Proc. Natl. Acad. Sci. USA Vol. 89, pp. 1159-1163, February 1992 Genetics Mechanism of SOS mutagenesis of UV-irradiated DNA: Mostly error-free processing of deaminated cytosine (A-rule/umusDC/groE/pyrimidine dimers/phages S13 and A) IRWIN TESSMAN, SHI-KAU Liu, AND MATTHEW A. KENNEDY Department of Biological Sciences, Purdue University, West Lafayette, IN 47907 Communicated by R. B. Setlow, November 8, 1991 (receivedfor review July 16, 1991)

ABSTRACT We measured the kinetics of growth and correctly during bypass of the dimers, adenines would be mutagenesis of UV-irradiated DNA of phages S13 and A that found in the newly synthesized complementary strand. If were undergoing SOS repair; the kinetics strongly suggest that there is adequate time for to occur, and that is most of SOS mutagenesis arises from the deamination of the very central issue here, bypass of lesions will usually cytosine in cyclobutane pyrimidine dimers, producing C -3 T involve the of the sequence A-A opposite cyclob- transitions. This occurs because the SOS mechanism bypasses utane dimers, even opposite those dimers that had originally T'T dimers promptly, while bypass of cytosine-containing contained C. Because of the distorted template, partially dimers is delayed long enough for deamination to occur. The relaxed proofreading might be needed to facilitate the accu- are thus primarily the product of a faithful mech- rate bypass (6, 7). anism of lesion bypass by a DNA polymerase and are not, as The foundation for the current work lies in our study of had been generally thought, the product of an error-prone mutagenesis by delayed photoreactivation of UV-irradiated mechanism. All of these observations are explained by the phage S13 and its naked DNA (8). In that work we attributed A-rule, which is that adenine nucleotides are inserted nonin- all of the mutagenesis to the deamination of cytosine, which structionally opposite DNA lesions. occurs readily when the 5,6 bond of the base is saturated (9, 10), as it is in cyclobutane dimers. We proposed, therefore, SOS repair and SOS mutagenesis were discovered nearly 40 that most SOS mutagenesis of UV-irradiated DNA might Jean result from error-free translesion DNA synthesis after deam- years ago by Weigle, who observed that preirradiation ination occurs. The kinetics ofthe mutagenic effect produced of the bacterial host increased survival of UV-irradiated by delayed photoreactivation is an unusual step function: phage A (1). When applied to phage DNA these SOS phe- storage of UV-irradiated phage for 15 min at 37°C is not nomena are now called Weigle reactivation and Weigle mutagenic, indicating that deamination is negligible in that mutagenesis; their mechanism has thus been an intriguing time, but an additional 15 min of storage is highly mutagenic, puzzle for some time. We show here that the primary, but not suggesting that deamination of practically all available sites the sole, mechanism of UV mutagenesis in phages S13 and A occurs by 30 min (8). More detailed experiments with infec- is simply the spontaneous deamination of cytosine in cyclo- tious S13 DNA show that the from negligible to butane pyrimidine dimers. This deamination occurs during a nearly complete mutagenesis occurs between 25 and 30 min deliberate, and arguably an unnecessary, delay in the bypass ofDNA storage at 37°C (unpublished data). Earlier studies of of the lesion by DNA polymerase; the delay is caused by mutagenesis by delayed photoreactivation had already indi- cytosine-containing dimers, possibly because of the mispair- cated that dimerized cytosines in the double-stranded DNA ing of the cytosine with adenine. of E. coli are deaminated in vivo within 90 min at 37°C (11). SOS repair is induced by damage to DNA. In Escherichia We predicted that if a sufficiently long delay were an intrinsic coli, the damage results in an activated form of the RecA aspect of the SOS repair process, deamination of the cy- protein, which can mediate the cleavage of LexA, the re- tosines in UV-induced dimers could have time to occur and pressor ofthe SOS regulon. The principal known components would then be a source of the C -+ T mutations that are so of the repair system are the products of the recA and the commonly induced by UV irradiation. We show here that umuDC genes. The activated RecA protein is needed for Weigle mutagenesis is indeed characterized by a delay in cleavage of the UmuD protein to produce UmuD', the active repair that is on the order of the time needed for deamination C-terminal fragment (2-4). The RecA protein can also be of dimerized cytosines. activated by mutations designated recA(Prtc) to what is Weigle reactivation of phage S13 occurs effectively in a termed the protease constitutive state (5); with mutants such groE mutant, but 80% of the mutagenesis is eliminated (7). as recAl202(Prtc) and recAJ237(Prtc), which were used in the We will describe here a recA mutant that accomplishes work described here, Weigle reactivation is achieved without essentially the same thing, namely, efficient reactivation of irradiation of the host cell. S13, and of phage A too, but with severely reduced muta- It is generally thought that a high frequency of erroneous genesis. Thus, mutagenesis is not an essential aspect of SOS base pairing during DNA synthesis opposite the distorted repair. We will explain the effect of these novel mutants by lesion is responsible for the mutagenesis that accompanies showing that they are characterized by reactivation of UV SOS repair, and that idea has evoked the term error-prone lesions in a time too short for deamination to occur. repair. We propose, instead, that most, though certainly not One of us has proposed a theory of the mechanism of all, ofthe mutations arise through accurate base pairing-i.e., Weigle reactivation (12) that is sometimes called the A-rule. an error-free bypass mechanism. This paradox will be re- It was proposed in order to explain the specificity of SOS solved by the fact that mutations can arise by deamination of mutagenesis of UV-irradiated DNA, which was first demon- cytosine in cyclobutane dimers; ifthe uracils formed are read strated in phage S13 (13); since the TT dimer is expected to be the major dimer formed, it is notable that the major The publication costs of this article were defrayed in part by page charge mutational change is nevertheless C -* T. That striking result payment. This article must therefore be hereby marked "advertisement" could be explained if the SOS system blindly inserts adenine in accordance with 18 U.S.C. §1734 solely to indicate this fact. nucleotides opposite pyrimidine dimers regardless ofthe base 1159 Downloaded by guest on September 28, 2021 1160 Genetics: Tessman et al. Proc. Natl. Acad. Sci. USA 89 (1992) composition of the dimers, a process that would ensure RESULTS proper pairing for most of the dimerized sites. A similar A-rule has been proposed to explain the successful bypass of The experimental approach was simple. The latent period apurinic/apyrimidinic sites in DNA (14, 15). We will see that between infection and phage burst indicated the time delay associated with the of a lesion. We will see that the A-rule not only explains the C -- T specificity of SOS bypass mutagenesis, but that it also provides a tidy explanation for unirradiated phage S13 has a latent period of 15-20 min. Of the critical delay in the dimer bypass that is a key to the C this, <2 min is needed to synthesize the complement of the T mutations. infecting DNA molecule (20). In the case of UV-damaged DNA we make the reasonable assumption that once the lesion is bypassed and the complement completely synthe- MATERIALS AND METHODS sized, the phage will require no more than an additional 15-20 Strains and Medium. The bacterial strains used in this min to produce a burst. Therefore, ifL is the measured latent study were isogenic derivatives of the E. coli K-12 strain period ofreactivated phage, then L - 20 is the minimum time AB1157. All strains contained the sulAIl dinDl::Mu d(lac) needed for the bypass of the lesion. If this delay time is >30 alleles. The two high-copy-number plasmids pSE117 min, then deamination ofcytosines containing a saturated 5,6 (umuD+C+) and pGW2123 (umuD) were provided by G. C. bond should occur. Walker (4, 16). The original groES30 strain was obtained We first measured the latent period of unirradiated S13 in from C. P. Georgopolous (University of Utah). The chromo- some of the strains of bacteria that were used for Weigle somal recAl202 and recAl237 strains were constructed by reactivation and found (Table 1, lines 1-3) that the latent crossing the ArecA1202 and ArecA1237 phages with a recA' period was between 15 and 20 min, in agreement with A(gal-attA-bio) strain. All ArecA lysogens contained the previous data (22). The latent period of irradiated phage chromosomal ArecA306 allele (5); the ArecAl237 strains reactivated in strain IT1993, which contains constitutively also activated RecA, was between 115 and 130 min (line 4). We IT3663 and IT3665 contained the IexA(Def)71::TnS al- infer that the delay before the initial bypass occurred was lele. Tryptone broth contained 13 g of Bacto-tryptone and 7 between 95 and 110 min, which should allow adequate time g of NaCl per liter. for dimerized cytosines to be deaminated and thereby be a Determination of Latent Periods for Phages S13 and A. To major source of Weigle mutagenesis. The degree of muta- measure replication delay, bacterial cells were grown in genesis is quantitatively represented in Table 1 by the specific tryptone broth (with 0.2% maltose in the A experiments) to frequency, Ms, which is defined as the number of =4 x 108 per ml. CaCl2 was added to a final concentration of mutations per lesion reactivated and provides the basis for 10 mM before the cells were infected with phage at a comparing the mutagenicity oflesions (7); it is independent of multiplicity of infection of 0.1. For unirradiated phage the the efficiency of reactivation as defined by the repair sector, latent period can be measured in a standard one-step growth W. The value M. 0.07 for temperature-sensitive mutations experiment (17). Determination of the latent period of UV- is about the maximum observed for Weigle reactivation (7, irradiated phage involved bursts from two types of phage: (i) 23). reactivated damaged phage and (ii) the original undamaged At 350C the groES30 mutation eliminates most of the phage. Since the latter was only about 30 times lower in mutagenesis from a ArecA1202 strain without greatly affect- concentration than the reactivated phage under the condi- ing reactivation (7). Accordingly, we observed that the latent tions of our experiments, bursts from undamaged phage period measured in single bursts was shortened to between 30 could mask any delayed bursts. This problem was overcome and 35 min (Table 1, line 5), implying a delay of only 10-15 by diluting the infective centers until a burst tube contained, min before the lesion was bypassed. Our photoreactivation on the average, only about 3 inactivated phage that would be studies (ref. 8 and unpublished data) indicate that no signif- reactivated and 0.1 live phage; since we typically used 10 icant amount ofdeamination occurs in that short time, which burst tubes for a latent period determination, only about 1 of is entirely consistent with the relatively low value of the the 10 would contain a burst characteristic of undamaged specific mutation frequency, M,. At 300C the temperature- phage. These elaborate precautions were not needed when sensitive groE mutation produced no significant reduction of the latent period of irradiated naked DNA was determined. M. and that was matched by the extended latent period of Naked DNA was assayed in spheroplasts (18), and intracel- between 115 and 130 min (line 6). At 35°C, the presence ofthe lular phage production was measured by lysing the sphero- high-copy-number umuD+C+ plasmid pSE117 (pumuD+C+, plasts with CHC13 at various times. Since the repair sector line 7) partially restored the mutagenesis (also see Table 1 of for naked single-stranded DNA is as high as 0.5-0.6 (19), it ref. 7); the small lengthening of the latent period to between was easy to exclude undamaged DNA from the burst tubes 45 and 50 min implies that the bypass was delayed 25-30 min. and still have many reactivated DNA molecules. This time is marginal for deamination and is therefore con- Each latent period is reported as a range; generally, the sistent with the intermediate value of M, that was observed. upper limit was the first time point at which a burst could be The recA1237(Prtc) mutant is unusual. Although it has detected, the lower limit was the previous time point. When weak constitutive protease activity (5), it shows very sub- multiple burst tubes were used, the upper limit was taken as stantial Weigle reactivation (ref. 24 and line 8 of Table 1). the first time point at which >50% of the burst tubes showed There was little mutagenesis, however; the critical value M, a burst, and the lower limit was the previous time point. In was greatly reduced from the value observed here for the nearly all cases, 90-100% of the bursts occurred between recA1202 mutant (line 8 compared with line 4). In full those two times. Two or three experiments were usually used agreement with the reduced mutagenesis in the ArecA1237 to narrow the range. strain was the relatively short latent period of 25-35 min, Weigle Reactivation and Mutagenesis. The recA1202 and which allows the inadequate time of only 10-15 min for recA1237 strains have protease constitutive recA alleles and deamination of cytosine. therefore reactivate the irradiated phages without irradiation What distinctive property of the RecA1237 protein can of the host cells. Temperature-sensitive mutants of S13 and explain why M, was severely reduced even though there was A, unable to grow at 430C relative to growth at 35°C, were effective Weigle reactivation as indicated by the large repair assayed to determine the specific mutation frequency, which sector? Much of the mutagenicity missing from the is defined in Table 1. Because the DNA is single-stranded, ArecAl237 strain was restored in strain 1T3665 (ArecAl237/ UV-damaged S13 cannot be reactivated by excision repair. pumuD', line 9); both strains can also be compared with Downloaded by guest on September 28, 2021 Genetics: Tessman et al. Proc. Natl. Acad. Sci. USA 89 (1992) 1161

Table 1. Latent period of phages S13 and A as a function of Weigle mutagenesis Temper- Repair Specific mutation Latent Strain Relevant genotype ature, 'C UV* sector (W)t frequency (Mj)t period, min S13 phage 1. IT1993 ArecA1202 37 15-20 2. IT3043 ArecA1202 groES30 35 15-20 3. IT3057 ArecA1202 groES30/pumuD+C' 35 15-20 4. IT1993 ArecA1202 37 G 0.28 0.067 ± 0.009 115-130 5. IT3043 ArecA1202 groES30 35 G 0.25 0.015 ± 0.004 30-35 6. IT3043 ArecA1202 groES30 30 G 0.27 0.065 ± 0.009 115-130 7. IT3057 ArecA1202 groES30/pumuD'C+ 35 G 0.29 0.036 + 0.005 45-50 8. IT3663 ArecA1237 37 G 0.23 0.019 ± 0.008 25-35 9. IT3665 ArecA1237/pumuD' 37 G 0.24 0.043 ± 0.007 90-100 10. EST2422§ ArecA+ groE+ 35 G 0.23 0.066 ± 0.009 100-115 11. IT2941§ ArecA+ groES30 35 G 0.10 0.030 ± 0.005 40-55 A phage 12. IT2906 recA+ uvrA294 35 50-55 13. IT2905 recA1202 uvrA294 35 50-55 14. IT2905 recA1202 uvrA294 35 G 0.17 0.059 ± 0.012 105-110 15. IT3483 recA1237 uvrA294 35 G 0.21 0.018 ± 0.005 80-90 S13 DNA1 16. IT2228 ArecA1202 37 +A 0.51 0.020 ± 0.005 10-15 17. IT2228 ArecA1202 37 -A 0.52 0.063 ± 0.008 50-55 18. IT2228 ArecAJ20211 37 -A 0.53 0.063 ± 0.008 10-15 *Blank, unirradiated phage; G, phage irradiated by a germicidal lamp with a flux of 180 J/m2 (S13 survival - 10-7) or 120 J/m2 (A survival 10-5); A, irradiated 25 min at 25°C at wavelengths .290 nm with (+A) or without (-A) 0.2% acetophenone (8). Irradiation of naked DNA with long-wavelength UV light in the presence of acetophenone provides a test of the effect of cytosine-containing dimers (21). tBecause the survival curves are exponential, the repair sector can be unambiguously defined as W = 1 - log Sa/log Sb, where Sa and Sb are the fractions of surviving after (with) and before (without) reactivation. Sb was measured in the reference strain IT1865, for which W = 0. The value of W represents the fraction of lethal damages that are repaired. For S13, Sb was _10-7 except for one case (IT3483, 2 x 10-6); for each strain infected with A, two values of Sb were used: 1 x 10-5 and 3 x 10-3. For S13 DNA irradiated with wavelengths .290 nm, Sb was 2 x 10-4. The standard deviation of W was ±0.01. Values of W and M. for IT1993, IT3043, IT3057, EST2422, and IT2941 are from Liu and Tessman (7). The other values of W are new. W. = M/(- W ln Sb); it is the number of mutations per lesion repaired (7). M is the number of temperature-sensitive mutations per scored at 43°C relative to growth at 35°C and was corrected for multiple mutations. The error in Ms was derived from the errors in the component quantities. §Cells UV-irradiated with 50 J/m2. 1Assayed in spheroplasts (18). The reference strain (W = 0) for measurement of survival was EST1262 (ArecA+). IIrradiated DNA stored 1 hr at 37°C before of spheroplasts. strain IT3664 (ArecA1202/pumuD', not in Table 1), for which 14). The increased latent period of A under conditions of M, = 0.062 ± 0.009. The restoration of mutagenesis by the Weigle reactivation has been seen previously (25). Not sur- umuD' plasmid implies that inadequate cleavage of the prisingly, M. was about the same as for S13. As with the UmuD protein was responsible for lowering the specific single-stranded DNA of S13, mutagenesis of the double- mutagenicity of the recA1237 mutant without having much stranded DNA of A correlated with the extension ofthe latent effect on the repair sector. Such a defect in cleavage ofUmuD period. We also see that the recAl237 allele affected UV- could also be suspected because the recA1237 strain is known irradiated A DNA in the same way as it did the S13 DNA: to have weak constitutive protease activity (5). In the case of Weigle mutagenesis, as indicated by M., was substantially IT3665, we see once again that increased mutagenicity was reduced whereas reactivation (W) was unaffected (line 15). accompanied by an increase in the latent period of the Accordingly, the latent period in the recAl237 strain was reactivated phage (line 9), in complete agreement with our 15-20 min shorter than in the recAl202 strain. proposed mechanism of mutagenesis by deamination. We determined what we infer to be the kinetics of cytosine Up to this point, activation of the RecA protein and deamination in UV-irradiated encapsidated double-stranded subsequent induction of the SOS system was accomplished A DNA, as we did previously for the single-stranded DNA of by recA mutations that produce constitutive protease activ- S13 (8), by measuring the specific mutation frequency as a ity. The latent period of reactivated phage was also length- function of storage time before photoreactivation (Fig. 1). As ened significantly when the RecA' protein was activated by with single-stranded DNA, the kinetics followed a step UV irradiation, a result consistent with the large value of M. function: for 50 min there was apparently hardly any deam- (line 10). In the irradiated strain IT2941, the groES30 muta- ination, but in the next 5 min nearly complete deamination tion reduced the latent period at 350C to a time that is marginal seemed to occur. Thus, deamination of cytosine in irradiated for deamination (line 11), which was expected in view of the double-stranded DNA appears to take about twice as long as partial reduction of Ms. Thus for the single-stranded DNA it does in single-stranded DNA. The extended length of the phage there is a complete correlation between the length of A latent period under conditions of Weigle mutagenesis (line the latent period and the degree of mutagenesis. 14) is consistent with the time needed in vitro for deami- We briefly examined the same phenomenon in the double- nation. stranded DNA ofphage A. Excision repair was eliminated by To confirm that cyclobutane dimers were responsible for the uvrA294 allele. Unirradiated A showed a latent period of most of the observed mutations, we examined the ability of 50-55 min (lines 12 and 13). The Weigle reactivation and photoreactivation to reverse the mutagenesis. S13 irradiated mutagenesis of irradiated A in a recAJ202(Prtc) strain was to a survival of lo-7 was plated without delay on the accompanied by a latent period extension of 55-60 min (line ArecAl202(Prtc) strain IT1993 and immediately placed under Downloaded by guest on September 28, 2021 1162 Genetics: Tessman et al. Proc. Natl. Acad. Sci. USA 89 (1992) specific mutation frequency, Ms, are negligible in recA(Def) or in umuC(Def) cells (refs. 23, 24, and 27 and unpublished data). 0 Of special interest was the behavior of the strain (IT3663) 4-Co 0.030 containing the recAJ237(PrtC) allele, which confers weak constitutive protease activity (5). The mutant has a split SOS < 0.020 L phenotype similar to that of the groES30 strain: it provides highly effective Weigle reactivation of S13 (24) and A (Table 0" 0.010 1) but produces a greatly reduced specific mutation fre- quency. The ability of the umuD' plasmid to restore muta- genesis to the recAl237 strain suggests that reactivation and 0.OOG 0 50 100 mutagenesis differ somewhat in their requirements for the UmuD' protein. Storage time (min) The A-rule-that adenine nucleotides are inserted nonin- FIG. 1. Mutation to temperature sensitivity by delayed photore- structionally opposite the UV lesion in the newly synthesized activation as a function of preinfection storage time of UV-irradiated DNA strand (12)-provides an explanation for the apparent phage A. delay in SOS bypass ofthe lesion: DNA synthesis stalls when the dimer contains the noncomplementary cytosine. Bypass a bank of six 15-W blacklight lamps (8). The combined repair may fail either because the proofreading function of the sector for photoreactivation and Weigle reactivation was polymerase repeatedly removes the mispaired adenine nucle- 0.32; the value of M, was reduced by the photoreactivation otide (6) or because the polymerization step is inhibited by from 0.067 + 0.014 to 0.019 + 0.005. Cyclobutane dimers are the mispairing (28). The reason the delay seems to match the the only UV lesions known to be photoreactivated (26); thus time needed for deamination may simply be that DNA roughly 70%, or more, of the specific mutation frequency synthesis remains stalled until the dimerized cytosines are associated with Weigle reactivation can be attributed to these deaminated. There would then seem to be some discrepancy dimers. between the time measured in vitro for deamination to occur Finally, we investigated an idea that the A-rule might be (=30 min in single-stranded DNA) and the bypass delay time responsible for the delayed bypass in an unusual way. Since (-100 min). This could be explained if the time needed for the A-rule implies an A-C mismatch at cytosine-containing deamination were lengthened by the intracellular environ- dimers but not at T1T dimers, we thought that the bypass ment of the lesion, reminiscent of the inhibition of deami- delay might be caused by the mismatch and therefore occur nation in S13 DNA by the viral capsid (8) and in E. coli DNA only at cytosine-containing dimers. When naked S13 DNA by DNA photolyase (11). was irradiated with wavelengths of .290 nm in the presence Fig. 2 illustrates our concept of lesion bypass under the of the photosensitizer acetophenone to produce practically A-rule. The UmuDC proteins appear to be key components only TFT dimers, as in the mutagenic photoreactivation of the SOS repair complex in the implementation of the in trans- A-rule; they are implicated by the effects of the groES30 and experiments (8), intracellular phage appeared the recA1237 mutants. The groES defect works through some fected spheroplasts promptly between 10 and 15 min after deficiency produced in the UmuDC proteins (29, 30), and we infection (line 16). But for DNA irradiated in the absence of have shown here that the recA1237 defect specifically in- acetophenone, which results in C-containing as well as T^T volves the UmuD' component. We suggest that the UmuD' dimers, intracellular phage did not appear until 50-55 min C complex interacts with PolC to create a less fastidious after infection of the spheroplasts (line 17). Therefore, the polymerase, particularly one that might insert adenine nu- delay does indeed seem to be due to the presence ofcytosine- containing dimers. When we stored the DNA that had been SOS repair irradiated in the absence of acetophenone for 1 hr at 37°C, ~ .-V complex there was no delay in the appearance of intracellular phage TT TA (line 18). That was the expected result because the storage _- --A _ --A time would allow all the cytosines in pyrimidine dimers to be deaminated to yield dimers that would pair correctly with the adenine nucleotides. The values of in the last opposing M. Bypassed S three lines of Table 1 are also in complete agreement with talled (>30 min) expectations. These results with acetophenone are consistent by mismatch with the A-rule and reinforce the conclusion that it is the presence of cytosine in the dimers that causes the bypass delay and most of the mutagenesis. TT TU A A -MM DISCUSSION Our previous study of the mutagenesis that accompanies delayed photoreactivation of UV-irradiated S13 DNA led us, Bypassed | by analogy, to the idea that if an appropriate time delay were promptly somehow intrinsic to the SOS repair process, the delay would provide an opportunity for the spontaneous deamination of cytosine in cyclobutane dimers, and mutagenesis would result (8). This idea would explain most, though certainly not - rTU_A all, of SOS mutagenesis. The results observed here with SOS _- -- A- repair ofUV-irradiated S13 were in complete accord with our expectations. The reactivation and mutagenesis studied here were indeed a result of SOS processing, inasmuch as the FIG. 2. A-rule model of lesion bypass. Broken line represents Weigle reactivation repair sector, W, and the accompanying newly synthesized DNA. Downloaded by guest on September 28, 2021 Genetics: Tessman et al. Proc. Natl. Acad. Sci. USA 89 (1992) 1163

cleotides noninstructionally opposite a lesion. The behavior 1. Weigle, J. J. (1953) Proc. Natl. Acad. Sci. USA 39, 628-636. of the groES30 and recA1237 mutants also demonstrates that 2. Shinagawa, H., Iwasaki, H., Kato, T. & Nakata, A. (1988) effective SOS repair is possible without a substantial delay Proc. Natl. Acad. Sci. USA 85, 1806-1810. and without the accompanying mutagenesis. To be consistent 3. Burckhardt, S. E., Woodgate, R., Scheuermann, R. H. & with our theory of the delay, we suggest that repair without Echols, H. (1988) Proc. Natl. Acad. Sci. USA 85, 1811-1815. the delay may occur if the polymerase in the groES30 and 4. Nohmi, T., Battista, J. R., Dodson, L. A. & Walker, G. C. recA1237 mutants correctly inserts a guanine nucleotide (1988) Proc. Natl. Acad. Sci. USA 85, 1816-1820. opposite a cytosine in the lesion; thus no delay in bypass and 5. Tessman, E. S. & Peterson, P. (1985) J. Bacteriol. 163, 677- no mutagenesis would occur. It is notable, therefore, that 687. 6. Villani, G., Boiteux, S. & Radman, M. (1978)Proc. Natl. Acad. under SOS conditions the cell deliberately imposes what Sci. USA 75, 3037-3041. seems at first glance to be an unnecessary delay in repair. One 7. Liu, S.-K. & Tessman, I. (1990) J. Mol. Biol. 216, 803-807. way to view this delay is to see it as part of a strategy to 8. Tessman, I. & Kennedy, M. A. (1991) Mol. Gen. Genet. 226, increase the mutation frequency when the cell is subjected to 144-148. adverse environmental conditions to which it must adapt (7, 9. Green, M. & Cohen, S. S. (1958) J. Biol. Chem. 228, 601-609. 31, 32). 10. Douki, T., Voituriez, L. & Cadet, J. (1991) Photochem. Pho- The A-rule was independently applied to explain the mu- tobiol. 53, 293-297. tations that accompany the SOS bypass of abasic lesions (14, 11. Ruiz-Rubio, M., Yamamoto, K. & Bockrath, R. (1988) J. 33), and the rule also accounts for the specificity of sponta- Bacteriol. 170, 5371-5374. neous mutations that are RecA(Prtc)-mediated (unpublished 12. Tessman, I. (1976) in Bacteriophage Meeting, eds. Bukhari, data). Thus, the A-rule may well account for most aspects of A. I. & Ljungquist, E. (Cold Spring Harbor Lab., Cold Spring SOS mutagenesis. Harbor, NY), p. 87. Clearly, the term "error-prone" repair, when generally 13. Howard, B. D. & Tessman, I. (1964) J. Mol. Biol. 9, 372-375. applied to the SOS process, may be a misnomer. However, 14. Sagher, D. & Strauss, B. (1983) Biochemistry 22, 4518-4526. not all UV-induced mutations can be attributed to deami- 15. Strauss, B. S. (1991) BioEssays 13, 79-84. nation of cytosine; roughly 30% of the SOS-induced muta- 16. Marsh, L. & Walker, G. C. (1985) J. Bacteriol. 162, 155-161. tions of S13 are not C T (13). It is striking that this 17. Ellis, E. L. & Delbruck, M. (1939) J. Gen. Physiol. 22, 365-384. percentage is about the same as the residual specific muta- 18. Tessman, I., Morrison, H., Bernasconi, C., Pandey, G. & Ekanayake, L. (1983) Photochem. Photobiol. 38, 29-35. genesis seen in the groES30 and recA1237 strains (Table 1) as 19. Tessman, 1. (1990) J. Bacteriol. 172, 5503-5505. well as that observed after photoreversal of Weigle muta- 20. Tessman, E. S. (1966) J. Mol. Biol. 17, 218-236. genesis. Although a major portion ofSOS mutagenesis can be 21. Fix, D. & Bockrath, R. (1981) Mol. Gen. Genet. 182, 7-11. attributed to an error-free mechanism involving the pairing of 22. Zahler, S. A. (1958) J. Bacteriol. 75, 310-315. adenine with uracil, a significant minor portion may yet be 23. Tessman, I. (1985) Proc. Natl. Acad. Sci. USA 82, 6614-6618. produced by an error-prone mechanism involving misincor- 24. Tessman, E. S., Tessman, I., Peterson, P. K. & Forestal, J. D. poration errors that fail to be corrected by the proofreading (1986) J. Bacteriol. 168, 1159-1164. function of DNA polymerase III (32, 34). 25. Caillet-Fauquet, P. & Defais, M. (1977) Mutat. Res. 45, 161- UV commonly induces C -* T transitions. It was first 167. observed in UV-irradiated phage T4; deamination of 5-hy- 26. Harm, H. (1976) in Photochemistry and Photobiology of Nu- droxymethylcytosine was offered as a possible explanation cleic Acids, ed. Wang, S. Y. (Academic, New York), pp. for the specificity (35). That possibility was strengthened by 219-263. the in 27. Kuan, C.-T., Liu, S.-K. & Tessman, I. (1991) Genetics 128, evidence for deamination of cytosine UV-irradiated 45-57. double-stranded dI-dC homopolymers (36). SOS mutagenesis 28. Echols, H. & Goodman, M. F. (1991) Annu. Rev. Biochem. 60, of phage S13 produces mostly, but not entirely, C -+ T 477-511. mutations (13). Experiments in a variety of systems, includ- 29. Donnelly, C. E. & Walker, G. C. (1989) J. Bacteriol. 171, ing mammalian cells, have now established that the major 6117-6125. effect of UV-induced mutagenesis is to produce C -- T 30. Liu, S.-K. & Tessman, I. (1990) J. Bacteriol. 172, 6135-6138. mutations (reviewed in ref. 37). Our experiments here and 31. Radman, M. (1980) Photochem. Photobiol. 32, 823-830. previous experiments (8) suggest that cytosine deamination is 32. Lu, C., Scheuermann, R. H. & Echols, H. (1986) Proc. Natl. a likely cause of this effect. Acad. Sci. USA 83, 619-623. 33. Schaaper, R. M., Kunkel, T. A. & Loeb, L. A. (1983) Proc. We dedicate this paper to the memory of Salvador E. Luria, who Natl. Acad. Sci. USA 80, 487-491. died February 6, 1991. The first edition ofhis book General Virology, 34. Fersht, A. R. & Knill-Jones, J. W. (1983) J. Mol. Biol. 165, published in 1953, inspired one ofus (I.T.) to be a biologist. We thank 669-682. C. Georgopolous, from whose laboratory we obtained the groES30 35. Drake, J. W. (1963) J. Mol. Biol. 6, 268-283. allele, and G. C. Walker, from whose laboratory we obtained the 36. Setlow, R. B., Carrier, W. L. & Bollum, F. J. (1965) Proc. plasmids pSE117 and pGW2123. The research was supported by Natl. Acad. Sci. USA 53, 1111-1118. Public Health Service Grant GM35850. 37. Brash, D. E. (1988) Photochem. Photobiol. 48, 59-66. Downloaded by guest on September 28, 2021