Proc. Nati. Acad. Sci. USAVol. 82, pp. 4193-4197, June 1985Genetics

Mutagenic repair in Escherichia coli: Products of the recA geneand of the umuD and umuC genes act at different stepsin UV-induced mutagenesis

(photoreversal/bypass replication/misincorporation)

BRYN A. BRIDGES AND ROGER WOODGATEMedical Research Council Cell Mutation Unit, University of Sussex, Falmer, Brighton, Sussex BN1 9RR, England

Communicated by C. Auerbach, February 25, 1985

ABSTRACT When excision-deficient Escherichia coli car-rying umuC or umuD alleles were exposed to visible light sev-eral hours after ultraviolet irradiation, base-pair-substitutionmutations were induced in these normally non-UV-mutablebacteria. It is argued that delayed photoreversal of pyrimidinedimers removes blocks to DNA replication and allows the"survival" and expression of misincorporated bases. A modelfor UV mutagenesis is proposed with two steps: (i) misincor-poration opposite a photoproduct, which can be mediated di-rectly by RecA protein, and (ii) bypass, only the latter processrequiring umuD+ and umuC+ alleles. Basal levels of geneproducts are sufficient for at least some misincorporationevents, although induced levels of umuD and umuC gene prod-ucts are necessary for the bypass step. umuC bacteria contain-ing the recA441 allele showed a greater yield of mutants, andthose containing recA430 a reduced yield, following delayedphotoreversal. The lexA51 allele (which results in constitutivederepression of RecA protein production) did not significantlyalter the yield of mutants but caused them to appear marginal-ly sooner in a recA441 umuC strain. These results emphasizethat the nature of the RecA protein and not its concentration isparamount in determining the level of misincorporation. Ex-periments with recA441 umuC bacteria at 43°C and 30°C sug-gest that the misincorporation effect is unlikely to be attribut-able to cleavage of a DNA binding protein such as a repressoror a component of the polymerase complex. Moreover, misin-corporation seems to occur without the need for induced syn-thesis of any other protein under recA control.

A number ofDNA repair genes are involved in the inductionof base-pair-substitution mutations in Escherichia coli by ul-traviolet light and most chemical mutagens. The recA geneoccupies a crucial role, as it is activated following DNAdamage to a form that is able to mediate the cleavage of thelexA gene product, a repressor of a number of genes includ-ing umuC and -D, both of which are required for UV muta-genesis (for review, see ref. 1). It has previously been pro-posed (2) that the RecA protein may, in addition to cleavingthe LexA repressor, play a more direct role, in particular bystabilizing the photoproduct-primer terminus configurationso as to allow DNA polymerase to continually insert andthen remove nucleotides without chain elongation. The ele-vated deoxynucleoside monophosphate levels resulting fromthis would then inhibit the 3' to 5' exonuclease proofreadingfunction of the polymerase so that a mismatched base thatbecame inserted would be less likely to be removed (see ref.3).That activated RecA protein has a vital direct role in UV

mutagenesis in addition to its function in derepressing umuCand umuD was established by Blanco et al. (4), who showed

that in a number of strains, the level of activated RecA pro-tein correlated with the level of mutagenesis even in the ab-sence of functional LexA repressor. This excluded the possi-bility that derepression of LexA-repressible genes was re-sponsible.Evidence consistent with these ideas has recently been

published by Fersht and Knill-Jones (5), who showed thatdeoxynucleoside monophosphates, particularly dGMP, in-hibited the proofreading activity ofDNA polymerase III ho-loenzyme with a concomitant decrease in fidelity of DNAreplication. In addition, RecA protein inhibited the proof-reading activity of DNA polymerase III on synthetic tem-plates containing mismatched 3' termini, and the inhibitionwas increased in the presence of dGMP or dAMP; it alsodecreased the fidelity ofDNA replication in vitro. Even thepresence of both purine deoxynucleoside monophosphatesand RecA protein did not, however, enable DNA polymer-ase III to copy past pyrimidine dimers.We have dissected the process of UV mutagenesis into

two steps (6). The first is a misincorporation event, presum-ably opposite a photoproduct such as a pyrimidine dimer,and the second requires the UmuC protein and allows DNAsynthesis to continue beyond the misincorporation (bypass).When the pyrimidine dimer is removed by photoreversalsome time after UV irradiation of a umuC strain, the bypassstep is made redundant and mutations arising from misincor-porations are able to "survive."

In the present work, in which we use strains lacking exci-sion repair, we show that umuD is also involved in the sec-ond step of dimer bypass and that induced levels of umuCand umuD gene products are necessary to mediate this. Weshow that the misincorporation step can be mediated directlyby RecA protein and that uninduced levels are sufficient toenable a detectable proportion of misincorporations to takeplace.

MATERIALS AND METHODS

Bacterial Strains. The strains used are shown in Table 1.CM1140 was made by transducing umuCJ22::Tn5 fromGW2100 (kindly provided by Graham Walker) into WP2uvrA (originally obtained from Ruth Hill); CM1142 was madeby a similar transduction into WP44S NF (equivalent to WP2uvrA recA441 sfiA, originally obtained from Evelyn Witkin).CM1143 was constructed from CM1142 by transducinglexA3 lexASI from DM1187 (given by S. G. Sedgwick), usingselection for malB+ and screening for increased resistance toUV and to 8-methoxypsoralen crosslinks. CM1144 was de-rived from CM1143 by transducing recA430 linked to a TnWOinsertion in srlC from IC1139; transductants selected for tet-racycline resistance were screened for UV sensitivity. Thegenotype of CM1144 was confirmed by transducing recA430from CM1144 into WP2 uvrA and screening tetracycline-re-sistant transductants for immutability to UV. CM1146 was

4193

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

4194 Genetics: Bridges and Woodgate

Table 1. Strains used

Strain Repair markers*

TK610a uvrA6 umuC36TK612a uvrA6 umuD44TK614a uvrA6 umuD77CM611b uvrA155 lexA102 (Ind-)CM114Ob uvrA155 umuCl22::Tn5CM1142b uvrA155 umuCl22::Tn5

recA441 sfiCM1143b uvrA155 umuCl22::TnS

recA441 lexA51CM1144C uvrA155 umuC122::Tn5

recA430 lexA51 sftCM1146C uvrA155 umuC122::Tn5

recA+ lexA51 sfiDM1187d lexA3 lexA51 sfiAllIC1139C uvrB5 recA430 lexA51IC1141e uvrB5 recA+ lexAGW2100f umuC122::Tn5

*Additional markers are given in footnotetAll due to ochre nonsense mutations.tInstituto de Investigaciones CitologicasSP-Valencia, Spain.§Biology Department, Massachusetts Instibridge, MA.

athrl leu6 proA2 his4 ilv325 thil lacYl galstrA31 supE44blon]i sulA32clonll sulA32 srlC300::TnlOdhis4 strA31 recA441ethr thiproA2 his4 argE3 lac Yl galK2 srlC3strA31 sup37 ilv TSfthrl leu6 proA3 his4 thil argE3 lacYl galrpsL31 supE44

derived from CM1143 by transduciimanner from IC1141 and screening foat 43°C attributable to the recA441 a]were carried out with P1 cml clrlOO.

Irradiation. Ultraviolet liaht (UV)

Revertibleauxotrophyt

his4his4his4trpE65f DCnPi

Reference

7778

Thico nenira

calculated according to the formula of Sedgwick and Bridges(11). Incubation temperature for initial cultures and forplates after photoreversal was 370C unless otherwise stated.Experiments were carried out at least three times and wereperformed under yellow light to avoid unwanted photoreacti-vation.

RESULTStrpE65J i nis paper We have previously demonstrated the induction of His' mu-trpE65 This paper tants in a umuC36 strain of E. coli K-12 by giving photore-

trpE65 This paper versing light some hours after exposure to UV (6). We haveinterpreted this to indicate that in the absence of the umuC

trpE65 This paper gene product, DNA synthesis cannot continue beyond cer-tain pyrimidine dimers but that when these are removed byenzymic photoreversal, misincorporated bases that have al-

trpE6S This paper ready been inserted in a UmuC-independent step persist and

9 are seen as mutations. In what follows, mutations seen onlyM. Blancot after delayed photoreversal will be assumed to reflect suchM. Blanco misincorporation events. An almost identical effect for theG. Walker§ induction of Trp' mutants has been demonstrated in a series

*se-f. of experiments with CM1140, a umuC122::Tn5 derivative ofas a-f. WP2 uvrA (data not shown and Figs. 2 and 3). The maximum

effect required at least 4 hr of incubation in minimal mediumAmadeo de Saboya 4, at 37°C after UV and was roughly proportional to the dose of

tute of Technology, Cam- UV. TK612 (umuD44) and TK614 (umuD77), both K-12strains, show a similar effect for the induction of His+ mu-

K2 aral4 xy15 mtll tsx33 tants (Fig. 1). Thus both umuC and umuD gene productsseem to be required for the dimer-bypass step that is pre-sumed to be alleviated by removing the dimer with photore-versing light.

In umuC and -D mutants, these gene products are absent.00::TnlO malE7::Tn5sfiB In lexA(Ind-) bacteria, although these genes cannot be dere-K2 ara14 xy15 m1 tsx33 pressed, basal levels of the products should be present. The

misincorporation step could still be demonstrated in WP2-derived lexA102(Ind-) bacteria by using the delayed photo-reversal approach on plates lacking Casamino acids (Fig. 2).

ng recA+ in a similar Because of its greater UV sensitivity, the dose given tonr loss ofUV resistance CM611 [carrying lexAl02(Ind )] was less than that given toIlele. All transductions CM1140 (carrying umuC122::TnS). This result suggests that

some misincorporations can occur without the need for in-nof nredominant1v 254 duced levels of gene products under the control of the LexAAl I 9UR99ILRVIE %1t IL CL SV 1It s1151t kW Vs/sssasWkF _A1JL1"A'

nm wavelength was obtained from an Hanovia type 12555lamp. The incident fluence was measured with a Latarjetmeter. Photoreversing light was obtained from two parallelOsram 20-W "Warmlight" tubes, 11 cm apart. Bacteria wereexposed on the surface of agar plates at a distance of 10 cmfor 40 min.

Experimental Protocol. Overnight cultures in L broth werediluted approximately 1:50 in L broth and Irown with shak-ing to a concentration between 1 and 2 x 10 bacteria per ml.They then were centrifuged and resuspended in 10 mMMgSO4 for exposure to UV. After irradiation, K-12 strainswere seeded on Davis and Mingioli (10) salts plus glucose(0.4% wt/vol); agar (1.5% wt/vol); proline, threonine, va-line, leucine, and isoleucine (all at 100 pg/ml); thiamine (0.25pg/ml); and histidine (1 pg/ml). For strains derived fromWP2, the supplements were (except where stated otherwise)tryptophan (0.25 pg/ml), adenine (75 Ag/ml) and Casaminoacids (Difco) (0.4%). On these plates, 10 or less viable bac-teria grew on the low level of histidine or tryptophan to formsmall detectable colonies after 2 days. When up to 4 x 107bacteria were seeded, they grew to form a lawn, concomi-tantly exhausting the low level of required amino acid. Histi-dine- or tryptophan-independent mutants grew up throughthe lawn and were counted after 4 and 3 days, respectively.The number of preexisting spontaneous mutants was deter-mined by seeding on minimal glucose plates lacking all sup-plements, except required amino acids other than histidinefor the K-12 strains. The induced mutation frequencies were

-o-.

*-

2 3 45 1UV, J/m2

FIG. 1. Induction of His+ mutants in umuD strains ofE. coli as afunction of initial UV exposure. After UV treatment, cells on mini-mal agar plates were incubated for 4 hr and then exposed to visiblelight for 40 min. (A) TK614 (uvrA6 umuD77). (B) TK612 (uvrA6umuD44). Broken line in both A and B is curve for TK610 (uvrA6umuC36) (from ref. 6). Each point represents the mean (±SEM) ofthree independent experiments; lines were fitted by eye.

Proc. Natl. Acad Sd USA 82 (1985)

Proc. Natl. Acad. Sci. USA 82 (1985) 4195

1-

x 8ur_

0

ce

v

`~ 6._

2

4-

E

- 4

0

0 1 2 3 4Time after UV, hr

5

FIG. 2. Mutagenic effect of delayed photoreversal of CM1140(uvrA155 umuC122) (i) and CM611 (uvrA155 IexA102) (v) as a func-tion of time of incubation at 370C on minimal plates supplementedwith tryptophan (0.75 Mg/ml) after UV (1 and 0.4 J/m2, respective-ly). Each point represents the mean (±SEM) of three independentexperiments.

repressor. The result also implies that, in contrast, full dimerbypass requires elevated (induced) levels of UmuC andUmuD proteins.The hypothesis that the misincorporation is mediated by a

gene under LexA repression was examined by carrying outdelayed-photoreversal experiments with a umuC strain car-

rying recA441, a temperature-sensitive allele formerly knownas tif-1, the product of which causes cleavage of LexA re-pressor at 430C without needing the usual SOS (i.e., dam-age-induced) induction signal. In this strain the productsof recA441 itself and the other LexA-repressible genes are

30 -

x

C

20-

0~

lo-

0

0 1 2 3 4 5Time after UV, hr

FIG. 3. Mutagenic effect of delayed photoreversal of CM1140(uvrA155 umuC122) (A), CM1142 (uvrA]55 umuC122 recA441) (v),CM1143 (uvrA155 umuC122 recA441 lexA51) (o), CM1144 (uvrA155umuC122 recA430 lexA51) (A), and CM1146 (uvrA]55 umuC122lexA51) Mo as a function of time of incubation on plates at 300C afterI J/m0of UV (plating medium was as described in the text5Initialand final incubation temperature was also30oC. Each point repre-sents the mean of three independent experiments. Standard errorsare omitted (for clarity).

present at elevated levels within 30 min or so of incubating at430C, particularly when adenine is present. In these experi-ments we used plates supplemented with Casamino acidsand found that the misincorporation step occurred much ear-lier than on minimal plates (Fig. 3). The mutant yield withthe recA441 umuC strain incubated after UV at 30'C wasabout 1.6 times the yield with the recA' umuC strain, sug-gesting that the misincorporation step is mediated by RecAprotein, either directly or via inactivation ofLexA protein orsome other repressor. Consistent with this conclusion is theobservation that mutants appear even earlier in a recA441lexASi umuC strain. In this strain, RecA441 protein is fullyinduced at the time ofUV exposure, whereas in the recA441umuC strain at 30'C, there is an inevitable delay before thefully induced level of RecA441 is attained.When plates were incubated at 430C after UV exposure,

the overall picture for strains containing the recA+ andrecA441 alleles was rather similar except that everythingoccurred more quickly (data not shown). Interestingly, themutant yields were similar to those obtained at 30'C. This issurprising, since the ability of RecA441 protein to cause re-pressor cleavage is very much greater at 43TC than at 30'C. Itsuggests that the ability of the RecA protein to mediate mis-incorporation is not simply a reflection of its ability to causerepressor cleavage. The biggest difference in mutant yieldwas between recA430 and recA441 in the lexASi uvrA umuCbackground. Although levels of RecA protein are high be-cause of the lexA51 allele, much less misincorporation oc-curred in the recA430 strain (Fig. 3). The RecA430 protein isknown to be defective in proteolytic activity (12), and its rel-ative ineffectiveness in promoting misincorporation con-firms the critical importance of the structure of the RecAprotein.

Since the demonstration of misincorporation in the lexA(Ind-) strain had shown that induction of genes controlledby the LexA repressor is not an absolute requirement formisincorporation, we sought to approach the question ofwhether RecA protein caused misincorporation via inactiva-tion of some other repressor; we did this by having chloram-phenicol present between UV exposure and exposure to visi-ble light. Under these conditions, RecA protein assumes itsactivated form and could, in principle, cleave such a hypo-thetical repressor without, however, resulting in significantsynthesis of the product of the derepressed gene. WithCM1143 (uvrA recA441 lexASi umuC) grown at 30°C, ex-posed to UV, and then incubated for 1 hr at 43°C with chlor-amphenicol, there was a significant induction of mutationsafter photoreversal, indicating that no other gene productcontrolled by RecA needs to be synthesized to get misincor-poration (Table 2). A similar result was obtained withCM1142, a recA441 strain that lacks lexASI.

DISCUSSIONIn these experiments, all the strains carried uvrA mutationsso that excision of UV photoproducts was eliminated, andwe assume that all UV-induced mutations arose as a conse-quence of photoproducts encountering a replication fork (13,14). Our experiments are consistent with the idea that umuD,like umuC, specifies a gene product that is concerned withcontinuation of replication beyond certain photoproducts.When the photoproduct is removed by delayed photorever-sal, replication in umuD bacteria can continue and the occur-rence of mutations among the survivors shows that wrongbases have already been incorporated even though (in theabsence of photoreversal) they are in DNA that cannot bereplicated further.We therefore suggest that error-prone repair is at least a

two-stage process. The first stage involves the incorporationof one or more wrong bases opposite a DNA photoproduct,

10

lH

Genetics: Bridges and Woodgate

4196 Genetics: Bridges and Woodgate

Table 2. Induction of Trp+ mutants following photoreversal of bacteria incubated for 1 hrof chloramphenicol at 430C

in the presence or absence

Trp+ revertants per 108viable cells in final broth

Chloram- Photo- Trp+ colonies per plate cultures (mean ± SEM)

phenicol reversal No UV UV, 1.0 J/m2 No UV UV, 1.0 J/m2

CM1143 (recA441 lexASiumuCJ22::TnS) - - 4 3 6 10 14 6 1.9 ± 1.3 3.0 ± 1.9

710 3 4 6 12 1 1 7 10 2

- + 6 7 3 10081 100 1.7 ± 0.9 35.0 ± 8.92 1 4 7494 743 2 5 6959 41

+ - 9 5 10 2 2 1 3.0 ± 1.5 2.0 ± 1.21 2 5 6 4 3910 8 9 7 7

+ + 9 2 3 2839 30 2.0 ± 1.1 15.0 ± 2.75 2 5 4030 421 5 7 2638 27

CM1142 (recA441umuCl22::TnS) - - 1 5 2 5 9 5 1.1 ± 0.7 2.8 ± 1.0

5 3 1 2 6 82 1 1 6 6 8

- + 8 8 4 7969 65 2.8 ± 1.3 31.4 ± 6.13 2 9 7469 726 11 4 50 73 71

+ - 5 5 4 913 5 3.9 ± 0.7 2.3 ± 1.62 4 6 5 3 52 2 5 2 2 2

+ + 7 2 4 2830 22 2.5 ± 1.0 13.0 ± 2.98 8 4 2924 393 8 5 23 28 28

Aliquots (0.2 ml) of logarithmic-phase cells grown at 30°C, UV-irradiated or not, were spread on prewarmed plates withor without chloramphenicol (20 Ag/ml) and incubated at 430C for 1 hr. The bacteria were photoreversed by exposure tovisible light for 40 min; bacteria from each set of plates then were harvested, centrifuged, resuspended in buffer, and usedas sources of inocula for nutrient-broth cultures. One milliliter of suspension was inoculated into each of three tubescontaining 9 ml of nutrient broth and incubated at 30'C overnight; 0.1 ml of nutrient broth from each tube then was spreadon each of three unsupplemented glucose/minimal medium plates, and the plates were incubated at 30'C for three days.Viable bacteria were spread on minimal agar plus tryptophan (0.75 gg/ml). The average titer was 2.2 x 109 per ml. The datashown are representative of three similar experiments.

the second involves further chain elongation beyond thephotoproduct. Given that the second step requires the prod-ucts of the umuC and umuD genes, what may be said aboutthe first, misincorporation step? Although our results do nottotally exclude the possibility that UmuD and UmuC pro-teins may also have a misincorporation role in wild-typestrains, the data on the effect of different recA alleles on themutation frequency after delayed photoreversal in umuCbacteria indicate that misincorporation can be mediated byRecA protein itself. That this is a direct function of RecAprotein is the only reasonable interpretation of the experi-ment (see Table 2) in which an elevated level of misincorpo-ration was seen after UV exposure in a recA441 umuC strainincubated at restrictive temperature in the presence of chlor-amphenicol. Under these conditions, the constitutive RecAprotein assumes a different (self-activated) configuration,but no induction of other genes controlled by RecA can oc-

cur. Therefore, no induction of a gene repressed by any re-

pressor that can be cleaved by RecA protein seems to berequired.A similar conclusion for UV mutagenesis in umuC+ bacte-

ria has been reached by Witkin (15) on the basis of a concep-tually similar experiment. In Witkin's experiments, howev-er, the yield of mutants in the presence of chloramphenicolwas the same as that in its absence. In the experiments re-

ported here, the yield of mutants after delayed photoreversalwhen chloramphenicol had been present was always lessthan half that obtained in its absence. The difference might

reflect the fact that in her experiments, misincorporation israpidly followed by dimer bypass. In our experiments, theabsence of the UmuC protein means that bypass cannot oc-cur; sites at which misincorporation has occurred still blockfurther replication until the time of photoreversal. In this un-natural state, there may well be a requirement for some pro-tein synthesis to occur to stabilize the photoproduct-primerconfiguration.The results with strains containing lexASI, recA+,

recA441, and recA430 in various combinations show clearlythat the extent of misincorporation is related to the natureand not the amount of the RecA protein present within thecell. Although the number of mutants seen after delayedphotoreversal increases in the order recA430 > recA+ >recA441, in parallel with the proteolytic activity of the re-spective proteins, the relation is not straightforward. Theyield in recA441 bacteria, for example, was the same at 30°Cas at 43°C, whereas proteolytic activity is generally sup-posed (on the basis of SOS induction) to be near the recA+level at 30°C. Thus, although the ability to mediate misincor-poration seems to be determined by the same structural fea-tures of the RecA protein as those that determine proteolyticactivity, the relation is not simple and is not necessarilycausal in nature.There are indications of an interaction of RecA with DNA

polymerase III in vivo. A DNA polymerase III temperature-sensitive mutant, for example, showed no loss of photore-versibility of mutations at restrictive temperature, arguing

Proc. NatL Acad Sci. USA 82 (1985)

Proc. NatL Acad. Sci. USA 82 (1985) 4197

for the involvement of polymerase III rather than polymer-ases I and II in UV mutagenesis (16). It could, of course, beargued that the polymerase III complex, although inactiveunder these conditions, nevertheless impedes access by theother polymerases. Other evidence for a specific interactionofDNA polymerase III with the RecA protein is the findingthat the polC74 mutant polymerase no longer requires RecAprotein for SOS untargeted mutagenesis of phage X (Brot-corne-Lannoye, G. Maenhaut-Michel, and M. Radman, per-sonal communication). It should not be overlooked, howev-er, that in vitro, two-step bypass synthesis of DNA lesionscan be artificially elicited with DNA polymerase I by se-quentially treating with a high concentration of manganeseions and high levels of deoxyribonucleoside triphosphates(17). Thus, whether activated RecA protein exerts its effectin misincorporation opposite a damaged template by inter-acting (perhaps proteolytically) with a component of thepolymerase complex or by modifying the DNA substrate re-mains to be established.

If RecA (without UmuC and UmuD) is able to relax thestringency of a DNA polymerase complex opposite a photo-product, one might expect it to lower the fidelity of replica-tion on an undamaged template in vivo, as has been observedfor in vitro replication of 4X174 DNA (5). The work of De-fais (18) on undamaged single-stranded DNA phage 4X174infecting bacteria in which LexA-controlled functions werederepressed after exposure to UV is consistent with this ex-pectation: a mutator effect on reversion of an amber muta-tion was observed that was independent of umuC. In thedouble-stranded DNA phage X, Maenhaut-Michel and Cail-let-Fauquet (19) have likewise shown that the SOS mutatoreffect, which is responsible for the production of mixedclones containing both wild-type and mutant phages and actson both irradiated and unirradiated DNA, does not requirethe UmuC gene product.

It has been assumed in this model that the misincorpora-tion event requiring RecA protein occurs opposite the photo-product in the template strand. However, the misincorpora-tion might conceivably occur at other sites (untargetedevents) and the pyrimidine dimer might block the furtherreplication of the strand, unless photoreversal occurs. If mis-incorporations and replication blocks were independent,however, then the mutation frequency should not be affectedby the presence or absence of a replication block. That this isnot so indicates that the two events are not independent andthat misincorporations are associated with the replication-blocking photoproduct. Thus, although misincorporationscould occur associated with but not opposite the photoprod-uct, it is simpler to assume that they are opposite.

One problem that must be mentioned here is the apparent-ly contradictory observation that in bacteria where LexA-repressible functions are constitutively expressed (recA441at 43°C and lexA51) the spontaneous mutator effect is umuC-dependent (20). If the additional mutations were simply dueto infidelity, we would not expect a requirement for UmuC.An explanation may perhaps be found in the suggestion ofCaillet-Fauquet et al. (21) that untargeted mutation in SOS-induced E. coli may be largely removed by the mismatch-correction system. One could then argue that the mutatoreffect in recA441 and lexA51 bacteria is actually targeted andreflects a mutagenic repair of spontaneously arising (cryptic)lesions that block DNA replication.

We thank Evelyn Witkin for communicating unpublished informa-tion and Manuel Blanco, Takesi Kato, Steve Sedgwick, and GrahamWalker for bacterial strains. This is paper no. 11 in a series; paperno. 10 is ref. 6.

1. Walker, G. C. (1984) Microbiol. Rev. 48, 60-93.2. Bridges, B. A. (1978) Nature (London) 275, 591-592.3. Villani, G., Boiteux, S. & Radman, M. (1978) Proc. Nati.

Acad. Sci. USA 75, 3037-3041.4. Blanco, M., Herrera, G., Collado, P., Rabollo, J. E. & Botella,

L. M. (1982) Biochimie 64, 633-636.5. Fersht, A. R. & Knill-Jones, J. W. (1983) J. Mol. Biol. 165,

669-682.6. Bridges, B. A. & Woodgate, R. (1984) Mol. Gen. Genet. 196,

364-366.7. Kato, T. & Shinoura, Y. (1977) Mol. Gen. Genet. 156, 121-131.8. Bridges, B. A., Mottershead, R. P., Rothwell, M. A. & Green,

M. H. L. (1972) Chem. Biol. Interact. 5, 77-84.9. Mount, D. W. (1977) Proc. Nati. Acad. Sci. USA 74, 300-304.

10. Davis, B. D. & Mingioli, E. S. (1950) J. Bacteriol. 60, 17-28.11. Sedgwick, S. G. & Bridges, B. A. (1972) Mol. Gen. Genet.

119, 93-102.12. McEntee, K. (1978) DNA Repair Mechanisms (Academic,

New York), pp. 349-360.13. Witkin, E. M. (1967) Brookhaven Symp. Biol. 23, 17-55.14. Bridges, B. A., Dennis, R. E. & Munson, R. J. (1967) Genet-

ics 57, 892-906.15. Witkin, E. M. (1984) Proc. Nati. Acad. Sci. USA 81, 7539-

7543.16. Bridges, B. A., Mottershead, R. P. & Sedgwick, S. G. (1976)

Mol. Gen. Genet. 144, 53-58.17. Rabkin, S. D., Moore, P. D. & Strauss, B. S. (1983) Proc.

Nati. Acad. Sci. USA 80, 1541-1545.18. Defais, M. (1983) Mol. Gen. Genet. 192, 509-511.19. Maenhaut-Michel, G. & Caillet-Fauquet, P. (1984) J. Mol.

Biol. 177, 181-187.20. Ciesla, Z. (1982) Mol. Gen. Genet. 186, 298-300.21. Caillet-Fauquet, P., Maenhaut-Michel, G. & Radman, M.

(1984) EMBO J. 3, 707-712.

Genetics: Bridges and Woodgate