BAcERIOLOGICAL REVIEWS, Dec. 1976, p. 869-907 Vol. 40, No. 4 Copyright © 1976 American Society for Microbiology Printed in U.S.A. and Inducible DNA Repair in Escherichia coli EVELYN M. WITKIN Department ofBiological Sciences, Douglass College, Rutgers University, New Brunswick, New Jersey 08903 INTRODUCTION ...... 869 Enzymatic Repair of UV Damage in E. coli ...... 869 Error-Proof and Error-Prone Pathways of DNA Repair ...... 871 The "SOS" Hypothesis: the Regulatory Role of DNA Damage ...... 874 EVIDENCE FOR THE INDUCIBILITY OF ERROR-PRONE REPAIR ("SOS REPAIR") ACTIVITY ...... 875 SOS Repair of Bacteriophage DNA ...... 875 SOS Repair of Bacterial DNA ...... 876 Evidence from studies of repair-deficient mutants ...... 876 Evidence from studies of ...... 879 MANIFESTATIONS OF SOS REPAIR AND THEIR SIGNIFICANCE ...... 879 Frequency Response to Increasing Fluence of UV Radiation ...... 879 The UV Lesion Responsible for Induction of SOS Functions ...... 880 Kinetics of Induction and Decay of SOS Repair Activity ...... 882 Time of Action of SOS Repair in UV Mutagenesis ...... 882 Cryptic Premutational Lesions Susceptible to SOS Repair ...... 884 Mutator Effect of SOS Repair on Undamaged DNA ...... 885 PLASMIDS AND SOS REPAIR ...... 886 MECHANISM OF SOS REPAIR ...... 886 Mechanism of SOS Repair in Bacteriophage ...... 886 Mechanism of SOS Repair in Bacteria ...... 887 IMPLICATIONS OF SOS REPAIR FOR CARCINOGENESIS ...... 888 REGULATION OF SOS REPAIR AND OTHER UV-INDUCIBLE FUNCTIONS . 889 PROTEASES AND THE EXPRESSION OF SOS FUNCTIONS ...... 894 CONCLUSIONS ...... 895 APPENDIX...... 896 LITERATURE CITED ...... 898 INTRODUCTION of UV radiation are described in detail else- change the genetic script by pro- where (88, 91, 93, 102). The brief summary moting errors in replication or in repair of de- given here is intended only to provide sufficient oxyribonucleic acid (DNA). Ultraviolet (UV) background for discussion of UV mutagenesis. light owes its powerful mutagenic effect in UV radiation produces a variety of photo- Escherichia coli primarily to misrepair of ra- products in DNA (169, 197, 202), among which damage the intrastrand -type dimer of adja- diation to the bacterial chromosome cent (8) has been identified as a (26, 125, 230). Recent developments indicate major cause of lethal, mutagenic and tumoro- that the error-prone repair activity responsible genic effects in a wide spectrum of organisms for UV mutagenesis in E. coli and some of its (92, 95, 216, 225). Wild-type strains of E. coli phages is repressed in undamaged wild-type effectively neutralize the potentially lethal ef- cells, but is expressed, as one of a metabolically fects of as many as a thousand di- diverse but coordinately regulated group of in- mers by utilizing one or more of three types of ducible functions, in response to UV radiation enzymatic DNA repair, which are represented or other agents that damage DNA or interrupt in Fig. 1. Photoreactivation (120) is accom- its replication (53, 77, 174, 240). This article plished by "photoreactivating enzyme" (PRE), summarizes current evidence for the inducibil- which binds pyrimidine dimers in the dark but ity ofmutagenic DNA repair and explores some requires "photoreactivating light" (PRL) in the of its implications. Other aspects of UV muta- range of wavelengths mainly between 310 and genesis in microorganisms have recently been 400 nm to "split" or monomerize them in situ reviewed (3, 62, 66, 67, 130). (183, 184). Since PRE acts specifically on pyrim- idine dimers (193), photoreversibility by this Enzymatic Repair of UV Damage in E. coli enzyme indicates that a UV-initiated biological Enzymatic repair mechanisms operating in phenomenon requires the more or less persist- bacteria which minimize the damaging effects ent presence of at least one pyrimidine dimer. 869 870 WITKIN BACTERIOL. REV. A B C PHOTOREACTIVATION EXCISION REPAIR POSTREPLICATION REPAIR ("short patch" pathway) (recombinational) UV pyrimidine UV UV A climer A'A n A ~~~~~~~~~~3'.

photoreativoting incision by replication enzyme bindsI dimer correndonucleose II 5- -3: 3 5 5' : 3' 3 V excision and repair (discontinuous daughter strands) photoreactivating light replication by DNA polymerose I recombinational exchanges I by recombination enzymes _D _ dimer _ split 3' 5 5 3: 3'- V 5 enzyme sealing by polynu- repair replication r released cleotide ligase and ligase sealing rv I 5, A ntact repaired DNA DNAlIA (continuous daughter strands) replication replication replication (dimers split) (dimers removed) (dimers still present) FIG. 1. Major pathways ofenzymatic repair ofdamage to DNA caused by UV radiation in E. coli. All three of these repair mechanisms are error-proof in principle. Symbols: -A... or -v, pyrimidine dimer or other noninstructive UV photoproduct; heavy lines, daughter strands produced in first postirradiation DNA replication; light lines, parental (UV-irradiated) strands of DNA; wavy lines, DNA polymerized during repair replication. Adapted from Hanawalt (91). See text for additional references. In the dark, pyrimidine dimers are not chem- of the sugar-phosphate linkage by polynucleo- ically reversed, but may be removed from the tide ligase. Alternate pathways of excision re- DNA by a multienzymatic mechanism, excision pair, utilizing DNA polymerase III (244) or repair (20, 195). The distortion of the double DNA polymerase II (144), have been demon- helix caused by a pyrimidine dimer (or by some strated in mutants lacking activity of DNA other kinds of DNA lesions) is recognized by polymerase I. In addition to the "short-patch" correndonuclease II, the product of the uvrA type of excision repair shown in Fig. 1B, a and uvrB genes ofE. coli (22), which makes a "long-patch" pathway has been identified single-strand nick, or incision, on the 5' side of which requires the products of the recA+ and the dimer. An exonucleolytic excision releases lexA + gene products, among others, and occurs the oligonucleotide bearing the pyrimidine di- only in growth-supporting media (49, 245, 248). mer plus some bases on either side ofit, and the Pyrimidine dimers, that are neither photoen- "excision gap" is patched by repair replication zymatically split nor removed from the DNA by (170), the intact region of the strand opposite excision repair (as in excision-deficient Uvr- the gap serving as template for accurate re- mutants kept in the dark) block the continuous placement of the missing . In wild- progress of the DNA replication complex, but type E. coli, the excision and resynthesis steps do not prevent subsequent reinitiation of DNA are probably effected simultaneously by DNA synthesis at a point beyond the dimer (185). polymerase I (50, 78, 119), although efficient When pyrimidine dimers or other noninstruc- excision requires additional gene products, in- tive lesions are present in the template strand, cluding those ofuvrC (103), uvrE (200, 201, 213) daughter strands are detected initially as seg- and mfd (73, 74) genes. The final step is sealing ments ofrelatively low molecular weight, their VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLI 871 continuity interrupted by gaps about 1,000 nu- some pathways of postreplication repair will be cleotides long (114) which correspond approxi- considered below, as they may relate to UV mately in number and position to the dimers in mutagenesis. the parental strand (7, 105). These "daughter- strand gaps" are secondary UV lesions, caused Error-Proof and Error-Prone Pathways of by the replication of DNA containing primary DNA Repair UV photoproducts that prevent base pairing. The possibility that UV-induced The third major type of enzymatic DNA repair, originate as errors in the repair of radiation postreplication repair (Fig. 1C), operates to damage (223) has been substantiated by the connect the daughter-strand segments, thereby finding that UV mutability is eliminated by endowing them with the continuity required for mutations in either the recA (124, 151, 229) or further replication (185). The formation and lexA (159, 227) genes of E. coli, which also joining of daughter-strand segments synthe- cause extreme UV sensitivity and abolish some sized on UV-irradiated templates should not be pathways of DNA repair. Although they are confused with the discontinuous synthesis and UV-nonmutable, recA- and lexA- strains are much more rapid joining of Okazaki fragments spontaneously mutable and are also mutable by at normal DNA replication forks (128). Al- agents believed to cause only replication errors though first demonstrated in Uvr- strains, (125, 241). Since recA- and lexA- mutants per- postreplication repair occurs also in Uvr+ form some kinds ofDNA repair efficiently (e.g., strains capable of excising pyrimidine dimers the "short-patch" pathway of excision repair, efficiently (192, 203), implying that ability to photoreactivation) after UV irradiation, yet perform excision repair does not prevent some produce no UV-induced mutations, the enzy- noninstructive UV lesions from passing matic repair occurring in these strains must be through a replication fork. accurate, whereas one or more of the pathways In excision-deficient (Uvr-) strains that are of DNA repair requiring both the recA + and otherwise normal, the major mechanism of lexA + gene products are probably error-prone postreplication repair is recombinational. After and responsible for UV mutability (26, 230). UV irradiation followed by DNA replication Enzymatic photoreversal of pyrimidine di- and postreplication repair, specifically labeled mers is accurate in principle and unlikely to parental DNA is found covalently inserted into cause changes in the base sequence of photore- the daughter strands, the number of insertions paired regions of the DNA. In fact, exposure to corresponding approximately to the number of PRL after UV irradiation prevents most of the daughter-strand gaps initially produced (186). mutagenic effect, as well as most of the lethal The precise mechanism of recombinational ex- effect, caused by the same UV treatment in change operating to rejoin daughter strands is "dark" controls (121, 167). There is no reason to not known, but it must be such that some py- believe that monomerization of pyrimidine di- rimidine dimers originally present in the pa- mers by PRE is mutagenic, per se, although the rental strands are exchanged into daughter visible light used to activate PRE may exert strands during postreplication repair (71). Fi- mutagenic effects of its own (218, Fig. 5). gure 1C shows a plausible but hypothetical The "short-patch" pathway ofexcision repair, scheme for recombinational postreplication re- at least as it occurs in recA - and lexA - strains, pair. must be accurate, since it does not contribute to E. coli is capable of performing postreplica- a detectable level ofUV-induced mutation. The tion repair via a number of distinct pathways low frequency of mutations which has been as- (72, 182, 189, 190, 192, 209, 247), all of which, cribed to errors in the repair of excision gaps however, require the recA + genotype. In recA - (30, 164, 165) may be associated with "long- mutants, no postreplication joining ofdaughter patch" excision repair, which, like UV muta- strands occurs (203), and the UV sensitivity of genesis, requires the recA + lexA + genotype. uvrA - recA double mutants, which can per- When Uvr+ and Uvr- strains are exposed to the form neither excision repair nor postreplication same low fluence of UV radiation, causing nei- repair, is such that a single pyrimidine dimer ther detectable killing nor detectable - per strand of DNA kills (103). The requirement esis in the excision-proficient Uvr+ strain, the for the recA + gene product, which is necessary Uvr- strain, unable to excise UV photoprod- for any type of genetic recombination in E. coli ucts, not only shows drastically reduced sur- (42), does not necessarily imply that all path- vival but also produces a high frequency ofUV- ways of postreplication repair are recombina- induced mutations among the survivors (32, 98, tional. The phenotype of recA- mutants is 225). The UV hypermutability of Uvr- strains highly pleiotropic, and these mutants lack demonstrates that unexcised UV photoproducts many wild-type functions in addition to recom- are much more mutagenic than the same photo- bination ability (Table 1). The specific nature of products removed from the DNA by excision, 872 WITKIN BACTERIOL. REV.

Utj cq t-- 1 '--4 ce '--4 o r-q r.4 eq - cn co CO m cq~ c~~~~~~ o - o 00 CD Q t- t I" LO o4 '4^ r~ - ^ CDO Fs" 00 cq4 '-4 CY "o rI ,- =N4O c m C Le ozscvq 1:- 00 0)00 &o~Q , o- - .!2 I" r- LO T-4 Lo 00 C I" I" I ^00 Co C8 2

O . Q o4

+ -4 CZ Cd W -W t 0- t 2 E. L. 04

O a 4. o o C o 0 co - C.) M 0 com At En 4

+ + Co + + + c*-. + +l1+I+ Co + '0 + + + + + + + + +

C.) 0 ++ L. ++ +++ + +~~~~~~( ++++ .0 ++ + + +

soa) Co C C.) Cd aBioa), ,0, .k- *S °) > CO >

' 0 0OC mO C d 0 z EH 14-0 > ,; C CO CO~~~~~~~'i - .e CO, -4- .^ > > VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLI 873 and implies that postreplication repair (at least discourage this hypothesis, at least until the as it occurs in Uvr- strains) is far more error- requirement for an inducible function was rec- prone than excision repair. In Uvr- strains, in ognized. which postreplication repair is the only type of A second hypothesis (27, 125, 228) proposed dark repair of UV damage known to occur, it is that UV mutagenesis may be due to errors in a probable that UV mutagenesis is due primarily lexA +-dependent error-prone pathway of re- to errors introduced into the DNA in the course combinational repair. Three experimental ob- of joining discontinuous daughter strands (26, servations made this interpretation seem rea- 230). sonable. One was the dependence of both re- The existence of two distinct pathways of combination ability and UV mutagenesis on postreplication repair, one error-proof and the recA+ gene product (43, 124, 151, 229) and jeA + independent, the other error-prone and the apparently commensurate reduction ofboth 1exA+ dependent, was postulated (227) to ac- by recB- and recC- mutations (232). Another count for the properties ofcertain derivatives of was the reduction of the yield of recombinants UV-sensitive, UV-nonmutable lexA- strains. to as little as 30% of normal when lexA- mu- These derivatives had been selected for in- tants were used as recipients in conjugation creased UV resistance, and although some (230), an observation later confirmed but as- were nearly as UV resistant as jDA + strains, cribed to effects other than reduced ability to they were still nonmutable by UV radiation. perform genetic recombination (159). Most im- The UV-resistant, UV-nonmutable strains portant was the approximate equality observed were assumed to have acquired improved abil- between the number of daughter-strand gaps ity to perform an error-prooftype ofpostreplica- repaired and the number of recombinational tion repair, enabling them to tolerate nearly as exchanges detected after postreplication repair many unexcised pyrimidine dimers as lexA+ (186). This relation implied that either all or strains without undergoing UV mutagenesis. nearly all postreplication repair is associated The ability of similar UV-resistant, UV-non- with recombination, although it did not rule mutable strains to perform postreplication re- out the possibility that a small fraction of the pair much more efficiently than their lexA- daughter-strand gaps is repaired by a nonre- parents has been confirmed (215). Biochemical combinational mechanism. evidence for a pathway ofpostreplication repair Evidence against recombinational postrepli- requiring the recA+ and lexA + gene products cation repair as the error-prone gap-filling and associated with UV mutagenesis has re- mechanism, however, began to accumulate in cently been obtained (189). 1971, when Bridges and Mottershead (30) Two hypotheses concerning the nature of er- showed that the UV mutability of chemostat- ror-proofand error-prone pathways ofpostrepli- grown E. coli is not enhanced by chromosomal cation repair have been considered. Witkin multiplicity. In recB and recC mutants, which (227) initially proposed that recombinational show reduced ability to perform genetic recom- repair is accurate, and that the lexA+-depend- bination (42, 68) and postreplication repair ent mechanism of daughter-strand gap-filling (247), yields of UV-induced mutations per sur- is an error-prone DNA polymerase activity ca- vivor at a given UV fluence were either the pable of inserting nucleotides opposite pyrimi- same as those in wild type (99; T. Kato, R. dine dimers or other noncoding UV photoprod- Rothman, and A. J. Clark, personal communi- ucts without template instruction. A similar cation) or were somewhat lower (232), depend- mechanism was suggested by Bridges et al. ing upon the strain used and on the mutation (27). The lexA + gene was envisaged (227) either scored. Even when reduced yields of UV-in- as conferring template independence upon a duced mutations were obtained, neither Witkin normal DNA polymerase or as itself coding an (232) nor Hill and Nestmann (99) considered error-prone DNA polymerase such as terminal the reduction to be conclusive evidence that the deoxynucleotidyl transferase (TDT), an enzyme probability of error per mutagenic lesion re- capable ofrandom "end-addition" ofnucleotides paired was affected by absence of exonuclease to single-strand ends of DNA (15). In either V, now known to be the product ofthe recB and case, gaps opposite pyrimidine dimers could be recC genes (211). The demonstration that E. filled by repair replication with a high probabil- coli can utilize a second pathway of genetic ity of error. Failure to find TDT activity in E. recombination requiring the recF+ but not the coli, and the later demonstration that DNA recB+ recC+ genotype (101) left open the possi- polymerases I, II, and III (the only known con- bility that UV mutagenesis might be associated stitutive DNA polymerases) are strictly de- with the recF+-dependent pathway. Recent ex- pendent upon template instruction (7, 128; M. periments (T. Kato, R. Rothman, and A. J. Radman, personal communication), tended to Clark, personal communication) have shown 874 WITKIN BACTERIOL. REV. that UV mutagenesis occurs in recA + recB - or other agents having in common the ability to recF- strains, which are unable to perform damage DNA or interrupt its synthesis. either of the two pathways of genetic recombi- The SOS hypothesis incorporates and inte- nation. This important result suggests that the grates many observations and ideas that go recA + gene product is required for UV muta- back as far as a quarter of a century to the genesis for some reason other than its role in demonstration by Lwoff et al. (134) that UV promoting genetic recombination and that nei- radiation initiates mass induction of prophage ther ofthese pathways of recombinational post- in lysogenic bacteria. Other treatments, shar- replication repair is very error-prone. However, ing with UV radiation the ability to arrest these experiments do not rule out the possibil- DNA replication, were also found to cause lyso- ity that residual recombinational repair, either genic induction, including exposure to X rays due to leakiness of the rec mutations or inde- (129), starvation for (147, 199a), incu- pendent of both recB and recF gene products, bation with mitomycin C (167c), and tempera- may be at least partly responsible for bacterial ture elevation in certain mutants unable to UV mutagenesis. Recent work has demon- synthesize DNA at high temperatures (154, strated that error-prone DNA repair activity, 166, 188). whatever its mechanism, is not a constitutive Another early observation contributing to property of undamaged wild-type E. coli, but is the SOS hypothesis was Weigle's finding (220) induced in response to UV radiation and other that exposure of the host population of E. coli mutagenic treatments. A full description ofthe cells to a low fluence of UV radiation prior to experimental evidence which has led to this infection substantially increases the survival of conclusion will follow an account of the "SOS" UV-irradiated X phage, and is required for a hypothesis, an hypothesis that has stimulated high level ofphage UV mutagenesis. This "UV- much of the work to be discussed in this article reactivation" and the mutagenesis that accom- and that continues to provide a unifying theo- panies it require the recA + (151) and lexA+ (53) retical framework for an impressively diverse genotype in the host. Defais et al. suggested yet interconnected array of biological and bio- that UV irradiation of the bacteria induces the chemical investigations. functions responsible for UV reactivation and UV mutagenesis ofthe irradiated phage, via an induction pathway at least partially common to The "SOS" Hypothesis: the Regulatory Role other recA + lexA +-dependent phenomena, such of DNA Damage as prophage induction and bacterial UV muta- Defais et al. (53) proposed that UV mutagen- genesis. Radman (174, 175) proposed that a sin- esis in E. coli might depend upon an inducible gle inducible error-prone repair replication sys- function which, like X prophage, is repressed in tem (SOS repair) may be responsible for both healthy wild-type cells but is expressed in re- the mutagenic reactivation of UV-irradiated sponse to UV irradiation. One basis for this phage and for error-prone repair of bacterial suggestion was the dependence ofUV mutagen- DNA. I shall use the term "SOS repair" to esis upon recA + and lexA + gene products, which denote inducible error-prone repair activity act- are also required for prophage induction by UV ing either on phage DNA or on bacterial DNA radiation, and for the pleiotropic expression of without implying that only one such activity is other UV-inducible functions now recognized to induced in E. coli, although there is at present be numerous (Table 1). The hypothesis that the no reason to invoke more than one. Since UV recA + and lexA + gene products jointly control a radiation is only one of many host treatments coordinately regulated group of inducible func- capable of stimulating mutagenic reactivation tions, including an error-prone DNA repair ac- of UV-irradiated phage, the term "UV-reacti- tivity, was further developed by Radman as the vation" is misleading. Following Radman's sug- "SOS" hypothesis (174, 175). The designation gestion (174), the terms "W-reactivation" and "SOS" (the international distress signal) im- "W-mutagenesis" are used in Table 1 and else- plies that damage to DNA (or stalled DNA where to refer to the Weigle effect. replication) initiates a regulatory signal that Another facet of the SOS hypothesis was an- causes the simultaneous derepression of var- ticipated in an interpretation of the parallels ious functions, all of which presumably pro- observed between the conditions that promote mote the survival of the cell or of its phages. I or prevent prophage induction in E. coli lyso- shall refer to the group of inducible functions gens and those that promote or prevent fila- believed to belong to this regulatory unit as mentous growth in lon- strains such as E. coli "SOS functions." "SOS repair" is defined here B (226). Witkin proposed that certain bacterial as recA + lexA+-dependent error-prone DNA re- functions, including the synthesis of a septum- pair activity induced in E. coli by UV radiation inhibiting protein, may be governed by repres- VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLI 875 sors similar enough to A repressor to respond to sis (220). A comparable degree of reactivation the same inducer, an inducer produced only and mutagenesis of UV-irradiated X phage oc- when DNA replication is interrupted. The sub- curs in unirradiated tit hosts incubated before sequent report (122) thatE. coli K-12 strain T44 infection at 420C for about 40 min (39), but not not only induces prophage spontaneously at ele- in recA or lexA- derivatives of the tit strain vated temperatures but also grows in long fila- (40). When wild-type strains are UV-irradiated ments at 420C, without requiring any of the and incubated in growth-supporting medium usual inducing treatments, further indicated for 30 min before infection, W-reactivation and coordinate regulation ofthese two functions. W-mutagenesis of UV-irradiated X are maxi- Table 1 lists some of the extremely pleio- mal, but neither occurs if chloramphenicol is tropic effects of recA and lezA- mutations, present during the preinfection incubation (52). only one of which (loss of ability to perform Like other recA+ lexA+-dependent functions, genetic recombination) is caused by recA but W-reactivation of X requires protein synthesis not by lexA mutations, and for this and other after the inducing treatment for its expression. reasons is not considered to be an inducible Further evidence for common regulation of SOS function. Table 1 also shows that the tif-1 W-reactivation and prophage induction is that mutation, which is responsible for the thermal both can be induced "indirectly." "Indirect" in- inducibility of prophage and filamentous duction of prophage occurs when an unirra- growth in strain T44 (39), causes constitutive diated F- ("female") lysogen is mated with a expression of other (perhaps of all) SOS func- UV-irradiated F+ or F' ("male") donor strain tions at elevated temperature, without requir- (18, 19, 56). The UV-damaged DNA of the F ing any other inducing treatment. When incu- episome (an E. coli sex factor), transferred by bated at 420C, tif-1 mutants show no abnormal- conjugation into a recipient lysogen whose own ity of DNA structure or replication (39), but DNA is undamaged, triggers prophage induc- their ability to express SOS functions at this tion. Indirect induction of prophage is also pro- temperature requires the recA+ kexA+ genotype moted by transfer of other UV-irradiated repli- (40). The tif-1 mutation maps in or very near cons, such as colicin I factor (152, 153) or P1 the recA locus (39), and may affect the same bacteriophage (181). Indirect induction of W- protein that is altered by recA mutations. Two reactivation and W-mutagenesis has been dem- other mutations that are tightly linked to the onstrated by George et al. (77), who mated an recA locus and may map within it are lexB- (11) unirradiated F- host with UV-irradiated do- and zab- (40). Both lexWB- and zab- mutations nors (F' or Hfr) before infection with UV-irrad- cause a phenotype like that oflexA - mutations: ited X phage. Mutagenic reactivation of the noninducibility ofSOS functions without loss of phage occurred as efficiently as ifthe host itself recombination ability. The recA locus maps at had been irradiated, whether the DNA trans- 58 min on the recently recalibrated linkage ferred during conjugation was that of a UV- map ofE. coli K-12 (4), and the lexA locus maps irradiated sex factor or a fragment of UV-irra- at 90 min. diated Hfr chromosome, and even if the DNA Evidence that SOS functions are actually de- transferred was restricted in the recipient. In repressed in response to UV radiation and spite of some puzzling differences (e.g., pro- other inducing agents (or in tit mutants at phage is indirectly inducible only by transfer of 4200) is indirect in all cases except X prophage a complete UV-irradiated replicon and not by a induction, in which inactivation (198) and pro- fragment of Hfr DNA), the indirect inducibility teolytic cleavage (180) of X repressor have been ofboth prophage and W-reactivation points to a demonstrated after such treatments. A require- common induction pathway. Indirect induction ment for new protein synthesis after the induc- of another SOS function (filamentous growth) ing treatment has been shown for the expres- has also been observed (152; J. Donch, personal sion of most of the SOS functions, and is ob- communication). vious in the case of protein X. W-reactivation (SOS repair) of UV-irradi- ated X phage occurs in both Uvr+ and Uvr- EVIDENCE FOR THE INDUCIBILITY OF strains, and is independent ofability to perform ERROR-PRONE REPAIR ("SOS excision repair (176). However, the range of UV REPAIR") ACTIVITY fluences in which both W-reactivation and X prophage induction can be expressed is about 10 SOS Repair of Bacteriophage DNA times lower in Uvr- strains (53, 155), as shown The increased survival of UV-irradiated in Fig. 2. The photoreversibility of prophage phage that occurs when the host is also irradi- induction (115), W-reactivation (220), and other ated before infection (W-reactivation) is accom- SOS fimctions indicates that pyrimidine dimer panied by a high level of phage UV mutagene- is necessary for their induction. Dimers that 876 WITKIN BACTERIOL. REV. dence for inducibility ofa function necessary for bacterial UV mutagenesis has come from two kinds of studies: those in which mutant strains exhibiting anomalous induction of prophage 0~~~~~~~~~~~~~~~~~1 and other SOS functions were examined for WV<10 corresponding anomalies of UV mutability pre- dicted by the SOS hypothesis, and those in

°1: t&P A or poIA-1 -0 which an inducible pathway of postreplication repair has been demonstrated biochemically wvrA po/A-i \ l1. and linked to UV mutability. Evidence from studies of repair-deficient mutants. (i) polA-1 mutants. Mutants of E. coli carrying a polA-1 mutation are deficient in the DNA polymerizing activity of DNA polym- i 0.01 erase I (54, 87). The repair replication usually 10 20 30 40 performed by this enzyme can be taken over by UV FLUENCE (J/m2) DNA polymerase III (192, 209, 244, 247) and to FIG. 2. Relation between survival and induction some extent by DNA polymerase II (144). Al- of two UV-inducible functions in polA-i, uvrA, and wild-type strains of E. coli. Dashed lines, survival though polA-1 mutants excise pyrimidine di- percentage. Solid lines, composite curves for induc- mers more or less normally (21), they are slow tion of X prophage (refer to "free phage") and for W- to complete the patching and sealing ofexcision reactivation of UV-irradiated X (refer to 'W-reactiva- gaps (117, 168). The UV sensitivity of polA-1 tion factor") (53, 70, 155, 240). mutants is relatively slight (4 to 5 times wild type) compared with that of Uvr- strains una- ble to excise pyrimidine dimers (10 to 20 times are repaired rapidly by the excision repair sys- wild type). Nevertheless, as Fig. 2 shows, exci- tem, however, evidently fail to generate an sion-proficient polA-1 strains express at least effective SOS induction signal. two UV-inducible SOS functions (prophage in- Blanco and Devoret (10) and Devoret et al. duction and W-reactivation) in the same low (55) have ruled out recombinational repair as range of UV fluences as Uvr- strains. Filamen- the mechanism of W-reactivation, and have tous growth, another SOS function, is also in- concluded that it expresses the activity of a duced by UV in the same low range of UV previously unrecognized type of dark repair, fluences in excision-proficient polA-1 mutants involving neither excision of DNA damage nor (J. Donch, personal communication). Since recombinational exchanges, which is either in- polA-1 strains are slow to complete the repair duced or activated in the host by W-reactivat- replication step of excision repair, it appears ing treatments. The occurrence of W-reactiva- that a persistently open excision gap is equiva- tion in single-stranded phages (167a, 209a) im- lent to an unexcised pyrimidine dimer as a plies a nonrecombinational mechanism, bar- means of initiating an effective SOS induction ring multiple infection, since only one copy of signal. the phage genome is present even after a repli- Witkin and George (240) suggested that both cation. Experiments with the single-stranded the UV sensitivity and the UV mutability of phage qX174 (M. Radman, P. Caillet-Fauquet, polA-1 strains can be explained by the anoma- M. Defais, and G. Villani, personal communi- lously low range of UV fluences in which these cation) have demonstrated that W-reactivation strains express inducible SOS functions, as- is accomplished by an inducible DNA replica- suming that all SOS functions are expressed in tion mechanism, capable of polymerizing DNA the same fluence range. The UV survival curve past pyrimidine dimers in the template strand. of a Uvr+polA-1 strain is superimposable upon These crucial studies will be described in detail the falling portion of induction versus fluence in connection with the mechanism of SOS re- curves for prophage and W-reactivation, as pair. shown in Fig. 2. If SOS repair or any other inducible SOS function is necessary for survival SOS Repair of Bacterial DNA at higher UV fluences, the survival of UV- The proposal that bacterial UV mutagenesis irradiated polA-1 populations may be limited and that of UV-irradiated X phage undergoing by their ability to induce the vital function(s). W-reactivation may depend upon the same in- Survivors, at any UV fluence, would then con- ducible error-prone repair activity (174, 175) sist only ofthe fraction ofthepolA-1 population was initially based only upon their common in which SOS functions are expressed. Since an requirement for the recA + lexA + genotype. Evi- inducible inhibitor of exonuclease V is one of VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLI 877 the SOS functions shown in Table 1, the imme- tion is required for bacterial UV mutagenesis diate cause of death inpoLA-1 strains may well (234, 238). Induction of the postulated SOS re- be double-strand breaks in DNA, produced at pair activity by incubation at 420C should en- lower UV fluences than in wild-type strains hance the UV mutability ofa tif- mutant, when (16, 17). However, the increased probability of compared with a largely uninduced population producing a lethal break may be traced to a containing the same amount ofpotentially mu- more primary cause: the inability of poLA-1 tagenic UV damage. UV damage is obviously cells to express SOS functions in the same necessary in an assay for enhanced UV muta- range of UV fluences as wild-type strains. genesis, and therefore the tif- bacteria and tiff The UV mutability of polA-1 strains (231) controls must be UV irradiated either before or and of other strains now known to be deficient after the incubation at 42"C. Since UV radia- in the polymerizing activity of DNA polymer- tion itself is an efficient inducer of SOS func- ase I (125, 127) is the same as that of the wild tions, thermally induced SOS repair activity type, when the frequency of induced mutations should be expected to enhance UV mutagenesis per survivor is compared at the same UV flu- only at extremely low fluences, well below ence. This means that a polA-1 survivor and a those required for mass UV induction of pro- Poll survivor, exposed to the same UV fluence, phage or W-reactivation (Fig. 2). In Uvr- have the same probability ofundergoing muta- strains, UV fluences below about 0.5 J/m2 in- genesis, although the surviving fraction is duce prophage or W-reactivation in fewer than much smaller in the polA-1 population. Al- 1% of the irradiated population, yet about five though other interpretations are possible (231), to six pyrimidine dimers per genome are pro- a polA-1 survivor in which SOS functions are duced inE. coli per 0.1 JWm2. Assuming that all expressed may be no different from an SOS- SOS functions are induced at a given UV flu- induced Pole survivor exposed to the same UV ence to the same extent, exposure of tif bacte- fluence in those parameters that determine the ria to UV fluences below about 0.5 J/m2 (Uvr-) probability of undergoing UV mutagenesis. or 5 JWm2 (Uvr+) should introduce some poten- The relevant parameters are: (i) the amount of tially mutagenic UV photoproducts into the premutational UV damage per survivor, (ii) DNA of almost every cell, while inducing SOS the probability that such damage will be re- functions (including the postulated SOS repair paired by an error-prone mechanism, and (iii) activity) in an extremely small fraction of the the probability of error per lesion repaired by population. the error-prone repair system. Unless the sur- A 10-fold enhancement of the yield of UV- viving fraction inpolA-1 populations is selected induced Trp+ mutations was obtained in a tryp- by UV radiation as the least damaged fraction, tophane-requiring uvrA- tif-1 derivative of E. polA-1 and Polk survivors exposed to the same coli B/r (strain WP44) when UV irradiation UV fluence should be alike in these respects. with about 0.1 J/m2 was followed by a 45-min They would be alike, for instance, if successful incubation at 420C, compared with the yield SOS induction in a polA-1 survivor depends obtained from similarly treated tiff controls or upon metabolic factors unrelated to the amount from tif- cells maintained after UV irradiation of DNA damage a particular cell may have at 300C (234). The magnitude of the enhance- sustained. Cells that start late, ment was later increased to a maximum of for example, may have the best chance of tran- about 50-fold at the lowest UV fluence used scribing information for SOS functions from (238). Typical results are shown in Fig. 3. The fully repaired templates and surviving. amount of enhancement of UV mutagenesis Since the induction of SOS functions is more provided by thermal post-treatment in the tif effective inpolA-1 strains that in the wild type strain diminishes as the UV fluence increases, at UV fluences below about 5 J/m2 (Fig. 2), and no significant enhancement is observed at Witkin and George postulated that the DNA UV fluences above about 3 J/m2. UV induction polymerase I-deficient mutant should be more of X lysogens and of W-reactivation in Uvr- UV mutable at these very low UV fluences, if strains approaches 100% at about the same flu- UV mutagenesis depends upon inducible SOS ence (Fig. 2). Once the UV treatment itself is repair. Significantly elevated UV mutability at sufficient to induce SOS functions in every cell, sublethal UV fluences was demonstrated in thermal post-treatment should no longer en- polA-1 strains, as were other properties pre- hance the yield of UV-induced mutations. dicted by the SOS hypothesis (235, 240). These results are in good agreement with the (ii) tiftI mutants. Studies oftif-1 mutants, in expectation that postirradiation incubation at which SOS functions are thermally inducible 420C (which should induce SOS functions in without insult to DNA, have provided the every tif cell) should enhance UV mutagenesis strongest evidence that an inducible SOS func- at UV fluences too low to cause mass UV-induc- 878 WITKIN BACTERIOL. REV. other predictions generated by the SOS hypoth- esis were fulfilled as well. Thermal enhance- ment of UV mutability was abolished by the presence of chloramphenicol during the heat treatment, indicating a requirement for new protein synthesis after the inducing treatment. Additives that had been shown to promote or 0 U(/ prevent thermal induction of prophage in tif- M >~~~~~~~/1 lysogens (80, 123) had parallel effects on ther- +S1037 A/-o10 mal enhancement of UV mutability: the en- hancement was abolished by pantoyl lactone or by a mixture ofcytidine and guanosine and was .~~~~~~~~Ti- promoted even at 300C by adenine. Although the basis ofthese effects is not understood, they provide strong evidence for common regulation of prophage and an inducible SOS function re- quired for bacterial UV mutagenesis. loo, (iii) dnaB's mutants. Temperature-sensitive dnaB mutants ofE. coli, which stop synthesiz- ing DNA instantly at 42CC (219), also induce prophage at elevated temperatures when lyso- .1 0.2 0.4 0.8 1.6 3.2 6.4 1.28 genic (166, 188). The SOS hypothesis predicts UV FLUENCE WJ that all SOS functions should be thermally in- FIG. 3. Effect of postirradiation temperature ele- ducible in dnaBts strains, including SOS repair. vation on the UV mutability of strains WP2, (tit) Enhancement of UV mutagenesis at very low and WP44j-NF (tif-1 Sfi-). Symbols; A, survival per- fluences, virtually identical with that described centage (average for both strains., whether or not for tiff mutants, has been demonstrated in thermal post-treatment was given); *, frequency of dnaB-43 derivatives of E. coli B/r (237). Al- UV-induaced Trp+ mutations (average ofdata for tif+ though tif and dnaBIs strains are very similar strain, whether or not thermal post-treatment was in their expression of SOS functions at 420C, given, and for tif-1 strain not given thermal post- the reasons for their thermal inducibility are treatment); O. frequency of UV-induced Trp+ muta- quite different. In dnaBIs mutants, induction is tions (average of data for tif-1 strain given thermal triggered by interrupted DNA replication, post-treatment). (a) - - -, theoretical 'one-hit" curve (slope I); (b) ---, theoretical 'two-hit' curve whereas heat treatment oftiF mutants induces (slope =2). Thermal post-treatment consisted of in- SOS functions without damaging the DNA or cubation for 60 min at 42C immediately after UV arresting its synthesis. irradiation and plating on 5% SEM agar. All plates (iv) Other mutant strains. The hyperinduci- were incubated at 301C for2 days (survival) or 3 days bility of polA-1 strains for SOS functions at (mutations) starting either immediately after ther- very low UV fluences suggests that persistently mal post-treatment (if given) or immediately after open gaps in DNA may generate effective SOS plating, if thermal post-treatment was not given. signals, whereas gaps that are rapidly closed do Points are averages of data from four experiments. not. elevated UV at Trp+ mutation frequencies shown have been cor- Significantly mutability rected for spontaneous mutations by subtracting fre- extremely low fluences (possibly indicative of quencies obtained from similarly treated unirra- hyperinducibility for SOS repair activity) has diated controls. Titers oflog cultures used were 1 x been observed in mfd mutants, which close ex- 10e to1i3 x 0 cells per ml. Forda(uivalescription of cision gaps very slowly (73) and are also some- strains, media and methods used, see references 234 what hyperinducible at low UV fluences for and 238. prophage and W-reactivation (74). Elevated UV mutability in the extremely low UV fluence tion, by permitting error-prone repair ofgV range has been found as well in recF mutants lesions that could otherwise not cause muta- (E. Witkin, unpublished observation), which tions. In uninduced cells, such lesions presum- are deficient in postreplication repair (72), and ably either remain unrepaired or are repaired in UVrA poLA-1 mutants (235), in which post- by error-proof mechanisms. The significance of replication repair may be slow. All of these the slopes of the curves in Fig. 3 will be consid- observations support the hypothesis that an in- ered later. ducible function is required for bacterial UV Not only was UV mutability enhanced in the mutability, its induction coordinated with that tifa mutant by thermal post-treatment, but of prophage and other SOS functions. In addi- VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLI 879 tion, the hyperinducibility of strains in which genesis is a pyrimidine dimer and that at least gaps in DNA are unusually persistent impli- one must be located in the gene in which the cates such gaps as possible sources of the SOS particular mutation scored occurs. Two hy- induction signal. potheses have been argued most recently. One Evidence from studies of postreplication re- is the "two-lesion" hypothesis, according to pair. A correlation between UV mutagenesis in which a UV-induced mutation requires the in- E. coli and an inducible lexA+-dependent path- teraction of two closely spaced UV photoprod- way of postreplication repair has been demon- ucts near the site of the mutation (25, 64, 148). strated by Sedgwick (189), who showed that The other is the "one lesion + SOS induction" chloramphenicol, added just before UV irradia- hypothesis (61, 240), which proposes that only tion of strain WP2S (a uvrA- trp- derivative of one UV photoproduct is a premutational lesion B/r), prevents both the irreversible "fixation" of in the gene scored, whereas the other is an UV-induced Trp+ mutations against enzymatic SOS-inducing "hit," required for induction of photoreversal and the occurrence of a small SOS repair activity. The tit mutant, in which fraction of postreplication repair. Inhibition of SOS functions are thermally inducible, seemed protein synthesis after UV irradiation thus an ideal instrument for determining which of causes a small number ofdaughter-strand gaps the two hypotheses is correct. The "one-lesion + to remain open, and also blocks the occurrence SOS induction" hypothesis predicts that tit of UV mutagenesis. Chloramphenicol does not bacteria, UV-irradiated and then given a ther- prevent the completion ofdaughter-strand join- mal post-treatment, will not require an SOS- ing, however, in irradiated bacteria which have inducing "hit," but will show a linear increase been allowed a 20-min period of growth before of mutation frequency as the UV fluence in- addition ofthe antibiotic, or in uvrA- tit bacte- creases, indicative of the accumulation of pre- ria which have been incubated for 70 min at mutational UV photoproducts caused by the 420C before UV irradiation. No chlorampheni- absorption of single photons. The frequency of col-sensitive pathway of postreplication repair thermally enhanced yields of UV-induced Trp+ was detected in UV-nonmutable uvrA- lexA- mutations rises with an unmistakably linear bacteria. These observations establish that UV slope, whereas controls show the typical F2 re- mutagenesis is associated with the occurrence sponse, in both tif (233, 234) and dnaBts (237) of a minor lexA +-dependent pathway ofpostrep- strains, as can be seen for a tit strain in Fig. 3. lication repair that is UV inducible, or requires This result seemed to provide unambiguous a UV-inducible component. Sedgwick's study support for the idea that UV mutagenesis re- also shows that the major pathway of postrepli- quires an SOS-inducing "hit" plus a premuta- cation repair in Uvr- E. coli, which is known to tional photoproduct, the latter remaining cryp- be recombinational (186), is constitutive, lexA+- tic in uninduced cells. However, if all UV mu- independent and error-free, at least in the pres- tagenesis involves SOS repair of damage ence of chloramphenicol. The existence of a caused by single UV photoproducts, mutation 1exA + - dependent chloramphenicol - sensitive frequencies in normal strains should begin to pathway of postreplication repair has also been rise linearly with increasing UV fluence once demonstrated by Youngs and Smith (247). the level ofUV radiation required for mass SOS induction is reached. The mass-induction flu- MANIFESTATIONS OF SOS REPAIR AND ence for both X prophage induction and W-reac- THEIR SIGNIFICANCE tivation is approached at about 3 J/m2 in Uvr- strains (Fig. 2), the same level at which ther- Mutation Frequency Response to Increasing mal post-treatments no longer enhance UV mu- Fluence of UV Radiation tagenesis in tit (Fig. 3) and dnaB's (237) popu- At low fluences of UV radiation, UV-induced lations. However, as Fig. 3 shows, the yield of mutations in bacteria usually increase as the UV-induced mutations continues to rise ap- square of the UV fluence, as in the lower curve proximately as the square ofthe UV fluence, in in Fig. 3. This "fluence-squared" (F2) relation both tip and tifr strains, thermally post- implies that two independent photon absorp- treated or not, up to UV fluences as high as 9 J/ tions (UV "hits") are required to induce a muta- m2. Bridges et al. (33) have also reported muta- tion. The nature of the two "hits" has been a tion frequency response curves that continue source ofspeculation for nearly two decades (25, their P slope well above the mass-induction 33, 61, 64, 148, 224, 240, 249). fluence level. The data shown in Fig. 3 seem to Since most UV-induced mutations are photo- support the "one-lesion + SOS induction" hy- reversible, it is generally assumed that at least pothesis at low UV fluences and the "two-le- one ofthe UV photoproducts required for muta- sion" hypothesis at higher UV fluences. Both 880 WITKIN BACTERIOL. REV. Witkin (238) and Doudney (63) have recently in both Uvr+ (141) and Uvr- (E. Witkin, unpub- concluded that some UV-induced mutations in lished data) E. coli lysogens, in striking con- E. coli are caused by single UV photoproducts, trast to the linear increase of induced lysogens whereas others are caused by the interaction of with increasing exposure to X rays (140). If all two closely spaced UV lesions, and that neither SOS functions are assumed to respond to a hypothesis is wholly correct nor wholly incor- common induction signal, UV radiation clearly rect. requires two photon absorptions to generate an Witkin (238) has distinguished between sin- effective induction signal. gle premutational UV lesions that require er- Sedgwick (189, 190a) has suggested that a ror-prone SOS repair but are not, in them- target for SOS repair (and, by implication, an selves, sufficient signals for the induction of SOS-inducing lesion) may be a daughter-strand SOS repair activity ("SOS-mutable, SOS-non- gap that partially overlaps another daughter- inducing" lesions), and UV lesions generated strand gap in the sister DNA molecule. Such a by the interaction of two UV photoproducts, pair of overlapping daughter-strand gaps which are not only targets for SOS repair activ- should be refractory to any type of recombina- ity but are also effective inducers of SOS repair tional repair, in principle, and could not be ("SOS-mutable, SOS-inducing" lesions). The patched by the repair replication activity ofthe distinction between these two types ofpremuta- known constitutive DNA polymerases, because tional lesions was based on the kinetics ofther- ofthe pyrimidine dimer or other noncoding UV mal enhancement of UV mutagenesis at 42°C photoproduct located opposite each of the gaps. and the kinetics of decay at 300C of susceptibil- This type of structure (Fig. 4A) would be ity to such enhancement in tif strains. It was formed when two pyrimidine dimers on oppo- concluded that in uninduced cells SOS-muta- site strands of the parent molecule are close ble, SOS-noninducing lesions are cryptic, but enough so that the daughter-strand gaps pro- contribute to overt UV mutagenesis if SOS ac- duced by them overlap, either as they are origi- tivity is induced by some other means, such as nally formed or after some DNA degradation. temperature elevation in tif mutants or expo- In Uvr- strains, such gap structures could ac- sure to UV fluences high enough for mass UV count entirely for the F2 relation between induction of SOS repair. Unless such cryptic expression of SOS functions and UV fluence in premutational lesions are present in UV-irradi- the low fluence range. Mass SOS induction, on ated but uninduced cells, thermal post-treat- this basis, should occur at the fluence at which ments could not enhance UV mutagenesis in every irradiated cell, on the average, will con- tif or dnaBIs strains. Unless the cryptic lesions tain one pair of overlapping daughter-strand are due to single UV photoproducts, the ther- gaps after DNA replication. mally enhanced yield should not rise linearly In Uvr+ strains, unexcised UV photoproducts with increasing UV fluence. However, unless could also generate overlapping daughter- one assumes that some UV-induced mutations strand gaps that would be refractory to consti- are caused by two UV photoproducts, the F2 tutive repair systems, as in Uvr- strains. How- accumulation of induced mutations at fluences ever, Uvr+ strains could, in addition, form the far above the mass-induction level (33, 63; Fig. kind of gap structure shown in Fig. 4B, which 3) would be hard to explain. Variation in the would also require the production oftwo excisa- proportions of the two kinds of premutational ble UV photoproducts close to each other on lesions might be expected in different genes, or opposite strands of the irradiated DNA mole- even in the same gene under different condi- cule. Excision of one of the lesions, perhaps tions, and could account for the complexity and with some DNA degradation enlarging the ex- variability characteristic ofmutation frequency cision gap, could place the second lesion oppo- response curves (63). site the excision gap, thus preventing patching by DNA polymerase I. Since correndonuclease The UV Lesion Responsible for Induction of II requires a double-stranded substrate for inci- SOS Functions sion near a pyrimidine dimer, the second lesion The composite induction curves for prophage would remain unexcised at least until the first induction and W-reactivation shown in Fig. 2 excision gap is somehow filled and sealed. Such show a steep rise in the low UV fluence range structures, if produced in a part of the DNA similar to the increase in frequency of UV- replicated before UV irradiation, might be re- induced mutations with the square of the UV pairable by recombination, but should be re- fluence in the same range. When measured fractory to any known type of constitutive re- carefully, prophage induction can be seen to pair if produced in an unreplicated part of the increase strictly as the square ofthe UV fluence chromosome. Bresler (23) has suggested that VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLI 881 A B POSTREPLICATION REPAIR EXCISION REPAIR (chloramphenicol-sensitive pathway) ("long-patch' pathway) UV UV

3' ,5' 3 5 ; replication s excision,degradation

5'- /.3 5- - 3: 3' 55' 3' 5' daughter-strand gaps overlap repair replication stops s recombinational repair impossible sat 2 nd dimer,degroda- gaps enlarged by degradation tion continues

5 - 3, 3' -V , 5'..5 5 ' 3 ------V SOS repair induced, 3' 5' ; polymerizes DNA post SOS repair induced, poly- dimer, causing mutation ; merizes DNA past dimer, (x) causing mutation (x) 5' 3' 2nd gap repaired by 2nd also repaired 31 5' recombination or by SOSgap polymerase 2nd dimer r,%pairedbyexcision 5' 3' 1f short-patch 5i 33x repair 3' ' 31' 3 3' 5' one daughter molecule no overlapping gaps produces overlapping produced in next re- (both strands carry gaps in next replication plication mutant information) FIG. 4. Possible pathways oferror-prone repair (SOS repair) ofdamage to DNA caused by UW radiation in E. coli. Symbols: -- oror- pyrimidine dimer or other noninstructive UV photoproduct; heavy lines, daughter strands produced during first postirradiation DNA replication; light lines, parental (UV-irradi- ated) strands ofDNA; wavy lines, DNA polymerized during repair replication; X = mutation ('wrong" base sequence). SOS repair activity is assumed to permit replication pastpyrimidine dimers or other noninstructive LW photoproducts with a high probability of error (see section of text under heading, Mechanism of SOS Repair).

this type of structure could cause UV mutagen- closing such gaps. Unless and until such activ- esis by promoting replication errors during ity was induced, these otherwise nonrepairable recA+-dependent excision gap resynthesis, but gaps would tend to persist. Several strains in he does not believe that error-prone repair ac- which repair defects cause unusually slow clo- tivity is inducible. Polymerization past a py- sure of gaps in DNA (polA-1, recF, mfd) have rimidine dimer, with or without errors, is not been found to be hyperinducible for at least within the capacity of any of the three known some SOS functions (see above). It is a reasona- constitutive E. coli DNA polymerases (7, 128; ble possibility that an SOS signal may be initi- P. Caillet-Fauquet, M. Defais, and M. Rad- ated by either ofthe types of gaps shown in Fig. man, personal communication). 4, although there is no direct evidence for this Excision gaps opposite a dimer (Fig. 4B) or idea. It should be emphasized that the probabil- overlapping daughter-strand gaps (Fig. 4A) ity that two UV photoproducts would generate would seem to confront E. coli with types of such a gap structure would depend not only damage not repairable by constitutive repair upon the distance between them, but also on systems, and therefore lethal in strains not able the extent of DNA degradation occurring after to induce a type of repair activity capable of UV irradiation. 882 WITKIN BACTERIOL. REV. Several otherwise puzzling features of UV Kinetics of Induction and Decay of SOS mutagenesis can be explained if it is assumed Repair Activity that the two kinds of gaps shown in Fig. 4 are The kinetics of induction and decay of induc- the primary targets for SOS repair activity in ible error-prone repair activity have been deter- E. coli. Although all UV-induced mutations mined in two very different systems with re- examined in Uvr- strains become irreversibly markably similar results. "fixed" against initially antimutagenic post- Witkin (238) investigated thermally induced treatments such as PRL (164) or caffeine (239) SOS repair activity in tif- B/r derivatives that coincidentally with DNA replication, some UV- had been rendered viable at 420C by selection induced mutations in Uvr+ strains lose their for a secondary mutation (Sfi-) that suppresses photoreversibility as dimer excision is com- thermally inducible filamentous growth with- pleted, well before DNA replication, indicating out altering other thermally inducible tif-me- that some mutations in Uvr+ strains originate diated functions (75). Kinetics of induction of as rare errors in the repair of excision gaps SOS repair activity were assayed in both Uvr+ before DNA replication (30, 164, 165). The kind and Uvr- tif- Sfi- derivatives by determining ofexcision gap shown in Fig. 4B could provide a the degree of enhancement of UV mutability site requiring prereplication SOS repair in obtained, after exposure to a low fluence of UV Uvr+ strains. Since induction of SOS repair radiation, with increasing periods of incubation requires considerable time (see below), gaps at 420C (0 to 30 h) before completing incubation requiring this repair should persist as targets at 300C. No significant thermal enhancement for exonuclease V degradation and could be- was obtained if the incubation at 42°C was come considerably enlarged by the time the shorter than 30 min. Optimal enhancement inducible repair system is available. This could was obtained with incubations at 420C for 45 account for the "long patch" character of the min after UV irradiation or longer. The ability recA+ lexA+-dependent (i.e., inducible) path- to enhance UV mutagenesis induced by a 60- way of excision repair, which may well be the min incubation at 420C was found to decay at same SOS repair activity that also repairs over- 300C (the noninducing temperature) to about lapping daughter-strand gaps. As Bresler (23) half its optimal level in 30 min, and was no has pointed out, subsequent excision of the sec- longer detectable after 90 min at 300C. Kinetics ond UV photoproduct, once the mutagenic of induction at 42°C and of decay at 300C of event has closed the first excision gap, would thermally induced ability to enhance UV muta- lead to a "pure clone" of the mutant type, at bility (i.e., of thermally induced SOS repair least in cells containing only one surviving copy ability) were virtually identical in Uvr+ and of the genome. Uvr- tif Sfi- strains. The time required for Sedgwick's overlapping daughter-strand gap induction of SOS repair is consistent with the hypothesis (190a) could also explain why detect- observation (69) that no error-prone gap-filling able UV-induced mutations are produced pri- occurs during the first 20 min after UV irradia- marily and probably exclusively in the first tion. postirradiation cell division (60, 230, 238), al- Defais et al. (52) determined the kinetics of though pyrimidine dimers persist and are ex- induction and decay ofUV-inducible capacity to changed into daughter strands in Uvr- strains W-reactivate UV-irradiated X phage. Irradi- (71). Closely spaced UV photoproducts that pro- ated X were reactivated maximally when infec- duce overlapping daughter-strand gaps in the tion was delayed for 30 min after UV irradia- first postirradiation cell division would either tion of the host cells. The UV-induced capacity fail to do so at all in subsequent replications, or for W-reactivation decayed with a half-life of 30 would do so with a reduced probability, depend- min once the optimal level was achieved. Al- ing upon the mechanism of SOS repair (e.g., if though there are some differences, not surpris- both gaps are repaired by error-prone polymeri- ing in view of the radically different inducing zation, the photoproducts would remain in the treatments and assay systems used, the two parental strands and no overlapping daughter- sets of kinetics obtained by Witkin and by De- strand gaps would be produced after the first fais et al. are similar enough to encourage the replication). Even if mutagenic gap structures hypothesis that the same inducible error-prone are generated after the first DNA replication, repair activity effects bacterial UV mutagene- the more rapid proliferation of DNA molecules sis and W-reactivation of UV-irradiated X. not containing such damage might prevent the expression of potential second-round muta- Time of Action of SOS Repair in UV tions, since most methods used to detect in- Mutagenesis duced mutants limit postirradiation growth in Two recent studies have attempted to deter- one way or another (3). mine just when UV-induced mutations are es- VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLI 883 tablished by the action of error-prone repair. of cells not containing SOS repair activity (be- Doubleday et al. (60) have used the kinetics of cause they were exposed to an extremely low loss ofphotoreversibility (LOP) as a measure of fluence of UV radiation and were incubated irreversible commitment to UV-induced muta- afterwards at 3000), the capacity to produce an tion caused by pyrimidine dimers in E. coli optimally enhanced yield of UV-induced Trp+ WP2S (uvrA- trp-). Irradiated bacteria, incu- mutations in response to a delayed thermal bated for two generations in each of three dif- post-treatment was fully maintained at 300C for ferent media before plating on medium selec- at least 2 to 3 h. tive for Trp+ mutants, produced the same num- Photoreversibility of the enhanced yield of ber of induced mutations at the same UV flu- UV-induced mutations, rapidly lost when incu- ence, but lost photoreversibility of the induced bation after UV irradiation is at 420C, was also mutations at very different rates. In rich me- fully maintained for at least 2 h at 300C. It was dium, photoreversibility was lost before the concluded that sites for SOS repair activity are population had completed one cell division; in stable in uninduced cells. If such sites are minimal medium plus tryptophane, some pho- daughter-strand gaps, they must be refractory toreversibility of the UV-induced Trp+ muta- to constitutive error-proof recombinational re- tions was still present after an average of two pair, remaining open and unrepaired for at cell divisions; and in rich medium plus pantoyl least 2 to 3 h in the absence of SOS repair lactone, LOP kinetics were extremely slow. UV activity. Since the thermally enhanced yields of fluence had no effect on LOP kinetics, but only UV-induced mutations increase linearly with on the final yield of induced mutations. The UV fluence, however, the premutational sites authors considered their results compatible must be gaps produced by single UV photoprod- with either of two interpretations: (i) that er- ucts, which, in spite of being refractory to con- ror-prone repair may occur after one or more stitutive postreplication repair, are unable to rounds of DNA replication in which daughter- generate an SOS repair-inducing signal. These strand gaps are repaired by an error-free mech- are the "SOS-mutable, SOS-noninducing" le- anism, or (ii) that error-prone repair occurs sions discussed above in connection with muta- only in the first round of postreplication repair tion frequency response curves. at daughter-strand gaps, which are refractory Ifthe target of SOS repair activity is a gap in to error-free postreplication repair, perhaps be- DNA, the "SOS-mutable, SOS-noninducing" le- cause they form part of an overlapping gap sion responsible for thermally enhanced yields structure such as that proposed by Sedgwick as of Trp+ mutations must be a single daughter- the target for SOS repair activity. In that case, strand gap, not overlapping another in the sis- the premutational gaps would remain open un- ter molecule of DNA. In principle, nonoverlap- til the error-prone system is induced and/or ping gaps should be susceptible to constitutive completes its action. According to the latter recombinational repair, which Sedgwick has interpretation, the different rates of LOP re- found to occur efficiently at 300C in the same flect medium effects on the rate of induction tif Sfi- strain (189). The refractoriness of this and/or on the rate of action of the error-prone particular group of lesions to constitutive post- repair system. Overall population increase, replication repair may be related to unique then, need not reflect the specifically delayed properties of transfer ribonucleic acid-coding division of potential mutants, and all detecta- genes, in which most of the Trp+ mutations ble UV mutagenesis may actually occur only induced by UV in these strains (primarily during the first round of postreplication repair. ochre suppressors) probably occur. UV-induced This study also provided clear evidence, based suppressor mutations are uniquely refractory on segregational analysis, that most or all UV to excision repair when incubation after UV mutagenesis is "terminal," occurring only once irradiation is in rich medium, a phenomenon at a given site within a clone. related to "mutation frequency decline" (12, Witkin (238) used tif Mfi strains (both Uvr- 230, 242). The same specific features that make and Uvr+) to analyze the timing of thermally some transfer ribonucleic acid-coding genes less induced error-prone repair activity, confirming susceptible to excision repair than other genes the conclusion of Doubleday et al. that UV under certain conditions may also interfere mutagenesis is "terminal." These experiments with recombinational postreplication repair. also showed that thermally enhanced yields of For instance, possible "cloverleafing" of the UV-induced Trp+ mutations are produced pri- nontranscribed strand during transcription marily (possibly exclusively) before the comple- could place the premutational UV photoproduct tion of the first round of postirradiation cell in a configuration inaccessible to repair en- division, when SOS activity is present (at zymes. Another possibility is that UV photo- 420C). In populations consisting almost entirely products causing this type of suppressor muta- 884 WITKIN BACTERIOL. REV. tion have an unusual primary structure, as indicated by the action spectrum for their in- duction, which is quite different from the action a. spectra for the induction of other mutations or for the production ofpyrimidine dimers (194). 1-J Cryptic Premutational Lesions Susceptible to SOS Repair 0 The existence ofSOS-mutable lesions incapa- ble of inducing error-prone repair activity has some important implications. Although such 20& lesions are cryptic in uninduced cells, they are 0 capable ofundergoing mutagenesis in cells that 4 also receive an effective induction signal. A case in point is the fluorescent light used in my laboratory as a source of PRL. Visible light is +k known to be mutagenic (218), but the exposures I0- required to demonstrate this are usually much 10~~ ~ ~ ~ ~ ~~~- longer than the times used in photoreversal z experiments. However, as Fig. 5 shows, the 0- same exposure to PRL is far more mutagenic in 0 Cr -- tif cells given a 60-min incubation at 420C 0 5 10 15 20 25 30 immediately after illumination than in controls PRL EXPOSURE (MIN) incubated at 300C. This implies that PRL may FIG. 5. Mutagenic effect of visible light (PRL) in produce SOS-mutable, SOS-noninducing le- strain WP44,-NF (tif-1 Sfi-). Aliquots (1 x 107 to 3 x sions in DNA which are cryptic in uninduced 107 cells) from log culture were plated on 5% SEM cells, but contribute to overt mutagenesis in agar, exposed to PRL 0 to 30 min, and then incu- SOS-induced cells. Recognition that some bated at 42TC for 60 min (a) or not (0). All plates sources of visible light may produce such le- were incubated for 3 days at 309C starting either sions can provide an explanation for some oth- immediately after post-treatment (ifgiven) or imme- erwise enigmatic effects in experiments such as diately after illumination (if no thermal post-treat- those 148) in which UV irradiation is fol- ment was given). Source ofPRL was a pair of West- (33, inghouse "cool white" fluorescent bulbs. Trp+ muta- lowed by exposure to PRL and then by a second tion frequencies shown have been corrected for spon- UV irradiation. In both studies, pretreatment taneous mutations by subtracting frequencies ob- with UV + PRL enhanced the mutagenic effect tained in similarly treated 'dark" controls. Survival of the second UV exposure and this effect was (monitored by plate-wash assays) did not depart sig- attributed to nonphotoreversible UV photo- nificantly from 100% under any of the conditions products, still present in the DNA after the shown. (For detailed description ofstrain, methods, PRL treatment. Actually, SOS-mutable, SOS- and media used, see references 234, 238, and 240.) noninducing lesions produced by the PRL treat- No significant effect of thermal post-treatment was ment itself may have caused or contributed to obtained in similar experiments using the tiff strain the enhanced mutagenesis. SOS repair activ- WP2, (not shown). ity, induced by the second UV exposure, could have potentiated the otherwise cryptic muta- cantly to the mutations scored as nonphotore- genic effect of the PRL. versible UV-induced mutations. A suitable con- The production by PRL of SOS-mutable, trol would be equivalent PRL treatment of a SOS-noninducing photoproducts in DNA re- heated but unirradiated tiff population, which quires caution in the interpretation of photo- would permit enumeration of the PRL-induced reversal experiments, especially those deter- contribution to the "nonphotoreversible" frac- mining kinetics of LOP. When PRL is given tion of mutations. Various sources of PRL may immediately after UV irradiation, dimers are differ in the efficiency with which they produce split before SOS repair activity can be induced, SOS-mutable, SOS-noninducing photoprod- and potentially mutagenic DNA lesions pro- ucts. The mutagenic effect in heated tif cells of duced by the PRL treatment, which do not trig- the PRL source used in my laboratory was re- ger SOS induction, remain cryptic. However, duced to about 50% by a filter that eliminated when the PRL treatment is delayed for about wave lengths below 350 nm. an hour after UV irradiation, UV-induced SOS Heated tif populations may be useful in repair activity is fully expressed by that time, screening tests for environmental mutagens, and PRL treatment could contribute signifi- some of which, like fluorescent light, may be VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLI 885 far more mutagenic in combination with an during repair replication steps of the otherwise efficient SOS-inducing agent that could be sus- error-proof repair pathways (e.g., Fig. 1B and pected from their nonmutagenic or weakly mu- C) when these pathways operate in SOS-in- tagenic response in the usual test systems. duced cells. It should be emphasized that photoreversibil- A mutator effect of thermally induced SOS ity of UV-induced mutations gives no informa- repair activity has also been reported in a lig's tion about the nature ofthe premutational pho- mutant in which DNA ligase is inactive at high toproduct. Pyrimidine dimers may be required temperatures (157). Since prophage is ther- for the induction of SOS repair activity, but moinducible in lig's lysogens (81), SOS repair once induced, the error-prone repair system activity should be expressed at 420C as well. may cause mutations during repair of nondi- However, the data in the paper cited are not mer damage as well. In Uvr+ strains, the split- sufficient to establish that bacterial mutability ting of pyrimidine dimers may also reduce UV is increased by heat treatment of this mutant. mutagenesis indirectly by promoting subse- The increase reported is an apparent elevation quent excision of nondimer UV photoproducts of the frequency of mutants among survivors of that would otherwise compete poorly with di- more or less lethal heat treatments. Under con- mers for excision enzymes. ditions promoting filamentous growth, as heat treatment of lig's is known to do (72a), seem- Mutator Effect of SOS Repair on Undamaged ingly large mutagenic effects can often be DNA traced to a methodological artifact (3, 227) An unexpected byproduct of studies of tif-1 which, if recognized, can be satisfactorily con- derivatives of B/r was that incubation of un- trolled (see Appendix). treated control populations under SOS-induc- If SOS repair activity is mutagenic in cells ing conditions (i.e., at 4200 for 45 min or longer) containing no structural damage in their DNA, resulted in a significant increase in sponta- one might expect that all agents capable of neous mutability, estimated as about 60 times inducing prophage and other SOS functions the normal rate of mutation from Trp- to Trp+, should be at least somewhat mutagenic. In fact, indicating that the tif-1 mutation is a mutator a high correlation exists in E. coli between the at 420C (233, 234, 238). Mutator activity has mutagenicity of various treatments and their been described also in K-12 tif-1 strains (75). ability to induce prophage (127). SOS-inducing The elevated spontaneous mutability in unirra- agents should be divisible into two distinct diated controls does not obscure thermal en- classes with respect to their mutagenic potency: hancement of UV mutability, since thermal potent mutagens, like UV radiation, which not post-treatments required to demonstrate such only induce SOS repair activity but also pro- enhancement are brief (45 to 60 min), and the duce numerous sites in the DNA requiring its heated populations are small enough (2 x 107 to error-prone action, and weak mutagens, which 3 x 107 cells per plate), so that the actual num- induce SOS repair activity by arresting DNA ber of mutant colonies on heated but unirra- replication but do not otherwise damage the diated control plates is small compared with the genetic material. SOS-inducing agents of the number on platings of UV-irradiated popula- latter type should exert only the relatively tions, and is subtracted before plotting UV- weak mutagenic effect seen in dnaB's unirra- induced yields as in Fig. 3. Prolonged incuba- diated populations incubated at 4200 for 1 h, tion at 4200 ofunirradiated tif-1 cells, however, and then returned to the permissive tempera- reveals the mutator effect much more dramati- ture (237). Thymine starvation, nalidixic acid cally. inhibition, and temperature elevation in ligts When SOS repair activity is induced by UV mutants should act as similarly weak muta- radiation, the mutations caused by the radia- gens, since these treatments arrest DNA repli- tion should include some that are not due to cation but probably do not introduce any struc- error-prone repair of UV damage, but to the tural damage requiring SOS repair activity, mutator activity of the SOS repair system in except perhaps at the replication fork(s). Since undamaged portions of the DNA. The magni- SOS activity decays with a half-life of about 30 tude of the mutator effect in thermally post- min, the mutator effect should operate exclu- treated tif- populations (238) is such that the sively or at least primarily during the first fraction ofUV-induced mutations caused by the round of DNA replication after termination of SOS mutator activity does not exceed 10% un- these inhibitory treatments. Valid demonstra- der the conditions usually used, at least with tions of thymineless mutagenesis (those in respect to the Trp- to Trp+ mutations scored. which an absolute increase in the number of This maximum estimate includes mutations mutant colonies is obtained after starvation that may be caused by SOS mutator activity treatments causing little or no thymineless 886 WITKIN BACTERIOL. REV. death) usually show the expected relatively undamaged bacterial DNA and on untreated weak effect (e.g., references 29, 116). Extremely phage infecting mutagen-treated hosts. high mutation yields, calculated from data ob- tained after long periods of starvation which PLASMIDS AND SOS REPAIR cause much thymineless death, have been re- In Salmonella typhimurium, certain plas- ported, but in these cases the frequency of mu- mids, including colIb and the drug-resistance tants in thymine-starved populations may have factor R-Utrecht, have been found to enhance been grossly overestimated (see Appendix). UV resistance and increase UV mutagenesis in Sublethal starvation for thymine in strain wild-type cells (65, 107, 136). These effects have WP2. thyA , a derivative ofE. coli B/r, causes a generally been interpreted as indicative ofplas- weak but genuine mutagenic effect, which is mid-borne genes that determine or enhance er- absent in its lexA - and recA - derivatives, and ror-prone repair of UV-damaged hosts. Expres- enhances UV mutability at low fluences as ef- sion of this effect by the plasmid R-Utrecht fectively as does thermal post-treatment of tif depends upon the presence in Salmonella of a or dnaB's strains (E. Witkin, manuscript in gene product that appears to be similar to the preparation). These observations confirm the recA+ product of E. coli (137). In the Salmo- conclusion (29) that UV radiation and thymine nella tester strains developed by Ames et al. starvation share a common pathway of muta- (2), some R factors enhance the mutagenic ef- genesis. Similar weak mutagenesis and strong fect of many carcinogens in recA + but not in enhancement of low-fluence UV mutagenesis recA - hosts (146). In recA - strains, these car- have been observed after sublethal treatment cinogens are nonmutagenic with or without the with nalidixic acid (E. Witkin, unpublished R factor. An R factor introduced into E. coli K- data). Both thymine starvation and nalidixic 12 enhances UV resistance and spontaneous acid treatment, as well as , mutability in wild-type hosts, but not in recA - which is also virtually nonmutagenic in lexA- or lexA - mutants (G. Walker, personal commu- strains (29), appear to be effective inducers of nication). It is possible that some plasmids SOS repair activity. The unusual kind of thy- carry genes specifying a unique SOS repair mineless mutagenesis described by Bresler et activity, responsive to host regulation, or genes al. (24), however, which operates only under that amplify the expression of the host's own special conditions and is independent of lexA+ inducible repair system. and recA+ gene products, is clearly not due to inducible SOS repair activity. MECHANISM OF SOS REPAIR Mutations have been produced in untreated phage by irradiating the host with UV radia- Although current evidence is consistent with tion before infection (109, 115a) or by pretreat- the possibility that a single SOS repair activity ing the host with N-methyl-N'-nitro-N-nitroso- is responsible for bacterial and phage UV mu- guanidine (126). These observations have been tagenesis, there is no proofthat only one type of interpreted by Ichikawa-Ryo and Kondo (109) error-prone DNA repair is induced by UV ra- as expression of an inducible misreplication diation in E. coli. The possible mechanism(s) of system in the host that elevates the yield of SOS repair of phage and bacterial DNA will mutations occurring during the replication of therefore be discussed separately. untreated phage. These authors, however, doubt that the induced misrepair system is the Mechanism of SOS Repair in Bacteriophages same one that is responsible for UV mutagene- Devoret et al. (55) have reviewed the evi- sis of either bacterial or phage DNA, since UV dence that mutagenic W-reactivation of bacte- irradiation of a lexA - mutant caused some mu- riophage X is independent of both excision re- tagenesis of the untreated phage, although not pair and recombinational repair and that it as much as that observed in an irradiated lexA + depends upon a novel error-prone repair activ- host. It seems possible (as suggested by M. ity induced or activated in the host cell by UV Volkert) that a low level of SOS repair activity radiation and .other SOS-inducing treatments. may be expressed in some lexA- strains, too low The requirement for new protein synthesis to be readily detectable in assays for UV muta- after the inducing treatment (52, 167a) suggests genesis or W-reactivation, but detectable in the induction rather than activation of this error- much more sensitive assay system provided by prone repair activity, or of one or more of its unirradiated phage. If so, the same SOS repair components. system induced in E. coli could be responsible Proof that at least one type of UV-inducible for bacterial UV mutagenesis, for mutagenic SOS repair is an error-prone DNA polymeriz- W-reactivation and for the mutator effect on ing activity has come recently from work done VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLI 887 by Radman and his collaborators with the sin- cause transient mutator activity during repli- gle-stranded DNA phage qX174 and with syn- cation of undamaged DNA, as well as to pro- thetic template-primers (M. Radman, personal mote mutagenesis at the site of any noninstruc- communication, data summarized in M. Rad- tive lesion. man, P. Caillet-Fauquet, M. Defais, and G. In experiments using a synthetic template- Villani, 1976, p. 537-545. In Screening tests in primer, poly(dT)-oligo(dA), Villani and Rad- chemical carcinogenesis. R. Montesano, H. man (personal communication) have developed Bartsch, and L. Tomatis [ed.], IARC Publica- an assay for inducible SOS repair activity. tions, no. 12, Lyon). When UV-irradiated When the template-primer is UV-irradiated, 4X174 infects untreated E. coli, the phage crude extracts from "SOS-" cultures (tif-1 cells DNA remains largely unreplicated (i.e., single incubated at 32°C) fail to promote DNA synthe- stranded), and the amount of phage DNA syn- sis. When crude extracts from "SOS" cultures thesis that does occur is consistent with the (tif-1 cells incubated at 42°C for 45 min in me- interpretation that replication is arrested by dium containing adenine) are used, DNA syn- the first pyrimidine dimer encountered in the thesis occurs on the UV-irradiated templates, template (7). It appears, therefore, that none of and massive misincorporation (uptake of dGTP the constitutive DNA polymerases can copy and dCTP) is observed. Some misincorporation UV-irradiated DNA past pyrimidine dimers. occurs when the "SOS" crude extract is used However, if the host is exposed before infection even with unirradiated template-primer, corre- to UV radiation, under optimal SOS-inducing sponding to the mutator effect of SOS activity conditions, much of the single-stranded qX174 in undamaged E. coli cells. The in vitro misin- DNA is converted to the covalently closed dou- corporation promoted by crude extracts from ble-stranded replicative form, and the extent of heated tif-1 mutants is recA+ dependent, as is the UV stimulation of phage DNA synthesis is SOS repair activity in vivo. The in vitro assay correlated with the extent of W-reactivation. provides a basis for possible purification of the When primed 4X174 DNA is used as an in vitro protein(s) responsible for SOS repair activity. template with Kornberg's clear extracts of E. coli (128), DNA synthesis is largely prevented Mechanism of SOS Repair in Bacteria by prior UV irradiation ofthe phage DNA. The Since SOS-inducing treatments induce an er- extent of the inhibition and its photoreversibil- ror-prone DNA-polymerizing activity that is ca- ity by PRE indicate that pyrimidine dimers are pable of promoting replication past pyrimidine blocks to this in vitro DNA synthesis. The rea- in 4X174 DNA and in synthetic template- son for blockage by dimers ofDNA synthesis by primer, the simplest assumption is that the DNA polymerase I appears to be the "proof- same inducible activity is responsible for error- reading" 3'-5'-exonuclease activity of this en- prone repair of UV-damaged bacterial DNA as zyme, which increases two to three orders of well. The same activity (or another activity of magnitude, relative to the polymerizing activ- the same kind) could effect error-prone repair ity, when the 4X174 DNA template is exposed synthesis to fill gaps such as those shown in to a UV fluence of 500 J/m2 (G. Villani and M. Fig. 4. However, other possibilities (236), in- Radman, personal communication). Because of cluding a minor inducible pathway of recombi- this result, and because they have failed to national repair, have not been ruled out. identify a new induced DNA polymerase, Vil- Bridges et al. (31) have concluded that DNA lani and Radman have suggested that an induc- polymerase III is involved in bacterial UV mu- ible inhibitor ofthe activity of one tagenesis, and that its ability to effect error- or more constitutive DNA polymerases may be prone repair may depend upon an inducible the mutagenic effector. In the absence of this factor. The experiments upon which this con- inhibitor, nucleotides inserted opposite pyrimi- clusion was based were done with a polCts de- dine dimers or other noninstructive lesions rivative ofE. coli WP2M, in which DNA polym- would be promptly removed. In its presence, erase III is temperature sensitive. UV-induced the stable insertion of one or more "wrong" mutations in this strain lose their photoreversi- nucleotides would permit replication to con- bility progressively at the permissive tempera- tinue, albeit with a high probability of muta- ture (34°C), but not when the irradiated bacte- tion. There is considerable evidence (66, 128) ria are transferred to the restrictive tempera- that spontaneous mutation rates are grossly ture (42°C) after an initial postirradiation incu- affected by changes in the relative efficiency of bation for 15 min at 34°C. Bridges et al. equate polymerizing and proofreading activities in the LOP with the occurrence of error-prone DNA polymerases. Thus, an induced inhibitor postreplication repair. Their expectation, if er- of proofreading activity would be expected to ror-prone gap-filling does not require active 888 WITKIN BACTERIOL. REV. DNA polymerase III, was that some LOP would (110). In the Salmonella test system (2), most occur at the restrictive temperature, reflecting mammalian carcinogens are mutagenic, and error-prone repair of gaps formed during the 15 much of this mutagenesis requires the product min between UV irradiation and temperature of a Salmonella equivalent of the E. coli recA + elevation. Since LOP stopped abruptly as soon gene (146). The mutagenicity of many carcino- as the temperature was raised to 420C, they gens in bacteria may therefore depend upon the concluded that error-prone gap-filling requires induction of SOS repair activity. This, of active DNA polymerase III, and proposed that course, does not imply that inducible muta- its ability to insert bases opposite pyrimidine genic DNA polymerase activity is involved in dimers in the template may depend upon an mammalian carcinogenesis. Speculation to this inducible factor. These results are suggestive, effect may be useful, however, if it stimulates a and, if substantiated by further evidence, may deliberate search for induction of similar activ- implicate DNA polymerase III, as modified by ity in normal mammalian cells after treatment an inducible factor (e.g., a proofreading inhibi- with carcinogens. tor), as the agent of SOS repair of bacterial Mutagenic DNA polymerases, which pro- DNA. mote misincorporation of"wrong" bases in vitro Analysis of the molecular basis of UV muta- in assays using synthetic template-primers, genesis inE. coli and some ofits phages reveals have been found in human cancer cells (204). that base transitions, primarily from guanine- The enzyme TDT, which polymerizes DNA by (GO) to adenine-thymine (AT) (3, 40a, random "end-addition" without template in- 167b), are efficiently induced. In one study of struction (15), has also been detected in human UV mutagenesis in a single-stranded phage, leukemic cells (48, 145, 187, 205). TDT is nor- the mutations induced were predominantly C mally found only in thymus, where it may act to T transitions (l0la). If polymerization past as a somatic mutator in generating diversity of noninstructive UV photoproducts is the pri- antigenic responsiveness in T cells (5). The mary mechanism of UV mutagenesis in E. coli, presence of TDT or other error-prone DNA po- the limited information available at the molec- lymerase does not imply a causative role for ular level suggests that photoproducts involv- these enzymes in carcinogenesis, although er- ing C (not necessarily pyrimidine dimers) may rors in DNA replication have been proposed as be the most common premutational lesions, at possibly important factors in initiation of ma- least in wild-type strains in which most pyrimi- lignant transformation (e.g., references 133, dine dimers are excised. Even if such lesions 163, 187). The response of bacteria to carcino- are C-containing dimers, requiring two nucleo- gens suggests that mammalian cells, as well, tide insertions by SOS activity without tem- may react to DNA-damaging agents by induc- plate instruction, most would produce muta- ing mutagenic DNA polymerase activity as a tions meeting the operational definition of sin- transient repair function. gle-base transitions, owing to the degeneracy of Mammalian cells are capable of repairing the genetic code, the probability that either or photochemical damage to their DNA (46), as both of the uninstructed insertions will be cor- well as DNA damage caused by chemical car- rect, and the probability that only one of two cinogens and mutagens (177). UV radiation amino acids that might be affected will matter. produces pyrimidine dimers in the DNA of mammalian cells (212), and most mammalian IMPLICATIONS OF SOS REPAIR FOR cells, including normal human cells, are capa- CARCINOGENESIS ble of excising them (178). A human hereditary disease, (XP), is asso- It is not known whether DNA damage initi- ciated with various deficiencies in DNA repair ates the induction of an SOS-like cluster of capacity (44, 45, 47). The clinical symptoms of functions, including an error-prone repair ac- XP include extreme sensitivity to UV radiation tivity, in eukaryotic cells. In bacteria, most of and susceptibility to development of multiple the agents that are effective SOS inducers are skin carcinomas, usually the cause of early also known to be carcinogenic in mammals. death. In one type of XP, caused by an autoso- This is true not only for such long-known muta- mal recessive mutation, ability to perform the gen-carcinogens as ionizing radiations, UV first step of excision repair (incision) is absent light, and alkylating agents. Aflatoxin B1, a (196), as in Uvr- mutants ofE. coli. Skin fibro- potent carcinogen, induces X prophage and is blast cultures taken from this type of XP pa- mutagenic for X in E. coli K-12 (82). The carcin- tient are not only much more sensitive to UV ogen 4-nitroquinoline-1-oxide is UV mimetic in radiation than normal skin fibroblasts, but E. coli, producing a type of DNA damage that they are also UV hypermutable at low UV flu- is biologically equivalent to pyrimidine dimer ences (139) as are Uvr- mutants of E. coli. VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLI 889 There is no direct evidence that the cancer haps unique, in E. coli. Most bacterial control proneness of XP patients is a consequence of systems regulate blocks of genes, either linked their inability to excise pyrimidine dimers. (operons) or unlinked (regulons), all of which However, since unexcised dimers cause hyper- code products active in the same metabolic inducibility for SOS functions in E. coli, it pathway, or in closely related pathways (6, seems possible that the induction of an error- 179). Regulatory molecules tend to be sub- prone repair activity in response to unexcised strates, end products, or other metabolites UV damage may account for the hypermutabil- bearing some direct biochemical relation to the ity of XP cells, and may also contribute to their pathway itself. SOS functions, at least concep- high rate of carcinogenesis. Pyrimidine dimers tually, are more like the metabolically diverse are known to be causative agents of neoplastic but teleonomically (i.e., adaptively) related transformation in fish (95). Carcinogens induce groups of functions that are simultaneously ac- prophage inE. coli (156a), as well as mutagenic tivated during eukaryotic development and dif- DNA polymerase activity, and Uvr- mutants ferentiation (34, 100). The regulation of SOS are hyperinducible for both functions. Ifcompa- functions is the answer from E. coli to a typi- rable SOS-type activities are induced by carcin- cally eukaryotic question: how to turn on a ogens in mammalian cells, the induction of la- group ofgenes simultaneously when their coor- tent virus and of error-prone DNA polymerase dinate expression is beneficial under particular activity should occur simultaneously, and conditions, although their products may act in either or both could contribute to carcinogene- unrelated pathways. An understanding of SOS sis. regulation may therefore be of interest in con- There are some indications that UV and X sidering the evolution ofeukaryotic control sys- radiation induce repair functions in mamma- tems. lian cells. UV- or X-irradiated human viruses Goldthwait and Jacob (80) proposed that X are reactivated (as in W-reactivation of bacte- repressor is inactivated, in lysogens treated riophage) when the mammalian host cells are .with prophage-inducing agents, by a precursor also exposed to X or UV radiation (13, 14, 96, of DNA synthesis (possibly a derivative of ade- 135). It is not known whether this radiation- nine), which accumulates when DNA replica- induced reactivation is accompanied by viral tion is interrupted. Hertman and Luria (97), mutagenesis. A UV-inducible enhancement of who found that recA- mutations prevent UV postreplicative repair, inducible also by N-ace- induction of X prophage, suggested that exten- toxy-acetylaminofluorene, has recently been sive DNA degradation (a characteristic of demonstrated in Chinese hamster cells (51a). recA- mutants) produces breakdown products In some mammalian cell lines, including nor- that antagonize the ability of a DNA precursor- mal human cells, the DNA made shortly after inducer to inactivate the phage repressor. Tom- UV irradiation is produced in segments of low izawa and Ogawa (210) concluded that the pro- molecular weight, which are elongated and duction of X inducer is a complex biochemical joined with further incubation to produce pathway, requiring postirradiation protein syn- strands of normal molecular weight (131). The thesis. process involved in this postreplication repair To explain the similar requirements for in- appears to be primarily a form ofrepair replica- duction of prophage in lysogenic strains and of tion, since recombinational exchanges have not filamentous growth of E. coli B after UV irra- been detected. The DNA synthesized on UV- diation, Witkin (226) proposed that a bacterial irradiated templates at long times after irradia- gene coding a septum-inhibiting protein is gov- tion (e.g., 20 h) is produced in strands ofnormal erned by a repressor similar enough to X repres- molecular weight, even when the UV photo- sor to respond to the same inducer, and that the products causing the initial discontinuities are common inducer is synthesized when DNA rep- still present in the template strands (36, 132, lication is interrupted. As additional bacterial 149). Although other explanations are possible, functions were identified as belonging to the the long time required to develop the capacity same regulatory system, and as their common to synthesize DNA continuously on damaged requirement for the recA + lexA + genotype was templates could indicate induction of a type of recognized, this model was further developed repair activity (possibly error prone) that is (215, 233, 235, 237, 240) to assume that all SOS repressed in undamaged mammalian cells. functions have evolved repressors similar enough to respond to the same inducer, and REGULATION OF SOS REPAIR AND that the recA + and lexA + products are neces- OTHER UV-INDUCIBLE FUNCTIONS sary for their induction and/or expression. The The functions activated by DNA damage con- tif-1 mutation was believed to cause constitu- stitute a regulatory unit that is unusual, per- tive synthesis at 420C of an intermediate in the 890 WITKIN BACTERIOL. REV. induction pathway, thereby bypassing the re- as an inducer in SOS regulation, a possibility quirement for interruption of DNA synthesis as that is supported by the failure of nalidixic the initiating signal. acid to induce protein X synthesis in recB or X repressor, which binds specifically to X recC mutants lacking exonuclease V activity, DNA (173), is reported to bind selectively to although bleomycin, an antibiotic that pro- nicked DNA, an observation that has led to the motes DNA degradation directly, induces pro- hypothesis that SOS functions are induced by tein X in these strains (89). The properties of agents that introduce single-strand nicks into the dam-3 mutant (142), which lacks an ade- DNA, thereby providing sites for competitive nine methylase and degrades its DNA chroni- binding of SOS repressors (206). This model cally, are also consistent with this idea, since does not account for thermal inducibility ofSOS dam-3 strains express at least some SOS func- functions in tif mutants, in which DNA is tions constitutively. However, there is no sim- intact at the inducing temperature, or for the ple correlation between DNA degradation and failure of SOS induction in recA - and lexA - thermoinducibility of prophage in different mutants when their DNA is nicked. Selective dnats mutants (188). binding of SOS repressors to nicked DNA may The Gudas-Pardee model is consistent with enhance induction by "soaking up" excess re- the dominance of lexA- mutations (159) in pressor molecules, but it seems inadequate to merodiploids with the wild-type allele. It also account for the primary derepression of SOS accounts satisfactorily for the synthesis ofpro- functions. tein X in a number ofdifferent mutant strains. The first comprehensive model of SOS regu- However, the assumption that the lexA+ gene lation, specifying the roles of recA and lexA product functions as a repressor of SOS func- gene products, was proposed by Gudas and Par- tions, or that it controls a product required for dee (89), based largely upon their study of the the induction ofthese functions, is not consist- induction of protein X (112, 113) in various ent with some other facts. A major problem is strains. As recently elaborated by Gudas (88a), X prophage, which codes its own repressor, yet this model assumes that the lexA + gene product requires the lexA+ genotype for induction by is a repressor of protein X and possibly of other SOS-inducing treatments. Gudas (88a) ex- SOS functions, which may be linked in a single plains this by proposing that X induction re- operon or scattered throughout the chromo- quires the participation of a protein or small some. Inactivation of the lexA-coded repressor metabolite synthesized via the lexA+-con- results, directly or indirectly, in the expression trolled SOS induction pathway or after its of all SOS functions. In wild-type strains, SOS- expression. However, in tif-1 X lysogens, the inducing conditions initiate induction by caus- inactivation of X repressor begins without a ing the accumulation of breakdown products of lag as soon as the temperature is raised to 42°C DNA degradation, one of which interacts with and proceeds without requiring new protein the lexA +-coded repressor so as to sensitize it to synthesis (221). Thus, in tif-1 strains all pro- inactivation by an antirepressor (possibly a teins required for thermal inactivation of X protease) coded by the recA gene. The possibil- repressor appear to have been constitutively ity that the recA gene may code a protease was synthesized before the temperature elevation. proposed earlier by Roberts and Roberts (180), Nevertheless, the lexA + gene product is who found that X repressor is proteolytically strictly required for thermal induction of X cleaved following prophage-inducing treat- prophage in tif-1 strains, and this requirement ments. According to the Gudas-Pardee model, is for the actual inactivation of X repressor the tif-1 mutation is interpreted either as rather than for any subsequent step leading to a temperature-sensitive operator-constitutive cell lysis. This has been demonstrated by mutation at the binding site of the lexA-coded showing rapid lysis oftif-1 lexA- cells lysogen- repressor or as a mutation affecting the activity ized with a X mutant (XcI857) which codes a of the recA gene product. Gudas (88a) considers temperature-sensitive repressor, whereas the the latter possibility more probable, and it is same strain lysogenized with wild-type X does consistent with the tight linkage between tif-1 not induce its prophage at 420C (E. Witkin, and recA mutations (40). The recA product, as unpublished observation). Many lexA- muta- altered by a tif-1 mutation, would presumably tions severely depress and delay UV induction function as an antirepressor at 420C, inactivat- ofX prophage, and this effect, too, is exerted at ing the lexA + product without requiring the the level of X repressor inactivation (59). The participation of a metabolite of DNA degrada- Gudas-Pardee model, as well as another model tion. (162) in which the lexA+ product is assigned A valuable aspect of this model is the pro- the role of repressor of some SOS functions, posal that a DNA breakdown product may act must assume that this gene product acts in an VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLJ 891 entirely different capacity to influence X pro- tive level of recA product is a critical factor in phage induction. Although this is not impossi- SOS induction, can explain all-or-none nonin- ble, it seems simpler to assume that the lexA+ ducibility or all-or-none constitutive expres- gene product participates in the regulation of sion of SOS functions. Such models, however, all the functions it controls in the same way. cannot easily explain the "split" phenotypes of Another set of facts that must be explained tsl- or rnm- mutants (described above), or the by any model for SOS regulation is the diver- quite different pattern of expression of SOS sity of patterns of expression of different SOS functions seen in lex-113 mutants (58). For any functions caused by mutations that probably one lexA allele, there can be only one level of map within the lexA locus, such as tsl- (158, recA + product produced under its control, 160) and rnm- (215). For example, in tsl- mu- whether that level be normal, abnormally tants, filamentous growth (160) and synthesis high or abnormally low. Yet tsl- and rnm- of protein X (89) are expressed constitutively mutants are noninducible for some SOS func- at 420C, but X prophage is not thermally induc- tions, whereas they express others normally or ible, and UV mutability (i.e., SOS repair ac- are constitutive for still others under the same tivity) is greatly reduced compared to the wild conditions. In lex-113 mutants, the phenotype type (158). In rnm- mutants, no UV mutabil- is like lexA- in all respects, except that fila- ity is detectable, but exonuclease V activity mentous growth is expressed constitutively. after UV irradiation is controlled to a nearly One would have to make some complicated normal extent (215), and synthesis of protein assumptions to explain how various lexA al- X is noninducible (D. Spencer, personal com- leles (which can only either raise or lower the munication). In rnm- mutants, synthesis of amount ofrecA product synthesized according the inhibitor of exonuclease V is constitutive to this model) can produce such a variety of (M. Volkert, personal communication). patterns of "split" phenotypes with respect to Models assuming that the lexA gene codes the SOS functions. It should be emphasized, how- common repressor of SOS functions could ac- ever, that tsl-, rnm-, and lex-113 mutations count for such "split" phenotypes only by as- have not been definitively mapped within the suming numerous nonidentical operator sites lexA gene, although the linkage data (espe- that respond differentially to an altered re- cially for tsl- and rnm-) are consistent with pressor. The assumption that all SOS func- this possibility. tions belong to a single operon (89) cannot Figure 6 illustrates another way of inter- account for them in any obvious way. preting SOS regulation. The essential features A different type ofmodel is based on Clark's of the model are as follows. (i) Each SOS func- suggestion (41) that the lexA gene codes a tion expresses the activity of a separate op- regulator of the recA gene, an attempt to ac- eron, controlled by its own repressor. (ii) SOS count for the extensive copleiotropic effects of repressors are similar enough to bind the mutations in these two genes. Mount (per- lexA+ gene product when bound to their re- sonal communication) has developed a model spective operator sites, but are not identical. in which the quantitative level of the recA + (iii) In undamaged wild-type cells, the lexA+ gene product (assumed to act as an antirepres- product is bound to all SOS repressors, sor that destroys SOS repressors under SOS- thereby preventing their inactivation by con- inducing conditions) is one of the critical fac- stitutive antirepressors. In Fig. 6, the antire- tors determining whether or not a strain can pressors are shown as cellular proteases, on express SOS functions normally. The lexA + the assumption that proteolytic cleavage is the gene product is assumed to act as a regulator primary mechanism of SOS repressor inacti- ofrecA + synthesis, resulting in a level ofrecA + vation. In that case, the lexA+ product may product that permits normal SOS induction in serve to block protease-sensitive sites on the wild-type strains. Mutations in the lexA gene SOS repressors, or it may act more positively that result in a reduced level ofrecA+ product as a protease inhibitor. Whatever the primary cause noninducibility ofall SOS functions (the mechanism of derepression, the role of the LexA phenotype). Certain other mutations in lexA+ gene product is protection ofSOS repres- the lexA gene (spr-) are assumed to increase sors against inactivation in the absence of an the synthesis ofthe recA + product to an abnor- SOS induction signal. (iv) The recA-tif com- mally high level, and thereby cause constitu- plex (so designated because the tif-1 mutation tive expression of SOS functions even at 300C, maps in the recA region and may affect the if other requirements for induction are pro- same protein) is inactive in the absence of an vided by a tif-1 sfi- genotype. Any model in SOS signal, in the sense that it is unable to which the 1exA product regulates synthesis of remove the lexA+ product from its binding the recA product, and in which the quantita- with SOS repressors. In the inactive state, the 892 WITKIN BACTERIOL. REV.

I. SOS FUNCTIONS EPESD (rcAA+hArA +adaoPsd cells, lid-i at 30) SOS repair Protein rerirm pat~ editorsewum protein X Emxoncleouinhibitor

k\ haI 2 TWAK I2 PTEm 3 PSOTEAEO 4 PRISE 5 AexA+ pvdsd WAS ed SWS represeors protects thewr proteine-sesiive sites opint cleva by coruttive p.iamse

Ar-cttif+ complex m torm iater for protose- site$ SOS wwkx ~ (wgafoj-own jwli) etuupmeoseidotve "~)

X proho

,/ 'POEAKIM 1 POIASE I 2. SOS FUNCTIONS IDUu a. recA~tf om"ple activated by f v UV, X-adiation MMC, etc. octw +r*CAkTtif K DN J 0 it S activator (4*" AAAA X Propho D activaed b. rucA+-td9- complex activatd by complex incublion at 42*C withiW dwnap complex i to DNA (no SOS activiator required) activated erAW prnP ct,wn Kmabenait ~~fragnmets from SOS rew ssors of repressorffter cleavag prat*eeby FIG. 6. Model to account for the regulation of SOS functions in E. coli (see text for explanation and alternative versions). recA-tif complex may be free, as shown in Fig. repressors, thus exposing their protease-sensi- 6, or bound to the lexA + product. In either tive sites to proteolytic cleavage (or permit- case, activation of this complex, enabling it to ting their inactivation by some other constitu- remove the lexA + product from SOS repres- tive antirepressor). sors, occurs in tifr strains only when an SOS The model in Fig. 6 explains currently activator is produced as a consequence of ar- known facts satisfactorily and generates a rested DNA replication. The activator may be number of testable predictions. Like any a precursor ofDNA replication (80) or a break- model offered when knowledge is incomplete, down product of DNA degradation (89), or (as it may require drastic revision or rejection as suggested by H. Ginsberg) activation may be additional facts become known, and it is pro- accomplished under some conditions by a pre- posed here as a working hypothesis. The domi- cursor and under other conditions by a break- nance oflexA - mutations in merodiploids with down product. In the tif-1 mutant, activation lexA+, regardless of which allele is located on of the recA-tif complex occurs when the tem- the F' factor (159), is readily explained. Muta- perature is raised to 420C without requiring an tions causing the noninducible phenotype for activator, or at lower temperatures when ade- all SOS functions (the LexA- phenotype), ac- nine is present (80, 123, 234). Activation of the cording to this model, are those eliminating complex permits its binding to the lexA + prod- the ability of the lexA + product to bind the uct unless it is already so bound before activa- activated recA-tif complex or to be released tion. In either case, the binding of activated from SOS repressors when bound to the acti- recA-tifcomplex to the lexA + product results in vated complex. In merodiploids, any lexA + the release of the lexA + product from SOS product bound to an SOS repressor would be VOL. 410, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLI 893 removed under SOS-inducing conditions, and and responds normally to activated recA-tif could be replaced by a lexA- product, which complex when so bound. Assuming nonidenti- would remain irreversibly bound to that re- cal repressors, changes at different sites pressor. If the inactivation of SOS repressors within the lexA gene could thus cause a multi- and/or the expression of SOS functions is slow plicity of different patterns of constitutivity, compared with the rate of binding of SOS re- inducibility, and noninducibility of the var- pressors by the lexA product, all SOS repres- ious SOS functions. sors would ultimately become saturated with A prediction generated by the model in Fig. the irreversibly bound product in a mero- 6 is that lexA mutations that cause "split" diploid. phenotypes for SOS functions should cause a Mutations that map in the recA region but corresponding pattern of "split" dominance, have a Lex- phenotype (noninducibility for such that any function that is noninducible in SOS functions but normal recombination abil- the mutant should be dominant in merodip- ity), such as zab- (40) or lexB- (11), may be loids with lexA +, whereas any function that is recA alleles affecting a part of the recA gene expressed constitutively in the mutant should product not essential for genetic recombina- be recessive. Insofar as the dominance/reces- tion. Both of these mutations are recessive siveness of individual components of "split" with respect to UV sensitivity in merodiploids phenotypes has been determined, the results with the wild-type allele (J. George, personal agree with this prediction. In tslh mutants, communication). This is consistent with the low UV mutability (i.e., poor inducibility of model in Fig. 6, since the noninducible pheno- SOS repair activity) is dominant, but constitu- type in these mutants is ascribed to changes in tive filamentous growth at 420C is recessive in the rec-tif complex, such as to prevent either merodiploids with the wild-type allele (158). In its activation or its interaction with the lexA+ the lex-113 mutant (which is closely linked to product. Presence of the normal allele should lexA but not unambiguously located within restore inducibility in the merodiploid. The tif- the lexA locus) the constitutive filamentous 1 mutation, on the other hand, should be dom- growth of the mutant is recessive in merodip- inant, or at least partly dominant. Although loids with the wild-type allele (J. Donch, per- most recA - mutations would be expected to be sonal communication), but its UV sensitivity recessive, partial dominance of some recA al- and UV nonmutability (i.e., noninducibility of leles would be consistent with the model, e.g., SOS repair activity) are dominant (86). The if the mutation caused unactivated recA prod- constitutive filamentous growth of the lex-113 uct to bind irreversibly to the lexA + product. mutant is suppressed by a lexA - mutation This model can explain the variety of "split" (59), an observation that will add further sup- phenotypes for SOS functions caused by muta- port to the model if these mutations prove to tions that are tightly linked to lexA- muta- be allelic. tions, and that may affect the same protein. Another prediction based on this model is These include tsl-, rnm-, and possibly lex-113 that the phenotype ofa deletion mutant lacking (see above). Additional examples of "split" the lexA gene should be constitutive for all SOS phenotypes in derivatives of lexA- strains se- functions, whereas the phenotype of a recA lected for increased UV resistance have been deletion mutant should be noninducible for all reported (28, 191). A single mutation in the SOS functions, assuming that such mutants lexA gene could cause a variety of different are viable. If recessiveness is interpreted as patterns of expression of SOS functions, ac- indicative of loss of function of the type one cording to the model in Fig. 6. If the mutation might expect of a deletion mutant, the reces- permits normal binding of the altered lexA siveness of noninducible zab- and lexB- muta- product to a particular SOS repressor, but pre- tions indicates that the positive action ofa func- vents its response (when so bound) to an acti- tional recA allele is necessary for SOS induc- vated recA-tif complex, the mutant will be tion, whereas the recessiveness of the constitu- noninducible for that SOS function. If the tive filamentous growth in tsl- (and possibly in same altered lexA product is unable to bind lex-113) mutants indicates that the positive ac- one or more other SOS repressors, the mutant tion of a functional lexA allele prevents induc- will express the corresponding function(s) con- tion. stitutively, since the repressor(s) will be ex- The model in Fig. 6 can also explain the posed continuously to inactivation by antire- unique phenotype of the double mutant tsl- pressor (e.g., cleavage by protease). The mu- recA- (108, 158a, 161, 162), which is different tant may exhibit normal inducibility of still with respect to some SOS functions from either other SOS functions, if the altered lexA prod- of the single mutants. The double mutant is uct binds normally to some SOS repressors unlike any other recA- strain in its UV resist- 894 WITKIN BACTERIOL. REV. ance and ability to perform some error-free sarily requiring that every SOS repressor be postreplication repair (in Uvr+ strains) (158a, susceptible to cleavage by a different protease, 161, 162), and in its constitutive synthesis of this version of the model would require that protein X (88a) and filamentous growth at 420C more than one protease have the capacity to (161, 162). The tsl- recA- strain differs from the cleave SOS suppressors. tsl- recA + mutant in its inability to control A puzzling set of observations concerns the DNA degradation after UV irradiation (i.e., its need for new protein synthesis after various failure to induce the inhibitor ofexonuclease V) SOS-inducing treatments in order to inactivate and in its inability to produce any UV-induced X repressor in various lysogenic strains. The mutations (158, 158a). The phenotype of the presence of chloramphenicol during incubation double mutant suggests an interaction between at 420C does not prevent repressor inactivation the recA and lexA gene products such that mu- in tif-1 (221) or dnaBts (166) lysogens, nor does tations in each can affect the role ofthe other in chloramphenicol, added immediately after SOS induction. An interaction of this type is gamma irradiation, prevent repressor inactiva- expected in the model in Fig. 6. Devoret (per- tion in wild-type lysogens (221). After treat- sonal communication) has proposed a model in ment with MMC or UV radiation, however, X which the recA and lexA gene products form a repressor is not inactivated in the presence of complex which acts as a "vise" or clamp, detect- chloramphenicol (198, 210), indicating that new ing SOS repressors after SOS-inducing treat- protein synthesis is needed to accomplish de- ments and positioning them so as to permit repression ofprophage when these inducers are their inactivation by an antirepressor. Such a used. These results imply that different SOS- model (if one assumes different repressors for inducing treatments are not alike in the man- SOS functions) could also account for "split" ner in which they generate an effective SOS- phenotypes and for unique phenotypes in dou- inducing signal. Clarification of this question ble mutants, but could not easily account for Inay be crucial for evaluation ofSOS regulation the apparently invariable dominance of lexA- models. alleles, unless it is assumed that complete inac- tivation of the lexA product is lethal. PROTEASES AND THE EXPRESSION OF A final aspect of the model in Fig. 6 is that it SOS FUNCTIONS assigns the primary repressor inactivation to The initial basis for considering protease ac- proteases coded by genes outside the recA and tivity as a possible primary SOS repressor-inac- lexA loci. The proteases are shown as different tivating event was the proteolytic cleavage of X for each SOS repressor. This degree ofdiversity repressor that occurs when X prophage is in- is not a necessary feature of the model, al- duced, although it is not known whether this though more than one antirepressor must be cleavage is the cause or the consequence of postulated. If there were only one SOS antire- repressor inactivation (180). The specificity of pressor, there should be a third locus (the locus proteases and their importance in both bacte- coding the antirepressor) in which mutations rial and mammalian cell economy are receiving could eliminate inducibility of all SOS func- increasing attention (79, 79a, 150, 178a, 207). tions. Since no such mutations have been found A strong reason for invoking protease activ- unlinked to recA and lexA, it is assumed that ity as a major factor in modulating the expres- more than one protease is required to inactivate sion of SOS functions, as well as in their induc- all SOS repressors. tion, is the possibility that the ion gene, muta- A variation of this model, which fits the facts tions in which cause extreme filamentous equally well, could operate if the lexA+ product growth in response to SOS-inducing treatments functions actively as a protease inhibitor, (1, 104), may code a protease or a product that rather than as a passive protector of protease- regulates protease activity. The Ion gene is ap- sensitive sites on SOS repressors. In that case, parently identical with the gene degT (199). the lexA + product need not be assumed to inter- Mutations in lonldegT greatly increase the sta- act with SOS repressors, but could be bound bility of specific polypeptide fragments that instead to the antirepressor proteases in the would otherwise be rapidly degraded, as well as uninduced state, and its inactivation by acti- the stability of some other proteins (79, 79a). vated recA-tif complex could result in the Mutations designated degR have been reported release of proteases and in the cleavage of SOS to alter the stability of different polypeptides, repressors. "Split" phenotypes exhibited by suggesting a unique pattern of specificity (2a), tslh, rnm- and other mutants believed to alter although it is apparently not yet certain the lexA product would then be due to specific- whether degT and degR are in fact distinct loci ity of interaction of the altered gene product (79a). The products of the deg gene(s) have not with different proteases. Although not neces- been identified, but they could be proteases, VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLI 895 protease inhibitors, or products that otherwise lizing the postulated septum inhibitor (if loni regulate selectively the rate of degradation of degT codes a protease) or by increasing the rate various proteins. The pleiotropy of lon/degT of its synthesis. Mutations affecting protease mutations includes (among others) effects on activity and/or specificity could mimic regula- the rate of transcription of the gal operon (138), tory mutations by affecting the stability of reg- on the synthesis of capsular polysaccharides ulatory proteins. The apparent regulatory role (37), and on the capacity to be lysogenized by X of the ion product in the galactose operon (138) phage (217). All of these effects could be the may exemplify the difficulty of distinguishing consequence of altered stability of a variety of between mutations directly affecting regula- metabolically unrelated gene products due to a tory protein structure and those affecting their lonldegT mutation affecting (directly or indi- metabolism (6). rectly) the specificity and activity of a particu- The limited and often very brief period dur- lar protease or protease inhibitor. Sfi- muta- ing which SOS operons are derepressed makes tions (suppressors of filamentous growth), the expression of SOS functions particularly which eliminate the filamentous growth ex- susceptible to mutations that increase or de- pressed at 4200 by tif-1 mutants (75, 238), may crease rates of transcription, rates of transla- also affect the structure or activity of proteases tion, stability of mRNA or stability ofproteins. or protease inhibitors, imposing a new pattern To the extent that such mutations act pleiotrop- of stability upon numerous proteins. Independ- ically on specific spectra of substrates, they ent Sfi mutations which suppress the filamen- may amplify or diminish the expression of dif- tous growth of the ion- wild-type strain E. coli ferent SOS functions in a variety of patterns. B may or may not also alter sensitivity to peni- Mutations at any ofthese levels, perhaps exem- cillin, and at least one such mutation (responsi- plified by ion and sfi mutations, may not only ble for the nonfilamentous phenotype of strain 'fine-tune" the expression of SOS functions but B/r) radically changes the kinetics of cell divi- could also represent a means of generating sion and death in the stationary phase of asexually, via a single mutation, a rich diver- growth (222). sity of phenotypic variation. Their importance A change in the half-life of most proteins in in evolution should not be underestimated. E. coli may not cause a strikingly obvious change in the phenotype, as long as the protein CONCLUSIONS does not become so unstable that it is essen- UV-induced mutations in E. coli and some of tially unable to perform its function. Altered its phages are caused primarily by inaccurate stability ofthe proteins synthesized in response repair of UV-damaged DNA. Most of the UV to an SOS induction signal, however, could damage produced in the DNA of surviving bac- drastically alter the expression of at least some teria is repaired by relatively "error-proof" SOS functions. SOS operons are transcribed mechanisms (e.g., photoreactivation, "short- only during a limited period oftime while SOS- patch" excision repair, the major pathways of inducing conditons prevail, and rapid degrada- recombinational postreplication repair), which tion of their products would, a priori, be ex- do not contribute substantially to UV mutagen- pected. This expectation is consistent with the esis. Some kinds of DNA damage (probably 30-min half-life of SOS repair activity (52, 238) single-strand gaps not repairable by any consti- and with the existence of a class of unstable tutive accurate mechanism) are targets for the proteins in bacteria with half-lives of 20 to 60 activity of inducible "error-prone" repair activ- min (79a). When UV radiation is the inducing ity ("SOS" repair), which is entirely responsible agent, transcription ofthese operons is possible for UV mutagenesis inE. coli and in X bacterio- only until DNA repair is completed and normal phage. SOS repair may participate in two mi- DNA replication resumes. The length of the nor DNA repair pathways: a "long-patch" exci- filaments produced after exposure ofE. coli B sion repair and chloramphenicol-sensitive post- (a ion mutant) to UV radiation is strictly pro- replication repair. portional to the amount of protein synthesis SOS repair activity is repressed in undam- that is allowed to occur before DNA repair is aged wild-type cells, but is induced in response completed, an observation that led to the hy- to UV radiation and other agents (including pothesis that the duration ofthe delay in septa- many mutagens and carcinogens) that damage tion is quantitatively related to the amount ofa DNA or interrupt its replication. Such treat- septum-inhibiting protein synthesized during ments generate a regulatory signal (the "SOS" the limited period ofderepression (226). George signal), which initiates a complex induction et al. (75) have pointed out that Ion mutations process culminating in the derepression of a could cause extreme filamentous growth in re- group of metabolically diverse but coordinately sponse to SOS-inducing agents either by stabi- regulated functions ("SOS" functions), all of 896 WITKIN BACTERIOL. REV. which presumably promote the survival of the activity is about 30 min. These observations are damaged cell or that of its phages. Included consistent with the possibility that the same among SOS functions, in addition to error- inducible error-prone DNA polymerase activity prone DNA repair activity, are prophage induc- is responsible for both bacterial and phage UV tion, cell division delay (sometimes leading to mutagenesis. filamentous growth), inhibition of the DNA- The induction of at least some SOS functions degrading activity of exonuclease V and "aber- by UV radiation requires the absorption of two rant" reinitiation of DNA synthesis at the chro- photons. UV-induced mutations, however, may mosomal origin. Expression of these and other be caused either by single UV photoproducts or inducible SOS functions requires the recA + and by the cooperative interaction of two UV photo- lexA + gene products and new protein synthesis products. All or nearly all detectable UV-in- during or after the inducing treatment. The duced mutations occur before the second postir- nature of the SOS induction signal is not yet radiation DNA replication. In wild-type understood, nor is the mode of regulation of strains, some UV-induced mutations occur be- SOS functions. fore DNA replication via SOS repair of gaps At least one kind of SOS repair activity in- resulting from excision of UV photoproducts; duced in E. coli has been identified as a DNA- others occur during or after the first postirra- polymerizing activity that is distinguishable diation DNA replication via SOS repair ofdam- from the activities of the constitutive DNA po- age (probably gaps in the daughter strands) lymerases I, II, and III by its ability to polymer- caused by unexcised UV photoproducts passing ize DNA past pyrimidine dimers in the tem- through a replication fork. plate strand, with a high probability of error. There is some evidence that mutagens and UV mutagenesis in E. coli and X bacteriophage carcinogens induce DNA repair activities in may be due primarily to the insertion of mammalian cells. It is not known whether "wrong" bases by this UV-inducible SOS repair these activities are error-prone or whether they activity as it replicates DNA past noninstruc- are induced in coordination with other SOS-like tive UV photoproducts, which are not necessar- functions in resonse to DNA damage. If so, at ily pyrimidine dimers. The inducible compo- least two such functions (induction of latent nent of this SOS repair system need not be a virus and of error-prone DNA repair activity) new DNA polymerase, but may be a factor that could contribute to carcinogenesis. inhibits the 3'-5'-exonuclease "proofreading" activity of one or more of the constitutive DNA APPENDIX: THE CELL polymerases, or one that relaxes their template DENSITY ARTIFACT dependence in some other way. It is possible Many studies of bacterial mutagenesis are that more than one type of SOS repair activity flawed by an artifact that has not been gener- is induced in E. coli. In that case, some or all of ally recognized. The problem stems from the the UV mutagenesis occurring in bacterial and common assumption that percentage of sur- in X phage DNA could be due to error-prone vival, after termination of a more or less lethal repair by a minor inducible pathway of recom- treatment, is independent of the cell density at binational repair, although there is no evidence which the bacteria are subsequently plated. that such a system is induced. The "cell density artifact" can be illustrated Anomalous induction of SOS repair activity by describing a protocol widely used to measure in undamaged cells containing intact DNA ca- the mutagenic effect of UV radiation in an pable of replication results in a mutator effect, auxotrophic (e.g., Trp-) strain, scoring muta- i.e., greatly increased generalized spontaneous tions to prototrophy (Trp+). This kind of sys- mutation rates. SOS repair activity in crude tem, when used properly (3), allows accurate extracts of UV-irradiated E. coli also promotes calculation of the frequency of induced muta- misincorporation of bases during replication of tions per survivor. A key feature ofthe auxotro- unirradiated synthetic template-primer DNA. phy to prototrophy system is the use of the Both in vivo and in vitro, however, the pres- same medium for assaying auxotrophic sur- ence of UV photoproducts in the DNA greatly vival and for selecting prototrophic mutants. increases the mutagenic effect of SOS repair The medium used (SEM) is a minimal agar activity. partially enriched with nutrient broth (1 to 5%, The kinetics of induction and decay of SOS vol/vol). When a highly diluted portion of a repair activity are similar, whether the error- Trp- culture (100 to 300 cells) is plated on SEM, prone activity is assayed by its capacity to en- the tryptophane in the supplement allows this hance bacterial UV mutagenesis or to promote small number of cells to undergo many cell X UV mutagenesis in UV-irradiated hosts. In divisions, so that each cell forms a small but both types of assay, the half-life of SOS repair visible colony before its growth is arrested by VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLI 897 exhaustion of the medium for the required population. It is usually taken for granted that growth factor. When a vary large population of percentage of survival, as indicated by diluted Trp- cells (107 to 108) is plated on the same platings, is the same as percentage of survival medium, the required amino acid is used up in the undiluted populations plated to select after relatively few divisions, and the auxo- Trp+ mutants. The assumption that percentage trophic population forms a confluent but barely of survival is independent of cell density on the visible lawn. Any Trp+ mutants either present postirradiation plating medium has been in the inoculum or arising during the divisions shown to be invalid for UV-treated populations on the plate can continue to grow to form large of the lon- strain B, but is valid for one of its colonies. This system was originally developed nonfilamentous derivatives, strain B10/r (227), by Demerec and Cahn (54a). and is also valid for strain B/r over a wide The use ofSEM medium in the auxotrophy to range of cell densities (E. Witkin, unpublished prototrophy system automatically satisfies observation). three major requirements of sound methodol- Under some conditions of postirradiation cul- ogy for bacterial mutagenesis. (i) Survival and tivation, strain B, like lon- mutants of K-12, mutation are scored on the same medium, thus forms long filamentous cells and is conse- eliminating any medium effect on survival as a quently UV sensitive (104, 222). Under differ- possible source of error. (ii) The mutagen- ent postirradiation conditions, however, these treated population undergoes a few cell divi- bacteria divide normally and are UV resistant sions before stringent selection for the proto- (lb, lc, 2a, 226). Survival after UV irradiation, trophic phenotype begins, permitting pheno- or after treatment with other SOS-inducing typic expression of any newly induced muta- agents, is not irreversibly determined in lon- tions. This requirement is not met when muta- strains at the time the treatment is terminated, gen-treated auxotrophs are plated directly on but can be influenced subsequently over several minimal agar, since the phenotype of potential orders of magnitude by manipulating the com- mutants remains auxotrophic until new jnfor- position of the medium, the temperature of in- mation in the DNA can be transferred to the cubation, or the cell density. Filamentous gene product. (iii) The treated population is growth itself appears to be lethal unless cell allowed to undergo the divisions required for division is resumed before a certain "critical phenotypic expression while immobilized on length" is reached (117a), and is the cause ofthe solid medium, thus preserving the essential 1:1 UV sensitivity ofE. coli B and of lon- mutants relation between the number of mutant colo- of K-12 to most SOS-inducing agents. High cell nies counted (above the number produced by density is one ofthe factors (among others) that unirradiated controls) and the number ofmuta- can prevent extreme (i.e., lethal) filamentous tions induced. Allowing phenotypic expression growth by promoting early resumption of nor- to occur during a period of liquid cultivation mal cell division and radiation resistance in long enough to guarantee complete expression lon- strains, a phenomenon described as permits division of induced mutants before "neighbor restoration" (1, 53a) or "crowding re- plating, and makes accurate determination of covery" (227). Thus, the "cell density artifact" mutation frequency impossible. applies to any strain treated with an agent that Before UV irradiation and after each incre- causes extensive filamentous growth under the ment of radiation exposure, both diluted and conditions used to determine percentage of sur- undiluted aliquots of the Trp- cell suspension vival, but not under the conditions used to se- are plated on SEM agar. The diluted samples lect mutants. Under these conditions, gross are used to calculate percent survival for each overestimation of the frequency of induced mu- UV fluence. The undiluted samples are used to tations per survivor results from extrapolating enumerate Trp+ colonies after 2 to 3 days of percentage of survival from diluted to undi- incubation. The frequency of UV-induced mu- luted platings. The actual size of the viable tations is then calculated, after subtracting the population from which the observed number of number of Trp+ colonies on plates seeded with induced mutations is arising may be 100 times unirradiated controls. (The number of Trp+ col- larger than estimated, if filamentous growth onies on control plates depends primarily upon kills 99% of the cells plated at low density, but the final size of the plated population, and kills none of the cells plated at high density. therefore is constant over a wide range ofinocu- The cell density artifact due to filamentous lum sizes, at a given level of nutrient broth growth can sometimes apply to lon+ strains, enrichment.) Any excess of Trp+ colonies over such as K-12, 15TAU, and others that are as the number on control plates represents the UV-resistant as strain B/r. E. coli K-12 wild number of Trp+ mutations induced by UV ra- type does not form long filaments after UV diation in the surviving fraction of the plated irradiation, and the cell density artifact proba- 898 WITKIN BACTERIOL. REV. bly does not affect studies of UV mutagenesis in ACKNOWLEDGMENTS this or other lon+ strains. However, lon+ UV- Most of the work described in papers by the au- resistant strains are capable of producing long thor and co-workers cited in this review was sup- filamentous cells, and may well be subject to ported by Public Health Service grant AI-10778 from the cell density artifact, after relatively pro- the National Institute ofAllergy and Infectious Dis- longed starvation for thymine (243a), or after eases. incubation for 1 h or more at 420C in tif-1 (39, 122), dnaBts (99a), or lig's (72a) derivatives. ADDENDUM IN PROOF Thus, lon+ K-12 and similar strains appear to S. Meyn, T. Rossman, and W. Troll (Proc. Natl. be capable of undergoing "filamentous death" Acad. Sci U.S.A., in press) have shown that tif-1- after some SOS-inducing treatments. In con- mediated thermal inactivation of X repressor is pre- trast, strain B/r does not form filaments when vented by a protease inhibitor, antipain, which also its dnaB's or tif-1 derivatives are incubated at blocks the induction and/or expression ofSOS repair activity and filamentous growth. Antipain does not 420C, and seems to be immune to the cell den- promote uncontrolled DNA degradation after UV sity artifact under any condition examined (E. irradiation of wild-type cells, however, and there- Witkin, unpublished observations). fore does not seem to interfere with induction of the Microscopic observation of treated bacteria, inhibitor of exonuclease V. These observations are incubated for several hours at relatively high consistent with the model of SOS regulation in Fig. and low cell densities after an SOS-inducing 6, but do not support models in which a single pro- treatment, can often indicate whether the cell tease is the antirepressor for all SOS functions. density artifact is likely to operate. The prob- Studies of merodiploids tif-1 Itifr and tif-1 IrecA (J. lem can be avoided entirely by the use of a George, personal communication) indicate that tif-1 and tifr are codominant and that tif-1 and recA mutation scored without selection, such as mu- mutations probably affect the same gene product. tations from Lac+ to Lac- detected on indicator These observations support the version of the SOS medium, since survival and mutations are regulation model in Fig. 6 in which the wild-type scored on the same plates at a single cell den- rec-tif complex (probably a single protein) is bound sity. This kind of system is slow and expensive, to the lexA+ product in the uninduced state (see however, since large numbers of plates are re- text). Complete dominance of the tif-1 allele would quired to obtain accurate mutation frequency be expected if the recA-tif complex were free, as data. The use of selected mutants (auxotrophy shown in Fig. 6. to prototrophy, streptomycin sensitivity to re- sistance) can provide reliable mutation fre- LITERATURE CITED quency data in strains not subject to filamen- 1. Adler, H. I., W. D. Fisher, A. A. Hardigree, tous growth after the treatment used. Even in and G. S. Stapleton. 1966. Repair of radia- strains capable of undergoing lethal filamen- tion-induced damage to the cell division mechanism ofEscherichia coli. J. Bacteriol. tous growth at low cell density after a particu- 91:737-742. lar treatment, accurate induced mutation fre- la. Adler, H. I., and A. A. Hardigree. 1964. Anal- quencies can be obtained by limiting the treat- ysis of a gene controlling cell division and ment to levels still permitting 100% survival as sensitivity to radiation in Escherichia coli. J. determined by plating dilutions, or by plate- Bacteriol. 87:720-726. wash assays (227, 234) of cells harvested from lb. Adler, H. I., and A. A. Hardigree. 1965. densely populated plates after an incubation Postirradiation growth, division and recov- long enough to allow DNA replication but not ery in bacteria. Radiat. Res. 25:92-102. cell division. Many SOS-inducing treatments ic. Alper, T., and N. E. Gillies. 1958. Restoration ofEscherichia coli strain B after irradiation: are demonstrably mutagenic even in this high its dependence on suboptimal growth condi- survival range, at least in strains exhibiting a tions. J. Gen. Microbiol. 18:461-472. "shoulder" in their survival curves. Induced 2. Ames, B. N., W. E. Durston, E. Yamasaki, mutations scored in the 100% survival range of and F. D. Lee. 1973. Carcinogens are muta- treatment can be assumed to have arisen from gens: a simple test system combining liver the whole plated population (frequency per sur- homogenates for activation and bacteria for vivor = frequency per bacterium plated). Thus, detection. Proc. Natl. Acad. Sci. U.S.A. in lon+ K-12 and similar strains, mutagenesis 70:2281-2285. by treatments such as thymine starvation or by 2a. Anderson, E. H. 1951. Heat reactivation of ultraviolet inactivated bacteria. J. Bacteriol. temperature elevation in dnaBts, ligs or tif-1 61:389-394. derivatives is likely to be highly overestimated, 2b. Apte, B. N., H. Rhodes, and D. Zipser. 1975. if indeed it is demonstrated at all, unless treat- Mutation blocking the specific degradation ment is limited to levels that are sublethal in ofreinitiation polypeptides in E. coli. Nature survival assays. (London) 257:329-331. VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLI 899 3. Auerbach, C. 1976. Mutation research. Halsted Progess in nucleic acid research and molecu- Press, New York. lar biology, vol. 13. Academic Press Inc., 4. Bachmann, B., K. B. Low, and A. L. Taylor. New York. 1976. Recalibrated linkage map of Esche- 20. Boyce, R. P., and P. Howard-Flanders. 1964. richia coli K-12. Bacteriol. Rev. 40:116-167. Release of ultraviolet light-induced thymine 5. Baltimore, D. 1974. Is terminal deoxynucleoti- dimers from DNA inE. coli K-12. Proc. Natl. dyl transferase a somatic mutagen in lym- Acad. Sci. U.S.A. 51:293-300. phocytes? Nature (London) 248:409-411. 21. Boyle, J. M., M. C. Paterson, and R. B. Set- 6. Beckwith, J., and P. Rossow. 1974. Analysis of low. 1970. Excision-repair properties of an genetic regulatory mechanisms. Annu. Rev. Escherichia coli mutant deficient in DNA Genet. 8:1-13. polymerase. Nature (London) 226:708-710. 7. Benbow, R., A. Zuccarelli, and R. L. Sinshei- 22. Braun, A., and L. Grossman. 1974. An endonu- mer. 1974. A role for single-strand breaks in clease from Escherichia coli that acts prefer- bacteriophage 4)X174 genetic recombination. entially on UV-irradiated DNA and is absent J. Mol. Biol. 88:629-651. from the uvrA and uvrB mutants. Proc. Natl. 8. Beukers, R., and W. Berends. 1960. Isolation Acad. Sci. U.S.A. 71:1838-1842. and identification of the irradiation product 23. Bresler, S. E. 1975. Theory of misrepair muta- of thymine. Biochim. Biophys. Acta 41:550- genesis. Mutat. Res. 29:467-474. 551. 24. Bresler, S. E., M. I. Mosevitsky, and L. G. 9. Billen, D. 1969. Replication of the bacterial Vyacheslavov. 1973. Mutations as possible chromosome: location of new initiation sites replication errors in bacteria growing under after irradiation. J. Bacteriol. 97:1169-1175. conditions of thymine deficiency. Mutat. 10. Blanco, M., and R. Devoret. 1973. Repair Res. 19:281-293. mechanisms involved in prophage reactiva- 25. Bridges, B. A. 1966. A note on the mechanism tion and UV reactivation of UV-irradiated of UV mutagenesis in Escherichia coli. Mu- phage X. Mutat. Res. 17:293-305. tat. Res. 3:273-279. 11. Blanco, M., A. Levine, and R. Devoret. 1975. 26. Bridges, B. A. 1969. Mechanisms of radiation lexB: a new gene governing radiation sensi- mutagenesis in cellular and subcellular sys- tivity and lysogenic induction in Escherichia tems. Annu. Rev. Nucl. Sci. 19:139-178. coli K12, p. 379-382. In P. Hanawalt and R. 27. Bridges, B. A., R. E. Dennis, and R. J. Mun- B. Setlow (ed.), Molecular mechanisms for son. 1967. Differential induction and repair the repair of DNA. Plenum Press, New of ultraviolet damage leading to true rever- York. sions and external suppressor mutations of 12. Bockrath, R., and M. K. Cheung. 1975. The an ochre codon in Escherichia coli B/rWP2. role of nutrient broth supplementation in Genetics 57:897-908. UV mutagenesis of E. coli. Mutat. Res. 28. Bridges, B. A., W. J. H. Gray, M. H. L. Green, 19:23. M. A. Rothwell, and S. G. Sedgwick. 1973. 13. Bockstahler, L. E., C. D. Lytle, J. E. Stafford, Genetic and physiological separation of the and K. F. Haynes. 1976. Ultraviolet en- repair and mutagenic functions of the exrA hanced reactivation of a human virus: effect gene inEscherichia coli. Genetics 73(Suppl.): of delayed infection. Mutat. Res. 35:189-198. 123-129. 14. Bockstahler, L. E., and C. D. Lytle. 1971. X- 29. Bridges, B. A., J. Law, and R. J. Munson. ray enhanced reactivation of UV-irradiated 1968. Mutagenesis inEscherichia coli. 11. Ev- human virus. J. Virol. 8:601-602. idence for a common pathway for mutagene- 15. Bollum, F. J. 1974. Terminal deoxynucleotidyl sis by ultraviolet light, ionizing radiation transferase, p. 145-171. In P. D. Boyer (ed.), and thymine deprivation. Mol. Gen. Genet. The enzymes, vol. 1, 3rd ed. Academic Press 103:266-273. Inc., New York. 30. Bridges, B. A., and R. Mottershead. 1971. 16. Bonura, T., and K. C. Smith. 1975. Quantita- RecA+-dependent mutagenesis occurring be- tive evidence for enzymatically induced fore DNA replication in UV- and y-irradi- DNA double-strand breaks as lethal lesions ated Escherichia coli. Mutat. Res. 13:1-8. in UV-irradiatedpol+ andpolI strains ofE. 31. Bridges, B. A., R. P. Mottershead, and S. G. coli K-12. Photochem. Photobiol. 22:243-248. Sedgwick. 1976. Mutagenic DNA repair in 17. Bonura, T., and K. C. Smith. 1975. Enzymatic Escherichia coli. HI. Requirement for a func- production of deoxyribonucleic acid double- tion of DNA polymerase III in ultraviolet- strand breaks after ultraviolet irradiation of light mutagenesis. Mol. Gen. Genet. 144:53- Escherichia coli K-12. J. Bacteriol. 121:511- 58. 517. 32. Bridges, B. A., and R. J. Munson. 1966. Exci- 18. Borek, E., and A. Ryan. 1958. The transfer of sion-repair of DNA damage in an auxo- irradiation-elicited induction in a lysogenic trophic strain of E. coli. Biochem. Biophys. organism. Proc. Natl. Acad. Sci. U.S.A. Res. Commun. 22:268-273. 44:374-377. 33. Bridges, B. A., M. A. Rothwell, and M. H. L. 19. Borek, E., and A. Ryan. 1973. Lysogenic in- Green. 1973. Repair processes and dose-re- duction. In J. Davidson and W. Cohn (ed.), sponse curves in ultraviolet mutagenesis of 900 WITKIN BACTERIOL. REV. bacteria. An. Acad. Bras. Cienc. 45(Suppl): DNA in Escherichia coli. J. Mol. Biol. 67:1- 203-209. 10. 34. Britten, R. J., and E. H. Davidson. 1969. Gene 50. Cooper, P. K., and P. C. Hanawalt. 1972. Role regulation for higher cells: a theory. Science of DNA polymerase I and the rec system in 165:349-357. excision repair in Escherichia coli. Proc. 35. Brooks, K., and A. J. Clark. 1967. Behavior of Natl. Acad. Sci. U.S.A. 69:1156-1160. X bacteriophage in a recombination deficient 51. d'Ambrosio, S. M., and R. B. Setlow. 1976. strain ofEscherichia coli. J. Virol. 1:283-293. Enhancement of postreplicative repair in 36. Buhl, S. N., R. B. Setlow, and J. D. Regan. Chinese hamster cells. Proc. Natl. Acad. Sci. 1973. Recovery of the ability to synthesize U.S.A. 73:2396-2400. DNA in segments of normal size at long 52. Defais, M., P. Caillet-Fauquet, M. S. Fox, and times after ultraviolet irradiation of human M. Radman. 1976. Induction kinetics of mu- cells. Biophys. J. 13:1265-1275. tagenic DNA repair activity in E. coli follow- 37. Bush, J. W., and A. Markovitz. 1973. The ge- ing ultraviolet irradiation. Mol. Gen. Genet. netic basis for mucoidy and radiation sensi- 148:125-130. tivity in capR (lon) mutants ofE. coli K-12. 53. Defais, M., P. Fauquet, M. Radman, and M. Genetics 74:215-225. Errera. 1971. Ultraviolet reactivation and 39. Castellazzi, M., J. George, and G. Buttin. ultraviolet mutagenesis of X in different ge- 1972. Prophage induction and cell division in netic systems. Virology 43:495-503. E. coli. I. Further characterization of the 53a. Delaporte, B. 1956. La "restauration par vois- thermosensitive mutation tif-1 whose expres- inage" chez des bacteries irradies par des sion mimics the effect of UV irradiation. rayons X. Ann. Inst. Pasteur 91:727-735. Mol. Gen. Genet. 119:139-152. 54. De Lucia, P., and J. Cairns. 1969. Isolation of 40. Castellazzi, M., J. George, and G. Buttin. an E. coli strain with a mutation affecting 1972. Prophage induction and cell division in DNA polymerase. Nature (London) 224: E. coli. II. Linked (recA zab) and unlinked 1164-1166. (lex) suppressors of tif-1-mediated induction 54a. Demerec, M., and E. Cahn. 1953. Studies of and filamentation. Mol. Gen. Genet. mutability in nutritionally deficient strains 119:153-174. ofEscherichia coli. J. Bacteriol. 65:27-36. 40a. Cheung, M. K., and R. C. Bockrath. 1970. On 55. Devoret, R., M. Blanco, J. George, and M. the specificity of UV mutagenesis in E. coli. Radman. 1975. Recovery of phage X from Mutat. Res. 10:521-523. ultraviolet damage, p. 155-171. In P. Hana- 41. Clark, A. J. 1971. Toward a metabolic interpre- walt and R. B. Setlow (ed.), Molecular mech- tation of genetic recombination ofE. coli and anisms for repair of DNA, part A. Plenum its phages. Annu. Rev. Microbiol. 25:437- Press, New York. 464. 56. Devoret, R., and J. George. 1967. Induction 42. Clark, A. J. 1973. Recombination deficient mu- indirecte du prophage X par le rayonnement tants of E. coli and other bacteria. Annu. ultraviolet. Mutat. Res. 4:713-734. Rev. Genet. 7:67-86. 57. Donch, J. J., and J. Greenberg. 1974. The effect 43. Clark, A. J., and A. D. Margulies. 1965. Isola- of lex on UV sensitivity, filament formation tion and characterization of recombination- and X induction in lon mutants of Esche- deficient mutants ofE. coli K-12. Proc. Natl. richia coli. Mol. Gen. Genet. 128:277-281. Acad. Sci. U.S.A. 53:451-459. 58. Donch, J. J., and J. Greenberg. 1976. Suppres- 44. Cleaver, J. E. 1968. Defective repair replication sion of filamentation in a new lex mutant by of DNA in xeroderma pigmentosum. Nature a linked (lexA) mutation in Escherichia coli. (London) 218:652-656. Mutat. Res. 34:533-538. 45. Cleaver, J. E. 1969. Xeroderma pigmentosum: 59. Donch, J., J. Greenberg, and M. H. L. Green. a human disease in which an initial stage of 1970. Repression of induction by u.v. of X DNA repair is defective. Proc. Natl. Acad. phage by exrA mutations in E. coli. Genet. Sci. U.S.A. 63:428-435. Res. 15:87-97. 46. Cleaver, J. E. 1974. Repair processes for photo- 60. Doubleday, 0. P., B. A. Bridges, and M. H. L. chemical damage in mammalian cells, p. 1- Green. 1975. Mutagenic DNA repair inEsch- 75. In J. T. Lett, H. Adler, and M. R. Zelle erichia coli. II. Factors affecting loss of pho- (ed.), Advances in radiation biology, vol. 4. toreversibility of UV induced mutations. Academic Press Inc., New York. Mol. Gen. Genet. 140:221-230. 47. Cleaver, J. E., and D. Bootsma. 1975. Xero- 61. Doudney, C. 0. 1975. The two lesion hypothesis derma pigmentosum: biochemical and ge- for UV-induced mutation in relation to re- netic characteristics. Annu. Rev. Genet. covery of capacity for DNA replication, p. 9:19-38. 389-392. In P. Hanawalt and R. B. Setlow 48. Coleman, M. S., J. J. Hutton, P. DeSimone, (ed.), Molecular mechanisms for the repair of and F. J. Bollum. 1974. Terminal deoxynu- DNA, part A. Plenum Press, New York. cleotidyl transferase in human leukemia. 62. Doudney, C. 0. 1976. Mutation in ultraviolet Proc. Natl. Acad. Sci. U.S.A. 71:4404-4408. light-damaged microorganisms, p. 309-374. 49. Copper, P. K., and P. C. Hanawalt. 1972. Het- In. S. Y. Wang (ed.), and erogeneity of patch size in repair replication photobiology of nucleic acids, vol. II. Aca- VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLI 901 demic Press Inc., New York. ase I in pyrimidine dimer excision and repair 63. Doudney, C. 0. 1976. Complexity of the ultra- replication in Escherichia coli K12 following violet mutation frequency response curve in ultraviolet irradiation. Biochim. Biophys. Escherichia coli B/r: SOS induction, one-le- Acta 335:115-122. sion and two-lesion mutagenesis. J. Bacte- 79. Goldberg, A. L., and J. F. Dice. 1974. Intracel- riol. 128:815-826. lular protein degradation in mammalian and 64. Doudney, C. O., and C. S. Young. 1962. Ultra- bacterial cells. Annu. Rev. Biochem. 43:835- violet light induced mutation and deoxyribo- 869. nucleic acid replication in bacteria. Genetics 79a. Goldberg, A. L., and A. C. St. John. 1976. 47:1125-1138. Intracellular protein degradation in mam- 65. Drabble, W. T., and B. A. D. Stocker. 1968. R malian and bacterial cells: part 2. Annu. (transmissible drug resistance) factors in Rev. Biochem. 45:747-803. Salmonella typhimurium: pattern of trans- 80. Goldthwait, D., and F. Jacob. 1964. Sur le duction by-phage P22 and ultraviolet-protec- m6canisme de l'induction du d6veloppement tion effect. J. Gen. Microbiol. 53:109-123. du prophage chez les bacteries lysogenes. C. 66. Drake, J. W. 1970. The molecular basis of mu- R. Acad. Sci. Paris D 259:661-664. tation. Holden-Day, San Francisco. 81. Gottesman, M. M., M. L. Hicks, and M. Gel- 67. Drake, J. W., and R. H. Baltz. 1976. The bio- lert. 1973. Genetics and function of DNA li- chemistry of mutagenesis. Annu. Rev. Bio- gase in Escherichia coli. J. Mol. Biol. 77:531- chem. 45:11-37. 547. 68. Emmerson, P. T. 1968. Recombination-defi- 82. Goze, A., A. Sarasin, Y. Moule, and R. Devo- cient mutants of Escherichia coli that map ret. 1975. Induction and mutagenesis of pro- between thyA and argA. Genetics 60:19-30. phage X in Escherichia coli K12 by metabo- 69. Eyfjord, J. E., M. H. L. Green, and B. A. lites of aflatoxin B.. Mutat. Res. 28:1-7. Bridges. 1975. Mutagenic DNA repair in 83. Green, M. H. L., J. Greenberg, and J. Donch. Escherichia coli: conditions for error-free fill- 1969. Effect of a recA gene on cell division ing of daughter-strand gaps. J. Gen. Micro- and capsular polysaccharide production in a biol. 91:369-375. lon strain of Escherichia coli. Genet. Res. 70. Fauquet, P., and M. Defais. 1972. UV reactiva- 14:159-162. tion of phage X in a polA mutant of E. coli. 84. Greenberg, J., L. J. Berends, J. Donch, and M. Mutat. Res. 15:353-355. H. L. Green. 1974. exrB: a malB-linked gene 71. Ganesan, A. K. 1974. Persistence ofpyrimidine in Escherichia coli B involved in sensitivity dimers during postreplication repair in ul- to radiation and filament formation. Genet. traviolet light-irradiated Escherichia coli Res. 23:175-184. K12. J. Mol. Biol. 87:103-119. 85. Greenberg, J., L. Berends, J. Donch, and B. 72. Ganesan, A. K., and P. C. Seawell. 1975. The Johnson. 1975. Reversion studies with exrB effect oflexA and recF mutations on postrep- in Escherichia coli. Genet. Res. 25:109-117. lication repair and DNA synthesis in Esche- 86. Greenberg, J., J. Donch, and L. Berends. 1975. richia coli K12. Mol. Gen. Genet. 141:189- The dominance ofexrB over exrB+ in hetero- 206. diploids of Escherichia coli. Genet. Res. 72a. Gellert, M., and M. L. Bullock. 1970. DNA 25:39-44. ligase mutants of Escherichia coli. Proc. 87. Gross, J., and M. Gross. 1969. Genetic analysis Natl. Acad. Sci. U.S.A. 67:1580-1587. of an E. coli strain with a mutation affecting 73. George, D. L., and E. M. Witkin. 1974. Slow DNA polymerase. Nature (London) 224: excision repair in an mfd mutant of Esche- 1166-1168. richia coli B/r. Mol. Gen. Genet. 133:283-291. 88. Grossman, L., A. Braun, R. Feldberg, and I. 74. George, D. L., and E. M. Witkin. 1975. Ultra- Mahler. 1975. Enzymatic repair of DNA. violet light-induced responses of an mfd mu- Annu. Rev. Biochem. 44:19-43. tant of Escherichia coli B/r having a slow 88a. Gudas, L. J. 1976. The induction of protein X rate of dimer excision. Mutat. Res. 28:347- in DNA repair and cell division mutants of 354. Escherichia coli. J. Mol. Biol. 104:567-587. 75. George, J., M. Castellazzi, and G. Buttin. 89. Gudas, L. J., and A. B. Pardee. 1975. Model for 1975. Prophage induction and cell division in the regulation of Escherichia coli DNA re- E. coli. III. Mutations sfiA and sfiB restore pair functions. Proc. Natl. Acad. Sci. U.S.A. division in tif and Ion strains and permit the 72:2330-2334. expression of mutator properties of tif. Mol. 90. Gudas, L. J., and A. B. Pardee. 1976. DNA Gen. Genet. 140:309-332. synthesis inhibition and the induction ofpro- 76. George, J., and R. Devoret. 1971. Conjugal tein X in Escherichia coli. J. Mol. Biol. transfer of UV-damaged F-prime sex factors 101:459-477. and indirect induction of prophage X. Mol. 91. Hanawalt, P. C. 1975. Molecular mechanisms Gen. Genet. 111:103-119. involved in DNA repair. Genetics 79:179- 77. George, J., R. Devoret, and M. Radman. 1974. 197. Indirect ultraviolet-reactivation of phage X. 92. Hanawalt, P. 1975. Repair processes in diverse Proc. Natl. Acad. Sci. U.S.A. 71:144-147. systems: overview, p. 503-506. In P. Hana- 78. Glickman, B. 1974. The role of DNA polymer- walt and R. B. Setlow (ed.), Molecular mech- 902 WITKIN BACTERIOL. REV. anisms for repair of DNA, part B. Plenum 105. Howard-Flanders, P., W. D. Rupp, B. M. Wil- Press, New York. kins, and R. S. Cole. 1968. DNA replication 93. Harm, H. 1976. Repair ofUV-irradiated biolog- and recombination after UV irradiation. ical systems: photoreactivation, p. 219-263. Cold Spring Harbor Symp. Quant. Biol. In S. Y. Wang (ed.), Photochemistry and 33:195-205. photobiology of nucleic acids, vol. II. Aca- 106. Howard-Flanders, P., and L. Theriot. 1966. demic Press Inc., New York. Mutants ofEscherichia coli K-12 defective in 94. Harm, W., C. S. Rupert, and H. Harm. 1971. DNA repair and in genetic recombination. The study of photoenzymatic repair of UV Genetics 53:1137-1150. lesions in DNA by flash photolysis, p. 279- 107. Howarth, S. 1966. Increase in frequency of ul- 324. In A. C. Giese (ed.), Photophysiology, traviolet-induced mutation brought about by vol. 6. Academic Press Inc., New York. the colicine factor colI in Salmonella typhi- 95. Hart, R. W., and R. B. Setlow. 1975. Direct murium. Mutat. Res. 3:129-134. evidence that pyrimidine dimers in DNA re- 108. Hull, R. A. 1975. Effect of tsl mutations on Col sult in neoplastic transformation, p. 719-724. El expression in a recA strain ofE. coli K-12. In P. Hanawalt and R. B. Setlow (ed.), Mo- J. Bacteriol. 123:775-776. lecular mechanisms for repair of DNA, part 109. Ichikawa-Ryo, H., and S. Kondo. 1975. Indi- B. Plenum Press, New York. rect mutagenesis in phage lambda by ultra- 96. Hellman, K. B., K. F. Haynes, and L. E. Bock- violet preirradiation of host bacteria. J. Mol. stahler. 1974. Radiation enhanced survival Biol. 97:77-92. of a human virus in normal and malignant 110. Ikenaga, M., H. Ichikawa-Ryo, and S. Kondo. rat cells. Proc. Exp. Biol. Med. 145:255-262. 1975. The major cause of inactivation and 97. Hertman, I., and S. E. Luria. 1967. Transduc- mutation by 4-nitroquinoline-1-oxide in tion studies on the role of a rec+ gene in Escherichia coli: excisable 4NQO-purine ad- ultraviolet induction of prophage lambda. J. ducts. J. Mol. Biol. 92:341-356. Mol. Biol. 23:117-133. 111. Inouye, M. 1971. Pleiotropic effect of the recA 98. Hill, R. F. 1965. Ultraviolet-induced lethality gene of Escherichia coli: uncoupling of cell and reversion to prototrophy in Escherichia division from deoxyribonucleic acid replica- coli strains with normal and reduced dark tion. J. Bacteriol. 106:539-542. repair ability. Photochem. Photobiol. 4:563- 112. Inouye, M., and J. P. Guthrie. 1969. A muta- 568. tion which changes a membrane protein ofE. 99. Hill, R. F., and E. R. Nestmann. 1973. Effect of coli. Proc. Natl. Acad. Sci. U.S.A. 64:957- the recC gene in Escherichia coli on frequen- 961. cies of ultraviolet-induced mutants. Mutat. 113. Inouye, M., and A. B. Pardee. 1970. Changes of Res. 17:27-36. membrane proteins and their relation to 99a. Hirota, Y., A. Ryter, and F. Jacob. 1968. DNA synthesis and cell division ofE. coli. J. Thermosensitive mutants of E. coli affected Biol. Chem. 245:5813-5819. in the processes of DNA synthesis and cellu- 114. Iyer, V. N., and W. D. Rupp. 1971. Usefulness lar division. Cold Spring Harbor Symp. of benzoylated napthoylated DEAE-cellulose Quant. Biol. 33:677-693. to distinguish and fractionate double- 100. Holliday, R., and J. E. Pugh. 1975. DNA modi- stranded DNA bearing different extents of fication mechanisms and gene activity dur- single-stranded regions. Biochim. Biophys. ing development. Science 187:226-232. Acta 228:117-126. 101. Horii, Z. I., and A. J. Clark. 1973. Genetic 115. Jacob, F. 1950. Induction de la lyse et de la analysis of the RecF pathway to genetic re- production de bacteriophages chez un Pseu- combination in Escherichia coli K-12: isola- domonas pyocyanea lysogene. C. R. Acad. tion and characterization of mutants. J. Mol. Sci. Paris D 231:1585-1587. Biol. 80:327-344. 115a. Jacob, F. 1954. Mutation d'un bacteriophage lOla. Howard, B. D., and I. Tessman. 1964. Identi- induite par l'irradiation des seules bacteries- fication of the altered bases in mutated sin- h6tes avant l'infection: C. R. Acad. Sci. Paris gle-stranded DNA. J. Mol. Biol. 9:372-375. D 238:732-734. 102. Howard-Flanders, P. 1975. Repair by genetic 116. Kanazir, D. 1958. The apparent mutagenicity recombination in bacteria: overview, p. 265- of thymine deficiency. Biochim. Biophys. 274. In P. Hanawalt and R. B. Setlow (ed.), Acta 30:20-23. Molecular mechanisms for repair of DNA, 117. Kanner, L., and P. Hanawalt. 1970. Repair part A. Plenum Press Inc., New York. deficiency in a bacterial mutant defective in 103. Howard-Flanders, P., and R. P. Boyce. 1966. DNA polymerase. Biochem. Biophys. Res. DNA repair and genetic recombination; Commun. 39:149-155. studies on mutants ofEscherichia coli defec- 117a. Kantor, G. C., and R. A. Deering. 1966. Ul- tive in these processes. Radiat. Res. traviolet radiation studies on filamentous 6:(Suppl)156-184. Escherichia coli. J. Bacteriol. 92:1062-1069. 104. Howard-Flanders, P., E. Simson, and L. Ther- 119. Kelley, R. B., M. R. Atkinson, J. A. Huber- iot. 1964. A locus that controls filament for- man, and A. Kornberg. 1969. Excision ofthy- mation and sensitivity to radiation in Esche- mine dimers and other mismatched se- richia coli K-12. Genetics 49:237-246. quences by DNA polymerase of Escherichia VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLu 903 coli. Nature (London) 224:495-501. Photochem. Photobiol. 20:91-94. 120. Kelner, A. 1949. Effect of visible light on the 136. MacPhee, D. G. 1973. Effect of an R factor and recovery of Streptomyces griseus conidia caffeine on ultraviolet mutability on Salmo- from ultraviolet irradiation injury. Proc. nella typhimurium. Mutat. Res. 18:367-370. Natl. Acad. Sci. U.S.A. 35:73-79. 137. MacPhee, D. G. 1973. Effect of rec mutations 121. Kelner, A. 1949. Photoreactivation of ultravi- on the ultraviolet protecting and mutation- olet-irradiated Escherichia coli with special enhancing properties of the plasmid R- reference to the dose-reduction principle and Utrecht in Salmonella typhimurium. Mutat. to ultraviolet-induced mutation. J. Bacte- Res. 19:357-359. riol. 58:511-522. 138. Mackie, G., and D. B. Wilson. 1972. Regula- 122. Kirby, E. P., F. Jacob, and D. A. Goldthwait. tion of the gal operon of Escherichia coli by 1967. Prophage induction and filament for- the capR gene. J. Biol. Chem. 247:2973-2978. mation in a mutant strain of Escherichia 139. Maher, V. M., L. M. Ouellette, R. D. Curren, coli. Proc. Natl. Acad. Sci. U.S.A. 58:1903- and J. J. McCormick. 1976. Frequency of 1910. ultraviolet light-induced mutations is higher 123. Kirby, E. P., W. L. Ruff, and D. A. Gold- in xeroderma pigmentosum variant cells thwait. 1972. Cell division and prophage in- than in normal human cells. Nature (Lon- duction in Escherichia coli: effects of pantoyl don) 201:593-595. lactone and various furan derivatives. J. 140. Marcovich, H. 1956. Etude radiobiologique du Bacteriol. 111:447-453. systeme lysogene d'Escherichia coli K12. 1. 124. Kondo, S. 1969. Mutagenicity versus radiosen- Rayons X. Ann. Inst. Pasteur Paris 90:303- sitivity in Escherichia coli, p. 126-127. Pro- 319. ceedings ofthe XIIth International Congress 141. Marcovich, H. 1956. Etude de l'action des ray- on Genetics, vol. II. ons ultraviolets sur le systeme lysogene 125. Kondo, S. 1973. Evidence that mutations are Escherichia coli K12(X), K12S, X. Ann. Inst. induced by errors in repair and replication. Pasteur Paris 91:511-522. Genetics 73:109-122. 142. Marinus, M. G., and N. R. Morris. 1975. Pleio- 126. Kondo, S., and H. Ichikawa. 1973. Evidence tropic effects of a DNA adenine methylation that pretreatment of Escherichia coli cells mutation (dam-3) in Escherichia coli K12. with N-methyl-N'-nitro-N-nitrosoguanidine Mutat. Res. 28:15-26. enhances mutability of subsequently infect- 143. Marsden, H. S., E. C. Pollard, W. Ginoza, and ing phage X. Mol. Gen. Genet. 126:319-324. E. P. Randall. 1974. Involvement ofrecA and 127. Kondo, S., H. Ichikawa, K. Iwo, and T. Kato. exr genes in the in vivo inhibition of the 1970. Base-change mutagenesis and pro- recBC nuclease. J. Bacteriol. 118:465-470. phage induction in strains ofEscherichia coli 144. Masker, W., P. Hanawalt, and H. Shizuya. with different DNA repair capacities. Ge- role of DNA polymerase II in repair netics 66:187-217. replication in Escherichia coli. Nature (Lon- 128. Kornberg, A. 1974. DNA synthesis. W. H. don) New Biol. 244:242-243. Freeman and Co., San Francisco. 145. McCaffrey, R., D. F. Smoler, and D. Balti- 129. Latarjet, R. 1951. Induction, par les rayons X, more. 1973. Terminal deoxynucleotidyl de la production d'un bacteriophage chez B. transferase in a case of childhood acute lym- megatherium lysogene. Ann. Inst. Pasteur phoblastic leukemia. Proc. Natl. Acad. Sci. Paris 81:389-393. U.S.A. 70:521-525. 130. Lawrence, C. W., and R. Christensen. 1976. 146. McCann, J., N. E. Springarn, J. Kobori, and UV mutagenesis in radiation sensitive B. Ames. 1975. Detection of carcinogens as strains of yeast. Genetics 82:207-232. mutagens: bacterial tester strains with R 131. Lehmann, A. R. 1972. Postreplication repair of factor plasmids. Proc. Natl. Acad. Sci. DNA in ultraviolet-irradiated mammalian U.S.A. 72:979-983. cells. J. Mol. Biol. 66:319-337. 147. Melechen, N. E., and P. D. Skaar. 1962. The 132. Lehmann, A. R., and S. Kirk-Bell. 1972. Post- provocation of an early step of induction by replication repair ofDNA in ultraviolet-irra- thymine deprivation. Virology 16:21-29. diated mammalian cells: no gaps in DNA 148. Mennigmann, H. D. 1972. Pyrimidine dimers synthesized late after ultraviolet irradiation. as pre-mutational lesions in Escherichia coli Eur. J. Biochem. 31:438-445. WP2 Hcr-. Mol. Gen. Genet. 117:167-186. 133. Loeb, L. A., C. F. Springgate, and N. Battula. 149. Meyn, R. E., and R. M. Humphrey. 1971. De- 1974. Errors in DNA replication as a basis of oxyribonucleic acid synthesis in ultraviolet- malignant changes. Cancer Res. 34:2311- light-irradiated Chinese hamster cells. Bio- 2321. phys. J. 11:295-301. 134. Lwoff, A., L. Simonovitch, and N. Kjeldgaard. 150. Miller, C. G. 1975. Peptidases and proteases of 1950. Induction de la production de bacterio- Escherichia coli and Salmonella typhimu- phages chez une bacterie lysogene. Ann. rium. Annu. Rev. Microbiol. 29:485-504. Inst. Pasteur Paris 79:815-859. 151. Miura, A., and J. Tomizawa. 1968. Studies on 135. Lytle, C. D., S. G. Benane, and L. E. Bock- radiation-sensitive mutants of E. coli. 1I. stahler. 1974. Ultraviolet-enhanced reactiva- Participation of the Rec system in induction tion of herpes virus in human tumor cells. of mutation by ultraviolet irradiation. Mol. 904 WITKIN BACTERIOL. REV. Gen. Genet. 103:1-10. resistant and sensitive bacteria. Mutat. Res. 152. Monk, M. 1967. Observations on the mecha- 8:215-228. nism of indirect induction by mating with 165. Nishioka, H., and C. 0. Doudney. 1970. Differ- ultraviolet colU donors. Mol. Gen. Genet. ent modes of loss of photoreversibility of ul- 100:264-274. traviolet light-induced true and suppressor 153. Monk, M. 1969. Induction of phage X by trans- mutations to tryptophan independence in an ferred irradiated coil DNA. Mol. Gen. Genet. auxotrophic strain of Escherichia coli. Mu- 106:14-24. tat. Res. 9:349-358. 154. Monk, M., and J. Gross. 1971. Induction of 166. Noack, D., and S. Klaus. 1972. Inactivation prophage X in a mutant ofE. coli K12 defec- kinetics of lambda phage repressors in a mu- tive in initiation of DNA replication at high tant ofE. coli temperature sensitive in DNA temperatures. Mol. Gen. Genet. 110:299-306. replication. Mol. Gen. Genet. 115:216-224. 155. Monk, M., M. Peacey, and J. D. Gross. 1971. 167. Novick, A., and L. Szilard. 1949. Experiments Repair of damage induced by ultraviolet on light reactivation of ultraviolet-inacti- light in DNA polymerase-defective Esche- vated bacteria. Proc. Natl. Acad. Sci. U.S.A. richia coli cells. J. Mol. Biol. 58:623-630. 35:591-600. 156. Moody, E. E. M., K. B. Low, and D. W. 167a. Ono, J., and Y. Shimazu. 1966. Ultraviolet Mount. 1973. Properties of strains ofEsche- reactivation of a bacteriophage containing a richia coli K12 carrying mutant lex and rec single-stranded deoxyribonucleic acid as a alleles. Mol. Gen. Genet. 121:197-205. genetic element. Virology 29:295-302. 156a. Moreau, P., A. Bailone, and R. Devoret. 1976. 167b. Osborn, M., S. Person, S. Phillips, and F. Prophage X induction in Escherichia coli Funk. 1967: A determination of mutagen envA uvrB: a highly sensitive test for poten- specificity in bacteria using nonsense mu- tial carcinogens. Proc. Natl. Acad. Sci. tants of bacteriophage T4. J. Mol. Biol. U.S.A., 73:3700-3704. 26:437-447. 157. Morse, L. S., and C. Pauling. 1975. The induc- 167c. Otsuji, N., M. Sekiguchi, T. Ijima, and Y. tion oferror-prone repair as a consequence of Takagi. 1959. Induction of phage formation DNA ligase deficiency in Escherichia coli. in the lysogenic Escherichia coli by mitomy- Proc. Natl. Acad. Sci. U.S.A. 72:4645-4649. cin C. Nature (London) 184:1079-1080. 158. Mount, D. W., and C. Kosel. 1975. Ultraviolet- 168. Paterson, M. C., J. M. Boyle, and R. B. Set- induced mutation in UV-resistant, thermo- low. 1971. Ultraviolet and X-ray-induced re- sensitive derivatives oflexA strains ofE. coli sponses of a deoxyribonucleic acid polymer- K-12. Mol. Gen. Genet. 136:95-106. ase-deficient mutant of Escherichia coli. J. 158a. Mount, D. W., C. Kosel, and A. Walker. 1976. Bacteriol. 107:61-67. Inducible error-free DNA repair in tsl recA 169. Patrick, M. H., and R. 0. Rahn. 1976. Photo- mutants ofE. coli. Mol. Gen. Genet. 146:37- chemistry of DNA and polynucleotides: pho- 42. toproducts, p. 35-95. In S. Y. Wang (ed.), 159. Mount, D. W., K. B. Low, and S. J. Edmiston. Photochemistry and photobiology of nucleic 1972. Dominant mutations (lex) in Esche- acids, vol. II. Academic Press Inc., New richia coli K-12 which affect radiation sensi- York. tivity and frequency of ultraviolet light-in- 170. Pettijohn, D., and P. Hanawalt. 1964. Evi- duced mutations. J. Bacteriol. 112:886-893. dence for repair replication of ultraviolet 160. Mount, D. W., A. C. Walker, and C. Kosel. damaged DNA in bacteria. J. Mol. Biol. 1973. Suppression of lex mutations affecting 9:395-410. deoxyribonucleic acid repair in Escherichia 171. Pollard, E., and E. P. Randall. 1973. Studies coli K-12 by closely linked thermosensitive on the inducible inhibitor of radiation-in- mutations. J. Bacteriol. 116:950-956. duced DNA degradation of E. coli. Radiat. 161. Mount, D. W., A. C. Walker, and C. Kosel. Res. 55:265-279. 1975. Effect of tsl mutations in decreasing 172. Pritchard, R. H., and K. G. Lark. 1964. Induc- radiation sensitivity of a recA strain of tion of replication by thymine starvation at Escherichia coli K-12. J. Bacteriol. 121:1203- the chromosome origin in Escherichia coli. J. 1207. Mol. Biol. 9:288-307. 162. Mount, D. W., A. C. Walker, and C. Kosel. 173. Ptashne, M. 1967. Specific binding of the X 1975. Indirect suppression of radiation sensi- phage repressor to X DNA. Nature (London) tivity of a recA strain of Escherichia coli 214:232-234. K12, p. 383-388. In P. Hanawalt and R. B. 174. Radman, M. 1974. Phenomenology ofan induc- Setlow (ed.), Molecular mechanisms for re- ible mutagenic DNA repair pathway in pair of DNA, part A. Plenum Press, New Escherichia coli: SOS repair hypothesis, p. York. 128-142. In L. Prokash, F. Sherman, M. 163. Nelson, R. L., and H. S. Mason. 1972. An Miller, C. Lawrence, and H. W. Tabor (ed.), explicit hypothesis for chemical carcinogene- Molecular and environmental aspects of mu- sis. J. Theor. Biol. 37:197-200. tagenesis. Charles C Thomas Publisher, 164. Nishioka, H., and C. 0. Doudney. 1969. Differ- Springfield, Ill. ent modes of loss of photoreversibility of mu- 175. Radman, M. 1975. SOS repair hypothesis: phe- tation and lethal damage in ultraviolet-light nomenology of an inducible DNA repair VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLu 905 which is accompanied by mutagenesis, p. lambda. J. Virol. 11:879-885. 355-367. In P. Hanawalt and R. B. Setlow 189. Sedgwick, S. G. 1975. Inducible error-prone re- (ed.), Molecular mechanisms for repair of pair in Escherichia coli. Proc. Natl. Acad. DNA, part A. Plenum Press, New York. Sci. U.S.A. 72:2753-2757. 176. Radman, M., and R. Devoret. 1971. UV-reacti- 190. Sedgwick, S. G. 1975. Genetic and kinetic evi- vation of bacteriophage X in excision repair- dence for different types of postreplication deficient hosts: independence of red func- repair in Escherichia coli B. J. Bacteriol. tions and attachment regions. Virology 123:154-161. 43:504-506. 190a. Sedgwick, S. G. 1976. Misrepair of overlap- 177. Regan, J. D., and R. B. Setlow. 1974. Two ping daughter-strand gaps as a possible forms of repair in the DNA of human cells mechanism for UV induced mutagenesis in damaged by chemical carcinogens and muta- uvr strains of Escherichia coli: a general gens. Cancer Res. 34:3318-3325. model for induced mutagenesis by misrepair 178. Regan, J. D., J. E. Trosko, and W. L. Carrier. (SOS repair) of closely spaced DNA lesions. 1968. Evidence for excision of ultraviolet-in- Mutat. Res., in press. duced pyrimidine dimers from the DNA of 191. Sedgwick, S. G., and B. A. Bridges. 1972. Sur- human cells in vitro. Biophys. J. 8:319-325. vival, mutation and capacity to repair sin- 178a. Reich, E., D. B. Rifkin, and E. Shaw (ed.). gle-strand breaks after gamma-irradiation 1975. Proteases and biological control. Cold in different Exr strains of Escherichia coli. Spring Harbor Laboratory, Cold Spring Har- Mol. Gen. Genet. 119:93-102. bor, N.Y. 192. Sedgwick, S., and B. A. Bridges. 1974. Re- 179. Reznikoff, W. S. 1972. The operon revisited. quirement for either DNA polymerase I or Annu. Rev. Genet. 6:133-156. DNA polymerase m in post-replication re- 180. Roberts, J. W., and C. W. Roberts. 1975. Prote- pair in excision proficient Escherichia coli. olytic cleavage of bacteriophage lambda re- Nature (London) 249:348-349. pressor in induction. Proc. Natl. Acad. Sci. 193. Setlow, J. K. 1966. Photoreactivation. Radiat. U.S.A. 72:147-151. Res. 6(Suppl.):141-155. 181. Rosner, J. L., L. R. Kass, and M. B. Yarmolin- 194. Setlow, J. K., and M. E. Boling. 1970. Ultravi- sky. 1968. Parallel behavior of F and P1 in olet action spectra for mutation in Esche- causing indirect induction oflysogenic bacte- richia coli. Mutat. Res. 9:437-442. ria. Cold Spring Harbor Symp. Quant. Biol. 195. Setlow, R. B., and W. L. Carrier. 1964. The 33:785-789. disappearance ofthymine dimers from DNA: 182. Rothman, R. H., T. Kato, and A. J. Clark. an error-correcting mechanism. Proc. Natl. 1975. The beginning of an investigation of Acad. Sci. U.S.A. 51:226-231. the role of recF in the pathways of metabo- 196. Setlow, R. B., J. D. Regan, J. German, and W. lism of ultraviolet-irradiated DNA in Esche- L. Carrier. 1969. Evidence that xeroderma richia coli, p. 283-291. In P. C. Hanawalt and pigmentosum cells do not perform to first R. B. Setlow (ed.), Molecular mechanisms step in the repair of ultraviolet damage to for repair of DNA, part A. Plenum Press, their DNA. Proc. Natl. Acad. Sci. U.S.A. New York. 64:1035-1041. 183. Rupert, C. S., H. Harm, and K. To. 1973. The 197. Setlow, R., and J. K. Setlow. 1972. Effects of anatomy of direct repair. An. Acad. Bras. radiation on polynucleotides. Annu. Rev. Cienc. 45(Suppl.):151-159. Biophys. Bioeng. 1:293-346. 184. Rupert, C. S. 1975. Enzymatic photoreactiva- 198. Shinagawa, K., and T. Itoh. 1973. Inactivation tion: overview, p. 73-87. In P. Hanawalt and of DNA binding activity of repressor in ex- R. B. Setlow (ed.), Molecular mechanisms tracts of lambda lysogens treated with mito- for repair of DNA, part A. Plenum Press, mycin C. Mol. Gen. Genet. 126:103-110. New York. 199. Shineberg, B., and D. Zipser. 1973. The lon 185. Rupp, W. D., and P. Howard-Flanders. 1968. gene and degradation of3l-galactosidase non- Discontinuities in the DNA synthesized in sense fragments. J. Bacteriol. 116:1469-1471. an excision-defective strain of Escherichia 199a. Sicard, N., and R. Devoret. 1962. Effects de la coli following ultraviolet irradiation. J. Mol. carence en thymine sur des souches lyso- Biol. 31:291-304. genes K12T- et 15T-. C. R. Acad. Sci. Paris 186. Rupp, W. D., C. E. Wilde III, D. L. Reno, and D 255:1417-1419. P. Howard-Flanders. 1971. Exchanges be- 200. Siegel, E. C. 1973. Ultraviolet-sensitive muta- tween DNA strands in ultraviolet-irradiated tor strain ofEscherichia coli K-12. J. Bacte- Escherichia coli. J. Mol. Biol. 61:25-44. riol. 113:145-160. 187. Sarin, P. S., and R. C. Gallo. 1974. Terminal 201. Sinzinis, B. I., G. B. Smirnov, and A. S. deoxynucleotidyl transferase in chronic mye- Saenko. 1973. Repair deficiency in an Esche- logenous leukemia. J. Biol. Chem. 249:8051- richia coli UV-sensitive mutator strain 8053. uvr5O2. Biochem. Biophys. Res. Commun. 188. Schuster, H., D. Beyersmann, M. Miko- 53:309-317. lajczyk, and M. Schlicht. 1973. Prophage in- 202. Smith, K. C. Ultraviolet radiation effects. In duction by high temperature in thermosensi- K. C. Smith (ed.), The science of photobiol- tive mutants lysogenic for bacteriophage ogy, in press. 906 WITKIN BACTERIOL. REV. 203. Smith, K. C., and D. H. C. Meun. 1970. Repair mutants ofEscherichia coli by bacteriophage of radiation-induced damage in Escherichia lambda. J. Bacteriol. 113:1326-1332. coli. I. Effect ofrec mutations on postreplica- 218. Webb, R. B., and M. M. Malina. 1967. Muta- tion repair of damage due to ultraviolet ra- genesis in E. coli by visible light. Science diation. J. Mol. Biol. 51:459-472. 156:1104-1105. 204. Springgate, C., and L. A. Loeb. 1973. Muta- 219. Wechsler, J. A., and J. D. Gross. 1971. Esche- genic DNA polymerases in human leukemic richia coli mutants temperature-sensitive for cells. Proc. Natl. Acad. Sci. U.S.A. 70:245- DNA synthesis. Mol. Gen. Genet. 113:273- 249. 284. 205. Srivastava, B. I. S., and J. Minowada. 1973. 220. Weigle, J. J. 1953. Induction of mutation in a Terminal deoxynucleotidyl transferase ac- bacterial virus. Proc. Natl. Acad. Sci. U.S.A. tivity in a cell line (molt-4) derived from the 39:628-636. peripheral blood of a patient with acute lym- 221. West, S. C., K. A. Powell, and P. T. Emmer- phoblastomic leukemia. Biochem. Biophys. son. 1975. recAl-dependent inactivation of Res. Commun. 51:529-535. the lambda repressor in Escherichia coli ly- 206. Sussman, R., and H. BenZeev. 1975. Proposed sogens by y-radiation and by tif expression. mechanism of bacteriophage lambda induc- Mol. Gen. Genet. 141:1-8. tion: acquisition of binding sites for lambda 222. Witkin, E. M. 1947. Genetics of resistance to repressor by DNA of the host. Proc. Natl. radiation in Escherichia coli. Genetics Acad. Sci. U.S.A. 72:1973-1976. 32:221-248. 207. Sutton, H. E., and R. P. Wagner. 1975. Muta- 223. Witkin, E. M. 1956. Time, temperature and tion and enzyme functions in humans. Annu. protein synthesis: a study of ultraviolet-in- Rev. Genet. 9:187-212. duced mutation in bacteria. Cold Spring 208. Swenson, P. A., and R. L. Schenley. 1974. Harbor Symp. Quant. Biol. 21:123-140. Respiration, growth and viability of repair- 224. Witkin, E. M. 1959. Post-irradiation metabo- deficient mutants of Escherichia coli after lism and the timing of ultraviolet-induced ultraviolet irradiation. Int. J. Radiat. Biol. mutations in bacteria, p. 280-299. In Pro- 25:51-60. ceedings of the X International Congress on 209. Tait, R. C., A. L. Harris, and D. W. Smith. Genetics, vol. I. 1974. DNA repair in Escherichia coli mu- 225. Witkin, E. M. 1966. Radiation-induced muta- tants deficient in DNA polymerases I, II and/ tions and their repair. Science 152:1345- or III. Proc. Natl. Acad. Sci. U.S.A. 71:675- 1353. 679. 226. Witkin, E. M. 1967. The radiation sensitivity of 209a. Tessman, E. S., and T. Ozaki. 1960. The in- Escherichia coli B: a hypothesis relating fila- teraction of phage S13 with ultraviolet irra- ment formation and prophage induction. diated host cells and properties of the ultra- Proc. Natl. Acad. Sci. U.S.A. 57:1275-1279. violet irradiated phage. Virology 12:431-449. 227. Witkin, E. M. 1967. Mutation-proof and muta- 210. Tomizawa, J., and T. Ogawa. 1967. Effect of tion-prone modes ofsurvival in derivatives of ultraviolet irradiation on bacteriophage Escherichia coli B differing in sensitivity to lambda immunity. J. Mol. Biol. 23:247-263. ultraviolet light. Brookhaven Symp. Biol. 211. Tomizawa, J., and H. Ogawa. 1972. Structural 20:17-55. genes of ATP-dependent deoxyribonuclease 228. Witkin, E. M. 1969. The role of DNA repair and ofE. coli. Nature (London) New Biol. 239:14- recombination in mutagenesis, p. 225-245. In 16. Proceedings of XII International Congress on 212. Trosko, J. E., E. H. Y. Chu, and W. L. Carrier. Genetics, vol. 3. 1965. The induction of thymine dimers in 229. Witkin, E. M. 1969. The mutability toward ultraviolet-irradiated mammalian cells. Ra- ultraviolet light of recombination-deficient diat. Res. 24:667-672. strains of Escherichia coli. Mutat. Res. 8:9- 213. Van Sluis, C. A., I. E. Mattern, and M. C. 14. Paterson. 1974. Properties of uvrE mutants 230. Witkin, E. M. 1969. Ultraviolet-induced muta- ofEscherichia coli K12. I. Effects of UV irra- tion and DNA repair. Annu. Rev. Microbiol. diation on DNA metabolism. Mutat. Res. 23:487-514. 25:273-279. 231. Witkin, E. M. 1971. Ultraviolet mutagenesis in 215. Volkert, M., D. L. George, and E. M. Witkin. strains of E. coli deficient in DNA polymer- 1976. Partial suppression of the LexA pheno- ase. Nature (London) New Biol. 229:81-82. type by mutations (rnm) which restore ultra- 232. Witkin, E. M. 1972. Ultraviolet mutagenesis in violet resistance but not ultraviolet mutabil- repair-deficient derivatives of Escherichia ity to Escherichia coli B/r uvrA lexA. Mutat. coli B/r: uvrA recB and uvrA recC strains. Res. 36:17-28. Mutat. Res. 16:235-242. 216. Wacker, A., H. Dellweg, and D. Jackerts. 1962. 233. Witkin, E. M. 1973. Ultraviolet mutagenesis Thymin-dimerisierung und Uberlebensrate in bacteria: the inducible nature of error- bei Bakterien. J. Mol. Biol. 4:410-412. prone repair. An. Acad. Bras. Cienc. 217. Walker, J. R., C. L. Ussery, and J. S. Allen. 45(Suppl.):188-192. 1973. Bacterial cell division regulation: lyso- 234. Witkin, E. M. 1974. Thermal enhancement of genization of conditional cell division lon ultraviolet mutability in a tif-1 uvrA deriva- VOL. 40, 1976 UV MUTAGENESIS AND DNA REPAIR IN E. COLI 907 tive of Escherichia coli B/r: evidence that Mutat. Res. 25:407-409. ultraviolet mutagenesis depends upon an in- 242. Witkin, E. M., and I. E. Wermundsen. 1973. ducible function. Proc. Natl. Acad. Sci. Do ultraviolet-induced mutations to strepto- U.S.A. 71:1930-1934. mycin resistance exhibit susceptibility to 235. Witkin, E. M. 1975. Elevated mutability of mutation frequency decline? Mutat. Res. polA and uvrA poLA derivatives of Esche- 19:261-264. richia coli B/r at sublethal doses of ultravi- 243. Worcel, A. 1970. Induction of chromosome re- olet light: evidence for an inducible error- initiation in a thermosensitive DNA mutant prone repair system ("SOS repair") and its ofE. coli. J. Mol. Biol. 52:371-386. anomalous expression in these strains. Ge- 243a. Yoshinaga, K., M. Fusaya, and M. Shimo- netics 79(Suppl.):199-213. mura. 1971. On the mechanism of thymine- 236. Witkin, E. M. 1975. Relationships among re- less death in E. coli strain 15 TAU, p. 41-53. pair, mutagenesis and survival: overview, p. Reports of Faculty of Science, Shizuoka Uni- 347-353. In P. Hanawalt and R. B. Setlow versity, vol. 6. (ed.), Molecular mechanisms for repair of 244. Youngs, D. A., and K. C. Smith. 1973. Involve- DNA, part A. Plenum Press, New York. ment of DNA polymerase III in excision re- 237. Witkin, E. M. 1975. Thermal enhancement of pair after ultraviolet irradiation. Nature ultraviolet mutability in a dnaB uvrA deriv- (London) New Biol. 244:240-241. ative of Escherichia coli B/r: evidence for 245. Youngs, D. A., and K. C. Smith. 1973. Evi- inducible error-prone repair, p. 369-378. In dence for the control by exrA andpolA genes P. Hanawalt and R. B. Setlow (ed.), Molecu- of two branches of the uvr gene-dependent lar mechanisms for repair of DNA, part A. excision repair pathway in Escherichia coli Plenum Press, New York. K-12. J. Bacteriol. 116:175-182. 238. Witkin, E. M. 1975. Persistence and decay of 246. Youngs, D. A., and K. C. Smith. 1973. X-ray thermo-inducible error-prone repair activity sensitivity and repair capacity of a polAl in nonfilamentous derivatives of tif-1 Esche- exrA strain ofEscherichia coli K-12. J. Bac- richia coli B/r: the timing of some critical teriol. 114:121-127. events in ultraviolet mutagenesis. Mol. Gen. 247. Youngs, D. A., and K. C. Smith. 1976. Genetic Genet. 142:87-103. control of multiple pathways of post-replica- 239. Witkin, E. M., and E. L. Farquharson. 1969. tional repair in uvrB strains of Escherichia Enhancement and diminution of ultraviolet coli K-12. J. Bacteriol. 125:102-110. light-initiated mutagenesis by posttreat- 248. Youngs, D. A., E. van der Scheuren, and K. C. ment with caffeine inEscherichia coli, p. 36- Smith. 1974. Separate branches of the uvr 49. In G. E. W. Wolstenholme and M. gene-dependent excision repair process in ul- O'Connor (ed.), Ciba Foundation Symposium traviolet-irradiated Escherichia coli K-12 on Mutation as Cellular Process. J. & A. cells; their dependence upon growth medium Churchill, Ltd., London. and the polA, recA, recB, and exrA genes. J. 240. Witkin, E. M., and D. L. George. 1973. Ultra- Bacteriol. 117:717-725. violet mutagenesis in poIA and uvrA poIA 249. Zelle, M. R., J. E. Ogg, and A. Hollaender. derivatives of Escherichia coli B/r: evidence 1958. Photoreactivation of induced mutation for an inducible error-prone repair system. and inactivation of Escherichia coli exposed Genetics 73(Suppl.):91-108. to various wave lengths of monochromatic 241. Witkin, E. M., and E. C. Parisi. 1974. Bromou- ultraviolet radiation. J. Bacteriol. 75:190- racil mutagenesis: mispairing or misrepair? 198.