Proc. Natl. Acad. Sci. USA Vol. 73, No. 11, pp. 4159-4163, November 1976 Genetics

Heat mutagenesis in bacteriophage T4: The transversion pathway (spontaneous mutation/acid mutagenesis/glycosidic bond migration) PAUL M. BINGHAM, RICHARD H. BALTZ*, LYNN S. RIPLEY, AND JOHN W. DRAKEt Department of Microbiology, University of Illinois, Urbana, Ill. 61801 Communicated by Matthew S. Meselson, September 3, 1976

ABSTRACT Heat induces transversions (as well as transi- produce the homologous ochre (UAA) mutant and therefore tions) at GC base pairs in bacteriophage T4. The target base for go undetected, the heat-induced reversion of amber mutants transversions is guanine, which is converted to a product which could occur only by transversions at the G-C base pair (pro- is sometimes replicated and transcribed as a pyrimidine. A model for this process is proposed in which the deoxyguanosine ducing the tyrosine codons UAU or UAC). Table 2 shows that glycosidic bond migrates from N9 to N2: the resulting deoxy- amber mutants readily revert when heated. neoguanosine may pair with normal guanine to produce GC Amber mutants might, however, revert by extracistronic _ CG transversions. suppressors arising by G-C A-T transitions. Hydroxylamine specifically induces this transition, and a considerable number We have previously described the induction by heat of G-C - of amber mutants in the WI region have already been shown A-T transitions in bacteriophage T4 by deamination of cytosine to be refractory to hydroxylamine-induced reversion (3, 4). We to produce uracil (1). Earlier studies of heat mutagenesis in have extended this result by showing that the amber mutants bacteriophage T4 suggested that transversions were also in- described in Table 2 are also insusceptible to hydroxylamine- duced at G-C base pairs (2), and this conjecture is now con- induced reversion even when the treated stocks are grown firmed. We have also confirmed previous results indicating that under permissive conditions before selective plating for the rII+ neither base pair substitutions at A-T sites nor frameshift revertants. (Hydroxylamine-treated HB129AM produces a few mutations are induced by heat mutagenesis. revertants that are not wild type, but these are readily distin- guishable from the revertants of normal appearance induced MATERIALS AND METHODS by heat.) Most materials and methods have been described previously Guanylate Residues Are the Mutagenic Targets for the (ref. 1 and references therein). Drake top and bottom agars Transversion Pathway. Among the revertants of rII mutants were usually used, but enriched H agar was used in experiments arising as rII/r + heteroduplex heterozygotes, only those con- involving amE509; supersoft enriched H top agar contained taining the r+ allele on the transcribed strand are detected by only 0.42% agar. Escherichia coli was used throughout: strain direct plating on a nonpermissive host; those containing the r+ BB to grow and titer T4 wild-type and rnI strains and to score allele on the complementary strand must be grown for one cycle revertants of amE509, strain KB to score rnI + revertants, and ("passaged") on a permissive host before revertants can be strain CR63 (an amber suppressor) to grow and titer amE509. detected by plating on a nonpermissive host. At the G-C site of All T4rII mutants are in a T4B background; amE509 is in a an amber (UAG) mutant the transcribed strand contains the T4D background. Phage stocks were purified by differential cytosine and the complementary strand the guanine residue. centrifugation and resuspended in 1 mM sodium phosphate, Table 2 shows that heat-induced reversion of rnI amber mutants pH 7.0, plus 0.5 mM MgCI2, except that the stocks used in the is detected mainly or exclusively only after passaging, thus experiments described in Tables 1 and 2 (excluding those entries implicating the guanylate residue as the primary mutagenic in Table 1 quoted from previous work) were resuspended di- target for the transversion pathway. [Cytosine is the mutagenic rectly in the heating buffers. During heating, pH was deter- target for the transition pathway (1).] mined under experimental conditions, and heating was ter- The alternative hypothesis, that cytosine is the mutagenic minated by dilution into chilled 100 mM sodium phosphate, target in the transversion mechanism, is contradicted by the pH 7.0, plus 0.5 mM MgCl2. behavior of the rII missense mutant rUV74 (Table 2). rUV74 is insusceptible to reversion by mutagens producing transitions RESULTS (base analogues, hydroxylamine), and its heat-induced rever- Neither Base Pair Substitutions at A-T Sites nor Frameshift tants are expressed upon direct plating on the nonpermissive Mutations Are Induced by Heat. Data accumulated to date host. For cytosine to be the target in a reverting amber (UAG) on the reversion responses of T4rII mutants reverting only at codon, its heat-induced derivative would have to be replicated A-T site, or only by frameshift mutagenesis, are summarized as a purine, but transcribed either not at all or else as a pyrim- in Table 1. None of 15 A-T and 10 frameshift mutants was in- idine. In that case, no nIl mutant could exhibit heat-induced duced to revert by heat under a variety of conditions, even transversion reversion upon direct plating, whereas rUV74 when heated stocks were grown under permissive conditions does. before selective plating to detect rnI + revertants. Thus, the guanylate residue is the target for the heat-induced Heat Induces Transversion Mutations at G-C Sites. Since transversion and the product of the mutagenic reaction can be A-T base pairs are refractory to heat mutagenesis, and since transcribed and replicated as a pyrimidine. an mutant The Transversion Pathway May Be GC -- CG. transitions at the G-C base pair of amber (UAG) Transversion mutations arising at G-C sites could produce either * Present address: Biochemical Development Division, Eli Lilly and C-G or T-A base pairs. At present we still lack tester strains ca- Co., Indianapolis, Ind. 46206. pable of distinguishing between these two pathways. However; t To whom communications should be addressed. a previous collection (2) of heat-induced rnl mutants consisted, 4159 Downloaded by guest on September 30, 2021 4160 Genetics: Bingham et al. Proc. Nati. Acad. Sci. USA 73 (1976)

Table 1. Failure of A- T and frameshift mutants to revert Table 2. Induction of G- C-site transversions by heat when heated Revertants per 108 survivorst Revertants per 108 survivorst H(-)- H(-)- H(+) H(+)- H(-)- H(-)- H(+)- H(+)- Mutant* Seriest P(-) P(+) P(-) P(+) Mutant* Seriest P(-) P(+) P(-) P(+) rHB129AM VII 6 9 15 63 A T mutants rP7AM VIII 0.7 0.5 < 0.9 6.5 42 rP70C VI 2 2 2 3 rUV200AM IV 3 7 4 rSM120C VI 9 7 10 7 V 6 9 9 56 rSM41 OC VI 9 11 10 10 VI 8 6 3 30 rUV16 I 26 18 rUV74 IV 35 50 700 460 rUV69 I 0.1 0.1 * The AM suffix indicates an amber (UAG) mutant. The three am- rUVI1 7 II 2000 2000 ber mutants (but not rUV74) are revertible by base analogues, III 2000 1700 but none of the mutants is revertible by hydroxylamine, even IV 2500 4400 1700 4000 after passaging. V 2000 3800 1700 4800 t IV, V, and VI are described in a footnote to Table 1. VII = 26 hr rUVI99 II 23 20 at 600 in 10 SmM sodium phosphate, pH 7.0, plus 0.5 mM MgCl2. III 23 25 VIII = 28 hr at 570 in 1 mM sodium phosphate, pH 7.0, plus 0.5 IV - 320 400 440 440 mM MgCl2. 1. V 150 390 200 11:u t See footnote to Table rUV2000C IV 1 1 3 3 V 3 4 2 2 after the frameshift mutants contributed VI 3 1 3 3 proflavin-revertible rUV215 IV 35 52 50 20 by the spontaneous background was discarded, of two classes. V 21 24 33 17 The members of one class were revertible by base analogues and rUV236 V 50 130 74 92 might, therefore, have arisen by the G-C , A-T transitions rUV304 I 1 1 already known to be induced by heat (1). However, many of V 13 33 21 these mutants were also revertible by hydroxylamine, which rUV354 I 8 8 specifically induces G-C -- A-T transitions. It was therefore rUV3570C VI 5 7 5 proposed that these mutants were induced by G-C - C-G rUV373 II 2300 2800 transversions, but are revertible to T-A by hydroxylamine, III 2300 2800 producing an acceptable but usually nonwild-type amino acid rUV379 I 0.5 0.5 in the rIl polypeptide. Frameshift mutants The members of the second class were revertible by neither nor These mutants might have arisen rUV2 I 52 73 base analogues proflavin. IV 110 67 150 either by transversion mutagenesis, generating a codon inca- rUV20 I 4 4 pable of reversion along a transition pathway, or by frameshift rUV27 II 1200 1400 mutagenesis, generating mutants nevertheless insusceptible to III 1200 1200 proflavin-induced reversion. [Note that of 12 proflavin-induced IV 6500 6500 5000 rII mutants, four were nonrevertible by proflavin, 5-amino- V 6800 7500 4300 acridine, or both (5).] We therefore further examined five rUV28 IV 74 90 92 mutants of this class, chosen for their varied reversion rates. The rUV34 I 1900 2500 IV 1500 2400 1000 revert by A-T - G.C transitions and sometimes by A.T-site rUV81 I 1.5 <1 transversions, but not (according to the conclusions of this report) I 1 <1 by G.C-site transversions. rP70C and rUV117, however, revert rUV1 03 only by A-T-site transversions (25). OC mutants (such as rSM12- rUV113 V 41 100 90 OC) are ochre (UAA) mutants, and frequently revert by A-T-site rUV274 II 730 950 transversions (25). Frameshift mutants are revertible by profla- III 730 880 vin, but not by base analogues. rUV7 is revertible by hydroxyl- IV 1400 1600 1100 amine by the transition G-C - A-T at two sites, one of which V 920 1500 900 yields revertants that are not wild type; the values for this mutant rUV353 II 180 190 are the sum (at roughly equal frequencies) ofboth responses. III 180 110 t I = 50 months at 30 in L broth. II = 412 days at 00 in L broth. III IV 460 700 500 = 412 days at 200 in L broth. [I, II, and HI are quoted from a 190 250 280 previous report (26).] IV = 31 days at 450 in 10 mM sodium phos- V phate, pH 7.25, plus 1 mM MgCl2. V = 40 hr at 580 in 10 mM so- G-C mutant (positive control) dium phosphate, pH 7.0, plus 0.5 mM MgCl2. VI = 48 hr at 550 in rUV7 I 18 128 10mM sodium phosphate, pH 7.25, plus 0.5mM MgCl2. f H(-) = unheated, H(+) = heated, P(-) = unpassaged, P(+) = II 20 102 passaged. Phages to be passaged were adsorbed to E. coli BB at III 20 945 5 x 108 cells per ml. on a rotary shaker at 37°. The total multi- IV 23 56 2700 plicity of infection (active plus killed particles) was less than 10, V 72 57 1100 and the multiplicity of active particles was much less than one. (Heat-inactivated T4 particles are incapable of crossreactivation * The relevant properties of the mutants have been described (2, and multiplicity-reactivation.) After 5-7 min the complexes were 22-25). In general, A.T mutants are revertible by base analogues diluted 10- to 100-fold into warm L broth on the rotary shaker, (2-aminopurine and/or 5-bromouracil) and by thymine depriva- and lysis was completed after an additional 20 min by adding tion (24), but by neither hydroxylamine nor proflavin. They can chloroform. Downloaded by guest on September 30, 2021 Genetics: Bingham et al. Proc. Nati. Acad. Sci. USA 73 (1976) 4161

Table 3. Properties of heat-induced rII mutants revertible neither by proflavin nor by base analogues Revertants per 108 survivors * Complementation Mutant H(-)P(-) H(-)P(+) H(+)P(-) H(+)P(+) in 1589 testt rSM3 0.3 0.4 0.1 0.4 Positive rSM18 230 200 220 310 Not tested rSM57 5 .8 6 9 Negative rSM71 0.2 0.4 G0.9 < 0.2 Negative rSM90 11 21 12 24 Negative * See Table 1 footnote. rSM3 was heated for 72 hr at 56° in 10 mM sodium phosphate, pH 7.0, plus 0.05 mM MgCl2. The other mutants were heated for 39 hr at 550 in 10 mM sodium phosphate, pH 7.25, plus 0.5 mM MgCl2. t These tests were performed both in a nonsuppressor strain and in the suppressor strains QA1 (amber suppressor), CA165 (ochre suppressor), and CAJ64 (opal suppressor). CA165 contains an inefficient suppressor, however, and may fail to detect ochre codons in this test. rSM18 could not be tested because of its map location. mutants were tested for revertibility by heat; for suppressibility reversion of amE5O9 is shown in Table 4. The reaction displays by host strains inserting tryptophan into opal (UGA) codons, a small negative salt effect, increasing Mg2+ concentrations tyrosine or glutamine into ochre (UAA) codons, and tyrosine, reducing heat-induced mutation rates. serine, glutamine, or leucine into amber (UAG) codons; and for Since amE509 resides in the major T4 head-protein gene, its behavior in the "1589 test" (6). None of the mutants was re- revertants might be inactivated by heat at rates different from vertible by heat (Table 3), and none was suppressible. Mutants that of the mutant itself. Reconstruction controls were therefore that exhibit complementation in the 1589 test contain either performed. Inocula were prepared containing amE509 plus missense mutations or additions or deletions of 3n base pairs (n about 10-4 parts of a mixture of equal numbers of ten inde- an integer). Those that fail to complement in the nonsuppressing pendent am + revertants, a frequency well above the typical host but do complement in a suppressor strain contain UGA, heat-induced revertant content of treated amE509 stocks. High UAA, or UAG chain-termination codons. Those that fail to titer stocks were grown from this mixture and purified ac- complement in any of these strains are likely to contain either cording to standard procedures (1). No selection for or against frameshift mutations or inefficiently suppressed chain-termi- am + revertants was observed during the subsequent heating nation codons. Three of the four mutants that could be subjected of such stocks under a variety of conditions. to the 1589 test failed to exhibit complementation (Table 3). Tests were also conducted to determine whether amE509 We therefore suspect that most of the mutants of the second reverted by extracistronic suppression. Thirty to 40 revertants class merely contain frameshift mutations nonrevertible by were recovered from each of various measurements of revertant proflavin. The characteristics of mutants of the first class frequency, and each was purified, grown into stocks, and therefore suggest that the main heat-induced transversion backcrossed against wild-type T4D in a standard cross. Control pathway consists of G-C C-G. crosses consisting of psuI+ amE509 X wild type were also Kinetics of the Transversion Mechanism. Further aspects performed; psuI + is a suppressor of amber mutations in late- of heat-induced transversion mutagenesis at GC base pairs have acting T4 genes and is loosely linked to gene 23 (7), and such been explored using rUV74 and the gene-23 (head protein) crosses always produced 6-15% amE509 recombinants. The amber mutant amE509. Since gene 23 is expressed late in the lysates from all crosses were screened for amE509 recombinants course of T4 development, well after the initiation of DNA replication, reversion of amE509 can also be detected by direct T (0C) plating of heated stocks upon a nonpermissive host. The temperature dependencies of heat-induced reversion 55 45 35 25 of rUV74 and amE509 are shown in Fig. 1. Both mutants revert at very similar rates, with Arrhenius activation energies of about 145 kJ/mol (about 35 kcal/mol). While rUV74 and amE509 produce monophasic Arrhenius plots, other gene-23 amber QL0-718 mutants (such as amB17) produce biphasic curves, upwards deviations occurring in the neighborhood of 430. Preliminary q.I studies suggest that the anomalously high mutation rates ob- served at lower temperatures result from the induction of GC ,- A-T transitions producing weak extracistronic amber sup- pressors (7), coupled with selection on the assay plates for true c1°108 revertants in the psu + background. The pH dependence of the heat-induced reversion of amE509 is shown in Fig. 2. The rate is insensitive to hydrogen Ia-FI Of ion concentration in the pH region from 6 to 8, but is inversely 3.0 3.1 3.2 3.3 3.4 proportional to pH (with -1.0 slope) in the pH region from 3.5 to 5.5. Unit slopes were not always observed in this pH region, 103/T (K) however; based on limited data, rUV74 displayed a slope of FIG. 1. Heat-induced reversion rates as functions of temperature. approximately -0.7, and the gene-23 mutants amH36 and rUV74 (-) was heated in 10mM sodium phosphate, pH 6.4, plus 0.5 mM MgCl2; E. = 149 + 27 kJ/mol. amE509 (0) was heated in 10 mM amB272 displayed slopes of approximately -0.3 and -0.6, sodium phosphate, pH 7.4, plus 0.05 mM MgCl2; Ea = 141 + 14 kJ/ respectively. mol. Error flags and activation energy uncertainties indicate two The effect of Mg2+ concentration on the rate of heat-induced standard deviations. Downloaded by guest on September 30, 2021 4162 Genetics: Bingham et al. Proc. Natl. Acad. Sci. USA 73 (1976) T4 I5 '-T ' 1 chastic aspects of replication, however, render further dis- 10' crimination impossible by purely genetic tests. DISCUSSION t10-6._ Heat mutagenesis in bacteriophage T4 generates exclusively base pair substitutions at GCC base pairs, but these are of at least two types: GCC -- A-T transitions, produced by the deamination of cytosine (1), and transversions (probably G-C CCG), pro- q 10' -7§iduced by the conversion of guanylate residues to functional 0 0 o 0 pyrinudine analogues (probably cytosine analogues). Until more discriminating tester strains are developed, however, the in- 3D0 4.0 5.0 6.0 7.0 8.0 duction of G-C T-A transversions cannot be excluded. 3.0p4.0H6. Misrepair mutagenesis (8, 9) cannot contribute significantly pH to heat mutagenesis because of the limited base pair specificity FirG. 2. Heat-induced reversion rate of amE509 as a function of of the latter, compared to the ability of misrepair to generate pH.E3amples were heated at 550 in buffers of the following composi- both base pair substitutions at A-T sites and frameshift muta- tion: 10 mM sodium phosphate (pH 5.9-7.4), 10 mM sodium acetate tions. Furthermore, the mottled plaques (mottling indicates 4 and 10 mM (pH 4.0-5.5), sodium citrate (pH 3.5). A total sodium ion rir + heterozygosity in the parantal particle) produced by entration of 20 mM was in cases conch maintained all by the addition heated+ parozclesit2 inthe hgr partice) rodutants of NsaCi. 0.05 mM MgCl2 was also included in all cases. Error flags ted T4 particles (2) contain higher proportions of r mutants indicate one standard deviation. than do those produced by misrepair mutagenesis. Depurination, followed by the random insertion of a pro- on a mixed indicator by preadsorbing to CR63 amber-sup- geny-strand base during DNA replication, has frequently been presssing cells (0.4 ml at 108 cells per ml) at low multiplicities, invoked to explain the mutagenic action of heat (and acid) adding nonsuppressing BB cells (0.2 ml at 108 cells per ml), (10-14). Some of our results are at least superficially consistent platiing the mixture on enriched H agar using enriched H su- with this hypothesis. The reversion of amE509 is promoted by pers(oft top agar, and incubating at 37°. Amber mutants pro- hydrogen ions at weakly acidic pH values, and displays a duce strongly haloed plaques readily distinguishable from all pH-independent mechanism which becomes dominant near othe:r genotypes. Tests for suppressors were conducted during neutral pH values; and this pattern is similar to that observed the cCollection of the 250 and 550 data points of Fig. 1, the pH for depurination (15, 16). The activation energies of reversion 4.0,(6.9, and 7.4 points of Fig. 2, and the 0.05 and 10 mM points for rUV74 and amE509 are about 145 kJ/mol, and that of of TEable 4. Fewer than 0.5% of these revertants contained ex- depurination is about 130 kJ/mol (15). Many of our results, tractistronic suppressors. however, are inconsistent with such a simple random-insertion M[utagenic Efficiency. Let G* denote a heat-modified model. The model fails to explain the observation that A-T base guanylate residue capable of pairing with a purine during DNA pairs are refractory to heat mutagenesis; although earlier reports repliication, and consider the consequences of infecting a per- suggested that guanine might be lost selectively early in the missiive cell with a heated rII amber mutant containing G*-C course of depurination (14, 17), this selectivity was not con- at thoe reverting site. If G* always pairs with a purine, then the firmed in a more recent comprehensive study (15), in which phagre-cell complex will release equal numbers of mutant and guanine was observed to depurinate only some 1.5-fold more reveirtant progeny. If G* usually pairs with a pyrimidine, but rapidly than adenine. Furthermore, the ability of heated rUV74 pairs with a purine with a constant low probability at each DNA particles to express revertants upon immediate plating on a replhication, the complexes will release clones of diverse sizes; nonpermissive host requires that the primary lesion be tran- specifically, clones comprising '/2n of the total burst size (n = scribed as a pyrimidine before DNA replication can occur,

2, 3, 4 ... .) will appear with equal frequency. While such a whereas depurinated sites are not detectably transcribed by E. distribution would indicate that G* infrequently pairs with a coli RNA polymerase (18), at least in vitro. purbne, it would not reveal the actual efficiency of mispair- A comparison of the transversion and depurination reactions ing. nevertheless suggests that, although distinct, they may share Diistributions of revertant clone size produced by heated a common intermediate. A reaction that may mediate the rP7AIM particles have been measured, and were of the general mutagenic process is purine N-glycosidic bond migration (19). type expected from a mispairing frequency of 0.3 or less. (The P.M.B. has constructed a space-filling model with minimal aver'age induced revertant clone comprised about 10% of the distortion of the DNA double helix incorporating N2-#-D- average total burst size of about 132. Further details of these deoxyribofuranosyl guanine paired with normal guanine (Fig. expe,riments will be reported subsequently.) Numerous sto- 3). Since the corresponding "neoguanosine" ribonucleotide is formed during the acid hydrolysis of RNA (20, 21), this reaction 'able 4. Effect of Mg2 + concentration on the rate of represents a possible mechanism for the heat-induced T heat-induced reversion of amE509 transversion G-C C-G. The mutagenic efficiency of "G*" is clearly less than unity, Revertants per 101 but might be as high as 30%. This perceived efficiency reflects MgCl2 (mM) survivors per hr at least three factors: the innate mispairing probability, the frequency with which mispairs are rejected by the DNA 0.05 5.4 1.3 polymerase, and the probability of their removal by excision ± * repair. Excision repair does not affect the shape of the clone size ± 10 1.5 0.5 distribution if the mispairing efficiency is high, but causes a

Sa,mples were heated at 55° in 10 mM sodium pphosphate, pH 6.9 selective loss of smaller clones if the mispairing efficiency is low plus the indicated MgCl2 concentration. Revertants are two (the final shape depends upon the specific excision probability stan(dard deviations. function). The observed clone size distribution is more com- Downloaded by guest on September 30, 2021 Genetics: Bingham et al. Proc. Natl. Acad. Sci. USA 73 (1976) 4163

o 2. Drake, J. W. & McGuire, J. (1967) Genetics 55,387-398. 3. Champe, S. P. & Benzer, S. (1962) Proc. NatI. Acad. Sci. USA 48, 532-546. 4. Brenner, S., Stretton, A. 0. W. & Kaplan, S. (1965) Nature 206, 994-998. 5. Orgel, A. & Brenner, S. (1961) J. Mol. Biol. 3,762-768. 6. Benzer, S. & Champe, S. P. (1962) Proc. NatI. Acad. Sci. USA 48, 1114-1121. 7. McClain, W. H., Guthrie, C. & Barrell, B. G. (1973) J. Mol. Biol. 81, 157-171. FIG. 3. The base pair composed of normal guanine (right) and 8. Drake, J. W. (1973) Genetics (Suppl.) 73, 45-64. N2-/3-D-deoxyribofuranosyl guanine (left). 9. Green, R. R. & Drake, J. W. (1974) Genetics 78, 81-89. 10. Zamenhof, S. & Greer, S. (1958) Nature 182,611-613. patible with inefficient mispairing (including the effects of 11. Freese, E. (1959) Brookhaven Symp. Biol. 12, 63-75. polymerase copyediting) than with efficient mispairing mod- 12. Zamenhof, S. (1960) Proc. Natl. Acad. Sci. USA 46, 101-105. ified by excision repair. Cytosine itself may therefore in some 13. Freese, E. B. (1961) Proc. Natl. Acad. Sci. USA 47,540-545. 14. Greer, S. & Zamenhof, S. (1962) J. Mol. Biol. 4, 123-141. sense pair with G* (and might, for instance, form a single hy- 15. Lindahl, T. & Nyberg, B. (1972) Biochemistry 11, 3610-3618. drogen bond with neoguanosine). 16. Zoltewicz, J. A., Clark, D. F., Sharpless, J. W. & Grahe, G. (1970) Extrapolation from T4 to man (1) indicates that heat con- J. Am. Chem. Soc. 92, 1741-1750. stitutes a substantial mutagenic challenge, namely, a mutation 17. Tamm, C., Hodes, M. E. & Chargaff, E. (1952) J. Biol. Chem. rate (before the intervention of repair processes) on the order 195,49-63. of 100 events per diploid cell per day. Further, comparison of 18. Mamet-Bratley, M. D. (1974) Biochim. Biophys. Acta 340, reversion data herein and those published previously (1) 237-243. suggests that the contribution of the transversion and transition 19. Miyaki, M. & Shimizu, B. (1970) Chem. pharm. Bull. 18, processes to this total may be comparable. 1446-1456. 20. Hemmens, W. F. (1964) Biochim. Biophys. Acta 91,332-334. We are grateful to Linda Cossins and Gary Johnson for technical 21. Shapiro, R. & Gordon, C. N. (1964) Biochem. Biophys. Res. assistance. This work was supported by Grant VC5 from the American Commun. 17,160-164. Cancer Society, Grant BMS71-01243 from the National Science 22. Drake, J. W. (1963) J. Mol. Biol. 6,268-283. Foundation, and Grant 04886 from the National Institute of Allergy 23. Drake, J. W. & McGuire, J. (1967) J. Virol. 1, 260-267. and Infectious Diseases. 24. Smith, M. D., Green, R. R., Ripley, L. S. & Drake, J. W. (1973) Genetics 74, 393-403. 1. Baltz, R. H., Bingham, P. M. & Drake, J. W. (1976) Proc. Natl. 25. Ripley, L. S. (1975) Mol. Gen. Genet. 141, 23-40. Acad. Sci. USA 73, 1269-1273. 26. Drake, J. W. (1967) Proc. NatI. Acad. Sci. USA 55,738-743. Downloaded by guest on September 30, 2021