 1799 (1970). 3. R. W. Talley, Proc. A m . Assoc. Cancer Res. 11, 78 (1970). 4. R. J. Kociba, S. D. Sleight and B. Rosenberg, Cancer Chemother. Rep. 54, 325 (1970). 5. C. W. Welsh,]. Natl. Cancer Inst. 47, 1071 (1971). 6. J. A. Gottlieb and B. Drewinko, Cancer Chemother. Rep. 59,621 (1975). 7 . H. J. Wallace and D. J. Higby, Recent Results Cancer Res. 48, 167 (1974). 8. I. H. Krakoff and A. J. Lippman, Recent Results Cancer Res. 48, 183 (1974). 9. J. M. Hill, E. Loeb, A. S. Maclellan, N. A. Hill, A. Khan and J. Kogler, Recent Results Cancer Res. 48, 145 (1974). 10. J. M. Hill, E. Loeb, A. S. Maclellan, N. 0.Hill, A. Khan and J. J. King, Cancer Chernother. Rep. 59, 647 (1975). 11. E. Wiltshaw and B. Carr, Cancer Res. 48, 178 (1974). 12. D. Hayes, E. Cvitkovic, R. Golbey, E. Scheiner and I. H. Krakoff,Proc. A m . Assoc. Cancer Res. 17, 169 (1976).
130
J. J. ROBERTS AND A. J. THOMSON
13. E. Wiltshaw and T. Kroner, Cancer Treat. Rep. 60, 55 (1976). 14. H. W. Bruckner, C. C. Cohen, G. Deppe, B. Kabakow, R. C. Wallach, E. M. Greenspan, S. B. Gusberg and I. F. HollandJ. Clin. Hematol. Oncol. 7,619 (1977). 15. D. J. Higby, H. J. Wallace, Jr., D. J. Albert and J. F. Holland, Cancer 33, 1219 (1974). 16. L. H. Einhorn, B. E. Furnas and N. Powell, Proc. Am. SOC.Clin. Oncol. 17, 240 (1976). 17. M . Rozencweig, D. D. von Hoff, J. S. Penta and F. M. Muggia,J. Clin. Hematol. Oncol. 7, 672 (1977). 18. R. J. Woodman, A. E. Sirica, M. Gang, I. Kline and J. M. Venditti, Chemotherapy (Basel) 18, 169 (1973). 19. E. M. Walker, Jr. and G. R. Gale, Res. Comrnun. Chem. Pathol. Pharmacol. 6,419 (1973). 20. R. C. Richmond and E. L. Powers, Radiat. Res. 68, 251 (1976). 21. E. B. Douple, R. C. Richmond and M. E. Logan,J. Clin. Hematol. Oncol. 7, 585 (1977). 22. I. I. Szumiel and A. H. W. Nias, Br. J . Cancer 33, 450 (1976). 23. A. H. W. Nias and I. I. Szumie1,J. Clin. Hematol. Oncol. 7, 562 (1977). 24. A. J. Thomson, R. J. P. Williams and S. Reslova, Struct. Bonding (Berlin) 11, 1746 (1972). 25. A. J. Thomson, Recent Results Cancer Res. 48, 38 (1974). 26. M. J. Cleare,J. Clin. Hemutol. Oncol. 7, l(1977). 27. C. J. L. Lock, J. Bradford, R. Faggiani, R. A. Speranzini, G. Turner and M. Zvagulis, J. Clin. Hematol. Oncol. 7, 63 (1977). 28. M. J. Cleare and J. D. Hoeschele, Bioinorg. Chem. 2, 187 (1973). 29. M. J. Cleare and J. D. Hoeschele, Platinum Met. Reo. 17, 1 (1973). 30. T. A. Connors, M. Jones, W. C. J. Ross, P. D. Braddock, A. R. Khokhar and M. L. Tobe, Chem.-Biol. Interact. 5, 415 (1972). 31. P. D. Braddock, T. A. Connors, M. Jones, A. R. Khokhar, D. H. Melzack and M. L. Tobe, Chenz.-BioE.fnteruct. 11, 145 (1975). 32. M. L. Tobe and A. R. Khokhar,J. Clin. Hematol. Oncol. 7, 114 (1977). 33. G. R. Gale, J. A. Howle and E. M. Walker, Jr., Cancer Res. 31, 950 (1971). 34. G. R. Gale, M. G. Rosenblum, L. M. Atkins, E. M. Walker, Jr., A. B. Smith and S. J. Meischen,J. Natl. Cancer Znst. 51, 210 (1973). 35. R. J. Speer, H. Ridgeway, D. P. Stewart, L. M. Hall, A. Zapata and J. M. Hil1,J. Clin. Hematol. Oncol. 7, 210 (1977). 36. G. S. Muraveiskaya, G. A. Kukina, V. S. Orlova, 0. N. Erstafera and M. A. PoraiKoshits, Dokl. Akad. Nauk S S S R 226, 596 (1976). 37. J. P. Davidson, P. J. Faber, R. S. Fischer, Jr., S. Mansy, H. J. Peresie, B. Rosenberg and L. Van Camp, Cancer Chemother. Rep. 59,287 (1975). 38. A. J. Thomson, I. A. G. Roos and R. D. Graham,]. Clin. Hematol. Oncol. 7, 242 (1977). 39. B. Lippert, J . Clin. Hematol. Oncol. 7, 26 (1977). 40. J. K. Barton, H. N. Rabinowitz, D. J. Szalda and S. J. Lippard, JACS 99, 2827 (1977). 41. B. Rosenberg, L. Van Camp and T. Krigas, Nature 203, 698 (1965). 42. B. Rosenberg, L. Van Camp, E. B. Grimley and A. J. Thomson,/BC 242, 1347 (1967). 43. B. Rosenberg, E. Renshaw, L. Van Camp, J. Hartwick and J. Drobnik,J. Bact. 93, 1347 (1967).
ANTITUMOR PLATINUM COMPOUNDS
44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85.
131
E. Renshaw and A. J. Thonuon,]. Boct. 94, 1915 (1967). J. A. Howle and G. R. Gale,]. Bact. 103, 259 ( 1970). S. Reslova, Chem.-Biol. Znteruct. 4, 66 (1971-1972). B. Rosenberg, Cancer Chemother. Rep. 59, 589 (1975). V. Vonka, L. Kutinova, J. Drobnik and J. Branerova,]. Nut!. Cancer Znst. 48, 1277 (1972). E. Anisimova, J. Roubal, I. Vlckova, L. Kutinova and V. Vonka,Acta Virol. (Engl., Ed.) 18, 203 (1974). D. J. Beck and R. R. Brubaker, Mutot. Res. 27, 181 (1975). C.Monti-Bragadini, M. Tamiro and E. Banfi, Chem.-BioL Interact. 11,469(1975). P. Lecointe, J.-P. Macquet, J.-L. Butour and C . Paoletti, Mutat. Res. 48,139 (1977). J. E. Trosko, Recent Results Cancer Res. 48, 108 (1974). L. Kutinova, V. Vonka and J. Drobnik, Neoplusmu 19, 453 (1972). S. A. Aaronson,]. Virol. 6, 393 (1970). A. D. Altstein, G. I. Deichman, 0. F. Sarycheva, N. N. Dodonova, E. M. Tsetlin and N. N. Vissilieva, Virology 33, 747 (1967). R. I. Carp, S. Kit and J. L. Melnick, Virology 29, 503 (1966). R. Lajarjet, R. Cramer and L. Montagnier, Virology 33, 104 (1967). S. Reslova, M. Srogl and J. Drobnik, Adu. Antimicroh. Antineoplastic Chemother., Proc. Znt. Congr., Chemother., 7th, 1971 Vol. 11, p. 209. (1972). H. C. Harder and B. Rosenberg, lnt. ]. Cancer 6, 207 (1970). J. A. HowIe and G. R. Gale, Biochem. Pharmucol. 19,2757 (1970). J. J. Roberts and J. M. Pascoe, Nature 235, 282 (1972). J. M. Pascoe and J. J . Roberts, Biochem. Phurmucol. 23, 1345 (1974). J. M. Pascoe and J. J. Roberts, Biochem. Pharnincol. 23, 1359 (1974). J. A. Howle, H. S. Thompson, A. E. Stone and G. R. Gale,PSEBM 137,820 (1971). D. M. Taylor, K. D. Tew and J. D. Jones, Eur. J . Cancer 12,249 (1976). J. Kara, J. Svoboda and J. Drobnik, Ado. Antiniicrob. Antineoplastic Chemother., Proc. Znt. Congr. Chemother., 7th, 1971 Vol. 11, p. 205 (1972). J. J. Roberts, T. P. Brent and A. R. Crathom, Eur. J. Cancer 7, 515 (1971). B. Drewinko and J. A. Gottlieb, Cancer Chemother. Rep. 59, 665 (1975). M. Ogawa, R. G. Gale, S. J. Meischen and V. A. Cooke, Cuncer Res. 36, 3185 (1976). M. Zak, J. Drobnik and Z. Rezny, Cancer Res. 32, 595 (1972). H. C. Harder, R. G. Smith and A. F. Leroy, Cancer Res. 36,3821 (1976). J. E. Teggins and M. E. Friedman, BBA 350, 273 (1974). M. E. Friedman and J. E. Teggins, BBA 350, 263 (1974). M. E. Friedman and P. Melius,]. Clin. Hematol. Oncol. 7, 503 (1977). T. Giraldi and D. M. Taylor, Biochem.Pharmacol. 23, 1659 (1974). S. K. Aggarwal, R. W. Wagner, P. K. McAllister and B. Rosenberg, PNAS 72, 928 (1975). P. K. McAllister, B. Rosenberg, S. K. Aggarwal and R. W. Wagner,]. C h . Hemutol. Oncol. 7, 717 (1977). M. M. Millard, J. P. Macquet and T. Theophanides, BBA 402, 166 (1975). A. Loveless, Nature 223, 206 (1969). L. L. Gerchman and D. B. Ludlum, BBA 308, 310 (1973). B. Rosenberg,]. Clin. Hematol. Oncol. 7, 817 (1977). J. E. Plant and J. J. Roberts, Chem.-Bid.Interact. 3, 343 (1971). J. E. Plant and J. J. Roberts, Chem.-Bid.Interact. 3, 337 (1971). B. Rosenberg, Naturwissenschaften 60, 399 (1973).
132
J. J. ROBERTS AND A. J. THOMSON
86. C . R. Ball and J. J. Roberts, Chem.-Biol. Interact. 2, 321 (1970). 87. H. C. Harder, Chem.-Biol. Interact. 10, 27 (1975). 88. K. V. Shooter, R. Howse, R. K. Merrifield and A. B. Robbins, Chem.-Biol. Interact. 5, 289 (1972). 89. P. D. Lawley, J. H. Lethbridge, P. A. Edwards and K. V. Shooter, JMB 39, 181 (1969). 90. K. V. Shooter, P. A. Edwards and P. D. Lawley, BJ 125, 829 (1971). 91. P. Horacek and J. Drobnik, BBA 254, 341 (1971). 92. S. Mansy, R. Rosenberg and A. J, Thomson,JACS 95, 1633 (1973). 93. A. B. Robbins, Chem.-Biol. Interact. 6, 35 (1973). 94. P. J. Stone, A. D. Kelman and F. M. Sinex, Nature 251, 736 (1974). 95. L. Munchausen and R. 0. Rahn, BBA 414,242 (1975). 96. D. M. L. Goodgame, I. Jeeves, F. L. Phillips and A. C. Skapski, BBA 378, 153 (1975). 97. R. W. Gellert and R. Bau,JACS 97,7379 (1975). 98. R. E. Cramer and M. A. Dahlstrom,J. Clin. Hemutol. Oncol. 7, 330 (1977). 99. S. Louie and R. Bau,JACS 99,3874 (1977). 100. A. Jack, J. E. Ladner, D. Rhodes, R. S. Brown and A. Klug, JMB 111, 315 (1977). 101. J. P. Macquet and T.Theophanides, Bioinorg. Chem. 5, 59 (1975). 102. I. A. G . Roos, A. J. Thomson and S. Mansy,JACS 96,6484 (1974). 103. V. Kleinwachter and R. Zaludova, Chem-Biol. Interact. 16, 207 (1977). 104. I. A. G. Roos and M. C. Arnold,]. Clin. Hematol. Oncol. 7, 374 (1977). 105. I. A. G. Roos, Chern.-Eiol. Interuct. 16, 39 (1977). 106. P. J. Stone, A. D. Kelman, F. M. Sinex, M. M. Bhargava and H. 0. Halvorson,JMB 104, 793 (1976). 107. A. D. Kelman, H. J. Peresie and P.J. StoneJ. Clin. Hernatol. Oncol. 7,440(1977). 108. J.-L. Butour and J.-P. Macquet, 1.Clin. Hemutol. Oncol. 7, 469 (1977). 109. A. M. Tamburro, L. Celotti, D. Furlan and V. Guantieri, Chern.-Biol. Interact. 16, 1 (1977). 110. M. Eshaque, M. J. McKay and T.Theophanides, J. Clin. Hematol. Oncol. 7, 338 (1977). 1 1 1 . J. A. Stanko, Results quoted by M. J. Cleare, Recent Results Cancer Res. 48, 24 (1974). 112. J. E. Cleaver,Ado. Radiat. Biol. 4, l(1974). 113. J. Drobnik, M. Urbankova and A. Krekulova, Mutat. Res. 17, 13 (1973). 114. D. J. Beck and R. R. BrubakerJ. Boct. 116, 1247 (1973). 115. J. Drobnik, A. Blahuskova, S. Vasilukova and A. Krekulova, Chem.-Biol. Interact. 11, 365 (1975). 116. L. L. Munchausen, PNAS 71,4519 (1974). 11 7. D. M. Kirtikar and D. A. Goldthwait, PNAS 71,2022 (1974). 118. T. Lindahl, Noture 259,64 (1976). 119. H. W. Van den Berg and J. J. Roberts, Chem.-Biol. Interuct. 11, 493 (1975). 120. H. W. Van den Berg and J. J. Roberts, Mutot. Res. 33, 279 (1975). 121. J. J. Roberts, Ado. Radiut. Biol. 7, 212 (1978). 122. H. N. A. Fraval and J. J. Roberts, Chern.-Eiol. Znteruct. 23, 99 (1978). 123. H. N. A. Fraval and J. J. Roberts, Chem.-Biol. Interact. 23, 111 (1978). 124. H. N. A. Fraval and J. J. Roberts, submitted for publication. 125. H. N. A. Fraval, C. R. Rawlings and J. J. Roberts, Mutat. Res. 51, 121 (1978). 126. W. D. Rupp and P. Howard-Flanders, JMB 31, 291 (1968).
ANTITUMOR PLATINUM COMPOUNDS
133
127. W. D. Rupp, C. E. Wilde, III, D. L. Reno and P. Howard-Flanders,JMB 61, 25 (1971). 128. A. R. Lehmanm,JMB 66, 319 (1972). 129. R. Meneghini, BBA 425,419 (1976). 130. R. Meneghini and P. Hanawalt, BBA 425, 428 (1976). 131. R. B. Painter, Genetics 78, 139 (1974). 132. N. P. Higgins, K. Kato and B. Strauss,JMB 101, 417 (1976). 133. Y. Fugiwara and M. Tatsumi, Mutat. Res. 37, 91 (1976). 134. J. E. Cleaver and G. H. Thomas, BBRC 36,203 (1969). 135. J. E. Trosko, P. Frank, E. H. Y. Chu and J. E. Becker, Cancer Res. 33,2444 (1973). 136. J. J . Roberts, in “Molecular Mechanism for Repair of DNA” (P. C. Hanwalt and R. B. Setlow, eds.), Part B, p. 611. Plenum, New York, 1975. 137. J. J. Roberts, F. Friedlos, H. W. Van den Berg and D. J. Kirkland, Chem.-Bid. Interact. 17, 265 (1977). 138. A. M. Rauth, Radiat. Res. 31, 121 (1967). 139. A. M. Rauth, B. Barton and C. P. Y. Lee, Cancer Res. 30,2724 (1970). 140. J. J. Roberts and K. N. Ward, Chem.-Bid. Interact. 7, 241 (1973). 141. J. J. Roberts, J. E. Sturrock and K. N. Ward, Mutat. Res. 26, 129 (1974). 142. I. G. Walker and B: D. Reid, Mutat. Res. 12, 101 (1971). 143. S. J. Grote and S. H. Revell, Curr. Top. Radiat. Res. Q. 7, 334 (1972). 144. D. Scott, M. Fox and B. W. Fox, Mutat. Res. 22,207 (1974). 145. B. A. Kihlman, S. Sturelid, B. Hartley-Asp and K. Nilsson, Mutat. Res. 26, 105 (1974). 146. J. J. Roberts and J. E. Sturrock, Mutut. Res. 20,243 (1973). 147. M. Dornon and A. R. Rauth, Radiat. Res. 40, 414 (1969). 148. I. I. Szumiel and A. H. W. Nias, Chenz.-Bid.Interact. 14, 217 (1976). 149. M. A. Bender, H. G. Griggs and J. C. Bedford, Mutat. Res. 23, 197 (1974). 150. H. W. Van den Berg and J. J. Roberts, Chem.-Bid. Interact. 12, 375 (1976). 151. S. M. D’Ambrosio and R. B. Setlow, PNAS 73, 2396 (1976). 152. D. Gaudin and K. L. Yielding, PSEBM 131, 1413 (1969). 153. E. M. Witkin, Brookhauen Symp. B i d . 20, 17 (1968). 154. A. Rorsch, P. Van d e Putte, I. E. Mattem and H. Zwenk, “Genetic and Enzymatic Control of Radiation Sensitivity in Escherichia coli-Genetic Aspects of Radiosensitivity, Mechanism of Repair,” p. 105. IAEA, Vienna, 1966.
This Page Intentionally Left Blank
DNA GIyc osy Iases, Endonucleases for Apu rin ic/Apyrim idin ic Sites, and Base Excision-Repair TOMAS LINDAHL Deportment of Medicul Chemistry University of Gothenburg Gothenburg, Sweden
I. Introduction ..................................................... 11. Models for Excision-Repair of DNA ............................... A. Nucleotide Excision-Repair .................................... B. Base Excision-Repair .......................................... C. Classifications of DNA Lesions ................................ 111. DNA Glycosylases ............................................... A. Uracil-DNA Glycosylase ...................................... B. Hypoxanthine-DNA Glycosylase ............................... C. 3-Methyladenine-DNA Glycosylase ............................ D. Additional DNA Glycosylases ................................. E. Enzymic Hydrolysis of Nucleosides, Mononucleotides, and Related Compounds ...................................... IV. Endonucleases for Apurinic/Apyrimidillic Sites (AP Endonucleases) .............................................. A. Depurination and Depyriniidination of DNA ................... B. AP Endonucleases with Associated Exonuclease Activity ........ C. AP Endonucleases Without Associated Exonuclease Activity ..... D. Endonucleases Acting at Many Lesions, Including Apurinic/Apyrimidinic sites ................................... V. Repair of Apurinic Sites in DNA by Alternative Pathways ........... References.. .....................................................
135 136 136 140 143 145 145 162 165 172 173 173 173 178 183 186 187 188
1. Introduction Since DNA is the carrier of genetic information and spontaneous mutations occur only at low frequency, cellular DNA has often been regarded as an essentially stable entity. Recent developments have necessitated a revision of this view. With the discovery of insertion elements (I), it became clear that certain segments of DNA can move between many different chromosomal sites. Further, the susceptibility of DNA to heat-induced degradation at moderate temperatures and neutral pH leads to hydrolytic decay at a much faster rate than that expected from spontaneous mutation frequencies (2, 3 ) . The latter, somewhat paradoxical, observation can be rationalized by postulating 135 Pmgress in Nucleic Acid Reaearch and Molecular Binlngy, Vd. 22
Copyright 0 1970 Iiy Ac;idcniic Press. Inc. All right:. of reprodiiction in ;iny fnmi resewed. ISBN o-iz-s4nozz-.5
136
TOMAS LINDAHL
the existence of efficient repair mechanisms to maintain the integrity of DNA. In agreement with this notion, several enzymes that act specifically on hydrolytically damaged nucleotide residues in DNA have recently been discovered, purified, and characterized, and they are the main subject of the present review. Some of these enzymes, the DNA glycosylases, belong to a previously unrecognized class of enzymes that cleave base-sugar bonds in DNA (4):In addition to their role in surveying and removing DNA damage that would otherwise lead to unacceptable spontaneous mutation frequencies, the same enzymes may also play an important role in the repair of cellular lesions introduced by ionizing radiation or by exposure to chemical mutagens such as alkylating agents, nitrous acid, or bisulfite. Moreover, one of the mechanisms to ensure that newly synthesized DNA is free from misincorporated dUMP residues, and possibly other unusual nucleotide residues as well, depends on the rapid removal of such residues by enzymes that are also active in the repair of hydrolytically damaged DNA (5, 6).
II. Models for Excision-Repair of DNA A. Nucleotide Excision-Repair A major pathway of DNA repair involves the enzymic excision of damaged residues, followed by repair replication (7). This mode of repair was discovered during studies on the removal of cyclobutanetype pyrimidine dimers from DNA after ultraviolet irradiation (8, G ) , and subsequent work on the rate and detailed mechanism of excisionrepair has to a large extent been concerned with this particular form of damage. The relative ease with which pyrimidine dimers in DNA can be introduced and quantitatively analyzed has certainly contributed to this situation. The individual steps in such a repair process are shown schematically in Fig, 1. It is of interest to discuss excisionrepair of pyrimidine dimers in some detail in the present context, for reference and comparison with the type of excision-repair that occurs after the initial action of a DNA glycosylase. The reactions displayed in Fig. 1, which shows the classical excision-repair model, appear straightforward at first sight, but recent studies on the enzymes involved strongly indicate that the initial steps in this scheme are simplified accounts of complex processes depending on many different proteins. Small, monomeric endonucleases that act in DNA repair as UV-endonucleases' and specifically incise DNA A UV-endonuclease is defined here as an enzyme that incises (cleaves the chain 00 UV-irradiated DNA but does not attack unirradiated DNA.
137
DNA GLYCOSYLASES
UV Irradiation n
UV Endonuclease
1
Exonuclease
DNA Polymerase
I
DNA Ligase
FIG. 1. Scheme for nucleotide excision-repair of DNA containing pyrimidine dimers. The complementary DNA strand is not shown. (a) Specific recognition o f the pyrimidine dimer by a UV endonuclease, which catalyzes the formation of a chain break at the 5' side of the dimer. (b) Exonucleolytic excision of the dimer as part of a small oligonucleotide. The exonuclease additionally releases 20-50 residues in the form of mononucleotides andlor small oligonucleotides. (c) Repair replication catalyzed by a DNA polymerase. The 3' end serves as primer, and the complementary strand as template. (d) Joining catalyzed b y a DNA ligase. Step (c)might precede or occur simultaneously with step (b) in a strand displacement reaction.
138
T O M S LINDAHL
at pyrimidine dimers have been found in Micrococcus luteus (10)and in T4infected Escherichia coli ( 1 1 ) , but it now appears that such enzymes may not be widely distributed and that they are present only in a few unusually radiation-resistant organisms or virus-infected cells, in which they presumably provide an accessory repair pathway. The incision (chain-cleaving) activity at pyrimidine dimers in DNA instead seems to be due to the concerted action of several different gene products in both E . coli and man, and this apparently simple nucleolytic event may in fact be a quite complex process involving several discrete steps. In E . coli, the products of the uwrA, uwrB, and uwrC genes are all required for incision (12, 13), and in man at least five different gene products may be needed (14).The characterization of these interesting proteins is still at an initial stage, but in the E . coli system a promising approach has recently been taken by Seeberg (13,15),who has established conditions that permit the assaying and purification of the individual E . coli uwr gene products by an in witro complementation method. This technique has been instrumental for the characterization of the many different protein factors involved in DNA replication, and it seems likely that it will also yield new insight into the function of the UWT genes. The presently available data indicate that the E . coli uwrA, uwrB, and uwrC gene products are all required for incision in an ATP-dependent process and that the uwrA gene product is a DNA-binding, high-molecular-weight protein of 100,000 daltons, while the uwrB and uwrC gene products may occur as a complex of 70,000 daltons (15, 1 5 2 ) . The reason for such a surprisingly complicated mechanism being employed by E . coli for specific incision at pyrimidine dimers in DNA may depend on the fact that the nucleolytic activity governed by the uvr gene products is much more versatile than the one of lowmolecular-weight UV endonucleases. Thus, the M . luteus and T4 UV endonucleases appear to be specific for pyrimidine dimers, while the E . coli uwr gene products are involved in the repair of many different DNA lesions of quite different structures. In addition to pyrimidine dimers, the excision of several bulky DNA adducts, such as those found after treatment of cells with 4nitroquinoline 1-oxide (15b) or 7-bromomethylbenz[u]anthracene (16),requires functional uwr gene products, as does the removal of interstrand crosslinks in DNA introduced by treatment with psoralen, mitomycin C, or bifunctional alkylating agents (1 7). The apparent ability of the uwr gene products to recognize specifically all these various lesions could easily be explained if DNA were consistently incised at sites of structural distortion. However, such a hypothesis runs into new difficulties, as this
would presumably mean that replicating DNA, actively transcribed DNA, supercoiled DNA, kinked DNA, “breathing” DNA, and the opposite strand at a lesion in DNA would all be potential substrates. It is obvious that such an incision activity would be deleterious to the cell. Instead, the incision activity depending on the uur gene products must be able to discern in a subtle fashion the sites where alterations of the covalent DNA structure have been introduced. An additional problem with the model shown in Fig. 1 is that UV endonucleases such as the T4-induced enzyme catalyze the formation of 3’-OH, 5’-P strand-breaks at the 5’ side of pyrimidine dimers, and these breaks are easily repaired by DNA ligase (18). This situation might lead to an abortive repair process, in which an incision would be made but immediately rejoined before the pyrimidine dimer could be excised. It seems likely that a mechanism exists to avoid such abortive repair by preventing premature ligase action, and this would again involve a processing step additional to those shown in Fig. 1. For example, a DNA phosphatase could remove the terminal phosphate group at the site of incision, or a DNA-binding protein could attach to the incised, irradiated DNA in such a fashion that exonuclease action was permitted, but not ligation. Besides the endonuclease activity dependent on the uur gene products, and the small monomeric UV endonucleases exemplified by the phage T4 u-gene product, a number of other endonucleases that act specifically or preferentially on damaged DNA have been described, and these enzymes may well be active in excision-repair processes analogous to that shown in Fig. 1. The most studied of these enzymes are the endonucleases acting at apurinic and apyrimidinic sites in DNA, which are discussed in detail. There is presently no genetic evidence for the involvement of endonucleases other than those acting at pyrimidine dimers or apurinic sites in DNA repair. However, a mammalian endonuclease discovered by Bacchetti and Brent appears to be a good candidate for a repair enzyme, as it specifically incises DNA at a minor UV-induced lesion different from pyrimidine dimers (19-21 ). An enzyme apparently similar has been identified in E . coli and temied endonuclease I11 (22,23). Different enzymes are responsible for endonucleolytic attack at pyrimidine dimers and subsequent exonuclease action (7). In E . coli, the 5‘ + 3‘ exonuclease function of DNA polymerase I (EC 2.7.7.7) may be the most important excision activity in this regard (24), but other E . coli exonucleases can also perfom1 this reaction in uiuo (25). The latter enzymes presumably include exonuclease VII and the 5‘ + 3’ exonuclease function of DNA polymerase 111.’Exonucleases
140
TOMAS LINDAHL
that degrade pyrimidine-dimer-containing DNA efficiently and that may be active in excision-repair in wivo have also been found in M . luteus (26) and in mammalian cells (27-29). A somewhat puzzling observation with regard to the excision step in both E . coli and mammalian cells is that, in addition to the removal of pyrimidine dimers as part of small oligonucleotides, 20 to 50 nucleotide residues are usually excised in the form of mononucleotides and small oligonucleotides irA each repair event. Further, much longer patches of repair also occur in E . coli (7). It appears that excision continues well beyond the region of structural distortion at the pyrimidine dimer, but then ceases in a controlled fashion. The mechanism for control of exonuclease action is presently not well understood, although several different theories have been proposed (7,24,29-31),but again the simple model in Fig. 1 does not provide an adequate explanation. On the other hand, the two final steps of the excision-repair model, which involve gap-filling by repair replication and subsequent sealing, appear straightforward. The E . coli DNA polymerases and DNA ligase (EC 6.5.1.2)have been investigated in great detail, and are known to catalyze processes of this kind efficiently (32, 33). Similarly, the polymerases and ligases of mammalian cells can account clearly for the gap-filling and joining of DNA (34,35). It has been found in recent years that DNA replication involves many more enzymes and other protein factors than previously postulated, and a similar research development is presently taking place with regard to the initiation and control of removal of bulky DNA lesions such as pyrimidine dimers. The intricacy of the mechanisms that have evolved for this purpose presumably reflects the need for versatile but error-free DNA repair processes. 6. Base Excision-Repair With the discoveries of DNA glycosylases, which release abnormal base residues from DNA in free form, and endonucleases that introduce chain breaks at apurinic sites and apyrimidinic sites in DNA (AP endonucleases), it seemed obvious that the initial steps of an excisionrepair process could occur in a different fashion from that shown in Fig. 1.Thus, a model has been proposed for the repair of certain forms of DNA damage in which the initial enzymic event is the release of an altered base residue by a DNA glycosylase (36). The resulting apurinic or apyrimidinic site is then subject to specific endonucleolytic attack. Most observations indicate that the incision made by an AP endonuclease is at the 5' side of the apurinichpyrimidinic residue (37-39). The deoxyribose-5'-phosphate residue, and possibly a small
DNA GLYCOSYLASES
141
number of additional residues, are then released by exonuclease action, and the DNA is finally subject to repair replication and ligation as in the original excision-repair model. The individual steps are schematically shown in Fig. 2 for the correction of a deaminated cytosine residue in DNA. A somewhat similar model has been advanced by Kirtikar and Goldthwait to explain the action of E . coli endonuclease I1 on alkylated DNA, although in that case it was suggested that a single enzyme could both release alkylated purines and cleave phosphodiester bonds (40).The mode of DNA repair displayed in Fig. 2 has been termed “base excision-repair,” in contrast to the “ nucleotide excision-repair” of pyrimidine dimers and similar lesions
(411. The base excision-repair model shown in Fig. 2 has one more processing step than the diagram of nucleotide excision-repair in Fig. I, and this may create the impression that it is a more complicated form of repair. However, such a notion is in all likelihood incorrect, as many of the objections advanced above about oversimplifications in the classical scheme of nucleotide excision-repair do not apply to the newer model. Instead, it seems plausible that base excision-repair is a simpler, but less versatile, mode of repair than the process governed by the U D T genes in E . coli. With regard to size and specificity, the DNA glycosylases found to date are small monomeric proteins highly specific for one particular type of damaged residue. Consequently, it is probable that the base excision-repair process is restricted to the repair of a small number of commonly occurring lesions, and that there is one separate DNA glycosylase for each of those lesions. As already noted, an additional enzymic step might be required after incision next to a pyrimidine dimer in order to prevent rejoining by DNA ligase prior to excision of the dimer. With regard to incisions at apurinic and apyrimidinic sites it is doubtful that such an extra step would be needed because DNA ligase may not be able to join an AMP residue to a deoxyribose-5’-phosphate residue lacking a base at a 5’ end in DNA, and therefore sealing could not occur. No direct experimental evidence is presently available on this point, but for comparison it is noted that DNA ligases cannot join a 3‘-hydroxyl DNA chain to a 5’-phospho-terminated RNA chain on a DNA template (42,43),SO the structure of the residue at the 5’ side of a break is clearly of importance. Alternatively, it has been proposed (39) that the intrinsic 3‘ 3 5’ exonuclease activity associated with some of the AP endonucleases could have an “antiligase” function in DNA repair if incision and generation of a small gap next to the lesion are coupled events occurring prior to excision of the sugar-phosphate residue.
Deamination
(a)
Uracil-DNA Glycosylase
(b)
AP Endonuclease
(c)
Exonuclease
(d)
DNA Polymerase
(e)
DNA Ligase
I
d
J
FIG.2. Scheme for base excision-repair of DNA containing deaminated cytosine residues. The complementary DNA strand is not shown. (a) Specific recognition of the dUMP residue by an uracil-DNA glycosylase, which cleaves the base-sugar'bond by hydrolysis. Uracil is released in free form, and an apyrimidinic site is generated in DNA. (b) The apyrimidinic site is specifically recognized by an endonuclease that catalyzes the formation of a chain break at the 5' side of the lesion. (c) Excision of the deoxyribose-phosphate moiety. It is presently not known whether this residue is released as such or as part of a small oligonucleotide, but extensive exonucleolytic degradation does not occur. (d) Repair replication catalyzed by a DNA polymerase. (e) Joining catalyzed by a DNA ligase.
DNA GLYCOSYLASES
143
The removal of an apurinic site from DNA does not seem to be accompanied by the relatively extensive DNA degradation occurring in connection with removal of pyrimidine diniers, and at least in mammalian cells the removal of an apurinic site is a typical “shortpatch process” that involves only the excision of one or a very few residues (44, 45). Thus, the exonuclease(s) that liberates the deoxyribose-5’-phosphate residue may be active only within the region of local distortion of DNA structure, and in this case it is easy to see how exonuclease action could take place in a controlled fashion. The 5’ + 3’ exonuclease function ofE. coli DNA polymerase I has the ability to perform the required excision in vitro (46), but it is not 3’ exonuclease is primarily reknown if this enzyme or another 5’ sponsible for excision in vivo. C. Classifications of DNA Lesions Several different types of lesions are introduced in DNA after exposure to UV light, ionizing radiation, or chemical mutagens such as alkylating agents; on the other hand, the same type of defect in some cases can be obtained by treatment with different agents. Consequently, it clearly seems unsatisfactory to classify DNA lesions in terms of the exogenous agents that introduce the damage, although this has been a common procedure in early work on DNA repair. Cerutti (47) has instead proposed that different types of DNA base damage should be divided in three main classes: (a) monofunctional lesions causing negligible helix distortion; (b)monofunctional lesions causing minor helix distortion; and (c) monofunctional and difunctional lesions causing major helix distortion. He suggested that separate repair pathways may exist for these different classes, but that lesions within one class might be repaired in a similar fashion. In a different approach, Grossman et al. (7) have employed a division of DNA lesions into two main classes, monoadducts and diadducts, and proposed that endonucleases active in DNA repair (termed “correndonucleases”) could be classified a s of correiidonuclease I type, acting at monoadduct damage, or of correiidonuclease I1 type, acting at’diadduct damage. There is little doubt that these attempts at a general classification of repairable DNA damage are important improvements on earlier terms, such as “X-ray-induced damage” or “niethyl-methanesulfonateinduced damage.” On the other hand, the notions that repair enzymes may consistently recognize different degrees of helix distortion, or discriminate primarily between inoiioadduct and diadduct damage, have not been borne out by subsequent studies on the enzymology of DNA
144
TOMAS LINDAHL
excision-repair. In addition, Cerutti’s classification easily becomes somewhat arbitrary in the absence of information from X-ray diffraction or model-building studies on the structural effects of different types of lesions on DNA helix conformation. For example, Salkyladenine was classified as a lesion causing major helix distortion (47)although there is no direct evidence to support such a view, and it may well be better to consider this form of damage a “lesion causing minor helix distortion” (47) since the alkylated site is not directly involved in hydrogen-bonding with the complementary DNA chain. Knowledge about the enzymes that can initiate excision-repair and their substrate specificities has been considerably extended since the structurally based classifications above were originally proposed, and it now seems a useful approach to arrange different types of DNA damage into groups with respect to the particular type of enzymic event that initiates an excision-repair process. This focuses attention on the enzymes involved rather than on the degree of DNA distortion or the type of adduct formed. It is recognized that most of the enzymological work has been performed on bacteria, but it is assumed that excision-repair processes will turn out to be fundamentally similar in prokaryotic and eukaryotic cells. A classification of this type is suggested here, as an alternative to the previous models, and four main groups of DNA lesions are considered. 1. DNA damage dependent on uvr gene products for repair. The lesions repaired b y this pathway would include a large variety of bulky lesions, and both diadduct damage and certain types of monoadduct damage; e.g., the arylalkylated purine residues obtained after treatment with bromomethylbenzanthracene would fall in this class. In fact, this group becomes fairly similar to Cerutti’s class of “lesions causing major helix distortion,” which shows the versatility of the uvr-dependent repair system. 2. DNA damage involving repair by AP endonucleases. This group would cover the lesions corrected by base excision-repair initiated by DNA glycosylases, and also the repair of apurinic sites and apyrimidinic sites introduced by nonenzymic events. The lesions would be of monoadduct type and mainly of the type that causes minor helix distortion. 3. DNA damage dependent on other repair endonucleases for repair. In addition to the two groups described above, there are probably several accessory excision-repair pathways. It seems likely that certain minor UV- and X-ray-induced products are specifically recognized by DNA repair endonucleases, although direct proof for this
DNA
145
GLYCOSYLASES
notion has not yet been obtained by mutant studies. Thus, it has been hypothesized that E . coli endonuclease 111 and analogous enzymes in other organisms may recognize products of the 5, 6dihydrodihydroxythymine type, which can be removed by excision-repair (23, 47). In addition, certain radiation-resistant organisms, such as M . luteus, may repair pyrimidine diiners both by this pathway, employing a small UV-endonuclease, and by a mechanism analogous to the uvr gene-dependent pathway of E . coli. 4. Lesions not regularly recognized by excision-repair enzymes. This class would comprise a rather heterogeneous group of monofunctional lesions, usually causing minor or negligible helix distortion and being fairly harmless in nature. One lesion of this type is 7-niethylguanine, which is not actively excised from either E . coli, most other microorganisms, or mammalian cells treated with alkylating agents (48). Base analogs such as 5-bromouracil, which can be stably incorporated into DNA but cause increased mutation frequencies and decreased resistance to heat and radiation, are also included in this group. Other lesions that belong here would be potential new fornis of unrepairable damage introduced after treatment with recently developed chemical mutagens to which living cells have not been exposed during the course of evolution. The remainder of this review is concerned with the properties of the enzymes responsible for the early stages of repair within the second of these groups of lesions.
111. DNA Glycosylases A. Uracil-DNA Glycosylase2 Uracil does not normally occur as a component of DNA, and one of the reasons for its absence has recently been shown to be that uracilcontaining DNA is selectively attacked and degraded in uivo. Enzymes preferentially cleaving uracil-containing DNA were initially detected in bacterial cell extracts in several laboratories (49-52). Since most of these studies were concerned with other problems, the reaction products were not adequately characterized, and it was assumed that the degradation was due to cleavage of phosphodiester bonds adjacent to dUMP residues b y an endonuclease.
* By analogy with uridine nucleosidase (uridine ribohydrolase, EC 3.2.2.3), the enzyme may be named “deoxyuridine (in DNA) nucleosidase” (or deoxyribohydrolase). The names of other base-excising glycosylases may be constructed similarly. [Ed.].
146
TOMAS LINDAHL
In a more detailed investigation of the degradation of uracilcontaining DNA by an activity partly purified from E . coli extracts, an enzyme was found that catalyzes the cleavage of uracil-deoxyribose The reaction products were free uracil and DNA of bonds in DNA (4). unaltered chain length containing apyrimidinic sites. Since the basesugar bonds in DNA had long been called N-glycosidic bonds, the enzyme was initially referred to as an N-glycosidase. However, these designations are not in accord with current recommendations for carbohydrate nomenclature, in which a glycosidic bond is a linkage through oxygen (53).The DNA base-sugar bonds are termed glycosyl bonds, and the trivial name for an enzyme hydrolyzing such bonds is glycosylase. In the present case, the substrate is uracil-containing DNA, hence the name uracil-DNA glycosylase.2
1. MECHANISMSOF INTRODUCTION OF URACIL INTO DNA Several different pathways for the introduction of uracil residues into DNA exist, and the two most important ones appear to be deamination of cytosine residues and incorporation of dUMP instead of dTMP residues. Uracil-DNA glycosylase is active in the removal of uracil introduced in either fashion. Cytosine is much more susceptible to heat-induced degradation than the other three major base residues in DNA, and the hydrolytic deamination of cytosine and cytidine occurring at neutral pH has been investigated by several groups (54-57). Shapiro and Klein (54) first proposed that such deamination might have mutagenic implications in uioo, and t b t it could occur by one of two alternative routes: by direct hydroxyl ion attack at the Cposition, or by an addition-elimination reaction involving the intermediate formation of a dihydrocytosine derivative, known to be easily deaminated (58).The latter mechanism is presently favored, since NMR measurements on neutral cytidine solutions have provided some evidence for its existence (55) and it is analogous to the mode of action of bisulfite on cytosine: but both pathways of deamination could be of importance. Hydrolytic cytosine deamination has also been studied on the DNA level, employing purified bacterial DNA I4C-labeled in the cytosine residues (59).After prolonged incubation at different temperatures in buffers of physiological ionic strength and pH 7.4, the DNA was enzymically degraded and analyzed for deoxyuridine content. Single-stranded DNA, poly(dC), and monomeric dCMP are deami-
Hayatsu, this series, Vol. 16.
nated at very similar rates, and the reaction proceeds at a rate of = 2 x sec-' at 95"C, essentially independent of the buffer composition. Measurements on poly(dC) and dCMP at several lower temperatures showed that the reaction is associated with an activation energy of 120 kJ/mol, so the expected deamination rate at 37" is k = 2 X lo-'' sec-I. Double-stranded DNA is well prQtected against hydrolytic deamination, which apparently occurs at 0.3-0.5% of the rate observed with single-stranded DNA (T. Lindahl and B. Nyberg, unpublished data). The latter observation makes it somewhat difficult to assess the exact rate of deamination of cytosine residues in DNA that might occur in vivo. However, it is interesting to note that unwinding of DNA occurs during replication and transcription and that 1-2% ofthe parental DNA present in growing mammalian cells appears to be present in single-stranded form (60-62). Such DNA might be a target for hydrolytic deamination. This notion is supported by the recent results of Friedberg et uZ. (63) on the sensitivity of E . coli to bisulfite. This group-specific reagent deaminates cytosine residues in singlestranded DNA, and the bacteria are more bisulfite-sensitive during the logarithmic growth phase than in stationary phase. It is not known at present if hydrolytic degradation of cytosine residues in DNA takes place by direct deamination or b y the addition-elimination mechanism involving transient formation of dihydrocytosine and dihydrouracil, or if both pathways occur in parallel. This matter may be of some importance, because different forms of excision-repair could be employed by cells to delete a damaged residue in the two cases. A direct deamination event would require a repair enzyme that specifically recognizes uracil in DNA. On the other hand, dihydrocytosine residues in DNA have an average lifetime of about 1 hour at 37"C, after which they either revert to cytosine or are deaminated (64),and much of the potential deamination damage could be repaired at this stage if an efficient repair mechanism existed to remove dihydropyrimidines from DNA. Although no direct evidence is available as yet for the presence of such repair, it seems likely that dihydropyrimidines are actively removed from DNA in vivo in some fashion, as they may be the most common form of UV-induced damage after pyrimidine dimers. In addition to deamination by heat-induced hydrolysis, DNA cytosine residues can also be effectively deaminated by group-specific reagents such as bisulfite or nitrous acid. Bisulfite, the neutral aqueous form of sulfur dioxide, converts cytosine residues in DNA to uracil by an addition-elimination reaction with the intermediate formation of 5,6-dihydrocytosine-6-sulfonateand 5,6-dihydrouracil-6-sulfonate.3 It
k
148
TOMAS LINDAHL
selectively deaminates cytosine in DNA without simultaneous deamination of purine residues, and is a useful reagent from this point of view. The reaction is considerably more rapid at weakly acidic than at neutral pH, and, similarly to heat-induced hydrolysis, the bisulfitedependent reaction occurs much more readily with single-stranded DNA than with double-stranded DNA (65).4Nitrous acid efficiently deaminates cytosine to uracil in DNA, but also deaminates adenine to hypoxanthine, and guanine to xanthine and minor products. In contrast to bisulfite, this reagent deaminates both single-stranded and double-stranded DNA effectively (67). Thus, native DNA is deaminated only about half as fast as denatured DNA at pH 4.2,and the rates of reaction with the individual bases proceed in the order guanine > cytosine > adenine, with guanine reacting S-fold more rapidly than adenine. Nitrous acid treatment is clearly a convenient method to deaminate base residues in DNA, but side reactions such as depurination-in particular due to spontaneous cleavage of the labile xhthine-deoxyribose bond (67)-crosslinking (68),and formation of several degradation products of guanine (69) also occur and complicate the interpretation of biological experiments on nitrous-acidinduced mutation us inactivation rates. The genetic effects of nitrous acid have recently been reviewed (70). Introduction of uracil residues into DNA by radiation exposure does not appear to be a reaction of major significance. Uracil can be obtained as a minor secondary photoproduct of UV irradiation, as primary products such as cytosine-containing pyrimidine dimers and dihydrocytosine are readily and spontaneously deaminated and then could generate uracil after enzymic monomerization or dihydrouracil formation. UV-irradiation of DNA containing 5bromouracil or 5iodouracil instead of thymine leads to dehalogenation with the formation of a uracil radical, which abstracts a hydrogen atom from the adjacent deoxyribose and destroys the sugar residue, so the net result is usually a single-strand break in DNA with a uracil residue and a damaged sugar residue at one terminus. Such uracil residues are rapidly excised from DNA in vivo (71),perhaps by direct exonuclease action. If 5-bromouracil-containing DNA is UV-irradiated in the presence of the radical-scavenger cysteamine, dehalogenation still occurs, but the reaction leading to chain scission is suppressed (72); this It is worth noting in passing that an alternative, simple, and effective way of specifically deaminating cytosine residues in single-stranded DNA for experimental purposes is incubation in 1 M NaOH (661,a s cytosine deamination occurs at least l@fold more rapidly thnn other types of alkali-catalyzed DNA degradation, such as depurination and subsequent chain breakage or imidazole ring opening in adenine residues.
method has been employed to create adenine-uracil base-pairs in bacterial DNA for in vivo experiments on the removal of uracil (73). Similarly, 5bromouracil reacts with hydroxylamine giving rise to uracil via the 5,6-adduct7and the strong lethal effect of hydroxylamine on bacteria containing 5bromouracil incorporated in their DNA is probably related to this r e a ~ t i o n . ~ Ionizing radiation converts cytosine in DNA to a series of labile derivatives, but uracil is not a major product (74). Thus, 100 krad of X-irradiation under aerobic conditions of either native or denatured [l4C]cytosine-1abeledDNA in 0.1 M NaCl/O.05M TrisC1, pH 7 - 4 2mM EDTA/2 mM L-histidine caused the conversion of less than 0.02% of the deoxycytidine to deoxyuridine residues, as analyzed by the methods employed to quantitate heat-induced hydrolytic deamination of cytosine in DNA (T. Lindahl and B. Nyberg, unpublished data). Two naturally occurring derivatives of cytosine have been found in DNA, 5methylcytosine, and 5hydroxymethylcytosine. They are of considerable interest in the present context, because their deaminated forms, thymine and 5hydroxymethyluraci1, are not substrates for uracil-DNA glycosylase (4,75).5Methylcytosine apparently occurs in all higher cells and in many but not all bacteria, and modified residues are created by the conversion of a minor part of the cytosine in DNA to 5-methylcytosine by a DNA inethylase that einploys Sadenosylinethionine as methyl donor (76). The physiological functions of the 5methylcytosine residues in DNA are poorly understood, but the widespread occurrence of this type of DNA modification indicates that it is of biological importance. In mammalian cells, 2-3% of the DNA cytosine residues are present in Smethylated form, and modified residues are found both in repeated and unique DNA sequences (77). In E. coli K12, about 0.1% of the DNA cytosine residues are methylated. 5Hydroxymethylcytosine completely replaces cytosine in the DNA of the E. coli bacteriophages T2, T4, and T6. In this case, the modified residues are introduced by using 5-hydroxymethyl-dCTP instead of dCTP as a precursor, and the hydroxymethylated cytosine residues in DNA are subsequently enzymically modified to a large extent by addition of glucose (78). The heat-induced deamination of 5methylcytosine has not been studied at the DNA level, but 5-methyl-dCMP is deaminated 4 times more rapidly than dCMP by hydrolysis at pH 7.4 (59).As the cytosine residues in dCMP and single-stranded DNA are deaminated at very similar rates, it seems likely that 5methylated cytosine residues in Phillips and Brown, this series, Vol. 7.
150
TOMAS LINDAHL
DNA would be deaminated slightly more rapidly than unmodified cytosine residues also in DNA. This means that deamination of 5methylcytosine residues would account for about 10% of the total hydrolytic deamination of DNA in organisms that contain DNA with %3% of the cytosine residues in Smethylated form. Deamination of 5-hydroxymethylcytosine has not been studied by biochemical techniques in either the polynucleotides or mononucleotides, and it is not known if the presence of bound glucose would influence the rate of deamination of the base residues. It seems probable that the rate of deamination of such DNA residues would not differ greatly from that of 5-methyl-dCMP. Phage T4 particles accumulate transition mutations at a surprisingly high rate during incubation in neutral buffers at moderate temperatures, and these mutations are due to deamination of 5hydroxymethylcytosine residues to 5-hydroxymethyluracil (79). These experiments demonstrate that spontaneous mutations may arise in the absence of DNA replication. Uracil may occur in DNA owing to misincorporation of uracil instead of thymine. This pathway is discussed below, together with the description of uracil-DNA-glycosylase-deficient bacterial mutants.
2. OCCURRENCE OF URACIL-DNAGLYCOSYLASE Uracil-DNA glycosylase was first discovered in bacteria, and the E. coli (80) and B . subtilis (81)enzymes have been extensively purified and characterized. Subsequently, uracil-DNA glycosylase was also found in mammalian tissues. The calf thymus enzyme (80) has been partly purified, and the human enzyme has been identified in extracts from placenta and fibroblasts (82, 83). It appears that the enzyme is mainly present in cell nuclei (82). A uracil-DNA glycosylase activity that does not require Mg2+,Pi, or other cofactors is also present in extracts from yeast, wheat germ, and several bacteria besides those mentioned above, e.g., M . luteus and Bacillus stearothermophilus (ref. 63; T. Lindahl and B. Nyberg, unpubIished data). However, embryos or cultured cells of Drosophila melanogaster do not contain detectable uracil-DNA glycosylase activity (63).The E. coli, B . subtilis, and mammalian enzymes have very similar properties, and the present discussion is concerned mainly with E . coli uracil-DNA glycosylase. 3. PHYSICALPROPERTIES OF THE ENZYME E. coli uracil-DNA glycosylase has been purified 11 000-fold by a procedure involving ammonium sulfate fractionation, gel filtration, hydroxyapatite chromatography, and DNA-agarose chromatography.
The purified enzyme is of better than 95% homogeneity (80).B . subtilis uracil-DNA glycosylase is also available in a state close to homogeneity (81).The E . coli enzyme has a sedimentation coefficient of 2.44 S. The native enzyme has a molecular weight of 24,100, as determined from its sedimentation coefficient, Stokes radius, and calculated partial specific volume. In sodium dodecyl sulfate/ polyacrylamide gels, the reduced and denatured enzyme has a molecular weight of 24,900. These data show that E . coli uracil-DNA glycosylase contains a single subunit of molecular weight close of 24,500. It is a typical globular protein with a frictional ratio,f/fo, of 1.2. The amino acid composition shows no markedly unusual features, but the protein is low in cysteine and serine residues and high in glutamic acid and/or glutamine. The ultraviolet spectrum is also typical of a protein, and there is no reason to suspect that the active enzyme molecule contains nucleotides, metal ions, or other cofactors. The B . subtilis (81) and calf thymus (80) uracil-DNA glycosylases have molecular weights of 24,000-25,000 and thus do not differ appreciably from the E . coli enzyme in size. The E . coli enzyme is moderately heat-labile. It withstands heating for 5 minutes at 45" in several buffers but is about 80% inactivated after 5 minutes at 50".
4. CATALYTIC PROPERTIES OF THE ENZYME Uracil-DNA glycosylase is a small, apparently uncomplicated enzyme without cofactor requirement. Thus, the enzyme shows full activity in EDTA-containing reaction mixtures, and this property has pennitted measurements of enzyme activity to be performed by incubation of radioactive polynucleotide substrates or uracil-containing DNA with crude cell extracts from E . coli, B . subtilis, or human cells, followed by acid precipitation of the polymer and determinations of the radioactivity of the supernatant solutions, as the activity of most DNases is dependent on the presence of divalent metal ions (75,82). More stringent methods to assay the enzyme in crude extracts have involved paper chromatography of the released material followed by determination of radioactivity in the fomi of free uracil ( 4 ) or passage of the EDTA-containing reaction mixture through a small colump of Dowex-1 in HzO after incubation. In the latter method, the eluate contains the free uracil, which is recovered for assay, while mononucleotides and polynucleotides are adsorbed to the column (6). For assays of the enzyme during purification, the best method is to measure release of acid-soluble material from the DNA of bacteriophage PBS 1or PBSB, radioactively labeled i n the uracil residues (75).E . coli uracil-DNA glycosylase has a broad optimum around pH 8 and the
152
TOMAS LINDAHL
activity is resistant to 5 mM N-ethylmaleimide, The enzyme contains only a single cysteine residue, and this residue thus is not part of the active site. The large majority of enzymes that degrade DNA act by catalyzing the cleavage of phosphodiester bonds, and the term “nuclease” is presently used synonymously with “phosphodiesterase.” As uracilDNA glycosylase was the first exception to this rule, considerable effort has been spent on demonstrating that the enzyme cleaves glycosyl bonds but not phosphodiester bonds in DNA, and this mode of action has now been verified in several laboratories. By employing covalently closed circular DNA molecules from bacteriophage PM2, treated with bisulfite to introduce small numbers of deaminated cytosine residues, it could be shown that the enzyme does not have the ability to convert such DNA molecules to an open circular form. However, the DNA molecules contained apyrimidinic sites as a consequence of the exposure to the enzyme, because they had become alkali-labile as well as sensitive to an AP endonuclease (80).In other experiments it was shown that a poly(dC) polymer, 1% deaminated by alkali treatment, acquired about 1 alkali-labile site per 100 residues after exposure to uracil-DNA glycosylase (4),and that the enzyme released a large proportion of [3H]uracil but no 32Pin low-molecularweight form when doubly radioactively labeled PBS2 DNA was employed as substrate (38,41). The cleavage of uracil-deoxyribose bonds in DNA occurs by hydrolysis, and the reaction is practically irreversible. While mononucleotides are often enzymically degraded by phosphorolysis, such a mechanism has been eliminated in the present case, as there is no phosphate requirement of the reaction or incorporation of 32Piinto DNA when uracil is released (80).Further, transfer reactions with exchange of uracil for cytosine or other base residues, either in the form of free bases or as part of mononucleotides or nucleoside triphosphates, during ,the reaction have not been detected. While such an exchange reaction would appear as a convenient repair mechanism, in which an erroneous base residue could be replaced with the correct one, it may be a too complicated task to perform for a small enzyme of the present type. The homogeneous E . coli uracil-DNA glycosylase has a turnover number of 800 uracil residues released &om DNA per minute under the standard reaction conditions, which is neither remarkably high nor low for an enzyme. It is 11,00@foldpurified fiomE. coli extracts, and a rough estimate is that about 300 enzyme molecules are present per cell. The enzyme has a low K,, 4 x 10-*M, for dUMP residues in
DNA, and taken together these data indicate that uracil-DNA glycosylase should have the potential to rapidly and efficiently excise uracil residues from DNA in uivo. The enzyme is specific for dUMP residues in polymeric form, and does not degrade free dUMP or deoxyuridine. The shortest oligodeoxynucleotide attacked is a uracil-containing tetranucleotide, and it is presently unclear whether dUMP residues at termini of oligodeoxynucleotides or polydeoxynucleotides can be cleaved. Further, uracil is not released from RNA, even by high concentrations of enzyme (4). In agreement with the latter observation, it has been shown with B. subtilis uracil-DNA glycosylase (81)that the enzyme cannot release uracil from the double-stranded polymer poly(rU) poly(dA). The E . coli uracil-DNA glycosylase has a narrow substrate specificity, and does not release several derivatives or analogs of uracil from DNA at a detectable rate. These include thymine (5methyluracil), which is not released at all by high concentrations of enzyme, 5-hydroxymethyluracil, and 5-bromouracil (4, 75, 80). The catalytic properties seem advantageous for a DNA repair enzyme, which should not act on unaltered DNA or RNA in the cell. However, uracil-DNA glycosylase does not seem to recognize damaged residues other than uracil residues in DNA, and there is no action on ultraviolet-irradiated DNA containing thymine-thymine and thymine-cytosine dimers. Furthermore, purified uracil-DNA glycosylase does not release hypoxanthine or 3-methyladenine from deaminated or alkylated DNA, but these altered bases are excised by separate DNA glycosylases. While the enzyme seems to have a strict specificity for uracil in DNA, it is less sensitive to DNA conformation. Thus, uracil is effectively released from both single-stranded and double-stranded DNA and polydeoxynucleotides, and double-stranded DNA is attacked independent of whether the uracil residues are hydrogen-bonded to adenine, as in PBSl DNA or DNA containing uracil misincorporated instead of thymine, or if they are hydrogen-bonded to guanine, as in partly deaminated DNA (80).The versatility of the enzyme in this respect also appears to be a useful property of a repair enzyme, and these data on the substrate specificity of the purified enzyme in vitro accurately reflect the properties of the enzyme in viuo, as bacterial mutants deficient in uracil-DNA glycosylase have reduced ability to remove uracil hydrogen-bonded to either adenine or guanine in DNA. E . coli uracil-DNA glycosylase is product-inhibited by free uracil, M (80). which acts as a noncompetitive inhibitor with a K i of 1 x The calf thymus uracil-DNA glycosylase is also product-inhibited by
154
TOMAS LINDAHL
uracil, but less efficiently than the E . coli enzyme, as the mammalian enzyme has an approximate Ki of 1 x l t 3 M (T. Lindahl and B. Nyberg, unpublished data). Since intracellular uracil concentrations under normal conditions are unlikely to be high enough to cause significant inhibition of the enzyme, this property of the enzyme seems of little physiological relevance, but product inhibition affords another method to investigate the specificity of the uracil-DNA glycosylase. Thus, while uracil is an effective inhibitor of the enzyme, deoxyuridine, dUMP, thymine, Sbromouracil, Saminouracil, 2-thiouracil, and orotic acid do not inhibit E . coli uracil-DNA glycosylase. The absence of detectable inhibition of release of uracil from DNA by the presence of a large excess of deoxyuridine or dUMP agrees with the finding that the enzyme does not cleave nucleosides or mononucleotides, and the inability of 5substituted uracil derivatives to act as inhibitors is consistent with the absence of enzymic activity on DNA containing such derivatives (80). Addition of uracil to a DNA-synthesizing semi-in vitro system of nuclei from pol yomavirus-infected cells has been successfully used (84) to inhibit selectively uracil-DNA glycosylase activity during studies on DNA replication intermediates. 5. URACIL-DNAGLYCOSYLASE AS A REAGENT ENZYME The strict specificity of uracil-DNA glycosylase for uracil residues in DNA, and the ability of the enzyme to act on both double-stranded and single-stranded DNA, suggest that this enzyme may be useful for the detection of small amounts of uracil in DNA. In studies on the effect of bisulfite on native DNA at neutral pH it has been unclear if deamination of cytosine residues occurs to a significant extent at moderate temperatures, as the reaction is much slower than with single-stranded DNA (65).This work was performed by classical base analysis on bisulfite-treated DNA. It seems likely that the sensitivity of such tests could be greatly improved by the determination of uracil in DNA as apyrimidinic sites, introduced by incubation of the DNA with uracil-DNA glycosylase. Such an approach has recently been used to detect traces of uracil in newly synthesized DNA as alkali-labile sites after treatment with uracil-DNA glycosylase (85). A similar accurate method to determine small numbers of alkali-labile apurinic sites in double-stranded bacteriophage T7 DNA was devised by Verly et al. (46).They performed parallel determinations of DNA chain lengths in neutral, formamide-containing gradients and in alkaline sucrose gradients. Cleavage at apurinic and apyrimidinic sites in DNA occurs
as a consequence of alkali treatment, while the same sites survive formamide denaturation of the secondary structure of the DNA, so the presence of alkali-labile sites in DNA is reflected as a discrepancy between the chain-length estimates obtained in the two types of gradients. Small numbers of dUMP residues in covalently closed circular DNA, introduced by treatment with group-specific reagents such as bisulfite, can also easily be detected by treatment of the DNA with uracil-DNA glycosylase, followed by treatment with an APendonuclease or alkali and analysis by gradient centrifugation (80),but this approach has the disadvantage that the group-specific reagent may have reacted preferentially at the regions of destabilized secondary structure that occur in supercoiled DNA. Extensive bisulfite treatment can be used for the essentially complete conversion of cytosine to uracil in single-stranded DNA (65),and such DNA molecules are degraded by uracil-DNA glycosylase, which then attacks at altered cytosine residues. However, there seems to be no obvious present need for this base-specific enzyme in DNA sequencing studies, as convenient and precise sequencing methods are already available (86,87): While uracil-DNA glycosylase may be a helpful reagent enzyme for some purposes, this rather abundant activity may also interfere in work on other enzymes. Moreover, in studies on DNA replication or DNA repair with crude semi-in uitro systems comprised of partly purified enzymes, it may sometimes be important to be able to avoid complications caused by contaminating uracil-DNA glycosylase activity. There are presently three good methods available, (i) for E. coli or B. subtilis, uracil-DNA glycosylase-deficient mutants have been isolated (6, 7 3 ) ,and extracts from some of these strains contain very little enzyme activity; (ii) free uracil may be added to the system to inhibit the enzyme [as little as 2 mM uracil causes more than 90% inhibition of the E . coli enzyme in vitro (80),and the mammalian activity can be strongly suppressed by 6 mM uracil (84)l;(iii) the bacteriophage PBS 1-induced enzyme inhibitor (75) not only interacts with the uracilDNA glycosylase of its B. subtilis host but also inhibits the corresponding mammalian enzyme (E. C . Friedberg, personal communication).
6. BACTERIALMUTANTS DEFICIENTIN URACIL-DNAGLYCOSYLASE Duncan et al. (6,88)isolated eight E . coli K12 mutants deficient in uracil-DNA glycosylase, of which at least five were of different origin, See also Wu et ul. in Vol. 21 of this series.
156
TOMAS LINDAHL
by a nonselective procedure involving screening of cell extracts from 3000 single colony isolates of a heavily mutagenized culture for enzyme activity. Mass screening procedures of this type usually yield leaky mutants (B. Weiss, personal communication), and most of the mutants obtained by Duncan et al. had indeed retained reduced but measurable amounts of uracil-DNA glycosylase activity. The residual enzyme activity in several of the mutant strains was abnormally heatsensitive, showing that the mutations were in the structural gene for the enzyme, They were termed ung mutants. Only one strain, E . coli BD 10, yielded cell extracts without detectable uracil-DNA glycosylase activity (6). However, fractionation of extracts from this strain through two purification steps yielded a protein fraction that contained uracil-DNA glycosylase activity, although it was present only at 0.5%of the level of a corresponding fraction from a wild-type strain (89).It would consequently appear that no E . coli mutant strains totally deficient in this enzyme have so far been isolated, and it is not presently known if uracil-DNA glycosylase is an essential enzyme. Isolation of deletion mutants or, alternatively, temperature-sensitive conditional-lethal mutants in the ung gene should settle this question. The ung gene is located at about 55.6 minutes of the E . coli K12 genetic map (6, 90). This is the structural gene for uracil-DNA glycosylase, which has a monomeric structure. The ung- mutation from the tightest mutant so far isolated, BD10, has been transduced into a wild-type genetic background. The strain obtained in this fashion, as well as the BDlO strain, grow at normal rates in both minimal medium and in nutrient broth. Further, they show normal resistance to ultraviolet light, methyl methanesulfonate, mitomycin C, nalidixic acid, and thymine starvation (6,11). The resistance to thymine starvation actually may be greater than in wild-type strains (H. R. Warner, personal communication). Such starvation would be expected to result in a greatly increased intracellular ratio of dUTP to dTTP, and ungmutants tolerate dUMP residues in their DNA better than wild-type strains (if hydrogen-bonded to dAMP residues, see below). On the other hand, ung strains are abnormally sensitive to nitrous acid (91) and to bisulfite (63). Since the reagents cause deamination of DNA, these results lend strong support to the idea that uracil-DNA glycosylase is involved in the repair of deaminated cytosine residues in DNA. While ung mutants may be identified by their nitrous acid sensitivity, a more convenient method is a spot test with a bacteriophage T4 mutant containing uracil in its DNA, as such phages are only able to grow on an E. coli ung host (6). A B . subtilis mutant deficient in uracil-DNA glycosylase has been
isolated by nonselective screening of extracts from mutagenized cells for enzyme activity (73).This B . subtilis mutant is similar to the E . coli ung- mutants in that it has a reduced ability to remove uracil from its DNA, but grows at a normal rate and is normally resistant to ultraviolet light, X-irradiation, and methyl methanesulfonate treatment. The mutant strain exhibits less than 1%of the wild-type enzyme activity in crude extracts. Bromouracil-labeling of the bacterial DNA, followed by ultraviolet irradiation in the presence of cysteamine to convert bromouracil to uracil residues, demonstrated convincingly that this mutant strain is strongly impaired in its ability to introduce chain breaks in uracil-containing DNA in uivo (73). No mammalian cells deficient in uracil-DNA glycosylase activity have been described. The enzyme is present in normal amounts in human fibroblasts from xeroderma pigmentosum complementation groups A and D and from a case of ataxia-telangiectasia (82,83). The ung- mutation leads to an increase in the spontaneous mutation rate of E . coli K12 (6). Thus, mutation to resistance to nalidixic acid or to rifampicin was Eifold higher in an ung- mutant than in the corresponding ung+ strain. In more detailed studies by trpA reversion analysis, it could be shown that a 15fold increase in the frequency of spontaneous G . C += A * T transition mutations occurred in an ung strain, but other transition and transversion frequencies were not af. results show that uracil-DNA glycosylase is active fected ( 9 1 ~ )These in the repair of spontaneously deaminated cytosine residues in DNA, and.that G U base-pairs arise spontaneously at a sufficiently high rate to cause a marked rise in mutation frequency when repair is impaired. Recent studies (91b) on the molecular nature of spontaneous mutations in the lactose repressor gene of E . coli yield additional information on this point. Comparison of DNA sequences in a series of spontaneous mutants showed that both deletions and point mutations had occurred, and base substitutions were primarily found at two sites in the gene. These two “hot spots” correspond to the positions of two 5-MeCyt residues present in the gene, which apparently had been deaminated to thymine. Since dCMP and 5Me-DCMP are hydrolytically deaminated at similar rates (59), there is no obvious reason to believe that these residues were deaminated much faster than the surrounding DNA cytosine residues. Instead, the results suggest that deaminated 5-MeCyt residues in E . coli DNA cannot be repaired in viuo, in contrast to deaminated cytosine residues. Since deamination of a 5-MeCyt in DNA yields a mispaired thymine residue, which is not a substrate for uracil-DNA glycosylase, this hypothesis seems attractive. In line with this idea, earlier attempts to find a separate E . coli
-
158
TOMAS LINDAHL
DNA glycosylase activity or other enzyme activity that could remove mismatched thymine residues from DNA, employing the doublestranded polydeoxynucleotide poly (dc) - p~ly-(dC,[~H]dT) as a model substrate for DNA containing deaminated 5-methylcytosine residues, yielded entirely negative results (T. Lindahl and B. Nyberg, unpublished results). The strict substrate specificity of E . coli uracil-DNA glycosylase, which does not allow it to act on the deaminated form of unusual cytosine analogs in DNA, is also of relevance with regard to studies on the nature of spontaneous mutations in bacteriophage T4 (79),which showed that spontaneous mutations accumulate at a high rate during storage of phage suspensions at 0°C and that most transition mutations are caused by cytosine deamination. However, since bacteriophage T4 contains hydroxymethylcytosine instead of cytosine, the deamination product in this case is 5hydroxymethyluracil instead of uracil. Such DNA residues are not attacked by uracil-DNA glycosylase (751,as shown with DNA from the B . subtilis phage SPO 1, which contains Shydroxymethyluracil instead of thymine. The high spontaneous mutation frequency observed for phage T4 may thus be analogous to the high mutation frequency at 5-methylcytosine residues in E . coli DNA, and in both cases be related to the inability of uracil-DNA glycosylase to initiate repair of the spontaneously deaminated residues. It has been proposed previously that the reason that DNA contains thymine rather than uracil may be that living cells with genomes much larger than a viral genome must have a uracil-DNA glycosylase in order to prevent an accumulation of mutations due to spontaneous cytosine deamination (80, 92). For the same reason, a large genome containing 5hydroxymethylcytosine or a similar derivative instead of cytosine appears unacceptable, as the deaminated form would not be repaired. A further prediction of this theory would be that mammalian cells and plant cells, which contain relatively large amounts of Smethylcytosine residues in their DNA, have the ability to repair the deaminated form of this residue.
7. MISINCORPORATION OF URACILIN DNA Highly purified E . coli DNA polymerase I can effectively use dUTP instead of dTTP as a precursor and incorporate dUMP residues opposite to dAMP residues in DNA (93). Nevertheless, dUMP residues are not detected in DNA synthesized in t h o . One explanation of these observations has been that the enzyme dUTPase (EC 3.6.1.23) effectively hydrolyzes this deoxynucleoside triphosphate to dUMP
and PPi, so the effective pool size of dUTP in vivo should be quite small (94). In an attempt to obtain an E . coli strain with uracilcontaining DNA, Hochhauser and Weiss (95) isolated dUTPasedeficient mutants, dut-, by the mass screening procedure. However, as these mutants still did not have detectable amounts of uracil in their DNA, it was proposed that any uracil incorporated in DNA might be attacked and removed by uracil-DNA glycosylase. Selective enzymic degradation of newly synthesized DNA containing dUMP residues had in fact been previously reported by Geider (49), who showed that 4X174-infected E . coli cells, permeabilized by ether treatment, accumulated short fragments (3-4 S) of nascent DNA as a consequence of addition of dUTP. More detailed studies on this phenomenon were undertaken when it was discovered that dut- mutants were identical to sof- mutants, which apparently accumulated unusually short Okazaki fragments (4to 5 S instead of 10 S fragments) during DNA replication (5). It seemed likely from these experiments that dUMP residues could indeed be incorporated into DNA, at least in dUTPase-deficient E . coli mutants, and that the subsequent action of uracil-DNA glycosylase or a similar enzyme on the newly replicated, uracil-containing DNA generated strand breaks in the DNA, so that this DNA appeared unusually fragmented. The fragments could be chased into high-molecular weight DNA by a base excision-repair process, and would therefore be difficult to distinguish from the presumably ribonucleotide-primed, “true” replication intermediates described by Okazaki (96). From studies of E . coli DNA replication in vitro, employing lysates on cellophane disks, Olivera (97)has recently concluded that only one of the two DNA strands is synthesized in a discontinuous fashion, but that both strands may undergo excisionrepair soon after synthesis, owing to uracil incorporation. The above model for the generation of fragments of newly synthesized DNA due to uracil incorporation has recently been shown to be correct with the aid of E . coZi dut- ung- double mutants. The dutphenotype of accumulation of short fragments of nascent DNA is almost completely suppressed by the ung- mutation, and the double mutant incorporates substantial amounts of uracil instead of thymine in its DNA (85,96).In E . coli dut- polA-, ordut- lig- double mutants, the short fragments of nascent DNA persist longer than in dut- single mutants, indicating that DNA polymerase I and DNA ligase are involved in the repair of uracil-containing DNA, in agreement with the base excision-repair model (96). Further, dut- xth- double mutants, which are deficient in dUTPase and in the major AP endonuclease
160
TOMAS LINDAHL
activity of E . coli, display poor and filamentous growth and are inviable at temperatures above 30°C. This growth defect, which presumably is due to accumulation of apyrimidinic sites in DNA, is suppressed by the addition of an ung- mutation, so dut- ung- xthp triple mutants grow well (98). The finding that about 10% of the DNA thymine residues are replaced by uracil in E . coli dut ung double mutants, which grow only slightly more slowly than wild-type strains, shows that frequent and persistent substitution of uracil for thymine in DNA may occur in E . coli without serious adverse effects on DNA replication or transcription processes (6,85).Traces of dUMP (about 1residue per 2000-3000 nucleotides) are also incorporated into the DNA ofE. coli dut+ strains (85).These observations again lend support to the idea that the major physiological function of uracil-DNA glycosylase is the repair of deaminated cytosine residues in DNA, Additional evidence for the innocuous effect of replacement of thymine with uracil in DNA in the absence of repair comes from studies by Warner and co-workers with bacteriophages T 4 and T5 growing in E . coli dut- ung- mutants (92, 99). A multiple mutant of T4, deficient in the virus-coded dCTPase/dUTPase but which nevertheless can grow in wild-type E . coli, also grows in E . coli dutung- mutants. In the latter host, about 30%of the T4 DNA thymine is replaced by uracil, and such phage can then be propagated only in an ung- host, as infection of an ung+ strain leads to rapid degradation of the viral DNA, with the production of less than one progeny phage per infected cell. However, in an ung- host, the phage with uracilcontaining DNA grow well, so again the uracil substitution leads to no serious consequences with regard to DNA replication and transcription (92). Analogous results have been obtained with a phage T5 mutant unable to induce a virus-coded dUTPase (99). The mutant grows normally on E . coli dut+, and dut- ung- strains but does not form plaques on dut- ung+ strains because relatively large amounts of dUMP residues are incorporated into DNA in the absence of any dUTPase activity, and the uracil-containing DNA is degraded by the uracil-DNA glycosylase. In dut- ung- hosts, T5 DNA with about 12% of the DNA thymine replaced by uracil is synthesized. Uracil-containing T4 DNA is not completely stable in the E . coli BD 10 ung- mutant, but is acid-solubilized at a significant rate, although much more slowly than after infection of an ung+ host (92).As this E . coli mutant is leaky (89), the simplest interpretation of these observations is that the degradation observed is due to the residual uracil-DNA glycosylase activity. Alternatively, there may be a second,
less efficient E . coli enzyme that slowly degrades uracil-containing DNA in wiwo. It has been proposed that E . coli endonuclease V (100) may perform this role (5,6,85,96,l o o ) ,but there is presently no direct evidence available to support this hypothesis. Endonuclease V differs markedly from uracil-DNA glycosylase in having a very broad substrate specificity. It degrades single-stranded DNA without altered bases and introduces single-strand breaks at a slow rate in doublestranded DNA without lesions, and does so at an increased rate in double-stranded DNA containing dUMP residues, apurinic sites, or lesions introduced by UV light or OsOl treatment (100).A final evaluation of the role, if any, of endonuclease V in excision-repair will have to await the isolation of enzyme-defective mutant strains. A transient generation of short fragments of newly synthesized DNA due to uracil incorporation followed b y excision-repair may also occur in mammalian cells. When viral DNA synthesis was followed in a semiin witro system of isolated cell nuclei from polyoma virusinfected cells developed by Reichard and co-workers, replacement of dTTP by dUTP caused an accumulation of short nascent fragments of DNA (84).This newly synthesized DNA contained dUMP residues, but DNA replication ceased a short time after addition of dUTP. The deleterious effect of dUTP could be completely suppressed by the simultaneous addition of 6 mM uracil to the system. As free uracil is an effective inhibitor of uracil-DNA glycosylase, the results indicate that this enzyme was responsible for the fragmentation of the newly synthesized DNA (84).A transient incorporation of dUMP residues during DNA synthesis, with an accompanying accumulation of short fragments of nascent DNA, has also been observed with isolated HeLa cell nuclei (102).In that system, most of the radioactive acid-soluble niaterial released from [3H]dUMP-containingDNA shortly after replication was identified as free uracil, which suggests that uracil-DNA glycosylase initiates degradation of uracil-containing DNA in wivo in human cells (102). Similar results have recently been obtained with lysates of human lymphocytes supplemented with dUTP instead of dTTP (103).
8. BACTERIOPHAGE PBS I INDUCED INHIBITOROF URACIL-DNA GLYCOSYLASE
P B S l (and its variant PBS2) is a large €3. subtilis bacteriophage that contains uracil instead of thymine in its DNA (104).Several virusinduced proteins that alter DNA precursor metabolism have been detected, including a deoxythymidylate phosphohydrolase and an inhibitor of the B. subtilis dUTPase (105);in addition, the virus induces
162
TOMAS LINDAHL
a specific protein inhibitor of the host's uracil-DNA glycosylase. This inhibitor was first believed to interact with a nuclease specific for uracil-containing DNA (106), but Friedberg et al. (41, 75) correctly showed that the protein inhibits uracil-DNA glycosylase, and their work has now been confirmed in several other laboratories (73, 105, 107). When synthesis of virus-induced proteins is prevented by addition of chloramphenico1 or actinomycin D immediately before PBS2 infection, the viral DNA appears to be protected during the first 6 minutes after infection, but it is then rapidly degraded with release of free uracil (107).However, if the same experiment is performed with a B. subtilis mutant deficient in uracil-DNA glycosylase, degradation of the parental DNA does not occur (73). The viral inhibitor of uracilDNA glycosylase is induced as a very early function after infection, and essentially complete inhibition of the host enzyme is observed 4 minutes after infection (75). Extracts from PBSZinfected B. subtilis inhibit the uracil-DNA glycosylase activity in extracts from uninfected bacteria (75), and this observation provides an assay for the inhibitor. It is a heat-stable protein of molecular weight about 20,000 that binds tightly to uracil-DNA glycosylase, apparently in a 1: 1 complex (41, 63,106). Interestingly, the purified inhibitor not only binds to the B. subtilis enzyme, but also inhibits uracil-DNA glycosylase from mammalian cells. The natural existence of a bacteriophage with uracil instead of thymine in its DNA provides definite evidence for the notion that uracil-containing DNA can function well in replication and transcription processes. If the main role of uracil-DNA glycosylase in vivo is to counteract cytosine deamination in DNA, it may be predicted that PBSl is forced to grow under conditions favoring a high rate of G * C + A . U transitions in DNA. In this regard, it is interesting to note that PBSl DNA has an exceptionally low content (28%)of G C base-pairs. Possibly, all permissible G - C --* A U transitions have already occurred in the genome of this phage during evolution, so that a strong selective pressure against further genetic drift of this kind would now exist.
B. Hypoxa nthine-DN A GIycosy lase' A DNA glycosylase that removes deaminated adenine residues from DNA was recently discovered in E . coli extracts (108).The conversion of an adenine residue to hypoxanthine in DNA would result in a transition mutation in the absence of DNA repair, since dIMP residues would pair with dCMP rather than dTMP during replication;
DNA
GLYCOSYLASES
163
hence, hypoxanthine-DNA glycosylase may have a physiologic role similar to that of uracil-DNA glycosylase in correcting DNA deamination events.
1. MECHANISMSOF INTRODUCTION OF HYPOXANTHINE INTO DNA Adenine and guanine are spontaneously deaminated at neutral pH, but much more slowly than cytosine, and the reactions have usually been considered of little significance, In alkali, adenine is fairly readily converted to hypoxanthine ( l o g ) ,although again the rate does not approach the rapid deamination of cytosine in alkaline solution. We have recently measured the rate of deamination of adenine residues in single-stranded DNA in buffers of pH 7.4 by prolonged incubations of DNA radioactively labeled in the adenine residues at elevated temperatures, followed by chromatography of acid hydrolyzates (T. Lindahl and B. Nyberg, unpublished data). The conversion of adenine to hypoxanthine in DNA occurs at about 2% of the rate of the conversion of cytosine to uracil, and this reaction rate may well be high enough to necessitate active removal of hypoxanthine residues by a DNA repair process in vivo. It is also possible that illicit enzymic deamination of DNA adenine residues could occur, e.g., catalyzed by the enzyme that converts certain AMP residues to IMP in tRNA, but no experimental evidence is available. DNA adenine residues can be deaminated by nitrous acid at a rate similar to that of cytosine deamination (67),and this provides an easy method to convert a relatively large proportion of the adenine residues to hypoxanthine. After treatment of poly(dA-dT) with nitrous acid, replication of the polymer with E . coli DNA polymerase I required dCTP in addition to dATP and dTTP, and the amount of dCMP incorporated into DNA correlated well with the extent of adenine deamination (110). It is unclear at present if significant misincorporation of dIMP residues instead of dGMP residues into DNA occurs in vivo in a fashion analogous to the incorporation of d U M P instead of dTMP. While IMP is a key metabolite in purine nucleotide biosynthesis, E . coZi (and probably other cells as well) do not have a kinase to convert IMP to IDP, and, as a result, E . coli are essentially devoid of a pool of ITP ( 1 1 1 ) . If IDP occurred in cells, incorporation of hypoxanthine in DNA would presumably take place, since IDP could be reduced to dIDP, phosphorylated to dITP, and used as a precursor by a DNA polymerase. Consequently, it appears to be of critical importance to avoid phosphorylation of IMP in uivo. Similarly, no kinase seems to exist for the conversion of XMP to XDP (U. Lagerkvist, personal communication). The presence of several different nucleoside
164
TOMAS LINDAHL
monophosphate kinases with restricted substrate specificities may thus be obligatory in order to prevent the incorporation of hypoxanthine and xanthine into nucleic acids, while a much less specific kinase is acceptable for the conversion of nucleoside diphosphates to triphosphates. The fate of nascent DNA containing incorporated dIMP residues can be investigated by the addition of dITP to permeabilized cells, or perhaps by the isolation of bacterial mutants with nucleoside monophosphate kinases of reduced substrate specificity. Nuclei from polyoma-virus-infected mouse cells readily incorporate dIMP into DNA when supplied with dITP, but hypoxanthine is not released in the same rapid fashion as uracil from the nascent DNA in this system (P. Reichard, personal communication). However, addition of dITP to E . coli lysates that replicate DNA in vitro leads to a decrease in size of the newly synthesized DNA chains, so, in this case, the effects of dITP and dUTP appear similar ( I l l a ) . 2. PR0PERTE.S OF HYPOXANTHINE-DNA GLYCOSYLASE The hypoxanthine-DNA glycosylase activity of E . coli crude cell extracts is much lower than the uracil-DNA glycosylase activity, and the hypoxanthine-releasing enzyme has so far been only partly purified (108).It is a small enzyme ( M , 30,000) that does not require Mg2+,phosphate, or other cofactors for activity. While it resembles the previously known DNA glycosylases in these respects, it is clearly a separate enzyme, as it has different fractionation properties from either uracil-DNA glycosylase or 3-methyladenine-DNA glycosylase. Further, highly purified preparations of E . coli uracil-DNA glycosylase or 3-methyladenine-DNA glycosylase are unable to release hypoxanthine from DNA, and extracts of E . coli ung- mutants and tag- mutants [deficient in 3-MeAde-DNA glycosylase] have normal levels of hypoxanthine-DNA glycosylase activity. The hypoxanthine-DNA glycosylase is more sensitive to inhibition by neutral salts and unmodified DNA than the other DNA glycosylases, and these properties together with the apparently intrinsically low amount of enzyme activity initially made the enzyme rather difficult 'to detect. It has now been approximately 200-fold purified from extracts by removal of DNA by polyethylene glycol treatment in 1 M NaCl, followed by stepwise hydroxyapatite chromatography to remove the polyethylene glycol, ammonium sulfate fractionation, gel filtration, and DNA-cellulose chromatography. The partly purified enzyme releases hypoxanthine, but not xanthine, adenine, or guanine from nitrous acid-treated DNA. The best substrate detected so far is the
-
double-stranded polydeoxynucleotide poly(dA,[3H]dI) . poly(dT), as radioactive hypoxanthine is released 10 times faster from this polymer than from p~ly(dA,[~H]dI) without a complementary poly(dT) strand. However, no release of radioactive material in low-molecular-weight * poly(dT) in parallel experiform was obtained from p~ly(dA,[~H]dC) ments, so the enzyme does not act by removing any mismatched base residue, Further, DNA purine residues with other alterations in the 6 position are not substrates. No release of Ofi-methylguaninefrom N methyl-N-nitrosourea-treated DNA, or of Ng-methyladenine from enzymically methylated DNA could be detected under reaction conditions that yielded effective enzymic liberation of hypoxanthine from deaminated DNA, so the hypoxanthine-DNA glycosylase appears to be highly specific for deaminated adenine residues. The definition of the physiological role of hypoxanthine-DNA glycosylase will have to await the isolation of enzyme-deficient bacterial mutants, although it may well be a DNA repair enzyme acting in an analogous fashion to uracil-DNA glycosylase. The mutagenic effect of deamination of adenine in bacteriophage T4 DNA in vivo appears to be due to error-prone DNA repair, involving DNA ligase, rather than to mismatching of base-pairs during DNA replication, while 2-aminopurine acts as a mutagen by the latter mechanism (112).These observations imply that DNA repair processes may exist to handle deaminated adenine residues.
C , 3-Met hyladen ine- DNA GIycosy Iase' Several different forms of damage are introduced into DNA by treatment with alkylating agents, and some of these lesions are removed by DNA excision-repair processes. One of the major products obtained after exposure to methylating or ethylating agents is adenine alkylated at the N-3 position, and this particular lesion has been shown to be removed by a DNA glycosylase in E . coli (36, 89, 113). Thus, DNA glycosylases exist that act on other types of lesions than deaminated base residues.
1. MECHANISMSOF INTRODUCTION OF %METHYLADENINE INTO DNA The literature on nucleic acid alkylation has been reviewed both in and only a brief account wiIl be this series7p8and elsewhere (48,114), given here. Alkylation may occur at many different sites in DNA, and Lawley, this series, Vol. 5. Singer, this series, Vol. 15.
166
TOMAS LINDAHL
the rate of alkylation observed at a certain position depends to a large extent on the nature of the alkylating agent. Methylating agents may be subdivided into two classes without a distinct borderline, depending on whether they seem to act primarily by a bimolecular mechanism with formation of a transition complex with the nucleophile (s&) or by a unimolecular mechanism (&I) in which the formation of a reactive ion would be the rate-limiting step. The sN2 reagents show a strong preference for the most nucleophilic sites while sN1 reagents attack all nucleophiles and consequently yield a broader variety of products. Typical sN2 reagents are dimethyl sulfate and methyl methanesulfonate, which react with DNA to give two major products, 7-methylguanine and &methyladenine, two minor products, 7-methyladenine and 3-methylguanine, and trace quantities of several other species that are more effectively obtained by treatment with sN1 reagents. Typical sN1 reagents are N-methyl-N-nitrosourea and N-methyl-N'-nitro-N-nitrosoguanidine,which act through the methyldiazonium ion CH,N2+. In addition to the lesions also obtained with sN2 reagents, the sN1 reagents yield phosphotriesters and Osmethylguanine as major products and several types of alkylated pyrimidines as minor products. Slightly higher amounts of 3-methyladenine are obtained with sN2 than with s N 1 reagents, while the latter reagents are much more effective as mutagenic and carcinogenic agents. Consequently, 3-methyladenine does not appear to belong to the most important lesions with regard to alkylation mutagenesis. Products analogous to those obtained with methylating agents are obtained with ethylating agents, such as ethyl methanesulfonate or N-ethyl-N-nitrosourea.* On the other hand, agents giving bulky substituents react preferentially at other sites in DNA. Thus, alkylation occurs at the amino with 7-bromomethyl-12-methylbenz[a]anthracene groups of guanine and of adenine to yield N2-(12-methylbenz[a]anthracen-7-ylmethy1)guanine and Nfi-(12-niethylbenz[u]anthracen-7ylmethyl) adenine as products (115). The methyl groups of 3-methyladenine and 3-methylguanine are located in the minor groove of the DNA double helix, while the methyl groups of 7--methylguanine, 7--methyladenine, and 06methylguanine are situated in the major groove. The only one of these lesions that would clearly interfere with the hydrogen bonding of regular base-pairing is 06-methylguanine, and this is also the only product clearly implicated in mutagenesis. While there is no obvious reason to conclude that 3-methyladenine would induce miscoding, it has been suggested (116) that the presence of unsubstituted N-3 atoms of purines may be essential for the template function of DNA, and that
methylation at N-3 might interfere with the action of an enzyme binding in the narrow groove of the DNA helix. This hypothesis appears to predict correctly several of the properties of E . coli mutants deficient in 3-methyladenine-DNA glycosylase (see below), but sufficient data on the locations of the DNA binding sites of DNA polymerases, other protein factors involved in DNA replication, and RNA polymerases are not yet at hand to permit a more detailed evaluation of this interesting idea. In addition to the introduction of methyl groups at the N-3 position of DNA-adenine residues by treatment with alkylating agents, the reaction could conceivably also occur as an enzyme-catalyzed event. While Smethyladenine does not seem to occur naturally in tRNA or other nucleic acids, an enzyme that catalyzes the conversion of free adenine to 3-methyladenine in an S-adenosylmethionine-dependent reaction has been found in mammalian cells (117).An occasional illicit methylation of DNA by an enzyme of this type would nicely explain the need for a repair system to remove 3-inethyladenine from DNA rapidly, but there is presently no evidence for the occurrence of enzymic niethylation of DNA in this position under in vivo conditions.
2. ACTIVE REMOVALOF 3-METHYLADENINE FROM DNA The fates i n vivo of the usual alkylation products in DNA vary markedly. The glycosyl bonds of DNA purine mononucleotide residues are greatly labilized by niethylatioii in the N-3 or N-7 position, so the alkylated bases are spontaneously released by nonenzymic hydroIysis in 2 2 0 0 hours at 37°C and pH 7.4(1 16). Three of the common alkylated derivatives do not seem to be actively excised and have similar half-lives in DNA either in vivo or in uitro: 7-methylguanine (116, 118), 7-methyladenine (116), and phosphotriesters (119). On the other hand, &methyladenine is released very rapidly from DNA in vivo in an enzyme-catalyzed process (113, 116, 118, 120). Thus, the initial elimination rate of Smethyladenine from E. coli DNA in vivo is unknown, because after exposure of E . coli cells for a few minutes at 37°C to an alkylating agent such as dimethyl sulfate, most of the 3-methyladenine residues initially formed in DNA have already been released during the period of alkylation (113, 116). In rat liver DNA, the half-life of Smethyladenine residues in vivo is 3 hours, which is an elimination rate 8-10 times faster than that due to nonenzymic release of 3-methyladenine from' purified DNA by hydrolysis under similar external conditions (120). 06-Methylguanine (113, 118) and 3-methylguanine (116) are also actively released froni DNA in vivo,
168
TOMAS LINDAHL
but they are liberated considerably more slowly than 3-methyladenine. Similar results have been obtained with the ethylated derivatives. 3-Ethyladenine is very rapidly released from E . coli whereas 7-ethylguanine is not actively excised (121), and 06ethylguanine is slowly but actively released. The rate of release of the latter base from mammalian DNA by excision-repair varies between different organs in an inverse fashion with the carcinogenic effect of the alkylation treatment (122). The bulky adducts obtained by treatment with 7-bromomethylbenz[a]anthracene or similar agents are actively removed from DNA in vivo (123), but this excision differs from that of methylated or ethylated derivatives in being dependent on the uvr gene products in E . coli (16).
3. OCCURRENCE OF 3-METHYLADENINE-DNA GLYCOSYLASE Evidence for the repair of DNA alkylation damage by excisionrepair processes was obtained in the early work of Strauss’ group, and they further showed that crude cell extracts from B . subtilis and M . luteus specifically catalyze the formation of strand breaks at some, but not all, alkylated sites in methyl methanesulfonate-treated DNA (124-126). An E. coli activity, termed endonuclease 11, selectively introduces chain breaks in alkylated DNA (127).Moreover, E. coli cell extracts incise alkylated DNA at Smethyladenine but not at 7-methylguanine residues (128).All these studies, which were interpreted to reflect the existence of endonucleases specific for alkylated DNA, were performed before the discovery of DNA glycosylases, and in retrospect it seems likeIy that the reactions observed were due to the concerted action of $methyladenine-DNA glycosylase and an AP endonuclease. In subsequent studies on the endonuclease I1 activity, it was in fact found that the enzyme fraction could incise at apurinic sites in DNA (129) as well as at unknown sites in y-irradiated DNA (130), and that the preparation would also release Smethyladenine, 06-methylguanine, N6-( 12-rnethylbenz[a]anthracen-7-ylmethyl) adenine, and N2-(12-rnethylbenz[a]anthracen-7-ylmethyl) guanine in free form from alkylated DNA (40, 131). While the existence of the latter three DNA glycosylase activities has not been confirmed in other laboratories, it seems likely that the E . coli endonuclease I1 activity may be ascribed to several enzymes, including Smethyladenine-DNA glycosylase. It has been unfortunate that the many interesting activities first observed in endonuclease I1 preparations were erroneously interpreted as functions of a single enzyme. The E . coli DNA glycosylase activity that releases 3-methyl-
DNA
GLYCOSYLASES
169
adenine from alkylated DNA is different from either uracil-DNA glycosylase or an AP endonuclease, as the three types of activities have different fractionation properties and heat labilities (36). The E . coli %methyladenine-DNA glycosylase has recently been purified and characterized in detail (89). A 3-methyladenine-DNA glycosylase with properties very similar to those of the E . coli enzyme has also been found in M . luteus (132).Further, a SmethyladenineDNA glycosylase is present in human lymphocytes (21; T. P. Brent, personal communication). While the occurrence of 3-methyladenineDNA glycosylase in different types of cells has not been screened in the same systematic fashion as for uracil-DNA glycosylase, it nevertheless appears that both these enzymes are widely distributed, and that 3-methyladenine-DNA glycosylase occurs both in bacteria and in mammalian cells.
4. PROPERTIES OF 3-METHYLADENINE-DNA GLYCOSYLASE E. coli Smethyladenine-DNA glycosylase has been purified 2800fold, but is not available in a homogeneous state (89).The amount of enzyme activity in cell extracts is considerably lower than for uracilDNA glycosylase. The purification procedure for the enzyme involves streptomycin treatment, ammonium sulfate fractionation, gel filtration, phosphocellulose chromatography, and DNA-cellulose chromatography. The native enzyme has a molecular weight of about 20,000, as estimated from its sedimentation coefficient and Stokes radius. It is a relatively heat-labile enzyme with a half-life of about 10 minutes at 45" in several buffers, and it is also easily partly inactivated during purification. 3-Methyladenine-DNA glycosylase is most conveniently measured by its ability to release ethanol-soluble radioactive material from [3H]dimethyl-sulfate-treatedDNA. Acid precipitation of the alkylated DNA cannot be used, as it leads to rapid release of 3-methyladenine by nonenzymic hydrolysis. The enzymically liberated material has been identified as Smethyladenine in several chromatographic systems. The enzyme has no obligatory cofactor requirements, and it is helpful to assay the enzyme in EDTA-containing reaction mixtures during purification in order to avoid interference by nucleases. However, the purified enzyme is slightly stimulated (about 30%) by the addition of 5-15 mM MgCll to reaction mixtures. It has a pH optimum at 7 . 2 7 . 8 and is inhibited by N-ethylmaleimide or p-mercuribenzoate. Further, it shows a strong preference for double-stranded substrates, and markedly differs from uracil-DNA glycosylase in this regard. On incubation of 3-methyladenine-DNA glycosylase with di-
170
TOMAS LINDAHL
inethyl sulfate- or N-methyl-N-nitrosourea-treated DNA, practically all Smethyladenine residues are released, but 7-methylguanine, 7-methyladenine, and 06-methylguanine residues are not liberated. $Methylguanine also does not seem to be released, but a very slow excision of this minor alkylation product has not been ruled out. Analogous results have been obtained with N-ethyl-N-nitrosourea-treated DNA, as 3-ethyladenine, but not 7-ethylguanine or 06-ethylguanine, was enzymicall y liberated. Thus, the enzyme releases 3alkylated adenine having either a methyl or an ethyl group. No detectable excision of N + ( 12-methylbenz[a]anthracen-7-ylmethy1)guanine or Nfi(l&methylbenz[a]anthracen-7-ylrnethyl) adenine from DNA treated with 7-bromomethyl-l2-methylbenz[a]anthraceneis catalyzed by 3-methyladenine-DNA glycosylase. Moreover, there is no release of the naturally occurring N6-methyladenine, adenine, or guanine from DNA, nor can the enzyme liberate pyrimidine dimers from ultraviolet-irradiated DNA, or hypoxanthine or xanthine from nitrous-acid-treated DNA. E . coli $methyladenine-DNA glycosylase does not introduce chain breaks into alkylated, irradiated, or intact DNA, but simultaneously with the release of free 3-methyladenine from alkylated DNA, apurinic sites are introduced into the DNA; these have been identified by their sensitivity to alkali and to AP endonucleases. These results identify the enzyme as a DNA glycosylase. Similar data have also been obtained for the M . luteus (132)and human (21)enzymes. While the amount of &methyladenine-DNA glycosylase activity in E. coli extracts appears to be low, the enzyme has a low K, value for $methyl-dAMP residues in DNA, 6 x M, so it might well be able to excise Smethyladenine in an effective fashion in uiuo. The enzyme acts in a hydrolytic, not in a phosphorolytic, fashion, as there is no phosphate dependence or incorporation of 32Pi into DNA during catalysis. Further, the enzyme is product-inhibited by free Smethyladenine, with an apparent K , close to lC3M, while higher concentrations of 7-methylguanine, 7-methyladenine, 3-methylguanine, 06-methylguanine, adenine, caffeine, or uracil yield no significant inhibition. The properties of this enzyme with regard to its mechanism of action and product inhibition closely resemble those of uracil-DNA glycosylase.
5. BACTERIAL MUTANTS DEFICIENTIN 3-METHYLADENINE-DNA GLYCOSYLASE It is somewhat surprising that a widely distributed DNA glycosylase appears to have as its sole function the removal of an
alkylation product. It is far from clear that living cells have been so regularly exposed to alkylating agents during evolution that it became necessary to develop a repair mechanism to remove 3-methyladenine from DNA. For these reasons, it has been of interest to isolate E . coli mutants deficient in 3-methyladenine-DNA glycosylase in order to establish the physiological roles of the enzyme. Two independent mutants have now been found (113) and termed tag- (threemethyladenine-DNA glycosylase) mutants. The tug- mutants were initially isolated as double mutants from E. coli x t h - strains, which are deficient in exonuclease 111.E . coli x t h - mutants are slightly sensitive to methyl methanesulfonate. The xth-tug- double mutants are much more sensitive to methyl methanesulfonate than x t h single mutants, while the xth-tag- mutants show normal resistance to ultraviolet light, X-irradiation, or nitrous acid treatment. Recently, an xth+tag- mutant that has retained a marked sensitivity to methyl methanesulfonate has been obtained. The high sensitivity of xth-tug- double mutants to the alkylating agent may in fact be ascribed to the tag mutation, because six revertants of xth-tag- strains to xth-tag+ had simultaneously regained the low sensitivity to methyl methanesulfonate and 'normal 3-methyladenine-DNA glycosylase levels characteristic ofE. coli x t h strains. Extracts from tag- mutants have markedly decreased levels of 3-methyladenine-DNA glycosylase activity compared to wild-type strains, but it is presently unclear whether any of the mutants so far isolated is totally deficient in the enzyme. One xth-tag- mutant strain, PK432, has a thermolabile enzyme and is much more sensitive to methyl methanesulfonate at 43°C than at 30"C, so the t a g - mutation in this strain is clearly placed in the structural gene for Smethyladenine-DNA glycosylase. Another xth-tag- mutant strain, BK2012, is tighter, and the strain is highly sensitive to methyl methanesulfonate at both 30°C and 43°C. Preliminary mapping data have placed the tag mutation at 47-49 minutes on the E . coli K 1 2 genetic map (90, 113). The tag- mutation does not appear to lead to an increased spontaneous mutation frequency inE. coli. However, the high sensitivity of tag- mutants to alkylation clearly identifies 3-methyladenine as a lesion having a major contribution to lethality in methylniethanesulfonate-treatedE. coli when repair of the lesion is impaired. It is not yet settled if 3-methyladenine should be regarded as an inactivating rather than a mutagenic lesion. The active elimination of Smethyladenine from alkylated DNA in uivo is strongly suppressed in tag- mutants. Thus, while a wild-type E.
172
TOMAS LINDAHL
coli strain, or an xth-tag+ strain removes 7 0 4 0 % of the %methyladenine from its DNA after treatment with dimethyl sulfate in about 5 minutes at 3O"C,an xth-tag- double mutant retains 5 6 7 0 % of its %methyladenine for 45 minutes under the same external conditions (113).These experiments have been performed with low, nonlethal concentrations of dimethyl sulfate. In similar experiments with N-methyl-N-nitrosourea-treated cells, while the rate of 3methyladenine release was again strongly reduced, 06-methylguanine was liberated at a normal rate in an xth-tag- mutant. These results are in good agreement with the substrate specificity of the purified E. coli %methyladenine-DNA glycosylase, which is unable to release 06-methylguanine from alkylated DNA.
D. Additional DNA Glycosylases The DNA glycosylases found to date are small enzymes, and each one only seems to recognize one particular form of damage. Obviously, there cannot be one DNA glycosylase for every conceivable type of DNA lesion, so this class of enzymes may well turn out to be confined to the repair of just a few important lesions in DNA. A probable estimate is that less than 10 different types of DNA glycosylases active in DNA repair will occur. At this point, it is too early to speculate on the possible existence of DNA glycosylases involved in other aspects of DNA metabolism than repair. However, it seems unlikely that the three enzymes found to date are the only representatives of this class, and a search for additional DNA glycosylases is presently being carried out in several laboratories. As two of the three enzymes 'identified so far may primarily be active on spontaneous lesions in DNA, while the third one appears to correct alkylation damage, it is clearly possible that additional lesions due to spontaneous hydrolysis or alkylation may be susceptible to DNA glycosylases. One obvious candidate is xanthine (deaminated guanine), as both deaminated cytosine and deaminated adenine can be removed by separate DNA glycosylases. Another possibility is that 06-methylguanine is released by a DNA glycosylase, as an E. coli activity present in an endonuclease I1 preparation has already been reported to release 06methylguanine in free form from alkylated DNA (40). The release of Os-methylguanine is not dependent on the xth, tag, uurA, or uurB genes (113; B. Straws, personal communication), so the mechanism of its liberation is presently not understood. However, it seems to differ from the release of other alkylation lesions, as an increased ability to repair O6-methylgumine can apparently be induced by pretreatment ofE. coli with low concentrations of alkylating agents (133).Additional
DNA GLYCOSYLASES
173
lesions that may be actively removed by DNA glycosylases include several of the base lesions introduced by exposure to ionizing radiation, and several “minor” lesions introduced by ultraviolet light.
E. Enzymic Hydrolysis of Nucleosides, Mononucleotides, and Related Compounds
Cleavage of nucleosides usually takes place by phosphorolysis in uiuo, but hydrolysis can occur as an additional or alternative pathway of degradation. Several enzymes that catalyze the hydrolysis of nucleosides were described in the 1950s, but relatively little work has been performed on this group of enzymes during the last 20 years. The irreversible hydrolytic cleavage of nucleosides clearly is an enzymic reaction related to the release of free bases from DNA by DNA glycosylases. However, in contrast to DNA glycosyhses, the nucleosidases usually show broad substrate specificity,e.g., a LactobacilZus enzyme that hydrolyzes inosine to hypoxanthine and ribose also degrades many other ribonucleosides (134). The best-characterized nucleosidase is the uridine nucleosidase (EC 3.2.2.3)’from yeast (135), which was recently purified to hoinogeneity (136, 137). Uridine nucleosidase does not require phosphate or Mg’+ for activity, and it cleaves uridine and ribosylthymine, but not other nucleosides such as cyti-dine or deoxyribosylthymine. The native enzyme is a coppercontaining protein that is inhibited by EDTA; it has a molecular weight of 32,500 and contains two apparently identical subunits. With regard to these latter properties, uridine nucleosidase seems quite different from the DNA glycosylases. Enzymes that hydrolyze mononucleotides have also been found. The degradation of AMP to adenine and ribose 5phosphate is catalyzed by an enzyme from Azotobacter vinelandii (138), and a Streptomyces enzyme hydrolyzes several pyrimidine nucleotides (139). Further, the enzymic release of nicotinamide from NAD (140, 141), adenine from S-adenosyl-L-homocysteine (142) and l-methyladenine from l-methyladenosine (143) by hydrolysis of the substrate have been described.
IV. Endonucleases for Apurinic/Apyrimidinic Sites (AP E ndonuc Ie a ses) A. Depurination and Depyrimidination of D N A
The properties of apurinic and apyrimidinic sites in DNA have been reviewed (144). It is interesting that apurinic sites in DNA,
174
TOMAS LINDAHL
which are lesions that cause loss of genetic information, rarely if ever seem to cause mutations in DNA (3).A small amount of miscoding has been observed in an in vitro system with a heavily depurinated DNA template (145), but in spite of these latter observations it seems likely that an apurinic site should be. regarded as a potentially inactivating rather than mutagenic lesion. This notion is supported by the fate of phage T7 containing apurinic sites in their DNA (146) and by the poor growth of E . coli dut-xth- double mutants, which apparently accumulate apyrimidinic sites in their DNA (98). The existence of DNA glycosylases obviously leads to rapid depurination or depyrimidination of DNA under certain in vivo conditions, but in addition apurinic/apyrimidinic sites may be introduced nonenzymically, and these processes are summarized here.
1. SPONTANEOUSBASE RELEASE The base-sugar bonds in DNA are susceptible to acid hydrolysis, and standard methods of DNA base analysis have involved liberation of the pyrimidine bases by strong acid treatment, while the purine bases may be selectively released in weak acid. Greer and Zamenhof found in 1962 that purine bases are also slowly released from DNA during incubation at neutral pH and high temperatures (147). Further investigations of the rate of DNA depurination at neutral pH were carried out 10 years later, employing more sensitive techniques, and it could be shown directly that spontaneous depurination of doublestranded DNA occurs at temperatures far below the t , of DNA (148). There is only a 4fold difference in the rate of depurination between single-stranded and double-stranded DNA, so the double-stranded conformation confers relatively little protection against hydrolysis of the glycosyl bonds. This is not particularly surprising, as many bound water molecules are present in the grooves of the DNA double helix, which is a highly hydrated structure in vivo. Guanine and adenine are released at similar rates, with guanine being lost slightly faster. From a series of measurements on the hydrolytic liberation of guanine and adenine from DNA radioactively labeled in the purine residues, it was concluded that the rate of DNA depurination at pH 7.4 and ionic strength 0.15 is about k = 4 x sec-' at 70°C and that the activation energy is 130 kJ/mol(148).The rate of DNA depurination at 37OC may therefore be estimated to be close to k = 3 x lo-" sec-I. These data correspond to the loss of one purine residue every second generation from an E . coli cell growing with a generation time of 40 minutes at 37"C, and this appears to put a rather small load on cellular DNA repair processes. In comparison, wild-type E. coli cells have the ca-
pacity to remove many hundred pyrimidine dimers from DNA per generation (17 ) . DNA depurination becomes a larger problem for therniophilic bacteria, as a Bacillus steurothermophilus cell growing at 70°C would lose about 50 purines from its DNA per generation, whereas aThermus themophilus cell growing at 85°C would lose about 300 purines per generation. Repair of several kinds of spontaneous hydrolytic DNA damage in thermophiles must be extensive, and this subject has received too little attention in studies on such microorganisms. Mammalian cells would lose many purines from their DNA in each cell generation because they grow more slowly and have much larger genomes than bacteria. Thus, an actively growing human cell may be estimated to lose about 10,000 purines from its DNA in each generation. This number may at first sight appear high, but there is now clear evidence froin work with cells exposed to alkylating agents such as methyl methanesulfonate that mammalian cells can effectively handle a much higher load of DNA depurination. The introduction of large quantities of the major alkylation product 7-methylguanine into the DNA of mammalian cells is surprisingly well tolerated from the point of view that these residues are relatively rapidly released by nonenzymic hydrolysis, and approximately 106 apurinic sites per cell generation introduced in this fashion elin be well tolerated (114,120), ie., a depurination rate 100 times higher than the estimate for spontaneous depurination in normal cells not exposed to alkylating agents. These data strongly imply that effective, essentially error-free repair systems exist to cope with apurinic sites. The mechanism of hydrolytic DNA depurination at neutral pH apparently involves protonation of the base followed by direct cleavage of the glycosyl bond, which is the same mechanism as that established for acid hydrolysis of deoxyiiucleosides (148,149). It should be noted here that DNA depurination is also an alkali-catalyzed process (66, 150),but no mechanistic studies on this relatively slow reaction have been performed. Depyrimidination of DNA occurs in an analogous fashion to depurination at neutral pH, albeit at a 20 times slower rate. Cytosine and thymine are liberated in similar quantities (151).This means that a growing mammalian cell would lose several hundred pyrimidine residues per cell generation. Further, the naturally occurring methylated bases N6-methyladenine and Smethylcytosine are released at rates similar to those of the corresponding unmethylated bases (148,151 ). The heat- and acid-induced hydrolytic release of purines from DNA has been a convenient method to introduce apurinic sites into
176
TOMAS LINDAHL
double-stranded DNA with a minimum of side reactions. In a typical protocol, bacteriophage PM2 DNA (M = 6.4 x 10s) has been treated in 0.1 M NaCI/O.Ol M sodium citrate, pH 5.0, for 5 minutes at 70°C to introduce approximately 1 apurinic site per DNA molecule (152). 2. DEPURINATION OF ALKYLATEDDNA DNA purine residues niethylated at the N-3 or N-7 positions are positively charged, and such alkylated purines are relatively rapidly released by hydrolysis because the charged purine residue becomes a better leaving group than an uncharged purine (149).Thus, there is no fundamental mechanistic difference between the release of an alkylated purine such as 7-methylguanine us unmodified guanine. The common methylated purine derivatives are released in the order 7-methyladenine > Smethyladenine > 7-methylguanine > 3methylguanine + 06-methylguanine(116).As the hydrolytic release of 7-methylguanine from DNA is rapid, and the base can be specifically labeled by employing a radioactive alkylating agent, it has been possible to show that the rates of release of 7-methylguanine from DNA are practically the same in viuo as in vitro at pH 7.4 (116, 118, 120). It seems very likely that the same situation holds true for the hydrolytic release of unmodified purines, although for technical reasons it has not been possible to measure this slow reaction in uiuo. As discussed above, 3-methyladenine is enzymically removed in viuo, and this reaction further increases the initial rate of total depurination of alkylated DNA in living cells. 3. DEPYRIMIDINATION BY IONIZING RADIATION X-irradiation introduces several types of alkali-labile lesions in DNA, some of which are susceptible to AP endonucleases (153).DNA glycosyl bonds can apparently be cleaved by direct hydroxyl radical attack, but a more important type of radiation-induced damage is probably the alteration and destruction of pyrimidine residues in DNA, since some saturated forms of pyrimidines are very easily lost by hydrolysis (154,155). Further, hydroxyl attack at a pyrimidine may result in ring opening and fragmentation, and in this fashion only a remnant of a pyrimidine residue will often be left in DNA. Radiolysis of thymine residues can lead to DNA containing thymine glycol or fonnamide-, N-formylureido- or ureido-deoxyribose in DNA, and similar reactions presumably occur for cytosine. While it has been noted that these defects may be as important in radiobiology as thymine dimers are in photobiology (155), there is little information available on the alkali-lability or sensitivity to AP endonucleases of these various lesions.
DNA GLYCOSYLASES
177
It is difficult to obtain accurate estimates of the numbers of apurinic and apyrimidinic sites introduced by ionizing radiation by subsequent alkali treatment, and the same problem also occurs with alkylated DNA. The complication is that alkaline hydrolysis of apurinic sites requires a certain time to go to completion, and in the meantime altered base residues appear to be further degraded in alkali and yield new sites, so that a gradual increase of the total number of sensitive sites occurs with time. A better procedure is therefore to quantitate the number of AP sites as chain breaks introduced by treatment with an A P endonuclease at neutral pH (156, 157). Apyrimidinic sites are also introduced as secondary lesions after exposure of DNA to high doses of ultraviolet light, presumably due to the spontaneous release of oxidized pyrimidine derivatives (153).It is also noted here that the fungal peptide antibiotic bleomycin binds to DNA and promotes cleavage of thymine-deoxyribose bonds, thereby generating apyrimidinic sites in DNA.g
4. CHAIN BREAKAGEAT APUFUNIC/APYRIMIDINICSITES The deoxyribose residue at an apurinic or apyrimidinic site in DNA will exist in equilibrium between the furanose form and the aldehyde form. In alkaline solution, a p-elimination reaction consequently occurs in which the phosphate is eliminated from the 3’ position of an adjacent deoxyribose residue in its aldehyde form, so a chain break results at the 3’ side of the lesion. The same reaction also proceeds at a reduced rate at neutral pH. In buffered physiological saline, a DNA apurinic/apyrimidinic site has an average lifetime of about 400 hours at 37°C and pH 7.4 (152).The presence of M e + and primary amines promote the cleavage (158),and in their presence the average lifetime is reduced to 1OG200 hours. Polyamines further promote the rate of chain cleavage (152).However, it seems clear that the DNA chain at an apurinic or apyrimidinic site in DNA would be sufficiently stable to survive several generations in actively growing cells in the absence of DNA repair. The alkali-induced cleavage of apurinic/apyrimidinic sites in DNA requires relatively prolonged treatment at high pH in order to approach completion. In 1 M glycine-NaOH, pH 12.8, a DNA apurillic site has a half-life of about 30 minutes at 25T, and incubation under these conditions for 3-4 hours of DNA containing apurinic sites introduced by incubation at pH 5 leads to cleavage at such sites with no significant chain cleavage at other locations (152).However, with irradiated or alkylated DNA a shorter alkali treatment is preferable in Muller and Zahn, this series, Vol. 20.
178
TOMAS LINDAHL
order to avoid alkali-catalyzed further modifications of damaged bases (157). In addition to chain breaks, crosslinks may occur as secondary lesions at DNA apurinic sites. While the introduction of DNA crosslinks in this fashion has been demonstrated at low p H (158a),the reaction is a very minor one at neutral pH and appears to be of little significance.
B. AP
Endonucleases with Associated Exonuclease Activity
1. E. coli EXONUCLEASE I11 Although a DNA chain is relatively stable at an apurinic site at neutral pH, incubation of partly depurinated DNA with cell extracts leads to very rapid enzymic cleavage at these sites (152,159). In early studies, an activity in E . coli endonuclease I1 preparations was found to catalyze chain breakage at apurinic sites in DNA (129), but the enzyme fraction also contained several other activities, and the first clear evidence for an E . coli endonuclease that would specifically incise DNA at apurinic sites was obtained by Verly et al. (159, 160). They used a preparation procedure similar to that devised for endonuclease I1 (127),but were able to demonstrate that their purified enzyme did not attack intact or alkylated DNA. When alkylated DNA was incubated at neutral pH to effect the release of alkylated purines, the DNA was simultaneously converted to a substrate for the endonuclease, which only incised the DNA at apurinic sites. The enzyme, which accounts for most of the endonuclease activity at DNA apurinic sites by E . coli extracts, has subsequently been purified to homogeneity and shown to have a monomeric structure (161). In studies on E . coli mutants defective in either exonuclease I11 (EC 3.1.4.27) (162) or the major endonuclease for apurinic sites, it was found that these apparently very different activities are due to the same enzyme (37,163).A large body of biochemical and genetic data has since then been accumulated to confirm this finding (98,164,165), and it is now beyond dispute that the two kinds of activities are due to the same monomeric protein. Thus, recent reports (166, 167) that exonuclease I11 can be isolated free from associated endonuclease activity at apurinic sites, and that mutants deficient in both activities might be double mutants, are clearly in error. When exonuclease I11 is purified by standard procedures, an AP endonuclease activity cochromatographs exactly with the exonuclease activity, and both activities are present in the physically homogeneous enzyme (37,165). Mutants with a thennolabile AP endonuclease activity also have a
thermolabile exonuclease I11 activity (163). Further, E . coli mutants isolated as deficient in one or the other of these two enzyme activities are always simultaneously deficient in the other enzyme activity, the two activities exhibit complete cotransduction frequencies, and a revertant had simultaneously regained both activities (98,163,164). The nomenclature for the apurinic/apyrimidinic endonuclease activity of E . coli exonuclease I11 has been somewhat confused. It has sometimes been called “apurinic endonuclease,” but as this enzyme as well as all other endonucleases acting at apurinic sites also cleave at apyrimidinic sites, we have replaced this term with the abbreviation “AP endonuclease.” Since the major activity in early preparations of endonuclease I1 appeared to be that of AP endonuclease, the term endonuclease I1 was occasionally used a few years ago to describe the AP endonuclease activity of exonuclease 111 (36, 37). This use of the term should be discouraged, however, because endonuclease I1 has recently been redefined as an activity on alkylated DNA (166). While this definition still does not clarify the chemical reaction(s) that endonuclease I1 activity is supposed to catalyze, this activity should not presently be regarded as an AP endonuclease. It is not known if endonuclease I1 activity is due to one or several enzymes, and if contaminating DNA glycosylases are obligatory for activity. Gossard and Verly (39) have shown that highly purified preparations of their AP endonuclease contain an exonuclease III-like activity, and it may be concluded that the enzyme investigated by Verly’s group is very likely identical with exonuclease 111. However, they have recently preferred to use the term “endonuclease VI” to describe their endonuclease activity (39). While this term is shorter than the somewhat clumsy expression “AP endonuclease activity of exonuclease 111,” the latter nomenclature is preferred here, as both activities are integral parts of the same monomeric protein. This is in line with Kornberg’s suggestion that the exonuclease activities ofE. coli DNA polymerase I should be termed “the 5’+ 3’ us 3 ’ 4 5‘ nuclease activities of DNA polymerase I,” whereas the older terms “exonuclease VI” and “exonuclease 11” should be abandoned (32). E . coZi exonuclease I11 is a monomeric protein of molecular weight 28,000 that has a pH optimum at 8.0-8.5. It is quite heat-labile, with a half-life of about 2 minutes at 45°C in common buffers, and is inhibited by EDTA. This rather small enzyme possesses no fewer than 4 distinct activities: (i) a 3‘ + 5’ exonuclease activity on doublestranded DNA, (ii) an AP endonuclease activity, (iii) an RNase H activity, and (iv) a DNA 3‘-phosphatase activity. The DNA exonuclease activity occurs mainly in a nonprocessive fashion with frequent dis-
180
TOMAS LINDAHL
sociation of the enzyme from the substrate during the course of digestion (168),but exonuclease action is more processive at low temper. the AP endonuclease and exonuclease I11 atures ( 1 6 8 ~ )Further, activities exhibit similar rates (98, 169). Weiss (37) has proposed an ingenious model to explain how these diverse enzyme activities might share a single catalytic site. The model is shown in Fig. 3 for three of the activities. It is assumed that “the exonuclease recognizes a space created by the unwinding of a terminal base-pair that occurs because of reduced base-stacking forces at the ends of duplexes” (37).The idea that the enzyme might recognize a space created by the displacement or removal of a base would nicely account also for the 3’-phosphatase and AP-endonuclease activities of the enzyme. In order to explain the RNase H activity it is necessary to predict that the enzyme domain that recognizes duplex structure requires a chain containing deoxyribose, while the active site may interact with either a DNA or an RNA chain. A number of predictions may be made from Weiss’ model. First, the enzyme should be able to remove a mispaired 3’-terminal nucleotide residue from DNA, and this is in fact a known property of the enzyme (170).Second, the enzyme may be able to act as an endonuclease on DNA containing mismatched or looped-out nucleotide residues. (A) Enzyme 3 recognition sites
(C) J’-Phosphatase
B -P-ester
(active site)
4-space
(B) Endonylease
@) Exonucleas:
FIG.3. The common site model for the multiple functions ofE. coli exonuclease 111 (37). The enzyme (shaded area) cleaves phosphocliester bonds at sites indicated by arrows. It is hypothesized that the enzyme has three domains, the active site, a second site that recognizes double-helical structure, and a third site that recognizes B space. (Reproduced with permission of the American Society of Biological Chemists.)
There is presently little experimental evidence for such a mode of action. Third, the endonucleolytic incision at an apurinic/apyrimidinic site should occur at the 5’ side of the lesion (see Fig. 3). This prediction has recently been verified experimentally. The deoxyribose residue at an apurinic/apyrimidinic site may be reduced to an alcohol form by treatment with sodium borohydride, and DNA containing such reduced sites is still a substrate for the AP endonuclease activity of exonuclease 111 (39, 98, 129). By using radioactive sodium borohydride for reduction it could be shown that AP endonuclease activity and subsequent removal of several nucleotide residues at each site by exonuclease 111 activity occurred without release of the deoxyribosephosphate residue at the apurinic site (39).The initial endonucleolytic cleavage generates a 3’-OH nucleotide end and a base-free deoxyribose 5phosphate at the 5’ terminus (171).Thus, the enzymic incision occurs at a different location from that of alkali-induced chain cleavage, which takes place at the 3’ side of an apurinidapyrimidinic site. There are several precedents for the association of endonuclease and exonuclease activities with a single protein. In fact, while these terms are often useful in discussing reagent enzymes, the whole subdivision of DNases into endonucleases vs exonucleases may be largely artificial. One example of this is E . coli exonuclease V, which acts as an ATP-dependent exonuclease, but also cleaves single-stranded circular DNA in an endonucleolytic reaction (172). While the AP endonuclease function of exonuclease 111 cannot be physically separated from the exonuclease activity, conditions have been established that permit the assaying of the AP endonuclease function of the enzyme without measuring significant amounts of the exonuclease activity. This can be done by an assay procedure primarily sensitive to endonucleases, e.g., by using covalently closed circular DNA molecules containing apurinic sites, measuring their conversion to a “nicked” circular form by a filter-binding assay (38, 172). Another approach depends on the observation that the exonuclease 111activity of the enzyme, but not the AP endonuclease activity, is strongly inhibited in reaction mixtures containing the weak chelating agent citrate and no added Mg2+(169). Similarly, the substitution of Ca2+for Mg2+in reaction mixtures seems to inhibit the exonuclease activity of the enzyme much more strongly than the AP endonuclease activity (171 ). On the other hand, conditions that would permit exonuclease I11 activity while inhibiting the AP endonuclease activity have not been found. E . coli mutants deficient in the major AP endonuclease activity
182
TOMAS LINDAHL
have been isolated in two different laboratories (163,173).'ORepresentative mutant strains, such as strains NH5016 or BWSlOl, are slightly but significantly sensitive to methyl methanesulfonate, presumably because of the increase in DNA depurination rate that occurs as a consequence of alkylation, while they show normal growth rates and normal resistance to ultraviolet radiation (91, 173). Further, there is little, if any sensitivity to ionizing radiation compared to wild-type strains, but rejoining at some X-ray-induced lesions is slightly impaired (E. Seeberg, personal communication). With the realization that the major AP endonuclease and exonuclease 111ofE. coli were the same protein, many more mutants became available, as several exonuclease III-deficient mutants, xth, had been previously isolated (174).The gene for exonuclease I11 maps at 38.2 minutes of the E . cold K12 map (90,164).It can be easily transduced to other strains, as linked markers for drug resistance occur (164),and it has also been cloned with A and colE1 vectors (98).The latter procedure leads to a 30-fold increased intracellular level of the enzyme with no impairment of growth or viability. Exonuclease I11 seems to be a nonessential enzyme, as deletions extending into the xth gene have apparently been obtained (163). The latter observation may be explained by the existence of additional AP endonucleases in E . coli if excision-repair systems to handle apurinic and apyrimidinic sites are essential. 2. HEMOPHILUS INFLUENZAE EXONUCLEASE I11 An enzyme with very similar properties to the E . coli exonuclease I11 is present in H . influenzae (175). It has an associated DNA 3'phosphatase activity (17 5 ) as well as an associated AP endonuclease activity (171), which are physically inseparable. The enzyme is a monomer of molecular weight 30,000. It has a pH optimum at 8.0-9.0 and requires M$+ for activity. Further, it incises DNA at the 5' side of apurinic and apyrimidinic sites. An H . influenzae mutant deficient in exonuclease I11 has been isolated (176),and this mutant has simultaneously become deficient in A P endonuclease activity (171).In conclusion, it seems clear that H . influenzae exonuclease 111 has closely analogous properties to the E . coli enzyme. An exonuclease-III-like enzyme with an associated DNA-3'phosphatase activity from Diplococcus pneumoniae has been delo It should be noted here that the first two tight mutants of this type obtained are atypical and should no longer be used as representative strains. Thus, the strain BW2001 is radiation-sensitive and slow-growing, perhaps owing to the presence of a second unrelated mutation, while the strain AB3027 has a DNA polymerase I deficiency in addition to its AP endonuclease deficiency.
scribed (177). While it has not yet been tested for AP endonuclease activity, it would appear that this kind of enzyme may be found in many unrelated bacteria. On the other hand, exonuclease I11 does not occur in several bacterial species, including M . luteus (178), and an exonuclease-III-like enzyme is not present in mammalian cells ( I 69). However, AP endonuclease activities are universally distributed. Although the exonuclease function of exonuclease I11 was the first activity of this enzyme to be discovered, and it has been a widely used reagent enzyme, it now seems likely that the physiologically most important function of this enzyme is its AP endonuclease activity, while the exonuclease activity is an accessory function of unclear biological relevance. C. AP Endonucleases without Associated Exonuclease Activity
1. MAMMALIAN AP ENDONUCLEASE Calf thymus extracts contain an enzyme that specifically incises DNA at apurinic and apyrimidinic sites (30),and this enzyme has been extensively purified and characterized (153, 169,179). A very similar activity has been described from human fibroblasts (38), and briefer accounts have been published on the AP endonucleases of rat liver (180), calf liver (181), and human tissue (182). The mammalian AP endonuclease has a molecular weight of 29,000-32,000, and exhibits a pH optimum at 8.CL8.5. It requires Mg2+ for activity, with an optimal concentration of 2-3 mM, and in contrast to E . coli exonuclease 111its AP endonuclease activity is strongly inhibited by citrate in the absence of added Mg2+.The most striking difference between the mammalian enzyme and E . coli exonuclease I11 is that the mammalian AP endonuclease appears totally devoid of associated DNA exonuclease and phosphatase activities. The mammalian enzyme shows a strict specificity for apurinic and apyrimidinic sites in DNA and does not act on normal DNA, or on DNA containing pyrimidine dimers or alkylated purines. In resemblance to all other purified AP endonucleases described to date, it shows a strong preference for double-stranded DNA and may be unable to cleave at apurinic sites in single-stranded DNA (179). Further, the enzyme seems to cleave at the 5’ side of apurinic M) for apurinic sites in DNA and exhibits a very low K , (lo+ to sites (38). Chromatography on phosphocellulose usually resolves the mammalian AP endonuclease activity in several peaks (38), but the significance of this phenomenon remains uncertain. In our laboratory, multiple peaks of calf thymus AP endonuclease activity obtained after phosphocellulose chromatography have exhibited practically identi-
184
TOMAS LINDAHL
cal properties with regard to heat sensitivity, substrate specificity, absence of associated exonuclease activity, pH optima, Mg2+ requirement, inhibition by NaC1, and a number of other parameters. Further, the properties of AP endonuclease from calf liver were identical to the activity from calf thymus in all these respects (J. Satava, S. Ljungquist, and T. Lindahl, unpublished data), so a report that the liver and thymus AP endonucleases have different characteristics (181) could not be confirmed. It seems plausible that the heterogeneous chromatographic behavior of the mammalian AP endonuclease on phosphocellulose could be due merely to posttranslational modification; e.g., the same protein sequence could occur in different degrees of phosphorylation or acetylation, or could have been subject to different degrees of limited proteolysis. There is certainly no convincing biochemical evidence for the existence of more than one mammalian AP endonuclease to date, but this problem should be pursued as bacteria seem regularly to have at least two different gene products acting as AP endonucleases. Kuhnlein et al. have reported that extracts of fibroblasts from xeroderma pigmentosum complementation group D had reduced levels of AP endonuclease activity and that the K , for apurinic sites of the enzyme fraction was moderately increased (183).These alterations may be correlated with the loss of one of the AP endonuclease peaks observed by phosphocellulose chromatography (38). If confirmed, these observations imply that the human AP endonuclease activity may be due to several gene products. In a separate investigation, no reduction in AP endonuclease activity was detected in extracts of human fibroblasts derived from a case of ataxia-telangiectasia and a case of Bloom's syndrome (184), two inherited human diseases with deficient DNA repair after exposure to ionizing radiation and increased frequency of sister chromatid exchange, respectively. An enzyme,fraction, partly purified from human cells, can effectively remove apurinic sites from DNA in vitro, with the formation of small gaps in the DNA (185).The fraction apparently contains two different enzymes, an AP endonuclease and a 5' + 3' exonuclease. Addition of mammalian DNA polymerase a and phage T4 DNA ligase to the gapped DNA permits gap-filling and rejoining, which suggests that excision-repair of apurinic sites may proceed in a similar fashion in human cells in vivo.
2. AP ENDONUCLEASES FROM PLANTS Extracts of several sources of plant cells contain an AP endonuclease activity with properties fairly similar to those of the mammalian
enzyme (186,187).The purified enzyme from Phaseolus multijorus is totally devoid of associated exonuclease activity, and shows no activity on intact DNA or at alkylated residues in DNA. However, this plant AP endonuclease appears slightly larger than the AP endonucleases from bacteria or mammalian cells, with a reported molecular weight of 40,000. An interesting observation is that the AP endonuclease activities in nuclei and chloroplasts have significantly different properties (187).
3. E . coli ENDONUCLEASE IV E . coli xth- mutants, lacking detectable amounts of exonuclease 111, are not totally devoid of AP endonuclease activity. Most or all of the residual AP endonuclease activity in such mutants is due to an enzyme with properties very different from exonuclease 111, e.g., marked heat stability, resistance of the enzyme activity to addition of EDTA in the absence of Mg2+,and presence at normal levels in several xth- mutants (173).This AP endonuclease has been purified 3000-fold from E. coli but has not yet been obtained in homogeneous form (188). It has been termed E. coli endonuclease IV, as it is physically and genetically separable from exonuclease 111, and it has a molecular weight of 30,000-33,000 (165, 188). Endonuclease IV has a pH optimum at 8.0-8.5, shows no stimulation by addition of M$+, and is strongly inhibited by p-mercuribenzoate. On the other hand, it is markedly resistant to the addition of NaCl to reaction mixtures. It is strictly specific for apurinic and apyrimidinic sites, and shows no significant activity on intact or ultraviolet-irradiated DNA. Further, it seems free from any associated DNA glycosylase activity. A closer definition of the physiological role of endonuclease IV, and its potential ability partly to substitute for the AP endonuclease function of exonuclease 111 in the excision-repair of apurinic sites, will have to await the isolation of endonuclease IV-deficient E . coli mutants.
4. Bacillus stearothermophilus AP ENDONUCLEASE An AP endonuclease has been purified to homogeneity from B . stearothermophilus (189).This is the first AP endonuclease without associated exonuclease activity to be obtained in homogeneous form. The enzyme has a monomeric structure and a molecular weight of 28,000. It contains a high proportion of hydrophobic amino acids. As expected for an enzyme from a thermophile, it shows high heat resistance and is optimally active at 60°C. The enzyme specifically cleaves DNA at AP sites and shows no activity on intact or alkylated DNA. Further, it is fully active in the presence of EDTA and has no detecta-
186
TOMAS LINDAHL
ble exonucleolytic activity. Thus, the B . stearothermophilus AP endonuclease seems closely similar to E . coli endonuclease IV, while markedly different from exonuclease 111.
5. M . luteus AP ENDONUCLEASES Two different AP endonucleases have recently been purified from
M . luteus (178). Both enzymes are specific for apurinic and apyrimidinic sites and exhibit no associated exonuclease activity. In fact, M . luteus extracts, which are very low in DNase activity against intact DNA, do not seem to contain any exonuclease III-like enzyme activity. The two M . luteus activities show different fractionation properties and heat labilities and are in all likelihood due to two distinct AP endonucleases. In conclusion, it appears that AP endonucleases without associated exonuclease activities are more widely occurring than exonucleaseIII-like enzymes, as they have been consistently found in eukaryotic cells and also in many bacteria, including E . coli. D. Endonucleases Acting a t Many Lesions, Including Apurinic and Apyrimidinic Sites
A number of potential DNA repair enzymes that seem to act at a variety of lesions in DNA, including ultraviolet-induced damage, alkali-stable lesions introduced by OsO, treatment, and perhaps apurinic sites, have been described. These would include E . coli endonuclease I11 (22,23)and E . coli endonuclease V (loo),as well as the M . luteus endonuclease investigated by Tomilin et al. (190). E . coli endonuclease V also acts on normal single-stranded DNA and on uracil-containing DNA. No bacterial mutants deficient in these interesting enzymes have yet been isolated, and in no case has it been demonstrated that a physically homogeneous enzyme possesses several nuclease activities, including AP endonuclease activity. Consequently, the observations are clearly preliminary in nature, and it remains to be ascertained if E . coli has multiple AP endonucleases, or if exonuclease I11 and endonuclease IV account for all the enzyme activity of this type. A heat-labile, protease-sensitive, and nondialyzable activity that cleaves single-stranded DNA at depurinated sites has been detected in E . coli extracts (63).This activity, which is inhibited by tRNA, should be different from either exonuclease I11 or endonuclease IV, since the latter enzymes are specific for depurinated sites in double-stranded DNA. Basic proteins, like polyamines, promote the rate of chain breakage at DNA apurinic sites (191).Thus, it is possible markedly to accelerate
the rate of chain cleavage at apurinic sites by incubation of partly depurinated DNA with high concentrations of pancreatic ribonuclease or cytochrome c, and these findings should be kept in mind during investigations on AP endoiiuclease activities of crude cell extracts.
V. Repair of Apurinic Sites in DNA by Alternative Pathways Linn et al. (192) published a preliminary description of a fascinating enzyme activity from human fibroblasts that directly inserts missing bases into DNA. The activity is due to a protein of molecular weight 120,000 that specifically binds to partly depurinated DNA. No energy source except that of reformation of DNA secondary structure seems to be needed. Thus, the partly purified enzyme specifically adds adenine at apurinic sites in poly(dA-dT) and similarly guanine to poly(dG-dC). The activity sediments faster in glycerol gradients than AP endonucleases and can be freed from interfering endonuclease activities in this fashion. While enzymic chain cleavage at apurinic sites seems to occur very rapidly both in viuo and in crude extracts, these observations indicate that human fibroblasts may possess an additional pathway that does not involve chain cleavage for the repair of apurinic sites. A damaged or misincorporated base in DNA could conceivably be corrected in only two enzymic steps with no replacement of phosphate or sugar residues, DNA glycosylase action followed by insertion of the correct base. There is presently little information available on the ability of E . coli and other cells to cope with hydrolytic DNA lesions, such as deaminated base residues or apurinic sites, by recombinational repair and by induced mechanisms of DNA repair. As expected, E . coli ungmutants have retained normal ability to perform recombinational repair after UV irradiation (E. C. Friedberg, personal communication), so there is no general interference with alternative DNA repair pathways by a DNA glycosylase defect. There is no suitable in vitro assay of recombinational repair, and until recently no methods have been available to introduce large numbers of apurinic or apyrimidinic sites into D N A in uiuo except by treatment of cells with alkylating agents, which yield a variety of different lesions. It is now possible to introduce DNA apyrimidinic sites in uiuo specifically by relying on the ability of uracil-DNA glycosylase to convert dUMP residues to such sites. DNA containing dUMP residues may be obtained by employing E . coli dut- mutants (85) or by UV irradiation of bromouracilcontaining DNA in the presence of cysteamine (73). However, it is still difficult to evaluate the possible role of alternative pathways in the
188
TOMAS LINDAHL
repair of apurinic and apyrimidinic sites in such systems, because no bacterial mutants that totally lack ability to incise DNA at apurinic/ apyrimidinic sites seem to be available. E . coli dut xth double mutants grow very poorly, apparently because they tend to accumulate apyrimidinic sites in their DNA to some extent (Q8), but it is not known if the survival of these strains depends on the ability to perform a certain amount of residual excision-repair, perhaps relying on endonuclease IV, or if recombinational repair or other modes of repair can occur at the apyrimidinic sites. Isolation of mutants deficient both in exonuclease I11 and endonuclease IV, which might turn out to be conditionally lethal, or the introduction of a recA- mutation into dut-xth- mutants to prevent recombinational repair, could to some extent clarify the relative importance of DNA repair pathways different from excision-repair in the circumvention of DNA damage involving transient formation of apurinic and apyrimidinic sites.
REFERENCES 1 . A. I. Bukhari, J. A. Shapiro and S. Adhya, eds., “DNA Insertion Elements, PIasmids and Episomes.” Cold Spring Harbor Laboratory, New York, 1977. 2. T. Lindahl, in “DNA Repair Processes” (W. W. Nichols and D. G . Murphy, eds.), p. 225. Symposia Specialists, Miami, 1977. 3. J. W. Drake and R. H. Baltz, ARB 45, 11 (1976). 4. T. Lindahl, PNAS 71, 3649 (1974). 5. B. K. Tye, P. 0. Nyman, I. R. Lehman, S. Hochhnuser and B. Weiss, PNAS 74, 154 (1977). 6. B. K. Duncan, P. A. Rockstroh and H. R. Warner,]. Bact. 134, 1039 (1978). 7. L. Grossman, A. Braun, R. Feldberg and I. Mahler, ARB 44, 19 (1975). 8. R. B. Setlow and W. L. Carrier, PNAS 51, 226 (1964). 9. R. P. Boyce and P. Howard-Flanders, PNAS 51, 293 (1964). 10. S. Riazuddin and L. Grossman,]BC 252, 6280 (1977). 1 1 . E. C. Friedberg, Photochem. Photohiol. 21,277 (1975). 12. P. Howard-Flanders and L. Theriot, Genetics 53, 1137 (1965). 13. E. Seeberg, J. Nissen-Meyer and P. Strike, Nature 263, 524 (1976). 14. K. H. Kraemer, E. A. de Weerd-Kastelein, J. H. Robbins, W. Keijzet, S. F. Barrett, R. A. Petinga and D. Bootsma, Mutut. Res. 33, 327 (1975). 15. E. Seeberg, PNAS 75,2569 (1978). 15a. E. Seeberg, in “DNA Repair Mechanisms” (P. C. Hanawalt, E. C. Friedberg and C. F. Fox, eds.). Academic Press, 1978, in press. 15h. M. Ikenaga, H. Ichikawa-Ryo and S. Kondo,]MB 92, 341 (1975). 16. S. Venitt and E. M. Tarmy, BBA 287,38 (1972). 17. P. Howard-Flanders, Br. Med. Bull. 29, 226 (1973). 18. K. Minton, M. Durphy, R. Taylor and E. C. Friedberg, ]BC 250, 2823 (1975). 19. S. Bacchetti, A. van der Plas and C. Veldhuisen, BBRC 48, 662 (1972). 20. S. Bacchetti and R. Benne, BBA 390, 285 (1975). 21. T. P. Brent, NARes. 4, 2445 (1977). 22. M. Radman, JBC 251, 1438 (1976). 23. F. T. Gates and S. Linn,JBC 252, 2802 (1977).
24. R. B. Kelly, M. R. Atkinson, J. A. Huberman and A. Kornberg, Nature 224, 495 (1969). 25. P. Cooper, Mol. Gen. Genet. 150, 1 (1977). 26. J. C. Kaplan, S. R. Kushner and L. Grossman, Bchem. 10, 3315 (1971). 27. T. Lindahl, EJB 18, 407 (1971). 28. J. Doniger and L. Grossman,JBC 251, 4579 (1976). 29. E. C. Friedberg, K. H. Cook, J. Duncan and K. Mortelmans, Photochem. Photohiol. Reo. 2 263 (1977). 30. T. Lindahl, in “Molecular and Cellular Repair Processes” (R. F. Beers, R. M. Herriott and R. C. Tilghman, eds.), p. 3. Johns Hopkins Univ. Press, Baltimore, Maryland 1972. 31. J. E. Cleaver, Nature 270, 451 (1977). 32. A. Kornberg, “DNA Synthesis.” Freeman, San Francisco, 1974. 33. I. R. Lehrnan, Science 186, 790 (1974). 34. A. Weissbach, Cell 5, 101 (1975). 35. S. Soderhall and T. Lindahl, FEBS Lett. 67, 1 (1976). 36. T. Lindahl, Nature 259,64 (1976). 37. B. Weiss,JBC 251, 1896 (1976). 38. W. S. Linsey, E. E. Penhoet and S. Linn,JBC 252, 1235 (1977). 39. F. Gossard and W. G. Verly, EJB 82, 321 (1978). 40. D. M. Kirtikar and D. A. Goldthwait, PNAS 71, 2022 (1974). 41. J. Duncan, L. Hamilton and E. C. Friedberg,]. Virol. 19, 338 (1976). 42. K. Nath and J. Hurwitz,JBC 249, 3680 (1974). 43. E. Bedows, J. T. Wachsman and R. I. Gumport, Bchem 16,2231 (1977). 44. J. D. Regan and R. B. Setlow, Cancer Res. 34, 3318 (1974). 45. P. Karran, N. P. Higgins and B. Straws, Bchem 16, 4483 (1977). 46. W. G. Verly, F. Gossard and R. Crine, PNAS 71, 2273 (1974). 47. P. A. Cerutti, in “Molecular Mechanisms for Repair of DNA” (P. C . Hanawalt and R. B. Setlow, eds.), Part A p. 3. Plenum, New York, 1975. 48. P. D. Lawley, in “Screening Tests in Chemical Carcinogenesis” (R. Montesano, H. Bartsch, and L. Tomatis, eds.) p. 181. IARC, Lyon, 1976. 49. K. Geider, EJB 27, 554 (1972). 50. M. G. Wovcha and H. R. Warner,JBC 248, 1746 (1973). 5 1 . W. L. Carrier and R. B. Setlow, FP 33, 1599 (1974). 52. F. Tomita and I. Takahashi,/. Virol. 15, 1081 (1975). 53. Bchem 10, 3983 (1971);BJ 125,673 (1972;JBC 247,613 i1972). 54 R. Shapiro and R. S. Klein, Bchem 5, 2358 (1966). 55. W. J. Wechter and R. C. Kelly, Collect. Czech. Chem. Commun. 35, 1991 (1970). 56. R. E. Notari,J. Pharm. Sci. 56, 804 (1967). 57. E. R. Garrett and J. Tsau, J. Plzarrn. Sci. 61, 1052 (1972). 58. G. J. Fisher and H . E. Johns, in “Photochemistry and Photobiology of Nucleic Acids” (S. Y. Wang, ed.), Vol. 1, p. 169. Academic Press, New York, 1976. 59. T. Lindahl and B. Nyberg, Bchem 13, 3405 (1974). 60. H. Tapiero, R. Caneva and C. L. Schildkraut, BBA 272, 350 (1972). 61. J. M. Collins,JBC 252, 141 (1977). 62. P. Henson,JMB 119, 487 (1978). 63. E. C. Friedberg, C. Anderson, T. Bonura, R. Cone and R. Simmons, in “DNA Repair Mechanisms” (P. C. Hanawalt, E. C. Friedberg, and C. F. Fox, eds.). Academic Press, N.Y., 1978, in press. 64. J. Y. Vanclerhoek and P. A. Cerutti, BBRC 52, 1156 (1973). 65. R. Shapiro, B. Braverman, J. B. Louis and R. E. Servis,JBC 248, 4060 (1973).
190
TOMAS LINDAHL
66. J. S. Ullman and B. J. McCarthy, BBA 294,396 (1973). 67. H. Schuster, Z. Naturforsch., Teil B 15, 298 (1960). 68. R. Shapiro, S. Dubelman, A. M. Feinberg, P. F. Crain and J. A. M. Closkey,JACS 99,302 (1977). 69. R. Shapiro and S. H. Pohl, Bchem. 7,448 (1968). 70. F. K. Zimmermann, Mutat. Res. 39, 127 (1977). 71. L. Smets and T. Cornelis, Int. J . Radiat. Biol. 19, 445 (1971). 72. W. D. Rupp and W. H. Prusoff, BBRC 18, 158 (1965). 73. F. Makino and N. Munakata,J. Bact. 131,438 (1977). 74. G. Scholes, in “Photochemistry and Photobiology of Nucleic Acids” (S. Y. Wang, ed.), p. 521. Academic Press, New York, 1976. 75. E. C. Friedberg, A. K. Ganesan and K. Minton,J. Virol. 16,315 (1975). 76. M. Gold and J. Hurwitz,JBC 209, 3858 (1964). 77. J. Singer, R. H. Stellwagen, J. Roberts-Ems and A. D. Riggs,JBC 252,5509 (1977). 78. J. Lichtenstein and S. S. Cohen,JBC 235, 1134 (1960). 79. R. H. Baltz, P. M. Bingham and J. W. Drake, PNAS 73, 1269 (1976). 80. T. Lindahl, S. Ljungquist, W. Siegert, B. Nyberg and B. Sperens,JBC 252, 3286 (1977). 81. R. Cone, J. Duncan, L. Hamilton and E. C. Friedberg, Bchem 16,3194 (1977). 82. M. Sekiguchi, H. Hayakawa, F. Makino, K. Tanaka and Y. Okada; BBRC 70, 293 (1976). 83. U. Kuhnlein, B. Lee and S . Linn, NARes. 5, 117 (1978). 84. K. Brynolf, R. Eliasson and P. Reichard, Cell 13, 573 (1978). 85. B. K. Tye, J. Chien, I. R. Lehman, B. K. Duncan and H. R. Warnw PNAS 75,233 (1978). 86. F. Sanger, S. Nicklen and A. R. Coulson, PNAS 74, 5463 (1977). 87. A. M. Maxam and W. Gilbert, PNAS 74, 560 (1977). 88. B. K. Duncan, P. A. Rockstroh and H. R. Warner, FP 35, 1493 (1976). 89. S. Riazuddin and T. Lindahl, Bchem 17,2110 (1978). 90. B. J. Bachmann, K. B. Low and A. L. Taylor, Bact. Rev. 40, 116 (1976). 91. R. DaRoza, E. C. Friedberg, B. K. Duncan and H. R. Warner, Bchem 16, 4934 (1977). 91a. B. K. Duncan and B. Weiss, in “DNA Repair Mechanisms” (P. C. Hanawalt, E. C. Friedberg, and C. F. Fox, eds.). Academic Press, N.Y., 1978, in press. 91b. C . Coulondre, J. H. Miller, P. J. Farabaugh, and W. Gilbert, Nature 274, 775 (1978). 92. B. K. Duncan and H. R. Warner, Nature 272, 32 (1978). 93. M. J. Bessman, I. R. Lehman, J. Adler, S. B. Zimmerman, E. S. Sinims and A. Kornberg, PNAS 44,633 (1958). 94. L. E. Bertani, A. Haggmark and P. Reichard, JBC 238, 3407 (1963). 95. S. J. Hochhauser and B. Weiss, FP 35, 1492 (1976). 96. B. K. Tye and I. R. Lehman,JMB 117,293 (1977). 97. B. M. Olivera, PNAS 75, 238 (1978). 98. B. Weiss, S. G. Rogers and A. F. Taylor, in “DNA Repair Mechanisms” (P. C. Hanawalt, E. C. Friedberg, and C. F. Fox, eds.). Academic Press, New York, 1978, in press. 99. R. B. Thompson, T. J. Mozer, B. K. Duncan and H. R. Warner, submitted. 100. F. T. Gates and S. Linn,]BC 252, 1647 (1977). 101. Deleted. 102. E. Wist, 0. Unhjem and H. Krokan, BBA, in press (1978). 103. R. H. Grafstrom, B. Y. Tseng and M. Goulian, Cell, in press (1978). 104. I. Takahashi and J. Marmur, Nature 197, 794 (1963).
191 105. G. E. Katz, A. R. Price and M. J. Pomerantz,J. Virol. 20, 535 (1976). 106. F. Tomita and 1. Takahashi,]. Virol. 15, 1073 (1975). 107. B. K. Duncan and H. R. Warner,J. Virol. 22, 835 (1977). 108. P. Karran and T. Lindahl,JBC 253, in press (1978). 109. A. S. Jones, A. M. Mian and R. T. Walker,JCS 692 (1966). 110. T. Kotaka and R. L. Baldwin,JMB 9, 323 (1964). 1 1 1 . H. 0. Kammen and S.J. Spengler, BBA 213, 352 (1970). l l l a . K. R. Thomas, P. Manlapaz-Ramos, R. Lundguist, and B. M. Olivera, CSHSQB 43, in press (1978). 112. C. Bernstein, D. Morgan, H. G. Gensler, S. Schneider and G. E. Holmes, Mol. Gen. Genet. 148,213 (1976). 113. P. Karran, T. Lindahl, I. Ofsteng and E. Seeberg, submitted for publication. 114. B. Strauss, D. Scudiero and E. Henderson, in “Molecular Mechanisms for Repair of DNA” (P. C. Hanawalt and R. B. Setlow, eds.), Part A p. 13. Plenum, New York, 1975. 115. M. P. Raynian and A. Dipple, Bchem 12, 1202 (1973). 116. P. D. Lawley and W. Warren, Ckeni.-Biol. lnterctct. 12,211 (1976). 1 1 7 . J. Axelrod and J. Daly, BBA 61, 855 (1962). 118. P. D. Lawley and D. J. On, Chem.-Biol. Interact. 2, 154 (1970). 119. K. V. Shooter and T. A. Slade, Cltem.-Biol. lnteruct. 19, 353 (1977). 120. G. P. Margison and P. J. O’Connor, BBA 331, 349 (1973). 121. P. D. Lawley and W. Warren, Chem.-Biol. Interact. 11, 55 (1975). 122. R. Goth and M. F. Rajewsky, PNAS 71, 639 (1974). 123. A. Dipple and J. J. Roberts, Bcheni 16, 1499 (1977). 124. B. Strmss, PNAS 48, 1670 (1962). 125. H. Reiter, B. Strauss, M. Robbins and R. Marone,J. Bact. 93, 1056 (1967). 126. B. Strauss and M. Robbins, BBA 161, 68 (1968). 127. E. Friedberg, S . M. Hadi and D. A. Goldthwait, JBC 244, 5879 (1969). 128. B. Papirmeister, J. K. Dorsey, C. L. Davison and C. L. Gross, FP 29, 726 (1970). 129. S. M. Hadi and D. A. Goldthwait, Bcliem 10, 4986 (1971). 130. D. M. Kiitikar, J. Slaughter and D. A. Coldthwait, Bchem 14, 1235 (1975). 131. D. M. Kirtikar, A. Dipple and D. A. Goldthwait, Bclzm 14, 5548 (1975). 132. J. Laval, Nntzlre 269, 829 (1977). 133. P. F. Schendel, M. Defais, P. Jeggo, L. Samson and J. Cairns, in “DNA Repair Mechanisms” (P. C. Hanawalt, E. C. Friedberg, and C. F. Fox, eds.) Academic Press, N.Y., 1978, in press. 134. Y. Takagi mid B. L. Horecker,JBC 225, 77 (1957). 135. E. C. Carter,JACS 73, 1508 (1951). 136. G. Magni, E. Fioretti, P. L. Ipata and P. Natdini, JBC 250, 9 (1975). 137. G. Magni, P. Natalini, S. Ruggieri and A. Vita, BBRC 69, 724 (1976). 138. J . Huiwitz, L. A. Heppel and B. L. Horecker,JBC 226, 525 (1957). 1.39. A. Imtids, J. Gen. Appl. Microhid. 13, 267 (1967). 140. K. Ueda, M. Fukushima, H. Okayamti and 0. Hayaishi,JBC 250, 7541 (1975). 141. A. J. Andreoli, T. W. Okita, R. Bloom and T. A. Grove, BBRC 49, 264 (1972). 142. J. A. Dueri-e,JBC 2.37, 3737 (1962). 143. H. L. A. Tarr,]. Fish. Res. Borird Cun. 30, 1861 (1973). 144. T. Lindahl and S . Ljungquist, in “Molecular Mechanisms for the Repair of DNA” (P. C. Hanawalt and R. €3. Setlow, eds.), Part A, p. 31. Plenum, New York, 1975. 145. C. W. Sheiirman and L. A. Loeb, Nnture 270, 537 (1977). 146. L. Brakier and W. C. Verly, BBA 213, 296 (1970). 147. S. Greer and S. Zamenhof,JMB 4, 123 (1962). 148. T. Lindahl and B. Nylierg, Bcliem 11, 3610 (1972).
192
TOMAS LINDAHL
149. J. A. Zoltewicz, F. 0. Clark, T. W. Sharpless and G. Grahe,JACS 92, 1741 (1970).
150. E. R. Garrett and P. J. Mehta,JACS 94, 8542 (1972). 151. T. Lindahl and 0. Karlstrom, Bchem 12, 5151 (1973). 152. T.Lindahl and A. Andersson, Bchem 11, 3618 (1972). 153. S. Ljungquist, A. Andersson and T.Lindahl,JBC 249, 1536 (1974). 154. B. D~iiil~ip imd P. Cerutti, FEBS Lett. 51, 188 (1975). 155. R.’Teoule, C. Bert and A. Bonicel, R~dicrtRES.72, 190 (1977). 156. N. J. Duker and G. W. Teebor, PNAS 73,2629 (1976 ). 157. T. P. Brent, G. W. Teebor and N. J. Duker, in “DNA Repair Mechanisms” (P. C. Hanawalt, E. C. Friedberg, and C. F. Fox, eds.), Academic Press, New York, in press. 158. C. Tamm, H. S. Shapiro, R. Lipschitz and E. Chargatf,JBC 203, 673 (1953). 1580. E. Freese and M. Cashel, BBA 91, 67 (1964). 159. W. G. Verly and Y. Paquette, Con. J. Biochem. SO, 217 (1972). 160. W. G. Verly, Y. P q u e t t e and L. Thibodeau, Nature NB 244, 67 (1973). 161. W. G. Verly and E. Rassait,JBC 250, 8214 (1975). 162. C. C. Richardson, I. R. Lehman and A. Kornberg,JBC 239, 251 (1964). 163. D. M. Yajko and B. Weiss, PNAS 72, 688 (1975). 164. B. J. White, S. J. Hochhauser, N . M. Cintron and B. WeissJ. Bact. 126,1082 (1976). 165. S. Ljungquist and T. Lindahl, NARes. 4, 2871 (1977). 166. D. M. Kirtikar, G. M. Cathcart and D. A. Goldthwait, PNAS 73, 4324 (1976). 167. D. M. Kirtikar, G. R. Cathcart, J. G. White, I. Ukstins and D. A. Goldthwait, Bchem 16, 5625 (1977). 168. K. R. Thomas and B. M. Olivera,JBC 253, 424 (1978). 1 6 8 ~ .R. Wu, G. Ruben, B. Siegel, E. Jay, P. Spiehnan and C. D. Tu, Bclzein 15, 734 (1976). 169. S. Ljungquist, B. Nyberg and T. Lindahl, FEBS Lett. 57, 169 (1975). 170. D. Brutlag and A. Kornberg, JBC 247, 241 (1972). 171. J. E. Clements, S. G. Rogers and B. Weiss,JBC 253, 2990 (1978). 172. P. J. Goldinark and S. Linn, PNAS 67, 434 (1970). 173. S. Ljungquist, T. Lindahl and P. Howard-Flanders,J. Bact. 126, 646 (1976). 174. C. Milcarek and B. Weiss, JMB 68, 303 (1972). 175. J. K. Gunther and S. H. Goodgal; JBC 245, 5341 (1970). 176. R. Gromkova, J. Bendler and S. H. Goodgal,J. Bact. 114, 1151 (1973). 177. S. Lacks and B. Greenberg,JBC 242, 3108 (1967). 178. J. Lava1 and J. Pierre, i n “DNA Repair Mechanisms” (P. C. Hanawalt, E. C. Friedberg, and C. F. Fox, eds.) Academic Press, N.Y., 1978, in press. 179. S. Ljungquist and T. Lindahl, JBC 249, 1530 (1974). 180. W. G. Verly and Y. Paquette, Can. J. Biochm. 51, 1003 (1973). 181. J. P. Kuebler and D. A. Goldthwait, Bchem 16, 1370 (1977). 182. G. W. Teebor and N. J. Duker, Nature 258, 544 (1975). 183. U. Kuhnlein, E. E. Penhoet and S. Linn, PNAS 73, 1169 (1976). 184. T. Inoue, K. Hirano, A. Y. Ama, T. Kada and H. Kato, BBA 479,497 (1977). 185. K. Bose, P. Karran and B. Strauss, PNAS 75, 794 (1978). 186. L. Thibodeau and W. G. Verly, FEBS Lett. 69, 183 (1976). 187. L. Thibodeau and W. G. Verly, JBC 252,3304 (1977). 188. S. Ljungquist,JBC 252, 2808 (1977). 189. V. Bibor and W. G. Verly,JBC 253, 850 (1978). 190. N. V. Tomilin, E. B. Paveltchuk aud T. V. Mosevitskaya, EJB 69, 265 (1976). 191. M. R. McDonald and B. P. Kaufmann,/. Histochem. Cytochem. 2, 387 (1954). 192. S. Linn, U. Kuhnlein, and W. A. Deutsch, in “DNA Repair Mechanisms” (P. C. Hanawalt, E. C. Friedberg and C. F. Fox, eds.), Academic Press, N.Y., in press.
Naturally Occurring Nucleoside and Nucleotide Antibiotics
ROBERT J. SUHADOLNIK Department of Biochemistry Temple Uniuerdty School of Medicine Philtidelphin, Petrnrylvaniu
Introduction .................................................... I. Inhibitors of Protein Synthesis ................................... A. Pyrimidine Nucleoside Antibiotics ............................ B. Purine Nucleoside Antibiotics ................................ 11. Inhibitors of RNA Synthesis ..................................... 111. Inhibitors of DNA Synthesis, DNA Viruses, and RNA Viruses ............................................... IV. Inhibitors of Adenosine Deaniinase and Effectors of the Immune Response ........................................ V. Inhibitors of Cell-Wall Synthesis and Antifungal Agents .............................................. VI. Inhibitors of Purine and Pyrimidine Interconversions ............. VII. Hyperesthetic and Hyperemic Agents ............................ VIII. Cyclic-AMP Phosphodiesterase Inhibitors ........................ IX. Miscellaneous Naturally Occurring Nucleosides .................. References .....................................................
193 196 197 200 209 239 245 249 266 269 269
270 272
Introduction In 1970, I reviewed the thirty naturally occurring nucleoside antibiotics that had been discovered from 1951 to 1970 ( 1 ) . In an earlier (1966) review in this series, Fox, Watanabe, and Bloch described twenty naturally occurring nucleoside antibiotics (2). Since 1970, thirty-two additional naturally occurring nucleoside antibiotics have been discovered. The progress involving their structural elucidations, syntheses, and use in studies with bacteria, mammalian cells in culture, and viruses has been truly impressive. Furthermore, three of the naturally occurring nucleoside antibiotics (tubercidin, 5-azacytidine, and ara-A) have been successfully used in humans as antineoplastic and antiviral agents. A fourth antibiotic, pyrazofurin, is undergoirig clinical testing as an antineoplastic agent for carcinoma of the breast. In addition to the health-related application of these naturally occurring nucleoside and nucleotide antibiotics, some have found application as antifungal agents (foiinycin, the polyoxins, and sinefungin). 193 Progress in Niicleic Acid Research and Molecular Biology, Vol. 22
Copylight @ 1979 by Academic Press. Inc. All tights of reproduction in any form reserved. ISBN 012-540022-5
194
ROBERT J. SUHADOLNIK
TABLE I ALPHABETICAL LIST OF NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS DISCUSSED IN THISREVIEW Name A201A A9145A A9145C Agrocin 84 Amicetin (amicetin A) 3'-Aminoadenosine Aminofluorene(s),bN-substituted 2'-Aminoguanosine Amipurimycin Angusbnycin C (see psicofuranine) Anthelmycin" (hikizimycin) Ara-A (see vidarabine") Ara-C (see cytarabine") Ara-T (see spongothymidine) Ara-U (see spongouridine) Aristeromycin Aspiculamycin (see gougerotin) Asteromycin (see gougerotin) BAzacytidine Bamicetin Blasticidin S Blasticidin H Bredinin Clindamycin ribonucleotides Clitidine Coformycin Cordycepin (3'-deoxyadenosine) Covidarabine (see deoxycoformycin) Crotonoside (isoguanosine) Cytarabine" (ara-C) Decoyinine (angustmycin A) Deoxycoformycin (co-vidarabine, pentostatin) 5,6-Dihydro-Sazathymidine (U-44590) Eritadenine (lentinacin, lentysine) Exotoxin (see thuringieiisin) Ezomycins Formycin Formycin B Gougerotin (aspiculaniycin, asteromycin, moroyamycin) Herbicidin A and B Hikizimycin (see anthelmycin") Homocitrullyladenosine Ileumycin
Section I,B,4
v, 10
v, 10 v, 1 I,A II,2 v, 9 I,B,5; II,I v,2
Structure 15 83 84 65 3 24 81 16 66
I,A
8
II,3
25
II,4 I,A I,A II,8 II,5 I,B,6 VI I
26 (27) 4
2
II,6
28 18-23 92 61 30
v,3
67
VI,1 (11,l); IV III,3 11.7
89 62 56 31-33
v,4 II,12 II,13 I,A
68-75 47 48 1
(11,l); IV
IX,3
132 V, 13
12
195
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
TABLE I (Continued) Name Isoguanosine (see crotonoside) Laurusin (see formycin 3) Lysylaminoadenosine 1-Methylpseudouridine Mildiomycin Minimycin (see oxazinomycin) Moroyamycin (see gougerotin) Nebularine Nikkom ycin Norplicacetin Nucleocidin (antibiotic T-3018) Octosyl acids 0-Clycosyl ribonucleosides Oxamicetin Oxazinomycin (minimycin) Oxoformycin B Pentopyranines Pentostatin (see deoxycoformycin) Platenocidin Plicacetin Polyoxins Psicofuranine (angustmycin C) Puromycin Puromycin aminonucleoside Pyrazofurin' (pyrazomycin) Raphanatin Reversed puromycin Ribavirin Sangivam ycin Septacidin Showdomycin Sinefungin (A9145) Spongoadenosine (see vidarabine) Spongosine Spongoth ymidine Spongouridine Streptovirudins Thraustomycin Thuringiensin (exotoxin) Toyocamycin (uramycin, vengicide) Tubercidin (7-deazaadenosine, sparsamycin A) Tunicamycin Vidarabine" (ara-A, spongoadenosine) a
USAN Not a nucleoside antibiotic.
Section
Structure
I,B,2 III,5 Footnote 1
13 58 (59)
III,6 v, 5 I,A I,BJ VIII IX, 1 I ,A III,4 II,14 II,8
60 76 7 14 9-3-95 96-98 5 57 49 34-42
V, 6 I,A v, 7 VI,1 I,B,l (I,B,l); II,16 VI,2 IX,2 LB,1 w 2 II,11 v, 9 v, 10
77 6 78 88 9 51 90 99 11 91 45 79 80 82
III,l III,2 III,1 Footnote 10 V,ll II,15 II,IO II,9 V,l2 III,1
55 53 54 86 50 44 43 87 52
v,8
196
ROBERT J. SUHADOLNIK
Once the structures of some of the nucleoside antibiotics were established, it was possible in some cases to predict the cellular reaction(s) that would be affected by them. For example, the polyoxins, which are structurally similar to UDP-N-acetylglucosamine, inhibit chitin synthetase and cell-wall formation. This had been predicted. The pyrimidine nucleoside antibiotics, which resemble the -CCA terminus of tRNA, bind to ribosomal peptidyltransferase and inhibit protein synthesis in prokaryotes and eukaryotes, as had been predicted. Some of the naturally occurring nucleosides cannot be assigned biological roles until they have been tested experimentally. For example, psicofuranine, decoyinine, pyrazofurin, and bredinin inhibit de no00 purine and pyrimidine synthesis per se. A second group must be anabolized to active forms before they can become effective anticancer or antiviral agents (i.e., phosphorylation of ara-A, cordycepin, tubercidin, and fonnycin). Two adenosine analogs (agrocin 84 and thuringiensin) have been isolated as nucleotides. Agrocin 84 prevents cell growth, most probably by interacting with receptor sites on bacterial cell surfaces. Thuringiensin, as a nucleotide, is taken up by mammalian cells and acts as an ATP analog. Tunicamycin, a new Streptomyces metabolite containing no phosphorus, has been extremely useful in elucidating the synthesis of glycoproteins and cell-wall formation in the eukaryotes, prokaryotes, and viruses. This type of biochemical probe has long been needed to elucidate the complex reactions involved in the assembly of the glycoproteins, cell walls, and viral coat formation. The continued growth of research related to the naturally occurring nucleoside antibiotics is illustrated by the reviews that have appeared (2) and are cited in this chapter. The aim here is to evaluate the progress that has been made in the nucleoside antibiotics. I have elected to arrange them according to their biological functions, integrating the structures with their biological properties in order to provide a comprehensive and up-to-date presentation. (An alphabetical listing is given in Table I.) Two nucleosides, eritadenine and clitidine, although not antibiotics, are included because of their biological properties.
1. Inhibitors of Protein Synthesis The naturally occurring antibiotics that inhibit protein synthesis can be divided into three groups. One group, the pyrimidine nu-
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
197
cleoside antibiotics, inhibits protein synthesis by binding to the ribosomal peptidyltransferase that is common to prokaryotic and eukaryotic ribosomes. This group of antibiotics affects the 3’ end of aminoacyl-tRNA in a noncompetitive manner. The second group that inhibits protein synthesis is comprised of the adenine nucleoside antibiotics, and the third includes the purine and pyrimidine phosphodiester ribonucleotides, the clindamycin ribonucleotides.
A. Pyrimidine Nucleoside Antibiotics Eight naturally occurring pyrimidine nucleoside antibiotics have been isolated from Streptomyces: gougerotin (l),blasticidin S (2),
Nq
A O
N
H :fy:O HOH2C
CHzNHCH3
GOUGEROTIN ( ASPICULAMYCIN,
ASTEROMYCIN,
NH
BLASTlClDlN S 2
MOROYAMYCIN 1 1
amicetin ( R ) , bamicetin (4), oxamicetin (5),plicacetin (6),norplicacetin (7), and hikizimycin (S).l These pyrimidine nucleoside antibiotics have similar structural features and similar inhibitory patterns (3, 4). Based on their “molecular architecture” (5), these antibiotics inhibit peptidyltransferase and block the transfer of amino acids from aminoacyl-tRNA to polypeptide (6). Aspiculamycin, isolated from S . toyocaensis var. aspiculam ycetius was reported to be a seryl homolog of gougerotin (7, 8). However, the synthesis of gougerotin and “seryl gougerotin” (i.e., aspiculamycin)
* Mildiomycin is another pyrimidine nucleoside antibiotic isolated from Streptomyces rimofuciens No, B-98891 (T.Kishi, personal communication).
198
ROBERT J. SUHADOLNIK
6 PLICACETIN
3 AMICETIN CAMICETIN A) R-CH3;
R,=H
RP CCH312N
4 BAMICETIN CAMICETIN C 1 R-H;
RPH
7 NORPLICACETIN R' CH3HN
5 OXAMICETIN RPCH3 ;S'OH CH20H I HOFH
r OH
(
HlKlZlMYClN ANTHELMYCIN 1 8
and their physical and chemical properties differed (9,10). Reinvestigation of the structure of the product isolated from S . toyocaensis by NMR and amino-acid analysis revealed that aspiculamycin is identical with gougerotin. Similarly, asteromycin (11) and moroyamycin (12) are identical with gougerotin. Hikizimycin (8), isolated from Streptomyces A-5 (13),is identical with anthelmycin isolated from S. Zongissimus (14) and contains a unique CI1 sugar (15, 16). Amicetin A and C, isolated from S . uinaceus-drappus (17,18),are identical with amicetin and bamicetin,
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
199
respectively (personal communication, Dr. A. Argoudelis). Oxamicetin (5) has been isolated from the culture filtrates of Arthrobacter oxamicetus (19, 20) .Because it differs from amicetin by an additional hydroxyl group in the disaccharide moiety, it is conceivable that oxamicetin is a precursor of amicetin. The most recent of the disaccharide cytosine antibiotics, norplicacetin (7), was isolated from a Streptomyces from a soil sample from Ghana (21). A number of laboratories have studied the mechanism by which the pyrimidine nucleoside antibiotics block protein synthesis as nonfunctional analogs of aminoacyl-tRNA by the inhibition of ribosomal peptidyltransferase. One technique used extensively involves the formation of a peptide bond between CCA-fMet and puromycin (the fragment reaction) (22).Studies on the effect of gougerotin, blasticidin S, amicetin, bamicetin, and plicacetin on peptide formation showed that blasticidin S inhibits the transfer of lysine peptides from (Lys),tRNA to puromycin (Fig. 1A) and the transfer of AcPhe from tRNAAcPhe to puromycin (Fig. 1B) (3, 23). The same type of inhibition was observed with the fragment reaction, -CACCA-(AcLeu). Clarification of the structural requirements essential for the inhibition of the traiispeptidation step was obtained by studying the increase in the binding of the donor substrate to the acceptor site ( 3 ) .Plicacetin was the least inhibitory. Similar findings were reported from other laboratories (24-27). The seryl moiety in amicetin and bamicetin appears to be a structural feature essential for the inhibition of transpeptidation. The absence of the seiyl group results in an antibiotic (plicacetin) with decreased properties. Similarly, the blasticidin S-gougerotin subgroup was only slightly inhibitory with the CAC CA-Phe. 'I
' e logh
FIG. 1. The eEect of blasticidin S (A), amicetiii (O), bamicetin
(a),and plicacetin
( 0 )on the transfer of Iysine peptides from (Lys),,-tRNA(A) and of the AcPhe residue
from tRNA (B) to puromycin. From c e r n i et
(11.
(3).
200
ROBERT J. SUHADOLNIK
Many compounds considered to be inhibitors of peptide-bond synthesis do not inhibit the peptidyl-puromycin synthesis when native Escherichia coli polyribosomes are used (28-30). The data were obtained in systems involved in the formation of fMet-, AcPhe-, and polylysyl-puromycin from synthetic donors with ammoniumchloride-washed ribosomes. Pestka cautions that the data obtained with model systems do not necessarily predict the behavior of antibiotics in the intact cell, The types of inhibition observed with amicetin, gougerotin, and blasticidin S on peptidyl-puromycin synthesis with native polyribosomes from E . coli are both competitive and noncompetitive (29,30). This could mean that either two sites exist for interaction with peptidyltransferase, or there are two classes of ribosomal states that are amenable to inhibition. There is one homogeneous binding site for gougerotin per E . coli ribosome; with Saccharomgces cerevisine ribosomes, the binding to gougerotin is heterogeneous (31).In addition, gougerotin has a much stronger affinity for washed E . coli ribosomes than for ribosomes reconstituted from subunits (Fig. 2). Blasticidin S completely inhibited the binding of [G-3H]gougerotin at the peptidyltransferase binding site of prokaryotic and eukaryotic ribosomes. Similar studies with ribosomes from rat liver or brain also showed two types of inhibition of peptidyl-puromycin synthesis (32). The binding of gougerotin to the ribosome is inhibited by blasticidin S and amicetin (3234). The effect of amicetin and gougerotin on total peptide chain termination has also been studied (35). B. Purine Nucleoside Antibiotics There are six naturally occurring purine nucleoside antibiotics that inhibit protein synthesis.
1. PUROMYCIN Puromycin (b), an aminoacyl nucleoside elaborated by Streptom yces alboniger, is a broad-spectrum antibiotic with antitumor activity (1). It specifically inhibits protein synthesis in vivo and in cell-free systems (36). Puromycin is structurally similar to the 3‘-0aminoacyladenylyl end of aminoacyl-tRNA (10). The “reversed” puromycin (11)incorporates all the features of puromycin but is devoid of the structural components toxic to animals (3739). Puromycin has been used to study (i) the mechanism of peptide bond formation (40,41), (ii) the mode of action of elongation (29), (iii) inhibitors of protein synthesis (39),and (iv) movement of the protein synthesis initiator (fMet-tRNAmet) (42). Because the level of
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
201
I
/
/
[Ribosomes or subunits] (pM)
FIG.2. Binding of [G-3H]gougerotin to Escherichia coli ribosomes. The concentration of gougerotin was 1 pM.Either untreated control ribosomes or ribosomes kept for 6 hours at 0°C under nondissociating conditions ( 0 4 ) .“Dissociated and reconstituted ribosomes” ( m a ) . Ribosomal complex reconstituted from preparations of 30 S and 50 S subunits, poly(U), and tRNA (0-0). Ribosomal complex reconstituted from preparations of 30 S and 50 S subunits and poly(U) (V- - -V).Ribosomes reconstituted from preparations of 30 S and 50 S subunits (A- - -A). 50 S ribosomal subunits (0-- -0).From Barbacid and Vazquez (31).
polysomes in cells reflects the capacity of their protein-synthesizing machinery, peptidyl-puromycin reaction has been used to quantitate the amount of polysomes (43). The puromycin-reactive 70 S initiation-complex is the working model to study the mechanism of polypeptide chain initiation (44, 44a). A hypothetical scheme for polypeptide chain initiation and the formation of this initiation complex is shown in Fig. 3. Methionylpuromycin is the first peptide bond formed in lysed rabbit reticulocytes (45,46),but its validity as a model is questionable because the formation of the first dipeptide is a unique process (47).Sparsomycin, an inhibitor of peptidyltransferase ( 4 8 5 0 ) , inhibits neither the first
202
ROBERT J. SUHADOLNIK
ny on c=o
PUROMYCIN
9
P IINOACYL-
. RNA
R tRNA R‘IALKYL GROW OF AMINO ACIDS
”
R EV ERSED” PUROMYCIN
11
10 dipeptide nor the pactamycin-induced dipeptide accumulation (47). Pactamycin inhibits methionyl-puromycin formation (51, 52), which causes an accumulation of methionyl-valine with globin mRNA (47, 53).These findings suggest that either a different peptidyltransferase is used to synthesize the dipeptide, or the ribosomes undergo a special conformation when the first peptide bond is formed. The addition of puromycin to bacterial cultures causes the breakdown of polyribosomes, an accumulation of 70 S monomers, and an increase in the exchange of ribosomal subunits (54-57).The dissociation of the 70 S ribosome requires the ribosome dissociation factor IF3 (5840). The addition of puromycin causes the ribosome to be detached as the 70 S particle, which then dissociates and equilibrates with the pool of subunits (56, 61, 62). Strains of mice resistant to audiogenic seizures can be rendered susceptible to sound-induced convulsions after exposure to an intense acoustic stimulus during a critical period of neural development. This phenomenon is referred to as “acoustic priming” (63).Puromycin or puromycin aminonucleoside (51) blocks this process. Although it appears that puromycin affects protein synthesis per se, it has been suggested that puromycin acts through an interference with normal neurohumoral transmission by blocking peptide synthesis (64). Puromycin has also been used to study the role of protein synthesis in the memory process in mice and goldfish ( 6 5 4 9 ) .Puromycin blocks
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS 30 S Subunit
+
203
+ IF3
mRNA IF2+GTP
1
+ 1, f Met
TRANSFER
Puramycin Unreoctive 70 S Intermediate
Purornycin Reactive 70 S Initiation Complex
FIG.3. Hypothetical mechanism for polypeptide chain initiation in E . coli. Symbols
are
1
, Met-tRNA,; I-A-U-G-I, messenger RNA showing A-U-G at initiation site.
f Mei
Although IF-1 is released concommitant with the 70 S complex formation, subsequent interaction of IF-1 with IF-2 bound to the 70 S complex is necessary for the release of IF-2, In the absence of IF-1, IF-2 remains bound to the puromycin-reactive 70 S initiation complex ( 4 4 0 ) . Modified from Dubnoff et d.(hl).
the postjunctional response to acetylcholine, and puromycin and puromycin aminonucleoside (51)are reversible mixed inhibitors of acetylcholinesterase (71). The observation that puromycin can interfere with the acetylcholine receptor (70) and also inhibit acetylcholinesterase (72) may be important to the interpretation of experiments in which puromycin is used to interfere with memory. The diaminonucleoside and an aromatic amino acid of puromycin are essential for maximum inhibition of protein synthesis (1).HOWever, the dimethyl groups, the methoxyl group, the furanosyl oxygen,
204
ROBERT J. SUHADOLNIK
and the 5’-hydroxyl group of puromycin are not necessary for biological activity (73-76). For example, a carbocyclic puromycin analog in which the furanosyl oxygen is replaced by a methylene moiety has antimicrobial and antitumor activity (73, 77). More recently, four cyclohexyl puromycin derivatives that inhibit protein synthesis have been synthesized (78).
2. HOMOCITRULLYLAMLNOADENOSINE AND LYSYLAMINOADENOSINE Homocitrullylaminoadenosine (12)and lysylaminoadenosine (13) can be isolated from the culture filtrates of Cordyceps miEitaris (79). Homocitrullylaminoadenosine is similar to puromycin as an inhibitor of protein synthesis (80).Although it does not inhibit protein synthesis by either blocking the activation of amino acids or the transfer of tRNA, it does inhibit the overall incorporation of amino acids from aminoacyl-tRNA into protein. It is assumed that lysylaminoadenosine inbibits i n much the &me way.
6)
HoH2$2 HoH2cd HN OH I
O=C-CH-KH,),-NH-~NH, NH2
O’C-CH-CH2-CH2-CH2-CH2NH2
0
HOMOCITRULLYLAMINOADENOSINE 12
AH,
LYSYLAMINCNDENOSINE
13
3. NUCLEOCIDIN(ANTIBIOTIC T-3018) Nucleocidin (4’-fluoro-5’-O-sulfamoyladenosine) (14) is elaborated by S. clavus nov. sp. The proof of structure of nucleocidin is based on NMR, mass spectral studies, and chemical syn’thesis (81, 82). Nucleocidin inhibits the incorporation of leucine into rat liver protein in vivo and in vitro. Although it is a more potent inhibitor than puromycin in in vivo studies, the inhibition of protein synthesis by nucleocidin and puromycin in vitro is essentially the same. The differences between the in vivo and in vitro inhibition of protein synthesis by nucleocidin and puromycin has been attributed to the slower metabolism and excretion of nucleocidin as compared to puromycin. The inhibition of protein synthesis by nucleocidin appears to involve its binding to the ribosomes, with subsequent inhibition of peptide bond formation. Nucleocidin does not affect the binding of tRNA to ribosomes, nor does it inhibit RNA synthesis.
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
205
NUCLEOClDlN 14
4. A201A A201A (15) has a structural resemblance to puromycin (9). It was isolated from a new strain of S. capreolus (84).Antibiotic A201B has been isolated from the same organism.
HoHzck9 HY
OH
OH
OCH3
A201A 15
2’-AMINOGUANOSINE 16
GUANOSINE
17
206
ROBERT J. SUHADOLNIK
A201A inhibits the incorporation of leucine into protein (85).However, unlike puromycin, A201A does not serve as an acceptor in polypeptide elongation. It stimulates the release of nascent polypeptides. Because puromycin (9) is an analog of aminoacyl-tRNA (lo), it can serve as an acceptor of the growing polypeptide chain. In contrast, A201A does not carry a reactive amino group; this precludes any acceptor activity. A201A does not inhibit peptidyltransferase, or RNA or DNA synthesis, but it does inhibit protein synthesis. When tested against cellfree poly(U)-directed synthesis of poly(Phe) on E . coli ribosomes, there was a marked inhibition of poly(Phe) synthesis. The antibiotic does not interfere with the binding of AcPhe-tRNA to the salt-washed ribosomes in the presence of the protein initiation factors. However, the reaction between bound AcPhe-tRNA and puromycin is very sensitive to A201A (85),which selectively inhibits dipeptide synthesis by interfering with the formation of a puromycin-reactive 70 S initiation complex. A201A does not inhibit the formation of the initiation complex, but does inhibit dipeptide formation. Similar results were obtained with A201A using tRNAmet.Epp and Allen (85)proposed that A201A interferes with the “joining” reaction, that is, the joining of an initiation complex to 50 S subunit. This should cause polyribosome “runoff ’; the ribosomes would finish one round of synthesis but fail to start to translate another message. Sucrose density gradient centrifugation of lysates from exponentially growing E . coli treated with puromycin or A201A showed that A201A caused a runoff of ribosomes much as did puromycin.
5. 2’-AMINOGUANOSINE The first evidence for the occurrence of 2‘-amino-2‘-deoxyguanosine [4(2-amino-2-deoxy-~-~-ribofuranosyl)guanine] (16) as a naturally occurring nucleoside antibiotic was its isolation from Aerobacter, which belongs to the Enterobacteriaceae (86). It and isoguanosine (crotonoside) (Section V, 3) are the first examples of naturally occurring analogs of guanosine (17). It inhibits E . coli strain KY3591 and has antitumor activity against HeLa cells and Sarcoma 180. The discovery, production, isolation, physical and chemical properties, and structural elucidation have been described (86, 87). The chemical synthesis has been reported (88,8Q). Of 20 strains of E. coli tested, only strain KY3591 was sensitive to 16 (0.1puglml). Growth-resistant colonies appear after 2 hours of exposure to 16. The inhibition is reversed by guanosine and adenosine, but
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
207
not by xanthosine, guanine, adenine, hypoxanthine, or xanthine. Guanosine prevents uptake of the analog. Purine nucleosides are hydrolyzed by purine nucleoside phosphorylase or purine phosphoribosyltransferase of E . coli (go), and the bases are released into the medium. With E . coli strain KY3591, 16 prevents the release of guanine. The aminoguanosine is rapidly taken up into the cells and converted to the 5’-mono-, di-, and triphosphates. It inhibits protein synthesis in E . coli KY3591 (91), but not until about 15 minutes. In contrast, chloramphenicol blocks protein synthesis in this E . coli strain in 5 minutes after its uptake. The effect of 2’-aminoguanosine on the incorporation of thymine, uracil, and leucine into the acid-insoluble fraction is shown in Fig. 4. It inhibits RNA and protein synthesis, but not DNA synthesis. When 16 is taken up into cells, 86% appears in the acid-soluble fraction as the nucleotides; of the acid-insoluble fraction, 95% is in the RNA. Nakanishi et al. (91) proposed the following three inhibitory mechanisms to explain the effect of 2’-aminoguanosine on protein synthesis. (i) Its 5’-triphosphate acts as an analog of GTP, which is re-
-30
0
30
60
I n c u b a t i o n time ( m i n )
90
-30
0
30
60
I n c u b a t i o n t i m e (niin)
FIG. 4. Effects of 2’-aminoguanosine (16)on macromolecular syntheses in Esclzerichiu coli KY3591. (A) Incorporation of [3H]thymine and [14C]uracilinto the acidinsoluble fraction. Incorporation of [3H]thyniine in the presence (1)or absence (2) of 16. Incorporation of [14C]uracilin the presence (3)or absence (4) of 16. (B) Incorporation of [14C]uracil and ~ - [ ~ H ] l e u c i n into e the acid-insoluble fraction. Incorporation of [14C]~racil in the presence (5)or absence (6)of 16. Incorporation of ~ - [ ~ H ] l e u c i nine the presence (7) or absence (8) of 16. The arrows indicate the time of addition. From Nakanishi et af. (91).
208
ROBERT J. SUHADOLNIK
quired for the initiation step of protein synthesis; initiation is blocked and the elongation reaction cannot contiiiue for several minutes. (ii) The 5’-triphosphate acts as an analog of GTP,which is required for the elongation step of protein synthesis. (iii) The 5’-triphosphate is incorporated into messenger, transfer, and ribosomal RNA in place of GMP; therefore, these RNAs do not function normally and protein synthesis is inhibited. Nakanishi et al. (91) proposed that the mode of action of 16 involves its incorporation into RNA. This results in the formation of nonfunctional RNA, and subsequently protein synthesis is prevented.
6. CLINDAMYCIN RIBONUCLEOTIDES Argoudelis et al. described the acylation of chloramphenicol by S . coelicotor (92), the phosphorylation of lincomycin by S . rochei (93), the conversion of clindamycin (18)to 1-demethylclindamycin and clindamycin sulfoxide by S. punipalus and S . armentosus (94), and the phosphorylation of clindamycin by whole cells and lysates of S . coelicolor (95).Clindamycin (18),a clinically useful antibiotic that is produced by the chlorination of lincomycin (92), inhibits protein synthesis. The addition of clindamycin to growing cultures of S . coelicolor yields inactive clindamycin &phosphate (19), and four new nucleotides: clindamycin 3-(5’-cytidylate) (20), clindamycin 3 (5’-adenylate)(21), clindamycin 3(5’-uridylate)(22), and clindamycin 3(5’-guanylate) (23) (95).
18 CLINDAMYCIN R= H
19 CLINDAMYCIN 3-PHOSPHATE
!
R= -P-OH OH
20 21 22 23
R = CYTIDYL-S‘-YL R = ADENYL-5I-YL
R’ URIDYL-5’-YL R = GUANYL-5’-YL
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
209
Clindamycin 3-ribonucleotides and clindamycin 3-phosphate do not inhibit cultures of S. aureus (94-96), but they do protect S. aureus-infected mice. The in vivo activity of clindamycin Sribonucleotides is presumably due to conversion to clindamycin by hydrolysis by phosphodiesterase and alkaline phosphatase.
II. Inhibitors of RNA Synthesis When the review on the naturally occurring nucleoside antibiotics was written in 1970, there were ten naturally occurring nucleoside antibiotics that inhibited RNA synthesis ( I ) . Today, that number has increased to one nucleotide and thirteen nucleoside analogs.
1. 2 ’-AMINOGUANOSINE 2’-Amino-2’-deoxyguanosine (2’-aminoguanosine, 16) was discussed above (Section I, B, 5) as an inhibitor of protein synthesis. As can be observed from the data in Fig. 4,it also inhibits RNA synthesis. Although 86% of the tritium labeled substance taken up by E. coli was found in the acid-soluble fraction as nucleotides, a small amount was isolated from the RNA. Nakanishi et al. proposed and favor the idea that the 5’-triphosphate of 16, after its incorporation into either mRNA, tRNA, or rRNA (in place of GMP residues), would produce nonfunctional RNAs that would inhibit protein synthesis (90, 91). Their reasoning is based on the observations that (i) 2’-aminoguanosine is incorporated into RNA, and (ii) poly(2’-aminouridylate) and poly(2’aminocytidylate) are not hydrolyzed by ribonuclease (88).Additional in vivo and/or in vitro studies with 16 and/or its phosphorylated derivatives will be necessary to clarify the exact mode of action of this guanosine analog. 2. Q‘-AMINOADENOSINE
3’-Amino-3’-deoxyadenosine (3’-aminoadenosine, 24) is a naturally occurring purine nucleoside antibiotic isolated from the culture filtrates of Cordyceps militaris, Aspergillus nidulans, and Helminthosporium . 3’-Aminoadenosine (24) inhibits RNA polymerase but not DNA polymerase (97).It has also been used to study the aminoacylation step in protein synthesis; it replaces the adenylyl residue at the 3’-terminus of tRNA (98). However, phenylalanine was covalently bound to the 3’-amino group in this modified tRNA. Although the phenylalanyl(3’-aminoadenosy1)-tRNA was bound to the ribosomes, the amide
210
ROBERT J. SUHADOLNIK
5
Hw2cd H0H2 CHZ
H2N
6H
3’-AMINOADENOSINE 24
Hi)
OH
ARISTEROMYCIN
25
bond was not cleaved. Therefore, the tRNA-n3’A-Phe has acceptor activity, but tRNA-n3’A-Phe does not have the donor activity essential for protein synthesis. The failure to act as a donor for protein synthesis is attributed to the amide linkage, which is considerably more stable than the ester bond formed in aminoacyl-tRNA.2Puromycin is not an analog of adenosine, but 3’-deoxyadenosine (cordycepin) and 3‘-aminoadenosine are.
3. ARISTEROMYCIN Aristeromycin, 4[(l R , 2S, 3R, 4R)-2,3-dihydroxy-4(hydroxymethyl) cyclopentyll adenine (25), is a carbocyclic analog of adenosine. The racemic mixture was first synthesized by Shealy and Clayton (99). The naturally occurring nucleoside antibiotic was isolated from S. citricolor (100). Human epithelial (H. Ep.) No. 2 cells phosphorylate aristeromycin (101). Aristeromycin is not incorporated into either RNA or DNA by these cells in in vitro studies; DNA-dependent RNA polymerase incorporates aristeromycin 5’-triphosphate into RNA in competition with ATP (102).Aristerornycin is toxic to H.Ep.#2 cells deficient in adenosine kinase. Therefore, nonphosphorylated aristeromycin is inhibitory. The toxicity is reversed by adenosine (103). Aristeromycin inhibits AMP synthesis, but does not interfere with the phosphorylation or deamination of adenosine. Aristeromycin 5‘triphosphate cannot replace ATP in the synthesis of NAD+ (103,104). One of the cellular reactions affected by aristeromycin is transmethylation. S-Aristeromycinyl-L-homocysteine (AriHcy) inhibits S-adenosylmethionine-dependent catechol-0-methyltransferase, phenethanolamine N-methyltransferase, histamine Nmethyltransferase, and hydroxyindole-O-methyltransferase (105, 106). Because of the sensitivity of the transmethylases to the See Sprinzl and Cramer in this volume.
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
211
S-aristeromycinyl analog of AdoHcy, it is possible that the “capping” of viral and eukaryotic mRNAs by the eukaryotic or viral mRNA methyltransferase would be inhibited (107).3The methylated structure, m’GpppN(m), at the 5‘ temiinus of mRNA is important for efficient translation and the binding of mRNA to the ribosomes (108-110). Although S-aristeromycinyl-L-homocysteine has marginal inhibitory properties against Newcastle disease guanine-7-methyl-transferase, this enzyme was strongly inhibited by S-tubercidinyl-L-homocysteine (106). Using the 2’-deoxy- and 3’-deoxyadenosines as well as aristeromycin (in which the 0-1of the ribose of adenosine has been replaced by a methylene group, Coward et al. showed that these AdoHcy analogs had low inhibitory properties (111, 112); they concluded that the methy lases have a very high degree of specificity. Aristeromycin (50 p M ) inhibits the growth of rice leaf, rice root, and many grasses; it inhibits cell division and elongation, and hence (at 2 g/acre) is used to control the growth of grass. Aristeromycin may thus be called a “chemical lawnmower” (113). Because aristeromycin is so cytotoxic to mammalian cells, several derivatives have been synthesized to overcome this toxicity. They are, 2‘-deoxy-, 3‘-deoxy-, 3‘-amino-3’-deoxy-, 3’-amino-3’-deoxyarabinofuranosyl-, 6-hydroxy-, emercapto-, 8-bromo-, 8-hydroxyaristeromycin and aristeromycin 3’, &cyclic phosphate (114-116). 3‘-Amino3’-deoxyaristeromycin inhibits herpes simplex virus and vaccinia virus. Aristeromycin 5‘-diphosphate and 8,2’-anhydro-8-mercapto4P-D-arabinofuranosyl 5’-diphosphate have been used to show that there are two or more binding sites for polynucleotide phosphorylase (102, 117).
4. ~AZACYTIDINE The s-triazene ribonucleoside antibiotic, 5-azacytidine (4aminol-P-~-ribofuranosyl-l,3-~triazin-2-one) (26), was synthesized in 1963 by Pliml and Sorm (118). It was subsequently isolated from culture filtrates of S. Zadakanus by Hahka et al. and Bergy and Herr (119, 120). The 5,edihydro derivative (27), has been synthesized. SAzacytidine is cytostatic, affecting a number of reactions in the cell (121, 122). Because it interferes with many cellular metabolic processes, its action is considered to be polyvalent (1, 123, 124). The incorporation of Sazacytidine into RNA and the subsequent See Part I of Vol. 19 of this series (articles by Furuichi et al., Rottman et al., Busch et al., Moss et d), all of which deal with the “cap” of mRNAs.
2 12
ROBERT J. SUHADOLNIX
HoH2cd HO OH
26 5-AZACYTIDINE XIN R'H
27 5,6-DIHYDRO-5-AZACYTlDlNE X=NH RgH2
effect on protein synthesis have been studied in many laboratories K, for the triphosphate of SazaC is 18fold greater than the K , for CTP. Several mammalian cell lines have been used to study the effect of SazaC on the transcription of a specific mRNA and the subsequent translation into protein. It is incorporated into mRNA and tRNA, which then become nonfunctional in normal protein synthesis. These abnormal RNAs show different elution patterns on DEAE-cellulose (125-130). The
(131-136).
SAzacytidine also inhibits the maturation of rRNA. The formation of 28 S and 18 S, but not of 38 S RNA, is severely inhibited (Fig. 5) (137-139). It increases the degradation of polysomes with the subsequent accumulation of monosomes (129,133,139,140). The kinetics of SazaCTP may be compared with CTP using DNA-dependent RNA-polymerase from calf thymus and E . coli (141).The kinetic data clearly show that SazaCTP is a weak inhibitor of CTP and does not compete for the incorporation of UTP into RNA. In human leukemic cells, a very active cytidine deaminase rapidly deaminates Sazacytidine to 5-azauridine. Compared with the aglycon of 5-azaC, following incorporation into RNA, the aglycon of 5-azaU is unstable (142). This deamination by cytidine deaminase is markedly inhibited by 04,4,5,6-tetrahydrouridine(142). 5-Azacytidine is phosphorylated, reduced, and incorporated into DNA. Cytidine inhibits this phosphorylation (125,126,143-150). Chroinosomal breakage occurs in the S and G2 phases after the incorporation of the 2'-deoxy analog of 5-azaC into DNA (147). The explanation offered for the chromosomal breakage is that the incorpo-
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
213
Distanceof migratirn,cm
FIG. 5. Inhibition of processing of rRNA maturation b y Sazacytidine in cultured Novikoff hepatoma cells. Cells were treated with 0.5 mM cytidine (A) or 5azacytidine (B) in the presence of [U-3H]guanosine (0.2pCi/O.OSpmoYml).RNA was isolated after 2 hours of labeling. From Gihik et al. (139).
ration of 5azadC results in a less stable secondary structure (127,147, 151 -153). The 5 azadC-containing DNA has a lower molecular weight and a lower melting point (151). The administration of 5-azaC an hour before partial hepatectomy inhibits thymidine and thymidylate kinase and DNA synthesis (153). In sea urchin embryos, 5azaC inhibits DNA synthesis by 90%;however, there is no effect on cell division (154). Based on these and other studies, Simpson and Baserga (155) suggest that there are two separate periods of RNA synthesis in the cell cycle that are needed for subsequent DNA synthesis. In prokaryotes, SazaC is incorporated into RNA and DNA and causes a marked decrease in transforming ability following such incorporation (125,146). Fi-Azacytidine changes the synthesis and activity of induced enzymes of liver, especially the amino-acid metabolizing enzymes. Induced enzyme formation is totally inhibited if 5azaC is given before or with the inducer (228). The tryptophan oxygenase of liver was higher for animals that received 5-azaC 1f3-30 hours before enzyme induction as compared to controls (156). 5Azacytidine decreased all the enzymes in the polyamine biosynthetic pathway in L-1210 leukemic mice. The accumulation of polyamines in leukemic mice is inhibited (157), but is restored to normal when the administration of 5-azaC is stopped (157,158). When 5aza-2’-deoxycytidine (5-azadC) is added to cultures of E . coZi deficient in cytidine deaminase, the nucleoside is deaminated and
2 14
ROBERT J. SUHADOLNIK
hydrolyzed, and the 5azauracil produced then enters the cell (159, 160). SAzadeoxycytidine lowers the level of the acid-soluble pool of 2’-dAMP in AKR mouse leukemic cells, inhibits the incorporation of deoxycytidine (161),and is incorporated into the DNA (162,163).The deamination of 5-aza-2’-deoxycytidine in Ehrlich ascites cells is prevented by 04,4,5,&tetrahydrouridine (163). Because 5-azaC is unstable and can be deaminated to the unstable 5-azauridine (142),the design of a suitable, stable analog that would have equal or better therapeutic effects with less toxic properties has been undertaken. The dihydro derivative, 5,6-H2-5-azaC(27) has biological antagonist properties that are similar to 5azaC against L-1210 leukemic cells (164-166). 5-Azacytidine is toxic to animals. Beagles (the animal most sensitive) show a decrease in leukocytes and necrosis of lymphatic organs (167, 168); the toxicity is reversed by cytidine (168).5-Azacytidine administered to beagles (i.v.) was excreted unchanged, also as 5-azacytosine7 5-azauracil, and urea- and guanidinelike compounds (169).In mice, tetrahydrouridine, araC, vincristine, or prednisone increased the amount of 5azaC in the urine &fold. Man can tolerate much higher doses of it than can beagles, rodents, and monkeys (1 70). Although early studies indicated that 26 is incorporated into the DNA of bacteria and mammalian tissues (151,171,172),subsequent studies showed no incorporation of 5-azaC into mammalian DNA. However, there is incorporation into the RNA (173, 174), but none appears as
co, (174).
5Azacytidine has two unusual effects upon G 1210 leukemic cells (175)-a biphasic dose response, and a prolonged antileukemic effect that lasts many days after administration. Of patients with acute myelogenous leukemia, 36% responded to 5-azaC (1 76-181 ). Combination therapy improved the response to 68% (180,182).Patients with acute lymphatic leukemia, treated with 5azaC, showed a partial remission (1 79,183-186).The drug has significant activity in acute nonlymphoblastic leukemia (176).The best response was to a regimen of intravenous administration at %hour intervals for 5 days. The complete remission rate was 27%. Toxicity included moderate-to-severe nausea, diarrhea, stomatitis, myelosuppression, and neurological side effects. 5-Azacytidine shows less encouraging results in solid breast tumors, lung, colon, rectum, malignant myeloma, and miscellaneous tumors (124, 183-191).
5. BREDININ Bredinin, 5-hydroxy-l-~-~-ribofuranosyI-1H-imidazole-4carboxamide (28), was isolated from Eupenicillium brefeldianum M2166
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
215
(192). It is a derivative of the 5-amino ribonucleoside.(AICA nucleoside (29)and is structurally similar to pyrazofurin (90) and the synthetic nucleoside antibiotic ribavirin (91). The 5-aminopyrazole analog of bredinin, (4amino-3-/3-~-ribofuranosylpyrazole-5-carboxamide) has been synthesized (193,194).
HoH2ct$HoH2ckY HO OH
BREDININ
28
HO OH
5-AMINO -4-IMIDAZOLECARBOXAMIDE RIBONUCLEOSIDE (AKA 1
29 Bredinin has beneficial effects on experimental rheumatoid arthritis (195). Suppression of adjuvant polyarthritis in rats by bredinin is due to the inhibition of antibody formation and/or the multiplication of sensitized lymphocytes by a possible antigen. Bredinin is also an excellent immunosuppressive nucleoside. It slightly decreases the peripheral leukocytes. Bredinin was not lethal to mice receiving 5000 mg/kg (i.p.) or 1500 mgkg (i.v.) for 9 days (195). The cytotoxic effect of bredinin on L5178Y cells and inhibition of multiplication of several cell lines can be reversed by AMP and GMP (196).Bredinin (at lop5M) causes chromosoinal aberrations in L5178Y cells in culture 1 hour after exposure (196, 197); these are prevented by GMP. Sakaguchi et al. (196) suggest that bredinin blocks the conversion of either IMP or XMP, or of XMP, to GMP. The inhibition of GMP synthesis by bredinin is similar to the inhibition of XMP aminase by psicofuranine and decoyinine and the synthetic nucleoside ribivarin, which inhibits IMP dehydrogenase (198). The aglycoii of bredinin is a s cytotoxic to L5178Y cells as bredinin itself. GMP, guanosine, and guanine reverse the cytotoxicity of both compounds. Both inhibit the incorporation of uridine and thymidine into RNA Lnd DNA, but the incorporation of leucine into protein is not inhibited (Fig. 6). Adenine, but not adenosine or AMP, reverses the growth inhibition by the aglycon (199). Because GMP does not completely reverse bredinin inhibition, the possibility exists that it acts on another site in the cell. GMP reverses the cytotoxic effect of bredinin on L5178Y cells (196, 200). The cytostatic effects of bredinin were reversed only by GMP if CAMP was
2 16
ROBERT J. SUHADOLNIK
hours
hours
hours
FIG. 6. Effects of bredinin and its aglycon on the synthesis of DNA, RNA, and protein in L5178Y cells. Bredinin and the aglycon were added at time zero at 2 x lV5 M. [3H]Thymidine (A and D), [3H]uridine (B and E), and [3H]leucine (C and F) were added simultaneously. A, B, and C, effect of bredinin. D, E, and F: effect of the aglycon. 0 4 , Cpm in the absence of bredinin or the aglycon; 0 4 , cpni in the presence of the aglycon. From Sakaguchi et u1. (199).
present; the cells did not survive in the absence of GMP. Although CAMP influences the secondary cytostatic effect of bredinin, it does not influence the primary cytotoxic effect, which is reversed by GMP. Bredinin is not incorporated into either RNA or DNA (201). The 5’-monophosphate of bredinin is not active against C. albicans, but it is toxic to L5178Y cells (200). Bredinin 5’-monophosphate, like ara-AMP in mammalian cell cultures, (Section II1,l) can enter the mammalian cell. The dose responses of bredinin and bredinin 5’monophosphate for anti-11210 activity in mice are the same (200). 6. CORDYCEPlN (Q’-DEOXYADENOSINE,3’-dA) Cordycepin (30) was the first naturally occurring nucleoside antibiotic to be isolated (202). It is isolated from the culture filtrates of Cordyceps militaris and Aspergillus nidulans (202-204). Although cordycepin is a cytostatic agent and an isomer of 2’-deoxyadenosine, biologically it competes with adenosine, but not with 2‘deoxyadenosine. The toxicity of cordycepin in eukaryotes and prokaryotes is reversed by adenosine, but not by 2’-deoxyadenosine, Cordycepin has been the subject of intense studies in prokaryotic and eukaryotic RNA, DNA, and protein synthesizing systems, and in
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
217
HoH2ckY H OH
CORDYCEPIN C 3‘-DEOXYADENOSINE
1
30 viral replication. It inhibits RNA and DNA synthesis (205),is cytotoxic to H.Ep. #1 cells, and decreases RNA and protein synthesis (206). Fetal calf serum has a very active adenosine deaminase (207),so it is not possible to calculate the level of cordycepin to which the H.Ep. #1 cells were exposed. Since cordycepin lacks a 3’-hydroxyl group, its incorporation into growing RNA chains would block further elongation. The idea that cordycepin is an RNA chain terminator was first approached experimentally by Cory et al. (207),who reported that [G-3H]3’-dAMP was incorporated into the 3‘-terminal position of RNA. More recently, the deamination of cordycepin in L cells was overcome by using the adenosine deaminase inhibitor erythro-9-(2hydroxy-3-nony1)adenine (64) (208). Cordycepin plus 64 reduces the incorporation of uridine into RNA and of thymidine into DNA. The inhibition of DNA synthesis was attributed to the inhibition of RNA primer synthesis. The antitumor activity of cordycepin in cell culture systems and in mice bearing P388 ascites leukemia is enhanced by the adenosine deaminase inhibitor, 2’-deoxycoformycin (62)(209)(Section IV). Toluene-treated E . coli cells show an inhibition of the ATP-dependent DNA replicative apparatus when 3’-deoxyATP (3‘-dATP)is added (Fig. 7). This inhibition is competitive with ATP, but not with 2‘-dATP. Gumport et al. state that “one possible mechanism by which this analog may interfere with DNA synthesis is at the initiation step involving the synthesis of the primer RNA” (210). DNA synthesis is not inhibited b y cordycepin (211).This observation is expanded by Muller et al., who added [G-3H]cordycepin to mouse L5178Y cells in culture (212). The primary inhibitory effects were directed toward RNA and protein synthesis; DNA synthesis was not inhibited. The cordycepin was incorporated into the 3’ terminus of RNA and was found in different RNA species (28 S, 10 S, 5 S, and 4 S), but the incorporation was not uniform. It was not hydrolyzed either intracellularly or in culture medium into adenine and 3-deoxyribose.
218
ROBERT J. SUHADOLNIK
A
Control
Time (min)
FIG.7. The kinetics of DNA synthesis in the presence of increasing amounts of 3’-dATP.ATP-dependent (filled symbols) and ATP-independent (open symbols) syntheses were tested. Control with no 3’-dATP (0,0), 0.1 mM 3‘-dATP (A,A), 0.2 mM 3‘-dATP (.,El), and 0.4 mM 3’-dATP (+,O). From Gumport et al. (210).
In contrast to the findings of Suhadolnik et al. (207),Muller et al. showed that DNA polymerase-cr and -p from mouse lymphoma cells are not inhibited by 3‘-dATP; they also showed that the inhibition of poly(A) polymerase by 3’-dATP was of the competitive type. These findings agree with the demonstration that 3’-dATP is a competitive inhibitor of rat liver nuclear-chromatin-free poly(A)-polymerase (213). However, with chromatin-associated poly(A)-polymerase, 3’-dATP is a noncompetitive inhibitor (214). At least 80 times more 3’-dATP is needed for a 50% inhibition of free nuclear poly(A)-polymerase compared to the chromatin-associated enzyme. The inhibition of DNAdependent RNA synthesis exhibits a dose-response similar to that of free poly(A) polymerase (213).These findings offer (i) a mechanism for the selective inhibition of initial polyadenylylation of hnRNA in vivo by cordycepin, and (ii) a satisfactory explanation for the indiscriminate effect of 3’-dATP on “free” poly(A) and RNA polymerase. It also appears that one 3‘-dAMP residue is incorporated into the 3’ end of oligo(A) (212).Therefore, 3’-dATP is an inhibitor and a substrate for poly(A) polymerase. Cordycepin triphosphate (3’-dATP) also inhibits the reactions catalyzed by E . coli RNA polymerase and by mouse myeloma RNA polymerase I1 (215).This effect is due to the competitive inhibition of
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
219
ATP and to the actual incorporation of 3'-dAMP into the growing RNA chains, thereby preventing further elongation. Chemical and enzymic hydrolyses of RNA chains labeled with either [GsH]3'-dAMP or [32P]3'-dAMPhave verified that the position of the labeled 3'-dAMP is at the 3' end. The powerful inhibitory effect of cordycepin nucleotides is demonstrated by the competitive inhibition of ATP binding to E . coli RNA polymerase (&,(ATP) = 21 pM), b y 3'-dATP (Km(3t-dATP) = 1 p M ) , and 3'-dADP (Ki(3t-dADp) = 143 p M ) as compared to the competitive inhibition shown by ADP (&ADPI = 1040 p M ) and the competitive inhibition of 2'-dATP (ki(Z,-dATP) = 390 p M ) and 2'-dADP (K1(2t-d,4DP) = 3630 pM). This striking ability of cordycepin nucleotides, but not of adenosine or 2'-deoxyadenosine nucleotides, to bind to the enzyme may originate from the preferred 3'-endo conformation of cordycepin over that of other adenine nucleosides (216,217). Although the initial studies on the maturation of cytoplasmic mRNA indicated that cordycepin acted by inhibiting RNA synthesis (218), more recent studies (219) show that cordycepin inhibits the synthesis of completed ribosomal RNA, ribosomal precursor 45 S RNA, and tRNA in HeLa cells; DNA and protein synthesis are not affected. Several workers subsequently showed that cordycepiii has little or no effect on the synthesis of hnRNA, but selectively blocks nuclear poly(A) synthesis (220-224), the enzyme for which was reported as early as 1960 (225). It is suggested that cordycepin is not incorporated into the 3' elid of poly(A) (221,226). This is in contrast to recent studies showing that 3'-dATP is incorporated at the 3' end of oligo(A) (212). The effect of cordycepin on nuclear synthesis and terminal turnover of poly(A) has been studied (227, 228). To determine if the syntheses of the approximately 230-nucleotide segment of nuclear poly(A) and of the terminal additional reaction are affected by cordycepin, Sawicki et al. (228)first added cordycepin to HeLa cells for 3 minutes, followed by [3H]adenosine for either 30 seconds or 2 minutes. The results4 may be summarized as follows (see Fig. 8): (i) the nucleus is the site of de no00 synthesis of poly(A); (ii) there is a nuclear and cytoplasmic 3' addition to poly(A); (iii) only those molecules bearing poly(A) (230 AMP residues or longer) exit from the nucleus to the cytoplasm; (iv) nuclear terminal addition is much more rapid than is cytoplasmic terminal addition. The addition of poly(A) sequences to mitochondria1 RNA in rat liver nuclei is also sensitive to 3'-dATP (229). See article by Darnell in this volume.
220
ROBERT J. SUHADOLNIK
Hn ANA
R
2
(AIO)iOH1
k
Nucleusi
Y Cytoplasm
3
+-(AnAonl
- A(230-n)AOH *+ATP
1
- A(2300-n+15)A
FIG. 8. Diagram of nuclear and cytoplasmic reactions involving poly(A). Terminal addition occurs in both nucleus and cytoplasm, but the de nouo synthesis of the occurs in the nucleus. This latter reaction is sensitive to cordycepin (3'dA is cordycepin). From Sawicki et al. (228).
After the discovery of poly(A) tracts in mRNA from various tissues (206-208, 230); three distinct forms of rat liver poly(A) polymerase were isolated (226).All three are inhibited by 3'-dATP, but 3'-dATP was not incorporated into the 3' end of RNA. It is suggested that the inhibition of nuclear poly(A) polymerase by 3'-dATP may be attributed to the incorporation of 3'-dAMP (231). Cordycepin has been used to elucidate the mechanism by which iron increases ferritin synthesis (232). The hypothesis that iron increases the level of ferritin in the liver by increasing polyribosomeferritin-mRNA has been refuted (233-236). Although cordycepin inhibits rRNA synthesis and polyadenylylation and the transfer of mature mRNA to cytoplasmic polysomes in many cells (218, 220, 237, 238), it has not found wide application in developmental systems. Cordycepin can block development of inner cell mass derivatives in postimplantation mouse embryos in culture (239). The addition of cordycepin (10 pg/ml) to mouse embryos explanted into culture at the two-cell, momla, and blastocyst stages caused a dose-responsive inhibition of cleavage and blastulation of these embryos (240). Most of the polyadenylylated mRNA is located in the nuclei of cotton seeds (241).After 6 hours of germination, cordycepin inhibits nuclear and polysomal poly(A) by 75%, but has little effect on protein synthesis. After 6 hours, cordycepin markedly inhibits protein syntheSee Vol. 19 of this series.
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
221
sis, presumably because the maturation of mRNA is inhibited. Cordycepin at 200 pg/ml totally inhibits RNA synthesis (242).At lower concentrations, rRNA and tRNA syntheses are selectively inhibited but protein synthesis is not affected. Cordycepin did not inhibit RNA polymerase I1 in germinating cotton cotyledons (238),nor did it inhibit polyribosome formation in wheat embryos (243). Cordycepin effectively inhibits viral replication in the presence of 5iodo-2’-deoxyuridine (244);it reduces the number of cells producing virus and has a more drastic effect on viral replication than on the transformation of normal rat kidney cells by murine sarcoma virus. Apparently, viral production is greatly reduced whereas viral transformation is only slightly inhibited (245).Cordycepin preferentially inhibits viral replication by blocking poly(A) synthesis. This inhibition interferes in the processing and maturation of cellular and viral mRNA. Cellular FWA synthesis in chick embryos infected with influenza virus is inhibited 75% by cordycepin. The replication of the virus is not inhibited (246). With parainfluenza virus, cordycepin inhibits poly(A) parainfluenza mRNA synthesis and causes an 80% inhibition of the synthesis of parainfluenza RNA. Similar findings were observed when cordycepin was added to KB cells infected with adenovirus. Cordycepin strongly inhibits virus-associated low-molecular-weight RNA, but has little effect on viral hnRNA synthesis (247).Apparently, the synthesis of viral RNAs involves either two different RNA polymerase activities or two different RNA species in the in vivo transcription of the adenovirus genome. The same conclusion was reached in studies with cordycepin with cultured feline leukemia virus (FeLV) and FeLV-associated cell-surface antigens (248).Cordycepin had no effect on either the size or the relative proportions of the 6-27 S and 18-22 S Newcastle disease virus mRNA species of embryonated hen’s eggs infected with Newcastle disease virus (249).Equally important is the observation that cordycepin does not affect poly(A) associated with RNA. This observation is in contrast to the inhibition of poly(A)-associated hnRNA in the eukaryotic cell. Replication of both human rhinovirus and poliovirus, is completely inhibited by cordycepin (250). The question concerning the mode of action of an inhibitor related to one specific reaction in the cells is always difficult to answer because the inhibitor may be pleotropic in its activity. Cordycepin is one such drug; in addition to 3’-dATP acting as an analog of ATP and inhibiting RNA synthesis, it also affects protein synthesis. Cordycepin is a potent inhibitor of a nucleotide-stimulated protein kinase from T.
222
ROBERT J. SUHADOLNIK
cruzi (251). It also competitively inhibits CAMP-dependent and CAMP-independent protein kinase from bovine heart and rat liver (252).These observations strongly suggest that cordycepin may function in vivo by affecting transcription via interference with phosphorylation of nonhistone chromosomal proteins ( 2 5 2 ~ )Cordycepin . causes a %fold stimulation of methionine incorporation into protein directed by myeloma mRNA and TMV mRNA (253). Leinwand speculates (private communication) that cordycepin stimulates protein synthesis by increasing the initiation sites on mRNA, which would increase the rate of translation. 2’,3’-Dideoxyadenosine is more stimulatory for protein synthesis than is cordycepin (Wu and Suhadolnik, unpublished results). The role of the 3’-hydroxyl group of adenosine is important in cellular reactions. Hampton and Sasaki (254) studied the substrate properties of several AMP analogs with respect to AMP aminohydrolase, snake venom 5’-nucleotidase, and AMP kinase to elucidate the adenine-ribose torsion angle of enzyme-bound AMP. They observed that structural changes impair catalysis. The anti-type adenine-ribose torsion angle is such that the H-8 is oriented in the vicinity of C-4’. Another utilization of cordycepin as a biochemical probe was the synthesis of the NAD analog, N(3’-dA)D, in which the bindings of NAD and the 3’-dA analog to the coenzyme domain of the dehydrogenases were compared (255).The data show that the K,’s andK,’s for NAD and the analog are the same; however, the V,,, of the analog is decreased by 80%.These results agree with those from X-ray diffraction, indicating that the 3’-hydroxyl group of the adenosine moiety of NAD is essential for hydrogen bonding. However, the energy of hydrogen bonding is used for the proper conformation of either the enzyme or the nicotinamide of NAD to form a productive complex. Another interesting reaction related to the role of the 2’- and 3’hydroxyl groups of NAD is in the addition of poly(ADP-ribose) to nuclear proteins. The ADP-ribose moiety of NAD is covalently attached to the y-carboxyl group of glutamate in histone-l(256), and the homopolymer of ADP-ribose is linked by an al“-+ 2’ glycosidic linkage (257).NAD, 0.5 M , inhibits DNA synthesis in nuclei isolated from rat liver by only 9%; however, 2‘- and 3’-dNAD inhibit DNA synthesis by 90% (Fig. 9). With nuclei from Novikoff hepatoma and fetal rat liver, NAD does not inhibit DNA synthesis. With N(2’-dA)D and N(S’-dA)D,DNA synthesis in nuclei from Novikoff and fetal rat liver is inhibited (258). Of the nucleoside antibiotics cordycepin, ara-A, coformycin, tubercidin, showdomycin, and formycin B, only formycin B and showdomycin inhibited the activity of polyADP-ribose synthase; cordycepin was without effect (259).
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
223
NTuD NOD
8-NAD
NFD HAD
I 0
: 0.5
2’dNAD Z‘dNAD
:
1.0
2.0
Concentration 01 NAD
3.0 or
Analog
4.0 (mM)
FIG.9. Effect of NAD and NAD analogs on the inhibition of template activity for DNA synthesis in nuclei isolated from rat liver. Replacement of [3HldTTP with [3H]dATP in the NAD experiment gave essentially the same percent inhibition. From Suhitdolnik et nl. (258).
Adenosine analogs have been used at the 3’-adenylate end of the -CCA of t W A to study the mechanism of aminoacylation (98, 260264). The use of adenosine aiialogs has elucidated a mechanism for “codon-anticodon checking” and transacylatioii that corrects misrecognition of amino acids by aminoacyl-tRNA synthases.6 Cordycepin and 5,6-dichloro-~-~-ribofuranosylbenzimidazole have been used to determine the subclasses of hnRNA (265, 266).
7. ERITADENINE Eritadenine, (2’R, 3’R)-9-(4-carboxy-2,3-dihydroxybutyl)adei1ine (31) has been isolated from the edible mushroom “Shiitake,” Lentinus edodes Sing. This edible species of mushroom has long been a delicacy in Japan because of its flavor and fragrance (267, 268). Deoxyeritadenine (32)and 9-(3carboxypropyl)adenine (33) have also been isolated from L. edodes. A diet containing 5% of the ground dried edible mushroom, L. edodes, markedly reduces the plasma cholesterol levels in rats (269). See Scheme 3 of Sprinzl and Cramer in this volume.
224
ROBERT J. SUHADOLNIK
y 2 HO-$-H HO-f-H
fH2 HO-F-H
COOH
ERITADENINE
fH2 COOH
COOH
DEOXYERITADEMNE
9-(4-CARBOXY -2,3- 9-(4 -CARBOXY-3DIHYDROXYBUTYL IHYDROXYBUTYL 1ADENINE ADENINE 31 32
9-(3-CARBOXYPROPYL 1ADENINE '
33
The subsequent isolation and crystallization of eritadenine (270) and the report (269) that natural and synthetic eritadenine had marked hypocholesterolemic effects in rats explained the biological properties of ingested, dried L. edodes. Eritadenine reduces total cholesterol levels in the serum of rats, effects an equilibration of cholesterol between plasma and tissues, lowers the liver cholesterol, and on a highfat diet depresses .serum lipid levels (270,271).The shift in the equilibrium in plasma cholesterol toward tissues by eritadenine may be related to either the similarities of the structures of eritadenine and the adenine nucleotides or the interference of eritadenine with the metabolic processes influenced by CAMP-dependent protein kinase
(272). Eritadenine does not induce fatty livers in rats (272). The free cholesterol level is lowered more than the esterified cholesterol; the cholesterol-rich lipoproteins are very sensitive to eritadenine. Administration (i.v.) of eritadenine does not elicit the hypocholesterolemic effect observed in oral administration. It may be that the intestinal wall or the liver is the site of action of eritadenine.
8. P E N T O P Y ~ I N E S Nine cytosine nucleoside antibiotics have been isolated from the culture medium of S . griseochromogenes (273-275). Five of these, designated pentopyranines A (34), B (35), C (36), D (37),and E (38), are variations of a-L-pentopyranosylcytosine; one, pentopyranine F (39), is a P-D-pentopyranosylcytosine. The related group of pD-hexopyranosylcytosine derivatives has also been isolated; these are, pentopyranamine D (40), pentopyranine G (41), and pentopyranic acid (42). Fox et al. confirmed the structures of 34,38,39,40, and 42 by total chemical synthesis or comparison with known nucleosides (276-
225
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
aN
O
Hd H
H
PENTOPYRANINE A 34
PENTOPYRANINE B
PENTOPYRANINE C
35
36
P (J
HO H
OH
OH
PENTOPYRANINE D
PENTOPYRANINE E
37
30
HO OH
PENTOPYRANAMINE D
PENTOPYRANINE G
40
41
PENTOPYRANINE
F
39
Ho@ no
OH
PENTOPYRANIC ACID
42
280). Pentopyranamine D is the nucleoside moiety of blasticidin H (278). The ribo- and deoxyribonucleosides occur in nature principally as the p-D-iSOrrIerS; the only naturally occurring nucleoside found with the (Y configuration is 5,Gdimethyl- 1-a-D-ribofuranosylbenzimidazole (in vitamin BIJ. The discovery, isolation, production, physical and chemical properties, proofs of structures, and chemical syntheses of the pentopyranines have been reported (273-282). Although the pentopyranines are not potent inhibitors of eukaryotes or prokaryotes, pentopyranines A-D inhibit the incorpora-
226
ROBERT J. SUHADOLNIK
tion of uridine into the RNA of Ehrlich ascites tumor cells (275).The toxicity of pentopyranine C injected (i.p,) into mice is LDo > 300 mg/kg* Although pentopyranines A-D are isolated from the culture filtrate that contains blasticidin S, they cannot be precursors for blasticidin because they contain one less carbon atom in the sugar moiety, i.e., blasticidin S contains a hexose. Set0 et al. (282)suggested that there is a common precursor for pentopyranines A and C and blasticidin S.
Pyrrolop yrimidine Nucleoside Antibiotics (Tubercidin, Toyocamycin, and Sangivamycin) The pyrrolopyrimidine nucleoside antibiotics, tubercidin ( 4 3 , toyocamycin (44), and sangivamycin (45), have stimulated consider-
6;) 6;; H O H Z C ~
HO OH
TUBERCIDIN 7-DEAZAADENOSINE 43
H o H 2 c j
HO OH
P
fJ$"
HoHz HO OH
TOYOCAMYCJN
SANGIVAMYCIN
7-CYANO-7DEAZA ADENOSINE
7-AMINOCARBONY L-7DEAZAADENOSINE
44 45 able research because of their action against bacteria, mammalian cells in culture, RNA and DNA viruses, and the treatment of cutaneous neoplasmas in humans. They are analogs of adenosine and are highly cytotoxic to mammalian cells in culture. Modification of the cyano group at carbon-7 of c'A(7-deazaadenosine) and the amino group at carbon-6 of adenosine markedly changes the antineoplastic and antiviral activities of these pyrrolopyrimidine nucleoside antibiotics. Although the occurrence of the pyrrolopyrimidine nucleoside antibiotics had been limited to their isolation from Streptomyces, the same ring system is found in the tRNAs of eukaryotes and prokaryotes in the nucleoside, originally called nucleoside Q' (46), that occupies tRNAAS", the first position of the anticodon ofE. coli tRNATYr,tFWAHis7 and tRNAASp(283). Based on the biosynthetic studies of the pyr'The names queuosine (symbols Quo or Q ) for the nucleoside, and queuine (Quu) for the base, were jointly proposed by Dunn and Cohn and by Nishimura, the discoverer. [Ed.]
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
227
A
H2N N'
HoH2c@ HO
OH
NUCLEOSIDE Q " ( QUEUOSINE 1 7-[( C&-4,5-DIHY DROXY-2-CYCLOPENTENI-Y L )AM1NOMETHY LI-7-DE AZ AGUANOSlNE 46 "
rolopyrimidine nucleoside antibiotics by Suhadolnik and Uematsu (284), Nishimura et al. (285) reported that guanine serves also as the carbon-nitrogen skeleton for the pyrrolpyrimidine aglycon of nucleoside Q. Of interest is the description of the synthesis of 7cyano-7-deazaguanosine from toyocamycin (286). This is a potential precursor for the biosynthesis of quevosine. 9. TUBERCIDIN Tubercidin blocks glucose utilization in bacteria and inhibits mitochondria1 respiration in Ehrlich-Lettre tumor cells (287, 288). Purine synthesis de no00 is inhibited as well as the anabolism of adenosine (289,290). Tubercidin inhibits rRNA processing (1), methylation of tRNA, protein and nucleic acid synthesis, and causes visible nuclear damage (291-294). In rat liver nuclei, tubercidin inhibits CAMP-dependent protein-kinase activity while causing a Sfold stimulation of protein-kinase activity in Trypanosoma crmxi and T. gambiense (251,295,296). Tubercidin is toxic to cells in all phases of the cell cycle (297). Tubercidin is a competitive inhibitor of ATP for the ATPpyrophosphate exchange catalyzed by Met-tRNA synthetase (Ki of tubercidin = 30 pM) (298). The most recent pleotropic eEects of tubercidin appear in the studies in which tubercidin replaces the adenosine moiety of NAD and of S-adenosylmethionine (292, 293). Tubercidin is not a substrate for nucleoside phosphorylase (299) or adenosine deaminase (300),but is phosphorylated at the 5' position by
228
ROBERT J. SUHADOLNIK
red blood cells or microorganisms (1, 287, 301, 302). Tubercidin is converted to the 5'-triphosphate in the red blood cell (289).It inhibits adenosine kinase, adenosine phosphoribosyltransferase, nucleoside phosphorylase, and other adenosine enzymes (303,304).The inhibition of bacterial cells by tubercidin can be attributed to faulty regulation of phosphofiuctokinase by tubercidin 5'-triphosphate (299).The 5'-diphosphate and 5'-triphosphate of tubercidin are substrates for ribonucleotide reductase (305,306).These findings would explain the incorporation of tubercidin into DNA. Tubercidin 5'-triphosphate can function as an initiating nucleotide for RNA synthesis, but cannot function well for the phosphodiester bond formation in the elongation step (307).The substitution of tubercidin for adenosine in S-adenosylhomocysteine yields a potent inhibitor of tRNA methylase, catechol-O-methyltransferase, indole ethylamine N-methyltransferase, methylation of tRNA in phytohemagglutinin-stimulated lymphocytes, and polyamine biosynthesis (308311). The K, and K D of NAD in which tubercidin replaces the adenosine is essentially the same as that of NAD; however, the V,,, decreased markedly (255). The kinetic data show that replacement of the adenosine in NAD with tubercidin (NTuD) does not change the binding of this NAD analog to the coenzyme domain of the dehydrogenases; it does decrease productive complex formation. The tubercidin-containing NAD is also a substrate for the ADPribosylation of elongation factor I1 by diphtheria toxin and the enhancement of protein synthesis in lysed rabbit reticulocytes (312,313). The NTuD is not effective as an inhibitor of DNA synthesis in nuclei isolated from rat liver (Fig. 9). Another form that markedly affects cellular processes is the 3',5'-cyclic tubercidin monophosphate. This analog of cAMP has much greater lipolytic activity than cAMP in adipose tissue (314). Although studies with various types of advanced neoplastic diseases treated with tubercidin have not been encouraging, it seems very effective against basal cell carcinoma, actinic keratoses, mycosis fungoides, reticulum cell sarcoma, and squamous cell carcinoma (315, 316). There was no recurrence of the basal cell carcinoma. One of the most promising clinical applications of tubercidin is as an antihelminthic agent. Adult bloodworms of Schistosoma mansoni cannot synthesize purine nucleotides and must rely on the purine salvage mechanisms for energy maintenance in the presence of tubercidin (317321). Poly(tubercidin) does not induce interferon production in the presence of poly(7-deazainosine), a potent inducer (322, 323). It ap-
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
229
pears that polytubercidin inhibits interferon production by binding to a cellular receptor site (323). SBrornotubercidin prevents induction of acetylcholine receptors in muscle cells from chick embryos, inhibits SV40 proliferation in African green monkey kidney cells and Rous sarcoma virus in chick embryo fibroblasts, and reversibly blocks the synthesis of cellular hnRNA, rRNA, and mRNA. The inhibition of viruses by 5-bromotubercidin is apparently not related to the interference of nucleic acid precursor synthesis, but rather its mode of action is at the level of nucleic acid polymerization (324). Additional evidence that encourages the application of nucleoside analogs to study DNA viruses is that those like tubercidin discriminate between different initiation species, which may limit the type of RNA formed (325). 10. TOYOCAMYCIN Toyocamycin (44) is converted to its 5’-triphosphate and incorporated into the RNA of Ehrlich ascites tumor cells (326), which seems to explain the mode of its antineoplastic activity. A more complete study showed that its incorporation into the RNA of L cells selectively inhibits rRNA synthesis (327,328).Studies on the effect of toyocamycin, tubercidin, and Gthioguanosine on rRNA maturation in cultured Novikoff hepatoma cells, showed that the processing of 45 S to the 38 S RNA is not inhibited; however, the formation of mature 28 S and 18 S RNA is inhibited ( 3 2 9 ~ )The . inhibition of rRNA maturation by toyocamycin in Novikoff hepatoma cells is shown in Fig. 10. The inhibition of rRNA maturation by tubercidin and toyocamycin suggests another possible mode for their antineoplastic activity. Toyocamycin interferes with RNA metabolism b y preventing polyadenylylation and/or methylation of adenosine residues (329b). Toyocamycin can also replace adenosine in the RNA of Saccharomyces cerevisiae (330). A limited synthesis of precursor rRNA occurs, but its processing and maturation are inhibited; 27 S and 20 S pre-rRNA accumulate. As the concentration of toyocamycin is increased, the later steps in processing are blocked, and the subsequent formation of mature ribosomes (i.e., the conversion of 27 S pre-rRNA, 25 S RNA, and 20 S pre-rRNA to 18 S rRNA) is stopped. Toyocamycin caused no increase in the newly formed RNA in the nucleolus of Chinese hamster cells in monolayer cultures (331). However, it was not clear if this was due to a turnover of RNA or if nucleolar RNA synthesis simply stopped. Toyocamycin profoundly inhibits the appearance of newly formed RNA in the cytoplasm. In toyocamycin-treated mouse and hamster cells, there is a gradual dis-
230
ROBERT J. SUHADOLNIK
CM FIG. 10. Effect of toyocamycin on [3H]uridine labeling of rRNA. Cells were treated with toyocamycin, 1 pg/ml, or adenosine, 1 pg/ml (control), in the presence of [3H]uridine (0.5 pCi/ml, 1 x lC5 M). Absorbance at 260 nm, AZm,and radioactivity, dpm, are plotted versus distance of migration, cm. (A) control, 1 hour; (B) control, 2 hours; (C) toyocamycin treated, 1 hour; (D) toyocamycin treated, 2 hours. From Weiss and Pitot (329a).
appearance of the 15@A granules from the particulate region in the nucleoli. Therefore, the region of the nucleolus that contains early rRNA precursor ultimately becomes fibrillar. Heine (332) reported similar findings. Actinomycin D, which inhibits all RNA synthesis, caused a segregation and finally a disaggregation of nucleolar components (331). Toyocamycin reversibly blocked the multiplication of embryonic chick fibroblasts infected with Rous sarcoma virus (333).At 0.1 pg/ml, it decreased the synthesis of cellular DNA and certain classes of RNA. Protein synthesis continued at disproportionately elevated rates for several days (333).Toyocamycin, although it inhibited the synthesis of ribosomal RNAs, blocked the antiviral activity of interferon in encephalomyocarditis virus-infected L cells (334). The isopentenyl, seleno, and alkylseleno derivatives of toyocamycin have been synthesized and studied for antitumor and antifungal activity (335-338).
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
231
11. SANGIVAMYCIN Sangivamycin (45) is one of the few nucleosides that have been selected for clinical studies. It has strong antileukemic activity (339). Sangivamycin 5’-triphosphate competes with ATP, is a substrate for RNA polymerase from Micrococcus lysodeikticus, and is incorporated into RNA with various DNA templates (340).It is a substrate for tFWA adenylyltransferase and is incorporated by it into the 3’ terminus of tRNA (341).This modified tRNA can function in the esterification of amino acids. When sangivamycin is injected (i.p,) into mice, its 5’mono-, di-, and triphosphates may be isolated from the red blood cells (342). Sangivamycin is incorporated into the RNA and DNA of all tissues except the brain, where only the RNA contained it. The halflife of sangivamycin in the red cell is 50 hours; this compares to a half-life of 43 hours for tubercidin (302).
P yrazolo p yrimidine Nucleoside Antibiotics The pyrazolopyrimidine nucleoside antibiotics include formycin or formycin A, 7-amino-S(~-~-ribofuranosyl)pyrazolo[4,Sd]pyrimidine or &aza-Sdeazaadenosine, 47, formycin B or laurosin, 8-aza-4 deazainosine (48), and oxoformycin B, 8-am-4deazaxanthosine (49). These were isolated from the culture filtrates of Nocardia interforma n. sp. (343), S. lavendulae (344), and S. gunmaences. Another important nucleoside isolated from N . interforma is coformycin (61) (Section IV).
FORMYCIN 8-AZA-9-DEAZAADENOSINE 47
FORMYCIN B 8-AZA-9-DEAZAINOSINE 40
OXOFORMYCIN B 8-AZA-9-DEAZAXANTHOSINE 49
12. FORMYCIN Formycin can be phosphorylated enzymically to the 5’-mono-, di-, and triphosphates, and de,aminated to fonnycin B. Formycin inhibits de novo purine synthesis in tumor cells. At 10 ,ug/ml, it inhibits DNA synthesis. At 0.1 &ml, it inhibits protein synthesis by lo%, whereas RNA synthesis is not affected; cell division is inhibited 40%. The
232
ROBERT J. SUHADOLNIK
5'-triphosphate (FTP) is incorporated into RNAin vitru and codes like ATP with bacterial and viral RNA polymerases (345,346).However, formycin polymers are very slowly hydrolyzed by spleen phosphodiesterase because the formycin residues exist in the syn-anti, not the anti, conformation. FTP is a substrate for aminoacyl-tRNA synthetase and tRNA-CCA pyrophosphorylase. The tRNA pools for certain amino acids of the silk gland of the silkworm undergoes massive changes for the production of fibroin and sericin; formycin did not affect the synthesis of 4.5 S tRNA (347).However, the synthesis of 4 S tRNA was greatly reduced owing to the existence of some degradation mechanisms of the 4.5 S precursor RNA that contains formycin. This results in a failure to process normal 4 S tRNA (347).These results suggest that the inhibition of tRNA synthesis by formycin may be attributed to the degradation of formycin-containing 4.5 S precursor RNAs so that they cannot be processed into normal 4 S tRNA. FTP reacts 17 times slower than ATP at the initiation site of RNA polymerase (307, 348). However, in the elongation step, it is incorporated into RNA at the same rate as ATP. Rigid specificity at the initiation site may explain the slow rate of incorporation by FTP. The peculiar behavior of poly (F) in enzymic reactions is attributed to the existence of poly(F) as a right-handed helix (349). The absence of a hydrogen at the position of the aglycon of formycin that corresponds to the C-8 of adenosine, plus the longer glycosidic bond (1.50A for formycin versus 1.47A for adenosine), serve to lower the energy barrier of rotation (349).This permits the syn-anti equilibrium (350).Whereas the formycins have free rotation about the C-C ribosyl-aglycon bond, the purine nucleosides have a hindered rotation due to the interaction of the hydroxyl at C-2' and the C-8 proton (351). This conformational inversion permits the deamination of formycin. The sterically constrained nucleoside, 8,5'-anhydroformycin (numbered as in adenosine), the anti conformation, is quantitatively deaminated by adenosine deaminase, whereas 3,5'-anhydroformycin (the syn conformation) is not deaminated (352, 353). Formycin is rapidly deaminated by erythrocytes (289, 301). Although nucleotide concentrations are low in human erythrocytes (301), their formation can be increased markedly by the addition of the adenosine deaminase inhibitor coformycin (61; Section IV), which decreases inosine nucleotides and increases ATP by a third. This observation immediately led to a study of the phosphorylation of formycin in humans suffering from erythrocytic and lymphocytic adenosine deaminase deficiency (ADA deficiency) (354).ADA deficiency is an inherited autosomal recessive trait associated with a se-
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
233
vere combined immunodeficiency disease characterized by defects in thymus-derived and bone marrow-derived cell-mediated immunity (355). ADA-deficient patients show an increased conversion of adenosine to ATP. Coformycin does not alter the nucleotide profiles in ADA-deficient cells because adenosine deaminase is already defective. One important finding of this study is that normal erythrocytes, incubated 4 hours with formycin, show no FTP accumulation, but the FTP level in the ADA-deficient erythrocytes was three times the ATP concentration in normal erythrocytes. The authors suggest that it should be possible to detect ADA-deficient patients by incubation of their erythrocytes with formycin (354). The activity of purine nucleoside phosphorylase in human erythrocytes is lo3times that of adenosine deaminase (301).Formycin B is not a substrate for the purine nucleoside phosphorylase of normal human erythrocytes (289).The hydrolysis of inosine is competitively inhibited by fonnycin B. Because nucleoside phosphorylase can prevent purine nucleoside analogs from entering the nucleotide pool and thereby limit their effectiveness, formycin and formycin B are excellent inhibitors of this enzyme for studies on the effectiveness of adenosine or inosine analogs that would normally be hydrolyzed. Cramer and co-workers have shown (261)that tRNA in which the adenosine at the 3’ end of tRNA is replaced with formycin accepts the amino acid from the AA-tRNA to the nonaccepting adenosyl hydroxyl (i.e., the 3’-hydroxyl) and occupies an position that is not accessible to the groups on the synthetase to make corrections for misactivationa6 Suhadolnik et al. (255) have reported on the activity of NAD analogs in which the adenine or the adenosine-ribose of NAD is modified. Kinetic and protein fluorescence quenching data show that replacement of the adenosine moiety of NAD with formycin (NFD) did not alter the K , or V,,, with horse liver, alcohol, or yeast glyceraldehyde-3-phosphate dehydrogenases. Lactic acid dehydrogenase showed a 10- and 27-fold decrease in the K , and Vma, of the NFD. Because the dehydrogenases follow compulsory ordered kinetics, the decrease in the ratio of V,,, to K , decreases the “ k l on”, while the decrease in V,,, decreases the dissociation of the enzyme * NADH complex, i.e., the “kg off)’. Apparently, the hydrophobic coenzyme domain that accommodates the adenine portion of NAD of lactic acid dehydrogenase is very specific. Hence, replacement of NAD with NFD does not permit productive complex formation with lactic acid. Adenine and adenosine derivatives substituted at N6 are cytokinins; that is, they promote cell division and growth (356, 357). In
234
ROBERT J. SUHADOLNIK
connection with studies of the interaction of certain cytokinins and related compounds with cAMP phosphodiesterase, Hecht et al. (358) described an excellent synthesis of formycin cyclic 3' :5'-monophosphate. Of the 23 compounds tested for the inhibition of cAMP conversion to 5'-AMP, only formycin 3' :5'-monophosphate failed to inhibit this enzyme (359). A recent utilization of formycin has been in studies of blood platelet aggregation. Human bloDd platelets contain higher concentrations of adenine nucleotides than do erythrocytes or leukocytes (360,361). Adenosine inhibits ADP-induced aggregation. Coformycin markedly prolongs this inhibition, while formycin, even in the presence of coformycin, has little effect on it (360).
13. FORMYCINB Formycin B (48), an analog of inosine, and related to formycin as inosine is related to adenosine, has been used as an inhibitor of tumors and viruses (1) and in studies of nucleotide metabolism in human erythrocytes, nuclear ADP-ribosylation, and fruiting-body transformation. At 16 p M , it causes a 50% inhibition of the growth of L5178Y mouse tumor cells (362).Incubation of formycin B with Ehrlich ascites tumor cells from mice yielded no formycin B nucleotides, whereas formycin was phosphorylated (362).The inability of cellular kinases to phosphorylate formycin B has led investigators to study the cellular processes that are affected by this nucleoside. The ADP-ribose moiety of NAD is involved in mono- and poly(ADP-ribosylattion) in prokaryotes and eukaryotes. The ability of adenosine or NAD to reverse the inhibition of exponentially growing L-5178Y mouse tumor cells by formycin B indicated that the inhibition of cellular poly(ADP-ribosylation) is more sensitive to formycin B than are DNA, RNA, and protein synthesis (362). Formycin B is a competitive inhibitor ofNAD in reactions catalyzed by purified chromatin-bound and soluble poly(ADP-ribose) polymerase isolated from quail oviduct (362).Formycin B and showdomycin inhibit poly(ADP-ribose) synthase, but cordycepin, tubercidin, and ara-A do not (259).NFD is equally effective in replacing NAD as an inhibitor of DNA synthesis (Fig. 9). Formycin B has been used as a biochemical probe to study the mechanism of proinsulin biosynthesis (363). Inosine and guanosine are hydrolyzed by purine nucleoside phosphorylase to yield D-ribose, which enhances the biosynthesis. Formycin B inhibits the hydrolysis, this inhibiting the biosynthesis and suggesting that intracellular nucleoside metabolic
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
235
products contribute to the primary signal that controls proinsulin biosynthesis (363). Foimycin B inhibits normal cellular disposition of guanosine in Dictyostelium discoideum by inhibiting purine nucleoside phosphorylase, thereby blocking construction of a fruiting body (364,365). Foimycin B is oxidized to oxoformycin B by rabbit liver aldehyde oxidase (366, 367). Because the K , for foiiiiycin B is 0.2 mM (in close agreement with its Ki), Parks et (11. (366) reasoned that formycin B is oxidized at the same catalytic site of aldehyde oxidase as is 1methylnicotinamide. The agIycon of formycin B (7-hydroxypyrazolo[ 4,3-d ]pyrimidine or 8-aza-4deazahypoxanthine) is also a substrate ( K , = 4mM). An explanation of these findings has been proposed by Johns (367), who postulates the necessity of an unsubstituted N-9 atom of a purine for binding to the enzyme prior to the oxidation step. The rapid oxidation of C-8 of azathiopurine but not that of 4butylazathiopurine illustrates this point (367). Formycin and formycin B, but not inosine, are competitive inhibitors, but not substrates of xanthine for milk xanthine oxidase (366). The homopolymer of poly(forn1ycin B), synthesized from polynucleotide phosphorylase and FDP, but not poly(F), forms a doublestranded complex in a 1: 1 ratio with the octanucleotide of 8,2'anhydro-8-mercapto-9-~-~-arabinofuranosyladenine (368,369). The synthesis of the 4amino derivative of pyrazofurin, 4 a m i n o - 3 (~-~-ribofuranosyl)pyrazole-5-carboxamide, may be effected from formycin B or the 1-oxide of formycin (193). This 4amino derivative of pyrazofurin can be viewed as a C-analog of Saminocarbamideimidazole ribonucleoside and may be a key intermediate in the biosynthesis or degradation of formycin (193). 14. OXOFORMYCIN B Mice can oxidize formycin B to oxoformycin B (366), which does not inhibit the growth ofX. oryzue, Yoshida sarcoma, or influenza virus ( 1 ) . Formycin B is also converted to oxoformycin B by rabbit liver aldehyde oxidase (366), whereas the 3methyl derivative of oxoformycin B is not oxidized (370).The glycosyl torsion angle of crystalline oxoformycin B is (x = 164.1') and places this nucleoside in the syn form (371).An intramolecular hydrogen bond between N-3 and 0-5' in oxoformycin B (adenosine numbering) stabilizes the syn conformation (371).Whereas xanthosine forms a 1: 2 complex with poly(A), surprisingly, oxoformycin B complexes in only a 1: 1ratio. This difference in complex formation is possibly due to steric hindrance by the 2-0x0 group (adenosine numbering) in oxoformycin B (372, 373).
236
ROBERT J. SUHADOLNIK
15. THURINGIENSIN Thuringiensins (50) is a thennostable nucleotide produced by certain strains of the insect pathogen Bacillus thuringiensis; it has been isolated, purified, and crystallized (374376,378).It contains adenine, ribose, glucose, allaric acid, and phosphoric acid (375). This unusual nucleotide has a disaccharide in which the ribose and glucose are linked through an ether bond; it can be characterized as 20-[3-0(5'-adenosyl)-a-D-g~ucopyranosyl]-~-al~aric 4-phosphate (377). $OOH
"OH2@H2N OH
HO OH
THURlNGlENSlN
50
PUROMYCIN AMINONUCLEOSIDE" 3'-AMINO-3'-DEOXY -N6, N6DIMETHYLADENOSINE 51
%
R= COH12
Thuringiensin is toxic to insects, animals, plants, and pathogenic nematodes (375, 379). Like tubercidin, it is also an effective chemosterilaqt when ingested by the house fly. The discovery (380),proof of structure (375,377,382,383), chemical syntheses (378), inhibition of growth (379,383,384),toxicity (379,384,385),and metabolism (379) of thuringiensin have been described (381). As early as 1966,Benz (376) proposed that thuringiensin might act as an antimetabolite of nucleic acids. The first reports (379) showed that the administration (i.p.) of thuringiensin to mice inhibits RNA synthesis by 60%. Polynucleotide phosphorylase and protein and The names pexotoxin, thuringiensin, thuringiensin B, and thuringiensin A have been proposed for this nucleotide. Since the nucleotide is not an exot$xin, but actually an antimetabolite, the trivial name thuringiensin, suggested by K. Sebesta (personal communication), has been proposed to IUPAC.
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
237
DNA synthesis are not inhibited (386,387).Although thuringiensin is a nucleotide, it can cross the mammalian cell wall intact (387). However, when given orally, thuringiensin is rapidly dephosphorylated to the inactive nucleoside. Thuringiensin is a competitive inhibitor of ATP with E . coli DNA-dependent RNA polymerase; the adenine moiety of thuringiensin participates in base-pairing with the complementary base of the template. The allaric acid moiety is essential for the inhibition of RNA synthesis (388). Deamination of thuringiensin produces the inosine analog, which is a competitive inhibitor of GTP (389);the l-oxide is not inhibitory. From these studies, Sebesta and Horska concluded that there is a GTP and ATP binding site on the RNA polymerase complex. Thuringiensin inhibits all RNA polymerases in isolated nuclei of rat liver. Administration (i.p.) of thuringiensin to mice preferentially inhibits the synthesis of rRNA and cytoplasmic RNA, but not of 45 S precursor RNA and tRNA (390). Nucleoplasmic RNA polymerase (RNA polymerase 11) is less sensitive to inhibition by thuringiensin than is nucleolar RNA polymerase (RNA polymerase I). In vivo treatment of mice with thuringiensin results in a marked reduction in the activity of subsequently separated nucleoplasmic RNA polymerase I1 (peak B), but not of that of the nucleolar RNA polymerase I (rRNA synthesis) (peak A) (Fig. 11;390,391). Therefore, thuringiensin complements a-amanitin in that it inhibits RNA polymerase I1 but not RNA polymerase I. The sensitivity of RNA polymerase isolated from the thuringiensin-producing organism is less in exponentially growing cells as compared to cells in the stationary phase of growth (392). Similarly, the RNA polymerase from adult S. bullata is inhibited by thuringiensin (393).Finally, the stimulation of mRNA and rRNA synthesis by the molting hormone, ecdysone, is inhibited by thuringiensin (393). Thuringiensin also inhibits mitotic spindle formation (394), is a competitive inhibitor of ATP with adenylate cyclase (395), and has been used to show differences in the sensitivity of the T3 and E . coli RNA polymerase (396). Although thuringiensin cannot cross the bacterial cell wall, it is toxic to E . coli. The inhibition of RNA synthesis by thuringiensin in E. coli rendered permeable with EDTA-Tris suggests that the effect of thuringiensin on RNA synthesis is external (397). The inhibition of bacteria by thuringiensin may be one of membrane transport.
238
ROBERT J. SUHADOLNIK
PreA
10
20
I \
A
3 0 4 0
50
60
Tube no. FIG. 11. Elution profile on DEAE-Sephadex of RNA polymerase isolated from (a) thuringiensin-treated mouse liver and (b) control. Peak A = nucleolar RNA polymerase I (rRNA synthesis). Peak B = nucleoplasmic RNA polymerase I1 (hnRNA synthesis). From Smuckler and Hadjilov (391).
16. PUROMYCIN AMINONUCLEOSIDE The aminonucleoside N6-dimethyl-$( 3’-amino-3’-deoxy-p-~-ribofuranosy1)adenine (51) is obtained from puromycin by hydrolytic removal of the p-methoxyphenylalanyl group. Although puromycin is a broad-spectrum antibiotic, its aminonucleoside form has no antibacterial activity; however, it inhibits cultured hamster embryo ceIls (398, 399). Although 51 is not a substrate for adenosine kinase, the monodemethylated compound is. This may explain its nephrotic effect in animals (400403).The formation of the nucleotide may also explain the inhibition of RNA synthesis in mammalian cells (398, 399). The
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
239
"reversed" puromycin (11) (39) is not a substrate for adenosine deaminase. The 5'-deoxy or demethylated analog of 51 shows no nephrotoxicity (75).
111. Inhibitors of DNA Synthesis, DNA Viruses, and RNA Viruses In this section are reviewed six naturally occurring nucleoside antibiotics that inhibit DNA replication, have antiviral activity, and are antineoplastic: 90-D-arabinofuranosyladenine (ara-A, vidarabine); l-p-D-arabinofuranosylthymine (ara-T); S76-dihydro-5azathymidine; oxazinomycin (minimycin); l-methylpseudouridine; and nebularine. The most exciting recent application of the naturally occurring nucleoside antibiotics has been the use of ara-A to reduce the number of deaths caused by herpes simplex encephalitis and to reduce the permanent central nervous system damage b y this virus (404).Undoubtedly, ara-A will also find application as an antineoplastic agent in humans.
1. ~-~-D-ARABINOFUMNOSYLADENINE (AM-A, VIDARABINE) Three naturally occurring purine and pyrimidine ribonucleosides have been isolated from the Caribbean sponge Tethya crypta (405407): ara-T (spongothymidine, arabinomethyluridine) (53), ara-U (spongouridine, arabinouridine) (54); and spongosine (2-methoxy-
HO
~-~-D-ARABINOFURANOSYLADENINE ( ARA-A
52
1
I-~D-ARABINOFURANOSYLTHYMINE ( ARA-T,
SPONGOTHYMIDINE 1
53 adenosine) (55). The last two have no known antibiotic activity. Ara-A (52) has been isolated from the culture filtrates of S . antibioticus; it has been synthesized (408,409). In addition to the wide use of ara-A as a biochemical probe and its use against human viruses, it is potentially useful as an antineoplastic agent and is currently employed in patients with chronic myelogenous leukemia in acute blast crisis (410).
240
ROBERT J. SUHADOLNIK
Ho"zw HO
I-P-D-ARABINOFURANOSYLURACIL ( ARA-U,
SPONGOURIDINE 1 54
9-p-D-RIBOFURANOSYL-2METHOXYADENINE C SPONGOSINE 1 55
Ara-A is a cytostatic analog of 2'-deoxyadenosine. L-Cells convert ara-A to ara-ATP, which inhibits DNA synthesis as well as cell viability (208, 4 1 1 4 1 4 ) . The injection of ara-A into tumor-bearing mice inhibits tumor cell DNA synthesis and increases the survival time of the host ( 4 1 5 4 1 7 ) .Although the exact mechanism of action as an antiviral or chemotherapeutic agent is not clear, it is known that ara-A may be converted to its corresponding 5'-mono, di-, and triphosphates, or deaminated and excreted as arabinoinosine (ara-H) (411,415419). Neither ara-A nor ara-H is a substrate for purified calf spleen nucleoside hydrolase (420). The reactions influenced by ara-A have been summarized (421), The inhibitory effects of ara-A seem to be due to the inhibition of DNA polymerase by ara-ATP (413,422,423).Enzymic studies suggest that the inhibition of DNA synthesis is due to (i) an inhibition of DNA replication and repair synthesis by ara-ATP, and (ii) a weak inhibition of ribonucleotide reductase by ara-ADP or ara-ATP (413, 4 2 2 4 2 5 ) . Ara-ATP is a competitive inhibitor of 2'-dATP and inhibits the DNA polymerase-a! and DNA polymerase-/3 of uninfected rabbit kidney cells, and the DNA-dependent DNA polymerase induced by HS virus type 1; it is incorporated into either the 3' end or into the internucleotide linkage of DNA (411,418,421, 4 2 6 4 2 9 ) . The incorporation of ara-A into DNA may explain the damaging effect of ara-A on chromosomes (430).Although ara-ATP appears to be incorporated also into RNA (426), all the tritium in RNA from [3H]ara-A is found in adenosine (418). Because of its rapid deamination by adenosine deaminase to ara-H, the effectiveness of ara-A is limited However, the therapeutic effectiveness of ara-A can be increased in two ways (208,412):it can be administered as ara-AMP, which enters the cell and is slowly deaminated (432), or administered simultaneously with erythro-9-(2'hydroxy-%nonyl)adenine, an inhibitor of adenosine deaminase (431 )
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
241
(Section IV). The advantage of ara-AMP is that ara-AMP is “(i) inert to adenosine deaminase, (ii)the physiological intermediate on the way to ara-ATP, (iii) more soluble than the nucleoside, and (iv) potentially capable of being active on cells incapable of phosphorylating the nucleoside, i.e., one type of analog-resistant cells.” (208). By using the nucleotides of ara-A, Cohen has overthrown the dictum that nucleotides “do not penetrate” (432, 433), which had been a powerful deterrent to research in the area of nucleotide biochemistry. The antiviral activities of ara-AMP and ara-HMP have been summarized (434,435). Ara-A is incorporated into the 3’ end of viral DNA of low molecular weight and is not assembled into high-molecular-weight aggregates (421, 429, 436). Ara-A does not inhibit eukaryotic DNA-dependent RNA-polymerases I and 11, nuclear poly(A)-polymerase, or poly(ADP-ribose)-polymerase. The viral RNA-directed DNApolymerase and DNA-directed DNA-polymerase were less affected by ara-ATP than were the cellular DNA polymerases (418).Ara-A preferentially inhibits the replication of the RNA turnor virus, murine leukemia virus, which has reverse transcriptase activity (437). However, ara-A does not inhibit the replication of most nononcogenic RNA viruses that do not contain the RNA-directed DNA-polymerase. Ara-A is rapidly deaminated by adenosine deaminase (438440), and fetal calf liver has a very active adenosine deaminase (441).Therefore, inhibition studies are distorted and inhibition is thwarted unless deamination of the adenosine analog is prevented. W-Methyladenosine is one analog of adenosine among several studied (442)that inhibits adenosine deaminase, and it increases the antiviral activity of ara-A (443). Without Ns-methyladenosine, its half-life in infected monkey kidney cells was 2-3 hours; the addition of Ns-methyladenosine increased the half-life to 22 hours. The increased inhibition of ara-A by Ns-methyladenosine may be attributed to Wmethyladenosine 5’-triphosphate, which also inhibits DNA synthesis (444). The malarial parasite, Plasmodium berghei, can phosphorylate ara-A to ara-ATP (456,457). While some protein synthesis was markedly inhibited, ara-A stimulated synthesis of some other proteins not normally expressed by the malarial genome. The toxicology, pharmacology, and antiviral activity of ara-A and ara-AMP have been reviewed (458, 510).
2. ~-~-D-ARABINOFURANOSYLTHYMINE (ARA-T,SPONGOTHYMLDINE) Ara-T (53) has little effect on the multiplication of tumor cells in culture (459, 460). It inhibits herpes virus replication in vitro (461-
242
ROBERT J. SUHADOLNIK
464). Cells infected with HS virus contain a pyrimidine deoxyribonucleoside kinase that has broad substrate specificity (465467),but it is not a substrate for normal mammalian cell thymidine kinase (466). Ara-T has no inhibitory effects on uninfected hamster cells in culture but when baby hamster kidney cells were cultured and infected with herpes simplex virus types 1 and 2, ara-T (2 x lW4 M) completely inhibited viral replication. Ara-T inhibits the replication of varicella zoster in human embryo fibroblasts in uitro. It does not inhibit cytomegalovirus (CMV) DNA synthesis in CMV-infected cells; the same relative proportions of cell DNA and viral DNA are synthesized with or without ara-T (467).Miller et al. (467) suggested that CMV does not induce a thymidine kinase as does HSV or VSV. In HSV-1-infected BHK cells, ara-T reduces viral DNA synthesis 91% and also inhibits cellular DNA synthesis of infected cell cultures to a large extent (468) (Fig. 12); the addition of ara-T to uninfected cells did not inhibit cellular DNA synthesis, Extracts of HS-virusinfected cells catalyze the phosphorylation of ara-T due to a virally induced deoxythymidine kinase (468),whereas noninfected cells do not phosphorylate ara-T. Ara-T has a very short half-life in Syrian hamsters, and appears to be phosphorolyzed to thymine and arabinose 1-phosphate (469).Urinary excretion seems to be the primary limiting factor in the in uivo anti-herpes viral activity of ara-T in these animals.
4’
+ omT
16
IHOST DNA 4
0 FR4CTION
FIG. 12. (A) Analysis of DNA synthesis in HSV-1 infected BHK cells. HSV-1 infected cells (3 x 10s cells, multiplicity of infection = 1) were pulse-labeled with 5 pCi of PHIhypoxanthine (@A) from 4 to 12 hours post-infection. The cell lysates, which included [14C]thymine-labeledDNA (0-- -0)from uninfected cells, were subjected to CsCl centrifugation. (B) Same as (A) except that ara-T (2 x lo-* M) was added at 2 hours Dost-infection. From Aswell et al. (468).
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
243
3. 5,6-DIHYDRO-5-AZATHYMIDINE (ANTIBIOTIC U-44590) 5,6-Dihydro-5-azathymidine (antibiotic U-44590)(56) was isolated from culture filtrates of S . platensis var. clarensis (470). Th'is nucleoside contains the same s-triazene ring as does 5-azacytidine (Section II,4). The isolation, production, physical and chemical properties, proof of structure, and inhibitory properties of 56 have been reported
(47 0 4 7 5 ) .
2Cd'
8
HNXN-CH~ 0 HOHZC
H0 H
HO H
5,6-DIHY DRO-5-AZATHYMIDINE 56
HO OH
OXAZINOMYCIN (MINIMYCIN 1 57
The dihydroazathymidine inhibits the replication of herpes simplex virus type 1 in primary rabbit kidney cells in culture (at 6-100 pg/ml). Its addition 4 hours after infection resulted in a 90% reduction in virus yields (476). Type 1 virus was more sensitive than type 2. Mice infected with HSV-1 were protected by 56 treatment (100-400 mg/kg) for 4-5 days. Antiviral activity was observed when drug therapy was initiated 48-72 hours after inoculation. Mice inoculated intracerebrally with HSV-1 were also protected. No .toxicity was detected in mice (476). The antibacterial activity of 56 in E . coli could be reversed by thymidine. Although E . coli is very sensitive to the antibiotic, the mode of action does not involve the direct inhibition of DNA synthesis. It appears to act by inhibiting the phosphorylation of endogenous thymidine (477).It suppresses bone marrow cells in dogs and cats, but not in rats (471). The resistance of an E . coli strain resistant to 56 could not be attributed to thymidine phosphorylase, which would hydrolyze 56 (478). 4. OXAZINOMYCIN (MINIMYCIN) Oxazinomycin (minimycin), a C-nucleoside antibiotic having the (57), was isolated structure 5-/?-ribofuranosyl-1,3-oxazine-2,4-dione
244
ROBERT J. SUHADOLNIK
from culture filtrates of S. hygroscopicus (479, 480). It is structurally similar to pseudouridine (59) (481). It inhibits both gram-positive and gram-negative bacteria, and has antitumor activity against transplantable tumors. Carbons 4,5,and 6 of the oxazine ring of oxazinomycin are derived from carbons 3, 4, and 5 of glutamic acid (482), which is also the precursor for the biosynthesis of the maleimide ring of showdomycin and formycin (483,484).The discovery, production, isolation, proof of structure, inhibitory properties, and chemical synthesis of oxazinomycin have been reported (479481,485487). Oxazinomycin shows significant activity against transplantable tumors. The survival of mice implanted with Ehrlich ascites carcinoma in ascitic form following five consecutive days of administration of oxazinomycin is 70-90%. The LD,, for mice is 10-20 mg/kg (i-p.)and 100-200 mglkg (i-v.) (479, 480). 5: 1-METHYLPSEUDOUFUDINE Argoudelis and Mizsak isolated l-methylpseudouridine (mW, 1MeYrd) (58) and pseudour,idine (Y, Yrd) (59) from S. platensis var.
58 I-METHYLPSEUDOURIDINE R=CH3
59 PSEUDOURIDINE
HO
OH
NEBULARINE 60
R=H
clarensis (488), the same organism from which was first isolated 5,6dihydro-Sazathymidine (see Section 111, 3) (470,471). Pseudouridine was first discovered as a natural constituent of RNA (489). Its biosynthesis by the Streptomyces that produces 5-azacytidine and oxazinomycin has since been shown (see Sections I I , 4 and 111, 4)(1, 490). The discovery, production, isolation, physical and chemical properties, proof of structure, and chemical synthesis of 1MeYrd have been reported (488,4914 9 3 ) . l-Methylpseudouridine and its triacetate did not inhibit grampositive or gram-negative organisms (488). Replication of herpes virus (42D type 1)is marginally inhibited by it, whereas L-1210 cells are not inhibited (488).
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
245
The isolation of pseudouridine and its l-methyl derivative from the culture filtrates of many Streptomyces poses many interesting questions concerning the biosynthesis of the C-nucleoside, W-d, as well as of all the C-nucleoside antibiotics. The biosynthesis of lIrMP in tRNA appears to occur via an intramolecular rearrangement of UMP residues (494496). Ofparticular interest is the observation that the q r d in tRNA is formed via an intramolecular rearrangement of uridine, whereas the biosynthesis of that isolated as the “free” nucleoside in the culture medium does not. Therefore, two mechanisms for pseudouridine biosynthesis must exist.
6. NEBULARINE Nebularine (SP-D-ribofuranosylpurine) (60) is a naturally occurring purine ribonucleoside isolated from Agaricus (Clitocybe) nebularis Batsch. and from S . yokusukanensis n sp. It has been studied in a number of bacterial and mammalian systems (1).It is very toxic to mice and Mycobacterium, to animal cells in culture, and to tumors. Nebularine inhibits influenza B virus infection (497). However, nebularine triphosphate, as an analog of ATP in in vitro studies with Mengovirus RNA polymerase, could not replace ATP (498). Apparently, nebularine must be phosphorylated before it becomes cytotoxic. Sublines of tumor ceIIs in culture devoid of adenosine kinase are not inhibited by nebularine (499).
IV. Inhibitors of Adenosine Deaminase and Effectors of the Immune Response
The therapeutic efficiency of adenine nucleoside antibiotics is markedly increased by the naturally occurring potent adenosine deaminase inhibitors, coformycin (61) and its 2‘-deoxy isomer, deoxycoformycin (co-vidarabine, pentostatin) (62). These two diazepin antibiotics block the deamination of ara-A, cordycepin, formycin, and other adenosine analogs. Coformycin was first isolated from the culture filtrates of N . interforma and S . kaniharaensis SF-557( 5 0 0 5 0 2 ) ; it is reported to be ( R )- 3-( P-D-erythro-pentofuranosyl)-3,4,7,8-tetrahydroimidazo-[ 4,5dI [ 1,3]diazepin-8-01 (503).The total synthesis of coformycin, starting with nebularine (9-P-D-ribofuranosylpurine) (60, see Section 111, 6) has been reported (504). The 2’-deoxy isomer of coformycin (62) has also been isolated as a naturally occurring nucleoside antibiotic from the culture filtrates of
246
ROBERT J. SUHADOLNIK
S. antibioticus (NRRL 3238) (505-507). Coformycin and deoxycoformycin may be considered as dihydrohomopurine analogs. Coformycin (61) and its deoxy analog (62) are the most active adenosine deaminase inhibitors known, with K,'s of 1 x lo-" M and 2.5 X 10+ M yrespectively (purified human erythrocytic adenosine deaminase) (453). Two chemically synthesized inhibitors of adenosine deaminase are 1,Bdihydro-6hydroxymethylpurine ribonucleoside (63) and erythr0-4(2-hydrbxy-Snonyl)adenine(64) (508,509); the K,'s are 1.3 x 1C6 M (453)and 1.6 x lCgM, respectively (453).
1(41
HOH2C
COFORMYCfN
61
DEOXYCOFORMYCIN (CO-VIDARABINE PENTOSTATIN f 62
H COH
I
CH3
1,6-DI HY DRO-6-HYDROXYMETHYLPURlNE NUCLEOSIDE
63
ERYTHRO-9-(2HYDROXY-3-NONYL 1ADENINE 64
In combination therapy with ara-A, 64 increases the lethality of ara-A on mice bearing the Ehrlich ascites carcinoma (208, 431).So does coformycin (445).SimiIarly, 62 potentiates the inhibitory effects of ara-A and cordycepin in cell cultures and in virus-infected animals by inhibiting the deamination of ara-A (446447b).With 62, cordycepin 5'-triphosphate, but not ara-A Ij'-triphosphate, causes an accumu-
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
247
lation of PRPP in Sarcoma 180 cells (447b).These findings may be valuable in designing drug regimens with 62 and adenosine analogs whose activation is PRPP-dependent (447b).Most striking are the reports (415,439,448)that L-1210 cells in culture and mice bearing the L-1210 leukemia, which are insensitive to ara-A, are inhibited when 62 is added. There are two reports on the type of inhibition of adenosine deaminase exhibited by 61, 62, and 64. The inhibition of intestinal adenosine deaminase with 61,62, and 64 has been reported to be of the noncompetitive type (449). However, these calculations did not take into account the extent to which these inhibitors bind to adenosine deaminase. Using the approach of “tight-binding ligands” with enzymes, Cha (452)has introduced a more reliable technique to study enzyme inhibition. The classical methods of kinetic analysis, based on steady-state assumptions, are inadequate with tight-binding inhibitors. Using this technique, Parks et ul. (450,451) speculate that coformycin and 62 are competitive inhibitors of adenosine deaminase. More recently, using the method of Cha to study the inhibition of adenosine deaminase with the four adenosine deaminase inhibitors, 63, 64, and 62 were classified as reversible, semi-tight binding, and tight-binding inhibitors. It is calculated (301, 450) that coformycin binds l06-lO’ times more tightly to adenosine deaminase than does inosine. The double-reciprocal plot of adenosine deaminase (human erythrocyte) in the presence of 64 or 63 shows a competitive type of inhibition. Caution must be exercised when using tight-binding inhibitors of adenosine deaminase. 2’-Deoxyadenosine, which at low concentrations is not an inhibitor of cells in culture, kills cells that have been exposed to 62, even if 62 is removed from the culture medium (454). Prolonged inhibition of adenosine deaminase by 62 and subsequent host toxicity should make 64, the inhibitor with a more transient activity against adenosine deaminase, the agent of choice with ara-A ( 4 5 4 ~ ) . Whereas there is no inhibition of 5’-AMP deaminase by 63,64, and 2-fluoroadenosine, 62 inhibits 5‘-AMP deaminase by 66% (455). A prior incubation of the deaminase with 62 showed no inhibition, whereas such an incubation with coforniycin inhibited it 99.5%. The inhibitions with coformyciii and 62 were of the noncompetitive type. Coformycin clears fiom the plasma of dogs with a tl,2 of 75-120 minutes. The extent of inhibition of adenosine deaminase varies among tissues: in bone marrow and jejund inucosa, it is 25%and 19%, respectively, which compares with only 9%,4%, and 2%,respectively, in spleen, liver, and kidney (511). A photometric enzyme assay
248
ROBERT J. SUHADOLNIK
to measure coformycin in tissues and body fluids of mice is based on the inhibition of calf intestinal adenosine deaminase by the amounts recovered from plasma and tissues. The biological half-life of 62 in mouse plasma was approximately 20 minutes, being rapidly eliminated from the body in the urine; approximately 87% is recovered in the urine in 2 hours (512).Another method for the determination of 62 in animal fluids and tissues is by means of a purine-requiring strain of S. fueculis; 62 is toxic for this organism only in the presence of adenine or diaminopurine (513). Deoxycoformycin, in combination therapy with ara-A and ara-C for iiitracerebral leukemia L-1210 in mice, increases the toxicity of ara-A and ara-C to the tumor without an increase in host toxicity. The Ki for partially purified mouse brain adenosine deaminase for 62 is 2 x l t l ' M; the K , for ara-A is 1 x M (514). Mice bearing L-1210 were treated with one dose of 62 plus ara-A or one dose of 64 plus ara-A. The former treatment gave high intracellular levels of ara-ATP, which fell rapidly (tl,z= 2 hours ). In the latter case, ara-ATP concentrations remained high for 9-12 hours. The failure of 64 to sustain ara-ATP levels indicated a rapid reversal of the inhibition of adenosine deaminase in vivo. Ara-A alone was ineffective against L-1210 cells, but 62 plus ara-A or 64 plus ara-A were highly effective against mouse transplanted with L- 1210 cells (515). Ara-A alone inhibited the DNA synthetic capacity of mouse leukemia cells by 50% in 1hour, with a return to normal after 3 hours (516). There was no significant inhibition of DNA synthesis in the bone marrow and gastrointestinal mucosa. However, ara-A plus deoxycoformycin increased the inhibition in both tissues from 50% to 90% in 6 hours, and recovery was delayed 18hours. The DNA synthetic capacity of bone marrow and gastrointestinal mucosa was inhibited 30% to 80% in 1hour. Recovery was complete by 3 hours. Another biological function for the adenosine deaminase inhibitors is as immunosuppressants. For example, they increase the success of tumor grafts, which suggests useful application as immunosuppressants in organ transplantation. Adenosine deaminase inhibitors may act as immunosuppressive agents by impairing maturation of mitogen-stimulated blastogenesis of lymphocytes (517, 518). The advantage of 62 or 64 as immunosuppressants is that they are not myelosuppressive; they act via adenosine deaminases that are limited to lymphocytes and monocytes. They mimic the rare, but known and manageable, immunodeficiency syndrome in certain individuals lacking adenosine deaminase (519).
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
249
V. Inhibitors of Cell- Wall Synthesis and Antifungal Agents There are one naturally occurring nucleotide antibiotic and twelve nucleoside antibiotics that act by affecting cell-wall receptors or inhibiting cell-wall synthesis in the eukaryotes and prokaryotes. They are agrocin 84, amipurimycin, crotonoside, the ezomycins, nikkomycin, platenocidin, the polyoxins, septacidin, showdomycin, sinefungin, thraustomycin, tunicamycin, and ileumycin.
1. AGROCIN 84 Agrocin 84 is a unique adenine nucleotide elaborated by the nonpathogenic strain Agrobacterium radiobacter var. radiobacter strain K-84 and by the virulent strain 396 (J. Schell, personal communication). Roberts et al. (520) and M. E. Tate (personal communication) proposed the structure of agrocin 84 shown (65),which was confirmed
R oH H 2 c d H H H
AGROCIN 84 R-UNIDENTIFIED PHOSPHORYLATED CARBOHYDRATE
66
by Thompson et al. (521). The phosphate on the anomeric carbon of the glucofuranosyl is unique. However, recent work by Tate, Kerr, and Roberts supports the structure as shown (65). Agrocin 84 is the first naturally occurring adenosine derivative with a 6-N-phosphoramidate to which glucose, in the furanose form, is covalently bound; also the ribofuranosyl moiety of adenosine is replaced by the rare pentose 3-deoxy-D-arabinofuranose.The discovery, production, isolation, physical and chemical properties, proof of structure, and inhibition of growth have been described (520, 522, 523). Agrocin 84 is one of several bacteriocins elaborated by A. radiobacter var. radiobacter strains K-84 and 396. Excretion of agrocin 84 by the Agrobacterium strain causes an interaction specifically with
250
ROBERT J. SUHADOLNIK
the cell wall of related varieties of A . radiobacter var. tumefaciens. Agrocin 84 does not interact with the surface of the nonpathogenic strains. It inhibits RNA, DNA, and protein synthesis completely, DNA and protein synthesis in 10 minutes, RNA synthesis after 20 minutes. It inhibits completely the uptake of amino acids and stops cell motility immediately (Fig. 13) (524).These observations indicate that agrocin 84 affects energy-generating reactions. Fifty percent of Agrobactedum cells were killed in 15 minutes after contact with agrocin 84.The surviving cells were inhibited. It is proposed that receptor sites for agrocin 84 on the nonpathogenic
n
c
8
r 10203040506070 Time (min)
f
2
3
8 1
z
10 20 30 40 50 60 70 Time (min)
Time (min)
FIG. 13. Effect of agrocin 84 on synthesis in Agrobacterium rudiobacter var. turnefaciens H-389. Agrocin added at zero time. (A) Ribonucleic acid synthesis at zero time. [‘*C]Uracil was added. (B) Deoxyribonucleic acid synthesis at zero time. ISH]Thymidinewas added. (C) Protein synthesis at zero time. W-Labeled mixed amino acids were added. 0, agrocin 84; 0, control. From McCardell and Pootjes (524).
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
251
strains are either missing or altered, in agreement with the postulate that a specific molecular configuration on the bacterial surface is essential for pathogenesis (523,525,526).It is this specific configuration of the cell-wall receptors of the pathogenic strains of the Agrohacterium that probably interacts with agrocin 84. These receptor sites probably interact with the plant and result in the neoplastic disease crown gall. Crown gall is the only natural system in which the genetic information is transmitted from bacterium to plant. The requirement for pathogenicity is that the Agrobacterium strain contain the Ti(tumor-inducing) plasmid, which initiates the neoplastic disease in plants. Therefore, the Ti plasmid codes for pathogenicity. Contact of agrocin 84 with the cell surface of the pathogenic strains of Agrobacterium brings about the killing effect. Once the Ti plasmid is lost, the pathogenic strain becomes nonpathogenic and is not sensitive to agrocin 84. The Ti plasmid is present in seven pathogenic strains of Agrohacterium (527,528).Therefore, those Agrohacterium strains carrying the Ti plasmid confer agrocin 84 sensitivity. Resistance to agrocin 84 can be due to the loss of the large Ti plasmid or to deletion in the Ti plasmid (522,528531).
2. AMIPUFUMYCIN Amipurimycin was isolated from culture filtrates of S. nouoguineesis (No. T-36496) in a soil sample collected in Rae Papua, New Guinea (534,533).The name amipurimycin, was given to the antibiotic because it contains the 2-aminopurine chromophore. Although the complete structure of amipurimycin is not known, it is composed of Baminopurine plus a G,HZ3,,NzO8 fragment (66).
HO OH
( C15H23-27N208) CROTONOSIDE
AM IPURIMY CIN 66
( ISOGUANOSINE1
67
Amipurimycin inhibits the growth of numerous phytopathogenic fungi of rice plants and the growth of the dermatophyte Trichophyton mentagrophytes, but has no antimicrobial activity against grampositive bacteria, gram-negative bacteria, yeasts, and saphrophytic
252
ROBERT J. SUHADOLNIK
fungi (532). In field tests, amipurimycin showed excellent curative effects against blast of rice plants. The type of inhibition by amipurimycin against blast disease resembles that of blasticidin S. No toxicity of amipurimycin to killifish was observed at 10 ppm in 2 days, but all test fish died in 3 days.
3. CROTONOSIDE Crotonoside (isoguanosine, 2-hydroxyadenosine) (67) was isolated from the seeds of the croton bean, Croton tiglium L (534);more recently, the aglycon isoguanine was isolated from the butterfly wings of Prioneris thestylis (535). The first studies on the effect of crotonoside on bacteria and in the rat showed that it is not incorporated into the nucleic acids of Lactobacillus casei (536538). Isoguanosine preferentially inhibits the inducible nucleoside binding sites whereas the nucleoside analog showdomycin interferes with the constitutive function (539). For example, isoguanosine is an effective inhibitor of the uptake of 2’-deoxycytidine in an E . coli mutant resistant to Sazacytidine and showdomycin. Although tubercidin, a structural analog of adenosine, is relatively ineffective in eliciting or preventing the adenosine-induced formation of CAMP, crotonoside causes a significant accumulation of CAMP (540). Adenosine analogs modified in the ribosyl moiety strongly inhibit the formation of CAMP, but do not cause the accumulation noted with crotonoside. Isoguanosine 5’-di- and triphosphates bind strongly to and inhibit glutamic acid dehydrogenase (541 ).
4. EZOMYCINS In 1971, a group of phytopathogenic nucleosides was isolated from a strain of Streptomyces similar to S . kitazuwaensis (542). Various studies (543552)yielded the structure, physical and chemical properties, and inhibitory properties of eight ezomycin nucleoside antibiotics. Ezomycins 4 (68) and Az (69) are N-nucleosides in which the sugar moiety is attached to the N-1 of cytosine. Ezomycins B, (70), & (71), C, (72),C, (73), D1 (74), and D, (75) are C-nucleosides in which the anomeric carbon of the sugar moiety is attached to the C-5 of uracil. Ezomycins B,-D2 are the first naturally occurring pseudouridine-type (i.e,, C-nucleoside) antibiotics to be identified, Ezomycins A*, AP,B1,and B2 occur as the p-anomers, ezomycins C, and C, as the a-anomers. The anomeric linkages of ezomycins Dl and D, have not been established. The octosyl-containing bicyclic system
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
y2
HOOC-C-CH2-S-CH
-CH
o=c
Y
- C-COOH
253
FZ R
Ill4
H
H N-
68
OH
EZOMYCIN A l R'~-I-CYTOSINE
69 EZOMYCIN A2 R-9-I-CYTOSINE
70
EZOMYCIN BI R = p-5-URAClL
72
EZOMYCIN CI R= d-5-URACIL
74
EZOMYCIN DI R = 5-URACIL
71 EZOMYCIN B2 R=P-S-URACIL
73 EZOMYCIN C2 R- (r-S-URACIL
75
EZOMYCIN 02 R = 5-URACIL
in the ezomycins is of the same type as that of octosyl acids A, B, and C (see Section VIII). The ezomycins inhibit the growth of a very limited number of species of fungi, including Sclerotinia and Botrytis (545).Ezomycin A1 and B, show distinct inhibitory activity whereas A2 and B, do not. The mode of inhibition on the growth of these fungi appears static; that is (in the case of B . cinerea), colonies were observed 2 weeks after inoculation with 100puglml of either A1 or B,, but mycelial growth and sporulation were strongly suppressed even at concentrations of 1.6 pglml. The amino group on the %position of the hexuronic moiety of the ezomycins may put them into the class of the pyrimidine nucleoside antibiotics (Section I, A). Although the ezomycins have a uronic acid O-carbamoyl moiety like the polyoxins (Section V, 7), they do not inhibit chitin synthetase.
5. NIKKOMYCIN Nikkomycin (76) was isolated from the fermentation broth of S. tendae TU 901 (553). Its structure was deduced from the products obtained by chemical degradation (553). Nikkomycin is a nucleoside-
254
ROBERT J. SUHADOLNIK
CH
I
OH
NH2
NIKKOMYCIN
76
PLATENOCIDIN 77
POLYOXINS 78 R,= -CHZOH,
-COOH, -CHS,
-H
COOH
%= -W,&NR 3 = -H, 5-O-CARBM(OYL-Z-AMlNO-2-DEOXY-LXYLONIC ACID OR THE 3-DEOXY DERIVATIVE
peptide consisting of uracil, a Saminohexuronic acid, and a new pyridine a-amino acid. Nikkomycin, a potent inhibitor of fungi, is structurally similar to the polyoxins (Section V, 7) and to ablastimycin, bulgerin, and antibiotic 24010 (554557). Nikkomycin and clitidine (92) are the first naturally occurring nucleoside antibiotics containing a
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
255
pyridine ring. The discovery, production, isolation, physical and chemical properties, proof of structure and inhibition of growth have been reported (553). Nikkomycin inhibits RNA synthesis in cultures of Mucor hiemalis (+) by 63%;this compares to 77% inhibition for cycloheximide (553). For protein synthesis, it is 40% as against 80%for cycloheximide. The inhibition of chitin synthesis by nikkomycin is very similar to that reported for the polyoxins (see Section V, 7 )(553),and is shown in Fig. 14. Cycloheximide does not inhibit chitin synthesis. At 0.5 pglml, nikkomycin completely inhibits the incorporation of N-acetylglucosamine, and it can serve as a negative allosteric effector in much the same way as N-acetylglucosamine. Chitin synthesis is more sensitive to inhibition by nikkomycin than is RNA synthesis. The in-
I
1
1
1
1
1
30 45 60 75 90
I
120 Minutes
I
I
1
165
225
285
-
FIG. 14. Effect of nikkomycin on chitin synthesis in Mucor hiemulis (+). N-acetylvs. time. 1, control; 2, nikkomycin (0.5 pg/ml); 3, ~-[~H]g~ucosamin incorporation e cycloheximide (10 pglnil). Diihn et (11. (553).
256
ROBERT J. SUHADOLNIK
hibition of RNA synthesis is not related to the inhibition of chitin synthesis.
6. PLATENOCIDIN Platenocidin (77), a 5-hydroxyinethyluracil antibiotic, has been isolated from the culture filtrates of S. H 273 N-SY and S. platensis (558), the same organism from which 5,6-dihydro-5azathymidine (antibiotic
U-44590),pseudouridine, and 1-methylpseudouridine (see Sections 111, 3 and 111, 5) were isolated (488). Platenocidin inhibits yeast, but not bacteria (558).Candida albicans was the most sensitive of the yeasts to this antibiotic. The isolation of 5-hydroxymethyluracil from a hydrolyzate of platenocidin suggests that polyoxins A, B, C, and I, which also contain 5 hydroxymethyluracil (554, 559561 ), may be part of the platenocidin molecule. However, the inhibitory properties of platenocidin differ markedly from those of polyoxins A and B, which inhibit phytopathogenic fungi, but not yeast; platenocidin inhibits yeast, but not pathogenic fungi. 7. POLYOXINS The polyoxins (78) are peptidyl-pyrimidine nucleoside antibiotics that have strong antifungal activity. The isolation, physical properties, and chemical degradations have been described, the latter leading to the structural elucidation of the polyoxiiis A-M (1,554,562).The aglycon in all the polyoxins contains a pyrimidine chromophore and the hexuronic acid, 5amino-5deoxy-allofuranuronose.Absolute proof of the nucleoside skeleton of the polyoxins has been reported (563),as has the total chemical synthesis of polyoxin J (564).In addition to the polyoxins, S. cacaoi also produces the octosyl acids, A, B, and C (see Section VIII) (684).There is a biogenetic interrelationship of the octosyl acids to the polyoxins (565). The polyoxins are excellent agricultural fungicides. Minor structural differences among them are important in their inhibitory activity. For example, polyoxin D, with 5-carboxyuracil as aglycon, is used to treat the Pellicularia disease of rice plants, whereas polyoxin L, with uracil as aglycon, is used against the Alternaria disease of fruit trees
(560). The first reports on the action of the polyoxins showed that they cause the fungal cell wall to swell (566, 567). Subsequent studies showed the polyoxins to be powerful competitive inhibitors of chitin UDP-N-acetylglucosaminotransferase (chitin synthetase) in cell-free extracts from T. oryzae, S. cereuisiae, and A. j a m s ( 5 6 8 5 7 4 ) .
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
257
Recent studies (575577),using the polyoxins and their analogs, show that the action of polyoxins can be considered as the actions of three parts of the molecule. It is the interaction of these three parts of the polyoxins with the active site of chitin synthetase that seems to explain the mechanism of their action as competitive inhibitors. The cell-wall synthesis of a polyoxin-resistant mutant of A. kikuchianu was not inhibited by polyoxin B (578). However, with cell-free extracts, the chitin synthetases from both the resistant strain and the sensitive strain were equally sensitive to polyoxin B. This dilemma was resolved when '*C-labeled polyoxins A, B, C , and I showed a decreased uptake by the polyoxin-resistant strain (579,580). These studies eliminated the possibility that the resistance should be attributed to an increased concentration of the substrate, UDP-Nacetylglucosamine, which competes with the polyoxins. Nucleated protoplasts of Schixophylluni commune can regenerate their walls, and 50% can revert to the hyphal mode of growth. Of the three main wall components, polyoxin D inhibited the synthesis of two, chitin and R-glucan (/+1,3-P-l,8glucan) (581,582).The carboxyl group on the 5'-C and the a-L-amino group of the polyoxins are absolutely required to inhibit chitin synthetase (583).
8. SEPTACIDIN Septacidin (79), although not a nucleoside, is included in this review because it contains adenine. A 4-aminoaldoheptose is covalently
SE PTACtDIN
79 bound to the amino group of adenine. The products of hydrolysis of septacidin are adenine, glycine, a 4aminoaldoheptose, and C, to C& fatty acids (mostly branched), isopalmitic acid predominating among them. Septadicin shows strong antifungal activity against Trichophyton menthagroph ytes and Fusarum bulbigenum Although composed of
.
*
258
ROBERT J. SUHADOLNIK
adenine and an amino sugar, it does not have an accessible free amino group or a peptide bond with a free amino group as do other inhibitors of protein synthesis, such as puromycin, homocitrullyladenosine, gougerotin, blasticidin S, etc. (1).
9. SHOWDOMYCIN Showdomycin, 2-(P-~-ribofuranosyl)maleimide (80), is a C-nucleoside antibiotic elaborated by S. showdoensis (584). Its struc-
J
HOH2C
HO OH
SHOWDOMYCIN 80
ture has been established (585, 586); its isolation, structural elucidation, and properties, and the biosynthesis of the bicyclic maleimide antibiotic maleimycin, isolated from the culture filtrates of S . showdoensis, have been described (587).The microbial transformation of 80 to isoshowdomycin has been reported (588). Although €40 is structurally similar to pseudouridine and uridine, its biological properties are markedly different. Showdomycin is not a substrate for nucleoside kinase or nucleoside phosphorylase (589), so it probably exerts its stimulatory effects on bacteria and tumors without conversion to the nucleotide (584, 590). Its cytotoxic action seems to depend on two properties. First, as a nucleoside, it can enter the cell easily. Second, once in the cell, the maleimide ring can act as an alkylating agent for sulfhydryl, amino acid, and imidazole groups (589,591596).The inhibitory effects of 80 are blocked by a prior incubation of E . coli with cytosine, cysteine, nucleosides or nucleoside analogs (597, 598). The entry of showdomycin into E . coli is competitively inhibited by nucleosides (599).E . coti mutants resistant to it take up very little and show an altered ability to take up nucleosides (600).These resistant mutants are very sensitive to N-ethylmaleimide (600).The difference between MalNEt and showdomycin is that the latter cannot overcome the inhibitory effects of MalNEt on the uptake of glucose, whereas most nucleosides can (597).This led to the speculation (597) that resistance to showdomycin by bacteria is due to an inhibition of transport of the drug into the cell.
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
259
Because some nucleosides protect E . coli against showdomycin, there must be competition between it and nucleosides for a common transport system. Showdoniycin-resistant E . coli have a marked reduction in uridine, cytidine, and showdoniyciii transport, whereas guanosine and adenosine transport is similar to that in wild-type cells (601, 604, 606). Showdomycin-resistant E . coli mutants also have an impaired ability to take up deoxyribonucleosides (602);transport systems for deoxyadenosine and adenosine are defective in resistant E . coli B (603). Binding sites are important for nucleoside transport in bacteria, and two types of binding sites are known: constitutive sites with high substrate affinity, and inducible sites, which have lower substrate affinity. In E . coli, 80 interferes more strongly with the constitutive function (539).E. coli K12 has two deoxycytidine transport systems (605). 80 is transported by the high-affinity deoxycytidine transport system, which is not affected by guanosine or xanthosine. In resistant mutants, this high affinity deoxycytidine transport system is no longer detectable, while the low-affinity deoxycytidine transport system is not changed. Showdomycin and its 5’-phosphate effectively alkylate the functional sulfhydryl group at the active site of thymidylate synthetase (607-609). This is consistent with a sulfhydryl addition-elimination mechanism of the enzyme-catalyzed reaction (Fig. 15). Under anoxic conditions, the radiosensitizing ability of 80 in E . coli
-
J-Jomi”” -’
HN
6
Rlb(P)
+o J-&T
(b)
Rib(P)
FIG. 15. Analogy between the postulated catalytic role of an active site SH-group and its interaction with showdomycin and showdomycin 5’-phosphate. Xf, electrophilic precursor of the methyl group of thymidylate. From Kalman (608).
260
ROBERT J. SUHADOLNIK
B/r is derived mainly from the maleimide moiety (610); in aerated conditions, the hydroxy groups of the ribose are essential. One hypothesis postulated for carcinogenesis is that a carcinogen damages the mitochondrial membrane and interferes with the flux of energy in the mitochondria, releasing mitochondrial genetic material, which behaves like an oncogenic virus and enters the genome of the cell (611,612). The well-known carcinogenic compound, 2-(N-acetylN-hydroxyl)aminofluorene, AcN(H0)Fln (Sl), when combined with
2-CN-ACETYL-N-HYDROXY AMINOFLUORENE ( Ac N(HO) Fln ) 81
I-
showdomycin, induces an ATP-energized mitochondrial volume change (611). Either 80 or 81 alone does not. The mitochondrial volume change by AcN(H0)Fln exposes the mitochondrial sulfhydryl groups to showdomycin, permitting alkylation, which allows the genetic material to leak out. This material may then eventually enter the genome and a cancerous cell may be generated. 2-(N-acetoxy-N-acetyl)aminofluorene,AcN(AcO)Fln, in combination with 80, also induces ATP-energized mitochondrial changes whose magnitudes parallel its carcinogenicity in rats (613, 614).The acetoxy derivative, favored by many investigators as the ultimate electrophilic carcinogenic agent, is enzymically converted to the hydroxy compound by mitochondria. Therefore, carcinogenesis by either substance is the result of a disruption of the symbiotic relationship between the mitochondria and cell established during the course of evolution. In effect, this hypothesis suggests that carcinogenesis represents a partial reversal of evolution. 10. SINEFUNGIN(A9145) A new strain of S . griseolus (NRRL 3739), isolated from a soil sample collected on the Ivory Coast region of Africa, produces the antifungal nucleosides sinefungin (A9145)(82), A9145A ( 8 3 , and A9145C (84) (61s-618). These nucleosides are unique in that they consist of an a-amino acid (2,5-diaminovaleric acid) linked through its carbon4 atom to the 5'-carbon of adenosine, and are thus structurally similar to S -adenosylmethionine (85). They inhibit Newcastle disease virion,
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
SINEFUNGlN
261
A9145A 83
( A9145 3
82
S-ADENOSYLME THlONlNE ( AdoMet 1 85
A9145C 84
vaccinia virion mRNA(guanine-7-)methyltransferase,vaccinia virion mRNA(nucleoside-2'-)methyltransferase, and other transferases
(618 a ) . 11. THRAUSTOMYCIN The structure of a new potent antifungal nucleoside antibiotic from S. exfoliatus, thraustomycin (619),has not been established; it appears to be composed of equimolar amounts of adenine, L-leucine, and a tetrahydroxymonocarboxylic acid with properties of a carbohydrate (86).The discovery, production, isolation, physical and chemical properties, evidence relating to structure, and inhibition of growth have been described (619).
ADENINE
CARBOHYDRATE
THRAUSTOMYCIN 86
LEUCINE
262
ROBERT J. SUHADOLNIK
TUNlCA M YClN 07 TUNICAMYCIN TUNICAMYCIN TUNICAMYCIN TUNICAMYCIN
A n.9 B n=lO C n-8 D n-11
12. TUNICAMYCIN Tunicamycin (87) was first isolated from either the mycelium or culture filtrates of S. Zysosuperi3cus nov. sp. (621).OIt is not a single compound, but a mixture of homologous antibiotics (622,623),at least four in number, called tunicamycins A, B, C, and D. Chemical degradations and physical data yield structure 87 for the main component of the complex. Each contains uracil, a fatty acid component that is a truns-(Y,P-unsaturatedbranched-chain acid, N-acetylglucosamine, and a new GI aminodeoxy-dialdose sugar, tunicamine (622). One of the acids is isopentadecanoic acid (623). Tunicamycin inhibits gram-positive bacteria (especially bacilli), yeast, fungi, protozoa, plants, viruses, and mammalian cells in culture (621). Because of its ability to inhibit the synthesis of Nacetylglucosaminyl lipids in prokaryotes, eukaryotes, and viruses, 8'7 is a most useful biochemical probe for elucidating the complex reactions involved in the assembly of glycoproteins and cell walls (for reviews, see 624-626). The evidence to date indicates that tunicamycin acts by inhibiting the incorporation of N-acetylglucosamine into glycoproteins (Fig. 16). The biosynthesis of glycoproteins proceeds via the transfer of an oligosaccharide from an oligosaccharide-lipid to endogenous membrane proteins (627-629).The synthesis of the polypeptide chain of ovalbumin has been reported in great detail (630).Nascent chains of ovalbuOIt also appears that the streptovirudin complex (A2, B2,C2,DJ (620) and antibiotic 24010 (557) are structurally similar to tunicamycin.
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
263
GlcNAC-GicNAC-P- P-Dol
7-
Do1 -P
UDP-GlcNAc
fGDPMon nGDP Yon
-
7
Man- GlCNAC-GlCNAC-P-P- DO1
nMan-P-Do1
nDol -P
(Mon),Mon- GlcNAC -GleNAc -P-P-Do1
J
flPolypeplide
(Yon)nMan-GlcNAc-GlcNAc- Polypeptidr
FIG.16. Pathway proposed for the assembly of asparagine-linked oligosaccharides of glycoproteins utilizing polyisoprenoid phosphate carriers. The known site of inhibition by tunicamycin is shown by the crosshatched bar. From Struck and Lennarz (627).
min are still attached to tRNA when the carbohydrate chain of mannose and N-acetylglucosamine units are added (631). Tunicamycin preferentially inhibits the incorporation of glucosamine in glycoproteins of chick embryo fibroblasts (632). 87, with rough microsomes of calf liver, inhibits the production of N-acetylglucosamine-lipid, but does not affect synthesis of mannolipid (633). With yeast, 87 specifically blocks the synthesis of N z
W
P
A I-ITUNICAMYCIN
B (4-1 TUNICAMYCIN
Monnoryl -Lipid
Ohgo-Lipid
0
5" Y
'0
10
20
30 0 10 TIME (MINUTES)
Monnoryl-L ipid
OIIgo-LIpld
20
30
FIG. 17. Effect of tunicamycin on inannose incorporation from GDP-mannose by oviduct memliranes. From Struck and Lennarz (627).
264
ROBERT J. SUHADOLNIK
0 P
3000
1000
B I+)TUNICAMYCIN
2000
z
2000
TIME (MINUTES)
P I
FIG.18. Effect of tunicamycin on N-acetylglucosaniine incorporation from UDP-Nacetylglucosamine by oviduct membranes. From Struck and Lennarz (627).
acetylglucosaminyl-lipid (634). Saccharide-lipids participate in the assembly of the oligosaccharide side chain of ovalbumin (627).Experiments in this area with 87 lead to five conclusions: (i) in vitro, 87 inhibits the synthesis of N-acetylglucosaminyl-lipid catalyzed by hen oviduct membrane preparations whereas the monosaccharide-lipid, mannosylphosphodolichol, and protein synthesis are not inhibited (Figs. 17 and 18); (ii), 87 does not inhibit elongation of preexisting N-acetylglucosamine-lipids to oligosaccharide-lipid, or the transfer of oligosaccharide from oligosaccharide-lipid to protein; (iii), 87 appears to block the synthesis of N-acetylglucosamixiylpyrophosphopolyisoprenol; (iv), 87 does not block other N-acetylglucosaminyl transferases present in oviduct membrane preparations; and (v), 87 can synthesize mannosylphosphodolichol, but cannot utilize this mannosyl donor for glycoprotein synthesis, Tunicamycin does not act as an analog of UDP-N-acetylglucosamine in the N-acetylglucosaminyl transfer reaction, nor does it inhibit chitin synthesis (627). Tunicamycin is associated with yeast membranes and can be removed by extraction with organic solvents (636),contrary to an earlier speculation (635). Tunicamycin inhibits cell-surface glycoproteins (637)and is also a potent antiviral antibiotic (621,638, 639). The carbohydrate moiety of the viral glycoproteins is necessary for virion formation (639). 87
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
265
causes detachment and death of cells transformed by Simian virus 40 and polyoma (640). Apparently tunicamycin interferes with the function of the cell-surface glycoproteins. It also inhibits the replication of the viral envelope. Encephalomyocarditis virus, which has no envelope, is not so inhibited (642). Of the eukaryotic microorganisms, the Saccharomyces have been the most extensively studied with tunicamycin (625,626). It blocks the synthesis of the external glycoprotein invertase, acid phosphatase, and mannin by yeast protoplasts and inhibits the synthesis of dolichyl N-acetylglucosamine pyrophosphate (642) (Fig. 16). Tunicamycin induces morphological changes such as the elongation, conversion from rod form to cocci form, and enlargement of microbial cells (621,638). Although RNA and DNA degradation occurred in B . subtilis, the incorporation of glucosamine was most extensively inhibited. The mode of inhibition must be directed on the cell surfaces. The addition of N-acetylglucosamine to undecaprenylphosphate from UDP-N-acetylglucosamine in isolated membranes of B . subtilis occurs in two different steps (645,646). Tunicamycin interferes with peptidoglycan synthesis in M . luteus and B . lichen$omnis by blocking the formation of the first lipid intermediate, undecaprenylmuramyl pentapeptide pyrosphophate; the enzyme inhibited is phospho-N-acetylmuramylpentapeptidetransferase (643, 644). B . Zicheniformis membrane can also incorporate N-acetylglucosamine from UDP-N-acetylglucosamine into a lipid and nonlipid fraction (624). Therefore, in the absence of UDP-muramylN-acetylpentapeptide, membranes are capable of transferring one N-acetylglucosamine from UDP-N-acetylglucosamine to undecaprenyl phosphate via transphosphorylation. There is a differential effect of UMP and UDP on the depolymerization of the lipid. UDP acts by breaking an O-glycoside linkage and UMP breaks the phosphodiester bond to the lipid. Tunicamycin inhibits the UMP and UDP lipid formationidepolymerization reactions by 74% and 78%, respectively, and there is a 10% inhibition in the formation of Css-pyrophosphate-muramyl-N-acetyl pentapeptide by 87 (694).These results are consistent with previous findings (643, 644) in which different systems were used. Tunicamycin inhibits N-acetylglucosamine-lipid formation in plants and mucopolysaccharide synthesis in chick embryo cells (632, 637, 647, 648). There is evidence for a linkage region between peptidoglycan and polyribitol phosphate consisting of N-acetylglucosamine and glycerol
266
ROBERT J. SUHADOLNIK
phosphate. Tunicamycin inhibits the synthesis of this linkage unit [accept~r-(P-glycerol)~]. Presumably it inhibits the synthesis of the linkage region between teichoic acid and peptidoglycan (641).
13. ILEUMYCIN Ileumycin was recently isolated from the culture broth of Streptornyces ZavenduZae ( 6 4 8 ~ )The . physical, chemical, and biological properties of ileumycin are similar to those of other nucleoside antibiotics. Although the structure has not been elucidated, ileumycin can be differentiated from the ezomycins, ablastomycin, bulgerin, and antibiotic SF1508 ( 6 4 8 ~ )Isoleucine . has been isolated following acid hydrolysis, whence the name. Ileumycin exhibits antimicrobial activity against plant pathogenic fungi, but not against yeast and bacteria (648~).
VI. Inhibitors of Purine and Pyrimidine lnterconversions 1. PSICOFURANINE AND DECOYWINE Psicofuranine and decoyinine are isolated from the culture filtrates of S . hygroscopicus var. decoyicus (649, 650). Psicofuranine (angustmycin C) is 9-P-D-psicofuranosyladenine(88) and decoyinine (angustmycin A) is 4/3-D-psicofuranos-5(6)enyladenine(89). Both are naturally occurring adenine-ketose antibiotics.
HoHzcw CHzOH
HO OH
HO OH
PSICOFURANINE
DECOYINlNE
88
89
Both 88 and 89 are noncompetitive inhibitors of the conversion of XMP to GMP. Guanine-containing compound's reverse this inhibition. Psicofuranine is not a substrate for adenosine deaminase; however, it can be phosphorylated to form the 6'-phosphate (1).The inhibition of XMP aminase by 88 and 89 is unique in that it occurs at the nucleoside level. Neither inhibits RNA synthesis in Euglena (651). Sporulation of bacilli occurs when rapidly metabolizable carbon or nitrogen compounds are replaced by more slowly metabolizable com-
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
267
pounds. One hypothesis suggests that a new synthetic balance takes place in the cell such that RNA, protein, and cell-wall synthesis decreases, but maintaining the rate of membrane synthesis (652). It is speculated (653) that the inhibition in the synthesis of any metabolite that maintains the proper biosynthetic balance might induce sporulation. Decoyinine blocks purine nucleotide biosynthesis and induces sporulation in B . subtilis even when metabolizable sources of carbon, nitrogen, and phosphate are present. Under these conditions, cells do not usually sporulate; the limitation of a nucleotide (GMP in this case) required for RNA or DNA synthesis is apparently sufficient to initiate sporulation. The chemical syntheses of the deoxy analogs and halo derivatives of psicofuranine and decoyinine have been reported (6544 5 7 ) , also of isocytidine 4',5'-unsaturated nucleoside (658)and of a number of saturated and unsaturated adenine nucleoside analogs (659-663). The nucleoside 4(Gdeoxy-p-L-idofuranosy1)adenineis a substrate for adenosine deaminase and may be somewhat related to 6deoxyp-D-dofuranosyladenine, which is an inhibitor of phosphorfbosyltran sferase (664),
2. PYRAZOFURIN Pyrazofurin (previously named pyrazomycin),4hydroxy-3-~-~ribofuranosyl-IH-pyrazole-Scarboxamide(90) is a C-nucleoside anti-
PYRAZOFURIN 90
RlBAVlRlN 91
biotic isolated from the culture filtrates of S. candidus (665,666). The chemical synthesis has been described (667,668). It has been of considerable interest because of its antitumor and broad-spectrum antiviral activity. Pyrazofurin suppresses the replication of vaccinia, herpes simplex, measles, rhino, influenza, and Rauscher leukemia viruses and of transplantable tumors (669-672). Pyrazofurin is currently being tested as an antitumor agent in man (670,673, 674). The 4amino analog has been synthesized from the 1-oxide(adenosine
268
ROBERT J. SUHADOLNIK
numbering) of formycin or formycin B (193). This replacement of the Chydroxyl group of pyrazofurin with an amino group results in a loss of cytotoxicity. Pyrazofurin is biologically phosphorylated to the 5'-phosphate, the active form, which inhibits orotidylate decarboxylase; this explains the increased urinary orotidine and orotic acid (675,676). Pyrazofurin 5'-phosphate, a structural analog of aminoimidazole carboxamide ribonucleotide (29) (a precursor of IMP), inhibits aminoimidazole carboxamide ribonucleotide transformylase in rat liver supernatants 97100%at a concentration 2.5 times that of aminoimidazole carboxamide ribonucleotide. Pyrafozurin (30 mg/kg) given to rats increases the concentration of 29 in the urine 330% over the controls (676~).Pyrazofurin triphosphate and AMP are converted by myokinase to ADP and pyrazofurin diphosphate (676). In patients with inoperable carcinoma and lymphoma, pyrazofurin is well tolerated. In several patients, mucositis, leucopenia, and anemia were observed. Of four patients with breast carcinoma, two responded (673). Pyrazofurin inhibits polio and Coxsackie viruses, although there is a report that pyrazofurin does not inhibit these two viruses (670). Uridine reverses the inhibition of pyrazofurin. Herpes simplex virus types 1 and 2 seem equally sensitive to pyrazofurin and ribavirin 91 (198, 677, 678). Pyrazofurin is 100-1000 times more effective than ribavirin in inhibiting vaccinia, polio, Coxsackie, measles, Sindbis, and VS viruses (677). In newborn mice inoculated subcutaneously with Coxsackie B4 virus, pyrazofurin failed to reduce the mortality rate (677). However, it was effective in inhibiting vaccinia virus (669, 670, 677). Pyrazofurin could not be administered at doses greater than 1mg/kg/day because of its toxicity. To reduce the toxicity, De CIercq et al. (679) synthesized the carbocyclic analog, but this compound had no antiviral activity. Pyrazofurin has been studied for its effect on Walker carcinoma 256, mammary carcinoma 755, Gardner lymphosarcoma, and X5563 plasma cell myeloma (672). The incorporations of adenosine and deoxyuridine into DNA were reduced by 85% and 50% respectively ( 6 7 2 ~ )There . was also an 85% inhibition of incorporation of adenosine into RNA, which supports the view that pyrazofurin inhibits de nova synthesis of pyrimidine nucleosides. The reversa1 of the inhibition of pyrazofurin by thymidine and deoxycytidine suggests that the inhibition of DNA synthesis by pyrazofurin is direct rather than through inhibition of RNA synthesis (672~).
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
269
VII. Hyperesthetic and Hyperemic Agents CLITIDINE It has long been known that ingestion of the toadstool Clitocybe acromelalga results in an increased sensitivity of the skin and an excess of blood in fingers and toes (hyperesthesia, hyperemia). Among the physiologically active substances is a pyridine nucleoside, clitidine [ 1,4dihydro-4imino-1-(P-~-ribofuranosyl)pyridine-3-carboxylic acid] (92). The discovery, isolation, production, physical and chemical properties, chemical and enzymic synthesis have been described (680, 681 ).
HoH2eiJ HO
OH
CLlTlDlNE I,4-DIHYDRO-4-IMINO-I-(~-D -RIBOFURANOSYL 1PYRIDINE-3-CARBOXYLIC ACID 92
The hyperesthetic and hyperemic activity of clitidine is now under investigation. Clitidine, at 0.05 g/kg, is toxic to mice (682).The incubation of NAD with methyl 4-aminonicotinate utilizing the transglycosidation action of pig brain NADase resulted in a 41% exchange of nicotinamide to form the “clitidine NAD” (681, 683).
VI 11. Cyclic-AMP Phosphodiesterase Inhi bitors 1. OCTOSYL ACIDS Three unique 5-substituted-uracil nucleosides have been isolated from S . cacaoi var. asoensis (684)and named octosyl acids A, B, and C (93, 94, 95). Octosyl acid A (93) is 1-(3,Sanhydr0-6-deoxy-~-glycero-P-D-a1!10octofuranosyluronic acid)-5carboxyuracil; octosyl acid B (94) is 1-(3,7anhydro-Sdeoxy-D-glycero-p-D-allo-octofuranosyluronic acid)-5hydroxymethyluracil; octosyl acid C (95) is 1-(3,7-anhydro-6-deoxy-~glycero -~-lyxo -octofuranosyl-5-ulosuronic acid)-5-carboxyuraci1.
270
ROBERT J. SUHADOLNIK
&@ J$. HOOC
93 OCTOSYL ACID A R = COOH
94 OCTOSYL ACID B
95 OCTOSYL ACID C R
= COOH
R = CH,OH
The octosyl acids are the first examples of naturally occurring trans-fused anhydro-octouronic acid nucleosides. The fused sugar skeleton of the octosyl acids is the same as that found in the ezomycins (Section V, 4; 68-75). The discovery, production, isolation, physical and chemical properties, and chemical synthesis have been described (684, 685). Octosyl acid A may be regarded as an analog of CAMP. The addition of cCMP of Ll210 cells in culture abolishes the temperature-dependent log phase and stimulates the resumption of growth or proliferation of those leukemic cells (686).The chemical transglycosylation of the glycan of octosyl acid A to an adenine (684) yields a competitive inhibitor of CAMPphosphodiesterase (K. Isono, personal communication).
IX. Miscellaneous Naturally Occurring Nucleosides 1. 0-GLYCOSYLF~BONUCLEOSIDES
A group of disaccharide nucleosides isolated from the culture medium of Brevibacterium ammoniagenes and Bacillus sp. 102 (687) have the structural elements of 5’~-hexosyl-ribonucleosides(96-98). The aglycon is either hypoxanthine or guanine and the added aldohexose is galactose, glucose or maltose in a or p linkage. The chemical synthesis of 0-glycosylpyranosylinosine has been reported (688). Nothing is known concerning any inhibitory properties these unique purine nucleosides may have.
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
R'H
0 -GLYCOSYL 96
2.
271
OR d-D-GLUCOSYL
RIBONUCLEOSIDES
- 98
RAPHANATIN AND Ns-BENZYL-7-P-D-GLUCOPYRANOSYJADENINE
Raphanatin (99)and N6-benzyl-7-~-~-glucopyranosyladenine (100) are active cytokinins and metabolites of zeatin [6-(4hydroxy-3 methylbut-trans-2-enylamino)purine] and 6-benzylarninopuriney respectively (689, 690). From mass spectral data, the metabolite of 6-benzylaminopurine was first assigned the structure Ns-benzyl-7glucofuranosyladenine (689). However, chemical syntheses show that 99 and 100 are 7-P-~-glucopyranosyl-N~-[(E )-4hydroxy-3-methyl-2 butenylladenine and N6-benzyl-7-~-~-glucopyranosyladenine, respectively (691).
Ql
HOH-C
HO
RAPHANATIN
99
-
N6 BENZYL-7-P-D- GLUCOPYRANOSYLADENINE
100
3. HERBICIDINS A AND B Two adenine nucleosides having herbicidal activity have been isolated from Streptornyces saganonensis (692,693). Herbicidins A and B inhibit Chinese cabbage germination at concentrations lower than those that affect rice germination. Herbicidin A selectively inhibits rice germination to a greater extent than does herbicidin B. Both, at 30 ppm, protect rice from bacterial blight. At 100 ppm, they are not toxic to mice (692, 693).
272
ROBERT J. SUHADOLNIK
ACKNOWLEDGMENTS I wish to express my sincere appreciation to Nancy L. Reichenbach, Diane M. Lichtenwalner, Joseph Wu, Chi P. Cheung, Alan Bruzel, and Michael B. Lennon for their valuable assistance and patience in preparing this review. I am also extremely grateful to those principal investigators throughout the world who have given their time so generously to my numerous requests for reprints to update all the naturally occurring nucleoside and nucleotide antibiotics and to those scientists who read, criticized, and made constructive comments relating to the material in this review. Research carried out in the author’s laboratory is supported by grants from the National Institutes of Health (No. AIl2066) and the National Science Foundation (No, PCM 77-24287).
REFERENCES 1. R. J. Suhadolnik, “Nucleoside Antibiotics,” Wiley, New York, 1970. 2. J. J. Fox, K. A. Watanabe and A. Bloch, This Series 5, 251 (1966); H. Umezawa, “Index of Antibiotics from Actinoniycetes,” Univ. Tokyo Press, Tokyo, 1970; A. Bloch, in “Drug Design” (E. J. Ariens, ed.), Vol. 4, p. 286. Academic Press, New York, 1973; J. A. Montgomery, in “Antineoplastic and Immunosuppressive Agents” (A. C. Sartorelli and D. C . Johns, eds.), Part I, p. 76. Springer-Verlag, Berlin and New York, 1974; J. W. Corcoran and F. E. Hahn, eds., “Mechanism of Action of Antimicrobial and Antitumor Agents.” Springer-Verlag, Berlin and New York, 1975; D. W. Visser, in “Antineoplastic and Immunosuppressive Agents” (A. C. Sartorelli and D. C . Johns, eds.), p. 343. Springer-Verlag, Berlin and New York, 1975; W. H. Prusoff and D. C. Ward, Biochem. Pharmacol. 25, 1233 (1976); H. Weissbach and S . Pestka, eds., “Molecular Mechanisms of Protein Biosynthesis,” Academic Press, New York, 1977; A. Bloch, i n “Handbook of Chemical Technology,” Wiley, New York, 1978, in press; S. Nakamura and H. Kondo,Heterocycles 8, 583 (1977). 3. J. cernti, I. Rychlik and F. W. Lichtenthaler, FEBS Lett. 30, 147 (1973). 4. F. W. Lichtenthaler, J. cernti and I. Rychlik, FEBS Lett. 53, 184 (1975). 5. F. W. Lichtenthaler and G. Trummlitz, F E B S Lett. 38, 237 (1974). 6 . J. M. Clark and A. Y. Chang,JBC 240, 4734 (1965). 7 . M. Arai, T. Haneishi, R. Enokita and H. Kayamori,J. Antibiot. 27, 329 (1974). 8. T. Haneishi, A. Terahara and M. Arai,J. Antibiot. 27, 334 (1974). 9. F. W. Lichtenthaler, T. Morino, W. Winterfeldt and Y. Sanemitsu, Tetrahedron Lett. p. 3527 (1975)i 10. F. W. Lichtenthaler, T. Morino and W. Winterfeldt, Nucleic Acids Res., Spec. Publ. 1, S33 (1975). 11. T. Ikeuchi, F. Kitame, M. Kikuchi and N. Ishida,J. Antibiot. 25, 548 (1972). 12. Y. Sakagami, R. L. Chang, K.Watanabe, S. Ichikawa and Y. S. Wang, Abstr. Pap., Ferment. Technol. Today, Proc. Int. Ferment. Symp., 4th, 1972, p. 212 (1972). 13. K. Uchida, T. Ichikawa, Y. Shimauchi, T. Ishikura and A. Ozaki,J. Antibiot. 24,259 (1971). 14. R. L. Hamill and M. M. Hoehn,]. Antibiot. 17, 100 (1964). 15. S. Ennifar, B. C. Das, S. M. Nash and R. Nagarajan,JCS, Chern. Commui. p. 41 (1977). 16. M. Vuilhorgne, S. Ennifar, B. C. Das, J. W. Paschal, R. Nagarajan, E. W. Hagaman and E. Wenkert, J. Org. Chem. 42, 3289 (1977).
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
273
17. C. DeBoer and J. W. Hinman, U.S. Patent. 2,909,463(1959);Chem. Ahstr. 54,
2669g (1960). 18. C. DeBoer and J. W. Hinman, U.S. Patent. 2,909,464(1959);Chem. Abstr. 54,26693
(1960). 19. M. Konishi, M. Kimeda, H. Tsukiura, H. Yamamoto, T. Hoshiya, T. Mijaki, K-1. Fujisawa, H. Koshiyama and H. Kawaguchi,]. Antibiot. 26, 752 (1973). 20. K. Toniita, Y. Uenoyama, K.-I. Fujisawa and H. Kawaguchi, J . Antibiot. 26, 765
(1973). J. R. Evans and G. Weare,]. Antibiot. 30,604 (1977). R. E. Monro, Nature 233,903 (1969). M. M. Sikorski, J. CernL, I. Rychlik and A. B. Legocki, BBA 475, 123 (1977). R. E.Monro, R. Fernandez-Munoz, M. L. Celma, A. Jiminez and D. Vazquez, Prog. Antimicrob. Anticancer T h r . 2, 473 (1970). 25. M. L. Celma, R. E. Monro and D. Vazquez, FEBS Lett. 6,273(1970). 26. J. Cerna, F. W. Lichtenthaler and I. Rychlik, FEBS Lett. 14,45 (1971). 27. T.Hishizawa and S. Pestka, ABB 147, 624 (1971). 28. S. Pestka, PNAS 69, 624 (1972). 29. S. Pestka,/BC 247, 4669 (1972). 30. S. Pestka, H. Rosenfeld, R. Harris and H. Hintikka,JBC 247, 6895 (1972). 31. M. Barbacid and D. Vazquez, EJB 44, 445 (1974). 32. L. Carrasco and D. Vazquez, J. Antibiot. 25, 732 (1972). 33. D.Vazquez, FEBS Lett. 40, S63 (1974). 34. E. Battaner and D. Vazquez, BBA 254, 316 (1971). 35. C.T. Caskey and A. L. Beaudet, in “Molecular Mechanisms of Antibiotic Action on Protein Biosynthesis and Membranes” (E. Munoz, F. Garcia-Ferrandiz and D. Vazquez, eds.), p. 326.Am. Elsevier, New York, 1972. 36. D. Nathans, in “Antibiotics” (D. Gottleib and P. D. Shaw, eds.), Vol. 1, p. 259. Springer-Verlag, Berlin and New York, 1967. 37. N . J . Leonard and K. L. Carraway,J. Heterocycl. Chem. 3, 485 (1966). 38. N. J. Leonard, F. C. Sciavolino and V. Nair,]. Org. Chem. 33, 3169 (1968). 39. V. Nair and D. J. Emmanue1,JACS 99, 1571 (1977). 40. R. J. Harris and S. Pestka, in “Molecular Mechanisms of Protein Biosynthesis” (H. Weissbach and S. Pestka, eds.), p. 246.Academic Press, New York, 1975. 41. B. Edens, H. A. Thompson and K. Moldave, Bchem 14, 54 (1975). 42. R. Benne and H. 0. Voorma, FEBS Lett. 20,347 (1972). 43. Y. Aharonowitz and E. Z. Ron, FEBS Lett. 52,25 (1975). 44. J. S. Dubnoff, A. H. Lockwood and U. Maitra,/BC 247, 2884 (1972). 44a. E.A. Stringer, P. Sarkar and U. Maitra,JBC 252, 1739 (1977). 45. R. G. Crystal, N. A. Elson and W. F. Anderson, in “Methods in Enzymology” (L.Grossman andK. Moldave,eds.),Vol.30,PartF,p.Il3.Academic Press,NewYork, 1974. 46. M. Tresno, L. Carrasco and D. Vazquez, EJB 68, 355 (1976). 47. C. P. Cheung, M. L. Stewart and N. K. Gupta, BBRC 54,1092 (1973). 48. R. E. Monro and D. Vazquez,JMB 28, 161 (1967). 49. J. Jayaraman and J. H. Goldberg, Bchem 7, 418 (1968). 50. J. H. Goldberg and K. Mitsugi, Bchem 6, 383 (1967). 51. S. N. Seal and A. Marcus, BBRC 46, 1895 (1972). 52. L. S. Kappen, H. Suzuki and J. H. Goldberg, PNAS 70,22 (1973). 53. J. Wu, C. P. Cheung and R. J. Suhadolnik, BBRC 78, 1079 (1977). 54. R. E.Kohler, E. Z. Ron and B. D. Davis,JMB 36, 71 (1968). 21. 22. 23. 24.
274
ROBERT J. SUHADOLNIK
A. R. Subramanian, B. D. Davis and R. J. Beller, CSHSQB 34,223 (1969). M. E. Azzam and I. D. Algranati, PNAS 70, 3862 (1973). A. R. Subramanian and B. D. Davis, PNAS 68,2453 (1971). S. Sabol, M. A. G. Sillero, K. Iwasaki and S. Ochoa, Nature 228, 1269 (1976). A. R. Subramanian and B. D. Davis, Nature 228, 1273 (1970). J. S. Dubnoff and U. Maitra, PNAS 68, 313 (1971). 61. R. Kaempfer, Nature 228, 534 (1970). 62. A. R. Subramanian and B. D. Davis, ]MB 74,45 (1973). 63. K. B. Henry, Science 158, 938 (1967). 64. B. D. Lieff, S . K. Sharpless and K. Schlesinger,]. Comp. Physiol. Psychol. 90, 773 (1976). 65. S. H. Barondes and H. D. Cohen, Science 151, 549 (1966). 66. L. B. Flexner and J. B. Flexner, PNAS 55, 369 (1966). 67. B. W. Agranoff, R. E. Davis and J. J. Brink, Brain Res. 1, 303 (1966). 68. S. H. Barondes, Znt. Rev. Neurobiol. 12, 177 (1970). 69. P. Paggi and G. Toschi,]. Neurobiol. 2, 119 (1971). 70. V. J. Wu& Pharmacol., Biochem. Behuu. 1,177 (1973). 71. D. E. Moss and D. Fahrney,]. Neurochem. 26, 1155 (1976). 72. D. R. Moss, D. E. Moss and D. Fahrney, BBA 350,95 (1974). 73. S. Daluge and R. Vince,]. Med. Chem. 15, 171 (1972). 74. R. Vince and R. G. Isakson,]. Med. Chem. 16, 37 (1973). 75. R. Vince, R. G. Almquist, C. L. Ritter, F. N. Shirota and H. T. Nagasawa, L$e Sci. 18, 345 (1976). 76. R. H. Symons, R. J. Harris, L. P. Clarke, J. F. Wheldrake and W. H. Elliot, BBA 179, 248 (1969). 77. R. Vince, S. Daluge and M. Palm, BBRC 46, 866 (1972). 78. R. Vince and S. Daluge,]. Med. Chem. 20,930 (1977). 79. A. J. Guarino and N. M. Kredich, FP 23, 371 (1964). 80. A. J. Guarino, M. L. Ibershof and R. Swain, BBA 72, 62 (1963). 81. G. 0. Morton, J. E. Lancaster, G. E. VanLear, W. Fulmor and W. E. MeyerJACS 91, 1535 (1969). 82. I. D. Jenkins, J. P. H. Verheyden and J. G. Moffatt,JACS 93, 4323 (1971). 83. J. R. Florini, H. H. Bird and P. H. Bel1,JBC 241, 1091 (1966). 84. R. L. Hamill and M. M. Hoehn, 16th Annu. ZCAAC Meet. 1976 Abstr. 60 (1976). 85. J. K. Epp and N. E. Allen, 16th Annu. ICACC Meet. 1976 Abstr. 63 (1976). 86, T. Nakanishi, F. Tomita and T. Suzuki, Agric. Biol. Chem. 38, 2465 (1974). 87. T. Nakanishi, T. Iida, F. Tomita and A. Furuya, Chem. Pharm. Bull. 24,2955 (1976). 88. M. Ikehara, T. Maruyma and H. Miki, Tetrahedron Lett. p. 4485 (1976). 89. J. B. Hobbs and F. Eckstein,]. Org. Chem. 42, 714 (1977). 90. J. Hochstadt-Ozer,]BC 247,2419 (1972). 91. T. Nakanishi, F. Tomita and A. Furuya, /. Antibiot. 30, 743 (1977). 92. A. D. Argoudelis and J. H. Coats,]. Antibiot. 24, 206 (1971). 93. A. D. Argoudelis and J. H. Coats,]. Antibiot. 22, 341 (1969). 94. A. D. Argoudelis, J. H. Coats, D. J. Mason and 0. K. Sebek,]. Antihiot. 22, 309 (1969). 95. J. H. Coats and A. D. Argoudelis,]. Bact. 108, 459 (1971). 96. A. D. Argoudelis, J. H. Coats and S. A. Mizsak,]. Antibiot. 30, 474 (1977). 97. J. T.Truman and H. Klenow, Mol. Pharmacol. 4, 77 (1968). 98. T. H. Fraser and A. Rich, PNAS 70, 2671 (1973). 99. Y. F. Shealy and J. D. Clayton,JACS 88, 3885 (1966); 91, 3075 (1969).
55. 56. 57. 58. 59. 60.
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
275
100. T. H. Kusaka, H. Yamamoto, M. Shibata, M. Muroi, T. Kishi and K. Mizuno, Chem. Commun. p. 852 (1967). 101. L. L. Bennett, Jr., P. W. Allan and D. L. Hill, Mol. Pharmucol. 4, 208 (1968). 102. M. Ikehara and T. Fukui,J. Biochem. (Tokyo) 73, 945 (1973). 103. T. Kusaka,J. Antihiot. 24, 756 (1971). 104. T. Kusaka,J. Takeda Res. Lab. 31, 85 (1972). 105. R. T. Borchardt and Y.4. Wu,J. Med. Chem. 19, 197 (1976). 106. C. S. G . Pugh, R. T. Borchardt and H. 0. Stone, Bchem 16, 3928 (1977). 107. A. J. Shatkin, This Series 19, 3 (1976). 108. G . W. Both, A. K. Banexjee and A. J. Shatkin, PNAS 72, 1189 (1975). 109. S . Muthukrishnan, G . W. Both, Y. Furuichi and A. J. Shatkin, Nature 255,33 (1975). 110. S. Muthukrishnan, W. Filipowicz, J. M. Sierra, G . W. Both, A. J. Shatkin and S. Ochoa, JBC 250, 9336 (1975). 111. S. K. Shapiro and F. Schlenk, eds., “Transmethylation and Methionine Biosynthesis.” Univ. of Chicago Press, Chicago, Illinois, 1965. 112. J. K. Coward and E. P. Slisz,J. Med. Chem. 16, 460 (1973). 113. H. Hagimoto, H. Yoshikawa and H. Taniura, 4th Annu. Meet. SOC.Chem. Regul. Plants, 1969 p. 27 (1969). 114. R. Marumoto, Y. Yoshioka, Y. Furukawa and M. Honjo, Chem. Pharm. Bull. 24, 2624 (1976). 115. S . Daluge and R. Vince, Tetrahedron Lett. 35, 3005 (1976). 116. R. Marumoto, T. Nishimura and M. Honjo, Chem. Pharm. Bull. 23, 2295 (1975). 117. M. Ikehara, I. Tanzawa and T. Fukui, Bchem 8, 736 (1969). 118. J. Pliml and F. Sorm, Collect. Czech. Chem. Commun. 29, 2576 (1964). 119. L. J. Halika, J. S. Evans, D. J. Mason and A. Dietz,Antimicrob. Agents Chemother. p. 619 (1966). 120. M. E. Bergy and R. R. Herr, Antimicroh. Agents Chemother. p. 625 (1966). 121. A. CihBk, Oncology 30, 405 (1974). 122. J. Veselp and A. Cihiik, Int. Encycl. Pharmucol. Ther., in press (1978). 123. J. Vesel>;and A. Cihrik, Cas. Cesk. Lek. 114, 607 (1975). 124. D. D. von Hoff, M. Salik and F. M. Muggia, Ann. Intern. Med. 85, 237 (1976). 125. V. Paces, J. DoskoEil and F. Sorm, BBA 161, 352 (1968). 126. K. Ralka, M. Juroveik, V. FuEk, R. Tykva, Z. SormovA and F. Somi, Collect. Czech. Chem. Commun. 31,2809 (1966). 127. A. Cihiik, J. Vesely and F. Sorm, BBA 134, 486 (1967). 128. A. Cihik, J. Vesely and F. Sorm, Collect. Czech. Chem. Commun. 32,3427 (1967). 129. A. Cihiik, H. Veselri and F. $om,BBA 166, 277 (1968). 130. B. S. Zain, R. L. P. Adams and R. C. Imrie, Cancer Res. 33, 40 (1973). 131. R. Shutt and R. G. Krueger,]. lmmunol. 108, 819 (1972). 132. T. T. Lee and R. L. Momparler, Med. Pediatr. Oncol. 2, 265 (1976). 133. M. Reichman and S . Penman, BBA 324,282 (1973). 134. T. Lee, Proc. Am. Assoc. Cancer Res. 14, 94 (1973). 135. R. L. Monparler, S. Siegel, F. Avila, T. Lee and M. Karon, Biochem. Pharmacol. 25, 389 (1976). 136. T. Lee and M. R. Karon, Biochem. Pharmacol. 25, 1737 (1976). 137. J. W. Weiss and H. C. Pitot, ABB 160, 119 (1974). 138. M. Reichman, D . Karlan and S. Penman, BBA 299, 173 (1973). 139. A. Cihiik, J. W. Weiss and H. C. Pitot, Cancer Res. 34, 3003 (1974). 140. A. Cihiik, L. M. Narurkar and H. C. Pitot, Collect. Czech. Chem. Commun. 08,948 ( 1973).
276
ROBERT J. SUHADOLNIK
141. T. T. Lee and R. L. Momparler, Biochm. Pharmacol. 26, 403 (1977). 142. B. A. Chabner, J. C. Drake and D. G. Johns, Biochm. Pharmacol. 22,2763 (1973). 143. L. H. Li, E. J. Olin, H. H. Buskirk and L. M. Reineke, Cancer Res. 30,2760 (1970). 144. J. Veself, R. Costof, A. Cihak and F. Sorm, Z . Naturforsch. 24B, 318 (1969). 145. P. Pithova, A. Piskala, J. Pitha and F. $om, Collect. Czech. Chem. Commun. 30, 2801 (1965).
146. S. Zadraiil, V. Fufik, M. JurovEik and Z. Sormovi, Collect. Czech. C h m . Commun. 37, 309 (1972). 147. M. Karon and W. F. Benedict, Science 178, 62 (1972). 148. K. RaSka, M. Jurovfik, Z. Sormovii and F. $om, Collect. Czech. Chem. Commun. 31, 2803 (1966).
149. P. Pithovs, V. Fufik, S. Zadraiil, Z. Sormovii and F. Sorm, Collect. Czech. Chem. Commun. 30, 2879 (1965). 150. J. Veselv, A. Cihik and F. Sorm, Cancer Res. 28, 1995 (1968). 151. S. Zadraiil, V. Fufik, P. B a d , Z. Sormovri and F. Sorm, BBA 108, 701 (1965). 152. V. FuCik, A. Michaelis and R. Rieger, Mutat. Res. 9, 599 (1970). 153. A. Cihiik, M. Seifertovs and P. Riches, Cancer Res. 36, 37 (1976). 154. R. Crkvenjakov, N. BajkoviE and V. GliSin, BBRC 39, 655 (1970). 155. J. A. V. Simpson and R. Baserga, Lab. Invest. 24,464 (1971). 156. A. Cihhk, J, Veself and F. Sorm, Collect. Czech. Chem. Commun., Engl. E d . 34, 1060 (1969).
157. 0.Heby and D. H. Russell, in “Polyamines in Normal and Neoplastic Growth” (D. H. Russell, ed.), p. 221. Raven, New York, 1973. 158. 0.Heby and D. H. Russell, Cancer Res. 33, 159 (1973). 159. J. Doskofil and V. Paces, Collect Czech. Chem. Commun. 33, 4369 (1968). 160. J. DoskoEil and F. Sorm, EJB 13, 180 (1970). 161. J. Veself, A. C i h g and F. Sorm, Collect. Czech. Chem. Commun. 34, 901 (1969). 162. J. Vesel9 and A. CihAk, Cancer Res. 37,3684 (1977). 163. A. tihAk, Eur.J. Cancer, 14, 117 (1978). 164. J. A. Beisler, M. M. Abbasi and J. S. Driscoll, Cancer Treat. Rep. 60, 1671 (1976). 165. J. A. Beisler, M. M. Abbasi, J. A. Kelley and J. S. DriscoII, J . M e d . Chem. 20, 806 (1977). 166. J. A. Beisler, M. M. Abbasi, J. A. Kelley and J. S. Driscol1,J. Carbohydr.Nucleosides
Nucleotides 4, 281 (1977). 167. P. E. Palm, E. P. Arnold, P. C. Rachwall and M. S. Nick, Toxicol.A p p l . Pharmacol. 25, 492 (1973). 168. P. E. Palm and C. J. Kensler, U.S. C . F. S. T. I . P B . R e p . 194791, 191 (1970). 169. E. Coles, P. S. Thayer, V. Reinhold, and L. Gaudio, Proc. Am. Assoc. Cancer Res. 15, 72 (1974).
170. D. D. von Hoff, H. Handelsman and M. Slavik, “Clinical Brochure.” Natl Cancer Inst., 1975. 171. L. J. Townsend, in “Handbook of Biochemistry and Molecular Biology” (G.’D. Fasman, ed.), 3rd ed., Vol. 1, p. 271. CRC Press, Cleveland, Ohio, 1975. 172. M. Jurovfik, K. RaSka, F. Sorm and Z. SormovR, Collect. Czech. Chem. Commun., E n g l . E d . 30, 3370 (1965).
173. W. M. Troetel, A. J. Weiss, J. E. Stambaugh, J. F. Laucius and R. W. Manthei, Cancer Chemother. R e p . 56,405 (1972). 174. Z. H. Israili, W. R. Vogler, E. S. Mingioli, J. L. Pirkle, R. W. Smithwick and J. H. Goldstein, Pharmacologist 16, 231 (1974). 175. C. A. Presant, T. Vietti and F. Valeriote, Cancer Res. 35, 1926 (1975).
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
277
176. J. H. Saiki, K. McCredie, T. Vietti, J. Hewlett and F. Morrison, Proc. Am. Assoc.
Cancer Res. 18, 399 (1977). 177. W. R. Vogler, D. S. Miller and J. W. Keller, Blood 48, 331 (1976). 178. K. B. McCredie, in “Minutes of New Drug Liason Meeting,” p. 31. Natl. Cancer
Inst., Bethesda, Maryland, 1975. 179. C. Tan, J. R. Burchenal, B. Clarkson, M. Feinstein, E. Garcia, J. Sidhu and I. H. Krakoff, Proc. Am. Assoc. Cancer Res. 14, 97 (1973). 180. J. A. Levi and P. H. Wiernik, Cancer Chernother. Rep. 59, 1043 (1975). 181. J. R. Bateman, in “Minutes of the New Drug Liason Meeting,” p. 32. Natl. Cancer
Inst., Bethesda, Maryland, 1975. 182. J. Z. Finkelstein, in “Minutes of the New Drug Liason Meeting,” p. 32. Natl.
Cancer Inst., Bethesda, Maryland, 1975. 183. J. M. Quagliana, J. Costanzi and R. O’Bryan, Proc. Am. Assoc. Cancer Res. 15, 121
(1974). 184. T. J. Cunningham, T. Nemoto, D. Rosner, E. Knight, S. Taylor, C. Rosenbaum, J.
Horton and T. Dao, Cancer Chemother. Rep. 58, 677 (1974). 185. R. E. Bellet, M. J. Mastrangelo, P. F. Engstrom, J. G. Strawitz, A. J. Weiss and J. W. Yarbro, Cancer Chemother. Rep. 58,217 (1974). 186. C. G. Moertel, A. J. Schutt, R. J. Reitemeir and R. G. Hahn, Cancer Chernother.
Rep. 56, 649 (1972). 187. R. W. Vogler, S. Arkun and E. Velez-Garcia, Cancer Chmother. Rep. 58, 895
(1974). 188. G. A. Omura, Cancer Treat. Rep. 61, 915 (1977). 189. G. A. Omura, Proc. Am. Assoc. Cancer Res. 18, 25 (1977). 190. A. J. Weiss, E. E. Stambaugh, M. J. Mastrangelo, J. F. Laucius and R. E. Bellet,
Cancer Chemother. Rep. 56, 413 (1972). 191. P. L. Lomen, V. K. Vaitnevicius and M. K. Samson, Proc. Am. Assoc. Cancer Res. 16,
52 (1975). 192. K. Mizuno, M. Tsujino, M. Takada, M. Hayaski, K. Atsumi, K. Asano and T. Mat-
suda, J. Antibiot. 27, 775 (1974). 193. A. F. Lewis, R. A. Long, L. W. Roti-Roti and L. B. Townsend,J. Heterocycl. Chem.
13, 1359 (1976). 194. L. B. Townsend, Chem. Rev. 67, 533 (1967). 195. H. Iwata, H. Iwaki, T. Masukawa, S. Kasmatsu and H. Okanioto, Experientia 33,
502 ( 1977). 196. K. Sakaguchi, M. Tsujino, M. Yoshizawa, K. Mizuno and K. Hayano, Cancer Res.
35, 1643 (1975). 197. K. Sakaguchi, M. Tsujino, K. Mizuno and K. Hayano, Jpn.J. Genet. 51, 61 (1976). 198. D. G. Streeter, J, T. Witkowski, G. P. Khare, R. W. Sidwell, R. J. Bauer, R. K. Robins
and L. N. Simon, PNAS 70, 1174 (1973). 199. K. Sakaguchi, M. Tsujino, K. Mizuno, K. Hayano, and N. Ishida, J. Antibiot. 28,798
(1975). 200. K. Sakaguchi, M. Tsujino, M. Hayashi, K. Kawai, K. Mizuno and K. Hayano, J. Antibiot. 29, 1302 (1976). 201. M. Hayashi, T. Hirano, M. Yaso, K. Mizuno and T. Ueda, Chem. Phann. Bull. 23,
245 (1975). 202. K. G. Cunningham, S. A. Hutchinson, W. Manson and F. S. Spring, JCS p. 2299
(1951). 203. E. A. Kaczka, E. L. Dulaney, C. 0. Gitterman, H. B. Woodruff and K. Folkers,
BBRC 14, 452 (1964).
278
ROBERT J. SUHADOLNIK
204. S . Frederiksen, BBA 76, 366 (1963). 205. H. Klenow and K. Overgaard-Hansen, BBA 80,500 (1964). 206. M. A. Rich, P. Meyers, G. Weinbaum, J. G. Cory and R. J. Suhadolnik, BBA 95,194 (1965). 207. J. G. Cory, R. J. Suhadolnik, B. Resnick and M. A. Rich, BBA 103, 646 (1965). 208. W. Plunkett and S. S. Cohen, Cancer Res. 35,1547 (1975). 209. D. G. Johns and R. H. Adamson, Biochem. Pharmacol. 25,1441 (1976). 210. R. I. Gumport, E. B. Edelheit, T. Uematsu and R. J. Suhadolnik, Bchem. 15, 2804 (1976). 211. N. Craig,]. Cell. Physiol. 82, 133 (1973). 212. W. E. G. Miiller, G. Seibert, R. Beyer, H. J. Breter, A. Maidhof and R. K. Zahn, Cancer Res. 37, 3824 (1977). 213. K. M. Rose, L. E. Bell and S. T. Jacob, BBA 475, 548 (1977). 214. K. M. Rose, L. E. Bell and S. T. Jacob, Nature 267,178 (1977). 215. A. Bruzel, R. J. Suhadolnik and J. K. Hoober, FP 36, 909 (Ac3364) (1977). 216. E. Westhof, H. Plach, I. Cunu and H.-D. Ludemann, Nucleic Acids Res. 4, 939 (1977). 217. J. E. Sylvester and D. Dennis, BBRC 75, 667 (1977). 218. H. Fouquet, R. Wick, R. Bohme, H. W. Sauer and K. Scheller,ABB 168,273 (1975). 219. M. Siev, R. Weinberg and S. Penman,./. Cell Bid. 41, 510 (1969). 220. S. Penman, M. Rosbash and M. Penman, PNAS 67,1878 (1970). 221. J. Mendecki, S . Y. Lee and G. Brawerman, Bchem. 11,792 (1972). 222. J. E. Darnell, W. R. Jelinek and G. R. Molloy, Science 181, 1215 (1973). 223. H. Nakazato, M. Edmonds and P. W. Kopp, PNAS 71,200 (1974). 224. A. J. Rizzo, C. Kelly and T. E. Webb, Can. J. Biochem. 50, 1010 (1972). 225. M. Edmonds and R. Abrams,JBC 235, 1142 (1960). 226. J. Niessing, EJB 59, 127 (1975). 227. W. Jelinek, M. Adesnik, M. Salditt, D. Sheiness, R. Wall, G. Molloy, L. Philipson and J. E. Darnel1,JMB 75, 515 (1973). 228. S. G. Sawicki, W. Jelinek and J. E. Damel1,JMB 113, 219 (1977). 229. K. M. Rose and S. T. Jacob, Bchem. 15, 5046 (1976). 230. M. Edmonds and M. A. Winters, This Series 17, 149 (1976). 231. S. P. Blatti, C. J. Ingles, T.J. Lindell, P. W. Morris, R. F. Weaver, F. Weinberg and W. J. Rutter, CSHSQB 35, 649 (1970). 232. M. Linder-Horowitz, R. T. Ruettinger and H. N. Munro, BBA 200, 442 (1970). 233. J. W. Drysdale and H. N. Munro,JBC 241, 3630 (1966). 234. J. Zahringer, A. M. Konijn, B. S. Baliga and H. N. Munro, BBRC 65, 583 (1975). 235. S. M. Tilghman, R. W. Hanson, L. Reshef, M. F.Hopgood and F. J. Ballard, PNAS 71, 1304 (1974). 236. J. Zahringer, B. S. Baliga and H. N. Munro, PNAS 73, 857 (1976). 237. G. C. Lavers, J. H. Chen and A. Spector,JMB 82,15 (1974). 238. B. Harris and L. S. Dure, Bchem. 13, 5463 (1974). 239. J. Rowinski, D. Solter and H. Koprowski,J. Exp. 2001.192, 133 (1975). 240. I. L. Levey and R. L. Brinster, Exp. Cell. Res. 109, 397 (1977). 241. J. R. Hammett and F. R. Katterman, Bchem 14, 4375 (1975). 242. M. Delseny, M. T. Peralta and Y. Guitton, BBRC 64, 1278 (1975). 243. S. Spiegel and A. Marcus, Nature 256, 228 (1975). 244. A. M. Wu, R. C. Ting, M. Paran and R. C. Gallo, PNAS 69, 3820 (1972). 245. L. S. Richardson, R. C. Ting, R. C. Gallo and A. M. Wu,Int.J. Cuncer 15,451 (1975).
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
279
246. B. W. J. Mahy, N. J. Cox, S. J. Armstrong and R. D. Barry, Nature N B 234, 172 (1973).
247. E. A. C. van Oortmerssen, B. A. Regensburg, J. G. Tasseron-de Jong, and L. Bosch, Virology 65, 238 (1975). 248. S. J. OBrien and C. W. Boone,]. Gen. Viral. 35, 511 (1977). 249. S. R. Weiss and M. A. Bratt, J. Viral. 16, 1575 (1975). 250. C. N. Nair and D. L. Panicali,J. Viral. 20, 170 (1976). 251. R. D. Walter and F. Ebert, Hoppe-Seyler’s 2. Physiol. Chem. 358, 23 (1977). 252. R. I. Glazer and J. F. Kuo, Biochem. PharmacoE. 26,1287 (1977). 252a. M. Legraverend, R. I. Glazer and D. G. Johns, Proc. Am. Assoc. Cancer Res. 19, Abstr. 437 (1978). 253. L. Leinwand and F. H. Ruddle, Science 197, 381 (1977). 254. A. Hampton and T. Sasaki, Bchem. 12, 2188 (1973). 255. R. J. Suhadolnik, M. B. Lemon, T. Uematsu, J. E. Monahan and R. BaurJBC 252, 4125 (1977).
256. P. T. Riquelme, L. 0. Burzio and S. S. Koide, FP 36, 785 (X2706) (1977). 257. P. Chambon, J. D. Weill, J. Doly, M. T. Strosser and P. Mandel, BBRC 75, 638 (1966).
258. R. J. Suhadolnik, R. Baur, D. M. Lichtenwalner, T. Uematsu, J. H. Roberts, S. Sudhakar and M. Sniulson,JBC 252, 4134 (1977). 259. W. E. G . Miiller and R. K. Zahn, Experientia 31, 1014 (1975). 260. M. Sprinzl and F. Cramer, Nature NB 245, 3 (1973). 261. F. von der Haar and F. Cranier, Bctzem. 15, 4131 (1976). 262. S. M. Hecht, J. W. Kozarich and F. J. Schmidt, PNAS 71,4317 (1974). 263. J. Ofengand and C. M. Chen, JBC 247, 2049 (1972). 264. R. S. Mulvey and A. R. Fersht, Bchem. 16, 4731 (1977). 265. P. B. Sehgal, E. Dernian, G. R. Molloy, I. Tamm and J. E. Darnell, Science 194,431 (1976).
266. P. B. Sehgal, E. Dernian, J. E. Darnell and I. Tamm,]. Cell Biol. 70, 244a (1976). 267. N. Nakajima, K. Ichikawa, M. Kaniada and E. FujitaJ. Agric. Chem. Soc.Jpn. 35, 797 (1961).
K. Morita and S. Kobayashi, Cheni. Pharm. B u l l . 15, 988 (1967). T. Kaneda and S. Tokuda,J. Nutr. 90, 371 (1966). I. Chibata, K. Okumura, S. Taneyama and K. Kotera, Experientia 25, 1237 (1969). T. Rokoju, H. Kikuchi, A. Tensho, Y. Tsukitani, T. Tiikenawa, K. Yashida and T. Kaniiya, Life Sci. 9, 379 (1970). 272. K. Takashima, C. Sato, Y. Sasaki, T. Morita and S. Takeyama, Biochem. Pharmncol.
268. 269. 270. 271.
23,433 (1974).
273. H. Seto, K. Furihata and H. Yonehara, Annu. Meet. Agric. Clzem. Sac. J p n . p. 258 (1976).
274. H. Seto and H. Yonehara, Annu. Meet. Agric. Chem. Sac. J p n . p. 87 (1975). 275. H. Seto, Agric. Biol. Chem. 37, 2415 (1973). 276. T. M. K. Chiu, H. Ohrui, K. A. Watanabe and J. J. Fox,J. Org. Chem. 38, 3622 (1973).
277. K. A. Watanabe, T. M. K. Chiu, D. H. Hollenberg and J. J. Fox,J.Org. Chem. 39, 2482 (1974).
278. K. A. Watanabe, T. M. K. Chiu, U. Reichman, C. K. Chiu and J. J. Fox, Tefrulzedron 32, 1493 (1976).
279. K. A. Watanabe, D. H. Hollenberg and J. J. Fox,]. Antihiof. 29, 597 (1976).
280
ROBERT J. SUHADOLNIK
J. J. Fox and I. Goodman,JACS 73,3256 (1951). H. Seto, N. atake and H. Yonehara,Agric. Bid. Chem. 37,2421 (1973). H. Seto, N. Otake and H. Yoneham, Tetruhedron Lett. p. 3991 (1972). H. Kasai, Y. Kuchino, K. Nihei and S. Nishimura,Nucleic Acids Res. 2, 1931 (1975); H. Kasai, Z. Ohashi, F. Harada, S. Nishimura, N. J. Oppenheimer, P. F. Crain, J. C . Liehr, D. L. von Minden and J. A. McCloskey, Bchem 14,4198 (1975). 284. R. J. Suhadolnik and T. Uematsu,JBC 245,4365 (1970). 285. Y. Kuchino, H. Kasai, K. Nihei and S. Nishimura,Nucleic Acids Res. 3,393 (1976). 286. C. S. Cheng, B. C. Hinshaw, R. P. Panzica and L. B. Townsend, JACS 98, 7870 ' (1976). 287. A. Bloch, R. J. Leonard and C. A. Nichol, BBA 138, 10 (1967). 288. M. Miko and L. Drobnica, Experientia 31, 832 (1975). 289. R. E. Parks and P. R. Brown, Bchem. 12, 3294 (197%. 290. R. J. Stegman, A. W. Senft, P. R. Brown and R. E. Parks, Biochem. Pharmacol. 22, 459 (1973). 291. E. Wainfan, J. Chu and G. B. Chheda, Biochem. Pharmucol. 24,83 (1973). 292. C. Baxter and P. Byvoet, Cancer Res. 34, 1418 (1974). 293. E. Wainfan and B. Landsberg, Biochem. Pharmacol. 22, 493 (1973). 294. R. Bassleer, A. Lepoint, F. De Paermentier and G. Goessens, Microsc. Bid. Cell 25, 33 (1976). 295. J. Hirsch and 0. J. Martelo, Life Sci. 19, 85 (1976). 296. R. D. Walter, BBA 429, 137 (1976). 297. K. B. Bhuyan, L. G . Scheidt and T. J. Fraser, Cancer Res. 32, 398 (1972). 298. F. Lawrence, D. J. Shire and J. P. Waller, EJB 41, 73 (1974). 299. A. Bloch, Ann. N.Y. Acad. Sci. 255,576 (1975). 300. M. Ikehara and T. Fukui, BBA 338,512 (1974). 301. R. P. Agarwal, S. M. Sagar and R. E. Parks, Biochem. Pharmacol. 24, 693 (1975). 302. C. G . Smith, L. M. Reineke, M. R. Burch, A. M. Shefner and E. E. Muirhead, Cancer Res. 30, 69 (1970). 303. A. F. Ross and J. J. Jaffe, Biochem. PharmacoZ. 21,3059 (1977). 304. J. F. Henderson, C. M. Smith, F. F. Snyder and G . Zombor,Ann. N.Y. Acad. Sci. 255,489 (1975). 305. R. J. Suhadolnik, S. I. Finkel and B. M. Chassy,JBC 243, 3532 (1968). 306. B. M. Chassy and R. J. Suhadolnik,]BC 243,3538 (1968). 307. S. A. Kumar, J. S. Kwkow and D. C. Ward, BBA 477, 112 (1977). 308. E. Borek and S. J. Kerr, Adu. Cancer Res. 15, 163 (1972). 309. J. K. Coward, D. L. Bussolotti and C. D. Chang,]. Med. Chem. 17, 1286 (1974). 310. C. D. Chang and J. K. Coward, Mol. PharmacoZ. 11, 701 (1975). 311. J. K. Coward, N. M. Motola and J. D. Moyer,J. Med. Chem. 20,500 (1977). 312. M. B. Lennon and R. J. Suhadolnik, BBA, in press (1978). 313. M. B. Lennon, J. Wu and R. J. Suhadolnik, BBRC 72,530 (1976). 314. M. Blecher, J. T. Ro'Ane and P. D. Flynn, Biochem. Pharmacol. 20,249 (1971). 315. C . H. Burgess, A. Bloch, H. Stoll, H. Milgram, F. Helm and E. Klein, Cancer 34, 250 (1974). 316. E. Klein, C . A. Burgess, A. Bloch, H. Milgram and 0. A. Halterman, Ann. N.Y. Acad. Sci. 255, 216 (1975). 317. A. W. Senft, R. P. Much, P. R. Brown and D. G. Senft,Znt.J. Parasitol. 2, l(1972). 318. A. W. Senft, G. W. Crabtree, K. C. Agarwal, E. M. Scholar and R.E. Parks,Biochem. Pharmacol. 22,449 (1973). 319. J. J. Jaffe, E. Meymarian and H. M. Doremus, Nature 230,408 (1971). 280. 281. 282. 283.
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
281
320. J. J. Jaffe, H. M. Dorernus, H. A. Dunsford and E. Meyrnarian,Am. J. Trop. Med. Hyg. 24, 835 (1975). 321. R. P. Much, A. W. Senft and D. G. Senft, Biochem. Pharmacol. 24,407 (1975). 322. E. De Clercq, P. F. Torrence and B. Witkop, PNAS 71, 182 (1974). 323. P. F. Torrence, E. D e Clercq, J. A. Waters and B. Witkop, Bchem. 13,4400 (1974). 324. B. Brdar, D. B. Riflcin and E. Reich,JBC 248,2397 (1973). 325. J. L. Darlix, P. Fromageot and E. Reich, Bchem 10, 1525 (1971). 326. R. J. Suhadolnik, T.Uematsu and H. Uernatsu, BBA 149,41 (1967). 327. A. Tavitian, S . C. Uretsky and G. Acs, BBA 157, 33 (1958). 328. A. Tavitian, S. C. Uretsky and G. Acs, BBA 179, 50 (1969). 329a. J. W. Weiss and H. C. Pitot, Cancer Res. 34, 581 (1974). 32%. C. Swart and L. D. Hodge, Virology 84,374 (1978). 330. P. V. Venkov, L. I. Stateva and A. A. Hadjiolov, BBA 474,245 (1977). 331. S. G. Phillips and D. M. Phillips,]. Cell Biol. 49, 785 (1971). 332. U. Heine, Cancer Res. 29, 1875 (1969). 333. B. Brdar and E. Reich, Period. Biol. 78, 51 (1976). 334. J. PBriBs, M. Canivet, M. Olivi6 and M. Tavitian, C.R. Acad. Sci., Ser. D 278,2079 (1974). 335. A. F. Lewis and L. B. Townsend,]. Heterocycl. Chem. 11,71 (1974). 336. K. H. Schrarnrn, S. J. Manning and L. B. Townsend,]. Heterocycl. Chem. 12,1021 (1975). 337. K. H. Schrarnm and L. B. Townsend, Tetrahedron Lett. 14, 1345 (1974). 338. L. B. Townsend and G. H. Milne, Ann. N.Y. Acad. Sci. 255,91 (1975). 339. J. A. Cairns, T. C. Hall, K. B. Olson, C. L. Khuang, J. Horton, J. Colsky and R. K. Shadduck, Cancer Chemother. Rep. 51,197 (1967). 340. R. J. Suhadolnik, T. Uernatsu, H. Uematsu and R. G. WilsonJBC 243,2761 (1968). 342. S. C. Uretsky, G. Acs, E. Reich, M. Mori and L. Altwerger,JBC 243,306 (1968). 342. C. T.Hardesty, N. A. Chaney, V. S. Waravdekar and J. A. R. Mead, Cancer Res. 34, 1005 (1974). 343. M. Hori, E. Ito, T. Takida, G. Koyama, T. Takeuchi and H. Urnezawa,]. Antibiot., Ser. A 17, 96 (1964). 344. S. Aizawa, T. Hidaka, N. Otake, H. Yonehara, K. Isono, N. Igarashi and S . Suzuki, Agric. Biol. Chem. 29, 375 (1965). 345. S. Asano, Y.Kurashina, Y. Anraku and D. Mizuno,]. Biochem. (Tokyo) 70,9 (1971). 346. S. Asano, Y. Anraku and D. Mizuno,]. Biochem. (Tokyo) 70,21 (1971). 347. R. Majirna, K. Tsutsurni, H. Suda and K. Shiniura,J. Biochem. (Tokyo) 82, 1161 (1977). 348. D. C. Ward, A. Cerarni, E. Reich, G. Acs and L. AltwergerJBC 244,3243 (1969). 349. D. C. Ward and E. Reich, Bchem. 61, 1494 (1968). 350. P. Prusiner, T. Brennan and M. Sundaralingam, Bchem. 12, 1196 (1973). 352. K. C. Agarwal and R. E. Parks, Jr., Biochem. Pharmacol. 24,2239 (1975). 352. J. ZernliEka,]ACS 97, 5896 (1975). 353. 0. Makabe, M. Nakamura and S. Umezawa,]. Antibiot. 28, 492 (1975). 354. R. P. Agarwal, G. W. Crabtree, R. E. Parks, Jr., J. A. Nelson, R. Keightley, R. Parkrnan, F. S. Rosen, R. C. Stem and S. H. Polmar.]. Clin. 2nuest. 57,1025 (1976). 355. R. Parkrnan, E. W. Gelfand, F. S. Rosen, A. Sanderson and R. Hirsthhorn, N. Engl. J. Med. 292,714 (1975). 356. F. Skoog, F. M. Strong and C. 0. Miller, Science 148, 532 (1965). 357. F. Skoog and D. J. Armstrong, Annu. Rev. Plant Physiol. 21, 359 (1970). 358. R. C. Gallo, S. M. Hecht, J. Whang-Peng and S. O'Hopp, BBA 281,488 (1972).
282
ROBERT J. SUHADOLNIK
359. S. M. Hecht, R. D. Faulkner and S. D. Hawrelak, PNAS 71,4670 (1974). 360, R. E. Parks, G. W. Crabtree, C. M. Kong, R. P. Agarwal, K. C. Agarwal and E. M. Scholar, Ann. N . Y. Acad. Sci. 255, 412 (1975). 361. E. M. Scholar, P. R. Brown, R. E. Parks, Jr. and P. Calabresi,Blood 41,927 (1973). 362, W. E. G. Muller, H. J. Rohde, R. Steffan, A. Maidhof, M. Lachman, R. K. Zahn and H. Umezawa, Cancer Res. 35,3673 (1974). 363. K. Jain and J. Logothetopoulos, Endocrinology 100, 923 (1977). 364, M. Sussman, S. Alexander, C. Boschwitz, R. Brackenbury, A. Cohen and J. Schindler, ICN-UCLA Symp. Mol. 61 Cell Biol. 2, 89 (1975). 365. A. Cohen and M. Sussman, PNAS 72,4479 (1975). 366. M. R. Sheen, H. F. Martin and R. E. Parks, Jr., Mol. Pharmacol. 6,255 (1970). 367. D. G . Johns, in “Antineoplastic and Immunosuppresive Agents” (A. C. Sartorelli and D. G. Johns, eds.), Part I, p. 277. Springer-Verlag, Berlin and New York, 1974. 368. P. F. Torrence, E. De Clercq, J. A. Waters and B. Witkop, BBRC 62, 658 (1975). 369. M. Ikehara and T. Tezuka, Nucleic Acids Res. 1, 907 (1974). 370. R. A. Long, J. F. Gerster and L. B. TownsendJ. Heterocycl. Chem. 7,863 (1970). 371. G. Koyama, H. Nakamura, H. Umezawa and Y. Iitaka,Acta Crystallogr. Sect. B 32, 813 (1976). 372. R. J. H. Davies, EJB 61, 225 (1976). 373. R. J. H. Davies,JMB 73,317 (1973). 374. K. Sebesta, K. Horski and J. Vaiikovl, Collect. Czech. Chem. Commun. 34, 891 (1969). 375. R. P. M. Bond, C. B. C. Boyce and S. J. French, BJ 114,477 (1969). 376. G. Benz, Experientia 22, 81 (1961). 377. M. Prystas and F. $om, Collect: Czech. Chem. Commun. 36, 1448 (1971). 378. M. Prystas, L. Kalvoda and F. Sonn, Collect. Czech. Chem. Commun. 41, 1426 (1976). 379. K. Sebesta, K. Horski and J. Vaiikovi, Collect. Czech. Chem. Commun. 34, 1786 (1969). 380. E. McConnell and A. G. Richards, Can. J. Microbiol. 5, 161j1959). 381. J. FarkaS, K. Sebesta, K. Horskl, Z. Samek, L. Dolejs and F. Sorm, Collect. Czech. Chem. Commun. 42,909 (1977). 382. L. Kalvoda, M. Prystas and F. Sorm, Tetrahedron Lett. p. 1873 (1973). 383. M. Prystas and F. Sonn, Collect. Czech. Chem. Commun. 36, 1472 (1971). 384. R. P. M. Bond, C. B. C. Boyce, M. H. Rogoff and T. R. Shieh,in “Microbial Control of Insects and Mites” (H. D. Burgess and N. W. Hussey, eds.), p. 275. Academic Press, New York, 1971. 385. J. Valikovi and K. Horska, Acta Entomol. Bohemoslov. 72, 7 (1975). 386. K. Sebesta, K. Horski and J. Valikov6,Abstr. 5th Meet. F E E S , 1968 p. 250 (1968). 387. T. Beebee, A. Korner and R. P. M. Bond, BJ 127,619 (1972). 388. K. Sebesta and K. Horskl, BBA 169,281 (1968). 389. K. Sebesta and K. Horski, BBA 209,357 (1970). 390. V. V. Mackedonski and A. A. Khadzihiolov, Dokl. Bolg. Akad. Nauk 27,1117 (1974). 391. E. A. Smuckler and A. A. Hadjilov, BJ 129, 153 (1972). 392. D. E. Johnson, Can. J . Microbiol., in press (1978). 393. T. J. C. Beebee and R. P. M. Bond, BJ 136,9 (1973). 394. C. B. S. R. Sharma, S. S. V. Prasad, S. B. Pai and S. Sharma, Experientia 32, 1465 (1976). 395. D. G. Grahame-Smith, P. Isaac and D. J. Heal, Nature 253, 58 (1975). 396. H. A. Kupper, W. T. McAllister and E. K. F. Bautz, EJB 38, 581 (1973).
NATURALLY OCCURFUNG NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
283
D. E. Johnson, Nature 260,333 (1976). J. M. Taylor and C. P. Stanners, BBA 155,424 (1968). P. K. Lewin and M. A. Moscarello, Lah. Inuest. 19, 265 (1968). S. G. F. Wilson, W. Heyman and D. Goldthwait, Pediatrics 25, 228 (1962). A. F. Michaels, Jr., H. D. Venters, H. B. Worthen and R. A. Good, Loh. lnuest. 11, 1266 (1962). 402. D. A. Karnofsky and B. D. Clarkson, Annu. Reo. Pharmucol. 3, 357 (1963). 403. E. Kinetec and A. Tirpack, Biochem. Pharmucol. 19, 1493 (1970). 404. R. J. Whitley, S. Soong, R. D o h , G. J. Galasso, L. T. Ch’ien and C. A. Alford, N. Engl. J. Med. 297, 289 (1977). 405. W. Bergmann and R. J. Feeney, JACS 72, 2809 (1950). 406. W. Bergmann and R. J. Feeney,]. Org. Chem. 16,981 (1951). 407. W. Bergmann and M. F. Stempien, Jr., J . Org. Chem. 22, 1575 (1957). 408. W. W. Lee, A. Benitez, L. Goodman and B. R. Baker, JACS 82,2648 (1960). 409. E. J. Reist, A. Benitez, L. Goodman, B. R. Baker and W. E. Lee,]. Org. Chern. 27, 3274 (1962). 410. G. P. Bodey, J. Gottlieb, K. B. McCredie and E. J. Freireich, in “Adenine Arahinoside: An Antiviral Agent” (D. Pavan-Langston, R. A. Buchanan, and C. A. Alford, Jr., eds.), p. 281. Raven, New York, 1975. 411. A. M. Doering, J. Keller and S . S. Cohen, Cancer Res. 26, 2444 (1966). 412. P. J. Ortiz, M. Manduka and S. S. Cohen, Cancer Res. 32, 1512 (1972). 413. J. L. York and G. A. LePage, Cnn. J . Biochem. 44, 19 (1966). 414. J. J . Brink and G. A. LePage, Can. J . Biochem. 43, 1 (1965). 415. J. J. Brink and G. A. LePage, Cancer Res. 24, 312 (1963). 416. G. A. LePage,Adu. Enzyme Regul. 8, 323 (1970). 41 7 . S. S. Cohen and W. Plunkett, Ann. N.Y. Acad. Sci. 255, 269 (1975). 418. W. E. G. Miiller, H. J. Rohde, R. Beyer, A. Maidhof, M. Lachman, H. Taschner and R. K. Zahn, Cancer Res. 35, 2160 (1975). 419. M. M. Anderson and R. J. Suhadolnik, Proc. Znt. Congr. Biochem., 9th, 1973 Abstr. Book 3r10 (1973). 420. L. Sweetman, J. D. Connor, R. Seshamani, M. A. Stuckey, S. Carey and R. Buchanan, in “Adenine Arabinoside: An Antiviral Agent” (D. Pavan-Langston, R. A. Buchanan, and C. A. Alford, Jr., eds.), p. 135. Raven, New York, 1975. 421. W. E. G . Miiller, R. K. Zahn, K. Bittlingmaier and D. Falke,Ann. N.Y. Acnd. Sci. 284,34 (1977). 422. J. J. Furth and S. S. Cohen, Cancer Res. 27, 1528 (1967). 423. J. J. Furth and S. S. Cohen, Cnncer Res. 28, 2061 (1968). 424. J. C. Drach and C. Shipman, Jr., Ann. N.Y. Acad. Sci. 284, 396 (1977). 425. E. C. Moore and S. S. Cohen, JBC 242,2116 (1968). 426. N. R. Cozzarelli,Annu. Reu. Biochem. 46, 641 (1977). 427. W. E. G. Miiller, A. Maidhof, R. K. Zahn and W. M. Shannon, Cancer Res. 37,2282 (1977). 428. W. Plunkett and S. S. Cohen, Cancer Res. 35, 415 (1975). 429. W. E. G. Muller, R. K. Zahn, R. Beyer and D. Falke, Virology 76, 787 (1977). 430. W. W. Nichols, Cnncer Res. 24, 1502 (1964). 431. W. Plunkett and S. S. Cohen,Ann. N.Y. Acad. Sci. 284, 91 (1977). 432. W. Plunkett, L. Lapi, P. J. Ortiz and S. S. Cohen, PNAS 71, 73 (1974). 433. S. S. Cohen, Biochem. Phannacol. 24, 1929 (1975). 434. B. J. Sloan, in “Adenine Arabinoside: An Antiviral Agent” (D. Pavan-Langston, R. A. Buchanan, and C. A. Alford, Jr., eds.), p. 45. Raven Press, New York, 1975. 397. 398. 399. 400. 401.
,
284
ROBERT J. SUHADOLNIK
435. L. B. Allen, J. H. Huffman, R. L. Tolman, G. R. Revankar, L. N. Simon, R. K. Robins and R. W. Sidwell, lntersci. Conf. Antimicrob. Agents Chemother., 14th, 1974, Absb 232 (1974). 436. M. A. Waqar, L. A. Burgoyne and M. R. Atkinson, B] 121,803(1971). 437. W. M.Shannon, L. Westbrook and F. M. Schabel, Jr., Proc. SOC.Exp. Biol. Med. 145,542 (1974). 438. J. G. Cory and R. J. Suhadolnik, Bchem. 4, 1729 (1965). 439. J. J. Brink and G. A. LePage, Cancer Res. 24, 1042 (1964). 440. A. Bloch, M. J. Robins and J. R. McCarthy, Jr.,]. Med. Chem. 10, 908 (1967). 441. J. G . Cory, G. Weinbaum and R. J. Suhadolnik, ABB 118,428 (1967). 442. B. M. Chassy and R. J. Suhadolnik,]BC 242,2655(1967). 443. J. D. Connor, L. Sweetman, S. Carey, M. A. Stuckey and R. Buchanan,Antimicrob. Agents 6 Chemother. 6,630 (1974). 444. R. W. Trewyn and S. J. Kerr, Proc. Am. Assoc. Cancer Res. 18,Absb. 84 (1977). 445. P. M.Schwartz, C. Shipman, Jr. and J. C. Drach,Antimicrob. Agents G Chemother. 10,64 (1976). 446. B. J. Sloan, J. K. Kieltz and F. A. Miller, Ann. N.Y. Acad. Sci. 264,60 (1977). 447. Y. Bryson, J. D. Connor, L. Sweetman, S. Carey, M. A. Stuckey and R. A. Buchanan, Antimicrob. Agents 6 Chemother. 6,98 (1974). 447a. R. I. Glazer, Proc. Am. Assoc. Cancer Res. 19,Abstr. 452 (1978). 447b. T.M.Savarese, G . W. Crabtree and R. E. Parks, Jr., Proc. Am. Assoc. Cancer Res. 19,Abstr 483 (1978). 448. C. E. Cass and T. H. Au-Yeung, Cancer Res. 36, 1486 (1976). 449. P. E. Borondy, T. Chang, E. Maschewske and A. J. Glazko, Ann. N.Y. Acad. Sci. 264,9 (1977). 450. S. Cha, R. P. Agarwal and R. E. Parks, Jr., Biochem. Pharmacol. 24, 2187 (1975). 451. R. P. Agarwal and R. E. Parks, Jr., Biochem. Phannacol., in press (1978). 452. S. Cha, Biochem. Pharmacol. 25,2695(1976). 453. R. P. Agarwal, T. Spector and R. E. Parks, Jr., Biochem. Pharmacol. 26,359(1977). 454. L.Lapi and S. S. Cohen, Biochem. Pharmacol. 26, 71 (1977). 454a. W. Plunkett, L. Alexander, S. Chubb and T. L. Loo,Proc. Am. Assoc. Cancer Res. 19,Abstr. 875 (1978). 455. R. P. Agarwal and R. E. Parks, Jr.. Biochem. Pharmacol. 26,663 (1977). 456. J. Ilan, D. R. Pierce and F. W. Miller, PNAS 74,3386 (1977). 457. J. Ilan, K. Tokuyasu and J. Ilan, Nature 226, 1300 (1970). 458. D. Pavan-Langston, R. A. Buchanan and C. ’A. Alford, Jr., eds., “Adenine Arabinoside: An Antiviral Agent.” Raven, New York, 1975. 459. M. Y. Chu and G. A. Fischer, Biochem. Pharmacol. 11,423(1962). 460. M. P. DeGarilhe and J. DeRudder, Prog. Antimicrob. Anticancer Chemother., Proc. Int. Congr. Chemother., 6th, 1969 Vol. 2,p. 180 (1970). 461. J. DeRudder and M. P. DeGarilhe,Antimicrob. Agents Chemother. p. 578 (1966). 462. J. DeRudder and M. P. DeGarilhe,Proc. Int. Congr. Chemother., 5th, 1967 Vol. 2, p. 29 (1968). 463. H. E. Renis and D. A. Buthala, Ann. N.Y. Acad. Sci. 130,345 (1965). 464. H.E.Renis, G. E. Underwood and J. H. Hunter,Antimicrob. Agents. Chemother. p. 675 (1968). 465. J. Hay, P. A. J. Perera, J. M. Morrison, C.A. Gentry and J. H. Subak-Sharpe, Strategy Viral Genome, Ciba Found. Symp., 1971 p. 335 (1971);A. T. Jamieson, G. A. Gentry and J. H. Subak-Sharpe,]. Gen. Virol. 24,465(1974);A. T. Jamieson and J. H. Subak-Sharpe, ihid. p. 481.
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
285
466. S . Kit, R. A. deTorres and D. R. Dubbs, Cancer Res. 26, 1859 (1966). 467. R. L. Miller, J. P. lltis and F. Rapp,]. Virol. 23, 679 (1977). 468. J. F. Aswell, G. P. Allen, A. T. Jamieson, D. E. Campbell and G. A. Gentry, Antimicrob. Agents G Chemother. 12, 243 (1977). 469. G. A. Gentry, J. F. Aswell, G. P. Allen and D. E. Campbell, Proc. Am. Assoc. Cancer Res. 18, Abstr. 530 (1977). 470. B. Bannister, L. Slechta and A. D. Argoudelis, Proc. Int. Symp. Chem. Nat. Prod., loth, 1976 Abstr. C28 (1976). 471. C. DeBoer, B. Bannister, A. Dietz, C. Lewis and J. E. Gray,Abstr. Annu. Meet. Am. SOC. Microbiol. Abstr. 029 (1976). 472. C. DeBoer and B. Bannister, U.S. Patent 3,907,643 (1975). 473. G. E. Zurenko and C. Lewis, 16th Annu. ICAAC Meet., 1976 Abstr. 420 (1976). 474. G. E. Underwood and S. D. Weed, Antimicrob. Agents Chemother. 11, 765 (1977). 475. K. F. Stern and C. Lewis, 16th Annu. lCAAC Meet., 1976 Abstr. 420 (1976). 476. H. E. Renis, B. A. Court and E. E. Eidson, 16th Annu. lCAAC Meet., 1976 Abstr. 420 (1976). 477. D. W. Stroman, 16th Annu. lCAAC Meet., 1976 Abstr. 423 (1976). 478. J. Cialdella and L. Slechta, 16th Annu. ICAAC Meet., 1976 Abstr. 421 (1976). 479. T. Haneishi, M. Nomura, T. Okazaki, A. Naito, I. Seki, M. Arai, T. Hata and C. Tamura, 174th Sci. Meet,]pn. Antibiot. Res. Assoc., 1970 (1970). 480. Y. Kusakabe, J. Nagatsu, M. Shibuya, 0. Kawaguchi, C. Hirose and S. Shirato,]. Antibiot. 25, 44 (1972). 481. K. Sasaki, Y. Kusakabe and S. Esumi,]. Antibiot. 25, 151 (1972). 482. K. lsono and R. J. Suhadolnik,]. Antibiot. 30, 272 (1977). 483. E. F. Elstner and R. J. Suhadolnik, Bchem. 11, 2578 (1972). 484. K. Ochi, personal communication. 485. T. Haneishi, T. Okazaki, T. Hata, C. Tamura, M. Nomura, A. Naito, I. Seki and M. Arai,]. Antibiot. 24, 797 (1971). 486. Kaken Kagaku Kabushiki Kaisha, German Patent 2,043,946 (1971). 487. S. De Bernard0 and M. WeigeleJ. Org. Chem. 42, 109 (1977). 488. A. D. Argoudelis and S. A. Mizsak,]. Antibiot. 29, 818 (1976). 489. W. E. Cohn,JBC235, 1488 (1960). 490. K. Isono and R. J. Suhadolnik,Ann. N.Y. Acad. Sci. 255,390 (1975). 491. C. DeBoer, B. Bannister, A. Dietz, C. Lewis and J. E. Gray, in preparation. 492. U. Reichman, K. Hirota, C. K. Chu, K. A. Watanabe and J. J. Fox,]. Antibiot. 30, 129 (1977). 493. R. A. Earl and L. B. Townsend,]. HeterocycL Chem. 14, 699 (1977). 494. T. Uematsu and R. J. Suhadolnik, Bchern 11, 4699 (1972). 495. M. S. Ciampi, F. Arena and R. Cortese, FEBS Lett. 77, 75 (1977). 496. R. Cortese, H. O., Kammen, S. J. Spengler and B. N. AmesJBC 249,1103 (1974). 497. I. Tamm, K. Folkers and C. H. Shunk,]. Bact. 72, 59 (1956). 498. A. Kapuler, D. C. Ward, N. Mendelsohn, H. Klett and G. Acs, Virology 37, 701 (1969). 499. L. L. Bennett, Jr., H. P. Schnebli, M. H. Vail, P. W. Allan and J. A. Montgomery, M o l . Pharmacol. 2,432 (1966). 500. T. Niida, T. Niwa, T. Tsuruoka, N. Ezaki, T. Shomura and H. Umezawa, 153rd Sci. Meet.]pn. Antibiot. Res. Assoc., 1967 (1967). 501. T. Sawa, Y. Fukagawa, I. Homma, T. Takeuchi and H. Umezawa,]. Antibiot., Ser. A 20,227 (1967).
286
ROBERT J. SUHADOLNIK
502. T. Tsuruoka, N. Ezaki, S. Amano, C. Uchida and T. Nuda, Meiji Seika Kenkyu Nempo 9, 17 (1967). 503. H. Nakamura, G. Koyama, Y. Iitaka, M. Ohno, N. Yagisawa, S. Kondo, K. Maeda and H. Umezawa,JACS 96,4328 (1974). 504. M. Ohno, N. Yagisawa, S. Shibahara, S. Kondo, K. Maeda and H. UmezawaJACS 96,4326 (1974). 505. P. W. K. Woo, H. W. Dion, S. M. Lange, L. F. Dahl and L. J. DurhamJ. Heterocycl. Chem. 11,641 (1974). 506. H. W. Dion, P. W. K. Woo and A. Ryder, Ann. N.Y. Acad. Sci. 284,21 (1977). 507. A. Ryder, H . W. Dion, P. W. K. Woo and J. D. Howells, U.S. Patent 3,923,785 (1975). 508. B. Evans and R. Wolfenden,JACS 92,4751 (1970). 509. H. J. Schaeffer and C. F. Schwender,]. Med. Chem. 1 7 , 6 (1974). 510. E. C. Herrmann, Jr., Ann. N.Y. A c Q ~Sci. . 284, (1977). 511. M. M. Chassin, R. H. Adamson and D. G. Johns, Proc. Am. Assoc. Cancer Res. 18, Abstr. 586 (1977). 512. W. R. McConnell, Proc. Am. Assoc. Cancer Res. 18, Abstr. 165 (1977). 513. W. J. Suling, L. S. Rice and W. M. Shannon,Proc. Am. Assoc. Cancer Res. 18, Abstr. 167 (1977). 514. A. P. Kimball, S. H. Lee and N. Caron, Proc. Am. Assoc. Cancer Res. 18, Abstr. 499 (1977). 515. R. W. Brockman, L. M. Rose, F. M. Schabel, Jr. and W. R. Laster, Proc. Am. Assoc. Cancer Res. 18, Abstr 192 (1977). 516. W. Plunkett, L. Alexander and Ti Li Loo, Proc. Am. Assoc. Cancer Res. 18, Abstr. 232 (1977). 517. M. M. Chassin, M. A. Chirigas, D. G. Johns and R. H. Adamson, N. Engl. J . Med. 296, 1232 (1977). 518. J. J. Ballet, R. Insel, E. Merler and F. S. Rosen,J. Exp. Med. 143, 1271 (1976). 519. G. C . Mills, F. C. Schmalstieg, K. B. Trimmer, A. S. Goldman and R. M. Goldblum, PNAS 73,2867 (1976). 520. W. P. Roberts, M. E. Tate and A. Kerr, Nature 265, 379 (1977). 521. R. J. Thompson, R. H. Hamilton and C. F. Pootjes, Plant Physiol. 59, Suppl. 110 (1977). 522. J. Heip, G . C. Chatterjee, J. Vandekerckhove, M. Van Montagu and J. Schel1,Arch. Int. Physiol. Biochim. 80, 974 (1975). 523. B. B. Lippincott and J. A. Lippincott,]. Bact. 97,620 (1969). 524. B. A. McCardell and C. F. Pootjes, Antimicrob. Agents 6 Chemother. 10, 498 (1976). 525. W. P. Roberts and A. Kerr, Physiol. Plant Pathol. 4, 81 (1974). 526. N. van Larebeke, C. Genetello, J. Schell, R. A. Schilperoort, A. K. Hermans, J. P. Hernalsteens and M. Van Montagu, Nature 255,742 (1975). 527. I. Zaenen, N. Van Larebeke, H. Teuchy, M. van Montagu and J. Schel1,JMB 86, 109 (1974). 528. N. van Larebeke, G. Engler, M. Holsters, S. van den Elsacker, I. Zaenen, R. A. Schilperoort and J. Schell, Nature 252, 169 (1974). 529. G. Engler, M. Holsters, M. van Montagu, J. Schell, J. P. Hernalsteens and R. Schilperoot, Mol. Gen. Genet. 138,345 (1975). 530. M. D. Chilton, S. K. Farrand, F. Eden, T. C. Currier, A. J. Bendich, M. P. Gordon and E. W. Nester, in “Modification of the Information Content of Plant Cells” (R. Markham et d., eds.), p. 297. Am. Elsevier, New York, 1974.
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
287
531. B. Watson, T. C. Currier, M. P. Gordon, M. D. Shilton and E. W. Nester,]. Bact. 123,255 (1975). 532. T. Iwasa, T. Kishi, K. Matsura and 0. Wakae,]. Antibiot. 30, 1 (1977). 533. S. Harada and T. Kishi,]. Antibiot. 30, 11 (1977). 534. E . Cherbuliez and K. Bernhard, Helu. Chim. Acta 15,464 (1932). 535. G. R. Pettit, R. H. Ode, R. M. Coomes and S. L. Ode, Lloydia 39,363 (1976). 536. M. E. Balis, D. H. Levin, G. B. Brown, G. B. Elion, H. VanderwerEand G. H. Hitchings,JBC 199, 277 (1952). 537. B. A. Lowy, J. Davoll and G. B. Brown,]BC 197, 591 (1952). 538. G. B. Elion, H. VanderwerE, G. H. Hitchings, M. E. Balis, D. H. Levin and G. B. Brown,]BC 200,7 (1953). 539. J. Doskoeil and A. Hal?, Nucleic Acids Res. 1, 645 (1974). 540. M. Huang, H. Shimizu and J. W. Daly,]. Med. Chem. 15,462 (1972). 541. H. M. Mantsch, I. Goia, M. Kezdi, 0. Blrzu, M. Dlngoreanu, G. Jebeleanu and N. G. Ty, Bchem. 14, 5593 (1975). 542. K. Takaoka, 'I. Kuwayama and A. Aoki, Japanese Patent 615,332 (1971). 543. K. Sakata, A. Sakurai and S. Tamura, Agric. Biol. Chem. 37,697 (1973). 544. K. Sakata, A. Sakurai and S. Tamura, Tetrahedron Lett. p. 1533 (1974). 545. K. Sakata, A. Sakurai and S. Tamura, Agric. Biol. Chem. 38, 1883 (1974). 546. K. Sakata, A. Sakurai and S. Tamura, Tetrahedron Lett. p. 4327 (1974). 547. K. Sakata, A. Sakurai and S. Tamura, Agric. Biol. C h e n . 39, 885 (1975). 548. K. Sakata, A. Sakurai and S. Tamura, Tetrahedron Left. p. 3191 (1975). 549. K. Sakata, A. Sakurai and S. Tamura, Agric. Biol. Chem. 40, 1993 (1976). 550. K. Sakata and J. Uzawa, Agric. Biol. Chem. 41, 413 (1977). 551. K. Sakata, A. Sakurai and S. Tamura, Agric. Biol. Chem. 41, 2027 (1977). 552. K. Sakata, A. Sakurai and S. Tarnura, Agric. Biol. Chen. 41,2033 (1977). 553. U. Dihn, J. H. Hagenmaier, H. Hohne, W. A. Konig, G. Wolf and H. Zahner,Arch. Microbiol. 107, 143 (1976). 554. K. Isono, K. Asahi and S. Suzuki,JACS 91, 7490 (1969). 555. T. Hashimoto, M. Kito, T. Takeuchi, M. Hamada, K. Maeda, Y. Okaini and H. Umezawa,]. Antihiot. Ser. A 21, 37 (1968). 556. J . Shoji, R. Sakazaki, M. Mayama, Y. Kawainura and Y. Yasuda,]. Antihiot. Ser. A 23,295 (1970). 557. M. Mizuno, Y. Shimojima, T. Sugawara and I. Takeda,]. Antibiot., Ser. A 24, 896 (1971). 558. T. Honke, M. Tanaka and S. Nakamura,]. Antihiot. 30, 439 (1977). 559. K. Isono, J. Nagatsu, Y. Kawashima and S. Suzuki, Agric. Biol. Chem. 29, 848 ( 1965). 560. K. Isono, J. Nagatsu, K. Kobinata, K. Sasaki and S. Suzuki, Agric. Biol. Chem. 31, 190 (1967). 561. K. lsono and S. Suzuki, Agric. Biol. Chem. 30, 813 (1966). 562. K. Isono, S. Suzuki, M. Tanaka, T. Nanbata and K. Shibuya, Tetrahedron 6, 425 (1970). 563. N. P. Darnodaran, G. H. Jones and J. G. Moffatt,]ACS 93,3812 (1971). 564. H. Kuzuhara, H. Ohrui and S. Emoto, Tetrahedron Lett. p. 5055 (1973). 565. K. Isono, T. Sato, K. Hirasawa, S. Funayama and S. SuzukiJACS 100,3937 (1978). 566. K. Isono and S. Suzuki, 156th Meet. Am. Chem. Soc. Atlantic Cit!], 1968 Abstr. Medi., p. 35 (1968). 567. J. Eguchi, S. Sasaki, N. Ohta, T. Akashiba, T. Tsuchiyama and S. Suzuki, Am. Phytopathol. SOC.34, 280 (1968).
288
ROBERT J. SUHADOLNIK
F. A. Keller and E. Cabib,JBC 246, 160 (1971). E. Cabib and F. A. Keller,JBC 246, 167 (1971). A. Endo and T. Misato, BBRC 37,718 (1969). C. W. Gooday, A. d e Rousset-Hall and D. Hunsley, Trans. Br. Mycol. Soc. 67, 193 (1976). 572. S. Bartnicki-Garcia and E. Lippman,J. Gen. Microbiol. 71,301 (1972). 573. N. Ohta, K. Kakiki and T. Misato, Agric. Bid. Chem. -34, 1224 (1970). 574. E. L6pez-Romero and J. Ruiz-Herrera, Antonie uun Leeuwenhoek 42,261 (1976). 575. M. Hori, K. Kakiki, S . Suzuki and T. Misato,Agric. Biol. Chem. 35, 1280 (1971). 576. M. Hori, K. Kakiki and T. Misato,Agric. Biol. Chem. 38, 691 (1974). 577. M. Hori, K. Kakiki and T.Misato, Agric. Biol. Chem. 38, 691 (1974). 578. M. Hori, J. Eguchi, K. Kakiki and T. Misato,J. Antibiot. 27, 260 (1974). 579. M. Hori, K. Kakiki and T. Misato,J. Antibiot. 28, 237 (1975). 580. M. Hori, K. Kakiki and T. Misato,J. Pestic. Sci. 1, 31 (1970). 581. 0. M. H. de Vries and J. G. H. Wessels,Arch. Microbiol. 102, 209 (1975). 582. P. van der Valk and J. G. H. Wessels, Protoplasma 90, 65 (1976). 583. K. Isono, T. Azuma and S. Suzuki, Chem. Pharm. Bull. 19, 505 (1971). 584. H. Nishimura, M. Moyama, Y. Komatsu, H. Katd, N. Shirnaoka and Y. Tanaka,J. Antibiot., Ser. A 17, 148 (1964). 585. K. R. Darnall, L. B. Townsend and R. K. Robins, PNAS 57,548 (1967). 586. Y. Nakagawa, H. Kand, Y. Tsukuda and H. Koyama, Tetrahedron Lett. p. 4105 (1967). 587. E. F. Elstner, D. M. Carnes, R. J. Suhadolnik, C. P. Krieshman, M. P. Schweizer and R. K. Robins, Bchem. 12,4992 (1973). 588. M. Ozaki, T. Kariya, H. Katd and T. Kimura, Agric. Biol. Chem. 36, 451 (1972). 589. S. Roy-Burman, P. Roy-Burman and D. W. Visser, Cancer Res. 28, 1605 (1968). 590. S. Matsuura, 0. Shiratori and K. Katagiri,J. Antibiot. 17, 234 (1964). 591. H. I. Hadler, B. E. Claybourn and T.P. Tschang, BBRC 31,25 (1968). 592. J. D. Gregory,JACS 77,3922 (1955). 593. J. Leslie, Anal. Biochem. 10, 162 (1965). 594. S . A. Morell, V. E. Ayers, T. J. Greenwalt and P. H o h a n , J B C 239,2696 (1964). 595. N. E. Sharpless and M. Flavin, Bchem. 5,2963 (1966). 596. S. WatanabeJ. Antibiot. 23, 313 (1970). 597. S. Roy-Burman, Y. H. Huang and D. W. Visser, BBRC 42,445 (1971). 598. J. Doskoeil and A. Holj~,Nucleic Acids Res. 1, 491 (1974). 599. Y. Komatsu, Agric. Bid. Chem. 35, 1328 (1971). 600. Y. Komatsu,J.,Antibiot. 24, 876 (1971). 601. S. Roy-Burman and D. W. Visser, BBA 282, 383 (1972). 602. Y. Komatsu and K. Tanaka, BBA 288,390 (1972). 603. S. Roy-Burman and D. W. Visser,JBC 250, 9270 (1975). 604. K.-K. Leung and D. W. Visser,JBC 252, 2492 (1977). 605. K. Komatsu and K. Tanaka, BBA 311,496 (1973). 606. P. J. von Dippe, S. Roy-Burman and D. W. Visser, BBA 318, 105 (1973). 607. T. I. Kalman, Bchem 10,2567 (1971). 608. T. I. Kalman, BBRC 49, 1007 (1972). 609. T. I. Kalman, Abstr. 168th Meet., Am. Chem. Soc. Biol. No. 128 (1974). 610. Y. Titani and Y. Katsube, BBA 192,367 (1969). 611. H. I. Hadler, B. G. Daniel and R. D. Pratt,]. Antibiot. 24,405 (1971). 612. H. I. Hadler, Medikon 3 , 2 2 (1974). 613. H. I. Hadler and J. M. Demetriou,J. Antibiot. 28, 809 (1975). 568. 569. 570. 571.
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
289
614. H. I. Hadler, B. G. Daniel, J . Demetriou and R. C. Pratt,]. Antibiot. 24,835 (1971). 615. L. D. Boeck, G. M. Clem, M. M. Wilson and J. E. Westhead, Ferment. Antimicrol. Agents Chemother. 3, 49 (1973). 616. R. L. Hamill and M. M. Hoehn,]. Antibiot. 26,463 (1973); R. S. Gordee and T. F. Butler, ibtd. p. 466. 617. R. L. Hamill, C. B. Carrel], S. M. Nash and R. Nagarajan,17thAnnu. lCAAC Meet. Abstr. No. 48 (1977). 618. R. Nagarajan, B. Chao, D. E. Dorman, S. M. Nash, J. L. Occolowitz and A. Schabel, 17th Annu. ICAAC Meet. Abstr. No. 50 (1977). 618n. C. Pugh, R. T. Borchardt and H. 0. Stone,JBC 253,4075 (1978). 619. H. Kneifel, W. A. Konig, G. Wolf and H. Zihner,]. Antibiot. 27, 20 (1974). 620. K. Eckhardt, H. Thrum, G. Bradler, E. Tonew and M. Tonew,]. Antibiot. 28,274 (1975). 621. A. Takatsuki, K. Arima and G. Tamura,]. Antibiot. 24,215 (1971). 622. T. Ito, Y. Kodama, K. Kawamura, K. Suzuki, A. Takatsuki and G. Tamura, Agric. Biol. Chem. 41, 2303 (1977). 623. A. Takatsuki, K. Kawamura, M. Okina, Y. Kodama, T. Ito and G. Tamura, Agric. B i d . Chem. 41, 2307 (1977). 624. P. A. Lambert, I. C. Hancock and J. Baddiley, BBA 472, l(1977). 625. J. S. Tkacz, S.-C. Kuo and J . 0. Lampen, in “Alcohol, Industry and Research” (0. Forsander et al., eds.), p. 197. Research Laboratories of the Finnish State Alcohol Monopoly, Helsinki, 1977. 626. S.-C. Kuo and J. 0. Lampen,]. Bact. 111,419 (1972). 627. D. K. Struck and W. J. Lennan,JBC 252, 1007 (1977). 628. J. J . Lucas, C. J. Waechter and W. J. LennarqJBC 250, 1992 (1975). 629. W. W. Chen and W. J. Lennarz,]BC 251, 7802 (1976). 630. R. D. Palmiter, T. Oka and R. T. Schimke,/BC 250, 724 (1971). 631. M. L. Keiley, G. S. McKnight and R. T. Schimke,]BC 251,5490 (1976). 632. A. Takatsuki, Y. Fukui and G. Tamura, Agric. Biol. Chem. 41, 425 (1977). 633. J. S. Tkacz and J. 0. Lampen, BBRC 65,248 (1975). 634. S.-C. Kuo and J. 0. Lampen, BBRC 58,287 (1974). 635. A. Takatsuki and G. Tamura, j . Antibiot. 25,362 (1972). 636. S . 4 . Kuo and J. 0. Lampen, ABB 172, 574 (1976). 637. K. Olden, R. M. Pratt and K. M. Yamada,J. Cell Biol. 75, Abstr. CJ885 (1977). 638. A. Takatsuki and G. Tamura,]. Antibiot. 24, 785 (1971). 639. R. Leavitt, S . Schlesinger and S. Kornfeld,]. Virol. 21, 375 (1977). 640. D. Duksin and P. Bornstein, PNAS 74, 3433 (1977). 641. R. Bracha and L. Glaser, BBRC 72, 1091 (1976). 642. A. Hasilik and W. Tanner, Antimicrob. Agents G Chemother. 10, 402 (1976). 643. G. Tamura, T. Saski, M. Matsuhashi, A. Takatsuki and M. Yamasaki, Agric. Biol. Chem. 40, 447 (1976). 644. J. B. Ward, F E B S Lett. 78, 151 (1977). 645. G. E. Bettinger, A. N. Chatterjee and F. E. Young,JBC 252,4118 (1977). 646. G . E. Bettinger and F. E. Young, in “Microbiology” (D. Schlessinger, ed.), p. 69. Am. SOC.Microbiol., New York, 1977. 647. M. C . Ericson, J. T. Gafford and A. D. Elbein,JBC 252, 7431 (1977). 648. M. L. Tanzer, F. N. Rowland, L. W. Murray and J. Kaplan, BBA 500, 187 (1977). 648a. Y. Kawakami, S. Matsuwaka, T. Otani, H. Kondo and S. Nakamura,]. Antibiot. (Tokyo) 31, 112 (1978). 649. H. Yuntsen, K. Ohkuma, Y. Ishii and H. Yonehara,]. Antibiot., Ser. A 9, 195 (1956).
290
ROBERT J. SUHADOLNIK
650. T. E. Elbe, H. Hoeksema, G. A. Boyack and G. M. Savage, Antibiot. Chemother. (Washington, D.C.) 9,419 (1959). 651. L. Ebringer, Experientia 27,586 (1971). 652. E. Freese, in “Spore Research 1976” (A. N. Barker, G. W. Gould, and J. Wolf, eds.), p. 1. Academic Press, New York, 1977. 653. T. Mitani, J. E. Heinze and E. Freese, BBRC 77, 1118 (1977). 654. E. J. Prisbe, J. Smejkal, J. P. H. Verheyden and J. G. Moffatt,/. Org. Chem. 41, 1836 (1976). 655. J. P. H. Verheyden and J . G . MoffattJACS 88, 5684 (1966). 656. H. Hiebabeckf and J. FarkaB, Collect. Czech. Chem. Commun. 39,2115 (1974). 657. H. Hiebabeckf and J. FarkaS, Collect. Czech. Chem. Commun. 39, 1098 (1974). 658. V. Skarii. and J. Matulii., Croat. Chem. Acta 47, 159 (1975). 659. L. M. Lerner,J. Org. Chem. 37,4386 (1972). 660. L. M. Lerner,J. Org. Chem. 37,473 and 477 (1972). 661. L. M. Lerner, Carbohydr. Res. 44, 13 (1975). 662. N. Suciu and L. M. Lerner, Carbohydr. Res. 44, 112 (1975). 663. L. M. Lerner,J. Org. Chem. 41,306 (1976). 664. J. F. Henderson, R. E. A. Gadd, H. M. Palser and M. Hori, Can.J.Biochem. 48,573 (1970). 665. K. Gerzon, R. H. Williams, M. Hoehn, M. Gorman and D. C. DeLong, Proc. Znt. Congr. Heterocycl. Chem., 2nd, 1969, Abstr. (3-30 (1969). 666. R. H. Williams and M. M. Hoehn, U.S. Patent 3,802,999 (1974). 667. J. Farkas, Z. Flegelova and F. Sorm, Tetrahedron Lett. p. 2279 (1972). 668. S. DeBernardo and M. Weigele,J. Org. Chern. 41, 287 (1976). 669. F. Streightoff, J. D. Nelson, J. C. Cline, K. Gerzon, R. H. Williams and D. C. DeLong, 9th Annu. lCAAC Meet. p. 18 (1969). 670. G . E. Gutowski, M. J. Sweeney, D. C. DeLong, R. L. Hamill, K. Gerzon and R. W. Dyke,Ann. N.Y. Acad. Sci. 255,544 (1975). 671. W. M. Shannon, Ann. N.Y. Acad. Sci. 284,472 (1977). 672. M. J. Sweeney, F. A. Davis, G. E. Gutowski, R. L. Hamill, D. H. Hoffman and G. A. Poore, Cancer Res. 33, 2619 (1973). 672a. P. G. W. Plagemann and M. Behrens, Cancer Res. 36,3807 (1976). 673. T. Ohnhma and J. F. Holland, Cancer Treat. Rep. 61, 389 (1977). 674. R. L. Nelson, R. W. Dyke, R. E. Crabtree and M. Zahir-Zafarzai, Proc. Am. Assoc. Cancer Res. 18, Abstr. 209 (1977). 675. T. Ohnuma, J. Roboz, M. L. Shapiro and J. F. Holland, Cancer Res. 37,2043 (1977). 676. A. Jakubowski, C. Lehman, J. Moyer and R. E. Handschumacher, Proc. Am. Assoc. Cancer Res. 18, Abstr. 865 (1977). 676a. J. F. Worzalla and M. J. Sweeney, Proc. Am. Assoc. Cancer Res. 19, Abstr. 224 (1978). 677. J.Descamps and E. De Clercq, in “Current Chemotherapy,” p. 354. Am. SOC. Microbiol., Washington D.C., 1978. 678. R. W. Sidwell, J. H. Huffman, G. P. Khare, L. B. Allen, J. T. Witkowski and R. K. Robins, Science 177, 705, (1972). 679. E. De Clercq, M. Luczak, J. C. Reepmeyer, K. L. Kirk and L. A. Cohen, Life Sci. 17, 187 (1975). 680. K. Konno, K, Hayano, H. Shirahama, H. Saito and T. Matsumoto, Tetrahedron Lett. p. 481 (1977). 681. S. Tono-Oka, A. Sasaki, H. Shirahama, T. Matsumoto and S. Kakimoto, Chem. Lett. p. 1449 (1972).
NATURALLY OCCURRING NUCLEOSIDE AND NUCLEOTIDE ANTIBIOTICS
291
682. T. Ushizawa, N. Katagiri, T. Kato and N. Taira, M e d . B i d . Tok!/o 94, 251 (1977). 683. S. Tono-Oka, A. Sasaki and S. Kakimoto,3fjth Anttu. Meet. Chern. Soc. J p n . Abstr. 11, p. 684 (1976). 684. K. Isono, P. F. Crain and J. A. McCloskey,JACS 97,943 (1975). 68fj. T. Azurna, K. Isono, P. F. Crain and J. A. McCloskey, Tetrdhedron Lett. p. 1687 ( 1976). 686. A. Bloch, BBRC 64,210 (1975). 687. T. Hayashi, K. Shirahata and I. Matsubara, Ahstr. Pap., Jpn. Chem. Soc. Meet., 1973 Abstr. 4L44 (1973). 688. K. Kitahara, B. Kraska, Y. Sanemitsu and F. W. Lichtenthaler, Nucleic Acids Res., Spec. Puhl. 1, SZl(l976). 689. G. G. Deleuze, J. D. McChesney and J. E. Fox, BBRC 48, 1426 (1972). 690. C. W. Parker, D. S. Letham, D. E. Cowley and J. K. MacLeod, BBRC 49, 460 ( 1972). 691. C. C. Duke, A. J. Liepa, J. K. MacLeod, D. S. Letharn and C. W. ParkerJCS, Chem. Commun. p. 964 (1975). 692. M. Arai, T. Haneishi, N. Kitahara, R. Enokita, K. Kawakubo and Y. Kondo, J . Antihiot. 29, 863 (1976). 693. T. Haneishi, A. Terahara, H. Kayamori, J. Yabe and M. Arai,J. Antihiot. 29, 870 (1976). 694. G . E. Bettinger and F. E. Young, B B R C 67, 16 (1975).
This Page Intentionally Left Blank
Genetically Controlled Variation in the Shapes of Enzymes
I
GEORGEJOHNSON Department of Biology Washington University St. Louis, Missouri
I. Introduction ..................................................... 11. Internal Standardization .......................................... 111. The Interaction of Protein Size and Charge during Electrophoresis ........................................... A. Criteria for Adequate Internal Standards ....................... B. Empirical Estimation of KR and M o ............................ C. Variation in Shape of a-Glycerophosphate Dehydrogenase ...... IV. Detecting “Hidden” Variation in Shape ........................... V. The Nature of Cryptic Variants ................................... Genetic Basis of Variation in Shape ............................... VI. Major Unresolved Issues: Posttranslational Modification ............ VII. Novel Approaches to Electrophoresis ............................. References .....................................................
293 294 295 297 299
302 303 305 311 315 321 325
1. Introduction Electrophoretic alleles of enzyme loci are typically detected and characterized by comparing their mobilities on starch or acrylamide gels. This has provided a particularly straightfonvard and convenient approach to detect genetic polymorphism, as mobility classes appear to be clearly discontinuous in most systems. In discussing electrophoretically detected enzyme polymorphism of this sort, differences between alleles are commonly ascribed to amino-acid substitutions involving charged residues. The observed discontinuities in gel mobility would be an integral property of such “charge state” variation. From a theoretical point of view, such a model is too simple. In ideal electrophoresis, a protein migrates in a field at a rate (the free electrophoretic mobility) determined by its net charge. However, gel electrophoresis does not conform to ideal conditions, as the gel and protein interact during the course of the protein’s migration, Whether one visualizes the interactions as frictional (the protein “bumps into” fibers) or hydrodynamic (sheer forces are generated by the protein’s movement past fibers), the shape and size of proteins should also affect their migration rates on gels. 293 Progress in Nucleic Acid Research mid Molecular Biology, Vnl. 22
Copyright @ 1979 b y Academic Press, Inc. All rights of‘ieproductioiiin any fnrni resewed. ISBN U12540022-5
294
GEORGE JOHNSON
Is allelic variation in the shape of proteins detectable by gel electrophoresis? Such variation might well be difficult to detect by electrophoresis if the changes in shape are subtle or have only minor effects on mobility. Investigation of small differences in mobility requires careful standardization in order to ensure gel-to-gel reproducibility. When small differences in mobility are seen, it is very important to know how much of the difference is attributable to experimental error.
II. Internal Standardization Approaches involving two internal standards have been described for both starch gels (1) and polyacrylamide gels (2). Particularly in starch gels, the correction has proved to be important. The two internal marker proteins are added to each biological preparation prior to the electrophoretic run and thus travel in the same path through the gel during the course of the run. These markers, usually commercially obtained proteins, are chosen on the basis of differences in physical characteristics; any change in experimental conditions is intended to affect the migration of one more than that of the other. The ratio of the mobilities of these two markers then provides sensitive standardization of the run, as any experimental error is reflected in an alteration in the standard ratio. Once it has been established that the standard ratio is within the range of values normally obtained, and thus that operating conditions for a particular experiment were normal, the mobility of a “variant” protein may be expressed relative to that of one of the internal standards. For many polymorphic variant proteins under study in our laboratory, the standardized numbers obtained have been highly reproducible by a variety of investigators over periods of years. The general criteria for choice of internal standards are easily stated: 1. To avoid confusion, a standard should have no activity in any contemplated assay unless the system is carefully characterized beforehand, and should not migrate with the same Rf as a band of experimental interest. 2. To avoid artifacts, a standard should not bind proteins in a generalized manner, should have a demonstrably constant mobility, and should not be used in concentrations so excessive as to affect the mobility of other migrating species. 3. To reflect changes in experimental factors that might cause variation in the observed R , of experimental bands, “internal standard” proteins should exhibit both high sensitivity and very different responses to experimental variables affecting electrophoretic mobility. Activity of standard proteins in enzyme assays, and similarity in mobilities of standards to experimental bands, are experimental mat-
295
VARIATION IN THE SHAPES OF ENZYMES
ters easily dealt with. Aggregation, generalized binding, and excessive concentration artifacts may readily be detected b y control experiments in which the proposed internal standard is omitted, or its concentration varied. Constancy of mobility, especially from lot to lot, should not be assumed, but is readily verifiable. The third criterion, which concerns the physicochemical characteristics of the standards, is not as simple to apply, although clearly of great experimental importance. In order adequately to formulate such a criterion, it is first necessary to examine in detail the differing contributions of protein charge and protein size to observed mobility.
111. The Interaction of Protein Size and Charge during Electrophoresis
The theoretical behavior of charged molecules migrating in an electric field has been thoroughly studied by physical chemists, and the essential elements can be stated quite simply. When the processes of translational diffusion and of electromigration are both taken into account, then, for a molecule of concentration C passing through a gel slice of thickness x over a time interval t , the observed band mobility may be expressed as dcldt
= D ( a z C / K 2)
MoE(dc/i3~)
( 1)
This is the fundamental equation for ideal electrophoresis (3).What it says is that, observing the movement of a band past a fixed point on a gel, the rate at which the band passes (expressed as dC/at)is a function of D (the diffusion constant), M o (the electrophoretic mobility of the protein), and E (the electric field strength). The diffusion constant, D, has provided a major focus of interest in physical chemistry and is usually expressed in terms of a complex equation involving net charge, molecular radius, counterion radius, viscosity, and a variety of constants. What is important here is that D may be expressed entirely in terms of empirical parameters of macromolecule, gel, and buffer. The fundamental assumption of electrophoretic analysis is that if one holds E constant, and i f D has been held constunt, then the observed mobility of the protein band is solely a function of the free electrophoretic mobility of the protein, M o . Within the past decade, physical chemists have begun to examine the theoretical behavior of proteins migrating in nonideal situations. Much of this work has been reviewed recently by Chrambach and Rodbard ( 4 ) . In its simplest forms, this work analyzes the effects of varying some of the component elements ofD in a controlled way. One important component of the diffusion constant results from interaction
296
GEORGE JOHNSON
of migrating proteins with molecules of the supporting medium. By assuming that the migration rate is retarded by frictional interactions with polyacrylamide gel fibers, Chrambach and Rodbard were able to write a simple equation expressing the migration of a protein species in polyacrylamide gel electrophoresis (5, 6).
where Rf= mobility of protein relative to the front; uf = apparent mobility of moving boundary in front of the resolving phase (a constant known for most common buffer systems); M , = free electrophoretic mobility of protein; K , = the retardation (frictional) coefficient( K R = K r x 2.303);and %T = % acrylamide (which determines pore size and is inversely proportional to it). The important aspect of the equation is that it completely describes electrophoretic behavior in terms of two measurable variables: the one the protein’s charge contribution and the other the protein’s conformational (frictional) contribution. These two parameters K , and M o provide a complete description of the contributions of the macromolecule to its migration rate. M o is the free eIectrophoretic mobility that the protein would have in an ideal system with infinitely large gel pore size and thus no gel friction to restrain migration; it thus reflects net charge independent of any size or shape effects. In the electrophoresis theory of Rodbard and Chrambach, the retardation coefficientK, is defined as a function of thefractional volume available to the protein as it passes through the gel. It is thus in a sense a frictional coefficient that arises due to nonideal behavior (the macromolecule is not a point in space and therefore hits the supporting medium). There are two components in the interaction. The first, the contribution of the gel, depends upon the concentration of acrylamide and upon the way in which gel pore size changes as acrylamide concentration changes. The second component, the contribution of the molecule, is simply a function of the area of interaction, and thus of the protein’s radius and that of the fiber with which it is interacting. Thus the retardation coefficient may be expressed as
K,
=n
(E + T ) ~JALJA , %T)
(3)
molecular gel contribution contribution
where K r = retardation coefficient; A = geometrical mean radius of the macromolecule; r = acrylamide fiber radius; (ALIA%T) = sensitivity of pore size to changes in acrylamide concentration.
VARIATION IN THE SHAPES OF ENZYMES
297
The form of this equation is of particular importance. The sole contribution of the protein to the frictional interaction is expressed by the geometrical mean radius, R, and K , is a polynomial function of R. It is possible to visualize alternative models in which K ,depends upon hydrodynamic factors rather than steric ones. In such models, K , would be a function of the Stokes radius (R,) rather than R . In either case, even small changes in protein size or conformation can be expected to produce significant changes in K,. In my work, the absolute concentration of the cross-linking agent, bisacrylamide, is kept constant as the concentration of acrylamide monomer, T, is varied from 3%to 12% (g/lOOml). This has the effect of systematically reducing the dimensions of the average gel pore (hLl A%T) as the value of T increases. As can be seen in Fig. la, high resolution can be obtained by this procedure. It is important to note that this contrasts to other procedures wherein the relative concentration of the bis crosslinker is held constant (7).The latter procedure has the effect of maintaining a constant pore size dimension ( A L / A % T )as the value of T increases, and results in greatly reduced resolution (Fig. lb). It is important to realize that the experimental factors contributing to errors in determining K R (KR = 2.303 x K,) and Mo are quite different. M o is a function of D and E : E will vary as the current varies; D will vary as a power function of temperature; D also varies as Q varies, and therefore as a function of pH. However, M o is not a function of any of the gel characteristics. The retardation coefficient is sensitive to quite a different set of experimental factors. K R is a function of %T and of the macromolecule’s size and shape. As ( h L I A % T )is relatively constant, and %T is a controlled variable, K R is sensitive only to characteristic of the protein under investigation and should in other respects be relatively insensitive to experimental conditions (although the protein’s conformation may itself be sensitive to changes in pH, temperature, ionic strength, etc.).
A. Criteria for Adequate Internal Standards When it is possible to estimate the differing contributions of protein charge and conformation to observed electrophoretic mobility, one can state simple criteria that ensure that two internal standard proteins will be maximally sensitive to experimental error. 1. To detect variation in gel pore size, use two internal standard proteins with very different K R values. This will ensure that any change in %T will affect the mobility of one protein far more than that of the other and thus significantly alter the mobility ratio of the two internal standards. Optimally the K R values of the two internal stan-
298
GEORGE JOHNSON
I
XANTHINI I
1
1 6
%
DlHYDROOENASt I
I
I
I
7
8
9
10
acrylamidc (TI
XANTHINI 5
b
DIHYDRO0tNASI 7
9
10
I acrylamida(T)
FIG.1. Gel-sieving electrophoresis of modified and wild-type xan. ..ine dehydrogenase (after 6a). (a) Analyses in which the dimensions of the gel pore are varied by maintaining a constant absolute concentration of bisacrylamide crosslinking agent, (0.184 g/100 ml). Relative to the concentration of acrylamide monomer, T, the level ofC (crosslinker) thus varies from 3.5% of total acrylamide for 5% T gels to 2% of total acrylamide for 10% T gels. As the crosslinking agent is the long fiber in the polyacrylamide pore, 10% T gels constructed in this manner have significantly smaller pore dimensions than 5% T gels. (b) Analyses in which the dimensions of the gel pore are held constant by maintaining a constant relative concentration of C. Relative to the concentration of acrylamide monomer, T, the level of C is maintained at 5% of total acrylamide (equivalent to commercial “cyanogum”). On the ordinate is the R f of xanthine dehydrogenase relative to that of the hemoglobin internal standard.
dards should bracket those of the experimental proteins of interest. 2. To detect variation in pH, select one of the two internal standards to be near its isoelectric point in the buffer employed. This will result in a steep pH-vs.-mobility curve at the operative pH, and thus maximal sensitivity to changes in pH. This usually means a standard molecule with a low M,. (Care must be taken in polyacrylamide gel
VARlATION IN THE SHAPES OF ENZYMES
299
stacking systems that the protein’s mobility not be so low as to be less than that of the counterion in the upper gel, or the molecule will not “stack,” and thus wiIl not produce the desired sharp band). 3. To detect variation in current, or in any of the experimental factors (particularly temperature) that affect D, select two internal standard proteins with widely divergent M,, values. Given the second criterion above, this means that the other internal standard protein should have a high M o value. If practical, the M o values of the two internal standards should bracket those of proteins of experimental interest. This will ensure that the greatest differential change in migration in response to an error in, e.g., temperature, will be between the two internal standard proteins, causing a maximal deflection of their mobility ratio from normal values.
B. Empirical Estimation of K R a n d Mo I have estimated the parameters K R and M o using disc gels similar to those originally described by Davis (8).Separate cylindrical gels are formed, each with a different acrylamide concentration. I have worked in the range of 3 to 12% acrylamide (%T),with the crosslinking agent bisacrylamide (%C) 5% of the total acrylamide at 7% T. For most enzymes I have examined, an experiment using six separately prepared gels with %T = 4,5,6,7,8, and 9 has been adequate to produce a statistically satisfactory result. In initial experiments with this approach, it is often advisable to run several replicates at each point, but familiarity with the technique soon renders this less necessary. For each gel, protein and tracking dye (bromophenol blue) are layered on top of the stacking gel in a buffered solution of 10% sucrose. Runs are conducted at controlled temperature for approximately 1-2 hours at 150-200 V. At the end of every run the position of the dye front is marked and each gel is assayed for enzyme activity or stained for protein [fixed in 12.5% trichloroacetic acid for 30 minutes and stained with 0.05% Coomassie Blue in 12.5% acid for 1 hour] and then is scanned spectrophotometrically at 540 nm ‘for blue-staining bands. The brown bands of hemoglobin and ferritin may be scanned without staining at 410 nm, although care must be taken not to use excessive concentrations of these standard proteins. After scanning, the distance to each peak from the interface of the running gel is measured, and the ratio of the distance traversed by any band to that of the tracking d y e is calculated. Thus for each protein run in a given gel, an Rfvalue that can be associated with a particular acrylamide concentration is obtained. The utility of Chrambach and Rodbard’s original formulation [Eq. (2)] can now be seen. The effect of taking the logarithm of both sides of
300
GEORGE JOHNSON
the equation is to produce a new equation of the linear form y = mx b, where rn is the slope and b the y intercept:
-
y axis
A
Y intercept
+ slope
+
+
x axis
One thus plots the log of each &value vs. the corresponding acrylamide concentration. A linear function is obtained with a slope equal toKR,and the anti-log of the y intercept equal to Mo divided by a buffer constant. These slopes and intercepts are determined by standard linear regression analysis. Rodbard (9) points out that the error variance of log Rfis not constant. Thus, while an unweighted regression does provide good estimates of K R and Yo, it may give poor estimates of their standard errors when R, is small (less than 0.1). As the Rf approaches zero, the logarithm of Rfbecomes infinite, and infinitesimal changes in the Rf will result in astronomical errors in the logarithm of Rf.In my own work, Rf values less than 0.1 were rarely encountered. Following the suggestion of Gonenne and Lebowitz (lo), I tested for variation in the error variance by plotting the sums of the squares of the deviation of log Rf versus Rf.I found no significant heterogeneity of variance. Thus, in my work, weighted regressions have been of minimal importance, as the reproducibility of the individual points has been very good. Assuming little conformational variability, K R ought to be directly relatable to molecular weight. The matter has been discussed previously (9-11 ). Examining commercially obtained molecules over a broad range of molecular weights, I obtain essentially the same relation as these prior workers (Fig. 2), despite slight differences in buffer systems.
1; 0.20
0.4-
0.3-
0.08
0.2I
20
40
60
(MWfi
80
100
FIG.2. Retardation coefficient ( K R )as a function of molecular weight.
301
VARIATION IN THE SHAPES OF ENZYMES
One may standardize determinations ofKRand M o for an enzyme of an individual by expressing the values relative to the corresponding value of the (similar) internal standard determined from the same gels. Because the standard is evaluated many hundreds of times, the procedure permits standardization of any error affecting both proteins. The procedure has been described in detail (12).For each individual gel, the mobility of the enzyme of interest is expressed relative to that of the internal standard. The appropriate regression equation is thus log (RFZ/Rltd)= log (Mgenz/MgStd) +
- KRad)T
(5) In accordance with Eq. (l),the error in M , is taken as multiplicative, whereas the error in K R is taken as additive. One may use the known values of KRSa and Mostddetermined from many experiments to estimate standardized values for experimental bands. In comparing estimates of K Rand of M o , it is important to note that KR and M o are determined from a single linear regression as slope and intercept and that error in the estimation of an intercept is not independent of error in the estimation of the corresponding slope. One simple approach to estimating the error associated with a mobility estimate independent of that associated with a corresponding K R estimate is to express mobility in terms of the Rf observed at the mean value of T . A linear regression may be considered to rotate around such a midpoint as its slope varies, the midpoint remaining unchanged despite great changes in the Y intercept. The error in the estimate of this parameter is independent of the error in the estimate of the slope (KRenz
KR.
To compare two proteins, one may plot the 95% confidence interval ellipses of the K R mid-Y pairs. If, on a plot of mid-Y vs. K R , the boundary of one ellipse does not overlap the center of the other, the proteins may be taken as not identical. When sampling from a natural population, there will be an error variance in K Rand in mid-Y associated with each protein type present in the sample. To document the existence of multiple classes requires an independent estimate of experimental error. This estimate may be readily obtained from the internal standards run in the same gels. In practice, the behavior of both experimental proteins and the similar internal standard are normalized to that of a third internal standard chosen to closely resembling them, and the error in the estimate of the standard is taken as a measure of experimental error. In plotting K R , mid-Y estimates from a natural population, points reffecting homologous proteins should have a distribution no greater than that seen for the standard. A significantly greater distribution is evidence of heterogeneity. Distributions may be compared in terms of coefficients of varia-
302
GEORGE JOHNSON
tion [(standard deviatiodmean) x 1001 associated with mid-Y and K values.
R
C. Variation in Shape of a-Glycerophosphate Dehydrogenase When gel-sieving analysis such as described above was carried out on individuals sampled from natural populations of butterflies, it was immediately apparent that the variation in mobility observed previously reflected more than simple charge differences. If the only source of difference is net charge (presumably produced by amino acid substitutions involving charged residues), then one would expect variants to have similar retardation coefficients, K R , and to'differ primarily in free electrophoretic mobility, M o . The range of variability in their K R values would be expected to be limited to about that observed for the hemoglobin internal standard. In fact, the distribution of K R values obtained is very much broader than the corresponding hemoglobin distribution, as seen in Fig. 3. Noncharge differences clearly contribute to the differences seen in mobility on 7% acrylamide gels. This proved to be an important result: Charge and size/ conformation interact in determining mobility, and these two protein properties vary concordantly-as M o values increase, so do the absolute values of K R (e.g., bigger or more asymmetric proteins have greater net charge). The result is that the mobility functions described by Eq. (4)intersect at intermediate gel pore sizes. The nature of the mobility variation at the a-glycerophosphate dehydrogenase (aGroPdH) locus is now clear (Fig. 4):A survey conducted at 5% acrylamide (equivalent to 10-11%starch) will not discriminate between variants, and will reveal a single uniform mobility type. Such a survey would
QGroP-dH
AA 0.05
0.06
0.05
0.06
0.01
Kr
FIG.3. Retardation coefficients ( K R )of a-glycerophosphate dehydrogenase in Colias compared to hemoglobin standards run in the same gels (from 26).
303
VARIATION I N THE SHAPES OF ENZYMES
2
4
6
8
1
0
%1
FIG.4. Gel sieving analysis of the five a-glycerophosphate dehydrogenase alleles in Colius (from 26).
classify this locus as uniformly homozygous. A survey conducted at 7% acrylamide, as were my previous surveys, would report two segregating alleles (see, for instance, 30,33) with considerable variation in the exact mobility observed. In fact there are a minimum ofjive alleles segregating at this locus. Gel-sieving analysis thus provides direct evidence of protein heterogeneity within electrophoretic classes. Note also that there are alleles that do not differ in net charge, differing only in K R .
IV. Detecting "Hidden ' Variation in Shape Heritable electrophoretic variants of enzymes have been typically detected and characterized by comparing their mobilities on starch or acrylamide gels. Over the last decade, this has been a straightforward and convenient approach to detect enzyme polymorphism, since triobility classes appear to be clearly discontinuous in a wide range of mammalian, insect, and plant systems (13-15). However, within the past few years a variety of studies concerned with electrophoretically detectable enzyme variation in insects, such as the one described above for aGroP-dH, have demonstrated the presence of additional variant classes not normally resolved by standard electrophoretic techniques. Early evidence of hidden variation came from studies of thermal lability. At the esterase-6 locus of Drosophila melanogaster, com-
304
GEORGE JOHNSON
parisons of the heat stability of electrophoretic variants suggested the existence of alleles not distinguishable by electrophoresis (16). More recently, when individuals of a presumptively homogeneous line of octanol dehydrogenase in Drosophila pseudoobscura were analyzed electrophoretically and the gels were incubated at high temperature, some individuals displayed a more stable enzyme than others (17); subsequent studies of the segregation patterns among F, progeny sugSince gested that such variation in thermal lability is heritable (18,19). these reports, other studies confirmed the existence of thermal lability variants in Drosophila: phosphoglucomutase (PGlc-M) (20), alcohol dehydrogenase (Alc-dH) (21),esterase (22).Particularly careful analyses of thermal variants of Alc-dH from D.m e h o g a s t e r (23,24) confirmed in these instances that the heat-labile variants map to the same genetic locus as the Alc-dH structural gene. The discovery of extensive variation in protein shape among aGroP-dH alleles raises the question whether this enzyme locus is typical in this respect. Certain loci, particularly esterases, typically exhibit far more electrophoretically detectable variationkhan others. Is the newly detected variation concentrated among a particular subset of enzyme loci? To address this question, gel-sieving analyses were carried out on 14 enzyme loci of the alpine butterfly Colias meadii (25). For all loci but malate dehydrogenase (Mal-dH), several common variants that cannot be distinguished from one another on routine 7% acrylamide gels are detected. In addition, other variants occur at low frequency at all the loci examined. These rare variants are clearly distinct from the common forms. A typical locus, GlcGP-dH, is illustrated in Fig. 5. Among the 14loci examined, a total of 32 variants occur at frequencies of greater than 10%. Sixteen of the 32 variants would not have been distinguished on 7% acrylamide gels. In several cases, variants occur that differ only in K R , possessing the same free electrophoretic mobilities. On the 32 common variants, 20%differ only in charge, 10% differ only in K h and 70% differ in both M, and K R . Thus fully 80% of the common variants differ significantly in KR. If differences in K R reflect conformational differences, as seems likely (the matter is discussed below), then it seems quite unlikely that this widespread variation in shape does not affect the functioning of the enzyme. Among the 14 loci, a total of 97 “rare” variants occur; 91 of them occur only once in a sample of 20 individuals; 6 others occur twice. Fully 30%of the genes analyzed in this survey code for proteins that appear only once in the sample. Perhaps these variants all occur typically at frequencies of 5%; alternatively, they may be unique alleles
305
VARIATION IN THE SHAPES O F ENZYMES
0.65
-
0.60
-
0.55
-
0.50
-
Hb (Glc 6P-dH)
I
I
I
> I
E E
FIG.5. Electrophoretic gel-sieving survey of glucose-&phosphate dehydrogenase in a natural population ofColias meadii. Values for hemoglobin (Hb) were determined from the same gels (after 25).
occurring only once. A larger sample is required to resolve this issue.
All 14 loci exhibit rare variants. Fully 70% of the rare variants are not detected on 7% acrylamide gels. Of the 97 rare variants, 34% differ solely in charge, 21% differ solely in K R , and 45% involve differences in both M o and K R . Thus fully two-thirds of the rare variants involve significant differences in K R . Again, conformational variation seems very prevalent.
V. The Nature of Cryptic Variants There have been several speculations recently in the literature concerning cryptic variation at enzyme loci, and what sorts of molecules might be expected to exhibit similar mobilities: The “charge
306
GEORGE JOHNSON
state” model of electrophoretic variation concerns families of alleles with identical net charge, and postulates that they thus migrate to the same band position despite individual differences between them. Polymorphism in heat stability is seen as reflecting internal differences between proteins, properties that have no influence on electrophoretic mobility. Gel sieving permits a more pointed experimental analysis than has been possible to date, as it provides data on physical properties of proteins, 2 explicit experimental error, rather than single-point rate measurements. The “charge state” model, for instance, may be evaluated directly by this approach, as the contribution of charge to mobility may be estimated independently of other gel-interactive effects. When this analysis is carried out for an electrophoretic mobility “ladder” of alleles at the esterase-5 locus of D.pseudoobscuru (26),it is seen that not only charge, but also conformation, plays an important role, and that “even” differences in Rfresult from the interaction of these two effects (Fig. 6). The gel-sieving analysis thus clearly indicates that the charge-state model is inappropriate for these data. Other models may also be evaluated explicitly by gel-sieving analysis. The prediction under the neutral hypothesis that larger genes
I
-0.065
1
I
,
I
I
1
I I
I
I
--
-1.171--
K R -1.071--
-0.010
--
FIG. 6. Conformational and charge properties of alleles at the esterase-5 locus of Drosophila pseudoobscura (after 26). Lines connect forms with identical R , values.
307
VARIATION I N THE SHAPES OF ENZYMES
should be more polymorphic, and thus that heterozygosity should be a function of subunit molecular weight [pointed out by Koehn and Eanes (27)], may be conveniently evaluated, as molecular weight may be determined from K R values (Fig. 2). The predicted relationship is not seen among Colias enzyme loci (Fig. 7 ) . [The matter is discussed further in Johnson (ZS).] The most straightforward interpretation of the K R differences reported here is that they reflect conformational differences between the proteins. That gel-sieving analysis is capable of detecting such variants may be demonstrated directly by analyzing proteins such as bovine fibrinogen, which are known to have very asymmetric shapes. If K R is a sensitive function of conformation, then it should not be a good predictor of molecular weight for very asymmetric molecules (e.g., the K R values obtained by gel sieving should be significantly different than predicted by the value corresponding to its molecular weight in Fig. 2). For several molecules of known asymmetry, this proves to be the case (myosin, fibrinogen, y-globulin; G . Johnson, unpublished). A particularly clear example is provided by a-chymotrypsin (12).This dimeric protein is known from single-crystal
AM-2 0.4
-
()EST-6
-
-
A0
0.3
0.2
-
AM-1.
0.1
i
[APH
PCM
AdH!
]IxdH
C6PdH
r.,
OdH
-
1
0
4.5
4.0
log
5.0
5.5
Subunit Molecular W e i g h t
FIG.7. Enzyme polymorphism in Drosophila melanogaster as a function of subunit molecular weight (after 28).
308
GEORGE JOHNSON
X-ray crystallography to undergo an asymmetric conformational change (only one of the two subunits changes) at pH 8.0. The molecular weight of dimeric a-chymotrypsin (50,000)corresponds on Fig. 2 to a K R value of -0.058. When analyzed at pH 7.5, the value obtained is -0.061, in good agreement. When the same sample is analyzed at pH 9.5, the value ofKRis -0.044. Thus the asymmetric change in conformation produced a marked change in the value of K R . Because a change in conformation can alter K R does not, of course, imply that all differences in K R need reflect conformational differences. There are in principle at least six ways in which K R values might differ between variant proteins. 1. Difference in subunit molecular weight. Protease degradation might be expected to produce changes in protein size, which would be detected as changes in K R . However, grinding and electrophoresis in the presence of protease inhibitors do not appear to alter the K R values of enzyme variants in Colias, which argues that this is not the case. The possibility remains that there is in vivo proteolytic action which itself is segregating in the population. 2. Difference in quaternary structure. Aggregation and dissociation of subunits will of course alter the observed value of K R , but such changes in the number of subunits would be expected to produce far greater changes in K R than are observed. 3. Difference in hydration shell. Variation in K R may reflect changes in the hydration shells of the proteins. Much of the sieving behavior of proteins in electrophoresis may be hydrodynamic, and could in principle be sensitive to the diameter of the hydration shell. This hypothesis would require that single amino-acid substitutions produce rather large changes in the volume of the hydration shell (up to 20%). 4.Difference in the dissociation constant of dimers, aa 2a.Most of the molecules reported here exhibit a third hybrid band in heterozygous individuals, and presumably are dimers. In a homozygote, if the rate is fast and the equilibrium is toward dimer formation, only one band (the dimer) will be seen in electrophoresis. A slight decrease in such a rate has the effect of lessening the fraction of time the protein molecules are in the dimer form. This will decrease the sieving retardation of the protein by the gel fibers in electrophoresis. The perceived result will be a K R suggesting a smaller molecule. 5. Difference in conformation due to epigenetic modification. Binding of carbohydrate or sialic acid to a protein might produce poresize-dependent changes in electrophoretic mobility and thus alter K R values. Staining for glycosides has revealed few glycoproteins in our studies in detectable concentration, and their K R values do not corre-
+
309
VARIATION IN THE SHAPES OF ENZYMES
spond to any of the variants. We are currently attempting to detect bound carbohydrate with more sensitive techniques. 6. Difference in conformation due to difference in amino acid sequence. This is the result predicted by the hypothesis of allelic variation at structural loci. Consistent with it is the observation that when a-chymotrypsin is made asymmetrical its K R value changes significantly. In general, proteins of greater axial ratio exhibit K R values increasingly at variance with that predicted by their molecular weight. The potential involvement of hydration in determining K R may be assessed by altering the conditions of solvation during electrophoresis. The hydration shell of a protein is produced largely by the hydrogen bonding of unchanged polar surface amino acids with the polar solvent water. A significant increase in the polarity of the water solvent would be expected to contract the water envelope. Thus a protein in 99.8% Sample
Kr
2
4Gro P-dH
-0.073 5 .016
0.22
0.01
-0.062 2 .009
-0.068 5 -016
0.10
0.04
3
-0.067 5 ,004
-0.075 5 0.16
0.12
0.04
4
-0.063
2
-007
-0.0715 .010
0.13
0.03
5
-0.065
2
.010
-0.069
.004
0.06
0.11
6
-0.066 5 .010
-0.0715 .011
0.08
0.03
1
-0.060
2
Sample t Mo
P DL-
B.o
I
.003
H,O
2
r DL;
-
olGroP-dH
1
1.56
2.05
0.32
0.04
2
1.66
1.98
0.19
0.01
3
1.77
2.23
0.26
0.01
4
1.59
1.99
0.25
0.01
5
1.67
1.95
0.17
0.14
6
1.69
2.01
0.19
0.02
-
FIG. 8. Comparison of the 4 0 - i n d u c e d alteration in retardation coefficient of aGroP-dehydrogenase with that ofhemoglobin. Both proteins were run in the same gels. The effect of D 2 0 was examined by running eight gels for each individual: 5%,6%, 7%, 8% acrylamide in HzO, and 5%, 6%, 7%, 8% acrylamide in 99.8% DZO. For the D 2 0 runs, all gel reagents and running buffers were made up in 99.8% DzO. The two sets of four gels were run in parallel simultaneously under controlled temperature (lVC), assayed, and scanned; K R values were calculated for the matched treatments. Data are presented as the fractional difference p observed in K R [e.g., (1+ p)KR(inH,O) = Kn(in
DD)l.
310
GEORGE JOHNSON
EFFECT OF
TABLE I DILUTIONON K R
SERIAL
Alcohol dehydrogenase (D.melanogaster) Dilution
Hemoglobin
Ferritin
Alc-dH-1
Alc-dH-2
1.0 0.5 0.25 ,0.12 0.06 0.01
-0.055 -0.057 -0.056 -0.058 -0.057 -0.055
-0.210 -0.217 -0.212 -0.205 -0.217 -0.219
-0.056 -0.056 -0.056 -0.053 -0.056 -0.054
-0.055 -0.057 -0.057 -0.056 -0.056 - 0.055
a-Glycerophosphate dehydrogenase (Colias meadii) Dilution
Hemoglobin
Ferritin
1.o
-0.056 -0.053 -0.055
-0.210 -0.207 -0.208 -0.204 -0.207 -0.207
0.5 0.25 0.12 0.06 0.01
-0.053
-0.056 -0.054
ct GroP-dH
-0.068 -0.066 -0.069 -0.068 -0.069 -0.068
D20rather than H 2 0 would exhibit a smaller hydration shell. If the magnitude of K R reflects the volume of the hydration shell, then electrophoresis in D20should alter K R . More to the point, two proteins with similar molecular weights but with different K R values ought to respond differently to D20, one exhibitng a greater K R alteration than the other. Two such similar protein pairs are bovine hemoglobin and Colias meadii a-glycerophosphate dehydrogenase. The subunit molecular weights are 64,500, and 65,000, respectively. When the two proteins are run simultaneously in water and also in 99.8% D20, and the results are compared, the responses of the two proteins to D 2 0are precisely parallel, despite their quite different K R values. This relationship is illustrated in Fig. 8: very little difference is seen between the two proteins in their response to Q O , despite their difference in K R . These results are not consistent with the hypothesis that the KR differences between the protein pairs reflect hydration shell differenceb. The potential involvement of subunit dissociation in determining K R may be evaluated directly by serial dilution. The rate of dissociation/reassociation is concentration dependent. If& reflects this rate, then it too should vary with concentration. It does not. Serial
VARIATION IN THE SHAPES OF ENZYMES
311
dilution over 2 orders of magnitude produces no change in K R values (Table I). These results are not consistent with the hypothesis that K R differences reflect altered dissociation rates. Thus allelic variation in amino-acid sequence and epigenetic modification remain the two most likely causes of the variation in KR, with no clear evidence yet ruling out either. It is particularly important in this context to note that either hypothesis entails large amounts of conformational variation among the proteins of natural populations. Genetic Basis of Variation in
Shape
1. VARIABILITYIN INBREDLINESOF Drosophilu In order to assess the hypotheses of genetic and epigenetic modification, it is necessary to demonstrate that the large amounts of variation detected by gel sieving analysis do not result from experimental error or some uncontrolled physiological variable. These possibilities were ruled out by testing for variation in inbred lines of Drosophilu. Two such tests were performed. In one, R. C. Lewontin provided me a blind series of ten D.pseudoobscura esterase-6 allele 1.00 lines, some of which were highly inbred, to verify objectively that variants would not be indicated where none could exist. No variants were detected among lines subsequently identified by Lewontin as inbred. In all cases the variation in esterase in inbred lines was no greater than that of the hemoglobin standard run in the same gels (Fig. 9). A second examination of variation in inbred lines was carried out for the alcohol dehydrogenase locus of D.melanogaster. The line examined had been inbred by single sib matings for 456 generations (strain Y49.455, kindly provided by S. Barker). TOdate, over 210 individual flies have been analyzed from this line: In both the NAD-fiee and the NAD-bound form of AdH, no variation has been detected in excess of that seen for the Hb internal standard run in the same gels. These results thus clearly indicate that gel-sieving variation is a phenomenon that does not occur in the absence of genic variation, and suggest strongly a heritable basis to the variation. 2. CONTROLLED CROSSES Previous crosses carried out in Colius butterflies indicated that gel sieving variants at the a-glycerophosphate dehydrogenase locus segregate in the F2generation in Mendelian proportions. Crosses were subsequently carried out in D. pseudoohscuru involving gel sieving variants of line 1.00. Both variants observed within this strain have the same charge, differing only in conformation. The results of these
3 12
GEORGE JOHNSON r
0.65
0.s 0
Hb c
-
0.55
0.5 0
I
I
I
0.1 0
0.6 5 0.60
0.55
0.50
0.45
-0.06
-0.01
-0
.Ol
K R
FIG. 9. Electrophoretic variation at the esterase-5 locus in an inbred line of Drosophila pseudoobscura. Values for the hemoglobin (Hb) internal standards were obtained from the same gels (26).
crosses also support a genetic interpretation: When variant 0.76 is crossed with itself, only this variant type is found among the F2 progeny. Similarly, variant 0.80, when crossed with itself, yields only 0.80 progeny. However, when the two variants are crossed with each other both variant types segregate among the F, progeny.
3. PROGENY TESTS The above results were all consistent with a heritable basis to the gel sieving variation. No crosses, however, have been carried out on the rare variant types, whose genetic nature is of particular interest. In order to assess the heritability of such variants, direct progeny tests
3 13
VARIATION IN THE SHAPES OF ENZYMES
were carried out of variants at four esterase loci of the ocean fish Zoarces, in collaboration with F. Christiansen and V. Simonsen of Aarhus University, Denmark. The fish is viviparous, each gravid female carrying up to 50 live progeny. Thus a survey of female fish permits one to carry out a direct progeny test of any variants detected by gel-sieving analysis. A total of 80 adult fish were surveyed, the four loci being analyzed in each fish. Each of the four loci exhibited several variant classes (Fig. 10).Every detected variant proved to be heritable. One of the progeny analyses is presented for illustration in Table 11. Variant ratios among the progeny were in all cases consistent with simple Mendelian ratios. These results thus establish that the preponderance of esterase variation detected by gel sieving in Zoarces is heritable variation. As the pattern of variation is not substantially different from that seen for Drosophila or Colias, these results suggest that gel sieving variants in general will prove heritable.
4. GENETICMAPPING:CO-ISOGENIC LINES For particular loci, it is possible to test the hypothesis of heritable modification due to the action of an unlinked second site by mapping the resultant variants and ascertaining whether or not they map at or close to the structural gene locus. In collaboration with D. Hartl of Purdue University, we have undertaken such a study for alcohol dehydrogenase variants of D . rnelanogaster. Forty lines of D. rnelanogaster were provided by Hartl in each of which the chromosome carrying Ale-dH had been placed in a common genetic background. If the heritable gel-sieving variation previously observed at this locus results
0.16
0.54
0.12
HtMOOLOllN 0.10
0.M
STANDA 0.1.
0.40
0.44
0.48
0.10
0.14
0.81)
FIG, 10. Gel-sieving analyses of two esterase loci in female Zoarces. Heterozygotes are represented by lines connecting two points. Individual No. 45 is the parent of the progeny shown in Table 11.
314
GEORGE JOHNSON
TABLE I1 ANALYSES OF 12 PROGENY OF INDIVIDUAL NO. 45 (FIGURE10) FOR FOURESTERASE (EST) LOCI" Progeny No.
Est-1
1
-
2
0.043 0.049 (0.064) (0.079) 0.042 (0.062)
3
4
-
5
0.043 (0.064)
6
0.052 (0.087) (0.055) (?.048) (0.077) 0.040 (0.058) 0.046 (0.072) 0.046 (0.072) 0.045 0.047 (0.068) (0.074) 0.042 (0.062) 0.046 (0.072) bh
7
8 9 10 11
12 Female parent Presumptive 9 genotype Presumptive 6 genotype Expected offspring
Est-2 0.056 (0.099) 0.061 (0.115) 0.057 (0.102) 0.055 0.059 (0.097) (0.107) 0.057 (0.102) 0.055 (0.096) 0.061 (0.115) 0.056 (0.099) 0.056 0.058 (0.099) (0.105) 0.056 (0.099)
Est-3 0.054 (0.093) 0.059 (0.108) 0.054 (0.093) 0.055 (0.095) 0.059 (0.108)
0.070 (0.148) 0.066 (0.132) 0.066 (0.132) 0.072 (0.156)
-
0.054 (0.093)
0.053 (0.091)
a/b
0.058 (0.105) b/b
da
ah
a!a
da; a h ; b/b
d a ; ah; blb
da;
9; bib N.O.
HET
-
0.057 (0.103) 0.057 (0.103)
0.054 0.059 (0.094) (0,108)
Est-2 (starch)
0.064 (0.125)
-
0.056 (0.099)
Est-4
0.067 (0.136) 0.066 (0.132) 0.066 (0.132) da
HET
HET
bh $a;
e;b h
N.O.
" Heterozygotes do not form hybrid third bands. Estimates of shape and charge are presented as K R (mid-Y). Only Est-2 may be scored on starch. A marked deficiency in heterozygotes is evident. N.O., not observed. from the action of a locus on a different chromosome, then such variation should not be evident in the 40 lines co-isogenic for the third and the X chromosomes. Each line was subjected to four independent analyses to ensure reproducibility of results. The analyses were in excellent agreement, and the results indicate that the full range of shape variation is exhibited in this sample (Table 111). Thus the source
315
VARIATION IN THE SHAPES OF ENZYMES
TABLE 111 GELSIEVING ANALYSISOF FORTYLINESOF Drosophih melonogoster ISOCHROMOSOMAL FOR CHROMOSOMES 2 abd X“ Range of variation in K R Alcohol dehydrogenase With NAD+ Without N A D +
F allele
S allele
0.059-0.052
0.045-0.048 0.033-0.045
0.045-0.057
G. Johnson and D. Hartl, unpublished results.
of the heritable variation in Alc-dH is on the same chromosome as the Alc-dH structural gene. We are now proceeding to map the variants with respect to the structural gene.
5. NONHERITABLEPOSTTRANSLATIONAL MODIFICATIONS In some cases, commonly employed experimental conditions can act to mask variation that would otherwise be detected. An example is the widely studied alcohol dehydrogenase locus of D. melanogaster. Two isozymes of this enzyme are commonly observed. These two isozymes are interconvertible, representing NAD-bound and NADfree forms (29).Twenty independent lines of Drosophila (kindly provided by w. Heed) isolated from wild populations were examined for variation at the Alc-dH locus. Considerable conformational variation was detected in the NAD-free form of Alc-dH (Fig. 11).This variation was not seen in highly inbred lines, and thus appears to represent genetic polymorphism. It is important that the NAD-bound form of AdH, present in the same individuals, does not exhibit the variation. This suggests that the binding of the cofactor acts to stabilize a particular conformation of the protein, and in so doing may prevent the detection of polymorphism in protein shape. As cofactors are routinely added to gels and homogenizing buffers (to eliminate the “artifact” NAD-free band, or to increase “stability” of the preparation), the binding of such cofactors to enzymes may thus conceal considerable variation.
VI. Major Unresolved Issues: Posttranslational Modification
An extensive survey of enzyme variation at fourteen enzyme loci has been carried out using the analytic gel sieving procedures described above (25,30).The data we have obtained on the amounts and
0.58,
0.54
I @
Hb
Ale-dH
0.40
c
Ale-dH
0.35
0.30
0.25
0.20
0.15
-0.05
-0.06
-0.07
0.04
(b)
0.08
0.1 2
0.16
-KR
FIG. 11. Gel sieving behavior of alcohol dehydrogenase in Drosophilu. a. A highly inbred line; b. individuals from a natural population. Each point represents a different individual. Hb indicates the values obtained for hemoglobin internal standards run in the same gels.
VARIATION I N THE SHAPES OF ENZYMES
317
genetic character of heritable enzyme variation are puzzling in at least three major respects. 1. There is a great deal of variation. For all loci but one, we detected several common variants that cannot be distinguished from one another on routine 7% acrylamide gels. For fourteen loci, a total of 103 variants were detected by gel-sieving electrophoresis. Fully 70%of all variants differ significantly in K R (and thus presumably in shape); 10% differ only in K R . This result is particularly significant in light of data indicating that differences in K R are associated with differences in enzyme activity. It is difficult to reconcile these high levels of polymorphism with the predictions of current theory, under either a selective or a neutral interpretation: under a neutral hypothesis, the new data require that the variants arise from much higher mutation rates than we had previously assumed (31).Under common selection hypotheses, it is difficult to understand how so many heterozygote combinations could be selected simultaneously (32).The surprising amounts of heritable variation revealed by gel-sieving procedures suggest the need to reevaluate our assumptions concerning the mode of this generation. 2. The number of heterozygotes is low. In all the gel-sieving surveys of enzyme polymorphism, there is a very serious deficiency in the number of observed heterozygotes. This has been true of Colias (25, 33 ), D. melanogaster (Johnson and Hart1 unpublished), D. mojavensis (34a),Zoarces (35),and the C4 salt grass Distichlis spicata (36).Many o f the less common variants reported in these studies are detected only once or twice in a sample of 50 or more genes. If such variants represent alleles of structural genes, they would be expected to be observed only in heterozygous combination with more common alleles. Instead, the rare variants are often apparently homozygous, producing single gel bands of differing charge or shape. A similar lack of heterozygotes is sometimes observed for two or more common variant classes at a locus: the predicted Hardy-Weinberg proportions of heterozygotes are not seen, despite the presence of several variants in high frequency. Not all loci are abnormal in this regard; some, such as Malate dehydrogenase (Mal-dH) or phosphoglucomutase produce heterozygotes between common alleles in nearly Hardy-Weinberg proportions (25). The general rule, however, is that far fewer heterozygotes are observed than would be expected on the basis of the observed variant frequencies. 3. There are too many rare alleles. The frequency distribution of variants is highly skewed, most variants occurring only once (25).The general situation is to observe one or two common variants at a locus,
318
GEORGE JOHNSON
and many rare ones. While some skew in allele frequency distribution is expected ( 3 7 3 9 ) ,the amounts observed in surveys of gel sieving variation seem excessive. Data have recently been reported for “hidden” variants at the xanthine dehydrogenase (Xari-dH)locus in D. pseudoobscuru (40)and in D.persirnilis (41), as well as at other loci (7,42),which are in broad agreement with our results, although at least one locus (hexokinase) has been reported to exhibit little “hidden” variation in Drosophila (43). Lewontin and his colleagues have studied the Xan-dH locus using two-gel comparison approaches that, while not as painstaking as the gel sieving procedures described above, detect analogous classes of variation: instead of estimating R f at six gel concentrations and evaluating the mobility function, two dissimilar gel concentrations are simply compared for R , difference. Instead of estimating isoelectric point (33),two buffer pHs are compared for their effect on Itfi Xan-dH is reported to be segregating for 23 alleles in D.persirnilis and 37 alleles in D. pseudoobscuru. The suggestion has been made that the large numbers of variants and very skewed frequency distribution at the XdH locus may reflect multiple niche selection (32). These three puzzling characteristics of “hidden” variation, while difficult to understand under the conventional hypothesis of allelic variation at enzyme loci, are readily understood in terms of an altemative hypothesis: the high degree of enzyme variation may reflect extensive modifications of the enzymes by the products of other genes. Such second-site posttranslational modification would have all the major characteristics reported for enzyme “polymorphism”: (1) it will appear as heritable enzyme variation; (2) if several modifying loci are active, very many enzyme variant classes may be produced; (3) if several such loci act independently (e.g., produce different modifications)¶ their effect will be multiplicative and a highly skewed frequency distribution may result, with many rare and few common classes; (4) if polymorphism at modifying loci exhibits simple dominance¶ heterozygote types will not be observed for the enzyme that is the target of the modification-what will be seen instead are different single-band types. This is illustrated in Fig. 12, which outlines three alternative hypotheses concerning the potential effects of a second-site posttranslational modifier on realized enzyme banding patterns: (a) no variation at the modifying locus; (b) codominant variation at the modifying locus; (c) dominanthecessive variation at the modifying locus. Dominanthecessive variation at a modifying locus would be expected to produce a pattern of heterozygote-deficient enzyme variation similar to that observed in gel-sieving analyses of enzyme polymorphism in natural populations of insects.
--
VARIATION IN THE SHAPES OF ENZYMES
XdH
maA*
XdH
ma-I
3 19
3-
c
4
0
modifiad
0
'ubuni*'
1
FIG. 12. Heterozygote phenotypes produced by enzyme alleles and by second-site modifiers. The gel pattern produced by heterozygosity at a modifying locus depends upon the degree of dominance of the modification. If modifier alleles lack dominance, a multiple-band gel pattern will be obtained just as expected for heterozygosity of enzyme alleles. If a modifier allele is dominant, a single gel band will be obtained. In surveys of populations containing such dominant modifiers, different single bands will be seen for each dominant allele; no heterozygotes will be detected.
Since heritable polymorphisms at modifying loci can in principle provide reasonable explanations for several aspects of our data that hypotheses of allelic structural gene variation do not, it is of real interest to determine if such inodifications are indeed capable of producing the patterns of variation actually observed in natural populations. To investigate this issue we have examined the Xan-dH locus of 0. melanogaster. This system is of particular interest because (1)Xan-dH exhibits considerable variation in populations of D. pseudoohscura (40,44),( 2 )from a genetic viewpoint, the Xan-dH structural genein D. melanogaster is one of the best-understood loci among higher eukaryotic organisms (see 45 for a detailed review), and ( 3 )three other distinct loci are known to affect Xan-dH activity, most likely by posttranslational modification ( 4 6 , 4 7 ) . My studies with Finnerty (6a)thus far indicate that mu-] mediates a posttranslational modification of these proteins: examination by gel sieving of a series of ma-I heteroallelic combinations indicates that in every cuse there is a significant alterution in protein .shape (Fig. 13). At least for one ma-I heteroallelic combination, alteration in KH is SUB-
320
GEORGE JOHNSON
3.8
XANTHINE
-a\ ,,_-, \/63/$62(
DEHYDROGENASE
-
,&
1
3.0
7
-
.;yb\
/’
‘--/
f
,’
2.2-
’
1.4
..
\U?Of3/27 \o ‘ p d type (+4) 10.
/
-
m
5
\
\
/
f3/1
\ \ c
/
-
A
f3/2
‘ A A
A
A
\
/
/
‘--’ I
0.08
II
, ‘\--/’
/
A
\
.
“\,
.I.. I
.--I,
/ /
-.
0.10
I
t
1
0.12
0.14
I
I
0.16
1
I
I
0.18
FIG.13. Gel-sieving analyses of wild-type xanthine dehydrogenase (Xan-dH) and of four different heteroallelic combinations of ma4 mutant alleles. A11 of the stocks examined, ma-1 and wild type, are co-isogenic for chromosome No. 3, which contains the structural gene for Xan-dH. The alleles at the ma-1 locus were selected as Xan-dH negatives (ma-1 function is required for Xan-dH activity); for analysis we have taken advantage of intra-allelic complementation, which occurs at the ma-1 locus (presumably two ma-1 mutant protein subunits combine to form a partially active multimer). The heteroallelic combinations of mu-1 analyzed here were constructed by crossing the individual mutant alleles, so that each heteroallelic stock is heterozygous for a particular pair of complementing alleles (6~).
cient to produce a detectable difference on routine starch or 5% acrylamide gels (Fig. l),although for most mu-1 alleles the alteration is not detected by such gels, and is instead revealed as “hidden” variation by gel sieving analysis. The magnitude of the posttranslational K R differences in XdH and aldehyde oxidase (AldOx) are consistent with the differences reported in earlier gel-sieving surveys. The implication is that heritable hidden variation at this locus, and potentially at other loci, may not be allelic to the structural gene. Such a hypothesis would explain the heterozygote deficiency among heritable “alleles” sampled from natural populations: posttranslational modifiers would act like simple dominants, and would not be expected to produce intermediate “heterozygote” bands (see case 3, Fig. 12). Over the last decade, electrophoresis has been employed widely in
VARIATION IN THE SHAPES OF ENZYMES
32 1
population genetics as a tool for examining primary gene products. Its great utility has been in directly visualizing the results of mutational change: single nucleotide substitutions could produce a new allele whose protein product has an altered electrophoretic mobility. The results obtained in this study warn that such a view may be overly simple. For Xan-dH and AldOx, the electrophoretic gel pattern must be considered a phenotype existing within a physiological framework, subject to epistatic interaction and modijication by other loci. For an unknown but perhaps significant proportion of other enzyme polymorphisms, similar degrees of phenotypic complexity will occur. Uy and Wold (48)list 140 different classes of covalent posttranslational modification reported to date and argue that the phenomenon is very widespread in nature. The properties of variation detected by gel sieving electrophoresis at many other loci also support the hypothesis of widespread occurrence. Several clear cases of modifiers affecting particular loci have been recently reported ( 4 9 5 2 ) . Owing to the almost universal use of electrophoresis in population genetics, and the importance of the issues addressed with the resulting data, it is highly desirable that the possibility of extensive polymorphisms for posttranslational modifiers in natural populations be evaluated in a direct empirical fashion.
VII. Novel Approaches to Electrophoresis To these studies I have recently applied two new high-resolution analytic procedures. Their utility can perhaps best be appreciated by consideration of the fundamental equations that govern electrophoresis. The most basic of these says that observed mobility is a function of the diffusion constant, D , the electric field strength, E , and the free electrophoretic mobility of the protein [Eq. ( l ) ] The . important thing is that the mobility is entirely a hnction of the three variables D, I , and M,. All the analytical applications of electrophoresis are based on this rekitionship: if one holds E constant, then, for a defined D , the observed mobility should be a hnction of the protein’s mobility, M,. D need not, however, be kept constant; most of our progress in detecting new electrophoretic variants has come through alteration of parameters that interact to determine D:
D thus is a function of net charge, Q; molecular radius, R; counterion radius, ri; viscosity, r); and several constants. Any of these four vari-
322
GEORGE JOHNSON
ables are subject to experimental manipulation. When one changes the buffer pH to detect new variants, or, more precisely, determines and compares isoelectric points, one is varying Q . When one changes buffer ions to detect new variants, one is varying Ti. When one changes acrylamide concentration in gel sieving analysis, one is varying the steric viscosity term 7.The sorts of variation detected by these and other manipulations of D are outlined in Table IV. We have developed two new procedures in an attempt to detect polarhonpolar amino-acid substitutions that do not alter charge or protein shape. The first of these involves changing solvent polarity. The polarity of the solvent is readily manipulated by running gels in 99.8% D,O. The results of one such screening procedure are presented in Table V: one individual is clearly variant, and this individual is indistinguishable from other individuals by conventional approaches, and by both pZ and gel-sieving analyses. Even greater resolution may be obtained by varying the temperature of gel-sieving analysis: While DzO increases K R by increasing hydrophobic interactions with solvent (the strength of those interactions depending upon the polar nature of the protein-hence the detection of polarity variants), temperature decreases K , as it interferes with hydrophobic interactions with solvent. Ae is a log-linear function of temperature,
One may plot log K R as a function of Ytemperature. The plot is linear, and any polarity variant will have a differing intercept. An example (the same variant individual analyzed with DzOabove) is presented in Fig. 14. TABLE IV NOVEL APPROACHES TO ELECTROPHORESIS Protein alteration Charge change Shape change Shape change Polar
* nonpolar substitution
Subunit binding site change
Procedure
1. 2. 1. 1. 2. 1. 2. 1.
Change pH Determine isoelectric point Change buffer ion Change gel concentration Determine K,by gel sieving Change polarity of solvent Change temperature Dilute sample
Variable
Q TI
r) r)
steric
hydrodynamic
R
323
VARIATION IN THE SHAPES OF ENZYMES
TABLE V EFFECT O F D,O ON THE ELECTROPHORETIC BEHAVIOROF a-GLYCEROPHOSPHATEDEHYDROGENASE (aGroP-dH) I N A SAMPLEOF C o h s ntecidii"
H@
D@
p
-0.060 t 0.003 -0.062 2 0.009 -0.067 t 0.004 -0.063 ? 0.007 -0.065 f 0.010 -0.066 C 0.010
-0.073 t 0.016 -0.068 ? 0.016 -0.075 ? 0.16 -0.071 t 0.010 -0.069 -C 0.004 -0.071 ? 0.011
0.22 0.10 0.12 0.13 0.06 0.08
Sample No.
H,O
D,O
A
1 2 3 4 5 6
1.56 1.66 1.77 1.59 1.67 1.69
2.05 1.98 2.23 1.99 1.95 2.01
0.32 0.19 0.26 0.25 0.17 0.19
Sample No.
K,
Mo
1 2 3 4 5 6
aGroP-dHHh
0.01 0.04 0.04 0.03 0.11 0.03 aGroP-dHHh
0.04 0.01 0.01 0.01 0.14 0.02
" Sixteen additional individuals were surveyed with no evidence of variation (Johnson, unpublished). Determinations on H,O and on 99.8% D,O are performed simultaneously on identical samples. p represents the fractional difference in K, produced by D,O, A the fractional difference in M,. aGroP-dHHhrepresents point-to-point comparison ofaGroP-dH to hemoglobin run in the same gels (Johnson, 1977).Run-torun variation is considerable unless standardized to an internal standard protein such as hemoglobin (Hb). Individual No. 5, normal by other criteria, shows a significant D,O effect when standardized to hemoglobin.
.
¶o
.
.
10
6
0.1
b
a2
FIG.14. High-resolution analysis of variation in a-glycerophosphate dehydrogenase. Gel sieving is carried out at each of four temperatures. Three normal individuals and one variant individual are presented. The four individuals are indistinguishable in normal electrophoresis, isoelectric focusing, or gel-sieving analysis.
324
GEORGE JOHNSON
I regard these new procedures as very promising, although I have only begun to apply them to surveys of naturally occurring variation. One class of polar-nonpolar amino-acid substitution that might go undetected by these procedures consists of substitutions at the hydrophobic subunit binding sites of multimers, as such residues are not normally exposed to solvent. Polymorphic variation in subunit binding has been reported for esterase variants in D.pseudoobscura (53).To the extent that such substitutions alter subunit dissociation rates, they will be detected in gel-sieving analysis as exhibiting differing K R values (as the proportion of time spent dissociated is altered, the effective size in gel sieving is altered accordingly). Subunit binding-site changes are readily distinguishable from changes in protein shape, however, as subunit exchange rates are concentration dependent. One simply compares K R values at two differing dilutions of sample. In my surveys of thermal stability variation among Zoarces esterase loci, carried out coordinately with gel sieving study, every case of unusual thermal liability or stability was also detected as a K R variant. It is important to note that posttranslational modifications such as carried out by ma-l may produce thermal stability variants indistinguishable from the sort detected by surveying stability on heated gels (Fig. 15).Thus reported enzyme polymorphisms for heat stability are also subject to reservations concerning possible second-site modification. Whether the shape variation is allelic to structural genes or represents second-site modification, it appears that most if not all of it is heritable. Thus any functional differences resulting from these variants are of real genetic and evolutionary significance. The correspon-
FIG.15.Thermal stability of xanthine dehydrogenase (Xan-dH) variants. For each of the mu-1 lines described in Fig. 13, the thermal stability of Xan-dH was deterniined as t l E ,the incubation time at 65°C required to produce H loss of 50%ofXan-dH activity (47). The corresponding degrees of Xan-dH shape modification produced by each of the difference in KR from wild type, estimated from Fig. 13. Shape and stability are clearly correlated.
325
VARIATION IN THE SHAPES OF ENZYMES
loo)
Y Y
LO
P
t
40
43
a0
-0-
c 0
20
10
%DIFFERENCE
IN
40
30
KR F R O M
WILD
SO
TYPE
FIG. 16. Activity of conformational variants. The four variants represent four heteroallelic combinations of ma-1 alleles, each having modified the shape (KR)of the xanthine dehydrogenase (Xan-dH) proteins to different degrees. The amount of enzyme activity is clearly related to estimates of enzyme K,+ The levels of Xan-dH activity have previously been characterized for each of the various ma-1 heteroallelic combinations (47).
dence for Xan-dH and AldOx between difference in shape as assessed by gel sieving and level of activity (Fig. 16) encourages me to expect that the widespread differences in shape detected in my previous gelsieving surveys may reflect similar differences in function. I regard the result of Fig. 16 as of particular importance for this reason, as it establishes that differences in K R may indeed reflect activity differences.
REFERENCES 1. K. Ferguson, Metaholism 13,985 (1964). 2. G. Johnson, PNAS 68, 997 (1971). 3. L. Ornstein, Ann. N.Y. Acad. Sci. 121, 321 (1964). 4. A. Chrambach and D. Rodbard, Science 172, 440 (1971). 5. D. Rodbard and A. Chrambach, PNAS 65, 970 (1970). 6. D. Rodbard, G. Kapadia and A. Chrambach, Anal. Biochem. 40,135 (1971). 6a. V. Finnerty and G. Johnson, Genetics 91, No. 2 (1979). 7. G. Cobbs and S. Prakash, Genetics 85, 697 (1977). 8. B. Davis, Ann. N.Y. Acad. Sci. 121, 404 (1964). 9. D. Rodbard and A. Chrambach, Anal. Biochem. 49, 95 (1971). 10. A. Gonenne and J. Lebowitz, Anal. Biochem. 64, 414 (1975). 11. J. Hedrick and A. Smith, ABB 126, 154 (1968). 12. G. Johnson, Biochem. Genet. 14, 403 (1976). 13. H. Harris, Proc. Roy. Soc. Ser. B 164, 298 (1966). 14. J. Powell, Eoolzrt. Biol. 8, 79 (1975). 15. J. Hamrick, in “Population Biology of Plants” (0. Solbrig, S. Jain, G. Johnson and P. Raven, eds.), Columbia Univ. Press, New York, 1978. 16. T. Wright and R. MacIntyre,J. Elisha Mitchell Sci. SOC. 81, 17-19 (1965). 17. S. Bemstein, L. Throckmorton and J. Hubby, PNAS 70, 3928 (1973).
326
GEORGE JOHNSON
18. R. Singh, J. Hubby and R. Lewontin, PNAS 71, 1808 (1974). 19. R. Singh, J. Hubby and L. Throckmorton, Genetics 80,637 (1975). 20. G. Trippa, A. Loverre and A. Gatamo, Nature 260, 42 (1976). 21. R. Milkman, Biochem. Genet. 14, 383 (1976). 22. B. Cochrane, Nature 263, 131 (1976). 23. G. Thorig, A. Schoone and W. Scharloo, Biochem. Genet. 13, 721 (1975). 24. B. Sampsell, Biochem. Genet. 15, 971 (1977). 25. G. Johnson, Biochem. Genet. 15,665 (1977). 26. G. Johnson, Genetics 87, 139 (1977). 27. R. Koehn and W. Eanes, Theor. Popul. Biol. 11, 301 (1977). 28. G. Johnson, Cum. Top. Biol. Med. Res. 2,1(1977). 29. K. B. Jacobson, J. B. Murphy, J. A. Knopp and J. R. Ortiz, ABB 149, 22 (1972). 30. G. Johnson, Stadler Symp. 7, 91 (1975). 31. G. Watterson, Genetics 88,171 (1978). 32. R. Lewontin, L. Ginzburg and S. Tuljapurkar, Genetics 88, 149 (1978). 33. G. Johnson, Genetics 83, 149 (1976). 34. G. Johnson, PNAS 75, 395 (1978). 35. G. Johnson, V. Simonsen and F. Christiansen, Heriditus, in press (1978). 36. M. Emma, Curnegie Inst. Wash. Yearh. 75, 409 (1977). 37. W. Ewens, Theor. Pop. Biol. 3, 87 (1972). 38. G. Johnson and M. Feldman, Theoret. Pop. Baal. 4, 209 (1973). 39. M. Nei, “Molecular Population Genetics and Evolution.” North-Holland Press, Amsterdam, 1975. 40. R. Singh, R. Lewontin and A. Felton, Genetics 84, 609 (1976). 41. J. Coyne, Genetics 84, 593 (1976). 42. R. McDowell and S. Prakash, PNAS 73, 4150 (1976). 43. A. Beckenbach and S. Prakash, Genetics 87, 743 (1977). 44. S. Prakash, Genetics 87, 159 (1977). 45. A. Chovnick, W. Gelbart and M. McCarron, Cell 11, l(1977). 46. V. Finnerty, in “The Genetics and Biology of Drosophih” (M. Ashburner and E. Novitski, eds.) Academic Press, New York, 1976. 47. V. Finnerty, M. McCarroii and G. Johnson, submitted to MoZ. Gen. Genet. 48. R. Uy and F. Wold, Science 198, 890 (1977). 49. W. Watt, Genetics 87, 177 (1977). 50. B. Cochrane, Genetics 83, 516 (1976). 51. M. Dizik and R. Elliott, B i o c l ~ mGenet. . 15, 31 (1977). 52. M. Dizik and R. Elliot, Biochem. Genet. 16, 247 (1978). 53. G. Cobbs, Genetics 82, 53 (1976).
Transcription Units for mRNA Production in Eukaryotic Cells and Their DNA Viruse&
I
JAMES
E. DARNELL, JR.
The Rockefeller Unioersity New York, New York
I. Introduction ..................................................... 11. Techniques for Defining Transcription Units ...................... A. Kinetic Labeling ............................................. B. Electron Microscope Visualization ............................. C. Ultraviolet Transcription Mapping ............................. 111. Definition of Transcription Units for mRNA ....................... A. The Average Transcription Unit ............................... B. Transcription Units for Specific mRNAs ........................ IV. The Precursor Relationship of Large Nuclear Primary Transcripts to mRNA ............................................. A. Kinetic Labeling Studies . . . . . . ............................ Primary Transcripts for B. Ultraviolet Evidence for the Siz mRNA Production ............................................ V. Recent Evidence on the Formation of mRNA from Primary Transcrivts ...................................................... VI. Models for the Regulation of Eukaryotic Gene Expression ...................................................... A. Process vs. Discard Decisions ................................. B. Selective Processing Decisions ................................ C. Present Knowledge about Transcription Unit Design . . . . . . . . . . . . VII. Conclusion ...................................................... References ...................................................... Note Added in Proof ..........................................
327 329 330 330 330 331 331 333 337 337 338 339 344 344 345 347 349 350 353
1. Introduction In bacteria the rate of proteih synthesis for any specific protein is controlled by the supply of mRNA for that protein (1). In turn, the supply of any one mRNA is governed mainly by the decision to promote (2, 3 ) or restrict transcription of the gene corresponding to that particular mRNA (4-6). It is natural, therefore, in considering how animal cells might control gene expression to suppose that transcriptional regulation will also prove to be important in controlling mRNA supply. Indeed, in a number of instances where qualitative or quantitative changes occur in protein synthesis in eukaryotic cells, corre'This work is supported by grants froni the National Institutes of Health (CA 16006-05)and the American Cancer Society (NP 213G). 327 Progress in Niicleic Acid Research and Molecular Biology, Vol. 22
Copylight @ 1979 by Academic Press. Inc. All tights of reproduction in any form reserved. ISBN 012-540022-5
328
JAMES E. DARNELL, JR.
sponding shifts of transcription also appear to occur. Two examples are: (a) during insect development, waves of transcription result in microscopically visible structures called “puffs” develop at specific chromosomal loci as development proceeds (7);(b) DNA-containing animal viruses, like bacteriophages, have an “early” and “late” phase: the genes that function early in infection are not the same as the genes that function late, and the mRNA that accumulates is qualitatively different “early” and “late” (8, 9). However, the assumption of a complete parallel in the possible mechanism(s) of gene control between prokaryotic and eukaryotic cells seems unwarranted when the details of mRNA production and utilization are considered. In bacteria, the primary RiVA transcript is the mRNA. As soon as the new RNA molecule begins to be transcribed, it is engaged by ribosomes, which then begin protein synthesis (10). Indeed a recent report supports an older suggestion that transcription may in fact be dependent on simultaneous translation ( I O U ) .In contrast, in eukaryotic cells, the primary transcript is not used directly as mRNA. Both the 5’ and 3‘ ends of most mRNA molecules undergo posttranscriptional modifications. The 5’ end consists of a methylated blocked oligonucleotide structure termed a “cap” (11 ) while the 3’ end of most mRNAs consists of poly(adeny1ic acid) (12) that is added after transcription.2 In addition, recent evidence to be reviewed here suggests that endonucleolytic cleavages of a primary transcript larger than mRNA is followed by RNA-RNA “splicing” or “ligation” of selected RNA segments during mRNA formation. These posttranscriptional steps in eukaryotic mRNA formation, which do not qccur in prokaryotes, raise the possibility that there may be multiple decision points for mRNA regulation, possibly at posttranscriptional levels as well as at the level of transcription. Such possibilities have been discussed many times in the past (12-16), but recent results enable much more specific models of regulation to be proposed. Many studies suggest that, in fact, different types of controls on eukaryotic mRNA production may exist. Investigations using reassociation kinetics between mRNA and radioactively labeled cDNA (complementary DNA produced by the viral enzyme reverse transcriptase) have established a wide diversity of mRNAs in various mammalian tissues (17-19). Cells from the liver, kidney, and brain, for example, have as many as 10,000 different mRNAs, some (estimates range from 20%to 50%) being confined to one of the three tissues. More See articles in Part I of “mRNA: Structure and Function,” Vol. 19 of this series.
[Ed.]
TRANSCRIPTION UNITS FOR
rnRNA
329
striking, perhaps, is the extensive similarity ofthe mRNA populations in all three tissues. The relativeabundance of mRNA changes, however, so that what is abundant in liver may be rare in kidney and brain. These changes in concentration of the shared mRNAs can be as large as 50-fold. The nuclear RNAs in these tissues do not differ nearly so much in the total range of sequences present (18,19a). Furthermore, while the cytoplasmic mRNA of developing sea urchins contain some overlapping and some nonoverlapping sequences during blastula and pluteus stages, the nuclei apparently contain the same RNA sequences during both stages of development (20,21).Thus both qualitative and quantitative changes in mRNA content can be expected in different cells when individual mRNAs are measured, and it is not yet known at what level control lies. Finally, it has been reported that RNA containing histone-mRNA sequences is transcribed at all times in the cell cycle, but histone mRNA appears only during S phase (22),suggesting possible posttranscriptional regulation in this case. The first requisite in determining the level of reguhtion of any gene encoding a particular mRNA is the identification of the primary RNA transcript that, after modification, becomes mRNA. The fateconservation, stabilization, and utilization-of any particular primary transcript must then be followed through the various posttranscriptional steps and into the cytoplasm to conclude unequivocally whether the gene is regulated at the transcriptional or posttranscriptional level or both levels. Considerable recent progress in defining the primary transcripts that serve in mRNA production and how these transcripts are processed into mRNA has been achieved, These results allow fairly specific models of gene regulation to be proposed. This paper deals with the identification of transcription units, the proof necessary to show that a primary transcript is an mRNA precursor, new developments in mRNA processing, and, finally, models for gene regulation.
I I . Techniques for Defining Transcription Units The region of a DNA genome between a start and stop signal for RNA polymerase is a transcription unit. Within a given region of DNA, more than one start and/or stop signal may exist, so the boundaries of a transcription unit might better be thought of as functioning start and stop signals. The RNA produced from a functioning transcription unit is the primary RNA transcript, or the primary transcript. Theoretically, the identification of a transcription unit might be made either by examining the length of DNA being transcribed at any site, or by identifying the unmodified primary transcript.
330
JAMES E. DARNELL, JR.
Three experimental techniques have been developed for defining transcription units in mammalian cells: (a) kinetic labeling; (b) electron microscope visualization; (c) ultraviolet transcriptional mapping. All of these were first applied to ribosomal RNA biosynthesis. A large precursor to rRNA was first identified as a primary rRNA transcript by kinetic labeling (23,24 ) and then confirmed by both of the other techniques (25,26).The early findings that rRNA and tRNA (27-29) were formed in cultured animal cells by reducing the size (“processing”) of primary transcripts has influenced much of the work on the definition of primary transcripts for mRNA. The logic involved in the use of the three techniques is summarized below. A. Kinetic labeling Exposure of cells to radioactive RNA precursors is limited to shorter and shorter times, and radioactivity in a specific RNA sequence is assayed. Identification of all or almost all of the sequences characteristic of a specific type of RNA in an initially labeled RNA species larger than the final product is presumptive evidence of an RNA precursor molecule. A particular strength of the kinetic labeling technique involves shortening the exposure of cells to a labeled RNA precursor to a time less than that required to finish RNA chains. Analysis of the terminally labeled, growing chains from a specific region of DNA allows a deduction of the direction of chain growth as well as a prediction of the length of transcription units (7, 30, 31). Thus far, this is the most widely used approach in the definition of specific transcription units. B. Electron Microscope Visualization If one type of active transcription unit can be isolated in sufficiently pure form or is sufficiently distinctive in character, the boundaries and direction of transcription can be visualized in the electron microscope (EM). This technique has not yet been applicable to specific mRNA formation except for silk fibroin mRNA (32),but it has been used to examine the total range of transcription units operating in insect nuclei (33, 34). C
. UItraviolet Tra nscription Mapping
Pyrimidine dimers, which accumulate in DNA as a result of UV irradiation, act as premature RNA chain terminators, but initiation of RNA synthesis proximal to the UV damage continues (35, 3 6 ) . Two features of UV damage to transcription have been most useful: (1)
TRANSCRIPTION UNITS FOR
mRNA
331
larger transcription units have a greater target size than smaller transcription units, frequently diagnosed by relative survival as a function of increasing dose; (2)within a given transcriptional unit, a single UV dose causes a greater decrease of RNA synthesis in promoter-distal than in promoter-proximal sequences. After UV irradiation, an exponential decrease in RNA synthesis along a particular segment of DNA indicates a single transcription unit (36). Unfortunately, the original and perhaps most elegant and effective means of transcription unit mapping in bacterial cells, the isolation and mapping of mutations that affect transcription (2-6), is not available for mammalian cells or their viruses. Not only are mutants more difficult to select in higher eukaryotic cells, but genetic manipulation of any mutated region, as was done in bacteria to prove the nature of mutation, is at the moment all but impossible in cells of higher eukaryotes. Even in lower eukaryotes, progress in defining and characterizing regulatory genes has been slow.
111. Definition of Transcription Units for mRNA A. The Average Transcription Unit
Before describing the available results on specific primary transcription units related to mRNA formation, we will consider the present evidence on the total range of transcription unit sizes from which mRNAs may be derived. If most primary transcription units were the same size as their mRNAs (about 1500 nucleotides), any technique of transcription unit examination should reveal only the long pre-rRNA transcript (ca. 14,000 nucleotides) and short mRNA primary transcripts (ca. 1500; ignoring the smaller 4 S and 5 S RNAs). Furthermore, in such a case, the only required posttranscriptional processing steps in the formation of mRNA would be the addition of the 5' cap I and the 3' poly(A) segment followed by transport to the cytoplasm.2 However, it was found years ago that eukaryotic cell nuclei produce as their major product a high-molecular-weight RNA, substantially larger than mRNA, with molecules ranging in size from 2000 to 20,000 nucleotides (23,24,37).Also, for many years it has been conjectured that at least some of this so-called heterogeneous nuclear RNA (hnRNA) might give rise to mRNA after endonucleolytic cleavage, preserving only a portion of the larger primary transcript as mRNA. In recent years, hnRNA has been shown to contain sequences present in mRNA ( 3 8 4 0 )as well as the terminal sequences added posttranscriptionally
332
JAMES E. DARNELL, JR.
that are characteristic of mRNA molecules, the 5’ methylated “cap” structure (41,42)and the 3‘ polyadenylate (12).These results clearly enhanced, but did not prove, the possibility that hn IWA represents the primary transcripts for mRNA. One source of confusion and perhaps the basis for a widespread reluctance to accept hnRNA as precursor to mRNA is the extremely large size of some of the hnRNA molecules and the overwhelming resultant waste that would be entailed if molecules of 20,000-50,000 nucleotides were obligatory primary transcripts for every mRNA molecule of 1000-2000 nucleotides. The uncertainty of the nature of the smaller hnRNA molecules-i.e., are the smaller, actually more numerous, hnRNA molecules breakdown products? or are they also primary transcripts?-has recently been clarified in cultured cells by several different types of experiments. The first definite evidence on the average transcription unit size for hnRNA molecules came from a quantitative analysis of the size distribution of hnRNA labeled only as growing or “nascent” chains (43).The ideas underlying this analysis follow. (a) If exposure of cells to a 3H-labeled nucleoside is short enough, the labeled hnRNA will exhibit a shorter average sedimentation profile since all growing chains, short and long, would contribute equal radioactivity, i.e., the sedimentation profile would not represent the “weight-average” distribution but the “number average” for “nascent” RNA. Label times of less than 30-40 seconds with [3H]uridine produced such “nascent chain” patterns. (b) In a nascent chain distribution from a heterogeneous collection of transcription unit sizes, all growing chains contribute equally to the smallest size class, but fewer labeled chains will be found in each succeeding larger size class. (c) The ratio ( L / l )of the number of chains in any larger size class, L, to the number in the smallest size class, 1, predicts the number of chains that, when complete, will reach or exceed L in length. An analysis of the labeled, nascent HeLa cell hnRNA employing these principles revealed that half of the active transcription units in HeLa cells exceed about 5000, and half are about that size or shorter (43). A further test of the average size of the transcription units in cultured mammalian cells is available by UV transcription analysis. If the labeled, nascent “short” (ca. 1000-2000 nucleotides) hnRNA chains represent breakdown products of longer transcriptional units, then the UV target size of short, nascent labeled RNA should be equal to that for the larger nascent RNAs. In both L cells (44)and HeLa cells (43, 4 5 ) the UV target size of the hnRNA corresponded to its size determined by electrophoresis or sedimentation, indicating that most of the hnRNA observed represents primary transcription products in the size
TRANSCRIPTION UNITS FOR
mRNA
333
range of about 2000-20,000 nucleotides. For HeLa cells, the UV analysis showed, as did the nascent-chain analysis, that about one-half of the primary transcripts were longer than about 5000 nucleotides and one-half were shorter (45). Other recent work on the average transcription unit size in eukaryotic cells has involved EM visualization of hnRNA transcription in insect cell nuclei (33, 34). Primary transcriptional units that averaged well over 10,000 nucleotides with very few shorter units were reported. However, a transcription unit of only a few thousand nucleotides in length that contained only a few polymerase molecules would be hard to detect visually, so a quantitative estimation of sparsely occupied, shorter transcription units would be difficult. Therefore the average size of transcription unit observed in these EM studies could be larger than the true average. Nevertheless, it is clear that many transcription units in insects exceed 10,000nucleotides, and few that were completely occupied by RNA polymerases were observed in the range below 5000 nucleotides. An outstanding fundamental question arises in considering the average transcript size of hnRNA: For any individual region of the genome, is one and only one primary transcript possible? In reviewing the experimental definition of transcription units containing specific mRNA sequences, it is most important first to ask: Is a primary transcript observed that is larger than the specific mRNA, and then, if such a larger transcript is observed, is it the only primary transcript? The possibility of a long transcript not productive of mRNA and a shorter transcript that is the mRNA precursor must eventually be satisfactorily resolved. 6. Transcription Units for Specific mRNAs 1. “75 S” RNA FROM Chironornus tentans
Perhaps the earliest studies on a specific cellular primary transcript related to an mRNA were performed with salivary glands or explants from Chironomus tentans [reviewed by Daneholt (7)]. In these cells, a localized region, a “Balbiani ring,” in the second chromosome expands into a center of rapid RNA synthesis. Daneholt, Edstrom, and their colleagues manually dissected this Balbiani ring and showed that it generates a single species of RNA about 30,000 nucleotides in length. This “75 S” RNA, tenninated by poly(A), is found in large cytoplasmic polyribosomes, presumably directing the synthesis of salivary gland protein. Labeled nascent (unfinished) chains shorter than 75 S could be detected in the Balbiani ring RNA labeled for a few
334
JAMES E. DARNELL, JR.
minutes, but no chains longer than “75 S” were observed. Thus it was concluded that the primary transcript in this case is very near the size of the functioning mRNA. However, since the 75 S mRNA is at least 30,000 nucleotides long, a primary transcript only a few thousand nucleotides longer than the mRNA would be difficult to detect.
2. THE SILKFIBROIN mRNA OF Bombyx mori Suzuki and Brown (46)isolated pure silk fibroin mRNA, which, like the Chironomus salivary gland mRNA, has an unusually large size over 16,000 nucleotides in length (47)as well as an unusual oligonucleotide makeup including a base composition high in guanine and cytosine related to the repetitious arrangement of the amino acids in silk. Lizzardi (30)took advantage of the high guanylate by using an artificial heteropolymer, poly(C-C-A)-Sephadex, which he found would specifically select the silk fibroin mRNA. Very briefly labeled silk fibroin RNA that bound to this absorbent revealed a nuclear RNA about the same size as the fibroin mRNA. When the silk worms were chilled below 15°C before a pulse-label, a specific RNA slightly larger than the final fibroin mRNA in the cytoplasm was detected. Because the posterior silk gland forms such a large amount of silk fibroin mRNA and because the size of the mRNA was known to be very large, Miller et al. spread silk gland nuclei in an attempt to visualize silk fibroin gene transcription. They observed long DNA fibrils undergoing continuous transcription over a region approximately the length of the silk fibroin mRNA (32). The nuclear RNA upon completion of this transcriptional unit was estimated to be some 10-15% longer than the cytoplasmic mRNA. These very long mRNAs from secretory glands of insects may or may not be good models for mRNA production in general; nevertheless, the evidence is that they are produced as molecules in excess of 15,000 bases in length with at most a few thousand extra bases in their primary transcript. 3. HEMOGLOBIN MRNA Hemoglobin mRNA is the most well-characterized single mRNA species from eukaryotic cells, having been completely or nearly completely sequenced from several species (4850).DNA containing hemoglobin-specific sequences has been purified in several ways. Reverse transcriptase was used first with hemoglobin mRNA to produce a DNA complementary to the mRNA (51). This so-called complementary DNA (cDNA) allows hybridization and direct measurement of newly labeled hemoglobin-specific RNA. Several laboratories report that in cells producing hemoglobin mRNA, the first hemoglobin-
TRANSCRIPTION UNITS FOR
mRNA
335
specific molecule (P-globin) to become labeled is a nuclear RNA molecule 2.5 to over 5 times as long as the hemoglobin mRNA ( 5 2 3 5 ) .In a detailed study, Bastos and Aviv (54)describe a hemoglobin-specific molecule of around 5000 nucleotides in length with a half-life less than 10 minutes, as well as a second hemoglobin-specific molecule of about 1500 nucleotides, which is also larger than the 9 S (700 nucleotides) hemoglobin mRNA that functions in the cytoplasm. The molecule of 1500 nucleotides has poly(A) attached to it. Others have observed hemoglobin-specific molecules of about 1500 nucleotides, but they have not observed any larger molecules (52, 56). Ross recently described a nuclear RNA containing sequences specific for the alpha chain of hemoglobin that was only slightly larger (100-200 nucleotides) than the cytoplasmic mRNA (56a).These results emphasize the importance of being able to examine newly labeled RNA within the first few minutes after labeling begins; after extensive periods of labeling, so much 9 S mRNA accumulates in both the cytoplasm and the nucleus that observation of a longer putative precursor molecule is difficult if not impossible. Nevertheless, it should be noted that Spohr st al. (57) had earlier shown, with painstaking preparation of completely denatured unlabeled nuclear RNA from duck cells, that a stable minority class of large RNA complementary to labeled cDNA is present. These molecules include a 5000-nucleotide species as well as some larger RNA. All the evidence therefore supports a hemoglobin primary transcript larger than the mRNA, at least in the range of 1500 for p-globin, shorter for a-globin, and no direct synthesis of the hemoglobin mRNA of 700 nucleotides has been detected.
4. DETECTION OF SPECIFICNUCLEARSEQUENCES BY RECOMBINANT DNA TECHNIQUES: IMMUNOGLOBULIN MRNA,THE RAT GROWTH HORMONE MRNA,AND CHINESEHAMSTER CELLSMRNAs The very powerful techniques of recombinant DNA research and the “cloning” and amplification of the recombinant DNA in bacteria allow the production of large quantities of DNA complementary to virtually any specific mRNA (58, 59). With purified DNA spepific for one mRNA, labeled, newly formed nuclear RNA complementary to the cloned DNA can be identified by RNA * DNA hybridization. Employing cloned DNA complementary to immunoglobulin “light” chain mRNA, Gilmore-Herbert and Wall (60)detected a nuclear RNA of about 9000 nucleoti,des containing the sequences of the immunoglobulin light-chain mRNA, which itself is about 1500nucleotides. Well over 90% of the immunoglobulin-specific sequences were in the large molecule, presumably the primary transcript, after a 5-minute label.
336
JAMES E. DARNELL, JR.
The immunoglobulin-specific nuclear RNA molecule also has a short (10-15 min) half-life and apparently undergoes cleavage to yield a 25 S molecule prior to becoming the final 14 S product. The same techniques have been used with cultured Chinese hamster cells to prepare a series of bacterial clones containing sequences complementary to randomly selected mRNAs (61). Sequences representing most of the rat growth hormone mRNA'were also cloned from partially purified mRNA (62). Nuclear RNA prepared after 5 minutes of labeling was hybridized to these cloned DNAs, revealing nuclear molecules from 3 to 5 times the length of the specific cytoplasmic mRNA from both rat and Chinese hainster genes.
5. HISTONEMRNA Melli et al. observed that histone RNA sequences are present in high-molecular-weight nuclear RNA greater than 14,000 nucleotides in length (63).These experiments rely on a similarity in sequence between all histone mRNA sequences, so that cloned sea urchin DNA containing the histone genes was used to hybridize HeLa cell histone-specific RNA. While these experiments did not show more than about 50% of the total histone-specific RNA in the largest form, there was a distinct correspondence between the average repeat length in histone DNA and the maximum size of the briefly labeled nuclear RNA containing histone sequences. It should be noted that while HeLa cell histone mRNA contains 5'-terminal cap I (64)it does not contain the 3'-terminal poly(A) (65),so that histone mRNA might not necessarily serve as the prototype of the average mammalian cell transcription unit. 6. ADENOVIRUSMRNA The most detailed information available about any eukaryotic transcriptiori units concerns virus-specific mRNA formation in adenovirus-2-infected HeLa cells. This is chiefly because of the availability of large quantities of pure virus DNA that can be subdivided with restriction enzymes to produce a very detailed physical map (8). In addition, a very high percentage of RNA synthesis in infected cells is virus-specific (66),and this allows the study of labeled nascent RNA chains so that the site of initiation of RNA synthesis can be mapped according to the following scheme: in an array of labeled nascent molecules, the shortest labeled RNA is promoter-proximal, while increasingly longer chains containing labeled termini are compIementary to an ordered array of DNA sites all the way to the terminus of the transcription unit (67).Experiments using this approach reveal at least four early transcription units of 2000 to -6000 or 7000 nucleotides
TRANSCRIPTION UNITS FOR
0
10
20
337
inRNA
30
40
50
60
70
80
90
100
I
to
fo
FIG.1. The transcription map of adenovirus type 2 (AD2). The genome of the AD2 virus is depicted as a linear molecule 100 units long. Each unit is 350 bases. 0 is at the left end of the strand, which is transcribed to the right ( r ) .The leftward transcribing strand ( 1 ) is therefore read from 100 + 0. The approximate map positions of early and late mRNA are shown on the I or r strand in dark blocks for late, light blocks for early (8). Pulse-labeled RNA was subjected to size separation and hybridized to various regions of the AD2 genome. Short, nascent chains were used to diagnose start sites for viral RNA synthesis (see text) 0- are early transcription units and O+ show the late transcription unit (68, 69).
(69-69b) that are distinct from one major late transcription unit of about 28,000 nucleotides (67-69) (Fig. 1).In addition to mapping by nascent-chain labeling, the major late transcription unit has also been defined by UV transcription mapping (70).In accord with the nascentlabel data, the entire right-hand 8 0 4 5 % of the Ad2 genome behaves as one transcription unit late in infection; UV irradiation affects the transcription of RNA complementary to each of a series of DNA fragments in exponentially increasing fashion beginning at about 1520% from the “left” end and extending to the “right” end of the DNA molecule.
IV. The Precursor Relationship of Large Nuclear Primary Transcripts to mRNA
A major aim of transcription unit definition is to determine which primary transcription units are responsible for mRNA production. Two kinds of experiments support the general conclusion that primary nuclear transcripts larger than mRNA are, in fact, precursors of the majority of shorter cytoplasmic mRNAs: (a) kinetic labeling studies for specific mRNA pathways; and (b) UV target-size analyses of specific and general cellular mRNAs. A. Kinetic Labeling Studies It is difficult to conclude unequivocally on the basis of kinetic labeling studies alone that the only means ofproducing a shorter mole-
338
JAMES E. DARNELL,JR.
cule is first to produce a longer one. There are two stages in the development of kinetic labeling evidence about precursor-product relationships that should be distinguished in reviewing recent work on RNA transcription. 1. When label first enters RNA, a putative mRNA precursor should contain all (or very close to all) the detectable radioactivity specific for that gene. As the label time increases, any intermediates between the precursor and the final product become labeled, and eventually the final product becomes the dominant labeled species. In experiments where cells were labeled for between 1and 10 minutes, a larger nuclear molecule containing all (or almost all) the labeled hemoglobin (52-56), immunoglobulin (60), specific Chinese hamster (61), rat growth-hormone (62), and late adenovirus mRNA sequences (68-70) have been observed. In the case of the nuclear RNA containing the “late” adenovirus mRNA sequences, the analysis has perhaps been most extensive; here the DNA fragment containing the initiation site for synthesis of a large nuclear molecule was identified and shown to be essentially the only fragment to hybridize very short nascent labeled chains. Thus, no individual RNA initiation events could be detected for any of the more than 10 Ad-Zspecific mRNAs (69) contained within the region spanned by the large transcript. Finally, substantial fractions (20-30%) of the primary transcript that represents mRNA exits to the cytoplasm (71). At least in this case, the question of a possible long transcript not productive of mRNA and shorter transcripts that are the mRNA precursor seems to be ruled out, and RNA processing must occur. 2. The second step in developing evidence of a precursor-product relationship between two labeled molecules is the demonstration of a transfer of label from the putative precursor to the final product. Even with the ribosomal RNA or tRNA, generally agreed to arise by a processing of precursor molecules (72), such experiments are at best only partially successful. The difficulty is simply that 100%of the incorporated radioactivity is never in a single molecular species, and metabolic “chases,” designed to stop further labeling and allow processing, are never instantaneously and completely effective. Nevertheless, it should be noted that Bastos and Aviv (54) and Ross (52) both report results consistent with a chase from a larger hemoglobin precursor molecule to the smaller hemoglobin mRNA. B. UV Evidence for the Size of Primary Transcripts for mRNA Production
A most useful addition to experiments aimed at determining the obligatory role of larger precursors in the formation of mRNA is UV
TRANSCRIPTION UNITS FOR
inRNA
339
transcription analysis (26,35,36).This technique can be used not only to define the primary transcript sizes of labeled nuclear molecules, but also to detemiine whether the UV target size of niRNA corresponds to the target size of the putative precursor. UV target-size experiments have been performed for both early and late Ad2 mRNAs. The mRNAs that derived from the large, late Ad-2 transcription unit all have at least as great a sensitivity to UV as the corresponding sequence in the large late primary transcript (70, 7 3 ) . One late Ad2 mRNA derived from a different short transcription unit (73)and all the early Ad2 mRNAs that arise from a series of shorter transcription units (45, 7 4 ) have much smaller target sizes than do the later mRNAs derived from the large late transcript. These results lend strong support to the conclusion that the primary transcript of about 28,000 nucleotides is the obligatory precursor in the formation of most late Ad2 mRNA. Since the study of early and late Ad2 mRNA formation indicates that the transcription unit size determines the UV target size for inRNA formation, UV transcription analysis was applied to cell mRNA formation. The UV target size for the cellular mRNA was estimated by comparison with rRNA, hnRNA, and early Ad2 mRNA fomiation. The average size of the UV target for mRNA formation was found to be at least three times the size of the mRNA itself (1500nucleotides) in both HeLa cells (45) and mouse cells (75). Thus the UV target size for mRNA is very similar to the average size transcription unit for hnRNA (5000 nucleotides) determined by both pulse label and UV studies. Recent experiments with UV transcription mapping of the mRNA for immunoglobulin light chains add further strength to these conclusions (R. Wall, personal communication). As described earlier, a 9000.nucleotide light-chain-specific nuclear RNA molecule has been identified by hybridization of pulse-labeled RNA to cloned DNA containing mRNA sequences specific for immunoglobulin light chains (60). The UV target size of the nuclear RNA complementary to this cloned DNA was estimated by comparison to 28 S and 18 S RNA to be about 9000 nucleotides long, in agreement with the analysis of pulselabeled RNA. In addition, the labeling of cytoplasmic mRNA sequences complementary to the cloned DNA had a very similar UV sensitivity (60a).
V. Recent Evidence on the Formation of mRNA from Primary Transcripts
The general conclusion from all the experiments reviewed thus far in this paper is that nonribosomal primary transcripts in eukaryotic cells range in size from 2000 to 20,000 nucleotides, the majority being
340
JAMES E. DARNELL, JR.
substantially longer than mRNA. While some mRNA molecules, e.g., silk fibroin mRNA and the mRNA for a salivary gland protein in Chironomus tentans may derive from primary transcripts no longer or only 10%longer than the finished mRNA, the majority of mRNAs are probably derived from transcription units several times larger than the mRNA itself. This conclusion, of course, does not comment on the problems of whether every hnRNA gives rise to mRNA or whether some hnRNAs may give rise to more than one variety of mRNA. Such “processing” to yield mRNA raises many questions, especially in view of the eventual necessity of modifying both the 5’ and 3’ ends of an mRNA (11,12,40).2Over two-thirds of both the 5’ methylated cap 1 structure (40,41,75a) and poly(A) (12, 76) in the nucleus exist in hnRNA molecules larger than the cytoplasmic mRNA. Furthermore, the evidence suggests the possibility that both cap (77) and poly(A) (78, 79) are conserved when nuclear molecules move to the cytoplasm. An earlier suggestion that nuclear poly(A) largely turned -over in the nucleus and therefore failed to reach the cytoplasm (79a) was based on the erroneous assumptions that all mRNA in mammalian cells had a very long half-life and that there was a uniform exit time from the nucleus to the cytoplasm for all the mRNAs. It is now clear that some mammalian cell mRNAs have a turnover time of less than 1-2 hours compared to an average of &lo hours (78,79,79b), and that some mRNAs exit to the cytoplasm almost instantly whereas others may take well over an hour (79b, 79c). This variability in exit time and turnover rate greatly complicates any analysis designed to determine the precursor-product relationship of the total nuclear and cytoplasmic molecules based on labeling kinetics (79). However, if the mRNA is taken to be approximately an equal mixture of long- and short-lived molecules, then it appears possible that both the nuclear caps (77) and the nuclear poly(A) (79) are conserved. The conclusion that both ends of some hnRNA molecules might be conserved prompted the speculation that each end of some primary transcripts might yield mRNAs. Recently, a most astonishing discovery about RNA processing in Ad2 infection that holds great’promise for explaining how hnRNA is converted into mRNA was made. As previously discussed, late during the course of adenovirus type 2 (Ad2) infection of human cells, at least 10 different adenovirus mRNAs that map in the right-hand 7585% of the adenovirus genome are made (8) (Fig. 2). Berget et al. (80) and Chow et al. (81) used E M visualization of mRNA * DNA hybrids, and Gelinas and Roberts (82) and Klessig (83)used RNA oligonucleotide analysis to show that the 5‘ end of each of these late Ad2 mRNAs consists of about 200 nucleotides that are not complementary to the
- -- - -
mRNA Products I
.
0
I
10
I
341
mRNA
TRANSCRIPTION UNITS FOR
* - -
L
8
16 19
,
I
26 30
,
40
,
,
,
50
,
60
,
.
,
70
,
80
,
.
90
,
100
Primary tronscript
1 1 1
Cleovoge ot any X
Nucleor Poly ( A ) +
-
Csp I sp I cr cr un I I I 1 I I
I
I ) (
- -
C Cop I Leoder
cr
Coding region
2Untronsloted region
mRNA
FIG. 2. Biosynthesis of late adenovirus type 2 mRNA. The figure shows the late mRNA products that derive from the major late transcription unit that extends from approximately 16 to 90 on the physical map (68-70, 73). There are five 3’ coterminal regions, which end at 38, 50, 62, 79, and 91 (71,103,104).The coding region (cv) at the bottom of the diagram is represented twice because two, three, or four configurations are possible within the various 3’ coterminal groups of mRNAs.
sequences that adjoin the major portion of each of the late mRNAs. Rather, all the 5’ ends are complementary to the same three, distinct distant sites on the genome (16.6,19.6,and 26.6 on the physical map of the genome). In rapid order, gaps in the DNA encoding other mRNAs have also been demonstrated: early adenovirus mRNA (84),SV40 mRNA (85-87), RNA tumor virus mRNAs [from integrated viral DNA (87a), hemoglobin mRNA (88, 89), ovalbumin mFWA (90, 91), and immunoglobin mRNA (92,93)] were also shown by either E M or biochemical analysis to derive from noncontiguous sites in the DNA. Explanations that might be offered for how an mRNA could be assembled from DNA containing gaps between the mRNA sequences fall into two categories: (1)RNA-RNA ligation; or (2) mechanisms that do not involve RNA-RNA ligation (Fig. 3). This simple division reflects the profound sense of reservation felt by many biologists in accepting the possibility that enzymic joining of RNA molecules is a central event in eukaryotic mRNA biosynthesis. The proposed RNARNA ligation (80-83)involves transcription of a region of DNA including all the sequences necessary to make an mRNA molecule; the regions that are not conserved in the mRNA are removed by enzymic
342
JAMES E. DARNELL, JR.
a b c d e f g h i j k l m n o p q r s t u v w x y z Transcribe
/
I
Recombined Template
Folded Template
‘ k l w
transcribe
-j-k-
b”
1
I-W-~, Y\
0’
RNA : RNA Ligation
“Jumping“ Polymerase
DNA Recombinotion
FIG. 3. Possible mechanisms of mRNA formation. The biosynthesis of an mRNA containing sequences not contiguous in the DNA could theoretically occur in the three ways depicted in the figure. Present evidence (see text) strongly favors RNA-RNA ligation or splicing.
cleavage, and the regions that are conserved become rejoined“spliced” or “ligated.” The alternatives to RNA-RNA ligation would be either (1)transcription of the DNA in the “correct” places, that is, the RNA polymerase “jumps” to transcribe noncontiguous regions, or (2) the DNA might be recombined before transcription, excluding the gaps between the mRNA-specific regions (Fig. 3). All the presently available evidence favors RNA-RNA ligation in mRNA formation, the strongest present evidence coming from the study of Ad2 mRNA formation late in infection. The great majority, if not all, the late Ad2 primary transcripts contain all the sequences (both the noncontiguous “leader” sequences and the main body of the mRNA) for all the previously mentioned groups of late Ad2 mRNAs (68-70). The start site for RNA synthesis (69) appears to contain the cap-containing region present in each mRNA (82). Moreover about
TRANSCRIPTION UNITS FOR
mRNA
343
20% of each of five different regions of the late primary transcript is conserved in transport from the nucleus to the cytoplasm (71 ). Thus it seems that each large transcript gives rise to one of a possible five mRNAs (or one of a possible 5 groups of 3’ coterminal mRNAs). Since all (or at least the majority) of the late mRNAs that map between 16and 98 on the Ad2 genome contain the three spliced regions from 16.6,19.6, and 26.5 (71, 81 ), the RNA (Fig. 2) of the primary transcript must not only be broken (as concluded in the previous section) but also rejoined during the biogenesis of late adenovirus mRNA. An additional 4 or 5 Ad2 transcription units that function early in infection have been identified by the hybridization of pulse-labeled RNA to DNA fragments derived with restriction endonucleases (69) and by UV transcription mapping (74 ). While these early transcription units are short compared to the long, late transcription unit, they produce primary transcripts that are clearly larger than the mRNAs derived from these same regions of the genome (94, 95). All the early mRNAs contain sequences that are noncontiguous in the Ad2 DNA, but all the sequences within one mRNA lie within one transcription unit (84). Thus there is no evidence of ligation of sequences between transcriptional units. Similar studies on the synthesis and processing of hemoglobin (pchain) mRNA and immunoglobulin (light-chain) mRNA also suggest that RNA-RNA ligation is involved during mRNA biogenesis. As described previously, the pulse-labeled nuclear RNA containing either hemoglobin sequences ( 5 2 5 6 ) or immunoglobulin sequences (GO) is larger than the finished mRNA, and in both cases the finished mRNA contains sequences that are not contiguous in the DNA (88,89,92,93). Also, late in SV40 infection, the primary virus-specific transcripts (96) appear to include both sequences that are discarded and noncontiguous sequences that are preserved in the manufacture of mRNA (85, 86). In contrast to the evidence cited above, there is at the moment no positive evidence in favor of either a “jumping polymerase” or DNA rearrangement before transcription to explain the derivation of mRNA from noncontiguous sites in the DNA of both cells that are pro,ducing these proteins as well as several other cell types not specialized to produce these proteins (88,90,91). Even in the case of myeloma cells making immunoglobulin light chains, an initial recombination even brought V (variable) and C (constant) regions of the gene closer together than in embryonic DNA, but left the V and C regions 1250 nucleotides apart (92, 93). Therefore, there is no experimental support for DNA rearrange-
344
JAMES E. DARNELL, JR.
ment bringing the mRNA-specific regions of a transcriptional unit into contiguity, and no evidence of a “jumping polymerase” has ever been described in vivo or in vitro, Finally, either of these two mechanisms would result in primary RNA transcripts equal in size to mRNA, whereas in all the cases discussed above, no such primary transcripts have been found. [See Note Added in Proof, p. 353.1
V1. Models for the Regulation of Eukaryotic Gene Expression Based on what is now known about eukaryotic transcription and mRNA structure, three types of problems that do not exist in prokaryotes must be addressed in considering models of eukaryotic gene control. (Needless to say, regulation of the start of transcription is implied to underlie any other models for gene regulation. Without the beginning of RNA chains there would be nothing to regulate. What we wish to consider here are possible decisions that are made during or after an RNA chain has been started.) 1. Are primary transcripts containing potential mRNA produced that are not processed? Are “process vs. discard” decisions necessary? 2. Some transcription units produce primary transcripts that might give rise to two or more different mRNAs. Is the ratio of these different mRNAs always constant, or can regulated changes in processing occur? This would be selective processing. 3. Is the regulatory eyent in either the process vs. discard or the selective processing case made after the completion of transcription or during the act of transcription? These problems are first be described more fully, then models of transcription units illustrating the problems are presented, followed by reference to specific cases in AD2 and SV40 mRNA formation that have the properties of the models. A. Process vs. Discard Decisions For many years the possibility has been discussed that process vs. discard decisions might be occurring in eukaryotic nuclear RNA synthesis. The initial basis for these earlier speculations was the wellestablished fact that the rate of nuclear RNA synthesis in cultured eukaryotic cells greatly exceeds the amount of RNA that appears in the cytoplasm (72, 97). Even if the average size of the primary nuclear transcripts for mRNA is five times the length of the finished mRNA (43, 95), it still appears possible, but not proven, that more nuclear transcripts are made than do, in fact, contribute to cytoplasmic mRNA.
TRANSCRIPTION UNITS FOR
InRNA
34s
Another recent finding with specific mRNA synthesis raises the possibility of extra, nonprocessed transcripts. In cells where the synthesis of specialized proteins (hemoglobin, immunoglobulin, or ovalbumin) constitute 10-20% of total protein synthesis, only about 0.1% of the total nuclear RNA or about 0.2%of nonribosomal RNA is represented Also, it has been by sequences for the specialized protein (52,53,60). reported that nuclear RNA, which is more complex than mRNA, i.e., has a greater sequence diversity (18,19a), may be similar in various cells that have a distinguishably different mRNA makeup (17, 18). Thus earlier suggestions about posttranscriptional regulation in eukaryotic cells focused on the possibility that, after synthesis, some transcripts were selected for processing to mRNA and others were discarded. This hypothetical process vs. discard type of selection must still be borne in mind as a possibility to prove or disprove in the future.
B. Selective Processing Decisions The more recent work with Ad2 and SV40 virus provides cause to propose more explicit models of mRNA formation that could require regulated decisions in addition to the control of initiation of transcription and/or process vs. discard mechanisms. Three hypothetical models of eukaryotic transcription units that function in the production of mRNA are shown in Fig. 4. The “simplest” of these models would require all the posttranscriptional modifications previously discussed-cap addition at the 5’ terminus, poly(A) addition at an internal site or at the 3’ end of primary transcript; and, finally, cleavage and ligation of all RNA sections necessary to make a functioning mRNA. It should be noted that all mRNAs in eukaryotic cells would, of course, not have to come through this pathway; for example, histone genes do not have poly(A) (65)and may, at least in Drosophila cells, be direct transcripts (D. S. Hogness, personal communication), not processed from larger precursors. The justification for calling model I a “simple” transcription unit is that only one mRNA could arise from transcription of such a region of DNA. All the necessary biochemical steps might be envisioned as automatic once the decision was made to synthesize this primary transcript. If so, such a hypothetical transcription unit would be exclusively under transcriptional control just as a bacterial gene is under transcriptional control. Failure to perform the correct processing steps would constitute a process vs. discard decision as described above. The second model (11)presents a “complex” transcription unit containing one cap site and two (three or more could obviously also exist) poly(A)-addition sites. In this case it is clear that aselective processing
346
JAMES E. DARNELL, JR.
“Simple” Eukoryotic Transcription Unit i = initiation
t
I
t =termination
I
Complex Transcription Unil: Two 3’ Ends
t
I
Complex Transcription Unit: Two 3’ Coterminal mRNAs I
-
t
0
or
C
-COP
I -Leader
Cr Coding g!Untronsloted -region
FIG. 4. Models of eukaryotic transcription units. Three possible models for transcription units. “Simple” indicates only 1 possible mRNA; “complex” indicates two or more possible mRNAs from same transcription unit containing either coding regions a or b.
TRANSCRIPTION UNITS FOR
IiiRNA
347
choice must be made, since only one mRNA can emerge successfully from each primary RNA transcript. However, if the proportion selected of the two possible mRNAs were invariant in all cells where this transcription unit functioned, then again the transcription unit could be regulated entirely by transcriptional or process vs. discurd control. The third model (111)presents a different “complex” configuration of multiple mRNAs within a transcription unit, but is subject to the same logic as model I1 with respect to control. Here, two inRNA products are possible from the same transcription unit, but they are 3’ coterminal and are different because splicing would remove different portions of the primary transcript (a is shorter than b in the diagram). In such a situation, the 5’-most portion of the coding sequence would most probably function in the longer mRNA and a new initiation site for protein synthesis would be exposed in the smaller mRNA, as is the case in many partially overlapping animal and plant virus mRNAs (98-102). In model 111again, the opportunity exists for regulatory decisions that would select more a than b mRNA or vice versa, but again the same proportion of a and h might always be chosen, making the selective processing of a or b an automatic, not a regulatory, decision. Regardless of whether the proportions of a or b are invariant in Models I1 or 111, selective processing must occur for each individual transcript. This choice could be made in either of two general ways: (1) completed transcripts could be shuttled into the a or b processing pathway, in which case the selection of a or b and any regulated variation in this choice would be posttranscriptional; ( 2 )the transcription complex might include the necessary factors (RNA polymerase subunits plus processing enzymes) to decide whether the a or h mRNA would result from any particular transit of the transcription compIex. In this case the choice between a or b or any regulated variation in this choice would remain at the transcriptional level. In any event, proteins that recognize sites in RNA would appear necessary.
C. Present Knowledge about Transcription Unit Design What is known at present that fits any of these models for the design of transcription units and the manufacture of mRNA? Transcription units similar to both models I1 and I11 function in the production of adenovirus mRNA (71, 103-104a) and model I11 occurs in SV40 (85, 86, 105) and retrovirus mRNA formation (87q 98,101). The large late AD2 transcript gives rise to five groups of mRNAs. The mRNAs within each group are 3’ coterminal (71, 103-104a) and contain the common 200 nucleotides at the 5’ end (81). However,
348
JAMES E. DARNELL, JR.
within a 3’ coterminal group, variable amounts of RNA are removed by splicing (Fig. 2). It appears very likely that each primary transcript can give rise to only one mRNA, and the percentage varies slightly from one 3’ coterminal group to another. Probably within a group there is also variation as to the frequency of the various spliced-out forms, If, however, the proportion among (or within) groups of mRNAs proved to be an invariant result for the long late AD2 primary transcript (for example throughout the period of infection when the long transcript is produced), then no differential selection would occur in the formation of late Ad2 mRNAs. An interesting point in considering the design and function of AD2 transcription units is that one of the early AD2 transcription units (-75-85 on the genome) lies entirely within the region of the large late transcripts (Fig. 2) (69).This early transcription unit produces an early mRNA but no mRNA is produced late from this region even though the appropriate sequences are part of the large late transcript (8,71; also J. Nevins, unpublished observations), Thus only if the RNA initiation site is the correct one does this early transcription unit produce mRNA. The only regulatory decision in making this AD2 mRNA might therefore remain a transcriptional decision-starting at the correct site. Both Ad2 and SV40 transcription units exhibit the features of the third model of a transcriptional unit-3’ coterminal mRNAs with splicing of the 5’ cap and leader sequence(s)to the presumed variable start sites for protein synthesis. For example, the late SV40 mRNAs code for three proteins: VP1,2, and 3. All three mRNAs terminate at the same region, but the 5‘ start sites for protein synthesis are staggered by several hundred nucleotides (86,85,105). A slight variation in model I11 is proposed for the early SV40 mRNAs (87,106). The “early” SV40 proteins, the two so-called “T” antigens, are probably made from mRNAs that are 3‘ coterminal, but protein termination signals stop the little “T” antigen after -150 amino acids, and it seems likely that the termination signals are spliced out in some mRNA molecules resulting in the translation of a “big T” antigen of about 800-900 amino acids, including those at the N-terminal region of “little T.” Thus the mRNAs for “big” and “little” “T” are probably 5’ and 3‘ coterminal and utilize the same start site for protein synthesis, but one has a short gap in the middle that removes termination signals for translation (105). Each of the early Ad2 transcriptional units has been reported to produce mRNAs that are spliced in at least two ways (84) so at the moment there is no certain candidate as an example of model I, i.e., a transcription unit that can produce only one mRNA, although it would of course be very surprising if no such transcription unit existed.
TRANSCRIPTION UNITS FOR
mRNA
349
In the AD2 and SV40 genomes, the amount of DNA is limited and all the genes function within a relatively short time after infection. Thus multiple forms of mRNA molecules arising from one transcription unit (models I1 and 111) can be readily detected. Trying to assess whether cell transcription units will prove to be “complex,” that is, to produce more than one mRNA, or “simple,” i.e., each produces only a single mRNA, will be a very difficult task. “Complex” genes might exist, but function at different times in development or in different tissues to produce different mRNA products, rendering experimental detection of multiple mRNA from the same DNA segment much more difficult. Such possible developmental variations in mRNA production could be regulated by changing transcription unit boundaries within the same region of DNA, as is the case for early and late AD-2 mRNAs in the 75-85 region of the AD2 genome. A single cellular transcriptional unit that might produce different mRNA molecules during development is the region of DNA that encodes (in order 5’ + 3’) the gamma, delta, and beta polypeptides of human hemoglobin (107). The DNA for these polypeptides has been inferred to be adjoining because certain individuals produce a fusion product of two of the neighboring peptides (208,109). In other individuals with the beta region deleted, the gamma region continues to function [the so-called hereditary persistence of fetal hemoglobin (107)l.These genetic changes in globin chains have previously been considered to be viable genetic deletions of DNA; of course, they might represent, as pointed out by Jeffreys and Flavell (88),different splicing events in mRNA production. At any rate, it will be most interesting to see whether the primary transcript from this region of human chromosomes might contain information for all three polypeptide chains.
VII. Conclusion The newly understood developments in mRNA formation make it possible to define in molecular terms that are not wholly speculative, how and at what level eukaryotic gene expression may be regulated. Moreover, since individual primary transcripts can be measured, experiments can now settle whether process vs. discard decisions occur or whether differential use of multiple mRNAs of complex primary transcripts occurs. Once the manner of regulation is settled for several eukaryotic genes a much more coherent plan to search for regulatory proteins can be designed. Success in the identification and mode of action of regulatory proteins for eukaryotic genes is therefore not too much to hope for in the next few years.
350
JAMES E. DARNELL, JR.
ACKNOWLEDGMENTS The author wishes to give special thanks to Dr. J. Nevins among other colleagues for many useful conversations that underlie the ideas of transcriptional unit design and function described here.
REFERENCES 1. F. Jacob and F. Monod,JMB 3, 318 (1961).
E. Englesberg, D. Sheppard, C. Squires and F. Meronik, Jr.,JMB 43, 281 (1969). J. Greenblatt and R. Schlief, Nature NB 273, 166 (1971). G. Zubay, D. Schwartz and J. Beckwith, PNAS 66,104 (1970). W. Gilbert, N. Maizels and A. Maxarn, CSHSQB 38, 845 (1973). T. Maniatis, M. Ptashne, K. Backman, D. Kleid, S. Flashman, A. Jeffrey and R. Maurer, Cell 5, 109 (1975). 7. B. Daneholt, Cell 4,1(1975). 8. J . Flint, Cell 10, 153 (1977). 9. N. H. Acheson, Cell 8, l(1976). 10. 0. L. Miller, Jr., B. R. Beatty, B. A. Hamkalo and C. A. Thomas, Jr., CSHSQB 35, 505 (1970). 1Oa. K. A. Jacobs, V. Shen and D. Schlessinger, PNAS 75,158 (1978). 11. A. J. Shatkin, Cell 9, 645 (1976). 12. J. E. Darnell, W. Jelinek and G. Molloy, Science 181, 1215 (1973). 13. K. Scherrer and L. Marcaud,J. Cell. Physiol. 72, Suppl., 181, 212 (1968). 14. S. P. Georgiev,J. Theor. Biol. 25, 473 (1969). 15. B. Lewin, Cell 4, 11(1975). 16. B. Lewin, Cell 4, 77 (1975). 17. N. D. Hastie and J. 0. Bishop, Cell 9, 761 (1976). 18. J. A. Bantle and W. E. Hahn, Cell 8, 139 (1976). 19. A. E. Sippel, B. Groner, N. Hynes and G. Schute, EJB 77, 141(1977). 1%. D. M. Chikaraishi, S. S. Deeb and N. Sueoka, Cell 13,111 (1978). 20. K. C. Kleene and T. Humphreys, Cell 12, 143 (1977). 21. S. A. Galau, W. H. Klein, M. M. Davis, B. J. Wold, R. J. Britten and E. H. Davidson, Cell 7, 487 (1976). 22. M. Melli, G. Spinelli and E. Arnold, Cell 12, 167 (1977). 23. K. Scherrer and J. E. Darnell, BBRC 7, 486 (1962). 24. K. Scherrer, H. Latharn and J. E. Darnell, PNAS 49, 240 (1963). 25. 0. L. Miller, Jr. and B. R. Beatty, Science 164, 955 (1969). 26. P. B. Hackett and W. SauerbierJMB 91,235 (1975). 27. R. H. Burdon, and A. E. Clason,JMB 39, 113 (1969). 28. D. Bernhardt and J. E. Darnel1,JMB 42, 43 (1969). 29. D. B. Mowshowitz,JMB 50, 143 (1970). 30. P. M. Lizzardi, This Series 19, 301 (1976). 31. J. E. Darnell, R. Evans, N. Fraser, S. Goldberg, J. Nevins, M. Salditt-Georgieff, H. Schwartz, J. Weber and E. Ziff, CSHSQB 42 in press (1977). 32. S. L. McKnight, N. L. Sullivan and 0. L. Miller, Jr., This Series. 19, 313 (1976). 33. S. L. McKnight and 0. L. Miller, Jr.. Cell 8, 305 (1976). 34. V. E. Foe, L. E. Wilinson and C. D. Laird, Cell 9, 131 (1976). 35. W. Sauerbier, R. L. Millette and P. B. Hackett, BBA 209, 368 (1970). 36. A. R. Brautigam and W. Sauerbier,J. Virol. 12, 882 (1973). 2. 3. 4. 5. 6.
TRANSCRIPTION UNITS FOR
mRNA
351
G. Georgiev and V. L. Mantieva, BBA 61, 153 (1962). R. H. Herman, J. G. Williams and S. Penman, Cell 7, 429 (1976). A. E. Sippel, B. Groner, N. Hynes and G. Schutz, FJB 77, 153 (1977). R. P. Perry, E. Bard, B. D. Hanes, D. E. Kelley and U. Schibler, This Series 19,275 (1976). 41. R. P. Perry, D. E. Kelley, K. Frederici and F. R o m a n , CeEE 6, 13 (1975). 42. M. Salditt-Georgieff, W. Jelinek, J. E. Damell, H. Furuichi, M. Morgan and A. Shatkin, Cell 7, 227 (1976). 43. E. Derman, S. Goldberg and J. E. Darnell, Cell 9, 465 (1976). 44. R. Giorno and W. Sauerbier, Cell 9, 775 (1976). 45. S. Goldberg, H. Schwartz and J. E. Darnell, PNAS 74, 4520 (1977). 46. Y. Suzuki and D. D. Brown,JMB 63, 409 (1972). 47. P. M. Lizzardi, R. Williamson and D. D. Brown, Cell 4, 199 (1976). 48. F. E. Barelle, Cell 10, 549 (1977). 49. N. J. Proudfoot, Cell 10, 559 (1977). 50. A. Efstratiadis, F. C. Kafatos and T. Maniatis, Cell 10, 571 (1977). 51. D. L. Kacian, S. Spiegelman, A. Bank, M. Terada, J. Metafora, L. Dow and P. A. Marks, Nature NB 235, 167 (1972). 52. J. Ross,JMB 106, 403 (1976). 53. P. J. Curtiss and C. Weissman,JMB 106, 1061 (1976). 54. R. N. Bastos and H. Aviv, Cell 11, 641 (1977). 55. S. P. Kwan, T. G. Wood and J. B. Lingrel, PNAS 74, 178 (1977). 56. P. J. Curtiss, N. Mantel, J. van den Berg and C . Weissman, PNAS 74, 3184 (1977). 560. J. Ross and D. A. Knecht,]MB 119, l(1978). 57. G. Spohr, T. Iniaizumi and K. Schemer, PNAS 71, 5009 (1974). 58. P. C. Wensink, D. J. Finnegan, J. E. Donelson and D. S. Hogness,Cell 0,315 (1974). 59. R. Higuchi, G. V. Paddock, R. Wall and W. Salser, PNAS 73, 3146 (1976). 60. M. Gilmore-Hebeit and R. Wall, PNAS 75, 342 (1978). 600. M. Gilniore-Hebert, K. Hercules, M. Komaromy, and R. Wall, PNAS 75, in press. 61. M. Harpold, R. Evans, M. Salditt-Georgieff and J . E. Damell, in preparation. 62. M. Harpold, P. R. Dobner, R. Evans and F. C. Bancroft, NARes in press (1978). 63. M. Melli, G. Spinelli, H. Wyssling and E. Arnold, Cell 11, 651 (1977). 64. B. Moss, A. Gershowitz, L. A. Webber and C. Baglioni, Cell 10, 113 (1977). 65. M. Adesnik and J. E. Darnel1,JMB 67, 397 (1972). 66. L. Philipson, U. Pettersson, U. Lindberg, C. Tibbetts, B. Venstrom and T. Persson, C S H S Q B 39, 447 (1974). 67. S. Bachenheimer and J. E. Darnell, PNAS 72, 4445 (1975). 68. J. Weber, W. Jelinek and J. E . Darnell, Cell 10, 612 (1977). 69. R. Evans, N. W. Fraser, E. Ziff, J. Weber, M. Wilson and J. E. Darnell, CeEf 12, 133 (1977). 69a. E. A. Craig and H. J. Raskas, Cell 8, 205 (1978). 6911. S. Bachenheimer,j. Virnl. 22, 577 (1977). 70. S. Goldberg, J. Weber and J. E. Darnell, Cell 10, 617 (1977). 71. J. Nevins and J . E. Darnell,]. Virol. 25, 811 (1978). 72. J. E. Darnell, Boct. Rev. 32, 262 (1968). 73. S. Goldberg, J. Nevins and J, E. Damel1,J. Virof. 25, 806 (1978). 74. A. W. Berk and P. A. Sharp, Cell 12, 45 (1977). 75. R. Giorno and W. Sauerbier, PNAS 75, in press (1978). 75n. R. F. Fernandez-Munoz, U. Lavi and J. E. Darnell, NARes 4, 3357 (1977). 76. J. E. Damell, W. Jelinek, L. Puckett, E. Derman and S. Bachenheiiner, Symp. Sac. D e n B i d . 34, 53-74 (1976). 37. 38. 39. 40.
352
JAMES E. DARNELL, JR.
77. R. P. Perry and D. E. Kelley, Cell 8, 433 (1977). 78. L. Puckett, S. Chambers and J. E. Darnell, PNAS 72, 389 (1975). 79. L. Puckett and J. E. Darnell,]. Cell. Phys. 90, 521 (1977). 790. R. P. Perry, D. E. Kelley and J. LaTorre,]MB 82, 315 (1974). 79b. M. Wilson, S. Sawicki, M. Salditt-Georgieff and J. E. Darnell,]. Virol. 25, 97 (1978). 79c. R. Herman and S. Penman, Bchem 16, 3460 (1977). 80. S. N. Berget, C. Moore and P. A. Sharp, PNAS 74, 3171 (1977). 81. L. T. Chow, R. E. Gelinas, T. R. Broker and R. J. Roberts, Cell 12, 10977). 82. R. E. Gelinas and R. J. Roberts, Cell 11, 533 (1977). 83. D. F. Klessig, Cell 12, 9 (1977). 84. G. R. Kitchingman, S. P. Lai and H. Westphal, PNAS 74, 4392 (1977). 85. Y. Aloni, 0. Dhar, M. Laub, M. Horowitz and G. Khoury, PNAS 74, 3686 (1977). 86. M.-T. Hsu and J. Ford, PNAS 74, 4982 (1977). 87. A. Berk and P. A. Sharp, PNAS 75, 1274 (1978). 87a. E. Rothenberg, D. J. Donoghue and D. Baltimore, Cell 13,435 (1978). 88. A. J. Jeffreys and R. A. Flavell, Cell 12, 1097 (1977). 89. S. Tilghman, D. C. Tiemier, J. G. Seidman, B. M. Peterlin, M. Sullivan, J. V. Maize1 and P. Leder, PNAS 78, 125 (1978). 90. R. Breathnach, J. L. Mandel and P. Chambon, Nuture 270, 314 (1977). 91. R. Weinstock, R. Sweet, M. Weiss, H. Cedar and R. Axel, PNAS 75, 1299 (1978). 92. C. Brack and J. Tonegawa, PNAS 74, 5652 (1977). 93. S. Tonegawa, A. M. Maxam, R. Tizard, 0. Bernhard and W. Gilbert, PNAS 75,1485 (1978). 95. E. A. Craig and H. J. Raskas, Cell 8, 205 (1976). 96. J. Ford and M.-T. Hsu,]. Virol.,in press (1978). 97, J. Lengyel and S. Penman, Cell 5,281 (1975). 98. A. L. J. Gielkens, D. Van Zaane, H. P. J. Bloemers, and H. Bloemendal, PNAS 73, 356 (1976). 99. D. T. Simmons and J. H. Strauss,JMB 86, 397 (1974). 100. S. I. T. Kennedy, ]ME 108, 491 (1976). 101. D. W. Stacey, V. G. Allfrey and H. Hanafusa, PNAS 74, 1614 (1978). 102. P. Kaesberg, This Series 19, 465 (1976). 103. M. McGrogan and H. J. Raskas, PNAS 75, 625 (1978). 104. E. Ziff and p. Fraser, 1. Virol. 25, 897 (1978). 1 0 4 ~ N. . Fraser and E. Ziff,JMB 121, in press (1978). 105. V. B. Reddy, B. Thimmeppaya, R. Dhar, K. N. Subramanian, B. Zain, J. Pan, M. L. Celman and S. M. Weissman, Science, in press. 106. L. V. Crawford, C. N. Colec, A. E. Smith, E. Paucha, P. Tegtmeyer, K. Rundell and P. Berg, PNAS 75,117 (1978). 107. B. G. Forget, D. G. Hilman, H. Lazarus, F. Barell, E. J. Benz, Jr., C. T. Caskey, T. H. J. Housman, W. A. Schroeder and D. Housman, Cell 7,323 (1976). 108. A. MotuIsky, Ann. N.Y. Acad. Sci. 241, 7 (1974). 109. H. Lehman, and R. G. Huntsman, in “The Metabolic Basis of Inherited Disease” J. B. Stanbury, J. B. Wyngaarden, and D. S. Frederickson, eds.), 3rd eds., pp. 1398-1431. McGraw-Hill, New York, 1972. 110. G. Knapp, J. S. Beckmann, P. F. Johnson, S. A. Fuhrman, and J. Abelson, Celt 14, 221 (1978). 111. J. N. Blanchard, J. Weber, W. R. Jelinek, and J. E. Darnell, PNAS 75, Dec. (1978).
TRANSCRIPTION UNITS FOR
mRNA
353
NOTE ADDEDIN PROOF Two cases of RNA-RNA splicing in cell-free extracts have been reported. Knapp et ul. (110) described the removal by extracts of yeast cell of 15 nucleotides from the middle of two different pre-tRNA molecules and the reunion of the remaining fragments to produce tRNA. Blanchard et uZ. ( 1 1 1 ) showed that a 5000 nucleotide AD-2 specific nuclear RNA molecule labeledin d u o would be converted within isolated nuclei to the size of a specific mRNA of 2000 nucleotides that retained sequences from both ends of the 5000 nucleotide region.
This Page Intentionally Left Blank
Subject Index A Adenosine deaminase, inhibitors of, 245-248 Aminoacyl-tRNA binding to ribosomes, 46 (EF-Tu) dependent, 51-57 formation of (EF-Tu).GTP complexes, 47-51 transacylation and, 57-59 Antibiotics adenosine deaminase inhibitors, 245248 cell-wall synthesis inhibitors, 249266 cyclic-AMP phosphodiesterase and, 269-270 DNA synthesis inhibitors, 239-245 DNA virus inhibitors, 239-245 as hyperemic agents, 269 as hypoesthetic agents, 269 immune response effectors and, 245248 protein synthesis inhibitors, 197-209 purine and pyrimidine interconversion and, 266-268 RNA synthesis inhibitors, 209-239 RNA virus inhibitors, 239-245 Antifungal agents, antibiotic, 249-266
B
biological effects of platinum complexes bacterial filament formation, 85-86 induction of lysogeny, 86-87 mutagenesis, 87 reaction with viruses and transforming DNA, 87-88 depurination and depyrimidination of, 173-178 interaction with platinum complexes, in uitro, 104-109 in uiuo, 94-103 models for excision-repair base excision-repair, 140-143 classification of lesions, 143-145 nucleotide excision-repair, 136-140 repair of apurinic sites by alternative pathways, 187-188 repair of platinum complex damage excision repair, 110-116 general comments, 110 postreplication repair, 116-128 Deoxyribonucleic acid glycosylases additional, 172-173 hydrolysis of nucleosides etc., 173 hypoxanthine, 162-165 3-niethyladenine, 165-172 uracil, 145-162 Deoxyribonucleic acid synthesis, inhibitors of, 239-245 Deoxyribonucleic acid viruses, inhibitors of, 239-245
Bacteria, filament formation, platinum complexes and, 85-86
C Cell-wall synthesis, inhibitors of, 249-266 Cyclic adenosine monophosphate phosphodiesterase, inhibitors of, 269-270
D Deoxyribonucleic acid biochemical effects of platinum complexes enzyme shidies, 89-93 inhibition of synthesis, 88-89 other proposed mechanisms of action, 93-94
E Electrophoresis interaction of protein size and charge during, 295-303 novel approach to, 321-325 Endonucleases, acting at many lesions, 186-187 Endonucleases (AP) associated exonuclease activity, 178183 without exonuclease activity, 183-186 Enzymes cryptic variants, nature of, 305-311 hidden variation in shape, detection of, 303-305 posttranslational modification, major unresolved issues, 315-321
355
356
SUBJECT INDEX
Enzyme shape genetic basis of variation, 311-315 internal standardization, 294-295 Eukaryotic gene models for regulation present knowledge about transcription designs, 347-349 process vs. discard decisions, 344345 selective processing decisions, 345347
G Gene, eukaryotic, models for regulation, 344-349 a-Glycerophosphate dehydrogenase, variation in shape of, 302-303
H Hyperemic agents, antibiotics as, 269 Hypoesthetic agents, antibiotics as, 269 Hypoxanthine-DNA glycosylase, 162-165
I Immune response, effectors, inhibitors of, 245-248 1
Large nuclear primary transcripts mRNA and kinetic IabeIing studies, 337-338 uItraviolet evidence for size; 338-339 Lysogeny, platinum complexes and, 86-87
M 3-Methyladenine-DNA glycosylase, 165-172 Mononucleotides, hydrolysis, DNA glycosylases and, 173 Mutagenesis, platinum complexes and, 87
N Nucleosides antibiotic, miscellaneous, 270-271 hydrolysis, DNA glycosylases and, 173
P Platinum coordination complexes biochemical effects indicating reaction with DNA enzyme studies, 89-93 inhibition of synthesis, 88-89 other proposed mechanisms of action, 93-94 biological effects indicating reactions with DNA bacterial filament formation, 85-86 induction of lysogeny, 86-87 mutagenesis, 87 reaction with viruses and transforming DNA, 87-88 interaction with DNA in vitro, 104-109 in vivo, 94-103 repair of DNA damage in viuo excision repair, 110-116 general, 110 postreplication repair, 116-128 Platinum drugs, chemical features of, 75-84 Protein size and charge, interaction during electrophoresis, 295-297 criteria for adequate internal standards, 297-299 empirical estimation of KR and Mo, 299-302 variation of shape of aglycerophosphate dehydrogenase, 302-303 Protein synthesis inhibitors, 196-197 purine nucleoside antibiotics, 200209 pyrimidine nucleoside antibiotics, 197-200 Purines, interconversion, inhibitors of, 266-268 Pyrimidines, interconversion, inhibitors of, 266-268
R Ribonucleic acid, see also Transfer RNA messenger definition of transcription units, 331337
357
SUBJECT INDEX
large nuclear primary transcripts and, 337-339 recent evidence on formation from primary transcripts, 339-344 Ribonucleic acid synthesis, inhibitors, 209-239 Ribonucleic acid viruses, inhibitors of, 239-245 Pibosomes aminoacyl-tRNA binding to, 46-59 peptidyltransferase center interaction with-C-C-A end, 60-63 peptide bond formation, 59-60
T Transcription units definition for mRNA average units, 331-333 specific mRNAs, 333-337 technique for defining, 329-330 electron microscopic visualization, 330 kinetic labeling, 330 ultraviolet mapping, 330-331 Transfer ribonucleic acid aminoacylation role of 3‘-terminal adenosine, 28-29
site of, 29-39 substrate properties of modified tRNAs, 23-28 aminoacyl structure conformation of aminoacyl residue, 11 influence of aminoacyl residue on tRNA conformation, 12-14 positional isomers, 10-11 enzymic modification of 3‘ end nucleotidyltransferase and, 16-22 polynncleotide phosphorylase and, 22-23 shortened tRNAs, 14-16 positional specificity of aminoacylation chemical “proofreading,” 45-46 site of aminoacylation and mischarging, 39-45 structure of -N-C-C-A terminus, 3 chemical modification, 4-6 physicochemical studies, 6-10 x-ray diffraction studies, 4
U Uracil-DNA glycosylase, 145-162
V Viruses, platinum complexes and, 87-88
Contents of Previous Volumes Volume 1
J.
"Primer" in DNA Polymerase Reactions-F. Bollum The Biosynthesis of Ribonucleic Acid in Animal Systems-R. M . Smellie The Role of DNA in RNA Synthesis-Jerard Hurwitz and]. T. August Polynucleotide Phosphorylase-M. Grunberg-Manago Messenger Ribonucleic Acid-Fritz Lipmann The Recent Excitement in the Coding Problem-F. H. Crick
s.
c.
Some Thoughts on the Double-Stranded Model of Deoxyribonucleic Acid-Aaron Bendich a n d H e r b e r t S . Rosenkranz Denaturation and Renaturation of Deoxyribonucleic Acid-J. Murmur, R. Rownd, a n d L.
c.
Schildkraut
s.
Some Problems Concerning the Macromolecular Structure of Ribonucleic Acids-A. Spirin The Structure of DNA as Determined by X-Ray Scattering Techniques-VittOriO L u z z a t i Molecular Mechanisms of Radiation Effects-A. Wacker
Volume 2 Nucleic Acids and Information Transfer-Liebe F. Cavalieri a n d B a r b a r a H. Rosenberg Nuclear Ribonucleic Acid-Henry H a r r i s Plant Virus Nucleic Acids-Roy M a r k h a m The Nucleases of Eschwichio coli-Z. R. L e h m a n R. Kdeg Specificity of Chemical Mutagenesis-huid Column Chromatography of Oligonucleotides and Polynucleotides-Matthy8 Staehelin Mechanism of Action ond Application of Arapyrimidines-1. Skoda The Function of the Pyrimidine Base in the Ribonucleose Reaction-Herbert Witzel Preporation, Fractionation, ond Properties of sRNA-4. L, B r o w n
Volume 3 Isolation and Fractionation of Nucleic Acids-K. S . Kirby Cellular Sites of RNA Synthesis-David M.Prescott Ribonucleases in Taka-Diastase: Properties, Chemical Nature, and Applications-Fuji0 Egami, K e n j i Takahashi, and Tsuneko U c h i d a Chemical Effects of Ionizing Radiations on Nucleic Acids and Related Compounds-JosephJ. Weiss The Regulation of RNA Synthesis in Bacteria-Frederick C. N e i d h a r d t Actinomycin and Nucleic Acid Function-E. Reich a n d I . H. Goldberg De Novo Protein Synthesis in Vitro-B. N i s m a n a n d J. Pelmont Free Nucleotides in Animal Tissues-P. M a n d e l
Volume 4 Fluorinated Pyrimidines-Charles Heidelberger Genetic Recombination in Bacteriophage-E. Volkin DNA Polyrneroses from Mammalian Cells-H. M . K e i r The Evolution of Base Sequences in Polynucleotides-B. J . M c C a r t h y Biosynthesis of Ribosomes in Bacterial Cells-Syozo Osawa 5-Hydroxymethylpyrimidines and Their Derivatives-T. L. V. Ulbright Amino Acid Esters of RNA, Nucleotides, and Related Compounds-H. Feldmann Uptake of DNA by living Cells-L. Ledous
358
G . Zachau
and
H.
CONTENTS OF PREVIOUS VOLUMES
359
Volume 5 Introduction to the Biochemistry of 4-Arabinosyl Nucleosides-Seymour S. Cohen Effects of Some Chemical Mutagens and Carcinogens on Nucleic Acids-P. D. L a w l e y Nucleic Acids in Chloroplasts and Metabolic DNA-Tatsuichi Zwamura Enzymatic Alteration of Macromolecular Structure-P. R. Srinioasan a n d E r n e s t Borek Hormones and the Synthesis and Utilization .of Ribonucleic Acids-]. R. Tata . FOX, K y o i c h i A. Wutanube, a n d Alexander B l o c h Nucleoside Antibiotics-Jack I Recombination of DNA Molecules-Charles A. Thomas, ]r. Appendix I. Recombination of o Pool of DNA Fragmentswith Complementary Single-Chain EndsG. S. Watson, W. K . Smith, a n d Charles A. Thomas, ]r. Appendix II. Proof that Sequences of A, C, G, and T Con Be Assembled to Produce Chains of Ultimate length, Avoiding Repetitions Everywhere-A. S. Fruenkel a n d J . G i l l i s The Chemistry of Pseudouridine-Robert Warner Chambers The Biochemistry of Pseudouridine-Eugene Goldwasser a n d Robert L. H e i n r i k s o n Volume 6 Nucleic Acids and M u t a b i l i t y S t e p h e n Zamenhof Specificity in the Structure of Transfer RNA-Kin-ichiro Miura Synthetic Polynucleotides-A. M. Michelson, J. Massoulif, and W. Guschbauer The DNA of Chloroplasts, Mitochondria, and Centrioles-S. G r a n i c k a n d A h a r o n G i b o r Behavior, Neural Function, and RNA-H. HydSn The Nucleolus and the Synthesis of Ribosomes-Robert P. Perry The Nature and Biosynthesis of Nuclear Ribonucleic Acids-G. P. Georgieo Replication of Phage RNA-Charles Weissmann a n d Seuero Ochoa Volume 7 Autoradiographic Studies on DNA Replication in Normal and leukemic Human ChromosomesFelice Cavosto Proteins of the Cell Nucleus-Lubomir S. H n i l i c a The Present Status of the Genetic C o d e - C n r l R. Woese The Search for the Messenger RNA of Hemoglobin-€€. Chantrenne, A. Burny, a n d G. M a r b a i x Ribonucleic Acids and Information Transfer in Animal Cells-A. A. Hadjiolov Transfer of Genetic Information during Embryogenesis-Martin Nemer Enzymatic Reduction of Ribonucleotides-Agne Larsson a n d Peter Reichard The Mutagenic Action of Hydroxylamine-1. H. P h i l l i p s and D. M. B r o w n Mammalian Nucleolytic Enzymes and Their Localization-Dauid Shugar a n d H u l i n a Sierakowska Volume 8 Nucleic Acids-The First Hundred Years-]. N. Davidson Nucleic Acids and Protamine in Salmon TestesGOrdOn H. D i x o n a n d M i c h a e l Smith Experimental Approaches to the Determination of the Nucleotide Sequences of large Oligonucleotides and Small Nucleic Acids-Robert W. H o l l e y Alterations of DNA Base Composition in B a c t e r i a x . F. Gause Chemistry of Guanine and Its Biologically Significant Derivatives-Robert Shapiro Bacteriophage 4x174 and Related Viruses-Robert L. Sinsheimer The Preparation and Characterization of Large Oligonucleotides-George W. Rushizky a n d H e r b e r t A . Sober Purine N-Oxides and Cancer-George B o s w o r t h B r o w n The Photochemistry, Photobiology, and Repair of Polynucleotides-R. B . Setlow
360
CONTENTS OF PREVIOUS VOLUMES
What Really Is DNA? Remarks on the Changing Aspects of a Scientific Concept-Erwin Recent Nucleic Acid Research in China-Tien-Hsi Cheng a n d R o y H. D o i
Chargaff
Volume 9 The Role of Conformation in Chemical Mutageneais-B. Singer a n d H. Fraenkel-Conrat Polarographic Techniques in Nucleic Acid Research-E. PaleEek RNA Polymerase and the Control of RNA Synthesis-John p. Richardson Radiation-Induced Alterations in the Structure of Deoxyribonucleic Acid and Their Biological T. K a n a z i r tonsequencer-D. Optical Rotatory Dispersion and Circular Dichroism of Nucleic Acids-.en Tsi Yang a n d Tatsuya Samejima M . B. The Specificity of Molecular Hybridization in Relation to Studies on Higher Organisms-P. Walker Quantum-Mechanical Investigations of the Electronic Structure of Nucleic Acids and Their constituents--Bernard P u l l m a n a n d Alberte P u l l m a n The Chemical Modification of Nucleic Acids-N. K . Kochetkou a n d E. I . Budowsky Volume TO Induced Activation of Amino Acid Activating Enzymes by Amino Acids and tRNA-Alan H. M e h l e r Transfer RNA and Cell Differentiation-Nobow Sueoka a n d Tamiko Kano-Sueoka N'-(A*-Isopentenyl)adenosine: Chemical Reactions, Biosyntheais, Metabolism, and Significance to the Structure and Function of tRNA-Ross H.Hall Nucleotide Biosynthesis from Preformed Purines in Mammalian Cells: Regulatory Mechanisms and Biological Significance-A. W. M u r r a y , D a p h n e C . E l l i o t t , a n d M . R. Atkinson Ribosome Specificity of Protein Synthesis in V i t r o - O r i o Ciferri und B r u n o Parisi Synthetic Nucleotide-peptides-Zoe A. Shabaroua The Crystal Structures of Purines, Pyrimidines and Their Intermolecular Complexes-Donald Voet a n d Alexander R i c h Volume 11 The Induction of interferon by Natural and Synthetic Polynucleotides-Clarence Colby, Jr. Ribonucleic Acid Maturation in Animal Cells-R. H . B u r d o n Liporibonucleoprotein as an Integral Part of Animal Cell Membranes-v. S. Shapot and S. Yu. Dauidoua Uptake of Nonviral Nucleic Acids by Mammalian Cells-Pushpa M . Bhargaua a n d G. Shan-
mugum The Relaxed Control Phenomenon-Ann M . R y a n and E r n e s t Borek Molecular Aspects of Genetic Recombination-Cedric I . D a u e r n Principles and Practices of Nucleic Acid Hybridization-Dauid E . Kennel1 Recent Studies Concerning the Coding Mechaniam-7'homas H. Jukes a n d L i l a G a t l i n The Ribosomal RNA Cistrons-M. L. Birnstiel, M . Chipchase, a n d J. Speirs Three-Dimensional Structure of tRNA-Friedrich Cramer Current Thoughts on the Replication of D N A d n d r e w Becker a n d Jerard Hurwitz Reaction of Aminoacyl-tRNA Synthetares with Heterologous tRNA's-K. B r u c e Jacobson On the Recognition of tRNA by Its Aminoacyl-tRNA Ligase-Robert Chambers
w.
Volume 12 Ultraviolet Photochemistry as o Probe of Polyribonucleotide Conformation-A. Jacques R. Fresco Some Recent Developments in DNA Enzymology-Mehran G o u l i a n
J. L o m a n t and
CONTENTS OF PREVIOUS VOLUMES
361
Minor Components in Transfer RNA: Their Characterization, Location, and Function-susumu Nishimura The Mechanism of Aminoacylotion of Transfer RNA-Robert B. Loftfield Regulation of RNA Synthesis-Ekkehard K . F. B a u t z The Poly(dA-dT) of Crab-M. Laskowski, Sr. The Chemical Synthesis and the Biochemical Properties of Peptidyl-tRNA-Yehuda N a t h a n de Groot
Lapidot a n d
Valume 13 Reactions of Nucleic Acids and Nucleoproteins with Formaldehyde-M. Ya. F e l d m a n Synthesis and Functions of the -C-C-A Terminus of Transfer RNA-Murray P. Deutscher Mammalian RNA Pofymerases-Samson T. Jacob Poly(adenosine and diphosphate ribose)-Z'akashi Sugimura The Stereochemistry of Actinomycin Binding to DNA and I t s Implications in Molecular BiologyH e n r y M. Sobell Resistance Factors and Their Ecological Importance to Bacteria and to Man-M. H. R i c h m o n d Lysogenic Induction-Ernest Borek a n d Ann R y a n Recognition in Nucleic Acids and the Anticodon Families-Jacques N i n i o Translation and Transcription of the Tryptophan Operon-Fumio Imamoto Lymphoid Cell RNA's and Immunity-A. Arthur G o t t l i e b Volume 14 DNA Modification and Restriction-Werner A r b e r Mechanism of Bacterial Transformotion and Tronsfection-Nihd K . N o t a n i andJane K . Setlow DNA Polymerases II and 111 of Escherichia coli--2\.lakolm L. Gefter The Primary Structure of DNA-Kenneth M u r r a y a n d Robert W. Old RNA-Directed DNA Polymerase-Properties and Functions in Oncogenic RNA Viruses and CellsM a u r i c e Green a n d G r a y F. G e r a r d Volume 15 Information Transfer in Cells Infected by RNA Tumor Viruses and Extension to Human Neoplasio-D. Gillespie, W. C. Saxinger, a n d R . C. Gallo Mammalian DNA Polymerases-F. J. Bollum Eukaryatic RNA Polymerases and the Factors That Control Them-B. B . Biswas,A. Ganguly, and D. D a s Structural and Energetic Consequences of Noncomplementary Base Oppositions in Nucleic Acid Helices-A. J . L o m a n t and Jacques R. Fresco The Chemical Effects of Nucleic Acid Alkylation and Their Relation to Mutagenesis and Carcinogenesis-B. Singer Effects of the Antibiotics Netropsin and Distamycin A on the Structure and Function of Nucleic AcidsAhristoph Zimmer Volume 16 Initiation of Enzymic Synthesis of Deoxyribonucleic Acid by Ribonucleic Acid Primers-Erwin Chargafl Transcription and Processing of Transfer RNA Precursors-John D. Smith Bisulfite Modification of Nucleic Acids and Their Constituents-Hikoya H a y a t s u The Mechanism of the Mutagenic Action of Hydroxylamines-E. 1. Budowsky Diethyl Pyrocarbonate in Nucleic Acid Research-L. Ehrenberg, I . Fedorcsrik, and F. SolYmosY
362
CONTENTS OF PREVIOUS VOLUMES
Volume 17 The Enzymic Mechanism of Guanosine 5‘,3’-Polyphorphate
Synthesis-Fritz
L i p m a n n a n d Jose
SY
s.
Effects of Polyamines on the Structure and Reactivity of tRNA-Ted T. Sakai and Seymour Cohen Information Transfer and Sperm Uptake by Mammalian Somatic Cells-Aaron Bendich, Ellen Borenfreund, Steven S . Witkins, D e l i a Beju, a n d P a u l J . H i g g i n s Studies on the Ribosome and Its Components-Pnha Spitnik-Elson a n d D a v i d E l s o n Classical and Postclassical Modes of Regulation of the Synthesis of Degradative Bacterial Enzymes-Boris Magasanik Characteristics and Significance of the Polyadenylate Sequence in Mammalian Messenger RNAGeorge B r a w e r m a n Polyadenylate Polymerases-Mary Edmonds a n d M a r y Ann Winters Three-Dimensional Structure of Transfer RNA-sung-Hou Kim Insights into Protein Biosynthesis and Ribosome Function through Inhibitors-Sidney Pestka Interaction with Nucleic Acids of Carciqgenic and Mutagenic N-Nitroso Compounds-W. L i j i n s k y Biochemistry and Physiology of Bacterial Ribonuclease-Abk K . D a t t a a n d S a h l K. N i y o g i Volume 18 The Ribosome of Escherichia coli-R. Brimacornbe, K . H. Nierhaus, R. A. G a r r e t t a n d H. G. Wittmann Structure and Function of 5 S and 5.8 S RNA-Volker A. E r d m a n n High-Resolution Nuclear Magnetic Resonance Investigations of the Structure of tRNA in SolutionD a v i d R. Kearns Premelting Changes in DNA Conformation-E. PaleEek Quantum-Mechanical Studies on the Conformation of Nucleic Acids and Their ConstituentsB e r n a r d P u l l m a n a n d Anil Saran Volume 19 I. The 5’-Terminal Sequence (“Cap”) of mRNAr Caps in Eukaryotic mRNAs: Mechanism of Formation of Reovirus mRNA 5’-Terminal m’GpppGmC-Y. Furuichi, S. Muthukrishnan, J . Tomasz a n d A. J . S h a t k i n Nucleotide Methylation Patterns in Eukaryotic mRNA-Fritz M. Rottman, R o n a l d Desrosiers and Karen F r i d e r i c i Structural and Functional Studies on the “5’-Cap”: A Survey Method of mRNA-Harris Busch, F r i e d r i c h Hirsch, Kaushal K u m a r Gupta, Manchanahalli Rao, W i l l i a m Spohn a n d B e n j a m i n C. Wu Modification of the 5’-Terminals of mRNAs by Viral and Cellular Enzymes-Bernard Moss, Scott A. Martin, M a r c i a J . Ensinger, Robert F. Boone and Cha-Mer Wei Blocked and Unblocked 5‘ Termini in Vesicular Stomatitis Virus Product RNA in Vitra: Their Possible Role in mRNA Biosynthesis-Richard J . Colonno, G o r d o n A b r a h a m a n d A m i y a K. Banerjee The Genome of Poliovirus Is an Exceptional Eukaryotic mRNA-Yuan F o n Lee, A k i o Nomoto a n d
c.
Eckard Wimmer 11. Sequences and Conformations of mRNAs Transcribed Oligonucleotide Sequences in Hela Cell hnRNA and mRNA-Mary Edmonds, H i r o s h i Nakazato, E. L. K o r w e k a n d S. Venkatesan Polyadenylylation of Stored mRNA in Cotton Seed Germination-Barry Harris a n d L e o n D u r e
111
363
CONTENTS OF PREVIOUS VOLUMES
mRNAs Containing and Lacking Poly(A) Function a s Separate and Distinct Ciosses during Embryonic
Development-Martin
Nenier and Saul Surrey
Sequence Analysis of Eukaryotic mRNA-N. J. Proudfoot, C. C. Cheng and G. G. Brownlee T h e Structure and Function of P r o t a m i n e mRNA from Developing Trout Testis-P. L. Davies, G. H.
Dixon, L. N. Ferrier, L. Gedamu und K. Iatrou The Primary Structure of Regions of SV40 DNA Encoding the Ends of mRNA-Kiranur
N . Subrarnrtniun, Prabhat K. Ghorhi, Ruvi Dhar, Bayur Thimmuppciya, Sayeeda B. Zain, Julian Pan and Sherman M. Weissman
Nucleotide Sequence Analysis of Coding and Noncoding Regions of H u m a n /3-Globin mRNA-
Charles A. Marotta, Bernard G. Forget, Michael CohenlSolal and Sherman M. Weissman Determination of Globin mRNA Sequences and Their Insertion into Bacterial Plasmids-Winston Salser, Jeff Browne, Pat Clarke, Howard Heindell, Russell Higuchi, Gary Paddock, John Roberts, Gory Studnicku a n d Paul Zakar The Chromosomal Arrangement of Coding Sequences in a Family of Repeated Genes-G. M. Rubin, D. J. Finnegan and D. S. Hogness Mutation Rates in Globin Genes: T h e Genetic load a n d Haldane's Dilemma-Winston Salser and Judith Strommer lsaacson Heterogeneity of t h e 3' Portion of Sequences Related to Immunoglobulin K-Chain mRNA-UrsulU Stwb Structural Studies o n Intact and Deadenylylated Rabbit Globin mRNA-John N . Vournakis, Marcia S. Flashner, MaryAnn Katopes, Gary A. Kitos, Nikos C. Vamvakopoulos, Matthew S. Sell and Regina M. Wur.st Molecular Weight Distribution of RNA Fractionated o n Aqueous a n d 70% Formomide Sucrose Gradients-Helga Boedtker rind Hans Lehrach 111. Processing of mRNAs Bacteriophages T7 and T3 as Model Systems for RNA Synthesis and
Processing-J.J. Dunn, c.w. Anderson, J . F. Atkins, D. C. Bartelt and W. C. Crockett The Relationship between hnRNA and mRNA-Robert P. Perry, Enzo Bard, B. David Hames, Dawn E. Kelley und Ueli Schibler A Comparison of Nuclear and Cytoplasmic Viral RNAs Synthesized E a r l y in Productive Infection with Adenovirus 2-Heschel J. Raskas and Elizabeth A. Craig Biogenesis of Silk Fibroin mRNA: An Example of Very Rapid Processing?-Pd M. Lizardi Visualization of the Silk Fibroin Transcription Unit and Nascent Silk Fibroin Molecules o n Polyribosomes of Bombyx mori-Steven L. McKnight, Nelda L. Sullivan und Oscar L.
Miller, Jr. Production and Fate of Balbiani Ring Products-B.
Daneholt, S. T. Case, J. Hyde, L. Nelson
and L. Wieslander Distribution of hnRNA and mRNA Sequences i n Nuclear Ribonucleoprotein Complexes-Ala?i
J.
Kinniburgh, Peter B. Billings, Thomas J. Quinlan and Terence E. Martin IV. Chromatin Structure and Template Activity The Structure of Specific Genes in Chromatin-Richard Axel The Structure of DNA in Native Chromatin a s Determined by Ethidium Bromide Binding-J.
Paoletti, B. B. Magee and P. T. Magee Cellular Skeletons and RNA Messages-Ronald
Herman, Gary Zieve, Jeffrey Williams,
Robert Lenk and Sheldon Penman The Mechonism of Steroid-Hormone Regulation of Transcription of Specific Eukaryotic Genes-Bert
W. O'Malley and Anthony R. Means Nonhistone Chromosomal Proteins and Histone Gene Transcription-Gary
Stein, Janet Stein, Lewis Kleinsmith, William Park, Robert Jansing and Judith Thomson
364
CONTENTS OF PREVIOUS VOLUMES
Selective Transcription of DNA Mediated by Nonhistone Proteins-Tung Y. Wang, Nina C. Kost r a b a a n d Ruth S. N e w m a n V. Control of Translation Structure and Function of the RNAs of Brome Mosaic Virus-Paul Kaesberg Effect of 5’-Tenninal Structures on the Binding of Ribopolymers to Eukaryotic Ribosomes-s. M u t h u k r i s h n a n , Y. F u r u i c h i , G. W. B o t h a n d A . J. S h a t k i n Translational Control in Embryonic M u s c l e S t u a r t M. H e y w o o d and D o r i s S . K e n n e d y G. Whalen, M. E . B u c k i n g h a m a n d F. Protein and mRNA Synthesis in Cultured Muscle Cell-R.
Gros VI. Summary mRNA Structure and Function-James
Volume 20
E. Darnell
c.
Correlation of Biological Activities with Structural Features of Transfer RNA-B. F. Clark Bleomycin, a n Antibiotic That Removes Thymine from Double-Stranded DNA-Werner E . G. Mdler a n d Rudolf K . Z a h n Mammalian Nucleolytic Enzymes-&zlina Sierakowska and David S h u g a r Transfer RNA in RNA Tumor Viruses-hrw C. Waters a n d B e t h C. Mullin I?tegration versus Degradation of Exogenous DNA in Plants: An Open Question-Paul F. Lurquin Initiation Mechanisms of Protein Synthesis-Marianne G r u n b e r g - M a n a g o and FranEois Gros
Volume 21 lnfonnosomes and Their Protein Componentr: The Present State o f Know1edge-A. A. Preobrazhensky a n d A. S. S p i r i n Energetics of the R i b o s o m e 4 . S . S p i r i n Mechonisms in Polypeptide Chain Elongation on Ribosomes-Engin B e r m e k Synthetic Oligodeoxynucleotides for Analysis of DNA Structure and F u n c t i o n 4 a y WU, C h a n d e r P. Bahl, and S a r a n A. N a r a n g The Transfer RNAs of Eukaryotic Organelles-W. E d g a r B a r n e t t , S. D. Schwartzhach, a n d L. I . H e c k e r Regulation of the Biosynthesis of Aminoacid:tRNA Ligases and of t R N A 4 u s a n D. M o r g a n a n d D i e t e r Sol1